From 641d7abefdfb68455764742ca39787da1966a4e3 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Tue, 3 Jun 2025 23:52:59 +0000 Subject: [PATCH 01/32] test commit --- test.py | 1 + 1 file changed, 1 insertion(+) create mode 100644 test.py diff --git a/test.py b/test.py new file mode 100644 index 0000000..3b18e51 --- /dev/null +++ b/test.py @@ -0,0 +1 @@ +hello world From 3b05347e6790f76f58f1bef2f77b64950c305927 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Wed, 4 Jun 2025 17:33:42 +0000 Subject: [PATCH 02/32] scratch folder added --- scratch/fiddle.py | 0 1 file changed, 0 insertions(+), 0 deletions(-) create mode 100644 scratch/fiddle.py diff --git a/scratch/fiddle.py b/scratch/fiddle.py new file mode 100644 index 0000000..e69de29 From 566ae05c890f46a4f5215dcd79b52373babe8477 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Thu, 5 Jun 2025 16:50:28 +0000 Subject: [PATCH 03/32] scratch files for JAMUN run on alanine dipeptide uncapped --- configs/experiment/sample_custom.yaml | 1 + .../sample_uncapped_single_shape.yaml | 45 ++++++++++++++++ .../experiment/train_test_single_shape.yaml | 21 ++++---- scratch/fiddle.py | 52 +++++++++++++++++++ src/jamun/cmdline/sample.py | 4 +- src/jamun/cmdline/train.py | 2 +- 6 files changed, 113 insertions(+), 12 deletions(-) create mode 100644 configs/experiment/sample_uncapped_single_shape.yaml diff --git a/configs/experiment/sample_custom.yaml b/configs/experiment/sample_custom.yaml index 9b657e3..f1f23c9 100644 --- a/configs/experiment/sample_custom.yaml +++ b/configs/experiment/sample_custom.yaml @@ -43,3 +43,4 @@ sampler: logger: wandb: group: sample_custom + notes: sampling on AA diff --git a/configs/experiment/sample_uncapped_single_shape.yaml b/configs/experiment/sample_uncapped_single_shape.yaml new file mode 100644 index 0000000..280beb0 --- /dev/null +++ b/configs/experiment/sample_uncapped_single_shape.yaml @@ -0,0 +1,45 @@ +# @package _global_ + +defaults: + - override /callbacks: + - sampler/save_trajectory.yaml + - _self_ + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/timewarp/2AA-1-large/train/" + traj_pattern: "^(.*)-traj-arrays.npz" + pdb_file: AA-traj-state0.pdb + filter_codes: + - AA + subsample: 100 + max_datasets: 1 + +num_sampling_steps_per_batch: 20 +num_batches: 5 +num_init_samples_per_dataset: 1 +repeat_init_samples: 1 +continue_chain: true + +# New 2AA +wandb_train_run_path: sule-shashank/jamun/7spefobw + +# Old 2AA +# wandb_train_run_path: prescient-design/jamun/zzt8s3rc + +checkpoint_type: best_so_far +sigma: 0.04 +M: 1.0 +delta: 0.04 +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: 100.0 + +sampler: + _target_: jamun.sampling.Sampler + devices: 1 + +logger: + wandb: + group: sample_uncapped_2AA + notes: single_shape_AA diff --git a/configs/experiment/train_test_single_shape.yaml b/configs/experiment/train_test_single_shape.yaml index 8d9d1ef..e674be4 100644 --- a/configs/experiment/train_test_single_shape.yaml +++ b/configs/experiment/train_test_single_shape.yaml @@ -22,35 +22,36 @@ data: train: - _target_: jamun.data.MDtrajDataset root: "${paths.data_path}/timewarp/2AA-1-large/train/" - trajfiles: + traj_files: - AA-traj-arrays.npz - pdbfile: AA-traj-state0.pdb + pdb_file: AA-traj-state0.pdb subsample: 100 - label: "AA" + label: AA val: - _target_: jamun.data.MDtrajDataset root: "${paths.data_path}/timewarp/2AA-1-large/train/" - trajfiles: + traj_files: - AA-traj-arrays.npz - pdbfile: AA-traj-state0.pdb + pdb_file: AA-traj-state0.pdb subsample: 100 - label: "AA" + label: AA test: - _target_: jamun.data.MDtrajDataset root: "${paths.data_path}/timewarp/2AA-1-large/train/" - trajfiles: + traj_files: - AA-traj-arrays.npz - pdbfile: AA-traj-state0.pdb + pdb_file: AA-traj-state0.pdb subsample: 100 - label: "AA" + label: AA trainer: val_check_interval: 0.5 - max_epochs: 1 + max_epochs: 200 logger: wandb: group: train_test + notes: "alanine dipeptide" \ No newline at end of file diff --git a/scratch/fiddle.py b/scratch/fiddle.py index e69de29..9bfd1e3 100644 --- a/scratch/fiddle.py +++ b/scratch/fiddle.py @@ -0,0 +1,52 @@ +import functools +import logging +import os + +import dotenv +import tqdm + +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("jamun") + +import torch + +torch.set_float32_matmul_precision("high") + +import e3nn +import e3tools.nn + +e3nn.set_optimization_defaults(jit_script_fx=False) + +import jamun +import jamun.data +import jamun.distributions +import jamun.model +import jamun.model.arch + +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") + +# Device. +device = torch.device("cuda:0") + +datasets = { + "train": jamun.data.parse_datasets_from_directory( + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, + subsample=100, + max_datasets=1, + ) +} + +datamodule = jamun.data.MDtrajDataModule( + datasets=datasets, + batch_size=32, + num_workers=2, +) +datamodule.setup(None) + +print("hello world") + diff --git a/src/jamun/cmdline/sample.py b/src/jamun/cmdline/sample.py index 7479906..8601981 100644 --- a/src/jamun/cmdline/sample.py +++ b/src/jamun/cmdline/sample.py @@ -2,6 +2,7 @@ import sys import traceback from typing import Sequence +import pdb import dotenv import e3nn @@ -71,8 +72,9 @@ def run(cfg): # Overwrite the checkpoint path in the config. cfg.model.checkpoint_path = checkpoint_path model = hydra.utils.instantiate(cfg.model) - + print(f'Checkpoint path at: {checkpoint_path}') init_datasets = hydra.utils.instantiate(cfg.init_datasets) + breakpoint() init_graphs = get_initial_graphs( init_datasets, num_init_samples_per_dataset=cfg.num_init_samples_per_dataset, diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 5ec8883..86e4e7b 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -85,7 +85,7 @@ def run(cfg): ) else: checkpoint_path = None - + print(f'Saving checkpoints @ {checkpoint_path}') trainer.fit(model, datamodule=datamodule, ckpt_path=checkpoint_path) if wandb_logger and isinstance(trainer.profiler, lightning.pytorch.profilers.PyTorchProfiler): From 6d658e9850b1f14fe54eedac18cdac08e0b09e14 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Thu, 5 Jun 2025 23:58:26 +0000 Subject: [PATCH 04/32] prep for hidden state introduction to denoiser --- .../sample_uncapped_single_shape.yaml | 18 ++-- scratch/__init__.py | 0 scratch/fiddle.py | 88 +++++++++++++++++-- 3 files changed, 93 insertions(+), 13 deletions(-) create mode 100644 scratch/__init__.py diff --git a/configs/experiment/sample_uncapped_single_shape.yaml b/configs/experiment/sample_uncapped_single_shape.yaml index 280beb0..1553a52 100644 --- a/configs/experiment/sample_uncapped_single_shape.yaml +++ b/configs/experiment/sample_uncapped_single_shape.yaml @@ -1,9 +1,13 @@ # @package _global_ -defaults: - - override /callbacks: - - sampler/save_trajectory.yaml - - _self_ +# defaults: +# - override /callbacks: +# - sampler/save_trajectory.yaml +# - _self_ + +# callbacks: +# viz: +# sigma_list: ["${model.sigma_distribution.sigma}"] init_datasets: _target_: jamun.data.parse_datasets_from_directory @@ -12,11 +16,11 @@ init_datasets: pdb_file: AA-traj-state0.pdb filter_codes: - AA - subsample: 100 + subsample: 1 max_datasets: 1 -num_sampling_steps_per_batch: 20 -num_batches: 5 +num_sampling_steps_per_batch: 100 +num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true diff --git a/scratch/__init__.py b/scratch/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/scratch/fiddle.py b/scratch/fiddle.py index 9bfd1e3..8c58622 100644 --- a/scratch/fiddle.py +++ b/scratch/fiddle.py @@ -1,3 +1,4 @@ +# %% import functools import logging import os @@ -23,14 +24,13 @@ import jamun.model import jamun.model.arch +# %% +# dataset dotenv.load_dotenv("../.env", verbose=True) JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") - -# Device. -device = torch.device("cuda:0") - +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") datasets = { - "train": jamun.data.parse_datasets_from_directory( + "test": jamun.data.parse_datasets_from_directory( root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", traj_pattern="^(.*)-traj-arrays.npz", pdb_file="AA-traj-state0.pdb", @@ -46,7 +46,83 @@ batch_size=32, num_workers=2, ) -datamodule.setup(None) +datamodule.setup('test') + + +# %% +arch = functools.partial( + jamun.model.arch.E3Conv, + irreps_out="1x1e", + irreps_hidden="120x0e + 32x1e", + irreps_sh="1x0e + 1x1e", + n_layers=5, + edge_attr_dim=64, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=32, + residue_index_embedding_dim=8, + use_residue_information=True, + use_residue_sequence_index=False, + hidden_layer_factory=functools.partial( + e3tools.nn.ConvBlock, + conv=e3tools.nn.Conv, + ), + output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), +) + + +# %% +# Device. +device = torch.device("cuda:0") + +# xplore generated trajectories +import mdtraj as md +# --- Option 2: Loading your own DCD and Topology file --- +print("\n--- Loading Your Own DCD and Topology File (Example) ---") +# Replace these with the actual paths to your files +dcd_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/predicted_samples/dcd/joined.dcd" # Your DCD trajectory file +topology_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/topology.pdb" # Your topology file (e.g., .pdb, .prmtop, .psf) + +# Create dummy files for this example to run without error if you don't have them +# In a real scenario, you would have your actual DCD and PDB files. +print(f'DCD file path exists: {os.path.exists(dcd_file_path)}') +print(f'Topology file path exists: {os.path.exists(topology_file_path)}') +try: + print(f"Attempting to load trajectory: {dcd_file_path}") + print(f"Using topology: {topology_file_path}") + + # The 'top' argument is crucial for DCD files + traj_custom = md.load_dcd(dcd_file_path, top=topology_file_path) + + print(f"Successfully loaded custom trajectory!") + print(f"Number of frames: {traj_custom.n_frames}") + print(f"Number of atoms: {traj_custom.n_atoms}") + # You can now perform analysis on traj_custom + # For example, calculate RMSD, distances, angles, etc. + +except FileNotFoundError: + print(f"Error: One or both files not found: {dcd_file_path}, {topology_file_path}") +except Exception as e: + print(f"An error occurred while loading your files: {e}") +print("-" * 30) + +# %% +from jamun.metrics._ramachandran import plot_ramachandran + +phi = md.compute_phi(traj_custom) +psi = md.compute_psi(traj_custom) + +import matplotlib.pyplot as plt +import numpy as np +fig = plt.figure() +ax = fig.add_subplot() +s = ax.scatter(phi[1], psi[1], cmap='hot', alpha=1.0) +ax.set_xlim((-np.pi, np.pi)) +ax.set_ylim((-np.pi, np.pi)) +c = fig.colorbar(s) print("hello world") +# load a jamun trained model + +# %% From 9184121ba6a30b47f77ef7bc8118c285c81d5124 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Mon, 9 Jun 2025 23:47:06 +0000 Subject: [PATCH 05/32] introduced hidden state in data, made a test denoiser module with conditioning --- scratch/config.yaml | 125 +++++++ scratch/denoiser_test.py | 399 ++++++++++++++++++++++ scratch/e3conv_test.py | 143 ++++++++ scratch/fiddle.py | 197 ++++++----- scratch/visualize_traj_data.py | 48 +++ src/jamun/data/_mdtraj.py | 5 +- src/jamun/model/arch/e3conv.py | 2 +- src/jamun/utils/__init__.py | 2 +- src/jamun/utils/data_with_residue_info.py | 5 +- 9 files changed, 845 insertions(+), 81 deletions(-) create mode 100644 scratch/config.yaml create mode 100644 scratch/denoiser_test.py create mode 100644 scratch/e3conv_test.py create mode 100644 scratch/visualize_traj_data.py diff --git a/scratch/config.yaml b/scratch/config.yaml new file mode 100644 index 0000000..5318d5f --- /dev/null +++ b/scratch/config.yaml @@ -0,0 +1,125 @@ +float32_matmul_precision: high +task_name: train +run_group: dev +run_key: ${now:%Y-%m-%d}_${now:%H-%M-%S} +python: + version: ${python_version:micro} +init_time: ${now:%y-%m-%d_%H:%M:%S} +compute_average_squared_distance_from_data: true +data: + datamodule: + _target_: jamun.data.MDtrajDataModule + batch_size: 32 + num_workers: 4 + datasets: + train: + - _target_: jamun.data.MDtrajDataset + root: ${paths.data_path}/timewarp/2AA-1-large/train/ + traj_files: + - AA-traj-arrays.npz + pdb_file: AA-traj-state0.pdb + subsample: 100 + label: AA + val: + - _target_: jamun.data.MDtrajDataset + root: ${paths.data_path}/timewarp/2AA-1-large/train/ + traj_files: + - AA-traj-arrays.npz + pdb_file: AA-traj-state0.pdb + subsample: 100 + label: AA + test: + - _target_: jamun.data.MDtrajDataset + root: ${paths.data_path}/timewarp/2AA-1-large/train/ + traj_files: + - AA-traj-arrays.npz + pdb_file: AA-traj-state0.pdb + subsample: 100 + label: AA + use_residue_information: true +model: + arch: + _target_: jamun.model.arch.E3Conv + _partial_: true + irreps_out: 1x1e + irreps_hidden: 120x0e + 32x1e + irreps_sh: 1x0e + 1x1e + n_layers: 2 + edge_attr_dim: 64 + atom_type_embedding_dim: 8 + atom_code_embedding_dim: 8 + residue_code_embedding_dim: 32 + residue_index_embedding_dim: 8 + use_residue_information: ${data.use_residue_information} + use_residue_sequence_index: false + num_atom_types: 20 + max_sequence_length: 10 + num_atom_codes: 10 + num_residue_types: 25 + hidden_layer_factory: + _target_: e3tools.nn.ConvBlock + _partial_: true + conv: + _target_: e3tools.nn.Conv + _partial_: true + output_head_factory: + _target_: e3tools.nn.EquivariantMLP + _partial_: true + irreps_hidden_list: + - ${model.arch.irreps_hidden} + optim: + _target_: torch.optim.Adam + _partial_: true + lr: 0.002 + weight_decay: 0.0 + max_radius: 1000.0 + average_squared_distance: null + add_fixed_noise: false + add_fixed_ones: false + align_noisy_input_during_training: true + align_noisy_input_during_evaluation: true + mean_center: true + mirror_augmentation_rate: 0.0 + use_torch_compile: true + torch_compile_kwargs: + fullgraph: true + dynamic: true + mode: default + _target_: jamun.model.Denoiser + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 +trainer: + _target_: lightning.Trainer + limit_train_batches: 1.0 + val_check_interval: 0.5 + max_epochs: 200 +logger: + wandb: + _target_: lightning.pytorch.loggers.WandbLogger + project: jamun + entity: null + offline: false + log_model: all + group: train_test + notes: alanine dipeptide +callbacks: + timing: + _target_: jamun.callbacks.Timing + lr_monitor: + _target_: lightning.pytorch.callbacks.LearningRateMonitor + model_checkpoint: + _target_: lightning.pytorch.callbacks.ModelCheckpoint + dirpath: ${hydra:runtime.output_dir}/checkpoints + save_top_k: 5 + save_last: true + monitor: val/loss + viz: + _target_: jamun.callbacks.VisualizeDenoise + datasets: ${data.datamodule.datasets.val} + sigma_list: + - ${model.sigma_distribution.sigma} +paths: + root_path: ${oc.env:JAMUN_ROOT_PATH, "."} + data_path: ${oc.env:JAMUN_DATA_PATH, ${paths.root_path}/data} + run_path: ${paths.root_path}/outputs/${task_name}/${run_group}/runs/${run_key} diff --git a/scratch/denoiser_test.py b/scratch/denoiser_test.py new file mode 100644 index 0000000..8c42642 --- /dev/null +++ b/scratch/denoiser_test.py @@ -0,0 +1,399 @@ +import logging +from typing import Callable, Dict, Optional, Tuple, Union + +import lightning.pytorch as pl +import numpy as np +import torch +import torch_geometric +from e3tools import radius_graph, scatter + +from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing, +from jamun.utils.align import kabsch_algorithm + + +class Denoiser(pl.LightningModule): + """The main denoiser model.""" + + def __init__( + self, + arch: Callable[..., torch.nn.Module], + optim: Callable[..., torch.optim.Optimizer], + sigma_distribution: torch.distributions.Distribution, + max_radius: float, + average_squared_distance: float, + add_fixed_noise: bool, + add_fixed_ones: bool, + align_noisy_input_during_training: bool, + align_noisy_input_during_evaluation: bool, + mean_center: bool, + mirror_augmentation_rate: float, + bond_loss_coefficient: float = 1.0, + normalization_type: Optional[str] = "JAMUN", + sigma_data: Optional[float] = None, # Only used if normalization_type is "EDM" + lr_scheduler_config: Optional[Dict] = None, + use_torch_compile: bool = True, + torch_compile_kwargs: Optional[Dict] = None, + ): + super().__init__() + self.save_hyperparameters(logger=False) + + self.g = arch() + if use_torch_compile: + if torch_compile_kwargs is None: + torch_compile_kwargs = {} + + self.g = torch.compile(self.g, **torch_compile_kwargs) + + py_logger = logging.getLogger("jamun") + py_logger.info(self.g) + + self.optim_factory = optim + self.lr_scheduler_config = lr_scheduler_config + self.sigma_distribution = sigma_distribution + self.max_radius = max_radius + + self.add_fixed_noise = add_fixed_noise + self.add_fixed_ones = add_fixed_ones + if self.add_fixed_noise and self.add_fixed_ones: + raise ValueError("Can't add fixed noise and fixed ones at the same time") + if self.add_fixed_noise: + py_logger.info("Adding fixed noise") + if self.add_fixed_ones: + py_logger.info("Adding fixed ones") + + self.average_squared_distance = average_squared_distance + py_logger.info(f"Average squared distance = {self.average_squared_distance}") + + self.align_noisy_input_during_training = align_noisy_input_during_training + if self.align_noisy_input_during_training: + py_logger.info("Aligning noisy input during training.") + else: + py_logger.info("Not aligning noisy input during training.") + + self.align_noisy_input_during_evaluation = align_noisy_input_during_evaluation + if self.align_noisy_input_during_evaluation: + py_logger.info("Aligning noisy input during evaluation.") + else: + py_logger.info("Not aligning noisy input during evaluation.") + + self.mean_center = mean_center + if self.mean_center: + py_logger.info("Mean centering input and output.") + else: + py_logger.info("Not mean centering input and output.") + + self.mirror_augmentation_rate = mirror_augmentation_rate + py_logger.info(f"Mirror augmentation rate: {self.mirror_augmentation_rate}") + + self.normalization_type = normalization_type + if self.normalization_type is not None: + py_logger.info(f"Normalization type: {self.normalization_type}") + else: + py_logger.info("No normalization") + + self.sigma_data = sigma_data + if self.normalization_type == "EDM" and self.sigma_data is None: + raise ValueError("sigma_data must be provided when normalization_type is 'EDM'") + elif self.normalization_type != "EDM" and self.sigma_data is not None: + raise ValueError("sigma_data can only be used when normalization_type is 'EDM'") + + self.bond_loss_coefficient = bond_loss_coefficient + + def conditioner(self, y: torch_geometric.data.Batch) -> torch.Tensor: + conditioned_structures = [] + for positions in y.hidden_state: + aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) + conditioned_structures.append(aligned_positions) + return conditioned_structures + + def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: + # pos [B, ...] + sigma = unsqueeze_trailing(sigma, x.pos.ndim) + + y = x.clone("pos") + if self.add_fixed_ones: + noise = torch.ones_like(x.pos) + hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] + elif self.add_fixed_noise: + torch.manual_seed(0) + num_batches = x.batch.max().item() + 1 + if len(x.pos.shape) == 2: + num_nodes_per_batch = x.pos.shape[0] // num_batches + noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) + hidden_noise = [torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat(num_batches, 1) for i in range(len(x.hidden_state))] + if len(x.pos.shape) == 3: + num_nodes_per_batch = x.pos.shape[1] + noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) + hidden_noise = [torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state))] + else: + noise = torch.randn_like(x.pos) + hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] + y.pos = x.pos + sigma * noise + for i in range(len(y.hidden_state)): + y.hidden_state[i] = x.hidden_state[i] + hidden_noise[i] + if torch.rand(()) < self.mirror_augmentation_rate: + y.pos = -y.pos + return y + + def score(self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: + """Compute the score function.""" + sigma = torch.as_tensor(sigma).to(y.pos) + return (self.xhat(y, sigma).pos - y.pos) / (unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2) + + def normalization_factors(self, sigma: float, D: int = 3) -> Tuple[float, float, float, float]: + """Normalization factors for the input and output.""" + sigma = torch.as_tensor(sigma) + + if self.normalization_type is None: + return 1.0, 0.0, 1.0, sigma + + if self.normalization_type == "EDM": + c_skip = (self.sigma_data**2) / (sigma**2 + self.sigma_data**2) + c_out = sigma * self.sigma_data / torch.sqrt(self.sigma_data**2 + sigma**2) + c_in = 1 / torch.sqrt(sigma**2 + self.sigma_data**2) + c_noise = torch.log(sigma / self.sigma_data) * 0.25 + return c_in, c_skip, c_out, c_noise + + if self.normalization_type == "JAMUN": + A = torch.as_tensor(self.average_squared_distance) + B = torch.as_tensor(2 * D * sigma**2) + + c_in = 1.0 / torch.sqrt(A + B) + c_skip = A / (A + B) + c_out = torch.sqrt((A * B) / (A + B)) + c_noise = torch.log(sigma) / 4 + return c_in, c_skip, c_out, c_noise + + raise ValueError(f"Unknown normalization type: {self.normalization_type}") + + def loss_weight(self, sigma: float, D: int = 3) -> float: + """Loss weight for this graph.""" + _, _, c_out, _ = self.normalization_factors(sigma, D) + return 1 / (c_out**2) + + def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Tensor: + """Compute the effective radial cutoff for the noise level.""" + return torch.sqrt((self.max_radius**2) + 6 * (sigma**2)) + + def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torch_geometric.data.Batch: + """Add edges to the graph based on the effective radial cutoff.""" + if "batch" in y: + batch = y["batch"] + else: + batch = torch.zeros(y.num_nodes, dtype=torch.long, device=self.device) + + # Our dataloader already adds the bonded edges. + bonded_edge_index = y.edge_index + + with torch.cuda.nvtx.range("radial_graph"): + radial_edge_index = radius_graph(y.pos, radial_cutoff, batch) + + with torch.cuda.nvtx.range("concatenate_edges"): + edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) + if bonded_edge_index.numel() == 0: + bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device) + else: + bond_mask = torch.cat( + ( + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device), + torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=self.device), + ), + dim=0, + ) + + y.edge_index = edge_index + y.bond_mask = bond_mask + return y + + def xhat_normalized( + self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] + ) -> torch_geometric.data.Batch: + """Compute the denoised prediction using the normalization factors from JAMUN.""" + sigma = torch.as_tensor(sigma).to(y.pos) + D = y.pos.shape[-1] + + # Compute the normalization factors. + with torch.cuda.nvtx.range("normalization_factors"): + c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) + radial_cutoff = self.effective_radial_cutoff(sigma) / c_in + + # Adjust dimensions. + c_in = unsqueeze_trailing(c_in, y.pos.ndim - 1) + c_skip = unsqueeze_trailing(c_skip, y.pos.ndim - 1) + c_out = unsqueeze_trailing(c_out, y.pos.ndim - 1) + c_noise = c_noise.unsqueeze(0) + + # Add edges to the graph. + with torch.cuda.nvtx.range("add_edges"): + y = self.add_edges(y, radial_cutoff) + + with torch.cuda.nvtx.range("scale_y"): + y_scaled = y.clone("pos") + y_scaled.pos = y.pos * c_in + scaled_hidden_state = [] + for positions in y_scaled.hidden_state: + scaled_hidden_state.append(positions * c_in) + y_scaled.hidden_state = scaled_hidden_state + + with torch.cuda.nvtx.range("clone_y"): + xhat = y.clone("pos") + + with torch.cuda.nvtx.range("conditioning"): + conditioned_structures = self.conditioner(y_scaled) + + with torch.cuda.nvtx.range("g"): + g_pred = self.g(torch.cat([y_scaled.pos, *conditioned_structures]), topology=y_scaled, \ + c_noise=c_noise, effective_radial_cutoff=radial_cutoff) + + xhat.pos = c_skip * y.pos + c_out * g_pred + return xhat + + def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): + """Compute the denoised prediction.""" + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_y"): + y = mean_center(y) + + with torch.cuda.nvtx.range("xhat_normalized"): + xhat = self.xhat_normalized(y, sigma) + + # Mean center the prediction. + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_xhat"): + xhat = mean_center(xhat) + + return xhat + + def noise_and_denoise( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch]: + """Add noise to the input and denoise it.""" + with torch.no_grad(): + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_x"): + x = mean_center(x) + + sigma = torch.as_tensor(sigma).to(x.pos) + + with torch.cuda.nvtx.range("add_noise"): + y = self.add_noise(x, sigma) + + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_y"): + y = mean_center(y) + + # Aligning each batch. + if align_noisy_input: + with torch.cuda.nvtx.range("align_A_to_B_batched"): + y = align_A_to_B_batched(y, x) + + with torch.cuda.nvtx.range("xhat"): + xhat = self.xhat(y, sigma) + + return xhat, y + + def compute_loss( + self, + x: torch_geometric.data.Batch, + xhat: torch.Tensor, + sigma: Union[float, torch.Tensor], + ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + """Compute the loss.""" + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_x"): + x = mean_center(x) + + D = xhat.pos.shape[-1] + + # Compute the raw loss. + with torch.cuda.nvtx.range("raw_coordinate_loss"): + raw_coordinate_loss = (xhat.pos - x.pos).pow(2).sum(dim=-1) + + # Take the mean over each graph. + with torch.cuda.nvtx.range("mean_over_graphs"): + mse = scatter(raw_coordinate_loss, x.batch, dim=0, dim_size=x.num_graphs, reduce="mean") + + # Compute the scaled RMSD. + with torch.cuda.nvtx.range("scaled_rmsd"): + rmsd = torch.sqrt(mse) + scaled_rmsd = rmsd / (sigma * np.sqrt(D)) + + # Account for the loss weight across graphs and noise levels. + with torch.cuda.nvtx.range("loss_weight"): + loss = mse * x.loss_weight + loss = loss * self.loss_weight(sigma, D) + + return loss, { + "mse": mse, + "rmsd": rmsd, + "scaled_rmsd": scaled_rmsd, + } + + def noise_and_compute_loss( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + """Add noise to the input and compute the loss.""" + xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) + return self.compute_loss(x, xhat, sigma) + + def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during training.""" + with torch.cuda.nvtx.range("sample_sigma"): + sigma = self.sigma_distribution.sample().to(self.device) + + loss, aux = self.noise_and_compute_loss( + batch, + sigma, + align_noisy_input=self.align_noisy_input_during_training, + ) + + # Average the loss and other metrics over all graphs. + with torch.cuda.nvtx.range("mean_over_graphs"): + aux["loss"] = loss + for key in aux: + aux[key] = aux[key].mean() + + self.log(f"train/{key}", aux[key], prog_bar=False, batch_size=batch.num_graphs, sync_dist=False) + + return { + "sigma": sigma, + **aux, + } + + def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during validation.""" + sigma = self.sigma_distribution.sample().to(self.device) + loss, aux = self.noise_and_compute_loss(batch, sigma, align_noisy_input=self.align_noisy_input_during_training) + + # Average the loss and other metrics over all graphs. + aux["loss"] = loss + for key in aux: + aux[key] = aux[key].mean() + self.log( + f"val/{key}", aux[key], prog_bar=(key == "scaled_rmsd"), batch_size=batch.num_graphs, sync_dist=True + ) + + return { + "sigma": sigma, + **aux, + } + + def configure_optimizers(self): + """Set up the optimizer and learning rate scheduler.""" + optimizer = self.optim_factory(params=self.parameters()) + + out = {"optimizer": optimizer} + if self.lr_scheduler_config: + scheduler = self.lr_scheduler_config.pop("scheduler") + out["lr_scheduler"] = { + "scheduler": scheduler(optimizer), + **self.lr_scheduler_config, + } + + return out diff --git a/scratch/e3conv_test.py b/scratch/e3conv_test.py new file mode 100644 index 0000000..6cf9721 --- /dev/null +++ b/scratch/e3conv_test.py @@ -0,0 +1,143 @@ +from typing import Callable + +import e3nn +import torch +import torch_geometric +from e3nn import o3 +from e3nn.o3 import Irreps +from e3tools import scatter +from torch import Tensor +from jamun.model.atom_embedding import AtomEmbeddingWithResidueInformation, SimpleAtomEmbedding +from jamun.model.noise_conditioning import NoiseConditionalScaling, NoiseConditionalSkipConnection + + +class E3Conv(torch.nn.Module): + """A simple E(3)-equivariant convolutional neural network, similar to NequIP.""" + + def __init__( + self, + irreps_out: str | Irreps, + irreps_hidden: str | Irreps, + irreps_sh: str | Irreps, + hidden_layer_factory: Callable[..., torch.nn.Module], + output_head_factory: Callable[..., torch.nn.Module], + use_residue_information: bool, + n_layers: int, + edge_attr_dim: int, + atom_type_embedding_dim: int, + atom_code_embedding_dim: int, + residue_code_embedding_dim: int, + residue_index_embedding_dim: int, + use_residue_sequence_index: bool, + num_atom_types: int = 20, + max_sequence_length: int = 10, + num_atom_codes: int = 10, + num_residue_types: int = 25, + test_equivariance: bool = False, + reduce: str | None = None, + N_structures: int = 1 + ): + super().__init__() + + self.test_equivariance = test_equivariance + self.irreps_out = o3.Irreps(irreps_out) + self.irreps_hidden = o3.Irreps(irreps_hidden) + self.irreps_sh = o3.Irreps(irreps_sh) + self.n_layers = n_layers + self.edge_attr_dim = edge_attr_dim + self.N_structures = N_structures + self.sh = o3.SphericalHarmonics(irreps_out=self.irreps_sh, normalize=True, normalization="component") + self.bonded_edge_attr_dim, self.radial_edge_attr_dim = self.edge_attr_dim // 2, (self.edge_attr_dim + 1) // 2 + self.embed_bondedness = torch.nn.Embedding(2, self.bonded_edge_attr_dim) + + if use_residue_information: + self.atom_embedder = AtomEmbeddingWithResidueInformation( + atom_type_embedding_dim=atom_type_embedding_dim, + atom_code_embedding_dim=atom_code_embedding_dim, + residue_code_embedding_dim=residue_code_embedding_dim, + residue_index_embedding_dim=residue_index_embedding_dim, + use_residue_sequence_index=use_residue_sequence_index, + num_atom_types=num_atom_types, + max_sequence_length=max_sequence_length, + num_atom_codes=num_atom_codes, + num_residue_types=num_residue_types, + ) + else: + self.atom_embedder = SimpleAtomEmbedding( + embedding_dim=atom_type_embedding_dim + + atom_code_embedding_dim + + residue_code_embedding_dim + + residue_index_embedding_dim + ) + + self.initial_noise_scaling = NoiseConditionalScaling(self.atom_embedder.irreps_out) + self.initial_projector = hidden_layer_factory( + irreps_in=self.initial_noise_scaling.irreps_out, + irreps_out=self.irreps_hidden, + irreps_sh=N_structures*self.irreps_sh, + edge_attr_dim=edge_attr_dim, + ) + + self.layers = torch.nn.ModuleList() + self.noise_scalings = torch.nn.ModuleList() + self.skip_connections = torch.nn.ModuleList() + for _ in range(n_layers): + self.layers.append( + hidden_layer_factory( + irreps_in=self.irreps_hidden, + irreps_out=self.irreps_hidden, + irreps_sh=N_structures*self.irreps_sh, + edge_attr_dim=self.edge_attr_dim, + ) + ) + self.noise_scalings.append(NoiseConditionalScaling(self.irreps_hidden)) + self.skip_connections.append(NoiseConditionalSkipConnection(self.irreps_hidden)) + + self.output_head = output_head_factory(irreps_in=self.irreps_hidden, irreps_out=self.irreps_out) + self.output_gain = torch.nn.Parameter(torch.tensor(0.0)) + self.reduce = reduce + + def forward( + self, + pos: Tensor, # should be [batch_size*N, 3T], T is the number of previous time-steps + topology: torch_geometric.data.Batch, + c_noise: Tensor, + effective_radial_cutoff: float, + ) -> torch_geometric.data.Batch: + # Extract edge attributes. + edge_index = topology["edge_index"] + bond_mask = topology["bond_mask"] + + src, dst = edge_index # compute edge spherical harmonics over concat structures + positions = torch.split(pos, 3, dim=-1) + edge_sh = [] + for block in positions: + edge_vec = block[src] - block[dst] + edge_sh.append(self.sh(edge_vec)) + edge_sh = torch.cat(edge_sh, dim=-1) + + # print(f"Edge spherical harmonics: {type(edge_sh)}") + bonded_edge_attr = self.embed_bondedness(bond_mask) + edge_vec_main = positions[0][src] - positions[0][dst] + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec_main.norm(dim=1), + 0.0, + effective_radial_cutoff, + self.radial_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr), dim=-1) + + node_attr = self.atom_embedder(topology) + node_attr = self.initial_noise_scaling(node_attr, c_noise) + node_attr = self.initial_projector(node_attr, edge_index, edge_attr, edge_sh) + for scaling, skip, layer in zip(self.noise_scalings, self.skip_connections, self.layers): + node_attr = skip(node_attr, layer(scaling(node_attr, c_noise), edge_index, edge_attr, edge_sh), c_noise) + node_attr = self.output_head(node_attr) + node_attr = node_attr * self.output_gain + + if self.reduce is not None: + node_attr = scatter(node_attr, topology.batch, dim=0, reduce=self.reduce) + + return node_attr diff --git a/scratch/fiddle.py b/scratch/fiddle.py index 8c58622..85fc6ad 100644 --- a/scratch/fiddle.py +++ b/scratch/fiddle.py @@ -5,12 +5,12 @@ import dotenv import tqdm - +from typing import Union logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) py_logger = logging.getLogger("jamun") import torch - +torch.cuda.is_available() torch.set_float32_matmul_precision("high") import e3nn @@ -29,6 +29,8 @@ dotenv.load_dotenv("../.env", verbose=True) JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +# %% datasets = { "test": jamun.data.parse_datasets_from_directory( root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", @@ -43,86 +45,129 @@ datamodule = jamun.data.MDtrajDataModule( datasets=datasets, - batch_size=32, + batch_size=3, num_workers=2, ) datamodule.setup('test') - - -# %% -arch = functools.partial( - jamun.model.arch.E3Conv, - irreps_out="1x1e", - irreps_hidden="120x0e + 32x1e", - irreps_sh="1x0e + 1x1e", - n_layers=5, - edge_attr_dim=64, - atom_type_embedding_dim=8, - atom_code_embedding_dim=8, - residue_code_embedding_dim=32, - residue_index_embedding_dim=8, - use_residue_information=True, - use_residue_sequence_index=False, - hidden_layer_factory=functools.partial( - e3tools.nn.ConvBlock, - conv=e3tools.nn.Conv, - ), - output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), -) - - -# %% -# Device. -device = torch.device("cuda:0") - -# xplore generated trajectories -import mdtraj as md -# --- Option 2: Loading your own DCD and Topology file --- -print("\n--- Loading Your Own DCD and Topology File (Example) ---") -# Replace these with the actual paths to your files -dcd_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/predicted_samples/dcd/joined.dcd" # Your DCD trajectory file -topology_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/topology.pdb" # Your topology file (e.g., .pdb, .prmtop, .psf) - -# Create dummy files for this example to run without error if you don't have them -# In a real scenario, you would have your actual DCD and PDB files. -print(f'DCD file path exists: {os.path.exists(dcd_file_path)}') -print(f'Topology file path exists: {os.path.exists(topology_file_path)}') -try: - print(f"Attempting to load trajectory: {dcd_file_path}") - print(f"Using topology: {topology_file_path}") - - # The 'top' argument is crucial for DCD files - traj_custom = md.load_dcd(dcd_file_path, top=topology_file_path) - - print(f"Successfully loaded custom trajectory!") - print(f"Number of frames: {traj_custom.n_frames}") - print(f"Number of atoms: {traj_custom.n_atoms}") - # You can now perform analysis on traj_custom - # For example, calculate RMSD, distances, angles, etc. - -except FileNotFoundError: - print(f"Error: One or both files not found: {dcd_file_path}, {topology_file_path}") -except Exception as e: - print(f"An error occurred while loading your files: {e}") -print("-" * 30) +_, data_batch = next(enumerate(datamodule.test_dataloader())) # %% -from jamun.metrics._ramachandran import plot_ramachandran - -phi = md.compute_phi(traj_custom) -psi = md.compute_psi(traj_custom) - -import matplotlib.pyplot as plt -import numpy as np -fig = plt.figure() -ax = fig.add_subplot() -s = ax.scatter(phi[1], psi[1], cmap='hot', alpha=1.0) -ax.set_xlim((-np.pi, np.pi)) -ax.set_ylim((-np.pi, np.pi)) -c = fig.colorbar(s) +import hydra +from hydra import initialize, compose +from omegaconf import DictConfig, OmegaConf + +with initialize(config_path=f"../outputs/train/dev/runs/2025-06-04_20-07-32/.hydra", job_name="my_script_job", version_base=None): + # 2. Compose the configuration + # `config_name` refers to the base YAML file (e.g., "config.yaml" -> "config") + # You can also pass overrides programmatically, e.g., + # cfg = compose(config_name="config", overrides=["model.irreps_out='2x0e'"]) + cfg: DictConfig = compose(config_name="config") + + print("Full configuration loaded by Hydra:") + print(OmegaConf.to_yaml(cfg)) + print("-" * 50) + print(torch.cuda.is_available()) + # 3. Instantiate the model + try: + print("Attempting to instantiate model...") + # hydra.utils.instantiate is still the way to go + from jamun.utils import compute_average_squared_distance_from_datasets + average_squared_distance = compute_average_squared_distance_from_datasets(datasets['test'], cfg.model.max_radius) + cfg.model.average_squared_distance = average_squared_distance + model: jamun.model.Denoiser = hydra.utils.instantiate(cfg.model) + # print("\nSuccessfully instantiated model:") + # print(model) + # print("-" * 50) + + # # Example: Create some dummy input data + # batch_s = 2 + # dummy_input = torch.randn(batch_s, model.irreps_in.dim) + # print(f"Created dummy input with shape: {dummy_input.shape}") + + # # Perform a forward pass + # output = model(dummy_input) + # print(f"Model output shape: {output.shape}") + # assert output.shape == (batch_s, model.irreps_out.dim) + # print("Dummy forward pass successful!") + + except Exception as e: + print(f"Error during model instantiation or processing: {e}") + import traceback + traceback.print_exc() + + +# %% play w/ spherical harmonics +from e3nn import o3 +sh = o3.SphericalHarmonics(irreps_out="1x0e + 1x1e", normalize=True, normalization="component") +irreps_sh = e3nn.o3.Irreps("1x0e + 1x1e") +# %% +full_pos = torch.cat([molecule_pos, molecule_pos_extra], dim=-1) +blocks = torch.split(full_pos, 3, dim=-1) +src, dst = molecule.edge_index +edge_sh = [] +for block in blocks: + edge_vec = block[src] - block[dst] + edge_sh.append(sh(edge_vec)) +edge_sh = torch.cat(edge_sh, dim=-1) + + +# %% test the new e3conv_test class +from e3conv_test import E3Conv +import torch_geometric +from e3tools import radius_graph + +trial_model = E3Conv( + irreps_out="1x1e", + irreps_hidden="120x0e + 32x1e", + irreps_sh="1x0e + 1x1e", + n_layers=1, + edge_attr_dim=8, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=32, + residue_index_embedding_dim=8, + use_residue_information=False, + use_residue_sequence_index=False, + hidden_layer_factory=functools.partial( + e3tools.nn.ConvBlock, + conv=e3tools.nn.Conv, + ), + output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), + N_structures=2 + ) -print("hello world") +# %% postprocess data for plugging into model +def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_geometric.data.Batch: + radial_edge_index = radius_graph(y.pos, cutoff, batch=y["batch"]) + bonded_edge_index = y.edge_index + edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) + bond_mask = torch.cat( + ( + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), + torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), + ), + dim=0, + ) + y.edge_index = edge_index + y.bond_mask = bond_mask + return y + +# add bond mask--do this only once! +y = add_bond_mask(data_batch) -# load a jamun trained model +# %% +from jamun.utils.align import kabsch_algorithm +from jamun.utils import mean_center +from e3tools import scatter + +# y = mean_center(y) +def conditioner(y: torch_geometric.data.Batch) -> torch.Tensor: + conditioned_structures = [] + for positions in y.hidden_state: + aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) + conditioned_structures.append(aligned_positions) + return conditioned_structures + +print('hello world') # %% diff --git a/scratch/visualize_traj_data.py b/scratch/visualize_traj_data.py new file mode 100644 index 0000000..16c8141 --- /dev/null +++ b/scratch/visualize_traj_data.py @@ -0,0 +1,48 @@ +# xplore generated trajectories +import mdtraj as md +import os +# --- Option 2: Loading your own DCD and Topology file --- +print("\n--- Loading Your Own DCD and Topology File (Example) ---") +# Replace these with the actual paths to your files +dcd_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/predicted_samples/dcd/joined.dcd" # Your DCD trajectory file +topology_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/topology.pdb" # Your topology file (e.g., .pdb, .prmtop, .psf) + +# Create dummy files for this example to run without error if you don't have them +# In a real scenario, you would have your actual DCD and PDB files. +print(f'DCD file path exists: {os.path.exists(dcd_file_path)}') +print(f'Topology file path exists: {os.path.exists(topology_file_path)}') +try: + print(f"Attempting to load trajectory: {dcd_file_path}") + print(f"Using topology: {topology_file_path}") + + # The 'top' argument is crucial for DCD files + traj_custom = md.load_dcd(dcd_file_path, top=topology_file_path) + + print(f"Successfully loaded custom trajectory!") + print(f"Number of frames: {traj_custom.n_frames}") + print(f"Number of atoms: {traj_custom.n_atoms}") + # You can now perform analysis on traj_custom + # For example, calculate RMSD, distances, angles, etc. + +except FileNotFoundError: + print(f"Error: One or both files not found: {dcd_file_path}, {topology_file_path}") +except Exception as e: + print(f"An error occurred while loading your files: {e}") +print("-" * 30) + +# %% +from jamun.metrics._ramachandran import plot_ramachandran + +phi = md.compute_phi(traj_custom) +psi = md.compute_psi(traj_custom) + +import matplotlib.pyplot as plt +import numpy as np +fig = plt.figure() +ax = fig.add_subplot() +s = ax.scatter(phi[1], psi[1], cmap='hot', alpha=1.0) +ax.set_xlim((-np.pi, np.pi)) +ax.set_ylim((-np.pi, np.pi)) +c = fig.colorbar(s) + +print("hello world") \ No newline at end of file diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index c244e7f..0fc4aa6 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -41,6 +41,7 @@ def preprocess_topology(topology: md.Topology) -> Tuple[torch_geometric.data.Dat num_residues=residue_sequence_index.max().item() + 1, edge_index=bonds, pos=None, + hidden_state=None ) graph.residues = [x.residue.name for x in top.atoms] graph.atom_names = [x.name for x in top.atoms] @@ -83,7 +84,7 @@ def __init__( self.graph, self.top, self.top_withH = preprocess_topology(topology) self.graph.dataset_label = self.label() self.graph.loss_weight = torch.tensor([loss_weight], dtype=torch.float32) - + self.graph.hidden_state = None # self.save_topology_pdb() if verbose: @@ -179,6 +180,7 @@ def __init__( self.traj = self.traj.atom_slice(topology.select("protein and not type H")) self.graph.pos = torch.tensor(self.traj.xyz[0], dtype=torch.float32) + self.graph.hidden_state = torch.tensor(self.traj.xyz[-1], dtype=torch.float32) self.graph.loss_weight = torch.tensor([loss_weight], dtype=torch.float32) self.graph.dataset_label = self.label() @@ -201,6 +203,7 @@ def save_topology_pdb(self): def __getitem__(self, idx): graph = self.graph.clone() graph.pos = torch.tensor(self.traj.xyz[idx]) + graph.hidden_state = [torch.tensor(self.traj.xyz[idx-1])] if self.transform: graph = self.transform(graph) return graph diff --git a/src/jamun/model/arch/e3conv.py b/src/jamun/model/arch/e3conv.py index 3ca8c04..0c7fc8d 100644 --- a/src/jamun/model/arch/e3conv.py +++ b/src/jamun/model/arch/e3conv.py @@ -111,7 +111,7 @@ def forward( src, dst = edge_index edge_vec = pos[src] - pos[dst] edge_sh = self.sh(edge_vec) - + # print(f"Edge spherical harmonics: {type(edge_sh)}") bonded_edge_attr = self.embed_bondedness(bond_mask) radial_edge_attr = e3nn.math.soft_one_hot_linspace( edge_vec.norm(dim=1), diff --git a/src/jamun/utils/__init__.py b/src/jamun/utils/__init__.py index 340f6c6..9207558 100644 --- a/src/jamun/utils/__init__.py +++ b/src/jamun/utils/__init__.py @@ -1,4 +1,4 @@ -from .align import align_A_to_B, align_A_to_B_batched +from .align import align_A_to_B, align_A_to_B_batched, kabsch_algorithm from .average_squared_distance import compute_average_squared_distance, compute_average_squared_distance_from_datasets from .checkpoint import find_checkpoint, find_checkpoint_directory, get_run_path_for_wandb_run, get_wandb_run_config from .data_with_residue_info import DataWithResidueInformation diff --git a/src/jamun/utils/data_with_residue_info.py b/src/jamun/utils/data_with_residue_info.py index a180f49..05c335d 100644 --- a/src/jamun/utils/data_with_residue_info.py +++ b/src/jamun/utils/data_with_residue_info.py @@ -1,6 +1,6 @@ import torch import torch_geometric - +from typing import Any class DataWithResidueInformation(torch_geometric.data.Data): """Graph with residue-level information.""" @@ -13,7 +13,7 @@ class DataWithResidueInformation(torch_geometric.data.Data): residue_index: torch.Tensor # batched version of residue_sequence_index num_residues: int loss_weight: float - + hidden_state: Any def __inc__(self, key, value, *args, **kwargs): del value, args, kwargs if key in [ @@ -24,6 +24,7 @@ def __inc__(self, key, value, *args, **kwargs): "residue_sequence_index", "num_residues", "loss_weight", + "hidden_state" ]: return 0 if key in ["edge_index"]: From 76af63d7e192c58cf06e6f3e8001a129c447335a Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 13 Jun 2025 20:01:10 +0000 Subject: [PATCH 06/32] conditional generation models, testing, and hydra debugging script --- configs/experiment/sample_custom.yaml | 53 +-- ...le_uncapped_single_shape_conditioning.yaml | 53 +++ scratch/check_denoiser.py | 231 ++++++++++++ scratch/check_model_arch.py | 143 +++++++ scratch/conditioners.py | 33 ++ scratch/config.yaml | 5 +- scratch/denoiser_test.py | 28 +- scratch/fiddle.py | 173 --------- scratch/hydra_trials.py | 69 ++++ scratch/override_config.yaml | 32 ++ scratch/run_single_shape_AA_conditional.sh | 52 +++ scratch/sample.py | 177 +++++++++ scratch/sampling_prototype.py | 261 +++++++++++++ scratch/training_prototype.py | 356 ++++++++++++++++++ src/jamun/cmdline/train.py | 5 +- src/jamun/hydra_config/sample.yaml | 2 +- 16 files changed, 1465 insertions(+), 208 deletions(-) create mode 100644 configs/experiment/sample_uncapped_single_shape_conditioning.yaml create mode 100644 scratch/check_denoiser.py create mode 100644 scratch/check_model_arch.py create mode 100644 scratch/conditioners.py delete mode 100644 scratch/fiddle.py create mode 100644 scratch/hydra_trials.py create mode 100644 scratch/override_config.yaml create mode 100644 scratch/run_single_shape_AA_conditional.sh create mode 100644 scratch/sample.py create mode 100644 scratch/sampling_prototype.py create mode 100644 scratch/training_prototype.py diff --git a/configs/experiment/sample_custom.yaml b/configs/experiment/sample_custom.yaml index f1f23c9..fe72cea 100644 --- a/configs/experiment/sample_custom.yaml +++ b/configs/experiment/sample_custom.yaml @@ -1,24 +1,32 @@ # @package _global_ -num_sampling_steps_per_batch: 1000 -num_batches: 1 +# defaults: +# - override /callbacks: +# - sampler/save_trajectory.yaml +# - _self_ + +# callbacks: +# viz: +# sigma_list: ["${model.sigma_distribution.sigma}"] + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/timewarp/2AA-1-large/train/" + traj_pattern: "^(.*)-traj-arrays.npz" + pdb_file: AA-traj-state0.pdb + filter_codes: + - AA + subsample: 1 + max_datasets: 1 + +num_sampling_steps_per_batch: 100 +num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true -# wandb_train_run_path: prescient-design/jamun/zzt8s3rc - -# Old 4AA -# wandb_train_run_path: prescient-design/jamun/ibtxmwcr - -# New 4AA -wandb_train_run_path: prescient-design/jamun/6297yugb - -# Finetuned new 4AA -# wandb_train_run_path: prescient-design/jamun/x6rwt91k - -# init_pdb: /data/bucket/kleinhej/fast-folding/processed/chignolin/filtered.pdb -init_pdbs: ??? +# Add your wandb run path here +# wandb_train_run_path: your-wandb-run-path checkpoint_type: best_so_far sigma: 0.04 @@ -28,13 +36,12 @@ friction: 1.0 inverse_temperature: 1.0 score_fn_clip: 100.0 -init_datasets: - _target_: jamun.data.create_dataset_from_pdbs - pdbfiles: ${init_pdbs} - -finetune_on_init: - num_steps: ??? - batch_size: 16 +# Model configuration +model: + _target_: scratch.denoiser_test.Denoiser + conditioner: + _target_: scratch.conditioners.SelfConditioner + hidden_dim: 1 sampler: _target_: jamun.sampling.Sampler @@ -43,4 +50,4 @@ sampler: logger: wandb: group: sample_custom - notes: sampling on AA + notes: Custom sampling with denoiser and conditioner \ No newline at end of file diff --git a/configs/experiment/sample_uncapped_single_shape_conditioning.yaml b/configs/experiment/sample_uncapped_single_shape_conditioning.yaml new file mode 100644 index 0000000..346ca5c --- /dev/null +++ b/configs/experiment/sample_uncapped_single_shape_conditioning.yaml @@ -0,0 +1,53 @@ +# @package _global_ + +# defaults: +# - override /callbacks: +# - sampler/save_trajectory.yaml +# - _self_ + +# callbacks: +# viz: +# sigma_list: ["${model.sigma_distribution.sigma}"] + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/timewarp/2AA-1-large/train/" + traj_pattern: "^(.*)-traj-arrays.npz" + pdb_file: AA-traj-state0.pdb + filter_codes: + - AA + subsample: 1 + max_datasets: 1 + +num_sampling_steps_per_batch: 100 +num_batches: 10 +num_init_samples_per_dataset: 1 +repeat_init_samples: 1 +continue_chain: true + +# Add your wandb run path here +wandb_train_run_path: sule-shashank/jamun/3wctzcjp + +checkpoint_type: best_so_far +sigma: 0.04 +M: 1.0 +delta: 0.04 +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: 100.0 + +# Model configuration +model: + _target_: scratch.denoiser_test.Denoiser + conditioner: + _target_: scratch.conditioners.SelfConditioner + hidden_dim: 1 + +sampler: + _target_: jamun.sampling.Sampler + devices: 1 + +logger: + wandb: + group: sample_custom + notes: Custom sampling with denoiser and conditioner diff --git a/scratch/check_denoiser.py b/scratch/check_denoiser.py new file mode 100644 index 0000000..f546c1b --- /dev/null +++ b/scratch/check_denoiser.py @@ -0,0 +1,231 @@ +# %% +import functools +import logging +import os + +import dotenv +import tqdm +from typing import Union +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("jamun") + +import torch +torch.cuda.is_available() +torch.set_float32_matmul_precision("high") + +import e3nn +import e3tools.nn + +e3nn.set_optimization_defaults(jit_script_fx=False) + +import jamun +import jamun.data +import jamun.distributions +import jamun.model +import jamun.model.arch + +# %% +# dataset +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +# %% +datasets = { + "test": jamun.data.parse_datasets_from_directory( + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, + subsample=100, + max_datasets=1, + ) +} + +datamodule = jamun.data.MDtrajDataModule( + datasets=datasets, + batch_size=3, + num_workers=2, +) +datamodule.setup('test') +_, data_batch = next(enumerate(datamodule.test_dataloader())) + +# %% check paths +import sys +project_root = "/homefs/home/sules/jamun" # Or use os.path.abspath("..") if your notebook is in a subdir of jamun + +if project_root not in sys.path: + sys.path.insert(0, project_root) + py_logger.info(f"Added '{project_root}' to sys.path for module discovery.") +else: + py_logger.info(f"'{project_root}' is already in sys.path.") + +# %% get configs +from hydra import compose, initialize +import hydra +from omegaconf import OmegaConf +# Load the configuration file +with initialize(config_path="", job_name="jamun_test"): + cfg = compose( + config_name="config", + overrides=[ + "model.arch._target_=scratch.e3conv_test.E3Conv", # Override to use E3Conv from e3conv_test.py + "model._target_=scratch.denoiser_test.Denoiser", # Override to use Denoiser from denoiser_test.py + "+model.arch.N_structures=2", # Ensure N_structures is set, defaulting to 2 + ] + ) +# Log the configuration +py_logger.info("Loaded configuration:") +py_logger.info(OmegaConf.to_yaml(cfg)) + + +# %% Re-instantiate the model with the updated configuration +try: + py_logger.info("Attempting to re-instantiate model with updated arch...") + # Ensure average_squared_distance is still set correctly + if not hasattr(cfg.model, "average_squared_distance") or cfg.model.average_squared_distance is None: + from jamun.utils import compute_average_squared_distance_from_datasets # Ensure import + average_squared_distance = compute_average_squared_distance_from_datasets(datasets['test'], cfg.model.max_radius) + cfg.model.average_squared_distance = average_squared_distance + py_logger.info(f"Set cfg.model.average_squared_distance to {cfg.model.average_squared_distance}") + # Provide conditioner if needed + if not hasattr(cfg.model, "conditioner"): + OmegaConf.set_struct(cfg.model, False) # Allow modification + cfg.model.conditioner = OmegaConf.create({}) + cfg.model.conditioner._target_ = 'scratch.conditioners.PositionConditioner' # Use the PositionConditioner from scratch.conditioners + OmegaConf.set_struct(cfg.model, True) # Lock structure again + py_logger.info("Set cfg.model.conditioner to instantiate 'scratch.conditioners.PositionConditioner'") + model = hydra.utils.instantiate(cfg.model) + py_logger.info("Successfully re-instantiated model with E3Conv from e3conv_test.py:") + print(model) + # You can inspect model.arch to confirm it's an instance of E3Conv + py_logger.info(f"Instantiated model architecture type: {type(model.g)}") + +except Exception as e: + py_logger.error(f"Error during model re-instantiation: {e}") + import traceback + traceback.print_exc() + +# %% Tests for Denoiser.noise_and_denoise + +# Ensure 'model' (your Denoiser instance) and 'data_batch' are available from previous cells. +# If 'model' is not the correct Denoiser instance, re-instantiate it as needed. +# For example, if you were using the custom_denoiser_model: +# denoiser_to_test = custom_denoiser_model +# Or if you are using the one from the cfg re-instantiation: +denoiser_to_test = model + +# Make sure data_batch is on the same device as the model +if hasattr(denoiser_to_test, 'device'): + data_batch = data_batch.to(denoiser_to_test.device) +elif next(denoiser_to_test.parameters()).is_cuda: + data_batch = data_batch.to(next(denoiser_to_test.parameters()).device) + + +py_logger.info(f"Testing Denoiser instance of type: {type(denoiser_to_test)}") +py_logger.info(f"Data batch has {data_batch.num_graphs} graphs and {data_batch.num_nodes} nodes.") + +# %% Tests for Denoiser object (denoiser_to_test) + +import torch_geometric.data # For isinstance checks + +py_logger.info("Starting tests for Denoiser object...") +_, data_batch = next(enumerate(datamodule.test_dataloader())) +# Ensure data_batch has hidden_state for the tests, matching model's N_structures +if not hasattr(data_batch, 'hidden_state') or \ + not isinstance(data_batch.hidden_state, list) or \ + len(data_batch.hidden_state) != denoiser_to_test.g._orig_mod.N_structures - 1: # Use _orig_mod to access N_structures if g is compiled + + n_structures = denoiser_to_test.g._orig_mod.N_structures + py_logger.info(f"data_batch.hidden_state is missing or incorrect. Re-creating with {n_structures} structures.") + data_batch.hidden_state = [torch.randn_like(data_batch.pos) for _ in range(n_structures)] + data_batch.hidden_state = [hs.to(data_batch.pos.device) for hs in data_batch.hidden_state] +else: + py_logger.info(f"data_batch.hidden_state already exists with {len(data_batch.hidden_state)} structures.") + + +# %% Test 1: Denoiser.noise_and_denoise (align_noisy_input=False) + +try: + py_logger.info("Test 1: Denoiser.noise_and_denoise (align_noisy_input=False)") + original_x = data_batch.clone() + # sigma_test1 = torch.tensor(0.5, device=denoiser_to_test.device) + sigma = denoiser_to_test.sigma_distribution.sample()*1e-5 + xhat1, y_processed1 = denoiser_to_test.noise_and_denoise(original_x.clone(), sigma, \ + align_noisy_input=True) + + # assert isinstance(xhat1, torch_geometric.data.Batch), "xhat1 is not a PyG Batch object" + # assert isinstance(y_processed1, torch_geometric.data.Batch), "y_processed1 is not a PyG Batch object" + + assert xhat1.pos.shape == original_x.pos.shape, "xhat1.pos shape mismatch" + assert y_processed1.pos.shape == original_x.pos.shape, "y_processed1.pos shape mismatch" + + assert not torch.allclose(y_processed1.pos, original_x.pos), "y_processed1.pos should be different from original x.pos" + + assert xhat1.num_graphs == original_x.num_graphs, "xhat1 num_graphs mismatch" + assert y_processed1.num_graphs == original_x.num_graphs, "y_processed1 num_graphs mismatch" + assert xhat1.num_nodes == original_x.num_nodes, "xhat1 num_nodes mismatch" + assert y_processed1.num_nodes == original_x.num_nodes, "y_processed1 num_nodes mismatch" + assert torch.allclose(xhat1.batch, original_x.batch), "xhat1.batch mismatch" + assert torch.allclose(y_processed1.batch, original_x.batch), "y_processed1.batch mismatch" + + # Check hidden_state in y_processed1 (noisy input) + if hasattr(original_x, 'hidden_state') and original_x.hidden_state: + assert hasattr(y_processed1, 'hidden_state') and len(y_processed1.hidden_state) == len(original_x.hidden_state), "y_processed1.hidden_state length mismatch" + for i in range(len(original_x.hidden_state)): + assert not torch.allclose(y_processed1.hidden_state[i], original_x.hidden_state[i]), f"y_processed1.hidden_state[{i}] should be different" + + # xhat inherits attributes from the input to xhat_normalized, which is the noisy graph (y_processed1) + # So, xhat1 should also have hidden_state if y_processed1 does. + if hasattr(y_processed1, 'hidden_state') and y_processed1.hidden_state: + assert hasattr(xhat1, 'hidden_state') and len(xhat1.hidden_state) == len(y_processed1.hidden_state), "xhat1.hidden_state length mismatch with y_processed1" + + py_logger.info("Test 1 PASSED.") +except Exception as e: + py_logger.error(f"Test 1 FAILED: {e}") + import traceback + traceback.print_exc() + +# %% Test 3: Denoiser.training_step +try: + py_logger.info("Test 3: Denoiser.training_step") + # Get a fresh batch for training_step to avoid issues with modified data_batch from other tests + _, train_batch = next(enumerate(datamodule.test_dataloader())) # Using test_dataloader for convenience + + # Ensure train_batch has hidden_state + if not hasattr(train_batch, 'hidden_state') or \ + not isinstance(train_batch.hidden_state, list) or \ + len(train_batch.hidden_state) != denoiser_to_test.g._orig_mod.N_structures-1: + n_structures = denoiser_to_test.g._orig_mod.N_structures + train_batch.hidden_state = [torch.randn_like(train_batch.pos) for _ in range(n_structures)] + else: + py_logger.info(f"train_batch.hidden_state already exists with {len(train_batch.hidden_state)} structures.") + train_batch.hidden_state = [hs.to(denoiser_to_test.device) for hs in train_batch.hidden_state] + train_batch = train_batch.to(denoiser_to_test.device) + + + # Manually set align_noisy_input_during_training if not set (it's a param of Denoiser) + if not hasattr(denoiser_to_test, 'align_noisy_input_during_training'): + py_logger.warning("Denoiser missing 'align_noisy_input_during_training', defaulting to False for this test.") + denoiser_to_test.align_noisy_input_during_training = False # Or True, as needed + + logs_dict = denoiser_to_test.training_step(train_batch, 0) + + assert isinstance(logs_dict, dict), "Logs is not a dictionary" + expected_keys = ["mse", "rmsd", "scaled_rmsd", "loss"] + for key in expected_keys: + assert key in logs_dict, f"Key '{key}' missing in logs" + assert isinstance(logs_dict[key], torch.Tensor), f"Log value for '{key}' is not a tensor" + if key == 'loss': + assert logs_dict[key].ndim == 0, f"logs_dict['loss'] is not scalar, shape: {logs_dict[key].shape}" + else: # mse, rmsd, scaled_rmsd are averaged in training_step's aux_mean + assert logs_dict[key].ndim == 0, f"logs_dict['{key}'] is not scalar, shape: {logs_dict[key].shape}" + + + py_logger.info(f"Test 3 PASSED. Loss: {logs_dict['mse'].item()}") +except Exception as e: + py_logger.error(f"Test 3 FAILED: {e}") + import traceback + traceback.print_exc() +# %% diff --git a/scratch/check_model_arch.py b/scratch/check_model_arch.py new file mode 100644 index 0000000..7ec7426 --- /dev/null +++ b/scratch/check_model_arch.py @@ -0,0 +1,143 @@ +# %% +import functools +import logging +import os + +import dotenv +import tqdm +from typing import Union +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("jamun") + +import torch +torch.cuda.is_available() +torch.set_float32_matmul_precision("high") + +import e3nn +import e3tools.nn + +e3nn.set_optimization_defaults(jit_script_fx=False) + +import jamun +import jamun.data +import jamun.distributions +import jamun.model +import jamun.model.arch + +# %% +# dataset +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +# %% +datasets = { + "test": jamun.data.parse_datasets_from_directory( + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, + subsample=100, + max_datasets=1, + ) +} + +datamodule = jamun.data.MDtrajDataModule( + datasets=datasets, + batch_size=3, + num_workers=2, +) +datamodule.setup('test') +_, data_batch = next(enumerate(datamodule.test_dataloader())) + +# %% test the new e3conv_test class +from e3conv_test import E3Conv +import torch_geometric +from e3tools import radius_graph + +trial_model = E3Conv( + irreps_out="1x1e", + irreps_hidden="120x0e + 32x1e", + irreps_sh="1x0e + 1x1e", + n_layers=1, + edge_attr_dim=8, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=32, + residue_index_embedding_dim=8, + use_residue_information=False, + use_residue_sequence_index=False, + hidden_layer_factory=functools.partial( + e3tools.nn.ConvBlock, + conv=e3tools.nn.Conv, + ), + output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), + N_structures=2 + ) + +# %% postprocess data for plugging into model +def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_geometric.data.Batch: + radial_edge_index = radius_graph(y.pos, cutoff, batch=y["batch"]) + bonded_edge_index = y.edge_index + edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) + bond_mask = torch.cat( + ( + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), + torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), + ), + dim=0, + ) + y.edge_index = edge_index + y.bond_mask = bond_mask + return y + +# add bond mask--do this only once! +bond_mask_exists = hasattr(data_batch, 'bond_mask') and data_batch.bond_mask is not None +if not bond_mask_exists: + py_logger.info("Adding bond mask to data_batch...") + # Ensure data_batch is a torch_geometric.data.Batch + if not isinstance(data_batch, torch_geometric.data.Batch): + raise TypeError(f"Expected data_batch to be a torch_geometric.data.Batch, got {type(data_batch)}") + + # Add bond mask + data_batch = add_bond_mask(data_batch) +else: + py_logger.info("Bond mask already exists in data_batch, skipping addition.") + # If bond mask already exists, we can still use it + # but we should ensure it's in the correct format + if not isinstance(data_batch.bond_mask, torch.Tensor): + raise TypeError(f"Expected data_batch.bond_mask to be a torch.Tensor, got {type(data_batch.bond_mask)}") + if data_batch.bond_mask.dtype != torch.long: + raise ValueError(f"Expected data_batch.bond_mask to be of dtype torch.long, got {data_batch.bond_mask.dtype}") + + # Ensure edge_index is set correctly + if not hasattr(data_batch, 'edge_index') or data_batch.edge_index is None: + raise ValueError("data_batch.edge_index is not set. Please ensure it is initialized before adding bond mask.") + + # If everything is fine, we can proceed with the existing bond mask +py_logger.info(f"data_batch has {data_batch.num_graphs} graphs and {data_batch.num_nodes} nodes.") +# Ensure data_batch is on the same device as the model +if hasattr(trial_model, 'device'): + data_batch = data_batch.to(trial_model.device) +elif next(trial_model.parameters()).is_cuda: + data_batch = data_batch.to(next(trial_model.parameters()).device) + +# %% Test the E3Conv model with the data_batch +py_logger.info("Testing E3Conv model with data_batch...") +# Ensure data_batch is on the same device as the model +y = data_batch +if hasattr(trial_model, 'device'): + y = y.to(trial_model.device) +elif next(trial_model.parameters()).is_cuda: + y = y.to(next(trial_model.parameters()).device) +# Run a forward pass through the model +try: + py_logger.info("Running forward pass through E3Conv model...") + output = trial_model(torch.cat([y.pos, *y.hidden_state], dim=-1), y, torch.Tensor([1e-5]), 100.0) + py_logger.info(f"Output shape: {output.shape}") +except Exception as e: + py_logger.error(f"Error during forward pass: {e}") + import traceback + traceback.print_exc() + diff --git a/scratch/conditioners.py b/scratch/conditioners.py new file mode 100644 index 0000000..1afeb35 --- /dev/null +++ b/scratch/conditioners.py @@ -0,0 +1,33 @@ +import logging +from typing import Callable, Dict, Optional, Tuple, Union + +import lightning.pytorch as pl +import numpy as np +import torch +import torch_geometric +from e3tools import radius_graph, scatter + +from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils.align import kabsch_algorithm + + + +class PositionConditioner(pl.LightningModule): + def __init__(self, **kwargs): + super().__init__() + + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + conditioned_structures = [] + for positions in y.hidden_state: + aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) + conditioned_structures.append(aligned_positions) + return conditioned_structures + +class SelfConditioner(pl.LightningModule): + def __init__(self, **kwargs): + super().__init__() + + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + conditioned_structures = [] + conditioned_structures.append(y.pos) + return conditioned_structures \ No newline at end of file diff --git a/scratch/config.yaml b/scratch/config.yaml index 5318d5f..f1446de 100644 --- a/scratch/config.yaml +++ b/scratch/config.yaml @@ -39,7 +39,7 @@ data: use_residue_information: true model: arch: - _target_: jamun.model.arch.E3Conv + _target_: scratch.e3conv_test.E3Conv _partial_: true irreps_out: 1x1e irreps_hidden: 120x0e + 32x1e @@ -102,7 +102,8 @@ logger: offline: false log_model: all group: train_test - notes: alanine dipeptide + notes: alanine dipeptide, without conditioning. + save_dir: callbacks: timing: _target_: jamun.callbacks.Timing diff --git a/scratch/denoiser_test.py b/scratch/denoiser_test.py index 8c42642..ab93938 100644 --- a/scratch/denoiser_test.py +++ b/scratch/denoiser_test.py @@ -7,7 +7,7 @@ import torch_geometric from e3tools import radius_graph, scatter -from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing, +from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm @@ -33,6 +33,7 @@ def __init__( lr_scheduler_config: Optional[Dict] = None, use_torch_compile: bool = True, torch_compile_kwargs: Optional[Dict] = None, + conditioner: Callable[..., list[torch.Tensor]] = None ): super().__init__() self.save_hyperparameters(logger=False) @@ -98,14 +99,27 @@ def __init__( raise ValueError("sigma_data can only be used when normalization_type is 'EDM'") self.bond_loss_coefficient = bond_loss_coefficient - - def conditioner(self, y: torch_geometric.data.Batch) -> torch.Tensor: + self.conditioning_module = conditioner + if self.conditioning_module is not None and not callable(self.conditioning_module): + raise ValueError("Conditioner must be a callable or None") + py_logger.info(f"Conditioner: {self.conditioning_module}") + + def conditioner_default(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures = [] - for positions in y.hidden_state: - aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) - conditioned_structures.append(aligned_positions) + # for positions in y.hidden_state: + # aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) + # conditioned_structures.append(aligned_positions) + conditioned_structures.append(y.pos) return conditioned_structures + def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + if self.conditioning_module is None: + return self.conditioner_default(y) + elif callable(self.conditioning_module): + return self.conditioning_module(y) + else: + raise ValueError("Conditioner must be a callable or None") + def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.pos.ndim) @@ -242,7 +256,7 @@ def xhat_normalized( conditioned_structures = self.conditioner(y_scaled) with torch.cuda.nvtx.range("g"): - g_pred = self.g(torch.cat([y_scaled.pos, *conditioned_structures]), topology=y_scaled, \ + g_pred = self.g(torch.cat([y_scaled.pos, *conditioned_structures], dim=-1), topology=y_scaled, \ c_noise=c_noise, effective_radial_cutoff=radial_cutoff) xhat.pos = c_skip * y.pos + c_out * g_pred diff --git a/scratch/fiddle.py b/scratch/fiddle.py deleted file mode 100644 index 85fc6ad..0000000 --- a/scratch/fiddle.py +++ /dev/null @@ -1,173 +0,0 @@ -# %% -import functools -import logging -import os - -import dotenv -import tqdm -from typing import Union -logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) -py_logger = logging.getLogger("jamun") - -import torch -torch.cuda.is_available() -torch.set_float32_matmul_precision("high") - -import e3nn -import e3tools.nn - -e3nn.set_optimization_defaults(jit_script_fx=False) - -import jamun -import jamun.data -import jamun.distributions -import jamun.model -import jamun.model.arch - -# %% -# dataset -dotenv.load_dotenv("../.env", verbose=True) -JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") -JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") - -# %% -datasets = { - "test": jamun.data.parse_datasets_from_directory( - root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", - traj_pattern="^(.*)-traj-arrays.npz", - pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], - as_iterable=False, - subsample=100, - max_datasets=1, - ) -} - -datamodule = jamun.data.MDtrajDataModule( - datasets=datasets, - batch_size=3, - num_workers=2, -) -datamodule.setup('test') -_, data_batch = next(enumerate(datamodule.test_dataloader())) - -# %% -import hydra -from hydra import initialize, compose -from omegaconf import DictConfig, OmegaConf - -with initialize(config_path=f"../outputs/train/dev/runs/2025-06-04_20-07-32/.hydra", job_name="my_script_job", version_base=None): - # 2. Compose the configuration - # `config_name` refers to the base YAML file (e.g., "config.yaml" -> "config") - # You can also pass overrides programmatically, e.g., - # cfg = compose(config_name="config", overrides=["model.irreps_out='2x0e'"]) - cfg: DictConfig = compose(config_name="config") - - print("Full configuration loaded by Hydra:") - print(OmegaConf.to_yaml(cfg)) - print("-" * 50) - print(torch.cuda.is_available()) - # 3. Instantiate the model - try: - print("Attempting to instantiate model...") - # hydra.utils.instantiate is still the way to go - from jamun.utils import compute_average_squared_distance_from_datasets - average_squared_distance = compute_average_squared_distance_from_datasets(datasets['test'], cfg.model.max_radius) - cfg.model.average_squared_distance = average_squared_distance - model: jamun.model.Denoiser = hydra.utils.instantiate(cfg.model) - # print("\nSuccessfully instantiated model:") - # print(model) - # print("-" * 50) - - # # Example: Create some dummy input data - # batch_s = 2 - # dummy_input = torch.randn(batch_s, model.irreps_in.dim) - # print(f"Created dummy input with shape: {dummy_input.shape}") - - # # Perform a forward pass - # output = model(dummy_input) - # print(f"Model output shape: {output.shape}") - # assert output.shape == (batch_s, model.irreps_out.dim) - # print("Dummy forward pass successful!") - - except Exception as e: - print(f"Error during model instantiation or processing: {e}") - import traceback - traceback.print_exc() - - -# %% play w/ spherical harmonics -from e3nn import o3 -sh = o3.SphericalHarmonics(irreps_out="1x0e + 1x1e", normalize=True, normalization="component") -irreps_sh = e3nn.o3.Irreps("1x0e + 1x1e") -# %% -full_pos = torch.cat([molecule_pos, molecule_pos_extra], dim=-1) -blocks = torch.split(full_pos, 3, dim=-1) -src, dst = molecule.edge_index -edge_sh = [] -for block in blocks: - edge_vec = block[src] - block[dst] - edge_sh.append(sh(edge_vec)) -edge_sh = torch.cat(edge_sh, dim=-1) - - -# %% test the new e3conv_test class -from e3conv_test import E3Conv -import torch_geometric -from e3tools import radius_graph - -trial_model = E3Conv( - irreps_out="1x1e", - irreps_hidden="120x0e + 32x1e", - irreps_sh="1x0e + 1x1e", - n_layers=1, - edge_attr_dim=8, - atom_type_embedding_dim=8, - atom_code_embedding_dim=8, - residue_code_embedding_dim=32, - residue_index_embedding_dim=8, - use_residue_information=False, - use_residue_sequence_index=False, - hidden_layer_factory=functools.partial( - e3tools.nn.ConvBlock, - conv=e3tools.nn.Conv, - ), - output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), - N_structures=2 - ) - -# %% postprocess data for plugging into model -def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_geometric.data.Batch: - radial_edge_index = radius_graph(y.pos, cutoff, batch=y["batch"]) - bonded_edge_index = y.edge_index - edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) - bond_mask = torch.cat( - ( - torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), - torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), - ), - dim=0, - ) - y.edge_index = edge_index - y.bond_mask = bond_mask - return y - -# add bond mask--do this only once! -y = add_bond_mask(data_batch) - -# %% -from jamun.utils.align import kabsch_algorithm -from jamun.utils import mean_center -from e3tools import scatter - -# y = mean_center(y) -def conditioner(y: torch_geometric.data.Batch) -> torch.Tensor: - conditioned_structures = [] - for positions in y.hidden_state: - aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) - conditioned_structures.append(aligned_positions) - return conditioned_structures - -print('hello world') - -# %% diff --git a/scratch/hydra_trials.py b/scratch/hydra_trials.py new file mode 100644 index 0000000..106ec3f --- /dev/null +++ b/scratch/hydra_trials.py @@ -0,0 +1,69 @@ +# %% imports +import hydra +from omegaconf import OmegaConf +import os +import sys +import dotenv +import logging + +# --- Basic Setup --- +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("jamun_sampling_script") + +# Add project root to path for custom modules +project_root = "/homefs/home/sules/jamun" +if project_root not in sys.path: + sys.path.insert(0, project_root) + py_logger.info(f"Added '{project_root}' to sys.path for module discovery.") +else: + py_logger.info(f"'{project_root}' is already in sys.path.") + +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +def print_config_sections(cfg): + """Print different sections of the configuration.""" + print("\nFull Configuration:") + print(OmegaConf.to_yaml(cfg)) + + if hasattr(cfg, 'model'): + print("\nModel Configuration:") + print(OmegaConf.to_yaml(cfg.model)) + + if hasattr(cfg, 'init_datasets'): + print("\nDataset Configuration:") + print(OmegaConf.to_yaml(cfg.init_datasets)) + + if hasattr(cfg, 'sampler'): + print("\nSampler Configuration:") + print(OmegaConf.to_yaml(cfg.sampler)) + +def run(cfg): + """Main function to run the config loading and printing.""" + # Print the loaded configuration + print_config_sections(cfg) + + # Print specific config values + if hasattr(cfg, 'model'): + print("\nModel target:", cfg.model._target_) + if hasattr(cfg, 'sampler'): + print("Sampler target:", cfg.sampler._target_) + +def main(): + # Initialize Hydra + with hydra.initialize(config_path="../src/jamun/hydra_config"): + # Compose the base configuration + base_cfg = hydra.compose(config_name="sample") + + # Compose the experiment configuration + experiment_cfg = hydra.compose(config_name="sample", overrides=["experiment=sample_uncapped_single_shape_conditioning"]) + + # Merge configurations (experiment overrides base) + cfg = OmegaConf.merge(base_cfg, experiment_cfg) + + # Run with the merged configuration + run(cfg) + +if __name__ == "__main__": + main() diff --git a/scratch/override_config.yaml b/scratch/override_config.yaml new file mode 100644 index 0000000..826332f --- /dev/null +++ b/scratch/override_config.yaml @@ -0,0 +1,32 @@ +_target_: scratch.denoiser_test.Denoiser +arch: + _target_: scratch.e3conv_test.E3Conv + _partial_: true + irreps_out: 1x1e + irreps_hidden: 120x0e + 32x1e + irreps_sh: 1x0e + 1x1e + n_layers: 2 + edge_attr_dim: 64 + atom_type_embedding_dim: 8 + atom_code_embedding_dim: 8 + residue_code_embedding_dim: 32 + residue_index_embedding_dim: 8 + use_residue_information: ${data.use_residue_information} + use_residue_sequence_index: false + num_atom_types: 20 + max_sequence_length: 10 + num_atom_codes: 10 + num_residue_types: 25 + hidden_layer_factory: + _target_: e3tools.nn.ConvBlock + _partial_: true + conv: + _target_: e3tools.nn.Conv + _partial_: true + output_head_factory: + _target_: e3tools.nn.EquivariantMLP + _partial_: true + irreps_hidden_list: + - ${model.arch.irreps_hidden} +conditioner: + _target_: scratch.conditioners.SelfConditioner \ No newline at end of file diff --git a/scratch/run_single_shape_AA_conditional.sh b/scratch/run_single_shape_AA_conditional.sh new file mode 100644 index 0000000..187361c --- /dev/null +++ b/scratch/run_single_shape_AA_conditional.sh @@ -0,0 +1,52 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu2 +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node 1 # Adjusted to 1 as typically one training script runs per job +#SBATCH --gpus-per-node 1 # Assuming your script uses 1 GPU. Adjust if it uses more. +#SBATCH --cpus-per-task 8 # Number of CPUs for your task +#SBATCH --time 08:00:00 # 7 days runtime limit +#SBATCH --mem-per-cpu=32G # Memory per CPU +#SBATCH --job-name=train_prototype # Descriptive job name +#SBATCH --output=slurm_logs/train_prototype_%A_%a.out # Standard output file +#SBATCH --error=slurm_logs/train_prototype_%A_%a.err # Standard error file + +# --- Environment Setup --- +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_JOB_NODELIST = ${SLURM_JOB_NODELIST}" +echo "hostname = $(hostname)" +echo "Running on partition: ${SLURM_JOB_PARTITION}" +echo "Allocated GPUs: ${CUDA_VISIBLE_DEVICES:-"Not set"}" # SLURM usually sets CUDA_VISIBLE_DEVICES + +# Activate Conda environment +eval "$(conda shell.bash hook)" +conda activate jamun +echo "Conda environment 'jamun' activated." +echo "Python version: $(python --version)" +echo "PyTorch version: $(python -c 'import torch; print(torch.__version__)')" +echo "CUDA available: $(python -c 'import torch; print(torch.cuda.is_available())')" + +# --- Create Log Directory --- +# Create a directory for SLURM logs if it doesn't exist +# This should be relative to where you submit the job from, or an absolute path. +# Assuming you submit from /homefs/home/sules/jamun/ +LOG_DIR_BASE="/homefs/home/sules/jamun/slurm_logs" +mkdir -p "${LOG_DIR_BASE}" +# The %A_%a in sbatch output/error directives will be replaced by JobID and TaskID + +# --- Application Execution --- +# Navigate to the directory containing your script, if necessary +# Assuming training_prototype.py is in /homefs/home/sules/jamun/scratch/ +SCRIPT_DIR="/homefs/home/sules/jamun/scratch" +PYTHON_SCRIPT="training_prototype.py" + +echo "Changing directory to ${SCRIPT_DIR}" +cd "${SCRIPT_DIR}" || { echo "Failed to cd to ${SCRIPT_DIR}"; exit 1; } + +echo "Starting Python script: ${PYTHON_SCRIPT}" +# Run the Python script +# Add any necessary command-line arguments for your script here +python "${PYTHON_SCRIPT}" + +echo "Python script finished." +echo "Job finished at: $(date)" diff --git a/scratch/sample.py b/scratch/sample.py new file mode 100644 index 0000000..f526000 --- /dev/null +++ b/scratch/sample.py @@ -0,0 +1,177 @@ +import os +import sys +import traceback +from typing import Sequence +import pdb # For debugging + +import dotenv +import e3nn +import hydra +import lightning.pytorch as pl +import torch +import torch_geometric +from lightning.pytorch.utilities import rank_zero_only +from omegaconf import OmegaConf + +e3nn.set_optimization_defaults(jit_script_fx=False) + +import jamun +from jamun.data import MDtrajDataModule, MDtrajDataset +from jamun.hydra import instantiate_dict_cfg +from jamun.hydra.utils import format_resolver +from jamun.utils import dist_log, find_checkpoint + +# Add project root to path for custom modules +project_root = "/homefs/home/sules/jamun" +if project_root not in sys.path: + sys.path.insert(0, project_root) + +dotenv.load_dotenv(".env", verbose=True) +OmegaConf.register_new_resolver("format", format_resolver) + + +def get_initial_graphs( + datasets: Sequence[MDtrajDataset], num_init_samples_per_dataset: int, repeat: int = 1 +) -> torch_geometric.data.Batch: + """Get initial graphs for sampling.""" + init_graphs = [] + for dataset in datasets: + random_indices = torch.randperm(len(dataset))[:num_init_samples_per_dataset] + for index in random_indices: + init_graph = dataset[index] + # Ensure graph has hidden state + if not hasattr(init_graph, 'hidden_state'): + init_graph.hidden_state = torch.zeros(init_graph.num_nodes, 1) + for _ in range(repeat): + init_graphs.append(init_graph) + return torch_geometric.data.Batch.from_data_list(init_graphs) + + +def run(cfg): + log_cfg = OmegaConf.to_container(cfg, throw_on_missing=True, resolve=True) + + # Breakpoint 0: After config is loaded + pdb.set_trace() # Inspect full configuration + + dist_log(f"{OmegaConf.to_yaml(log_cfg)}") + dist_log(f"{os.getcwd()=}") + dist_log(f"{torch.__config__.parallel_info()}") + if hasattr(os, "sched_getaffinity"): + dist_log(f"{os.sched_getaffinity(0)=}") + + if matmul_prec := cfg.get("float32_matmul_precision"): + dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") + torch.set_float32_matmul_precision(matmul_prec) + + loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) + wandb_logger = None + for logger in loggers: + if isinstance(logger, pl.loggers.WandbLogger): + wandb_logger = logger + + if rank_zero_only.rank == 0 and wandb_logger: + dist_log(f"{wandb_logger.experiment.name=}") + wandb_logger.experiment.config.update({"cfg": log_cfg, "version": jamun.__version__, "cwd": os.getcwd()}) + + # Load the checkpoint + checkpoint_path = find_checkpoint( + wandb_train_run_path=cfg.get("wandb_train_run_path"), + checkpoint_dir=cfg.get("checkpoint_dir"), + checkpoint_type=cfg.get("checkpoint_type"), + ) + dist_log(f"Found checkpoint at: {checkpoint_path}") + + # Load checkpoint and create model + checkpoint = torch.load(checkpoint_path, map_location='cpu') + dist_log("Loaded checkpoint successfully") + + # Create model with conditioner + cfg.model._target_ = "scratch.denoiser_test.Denoiser" + if not hasattr(cfg.model, 'conditioner'): + cfg.model.conditioner = OmegaConf.create({ + "_target_": "scratch.conditioners.SelfConditioner", + "hidden_dim": 1 # Adjust based on your needs + }) + + model = hydra.utils.instantiate(cfg.model) + dist_log("Created model instance") + + # Load state dict into model + model.load_state_dict(checkpoint['state_dict']) + dist_log("Loaded state dict into model") + + # Breakpoint 1: After model instantiation and loading + pdb.set_trace() # Check model architecture and weights + + init_datasets = hydra.utils.instantiate(cfg.init_datasets) + init_graphs = get_initial_graphs( + init_datasets, + num_init_samples_per_dataset=cfg.num_init_samples_per_dataset, + repeat=cfg.repeat_init_samples, + ) + + # Breakpoint 2: After dataset loading + pdb.set_trace() # Check dataset and initial graphs + + callbacks = instantiate_dict_cfg(cfg.get("callbacks"), verbose=(rank_zero_only.rank == 0)) + sampler = hydra.utils.instantiate(cfg.sampler, callbacks=callbacks, loggers=loggers) + batch_sampler = hydra.utils.instantiate(cfg.batch_sampler) + + if seed := cfg.get("seed"): + # During sampling, we want ranks to generate different chains + pl.seed_everything(seed + sampler.fabric.global_rank) + + # Run test-time adaptation, if specified + if finetuning_cfg := cfg.get("finetune_on_init"): + num_finetuning_steps = finetuning_cfg.get("num_steps") + dist_log(f"Finetuning for {num_finetuning_steps} steps.") + + # Check that model parameters changed + param_sum = sum(p.sum() for p in model.parameters()) + + # Train the model for a fixed number of steps + trainer = pl.Trainer( + logger=loggers, + max_steps=num_finetuning_steps, + min_steps=num_finetuning_steps, + log_every_n_steps=1, + check_val_every_n_epoch=1, + ) + + # Breakpoint 3: Before finetuning + pdb.set_trace() # Check model state before finetuning + + trainer.fit( + model, + datamodule=MDtrajDataModule( + datasets={"train": init_datasets, "val": init_datasets}, + batch_size=finetuning_cfg.batch_size, + ), + ) + + # Check that model parameters changed + new_param_sum = sum(p.sum() for p in model.parameters()) + dist_log(f"Model parameters changed: {param_sum} -> {new_param_sum}") + + # Breakpoint 4: Before sampling + pdb.set_trace() # Check final model state before sampling + + sampler.sample( + model=model, + batch_sampler=batch_sampler, + init_graphs=init_graphs, + num_batches=cfg.num_batches, + continue_chain=cfg.continue_chain, + ) + + if wandb_logger: + wandb_logger.finalize(status="finished") + + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample") +def main(cfg): + run(cfg) + + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/sampling_prototype.py b/scratch/sampling_prototype.py new file mode 100644 index 0000000..22e9a4a --- /dev/null +++ b/scratch/sampling_prototype.py @@ -0,0 +1,261 @@ + +# %% Imports and Basic Setup +import functools +import logging +import os +import sys +from typing import Union, Sequence + +import dotenv +import tqdm +import torch +import e3nn +import e3tools.nn +import hydra +from hydra import compose, initialize +from omegaconf import OmegaConf +import lightning.pytorch as pl +import torch_geometric.data + +import jamun +import jamun.data +import jamun.distributions +import jamun.model +import jamun.model.arch +import jamun.sampling +from jamun.utils import compute_average_squared_distance_from_datasets, find_checkpoint +from jamun.hydra import instantiate_dict_cfg +from jamun.data import MDtrajDataModule, MDtrajDataset + +# --- Basic Setup --- +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("jamun_sampling_script") + +torch.cuda.is_available() +torch.set_float32_matmul_precision("high") +e3nn.set_optimization_defaults(jit_script_fx=False) + +# %% Environment and Paths +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + py_logger.info(f"Added '{project_root}' to sys.path for module discovery.") +else: + py_logger.info(f"'{project_root}' is already in sys.path.") + +# %% Load Configuration from Specific File +py_logger.info("Loading configuration from specific training run...") +config_file_path = "../outputs/train/dev/runs/2025-06-11_20-16-04/final_resolved_script_config.yaml" + +# Load the configuration directly from the YAML file +cfg = OmegaConf.load(config_file_path) +py_logger.info("Loaded configuration from training run:") +py_logger.info(f"Config file: {config_file_path}") + +# Modify config for sampling purposes +# Override model target to use our local architectures +cfg.model._target_ = "scratch.denoiser_test.Denoiser" +cfg.model.arch._target_ = "scratch.e3conv_test.E3Conv" + +# Add sampling-specific configuration +if not hasattr(cfg, 'sampling'): + cfg.sampling = OmegaConf.create({}) + +# Set sampling parameters (adapt from sample.yaml structure) +cfg.sampling.repeat_init_samples = 1 +cfg.sampling.num_batches = 10 +cfg.sampling.continue_chain = True +cfg.sampling.num_init_samples_per_dataset = 5 +cfg.sampling.seed = 42 + +# Add sampler configuration +cfg.sampling.sampler = OmegaConf.create({ + "_target_": "jamun.sampling.Sampler", + "precision": "32-true" +}) + +# Add batch sampler configuration +cfg.sampling.batch_sampler = OmegaConf.create({ + "_target_": "jamun.sampling.SingleMeasurementSampler", + "num_steps": 100, + "sigma_min": 0.001, + "sigma_max": 0.1 +}) + +py_logger.info("Modified configuration for sampling:") +py_logger.info(OmegaConf.to_yaml(cfg)) + +# %% Helper Functions +def get_initial_graphs( + datasets: Sequence[MDtrajDataset], num_init_samples_per_dataset: int, repeat: int = 1 +) -> torch_geometric.data.Batch: + """Get initial graphs for sampling.""" + init_graphs = [] + for dataset in datasets: + random_indices = torch.randperm(len(dataset))[:num_init_samples_per_dataset] + for index in random_indices: + init_graph = dataset[index] + for _ in range(repeat): + init_graphs.append(init_graph) + return torch_geometric.data.Batch.from_data_list(init_graphs) + +# %% Load Model +py_logger.info("Loading model from checkpoint...") + +cfg.model._target_ = 'scratch.denoiser_test.Denoiser' +model = hydra.utils.instantiate(cfg.model) + +# Device Setup +if torch.cuda.is_available(): + device = torch.device("cuda") + py_logger.info("CUDA is available. Using GPU.") +else: + device = torch.device("cpu") + py_logger.info("CUDA not available. Using CPU.") + +model = model.to(device) +py_logger.info(f"Model moved to device: {device}") + +# %% Load from ckpoint +sys.path.append("../") +checkpoint_dir = "../outputs/train/dev/runs/2025-06-11_20-16-04/wandb/latest-run/checkpoints" +try: + checkpoint_path = find_checkpoint( + checkpoint_dir=checkpoint_dir, + checkpoint_type="last" # or "best" + ) + py_logger.info(f"Found checkpoint: {checkpoint_path}") +except Exception as e: + py_logger.error(f"Could not find checkpoint in {checkpoint_dir}: {e}") + # Try to find checkpoint in the checkpoints subdirectory + checkpoint_subdir = os.path.join(checkpoint_dir, "run-*/checkpoints") + import glob + checkpoint_files = glob.glob(os.path.join(checkpoint_subdir, "*.ckpt")) + if checkpoint_files: + checkpoint_path = checkpoint_files[-1] # Use the last one + py_logger.info(f"Using checkpoint: {checkpoint_path}") + else: + py_logger.error("No checkpoint files found!") + # sys.exit(1) + +# %% Load from checkpoint +if checkpoint_path: + + checkpoint = torch.load(checkpoint_path, map_location=model.device, weights_only=False) + model.load_state_dict(checkpoint['state_dict']) + py_logger.info("Successfully loaded model from checkpoint.") + py_logger.info(f"Model architecture type: {type(model.g)}") +else: + py_logger.error("No checkpoint files found!") + # sys.exit(1) + +# %% checkpoint path update +cfg.model.checkpoint_path = checkpoint_path +py_logger.info(f"Updated checkpoint path: {cfg.model.checkpoint_path}") + +# %% Setup Initial Datasets for Sampling +py_logger.info("Setting up initial datasets for sampling...") + +# Use the same data configuration as the training +init_datasets = jamun.data.parse_datasets_from_directory( + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, + subsample=20, # Smaller subset for sampling + max_datasets=1, +) + +py_logger.info(f"Loaded {len(init_datasets)} samples for initial configurations") + +# %% Generate Initial Graphs +py_logger.info("Generating initial graphs for sampling...") +init_graphs = get_initial_graphs( + init_datasets, + num_init_samples_per_dataset=cfg.sampling.num_init_samples_per_dataset, + repeat=cfg.sampling.repeat_init_samples, +) +py_logger.info(f"Generated {len(init_graphs)} initial graphs") + +# %% Setup Sampling Components +py_logger.info("Setting up sampling components...") + +# Set random seed +if cfg.sampling.seed: + pl.seed_everything(cfg.sampling.seed) + py_logger.info(f"Set random seed to {cfg.sampling.seed}") + +# Setup loggers for sampling +loggers_list = [] +if cfg.get("logger"): + # Create a sampling-specific logger config + sampling_logger_cfg = OmegaConf.create({ + "wandb": { + "_target_": "lightning.pytorch.loggers.WandbLogger", + "project": "jamun-sampling", + "entity": None, + "offline": False, + "group": "sampling_test", + "notes": "Sampling from trained model", + "save_dir": "./outputs/sample/" + } + }) + loggers_list = instantiate_dict_cfg(sampling_logger_cfg) + +# Instantiate sampler +try: + sampler = hydra.utils.instantiate(cfg.sampling.sampler, callbacks=[], loggers=loggers_list) + py_logger.info("Successfully instantiated sampler") +except Exception as e: + py_logger.error(f"Error instantiating sampler: {e}") + # Fallback to direct instantiation + sampler = jamun.sampling.Sampler(precision="32-true", callbacks=[], loggers=loggers_list) + py_logger.info("Using fallback sampler instantiation") + +# Instantiate batch sampler +try: + batch_sampler = hydra.utils.instantiate(cfg.sampling.batch_sampler) + py_logger.info("Successfully instantiated batch sampler") +except Exception as e: + py_logger.error(f"Error instantiating batch sampler: {e}") + # Fallback to direct instantiation + batch_sampler = jamun.sampling.Sampler( + num_steps=100, + sigma_min=0.001, + sigma_max=0.1 + ) + py_logger.info("Using fallback batch sampler instantiation") + +# %% Run Sampling +py_logger.info("Starting sampling...") +try: + sampler.sample( + model=model, + batch_sampler=batch_sampler, + init_graphs=init_graphs, + num_batches=cfg.sampling.num_batches, + continue_chain=cfg.sampling.continue_chain, + ) + py_logger.info("Sampling completed successfully!") + +except Exception as e: + py_logger.error(f"Sampling FAILED: {e}") + import traceback + traceback.print_exc() + +# %% Cleanup and Finish +py_logger.info("Sampling script finished.") + +# Finalize wandb if used +if loggers_list: + for logger in loggers_list: + if isinstance(logger, pl.loggers.WandbLogger): + logger.finalize(status="finished") + py_logger.info("Finalized WandB logger") + +py_logger.info("All done!") \ No newline at end of file diff --git a/scratch/training_prototype.py b/scratch/training_prototype.py new file mode 100644 index 0000000..35f73d7 --- /dev/null +++ b/scratch/training_prototype.py @@ -0,0 +1,356 @@ +# %% Imports and Basic Setup +import functools +import logging +import os +import sys +from typing import Union + +import dotenv +import tqdm # Often used in notebooks, can be optional in scripts +import torch +import e3nn +import e3tools.nn +import hydra +from hydra import compose, initialize +from omegaconf import OmegaConf +import lightning.pytorch as pl +import torch_geometric.data + +import jamun +import jamun.data +import jamun.distributions +import jamun.model +import jamun.model.arch +# Assuming these are in jamun.utils and jamun.hydra respectively +from jamun.utils import compute_average_squared_distance_from_datasets, find_checkpoint +from jamun.hydra import instantiate_dict_cfg + +# --- Basic Setup --- +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("jamun_script") + +torch.cuda.is_available() # Good to check, but PL trainer will also handle device +torch.set_float32_matmul_precision("high") +e3nn.set_optimization_defaults(jit_script_fx=False) + +# %% Environment and Paths +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + py_logger.info(f"Added '{project_root}' to sys.path for module discovery.") +else: + py_logger.info(f"'{project_root}' is already in sys.path.") + +# %% Load Hydra Configuration +py_logger.info("Loading Hydra configuration...") +# Adjust config_path relative to the script's location if it's not in scratch/ +# If jamun_training_script.py is in scratch/, and configs are in jamun/configs/ +# then config_path should be "../configs" +with initialize(config_path=".", job_name="conditioning_initial_run"): # Corrected job_name from previous context + cfg = compose( + config_name="config", # Main config file + overrides=[ + "model.arch._target_=scratch.e3conv_test.E3Conv", # Relative to project_root + "model._target_=scratch.denoiser_test.Denoiser", # Relative to project_root + "+model.arch.N_structures=2", + "trainer.max_epochs=100", # Example: train for 10 epochs + # Add other overrides, e.g. "logger=null" if you don't want default loggers for a quick test + ] + ) +py_logger.info("Loaded configuration:") +py_logger.info(OmegaConf.to_yaml(cfg)) + + +# %% Initial Dataset Setup (for model properties like average_squared_distance) +py_logger.info("Setting up initial dataset for model properties...") +initial_datasets_for_props = { + "props_dataset": jamun.data.parse_datasets_from_directory( # Renamed key for clarity + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, + subsample=100, # Keep this small for this purpose + max_datasets=1, + ) +} + +# %% Model Instantiation +py_logger.info("Instantiating model...") +try: + if not hasattr(cfg.model, "average_squared_distance") or cfg.model.average_squared_distance is None: + py_logger.info("Computing average_squared_distance for the model...") + average_squared_distance = compute_average_squared_distance_from_datasets( + initial_datasets_for_props['props_dataset'], # Use the small dataset for this + cfg.model.max_radius + ) + cfg.model.average_squared_distance = average_squared_distance + py_logger.info(f"Set cfg.model.average_squared_distance to {cfg.model.average_squared_distance}") + + # Provide conditioner if needed + if not hasattr(cfg.model, "conditioner"): + OmegaConf.set_struct(cfg.model, False) # Allow modification + cfg.model.conditioner = OmegaConf.create({}) + cfg.model.conditioner._target_ = 'scratch.conditioners.SelfConditioner' # Use the SelfConditioner from scratch.conditioners + OmegaConf.set_struct(cfg.model, True) # Lock structure again + py_logger.info(f"Set cfg.model.conditioner to instantiate {cfg.model.conditioner._target_}.") + + model = hydra.utils.instantiate(cfg.model) + + py_logger.info("Successfully instantiated model.") + py_logger.info(f"Instantiated model architecture type: {type(model.g)}") +except Exception as e: + py_logger.error(f"Error during model instantiation: {e}") + import traceback + traceback.print_exc() + sys.exit(1) + +# %% Device Setup for Model +# Determine the target device, preferring GPU if available. +if torch.cuda.is_available(): + device = torch.device("cuda") + py_logger.info("CUDA is available. Attempting to use GPU.") +else: + device = torch.device("cpu") + py_logger.info("CUDA not available. Using CPU.") + +# Move the model to the determined device. +model = model.to(device) + +# Verify and log the model's actual device. +# After .to(device), the model's internal device attribute should update. +# We can also check a parameter's device as a fallback verification. +final_model_device = None +if hasattr(model, 'device') and model.device is not None: + final_model_device = model.device +elif next(model.parameters(), None) is not None: + final_model_device = next(model.parameters()).device + +py_logger.info(f"Model '{type(model).__name__}' is now on device: {final_model_device}") + + +# %% Setup for Actual Training +py_logger.info("Setting up for actual training...") + +# 1. Prepare datasets for training, validation, and testing +# Using the same dataset source for all splits as a placeholder. +# In a real scenario, these would be different datasets or splits. +# You might want to parse a larger dataset here for actual training. +py_logger.info("Parsing dataset for training...") +training_dataset_source = jamun.data.parse_datasets_from_directory( + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", # Consider using full dataset + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, # Set to True for very large datasets if memory is an issue + subsample=cfg.data.datamodule.datasets.train[0].subsample, # Use subsample from config or None for full + max_datasets=1, # Use from config or None for all +) + +datasets_for_training = { + "train": training_dataset_source, + "val": training_dataset_source, # Replace with actual validation set + "test": training_dataset_source, # Replace with actual test set +} +py_logger.info(f"Prepared datasets for training: { {k: type(v).__name__ for k, v in datasets_for_training.items()} }") +if isinstance(training_dataset_source, torch_geometric.data.Dataset): + py_logger.info(f"Training dataset size: {len(training_dataset_source)}") + + +# 2. Initialize DataModule for training +datamodule_for_training = jamun.data.MDtrajDataModule( + datasets=datasets_for_training, + batch_size=cfg.data.datamodule.batch_size, + num_workers=cfg.data.datamodule.num_workers, +) + +# 3. Model is already instantiated and on device +py_logger.info(f"Model '{type(model).__name__}' is ready for training.") + +# %% Loggers and Callbacks Setup +py_logger.info("Setting up loggers and callbacks...") + +if cfg.get("logger") and cfg.logger.get("wandb"): + try: + # This requires JAMUN_ROOT_PATH, task_name, run_group, and run_key to be correctly + # defined and resolvable in your configuration. + wandb_save_dir = str(cfg.paths.run_path) # Resolve the path from OmegaConf + + # Update the logger config before instantiation + OmegaConf.update(cfg, "logger.wandb.save_dir", wandb_save_dir, merge=False) + py_logger.info(f"Explicitly setting WandbLogger save_dir to: {wandb_save_dir}") + + # Ensure the target directory for wandb files exists. + # WandbLogger will create its 'wandb/' subdirectory and run-specific folders inside this save_dir. + os.makedirs(wandb_save_dir, exist_ok=True) + py_logger.info(f"Ensured WandbLogger save_dir exists: {wandb_save_dir}") + + except Exception as e: + py_logger.error(f"Could not resolve or set wandb save_dir from cfg.paths.run_path: {e}") + py_logger.warning(f"Wandb will use default save directory (likely ./wandb in CWD: {os.getcwd()}).") + +# 1. Instantiate Loggers and Callbacks from Hydra config +loggers_list = [] +if cfg.get("logger"): + if hasattr(cfg.logger, '_target_'): + loggers_list.append(hydra.utils.instantiate(cfg.logger)) + else: + # Assuming instantiate_dict_cfg iterates and calls hydra.utils.instantiate for each logger config + loggers_list = instantiate_dict_cfg(cfg.logger) +py_logger.info(f"Instantiated loggers: {[type(l).__name__ for l in loggers_list]}") + +# %% 2. Determine and set the ModelCheckpoint directory path +final_checkpoint_dir = None +# Check if the first logger is a WandbLogger and provides a directory +if (loggers_list and + isinstance(loggers_list[0], pl.loggers.WandbLogger) and + hasattr(loggers_list[0], 'experiment') and loggers_list[0].experiment and + hasattr(loggers_list[0].experiment, 'dir') and loggers_list[0].experiment.dir): + + wandb_run_root_dir = loggers_list[0].experiment.dir + # Adjust if wandb_run_root_dir points to a 'files' subdirectory + if os.path.basename(wandb_run_root_dir) == "files": + wandb_run_root_dir = os.path.dirname(wandb_run_root_dir) + final_checkpoint_dir = os.path.join(wandb_run_root_dir, "checkpoints") + py_logger.info(f"Using WandB logger's experiment directory for checkpoints: {final_checkpoint_dir}") +else: + # Default path if no suitable WandB logger is found or no loggers are configured + final_checkpoint_dir = os.path.join(os.getcwd(), "outputs", "checkpoints") + if not loggers_list: + py_logger.info(f"No loggers configured. Defaulting checkpoint directory: {final_checkpoint_dir}") + else: + py_logger.info(f"First logger is not a suitable WandB logger. Defaulting checkpoint directory: {final_checkpoint_dir}") + +# Update the config if model_checkpoint callback is defined +if cfg.get("callbacks") and cfg.callbacks.get("model_checkpoint"): + # Use OmegaConf.update to safely set the possibly nested key, + # this will create it if it doesn't exist or overwrite if it does. + OmegaConf.update(cfg, "callbacks.model_checkpoint.dirpath", final_checkpoint_dir, merge=False) + py_logger.info(f"Set cfg.callbacks.model_checkpoint.dirpath to: {final_checkpoint_dir}") + # Ensure the directory exists + os.makedirs(final_checkpoint_dir, exist_ok=True) +else: + py_logger.info("ModelCheckpoint callback not configured in cfg.callbacks, dirpath not set.") + + +# %% Instantiate Callbacks +callbacks_list = [] +if cfg.get("callbacks"): + if hasattr(cfg.callbacks, '_target_'): # Single callback config + callbacks_list.append(hydra.utils.instantiate(cfg.callbacks)) + else: # Dictionary of callback configs + callbacks_list = instantiate_dict_cfg(cfg.callbacks) # This will now use the modified dirpath +py_logger.info(f"Instantiated callbacks: {[type(c).__name__ for c in callbacks_list]}") + +# 2. Instantiate PyTorch Lightning Trainer +trainer_config = cfg.trainer +if not hasattr(trainer_config, "_target_") and isinstance(trainer_config, dict): + trainer_config = OmegaConf.merge(trainer_config, {"_target_": "lightning.pytorch.Trainer"}) + +trainer: pl.Trainer = hydra.utils.instantiate( + trainer_config, + logger=loggers_list if loggers_list else True, + callbacks=callbacks_list, +) +py_logger.info(f"Instantiated Trainer: {type(trainer)}") +py_logger.info(f"Trainer will run for {trainer.max_epochs} epochs.") + +# 3. Handle checkpoint resumption (optional) +checkpoint_path = None +if resume_checkpoint_cfg := cfg.get("resume_from_checkpoint"): + if resume_checkpoint_cfg.get("enabled", False): + py_logger.info(f"Attempting to resume from checkpoint with config: {resume_checkpoint_cfg}") + try: + checkpoint_path = find_checkpoint( + wandb_train_run_path=resume_checkpoint_cfg.get("wandb_train_run_path"), + checkpoint_dir=resume_checkpoint_cfg.get("checkpoint_dir"), + checkpoint_type=resume_checkpoint_cfg.get("checkpoint_type", "last"), + ) + if checkpoint_path: + py_logger.info(f"Found checkpoint to resume from: {checkpoint_path}") + else: + py_logger.warning("No checkpoint found for resumption based on config.") + except Exception as e: + py_logger.error(f"Error finding checkpoint: {e}. Starting training from scratch.") + checkpoint_path = None + else: + py_logger.info("Checkpoint resumption is configured but not enabled.") + +# %% Start Training +py_logger.info("Starting training...") +try: + trainer.fit( + model=model, # Use the main model instance + datamodule=datamodule_for_training, + ckpt_path=checkpoint_path if checkpoint_path else None + ) + py_logger.info("Training finished.") + + if cfg.get("run_test_after_train", False): + py_logger.info("Running test phase...") + trainer.test(model=model, datamodule=datamodule_for_training) + py_logger.info("Test phase finished.") + +except Exception as e: + py_logger.error(f"Training FAILED: {e}") + traceback.print_exc() + +py_logger.info("Script finished.") + +# %% Log the final configuration and save it locally +wandb_logger_instance = None +# loggers_list should still be in scope from when it was passed to the Trainer +for logger_from_list in loggers_list: # Use a different variable name to avoid conflict if logger is defined elsewhere + if isinstance(logger_from_list, pl.loggers.WandbLogger): + wandb_logger_instance = logger_from_list + break + +if wandb_logger_instance and hasattr(wandb_logger_instance, 'experiment') and wandb_logger_instance.experiment: + py_logger.info(f"WandbLogger experiment active (run_id: {wandb_logger_instance.experiment.id}). Logging final script config to wandb.") + # Convert the current OmegaConf object 'cfg' to a plain dictionary + # This 'cfg' includes all modifications made throughout the script + final_script_cfg_dict = OmegaConf.to_container(cfg, resolve=True, throw_on_missing=True) + + try: + wandb_logger_instance.experiment.config.update( + {"final_script_cfg": final_script_cfg_dict, "jamun_version_at_end": jamun.__version__, "script_cwd_at_end": os.getcwd()} + ) + py_logger.info("Updated wandb.config with final_script_cfg.") + except Exception as e: + py_logger.error(f"Failed to update wandb.config with final script config: {e}") +else: + if cfg.get("logger") and cfg.logger.get("wandb"): + py_logger.warning("WandbLogger was configured but not found or experiment not active at script end. Final script config not logged to wandb.config.") + +# 2. Explicitly save the final state of the OmegaConf object 'cfg' to a local file + +final_config_output_dir = None +if cfg.get("logger") and cfg.logger.get("wandb") and cfg.logger.wandb.get("save_dir"): + final_config_output_dir = cfg.logger.wandb.save_dir +elif 'wandb_save_dir' in locals() and wandb_save_dir: # If it was set in a previous cell + final_config_output_dir = wandb_save_dir +else: + # Fallback if a specific run directory isn't easily available + # This might not be ideal as it won't be co-located with W&B run files if save_dir wasn't set + final_config_output_dir = os.path.join(os.getcwd(), "outputs", cfg.get("task_name", "unknown_task"), cfg.get("run_key", "unknown_run")) + os.makedirs(final_config_output_dir, exist_ok=True) + + +if final_config_output_dir: + final_config_path = os.path.join(final_config_output_dir, "final_resolved_script_config.yaml") + try: + with open(final_config_path, 'w') as f: + OmegaConf.save(config=cfg, f=f) + py_logger.info(f"Final script configuration saved locally to: {final_config_path}") + except Exception as e: + py_logger.error(f"Failed to save final script configuration locally: {e}") +else: + py_logger.warning("Could not determine a definitive output directory for final_resolved_script_config.yaml. Not saving locally.") + +loggers_list[0].experiment.finish() if loggers_list and isinstance(loggers_list[0], pl.loggers.WandbLogger) else None +py_logger.info("Script finished.") +# %% diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 86e4e7b..236a38b 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -40,13 +40,14 @@ def run(cfg): dist_log(f"{os.getcwd()=}") dist_log(f"{torch.__config__.parallel_info()}") dist_log(f"{os.sched_getaffinity(0)=}") - + sys.exit() + print(f"Terminating...") # Set the start method to spawn to avoid issues with the default fork method. torch.multiprocessing.set_start_method("spawn", force=True) # Compute data normalization. if cfg.get("compute_average_squared_distance_from_data"): - average_squared_distance = compute_average_squared_distance_from_config(cfg) + average_squared_distance = compute_average_squared_distance_from_config(cfg)s dist_log( f"Overwriting average_squared_distance in config from {cfg.model.average_squared_distance} to {average_squared_distance}." ) diff --git a/src/jamun/hydra_config/sample.yaml b/src/jamun/hydra_config/sample.yaml index dc97330..e7f53d2 100644 --- a/src/jamun/hydra_config/sample.yaml +++ b/src/jamun/hydra_config/sample.yaml @@ -6,7 +6,7 @@ defaults: - paths: default - hydra: default - callbacks: sampler/default - - experiment: null + - experiment: sample_uncapped_single_shape_conditioning float32_matmul_precision: "high" From 8e1bd9c87836dabf4be6c22a11a8e0078cec876c Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Mon, 16 Jun 2025 22:23:44 +0000 Subject: [PATCH 07/32] checks on conditioned denoising model --- configs/experiment/sample_custom.yaml | 16 ++-- scratch/check_hidden_state.py | 108 ++++++++++++++++++++++++++ scratch/conditioners.py | 6 ++ scratch/hydra_trials.py | 42 +++++----- scratch/load_wandb_checkpoint.py | 60 ++++++++++++++ scratch/sample.py | 20 ++--- scratch/training_prototype.py | 2 +- src/jamun/cmdline/sample.py | 17 +++- src/jamun/utils/checkpoint.py | 3 +- 9 files changed, 232 insertions(+), 42 deletions(-) create mode 100644 scratch/check_hidden_state.py create mode 100644 scratch/load_wandb_checkpoint.py diff --git a/configs/experiment/sample_custom.yaml b/configs/experiment/sample_custom.yaml index fe72cea..3052289 100644 --- a/configs/experiment/sample_custom.yaml +++ b/configs/experiment/sample_custom.yaml @@ -1,5 +1,7 @@ # @package _global_ +name: cfg # This ensures the config is saved with the 'cfg' key + # defaults: # - override /callbacks: # - sampler/save_trajectory.yaml @@ -19,14 +21,15 @@ init_datasets: subsample: 1 max_datasets: 1 -num_sampling_steps_per_batch: 100 -num_batches: 10 +num_sampling_steps_per_batch: 1000 +num_batches: 100 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your wandb run path here -# wandb_train_run_path: your-wandb-run-path +wandb_train_run_path: sule-shashank/jamun/y4rm5488 +# checkpoint_dir: outputs/train/dev/runs/2025-06-11_20-16-04/wandb/latest-run/checkpoints checkpoint_type: best_so_far sigma: 0.04 @@ -38,10 +41,7 @@ score_fn_clip: 100.0 # Model configuration model: - _target_: scratch.denoiser_test.Denoiser - conditioner: - _target_: scratch.conditioners.SelfConditioner - hidden_dim: 1 + _target_: scratch.denoiser_test.Denoiser.load_from_checkpoint sampler: _target_: jamun.sampling.Sampler @@ -50,4 +50,4 @@ sampler: logger: wandb: group: sample_custom - notes: Custom sampling with denoiser and conditioner \ No newline at end of file + notes: Custom sampling with denoiser, conditioner \ No newline at end of file diff --git a/scratch/check_hidden_state.py b/scratch/check_hidden_state.py new file mode 100644 index 0000000..9abfb69 --- /dev/null +++ b/scratch/check_hidden_state.py @@ -0,0 +1,108 @@ +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import logging +import os +import sys +import torch +import torch_geometric.data +import hydra +from hydra import compose, initialize +from omegaconf import OmegaConf +import wandb +from jamun.utils import find_checkpoint +import jamun.data +import dotenv +from denoiser_test import Denoiser + + +# Setup logging +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("check_hidden_state") + +# Add project root to path +project_root = "/homefs/home/sules/jamun" +if project_root not in sys.path: + sys.path.insert(0, project_root) + +# Load configuration +with initialize(config_path="", job_name="check_hidden_state"): + cfg = compose( + config_name="config", + overrides=[ + "model.arch._target_=scratch.e3conv_test.E3Conv", + "model._target_=scratch.denoiser_test.Denoiser", + "+model.arch.N_structures=2", # We need at least 2 structures to test hidden state + "model.use_torch_compile=false", # Disable torch.compile to avoid ScriptModule issues + "+model.conditioner._target_=scratch.conditioners.SelfConditioner", + ] + ) + +# Load checkpoint +checkpoint_path = find_checkpoint(wandb_train_run_path="sule-shashank/jamun/y4rm5488", checkpoint_type='last') +checkpoint = torch.load(checkpoint_path, map_location='cpu', weights_only=False) + +# Modify hyperparameters to disable torch.compile +if 'hyper_parameters' in checkpoint: + checkpoint['hyper_parameters']['use_torch_compile'] = False + checkpoint['hyper_parameters']['torch_compile_kwargs'] = None + +# Load model with modified hyperparameters +breakpoint() +model = Denoiser.load_from_checkpoint(checkpoint_path) +model.eval() + +# Get test data +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +datasets = { + "test": jamun.data.parse_datasets_from_directory( + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", + traj_pattern="^(.*)-traj-arrays.npz", + pdb_file="AA-traj-state0.pdb", + filter_codes=['AA'], + as_iterable=False, + subsample=100, + max_datasets=1, + ) +} + +datamodule = jamun.data.MDtrajDataModule( + datasets=datasets, + batch_size=3, + num_workers=2, +) +datamodule.setup('test') +_, test_data = next(enumerate(datamodule.test_dataloader())) +test_data = test_data.to(model.device) + +# Ensure test_data has hidden_state +if not hasattr(test_data, 'hidden_state') or not test_data.hidden_state: + py_logger.info("Adding hidden state to test data") + test_data.hidden_state = [torch.randn_like(test_data.pos) for _ in range(model.g.N_structures - 1)] + +breakpoint() +# Add noise and denoise +sigma = torch.tensor(0.04) # Same sigma as in config +with torch.no_grad(): + xhat, y = model.noise_and_denoise(test_data, sigma, align_noisy_input=False) + +# Check if hidden state is preserved +print("\nChecking hidden state preservation:") +print(f"Original hidden state shapes: {[hs.shape for hs in test_data.hidden_state]}") +print(f"Noisy hidden state shapes: {[hs.shape for hs in y.hidden_state]}") +print(f"Denoised hidden state shapes: {[hs.shape for hs in xhat.hidden_state]}") + +# Check if hidden state values are preserved +for i in range(len(test_data.hidden_state)): + hidden_state_diff = torch.abs(xhat.hidden_state[i] - test_data.hidden_state[i]).mean() + print(f"\nMean absolute difference between original and denoised hidden state {i}: {hidden_state_diff.item()}") + +# Check if positions are actually denoised +pos_diff = torch.abs(xhat.pos - test_data.pos).mean() +print(f"Mean absolute difference between original and denoised positions: {pos_diff.item()}") + +# Check if noisy positions are different from original +noisy_pos_diff = torch.abs(y.pos - test_data.pos).mean() +print(f"Mean absolute difference between original and noisy positions: {noisy_pos_diff.item()}") \ No newline at end of file diff --git a/scratch/conditioners.py b/scratch/conditioners.py index 1afeb35..9dafc01 100644 --- a/scratch/conditioners.py +++ b/scratch/conditioners.py @@ -13,6 +13,9 @@ class PositionConditioner(pl.LightningModule): + """ + Condition the hidden state on the position of the structure. + """ def __init__(self, **kwargs): super().__init__() @@ -24,6 +27,9 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: return conditioned_structures class SelfConditioner(pl.LightningModule): + """ + No conditioning, but add the position of the structure to itself to make it compatible with the denoiser. + """ def __init__(self, **kwargs): super().__init__() diff --git a/scratch/hydra_trials.py b/scratch/hydra_trials.py index 106ec3f..832011d 100644 --- a/scratch/hydra_trials.py +++ b/scratch/hydra_trials.py @@ -5,6 +5,7 @@ import sys import dotenv import logging +import traceback # --- Basic Setup --- logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) @@ -41,29 +42,26 @@ def print_config_sections(cfg): def run(cfg): """Main function to run the config loading and printing.""" - # Print the loaded configuration - print_config_sections(cfg) - - # Print specific config values - if hasattr(cfg, 'model'): - print("\nModel target:", cfg.model._target_) - if hasattr(cfg, 'sampler'): - print("Sampler target:", cfg.sampler._target_) + try: + # Print the loaded configuration + print_config_sections(cfg) + + # Print specific config values + if hasattr(cfg, 'model'): + print("\nModel target:", cfg.model._target_) + if hasattr(cfg, 'sampler'): + print("Sampler target:", cfg.sampler._target_) + except Exception: + traceback.print_exc(file=sys.stderr) + raise -def main(): - # Initialize Hydra - with hydra.initialize(config_path="../src/jamun/hydra_config"): - # Compose the base configuration - base_cfg = hydra.compose(config_name="sample") - - # Compose the experiment configuration - experiment_cfg = hydra.compose(config_name="sample", overrides=["experiment=sample_uncapped_single_shape_conditioning"]) - - # Merge configurations (experiment overrides base) - cfg = OmegaConf.merge(base_cfg, experiment_cfg) - - # Run with the merged configuration - run(cfg) +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample") +def main(cfg): + try: + run(cfg) + except Exception: + traceback.print_exc(file=sys.stderr) + raise if __name__ == "__main__": main() diff --git a/scratch/load_wandb_checkpoint.py b/scratch/load_wandb_checkpoint.py new file mode 100644 index 0000000..06ba3ec --- /dev/null +++ b/scratch/load_wandb_checkpoint.py @@ -0,0 +1,60 @@ +import sys +import os +import torch +import logging +import dotenv +import hydra +from omegaconf import OmegaConf +from denoiser_test import Denoiser +from jamun.utils.checkpoint import find_checkpoint + +# Setup logging +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +logger = logging.getLogger("load_wandb_checkpoint") + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + logger.info(f"Added '{project_root}' to sys.path for module discovery.") +else: + logger.info(f"'{project_root}' is already in sys.path.") + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample") +def load_model_from_wandb(cfg, wandb_path: str, device: str = "cuda" if torch.cuda.is_available() else "cpu"): + """ + Load a model checkpoint from wandb. + + Args: + cfg: Hydra configuration + wandb_path (str): Path to the wandb checkpoint (e.g., "entity/project/run_id") + device (str): Device to load the model on + + Returns: + Denoiser: The loaded model + """ + # Find the checkpoint path using the utility function + checkpoint_path = find_checkpoint( + wandb_train_run_path=wandb_path, + checkpoint_type="last" # or "best_so_far" if you want the best checkpoint + ) + logger.info(f"Found checkpoint at: {checkpoint_path}") + + # Update the config with the checkpoint path + cfg.model.checkpoint_path = checkpoint_path + + # Load the model using Hydra + model = hydra.utils.instantiate(cfg.model) + model = model.to(device) + model.eval() + + logger.info("Model loaded successfully") + return model + +if __name__ == "__main__": + # Example usage + wandb_path = "sule-shashank/jamun/y4rm5488" # Replace with actual wandb path + model = load_model_from_wandb(wandb_path) \ No newline at end of file diff --git a/scratch/sample.py b/scratch/sample.py index f526000..39c4792 100644 --- a/scratch/sample.py +++ b/scratch/sample.py @@ -51,7 +51,7 @@ def run(cfg): log_cfg = OmegaConf.to_container(cfg, throw_on_missing=True, resolve=True) # Breakpoint 0: After config is loaded - pdb.set_trace() # Inspect full configuration + # pdb.set_trace() # Inspect full configuration dist_log(f"{OmegaConf.to_yaml(log_cfg)}") dist_log(f"{os.getcwd()=}") @@ -63,17 +63,19 @@ def run(cfg): dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") torch.set_float32_matmul_precision(matmul_prec) - loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) - wandb_logger = None - for logger in loggers: - if isinstance(logger, pl.loggers.WandbLogger): - wandb_logger = logger + # # Turned off wandb logging for now + # loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) + # wandb_logger = None + # for logger in loggers: + # if isinstance(logger, pl.loggers.WandbLogger): + # wandb_logger = logger - if rank_zero_only.rank == 0 and wandb_logger: - dist_log(f"{wandb_logger.experiment.name=}") - wandb_logger.experiment.config.update({"cfg": log_cfg, "version": jamun.__version__, "cwd": os.getcwd()}) + # if rank_zero_only.rank == 0 and wandb_logger: + # dist_log(f"{wandb_logger.experiment.name=}") + # wandb_logger.experiment.config.update({"cfg": log_cfg, "version": jamun.__version__, "cwd": os.getcwd()}) # Load the checkpoint + print(f'Current working directory: {os.getcwd()}') checkpoint_path = find_checkpoint( wandb_train_run_path=cfg.get("wandb_train_run_path"), checkpoint_dir=cfg.get("checkpoint_dir"), diff --git a/scratch/training_prototype.py b/scratch/training_prototype.py index 35f73d7..5d99f02 100644 --- a/scratch/training_prototype.py +++ b/scratch/training_prototype.py @@ -317,7 +317,7 @@ try: wandb_logger_instance.experiment.config.update( - {"final_script_cfg": final_script_cfg_dict, "jamun_version_at_end": jamun.__version__, "script_cwd_at_end": os.getcwd()} + {"cfg": final_script_cfg_dict, "jamun_version_at_end": jamun.__version__, "script_cwd_at_end": os.getcwd()} ) py_logger.info("Updated wandb.config with final_script_cfg.") except Exception as e: diff --git a/src/jamun/cmdline/sample.py b/src/jamun/cmdline/sample.py index 8601981..87869db 100644 --- a/src/jamun/cmdline/sample.py +++ b/src/jamun/cmdline/sample.py @@ -23,7 +23,22 @@ dotenv.load_dotenv(".env", verbose=True) OmegaConf.register_new_resolver("format", format_resolver) +import logging +# Setup logging +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +logger = logging.getLogger("load_wandb_checkpoint") + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + logger.info(f"Added '{project_root}' to sys.path for module discovery.") +else: + logger.info(f"'{project_root}' is already in sys.path.") def get_initial_graphs( datasets: Sequence[MDtrajDataset], num_init_samples_per_dataset: int, repeat: int = 1 @@ -51,7 +66,7 @@ def run(cfg): if matmul_prec := cfg.get("float32_matmul_precision"): dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") torch.set_float32_matmul_precision(matmul_prec) - + breakpoint() loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) wandb_logger = None for logger in loggers: diff --git a/src/jamun/utils/checkpoint.py b/src/jamun/utils/checkpoint.py index e6f26d9..663c1c7 100644 --- a/src/jamun/utils/checkpoint.py +++ b/src/jamun/utils/checkpoint.py @@ -16,7 +16,8 @@ def get_wandb_run_config(wandb_run_path: str) -> Dict[str, Any]: run = wandb.Api().run(wandb_run_path) py_logger = logging.getLogger("jamun") py_logger.info(f"Loading checkpoint corresponding to wandb run {run.name} at {run.url}") - return run.config["cfg"] + key = next(iter(run.config)) # the key might be named differently in the future + return run.config[key] def get_run_path_for_wandb_run(wandb_run_path: str) -> str: From c82d4287df8e6f3b52d35af57b2dffaa1a95dcf1 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Tue, 24 Jun 2025 00:07:49 +0000 Subject: [PATCH 08/32] conditioning module set up, dataloaders with time lags, and adding multimeasurement noise --- .../train_test_single_shape_conditional.yaml | 71 ++++++ configs/sweep.yaml | 18 ++ scratch/check_model_arch.py | 16 +- scratch/run_single_shape_AA_conditional.sh | 8 +- scratch/test_conditional.py | 88 +++++++ scripts/slurm/sweep.sh | 33 +++ .../train_capped_2AA_ALA_ALA_conditional.sh | 38 +++ src/jamun/cmdline/train.py | 7 +- src/jamun/data/_mdtraj.py | 82 ++++++- src/jamun/data/_subsample.py | 87 +++++++ src/jamun/data/tests/test_subsample.py | 229 ++++++++++++++++++ .../model/arch/e3conv_conditional.yaml | 31 +++ .../model/denoiser_conditional.yaml | 29 +++ src/jamun/model/__init__.py | 2 +- src/jamun/model/arch/__init__.py | 1 + .../jamun/model/arch/e3conv_conditional.py | 2 +- src/jamun/model/conditioners/__init__.py | 2 + .../jamun/model/conditioners}/conditioners.py | 21 +- .../jamun/model/denoiser_conditional.py | 113 +++++++-- src/jamun/model/noise_test.py | 110 +++++++++ wandb/sweep-01dho9mz/config-o4x6y3nm.yaml | 6 + wandb/sweep-01dho9mz/config-wrqb60uz.yaml | 6 + wandb/sweep-5tie5khj/config-lhzhrmk7.yaml | 4 + wandb/sweep-gze9gptu/config-sxlo5fvy.yaml | 6 + wandb/sweep-gze9gptu/config-vfd1ntbz.yaml | 6 + wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml | 6 + wandb/sweep-rxcjaig2/config-zvdgfxss.yaml | 6 + 27 files changed, 978 insertions(+), 50 deletions(-) create mode 100644 configs/experiment/train_test_single_shape_conditional.yaml create mode 100644 configs/sweep.yaml create mode 100644 scratch/test_conditional.py create mode 100644 scripts/slurm/sweep.sh create mode 100644 scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh create mode 100644 src/jamun/data/_subsample.py create mode 100644 src/jamun/data/tests/test_subsample.py create mode 100644 src/jamun/hydra_config/model/arch/e3conv_conditional.yaml create mode 100644 src/jamun/hydra_config/model/denoiser_conditional.yaml rename scratch/e3conv_test.py => src/jamun/model/arch/e3conv_conditional.py (99%) create mode 100644 src/jamun/model/conditioners/__init__.py rename {scratch => src/jamun/model/conditioners}/conditioners.py (68%) rename scratch/denoiser_test.py => src/jamun/model/denoiser_conditional.py (79%) create mode 100644 src/jamun/model/noise_test.py create mode 100644 wandb/sweep-01dho9mz/config-o4x6y3nm.yaml create mode 100644 wandb/sweep-01dho9mz/config-wrqb60uz.yaml create mode 100644 wandb/sweep-5tie5khj/config-lhzhrmk7.yaml create mode 100644 wandb/sweep-gze9gptu/config-sxlo5fvy.yaml create mode 100644 wandb/sweep-gze9gptu/config-vfd1ntbz.yaml create mode 100644 wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml create mode 100644 wandb/sweep-rxcjaig2/config-zvdgfxss.yaml diff --git a/configs/experiment/train_test_single_shape_conditional.yaml b/configs/experiment/train_test_single_shape_conditional.yaml new file mode 100644 index 0000000..519a6bd --- /dev/null +++ b/configs/experiment/train_test_single_shape_conditional.yaml @@ -0,0 +1,71 @@ +# @package _global_ + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + num_workers: 8 + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 + total_lag_time: 5 + lag_subsample_rate: 100 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 + total_lag_time: 5 + lag_subsample_rate: 100 + + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 + total_lag_time: 5 + lag_subsample_rate: 100 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 1 + + +logger: + wandb: + group: ALA_ALA, capped diamines, conditional denoiser + notes: "Running on ALA_ALA capped diamine, self-conditioned denoiser" \ No newline at end of file diff --git a/configs/sweep.yaml b/configs/sweep.yaml new file mode 100644 index 0000000..e264c3f --- /dev/null +++ b/configs/sweep.yaml @@ -0,0 +1,18 @@ +program: jamun_train +method: grid +project: jamun +name: conditional_vs_self +metric: + name: val/loss + goal: minimize +parameters: + model.conditioner._target_: + values: + - jamun.model.conditioners.PositionConditioner + - jamun.model.conditioners.SelfConditioner + +command: + - ${program} + - "--config-dir=configs" + - "experiment=train_test_single_shape_conditional" + - ${args_no_hyphens} \ No newline at end of file diff --git a/scratch/check_model_arch.py b/scratch/check_model_arch.py index 7ec7426..1830cec 100644 --- a/scratch/check_model_arch.py +++ b/scratch/check_model_arch.py @@ -33,24 +33,28 @@ # %% datasets = { "test": jamun.data.parse_datasets_from_directory( - root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", - traj_pattern="^(.*)-traj-arrays.npz", - pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], + root=f"{JAMUN_DATA_PATH}/capped_diamines/timewarp_splits/train/", + traj_pattern="^(.*).xtc", + pdb_file="ALA_ALA.pdb", + filter_codes=['ALA_ALA'], as_iterable=False, subsample=100, + total_lag_time=10, + lag_subsample_rate=100, max_datasets=1, ) } +# %% datamodule = jamun.data.MDtrajDataModule( datasets=datasets, - batch_size=3, + batch_size=5, num_workers=2, ) datamodule.setup('test') _, data_batch = next(enumerate(datamodule.test_dataloader())) - +print(f'Number of hidden states: {len(data_batch.hidden_state)}') +print(f'Size of one hidden state: {data_batch.hidden_state[0].shape}') # %% test the new e3conv_test class from e3conv_test import E3Conv import torch_geometric diff --git a/scratch/run_single_shape_AA_conditional.sh b/scratch/run_single_shape_AA_conditional.sh index 187361c..e92a26f 100644 --- a/scratch/run_single_shape_AA_conditional.sh +++ b/scratch/run_single_shape_AA_conditional.sh @@ -37,11 +37,11 @@ mkdir -p "${LOG_DIR_BASE}" # --- Application Execution --- # Navigate to the directory containing your script, if necessary # Assuming training_prototype.py is in /homefs/home/sules/jamun/scratch/ -SCRIPT_DIR="/homefs/home/sules/jamun/scratch" -PYTHON_SCRIPT="training_prototype.py" +# SCRIPT_DIR="/homefs/home/sules/jamun/scratch" +# PYTHON_SCRIPT="training_prototype.py" -echo "Changing directory to ${SCRIPT_DIR}" -cd "${SCRIPT_DIR}" || { echo "Failed to cd to ${SCRIPT_DIR}"; exit 1; } +# echo "Changing directory to ${SCRIPT_DIR}" +# cd "${SCRIPT_DIR}" || { echo "Failed to cd to ${SCRIPT_DIR}"; exit 1; } echo "Starting Python script: ${PYTHON_SCRIPT}" # Run the Python script diff --git a/scratch/test_conditional.py b/scratch/test_conditional.py new file mode 100644 index 0000000..badc01f --- /dev/null +++ b/scratch/test_conditional.py @@ -0,0 +1,88 @@ +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +from jamun.hydra import instantiate_dict_cfg +import pdb +import jamun +from jamun.utils import compute_average_squared_distance_from_datasets + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + print(f"Added '{project_root}' to sys.path for module discovery.") +else: + print(f"'{project_root}' is already in sys.path.") + +def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: + """Computes the average squared distance for normalization from the data.""" + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + cutoff = cfg.model.max_radius + average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) + return average_squared_distance + + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") +def main(cfg): + # Load the test config + average_squared_distance = compute_average_squared_distance_from_config(cfg) + cfg.model.average_squared_distance = average_squared_distance + # breakpoint() + + # # First merge test config into base config, then override with test config + # cfg = OmegaConf.merge(cfg, test_cfg) + # cfg = OmegaConf.merge(cfg, test_cfg, override=True) + # breakpoint() + + print("Loading datamodule...") + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup('test') + # breakpoint() + + print("Loading model...") + model = hydra.utils.instantiate(cfg.model) + breakpoint() + + # Get a single batch + print("Getting a batch of data...") + train_loader = datamodule.train_dataloader() + _, batch = next(enumerate(train_loader)) + # breakpoint() + + # # Move to CPU + # batch = batch.to("cpu") + # model = model.to("cpu") + + print(f"Batch shape: {batch.pos.shape}") + print(f"Hidden state shape: {[h.shape for h in batch.hidden_state]}") + breakpoint() + + # Test forward pass + print("Testing forward pass...") + with torch.no_grad(): + sigma = model.sigma_distribution.sample() + _, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + + print(f"Input shape: {batch.pos.shape}") + print(f"Noisy shape: {y.pos.shape}") + print(f"Output shape: {xhat.pos.shape}") + + # Test loss computation + print("Testing loss computation...") + loss, aux = model.compute_loss(batch, xhat, sigma) + print(f"Loss: {loss.mean().item():.4f}") + print(f"Metrics: {aux}") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scripts/slurm/sweep.sh b/scripts/slurm/sweep.sh new file mode 100644 index 0000000..74c167e --- /dev/null +++ b/scripts/slurm/sweep.sh @@ -0,0 +1,33 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu3 +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node=4 # Number of agents to run in parallel on this node +#SBATCH --gpus-per-task=1 # Assign one GPU to each agent +#SBATCH --cpus-per-task=8 +#SBATCH --time 3-0 +#SBATCH --mem-per-cpu=32G + +# Check if a Sweep ID is provided as an argument +if [ -z "$1" ]; then + echo "Error: Please provide the W&B Sweep ID as the first argument." + echo "Usage: sbatch scripts/slurm/sweep.sh " + exit 1 +fi + +SWEEP_ID=$1 + +# Set up the environment +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID: ${SLURM_JOB_ID}" +echo "Running on hostname: $(hostname)" +echo "Starting ${SLURM_NTASKS} agents for sweep: ${SWEEP_ID}" + +# Launch multiple wandb agents in parallel using srun. +# Each agent will poll the sweep server, get a configuration, and run one training job. +# PyTorch Lightning will automatically use the single GPU assigned by Slurm to each task. +srun wandb agent "${SWEEP_ID}" \ No newline at end of file diff --git a/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh b/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh new file mode 100644 index 0000000..de0c4d8 --- /dev/null +++ b/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh @@ -0,0 +1,38 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu3 +#SBATCH --qos=preempt +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node 4 +#SBATCH --gpus-per-node 4 +#SBATCH --cpus-per-task 8 +#SBATCH --time 3-0 +#SBATCH --mem-per-cpu=32G + +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 +# export TORCH_COMPILE_DEBUG=1 +# export TORCH_LOGS="+dynamo" +# export TORCHDYNAMO_VERBOSE=1 + +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +nvidia-smi + +srun --cpus-per-task 8 --cpu-bind=cores,verbose \ + jamun_train --config-dir=/homefs/home/sules/jamun/configs \ + experiment=train_test_single_shape_conditional.yaml \ + ++model.conditioner._target_=jamun.model.conditioners.SelfConditioner \ + ++trainer.devices=$SLURM_GPUS_PER_NODE \ + ++trainer.num_nodes=$SLURM_JOB_NUM_NODES \ + ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","train","capped_2AA, ALA_ALA, conditional denoiser"] \ + ++run_key=$RUN_KEY \ No newline at end of file diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 236a38b..32b5e4f 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -40,14 +40,13 @@ def run(cfg): dist_log(f"{os.getcwd()=}") dist_log(f"{torch.__config__.parallel_info()}") dist_log(f"{os.sched_getaffinity(0)=}") - sys.exit() - print(f"Terminating...") + # Set the start method to spawn to avoid issues with the default fork method. - torch.multiprocessing.set_start_method("spawn", force=True) + # torch.multiprocessing.set_start_method("spawn", force=True) # Compute data normalization. if cfg.get("compute_average_squared_distance_from_data"): - average_squared_distance = compute_average_squared_distance_from_config(cfg)s + average_squared_distance = compute_average_squared_distance_from_config(cfg) dist_log( f"Overwriting average_squared_distance in config from {cfg.model.average_squared_distance} to {average_squared_distance}." ) diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index 0fc4aa6..45a8375 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -1,6 +1,6 @@ import functools import os -from typing import Callable, Optional, Sequence, Tuple +from typing import Callable, Optional, Sequence, Tuple, List import mdtraj as md import numpy as np @@ -11,6 +11,47 @@ from jamun import utils +def get_subsampled_indices( + N: int, + subsample_rate: int, + total_lag_time: int, + lag_subsample_rate: int, +) -> List[np.ndarray]: + """ + Generate subsampled indices and their corresponding lagged indices. + + Args: + N: Total number of frames + subsample_rate: Rate at which to subsample the frames + total_lag_time: Number of lagged frames to generate for each subsampled frame + lag_subsample_rate: Rate at which to subsample the lagged frames + + Returns: + List of arrays, where each array contains the lagged indices for a subsampled frame + + Raises: + ValueError: If the input parameters don't satisfy the required constraints + """ + # Check guardrails + if N / subsample_rate < 1: + raise ValueError(f"Number of samples (N/subsample_rate = {N/subsample_rate}) must be >= 1") + + # Generate subsampled indices + subsampled_indices = np.arange(0, N, subsample_rate) + + # Generate lagged indices for each subsampled index + lagged_indices = [] + for idx in subsampled_indices: + # Calculate lagged indices + lagged = [int(idx - j * lag_subsample_rate) for j in range(total_lag_time)] + + # Check if we have enough lagged indices + if len(lagged) == total_lag_time and all(x >= 0 for x in lagged): + lagged_indices.append(lagged) + + return lagged_indices + + def preprocess_topology(topology: md.Topology) -> Tuple[torch_geometric.data.Data, md.Topology, md.Topology]: """Preprocess the MDtraj topology, returning a PyTorch Geometric graph, the topology with protein only, and the topology with hydrogenated protein.""" # Select all heavy atoms in the protein. @@ -138,6 +179,8 @@ def __init__( start_frame: Optional[int] = None, transform: Optional[Callable] = None, subsample: Optional[int] = None, + total_lag_time: Optional[int] = None, + lag_subsample_rate: Optional[int] = None, loss_weight: float = 1.0, verbose: bool = False, ): @@ -173,19 +216,35 @@ def __init__( if subsample is None or subsample == 0: subsample = 1 - # Subsample the trajectory. - self.traj = self.traj[start_frame : start_frame + num_frames : subsample] + # Get lagged indices if lag parameters are provided + if total_lag_time is not None and lag_subsample_rate is not None: + lagged_indices = get_subsampled_indices( + self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate + ) + # Extract subsampled indices (first element of each list) + subsampled_indices = [indices[0] for indices in lagged_indices] + # Extract lagged indices (all except first element) + self.lagged_indices = [indices[1:] for indices in lagged_indices] + # Subsample the trajectory using the subsampled indices + self.hidden_state = [self.traj[indices] for indices in self.lagged_indices] + self.traj = self.traj[subsampled_indices] # self.traj is permanently modified. + else: + # Regular subsampling without lag + self.traj = self.traj[start_frame : start_frame + num_frames : subsample] + self.hidden_state = None + self.lagged_indices = None + topology = self.traj.topology self.graph, self.top, self.top_withH = preprocess_topology(topology) - self.traj = self.traj.atom_slice(topology.select("protein and not type H")) + atom_selection = topology.select("protein and not type H") + self.traj = self.traj.atom_slice(atom_selection) + if self.hidden_state is not None: + self.hidden_state = [traj.atom_slice(atom_selection) for traj in self.hidden_state] # select protein atoms for hidden state(s) self.graph.pos = torch.tensor(self.traj.xyz[0], dtype=torch.float32) - self.graph.hidden_state = torch.tensor(self.traj.xyz[-1], dtype=torch.float32) self.graph.loss_weight = torch.tensor([loss_weight], dtype=torch.float32) self.graph.dataset_label = self.label() - - # self.save_topology_pdb() - + self.graph.hidden_state = [self.hidden_state[0].xyz[i] for i in range(self.hidden_state[0].n_frames)] if verbose: utils.dist_log(f"Dataset {self.label()}: Loading trajectory files {traj_files} and PDB file {pdb_file}.") utils.dist_log( @@ -203,7 +262,12 @@ def save_topology_pdb(self): def __getitem__(self, idx): graph = self.graph.clone() graph.pos = torch.tensor(self.traj.xyz[idx]) - graph.hidden_state = [torch.tensor(self.traj.xyz[idx-1])] + + if self.lagged_indices is not None: + graph.hidden_state = [torch.tensor(self.hidden_state[idx].xyz[i]) for i in range(self.hidden_state[idx].n_frames)] + else: + graph.hidden_state = [] + if self.transform: graph = self.transform(graph) return graph diff --git a/src/jamun/data/_subsample.py b/src/jamun/data/_subsample.py new file mode 100644 index 0000000..635d04c --- /dev/null +++ b/src/jamun/data/_subsample.py @@ -0,0 +1,87 @@ +import numpy as np +from typing import Tuple, List + + +def get_subsampled_indices( + N: int, + subsample_rate: int, + total_lag_time: int, + lag_subsample_rate: int, +) -> List[np.ndarray]: + """ + Generate subsampled indices and their corresponding lagged indices. + + Args: + N: Total number of frames + subsample_rate: Rate at which to subsample the frames + total_lag_time: Number of lagged frames to generate for each subsampled frame + lag_subsample_rate: Rate at which to subsample the lagged frames + + Returns: + List of arrays, where each array contains the lagged indices for a subsampled frame + + Raises: + ValueError: If the input parameters don't satisfy the required constraints + """ + # Check guardrails + if N / subsample_rate < 1: + raise ValueError(f"Number of samples (N/subsample_rate = {N/subsample_rate}) must be >= 1") + + # if total_lag_time * lag_subsample_rate > subsample_rate: + # raise ValueError( + # f"total_lag_time * lag_subsample_rate ({total_lag_time * lag_subsample_rate}) " + # f"must be <= subsample_rate ({subsample_rate})" + # ) + + # Generate subsampled indices + subsampled_indices = np.arange(0, N, subsample_rate) + + # Generate lagged indices for each subsampled index + lagged_indices = [] + for idx in subsampled_indices: + # Calculate lagged indices + lagged = [int(idx - j * lag_subsample_rate) for j in range(total_lag_time)] + + # Check if we have enough lagged indices + if len(lagged) == total_lag_time and all(x >= 0 for x in lagged): + lagged_indices.append(lagged) + + return lagged_indices + + +def get_subsampled_trajectory( + positions: np.ndarray, + subsample_rate: int, + total_lag_time: int, + lag_subsample_rate: int, +) -> Tuple[np.ndarray, List[np.ndarray]]: + """ + Subsample a trajectory and generate lagged states for each subsampled frame. + + Args: + positions: Array of shape (N, ...) containing trajectory positions + subsample_rate: Rate at which to subsample the frames + total_lag_time: Number of lagged frames to generate for each subsampled frame + lag_subsample_rate: Rate at which to subsample the lagged frames + + Returns: + Tuple containing: + - subsampled_positions: Array of subsampled positions + - lagged_positions: List of arrays, where each array contains the lagged positions + for the corresponding subsampled frame + + Raises: + ValueError: If the input parameters don't satisfy the required constraints + """ + N = len(positions) + + # Get the lagged indices + lagged_indices = get_subsampled_indices(N, subsample_rate, total_lag_time, lag_subsample_rate) + + # Extract subsampled positions (first element of each lagged indices list) + subsampled_positions = np.array([positions[indices[0]] for indices in lagged_indices]) + + # Generate lagged positions for each subsampled frame + lagged_positions = [ [positions[idx] for idx in indices[1:]] for indices in lagged_indices] + + return subsampled_positions, lagged_positions \ No newline at end of file diff --git a/src/jamun/data/tests/test_subsample.py b/src/jamun/data/tests/test_subsample.py new file mode 100644 index 0000000..23b3248 --- /dev/null +++ b/src/jamun/data/tests/test_subsample.py @@ -0,0 +1,229 @@ +import numpy as np + +from jamun.data._subsample import get_subsampled_indices, get_subsampled_trajectory + + +def test_basic_functionality(): + """Test basic functionality with valid inputs.""" + print("\nTesting basic functionality...") + N = 100 + subsample_rate = 10 + total_lag_time = 3 + lag_subsample_rate = 10 + + print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + + breakpoint() # Debug point 1: Check input parameters before function call + + lagged_indices = get_subsampled_indices( + N, subsample_rate, total_lag_time, lag_subsample_rate + ) + + breakpoint() # Debug point 2: Check function output + + # Extract subsampled indices (first element of each list) + subsampled_indices = np.array([indices[0] for indices in lagged_indices]) + print(f"Subsampled indices: {subsampled_indices}") + print(f"Number of lagged indices lists: {len(lagged_indices)}") + + # Check subsampled indices + expected_subsampled = np.array([20, 30, 40, 50, 60, 70, 80, 90]) + assert np.array_equal(subsampled_indices, expected_subsampled), \ + f"Expected {expected_subsampled}, got {subsampled_indices}" + + # Check lagged indices + for i, lagged in enumerate(lagged_indices): + expected_lagged = np.array([ + subsampled_indices[i], + subsampled_indices[i] - lag_subsample_rate, + subsampled_indices[i] - 2 * lag_subsample_rate + ]) + assert np.array_equal(lagged, expected_lagged), \ + f"For index {i}, expected {expected_lagged}, got {lagged}" + + print("Basic functionality test passed!") + + +def test_edge_cases(): + """Test edge cases and boundary conditions.""" + print("\nTesting edge cases...") + N = 10 + subsample_rate = 10 + total_lag_time = 1 + lag_subsample_rate = 1 + + print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + + breakpoint() # Debug point 5: Check edge case parameters before function call + + lagged_indices = get_subsampled_indices( + N, subsample_rate, total_lag_time, lag_subsample_rate + ) + + breakpoint() # Debug point 6: Check edge case results + + # Extract subsampled indices (first element of each list) + subsampled_indices = np.array([indices[0] for indices in lagged_indices]) + print(f"Subsampled indices: {subsampled_indices}") + print(f"Lagged indices: {lagged_indices}") + + assert len(subsampled_indices) == 1, f"Expected 1 subsampled index, got {len(subsampled_indices)}" + assert len(lagged_indices) == 1, f"Expected 1 lagged indices list, got {len(lagged_indices)}" + assert np.array_equal(subsampled_indices, np.array([0])), \ + f"Expected [0], got {subsampled_indices}" + assert np.array_equal(lagged_indices[0], np.array([0])), \ + f"Expected [0], got {lagged_indices[0]}" + + print("Edge cases test passed!") + + +def test_lagged_indices_filtering(): + """Test that lagged indices are properly filtered when they would go negative.""" + print("\nTesting lagged indices filtering...") + N = 20 + subsample_rate = 5 + total_lag_time = 3 + lag_subsample_rate = 3 + + print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + + breakpoint() # Debug point 7: Check filtering parameters before function call + + lagged_indices = get_subsampled_indices( + N, subsample_rate, total_lag_time, lag_subsample_rate + ) + + breakpoint() # Debug point 8: Check filtering results + + # Extract subsampled indices (first element of each list) + subsampled_indices = np.array([indices[0] for indices in lagged_indices]) + print(f"Subsampled indices: {subsampled_indices}") + print(f"Number of lagged indices lists: {len(lagged_indices)}") + + expected_subsampled = np.array([5, 10, 15]) + assert np.array_equal(subsampled_indices, expected_subsampled), \ + f"Expected {expected_subsampled}, got {subsampled_indices}" + + assert len(lagged_indices) == len(subsampled_indices), \ + f"Expected {len(subsampled_indices)} lagged indices lists, got {len(lagged_indices)}" + + expected_first_lagged = np.array([5, 2, -1]) + assert not any(np.array_equal(lagged, expected_first_lagged) for lagged in lagged_indices), \ + "Found unexpected lagged indices that should have been filtered out" + + print("Lagged indices filtering test passed!") + + +def test_large_numbers(): + """Test with larger numbers to ensure scalability.""" + print("\nTesting large numbers...") + N = 10000 + subsample_rate = 100 + total_lag_time = 5 + lag_subsample_rate = 10 + + print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + + breakpoint() # Debug point 9: Check large number parameters before function call + + lagged_indices = get_subsampled_indices( + N, subsample_rate, total_lag_time, lag_subsample_rate + ) + + breakpoint() # Debug point 10: Check large number results + + # Extract subsampled indices (first element of each list) + subsampled_indices = np.array([indices[0] for indices in lagged_indices]) + print(f"Number of subsampled indices: {len(subsampled_indices)}") + print(f"Number of lagged indices lists: {len(lagged_indices)}") + + assert len(subsampled_indices) == N // subsample_rate, \ + f"Expected {N // subsample_rate} subsampled indices, got {len(subsampled_indices)}" + + for i, lagged in enumerate(lagged_indices): + print(f"\nChecking lagged indices list {i}:") + print(f"Lagged indices: {lagged}") + print(f"Type of lagged indices: {type(lagged)}") + print(f"Individual values and their types:") + for j, val in enumerate(lagged): + print(f" Index {j}: value={val}, type={type(val)}") + + assert len(lagged) == total_lag_time, \ + f"Expected lagged indices length {total_lag_time}, got {len(lagged)}" + + # Check each value individually + for j, val in enumerate(lagged): + assert isinstance(val, (int, np.integer)), \ + f"Value at index {j} is not an integer: {val} (type: {type(val)})" + assert val >= 0, \ + f"Found negative value at index {j}: {val}" + + print("Large numbers test passed!") + + +def test_trajectory_subsampling(): + """Test subsampling of trajectory positions.""" + print("\nTesting trajectory subsampling...") + + # Create a random trajectory with 100 frames, 10 particles, and 3 coordinates + N = 100 + np.random.seed(42) # For reproducibility + positions = np.random.randn(N, 10, 3) + + subsample_rate = 10 + total_lag_time = 3 + lag_subsample_rate = 10 + + print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + + breakpoint() # Debug point 11: Check trajectory parameters + + subsampled_positions, lagged_positions = get_subsampled_trajectory( + positions, subsample_rate, total_lag_time, lag_subsample_rate + ) + + breakpoint() # Debug point 12: Check trajectory results + + print(f"Original positions shape: {positions.shape}") + print(f"Subsampled positions shape: {subsampled_positions.shape}") + print(f"Number of lagged position lists: {len(lagged_positions)}") + + # Check shapes + expected_num_subsampled = (N - 20) // subsample_rate # Starting from index 20 + assert subsampled_positions.shape[0] == expected_num_subsampled, \ + f"Expected {expected_num_subsampled} subsampled positions, got {subsampled_positions.shape[0]}" + assert subsampled_positions.shape[1:] == (10, 3), \ + f"Expected subsampled positions to have shape (N, 10, 3), got {subsampled_positions.shape}" + + # Check values + for i in range(expected_num_subsampled): + # Check subsampled position + expected_sub_pos = positions[20 + i*subsample_rate] + assert np.array_equal(subsampled_positions[i], expected_sub_pos), \ + f"For index {i}, expected subsampled position {expected_sub_pos}, got {subsampled_positions[i]}" + + # Check lagged positions + assert len(lagged_positions[i]) == total_lag_time, \ + f"For index {i}, expected {total_lag_time} lagged positions, got {len(lagged_positions[i])}" + + for j, lag_pos in enumerate(lagged_positions[i]): + expected_lag_pos = positions[20 + i*subsample_rate - j*lag_subsample_rate] + assert np.array_equal(lag_pos, expected_lag_pos), \ + f"For index {i}, lag {j}, expected position {expected_lag_pos}, got {lag_pos}" + + print("Trajectory subsampling test passed!") + + +if __name__ == "__main__": + print("Starting tests...") + test_basic_functionality() + # test_edge_cases() + # test_lagged_indices_filtering() + # test_large_numbers() + test_trajectory_subsampling() + print("\nAll tests completed!") \ No newline at end of file diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml new file mode 100644 index 0000000..c53a440 --- /dev/null +++ b/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml @@ -0,0 +1,31 @@ +_target_: jamun.model.arch.E3ConvConditional +_partial_: true +irreps_out: "1x1e" +irreps_hidden: "120x0e + 32x1e" +irreps_sh: "1x0e + 1x1e" +n_layers: 5 +edge_attr_dim: 64 +atom_type_embedding_dim: 8 +atom_code_embedding_dim: 8 +residue_code_embedding_dim: 32 +residue_index_embedding_dim: 8 +use_residue_information: ${data.use_residue_information} +use_residue_sequence_index: false +num_atom_types: 20 +max_sequence_length: 10 +num_atom_codes: 10 +num_residue_types: 25 +hidden_layer_factory: + _target_: e3tools.nn.ConvBlock + _partial_: true + conv: + _target_: e3tools.nn.Conv + _partial_: true +output_head_factory: + _target_: e3tools.nn.EquivariantMLP + _partial_: true + irreps_hidden_list: + - ${model.arch.irreps_hidden} +test_equivariance: false +reduce: null +N_structures: 2 \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_conditional.yaml b/src/jamun/hydra_config/model/denoiser_conditional.yaml new file mode 100644 index 0000000..386c72c --- /dev/null +++ b/src/jamun/hydra_config/model/denoiser_conditional.yaml @@ -0,0 +1,29 @@ +defaults: + - arch: e3conv_conditional.yaml + - optim: adam.yaml + - lr_scheduler_config: null + - _self_ + +max_radius: null +average_squared_distance: null +add_fixed_noise: false +add_fixed_ones: false +align_noisy_input_during_training: true +align_noisy_input_during_evaluation: true +mean_center: true +mirror_augmentation_rate: 0.0 +use_torch_compile: true +torch_compile_kwargs: + fullgraph: true + dynamic: true + mode: default + +conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + +multimeasurement: false +N_measurements_hidden: 1 +N_measurements: 1 + +_target_: jamun.model.denoiser_conditional.Denoiser \ No newline at end of file diff --git a/src/jamun/model/__init__.py b/src/jamun/model/__init__.py index de26ff3..c31fabc 100644 --- a/src/jamun/model/__init__.py +++ b/src/jamun/model/__init__.py @@ -1,2 +1,2 @@ from .denoiser import Denoiser -from .energy import EnergyModel +from .energy import EnergyModel \ No newline at end of file diff --git a/src/jamun/model/arch/__init__.py b/src/jamun/model/arch/__init__.py index 5675314..15b4e35 100644 --- a/src/jamun/model/arch/__init__.py +++ b/src/jamun/model/arch/__init__.py @@ -1,2 +1,3 @@ from .e3conv import E3Conv from .ophiuchus import Ophiuchus +from .e3conv_conditional import E3ConvConditional \ No newline at end of file diff --git a/scratch/e3conv_test.py b/src/jamun/model/arch/e3conv_conditional.py similarity index 99% rename from scratch/e3conv_test.py rename to src/jamun/model/arch/e3conv_conditional.py index 6cf9721..6365001 100644 --- a/scratch/e3conv_test.py +++ b/src/jamun/model/arch/e3conv_conditional.py @@ -11,7 +11,7 @@ from jamun.model.noise_conditioning import NoiseConditionalScaling, NoiseConditionalSkipConnection -class E3Conv(torch.nn.Module): +class E3ConvConditional(torch.nn.Module): """A simple E(3)-equivariant convolutional neural network, similar to NequIP.""" def __init__( diff --git a/src/jamun/model/conditioners/__init__.py b/src/jamun/model/conditioners/__init__.py new file mode 100644 index 0000000..768bbac --- /dev/null +++ b/src/jamun/model/conditioners/__init__.py @@ -0,0 +1,2 @@ +from .conditioners import Conditioner, PositionConditioner, SelfConditioner + diff --git a/scratch/conditioners.py b/src/jamun/model/conditioners/conditioners.py similarity index 68% rename from scratch/conditioners.py rename to src/jamun/model/conditioners/conditioners.py index 9dafc01..56baedc 100644 --- a/scratch/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -10,15 +10,21 @@ from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm - +class Conditioner(pl.LightningModule): + """ + Base class for conditioners. + """ + def __init__(self, N_structures: int, **kwargs): + super().__init__() + self.N_structures = N_structures class PositionConditioner(pl.LightningModule): """ Condition the hidden state on the position of the structure. """ - def __init__(self, **kwargs): + def __init__(self, N_structures: int, **kwargs): super().__init__() - + self.N_structures = N_structures def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures = [] for positions in y.hidden_state: @@ -30,10 +36,9 @@ class SelfConditioner(pl.LightningModule): """ No conditioning, but add the position of the structure to itself to make it compatible with the denoiser. """ - def __init__(self, **kwargs): + def __init__(self, N_structures: int, **kwargs): super().__init__() - + self.N_structures = N_structures def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: - conditioned_structures = [] - conditioned_structures.append(y.pos) - return conditioned_structures \ No newline at end of file + conditioned_structures = [y.pos for _ in range(self.N_structures-1)] + return conditioned_structures \ No newline at end of file diff --git a/scratch/denoiser_test.py b/src/jamun/model/denoiser_conditional.py similarity index 79% rename from scratch/denoiser_test.py rename to src/jamun/model/denoiser_conditional.py index ab93938..7b900eb 100644 --- a/scratch/denoiser_test.py +++ b/src/jamun/model/denoiser_conditional.py @@ -12,7 +12,7 @@ class Denoiser(pl.LightningModule): - """The main denoiser model.""" + """The main denoiser mode with conditional architecture.""" def __init__( self, @@ -33,7 +33,10 @@ def __init__( lr_scheduler_config: Optional[Dict] = None, use_torch_compile: bool = True, torch_compile_kwargs: Optional[Dict] = None, - conditioner: Callable[..., list[torch.Tensor]] = None + conditioner: Callable[..., list[torch.Tensor]] = None, + multimeasurement: bool = False, + N_measurements_hidden: int = 1, + N_measurements: int = 1, ): super().__init__() self.save_hyperparameters(logger=False) @@ -104,6 +107,10 @@ def __init__( raise ValueError("Conditioner must be a callable or None") py_logger.info(f"Conditioner: {self.conditioning_module}") + self.multimeasurement = multimeasurement + self.N_measurements_hidden = N_measurements_hidden + self.N_measurements = N_measurements + def conditioner_default(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures = [] # for positions in y.hidden_state: @@ -124,7 +131,7 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.pos.ndim) - y = x.clone("pos") + y = x.clone() if self.add_fixed_ones: noise = torch.ones_like(x.pos) hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] @@ -149,6 +156,54 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten y.pos = -y.pos return y + def add_noise_hiddens( + self, + x: torch_geometric.data.Batch, + N_measurements_hidden: int, + N_measurements: int, + sigma: Union[float, torch.Tensor], + ) -> torch_geometric.data.Batch: + """ + Makes N_measurements_hidden number of noisy copies of the hidden states of x + and then for every noisy copy, makes N_measurements number of noisy copies of the positions of x. + + Args: + x (Batch): A torch_geometric Batch object. Must have `pos` and `hidden_state` attributes. + `hidden_state` is expected to be a list of tensors. + N_measurements_hidden (int): Number of noisy copies of hidden states. + N_measurements (int): Number of noisy copies of positions for each noisy hidden state. + sigma (float or torch.Tensor): The standard deviation of the Gaussian noise to add. + + Returns: + Batch: A new Batch object containing all the noisy copies. + """ + x_list = x.to_data_list() + noisy_y_list = [] + + for graph in x_list: + for _ in range(N_measurements_hidden): + # Create a noisy version of the hidden state + noisy_hidden_state = [] + if hasattr(graph, "hidden_state") and graph.hidden_state is not None: + for hs_tensor in graph.hidden_state: + noise = torch.randn_like(hs_tensor) * sigma + noisy_hidden_state.append(hs_tensor + noise) + + for _ in range(N_measurements): + noisy_graph = graph.clone() + + # Add noise to positions + pos_noise = torch.randn_like(graph.pos) * sigma + noisy_graph.pos = graph.pos + pos_noise + + # Assign the noisy hidden state + if hasattr(graph, "hidden_state") and graph.hidden_state is not None: + noisy_graph.hidden_state = [hs.clone() for hs in noisy_hidden_state] + + noisy_y_list.append(noisy_graph) + + return torch_geometric.data.Batch.from_data_list(noisy_y_list) + def score(self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: """Compute the score function.""" sigma = torch.as_tensor(sigma).to(y.pos) @@ -242,7 +297,7 @@ def xhat_normalized( y = self.add_edges(y, radial_cutoff) with torch.cuda.nvtx.range("scale_y"): - y_scaled = y.clone("pos") + y_scaled = y.clone() y_scaled.pos = y.pos * c_in scaled_hidden_state = [] for positions in y_scaled.hidden_state: @@ -250,7 +305,7 @@ def xhat_normalized( y_scaled.hidden_state = scaled_hidden_state with torch.cuda.nvtx.range("clone_y"): - xhat = y.clone("pos") + xhat = y.clone() with torch.cuda.nvtx.range("conditioning"): conditioned_structures = self.conditioner(y_scaled) @@ -283,17 +338,41 @@ def noise_and_denoise( x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor], align_noisy_input: bool, - ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch]: - """Add noise to the input and denoise it.""" + ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: + """ + Add noise to the input and denoise it. + Returns the target for the loss, the prediction, and the noisy input. + """ with torch.no_grad(): if self.mean_center: - with torch.cuda.nvtx.range("mean_center_x"): - x = mean_center(x) - - sigma = torch.as_tensor(sigma).to(x.pos) + # Operate on a clone to avoid side effects on the original batch object. + x_processed = mean_center(x) + else: + x_processed = x + + sigma = torch.as_tensor(sigma).to(x_processed.pos) + + if self.multimeasurement: + with torch.cuda.nvtx.range("add_noise_hiddens"): + y = self.add_noise_hiddens( + x_processed, self.N_measurements_hidden, self.N_measurements, sigma + ) + + # Repeat x_processed to match y's batch size for alignment and loss calculation. + x_list = x_processed.to_data_list() + repeated_x_list = [ + graph.clone() + for graph in x_list + for _ in range(self.N_measurements_hidden * self.N_measurements) + ] + x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to( + x_processed.pos.device + ) - with torch.cuda.nvtx.range("add_noise"): - y = self.add_noise(x, sigma) + else: + with torch.cuda.nvtx.range("add_noise"): + y = self.add_noise(x_processed, sigma) + x_target = x_processed.clone() if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): @@ -302,12 +381,12 @@ def noise_and_denoise( # Aligning each batch. if align_noisy_input: with torch.cuda.nvtx.range("align_A_to_B_batched"): - y = align_A_to_B_batched(y, x) + y = align_A_to_B_batched(y, x_target) with torch.cuda.nvtx.range("xhat"): xhat = self.xhat(y, sigma) - return xhat, y + return x_target, xhat, y def compute_loss( self, @@ -353,8 +432,8 @@ def noise_and_compute_loss( align_noisy_input: bool, ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Add noise to the input and compute the loss.""" - xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) - return self.compute_loss(x, xhat, sigma) + x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) + return self.compute_loss(x_target, xhat, sigma) def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): """Called during training.""" diff --git a/src/jamun/model/noise_test.py b/src/jamun/model/noise_test.py new file mode 100644 index 0000000..5c316b4 --- /dev/null +++ b/src/jamun/model/noise_test.py @@ -0,0 +1,110 @@ +import torch +import torch_geometric +from torch_geometric.data import Data, Batch +from typing import List + +def add_noise_hiddens(x: Batch, N_measurements_hidden: int, N_measurements: int, sigma: float) -> Batch: + """ + Makes N_measurements_hidden number of noisy copies of the hidden states of x + and then for every noisy copy, makes N_measurements number of noisy copies of the positions of x. + + Args: + x (Batch): A torch_geometric Batch object. Must have `pos` and `hidden_state` attributes. + `hidden_state` is expected to be a list of tensors. + N_measurements_hidden (int): Number of noisy copies of hidden states. + N_measurements (int): Number of noisy copies of positions for each noisy hidden state. + sigma (float): The standard deviation of the Gaussian noise to add. + + Returns: + Batch: A new Batch object containing all the noisy copies. + """ + x_list = x.to_data_list() + noisy_y_list = [] + + for graph in x_list: + for _ in range(N_measurements_hidden): + # Create a noisy version of the hidden state + noisy_hidden_state = [] + if hasattr(graph, 'hidden_state') and graph.hidden_state is not None: + for hs_tensor in graph.hidden_state: + noise = torch.randn_like(hs_tensor) * sigma + noisy_hidden_state.append(hs_tensor + noise) + + for _ in range(N_measurements): + noisy_graph = graph.clone() + + # Add noise to positions + pos_noise = torch.randn_like(graph.pos) * sigma + noisy_graph.pos = graph.pos + pos_noise + + # Assign the noisy hidden state + if hasattr(graph, 'hidden_state') and graph.hidden_state is not None: + noisy_graph.hidden_state = [hs.clone() for hs in noisy_hidden_state] + + noisy_y_list.append(noisy_graph) + + return Batch.from_data_list(noisy_y_list) + + +# --- Testing Script --- +def run_test(): + print("Running test for add_noise_hiddens...") + + # 1. Create dummy data + num_nodes = 5 + # single data object + data1 = Data( + pos=torch.randn(num_nodes, 3), + hidden_state=[torch.randn(num_nodes, 4), torch.randn(num_nodes, 8)] + ) + # another data object + data2 = Data( + pos=torch.randn(num_nodes + 2, 3), + hidden_state=[torch.randn(num_nodes + 2, 4), torch.randn(num_nodes + 2, 8)] + ) + + original_batch = Batch.from_data_list([data1, data2]) + + # 2. Set parameters + N_measurements_hidden = 2 + N_measurements = 3 + sigma = 0.1 + + # 3. Call the function + noisy_batch = add_noise_hiddens(original_batch, N_measurements_hidden, N_measurements, sigma) + + # 4. Assertions + # Check total number of graphs + expected_num_graphs = original_batch.num_graphs * N_measurements_hidden * N_measurements + assert noisy_batch.num_graphs == expected_num_graphs, \ + f"Expected {expected_num_graphs} graphs, but got {noisy_batch.num_graphs}" + print(f"Correct number of graphs in output batch: {noisy_batch.num_graphs}") + + noisy_graphs = noisy_batch.to_data_list() + + # Check that noise was added + assert not torch.allclose(noisy_graphs[0].pos, data1.pos) + assert not torch.allclose(noisy_graphs[0].hidden_state[0], data1.hidden_state[0]) + print("Noise was added to pos and hidden_state.") + + # Check hidden state logic + # The first N_measurements graphs (for the first original graph) should have the same hidden state + first_hidden_state_set = noisy_graphs[0].hidden_state + for i in range(1, N_measurements): + assert torch.allclose(noisy_graphs[i].hidden_state[0], first_hidden_state_set[0]) + assert torch.allclose(noisy_graphs[i].hidden_state[1], first_hidden_state_set[1]) + + # But their positions should be different + assert not torch.allclose(noisy_graphs[0].pos, noisy_graphs[1].pos) + + # The (N_measurements+1)-th graph should have a different hidden state (from the second hidden measurement) + next_hidden_state_set = noisy_graphs[N_measurements].hidden_state + assert not torch.allclose(next_hidden_state_set[0], first_hidden_state_set[0]) + + print("Hidden state noise logic seems correct.") + + print("Test passed!") + + +if __name__ == "__main__": + run_test() \ No newline at end of file diff --git a/wandb/sweep-01dho9mz/config-o4x6y3nm.yaml b/wandb/sweep-01dho9mz/config-o4x6y3nm.yaml new file mode 100644 index 0000000..d1b968c --- /dev/null +++ b/wandb/sweep-01dho9mz/config-o4x6y3nm.yaml @@ -0,0 +1,6 @@ +wandb_version: 1 + +model.conditioner: + value: + N_structures: ${model.arch.N_structures} + _target_: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-01dho9mz/config-wrqb60uz.yaml b/wandb/sweep-01dho9mz/config-wrqb60uz.yaml new file mode 100644 index 0000000..dc44beb --- /dev/null +++ b/wandb/sweep-01dho9mz/config-wrqb60uz.yaml @@ -0,0 +1,6 @@ +wandb_version: 1 + +model.conditioner: + value: + N_structures: ${model.arch.N_structures} + _target_: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-5tie5khj/config-lhzhrmk7.yaml b/wandb/sweep-5tie5khj/config-lhzhrmk7.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-5tie5khj/config-lhzhrmk7.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-gze9gptu/config-sxlo5fvy.yaml b/wandb/sweep-gze9gptu/config-sxlo5fvy.yaml new file mode 100644 index 0000000..d1b968c --- /dev/null +++ b/wandb/sweep-gze9gptu/config-sxlo5fvy.yaml @@ -0,0 +1,6 @@ +wandb_version: 1 + +model.conditioner: + value: + N_structures: ${model.arch.N_structures} + _target_: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-gze9gptu/config-vfd1ntbz.yaml b/wandb/sweep-gze9gptu/config-vfd1ntbz.yaml new file mode 100644 index 0000000..dc44beb --- /dev/null +++ b/wandb/sweep-gze9gptu/config-vfd1ntbz.yaml @@ -0,0 +1,6 @@ +wandb_version: 1 + +model.conditioner: + value: + N_structures: ${model.arch.N_structures} + _target_: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml b/wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml new file mode 100644 index 0000000..dc44beb --- /dev/null +++ b/wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml @@ -0,0 +1,6 @@ +wandb_version: 1 + +model.conditioner: + value: + N_structures: ${model.arch.N_structures} + _target_: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-rxcjaig2/config-zvdgfxss.yaml b/wandb/sweep-rxcjaig2/config-zvdgfxss.yaml new file mode 100644 index 0000000..d1b968c --- /dev/null +++ b/wandb/sweep-rxcjaig2/config-zvdgfxss.yaml @@ -0,0 +1,6 @@ +wandb_version: 1 + +model.conditioner: + value: + N_structures: ${model.arch.N_structures} + _target_: jamun.model.conditioners.PositionConditioner From a51de44e384d391dc1706f656e1535f0c7056a89 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Wed, 25 Jun 2025 15:57:30 +0000 Subject: [PATCH 09/32] multimeasurement loss added and tested with sweeps --- .../train_test_single_shape_conditional.yaml | 6 +- ...est_single_shape_conditional_one_traj.yaml | 80 +++++++++++++++++++ configs/sweep.yaml | 3 +- scratch/test_conditional.py | 4 +- scripts/slurm/sweep.sh | 7 +- src/jamun/cmdline/train.py | 17 +++- src/jamun/data/_mdtraj.py | 1 + wandb/sweep-199gq3m7/config-iw7qu9pc.yaml | 4 + wandb/sweep-199gq3m7/config-xn0ylvd3.yaml | 4 + wandb/sweep-2b4ndoa2/config-29fpum8f.yaml | 4 + wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml | 4 + wandb/sweep-2chv306p/config-dfa7a6n1.yaml | 4 + wandb/sweep-2chv306p/config-e15fh1bu.yaml | 4 + wandb/sweep-3945fdiz/config-bj96w41e.yaml | 4 + wandb/sweep-3945fdiz/config-ioayvefo.yaml | 4 + wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml | 4 + wandb/sweep-4y3t4lnj/config-irjpk449.yaml | 4 + wandb/sweep-5tie5khj/config-lzq16xjh.yaml | 4 + wandb/sweep-7h8tmiyj/config-14ueynti.yaml | 4 + wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml | 4 + wandb/sweep-94u6xhkc/config-q8kcodq4.yaml | 4 + wandb/sweep-94u6xhkc/config-zrd7urfq.yaml | 4 + wandb/sweep-g101608w/config-htlax8nq.yaml | 4 + wandb/sweep-g101608w/config-pp1wtj1t.yaml | 4 + wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml | 4 + wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml | 4 + wandb/sweep-mebia8e6/config-0t2gzjry.yaml | 4 + wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml | 4 + wandb/sweep-x0n6d1ps/config-r1havcvy.yaml | 4 + wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml | 4 + wandb/sweep-z467ihn2/config-r1jppjka.yaml | 4 + 31 files changed, 204 insertions(+), 10 deletions(-) create mode 100644 configs/experiment/train_test_single_shape_conditional_one_traj.yaml create mode 100644 wandb/sweep-199gq3m7/config-iw7qu9pc.yaml create mode 100644 wandb/sweep-199gq3m7/config-xn0ylvd3.yaml create mode 100644 wandb/sweep-2b4ndoa2/config-29fpum8f.yaml create mode 100644 wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml create mode 100644 wandb/sweep-2chv306p/config-dfa7a6n1.yaml create mode 100644 wandb/sweep-2chv306p/config-e15fh1bu.yaml create mode 100644 wandb/sweep-3945fdiz/config-bj96w41e.yaml create mode 100644 wandb/sweep-3945fdiz/config-ioayvefo.yaml create mode 100644 wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml create mode 100644 wandb/sweep-4y3t4lnj/config-irjpk449.yaml create mode 100644 wandb/sweep-5tie5khj/config-lzq16xjh.yaml create mode 100644 wandb/sweep-7h8tmiyj/config-14ueynti.yaml create mode 100644 wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml create mode 100644 wandb/sweep-94u6xhkc/config-q8kcodq4.yaml create mode 100644 wandb/sweep-94u6xhkc/config-zrd7urfq.yaml create mode 100644 wandb/sweep-g101608w/config-htlax8nq.yaml create mode 100644 wandb/sweep-g101608w/config-pp1wtj1t.yaml create mode 100644 wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml create mode 100644 wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml create mode 100644 wandb/sweep-mebia8e6/config-0t2gzjry.yaml create mode 100644 wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml create mode 100644 wandb/sweep-x0n6d1ps/config-r1havcvy.yaml create mode 100644 wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml create mode 100644 wandb/sweep-z467ihn2/config-r1jppjka.yaml diff --git a/configs/experiment/train_test_single_shape_conditional.yaml b/configs/experiment/train_test_single_shape_conditional.yaml index 519a6bd..c922f4a 100644 --- a/configs/experiment/train_test_single_shape_conditional.yaml +++ b/configs/experiment/train_test_single_shape_conditional.yaml @@ -9,8 +9,8 @@ defaults: data: datamodule: - num_workers: 8 batch_size: 32 + num_workers: 2 datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -62,10 +62,10 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 1 + max_epochs: 1000 logger: wandb: group: ALA_ALA, capped diamines, conditional denoiser - notes: "Running on ALA_ALA capped diamine, self-conditioned denoiser" \ No newline at end of file + notes: "Running on ALA_ALA capped diamine, conditional denoiser" \ No newline at end of file diff --git a/configs/experiment/train_test_single_shape_conditional_one_traj.yaml b/configs/experiment/train_test_single_shape_conditional_one_traj.yaml new file mode 100644 index 0000000..9b3dd90 --- /dev/null +++ b/configs/experiment/train_test_single_shape_conditional_one_traj.yaml @@ -0,0 +1,80 @@ +# @package _global_ + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 1 + num_workers: 2 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 10 + total_lag_time: 5 + lag_subsample_rate: 1 + start_frame: 0 + num_frames: 11 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 10 + total_lag_time: 5 + lag_subsample_rate: 1 + start_frame: 1 + num_frames: 11 + + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 10 + total_lag_time: 5 + lag_subsample_rate: 1 + start_frame: 2 + num_frames: 11 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + multimeasurement: true + N_measurements: 100 + N_measurements_hidden: 1 + +trainer: + val_check_interval: 0.5 + max_epochs: 1000 + + +logger: + wandb: + group: ALA_ALA, capped diamines, conditional denoiser + notes: "Running on ALA_ALA capped diamine, conditional denoiser" \ No newline at end of file diff --git a/configs/sweep.yaml b/configs/sweep.yaml index e264c3f..f86f9b3 100644 --- a/configs/sweep.yaml +++ b/configs/sweep.yaml @@ -14,5 +14,6 @@ parameters: command: - ${program} - "--config-dir=configs" - - "experiment=train_test_single_shape_conditional" + - "experiment=train_test_single_shape_conditional_one_traj" + - "++trainer.log_every_n_steps=10" - ${args_no_hyphens} \ No newline at end of file diff --git a/scratch/test_conditional.py b/scratch/test_conditional.py index badc01f..d34c4d0 100644 --- a/scratch/test_conditional.py +++ b/scratch/test_conditional.py @@ -72,7 +72,7 @@ def main(cfg): print("Testing forward pass...") with torch.no_grad(): sigma = model.sigma_distribution.sample() - _, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) print(f"Input shape: {batch.pos.shape}") print(f"Noisy shape: {y.pos.shape}") @@ -80,7 +80,7 @@ def main(cfg): # Test loss computation print("Testing loss computation...") - loss, aux = model.compute_loss(batch, xhat, sigma) + loss, aux = model.compute_loss(x_target, xhat, sigma) print(f"Loss: {loss.mean().item():.4f}") print(f"Metrics: {aux}") diff --git a/scripts/slurm/sweep.sh b/scripts/slurm/sweep.sh index 74c167e..5b15c4f 100644 --- a/scripts/slurm/sweep.sh +++ b/scripts/slurm/sweep.sh @@ -1,9 +1,10 @@ #!/usr/bin/env bash #SBATCH --partition gpu3 +#SBATCH --qos=preempt #SBATCH --nodes 1 -#SBATCH --ntasks-per-node=4 # Number of agents to run in parallel on this node -#SBATCH --gpus-per-task=1 # Assign one GPU to each agent +#SBATCH --ntasks-per-node=2 # Number of agents to run in parallel on this node +#SBATCH --gpus-per-node=2 # Assign one GPU to each agent #SBATCH --cpus-per-task=8 #SBATCH --time 3-0 #SBATCH --mem-per-cpu=32G @@ -30,4 +31,4 @@ echo "Starting ${SLURM_NTASKS} agents for sweep: ${SWEEP_ID}" # Launch multiple wandb agents in parallel using srun. # Each agent will poll the sweep server, get a configuration, and run one training job. # PyTorch Lightning will automatically use the single GPU assigned by Slurm to each task. -srun wandb agent "${SWEEP_ID}" \ No newline at end of file +wandb agent --count 1 "${SWEEP_ID}" \ No newline at end of file diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 32b5e4f..64b7d74 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -39,10 +39,11 @@ def run(cfg): dist_log(f"{OmegaConf.to_yaml(log_cfg)}") dist_log(f"{os.getcwd()=}") dist_log(f"{torch.__config__.parallel_info()}") + dist_log(f"CUDA_VISIBLE_DEVICES: {os.environ.get('CUDA_VISIBLE_DEVICES')}") dist_log(f"{os.sched_getaffinity(0)=}") # Set the start method to spawn to avoid issues with the default fork method. - # torch.multiprocessing.set_start_method("spawn", force=True) + torch.multiprocessing.set_start_method("spawn", force=True) # Compute data normalization. if cfg.get("compute_average_squared_distance_from_data"): @@ -58,6 +59,20 @@ def run(cfg): dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") torch.set_float32_matmul_precision(matmul_prec) + # # If running under Slurm, ensure the number of devices matches the allocation. + # if "SLURM_GPUS_PER_TASK" in os.environ and torch.cuda.is_available(): + # dist_log(f"torch.cuda.device_count(): {torch.cuda.device_count()}") + # try: + # num_gpus = int(os.environ["SLURM_GPUS_PER_TASK"]) + # dist_log(f"Slurm-allocated GPUs per task: {num_gpus}") + # # Explicitly create a list of device IDs [0, 1, ..., n-1] for Lightning. + # device_ids = list(range(num_gpus)) + # # This will override any value from the config file, ensuring it matches the Slurm allocation. + # cfg.trainer.devices = device_ids + # dist_log(f"Explicitly set cfg.trainer.devices to {cfg.trainer.devices}") + # except (ValueError, KeyError): + # dist_log("Could not parse or find SLURM_GPUS_PER_TASK.") + loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) wandb_logger = None for logger in loggers: diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index 45a8375..4a3638d 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -218,6 +218,7 @@ def __init__( # Get lagged indices if lag parameters are provided if total_lag_time is not None and lag_subsample_rate is not None: + self.traj = self.traj[start_frame : start_frame + num_frames] # accommodate for start_frame and num_frames lagged_indices = get_subsampled_indices( self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate ) diff --git a/wandb/sweep-199gq3m7/config-iw7qu9pc.yaml b/wandb/sweep-199gq3m7/config-iw7qu9pc.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-199gq3m7/config-iw7qu9pc.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-199gq3m7/config-xn0ylvd3.yaml b/wandb/sweep-199gq3m7/config-xn0ylvd3.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-199gq3m7/config-xn0ylvd3.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-2b4ndoa2/config-29fpum8f.yaml b/wandb/sweep-2b4ndoa2/config-29fpum8f.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-2b4ndoa2/config-29fpum8f.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml b/wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-2chv306p/config-dfa7a6n1.yaml b/wandb/sweep-2chv306p/config-dfa7a6n1.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-2chv306p/config-dfa7a6n1.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-2chv306p/config-e15fh1bu.yaml b/wandb/sweep-2chv306p/config-e15fh1bu.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-2chv306p/config-e15fh1bu.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-3945fdiz/config-bj96w41e.yaml b/wandb/sweep-3945fdiz/config-bj96w41e.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-3945fdiz/config-bj96w41e.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-3945fdiz/config-ioayvefo.yaml b/wandb/sweep-3945fdiz/config-ioayvefo.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-3945fdiz/config-ioayvefo.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml b/wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-4y3t4lnj/config-irjpk449.yaml b/wandb/sweep-4y3t4lnj/config-irjpk449.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-4y3t4lnj/config-irjpk449.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-5tie5khj/config-lzq16xjh.yaml b/wandb/sweep-5tie5khj/config-lzq16xjh.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-5tie5khj/config-lzq16xjh.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-7h8tmiyj/config-14ueynti.yaml b/wandb/sweep-7h8tmiyj/config-14ueynti.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-7h8tmiyj/config-14ueynti.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml b/wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-94u6xhkc/config-q8kcodq4.yaml b/wandb/sweep-94u6xhkc/config-q8kcodq4.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-94u6xhkc/config-q8kcodq4.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-94u6xhkc/config-zrd7urfq.yaml b/wandb/sweep-94u6xhkc/config-zrd7urfq.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-94u6xhkc/config-zrd7urfq.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-g101608w/config-htlax8nq.yaml b/wandb/sweep-g101608w/config-htlax8nq.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-g101608w/config-htlax8nq.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-g101608w/config-pp1wtj1t.yaml b/wandb/sweep-g101608w/config-pp1wtj1t.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-g101608w/config-pp1wtj1t.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml b/wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml b/wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-mebia8e6/config-0t2gzjry.yaml b/wandb/sweep-mebia8e6/config-0t2gzjry.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-mebia8e6/config-0t2gzjry.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml b/wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-x0n6d1ps/config-r1havcvy.yaml b/wandb/sweep-x0n6d1ps/config-r1havcvy.yaml new file mode 100644 index 0000000..8b7af48 --- /dev/null +++ b/wandb/sweep-x0n6d1ps/config-r1havcvy.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml b/wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-z467ihn2/config-r1jppjka.yaml b/wandb/sweep-z467ihn2/config-r1jppjka.yaml new file mode 100644 index 0000000..63aeac1 --- /dev/null +++ b/wandb/sweep-z467ihn2/config-r1jppjka.yaml @@ -0,0 +1,4 @@ +wandb_version: 1 + +model.conditioner._target_: + value: jamun.model.conditioners.PositionConditioner From 1bc39444ea79c48f816579fe492d2b2ecb25ff66 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 11 Jul 2025 17:11:56 +0000 Subject: [PATCH 10/32] added label override to dataset parser --- src/jamun/data/_utils.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/src/jamun/data/_utils.py b/src/jamun/data/_utils.py index 3ad7963..e0cf0ff 100644 --- a/src/jamun/data/_utils.py +++ b/src/jamun/data/_utils.py @@ -43,6 +43,7 @@ def parse_datasets_from_directory( max_datasets_offset: Optional[int] = None, filter_codes: Optional[Sequence[str]] = None, as_iterable: bool = False, + label_override: Optional[str] = None, **dataset_kwargs, ) -> List[MDtrajDataset]: """Helper function to create MDtrajDataset objects from a directory of trajectory files.""" @@ -106,11 +107,15 @@ def parse_datasets_from_directory( datasets = [] for code in tqdm(codes, desc="Creating datasets"): + if label_override is not None: + print(f"Label override: {label_override}") + code = str(label_override) + dataset = dataset_class( root, traj_files=traj_files[code], pdb_file=pdb_files[code], - label=code, + label=code, # TODO: add a label override here. **dataset_kwargs, ) datasets.append(dataset) From df42585ab51c147c1bbb35552ac558b1bb114f40 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Mon, 14 Jul 2025 21:59:08 +0000 Subject: [PATCH 11/32] Bug fix, label override, init graph recentering - Fixed noise level bug in denoiser_conditional - Cleaned up denoiser_conditional.py (now does not have multimeasurement) - Inserted a manual label override in parse_datasets_from_directory - New datasets and script for noise check experiment - Init graph recentering in ModelSamplingWrapper --- .gitignore | 2 + .../ala_ala_denoiser_experiment_model1.yaml | 73 ++ .../ala_ala_denoiser_experiment_model2.yaml | 72 ++ .../ala_ala_denoiser_experiment_model3.yaml | 73 ++ .../ala_ala_denoiser_experiment_model4.yaml | 74 ++ configs/experiment/sample_capped_2AA.yaml | 20 +- ...mple_capped_single_shape_conditioning.yaml | 51 ++ ...sample_enhanced_sampling_single_shape.yaml | 84 +++ configs/experiment/train_capped_2AA.yaml | 2 +- .../train_capped_2AA_conditional.yaml | 70 ++ .../experiment/train_test_single_shape.yaml | 50 +- .../train_test_single_shape_conditional.yaml | 24 +- ...est_single_shape_conditional_one_traj.yaml | 7 +- ...n_test_single_shape_enhanced_sampling.yaml | 69 ++ configs/sweep.yaml | 4 + configs/sweep_conditioning.yaml | 21 + configs/sweep_conditioning_noise.yaml | 24 + scratch/load_wandb_checkpoint.py | 83 +-- scratch/sample.py | 179 ----- scratch/sampling_prototype.py | 261 ------- scratch/test_conditional.py | 33 +- scripts/slurm/noise_check.sh | 29 + scripts/slurm/run_denoiser_experiments.py | 123 ++++ scripts/slurm/run_single_sample.sh | 6 + scripts/slurm/run_single_sample_from_sweep.sh | 86 +++ scripts/slurm/run_sweep_sampling.sh | 33 + scripts/slurm/show_sweep_combinations.py | 27 + scripts/slurm/sweep.sh | 13 +- scripts/slurm/train_capped_2AA.sh | 2 +- .../train_capped_2AA_ALA_ALA_conditional.sh | 5 +- scripts/slurm/train_capped_2AA_conditional.sh | 37 + .../slurm/train_enhanced_sampling_sweep.sh | 72 ++ .../slurm/train_enhanced_sampling_trial.sh | 76 ++ src/jamun/cmdline/sample.py | 5 +- src/jamun/cmdline/train.py | 18 +- src/jamun/data/__init__.py | 2 + src/jamun/data/_mdtraj.py | 11 +- src/jamun/data/_utils.py | 94 +++ src/jamun/data/noisy_position_dataset.py | 37 + .../batch_sampler/mcmc/aboba_memory.yaml | 12 + .../batch_sampler/mcmc/baoab_memory.yaml | 12 + .../single_measurement_sampler_memory.yaml | 7 + .../model/denoiser_conditional.yaml | 4 - .../denoiser_conditional_pretrained.yaml | 2 + src/jamun/hydra_config/sample_memory.yaml | 27 + src/jamun/hydra_config/train.yaml | 1 + src/jamun/model/conditioners/__init__.py | 2 +- src/jamun/model/conditioners/conditioners.py | 41 +- src/jamun/model/denoiser_conditional.py | 179 ++--- src/jamun/model/denoiser_multimeasurement.py | 691 ++++++++++++++++++ src/jamun/sampling/__init__.py | 2 +- src/jamun/sampling/_sampler.py | 54 ++ src/jamun/sampling/mcmc/__init__.py | 2 +- src/jamun/sampling/mcmc/_splitting.py | 16 +- .../sampling/mcmc/functional/__init__.py | 2 +- .../sampling/mcmc/functional/_splitting.py | 127 +++- src/jamun/sampling/walkjump/__init__.py | 2 +- .../sampling/walkjump/_single_measurement.py | 92 +++ src/jamun/utils/__init__.py | 2 +- src/jamun/utils/sampling_wrapper.py | 106 ++- test.py | 1 - wandb/sweep-01dho9mz/config-o4x6y3nm.yaml | 6 - wandb/sweep-01dho9mz/config-wrqb60uz.yaml | 6 - wandb/sweep-199gq3m7/config-iw7qu9pc.yaml | 4 - wandb/sweep-199gq3m7/config-xn0ylvd3.yaml | 4 - wandb/sweep-2b4ndoa2/config-29fpum8f.yaml | 4 - wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml | 4 - wandb/sweep-2chv306p/config-dfa7a6n1.yaml | 4 - wandb/sweep-2chv306p/config-e15fh1bu.yaml | 4 - wandb/sweep-3945fdiz/config-bj96w41e.yaml | 4 - wandb/sweep-3945fdiz/config-ioayvefo.yaml | 4 - wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml | 4 - wandb/sweep-4y3t4lnj/config-irjpk449.yaml | 4 - wandb/sweep-5tie5khj/config-lhzhrmk7.yaml | 4 - wandb/sweep-5tie5khj/config-lzq16xjh.yaml | 4 - wandb/sweep-7h8tmiyj/config-14ueynti.yaml | 4 - wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml | 4 - wandb/sweep-94u6xhkc/config-q8kcodq4.yaml | 4 - wandb/sweep-94u6xhkc/config-zrd7urfq.yaml | 4 - wandb/sweep-g101608w/config-htlax8nq.yaml | 4 - wandb/sweep-g101608w/config-pp1wtj1t.yaml | 4 - wandb/sweep-gze9gptu/config-sxlo5fvy.yaml | 6 - wandb/sweep-gze9gptu/config-vfd1ntbz.yaml | 6 - wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml | 4 - wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml | 4 - wandb/sweep-mebia8e6/config-0t2gzjry.yaml | 4 - wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml | 4 - wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml | 6 - wandb/sweep-rxcjaig2/config-zvdgfxss.yaml | 6 - wandb/sweep-x0n6d1ps/config-r1havcvy.yaml | 4 - wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml | 4 - wandb/sweep-z467ihn2/config-r1jppjka.yaml | 4 - 92 files changed, 2631 insertions(+), 841 deletions(-) create mode 100644 configs/experiment/ala_ala_denoiser_experiment_model1.yaml create mode 100644 configs/experiment/ala_ala_denoiser_experiment_model2.yaml create mode 100644 configs/experiment/ala_ala_denoiser_experiment_model3.yaml create mode 100644 configs/experiment/ala_ala_denoiser_experiment_model4.yaml create mode 100644 configs/experiment/sample_capped_single_shape_conditioning.yaml create mode 100644 configs/experiment/sample_enhanced_sampling_single_shape.yaml create mode 100644 configs/experiment/train_capped_2AA_conditional.yaml create mode 100644 configs/experiment/train_test_single_shape_enhanced_sampling.yaml create mode 100644 configs/sweep_conditioning.yaml create mode 100644 configs/sweep_conditioning_noise.yaml delete mode 100644 scratch/sample.py delete mode 100644 scratch/sampling_prototype.py create mode 100755 scripts/slurm/noise_check.sh create mode 100644 scripts/slurm/run_denoiser_experiments.py create mode 100755 scripts/slurm/run_single_sample.sh create mode 100755 scripts/slurm/run_single_sample_from_sweep.sh create mode 100644 scripts/slurm/run_sweep_sampling.sh create mode 100644 scripts/slurm/show_sweep_combinations.py create mode 100644 scripts/slurm/train_capped_2AA_conditional.sh create mode 100755 scripts/slurm/train_enhanced_sampling_sweep.sh create mode 100755 scripts/slurm/train_enhanced_sampling_trial.sh create mode 100644 src/jamun/data/noisy_position_dataset.py create mode 100644 src/jamun/hydra_config/batch_sampler/mcmc/aboba_memory.yaml create mode 100644 src/jamun/hydra_config/batch_sampler/mcmc/baoab_memory.yaml create mode 100644 src/jamun/hydra_config/batch_sampler/single_measurement_sampler_memory.yaml create mode 100644 src/jamun/hydra_config/model/denoiser_conditional_pretrained.yaml create mode 100644 src/jamun/hydra_config/sample_memory.yaml create mode 100644 src/jamun/model/denoiser_multimeasurement.py delete mode 100644 test.py delete mode 100644 wandb/sweep-01dho9mz/config-o4x6y3nm.yaml delete mode 100644 wandb/sweep-01dho9mz/config-wrqb60uz.yaml delete mode 100644 wandb/sweep-199gq3m7/config-iw7qu9pc.yaml delete mode 100644 wandb/sweep-199gq3m7/config-xn0ylvd3.yaml delete mode 100644 wandb/sweep-2b4ndoa2/config-29fpum8f.yaml delete mode 100644 wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml delete mode 100644 wandb/sweep-2chv306p/config-dfa7a6n1.yaml delete mode 100644 wandb/sweep-2chv306p/config-e15fh1bu.yaml delete mode 100644 wandb/sweep-3945fdiz/config-bj96w41e.yaml delete mode 100644 wandb/sweep-3945fdiz/config-ioayvefo.yaml delete mode 100644 wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml delete mode 100644 wandb/sweep-4y3t4lnj/config-irjpk449.yaml delete mode 100644 wandb/sweep-5tie5khj/config-lhzhrmk7.yaml delete mode 100644 wandb/sweep-5tie5khj/config-lzq16xjh.yaml delete mode 100644 wandb/sweep-7h8tmiyj/config-14ueynti.yaml delete mode 100644 wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml delete mode 100644 wandb/sweep-94u6xhkc/config-q8kcodq4.yaml delete mode 100644 wandb/sweep-94u6xhkc/config-zrd7urfq.yaml delete mode 100644 wandb/sweep-g101608w/config-htlax8nq.yaml delete mode 100644 wandb/sweep-g101608w/config-pp1wtj1t.yaml delete mode 100644 wandb/sweep-gze9gptu/config-sxlo5fvy.yaml delete mode 100644 wandb/sweep-gze9gptu/config-vfd1ntbz.yaml delete mode 100644 wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml delete mode 100644 wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml delete mode 100644 wandb/sweep-mebia8e6/config-0t2gzjry.yaml delete mode 100644 wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml delete mode 100644 wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml delete mode 100644 wandb/sweep-rxcjaig2/config-zvdgfxss.yaml delete mode 100644 wandb/sweep-x0n6d1ps/config-r1havcvy.yaml delete mode 100644 wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml delete mode 100644 wandb/sweep-z467ihn2/config-r1jppjka.yaml diff --git a/.gitignore b/.gitignore index df6fd73..10226f9 100644 --- a/.gitignore +++ b/.gitignore @@ -24,3 +24,5 @@ torch_compile_debug *.profile* **/*.log **/*.err +wandb/* +scratch/* \ No newline at end of file diff --git a/configs/experiment/ala_ala_denoiser_experiment_model1.yaml b/configs/experiment/ala_ala_denoiser_experiment_model1.yaml new file mode 100644 index 0000000..92c9a73 --- /dev/null +++ b/configs/experiment/ala_ala_denoiser_experiment_model1.yaml @@ -0,0 +1,73 @@ +# @package _global_ +# Model 1: Denoiser with self conditioner, two structures, noise level sigma + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: 2 # Two structures: current + 1 hidden state + lag_subsample_rate: 1 + num_frames: 10000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + num_frames: 100 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 # Base sigma level + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.SelfConditioner + N_structures: ${model.arch.N_structures} + +trainer: + max_epochs: 100 + val_check_interval: 0.5 + +logger: + wandb: + group: ALA_ALA_noise_check + notes: "Model 1: Denoiser with SelfConditioner, 2 structures, sigma=0.04" + tags: ["noise_check", "high noise", "identical measurements"] \ No newline at end of file diff --git a/configs/experiment/ala_ala_denoiser_experiment_model2.yaml b/configs/experiment/ala_ala_denoiser_experiment_model2.yaml new file mode 100644 index 0000000..5d47ce6 --- /dev/null +++ b/configs/experiment/ala_ala_denoiser_experiment_model2.yaml @@ -0,0 +1,72 @@ +# @package _global_ +# Model 2: Denoiser with self conditioner, two structures, noise level sigma/sqrt(2) + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: 2 # Two structures: current + 1 hidden state + lag_subsample_rate: 1 + num_frames: 10000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + num_frames: 100 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.02828427124 # sigma/sqrt(2) = 0.04/sqrt(2) ā‰ˆ 0.0283 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.SelfConditioner + N_structures: ${model.arch.N_structures} + +trainer: + max_epochs: 100 + +logger: + wandb: + group: ALA_ALA_noise_check + notes: "Model 2: Denoiser with SelfConditioner, 2 structures, sigma=0.04/sqrt(2)" + tags: ["noise_check", "low noise", "identical measurements"] \ No newline at end of file diff --git a/configs/experiment/ala_ala_denoiser_experiment_model3.yaml b/configs/experiment/ala_ala_denoiser_experiment_model3.yaml new file mode 100644 index 0000000..05a6220 --- /dev/null +++ b/configs/experiment/ala_ala_denoiser_experiment_model3.yaml @@ -0,0 +1,73 @@ +# @package _global_ +# Model 3: Denoiser with position conditioner, noise level sigma, +# but hidden states are repeated copies of y.pos (noise added by denoiser) + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: 2 # Two structures: current + 1 copy + lag_subsample_rate: 1 + num_frames: 10000 + + val: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + num_frames: 100 + + test: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 # Base sigma level for denoising + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + +trainer: + max_epochs: 100 + +logger: + wandb: + group: ALA_ALA_noise_check + notes: "Model 3: Denoiser with PositionConditioner, 2 structures, sigma=0.04" + tags: ["noise_check", "high noise", "non-identical, i.i.d. measurements"] \ No newline at end of file diff --git a/configs/experiment/ala_ala_denoiser_experiment_model4.yaml b/configs/experiment/ala_ala_denoiser_experiment_model4.yaml new file mode 100644 index 0000000..77e181a --- /dev/null +++ b/configs/experiment/ala_ala_denoiser_experiment_model4.yaml @@ -0,0 +1,74 @@ +# @package _global_ +# Model 3: Denoiser with position conditioner, noise level sigma, +# but hidden states are repeated copies of y.pos (noise added by denoiser) + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: 2 # Two structures: current + 1 copy + lag_subsample_rate: 1 + num_frames: 10000 + + val: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + num_frames: 100 + + test: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 # Base sigma level for denoising + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + align_hidden_states: false + +trainer: + max_epochs: 100 + +logger: + wandb: + group: ALA_ALA_noise_check + notes: "Model 4: Denoiser with PositionConditioner, 2 structures, sigma=0.04, hidden states not aligned" + tags: ["noise_check", "high noise", "non-identical, i.i.d. measurements", "hidden states not aligned"] \ No newline at end of file diff --git a/configs/experiment/sample_capped_2AA.yaml b/configs/experiment/sample_capped_2AA.yaml index f9b1fbf..3402e26 100644 --- a/configs/experiment/sample_capped_2AA.yaml +++ b/configs/experiment/sample_capped_2AA.yaml @@ -1,17 +1,18 @@ # @package _global_ -defaults: - - override /callbacks: - - sampler/save_trajectory.yaml - - _self_ +# defaults: +# - override /callbacks: +# - sampler/save_trajectory.yaml +# - _self_ init_datasets: _target_: jamun.data.parse_datasets_from_directory - root: "${paths.data_path}/capped_diamines/timewarp_splits/test" + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" subsample: 1 + filter_codes: ['ALA_ALA'] num_frames: 60000 @@ -22,15 +23,15 @@ repeat_init_samples: 1 continue_chain: true # New 2AA -wandb_train_run_path: prescient-design/jamun/yfz3vpzg +wandb_train_run_path: sule-shashank/jamun/vxpxronn checkpoint_type: best_so_far -sigma: 0.04 +sigma: 0.01 M: 1.0 -delta: 0.04 +delta: ${sigma} friction: 1.0 inverse_temperature: 1.0 -score_fn_clip: 100.0 +score_fn_clip: null sampler: _target_: jamun.sampling.Sampler @@ -39,3 +40,4 @@ sampler: logger: wandb: group: sample_capped_2AA + tags: ['ALA_ALA', 'sigma_0.01', 'standard JAMUN'] \ No newline at end of file diff --git a/configs/experiment/sample_capped_single_shape_conditioning.yaml b/configs/experiment/sample_capped_single_shape_conditioning.yaml new file mode 100644 index 0000000..911e2ee --- /dev/null +++ b/configs/experiment/sample_capped_single_shape_conditioning.yaml @@ -0,0 +1,51 @@ +# @package _global_ + +defaults: + - override /model: denoiser_conditional_pretrained.yaml +# defaults: +# - override /callbacks: +# - sampler/save_trajectory.yaml +# - _self_ + +# callbacks: +# viz: +# sigma_list: ["${model.sigma_distribution.sigma}"] + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 + total_lag_time: 10 + lag_subsample_rate: 100 + +num_sampling_steps_per_batch: 1000 +num_batches: 10 +num_init_samples_per_dataset: 10 +repeat_init_samples: 1 +continue_chain: false + +# Add your wandb run path here +wandb_train_run_path: sule-shashank/jamun/jqp09yv1 +# checkpoint_dir: /data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-01_16-15-23/checkpoints +# checkpoint_dir: /data2/sules/jamun-conditional-runs/old_outputs/outputs/train/dev/runs/2a381c2d310e3d0789831338/checkpoints +checkpoint_type: last + +sigma: 0.01 +M: 1.0 +delta: ${sigma} +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: 100.0 + +sampler: + _target_: jamun.sampling.SamplerMemory + devices: 1 + +logger: + wandb: + group: sample_ALA_ALA_conditional + notes: stellar-sweep-25 diff --git a/configs/experiment/sample_enhanced_sampling_single_shape.yaml b/configs/experiment/sample_enhanced_sampling_single_shape.yaml new file mode 100644 index 0000000..423aabf --- /dev/null +++ b/configs/experiment/sample_enhanced_sampling_single_shape.yaml @@ -0,0 +1,84 @@ +# @package _global_ + +defaults: + # - override /model: denoiser_conditional_pretrained.yaml + - override /callbacks: null + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" + traj_pattern: "^(.*).xtc" + pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + as_iterable: false + subsample: 10 + # total_lag_time: 5 + # lag_subsample_rate: 1 + max_datasets: 10 + +num_sampling_steps_per_batch: 10 +num_batches: 1 +num_init_samples_per_dataset: 10 +repeat_init_samples: 1 +continue_chain: false + +# Add your checkpoint path here - update with actual trained model path +wandb_train_run_path: sule-shashank/jamun/vxpxronn +# checkpoint_dir: /data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-02_00-37-08/checkpoints +checkpoint_type: last + +sigma: 0.01 +M: 1.0 +delta: ${sigma} +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: 100.0 + +sampler: + _target_: jamun.sampling.Sampler + devices: 1 + +# # Evaluation dataset - standard ALA_ALA from capped diamines for computing metrics +# eval_dataset: +# _target_: jamun.data.parse_datasets_from_directory +# root: "${paths.data_path}/capped_diamines/timewarp_splits/train" +# traj_pattern: "^(.*).xtc" +# pdb_pattern: "^(.*).pdb" +# filter_codes: ['ALA_ALA'] +# as_iterable: false +# subsample: 100 +# max_datasets: 1 + +# # Override ALL callbacks to use eval_dataset for metrics computation +# callbacks: +# measure_sampling_time: +# _target_: jamun.callbacks.sampler.MeasureSamplingTimeCallback +# chemical_validity: +# _target_: jamun.callbacks.sampler.ChemicalValidityMetricsCallback +# datasets: ${eval_dataset} +# bond_length_tolerance: 0.2 +# volume_exclusion_tolerance: 0.1 +# num_molecules_per_trajectory: 100 +# ramachandran_plot: +# _target_: jamun.callbacks.sampler.RamachandranPlotMetricsCallback +# datasets: ${eval_dataset} +# trajectory_visualizer: +# _target_: jamun.callbacks.sampler.TrajectoryVisualizerCallback +# datasets: ${eval_dataset} +# num_frames_to_animate: 100 +# sample_visualizer: +# _target_: jamun.callbacks.sampler.SampleVisualizerCallback +# datasets: ${eval_dataset} +# num_samples_to_plot: 16 +# subsample: 100 +# score_distribution: +# _target_: jamun.callbacks.sampler.ScoreDistributionCallback +# datasets: ${eval_dataset} +# save_trajectory: +# _target_: jamun.callbacks.sampler.SaveTrajectoryCallback +# datasets: ${eval_dataset} + +logger: + wandb: + group: sample_ALA_ALA_enhanced_sampling + notes: "Sampling from enhanced sampling trained conditional denoiser" + tags: ["sample", "enhanced_sampling", "conditional_denoiser", "ALA_ALA", "dutiful-fog-302"] \ No newline at end of file diff --git a/configs/experiment/train_capped_2AA.yaml b/configs/experiment/train_capped_2AA.yaml index 4b7c02c..fd2f985 100644 --- a/configs/experiment/train_capped_2AA.yaml +++ b/configs/experiment/train_capped_2AA.yaml @@ -43,7 +43,7 @@ data: trainer: val_check_interval: 0.1 - max_epochs: 10 + max_epochs: 20 logger: wandb: diff --git a/configs/experiment/train_capped_2AA_conditional.yaml b/configs/experiment/train_capped_2AA_conditional.yaml new file mode 100644 index 0000000..a2e6728 --- /dev/null +++ b/configs/experiment/train_capped_2AA_conditional.yaml @@ -0,0 +1,70 @@ +# @package _global_ +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + + +callbacks: + viz: + sigma_list: ["${model.sigma_distribution.sigma}"] + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + num_frames: 320000 + total_lag_time: 5 + lag_subsample_rate: 10 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/val/" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 100 + max_datasets: 20 + num_frames: 320000 + total_lag_time: 5 + lag_subsample_rate: 10 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/test/" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 100 + max_datasets: 20 + num_frames: 320000 + total_lag_time: 5 + lag_subsample_rate: 10 + +trainer: + val_check_interval: 0.1 + max_epochs: 20 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 3 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + multimeasurement: false + +logger: + wandb: + group: train_capped_2AA diff --git a/configs/experiment/train_test_single_shape.yaml b/configs/experiment/train_test_single_shape.yaml index e674be4..7a11821 100644 --- a/configs/experiment/train_test_single_shape.yaml +++ b/configs/experiment/train_test_single_shape.yaml @@ -3,7 +3,7 @@ model: sigma_distribution: _target_: jamun.distributions.ConstantSigma - sigma: 0.04 + sigma: 0.01 arch: n_layers: 2 max_radius: 1000.0 @@ -20,38 +20,40 @@ data: batch_size: 32 datasets: train: - - _target_: jamun.data.MDtrajDataset - root: "${paths.data_path}/timewarp/2AA-1-large/train/" - traj_files: - - AA-traj-arrays.npz - pdb_file: AA-traj-state0.pdb - subsample: 100 - label: AA + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 + val: - - _target_: jamun.data.MDtrajDataset - root: "${paths.data_path}/timewarp/2AA-1-large/train/" - traj_files: - - AA-traj-arrays.npz - pdb_file: AA-traj-state0.pdb - subsample: 100 - label: AA + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 test: - - _target_: jamun.data.MDtrajDataset - root: "${paths.data_path}/timewarp/2AA-1-large/train/" - traj_files: - - AA-traj-arrays.npz - pdb_file: AA-traj-state0.pdb - subsample: 100 - label: AA + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 trainer: val_check_interval: 0.5 - max_epochs: 200 + max_epochs: 100 logger: wandb: group: train_test - notes: "alanine dipeptide" \ No newline at end of file + notes: "alanine dipeptide" + tags: ["single_shape", "ALA_ALA", "standard JAMUN"] \ No newline at end of file diff --git a/configs/experiment/train_test_single_shape_conditional.yaml b/configs/experiment/train_test_single_shape_conditional.yaml index c922f4a..8fd33de 100644 --- a/configs/experiment/train_test_single_shape_conditional.yaml +++ b/configs/experiment/train_test_single_shape_conditional.yaml @@ -10,7 +10,6 @@ defaults: data: datamodule: batch_size: 32 - num_workers: 2 datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -19,9 +18,10 @@ data: pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] as_iterable: false - subsample: 100 + subsample: 1 total_lag_time: 5 - lag_subsample_rate: 100 + lag_subsample_rate: 1 + num_frames: 10 val: _target_: jamun.data.parse_datasets_from_directory @@ -30,10 +30,10 @@ data: pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] as_iterable: false - subsample: 100 - total_lag_time: 5 - lag_subsample_rate: 100 - + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 test: _target_: jamun.data.parse_datasets_from_directory @@ -42,10 +42,10 @@ data: pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] as_iterable: false - subsample: 100 - total_lag_time: 5 - lag_subsample_rate: 100 - + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 model: sigma_distribution: _target_: jamun.distributions.ConstantSigma @@ -62,7 +62,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 1000 + max_epochs: 500 logger: diff --git a/configs/experiment/train_test_single_shape_conditional_one_traj.yaml b/configs/experiment/train_test_single_shape_conditional_one_traj.yaml index 9b3dd90..c9a3de8 100644 --- a/configs/experiment/train_test_single_shape_conditional_one_traj.yaml +++ b/configs/experiment/train_test_single_shape_conditional_one_traj.yaml @@ -10,7 +10,7 @@ defaults: data: datamodule: batch_size: 1 - num_workers: 2 + num_workers: 4 datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -71,10 +71,11 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 1000 + max_epochs: 500 logger: wandb: group: ALA_ALA, capped diamines, conditional denoiser - notes: "Running on ALA_ALA capped diamine, conditional denoiser" \ No newline at end of file + notes: "Running on ALA_ALA capped diamine, conditional denoiser" + tags: ["train", "capped_diamines", "conditional denoiser", "one_traj"] \ No newline at end of file diff --git a/configs/experiment/train_test_single_shape_enhanced_sampling.yaml b/configs/experiment/train_test_single_shape_enhanced_sampling.yaml new file mode 100644 index 0000000..6f8185e --- /dev/null +++ b/configs/experiment/train_test_single_shape_enhanced_sampling.yaml @@ -0,0 +1,69 @@ +# @package _global_ + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" + traj_pattern: "^(.*).xtc" + pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + as_iterable: false + subsample: 10 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 5000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/fake_enhanced_data/ALA_ALA_organized/val" + traj_pattern: "^(.*).xtc" + pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + as_iterable: false + subsample: 10 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + max_datasets: 5000 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/fake_enhanced_data/ALA_ALA_organized/test" + traj_pattern: "^(.*).xtc" + pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + as_iterable: false + subsample: 10 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + max_datasets: 5000 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.SelfConditioner + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 100 + +logger: + wandb: + group: ALA_ALA, enhanced sampling, conditional denoiser + notes: "Training conditional denoiser on ALA_ALA enhanced sampling data" + tags: ["train", "enhanced_sampling", "conditional_denoiser", "ALA_ALA"] \ No newline at end of file diff --git a/configs/sweep.yaml b/configs/sweep.yaml index f86f9b3..6f8c3a1 100644 --- a/configs/sweep.yaml +++ b/configs/sweep.yaml @@ -10,6 +10,10 @@ parameters: values: - jamun.model.conditioners.PositionConditioner - jamun.model.conditioners.SelfConditioner + model.N_measurements_hidden: + values: [1, 5, 10, 20, 25] + model.sigma_distribution.sigma: + values: [0.04, 0.1085767, 0.29472252, 0.8] command: - ${program} diff --git a/configs/sweep_conditioning.yaml b/configs/sweep_conditioning.yaml new file mode 100644 index 0000000..bfa7b25 --- /dev/null +++ b/configs/sweep_conditioning.yaml @@ -0,0 +1,21 @@ +program: jamun_train +method: grid +project: jamun +name: conditional_vs_self +metric: + name: val/loss + goal: minimize +parameters: + model.conditioner._target_: + values: + - jamun.model.conditioners.PositionConditioner + - jamun.model.conditioners.SelfConditioner + data.datamodule.datasets.train.total_lag_time: + values: [4, 5, 6, 7, 8, 9, 10] +command: + - ${program} + - "--config-dir=configs" + - "experiment=train_test_single_shape_conditional" + - "++trainer.log_every_n_steps=10" + - "++paths.root_path=/data2/sules/jamun-conditional-runs" + - ${args_no_hyphens} \ No newline at end of file diff --git a/configs/sweep_conditioning_noise.yaml b/configs/sweep_conditioning_noise.yaml new file mode 100644 index 0000000..4f3f293 --- /dev/null +++ b/configs/sweep_conditioning_noise.yaml @@ -0,0 +1,24 @@ +program: jamun_train +method: grid +project: jamun +name: conditional_vs_self +metric: + name: val/loss + goal: minimize +parameters: + model.conditioner._target_: + values: + - jamun.model.conditioners.PositionConditioner + - jamun.model.conditioners.SelfConditioner + data.datamodule.datasets.train.total_lag_time: + values: [4, 6, 8, 10] + model.sigma_distribution.sigma: + values: [0.01, 0.03684031, 0.13572088, 0.5] +command: + - ${program} + - "--config-dir=configs" + - "experiment=train_test_single_shape_conditional" + - "++trainer.log_every_n_steps=10" + - "++trainer.max_epochs=100" + - "++paths.root_path=/data2/sules/jamun-conditional-runs" + - ${args_no_hyphens} \ No newline at end of file diff --git a/scratch/load_wandb_checkpoint.py b/scratch/load_wandb_checkpoint.py index 06ba3ec..78d8d37 100644 --- a/scratch/load_wandb_checkpoint.py +++ b/scratch/load_wandb_checkpoint.py @@ -1,60 +1,37 @@ -import sys import os -import torch -import logging -import dotenv -import hydra -from omegaconf import OmegaConf -from denoiser_test import Denoiser -from jamun.utils.checkpoint import find_checkpoint +from jamun.model.denoiser_conditional import Denoiser -# Setup logging -logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) -logger = logging.getLogger("load_wandb_checkpoint") +def load_model_from_local_checkpoint(checkpoint_dir: str): + """ + Loads a model from a local checkpoint directory. + """ + try: + checkpoint_file = None + for file_name in os.listdir(checkpoint_dir): + if file_name.endswith(".ckpt"): + if "last.ckpt" in file_name: + checkpoint_file = file_name + break + checkpoint_file = file_name # fallback to first .ckpt -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ -JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") -JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + if checkpoint_file: + checkpoint_path = os.path.join(checkpoint_dir, checkpoint_file) + print(f"Found checkpoint file: {checkpoint_path}") -project_root = "/homefs/home/sules/jamun" # Adjust if necessary -if project_root not in sys.path: - sys.path.insert(0, project_root) - logger.info(f"Added '{project_root}' to sys.path for module discovery.") -else: - logger.info(f"'{project_root}' is already in sys.path.") + # load model + model = Denoiser.load_from_checkpoint(checkpoint_path) + print("Model loaded successfully!") + print(model) -@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample") -def load_model_from_wandb(cfg, wandb_path: str, device: str = "cuda" if torch.cuda.is_available() else "cpu"): - """ - Load a model checkpoint from wandb. - - Args: - cfg: Hydra configuration - wandb_path (str): Path to the wandb checkpoint (e.g., "entity/project/run_id") - device (str): Device to load the model on - - Returns: - Denoiser: The loaded model - """ - # Find the checkpoint path using the utility function - checkpoint_path = find_checkpoint( - wandb_train_run_path=wandb_path, - checkpoint_type="last" # or "best_so_far" if you want the best checkpoint - ) - logger.info(f"Found checkpoint at: {checkpoint_path}") - - # Update the config with the checkpoint path - cfg.model.checkpoint_path = checkpoint_path - - # Load the model using Hydra - model = hydra.utils.instantiate(cfg.model) - model = model.to(device) - model.eval() - - logger.info("Model loaded successfully") - return model + return model + else: + print(f"No checkpoint file (.ckpt) found in directory: {checkpoint_dir}") + return None + + except Exception as e: + print(f"An error occurred: {e}") + return None if __name__ == "__main__": - # Example usage - wandb_path = "sule-shashank/jamun/y4rm5488" # Replace with actual wandb path - model = load_model_from_wandb(wandb_path) \ No newline at end of file + checkpoint_dir = "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-06-30_19-07-58/checkpoints" + load_model_from_local_checkpoint(checkpoint_dir) \ No newline at end of file diff --git a/scratch/sample.py b/scratch/sample.py deleted file mode 100644 index 39c4792..0000000 --- a/scratch/sample.py +++ /dev/null @@ -1,179 +0,0 @@ -import os -import sys -import traceback -from typing import Sequence -import pdb # For debugging - -import dotenv -import e3nn -import hydra -import lightning.pytorch as pl -import torch -import torch_geometric -from lightning.pytorch.utilities import rank_zero_only -from omegaconf import OmegaConf - -e3nn.set_optimization_defaults(jit_script_fx=False) - -import jamun -from jamun.data import MDtrajDataModule, MDtrajDataset -from jamun.hydra import instantiate_dict_cfg -from jamun.hydra.utils import format_resolver -from jamun.utils import dist_log, find_checkpoint - -# Add project root to path for custom modules -project_root = "/homefs/home/sules/jamun" -if project_root not in sys.path: - sys.path.insert(0, project_root) - -dotenv.load_dotenv(".env", verbose=True) -OmegaConf.register_new_resolver("format", format_resolver) - - -def get_initial_graphs( - datasets: Sequence[MDtrajDataset], num_init_samples_per_dataset: int, repeat: int = 1 -) -> torch_geometric.data.Batch: - """Get initial graphs for sampling.""" - init_graphs = [] - for dataset in datasets: - random_indices = torch.randperm(len(dataset))[:num_init_samples_per_dataset] - for index in random_indices: - init_graph = dataset[index] - # Ensure graph has hidden state - if not hasattr(init_graph, 'hidden_state'): - init_graph.hidden_state = torch.zeros(init_graph.num_nodes, 1) - for _ in range(repeat): - init_graphs.append(init_graph) - return torch_geometric.data.Batch.from_data_list(init_graphs) - - -def run(cfg): - log_cfg = OmegaConf.to_container(cfg, throw_on_missing=True, resolve=True) - - # Breakpoint 0: After config is loaded - # pdb.set_trace() # Inspect full configuration - - dist_log(f"{OmegaConf.to_yaml(log_cfg)}") - dist_log(f"{os.getcwd()=}") - dist_log(f"{torch.__config__.parallel_info()}") - if hasattr(os, "sched_getaffinity"): - dist_log(f"{os.sched_getaffinity(0)=}") - - if matmul_prec := cfg.get("float32_matmul_precision"): - dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") - torch.set_float32_matmul_precision(matmul_prec) - - # # Turned off wandb logging for now - # loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) - # wandb_logger = None - # for logger in loggers: - # if isinstance(logger, pl.loggers.WandbLogger): - # wandb_logger = logger - - # if rank_zero_only.rank == 0 and wandb_logger: - # dist_log(f"{wandb_logger.experiment.name=}") - # wandb_logger.experiment.config.update({"cfg": log_cfg, "version": jamun.__version__, "cwd": os.getcwd()}) - - # Load the checkpoint - print(f'Current working directory: {os.getcwd()}') - checkpoint_path = find_checkpoint( - wandb_train_run_path=cfg.get("wandb_train_run_path"), - checkpoint_dir=cfg.get("checkpoint_dir"), - checkpoint_type=cfg.get("checkpoint_type"), - ) - dist_log(f"Found checkpoint at: {checkpoint_path}") - - # Load checkpoint and create model - checkpoint = torch.load(checkpoint_path, map_location='cpu') - dist_log("Loaded checkpoint successfully") - - # Create model with conditioner - cfg.model._target_ = "scratch.denoiser_test.Denoiser" - if not hasattr(cfg.model, 'conditioner'): - cfg.model.conditioner = OmegaConf.create({ - "_target_": "scratch.conditioners.SelfConditioner", - "hidden_dim": 1 # Adjust based on your needs - }) - - model = hydra.utils.instantiate(cfg.model) - dist_log("Created model instance") - - # Load state dict into model - model.load_state_dict(checkpoint['state_dict']) - dist_log("Loaded state dict into model") - - # Breakpoint 1: After model instantiation and loading - pdb.set_trace() # Check model architecture and weights - - init_datasets = hydra.utils.instantiate(cfg.init_datasets) - init_graphs = get_initial_graphs( - init_datasets, - num_init_samples_per_dataset=cfg.num_init_samples_per_dataset, - repeat=cfg.repeat_init_samples, - ) - - # Breakpoint 2: After dataset loading - pdb.set_trace() # Check dataset and initial graphs - - callbacks = instantiate_dict_cfg(cfg.get("callbacks"), verbose=(rank_zero_only.rank == 0)) - sampler = hydra.utils.instantiate(cfg.sampler, callbacks=callbacks, loggers=loggers) - batch_sampler = hydra.utils.instantiate(cfg.batch_sampler) - - if seed := cfg.get("seed"): - # During sampling, we want ranks to generate different chains - pl.seed_everything(seed + sampler.fabric.global_rank) - - # Run test-time adaptation, if specified - if finetuning_cfg := cfg.get("finetune_on_init"): - num_finetuning_steps = finetuning_cfg.get("num_steps") - dist_log(f"Finetuning for {num_finetuning_steps} steps.") - - # Check that model parameters changed - param_sum = sum(p.sum() for p in model.parameters()) - - # Train the model for a fixed number of steps - trainer = pl.Trainer( - logger=loggers, - max_steps=num_finetuning_steps, - min_steps=num_finetuning_steps, - log_every_n_steps=1, - check_val_every_n_epoch=1, - ) - - # Breakpoint 3: Before finetuning - pdb.set_trace() # Check model state before finetuning - - trainer.fit( - model, - datamodule=MDtrajDataModule( - datasets={"train": init_datasets, "val": init_datasets}, - batch_size=finetuning_cfg.batch_size, - ), - ) - - # Check that model parameters changed - new_param_sum = sum(p.sum() for p in model.parameters()) - dist_log(f"Model parameters changed: {param_sum} -> {new_param_sum}") - - # Breakpoint 4: Before sampling - pdb.set_trace() # Check final model state before sampling - - sampler.sample( - model=model, - batch_sampler=batch_sampler, - init_graphs=init_graphs, - num_batches=cfg.num_batches, - continue_chain=cfg.continue_chain, - ) - - if wandb_logger: - wandb_logger.finalize(status="finished") - - -@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample") -def main(cfg): - run(cfg) - - -if __name__ == "__main__": - main() \ No newline at end of file diff --git a/scratch/sampling_prototype.py b/scratch/sampling_prototype.py deleted file mode 100644 index 22e9a4a..0000000 --- a/scratch/sampling_prototype.py +++ /dev/null @@ -1,261 +0,0 @@ - -# %% Imports and Basic Setup -import functools -import logging -import os -import sys -from typing import Union, Sequence - -import dotenv -import tqdm -import torch -import e3nn -import e3tools.nn -import hydra -from hydra import compose, initialize -from omegaconf import OmegaConf -import lightning.pytorch as pl -import torch_geometric.data - -import jamun -import jamun.data -import jamun.distributions -import jamun.model -import jamun.model.arch -import jamun.sampling -from jamun.utils import compute_average_squared_distance_from_datasets, find_checkpoint -from jamun.hydra import instantiate_dict_cfg -from jamun.data import MDtrajDataModule, MDtrajDataset - -# --- Basic Setup --- -logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) -py_logger = logging.getLogger("jamun_sampling_script") - -torch.cuda.is_available() -torch.set_float32_matmul_precision("high") -e3nn.set_optimization_defaults(jit_script_fx=False) - -# %% Environment and Paths -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ -JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") -JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") - -project_root = "/homefs/home/sules/jamun" # Adjust if necessary -if project_root not in sys.path: - sys.path.insert(0, project_root) - py_logger.info(f"Added '{project_root}' to sys.path for module discovery.") -else: - py_logger.info(f"'{project_root}' is already in sys.path.") - -# %% Load Configuration from Specific File -py_logger.info("Loading configuration from specific training run...") -config_file_path = "../outputs/train/dev/runs/2025-06-11_20-16-04/final_resolved_script_config.yaml" - -# Load the configuration directly from the YAML file -cfg = OmegaConf.load(config_file_path) -py_logger.info("Loaded configuration from training run:") -py_logger.info(f"Config file: {config_file_path}") - -# Modify config for sampling purposes -# Override model target to use our local architectures -cfg.model._target_ = "scratch.denoiser_test.Denoiser" -cfg.model.arch._target_ = "scratch.e3conv_test.E3Conv" - -# Add sampling-specific configuration -if not hasattr(cfg, 'sampling'): - cfg.sampling = OmegaConf.create({}) - -# Set sampling parameters (adapt from sample.yaml structure) -cfg.sampling.repeat_init_samples = 1 -cfg.sampling.num_batches = 10 -cfg.sampling.continue_chain = True -cfg.sampling.num_init_samples_per_dataset = 5 -cfg.sampling.seed = 42 - -# Add sampler configuration -cfg.sampling.sampler = OmegaConf.create({ - "_target_": "jamun.sampling.Sampler", - "precision": "32-true" -}) - -# Add batch sampler configuration -cfg.sampling.batch_sampler = OmegaConf.create({ - "_target_": "jamun.sampling.SingleMeasurementSampler", - "num_steps": 100, - "sigma_min": 0.001, - "sigma_max": 0.1 -}) - -py_logger.info("Modified configuration for sampling:") -py_logger.info(OmegaConf.to_yaml(cfg)) - -# %% Helper Functions -def get_initial_graphs( - datasets: Sequence[MDtrajDataset], num_init_samples_per_dataset: int, repeat: int = 1 -) -> torch_geometric.data.Batch: - """Get initial graphs for sampling.""" - init_graphs = [] - for dataset in datasets: - random_indices = torch.randperm(len(dataset))[:num_init_samples_per_dataset] - for index in random_indices: - init_graph = dataset[index] - for _ in range(repeat): - init_graphs.append(init_graph) - return torch_geometric.data.Batch.from_data_list(init_graphs) - -# %% Load Model -py_logger.info("Loading model from checkpoint...") - -cfg.model._target_ = 'scratch.denoiser_test.Denoiser' -model = hydra.utils.instantiate(cfg.model) - -# Device Setup -if torch.cuda.is_available(): - device = torch.device("cuda") - py_logger.info("CUDA is available. Using GPU.") -else: - device = torch.device("cpu") - py_logger.info("CUDA not available. Using CPU.") - -model = model.to(device) -py_logger.info(f"Model moved to device: {device}") - -# %% Load from ckpoint -sys.path.append("../") -checkpoint_dir = "../outputs/train/dev/runs/2025-06-11_20-16-04/wandb/latest-run/checkpoints" -try: - checkpoint_path = find_checkpoint( - checkpoint_dir=checkpoint_dir, - checkpoint_type="last" # or "best" - ) - py_logger.info(f"Found checkpoint: {checkpoint_path}") -except Exception as e: - py_logger.error(f"Could not find checkpoint in {checkpoint_dir}: {e}") - # Try to find checkpoint in the checkpoints subdirectory - checkpoint_subdir = os.path.join(checkpoint_dir, "run-*/checkpoints") - import glob - checkpoint_files = glob.glob(os.path.join(checkpoint_subdir, "*.ckpt")) - if checkpoint_files: - checkpoint_path = checkpoint_files[-1] # Use the last one - py_logger.info(f"Using checkpoint: {checkpoint_path}") - else: - py_logger.error("No checkpoint files found!") - # sys.exit(1) - -# %% Load from checkpoint -if checkpoint_path: - - checkpoint = torch.load(checkpoint_path, map_location=model.device, weights_only=False) - model.load_state_dict(checkpoint['state_dict']) - py_logger.info("Successfully loaded model from checkpoint.") - py_logger.info(f"Model architecture type: {type(model.g)}") -else: - py_logger.error("No checkpoint files found!") - # sys.exit(1) - -# %% checkpoint path update -cfg.model.checkpoint_path = checkpoint_path -py_logger.info(f"Updated checkpoint path: {cfg.model.checkpoint_path}") - -# %% Setup Initial Datasets for Sampling -py_logger.info("Setting up initial datasets for sampling...") - -# Use the same data configuration as the training -init_datasets = jamun.data.parse_datasets_from_directory( - root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", - traj_pattern="^(.*)-traj-arrays.npz", - pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], - as_iterable=False, - subsample=20, # Smaller subset for sampling - max_datasets=1, -) - -py_logger.info(f"Loaded {len(init_datasets)} samples for initial configurations") - -# %% Generate Initial Graphs -py_logger.info("Generating initial graphs for sampling...") -init_graphs = get_initial_graphs( - init_datasets, - num_init_samples_per_dataset=cfg.sampling.num_init_samples_per_dataset, - repeat=cfg.sampling.repeat_init_samples, -) -py_logger.info(f"Generated {len(init_graphs)} initial graphs") - -# %% Setup Sampling Components -py_logger.info("Setting up sampling components...") - -# Set random seed -if cfg.sampling.seed: - pl.seed_everything(cfg.sampling.seed) - py_logger.info(f"Set random seed to {cfg.sampling.seed}") - -# Setup loggers for sampling -loggers_list = [] -if cfg.get("logger"): - # Create a sampling-specific logger config - sampling_logger_cfg = OmegaConf.create({ - "wandb": { - "_target_": "lightning.pytorch.loggers.WandbLogger", - "project": "jamun-sampling", - "entity": None, - "offline": False, - "group": "sampling_test", - "notes": "Sampling from trained model", - "save_dir": "./outputs/sample/" - } - }) - loggers_list = instantiate_dict_cfg(sampling_logger_cfg) - -# Instantiate sampler -try: - sampler = hydra.utils.instantiate(cfg.sampling.sampler, callbacks=[], loggers=loggers_list) - py_logger.info("Successfully instantiated sampler") -except Exception as e: - py_logger.error(f"Error instantiating sampler: {e}") - # Fallback to direct instantiation - sampler = jamun.sampling.Sampler(precision="32-true", callbacks=[], loggers=loggers_list) - py_logger.info("Using fallback sampler instantiation") - -# Instantiate batch sampler -try: - batch_sampler = hydra.utils.instantiate(cfg.sampling.batch_sampler) - py_logger.info("Successfully instantiated batch sampler") -except Exception as e: - py_logger.error(f"Error instantiating batch sampler: {e}") - # Fallback to direct instantiation - batch_sampler = jamun.sampling.Sampler( - num_steps=100, - sigma_min=0.001, - sigma_max=0.1 - ) - py_logger.info("Using fallback batch sampler instantiation") - -# %% Run Sampling -py_logger.info("Starting sampling...") -try: - sampler.sample( - model=model, - batch_sampler=batch_sampler, - init_graphs=init_graphs, - num_batches=cfg.sampling.num_batches, - continue_chain=cfg.sampling.continue_chain, - ) - py_logger.info("Sampling completed successfully!") - -except Exception as e: - py_logger.error(f"Sampling FAILED: {e}") - import traceback - traceback.print_exc() - -# %% Cleanup and Finish -py_logger.info("Sampling script finished.") - -# Finalize wandb if used -if loggers_list: - for logger in loggers_list: - if isinstance(logger, pl.loggers.WandbLogger): - logger.finalize(status="finished") - py_logger.info("Finalized WandB logger") - -py_logger.info("All done!") \ No newline at end of file diff --git a/scratch/test_conditional.py b/scratch/test_conditional.py index d34c4d0..f865fd6 100644 --- a/scratch/test_conditional.py +++ b/scratch/test_conditional.py @@ -44,11 +44,10 @@ def main(cfg): # cfg = OmegaConf.merge(cfg, test_cfg) # cfg = OmegaConf.merge(cfg, test_cfg, override=True) # breakpoint() - print("Loading datamodule...") datamodule = hydra.utils.instantiate(cfg.data.datamodule) datamodule.setup('test') - # breakpoint() + breakpoint() print("Loading model...") model = hydra.utils.instantiate(cfg.model) @@ -58,7 +57,7 @@ def main(cfg): print("Getting a batch of data...") train_loader = datamodule.train_dataloader() _, batch = next(enumerate(train_loader)) - # breakpoint() + breakpoint() # # Move to CPU # batch = batch.to("cpu") @@ -70,19 +69,25 @@ def main(cfg): # Test forward pass print("Testing forward pass...") - with torch.no_grad(): - sigma = model.sigma_distribution.sample() - x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + # with torch.no_grad(): + # sigma = model.sigma_distribution.sample() + # x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) - print(f"Input shape: {batch.pos.shape}") - print(f"Noisy shape: {y.pos.shape}") - print(f"Output shape: {xhat.pos.shape}") + # print(f"Input shape: {batch.pos.shape}") + # print(f"Noisy shape: {y.pos.shape}") + # print(f"Output shape: {xhat.pos.shape}") - # Test loss computation - print("Testing loss computation...") - loss, aux = model.compute_loss(x_target, xhat, sigma) - print(f"Loss: {loss.mean().item():.4f}") - print(f"Metrics: {aux}") + # Test single backward pass computation + trainer = hydra.utils.instantiate(cfg.trainer) + trainer.fit(model, datamodule=datamodule, ckpt_path=None) + + + # loss = model.training_step(batch, 0) + # breakpoint() + # print("Testing loss computation...") + # loss, aux = model.compute_loss(x_target, xhat, sigma) + # print(f"Loss: {loss.mean().item():.4f}") + # print(f"Metrics: {aux}") if __name__ == "__main__": main() \ No newline at end of file diff --git a/scripts/slurm/noise_check.sh b/scripts/slurm/noise_check.sh new file mode 100755 index 0000000..0286321 --- /dev/null +++ b/scripts/slurm/noise_check.sh @@ -0,0 +1,29 @@ +#!/bin/bash + +#SBATCH --job-name=noise_check +#SBATCH --partition gpu2 +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node +#SBATCH --gpus-per-node=1 # Assign one GPU to each agent +#SBATCH --cpus-per-task=12 +#SBATCH --time 1-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array 3-4 + +# Print job information +echo "Starting job $SLURM_JOB_ID, array task $SLURM_ARRAY_TASK_ID" +echo "Running on node: $(hostname)" +echo "Job started at: $(date)" + +# Set up environment +source ~/.bashrc +conda activate jamun + +# Change to project directory +cd /homefs/home/sules/jamun + +# Run training with the corresponding model config +echo "Training with experiment: ala_ala_denoiser_experiment_model${SLURM_ARRAY_TASK_ID}" +jamun_train --config-dir=configs experiment=ala_ala_denoiser_experiment_model${SLURM_ARRAY_TASK_ID} + +echo "Job completed at: $(date)" \ No newline at end of file diff --git a/scripts/slurm/run_denoiser_experiments.py b/scripts/slurm/run_denoiser_experiments.py new file mode 100644 index 0000000..457201a --- /dev/null +++ b/scripts/slurm/run_denoiser_experiments.py @@ -0,0 +1,123 @@ +#!/usr/bin/env python3 +""" +Script to run the three ALA_ALA denoiser experiments. + +Experiment Setup: +================= + +1. Model 1: Denoiser with SelfConditioner, 2 structures, noise level sigma=0.04 + - Uses real lagged frames from trajectory as hidden states + - SelfConditioner just repeats the current position + +2. Model 2: Denoiser with SelfConditioner, 2 structures, noise level sigma/sqrt(2)ā‰ˆ0.0283 + - Same as Model 1 but with reduced noise level + - Uses real lagged frames from trajectory as hidden states + +3. Model 3: Denoiser with PositionConditioner, 2 structures, noise level sigma=0.04 + - Hidden states are repeated copies of current position (not real trajectory frames) + - PositionConditioner aligns these copies to current position + - Noise is added by the denoiser during training + +Usage: +====== +python run_denoiser_experiments.py [model_number] + +Where model_number is 1, 2, or 3. If no number is provided, all models will be run. +""" + +import subprocess +import sys +import time +from pathlib import Path + +def run_experiment(model_num: int, root_path: str = "/data2/sules/jamun-denoiser-experiments"): + """Run a specific experiment model.""" + config_name = f"ala_ala_denoiser_experiment_model{model_num}" + + # Map model numbers to descriptions + descriptions = { + 1: "Model 1: SelfConditioner, sigma=0.04", + 2: "Model 2: SelfConditioner, sigma/sqrt(2)ā‰ˆ0.0283", + 3: "Model 3: PositionConditioner with repeated position copies, sigma=0.04" + } + + print(f"\n{'='*60}") + print(f"Starting {descriptions[model_num]}") + print(f"Config: {config_name}") + print(f"Output path: {root_path}/model{model_num}") + print(f"{'='*60}\n") + + cmd = [ + "python", "jamun_train.py", + "--config-dir=configs", + f"experiment={config_name}", + f"++paths.root_path={root_path}/model{model_num}", + "++trainer.max_epochs=500", + "++trainer.log_every_n_steps=10" + ] + + print(f"Running command: {' '.join(cmd)}") + + try: + result = subprocess.run(cmd, check=True, cwd=Path(__file__).parent) + print(f"\nāœ… Model {model_num} completed successfully!") + return True + except subprocess.CalledProcessError as e: + print(f"\nāŒ Model {model_num} failed with error: {e}") + return False + except KeyboardInterrupt: + print(f"\nāš ļø Model {model_num} interrupted by user") + return False + +def main(): + """Main function to run experiments.""" + print(__doc__) + + # Parse command line arguments + if len(sys.argv) > 1: + try: + model_num = int(sys.argv[1]) + if model_num not in [1, 2, 3]: + raise ValueError() + models_to_run = [model_num] + except ValueError: + print("Error: Please provide a valid model number (1, 2, or 3)") + sys.exit(1) + else: + models_to_run = [1, 2, 3] + print("No model specified. Running all three models...") + + # Check if we're in the right directory + if not Path("jamun_train.py").exists(): + print("Error: jamun_train.py not found. Please run this script from the jamun root directory.") + sys.exit(1) + + # Run experiments + start_time = time.time() + results = {} + + for model_num in models_to_run: + print(f"\n\nStarting Model {model_num}...") + results[model_num] = run_experiment(model_num) + + if len(models_to_run) > 1 and model_num != models_to_run[-1]: + print(f"\nWaiting 10 seconds before starting next model...") + time.sleep(10) + + # Print summary + elapsed = time.time() - start_time + print(f"\n\n{'='*60}") + print(f"EXPERIMENT SUMMARY") + print(f"{'='*60}") + print(f"Total time: {elapsed/3600:.2f} hours") + print() + + for model_num in models_to_run: + status = "āœ… SUCCESS" if results[model_num] else "āŒ FAILED" + print(f"Model {model_num}: {status}") + + print(f"\nResults saved to: /data2/sules/jamun-denoiser-experiments/") + print(f"{'='*60}") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scripts/slurm/run_single_sample.sh b/scripts/slurm/run_single_sample.sh new file mode 100755 index 0000000..4730be3 --- /dev/null +++ b/scripts/slurm/run_single_sample.sh @@ -0,0 +1,6 @@ +#!/bin/bash + +# This script runs sampling for a specific run from a wandb sweep, selected by an index. + +jamun_sample --config-dir=configs experiment=sample_capped_single_shape_conditioning ++wandb_train_run_path=sule-shashank/jamun/zchesftt ++logger.wandb.notes=jumping-sweep-29 +jamun_sample --config-dir=configs experiment=sample_capped_single_shape_conditioning ++wandb_train_run_path=sule-shashank/jamun/jqp09yv1 ++logger.wandb.notes=stellar-sweep-25 \ No newline at end of file diff --git a/scripts/slurm/run_single_sample_from_sweep.sh b/scripts/slurm/run_single_sample_from_sweep.sh new file mode 100755 index 0000000..a53eb1f --- /dev/null +++ b/scripts/slurm/run_single_sample_from_sweep.sh @@ -0,0 +1,86 @@ +#!/bin/bash + +# This script runs sampling for a specific run from a wandb sweep, selected by an index. + +set -e + +# --- Configuration --- +SWEEP_ID="sule-shashank/jamun/evgtrff4" + +# --- Argument Parsing --- +if [ "$#" -ne 1 ]; then + echo "Usage: $0 " + echo "Please provide the 0-based index of the run to process." + exit 1 +fi +RUN_INDEX=$1 + +# --- Main Logic --- +echo "Fetching run at index $RUN_INDEX from sweep $SWEEP_ID..." + +python -c " +import wandb +import subprocess +import sys + +# The sweep ID and run index are passed as command-line arguments +if len(sys.argv) < 3: + print('Usage: python_script.py ', file=sys.stderr) + sys.exit(1) +sweep_id = sys.argv[1] +run_index = int(sys.argv[2]) + +try: + api = wandb.Api() + sweep = api.sweep(sweep_id) + # wandb runs are often ordered from newest to oldest; reverse to make index stable + runs_list = list(reversed(list(sweep.runs))) + + if run_index >= len(runs_list) or run_index < 0: + print(f'Error: Run index {run_index} is out of bounds. The sweep has {len(runs_list)} runs (indices 0 to {len(runs_list) - 1}).', file=sys.stderr) + sys.exit(1) + + run = runs_list[run_index] + run_path = '/'.join(run.path) + conditioner = run.config.get('cfg', {}).get('model', {}).get('conditioner', {}).get('_target_') + total_lag_time = run.config.get('cfg', {}).get('data', {}).get('datamodule', {}).get('datasets', {}).get('train', {}).get('total_lag_time') + sigma = run.config.get('cfg', {}).get('model', {}).get('sigma_distribution', {}).get('sigma') + + # Ensure all required parameters are present + if not all(v is not None for v in [run_path, conditioner, total_lag_time, sigma]): + print(f'Error: Could not extract all required parameters for run at index {run_index}.', file=sys.stderr) + sys.exit(1) + + print('========================================') + print(f'Starting sampling for run at index {run_index}: {run_path}') + print(f' Conditioner: {conditioner}') + print(f' Total Lag Time: {total_lag_time}') + print(f' Sigma: {sigma}') + print('========================================') + + # Execute jamun_sample with the extracted parameters + tags_string = '[' + f'\"{str(conditioner)}\", \"{str(total_lag_time)}\", \"{str(sigma)}\"' + ']' + cmd = [ + 'jamun_sample', + '--config-dir=configs', + 'experiment=sample_capped_single_shape_conditioning.yaml', + f'wandb_train_run_path={run_path}', + f'++init_datasets.total_lag_time={total_lag_time}', + f'++sigma={sigma}', + f'++delta={sigma}', + f'++logger.wandb.group=sampling_from_sweep_{run_index}', + f'++logger.wandb.tags={tags_string}' + ] + + result = subprocess.run(cmd, check=True) + + print('----------------------------------------') + print(f'Finished sampling for run index {run_index}.') + +except subprocess.CalledProcessError as e: + print(f'Error running jamun_sample: {e}', file=sys.stderr) + sys.exit(1) +except Exception as e: + print(f'Error fetching data from wandb: {e}', file=sys.stderr) + sys.exit(1) +" "$SWEEP_ID" "$RUN_INDEX" \ No newline at end of file diff --git a/scripts/slurm/run_sweep_sampling.sh b/scripts/slurm/run_sweep_sampling.sh new file mode 100644 index 0000000..c0a6011 --- /dev/null +++ b/scripts/slurm/run_sweep_sampling.sh @@ -0,0 +1,33 @@ +#!/bin/bash +#SBATCH --job-name=sweep_sampling +#SBATCH --partition gpu2 +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node +#SBATCH --gpus-per-node=1 # Assign one GPU to each agent +#SBATCH --cpus-per-task=12 +#SBATCH --time 1-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array 2-31 + +# Create logs directory if it doesn't exist +mkdir -p logs + +# Print job information +echo "Job ID: $SLURM_JOB_ID" +echo "Array Task ID: $SLURM_ARRAY_TASK_ID" +echo "Running on node: $HOSTNAME" +echo "Starting time: $(date)" + +# Activate conda environment (adjust the environment name as needed) +source ~/.bashrc +conda activate jamun + +# Change to the working directory +cd /homefs/home/sules/jamun + +# Run the sampling script with the array task ID as the run index +echo "Running sampling for sweep run index: $SLURM_ARRAY_TASK_ID" +bash run_single_sample_from_sweep.sh $SLURM_ARRAY_TASK_ID + +echo "Finished sampling for run index: $SLURM_ARRAY_TASK_ID" +echo "End time: $(date)" \ No newline at end of file diff --git a/scripts/slurm/show_sweep_combinations.py b/scripts/slurm/show_sweep_combinations.py new file mode 100644 index 0000000..b89a908 --- /dev/null +++ b/scripts/slurm/show_sweep_combinations.py @@ -0,0 +1,27 @@ +#!/usr/bin/env python3 +""" +Show all parameter combinations for the enhanced sampling sweep. +""" + +# Define parameter arrays (same as in the bash script) +CONDITIONERS = ["PositionConditioner", "SelfConditioner"] +SIGMAS = [0.01, 0.04, 0.08, 0.1] +LAG_TIMES = [2, 5, 8] + +print("Enhanced Sampling Training Sweep - Parameter Combinations") +print("=" * 60) +print(f"Total combinations: {len(CONDITIONERS)} Ɨ {len(SIGMAS)} Ɨ {len(LAG_TIMES)} = {len(CONDITIONERS) * len(SIGMAS) * len(LAG_TIMES)}") +print() +print(f"{'Task ID':<8} {'Conditioner':<18} {'Sigma':<8} {'Lag Time':<10}") +print("-" * 45) + +task_id = 0 +for cond_idx, conditioner in enumerate(CONDITIONERS): + for sigma_idx, sigma in enumerate(SIGMAS): + for lag_idx, lag_time in enumerate(LAG_TIMES): + print(f"{task_id:<8} {conditioner:<18} {sigma:<8} {lag_time:<10}") + task_id += 1 + +print() +print("To run the sweep:") +print("sbatch scripts/slurm/train_enhanced_sampling_sweep.sh") \ No newline at end of file diff --git a/scripts/slurm/sweep.sh b/scripts/slurm/sweep.sh index 5b15c4f..907db55 100644 --- a/scripts/slurm/sweep.sh +++ b/scripts/slurm/sweep.sh @@ -1,15 +1,16 @@ #!/usr/bin/env bash -#SBATCH --partition gpu3 -#SBATCH --qos=preempt +#SBATCH --partition gpu2 #SBATCH --nodes 1 -#SBATCH --ntasks-per-node=2 # Number of agents to run in parallel on this node -#SBATCH --gpus-per-node=2 # Assign one GPU to each agent -#SBATCH --cpus-per-task=8 +#SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node +#SBATCH --gpus-per-node=1 # Assign one GPU to each agent +#SBATCH --cpus-per-task=12 #SBATCH --time 3-0 #SBATCH --mem-per-cpu=32G +#SBATCH --array 0-15 # Check if a Sweep ID is provided as an argument +export JAMUN_ROOT_PATH=/data2/sules/jamun-conditional-runs if [ -z "$1" ]; then echo "Error: Please provide the W&B Sweep ID as the first argument." echo "Usage: sbatch scripts/slurm/sweep.sh " @@ -31,4 +32,4 @@ echo "Starting ${SLURM_NTASKS} agents for sweep: ${SWEEP_ID}" # Launch multiple wandb agents in parallel using srun. # Each agent will poll the sweep server, get a configuration, and run one training job. # PyTorch Lightning will automatically use the single GPU assigned by Slurm to each task. -wandb agent --count 1 "${SWEEP_ID}" \ No newline at end of file +srun wandb agent "${SWEEP_ID}" \ No newline at end of file diff --git a/scripts/slurm/train_capped_2AA.sh b/scripts/slurm/train_capped_2AA.sh index f8117be..4a54eeb 100644 --- a/scripts/slurm/train_capped_2AA.sh +++ b/scripts/slurm/train_capped_2AA.sh @@ -28,7 +28,7 @@ echo "RUN_KEY = ${RUN_KEY}" nvidia-smi srun --cpus-per-task 8 --cpu-bind=cores,verbose \ - jamun_train --config-dir=/homefs/home/daigavaa/jamun/configs \ + jamun_train --config-dir=configs \ experiment=train_capped_2AA.yaml \ ++trainer.devices=$SLURM_GPUS_PER_NODE \ ++trainer.num_nodes=$SLURM_JOB_NUM_NODES \ diff --git a/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh b/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh index de0c4d8..7d10607 100644 --- a/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh +++ b/scripts/slurm/train_capped_2AA_ALA_ALA_conditional.sh @@ -3,11 +3,12 @@ #SBATCH --partition gpu3 #SBATCH --qos=preempt #SBATCH --nodes 1 -#SBATCH --ntasks-per-node 4 -#SBATCH --gpus-per-node 4 +#SBATCH --ntasks-per-node 1 +#SBATCH --gpus-per-node 1 #SBATCH --cpus-per-task 8 #SBATCH --time 3-0 #SBATCH --mem-per-cpu=32G +#SBATCH --array 4-10 eval "$(conda shell.bash hook)" conda activate jamun diff --git a/scripts/slurm/train_capped_2AA_conditional.sh b/scripts/slurm/train_capped_2AA_conditional.sh new file mode 100644 index 0000000..5384bd9 --- /dev/null +++ b/scripts/slurm/train_capped_2AA_conditional.sh @@ -0,0 +1,37 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu3 +#SBATCH --qos=preempt +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node 4 +#SBATCH --gpus-per-node 4 +#SBATCH --cpus-per-task 8 +#SBATCH --time 3-0 +#SBATCH --mem-per-cpu=32G + +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 +# export TORCH_COMPILE_DEBUG=1 +# export TORCH_LOGS="+dynamo" +# export TORCHDYNAMO_VERBOSE=1 + +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +nvidia-smi + +# srun --cpus-per-task 8 --cpu-bind=cores,verbose \ +jamun_train --config-dir=configs \ + experiment=train_capped_2AA_conditional.yaml \ + ++trainer.devices=$SLURM_GPUS_PER_NODE \ + ++trainer.num_nodes=$SLURM_JOB_NUM_NODES \ + ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","train","capped_2AA"] \ + ++run_key=$RUN_KEY diff --git a/scripts/slurm/train_enhanced_sampling_sweep.sh b/scripts/slurm/train_enhanced_sampling_sweep.sh new file mode 100755 index 0000000..bd14fb7 --- /dev/null +++ b/scripts/slurm/train_enhanced_sampling_sweep.sh @@ -0,0 +1,72 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu2 +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node 1 +#SBATCH --gpus-per-node 1 +#SBATCH --cpus-per-task 8 +#SBATCH --time 3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-23 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 + +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +# Define parameter arrays +CONDITIONERS=("jamun.model.conditioners.PositionConditioner" "jamun.model.conditioners.SelfConditioner") +CONDITIONER_NAMES=("PositionConditioner" "SelfConditioner") +SIGMAS=(0.01 0.04 0.08 0.1) +LAG_TIMES=(2 5 8) + +# Calculate parameter indices from SLURM_ARRAY_TASK_ID +# Total combinations: 2 conditioners * 4 sigmas * 3 lag_times = 24 +COND_IDX=$((SLURM_ARRAY_TASK_ID / 12)) +SIGMA_IDX=$(((SLURM_ARRAY_TASK_ID % 12) / 3)) +LAG_IDX=$((SLURM_ARRAY_TASK_ID % 3)) + +# Get parameter values +CONDITIONER=${CONDITIONERS[$COND_IDX]} +CONDITIONER_NAME=${CONDITIONER_NAMES[$COND_IDX]} +SIGMA=${SIGMAS[$SIGMA_IDX]} +LAG_TIME=${LAG_TIMES[$LAG_IDX]} + +echo "Parameter combination ${SLURM_ARRAY_TASK_ID}:" +echo " Conditioner: ${CONDITIONER_NAME}" +echo " Sigma: ${SIGMA}" +echo " Total lag time: ${LAG_TIME}" + +nvidia-smi + +# Run training with parameter overrides +jamun_train --config-dir=configs \ + experiment=train_test_single_shape_enhanced_sampling.yaml \ + ++trainer.max_epochs=100 \ + ++data.datamodule.datasets.train.subsample=1 \ + ++data.datamodule.datasets.val.subsample=1 \ + ++data.datamodule.datasets.test.subsample=1 \ + ++model.conditioner._target_=${CONDITIONER} \ + ++model.sigma_distribution.sigma=${SIGMA} \ + ++data.datamodule.datasets.train.total_lag_time=${LAG_TIME} \ + ++model.arch.N_structures=${LAG_TIME} \ + ++logger.wandb.group="fake_enhanced_data_jul_11_sweep" \ + ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","train","enhanced_sampling","${CONDITIONER_NAME}","sigma_${SIGMA}","lag_${LAG_TIME}"] \ + ++run_key=$RUN_KEY \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_sampling_trial.sh b/scripts/slurm/train_enhanced_sampling_trial.sh new file mode 100755 index 0000000..af55511 --- /dev/null +++ b/scripts/slurm/train_enhanced_sampling_trial.sh @@ -0,0 +1,76 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu3 +#SBATCH --qos=preempt +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node 1 +#SBATCH --gpus-per-node 1 +#SBATCH --cpus-per-task 8 +#SBATCH --time 0-2 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 + +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +# Define parameter arrays +CONDITIONERS=("jamun.model.conditioners.PositionConditioner" "jamun.model.conditioners.SelfConditioner") +CONDITIONER_NAMES=("PositionConditioner" "SelfConditioner") +SIGMAS=(0.01 0.04 0.08 0.1) +LAG_TIMES=(2 5 8) + +# Calculate parameter indices from SLURM_ARRAY_TASK_ID +# Total combinations: 2 conditioners * 4 sigmas * 3 lag_times = 24 +COND_IDX=$((SLURM_ARRAY_TASK_ID / 12)) +SIGMA_IDX=$(((SLURM_ARRAY_TASK_ID % 12) / 3)) +LAG_IDX=$((SLURM_ARRAY_TASK_ID % 3)) + +# Get parameter values +CONDITIONER=${CONDITIONERS[$COND_IDX]} +CONDITIONER_NAME=${CONDITIONER_NAMES[$COND_IDX]} +SIGMA=${SIGMAS[$SIGMA_IDX]} +LAG_TIME=${LAG_TIMES[$LAG_IDX]} + +echo "TRIAL RUN - Parameter combination ${SLURM_ARRAY_TASK_ID}:" +echo " Conditioner: ${CONDITIONER_NAME}" +echo " Sigma: ${SIGMA}" +echo " Total lag time: ${LAG_TIME}" + +nvidia-smi + +# Run training with parameter overrides (reduced epochs for trial) +jamun_train --config-dir=configs \ + experiment=train_test_single_shape_enhanced_sampling.yaml \ + ++trainer.max_epochs=2 \ + ++data.datamodule.datasets.train.subsample=10 \ + ++data.datamodule.datasets.val.subsample=10 \ + ++data.datamodule.datasets.test.subsample=10 \ + ++data.datamodule.datasets.train.max_datasets=5 \ + ++data.datamodule.datasets.val.max_datasets=5 \ + ++data.datamodule.datasets.test.max_datasets=5 \ + ++model.conditioner._target_=${CONDITIONER} \ + ++model.sigma_distribution.sigma=${SIGMA} \ + ++data.datamodule.datasets.train.total_lag_time=${LAG_TIME} \ + ++model.arch.N_structures=${LAG_TIME} \ + ++logger.wandb.group="fake_enhanced_data_trial" \ + ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","trial","enhanced_sampling","${CONDITIONER_NAME}","sigma_${SIGMA}","lag_${LAG_TIME}"] \ + ++run_key=$RUN_KEY \ No newline at end of file diff --git a/src/jamun/cmdline/sample.py b/src/jamun/cmdline/sample.py index 87869db..4c5b070 100644 --- a/src/jamun/cmdline/sample.py +++ b/src/jamun/cmdline/sample.py @@ -66,7 +66,6 @@ def run(cfg): if matmul_prec := cfg.get("float32_matmul_precision"): dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") torch.set_float32_matmul_precision(matmul_prec) - breakpoint() loggers = instantiate_dict_cfg(cfg.get("logger"), verbose=(rank_zero_only.rank == 0)) wandb_logger = None for logger in loggers: @@ -95,15 +94,13 @@ def run(cfg): num_init_samples_per_dataset=cfg.num_init_samples_per_dataset, repeat=cfg.repeat_init_samples, ) - callbacks = instantiate_dict_cfg(cfg.get("callbacks"), verbose=(rank_zero_only.rank == 0)) sampler = hydra.utils.instantiate(cfg.sampler, callbacks=callbacks, loggers=loggers) batch_sampler = hydra.utils.instantiate(cfg.batch_sampler) - if seed := cfg.get("seed"): # During sampling, we want ranks to generate different chains. pl.seed_everything(seed + sampler.fabric.global_rank) - + breakpoint() # Run test-time adapation, if specified. if finetuning_cfg := cfg.get("finetune_on_init"): num_finetuning_steps = finetuning_cfg.get("num_steps") diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 64b7d74..7ee67ee 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -45,6 +45,11 @@ def run(cfg): # Set the start method to spawn to avoid issues with the default fork method. torch.multiprocessing.set_start_method("spawn", force=True) + # Set random seed for reproducible training + if seed := cfg.get("seed"): + lightning.seed_everything(seed) + dist_log(f"Set random seed to {seed} for reproducible training") + # Compute data normalization. if cfg.get("compute_average_squared_distance_from_data"): average_squared_distance = compute_average_squared_distance_from_config(cfg) @@ -53,12 +58,20 @@ def run(cfg): ) cfg.model.average_squared_distance = average_squared_distance + # # do this for the sweep + # if cfg.model.N_measurements_hidden is not None: + # dist_log(f"Number of hidden measurements: {cfg.model.N_measurements_hidden}") + # dist_log(f"Overwriting N_measurements...") + # cfg.model.N_measurements = 100 // cfg.model.N_measurements_hidden + # dist_log(f"New num of measurements: {cfg.model.N_measurements=}") + datamodule = hydra.utils.instantiate(cfg.data.datamodule) model = hydra.utils.instantiate(cfg.model) if matmul_prec := cfg.get("float32_matmul_precision"): dist_log(f"Setting float_32_matmul_precision to {matmul_prec}") torch.set_float32_matmul_precision(matmul_prec) + # breakpoint() # # If running under Slurm, ensure the number of devices matches the allocation. # if "SLURM_GPUS_PER_TASK" in os.environ and torch.cuda.is_available(): # dist_log(f"torch.cuda.device_count(): {torch.cuda.device_count()}") @@ -85,7 +98,7 @@ def run(cfg): callbacks = instantiate_dict_cfg(cfg.get("callbacks"), verbose=(rank_zero_only.rank == 0)) trainer = hydra.utils.instantiate(cfg.trainer, callbacks=callbacks, logger=loggers) - + # breakpoint() # TODO support wandb notes/description if rank_zero_only.rank == 0 and wandb_logger: wandb_logger.experiment.config.update({"cfg": log_cfg, "version": jamun.__version__, "cwd": os.getcwd()}) @@ -101,8 +114,9 @@ def run(cfg): else: checkpoint_path = None print(f'Saving checkpoints @ {checkpoint_path}') - trainer.fit(model, datamodule=datamodule, ckpt_path=checkpoint_path) + trainer.fit(model, datamodule=datamodule, ckpt_path=checkpoint_path) + # breakpoint() if wandb_logger and isinstance(trainer.profiler, lightning.pytorch.profilers.PyTorchProfiler): profile_art = wandb.Artifact("trace", type="profile") for trace in pathlib.Path(trainer.profiler.dirpath).glob("*.pt.trace.json"): diff --git a/src/jamun/data/__init__.py b/src/jamun/data/__init__.py index 8800cc2..ab98bb5 100644 --- a/src/jamun/data/__init__.py +++ b/src/jamun/data/__init__.py @@ -1,11 +1,13 @@ from ._dloader import MDtrajDataModule, RandomChainDataset, StreamingRandomChainDataset from ._mdtraj import MDtrajDataset, MDtrajIterableDataset from ._sdf import MDtrajSDFDataset +from .noisy_position_dataset import RepeatedPositionDataset from ._utils import ( concatenate_datasets, create_dataset_from_pdbs, dloader_map_reduce, parse_datasets_from_directory, parse_datasets_from_directory_new, + parse_repeated_position_datasets_from_directory, parse_sdf_datasets_from_directory, ) diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index 4a3638d..45096b8 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -215,9 +215,10 @@ def __init__( if subsample is None or subsample == 0: subsample = 1 - + # Get lagged indices if lag parameters are provided if total_lag_time is not None and lag_subsample_rate is not None: + print(f"total_lag_time: {total_lag_time}, lag_subsample_rate: {lag_subsample_rate}") self.traj = self.traj[start_frame : start_frame + num_frames] # accommodate for start_frame and num_frames lagged_indices = get_subsampled_indices( self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate @@ -231,6 +232,7 @@ def __init__( self.traj = self.traj[subsampled_indices] # self.traj is permanently modified. else: # Regular subsampling without lag + print(f"subsample: {subsample}, regular subsampling") self.traj = self.traj[start_frame : start_frame + num_frames : subsample] self.hidden_state = None self.lagged_indices = None @@ -245,7 +247,10 @@ def __init__( self.graph.pos = torch.tensor(self.traj.xyz[0], dtype=torch.float32) self.graph.loss_weight = torch.tensor([loss_weight], dtype=torch.float32) self.graph.dataset_label = self.label() - self.graph.hidden_state = [self.hidden_state[0].xyz[i] for i in range(self.hidden_state[0].n_frames)] + if self.hidden_state is not None: + self.graph.hidden_state = [self.hidden_state[0].xyz[i] for i in range(self.hidden_state[0].n_frames)] + else: + self.graph.hidden_state = [] if verbose: utils.dist_log(f"Dataset {self.label()}: Loading trajectory files {traj_files} and PDB file {pdb_file}.") utils.dist_log( @@ -264,7 +269,7 @@ def __getitem__(self, idx): graph = self.graph.clone() graph.pos = torch.tensor(self.traj.xyz[idx]) - if self.lagged_indices is not None: + if self.hidden_state is not None: graph.hidden_state = [torch.tensor(self.hidden_state[idx].xyz[i]) for i in range(self.hidden_state[idx].n_frames)] else: graph.hidden_state = [] diff --git a/src/jamun/data/_utils.py b/src/jamun/data/_utils.py index e0cf0ff..4360a6d 100644 --- a/src/jamun/data/_utils.py +++ b/src/jamun/data/_utils.py @@ -353,3 +353,97 @@ def create_dataset_from_pdbs(pdbfiles: str, label_prefix: Optional[str] = None) datasets.append(dataset) return datasets + + +def parse_repeated_position_datasets_from_directory( + root: str, + traj_pattern: str, + pdb_pattern: Optional[str] = None, + pdb_file: Optional[Sequence[str]] = None, + max_datasets: Optional[int] = None, + max_datasets_offset: Optional[int] = None, + filter_codes: Optional[Sequence[str]] = None, + as_iterable: bool = False, + label_override: Optional[str] = None, + **dataset_kwargs, +) -> List: + """Helper function to create RepeatedPositionDataset objects from a directory of trajectory files.""" + # Import here to avoid circular imports + from jamun.data.noisy_position_dataset import RepeatedPositionDataset + + # Print the dataset_kwargs for debugging + print(f"=== parse_repeated_position_datasets_from_directory dataset_kwargs ===") + print(f"dataset_kwargs: {dataset_kwargs}") + print(f"=== End dataset_kwargs ===") + + if pdb_file is not None and pdb_pattern is not None: + raise ValueError("Exactly one of pdb_file and pdb_pattern should be provided.") + + traj_prefix, traj_pattern = os.path.split(traj_pattern) + traj_pattern_compiled = re.compile(traj_pattern) + if "*" in traj_prefix or "?" in traj_prefix: + raise ValueError("traj_prefix should not contain wildcards.") + + traj_files = collections.defaultdict(list) + codes = set() + for entry in os.scandir(os.path.join(root, traj_prefix)): + match = traj_pattern_compiled.match(entry.name) + if not match: + continue + + code = match.group(1) + codes.add(code) + traj_files[code].append(os.path.join(traj_prefix, entry.name)) + + if len(codes) == 0: + raise ValueError("No codes found in directory.") + + pdb_files = {} + if pdb_pattern is not None: + pdb_prefix, pdb_pattern = os.path.split(pdb_pattern) + pdb_pattern_compiled = re.compile(pdb_pattern) + if "*" in pdb_prefix or "?" in pdb_prefix: + raise ValueError("pdb_prefix should not contain wildcards.") + + for entry in os.scandir(os.path.join(root, pdb_prefix)): + match = pdb_pattern_compiled.match(entry.name) + if not match: + continue + + code = match.group(1) + if code not in codes: + continue + pdb_files[code] = os.path.join(pdb_prefix, entry.name) + else: + for code in codes: + pdb_files[code] = pdb_file + + # Filter out codes. + if filter_codes is not None: + codes = [code for code in codes if code in set(filter_codes)] + + # Sort the codes and offset them, if necessary. + codes = list(sorted(codes)) + if max_datasets_offset is not None: + codes = codes[max_datasets_offset:] + if max_datasets is not None: + codes = codes[:max_datasets] + + if as_iterable: + raise ValueError("RepeatedPositionDataset does not support iterable mode") + + datasets = [] + for code in tqdm(codes, desc="Creating RepeatedPositionDatasets"): + if label_override is not None: + print(f"Label override: {label_override}") + code = str(label_override) + + dataset = RepeatedPositionDataset( + root, + traj_files=traj_files[code], + pdb_file=pdb_files[code], + label=code, + **dataset_kwargs, + ) + datasets.append(dataset) + return datasets diff --git a/src/jamun/data/noisy_position_dataset.py b/src/jamun/data/noisy_position_dataset.py new file mode 100644 index 0000000..62687f6 --- /dev/null +++ b/src/jamun/data/noisy_position_dataset.py @@ -0,0 +1,37 @@ +import torch +from jamun.data._mdtraj import MDtrajDataset + + +class RepeatedPositionDataset(MDtrajDataset): + """ + Dataset that replaces hidden states with copies of the current position. + This is used for Model 3 experiment where the structures passed to the denoiser + are copies of the same structure given by y.pos. The denoiser will add noise during training. + """ + + def __init__(self, *args, **kwargs): + """Initialize but store total_lag_time before modifying parent behavior.""" + # Store the total_lag_time for our own use + self._target_total_lag_time = kwargs.get('total_lag_time', 2) + + # Prevent parent from doing lag processing by removing lag parameters + kwargs_no_lag = kwargs.copy() + kwargs_no_lag['total_lag_time'] = None + kwargs_no_lag['lag_subsample_rate'] = None + + super().__init__(*args, **kwargs_no_lag) + + def __getitem__(self, idx): + """Override to create position copies instead of using real hidden states.""" + # Get the normal item from parent class (without lag processing) + graph = super().__getitem__(idx) + + # Create the number of hidden states we want based on our target total_lag_time + num_hidden_states = self._target_total_lag_time - 1 + + graph.hidden_state = [] + for _ in range(num_hidden_states): + # Create a copy of the current position (no noise added here) + graph.hidden_state.append(graph.pos.clone()) + + return graph \ No newline at end of file diff --git a/src/jamun/hydra_config/batch_sampler/mcmc/aboba_memory.yaml b/src/jamun/hydra_config/batch_sampler/mcmc/aboba_memory.yaml new file mode 100644 index 0000000..40ea97d --- /dev/null +++ b/src/jamun/hydra_config/batch_sampler/mcmc/aboba_memory.yaml @@ -0,0 +1,12 @@ +_target_: jamun.sampling.mcmc.ABOBA_memory +delta: ${delta} +friction: ${friction} +steps: ${num_sampling_steps_per_batch} +save_trajectory: true +cpu_offload: true +verbose: true +inverse_temperature: ${inverse_temperature} +score_fn_clip: ${score_fn_clip} +M: ${M} +burn_in_steps: 0 +v_init: "zero" \ No newline at end of file diff --git a/src/jamun/hydra_config/batch_sampler/mcmc/baoab_memory.yaml b/src/jamun/hydra_config/batch_sampler/mcmc/baoab_memory.yaml new file mode 100644 index 0000000..01e4a04 --- /dev/null +++ b/src/jamun/hydra_config/batch_sampler/mcmc/baoab_memory.yaml @@ -0,0 +1,12 @@ +_target_: jamun.sampling.mcmc.BAOAB_memory +delta: ${delta} +friction: ${friction} +steps: ${num_sampling_steps_per_batch} +save_trajectory: true +cpu_offload: true +verbose: true +inverse_temperature: ${inverse_temperature} +score_fn_clip: ${score_fn_clip} +M: ${M} +burn_in_steps: 0 +v_init: "zero" \ No newline at end of file diff --git a/src/jamun/hydra_config/batch_sampler/single_measurement_sampler_memory.yaml b/src/jamun/hydra_config/batch_sampler/single_measurement_sampler_memory.yaml new file mode 100644 index 0000000..a58a9cb --- /dev/null +++ b/src/jamun/hydra_config/batch_sampler/single_measurement_sampler_memory.yaml @@ -0,0 +1,7 @@ +defaults: + - mcmc: baoab_memory.yaml + - callbacks: null + - _self_ + +_target_: jamun.sampling.walkjump.SingleMeasurementSamplerMemory +sigma: ${sigma} diff --git a/src/jamun/hydra_config/model/denoiser_conditional.yaml b/src/jamun/hydra_config/model/denoiser_conditional.yaml index 386c72c..59143b3 100644 --- a/src/jamun/hydra_config/model/denoiser_conditional.yaml +++ b/src/jamun/hydra_config/model/denoiser_conditional.yaml @@ -22,8 +22,4 @@ conditioner: _target_: jamun.model.conditioners.PositionConditioner N_structures: ${model.arch.N_structures} -multimeasurement: false -N_measurements_hidden: 1 -N_measurements: 1 - _target_: jamun.model.denoiser_conditional.Denoiser \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_conditional_pretrained.yaml b/src/jamun/hydra_config/model/denoiser_conditional_pretrained.yaml new file mode 100644 index 0000000..b473d6b --- /dev/null +++ b/src/jamun/hydra_config/model/denoiser_conditional_pretrained.yaml @@ -0,0 +1,2 @@ +_target_: jamun.model.denoiser_conditional.Denoiser.load_from_checkpoint +checkpoint_path: null diff --git a/src/jamun/hydra_config/sample_memory.yaml b/src/jamun/hydra_config/sample_memory.yaml new file mode 100644 index 0000000..4d00052 --- /dev/null +++ b/src/jamun/hydra_config/sample_memory.yaml @@ -0,0 +1,27 @@ +defaults: + - _self_ + - model: denoiser_conditional_pretrained + - batch_sampler: single_measurement_sampler_memory + - logger: default + - paths: default + - hydra: default + - callbacks: sampler/default + - experiment: sample_uncapped_single_shape_conditioning + +float32_matmul_precision: "high" + +sample_pdb: null +repeat_init_samples: 1 +num_batches: 1 +continue_chain: true +finetune_on_init: false + +seed: 42 +task_name: "sample" +run_group: "dev" +run_key: ${now:%Y-%m-%d}_${now:%H-%M-%S} # NOTE in DDP this must be set consistently across ranks + +sampler: + _target_: jamun.sampling.SamplerMemory + _convert_: "partial" # loggers argument must be passed as plain list + precision: "32-true" diff --git a/src/jamun/hydra_config/train.yaml b/src/jamun/hydra_config/train.yaml index 3255d93..92623cd 100644 --- a/src/jamun/hydra_config/train.yaml +++ b/src/jamun/hydra_config/train.yaml @@ -12,6 +12,7 @@ defaults: float32_matmul_precision: "high" +seed: 42 task_name: "train" run_group: "dev" run_key: ${now:%Y-%m-%d}_${now:%H-%M-%S} # NOTE in DDP this must be set consistently across ranks diff --git a/src/jamun/model/conditioners/__init__.py b/src/jamun/model/conditioners/__init__.py index 768bbac..56292f6 100644 --- a/src/jamun/model/conditioners/__init__.py +++ b/src/jamun/model/conditioners/__init__.py @@ -1,2 +1,2 @@ -from .conditioners import Conditioner, PositionConditioner, SelfConditioner +from .conditioners import Conditioner, PositionConditioner, SelfConditioner, MeanConditioner diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 56baedc..2dc2445 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -22,14 +22,18 @@ class PositionConditioner(pl.LightningModule): """ Condition the hidden state on the position of the structure. """ - def __init__(self, N_structures: int, **kwargs): + def __init__(self, N_structures: int, align_hidden_states: bool = True, **kwargs): super().__init__() self.N_structures = N_structures + self.align_hidden_states = align_hidden_states def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: - conditioned_structures = [] + conditioned_structures = [y.pos] # Start with current position for positions in y.hidden_state: - aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) - conditioned_structures.append(aligned_positions) + if self.align_hidden_states: + aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) + conditioned_structures.append(aligned_positions) + else: + conditioned_structures.append(positions) return conditioned_structures class SelfConditioner(pl.LightningModule): @@ -40,5 +44,32 @@ def __init__(self, N_structures: int, **kwargs): super().__init__() self.N_structures = N_structures def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: - conditioned_structures = [y.pos for _ in range(self.N_structures-1)] + conditioned_structures = [y.pos for _ in range(self.N_structures)] # Include current position + return conditioned_structures + +class MeanConditioner(pl.LightningModule): + """ + Condition on the mean across time steps of positions and hidden states. + For each atom and coordinate, averages across all T+1 structures (current + hidden states). + """ + def __init__(self, N_structures: int, **kwargs): + super().__init__() + self.N_structures = N_structures + + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + # Start with current position + all_positions = [y.pos] + + # Add all hidden states if they exist + if hasattr(y, "hidden_state") and y.hidden_state is not None: + all_positions.extend(y.hidden_state) + + # Stack all positions along a new dimension and compute mean across time steps + # Shape: (T+1, N, 3) -> (N, 3) where T is number of hidden states + stacked_positions = torch.stack(all_positions, dim=0) # (T+1, N, 3) + mean_positions = torch.mean(stacked_positions, dim=0) # (N, 3) + + # Return the mean repeated N_structures times + conditioned_structures = [mean_positions for _ in range(self.N_structures)] + return conditioned_structures \ No newline at end of file diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index 7b900eb..a6ab019 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -6,6 +6,7 @@ import torch import torch_geometric from e3tools import radius_graph, scatter +from tqdm import tqdm from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm @@ -34,9 +35,6 @@ def __init__( use_torch_compile: bool = True, torch_compile_kwargs: Optional[Dict] = None, conditioner: Callable[..., list[torch.Tensor]] = None, - multimeasurement: bool = False, - N_measurements_hidden: int = 1, - N_measurements: int = 1, ): super().__init__() self.save_hyperparameters(logger=False) @@ -107,16 +105,8 @@ def __init__( raise ValueError("Conditioner must be a callable or None") py_logger.info(f"Conditioner: {self.conditioning_module}") - self.multimeasurement = multimeasurement - self.N_measurements_hidden = N_measurements_hidden - self.N_measurements = N_measurements - def conditioner_default(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: - conditioned_structures = [] - # for positions in y.hidden_state: - # aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) - # conditioned_structures.append(aligned_positions) - conditioned_structures.append(y.pos) + conditioned_structures = [y.pos] # Return complete list starting with current position return conditioned_structures def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: @@ -127,6 +117,31 @@ def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: else: raise ValueError("Conditioner must be a callable or None") + def _align_A_to_B_batched_with_hidden_states( + self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch + ) -> torch_geometric.data.Batch: + """Aligns each graph of A to the corresponding graph in B, including hidden states.""" + A_aligned = A.clone() + + # Align positions + A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) + + # Align hidden states + if hasattr(A, "hidden_state") and A.hidden_state is not None: + A_aligned.hidden_state = [] + for i in range(len(A.hidden_state)): + A_aligned.hidden_state.append(kabsch_algorithm( + A.hidden_state[i], B.pos, A.batch, A.num_graphs + )) + return A_aligned + + def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): + if hasattr(data, "hidden_state") and data.hidden_state is not None: + for i in range(len(data.hidden_state)): + mean = scatter(data.hidden_state[i], data.batch, dim=0, reduce="mean") + data.hidden_state[i] = data.hidden_state[i] - mean[data.batch] + return data + def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.pos.ndim) @@ -151,59 +166,11 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] y.pos = x.pos + sigma * noise for i in range(len(y.hidden_state)): - y.hidden_state[i] = x.hidden_state[i] + hidden_noise[i] + y.hidden_state[i] = x.hidden_state[i] + sigma * hidden_noise[i] if torch.rand(()) < self.mirror_augmentation_rate: y.pos = -y.pos return y - def add_noise_hiddens( - self, - x: torch_geometric.data.Batch, - N_measurements_hidden: int, - N_measurements: int, - sigma: Union[float, torch.Tensor], - ) -> torch_geometric.data.Batch: - """ - Makes N_measurements_hidden number of noisy copies of the hidden states of x - and then for every noisy copy, makes N_measurements number of noisy copies of the positions of x. - - Args: - x (Batch): A torch_geometric Batch object. Must have `pos` and `hidden_state` attributes. - `hidden_state` is expected to be a list of tensors. - N_measurements_hidden (int): Number of noisy copies of hidden states. - N_measurements (int): Number of noisy copies of positions for each noisy hidden state. - sigma (float or torch.Tensor): The standard deviation of the Gaussian noise to add. - - Returns: - Batch: A new Batch object containing all the noisy copies. - """ - x_list = x.to_data_list() - noisy_y_list = [] - - for graph in x_list: - for _ in range(N_measurements_hidden): - # Create a noisy version of the hidden state - noisy_hidden_state = [] - if hasattr(graph, "hidden_state") and graph.hidden_state is not None: - for hs_tensor in graph.hidden_state: - noise = torch.randn_like(hs_tensor) * sigma - noisy_hidden_state.append(hs_tensor + noise) - - for _ in range(N_measurements): - noisy_graph = graph.clone() - - # Add noise to positions - pos_noise = torch.randn_like(graph.pos) * sigma - noisy_graph.pos = graph.pos + pos_noise - - # Assign the noisy hidden state - if hasattr(graph, "hidden_state") and graph.hidden_state is not None: - noisy_graph.hidden_state = [hs.clone() for hs in noisy_hidden_state] - - noisy_y_list.append(noisy_graph) - - return torch_geometric.data.Batch.from_data_list(noisy_y_list) - def score(self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: """Compute the score function.""" sigma = torch.as_tensor(sigma).to(y.pos) @@ -299,22 +266,28 @@ def xhat_normalized( with torch.cuda.nvtx.range("scale_y"): y_scaled = y.clone() y_scaled.pos = y.pos * c_in - scaled_hidden_state = [] - for positions in y_scaled.hidden_state: - scaled_hidden_state.append(positions * c_in) - y_scaled.hidden_state = scaled_hidden_state + # Manually copy hidden state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + y_scaled.hidden_state = [] + for positions in y.hidden_state: + y_scaled.hidden_state.append(positions * c_in) with torch.cuda.nvtx.range("clone_y"): xhat = y.clone() + # Manually copy hidden state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [h.clone() for h in y.hidden_state] with torch.cuda.nvtx.range("conditioning"): conditioned_structures = self.conditioner(y_scaled) with torch.cuda.nvtx.range("g"): - g_pred = self.g(torch.cat([y_scaled.pos, *conditioned_structures], dim=-1), topology=y_scaled, \ + g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), topology=y_scaled, \ c_noise=c_noise, effective_radial_cutoff=radial_cutoff) xhat.pos = c_skip * y.pos + c_out * g_pred + if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] return xhat def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): @@ -322,6 +295,7 @@ def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): y = mean_center(y) + y = self._mean_center_hidden_states(y) with torch.cuda.nvtx.range("xhat_normalized"): xhat = self.xhat_normalized(y, sigma) @@ -347,41 +321,28 @@ def noise_and_denoise( if self.mean_center: # Operate on a clone to avoid side effects on the original batch object. x_processed = mean_center(x) + x_processed = self._mean_center_hidden_states(x_processed) else: x_processed = x sigma = torch.as_tensor(sigma).to(x_processed.pos) - if self.multimeasurement: - with torch.cuda.nvtx.range("add_noise_hiddens"): - y = self.add_noise_hiddens( - x_processed, self.N_measurements_hidden, self.N_measurements, sigma - ) - - # Repeat x_processed to match y's batch size for alignment and loss calculation. - x_list = x_processed.to_data_list() - repeated_x_list = [ - graph.clone() - for graph in x_list - for _ in range(self.N_measurements_hidden * self.N_measurements) - ] - x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to( - x_processed.pos.device - ) - - else: - with torch.cuda.nvtx.range("add_noise"): - y = self.add_noise(x_processed, sigma) - x_target = x_processed.clone() + with torch.cuda.nvtx.range("add_noise"): + y = self.add_noise(x_processed, sigma) + x_target = x_processed.clone() + # Manually copy hidden state + if hasattr(x_processed, "hidden_state") and x_processed.hidden_state is not None: + x_target.hidden_state = [h.clone() for h in x_processed.hidden_state] if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): y = mean_center(y) + y = self._mean_center_hidden_states(y) # Aligning each batch. if align_noisy_input: with torch.cuda.nvtx.range("align_A_to_B_batched"): - y = align_A_to_B_batched(y, x_target) + y = self._align_A_to_B_batched_with_hidden_states(y, x_target) with torch.cuda.nvtx.range("xhat"): xhat = self.xhat(y, sigma) @@ -435,15 +396,15 @@ def noise_and_compute_loss( x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) return self.compute_loss(x_target, xhat, sigma) - def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): - """Called during training.""" - with torch.cuda.nvtx.range("sample_sigma"): - sigma = self.sigma_distribution.sample().to(self.device) - + def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): + """The standard step for automatic optimization.""" + align_noisy_input = self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation + sigma = self.sigma_distribution.sample().to(self.device) + loss, aux = self.noise_and_compute_loss( batch, sigma, - align_noisy_input=self.align_noisy_input_during_training, + align_noisy_input=align_noisy_input, ) # Average the loss and other metrics over all graphs. @@ -451,31 +412,29 @@ def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): aux["loss"] = loss for key in aux: aux[key] = aux[key].mean() + if stage == "train": + self.log(f"train/{key}", aux[key], prog_bar=False, batch_size=batch.num_graphs, sync_dist=False) + elif stage == "val": + self.log( + f"val/{key}", aux[key], prog_bar=(key == "scaled_rmsd"), batch_size=batch.num_graphs, sync_dist=True + ) + else: + continue - self.log(f"train/{key}", aux[key], prog_bar=False, batch_size=batch.num_graphs, sync_dist=False) return { "sigma": sigma, **aux, } - def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): - """Called during validation.""" - sigma = self.sigma_distribution.sample().to(self.device) - loss, aux = self.noise_and_compute_loss(batch, sigma, align_noisy_input=self.align_noisy_input_during_training) - # Average the loss and other metrics over all graphs. - aux["loss"] = loss - for key in aux: - aux[key] = aux[key].mean() - self.log( - f"val/{key}", aux[key], prog_bar=(key == "scaled_rmsd"), batch_size=batch.num_graphs, sync_dist=True - ) + def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during training.""" + return self._automatic_step(batch, "train") - return { - "sigma": sigma, - **aux, - } + def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during validation.""" + self._automatic_step(batch, "val") def configure_optimizers(self): """Set up the optimizer and learning rate scheduler.""" diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py new file mode 100644 index 0000000..1b8ec7f --- /dev/null +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -0,0 +1,691 @@ +import logging +from typing import Callable, Dict, Optional, Tuple, Union + +import lightning.pytorch as pl +import numpy as np +import torch +import torch_geometric +from e3tools import radius_graph, scatter +from tqdm import tqdm + +from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils.align import kabsch_algorithm + + +class DenoiserMultimeasurement(pl.LightningModule): + """The main denoiser mode with conditional architecture.""" + + def __init__( + self, + arch: Callable[..., torch.nn.Module], + optim: Callable[..., torch.optim.Optimizer], + sigma_distribution: torch.distributions.Distribution, + max_radius: float, + average_squared_distance: float, + add_fixed_noise: bool, + add_fixed_ones: bool, + align_noisy_input_during_training: bool, + align_noisy_input_during_evaluation: bool, + mean_center: bool, + mirror_augmentation_rate: float, + bond_loss_coefficient: float = 1.0, + normalization_type: Optional[str] = "JAMUN", + sigma_data: Optional[float] = None, # Only used if normalization_type is "EDM" + lr_scheduler_config: Optional[Dict] = None, + use_torch_compile: bool = True, + torch_compile_kwargs: Optional[Dict] = None, + conditioner: Callable[..., list[torch.Tensor]] = None, + multimeasurement: bool = False, + N_measurements_hidden: int = 1, + N_measurements: int = 1, + max_graphs_per_batch: int = None, + ): + super().__init__() + self.save_hyperparameters(logger=False) + + # Let us control the optimization process only if we need to chunk batches. + self.automatic_optimization = max_graphs_per_batch is None + + self.g = arch() + if use_torch_compile: + if torch_compile_kwargs is None: + torch_compile_kwargs = {} + + self.g = torch.compile(self.g, **torch_compile_kwargs) + + py_logger = logging.getLogger("jamun") + py_logger.info(self.g) + + self.optim_factory = optim + self.lr_scheduler_config = lr_scheduler_config + self.sigma_distribution = sigma_distribution + self.max_radius = max_radius + + self.add_fixed_noise = add_fixed_noise + self.add_fixed_ones = add_fixed_ones + if self.add_fixed_noise and self.add_fixed_ones: + raise ValueError("Can't add fixed noise and fixed ones at the same time") + if self.add_fixed_noise: + py_logger.info("Adding fixed noise") + if self.add_fixed_ones: + py_logger.info("Adding fixed ones") + + self.average_squared_distance = average_squared_distance + py_logger.info(f"Average squared distance = {self.average_squared_distance}") + + self.align_noisy_input_during_training = align_noisy_input_during_training + if self.align_noisy_input_during_training: + py_logger.info("Aligning noisy input during training.") + else: + py_logger.info("Not aligning noisy input during training.") + + self.align_noisy_input_during_evaluation = align_noisy_input_during_evaluation + if self.align_noisy_input_during_evaluation: + py_logger.info("Aligning noisy input during evaluation.") + else: + py_logger.info("Not aligning noisy input during evaluation.") + + self.mean_center = mean_center + if self.mean_center: + py_logger.info("Mean centering input and output.") + else: + py_logger.info("Not mean centering input and output.") + + self.mirror_augmentation_rate = mirror_augmentation_rate + py_logger.info(f"Mirror augmentation rate: {self.mirror_augmentation_rate}") + + self.normalization_type = normalization_type + if self.normalization_type is not None: + py_logger.info(f"Normalization type: {self.normalization_type}") + else: + py_logger.info("No normalization") + + self.sigma_data = sigma_data + if self.normalization_type == "EDM" and self.sigma_data is None: + raise ValueError("sigma_data must be provided when normalization_type is 'EDM'") + elif self.normalization_type != "EDM" and self.sigma_data is not None: + raise ValueError("sigma_data can only be used when normalization_type is 'EDM'") + + self.bond_loss_coefficient = bond_loss_coefficient + self.conditioning_module = conditioner + if self.conditioning_module is not None and not callable(self.conditioning_module): + raise ValueError("Conditioner must be a callable or None") + py_logger.info(f"Conditioner: {self.conditioning_module}") + + self.multimeasurement = multimeasurement + self.N_measurements_hidden = N_measurements_hidden + self.N_measurements = N_measurements + self.max_graphs_per_batch = max_graphs_per_batch + if not self.automatic_optimization: + py_logger.info( + f"Manual optimization enabled with micro-batch size of {self.max_graphs_per_batch} graphs." + ) + + def _align_A_to_B_batched_with_hidden_states( + self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch + ) -> torch_geometric.data.Batch: + """Aligns each graph of A to the corresponding graph in B, including hidden states.""" + A_aligned = A.clone() + + # Align positions + A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) + + # Align hidden states + if hasattr(A, "hidden_state") and A.hidden_state is not None: + for i in range(len(A.hidden_state)): + A_aligned.hidden_state[i] = kabsch_algorithm( + A.hidden_state[i], B.pos, A.batch, A.num_graphs + ) + return A_aligned + + def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): + if hasattr(data, "hidden_state") and data.hidden_state is not None: + for i in range(len(data.hidden_state)): + mean = scatter(data.hidden_state[i], data.batch, dim=0, reduce="mean") + data.hidden_state[i] = data.hidden_state[i] - mean[data.batch] + return data + + def _prepare_noisy_batch( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ): + """Prepare a batch of noisy graphs and their targets.""" + with torch.no_grad(): + if self.mean_center: + x_processed = mean_center(x) + x_processed = self._mean_center_hidden_states(x_processed) + else: + x_processed = x + + sigma_tensor = torch.as_tensor(sigma).to(x_processed.pos.device) + + y = self.add_noise_hiddens( + x_processed, self.N_measurements_hidden, self.N_measurements, sigma_tensor + ) + + x_list = x_processed.to_data_list() + repeated_x_list = [ + graph.clone() + for graph in x_list + for _ in range(self.N_measurements_hidden * self.N_measurements) + ] + x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to( + x_processed.pos.device + ) + + if self.mean_center: + y = mean_center(y) + y = self._mean_center_hidden_states(y) + + if align_noisy_input: + y = self._align_A_to_B_batched_with_hidden_states(y, x_target) + + return y, x_target + + def conditioner_default(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + conditioned_structures = [y.pos] # Return complete list starting with current position + return conditioned_structures + + def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + if self.conditioning_module is None: + return self.conditioner_default(y) + elif callable(self.conditioning_module): + return self.conditioning_module(y) + else: + raise ValueError("Conditioner must be a callable or None") + + def add_noise( + self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] + ) -> torch_geometric.data.Batch: + # pos [B, ...] + sigma = unsqueeze_trailing(sigma, x.pos.ndim) + + y = x.clone() + if self.add_fixed_ones: + noise = torch.ones_like(x.pos) + hidden_noise = [ + torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) + ] + elif self.add_fixed_noise: + torch.manual_seed(0) + num_batches = x.batch.max().item() + 1 + if len(x.pos.shape) == 2: + num_nodes_per_batch = x.pos.shape[0] // num_batches + noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) + hidden_noise = [ + torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat( + num_batches, 1 + ) + for i in range(len(x.hidden_state)) + ] + if len(x.pos.shape) == 3: + num_nodes_per_batch = x.pos.shape[1] + noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) + hidden_noise = [ + torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) + for i in range(len(x.hidden_state)) + ] + else: + noise = torch.randn_like(x.pos) + hidden_noise = [ + torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) + ] + y.pos = x.pos + sigma * noise + for i in range(len(y.hidden_state)): + y.hidden_state[i] = x.hidden_state[i] + hidden_noise[i] + if torch.rand(()) < self.mirror_augmentation_rate: + y.pos = -y.pos + return y + + def add_noise_hiddens( + self, + x: torch_geometric.data.Batch, + N_measurements_hidden: int, + N_measurements: int, + sigma: Union[float, torch.Tensor], + ) -> torch_geometric.data.Batch: + """ + Makes N_measurements_hidden number of noisy copies of the hidden states of x + and then for every noisy copy, makes N_measurements number of noisy copies of the positions of x. + + Args: + x (Batch): A torch_geometric Batch object. Must have `pos` and `hidden_state` attributes. + `hidden_state` is expected to be a list of tensors. + N_measurements_hidden (int): Number of noisy copies of hidden states. + N_measurements (int): Number of noisy copies of positions for each noisy hidden state. + sigma (float or torch.Tensor): The standard deviation of the Gaussian noise to add. + + Returns: + Batch: A new Batch object containing all the noisy copies. + """ + x_list = x.to_data_list() + noisy_y_list = [] + + for graph in x_list: + for _ in range(N_measurements_hidden): + # Create a noisy version of the hidden state + noisy_hidden_state = [] + if hasattr(graph, "hidden_state") and graph.hidden_state is not None: + for hs_tensor in graph.hidden_state: + noise = torch.randn_like(hs_tensor) * sigma + noisy_hidden_state.append(hs_tensor + noise) + + for _ in range(N_measurements): + noisy_graph = graph.clone() + + # Add noise to positions + pos_noise = torch.randn_like(graph.pos) * sigma + noisy_graph.pos = graph.pos + pos_noise + + # Assign the noisy hidden state + if hasattr(graph, "hidden_state") and graph.hidden_state is not None: + noisy_graph.hidden_state = [hs.clone() for hs in noisy_hidden_state] + + noisy_y_list.append(noisy_graph) + + return torch_geometric.data.Batch.from_data_list(noisy_y_list) + + def score( + self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] + ) -> torch_geometric.data.Batch: + """Compute the score function.""" + sigma = torch.as_tensor(sigma).to(y.pos) + return (self.xhat(y, sigma).pos - y.pos) / ( + unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2 + ) + + def normalization_factors( + self, sigma: float, D: int = 3 + ) -> Tuple[float, float, float, float]: + """Normalization factors for the input and output.""" + sigma = torch.as_tensor(sigma) + + if self.normalization_type is None: + return 1.0, 0.0, 1.0, sigma + + if self.normalization_type == "EDM": + c_skip = (self.sigma_data**2) / (sigma**2 + self.sigma_data**2) + c_out = sigma * self.sigma_data / torch.sqrt(self.sigma_data**2 + sigma**2) + c_in = 1 / torch.sqrt(sigma**2 + self.sigma_data**2) + c_noise = torch.log(sigma / self.sigma_data) * 0.25 + return c_in, c_skip, c_out, c_noise + + if self.normalization_type == "JAMUN": + A = torch.as_tensor(self.average_squared_distance) + B = torch.as_tensor(2 * D * sigma**2) + + c_in = 1.0 / torch.sqrt(A + B) + c_skip = A / (A + B) + c_out = torch.sqrt((A * B) / (A + B)) + c_noise = torch.log(sigma) / 4 + return c_in, c_skip, c_out, c_noise + + raise ValueError(f"Unknown normalization type: {self.normalization_type}") + + def loss_weight(self, sigma: float, D: int = 3) -> float: + """Loss weight for this graph.""" + _, _, c_out, _ = self.normalization_factors(sigma, D) + return 1 / (c_out**2) + + def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Tensor: + """Compute the effective radial cutoff for the noise level.""" + return torch.sqrt((self.max_radius**2) + 6 * (sigma**2)) + + def add_edges( + self, y: torch_geometric.data.Batch, radial_cutoff: float + ) -> torch_geometric.data.Batch: + """Add edges to the graph based on the effective radial cutoff.""" + if "batch" in y: + batch = y["batch"] + else: + batch = torch.zeros(y.num_nodes, dtype=torch.long, device=self.device) + + # Our dataloader already adds the bonded edges. + bonded_edge_index = y.edge_index + + with torch.cuda.nvtx.range("radial_graph"): + radial_edge_index = radius_graph(y.pos, radial_cutoff, batch) + + with torch.cuda.nvtx.range("concatenate_edges"): + edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) + if bonded_edge_index.numel() == 0: + bond_mask = torch.zeros( + radial_edge_index.shape[1], dtype=torch.long, device=self.device + ) + else: + bond_mask = torch.cat( + ( + torch.zeros( + radial_edge_index.shape[1], dtype=torch.long, device=self.device + ), + torch.ones( + bonded_edge_index.shape[1], dtype=torch.long, device=self.device + ), + ), + dim=0, + ) + + y.edge_index = edge_index + y.bond_mask = bond_mask + return y + + def xhat_normalized( + self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] + ) -> torch_geometric.data.Batch: + """Compute the denoised prediction using the normalization factors from JAMUN.""" + sigma = torch.as_tensor(sigma).to(y.pos) + D = y.pos.shape[-1] + + # Compute the normalization factors. + with torch.cuda.nvtx.range("normalization_factors"): + c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) + radial_cutoff = self.effective_radial_cutoff(sigma) / c_in + + # Adjust dimensions. + c_in = unsqueeze_trailing(c_in, y.pos.ndim - 1) + c_skip = unsqueeze_trailing(c_skip, y.pos.ndim - 1) + c_out = unsqueeze_trailing(c_out, y.pos.ndim - 1) + c_noise = c_noise.unsqueeze(0) + + # Add edges to the graph. + with torch.cuda.nvtx.range("add_edges"): + y = self.add_edges(y, radial_cutoff) + + with torch.cuda.nvtx.range("scale_y"): + y_scaled = y.clone() + y_scaled.pos = y.pos * c_in + scaled_hidden_state = [] + for positions in y_scaled.hidden_state: + scaled_hidden_state.append(positions * c_in) + y_scaled.hidden_state = scaled_hidden_state + + with torch.cuda.nvtx.range("clone_y"): + xhat = y.clone() + + with torch.cuda.nvtx.range("conditioning"): + conditioned_structures = self.conditioner(y_scaled) + + with torch.cuda.nvtx.range("g"): + g_pred = self.g( + torch.cat([*conditioned_structures], dim=-1), + topology=y_scaled, + c_noise=c_noise, + effective_radial_cutoff=radial_cutoff, + ) + + xhat.pos = c_skip * y.pos + c_out * g_pred + if hasattr(y, "hidden_state") and xhat.hidden_state is not None: + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] # the hidden state updates! + return xhat + + def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): + """Compute the denoised prediction.""" + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_y"): + y = mean_center(y) + y = self._mean_center_hidden_states(y) + + with torch.cuda.nvtx.range("xhat_normalized"): + xhat = self.xhat_normalized(y, sigma) + + # Mean center the prediction. + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_xhat"): + xhat = mean_center(xhat) + xhat = self._mean_center_hidden_states(xhat) + + return xhat + + def noise_and_denoise( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: + """ + Add noise to the input and denoise it. + Returns the target for the loss, the prediction, and the noisy input. + """ + with torch.no_grad(): + if self.mean_center: + # Operate on a clone to avoid side effects on the original batch object. + x_processed = mean_center(x) + else: + x_processed = x + + sigma = torch.as_tensor(sigma).to(x_processed.pos) + + if self.multimeasurement: + with torch.cuda.nvtx.range("add_noise_hiddens"): + y = self.add_noise_hiddens( + x_processed, self.N_measurements_hidden, self.N_measurements, sigma + ) + + # Repeat x_processed to match y's batch size for alignment and loss calculation. + x_list = x_processed.to_data_list() + repeated_x_list = [ + graph.clone() + for graph in x_list + for _ in range(self.N_measurements_hidden * self.N_measurements) + ] + x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to( + x_processed.pos.device + ) + + else: + with torch.cuda.nvtx.range("add_noise"): + y = self.add_noise(x_processed, sigma) + x_target = x_processed.clone() + + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_y"): + y = mean_center(y) + + # Aligning each batch. + if align_noisy_input: + with torch.cuda.nvtx.range("align_A_to_B_batched"): + y = self._align_A_to_B_batched_with_hidden_states(y, x_target) + + with torch.cuda.nvtx.range("xhat"): + xhat = self.xhat(y, sigma) + + return x_target, xhat, y + + def compute_loss( + self, + x: torch_geometric.data.Batch, + xhat: torch.Tensor, + sigma: Union[float, torch.Tensor], + ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + """Compute the loss.""" + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_x"): + x = mean_center(x) + x = self._mean_center_hidden_states(x) + + D = xhat.pos.shape[-1] + + # Compute the raw loss. + with torch.cuda.nvtx.range("raw_coordinate_loss"): + raw_coordinate_loss = (xhat.pos - x.pos).pow(2).sum(dim=-1) + + # Take the mean over each graph. + with torch.cuda.nvtx.range("mean_over_graphs"): + mse = scatter(raw_coordinate_loss, x.batch, dim=0, dim_size=x.num_graphs, reduce="mean") + + # Compute the scaled RMSD. + with torch.cuda.nvtx.range("scaled_rmsd"): + rmsd = torch.sqrt(mse) + scaled_rmsd = rmsd / (sigma * np.sqrt(D)) + + # Account for the loss weight across graphs and noise levels. + with torch.cuda.nvtx.range("loss_weight"): + loss = mse * x.loss_weight + loss = loss * self.loss_weight(sigma, D) + + return loss, { + "mse": mse, + "rmsd": rmsd, + "scaled_rmsd": scaled_rmsd, + } + + def noise_and_compute_loss( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + """Add noise to the input and compute the loss.""" + x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) + return self.compute_loss(x_target, xhat, sigma) + + def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): + """The standard step for automatic optimization.""" + align_noisy_input = ( + self.align_noisy_input_during_training + if stage == "train" + else self.align_noisy_input_during_evaluation + ) + sigma = self.sigma_distribution.sample().to(self.device) + + loss, aux = self.noise_and_compute_loss( + batch, + sigma, + align_noisy_input=align_noisy_input, + ) + + # Average the loss and other metrics over all graphs. + with torch.cuda.nvtx.range("mean_over_graphs"): + aux["loss"] = loss + for key in aux: + aux[key] = aux[key].mean() + if stage == "train": + self.log( + f"train/{key}", + aux[key], + prog_bar=False, + batch_size=batch.num_graphs, + sync_dist=False, + ) + elif stage == "val": + self.log( + f"val/{key}", + aux[key], + prog_bar=(key == "scaled_rmsd"), + batch_size=batch.num_graphs, + sync_dist=True, + ) + else: + continue + + return { + "sigma": sigma, + **aux, + } + + def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): + """A shared step for training and validation with manual optimization.""" + sigma = self.sigma_distribution.sample().to(self.device) + align_noisy_input = ( + self.align_noisy_input_during_training + if stage == "train" + else self.align_noisy_input_during_evaluation + ) + + y, x_target = self._prepare_noisy_batch(batch, sigma, align_noisy_input) + + y_list = y.to_data_list() + x_target_list = x_target.to_data_list() + + chunk_size = self.max_graphs_per_batch + num_chunks = (len(y_list) + chunk_size - 1) // chunk_size + + all_aux = [] + opt = self.optimizers() if stage == "train" else None + + print(f"Processing {num_chunks} chunks of size {chunk_size} for {stage}...") + for i in tqdm(range(num_chunks)): + start_index = i * chunk_size + end_index = min(start_index + chunk_size, len(y_list)) + + y_micro_batch_list = y_list[start_index:end_index] + x_target_micro_batch_list = x_target_list[start_index:end_index] + + if not y_micro_batch_list: + continue + + y_micro_batch = torch_geometric.data.Batch.from_data_list(y_micro_batch_list) + x_target_micro_batch = torch_geometric.data.Batch.from_data_list( + x_target_micro_batch_list + ) + + xhat_micro_batch = self.xhat(y_micro_batch, sigma) + + loss, aux = self.compute_loss(x_target_micro_batch, xhat_micro_batch, sigma) + + with torch.cuda.nvtx.range("mean_over_graphs"): + aux["loss"] = loss + for key in aux: + aux[key] = aux[key].mean() + self.log( + f"train/{key}", + aux[key], + prog_bar=False, + batch_size=batch.num_graphs, + sync_dist=False, + ) + if stage == "train": + opt.zero_grad() + self.manual_backward(aux["loss"]) + opt.step() + all_aux.append(aux) + + avg_aux = {} + with torch.no_grad(): + if all_aux: + for key in all_aux[0]: + avg_aux[key] = torch.tensor([d[key] for d in all_aux]).mean() + log_opts = { + "prog_bar": (stage == "val" and "scaled_rmsd" in avg_aux), + "batch_size": len(y_list), + } + if stage == "train": + log_opts["sync_dist"] = True + else: + log_opts["sync_dist"] = True + + for key, value in avg_aux.items(): + self.log(f"{stage}/{key}", value, **log_opts) + + return {"sigma": sigma, **avg_aux} + + def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during training.""" + if self.automatic_optimization: + return self._automatic_step(batch, "train") + else: + print(f"Manual optimization enabled for training step {batch_idx}.") + return self._manual_step(batch, "train") + + def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during validation.""" + if self.automatic_optimization: + self._automatic_step(batch, "val") + else: + self._manual_step(batch, "val") + + def configure_optimizers(self): + """Set up the optimizer and learning rate scheduler.""" + optimizer = self.optim_factory(params=self.parameters()) + + out = {"optimizer": optimizer} + if self.lr_scheduler_config: + scheduler = self.lr_scheduler_config.pop("scheduler") + out["lr_scheduler"] = { + "scheduler": scheduler(optimizer), + **self.lr_scheduler_config, + } + + return out \ No newline at end of file diff --git a/src/jamun/sampling/__init__.py b/src/jamun/sampling/__init__.py index 3f14de8..7e4059e 100644 --- a/src/jamun/sampling/__init__.py +++ b/src/jamun/sampling/__init__.py @@ -1,2 +1,2 @@ from . import diffusion, mcmc, walkjump -from ._sampler import Sampler +from ._sampler import Sampler, SamplerMemory diff --git a/src/jamun/sampling/_sampler.py b/src/jamun/sampling/_sampler.py index 958861f..260dbd4 100644 --- a/src/jamun/sampling/_sampler.py +++ b/src/jamun/sampling/_sampler.py @@ -96,3 +96,57 @@ def sample( self.fabric.log("sampler/global_step", batch_idx) self.fabric.call("on_sample_end", sampler=self) + +class SamplerMemory(Sampler): + """A sampler for molecular dynamics simulations that uses memory.""" + + def sample( + self, + model, + batch_sampler, + num_batches: int, + init_graphs: torch_geometric.data.Data, + continue_chain: bool = False, + ): + + self.fabric.launch() + self.fabric.setup(model) + model.eval() + + init_graphs = init_graphs.to(self.fabric.device) + model_wrapped = utils.ModelSamplingWrapperMemory( + model=model, + init_graphs=init_graphs, + sigma=batch_sampler.sigma, + ) + + y_init = model_wrapped.sample_initial_noisy_positions() + y_hist_init = model_wrapped.sample_initial_noisy_history() + v_init = "gaussian" + + self.fabric.call("on_sample_start", sampler=self) + + batches = torch.arange(num_batches) + iterable = self.progbar_wrapper(batches, desc="Sampling", total=len(batches), leave=False) + + with torch.inference_mode(): + for batch_idx in iterable: + self.global_step = batch_idx + + out = batch_sampler.sample(model=model_wrapped, y_init=y_init, v_init=v_init, y_hist_init=y_hist_init) + samples = model_wrapped.unbatch_samples(out) + + # Start next chain from the end state of the previous chain? + if continue_chain: + y_init = out["y"] + v_init = out["v"] + y_hist_init = out["y_hist"] + else: + y_init = model_wrapped.sample_initial_noisy_positions() + y_hist_init = model_wrapped.sample_initial_noisy_history() + v_init = "gaussian" + + self.fabric.call("on_after_sample_batch", sample=samples, sampler=self) + self.fabric.log("sampler/global_step", batch_idx) + + self.fabric.call("on_sample_end", sampler=self) \ No newline at end of file diff --git a/src/jamun/sampling/mcmc/__init__.py b/src/jamun/sampling/mcmc/__init__.py index b984d4a..1558500 100644 --- a/src/jamun/sampling/mcmc/__init__.py +++ b/src/jamun/sampling/mcmc/__init__.py @@ -1 +1 @@ -from ._splitting import ABOBA, BAOAB +from ._splitting import ABOBA, BAOAB, BAOAB_memory, ABOBA_memory diff --git a/src/jamun/sampling/mcmc/_splitting.py b/src/jamun/sampling/mcmc/_splitting.py index 8338991..f2e52c5 100644 --- a/src/jamun/sampling/mcmc/_splitting.py +++ b/src/jamun/sampling/mcmc/_splitting.py @@ -5,7 +5,7 @@ import torch from torch import Tensor -from jamun.sampling.mcmc.functional import aboba, baoab +from jamun.sampling.mcmc.functional import aboba, baoab, aboba_memory, baoab_memory @dataclass @@ -56,3 +56,17 @@ def __post_init__(self): def __call__(self, y: torch.Tensor, score_fn: Callable, **kwargs): kwargs = dataclasses.asdict(self) | kwargs return baoab(y, score_fn, **kwargs) + + +@dataclass +class ABOBA_memory(ABOBA): + def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): + kwargs = dataclasses.asdict(self) | kwargs + return aboba_memory(y=y, y_hist=y_hist, score_fn=score_fn, **kwargs) + + +@dataclass +class BAOAB_memory(BAOAB): + def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): + kwargs = dataclasses.asdict(self) | kwargs + return baoab_memory(y=y, y_hist=y_hist, score_fn=score_fn, **kwargs) diff --git a/src/jamun/sampling/mcmc/functional/__init__.py b/src/jamun/sampling/mcmc/functional/__init__.py index 37ba462..a91fb53 100644 --- a/src/jamun/sampling/mcmc/functional/__init__.py +++ b/src/jamun/sampling/mcmc/functional/__init__.py @@ -1 +1 @@ -from ._splitting import aboba, baoab +from ._splitting import aboba, baoab, aboba_memory, baoab_memory diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index 55018ee..ca38c7d 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -1,7 +1,7 @@ import logging import math from typing import Callable, Optional, Tuple, Union - +from copy import deepcopy import torch from tqdm.auto import tqdm @@ -26,9 +26,9 @@ def initialize_velocity(v_init: Union[str, torch.Tensor], y: torch.Tensor, u: fl def create_score_fn(score_fn: Callable, inverse_temperature: float, score_fn_clip: Optional[float]) -> Callable: """Create a score function that is clipped and scaled by the inverse temperature.""" - def score_fn_processed(y: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]: + def score_fn_processed(y: torch.Tensor, *args, **kwargs) -> Tuple[torch.Tensor, torch.Tensor]: """Score function clipped and scaled by the inverse temperature.""" - orig_score = score_fn(y).to(dtype=y.dtype) + orig_score = score_fn(y, *args, **kwargs).to(dtype=y.dtype) # Clip the score by norm. score = orig_score if score_fn_clip is not None: @@ -176,3 +176,124 @@ def baoab( score_traj = torch.stack(score_traj) return y, v, y_traj, score_traj + + +def aboba_memory( + y: torch.Tensor, + y_hist: list, + score_fn: Callable, + steps: int, + v_init: Union[str, torch.Tensor] = "zero", + save_trajectory=False, + save_every_n_steps=1, + burn_in_steps=0, + verbose=False, + cpu_offload=False, + delta: float = 1.0, + friction: float = 1.0, + M: float = 1.0, + inverse_temperature: float = 1.0, + score_fn_clip: Optional[float] = None, + **_, +): + """ABOBA splitting scheme that updates a state history.""" + i = 0 + y_traj = [] if save_trajectory else None + score_traj = [] + y_hist_traj = [] + + # Initialize trajectory with initial state + if y_traj is not None and i >= burn_in_steps: + y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) + score_traj.append(torch.zeros_like(y).detach().cpu() if cpu_offload else torch.zeros_like(y).detach()) + y_hist_traj.append(list(y_hist)) + + u = pow(M, -1) + zeta2 = math.sqrt(1 - math.exp(-2 * friction)) + v = initialize_velocity(v_init=v_init, y=y, u=u) + score_fn_processed = create_score_fn(score_fn, inverse_temperature, score_fn_clip) + + steps_iter = range(1, steps) + if verbose: + steps_iter = tqdm(steps_iter, leave=False, desc="ABOBA Memory") + + for i in steps_iter: + y_current = y.clone().detach() + y = y + (delta / 2) * v + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + v = v + u * (delta / 2) * psi + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R + v = vhat + (delta / 2) * psi + y = y + (delta / 2) * v + + if save_trajectory and ((i % save_every_n_steps) == 0) and (i >= burn_in_steps): + y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) + score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) + y_hist_traj.append(list(y_hist)) + + y_hist.pop(-1) + y_hist.insert(0, y_current) + + return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj + + +def baoab_memory( + y: torch.Tensor, + y_hist: list, + score_fn: Callable, + steps: int, + v_init: Union[str, torch.Tensor] = "zero", + save_trajectory=False, + save_every_n_steps=1, + burn_in_steps=0, + verbose=False, + cpu_offload=False, + delta: float = 1.0, + friction: float = 1.0, + M: float = 1.0, + inverse_temperature: float = 1.0, + score_fn_clip: Optional[float] = None, + **_, +): + """BAOAB splitting scheme that updates a state history.""" + i = 0 + y_traj = [] if save_trajectory else None + score_traj = [] + y_hist_traj = [] + + u = pow(M, -1) + zeta2 = math.sqrt(1 - math.exp(-2 * friction)) + v = initialize_velocity(v_init=v_init, y=y, u=u) + score_fn_processed = create_score_fn(score_fn, inverse_temperature, score_fn_clip) + + steps_iter = range(1, steps) + if verbose: + steps_iter = tqdm(steps_iter, leave=False, desc="BAOAB Memory") + + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + + # Initialize trajectory with initial state + if y_traj is not None and i >= burn_in_steps: + y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) + score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) + y_hist_traj.append(list(y_hist)) + + for i in steps_iter: + y_current = y.clone().detach() + v = v + u * (delta / 2) * psi # update with previous psi + y = y + (delta / 2) * v # update with previous v + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R + y = y + (delta / 2) * vhat + y_hist.pop(-1) # remove the last element of the history + y_hist.insert(0, y_current) # present point is the first element of the history + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + v = vhat + (delta / 2) * psi + + if save_trajectory and ((i % save_every_n_steps) == 0) and (i >= burn_in_steps): + y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) + score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) + y_hist_traj.append(list(y_hist)) + + return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj diff --git a/src/jamun/sampling/walkjump/__init__.py b/src/jamun/sampling/walkjump/__init__.py index 8c197ba..baf3c3c 100644 --- a/src/jamun/sampling/walkjump/__init__.py +++ b/src/jamun/sampling/walkjump/__init__.py @@ -1,2 +1,2 @@ from ._callbacks import InterpolateParametersCallback, MeasurementDependentParametersCallback -from ._single_measurement import SingleMeasurementSampler +from ._single_measurement import SingleMeasurementSampler, SingleMeasurementSamplerMemory diff --git a/src/jamun/sampling/walkjump/_single_measurement.py b/src/jamun/sampling/walkjump/_single_measurement.py index 8fcad21..54980a5 100644 --- a/src/jamun/sampling/walkjump/_single_measurement.py +++ b/src/jamun/sampling/walkjump/_single_measurement.py @@ -87,3 +87,95 @@ def sample( out = self.walk_jump(model, batch_size=batch_size, y_init=y_init, v_init=v_init) out["sample"] = out["xhat"] return out + + +class SingleMeasurementSamplerMemory: + """Single Measurement Walk-Jump Sampler.""" + + def __init__( + self, + mcmc, + sigma: float, + y_init_distribution: Optional[torch.distributions.Distribution] = None + ): + self.mcmc = mcmc + self.sigma = float(sigma) + self.y_init_distribution = y_init_distribution + + def walk( + self, + model, + batch_size: Optional[int] = None, + y_init: Optional[torch.Tensor] = None, + v_init: str | Tensor = "gaussian", + y_hist_init: Optional[list] = None, + ): + if y_init is None: + if self.y_init_distribution is None: + raise RuntimeError("either y_init and y_init_distribution must be supplied") + y_init = self.y_init_distribution.sample(sample_shape=(batch_size,)).to(model.device) + if y_hist_init is None: + raise RuntimeError("y_hist_init must be supplied") + y, v, y_hist,y_traj, score_traj, y_hist_traj = self.mcmc(y_init, y_hist_init, lambda y, y_hist: model.score(y, y_hist, self.sigma), \ + v_init=v_init) + + if y_traj is not None: + t_traj = torch.ones(y_traj.size(0), device=y_traj.device, dtype=int) + else: + t_traj = None + + return {"y": y, "v": v, "y_hist": y_hist, "y_traj": y_traj, "t_traj": t_traj, "score_traj": score_traj, "y_hist_traj": y_hist_traj} + + def walk_jump( + self, + model, + batch_size: Optional[int] = None, + y_init: Optional[torch.Tensor] = None, + v_init: str | Tensor = "gaussian", + y_hist_init: Optional[list] = None, + ): + out = self.walk( + model, + batch_size=batch_size, + y_init=y_init, + v_init=v_init, + y_hist_init=y_hist_init, + ) + y, v, y_hist, y_traj, t_traj, score_traj, y_hist_traj = out["y"], out["v"], out["y_hist"], out["y_traj"], out["t_traj"], out["score_traj"], out["y_hist_traj"] + + xhat = model.xhat(y, y_hist, sigma=self.sigma) + + if y_traj is not None: + xhat_traj = torch.stack( + [ + model.xhat(y_traj[i, :].to(model.device), y_hist_traj[i], sigma=self.sigma) + for i in tqdm(range(y_traj.size(0)), leave=False, desc="Jump") + ], + dim=0, + ) + else: + xhat_traj = None + + return { + "xhat": xhat, + "y": y, + "v": v, + "y_hist": y_hist, + "xhat_traj": xhat_traj, + "y_traj": y_traj, + "y_hist_traj": y_hist_traj, + "t_traj": t_traj, + "score_traj": score_traj, + } + + def sample( + self, + model, + batch_size: Optional[int] = None, + y_init: Optional[torch.Tensor] = None, + v_init: str | Tensor = "gaussian", + y_hist_init: Optional[list] = None, + ): + out = self.walk_jump(model, batch_size=batch_size, y_init=y_init, v_init=v_init, y_hist_init=y_hist_init) + out["sample"] = out["xhat"] + return out \ No newline at end of file diff --git a/src/jamun/utils/__init__.py b/src/jamun/utils/__init__.py index 9207558..e884159 100644 --- a/src/jamun/utils/__init__.py +++ b/src/jamun/utils/__init__.py @@ -25,7 +25,7 @@ encode_atom_type, encode_residue, ) -from .sampling_wrapper import ModelSamplingWrapper +from .sampling_wrapper import ModelSamplingWrapper, ModelSamplingWrapperMemory from .scaled_rmsd import scaled_rmsd from .simple_ddp import SimpleDDPStrategy from .singleton import singleton diff --git a/src/jamun/utils/sampling_wrapper.py b/src/jamun/utils/sampling_wrapper.py index c4bf04d..77ac604 100644 --- a/src/jamun/utils/sampling_wrapper.py +++ b/src/jamun/utils/sampling_wrapper.py @@ -5,14 +5,20 @@ import torch.nn as nn import torch_geometric +from jamun.utils import mean_center + class ModelSamplingWrapper: """Wrapper to sample positions from a model.""" - def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float): + def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = False): self._model = model self.init_graphs = init_graphs self.sigma = sigma + + # Apply mean centering if requested + if recenter_on_init: + self.init_graphs = mean_center(self.init_graphs) @property def device(self) -> torch.device: @@ -81,3 +87,101 @@ def unbatch_samples(self, samples: Dict[str, torch.Tensor]) -> List[torch_geomet output_graph[key] = unbatched_value return output_graphs + + +class ModelSamplingWrapperMemory: + """Wrapper for models that depend on a memory of states.""" + + def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = False): + self._model = model + self.init_graphs = init_graphs + self.sigma = sigma + + # Apply mean centering if requested + if recenter_on_init: + # Mean center positions + self.init_graphs = mean_center(self.init_graphs) + + # Mean center hidden states if they exist and aren't empty + if hasattr(self.init_graphs, 'hidden_state') and self.init_graphs.hidden_state: + for i in range(len(self.init_graphs.hidden_state)): + # Mean center each hidden state in-place + self.init_graphs.hidden_state[i] = self.init_graphs.hidden_state[i] - self.init_graphs.hidden_state[i].mean(dim=0, keepdim=True) + + @property + def device(self) -> torch.device: + return next(self._model.parameters()).device + def sample_initial_noisy_positions(self) -> torch.Tensor: + pos = self.init_graphs.pos + pos = pos + torch.randn_like(pos) * self.sigma + return pos + + def sample_initial_noisy_history(self) -> list: + noisy_history = [] + for hidden_state in self.init_graphs.hidden_state: + noisy_history.append(hidden_state + torch.randn_like(hidden_state) * self.sigma) + return noisy_history + + def __getattr__(self, name): + return getattr(self._model, name) + + def score(self, y, y_hist, sigma): + graph = self.positions_to_graph(y, y_hist).to(self.device) + return self._model.score(graph, sigma) + + def xhat(self, y, y_hist, sigma): + graph = self.positions_to_graph(y, y_hist).to(self.device) + xhat_graph = self._model.xhat(graph, sigma) + return xhat_graph.pos + + def positions_to_graph(self, positions: torch.Tensor, y_hist: list) -> torch_geometric.data.Data: + """Wraps positions to a graph and attaches the historical states.""" + assert len(positions) == self.init_graphs.num_nodes + assert positions.shape[1] == 3 + input_graph = self.init_graphs.clone() + input_graph.pos = positions + input_graph.hidden_state = y_hist + return input_graph.to(positions.device) + + def unbatch_samples(self, samples: Dict[str, torch.Tensor]) -> List[torch_geometric.data.Data]: + """Unbatch samples.""" + if "batch" not in self.init_graphs: + raise ValueError("The initial graph does not have a batch attribute.") + + # Copy off the input graphs, to update attributes later. + output_graphs = self.init_graphs.clone() + output_graphs = torch_geometric.data.Batch.to_data_list(output_graphs) + + for key, value in samples.items(): + if key == "y_hist" or key == "y_hist_traj": + if key == "y_hist": + value = [value] + value = torch.stack([torch.stack(traj, dim=1) for traj in value], dim=1) + else: + if hasattr(value, "ndim") and value.ndim not in [2, 3]: + # py_logger = logging.getLogger("jamun") + # py_logger.info(f"Skipping unbatching of key {key} with shape {value.shape} as it is not 2D or 3D.") + continue + if hasattr(value, "ndim") and value.ndim == 3: + value = einops.rearrange( + value, + "num_frames atoms coords -> atoms num_frames coords", + ) + + unbatched_values = torch_geometric.utils.unbatch(value, self.init_graphs.batch) + for output_graph, unbatched_value in zip(output_graphs, unbatched_values, strict=True): + if key in output_graph: + raise ValueError(f"Key {key} already exists in the output graph.") + + if unbatched_value.shape[0] != output_graph.num_nodes: + raise ValueError( + f"Number of nodes in unbatched value ({unbatched_value.shape[0]}) for key {key} does not match " + f"number of nodes in output graph ({output_graph.num_nodes})." + ) + if key == "y_hist": + unbatched_value = [t.squeeze(-2).squeeze(1) for t in torch.split(unbatched_value, 1, dim=-2)] + if key == "y_hist_traj": + unbatched_value = [t.squeeze(-2) for t in torch.split(unbatched_value, 1, dim=-2)] + output_graph[key] = unbatched_value + + return output_graphs \ No newline at end of file diff --git a/test.py b/test.py deleted file mode 100644 index 3b18e51..0000000 --- a/test.py +++ /dev/null @@ -1 +0,0 @@ -hello world diff --git a/wandb/sweep-01dho9mz/config-o4x6y3nm.yaml b/wandb/sweep-01dho9mz/config-o4x6y3nm.yaml deleted file mode 100644 index d1b968c..0000000 --- a/wandb/sweep-01dho9mz/config-o4x6y3nm.yaml +++ /dev/null @@ -1,6 +0,0 @@ -wandb_version: 1 - -model.conditioner: - value: - N_structures: ${model.arch.N_structures} - _target_: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-01dho9mz/config-wrqb60uz.yaml b/wandb/sweep-01dho9mz/config-wrqb60uz.yaml deleted file mode 100644 index dc44beb..0000000 --- a/wandb/sweep-01dho9mz/config-wrqb60uz.yaml +++ /dev/null @@ -1,6 +0,0 @@ -wandb_version: 1 - -model.conditioner: - value: - N_structures: ${model.arch.N_structures} - _target_: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-199gq3m7/config-iw7qu9pc.yaml b/wandb/sweep-199gq3m7/config-iw7qu9pc.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-199gq3m7/config-iw7qu9pc.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-199gq3m7/config-xn0ylvd3.yaml b/wandb/sweep-199gq3m7/config-xn0ylvd3.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-199gq3m7/config-xn0ylvd3.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-2b4ndoa2/config-29fpum8f.yaml b/wandb/sweep-2b4ndoa2/config-29fpum8f.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-2b4ndoa2/config-29fpum8f.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml b/wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-2b4ndoa2/config-h5bhhw56.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-2chv306p/config-dfa7a6n1.yaml b/wandb/sweep-2chv306p/config-dfa7a6n1.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-2chv306p/config-dfa7a6n1.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-2chv306p/config-e15fh1bu.yaml b/wandb/sweep-2chv306p/config-e15fh1bu.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-2chv306p/config-e15fh1bu.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-3945fdiz/config-bj96w41e.yaml b/wandb/sweep-3945fdiz/config-bj96w41e.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-3945fdiz/config-bj96w41e.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-3945fdiz/config-ioayvefo.yaml b/wandb/sweep-3945fdiz/config-ioayvefo.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-3945fdiz/config-ioayvefo.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml b/wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-4y3t4lnj/config-ihttgv3p.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-4y3t4lnj/config-irjpk449.yaml b/wandb/sweep-4y3t4lnj/config-irjpk449.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-4y3t4lnj/config-irjpk449.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-5tie5khj/config-lhzhrmk7.yaml b/wandb/sweep-5tie5khj/config-lhzhrmk7.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-5tie5khj/config-lhzhrmk7.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-5tie5khj/config-lzq16xjh.yaml b/wandb/sweep-5tie5khj/config-lzq16xjh.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-5tie5khj/config-lzq16xjh.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-7h8tmiyj/config-14ueynti.yaml b/wandb/sweep-7h8tmiyj/config-14ueynti.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-7h8tmiyj/config-14ueynti.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml b/wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-7h8tmiyj/config-2pq6zc7e.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-94u6xhkc/config-q8kcodq4.yaml b/wandb/sweep-94u6xhkc/config-q8kcodq4.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-94u6xhkc/config-q8kcodq4.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-94u6xhkc/config-zrd7urfq.yaml b/wandb/sweep-94u6xhkc/config-zrd7urfq.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-94u6xhkc/config-zrd7urfq.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-g101608w/config-htlax8nq.yaml b/wandb/sweep-g101608w/config-htlax8nq.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-g101608w/config-htlax8nq.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-g101608w/config-pp1wtj1t.yaml b/wandb/sweep-g101608w/config-pp1wtj1t.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-g101608w/config-pp1wtj1t.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-gze9gptu/config-sxlo5fvy.yaml b/wandb/sweep-gze9gptu/config-sxlo5fvy.yaml deleted file mode 100644 index d1b968c..0000000 --- a/wandb/sweep-gze9gptu/config-sxlo5fvy.yaml +++ /dev/null @@ -1,6 +0,0 @@ -wandb_version: 1 - -model.conditioner: - value: - N_structures: ${model.arch.N_structures} - _target_: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-gze9gptu/config-vfd1ntbz.yaml b/wandb/sweep-gze9gptu/config-vfd1ntbz.yaml deleted file mode 100644 index dc44beb..0000000 --- a/wandb/sweep-gze9gptu/config-vfd1ntbz.yaml +++ /dev/null @@ -1,6 +0,0 @@ -wandb_version: 1 - -model.conditioner: - value: - N_structures: ${model.arch.N_structures} - _target_: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml b/wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-lxzn3ea1/config-9d3h4wqy.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml b/wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-lxzn3ea1/config-ym4s8uwp.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-mebia8e6/config-0t2gzjry.yaml b/wandb/sweep-mebia8e6/config-0t2gzjry.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-mebia8e6/config-0t2gzjry.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml b/wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-rgkrmz7f/config-fofnbbk4.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml b/wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml deleted file mode 100644 index dc44beb..0000000 --- a/wandb/sweep-rxcjaig2/config-8ox9n7n2.yaml +++ /dev/null @@ -1,6 +0,0 @@ -wandb_version: 1 - -model.conditioner: - value: - N_structures: ${model.arch.N_structures} - _target_: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-rxcjaig2/config-zvdgfxss.yaml b/wandb/sweep-rxcjaig2/config-zvdgfxss.yaml deleted file mode 100644 index d1b968c..0000000 --- a/wandb/sweep-rxcjaig2/config-zvdgfxss.yaml +++ /dev/null @@ -1,6 +0,0 @@ -wandb_version: 1 - -model.conditioner: - value: - N_structures: ${model.arch.N_structures} - _target_: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-x0n6d1ps/config-r1havcvy.yaml b/wandb/sweep-x0n6d1ps/config-r1havcvy.yaml deleted file mode 100644 index 8b7af48..0000000 --- a/wandb/sweep-x0n6d1ps/config-r1havcvy.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.SelfConditioner diff --git a/wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml b/wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-x0n6d1ps/config-uz7qu14d.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner diff --git a/wandb/sweep-z467ihn2/config-r1jppjka.yaml b/wandb/sweep-z467ihn2/config-r1jppjka.yaml deleted file mode 100644 index 63aeac1..0000000 --- a/wandb/sweep-z467ihn2/config-r1jppjka.yaml +++ /dev/null @@ -1,4 +0,0 @@ -wandb_version: 1 - -model.conditioner._target_: - value: jamun.model.conditioners.PositionConditioner From 3a3bf8750607a33513b737859424bcaa9509e2ae Mon Sep 17 00:00:00 2001 From: Vani Date: Sat, 19 Jul 2025 15:00:37 +0000 Subject: [PATCH 12/32] Added a data generation protocol for generating equilibrium structures and then swarms for a set of pdbs. Folder management is a bit janky right now and needs cleanup if you want to do it stepwise or if you want to add more swarms. --- scripts/generate_data/generate_swarms.py | 584 +++++++++++++++++++ scripts/generate_data/openmm_utils.py | 56 +- scripts/generate_data/run_swarms_parallel.sh | 294 ++++++++++ scripts/slurm/generate_swarms_batch_final.sh | 183 ++++++ 4 files changed, 1115 insertions(+), 2 deletions(-) create mode 100755 scripts/generate_data/generate_swarms.py create mode 100755 scripts/generate_data/run_swarms_parallel.sh create mode 100755 scripts/slurm/generate_swarms_batch_final.sh diff --git a/scripts/generate_data/generate_swarms.py b/scripts/generate_data/generate_swarms.py new file mode 100755 index 0000000..32486b6 --- /dev/null +++ b/scripts/generate_data/generate_swarms.py @@ -0,0 +1,584 @@ +#!/usr/bin/env python3 + +import argparse +import logging +import os +import glob +from dataclasses import dataclass +from typing import Optional, List, Tuple +from pathlib import Path + +import openmm_utils as op +from openmm.app import ForceField, Simulation, Topology + +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("generate_swarms") + + +@dataclass +class SwarmConfig: + """Configuration parameters for swarm trajectory generation""" + + # Input/Output + input_pdbs: List[str] + output_dir: str + + # MD simulation parameters (similar to run_simulation.py) + dt_ps: float = 0.002 + temp_K: float = 300 + pressure_bar: float = 1.0 + position_restraint_k: float = 10.0 # kJ/(mol.A^2) + forcefield: tuple[str, str] = ("amber99sbildn.xml", "tip3p.xml") + padding_nm: float = 1.0 + water_model: str = "tip3p" + positive_ion: str = "Na+" + negative_ion: str = "Cl-" + + # Equilibration parameters + energy_minimization_steps: int = 1500 + nvt_restraint_steps: int = 75_000 # Reduced from run_simulation defaults + npt_restraint_steps: int = 75_000 # Reduced from run_simulation defaults + nvt_equil_steps: int = 100_000 # Reduced from run_simulation defaults + npt_equil_steps: int = 100_000 # Reduced from run_simulation defaults + + # Swarm generation parameters + num_swarms: int = 10 + swarm_steps: int = 10_000 + save_frequency: int = 10 + + # Processing options + save_intermediate_files: bool = False + single_structure_mode: bool = False # For processing just one structure + structure_index: Optional[int] = None # For processing a specific structure by index + + # New options for separated workflow + skip_equilibration: bool = False # Skip equilibration if already done + equilibrate_only: bool = False # Only do equilibration, no swarms + append_swarms: bool = True # Start swarm indexing from existing trajectories + + +def parse_args() -> SwarmConfig: + """Parse command line arguments.""" + parser = argparse.ArgumentParser( + description="Generate swarm trajectories from equilibrated structures", + formatter_class=argparse.RawDescriptionHelpFormatter, + epilog=""" +Examples: + # Process folder of PDBs + %(prog)s --input-folder /path/to/pdbs --output-dir results --num-swarms 50 --swarm-steps 10000 + + # Process specific PDB files + %(prog)s --input-pdbs struct1.pdb struct2.pdb --output-dir results --num-swarms 20 --swarm-steps 5000 + + # Process single structure (for parallelization) + %(prog)s --input-pdbs struct1.pdb --output-dir results --single-structure --structure-index 1 + """, + ) + + # Input arguments (mutually exclusive) + input_group = parser.add_mutually_exclusive_group(required=True) + input_group.add_argument( + "--input-folder", + type=str, + help="Folder containing PDB files to process" + ) + input_group.add_argument( + "--input-pdbs", + nargs="+", + help="List of PDB files to process" + ) + + # Output + parser.add_argument( + "--output-dir", + type=str, + required=True, + help="Output directory for swarm trajectories" + ) + + # Simulation parameters + sim_group = parser.add_argument_group("Simulation Parameters") + sim_group.add_argument( + "--dt", type=float, default=SwarmConfig.dt_ps, + help="Timestep in ps (default: %(default)s)" + ) + sim_group.add_argument( + "--temp", type=float, default=SwarmConfig.temp_K, + help="Temperature in K (default: %(default)s)" + ) + sim_group.add_argument( + "--pressure", type=float, default=SwarmConfig.pressure_bar, + help="Pressure in bar (default: %(default)s)" + ) + sim_group.add_argument( + "--position-restraint-k", type=float, default=SwarmConfig.position_restraint_k, + help="Position restraint force constant in kJ/(mol.A^2) (default: %(default)s)" + ) + + # Forcefield options + ff_group = parser.add_argument_group("Forcefield Options") + ff_group.add_argument( + "--forcefield", nargs=2, default=SwarmConfig.forcefield, + metavar=("FF1", "FF2"), help="Forcefield XML files (default: %(default)s)" + ) + + # Equilibration steps + equil_group = parser.add_argument_group("Equilibration Steps") + equil_group.add_argument( + "--energy-minimization-steps", type=int, default=SwarmConfig.energy_minimization_steps, + help="Steps for energy minimization (default: %(default)s)" + ) + equil_group.add_argument( + "--nvt-restraint-steps", type=int, default=SwarmConfig.nvt_restraint_steps, + help="Steps for NVT equilibration with restraints (default: %(default)s)" + ) + equil_group.add_argument( + "--npt-restraint-steps", type=int, default=SwarmConfig.npt_restraint_steps, + help="Steps for NPT equilibration with restraints (default: %(default)s)" + ) + equil_group.add_argument( + "--nvt-equil-steps", type=int, default=SwarmConfig.nvt_equil_steps, + help="Steps for NVT equilibration without restraints (default: %(default)s)" + ) + equil_group.add_argument( + "--npt-equil-steps", type=int, default=SwarmConfig.npt_equil_steps, + help="Steps for NPT equilibration without restraints (default: %(default)s)" + ) + + # Swarm parameters + swarm_group = parser.add_argument_group("Swarm Parameters") + swarm_group.add_argument( + "--num-swarms", type=int, default=SwarmConfig.num_swarms, + help="Number of swarm trajectories to generate per structure (default: %(default)s)" + ) + swarm_group.add_argument( + "--swarm-steps", type=int, default=SwarmConfig.swarm_steps, + help="Number of steps per swarm trajectory (default: %(default)s)" + ) + swarm_group.add_argument( + "--save-frequency", type=int, default=SwarmConfig.save_frequency, + help="Frequency of saving frames in swarm trajectories (default: %(default)s)" + ) + + # Processing options + proc_group = parser.add_argument_group("Processing Options") + proc_group.add_argument( + "--save-intermediate-files", action="store_true", + help="Save intermediate files during equilibration (default: False)" + ) + proc_group.add_argument( + "--single-structure", action="store_true", + help="Process only a single structure (for parallelization)" + ) + proc_group.add_argument( + "--structure-index", type=int, + help="Index of structure to process (0-based, for parallelization)" + ) + + # New workflow options + proc_group.add_argument( + "--skip-equilibration", action="store_true", + help="Skip equilibration if equilibrated_start.pdb already exists (default: False)" + ) + proc_group.add_argument( + "--equilibrate-only", action="store_true", + help="Only perform equilibration, do not generate swarms (default: False)" + ) + proc_group.add_argument( + "--append-swarms", action="store_true", default=True, + help="Start swarm indexing from existing trajectories rather than overwriting (default: True)" + ) + + args = parser.parse_args() + + # Handle input parsing + if args.input_folder: + # Find all PDB files in the folder + pdb_pattern = os.path.join(args.input_folder, "*.pdb") + input_pdbs = sorted(glob.glob(pdb_pattern)) + if not input_pdbs: + raise ValueError(f"No PDB files found in {args.input_folder}") + py_logger.info(f"Found {len(input_pdbs)} PDB files in {args.input_folder}") + else: + input_pdbs = args.input_pdbs + # Verify all files exist + for pdb_file in input_pdbs: + if not os.path.exists(pdb_file): + raise FileNotFoundError(f"PDB file not found: {pdb_file}") + + # Handle single structure processing + if args.single_structure: + if args.structure_index is not None: + if args.structure_index >= len(input_pdbs): + raise ValueError(f"Structure index {args.structure_index} out of range (0-{len(input_pdbs)-1})") + input_pdbs = [input_pdbs[args.structure_index]] + else: + if len(input_pdbs) > 1: + py_logger.warning("Single structure mode with multiple PDbs - processing only the first one") + input_pdbs = [input_pdbs[0]] + + return SwarmConfig( + input_pdbs=input_pdbs, + output_dir=args.output_dir, + dt_ps=args.dt, + temp_K=args.temp, + pressure_bar=args.pressure, + position_restraint_k=args.position_restraint_k, + forcefield=tuple(args.forcefield), + energy_minimization_steps=args.energy_minimization_steps, + nvt_restraint_steps=args.nvt_restraint_steps, + npt_restraint_steps=args.npt_restraint_steps, + nvt_equil_steps=args.nvt_equil_steps, + npt_equil_steps=args.npt_equil_steps, + num_swarms=args.num_swarms, + swarm_steps=args.swarm_steps, + save_frequency=args.save_frequency, + save_intermediate_files=args.save_intermediate_files, + single_structure_mode=args.single_structure, + structure_index=args.structure_index, + skip_equilibration=args.skip_equilibration, + equilibrate_only=args.equilibrate_only, + append_swarms=args.append_swarms, + ) + + +def get_structure_name(pdb_file: str) -> str: + """Get a clean structure name from PDB filename.""" + return os.path.splitext(os.path.basename(pdb_file))[0] + + +def setup_structure_directory(pdb_file: str, config: SwarmConfig, structure_idx: int) -> Tuple[str, str]: + """Create output directory for a structure and return paths.""" + structure_name = get_structure_name(pdb_file) + structure_dir = os.path.join(config.output_dir, f"AA_{structure_idx:03d}") + + os.makedirs(structure_dir, exist_ok=True) + py_logger.info(f"Created structure directory: {structure_dir}") + + return structure_dir, structure_name + + +def find_existing_swarms(structure_dir: str, swarm_steps: int, dt_ps: float) -> int: + """Find existing swarm trajectories and return the next available index.""" + trajectory_time_ps = swarm_steps * dt_ps + pattern = os.path.join(structure_dir, f"swarm_{trajectory_time_ps:.0f}ps_*.xtc") + existing_swarms = glob.glob(pattern) + + if not existing_swarms: + return 0 # Start from 1 if no existing swarms + + # Extract indices from existing filenames + indices = [] + for swarm_file in existing_swarms: + filename = os.path.basename(swarm_file) + # Extract index from filename like "swarm_1ps_001.xtc" + try: + index_part = filename.split('_')[-1].split('.')[0] # Get "001" part + indices.append(int(index_part)) + except (ValueError, IndexError): + continue + + if indices: + next_index = max(indices) + 1 + py_logger.info(f"Found {len(indices)} existing swarm trajectories, starting from index {next_index}") + return next_index + else: + return 0 + + +def check_equilibration_exists(structure_dir: str) -> bool: + """Check if equilibration has already been completed.""" + equilibrated_pdb = os.path.join(structure_dir, "equilibrated_start.pdb") + return os.path.exists(equilibrated_pdb) + + +def equilibrate_structure( + pdb_file: str, + structure_dir: str, + structure_name: str, + config: SwarmConfig +) -> Tuple[op.Positions, op.Velocities, Simulation]: + """ + Equilibrate a single structure starting from solvation. + Returns the equilibrated positions, velocities, and simulation object. + """ + py_logger.info(f"Starting equilibration for {structure_name}") + + # Convert to absolute path before changing directories + pdb_file_abs = os.path.abspath(pdb_file) + + # Change to structure directory + original_dir = os.getcwd() + os.chdir(structure_dir) + + try: + # Load the initial structure (assume it's already fixed and hydrogenated) + from openmm.app import PDBFile + pdb = PDBFile(pdb_file_abs) + positions = pdb.positions + topology = pdb.topology + + # Create forcefield + ff = ForceField(*config.forcefield) + + # Solvate the structure + py_logger.info("Solvating structure...") + positions, topology = op.solvate( + positions, + topology, + ff, + padding_nm=config.padding_nm, + water_model=config.water_model, + positive_ion=config.positive_ion, + negative_ion=config.negative_ion, + output_file_prefix=f"{structure_name}_solvated", + save_file=config.save_intermediate_files, + ) + + # Create simulation + simulation = op.get_system_with_Langevin_integrator( + topology, ff, config.temp_K, dt_ps=config.dt_ps + ) + + # Add position restraints for equilibration + simulation = op.add_position_restraints( + positions, topology, simulation, k=config.position_restraint_k + ) + + # Energy minimization + py_logger.info("Energy minimization...") + positions, simulation = op.minimize_energy( + positions, + simulation, + num_steps=config.energy_minimization_steps, + output_file_prefix=f"{structure_name}_minimized", + save_file=config.save_intermediate_files, + save_protein_only_file=False, # Don't need protein-only file here + ) + + # NVT equilibration with restraints + py_logger.info("NVT equilibration with restraints...") + positions, velocities, simulation = op.run_simulation( + positions=positions, + simulation=simulation, + velocities=None, + output_frequency=1000, # Less frequent output for equilibration + save_intermediate_files=config.save_intermediate_files, + ensemble="NVT", + output_file_prefix=f"{structure_name}_restrainedNVT", + num_steps=config.nvt_restraint_steps, + ) + + # NPT equilibration with restraints + py_logger.info("NPT equilibration with restraints...") + positions, velocities, simulation = op.run_simulation( + positions=positions, + simulation=simulation, + velocities=velocities, + temp_K=config.temp_K, + pressure_bar=config.pressure_bar, + output_frequency=1000, + save_intermediate_files=config.save_intermediate_files, + ensemble="NPT", + output_file_prefix=f"{structure_name}_restrainedNPT", + num_steps=config.npt_restraint_steps, + ) + + # Remove position restraints + py_logger.info("Removing position restraints...") + simulation.context.getSystem().removeForce(simulation.context.getSystem().getNumForces() - 1) + + # NVT equilibration without restraints + py_logger.info("NVT equilibration without restraints...") + positions, velocities, simulation = op.run_simulation( + positions=positions, + simulation=simulation, + velocities=velocities, + output_frequency=1000, + save_intermediate_files=config.save_intermediate_files, + ensemble="NVT", + output_file_prefix=f"{structure_name}_equilNVT", + num_steps=config.nvt_equil_steps, + ) + + # Final NPT equilibration + py_logger.info("Final NPT equilibration...") + positions, velocities, simulation = op.run_simulation( + positions=positions, + simulation=simulation, + velocities=velocities, + temp_K=config.temp_K, + pressure_bar=config.pressure_bar, + output_frequency=1000, + save_intermediate_files=config.save_intermediate_files, + ensemble="NPT", + output_file_prefix=f"{structure_name}_equilNPT", + num_steps=config.npt_equil_steps, + save_pdb=True, + pdb_output_file="equilibrated_start.pdb", # Save the starting structure for swarms + ) + + py_logger.info(f"Equilibration completed for {structure_name}") + return positions, velocities, simulation + + finally: + # Always return to original directory + os.chdir(original_dir) + + +def generate_swarms( + positions: op.Positions, + velocities: op.Velocities, + simulation: Simulation, + structure_dir: str, + structure_name: str, + config: SwarmConfig +) -> None: + """Generate swarm trajectories from equilibrated structure.""" + py_logger.info(f"Generating {config.num_swarms} swarm trajectories for {structure_name}") + + # Change to structure directory + original_dir = os.getcwd() + os.chdir(structure_dir) + + try: + # Calculate trajectory time in picoseconds + trajectory_time_ps = config.swarm_steps * config.dt_ps + + # Determine starting index based on existing swarms + if config.append_swarms: + start_idx = find_existing_swarms(structure_dir, config.swarm_steps, config.dt_ps) + else: + start_idx = 1 + + for swarm_count in range(config.num_swarms): + swarm_idx = start_idx + swarm_count + py_logger.info(f"Generating swarm {swarm_idx + 1}/{config.num_swarms}") + + # Set initial conditions (same positions, slightly perturbed velocities for variation) + simulation.context.setPositions(positions) + + # Add small random perturbation to velocities for each swarm + import numpy as np + from openmm.unit import nanometer, picosecond + np.random.seed(swarm_idx) # Reproducible but different per swarm + + # Get original velocities as numpy array + velocities_array = np.array(velocities.value_in_unit(nanometer/picosecond)) + + # Add small random perturbation (0.1% of thermal velocity) + perturbation_scale = 0.001 + thermal_velocity = np.sqrt(3 * 8.314 * config.temp_K / 1000) # Approximate thermal velocity + perturbation = np.random.normal(0, perturbation_scale * thermal_velocity, velocities_array.shape) + perturbed_velocities = velocities_array + perturbation + + # Convert back to OpenMM format + from openmm.unit import Quantity + perturbed_velocities_unit = Quantity(perturbed_velocities, nanometer/picosecond) + simulation.context.setVelocities(perturbed_velocities_unit) + + # Generate swarm trajectory + swarm_filename = f"swarm_{trajectory_time_ps:.0f}ps_{swarm_idx + 1:03d}.xtc" + + _, _, simulation = op.run_simulation( + positions=positions, + simulation=simulation, + velocities=velocities, + temp_K=config.temp_K, + pressure_bar=config.pressure_bar, + output_frequency=config.save_frequency, + save_intermediate_files=False, # No intermediate files for swarms + ensemble="NPT", + output_file_prefix=f"swarm_{swarm_idx + 1:03d}", + num_steps=config.swarm_steps, + save_xtc=True, + xtc_output_file=swarm_filename, + save_pdb=False, # Don't save PDB for each swarm + ) + + py_logger.info(f"Completed swarm {swarm_idx + 1}: {swarm_filename}") + + finally: + # Always return to original directory + os.chdir(original_dir) + + +def process_structure(pdb_file: str, structure_idx: int, config: SwarmConfig) -> None: + """Process a single structure: equilibrate and/or generate swarms.""" + structure_name = get_structure_name(pdb_file) + py_logger.info(f"Processing structure {structure_idx + 1}: {structure_name}") + + # Setup structure directory + structure_dir, structure_name = setup_structure_directory(pdb_file, config, structure_idx) + + # Check if equilibration exists and should be skipped + equilibration_exists = check_equilibration_exists(structure_dir) + + if config.skip_equilibration and not equilibration_exists: + py_logger.warning(f"--skip-equilibration set but no equilibrated_start.pdb found in {structure_dir}") + py_logger.info("Proceeding with equilibration...") + config.skip_equilibration = False + + # Handle equilibration + if config.skip_equilibration and equilibration_exists: + py_logger.info(f"Skipping equilibration for {structure_name} (equilibrated_start.pdb exists)") + # TODO: Load from equilibrated state if needed for swarm generation + positions, velocities, simulation = None, None, None + else: + # Perform equilibration + py_logger.info(f"Starting equilibration for {structure_name}") + positions, velocities, simulation = equilibrate_structure( + pdb_file, structure_dir, structure_name, config + ) + + # Stop here if only equilibrating + if config.equilibrate_only: + py_logger.info(f"Equilibration-only mode: completed equilibration for {structure_name}") + return + + # Generate swarms (need to implement loading from equilibrated state if skipped equilibration) + if config.skip_equilibration and equilibration_exists: + py_logger.info("Loading equilibrated state for swarm generation...") + # TODO: Implement loading from saved equilibrated state + py_logger.warning("Loading from saved equilibrated state not yet implemented!") + py_logger.warning("Please run without --skip-equilibration for now") + return + + # Generate swarm trajectories + generate_swarms(positions, velocities, simulation, structure_dir, structure_name, config) + + py_logger.info(f"Completed processing structure {structure_idx + 1}: {structure_name}") + + +def main(): + """Main execution function.""" + config = parse_args() + + # Create main output directory + os.makedirs(config.output_dir, exist_ok=True) + py_logger.info(f"Output directory: {config.output_dir}") + py_logger.info(f"Processing {len(config.input_pdbs)} structure(s)") + + # Process each structure + for idx, pdb_file in enumerate(config.input_pdbs): + try: + # Use global structure index in single-structure mode, otherwise use enumeration index + if config.single_structure_mode and config.structure_index is not None: + structure_idx = config.structure_index + else: + structure_idx = idx + + process_structure(pdb_file, structure_idx, config) + except Exception as e: + py_logger.error(f"Error processing {pdb_file}: {str(e)}") + if config.single_structure_mode: + raise # Re-raise in single structure mode for debugging + else: + py_logger.warning("Continuing with next structure...") + continue + + py_logger.info("Swarm generation completed!") + + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scripts/generate_data/openmm_utils.py b/scripts/generate_data/openmm_utils.py index 5fe640e..a2a8652 100644 --- a/scripts/generate_data/openmm_utils.py +++ b/scripts/generate_data/openmm_utils.py @@ -145,6 +145,46 @@ def solvate( return modeller.positions, modeller.topology +def select_best_platform(): + """Select the best available OpenMM platform (preferably GPU) with fallback logic.""" + try: + import openmm + platforms = [] + for i in range(openmm.Platform.getNumPlatforms()): + platform = openmm.Platform.getPlatform(i) + platforms.append((platform.getName(), platform.getSpeed(), platform)) + + # Sort by speed (higher is better) and prefer CUDA > OpenCL > CPU + platform_priority = {'CUDA': 3, 'OpenCL': 2, 'CPU': 1, 'Reference': 0} + platforms.sort(key=lambda x: (x[1], platform_priority.get(x[0], 0)), reverse=True) + + # Try each platform in order until one works + for platform_name, speed, platform in platforms: + try: + # Quick test to see if platform can create a context + # We'll use a minimal system for testing + test_system = openmm.System() + test_system.addParticle(1.0) # Add one particle + test_integrator = openmm.LangevinMiddleIntegrator(300, 1.0, 0.002) + test_context = openmm.Context(test_system, test_integrator, platform) + del test_context # Clean up + del test_integrator + del test_system + + py_logger.info(f"Using OpenMM platform: {platform_name} (speed: {speed})") + return platform + except Exception as e: + py_logger.warning(f"Platform {platform_name} failed test: {e}") + continue + + py_logger.warning("No working OpenMM platforms found, using default") + return None + + except Exception as e: + py_logger.warning(f"Could not select optimal platform: {e}, using default") + return None + + def get_system_with_Langevin_integrator( topology: Topology, forcefield: ForceField, temp_K: float, dt_ps: float, state: Optional[str] = None ) -> Simulation: @@ -157,7 +197,13 @@ def get_system_with_Langevin_integrator( constraints=HBonds, ) integrator = LangevinMiddleIntegrator(temp_K * kelvin, 1 / picoseconds, dt_ps * picoseconds) - simulation = Simulation(topology, system, integrator) + + platform = select_best_platform() + if platform is not None: + simulation = Simulation(topology, system, integrator, platform) + else: + simulation = Simulation(topology, system, integrator) + if state is not None: simulation.loadState(state) return simulation @@ -175,7 +221,13 @@ def get_system_with_NoseHoover_integrator( constraints=HBonds, ) integrator = NoseHooverIntegrator(temp_K * kelvin, 1 / picoseconds, dt_ps * picoseconds) - simulation = Simulation(topology, system, integrator) + + platform = select_best_platform() + if platform is not None: + simulation = Simulation(topology, system, integrator, platform) + else: + simulation = Simulation(topology, system, integrator) + return simulation diff --git a/scripts/generate_data/run_swarms_parallel.sh b/scripts/generate_data/run_swarms_parallel.sh new file mode 100755 index 0000000..fe55b46 --- /dev/null +++ b/scripts/generate_data/run_swarms_parallel.sh @@ -0,0 +1,294 @@ +#!/bin/bash + +# Helper script for parallelizing swarm generation +# This script demonstrates different approaches for running generate_swarms.py in parallel + +# Default parameters +INPUT_FOLDER="" +INPUT_PDBS="" +OUTPUT_DIR="" +NUM_SWARMS=10 +SWARM_STEPS=10000 +SAVE_FREQUENCY=10 + +# Parse command line arguments +while [[ $# -gt 0 ]]; do + case $1 in + --input-folder) + INPUT_FOLDER="$2" + shift 2 + ;; + --input-pdbs) + shift + INPUT_PDBS="" + while [[ $# -gt 0 && $1 != --* ]]; do + INPUT_PDBS="$INPUT_PDBS $1" + shift + done + ;; + --output-dir) + OUTPUT_DIR="$2" + shift 2 + ;; + --num-swarms) + NUM_SWARMS="$2" + shift 2 + ;; + --swarm-steps) + SWARM_STEPS="$2" + shift 2 + ;; + --save-frequency) + SAVE_FREQUENCY="$2" + shift 2 + ;; + --help) + echo "Usage: $0 [OPTIONS]" + echo "" + echo "Options:" + echo " --input-folder DIR Folder containing PDB files" + echo " --input-pdbs FILE... List of PDB files" + echo " --output-dir DIR Output directory" + echo " --num-swarms N Number of swarms per structure (default: 10)" + echo " --swarm-steps N Steps per swarm (default: 10000)" + echo " --save-frequency N Save frequency (default: 10)" + echo " --help Show this help" + echo "" + echo "Examples:" + echo " # Using SLURM job arrays:" + echo " $0 --input-folder /path/to/pdbs --output-dir results" + echo "" + echo " # Using GNU parallel:" + echo " $0 --input-pdbs *.pdb --output-dir results" + exit 0 + ;; + *) + echo "Unknown option: $1" + exit 1 + ;; + esac +done + +# Validate required arguments +if [[ -z "$OUTPUT_DIR" ]]; then + echo "Error: --output-dir is required" + exit 1 +fi + +if [[ -z "$INPUT_FOLDER" && -z "$INPUT_PDBS" ]]; then + echo "Error: Either --input-folder or --input-pdbs is required" + exit 1 +fi + +# Get list of PDB files +if [[ -n "$INPUT_FOLDER" ]]; then + PDB_FILES=($(find "$INPUT_FOLDER" -name "*.pdb" | sort)) + echo "Found ${#PDB_FILES[@]} PDB files in $INPUT_FOLDER" +else + PDB_FILES=($INPUT_PDBS) + echo "Processing ${#PDB_FILES[@]} specified PDB files" +fi + +if [[ ${#PDB_FILES[@]} -eq 0 ]]; then + echo "Error: No PDB files found" + exit 1 +fi + +# Create output directory +mkdir -p "$OUTPUT_DIR" + +echo "==========================================" +echo "Swarm Generation Parallelization Helper" +echo "==========================================" +echo "PDB files to process: ${#PDB_FILES[@]}" +echo "Output directory: $OUTPUT_DIR" +echo "Swarms per structure: $NUM_SWARMS" +echo "Steps per swarm: $SWARM_STEPS" +echo "Save frequency: $SAVE_FREQUENCY" +echo "" + +# Method 1: SLURM Job Array +echo "=== SLURM Job Array Approach ===" +echo "To submit as a SLURM job array, create a file 'submit_swarms.sh':" +echo "" +cat << 'EOF' +#!/bin/bash +#SBATCH --job-name=swarms +#SBATCH --array=0-NUM_STRUCTURES-1 +#SBATCH --ntasks=1 +#SBATCH --cpus-per-task=1 +#SBATCH --mem=8GB +#SBATCH --time=24:00:00 +#SBATCH --output=swarms_%A_%a.out +#SBATCH --error=swarms_%A_%a.err + +# Get PDB file for this array job +PDB_FILES=(PDB_FILE_LIST) +PDB_FILE=${PDB_FILES[$SLURM_ARRAY_TASK_ID]} + +# Run swarm generation for single structure +python scripts/generate_data/generate_swarms.py \ + --input-pdbs "$PDB_FILE" \ + --output-dir OUTPUT_DIR \ + --single-structure \ + --structure-index $SLURM_ARRAY_TASK_ID \ + --num-swarms NUM_SWARMS \ + --swarm-steps SWARM_STEPS \ + --save-frequency SAVE_FREQUENCY +EOF + +# Create actual SLURM script +SLURM_SCRIPT="submit_swarms_$(date +%Y%m%d_%H%M%S).sh" +sed "s/NUM_STRUCTURES/$((${#PDB_FILES[@]}-1))/" << 'EOF' > "$SLURM_SCRIPT" +#!/bin/bash +#SBATCH --job-name=swarms +#SBATCH --array=0-NUM_STRUCTURES +#SBATCH --ntasks=1 +#SBATCH --cpus-per-task=1 +#SBATCH --mem=8GB +#SBATCH --time=24:00:00 +#SBATCH --output=swarms_%A_%a.out +#SBATCH --error=swarms_%A_%a.err + +# Get PDB file for this array job +EOF + +echo "PDB_FILES=(" >> "$SLURM_SCRIPT" +for pdb in "${PDB_FILES[@]}"; do + echo " \"$pdb\"" >> "$SLURM_SCRIPT" +done +echo ")" >> "$SLURM_SCRIPT" + +cat << EOF >> "$SLURM_SCRIPT" +PDB_FILE=\${PDB_FILES[\$SLURM_ARRAY_TASK_ID]} + +# Run swarm generation for single structure +python scripts/generate_data/generate_swarms.py \\ + --input-pdbs "\$PDB_FILE" \\ + --output-dir "$OUTPUT_DIR" \\ + --single-structure \\ + --structure-index \$SLURM_ARRAY_TASK_ID \\ + --num-swarms $NUM_SWARMS \\ + --swarm-steps $SWARM_STEPS \\ + --save-frequency $SAVE_FREQUENCY +EOF + +echo "Created SLURM script: $SLURM_SCRIPT" +echo "To submit: sbatch $SLURM_SCRIPT" +echo "" + +# Method 2: GNU Parallel +echo "=== GNU Parallel Approach ===" +echo "To run with GNU parallel:" + +# Create parallel command file +PARALLEL_SCRIPT="run_swarms_parallel_$(date +%Y%m%d_%H%M%S).sh" +cat << EOF > "$PARALLEL_SCRIPT" +#!/bin/bash + +# Function to process a single PDB file +process_pdb() { + local pdb_file="\$1" + local structure_idx="\$2" + + echo "Processing \$pdb_file (structure \$structure_idx)" + + python scripts/generate_data/generate_swarms.py \\ + --input-pdbs "\$pdb_file" \\ + --output-dir "$OUTPUT_DIR" \\ + --single-structure \\ + --structure-index "\$structure_idx" \\ + --num-swarms $NUM_SWARMS \\ + --swarm-steps $SWARM_STEPS \\ + --save-frequency $SAVE_FREQUENCY +} + +export -f process_pdb + +# Run in parallel (adjust -j for number of parallel jobs) +parallel -j 4 process_pdb {1} {#} ::: \\ +EOF + +for pdb in "${PDB_FILES[@]}"; do + echo " \"$pdb\" \\" >> "$PARALLEL_SCRIPT" +done + +# Remove last backslash +sed -i '$ s/ \\$//' "$PARALLEL_SCRIPT" + +chmod +x "$PARALLEL_SCRIPT" +echo "Created parallel script: $PARALLEL_SCRIPT" +echo "To run: ./$PARALLEL_SCRIPT" +echo "" + +# Method 3: Simple Background Jobs +echo "=== Background Jobs Approach ===" +BACKGROUND_SCRIPT="run_swarms_background_$(date +%Y%m%d_%H%M%S).sh" +cat << EOF > "$BACKGROUND_SCRIPT" +#!/bin/bash + +echo "Running swarm generation with background jobs..." + +# Process each PDB file in background (limit concurrent jobs) +max_jobs=4 # Adjust based on your system +job_count=0 + +EOF + +for i in "${!PDB_FILES[@]}"; do + cat << EOF >> "$BACKGROUND_SCRIPT" +# Wait if we've hit the job limit +while [ \$(jobs -r | wc -l) -ge \$max_jobs ]; do + sleep 1 +done + +echo "Starting structure $i: ${PDB_FILES[$i]}" +python scripts/generate_data/generate_swarms.py \\ + --input-pdbs "${PDB_FILES[$i]}" \\ + --output-dir "$OUTPUT_DIR" \\ + --single-structure \\ + --structure-index $i \\ + --num-swarms $NUM_SWARMS \\ + --swarm-steps $SWARM_STEPS \\ + --save-frequency $SAVE_FREQUENCY & + +EOF +done + +cat << 'EOF' >> "$BACKGROUND_SCRIPT" + +# Wait for all background jobs to complete +echo "Waiting for all jobs to complete..." +wait + +echo "All swarm generation jobs completed!" +EOF + +chmod +x "$BACKGROUND_SCRIPT" +echo "Created background jobs script: $BACKGROUND_SCRIPT" +echo "To run: ./$BACKGROUND_SCRIPT" +echo "" + +# Method 4: Single command (no parallelization) +echo "=== Single Process Approach ===" +echo "To run all structures in a single process:" +if [[ -n "$INPUT_FOLDER" ]]; then + SINGLE_CMD="python scripts/generate_data/generate_swarms.py --input-folder \"$INPUT_FOLDER\"" +else + SINGLE_CMD="python scripts/generate_data/generate_swarms.py --input-pdbs" + for pdb in "${PDB_FILES[@]}"; do + SINGLE_CMD="$SINGLE_CMD \"$pdb\"" + done +fi +SINGLE_CMD="$SINGLE_CMD --output-dir \"$OUTPUT_DIR\" --num-swarms $NUM_SWARMS --swarm-steps $SWARM_STEPS --save-frequency $SAVE_FREQUENCY" + +echo "$SINGLE_CMD" +echo "" + +echo "==========================================" +echo "Choose the parallelization method that best fits your computing environment:" +echo "1. SLURM job arrays - Best for HPC clusters" +echo "2. GNU parallel - Good for multi-core workstations" +echo "3. Background jobs - Simple shell-based parallelization" +echo "4. Single process - No parallelization, simplest approach" +echo "==========================================" \ No newline at end of file diff --git a/scripts/slurm/generate_swarms_batch_final.sh b/scripts/slurm/generate_swarms_batch_final.sh new file mode 100755 index 0000000..1364de1 --- /dev/null +++ b/scripts/slurm/generate_swarms_batch_final.sh @@ -0,0 +1,183 @@ +#!/bin/bash + +# FINAL CORRECTED: Master script to generate SLURM batch jobs for swarm generation +# Correct equilibration steps: 50k restrained, 10 unrestrained +# Sequential processing: equilibration + swarms per structure before moving to next + +echo "šŸš€ FINAL CORRECTED Swarm Batch Generation Script" +echo "================================================" + +# Configuration +STRUCTURES_PER_BATCH=20 +INPUT_DIR="data/swarm_data/test" +OUTPUT_DIR="data/swarm_data/test/swarm_results" +SCRIPT_DIR="scripts/slurm/batches_final" + +# Equilibration settings (CORRECTED) +NVT_RESTRAINT_STEPS=50000 # 50k as requested +NPT_RESTRAINT_STEPS=50000 # 50k as requested +NVT_EQUIL_STEPS=10 # 10 steps (not 10k!) as requested +NPT_EQUIL_STEPS=10 # 10 steps (not 10k!) as requested + +# Swarm settings +NUM_SWARMS=5 +SWARM_STEPS=500 # 1ps Ć· 2fs/step = 500 steps per swarm +SAVE_FREQUENCY=10 + +# Create batch script directory +mkdir -p "$SCRIPT_DIR" + +# Get list of all PDB files +PDB_FILES=($(ls -1 "$INPUT_DIR"/*.pdb | sort)) +TOTAL_STRUCTURES=${#PDB_FILES[@]} + +echo "šŸ“Š Configuration:" +echo " Total structures: $TOTAL_STRUCTURES" +echo " Structures per batch: $STRUCTURES_PER_BATCH" +echo " Equilibration steps: NVT/NPT restrained=50k, unrestrained=10 (CORRECTED)" +echo " Swarms: $NUM_SWARMS Ɨ 1ps ($SWARM_STEPS steps) per structure" +echo " Workflow: Sequential (equil+swarms per structure)" +echo "" + +# Calculate number of batches needed +NUM_BATCHES=$(( (TOTAL_STRUCTURES + STRUCTURES_PER_BATCH - 1) / STRUCTURES_PER_BATCH )) + +echo "šŸ“ Generating $NUM_BATCHES FINAL CORRECTED batch scripts..." + +for ((batch=1; batch<=NUM_BATCHES; batch++)); do + # Calculate structure range for this batch + start_idx=$(( (batch - 1) * STRUCTURES_PER_BATCH )) + end_idx=$(( start_idx + STRUCTURES_PER_BATCH - 1 )) + + # Don't exceed total number of structures + if [ $end_idx -ge $TOTAL_STRUCTURES ]; then + end_idx=$(( TOTAL_STRUCTURES - 1 )) + fi + + structures_in_batch=$(( end_idx - start_idx + 1 )) + + echo " Batch $batch: global indices $start_idx-$end_idx ($structures_in_batch structures)" + + # Create SLURM script for this batch + script_file="$SCRIPT_DIR/swarms_batch_${batch}_final.sh" + + cat > "$script_file" << EOF +#!/bin/bash +#SBATCH --job-name=swarms_batch_${batch}_final +#SBATCH --partition=gpu2 +#SBATCH --gres=gpu:1 +#SBATCH --cpus-per-task=4 +#SBATCH --mem=16GB +#SBATCH --time=4:00:00 +#SBATCH --output=swarms_batch_${batch}_final_%j.out +#SBATCH --error=swarms_batch_${batch}_final_%j.err + +echo "SLURM_JOB_ID = \$SLURM_JOB_ID" +echo "hostname = \$(hostname)" +echo "Starting FINAL CORRECTED swarm batch ${batch} on GPU..." +echo "" + +# Print GPU info +nvidia-smi + +# Activate conda environment +echo "Activating conda environment..." +source /homefs/home/vanib/miniforge3/etc/profile.d/conda.sh +conda activate jamun +echo "Python path: \$(which python)" +echo "Conda environment: \$CONDA_DEFAULT_ENV" +echo "" + +# Change to working directory +cd /homefs/home/vanib/jamun + +# Create full PDB file list +ALL_PDB_FILES=( +EOF + + # Add ALL PDB files to each script (needed for structure index validation) + for ((i=0; i> "$script_file" + done + + cat >> "$script_file" << EOF +) + +echo "šŸ§¬ Processing structures with GLOBAL indices $start_idx to $end_idx:" +echo "Using full PDB list of \${#ALL_PDB_FILES[@]} files for proper indexing" +echo "Sequential workflow: equilibration + swarms per structure" +echo "" + +# Process each structure in this batch (SEQUENTIAL: equil+swarms per structure) +EOF + + # Add individual structure processing (SEQUENTIAL) + for ((global_idx=start_idx; global_idx<=end_idx; global_idx++)); do + pdb_file="${PDB_FILES[$global_idx]}" + cat >> "$script_file" << EOF + +echo "āš–ļø Processing structure $global_idx: \$(basename "${pdb_file}")" +echo "============================================================" + +# SINGLE COMMAND: Do both equilibration AND swarms for this structure +echo "šŸ”„ Processing structure $global_idx: equilibration + $NUM_SWARMS Ɨ 1ps swarms..." +python scripts/generate_data/generate_swarms.py \\ + --input-pdbs "\${ALL_PDB_FILES[@]}" \\ + --output-dir "$OUTPUT_DIR" \\ + --single-structure \\ + --structure-index $global_idx \\ + --nvt-restraint-steps $NVT_RESTRAINT_STEPS \\ + --npt-restraint-steps $NPT_RESTRAINT_STEPS \\ + --nvt-equil-steps $NVT_EQUIL_STEPS \\ + --npt-equil-steps $NPT_EQUIL_STEPS \\ + --num-swarms $NUM_SWARMS \\ + --swarm-steps $SWARM_STEPS \\ + --save-frequency $SAVE_FREQUENCY \\ + --save-intermediate-files + +if [ \$? -ne 0 ]; then + echo "āŒ Processing failed for structure $global_idx" + exit 1 +fi + +echo "āœ… COMPLETED structure $global_idx: \$(basename "${pdb_file}") (equilibration + $NUM_SWARMS swarms)" +echo "" +EOF + done + + cat >> "$script_file" << EOF + +# Summary +echo "šŸ“ˆ BATCH ${batch} SUMMARY" +echo "======================" +echo " Global structure indices: $start_idx to $end_idx" +echo " Structures processed: $structures_in_batch" +echo " Swarms per structure: $NUM_SWARMS" +echo " Total swarms generated: $(( structures_in_batch * NUM_SWARMS ))" +echo " Swarm duration: 1ps each" +echo " Equilibration: 50k restrained, 10 unrestrained steps" +echo " Workflow: Sequential (equil+swarms per structure)" +echo "" +echo "šŸŽ‰ Batch ${batch} completed successfully!" +EOF + + chmod +x "$script_file" +done + +echo "" +echo "āœ… Generated $NUM_BATCHES FINAL CORRECTED batch scripts in $SCRIPT_DIR/" +echo "" +echo "šŸ“‹ To submit jobs:" +echo " # Submit first batch only (for testing):" +echo " sbatch $SCRIPT_DIR/swarms_batch_1_final.sh" +echo "" +echo " # After verification, submit remaining batches:" +echo " for i in {2..$NUM_BATCHES}; do" +echo " sbatch $SCRIPT_DIR/swarms_batch_\${i}_final.sh" +echo " done" +echo "" +echo "šŸ”§ FINAL CORRECTIONS APPLIED:" +echo " āœ… Correct equilibration steps: 50k restrained, 10 unrestrained" +echo " āœ… Sequential workflow: equilibration + swarms per structure" +echo " āœ… Passes all PDB files for proper structure index validation" +echo " āœ… Each structure gets unique AA_XXX directory" \ No newline at end of file From 4982bf65d5e524fa5cdf7dffcff20a1df93e3a66 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Thu, 24 Jul 2025 18:22:14 +0000 Subject: [PATCH 13/32] New configs for noise check experiments, debugged multimeasurement (still slow), sweep scripts --- ...sample_ala_ala_enhanced_sampling_data.yaml | 87 ++++ configs/experiment/sample_capped_2AA.yaml | 14 +- ...sample_enhanced_sampling_single_shape.yaml | 84 ---- ...e_fake_enhanced_sampling_single_shape.yaml | 86 ++++ ...a_enhanced_full_grid_multimeasurement.yaml | 60 +++ .../train_enhanced_denoised_conditioner.yaml | 60 +++ .../train_enhanced_mean_conditioner.yaml | 58 +++ .../train_enhanced_position_conditioner.yaml | 56 +++ .../train_enhanced_self_conditioner.yaml | 56 +++ .../train_enhanced_spiked_conditioner.yaml | 59 +++ .../train_enhanced_standard_jamun.yaml | 52 ++ .../experiment/train_test_single_shape.yaml | 10 +- .../train_test_single_shape_conditional.yaml | 5 +- ..._single_shape_fake_enhanced_sampling.yaml} | 8 +- scratch/test_conditional.py | 93 ---- scratch/visualize_traj_data.py | 103 ++-- scripts/concatenate_trajectories.py | 112 +++++ scripts/slurm/check_sweep_runs.sh | 148 ++++++ .../slurm/debug_sweep_enhanced_sampling.sh | 189 ++++++++ scripts/slurm/run_conditioner_experiment.sh | 81 ++++ scripts/slurm/sweep_enhanced_sampling.sh | 179 +++++++ .../slurm/train_enhanced_sampling_sweep.sh | 46 +- src/jamun/cmdline/sample.py | 6 +- src/jamun/cmdline/train.py | 21 +- src/jamun/data/_mdtraj.py | 4 +- src/jamun/data/_utils.py | 14 +- .../model/denoiser_multimeasurement.yaml | 31 ++ src/jamun/model/__init__.py | 5 +- src/jamun/model/conditioner_usage_example.py | 326 +++++++++++++ src/jamun/model/conditioners/__init__.py | 2 +- src/jamun/model/conditioners/conditioners.py | 170 ++++++- src/jamun/model/denoiser_conditional.py | 2 +- src/jamun/model/denoiser_multimeasurement.py | 89 ++-- src/jamun/model/denoiser_spiked.py | 458 ++++++++++++++++++ src/jamun/utils/_normalizations.py | 53 ++ src/jamun/utils/sampling_wrapper.py | 4 +- 36 files changed, 2513 insertions(+), 318 deletions(-) create mode 100644 configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml delete mode 100644 configs/experiment/sample_enhanced_sampling_single_shape.yaml create mode 100644 configs/experiment/sample_fake_enhanced_sampling_single_shape.yaml create mode 100644 configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml create mode 100644 configs/experiment/train_enhanced_denoised_conditioner.yaml create mode 100644 configs/experiment/train_enhanced_mean_conditioner.yaml create mode 100644 configs/experiment/train_enhanced_position_conditioner.yaml create mode 100644 configs/experiment/train_enhanced_self_conditioner.yaml create mode 100644 configs/experiment/train_enhanced_spiked_conditioner.yaml create mode 100644 configs/experiment/train_enhanced_standard_jamun.yaml rename configs/experiment/{train_test_single_shape_enhanced_sampling.yaml => train_test_single_shape_fake_enhanced_sampling.yaml} (94%) delete mode 100644 scratch/test_conditional.py create mode 100755 scripts/concatenate_trajectories.py create mode 100755 scripts/slurm/check_sweep_runs.sh create mode 100755 scripts/slurm/debug_sweep_enhanced_sampling.sh create mode 100644 scripts/slurm/run_conditioner_experiment.sh create mode 100755 scripts/slurm/sweep_enhanced_sampling.sh create mode 100644 src/jamun/hydra_config/model/denoiser_multimeasurement.yaml create mode 100644 src/jamun/model/conditioner_usage_example.py create mode 100644 src/jamun/model/denoiser_spiked.py create mode 100644 src/jamun/utils/_normalizations.py diff --git a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml b/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml new file mode 100644 index 0000000..5b599a9 --- /dev/null +++ b/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml @@ -0,0 +1,87 @@ +# @package _global_ + +defaults: + # - override /model: denoiser_conditional_pretrained.yaml + - override /callbacks: null + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + as_iterable: false + subsample: 1 + total_lag_time: 2 + lag_subsample_rate: 1 + max_datasets: 10 + label_override: "ALA_ALA" + +num_sampling_steps_per_batch: 10000 +num_batches: 1 +dnum_init_samples_per_dataset: 10 +repeat_init_samples: 1 +continue_chain: true + +# Add your checkpoint path here - update with actual trained model path +wandb_train_run_path: sule-shashank/jamun/l8jwx7mx +checkpoint_type: last + +sigma: 0.04 +M: 1.0 +delta: ${sigma} +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: null + +sampler: + _target_: jamun.sampling.SamplerMemory + devices: 1 + +# Evaluation dataset - standard ALA_ALA from capped diamines for computing metrics +eval_dataset: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 10 + total_lag_time: 2 + lag_subsample_rate: 1 + max_datasets: 1 + label_override: "ALA_ALA" + +# Override ALL callbacks to use eval_dataset for metrics computation +callbacks: + measure_sampling_time: + _target_: jamun.callbacks.sampler.MeasureSamplingTimeCallback + chemical_validity: + _target_: jamun.callbacks.sampler.ChemicalValidityMetricsCallback + datasets: ${eval_dataset} + bond_length_tolerance: 0.2 + volume_exclusion_tolerance: 0.1 + num_molecules_per_trajectory: 100 + ramachandran_plot: + _target_: jamun.callbacks.sampler.RamachandranPlotMetricsCallback + datasets: ${eval_dataset} + trajectory_visualizer: + _target_: jamun.callbacks.sampler.TrajectoryVisualizerCallback + datasets: ${eval_dataset} + num_frames_to_animate: 100 + sample_visualizer: + _target_: jamun.callbacks.sampler.SampleVisualizerCallback + datasets: ${eval_dataset} + num_samples_to_plot: 16 + subsample: 100 + score_distribution: + _target_: jamun.callbacks.sampler.ScoreDistributionCallback + datasets: ${eval_dataset} + save_trajectory: + _target_: jamun.callbacks.sampler.SaveTrajectoryCallback + datasets: ${eval_dataset} + +logger: + wandb: + group: sample_enhanced_sampling_data + notes: "Sampling from enhanced sampling data using memory sampler" + tags: ["sample", "enhanced_sampling", "memory_sampler", "ALA_ALA"] \ No newline at end of file diff --git a/configs/experiment/sample_capped_2AA.yaml b/configs/experiment/sample_capped_2AA.yaml index 3402e26..af989c9 100644 --- a/configs/experiment/sample_capped_2AA.yaml +++ b/configs/experiment/sample_capped_2AA.yaml @@ -13,20 +13,20 @@ init_datasets: pdb_pattern: "^(.*).pdb" subsample: 1 filter_codes: ['ALA_ALA'] - num_frames: 60000 + num_frames: 320000 -num_sampling_steps_per_batch: 20000 -num_batches: 5 -num_init_samples_per_dataset: 1 +num_sampling_steps_per_batch: 1000 +num_batches: 10 +num_init_samples_per_dataset: 50 repeat_init_samples: 1 continue_chain: true # New 2AA -wandb_train_run_path: sule-shashank/jamun/vxpxronn +wandb_train_run_path: sule-shashank/jamun/370wpt17 checkpoint_type: best_so_far -sigma: 0.01 +sigma: 0.04 M: 1.0 delta: ${sigma} friction: 1.0 @@ -40,4 +40,4 @@ sampler: logger: wandb: group: sample_capped_2AA - tags: ['ALA_ALA', 'sigma_0.01', 'standard JAMUN'] \ No newline at end of file + tags: ['ALA_ALA', 'sigma_0.04', 'standard JAMUN'] \ No newline at end of file diff --git a/configs/experiment/sample_enhanced_sampling_single_shape.yaml b/configs/experiment/sample_enhanced_sampling_single_shape.yaml deleted file mode 100644 index 423aabf..0000000 --- a/configs/experiment/sample_enhanced_sampling_single_shape.yaml +++ /dev/null @@ -1,84 +0,0 @@ -# @package _global_ - -defaults: - # - override /model: denoiser_conditional_pretrained.yaml - - override /callbacks: null - -init_datasets: - _target_: jamun.data.parse_datasets_from_directory - root: "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" - traj_pattern: "^(.*).xtc" - pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" - as_iterable: false - subsample: 10 - # total_lag_time: 5 - # lag_subsample_rate: 1 - max_datasets: 10 - -num_sampling_steps_per_batch: 10 -num_batches: 1 -num_init_samples_per_dataset: 10 -repeat_init_samples: 1 -continue_chain: false - -# Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/vxpxronn -# checkpoint_dir: /data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-02_00-37-08/checkpoints -checkpoint_type: last - -sigma: 0.01 -M: 1.0 -delta: ${sigma} -friction: 1.0 -inverse_temperature: 1.0 -score_fn_clip: 100.0 - -sampler: - _target_: jamun.sampling.Sampler - devices: 1 - -# # Evaluation dataset - standard ALA_ALA from capped diamines for computing metrics -# eval_dataset: -# _target_: jamun.data.parse_datasets_from_directory -# root: "${paths.data_path}/capped_diamines/timewarp_splits/train" -# traj_pattern: "^(.*).xtc" -# pdb_pattern: "^(.*).pdb" -# filter_codes: ['ALA_ALA'] -# as_iterable: false -# subsample: 100 -# max_datasets: 1 - -# # Override ALL callbacks to use eval_dataset for metrics computation -# callbacks: -# measure_sampling_time: -# _target_: jamun.callbacks.sampler.MeasureSamplingTimeCallback -# chemical_validity: -# _target_: jamun.callbacks.sampler.ChemicalValidityMetricsCallback -# datasets: ${eval_dataset} -# bond_length_tolerance: 0.2 -# volume_exclusion_tolerance: 0.1 -# num_molecules_per_trajectory: 100 -# ramachandran_plot: -# _target_: jamun.callbacks.sampler.RamachandranPlotMetricsCallback -# datasets: ${eval_dataset} -# trajectory_visualizer: -# _target_: jamun.callbacks.sampler.TrajectoryVisualizerCallback -# datasets: ${eval_dataset} -# num_frames_to_animate: 100 -# sample_visualizer: -# _target_: jamun.callbacks.sampler.SampleVisualizerCallback -# datasets: ${eval_dataset} -# num_samples_to_plot: 16 -# subsample: 100 -# score_distribution: -# _target_: jamun.callbacks.sampler.ScoreDistributionCallback -# datasets: ${eval_dataset} -# save_trajectory: -# _target_: jamun.callbacks.sampler.SaveTrajectoryCallback -# datasets: ${eval_dataset} - -logger: - wandb: - group: sample_ALA_ALA_enhanced_sampling - notes: "Sampling from enhanced sampling trained conditional denoiser" - tags: ["sample", "enhanced_sampling", "conditional_denoiser", "ALA_ALA", "dutiful-fog-302"] \ No newline at end of file diff --git a/configs/experiment/sample_fake_enhanced_sampling_single_shape.yaml b/configs/experiment/sample_fake_enhanced_sampling_single_shape.yaml new file mode 100644 index 0000000..9f3c670 --- /dev/null +++ b/configs/experiment/sample_fake_enhanced_sampling_single_shape.yaml @@ -0,0 +1,86 @@ +# @package _global_ + +defaults: + # - override /model: denoiser_conditional_pretrained.yaml + - override /callbacks: null + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" + traj_pattern: "^(.*).xtc" + pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + as_iterable: false + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 10 + label_override: "ALA_ALA" + +num_sampling_steps_per_batch: 100000 +num_batches: 1 +num_init_samples_per_dataset: 1 +repeat_init_samples: 1 +continue_chain: false + +# Add your checkpoint path here - update with actual trained model path +wandb_train_run_path: sule-shashank/jamun/0mu06yg4 +# checkpoint_dir: /data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-02_00-37-08/checkpoints +checkpoint_type: last + +sigma: 0.04 +M: 1.0 +delta: ${sigma} +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: 100.0 + +sampler: + _target_: jamun.sampling.SamplerMemory + devices: 1 + +# Evaluation dataset - standard ALA_ALA from capped diamines for computing metrics +eval_dataset: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 100 + max_datasets: 1 + label_override: "ALA_ALA" + +# Override ALL callbacks to use eval_dataset for metrics computation +callbacks: + measure_sampling_time: + _target_: jamun.callbacks.sampler.MeasureSamplingTimeCallback + chemical_validity: + _target_: jamun.callbacks.sampler.ChemicalValidityMetricsCallback + datasets: ${eval_dataset} + bond_length_tolerance: 0.2 + volume_exclusion_tolerance: 0.1 + num_molecules_per_trajectory: 100 + ramachandran_plot: + _target_: jamun.callbacks.sampler.RamachandranPlotMetricsCallback + datasets: ${eval_dataset} + trajectory_visualizer: + _target_: jamun.callbacks.sampler.TrajectoryVisualizerCallback + datasets: ${eval_dataset} + num_frames_to_animate: 100 + sample_visualizer: + _target_: jamun.callbacks.sampler.SampleVisualizerCallback + datasets: ${eval_dataset} + num_samples_to_plot: 16 + subsample: 100 + score_distribution: + _target_: jamun.callbacks.sampler.ScoreDistributionCallback + datasets: ${eval_dataset} + save_trajectory: + _target_: jamun.callbacks.sampler.SaveTrajectoryCallback + datasets: ${eval_dataset} + +logger: + wandb: + group: sample_ALA_ALA_enhanced_sampling + notes: "Sampling from enhanced sampling trained conditional denoiser" + tags: ["sample", "enhanced_sampling", "conditional_denoiser", "ALA_ALA", "sunny-paper-304"] \ No newline at end of file diff --git a/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml b/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml new file mode 100644 index 0000000..e784b2f --- /dev/null +++ b/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml @@ -0,0 +1,60 @@ +# @package _global_ +defaults: + - override /model: denoiser_multimeasurement + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + # max_datasets: 1 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + #max_datasets: ${data.datamodule.datasets.train.max_datasets} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 3 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + # Override multimeasurement parameters for this experiment + N_measurements: 5 + N_measurements_hidden: 5 + max_graphs_per_batch: ${data.datamodule.batch_size} + +trainer: + val_check_interval: 0.5 + max_epochs: 100 + devices: 1 + +logger: + wandb: + group: ALA_ALA_enhanced_full_grid_multimeasurement + notes: "Training multimeasurement model on ALA_ALA enhanced dataset with trajectory-based split" + tags: ["multimeasurement", "ala_ala", "enhanced_dataset", "trajectory_split"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_denoised_conditioner.yaml b/configs/experiment/train_enhanced_denoised_conditioner.yaml new file mode 100644 index 0000000..099480a --- /dev/null +++ b/configs/experiment/train_enhanced_denoised_conditioner.yaml @@ -0,0 +1,60 @@ +# @package _global_ +# Training DenoisedConditioner on enhanced sampling data + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + # max_datasets: 1 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + #max_datasets: ${data.datamodule.datasets.train.max_datasets} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.DenoisedConditioner + N_structures: ${model.arch.N_structures} + pretrained_model_path: "sule-shashank/jamun/370wpt17" + c_in: null # Will be computed automatically by training script + +trainer: + val_check_interval: 0.5 + max_epochs: 50 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "DenoisedConditioner pretrained from 370wpt17" + tags: ["denoised_conditioner", "enhanced_sampling", "pretrained"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_mean_conditioner.yaml b/configs/experiment/train_enhanced_mean_conditioner.yaml new file mode 100644 index 0000000..84dd0c8 --- /dev/null +++ b/configs/experiment/train_enhanced_mean_conditioner.yaml @@ -0,0 +1,58 @@ +# @package _global_ +# Training DenoisedConditioner on enhanced sampling data + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + # max_datasets: 1 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + #max_datasets: ${data.datamodule.datasets.train.max_datasets} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.MeanConditioner + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 50 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "Mean conditioner" + tags: ["mean_conditioner", "enhanced_sampling"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_position_conditioner.yaml b/configs/experiment/train_enhanced_position_conditioner.yaml new file mode 100644 index 0000000..74535b3 --- /dev/null +++ b/configs/experiment/train_enhanced_position_conditioner.yaml @@ -0,0 +1,56 @@ +# @package _global_ +# Training PositionConditioner on enhanced sampling data + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 50 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "PositionConditioner on enhanced sampling data" + tags: ["position_conditioner", "enhanced_sampling"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_self_conditioner.yaml b/configs/experiment/train_enhanced_self_conditioner.yaml new file mode 100644 index 0000000..31f6275 --- /dev/null +++ b/configs/experiment/train_enhanced_self_conditioner.yaml @@ -0,0 +1,56 @@ +# @package _global_ +# Training SelfConditioner on enhanced sampling data + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.SelfConditioner + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 50 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "SelfConditioner on enhanced sampling data" + tags: ["self_conditioner", "enhanced_sampling"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_spiked_conditioner.yaml b/configs/experiment/train_enhanced_spiked_conditioner.yaml new file mode 100644 index 0000000..347e05d --- /dev/null +++ b/configs/experiment/train_enhanced_spiked_conditioner.yaml @@ -0,0 +1,59 @@ +# @package _global_ +# Training DenoiserSpiked with ConditionerSpiked on enhanced sampling data + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + # max_datasets: 10 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + #max_datasets: ${data.datamodule.datasets.train.max_datasets} + +model: + _target_: jamun.model.denoiser_spiked.DenoiserSpiked + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.ConditionerSpiked + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 25 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_spiked_conditioner + notes: "DenoiserSpiked with ConditionerSpiked - clean structure conditioning" + tags: ["spiked_conditioner", "enhanced_sampling", "clean_conditioning"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_standard_jamun.yaml b/configs/experiment/train_enhanced_standard_jamun.yaml new file mode 100644 index 0000000..79c85d8 --- /dev/null +++ b/configs/experiment/train_enhanced_standard_jamun.yaml @@ -0,0 +1,52 @@ +# @package _global_ +# Training standard JAMUN Denoiser on enhanced sampling data + +defaults: + - override /model: denoiser + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + max_radius: 1000.0 + optim: + lr: 0.002 + +trainer: + val_check_interval: 0.5 + max_epochs: 50 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_conditioning_comparison + notes: "Standard JAMUN Denoiser on enhanced sampling data" + tags: ["standard_jamun", "enhanced_sampling", "no_conditioning"] \ No newline at end of file diff --git a/configs/experiment/train_test_single_shape.yaml b/configs/experiment/train_test_single_shape.yaml index 7a11821..9f15a0f 100644 --- a/configs/experiment/train_test_single_shape.yaml +++ b/configs/experiment/train_test_single_shape.yaml @@ -3,9 +3,9 @@ model: sigma_distribution: _target_: jamun.distributions.ConstantSigma - sigma: 0.01 + sigma: 0.04 arch: - n_layers: 2 + n_layers: 4 max_radius: 1000.0 optim: lr: 0.002 @@ -26,7 +26,7 @@ data: pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] as_iterable: false - subsample: 100 + subsample: 10 val: @@ -36,7 +36,7 @@ data: pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] as_iterable: false - subsample: 100 + subsample: 10 test: _target_: jamun.data.parse_datasets_from_directory @@ -45,7 +45,7 @@ data: pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] as_iterable: false - subsample: 100 + subsample: 10 trainer: diff --git a/configs/experiment/train_test_single_shape_conditional.yaml b/configs/experiment/train_test_single_shape_conditional.yaml index 8fd33de..a1732de 100644 --- a/configs/experiment/train_test_single_shape_conditional.yaml +++ b/configs/experiment/train_test_single_shape_conditional.yaml @@ -59,10 +59,13 @@ model: conditioner: _target_: jamun.model.conditioners.PositionConditioner N_structures: ${model.arch.N_structures} + pretrained_model_path: null + c_in: null + trainer: val_check_interval: 0.5 - max_epochs: 500 + max_epochs: 1 logger: diff --git a/configs/experiment/train_test_single_shape_enhanced_sampling.yaml b/configs/experiment/train_test_single_shape_fake_enhanced_sampling.yaml similarity index 94% rename from configs/experiment/train_test_single_shape_enhanced_sampling.yaml rename to configs/experiment/train_test_single_shape_fake_enhanced_sampling.yaml index 6f8185e..3bae047 100644 --- a/configs/experiment/train_test_single_shape_enhanced_sampling.yaml +++ b/configs/experiment/train_test_single_shape_fake_enhanced_sampling.yaml @@ -17,7 +17,7 @@ data: traj_pattern: "^(.*).xtc" pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" as_iterable: false - subsample: 10 + subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 max_datasets: 5000 @@ -28,7 +28,7 @@ data: traj_pattern: "^(.*).xtc" pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" as_iterable: false - subsample: 10 + subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: 1 max_datasets: 5000 @@ -39,7 +39,7 @@ data: traj_pattern: "^(.*).xtc" pdb_file: "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" as_iterable: false - subsample: 10 + subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: 1 max_datasets: 5000 @@ -57,6 +57,8 @@ model: conditioner: _target_: jamun.model.conditioners.SelfConditioner N_structures: ${model.arch.N_structures} + pretrained_model_path: null + c_in: null trainer: val_check_interval: 0.5 diff --git a/scratch/test_conditional.py b/scratch/test_conditional.py deleted file mode 100644 index f865fd6..0000000 --- a/scratch/test_conditional.py +++ /dev/null @@ -1,93 +0,0 @@ -import e3nn -e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys -import os -import hydra -from omegaconf import OmegaConf -import torch -import torch_geometric -from jamun.hydra import instantiate_dict_cfg -import pdb -import jamun -from jamun.utils import compute_average_squared_distance_from_datasets - -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ -JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") -JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") - -project_root = "/homefs/home/sules/jamun" # Adjust if necessary -if project_root not in sys.path: - sys.path.insert(0, project_root) - print(f"Added '{project_root}' to sys.path for module discovery.") -else: - print(f"'{project_root}' is already in sys.path.") - -def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: - """Computes the average squared distance for normalization from the data.""" - datamodule = hydra.utils.instantiate(cfg.data.datamodule) - datamodule.setup("compute_normalization") - train_datasets = datamodule.datasets["train"] - cutoff = cfg.model.max_radius - average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) - return average_squared_distance - - -@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") -def main(cfg): - # Load the test config - average_squared_distance = compute_average_squared_distance_from_config(cfg) - cfg.model.average_squared_distance = average_squared_distance - # breakpoint() - - # # First merge test config into base config, then override with test config - # cfg = OmegaConf.merge(cfg, test_cfg) - # cfg = OmegaConf.merge(cfg, test_cfg, override=True) - # breakpoint() - print("Loading datamodule...") - datamodule = hydra.utils.instantiate(cfg.data.datamodule) - datamodule.setup('test') - breakpoint() - - print("Loading model...") - model = hydra.utils.instantiate(cfg.model) - breakpoint() - - # Get a single batch - print("Getting a batch of data...") - train_loader = datamodule.train_dataloader() - _, batch = next(enumerate(train_loader)) - breakpoint() - - # # Move to CPU - # batch = batch.to("cpu") - # model = model.to("cpu") - - print(f"Batch shape: {batch.pos.shape}") - print(f"Hidden state shape: {[h.shape for h in batch.hidden_state]}") - breakpoint() - - # Test forward pass - print("Testing forward pass...") - # with torch.no_grad(): - # sigma = model.sigma_distribution.sample() - # x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) - - # print(f"Input shape: {batch.pos.shape}") - # print(f"Noisy shape: {y.pos.shape}") - # print(f"Output shape: {xhat.pos.shape}") - - # Test single backward pass computation - trainer = hydra.utils.instantiate(cfg.trainer) - trainer.fit(model, datamodule=datamodule, ckpt_path=None) - - - # loss = model.training_step(batch, 0) - # breakpoint() - # print("Testing loss computation...") - # loss, aux = model.compute_loss(x_target, xhat, sigma) - # print(f"Loss: {loss.mean().item():.4f}") - # print(f"Metrics: {aux}") - -if __name__ == "__main__": - main() \ No newline at end of file diff --git a/scratch/visualize_traj_data.py b/scratch/visualize_traj_data.py index 16c8141..4da7d24 100644 --- a/scratch/visualize_traj_data.py +++ b/scratch/visualize_traj_data.py @@ -1,48 +1,59 @@ -# xplore generated trajectories import mdtraj as md -import os -# --- Option 2: Loading your own DCD and Topology file --- -print("\n--- Loading Your Own DCD and Topology File (Example) ---") -# Replace these with the actual paths to your files -dcd_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/predicted_samples/dcd/joined.dcd" # Your DCD trajectory file -topology_file_path = f"{JAMUN_ROOT_PATH}/outputs/sample/dev/runs/2025-06-04_22-46-33/sampler/AA/topology.pdb" # Your topology file (e.g., .pdb, .prmtop, .psf) - -# Create dummy files for this example to run without error if you don't have them -# In a real scenario, you would have your actual DCD and PDB files. -print(f'DCD file path exists: {os.path.exists(dcd_file_path)}') -print(f'Topology file path exists: {os.path.exists(topology_file_path)}') -try: - print(f"Attempting to load trajectory: {dcd_file_path}") - print(f"Using topology: {topology_file_path}") - - # The 'top' argument is crucial for DCD files - traj_custom = md.load_dcd(dcd_file_path, top=topology_file_path) - - print(f"Successfully loaded custom trajectory!") - print(f"Number of frames: {traj_custom.n_frames}") - print(f"Number of atoms: {traj_custom.n_atoms}") - # You can now perform analysis on traj_custom - # For example, calculate RMSD, distances, angles, etc. - -except FileNotFoundError: - print(f"Error: One or both files not found: {dcd_file_path}, {topology_file_path}") -except Exception as e: - print(f"An error occurred while loading your files: {e}") -print("-" * 30) - -# %% -from jamun.metrics._ramachandran import plot_ramachandran - -phi = md.compute_phi(traj_custom) -psi = md.compute_psi(traj_custom) - +import os import matplotlib.pyplot as plt -import numpy as np -fig = plt.figure() -ax = fig.add_subplot() -s = ax.scatter(phi[1], psi[1], cmap='hot', alpha=1.0) -ax.set_xlim((-np.pi, np.pi)) -ax.set_ylim((-np.pi, np.pi)) -c = fig.colorbar(s) - -print("hello world") \ No newline at end of file +import numpy as np +import itertools +import matplotlib.colors as colors + +# Set file paths for ALA_ALA in capped diamines +xtc_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.xtc" +pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + +output_dir = "/data2/sules/ramachandran_plots_ala_ala_fake_enhanced_data" +os.makedirs(output_dir, exist_ok=True) + +print(f"XTC file exists: {os.path.exists(xtc_file)}") +print(f"PDB file exists: {os.path.exists(pdb_file)}") + +# Load the trajectory (subsample=1 means load all frames) +traj = md.load(xtc_file, top=pdb_file) +print(f"Loaded trajectory with {traj.n_frames} frames and {traj.n_atoms} atoms.") + +# Compute backbone dihedrals (phi and psi) +phi_indices, phi_angles = md.compute_phi(traj) +psi_indices, psi_angles = md.compute_psi(traj) + +num_phi = phi_angles.shape[1] +num_psi = psi_angles.shape[1] + +# Collect all dihedral arrays in a dict for easy access +# Each entry is (n_frames,) +dihedrals = {} +for i in range(num_phi): + dihedrals[f'phi_{i+1}'] = phi_angles[:, i] +for i in range(num_psi): + dihedrals[f'psi_{i+1}'] = psi_angles[:, i] + +dihedral_names = list(dihedrals.keys()) + +# Make 2D histograms for all pairs +for name1, name2 in itertools.combinations(dihedral_names, 2): + x = dihedrals[name1] + y = dihedrals[name2] + plt.figure(figsize=(8, 8)) + plt.hist2d(x, y, bins=100, range=((-np.pi, np.pi), (-np.pi, np.pi)), cmap='viridis', norm=colors.LogNorm()) + plt.colorbar(label='Density') + plt.title(f'2D Histogram: {name1} vs {name2}') + plt.xlabel(f'{name1} (radians)') + plt.ylabel(f'{name2} (radians)') + plt.xlim(-np.pi, np.pi) + plt.ylim(-np.pi, np.pi) + plt.grid(True, linestyle='--', alpha=0.6) + plt.axhline(0, color='k', linestyle='--', linewidth=0.5) + plt.axvline(0, color='k', linestyle='--', linewidth=0.5) + output_filename = f"hist2d_{name1}_vs_{name2}_true_distribution.png" + output_path = os.path.join(output_dir, output_filename) + plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.close() + +print(f"All 2D histograms saved in {output_dir}") \ No newline at end of file diff --git a/scripts/concatenate_trajectories.py b/scripts/concatenate_trajectories.py new file mode 100755 index 0000000..9d4955e --- /dev/null +++ b/scripts/concatenate_trajectories.py @@ -0,0 +1,112 @@ +#!/usr/bin/env python3 + +import os +import glob +import mdtraj as md +import numpy as np +from pathlib import Path +import argparse +from tqdm import tqdm + +def concatenate_trajectories(folder_path, pdb_file, output_name="ALA_ALA.xtc"): + """ + Concatenate all .xtc trajectories in a folder into a single long trajectory. + + Args: + folder_path (str): Path to the folder containing .xtc files + pdb_file (str): Path to the PDB topology file + output_name (str): Name of the output trajectory file + """ + folder_path = Path(folder_path) + + # Find all .xtc files in the folder + xtc_files = sorted(glob.glob(str(folder_path / "*.xtc"))) + + # Filter out any existing ALA_ALA.xtc to avoid including it in concatenation + xtc_files = [f for f in xtc_files if not f.endswith("ALA_ALA.xtc")] + + if not xtc_files: + print(f"No .xtc files found in {folder_path}") + return + + print(f"Found {len(xtc_files)} .xtc files in {folder_path}") + print(f"First few files: {xtc_files[:5]}") + + # Load the first trajectory to get the topology + print("Loading first trajectory to get topology...") + first_traj = md.load(xtc_files[0], top=pdb_file) + print(f"Topology: {first_traj.n_atoms} atoms, {first_traj.n_frames} frames") + + # Initialize the concatenated trajectory with the first one + concat_traj = first_traj + + # Load and concatenate the rest of the trajectories + for xtc_file in tqdm(xtc_files[1:], desc="Concatenating trajectories", unit="file"): + try: + traj = md.load(xtc_file, top=pdb_file) + concat_traj = concat_traj.join(traj) + + except Exception as e: + tqdm.write(f"Error loading {os.path.basename(xtc_file)}: {e}") + continue + + # Save the concatenated trajectory + output_path = folder_path / output_name + print(f"Saving concatenated trajectory to {output_path}") + print(f"Final trajectory: {concat_traj.n_frames} frames, {concat_traj.n_atoms} atoms") + + concat_traj.save_xtc(str(output_path)) + print(f"Successfully saved {output_path}") + + return concat_traj + +def main(): + parser = argparse.ArgumentParser(description="Concatenate .xtc trajectories in folders") + parser.add_argument("--base-dir", + default="/data2/sules/fake_enhanced_data/ALA_ALA_organized", + help="Base directory containing train/val/test folders") + parser.add_argument("--pdb-file", + default="/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb", + help="PDB topology file") + parser.add_argument("--folders", nargs='+', + default=["train", "val", "test"], + help="Folders to process") + + args = parser.parse_args() + + base_dir = Path(args.base_dir) + pdb_file = args.pdb_file + + # Check if PDB file exists + if not os.path.exists(pdb_file): + print(f"Error: PDB file not found: {pdb_file}") + return + + print(f"Using PDB file: {pdb_file}") + print(f"Base directory: {base_dir}") + + # Process each folder + for folder in args.folders: + folder_path = base_dir / folder + + if not folder_path.exists(): + print(f"Folder {folder_path} does not exist, skipping...") + continue + + print(f"\n{'='*60}") + print(f"Processing folder: {folder}") + print(f"{'='*60}") + + try: + concatenate_trajectories(folder_path, pdb_file) + except Exception as e: + print(f"Error processing {folder}: {e}") + import traceback + traceback.print_exc() + + print(f"\n{'='*60}") + print("Concatenation completed!") + print(f"{'='*60}") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scripts/slurm/check_sweep_runs.sh b/scripts/slurm/check_sweep_runs.sh new file mode 100755 index 0000000..277e2e9 --- /dev/null +++ b/scripts/slurm/check_sweep_runs.sh @@ -0,0 +1,148 @@ +#!/bin/bash + +# Helper script to check how many runs are in the enhanced sampling sweep +# and preview the array range for the SLURM script + +set -e + +# Configuration +WANDB_GROUP="fake_enhanced_data_jul_11_sweep" +ENTITY="sule-shashank" +PROJECT="jamun" + +echo "=========================================" +echo "Enhanced Sampling Sweep - Run Check" +echo "Group: $WANDB_GROUP" +echo "Entity/Project: $ENTITY/$PROJECT" +echo "=========================================" + +# Initialize conda +echo "Initializing conda..." +source ~/.bashrc +eval "$(conda shell.bash hook)" + +# Activate conda environment +echo "Activating jamun environment..." +conda activate jamun + +# Python script to fetch wandb runs and show summary +python -c " +import wandb +import sys +from collections import defaultdict + +# Configuration +entity = '$ENTITY' +project = '$PROJECT' +group = '$WANDB_GROUP' + +try: + # Initialize wandb API + api = wandb.Api() + + # Get all runs from the specified group + print(f'Fetching runs from {entity}/{project} with group \"{group}\"...') + runs = api.runs(f'{entity}/{project}', filters={'group': group}) + runs_list = list(runs) + + print(f'\\nFound {len(runs_list)} runs in group \"{group}\"') + + if len(runs_list) == 0: + print('No runs found in this group!') + sys.exit(1) + + # Collect parameter combinations + param_combinations = [] + conditioner_counts = defaultdict(int) + sigma_counts = defaultdict(int) + lag_time_counts = defaultdict(int) + + for i, run in enumerate(runs_list): + try: + config = run.config + cfg = config.get('cfg', {}) + + conditioner = cfg.get('model', {}).get('conditioner', {}).get('_target_', 'Unknown') + sigma = cfg.get('model', {}).get('sigma_distribution', {}).get('sigma', 'Unknown') + total_lag_time = cfg.get('data', {}).get('datamodule', {}).get('datasets', {}).get('train', {}).get('total_lag_time', 'Unknown') + + conditioner_name = conditioner.split('.')[-1] if conditioner != 'Unknown' else 'Unknown' + + param_combinations.append({ + 'index': i, + 'name': run.name, + 'run_path': '/'.join(run.path), + 'conditioner': conditioner_name, + 'sigma': sigma, + 'lag_time': total_lag_time, + 'state': run.state + }) + + conditioner_counts[conditioner_name] += 1 + sigma_counts[sigma] += 1 + lag_time_counts[total_lag_time] += 1 + + except Exception as e: + print(f'Warning: Could not extract parameters for run {i}: {e}') + param_combinations.append({ + 'index': i, + 'name': run.name, + 'run_path': '/'.join(run.path), + 'conditioner': 'Error', + 'sigma': 'Error', + 'lag_time': 'Error', + 'state': run.state + }) + + # Print summary + print('\\n========================================') + print('PARAMETER DISTRIBUTION SUMMARY:') + print('========================================') + + print('\\nConditioner types:') + for conditioner, count in sorted(conditioner_counts.items()): + print(f' {conditioner}: {count} runs') + + print('\\nSigma values:') + for sigma, count in sorted(sigma_counts.items()): + print(f' {sigma}: {count} runs') + + print('\\nLag time values:') + for lag_time, count in sorted(lag_time_counts.items()): + print(f' {lag_time}: {count} runs') + + # Print first 5 runs as examples + print('\\n========================================') + print('FIRST 5 RUNS (EXAMPLES):') + print('========================================') + print(f'{'Index':<6} {'Name':<25} {'Conditioner':<18} {'Sigma':<8} {'Lag':<5} {'State':<10}') + print('-' * 75) + + for combo in param_combinations[:5]: + print(f'{combo[\"index\"]:<6} {combo[\"name\"]:<25} {combo[\"conditioner\"]:<18} {combo[\"sigma\"]:<8} {combo[\"lag_time\"]:<5} {combo[\"state\"]:<10}') + + if len(param_combinations) > 5: + print(f'... and {len(param_combinations) - 5} more runs') + + print('\\n========================================') + print('SLURM ARRAY CONFIGURATION:') + print('========================================') + print(f'Total runs: {len(runs_list)}') + print(f'Array range: 0-{len(runs_list) - 1}') + print(f'\\nUpdate your SLURM script with:') + print(f'#SBATCH --array=0-{len(runs_list) - 1}') + print('\\nTo submit the job:') + print('sbatch scripts/slurm/sweep_enhanced_sampling.sh') + print('\\nTo submit a subset (e.g., first 5 runs):') + print('sbatch --array=0-4 scripts/slurm/sweep_enhanced_sampling.sh') + +except Exception as e: + print(f'Error: {e}', file=sys.stderr) + import traceback + traceback.print_exc() + sys.exit(1) +" + +echo "=========================================" +echo "Run check completed!" +echo "=========================================" \ No newline at end of file diff --git a/scripts/slurm/debug_sweep_enhanced_sampling.sh b/scripts/slurm/debug_sweep_enhanced_sampling.sh new file mode 100755 index 0000000..bfd8b34 --- /dev/null +++ b/scripts/slurm/debug_sweep_enhanced_sampling.sh @@ -0,0 +1,189 @@ +#!/bin/bash + +# Debug version of the enhanced sampling sweep script +# Usage: ./debug_sweep_enhanced_sampling.sh + +set -e + +# Check if run index is provided +if [ $# -ne 1 ]; then + echo "Usage: $0 " + echo "Please provide the 0-based index of the run to process." + exit 1 +fi + +RUN_INDEX=$1 + +# Set up environment +export JAMUN_ROOT_PATH=/homefs/home/sules/jamun +cd $JAMUN_ROOT_PATH + +# Initialize conda +echo "Initializing conda..." +source ~/.bashrc +eval "$(conda shell.bash hook)" + +# Activate conda environment +echo "Activating jamun environment..." +conda activate jamun + +# Configuration +WANDB_GROUP="fake_enhanced_data_jul_11_sweep" +ENTITY="sule-shashank" +PROJECT="jamun" + +echo "=========================================" +echo "DEBUG MODE - Enhanced Sampling Sweep" +echo "Working directory: $(pwd)" +echo "Processing run index: $RUN_INDEX from group: $WANDB_GROUP" +echo "=========================================" + +# Python script to fetch wandb runs and build jamun_sample command +python -c " +import wandb +import sys +import os + +# Configuration +entity = '$ENTITY' +project = '$PROJECT' +group = '$WANDB_GROUP' +run_index = $RUN_INDEX + +try: + # Initialize wandb API + api = wandb.Api() + + # Get all runs from the specified group + print(f'Fetching runs from {entity}/{project} with group \"{group}\"...') + runs = api.runs(f'{entity}/{project}', filters={'group': group}) + runs_list = list(runs) + + print(f'Found {len(runs_list)} runs in group \"{group}\"') + + # Check if run_index is valid + if run_index >= len(runs_list) or run_index < 0: + print(f'Error: Run index {run_index} is out of bounds. The group has {len(runs_list)} runs (indices 0 to {len(runs_list) - 1}).', file=sys.stderr) + sys.exit(1) + + # Get the specific run + run = runs_list[run_index] + run_path = '/'.join(run.path) + + print(f'\\nProcessing run: {run.name} ({run_path})') + print(f'Run URL: {run.url}') + print(f'Run state: {run.state}') + + # Extract parameters from the run config + config = run.config + print(f'\\nAvailable config keys: {list(config.keys())}') + + cfg_key = 'cfg' # This is the key used in jamun configs + + if cfg_key not in config: + print(f'Error: Config key \"{cfg_key}\" not found in run config.', file=sys.stderr) + sys.exit(1) + + cfg = config[cfg_key] + print(f'Config structure keys: {list(cfg.keys())}') + + # Extract required parameters with detailed debugging + try: + print(f'\\nExtracting parameters...') + + # Extract conditioner + conditioner = cfg['model']['conditioner']['_target_'] + print(f' āœ“ Conditioner: {conditioner}') + + # Extract sigma + sigma = cfg['model']['sigma_distribution']['sigma'] + print(f' āœ“ Sigma: {sigma}') + + # Extract total_lag_time + total_lag_time = cfg['data']['datamodule']['datasets']['train']['total_lag_time'] + print(f' āœ“ Total Lag Time: {total_lag_time}') + + # Optional: Extract other useful parameters + model_arch_N_structures = cfg.get('model', {}).get('arch', {}).get('N_structures', total_lag_time) + print(f' āœ“ Model N_structures: {model_arch_N_structures}') + + except KeyError as e: + print(f'Error: Could not extract required parameter: {e}', file=sys.stderr) + print(f'Available model keys: {cfg.get(\"model\", {}).keys()}', file=sys.stderr) + if 'model' in cfg: + print(f'Available model.conditioner keys: {cfg[\"model\"].get(\"conditioner\", {}).keys()}', file=sys.stderr) + print(f'Available model.sigma_distribution keys: {cfg[\"model\"].get(\"sigma_distribution\", {}).keys()}', file=sys.stderr) + if 'data' in cfg: + print(f'Available data keys: {cfg[\"data\"].keys()}', file=sys.stderr) + sys.exit(1) + + # Ensure all required parameters are present + if not all(v is not None for v in [run_path, conditioner, total_lag_time, sigma]): + print(f'Error: Could not extract all required parameters for run at index {run_index}.', file=sys.stderr) + sys.exit(1) + + # Create a meaningful run group name for sampling + conditioner_name = conditioner.split('.')[-1] # Get class name without module path + sampling_group = 'sample_enhanced_sampling_from_jul_11' + + # Create tags for better organization + tags = [ + f'sweep_run_{run_index}', + f'conditioner_{conditioner_name}', + f'sigma_{sigma}', + f'lag_time_{total_lag_time}', + 'enhanced_sampling', + 'sample_from_sweep' + ] + tags_string = '[' + ', '.join(f'\"{tag}\"' for tag in tags) + ']' + + print('\\n========================================') + print(f'Parameters extracted successfully!') + print(f' Run Path: {run_path}') + print(f' Conditioner: {conditioner}') + print(f' Conditioner Name: {conditioner_name}') + print(f' Sigma: {sigma}') + print(f' Total Lag Time: {total_lag_time}') + print(f' Sample Group: {sampling_group}') + print('========================================\\n') + + # Build the jamun_sample command + # Note: We don't override model.conditioner._target_ because the checkpoint + # already contains the correct conditioner configuration with all parameters + cmd_parts = [ + 'jamun_sample', + '--config-dir=configs', + 'experiment=sample_enhanced_sampling_single_shape.yaml', + f'++wandb_train_run_path={run_path}', + f'++init_datasets.total_lag_time={total_lag_time}', + f'++sigma={sigma}', + f'++delta={sigma}', + f'++logger.wandb.group={sampling_group}', + f'++logger.wandb.tags={tags_string}', + f'++logger.wandb.notes=\"Sampling from enhanced sampling sweep run {run_index} - {conditioner_name} sigma={sigma} lag={total_lag_time}\"', + f'++run_key=sweep_sample_{run_index}_{conditioner_name}_sigma_{sigma}_lag_{total_lag_time}' + ] + + print('DEBUG: Generated jamun_sample command:') + print('=' * 80) + cmd_string = ' \\\\\\n '.join(cmd_parts) + print(cmd_string) + print('=' * 80) + + print('\\nDEBUG: Single line command:') + print('=' * 80) + print(' '.join(cmd_parts)) + print('=' * 80) + + print(f'\\nDEBUG: Successfully processed run index {run_index}') + +except Exception as e: + print(f'Error: {e}', file=sys.stderr) + import traceback + traceback.print_exc() + sys.exit(1) +" + +echo "=========================================" +echo "DEBUG: Finished processing run index: $RUN_INDEX" +echo "=========================================" \ No newline at end of file diff --git a/scripts/slurm/run_conditioner_experiment.sh b/scripts/slurm/run_conditioner_experiment.sh new file mode 100644 index 0000000..8812440 --- /dev/null +++ b/scripts/slurm/run_conditioner_experiment.sh @@ -0,0 +1,81 @@ +#!/bin/bash + +#SBATCH --job-name=conditioner_lag_sweep +#SBATCH --array=0-5 +#SBATCH --partition=gpu2 +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=12 +#SBATCH --time=1-0 + +# Experiment: Sweep over SelfConditioner, PositionConditioner, and DenoisedConditioner +# with different total lag times (2, 5, 8) on enhanced sampling data with 2 layers +# Testing mode: 1 epoch, max_datasets=1 + +# Array of config names to run (base configs) +CONFIG_NAMES=( + "train_enhanced_self_conditioner" + "train_enhanced_position_conditioner" + "train_enhanced_denoised_conditioner" +) + +# Array of conditioner names for logging +CONDITIONER_NAMES=( + "SelfConditioner" + "PositionConditioner" + "DenoisedConditioner" +) + +# Array of lag times to test +LAG_TIMES=(2 5) + +# Calculate which conditioner and lag time based on array index +# Array index 0-5 maps to: +# 0-1: SelfConditioner with lag_time 2,5 +# 2-3: PositionConditioner with lag_time 2,5 +# 4-5: DenoisedConditioner with lag_time 2,5 +CONDITIONER_IDX=$((SLURM_ARRAY_TASK_ID / 2)) +LAG_TIME_IDX=$((SLURM_ARRAY_TASK_ID % 2)) + +CONFIG_NAME=${CONFIG_NAMES[$CONDITIONER_IDX]} +CONDITIONER_NAME=${CONDITIONER_NAMES[$CONDITIONER_IDX]} +LAG_TIME=${LAG_TIMES[$LAG_TIME_IDX]} + +echo "=== SLURM Array Job ${SLURM_ARRAY_TASK_ID}: Training ${CONDITIONER_NAME} with lag_time=${LAG_TIME} ===" +echo "Job ID: ${SLURM_JOB_ID}" +echo "Array Task ID: ${SLURM_ARRAY_TASK_ID}" +echo "Conditioner Index: ${CONDITIONER_IDX}" +echo "Lag Time Index: ${LAG_TIME_IDX}" +echo "Config: ${CONFIG_NAME}" +echo "Lag Time: ${LAG_TIME}" +echo "Starting at $(date)" +echo "" + +# Set environment variables +export JAMUN_DATA_PATH=/data/bucket/kleinhej/ +export WANDB_PROJECT=jamun + +# Activate conda environment +source ~/.bashrc +conda activate jamun + +# Create logs directory if it doesn't exist +mkdir -p logs + +# Build command with testing overrides and lag time sweep +CMD="jamun_train --config-dir=configs experiment=${CONFIG_NAME}" +CMD="${CMD} ++data.datamodule.datasets.train.total_lag_time=${LAG_TIME}" +CMD="${CMD} ++data.datamodule.datasets.val.total_lag_time=${LAG_TIME}" +CMD="${CMD} ++trainer.max_epochs=100" + +echo "Running command:" +echo "${CMD}" +echo "" + +eval ${CMD} + +echo "" +echo "=== ${CONDITIONER_NAME} (lag_time=${LAG_TIME}) Training Complete ===" +echo "Finished at $(date)" +echo "Check Weights & Biases group 'conditioner_lag_sweep_test' for results" \ No newline at end of file diff --git a/scripts/slurm/sweep_enhanced_sampling.sh b/scripts/slurm/sweep_enhanced_sampling.sh new file mode 100755 index 0000000..a3fdea4 --- /dev/null +++ b/scripts/slurm/sweep_enhanced_sampling.sh @@ -0,0 +1,179 @@ +#!/bin/bash +#SBATCH --job-name=sweep_enhanced_sampling +#SBATCH --partition=gpu2 +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=12 +#SBATCH --time=2-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-23 # Adjust this range based on number of runs in your sweep +#SBATCH --output=logs/%A_%a_sweep_enhanced_sampling.log +#SBATCH --error=logs/%A_%a_sweep_enhanced_sampling.err + +# Set up environment +set -e +export JAMUN_ROOT_PATH=/homefs/home/sules/jamun +cd $JAMUN_ROOT_PATH + +# Create logs directory if it doesn't exist +mkdir -p logs + +# Initialize conda +echo "Initializing conda..." +source ~/.bashrc +eval "$(conda shell.bash hook)" + +# Activate conda environment +echo "Activating jamun environment..." +conda activate jamun + +# Configuration +WANDB_GROUP="fake_enhanced_data_jul_11_sweep" +ENTITY="sule-shashank" +PROJECT="jamun" +RUN_INDEX=$SLURM_ARRAY_TASK_ID + +echo "=========================================" +echo "Enhanced Sampling Sweep - Production Run" +echo "SLURM Job ID: $SLURM_JOB_ID" +echo "Array Task ID: $SLURM_ARRAY_TASK_ID" +echo "Running on hostname: $(hostname)" +echo "Working directory: $(pwd)" +echo "Starting time: $(date)" +echo "Processing run index: $RUN_INDEX from group: $WANDB_GROUP" +echo "=========================================" + +# Python script to fetch wandb runs and execute jamun_sample +python -c " +import wandb +import subprocess +import sys +import os + +# Configuration +entity = '$ENTITY' +project = '$PROJECT' +group = '$WANDB_GROUP' +run_index = $RUN_INDEX + +try: + # Initialize wandb API + api = wandb.Api() + + # Get all runs from the specified group + print(f'Fetching runs from {entity}/{project} with group \"{group}\"...') + runs = api.runs(f'{entity}/{project}', filters={'group': group}) + runs_list = list(runs) + + print(f'Found {len(runs_list)} runs in group \"{group}\"') + + # Check if run_index is valid + if run_index >= len(runs_list) or run_index < 0: + print(f'Error: Run index {run_index} is out of bounds. The group has {len(runs_list)} runs (indices 0 to {len(runs_list) - 1}).', file=sys.stderr) + sys.exit(1) + + # Get the specific run + run = runs_list[run_index] + run_path = '/'.join(run.path) + + print(f'\\nProcessing run: {run.name} ({run_path})') + print(f'Run URL: {run.url}') + print(f'Run state: {run.state}') + + # Extract parameters from the run config + config = run.config + cfg_key = 'cfg' # This is the key used in jamun configs + + if cfg_key not in config: + print(f'Error: Config key \"{cfg_key}\" not found in run config. Available keys: {list(config.keys())}', file=sys.stderr) + sys.exit(1) + + cfg = config[cfg_key] + + # Extract required parameters + try: + conditioner = cfg['model']['conditioner']['_target_'] + sigma = cfg['model']['sigma_distribution']['sigma'] + total_lag_time = cfg['data']['datamodule']['datasets']['train']['total_lag_time'] + + print(f'\\nExtracted parameters:') + print(f' Conditioner: {conditioner}') + print(f' Sigma: {sigma}') + print(f' Total Lag Time: {total_lag_time}') + + except KeyError as e: + print(f'Error: Could not extract required parameter: {e}', file=sys.stderr) + print(f'Available config structure: {cfg.keys()}', file=sys.stderr) + sys.exit(1) + + # Ensure all required parameters are present + if not all(v is not None for v in [run_path, conditioner, total_lag_time, sigma]): + print(f'Error: Could not extract all required parameters for run at index {run_index}.', file=sys.stderr) + sys.exit(1) + + # Create a meaningful run group name for sampling + conditioner_name = conditioner.split('.')[-1] # Get class name without module path + sampling_group = 'sample_enhanced_sampling_from_jul_11' + + # Create tags for better organization + tags = [ + f'sweep_run_{run_index}', + f'conditioner_{conditioner_name}', + f'sigma_{sigma}', + f'lag_time_{total_lag_time}', + 'enhanced_sampling', + 'sample_from_sweep' + ] + tags_string = '[' + ', '.join(f'\"{tag}\"' for tag in tags) + ']' + + print('\\n========================================') + print(f'Starting sampling for run at index {run_index}: {run_path}') + print(f' Conditioner: {conditioner}') + print(f' Sigma: {sigma}') + print(f' Total Lag Time: {total_lag_time}') + print(f' Sample Group: {sampling_group}') + print('========================================\\n') + + # Build the jamun_sample command + # Note: We don't override model.conditioner._target_ because the checkpoint + # already contains the correct conditioner configuration with all parameters + cmd = [ + 'jamun_sample', + '--config-dir=configs', + 'experiment=sample_enhanced_sampling_single_shape.yaml', + f'++wandb_train_run_path={run_path}', + f'++init_datasets.total_lag_time={total_lag_time}', + f'++sigma={sigma}', + f'++delta={sigma}', + f'++logger.wandb.group={sampling_group}', + f'++logger.wandb.tags={tags_string}', + f'++logger.wandb.notes=\"Sampling from enhanced sampling sweep run {run_index} - {conditioner_name} sigma={sigma} lag={total_lag_time}\"', + f'++run_key=sweep_sample_{run_index}_{conditioner_name}_sigma_{sigma}_lag_{total_lag_time}' + ] + + print('Executing command:') + print(' '.join(cmd)) + print('\\n' + '='*50 + '\\n') + + # Execute the command + result = subprocess.run(cmd, check=True, env=os.environ.copy()) + + print('\\n' + '='*50) + print(f'Successfully completed sampling for run index {run_index}') + print(f'End time: {os.popen(\"date\").read().strip()}') + +except subprocess.CalledProcessError as e: + print(f'Error running jamun_sample: {e}', file=sys.stderr) + sys.exit(1) +except Exception as e: + print(f'Error: {e}', file=sys.stderr) + import traceback + traceback.print_exc() + sys.exit(1) +" + +echo "=========================================" +echo "Finished processing run index: $RUN_INDEX" +echo "End time: $(date)" +echo "=========================================" \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_sampling_sweep.sh b/scripts/slurm/train_enhanced_sampling_sweep.sh index bd14fb7..db22dd6 100755 --- a/scripts/slurm/train_enhanced_sampling_sweep.sh +++ b/scripts/slurm/train_enhanced_sampling_sweep.sh @@ -7,7 +7,7 @@ #SBATCH --cpus-per-task 8 #SBATCH --time 3-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array=0-23 +#SBATCH --array=0-2 # Initialize conda source ~/.bashrc @@ -32,27 +32,30 @@ RUN_KEY=$(openssl rand -hex 12) echo "RUN_KEY = ${RUN_KEY}" # Define parameter arrays -CONDITIONERS=("jamun.model.conditioners.PositionConditioner" "jamun.model.conditioners.SelfConditioner") -CONDITIONER_NAMES=("PositionConditioner" "SelfConditioner") -SIGMAS=(0.01 0.04 0.08 0.1) -LAG_TIMES=(2 5 8) +CONDITIONERS=("jamun.model.conditioners.PositionConditioner") +CONDITIONER_NAMES=("PositionConditioner") +LAG_SUBSAMPLE_RATES=(1 5 10) + +# Fixed parameters +SIGMA=0.04 +TOTAL_LAG_TIME=5 +SUBSAMPLE=1 # Calculate parameter indices from SLURM_ARRAY_TASK_ID -# Total combinations: 2 conditioners * 4 sigmas * 3 lag_times = 24 -COND_IDX=$((SLURM_ARRAY_TASK_ID / 12)) -SIGMA_IDX=$(((SLURM_ARRAY_TASK_ID % 12) / 3)) -LAG_IDX=$((SLURM_ARRAY_TASK_ID % 3)) +# Total combinations: 1 conditioner * 3 lag_subsample_rates = 3 +LAG_SUBSAMPLE_IDX=${SLURM_ARRAY_TASK_ID} # Get parameter values -CONDITIONER=${CONDITIONERS[$COND_IDX]} -CONDITIONER_NAME=${CONDITIONER_NAMES[$COND_IDX]} -SIGMA=${SIGMAS[$SIGMA_IDX]} -LAG_TIME=${LAG_TIMES[$LAG_IDX]} +CONDITIONER=${CONDITIONERS[0]} +CONDITIONER_NAME=${CONDITIONER_NAMES[0]} +LAG_SUBSAMPLE_RATE=${LAG_SUBSAMPLE_RATES[$LAG_SUBSAMPLE_IDX]} echo "Parameter combination ${SLURM_ARRAY_TASK_ID}:" echo " Conditioner: ${CONDITIONER_NAME}" echo " Sigma: ${SIGMA}" -echo " Total lag time: ${LAG_TIME}" +echo " Total lag time: ${TOTAL_LAG_TIME}" +echo " Lag subsample rate: ${LAG_SUBSAMPLE_RATE}" +echo " Subsample: ${SUBSAMPLE}" nvidia-smi @@ -60,13 +63,14 @@ nvidia-smi jamun_train --config-dir=configs \ experiment=train_test_single_shape_enhanced_sampling.yaml \ ++trainer.max_epochs=100 \ - ++data.datamodule.datasets.train.subsample=1 \ - ++data.datamodule.datasets.val.subsample=1 \ - ++data.datamodule.datasets.test.subsample=1 \ + ++data.datamodule.datasets.train.subsample=${SUBSAMPLE} \ + ++data.datamodule.datasets.val.subsample=${SUBSAMPLE} \ + ++data.datamodule.datasets.test.subsample=${SUBSAMPLE} \ ++model.conditioner._target_=${CONDITIONER} \ ++model.sigma_distribution.sigma=${SIGMA} \ - ++data.datamodule.datasets.train.total_lag_time=${LAG_TIME} \ - ++model.arch.N_structures=${LAG_TIME} \ - ++logger.wandb.group="fake_enhanced_data_jul_11_sweep" \ - ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","train","enhanced_sampling","${CONDITIONER_NAME}","sigma_${SIGMA}","lag_${LAG_TIME}"] \ + ++data.datamodule.datasets.train.total_lag_time=${TOTAL_LAG_TIME} \ + ++data.datamodule.datasets.train.lag_subsample_rate=${LAG_SUBSAMPLE_RATE} \ + ++model.arch.N_structures=${TOTAL_LAG_TIME} \ + ++logger.wandb.group="fake_enhanced_data_jul17_sweep_lag_times" \ + ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","train","enhanced_sampling","${CONDITIONER_NAME}","sigma_${SIGMA}","lag_${TOTAL_LAG_TIME}","lag_subsample_${LAG_SUBSAMPLE_RATE}"] \ ++run_key=$RUN_KEY \ No newline at end of file diff --git a/src/jamun/cmdline/sample.py b/src/jamun/cmdline/sample.py index 4c5b070..096748e 100644 --- a/src/jamun/cmdline/sample.py +++ b/src/jamun/cmdline/sample.py @@ -88,7 +88,7 @@ def run(cfg): model = hydra.utils.instantiate(cfg.model) print(f'Checkpoint path at: {checkpoint_path}') init_datasets = hydra.utils.instantiate(cfg.init_datasets) - breakpoint() + # breakpoint() init_graphs = get_initial_graphs( init_datasets, num_init_samples_per_dataset=cfg.num_init_samples_per_dataset, @@ -100,7 +100,7 @@ def run(cfg): if seed := cfg.get("seed"): # During sampling, we want ranks to generate different chains. pl.seed_everything(seed + sampler.fabric.global_rank) - breakpoint() + # breakpoint() # Run test-time adapation, if specified. if finetuning_cfg := cfg.get("finetune_on_init"): num_finetuning_steps = finetuning_cfg.get("num_steps") @@ -143,7 +143,7 @@ def run(cfg): # Needed for submitit error output. # See https://github.com/facebookresearch/hydra/issues/2664 -@hydra.main(version_base=None, config_path="../hydra_config", config_name="sample") +@hydra.main(version_base=None, config_path="../hydra_config", config_name="sample_memory") def main(cfg): try: run(cfg) diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 7ee67ee..df9ffcf 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -18,6 +18,7 @@ from jamun.hydra import instantiate_dict_cfg # noqa: E402 from jamun.hydra.utils import format_resolver # noqa: E402 from jamun.utils import compute_average_squared_distance_from_datasets, dist_log, find_checkpoint # noqa: E402 +from jamun.utils._normalizations import normalization_factors # noqa: E402 dotenv.load_dotenv(".env", verbose=True) OmegaConf.register_new_resolver("format", format_resolver) @@ -58,13 +59,31 @@ def run(cfg): ) cfg.model.average_squared_distance = average_squared_distance + # Compute normalization factors for conditioner c_in parameter + if cfg.model.get("conditioner") and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.DenoisedConditioner": + if hasattr(cfg.model.sigma_distribution, "sigma"): + sigma = cfg.model.sigma_distribution.sigma + average_squared_distance = cfg.model.average_squared_distance + c_in, c_skip, c_out, c_noise = normalization_factors(sigma, average_squared_distance) + c_in_float = float(c_in) + + dist_log(f"Computing normalization factors for DenoisedConditioner with sigma={sigma}") + dist_log(f" average_squared_distance: {average_squared_distance}") + dist_log(f" c_in: {c_in_float}") + dist_log(f" c_skip: {c_skip}") + dist_log(f" c_out: {c_out}") + dist_log(f" c_noise: {c_noise}") + + cfg.model.conditioner.c_in = c_in_float + dist_log(f"Set cfg.model.conditioner.c_in to {c_in_float}") + # # do this for the sweep # if cfg.model.N_measurements_hidden is not None: # dist_log(f"Number of hidden measurements: {cfg.model.N_measurements_hidden}") # dist_log(f"Overwriting N_measurements...") # cfg.model.N_measurements = 100 // cfg.model.N_measurements_hidden # dist_log(f"New num of measurements: {cfg.model.N_measurements=}") - + # breakpoint() datamodule = hydra.utils.instantiate(cfg.data.datamodule) model = hydra.utils.instantiate(cfg.model) if matmul_prec := cfg.get("float32_matmul_precision"): diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index 45096b8..2c25943 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -218,7 +218,7 @@ def __init__( # Get lagged indices if lag parameters are provided if total_lag_time is not None and lag_subsample_rate is not None: - print(f"total_lag_time: {total_lag_time}, lag_subsample_rate: {lag_subsample_rate}") + # print(f"total_lag_time: {total_lag_time}, lag_subsample_rate: {lag_subsample_rate}") self.traj = self.traj[start_frame : start_frame + num_frames] # accommodate for start_frame and num_frames lagged_indices = get_subsampled_indices( self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate @@ -232,7 +232,7 @@ def __init__( self.traj = self.traj[subsampled_indices] # self.traj is permanently modified. else: # Regular subsampling without lag - print(f"subsample: {subsample}, regular subsampling") + # print(f"subsample: {subsample}, regular subsampling") self.traj = self.traj[start_frame : start_frame + num_frames : subsample] self.hidden_state = None self.lagged_indices = None diff --git a/src/jamun/data/_utils.py b/src/jamun/data/_utils.py index 4360a6d..f020ca9 100644 --- a/src/jamun/data/_utils.py +++ b/src/jamun/data/_utils.py @@ -107,15 +107,18 @@ def parse_datasets_from_directory( datasets = [] for code in tqdm(codes, desc="Creating datasets"): + # Use label_override for the dataset label, but keep original code for file lookups if label_override is not None: print(f"Label override: {label_override}") - code = str(label_override) + dataset_label = str(label_override) + else: + dataset_label = code dataset = dataset_class( root, traj_files=traj_files[code], pdb_file=pdb_files[code], - label=code, # TODO: add a label override here. + label=dataset_label, **dataset_kwargs, ) datasets.append(dataset) @@ -434,15 +437,18 @@ def parse_repeated_position_datasets_from_directory( datasets = [] for code in tqdm(codes, desc="Creating RepeatedPositionDatasets"): + # Use label_override for the dataset label, but keep original code for file lookups if label_override is not None: print(f"Label override: {label_override}") - code = str(label_override) + dataset_label = str(label_override) + else: + dataset_label = code dataset = RepeatedPositionDataset( root, traj_files=traj_files[code], pdb_file=pdb_files[code], - label=code, + label=dataset_label, **dataset_kwargs, ) datasets.append(dataset) diff --git a/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml b/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml new file mode 100644 index 0000000..2def981 --- /dev/null +++ b/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml @@ -0,0 +1,31 @@ +defaults: + - arch: e3conv_conditional.yaml + - optim: adam.yaml + - lr_scheduler_config: null + - _self_ + +max_radius: null +average_squared_distance: null +add_fixed_noise: false +add_fixed_ones: false +align_noisy_input_during_training: true +align_noisy_input_during_evaluation: true +mean_center: true +mirror_augmentation_rate: 0.0 +use_torch_compile: true +torch_compile_kwargs: + fullgraph: true + dynamic: true + mode: default + +conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + +# Multimeasurement specific parameters +multimeasurement: true +N_measurements: 1 +N_measurements_hidden: 1 +max_graphs_per_batch: null + +_target_: jamun.model.DenoiserMultimeasurement \ No newline at end of file diff --git a/src/jamun/model/__init__.py b/src/jamun/model/__init__.py index c31fabc..b237bc2 100644 --- a/src/jamun/model/__init__.py +++ b/src/jamun/model/__init__.py @@ -1,2 +1,5 @@ from .denoiser import Denoiser -from .energy import EnergyModel \ No newline at end of file +from .energy import EnergyModel +from .denoiser_multimeasurement import DenoiserMultimeasurement +from .denoiser_spiked import DenoiserSpiked +from .conditioners import ConditionerSpiked diff --git a/src/jamun/model/conditioner_usage_example.py b/src/jamun/model/conditioner_usage_example.py new file mode 100644 index 0000000..d00387b --- /dev/null +++ b/src/jamun/model/conditioner_usage_example.py @@ -0,0 +1,326 @@ +""" +Test for ConditionerSpiked with DenoiserSpiked using ALA_ALA data. + +This file demonstrates and tests the DenoiserSpiked model with ConditionerSpiked. +""" + +import functools +import os +import torch +import numpy as np +from pathlib import Path + +import jamun +from jamun.model import DenoiserSpiked +from jamun.model.conditioners import ConditionerSpiked +from jamun.model.arch import E3ConvConditional +import jamun.distributions +import jamun.data + + +def get_ala_ala_data(num_frames=20, total_lag_time=5): + """ + Load ALA_ALA data with specified parameters. + + Args: + num_frames: Number of frames to load per dataset + total_lag_time: Number of hidden states (total time lag) + + Returns: + List of datasets + """ + # Check if data path exists + data_path = os.getenv("JAMUN_DATA_PATH") + if data_path is None: + # Try common locations + possible_paths = [ + "/data/bucket/kleinhej/", + "/data2/sules/", + "/path/to/data/" + ] + for path in possible_paths: + ala_path = Path(path) / "capped_diamines/timewarp_splits/train" + if ala_path.exists(): + data_path = path + break + + if data_path is None: + raise ValueError("JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable.") + + print(f"Using data path: {data_path}") + root_path = f"{data_path}/capped_diamines/timewarp_splits/train" + + datasets = jamun.data.parse_datasets_from_directory( + root=root_path, + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + filter_codes=['ALA_ALA'], + as_iterable=False, + subsample=1, + total_lag_time=total_lag_time, + lag_subsample_rate=1, + num_frames=num_frames, + max_datasets=1 # Just use one dataset for testing + ) + + return datasets + + +def create_test_denoiser_spiked(total_lag_time=5): + """ + Create a simple DenoiserSpiked model for testing. + + Args: + total_lag_time: Number of structures for conditioning + + Returns: + DenoiserSpiked model + """ + import e3tools.nn + + # Note: The actual data has 4 hidden states, so we'll have 4 + 1 clean = 5 structures + actual_n_structures = 5 # 4 hidden states + 1 clean structure + + arch = functools.partial( + E3ConvConditional, + irreps_out="1x1e", + irreps_hidden="32x0e + 8x1e", # Smaller for testing + irreps_sh="1x0e + 1x1e", + n_layers=2, # Fewer layers for faster testing + edge_attr_dim=32, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=16, + residue_index_embedding_dim=8, + use_residue_information=True, + use_residue_sequence_index=False, + N_structures=actual_n_structures, # Match actual data structure count + hidden_layer_factory=functools.partial( + e3tools.nn.ConvBlock, + conv=e3tools.nn.Conv, + ), + output_head_factory=functools.partial( + e3tools.nn.EquivariantMLP, + irreps_hidden_list=["32x0e + 8x1e"] + ), + ) + + conditioner = ConditionerSpiked(N_structures=actual_n_structures) + + denoiser = DenoiserSpiked( + arch=arch, + optim=functools.partial(torch.optim.Adam, lr=1e-3), + sigma_distribution=jamun.distributions.ConstantSigma(sigma=0.04), + max_radius=1000.0, # Large radius for testing + average_squared_distance=10.0, + add_fixed_noise=False, + add_fixed_ones=False, + align_noisy_input_during_training=True, + align_noisy_input_during_evaluation=True, + mean_center=True, + mirror_augmentation_rate=0.0, + conditioner=conditioner, + ) + + return denoiser + + +def test_noise_and_denoise(): + """ + Test the noise_and_denoise method with ALA_ALA data. + """ + print("=" * 60) + print("Testing DenoiserSpiked with ConditionerSpiked on ALA_ALA data") + print("=" * 60) + + # Load data + try: + total_lag_time = 5 + datasets = get_ala_ala_data(num_frames=10, total_lag_time=total_lag_time) + print(f"āœ… Successfully loaded {len(datasets)} datasets") + + dataset = datasets[0] + print(f" Dataset label: {dataset.label()}") + print(f" Dataset length: {len(dataset)}") + + # Get a sample + sample = dataset[0] + print(f" Sample positions shape: {sample.pos.shape}") + print(f" Sample hidden states: {len(sample.hidden_state) if hasattr(sample, 'hidden_state') and sample.hidden_state else 0}") + if hasattr(sample, 'hidden_state') and sample.hidden_state: + for i, h in enumerate(sample.hidden_state): + print(f" Hidden state {i} shape: {h.shape}") + + except Exception as e: + print(f"āŒ Failed to load data: {e}") + return False + + # Create model + try: + denoiser = create_test_denoiser_spiked(total_lag_time=total_lag_time) + print(f"āœ… Successfully created DenoiserSpiked model") + print(f" Conditioner: {type(denoiser.conditioning_module).__name__}") + + except Exception as e: + print(f"āŒ Failed to create model: {e}") + return False + + # Test noise_and_denoise + try: + print("\n" + "-" * 40) + print("Testing noise_and_denoise method...") + print("-" * 40) + + # Convert to batch for testing + import torch_geometric.data + batch = torch_geometric.data.Batch.from_data_list([sample]) + print(f" Batch positions shape: {batch.pos.shape}") + print(f" Batch num_graphs: {batch.num_graphs}") + + # Set model to eval mode + denoiser.eval() + + # Test with different sigma values + sigma_values = [0.01, 0.04, 0.1] + + for sigma in sigma_values: + print(f"\n Testing with sigma = {sigma}") + + # Run noise_and_denoise + with torch.no_grad(): + x_target, xhat, y_noisy = denoiser.noise_and_denoise( + batch, + sigma=sigma, + align_noisy_input=True + ) + + print(f" āœ… noise_and_denoise completed successfully") + print(f" Target shape: {x_target.pos.shape}") + print(f" Prediction shape: {xhat.pos.shape}") + print(f" Noisy input shape: {y_noisy.pos.shape}") + + # Check that shapes match + assert x_target.pos.shape == xhat.pos.shape == y_noisy.pos.shape + + # Test conditioning + print(f" Testing conditioner...") + print(f" y_noisy has hidden_state: {hasattr(y_noisy, 'hidden_state') and y_noisy.hidden_state is not None}") + if hasattr(y_noisy, 'hidden_state') and y_noisy.hidden_state is not None: + print(f" y_noisy hidden_state count: {len(y_noisy.hidden_state)}") + print(f" x_target is not None: {x_target is not None}") + if x_target is not None: + print(f" x_target.pos shape: {x_target.pos.shape}") + print(f" x_target has hidden_state: {hasattr(x_target, 'hidden_state') and x_target.hidden_state is not None}") + + conditioned_structures = denoiser.conditioner(y_noisy, x_target) + print(f" Conditioned structures count: {len(conditioned_structures)}") + + for i, struct in enumerate(conditioned_structures): + print(f" Structure {i} shape: {struct.shape}") + + # Verify that the last structure is the clean structure + if len(conditioned_structures) > 0: + last_structure = conditioned_structures[-1] + clean_structure = x_target.pos + if torch.allclose(last_structure, clean_structure, atol=1e-6): + print(f" āœ… Last conditioned structure matches x_clean.pos") + else: + print(f" āŒ Last conditioned structure does NOT match x_clean.pos") + print(f" Max difference: {torch.max(torch.abs(last_structure - clean_structure)).item():.8f}") + + # Calculate some basic metrics + noise_level = torch.mean(torch.norm(y_noisy.pos - x_target.pos, dim=-1)) + prediction_error = torch.mean(torch.norm(xhat.pos - x_target.pos, dim=-1)) + print(f" Average noise level: {noise_level:.4f}") + print(f" Average prediction error: {prediction_error:.4f}") + + print(f"\nāœ… All noise_and_denoise tests passed!") + return True + + except Exception as e: + print(f"āŒ Failed during noise_and_denoise testing: {e}") + import traceback + traceback.print_exc() + return False + + +def test_conditioning_shapes(): + """ + Test that conditioning produces expected shapes. + """ + print("\n" + "=" * 60) + print("Testing ConditionerSpiked shape outputs") + print("=" * 60) + + # Create dummy data + N_atoms = 22 # ALA_ALA has 22 atoms + N_structures = 5 + + # Create fake batch + pos = torch.randn(N_atoms, 3) + hidden_states = [torch.randn(N_atoms, 3) for _ in range(N_structures - 2)] # -2 for current pos and clean pos + + # Create fake torch_geometric batch + import torch_geometric.data + y = torch_geometric.data.Data(pos=pos, hidden_state=hidden_states) + x_clean = torch_geometric.data.Data(pos=torch.randn(N_atoms, 3)) + + # Test conditioner + conditioner = ConditionerSpiked(N_structures=N_structures) + conditioned_structures = conditioner.forward(y, x_clean) + + print(f"Input shapes:") + print(f" y.pos: {y.pos.shape}") + print(f" y.hidden_state: {[h.shape for h in y.hidden_state]}") + print(f" x_clean.pos: {x_clean.pos.shape}") + + print(f"\nConditioned structures:") + for i, struct in enumerate(conditioned_structures): + print(f" Structure {i}: {struct.shape}") + + # Test concatenation (like in the model) + concatenated = torch.cat(conditioned_structures, dim=-1) + print(f"\nConcatenated shape: {concatenated.shape}") + expected_dim = len(conditioned_structures) * 3 # Each structure has 3D coordinates + print(f"Expected last dimension: {expected_dim}") + + # Verify that the last structure is the clean structure + if len(conditioned_structures) > 0: + last_structure = conditioned_structures[-1] + clean_structure = x_clean.pos + if torch.allclose(last_structure, clean_structure, atol=1e-6): + print(f"āœ… Last conditioned structure matches x_clean.pos") + else: + print(f"āŒ Last conditioned structure does NOT match x_clean.pos") + print(f" Max difference: {torch.max(torch.abs(last_structure - clean_structure)).item():.8f}") + assert False, "Last conditioned structure should match x_clean.pos" + + assert concatenated.shape == (N_atoms, expected_dim) + print(f"āœ… Shape test passed!") + + +if __name__ == "__main__": + # Run tests + try: + # Test data loading and noise_and_denoise + success = test_noise_and_denoise() + + # Test conditioning shapes + test_conditioning_shapes() + + if success: + print("\n" + "=" * 60) + print("šŸŽ‰ ALL TESTS PASSED! šŸŽ‰") + print("DenoiserSpiked with ConditionerSpiked is working correctly!") + print("=" * 60) + else: + print("\n" + "=" * 60) + print("āŒ SOME TESTS FAILED") + print("=" * 60) + + except KeyboardInterrupt: + print("\nāš ļø Tests interrupted by user") + except Exception as e: + print(f"\nāŒ Unexpected error: {e}") + import traceback + traceback.print_exc() \ No newline at end of file diff --git a/src/jamun/model/conditioners/__init__.py b/src/jamun/model/conditioners/__init__.py index 56292f6..c43d2ab 100644 --- a/src/jamun/model/conditioners/__init__.py +++ b/src/jamun/model/conditioners/__init__.py @@ -1,2 +1,2 @@ -from .conditioners import Conditioner, PositionConditioner, SelfConditioner, MeanConditioner +from .conditioners import Conditioner, PositionConditioner, SelfConditioner, MeanConditioner, DenoisedConditioner, ConditionerSpiked diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 2dc2445..95e078f 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -1,14 +1,19 @@ import logging -from typing import Callable, Dict, Optional, Tuple, Union +from typing import Callable, Dict, Tuple, Union import lightning.pytorch as pl import numpy as np import torch import torch_geometric -from e3tools import radius_graph, scatter +import e3nn +from e3tools import radius_graph +from jamun.model.denoiser_conditional import Denoiser +# Fix e3nn optimization for avoiding script issues +e3nn.set_optimization_defaults(jit_script_fx=False) from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm +from jamun.utils.checkpoint import find_checkpoint class Conditioner(pl.LightningModule): """ @@ -68,8 +73,163 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: # Shape: (T+1, N, 3) -> (N, 3) where T is number of hidden states stacked_positions = torch.stack(all_positions, dim=0) # (T+1, N, 3) mean_positions = torch.mean(stacked_positions, dim=0) # (N, 3) - + # mean center the mean positions + dummy_graph = y.clone() + dummy_graph.pos = mean_positions + # mean center the mean positions + mean_positions = mean_center(dummy_graph).pos + # align the mean positions to the current positions + aligned_mean_positions = kabsch_algorithm(mean_positions, y.pos, y.batch, y.num_graphs) + # Return the mean repeated N_structures times - conditioned_structures = [mean_positions for _ in range(self.N_structures)] + conditioned_structures = [aligned_mean_positions for _ in range(self.N_structures)] + + return conditioned_structures + +class DenoisedConditioner(pl.LightningModule): + """ + Conditioner that uses a pretrained denoiser to denoise hidden states. + + Takes hidden states, unscales them using c_in, denoises each structure, + then recenters and aligns them to the current noisy positions. + """ + def __init__(self, N_structures: int, pretrained_model_path: str, c_in: float, **kwargs): + super().__init__() + self.N_structures = N_structures + self.c_in = c_in + self.pretrained_model_path = pretrained_model_path + + # Load the pretrained denoiser + py_logger = logging.getLogger("jamun") + py_logger.info(f"Loading pretrained denoiser from wandb run: {pretrained_model_path}") + + # Find the checkpoint for the wandb run + checkpoint_path = find_checkpoint( + wandb_train_run_path=pretrained_model_path, + checkpoint_type="best_so_far" + ) + + # Load the denoiser from checkpoint + self.pretrained_denoiser = Denoiser.load_from_checkpoint(checkpoint_path, strict=False) + self.pretrained_denoiser.eval() # Set to evaluation mode + + # Freeze the pretrained model parameters + for param in self.pretrained_denoiser.parameters(): + param.requires_grad = False + + # Extract sigma from the pretrained denoiser + self.denoiser_sigma = self._extract_sigma_from_denoiser() + py_logger.info(f"Extracted sigma from pretrained denoiser: {self.denoiser_sigma}") + py_logger.info(f"Successfully loaded pretrained denoiser with c_in={c_in}") + + def _extract_sigma_from_denoiser(self) -> float: + """Extract sigma value from the pretrained denoiser's sigma distribution.""" + sigma_distribution = self.pretrained_denoiser.sigma_distribution + + # Handle different types of sigma distributions + if hasattr(sigma_distribution, 'sigma'): + # For ConstantSigma distribution + return float(sigma_distribution.sigma) + elif hasattr(sigma_distribution, 'mean'): + # For other distributions that might have a mean + return float(sigma_distribution.mean) + else: + # Fallback - sample from the distribution + sample = sigma_distribution.sample() + return float(sample) + + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + """ + Forward pass that denoises hidden states and returns conditioned structures. + + Args: + y: Batch containing current positions and hidden states + + Returns: + List of tensors: [y.pos, *denoised_hidden_states] + """ + # Use the sigma from the pretrained denoiser + sigma_to_use = self.denoiser_sigma + + conditioned_structures = [y.pos] # Start with current position + + # Check if we have hidden states to process + if not hasattr(y, "hidden_state") or y.hidden_state is None: + # If no hidden states, just repeat current position + conditioned_structures.extend([y.pos for _ in range(self.N_structures - 1)]) + return conditioned_structures + + # # Move pretrained denoiser to same device as input + # device = y.pos.device + # self.pretrained_denoiser = self.pretrained_denoiser.to(device) + + # Process each hidden state + for i, hidden_positions in enumerate(y.hidden_state): + # Unscale the hidden state positions + unscaled_positions = hidden_positions / self.c_in + + # Create a batch for denoising + denoising_batch = y.clone() + denoising_batch.pos = unscaled_positions + + # Remove hidden states from the denoising batch to avoid recursion + if hasattr(denoising_batch, "hidden_state"): + delattr(denoising_batch, "hidden_state") + + # Denoise the unscaled positions using the denoiser's sigma + with torch.no_grad(): + denoised_batch = self.pretrained_denoiser.xhat(denoising_batch, sigma_to_use) + denoised_positions = denoised_batch.pos + + # Align the denoised positions to the current noisy positions + aligned_positions = kabsch_algorithm(denoised_positions, y.pos, y.batch, y.num_graphs) + + conditioned_structures.append(aligned_positions) + + # Break if we've processed enough structures + if len(conditioned_structures) >= self.N_structures: + break + + # If we don't have enough hidden states, pad with the last denoised structure + while len(conditioned_structures) < self.N_structures: + conditioned_structures.append(conditioned_structures[-1]) + + return conditioned_structures + + +class ConditionerSpiked(Conditioner): + """ + A conditioner that concatenates hidden states with the clean structure. + + The conditioning order is: + 1. Hidden states (y.hidden_state) - if present + 2. Clean structure positions (x_clean.pos) - if provided at the end + """ + + def __init__(self, N_structures: int, **kwargs): + super().__init__(N_structures, **kwargs) + + def forward(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: + """ + Create conditioning structures by concatenating hidden states with clean structure. + + Args: + y: The noisy sample batch containing positions and hidden states + x_clean: The clean sample batch containing ground truth positions + + Returns: + List of tensors to be concatenated for conditioning + """ + conditioned_structures = [y.pos] + + # Add hidden states if they exist + if hasattr(y, "hidden_state") and y.hidden_state is not None: + for hidden_pos in y.hidden_state: + conditioned_structures.append(hidden_pos) + + # Add clean structure positions at the end if provided + if x_clean is not None: + conditioned_structures.pop(-1) + conditioned_structures.append(x_clean.pos) - return conditioned_structures \ No newline at end of file + return conditioned_structures \ No newline at end of file diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index a6ab019..e25e352 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -280,7 +280,7 @@ def xhat_normalized( with torch.cuda.nvtx.range("conditioning"): conditioned_structures = self.conditioner(y_scaled) - + # print(f"Conditioner is working, number of conditioned structures: {len(conditioned_structures)}") with torch.cuda.nvtx.range("g"): g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), topology=y_scaled, \ c_noise=c_noise, effective_radial_cutoff=radial_cutoff) diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py index 1b8ec7f..4702f38 100644 --- a/src/jamun/model/denoiser_multimeasurement.py +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -205,36 +205,49 @@ def add_noise( y = x.clone() if self.add_fixed_ones: noise = torch.ones_like(x.pos) - hidden_noise = [ - torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) - ] + if hasattr(x, "hidden_state") and x.hidden_state is not None: + hidden_noise = [ + torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) + ] + else: + hidden_noise = [] elif self.add_fixed_noise: torch.manual_seed(0) num_batches = x.batch.max().item() + 1 if len(x.pos.shape) == 2: num_nodes_per_batch = x.pos.shape[0] // num_batches noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) - hidden_noise = [ - torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat( - num_batches, 1 - ) - for i in range(len(x.hidden_state)) - ] + if hasattr(x, "hidden_state") and x.hidden_state is not None: + hidden_noise = [ + torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat( + num_batches, 1 + ) + for i in range(len(x.hidden_state)) + ] + else: + hidden_noise = [] if len(x.pos.shape) == 3: num_nodes_per_batch = x.pos.shape[1] noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) - hidden_noise = [ - torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) - for i in range(len(x.hidden_state)) - ] + if hasattr(x, "hidden_state") and x.hidden_state is not None: + hidden_noise = [ + torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) + for i in range(len(x.hidden_state)) + ] + else: + hidden_noise = [] else: noise = torch.randn_like(x.pos) - hidden_noise = [ - torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) - ] + if hasattr(x, "hidden_state") and x.hidden_state is not None: + hidden_noise = [ + torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) + ] + else: + hidden_noise = [] y.pos = x.pos + sigma * noise - for i in range(len(y.hidden_state)): - y.hidden_state[i] = x.hidden_state[i] + hidden_noise[i] + if hasattr(y, "hidden_state") and y.hidden_state is not None and hidden_noise: + for i in range(len(y.hidden_state)): + y.hidden_state[i] = x.hidden_state[i] + sigma*hidden_noise[i] if torch.rand(()) < self.mirror_augmentation_rate: y.pos = -y.pos return y @@ -396,10 +409,11 @@ def xhat_normalized( with torch.cuda.nvtx.range("scale_y"): y_scaled = y.clone() y_scaled.pos = y.pos * c_in - scaled_hidden_state = [] - for positions in y_scaled.hidden_state: - scaled_hidden_state.append(positions * c_in) - y_scaled.hidden_state = scaled_hidden_state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + scaled_hidden_state = [] + for positions in y.hidden_state: + scaled_hidden_state.append(positions * c_in) + y_scaled.hidden_state = scaled_hidden_state with torch.cuda.nvtx.range("clone_y"): xhat = y.clone() @@ -416,7 +430,7 @@ def xhat_normalized( ) xhat.pos = c_skip * y.pos + c_out * g_pred - if hasattr(y, "hidden_state") and xhat.hidden_state is not None: + if hasattr(y, "hidden_state") and y.hidden_state is not None: xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] # the hidden state updates! return xhat @@ -452,6 +466,7 @@ def noise_and_denoise( if self.mean_center: # Operate on a clone to avoid side effects on the original batch object. x_processed = mean_center(x) + x_processed = self._mean_center_hidden_states(x_processed) else: x_processed = x @@ -482,6 +497,7 @@ def noise_and_denoise( if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): y = mean_center(y) + y = self._mean_center_hidden_states(y) # Aligning each batch. if align_noisy_input: @@ -605,8 +621,8 @@ def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): all_aux = [] opt = self.optimizers() if stage == "train" else None - print(f"Processing {num_chunks} chunks of size {chunk_size} for {stage}...") - for i in tqdm(range(num_chunks)): + # print(f"Processing {num_chunks} chunks of size {chunk_size} for {stage}...") + for i in range(num_chunks): start_index = i * chunk_size end_index = min(start_index + chunk_size, len(y_list)) @@ -629,16 +645,11 @@ def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): aux["loss"] = loss for key in aux: aux[key] = aux[key].mean() - self.log( - f"train/{key}", - aux[key], - prog_bar=False, - batch_size=batch.num_graphs, - sync_dist=False, - ) if stage == "train": + # Scale loss by number of chunks to match automatic optimization gradients + scaled_loss = aux["loss"] / num_chunks opt.zero_grad() - self.manual_backward(aux["loss"]) + self.manual_backward(scaled_loss) opt.step() all_aux.append(aux) @@ -650,11 +661,13 @@ def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): log_opts = { "prog_bar": (stage == "val" and "scaled_rmsd" in avg_aux), "batch_size": len(y_list), + "sync_dist": (stage == "val"), # Only sync for validation } + + # Ensure training metrics are always logged if stage == "train": - log_opts["sync_dist"] = True - else: - log_opts["sync_dist"] = True + log_opts["on_step"] = True + log_opts["on_epoch"] = True for key, value in avg_aux.items(): self.log(f"{stage}/{key}", value, **log_opts) @@ -672,9 +685,9 @@ def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): """Called during validation.""" if self.automatic_optimization: - self._automatic_step(batch, "val") + return self._automatic_step(batch, "val") else: - self._manual_step(batch, "val") + return self._manual_step(batch, "val") def configure_optimizers(self): """Set up the optimizer and learning rate scheduler.""" diff --git a/src/jamun/model/denoiser_spiked.py b/src/jamun/model/denoiser_spiked.py new file mode 100644 index 0000000..13f44c8 --- /dev/null +++ b/src/jamun/model/denoiser_spiked.py @@ -0,0 +1,458 @@ +import logging +from typing import Callable, Dict, Optional, Tuple, Union + +import lightning.pytorch as pl +import numpy as np +import torch +import torch_geometric +from e3tools import radius_graph, scatter +from tqdm import tqdm + +from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils.align import kabsch_algorithm + + +class DenoiserSpiked(pl.LightningModule): + """The main denoiser model with conditional architecture that includes clean sample conditioning.""" + + def __init__( + self, + arch: Callable[..., torch.nn.Module], + optim: Callable[..., torch.optim.Optimizer], + sigma_distribution: torch.distributions.Distribution, + max_radius: float, + average_squared_distance: float, + add_fixed_noise: bool, + add_fixed_ones: bool, + align_noisy_input_during_training: bool, + align_noisy_input_during_evaluation: bool, + mean_center: bool, + mirror_augmentation_rate: float, + bond_loss_coefficient: float = 1.0, + normalization_type: Optional[str] = "JAMUN", + sigma_data: Optional[float] = None, # Only used if normalization_type is "EDM" + lr_scheduler_config: Optional[Dict] = None, + use_torch_compile: bool = True, + torch_compile_kwargs: Optional[Dict] = None, + conditioner: Callable[..., list[torch.Tensor]] = None, + ): + super().__init__() + self.save_hyperparameters(logger=False) + + self.g = arch() + if use_torch_compile: + if torch_compile_kwargs is None: + torch_compile_kwargs = {} + + self.g = torch.compile(self.g, **torch_compile_kwargs) + + py_logger = logging.getLogger("jamun") + py_logger.info(self.g) + + self.optim_factory = optim + self.lr_scheduler_config = lr_scheduler_config + self.sigma_distribution = sigma_distribution + self.max_radius = max_radius + + self.add_fixed_noise = add_fixed_noise + self.add_fixed_ones = add_fixed_ones + if self.add_fixed_noise and self.add_fixed_ones: + raise ValueError("Can't add fixed noise and fixed ones at the same time") + if self.add_fixed_noise: + py_logger.info("Adding fixed noise") + if self.add_fixed_ones: + py_logger.info("Adding fixed ones") + + self.average_squared_distance = average_squared_distance + py_logger.info(f"Average squared distance = {self.average_squared_distance}") + + self.align_noisy_input_during_training = align_noisy_input_during_training + if self.align_noisy_input_during_training: + py_logger.info("Aligning noisy input during training.") + else: + py_logger.info("Not aligning noisy input during training.") + + self.align_noisy_input_during_evaluation = align_noisy_input_during_evaluation + if self.align_noisy_input_during_evaluation: + py_logger.info("Aligning noisy input during evaluation.") + else: + py_logger.info("Not aligning noisy input during evaluation.") + + self.mean_center = mean_center + if self.mean_center: + py_logger.info("Mean centering input and output.") + else: + py_logger.info("Not mean centering input and output.") + + self.mirror_augmentation_rate = mirror_augmentation_rate + py_logger.info(f"Mirror augmentation rate: {self.mirror_augmentation_rate}") + + self.normalization_type = normalization_type + if self.normalization_type is not None: + py_logger.info(f"Normalization type: {self.normalization_type}") + else: + py_logger.info("No normalization") + + self.sigma_data = sigma_data + if self.normalization_type == "EDM" and self.sigma_data is None: + raise ValueError("sigma_data must be provided when normalization_type is 'EDM'") + elif self.normalization_type != "EDM" and self.sigma_data is not None: + raise ValueError("sigma_data can only be used when normalization_type is 'EDM'") + + self.bond_loss_coefficient = bond_loss_coefficient + self.conditioning_module = conditioner + if self.conditioning_module is not None and not callable(self.conditioning_module): + raise ValueError("Conditioner must be a callable or None") + py_logger.info(f"Conditioner: {self.conditioning_module}") + + def conditioner_default(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: + conditioned_structures = [y.pos] # Return complete list starting with current position + if x_clean is not None: + conditioned_structures.append(x_clean.pos) # Add clean sample positions + return conditioned_structures + + def conditioner(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: + if self.conditioning_module is None: + return self.conditioner_default(y, x_clean) + elif callable(self.conditioning_module): + return self.conditioning_module(y, x_clean) + else: + raise ValueError("Conditioner must be a callable or None") + + def _align_A_to_B_batched_with_hidden_states( + self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch + ) -> torch_geometric.data.Batch: + """Aligns each graph of A to the corresponding graph in B, including hidden states.""" + A_aligned = A.clone() + + # Align positions + A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) + + # Align hidden states + if hasattr(A, "hidden_state") and A.hidden_state is not None: + A_aligned.hidden_state = [] + for i in range(len(A.hidden_state)): + A_aligned.hidden_state.append(kabsch_algorithm( + A.hidden_state[i], B.pos, A.batch, A.num_graphs + )) + return A_aligned + + def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): + if hasattr(data, "hidden_state") and data.hidden_state is not None: + for i in range(len(data.hidden_state)): + mean = scatter(data.hidden_state[i], data.batch, dim=0, reduce="mean") + data.hidden_state[i] = data.hidden_state[i] - mean[data.batch] + return data + + def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: + # pos [B, ...] + sigma = unsqueeze_trailing(sigma, x.pos.ndim) + + y = x.clone() + if self.add_fixed_ones: + noise = torch.ones_like(x.pos) + hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] + elif self.add_fixed_noise: + torch.manual_seed(0) + num_batches = x.batch.max().item() + 1 + if len(x.pos.shape) == 2: + num_nodes_per_batch = x.pos.shape[0] // num_batches + noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) + hidden_noise = [torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat(num_batches, 1) for i in range(len(x.hidden_state))] + if len(x.pos.shape) == 3: + num_nodes_per_batch = x.pos.shape[1] + noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) + hidden_noise = [torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state))] + else: + noise = torch.randn_like(x.pos) + hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] + y.pos = x.pos + sigma * noise + for i in range(len(y.hidden_state)): + y.hidden_state[i] = x.hidden_state[i] + sigma * hidden_noise[i] + if torch.rand(()) < self.mirror_augmentation_rate: + y.pos = -y.pos + return y + + def score(self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor], x_clean: torch_geometric.data.Batch) -> torch_geometric.data.Batch: + """Compute the score function.""" + sigma = torch.as_tensor(sigma).to(y.pos) + return (self.xhat(y, sigma, x_clean).pos - y.pos) / (unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2) + + def normalization_factors(self, sigma: float, D: int = 3) -> Tuple[float, float, float, float]: + """Normalization factors for the input and output.""" + sigma = torch.as_tensor(sigma) + + if self.normalization_type is None: + return 1.0, 0.0, 1.0, sigma + + if self.normalization_type == "EDM": + c_skip = (self.sigma_data**2) / (sigma**2 + self.sigma_data**2) + c_out = sigma * self.sigma_data / torch.sqrt(self.sigma_data**2 + sigma**2) + c_in = 1 / torch.sqrt(sigma**2 + self.sigma_data**2) + c_noise = torch.log(sigma / self.sigma_data) * 0.25 + return c_in, c_skip, c_out, c_noise + + if self.normalization_type == "JAMUN": + A = torch.as_tensor(self.average_squared_distance) + B = torch.as_tensor(2 * D * sigma**2) + + c_in = 1.0 / torch.sqrt(A + B) + c_skip = A / (A + B) + c_out = torch.sqrt((A * B) / (A + B)) + c_noise = torch.log(sigma) / 4 + return c_in, c_skip, c_out, c_noise + + raise ValueError(f"Unknown normalization type: {self.normalization_type}") + + def loss_weight(self, sigma: float, D: int = 3) -> float: + """Loss weight for this graph.""" + _, _, c_out, _ = self.normalization_factors(sigma, D) + return 1 / (c_out**2) + + def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Tensor: + """Compute the effective radial cutoff for the noise level.""" + return torch.sqrt((self.max_radius**2) + 6 * (sigma**2)) + + def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torch_geometric.data.Batch: + """Add edges to the graph based on the effective radial cutoff.""" + if "batch" in y: + batch = y["batch"] + else: + batch = torch.zeros(y.num_nodes, dtype=torch.long, device=self.device) + + # Our dataloader already adds the bonded edges. + bonded_edge_index = y.edge_index + + with torch.cuda.nvtx.range("radial_graph"): + radial_edge_index = radius_graph(y.pos, radial_cutoff, batch) + + with torch.cuda.nvtx.range("concatenate_edges"): + edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) + if bonded_edge_index.numel() == 0: + bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device) + else: + bond_mask = torch.cat( + ( + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device), + torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=self.device), + ), + dim=0, + ) + + y.edge_index = edge_index + y.bond_mask = bond_mask + return y + + def xhat_normalized( + self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor], x_clean: torch_geometric.data.Batch + ) -> torch_geometric.data.Batch: + """Compute the denoised prediction using the normalization factors from JAMUN.""" + sigma = torch.as_tensor(sigma).to(y.pos) + D = y.pos.shape[-1] + + # Compute the normalization factors. + with torch.cuda.nvtx.range("normalization_factors"): + c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) + radial_cutoff = self.effective_radial_cutoff(sigma) / c_in + + # Adjust dimensions. + c_in = unsqueeze_trailing(c_in, y.pos.ndim - 1) + c_skip = unsqueeze_trailing(c_skip, y.pos.ndim - 1) + c_out = unsqueeze_trailing(c_out, y.pos.ndim - 1) + c_noise = c_noise.unsqueeze(0) + + # Add edges to the graph. + with torch.cuda.nvtx.range("add_edges"): + y = self.add_edges(y, radial_cutoff) + + with torch.cuda.nvtx.range("scale_y"): + y_scaled = y.clone() + y_scaled.pos = y.pos * c_in + # Manually copy hidden state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + y_scaled.hidden_state = [] + for positions in y.hidden_state: + y_scaled.hidden_state.append(positions * c_in) + + # Keep clean sample unscaled + with torch.cuda.nvtx.range("clone_y"): + xhat = y.clone() + # Manually copy hidden state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [h.clone() for h in y.hidden_state] + + with torch.cuda.nvtx.range("conditioning"): + conditioned_structures = self.conditioner(y_scaled, x_clean) + # print(f"Conditioner is working, number of conditioned structures: {len(conditioned_structures)}") + with torch.cuda.nvtx.range("g"): + g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), topology=y_scaled, \ + c_noise=c_noise, effective_radial_cutoff=radial_cutoff) + + xhat.pos = c_skip * y.pos + c_out * g_pred + if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] + return xhat + + def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor], x_clean: torch_geometric.data.Batch): + """Compute the denoised prediction.""" + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_y"): + y = mean_center(y) + y = self._mean_center_hidden_states(y) + with torch.cuda.nvtx.range("mean_center_x_clean"): + x_clean = mean_center(x_clean) + x_clean = self._mean_center_hidden_states(x_clean) + + with torch.cuda.nvtx.range("xhat_normalized"): + xhat = self.xhat_normalized(y, sigma, x_clean) + + # Mean center the prediction. + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_xhat"): + xhat = mean_center(xhat) + + return xhat + + def noise_and_denoise( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: + """ + Add noise to the input and denoise it. + Returns the target for the loss, the prediction, and the noisy input. + """ + with torch.no_grad(): + if self.mean_center: + # Operate on a clone to avoid side effects on the original batch object. + x_processed = mean_center(x) + x_processed = self._mean_center_hidden_states(x_processed) + else: + x_processed = x + + sigma = torch.as_tensor(sigma).to(x_processed.pos) + + with torch.cuda.nvtx.range("add_noise"): + y = self.add_noise(x_processed, sigma) + x_target = x_processed.clone() + # Manually copy hidden state + if hasattr(x_processed, "hidden_state") and x_processed.hidden_state is not None: + x_target.hidden_state = [h.clone() for h in x_processed.hidden_state] + + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_y"): + y = mean_center(y) + y = self._mean_center_hidden_states(y) + + # Aligning each batch. + if align_noisy_input: + with torch.cuda.nvtx.range("align_A_to_B_batched"): + y = self._align_A_to_B_batched_with_hidden_states(y, x_target) + + # KEY CHANGE: Pass both noisy sample (y) AND clean sample (x_target) to xhat + with torch.cuda.nvtx.range("xhat"): + xhat = self.xhat(y, sigma, x_target) + + return x_target, xhat, y + + def compute_loss( + self, + x: torch_geometric.data.Batch, + xhat: torch.Tensor, + sigma: Union[float, torch.Tensor], + ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + """Compute the loss.""" + if self.mean_center: + with torch.cuda.nvtx.range("mean_center_x"): + x = mean_center(x) + + D = xhat.pos.shape[-1] + + # Compute the raw loss. + with torch.cuda.nvtx.range("raw_coordinate_loss"): + raw_coordinate_loss = (xhat.pos - x.pos).pow(2).sum(dim=-1) + + # Take the mean over each graph. + with torch.cuda.nvtx.range("mean_over_graphs"): + mse = scatter(raw_coordinate_loss, x.batch, dim=0, dim_size=x.num_graphs, reduce="mean") + + # Compute the scaled RMSD. + with torch.cuda.nvtx.range("scaled_rmsd"): + rmsd = torch.sqrt(mse) + scaled_rmsd = rmsd / (sigma * np.sqrt(D)) + + # Account for the loss weight across graphs and noise levels. + with torch.cuda.nvtx.range("loss_weight"): + loss = mse * x.loss_weight + loss = loss * self.loss_weight(sigma, D) + + return loss, { + "mse": mse, + "rmsd": rmsd, + "scaled_rmsd": scaled_rmsd, + } + + def noise_and_compute_loss( + self, + x: torch_geometric.data.Batch, + sigma: Union[float, torch.Tensor], + align_noisy_input: bool, + ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + """Add noise to the input and compute the loss.""" + x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) + return self.compute_loss(x_target, xhat, sigma) + + def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): + """The standard step for automatic optimization.""" + align_noisy_input = self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation + sigma = self.sigma_distribution.sample().to(self.device) + + loss, aux = self.noise_and_compute_loss( + batch, + sigma, + align_noisy_input=align_noisy_input, + ) + + # Average the loss and other metrics over all graphs. + with torch.cuda.nvtx.range("mean_over_graphs"): + aux["loss"] = loss + for key in aux: + aux[key] = aux[key].mean() + if stage == "train": + self.log(f"train/{key}", aux[key], prog_bar=False, batch_size=batch.num_graphs, sync_dist=False) + elif stage == "val": + self.log( + f"val/{key}", aux[key], prog_bar=(key == "scaled_rmsd"), batch_size=batch.num_graphs, sync_dist=True + ) + else: + continue + + + return { + "sigma": sigma, + **aux, + } + + + def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during training.""" + return self._automatic_step(batch, "train") + + def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): + """Called during validation.""" + self._automatic_step(batch, "val") + + def configure_optimizers(self): + """Set up the optimizer and learning rate scheduler.""" + optimizer = self.optim_factory(params=self.parameters()) + + out = {"optimizer": optimizer} + if self.lr_scheduler_config: + scheduler = self.lr_scheduler_config.pop("scheduler") + out["lr_scheduler"] = { + "scheduler": scheduler(optimizer), + **self.lr_scheduler_config, + } + + return out \ No newline at end of file diff --git a/src/jamun/utils/_normalizations.py b/src/jamun/utils/_normalizations.py new file mode 100644 index 0000000..43fc66a --- /dev/null +++ b/src/jamun/utils/_normalizations.py @@ -0,0 +1,53 @@ +""" +Normalization utilities for jamun models. +""" + +from typing import Tuple +import torch + + +def normalization_factors( + sigma: float, + average_squared_distance: float, + normalization_type: str = "JAMUN", + sigma_data: float = None, + D: int = 3 +) -> Tuple[float, float, float, float]: + """ + Compute normalization factors for the input and output. + + Args: + sigma: Noise level + average_squared_distance: Average squared distance from the dataset + normalization_type: Type of normalization ("JAMUN", "EDM", or None) + sigma_data: Sigma data parameter (only used for EDM normalization) + D: Dimensionality (default: 3) + + Returns: + Tuple of (c_in, c_skip, c_out, c_noise) normalization factors + """ + sigma = torch.as_tensor(sigma) + + if normalization_type is None: + return 1.0, 0.0, 1.0, sigma + + if normalization_type == "EDM": + if sigma_data is None: + raise ValueError("sigma_data must be provided when normalization_type is 'EDM'") + c_skip = (sigma_data**2) / (sigma**2 + sigma_data**2) + c_out = sigma * sigma_data / torch.sqrt(sigma_data**2 + sigma**2) + c_in = 1 / torch.sqrt(sigma**2 + sigma_data**2) + c_noise = torch.log(sigma / sigma_data) * 0.25 + return c_in, c_skip, c_out, c_noise + + if normalization_type == "JAMUN": + A = torch.as_tensor(average_squared_distance) + B = torch.as_tensor(2 * D * sigma**2) + + c_in = 1.0 / torch.sqrt(A + B) + c_skip = A / (A + B) + c_out = torch.sqrt((A * B) / (A + B)) + c_noise = torch.log(sigma) / 4 + return c_in, c_skip, c_out, c_noise + + raise ValueError(f"Unknown normalization type: {normalization_type}") \ No newline at end of file diff --git a/src/jamun/utils/sampling_wrapper.py b/src/jamun/utils/sampling_wrapper.py index 77ac604..5e461d6 100644 --- a/src/jamun/utils/sampling_wrapper.py +++ b/src/jamun/utils/sampling_wrapper.py @@ -11,7 +11,7 @@ class ModelSamplingWrapper: """Wrapper to sample positions from a model.""" - def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = False): + def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True): self._model = model self.init_graphs = init_graphs self.sigma = sigma @@ -92,7 +92,7 @@ def unbatch_samples(self, samples: Dict[str, torch.Tensor]) -> List[torch_geomet class ModelSamplingWrapperMemory: """Wrapper for models that depend on a memory of states.""" - def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = False): + def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True): self._model = model self.init_graphs = init_graphs self.sigma = sigma From c1d7e26b424ff1c3a4ec99483d8df2b4c6a7a757 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Thu, 24 Jul 2025 18:33:52 +0000 Subject: [PATCH 14/32] added kwargs to score function processed --- src/jamun/sampling/mcmc/functional/_splitting.py | 5 ----- 1 file changed, 5 deletions(-) diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index 0eac75a..f02d27b 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -1,12 +1,7 @@ import logging import math -<<<<<<< HEAD from typing import Callable, Optional, Tuple, Union from copy import deepcopy -======= -from collections.abc import Callable - ->>>>>>> main import torch from tqdm.auto import tqdm From 67c6c13cc3d32ac1067f695293635daed23ebf3a Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Sat, 26 Jul 2025 01:51:49 +0000 Subject: [PATCH 15/32] catch up with main --- ...a_enhanced_full_grid_multimeasurement.yaml | 10 ++++----- src/jamun/data/_mdtraj.py | 19 +++++----------- src/jamun/data/_utils.py | 1 + src/jamun/model/denoiser_conditional.py | 22 ++++++++++--------- src/jamun/model/denoiser_multimeasurement.py | 22 ++++++++++--------- src/jamun/model/denoiser_spiked.py | 22 ++++++++++--------- .../sampling/mcmc/functional/_splitting.py | 4 ---- src/jamun/utils/sampling_wrapper.py | 4 ++-- 8 files changed, 50 insertions(+), 54 deletions(-) diff --git a/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml b/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml index e784b2f..3fb129e 100644 --- a/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml +++ b/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml @@ -8,7 +8,7 @@ defaults: data: datamodule: - batch_size: 32 + batch_size: 2 datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -44,13 +44,13 @@ model: _target_: jamun.model.conditioners.PositionConditioner N_structures: ${model.arch.N_structures} # Override multimeasurement parameters for this experiment - N_measurements: 5 - N_measurements_hidden: 5 - max_graphs_per_batch: ${data.datamodule.batch_size} + N_measurements: 4 + N_measurements_hidden: 4 + max_graphs_per_batch: null trainer: val_check_interval: 0.5 - max_epochs: 100 + max_epochs: 50 devices: 1 logger: diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index 87960c7..4e21b8d 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -52,17 +52,10 @@ def get_subsampled_indices( return lagged_indices -def preprocess_topology(topology: md.Topology) -> Tuple[torch_geometric.data.Data, md.Topology, md.Topology]: - """Preprocess the MDtraj topology, returning a PyTorch Geometric graph, the topology with protein only, and the topology with hydrogenated protein.""" - # Select all heavy atoms in the protein. - # This also removes all waters. - select = topology.select("protein and not type H") - top = topology.subset(select) - - # Select all atoms in the protein. - select_withH = topology.select("protein") - top_withH = topology.subset(select_withH) - +def make_graph_from_topology( + topology: md.Topology, +) -> torch_geometric.data.Data: + """Create a PyTorch Geometric graph from an MDTraj topology.""" # Encode the atom types, residue codes, and residue sequence indices. atom_type_index = torch.tensor( [utils.encode_atom_type(x.element.symbol) for x in topology.atoms], dtype=torch.int32 @@ -274,8 +267,8 @@ def __init__( self.lagged_indices = None topology = self.traj.topology - self.graph, self.top, self.top_withH = preprocess_topology(topology) - atom_selection = topology.select("protein and not type H") + self.top, atom_selection = preprocess_topology(topology, keep_hydrogens=False) + self.graph = make_graph_from_topology(self.top) self.traj = self.traj.atom_slice(atom_selection) if self.hidden_state is not None: self.hidden_state = [traj.atom_slice(atom_selection) for traj in self.hidden_state] # select protein atoms for hidden state(s) diff --git a/src/jamun/data/_utils.py b/src/jamun/data/_utils.py index 22967b7..75bef0f 100644 --- a/src/jamun/data/_utils.py +++ b/src/jamun/data/_utils.py @@ -11,6 +11,7 @@ from jamun.data._mdtraj import MDtrajDataset, MDtrajIterableDataset from jamun.data._sdf import MDtrajSDFDataset +from typing import Optional, List, Sequence, Dict, Any def dloader_map_reduce(f, dloader, reduce_fn=torch.cat, verbose: bool = False): diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index e25e352..6724949 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -5,7 +5,8 @@ import numpy as np import torch import torch_geometric -from e3tools import radius_graph, scatter +import e3tools +from e3tools import scatter from tqdm import tqdm from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing @@ -213,26 +214,27 @@ def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Te def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torch_geometric.data.Batch: """Add edges to the graph based on the effective radial cutoff.""" + if y.get("edge_index") is not None: + return y + + y = y.clone() if "batch" in y: batch = y["batch"] else: batch = torch.zeros(y.num_nodes, dtype=torch.long, device=self.device) - # Our dataloader already adds the bonded edges. - bonded_edge_index = y.edge_index - with torch.cuda.nvtx.range("radial_graph"): - radial_edge_index = radius_graph(y.pos, radial_cutoff, batch) + radial_edge_index = e3tools.radius_graph(y.pos, radial_cutoff, batch) with torch.cuda.nvtx.range("concatenate_edges"): - edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) - if bonded_edge_index.numel() == 0: - bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device) + edge_index = torch.cat((radial_edge_index, y.bonded_edge_index), dim=-1) + if y.bonded_edge_index.numel() == 0: + bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device) else: bond_mask = torch.cat( ( - torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device), - torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=self.device), + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device), + torch.ones(y.bonded_edge_index.shape[1], dtype=torch.long, device=y.pos.device), ), dim=0, ) diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py index 4702f38..1542d21 100644 --- a/src/jamun/model/denoiser_multimeasurement.py +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -1,11 +1,12 @@ import logging from typing import Callable, Dict, Optional, Tuple, Union +import e3tools import lightning.pytorch as pl import numpy as np import torch import torch_geometric -from e3tools import radius_graph, scatter +from e3tools import scatter from tqdm import tqdm from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing @@ -350,31 +351,32 @@ def add_edges( self, y: torch_geometric.data.Batch, radial_cutoff: float ) -> torch_geometric.data.Batch: """Add edges to the graph based on the effective radial cutoff.""" + if y.get("edge_index") is not None: + return y + + y = y.clone() if "batch" in y: batch = y["batch"] else: batch = torch.zeros(y.num_nodes, dtype=torch.long, device=self.device) - # Our dataloader already adds the bonded edges. - bonded_edge_index = y.edge_index - with torch.cuda.nvtx.range("radial_graph"): - radial_edge_index = radius_graph(y.pos, radial_cutoff, batch) + radial_edge_index = e3tools.radius_graph(y.pos, radial_cutoff, batch) with torch.cuda.nvtx.range("concatenate_edges"): - edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) - if bonded_edge_index.numel() == 0: + edge_index = torch.cat((radial_edge_index, y.bonded_edge_index), dim=-1) + if y.bonded_edge_index.numel() == 0: bond_mask = torch.zeros( - radial_edge_index.shape[1], dtype=torch.long, device=self.device + radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device ) else: bond_mask = torch.cat( ( torch.zeros( - radial_edge_index.shape[1], dtype=torch.long, device=self.device + radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device ), torch.ones( - bonded_edge_index.shape[1], dtype=torch.long, device=self.device + y.bonded_edge_index.shape[1], dtype=torch.long, device=y.pos.device ), ), dim=0, diff --git a/src/jamun/model/denoiser_spiked.py b/src/jamun/model/denoiser_spiked.py index 13f44c8..ed28542 100644 --- a/src/jamun/model/denoiser_spiked.py +++ b/src/jamun/model/denoiser_spiked.py @@ -1,11 +1,12 @@ import logging from typing import Callable, Dict, Optional, Tuple, Union +import e3tools import lightning.pytorch as pl import numpy as np import torch import torch_geometric -from e3tools import radius_graph, scatter +from e3tools import scatter from tqdm import tqdm from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing @@ -215,26 +216,27 @@ def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Te def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torch_geometric.data.Batch: """Add edges to the graph based on the effective radial cutoff.""" + if y.get("edge_index") is not None: + return y + + y = y.clone() if "batch" in y: batch = y["batch"] else: batch = torch.zeros(y.num_nodes, dtype=torch.long, device=self.device) - # Our dataloader already adds the bonded edges. - bonded_edge_index = y.edge_index - with torch.cuda.nvtx.range("radial_graph"): - radial_edge_index = radius_graph(y.pos, radial_cutoff, batch) + radial_edge_index = e3tools.radius_graph(y.pos, radial_cutoff, batch) with torch.cuda.nvtx.range("concatenate_edges"): - edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) - if bonded_edge_index.numel() == 0: - bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device) + edge_index = torch.cat((radial_edge_index, y.bonded_edge_index), dim=-1) + if y.bonded_edge_index.numel() == 0: + bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device) else: bond_mask = torch.cat( ( - torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=self.device), - torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=self.device), + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device), + torch.ones(y.bonded_edge_index.shape[1], dtype=torch.long, device=y.pos.device), ), dim=0, ) diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index f02d27b..7c5f130 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -26,11 +26,7 @@ def initialize_velocity(v_init: str | torch.Tensor, y: torch.Tensor, u: float) - def create_score_fn(score_fn: Callable, inverse_temperature: float, score_fn_clip: float | None) -> Callable: """Create a score function that is clipped and scaled by the inverse temperature.""" -<<<<<<< HEAD def score_fn_processed(y: torch.Tensor, *args, **kwargs) -> Tuple[torch.Tensor, torch.Tensor]: -======= - def score_fn_processed(y: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]: ->>>>>>> main """Score function clipped and scaled by the inverse temperature.""" orig_score = score_fn(y, *args, **kwargs).to(dtype=y.dtype) # Clip the score by norm. diff --git a/src/jamun/utils/sampling_wrapper.py b/src/jamun/utils/sampling_wrapper.py index 8850a6c..c50f2a6 100644 --- a/src/jamun/utils/sampling_wrapper.py +++ b/src/jamun/utils/sampling_wrapper.py @@ -4,7 +4,7 @@ import torch_geometric from jamun.utils import mean_center - +from typing import Dict, List, Optional class ModelSamplingWrapper: """Wrapper to sample positions from a model.""" @@ -141,7 +141,7 @@ def positions_to_graph(self, positions: torch.Tensor, y_hist: list) -> torch_geo input_graph.hidden_state = y_hist return input_graph.to(positions.device) - def unbatch_samples(self, samples: Dict[str, torch.Tensor]) -> List[torch_geometric.data.Data]: + def unbatch_samples(self, samples: dict[str, torch.Tensor]) -> list[torch_geometric.data.Data]: """Unbatch samples.""" if "batch" not in self.init_graphs: raise ValueError("The initial graph does not have a batch attribute.") From 02219ab8e1ab75b8f263484ac6b2817f16de81c1 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Sun, 27 Jul 2025 04:14:20 +0000 Subject: [PATCH 16/32] Updated configs/sbatch script for single run - Reorganized train scripts for enhanced sampling data, with a new script for running a single conditional model in train_enhanced_sampling_single.sh - Sample training script in configs/experiment/train_enhanced_position_conditioner.yaml --- .../train_enhanced_position_conditioner.yaml | 4 +- ...> train_enhanced_sampling_conditioners.sh} | 0 ...anced_sampling_noise_conditioner_sweep.sh} | 4 +- .../slurm/train_enhanced_sampling_single.sh | 23 ++++++ .../slurm/train_enhanced_sampling_sweep.sh | 76 ------------------- 5 files changed, 28 insertions(+), 79 deletions(-) rename scripts/slurm/{run_conditioner_experiment.sh => train_enhanced_sampling_conditioners.sh} (100%) rename scripts/slurm/{train_enhanced_sampling_trial.sh => train_enhanced_sampling_noise_conditioner_sweep.sh} (98%) create mode 100755 scripts/slurm/train_enhanced_sampling_single.sh delete mode 100755 scripts/slurm/train_enhanced_sampling_sweep.sh diff --git a/configs/experiment/train_enhanced_position_conditioner.yaml b/configs/experiment/train_enhanced_position_conditioner.yaml index 74535b3..1e02bfe 100644 --- a/configs/experiment/train_enhanced_position_conditioner.yaml +++ b/configs/experiment/train_enhanced_position_conditioner.yaml @@ -20,6 +20,7 @@ data: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 + max_datasets: 1 val: _target_: jamun.data.parse_datasets_from_directory @@ -29,6 +30,7 @@ data: subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + max_datasets: ${data.datamodule.datasets.train.max_datasets} model: sigma_distribution: @@ -46,7 +48,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 50 + max_epochs: 1 devices: 1 logger: diff --git a/scripts/slurm/run_conditioner_experiment.sh b/scripts/slurm/train_enhanced_sampling_conditioners.sh similarity index 100% rename from scripts/slurm/run_conditioner_experiment.sh rename to scripts/slurm/train_enhanced_sampling_conditioners.sh diff --git a/scripts/slurm/train_enhanced_sampling_trial.sh b/scripts/slurm/train_enhanced_sampling_noise_conditioner_sweep.sh similarity index 98% rename from scripts/slurm/train_enhanced_sampling_trial.sh rename to scripts/slurm/train_enhanced_sampling_noise_conditioner_sweep.sh index af55511..043b0e2 100755 --- a/scripts/slurm/train_enhanced_sampling_trial.sh +++ b/scripts/slurm/train_enhanced_sampling_noise_conditioner_sweep.sh @@ -1,12 +1,12 @@ #!/usr/bin/env bash -#SBATCH --partition gpu3 +#SBATCH --partition gpu2 #SBATCH --qos=preempt #SBATCH --nodes 1 #SBATCH --ntasks-per-node 1 #SBATCH --gpus-per-node 1 #SBATCH --cpus-per-task 8 -#SBATCH --time 0-2 +#SBATCH --time 1-0 #SBATCH --mem-per-cpu=32G #SBATCH --array=0 diff --git a/scripts/slurm/train_enhanced_sampling_single.sh b/scripts/slurm/train_enhanced_sampling_single.sh new file mode 100755 index 0000000..ee6d6ae --- /dev/null +++ b/scripts/slurm/train_enhanced_sampling_single.sh @@ -0,0 +1,23 @@ +#!/usr/bin/env bash + +#SBATCH --partition gpu2 +#SBATCH --nodes 1 +#SBATCH --ntasks-per-node 1 +#SBATCH --gpus-per-node 1 +#SBATCH --cpus-per-task 8 +#SBATCH --time 3-0 +#SBATCH --mem-per-cpu=32G + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +# Run training with parameter overrides +jamun_train --config-dir=configs experiment=train_enhanced_position_conditioner \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_sampling_sweep.sh b/scripts/slurm/train_enhanced_sampling_sweep.sh deleted file mode 100755 index db22dd6..0000000 --- a/scripts/slurm/train_enhanced_sampling_sweep.sh +++ /dev/null @@ -1,76 +0,0 @@ -#!/usr/bin/env bash - -#SBATCH --partition gpu2 -#SBATCH --nodes 1 -#SBATCH --ntasks-per-node 1 -#SBATCH --gpus-per-node 1 -#SBATCH --cpus-per-task 8 -#SBATCH --time 3-0 -#SBATCH --mem-per-cpu=32G -#SBATCH --array=0-2 - -# Initialize conda -source ~/.bashrc -eval "$(conda shell.bash hook)" -conda activate jamun - -# Verify conda activation worked -which python -echo "Python path: $(which python)" -echo "Conda environment: $CONDA_DEFAULT_ENV" - -set -eux - -echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" -echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" -echo "hostname = $(hostname)" - -export HYDRA_FULL_ERROR=1 - -# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. -RUN_KEY=$(openssl rand -hex 12) -echo "RUN_KEY = ${RUN_KEY}" - -# Define parameter arrays -CONDITIONERS=("jamun.model.conditioners.PositionConditioner") -CONDITIONER_NAMES=("PositionConditioner") -LAG_SUBSAMPLE_RATES=(1 5 10) - -# Fixed parameters -SIGMA=0.04 -TOTAL_LAG_TIME=5 -SUBSAMPLE=1 - -# Calculate parameter indices from SLURM_ARRAY_TASK_ID -# Total combinations: 1 conditioner * 3 lag_subsample_rates = 3 -LAG_SUBSAMPLE_IDX=${SLURM_ARRAY_TASK_ID} - -# Get parameter values -CONDITIONER=${CONDITIONERS[0]} -CONDITIONER_NAME=${CONDITIONER_NAMES[0]} -LAG_SUBSAMPLE_RATE=${LAG_SUBSAMPLE_RATES[$LAG_SUBSAMPLE_IDX]} - -echo "Parameter combination ${SLURM_ARRAY_TASK_ID}:" -echo " Conditioner: ${CONDITIONER_NAME}" -echo " Sigma: ${SIGMA}" -echo " Total lag time: ${TOTAL_LAG_TIME}" -echo " Lag subsample rate: ${LAG_SUBSAMPLE_RATE}" -echo " Subsample: ${SUBSAMPLE}" - -nvidia-smi - -# Run training with parameter overrides -jamun_train --config-dir=configs \ - experiment=train_test_single_shape_enhanced_sampling.yaml \ - ++trainer.max_epochs=100 \ - ++data.datamodule.datasets.train.subsample=${SUBSAMPLE} \ - ++data.datamodule.datasets.val.subsample=${SUBSAMPLE} \ - ++data.datamodule.datasets.test.subsample=${SUBSAMPLE} \ - ++model.conditioner._target_=${CONDITIONER} \ - ++model.sigma_distribution.sigma=${SIGMA} \ - ++data.datamodule.datasets.train.total_lag_time=${TOTAL_LAG_TIME} \ - ++data.datamodule.datasets.train.lag_subsample_rate=${LAG_SUBSAMPLE_RATE} \ - ++model.arch.N_structures=${TOTAL_LAG_TIME} \ - ++logger.wandb.group="fake_enhanced_data_jul17_sweep_lag_times" \ - ++logger.wandb.tags=["'${SLURM_JOB_ID}'","'${RUN_KEY}'","train","enhanced_sampling","${CONDITIONER_NAME}","sigma_${SIGMA}","lag_${TOTAL_LAG_TIME}","lag_subsample_${LAG_SUBSAMPLE_RATE}"] \ - ++run_key=$RUN_KEY \ No newline at end of file From c6b04d184a5a1d06a297e451aad97b3ca437552f Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Mon, 28 Jul 2025 23:10:45 +0000 Subject: [PATCH 17/32] feat: Add spatiotemporal conditioning with input attributes for enhanced denoising - Extended E3ConvConditional to accept additional input attributes - Added input_irrep_aggregator for combining node_attr with input_attr - Supports "3x1e" input attributes from spatiotemporal features - Maintains full E(3)-equivariance throughout processing pipeline - Subclass of Denoiser that extracts spatiotemporal features as input_attr - Modified xhat_normalized() to integrate spatiotemporal conditioning - Automatically extracts features from SpatioTemporalConditioner - Graceful fallback when spatiotemporal model unavailable - Updated to always return exactly 1 structure (not N_structures) - Improved documentation and error handling - Better integration with new input attribute system - src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml - src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml - src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml - configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml - Fixed E3SpatioTemporal to use E3Conv instead of E3ConvConditional - Corrected import statements throughout spatiotemporal.py - Updated all irreps to use "3x1e" consistently for proper dimensionality - Fixed circular config references in YAML files - Corrected architecture call signatures (positional vs keyword args) - Fixed missing imports in average_squared_distance.py - Updated return type annotations for consistency - Improved edge addition logic in test scripts - Moved helper functions from scratch/transformer/helpers.py to appropriate utils modules - Enhanced test script with proper DenoiserWithInputAttr testing - Added comprehensive error handling and validation - Improved device handling and tensor shape management - All irreps configurations updated to "3x1e" for consistency - Factory functions properly configured for all architectures - Experiment configs optimized for training stability This implementation enables conditioning denoising models on rich spatiotemporal features while maintaining full equivariance and providing a clean, extensible architecture for future enhancements. --- ...n_enhanced_spatiotemporal_conditioner.yaml | 55 +++ ...otemporal_conditioner_with_input_attr.yaml | 55 +++ .../e3conv_conditional_with_input_attr.yaml | 44 ++ .../model/arch/spatiotemporal.yaml | 46 ++ .../model/conditioners/denoised.yaml | 5 + .../hydra_config/model/conditioners/mean.yaml | 3 + .../model/conditioners/position.yaml | 4 + .../hydra_config/model/conditioners/self.yaml | 3 + .../model/conditioners/spatiotemporal.yaml | 51 ++ .../spatiotemporal_pretrained.yaml | 51 ++ .../spatiotemporal_with_input_attr.yaml | 53 +++ .../model/conditioners/spiked.yaml | 3 + .../model/denoiser_conditional.yaml | 6 +- .../denoiser_conditional_with_input_attr.yaml | 26 ++ .../denoiser_spatiotemporal_conditioner.yaml | 26 ++ src/jamun/model/arch/__init__.py | 1 + src/jamun/model/arch/e3conv_conditional.py | 166 +++++++ src/jamun/model/arch/spatiotemporal.py | 441 ++++++++++++++++++ src/jamun/model/conditioners/conditioners.py | 101 +++- src/jamun/model/denoiser_conditional.py | 151 ++++++ src/jamun/model/pooling.py | 217 +++++++++ src/jamun/utils/__init__.py | 3 +- src/jamun/utils/average_squared_distance.py | 51 ++ test_spatiotemporal_conditioner.py | 394 ++++++++++++++++ 24 files changed, 1951 insertions(+), 5 deletions(-) create mode 100644 configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml create mode 100644 configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml create mode 100644 src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml create mode 100644 src/jamun/hydra_config/model/arch/spatiotemporal.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/denoised.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/mean.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/position.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/self.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/spatiotemporal_pretrained.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml create mode 100644 src/jamun/hydra_config/model/conditioners/spiked.yaml create mode 100644 src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml create mode 100644 src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml create mode 100644 src/jamun/model/arch/spatiotemporal.py create mode 100644 src/jamun/model/pooling.py create mode 100755 test_spatiotemporal_conditioner.py diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml new file mode 100644 index 0000000..7a9842f --- /dev/null +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml @@ -0,0 +1,55 @@ +# @package _global_ +# Training SpatioTemporalConditioner on enhanced sampling data + +defaults: + - override /model: denoiser_spatiotemporal_conditioner + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 16 # Reduced batch size due to increased model complexity + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 2 # Increased for more training data + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + max_datasets: 1 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + max_radius: 1000.0 + optim: + lr: 0.001 # Slightly reduced learning rate for stability + +trainer: + val_check_interval: 0.5 + max_epochs: 2 # Increased due to model complexity + devices: 1 + gradient_clip_val: 1.0 # Add gradient clipping for stability + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "SpatioTemporalConditioner on enhanced sampling data - processes temporal sequences through spatial and temporal modules" + tags: ["spatiotemporal_conditioner", "enhanced_sampling", "transformer", "e3conv"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml new file mode 100644 index 0000000..807e515 --- /dev/null +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml @@ -0,0 +1,55 @@ +# @package _global_ +# Training SpatioTemporalConditioner with E3ConvConditionalWithInputAttr on enhanced sampling data + +defaults: + - override /model: denoiser_conditional_with_input_attr + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 16 # Reduced batch size due to increased model complexity + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 2 # Increased for more training data + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + max_datasets: ${data.datamodule.datasets.train.max_datasets} + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 2 + N_structures: 1 # E3ConvConditionalWithInputAttr + SpatioTemporalConditioner + max_radius: 1000.0 + optim: + lr: 0.0008 # Slightly reduced learning rate for stability with input attributes + +trainer: + val_check_interval: 0.5 + max_epochs: 10 + devices: 1 + +logger: + wandb: + group: enhanced_sampling_spatiotemporal_with_input_attr + notes: "SpatioTemporalConditioner with E3ConvConditionalWithInputAttr on enhanced sampling data" + tags: ["spatiotemporal_conditioner", "enhanced_sampling", "input_attr", "e3conv_conditional"] \ No newline at end of file diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml new file mode 100644 index 0000000..d48182f --- /dev/null +++ b/src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml @@ -0,0 +1,44 @@ +# E3ConvConditionalWithInputAttr architecture configuration +# This extends E3ConvConditional to accept additional input attributes + +_target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalWithInputAttr + +# Standard E3ConvConditional parameters +irreps_out: "3x1e" +irreps_hidden: "120x0e + 32x1e" +irreps_sh: "1x0e + 1x1e" +n_layers: 2 +edge_attr_dim: 64 +use_residue_information: ${data.use_residue_information} +atom_type_embedding_dim: 8 +atom_code_embedding_dim: 8 +residue_code_embedding_dim: 32 +residue_index_embedding_dim: 8 +use_residue_sequence_index: false +num_atom_types: ${data.num_atom_types} +max_sequence_length: ${data.max_sequence_length} +num_atom_codes: ${data.num_atom_codes} +num_residue_types: ${data.num_residue_types} +test_equivariance: false +reduce: null +N_structures: ${model.arch.N_structures} + +# New parameter for input attributes +input_attr_irreps: "3x1e" # 3D vector input attributes (matching transformer output) + +# Factory functions for layers +hidden_layer_factory: + _target_: functools.partial + _args_: + - _target_: e3tools.nn.ConvBlock + conv: + _target_: functools.partial + _args_: + - _target_: e3tools.nn.Conv + +output_head_factory: + _target_: functools.partial + _args_: + - _target_: e3tools.nn.EquivariantMLP + irreps_hidden_list: + - "120x0e + 32x1e" \ No newline at end of file diff --git a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml new file mode 100644 index 0000000..109a67f --- /dev/null +++ b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml @@ -0,0 +1,46 @@ +# Configuration for E3SpatioTemporal architecture +_target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal + +# Cutoff parameters +radial_cutoff: 0.05 +temporal_cutoff: 1.0 + +# Spatial module (E3Conv) +spatial_module: + _target_: jamun.model.arch.e3conv_conditional.E3Conv + irreps_out: "1x1e" + irreps_hidden: "120x0e + 32x1e" + irreps_sh: "1x0e + 1x1e" + n_layers: 1 + edge_attr_dim: 64 + use_residue_information: true + atom_type_embedding_dim: 8 + atom_code_embedding_dim: 8 + residue_code_embedding_dim: 32 + residue_index_embedding_dim: 8 + use_residue_sequence_index: false + num_atom_types: 20 + max_sequence_length: 10 + num_atom_codes: 10 + num_residue_types: 25 + test_equivariance: false + reduce: null + +# Temporal module (E3Transformer) +temporal_module: + _target_: jamun.model.arch.spatiotemporal.E3Transformer + irreps_out: "3x1e" + irreps_hidden: "8x0e + 4x1e" + irreps_sh: "1x0e + 1x1e" + irreps_node_attr: "1x1e" # Match spatial module output + num_layers: 2 + edge_attr_dim: 24 + num_attention_heads: 1 + reduce: null + +# Pooling modules +spatial_to_temporal_pooler: + _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr + +temporal_to_spatial_pooler: + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/denoised.yaml b/src/jamun/hydra_config/model/conditioners/denoised.yaml new file mode 100644 index 0000000..c3c242f --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/denoised.yaml @@ -0,0 +1,5 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.DenoisedConditioner +N_structures: 1 +pretrained_model_path: "wandb_run_path/here" # Replace with actual wandb run path +c_in: 1.0 \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/mean.yaml b/src/jamun/hydra_config/model/conditioners/mean.yaml new file mode 100644 index 0000000..8157c8e --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/mean.yaml @@ -0,0 +1,3 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.MeanConditioner +N_structures: 1 \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/position.yaml b/src/jamun/hydra_config/model/conditioners/position.yaml new file mode 100644 index 0000000..0cb184c --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/position.yaml @@ -0,0 +1,4 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.PositionConditioner +N_structures: 1 +align_hidden_states: true \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/self.yaml b/src/jamun/hydra_config/model/conditioners/self.yaml new file mode 100644 index 0000000..f6334c7 --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/self.yaml @@ -0,0 +1,3 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.SelfConditioner +N_structures: 1 \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml b/src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml new file mode 100644 index 0000000..6ef8c49 --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml @@ -0,0 +1,51 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner +N_structures: 1 +c_noise: 0.0 +freeze_spatiotemporal_model: false # Trainable by default + +# Spatiotemporal model configuration +spatiotemporal_model: + _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal + radial_cutoff: 0.05 + temporal_cutoff: 1.0 + + # Spatial module (E3Conv) + spatial_module: + _target_: jamun.model.arch.e3conv.E3Conv + irreps_out: "3x1e" + irreps_hidden: "120x0e + 32x1e" + irreps_sh: "1x0e + 1x1e" + n_layers: 1 + edge_attr_dim: 64 + use_residue_information: true + atom_type_embedding_dim: 8 + atom_code_embedding_dim: 8 + residue_code_embedding_dim: 32 + residue_index_embedding_dim: 8 + use_residue_sequence_index: false + num_atom_types: 20 + max_sequence_length: 10 + num_atom_codes: 10 + num_residue_types: 25 + test_equivariance: false + reduce: null + + # Temporal module (E3Transformer) + temporal_module: + _target_: jamun.model.arch.spatiotemporal.E3Transformer + irreps_out: "3x1e" + irreps_hidden: "8x0e + 4x1e" + irreps_sh: "1x0e + 1x1e" + irreps_node_attr: "3x1e" # Match spatial module output + num_layers: 2 + edge_attr_dim: 24 + num_attention_heads: 1 + reduce: null + + # Pooling modules + spatial_to_temporal_pooler: + _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr + + temporal_to_spatial_pooler: + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/spatiotemporal_pretrained.yaml b/src/jamun/hydra_config/model/conditioners/spatiotemporal_pretrained.yaml new file mode 100644 index 0000000..38b3f39 --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/spatiotemporal_pretrained.yaml @@ -0,0 +1,51 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner +N_structures: 1 +c_noise: 0.0 +freeze_spatiotemporal_model: true # Frozen for pretrained models + +# Spatiotemporal model configuration +spatiotemporal_model: + _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal + radial_cutoff: 0.05 + temporal_cutoff: 1.0 + + # Spatial module (E3Conv) + spatial_module: + _target_: jamun.model.arch.e3conv_conditional.E3Conv + irreps_out: "1x1e" + irreps_hidden: "120x0e + 32x1e" + irreps_sh: "1x0e + 1x1e" + n_layers: 1 + edge_attr_dim: 64 + use_residue_information: true + atom_type_embedding_dim: 8 + atom_code_embedding_dim: 8 + residue_code_embedding_dim: 32 + residue_index_embedding_dim: 8 + use_residue_sequence_index: false + num_atom_types: 20 + max_sequence_length: 10 + num_atom_codes: 10 + num_residue_types: 25 + test_equivariance: false + reduce: null + + # Temporal module (E3Transformer) + temporal_module: + _target_: jamun.model.arch.spatiotemporal.E3Transformer + irreps_out: "3x1e" + irreps_hidden: "8x0e + 4x1e" + irreps_sh: "1x0e + 1x1e" + irreps_node_attr: "1x1e" # Match spatial module output + num_layers: 2 + edge_attr_dim: 24 + num_attention_heads: 1 + reduce: null + + # Pooling modules + spatial_to_temporal_pooler: + _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr + + temporal_to_spatial_pooler: + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml b/src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml new file mode 100644 index 0000000..81df378 --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml @@ -0,0 +1,53 @@ +# @package _global_ +# SpatioTemporalConditioner configuration that provides input attributes to E3ConvConditionalWithInputAttr + +_target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner +N_structures: 1 +c_noise: 0.0 +freeze_spatiotemporal_model: false # Trainable by default + +# Spatiotemporal model configuration +spatiotemporal_model: + _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal + radial_cutoff: 0.05 + temporal_cutoff: 1.0 + + # Spatial module (E3Conv) - outputs features that will be used as input_attr + spatial_module: + _target_: jamun.model.arch.e3conv.E3Conv + irreps_out: "3x1e" # Match input_attr_irreps in the main architecture + irreps_hidden: "120x0e + 32x1e" + irreps_sh: "1x0e + 1x1e" + n_layers: 1 + edge_attr_dim: 64 + use_residue_information: true + atom_type_embedding_dim: 8 + atom_code_embedding_dim: 8 + residue_code_embedding_dim: 32 + residue_index_embedding_dim: 8 + use_residue_sequence_index: false + num_atom_types: 20 + max_sequence_length: 10 + num_atom_codes: 10 + num_residue_types: 25 + test_equivariance: false + reduce: null + + # Temporal module (E3Transformer) + temporal_module: + _target_: jamun.model.arch.spatiotemporal.E3Transformer + irreps_out: "3x1e" # Match spatial module output + irreps_hidden: "8x0e + 4x1e" + irreps_sh: "1x0e + 1x1e" + irreps_node_attr: "3x1e" # Match spatial module output + num_layers: 2 + edge_attr_dim: 24 + num_attention_heads: 1 + reduce: null + + # Pooling modules + spatial_to_temporal_pooler: + _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr + + temporal_to_spatial_pooler: + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/spiked.yaml b/src/jamun/hydra_config/model/conditioners/spiked.yaml new file mode 100644 index 0000000..0189deb --- /dev/null +++ b/src/jamun/hydra_config/model/conditioners/spiked.yaml @@ -0,0 +1,3 @@ +# @package _global_ +_target_: jamun.model.conditioners.conditioners.ConditionerSpiked +N_structures: 1 \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_conditional.yaml b/src/jamun/hydra_config/model/denoiser_conditional.yaml index 59143b3..3c8832d 100644 --- a/src/jamun/hydra_config/model/denoiser_conditional.yaml +++ b/src/jamun/hydra_config/model/denoiser_conditional.yaml @@ -2,6 +2,7 @@ defaults: - arch: e3conv_conditional.yaml - optim: adam.yaml - lr_scheduler_config: null + - conditioners: position.yaml # Default conditioner, can be overridden - _self_ max_radius: null @@ -18,8 +19,7 @@ torch_compile_kwargs: dynamic: true mode: default -conditioner: - _target_: jamun.model.conditioners.PositionConditioner - N_structures: ${model.arch.N_structures} +# Conditioner configuration now comes from defaults above +# Use conditioners=spatiotemporal to use the spatio-temporal conditioner _target_: jamun.model.denoiser_conditional.Denoiser \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml b/src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml new file mode 100644 index 0000000..848dfd7 --- /dev/null +++ b/src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml @@ -0,0 +1,26 @@ +defaults: + - arch: e3conv_conditional_with_input_attr.yaml + - optim: adam.yaml + - lr_scheduler_config: null + - conditioners: spatiotemporal.yaml + - _self_ + +max_radius: null +average_squared_distance: null +add_fixed_noise: false +add_fixed_ones: false +align_noisy_input_during_training: true +align_noisy_input_during_evaluation: true +mean_center: true +mirror_augmentation_rate: 0.0 +use_torch_compile: true +torch_compile_kwargs: + fullgraph: true + dynamic: true + mode: default + +# Architecture settings +arch: + N_structures: 1 # Compatible with SpatioTemporalConditioner + +_target_: jamun.model.denoiser_conditional.DenoiserWithInputAttr \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml b/src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml new file mode 100644 index 0000000..e3e84a3 --- /dev/null +++ b/src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml @@ -0,0 +1,26 @@ +defaults: + - arch: e3conv_conditional.yaml + - optim: adam.yaml + - lr_scheduler_config: null + - conditioners: spatiotemporal.yaml # Use spatiotemporal conditioner + - _self_ + +max_radius: null +average_squared_distance: null +add_fixed_noise: false +add_fixed_ones: false +align_noisy_input_during_training: true +align_noisy_input_during_evaluation: true +mean_center: true +mirror_augmentation_rate: 0.0 +use_torch_compile: true +torch_compile_kwargs: + fullgraph: true + dynamic: true + mode: default + +# Architecture settings for spatiotemporal conditioner +arch: + N_structures: 1 # SpatioTemporalConditioner always returns 1 structure + +_target_: jamun.model.denoiser_conditional.Denoiser \ No newline at end of file diff --git a/src/jamun/model/arch/__init__.py b/src/jamun/model/arch/__init__.py index 33b8eeb..313896a 100644 --- a/src/jamun/model/arch/__init__.py +++ b/src/jamun/model/arch/__init__.py @@ -1,3 +1,4 @@ from .e3conv import E3Conv from .ophiuchus import Ophiuchus from .e3conv_conditional import E3ConvConditional +from .spatiotemporal import E3Transformer, E3SpatioTemporal diff --git a/src/jamun/model/arch/e3conv_conditional.py b/src/jamun/model/arch/e3conv_conditional.py index 6365001..1173aa2 100644 --- a/src/jamun/model/arch/e3conv_conditional.py +++ b/src/jamun/model/arch/e3conv_conditional.py @@ -9,6 +9,7 @@ from torch import Tensor from jamun.model.atom_embedding import AtomEmbeddingWithResidueInformation, SimpleAtomEmbedding from jamun.model.noise_conditioning import NoiseConditionalScaling, NoiseConditionalSkipConnection +import e3tools.nn class E3ConvConditional(torch.nn.Module): @@ -141,3 +142,168 @@ def forward( node_attr = scatter(node_attr, topology.batch, dim=0, reduce=self.reduce) return node_attr + + +class E3ConvConditionalWithInputAttr(E3ConvConditional): + """ + Extension of E3ConvConditional that can accept additional input attributes + and combine them with the computed node attributes. + """ + + def __init__( + self, + irreps_out: str | Irreps, + irreps_hidden: str | Irreps, + irreps_sh: str | Irreps, + hidden_layer_factory: Callable[..., torch.nn.Module], + output_head_factory: Callable[..., torch.nn.Module], + use_residue_information: bool, + n_layers: int, + edge_attr_dim: int, + atom_type_embedding_dim: int, + atom_code_embedding_dim: int, + residue_code_embedding_dim: int, + residue_index_embedding_dim: int, + use_residue_sequence_index: bool, + num_atom_types: int = 20, + max_sequence_length: int = 10, + num_atom_codes: int = 10, + num_residue_types: int = 25, + test_equivariance: bool = False, + reduce: str | None = None, + N_structures: int = 1, + input_attr_irreps: str | Irreps | None = None, + ): + """ + Initialize E3ConvConditionalWithInputAttr. + + Args: + input_attr_irreps: Irreps of the input attributes that will be combined with node_attr. + If None, the model behaves like the parent class. + All other args: Same as parent E3ConvConditional class. + """ + super().__init__( + irreps_out=irreps_out, + irreps_hidden=irreps_hidden, + irreps_sh=irreps_sh, + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + use_residue_information=use_residue_information, + n_layers=n_layers, + edge_attr_dim=edge_attr_dim, + atom_type_embedding_dim=atom_type_embedding_dim, + atom_code_embedding_dim=atom_code_embedding_dim, + residue_code_embedding_dim=residue_code_embedding_dim, + residue_index_embedding_dim=residue_index_embedding_dim, + use_residue_sequence_index=use_residue_sequence_index, + num_atom_types=num_atom_types, + max_sequence_length=max_sequence_length, + num_atom_codes=num_atom_codes, + num_residue_types=num_residue_types, + test_equivariance=test_equivariance, + reduce=reduce, + N_structures=N_structures, + ) + + self.input_attr_irreps = o3.Irreps(input_attr_irreps) if input_attr_irreps is not None else None + + # Create input irrep aggregator if input attributes are provided + if self.input_attr_irreps is not None: + # Combined irreps: node_attr irreps + input_attr irreps + combined_irreps = self.irreps_hidden + self.input_attr_irreps + + # Create aggregator that takes combined input and outputs node_attr irreps + self.input_irrep_aggregator = e3tools.nn.EquivariantMLP( + irreps_in=combined_irreps, + irreps_out=self.irreps_hidden, + irreps_hidden_list=[self.irreps_hidden], # Single hidden layer + ) + else: + self.input_irrep_aggregator = None + + def forward( + self, + pos: Tensor, + topology: torch_geometric.data.Batch, + c_noise: Tensor, + effective_radial_cutoff: float, + input_attr: Tensor | None = None, + ) -> Tensor: + """ + Forward pass with optional input attributes. + + Args: + pos: Node positions + topology: Graph topology + c_noise: Noise conditioning + effective_radial_cutoff: Radial cutoff for edges + input_attr: Optional input attributes to combine with node_attr. + Should have shape [N, input_attr_irreps.dim] where N is number of nodes. + + Returns: + Node attributes after processing + """ + # Extract edge attributes. + edge_index = topology["edge_index"] + bond_mask = topology["bond_mask"] + + src, dst = edge_index # compute edge spherical harmonics over concat structures + positions = torch.split(pos, 3, dim=-1) + edge_sh = [] + for block in positions: + edge_vec = block[src] - block[dst] + edge_sh.append(self.sh(edge_vec)) + edge_sh = torch.cat(edge_sh, dim=-1) + + # print(f"Edge spherical harmonics: {type(edge_sh)}") + bonded_edge_attr = self.embed_bondedness(bond_mask) + edge_vec_main = positions[0][src] - positions[0][dst] + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec_main.norm(dim=1), + 0.0, + effective_radial_cutoff, + self.radial_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr), dim=-1) + + node_attr = self.atom_embedder(topology) + node_attr = self.initial_noise_scaling(node_attr, c_noise) + node_attr = self.initial_projector(node_attr, edge_index, edge_attr, edge_sh) + + # Combine with input attributes if provided + if input_attr is not None and self.input_irrep_aggregator is not None: + # Validate input_attr shape + expected_dim = self.input_attr_irreps.dim + if input_attr.shape[-1] != expected_dim: + raise ValueError( + f"Expected input_attr to have dimension {expected_dim}, " + f"but got {input_attr.shape[-1]}" + ) + if input_attr.shape[0] != node_attr.shape[0]: + raise ValueError( + f"Expected input_attr to have {node_attr.shape[0]} nodes, " + f"but got {input_attr.shape[0]}" + ) + + # Concatenate node_attr with input_attr + combined_attr = torch.cat([node_attr, input_attr], dim=-1) + + # Aggregate to get back to node_attr irreps + node_attr = self.input_irrep_aggregator(combined_attr) + elif input_attr is not None and self.input_irrep_aggregator is None: + raise ValueError( + "input_attr provided but input_attr_irreps was not specified during initialization" + ) + + # Continue with normal processing + for scaling, skip, layer in zip(self.noise_scalings, self.skip_connections, self.layers): + node_attr = skip(node_attr, layer(scaling(node_attr, c_noise), edge_index, edge_attr, edge_sh), c_noise) + node_attr = self.output_head(node_attr) + node_attr = node_attr * self.output_gain + + if self.reduce is not None: + node_attr = scatter(node_attr, topology.batch, dim=0, reduce=self.reduce) + + return node_attr diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py new file mode 100644 index 0000000..2ca7a6d --- /dev/null +++ b/src/jamun/model/arch/spatiotemporal.py @@ -0,0 +1,441 @@ +""" +E(3)-equivariant spatio-temporal models and conversion functions. + +This module contains: +- E3Transformer: E(3)-equivariant transformer for temporal graph processing +- E3SpatioTemporal: Unified spatio-temporal processing model +- Spatial-temporal graph conversion utilities +""" + +from typing import Dict, Union + +import e3nn +import torch +import torch.nn as nn +from e3nn import o3 +import torch_geometric +import torch_geometric.data +import e3tools +import e3tools.nn + +from jamun.model.arch.e3conv import E3Conv + + +def calculate_temporal_positions(temporal_length, device=None): + """ + Calculate normalized temporal positions for nodes in a temporal graph. + + Args: + temporal_length: Total number of nodes in the temporal sequence + device: Device to create tensors on + + Returns: + torch.Tensor: Normalized positions [0, 1/T, 2/T, ..., (T-1)/T] + """ + if temporal_length <= 1: + return torch.tensor([0.0], device=device) + + # Create positions [0, 1, 2, ..., T-1] and normalize by T + positions = torch.arange(temporal_length, dtype=torch.float32, device=device) + normalized_positions = positions / temporal_length + + return normalized_positions + + +def spatial_to_temporal_graphs(batch): + """ + Convert a batch of spatial graphs to temporal graphs. + + For each spatial node with position + hidden states, create a temporal graph where: + - Node 0: current position + - Nodes 1-T: hidden state positions + - Connectivity: Node 0 connects to all others, sequential connections 1->2->3->... + """ + # Get device from input batch + device = batch.pos.device + + # Get dimensions + num_spatial_nodes = batch.pos.shape[0] + + # Check if we have hidden states + if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + num_hidden_states = len(batch.hidden_state) + temporal_length = 1 + num_hidden_states # current + hidden + else: + # If no hidden states, just use current position + num_hidden_states = 0 + temporal_length = 1 + + # Store reference to spatial graph + spatial_graph = batch.clone() + + temporal_graphs = [] + + for node_idx in range(num_spatial_nodes): + # Build temporal positions: [current_pos, hidden_1, hidden_2, ...] + temporal_positions = [batch.pos[node_idx]] # Start with current position + + # Add hidden state positions + if num_hidden_states > 0: + for hidden_pos in batch.hidden_state: + temporal_positions.append(hidden_pos[node_idx]) + + temporal_pos = torch.stack(temporal_positions) # Shape: [T, 3] + + # Calculate temporal positions for this sequence + temporal_position = calculate_temporal_positions(temporal_length, device=device) + + # Create edge connectivity + if temporal_length > 1: + # Node 0 connects to all others: 0->1, 0->2, 0->3, ..., 0->T-1 + hub_src = [0] * (temporal_length - 1) + hub_dst = list(range(1, temporal_length)) + + # Sequential connections: 1->2, 2->3, ..., (T-2)->(T-1) + seq_src = list(range(1, temporal_length - 1)) + seq_dst = list(range(2, temporal_length)) + + # Combine all edges + all_src = hub_src + seq_src + all_dst = hub_dst + seq_dst + + edge_index = torch.tensor([all_src, all_dst], dtype=torch.long, device=device) + else: + # Single node, no edges + edge_index = torch.tensor([[], []], dtype=torch.long, device=device) + + # Create temporal graph for this spatial node + temporal_graph = torch_geometric.data.Data( + pos=temporal_pos, + edge_index=edge_index, + spatial_node_idx=torch.tensor([node_idx], device=device), # Track which spatial node this came from + temporal_length=torch.tensor([temporal_length], device=device), + temporal_position=temporal_position # Normalized position in sequence [0, 1/T, 2/T, ...] + ) + temporal_graphs.append(temporal_graph) + + # Batch all temporal graphs + temporal_batch = torch_geometric.data.Batch.from_data_list(temporal_graphs) + + # Store spatial graph reference + temporal_batch.spatial_graph = spatial_graph + + return temporal_batch + + +def temporal_to_spatial_graphs(temporal_batch): + """ + Convert temporal graphs back to spatial graphs. + Take the 0th node position from each temporal graph as the updated spatial position. + """ + # Get the spatial graph template + spatial_graph = temporal_batch.spatial_graph.clone() + + # Extract 0th node positions from each temporal graph + num_temporal_graphs = temporal_batch.num_graphs + updated_positions = [] + + # Iterate through each temporal graph in the batch + for graph_idx in range(num_temporal_graphs): + # Get the node range for this temporal graph + start_idx = temporal_batch.ptr[graph_idx] + + # The 0th node of each temporal graph is at the start of its range + updated_positions.append(temporal_batch.pos[start_idx]) + + # Stack to create new position tensor + updated_positions = torch.stack(updated_positions) + + # Update spatial graph with new positions + spatial_graph.pos = updated_positions + + return spatial_graph + + +class E3Transformer(nn.Module): + """E(3)-equivariant transformer with temporal graph support.""" + + def __init__( + self, + irreps_out: Union[str, e3nn.o3.Irreps], + irreps_hidden: Union[str, e3nn.o3.Irreps], + irreps_sh: Union[str, e3nn.o3.Irreps], + irreps_node_attr: Union[str, e3nn.o3.Irreps], + num_layers: int, + edge_attr_dim: int, + num_attention_heads: int, + reduce: str | None = None, + ): + super().__init__() + + self.irreps_out = o3.Irreps(irreps_out) + self.irreps_hidden = o3.Irreps(irreps_hidden) + self.irreps_sh = o3.Irreps(irreps_sh) + self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps + self.num_layers = num_layers + self.edge_attr_dim = edge_attr_dim + self.num_attention_heads = num_attention_heads + self.reduce = reduce + self.sh = o3.SphericalHarmonics( + irreps_out=self.irreps_sh, normalize=True, normalization="component" + ) + # Split edge attribute dimensions: radial and temporal (bondedness is optional) + self.radial_edge_attr_dim = self.edge_attr_dim // 2 + self.temporal_edge_attr_dim = self.edge_attr_dim - self.radial_edge_attr_dim + + # Optional bondedness embedding (only used if bond_mask exists in graph) + self.embed_bondedness = nn.Embedding(2, self.edge_attr_dim // 3) + + # Gate for combining node attributes with temporal position + # Input: node_attr (from data) + temporal_position (1x0e scalar) + irreps_with_temporal = self.irreps_node_attr + o3.Irreps("1x0e") + self.temporal_gate = e3tools.nn.GateWrapper(irreps_in=irreps_with_temporal, \ + irreps_out=self.irreps_hidden, \ + irreps_gate=irreps_with_temporal,) + + self.layers = nn.ModuleList() + for _ in range(num_layers): + self.layers.append( + e3tools.nn.TransformerBlock( + irreps_in=self.irreps_hidden, + irreps_out=self.irreps_hidden, + irreps_sh=self.irreps_sh, + edge_attr_dim=self.edge_attr_dim, + num_heads=self.num_attention_heads, + ) + ) + self.output_head = e3tools.nn.EquivariantMLP( + irreps_in=self.irreps_hidden, + irreps_out=self.irreps_out, + irreps_hidden_list=[self.irreps_hidden], + ) + + def forward( + self, + node_attr: torch.Tensor, + temporal_graph: torch_geometric.data.Batch, + effective_radial_cutoff: float, + temporal_cutoff: float = 1.0, + ) -> torch.Tensor: + """Forward pass of the E3Transformer model.""" + # Extract graph data + pos = temporal_graph.pos + edge_index = temporal_graph.edge_index + temporal_position = temporal_graph.temporal_position + batch = temporal_graph.batch + num_graphs = temporal_graph.num_graphs + + src, dst = edge_index + edge_vec = pos[src] - pos[dst] + edge_sh = self.sh(edge_vec) + + # Compute edge attributes: radial and temporal + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec.norm(dim=1), + 0.0, + effective_radial_cutoff, + self.radial_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + + # Temporal edge attributes from temporal_position differences + temporal_edge_vec = temporal_position[src] - temporal_position[dst] + temporal_edge_attr = e3nn.math.soft_one_hot_linspace( + temporal_edge_vec.abs(), # Use absolute difference + 0.0, + temporal_cutoff, + self.temporal_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + + # Optional bondedness (if bond_mask exists in the temporal graph) + if hasattr(temporal_graph, 'bond_mask') and temporal_graph.bond_mask is not None: + bonded_edge_attr = self.embed_bondedness(temporal_graph.bond_mask) + edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr, temporal_edge_attr), dim=-1) + else: + edge_attr = torch.cat((radial_edge_attr, temporal_edge_attr), dim=-1) + + # Process node attributes with temporal gating + + # Concatenate node_attr with temporal_position (scalar) + temporal_position_expanded = temporal_position.unsqueeze(-1) # [N, 1] for concatenation + node_attr_with_temporal = torch.cat([node_attr, temporal_position_expanded], dim=-1) + + # Apply temporal gate + node_attr_processed = self.temporal_gate(node_attr_with_temporal) + + # Perform message passing with gated node attributes + for layer in self.layers: + node_attr_processed = layer(node_attr_processed, edge_index, edge_attr, edge_sh) + node_attr_processed = self.output_head(node_attr_processed) + + # Pool over nodes. + if self.reduce is not None: + node_attr_processed = e3tools.scatter( + node_attr_processed, + index=batch, + dim=0, + dim_size=num_graphs, + reduce=self.reduce, + ) + + return node_attr_processed + + +class E3SpatioTemporal(nn.Module): + """ + E(3)-equivariant spatio-temporal model that combines spatial and temporal processing. + + This model implements the complete workflow: + 1. Process input spatial graph and hidden states through spatial module + 2. Pool spatial features to temporal graph representation + 3. Process temporal graph through temporal module + 4. Pool temporal features back to spatial representation + 5. Convert temporal graph back to spatial graph + """ + + def __init__( + self, + spatial_module: nn.Module, + temporal_module: nn.Module, + spatial_to_temporal_pooler: nn.Module, + temporal_to_spatial_pooler: nn.Module, + radial_cutoff: float, + temporal_cutoff: float = 1.0, + ): + """ + Initialize the E3SpatioTemporal model. + + Args: + spatial_module: Module for processing spatial positions (e.g., E3Conv) + temporal_module: Module for processing temporal graphs (e.g., E3Transformer) + spatial_to_temporal_pooler: Module to convert spatial-temporal features to temporal node attributes + temporal_to_spatial_pooler: Module to convert temporal features back to spatial features + radial_cutoff: Cutoff for spatial radial edge weights + temporal_cutoff: Cutoff for temporal edge weights + """ + super().__init__() + + self.spatial_module = spatial_module + self.temporal_module = temporal_module + self.spatial_to_temporal_pooler = spatial_to_temporal_pooler + self.temporal_to_spatial_pooler = temporal_to_spatial_pooler + self.radial_cutoff = radial_cutoff + self.temporal_cutoff = temporal_cutoff + + def forward( + self, + batch: torch_geometric.data.Batch, + c_noise: torch.Tensor, + return_temporal_features: bool = False, + return_temporal_graph: bool = False, + ) -> Union[torch.Tensor, Dict[str, torch.Tensor]]: + """ + Forward pass implementing the complete spatio-temporal workflow. + + Args: + batch: Input spatial graph batch with pos, batch, num_graphs, and optionally hidden_state + c_noise: Noise conditioning tensor + return_temporal_features: Whether to return intermediate temporal features + return_temporal_graph: Whether to return the temporal graph + + Returns: + If return_temporal_features or return_temporal_graph is True, returns dict with: + - 'spatial_features': Final spatial features + - 'spatial_graph': Output spatial graph + - 'temporal_features': Temporal features (if requested) + - 'temporal_graph': Temporal graph (if requested) + Otherwise returns just the final spatial features tensor + """ + # Store original device + device = batch.pos.device + + # Step 1: Convert spatial graph to temporal graphs + temporal_batch = spatial_to_temporal_graphs(batch) + + # Step 2: Process all positions (current + hidden states) with spatial module + # Create topology for spatial processing (without positions) + topology = batch.clone() + # Remove position-dependent attributes but keep graph structure + if hasattr(topology, 'pos'): + del topology.pos + if hasattr(topology, 'batch'): + del topology.batch + if hasattr(topology, 'num_graphs'): + del topology.num_graphs + + node_attr_list = [] + + # Process current positions + node_attr_current = self.spatial_module( + pos=batch.pos, + topology=topology, + batch=batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff + ).unsqueeze(1) # [N, 1, features] + node_attr_list.append(node_attr_current) + + # Process hidden state positions if they exist + if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + for hidden_pos in batch.hidden_state: + node_attr_hidden = self.spatial_module( + pos=hidden_pos, + topology=topology, + batch=batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff + ).unsqueeze(1) # [N, 1, features] + node_attr_list.append(node_attr_hidden) + + # Step 3: Stack spatial-temporal features + node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) # [N, T, features] + + # Step 4: Convert spatial-temporal features to temporal node attributes + temporal_node_attr = self.spatial_to_temporal_pooler(node_attr_spatial_temporal, temporal_batch) + + # Step 5: Process temporal graph through temporal module + temporal_output = self.temporal_module( + temporal_node_attr, + temporal_batch, + self.radial_cutoff, + self.temporal_cutoff + ) + + # Step 6: Pool temporal features back to spatial features + spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) + + # Step 7: Convert temporal graph back to spatial graph + output_spatial_graph = temporal_to_spatial_graphs(temporal_batch) + + # Prepare return values + if return_temporal_features or return_temporal_graph: + result = { + 'spatial_features': spatial_features, + 'spatial_graph': output_spatial_graph, + } + if return_temporal_features: + result['temporal_features'] = temporal_output + if return_temporal_graph: + result['temporal_graph'] = temporal_batch + return result + else: + return spatial_features + + def get_spatial_output_irreps(self): + """Get the irreps of the spatial module output.""" + if hasattr(self.spatial_module, 'irreps_out'): + return self.spatial_module.irreps_out + else: + raise AttributeError("Spatial module does not have irreps_out attribute") + + def get_temporal_output_irreps(self): + """Get the irreps of the temporal module output.""" + if hasattr(self.temporal_module, 'irreps_out'): + return self.temporal_module.irreps_out + else: + raise AttributeError("Temporal module does not have irreps_out attribute") \ No newline at end of file diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 95e078f..92835f1 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -232,4 +232,103 @@ def forward(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.B conditioned_structures.pop(-1) conditioned_structures.append(x_clean.pos) - return conditioned_structures \ No newline at end of file + return conditioned_structures + + +class SpatioTemporalConditioner(pl.LightningModule): + """ + Conditioner that uses a spatio-temporal model to process hidden states. + + This conditioner takes the current positions and hidden states, processes them + through a spatio-temporal model, and returns a single conditioned structure. + Always returns exactly one structure regardless of N_structures parameter. + + By default, the spatiotemporal model is trainable. Set freeze_spatiotemporal_model=True + to freeze the parameters (e.g., when using a pretrained model). + """ + + def __init__( + self, + N_structures: int, + spatiotemporal_model: torch.nn.Module, + c_noise: float = 0.0, + freeze_spatiotemporal_model: bool = False, + **kwargs + ): + """ + Initialize the SpatioTemporalConditioner. + + Args: + N_structures: Number of structures parameter (ignored - this conditioner always returns 1 structure) + spatiotemporal_model: The E3SpatioTemporal model to use for processing + c_noise: Noise conditioning parameter + freeze_spatiotemporal_model: Whether to freeze spatiotemporal model parameters + **kwargs: Additional arguments passed to parent class + """ + super().__init__() + self.N_structures = N_structures + self.spatiotemporal_model = spatiotemporal_model + self.c_noise = c_noise + self.freeze_spatiotemporal_model = freeze_spatiotemporal_model + + # Only freeze parameters if explicitly requested + if self.freeze_spatiotemporal_model: + self.freeze_spatiotemporal_parameters() + # Set to evaluation mode when frozen + self.spatiotemporal_model.eval() + + def freeze_spatiotemporal_parameters(self): + """Freeze the spatiotemporal model parameters.""" + for param in self.spatiotemporal_model.parameters(): + param.requires_grad = False + + def unfreeze_spatiotemporal_parameters(self): + """Unfreeze the spatiotemporal model parameters.""" + for param in self.spatiotemporal_model.parameters(): + param.requires_grad = True + + def configure_for_inference(self): + """Configure the conditioner for inference (freeze parameters and set eval mode).""" + self.freeze_spatiotemporal_model = True + self.freeze_spatiotemporal_parameters() + self.spatiotemporal_model.eval() + + def configure_for_training(self): + """Configure the conditioner for training (unfreeze parameters and set train mode).""" + self.freeze_spatiotemporal_model = False + self.unfreeze_spatiotemporal_parameters() + self.spatiotemporal_model.train() + + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + """ + Forward pass that processes the batch through the spatio-temporal model. + + Args: + y: Batch containing current positions and hidden states + + Returns: + List containing a single conditioned structure tensor (always length 1) + """ + # Prepare noise conditioning + device = y.pos.device + sigma = torch.tensor(self.c_noise, device=device) + sigma = unsqueeze_trailing(sigma, 1) + + # Process through spatio-temporal model + # Only disable gradients if the model is frozen + if self.freeze_spatiotemporal_model: + with torch.no_grad(): + spatial_features = self.spatiotemporal_model(y, sigma) + else: + # Allow gradients to flow when training + spatial_features = self.spatiotemporal_model(y, sigma) + + # The spatiotemporal model returns spatial features, not positions + # We need to use these features to condition the structure + # For now, we'll use the current position as the conditioned structure + # In a more sophisticated implementation, we might use the features + # to modify the positions or use them in other ways + conditioned_position = y.pos + + # Return list with single conditioned structure (always length 1) + return [conditioned_position] \ No newline at end of file diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index 6724949..a1f8734 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -451,3 +451,154 @@ def configure_optimizers(self): } return out + + +class DenoiserWithInputAttr(Denoiser): + """ + Extension of Denoiser that works with E3ConvConditionalWithInputAttr. + + This subclass modifies the xhat_normalized method to extract spatiotemporal + features from the conditioner and pass them as input_attr to the architecture. + """ + + def __init__( + self, + arch: Callable[..., torch.nn.Module], + optim: Callable[..., torch.optim.Optimizer], + sigma_distribution: torch.distributions.Distribution, + max_radius: float, + average_squared_distance: float, + add_fixed_noise: bool, + add_fixed_ones: bool, + align_noisy_input_during_training: bool, + align_noisy_input_during_evaluation: bool, + mean_center: bool, + mirror_augmentation_rate: float, + bond_loss_coefficient: float = 1.0, + normalization_type: Optional[str] = "JAMUN", + sigma_data: Optional[float] = None, + lr_scheduler_config: Optional[Dict] = None, + use_torch_compile: bool = True, + torch_compile_kwargs: Optional[Dict] = None, + conditioner: Callable[..., list[torch.Tensor]] = None, + ): + """ + Initialize DenoiserWithInputAttr. + + Args: + All arguments are the same as the parent Denoiser class. + The conditioner should be a SpatioTemporalConditioner that provides + features that can be used as input_attr. + """ + super().__init__( + arch=arch, + optim=optim, + sigma_distribution=sigma_distribution, + max_radius=max_radius, + average_squared_distance=average_squared_distance, + add_fixed_noise=add_fixed_noise, + add_fixed_ones=add_fixed_ones, + align_noisy_input_during_training=align_noisy_input_during_training, + align_noisy_input_during_evaluation=align_noisy_input_during_evaluation, + mean_center=mean_center, + mirror_augmentation_rate=mirror_augmentation_rate, + bond_loss_coefficient=bond_loss_coefficient, + normalization_type=normalization_type, + sigma_data=sigma_data, + lr_scheduler_config=lr_scheduler_config, + use_torch_compile=use_torch_compile, + torch_compile_kwargs=torch_compile_kwargs, + conditioner=conditioner, + ) + + # Verify that we have a conditioner (required for this subclass) + if self.conditioning_module is None: + raise ValueError( + "DenoiserWithInputAttr requires a conditioner to provide input attributes" + ) + + def xhat_normalized( + self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] + ) -> torch_geometric.data.Batch: + """ + Compute the denoised prediction using the normalization factors from JAMUN. + + This version extracts spatiotemporal features from the conditioner and + passes them as input_attr to E3ConvConditionalWithInputAttr. + """ + sigma = torch.as_tensor(sigma).to(y.pos) + D = y.pos.shape[-1] + + # Compute the normalization factors. + with torch.cuda.nvtx.range("normalization_factors"): + c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) + radial_cutoff = self.effective_radial_cutoff(sigma) / c_in + + # Adjust dimensions. + c_in = unsqueeze_trailing(c_in, y.pos.ndim - 1) + c_skip = unsqueeze_trailing(c_skip, y.pos.ndim - 1) + c_out = unsqueeze_trailing(c_out, y.pos.ndim - 1) + c_noise = c_noise.unsqueeze(0) + + # Add edges to the graph. + with torch.cuda.nvtx.range("add_edges"): + y = self.add_edges(y, radial_cutoff) + + with torch.cuda.nvtx.range("scale_y"): + y_scaled = y.clone() + y_scaled.pos = y.pos * c_in + # Manually copy hidden state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + y_scaled.hidden_state = [] + for positions in y.hidden_state: + y_scaled.hidden_state.append(positions * c_in) + + with torch.cuda.nvtx.range("clone_y"): + xhat = y.clone() + # Manually copy hidden state + if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [h.clone() for h in y.hidden_state] + + # Extract spatiotemporal features from conditioner + with torch.cuda.nvtx.range("conditioning"): + conditioned_structures = self.conditioner(y_scaled) + # print(f"Conditioner is working, number of conditioned structures: {len(conditioned_structures)}") + + # Extract spatiotemporal features as input_attr + with torch.cuda.nvtx.range("extract_input_attr"): + # The conditioner should provide spatiotemporal features + # For SpatioTemporalConditioner, we need to extract the features from the model + if hasattr(self.conditioning_module, 'spatiotemporal_model'): + # Get spatiotemporal features by calling the model with return_spatial_features=True + spatiotemporal_result = self.conditioning_module.spatiotemporal_model( + y_scaled, + unsqueeze_trailing(torch.tensor(self.conditioning_module.c_noise, device=y.pos.device), 1) + ) + # Extract spatial features as input_attr + if isinstance(spatiotemporal_result, dict): + input_attr = spatiotemporal_result.get('spatial_features', None) + else: + # If not dict, assume the result is the spatial features + input_attr = spatiotemporal_result + else: + # Fallback: use zeros with correct shape and irreps + # This allows the model to work even without a proper spatiotemporal conditioner + py_logger = logging.getLogger("jamun") + py_logger.warning("No spatiotemporal_model found in conditioner, using zero input_attr") + # Assume input_attr_irreps is "3x1e" (9D vector) as specified in config + input_attr = torch.zeros(y.pos.shape[0], 9, device=y.pos.device, dtype=y.pos.dtype) + + # Call the architecture with input_attr + with torch.cuda.nvtx.range("g"): + g_pred = self.g( + torch.cat([*conditioned_structures], dim=-1), # Concatenated positions as first arg + topology=y_scaled, + c_noise=c_noise, + effective_radial_cutoff=radial_cutoff, + input_attr=input_attr # Pass spatiotemporal features as input_attr + ) + + xhat.pos = c_skip * y.pos + c_out * g_pred + if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] + return xhat diff --git a/src/jamun/model/pooling.py b/src/jamun/model/pooling.py new file mode 100644 index 0000000..b27dd53 --- /dev/null +++ b/src/jamun/model/pooling.py @@ -0,0 +1,217 @@ +#!/usr/bin/env python3 +""" +Lightning modules for converting node attributes between spatial and temporal representations. +""" + +import torch +import torch_geometric +import pytorch_lightning as pl + + +class SpatialToTemporalNodeAttr(pl.LightningModule): + """ + Lightning module to transfer node attributes from spatial nodes to temporal nodes + by repeating first temporal feature. + """ + + def __init__(self): + super().__init__() + + def forward(self, spatial_node_attr_temporal, temporal_batch): + """ + Transfer node attributes from spatial nodes to temporal nodes by repeating first temporal feature. + Takes the first temporal feature (t=0) and repeats it T times for each spatial node. + + Args: + spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs + + Returns: + torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] + """ + num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape + num_temporal_graphs = temporal_batch.num_graphs + + # Verify consistency + assert num_spatial_nodes == num_temporal_graphs, \ + f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" + + # Verify temporal length consistency + expected_temporal_nodes = temporal_batch.pos.shape[0] + expected_total_nodes = num_spatial_nodes * temporal_length + assert expected_total_nodes == expected_temporal_nodes, \ + f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" + + # Extract first temporal feature (t=0) and repeat it T times for each spatial node + first_temporal_features = spatial_node_attr_temporal[:, 0, :] # [N, attr_dim] + + # Repeat each spatial node's first temporal feature T times + temporal_node_attr = first_temporal_features.repeat_interleave(temporal_length, dim=0) # [N*T, attr_dim] + + # Verify the output shape matches the temporal batch + assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" + + return temporal_node_attr + + +class TemporalToSpatialNodeAttr(pl.LightningModule): + """ + Lightning module to convert temporal node attributes back to spatial node attributes. + Takes the first temporal node attribute from each temporal graph. + """ + + def __init__(self): + super().__init__() + + def forward(self, temporal_node_attr, temporal_batch): + """ + Convert temporal node attributes back to spatial node attributes. + Takes the first temporal node attribute from each temporal graph. + + Args: + temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs + + Returns: + torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] + """ + num_temporal_graphs = temporal_batch.num_graphs + attr_dim = temporal_node_attr.shape[1] + + # Extract the first node attribute from each temporal graph + spatial_node_attr = [] + + for graph_idx in range(num_temporal_graphs): + # Get the node range for this temporal graph + start_idx = temporal_batch.ptr[graph_idx] + + # The 0th node of each temporal graph is at the start of its range + first_node_attr = temporal_node_attr[start_idx] + spatial_node_attr.append(first_node_attr) + + # Stack to create spatial node attribute tensor + spatial_node_attr = torch.stack(spatial_node_attr) + + # Verify output shape + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" + + return spatial_node_attr + + +class TemporalToSpatialNodeAttrMean(pl.LightningModule): + """ + Lightning module to convert temporal node attributes back to spatial node attributes by averaging. + Takes the mean of all temporal node attributes for each temporal graph. + """ + + def __init__(self): + super().__init__() + + def forward(self, temporal_node_attr, temporal_batch): + """ + Convert temporal node attributes back to spatial node attributes by averaging. + Takes the mean of all temporal node attributes for each temporal graph. + + Args: + temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs + + Returns: + torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] + """ + num_temporal_graphs = temporal_batch.num_graphs + attr_dim = temporal_node_attr.shape[1] + + # Extract the mean node attributes from each temporal graph + spatial_node_attr = [] + + for graph_idx in range(num_temporal_graphs): + # Get the node range for this temporal graph + start_idx = temporal_batch.ptr[graph_idx] + end_idx = temporal_batch.ptr[graph_idx + 1] if graph_idx + 1 < len(temporal_batch.ptr) else len(temporal_node_attr) + + # Take the mean of all temporal nodes for this spatial node + temporal_nodes_attr = temporal_node_attr[start_idx:end_idx] # [temporal_length, attr_dim] + mean_node_attr = temporal_nodes_attr.mean(dim=0) # [attr_dim] + spatial_node_attr.append(mean_node_attr) + + # Stack to create spatial node attribute tensor + spatial_node_attr = torch.stack(spatial_node_attr) + + # Verify output shape + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" + + return spatial_node_attr + + +class SpatialTemporalToTemporalNodeAttr(pl.LightningModule): + """ + Lightning module to convert spatial node attributes arranged temporally to temporal node attributes. + Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. + """ + + def __init__(self): + super().__init__() + + def forward(self, spatial_node_attr_temporal, temporal_batch): + """ + Convert spatial node attributes arranged temporally to temporal node attributes. + Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. + + Args: + spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs for validation + + Returns: + torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] + """ + num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape + num_temporal_graphs = temporal_batch.num_graphs + + # Verify consistency with temporal batch + assert num_spatial_nodes == num_temporal_graphs, \ + f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" + + # Verify temporal length consistency + expected_temporal_nodes = temporal_batch.pos.shape[0] + expected_total_nodes = num_spatial_nodes * temporal_length + assert expected_total_nodes == expected_temporal_nodes, \ + f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" + + # Reshape to match temporal graph ordering: [N, T, features] -> [N*T, features] + # Temporal graph arranges nodes as: [node0_t0, node0_t1, ..., node0_tT-1, node1_t0, ...] + temporal_node_attr = spatial_node_attr_temporal.reshape(num_spatial_nodes * temporal_length, attr_dim) + + # Verify the output shape matches the temporal batch + assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" + + return temporal_node_attr + + +# Legacy function interfaces for backward compatibility +def spatial_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = SpatialToTemporalNodeAttr() + return module(spatial_node_attr_temporal, temporal_batch) + + +def temporal_to_spatial_node_attr(temporal_node_attr, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = TemporalToSpatialNodeAttr() + return module(temporal_node_attr, temporal_batch) + + +def temporal_to_spatial_node_attr_mean(temporal_node_attr, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = TemporalToSpatialNodeAttrMean() + return module(temporal_node_attr, temporal_batch) + + +def spatial_temporal_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = SpatialTemporalToTemporalNodeAttr() + return module(spatial_node_attr_temporal, temporal_batch) \ No newline at end of file diff --git a/src/jamun/utils/__init__.py b/src/jamun/utils/__init__.py index c433b9c..d38593f 100644 --- a/src/jamun/utils/__init__.py +++ b/src/jamun/utils/__init__.py @@ -1,6 +1,6 @@ from .align import align_A_to_B, align_A_to_B_batched, align_A_to_B_batched_f, find_rigid_alignment from .atom_graphs import to_atom_graphs -from .average_squared_distance import compute_average_squared_distance, compute_average_squared_distance_from_datasets +from .average_squared_distance import compute_average_squared_distance, compute_average_squared_distance_from_datasets, compute_temporal_average_squared_distance_from_dataset from .checkpoint import find_checkpoint, find_checkpoint_directory, get_run_path_for_wandb_run, get_wandb_run_config from .data_with_residue_info import DataWithResidueInformation from .dist_log import dist_log, wandb_dist_log @@ -12,6 +12,7 @@ get_side_chain_torsion_idxs, one_k_encoding, ) + from .mdtraj import coordinates_to_trajectories, save_pdb from .mean_center import mean_center, mean_center_f from .plot import animate_trajectory_with_py3Dmol, plot_molecules_with_py3Dmol diff --git a/src/jamun/utils/average_squared_distance.py b/src/jamun/utils/average_squared_distance.py index 43e8cd8..e609561 100644 --- a/src/jamun/utils/average_squared_distance.py +++ b/src/jamun/utils/average_squared_distance.py @@ -3,6 +3,7 @@ import numpy as np import torch +import torch_geometric from jamun import utils @@ -71,3 +72,53 @@ def compute_average_squared_distance_from_datasets( ) return float(mean_avg_sq_dist) + + +def compute_temporal_average_squared_distance_from_dataset( + dataset, + num_samples: int = 100, + verbose: bool = False +) -> float: + """ + Compute average squared distance between neighboring vertices in temporal graphs. + + Args: + dataset: Dataset containing spatial graphs with hidden states + num_samples: Number of samples to use for estimation + verbose: Whether to print verbose output + + Returns: + float: Average squared distance between temporal neighbors + """ + from jamun.model.arch.spatiotemporal import spatial_to_temporal_graphs + + avg_sq_dists = [] + num_graphs = 0 + + # Follow pattern from existing functions in this module + for item in dataset: + if num_graphs >= num_samples: + break + for graph in item: + if num_graphs >= num_samples: + break + # Convert to temporal graphs + temporal_batch = spatial_to_temporal_graphs(graph) + temporal_graphs = torch_geometric.data.Batch.to_data_list(temporal_batch) + graph_mean = 0.0 + num_nodes = graph.pos.shape[0] + for temporal_graph in temporal_graphs: + avg_sq_dist = compute_average_squared_distance(temporal_graph.pos, cutoff=None) + graph_mean += avg_sq_dist / num_nodes + avg_sq_dists.append(graph_mean) + num_graphs += 1 + mean_avg_sq_dist = sum(avg_sq_dists) / num_graphs + + + if verbose: + print(f"Total graphs processed: {num_graphs}") + print(f"Total temporal graphs processed: {len(avg_sq_dists)}") + print(f"Mean average squared distance between temporal nodes: {mean_avg_sq_dist:.6f}") + print(f"Standard deviation: {np.std(avg_sq_dists):.6f}") + + return float(mean_avg_sq_dist) diff --git a/test_spatiotemporal_conditioner.py b/test_spatiotemporal_conditioner.py new file mode 100755 index 0000000..94bef75 --- /dev/null +++ b/test_spatiotemporal_conditioner.py @@ -0,0 +1,394 @@ +#!/usr/bin/env python3 +""" +Test script for loading and testing a conditional denoiser with spatiotemporal conditioner. +Uses the ALA_ALA enhanced sampling dataset for testing. +""" + +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) + +import torch +import torch_geometric +from typing import Dict, Any +import sys +import os + +# Add the src directory to path to import jamun modules +sys.path.insert(0, 'src') + +from jamun.data import parse_datasets_from_directory +from jamun.model.denoiser_conditional import DenoiserWithInputAttr +from jamun.model.conditioners.conditioners import SpatioTemporalConditioner +from jamun.model.arch.spatiotemporal import E3SpatioTemporal, E3Transformer +from jamun.model.arch.e3conv import E3Conv +from jamun.model.arch.e3conv_conditional import E3ConvConditionalWithInputAttr +from jamun.model.pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean +from jamun.distributions._distributions import ConstantSigma +from jamun.utils import unsqueeze_trailing + +# Setup device +device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +print(f"Using device: {device}") + +def create_spatial_module() -> E3Conv: + """Create E3Conv spatial module with reasonable parameters.""" + import functools + import e3tools + + # Create factory functions + hidden_layer_factory = functools.partial( + e3tools.nn.ConvBlock, + conv=functools.partial(e3tools.nn.Conv) + ) + + output_head_factory = functools.partial( + e3tools.nn.EquivariantMLP, + irreps_hidden_list=["120x0e + 32x1e"] + ) + + return E3Conv( + irreps_out="1x1e", + irreps_hidden="120x0e + 32x1e", + irreps_sh="1x0e + 1x1e", + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + n_layers=1, + edge_attr_dim=64, + use_residue_information=True, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=32, + residue_index_embedding_dim=8, + use_residue_sequence_index=False, + num_atom_types=20, + max_sequence_length=10, + num_atom_codes=10, + num_residue_types=25, + test_equivariance=False, + reduce=None + ) + +def create_temporal_module() -> E3Transformer: + """Create E3Transformer temporal module.""" + return E3Transformer( + irreps_out="3x1e", + irreps_hidden="8x0e + 4x1e", + irreps_sh="1x0e + 1x1e", + irreps_node_attr="1x1e", # Match spatial module output + num_layers=2, + edge_attr_dim=24, + num_attention_heads=1, + reduce=None + ) + +def create_spatiotemporal_model() -> E3SpatioTemporal: + """Create the complete E3SpatioTemporal model.""" + spatial_module = create_spatial_module() + temporal_module = create_temporal_module() + + # Create pooling modules + spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr() + temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean() + + return E3SpatioTemporal( + spatial_module=spatial_module, + temporal_module=temporal_module, + spatial_to_temporal_pooler=spatial_to_temporal_pooler, + temporal_to_spatial_pooler=temporal_to_spatial_pooler, + radial_cutoff=0.05, + temporal_cutoff=1.0 + ) + +def create_spatiotemporal_conditioner() -> SpatioTemporalConditioner: + """Create SpatioTemporalConditioner with E3SpatioTemporal model.""" + spatiotemporal_model = create_spatiotemporal_model() + + return SpatioTemporalConditioner( + N_structures=1, + spatiotemporal_model=spatiotemporal_model, + c_noise=0.0, + freeze_spatiotemporal_model=False # Keep trainable + ) + +def create_conditional_denoiser_config() -> Dict[str, Any]: + """Create configuration for DenoiserWithInputAttr with spatiotemporal conditioner.""" + import functools + import e3tools.nn + + def create_arch(): + """Create the E3ConvConditionalWithInputAttr architecture module.""" + # Hidden layer factory + hidden_layer_factory = functools.partial( + e3tools.nn.ConvBlock, + conv=functools.partial(e3tools.nn.Conv) + ) + + # Output head factory + output_head_factory = functools.partial( + e3tools.nn.EquivariantMLP, + irreps_hidden_list=["16x0e + 8x1e"] + ) + + return E3ConvConditionalWithInputAttr( + irreps_out="3x1e", + irreps_hidden="16x0e + 8x1e", + irreps_sh="1x0e + 1x1e", + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + n_layers=2, + edge_attr_dim=32, + use_residue_information=True, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=16, + residue_index_embedding_dim=8, + use_residue_sequence_index=False, + num_atom_types=20, + max_sequence_length=10, + num_atom_codes=10, + num_residue_types=25, + test_equivariance=False, + reduce=None, + N_structures=1, + input_attr_irreps="3x1e", # Accept 3D vector input attributes from spatiotemporal model + ) + + def create_optim(params): + """Create the optimizer.""" + return torch.optim.Adam(params, lr=0.001) + + return { + # Required DenoiserWithInputAttr parameters + 'arch': create_arch, + 'optim': create_optim, + 'sigma_distribution': ConstantSigma(sigma=0.1), + 'max_radius': 1000.0, + 'average_squared_distance': 10.0, # Dummy value for testing + 'add_fixed_noise': False, + 'add_fixed_ones': False, + 'align_noisy_input_during_training': True, + 'align_noisy_input_during_evaluation': True, + 'mean_center': True, + 'mirror_augmentation_rate': 0.0, + 'bond_loss_coefficient': 1.0, + 'normalization_type': "JAMUN", + 'sigma_data': None, + 'lr_scheduler_config': None, + 'use_torch_compile': False, # Disable for testing + 'torch_compile_kwargs': None, + 'conditioner': create_spatiotemporal_conditioner() + } + +def add_edges_to_batch(batch: torch_geometric.data.Batch, cutoff: float = 0.05) -> torch_geometric.data.Batch: + """Add edges to batch using existing utility from denoiser.""" + # Use e3tools radius_graph directly since we don't need the full denoiser add_edges logic + import e3tools + + if hasattr(batch, 'edge_index') and batch.edge_index is not None: + return batch + + # Add radius-based edges + edge_index = e3tools.radius_graph( + x=batch.pos, + r=cutoff, + batch=batch.batch, + loop=False, + max_num_neighbors=32 + ) + batch.edge_index = edge_index + + # Add bonded edges if they exist + if hasattr(batch, 'bonded_edge_index') and batch.bonded_edge_index is not None: + bond_mask = torch.cat([ + torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device), + torch.ones(batch.bonded_edge_index.shape[1], dtype=torch.long, device=batch.pos.device) + ]) + batch.edge_index = torch.cat([edge_index, batch.bonded_edge_index], dim=1) + batch.bond_mask = bond_mask + else: + batch.bond_mask = torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device) + + return batch + +def load_test_data(): + """Load ALA_ALA test dataset.""" + print("Loading ALA_ALA dataset...") + + dataset = parse_datasets_from_directory( + root="/data2/sules/ALA_ALA_enhanced_full_grid/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + subsample=1, + total_lag_time=5, + lag_subsample_rate=1, + max_datasets=2, # Keep small for testing + num_frames=5 # Small number of frames + ) + + print(f"Loaded dataset with {len(dataset)} samples") + + # Get a sample and create batch + graph = dataset[0].__getitem__(0) + batch = torch_geometric.data.Batch.from_data_list([graph]) + + # Add edges + batch = add_edges_to_batch(batch, cutoff=0.05) + + # Move to device + batch = batch.to(device) + + print(f"Batch info:") + print(f" - pos shape: {batch.pos.shape}") + print(f" - edge_index shape: {batch.edge_index.shape}") + print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") + if hasattr(batch, 'hidden_state') and batch.hidden_state: + print(f" - hidden_state[0] shape: {batch.hidden_state[0].shape}") + + return batch + +def test_spatiotemporal_conditioner(conditioner: SpatioTemporalConditioner, batch: torch_geometric.data.Batch): + """Test the spatiotemporal conditioner.""" + print("\n" + "="*50) + print("TESTING SPATIOTEMPORAL CONDITIONER") + print("="*50) + + try: + # Test forward pass + conditioned_structures = conditioner(batch) + + print(f"āœ… Conditioner forward pass successful!") + print(f"Number of conditioned structures: {len(conditioned_structures)} (expected: 1)") + print(f"Conditioned structure shape: {conditioned_structures[0].shape}") + print(f"Original position shape: {batch.pos.shape}") + print(f"Position difference norm: {torch.norm(conditioned_structures[0] - batch.pos):.6f}") + + # Verify we got exactly one structure + assert len(conditioned_structures) == 1, f"Expected 1 structure, got {len(conditioned_structures)}" + + return True, conditioned_structures + + except Exception as e: + print(f"āŒ Conditioner test failed: {e}") + import traceback + traceback.print_exc() + return False, None + +def test_conditional_denoiser_creation(): + """Test creating DenoiserWithInputAttr with spatiotemporal conditioner.""" + print("\n" + "="*50) + print("TESTING DENOISER WITH INPUT ATTR CREATION") + print("="*50) + + try: + # Create configuration + config = create_conditional_denoiser_config() + + # Create denoiser (this will instantiate all components) + denoiser = DenoiserWithInputAttr(**config) + denoiser = denoiser.to(device) + + print(f"āœ… DenoiserWithInputAttr created successfully!") + print(f"Denoiser device: {next(denoiser.parameters()).device}") + print(f"Has conditioner: {hasattr(denoiser, 'conditioning_module')}") + print(f"Architecture type: {type(denoiser.g).__name__}") + print(f"Conditioner type: {type(denoiser.conditioning_module).__name__}") + + # Check if spatiotemporal model is properly set up + if hasattr(denoiser.conditioning_module, 'spatiotemporal_model'): + st_model = denoiser.conditioning_module.spatiotemporal_model + print(f"SpatioTemporal model type: {type(st_model).__name__}") + print(f"Spatial module type: {type(st_model.spatial_module).__name__}") + print(f"Temporal module type: {type(st_model.temporal_module).__name__}") + + return True, denoiser + + except Exception as e: + print(f"āŒ DenoiserWithInputAttr creation failed: {e}") + import traceback + traceback.print_exc() + return False, None + +def test_denoiser_forward_pass(denoiser: DenoiserWithInputAttr, batch: torch_geometric.data.Batch): + """Test the complete denoiser forward pass.""" + print("\n" + "="*50) + print("TESTING DENOISER WITH INPUT ATTR FORWARD PASS") + print("="*50) + + try: + # Test with sigma = 0.1 + sigma = 0.1 + print(f"Testing with sigma = {sigma}") + + # Forward pass - this should now extract spatiotemporal features and use them as input_attr + with torch.no_grad(): + xhat_batch = denoiser.xhat(batch, sigma) + + print(f"āœ… DenoiserWithInputAttr forward pass successful!") + print(f"Input shape: {batch.pos.shape}") + print(f"Output shape: {xhat_batch.pos.shape}") + print(f"Position difference norm: {torch.norm(xhat_batch.pos - batch.pos):.6f}") + + # Check that the spatiotemporal features were properly extracted and used + print(f"āœ… Spatiotemporal features successfully integrated as input_attr!") + + return True + + except Exception as e: + print(f"āŒ DenoiserWithInputAttr forward pass failed: {e}") + import traceback + traceback.print_exc() + return False + +def main(): + """Main test function.""" + print("="*60) + print("CONDITIONAL DENOISER WITH SPATIOTEMPORAL CONDITIONER TEST") + print("="*60) + + # Load test data + batch = load_test_data() + + # Test conditioner creation and forward pass + conditioner = create_spatiotemporal_conditioner() + conditioner = conditioner.to(device) + + conditioner_success, conditioned_structures = test_spatiotemporal_conditioner(conditioner, batch) + + if not conditioner_success: + print("āŒ Conditioner test failed, stopping here.") + return + + # Test complete denoiser creation + denoiser_success, denoiser = test_conditional_denoiser_creation() + + if not denoiser_success: + print("āŒ Denoiser creation failed, stopping here.") + return + + # Test complete forward pass + forward_success = test_denoiser_forward_pass(denoiser, batch) + + # Final summary + print("\n" + "="*60) + print("FINAL SUMMARY") + print("="*60) + + print(f"Test Results:") + print(f" - Conditioner test: {'āœ… PASSED' if conditioner_success else 'āŒ FAILED'}") + print(f" - Denoiser creation: {'āœ… PASSED' if denoiser_success else 'āŒ FAILED'}") + print(f" - Forward pass test: {'āœ… PASSED' if forward_success else 'āŒ FAILED'}") + + if conditioner_success and denoiser_success and forward_success: + print("\nšŸŽ‰ ALL TESTS PASSED!") + print("The conditional denoiser with spatiotemporal conditioner is working correctly!") + else: + print("\nāš ļø Some tests failed. Check the output above for details.") + + # Device memory summary + if torch.cuda.is_available(): + print(f"\nCUDA Memory:") + print(f" - Allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") + print(f" - Cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") + +if __name__ == "__main__": + main() \ No newline at end of file From 73ff35db90c0e25b74e6260027973bd61b9367fa Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Wed, 30 Jul 2025 21:33:56 +0000 Subject: [PATCH 18/32] added spatiotemporal transformer with configs --- .../ala_ala_denoiser_experiment_model3.yaml | 2 +- ...a_ala_denoiser_experiment_spike_check.yaml | 73 ++++++++ ...sample_ala_ala_enhanced_sampling_data.yaml | 33 ++-- ...n_enhanced_spatiotemporal_conditioner.yaml | 28 ++- ...otemporal_conditioner_with_input_attr.yaml | 55 ------ debug_model3_simple.sbatch | 56 ++++++ run_model3_beta_comparison.sbatch | 57 +++++++ run_model3_beta_comparison.sh | 54 ++++++ run_model3_beta_simple.sbatch | 55 ++++++ src/jamun/cmdline/train.py | 78 ++++++++- .../e3conv_conditional_spatiotemporal.yaml | 32 ++++ .../e3conv_conditional_with_input_attr.yaml | 44 ----- .../model/arch/spatiotemporal.yaml | 24 ++- .../denoised.yaml | 0 .../{conditioners => conditioner}/mean.yaml | 0 .../position.yaml | 0 .../{conditioners => conditioner}/self.yaml | 0 .../spatiotemporal.yaml | 38 +++-- .../spatiotemporal_pretrained.yaml | 0 .../{conditioners => conditioner}/spiked.yaml | 0 .../spatiotemporal_with_input_attr.yaml | 53 ------ .../model/denoiser_conditional.yaml | 2 +- .../denoiser_conditional_spatiotemporal.yaml | 31 ++++ .../denoiser_conditional_with_input_attr.yaml | 26 --- .../denoiser_spatiotemporal_conditioner.yaml | 26 --- src/jamun/model/arch/e3conv_conditional.py | 161 ++++++++++++++++++ src/jamun/model/conditioners/conditioners.py | 44 +++-- src/jamun/model/denoiser_conditional.py | 151 ---------------- .../sampling/walkjump/_single_measurement.py | 1 + src/jamun/utils/__init__.py | 2 +- src/jamun/utils/average_squared_distance.py | 8 +- test_spatiotemporal_conditioner.py | 119 ++++++++----- 32 files changed, 783 insertions(+), 470 deletions(-) create mode 100644 configs/experiment/ala_ala_denoiser_experiment_spike_check.yaml delete mode 100644 configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml create mode 100644 debug_model3_simple.sbatch create mode 100644 run_model3_beta_comparison.sbatch create mode 100644 run_model3_beta_comparison.sh create mode 100644 run_model3_beta_simple.sbatch create mode 100644 src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml delete mode 100644 src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml rename src/jamun/hydra_config/model/{conditioners => conditioner}/denoised.yaml (100%) rename src/jamun/hydra_config/model/{conditioners => conditioner}/mean.yaml (100%) rename src/jamun/hydra_config/model/{conditioners => conditioner}/position.yaml (100%) rename src/jamun/hydra_config/model/{conditioners => conditioner}/self.yaml (100%) rename src/jamun/hydra_config/model/{conditioners => conditioner}/spatiotemporal.yaml (60%) rename src/jamun/hydra_config/model/{conditioners => conditioner}/spatiotemporal_pretrained.yaml (100%) rename src/jamun/hydra_config/model/{conditioners => conditioner}/spiked.yaml (100%) delete mode 100644 src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml create mode 100644 src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml delete mode 100644 src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml delete mode 100644 src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml diff --git a/configs/experiment/ala_ala_denoiser_experiment_model3.yaml b/configs/experiment/ala_ala_denoiser_experiment_model3.yaml index 05a6220..1f92fa5 100644 --- a/configs/experiment/ala_ala_denoiser_experiment_model3.yaml +++ b/configs/experiment/ala_ala_denoiser_experiment_model3.yaml @@ -64,7 +64,7 @@ model: N_structures: ${model.arch.N_structures} trainer: - max_epochs: 100 + max_epochs: 200 logger: wandb: diff --git a/configs/experiment/ala_ala_denoiser_experiment_spike_check.yaml b/configs/experiment/ala_ala_denoiser_experiment_spike_check.yaml new file mode 100644 index 0000000..b99df8a --- /dev/null +++ b/configs/experiment/ala_ala_denoiser_experiment_spike_check.yaml @@ -0,0 +1,73 @@ +# @package _global_ +# Model 3: Denoiser with position conditioner, noise level sigma, +# but hidden states are repeated copies of y.pos (noise added by denoiser) + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: 2 # Two structures: current + 1 copy + lag_subsample_rate: 1 + num_frames: 10000 + + val: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + num_frames: 100 + + test: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 10 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 # Base sigma level for denoising + arch: + n_layers: 2 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + +trainer: + max_epochs: 100 + +logger: + wandb: + group: ALA_ALA_noise_check_spike_check + notes: "Model 3: Denoiser with PositionConditioner, 2 structures, sigma=0.04" + tags: ["noise_check", "high noise", "non-identical, i.i.d. measurements", "spike check"] \ No newline at end of file diff --git a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml b/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml index 5b599a9..4cc446a 100644 --- a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml +++ b/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml @@ -1,9 +1,22 @@ # @package _global_ defaults: - # - override /model: denoiser_conditional_pretrained.yaml + - override /model: denoiser_conditional_pretrained.yaml - override /callbacks: null +# init_datasets: +# _target_: jamun.data.parse_datasets_from_directory +# root: "${paths.data_path}/capped_diamines/timewarp_splits/train" +# traj_pattern: "^(.*).xtc" +# pdb_pattern: "^(.*).pdb" +# filter_codes: ['ALA_ALA'] +# as_iterable: false +# subsample: 10 +# total_lag_time: 5 +# lag_subsample_rate: 1 +# max_datasets: 1 +# label_override: "ALA_ALA" + init_datasets: _target_: jamun.data.parse_datasets_from_directory root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" @@ -11,27 +24,27 @@ init_datasets: pdb_pattern: "^(.*).pdb" as_iterable: false subsample: 1 - total_lag_time: 2 + total_lag_time: 10 lag_subsample_rate: 1 max_datasets: 10 label_override: "ALA_ALA" -num_sampling_steps_per_batch: 10000 -num_batches: 1 -dnum_init_samples_per_dataset: 10 +num_sampling_steps_per_batch: 100 +num_batches: 10 +num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/l8jwx7mx +wandb_train_run_path: sule-shashank/jamun/tw4vjtqo checkpoint_type: last sigma: 0.04 M: 1.0 -delta: ${sigma} +delta: 0.04 friction: 1.0 inverse_temperature: 1.0 -score_fn_clip: null +score_fn_clip: 100.0 sampler: _target_: jamun.sampling.SamplerMemory @@ -46,7 +59,7 @@ eval_dataset: filter_codes: ['ALA_ALA'] as_iterable: false subsample: 10 - total_lag_time: 2 + total_lag_time: 5 lag_subsample_rate: 1 max_datasets: 1 label_override: "ALA_ALA" @@ -84,4 +97,4 @@ logger: wandb: group: sample_enhanced_sampling_data notes: "Sampling from enhanced sampling data using memory sampler" - tags: ["sample", "enhanced_sampling", "memory_sampler", "ALA_ALA"] \ No newline at end of file + tags: ["sample", "enhanced_sampling", "memory_sampler", "ALA_ALA", "spatiotemporal transformer", "mean temporal pooler"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml index 7a9842f..17e7043 100644 --- a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml @@ -1,12 +1,10 @@ # @package _global_ -# Training SpatioTemporalConditioner on enhanced sampling data - defaults: - - override /model: denoiser_spatiotemporal_conditioner - - override /callbacks: - - timing.yaml - - lr_monitor.yaml - - model_checkpoint.yaml + - override /model: denoiser_conditional_spatiotemporal + # - override /callbacks: + # - timing.yaml + # - lr_monitor.yaml + # - model_checkpoint.yaml data: datamodule: @@ -20,7 +18,7 @@ data: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 2 # Increased for more training data + # max_datasets: 2 # Increased for more training data val: _target_: jamun.data.parse_datasets_from_directory @@ -30,23 +28,23 @@ data: subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} - max_datasets: 1 + # max_datasets: 1 model: sigma_distribution: _target_: jamun.distributions.ConstantSigma sigma: 0.04 arch: - n_layers: 2 - max_radius: 1000.0 + n_layers: 1 + max_radius: 10.0 optim: - lr: 0.001 # Slightly reduced learning rate for stability + lr: 0.002 # Slightly reduced learning rate for stability trainer: val_check_interval: 0.5 - max_epochs: 2 # Increased due to model complexity - devices: 1 - gradient_clip_val: 1.0 # Add gradient clipping for stability + max_epochs: 50 # Increased due to model complexity + # devices: 1 + # gradient_clip_val: 1.0 # Add gradient clipping for stability logger: wandb: diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml deleted file mode 100644 index 807e515..0000000 --- a/configs/experiment/train_enhanced_spatiotemporal_conditioner_with_input_attr.yaml +++ /dev/null @@ -1,55 +0,0 @@ -# @package _global_ -# Training SpatioTemporalConditioner with E3ConvConditionalWithInputAttr on enhanced sampling data - -defaults: - - override /model: denoiser_conditional_with_input_attr - - override /callbacks: - - timing.yaml - - lr_monitor.yaml - - model_checkpoint.yaml - -data: - datamodule: - batch_size: 16 # Reduced batch size due to increased model complexity - datasets: - train: - _target_: jamun.data.parse_datasets_from_directory - root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" - traj_pattern: "^(.*).xtc" - pdb_pattern: "^(.*).pdb" - subsample: 1 - total_lag_time: 5 - lag_subsample_rate: 1 - max_datasets: 2 # Increased for more training data - - val: - _target_: jamun.data.parse_datasets_from_directory - root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" - traj_pattern: "^(.*).xtc" - pdb_pattern: "^(.*).pdb" - subsample: 1 - total_lag_time: ${data.datamodule.datasets.train.total_lag_time} - lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} - max_datasets: ${data.datamodule.datasets.train.max_datasets} - -model: - sigma_distribution: - _target_: jamun.distributions.ConstantSigma - sigma: 0.04 - arch: - n_layers: 2 - N_structures: 1 # E3ConvConditionalWithInputAttr + SpatioTemporalConditioner - max_radius: 1000.0 - optim: - lr: 0.0008 # Slightly reduced learning rate for stability with input attributes - -trainer: - val_check_interval: 0.5 - max_epochs: 10 - devices: 1 - -logger: - wandb: - group: enhanced_sampling_spatiotemporal_with_input_attr - notes: "SpatioTemporalConditioner with E3ConvConditionalWithInputAttr on enhanced sampling data" - tags: ["spatiotemporal_conditioner", "enhanced_sampling", "input_attr", "e3conv_conditional"] \ No newline at end of file diff --git a/debug_model3_simple.sbatch b/debug_model3_simple.sbatch new file mode 100644 index 0000000..15f07d5 --- /dev/null +++ b/debug_model3_simple.sbatch @@ -0,0 +1,56 @@ +#!/usr/bin/env bash + +#SBATCH --job-name=debug_model3 +#SBATCH --array=1-2 +#SBATCH --output=logs/debug_model3_%A_%a.out +#SBATCH --error=logs/debug_model3_%A_%a.err +#SBATCH --time=1:00:00 +#SBATCH --partition=gpu +#SBATCH --gres=gpu:1 +#SBATCH --cpus-per-task=2 +#SBATCH --mem=16G + +mkdir -p logs + +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 + +cd /homefs/home/sules/jamun + +# Test basic command first +echo "Testing jamun_train command..." +jamun_train --help + +echo "Testing config validation..." +jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + --cfg job + +# Now test with minimal overrides +if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then + echo "Testing with betas=[0.9,0.9]" + jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + trainer.max_epochs=1 \ + data.datamodule.datasets.train.num_frames=10 \ + 'model.optim.betas=[0.9,0.9]' \ + --dry-run +else + echo "Testing with betas=[0.9,0.999]" + jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + trainer.max_epochs=1 \ + data.datamodule.datasets.train.num_frames=10 \ + 'model.optim.betas=[0.9,0.999]' \ + --dry-run +fi + +echo "Debug test completed for task $SLURM_ARRAY_TASK_ID" \ No newline at end of file diff --git a/run_model3_beta_comparison.sbatch b/run_model3_beta_comparison.sbatch new file mode 100644 index 0000000..43f5a78 --- /dev/null +++ b/run_model3_beta_comparison.sbatch @@ -0,0 +1,57 @@ +#!/usr/bin/env bash + +#SBATCH --job-name=model3_beta_comparison +#SBATCH --array=1-2 +#SBATCH --output=logs/model3_beta_%A_%a.out +#SBATCH --error=logs/model3_beta_%A_%a.err +#SBATCH --time=24:00:00 +#SBATCH --partition=gpu +#SBATCH --gres=gpu:1 +#SBATCH --cpus-per-task=4 +#SBATCH --mem=32G + +# Create logs directory if it doesn't exist +mkdir -p logs + +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 + +# Generate unique run key for this experiment +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +nvidia-smi + +# Navigate to project directory +cd /homefs/home/sules/jamun + +# Determine which beta configuration to run based on array task ID +if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then + # First run: Adam betas (0.9, 0.9) + echo "Running model3 experiment with Adam betas (0.9, 0.9)" + jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + ++run_key=$RUN_KEY \ + 'model.optim.betas=[0.9,0.9]' \ + ++logger.wandb.name="model3_beta09_09" \ + ++logger.wandb.tags="[${SLURM_JOB_ID},${RUN_KEY},model3,beta09_09]" +else + # Second run: Adam betas (0.9, 0.999) - PyTorch default + echo "Running model3 experiment with Adam betas (0.9, 0.999)" + jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + ++run_key=$RUN_KEY \ + 'model.optim.betas=[0.9,0.999]' \ + ++logger.wandb.name="model3_beta09_0999" \ + ++logger.wandb.tags="[${SLURM_JOB_ID},${RUN_KEY},model3,beta09_0999]" +fi + +echo "Model3 beta comparison experiment $SLURM_ARRAY_TASK_ID completed" \ No newline at end of file diff --git a/run_model3_beta_comparison.sh b/run_model3_beta_comparison.sh new file mode 100644 index 0000000..fc96d45 --- /dev/null +++ b/run_model3_beta_comparison.sh @@ -0,0 +1,54 @@ +#!/bin/bash +#SBATCH --job-name=model3_beta_comparison +#SBATCH --partition=gpu2 +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node +#SBATCH --gpus-per-node=1 # Assign one GPU to each agent +#SBATCH --cpus-per-task=12 +#SBATCH --time=1-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=1-2 + +# Create logs directory if it doesn't exist +mkdir -p logs + +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 + +# Generate unique run key for this experiment +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +nvidia-smi + +# Navigate to project directory +cd /homefs/home/sules/jamun + +# Determine which beta configuration to run based on array task ID +if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then + # First run: Adam betas (0.9, 0.9) + echo "Running model3 experiment with Adam betas (0.9, 0.9)" + jamun_train --config-dir=configs \ + experiment=ala_ala_denoiser_experiment_model3 \ + ++run_key=$RUN_KEY \ + '++model.optim.betas=[0.9,0.9]' \ + ++logger.wandb.name="model3_beta09_09" +else + # Second run: Adam betas (0.9, 0.999) - PyTorch default + echo "Running model3 experiment with Adam betas (0.9, 0.999)" + jamun_train --config-dir=configs \ + experiment=ala_ala_denoiser_experiment_model3 \ + ++run_key=$RUN_KEY \ + '++model.optim.betas=[0.9,0.999]' \ + ++logger.wandb.name="model3_beta09_0999" +fi + +echo "Model3 beta comparison experiment $SLURM_ARRAY_TASK_ID completed" \ No newline at end of file diff --git a/run_model3_beta_simple.sbatch b/run_model3_beta_simple.sbatch new file mode 100644 index 0000000..1d63dcb --- /dev/null +++ b/run_model3_beta_simple.sbatch @@ -0,0 +1,55 @@ +#!/usr/bin/env bash + +#SBATCH --job-name=model3_beta_simple +#SBATCH --array=1-2 +#SBATCH --output=logs/model3_simple_%A_%a.out +#SBATCH --error=logs/model3_simple_%A_%a.err +#SBATCH --time=24:00:00 +#SBATCH --partition=gpu +#SBATCH --gres=gpu:1 +#SBATCH --cpus-per-task=4 +#SBATCH --mem=32G + +# Create logs directory if it doesn't exist +mkdir -p logs + +eval "$(conda shell.bash hook)" +conda activate jamun + +set -eux + +echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" +echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" +echo "hostname = $(hostname)" + +export HYDRA_FULL_ERROR=1 + +# Generate unique run key for this experiment +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +nvidia-smi + +# Navigate to project directory +cd /homefs/home/sules/jamun + +# Determine which beta configuration to run based on array task ID +if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then + # First run: Adam betas (0.9, 0.9) + echo "Running model3 experiment with Adam betas (0.9, 0.9)" + jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + ++run_key=$RUN_KEY \ + 'model.optim.betas=[0.9,0.9]' \ + ++logger.wandb.name="model3_beta09_09" +else + # Second run: Adam betas (0.9, 0.999) - PyTorch default + echo "Running model3 experiment with Adam betas (0.9, 0.999)" + jamun_train --config-dir=src/jamun/hydra_config \ + experiment=ala_ala_denoiser_experiment_model3 \ + ++run_key=$RUN_KEY \ + 'model.optim.betas=[0.9,0.999]' \ + ++logger.wandb.name="model3_beta09_0999" +fi + +echo "Model3 beta comparison experiment $SLURM_ARRAY_TASK_ID completed" \ No newline at end of file diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index df9ffcf..5a2b74c 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -19,10 +19,53 @@ from jamun.hydra.utils import format_resolver # noqa: E402 from jamun.utils import compute_average_squared_distance_from_datasets, dist_log, find_checkpoint # noqa: E402 from jamun.utils._normalizations import normalization_factors # noqa: E402 +from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets # noqa: E402 +import math +from jamun.utils._normalizations import normalization_factors dotenv.load_dotenv(".env", verbose=True) OmegaConf.register_new_resolver("format", format_resolver) +def compute_radial_cutoff(max_radius: float, average_squared_distance: float, sigma: float, D: int = 3) -> float: + """ + Compute radial cutoff using the same formula as the denoiser. + + This replicates the computation from denoiser_conditional.py: + radial_cutoff = effective_radial_cutoff(sigma) / c_in + where: + - effective_radial_cutoff = sqrt(max_radiusĀ² + 6ĻƒĀ²) + - c_in = 1.0 / sqrt(average_squared_distance + 2DĻƒĀ²) + + Args: + max_radius: Maximum radius parameter + average_squared_distance: Average squared distance from dataset + sigma: Noise level + D: Dimensionality (default 3 for 3D coordinates) + + Returns: + Computed radial cutoff + """ + # Effective radial cutoff based on noise level + effective_radial_cutoff = math.sqrt(max_radius**2 + 6 * sigma**2) + + # JAMUN normalization factor c_in + A = average_squared_distance + B = 2 * D * sigma**2 + c_in = 1.0 / math.sqrt(A + B) + + # Final radial cutoff + radial_cutoff = effective_radial_cutoff / c_in + + print(f"Radial cutoff computation:") + print(f" max_radius: {max_radius}") + print(f" average_squared_distance: {average_squared_distance}") + print(f" sigma: {sigma}") + print(f" D: {D}") + print(f" effective_radial_cutoff: {effective_radial_cutoff}") + print(f" c_in: {c_in}") + print(f" final radial_cutoff: {radial_cutoff}") + + return radial_cutoff def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the average squared distance for normalization from the data.""" @@ -33,6 +76,18 @@ def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) return average_squared_distance +def compute_temporal_average_squared_distance_from_config(cfg: OmegaConf) -> float: + """Computes the temporal average squared distance for normalization from the data.""" + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + + average_squared_distance = compute_temporal_average_squared_distance_from_datasets( + train_datasets, + num_samples=100, # Use reasonable number of samples + verbose=True + ) + return average_squared_distance def run(cfg): log_cfg = OmegaConf.to_container(cfg, throw_on_missing=True, resolve=True) @@ -59,14 +114,15 @@ def run(cfg): ) cfg.model.average_squared_distance = average_squared_distance + sigma = cfg.model.sigma_distribution.sigma + average_squared_distance = cfg.model.average_squared_distance + c_in, c_skip, c_out, c_noise = normalization_factors(sigma, average_squared_distance) + c_in_float = float(c_in) + c_noise_float = float(c_noise) + # Compute normalization factors for conditioner c_in parameter if cfg.model.get("conditioner") and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.DenoisedConditioner": if hasattr(cfg.model.sigma_distribution, "sigma"): - sigma = cfg.model.sigma_distribution.sigma - average_squared_distance = cfg.model.average_squared_distance - c_in, c_skip, c_out, c_noise = normalization_factors(sigma, average_squared_distance) - c_in_float = float(c_in) - dist_log(f"Computing normalization factors for DenoisedConditioner with sigma={sigma}") dist_log(f" average_squared_distance: {average_squared_distance}") dist_log(f" c_in: {c_in_float}") @@ -77,6 +133,18 @@ def run(cfg): cfg.model.conditioner.c_in = c_in_float dist_log(f"Set cfg.model.conditioner.c_in to {c_in_float}") + if cfg.model.conditioner._target_ == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner": + cfg.model.conditioner.spatiotemporal_model.radial_cutoff = average_squared_distance + max_radius = cfg.model.max_radius + temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) + temporal_radial_cutoff = compute_radial_cutoff( + max_radius=max_radius, + average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model + sigma=sigma, + D=3) + cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff + cfg.model.conditioner.c_noise = c_noise_float + dist_log(f"Set cfg.model.conditioner.spatiotemporal_model.c_noise to {c_noise_float}") # # do this for the sweep # if cfg.model.N_measurements_hidden is not None: # dist_log(f"Number of hidden measurements: {cfg.model.N_measurements_hidden}") diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml new file mode 100644 index 0000000..96bcfec --- /dev/null +++ b/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml @@ -0,0 +1,32 @@ +_target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalSpatioTemporal +_partial_: true +irreps_out: "1x1e" # Output 3 components (1x1e) to match position +irreps_hidden: "120x0e + 32x1e" +irreps_sh: "1x0e + 1x1e" +n_layers: 1 +edge_attr_dim: 64 # Match spatiotemporal spatial module +atom_type_embedding_dim: 8 +atom_code_embedding_dim: 8 +residue_code_embedding_dim: 32 # Match spatiotemporal spatial module +residue_index_embedding_dim: 8 +use_residue_information: ${data.use_residue_information} +use_residue_sequence_index: false +num_atom_types: 20 +max_sequence_length: 10 +num_atom_codes: 10 +num_residue_types: 25 +test_equivariance: false +reduce: null +hidden_layer_factory: + _target_: e3tools.nn.ConvBlock + _partial_: true + conv: + _target_: e3tools.nn.Conv + _partial_: true +output_head_factory: + _target_: e3tools.nn.EquivariantMLP + _partial_: true + irreps_hidden_list: + - ${model.arch.irreps_hidden} +N_structures: 1 # For [y.pos, spatial_features] input +input_attr_irreps: "120x0e + 32x1e" # spatial_features (9 components = 3x1e) \ No newline at end of file diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml deleted file mode 100644 index d48182f..0000000 --- a/src/jamun/hydra_config/model/arch/e3conv_conditional_with_input_attr.yaml +++ /dev/null @@ -1,44 +0,0 @@ -# E3ConvConditionalWithInputAttr architecture configuration -# This extends E3ConvConditional to accept additional input attributes - -_target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalWithInputAttr - -# Standard E3ConvConditional parameters -irreps_out: "3x1e" -irreps_hidden: "120x0e + 32x1e" -irreps_sh: "1x0e + 1x1e" -n_layers: 2 -edge_attr_dim: 64 -use_residue_information: ${data.use_residue_information} -atom_type_embedding_dim: 8 -atom_code_embedding_dim: 8 -residue_code_embedding_dim: 32 -residue_index_embedding_dim: 8 -use_residue_sequence_index: false -num_atom_types: ${data.num_atom_types} -max_sequence_length: ${data.max_sequence_length} -num_atom_codes: ${data.num_atom_codes} -num_residue_types: ${data.num_residue_types} -test_equivariance: false -reduce: null -N_structures: ${model.arch.N_structures} - -# New parameter for input attributes -input_attr_irreps: "3x1e" # 3D vector input attributes (matching transformer output) - -# Factory functions for layers -hidden_layer_factory: - _target_: functools.partial - _args_: - - _target_: e3tools.nn.ConvBlock - conv: - _target_: functools.partial - _args_: - - _target_: e3tools.nn.Conv - -output_head_factory: - _target_: functools.partial - _args_: - - _target_: e3tools.nn.EquivariantMLP - irreps_hidden_list: - - "120x0e + 32x1e" \ No newline at end of file diff --git a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml index 109a67f..275efb6 100644 --- a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml @@ -2,29 +2,39 @@ _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal # Cutoff parameters -radial_cutoff: 0.05 -temporal_cutoff: 1.0 +radial_cutoff: 0.05 # radial cutoff for the spatial module +temporal_cutoff: 1.0 # radial cutoff for the temporal module # Spatial module (E3Conv) spatial_module: - _target_: jamun.model.arch.e3conv_conditional.E3Conv + _target_: jamun.model.arch.E3Conv + _partial_: true irreps_out: "1x1e" irreps_hidden: "120x0e + 32x1e" irreps_sh: "1x0e + 1x1e" - n_layers: 1 + n_layers: 5 edge_attr_dim: 64 - use_residue_information: true atom_type_embedding_dim: 8 atom_code_embedding_dim: 8 residue_code_embedding_dim: 32 residue_index_embedding_dim: 8 + use_residue_information: ${data.use_residue_information} use_residue_sequence_index: false num_atom_types: 20 max_sequence_length: 10 num_atom_codes: 10 num_residue_types: 25 - test_equivariance: false - reduce: null + hidden_layer_factory: + _target_: e3tools.nn.ConvBlock + _partial_: true + conv: + _target_: e3tools.nn.Conv + _partial_: true + output_head_factory: + _target_: e3tools.nn.EquivariantMLP + _partial_: true + irreps_hidden_list: + - ${model.arch.irreps_hidden} # Temporal module (E3Transformer) temporal_module: diff --git a/src/jamun/hydra_config/model/conditioners/denoised.yaml b/src/jamun/hydra_config/model/conditioner/denoised.yaml similarity index 100% rename from src/jamun/hydra_config/model/conditioners/denoised.yaml rename to src/jamun/hydra_config/model/conditioner/denoised.yaml diff --git a/src/jamun/hydra_config/model/conditioners/mean.yaml b/src/jamun/hydra_config/model/conditioner/mean.yaml similarity index 100% rename from src/jamun/hydra_config/model/conditioners/mean.yaml rename to src/jamun/hydra_config/model/conditioner/mean.yaml diff --git a/src/jamun/hydra_config/model/conditioners/position.yaml b/src/jamun/hydra_config/model/conditioner/position.yaml similarity index 100% rename from src/jamun/hydra_config/model/conditioners/position.yaml rename to src/jamun/hydra_config/model/conditioner/position.yaml diff --git a/src/jamun/hydra_config/model/conditioners/self.yaml b/src/jamun/hydra_config/model/conditioner/self.yaml similarity index 100% rename from src/jamun/hydra_config/model/conditioners/self.yaml rename to src/jamun/hydra_config/model/conditioner/self.yaml diff --git a/src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml similarity index 60% rename from src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml rename to src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml index 6ef8c49..9f3ad70 100644 --- a/src/jamun/hydra_config/model/conditioners/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml @@ -1,51 +1,59 @@ -# @package _global_ _target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner -N_structures: 1 +N_structures: 2 # Now returns [y.pos, spatial_features] c_noise: 0.0 -freeze_spatiotemporal_model: false # Trainable by default +freeze_spatiotemporal_model: false # Trainable by default # Spatiotemporal model configuration spatiotemporal_model: _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal radial_cutoff: 0.05 temporal_cutoff: 1.0 - + # Spatial module (E3Conv) spatial_module: _target_: jamun.model.arch.e3conv.E3Conv - irreps_out: "3x1e" + irreps_out: "1x1e" irreps_hidden: "120x0e + 32x1e" irreps_sh: "1x0e + 1x1e" n_layers: 1 edge_attr_dim: 64 - use_residue_information: true atom_type_embedding_dim: 8 atom_code_embedding_dim: 8 residue_code_embedding_dim: 32 residue_index_embedding_dim: 8 + use_residue_information: ${data.use_residue_information} use_residue_sequence_index: false num_atom_types: 20 max_sequence_length: 10 num_atom_codes: 10 num_residue_types: 25 - test_equivariance: false - reduce: null - + hidden_layer_factory: + _target_: e3tools.nn.ConvBlock + _partial_: true + conv: + _target_: e3tools.nn.Conv + _partial_: true + output_head_factory: + _target_: e3tools.nn.EquivariantMLP + _partial_: true + irreps_hidden_list: + - "120x0e + 32x1e" + # Temporal module (E3Transformer) temporal_module: _target_: jamun.model.arch.spatiotemporal.E3Transformer - irreps_out: "3x1e" - irreps_hidden: "8x0e + 4x1e" + irreps_out: "120x0e + 32x1e" # Final spatial features output + irreps_hidden: "120x0e + 32x1e" irreps_sh: "1x0e + 1x1e" - irreps_node_attr: "3x1e" # Match spatial module output + irreps_node_attr: "1x1e" # Match spatial module output num_layers: 2 edge_attr_dim: 24 num_attention_heads: 1 reduce: null - + # Pooling modules spatial_to_temporal_pooler: _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr - + temporal_to_spatial_pooler: - _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file + _target_: jamun.model.pooling.TemporalToSpatialNodeAttr \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioners/spatiotemporal_pretrained.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml similarity index 100% rename from src/jamun/hydra_config/model/conditioners/spatiotemporal_pretrained.yaml rename to src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml diff --git a/src/jamun/hydra_config/model/conditioners/spiked.yaml b/src/jamun/hydra_config/model/conditioner/spiked.yaml similarity index 100% rename from src/jamun/hydra_config/model/conditioners/spiked.yaml rename to src/jamun/hydra_config/model/conditioner/spiked.yaml diff --git a/src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml b/src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml deleted file mode 100644 index 81df378..0000000 --- a/src/jamun/hydra_config/model/conditioners/spatiotemporal_with_input_attr.yaml +++ /dev/null @@ -1,53 +0,0 @@ -# @package _global_ -# SpatioTemporalConditioner configuration that provides input attributes to E3ConvConditionalWithInputAttr - -_target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner -N_structures: 1 -c_noise: 0.0 -freeze_spatiotemporal_model: false # Trainable by default - -# Spatiotemporal model configuration -spatiotemporal_model: - _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal - radial_cutoff: 0.05 - temporal_cutoff: 1.0 - - # Spatial module (E3Conv) - outputs features that will be used as input_attr - spatial_module: - _target_: jamun.model.arch.e3conv.E3Conv - irreps_out: "3x1e" # Match input_attr_irreps in the main architecture - irreps_hidden: "120x0e + 32x1e" - irreps_sh: "1x0e + 1x1e" - n_layers: 1 - edge_attr_dim: 64 - use_residue_information: true - atom_type_embedding_dim: 8 - atom_code_embedding_dim: 8 - residue_code_embedding_dim: 32 - residue_index_embedding_dim: 8 - use_residue_sequence_index: false - num_atom_types: 20 - max_sequence_length: 10 - num_atom_codes: 10 - num_residue_types: 25 - test_equivariance: false - reduce: null - - # Temporal module (E3Transformer) - temporal_module: - _target_: jamun.model.arch.spatiotemporal.E3Transformer - irreps_out: "3x1e" # Match spatial module output - irreps_hidden: "8x0e + 4x1e" - irreps_sh: "1x0e + 1x1e" - irreps_node_attr: "3x1e" # Match spatial module output - num_layers: 2 - edge_attr_dim: 24 - num_attention_heads: 1 - reduce: null - - # Pooling modules - spatial_to_temporal_pooler: - _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr - - temporal_to_spatial_pooler: - _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_conditional.yaml b/src/jamun/hydra_config/model/denoiser_conditional.yaml index 3c8832d..d4369ee 100644 --- a/src/jamun/hydra_config/model/denoiser_conditional.yaml +++ b/src/jamun/hydra_config/model/denoiser_conditional.yaml @@ -2,7 +2,7 @@ defaults: - arch: e3conv_conditional.yaml - optim: adam.yaml - lr_scheduler_config: null - - conditioners: position.yaml # Default conditioner, can be overridden + - conditioner: position.yaml # Default conditioner, can be overridden - _self_ max_radius: null diff --git a/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml b/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml new file mode 100644 index 0000000..e4f18cd --- /dev/null +++ b/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml @@ -0,0 +1,31 @@ +defaults: + - arch: e3conv_conditional_spatiotemporal.yaml + - optim: adam.yaml + - conditioner: spatiotemporal.yaml + - _self_ + +# Model configuration +sigma_distribution: + _target_: jamun.model.sigma_distribution.ConstantSigma + sigma: 0.1 + +max_radius: 1000.0 +average_squared_distance: 10.0 +add_fixed_noise: false +add_fixed_ones: false +align_noisy_input_during_training: true +align_noisy_input_during_evaluation: true +mean_center: true +mirror_augmentation_rate: 0.0 +bond_loss_coefficient: 1.0 +normalization_type: "JAMUN" +sigma_data: null +lr_scheduler_config: null +use_torch_compile: false +torch_compile_kwargs: null + +# Override to ensure N_structures matches spatiotemporal conditioner output +arch: + N_structures: 1 # For [y.pos, spatial_features] + +_target_: jamun.model.denoiser_conditional.Denoiser diff --git a/src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml b/src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml deleted file mode 100644 index 848dfd7..0000000 --- a/src/jamun/hydra_config/model/denoiser_conditional_with_input_attr.yaml +++ /dev/null @@ -1,26 +0,0 @@ -defaults: - - arch: e3conv_conditional_with_input_attr.yaml - - optim: adam.yaml - - lr_scheduler_config: null - - conditioners: spatiotemporal.yaml - - _self_ - -max_radius: null -average_squared_distance: null -add_fixed_noise: false -add_fixed_ones: false -align_noisy_input_during_training: true -align_noisy_input_during_evaluation: true -mean_center: true -mirror_augmentation_rate: 0.0 -use_torch_compile: true -torch_compile_kwargs: - fullgraph: true - dynamic: true - mode: default - -# Architecture settings -arch: - N_structures: 1 # Compatible with SpatioTemporalConditioner - -_target_: jamun.model.denoiser_conditional.DenoiserWithInputAttr \ No newline at end of file diff --git a/src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml b/src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml deleted file mode 100644 index e3e84a3..0000000 --- a/src/jamun/hydra_config/model/denoiser_spatiotemporal_conditioner.yaml +++ /dev/null @@ -1,26 +0,0 @@ -defaults: - - arch: e3conv_conditional.yaml - - optim: adam.yaml - - lr_scheduler_config: null - - conditioners: spatiotemporal.yaml # Use spatiotemporal conditioner - - _self_ - -max_radius: null -average_squared_distance: null -add_fixed_noise: false -add_fixed_ones: false -align_noisy_input_during_training: true -align_noisy_input_during_evaluation: true -mean_center: true -mirror_augmentation_rate: 0.0 -use_torch_compile: true -torch_compile_kwargs: - fullgraph: true - dynamic: true - mode: default - -# Architecture settings for spatiotemporal conditioner -arch: - N_structures: 1 # SpatioTemporalConditioner always returns 1 structure - -_target_: jamun.model.denoiser_conditional.Denoiser \ No newline at end of file diff --git a/src/jamun/model/arch/e3conv_conditional.py b/src/jamun/model/arch/e3conv_conditional.py index 1173aa2..dcbc82d 100644 --- a/src/jamun/model/arch/e3conv_conditional.py +++ b/src/jamun/model/arch/e3conv_conditional.py @@ -307,3 +307,164 @@ def forward( node_attr = scatter(node_attr, topology.batch, dim=0, reduce=self.reduce) return node_attr + + +class E3ConvConditionalSpatioTemporal(E3ConvConditional): + """ + E3ConvConditional specifically designed for spatiotemporal conditioning. + + This class expects input positions to be concatenated as [y.pos, spatial_features] + where y.pos are the physical 3D coordinates and spatial_features are additional + attributes from the spatiotemporal model. + + Key differences from E3ConvConditional: + - Edge spherical harmonics are only computed for the first 3 coordinates (y.pos) + - Remaining coordinates are treated as per-node input attributes + - Input attributes are combined with computed node attributes + """ + + def __init__( + self, + irreps_out: str | Irreps, + irreps_hidden: str | Irreps, + irreps_sh: str | Irreps, + hidden_layer_factory: Callable[..., torch.nn.Module], + output_head_factory: Callable[..., torch.nn.Module], + use_residue_information: bool, + n_layers: int, + edge_attr_dim: int, + atom_type_embedding_dim: int, + atom_code_embedding_dim: int, + residue_code_embedding_dim: int, + residue_index_embedding_dim: int, + use_residue_sequence_index: bool, + num_atom_types: int = 20, + max_sequence_length: int = 10, + num_atom_codes: int = 10, + num_residue_types: int = 25, + test_equivariance: bool = False, + reduce: str | None = None, + N_structures: int = 1, # Should be 2 for [y.pos, spatial_features] + input_attr_irreps: str | Irreps = "3x1e", # Default for spatial features + ): + """ + Initialize E3ConvConditionalSpatioTemporal. + + Args: + input_attr_irreps: Irreps of the spatial features from spatiotemporal model. + Should match the irreps_out of the spatiotemporal model. + N_structures: Should be 2 for [y.pos, spatial_features] + All other args: Same as parent E3ConvConditional class. + """ + super().__init__( + irreps_out=irreps_out, + irreps_hidden=irreps_hidden, + irreps_sh=irreps_sh, + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + use_residue_information=use_residue_information, + n_layers=n_layers, + edge_attr_dim=edge_attr_dim, + atom_type_embedding_dim=atom_type_embedding_dim, + atom_code_embedding_dim=atom_code_embedding_dim, + residue_code_embedding_dim=residue_code_embedding_dim, + residue_index_embedding_dim=residue_index_embedding_dim, + use_residue_sequence_index=use_residue_sequence_index, + num_atom_types=num_atom_types, + max_sequence_length=max_sequence_length, + num_atom_codes=num_atom_codes, + num_residue_types=num_residue_types, + test_equivariance=test_equivariance, + reduce=reduce, + N_structures=N_structures, + ) + + # Set up input attribute handling + self.input_attr_irreps = o3.Irreps(input_attr_irreps) + + # Create input irrep aggregator to combine node_attr with input_attr + # Combined irreps: node_attr irreps + input_attr irreps + combined_irreps = self.irreps_hidden + self.input_attr_irreps + + # Create aggregator that takes combined input and outputs node_attr irreps + self.input_irrep_aggregator = e3tools.nn.EquivariantMLP( + irreps_in=combined_irreps, + irreps_out=self.irreps_hidden, + irreps_hidden_list=[self.irreps_hidden], # Single hidden layer + ) + + def forward( + self, + pos: Tensor, # should be [N, 3 + spatial_features_dim] from [y.pos, spatial_features] + topology: torch_geometric.data.Batch, + c_noise: Tensor, + effective_radial_cutoff: float, + ) -> Tensor: + """ + Forward pass with spatiotemporal conditioning. + + Args: + pos: Concatenated positions [y.pos, spatial_features] with shape [N, 3 + spatial_features_dim] + topology: Graph topology + c_noise: Noise conditioning + effective_radial_cutoff: Radial cutoff for edges + + Returns: + Node attributes after processing + """ + # Extract edge attributes. + edge_index = topology["edge_index"] + bond_mask = topology["bond_mask"] + + src, dst = edge_index + + # Split positions: first 3 coords are physical positions, rest are spatial features + pos_physical = pos[:, :3] # [N, 3] - physical coordinates + pos_features = pos[:, 3:] # [N, spatial_features_dim] - spatial features + + # Compute edge spherical harmonics ONLY for physical positions + edge_vec_physical = pos_physical[src] - pos_physical[dst] + edge_sh = self.sh(edge_vec_physical) + + # Compute edge attributes using physical positions + bonded_edge_attr = self.embed_bondedness(bond_mask) + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec_physical.norm(dim=1), + 0.0, + effective_radial_cutoff, + self.radial_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr), dim=-1) + + # Compute initial node attributes + node_attr = self.atom_embedder(topology) + node_attr = self.initial_noise_scaling(node_attr, c_noise) + node_attr = self.initial_projector(node_attr, edge_index, edge_attr, edge_sh) + + # Combine node_attr with spatial features (input_attr) + # Validate spatial features shape + expected_dim = self.input_attr_irreps.dim + if pos_features.shape[-1] != expected_dim: + raise ValueError( + f"Expected spatial features to have dimension {expected_dim}, " + f"but got {pos_features.shape[-1]}" + ) + + # Concatenate node_attr with spatial features + combined_attr = torch.cat([node_attr, pos_features], dim=-1) + + # Aggregate to get back to node_attr irreps + node_attr = self.input_irrep_aggregator(combined_attr) + + # Continue with normal processing using only physical positions for edge computations + for scaling, skip, layer in zip(self.noise_scalings, self.skip_connections, self.layers): + node_attr = skip(node_attr, layer(scaling(node_attr, c_noise), edge_index, edge_attr, edge_sh), c_noise) + node_attr = self.output_head(node_attr) + node_attr = node_attr * self.output_gain + + if self.reduce is not None: + node_attr = scatter(node_attr, topology.batch, dim=0, reduce=self.reduce) + + return node_attr diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 92835f1..460b025 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -240,8 +240,8 @@ class SpatioTemporalConditioner(pl.LightningModule): Conditioner that uses a spatio-temporal model to process hidden states. This conditioner takes the current positions and hidden states, processes them - through a spatio-temporal model, and returns a single conditioned structure. - Always returns exactly one structure regardless of N_structures parameter. + through a spatio-temporal model, and returns [y.pos, spatial_features]. + Always returns exactly 2 structures: the original positions and computed spatial features. By default, the spatiotemporal model is trainable. Set freeze_spatiotemporal_model=True to freeze the parameters (e.g., when using a pretrained model). @@ -307,28 +307,40 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: y: Batch containing current positions and hidden states Returns: - List containing a single conditioned structure tensor (always length 1) + List containing [y.pos, spatial_features] for concatenation by the denoiser """ + # Align hidden positions with current position before processing + if hasattr(y, 'hidden_state') and y.hidden_state is not None and len(y.hidden_state) > 0: + # Create a copy of the batch to avoid modifying the original + y_aligned = y.clone() + + # Align each hidden state to the current position + aligned_hidden_states = [] + for hidden_pos in y.hidden_state: + # Align hidden_pos to y.pos using Kabsch algorithm (same as PositionConditioner) + aligned_hidden_pos = kabsch_algorithm(hidden_pos, y.pos, y.batch, y.num_graphs) + aligned_hidden_states.append(aligned_hidden_pos) + + # Update the hidden states in the aligned batch + y_aligned.hidden_state = aligned_hidden_states + else: + # If no hidden states, use original batch + y_aligned = y + # Prepare noise conditioning - device = y.pos.device + device = y_aligned.pos.device sigma = torch.tensor(self.c_noise, device=device) sigma = unsqueeze_trailing(sigma, 1) - # Process through spatio-temporal model + # Process through spatio-temporal model with aligned hidden states # Only disable gradients if the model is frozen if self.freeze_spatiotemporal_model: with torch.no_grad(): - spatial_features = self.spatiotemporal_model(y, sigma) + spatial_features = self.spatiotemporal_model(y_aligned, sigma) else: # Allow gradients to flow when training - spatial_features = self.spatiotemporal_model(y, sigma) + spatial_features = self.spatiotemporal_model(y_aligned, sigma) - # The spatiotemporal model returns spatial features, not positions - # We need to use these features to condition the structure - # For now, we'll use the current position as the conditioned structure - # In a more sophisticated implementation, we might use the features - # to modify the positions or use them in other ways - conditioned_position = y.pos - - # Return list with single conditioned structure (always length 1) - return [conditioned_position] \ No newline at end of file + # Return list containing [y.pos, spatial_features] for concatenation + # The denoiser will concatenate these along the feature dimension + return [y.pos, spatial_features] \ No newline at end of file diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index a1f8734..6724949 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -451,154 +451,3 @@ def configure_optimizers(self): } return out - - -class DenoiserWithInputAttr(Denoiser): - """ - Extension of Denoiser that works with E3ConvConditionalWithInputAttr. - - This subclass modifies the xhat_normalized method to extract spatiotemporal - features from the conditioner and pass them as input_attr to the architecture. - """ - - def __init__( - self, - arch: Callable[..., torch.nn.Module], - optim: Callable[..., torch.optim.Optimizer], - sigma_distribution: torch.distributions.Distribution, - max_radius: float, - average_squared_distance: float, - add_fixed_noise: bool, - add_fixed_ones: bool, - align_noisy_input_during_training: bool, - align_noisy_input_during_evaluation: bool, - mean_center: bool, - mirror_augmentation_rate: float, - bond_loss_coefficient: float = 1.0, - normalization_type: Optional[str] = "JAMUN", - sigma_data: Optional[float] = None, - lr_scheduler_config: Optional[Dict] = None, - use_torch_compile: bool = True, - torch_compile_kwargs: Optional[Dict] = None, - conditioner: Callable[..., list[torch.Tensor]] = None, - ): - """ - Initialize DenoiserWithInputAttr. - - Args: - All arguments are the same as the parent Denoiser class. - The conditioner should be a SpatioTemporalConditioner that provides - features that can be used as input_attr. - """ - super().__init__( - arch=arch, - optim=optim, - sigma_distribution=sigma_distribution, - max_radius=max_radius, - average_squared_distance=average_squared_distance, - add_fixed_noise=add_fixed_noise, - add_fixed_ones=add_fixed_ones, - align_noisy_input_during_training=align_noisy_input_during_training, - align_noisy_input_during_evaluation=align_noisy_input_during_evaluation, - mean_center=mean_center, - mirror_augmentation_rate=mirror_augmentation_rate, - bond_loss_coefficient=bond_loss_coefficient, - normalization_type=normalization_type, - sigma_data=sigma_data, - lr_scheduler_config=lr_scheduler_config, - use_torch_compile=use_torch_compile, - torch_compile_kwargs=torch_compile_kwargs, - conditioner=conditioner, - ) - - # Verify that we have a conditioner (required for this subclass) - if self.conditioning_module is None: - raise ValueError( - "DenoiserWithInputAttr requires a conditioner to provide input attributes" - ) - - def xhat_normalized( - self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] - ) -> torch_geometric.data.Batch: - """ - Compute the denoised prediction using the normalization factors from JAMUN. - - This version extracts spatiotemporal features from the conditioner and - passes them as input_attr to E3ConvConditionalWithInputAttr. - """ - sigma = torch.as_tensor(sigma).to(y.pos) - D = y.pos.shape[-1] - - # Compute the normalization factors. - with torch.cuda.nvtx.range("normalization_factors"): - c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) - radial_cutoff = self.effective_radial_cutoff(sigma) / c_in - - # Adjust dimensions. - c_in = unsqueeze_trailing(c_in, y.pos.ndim - 1) - c_skip = unsqueeze_trailing(c_skip, y.pos.ndim - 1) - c_out = unsqueeze_trailing(c_out, y.pos.ndim - 1) - c_noise = c_noise.unsqueeze(0) - - # Add edges to the graph. - with torch.cuda.nvtx.range("add_edges"): - y = self.add_edges(y, radial_cutoff) - - with torch.cuda.nvtx.range("scale_y"): - y_scaled = y.clone() - y_scaled.pos = y.pos * c_in - # Manually copy hidden state - if hasattr(y, "hidden_state") and y.hidden_state is not None: - y_scaled.hidden_state = [] - for positions in y.hidden_state: - y_scaled.hidden_state.append(positions * c_in) - - with torch.cuda.nvtx.range("clone_y"): - xhat = y.clone() - # Manually copy hidden state - if hasattr(y, "hidden_state") and y.hidden_state is not None: - xhat.hidden_state = [h.clone() for h in y.hidden_state] - - # Extract spatiotemporal features from conditioner - with torch.cuda.nvtx.range("conditioning"): - conditioned_structures = self.conditioner(y_scaled) - # print(f"Conditioner is working, number of conditioned structures: {len(conditioned_structures)}") - - # Extract spatiotemporal features as input_attr - with torch.cuda.nvtx.range("extract_input_attr"): - # The conditioner should provide spatiotemporal features - # For SpatioTemporalConditioner, we need to extract the features from the model - if hasattr(self.conditioning_module, 'spatiotemporal_model'): - # Get spatiotemporal features by calling the model with return_spatial_features=True - spatiotemporal_result = self.conditioning_module.spatiotemporal_model( - y_scaled, - unsqueeze_trailing(torch.tensor(self.conditioning_module.c_noise, device=y.pos.device), 1) - ) - # Extract spatial features as input_attr - if isinstance(spatiotemporal_result, dict): - input_attr = spatiotemporal_result.get('spatial_features', None) - else: - # If not dict, assume the result is the spatial features - input_attr = spatiotemporal_result - else: - # Fallback: use zeros with correct shape and irreps - # This allows the model to work even without a proper spatiotemporal conditioner - py_logger = logging.getLogger("jamun") - py_logger.warning("No spatiotemporal_model found in conditioner, using zero input_attr") - # Assume input_attr_irreps is "3x1e" (9D vector) as specified in config - input_attr = torch.zeros(y.pos.shape[0], 9, device=y.pos.device, dtype=y.pos.dtype) - - # Call the architecture with input_attr - with torch.cuda.nvtx.range("g"): - g_pred = self.g( - torch.cat([*conditioned_structures], dim=-1), # Concatenated positions as first arg - topology=y_scaled, - c_noise=c_noise, - effective_radial_cutoff=radial_cutoff, - input_attr=input_attr # Pass spatiotemporal features as input_attr - ) - - xhat.pos = c_skip * y.pos + c_out * g_pred - if hasattr(y, "hidden_state") and y.hidden_state is not None: - xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] - return xhat diff --git a/src/jamun/sampling/walkjump/_single_measurement.py b/src/jamun/sampling/walkjump/_single_measurement.py index 1b3c4b2..c3c3c6b 100644 --- a/src/jamun/sampling/walkjump/_single_measurement.py +++ b/src/jamun/sampling/walkjump/_single_measurement.py @@ -1,5 +1,6 @@ import torch from torch import Tensor +from typing import Optional from tqdm.auto import tqdm diff --git a/src/jamun/utils/__init__.py b/src/jamun/utils/__init__.py index d38593f..dcb87ad 100644 --- a/src/jamun/utils/__init__.py +++ b/src/jamun/utils/__init__.py @@ -1,6 +1,6 @@ from .align import align_A_to_B, align_A_to_B_batched, align_A_to_B_batched_f, find_rigid_alignment from .atom_graphs import to_atom_graphs -from .average_squared_distance import compute_average_squared_distance, compute_average_squared_distance_from_datasets, compute_temporal_average_squared_distance_from_dataset +from .average_squared_distance import compute_average_squared_distance, compute_average_squared_distance_from_datasets, compute_temporal_average_squared_distance_from_datasets from .checkpoint import find_checkpoint, find_checkpoint_directory, get_run_path_for_wandb_run, get_wandb_run_config from .data_with_residue_info import DataWithResidueInformation from .dist_log import dist_log, wandb_dist_log diff --git a/src/jamun/utils/average_squared_distance.py b/src/jamun/utils/average_squared_distance.py index e609561..5f9c616 100644 --- a/src/jamun/utils/average_squared_distance.py +++ b/src/jamun/utils/average_squared_distance.py @@ -74,8 +74,8 @@ def compute_average_squared_distance_from_datasets( return float(mean_avg_sq_dist) -def compute_temporal_average_squared_distance_from_dataset( - dataset, +def compute_temporal_average_squared_distance_from_datasets( + datasets, num_samples: int = 100, verbose: bool = False ) -> float: @@ -83,7 +83,7 @@ def compute_temporal_average_squared_distance_from_dataset( Compute average squared distance between neighboring vertices in temporal graphs. Args: - dataset: Dataset containing spatial graphs with hidden states + datasets: Collection of datasets containing spatial graphs with hidden states num_samples: Number of samples to use for estimation verbose: Whether to print verbose output @@ -96,7 +96,7 @@ def compute_temporal_average_squared_distance_from_dataset( num_graphs = 0 # Follow pattern from existing functions in this module - for item in dataset: + for item in datasets: if num_graphs >= num_samples: break for graph in item: diff --git a/test_spatiotemporal_conditioner.py b/test_spatiotemporal_conditioner.py index 94bef75..a86f974 100755 --- a/test_spatiotemporal_conditioner.py +++ b/test_spatiotemporal_conditioner.py @@ -1,7 +1,8 @@ #!/usr/bin/env python3 """ Test script for loading and testing a conditional denoiser with spatiotemporal conditioner. -Uses the ALA_ALA enhanced sampling dataset for testing. +Uses the new approach where SpatioTemporalConditioner outputs [y.pos, spatial_features] +and E3ConvConditionalSpatioTemporal handles concatenated inputs. """ import e3nn @@ -17,14 +18,16 @@ sys.path.insert(0, 'src') from jamun.data import parse_datasets_from_directory -from jamun.model.denoiser_conditional import DenoiserWithInputAttr +from jamun.model.denoiser_conditional import Denoiser # Changed from DenoiserWithInputAttr +from jamun.model.denoiser import add_edges # Import add_edges function from jamun.model.conditioners.conditioners import SpatioTemporalConditioner from jamun.model.arch.spatiotemporal import E3SpatioTemporal, E3Transformer from jamun.model.arch.e3conv import E3Conv -from jamun.model.arch.e3conv_conditional import E3ConvConditionalWithInputAttr +from jamun.model.arch.e3conv_conditional import E3ConvConditionalSpatioTemporal # Changed from E3ConvConditionalWithInputAttr from jamun.model.pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean from jamun.distributions._distributions import ConstantSigma from jamun.utils import unsqueeze_trailing +from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets # Import temporal function # Setup device device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') @@ -47,7 +50,7 @@ def create_spatial_module() -> E3Conv: ) return E3Conv( - irreps_out="1x1e", + irreps_out="3x1e", # Changed to match temporal module input irreps_hidden="120x0e + 32x1e", irreps_sh="1x0e + 1x1e", hidden_layer_factory=hidden_layer_factory, @@ -71,10 +74,10 @@ def create_spatial_module() -> E3Conv: def create_temporal_module() -> E3Transformer: """Create E3Transformer temporal module.""" return E3Transformer( - irreps_out="3x1e", + irreps_out="3x1e", # Final spatial features output irreps_hidden="8x0e + 4x1e", irreps_sh="1x0e + 1x1e", - irreps_node_attr="1x1e", # Match spatial module output + irreps_node_attr="3x1e", # Match spatial module output num_layers=2, edge_attr_dim=24, num_attention_heads=1, @@ -90,12 +93,44 @@ def create_spatiotemporal_model() -> E3SpatioTemporal: spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr() temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean() + # Compute radial cutoff using temporal average squared distance + print("Computing radial cutoff from temporal dataset...") + try: + # Load dataset to compute temporal average squared distance + dataset = parse_datasets_from_directory( + root="/data2/sules/ALA_ALA_enhanced_full_grid/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + subsample=1, + total_lag_time=5, + lag_subsample_rate=1, + max_datasets=2, # Keep small for testing + num_frames=5 # Small number of frames + ) + + # Compute temporal average squared distance + temporal_avg_sq_dist = compute_temporal_average_squared_distance_from_datasets( + [dataset], # Pass as list since function expects multiple datasets + num_samples=50, # Use fewer samples for testing + verbose=True + ) + + # Use a multiple of the temporal average squared distance as the radial cutoff + # Typically we might use sqrt(temporal_avg_sq_dist) * some_factor + import math + radial_cutoff = math.sqrt(temporal_avg_sq_dist) * 2.0 # Scale factor of 2.0 + print(f"Computed radial cutoff: {radial_cutoff:.6f} nm") + + except Exception as e: + print(f"Warning: Failed to compute temporal cutoff ({e}), using default value 0.05") + radial_cutoff = 0.05 + return E3SpatioTemporal( spatial_module=spatial_module, temporal_module=temporal_module, spatial_to_temporal_pooler=spatial_to_temporal_pooler, temporal_to_spatial_pooler=temporal_to_spatial_pooler, - radial_cutoff=0.05, + radial_cutoff=radial_cutoff, temporal_cutoff=1.0 ) @@ -104,19 +139,19 @@ def create_spatiotemporal_conditioner() -> SpatioTemporalConditioner: spatiotemporal_model = create_spatiotemporal_model() return SpatioTemporalConditioner( - N_structures=1, + N_structures=1, # Changed to 2 for [y.pos, spatial_features] spatiotemporal_model=spatiotemporal_model, c_noise=0.0, freeze_spatiotemporal_model=False # Keep trainable ) def create_conditional_denoiser_config() -> Dict[str, Any]: - """Create configuration for DenoiserWithInputAttr with spatiotemporal conditioner.""" + """Create configuration for Denoiser with spatiotemporal conditioner.""" import functools import e3tools.nn def create_arch(): - """Create the E3ConvConditionalWithInputAttr architecture module.""" + """Create the E3ConvConditionalSpatioTemporal architecture module.""" # Hidden layer factory hidden_layer_factory = functools.partial( e3tools.nn.ConvBlock, @@ -129,8 +164,8 @@ def create_arch(): irreps_hidden_list=["16x0e + 8x1e"] ) - return E3ConvConditionalWithInputAttr( - irreps_out="3x1e", + return E3ConvConditionalSpatioTemporal( + irreps_out="1x1e", # Output should be 3 components (1x1e) to match position irreps_hidden="16x0e + 8x1e", irreps_sh="1x0e + 1x1e", hidden_layer_factory=hidden_layer_factory, @@ -149,8 +184,8 @@ def create_arch(): num_residue_types=25, test_equivariance=False, reduce=None, - N_structures=1, - input_attr_irreps="3x1e", # Accept 3D vector input attributes from spatiotemporal model + N_structures=1, # Changed to 2 for [y.pos, spatial_features] + input_attr_irreps="3x1e", # spatial_features only (9 components = 3x1e) ) def create_optim(params): @@ -158,7 +193,7 @@ def create_optim(params): return torch.optim.Adam(params, lr=0.001) return { - # Required DenoiserWithInputAttr parameters + # Required Denoiser parameters (changed from DenoiserWithInputAttr) 'arch': create_arch, 'optim': create_optim, 'sigma_distribution': ConstantSigma(sigma=0.1), @@ -188,13 +223,7 @@ def add_edges_to_batch(batch: torch_geometric.data.Batch, cutoff: float = 0.05) return batch # Add radius-based edges - edge_index = e3tools.radius_graph( - x=batch.pos, - r=cutoff, - batch=batch.batch, - loop=False, - max_num_neighbors=32 - ) + edge_index = e3tools.radius_graph(batch.pos, cutoff, batch.batch) batch.edge_index = edge_index # Add bonded edges if they exist @@ -257,13 +286,14 @@ def test_spatiotemporal_conditioner(conditioner: SpatioTemporalConditioner, batc conditioned_structures = conditioner(batch) print(f"āœ… Conditioner forward pass successful!") - print(f"Number of conditioned structures: {len(conditioned_structures)} (expected: 1)") - print(f"Conditioned structure shape: {conditioned_structures[0].shape}") + print(f"Number of conditioned structures: {len(conditioned_structures)} (expected: 2)") + print(f"First structure (y.pos) shape: {conditioned_structures[0].shape}") + print(f"Second structure (spatial_features) shape: {conditioned_structures[1].shape}") print(f"Original position shape: {batch.pos.shape}") print(f"Position difference norm: {torch.norm(conditioned_structures[0] - batch.pos):.6f}") - # Verify we got exactly one structure - assert len(conditioned_structures) == 1, f"Expected 1 structure, got {len(conditioned_structures)}" + # Verify we got exactly two structures + assert len(conditioned_structures) == 2, f"Expected 2 structures, got {len(conditioned_structures)}" return True, conditioned_structures @@ -274,9 +304,9 @@ def test_spatiotemporal_conditioner(conditioner: SpatioTemporalConditioner, batc return False, None def test_conditional_denoiser_creation(): - """Test creating DenoiserWithInputAttr with spatiotemporal conditioner.""" + """Test creating Denoiser with spatiotemporal conditioner.""" print("\n" + "="*50) - print("TESTING DENOISER WITH INPUT ATTR CREATION") + print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER CREATION") print("="*50) try: @@ -284,10 +314,10 @@ def test_conditional_denoiser_creation(): config = create_conditional_denoiser_config() # Create denoiser (this will instantiate all components) - denoiser = DenoiserWithInputAttr(**config) + denoiser = Denoiser(**config) denoiser = denoiser.to(device) - print(f"āœ… DenoiserWithInputAttr created successfully!") + print(f"āœ… Denoiser created successfully!") print(f"Denoiser device: {next(denoiser.parameters()).device}") print(f"Has conditioner: {hasattr(denoiser, 'conditioning_module')}") print(f"Architecture type: {type(denoiser.g).__name__}") @@ -303,38 +333,47 @@ def test_conditional_denoiser_creation(): return True, denoiser except Exception as e: - print(f"āŒ DenoiserWithInputAttr creation failed: {e}") + print(f"āŒ Denoiser creation failed: {e}") import traceback traceback.print_exc() return False, None -def test_denoiser_forward_pass(denoiser: DenoiserWithInputAttr, batch: torch_geometric.data.Batch): +def test_denoiser_forward_pass(denoiser: Denoiser, batch: torch_geometric.data.Batch): """Test the complete denoiser forward pass.""" print("\n" + "="*50) - print("TESTING DENOISER WITH INPUT ATTR FORWARD PASS") + print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER FORWARD PASS") print("="*50) try: # Test with sigma = 0.1 sigma = 0.1 - print(f"Testing with sigma = {sigma}") - # Forward pass - this should now extract spatiotemporal features and use them as input_attr + # Debug: check conditioned structures shapes + conditioned_structures = denoiser.conditioning_module(batch) + print(f"DEBUG: Conditioned structures shapes:") + for i, struct in enumerate(conditioned_structures): + print(f" Structure {i}: {struct.shape}") + + concatenated = torch.cat([*conditioned_structures], dim=-1) + print(f"DEBUG: Concatenated shape: {concatenated.shape}") + print(f"DEBUG: Expected irreps: 4x1e = 12 components") + with torch.no_grad(): xhat_batch = denoiser.xhat(batch, sigma) - print(f"āœ… DenoiserWithInputAttr forward pass successful!") + print(f"āœ… Denoiser forward pass successful!") print(f"Input shape: {batch.pos.shape}") print(f"Output shape: {xhat_batch.pos.shape}") - print(f"Position difference norm: {torch.norm(xhat_batch.pos - batch.pos):.6f}") + print(f"Output norm: {torch.norm(xhat_batch.pos):.6f}") + print(f"Used sigma: {sigma}") - # Check that the spatiotemporal features were properly extracted and used - print(f"āœ… Spatiotemporal features successfully integrated as input_attr!") + # Verify output shapes match input + assert xhat_batch.pos.shape == batch.pos.shape, f"Shape mismatch: {xhat_batch.pos.shape} vs {batch.pos.shape}" return True except Exception as e: - print(f"āŒ DenoiserWithInputAttr forward pass failed: {e}") + print(f"āŒ Denoiser forward pass failed: {e}") import traceback traceback.print_exc() return False From 265e75356133666972309735ada24c1007b2c5f2 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Sun, 3 Aug 2025 22:58:51 +0000 Subject: [PATCH 19/32] wrapper for pretrained_denoiser --- ...sample_ala_ala_enhanced_sampling_data.yaml | 10 +- ...pretrained_spatiotemporal_conditioner.yaml | 54 +++ ...n_enhanced_spatiotemporal_conditioner.yaml | 14 +- docs/pretrained_spatiotemporal.md | 337 ++++++++++++++++++ run_model3_beta_comparison.sbatch | 57 --- run_model3_beta_simple.sbatch | 55 --- .../slurm/run_model3_beta_comparison.sh | 0 .../slurm/train_enhanced_sampling_single.sh | 2 +- ...nced_sampling_spatiotemporal_comparison.sh | 91 +++++ src/jamun/cmdline/train.py | 2 + .../e3conv_conditional_spatiotemporal.yaml | 2 +- .../model/arch/spatiotemporal.yaml | 4 +- .../model/conditioner/spatiotemporal.yaml | 9 +- .../spatiotemporal_pretrained.yaml | 47 +-- src/jamun/model/arch/spatiotemporal.py | 17 +- src/jamun/model/conditioners/conditioners.py | 3 +- src/jamun/model/denoiser_conditional.py | 3 + src/jamun/model/denoiser_multimeasurement.py | 3 + src/jamun/model/pooling.py | 46 ++- src/jamun/utils/inspect_pretrained.py | 250 +++++++++++++ src/jamun/utils/pretrained.py | 235 ++++++++++++ src/jamun/utils/pretrained_wrapper.py | 216 +++++++++++ src/jamun/utils/sampling_wrapper.py | 4 +- test_spatiotemporal_conditioner.py | 4 +- 24 files changed, 1283 insertions(+), 182 deletions(-) create mode 100644 configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml create mode 100644 docs/pretrained_spatiotemporal.md delete mode 100644 run_model3_beta_comparison.sbatch delete mode 100644 run_model3_beta_simple.sbatch rename run_model3_beta_comparison.sh => scripts/slurm/run_model3_beta_comparison.sh (100%) create mode 100755 scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh create mode 100644 src/jamun/utils/inspect_pretrained.py create mode 100644 src/jamun/utils/pretrained.py create mode 100644 src/jamun/utils/pretrained_wrapper.py diff --git a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml b/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml index 4cc446a..8daa916 100644 --- a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml +++ b/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml @@ -24,19 +24,21 @@ init_datasets: pdb_pattern: "^(.*).pdb" as_iterable: false subsample: 1 - total_lag_time: 10 + total_lag_time: 5 lag_subsample_rate: 1 max_datasets: 10 label_override: "ALA_ALA" -num_sampling_steps_per_batch: 100 +num_sampling_steps_per_batch: 1000 num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/tw4vjtqo +wandb_train_run_path: sule-shashank/jamun/2k3e9l7x +# checkpoint_type: last +# checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last sigma: 0.04 @@ -44,7 +46,7 @@ M: 1.0 delta: 0.04 friction: 1.0 inverse_temperature: 1.0 -score_fn_clip: 100.0 +score_fn_clip: null sampler: _target_: jamun.sampling.SamplerMemory diff --git a/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml new file mode 100644 index 0000000..bd179be --- /dev/null +++ b/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml @@ -0,0 +1,54 @@ +# @package _global_ +defaults: + - override /model: denoiser_conditional_spatiotemporal + - override /model/conditioner: spatiotemporal_pretrained + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 8 # Reduced batch size due to increased model complexity + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + # max_datasets: 2 # Increased for more training data + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + # max_datasets: 1 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 1 + max_radius: 1.0 + optim: + lr: 0.002 # Slightly reduced learning rate for stability + +trainer: + val_check_interval: 0.5 + max_epochs: 100 # Increased due to model complexity + # devices: 1 + # gradient_clip_val: 1.0 # Add gradient clipping for stability + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "SpatioTemporalConditioner on enhanced sampling data - processes temporal sequences through spatial and temporal modules" + tags: ["spatiotemporal_conditioner", "enhanced_sampling", "transformer", "e3conv"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml index 17e7043..e2aaf54 100644 --- a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml @@ -1,14 +1,14 @@ # @package _global_ defaults: - override /model: denoiser_conditional_spatiotemporal - # - override /callbacks: - # - timing.yaml - # - lr_monitor.yaml - # - model_checkpoint.yaml + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml data: datamodule: - batch_size: 16 # Reduced batch size due to increased model complexity + batch_size: 8 # Reduced batch size due to increased model complexity datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -36,13 +36,13 @@ model: sigma: 0.04 arch: n_layers: 1 - max_radius: 10.0 + max_radius: 1.0 optim: lr: 0.002 # Slightly reduced learning rate for stability trainer: val_check_interval: 0.5 - max_epochs: 50 # Increased due to model complexity + max_epochs: 100 # Increased due to model complexity # devices: 1 # gradient_clip_val: 1.0 # Add gradient clipping for stability diff --git a/docs/pretrained_spatiotemporal.md b/docs/pretrained_spatiotemporal.md new file mode 100644 index 0000000..86700cb --- /dev/null +++ b/docs/pretrained_spatiotemporal.md @@ -0,0 +1,337 @@ +# Pretrained Spatiotemporal Models + +This document describes how to use pretrained denoiser models as spatial and temporal modules in the spatiotemporal transformer architecture. + +## Overview + +The spatiotemporal transformer consists of four main components: +- **Spatial Module**: Processes spatial positions (can be a pretrained denoiser) +- **Temporal Module**: Processes temporal graphs (can be a pretrained denoiser) +- **Spatialā†’Temporal Pooler**: Converts spatial features to temporal representation +- **Temporalā†’Spatial Pooler**: Converts temporal features back to spatial representation + +You can now use pretrained denoiser models directly as spatial/temporal modules using the wrapper approach, with control over: +- Whether to freeze each module (trainable: true/false) +- Loading from WandB runs or direct checkpoint paths +- **No architecture matching required** - pretrained denoisers are wrapped to expose only their `xhat` function + +## Quick Start + +### 1. Basic Usage + +The simplest way to use a pretrained denoiser is to specify it directly in your config: + +```yaml +# configs/experiment/my_experiment.yaml +defaults: + - override /model/conditioner: spatiotemporal_pretrained_wrapper_example + +model: + conditioner: + spatiotemporal_model: + spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "your_entity/your_project/spatial_run_id" + trainable: false # Freeze the pretrained spatial module + + temporal_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "your_entity/your_project/temporal_run_id" + trainable: true # Fine-tune the temporal module +``` + +### 2. Mixed Approach + +You can also mix pretrained modules with modules trained from scratch: + +```yaml +model: + conditioner: + spatiotemporal_model: + # Use pretrained spatial module + spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "entity/project/spatial_run_id" + trainable: false + + # Train temporal module from scratch + temporal_module: + _target_: jamun.model.arch.spatiotemporal.E3Transformer + irreps_out: "3x1e" + irreps_hidden: "8x0e + 4x1e" + # ... other E3Transformer parameters +``` + +### 3. Available Config Templates + +Use one of the provided config templates: + +- `spatiotemporal_pretrained_wrapper_example.yaml`: Both spatial and temporal modules from pretrained denoisers +- `spatiotemporal_mixed_example.yaml`: Pretrained spatial module + temporal module trained from scratch + +## Configuration Options + +### Wrapper Parameters + +The pretrained wrapper accepts these parameters: + +```yaml +spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "username/project/run_id" # Load from WandB run + checkpoint_path: "/path/to/checkpoint.ckpt" # OR load from direct path + checkpoint_type: "best_so_far" # "last", "best_so_far", or specific .ckpt filename + trainable: true # Whether to keep trainable (false = freeze) +``` + +### Key Parameters + +- **`wandb_run_path`**: Load from WandB run (format: "username/project/run_id") +- **`checkpoint_path`**: Direct path to checkpoint file (mutually exclusive with wandb_run_path) +- **`checkpoint_type`**: Which checkpoint to load ("best_so_far", "last", or specific filename) +- **`trainable`**: Whether the module should be trainable (default: true, false = freeze) + +## Common Use Cases + +### 1. Load Pretrained Spatial, Train Temporal from Scratch + +```yaml +# Use: spatiotemporal_mixed_example.yaml +model: + conditioner: + spatiotemporal_model: + spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "username/project/spatial_pretrained_run" + trainable: false # Freeze pretrained spatial + + temporal_module: + _target_: jamun.model.arch.spatiotemporal.E3Transformer + irreps_out: "3x1e" + irreps_hidden: "8x0e + 4x1e" + # ... standard E3Transformer configuration +``` + +### 2. Fine-tune Both Modules + +```yaml +# Use: spatiotemporal_pretrained_wrapper_example.yaml +model: + conditioner: + spatiotemporal_model: + spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "username/project/spatial_run" + trainable: true # Fine-tune pretrained spatial + + temporal_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "username/project/temporal_run" + trainable: true # Fine-tune pretrained temporal +``` + +### 3. Load from Different Checkpoint Sources + +```yaml +model: + conditioner: + spatiotemporal_model: + spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + checkpoint_path: "/shared/models/spatial_best.ckpt" + trainable: true + + temporal_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "username/project/temporal_run" + trainable: true +``` + +## How It Works + +### Wrapper Mechanism + +The `DenoiserWrapper` class replicates the full denoiser logic including normalization: + +```python +class DenoiserWrapper(nn.Module): + def __init__(self, denoiser_model: nn.Module, trainable: bool = True): + super().__init__() + self.denoiser = denoiser_model + + # Control trainability + if not trainable: + for param in self.denoiser.parameters(): + param.requires_grad = False + + def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cutoff): + # Replicates xhat and xhat_normalized logic from the denoiser + # Uses denoiser's own normalization parameters + # Returns properly normalized denoised positions +``` + +### Loading Process + +The `return_wrapped_denoiser` function: +1. Loads the full pretrained PyTorch Lightning model from checkpoint +2. Wraps it with `DenoiserWrapper` +3. Sets trainability based on the `trainable` parameter +4. Returns the wrapped model ready for use as a spatial/temporal module + +### Advantages + +- **No architecture matching required**: Works with any pretrained denoiser +- **Simple configuration**: Just specify wandb run path and trainability +- **Authentic denoiser behavior**: Replicates exact `xhat` and `xhat_normalized` logic +- **Automatic normalization**: Uses the denoiser's own training parameters +- **Flexible**: Mix pretrained and from-scratch modules easily + +### Normalization Handling + +The wrapper automatically uses the denoiser's own normalization parameters: + +- **`normalization_type`**: The type used during training ("JAMUN", "EDM", or None) +- **`average_squared_distance`**: For JAMUN normalization +- **`sigma_data`**: For EDM normalization +- **`mean_center`**: Whether to apply mean centering + +This ensures the pretrained denoiser behaves exactly as it was trained, without any external rescaling parameters needed. + +## Utility Commands (Optional) + +### Inspect Checkpoint (if needed) + +```bash +# Basic inspection to verify model can be loaded +python -c " +from jamun.utils.pretrained_wrapper import return_wrapped_denoiser +model = return_wrapped_denoiser(wandb_run_path='username/project/run_id') +print('āœ“ Model loaded successfully') +print(f'Model type: {type(model.denoiser)}') +" +``` + +## Module Paths + +Common module paths for extracting from loaded models: + +| Module | Typical Path | +|--------|-------------| +| Spatial module from spatiotemporal model | `conditioner.spatiotemporal_model.spatial_module` | +| Temporal module from spatiotemporal model | `conditioner.spatiotemporal_model.temporal_module` | +| Entire spatiotemporal model | `conditioner.spatiotemporal_model` | +| Architecture from denoiser models | `arch` | +| Conditioner from denoiser models | `conditioner` | + +## Troubleshooting + +### Model Loading Issues + +If the checkpoint cannot be loaded as a complete model: +- Check that the checkpoint is a valid PyTorch Lightning checkpoint +- Ensure the model class can be auto-detected or specify `model_class` explicitly +- Verify that all required dependencies are available + +### Module Path Errors + +If a module path cannot be found: +- Use the `inspect` command to see the actual model structure +- Check that the path uses dot notation (e.g., `module.submodule`) +- Verify the checkpoint contains the expected model architecture + +### Learning Rate Issues + +- Use standard learning rates - freezing/unfreezing is now the main control mechanism +- Use gradient clipping when fine-tuning: `trainer.gradient_clip_val: 1.0` +- Monitor training closely in the first few epochs + +### Memory Issues + +- Reduce batch size when using complex spatiotemporal models +- Consider freezing larger modules if GPU memory is limited + +## Example Workflows + +### Workflow 1: Progressive Training + +1. Train spatial module alone +2. Freeze spatial, train temporal module +3. Fine-tune both together with low learning rates + +### Workflow 2: Transfer Learning + +1. Use spatial module pretrained on molecular dynamics +2. Train temporal module for your specific task +3. Fine-tune both on your target dataset + +### Workflow 3: Architecture Search + +1. Extract and test different pretrained modules +2. Mix and match spatial/temporal components +3. Find optimal combination for your task + +## Configuration Templates + +The following configuration templates are available: + +- `spatiotemporal_pretrained_advanced.yaml`: Full configuration with all options +- `spatiotemporal_pretrained_spatial_only.yaml`: Load spatial only, train temporal +- `spatiotemporal_pretrained_finetune.yaml`: Fine-tune both with different learning rates +- `spatiotemporal.yaml`: Standard configuration without pretrained loading + +Choose the template that best matches your use case and customize the pretrained paths. + +## New Approach: Complete Model Loading + +This implementation uses a **complete model loading** approach rather than state dict matching: + +### Benefits: +- **No Architecture Matching**: The exact architecture from the checkpoint is loaded directly +- **Eliminates Parameter Mismatches**: No missing/unexpected keys issues +- **Preserves Model Structure**: The complete model hierarchy is maintained +- **Automatic Class Detection**: Model classes are auto-detected from checkpoints +- **Simpler Configuration**: No need to specify complex parameter prefixes or strict loading + +### How It Works: +1. **Simple Logic**: If `checkpoint_path` or `wandb_run_path` is provided ā†’ load pretrained module, otherwise use config +2. **Load Complete Model**: `Model.load_from_checkpoint()` loads the entire saved model +3. **Extract Module**: Use `module_path` to extract the specific module (defaults to `"arch"`) +4. **Replace & Configure**: Module replaces the config version, with `trainable` flag applied + +This approach is much more robust and eliminates the common issues with architecture mismatches that plague state dict loading approaches. + +## Simplified Configuration + +The configuration has been simplified to follow a clear principle: + +**Rule: If you provide a checkpoint path ā†’ load from checkpoint, otherwise use the config architecture** + +### Simple Parameters: +- **`wandb_run_path`** or **`checkpoint_path`**: Specify where to load from (if neither provided, uses config) +- **`trainable`**: `true` = trainable, `false` = freeze (default: `true`) +- **`module_path`**: What to extract from checkpoint (default: `"arch"`) + +### No More: +- āŒ Complex `freeze` vs `trainable` logic +- āŒ `strict` loading parameters +- āŒ `model_class` specifications +- āŒ Complex module prefix matching +- āŒ Architecture compatibility checking +- āŒ Learning rate multipliers and parameter groups + +### Examples: +```yaml +# Load pretrained spatial, freeze it +spatial_module: + wandb_run_path: "user/project/run_id" + trainable: false + +# Load pretrained temporal, fine-tune it +temporal_module: + checkpoint_path: "/path/to/model.ckpt" + trainable: true + +# No checkpoint = use config architecture +spatial_module: + trainable: true # Just normal training +``` \ No newline at end of file diff --git a/run_model3_beta_comparison.sbatch b/run_model3_beta_comparison.sbatch deleted file mode 100644 index 43f5a78..0000000 --- a/run_model3_beta_comparison.sbatch +++ /dev/null @@ -1,57 +0,0 @@ -#!/usr/bin/env bash - -#SBATCH --job-name=model3_beta_comparison -#SBATCH --array=1-2 -#SBATCH --output=logs/model3_beta_%A_%a.out -#SBATCH --error=logs/model3_beta_%A_%a.err -#SBATCH --time=24:00:00 -#SBATCH --partition=gpu -#SBATCH --gres=gpu:1 -#SBATCH --cpus-per-task=4 -#SBATCH --mem=32G - -# Create logs directory if it doesn't exist -mkdir -p logs - -eval "$(conda shell.bash hook)" -conda activate jamun - -set -eux - -echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" -echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" -echo "hostname = $(hostname)" - -export HYDRA_FULL_ERROR=1 - -# Generate unique run key for this experiment -RUN_KEY=$(openssl rand -hex 12) -echo "RUN_KEY = ${RUN_KEY}" - -nvidia-smi - -# Navigate to project directory -cd /homefs/home/sules/jamun - -# Determine which beta configuration to run based on array task ID -if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then - # First run: Adam betas (0.9, 0.9) - echo "Running model3 experiment with Adam betas (0.9, 0.9)" - jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - ++run_key=$RUN_KEY \ - 'model.optim.betas=[0.9,0.9]' \ - ++logger.wandb.name="model3_beta09_09" \ - ++logger.wandb.tags="[${SLURM_JOB_ID},${RUN_KEY},model3,beta09_09]" -else - # Second run: Adam betas (0.9, 0.999) - PyTorch default - echo "Running model3 experiment with Adam betas (0.9, 0.999)" - jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - ++run_key=$RUN_KEY \ - 'model.optim.betas=[0.9,0.999]' \ - ++logger.wandb.name="model3_beta09_0999" \ - ++logger.wandb.tags="[${SLURM_JOB_ID},${RUN_KEY},model3,beta09_0999]" -fi - -echo "Model3 beta comparison experiment $SLURM_ARRAY_TASK_ID completed" \ No newline at end of file diff --git a/run_model3_beta_simple.sbatch b/run_model3_beta_simple.sbatch deleted file mode 100644 index 1d63dcb..0000000 --- a/run_model3_beta_simple.sbatch +++ /dev/null @@ -1,55 +0,0 @@ -#!/usr/bin/env bash - -#SBATCH --job-name=model3_beta_simple -#SBATCH --array=1-2 -#SBATCH --output=logs/model3_simple_%A_%a.out -#SBATCH --error=logs/model3_simple_%A_%a.err -#SBATCH --time=24:00:00 -#SBATCH --partition=gpu -#SBATCH --gres=gpu:1 -#SBATCH --cpus-per-task=4 -#SBATCH --mem=32G - -# Create logs directory if it doesn't exist -mkdir -p logs - -eval "$(conda shell.bash hook)" -conda activate jamun - -set -eux - -echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" -echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" -echo "hostname = $(hostname)" - -export HYDRA_FULL_ERROR=1 - -# Generate unique run key for this experiment -RUN_KEY=$(openssl rand -hex 12) -echo "RUN_KEY = ${RUN_KEY}" - -nvidia-smi - -# Navigate to project directory -cd /homefs/home/sules/jamun - -# Determine which beta configuration to run based on array task ID -if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then - # First run: Adam betas (0.9, 0.9) - echo "Running model3 experiment with Adam betas (0.9, 0.9)" - jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - ++run_key=$RUN_KEY \ - 'model.optim.betas=[0.9,0.9]' \ - ++logger.wandb.name="model3_beta09_09" -else - # Second run: Adam betas (0.9, 0.999) - PyTorch default - echo "Running model3 experiment with Adam betas (0.9, 0.999)" - jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - ++run_key=$RUN_KEY \ - 'model.optim.betas=[0.9,0.999]' \ - ++logger.wandb.name="model3_beta09_0999" -fi - -echo "Model3 beta comparison experiment $SLURM_ARRAY_TASK_ID completed" \ No newline at end of file diff --git a/run_model3_beta_comparison.sh b/scripts/slurm/run_model3_beta_comparison.sh similarity index 100% rename from run_model3_beta_comparison.sh rename to scripts/slurm/run_model3_beta_comparison.sh diff --git a/scripts/slurm/train_enhanced_sampling_single.sh b/scripts/slurm/train_enhanced_sampling_single.sh index ee6d6ae..6e5db78 100755 --- a/scripts/slurm/train_enhanced_sampling_single.sh +++ b/scripts/slurm/train_enhanced_sampling_single.sh @@ -20,4 +20,4 @@ echo "Conda environment: $CONDA_DEFAULT_ENV" nvidia-smi # Run training with parameter overrides -jamun_train --config-dir=configs experiment=train_enhanced_position_conditioner \ No newline at end of file +jamun_train --config-dir=configs experiment=train_enhanced_spatiotemporal_conditioner.yaml \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh b/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh new file mode 100755 index 0000000..16b0086 --- /dev/null +++ b/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh @@ -0,0 +1,91 @@ +#!/usr/bin/env bash + +#SBATCH --partition=gpu2 +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-5 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# Define configurations for each job +case ${SLURM_ARRAY_TASK_ID} in + 0) + echo "Job 0: Initialized spatial module + TemporalToSpatialNodeAttrMean" + CONFIG="train_enhanced_spatiotemporal_conditioner" + POOLER="jamun.model.pooling.TemporalToSpatialNodeAttrMean" + TRAINABLE_OVERRIDE="" + ;; + 1) + echo "Job 1: Initialized spatial module + TemporalToSpatialNodeAttr" + CONFIG="train_enhanced_spatiotemporal_conditioner" + POOLER="jamun.model.pooling.TemporalToSpatialNodeAttr" + TRAINABLE_OVERRIDE="" + ;; + 2) + echo "Job 2: Pretrained trainable spatial module + TemporalToSpatialNodeAttrMean" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + POOLER="jamun.model.pooling.TemporalToSpatialNodeAttrMean" + TRAINABLE_OVERRIDE="model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + ;; + 3) + echo "Job 3: Pretrained trainable spatial module + TemporalToSpatialNodeAttr" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + POOLER="jamun.model.pooling.TemporalToSpatialNodeAttr" + TRAINABLE_OVERRIDE="model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + ;; + 4) + echo "Job 4: Pretrained non-trainable spatial module + TemporalToSpatialNodeAttrMean" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + POOLER="jamun.model.pooling.TemporalToSpatialNodeAttrMean" + TRAINABLE_OVERRIDE="model.conditioner.spatiotemporal_model.spatial_module.trainable=false" + ;; + 5) + echo "Job 5: Pretrained non-trainable spatial module + TemporalToSpatialNodeAttr" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + POOLER="jamun.model.pooling.TemporalToSpatialNodeAttr" + TRAINABLE_OVERRIDE="model.conditioner.spatiotemporal_model.spatial_module.trainable=false" + ;; + *) + echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" + exit 1 + ;; +esac + +# Build the command with overrides +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add pooler override (keeping the irreps_out parameter from base config) +CMD="$CMD model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=${POOLER}" + +# Add trainable override if needed +if [ -n "$TRAINABLE_OVERRIDE" ]; then + CMD="$CMD $TRAINABLE_OVERRIDE" +fi + +# Add dataset and training overrides +# CMD="$CMD data.datamodule.datasets.train.max_datasets=1" +# CMD="$CMD data.datamodule.datasets.val.max_datasets=1" +CMD="$CMD trainer.max_epochs=100" +CMD="$CMD ++wandb.logger.group=spatiotemporal_comparison" + +# Add job-specific wandb tags +WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" +CMD="$CMD +logger.wandb.tags=[\"${WANDB_TAG}\",\"spatiotemporal_comparison\"]" + +echo "Running command: $CMD" +exec $CMD \ No newline at end of file diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 5a2b74c..08ab7cb 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -144,7 +144,9 @@ def run(cfg): D=3) cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff cfg.model.conditioner.c_noise = c_noise_float + cfg.model.conditioner.c_in = c_in_float dist_log(f"Set cfg.model.conditioner.spatiotemporal_model.c_noise to {c_noise_float}") + dist_log(f"Set cfg.model.conditioner.c_in to {c_in_float}") # # do this for the sweep # if cfg.model.N_measurements_hidden is not None: # dist_log(f"Number of hidden measurements: {cfg.model.N_measurements_hidden}") diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml index 96bcfec..b125f54 100644 --- a/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml @@ -21,7 +21,7 @@ hidden_layer_factory: _target_: e3tools.nn.ConvBlock _partial_: true conv: - _target_: e3tools.nn.Conv + _target_: e3tools.nn.Conv # replace with Conv for non-separable case _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP diff --git a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml index 275efb6..ef23d79 100644 --- a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml @@ -51,6 +51,8 @@ temporal_module: # Pooling modules spatial_to_temporal_pooler: _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr + irreps_out: "1x1e" # Match spatial module output temporal_to_spatial_pooler: - _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean + irreps_out: "3x1e" # Match temporal module output \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml index 9f3ad70..b10ac95 100644 --- a/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml @@ -1,6 +1,7 @@ _target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner N_structures: 2 # Now returns [y.pos, spatial_features] c_noise: 0.0 +c_in: 1.0 freeze_spatiotemporal_model: false # Trainable by default # Spatiotemporal model configuration @@ -30,8 +31,8 @@ spatiotemporal_model: hidden_layer_factory: _target_: e3tools.nn.ConvBlock _partial_: true - conv: - _target_: e3tools.nn.Conv + conv:--- + _target_: e3tools.nn.Conv # replace with Conv for non-separable case _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP @@ -54,6 +55,8 @@ spatiotemporal_model: # Pooling modules spatial_to_temporal_pooler: _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr + irreps_out: "1x1e" # Match spatial module output temporal_to_spatial_pooler: - _target_: jamun.model.pooling.TemporalToSpatialNodeAttr \ No newline at end of file + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean + irreps_out: "120x0e + 32x1e" # Match temporal module output \ No newline at end of file diff --git a/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml index 38b3f39..391f84d 100644 --- a/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml +++ b/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml @@ -1,51 +1,40 @@ -# @package _global_ _target_: jamun.model.conditioners.conditioners.SpatioTemporalConditioner -N_structures: 1 +N_structures: 2 # Now returns [y.pos, spatial_features] c_noise: 0.0 -freeze_spatiotemporal_model: true # Frozen for pretrained models - +freeze_spatiotemporal_model: false # Trainable by default +c_in: 1.0 # Spatiotemporal model configuration spatiotemporal_model: _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal radial_cutoff: 0.05 temporal_cutoff: 1.0 - + # Spatial module (E3Conv) spatial_module: - _target_: jamun.model.arch.e3conv_conditional.E3Conv - irreps_out: "1x1e" - irreps_hidden: "120x0e + 32x1e" - irreps_sh: "1x0e + 1x1e" - n_layers: 1 - edge_attr_dim: 64 - use_residue_information: true - atom_type_embedding_dim: 8 - atom_code_embedding_dim: 8 - residue_code_embedding_dim: 32 - residue_index_embedding_dim: 8 - use_residue_sequence_index: false - num_atom_types: 20 - max_sequence_length: 10 - num_atom_codes: 10 - num_residue_types: 25 - test_equivariance: false - reduce: null - + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "sule-shashank/jamun/sxcdx4wf" + checkpoint_type: "last" + trainable: false + c_in: ${model.conditioner.c_in} + + # Temporal module (E3Transformer) temporal_module: _target_: jamun.model.arch.spatiotemporal.E3Transformer - irreps_out: "3x1e" - irreps_hidden: "8x0e + 4x1e" + irreps_out: "120x0e + 32x1e" # Final spatial features output + irreps_hidden: "120x0e + 32x1e" irreps_sh: "1x0e + 1x1e" irreps_node_attr: "1x1e" # Match spatial module output num_layers: 2 edge_attr_dim: 24 num_attention_heads: 1 reduce: null - + # Pooling modules spatial_to_temporal_pooler: _target_: jamun.model.pooling.SpatialTemporalToTemporalNodeAttr - + irreps_out: "1x1e" # Match spatial module output (from temporal module irreps_node_attr) + temporal_to_spatial_pooler: - _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean \ No newline at end of file + _target_: jamun.model.pooling.TemporalToSpatialNodeAttrMean + irreps_out: "120x0e + 32x1e" # Match temporal module output \ No newline at end of file diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index 2ca7a6d..b0ac51f 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -17,7 +17,7 @@ import torch_geometric.data import e3tools import e3tools.nn - +import logging from jamun.model.arch.e3conv import E3Conv @@ -67,7 +67,7 @@ def spatial_to_temporal_graphs(batch): temporal_length = 1 # Store reference to spatial graph - spatial_graph = batch.clone() + # spatial_graph = batch.clone() temporal_graphs = [] @@ -118,7 +118,7 @@ def spatial_to_temporal_graphs(batch): temporal_batch = torch_geometric.data.Batch.from_data_list(temporal_graphs) # Store spatial graph reference - temporal_batch.spatial_graph = spatial_graph + # temporal_batch.spatial_graph = spatial_graph return temporal_batch @@ -249,7 +249,8 @@ def forward( basis="gaussian", cutoff=True, ) - + temporal_edge_attr = torch.ones_like(temporal_edge_attr) # TODO: remove this, this is hacking. + # Optional bondedness (if bond_mask exists in the temporal graph) if hasattr(temporal_graph, 'bond_mask') and temporal_graph.bond_mask is not None: bonded_edge_attr = self.embed_bondedness(temporal_graph.bond_mask) @@ -324,7 +325,8 @@ def __init__( self.temporal_to_spatial_pooler = temporal_to_spatial_pooler self.radial_cutoff = radial_cutoff self.temporal_cutoff = temporal_cutoff - + + def forward( self, batch: torch_geometric.data.Batch, @@ -410,8 +412,9 @@ def forward( spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) # Step 7: Convert temporal graph back to spatial graph - output_spatial_graph = temporal_to_spatial_graphs(temporal_batch) - + # output_spatial_graph = temporal_to_spatial_graphs(temporal_batch) + output_spatial_graph = batch + # Prepare return values if return_temporal_features or return_temporal_graph: result = { diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 460b025..64f6ac2 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -343,4 +343,5 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: # Return list containing [y.pos, spatial_features] for concatenation # The denoiser will concatenate these along the feature dimension - return [y.pos, spatial_features] \ No newline at end of file + return [y.pos, spatial_features] + diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index 6724949..253fe1b 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -36,6 +36,9 @@ def __init__( use_torch_compile: bool = True, torch_compile_kwargs: Optional[Dict] = None, conditioner: Callable[..., list[torch.Tensor]] = None, + rotational_augmentation: bool = False, + alignment_correction_order: int = 0, + pass_topology_as_atom_graphs: bool = False, ): super().__init__() self.save_hyperparameters(logger=False) diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py index 1542d21..79898a5 100644 --- a/src/jamun/model/denoiser_multimeasurement.py +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -40,6 +40,9 @@ def __init__( N_measurements_hidden: int = 1, N_measurements: int = 1, max_graphs_per_batch: int = None, + rotational_augmentation: bool = False, + alignment_correction_order: int = 0, + pass_topology_as_atom_graphs: bool = False, ): super().__init__() self.save_hyperparameters(logger=False) diff --git a/src/jamun/model/pooling.py b/src/jamun/model/pooling.py index b27dd53..0e90cf9 100644 --- a/src/jamun/model/pooling.py +++ b/src/jamun/model/pooling.py @@ -6,7 +6,7 @@ import torch import torch_geometric import pytorch_lightning as pl - +from e3tools.nn import LayerNorm class SpatialToTemporalNodeAttr(pl.LightningModule): """ @@ -14,8 +14,10 @@ class SpatialToTemporalNodeAttr(pl.LightningModule): by repeating first temporal feature. """ - def __init__(self): + def __init__(self, irreps_out): super().__init__() + self.irreps_out = irreps_out + self.layer_norm = LayerNorm(irreps_out) def forward(self, spatial_node_attr_temporal, temporal_batch): """ @@ -52,6 +54,9 @@ def forward(self, spatial_node_attr_temporal, temporal_batch): assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" + # Apply layer normalization before returning + temporal_node_attr = self.layer_norm(temporal_node_attr) + return temporal_node_attr @@ -61,8 +66,10 @@ class TemporalToSpatialNodeAttr(pl.LightningModule): Takes the first temporal node attribute from each temporal graph. """ - def __init__(self): + def __init__(self, irreps_out): super().__init__() + self.irreps_out = irreps_out + self.layer_norm = LayerNorm(irreps_out) def forward(self, temporal_node_attr, temporal_batch): """ @@ -97,6 +104,9 @@ def forward(self, temporal_node_attr, temporal_batch): assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" + # Apply layer normalization before returning + spatial_node_attr = self.layer_norm(spatial_node_attr) + return spatial_node_attr @@ -106,8 +116,10 @@ class TemporalToSpatialNodeAttrMean(pl.LightningModule): Takes the mean of all temporal node attributes for each temporal graph. """ - def __init__(self): + def __init__(self, irreps_out): super().__init__() + self.irreps_out = irreps_out + self.layer_norm = LayerNorm(irreps_out) def forward(self, temporal_node_attr, temporal_batch): """ @@ -144,6 +156,9 @@ def forward(self, temporal_node_attr, temporal_batch): assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" + # Apply layer normalization before returning + spatial_node_attr = self.layer_norm(spatial_node_attr) + return spatial_node_attr @@ -153,8 +168,10 @@ class SpatialTemporalToTemporalNodeAttr(pl.LightningModule): Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. """ - def __init__(self): + def __init__(self, irreps_out): super().__init__() + self.irreps_out = irreps_out + self.layer_norm = LayerNorm(irreps_out) def forward(self, spatial_node_attr_temporal, temporal_batch): """ @@ -189,29 +206,32 @@ def forward(self, spatial_node_attr_temporal, temporal_batch): assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" + # Apply layer normalization before returning + temporal_node_attr = self.layer_norm(temporal_node_attr) + return temporal_node_attr # Legacy function interfaces for backward compatibility -def spatial_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch): +def spatial_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch, irreps_out): """Legacy function interface for backward compatibility.""" - module = SpatialToTemporalNodeAttr() + module = SpatialToTemporalNodeAttr(irreps_out) return module(spatial_node_attr_temporal, temporal_batch) -def temporal_to_spatial_node_attr(temporal_node_attr, temporal_batch): +def temporal_to_spatial_node_attr(temporal_node_attr, temporal_batch, irreps_out): """Legacy function interface for backward compatibility.""" - module = TemporalToSpatialNodeAttr() + module = TemporalToSpatialNodeAttr(irreps_out) return module(temporal_node_attr, temporal_batch) -def temporal_to_spatial_node_attr_mean(temporal_node_attr, temporal_batch): +def temporal_to_spatial_node_attr_mean(temporal_node_attr, temporal_batch, irreps_out): """Legacy function interface for backward compatibility.""" - module = TemporalToSpatialNodeAttrMean() + module = TemporalToSpatialNodeAttrMean(irreps_out) return module(temporal_node_attr, temporal_batch) -def spatial_temporal_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch): +def spatial_temporal_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch, irreps_out): """Legacy function interface for backward compatibility.""" - module = SpatialTemporalToTemporalNodeAttr() + module = SpatialTemporalToTemporalNodeAttr(irreps_out) return module(spatial_node_attr_temporal, temporal_batch) \ No newline at end of file diff --git a/src/jamun/utils/inspect_pretrained.py b/src/jamun/utils/inspect_pretrained.py new file mode 100644 index 0000000..f95a811 --- /dev/null +++ b/src/jamun/utils/inspect_pretrained.py @@ -0,0 +1,250 @@ +#!/usr/bin/env python3 +""" +Utility script for inspecting pretrained checkpoints and checking module compatibility. +""" + +import argparse +import sys +from pathlib import Path +from typing import Optional, Dict, Any + +import torch +import hydra +from omegaconf import DictConfig + +from jamun.utils.pretrained import ( + load_checkpoint_state_dict, + load_pretrained_model_from_checkpoint, + extract_module_from_model, + inspect_model_structure, + check_model_compatibility +) +from jamun.utils.checkpoint import find_checkpoint + + +def print_checkpoint_structure(checkpoint_path: str, max_depth: int = 2, use_model_loading: bool = True): + """Print the structure of a checkpoint file.""" + print(f"\nšŸ“ Checkpoint structure: {checkpoint_path}") + print("=" * 60) + + if use_model_loading: + # Try to load as a complete model first + try: + model = load_pretrained_model_from_checkpoint(checkpoint_path=checkpoint_path) + if model is not None: + inspect_model_structure(model, max_depth) + return + else: + print("āš ļø Could not load as complete model, falling back to state_dict inspection") + except Exception as e: + print(f"āš ļø Model loading failed ({e}), falling back to state_dict inspection") + + # Fallback to state_dict inspection + try: + state_dict = load_checkpoint_state_dict(checkpoint_path) + + # Group keys by their prefixes + key_groups: Dict[str, list] = {} + for key in state_dict.keys(): + parts = key.split('.') + if len(parts) >= max_depth: + prefix = '.'.join(parts[:max_depth]) + else: + prefix = key + + if prefix not in key_groups: + key_groups[prefix] = [] + key_groups[prefix].append(key) + + # Print grouped structure + for prefix in sorted(key_groups.keys()): + keys = key_groups[prefix] + if len(keys) == 1 and keys[0] == prefix: + # Single parameter + tensor = state_dict[prefix] + print(f" {prefix}: {list(tensor.shape)} ({tensor.dtype})") + else: + # Group of parameters + total_params = sum(state_dict[key].numel() for key in keys) + print(f" {prefix}.* : {len(keys)} parameters ({total_params:,} total elements)") + + # Show a few example keys + if len(keys) <= 5: + for key in sorted(keys)[:5]: + tensor = state_dict[key] + sub_key = key[len(prefix)+1:] if key.startswith(prefix + '.') else key + print(f" ā””ā”€ {sub_key}: {list(tensor.shape)}") + else: + for key in sorted(keys)[:3]: + tensor = state_dict[key] + sub_key = key[len(prefix)+1:] if key.startswith(prefix + '.') else key + print(f" ā””ā”€ {sub_key}: {list(tensor.shape)}") + print(f" ā””ā”€ ... and {len(keys)-3} more") + + total_params = sum(tensor.numel() for tensor in state_dict.values()) + print(f"\nšŸ“Š Total parameters: {total_params:,}") + + except Exception as e: + print(f"āŒ Error loading checkpoint: {e}") + + +def check_compatibility_with_config( + checkpoint_path: str, + config_path: str, + module_path: Optional[str] = None +): + """Check if a checkpoint is compatible and can be loaded.""" + print(f"\nšŸ” Checking compatibility...") + print(f"Checkpoint: {checkpoint_path}") + print(f"Config: {config_path}") + if module_path: + print(f"Module path: {module_path}") + print("=" * 60) + + try: + # Check if checkpoint can be loaded as a model + compatibility = check_model_compatibility(checkpoint_path=checkpoint_path) + + if compatibility['loadable']: + print("āœ… Checkpoint can be loaded as a complete model!") + print(f"Model class: {compatibility['model_class'].__name__}") + print(f"Total parameters: {compatibility['total_params']:,}") + print(f"Trainable parameters: {compatibility['trainable_params']:,}") + + # If a module path is specified, try to extract it + if module_path: + try: + model = load_pretrained_model_from_checkpoint(checkpoint_path=checkpoint_path) + extracted_module = extract_module_from_model(model, module_path) + if extracted_module is not None: + print(f"āœ… Module at path '{module_path}' can be extracted!") + module_params = sum(p.numel() for p in extracted_module.parameters()) + print(f"Module parameters: {module_params:,}") + else: + print(f"āŒ Module at path '{module_path}' not found in model") + except Exception as e: + print(f"āŒ Error extracting module: {e}") + + # Try loading with the config if provided + if config_path and Path(config_path).exists(): + try: + with hydra.initialize_config_dir(config_dir=str(Path(config_path).parent.absolute())): + cfg = hydra.compose(config_name=Path(config_path).stem) + if isinstance(cfg, DictConfig): + target_model = hydra.utils.instantiate(cfg) + if isinstance(target_model, compatibility['model_class']): + print("āœ… Config model class matches checkpoint model class!") + else: + print(f"āš ļø Config model class ({type(target_model).__name__}) differs from checkpoint ({compatibility['model_class'].__name__})") + except Exception as e: + print(f"āš ļø Could not instantiate model from config: {e}") + else: + print("āŒ Checkpoint cannot be loaded as a model") + if 'error' in compatibility: + print(f"Error: {compatibility['error']}") + + except Exception as e: + print(f"āŒ Error checking compatibility: {e}") + + +def extract_and_save_module( + checkpoint_path: str, + module_path: str, + output_path: str +): + """Extract a specific module from a checkpoint and save it separately.""" + print(f"\nšŸ“¤ Extracting module: {module_path}") + print(f"From: {checkpoint_path}") + print(f"To: {output_path}") + print("=" * 60) + + try: + # Load the full model + model = load_pretrained_model_from_checkpoint(checkpoint_path=checkpoint_path) + if model is None: + print("āŒ Could not load model from checkpoint") + return + + # Extract the specific module + extracted_module = extract_module_from_model(model, module_path) + if extracted_module is None: + print(f"āŒ Module at path '{module_path}' not found in model") + return + + # Save the extracted module + # We'll save it as a state dict that can be loaded later + module_state_dict = extracted_module.state_dict() + + save_data = { + 'state_dict': module_state_dict, + 'module_class': type(extracted_module).__name__, + 'module_path': module_path, + 'source_checkpoint': checkpoint_path + } + + torch.save(save_data, output_path) + + param_count = sum(tensor.numel() for tensor in module_state_dict.values()) + print(f"āœ… Extracted {len(module_state_dict)} parameters ({param_count:,} elements)") + print(f"Module class: {type(extracted_module).__name__}") + print(f"Saved to: {output_path}") + + except Exception as e: + print(f"āŒ Error extracting module: {e}") + + +def main(): + parser = argparse.ArgumentParser(description="Inspect pretrained checkpoints for spatiotemporal models") + parser.add_argument("command", choices=["inspect", "check", "extract"], help="Command to run") + + # Common arguments + parser.add_argument("--checkpoint", type=str, help="Path to checkpoint file") + parser.add_argument("--wandb_run", type=str, help="WandB run path (e.g., user/project/run_id)") + parser.add_argument("--checkpoint_type", type=str, default="best_so_far", + help="Type of checkpoint to load") + + # Inspect command arguments + parser.add_argument("--max_depth", type=int, default=2, + help="Maximum depth for structure inspection") + + # Check command arguments + parser.add_argument("--config", type=str, help="Path to model config file") + parser.add_argument("--module_path", type=str, + help="Module path to extract (e.g., 'conditioner.spatiotemporal_model.spatial_module')") + + # Extract command arguments + parser.add_argument("--output", type=str, help="Output path for extracted module") + + args = parser.parse_args() + + # Get checkpoint path + if args.checkpoint: + checkpoint_path = args.checkpoint + elif args.wandb_run: + checkpoint_path = find_checkpoint( + wandb_train_run_path=args.wandb_run, + checkpoint_type=args.checkpoint_type + ) + else: + print("āŒ Must specify either --checkpoint or --wandb_run") + sys.exit(1) + + # Execute command + if args.command == "inspect": + print_checkpoint_structure(checkpoint_path, args.max_depth) + + elif args.command == "check": + if not args.config: + print("āŒ --config is required for check command") + sys.exit(1) + check_compatibility_with_config(checkpoint_path, args.config, args.module_path) + + elif args.command == "extract": + if not args.module_path or not args.output: + print("āŒ --module_path and --output are required for extract command") + sys.exit(1) + extract_and_save_module(checkpoint_path, args.module_path, args.output) + + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/src/jamun/utils/pretrained.py b/src/jamun/utils/pretrained.py new file mode 100644 index 0000000..57788cf --- /dev/null +++ b/src/jamun/utils/pretrained.py @@ -0,0 +1,235 @@ +"""Utilities for loading pretrained models from checkpoints.""" + +import logging +import os +import torch +import torch.nn as nn +import lightning.pytorch as pl +from typing import Dict, Optional, Union, Any +from pathlib import Path + +from jamun.utils.checkpoint import find_checkpoint + +py_logger = logging.getLogger("jamun") + + +def load_checkpoint_state_dict(checkpoint_path: str) -> Dict[str, torch.Tensor]: + """Load state dict from a checkpoint file.""" + if not os.path.exists(checkpoint_path): + raise FileNotFoundError(f"Checkpoint file not found: {checkpoint_path}") + + checkpoint = torch.load(checkpoint_path, map_location='cpu') + + # Handle different checkpoint formats + if 'state_dict' in checkpoint: + return checkpoint['state_dict'] + elif isinstance(checkpoint, dict) and any(k.startswith('model.') for k in checkpoint.keys()): + return checkpoint + else: + raise ValueError(f"Unrecognized checkpoint format in {checkpoint_path}") + + +def load_pretrained_model_from_checkpoint( + checkpoint_path: Optional[str] = None, + wandb_run_path: Optional[str] = None, + checkpoint_type: str = "best_so_far", + model_class: Optional[type] = None +) -> Optional[pl.LightningModule]: + """ + Load an entire pretrained model from checkpoint. + + Args: + checkpoint_path: Direct path to checkpoint file (mutually exclusive with wandb_run_path) + wandb_run_path: WandB run path to find checkpoint (mutually exclusive with checkpoint_path) + checkpoint_type: Type of checkpoint to load ("best_so_far", "last", etc.) + model_class: Optional model class to use for loading (if checkpoint doesn't contain class info) + + Returns: + Loaded model or None if loading failed + """ + if not checkpoint_path and not wandb_run_path: + py_logger.warning("No checkpoint path or wandb run path provided, skipping pretrained loading") + return None + + if checkpoint_path and wandb_run_path: + raise ValueError("Cannot specify both checkpoint_path and wandb_run_path") + + try: + # Find the checkpoint file + if wandb_run_path: + checkpoint_path = find_checkpoint( + wandb_train_run_path=wandb_run_path, + checkpoint_type=checkpoint_type + ) + + py_logger.info(f"Loading pretrained model from: {checkpoint_path}") + + # Load the entire model from checkpoint + if model_class: + model = model_class.load_from_checkpoint(checkpoint_path, strict=False) + else: + # Try to auto-detect model class from checkpoint + checkpoint = torch.load(checkpoint_path, map_location='cpu') + if 'hyper_parameters' in checkpoint and '_target_' in checkpoint['hyper_parameters']: + # Try to import and use the model class from checkpoint + import importlib + target = checkpoint['hyper_parameters']['_target_'] + module_path, class_name = target.rsplit('.', 1) + module = importlib.import_module(module_path) + model_class = getattr(module, class_name) + model = model_class.load_from_checkpoint(checkpoint_path, strict=False) + else: + py_logger.error("Cannot determine model class from checkpoint and no model_class provided") + return None + + py_logger.info(f"Successfully loaded pretrained model of type {type(model).__name__}") + return model + + except Exception as e: + py_logger.error(f"Error loading pretrained model: {e}") + return None + + +def extract_module_from_model(model: pl.LightningModule, module_path: str) -> Optional[nn.Module]: + """ + Extract a specific module from a loaded model using dot notation. + + Args: + model: Loaded PyTorch Lightning model + module_path: Dot-separated path to module (e.g., "conditioner.spatiotemporal_model.spatial_module") + + Returns: + Extracted module or None if not found + """ + try: + current = model + for attr in module_path.split('.'): + if hasattr(current, attr): + current = getattr(current, attr) + else: + py_logger.warning(f"Module path '{module_path}' not found in model") + return None + + py_logger.info(f"Successfully extracted module at path: {module_path}") + return current + + except Exception as e: + py_logger.error(f"Error extracting module '{module_path}': {e}") + return None + + +def load_pretrained_module_from_checkpoint( + checkpoint_path: Optional[str] = None, + wandb_run_path: Optional[str] = None, + checkpoint_type: str = "best_so_far", + module_path: Optional[str] = None, + model_class: Optional[type] = None +) -> Optional[nn.Module]: + """ + Load a specific module from a pretrained model checkpoint. + + Args: + checkpoint_path: Direct path to checkpoint file + wandb_run_path: WandB run path to find checkpoint + checkpoint_type: Type of checkpoint to load + module_path: Dot notation path to extract specific module (e.g., "conditioner.spatiotemporal_model.spatial_module") + model_class: Optional model class for loading + + Returns: + Extracted module or None if loading failed + """ + # Load the full model + model = load_pretrained_model_from_checkpoint( + checkpoint_path=checkpoint_path, + wandb_run_path=wandb_run_path, + checkpoint_type=checkpoint_type, + model_class=model_class + ) + + if model is None: + return None + + # Extract the specific module if path provided + if module_path: + return extract_module_from_model(model, module_path) + else: + # Return the entire model if no specific module path + return model + + +def inspect_model_structure(model: pl.LightningModule, max_depth: int = 3) -> None: + """Print the structure of a loaded model.""" + print(f"\nšŸ“ Model structure: {type(model).__name__}") + print("=" * 60) + + def print_module_tree(module, prefix="", depth=0): + if depth >= max_depth: + return + + for name, child in module.named_children(): + full_name = f"{prefix}.{name}" if prefix else name + param_count = sum(p.numel() for p in child.parameters()) + + print(f"{' ' * depth}ā”œā”€ {name}: {type(child).__name__} ({param_count:,} params)") + + if depth < max_depth - 1: + print_module_tree(child, full_name, depth + 1) + + print_module_tree(model) + + total_params = sum(p.numel() for p in model.parameters()) + trainable_params = sum(p.numel() for p in model.parameters() if p.requires_grad) + print(f"\nšŸ“Š Total parameters: {total_params:,}") + print(f"šŸ“Š Trainable parameters: {trainable_params:,}") + + +def check_model_compatibility( + checkpoint_path: Optional[str] = None, + wandb_run_path: Optional[str] = None, + checkpoint_type: str = "best_so_far", + expected_model_class: Optional[type] = None +) -> Dict[str, Any]: + """ + Check if a checkpoint can be loaded and optionally verify model class. + + Returns: + Dict with 'loadable', 'model_class', 'error' info + """ + try: + if wandb_run_path: + checkpoint_path = find_checkpoint( + wandb_train_run_path=wandb_run_path, + checkpoint_type=checkpoint_type + ) + + # Try loading the model + model = load_pretrained_model_from_checkpoint(checkpoint_path=checkpoint_path) + + if model is None: + return { + 'loadable': False, + 'model_class': None, + 'error': 'Failed to load model from checkpoint' + } + + model_class = type(model) + class_compatible = True + + if expected_model_class: + class_compatible = isinstance(model, expected_model_class) + + return { + 'loadable': True, + 'model_class': model_class, + 'class_compatible': class_compatible, + 'checkpoint_path': checkpoint_path, + 'total_params': sum(p.numel() for p in model.parameters()), + 'trainable_params': sum(p.numel() for p in model.parameters() if p.requires_grad) + } + + except Exception as e: + return { + 'loadable': False, + 'model_class': None, + 'error': str(e) + } \ No newline at end of file diff --git a/src/jamun/utils/pretrained_wrapper.py b/src/jamun/utils/pretrained_wrapper.py new file mode 100644 index 0000000..d1ea1e6 --- /dev/null +++ b/src/jamun/utils/pretrained_wrapper.py @@ -0,0 +1,216 @@ +""" +Pretrained model wrapper utilities for seamless integration with Hydra configs. +""" + +import torch +import torch.nn as nn +from typing import Optional, Union +import logging + +from jamun.utils.pretrained import load_pretrained_model_from_checkpoint +from jamun.utils import mean_center_f, unsqueeze_trailing +from jamun.model import Denoiser +from jamun.utils import find_checkpoint + + +def compute_normalization_factors( + sigma: float | torch.Tensor, + *, + average_squared_distance: float, + normalization_type: str | None, + sigma_data: float | None = None, + D: int = 3, + device: torch.device | None = None, +) -> tuple[float, float, float, float]: + """Compute the normalization factors for the input, skip connection, output, and noise.""" + sigma = torch.as_tensor(sigma, device=device) + + if normalization_type is None: + c_in = torch.as_tensor(1.0, device=device) + c_skip = torch.as_tensor(0.0, device=device) + c_out = torch.as_tensor(1.0, device=device) + c_noise = torch.as_tensor(sigma, device=device) + return c_in, c_skip, c_out, c_noise + + if normalization_type == "EDM": + c_skip = (sigma_data**2) / (sigma**2 + sigma_data**2) + c_out = sigma * sigma_data / torch.sqrt(sigma_data**2 + sigma**2) + c_in = 1 / torch.sqrt(sigma**2 + sigma_data**2) + c_noise = torch.log(sigma / sigma_data) * 0.25 + return c_in, c_skip, c_out, c_noise + + if normalization_type == "JAMUN": + A = torch.as_tensor(average_squared_distance, device=device) + B = torch.as_tensor(2 * D * sigma**2, device=device) + + c_in = 1.0 / torch.sqrt(A + B) + c_skip = A / (A + B) + c_out = torch.sqrt((A * B) / (A + B)) + c_noise = torch.log(sigma) / 4 + return c_in, c_skip, c_out, c_noise + + raise ValueError(f"Unknown normalization type: {normalization_type}") + + + +class DenoiserWrapper(nn.Module): + """ + Wrapper around a denoiser model that matches the spatial module interface. + + This allows pretrained denoiser models to be used as spatial/temporal modules + in the spatiotemporal architecture by replicating the full denoiser logic + including normalization factors computed from the denoiser's own parameters. + """ + + def __init__(self, denoiser_model: nn.Module, c_in: float = 1.0, trainable: bool = True): + """ + Initialize the wrapper. + + Args: + denoiser_model: The pretrained denoiser model + c_in: Rescaling factor to convert positions from overlaying model scale + trainable: Whether to keep the model trainable (default: True) + """ + super().__init__() + self.denoiser = denoiser_model + self.c_in = c_in + + # Set trainability + if not trainable: + for param in self.denoiser.parameters(): + param.requires_grad = False + + def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cutoff): + """ + Forward pass that replicates the denoiser's xhat and xhat_normalized methods. + + Args: + pos: Node positions (input to spatial module) + topology: Graph topology information (already contains bonded edges) + batch: Batch indices + num_graphs: Number of graphs in batch + c_noise: Noise conditioning parameter (already computed) + effective_radial_cutoff: Radial cutoff + + Returns: + Denoised positions from the pretrained model + """ + # Sample sigma from the denoiser's own sigma distribution + sigma = self.denoiser.sigma_distribution.sample().to(pos.device) + + # Rescale positions from overlaying model scale + y = pos / self.c_in + + # Replicate xhat logic + if self.denoiser.mean_center: + y = mean_center_f(y, batch, num_graphs) + + # Replicate xhat_normalized logic + # Compute the normalization factors for the rescaled positions + c_in, c_skip, c_out, _ = compute_normalization_factors( + sigma, + average_squared_distance=self.denoiser.average_squared_distance, + normalization_type=self.denoiser.normalization_type, + sigma_data=self.denoiser.sigma_data, + D=y.shape[-1], + device=y.device, + ) + + # Adjust dimensions + c_in = unsqueeze_trailing(c_in, y.ndim - 1) + c_skip = unsqueeze_trailing(c_skip, y.ndim - 1) + c_out = unsqueeze_trailing(c_out, y.ndim - 1) + c_noise = c_noise.unsqueeze(0) if c_noise.dim() == 0 else c_noise + + # Scale input positions by c_in + y_scaled = y * c_in + + # Call the denoiser's architecture (topology already has edges) + g_pred = self.denoiser.g( + pos=y_scaled, + topology=topology, + c_noise=c_noise, + effective_radial_cutoff=effective_radial_cutoff, + batch=batch, + num_graphs=num_graphs, + ) + + # Compute final prediction with skip connection + xhat = c_skip * y + c_out * g_pred + + # Mean center the prediction if needed + if self.denoiser.mean_center: + xhat = mean_center_f(xhat, batch, num_graphs) + + return xhat + + +def return_wrapped_denoiser( + wandb_run_path: Optional[str] = None, + checkpoint_dir: Optional[str] = None, + checkpoint_type: str = "best_so_far", + c_in: float = 1.0, + trainable: bool = True +) -> DenoiserWrapper: + """ + Load a pretrained denoiser model and return it wrapped for use in spatiotemporal architecture. + + This function is designed to be used directly as a _target_ in Hydra configs. + The wrapper replicates the full denoiser logic including normalization factors + computed from the denoiser's own training parameters. + + Args: + wandb_run_path: Path to wandb run (e.g., "entity/project/run_id") + checkpoint_path: Direct path to checkpoint file + checkpoint_type: Type of checkpoint to load ("best_so_far", "latest", etc.) + c_in: Rescaling factor to convert positions from overlaying model scale + trainable: Whether to keep the loaded model trainable + + Returns: + DenoiserWrapper containing the pretrained model + + Example usage in config: + spatial_module: + _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser + wandb_run_path: "your_entity/your_project/run_id" + c_in: 1.0 + trainable: false + """ + py_logger = logging.getLogger("jamun") + + if not wandb_run_path and not checkpoint_dir: + raise ValueError("Either wandb_run_path or checkpoint_path must be provided") + + # Load the pretrained model + py_logger.info(f"Loading pretrained denoiser from: {wandb_run_path or checkpoint_dir}") + + # pretrained_model = load_pretrained_model_from_checkpoint( + # checkpoint_path=checkpoint_path, + # wandb_run_path=wandb_run_path, + # checkpoint_type=checkpoint_type + # ) + checkpoint_path = find_checkpoint(wandb_train_run_path=wandb_run_path, checkpoint_dir=checkpoint_dir, checkpoint_type=checkpoint_type) + pretrained_model = Denoiser.load_from_checkpoint(checkpoint_path) + + if pretrained_model is None: + raise RuntimeError(f"Failed to load pretrained model from {wandb_run_path or checkpoint_path}") + + py_logger.info("āœ“ Successfully loaded pretrained denoiser") + + # Wrap the model + wrapped_model = DenoiserWrapper(pretrained_model, c_in=c_in, trainable=trainable) + + py_logger.info(f"āœ“ Using c_in rescaling factor: {c_in}") + py_logger.info(f"āœ“ Using denoiser's own normalization parameters:") + py_logger.info(f" - normalization_type: {pretrained_model.normalization_type}") + py_logger.info(f" - average_squared_distance: {pretrained_model.average_squared_distance}") + if hasattr(pretrained_model, 'sigma_data') and pretrained_model.sigma_data is not None: + py_logger.info(f" - sigma_data: {pretrained_model.sigma_data}") + py_logger.info(f" - mean_center: {pretrained_model.mean_center}") + + if not trainable: + py_logger.info("āœ“ Frozen pretrained denoiser (not trainable)") + else: + py_logger.info("āœ“ Pretrained denoiser is trainable") + + return wrapped_model \ No newline at end of file diff --git a/src/jamun/utils/sampling_wrapper.py b/src/jamun/utils/sampling_wrapper.py index c50f2a6..3a72d5f 100644 --- a/src/jamun/utils/sampling_wrapper.py +++ b/src/jamun/utils/sampling_wrapper.py @@ -5,6 +5,7 @@ from jamun.utils import mean_center from typing import Dict, List, Optional +from e3tools import scatter class ModelSamplingWrapper: """Wrapper to sample positions from a model.""" @@ -104,7 +105,8 @@ def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sig if hasattr(self.init_graphs, 'hidden_state') and self.init_graphs.hidden_state: for i in range(len(self.init_graphs.hidden_state)): # Mean center each hidden state in-place - self.init_graphs.hidden_state[i] = self.init_graphs.hidden_state[i] - self.init_graphs.hidden_state[i].mean(dim=0, keepdim=True) + mean = scatter(self.init_graphs.hidden_state[i], self.init_graphs.batch, dim=0, reduce="mean") + self.init_graphs.hidden_state[i] = self.init_graphs.hidden_state[i] - mean[self.init_graphs.batch] @property def device(self) -> torch.device: diff --git a/test_spatiotemporal_conditioner.py b/test_spatiotemporal_conditioner.py index a86f974..055fc8b 100755 --- a/test_spatiotemporal_conditioner.py +++ b/test_spatiotemporal_conditioner.py @@ -90,8 +90,8 @@ def create_spatiotemporal_model() -> E3SpatioTemporal: temporal_module = create_temporal_module() # Create pooling modules - spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr() - temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean() + spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr(irreps_out="3x1e") # Match spatial module output + temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean(irreps_out="3x1e") # Match temporal module output # Compute radial cutoff using temporal average squared distance print("Computing radial cutoff from temporal dataset...") From 5d3e3b32ead00743bc852a708395b881cab64387 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 8 Aug 2025 21:51:57 +0000 Subject: [PATCH 20/32] Fixes: Spatiotemporal model has "ones" facility, for more elegantly turning off positional interactions The default positional attributes are now 3x0e--this should ensure full equivariance Spatiotemporal config has been broken down into standard and conditional Multimeasurement bug fix Sweep script for delta, friction sweep --- ...yaml => sample_enhanced_conditioning.yaml} | 10 +- .../sample_enhanced_conditioning_sweep.yaml | 106 + .../experiment/sample_enhanced_standard.yaml | 106 + ...l => train_enhanced_multimeasurement.yaml} | 0 ...pretrained_spatiotemporal_conditioner.yaml | 4 +- ...n_enhanced_spatiotemporal_conditioner.yaml | 6 +- .../train_enhanced_standard_jamun.yaml | 2 +- .../ala_ala_denoiser_experiment_model1.yaml | 0 .../ala_ala_denoiser_experiment_model2.yaml | 0 .../ala_ala_denoiser_experiment_model3.yaml | 0 .../ala_ala_denoiser_experiment_model4.yaml | 0 ...a_ala_denoiser_experiment_spike_check.yaml | 0 scripts/analyze_sweep_results.py | 181 + scripts/slurm/sample_delta_friction_sweep.sh | 39 + .../slurm/train_enhanced_sampling_single.sh | 4 +- ...nced_sampling_spatiotemporal_comparison.sh | 6 +- src/jamun/cmdline/train.py | 4 +- .../callbacks/model_checkpoint.yaml | 1 + .../model/conditioner/spatiotemporal.yaml | 3 +- .../model/denoiser_multimeasurement.yaml | 13 +- src/jamun/model/arch/spatiotemporal.py | 115 +- src/jamun/model/conditioners/conditioners.py | 2 +- src/jamun/model/denoiser.py | 15 +- src/jamun/model/denoiser_conditional.py | 32 +- src/jamun/model/denoiser_multimeasurement.py | 39 +- src/jamun/model/denoiser_spiked.py | 7 + src/jamun/utils/pretrained_wrapper.py | 19 +- src/jamun/utils/sampling_wrapper.py | 8 +- test_spatiotemporal_conditioner.py | 433 -- uv.lock | 3618 ++++++++++------- 30 files changed, 2855 insertions(+), 1918 deletions(-) rename configs/experiment/{sample_ala_ala_enhanced_sampling_data.yaml => sample_enhanced_conditioning.yaml} (94%) create mode 100644 configs/experiment/sample_enhanced_conditioning_sweep.yaml create mode 100644 configs/experiment/sample_enhanced_standard.yaml rename configs/experiment/{train_ala_ala_enhanced_full_grid_multimeasurement.yaml => train_enhanced_multimeasurement.yaml} (100%) rename configs/{experiment => noise_check}/ala_ala_denoiser_experiment_model1.yaml (100%) rename configs/{experiment => noise_check}/ala_ala_denoiser_experiment_model2.yaml (100%) rename configs/{experiment => noise_check}/ala_ala_denoiser_experiment_model3.yaml (100%) rename configs/{experiment => noise_check}/ala_ala_denoiser_experiment_model4.yaml (100%) rename configs/{experiment => noise_check}/ala_ala_denoiser_experiment_spike_check.yaml (100%) create mode 100755 scripts/analyze_sweep_results.py create mode 100644 scripts/slurm/sample_delta_friction_sweep.sh delete mode 100755 test_spatiotemporal_conditioner.py diff --git a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml b/configs/experiment/sample_enhanced_conditioning.yaml similarity index 94% rename from configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml rename to configs/experiment/sample_enhanced_conditioning.yaml index 8daa916..aa39cfe 100644 --- a/configs/experiment/sample_ala_ala_enhanced_sampling_data.yaml +++ b/configs/experiment/sample_enhanced_conditioning.yaml @@ -29,14 +29,18 @@ init_datasets: max_datasets: 10 label_override: "ALA_ALA" +# model: +# conditioner: +# N_structures: ${init_datasets.total_lag_time} + num_sampling_steps_per_batch: 1000 -num_batches: 10 +num_batches: 1 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/2k3e9l7x +wandb_train_run_path: "sule-shashank/jamun/i0rd0uoa" # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last @@ -99,4 +103,4 @@ logger: wandb: group: sample_enhanced_sampling_data notes: "Sampling from enhanced sampling data using memory sampler" - tags: ["sample", "enhanced_sampling", "memory_sampler", "ALA_ALA", "spatiotemporal transformer", "mean temporal pooler"] \ No newline at end of file + tags: ["sample", "enhanced_sampling", "memory_sampler", "ALA_ALA", "spatiotemporal conditioner"] \ No newline at end of file diff --git a/configs/experiment/sample_enhanced_conditioning_sweep.yaml b/configs/experiment/sample_enhanced_conditioning_sweep.yaml new file mode 100644 index 0000000..223d0a9 --- /dev/null +++ b/configs/experiment/sample_enhanced_conditioning_sweep.yaml @@ -0,0 +1,106 @@ +# @package _global_ + +defaults: + - override /model: denoiser_conditional_pretrained.yaml + - override /callbacks: null + +# init_datasets: +# _target_: jamun.data.parse_datasets_from_directory +# root: "${paths.data_path}/capped_diamines/timewarp_splits/train" +# traj_pattern: "^(.*).xtc" +# pdb_pattern: "^(.*).pdb" +# filter_codes: ['ALA_ALA'] +# as_iterable: false +# subsample: 10 +# total_lag_time: 5 +# lag_subsample_rate: 1 +# max_datasets: 1 +# label_override: "ALA_ALA" + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + as_iterable: false + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 1 + label_override: "ALA_ALA" + +# model: +# conditioner: +# N_structures: ${init_datasets.total_lag_time} + +num_sampling_steps_per_batch: 1 +num_batches: 1 +num_init_samples_per_dataset: 1 +repeat_init_samples: 1 +continue_chain: true + +# Add your checkpoint path here - update with actual trained model path +wandb_train_run_path: "sule-shashank/jamun/i0rd0uoa" +# checkpoint_type: last +# checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" +checkpoint_type: last + +sigma: 0.04 +M: 1.0 +delta: 0.04 +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: null + +sampler: + _target_: jamun.sampling.SamplerMemory + devices: 1 + +# Evaluation dataset - standard ALA_ALA from capped diamines for computing metrics +eval_dataset: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 10 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 1 + label_override: "ALA_ALA" + +# Override ALL callbacks to use eval_dataset for metrics computation +callbacks: + measure_sampling_time: + _target_: jamun.callbacks.sampler.MeasureSamplingTimeCallback + chemical_validity: + _target_: jamun.callbacks.sampler.ChemicalValidityMetricsCallback + datasets: ${eval_dataset} + bond_length_tolerance: 0.2 + volume_exclusion_tolerance: 0.1 + num_molecules_per_trajectory: 100 + ramachandran_plot: + _target_: jamun.callbacks.sampler.RamachandranPlotMetricsCallback + datasets: ${eval_dataset} + trajectory_visualizer: + _target_: jamun.callbacks.sampler.TrajectoryVisualizerCallback + datasets: ${eval_dataset} + num_frames_to_animate: 100 + sample_visualizer: + _target_: jamun.callbacks.sampler.SampleVisualizerCallback + datasets: ${eval_dataset} + num_samples_to_plot: 16 + subsample: 100 + score_distribution: + _target_: jamun.callbacks.sampler.ScoreDistributionCallback + datasets: ${eval_dataset} + save_trajectory: + _target_: jamun.callbacks.sampler.SaveTrajectoryCallback + datasets: ${eval_dataset} + +logger: + wandb: + group: sampling_hyperparameter_sweep + notes: "Tuning hyperparameter sweep for memory sampler" + tags: ["sample", "enhanced_sampling", "memory_sampler", "ALA_ALA", "spatiotemporal conditioner", "sweep"] \ No newline at end of file diff --git a/configs/experiment/sample_enhanced_standard.yaml b/configs/experiment/sample_enhanced_standard.yaml new file mode 100644 index 0000000..ed6bcd8 --- /dev/null +++ b/configs/experiment/sample_enhanced_standard.yaml @@ -0,0 +1,106 @@ +# @package _global_ + +defaults: + - override /model: denoiser_pretrained.yaml + - override /callbacks: null + +# init_datasets: +# _target_: jamun.data.parse_datasets_from_directory +# root: "${paths.data_path}/capped_diamines/timewarp_splits/train" +# traj_pattern: "^(.*).xtc" +# pdb_pattern: "^(.*).pdb" +# filter_codes: ['ALA_ALA'] +# as_iterable: false +# subsample: 10 +# total_lag_time: 5 +# lag_subsample_rate: 1 +# max_datasets: 1 +# label_override: "ALA_ALA" + +init_datasets: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + as_iterable: false + subsample: 1 + total_lag_time: 2 + lag_subsample_rate: 1 + max_datasets: 10 + label_override: "ALA_ALA" + +# model: +# conditioner: +# N_structures: ${init_datasets.total_lag_time} + +num_sampling_steps_per_batch: 1000 +num_batches: 10 +num_init_samples_per_dataset: 1 +repeat_init_samples: 1 +continue_chain: true + +# Add your checkpoint path here - update with actual trained model path +wandb_train_run_path: sule-shashank/jamun/cz5a6xo8 +# checkpoint_type: last +# checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" +checkpoint_type: last + +sigma: 0.04 +M: 1.0 +delta: 0.04 +friction: 1.0 +inverse_temperature: 1.0 +score_fn_clip: null + +sampler: + _target_: jamun.sampling.Sampler + devices: 1 + +# Evaluation dataset - standard ALA_ALA from capped diamines for computing metrics +eval_dataset: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 10 + total_lag_time: 5 + lag_subsample_rate: 1 + max_datasets: 1 + label_override: "ALA_ALA" + +# Override ALL callbacks to use eval_dataset for metrics computation +callbacks: + measure_sampling_time: + _target_: jamun.callbacks.sampler.MeasureSamplingTimeCallback + chemical_validity: + _target_: jamun.callbacks.sampler.ChemicalValidityMetricsCallback + datasets: ${eval_dataset} + bond_length_tolerance: 0.2 + volume_exclusion_tolerance: 0.1 + num_molecules_per_trajectory: 100 + ramachandran_plot: + _target_: jamun.callbacks.sampler.RamachandranPlotMetricsCallback + datasets: ${eval_dataset} + trajectory_visualizer: + _target_: jamun.callbacks.sampler.TrajectoryVisualizerCallback + datasets: ${eval_dataset} + num_frames_to_animate: 100 + sample_visualizer: + _target_: jamun.callbacks.sampler.SampleVisualizerCallback + datasets: ${eval_dataset} + num_samples_to_plot: 16 + subsample: 100 + score_distribution: + _target_: jamun.callbacks.sampler.ScoreDistributionCallback + datasets: ${eval_dataset} + save_trajectory: + _target_: jamun.callbacks.sampler.SaveTrajectoryCallback + datasets: ${eval_dataset} + +logger: + wandb: + group: sample_enhanced_sampling_data + notes: "Benchmark sampling from enhanced sampling data via standard jamun" + tags: ["sample", "enhanced_sampling", "standard_jamun", "ALA_ALA"] \ No newline at end of file diff --git a/configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml b/configs/experiment/train_enhanced_multimeasurement.yaml similarity index 100% rename from configs/experiment/train_ala_ala_enhanced_full_grid_multimeasurement.yaml rename to configs/experiment/train_enhanced_multimeasurement.yaml diff --git a/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml index bd179be..9e31096 100644 --- a/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml +++ b/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner.yaml @@ -9,7 +9,7 @@ defaults: data: datamodule: - batch_size: 8 # Reduced batch size due to increased model complexity + batch_size: 32 # Reduced batch size due to increased model complexity datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -43,7 +43,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 100 # Increased due to model complexity + max_epochs: 1 # Increased due to model complexity # devices: 1 # gradient_clip_val: 1.0 # Add gradient clipping for stability diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml index e2aaf54..45ef62f 100644 --- a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml @@ -8,7 +8,7 @@ defaults: data: datamodule: - batch_size: 8 # Reduced batch size due to increased model complexity + batch_size: 32 # Reduced batch size due to increased model complexity datasets: train: _target_: jamun.data.parse_datasets_from_directory @@ -42,7 +42,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 100 # Increased due to model complexity + max_epochs: 1 # Increased due to model complexity # devices: 1 # gradient_clip_val: 1.0 # Add gradient clipping for stability @@ -50,4 +50,4 @@ logger: wandb: group: enhanced_sampling_conditioner_comparison notes: "SpatioTemporalConditioner on enhanced sampling data - processes temporal sequences through spatial and temporal modules" - tags: ["spatiotemporal_conditioner", "enhanced_sampling", "transformer", "e3conv"] \ No newline at end of file + tags: ["spatiotemporal_conditioner", "enhanced_sampling", "transformer", "debug run"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_standard_jamun.yaml b/configs/experiment/train_enhanced_standard_jamun.yaml index 79c85d8..f05a21d 100644 --- a/configs/experiment/train_enhanced_standard_jamun.yaml +++ b/configs/experiment/train_enhanced_standard_jamun.yaml @@ -42,7 +42,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 50 + max_epochs: 100 devices: 1 logger: diff --git a/configs/experiment/ala_ala_denoiser_experiment_model1.yaml b/configs/noise_check/ala_ala_denoiser_experiment_model1.yaml similarity index 100% rename from configs/experiment/ala_ala_denoiser_experiment_model1.yaml rename to configs/noise_check/ala_ala_denoiser_experiment_model1.yaml diff --git a/configs/experiment/ala_ala_denoiser_experiment_model2.yaml b/configs/noise_check/ala_ala_denoiser_experiment_model2.yaml similarity index 100% rename from configs/experiment/ala_ala_denoiser_experiment_model2.yaml rename to configs/noise_check/ala_ala_denoiser_experiment_model2.yaml diff --git a/configs/experiment/ala_ala_denoiser_experiment_model3.yaml b/configs/noise_check/ala_ala_denoiser_experiment_model3.yaml similarity index 100% rename from configs/experiment/ala_ala_denoiser_experiment_model3.yaml rename to configs/noise_check/ala_ala_denoiser_experiment_model3.yaml diff --git a/configs/experiment/ala_ala_denoiser_experiment_model4.yaml b/configs/noise_check/ala_ala_denoiser_experiment_model4.yaml similarity index 100% rename from configs/experiment/ala_ala_denoiser_experiment_model4.yaml rename to configs/noise_check/ala_ala_denoiser_experiment_model4.yaml diff --git a/configs/experiment/ala_ala_denoiser_experiment_spike_check.yaml b/configs/noise_check/ala_ala_denoiser_experiment_spike_check.yaml similarity index 100% rename from configs/experiment/ala_ala_denoiser_experiment_spike_check.yaml rename to configs/noise_check/ala_ala_denoiser_experiment_spike_check.yaml diff --git a/scripts/analyze_sweep_results.py b/scripts/analyze_sweep_results.py new file mode 100755 index 0000000..3e26e65 --- /dev/null +++ b/scripts/analyze_sweep_results.py @@ -0,0 +1,181 @@ +#!/usr/bin/env python3 +""" +Script to analyze results from the delta-friction parameter sweep. +""" + +import numpy as np +import math +import argparse +import wandb +import pandas as pd +import matplotlib.pyplot as plt +import seaborn as sns + +def calculate_parameter_grid(): + """Calculate the parameter grid used in the sweep.""" + sigma = 0.04 + + # Delta values + delta_min = sigma / math.sqrt(5) + delta_max = math.sqrt(5) * sigma + deltas = np.linspace(delta_min, delta_max, 5) + + # Friction values + linear_points = np.linspace(0.01, 0.99, 5) + frictions = [-math.log(p) for p in linear_points] + + return deltas, frictions, sigma + +def fetch_sweep_results(project_name="sule-shashank/jamun"): + """Fetch results from wandb for the parameter sweep.""" + api = wandb.Api() + + # Get runs with the sweep tag + runs = api.runs(project_name, filters={"tags": {"$in": ["sweep", "delta_friction"]}}) + + results = [] + for run in runs: + if run.state == "finished": + # Extract parameters from tags or config + delta = None + friction = None + + # Try to extract from tags first + for tag in run.tags: + if tag.startswith("delta_"): + try: + delta = float(tag.replace("delta_", "")) + except ValueError: + pass + elif tag.startswith("friction_"): + try: + friction = float(tag.replace("friction_", "")) + except ValueError: + pass + + # Try to extract from config if not found in tags + if delta is None: + delta = run.config.get("delta") + if friction is None: + friction = run.config.get("friction") + + if delta is not None and friction is not None: + # Get metrics (adjust these based on what metrics you're interested in) + metrics = {} + if run.summary: + # Add metrics you want to analyze + for key in ["sampling_time", "chemical_validity", "ramachandran_score"]: + if key in run.summary: + metrics[key] = run.summary[key] + + results.append({ + "run_id": run.id, + "run_name": run.name, + "delta": delta, + "friction": friction, + **metrics + }) + + return pd.DataFrame(results) + +def create_heatmaps(df, deltas, frictions): + """Create heatmaps for each metric.""" + if df.empty: + print("No results found to plot.") + return + + # Get metric columns (exclude parameter and metadata columns) + metric_cols = [col for col in df.columns if col not in ["run_id", "run_name", "delta", "friction"]] + + if not metric_cols: + print("No metrics found in the data.") + return + + # Create a figure with subplots for each metric + n_metrics = len(metric_cols) + fig, axes = plt.subplots(1, n_metrics, figsize=(6*n_metrics, 5)) + if n_metrics == 1: + axes = [axes] + + for i, metric in enumerate(metric_cols): + # Create a pivot table for the heatmap + pivot_data = df.pivot(index="friction", columns="delta", values=metric) + + # Create heatmap + sns.heatmap( + pivot_data, + ax=axes[i], + annot=True, + fmt=".4f", + cmap="viridis", + cbar_kws={"label": metric} + ) + axes[i].set_title(f"{metric.replace('_', ' ').title()}") + axes[i].set_xlabel("Delta") + axes[i].set_ylabel("Friction") + + plt.tight_layout() + plt.savefig("sweep_results_heatmap.png", dpi=300, bbox_inches="tight") + plt.show() + +def print_summary(df, deltas, frictions): + """Print a summary of the sweep results.""" + print("\n" + "="*60) + print("PARAMETER SWEEP SUMMARY") + print("="*60) + + print(f"Parameter grid:") + print(f" Deltas: {len(deltas)} values from {deltas[0]:.6f} to {deltas[-1]:.6f}") + print(f" Frictions: {len(frictions)} values from {frictions[0]:.6f} to {frictions[-1]:.6f}") + print(f" Total combinations: {len(deltas) * len(frictions)}") + + print(f"\nResults found: {len(df)} / {len(deltas) * len(frictions)}") + + if not df.empty: + print(f"\nMetrics available:") + metric_cols = [col for col in df.columns if col not in ["run_id", "run_name", "delta", "friction"]] + for metric in metric_cols: + print(f" - {metric}") + + print(f"\nBest performing combinations:") + for metric in metric_cols: + if metric in df.columns: + if "time" in metric.lower(): + # For time metrics, lower is better + best_idx = df[metric].idxmin() + print(f" {metric} (lowest): delta={df.loc[best_idx, 'delta']:.6f}, friction={df.loc[best_idx, 'friction']:.6f}, value={df.loc[best_idx, metric]:.6f}") + else: + # For other metrics, higher is usually better + best_idx = df[metric].idxmax() + print(f" {metric} (highest): delta={df.loc[best_idx, 'delta']:.6f}, friction={df.loc[best_idx, 'friction']:.6f}, value={df.loc[best_idx, metric]:.6f}") + +def main(): + parser = argparse.ArgumentParser(description="Analyze delta-friction parameter sweep results") + parser.add_argument("--project", default="sule-shashank/jamun", help="Wandb project name") + parser.add_argument("--plot", action="store_true", help="Create heatmap plots") + parser.add_argument("--save-csv", help="Save results to CSV file") + + args = parser.parse_args() + + # Calculate parameter grid + deltas, frictions, sigma = calculate_parameter_grid() + + print("Fetching results from wandb...") + df = fetch_sweep_results(args.project) + + # Print summary + print_summary(df, deltas, frictions) + + # Save to CSV if requested + if args.save_csv: + df.to_csv(args.save_csv, index=False) + print(f"\nResults saved to {args.save_csv}") + + # Create plots if requested + if args.plot and not df.empty: + create_heatmaps(df, deltas, frictions) + + print("\nAnalysis complete!") + +if __name__ == "__main__": + main() diff --git a/scripts/slurm/sample_delta_friction_sweep.sh b/scripts/slurm/sample_delta_friction_sweep.sh new file mode 100644 index 0000000..59acb70 --- /dev/null +++ b/scripts/slurm/sample_delta_friction_sweep.sh @@ -0,0 +1,39 @@ +#!/bin/bash +#SBATCH --job-name=sweep_delta_friction +#SBATCH --partition=gpu2 +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=1-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-24 + +# Load environment +source ~/.bashrc +conda activate jamun +cd /homefs/home/sules/jamun +mkdir -p logs + +# Precomputed values for sigma=0.04 +# Delta: 5 values from sigma/sqrt(5) to sqrt(5)*sigma +DELTAS=(0.017889 0.026833 0.040000 0.059665 0.089443) + +# Friction: -log of 5 values from 0.01 to 0.99 +FRICTIONS=(2.52572864 1.2552661 0.71334989 0.36384343 0.10536052) + +# Get parameter values based on array index +DELTA_INDEX=$((SLURM_ARRAY_TASK_ID / 5)) +FRICTION_INDEX=$((SLURM_ARRAY_TASK_ID % 5)) +DELTA=${DELTAS[$DELTA_INDEX]} +FRICTION=${FRICTIONS[$FRICTION_INDEX]} + +echo "Running: delta=$DELTA, friction=$FRICTION (job $SLURM_ARRAY_TASK_ID)" + +# Run experiment +jamun_sample \ + --config-dir=configs \ + experiment=sample_enhanced_conditioning_sweep \ + ++delta=$DELTA \ + ++friction=$FRICTION \ + ++logger.wandb.name="sweep_d${DELTA}_f${FRICTION}_${SLURM_ARRAY_JOB_ID}_${SLURM_ARRAY_TASK_ID}" \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_sampling_single.sh b/scripts/slurm/train_enhanced_sampling_single.sh index 6e5db78..a57ad75 100755 --- a/scripts/slurm/train_enhanced_sampling_single.sh +++ b/scripts/slurm/train_enhanced_sampling_single.sh @@ -5,7 +5,7 @@ #SBATCH --ntasks-per-node 1 #SBATCH --gpus-per-node 1 #SBATCH --cpus-per-task 8 -#SBATCH --time 3-0 +#SBATCH --time 12:00:00 #SBATCH --mem-per-cpu=32G # Initialize conda @@ -20,4 +20,4 @@ echo "Conda environment: $CONDA_DEFAULT_ENV" nvidia-smi # Run training with parameter overrides -jamun_train --config-dir=configs experiment=train_enhanced_spatiotemporal_conditioner.yaml \ No newline at end of file +jamun_train --config-dir=configs experiment=train_enhanced_pretrained_spatiotemporal_conditioner.yaml model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean model.conditioner.spatiotemporal_model.spatial_module.trainable=false trainer.max_epochs=10 logger.wandb.tags=[job_4,spatiotemporal_comparison] \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh b/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh index 16b0086..f38a064 100755 --- a/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh +++ b/scripts/slurm/train_enhanced_sampling_spatiotemporal_comparison.sh @@ -70,7 +70,7 @@ esac CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" # Add pooler override (keeping the irreps_out parameter from base config) -CMD="$CMD model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=${POOLER}" +CMD="$CMD ++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=${POOLER}" # Add trainable override if needed if [ -n "$TRAINABLE_OVERRIDE" ]; then @@ -80,12 +80,12 @@ fi # Add dataset and training overrides # CMD="$CMD data.datamodule.datasets.train.max_datasets=1" # CMD="$CMD data.datamodule.datasets.val.max_datasets=1" -CMD="$CMD trainer.max_epochs=100" +CMD="$CMD ++trainer.max_epochs=100" CMD="$CMD ++wandb.logger.group=spatiotemporal_comparison" # Add job-specific wandb tags WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" -CMD="$CMD +logger.wandb.tags=[\"${WANDB_TAG}\",\"spatiotemporal_comparison\"]" +# CMD="$CMD ++wandb.logger.tags=[${WANDB_TAG},spatiotemporal_comparison]" echo "Running command: $CMD" exec $CMD \ No newline at end of file diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 08ab7cb..5a461ac 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -132,8 +132,8 @@ def run(cfg): cfg.model.conditioner.c_in = c_in_float dist_log(f"Set cfg.model.conditioner.c_in to {c_in_float}") - - if cfg.model.conditioner._target_ == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner": + # breakpoint() + if cfg.model.get("conditioner") and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner": cfg.model.conditioner.spatiotemporal_model.radial_cutoff = average_squared_distance max_radius = cfg.model.max_radius temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) diff --git a/src/jamun/hydra_config/callbacks/model_checkpoint.yaml b/src/jamun/hydra_config/callbacks/model_checkpoint.yaml index da57e74..ddad6cf 100644 --- a/src/jamun/hydra_config/callbacks/model_checkpoint.yaml +++ b/src/jamun/hydra_config/callbacks/model_checkpoint.yaml @@ -4,3 +4,4 @@ model_checkpoint: save_top_k: 5 save_last: true monitor: "val/loss" + every_n_epochs: 1 diff --git a/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml index b10ac95..b34a68a 100644 --- a/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml @@ -31,7 +31,7 @@ spatiotemporal_model: hidden_layer_factory: _target_: e3tools.nn.ConvBlock _partial_: true - conv:--- + conv: _target_: e3tools.nn.Conv # replace with Conv for non-separable case _partial_: true output_head_factory: @@ -47,6 +47,7 @@ spatiotemporal_model: irreps_hidden: "120x0e + 32x1e" irreps_sh: "1x0e + 1x1e" irreps_node_attr: "1x1e" # Match spatial module output + irreps_node_attr_temporal: "3x0e" num_layers: 2 edge_attr_dim: 24 num_attention_heads: 1 diff --git a/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml b/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml index 2def981..e098a80 100644 --- a/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml +++ b/src/jamun/hydra_config/model/denoiser_multimeasurement.yaml @@ -1,6 +1,7 @@ defaults: - arch: e3conv_conditional.yaml - optim: adam.yaml + - conditioner: position_conditioner.yaml - lr_scheduler_config: null - _self_ @@ -18,14 +19,14 @@ torch_compile_kwargs: dynamic: true mode: default -conditioner: - _target_: jamun.model.conditioners.PositionConditioner - N_structures: ${model.arch.N_structures} +# conditioner: +# _target_: jamun.model.conditioners.PositionConditioner +# N_structures: ${model.arch.N_structures} # Multimeasurement specific parameters multimeasurement: true -N_measurements: 1 -N_measurements_hidden: 1 -max_graphs_per_batch: null +N_measurements: 4 +N_measurements_hidden: 4 +max_graphs_per_batch: 32 _target_: jamun.model.DenoiserMultimeasurement \ No newline at end of file diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index b0ac51f..7e5612b 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -21,7 +21,7 @@ from jamun.model.arch.e3conv import E3Conv -def calculate_temporal_positions(temporal_length, device=None): +def calculate_temporal_positions(temporal_length, mode="linear", device=None): """ Calculate normalized temporal positions for nodes in a temporal graph. @@ -35,14 +35,19 @@ def calculate_temporal_positions(temporal_length, device=None): if temporal_length <= 1: return torch.tensor([0.0], device=device) - # Create positions [0, 1, 2, ..., T-1] and normalize by T - positions = torch.arange(temporal_length, dtype=torch.float32, device=device) - normalized_positions = positions / temporal_length + if mode == "linear": + # Create positions [0, 1, 2, ..., T-1] and normalize by T + positions = torch.arange(temporal_length, dtype=torch.float32, device=device) + normalized_positions = positions / temporal_length + elif mode == "zeros": + # Create positions [0, 1, 2, ..., T-1] and normalize by T + positions = torch.arange(temporal_length, dtype=torch.float32, device=device) + positions = torch.zeros_like(positions) return normalized_positions -def spatial_to_temporal_graphs(batch): +def spatial_to_temporal_graphs(batch, temporal_position_mode="linear"): """ Convert a batch of spatial graphs to temporal graphs. @@ -83,7 +88,7 @@ def spatial_to_temporal_graphs(batch): temporal_pos = torch.stack(temporal_positions) # Shape: [T, 3] # Calculate temporal positions for this sequence - temporal_position = calculate_temporal_positions(temporal_length, device=device) + temporal_position = calculate_temporal_positions(temporal_length, mode=temporal_position_mode, device=device) # Create edge connectivity if temporal_length > 1: @@ -108,8 +113,8 @@ def spatial_to_temporal_graphs(batch): temporal_graph = torch_geometric.data.Data( pos=temporal_pos, edge_index=edge_index, - spatial_node_idx=torch.tensor([node_idx], device=device), # Track which spatial node this came from - temporal_length=torch.tensor([temporal_length], device=device), + spatial_node_idx=torch.tensor([node_idx], dtype=torch.long, device=device), # Track which spatial node this came from + temporal_length=torch.tensor([temporal_length], dtype=torch.long, device=device), temporal_position=temporal_position # Normalized position in sequence [0, 1/T, 2/T, ...] ) temporal_graphs.append(temporal_graph) @@ -165,13 +170,18 @@ def __init__( edge_attr_dim: int, num_attention_heads: int, reduce: str | None = None, + irreps_node_attr_temporal: Union[str, e3nn.o3.Irreps] = "1x1e", + radial_edge_attr_encoding_function: str = "gaussian", + node_attr_temporal_encoding_function: str = "gaussian", + edge_attr_temporal_encoding_function: str = "gaussian", ): super().__init__() self.irreps_out = o3.Irreps(irreps_out) self.irreps_hidden = o3.Irreps(irreps_hidden) self.irreps_sh = o3.Irreps(irreps_sh) - self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps + self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps + self.irreps_node_attr_temporal = o3.Irreps(irreps_node_attr_temporal) self.num_layers = num_layers self.edge_attr_dim = edge_attr_dim self.num_attention_heads = num_attention_heads @@ -182,13 +192,16 @@ def __init__( # Split edge attribute dimensions: radial and temporal (bondedness is optional) self.radial_edge_attr_dim = self.edge_attr_dim // 2 self.temporal_edge_attr_dim = self.edge_attr_dim - self.radial_edge_attr_dim - + self.temporal_node_attr_dim = self.irreps_node_attr_temporal.dim # Optional bondedness embedding (only used if bond_mask exists in graph) self.embed_bondedness = nn.Embedding(2, self.edge_attr_dim // 3) - + self.edge_attr_temporal_encoding_function = edge_attr_temporal_encoding_function + self.node_attr_temporal_encoding_function = node_attr_temporal_encoding_function + self.radial_edge_attr_encoding_function = radial_edge_attr_encoding_function # Gate for combining node attributes with temporal position # Input: node_attr (from data) + temporal_position (1x0e scalar) - irreps_with_temporal = self.irreps_node_attr + o3.Irreps("1x0e") + # irreps_with_temporal = self.irreps_node_attr + o3.Irreps("1x0e") + irreps_with_temporal = self.irreps_node_attr + self.irreps_node_attr_temporal self.temporal_gate = e3tools.nn.GateWrapper(irreps_in=irreps_with_temporal, \ irreps_out=self.irreps_hidden, \ irreps_gate=irreps_with_temporal,) @@ -230,26 +243,49 @@ def forward( edge_sh = self.sh(edge_vec) # Compute edge attributes: radial and temporal - radial_edge_attr = e3nn.math.soft_one_hot_linspace( - edge_vec.norm(dim=1), - 0.0, - effective_radial_cutoff, - self.radial_edge_attr_dim, - basis="gaussian", - cutoff=True, - ) + if self.radial_edge_attr_encoding_function is not "ones": + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec.norm(dim=1), + 0.0, + temporal_cutoff, + self.radial_edge_attr_dim, + basis=self.radial_edge_attr_encoding_function, + cutoff=True, + ) + else: + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec.norm(dim=1), + 0.0, + temporal_cutoff, + self.radial_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + radial_edge_attr = torch.ones_like(radial_edge_attr) # Temporal edge attributes from temporal_position differences temporal_edge_vec = temporal_position[src] - temporal_position[dst] - temporal_edge_attr = e3nn.math.soft_one_hot_linspace( - temporal_edge_vec.abs(), # Use absolute difference - 0.0, - temporal_cutoff, - self.temporal_edge_attr_dim, - basis="gaussian", - cutoff=True, - ) - temporal_edge_attr = torch.ones_like(temporal_edge_attr) # TODO: remove this, this is hacking. + if self.edge_attr_temporal_encoding_function is not "ones": + temporal_edge_attr = e3nn.math.soft_one_hot_linspace( + temporal_edge_vec.abs(), # Use absolute difference + 0.0, + 2.0, + self.temporal_edge_attr_dim, + basis=self.edge_attr_temporal_encoding_function, + cutoff=True, + ) + else: + temporal_edge_attr = e3nn.math.soft_one_hot_linspace( + temporal_edge_vec.abs(), # Use absolute difference + 0.0, + 2.0, + self.temporal_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + temporal_edge_attr = torch.ones_like(temporal_edge_attr) + + # temporal_edge_attr = torch.ones_like(temporal_edge_attr) # TODO: remove this, this is hacking. # Optional bondedness (if bond_mask exists in the temporal graph) if hasattr(temporal_graph, 'bond_mask') and temporal_graph.bond_mask is not None: @@ -261,7 +297,26 @@ def forward( # Process node attributes with temporal gating # Concatenate node_attr with temporal_position (scalar) - temporal_position_expanded = temporal_position.unsqueeze(-1) # [N, 1] for concatenation + if self.node_attr_temporal_encoding_function is not "ones": + temporal_position = e3nn.math.soft_one_hot_linspace( + temporal_position, # Use absolute difference + 0.0, # time always starts at 0 + 1.0, # time always ends at 1 + self.temporal_node_attr_dim, + basis=self.node_attr_temporal_encoding_function, + cutoff=True, + ) + else: + temporal_position = e3nn.math.soft_one_hot_linspace( + temporal_position, # Use absolute difference + 0.0, # time always starts at 0 + 1.0, # time always ends at 1 + self.temporal_node_attr_dim, + basis="gaussian", + cutoff=True, + ) + temporal_position = torch.ones_like(temporal_position) + temporal_position_expanded = temporal_position # [N, 1] for concatenation node_attr_with_temporal = torch.cat([node_attr, temporal_position_expanded], dim=-1) # Apply temporal gate diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 64f6ac2..8879c12 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -330,7 +330,7 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: # Prepare noise conditioning device = y_aligned.pos.device sigma = torch.tensor(self.c_noise, device=device) - sigma = unsqueeze_trailing(sigma, 1) + sigma = unsqueeze_trailing(sigma, 1) # actually this is correct, but this is bad variable naming. the positional e3conv will take a c_noise, so this is right, but it is not right to call it sigma. # Process through spatio-temporal model with aligned hidden states # Only disable gradients if the model is frozen diff --git a/src/jamun/model/denoiser.py b/src/jamun/model/denoiser.py index a0a8567..c0f5b0f 100644 --- a/src/jamun/model/denoiser.py +++ b/src/jamun/model/denoiser.py @@ -182,6 +182,13 @@ def __init__( if self.rotational_augmentation: py_logger.info("Rotational augmentation is enabled.") + def on_before_optimizer_step(self, optimizer): + # Log gradients and parameters. + for name, param in self.named_parameters(): + self.log(f"parameter_norms/{name}", param.norm(), sync_dist=True) + if param.grad is not None: + self.log(f"gradient_norms/{name}", param.grad.norm(), sync_dist=True) + def add_noise(self, x: torch.Tensor, sigma: float | torch.Tensor, num_graphs: int) -> torch.Tensor: # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.ndim) @@ -317,8 +324,8 @@ def noise_and_denoise( x = align_A_to_B_batched_f( x, y, - topology.batch, - topology.num_graphs, + batch, + num_graphs, sigma=sigma, correction_order=self.alignment_correction_order, ) @@ -386,8 +393,8 @@ def noise_and_compute_loss( align_noisy_input: bool, ) -> tuple[torch.Tensor, dict[str, torch.Tensor]]: """Add noise to the input and compute the loss.""" - xhat, x, _ = self.noise_and_denoise(x, topology, sigma, align_noisy_input=align_noisy_input) - return self.compute_loss(x, xhat, topology, sigma) + xhat, x, _ = self.noise_and_denoise(x, topology, batch, num_graphs, sigma, align_noisy_input=align_noisy_input) + return self.compute_loss(x, xhat, topology, batch, num_graphs, sigma) def training_step(self, data: torch_geometric.data.Batch, data_idx: int): """Called during training.""" diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index 253fe1b..60761de 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -11,6 +11,7 @@ from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm +import os class Denoiser(pl.LightningModule): @@ -109,6 +110,13 @@ def __init__( raise ValueError("Conditioner must be a callable or None") py_logger.info(f"Conditioner: {self.conditioning_module}") + def on_before_optimizer_step(self, optimizer): + # Log gradients and parameters. + for name, param in self.named_parameters(): + self.log(f"parameter_norms/{name}", param.norm(), sync_dist=True) + if param.grad is not None: + self.log(f"gradient_norms/{name}", param.grad.norm(), sync_dist=True) + def conditioner_default(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures = [y.pos] # Return complete list starting with current position return conditioned_structures @@ -410,8 +418,28 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): batch, sigma, align_noisy_input=align_noisy_input, - ) - + ) # check if the loss is nan. if nan then save the model, and the batch and see what went on. + if torch.isnan(loss.sum()): + print(f"Loss is nan at step {self.global_step}") + print(f"Batch: {batch}") + print(f"Sigma: {sigma}") + print(f"Align noisy input: {align_noisy_input}") + print(f"Loss: {loss}") + print(f"Aux: {aux}") + # Create debug directory if it doesn't exist + debug_dir = f"/homefs/home/sules/jamun/debug_nan_loss_step_{self.global_step}" + os.makedirs(debug_dir, exist_ok=True) + + # Save model checkpoint + checkpoint_path = os.path.join(debug_dir, "model_nan_loss.ckpt") + self.trainer.save_checkpoint(checkpoint_path) + print(f"Model saved to {checkpoint_path}") + + torch.save(batch, debug_dir + "/batch_nan_loss.pt") + + # Optionally raise an exception to stop training + raise RuntimeError(f"NaN loss detected at step {self.global_step}. Debug files saved to {debug_dir}") + # Average the loss and other metrics over all graphs. with torch.cuda.nvtx.range("mean_over_graphs"): aux["loss"] = loss diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py index 79898a5..0c694bb 100644 --- a/src/jamun/model/denoiser_multimeasurement.py +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -11,6 +11,7 @@ from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm +import os class DenoiserMultimeasurement(pl.LightningModule): @@ -125,6 +126,13 @@ def __init__( f"Manual optimization enabled with micro-batch size of {self.max_graphs_per_batch} graphs." ) + def on_before_optimizer_step(self, optimizer): + # Log gradients and parameters. + for name, param in self.named_parameters(): + self.log(f"parameter_norms/{name}", param.norm(), sync_dist=True) + if param.grad is not None: + self.log(f"gradient_norms/{name}", param.grad.norm(), sync_dist=True) + def _align_A_to_B_batched_with_hidden_states( self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch ) -> torch_geometric.data.Batch: @@ -134,12 +142,13 @@ def _align_A_to_B_batched_with_hidden_states( # Align positions A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) - # Align hidden states + # Align hidden states if hasattr(A, "hidden_state") and A.hidden_state is not None: + A_aligned.hidden_state = [] for i in range(len(A.hidden_state)): - A_aligned.hidden_state[i] = kabsch_algorithm( + A_aligned.hidden_state.append(kabsch_algorithm( A.hidden_state[i], B.pos, A.batch, A.num_graphs - ) + )) return A_aligned def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): @@ -575,8 +584,28 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): batch, sigma, align_noisy_input=align_noisy_input, - ) - + ) # check if the loss is nan. if nan then save the model, and the batch and see what went on. + # if torch.isnan(loss.sum()): + # print(f"Loss is nan at step {self.global_step}") + # print(f"Batch: {batch}") + # print(f"Sigma: {sigma}") + # print(f"Align noisy input: {align_noisy_input}") + # print(f"Loss: {loss}") + # print(f"Aux: {aux}") + # # Create debug directory if it doesn't exist + # debug_dir = f"/homefs/home/sules/jamun/debug_nan_loss_step_{self.global_step}" + # os.makedirs(debug_dir, exist_ok=True) + # + # # Save model checkpoint + # checkpoint_path = os.path.join(debug_dir, "model_nan_loss.ckpt") + # self.trainer.save_checkpoint(checkpoint_path) + # print(f"Model saved to {checkpoint_path}") + # + # torch.save(batch, debug_dir + "/batch_nan_loss.pt") + # + # # Optionally raise an exception to stop training + # raise RuntimeError(f"NaN loss detected at step {self.global_step}. Debug files saved to {debug_dir}") + # Average the loss and other metrics over all graphs. with torch.cuda.nvtx.range("mean_over_graphs"): aux["loss"] = loss diff --git a/src/jamun/model/denoiser_spiked.py b/src/jamun/model/denoiser_spiked.py index ed28542..4dd6bd9 100644 --- a/src/jamun/model/denoiser_spiked.py +++ b/src/jamun/model/denoiser_spiked.py @@ -106,6 +106,13 @@ def __init__( raise ValueError("Conditioner must be a callable or None") py_logger.info(f"Conditioner: {self.conditioning_module}") + def on_before_optimizer_step(self, optimizer): + # Log gradients and parameters. + for name, param in self.named_parameters(): + self.log(f"parameter_norms/{name}", param.norm(), sync_dist=True) + if param.grad is not None: + self.log(f"gradient_norms/{name}", param.grad.norm(), sync_dist=True) + def conditioner_default(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: conditioned_structures = [y.pos] # Return complete list starting with current position if x_clean is not None: diff --git a/src/jamun/utils/pretrained_wrapper.py b/src/jamun/utils/pretrained_wrapper.py index d1ea1e6..75570a7 100644 --- a/src/jamun/utils/pretrained_wrapper.py +++ b/src/jamun/utils/pretrained_wrapper.py @@ -121,11 +121,28 @@ def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cu c_skip = unsqueeze_trailing(c_skip, y.ndim - 1) c_out = unsqueeze_trailing(c_out, y.ndim - 1) c_noise = c_noise.unsqueeze(0) if c_noise.dim() == 0 else c_noise + + # Ensure c_noise is float type (fix for dtype mismatch) + c_noise = c_noise.float() # Scale input positions by c_in y_scaled = y * c_in - # Call the denoiser's architecture (topology already has edges) + # # Call the denoiser's architecture (topology already has edges) + # # Add this right before line 129 in the pretrained wrapper call + # print("=== Debugging pretrained denoiser input types ===") + # print(f"y_scaled dtype: {y_scaled.dtype}, shape: {y_scaled.shape}") + # print(f"topology.edge_index dtype: {topology.edge_index.dtype if hasattr(topology, 'edge_index') else 'N/A'}") + # print(f"c_noise dtype: {c_noise.dtype}, shape: {c_noise.shape}") + # print(f"batch dtype: {batch.dtype}, shape: {batch.shape}") + # print(f"effective_radial_cutoff dtype: {type(effective_radial_cutoff)}") + + # # Check if topology has any Long tensors + # for attr_name in dir(topology): + # if not attr_name.startswith('_'): + # attr = getattr(topology, attr_name) + # if isinstance(attr, torch.Tensor): + # print(f"topology.{attr_name} dtype: {attr.dtype}") g_pred = self.denoiser.g( pos=y_scaled, topology=topology, diff --git a/src/jamun/utils/sampling_wrapper.py b/src/jamun/utils/sampling_wrapper.py index 3a72d5f..3be05e8 100644 --- a/src/jamun/utils/sampling_wrapper.py +++ b/src/jamun/utils/sampling_wrapper.py @@ -35,8 +35,12 @@ def score(self, y, sigma, *args, **kwargs): return self._model.score(self.positions_to_graph(y), sigma) def xhat(self, y, sigma, *args, **kwargs): - xhat_graph = self._model.xhat(self.positions_to_graph(y), sigma) - return xhat_graph.pos + data = self.positions_to_graph(y) + y, topology, batch, num_graphs = data.pos, data.clone(), data.batch, data.num_graphs + del topology.batch, topology.num_graphs + sigma = torch.as_tensor(sigma).to(y) + xhat_pos = self._model.xhat(y, topology, batch, num_graphs, sigma) + return xhat_pos def positions_to_graph(self, positions: torch.Tensor) -> torch_geometric.data.Data: """Wraps a tensor of positions to a graph with these positions as an attribute.""" diff --git a/test_spatiotemporal_conditioner.py b/test_spatiotemporal_conditioner.py deleted file mode 100755 index 055fc8b..0000000 --- a/test_spatiotemporal_conditioner.py +++ /dev/null @@ -1,433 +0,0 @@ -#!/usr/bin/env python3 -""" -Test script for loading and testing a conditional denoiser with spatiotemporal conditioner. -Uses the new approach where SpatioTemporalConditioner outputs [y.pos, spatial_features] -and E3ConvConditionalSpatioTemporal handles concatenated inputs. -""" - -import e3nn -e3nn.set_optimization_defaults(jit_script_fx=False) - -import torch -import torch_geometric -from typing import Dict, Any -import sys -import os - -# Add the src directory to path to import jamun modules -sys.path.insert(0, 'src') - -from jamun.data import parse_datasets_from_directory -from jamun.model.denoiser_conditional import Denoiser # Changed from DenoiserWithInputAttr -from jamun.model.denoiser import add_edges # Import add_edges function -from jamun.model.conditioners.conditioners import SpatioTemporalConditioner -from jamun.model.arch.spatiotemporal import E3SpatioTemporal, E3Transformer -from jamun.model.arch.e3conv import E3Conv -from jamun.model.arch.e3conv_conditional import E3ConvConditionalSpatioTemporal # Changed from E3ConvConditionalWithInputAttr -from jamun.model.pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean -from jamun.distributions._distributions import ConstantSigma -from jamun.utils import unsqueeze_trailing -from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets # Import temporal function - -# Setup device -device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') -print(f"Using device: {device}") - -def create_spatial_module() -> E3Conv: - """Create E3Conv spatial module with reasonable parameters.""" - import functools - import e3tools - - # Create factory functions - hidden_layer_factory = functools.partial( - e3tools.nn.ConvBlock, - conv=functools.partial(e3tools.nn.Conv) - ) - - output_head_factory = functools.partial( - e3tools.nn.EquivariantMLP, - irreps_hidden_list=["120x0e + 32x1e"] - ) - - return E3Conv( - irreps_out="3x1e", # Changed to match temporal module input - irreps_hidden="120x0e + 32x1e", - irreps_sh="1x0e + 1x1e", - hidden_layer_factory=hidden_layer_factory, - output_head_factory=output_head_factory, - n_layers=1, - edge_attr_dim=64, - use_residue_information=True, - atom_type_embedding_dim=8, - atom_code_embedding_dim=8, - residue_code_embedding_dim=32, - residue_index_embedding_dim=8, - use_residue_sequence_index=False, - num_atom_types=20, - max_sequence_length=10, - num_atom_codes=10, - num_residue_types=25, - test_equivariance=False, - reduce=None - ) - -def create_temporal_module() -> E3Transformer: - """Create E3Transformer temporal module.""" - return E3Transformer( - irreps_out="3x1e", # Final spatial features output - irreps_hidden="8x0e + 4x1e", - irreps_sh="1x0e + 1x1e", - irreps_node_attr="3x1e", # Match spatial module output - num_layers=2, - edge_attr_dim=24, - num_attention_heads=1, - reduce=None - ) - -def create_spatiotemporal_model() -> E3SpatioTemporal: - """Create the complete E3SpatioTemporal model.""" - spatial_module = create_spatial_module() - temporal_module = create_temporal_module() - - # Create pooling modules - spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr(irreps_out="3x1e") # Match spatial module output - temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean(irreps_out="3x1e") # Match temporal module output - - # Compute radial cutoff using temporal average squared distance - print("Computing radial cutoff from temporal dataset...") - try: - # Load dataset to compute temporal average squared distance - dataset = parse_datasets_from_directory( - root="/data2/sules/ALA_ALA_enhanced_full_grid/train", - traj_pattern="^(.*).xtc", - pdb_pattern="^(.*).pdb", - subsample=1, - total_lag_time=5, - lag_subsample_rate=1, - max_datasets=2, # Keep small for testing - num_frames=5 # Small number of frames - ) - - # Compute temporal average squared distance - temporal_avg_sq_dist = compute_temporal_average_squared_distance_from_datasets( - [dataset], # Pass as list since function expects multiple datasets - num_samples=50, # Use fewer samples for testing - verbose=True - ) - - # Use a multiple of the temporal average squared distance as the radial cutoff - # Typically we might use sqrt(temporal_avg_sq_dist) * some_factor - import math - radial_cutoff = math.sqrt(temporal_avg_sq_dist) * 2.0 # Scale factor of 2.0 - print(f"Computed radial cutoff: {radial_cutoff:.6f} nm") - - except Exception as e: - print(f"Warning: Failed to compute temporal cutoff ({e}), using default value 0.05") - radial_cutoff = 0.05 - - return E3SpatioTemporal( - spatial_module=spatial_module, - temporal_module=temporal_module, - spatial_to_temporal_pooler=spatial_to_temporal_pooler, - temporal_to_spatial_pooler=temporal_to_spatial_pooler, - radial_cutoff=radial_cutoff, - temporal_cutoff=1.0 - ) - -def create_spatiotemporal_conditioner() -> SpatioTemporalConditioner: - """Create SpatioTemporalConditioner with E3SpatioTemporal model.""" - spatiotemporal_model = create_spatiotemporal_model() - - return SpatioTemporalConditioner( - N_structures=1, # Changed to 2 for [y.pos, spatial_features] - spatiotemporal_model=spatiotemporal_model, - c_noise=0.0, - freeze_spatiotemporal_model=False # Keep trainable - ) - -def create_conditional_denoiser_config() -> Dict[str, Any]: - """Create configuration for Denoiser with spatiotemporal conditioner.""" - import functools - import e3tools.nn - - def create_arch(): - """Create the E3ConvConditionalSpatioTemporal architecture module.""" - # Hidden layer factory - hidden_layer_factory = functools.partial( - e3tools.nn.ConvBlock, - conv=functools.partial(e3tools.nn.Conv) - ) - - # Output head factory - output_head_factory = functools.partial( - e3tools.nn.EquivariantMLP, - irreps_hidden_list=["16x0e + 8x1e"] - ) - - return E3ConvConditionalSpatioTemporal( - irreps_out="1x1e", # Output should be 3 components (1x1e) to match position - irreps_hidden="16x0e + 8x1e", - irreps_sh="1x0e + 1x1e", - hidden_layer_factory=hidden_layer_factory, - output_head_factory=output_head_factory, - n_layers=2, - edge_attr_dim=32, - use_residue_information=True, - atom_type_embedding_dim=8, - atom_code_embedding_dim=8, - residue_code_embedding_dim=16, - residue_index_embedding_dim=8, - use_residue_sequence_index=False, - num_atom_types=20, - max_sequence_length=10, - num_atom_codes=10, - num_residue_types=25, - test_equivariance=False, - reduce=None, - N_structures=1, # Changed to 2 for [y.pos, spatial_features] - input_attr_irreps="3x1e", # spatial_features only (9 components = 3x1e) - ) - - def create_optim(params): - """Create the optimizer.""" - return torch.optim.Adam(params, lr=0.001) - - return { - # Required Denoiser parameters (changed from DenoiserWithInputAttr) - 'arch': create_arch, - 'optim': create_optim, - 'sigma_distribution': ConstantSigma(sigma=0.1), - 'max_radius': 1000.0, - 'average_squared_distance': 10.0, # Dummy value for testing - 'add_fixed_noise': False, - 'add_fixed_ones': False, - 'align_noisy_input_during_training': True, - 'align_noisy_input_during_evaluation': True, - 'mean_center': True, - 'mirror_augmentation_rate': 0.0, - 'bond_loss_coefficient': 1.0, - 'normalization_type': "JAMUN", - 'sigma_data': None, - 'lr_scheduler_config': None, - 'use_torch_compile': False, # Disable for testing - 'torch_compile_kwargs': None, - 'conditioner': create_spatiotemporal_conditioner() - } - -def add_edges_to_batch(batch: torch_geometric.data.Batch, cutoff: float = 0.05) -> torch_geometric.data.Batch: - """Add edges to batch using existing utility from denoiser.""" - # Use e3tools radius_graph directly since we don't need the full denoiser add_edges logic - import e3tools - - if hasattr(batch, 'edge_index') and batch.edge_index is not None: - return batch - - # Add radius-based edges - edge_index = e3tools.radius_graph(batch.pos, cutoff, batch.batch) - batch.edge_index = edge_index - - # Add bonded edges if they exist - if hasattr(batch, 'bonded_edge_index') and batch.bonded_edge_index is not None: - bond_mask = torch.cat([ - torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device), - torch.ones(batch.bonded_edge_index.shape[1], dtype=torch.long, device=batch.pos.device) - ]) - batch.edge_index = torch.cat([edge_index, batch.bonded_edge_index], dim=1) - batch.bond_mask = bond_mask - else: - batch.bond_mask = torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device) - - return batch - -def load_test_data(): - """Load ALA_ALA test dataset.""" - print("Loading ALA_ALA dataset...") - - dataset = parse_datasets_from_directory( - root="/data2/sules/ALA_ALA_enhanced_full_grid/train", - traj_pattern="^(.*).xtc", - pdb_pattern="^(.*).pdb", - subsample=1, - total_lag_time=5, - lag_subsample_rate=1, - max_datasets=2, # Keep small for testing - num_frames=5 # Small number of frames - ) - - print(f"Loaded dataset with {len(dataset)} samples") - - # Get a sample and create batch - graph = dataset[0].__getitem__(0) - batch = torch_geometric.data.Batch.from_data_list([graph]) - - # Add edges - batch = add_edges_to_batch(batch, cutoff=0.05) - - # Move to device - batch = batch.to(device) - - print(f"Batch info:") - print(f" - pos shape: {batch.pos.shape}") - print(f" - edge_index shape: {batch.edge_index.shape}") - print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") - if hasattr(batch, 'hidden_state') and batch.hidden_state: - print(f" - hidden_state[0] shape: {batch.hidden_state[0].shape}") - - return batch - -def test_spatiotemporal_conditioner(conditioner: SpatioTemporalConditioner, batch: torch_geometric.data.Batch): - """Test the spatiotemporal conditioner.""" - print("\n" + "="*50) - print("TESTING SPATIOTEMPORAL CONDITIONER") - print("="*50) - - try: - # Test forward pass - conditioned_structures = conditioner(batch) - - print(f"āœ… Conditioner forward pass successful!") - print(f"Number of conditioned structures: {len(conditioned_structures)} (expected: 2)") - print(f"First structure (y.pos) shape: {conditioned_structures[0].shape}") - print(f"Second structure (spatial_features) shape: {conditioned_structures[1].shape}") - print(f"Original position shape: {batch.pos.shape}") - print(f"Position difference norm: {torch.norm(conditioned_structures[0] - batch.pos):.6f}") - - # Verify we got exactly two structures - assert len(conditioned_structures) == 2, f"Expected 2 structures, got {len(conditioned_structures)}" - - return True, conditioned_structures - - except Exception as e: - print(f"āŒ Conditioner test failed: {e}") - import traceback - traceback.print_exc() - return False, None - -def test_conditional_denoiser_creation(): - """Test creating Denoiser with spatiotemporal conditioner.""" - print("\n" + "="*50) - print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER CREATION") - print("="*50) - - try: - # Create configuration - config = create_conditional_denoiser_config() - - # Create denoiser (this will instantiate all components) - denoiser = Denoiser(**config) - denoiser = denoiser.to(device) - - print(f"āœ… Denoiser created successfully!") - print(f"Denoiser device: {next(denoiser.parameters()).device}") - print(f"Has conditioner: {hasattr(denoiser, 'conditioning_module')}") - print(f"Architecture type: {type(denoiser.g).__name__}") - print(f"Conditioner type: {type(denoiser.conditioning_module).__name__}") - - # Check if spatiotemporal model is properly set up - if hasattr(denoiser.conditioning_module, 'spatiotemporal_model'): - st_model = denoiser.conditioning_module.spatiotemporal_model - print(f"SpatioTemporal model type: {type(st_model).__name__}") - print(f"Spatial module type: {type(st_model.spatial_module).__name__}") - print(f"Temporal module type: {type(st_model.temporal_module).__name__}") - - return True, denoiser - - except Exception as e: - print(f"āŒ Denoiser creation failed: {e}") - import traceback - traceback.print_exc() - return False, None - -def test_denoiser_forward_pass(denoiser: Denoiser, batch: torch_geometric.data.Batch): - """Test the complete denoiser forward pass.""" - print("\n" + "="*50) - print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER FORWARD PASS") - print("="*50) - - try: - # Test with sigma = 0.1 - sigma = 0.1 - - # Debug: check conditioned structures shapes - conditioned_structures = denoiser.conditioning_module(batch) - print(f"DEBUG: Conditioned structures shapes:") - for i, struct in enumerate(conditioned_structures): - print(f" Structure {i}: {struct.shape}") - - concatenated = torch.cat([*conditioned_structures], dim=-1) - print(f"DEBUG: Concatenated shape: {concatenated.shape}") - print(f"DEBUG: Expected irreps: 4x1e = 12 components") - - with torch.no_grad(): - xhat_batch = denoiser.xhat(batch, sigma) - - print(f"āœ… Denoiser forward pass successful!") - print(f"Input shape: {batch.pos.shape}") - print(f"Output shape: {xhat_batch.pos.shape}") - print(f"Output norm: {torch.norm(xhat_batch.pos):.6f}") - print(f"Used sigma: {sigma}") - - # Verify output shapes match input - assert xhat_batch.pos.shape == batch.pos.shape, f"Shape mismatch: {xhat_batch.pos.shape} vs {batch.pos.shape}" - - return True - - except Exception as e: - print(f"āŒ Denoiser forward pass failed: {e}") - import traceback - traceback.print_exc() - return False - -def main(): - """Main test function.""" - print("="*60) - print("CONDITIONAL DENOISER WITH SPATIOTEMPORAL CONDITIONER TEST") - print("="*60) - - # Load test data - batch = load_test_data() - - # Test conditioner creation and forward pass - conditioner = create_spatiotemporal_conditioner() - conditioner = conditioner.to(device) - - conditioner_success, conditioned_structures = test_spatiotemporal_conditioner(conditioner, batch) - - if not conditioner_success: - print("āŒ Conditioner test failed, stopping here.") - return - - # Test complete denoiser creation - denoiser_success, denoiser = test_conditional_denoiser_creation() - - if not denoiser_success: - print("āŒ Denoiser creation failed, stopping here.") - return - - # Test complete forward pass - forward_success = test_denoiser_forward_pass(denoiser, batch) - - # Final summary - print("\n" + "="*60) - print("FINAL SUMMARY") - print("="*60) - - print(f"Test Results:") - print(f" - Conditioner test: {'āœ… PASSED' if conditioner_success else 'āŒ FAILED'}") - print(f" - Denoiser creation: {'āœ… PASSED' if denoiser_success else 'āŒ FAILED'}") - print(f" - Forward pass test: {'āœ… PASSED' if forward_success else 'āŒ FAILED'}") - - if conditioner_success and denoiser_success and forward_success: - print("\nšŸŽ‰ ALL TESTS PASSED!") - print("The conditional denoiser with spatiotemporal conditioner is working correctly!") - else: - print("\nāš ļø Some tests failed. Check the output above for details.") - - # Device memory summary - if torch.cuda.is_available(): - print(f"\nCUDA Memory:") - print(f" - Allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") - print(f" - Cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") - -if __name__ == "__main__": - main() \ No newline at end of file diff --git a/uv.lock b/uv.lock index 2ee8e36..0cb5bfc 100644 --- a/uv.lock +++ b/uv.lock @@ -2,8 +2,10 @@ version = 1 revision = 2 requires-python = ">=3.10" resolution-markers = [ - "python_full_version >= '3.12' and sys_platform == 'linux'", - "python_full_version >= '3.12' and sys_platform != 'linux'", + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", "python_full_version == '3.11.*' and sys_platform == 'linux'", "python_full_version == '3.11.*' and sys_platform != 'linux'", "python_full_version < '3.11' and sys_platform == 'linux'", @@ -12,7 +14,7 @@ resolution-markers = [ [[package]] name = "aiobotocore" -version = "2.21.1" +version = "2.23.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "aiohttp" }, @@ -23,9 +25,9 @@ dependencies = [ { name = "python-dateutil" }, { name = "wrapt" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/d2/dc/f5f872fb01ce37c09525cedc7ecfad7002ffe2a8a23f77d7d2c234399b51/aiobotocore-2.21.1.tar.gz", hash = "sha256:010357f43004413e92a9d066bb0db1f241aeb29ffed306e9197061ffc94e6577", size = 108900, upload-time = "2025-03-04T18:30:58.945Z" } +sdist = { url = "https://files.pythonhosted.org/packages/1d/ae/523d48504902a6f17f6ec94311899f217f1bf64b9ca394c89c690c37434c/aiobotocore-2.23.2.tar.gz", hash = "sha256:9c2cbd6e813bb6c60b7f20fc11897976a583c57b0093a87bebfe80a9b08746b2", size = 115881, upload-time = "2025-07-24T17:48:15.957Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/95/67/026598918f92145156f2feb7957f57daefda20375cc2ac1a0692a9b8010b/aiobotocore-2.21.1-py3-none-any.whl", hash = "sha256:bd7c49a6d6f8a3d9444b0a94417c8da13813b5c7eec1c4f0ec2db7e8ce8f23e7", size = 78313, upload-time = "2025-03-04T18:30:56.493Z" }, + { url = "https://files.pythonhosted.org/packages/c9/2e/55780065672a69ab3d16062368d358ace7eac196b246e6b15b07301f8fbf/aiobotocore-2.23.2-py3-none-any.whl", hash = "sha256:5ca24feb49be73bd6cd92e82e95aefb0647c07bb85ca57000a0361b9554503d8", size = 84301, upload-time = "2025-07-24T17:48:14.494Z" }, ] [package.optional-dependencies] @@ -44,7 +46,7 @@ wheels = [ [[package]] name = "aiohttp" -version = "3.11.18" +version = "3.12.15" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "aiohappyeyeballs" }, @@ -56,72 +58,76 @@ dependencies = [ { name = "propcache" }, { name = "yarl" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/63/e7/fa1a8c00e2c54b05dc8cb5d1439f627f7c267874e3f7bb047146116020f9/aiohttp-3.11.18.tar.gz", hash = "sha256:ae856e1138612b7e412db63b7708735cff4d38d0399f6a5435d3dac2669f558a", size = 7678653, upload-time = "2025-04-21T09:43:09.191Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/c7/c3/e5f64af7e97a02f547020e6ff861595766bb5ecb37c7492fac9fe3c14f6c/aiohttp-3.11.18-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:96264854fedbea933a9ca4b7e0c745728f01380691687b7365d18d9e977179c4", size = 711703, upload-time = "2025-04-21T09:40:25.487Z" }, - { url = "https://files.pythonhosted.org/packages/5f/2f/53c26e96efa5fd01ebcfe1fefdfb7811f482bb21f4fa103d85eca4dcf888/aiohttp-3.11.18-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:9602044ff047043430452bc3a2089743fa85da829e6fc9ee0025351d66c332b6", size = 471348, upload-time = "2025-04-21T09:40:27.569Z" }, - { url = "https://files.pythonhosted.org/packages/80/47/dcc248464c9b101532ee7d254a46f6ed2c1fd3f4f0f794cf1f2358c0d45b/aiohttp-3.11.18-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:5691dc38750fcb96a33ceef89642f139aa315c8a193bbd42a0c33476fd4a1609", size = 457611, upload-time = "2025-04-21T09:40:28.978Z" }, - { url = "https://files.pythonhosted.org/packages/4c/ca/67d816ef075e8ac834b5f1f6b18e8db7d170f7aebaf76f1be462ea10cab0/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:554c918ec43f8480b47a5ca758e10e793bd7410b83701676a4782672d670da55", size = 1591976, upload-time = "2025-04-21T09:40:30.804Z" }, - { url = "https://files.pythonhosted.org/packages/46/00/0c120287aa51c744438d99e9aae9f8c55ca5b9911c42706966c91c9d68d6/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8a4076a2b3ba5b004b8cffca6afe18a3b2c5c9ef679b4d1e9859cf76295f8d4f", size = 1632819, upload-time = "2025-04-21T09:40:32.731Z" }, - { url = "https://files.pythonhosted.org/packages/54/a3/3923c9040cd4927dfee1aa017513701e35adcfc35d10729909688ecaa465/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:767a97e6900edd11c762be96d82d13a1d7c4fc4b329f054e88b57cdc21fded94", size = 1666567, upload-time = "2025-04-21T09:40:34.901Z" }, - { url = "https://files.pythonhosted.org/packages/e0/ab/40dacb15c0c58f7f17686ea67bc186e9f207341691bdb777d1d5ff4671d5/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f0ddc9337a0fb0e727785ad4f41163cc314376e82b31846d3835673786420ef1", size = 1594959, upload-time = "2025-04-21T09:40:36.714Z" }, - { url = "https://files.pythonhosted.org/packages/0d/98/d40c2b7c4a5483f9a16ef0adffce279ced3cc44522e84b6ba9e906be5168/aiohttp-3.11.18-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f414f37b244f2a97e79b98d48c5ff0789a0b4b4609b17d64fa81771ad780e415", size = 1538516, upload-time = "2025-04-21T09:40:38.263Z" }, - { url = "https://files.pythonhosted.org/packages/cf/10/e0bf3a03524faac45a710daa034e6f1878b24a1fef9c968ac8eb786ae657/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:fdb239f47328581e2ec7744ab5911f97afb10752332a6dd3d98e14e429e1a9e7", size = 1529037, upload-time = "2025-04-21T09:40:40.349Z" }, - { url = "https://files.pythonhosted.org/packages/ad/d6/5ff5282e00e4eb59c857844984cbc5628f933e2320792e19f93aff518f52/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:f2c50bad73ed629cc326cc0f75aed8ecfb013f88c5af116f33df556ed47143eb", size = 1546813, upload-time = "2025-04-21T09:40:42.106Z" }, - { url = "https://files.pythonhosted.org/packages/de/96/f1014f84101f9b9ad2d8acf3cc501426475f7f0cc62308ae5253e2fac9a7/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:0a8d8f20c39d3fa84d1c28cdb97f3111387e48209e224408e75f29c6f8e0861d", size = 1523852, upload-time = "2025-04-21T09:40:44.164Z" }, - { url = "https://files.pythonhosted.org/packages/a5/86/ec772c6838dd6bae3229065af671891496ac1834b252f305cee8152584b2/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:106032eaf9e62fd6bc6578c8b9e6dc4f5ed9a5c1c7fb2231010a1b4304393421", size = 1603766, upload-time = "2025-04-21T09:40:46.203Z" }, - { url = "https://files.pythonhosted.org/packages/84/38/31f85459c9402d409c1499284fc37a96f69afadce3cfac6a1b5ab048cbf1/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:b491e42183e8fcc9901d8dcd8ae644ff785590f1727f76ca86e731c61bfe6643", size = 1620647, upload-time = "2025-04-21T09:40:48.168Z" }, - { url = "https://files.pythonhosted.org/packages/31/2f/54aba0040764dd3d362fb37bd6aae9b3034fcae0b27f51b8a34864e48209/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:ad8c745ff9460a16b710e58e06a9dec11ebc0d8f4dd82091cefb579844d69868", size = 1559260, upload-time = "2025-04-21T09:40:50.219Z" }, - { url = "https://files.pythonhosted.org/packages/ca/d2/a05c7dd9e1b6948c1c5d04f1a8bcfd7e131923fa809bb87477d5c76f1517/aiohttp-3.11.18-cp310-cp310-win32.whl", hash = "sha256:8e57da93e24303a883146510a434f0faf2f1e7e659f3041abc4e3fb3f6702a9f", size = 418051, upload-time = "2025-04-21T09:40:52.272Z" }, - { url = "https://files.pythonhosted.org/packages/39/e2/796a6179e8abe267dfc84614a50291560a989d28acacbc5dab3bcd4cbec4/aiohttp-3.11.18-cp310-cp310-win_amd64.whl", hash = "sha256:cc93a4121d87d9f12739fc8fab0a95f78444e571ed63e40bfc78cd5abe700ac9", size = 442908, upload-time = "2025-04-21T09:40:54.345Z" }, - { url = "https://files.pythonhosted.org/packages/2f/10/fd9ee4f9e042818c3c2390054c08ccd34556a3cb209d83285616434cf93e/aiohttp-3.11.18-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:427fdc56ccb6901ff8088544bde47084845ea81591deb16f957897f0f0ba1be9", size = 712088, upload-time = "2025-04-21T09:40:55.776Z" }, - { url = "https://files.pythonhosted.org/packages/22/eb/6a77f055ca56f7aae2cd2a5607a3c9e7b9554f1497a069dcfcb52bfc9540/aiohttp-3.11.18-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2c828b6d23b984255b85b9b04a5b963a74278b7356a7de84fda5e3b76866597b", size = 471450, upload-time = "2025-04-21T09:40:57.301Z" }, - { url = "https://files.pythonhosted.org/packages/78/dc/5f3c0d27c91abf0bb5d103e9c9b0ff059f60cf6031a5f06f456c90731f42/aiohttp-3.11.18-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5c2eaa145bb36b33af1ff2860820ba0589e165be4ab63a49aebfd0981c173b66", size = 457836, upload-time = "2025-04-21T09:40:59.322Z" }, - { url = "https://files.pythonhosted.org/packages/49/7b/55b65af9ef48b9b811c91ff8b5b9de9650c71147f10523e278d297750bc8/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3d518ce32179f7e2096bf4e3e8438cf445f05fedd597f252de9f54c728574756", size = 1690978, upload-time = "2025-04-21T09:41:00.795Z" }, - { url = "https://files.pythonhosted.org/packages/a2/5a/3f8938c4f68ae400152b42742653477fc625d6bfe02e764f3521321c8442/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0700055a6e05c2f4711011a44364020d7a10fbbcd02fbf3e30e8f7e7fddc8717", size = 1745307, upload-time = "2025-04-21T09:41:02.89Z" }, - { url = "https://files.pythonhosted.org/packages/b4/42/89b694a293333ef6f771c62da022163bcf44fb03d4824372d88e3dc12530/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8bd1cde83e4684324e6ee19adfc25fd649d04078179890be7b29f76b501de8e4", size = 1780692, upload-time = "2025-04-21T09:41:04.461Z" }, - { url = "https://files.pythonhosted.org/packages/e2/ce/1a75384e01dd1bf546898b6062b1b5f7a59b6692ef802e4dd6db64fed264/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:73b8870fe1c9a201b8c0d12c94fe781b918664766728783241a79e0468427e4f", size = 1676934, upload-time = "2025-04-21T09:41:06.728Z" }, - { url = "https://files.pythonhosted.org/packages/a5/31/442483276e6c368ab5169797d9873b5875213cbcf7e74b95ad1c5003098a/aiohttp-3.11.18-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:25557982dd36b9e32c0a3357f30804e80790ec2c4d20ac6bcc598533e04c6361", size = 1621190, upload-time = "2025-04-21T09:41:08.293Z" }, - { url = "https://files.pythonhosted.org/packages/7b/83/90274bf12c079457966008a58831a99675265b6a34b505243e004b408934/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:7e889c9df381a2433802991288a61e5a19ceb4f61bd14f5c9fa165655dcb1fd1", size = 1658947, upload-time = "2025-04-21T09:41:11.054Z" }, - { url = "https://files.pythonhosted.org/packages/91/c1/da9cee47a0350b78fdc93670ebe7ad74103011d7778ab4c382ca4883098d/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:9ea345fda05bae217b6cce2acf3682ce3b13d0d16dd47d0de7080e5e21362421", size = 1654443, upload-time = "2025-04-21T09:41:13.213Z" }, - { url = "https://files.pythonhosted.org/packages/c9/f2/73cbe18dc25d624f79a09448adfc4972f82ed6088759ddcf783cd201956c/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:9f26545b9940c4b46f0a9388fd04ee3ad7064c4017b5a334dd450f616396590e", size = 1644169, upload-time = "2025-04-21T09:41:14.827Z" }, - { url = "https://files.pythonhosted.org/packages/5b/32/970b0a196c4dccb1b0cfa5b4dc3b20f63d76f1c608f41001a84b2fd23c3d/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:3a621d85e85dccabd700294494d7179ed1590b6d07a35709bb9bd608c7f5dd1d", size = 1728532, upload-time = "2025-04-21T09:41:17.168Z" }, - { url = "https://files.pythonhosted.org/packages/0b/50/b1dc810a41918d2ea9574e74125eb053063bc5e14aba2d98966f7d734da0/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:9c23fd8d08eb9c2af3faeedc8c56e134acdaf36e2117ee059d7defa655130e5f", size = 1750310, upload-time = "2025-04-21T09:41:19.353Z" }, - { url = "https://files.pythonhosted.org/packages/95/24/39271f5990b35ff32179cc95537e92499d3791ae82af7dcf562be785cd15/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d9e6b0e519067caa4fd7fb72e3e8002d16a68e84e62e7291092a5433763dc0dd", size = 1691580, upload-time = "2025-04-21T09:41:21.868Z" }, - { url = "https://files.pythonhosted.org/packages/6b/78/75d0353feb77f041460564f12fe58e456436bbc00cbbf5d676dbf0038cc2/aiohttp-3.11.18-cp311-cp311-win32.whl", hash = "sha256:122f3e739f6607e5e4c6a2f8562a6f476192a682a52bda8b4c6d4254e1138f4d", size = 417565, upload-time = "2025-04-21T09:41:24.78Z" }, - { url = "https://files.pythonhosted.org/packages/ed/97/b912dcb654634a813f8518de359364dfc45976f822116e725dc80a688eee/aiohttp-3.11.18-cp311-cp311-win_amd64.whl", hash = "sha256:e6f3c0a3a1e73e88af384b2e8a0b9f4fb73245afd47589df2afcab6b638fa0e6", size = 443652, upload-time = "2025-04-21T09:41:26.48Z" }, - { url = "https://files.pythonhosted.org/packages/b5/d2/5bc436f42bf4745c55f33e1e6a2d69e77075d3e768e3d1a34f96ee5298aa/aiohttp-3.11.18-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:63d71eceb9cad35d47d71f78edac41fcd01ff10cacaa64e473d1aec13fa02df2", size = 706671, upload-time = "2025-04-21T09:41:28.021Z" }, - { url = "https://files.pythonhosted.org/packages/fe/d0/2dbabecc4e078c0474abb40536bbde717fb2e39962f41c5fc7a216b18ea7/aiohttp-3.11.18-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:d1929da615840969929e8878d7951b31afe0bac883d84418f92e5755d7b49508", size = 466169, upload-time = "2025-04-21T09:41:29.783Z" }, - { url = "https://files.pythonhosted.org/packages/70/84/19edcf0b22933932faa6e0be0d933a27bd173da02dc125b7354dff4d8da4/aiohttp-3.11.18-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:7d0aebeb2392f19b184e3fdd9e651b0e39cd0f195cdb93328bd124a1d455cd0e", size = 457554, upload-time = "2025-04-21T09:41:31.327Z" }, - { url = "https://files.pythonhosted.org/packages/32/d0/e8d1f034ae5624a0f21e4fb3feff79342ce631f3a4d26bd3e58b31ef033b/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3849ead845e8444f7331c284132ab314b4dac43bfae1e3cf350906d4fff4620f", size = 1690154, upload-time = "2025-04-21T09:41:33.541Z" }, - { url = "https://files.pythonhosted.org/packages/16/de/2f9dbe2ac6f38f8495562077131888e0d2897e3798a0ff3adda766b04a34/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5e8452ad6b2863709f8b3d615955aa0807bc093c34b8e25b3b52097fe421cb7f", size = 1733402, upload-time = "2025-04-21T09:41:35.634Z" }, - { url = "https://files.pythonhosted.org/packages/e0/04/bd2870e1e9aef990d14b6df2a695f17807baf5c85a4c187a492bda569571/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3b8d2b42073611c860a37f718b3d61ae8b4c2b124b2e776e2c10619d920350ec", size = 1783958, upload-time = "2025-04-21T09:41:37.456Z" }, - { url = "https://files.pythonhosted.org/packages/23/06/4203ffa2beb5bedb07f0da0f79b7d9039d1c33f522e0d1a2d5b6218e6f2e/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:40fbf91f6a0ac317c0a07eb328a1384941872f6761f2e6f7208b63c4cc0a7ff6", size = 1695288, upload-time = "2025-04-21T09:41:39.756Z" }, - { url = "https://files.pythonhosted.org/packages/30/b2/e2285dda065d9f29ab4b23d8bcc81eb881db512afb38a3f5247b191be36c/aiohttp-3.11.18-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:44ff5625413fec55216da5eaa011cf6b0a2ed67a565914a212a51aa3755b0009", size = 1618871, upload-time = "2025-04-21T09:41:41.972Z" }, - { url = "https://files.pythonhosted.org/packages/57/e0/88f2987885d4b646de2036f7296ebea9268fdbf27476da551c1a7c158bc0/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7f33a92a2fde08e8c6b0c61815521324fc1612f397abf96eed86b8e31618fdb4", size = 1646262, upload-time = "2025-04-21T09:41:44.192Z" }, - { url = "https://files.pythonhosted.org/packages/e0/19/4d2da508b4c587e7472a032290b2981f7caeca82b4354e19ab3df2f51d56/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:11d5391946605f445ddafda5eab11caf310f90cdda1fd99865564e3164f5cff9", size = 1677431, upload-time = "2025-04-21T09:41:46.049Z" }, - { url = "https://files.pythonhosted.org/packages/eb/ae/047473ea50150a41440f3265f53db1738870b5a1e5406ece561ca61a3bf4/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:3cc314245deb311364884e44242e00c18b5896e4fe6d5f942e7ad7e4cb640adb", size = 1637430, upload-time = "2025-04-21T09:41:47.973Z" }, - { url = "https://files.pythonhosted.org/packages/11/32/c6d1e3748077ce7ee13745fae33e5cb1dac3e3b8f8787bf738a93c94a7d2/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:0f421843b0f70740772228b9e8093289924359d306530bcd3926f39acbe1adda", size = 1703342, upload-time = "2025-04-21T09:41:50.323Z" }, - { url = "https://files.pythonhosted.org/packages/c5/1d/a3b57bfdbe285f0d45572d6d8f534fd58761da3e9cbc3098372565005606/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:e220e7562467dc8d589e31c1acd13438d82c03d7f385c9cd41a3f6d1d15807c1", size = 1740600, upload-time = "2025-04-21T09:41:52.111Z" }, - { url = "https://files.pythonhosted.org/packages/a5/71/f9cd2fed33fa2b7ce4d412fb7876547abb821d5b5520787d159d0748321d/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ab2ef72f8605046115bc9aa8e9d14fd49086d405855f40b79ed9e5c1f9f4faea", size = 1695131, upload-time = "2025-04-21T09:41:53.94Z" }, - { url = "https://files.pythonhosted.org/packages/97/97/d1248cd6d02b9de6aa514793d0dcb20099f0ec47ae71a933290116c070c5/aiohttp-3.11.18-cp312-cp312-win32.whl", hash = "sha256:12a62691eb5aac58d65200c7ae94d73e8a65c331c3a86a2e9670927e94339ee8", size = 412442, upload-time = "2025-04-21T09:41:55.689Z" }, - { url = "https://files.pythonhosted.org/packages/33/9a/e34e65506e06427b111e19218a99abf627638a9703f4b8bcc3e3021277ed/aiohttp-3.11.18-cp312-cp312-win_amd64.whl", hash = "sha256:364329f319c499128fd5cd2d1c31c44f234c58f9b96cc57f743d16ec4f3238c8", size = 439444, upload-time = "2025-04-21T09:41:57.977Z" }, - { url = "https://files.pythonhosted.org/packages/0a/18/be8b5dd6b9cf1b2172301dbed28e8e5e878ee687c21947a6c81d6ceaa15d/aiohttp-3.11.18-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:474215ec618974054cf5dc465497ae9708543cbfc312c65212325d4212525811", size = 699833, upload-time = "2025-04-21T09:42:00.298Z" }, - { url = "https://files.pythonhosted.org/packages/0d/84/ecdc68e293110e6f6f6d7b57786a77555a85f70edd2b180fb1fafaff361a/aiohttp-3.11.18-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:6ced70adf03920d4e67c373fd692123e34d3ac81dfa1c27e45904a628567d804", size = 462774, upload-time = "2025-04-21T09:42:02.015Z" }, - { url = "https://files.pythonhosted.org/packages/d7/85/f07718cca55884dad83cc2433746384d267ee970e91f0dcc75c6d5544079/aiohttp-3.11.18-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:2d9f6c0152f8d71361905aaf9ed979259537981f47ad099c8b3d81e0319814bd", size = 454429, upload-time = "2025-04-21T09:42:03.728Z" }, - { url = "https://files.pythonhosted.org/packages/82/02/7f669c3d4d39810db8842c4e572ce4fe3b3a9b82945fdd64affea4c6947e/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a35197013ed929c0aed5c9096de1fc5a9d336914d73ab3f9df14741668c0616c", size = 1670283, upload-time = "2025-04-21T09:42:06.053Z" }, - { url = "https://files.pythonhosted.org/packages/ec/79/b82a12f67009b377b6c07a26bdd1b81dab7409fc2902d669dbfa79e5ac02/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:540b8a1f3a424f1af63e0af2d2853a759242a1769f9f1ab053996a392bd70118", size = 1717231, upload-time = "2025-04-21T09:42:07.953Z" }, - { url = "https://files.pythonhosted.org/packages/a6/38/d5a1f28c3904a840642b9a12c286ff41fc66dfa28b87e204b1f242dbd5e6/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f9e6710ebebfce2ba21cee6d91e7452d1125100f41b906fb5af3da8c78b764c1", size = 1769621, upload-time = "2025-04-21T09:42:09.855Z" }, - { url = "https://files.pythonhosted.org/packages/53/2d/deb3749ba293e716b5714dda06e257f123c5b8679072346b1eb28b766a0b/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f8af2ef3b4b652ff109f98087242e2ab974b2b2b496304063585e3d78de0b000", size = 1678667, upload-time = "2025-04-21T09:42:11.741Z" }, - { url = "https://files.pythonhosted.org/packages/b8/a8/04b6e11683a54e104b984bd19a9790eb1ae5f50968b601bb202d0406f0ff/aiohttp-3.11.18-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:28c3f975e5ae3dbcbe95b7e3dcd30e51da561a0a0f2cfbcdea30fc1308d72137", size = 1601592, upload-time = "2025-04-21T09:42:14.137Z" }, - { url = "https://files.pythonhosted.org/packages/5e/9d/c33305ae8370b789423623f0e073d09ac775cd9c831ac0f11338b81c16e0/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c28875e316c7b4c3e745172d882d8a5c835b11018e33432d281211af35794a93", size = 1621679, upload-time = "2025-04-21T09:42:16.056Z" }, - { url = "https://files.pythonhosted.org/packages/56/45/8e9a27fff0538173d47ba60362823358f7a5f1653c6c30c613469f94150e/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:13cd38515568ae230e1ef6919e2e33da5d0f46862943fcda74e7e915096815f3", size = 1656878, upload-time = "2025-04-21T09:42:18.368Z" }, - { url = "https://files.pythonhosted.org/packages/84/5b/8c5378f10d7a5a46b10cb9161a3aac3eeae6dba54ec0f627fc4ddc4f2e72/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:0e2a92101efb9f4c2942252c69c63ddb26d20f46f540c239ccfa5af865197bb8", size = 1620509, upload-time = "2025-04-21T09:42:20.141Z" }, - { url = "https://files.pythonhosted.org/packages/9e/2f/99dee7bd91c62c5ff0aa3c55f4ae7e1bc99c6affef780d7777c60c5b3735/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:e6d3e32b8753c8d45ac550b11a1090dd66d110d4ef805ffe60fa61495360b3b2", size = 1680263, upload-time = "2025-04-21T09:42:21.993Z" }, - { url = "https://files.pythonhosted.org/packages/03/0a/378745e4ff88acb83e2d5c884a4fe993a6e9f04600a4560ce0e9b19936e3/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:ea4cf2488156e0f281f93cc2fd365025efcba3e2d217cbe3df2840f8c73db261", size = 1715014, upload-time = "2025-04-21T09:42:23.87Z" }, - { url = "https://files.pythonhosted.org/packages/f6/0b/b5524b3bb4b01e91bc4323aad0c2fcaebdf2f1b4d2eb22743948ba364958/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9d4df95ad522c53f2b9ebc07f12ccd2cb15550941e11a5bbc5ddca2ca56316d7", size = 1666614, upload-time = "2025-04-21T09:42:25.764Z" }, - { url = "https://files.pythonhosted.org/packages/c7/b7/3d7b036d5a4ed5a4c704e0754afe2eef24a824dfab08e6efbffb0f6dd36a/aiohttp-3.11.18-cp313-cp313-win32.whl", hash = "sha256:cdd1bbaf1e61f0d94aced116d6e95fe25942f7a5f42382195fd9501089db5d78", size = 411358, upload-time = "2025-04-21T09:42:27.558Z" }, - { url = "https://files.pythonhosted.org/packages/1e/3c/143831b32cd23b5263a995b2a1794e10aa42f8a895aae5074c20fda36c07/aiohttp-3.11.18-cp313-cp313-win_amd64.whl", hash = "sha256:bdd619c27e44382cf642223f11cfd4d795161362a5a1fc1fa3940397bc89db01", size = 437658, upload-time = "2025-04-21T09:42:29.209Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/9b/e7/d92a237d8802ca88483906c388f7c201bbe96cd80a165ffd0ac2f6a8d59f/aiohttp-3.12.15.tar.gz", hash = "sha256:4fc61385e9c98d72fcdf47e6dd81833f47b2f77c114c29cd64a361be57a763a2", size = 7823716, upload-time = "2025-07-29T05:52:32.215Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/47/dc/ef9394bde9080128ad401ac7ede185267ed637df03b51f05d14d1c99ad67/aiohttp-3.12.15-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:b6fc902bff74d9b1879ad55f5404153e2b33a82e72a95c89cec5eb6cc9e92fbc", size = 703921, upload-time = "2025-07-29T05:49:43.584Z" }, + { url = "https://files.pythonhosted.org/packages/8f/42/63fccfc3a7ed97eb6e1a71722396f409c46b60a0552d8a56d7aad74e0df5/aiohttp-3.12.15-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:098e92835b8119b54c693f2f88a1dec690e20798ca5f5fe5f0520245253ee0af", size = 480288, upload-time = "2025-07-29T05:49:47.851Z" }, + { url = "https://files.pythonhosted.org/packages/9c/a2/7b8a020549f66ea2a68129db6960a762d2393248f1994499f8ba9728bbed/aiohttp-3.12.15-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:40b3fee496a47c3b4a39a731954c06f0bd9bd3e8258c059a4beb76ac23f8e421", size = 468063, upload-time = "2025-07-29T05:49:49.789Z" }, + { url = "https://files.pythonhosted.org/packages/8f/f5/d11e088da9176e2ad8220338ae0000ed5429a15f3c9dfd983f39105399cd/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2ce13fcfb0bb2f259fb42106cdc63fa5515fb85b7e87177267d89a771a660b79", size = 1650122, upload-time = "2025-07-29T05:49:51.874Z" }, + { url = "https://files.pythonhosted.org/packages/b0/6b/b60ce2757e2faed3d70ed45dafee48cee7bfb878785a9423f7e883f0639c/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:3beb14f053222b391bf9cf92ae82e0171067cc9c8f52453a0f1ec7c37df12a77", size = 1624176, upload-time = "2025-07-29T05:49:53.805Z" }, + { url = "https://files.pythonhosted.org/packages/dd/de/8c9fde2072a1b72c4fadecf4f7d4be7a85b1d9a4ab333d8245694057b4c6/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4c39e87afe48aa3e814cac5f535bc6199180a53e38d3f51c5e2530f5aa4ec58c", size = 1696583, upload-time = "2025-07-29T05:49:55.338Z" }, + { url = "https://files.pythonhosted.org/packages/0c/ad/07f863ca3d895a1ad958a54006c6dafb4f9310f8c2fdb5f961b8529029d3/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d5f1b4ce5bc528a6ee38dbf5f39bbf11dd127048726323b72b8e85769319ffc4", size = 1738896, upload-time = "2025-07-29T05:49:57.045Z" }, + { url = "https://files.pythonhosted.org/packages/20/43/2bd482ebe2b126533e8755a49b128ec4e58f1a3af56879a3abdb7b42c54f/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1004e67962efabbaf3f03b11b4c43b834081c9e3f9b32b16a7d97d4708a9abe6", size = 1643561, upload-time = "2025-07-29T05:49:58.762Z" }, + { url = "https://files.pythonhosted.org/packages/23/40/2fa9f514c4cf4cbae8d7911927f81a1901838baf5e09a8b2c299de1acfe5/aiohttp-3.12.15-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8faa08fcc2e411f7ab91d1541d9d597d3a90e9004180edb2072238c085eac8c2", size = 1583685, upload-time = "2025-07-29T05:50:00.375Z" }, + { url = "https://files.pythonhosted.org/packages/b8/c3/94dc7357bc421f4fb978ca72a201a6c604ee90148f1181790c129396ceeb/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:fe086edf38b2222328cdf89af0dde2439ee173b8ad7cb659b4e4c6f385b2be3d", size = 1627533, upload-time = "2025-07-29T05:50:02.306Z" }, + { url = "https://files.pythonhosted.org/packages/bf/3f/1f8911fe1844a07001e26593b5c255a685318943864b27b4e0267e840f95/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:79b26fe467219add81d5e47b4a4ba0f2394e8b7c7c3198ed36609f9ba161aecb", size = 1638319, upload-time = "2025-07-29T05:50:04.282Z" }, + { url = "https://files.pythonhosted.org/packages/4e/46/27bf57a99168c4e145ffee6b63d0458b9c66e58bb70687c23ad3d2f0bd17/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:b761bac1192ef24e16706d761aefcb581438b34b13a2f069a6d343ec8fb693a5", size = 1613776, upload-time = "2025-07-29T05:50:05.863Z" }, + { url = "https://files.pythonhosted.org/packages/0f/7e/1d2d9061a574584bb4ad3dbdba0da90a27fdc795bc227def3a46186a8bc1/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:e153e8adacfe2af562861b72f8bc47f8a5c08e010ac94eebbe33dc21d677cd5b", size = 1693359, upload-time = "2025-07-29T05:50:07.563Z" }, + { url = "https://files.pythonhosted.org/packages/08/98/bee429b52233c4a391980a5b3b196b060872a13eadd41c3a34be9b1469ed/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:fc49c4de44977aa8601a00edbf157e9a421f227aa7eb477d9e3df48343311065", size = 1716598, upload-time = "2025-07-29T05:50:09.33Z" }, + { url = "https://files.pythonhosted.org/packages/57/39/b0314c1ea774df3392751b686104a3938c63ece2b7ce0ba1ed7c0b4a934f/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:2776c7ec89c54a47029940177e75c8c07c29c66f73464784971d6a81904ce9d1", size = 1644940, upload-time = "2025-07-29T05:50:11.334Z" }, + { url = "https://files.pythonhosted.org/packages/1b/83/3dacb8d3f8f512c8ca43e3fa8a68b20583bd25636ffa4e56ee841ffd79ae/aiohttp-3.12.15-cp310-cp310-win32.whl", hash = "sha256:2c7d81a277fa78b2203ab626ced1487420e8c11a8e373707ab72d189fcdad20a", size = 429239, upload-time = "2025-07-29T05:50:12.803Z" }, + { url = "https://files.pythonhosted.org/packages/eb/f9/470b5daba04d558c9673ca2034f28d067f3202a40e17804425f0c331c89f/aiohttp-3.12.15-cp310-cp310-win_amd64.whl", hash = "sha256:83603f881e11f0f710f8e2327817c82e79431ec976448839f3cd05d7afe8f830", size = 452297, upload-time = "2025-07-29T05:50:14.266Z" }, + { url = "https://files.pythonhosted.org/packages/20/19/9e86722ec8e835959bd97ce8c1efa78cf361fa4531fca372551abcc9cdd6/aiohttp-3.12.15-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:d3ce17ce0220383a0f9ea07175eeaa6aa13ae5a41f30bc61d84df17f0e9b1117", size = 711246, upload-time = "2025-07-29T05:50:15.937Z" }, + { url = "https://files.pythonhosted.org/packages/71/f9/0a31fcb1a7d4629ac9d8f01f1cb9242e2f9943f47f5d03215af91c3c1a26/aiohttp-3.12.15-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:010cc9bbd06db80fe234d9003f67e97a10fe003bfbedb40da7d71c1008eda0fe", size = 483515, upload-time = "2025-07-29T05:50:17.442Z" }, + { url = "https://files.pythonhosted.org/packages/62/6c/94846f576f1d11df0c2e41d3001000527c0fdf63fce7e69b3927a731325d/aiohttp-3.12.15-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:3f9d7c55b41ed687b9d7165b17672340187f87a773c98236c987f08c858145a9", size = 471776, upload-time = "2025-07-29T05:50:19.568Z" }, + { url = "https://files.pythonhosted.org/packages/f8/6c/f766d0aaafcee0447fad0328da780d344489c042e25cd58fde566bf40aed/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bc4fbc61bb3548d3b482f9ac7ddd0f18c67e4225aaa4e8552b9f1ac7e6bda9e5", size = 1741977, upload-time = "2025-07-29T05:50:21.665Z" }, + { url = "https://files.pythonhosted.org/packages/17/e5/fb779a05ba6ff44d7bc1e9d24c644e876bfff5abe5454f7b854cace1b9cc/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:7fbc8a7c410bb3ad5d595bb7118147dfbb6449d862cc1125cf8867cb337e8728", size = 1690645, upload-time = "2025-07-29T05:50:23.333Z" }, + { url = "https://files.pythonhosted.org/packages/37/4e/a22e799c2035f5d6a4ad2cf8e7c1d1bd0923192871dd6e367dafb158b14c/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:74dad41b3458dbb0511e760fb355bb0b6689e0630de8a22b1b62a98777136e16", size = 1789437, upload-time = "2025-07-29T05:50:25.007Z" }, + { url = "https://files.pythonhosted.org/packages/28/e5/55a33b991f6433569babb56018b2fb8fb9146424f8b3a0c8ecca80556762/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3b6f0af863cf17e6222b1735a756d664159e58855da99cfe965134a3ff63b0b0", size = 1828482, upload-time = "2025-07-29T05:50:26.693Z" }, + { url = "https://files.pythonhosted.org/packages/c6/82/1ddf0ea4f2f3afe79dffed5e8a246737cff6cbe781887a6a170299e33204/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b5b7fe4972d48a4da367043b8e023fb70a04d1490aa7d68800e465d1b97e493b", size = 1730944, upload-time = "2025-07-29T05:50:28.382Z" }, + { url = "https://files.pythonhosted.org/packages/1b/96/784c785674117b4cb3877522a177ba1b5e4db9ce0fd519430b5de76eec90/aiohttp-3.12.15-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6443cca89553b7a5485331bc9bedb2342b08d073fa10b8c7d1c60579c4a7b9bd", size = 1668020, upload-time = "2025-07-29T05:50:30.032Z" }, + { url = "https://files.pythonhosted.org/packages/12/8a/8b75f203ea7e5c21c0920d84dd24a5c0e971fe1e9b9ebbf29ae7e8e39790/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:6c5f40ec615e5264f44b4282ee27628cea221fcad52f27405b80abb346d9f3f8", size = 1716292, upload-time = "2025-07-29T05:50:31.983Z" }, + { url = "https://files.pythonhosted.org/packages/47/0b/a1451543475bb6b86a5cfc27861e52b14085ae232896a2654ff1231c0992/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:2abbb216a1d3a2fe86dbd2edce20cdc5e9ad0be6378455b05ec7f77361b3ab50", size = 1711451, upload-time = "2025-07-29T05:50:33.989Z" }, + { url = "https://files.pythonhosted.org/packages/55/fd/793a23a197cc2f0d29188805cfc93aa613407f07e5f9da5cd1366afd9d7c/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:db71ce547012a5420a39c1b744d485cfb823564d01d5d20805977f5ea1345676", size = 1691634, upload-time = "2025-07-29T05:50:35.846Z" }, + { url = "https://files.pythonhosted.org/packages/ca/bf/23a335a6670b5f5dfc6d268328e55a22651b440fca341a64fccf1eada0c6/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:ced339d7c9b5030abad5854aa5413a77565e5b6e6248ff927d3e174baf3badf7", size = 1785238, upload-time = "2025-07-29T05:50:37.597Z" }, + { url = "https://files.pythonhosted.org/packages/57/4f/ed60a591839a9d85d40694aba5cef86dde9ee51ce6cca0bb30d6eb1581e7/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:7c7dd29c7b5bda137464dc9bfc738d7ceea46ff70309859ffde8c022e9b08ba7", size = 1805701, upload-time = "2025-07-29T05:50:39.591Z" }, + { url = "https://files.pythonhosted.org/packages/85/e0/444747a9455c5de188c0f4a0173ee701e2e325d4b2550e9af84abb20cdba/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:421da6fd326460517873274875c6c5a18ff225b40da2616083c5a34a7570b685", size = 1718758, upload-time = "2025-07-29T05:50:41.292Z" }, + { url = "https://files.pythonhosted.org/packages/36/ab/1006278d1ffd13a698e5dd4bfa01e5878f6bddefc296c8b62649753ff249/aiohttp-3.12.15-cp311-cp311-win32.whl", hash = "sha256:4420cf9d179ec8dfe4be10e7d0fe47d6d606485512ea2265b0d8c5113372771b", size = 428868, upload-time = "2025-07-29T05:50:43.063Z" }, + { url = "https://files.pythonhosted.org/packages/10/97/ad2b18700708452400278039272032170246a1bf8ec5d832772372c71f1a/aiohttp-3.12.15-cp311-cp311-win_amd64.whl", hash = "sha256:edd533a07da85baa4b423ee8839e3e91681c7bfa19b04260a469ee94b778bf6d", size = 453273, upload-time = "2025-07-29T05:50:44.613Z" }, + { url = "https://files.pythonhosted.org/packages/63/97/77cb2450d9b35f517d6cf506256bf4f5bda3f93a66b4ad64ba7fc917899c/aiohttp-3.12.15-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:802d3868f5776e28f7bf69d349c26fc0efadb81676d0afa88ed00d98a26340b7", size = 702333, upload-time = "2025-07-29T05:50:46.507Z" }, + { url = "https://files.pythonhosted.org/packages/83/6d/0544e6b08b748682c30b9f65640d006e51f90763b41d7c546693bc22900d/aiohttp-3.12.15-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:f2800614cd560287be05e33a679638e586a2d7401f4ddf99e304d98878c29444", size = 476948, upload-time = "2025-07-29T05:50:48.067Z" }, + { url = "https://files.pythonhosted.org/packages/3a/1d/c8c40e611e5094330284b1aea8a4b02ca0858f8458614fa35754cab42b9c/aiohttp-3.12.15-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:8466151554b593909d30a0a125d638b4e5f3836e5aecde85b66b80ded1cb5b0d", size = 469787, upload-time = "2025-07-29T05:50:49.669Z" }, + { url = "https://files.pythonhosted.org/packages/38/7d/b76438e70319796bfff717f325d97ce2e9310f752a267bfdf5192ac6082b/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2e5a495cb1be69dae4b08f35a6c4579c539e9b5706f606632102c0f855bcba7c", size = 1716590, upload-time = "2025-07-29T05:50:51.368Z" }, + { url = "https://files.pythonhosted.org/packages/79/b1/60370d70cdf8b269ee1444b390cbd72ce514f0d1cd1a715821c784d272c9/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:6404dfc8cdde35c69aaa489bb3542fb86ef215fc70277c892be8af540e5e21c0", size = 1699241, upload-time = "2025-07-29T05:50:53.628Z" }, + { url = "https://files.pythonhosted.org/packages/a3/2b/4968a7b8792437ebc12186db31523f541943e99bda8f30335c482bea6879/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3ead1c00f8521a5c9070fcb88f02967b1d8a0544e6d85c253f6968b785e1a2ab", size = 1754335, upload-time = "2025-07-29T05:50:55.394Z" }, + { url = "https://files.pythonhosted.org/packages/fb/c1/49524ed553f9a0bec1a11fac09e790f49ff669bcd14164f9fab608831c4d/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6990ef617f14450bc6b34941dba4f12d5613cbf4e33805932f853fbd1cf18bfb", size = 1800491, upload-time = "2025-07-29T05:50:57.202Z" }, + { url = "https://files.pythonhosted.org/packages/de/5e/3bf5acea47a96a28c121b167f5ef659cf71208b19e52a88cdfa5c37f1fcc/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd736ed420f4db2b8148b52b46b88ed038d0354255f9a73196b7bbce3ea97545", size = 1719929, upload-time = "2025-07-29T05:50:59.192Z" }, + { url = "https://files.pythonhosted.org/packages/39/94/8ae30b806835bcd1cba799ba35347dee6961a11bd507db634516210e91d8/aiohttp-3.12.15-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3c5092ce14361a73086b90c6efb3948ffa5be2f5b6fbcf52e8d8c8b8848bb97c", size = 1635733, upload-time = "2025-07-29T05:51:01.394Z" }, + { url = "https://files.pythonhosted.org/packages/7a/46/06cdef71dd03acd9da7f51ab3a9107318aee12ad38d273f654e4f981583a/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:aaa2234bb60c4dbf82893e934d8ee8dea30446f0647e024074237a56a08c01bd", size = 1696790, upload-time = "2025-07-29T05:51:03.657Z" }, + { url = "https://files.pythonhosted.org/packages/02/90/6b4cfaaf92ed98d0ec4d173e78b99b4b1a7551250be8937d9d67ecb356b4/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:6d86a2fbdd14192e2f234a92d3b494dd4457e683ba07e5905a0b3ee25389ac9f", size = 1718245, upload-time = "2025-07-29T05:51:05.911Z" }, + { url = "https://files.pythonhosted.org/packages/2e/e6/2593751670fa06f080a846f37f112cbe6f873ba510d070136a6ed46117c6/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:a041e7e2612041a6ddf1c6a33b883be6a421247c7afd47e885969ee4cc58bd8d", size = 1658899, upload-time = "2025-07-29T05:51:07.753Z" }, + { url = "https://files.pythonhosted.org/packages/8f/28/c15bacbdb8b8eb5bf39b10680d129ea7410b859e379b03190f02fa104ffd/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5015082477abeafad7203757ae44299a610e89ee82a1503e3d4184e6bafdd519", size = 1738459, upload-time = "2025-07-29T05:51:09.56Z" }, + { url = "https://files.pythonhosted.org/packages/00/de/c269cbc4faa01fb10f143b1670633a8ddd5b2e1ffd0548f7aa49cb5c70e2/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:56822ff5ddfd1b745534e658faba944012346184fbfe732e0d6134b744516eea", size = 1766434, upload-time = "2025-07-29T05:51:11.423Z" }, + { url = "https://files.pythonhosted.org/packages/52/b0/4ff3abd81aa7d929b27d2e1403722a65fc87b763e3a97b3a2a494bfc63bc/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:b2acbbfff69019d9014508c4ba0401822e8bae5a5fdc3b6814285b71231b60f3", size = 1726045, upload-time = "2025-07-29T05:51:13.689Z" }, + { url = "https://files.pythonhosted.org/packages/71/16/949225a6a2dd6efcbd855fbd90cf476052e648fb011aa538e3b15b89a57a/aiohttp-3.12.15-cp312-cp312-win32.whl", hash = "sha256:d849b0901b50f2185874b9a232f38e26b9b3d4810095a7572eacea939132d4e1", size = 423591, upload-time = "2025-07-29T05:51:15.452Z" }, + { url = "https://files.pythonhosted.org/packages/2b/d8/fa65d2a349fe938b76d309db1a56a75c4fb8cc7b17a398b698488a939903/aiohttp-3.12.15-cp312-cp312-win_amd64.whl", hash = "sha256:b390ef5f62bb508a9d67cb3bba9b8356e23b3996da7062f1a57ce1a79d2b3d34", size = 450266, upload-time = "2025-07-29T05:51:17.239Z" }, + { url = "https://files.pythonhosted.org/packages/f2/33/918091abcf102e39d15aba2476ad9e7bd35ddb190dcdd43a854000d3da0d/aiohttp-3.12.15-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:9f922ffd05034d439dde1c77a20461cf4a1b0831e6caa26151fe7aa8aaebc315", size = 696741, upload-time = "2025-07-29T05:51:19.021Z" }, + { url = "https://files.pythonhosted.org/packages/b5/2a/7495a81e39a998e400f3ecdd44a62107254803d1681d9189be5c2e4530cd/aiohttp-3.12.15-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:2ee8a8ac39ce45f3e55663891d4b1d15598c157b4d494a4613e704c8b43112cd", size = 474407, upload-time = "2025-07-29T05:51:21.165Z" }, + { url = "https://files.pythonhosted.org/packages/49/fc/a9576ab4be2dcbd0f73ee8675d16c707cfc12d5ee80ccf4015ba543480c9/aiohttp-3.12.15-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3eae49032c29d356b94eee45a3f39fdf4b0814b397638c2f718e96cfadf4c4e4", size = 466703, upload-time = "2025-07-29T05:51:22.948Z" }, + { url = "https://files.pythonhosted.org/packages/09/2f/d4bcc8448cf536b2b54eed48f19682031ad182faa3a3fee54ebe5b156387/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b97752ff12cc12f46a9b20327104448042fce5c33a624f88c18f66f9368091c7", size = 1705532, upload-time = "2025-07-29T05:51:25.211Z" }, + { url = "https://files.pythonhosted.org/packages/f1/f3/59406396083f8b489261e3c011aa8aee9df360a96ac8fa5c2e7e1b8f0466/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:894261472691d6fe76ebb7fcf2e5870a2ac284c7406ddc95823c8598a1390f0d", size = 1686794, upload-time = "2025-07-29T05:51:27.145Z" }, + { url = "https://files.pythonhosted.org/packages/dc/71/164d194993a8d114ee5656c3b7ae9c12ceee7040d076bf7b32fb98a8c5c6/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5fa5d9eb82ce98959fc1031c28198b431b4d9396894f385cb63f1e2f3f20ca6b", size = 1738865, upload-time = "2025-07-29T05:51:29.366Z" }, + { url = "https://files.pythonhosted.org/packages/1c/00/d198461b699188a93ead39cb458554d9f0f69879b95078dce416d3209b54/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f0fa751efb11a541f57db59c1dd821bec09031e01452b2b6217319b3a1f34f3d", size = 1788238, upload-time = "2025-07-29T05:51:31.285Z" }, + { url = "https://files.pythonhosted.org/packages/85/b8/9e7175e1fa0ac8e56baa83bf3c214823ce250d0028955dfb23f43d5e61fd/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5346b93e62ab51ee2a9d68e8f73c7cf96ffb73568a23e683f931e52450e4148d", size = 1710566, upload-time = "2025-07-29T05:51:33.219Z" }, + { url = "https://files.pythonhosted.org/packages/59/e4/16a8eac9df39b48ae102ec030fa9f726d3570732e46ba0c592aeeb507b93/aiohttp-3.12.15-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:049ec0360f939cd164ecbfd2873eaa432613d5e77d6b04535e3d1fbae5a9e645", size = 1624270, upload-time = "2025-07-29T05:51:35.195Z" }, + { url = "https://files.pythonhosted.org/packages/1f/f8/cd84dee7b6ace0740908fd0af170f9fab50c2a41ccbc3806aabcb1050141/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:b52dcf013b57464b6d1e51b627adfd69a8053e84b7103a7cd49c030f9ca44461", size = 1677294, upload-time = "2025-07-29T05:51:37.215Z" }, + { url = "https://files.pythonhosted.org/packages/ce/42/d0f1f85e50d401eccd12bf85c46ba84f947a84839c8a1c2c5f6e8ab1eb50/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:9b2af240143dd2765e0fb661fd0361a1b469cab235039ea57663cda087250ea9", size = 1708958, upload-time = "2025-07-29T05:51:39.328Z" }, + { url = "https://files.pythonhosted.org/packages/d5/6b/f6fa6c5790fb602538483aa5a1b86fcbad66244997e5230d88f9412ef24c/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:ac77f709a2cde2cc71257ab2d8c74dd157c67a0558a0d2799d5d571b4c63d44d", size = 1651553, upload-time = "2025-07-29T05:51:41.356Z" }, + { url = "https://files.pythonhosted.org/packages/04/36/a6d36ad545fa12e61d11d1932eef273928b0495e6a576eb2af04297fdd3c/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:47f6b962246f0a774fbd3b6b7be25d59b06fdb2f164cf2513097998fc6a29693", size = 1727688, upload-time = "2025-07-29T05:51:43.452Z" }, + { url = "https://files.pythonhosted.org/packages/aa/c8/f195e5e06608a97a4e52c5d41c7927301bf757a8e8bb5bbf8cef6c314961/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:760fb7db442f284996e39cf9915a94492e1896baac44f06ae551974907922b64", size = 1761157, upload-time = "2025-07-29T05:51:45.643Z" }, + { url = "https://files.pythonhosted.org/packages/05/6a/ea199e61b67f25ba688d3ce93f63b49b0a4e3b3d380f03971b4646412fc6/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ad702e57dc385cae679c39d318def49aef754455f237499d5b99bea4ef582e51", size = 1710050, upload-time = "2025-07-29T05:51:48.203Z" }, + { url = "https://files.pythonhosted.org/packages/b4/2e/ffeb7f6256b33635c29dbed29a22a723ff2dd7401fff42ea60cf2060abfb/aiohttp-3.12.15-cp313-cp313-win32.whl", hash = "sha256:f813c3e9032331024de2eb2e32a88d86afb69291fbc37a3a3ae81cc9917fb3d0", size = 422647, upload-time = "2025-07-29T05:51:50.718Z" }, + { url = "https://files.pythonhosted.org/packages/1b/8e/78ee35774201f38d5e1ba079c9958f7629b1fd079459aea9467441dbfbf5/aiohttp-3.12.15-cp313-cp313-win_amd64.whl", hash = "sha256:1a649001580bdb37c6fdb1bebbd7e3bc688e8ec2b5c6f52edbb664662b17dc84", size = 449067, upload-time = "2025-07-29T05:51:52.549Z" }, ] [[package]] @@ -135,14 +141,15 @@ wheels = [ [[package]] name = "aiosignal" -version = "1.3.2" +version = "1.4.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "frozenlist" }, + { name = "typing-extensions", marker = "python_full_version < '3.13'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/ba/b5/6d55e80f6d8a08ce22b982eafa278d823b541c925f11ee774b0b9c43473d/aiosignal-1.3.2.tar.gz", hash = "sha256:a8c255c66fafb1e499c9351d0bf32ff2d8a0321595ebac3b93713656d2436f54", size = 19424, upload-time = "2024-12-13T17:10:40.86Z" } +sdist = { url = "https://files.pythonhosted.org/packages/61/62/06741b579156360248d1ec624842ad0edf697050bbaf7c3e46394e106ad1/aiosignal-1.4.0.tar.gz", hash = "sha256:f47eecd9468083c2029cc99945502cb7708b082c232f9aca65da147157b251c7", size = 25007, upload-time = "2025-07-03T22:54:43.528Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/ec/6a/bc7e17a3e87a2985d3e8f4da4cd0f481060eb78fb08596c42be62c90a4d9/aiosignal-1.3.2-py2.py3-none-any.whl", hash = "sha256:45cde58e409a301715980c2b01d0c28bdde3770d8290b5eb2173759d9acb31a5", size = 7597, upload-time = "2024-12-13T17:10:38.469Z" }, + { url = "https://files.pythonhosted.org/packages/fb/76/641ae371508676492379f16e2fa48f4e2c11741bd63c48be4b12a6b09cba/aiosignal-1.4.0-py3-none-any.whl", hash = "sha256:053243f8b92b990551949e63930a839ff0cf0b0ebbe0597b0f3fb19e1a0fe82e", size = 7490, upload-time = "2025-07-03T22:54:42.156Z" }, ] [[package]] @@ -171,16 +178,18 @@ wheels = [ [[package]] name = "ase" -version = "3.24.0" +version = "3.25.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "matplotlib" }, - { name = "numpy" }, - { name = "scipy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/5c/c9/9adb9bc641bd7222367886e4e6c753b4c64da4ff2d9565ab39aee1e34734/ase-3.24.0.tar.gz", hash = "sha256:9acc93d6daaf48cd27b844c56f8bf49428b9db0542faa3cc30d9d5b8e1842195", size = 2383264, upload-time = "2024-12-28T22:20:33.711Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e8/a1/5735ced2f159979f5b27c4083126b7796a5750cee6f027864e59818a5b76/ase-3.25.0.tar.gz", hash = "sha256:374cf8ca9fe588f05d6e856da3c9c17ef262dc968027b231d449334140c962c2", size = 2400055, upload-time = "2025-04-11T17:14:39.975Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/1f/cd/b1253035a1da90e89f31947e052c558cd83df3bcaff34aa199e5e806d773/ase-3.24.0-py3-none-any.whl", hash = "sha256:974922df87ef4ec8cf1140359a55ab4c4dc55c38e26876bdd9c00968da1f463c", size = 2928893, upload-time = "2024-12-28T22:20:29.416Z" }, + { url = "https://files.pythonhosted.org/packages/07/f5/007d993fcf3b051acb304d5402e0bd103fd20816b47dee9531bdbfb3aa0c/ase-3.25.0-py3-none-any.whl", hash = "sha256:f9a5295e1154da355af04726d001fa76a311c076616d98e49cd9f34fc3afe188", size = 2951559, upload-time = "2025-04-11T17:14:37.617Z" }, ] [[package]] @@ -212,39 +221,65 @@ wheels = [ [[package]] name = "boto3" -version = "1.37.1" +version = "1.39.8" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "botocore" }, { name = "jmespath" }, { name = "s3transfer" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/21/8c/c2af03daafaacea1db1823d23073facffa75818b61d376c3be77dd297ae8/boto3-1.37.1.tar.gz", hash = "sha256:96d18f7feb0c1fcb95f8837b74b6c8880e1b4e35ce5f8a8f8cb243a090c278ed", size = 111175, upload-time = "2025-02-25T20:33:16.471Z" } +sdist = { url = "https://files.pythonhosted.org/packages/d1/ef/f8dbe6482bdf9eb0230f2639483cdd40ef5aaa89c2fb651f2edeee9c248a/boto3-1.39.8.tar.gz", hash = "sha256:456ea6baef037eb6205d64e012259d14f0c9300c9b30603890746c1a0882fa01", size = 111829, upload-time = "2025-07-17T19:19:14.828Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/63/ec/e722c53c9dc41e8df094587c32e19409bace8b43b5eb31fe3536ca57a38b/boto3-1.37.1-py3-none-any.whl", hash = "sha256:4320441f904435a1b85e6ecb81793192e522c737cc9ed6566014e29f0a11cb22", size = 139338, upload-time = "2025-02-25T20:33:11.935Z" }, + { url = "https://files.pythonhosted.org/packages/3c/f0/f3701472b2e6192e62d80e703186ae9c789b3d607ba22943702c500897d2/boto3-1.39.8-py3-none-any.whl", hash = "sha256:dcea5270ccced0b4b962eb5874cb71b6232ccfc6203e05bf834a314442e4a79c", size = 139886, upload-time = "2025-07-17T19:19:12.634Z" }, ] [[package]] name = "botocore" -version = "1.37.1" +version = "1.39.8" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "jmespath" }, { name = "python-dateutil" }, { name = "urllib3" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e5/01/3083bff25fd91193162298920cb093b9095609408416526d52b2826965b7/botocore-1.37.1.tar.gz", hash = "sha256:b194db8fb2a0ffba53568c364ae26166e7eec0445496b2ac86a6e142f3dd982f", size = 13578835, upload-time = "2025-02-25T20:32:56.63Z" } +sdist = { url = "https://files.pythonhosted.org/packages/32/57/16d3d21963975b9be180e96695abfb146695ae7db57f9a2d47e92d33ce9d/botocore-1.39.8.tar.gz", hash = "sha256:3848bd9057ea8dbc059e7764eda63bda575727ad1101dbd03636ab4a6f283fa5", size = 14205898, upload-time = "2025-07-17T19:19:03.832Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/3d/20/352b2bf99f93ba18986615841786cbd0d38f7856bd49d4e154a540f04afe/botocore-1.37.1-py3-none-any.whl", hash = "sha256:c1db1bfc5d8c6b3b6d1ca6794f605294b4264e82a7e727b88e0fef9c2b9fbb9c", size = 13359164, upload-time = "2025-02-25T20:32:52.347Z" }, + { url = "https://files.pythonhosted.org/packages/11/ac/51462dd35fc60d11cdce93ba82ccf1635a161ceadc646d89f67d666fff31/botocore-1.39.8-py3-none-any.whl", hash = "sha256:ab43f79c6893271934faba7ae1987a313d59576361c544c70a5391ade560891d", size = 13866818, upload-time = "2025-07-17T19:18:58.521Z" }, +] + +[[package]] +name = "cached-path" +version = "1.7.3" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "boto3" }, + { name = "filelock" }, + { name = "google-cloud-storage" }, + { name = "huggingface-hub" }, + { name = "requests" }, + { name = "rich" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/14/35/ce8d9b5821f83df1a412379623d1d365a42c9b65c2e9c96fb7b24ce521ba/cached_path-1.7.3.tar.gz", hash = "sha256:956d21b5ac92d64ae6d76b2a1a043c5d660e3421d513e735157d56aca9a31d8e", size = 32795, upload-time = "2025-05-07T16:31:27.516Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/bc/2c/787a39669287d69b57bf150d01f13cf5daedea76ab69eae7e7677780acca/cached_path-1.7.3-py3-none-any.whl", hash = "sha256:fe2b396b4816205c95d6e961efb35c66288966c4f96b82cd87c4fb03d4d037d1", size = 36839, upload-time = "2025-05-07T16:31:25.483Z" }, +] + +[[package]] +name = "cachetools" +version = "5.5.2" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/6c/81/3747dad6b14fa2cf53fcf10548cf5aea6913e96fab41a3c198676f8948a5/cachetools-5.5.2.tar.gz", hash = "sha256:1a661caa9175d26759571b2e19580f9d6393969e5dfca11fdb1f947a23e640d4", size = 28380, upload-time = "2025-02-20T21:01:19.524Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/72/76/20fa66124dbe6be5cafeb312ece67de6b61dd91a0247d1ea13db4ebb33c2/cachetools-5.5.2-py3-none-any.whl", hash = "sha256:d26a22bcc62eb95c3beabd9f1ee5e820d3d2704fe2967cbe350e20c8ffcd3f0a", size = 10080, upload-time = "2025-02-20T21:01:16.647Z" }, ] [[package]] name = "certifi" -version = "2025.1.31" +version = "2025.8.3" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/1c/ab/c9f1e32b7b1bf505bf26f0ef697775960db7932abeb7b516de930ba2705f/certifi-2025.1.31.tar.gz", hash = "sha256:3d5da6925056f6f18f119200434a4780a94263f10d1c21d032a6f6b2baa20651", size = 167577, upload-time = "2025-01-31T02:16:47.166Z" } +sdist = { url = "https://files.pythonhosted.org/packages/dc/67/960ebe6bf230a96cda2e0abcf73af550ec4f090005363542f0765df162e0/certifi-2025.8.3.tar.gz", hash = "sha256:e564105f78ded564e3ae7c923924435e1daa7463faeab5bb932bc53ffae63407", size = 162386, upload-time = "2025-08-03T03:07:47.08Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/38/fc/bce832fd4fd99766c04d1ee0eead6b0ec6486fb100ae5e74c1d91292b982/certifi-2025.1.31-py3-none-any.whl", hash = "sha256:ca78db4565a652026a4db2bcdf68f2fb589ea80d0be70e03929ed730746b84fe", size = 166393, upload-time = "2025-01-31T02:16:45.015Z" }, + { url = "https://files.pythonhosted.org/packages/e5/48/1549795ba7742c948d2ad169c1c8cdbae65bc450d6cd753d124b17c8cd32/certifi-2025.8.3-py3-none-any.whl", hash = "sha256:f6c12493cfb1b06ba2ff328595af9350c65d6644968e5d3a2ffd78699af217a5", size = 161216, upload-time = "2025-08-03T03:07:45.777Z" }, ] [[package]] @@ -318,7 +353,7 @@ name = "cftime" version = "1.6.4.post1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/ab/c8/1155d1d58003105307c7e5985f422ae5bcb2ca0cbc553cc828f3c5a934a7/cftime-1.6.4.post1.tar.gz", hash = "sha256:50ac76cc9f10ab7bd46e44a71c51a6927051b499b4407df4f29ab13d741b942f", size = 54631, upload-time = "2024-10-22T18:48:34.194Z" } wheels = [ @@ -350,75 +385,75 @@ wheels = [ [[package]] name = "charset-normalizer" -version = "3.4.1" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/16/b0/572805e227f01586461c80e0fd25d65a2115599cc9dad142fee4b747c357/charset_normalizer-3.4.1.tar.gz", hash = "sha256:44251f18cd68a75b56585dd00dae26183e102cd5e0f9f1466e6df5da2ed64ea3", size = 123188, upload-time = "2024-12-24T18:12:35.43Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/0d/58/5580c1716040bc89206c77d8f74418caf82ce519aae06450393ca73475d1/charset_normalizer-3.4.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:91b36a978b5ae0ee86c394f5a54d6ef44db1de0815eb43de826d41d21e4af3de", size = 198013, upload-time = "2024-12-24T18:09:43.671Z" }, - { url = "https://files.pythonhosted.org/packages/d0/11/00341177ae71c6f5159a08168bcb98c6e6d196d372c94511f9f6c9afe0c6/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7461baadb4dc00fd9e0acbe254e3d7d2112e7f92ced2adc96e54ef6501c5f176", size = 141285, upload-time = "2024-12-24T18:09:48.113Z" }, - { url = "https://files.pythonhosted.org/packages/01/09/11d684ea5819e5a8f5100fb0b38cf8d02b514746607934134d31233e02c8/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e218488cd232553829be0664c2292d3af2eeeb94b32bea483cf79ac6a694e037", size = 151449, upload-time = "2024-12-24T18:09:50.845Z" }, - { url = "https://files.pythonhosted.org/packages/08/06/9f5a12939db324d905dc1f70591ae7d7898d030d7662f0d426e2286f68c9/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:80ed5e856eb7f30115aaf94e4a08114ccc8813e6ed1b5efa74f9f82e8509858f", size = 143892, upload-time = "2024-12-24T18:09:52.078Z" }, - { url = "https://files.pythonhosted.org/packages/93/62/5e89cdfe04584cb7f4d36003ffa2936681b03ecc0754f8e969c2becb7e24/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b010a7a4fd316c3c484d482922d13044979e78d1861f0e0650423144c616a46a", size = 146123, upload-time = "2024-12-24T18:09:54.575Z" }, - { url = "https://files.pythonhosted.org/packages/a9/ac/ab729a15c516da2ab70a05f8722ecfccc3f04ed7a18e45c75bbbaa347d61/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4532bff1b8421fd0a320463030c7520f56a79c9024a4e88f01c537316019005a", size = 147943, upload-time = "2024-12-24T18:09:57.324Z" }, - { url = "https://files.pythonhosted.org/packages/03/d2/3f392f23f042615689456e9a274640c1d2e5dd1d52de36ab8f7955f8f050/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:d973f03c0cb71c5ed99037b870f2be986c3c05e63622c017ea9816881d2dd247", size = 142063, upload-time = "2024-12-24T18:09:59.794Z" }, - { url = "https://files.pythonhosted.org/packages/f2/e3/e20aae5e1039a2cd9b08d9205f52142329f887f8cf70da3650326670bddf/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:3a3bd0dcd373514dcec91c411ddb9632c0d7d92aed7093b8c3bbb6d69ca74408", size = 150578, upload-time = "2024-12-24T18:10:02.357Z" }, - { url = "https://files.pythonhosted.org/packages/8d/af/779ad72a4da0aed925e1139d458adc486e61076d7ecdcc09e610ea8678db/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:d9c3cdf5390dcd29aa8056d13e8e99526cda0305acc038b96b30352aff5ff2bb", size = 153629, upload-time = "2024-12-24T18:10:03.678Z" }, - { url = "https://files.pythonhosted.org/packages/c2/b6/7aa450b278e7aa92cf7732140bfd8be21f5f29d5bf334ae987c945276639/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:2bdfe3ac2e1bbe5b59a1a63721eb3b95fc9b6817ae4a46debbb4e11f6232428d", size = 150778, upload-time = "2024-12-24T18:10:06.197Z" }, - { url = "https://files.pythonhosted.org/packages/39/f4/d9f4f712d0951dcbfd42920d3db81b00dd23b6ab520419626f4023334056/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:eab677309cdb30d047996b36d34caeda1dc91149e4fdca0b1a039b3f79d9a807", size = 146453, upload-time = "2024-12-24T18:10:08.848Z" }, - { url = "https://files.pythonhosted.org/packages/49/2b/999d0314e4ee0cff3cb83e6bc9aeddd397eeed693edb4facb901eb8fbb69/charset_normalizer-3.4.1-cp310-cp310-win32.whl", hash = "sha256:c0429126cf75e16c4f0ad00ee0eae4242dc652290f940152ca8c75c3a4b6ee8f", size = 95479, upload-time = "2024-12-24T18:10:10.044Z" }, - { url = "https://files.pythonhosted.org/packages/2d/ce/3cbed41cff67e455a386fb5e5dd8906cdda2ed92fbc6297921f2e4419309/charset_normalizer-3.4.1-cp310-cp310-win_amd64.whl", hash = "sha256:9f0b8b1c6d84c8034a44893aba5e767bf9c7a211e313a9605d9c617d7083829f", size = 102790, upload-time = "2024-12-24T18:10:11.323Z" }, - { url = "https://files.pythonhosted.org/packages/72/80/41ef5d5a7935d2d3a773e3eaebf0a9350542f2cab4eac59a7a4741fbbbbe/charset_normalizer-3.4.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:8bfa33f4f2672964266e940dd22a195989ba31669bd84629f05fab3ef4e2d125", size = 194995, upload-time = "2024-12-24T18:10:12.838Z" }, - { url = "https://files.pythonhosted.org/packages/7a/28/0b9fefa7b8b080ec492110af6d88aa3dea91c464b17d53474b6e9ba5d2c5/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:28bf57629c75e810b6ae989f03c0828d64d6b26a5e205535585f96093e405ed1", size = 139471, upload-time = "2024-12-24T18:10:14.101Z" }, - { url = "https://files.pythonhosted.org/packages/71/64/d24ab1a997efb06402e3fc07317e94da358e2585165930d9d59ad45fcae2/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f08ff5e948271dc7e18a35641d2f11a4cd8dfd5634f55228b691e62b37125eb3", size = 149831, upload-time = "2024-12-24T18:10:15.512Z" }, - { url = "https://files.pythonhosted.org/packages/37/ed/be39e5258e198655240db5e19e0b11379163ad7070962d6b0c87ed2c4d39/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:234ac59ea147c59ee4da87a0c0f098e9c8d169f4dc2a159ef720f1a61bbe27cd", size = 142335, upload-time = "2024-12-24T18:10:18.369Z" }, - { url = "https://files.pythonhosted.org/packages/88/83/489e9504711fa05d8dde1574996408026bdbdbd938f23be67deebb5eca92/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd4ec41f914fa74ad1b8304bbc634b3de73d2a0889bd32076342a573e0779e00", size = 143862, upload-time = "2024-12-24T18:10:19.743Z" }, - { url = "https://files.pythonhosted.org/packages/c6/c7/32da20821cf387b759ad24627a9aca289d2822de929b8a41b6241767b461/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:eea6ee1db730b3483adf394ea72f808b6e18cf3cb6454b4d86e04fa8c4327a12", size = 145673, upload-time = "2024-12-24T18:10:21.139Z" }, - { url = "https://files.pythonhosted.org/packages/68/85/f4288e96039abdd5aeb5c546fa20a37b50da71b5cf01e75e87f16cd43304/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c96836c97b1238e9c9e3fe90844c947d5afbf4f4c92762679acfe19927d81d77", size = 140211, upload-time = "2024-12-24T18:10:22.382Z" }, - { url = "https://files.pythonhosted.org/packages/28/a3/a42e70d03cbdabc18997baf4f0227c73591a08041c149e710045c281f97b/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:4d86f7aff21ee58f26dcf5ae81a9addbd914115cdebcbb2217e4f0ed8982e146", size = 148039, upload-time = "2024-12-24T18:10:24.802Z" }, - { url = "https://files.pythonhosted.org/packages/85/e4/65699e8ab3014ecbe6f5c71d1a55d810fb716bbfd74f6283d5c2aa87febf/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:09b5e6733cbd160dcc09589227187e242a30a49ca5cefa5a7edd3f9d19ed53fd", size = 151939, upload-time = "2024-12-24T18:10:26.124Z" }, - { url = "https://files.pythonhosted.org/packages/b1/82/8e9fe624cc5374193de6860aba3ea8070f584c8565ee77c168ec13274bd2/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:5777ee0881f9499ed0f71cc82cf873d9a0ca8af166dfa0af8ec4e675b7df48e6", size = 149075, upload-time = "2024-12-24T18:10:30.027Z" }, - { url = "https://files.pythonhosted.org/packages/3d/7b/82865ba54c765560c8433f65e8acb9217cb839a9e32b42af4aa8e945870f/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:237bdbe6159cff53b4f24f397d43c6336c6b0b42affbe857970cefbb620911c8", size = 144340, upload-time = "2024-12-24T18:10:32.679Z" }, - { url = "https://files.pythonhosted.org/packages/b5/b6/9674a4b7d4d99a0d2df9b215da766ee682718f88055751e1e5e753c82db0/charset_normalizer-3.4.1-cp311-cp311-win32.whl", hash = "sha256:8417cb1f36cc0bc7eaba8ccb0e04d55f0ee52df06df3ad55259b9a323555fc8b", size = 95205, upload-time = "2024-12-24T18:10:34.724Z" }, - { url = "https://files.pythonhosted.org/packages/1e/ab/45b180e175de4402dcf7547e4fb617283bae54ce35c27930a6f35b6bef15/charset_normalizer-3.4.1-cp311-cp311-win_amd64.whl", hash = "sha256:d7f50a1f8c450f3925cb367d011448c39239bb3eb4117c36a6d354794de4ce76", size = 102441, upload-time = "2024-12-24T18:10:37.574Z" }, - { url = "https://files.pythonhosted.org/packages/0a/9a/dd1e1cdceb841925b7798369a09279bd1cf183cef0f9ddf15a3a6502ee45/charset_normalizer-3.4.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:73d94b58ec7fecbc7366247d3b0b10a21681004153238750bb67bd9012414545", size = 196105, upload-time = "2024-12-24T18:10:38.83Z" }, - { url = "https://files.pythonhosted.org/packages/d3/8c/90bfabf8c4809ecb648f39794cf2a84ff2e7d2a6cf159fe68d9a26160467/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dad3e487649f498dd991eeb901125411559b22e8d7ab25d3aeb1af367df5efd7", size = 140404, upload-time = "2024-12-24T18:10:44.272Z" }, - { url = "https://files.pythonhosted.org/packages/ad/8f/e410d57c721945ea3b4f1a04b74f70ce8fa800d393d72899f0a40526401f/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c30197aa96e8eed02200a83fba2657b4c3acd0f0aa4bdc9f6c1af8e8962e0757", size = 150423, upload-time = "2024-12-24T18:10:45.492Z" }, - { url = "https://files.pythonhosted.org/packages/f0/b8/e6825e25deb691ff98cf5c9072ee0605dc2acfca98af70c2d1b1bc75190d/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2369eea1ee4a7610a860d88f268eb39b95cb588acd7235e02fd5a5601773d4fa", size = 143184, upload-time = "2024-12-24T18:10:47.898Z" }, - { url = "https://files.pythonhosted.org/packages/3e/a2/513f6cbe752421f16d969e32f3583762bfd583848b763913ddab8d9bfd4f/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc2722592d8998c870fa4e290c2eec2c1569b87fe58618e67d38b4665dfa680d", size = 145268, upload-time = "2024-12-24T18:10:50.589Z" }, - { url = "https://files.pythonhosted.org/packages/74/94/8a5277664f27c3c438546f3eb53b33f5b19568eb7424736bdc440a88a31f/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ffc9202a29ab3920fa812879e95a9e78b2465fd10be7fcbd042899695d75e616", size = 147601, upload-time = "2024-12-24T18:10:52.541Z" }, - { url = "https://files.pythonhosted.org/packages/7c/5f/6d352c51ee763623a98e31194823518e09bfa48be2a7e8383cf691bbb3d0/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:804a4d582ba6e5b747c625bf1255e6b1507465494a40a2130978bda7b932c90b", size = 141098, upload-time = "2024-12-24T18:10:53.789Z" }, - { url = "https://files.pythonhosted.org/packages/78/d4/f5704cb629ba5ab16d1d3d741396aec6dc3ca2b67757c45b0599bb010478/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:0f55e69f030f7163dffe9fd0752b32f070566451afe180f99dbeeb81f511ad8d", size = 149520, upload-time = "2024-12-24T18:10:55.048Z" }, - { url = "https://files.pythonhosted.org/packages/c5/96/64120b1d02b81785f222b976c0fb79a35875457fa9bb40827678e54d1bc8/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:c4c3e6da02df6fa1410a7680bd3f63d4f710232d3139089536310d027950696a", size = 152852, upload-time = "2024-12-24T18:10:57.647Z" }, - { url = "https://files.pythonhosted.org/packages/84/c9/98e3732278a99f47d487fd3468bc60b882920cef29d1fa6ca460a1fdf4e6/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:5df196eb874dae23dcfb968c83d4f8fdccb333330fe1fc278ac5ceeb101003a9", size = 150488, upload-time = "2024-12-24T18:10:59.43Z" }, - { url = "https://files.pythonhosted.org/packages/13/0e/9c8d4cb99c98c1007cc11eda969ebfe837bbbd0acdb4736d228ccaabcd22/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e358e64305fe12299a08e08978f51fc21fac060dcfcddd95453eabe5b93ed0e1", size = 146192, upload-time = "2024-12-24T18:11:00.676Z" }, - { url = "https://files.pythonhosted.org/packages/b2/21/2b6b5b860781a0b49427309cb8670785aa543fb2178de875b87b9cc97746/charset_normalizer-3.4.1-cp312-cp312-win32.whl", hash = "sha256:9b23ca7ef998bc739bf6ffc077c2116917eabcc901f88da1b9856b210ef63f35", size = 95550, upload-time = "2024-12-24T18:11:01.952Z" }, - { url = "https://files.pythonhosted.org/packages/21/5b/1b390b03b1d16c7e382b561c5329f83cc06623916aab983e8ab9239c7d5c/charset_normalizer-3.4.1-cp312-cp312-win_amd64.whl", hash = "sha256:6ff8a4a60c227ad87030d76e99cd1698345d4491638dfa6673027c48b3cd395f", size = 102785, upload-time = "2024-12-24T18:11:03.142Z" }, - { url = "https://files.pythonhosted.org/packages/38/94/ce8e6f63d18049672c76d07d119304e1e2d7c6098f0841b51c666e9f44a0/charset_normalizer-3.4.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:aabfa34badd18f1da5ec1bc2715cadc8dca465868a4e73a0173466b688f29dda", size = 195698, upload-time = "2024-12-24T18:11:05.834Z" }, - { url = "https://files.pythonhosted.org/packages/24/2e/dfdd9770664aae179a96561cc6952ff08f9a8cd09a908f259a9dfa063568/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:22e14b5d70560b8dd51ec22863f370d1e595ac3d024cb8ad7d308b4cd95f8313", size = 140162, upload-time = "2024-12-24T18:11:07.064Z" }, - { url = "https://files.pythonhosted.org/packages/24/4e/f646b9093cff8fc86f2d60af2de4dc17c759de9d554f130b140ea4738ca6/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8436c508b408b82d87dc5f62496973a1805cd46727c34440b0d29d8a2f50a6c9", size = 150263, upload-time = "2024-12-24T18:11:08.374Z" }, - { url = "https://files.pythonhosted.org/packages/5e/67/2937f8d548c3ef6e2f9aab0f6e21001056f692d43282b165e7c56023e6dd/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2d074908e1aecee37a7635990b2c6d504cd4766c7bc9fc86d63f9c09af3fa11b", size = 142966, upload-time = "2024-12-24T18:11:09.831Z" }, - { url = "https://files.pythonhosted.org/packages/52/ed/b7f4f07de100bdb95c1756d3a4d17b90c1a3c53715c1a476f8738058e0fa/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:955f8851919303c92343d2f66165294848d57e9bba6cf6e3625485a70a038d11", size = 144992, upload-time = "2024-12-24T18:11:12.03Z" }, - { url = "https://files.pythonhosted.org/packages/96/2c/d49710a6dbcd3776265f4c923bb73ebe83933dfbaa841c5da850fe0fd20b/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:44ecbf16649486d4aebafeaa7ec4c9fed8b88101f4dd612dcaf65d5e815f837f", size = 147162, upload-time = "2024-12-24T18:11:13.372Z" }, - { url = "https://files.pythonhosted.org/packages/b4/41/35ff1f9a6bd380303dea55e44c4933b4cc3c4850988927d4082ada230273/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:0924e81d3d5e70f8126529951dac65c1010cdf117bb75eb02dd12339b57749dd", size = 140972, upload-time = "2024-12-24T18:11:14.628Z" }, - { url = "https://files.pythonhosted.org/packages/fb/43/c6a0b685fe6910d08ba971f62cd9c3e862a85770395ba5d9cad4fede33ab/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:2967f74ad52c3b98de4c3b32e1a44e32975e008a9cd2a8cc8966d6a5218c5cb2", size = 149095, upload-time = "2024-12-24T18:11:17.672Z" }, - { url = "https://files.pythonhosted.org/packages/4c/ff/a9a504662452e2d2878512115638966e75633519ec11f25fca3d2049a94a/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:c75cb2a3e389853835e84a2d8fb2b81a10645b503eca9bcb98df6b5a43eb8886", size = 152668, upload-time = "2024-12-24T18:11:18.989Z" }, - { url = "https://files.pythonhosted.org/packages/6c/71/189996b6d9a4b932564701628af5cee6716733e9165af1d5e1b285c530ed/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:09b26ae6b1abf0d27570633b2b078a2a20419c99d66fb2823173d73f188ce601", size = 150073, upload-time = "2024-12-24T18:11:21.507Z" }, - { url = "https://files.pythonhosted.org/packages/e4/93/946a86ce20790e11312c87c75ba68d5f6ad2208cfb52b2d6a2c32840d922/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:fa88b843d6e211393a37219e6a1c1df99d35e8fd90446f1118f4216e307e48cd", size = 145732, upload-time = "2024-12-24T18:11:22.774Z" }, - { url = "https://files.pythonhosted.org/packages/cd/e5/131d2fb1b0dddafc37be4f3a2fa79aa4c037368be9423061dccadfd90091/charset_normalizer-3.4.1-cp313-cp313-win32.whl", hash = "sha256:eb8178fe3dba6450a3e024e95ac49ed3400e506fd4e9e5c32d30adda88cbd407", size = 95391, upload-time = "2024-12-24T18:11:24.139Z" }, - { url = "https://files.pythonhosted.org/packages/27/f2/4f9a69cc7712b9b5ad8fdb87039fd89abba997ad5cbe690d1835d40405b0/charset_normalizer-3.4.1-cp313-cp313-win_amd64.whl", hash = "sha256:b1ac5992a838106edb89654e0aebfc24f5848ae2547d22c2c3f66454daa11971", size = 102702, upload-time = "2024-12-24T18:11:26.535Z" }, - { url = "https://files.pythonhosted.org/packages/0e/f6/65ecc6878a89bb1c23a086ea335ad4bf21a588990c3f535a227b9eea9108/charset_normalizer-3.4.1-py3-none-any.whl", hash = "sha256:d98b1668f06378c6dbefec3b92299716b931cd4e6061f3c875a71ced1780ab85", size = 49767, upload-time = "2024-12-24T18:12:32.852Z" }, +version = "3.4.2" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/e4/33/89c2ced2b67d1c2a61c19c6751aa8902d46ce3dacb23600a283619f5a12d/charset_normalizer-3.4.2.tar.gz", hash = "sha256:5baececa9ecba31eff645232d59845c07aa030f0c81ee70184a90d35099a0e63", size = 126367, upload-time = "2025-05-02T08:34:42.01Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/95/28/9901804da60055b406e1a1c5ba7aac1276fb77f1dde635aabfc7fd84b8ab/charset_normalizer-3.4.2-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:7c48ed483eb946e6c04ccbe02c6b4d1d48e51944b6db70f697e089c193404941", size = 201818, upload-time = "2025-05-02T08:31:46.725Z" }, + { url = "https://files.pythonhosted.org/packages/d9/9b/892a8c8af9110935e5adcbb06d9c6fe741b6bb02608c6513983048ba1a18/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b2d318c11350e10662026ad0eb71bb51c7812fc8590825304ae0bdd4ac283acd", size = 144649, upload-time = "2025-05-02T08:31:48.889Z" }, + { url = "https://files.pythonhosted.org/packages/7b/a5/4179abd063ff6414223575e008593861d62abfc22455b5d1a44995b7c101/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9cbfacf36cb0ec2897ce0ebc5d08ca44213af24265bd56eca54bee7923c48fd6", size = 155045, upload-time = "2025-05-02T08:31:50.757Z" }, + { url = "https://files.pythonhosted.org/packages/3b/95/bc08c7dfeddd26b4be8c8287b9bb055716f31077c8b0ea1cd09553794665/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:18dd2e350387c87dabe711b86f83c9c78af772c748904d372ade190b5c7c9d4d", size = 147356, upload-time = "2025-05-02T08:31:52.634Z" }, + { url = "https://files.pythonhosted.org/packages/a8/2d/7a5b635aa65284bf3eab7653e8b4151ab420ecbae918d3e359d1947b4d61/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8075c35cd58273fee266c58c0c9b670947c19df5fb98e7b66710e04ad4e9ff86", size = 149471, upload-time = "2025-05-02T08:31:56.207Z" }, + { url = "https://files.pythonhosted.org/packages/ae/38/51fc6ac74251fd331a8cfdb7ec57beba8c23fd5493f1050f71c87ef77ed0/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5bf4545e3b962767e5c06fe1738f951f77d27967cb2caa64c28be7c4563e162c", size = 151317, upload-time = "2025-05-02T08:31:57.613Z" }, + { url = "https://files.pythonhosted.org/packages/b7/17/edee1e32215ee6e9e46c3e482645b46575a44a2d72c7dfd49e49f60ce6bf/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:7a6ab32f7210554a96cd9e33abe3ddd86732beeafc7a28e9955cdf22ffadbab0", size = 146368, upload-time = "2025-05-02T08:31:59.468Z" }, + { url = "https://files.pythonhosted.org/packages/26/2c/ea3e66f2b5f21fd00b2825c94cafb8c326ea6240cd80a91eb09e4a285830/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:b33de11b92e9f75a2b545d6e9b6f37e398d86c3e9e9653c4864eb7e89c5773ef", size = 154491, upload-time = "2025-05-02T08:32:01.219Z" }, + { url = "https://files.pythonhosted.org/packages/52/47/7be7fa972422ad062e909fd62460d45c3ef4c141805b7078dbab15904ff7/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:8755483f3c00d6c9a77f490c17e6ab0c8729e39e6390328e42521ef175380ae6", size = 157695, upload-time = "2025-05-02T08:32:03.045Z" }, + { url = "https://files.pythonhosted.org/packages/2f/42/9f02c194da282b2b340f28e5fb60762de1151387a36842a92b533685c61e/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:68a328e5f55ec37c57f19ebb1fdc56a248db2e3e9ad769919a58672958e8f366", size = 154849, upload-time = "2025-05-02T08:32:04.651Z" }, + { url = "https://files.pythonhosted.org/packages/67/44/89cacd6628f31fb0b63201a618049be4be2a7435a31b55b5eb1c3674547a/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:21b2899062867b0e1fde9b724f8aecb1af14f2778d69aacd1a5a1853a597a5db", size = 150091, upload-time = "2025-05-02T08:32:06.719Z" }, + { url = "https://files.pythonhosted.org/packages/1f/79/4b8da9f712bc079c0f16b6d67b099b0b8d808c2292c937f267d816ec5ecc/charset_normalizer-3.4.2-cp310-cp310-win32.whl", hash = "sha256:e8082b26888e2f8b36a042a58307d5b917ef2b1cacab921ad3323ef91901c71a", size = 98445, upload-time = "2025-05-02T08:32:08.66Z" }, + { url = "https://files.pythonhosted.org/packages/7d/d7/96970afb4fb66497a40761cdf7bd4f6fca0fc7bafde3a84f836c1f57a926/charset_normalizer-3.4.2-cp310-cp310-win_amd64.whl", hash = "sha256:f69a27e45c43520f5487f27627059b64aaf160415589230992cec34c5e18a509", size = 105782, upload-time = "2025-05-02T08:32:10.46Z" }, + { url = "https://files.pythonhosted.org/packages/05/85/4c40d00dcc6284a1c1ad5de5e0996b06f39d8232f1031cd23c2f5c07ee86/charset_normalizer-3.4.2-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:be1e352acbe3c78727a16a455126d9ff83ea2dfdcbc83148d2982305a04714c2", size = 198794, upload-time = "2025-05-02T08:32:11.945Z" }, + { url = "https://files.pythonhosted.org/packages/41/d9/7a6c0b9db952598e97e93cbdfcb91bacd89b9b88c7c983250a77c008703c/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa88ca0b1932e93f2d961bf3addbb2db902198dca337d88c89e1559e066e7645", size = 142846, upload-time = "2025-05-02T08:32:13.946Z" }, + { url = "https://files.pythonhosted.org/packages/66/82/a37989cda2ace7e37f36c1a8ed16c58cf48965a79c2142713244bf945c89/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d524ba3f1581b35c03cb42beebab4a13e6cdad7b36246bd22541fa585a56cccd", size = 153350, upload-time = "2025-05-02T08:32:15.873Z" }, + { url = "https://files.pythonhosted.org/packages/df/68/a576b31b694d07b53807269d05ec3f6f1093e9545e8607121995ba7a8313/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:28a1005facc94196e1fb3e82a3d442a9d9110b8434fc1ded7a24a2983c9888d8", size = 145657, upload-time = "2025-05-02T08:32:17.283Z" }, + { url = "https://files.pythonhosted.org/packages/92/9b/ad67f03d74554bed3aefd56fe836e1623a50780f7c998d00ca128924a499/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fdb20a30fe1175ecabed17cbf7812f7b804b8a315a25f24678bcdf120a90077f", size = 147260, upload-time = "2025-05-02T08:32:18.807Z" }, + { url = "https://files.pythonhosted.org/packages/a6/e6/8aebae25e328160b20e31a7e9929b1578bbdc7f42e66f46595a432f8539e/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0f5d9ed7f254402c9e7d35d2f5972c9bbea9040e99cd2861bd77dc68263277c7", size = 149164, upload-time = "2025-05-02T08:32:20.333Z" }, + { url = "https://files.pythonhosted.org/packages/8b/f2/b3c2f07dbcc248805f10e67a0262c93308cfa149a4cd3d1fe01f593e5fd2/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:efd387a49825780ff861998cd959767800d54f8308936b21025326de4b5a42b9", size = 144571, upload-time = "2025-05-02T08:32:21.86Z" }, + { url = "https://files.pythonhosted.org/packages/60/5b/c3f3a94bc345bc211622ea59b4bed9ae63c00920e2e8f11824aa5708e8b7/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:f0aa37f3c979cf2546b73e8222bbfa3dc07a641585340179d768068e3455e544", size = 151952, upload-time = "2025-05-02T08:32:23.434Z" }, + { url = "https://files.pythonhosted.org/packages/e2/4d/ff460c8b474122334c2fa394a3f99a04cf11c646da895f81402ae54f5c42/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:e70e990b2137b29dc5564715de1e12701815dacc1d056308e2b17e9095372a82", size = 155959, upload-time = "2025-05-02T08:32:24.993Z" }, + { url = "https://files.pythonhosted.org/packages/a2/2b/b964c6a2fda88611a1fe3d4c400d39c66a42d6c169c924818c848f922415/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:0c8c57f84ccfc871a48a47321cfa49ae1df56cd1d965a09abe84066f6853b9c0", size = 153030, upload-time = "2025-05-02T08:32:26.435Z" }, + { url = "https://files.pythonhosted.org/packages/59/2e/d3b9811db26a5ebf444bc0fa4f4be5aa6d76fc6e1c0fd537b16c14e849b6/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:6b66f92b17849b85cad91259efc341dce9c1af48e2173bf38a85c6329f1033e5", size = 148015, upload-time = "2025-05-02T08:32:28.376Z" }, + { url = "https://files.pythonhosted.org/packages/90/07/c5fd7c11eafd561bb51220d600a788f1c8d77c5eef37ee49454cc5c35575/charset_normalizer-3.4.2-cp311-cp311-win32.whl", hash = "sha256:daac4765328a919a805fa5e2720f3e94767abd632ae410a9062dff5412bae65a", size = 98106, upload-time = "2025-05-02T08:32:30.281Z" }, + { url = "https://files.pythonhosted.org/packages/a8/05/5e33dbef7e2f773d672b6d79f10ec633d4a71cd96db6673625838a4fd532/charset_normalizer-3.4.2-cp311-cp311-win_amd64.whl", hash = "sha256:e53efc7c7cee4c1e70661e2e112ca46a575f90ed9ae3fef200f2a25e954f4b28", size = 105402, upload-time = "2025-05-02T08:32:32.191Z" }, + { url = "https://files.pythonhosted.org/packages/d7/a4/37f4d6035c89cac7930395a35cc0f1b872e652eaafb76a6075943754f095/charset_normalizer-3.4.2-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:0c29de6a1a95f24b9a1aa7aefd27d2487263f00dfd55a77719b530788f75cff7", size = 199936, upload-time = "2025-05-02T08:32:33.712Z" }, + { url = "https://files.pythonhosted.org/packages/ee/8a/1a5e33b73e0d9287274f899d967907cd0bf9c343e651755d9307e0dbf2b3/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cddf7bd982eaa998934a91f69d182aec997c6c468898efe6679af88283b498d3", size = 143790, upload-time = "2025-05-02T08:32:35.768Z" }, + { url = "https://files.pythonhosted.org/packages/66/52/59521f1d8e6ab1482164fa21409c5ef44da3e9f653c13ba71becdd98dec3/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fcbe676a55d7445b22c10967bceaaf0ee69407fbe0ece4d032b6eb8d4565982a", size = 153924, upload-time = "2025-05-02T08:32:37.284Z" }, + { url = "https://files.pythonhosted.org/packages/86/2d/fb55fdf41964ec782febbf33cb64be480a6b8f16ded2dbe8db27a405c09f/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d41c4d287cfc69060fa91cae9683eacffad989f1a10811995fa309df656ec214", size = 146626, upload-time = "2025-05-02T08:32:38.803Z" }, + { url = "https://files.pythonhosted.org/packages/8c/73/6ede2ec59bce19b3edf4209d70004253ec5f4e319f9a2e3f2f15601ed5f7/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4e594135de17ab3866138f496755f302b72157d115086d100c3f19370839dd3a", size = 148567, upload-time = "2025-05-02T08:32:40.251Z" }, + { url = "https://files.pythonhosted.org/packages/09/14/957d03c6dc343c04904530b6bef4e5efae5ec7d7990a7cbb868e4595ee30/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:cf713fe9a71ef6fd5adf7a79670135081cd4431c2943864757f0fa3a65b1fafd", size = 150957, upload-time = "2025-05-02T08:32:41.705Z" }, + { url = "https://files.pythonhosted.org/packages/0d/c8/8174d0e5c10ccebdcb1b53cc959591c4c722a3ad92461a273e86b9f5a302/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:a370b3e078e418187da8c3674eddb9d983ec09445c99a3a263c2011993522981", size = 145408, upload-time = "2025-05-02T08:32:43.709Z" }, + { url = "https://files.pythonhosted.org/packages/58/aa/8904b84bc8084ac19dc52feb4f5952c6df03ffb460a887b42615ee1382e8/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:a955b438e62efdf7e0b7b52a64dc5c3396e2634baa62471768a64bc2adb73d5c", size = 153399, upload-time = "2025-05-02T08:32:46.197Z" }, + { url = "https://files.pythonhosted.org/packages/c2/26/89ee1f0e264d201cb65cf054aca6038c03b1a0c6b4ae998070392a3ce605/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:7222ffd5e4de8e57e03ce2cef95a4c43c98fcb72ad86909abdfc2c17d227fc1b", size = 156815, upload-time = "2025-05-02T08:32:48.105Z" }, + { url = "https://files.pythonhosted.org/packages/fd/07/68e95b4b345bad3dbbd3a8681737b4338ff2c9df29856a6d6d23ac4c73cb/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:bee093bf902e1d8fc0ac143c88902c3dfc8941f7ea1d6a8dd2bcb786d33db03d", size = 154537, upload-time = "2025-05-02T08:32:49.719Z" }, + { url = "https://files.pythonhosted.org/packages/77/1a/5eefc0ce04affb98af07bc05f3bac9094513c0e23b0562d64af46a06aae4/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:dedb8adb91d11846ee08bec4c8236c8549ac721c245678282dcb06b221aab59f", size = 149565, upload-time = "2025-05-02T08:32:51.404Z" }, + { url = "https://files.pythonhosted.org/packages/37/a0/2410e5e6032a174c95e0806b1a6585eb21e12f445ebe239fac441995226a/charset_normalizer-3.4.2-cp312-cp312-win32.whl", hash = "sha256:db4c7bf0e07fc3b7d89ac2a5880a6a8062056801b83ff56d8464b70f65482b6c", size = 98357, upload-time = "2025-05-02T08:32:53.079Z" }, + { url = "https://files.pythonhosted.org/packages/6c/4f/c02d5c493967af3eda9c771ad4d2bbc8df6f99ddbeb37ceea6e8716a32bc/charset_normalizer-3.4.2-cp312-cp312-win_amd64.whl", hash = "sha256:5a9979887252a82fefd3d3ed2a8e3b937a7a809f65dcb1e068b090e165bbe99e", size = 105776, upload-time = "2025-05-02T08:32:54.573Z" }, + { url = "https://files.pythonhosted.org/packages/ea/12/a93df3366ed32db1d907d7593a94f1fe6293903e3e92967bebd6950ed12c/charset_normalizer-3.4.2-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:926ca93accd5d36ccdabd803392ddc3e03e6d4cd1cf17deff3b989ab8e9dbcf0", size = 199622, upload-time = "2025-05-02T08:32:56.363Z" }, + { url = "https://files.pythonhosted.org/packages/04/93/bf204e6f344c39d9937d3c13c8cd5bbfc266472e51fc8c07cb7f64fcd2de/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:eba9904b0f38a143592d9fc0e19e2df0fa2e41c3c3745554761c5f6447eedabf", size = 143435, upload-time = "2025-05-02T08:32:58.551Z" }, + { url = "https://files.pythonhosted.org/packages/22/2a/ea8a2095b0bafa6c5b5a55ffdc2f924455233ee7b91c69b7edfcc9e02284/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3fddb7e2c84ac87ac3a947cb4e66d143ca5863ef48e4a5ecb83bd48619e4634e", size = 153653, upload-time = "2025-05-02T08:33:00.342Z" }, + { url = "https://files.pythonhosted.org/packages/b6/57/1b090ff183d13cef485dfbe272e2fe57622a76694061353c59da52c9a659/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:98f862da73774290f251b9df8d11161b6cf25b599a66baf087c1ffe340e9bfd1", size = 146231, upload-time = "2025-05-02T08:33:02.081Z" }, + { url = "https://files.pythonhosted.org/packages/e2/28/ffc026b26f441fc67bd21ab7f03b313ab3fe46714a14b516f931abe1a2d8/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c9379d65defcab82d07b2a9dfbfc2e95bc8fe0ebb1b176a3190230a3ef0e07c", size = 148243, upload-time = "2025-05-02T08:33:04.063Z" }, + { url = "https://files.pythonhosted.org/packages/c0/0f/9abe9bd191629c33e69e47c6ef45ef99773320e9ad8e9cb08b8ab4a8d4cb/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e635b87f01ebc977342e2697d05b56632f5f879a4f15955dfe8cef2448b51691", size = 150442, upload-time = "2025-05-02T08:33:06.418Z" }, + { url = "https://files.pythonhosted.org/packages/67/7c/a123bbcedca91d5916c056407f89a7f5e8fdfce12ba825d7d6b9954a1a3c/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:1c95a1e2902a8b722868587c0e1184ad5c55631de5afc0eb96bc4b0d738092c0", size = 145147, upload-time = "2025-05-02T08:33:08.183Z" }, + { url = "https://files.pythonhosted.org/packages/ec/fe/1ac556fa4899d967b83e9893788e86b6af4d83e4726511eaaad035e36595/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:ef8de666d6179b009dce7bcb2ad4c4a779f113f12caf8dc77f0162c29d20490b", size = 153057, upload-time = "2025-05-02T08:33:09.986Z" }, + { url = "https://files.pythonhosted.org/packages/2b/ff/acfc0b0a70b19e3e54febdd5301a98b72fa07635e56f24f60502e954c461/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:32fc0341d72e0f73f80acb0a2c94216bd704f4f0bce10aedea38f30502b271ff", size = 156454, upload-time = "2025-05-02T08:33:11.814Z" }, + { url = "https://files.pythonhosted.org/packages/92/08/95b458ce9c740d0645feb0e96cea1f5ec946ea9c580a94adfe0b617f3573/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:289200a18fa698949d2b39c671c2cc7a24d44096784e76614899a7ccf2574b7b", size = 154174, upload-time = "2025-05-02T08:33:13.707Z" }, + { url = "https://files.pythonhosted.org/packages/78/be/8392efc43487ac051eee6c36d5fbd63032d78f7728cb37aebcc98191f1ff/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4a476b06fbcf359ad25d34a057b7219281286ae2477cc5ff5e3f70a246971148", size = 149166, upload-time = "2025-05-02T08:33:15.458Z" }, + { url = "https://files.pythonhosted.org/packages/44/96/392abd49b094d30b91d9fbda6a69519e95802250b777841cf3bda8fe136c/charset_normalizer-3.4.2-cp313-cp313-win32.whl", hash = "sha256:aaeeb6a479c7667fbe1099af9617c83aaca22182d6cf8c53966491a0f1b7ffb7", size = 98064, upload-time = "2025-05-02T08:33:17.06Z" }, + { url = "https://files.pythonhosted.org/packages/e9/b0/0200da600134e001d91851ddc797809e2fe0ea72de90e09bec5a2fbdaccb/charset_normalizer-3.4.2-cp313-cp313-win_amd64.whl", hash = "sha256:aa6af9e7d59f9c12b33ae4e9450619cf2488e2bbe9b44030905877f0b2324980", size = 105641, upload-time = "2025-05-02T08:33:18.753Z" }, + { url = "https://files.pythonhosted.org/packages/20/94/c5790835a017658cbfabd07f3bfb549140c3ac458cfc196323996b10095a/charset_normalizer-3.4.2-py3-none-any.whl", hash = "sha256:7f56930ab0abd1c45cd15be65cc741c28b1c9a34876ce8c17a2fa107810c0af0", size = 52626, upload-time = "2025-05-02T08:34:40.053Z" }, ] [[package]] name = "click" -version = "8.1.8" +version = "8.2.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "colorama", marker = "sys_platform == 'win32'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/b9/2e/0090cbf739cee7d23781ad4b89a9894a41538e4fcf4c31dcdd705b78eb8b/click-8.1.8.tar.gz", hash = "sha256:ed53c9d8990d83c2a27deae68e4ee337473f6330c040a31d4225c9574d16096a", size = 226593, upload-time = "2024-12-21T18:38:44.339Z" } +sdist = { url = "https://files.pythonhosted.org/packages/60/6c/8ca2efa64cf75a977a0d7fac081354553ebe483345c734fb6b6515d96bbc/click-8.2.1.tar.gz", hash = "sha256:27c491cc05d968d271d5a1db13e3b5a184636d9d930f148c50b038f0d0646202", size = 286342, upload-time = "2025-05-20T23:19:49.832Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/7e/d4/7ebdbd03970677812aac39c869717059dbb71a4cfc033ca6e5221787892c/click-8.1.8-py3-none-any.whl", hash = "sha256:63c132bbbed01578a06712a2d1f497bb62d9c1c0d329b7903a866228027263b2", size = 98188, upload-time = "2024-12-21T18:38:41.666Z" }, + { url = "https://files.pythonhosted.org/packages/85/32/10bb5764d90a8eee674e9dc6f4db6a0ab47c8c4d0d83c27f7c39ac415a4d/click-8.2.1-py3-none-any.whl", hash = "sha256:61a3265b914e850b85317d0b3109c7f8cd35a670f963866005d6ef1d5175a12b", size = 102215, upload-time = "2025-05-20T23:19:47.796Z" }, ] [[package]] @@ -432,78 +467,172 @@ wheels = [ [[package]] name = "comm" -version = "0.2.2" +version = "0.2.3" source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "traitlets" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/e9/a8/fb783cb0abe2b5fded9f55e5703015cdf1c9c85b3669087c538dd15a6a86/comm-0.2.2.tar.gz", hash = "sha256:3fd7a84065306e07bea1773df6eb8282de51ba82f77c72f9c85716ab11fe980e", size = 6210, upload-time = "2024-03-12T16:53:41.133Z" } +sdist = { url = "https://files.pythonhosted.org/packages/4c/13/7d740c5849255756bc17888787313b61fd38a0a8304fc4f073dfc46122aa/comm-0.2.3.tar.gz", hash = "sha256:2dc8048c10962d55d7ad693be1e7045d891b7ce8d999c97963a5e3e99c055971", size = 6319, upload-time = "2025-07-25T14:02:04.452Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e6/75/49e5bfe642f71f272236b5b2d2691cf915a7283cc0ceda56357b61daa538/comm-0.2.2-py3-none-any.whl", hash = "sha256:e6fb86cb70ff661ee8c9c14e7d36d6de3b4066f1441be4063df9c5009f0a64d3", size = 7180, upload-time = "2024-03-12T16:53:39.226Z" }, + { url = "https://files.pythonhosted.org/packages/60/97/891a0971e1e4a8c5d2b20bbe0e524dc04548d2307fee33cdeba148fd4fc7/comm-0.2.3-py3-none-any.whl", hash = "sha256:c615d91d75f7f04f095b30d1c1711babd43bdc6419c1be9886a85f2f4e489417", size = 7294, upload-time = "2025-07-25T14:02:02.896Z" }, +] + +[[package]] +name = "contourpy" +version = "1.3.2" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'linux'", + "python_full_version < '3.11' and sys_platform != 'linux'", +] +dependencies = [ + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/66/54/eb9bfc647b19f2009dd5c7f5ec51c4e6ca831725f1aea7a993034f483147/contourpy-1.3.2.tar.gz", hash = "sha256:b6945942715a034c671b7fc54f9588126b0b8bf23db2696e3ca8328f3ff0ab54", size = 13466130, upload-time = "2025-04-15T17:47:53.79Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/12/a3/da4153ec8fe25d263aa48c1a4cbde7f49b59af86f0b6f7862788c60da737/contourpy-1.3.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:ba38e3f9f330af820c4b27ceb4b9c7feee5fe0493ea53a8720f4792667465934", size = 268551, upload-time = "2025-04-15T17:34:46.581Z" }, + { url = "https://files.pythonhosted.org/packages/2f/6c/330de89ae1087eb622bfca0177d32a7ece50c3ef07b28002de4757d9d875/contourpy-1.3.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:dc41ba0714aa2968d1f8674ec97504a8f7e334f48eeacebcaa6256213acb0989", size = 253399, upload-time = "2025-04-15T17:34:51.427Z" }, + { url = "https://files.pythonhosted.org/packages/c1/bd/20c6726b1b7f81a8bee5271bed5c165f0a8e1f572578a9d27e2ccb763cb2/contourpy-1.3.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9be002b31c558d1ddf1b9b415b162c603405414bacd6932d031c5b5a8b757f0d", size = 312061, upload-time = "2025-04-15T17:34:55.961Z" }, + { url = "https://files.pythonhosted.org/packages/22/fc/a9665c88f8a2473f823cf1ec601de9e5375050f1958cbb356cdf06ef1ab6/contourpy-1.3.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8d2e74acbcba3bfdb6d9d8384cdc4f9260cae86ed9beee8bd5f54fee49a430b9", size = 351956, upload-time = "2025-04-15T17:35:00.992Z" }, + { url = "https://files.pythonhosted.org/packages/25/eb/9f0a0238f305ad8fb7ef42481020d6e20cf15e46be99a1fcf939546a177e/contourpy-1.3.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e259bced5549ac64410162adc973c5e2fb77f04df4a439d00b478e57a0e65512", size = 320872, upload-time = "2025-04-15T17:35:06.177Z" }, + { url = "https://files.pythonhosted.org/packages/32/5c/1ee32d1c7956923202f00cf8d2a14a62ed7517bdc0ee1e55301227fc273c/contourpy-1.3.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ad687a04bc802cbe8b9c399c07162a3c35e227e2daccf1668eb1f278cb698631", size = 325027, upload-time = "2025-04-15T17:35:11.244Z" }, + { url = "https://files.pythonhosted.org/packages/83/bf/9baed89785ba743ef329c2b07fd0611d12bfecbedbdd3eeecf929d8d3b52/contourpy-1.3.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:cdd22595308f53ef2f891040ab2b93d79192513ffccbd7fe19be7aa773a5e09f", size = 1306641, upload-time = "2025-04-15T17:35:26.701Z" }, + { url = "https://files.pythonhosted.org/packages/d4/cc/74e5e83d1e35de2d28bd97033426b450bc4fd96e092a1f7a63dc7369b55d/contourpy-1.3.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:b4f54d6a2defe9f257327b0f243612dd051cc43825587520b1bf74a31e2f6ef2", size = 1374075, upload-time = "2025-04-15T17:35:43.204Z" }, + { url = "https://files.pythonhosted.org/packages/0c/42/17f3b798fd5e033b46a16f8d9fcb39f1aba051307f5ebf441bad1ecf78f8/contourpy-1.3.2-cp310-cp310-win32.whl", hash = "sha256:f939a054192ddc596e031e50bb13b657ce318cf13d264f095ce9db7dc6ae81c0", size = 177534, upload-time = "2025-04-15T17:35:46.554Z" }, + { url = "https://files.pythonhosted.org/packages/54/ec/5162b8582f2c994721018d0c9ece9dc6ff769d298a8ac6b6a652c307e7df/contourpy-1.3.2-cp310-cp310-win_amd64.whl", hash = "sha256:c440093bbc8fc21c637c03bafcbef95ccd963bc6e0514ad887932c18ca2a759a", size = 221188, upload-time = "2025-04-15T17:35:50.064Z" }, + { url = "https://files.pythonhosted.org/packages/b3/b9/ede788a0b56fc5b071639d06c33cb893f68b1178938f3425debebe2dab78/contourpy-1.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:6a37a2fb93d4df3fc4c0e363ea4d16f83195fc09c891bc8ce072b9d084853445", size = 269636, upload-time = "2025-04-15T17:35:54.473Z" }, + { url = "https://files.pythonhosted.org/packages/e6/75/3469f011d64b8bbfa04f709bfc23e1dd71be54d05b1b083be9f5b22750d1/contourpy-1.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:b7cd50c38f500bbcc9b6a46643a40e0913673f869315d8e70de0438817cb7773", size = 254636, upload-time = "2025-04-15T17:35:58.283Z" }, + { url = "https://files.pythonhosted.org/packages/8d/2f/95adb8dae08ce0ebca4fd8e7ad653159565d9739128b2d5977806656fcd2/contourpy-1.3.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d6658ccc7251a4433eebd89ed2672c2ed96fba367fd25ca9512aa92a4b46c4f1", size = 313053, upload-time = "2025-04-15T17:36:03.235Z" }, + { url = "https://files.pythonhosted.org/packages/c3/a6/8ccf97a50f31adfa36917707fe39c9a0cbc24b3bbb58185577f119736cc9/contourpy-1.3.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:70771a461aaeb335df14deb6c97439973d253ae70660ca085eec25241137ef43", size = 352985, upload-time = "2025-04-15T17:36:08.275Z" }, + { url = "https://files.pythonhosted.org/packages/1d/b6/7925ab9b77386143f39d9c3243fdd101621b4532eb126743201160ffa7e6/contourpy-1.3.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:65a887a6e8c4cd0897507d814b14c54a8c2e2aa4ac9f7686292f9769fcf9a6ab", size = 323750, upload-time = "2025-04-15T17:36:13.29Z" }, + { url = "https://files.pythonhosted.org/packages/c2/f3/20c5d1ef4f4748e52d60771b8560cf00b69d5c6368b5c2e9311bcfa2a08b/contourpy-1.3.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3859783aefa2b8355697f16642695a5b9792e7a46ab86da1118a4a23a51a33d7", size = 326246, upload-time = "2025-04-15T17:36:18.329Z" }, + { url = "https://files.pythonhosted.org/packages/8c/e5/9dae809e7e0b2d9d70c52b3d24cba134dd3dad979eb3e5e71f5df22ed1f5/contourpy-1.3.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:eab0f6db315fa4d70f1d8ab514e527f0366ec021ff853d7ed6a2d33605cf4b83", size = 1308728, upload-time = "2025-04-15T17:36:33.878Z" }, + { url = "https://files.pythonhosted.org/packages/e2/4a/0058ba34aeea35c0b442ae61a4f4d4ca84d6df8f91309bc2d43bb8dd248f/contourpy-1.3.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d91a3ccc7fea94ca0acab82ceb77f396d50a1f67412efe4c526f5d20264e6ecd", size = 1375762, upload-time = "2025-04-15T17:36:51.295Z" }, + { url = "https://files.pythonhosted.org/packages/09/33/7174bdfc8b7767ef2c08ed81244762d93d5c579336fc0b51ca57b33d1b80/contourpy-1.3.2-cp311-cp311-win32.whl", hash = "sha256:1c48188778d4d2f3d48e4643fb15d8608b1d01e4b4d6b0548d9b336c28fc9b6f", size = 178196, upload-time = "2025-04-15T17:36:55.002Z" }, + { url = "https://files.pythonhosted.org/packages/5e/fe/4029038b4e1c4485cef18e480b0e2cd2d755448bb071eb9977caac80b77b/contourpy-1.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:5ebac872ba09cb8f2131c46b8739a7ff71de28a24c869bcad554477eb089a878", size = 222017, upload-time = "2025-04-15T17:36:58.576Z" }, + { url = "https://files.pythonhosted.org/packages/34/f7/44785876384eff370c251d58fd65f6ad7f39adce4a093c934d4a67a7c6b6/contourpy-1.3.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:4caf2bcd2969402bf77edc4cb6034c7dd7c0803213b3523f111eb7460a51b8d2", size = 271580, upload-time = "2025-04-15T17:37:03.105Z" }, + { url = "https://files.pythonhosted.org/packages/93/3b/0004767622a9826ea3d95f0e9d98cd8729015768075d61f9fea8eeca42a8/contourpy-1.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:82199cb78276249796419fe36b7386bd8d2cc3f28b3bc19fe2454fe2e26c4c15", size = 255530, upload-time = "2025-04-15T17:37:07.026Z" }, + { url = "https://files.pythonhosted.org/packages/e7/bb/7bd49e1f4fa805772d9fd130e0d375554ebc771ed7172f48dfcd4ca61549/contourpy-1.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:106fab697af11456fcba3e352ad50effe493a90f893fca6c2ca5c033820cea92", size = 307688, upload-time = "2025-04-15T17:37:11.481Z" }, + { url = "https://files.pythonhosted.org/packages/fc/97/e1d5dbbfa170725ef78357a9a0edc996b09ae4af170927ba8ce977e60a5f/contourpy-1.3.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d14f12932a8d620e307f715857107b1d1845cc44fdb5da2bc8e850f5ceba9f87", size = 347331, upload-time = "2025-04-15T17:37:18.212Z" }, + { url = "https://files.pythonhosted.org/packages/6f/66/e69e6e904f5ecf6901be3dd16e7e54d41b6ec6ae3405a535286d4418ffb4/contourpy-1.3.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:532fd26e715560721bb0d5fc7610fce279b3699b018600ab999d1be895b09415", size = 318963, upload-time = "2025-04-15T17:37:22.76Z" }, + { url = "https://files.pythonhosted.org/packages/a8/32/b8a1c8965e4f72482ff2d1ac2cd670ce0b542f203c8e1d34e7c3e6925da7/contourpy-1.3.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f26b383144cf2d2c29f01a1e8170f50dacf0eac02d64139dcd709a8ac4eb3cfe", size = 323681, upload-time = "2025-04-15T17:37:33.001Z" }, + { url = "https://files.pythonhosted.org/packages/30/c6/12a7e6811d08757c7162a541ca4c5c6a34c0f4e98ef2b338791093518e40/contourpy-1.3.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:c49f73e61f1f774650a55d221803b101d966ca0c5a2d6d5e4320ec3997489441", size = 1308674, upload-time = "2025-04-15T17:37:48.64Z" }, + { url = "https://files.pythonhosted.org/packages/2a/8a/bebe5a3f68b484d3a2b8ffaf84704b3e343ef1addea528132ef148e22b3b/contourpy-1.3.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:3d80b2c0300583228ac98d0a927a1ba6a2ba6b8a742463c564f1d419ee5b211e", size = 1380480, upload-time = "2025-04-15T17:38:06.7Z" }, + { url = "https://files.pythonhosted.org/packages/34/db/fcd325f19b5978fb509a7d55e06d99f5f856294c1991097534360b307cf1/contourpy-1.3.2-cp312-cp312-win32.whl", hash = "sha256:90df94c89a91b7362e1142cbee7568f86514412ab8a2c0d0fca72d7e91b62912", size = 178489, upload-time = "2025-04-15T17:38:10.338Z" }, + { url = "https://files.pythonhosted.org/packages/01/c8/fadd0b92ffa7b5eb5949bf340a63a4a496a6930a6c37a7ba0f12acb076d6/contourpy-1.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:8c942a01d9163e2e5cfb05cb66110121b8d07ad438a17f9e766317bcb62abf73", size = 223042, upload-time = "2025-04-15T17:38:14.239Z" }, + { url = "https://files.pythonhosted.org/packages/2e/61/5673f7e364b31e4e7ef6f61a4b5121c5f170f941895912f773d95270f3a2/contourpy-1.3.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:de39db2604ae755316cb5967728f4bea92685884b1e767b7c24e983ef5f771cb", size = 271630, upload-time = "2025-04-15T17:38:19.142Z" }, + { url = "https://files.pythonhosted.org/packages/ff/66/a40badddd1223822c95798c55292844b7e871e50f6bfd9f158cb25e0bd39/contourpy-1.3.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3f9e896f447c5c8618f1edb2bafa9a4030f22a575ec418ad70611450720b5b08", size = 255670, upload-time = "2025-04-15T17:38:23.688Z" }, + { url = "https://files.pythonhosted.org/packages/1e/c7/cf9fdee8200805c9bc3b148f49cb9482a4e3ea2719e772602a425c9b09f8/contourpy-1.3.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:71e2bd4a1c4188f5c2b8d274da78faab884b59df20df63c34f74aa1813c4427c", size = 306694, upload-time = "2025-04-15T17:38:28.238Z" }, + { url = "https://files.pythonhosted.org/packages/dd/e7/ccb9bec80e1ba121efbffad7f38021021cda5be87532ec16fd96533bb2e0/contourpy-1.3.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:de425af81b6cea33101ae95ece1f696af39446db9682a0b56daaa48cfc29f38f", size = 345986, upload-time = "2025-04-15T17:38:33.502Z" }, + { url = "https://files.pythonhosted.org/packages/dc/49/ca13bb2da90391fa4219fdb23b078d6065ada886658ac7818e5441448b78/contourpy-1.3.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:977e98a0e0480d3fe292246417239d2d45435904afd6d7332d8455981c408b85", size = 318060, upload-time = "2025-04-15T17:38:38.672Z" }, + { url = "https://files.pythonhosted.org/packages/c8/65/5245ce8c548a8422236c13ffcdcdada6a2a812c361e9e0c70548bb40b661/contourpy-1.3.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:434f0adf84911c924519d2b08fc10491dd282b20bdd3fa8f60fd816ea0b48841", size = 322747, upload-time = "2025-04-15T17:38:43.712Z" }, + { url = "https://files.pythonhosted.org/packages/72/30/669b8eb48e0a01c660ead3752a25b44fdb2e5ebc13a55782f639170772f9/contourpy-1.3.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c66c4906cdbc50e9cba65978823e6e00b45682eb09adbb78c9775b74eb222422", size = 1308895, upload-time = "2025-04-15T17:39:00.224Z" }, + { url = "https://files.pythonhosted.org/packages/05/5a/b569f4250decee6e8d54498be7bdf29021a4c256e77fe8138c8319ef8eb3/contourpy-1.3.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8b7fc0cd78ba2f4695fd0a6ad81a19e7e3ab825c31b577f384aa9d7817dc3bef", size = 1379098, upload-time = "2025-04-15T17:43:29.649Z" }, + { url = "https://files.pythonhosted.org/packages/19/ba/b227c3886d120e60e41b28740ac3617b2f2b971b9f601c835661194579f1/contourpy-1.3.2-cp313-cp313-win32.whl", hash = "sha256:15ce6ab60957ca74cff444fe66d9045c1fd3e92c8936894ebd1f3eef2fff075f", size = 178535, upload-time = "2025-04-15T17:44:44.532Z" }, + { url = "https://files.pythonhosted.org/packages/12/6e/2fed56cd47ca739b43e892707ae9a13790a486a3173be063681ca67d2262/contourpy-1.3.2-cp313-cp313-win_amd64.whl", hash = "sha256:e1578f7eafce927b168752ed7e22646dad6cd9bca673c60bff55889fa236ebf9", size = 223096, upload-time = "2025-04-15T17:44:48.194Z" }, + { url = "https://files.pythonhosted.org/packages/54/4c/e76fe2a03014a7c767d79ea35c86a747e9325537a8b7627e0e5b3ba266b4/contourpy-1.3.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:0475b1f6604896bc7c53bb070e355e9321e1bc0d381735421a2d2068ec56531f", size = 285090, upload-time = "2025-04-15T17:43:34.084Z" }, + { url = "https://files.pythonhosted.org/packages/7b/e2/5aba47debd55d668e00baf9651b721e7733975dc9fc27264a62b0dd26eb8/contourpy-1.3.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:c85bb486e9be652314bb5b9e2e3b0d1b2e643d5eec4992c0fbe8ac71775da739", size = 268643, upload-time = "2025-04-15T17:43:38.626Z" }, + { url = "https://files.pythonhosted.org/packages/a1/37/cd45f1f051fe6230f751cc5cdd2728bb3a203f5619510ef11e732109593c/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:745b57db7758f3ffc05a10254edd3182a2a83402a89c00957a8e8a22f5582823", size = 310443, upload-time = "2025-04-15T17:43:44.522Z" }, + { url = "https://files.pythonhosted.org/packages/8b/a2/36ea6140c306c9ff6dd38e3bcec80b3b018474ef4d17eb68ceecd26675f4/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:970e9173dbd7eba9b4e01aab19215a48ee5dd3f43cef736eebde064a171f89a5", size = 349865, upload-time = "2025-04-15T17:43:49.545Z" }, + { url = "https://files.pythonhosted.org/packages/95/b7/2fc76bc539693180488f7b6cc518da7acbbb9e3b931fd9280504128bf956/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c6c4639a9c22230276b7bffb6a850dfc8258a2521305e1faefe804d006b2e532", size = 321162, upload-time = "2025-04-15T17:43:54.203Z" }, + { url = "https://files.pythonhosted.org/packages/f4/10/76d4f778458b0aa83f96e59d65ece72a060bacb20cfbee46cf6cd5ceba41/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cc829960f34ba36aad4302e78eabf3ef16a3a100863f0d4eeddf30e8a485a03b", size = 327355, upload-time = "2025-04-15T17:44:01.025Z" }, + { url = "https://files.pythonhosted.org/packages/43/a3/10cf483ea683f9f8ab096c24bad3cce20e0d1dd9a4baa0e2093c1c962d9d/contourpy-1.3.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:d32530b534e986374fc19eaa77fcb87e8a99e5431499949b828312bdcd20ac52", size = 1307935, upload-time = "2025-04-15T17:44:17.322Z" }, + { url = "https://files.pythonhosted.org/packages/78/73/69dd9a024444489e22d86108e7b913f3528f56cfc312b5c5727a44188471/contourpy-1.3.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:e298e7e70cf4eb179cc1077be1c725b5fd131ebc81181bf0c03525c8abc297fd", size = 1372168, upload-time = "2025-04-15T17:44:33.43Z" }, + { url = "https://files.pythonhosted.org/packages/0f/1b/96d586ccf1b1a9d2004dd519b25fbf104a11589abfd05484ff12199cca21/contourpy-1.3.2-cp313-cp313t-win32.whl", hash = "sha256:d0e589ae0d55204991450bb5c23f571c64fe43adaa53f93fc902a84c96f52fe1", size = 189550, upload-time = "2025-04-15T17:44:37.092Z" }, + { url = "https://files.pythonhosted.org/packages/b0/e6/6000d0094e8a5e32ad62591c8609e269febb6e4db83a1c75ff8868b42731/contourpy-1.3.2-cp313-cp313t-win_amd64.whl", hash = "sha256:78e9253c3de756b3f6a5174d024c4835acd59eb3f8e2ca13e775dbffe1558f69", size = 238214, upload-time = "2025-04-15T17:44:40.827Z" }, + { url = "https://files.pythonhosted.org/packages/33/05/b26e3c6ecc05f349ee0013f0bb850a761016d89cec528a98193a48c34033/contourpy-1.3.2-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:fd93cc7f3139b6dd7aab2f26a90dde0aa9fc264dbf70f6740d498a70b860b82c", size = 265681, upload-time = "2025-04-15T17:44:59.314Z" }, + { url = "https://files.pythonhosted.org/packages/2b/25/ac07d6ad12affa7d1ffed11b77417d0a6308170f44ff20fa1d5aa6333f03/contourpy-1.3.2-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:107ba8a6a7eec58bb475329e6d3b95deba9440667c4d62b9b6063942b61d7f16", size = 315101, upload-time = "2025-04-15T17:45:04.165Z" }, + { url = "https://files.pythonhosted.org/packages/8f/4d/5bb3192bbe9d3f27e3061a6a8e7733c9120e203cb8515767d30973f71030/contourpy-1.3.2-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:ded1706ed0c1049224531b81128efbd5084598f18d8a2d9efae833edbd2b40ad", size = 220599, upload-time = "2025-04-15T17:45:08.456Z" }, + { url = "https://files.pythonhosted.org/packages/ff/c0/91f1215d0d9f9f343e4773ba6c9b89e8c0cc7a64a6263f21139da639d848/contourpy-1.3.2-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:5f5964cdad279256c084b69c3f412b7801e15356b16efa9d78aa974041903da0", size = 266807, upload-time = "2025-04-15T17:45:15.535Z" }, + { url = "https://files.pythonhosted.org/packages/d4/79/6be7e90c955c0487e7712660d6cead01fa17bff98e0ea275737cc2bc8e71/contourpy-1.3.2-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:49b65a95d642d4efa8f64ba12558fcb83407e58a2dfba9d796d77b63ccfcaff5", size = 318729, upload-time = "2025-04-15T17:45:20.166Z" }, + { url = "https://files.pythonhosted.org/packages/87/68/7f46fb537958e87427d98a4074bcde4b67a70b04900cfc5ce29bc2f556c1/contourpy-1.3.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:8c5acb8dddb0752bf252e01a3035b21443158910ac16a3b0d20e7fed7d534ce5", size = 221791, upload-time = "2025-04-15T17:45:24.794Z" }, ] [[package]] name = "contourpy" -version = "1.3.1" -source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "numpy" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/25/c2/fc7193cc5383637ff390a712e88e4ded0452c9fbcf84abe3de5ea3df1866/contourpy-1.3.1.tar.gz", hash = "sha256:dfd97abd83335045a913e3bcc4a09c0ceadbe66580cf573fe961f4a825efa699", size = 13465753, upload-time = "2024-11-12T11:00:59.118Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/b2/a3/80937fe3efe0edacf67c9a20b955139a1a622730042c1ea991956f2704ad/contourpy-1.3.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:a045f341a77b77e1c5de31e74e966537bba9f3c4099b35bf4c2e3939dd54cdab", size = 268466, upload-time = "2024-11-12T10:52:03.706Z" }, - { url = "https://files.pythonhosted.org/packages/82/1d/e3eaebb4aa2d7311528c048350ca8e99cdacfafd99da87bc0a5f8d81f2c2/contourpy-1.3.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:500360b77259914f7805af7462e41f9cb7ca92ad38e9f94d6c8641b089338124", size = 253314, upload-time = "2024-11-12T10:52:08.721Z" }, - { url = "https://files.pythonhosted.org/packages/de/f3/d796b22d1a2b587acc8100ba8c07fb7b5e17fde265a7bb05ab967f4c935a/contourpy-1.3.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b2f926efda994cdf3c8d3fdb40b9962f86edbc4457e739277b961eced3d0b4c1", size = 312003, upload-time = "2024-11-12T10:52:13.868Z" }, - { url = "https://files.pythonhosted.org/packages/bf/f5/0e67902bc4394daee8daa39c81d4f00b50e063ee1a46cb3938cc65585d36/contourpy-1.3.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:adce39d67c0edf383647a3a007de0a45fd1b08dedaa5318404f1a73059c2512b", size = 351896, upload-time = "2024-11-12T10:52:19.513Z" }, - { url = "https://files.pythonhosted.org/packages/1f/d6/e766395723f6256d45d6e67c13bb638dd1fa9dc10ef912dc7dd3dcfc19de/contourpy-1.3.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:abbb49fb7dac584e5abc6636b7b2a7227111c4f771005853e7d25176daaf8453", size = 320814, upload-time = "2024-11-12T10:52:25.053Z" }, - { url = "https://files.pythonhosted.org/packages/a9/57/86c500d63b3e26e5b73a28b8291a67c5608d4aa87ebd17bd15bb33c178bc/contourpy-1.3.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a0cffcbede75c059f535725c1680dfb17b6ba8753f0c74b14e6a9c68c29d7ea3", size = 324969, upload-time = "2024-11-12T10:52:30.731Z" }, - { url = "https://files.pythonhosted.org/packages/b8/62/bb146d1289d6b3450bccc4642e7f4413b92ebffd9bf2e91b0404323704a7/contourpy-1.3.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:ab29962927945d89d9b293eabd0d59aea28d887d4f3be6c22deaefbb938a7277", size = 1265162, upload-time = "2024-11-12T10:52:46.26Z" }, - { url = "https://files.pythonhosted.org/packages/18/04/9f7d132ce49a212c8e767042cc80ae390f728060d2eea47058f55b9eff1c/contourpy-1.3.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:974d8145f8ca354498005b5b981165b74a195abfae9a8129df3e56771961d595", size = 1324328, upload-time = "2024-11-12T10:53:03.081Z" }, - { url = "https://files.pythonhosted.org/packages/46/23/196813901be3f97c83ababdab1382e13e0edc0bb4e7b49a7bff15fcf754e/contourpy-1.3.1-cp310-cp310-win32.whl", hash = "sha256:ac4578ac281983f63b400f7fe6c101bedc10651650eef012be1ccffcbacf3697", size = 173861, upload-time = "2024-11-12T10:53:06.283Z" }, - { url = "https://files.pythonhosted.org/packages/e0/82/c372be3fc000a3b2005061ca623a0d1ecd2eaafb10d9e883a2fc8566e951/contourpy-1.3.1-cp310-cp310-win_amd64.whl", hash = "sha256:174e758c66bbc1c8576992cec9599ce8b6672b741b5d336b5c74e35ac382b18e", size = 218566, upload-time = "2024-11-12T10:53:09.798Z" }, - { url = "https://files.pythonhosted.org/packages/12/bb/11250d2906ee2e8b466b5f93e6b19d525f3e0254ac8b445b56e618527718/contourpy-1.3.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:3e8b974d8db2c5610fb4e76307e265de0edb655ae8169e8b21f41807ccbeec4b", size = 269555, upload-time = "2024-11-12T10:53:14.707Z" }, - { url = "https://files.pythonhosted.org/packages/67/71/1e6e95aee21a500415f5d2dbf037bf4567529b6a4e986594d7026ec5ae90/contourpy-1.3.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:20914c8c973f41456337652a6eeca26d2148aa96dd7ac323b74516988bea89fc", size = 254549, upload-time = "2024-11-12T10:53:19.42Z" }, - { url = "https://files.pythonhosted.org/packages/31/2c/b88986e8d79ac45efe9d8801ae341525f38e087449b6c2f2e6050468a42c/contourpy-1.3.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:19d40d37c1c3a4961b4619dd9d77b12124a453cc3d02bb31a07d58ef684d3d86", size = 313000, upload-time = "2024-11-12T10:53:23.944Z" }, - { url = "https://files.pythonhosted.org/packages/c4/18/65280989b151fcf33a8352f992eff71e61b968bef7432fbfde3a364f0730/contourpy-1.3.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:113231fe3825ebf6f15eaa8bc1f5b0ddc19d42b733345eae0934cb291beb88b6", size = 352925, upload-time = "2024-11-12T10:53:29.719Z" }, - { url = "https://files.pythonhosted.org/packages/f5/c7/5fd0146c93220dbfe1a2e0f98969293b86ca9bc041d6c90c0e065f4619ad/contourpy-1.3.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4dbbc03a40f916a8420e420d63e96a1258d3d1b58cbdfd8d1f07b49fcbd38e85", size = 323693, upload-time = "2024-11-12T10:53:35.046Z" }, - { url = "https://files.pythonhosted.org/packages/85/fc/7fa5d17daf77306840a4e84668a48ddff09e6bc09ba4e37e85ffc8e4faa3/contourpy-1.3.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3a04ecd68acbd77fa2d39723ceca4c3197cb2969633836ced1bea14e219d077c", size = 326184, upload-time = "2024-11-12T10:53:40.261Z" }, - { url = "https://files.pythonhosted.org/packages/ef/e7/104065c8270c7397c9571620d3ab880558957216f2b5ebb7e040f85eeb22/contourpy-1.3.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c414fc1ed8ee1dbd5da626cf3710c6013d3d27456651d156711fa24f24bd1291", size = 1268031, upload-time = "2024-11-12T10:53:55.876Z" }, - { url = "https://files.pythonhosted.org/packages/e2/4a/c788d0bdbf32c8113c2354493ed291f924d4793c4a2e85b69e737a21a658/contourpy-1.3.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:31c1b55c1f34f80557d3830d3dd93ba722ce7e33a0b472cba0ec3b6535684d8f", size = 1325995, upload-time = "2024-11-12T10:54:11.572Z" }, - { url = "https://files.pythonhosted.org/packages/a6/e6/a2f351a90d955f8b0564caf1ebe4b1451a3f01f83e5e3a414055a5b8bccb/contourpy-1.3.1-cp311-cp311-win32.whl", hash = "sha256:f611e628ef06670df83fce17805c344710ca5cde01edfdc72751311da8585375", size = 174396, upload-time = "2024-11-12T10:54:15.358Z" }, - { url = "https://files.pythonhosted.org/packages/a8/7e/cd93cab453720a5d6cb75588cc17dcdc08fc3484b9de98b885924ff61900/contourpy-1.3.1-cp311-cp311-win_amd64.whl", hash = "sha256:b2bdca22a27e35f16794cf585832e542123296b4687f9fd96822db6bae17bfc9", size = 219787, upload-time = "2024-11-12T10:54:18.836Z" }, - { url = "https://files.pythonhosted.org/packages/37/6b/175f60227d3e7f5f1549fcb374592be311293132207e451c3d7c654c25fb/contourpy-1.3.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:0ffa84be8e0bd33410b17189f7164c3589c229ce5db85798076a3fa136d0e509", size = 271494, upload-time = "2024-11-12T10:54:23.6Z" }, - { url = "https://files.pythonhosted.org/packages/6b/6a/7833cfae2c1e63d1d8875a50fd23371394f540ce809d7383550681a1fa64/contourpy-1.3.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:805617228ba7e2cbbfb6c503858e626ab528ac2a32a04a2fe88ffaf6b02c32bc", size = 255444, upload-time = "2024-11-12T10:54:28.267Z" }, - { url = "https://files.pythonhosted.org/packages/7f/b3/7859efce66eaca5c14ba7619791b084ed02d868d76b928ff56890d2d059d/contourpy-1.3.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ade08d343436a94e633db932e7e8407fe7de8083967962b46bdfc1b0ced39454", size = 307628, upload-time = "2024-11-12T10:54:33.418Z" }, - { url = "https://files.pythonhosted.org/packages/48/b2/011415f5e3f0a50b1e285a0bf78eb5d92a4df000553570f0851b6e309076/contourpy-1.3.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:47734d7073fb4590b4a40122b35917cd77be5722d80683b249dac1de266aac80", size = 347271, upload-time = "2024-11-12T10:54:38.816Z" }, - { url = "https://files.pythonhosted.org/packages/84/7d/ef19b1db0f45b151ac78c65127235239a8cf21a59d1ce8507ce03e89a30b/contourpy-1.3.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2ba94a401342fc0f8b948e57d977557fbf4d515f03c67682dd5c6191cb2d16ec", size = 318906, upload-time = "2024-11-12T10:54:44.132Z" }, - { url = "https://files.pythonhosted.org/packages/ba/99/6794142b90b853a9155316c8f470d2e4821fe6f086b03e372aca848227dd/contourpy-1.3.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:efa874e87e4a647fd2e4f514d5e91c7d493697127beb95e77d2f7561f6905bd9", size = 323622, upload-time = "2024-11-12T10:54:48.788Z" }, - { url = "https://files.pythonhosted.org/packages/3c/0f/37d2c84a900cd8eb54e105f4fa9aebd275e14e266736778bb5dccbf3bbbb/contourpy-1.3.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:1bf98051f1045b15c87868dbaea84f92408337d4f81d0e449ee41920ea121d3b", size = 1266699, upload-time = "2024-11-12T10:55:04.016Z" }, - { url = "https://files.pythonhosted.org/packages/3a/8a/deb5e11dc7d9cc8f0f9c8b29d4f062203f3af230ba83c30a6b161a6effc9/contourpy-1.3.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:61332c87493b00091423e747ea78200659dc09bdf7fd69edd5e98cef5d3e9a8d", size = 1326395, upload-time = "2024-11-12T10:55:20.547Z" }, - { url = "https://files.pythonhosted.org/packages/1a/35/7e267ae7c13aaf12322ccc493531f1e7f2eb8fba2927b9d7a05ff615df7a/contourpy-1.3.1-cp312-cp312-win32.whl", hash = "sha256:e914a8cb05ce5c809dd0fe350cfbb4e881bde5e2a38dc04e3afe1b3e58bd158e", size = 175354, upload-time = "2024-11-12T10:55:24.377Z" }, - { url = "https://files.pythonhosted.org/packages/a1/35/c2de8823211d07e8a79ab018ef03960716c5dff6f4d5bff5af87fd682992/contourpy-1.3.1-cp312-cp312-win_amd64.whl", hash = "sha256:08d9d449a61cf53033612cb368f3a1b26cd7835d9b8cd326647efe43bca7568d", size = 220971, upload-time = "2024-11-12T10:55:27.971Z" }, - { url = "https://files.pythonhosted.org/packages/9a/e7/de62050dce687c5e96f946a93546910bc67e483fe05324439e329ff36105/contourpy-1.3.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:a761d9ccfc5e2ecd1bf05534eda382aa14c3e4f9205ba5b1684ecfe400716ef2", size = 271548, upload-time = "2024-11-12T10:55:32.228Z" }, - { url = "https://files.pythonhosted.org/packages/78/4d/c2a09ae014ae984c6bdd29c11e74d3121b25eaa117eca0bb76340efd7e1c/contourpy-1.3.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:523a8ee12edfa36f6d2a49407f705a6ef4c5098de4f498619787e272de93f2d5", size = 255576, upload-time = "2024-11-12T10:55:36.246Z" }, - { url = "https://files.pythonhosted.org/packages/ab/8a/915380ee96a5638bda80cd061ccb8e666bfdccea38d5741cb69e6dbd61fc/contourpy-1.3.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ece6df05e2c41bd46776fbc712e0996f7c94e0d0543af1656956d150c4ca7c81", size = 306635, upload-time = "2024-11-12T10:55:41.904Z" }, - { url = "https://files.pythonhosted.org/packages/29/5c/c83ce09375428298acd4e6582aeb68b1e0d1447f877fa993d9bf6cd3b0a0/contourpy-1.3.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:573abb30e0e05bf31ed067d2f82500ecfdaec15627a59d63ea2d95714790f5c2", size = 345925, upload-time = "2024-11-12T10:55:47.206Z" }, - { url = "https://files.pythonhosted.org/packages/29/63/5b52f4a15e80c66c8078a641a3bfacd6e07106835682454647aca1afc852/contourpy-1.3.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:a9fa36448e6a3a1a9a2ba23c02012c43ed88905ec80163f2ffe2421c7192a5d7", size = 318000, upload-time = "2024-11-12T10:55:52.264Z" }, - { url = "https://files.pythonhosted.org/packages/9a/e2/30ca086c692691129849198659bf0556d72a757fe2769eb9620a27169296/contourpy-1.3.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3ea9924d28fc5586bf0b42d15f590b10c224117e74409dd7a0be3b62b74a501c", size = 322689, upload-time = "2024-11-12T10:55:57.858Z" }, - { url = "https://files.pythonhosted.org/packages/6b/77/f37812ef700f1f185d348394debf33f22d531e714cf6a35d13d68a7003c7/contourpy-1.3.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:5b75aa69cb4d6f137b36f7eb2ace9280cfb60c55dc5f61c731fdf6f037f958a3", size = 1268413, upload-time = "2024-11-12T10:56:13.328Z" }, - { url = "https://files.pythonhosted.org/packages/3f/6d/ce84e79cdd128542ebeb268f84abb4b093af78e7f8ec504676673d2675bc/contourpy-1.3.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:041b640d4ec01922083645a94bb3b2e777e6b626788f4095cf21abbe266413c1", size = 1326530, upload-time = "2024-11-12T10:56:30.07Z" }, - { url = "https://files.pythonhosted.org/packages/72/22/8282f4eae20c73c89bee7a82a19c4e27af9b57bb602ecaa00713d5bdb54d/contourpy-1.3.1-cp313-cp313-win32.whl", hash = "sha256:36987a15e8ace5f58d4d5da9dca82d498c2bbb28dff6e5d04fbfcc35a9cb3a82", size = 175315, upload-time = "2024-11-12T10:57:42.804Z" }, - { url = "https://files.pythonhosted.org/packages/e3/d5/28bca491f65312b438fbf076589dcde7f6f966b196d900777f5811b9c4e2/contourpy-1.3.1-cp313-cp313-win_amd64.whl", hash = "sha256:a7895f46d47671fa7ceec40f31fae721da51ad34bdca0bee83e38870b1f47ffd", size = 220987, upload-time = "2024-11-12T10:57:46.365Z" }, - { url = "https://files.pythonhosted.org/packages/2f/24/a4b285d6adaaf9746e4700932f579f1a7b6f9681109f694cfa233ae75c4e/contourpy-1.3.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:9ddeb796389dadcd884c7eb07bd14ef12408aaae358f0e2ae24114d797eede30", size = 285001, upload-time = "2024-11-12T10:56:34.483Z" }, - { url = "https://files.pythonhosted.org/packages/48/1d/fb49a401b5ca4f06ccf467cd6c4f1fd65767e63c21322b29b04ec40b40b9/contourpy-1.3.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:19c1555a6801c2f084c7ddc1c6e11f02eb6a6016ca1318dd5452ba3f613a1751", size = 268553, upload-time = "2024-11-12T10:56:39.167Z" }, - { url = "https://files.pythonhosted.org/packages/79/1e/4aef9470d13fd029087388fae750dccb49a50c012a6c8d1d634295caa644/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:841ad858cff65c2c04bf93875e384ccb82b654574a6d7f30453a04f04af71342", size = 310386, upload-time = "2024-11-12T10:56:44.594Z" }, - { url = "https://files.pythonhosted.org/packages/b0/34/910dc706ed70153b60392b5305c708c9810d425bde12499c9184a1100888/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4318af1c925fb9a4fb190559ef3eec206845f63e80fb603d47f2d6d67683901c", size = 349806, upload-time = "2024-11-12T10:56:49.565Z" }, - { url = "https://files.pythonhosted.org/packages/31/3c/faee6a40d66d7f2a87f7102236bf4780c57990dd7f98e5ff29881b1b1344/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:14c102b0eab282427b662cb590f2e9340a9d91a1c297f48729431f2dcd16e14f", size = 321108, upload-time = "2024-11-12T10:56:55.013Z" }, - { url = "https://files.pythonhosted.org/packages/17/69/390dc9b20dd4bb20585651d7316cc3054b7d4a7b4f8b710b2b698e08968d/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:05e806338bfeaa006acbdeba0ad681a10be63b26e1b17317bfac3c5d98f36cda", size = 327291, upload-time = "2024-11-12T10:56:59.897Z" }, - { url = "https://files.pythonhosted.org/packages/ef/74/7030b67c4e941fe1e5424a3d988080e83568030ce0355f7c9fc556455b01/contourpy-1.3.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:4d76d5993a34ef3df5181ba3c92fabb93f1eaa5729504fb03423fcd9f3177242", size = 1263752, upload-time = "2024-11-12T10:57:14.79Z" }, - { url = "https://files.pythonhosted.org/packages/f0/ed/92d86f183a8615f13f6b9cbfc5d4298a509d6ce433432e21da838b4b63f4/contourpy-1.3.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:89785bb2a1980c1bd87f0cb1517a71cde374776a5f150936b82580ae6ead44a1", size = 1318403, upload-time = "2024-11-12T10:57:31.326Z" }, - { url = "https://files.pythonhosted.org/packages/b3/0e/c8e4950c77dcfc897c71d61e56690a0a9df39543d2164040301b5df8e67b/contourpy-1.3.1-cp313-cp313t-win32.whl", hash = "sha256:8eb96e79b9f3dcadbad2a3891672f81cdcab7f95b27f28f1c67d75f045b6b4f1", size = 185117, upload-time = "2024-11-12T10:57:34.735Z" }, - { url = "https://files.pythonhosted.org/packages/c1/31/1ae946f11dfbd229222e6d6ad8e7bd1891d3d48bde5fbf7a0beb9491f8e3/contourpy-1.3.1-cp313-cp313t-win_amd64.whl", hash = "sha256:287ccc248c9e0d0566934e7d606201abd74761b5703d804ff3df8935f523d546", size = 236668, upload-time = "2024-11-12T10:57:39.061Z" }, - { url = "https://files.pythonhosted.org/packages/3e/4f/e56862e64b52b55b5ddcff4090085521fc228ceb09a88390a2b103dccd1b/contourpy-1.3.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:b457d6430833cee8e4b8e9b6f07aa1c161e5e0d52e118dc102c8f9bd7dd060d6", size = 265605, upload-time = "2024-11-12T10:57:51.188Z" }, - { url = "https://files.pythonhosted.org/packages/b0/2e/52bfeeaa4541889f23d8eadc6386b442ee2470bd3cff9baa67deb2dd5c57/contourpy-1.3.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cb76c1a154b83991a3cbbf0dfeb26ec2833ad56f95540b442c73950af2013750", size = 315040, upload-time = "2024-11-12T10:57:56.492Z" }, - { url = "https://files.pythonhosted.org/packages/52/94/86bfae441707205634d80392e873295652fc313dfd93c233c52c4dc07874/contourpy-1.3.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:44a29502ca9c7b5ba389e620d44f2fbe792b1fb5734e8b931ad307071ec58c53", size = 218221, upload-time = "2024-11-12T10:58:00.033Z" }, +version = "1.3.3" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", + "python_full_version == '3.11.*' and sys_platform == 'linux'", + "python_full_version == '3.11.*' and sys_platform != 'linux'", +] +dependencies = [ + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/58/01/1253e6698a07380cd31a736d248a3f2a50a7c88779a1813da27503cadc2a/contourpy-1.3.3.tar.gz", hash = "sha256:083e12155b210502d0bca491432bb04d56dc3432f95a979b429f2848c3dbe880", size = 13466174, upload-time = "2025-07-26T12:03:12.549Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/91/2e/c4390a31919d8a78b90e8ecf87cd4b4c4f05a5b48d05ec17db8e5404c6f4/contourpy-1.3.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:709a48ef9a690e1343202916450bc48b9e51c049b089c7f79a267b46cffcdaa1", size = 288773, upload-time = "2025-07-26T12:01:02.277Z" }, + { url = "https://files.pythonhosted.org/packages/0d/44/c4b0b6095fef4dc9c420e041799591e3b63e9619e3044f7f4f6c21c0ab24/contourpy-1.3.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:23416f38bfd74d5d28ab8429cc4d63fa67d5068bd711a85edb1c3fb0c3e2f381", size = 270149, upload-time = "2025-07-26T12:01:04.072Z" }, + { url = "https://files.pythonhosted.org/packages/30/2e/dd4ced42fefac8470661d7cb7e264808425e6c5d56d175291e93890cce09/contourpy-1.3.3-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:929ddf8c4c7f348e4c0a5a3a714b5c8542ffaa8c22954862a46ca1813b667ee7", size = 329222, upload-time = "2025-07-26T12:01:05.688Z" }, + { url = "https://files.pythonhosted.org/packages/f2/74/cc6ec2548e3d276c71389ea4802a774b7aa3558223b7bade3f25787fafc2/contourpy-1.3.3-cp311-cp311-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:9e999574eddae35f1312c2b4b717b7885d4edd6cb46700e04f7f02db454e67c1", size = 377234, upload-time = "2025-07-26T12:01:07.054Z" }, + { url = "https://files.pythonhosted.org/packages/03/b3/64ef723029f917410f75c09da54254c5f9ea90ef89b143ccadb09df14c15/contourpy-1.3.3-cp311-cp311-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:0bf67e0e3f482cb69779dd3061b534eb35ac9b17f163d851e2a547d56dba0a3a", size = 380555, upload-time = "2025-07-26T12:01:08.801Z" }, + { url = "https://files.pythonhosted.org/packages/5f/4b/6157f24ca425b89fe2eb7e7be642375711ab671135be21e6faa100f7448c/contourpy-1.3.3-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:51e79c1f7470158e838808d4a996fa9bac72c498e93d8ebe5119bc1e6becb0db", size = 355238, upload-time = "2025-07-26T12:01:10.319Z" }, + { url = "https://files.pythonhosted.org/packages/98/56/f914f0dd678480708a04cfd2206e7c382533249bc5001eb9f58aa693e200/contourpy-1.3.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:598c3aaece21c503615fd59c92a3598b428b2f01bfb4b8ca9c4edeecc2438620", size = 1326218, upload-time = "2025-07-26T12:01:12.659Z" }, + { url = "https://files.pythonhosted.org/packages/fb/d7/4a972334a0c971acd5172389671113ae82aa7527073980c38d5868ff1161/contourpy-1.3.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:322ab1c99b008dad206d406bb61d014cf0174df491ae9d9d0fac6a6fda4f977f", size = 1392867, upload-time = "2025-07-26T12:01:15.533Z" }, + { url = "https://files.pythonhosted.org/packages/75/3e/f2cc6cd56dc8cff46b1a56232eabc6feea52720083ea71ab15523daab796/contourpy-1.3.3-cp311-cp311-win32.whl", hash = "sha256:fd907ae12cd483cd83e414b12941c632a969171bf90fc937d0c9f268a31cafff", size = 183677, upload-time = "2025-07-26T12:01:17.088Z" }, + { url = "https://files.pythonhosted.org/packages/98/4b/9bd370b004b5c9d8045c6c33cf65bae018b27aca550a3f657cdc99acdbd8/contourpy-1.3.3-cp311-cp311-win_amd64.whl", hash = "sha256:3519428f6be58431c56581f1694ba8e50626f2dd550af225f82fb5f5814d2a42", size = 225234, upload-time = "2025-07-26T12:01:18.256Z" }, + { url = "https://files.pythonhosted.org/packages/d9/b6/71771e02c2e004450c12b1120a5f488cad2e4d5b590b1af8bad060360fe4/contourpy-1.3.3-cp311-cp311-win_arm64.whl", hash = "sha256:15ff10bfada4bf92ec8b31c62bf7c1834c244019b4a33095a68000d7075df470", size = 193123, upload-time = "2025-07-26T12:01:19.848Z" }, + { url = "https://files.pythonhosted.org/packages/be/45/adfee365d9ea3d853550b2e735f9d66366701c65db7855cd07621732ccfc/contourpy-1.3.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b08a32ea2f8e42cf1d4be3169a98dd4be32bafe4f22b6c4cb4ba810fa9e5d2cb", size = 293419, upload-time = "2025-07-26T12:01:21.16Z" }, + { url = "https://files.pythonhosted.org/packages/53/3e/405b59cfa13021a56bba395a6b3aca8cec012b45bf177b0eaf7a202cde2c/contourpy-1.3.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:556dba8fb6f5d8742f2923fe9457dbdd51e1049c4a43fd3986a0b14a1d815fc6", size = 273979, upload-time = "2025-07-26T12:01:22.448Z" }, + { url = "https://files.pythonhosted.org/packages/d4/1c/a12359b9b2ca3a845e8f7f9ac08bdf776114eb931392fcad91743e2ea17b/contourpy-1.3.3-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:92d9abc807cf7d0e047b95ca5d957cf4792fcd04e920ca70d48add15c1a90ea7", size = 332653, upload-time = "2025-07-26T12:01:24.155Z" }, + { url = "https://files.pythonhosted.org/packages/63/12/897aeebfb475b7748ea67b61e045accdfcf0d971f8a588b67108ed7f5512/contourpy-1.3.3-cp312-cp312-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:b2e8faa0ed68cb29af51edd8e24798bb661eac3bd9f65420c1887b6ca89987c8", size = 379536, upload-time = "2025-07-26T12:01:25.91Z" }, + { url = "https://files.pythonhosted.org/packages/43/8a/a8c584b82deb248930ce069e71576fc09bd7174bbd35183b7943fb1064fd/contourpy-1.3.3-cp312-cp312-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:626d60935cf668e70a5ce6ff184fd713e9683fb458898e4249b63be9e28286ea", size = 384397, upload-time = "2025-07-26T12:01:27.152Z" }, + { url = "https://files.pythonhosted.org/packages/cc/8f/ec6289987824b29529d0dfda0d74a07cec60e54b9c92f3c9da4c0ac732de/contourpy-1.3.3-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4d00e655fcef08aba35ec9610536bfe90267d7ab5ba944f7032549c55a146da1", size = 362601, upload-time = "2025-07-26T12:01:28.808Z" }, + { url = "https://files.pythonhosted.org/packages/05/0a/a3fe3be3ee2dceb3e615ebb4df97ae6f3828aa915d3e10549ce016302bd1/contourpy-1.3.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:451e71b5a7d597379ef572de31eeb909a87246974d960049a9848c3bc6c41bf7", size = 1331288, upload-time = "2025-07-26T12:01:31.198Z" }, + { url = "https://files.pythonhosted.org/packages/33/1d/acad9bd4e97f13f3e2b18a3977fe1b4a37ecf3d38d815333980c6c72e963/contourpy-1.3.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:459c1f020cd59fcfe6650180678a9993932d80d44ccde1fa1868977438f0b411", size = 1403386, upload-time = "2025-07-26T12:01:33.947Z" }, + { url = "https://files.pythonhosted.org/packages/cf/8f/5847f44a7fddf859704217a99a23a4f6417b10e5ab1256a179264561540e/contourpy-1.3.3-cp312-cp312-win32.whl", hash = "sha256:023b44101dfe49d7d53932be418477dba359649246075c996866106da069af69", size = 185018, upload-time = "2025-07-26T12:01:35.64Z" }, + { url = "https://files.pythonhosted.org/packages/19/e8/6026ed58a64563186a9ee3f29f41261fd1828f527dd93d33b60feca63352/contourpy-1.3.3-cp312-cp312-win_amd64.whl", hash = "sha256:8153b8bfc11e1e4d75bcb0bff1db232f9e10b274e0929de9d608027e0d34ff8b", size = 226567, upload-time = "2025-07-26T12:01:36.804Z" }, + { url = "https://files.pythonhosted.org/packages/d1/e2/f05240d2c39a1ed228d8328a78b6f44cd695f7ef47beb3e684cf93604f86/contourpy-1.3.3-cp312-cp312-win_arm64.whl", hash = "sha256:07ce5ed73ecdc4a03ffe3e1b3e3c1166db35ae7584be76f65dbbe28a7791b0cc", size = 193655, upload-time = "2025-07-26T12:01:37.999Z" }, + { url = "https://files.pythonhosted.org/packages/68/35/0167aad910bbdb9599272bd96d01a9ec6852f36b9455cf2ca67bd4cc2d23/contourpy-1.3.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:177fb367556747a686509d6fef71d221a4b198a3905fe824430e5ea0fda54eb5", size = 293257, upload-time = "2025-07-26T12:01:39.367Z" }, + { url = "https://files.pythonhosted.org/packages/96/e4/7adcd9c8362745b2210728f209bfbcf7d91ba868a2c5f40d8b58f54c509b/contourpy-1.3.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:d002b6f00d73d69333dac9d0b8d5e84d9724ff9ef044fd63c5986e62b7c9e1b1", size = 274034, upload-time = "2025-07-26T12:01:40.645Z" }, + { url = "https://files.pythonhosted.org/packages/73/23/90e31ceeed1de63058a02cb04b12f2de4b40e3bef5e082a7c18d9c8ae281/contourpy-1.3.3-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:348ac1f5d4f1d66d3322420f01d42e43122f43616e0f194fc1c9f5d830c5b286", size = 334672, upload-time = "2025-07-26T12:01:41.942Z" }, + { url = "https://files.pythonhosted.org/packages/ed/93/b43d8acbe67392e659e1d984700e79eb67e2acb2bd7f62012b583a7f1b55/contourpy-1.3.3-cp313-cp313-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:655456777ff65c2c548b7c454af9c6f33f16c8884f11083244b5819cc214f1b5", size = 381234, upload-time = "2025-07-26T12:01:43.499Z" }, + { url = "https://files.pythonhosted.org/packages/46/3b/bec82a3ea06f66711520f75a40c8fc0b113b2a75edb36aa633eb11c4f50f/contourpy-1.3.3-cp313-cp313-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:644a6853d15b2512d67881586bd03f462c7ab755db95f16f14d7e238f2852c67", size = 385169, upload-time = "2025-07-26T12:01:45.219Z" }, + { url = "https://files.pythonhosted.org/packages/4b/32/e0f13a1c5b0f8572d0ec6ae2f6c677b7991fafd95da523159c19eff0696a/contourpy-1.3.3-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4debd64f124ca62069f313a9cb86656ff087786016d76927ae2cf37846b006c9", size = 362859, upload-time = "2025-07-26T12:01:46.519Z" }, + { url = "https://files.pythonhosted.org/packages/33/71/e2a7945b7de4e58af42d708a219f3b2f4cff7386e6b6ab0a0fa0033c49a9/contourpy-1.3.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:a15459b0f4615b00bbd1e91f1b9e19b7e63aea7483d03d804186f278c0af2659", size = 1332062, upload-time = "2025-07-26T12:01:48.964Z" }, + { url = "https://files.pythonhosted.org/packages/12/fc/4e87ac754220ccc0e807284f88e943d6d43b43843614f0a8afa469801db0/contourpy-1.3.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ca0fdcd73925568ca027e0b17ab07aad764be4706d0a925b89227e447d9737b7", size = 1403932, upload-time = "2025-07-26T12:01:51.979Z" }, + { url = "https://files.pythonhosted.org/packages/a6/2e/adc197a37443f934594112222ac1aa7dc9a98faf9c3842884df9a9d8751d/contourpy-1.3.3-cp313-cp313-win32.whl", hash = "sha256:b20c7c9a3bf701366556e1b1984ed2d0cedf999903c51311417cf5f591d8c78d", size = 185024, upload-time = "2025-07-26T12:01:53.245Z" }, + { url = "https://files.pythonhosted.org/packages/18/0b/0098c214843213759692cc638fce7de5c289200a830e5035d1791d7a2338/contourpy-1.3.3-cp313-cp313-win_amd64.whl", hash = "sha256:1cadd8b8969f060ba45ed7c1b714fe69185812ab43bd6b86a9123fe8f99c3263", size = 226578, upload-time = "2025-07-26T12:01:54.422Z" }, + { url = "https://files.pythonhosted.org/packages/8a/9a/2f6024a0c5995243cd63afdeb3651c984f0d2bc727fd98066d40e141ad73/contourpy-1.3.3-cp313-cp313-win_arm64.whl", hash = "sha256:fd914713266421b7536de2bfa8181aa8c699432b6763a0ea64195ebe28bff6a9", size = 193524, upload-time = "2025-07-26T12:01:55.73Z" }, + { url = "https://files.pythonhosted.org/packages/c0/b3/f8a1a86bd3298513f500e5b1f5fd92b69896449f6cab6a146a5d52715479/contourpy-1.3.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:88df9880d507169449d434c293467418b9f6cbe82edd19284aa0409e7fdb933d", size = 306730, upload-time = "2025-07-26T12:01:57.051Z" }, + { url = "https://files.pythonhosted.org/packages/3f/11/4780db94ae62fc0c2053909b65dc3246bd7cecfc4f8a20d957ad43aa4ad8/contourpy-1.3.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:d06bb1f751ba5d417047db62bca3c8fde202b8c11fb50742ab3ab962c81e8216", size = 287897, upload-time = "2025-07-26T12:01:58.663Z" }, + { url = "https://files.pythonhosted.org/packages/ae/15/e59f5f3ffdd6f3d4daa3e47114c53daabcb18574a26c21f03dc9e4e42ff0/contourpy-1.3.3-cp313-cp313t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e4e6b05a45525357e382909a4c1600444e2a45b4795163d3b22669285591c1ae", size = 326751, upload-time = "2025-07-26T12:02:00.343Z" }, + { url = "https://files.pythonhosted.org/packages/0f/81/03b45cfad088e4770b1dcf72ea78d3802d04200009fb364d18a493857210/contourpy-1.3.3-cp313-cp313t-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:ab3074b48c4e2cf1a960e6bbeb7f04566bf36b1861d5c9d4d8ac04b82e38ba20", size = 375486, upload-time = "2025-07-26T12:02:02.128Z" }, + { url = "https://files.pythonhosted.org/packages/0c/ba/49923366492ffbdd4486e970d421b289a670ae8cf539c1ea9a09822b371a/contourpy-1.3.3-cp313-cp313t-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:6c3d53c796f8647d6deb1abe867daeb66dcc8a97e8455efa729516b997b8ed99", size = 388106, upload-time = "2025-07-26T12:02:03.615Z" }, + { url = "https://files.pythonhosted.org/packages/9f/52/5b00ea89525f8f143651f9f03a0df371d3cbd2fccd21ca9b768c7a6500c2/contourpy-1.3.3-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:50ed930df7289ff2a8d7afeb9603f8289e5704755c7e5c3bbd929c90c817164b", size = 352548, upload-time = "2025-07-26T12:02:05.165Z" }, + { url = "https://files.pythonhosted.org/packages/32/1d/a209ec1a3a3452d490f6b14dd92e72280c99ae3d1e73da74f8277d4ee08f/contourpy-1.3.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:4feffb6537d64b84877da813a5c30f1422ea5739566abf0bd18065ac040e120a", size = 1322297, upload-time = "2025-07-26T12:02:07.379Z" }, + { url = "https://files.pythonhosted.org/packages/bc/9e/46f0e8ebdd884ca0e8877e46a3f4e633f6c9c8c4f3f6e72be3fe075994aa/contourpy-1.3.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:2b7e9480ffe2b0cd2e787e4df64270e3a0440d9db8dc823312e2c940c167df7e", size = 1391023, upload-time = "2025-07-26T12:02:10.171Z" }, + { url = "https://files.pythonhosted.org/packages/b9/70/f308384a3ae9cd2209e0849f33c913f658d3326900d0ff5d378d6a1422d2/contourpy-1.3.3-cp313-cp313t-win32.whl", hash = "sha256:283edd842a01e3dcd435b1c5116798d661378d83d36d337b8dde1d16a5fc9ba3", size = 196157, upload-time = "2025-07-26T12:02:11.488Z" }, + { url = "https://files.pythonhosted.org/packages/b2/dd/880f890a6663b84d9e34a6f88cded89d78f0091e0045a284427cb6b18521/contourpy-1.3.3-cp313-cp313t-win_amd64.whl", hash = "sha256:87acf5963fc2b34825e5b6b048f40e3635dd547f590b04d2ab317c2619ef7ae8", size = 240570, upload-time = "2025-07-26T12:02:12.754Z" }, + { url = "https://files.pythonhosted.org/packages/80/99/2adc7d8ffead633234817ef8e9a87115c8a11927a94478f6bb3d3f4d4f7d/contourpy-1.3.3-cp313-cp313t-win_arm64.whl", hash = "sha256:3c30273eb2a55024ff31ba7d052dde990d7d8e5450f4bbb6e913558b3d6c2301", size = 199713, upload-time = "2025-07-26T12:02:14.4Z" }, + { url = "https://files.pythonhosted.org/packages/72/8b/4546f3ab60f78c514ffb7d01a0bd743f90de36f0019d1be84d0a708a580a/contourpy-1.3.3-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:fde6c716d51c04b1c25d0b90364d0be954624a0ee9d60e23e850e8d48353d07a", size = 292189, upload-time = "2025-07-26T12:02:16.095Z" }, + { url = "https://files.pythonhosted.org/packages/fd/e1/3542a9cb596cadd76fcef413f19c79216e002623158befe6daa03dbfa88c/contourpy-1.3.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:cbedb772ed74ff5be440fa8eee9bd49f64f6e3fc09436d9c7d8f1c287b121d77", size = 273251, upload-time = "2025-07-26T12:02:17.524Z" }, + { url = "https://files.pythonhosted.org/packages/b1/71/f93e1e9471d189f79d0ce2497007731c1e6bf9ef6d1d61b911430c3db4e5/contourpy-1.3.3-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:22e9b1bd7a9b1d652cd77388465dc358dafcd2e217d35552424aa4f996f524f5", size = 335810, upload-time = "2025-07-26T12:02:18.9Z" }, + { url = "https://files.pythonhosted.org/packages/91/f9/e35f4c1c93f9275d4e38681a80506b5510e9327350c51f8d4a5a724d178c/contourpy-1.3.3-cp314-cp314-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:a22738912262aa3e254e4f3cb079a95a67132fc5a063890e224393596902f5a4", size = 382871, upload-time = "2025-07-26T12:02:20.418Z" }, + { url = "https://files.pythonhosted.org/packages/b5/71/47b512f936f66a0a900d81c396a7e60d73419868fba959c61efed7a8ab46/contourpy-1.3.3-cp314-cp314-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:afe5a512f31ee6bd7d0dda52ec9864c984ca3d66664444f2d72e0dc4eb832e36", size = 386264, upload-time = "2025-07-26T12:02:21.916Z" }, + { url = "https://files.pythonhosted.org/packages/04/5f/9ff93450ba96b09c7c2b3f81c94de31c89f92292f1380261bd7195bea4ea/contourpy-1.3.3-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f64836de09927cba6f79dcd00fdd7d5329f3fccc633468507079c829ca4db4e3", size = 363819, upload-time = "2025-07-26T12:02:23.759Z" }, + { url = "https://files.pythonhosted.org/packages/3e/a6/0b185d4cc480ee494945cde102cb0149ae830b5fa17bf855b95f2e70ad13/contourpy-1.3.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:1fd43c3be4c8e5fd6e4f2baeae35ae18176cf2e5cced681cca908addf1cdd53b", size = 1333650, upload-time = "2025-07-26T12:02:26.181Z" }, + { url = "https://files.pythonhosted.org/packages/43/d7/afdc95580ca56f30fbcd3060250f66cedbde69b4547028863abd8aa3b47e/contourpy-1.3.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:6afc576f7b33cf00996e5c1102dc2a8f7cc89e39c0b55df93a0b78c1bd992b36", size = 1404833, upload-time = "2025-07-26T12:02:28.782Z" }, + { url = "https://files.pythonhosted.org/packages/e2/e2/366af18a6d386f41132a48f033cbd2102e9b0cf6345d35ff0826cd984566/contourpy-1.3.3-cp314-cp314-win32.whl", hash = "sha256:66c8a43a4f7b8df8b71ee1840e4211a3c8d93b214b213f590e18a1beca458f7d", size = 189692, upload-time = "2025-07-26T12:02:30.128Z" }, + { url = "https://files.pythonhosted.org/packages/7d/c2/57f54b03d0f22d4044b8afb9ca0e184f8b1afd57b4f735c2fa70883dc601/contourpy-1.3.3-cp314-cp314-win_amd64.whl", hash = "sha256:cf9022ef053f2694e31d630feaacb21ea24224be1c3ad0520b13d844274614fd", size = 232424, upload-time = "2025-07-26T12:02:31.395Z" }, + { url = "https://files.pythonhosted.org/packages/18/79/a9416650df9b525737ab521aa181ccc42d56016d2123ddcb7b58e926a42c/contourpy-1.3.3-cp314-cp314-win_arm64.whl", hash = "sha256:95b181891b4c71de4bb404c6621e7e2390745f887f2a026b2d99e92c17892339", size = 198300, upload-time = "2025-07-26T12:02:32.956Z" }, + { url = "https://files.pythonhosted.org/packages/1f/42/38c159a7d0f2b7b9c04c64ab317042bb6952b713ba875c1681529a2932fe/contourpy-1.3.3-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:33c82d0138c0a062380332c861387650c82e4cf1747aaa6938b9b6516762e772", size = 306769, upload-time = "2025-07-26T12:02:34.2Z" }, + { url = "https://files.pythonhosted.org/packages/c3/6c/26a8205f24bca10974e77460de68d3d7c63e282e23782f1239f226fcae6f/contourpy-1.3.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:ea37e7b45949df430fe649e5de8351c423430046a2af20b1c1961cae3afcda77", size = 287892, upload-time = "2025-07-26T12:02:35.807Z" }, + { url = "https://files.pythonhosted.org/packages/66/06/8a475c8ab718ebfd7925661747dbb3c3ee9c82ac834ccb3570be49d129f4/contourpy-1.3.3-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d304906ecc71672e9c89e87c4675dc5c2645e1f4269a5063b99b0bb29f232d13", size = 326748, upload-time = "2025-07-26T12:02:37.193Z" }, + { url = "https://files.pythonhosted.org/packages/b4/a3/c5ca9f010a44c223f098fccd8b158bb1cb287378a31ac141f04730dc49be/contourpy-1.3.3-cp314-cp314t-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:ca658cd1a680a5c9ea96dc61cdbae1e85c8f25849843aa799dfd3cb370ad4fbe", size = 375554, upload-time = "2025-07-26T12:02:38.894Z" }, + { url = "https://files.pythonhosted.org/packages/80/5b/68bd33ae63fac658a4145088c1e894405e07584a316738710b636c6d0333/contourpy-1.3.3-cp314-cp314t-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:ab2fd90904c503739a75b7c8c5c01160130ba67944a7b77bbf36ef8054576e7f", size = 388118, upload-time = "2025-07-26T12:02:40.642Z" }, + { url = "https://files.pythonhosted.org/packages/40/52/4c285a6435940ae25d7410a6c36bda5145839bc3f0beb20c707cda18b9d2/contourpy-1.3.3-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b7301b89040075c30e5768810bc96a8e8d78085b47d8be6e4c3f5a0b4ed478a0", size = 352555, upload-time = "2025-07-26T12:02:42.25Z" }, + { url = "https://files.pythonhosted.org/packages/24/ee/3e81e1dd174f5c7fefe50e85d0892de05ca4e26ef1c9a59c2a57e43b865a/contourpy-1.3.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:2a2a8b627d5cc6b7c41a4beff6c5ad5eb848c88255fda4a8745f7e901b32d8e4", size = 1322295, upload-time = "2025-07-26T12:02:44.668Z" }, + { url = "https://files.pythonhosted.org/packages/3c/b2/6d913d4d04e14379de429057cd169e5e00f6c2af3bb13e1710bcbdb5da12/contourpy-1.3.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:fd6ec6be509c787f1caf6b247f0b1ca598bef13f4ddeaa126b7658215529ba0f", size = 1391027, upload-time = "2025-07-26T12:02:47.09Z" }, + { url = "https://files.pythonhosted.org/packages/93/8a/68a4ec5c55a2971213d29a9374913f7e9f18581945a7a31d1a39b5d2dfe5/contourpy-1.3.3-cp314-cp314t-win32.whl", hash = "sha256:e74a9a0f5e3fff48fb5a7f2fd2b9b70a3fe014a67522f79b7cca4c0c7e43c9ae", size = 202428, upload-time = "2025-07-26T12:02:48.691Z" }, + { url = "https://files.pythonhosted.org/packages/fa/96/fd9f641ffedc4fa3ace923af73b9d07e869496c9cc7a459103e6e978992f/contourpy-1.3.3-cp314-cp314t-win_amd64.whl", hash = "sha256:13b68d6a62db8eafaebb8039218921399baf6e47bf85006fd8529f2a08ef33fc", size = 250331, upload-time = "2025-07-26T12:02:50.137Z" }, + { url = "https://files.pythonhosted.org/packages/ae/8c/469afb6465b853afff216f9528ffda78a915ff880ed58813ba4faf4ba0b6/contourpy-1.3.3-cp314-cp314t-win_arm64.whl", hash = "sha256:b7448cb5a725bb1e35ce88771b86fba35ef418952474492cf7c764059933ff8b", size = 203831, upload-time = "2025-07-26T12:02:51.449Z" }, + { url = "https://files.pythonhosted.org/packages/a5/29/8dcfe16f0107943fa92388c23f6e05cff0ba58058c4c95b00280d4c75a14/contourpy-1.3.3-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:cd5dfcaeb10f7b7f9dc8941717c6c2ade08f587be2226222c12b25f0483ed497", size = 278809, upload-time = "2025-07-26T12:02:52.74Z" }, + { url = "https://files.pythonhosted.org/packages/85/a9/8b37ef4f7dafeb335daee3c8254645ef5725be4d9c6aa70b50ec46ef2f7e/contourpy-1.3.3-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:0c1fc238306b35f246d61a1d416a627348b5cf0648648a031e14bb8705fcdfe8", size = 261593, upload-time = "2025-07-26T12:02:54.037Z" }, + { url = "https://files.pythonhosted.org/packages/0a/59/ebfb8c677c75605cc27f7122c90313fd2f375ff3c8d19a1694bda74aaa63/contourpy-1.3.3-pp311-pypy311_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:70f9aad7de812d6541d29d2bbf8feb22ff7e1c299523db288004e3157ff4674e", size = 302202, upload-time = "2025-07-26T12:02:55.947Z" }, + { url = "https://files.pythonhosted.org/packages/3c/37/21972a15834d90bfbfb009b9d004779bd5a07a0ec0234e5ba8f64d5736f4/contourpy-1.3.3-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5ed3657edf08512fc3fe81b510e35c2012fbd3081d2e26160f27ca28affec989", size = 329207, upload-time = "2025-07-26T12:02:57.468Z" }, + { url = "https://files.pythonhosted.org/packages/0c/58/bd257695f39d05594ca4ad60df5bcb7e32247f9951fd09a9b8edb82d1daa/contourpy-1.3.3-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:3d1a3799d62d45c18bafd41c5fa05120b96a28079f2393af559b843d1a966a77", size = 225315, upload-time = "2025-07-26T12:02:58.801Z" }, ] [[package]] @@ -517,27 +646,27 @@ wheels = [ [[package]] name = "debugpy" -version = "1.8.13" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/51/d4/f35f539e11c9344652f362c22413ec5078f677ac71229dc9b4f6f85ccaa3/debugpy-1.8.13.tar.gz", hash = "sha256:837e7bef95bdefba426ae38b9a94821ebdc5bea55627879cd48165c90b9e50ce", size = 1641193, upload-time = "2025-03-05T01:02:22.807Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/3f/32/901c7204cceb3262fdf38f4c25c9a46372c11661e8490e9ea702bc4ff448/debugpy-1.8.13-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:06859f68e817966723ffe046b896b1bd75c665996a77313370336ee9e1de3e90", size = 2076250, upload-time = "2025-03-05T01:02:26.028Z" }, - { url = "https://files.pythonhosted.org/packages/95/10/77fe746851c8d84838a807da60c7bd0ac8627a6107d6917dd3293bf8628c/debugpy-1.8.13-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cb56c2db69fb8df3168bc857d7b7d2494fed295dfdbde9a45f27b4b152f37520", size = 3560883, upload-time = "2025-03-05T01:02:28.207Z" }, - { url = "https://files.pythonhosted.org/packages/a1/ef/28f8db2070e453dda0e49b356e339d0b4e1d38058d4c4ea9e88cdc8ee8e7/debugpy-1.8.13-cp310-cp310-win32.whl", hash = "sha256:46abe0b821cad751fc1fb9f860fb2e68d75e2c5d360986d0136cd1db8cad4428", size = 5180149, upload-time = "2025-03-05T01:02:30.64Z" }, - { url = "https://files.pythonhosted.org/packages/89/16/1d53a80caf5862627d3eaffb217d4079d7e4a1df6729a2d5153733661efd/debugpy-1.8.13-cp310-cp310-win_amd64.whl", hash = "sha256:dc7b77f5d32674686a5f06955e4b18c0e41fb5a605f5b33cf225790f114cfeec", size = 5212540, upload-time = "2025-03-05T01:02:32.403Z" }, - { url = "https://files.pythonhosted.org/packages/31/90/dd2fcad8364f0964f476537481985198ce6e879760281ad1cec289f1aa71/debugpy-1.8.13-cp311-cp311-macosx_14_0_universal2.whl", hash = "sha256:eee02b2ed52a563126c97bf04194af48f2fe1f68bb522a312b05935798e922ff", size = 2174802, upload-time = "2025-03-05T01:02:34.607Z" }, - { url = "https://files.pythonhosted.org/packages/5c/c9/06ff65f15eb30dbdafd45d1575770b842ce3869ad5580a77f4e5590f1be7/debugpy-1.8.13-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4caca674206e97c85c034c1efab4483f33971d4e02e73081265ecb612af65377", size = 3133620, upload-time = "2025-03-05T01:02:36.203Z" }, - { url = "https://files.pythonhosted.org/packages/3b/49/798a4092bde16a4650f17ac5f2301d4d37e1972d65462fb25c80a83b4790/debugpy-1.8.13-cp311-cp311-win32.whl", hash = "sha256:7d9a05efc6973b5aaf076d779cf3a6bbb1199e059a17738a2aa9d27a53bcc888", size = 5104764, upload-time = "2025-03-05T01:02:38.64Z" }, - { url = "https://files.pythonhosted.org/packages/cd/d5/3684d7561c8ba2797305cf8259619acccb8d6ebe2117bb33a6897c235eee/debugpy-1.8.13-cp311-cp311-win_amd64.whl", hash = "sha256:62f9b4a861c256f37e163ada8cf5a81f4c8d5148fc17ee31fb46813bd658cdcc", size = 5129670, upload-time = "2025-03-05T01:02:40.371Z" }, - { url = "https://files.pythonhosted.org/packages/79/ad/dff929b6b5403feaab0af0e5bb460fd723f9c62538b718a9af819b8fff20/debugpy-1.8.13-cp312-cp312-macosx_14_0_universal2.whl", hash = "sha256:2b8de94c5c78aa0d0ed79023eb27c7c56a64c68217d881bee2ffbcb13951d0c1", size = 2501004, upload-time = "2025-03-05T01:02:42.602Z" }, - { url = "https://files.pythonhosted.org/packages/d6/4f/b7d42e6679f0bb525888c278b0c0d2b6dff26ed42795230bb46eaae4f9b3/debugpy-1.8.13-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:887d54276cefbe7290a754424b077e41efa405a3e07122d8897de54709dbe522", size = 4222346, upload-time = "2025-03-05T01:02:44.803Z" }, - { url = "https://files.pythonhosted.org/packages/ec/18/d9b3e88e85d41f68f77235112adc31012a784e45a3fcdbb039777d570a0f/debugpy-1.8.13-cp312-cp312-win32.whl", hash = "sha256:3872ce5453b17837ef47fb9f3edc25085ff998ce63543f45ba7af41e7f7d370f", size = 5226639, upload-time = "2025-03-05T01:02:47.144Z" }, - { url = "https://files.pythonhosted.org/packages/c9/f7/0df18a4f530ed3cc06f0060f548efe9e3316102101e311739d906f5650be/debugpy-1.8.13-cp312-cp312-win_amd64.whl", hash = "sha256:63ca7670563c320503fea26ac688988d9d6b9c6a12abc8a8cf2e7dd8e5f6b6ea", size = 5268735, upload-time = "2025-03-05T01:02:48.92Z" }, - { url = "https://files.pythonhosted.org/packages/b1/db/ae7cd645c1826aae557cebccbc448f0cc9a818d364efb88f8d80e7a03f41/debugpy-1.8.13-cp313-cp313-macosx_14_0_universal2.whl", hash = "sha256:31abc9618be4edad0b3e3a85277bc9ab51a2d9f708ead0d99ffb5bb750e18503", size = 2485416, upload-time = "2025-03-05T01:02:50.558Z" }, - { url = "https://files.pythonhosted.org/packages/ec/ed/db4b10ff3b5bb30fe41d9e86444a08bb6448e4d8265e7768450b8408dd36/debugpy-1.8.13-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a0bd87557f97bced5513a74088af0b84982b6ccb2e254b9312e29e8a5c4270eb", size = 4218784, upload-time = "2025-03-05T01:02:53.535Z" }, - { url = "https://files.pythonhosted.org/packages/82/82/ed81852a8d94086f51664d032d83c7f87cd2b087c6ea70dabec7c1ba813d/debugpy-1.8.13-cp313-cp313-win32.whl", hash = "sha256:5268ae7fdca75f526d04465931cb0bd24577477ff50e8bb03dab90983f4ebd02", size = 5226270, upload-time = "2025-03-05T01:02:56.241Z" }, - { url = "https://files.pythonhosted.org/packages/15/63/aa92fb341a78ec40f1c414ec7a7885c2ee17032eee00d12cee0cdc502af4/debugpy-1.8.13-cp313-cp313-win_amd64.whl", hash = "sha256:79ce4ed40966c4c1631d0131606b055a5a2f8e430e3f7bf8fd3744b09943e8e8", size = 5268621, upload-time = "2025-03-05T01:02:57.845Z" }, - { url = "https://files.pythonhosted.org/packages/37/4f/0b65410a08b6452bfd3f7ed6f3610f1a31fb127f46836e82d31797065dcb/debugpy-1.8.13-py2.py3-none-any.whl", hash = "sha256:d4ba115cdd0e3a70942bd562adba9ec8c651fe69ddde2298a1be296fc331906f", size = 5229306, upload-time = "2025-03-05T01:03:16.51Z" }, +version = "1.8.15" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/8c/8b/3a9a28ddb750a76eaec445c7f4d3147ea2c579a97dbd9e25d39001b92b21/debugpy-1.8.15.tar.gz", hash = "sha256:58d7a20b7773ab5ee6bdfb2e6cf622fdf1e40c9d5aef2857d85391526719ac00", size = 1643279, upload-time = "2025-07-15T16:43:29.135Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/69/51/0b4315169f0d945271db037ae6b98c0548a2d48cc036335cd1b2f5516c1b/debugpy-1.8.15-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:e9a8125c85172e3ec30985012e7a81ea5e70bbb836637f8a4104f454f9b06c97", size = 2084890, upload-time = "2025-07-15T16:43:31.239Z" }, + { url = "https://files.pythonhosted.org/packages/36/cc/a5391dedb079280d7b72418022e00ba8227ae0b5bc8b2e3d1ecffc5d6b01/debugpy-1.8.15-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7fd0b6b5eccaa745c214fd240ea82f46049d99ef74b185a3517dad3ea1ec55d9", size = 3561470, upload-time = "2025-07-15T16:43:32.515Z" }, + { url = "https://files.pythonhosted.org/packages/e8/92/acf64b92010c66b33c077dee3862c733798a2c90e7d14b25c01d771e2a0d/debugpy-1.8.15-cp310-cp310-win32.whl", hash = "sha256:8181cce4d344010f6bfe94a531c351a46a96b0f7987750932b2908e7a1e14a55", size = 5229194, upload-time = "2025-07-15T16:43:33.997Z" }, + { url = "https://files.pythonhosted.org/packages/3f/f5/c58c015c9ff78de35901bea3ab4dbf7946d7a4aa867ee73875df06ba6468/debugpy-1.8.15-cp310-cp310-win_amd64.whl", hash = "sha256:af2dcae4e4cd6e8b35f982ccab29fe65f7e8766e10720a717bc80c464584ee21", size = 5260900, upload-time = "2025-07-15T16:43:35.413Z" }, + { url = "https://files.pythonhosted.org/packages/d2/b3/1c44a2ed311199ab11c2299c9474a6c7cd80d19278defd333aeb7c287995/debugpy-1.8.15-cp311-cp311-macosx_14_0_universal2.whl", hash = "sha256:babc4fb1962dd6a37e94d611280e3d0d11a1f5e6c72ac9b3d87a08212c4b6dd3", size = 2183442, upload-time = "2025-07-15T16:43:36.733Z" }, + { url = "https://files.pythonhosted.org/packages/f6/69/e2dcb721491e1c294d348681227c9b44fb95218f379aa88e12a19d85528d/debugpy-1.8.15-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f778e68f2986a58479d0ac4f643e0b8c82fdd97c2e200d4d61e7c2d13838eb53", size = 3134215, upload-time = "2025-07-15T16:43:38.116Z" }, + { url = "https://files.pythonhosted.org/packages/17/76/4ce63b95d8294dcf2fd1820860b300a420d077df4e93afcaa25a984c2ca7/debugpy-1.8.15-cp311-cp311-win32.whl", hash = "sha256:f9d1b5abd75cd965e2deabb1a06b0e93a1546f31f9f621d2705e78104377c702", size = 5154037, upload-time = "2025-07-15T16:43:39.471Z" }, + { url = "https://files.pythonhosted.org/packages/c2/a7/e5a7c784465eb9c976d84408873d597dc7ce74a0fc69ed009548a1a94813/debugpy-1.8.15-cp311-cp311-win_amd64.whl", hash = "sha256:62954fb904bec463e2b5a415777f6d1926c97febb08ef1694da0e5d1463c5c3b", size = 5178133, upload-time = "2025-07-15T16:43:40.969Z" }, + { url = "https://files.pythonhosted.org/packages/ab/4a/4508d256e52897f5cdfee6a6d7580974811e911c6d01321df3264508a5ac/debugpy-1.8.15-cp312-cp312-macosx_14_0_universal2.whl", hash = "sha256:3dcc7225cb317469721ab5136cda9ff9c8b6e6fb43e87c9e15d5b108b99d01ba", size = 2511197, upload-time = "2025-07-15T16:43:42.343Z" }, + { url = "https://files.pythonhosted.org/packages/99/8d/7f6ef1097e7fecf26b4ef72338d08e41644a41b7ee958a19f494ffcffc29/debugpy-1.8.15-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:047a493ca93c85ccede1dbbaf4e66816794bdc214213dde41a9a61e42d27f8fc", size = 4229517, upload-time = "2025-07-15T16:43:44.14Z" }, + { url = "https://files.pythonhosted.org/packages/3f/e8/e8c6a9aa33a9c9c6dacbf31747384f6ed2adde4de2e9693c766bdf323aa3/debugpy-1.8.15-cp312-cp312-win32.whl", hash = "sha256:b08e9b0bc260cf324c890626961dad4ffd973f7568fbf57feb3c3a65ab6b6327", size = 5276132, upload-time = "2025-07-15T16:43:45.529Z" }, + { url = "https://files.pythonhosted.org/packages/e9/ad/231050c6177b3476b85fcea01e565dac83607b5233d003ff067e2ee44d8f/debugpy-1.8.15-cp312-cp312-win_amd64.whl", hash = "sha256:e2a4fe357c92334272eb2845fcfcdbec3ef9f22c16cf613c388ac0887aed15fa", size = 5317645, upload-time = "2025-07-15T16:43:46.968Z" }, + { url = "https://files.pythonhosted.org/packages/28/70/2928aad2310726d5920b18ed9f54b9f06df5aa4c10cf9b45fa18ff0ab7e8/debugpy-1.8.15-cp313-cp313-macosx_14_0_universal2.whl", hash = "sha256:f5e01291ad7d6649aed5773256c5bba7a1a556196300232de1474c3c372592bf", size = 2495538, upload-time = "2025-07-15T16:43:48.927Z" }, + { url = "https://files.pythonhosted.org/packages/9e/c6/9b8ffb4ca91fac8b2877eef63c9cc0e87dd2570b1120054c272815ec4cd0/debugpy-1.8.15-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:94dc0f0d00e528d915e0ce1c78e771475b2335b376c49afcc7382ee0b146bab6", size = 4221874, upload-time = "2025-07-15T16:43:50.282Z" }, + { url = "https://files.pythonhosted.org/packages/55/8a/9b8d59674b4bf489318c7c46a1aab58e606e583651438084b7e029bf3c43/debugpy-1.8.15-cp313-cp313-win32.whl", hash = "sha256:fcf0748d4f6e25f89dc5e013d1129ca6f26ad4da405e0723a4f704583896a709", size = 5275949, upload-time = "2025-07-15T16:43:52.079Z" }, + { url = "https://files.pythonhosted.org/packages/72/83/9e58e6fdfa8710a5e6ec06c2401241b9ad48b71c0a7eb99570a1f1edb1d3/debugpy-1.8.15-cp313-cp313-win_amd64.whl", hash = "sha256:73c943776cb83e36baf95e8f7f8da765896fd94b05991e7bc162456d25500683", size = 5317720, upload-time = "2025-07-15T16:43:53.703Z" }, + { url = "https://files.pythonhosted.org/packages/07/d5/98748d9860e767a1248b5e31ffa7ce8cb7006e97bf8abbf3d891d0a8ba4e/debugpy-1.8.15-py2.py3-none-any.whl", hash = "sha256:bce2e6c5ff4f2e00b98d45e7e01a49c7b489ff6df5f12d881c67d2f1ac635f3d", size = 5282697, upload-time = "2025-07-15T16:44:07.996Z" }, ] [[package]] @@ -551,23 +680,42 @@ wheels = [ [[package]] name = "distlib" -version = "0.3.9" +version = "0.4.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/0d/dd/1bec4c5ddb504ca60fc29472f3d27e8d4da1257a854e1d96742f15c1d02d/distlib-0.3.9.tar.gz", hash = "sha256:a60f20dea646b8a33f3e7772f74dc0b2d0772d2837ee1342a00645c81edf9403", size = 613923, upload-time = "2024-10-09T18:35:47.551Z" } +sdist = { url = "https://files.pythonhosted.org/packages/96/8e/709914eb2b5749865801041647dc7f4e6d00b549cfe88b65ca192995f07c/distlib-0.4.0.tar.gz", hash = "sha256:feec40075be03a04501a973d81f633735b4b69f98b05450592310c0f401a4e0d", size = 614605, upload-time = "2025-07-17T16:52:00.465Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/91/a1/cf2472db20f7ce4a6be1253a81cfdf85ad9c7885ffbed7047fb72c24cf87/distlib-0.3.9-py2.py3-none-any.whl", hash = "sha256:47f8c22fd27c27e25a65601af709b38e4f0a45ea4fc2e710f65755fa8caaaf87", size = 468973, upload-time = "2024-10-09T18:35:44.272Z" }, + { url = "https://files.pythonhosted.org/packages/33/6b/e0547afaf41bf2c42e52430072fa5658766e3d65bd4b03a563d1b6336f57/distlib-0.4.0-py2.py3-none-any.whl", hash = "sha256:9659f7d87e46584a30b5780e43ac7a2143098441670ff0a49d5f9034c54a6c16", size = 469047, upload-time = "2025-07-17T16:51:58.613Z" }, ] [[package]] -name = "docker-pycreds" -version = "0.4.0" +name = "dm-tree" +version = "0.1.8" source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "six" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/c5/e6/d1f6c00b7221e2d7c4b470132c931325c8b22c51ca62417e300f5ce16009/docker-pycreds-0.4.0.tar.gz", hash = "sha256:6ce3270bcaf404cc4c3e27e4b6c70d3521deae82fb508767870fdbf772d584d4", size = 8754, upload-time = "2018-11-29T03:26:50.996Z" } +sdist = { url = "https://files.pythonhosted.org/packages/f8/6d/f1997aac42e0f550c1e952a0b920eaa0bfc4d27d0421499881b934b969fc/dm-tree-0.1.8.tar.gz", hash = "sha256:0fcaabbb14e7980377439e7140bd05552739ca5e515ecb3119f234acee4b9430", size = 35384, upload-time = "2022-12-18T09:46:55.953Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/f5/e8/f6bd1eee09314e7e6dee49cbe2c5e22314ccdb38db16c9fc72d2fa80d054/docker_pycreds-0.4.0-py2.py3-none-any.whl", hash = "sha256:7266112468627868005106ec19cd0d722702d2b7d5912a28e19b826c3d37af49", size = 8982, upload-time = "2018-11-29T03:26:49.575Z" }, + { url = "https://files.pythonhosted.org/packages/be/3b/d5ef06ee302ecea27351b18c28f2bde7ac982c774967d7bc82f7765fa0cb/dm_tree-0.1.8-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:35cc164a79336bfcfafb47e5f297898359123bbd3330c1967f0c4994f9cf9f60", size = 167626, upload-time = "2022-12-18T09:46:03.126Z" }, + { url = "https://files.pythonhosted.org/packages/63/29/b7c77a2500742ebbc956c2e6c9c215abeb4348040ddda72a61c760999d64/dm_tree-0.1.8-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:39070ba268c0491af9fe7a58644d99e8b4f2cde6e5884ba3380bddc84ed43d5f", size = 115351, upload-time = "2022-12-18T09:46:05.517Z" }, + { url = "https://files.pythonhosted.org/packages/ab/b0/8bf47b99c302a01db55ec43645663a385b8d3dfeb94b5fe6adf03b1121dc/dm_tree-0.1.8-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:2869228d9c619074de501a3c10dc7f07c75422f8fab36ecdcb859b6f1b1ec3ef", size = 110653, upload-time = "2022-12-18T09:46:07.869Z" }, + { url = "https://files.pythonhosted.org/packages/4c/4b/046c634913643333b1cf8f0dedd45683278013c0fb187fe36915b233ac7b/dm_tree-0.1.8-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d20f2faa3672b52e5013f4077117bfb99c4cfc0b445d3bde1584c34032b57436", size = 146732, upload-time = "2023-01-21T08:49:45.871Z" }, + { url = "https://files.pythonhosted.org/packages/ea/79/8f65fee71f3cf8bd993031578425fb10f42840b5d9a7298da0c1d52281f7/dm_tree-0.1.8-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5483dca4d7eb1a0d65fe86d3b6a53ae717face83c1f17e0887b1a4a64ae5c410", size = 174704, upload-time = "2023-01-21T08:49:48.433Z" }, + { url = "https://files.pythonhosted.org/packages/3e/9e/20bdcf1953949d8aa1e614f5c6cc1f9b556d4d72e0731e5aa1d353423bb1/dm_tree-0.1.8-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1d7c26e431fc93cc7e0cba867eb000db6a05f6f2b25af11ac4e9dada88fc5bca", size = 150386, upload-time = "2023-01-21T08:49:50.439Z" }, + { url = "https://files.pythonhosted.org/packages/cc/2b/a13e3a44f9121ecab0057af462baeb64dc50eb269de52648db8823bc12ae/dm_tree-0.1.8-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e4d714371bb08839e4e5e29024fc95832d9affe129825ef38836b143028bd144", size = 152844, upload-time = "2022-12-18T09:46:10.308Z" }, + { url = "https://files.pythonhosted.org/packages/f0/5d/86eb4e071ff395fed0783076e94c56ad9a97ba7b6e49b5aaf1b651a4fcd3/dm_tree-0.1.8-cp310-cp310-win_amd64.whl", hash = "sha256:d40fa4106ca6edc66760246a08f500ec0c85ef55c762fb4a363f6ee739ba02ee", size = 101319, upload-time = "2022-12-18T09:46:12.352Z" }, + { url = "https://files.pythonhosted.org/packages/e2/64/901b324804793743f0fdc9e47db893bf0ded9e074850fab2440af330fe83/dm_tree-0.1.8-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:ad16ceba90a56ec47cf45b21856d14962ac314787975ef786efb5e6e9ca75ec7", size = 167628, upload-time = "2022-12-18T09:46:14.195Z" }, + { url = "https://files.pythonhosted.org/packages/b1/65/4f10a68dde5fa0c91043c9c899e9bc79b1657ba932d39a5f8525c0058e68/dm_tree-0.1.8-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:803bfc53b4659f447ac694dbd04235f94a73ef7c1fd1e0df7c84ac41e0bc963b", size = 115351, upload-time = "2022-12-18T09:46:16.467Z" }, + { url = "https://files.pythonhosted.org/packages/08/e2/4c29cb9876456517f21979ddcbb6048f28a3b52c61aa9d14d42adafcdca4/dm_tree-0.1.8-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:378cc8ad93c5fe3590f405a309980721f021c790ca1bdf9b15bb1d59daec57f5", size = 110661, upload-time = "2022-12-18T09:46:18.821Z" }, + { url = "https://files.pythonhosted.org/packages/fe/89/386332bbd7567c4ccc13aa2e58f733237503fc75fb389955d3b06b9fb967/dm_tree-0.1.8-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1607ce49aa42f010d1e5e616d92ce899d66835d4d8bea49679582435285515de", size = 146727, upload-time = "2023-01-21T08:49:52.992Z" }, + { url = "https://files.pythonhosted.org/packages/a3/e7/b0c04ea5af82c19fd5984bfe980f4012601c4708634c7c51a952b17c93b2/dm_tree-0.1.8-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:343a4a4ebaa127451ff971254a4be4084eb4bdc0b2513c32b46f6f728fd03f9e", size = 174689, upload-time = "2023-01-21T08:49:56.279Z" }, + { url = "https://files.pythonhosted.org/packages/13/0d/09a4ecb54c03db53d9eb5bbc81609d89de26e3762743f003282c1b48debb/dm_tree-0.1.8-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:fa42a605d099ee7d41ba2b5fb75e21423951fd26e5d50583a00471238fb3021d", size = 150338, upload-time = "2023-01-21T08:49:59.049Z" }, + { url = "https://files.pythonhosted.org/packages/4a/27/c5e3580a952a07e5a1428ae952874796870dc8db789f3d774e886160a9f4/dm_tree-0.1.8-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:83b7764de0d855338abefc6e3ee9fe40d301668310aa3baea3f778ff051f4393", size = 152800, upload-time = "2022-12-18T09:46:21.065Z" }, + { url = "https://files.pythonhosted.org/packages/e4/c1/522041457444b67125ac9527208bb3148f63d7dce0a86ffa589ec763a10e/dm_tree-0.1.8-cp311-cp311-win_amd64.whl", hash = "sha256:a5d819c38c03f0bb5b3b3703c60e4b170355a0fc6b5819325bf3d4ceb3ae7e80", size = 101336, upload-time = "2022-12-18T09:46:23.449Z" }, + { url = "https://files.pythonhosted.org/packages/72/2c/e33dfc96f974ae3cba82c9836371c93fcb4d59d5a82ebb853861618a0b0b/dm_tree-0.1.8-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:ea9e59e0451e7d29aece402d9f908f2e2a80922bcde2ebfd5dcb07750fcbfee8", size = 169495, upload-time = "2024-02-06T09:09:13.276Z" }, + { url = "https://files.pythonhosted.org/packages/17/af/4030827253a5d50eb8da6f7189bc33d3c850c4109cf3414910e9af677cb7/dm_tree-0.1.8-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:94d3f0826311f45ee19b75f5b48c99466e4218a0489e81c0f0167bda50cacf22", size = 116525, upload-time = "2024-02-06T09:09:15.529Z" }, + { url = "https://files.pythonhosted.org/packages/10/10/5f9eed00b1186921e447960443f03cda6374cba8cd5cf7aff2b42ecb8a0e/dm_tree-0.1.8-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:435227cf3c5dc63f4de054cf3d00183790bd9ead4c3623138c74dde7f67f521b", size = 111436, upload-time = "2024-02-06T09:09:16.781Z" }, + { url = "https://files.pythonhosted.org/packages/4a/da/3d3d04f7a572f7649f48edc9402ff5836e2f90e18445ffde110fd6142889/dm_tree-0.1.8-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:09964470f76a5201aff2e8f9b26842976de7889300676f927930f6285e256760", size = 146828, upload-time = "2024-02-13T21:25:21.639Z" }, + { url = "https://files.pythonhosted.org/packages/c4/12/0a8c2152655ca39c1059c762ea1dc12784166c735126eb0ab929c518ef4e/dm_tree-0.1.8-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:75c5d528bb992981c20793b6b453e91560784215dffb8a5440ba999753c14ceb", size = 175054, upload-time = "2024-02-13T21:25:23.532Z" }, + { url = "https://files.pythonhosted.org/packages/c9/d4/8cbb857612ca69763ee4f4f97c7b91659df1d373d62237cb9c772e55ae97/dm_tree-0.1.8-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c0a94aba18a35457a1b5cd716fd7b46c5dafdc4cf7869b4bae665b91c4682a8e", size = 152834, upload-time = "2024-02-06T09:09:18.536Z" }, + { url = "https://files.pythonhosted.org/packages/ad/e3/96f5267fe5a47c882dce7f3d06b26ddd756681fc4fbedd55d51b78b08bca/dm_tree-0.1.8-cp312-cp312-win_amd64.whl", hash = "sha256:96a548a406a6fb15fe58f6a30a57ff2f2aafbf25f05afab00c8f5e5977b6c715", size = 101754, upload-time = "2024-02-06T09:09:20.962Z" }, ] [[package]] @@ -576,7 +724,8 @@ version = "0.5.6" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "opt-einsum-fx" }, - { name = "scipy" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "sympy" }, { name = "torch" }, ] @@ -587,16 +736,18 @@ wheels = [ [[package]] name = "e3tools" -version = "0.1.1" +version = "0.1.3" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "e3nn" }, + { name = "einops" }, { name = "jaxtyping" }, + { name = "setuptools" }, { name = "torch" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/5a/af/e48b3be6fa30e83adb553869da760117eaa643b7c2af9946b805cf48b325/e3tools-0.1.1.tar.gz", hash = "sha256:23022378554b9ca73f1480e4c575088ca7a73ee4adee7c761eb29907e8cc2098", size = 75249, upload-time = "2025-03-07T00:05:40.901Z" } +sdist = { url = "https://files.pythonhosted.org/packages/1a/03/d3560619bc9c7d7bb1548a260087201a217400f125f6b8c68e5c5532be3e/e3tools-0.1.3.tar.gz", hash = "sha256:a49d919b6f754767ca3c09eaa6a6e1c12fbddace156572b878bbe40ad70ceaa8", size = 106214, upload-time = "2025-08-04T23:29:52.313Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/fe/fd/f3bada4eb07cbca953d403ca121cc8224554bdee905cb2bd176cd75929fa/e3tools-0.1.1-py3-none-any.whl", hash = "sha256:40346ef8e17d966e16d3754e8945c30013baf2c6c4b959c5dacc9b810946d169", size = 18236, upload-time = "2025-03-07T00:05:39.723Z" }, + { url = "https://files.pythonhosted.org/packages/78/c2/76b4ea3bb1426a13d769bf1ae5a13766e8398116f9f158ee5298970c6664/e3tools-0.1.3-py3-none-any.whl", hash = "sha256:39fd064c42f5fe2edd5b15955b5cc514bf64863a5178dd6c041e73423dd03239", size = 21918, upload-time = "2025-08-04T23:29:51.107Z" }, ] [[package]] @@ -610,11 +761,14 @@ wheels = [ [[package]] name = "exceptiongroup" -version = "1.2.2" +version = "1.3.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/09/35/2495c4ac46b980e4ca1f6ad6db102322ef3ad2410b79fdde159a4b0f3b92/exceptiongroup-1.2.2.tar.gz", hash = "sha256:47c2edf7c6738fafb49fd34290706d1a1a2f4d1c6df275526b62cbb4aa5393cc", size = 28883, upload-time = "2024-07-12T22:26:00.161Z" } +dependencies = [ + { name = "typing-extensions", marker = "python_full_version < '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/0b/9f/a65090624ecf468cdca03533906e7c69ed7588582240cfe7cc9e770b50eb/exceptiongroup-1.3.0.tar.gz", hash = "sha256:b241f5885f560bc56a59ee63ca4c6a8bfa46ae4ad651af316d4e81817bb9fd88", size = 29749, upload-time = "2025-05-10T17:42:51.123Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/02/cc/b7e31358aac6ed1ef2bb790a9746ac2c69bcb3c8588b41616914eb106eaf/exceptiongroup-1.2.2-py3-none-any.whl", hash = "sha256:3111b9d131c238bec2f8f516e123e14ba243563fb135d3fe885990585aa7795b", size = 16453, upload-time = "2024-07-12T22:25:58.476Z" }, + { url = "https://files.pythonhosted.org/packages/36/f4/c6e662dade71f56cd2f3735141b265c3c79293c109549c1e6933b0651ffc/exceptiongroup-1.3.0-py3-none-any.whl", hash = "sha256:4d111e6e0c13d0644cad6ddaa7ed0261a0b36971f6d23e7ec9b4b9097da78a10", size = 16674, upload-time = "2025-05-10T17:42:49.33Z" }, ] [[package]] @@ -646,121 +800,146 @@ wheels = [ [[package]] name = "fonttools" -version = "4.56.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/1c/8c/9ffa2a555af0e5e5d0e2ed7fdd8c9bef474ed676995bb4c57c9cd0014248/fonttools-4.56.0.tar.gz", hash = "sha256:a114d1567e1a1586b7e9e7fc2ff686ca542a82769a296cef131e4c4af51e58f4", size = 3462892, upload-time = "2025-02-07T13:46:29.026Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/1e/5e/6ac30c2cc6a29454260f13c9c6422fc509b7982c13cd4597041260d8f482/fonttools-4.56.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:331954d002dbf5e704c7f3756028e21db07097c19722569983ba4d74df014000", size = 2752190, upload-time = "2025-02-07T13:43:30.593Z" }, - { url = "https://files.pythonhosted.org/packages/92/3a/ac382a8396d1b420ee45eeb0f65b614a9ca7abbb23a1b17524054f0f2200/fonttools-4.56.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:8d1613abd5af2f93c05867b3a3759a56e8bf97eb79b1da76b2bc10892f96ff16", size = 2280624, upload-time = "2025-02-07T13:43:35.349Z" }, - { url = "https://files.pythonhosted.org/packages/8a/ae/00b58bfe20e9ff7fbc3dda38f5d127913942b5e252288ea9583099a31bf5/fonttools-4.56.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:705837eae384fe21cee5e5746fd4f4b2f06f87544fa60f60740007e0aa600311", size = 4562074, upload-time = "2025-02-07T13:43:38.799Z" }, - { url = "https://files.pythonhosted.org/packages/46/d0/0004ca8f6a200252e5bd6982ed99b5fe58c4c59efaf5f516621c4cd8f703/fonttools-4.56.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc871904a53a9d4d908673c6faa15689874af1c7c5ac403a8e12d967ebd0c0dc", size = 4604747, upload-time = "2025-02-07T13:43:41.831Z" }, - { url = "https://files.pythonhosted.org/packages/45/ea/c8862bd3e09d143ef8ed8268ec8a7d477828f960954889e65288ac050b08/fonttools-4.56.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:38b947de71748bab150259ee05a775e8a0635891568e9fdb3cdd7d0e0004e62f", size = 4559025, upload-time = "2025-02-07T13:43:45.525Z" }, - { url = "https://files.pythonhosted.org/packages/8f/75/bb88a9552ec1de31a414066257bfd9f40f4ada00074f7a3799ea39b5741f/fonttools-4.56.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:86b2a1013ef7a64d2e94606632683f07712045ed86d937c11ef4dde97319c086", size = 4728482, upload-time = "2025-02-07T13:43:49.296Z" }, - { url = "https://files.pythonhosted.org/packages/2a/5f/80a2b640df1e1bb7d459d62c8b3f37fe83fd413897e549106d4ebe6371f5/fonttools-4.56.0-cp310-cp310-win32.whl", hash = "sha256:133bedb9a5c6376ad43e6518b7e2cd2f866a05b1998f14842631d5feb36b5786", size = 2155557, upload-time = "2025-02-07T13:43:52.029Z" }, - { url = "https://files.pythonhosted.org/packages/8f/85/0904f9dbe51ac70d878d3242a8583b9453a09105c3ed19c6301247fd0d3a/fonttools-4.56.0-cp310-cp310-win_amd64.whl", hash = "sha256:17f39313b649037f6c800209984a11fc256a6137cbe5487091c6c7187cae4685", size = 2200017, upload-time = "2025-02-07T13:43:54.768Z" }, - { url = "https://files.pythonhosted.org/packages/35/56/a2f3e777d48fcae7ecd29de4d96352d84e5ea9871e5f3fc88241521572cf/fonttools-4.56.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:7ef04bc7827adb7532be3d14462390dd71287644516af3f1e67f1e6ff9c6d6df", size = 2753325, upload-time = "2025-02-07T13:43:57.855Z" }, - { url = "https://files.pythonhosted.org/packages/71/85/d483e9c4e5ed586b183bf037a353e8d766366b54fd15519b30e6178a6a6e/fonttools-4.56.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ffda9b8cd9cb8b301cae2602ec62375b59e2e2108a117746f12215145e3f786c", size = 2281554, upload-time = "2025-02-07T13:44:01.671Z" }, - { url = "https://files.pythonhosted.org/packages/09/67/060473b832b2fade03c127019794df6dc02d9bc66fa4210b8e0d8a99d1e5/fonttools-4.56.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e2e993e8db36306cc3f1734edc8ea67906c55f98683d6fd34c3fc5593fdbba4c", size = 4869260, upload-time = "2025-02-07T13:44:05.746Z" }, - { url = "https://files.pythonhosted.org/packages/28/e9/47c02d5a7027e8ed841ab6a10ca00c93dadd5f16742f1af1fa3f9978adf4/fonttools-4.56.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:003548eadd674175510773f73fb2060bb46adb77c94854af3e0cc5bc70260049", size = 4898508, upload-time = "2025-02-07T13:44:09.965Z" }, - { url = "https://files.pythonhosted.org/packages/bf/8a/221d456d1afb8ca043cfd078f59f187ee5d0a580f4b49351b9ce95121f57/fonttools-4.56.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:bd9825822e7bb243f285013e653f6741954d8147427aaa0324a862cdbf4cbf62", size = 4877700, upload-time = "2025-02-07T13:44:13.598Z" }, - { url = "https://files.pythonhosted.org/packages/a4/8c/e503863adf7a6aeff7b960e2f66fa44dd0c29a7a8b79765b2821950d7b05/fonttools-4.56.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b23d30a2c0b992fb1c4f8ac9bfde44b5586d23457759b6cf9a787f1a35179ee0", size = 5045817, upload-time = "2025-02-07T13:44:17.532Z" }, - { url = "https://files.pythonhosted.org/packages/2b/50/79ba3b7e42f4eaa70b82b9e79155f0f6797858dc8a97862428b6852c6aee/fonttools-4.56.0-cp311-cp311-win32.whl", hash = "sha256:47b5e4680002ae1756d3ae3b6114e20aaee6cc5c69d1e5911f5ffffd3ee46c6b", size = 2154426, upload-time = "2025-02-07T13:44:21.063Z" }, - { url = "https://files.pythonhosted.org/packages/3b/90/4926e653041c4116ecd43e50e3c79f5daae6dcafc58ceb64bc4f71dd4924/fonttools-4.56.0-cp311-cp311-win_amd64.whl", hash = "sha256:14a3e3e6b211660db54ca1ef7006401e4a694e53ffd4553ab9bc87ead01d0f05", size = 2200937, upload-time = "2025-02-07T13:44:24.607Z" }, - { url = "https://files.pythonhosted.org/packages/39/32/71cfd6877999576a11824a7fe7bc0bb57c5c72b1f4536fa56a3e39552643/fonttools-4.56.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:d6f195c14c01bd057bc9b4f70756b510e009c83c5ea67b25ced3e2c38e6ee6e9", size = 2747757, upload-time = "2025-02-07T13:44:28.021Z" }, - { url = "https://files.pythonhosted.org/packages/15/52/d9f716b072c5061a0b915dd4c387f74bef44c68c069e2195c753905bd9b7/fonttools-4.56.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:fa760e5fe8b50cbc2d71884a1eff2ed2b95a005f02dda2fa431560db0ddd927f", size = 2279007, upload-time = "2025-02-07T13:44:31.325Z" }, - { url = "https://files.pythonhosted.org/packages/d1/97/f1b3a8afa9a0d814a092a25cd42f59ccb98a0bb7a295e6e02fc9ba744214/fonttools-4.56.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d54a45d30251f1d729e69e5b675f9a08b7da413391a1227781e2a297fa37f6d2", size = 4783991, upload-time = "2025-02-07T13:44:34.888Z" }, - { url = "https://files.pythonhosted.org/packages/95/70/2a781bedc1c45a0c61d29c56425609b22ed7f971da5d7e5df2679488741b/fonttools-4.56.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:661a8995d11e6e4914a44ca7d52d1286e2d9b154f685a4d1f69add8418961563", size = 4855109, upload-time = "2025-02-07T13:44:40.702Z" }, - { url = "https://files.pythonhosted.org/packages/0c/02/a2597858e61a5e3fb6a14d5f6be9e6eb4eaf090da56ad70cedcbdd201685/fonttools-4.56.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:9d94449ad0a5f2a8bf5d2f8d71d65088aee48adbe45f3c5f8e00e3ad861ed81a", size = 4762496, upload-time = "2025-02-07T13:44:45.929Z" }, - { url = "https://files.pythonhosted.org/packages/f2/00/aaf00100d6078fdc73f7352b44589804af9dc12b182a2540b16002152ba4/fonttools-4.56.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:f59746f7953f69cc3290ce2f971ab01056e55ddd0fb8b792c31a8acd7fee2d28", size = 4990094, upload-time = "2025-02-07T13:44:49.004Z" }, - { url = "https://files.pythonhosted.org/packages/bf/dc/3ff1db522460db60cf3adaf1b64e0c72b43406717d139786d3fa1eb20709/fonttools-4.56.0-cp312-cp312-win32.whl", hash = "sha256:bce60f9a977c9d3d51de475af3f3581d9b36952e1f8fc19a1f2254f1dda7ce9c", size = 2142888, upload-time = "2025-02-07T13:44:54.127Z" }, - { url = "https://files.pythonhosted.org/packages/6f/e3/5a181a85777f7809076e51f7422e0dc77eb04676c40ec8bf6a49d390d1ff/fonttools-4.56.0-cp312-cp312-win_amd64.whl", hash = "sha256:300c310bb725b2bdb4f5fc7e148e190bd69f01925c7ab437b9c0ca3e1c7cd9ba", size = 2189734, upload-time = "2025-02-07T13:44:57.393Z" }, - { url = "https://files.pythonhosted.org/packages/a5/55/f06b48d48e0b4ec3a3489efafe9bd4d81b6e0802ac51026e3ee4634e89ba/fonttools-4.56.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:f20e2c0dfab82983a90f3d00703ac0960412036153e5023eed2b4641d7d5e692", size = 2735127, upload-time = "2025-02-07T13:44:59.966Z" }, - { url = "https://files.pythonhosted.org/packages/59/db/d2c7c9b6dd5cbd46f183e650a47403ffb88fca17484eb7c4b1cd88f9e513/fonttools-4.56.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:f36a0868f47b7566237640c026c65a86d09a3d9ca5df1cd039e30a1da73098a0", size = 2272519, upload-time = "2025-02-07T13:45:03.891Z" }, - { url = "https://files.pythonhosted.org/packages/4d/a2/da62d779c34a0e0c06415f02eab7fa3466de5d46df459c0275a255cefc65/fonttools-4.56.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:62b4c6802fa28e14dba010e75190e0e6228513573f1eeae57b11aa1a39b7e5b1", size = 4762423, upload-time = "2025-02-07T13:45:07.034Z" }, - { url = "https://files.pythonhosted.org/packages/be/6a/fd4018e0448c8a5e12138906411282c5eab51a598493f080a9f0960e658f/fonttools-4.56.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a05d1f07eb0a7d755fbe01fee1fd255c3a4d3730130cf1bfefb682d18fd2fcea", size = 4834442, upload-time = "2025-02-07T13:45:10.6Z" }, - { url = "https://files.pythonhosted.org/packages/6d/63/fa1dec8efb35bc11ef9c39b2d74754b45d48a3ccb2cf78c0109c0af639e8/fonttools-4.56.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:0073b62c3438cf0058488c002ea90489e8801d3a7af5ce5f7c05c105bee815c3", size = 4742800, upload-time = "2025-02-07T13:45:14.096Z" }, - { url = "https://files.pythonhosted.org/packages/dd/f4/963247ae8c73ccc4cf2929e7162f595c81dbe17997d1d0ea77da24a217c9/fonttools-4.56.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e2cad98c94833465bcf28f51c248aaf07ca022efc6a3eba750ad9c1e0256d278", size = 4963746, upload-time = "2025-02-07T13:45:17.479Z" }, - { url = "https://files.pythonhosted.org/packages/ea/e0/46f9600c39c644b54e4420f941f75fa200d9288c9ae171e5d80918b8cbb9/fonttools-4.56.0-cp313-cp313-win32.whl", hash = "sha256:d0cb73ccf7f6d7ca8d0bc7ea8ac0a5b84969a41c56ac3ac3422a24df2680546f", size = 2140927, upload-time = "2025-02-07T13:45:21.084Z" }, - { url = "https://files.pythonhosted.org/packages/27/6d/3edda54f98a550a0473f032d8050315fbc8f1b76a0d9f3879b72ebb2cdd6/fonttools-4.56.0-cp313-cp313-win_amd64.whl", hash = "sha256:62cc1253827d1e500fde9dbe981219fea4eb000fd63402283472d38e7d8aa1c6", size = 2186709, upload-time = "2025-02-07T13:45:23.719Z" }, - { url = "https://files.pythonhosted.org/packages/bf/ff/44934a031ce5a39125415eb405b9efb76fe7f9586b75291d66ae5cbfc4e6/fonttools-4.56.0-py3-none-any.whl", hash = "sha256:1088182f68c303b50ca4dc0c82d42083d176cba37af1937e1a976a31149d4d14", size = 1089800, upload-time = "2025-02-07T13:46:26.415Z" }, +version = "4.59.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/8a/27/ec3c723bfdf86f34c5c82bf6305df3e0f0d8ea798d2d3a7cb0c0a866d286/fonttools-4.59.0.tar.gz", hash = "sha256:be392ec3529e2f57faa28709d60723a763904f71a2b63aabe14fee6648fe3b14", size = 3532521, upload-time = "2025-07-16T12:04:54.613Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/1c/1f/3dcae710b7c4b56e79442b03db64f6c9f10c3348f7af40339dffcefb581e/fonttools-4.59.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:524133c1be38445c5c0575eacea42dbd44374b310b1ffc4b60ff01d881fabb96", size = 2761846, upload-time = "2025-07-16T12:03:33.267Z" }, + { url = "https://files.pythonhosted.org/packages/eb/0e/ae3a1884fa1549acac1191cc9ec039142f6ac0e9cbc139c2e6a3dab967da/fonttools-4.59.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:21e606b2d38fed938dde871c5736822dd6bda7a4631b92e509a1f5cd1b90c5df", size = 2332060, upload-time = "2025-07-16T12:03:36.472Z" }, + { url = "https://files.pythonhosted.org/packages/75/46/58bff92a7216829159ac7bdb1d05a48ad1b8ab8c539555f12d29fdecfdd4/fonttools-4.59.0-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e93df708c69a193fc7987192f94df250f83f3851fda49413f02ba5dded639482", size = 4852354, upload-time = "2025-07-16T12:03:39.102Z" }, + { url = "https://files.pythonhosted.org/packages/05/57/767e31e48861045d89691128bd81fd4c62b62150f9a17a666f731ce4f197/fonttools-4.59.0-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:62224a9bb85b4b66d1b46d45cbe43d71cbf8f527d332b177e3b96191ffbc1e64", size = 4781132, upload-time = "2025-07-16T12:03:41.415Z" }, + { url = "https://files.pythonhosted.org/packages/d7/78/adb5e9b0af5c6ce469e8b0e112f144eaa84b30dd72a486e9c778a9b03b31/fonttools-4.59.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b8974b2a266b54c96709bd5e239979cddfd2dbceed331aa567ea1d7c4a2202db", size = 4832901, upload-time = "2025-07-16T12:03:43.115Z" }, + { url = "https://files.pythonhosted.org/packages/ac/92/bc3881097fbf3d56d112bec308c863c058e5d4c9c65f534e8ae58450ab8a/fonttools-4.59.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:209b75943d158f610b78320eacb5539aa9e920bee2c775445b2846c65d20e19d", size = 4940140, upload-time = "2025-07-16T12:03:44.781Z" }, + { url = "https://files.pythonhosted.org/packages/4a/54/39cdb23f0eeda2e07ae9cb189f2b6f41da89aabc682d3a387b3ff4a4ed29/fonttools-4.59.0-cp310-cp310-win32.whl", hash = "sha256:4c908a7036f0f3677f8afa577bcd973e3e20ddd2f7c42a33208d18bee95cdb6f", size = 2215890, upload-time = "2025-07-16T12:03:46.961Z" }, + { url = "https://files.pythonhosted.org/packages/d8/eb/f8388d9e19f95d8df2449febe9b1a38ddd758cfdb7d6de3a05198d785d61/fonttools-4.59.0-cp310-cp310-win_amd64.whl", hash = "sha256:8b4309a2775e4feee7356e63b163969a215d663399cce1b3d3b65e7ec2d9680e", size = 2260191, upload-time = "2025-07-16T12:03:48.908Z" }, + { url = "https://files.pythonhosted.org/packages/06/96/520733d9602fa1bf6592e5354c6721ac6fc9ea72bc98d112d0c38b967199/fonttools-4.59.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:841b2186adce48903c0fef235421ae21549020eca942c1da773ac380b056ab3c", size = 2782387, upload-time = "2025-07-16T12:03:51.424Z" }, + { url = "https://files.pythonhosted.org/packages/87/6a/170fce30b9bce69077d8eec9bea2cfd9f7995e8911c71be905e2eba6368b/fonttools-4.59.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:9bcc1e77fbd1609198966ded6b2a9897bd6c6bcbd2287a2fc7d75f1a254179c5", size = 2342194, upload-time = "2025-07-16T12:03:53.295Z" }, + { url = "https://files.pythonhosted.org/packages/b0/b6/7c8166c0066856f1408092f7968ac744060cf72ca53aec9036106f57eeca/fonttools-4.59.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:37c377f7cb2ab2eca8a0b319c68146d34a339792f9420fca6cd49cf28d370705", size = 5032333, upload-time = "2025-07-16T12:03:55.177Z" }, + { url = "https://files.pythonhosted.org/packages/eb/0c/707c5a19598eafcafd489b73c4cb1c142102d6197e872f531512d084aa76/fonttools-4.59.0-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fa39475eaccb98f9199eccfda4298abaf35ae0caec676ffc25b3a5e224044464", size = 4974422, upload-time = "2025-07-16T12:03:57.406Z" }, + { url = "https://files.pythonhosted.org/packages/f6/e7/6d33737d9fe632a0f59289b6f9743a86d2a9d0673de2a0c38c0f54729822/fonttools-4.59.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d3972b13148c1d1fbc092b27678a33b3080d1ac0ca305742b0119b75f9e87e38", size = 5010631, upload-time = "2025-07-16T12:03:59.449Z" }, + { url = "https://files.pythonhosted.org/packages/63/e1/a4c3d089ab034a578820c8f2dff21ef60daf9668034a1e4fb38bb1cc3398/fonttools-4.59.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:a408c3c51358c89b29cfa5317cf11518b7ce5de1717abb55c5ae2d2921027de6", size = 5122198, upload-time = "2025-07-16T12:04:01.542Z" }, + { url = "https://files.pythonhosted.org/packages/09/77/ca82b9c12fa4de3c520b7760ee61787640cf3fde55ef1b0bfe1de38c8153/fonttools-4.59.0-cp311-cp311-win32.whl", hash = "sha256:6770d7da00f358183d8fd5c4615436189e4f683bdb6affb02cad3d221d7bb757", size = 2214216, upload-time = "2025-07-16T12:04:03.515Z" }, + { url = "https://files.pythonhosted.org/packages/ab/25/5aa7ca24b560b2f00f260acf32c4cf29d7aaf8656e159a336111c18bc345/fonttools-4.59.0-cp311-cp311-win_amd64.whl", hash = "sha256:84fc186980231a287b28560d3123bd255d3c6b6659828c642b4cf961e2b923d0", size = 2261879, upload-time = "2025-07-16T12:04:05.015Z" }, + { url = "https://files.pythonhosted.org/packages/e2/77/b1c8af22f4265e951cd2e5535dbef8859efcef4fb8dee742d368c967cddb/fonttools-4.59.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:f9b3a78f69dcbd803cf2fb3f972779875b244c1115481dfbdd567b2c22b31f6b", size = 2767562, upload-time = "2025-07-16T12:04:06.895Z" }, + { url = "https://files.pythonhosted.org/packages/ff/5a/aeb975699588176bb357e8b398dfd27e5d3a2230d92b81ab8cbb6187358d/fonttools-4.59.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:57bb7e26928573ee7c6504f54c05860d867fd35e675769f3ce01b52af38d48e2", size = 2335168, upload-time = "2025-07-16T12:04:08.695Z" }, + { url = "https://files.pythonhosted.org/packages/54/97/c6101a7e60ae138c4ef75b22434373a0da50a707dad523dd19a4889315bf/fonttools-4.59.0-cp312-cp312-manylinux1_x86_64.manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:4536f2695fe5c1ffb528d84a35a7d3967e5558d2af58b4775e7ab1449d65767b", size = 4909850, upload-time = "2025-07-16T12:04:10.761Z" }, + { url = "https://files.pythonhosted.org/packages/bd/6c/fa4d18d641054f7bff878cbea14aa9433f292b9057cb1700d8e91a4d5f4f/fonttools-4.59.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:885bde7d26e5b40e15c47bd5def48b38cbd50830a65f98122a8fb90962af7cd1", size = 4955131, upload-time = "2025-07-16T12:04:12.846Z" }, + { url = "https://files.pythonhosted.org/packages/20/5c/331947fc1377deb928a69bde49f9003364f5115e5cbe351eea99e39412a2/fonttools-4.59.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6801aeddb6acb2c42eafa45bc1cb98ba236871ae6f33f31e984670b749a8e58e", size = 4899667, upload-time = "2025-07-16T12:04:14.558Z" }, + { url = "https://files.pythonhosted.org/packages/8a/46/b66469dfa26b8ff0baa7654b2cc7851206c6d57fe3abdabbaab22079a119/fonttools-4.59.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:31003b6a10f70742a63126b80863ab48175fb8272a18ca0846c0482968f0588e", size = 5051349, upload-time = "2025-07-16T12:04:16.388Z" }, + { url = "https://files.pythonhosted.org/packages/2e/05/ebfb6b1f3a4328ab69787d106a7d92ccde77ce66e98659df0f9e3f28d93d/fonttools-4.59.0-cp312-cp312-win32.whl", hash = "sha256:fbce6dae41b692a5973d0f2158f782b9ad05babc2c2019a970a1094a23909b1b", size = 2201315, upload-time = "2025-07-16T12:04:18.557Z" }, + { url = "https://files.pythonhosted.org/packages/09/45/d2bdc9ea20bbadec1016fd0db45696d573d7a26d95ab5174ffcb6d74340b/fonttools-4.59.0-cp312-cp312-win_amd64.whl", hash = "sha256:332bfe685d1ac58ca8d62b8d6c71c2e52a6c64bc218dc8f7825c9ea51385aa01", size = 2249408, upload-time = "2025-07-16T12:04:20.489Z" }, + { url = "https://files.pythonhosted.org/packages/f3/bb/390990e7c457d377b00890d9f96a3ca13ae2517efafb6609c1756e213ba4/fonttools-4.59.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:78813b49d749e1bb4db1c57f2d4d7e6db22c253cb0a86ad819f5dc197710d4b2", size = 2758704, upload-time = "2025-07-16T12:04:22.217Z" }, + { url = "https://files.pythonhosted.org/packages/df/6f/d730d9fcc9b410a11597092bd2eb9ca53e5438c6cb90e4b3047ce1b723e9/fonttools-4.59.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:401b1941ce37e78b8fd119b419b617277c65ae9417742a63282257434fd68ea2", size = 2330764, upload-time = "2025-07-16T12:04:23.985Z" }, + { url = "https://files.pythonhosted.org/packages/75/b4/b96bb66f6f8cc4669de44a158099b249c8159231d254ab6b092909388be5/fonttools-4.59.0-cp313-cp313-manylinux1_x86_64.manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:efd7e6660674e234e29937bc1481dceb7e0336bfae75b856b4fb272b5093c5d4", size = 4890699, upload-time = "2025-07-16T12:04:25.664Z" }, + { url = "https://files.pythonhosted.org/packages/b5/57/7969af50b26408be12baa317c6147588db5b38af2759e6df94554dbc5fdb/fonttools-4.59.0-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:51ab1ff33c19e336c02dee1e9fd1abd974a4ca3d8f7eef2a104d0816a241ce97", size = 4952934, upload-time = "2025-07-16T12:04:27.733Z" }, + { url = "https://files.pythonhosted.org/packages/d6/e2/dd968053b6cf1f46c904f5bd409b22341477c017d8201619a265e50762d3/fonttools-4.59.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:a9bf8adc9e1f3012edc8f09b08336272aec0c55bc677422273e21280db748f7c", size = 4892319, upload-time = "2025-07-16T12:04:30.074Z" }, + { url = "https://files.pythonhosted.org/packages/6b/95/a59810d8eda09129f83467a4e58f84205dc6994ebaeb9815406363e07250/fonttools-4.59.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:37e01c6ec0c98599778c2e688350d624fa4770fbd6144551bd5e032f1199171c", size = 5034753, upload-time = "2025-07-16T12:04:32.292Z" }, + { url = "https://files.pythonhosted.org/packages/a5/84/51a69ee89ff8d1fea0c6997e946657e25a3f08513de8435fe124929f3eef/fonttools-4.59.0-cp313-cp313-win32.whl", hash = "sha256:70d6b3ceaa9cc5a6ac52884f3b3d9544e8e231e95b23f138bdb78e6d4dc0eae3", size = 2199688, upload-time = "2025-07-16T12:04:34.444Z" }, + { url = "https://files.pythonhosted.org/packages/a0/ee/f626cd372932d828508137a79b85167fdcf3adab2e3bed433f295c596c6a/fonttools-4.59.0-cp313-cp313-win_amd64.whl", hash = "sha256:26731739daa23b872643f0e4072d5939960237d540c35c14e6a06d47d71ca8fe", size = 2248560, upload-time = "2025-07-16T12:04:36.034Z" }, + { url = "https://files.pythonhosted.org/packages/d0/9c/df0ef2c51845a13043e5088f7bb988ca6cd5bb82d5d4203d6a158aa58cf2/fonttools-4.59.0-py3-none-any.whl", hash = "sha256:241313683afd3baacb32a6bd124d0bce7404bc5280e12e291bae1b9bba28711d", size = 1128050, upload-time = "2025-07-16T12:04:52.687Z" }, ] [[package]] name = "frozenlist" -version = "1.5.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/8f/ed/0f4cec13a93c02c47ec32d81d11c0c1efbadf4a471e3f3ce7cad366cbbd3/frozenlist-1.5.0.tar.gz", hash = "sha256:81d5af29e61b9c8348e876d442253723928dce6433e0e76cd925cd83f1b4b817", size = 39930, upload-time = "2024-10-23T09:48:29.903Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/54/79/29d44c4af36b2b240725dce566b20f63f9b36ef267aaaa64ee7466f4f2f8/frozenlist-1.5.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:5b6a66c18b5b9dd261ca98dffcb826a525334b2f29e7caa54e182255c5f6a65a", size = 94451, upload-time = "2024-10-23T09:46:20.558Z" }, - { url = "https://files.pythonhosted.org/packages/47/47/0c999aeace6ead8a44441b4f4173e2261b18219e4ad1fe9a479871ca02fc/frozenlist-1.5.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:d1b3eb7b05ea246510b43a7e53ed1653e55c2121019a97e60cad7efb881a97bb", size = 54301, upload-time = "2024-10-23T09:46:21.759Z" }, - { url = "https://files.pythonhosted.org/packages/8d/60/107a38c1e54176d12e06e9d4b5d755b677d71d1219217cee063911b1384f/frozenlist-1.5.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:15538c0cbf0e4fa11d1e3a71f823524b0c46299aed6e10ebb4c2089abd8c3bec", size = 52213, upload-time = "2024-10-23T09:46:22.993Z" }, - { url = "https://files.pythonhosted.org/packages/17/62/594a6829ac5679c25755362a9dc93486a8a45241394564309641425d3ff6/frozenlist-1.5.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e79225373c317ff1e35f210dd5f1344ff31066ba8067c307ab60254cd3a78ad5", size = 240946, upload-time = "2024-10-23T09:46:24.661Z" }, - { url = "https://files.pythonhosted.org/packages/7e/75/6c8419d8f92c80dd0ee3f63bdde2702ce6398b0ac8410ff459f9b6f2f9cb/frozenlist-1.5.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9272fa73ca71266702c4c3e2d4a28553ea03418e591e377a03b8e3659d94fa76", size = 264608, upload-time = "2024-10-23T09:46:26.017Z" }, - { url = "https://files.pythonhosted.org/packages/88/3e/82a6f0b84bc6fb7e0be240e52863c6d4ab6098cd62e4f5b972cd31e002e8/frozenlist-1.5.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:498524025a5b8ba81695761d78c8dd7382ac0b052f34e66939c42df860b8ff17", size = 261361, upload-time = "2024-10-23T09:46:27.787Z" }, - { url = "https://files.pythonhosted.org/packages/fd/85/14e5f9ccac1b64ff2f10c927b3ffdf88772aea875882406f9ba0cec8ad84/frozenlist-1.5.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:92b5278ed9d50fe610185ecd23c55d8b307d75ca18e94c0e7de328089ac5dcba", size = 231649, upload-time = "2024-10-23T09:46:28.992Z" }, - { url = "https://files.pythonhosted.org/packages/ee/59/928322800306f6529d1852323014ee9008551e9bb027cc38d276cbc0b0e7/frozenlist-1.5.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7f3c8c1dacd037df16e85227bac13cca58c30da836c6f936ba1df0c05d046d8d", size = 241853, upload-time = "2024-10-23T09:46:30.211Z" }, - { url = "https://files.pythonhosted.org/packages/7d/bd/e01fa4f146a6f6c18c5d34cab8abdc4013774a26c4ff851128cd1bd3008e/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:f2ac49a9bedb996086057b75bf93538240538c6d9b38e57c82d51f75a73409d2", size = 243652, upload-time = "2024-10-23T09:46:31.758Z" }, - { url = "https://files.pythonhosted.org/packages/a5/bd/e4771fd18a8ec6757033f0fa903e447aecc3fbba54e3630397b61596acf0/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:e66cc454f97053b79c2ab09c17fbe3c825ea6b4de20baf1be28919460dd7877f", size = 241734, upload-time = "2024-10-23T09:46:33.044Z" }, - { url = "https://files.pythonhosted.org/packages/21/13/c83821fa5544af4f60c5d3a65d054af3213c26b14d3f5f48e43e5fb48556/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:5a3ba5f9a0dfed20337d3e966dc359784c9f96503674c2faf015f7fe8e96798c", size = 260959, upload-time = "2024-10-23T09:46:34.916Z" }, - { url = "https://files.pythonhosted.org/packages/71/f3/1f91c9a9bf7ed0e8edcf52698d23f3c211d8d00291a53c9f115ceb977ab1/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:6321899477db90bdeb9299ac3627a6a53c7399c8cd58d25da094007402b039ab", size = 262706, upload-time = "2024-10-23T09:46:36.159Z" }, - { url = "https://files.pythonhosted.org/packages/4c/22/4a256fdf5d9bcb3ae32622c796ee5ff9451b3a13a68cfe3f68e2c95588ce/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:76e4753701248476e6286f2ef492af900ea67d9706a0155335a40ea21bf3b2f5", size = 250401, upload-time = "2024-10-23T09:46:37.327Z" }, - { url = "https://files.pythonhosted.org/packages/af/89/c48ebe1f7991bd2be6d5f4ed202d94960c01b3017a03d6954dd5fa9ea1e8/frozenlist-1.5.0-cp310-cp310-win32.whl", hash = "sha256:977701c081c0241d0955c9586ffdd9ce44f7a7795df39b9151cd9a6fd0ce4cfb", size = 45498, upload-time = "2024-10-23T09:46:38.552Z" }, - { url = "https://files.pythonhosted.org/packages/28/2f/cc27d5f43e023d21fe5c19538e08894db3d7e081cbf582ad5ed366c24446/frozenlist-1.5.0-cp310-cp310-win_amd64.whl", hash = "sha256:189f03b53e64144f90990d29a27ec4f7997d91ed3d01b51fa39d2dbe77540fd4", size = 51622, upload-time = "2024-10-23T09:46:39.513Z" }, - { url = "https://files.pythonhosted.org/packages/79/43/0bed28bf5eb1c9e4301003b74453b8e7aa85fb293b31dde352aac528dafc/frozenlist-1.5.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:fd74520371c3c4175142d02a976aee0b4cb4a7cc912a60586ffd8d5929979b30", size = 94987, upload-time = "2024-10-23T09:46:40.487Z" }, - { url = "https://files.pythonhosted.org/packages/bb/bf/b74e38f09a246e8abbe1e90eb65787ed745ccab6eaa58b9c9308e052323d/frozenlist-1.5.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2f3f7a0fbc219fb4455264cae4d9f01ad41ae6ee8524500f381de64ffaa077d5", size = 54584, upload-time = "2024-10-23T09:46:41.463Z" }, - { url = "https://files.pythonhosted.org/packages/2c/31/ab01375682f14f7613a1ade30149f684c84f9b8823a4391ed950c8285656/frozenlist-1.5.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:f47c9c9028f55a04ac254346e92977bf0f166c483c74b4232bee19a6697e4778", size = 52499, upload-time = "2024-10-23T09:46:42.451Z" }, - { url = "https://files.pythonhosted.org/packages/98/a8/d0ac0b9276e1404f58fec3ab6e90a4f76b778a49373ccaf6a563f100dfbc/frozenlist-1.5.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0996c66760924da6e88922756d99b47512a71cfd45215f3570bf1e0b694c206a", size = 276357, upload-time = "2024-10-23T09:46:44.166Z" }, - { url = "https://files.pythonhosted.org/packages/ad/c9/c7761084fa822f07dac38ac29f841d4587570dd211e2262544aa0b791d21/frozenlist-1.5.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a2fe128eb4edeabe11896cb6af88fca5346059f6c8d807e3b910069f39157869", size = 287516, upload-time = "2024-10-23T09:46:45.369Z" }, - { url = "https://files.pythonhosted.org/packages/a1/ff/cd7479e703c39df7bdab431798cef89dc75010d8aa0ca2514c5b9321db27/frozenlist-1.5.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1a8ea951bbb6cacd492e3948b8da8c502a3f814f5d20935aae74b5df2b19cf3d", size = 283131, upload-time = "2024-10-23T09:46:46.654Z" }, - { url = "https://files.pythonhosted.org/packages/59/a0/370941beb47d237eca4fbf27e4e91389fd68699e6f4b0ebcc95da463835b/frozenlist-1.5.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:de537c11e4aa01d37db0d403b57bd6f0546e71a82347a97c6a9f0dcc532b3a45", size = 261320, upload-time = "2024-10-23T09:46:47.825Z" }, - { url = "https://files.pythonhosted.org/packages/b8/5f/c10123e8d64867bc9b4f2f510a32042a306ff5fcd7e2e09e5ae5100ee333/frozenlist-1.5.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9c2623347b933fcb9095841f1cc5d4ff0b278addd743e0e966cb3d460278840d", size = 274877, upload-time = "2024-10-23T09:46:48.989Z" }, - { url = "https://files.pythonhosted.org/packages/fa/79/38c505601ae29d4348f21706c5d89755ceded02a745016ba2f58bd5f1ea6/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cee6798eaf8b1416ef6909b06f7dc04b60755206bddc599f52232606e18179d3", size = 269592, upload-time = "2024-10-23T09:46:50.235Z" }, - { url = "https://files.pythonhosted.org/packages/19/e2/39f3a53191b8204ba9f0bb574b926b73dd2efba2a2b9d2d730517e8f7622/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:f5f9da7f5dbc00a604fe74aa02ae7c98bcede8a3b8b9666f9f86fc13993bc71a", size = 265934, upload-time = "2024-10-23T09:46:51.829Z" }, - { url = "https://files.pythonhosted.org/packages/d5/c9/3075eb7f7f3a91f1a6b00284af4de0a65a9ae47084930916f5528144c9dd/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:90646abbc7a5d5c7c19461d2e3eeb76eb0b204919e6ece342feb6032c9325ae9", size = 283859, upload-time = "2024-10-23T09:46:52.947Z" }, - { url = "https://files.pythonhosted.org/packages/05/f5/549f44d314c29408b962fa2b0e69a1a67c59379fb143b92a0a065ffd1f0f/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:bdac3c7d9b705d253b2ce370fde941836a5f8b3c5c2b8fd70940a3ea3af7f4f2", size = 287560, upload-time = "2024-10-23T09:46:54.162Z" }, - { url = "https://files.pythonhosted.org/packages/9d/f8/cb09b3c24a3eac02c4c07a9558e11e9e244fb02bf62c85ac2106d1eb0c0b/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:03d33c2ddbc1816237a67f66336616416e2bbb6beb306e5f890f2eb22b959cdf", size = 277150, upload-time = "2024-10-23T09:46:55.361Z" }, - { url = "https://files.pythonhosted.org/packages/37/48/38c2db3f54d1501e692d6fe058f45b6ad1b358d82cd19436efab80cfc965/frozenlist-1.5.0-cp311-cp311-win32.whl", hash = "sha256:237f6b23ee0f44066219dae14c70ae38a63f0440ce6750f868ee08775073f942", size = 45244, upload-time = "2024-10-23T09:46:56.578Z" }, - { url = "https://files.pythonhosted.org/packages/ca/8c/2ddffeb8b60a4bce3b196c32fcc30d8830d4615e7b492ec2071da801b8ad/frozenlist-1.5.0-cp311-cp311-win_amd64.whl", hash = "sha256:0cc974cc93d32c42e7b0f6cf242a6bd941c57c61b618e78b6c0a96cb72788c1d", size = 51634, upload-time = "2024-10-23T09:46:57.6Z" }, - { url = "https://files.pythonhosted.org/packages/79/73/fa6d1a96ab7fd6e6d1c3500700963eab46813847f01ef0ccbaa726181dd5/frozenlist-1.5.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:31115ba75889723431aa9a4e77d5f398f5cf976eea3bdf61749731f62d4a4a21", size = 94026, upload-time = "2024-10-23T09:46:58.601Z" }, - { url = "https://files.pythonhosted.org/packages/ab/04/ea8bf62c8868b8eada363f20ff1b647cf2e93377a7b284d36062d21d81d1/frozenlist-1.5.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:7437601c4d89d070eac8323f121fcf25f88674627505334654fd027b091db09d", size = 54150, upload-time = "2024-10-23T09:46:59.608Z" }, - { url = "https://files.pythonhosted.org/packages/d0/9a/8e479b482a6f2070b26bda572c5e6889bb3ba48977e81beea35b5ae13ece/frozenlist-1.5.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:7948140d9f8ece1745be806f2bfdf390127cf1a763b925c4a805c603df5e697e", size = 51927, upload-time = "2024-10-23T09:47:00.625Z" }, - { url = "https://files.pythonhosted.org/packages/e3/12/2aad87deb08a4e7ccfb33600871bbe8f0e08cb6d8224371387f3303654d7/frozenlist-1.5.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:feeb64bc9bcc6b45c6311c9e9b99406660a9c05ca8a5b30d14a78555088b0b3a", size = 282647, upload-time = "2024-10-23T09:47:01.992Z" }, - { url = "https://files.pythonhosted.org/packages/77/f2/07f06b05d8a427ea0060a9cef6e63405ea9e0d761846b95ef3fb3be57111/frozenlist-1.5.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:683173d371daad49cffb8309779e886e59c2f369430ad28fe715f66d08d4ab1a", size = 289052, upload-time = "2024-10-23T09:47:04.039Z" }, - { url = "https://files.pythonhosted.org/packages/bd/9f/8bf45a2f1cd4aa401acd271b077989c9267ae8463e7c8b1eb0d3f561b65e/frozenlist-1.5.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:7d57d8f702221405a9d9b40f9da8ac2e4a1a8b5285aac6100f3393675f0a85ee", size = 291719, upload-time = "2024-10-23T09:47:05.58Z" }, - { url = "https://files.pythonhosted.org/packages/41/d1/1f20fd05a6c42d3868709b7604c9f15538a29e4f734c694c6bcfc3d3b935/frozenlist-1.5.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:30c72000fbcc35b129cb09956836c7d7abf78ab5416595e4857d1cae8d6251a6", size = 267433, upload-time = "2024-10-23T09:47:07.807Z" }, - { url = "https://files.pythonhosted.org/packages/af/f2/64b73a9bb86f5a89fb55450e97cd5c1f84a862d4ff90d9fd1a73ab0f64a5/frozenlist-1.5.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:000a77d6034fbad9b6bb880f7ec073027908f1b40254b5d6f26210d2dab1240e", size = 283591, upload-time = "2024-10-23T09:47:09.645Z" }, - { url = "https://files.pythonhosted.org/packages/29/e2/ffbb1fae55a791fd6c2938dd9ea779509c977435ba3940b9f2e8dc9d5316/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:5d7f5a50342475962eb18b740f3beecc685a15b52c91f7d975257e13e029eca9", size = 273249, upload-time = "2024-10-23T09:47:10.808Z" }, - { url = "https://files.pythonhosted.org/packages/2e/6e/008136a30798bb63618a114b9321b5971172a5abddff44a100c7edc5ad4f/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:87f724d055eb4785d9be84e9ebf0f24e392ddfad00b3fe036e43f489fafc9039", size = 271075, upload-time = "2024-10-23T09:47:11.938Z" }, - { url = "https://files.pythonhosted.org/packages/ae/f0/4e71e54a026b06724cec9b6c54f0b13a4e9e298cc8db0f82ec70e151f5ce/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:6e9080bb2fb195a046e5177f10d9d82b8a204c0736a97a153c2466127de87784", size = 285398, upload-time = "2024-10-23T09:47:14.071Z" }, - { url = "https://files.pythonhosted.org/packages/4d/36/70ec246851478b1c0b59f11ef8ade9c482ff447c1363c2bd5fad45098b12/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:9b93d7aaa36c966fa42efcaf716e6b3900438632a626fb09c049f6a2f09fc631", size = 294445, upload-time = "2024-10-23T09:47:15.318Z" }, - { url = "https://files.pythonhosted.org/packages/37/e0/47f87544055b3349b633a03c4d94b405956cf2437f4ab46d0928b74b7526/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:52ef692a4bc60a6dd57f507429636c2af8b6046db8b31b18dac02cbc8f507f7f", size = 280569, upload-time = "2024-10-23T09:47:17.149Z" }, - { url = "https://files.pythonhosted.org/packages/f9/7c/490133c160fb6b84ed374c266f42800e33b50c3bbab1652764e6e1fc498a/frozenlist-1.5.0-cp312-cp312-win32.whl", hash = "sha256:29d94c256679247b33a3dc96cce0f93cbc69c23bf75ff715919332fdbb6a32b8", size = 44721, upload-time = "2024-10-23T09:47:19.012Z" }, - { url = "https://files.pythonhosted.org/packages/b1/56/4e45136ffc6bdbfa68c29ca56ef53783ef4c2fd395f7cbf99a2624aa9aaa/frozenlist-1.5.0-cp312-cp312-win_amd64.whl", hash = "sha256:8969190d709e7c48ea386db202d708eb94bdb29207a1f269bab1196ce0dcca1f", size = 51329, upload-time = "2024-10-23T09:47:20.177Z" }, - { url = "https://files.pythonhosted.org/packages/da/3b/915f0bca8a7ea04483622e84a9bd90033bab54bdf485479556c74fd5eaf5/frozenlist-1.5.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:7a1a048f9215c90973402e26c01d1cff8a209e1f1b53f72b95c13db61b00f953", size = 91538, upload-time = "2024-10-23T09:47:21.176Z" }, - { url = "https://files.pythonhosted.org/packages/c7/d1/a7c98aad7e44afe5306a2b068434a5830f1470675f0e715abb86eb15f15b/frozenlist-1.5.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:dd47a5181ce5fcb463b5d9e17ecfdb02b678cca31280639255ce9d0e5aa67af0", size = 52849, upload-time = "2024-10-23T09:47:22.439Z" }, - { url = "https://files.pythonhosted.org/packages/3a/c8/76f23bf9ab15d5f760eb48701909645f686f9c64fbb8982674c241fbef14/frozenlist-1.5.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1431d60b36d15cda188ea222033eec8e0eab488f39a272461f2e6d9e1a8e63c2", size = 50583, upload-time = "2024-10-23T09:47:23.44Z" }, - { url = "https://files.pythonhosted.org/packages/1f/22/462a3dd093d11df623179d7754a3b3269de3b42de2808cddef50ee0f4f48/frozenlist-1.5.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6482a5851f5d72767fbd0e507e80737f9c8646ae7fd303def99bfe813f76cf7f", size = 265636, upload-time = "2024-10-23T09:47:24.82Z" }, - { url = "https://files.pythonhosted.org/packages/80/cf/e075e407fc2ae7328155a1cd7e22f932773c8073c1fc78016607d19cc3e5/frozenlist-1.5.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:44c49271a937625619e862baacbd037a7ef86dd1ee215afc298a417ff3270608", size = 270214, upload-time = "2024-10-23T09:47:26.156Z" }, - { url = "https://files.pythonhosted.org/packages/a1/58/0642d061d5de779f39c50cbb00df49682832923f3d2ebfb0fedf02d05f7f/frozenlist-1.5.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:12f78f98c2f1c2429d42e6a485f433722b0061d5c0b0139efa64f396efb5886b", size = 273905, upload-time = "2024-10-23T09:47:27.741Z" }, - { url = "https://files.pythonhosted.org/packages/ab/66/3fe0f5f8f2add5b4ab7aa4e199f767fd3b55da26e3ca4ce2cc36698e50c4/frozenlist-1.5.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ce3aa154c452d2467487765e3adc730a8c153af77ad84096bc19ce19a2400840", size = 250542, upload-time = "2024-10-23T09:47:28.938Z" }, - { url = "https://files.pythonhosted.org/packages/f6/b8/260791bde9198c87a465224e0e2bb62c4e716f5d198fc3a1dacc4895dbd1/frozenlist-1.5.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9b7dc0c4338e6b8b091e8faf0db3168a37101943e687f373dce00959583f7439", size = 267026, upload-time = "2024-10-23T09:47:30.283Z" }, - { url = "https://files.pythonhosted.org/packages/2e/a4/3d24f88c527f08f8d44ade24eaee83b2627793fa62fa07cbb7ff7a2f7d42/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:45e0896250900b5aa25180f9aec243e84e92ac84bd4a74d9ad4138ef3f5c97de", size = 257690, upload-time = "2024-10-23T09:47:32.388Z" }, - { url = "https://files.pythonhosted.org/packages/de/9a/d311d660420b2beeff3459b6626f2ab4fb236d07afbdac034a4371fe696e/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:561eb1c9579d495fddb6da8959fd2a1fca2c6d060d4113f5844b433fc02f2641", size = 253893, upload-time = "2024-10-23T09:47:34.274Z" }, - { url = "https://files.pythonhosted.org/packages/c6/23/e491aadc25b56eabd0f18c53bb19f3cdc6de30b2129ee0bc39cd387cd560/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:df6e2f325bfee1f49f81aaac97d2aa757c7646534a06f8f577ce184afe2f0a9e", size = 267006, upload-time = "2024-10-23T09:47:35.499Z" }, - { url = "https://files.pythonhosted.org/packages/08/c4/ab918ce636a35fb974d13d666dcbe03969592aeca6c3ab3835acff01f79c/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:140228863501b44b809fb39ec56b5d4071f4d0aa6d216c19cbb08b8c5a7eadb9", size = 276157, upload-time = "2024-10-23T09:47:37.522Z" }, - { url = "https://files.pythonhosted.org/packages/c0/29/3b7a0bbbbe5a34833ba26f686aabfe982924adbdcafdc294a7a129c31688/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:7707a25d6a77f5d27ea7dc7d1fc608aa0a478193823f88511ef5e6b8a48f9d03", size = 264642, upload-time = "2024-10-23T09:47:38.75Z" }, - { url = "https://files.pythonhosted.org/packages/ab/42/0595b3dbffc2e82d7fe658c12d5a5bafcd7516c6bf2d1d1feb5387caa9c1/frozenlist-1.5.0-cp313-cp313-win32.whl", hash = "sha256:31a9ac2b38ab9b5a8933b693db4939764ad3f299fcaa931a3e605bc3460e693c", size = 44914, upload-time = "2024-10-23T09:47:40.145Z" }, - { url = "https://files.pythonhosted.org/packages/17/c4/b7db1206a3fea44bf3b838ca61deb6f74424a8a5db1dd53ecb21da669be6/frozenlist-1.5.0-cp313-cp313-win_amd64.whl", hash = "sha256:11aabdd62b8b9c4b84081a3c246506d1cddd2dd93ff0ad53ede5defec7886b28", size = 51167, upload-time = "2024-10-23T09:47:41.812Z" }, - { url = "https://files.pythonhosted.org/packages/c6/c8/a5be5b7550c10858fcf9b0ea054baccab474da77d37f1e828ce043a3a5d4/frozenlist-1.5.0-py3-none-any.whl", hash = "sha256:d994863bba198a4a518b467bb971c56e1db3f180a25c6cf7bb1949c267f748c3", size = 11901, upload-time = "2024-10-23T09:48:28.851Z" }, +version = "1.7.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/79/b1/b64018016eeb087db503b038296fd782586432b9c077fc5c7839e9cb6ef6/frozenlist-1.7.0.tar.gz", hash = "sha256:2e310d81923c2437ea8670467121cc3e9b0f76d3043cc1d2331d56c7fb7a3a8f", size = 45078, upload-time = "2025-06-09T23:02:35.538Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/af/36/0da0a49409f6b47cc2d060dc8c9040b897b5902a8a4e37d9bc1deb11f680/frozenlist-1.7.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:cc4df77d638aa2ed703b878dd093725b72a824c3c546c076e8fdf276f78ee84a", size = 81304, upload-time = "2025-06-09T22:59:46.226Z" }, + { url = "https://files.pythonhosted.org/packages/77/f0/77c11d13d39513b298e267b22eb6cb559c103d56f155aa9a49097221f0b6/frozenlist-1.7.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:716a9973a2cc963160394f701964fe25012600f3d311f60c790400b00e568b61", size = 47735, upload-time = "2025-06-09T22:59:48.133Z" }, + { url = "https://files.pythonhosted.org/packages/37/12/9d07fa18971a44150593de56b2f2947c46604819976784bcf6ea0d5db43b/frozenlist-1.7.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:a0fd1bad056a3600047fb9462cff4c5322cebc59ebf5d0a3725e0ee78955001d", size = 46775, upload-time = "2025-06-09T22:59:49.564Z" }, + { url = "https://files.pythonhosted.org/packages/70/34/f73539227e06288fcd1f8a76853e755b2b48bca6747e99e283111c18bcd4/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3789ebc19cb811163e70fe2bd354cea097254ce6e707ae42e56f45e31e96cb8e", size = 224644, upload-time = "2025-06-09T22:59:51.35Z" }, + { url = "https://files.pythonhosted.org/packages/fb/68/c1d9c2f4a6e438e14613bad0f2973567586610cc22dcb1e1241da71de9d3/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:af369aa35ee34f132fcfad5be45fbfcde0e3a5f6a1ec0712857f286b7d20cca9", size = 222125, upload-time = "2025-06-09T22:59:52.884Z" }, + { url = "https://files.pythonhosted.org/packages/b9/d0/98e8f9a515228d708344d7c6986752be3e3192d1795f748c24bcf154ad99/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ac64b6478722eeb7a3313d494f8342ef3478dff539d17002f849101b212ef97c", size = 233455, upload-time = "2025-06-09T22:59:54.74Z" }, + { url = "https://files.pythonhosted.org/packages/79/df/8a11bcec5600557f40338407d3e5bea80376ed1c01a6c0910fcfdc4b8993/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f89f65d85774f1797239693cef07ad4c97fdd0639544bad9ac4b869782eb1981", size = 227339, upload-time = "2025-06-09T22:59:56.187Z" }, + { url = "https://files.pythonhosted.org/packages/50/82/41cb97d9c9a5ff94438c63cc343eb7980dac4187eb625a51bdfdb7707314/frozenlist-1.7.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:1073557c941395fdfcfac13eb2456cb8aad89f9de27bae29fabca8e563b12615", size = 212969, upload-time = "2025-06-09T22:59:57.604Z" }, + { url = "https://files.pythonhosted.org/packages/13/47/f9179ee5ee4f55629e4f28c660b3fdf2775c8bfde8f9c53f2de2d93f52a9/frozenlist-1.7.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1ed8d2fa095aae4bdc7fdd80351009a48d286635edffee66bf865e37a9125c50", size = 222862, upload-time = "2025-06-09T22:59:59.498Z" }, + { url = "https://files.pythonhosted.org/packages/1a/52/df81e41ec6b953902c8b7e3a83bee48b195cb0e5ec2eabae5d8330c78038/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:24c34bea555fe42d9f928ba0a740c553088500377448febecaa82cc3e88aa1fa", size = 222492, upload-time = "2025-06-09T23:00:01.026Z" }, + { url = "https://files.pythonhosted.org/packages/84/17/30d6ea87fa95a9408245a948604b82c1a4b8b3e153cea596421a2aef2754/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:69cac419ac6a6baad202c85aaf467b65ac860ac2e7f2ac1686dc40dbb52f6577", size = 238250, upload-time = "2025-06-09T23:00:03.401Z" }, + { url = "https://files.pythonhosted.org/packages/8f/00/ecbeb51669e3c3df76cf2ddd66ae3e48345ec213a55e3887d216eb4fbab3/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:960d67d0611f4c87da7e2ae2eacf7ea81a5be967861e0c63cf205215afbfac59", size = 218720, upload-time = "2025-06-09T23:00:05.282Z" }, + { url = "https://files.pythonhosted.org/packages/1a/c0/c224ce0e0eb31cc57f67742071bb470ba8246623c1823a7530be0e76164c/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:41be2964bd4b15bf575e5daee5a5ce7ed3115320fb3c2b71fca05582ffa4dc9e", size = 232585, upload-time = "2025-06-09T23:00:07.962Z" }, + { url = "https://files.pythonhosted.org/packages/55/3c/34cb694abf532f31f365106deebdeac9e45c19304d83cf7d51ebbb4ca4d1/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:46d84d49e00c9429238a7ce02dc0be8f6d7cd0cd405abd1bebdc991bf27c15bd", size = 234248, upload-time = "2025-06-09T23:00:09.428Z" }, + { url = "https://files.pythonhosted.org/packages/98/c0/2052d8b6cecda2e70bd81299e3512fa332abb6dcd2969b9c80dfcdddbf75/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:15900082e886edb37480335d9d518cec978afc69ccbc30bd18610b7c1b22a718", size = 221621, upload-time = "2025-06-09T23:00:11.32Z" }, + { url = "https://files.pythonhosted.org/packages/c5/bf/7dcebae315436903b1d98ffb791a09d674c88480c158aa171958a3ac07f0/frozenlist-1.7.0-cp310-cp310-win32.whl", hash = "sha256:400ddd24ab4e55014bba442d917203c73b2846391dd42ca5e38ff52bb18c3c5e", size = 39578, upload-time = "2025-06-09T23:00:13.526Z" }, + { url = "https://files.pythonhosted.org/packages/8f/5f/f69818f017fa9a3d24d1ae39763e29b7f60a59e46d5f91b9c6b21622f4cd/frozenlist-1.7.0-cp310-cp310-win_amd64.whl", hash = "sha256:6eb93efb8101ef39d32d50bce242c84bcbddb4f7e9febfa7b524532a239b4464", size = 43830, upload-time = "2025-06-09T23:00:14.98Z" }, + { url = "https://files.pythonhosted.org/packages/34/7e/803dde33760128acd393a27eb002f2020ddb8d99d30a44bfbaab31c5f08a/frozenlist-1.7.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:aa51e147a66b2d74de1e6e2cf5921890de6b0f4820b257465101d7f37b49fb5a", size = 82251, upload-time = "2025-06-09T23:00:16.279Z" }, + { url = "https://files.pythonhosted.org/packages/75/a9/9c2c5760b6ba45eae11334db454c189d43d34a4c0b489feb2175e5e64277/frozenlist-1.7.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:9b35db7ce1cd71d36ba24f80f0c9e7cff73a28d7a74e91fe83e23d27c7828750", size = 48183, upload-time = "2025-06-09T23:00:17.698Z" }, + { url = "https://files.pythonhosted.org/packages/47/be/4038e2d869f8a2da165f35a6befb9158c259819be22eeaf9c9a8f6a87771/frozenlist-1.7.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:34a69a85e34ff37791e94542065c8416c1afbf820b68f720452f636d5fb990cd", size = 47107, upload-time = "2025-06-09T23:00:18.952Z" }, + { url = "https://files.pythonhosted.org/packages/79/26/85314b8a83187c76a37183ceed886381a5f992975786f883472fcb6dc5f2/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4a646531fa8d82c87fe4bb2e596f23173caec9185bfbca5d583b4ccfb95183e2", size = 237333, upload-time = "2025-06-09T23:00:20.275Z" }, + { url = "https://files.pythonhosted.org/packages/1f/fd/e5b64f7d2c92a41639ffb2ad44a6a82f347787abc0c7df5f49057cf11770/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:79b2ffbba483f4ed36a0f236ccb85fbb16e670c9238313709638167670ba235f", size = 231724, upload-time = "2025-06-09T23:00:21.705Z" }, + { url = "https://files.pythonhosted.org/packages/20/fb/03395c0a43a5976af4bf7534759d214405fbbb4c114683f434dfdd3128ef/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a26f205c9ca5829cbf82bb2a84b5c36f7184c4316617d7ef1b271a56720d6b30", size = 245842, upload-time = "2025-06-09T23:00:23.148Z" }, + { url = "https://files.pythonhosted.org/packages/d0/15/c01c8e1dffdac5d9803507d824f27aed2ba76b6ed0026fab4d9866e82f1f/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:bcacfad3185a623fa11ea0e0634aac7b691aa925d50a440f39b458e41c561d98", size = 239767, upload-time = "2025-06-09T23:00:25.103Z" }, + { url = "https://files.pythonhosted.org/packages/14/99/3f4c6fe882c1f5514b6848aa0a69b20cb5e5d8e8f51a339d48c0e9305ed0/frozenlist-1.7.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:72c1b0fe8fe451b34f12dce46445ddf14bd2a5bcad7e324987194dc8e3a74c86", size = 224130, upload-time = "2025-06-09T23:00:27.061Z" }, + { url = "https://files.pythonhosted.org/packages/4d/83/220a374bd7b2aeba9d0725130665afe11de347d95c3620b9b82cc2fcab97/frozenlist-1.7.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:61d1a5baeaac6c0798ff6edfaeaa00e0e412d49946c53fae8d4b8e8b3566c4ae", size = 235301, upload-time = "2025-06-09T23:00:29.02Z" }, + { url = "https://files.pythonhosted.org/packages/03/3c/3e3390d75334a063181625343e8daab61b77e1b8214802cc4e8a1bb678fc/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:7edf5c043c062462f09b6820de9854bf28cc6cc5b6714b383149745e287181a8", size = 234606, upload-time = "2025-06-09T23:00:30.514Z" }, + { url = "https://files.pythonhosted.org/packages/23/1e/58232c19608b7a549d72d9903005e2d82488f12554a32de2d5fb59b9b1ba/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:d50ac7627b3a1bd2dcef6f9da89a772694ec04d9a61b66cf87f7d9446b4a0c31", size = 248372, upload-time = "2025-06-09T23:00:31.966Z" }, + { url = "https://files.pythonhosted.org/packages/c0/a4/e4a567e01702a88a74ce8a324691e62a629bf47d4f8607f24bf1c7216e7f/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:ce48b2fece5aeb45265bb7a58259f45027db0abff478e3077e12b05b17fb9da7", size = 229860, upload-time = "2025-06-09T23:00:33.375Z" }, + { url = "https://files.pythonhosted.org/packages/73/a6/63b3374f7d22268b41a9db73d68a8233afa30ed164c46107b33c4d18ecdd/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:fe2365ae915a1fafd982c146754e1de6ab3478def8a59c86e1f7242d794f97d5", size = 245893, upload-time = "2025-06-09T23:00:35.002Z" }, + { url = "https://files.pythonhosted.org/packages/6d/eb/d18b3f6e64799a79673c4ba0b45e4cfbe49c240edfd03a68be20002eaeaa/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:45a6f2fdbd10e074e8814eb98b05292f27bad7d1883afbe009d96abdcf3bc898", size = 246323, upload-time = "2025-06-09T23:00:36.468Z" }, + { url = "https://files.pythonhosted.org/packages/5a/f5/720f3812e3d06cd89a1d5db9ff6450088b8f5c449dae8ffb2971a44da506/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:21884e23cffabb157a9dd7e353779077bf5b8f9a58e9b262c6caad2ef5f80a56", size = 233149, upload-time = "2025-06-09T23:00:37.963Z" }, + { url = "https://files.pythonhosted.org/packages/69/68/03efbf545e217d5db8446acfd4c447c15b7c8cf4dbd4a58403111df9322d/frozenlist-1.7.0-cp311-cp311-win32.whl", hash = "sha256:284d233a8953d7b24f9159b8a3496fc1ddc00f4db99c324bd5fb5f22d8698ea7", size = 39565, upload-time = "2025-06-09T23:00:39.753Z" }, + { url = "https://files.pythonhosted.org/packages/58/17/fe61124c5c333ae87f09bb67186d65038834a47d974fc10a5fadb4cc5ae1/frozenlist-1.7.0-cp311-cp311-win_amd64.whl", hash = "sha256:387cbfdcde2f2353f19c2f66bbb52406d06ed77519ac7ee21be0232147c2592d", size = 44019, upload-time = "2025-06-09T23:00:40.988Z" }, + { url = "https://files.pythonhosted.org/packages/ef/a2/c8131383f1e66adad5f6ecfcce383d584ca94055a34d683bbb24ac5f2f1c/frozenlist-1.7.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:3dbf9952c4bb0e90e98aec1bd992b3318685005702656bc6f67c1a32b76787f2", size = 81424, upload-time = "2025-06-09T23:00:42.24Z" }, + { url = "https://files.pythonhosted.org/packages/4c/9d/02754159955088cb52567337d1113f945b9e444c4960771ea90eb73de8db/frozenlist-1.7.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:1f5906d3359300b8a9bb194239491122e6cf1444c2efb88865426f170c262cdb", size = 47952, upload-time = "2025-06-09T23:00:43.481Z" }, + { url = "https://files.pythonhosted.org/packages/01/7a/0046ef1bd6699b40acd2067ed6d6670b4db2f425c56980fa21c982c2a9db/frozenlist-1.7.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3dabd5a8f84573c8d10d8859a50ea2dec01eea372031929871368c09fa103478", size = 46688, upload-time = "2025-06-09T23:00:44.793Z" }, + { url = "https://files.pythonhosted.org/packages/d6/a2/a910bafe29c86997363fb4c02069df4ff0b5bc39d33c5198b4e9dd42d8f8/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa57daa5917f1738064f302bf2626281a1cb01920c32f711fbc7bc36111058a8", size = 243084, upload-time = "2025-06-09T23:00:46.125Z" }, + { url = "https://files.pythonhosted.org/packages/64/3e/5036af9d5031374c64c387469bfcc3af537fc0f5b1187d83a1cf6fab1639/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:c193dda2b6d49f4c4398962810fa7d7c78f032bf45572b3e04dd5249dff27e08", size = 233524, upload-time = "2025-06-09T23:00:47.73Z" }, + { url = "https://files.pythonhosted.org/packages/06/39/6a17b7c107a2887e781a48ecf20ad20f1c39d94b2a548c83615b5b879f28/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:bfe2b675cf0aaa6d61bf8fbffd3c274b3c9b7b1623beb3809df8a81399a4a9c4", size = 248493, upload-time = "2025-06-09T23:00:49.742Z" }, + { url = "https://files.pythonhosted.org/packages/be/00/711d1337c7327d88c44d91dd0f556a1c47fb99afc060ae0ef66b4d24793d/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8fc5d5cda37f62b262405cf9652cf0856839c4be8ee41be0afe8858f17f4c94b", size = 244116, upload-time = "2025-06-09T23:00:51.352Z" }, + { url = "https://files.pythonhosted.org/packages/24/fe/74e6ec0639c115df13d5850e75722750adabdc7de24e37e05a40527ca539/frozenlist-1.7.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b0d5ce521d1dd7d620198829b87ea002956e4319002ef0bc8d3e6d045cb4646e", size = 224557, upload-time = "2025-06-09T23:00:52.855Z" }, + { url = "https://files.pythonhosted.org/packages/8d/db/48421f62a6f77c553575201e89048e97198046b793f4a089c79a6e3268bd/frozenlist-1.7.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:488d0a7d6a0008ca0db273c542098a0fa9e7dfaa7e57f70acef43f32b3f69dca", size = 241820, upload-time = "2025-06-09T23:00:54.43Z" }, + { url = "https://files.pythonhosted.org/packages/1d/fa/cb4a76bea23047c8462976ea7b7a2bf53997a0ca171302deae9d6dd12096/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:15a7eaba63983d22c54d255b854e8108e7e5f3e89f647fc854bd77a237e767df", size = 236542, upload-time = "2025-06-09T23:00:56.409Z" }, + { url = "https://files.pythonhosted.org/packages/5d/32/476a4b5cfaa0ec94d3f808f193301debff2ea42288a099afe60757ef6282/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:1eaa7e9c6d15df825bf255649e05bd8a74b04a4d2baa1ae46d9c2d00b2ca2cb5", size = 249350, upload-time = "2025-06-09T23:00:58.468Z" }, + { url = "https://files.pythonhosted.org/packages/8d/ba/9a28042f84a6bf8ea5dbc81cfff8eaef18d78b2a1ad9d51c7bc5b029ad16/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:e4389e06714cfa9d47ab87f784a7c5be91d3934cd6e9a7b85beef808297cc025", size = 225093, upload-time = "2025-06-09T23:01:00.015Z" }, + { url = "https://files.pythonhosted.org/packages/bc/29/3a32959e68f9cf000b04e79ba574527c17e8842e38c91d68214a37455786/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:73bd45e1488c40b63fe5a7df892baf9e2a4d4bb6409a2b3b78ac1c6236178e01", size = 245482, upload-time = "2025-06-09T23:01:01.474Z" }, + { url = "https://files.pythonhosted.org/packages/80/e8/edf2f9e00da553f07f5fa165325cfc302dead715cab6ac8336a5f3d0adc2/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:99886d98e1643269760e5fe0df31e5ae7050788dd288947f7f007209b8c33f08", size = 249590, upload-time = "2025-06-09T23:01:02.961Z" }, + { url = "https://files.pythonhosted.org/packages/1c/80/9a0eb48b944050f94cc51ee1c413eb14a39543cc4f760ed12657a5a3c45a/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:290a172aae5a4c278c6da8a96222e6337744cd9c77313efe33d5670b9f65fc43", size = 237785, upload-time = "2025-06-09T23:01:05.095Z" }, + { url = "https://files.pythonhosted.org/packages/f3/74/87601e0fb0369b7a2baf404ea921769c53b7ae00dee7dcfe5162c8c6dbf0/frozenlist-1.7.0-cp312-cp312-win32.whl", hash = "sha256:426c7bc70e07cfebc178bc4c2bf2d861d720c4fff172181eeb4a4c41d4ca2ad3", size = 39487, upload-time = "2025-06-09T23:01:06.54Z" }, + { url = "https://files.pythonhosted.org/packages/0b/15/c026e9a9fc17585a9d461f65d8593d281fedf55fbf7eb53f16c6df2392f9/frozenlist-1.7.0-cp312-cp312-win_amd64.whl", hash = "sha256:563b72efe5da92e02eb68c59cb37205457c977aa7a449ed1b37e6939e5c47c6a", size = 43874, upload-time = "2025-06-09T23:01:07.752Z" }, + { url = "https://files.pythonhosted.org/packages/24/90/6b2cebdabdbd50367273c20ff6b57a3dfa89bd0762de02c3a1eb42cb6462/frozenlist-1.7.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:ee80eeda5e2a4e660651370ebffd1286542b67e268aa1ac8d6dbe973120ef7ee", size = 79791, upload-time = "2025-06-09T23:01:09.368Z" }, + { url = "https://files.pythonhosted.org/packages/83/2e/5b70b6a3325363293fe5fc3ae74cdcbc3e996c2a11dde2fd9f1fb0776d19/frozenlist-1.7.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:d1a81c85417b914139e3a9b995d4a1c84559afc839a93cf2cb7f15e6e5f6ed2d", size = 47165, upload-time = "2025-06-09T23:01:10.653Z" }, + { url = "https://files.pythonhosted.org/packages/f4/25/a0895c99270ca6966110f4ad98e87e5662eab416a17e7fd53c364bf8b954/frozenlist-1.7.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:cbb65198a9132ebc334f237d7b0df163e4de83fb4f2bdfe46c1e654bdb0c5d43", size = 45881, upload-time = "2025-06-09T23:01:12.296Z" }, + { url = "https://files.pythonhosted.org/packages/19/7c/71bb0bbe0832793c601fff68cd0cf6143753d0c667f9aec93d3c323f4b55/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dab46c723eeb2c255a64f9dc05b8dd601fde66d6b19cdb82b2e09cc6ff8d8b5d", size = 232409, upload-time = "2025-06-09T23:01:13.641Z" }, + { url = "https://files.pythonhosted.org/packages/c0/45/ed2798718910fe6eb3ba574082aaceff4528e6323f9a8570be0f7028d8e9/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:6aeac207a759d0dedd2e40745575ae32ab30926ff4fa49b1635def65806fddee", size = 225132, upload-time = "2025-06-09T23:01:15.264Z" }, + { url = "https://files.pythonhosted.org/packages/ba/e2/8417ae0f8eacb1d071d4950f32f229aa6bf68ab69aab797b72a07ea68d4f/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:bd8c4e58ad14b4fa7802b8be49d47993182fdd4023393899632c88fd8cd994eb", size = 237638, upload-time = "2025-06-09T23:01:16.752Z" }, + { url = "https://files.pythonhosted.org/packages/f8/b7/2ace5450ce85f2af05a871b8c8719b341294775a0a6c5585d5e6170f2ce7/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:04fb24d104f425da3540ed83cbfc31388a586a7696142004c577fa61c6298c3f", size = 233539, upload-time = "2025-06-09T23:01:18.202Z" }, + { url = "https://files.pythonhosted.org/packages/46/b9/6989292c5539553dba63f3c83dc4598186ab2888f67c0dc1d917e6887db6/frozenlist-1.7.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6a5c505156368e4ea6b53b5ac23c92d7edc864537ff911d2fb24c140bb175e60", size = 215646, upload-time = "2025-06-09T23:01:19.649Z" }, + { url = "https://files.pythonhosted.org/packages/72/31/bc8c5c99c7818293458fe745dab4fd5730ff49697ccc82b554eb69f16a24/frozenlist-1.7.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8bd7eb96a675f18aa5c553eb7ddc24a43c8c18f22e1f9925528128c052cdbe00", size = 232233, upload-time = "2025-06-09T23:01:21.175Z" }, + { url = "https://files.pythonhosted.org/packages/59/52/460db4d7ba0811b9ccb85af996019f5d70831f2f5f255f7cc61f86199795/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:05579bf020096fe05a764f1f84cd104a12f78eaab68842d036772dc6d4870b4b", size = 227996, upload-time = "2025-06-09T23:01:23.098Z" }, + { url = "https://files.pythonhosted.org/packages/ba/c9/f4b39e904c03927b7ecf891804fd3b4df3db29b9e487c6418e37988d6e9d/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:376b6222d114e97eeec13d46c486facd41d4f43bab626b7c3f6a8b4e81a5192c", size = 242280, upload-time = "2025-06-09T23:01:24.808Z" }, + { url = "https://files.pythonhosted.org/packages/b8/33/3f8d6ced42f162d743e3517781566b8481322be321b486d9d262adf70bfb/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:0aa7e176ebe115379b5b1c95b4096fb1c17cce0847402e227e712c27bdb5a949", size = 217717, upload-time = "2025-06-09T23:01:26.28Z" }, + { url = "https://files.pythonhosted.org/packages/3e/e8/ad683e75da6ccef50d0ab0c2b2324b32f84fc88ceee778ed79b8e2d2fe2e/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:3fbba20e662b9c2130dc771e332a99eff5da078b2b2648153a40669a6d0e36ca", size = 236644, upload-time = "2025-06-09T23:01:27.887Z" }, + { url = "https://files.pythonhosted.org/packages/b2/14/8d19ccdd3799310722195a72ac94ddc677541fb4bef4091d8e7775752360/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:f3f4410a0a601d349dd406b5713fec59b4cee7e71678d5b17edda7f4655a940b", size = 238879, upload-time = "2025-06-09T23:01:29.524Z" }, + { url = "https://files.pythonhosted.org/packages/ce/13/c12bf657494c2fd1079a48b2db49fa4196325909249a52d8f09bc9123fd7/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e2cdfaaec6a2f9327bf43c933c0319a7c429058e8537c508964a133dffee412e", size = 232502, upload-time = "2025-06-09T23:01:31.287Z" }, + { url = "https://files.pythonhosted.org/packages/d7/8b/e7f9dfde869825489382bc0d512c15e96d3964180c9499efcec72e85db7e/frozenlist-1.7.0-cp313-cp313-win32.whl", hash = "sha256:5fc4df05a6591c7768459caba1b342d9ec23fa16195e744939ba5914596ae3e1", size = 39169, upload-time = "2025-06-09T23:01:35.503Z" }, + { url = "https://files.pythonhosted.org/packages/35/89/a487a98d94205d85745080a37860ff5744b9820a2c9acbcdd9440bfddf98/frozenlist-1.7.0-cp313-cp313-win_amd64.whl", hash = "sha256:52109052b9791a3e6b5d1b65f4b909703984b770694d3eb64fad124c835d7cba", size = 43219, upload-time = "2025-06-09T23:01:36.784Z" }, + { url = "https://files.pythonhosted.org/packages/56/d5/5c4cf2319a49eddd9dd7145e66c4866bdc6f3dbc67ca3d59685149c11e0d/frozenlist-1.7.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:a6f86e4193bb0e235ef6ce3dde5cbabed887e0b11f516ce8a0f4d3b33078ec2d", size = 84345, upload-time = "2025-06-09T23:01:38.295Z" }, + { url = "https://files.pythonhosted.org/packages/a4/7d/ec2c1e1dc16b85bc9d526009961953df9cec8481b6886debb36ec9107799/frozenlist-1.7.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:82d664628865abeb32d90ae497fb93df398a69bb3434463d172b80fc25b0dd7d", size = 48880, upload-time = "2025-06-09T23:01:39.887Z" }, + { url = "https://files.pythonhosted.org/packages/69/86/f9596807b03de126e11e7d42ac91e3d0b19a6599c714a1989a4e85eeefc4/frozenlist-1.7.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:912a7e8375a1c9a68325a902f3953191b7b292aa3c3fb0d71a216221deca460b", size = 48498, upload-time = "2025-06-09T23:01:41.318Z" }, + { url = "https://files.pythonhosted.org/packages/5e/cb/df6de220f5036001005f2d726b789b2c0b65f2363b104bbc16f5be8084f8/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9537c2777167488d539bc5de2ad262efc44388230e5118868e172dd4a552b146", size = 292296, upload-time = "2025-06-09T23:01:42.685Z" }, + { url = "https://files.pythonhosted.org/packages/83/1f/de84c642f17c8f851a2905cee2dae401e5e0daca9b5ef121e120e19aa825/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:f34560fb1b4c3e30ba35fa9a13894ba39e5acfc5f60f57d8accde65f46cc5e74", size = 273103, upload-time = "2025-06-09T23:01:44.166Z" }, + { url = "https://files.pythonhosted.org/packages/88/3c/c840bfa474ba3fa13c772b93070893c6e9d5c0350885760376cbe3b6c1b3/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:acd03d224b0175f5a850edc104ac19040d35419eddad04e7cf2d5986d98427f1", size = 292869, upload-time = "2025-06-09T23:01:45.681Z" }, + { url = "https://files.pythonhosted.org/packages/a6/1c/3efa6e7d5a39a1d5ef0abeb51c48fb657765794a46cf124e5aca2c7a592c/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f2038310bc582f3d6a09b3816ab01737d60bf7b1ec70f5356b09e84fb7408ab1", size = 291467, upload-time = "2025-06-09T23:01:47.234Z" }, + { url = "https://files.pythonhosted.org/packages/4f/00/d5c5e09d4922c395e2f2f6b79b9a20dab4b67daaf78ab92e7729341f61f6/frozenlist-1.7.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b8c05e4c8e5f36e5e088caa1bf78a687528f83c043706640a92cb76cd6999384", size = 266028, upload-time = "2025-06-09T23:01:48.819Z" }, + { url = "https://files.pythonhosted.org/packages/4e/27/72765be905619dfde25a7f33813ac0341eb6b076abede17a2e3fbfade0cb/frozenlist-1.7.0-cp313-cp313t-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:765bb588c86e47d0b68f23c1bee323d4b703218037765dcf3f25c838c6fecceb", size = 284294, upload-time = "2025-06-09T23:01:50.394Z" }, + { url = "https://files.pythonhosted.org/packages/88/67/c94103a23001b17808eb7dd1200c156bb69fb68e63fcf0693dde4cd6228c/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:32dc2e08c67d86d0969714dd484fd60ff08ff81d1a1e40a77dd34a387e6ebc0c", size = 281898, upload-time = "2025-06-09T23:01:52.234Z" }, + { url = "https://files.pythonhosted.org/packages/42/34/a3e2c00c00f9e2a9db5653bca3fec306349e71aff14ae45ecc6d0951dd24/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:c0303e597eb5a5321b4de9c68e9845ac8f290d2ab3f3e2c864437d3c5a30cd65", size = 290465, upload-time = "2025-06-09T23:01:53.788Z" }, + { url = "https://files.pythonhosted.org/packages/bb/73/f89b7fbce8b0b0c095d82b008afd0590f71ccb3dee6eee41791cf8cd25fd/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:a47f2abb4e29b3a8d0b530f7c3598badc6b134562b1a5caee867f7c62fee51e3", size = 266385, upload-time = "2025-06-09T23:01:55.769Z" }, + { url = "https://files.pythonhosted.org/packages/cd/45/e365fdb554159462ca12df54bc59bfa7a9a273ecc21e99e72e597564d1ae/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:3d688126c242a6fabbd92e02633414d40f50bb6002fa4cf995a1d18051525657", size = 288771, upload-time = "2025-06-09T23:01:57.4Z" }, + { url = "https://files.pythonhosted.org/packages/00/11/47b6117002a0e904f004d70ec5194fe9144f117c33c851e3d51c765962d0/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:4e7e9652b3d367c7bd449a727dc79d5043f48b88d0cbfd4f9f1060cf2b414104", size = 288206, upload-time = "2025-06-09T23:01:58.936Z" }, + { url = "https://files.pythonhosted.org/packages/40/37/5f9f3c3fd7f7746082ec67bcdc204db72dad081f4f83a503d33220a92973/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:1a85e345b4c43db8b842cab1feb41be5cc0b10a1830e6295b69d7310f99becaf", size = 282620, upload-time = "2025-06-09T23:02:00.493Z" }, + { url = "https://files.pythonhosted.org/packages/0b/31/8fbc5af2d183bff20f21aa743b4088eac4445d2bb1cdece449ae80e4e2d1/frozenlist-1.7.0-cp313-cp313t-win32.whl", hash = "sha256:3a14027124ddb70dfcee5148979998066897e79f89f64b13328595c4bdf77c81", size = 43059, upload-time = "2025-06-09T23:02:02.072Z" }, + { url = "https://files.pythonhosted.org/packages/bb/ed/41956f52105b8dbc26e457c5705340c67c8cc2b79f394b79bffc09d0e938/frozenlist-1.7.0-cp313-cp313t-win_amd64.whl", hash = "sha256:3bf8010d71d4507775f658e9823210b7427be36625b387221642725b515dcf3e", size = 47516, upload-time = "2025-06-09T23:02:03.779Z" }, + { url = "https://files.pythonhosted.org/packages/ee/45/b82e3c16be2182bff01179db177fe144d58b5dc787a7d4492c6ed8b9317f/frozenlist-1.7.0-py3-none-any.whl", hash = "sha256:9a5af342e34f7e97caf8c995864c7a396418ae2859cc6fdf1b1073020d516a7e", size = 13106, upload-time = "2025-06-09T23:02:34.204Z" }, ] [[package]] name = "fsspec" -version = "2025.3.0" +version = "2025.7.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/34/f4/5721faf47b8c499e776bc34c6a8fc17efdf7fdef0b00f398128bc5dcb4ac/fsspec-2025.3.0.tar.gz", hash = "sha256:a935fd1ea872591f2b5148907d103488fc523295e6c64b835cfad8c3eca44972", size = 298491, upload-time = "2025-03-07T21:47:56.461Z" } +sdist = { url = "https://files.pythonhosted.org/packages/8b/02/0835e6ab9cfc03916fe3f78c0956cfcdb6ff2669ffa6651065d5ebf7fc98/fsspec-2025.7.0.tar.gz", hash = "sha256:786120687ffa54b8283d942929540d8bc5ccfa820deb555a2b5d0ed2b737bf58", size = 304432, upload-time = "2025-07-15T16:05:21.19Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/56/53/eb690efa8513166adef3e0669afd31e95ffde69fb3c52ec2ac7223ed6018/fsspec-2025.3.0-py3-none-any.whl", hash = "sha256:efb87af3efa9103f94ca91a7f8cb7a4df91af9f74fc106c9c7ea0efd7277c1b3", size = 193615, upload-time = "2025-03-07T21:47:54.809Z" }, + { url = "https://files.pythonhosted.org/packages/2f/e0/014d5d9d7a4564cf1c40b5039bc882db69fd881111e03ab3657ac0b218e2/fsspec-2025.7.0-py3-none-any.whl", hash = "sha256:8b012e39f63c7d5f10474de957f3ab793b47b45ae7d39f2fb735f8bbe25c0e21", size = 199597, upload-time = "2025-07-15T16:05:19.529Z" }, ] [package.optional-dependencies] @@ -782,14 +961,167 @@ wheels = [ [[package]] name = "gitpython" -version = "3.1.44" +version = "3.1.45" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "gitdb" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/c0/89/37df0b71473153574a5cdef8f242de422a0f5d26d7a9e231e6f169b4ad14/gitpython-3.1.44.tar.gz", hash = "sha256:c87e30b26253bf5418b01b0660f818967f3c503193838337fe5e573331249269", size = 214196, upload-time = "2025-01-02T07:32:43.59Z" } +sdist = { url = "https://files.pythonhosted.org/packages/9a/c8/dd58967d119baab745caec2f9d853297cec1989ec1d63f677d3880632b88/gitpython-3.1.45.tar.gz", hash = "sha256:85b0ee964ceddf211c41b9f27a49086010a190fd8132a24e21f362a4b36a791c", size = 215076, upload-time = "2025-07-24T03:45:54.871Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/01/61/d4b89fec821f72385526e1b9d9a3a0385dda4a72b206d28049e2c7cd39b8/gitpython-3.1.45-py3-none-any.whl", hash = "sha256:8908cb2e02fb3b93b7eb0f2827125cb699869470432cc885f019b8fd0fccff77", size = 208168, upload-time = "2025-07-24T03:45:52.517Z" }, +] + +[[package]] +name = "google-api-core" +version = "2.25.1" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "google-auth" }, + { name = "googleapis-common-protos" }, + { name = "proto-plus" }, + { name = "protobuf" }, + { name = "requests" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/dc/21/e9d043e88222317afdbdb567165fdbc3b0aad90064c7e0c9eb0ad9955ad8/google_api_core-2.25.1.tar.gz", hash = "sha256:d2aaa0b13c78c61cb3f4282c464c046e45fbd75755683c9c525e6e8f7ed0a5e8", size = 165443, upload-time = "2025-06-12T20:52:20.439Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/1d/9a/4114a9057db2f1462d5c8f8390ab7383925fe1ac012eaa42402ad65c2963/GitPython-3.1.44-py3-none-any.whl", hash = "sha256:9e0e10cda9bed1ee64bc9a6de50e7e38a9c9943241cd7f585f6df3ed28011110", size = 207599, upload-time = "2025-01-02T07:32:40.731Z" }, + { url = "https://files.pythonhosted.org/packages/14/4b/ead00905132820b623732b175d66354e9d3e69fcf2a5dcdab780664e7896/google_api_core-2.25.1-py3-none-any.whl", hash = "sha256:8a2a56c1fef82987a524371f99f3bd0143702fecc670c72e600c1cda6bf8dbb7", size = 160807, upload-time = "2025-06-12T20:52:19.334Z" }, +] + +[[package]] +name = "google-auth" +version = "2.40.3" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "cachetools" }, + { name = "pyasn1-modules" }, + { name = "rsa" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/9e/9b/e92ef23b84fa10a64ce4831390b7a4c2e53c0132568d99d4ae61d04c8855/google_auth-2.40.3.tar.gz", hash = "sha256:500c3a29adedeb36ea9cf24b8d10858e152f2412e3ca37829b3fa18e33d63b77", size = 281029, upload-time = "2025-06-04T18:04:57.577Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/17/63/b19553b658a1692443c62bd07e5868adaa0ad746a0751ba62c59568cd45b/google_auth-2.40.3-py2.py3-none-any.whl", hash = "sha256:1370d4593e86213563547f97a92752fc658456fe4514c809544f330fed45a7ca", size = 216137, upload-time = "2025-06-04T18:04:55.573Z" }, +] + +[[package]] +name = "google-cloud-core" +version = "2.4.3" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "google-api-core" }, + { name = "google-auth" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/d6/b8/2b53838d2acd6ec6168fd284a990c76695e84c65deee79c9f3a4276f6b4f/google_cloud_core-2.4.3.tar.gz", hash = "sha256:1fab62d7102844b278fe6dead3af32408b1df3eb06f5c7e8634cbd40edc4da53", size = 35861, upload-time = "2025-03-10T21:05:38.948Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/40/86/bda7241a8da2d28a754aad2ba0f6776e35b67e37c36ae0c45d49370f1014/google_cloud_core-2.4.3-py2.py3-none-any.whl", hash = "sha256:5130f9f4c14b4fafdff75c79448f9495cfade0d8775facf1b09c3bf67e027f6e", size = 29348, upload-time = "2025-03-10T21:05:37.785Z" }, +] + +[[package]] +name = "google-cloud-storage" +version = "2.19.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "google-api-core" }, + { name = "google-auth" }, + { name = "google-cloud-core" }, + { name = "google-crc32c" }, + { name = "google-resumable-media" }, + { name = "requests" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/36/76/4d965702e96bb67976e755bed9828fa50306dca003dbee08b67f41dd265e/google_cloud_storage-2.19.0.tar.gz", hash = "sha256:cd05e9e7191ba6cb68934d8eb76054d9be4562aa89dbc4236feee4d7d51342b2", size = 5535488, upload-time = "2024-12-05T01:35:06.49Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/d5/94/6db383d8ee1adf45dc6c73477152b82731fa4c4a46d9c1932cc8757e0fd4/google_cloud_storage-2.19.0-py2.py3-none-any.whl", hash = "sha256:aeb971b5c29cf8ab98445082cbfe7b161a1f48ed275822f59ed3f1524ea54fba", size = 131787, upload-time = "2024-12-05T01:35:04.736Z" }, +] + +[[package]] +name = "google-crc32c" +version = "1.7.1" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/19/ae/87802e6d9f9d69adfaedfcfd599266bf386a54d0be058b532d04c794f76d/google_crc32c-1.7.1.tar.gz", hash = "sha256:2bff2305f98846f3e825dbeec9ee406f89da7962accdb29356e4eadc251bd472", size = 14495, upload-time = "2025-03-26T14:29:13.32Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/eb/69/b1b05cf415df0d86691d6a8b4b7e60ab3a6fb6efb783ee5cd3ed1382bfd3/google_crc32c-1.7.1-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:b07d48faf8292b4db7c3d64ab86f950c2e94e93a11fd47271c28ba458e4a0d76", size = 30467, upload-time = "2025-03-26T14:31:11.92Z" }, + { url = "https://files.pythonhosted.org/packages/44/3d/92f8928ecd671bd5b071756596971c79d252d09b835cdca5a44177fa87aa/google_crc32c-1.7.1-cp310-cp310-macosx_12_0_x86_64.whl", hash = "sha256:7cc81b3a2fbd932a4313eb53cc7d9dde424088ca3a0337160f35d91826880c1d", size = 30311, upload-time = "2025-03-26T14:53:14.161Z" }, + { url = "https://files.pythonhosted.org/packages/33/42/c2d15a73df79d45ed6b430b9e801d0bd8e28ac139a9012d7d58af50a385d/google_crc32c-1.7.1-cp310-cp310-manylinux_2_12_x86_64.manylinux2010_x86_64.whl", hash = "sha256:1c67ca0a1f5b56162951a9dae987988679a7db682d6f97ce0f6381ebf0fbea4c", size = 37889, upload-time = "2025-03-26T14:41:27.83Z" }, + { url = "https://files.pythonhosted.org/packages/57/ea/ac59c86a3c694afd117bb669bde32aaf17d0de4305d01d706495f09cbf19/google_crc32c-1.7.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:fc5319db92daa516b653600794d5b9f9439a9a121f3e162f94b0e1891c7933cb", size = 33028, upload-time = "2025-03-26T14:41:29.141Z" }, + { url = "https://files.pythonhosted.org/packages/60/44/87e77e8476767a4a93f6cf271157c6d948eacec63688c093580af13b04be/google_crc32c-1.7.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:dcdf5a64adb747610140572ed18d011896e3b9ae5195f2514b7ff678c80f1603", size = 38026, upload-time = "2025-03-26T14:41:29.921Z" }, + { url = "https://files.pythonhosted.org/packages/c8/bf/21ac7bb305cd7c1a6de9c52f71db0868e104a5b573a4977cd9d0ff830f82/google_crc32c-1.7.1-cp310-cp310-win_amd64.whl", hash = "sha256:754561c6c66e89d55754106739e22fdaa93fafa8da7221b29c8b8e8270c6ec8a", size = 33476, upload-time = "2025-03-26T14:29:09.086Z" }, + { url = "https://files.pythonhosted.org/packages/f7/94/220139ea87822b6fdfdab4fb9ba81b3fff7ea2c82e2af34adc726085bffc/google_crc32c-1.7.1-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:6fbab4b935989e2c3610371963ba1b86afb09537fd0c633049be82afe153ac06", size = 30468, upload-time = "2025-03-26T14:32:52.215Z" }, + { url = "https://files.pythonhosted.org/packages/94/97/789b23bdeeb9d15dc2904660463ad539d0318286d7633fe2760c10ed0c1c/google_crc32c-1.7.1-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:ed66cbe1ed9cbaaad9392b5259b3eba4a9e565420d734e6238813c428c3336c9", size = 30313, upload-time = "2025-03-26T14:57:38.758Z" }, + { url = "https://files.pythonhosted.org/packages/81/b8/976a2b843610c211e7ccb3e248996a61e87dbb2c09b1499847e295080aec/google_crc32c-1.7.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ee6547b657621b6cbed3562ea7826c3e11cab01cd33b74e1f677690652883e77", size = 33048, upload-time = "2025-03-26T14:41:30.679Z" }, + { url = "https://files.pythonhosted.org/packages/c9/16/a3842c2cf591093b111d4a5e2bfb478ac6692d02f1b386d2a33283a19dc9/google_crc32c-1.7.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d68e17bad8f7dd9a49181a1f5a8f4b251c6dbc8cc96fb79f1d321dfd57d66f53", size = 32669, upload-time = "2025-03-26T14:41:31.432Z" }, + { url = "https://files.pythonhosted.org/packages/04/17/ed9aba495916fcf5fe4ecb2267ceb851fc5f273c4e4625ae453350cfd564/google_crc32c-1.7.1-cp311-cp311-win_amd64.whl", hash = "sha256:6335de12921f06e1f774d0dd1fbea6bf610abe0887a1638f64d694013138be5d", size = 33476, upload-time = "2025-03-26T14:29:10.211Z" }, + { url = "https://files.pythonhosted.org/packages/dd/b7/787e2453cf8639c94b3d06c9d61f512234a82e1d12d13d18584bd3049904/google_crc32c-1.7.1-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:2d73a68a653c57281401871dd4aeebbb6af3191dcac751a76ce430df4d403194", size = 30470, upload-time = "2025-03-26T14:34:31.655Z" }, + { url = "https://files.pythonhosted.org/packages/ed/b4/6042c2b0cbac3ec3a69bb4c49b28d2f517b7a0f4a0232603c42c58e22b44/google_crc32c-1.7.1-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:22beacf83baaf59f9d3ab2bbb4db0fb018da8e5aebdce07ef9f09fce8220285e", size = 30315, upload-time = "2025-03-26T15:01:54.634Z" }, + { url = "https://files.pythonhosted.org/packages/29/ad/01e7a61a5d059bc57b702d9ff6a18b2585ad97f720bd0a0dbe215df1ab0e/google_crc32c-1.7.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:19eafa0e4af11b0a4eb3974483d55d2d77ad1911e6cf6f832e1574f6781fd337", size = 33180, upload-time = "2025-03-26T14:41:32.168Z" }, + { url = "https://files.pythonhosted.org/packages/3b/a5/7279055cf004561894ed3a7bfdf5bf90a53f28fadd01af7cd166e88ddf16/google_crc32c-1.7.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b6d86616faaea68101195c6bdc40c494e4d76f41e07a37ffdef270879c15fb65", size = 32794, upload-time = "2025-03-26T14:41:33.264Z" }, + { url = "https://files.pythonhosted.org/packages/0f/d6/77060dbd140c624e42ae3ece3df53b9d811000729a5c821b9fd671ceaac6/google_crc32c-1.7.1-cp312-cp312-win_amd64.whl", hash = "sha256:b7491bdc0c7564fcf48c0179d2048ab2f7c7ba36b84ccd3a3e1c3f7a72d3bba6", size = 33477, upload-time = "2025-03-26T14:29:10.94Z" }, + { url = "https://files.pythonhosted.org/packages/8b/72/b8d785e9184ba6297a8620c8a37cf6e39b81a8ca01bb0796d7cbb28b3386/google_crc32c-1.7.1-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:df8b38bdaf1629d62d51be8bdd04888f37c451564c2042d36e5812da9eff3c35", size = 30467, upload-time = "2025-03-26T14:36:06.909Z" }, + { url = "https://files.pythonhosted.org/packages/34/25/5f18076968212067c4e8ea95bf3b69669f9fc698476e5f5eb97d5b37999f/google_crc32c-1.7.1-cp313-cp313-macosx_12_0_x86_64.whl", hash = "sha256:e42e20a83a29aa2709a0cf271c7f8aefaa23b7ab52e53b322585297bb94d4638", size = 30309, upload-time = "2025-03-26T15:06:15.318Z" }, + { url = "https://files.pythonhosted.org/packages/92/83/9228fe65bf70e93e419f38bdf6c5ca5083fc6d32886ee79b450ceefd1dbd/google_crc32c-1.7.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:905a385140bf492ac300026717af339790921f411c0dfd9aa5a9e69a08ed32eb", size = 33133, upload-time = "2025-03-26T14:41:34.388Z" }, + { url = "https://files.pythonhosted.org/packages/c3/ca/1ea2fd13ff9f8955b85e7956872fdb7050c4ace8a2306a6d177edb9cf7fe/google_crc32c-1.7.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6b211ddaf20f7ebeec5c333448582c224a7c90a9d98826fbab82c0ddc11348e6", size = 32773, upload-time = "2025-03-26T14:41:35.19Z" }, + { url = "https://files.pythonhosted.org/packages/89/32/a22a281806e3ef21b72db16f948cad22ec68e4bdd384139291e00ff82fe2/google_crc32c-1.7.1-cp313-cp313-win_amd64.whl", hash = "sha256:0f99eaa09a9a7e642a61e06742856eec8b19fc0037832e03f941fe7cf0c8e4db", size = 33475, upload-time = "2025-03-26T14:29:11.771Z" }, + { url = "https://files.pythonhosted.org/packages/b8/c5/002975aff514e57fc084ba155697a049b3f9b52225ec3bc0f542871dd524/google_crc32c-1.7.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32d1da0d74ec5634a05f53ef7df18fc646666a25efaaca9fc7dcfd4caf1d98c3", size = 33243, upload-time = "2025-03-26T14:41:35.975Z" }, + { url = "https://files.pythonhosted.org/packages/61/cb/c585282a03a0cea70fcaa1bf55d5d702d0f2351094d663ec3be1c6c67c52/google_crc32c-1.7.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e10554d4abc5238823112c2ad7e4560f96c7bf3820b202660373d769d9e6e4c9", size = 32870, upload-time = "2025-03-26T14:41:37.08Z" }, + { url = "https://files.pythonhosted.org/packages/0b/43/31e57ce04530794917dfe25243860ec141de9fadf4aa9783dffe7dac7c39/google_crc32c-1.7.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a8e9afc74168b0b2232fb32dd202c93e46b7d5e4bf03e66ba5dc273bb3559589", size = 28242, upload-time = "2025-03-26T14:41:42.858Z" }, + { url = "https://files.pythonhosted.org/packages/eb/f3/8b84cd4e0ad111e63e30eb89453f8dd308e3ad36f42305cf8c202461cdf0/google_crc32c-1.7.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fa8136cc14dd27f34a3221c0f16fd42d8a40e4778273e61a3c19aedaa44daf6b", size = 28049, upload-time = "2025-03-26T14:41:44.651Z" }, + { url = "https://files.pythonhosted.org/packages/16/1b/1693372bf423ada422f80fd88260dbfd140754adb15cbc4d7e9a68b1cb8e/google_crc32c-1.7.1-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:85fef7fae11494e747c9fd1359a527e5970fc9603c90764843caabd3a16a0a48", size = 28241, upload-time = "2025-03-26T14:41:45.898Z" }, + { url = "https://files.pythonhosted.org/packages/fd/3c/2a19a60a473de48717b4efb19398c3f914795b64a96cf3fbe82588044f78/google_crc32c-1.7.1-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6efb97eb4369d52593ad6f75e7e10d053cf00c48983f7a973105bc70b0ac4d82", size = 28048, upload-time = "2025-03-26T14:41:46.696Z" }, +] + +[[package]] +name = "google-resumable-media" +version = "2.7.2" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "google-crc32c" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/58/5a/0efdc02665dca14e0837b62c8a1a93132c264bd02054a15abb2218afe0ae/google_resumable_media-2.7.2.tar.gz", hash = "sha256:5280aed4629f2b60b847b0d42f9857fd4935c11af266744df33d8074cae92fe0", size = 2163099, upload-time = "2024-08-07T22:20:38.555Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/82/35/b8d3baf8c46695858cb9d8835a53baa1eeb9906ddaf2f728a5f5b640fd1e/google_resumable_media-2.7.2-py2.py3-none-any.whl", hash = "sha256:3ce7551e9fe6d99e9a126101d2536612bb73486721951e9562fee0f90c6ababa", size = 81251, upload-time = "2024-08-07T22:20:36.409Z" }, +] + +[[package]] +name = "googleapis-common-protos" +version = "1.70.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "protobuf" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/39/24/33db22342cf4a2ea27c9955e6713140fedd51e8b141b5ce5260897020f1a/googleapis_common_protos-1.70.0.tar.gz", hash = "sha256:0e1b44e0ea153e6594f9f394fef15193a68aaaea2d843f83e2742717ca753257", size = 145903, upload-time = "2025-04-14T10:17:02.924Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/86/f1/62a193f0227cf15a920390abe675f386dec35f7ae3ffe6da582d3ade42c7/googleapis_common_protos-1.70.0-py3-none-any.whl", hash = "sha256:b8bfcca8c25a2bb253e0e0b0adaf8c00773e5e6af6fd92397576680b807e0fd8", size = 294530, upload-time = "2025-04-14T10:17:01.271Z" }, +] + +[[package]] +name = "hf-xet" +version = "1.1.5" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/ed/d4/7685999e85945ed0d7f0762b686ae7015035390de1161dcea9d5276c134c/hf_xet-1.1.5.tar.gz", hash = "sha256:69ebbcfd9ec44fdc2af73441619eeb06b94ee34511bbcf57cd423820090f5694", size = 495969, upload-time = "2025-06-20T21:48:38.007Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/00/89/a1119eebe2836cb25758e7661d6410d3eae982e2b5e974bcc4d250be9012/hf_xet-1.1.5-cp37-abi3-macosx_10_12_x86_64.whl", hash = "sha256:f52c2fa3635b8c37c7764d8796dfa72706cc4eded19d638331161e82b0792e23", size = 2687929, upload-time = "2025-06-20T21:48:32.284Z" }, + { url = "https://files.pythonhosted.org/packages/de/5f/2c78e28f309396e71ec8e4e9304a6483dcbc36172b5cea8f291994163425/hf_xet-1.1.5-cp37-abi3-macosx_11_0_arm64.whl", hash = "sha256:9fa6e3ee5d61912c4a113e0708eaaef987047616465ac7aa30f7121a48fc1af8", size = 2556338, upload-time = "2025-06-20T21:48:30.079Z" }, + { url = "https://files.pythonhosted.org/packages/6d/2f/6cad7b5fe86b7652579346cb7f85156c11761df26435651cbba89376cd2c/hf_xet-1.1.5-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc874b5c843e642f45fd85cda1ce599e123308ad2901ead23d3510a47ff506d1", size = 3102894, upload-time = "2025-06-20T21:48:28.114Z" }, + { url = "https://files.pythonhosted.org/packages/d0/54/0fcf2b619720a26fbb6cc941e89f2472a522cd963a776c089b189559447f/hf_xet-1.1.5-cp37-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:dbba1660e5d810bd0ea77c511a99e9242d920790d0e63c0e4673ed36c4022d18", size = 3002134, upload-time = "2025-06-20T21:48:25.906Z" }, + { url = "https://files.pythonhosted.org/packages/f3/92/1d351ac6cef7c4ba8c85744d37ffbfac2d53d0a6c04d2cabeba614640a78/hf_xet-1.1.5-cp37-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:ab34c4c3104133c495785d5d8bba3b1efc99de52c02e759cf711a91fd39d3a14", size = 3171009, upload-time = "2025-06-20T21:48:33.987Z" }, + { url = "https://files.pythonhosted.org/packages/c9/65/4b2ddb0e3e983f2508528eb4501288ae2f84963586fbdfae596836d5e57a/hf_xet-1.1.5-cp37-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:83088ecea236d5113de478acb2339f92c95b4fb0462acaa30621fac02f5a534a", size = 3279245, upload-time = "2025-06-20T21:48:36.051Z" }, + { url = "https://files.pythonhosted.org/packages/f0/55/ef77a85ee443ae05a9e9cba1c9f0dd9241eb42da2aeba1dc50f51154c81a/hf_xet-1.1.5-cp37-abi3-win_amd64.whl", hash = "sha256:73e167d9807d166596b4b2f0b585c6d5bd84a26dea32843665a8b58f6edba245", size = 2738931, upload-time = "2025-06-20T21:48:39.482Z" }, +] + +[[package]] +name = "huggingface-hub" +version = "0.34.3" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "filelock" }, + { name = "fsspec" }, + { name = "hf-xet", marker = "platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'" }, + { name = "packaging" }, + { name = "pyyaml" }, + { name = "requests" }, + { name = "tqdm" }, + { name = "typing-extensions" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/91/b4/e6b465eca5386b52cf23cb6df8644ad318a6b0e12b4b96a7e0be09cbfbcc/huggingface_hub-0.34.3.tar.gz", hash = "sha256:d58130fd5aa7408480681475491c0abd7e835442082fbc3ef4d45b6c39f83853", size = 456800, upload-time = "2025-07-29T08:38:53.885Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/59/a8/4677014e771ed1591a87b63a2392ce6923baf807193deef302dcfde17542/huggingface_hub-0.34.3-py3-none-any.whl", hash = "sha256:5444550099e2d86e68b2898b09e85878fbd788fc2957b506c6a79ce060e39492", size = 558847, upload-time = "2025-07-29T08:38:51.904Z" }, ] [[package]] @@ -808,11 +1140,11 @@ wheels = [ [[package]] name = "identify" -version = "2.6.9" +version = "2.6.12" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/9b/98/a71ab060daec766acc30fb47dfca219d03de34a70d616a79a38c6066c5bf/identify-2.6.9.tar.gz", hash = "sha256:d40dfe3142a1421d8518e3d3985ef5ac42890683e32306ad614a29490abeb6bf", size = 99249, upload-time = "2025-03-08T15:54:13.632Z" } +sdist = { url = "https://files.pythonhosted.org/packages/a2/88/d193a27416618628a5eea64e3223acd800b40749a96ffb322a9b55a49ed1/identify-2.6.12.tar.gz", hash = "sha256:d8de45749f1efb108badef65ee8386f0f7bb19a7f26185f74de6367bffbaf0e6", size = 99254, upload-time = "2025-05-23T20:37:53.3Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/07/ce/0845144ed1f0e25db5e7a79c2354c1da4b5ce392b8966449d5db8dca18f1/identify-2.6.9-py2.py3-none-any.whl", hash = "sha256:c98b4322da415a8e5a70ff6e51fbc2d2932c015532d77e9f8537b4ba7813b150", size = 99101, upload-time = "2025-03-08T15:54:12.026Z" }, + { url = "https://files.pythonhosted.org/packages/7a/cd/18f8da995b658420625f7ef13f037be53ae04ec5ad33f9b718240dcfd48c/identify-2.6.12-py2.py3-none-any.whl", hash = "sha256:ad9672d5a72e0d2ff7c5c8809b62dfa60458626352fb0eb7b55e69bdc45334a2", size = 99145, upload-time = "2025-05-23T20:37:51.495Z" }, ] [[package]] @@ -835,14 +1167,14 @@ wheels = [ [[package]] name = "ipykernel" -version = "6.29.5" +version = "6.30.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "appnope", marker = "sys_platform == 'darwin'" }, { name = "comm" }, { name = "debugpy" }, - { name = "ipython", version = "8.34.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "ipython", version = "9.0.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "ipython", version = "8.37.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "ipython", version = "9.4.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "jupyter-client" }, { name = "jupyter-core" }, { name = "matplotlib-inline" }, @@ -853,14 +1185,14 @@ dependencies = [ { name = "tornado" }, { name = "traitlets" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e9/5c/67594cb0c7055dc50814b21731c22a601101ea3b1b50a9a1b090e11f5d0f/ipykernel-6.29.5.tar.gz", hash = "sha256:f093a22c4a40f8828f8e330a9c297cb93dcab13bd9678ded6de8e5cf81c56215", size = 163367, upload-time = "2024-07-01T14:07:22.543Z" } +sdist = { url = "https://files.pythonhosted.org/packages/bb/76/11082e338e0daadc89c8ff866185de11daf67d181901038f9e139d109761/ipykernel-6.30.1.tar.gz", hash = "sha256:6abb270161896402e76b91394fcdce5d1be5d45f456671e5080572f8505be39b", size = 166260, upload-time = "2025-08-04T15:47:35.018Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/94/5c/368ae6c01c7628438358e6d337c19b05425727fbb221d2a3c4303c372f42/ipykernel-6.29.5-py3-none-any.whl", hash = "sha256:afdb66ba5aa354b09b91379bac28ae4afebbb30e8b39510c9690afb7a10421b5", size = 117173, upload-time = "2024-07-01T14:07:19.603Z" }, + { url = "https://files.pythonhosted.org/packages/fc/c7/b445faca8deb954fe536abebff4ece5b097b923de482b26e78448c89d1dd/ipykernel-6.30.1-py3-none-any.whl", hash = "sha256:aa6b9fb93dca949069d8b85b6c79b2518e32ac583ae9c7d37c51d119e18b3fb4", size = 117484, upload-time = "2025-08-04T15:47:32.622Z" }, ] [[package]] name = "ipython" -version = "8.34.0" +version = "8.37.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ "python_full_version < '3.11' and sys_platform == 'linux'", @@ -879,18 +1211,20 @@ dependencies = [ { name = "traitlets", marker = "python_full_version < '3.11'" }, { name = "typing-extensions", marker = "python_full_version < '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/13/18/1a60aa62e9d272fcd7e658a89e1c148da10e1a5d38edcbcd834b52ca7492/ipython-8.34.0.tar.gz", hash = "sha256:c31d658e754673ecc6514583e7dda8069e47136eb62458816b7d1e6625948b5a", size = 5508477, upload-time = "2025-03-08T13:43:17.591Z" } +sdist = { url = "https://files.pythonhosted.org/packages/85/31/10ac88f3357fc276dc8a64e8880c82e80e7459326ae1d0a211b40abf6665/ipython-8.37.0.tar.gz", hash = "sha256:ca815841e1a41a1e6b73a0b08f3038af9b2252564d01fc405356d34033012216", size = 5606088, upload-time = "2025-05-31T16:39:09.613Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/04/78/45615356bb973904856808183ae2a5fba1f360e9d682314d79766f4b88f2/ipython-8.34.0-py3-none-any.whl", hash = "sha256:0419883fa46e0baa182c5d50ebb8d6b49df1889fdb70750ad6d8cfe678eda6e3", size = 826731, upload-time = "2025-03-08T13:43:15.004Z" }, + { url = "https://files.pythonhosted.org/packages/91/d0/274fbf7b0b12643cbbc001ce13e6a5b1607ac4929d1b11c72460152c9fc3/ipython-8.37.0-py3-none-any.whl", hash = "sha256:ed87326596b878932dbcb171e3e698845434d8c61b8d8cd474bf663041a9dcf2", size = 831864, upload-time = "2025-05-31T16:39:06.38Z" }, ] [[package]] name = "ipython" -version = "9.0.2" +version = "9.4.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ - "python_full_version >= '3.12' and sys_platform == 'linux'", - "python_full_version >= '3.12' and sys_platform != 'linux'", + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", "python_full_version == '3.11.*' and sys_platform == 'linux'", "python_full_version == '3.11.*' and sys_platform != 'linux'", ] @@ -907,9 +1241,9 @@ dependencies = [ { name = "traitlets", marker = "python_full_version >= '3.11'" }, { name = "typing-extensions", marker = "python_full_version == '3.11.*'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/7d/ce/012a0f40ca58a966f87a6e894d6828e2817657cbdf522b02a5d3a87d92ce/ipython-9.0.2.tar.gz", hash = "sha256:ec7b479e3e5656bf4f58c652c120494df1820f4f28f522fb7ca09e213c2aab52", size = 4366102, upload-time = "2025-03-08T15:04:52.885Z" } +sdist = { url = "https://files.pythonhosted.org/packages/54/80/406f9e3bde1c1fd9bf5a0be9d090f8ae623e401b7670d8f6fdf2ab679891/ipython-9.4.0.tar.gz", hash = "sha256:c033c6d4e7914c3d9768aabe76bbe87ba1dc66a92a05db6bfa1125d81f2ee270", size = 4385338, upload-time = "2025-07-01T11:11:30.606Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/20/3a/917cb9e72f4e1a4ea13c862533205ae1319bd664119189ee5cc9e4e95ebf/ipython-9.0.2-py3-none-any.whl", hash = "sha256:143ef3ea6fb1e1bffb4c74b114051de653ffb7737a3f7ab1670e657ca6ae8c44", size = 600524, upload-time = "2025-03-08T15:04:50.667Z" }, + { url = "https://files.pythonhosted.org/packages/63/f8/0031ee2b906a15a33d6bfc12dd09c3dfa966b3cb5b284ecfb7549e6ac3c4/ipython-9.4.0-py3-none-any.whl", hash = "sha256:25850f025a446d9b359e8d296ba175a36aedd32e83ca9b5060430fe16801f066", size = 611021, upload-time = "2025-07-01T11:11:27.85Z" }, ] [[package]] @@ -936,10 +1270,13 @@ dependencies = [ { name = "lightning" }, { name = "lovelyplots" }, { name = "matplotlib" }, - { name = "mdtraj" }, + { name = "mdtraj", version = "1.10.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "mdtraj", version = "1.11.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "ninja" }, - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "omegaconf" }, + { name = "orb-models" }, { name = "pandas" }, { name = "plotly" }, { name = "posebusters" }, @@ -949,7 +1286,8 @@ dependencies = [ { name = "rdkit" }, { name = "requests" }, { name = "s3fs", extra = ["boto3"] }, - { name = "scipy" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "statsmodels" }, { name = "tabulate" }, { name = "torch" }, @@ -964,8 +1302,8 @@ dependencies = [ [package.dev-dependencies] dev = [ { name = "ipykernel" }, - { name = "ipython", version = "8.34.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "ipython", version = "9.0.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "ipython", version = "8.37.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "ipython", version = "9.4.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "nbstripout" }, { name = "pre-commit" }, { name = "pytest" }, @@ -986,6 +1324,7 @@ requires-dist = [ { name = "ninja", specifier = ">=1.11.1.3" }, { name = "numpy", specifier = ">=2" }, { name = "omegaconf", specifier = ">=2.3.0" }, + { name = "orb-models", specifier = ">=0.5.4" }, { name = "pandas", specifier = ">=2.1.0" }, { name = "plotly", specifier = ">=5.24.1" }, { name = "posebusters", specifier = ">=0.3.1" }, @@ -1019,14 +1358,14 @@ dev = [ [[package]] name = "jaxtyping" -version = "0.3.0" +version = "0.3.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "wadler-lindig" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/65/bc/a3d6e865ed05e93c5f9c54050fb5f6239c024089db93c3e4f8d85465487d/jaxtyping-0.3.0.tar.gz", hash = "sha256:b334b56436295332addd0b6c451548404d3700c9c35c7fa877c6b3b30ea968de", size = 44793, upload-time = "2025-03-18T17:28:27.372Z" } +sdist = { url = "https://files.pythonhosted.org/packages/88/a8/416bd7ea110ec6b68e8868b0f99c985c735adcf7badc491d3c343937260a/jaxtyping-0.3.2.tar.gz", hash = "sha256:f30483fac4b42e915db8ad2414a85c3b63284aa7d3c100b96b59f755cf4a86ad", size = 44989, upload-time = "2025-04-23T09:48:16.907Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e9/6e/ce242e8f39919e1af817af194189c5ed2cd2a02dae242563932750787e6b/jaxtyping-0.3.0-py3-none-any.whl", hash = "sha256:4b20d4e7c94d6a2850d78d7849cf33e38a87b993f2f78977d8093efb42cdb892", size = 55199, upload-time = "2025-03-18T17:28:26.003Z" }, + { url = "https://files.pythonhosted.org/packages/c9/b9/281e10e2d967ea5e481683eaec99f55ac5a61085ee60551c36942ef32bef/jaxtyping-0.3.2-py3-none-any.whl", hash = "sha256:6a020fd276226ddb5ac4f5725323843dd65e3c7e85c64fd62431e5f738c74e04", size = 55409, upload-time = "2025-04-23T09:48:15.924Z" }, ] [[package]] @@ -1064,7 +1403,7 @@ wheels = [ [[package]] name = "jsonschema" -version = "4.23.0" +version = "4.25.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "attrs" }, @@ -1072,21 +1411,21 @@ dependencies = [ { name = "referencing" }, { name = "rpds-py" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/38/2e/03362ee4034a4c917f697890ccd4aec0800ccf9ded7f511971c75451deec/jsonschema-4.23.0.tar.gz", hash = "sha256:d71497fef26351a33265337fa77ffeb82423f3ea21283cd9467bb03999266bc4", size = 325778, upload-time = "2024-07-08T18:40:05.546Z" } +sdist = { url = "https://files.pythonhosted.org/packages/d5/00/a297a868e9d0784450faa7365c2172a7d6110c763e30ba861867c32ae6a9/jsonschema-4.25.0.tar.gz", hash = "sha256:e63acf5c11762c0e6672ffb61482bdf57f0876684d8d249c0fe2d730d48bc55f", size = 356830, upload-time = "2025-07-18T15:39:45.11Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/69/4a/4f9dbeb84e8850557c02365a0eee0649abe5eb1d84af92a25731c6c0f922/jsonschema-4.23.0-py3-none-any.whl", hash = "sha256:fbadb6f8b144a8f8cf9f0b89ba94501d143e50411a1278633f56a7acf7fd5566", size = 88462, upload-time = "2024-07-08T18:40:00.165Z" }, + { url = "https://files.pythonhosted.org/packages/fe/54/c86cd8e011fe98803d7e382fd67c0df5ceab8d2b7ad8c5a81524f791551c/jsonschema-4.25.0-py3-none-any.whl", hash = "sha256:24c2e8da302de79c8b9382fee3e76b355e44d2a4364bb207159ce10b517bd716", size = 89184, upload-time = "2025-07-18T15:39:42.956Z" }, ] [[package]] name = "jsonschema-specifications" -version = "2024.10.1" +version = "2025.4.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "referencing" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/10/db/58f950c996c793472e336ff3655b13fbcf1e3b359dcf52dcf3ed3b52c352/jsonschema_specifications-2024.10.1.tar.gz", hash = "sha256:0f38b83639958ce1152d02a7f062902c41c8fd20d558b0c34344292d417ae272", size = 15561, upload-time = "2024-10-08T12:29:32.068Z" } +sdist = { url = "https://files.pythonhosted.org/packages/bf/ce/46fbd9c8119cfc3581ee5643ea49464d168028cfb5caff5fc0596d0cf914/jsonschema_specifications-2025.4.1.tar.gz", hash = "sha256:630159c9f4dbea161a6a2205c3011cc4f18ff381b189fff48bb39b9bf26ae608", size = 15513, upload-time = "2025-04-23T12:34:07.418Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/d1/0f/8910b19ac0670a0f80ce1008e5e751c4a57e14d2c4c13a482aa6079fa9d6/jsonschema_specifications-2024.10.1-py3-none-any.whl", hash = "sha256:a09a0680616357d9a0ecf05c12ad234479f549239d0f5b55f3deea67475da9bf", size = 18459, upload-time = "2024-10-08T12:29:30.439Z" }, + { url = "https://files.pythonhosted.org/packages/01/0e/b27cdbaccf30b890c40ed1da9fd4a3593a5cf94dae54fb34f8a4b74fcd3f/jsonschema_specifications-2025.4.1-py3-none-any.whl", hash = "sha256:4653bffbd6584f7de83a67e0d620ef16900b390ddc7939d56684d6c81e33f1af", size = 18437, upload-time = "2025-04-23T12:34:05.422Z" }, ] [[package]] @@ -1107,16 +1446,16 @@ wheels = [ [[package]] name = "jupyter-core" -version = "5.7.2" +version = "5.8.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "platformdirs" }, { name = "pywin32", marker = "platform_python_implementation != 'PyPy' and sys_platform == 'win32'" }, { name = "traitlets" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/00/11/b56381fa6c3f4cc5d2cf54a7dbf98ad9aa0b339ef7a601d6053538b079a7/jupyter_core-5.7.2.tar.gz", hash = "sha256:aa5f8d32bbf6b431ac830496da7392035d6f61b4f54872f15c4bd2a9c3f536d9", size = 87629, upload-time = "2024-03-12T12:37:35.652Z" } +sdist = { url = "https://files.pythonhosted.org/packages/99/1b/72906d554acfeb588332eaaa6f61577705e9ec752ddb486f302dafa292d9/jupyter_core-5.8.1.tar.gz", hash = "sha256:0a5f9706f70e64786b75acba995988915ebd4601c8a52e534a40b51c95f59941", size = 88923, upload-time = "2025-05-27T07:38:16.655Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c9/fb/108ecd1fe961941959ad0ee4e12ee7b8b1477247f30b1fdfd83ceaf017f0/jupyter_core-5.7.2-py3-none-any.whl", hash = "sha256:4f7315d2f6b4bcf2e3e7cb6e46772eba760ae459cd1f59d29eb57b0a01bd7409", size = 28965, upload-time = "2024-03-12T12:37:32.36Z" }, + { url = "https://files.pythonhosted.org/packages/2f/57/6bffd4b20b88da3800c5d691e0337761576ee688eb01299eae865689d2df/jupyter_core-5.8.1-py3-none-any.whl", hash = "sha256:c28d268fc90fb53f1338ded2eb410704c5449a358406e8a948b75706e24863d0", size = 28880, upload-time = "2025-05-27T07:38:15.137Z" }, ] [[package]] @@ -1208,7 +1547,7 @@ wheels = [ [[package]] name = "lightning" -version = "2.5.1" +version = "2.5.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "fsspec", extra = ["http"] }, @@ -1221,23 +1560,23 @@ dependencies = [ { name = "tqdm" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/39/53/21a834e03317d04cc3db7307d4c19af94c0db43b9001e8fabb8d150c5e69/lightning-2.5.1.tar.gz", hash = "sha256:aca88f8abf3fc38d8b40c1f82ce481f4379c2b181a6eeeb9217db0aba8e40736", size = 630918, upload-time = "2025-03-19T20:28:24.517Z" } +sdist = { url = "https://files.pythonhosted.org/packages/42/3c/6a930ac7c64fb896adbe560a9141570732d9ca890a11e6d158edd5aece76/lightning-2.5.2.tar.gz", hash = "sha256:9550df613cfb22358ebf77b4a8ad45f3767cd7d26ba2d52b7f036bd3cdd701c4", size = 633391, upload-time = "2025-06-20T15:58:22.065Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/80/eb/45f6629b92cb4ed38854d5b76f9f668ff58404a4b9ec1abefa98502afd98/lightning-2.5.1-py3-none-any.whl", hash = "sha256:512cbf9e80859f331b329536b2e2f90776e6f8a399048745bb4dabacc36e2850", size = 818892, upload-time = "2025-03-19T20:28:21.036Z" }, + { url = "https://files.pythonhosted.org/packages/71/a9/5d39280e55dc5df9e98be074029f6b48f86fe3db4929cb9ada6401234b47/lightning-2.5.2-py3-none-any.whl", hash = "sha256:7e7f23245e214c8ec14d5d8119d3856c25cfe96f9856296fd5df4e29c2ff88a7", size = 821145, upload-time = "2025-06-20T15:58:18.609Z" }, ] [[package]] name = "lightning-utilities" -version = "0.14.2" +version = "0.15.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "packaging" }, { name = "setuptools" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e2/95/3654879a4213e29f2c17ce622ca289d67e35c22a35f63cc6d79d22456cfa/lightning_utilities-0.14.2.tar.gz", hash = "sha256:0466a4f1bb9dff1c7190d4c7a32d1a8a1109f94fb816931efe8fb8b12bb0ab8d", size = 30290, upload-time = "2025-03-20T12:53:38.95Z" } +sdist = { url = "https://files.pythonhosted.org/packages/10/c8/c1cccd09b1c8a63953cd158ed943b9ee054f553cec4d82974561ee3d20c9/lightning_utilities-0.15.1.tar.gz", hash = "sha256:e1052e9ab294a176f5efcfe6a57b2969b9675cfdae33b936612cbf23fffb4d4b", size = 31093, upload-time = "2025-08-04T20:10:16.658Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/ec/65/0243d079b4124840abd5fa5a472810c84b46aec67d1b05ba3ba96e3d01e8/lightning_utilities-0.14.2-py3-none-any.whl", hash = "sha256:da791fcaa731f651ec76a1a3b12994ed05af4d6841f2e78760233552709ef05d", size = 28891, upload-time = "2025-03-20T12:53:37.873Z" }, + { url = "https://files.pythonhosted.org/packages/c9/b5/2d118bb657a1877cea0fe7ebced4aaccf6eb1339603104e654cb9b782168/lightning_utilities-0.15.1-py3-none-any.whl", hash = "sha256:18437c86f8380fe80beee0cd3f7a7bde8a68d752060e6a2663c1e5784da768d9", size = 29446, upload-time = "2025-08-04T20:10:15.726Z" }, ] [[package]] @@ -1252,6 +1591,18 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/17/c8/dd7dd77f5022b0ae6d377d28dd49b470d820dc2fc51592c2c4860aa6cb2d/lovelyplots-1.0.2-py3-none-any.whl", hash = "sha256:3d778300203e546d7ff642b2ee16c91fa47f0980574b409148e1cb8976fe6832", size = 12968, upload-time = "2024-03-06T05:18:49.258Z" }, ] +[[package]] +name = "markdown-it-py" +version = "3.0.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "mdurl" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/38/71/3b932df36c1a044d397a1f92d1cf91ee0a503d91e470cbd670aa66b07ed0/markdown-it-py-3.0.0.tar.gz", hash = "sha256:e3f60a94fa066dc52ec76661e37c851cb232d92f9886b15cb560aaada2df8feb", size = 74596, upload-time = "2023-06-03T06:41:14.443Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/42/d7/1ec15b46af6af88f19b8e5ffea08fa375d433c998b8a7639e76935c14f1f/markdown_it_py-3.0.0-py3-none-any.whl", hash = "sha256:355216845c60bd96232cd8d8c40e8f9765cc86f46880e43a8fd22dc1a1a8cab1", size = 87528, upload-time = "2023-06-03T06:41:11.019Z" }, +] + [[package]] name = "markupsafe" version = "3.0.2" @@ -1312,54 +1663,77 @@ wheels = [ [[package]] name = "matplotlib" -version = "3.10.1" +version = "3.10.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "contourpy" }, + { name = "contourpy", version = "1.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "contourpy", version = "1.3.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "cycler" }, { name = "fonttools" }, { name = "kiwisolver" }, - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "packaging" }, { name = "pillow" }, { name = "pyparsing" }, { name = "python-dateutil" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/2f/08/b89867ecea2e305f408fbb417139a8dd941ecf7b23a2e02157c36da546f0/matplotlib-3.10.1.tar.gz", hash = "sha256:e8d2d0e3881b129268585bf4765ad3ee73a4591d77b9a18c214ac7e3a79fb2ba", size = 36743335, upload-time = "2025-02-27T19:19:51.038Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/ee/b1/f70e27cf1cd76ce2a5e1aa5579d05afe3236052c6d9b9a96325bc823a17e/matplotlib-3.10.1-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:ff2ae14910be903f4a24afdbb6d7d3a6c44da210fc7d42790b87aeac92238a16", size = 8163654, upload-time = "2025-02-27T19:18:10.961Z" }, - { url = "https://files.pythonhosted.org/packages/26/af/5ec3d4636106718bb62503a03297125d4514f98fe818461bd9e6b9d116e4/matplotlib-3.10.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:0721a3fd3d5756ed593220a8b86808a36c5031fce489adb5b31ee6dbb47dd5b2", size = 8037943, upload-time = "2025-02-27T19:18:16.742Z" }, - { url = "https://files.pythonhosted.org/packages/a1/3d/07f9003a71b698b848c9925d05979ffa94a75cd25d1a587202f0bb58aa81/matplotlib-3.10.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d0673b4b8f131890eb3a1ad058d6e065fb3c6e71f160089b65f8515373394698", size = 8449510, upload-time = "2025-02-27T19:18:19.56Z" }, - { url = "https://files.pythonhosted.org/packages/12/87/9472d4513ff83b7cd864311821793ab72234fa201ab77310ec1b585d27e2/matplotlib-3.10.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8e875b95ac59a7908978fe307ecdbdd9a26af7fa0f33f474a27fcf8c99f64a19", size = 8586585, upload-time = "2025-02-27T19:18:25.61Z" }, - { url = "https://files.pythonhosted.org/packages/31/9e/fe74d237d2963adae8608faeb21f778cf246dbbf4746cef87cffbc82c4b6/matplotlib-3.10.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:2589659ea30726284c6c91037216f64a506a9822f8e50592d48ac16a2f29e044", size = 9397911, upload-time = "2025-02-27T19:18:28.914Z" }, - { url = "https://files.pythonhosted.org/packages/b6/1b/025d3e59e8a4281ab463162ad7d072575354a1916aba81b6a11507dfc524/matplotlib-3.10.1-cp310-cp310-win_amd64.whl", hash = "sha256:a97ff127f295817bc34517255c9db6e71de8eddaab7f837b7d341dee9f2f587f", size = 8052998, upload-time = "2025-02-27T19:18:31.518Z" }, - { url = "https://files.pythonhosted.org/packages/a5/14/a1b840075be247bb1834b22c1e1d558740b0f618fe3a823740181ca557a1/matplotlib-3.10.1-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:057206ff2d6ab82ff3e94ebd94463d084760ca682ed5f150817b859372ec4401", size = 8174669, upload-time = "2025-02-27T19:18:34.346Z" }, - { url = "https://files.pythonhosted.org/packages/0a/e4/300b08e3e08f9c98b0d5635f42edabf2f7a1d634e64cb0318a71a44ff720/matplotlib-3.10.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a144867dd6bf8ba8cb5fc81a158b645037e11b3e5cf8a50bd5f9917cb863adfe", size = 8047996, upload-time = "2025-02-27T19:18:37.247Z" }, - { url = "https://files.pythonhosted.org/packages/75/f9/8d99ff5a2498a5f1ccf919fb46fb945109623c6108216f10f96428f388bc/matplotlib-3.10.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:56c5d9fcd9879aa8040f196a235e2dcbdf7dd03ab5b07c0696f80bc6cf04bedd", size = 8461612, upload-time = "2025-02-27T19:18:39.642Z" }, - { url = "https://files.pythonhosted.org/packages/40/b8/53fa08a5eaf78d3a7213fd6da1feec4bae14a81d9805e567013811ff0e85/matplotlib-3.10.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0f69dc9713e4ad2fb21a1c30e37bd445d496524257dfda40ff4a8efb3604ab5c", size = 8602258, upload-time = "2025-02-27T19:18:43.217Z" }, - { url = "https://files.pythonhosted.org/packages/40/87/4397d2ce808467af86684a622dd112664553e81752ea8bf61bdd89d24a41/matplotlib-3.10.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:4c59af3e8aca75d7744b68e8e78a669e91ccbcf1ac35d0102a7b1b46883f1dd7", size = 9408896, upload-time = "2025-02-27T19:18:45.852Z" }, - { url = "https://files.pythonhosted.org/packages/d7/68/0d03098b3feb786cbd494df0aac15b571effda7f7cbdec267e8a8d398c16/matplotlib-3.10.1-cp311-cp311-win_amd64.whl", hash = "sha256:11b65088c6f3dae784bc72e8d039a2580186285f87448babb9ddb2ad0082993a", size = 8061281, upload-time = "2025-02-27T19:18:48.919Z" }, - { url = "https://files.pythonhosted.org/packages/7c/1d/5e0dc3b59c034e43de16f94deb68f4ad8a96b3ea00f4b37c160b7474928e/matplotlib-3.10.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:66e907a06e68cb6cfd652c193311d61a12b54f56809cafbed9736ce5ad92f107", size = 8175488, upload-time = "2025-02-27T19:18:51.436Z" }, - { url = "https://files.pythonhosted.org/packages/7a/81/dae7e14042e74da658c3336ab9799128e09a1ee03964f2d89630b5d12106/matplotlib-3.10.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:e9b4bb156abb8fa5e5b2b460196f7db7264fc6d62678c03457979e7d5254b7be", size = 8046264, upload-time = "2025-02-27T19:18:54.344Z" }, - { url = "https://files.pythonhosted.org/packages/21/c4/22516775dcde10fc9c9571d155f90710761b028fc44f660508106c363c97/matplotlib-3.10.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1985ad3d97f51307a2cbfc801a930f120def19ba22864182dacef55277102ba6", size = 8452048, upload-time = "2025-02-27T19:18:56.536Z" }, - { url = "https://files.pythonhosted.org/packages/63/23/c0615001f67ce7c96b3051d856baedc0c818a2ed84570b9bf9bde200f85d/matplotlib-3.10.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c96f2c2f825d1257e437a1482c5a2cf4fee15db4261bd6fc0750f81ba2b4ba3d", size = 8597111, upload-time = "2025-02-27T19:18:59.439Z" }, - { url = "https://files.pythonhosted.org/packages/ca/c0/a07939a82aed77770514348f4568177d7dadab9787ebc618a616fe3d665e/matplotlib-3.10.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:35e87384ee9e488d8dd5a2dd7baf471178d38b90618d8ea147aced4ab59c9bea", size = 9402771, upload-time = "2025-02-27T19:19:01.944Z" }, - { url = "https://files.pythonhosted.org/packages/a6/b6/a9405484fb40746fdc6ae4502b16a9d6e53282ba5baaf9ebe2da579f68c4/matplotlib-3.10.1-cp312-cp312-win_amd64.whl", hash = "sha256:cfd414bce89cc78a7e1d25202e979b3f1af799e416010a20ab2b5ebb3a02425c", size = 8063742, upload-time = "2025-02-27T19:19:04.632Z" }, - { url = "https://files.pythonhosted.org/packages/60/73/6770ff5e5523d00f3bc584acb6031e29ee5c8adc2336b16cd1d003675fe0/matplotlib-3.10.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:c42eee41e1b60fd83ee3292ed83a97a5f2a8239b10c26715d8a6172226988d7b", size = 8176112, upload-time = "2025-02-27T19:19:07.59Z" }, - { url = "https://files.pythonhosted.org/packages/08/97/b0ca5da0ed54a3f6599c3ab568bdda65269bc27c21a2c97868c1625e4554/matplotlib-3.10.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:4f0647b17b667ae745c13721602b540f7aadb2a32c5b96e924cd4fea5dcb90f1", size = 8046931, upload-time = "2025-02-27T19:19:10.515Z" }, - { url = "https://files.pythonhosted.org/packages/df/9a/1acbdc3b165d4ce2dcd2b1a6d4ffb46a7220ceee960c922c3d50d8514067/matplotlib-3.10.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa3854b5f9473564ef40a41bc922be978fab217776e9ae1545c9b3a5cf2092a3", size = 8453422, upload-time = "2025-02-27T19:19:12.738Z" }, - { url = "https://files.pythonhosted.org/packages/51/d0/2bc4368abf766203e548dc7ab57cf7e9c621f1a3c72b516cc7715347b179/matplotlib-3.10.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7e496c01441be4c7d5f96d4e40f7fca06e20dcb40e44c8daa2e740e1757ad9e6", size = 8596819, upload-time = "2025-02-27T19:19:15.306Z" }, - { url = "https://files.pythonhosted.org/packages/ab/1b/8b350f8a1746c37ab69dda7d7528d1fc696efb06db6ade9727b7887be16d/matplotlib-3.10.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:5d45d3f5245be5b469843450617dcad9af75ca50568acf59997bed9311131a0b", size = 9402782, upload-time = "2025-02-27T19:19:17.841Z" }, - { url = "https://files.pythonhosted.org/packages/89/06/f570373d24d93503988ba8d04f213a372fa1ce48381c5eb15da985728498/matplotlib-3.10.1-cp313-cp313-win_amd64.whl", hash = "sha256:8e8e25b1209161d20dfe93037c8a7f7ca796ec9aa326e6e4588d8c4a5dd1e473", size = 8063812, upload-time = "2025-02-27T19:19:20.888Z" }, - { url = "https://files.pythonhosted.org/packages/fc/e0/8c811a925b5a7ad75135f0e5af46408b78af88bbb02a1df775100ef9bfef/matplotlib-3.10.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:19b06241ad89c3ae9469e07d77efa87041eac65d78df4fcf9cac318028009b01", size = 8214021, upload-time = "2025-02-27T19:19:23.412Z" }, - { url = "https://files.pythonhosted.org/packages/4a/34/319ec2139f68ba26da9d00fce2ff9f27679fb799a6c8e7358539801fd629/matplotlib-3.10.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:01e63101ebb3014e6e9f80d9cf9ee361a8599ddca2c3e166c563628b39305dbb", size = 8090782, upload-time = "2025-02-27T19:19:28.33Z" }, - { url = "https://files.pythonhosted.org/packages/77/ea/9812124ab9a99df5b2eec1110e9b2edc0b8f77039abf4c56e0a376e84a29/matplotlib-3.10.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3f06bad951eea6422ac4e8bdebcf3a70c59ea0a03338c5d2b109f57b64eb3972", size = 8478901, upload-time = "2025-02-27T19:19:31.536Z" }, - { url = "https://files.pythonhosted.org/packages/c9/db/b05bf463689134789b06dea85828f8ebe506fa1e37593f723b65b86c9582/matplotlib-3.10.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a3dfb036f34873b46978f55e240cff7a239f6c4409eac62d8145bad3fc6ba5a3", size = 8613864, upload-time = "2025-02-27T19:19:34.233Z" }, - { url = "https://files.pythonhosted.org/packages/c2/04/41ccec4409f3023a7576df3b5c025f1a8c8b81fbfe922ecfd837ac36e081/matplotlib-3.10.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:dc6ab14a7ab3b4d813b88ba957fc05c79493a037f54e246162033591e770de6f", size = 9409487, upload-time = "2025-02-27T19:19:36.924Z" }, - { url = "https://files.pythonhosted.org/packages/ac/c2/0d5aae823bdcc42cc99327ecdd4d28585e15ccd5218c453b7bcd827f3421/matplotlib-3.10.1-cp313-cp313t-win_amd64.whl", hash = "sha256:bc411ebd5889a78dabbc457b3fa153203e22248bfa6eedc6797be5df0164dbf9", size = 8134832, upload-time = "2025-02-27T19:19:39.431Z" }, - { url = "https://files.pythonhosted.org/packages/c8/f6/10adb696d8cbeed2ab4c2e26ecf1c80dd3847bbf3891f4a0c362e0e08a5a/matplotlib-3.10.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:648406f1899f9a818cef8c0231b44dcfc4ff36f167101c3fd1c9151f24220fdc", size = 8158685, upload-time = "2025-02-27T19:19:41.535Z" }, - { url = "https://files.pythonhosted.org/packages/3f/84/0603d917406072763e7f9bb37747d3d74d7ecd4b943a8c947cc3ae1cf7af/matplotlib-3.10.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:02582304e352f40520727984a5a18f37e8187861f954fea9be7ef06569cf85b4", size = 8035491, upload-time = "2025-02-27T19:19:44.186Z" }, - { url = "https://files.pythonhosted.org/packages/fd/7d/6a8b31dd07ed856b3eae001c9129670ef75c4698fa1c2a6ac9f00a4a7054/matplotlib-3.10.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d3809916157ba871bcdd33d3493acd7fe3037db5daa917ca6e77975a94cef779", size = 8590087, upload-time = "2025-02-27T19:19:46.709Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/43/91/f2939bb60b7ebf12478b030e0d7f340247390f402b3b189616aad790c366/matplotlib-3.10.5.tar.gz", hash = "sha256:352ed6ccfb7998a00881692f38b4ca083c691d3e275b4145423704c34c909076", size = 34804044, upload-time = "2025-07-31T18:09:33.805Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/d1/89/5355cdfe43242cb4d1a64a67cb6831398b665ad90e9702c16247cbd8d5ab/matplotlib-3.10.5-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:5d4773a6d1c106ca05cb5a5515d277a6bb96ed09e5c8fab6b7741b8fcaa62c8f", size = 8229094, upload-time = "2025-07-31T18:07:36.507Z" }, + { url = "https://files.pythonhosted.org/packages/34/bc/ba802650e1c69650faed261a9df004af4c6f21759d7a1ec67fe972f093b3/matplotlib-3.10.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:dc88af74e7ba27de6cbe6faee916024ea35d895ed3d61ef6f58c4ce97da7185a", size = 8091464, upload-time = "2025-07-31T18:07:38.864Z" }, + { url = "https://files.pythonhosted.org/packages/ac/64/8d0c8937dee86c286625bddb1902efacc3e22f2b619f5b5a8df29fe5217b/matplotlib-3.10.5-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:64c4535419d5617f7363dad171a5a59963308e0f3f813c4bed6c9e6e2c131512", size = 8653163, upload-time = "2025-07-31T18:07:41.141Z" }, + { url = "https://files.pythonhosted.org/packages/11/dc/8dfc0acfbdc2fc2336c72561b7935cfa73db9ca70b875d8d3e1b3a6f371a/matplotlib-3.10.5-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a277033048ab22d34f88a3c5243938cef776493f6201a8742ed5f8b553201343", size = 9490635, upload-time = "2025-07-31T18:07:42.936Z" }, + { url = "https://files.pythonhosted.org/packages/54/02/e3fdfe0f2e9fb05f3a691d63876639dbf684170fdcf93231e973104153b4/matplotlib-3.10.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:e4a6470a118a2e93022ecc7d3bd16b3114b2004ea2bf014fff875b3bc99b70c6", size = 9539036, upload-time = "2025-07-31T18:07:45.18Z" }, + { url = "https://files.pythonhosted.org/packages/c1/29/82bf486ff7f4dbedfb11ccc207d0575cbe3be6ea26f75be514252bde3d70/matplotlib-3.10.5-cp310-cp310-win_amd64.whl", hash = "sha256:7e44cada61bec8833c106547786814dd4a266c1b2964fd25daa3804f1b8d4467", size = 8093529, upload-time = "2025-07-31T18:07:49.553Z" }, + { url = "https://files.pythonhosted.org/packages/aa/c7/1f2db90a1d43710478bb1e9b57b162852f79234d28e4f48a28cc415aa583/matplotlib-3.10.5-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:dcfc39c452c6a9f9028d3e44d2d721484f665304857188124b505b2c95e1eecf", size = 8239216, upload-time = "2025-07-31T18:07:51.947Z" }, + { url = "https://files.pythonhosted.org/packages/82/6d/ca6844c77a4f89b1c9e4d481c412e1d1dbabf2aae2cbc5aa2da4a1d6683e/matplotlib-3.10.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:903352681b59f3efbf4546985142a9686ea1d616bb054b09a537a06e4b892ccf", size = 8102130, upload-time = "2025-07-31T18:07:53.65Z" }, + { url = "https://files.pythonhosted.org/packages/1d/1e/5e187a30cc673a3e384f3723e5f3c416033c1d8d5da414f82e4e731128ea/matplotlib-3.10.5-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:080c3676a56b8ee1c762bcf8fca3fe709daa1ee23e6ef06ad9f3fc17332f2d2a", size = 8666471, upload-time = "2025-07-31T18:07:55.304Z" }, + { url = "https://files.pythonhosted.org/packages/03/c0/95540d584d7d645324db99a845ac194e915ef75011a0d5e19e1b5cee7e69/matplotlib-3.10.5-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4b4984d5064a35b6f66d2c11d668565f4389b1119cc64db7a4c1725bc11adffc", size = 9500518, upload-time = "2025-07-31T18:07:57.199Z" }, + { url = "https://files.pythonhosted.org/packages/ba/2e/e019352099ea58b4169adb9c6e1a2ad0c568c6377c2b677ee1f06de2adc7/matplotlib-3.10.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:3967424121d3a46705c9fa9bdb0931de3228f13f73d7bb03c999c88343a89d89", size = 9552372, upload-time = "2025-07-31T18:07:59.41Z" }, + { url = "https://files.pythonhosted.org/packages/b7/81/3200b792a5e8b354f31f4101ad7834743ad07b6d620259f2059317b25e4d/matplotlib-3.10.5-cp311-cp311-win_amd64.whl", hash = "sha256:33775bbeb75528555a15ac29396940128ef5613cf9a2d31fb1bfd18b3c0c0903", size = 8100634, upload-time = "2025-07-31T18:08:01.801Z" }, + { url = "https://files.pythonhosted.org/packages/52/46/a944f6f0c1f5476a0adfa501969d229ce5ae60cf9a663be0e70361381f89/matplotlib-3.10.5-cp311-cp311-win_arm64.whl", hash = "sha256:c61333a8e5e6240e73769d5826b9a31d8b22df76c0778f8480baf1b4b01c9420", size = 7978880, upload-time = "2025-07-31T18:08:03.407Z" }, + { url = "https://files.pythonhosted.org/packages/66/1e/c6f6bcd882d589410b475ca1fc22e34e34c82adff519caf18f3e6dd9d682/matplotlib-3.10.5-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:00b6feadc28a08bd3c65b2894f56cf3c94fc8f7adcbc6ab4516ae1e8ed8f62e2", size = 8253056, upload-time = "2025-07-31T18:08:05.385Z" }, + { url = "https://files.pythonhosted.org/packages/53/e6/d6f7d1b59413f233793dda14419776f5f443bcccb2dfc84b09f09fe05dbe/matplotlib-3.10.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ee98a5c5344dc7f48dc261b6ba5d9900c008fc12beb3fa6ebda81273602cc389", size = 8110131, upload-time = "2025-07-31T18:08:07.293Z" }, + { url = "https://files.pythonhosted.org/packages/66/2b/bed8a45e74957549197a2ac2e1259671cd80b55ed9e1fe2b5c94d88a9202/matplotlib-3.10.5-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:a17e57e33de901d221a07af32c08870ed4528db0b6059dce7d7e65c1122d4bea", size = 8669603, upload-time = "2025-07-31T18:08:09.064Z" }, + { url = "https://files.pythonhosted.org/packages/7e/a7/315e9435b10d057f5e52dfc603cd353167ae28bb1a4e033d41540c0067a4/matplotlib-3.10.5-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:97b9d6443419085950ee4a5b1ee08c363e5c43d7176e55513479e53669e88468", size = 9508127, upload-time = "2025-07-31T18:08:10.845Z" }, + { url = "https://files.pythonhosted.org/packages/7f/d9/edcbb1f02ca99165365d2768d517898c22c6040187e2ae2ce7294437c413/matplotlib-3.10.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ceefe5d40807d29a66ae916c6a3915d60ef9f028ce1927b84e727be91d884369", size = 9566926, upload-time = "2025-07-31T18:08:13.186Z" }, + { url = "https://files.pythonhosted.org/packages/3b/d9/6dd924ad5616c97b7308e6320cf392c466237a82a2040381163b7500510a/matplotlib-3.10.5-cp312-cp312-win_amd64.whl", hash = "sha256:c04cba0f93d40e45b3c187c6c52c17f24535b27d545f757a2fffebc06c12b98b", size = 8107599, upload-time = "2025-07-31T18:08:15.116Z" }, + { url = "https://files.pythonhosted.org/packages/0e/f3/522dc319a50f7b0279fbe74f86f7a3506ce414bc23172098e8d2bdf21894/matplotlib-3.10.5-cp312-cp312-win_arm64.whl", hash = "sha256:a41bcb6e2c8e79dc99c5511ae6f7787d2fb52efd3d805fff06d5d4f667db16b2", size = 7978173, upload-time = "2025-07-31T18:08:21.518Z" }, + { url = "https://files.pythonhosted.org/packages/8d/05/4f3c1f396075f108515e45cb8d334aff011a922350e502a7472e24c52d77/matplotlib-3.10.5-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:354204db3f7d5caaa10e5de74549ef6a05a4550fdd1c8f831ab9bca81efd39ed", size = 8253586, upload-time = "2025-07-31T18:08:23.107Z" }, + { url = "https://files.pythonhosted.org/packages/2f/2c/e084415775aac7016c3719fe7006cdb462582c6c99ac142f27303c56e243/matplotlib-3.10.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b072aac0c3ad563a2b3318124756cb6112157017f7431626600ecbe890df57a1", size = 8110715, upload-time = "2025-07-31T18:08:24.675Z" }, + { url = "https://files.pythonhosted.org/packages/52/1b/233e3094b749df16e3e6cd5a44849fd33852e692ad009cf7de00cf58ddf6/matplotlib-3.10.5-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:d52fd5b684d541b5a51fb276b2b97b010c75bee9aa392f96b4a07aeb491e33c7", size = 8669397, upload-time = "2025-07-31T18:08:26.778Z" }, + { url = "https://files.pythonhosted.org/packages/e8/ec/03f9e003a798f907d9f772eed9b7c6a9775d5bd00648b643ebfb88e25414/matplotlib-3.10.5-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ee7a09ae2f4676276f5a65bd9f2bd91b4f9fbaedf49f40267ce3f9b448de501f", size = 9508646, upload-time = "2025-07-31T18:08:28.848Z" }, + { url = "https://files.pythonhosted.org/packages/91/e7/c051a7a386680c28487bca27d23b02d84f63e3d2a9b4d2fc478e6a42e37e/matplotlib-3.10.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ba6c3c9c067b83481d647af88b4e441d532acdb5ef22178a14935b0b881188f4", size = 9567424, upload-time = "2025-07-31T18:08:30.726Z" }, + { url = "https://files.pythonhosted.org/packages/36/c2/24302e93ff431b8f4173ee1dd88976c8d80483cadbc5d3d777cef47b3a1c/matplotlib-3.10.5-cp313-cp313-win_amd64.whl", hash = "sha256:07442d2692c9bd1cceaa4afb4bbe5b57b98a7599de4dabfcca92d3eea70f9ebe", size = 8107809, upload-time = "2025-07-31T18:08:33.928Z" }, + { url = "https://files.pythonhosted.org/packages/0b/33/423ec6a668d375dad825197557ed8fbdb74d62b432c1ed8235465945475f/matplotlib-3.10.5-cp313-cp313-win_arm64.whl", hash = "sha256:48fe6d47380b68a37ccfcc94f009530e84d41f71f5dae7eda7c4a5a84aa0a674", size = 7978078, upload-time = "2025-07-31T18:08:36.764Z" }, + { url = "https://files.pythonhosted.org/packages/51/17/521fc16ec766455c7bb52cc046550cf7652f6765ca8650ff120aa2d197b6/matplotlib-3.10.5-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3b80eb8621331449fc519541a7461987f10afa4f9cfd91afcd2276ebe19bd56c", size = 8295590, upload-time = "2025-07-31T18:08:38.521Z" }, + { url = "https://files.pythonhosted.org/packages/f8/12/23c28b2c21114c63999bae129fce7fd34515641c517ae48ce7b7dcd33458/matplotlib-3.10.5-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:47a388908e469d6ca2a6015858fa924e0e8a2345a37125948d8e93a91c47933e", size = 8158518, upload-time = "2025-07-31T18:08:40.195Z" }, + { url = "https://files.pythonhosted.org/packages/81/f8/aae4eb25e8e7190759f3cb91cbeaa344128159ac92bb6b409e24f8711f78/matplotlib-3.10.5-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:8b6b49167d208358983ce26e43aa4196073b4702858670f2eb111f9a10652b4b", size = 8691815, upload-time = "2025-07-31T18:08:42.238Z" }, + { url = "https://files.pythonhosted.org/packages/d0/ba/450c39ebdd486bd33a359fc17365ade46c6a96bf637bbb0df7824de2886c/matplotlib-3.10.5-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8a8da0453a7fd8e3da114234ba70c5ba9ef0e98f190309ddfde0f089accd46ea", size = 9522814, upload-time = "2025-07-31T18:08:44.914Z" }, + { url = "https://files.pythonhosted.org/packages/89/11/9c66f6a990e27bb9aa023f7988d2d5809cb98aa39c09cbf20fba75a542ef/matplotlib-3.10.5-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:52c6573dfcb7726a9907b482cd5b92e6b5499b284ffacb04ffbfe06b3e568124", size = 9573917, upload-time = "2025-07-31T18:08:47.038Z" }, + { url = "https://files.pythonhosted.org/packages/b3/69/8b49394de92569419e5e05e82e83df9b749a0ff550d07631ea96ed2eb35a/matplotlib-3.10.5-cp313-cp313t-win_amd64.whl", hash = "sha256:a23193db2e9d64ece69cac0c8231849db7dd77ce59c7b89948cf9d0ce655a3ce", size = 8181034, upload-time = "2025-07-31T18:08:48.943Z" }, + { url = "https://files.pythonhosted.org/packages/47/23/82dc435bb98a2fc5c20dffcac8f0b083935ac28286413ed8835df40d0baa/matplotlib-3.10.5-cp313-cp313t-win_arm64.whl", hash = "sha256:56da3b102cf6da2776fef3e71cd96fcf22103a13594a18ac9a9b31314e0be154", size = 8023337, upload-time = "2025-07-31T18:08:50.791Z" }, + { url = "https://files.pythonhosted.org/packages/ac/e0/26b6cfde31f5383503ee45dcb7e691d45dadf0b3f54639332b59316a97f8/matplotlib-3.10.5-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:96ef8f5a3696f20f55597ffa91c28e2e73088df25c555f8d4754931515512715", size = 8253591, upload-time = "2025-07-31T18:08:53.254Z" }, + { url = "https://files.pythonhosted.org/packages/c1/89/98488c7ef7ea20ea659af7499628c240a608b337af4be2066d644cfd0a0f/matplotlib-3.10.5-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:77fab633e94b9da60512d4fa0213daeb76d5a7b05156840c4fd0399b4b818837", size = 8112566, upload-time = "2025-07-31T18:08:55.116Z" }, + { url = "https://files.pythonhosted.org/packages/52/67/42294dfedc82aea55e1a767daf3263aacfb5a125f44ba189e685bab41b6f/matplotlib-3.10.5-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:27f52634315e96b1debbfdc5c416592edcd9c4221bc2f520fd39c33db5d9f202", size = 9513281, upload-time = "2025-07-31T18:08:56.885Z" }, + { url = "https://files.pythonhosted.org/packages/e7/68/f258239e0cf34c2cbc816781c7ab6fca768452e6bf1119aedd2bd4a882a3/matplotlib-3.10.5-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:525f6e28c485c769d1f07935b660c864de41c37fd716bfa64158ea646f7084bb", size = 9780873, upload-time = "2025-07-31T18:08:59.241Z" }, + { url = "https://files.pythonhosted.org/packages/89/64/f4881554006bd12e4558bd66778bdd15d47b00a1f6c6e8b50f6208eda4b3/matplotlib-3.10.5-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:1f5f3ec4c191253c5f2b7c07096a142c6a1c024d9f738247bfc8e3f9643fc975", size = 9568954, upload-time = "2025-07-31T18:09:01.244Z" }, + { url = "https://files.pythonhosted.org/packages/06/f8/42779d39c3f757e1f012f2dda3319a89fb602bd2ef98ce8faf0281f4febd/matplotlib-3.10.5-cp314-cp314-win_amd64.whl", hash = "sha256:707f9c292c4cd4716f19ab8a1f93f26598222cd931e0cd98fbbb1c5994bf7667", size = 8237465, upload-time = "2025-07-31T18:09:03.206Z" }, + { url = "https://files.pythonhosted.org/packages/cf/f8/153fd06b5160f0cd27c8b9dd797fcc9fb56ac6a0ebf3c1f765b6b68d3c8a/matplotlib-3.10.5-cp314-cp314-win_arm64.whl", hash = "sha256:21a95b9bf408178d372814de7baacd61c712a62cae560b5e6f35d791776f6516", size = 8108898, upload-time = "2025-07-31T18:09:05.231Z" }, + { url = "https://files.pythonhosted.org/packages/9a/ee/c4b082a382a225fe0d2a73f1f57cf6f6f132308805b493a54c8641006238/matplotlib-3.10.5-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:a6b310f95e1102a8c7c817ef17b60ee5d1851b8c71b63d9286b66b177963039e", size = 8295636, upload-time = "2025-07-31T18:09:07.306Z" }, + { url = "https://files.pythonhosted.org/packages/30/73/2195fa2099718b21a20da82dfc753bf2af58d596b51aefe93e359dd5915a/matplotlib-3.10.5-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:94986a242747a0605cb3ff1cb98691c736f28a59f8ffe5175acaeb7397c49a5a", size = 8158575, upload-time = "2025-07-31T18:09:09.083Z" }, + { url = "https://files.pythonhosted.org/packages/f6/e9/a08cdb34618a91fa08f75e6738541da5cacde7c307cea18ff10f0d03fcff/matplotlib-3.10.5-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1ff10ea43288f0c8bab608a305dc6c918cc729d429c31dcbbecde3b9f4d5b569", size = 9522815, upload-time = "2025-07-31T18:09:11.191Z" }, + { url = "https://files.pythonhosted.org/packages/4e/bb/34d8b7e0d1bb6d06ef45db01dfa560d5a67b1c40c0b998ce9ccde934bb09/matplotlib-3.10.5-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f6adb644c9d040ffb0d3434e440490a66cf73dbfa118a6f79cd7568431f7a012", size = 9783514, upload-time = "2025-07-31T18:09:13.307Z" }, + { url = "https://files.pythonhosted.org/packages/12/09/d330d1e55dcca2e11b4d304cc5227f52e2512e46828d6249b88e0694176e/matplotlib-3.10.5-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:4fa40a8f98428f789a9dcacd625f59b7bc4e3ef6c8c7c80187a7a709475cf592", size = 9573932, upload-time = "2025-07-31T18:09:15.335Z" }, + { url = "https://files.pythonhosted.org/packages/eb/3b/f70258ac729aa004aca673800a53a2b0a26d49ca1df2eaa03289a1c40f81/matplotlib-3.10.5-cp314-cp314t-win_amd64.whl", hash = "sha256:95672a5d628b44207aab91ec20bf59c26da99de12b88f7e0b1fb0a84a86ff959", size = 8322003, upload-time = "2025-07-31T18:09:17.416Z" }, + { url = "https://files.pythonhosted.org/packages/5b/60/3601f8ce6d76a7c81c7f25a0e15fde0d6b66226dd187aa6d2838e6374161/matplotlib-3.10.5-cp314-cp314t-win_arm64.whl", hash = "sha256:2efaf97d72629e74252e0b5e3c46813e9eeaa94e011ecf8084a971a31a97f40b", size = 8153849, upload-time = "2025-07-31T18:09:19.673Z" }, + { url = "https://files.pythonhosted.org/packages/e4/eb/7d4c5de49eb78294e1a8e2be8a6ecff8b433e921b731412a56cd1abd3567/matplotlib-3.10.5-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:b5fa2e941f77eb579005fb804026f9d0a1082276118d01cc6051d0d9626eaa7f", size = 8222360, upload-time = "2025-07-31T18:09:21.813Z" }, + { url = "https://files.pythonhosted.org/packages/16/8a/e435db90927b66b16d69f8f009498775f4469f8de4d14b87856965e58eba/matplotlib-3.10.5-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:1fc0d2a3241cdcb9daaca279204a3351ce9df3c0e7e621c7e04ec28aaacaca30", size = 8087462, upload-time = "2025-07-31T18:09:23.504Z" }, + { url = "https://files.pythonhosted.org/packages/0b/dd/06c0e00064362f5647f318e00b435be2ff76a1bdced97c5eaf8347311fbe/matplotlib-3.10.5-pp310-pypy310_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:8dee65cb1424b7dc982fe87895b5613d4e691cc57117e8af840da0148ca6c1d7", size = 8659802, upload-time = "2025-07-31T18:09:25.256Z" }, + { url = "https://files.pythonhosted.org/packages/dc/d6/e921be4e1a5f7aca5194e1f016cb67ec294548e530013251f630713e456d/matplotlib-3.10.5-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:160e125da27a749481eaddc0627962990f6029811dbeae23881833a011a0907f", size = 8233224, upload-time = "2025-07-31T18:09:27.512Z" }, + { url = "https://files.pythonhosted.org/packages/ec/74/a2b9b04824b9c349c8f1b2d21d5af43fa7010039427f2b133a034cb09e59/matplotlib-3.10.5-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:ac3d50760394d78a3c9be6b28318fe22b494c4fcf6407e8fd4794b538251899b", size = 8098539, upload-time = "2025-07-31T18:09:29.629Z" }, + { url = "https://files.pythonhosted.org/packages/fc/66/cd29ebc7f6c0d2a15d216fb572573e8fc38bd5d6dec3bd9d7d904c0949f7/matplotlib-3.10.5-pp311-pypy311_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:6c49465bf689c4d59d174d0c7795fb42a21d4244d11d70e52b8011987367ac61", size = 8672192, upload-time = "2025-07-31T18:09:31.407Z" }, ] [[package]] @@ -1378,12 +1752,16 @@ wheels = [ name = "mdtraj" version = "1.10.3" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'linux'", + "python_full_version < '3.11' and sys_platform != 'linux'", +] dependencies = [ - { name = "netcdf4" }, - { name = "numpy" }, - { name = "packaging" }, - { name = "pyparsing" }, - { name = "scipy" }, + { name = "netcdf4", marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "packaging", marker = "python_full_version < '3.11'" }, + { name = "pyparsing", marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/6a/ef/44c34823d2cfd935b76be102ea41bce3a2e688c3d84ccff9029df0c1ff96/mdtraj-1.10.3.tar.gz", hash = "sha256:d14a35009263725b784c436a8ac63fb6ceeb2bb366a526715dac6590d21025e5", size = 2548673, upload-time = "2025-02-07T15:52:11.953Z" } wheels = [ @@ -1405,6 +1783,46 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/26/b7/cd45c6bae1566572d96bda6e749c63886c9c6ded079e34615376de5fe26e/mdtraj-1.10.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1c083e080d1ddf3eb25acec343f4efe93671e1508e17f61b656db8c3a50a38d1", size = 7800597, upload-time = "2025-02-07T18:11:16.853Z" }, ] +[[package]] +name = "mdtraj" +version = "1.11.0" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", + "python_full_version == '3.11.*' and sys_platform == 'linux'", + "python_full_version == '3.11.*' and sys_platform != 'linux'", +] +dependencies = [ + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "packaging", marker = "python_full_version >= '3.11'" }, + { name = "pyparsing", marker = "python_full_version >= '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/49/87/04f21b321e7bb6f3e09b5339df51699b8caae4846796175845bbc44f298b/mdtraj-1.11.0.tar.gz", hash = "sha256:2cf0ed2ee9a603dc4599743b1806e1387c655e2a1e9d013841b99fb137ad852e", size = 2860049, upload-time = "2025-06-25T02:06:19.716Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/f8/bc/1a6d273467a24dad5048bf527288a7bbe997ecee8342f6e8fdd7465c9132/mdtraj-1.11.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:a6bbf2bc03a2ed23fd1fe696f6f6f970f495e843f3baaa71a0e99da086420de1", size = 2463123, upload-time = "2025-06-25T02:03:06.602Z" }, + { url = "https://files.pythonhosted.org/packages/09/dc/37bd8a22c66d05a293979d8d8cf1de5e6c03d21ebd6950b6a839c8d8a16b/mdtraj-1.11.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c34002c49282d208fdd34cd1b2578859bc7377a22ebd66a194efa1fc17395e5", size = 8023237, upload-time = "2025-06-25T02:03:28.425Z" }, + { url = "https://files.pythonhosted.org/packages/4c/48/6fc195907c056df9b6ff6442bb39c5dd433b971b9deaecf6d18a0761efa8/mdtraj-1.11.0-cp311-cp311-win_amd64.whl", hash = "sha256:4c794aac3db13937ed7e4f8c2b773564a89b2b539c15f43bae9dab6e1ffa5de1", size = 1355146, upload-time = "2025-06-25T02:03:31.524Z" }, + { url = "https://files.pythonhosted.org/packages/21/71/56adb24577052f2cd6f9b678da908e8e2d7963db3a88faff39792be43903/mdtraj-1.11.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:878fb8e62962b881ba75d4200637280b518e891e4aa5f213052ddb1905704558", size = 2448664, upload-time = "2025-06-25T02:03:43.052Z" }, + { url = "https://files.pythonhosted.org/packages/50/af/7d65be36619befb29b51b6118a97121de6685720af7f4cfd6ff1902ca7eb/mdtraj-1.11.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:01eb00d9b62f22056303fb8e4be0f85257a2ce1003f1e6274393bbcc977badaf", size = 8028112, upload-time = "2025-06-25T02:04:12.949Z" }, + { url = "https://files.pythonhosted.org/packages/d2/8f/fd05aa3613fad29b3a779e3384797d2b64aa329bcfd18d25f8fb9c0d725e/mdtraj-1.11.0-cp312-cp312-win_amd64.whl", hash = "sha256:f87117f7e6742b11a3a39e55fe5c9b4fc013d27691ac671bcc6dfb8193beae0f", size = 1341985, upload-time = "2025-06-25T02:04:19.133Z" }, + { url = "https://files.pythonhosted.org/packages/9c/8f/adbeabf14f27a2befd021548c5d8e696ca129924e0473863fb0e9e546f27/mdtraj-1.11.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:125db8d7b8f795ed2a771d5d6f419461e2f9dd21ef74d43311570b9dadbf0d46", size = 2432588, upload-time = "2025-06-25T02:04:25.409Z" }, + { url = "https://files.pythonhosted.org/packages/b5/c4/4a2e8db0f24c1deb85f4719d54a0a52c56403e17db61ddd72a95d39e07d6/mdtraj-1.11.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:46def167a429af5834f4f26a80bca372518952519847c1ceffd0163814b1164e", size = 7983336, upload-time = "2025-06-25T02:04:57.022Z" }, + { url = "https://files.pythonhosted.org/packages/ac/17/239ae1869fa0f6fe11e42388f5d1e094ac774f4d122767722f0451523989/mdtraj-1.11.0-cp313-cp313-win_amd64.whl", hash = "sha256:8ad81b0aabf090b1b0e60ee7974bd676046e48d35a7efdc9a37a8eacaf5b70e4", size = 1340489, upload-time = "2025-06-25T02:05:04.776Z" }, +] + +[[package]] +name = "mdurl" +version = "0.1.2" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/d6/54/cfe61301667036ec958cb99bd3efefba235e65cdeb9c84d24a8293ba1d90/mdurl-0.1.2.tar.gz", hash = "sha256:bb413d29f5eea38f31dd4754dd7377d4465116fb207585f97bf925588687c1ba", size = 8729, upload-time = "2022-08-14T12:40:10.846Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl", hash = "sha256:84008a41e51615a49fc9966191ff91509e3c40b939176e643fd50a5c2196b8f8", size = 9979, upload-time = "2022-08-14T12:40:09.779Z" }, +] + [[package]] name = "mpmath" version = "1.3.0" @@ -1416,98 +1834,113 @@ wheels = [ [[package]] name = "multidict" -version = "6.2.0" +version = "6.6.3" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "typing-extensions", marker = "python_full_version < '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/82/4a/7874ca44a1c9b23796c767dd94159f6c17e31c0e7d090552a1c623247d82/multidict-6.2.0.tar.gz", hash = "sha256:0085b0afb2446e57050140240a8595846ed64d1cbd26cef936bfab3192c673b8", size = 71066, upload-time = "2025-03-17T16:55:54.689Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/2d/ca/3ae4d9c9ba78e7bcb63e3f12974b8fa16b9a20de44e9785f5d291ccb823c/multidict-6.2.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:b9f6392d98c0bd70676ae41474e2eecf4c7150cb419237a41f8f96043fcb81d1", size = 49238, upload-time = "2025-03-17T16:53:32.192Z" }, - { url = "https://files.pythonhosted.org/packages/25/a4/55e595d2df586e442c85b2610542d1e14def4c6f641761125d35fb38f87c/multidict-6.2.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:3501621d5e86f1a88521ea65d5cad0a0834c77b26f193747615b7c911e5422d2", size = 29748, upload-time = "2025-03-17T16:53:34.057Z" }, - { url = "https://files.pythonhosted.org/packages/35/6f/09bc361a34bbf953e9897f69823f9c4b46aec0aaed6ec94ce63093ede317/multidict-6.2.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:32ed748ff9ac682eae7859790d3044b50e3076c7d80e17a44239683769ff485e", size = 30026, upload-time = "2025-03-17T16:53:35.378Z" }, - { url = "https://files.pythonhosted.org/packages/b6/c7/5b51816f7c38049fc50786f46e63c009e6fecd1953fbbafa8bfe4e2eb39d/multidict-6.2.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cc826b9a8176e686b67aa60fd6c6a7047b0461cae5591ea1dc73d28f72332a8a", size = 132393, upload-time = "2025-03-17T16:53:37.684Z" }, - { url = "https://files.pythonhosted.org/packages/1a/21/c51aca665afa93b397d2c47369f6c267193977611a55a7c9d8683dc095bc/multidict-6.2.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:214207dcc7a6221d9942f23797fe89144128a71c03632bf713d918db99bd36de", size = 139237, upload-time = "2025-03-17T16:53:39.287Z" }, - { url = "https://files.pythonhosted.org/packages/2e/9b/a7b91f8ed63314e7a3c276b4ca90ae5d0267a584ca2e42106baa728622d6/multidict-6.2.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:05fefbc3cddc4e36da209a5e49f1094bbece9a581faa7f3589201fd95df40e5d", size = 134920, upload-time = "2025-03-17T16:53:40.6Z" }, - { url = "https://files.pythonhosted.org/packages/c8/84/4b590a121b1009fe79d1ae5875b4aa9339d37d23e368dd3bcf5e36d27452/multidict-6.2.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e851e6363d0dbe515d8de81fd544a2c956fdec6f8a049739562286727d4a00c3", size = 129764, upload-time = "2025-03-17T16:53:41.881Z" }, - { url = "https://files.pythonhosted.org/packages/b8/de/831be406b5ab0dc0d25430ddf597c6ce1a2e23a4991363f1ca48f16fb817/multidict-6.2.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:32c9b4878f48be3e75808ea7e499d6223b1eea6d54c487a66bc10a1871e3dc6a", size = 122121, upload-time = "2025-03-17T16:53:43.848Z" }, - { url = "https://files.pythonhosted.org/packages/fa/2f/892334f4d3efc7cd11e3a64dc922a85611627380ee2de3d0627ac159a975/multidict-6.2.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:7243c5a6523c5cfeca76e063efa5f6a656d1d74c8b1fc64b2cd1e84e507f7e2a", size = 135640, upload-time = "2025-03-17T16:53:45.698Z" }, - { url = "https://files.pythonhosted.org/packages/6c/53/bf91c5fdede9406247dcbceaa9d7e7fa08e4d0e27fa3c76a0dab126bc6b2/multidict-6.2.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:0e5a644e50ef9fb87878d4d57907f03a12410d2aa3b93b3acdf90a741df52c49", size = 129655, upload-time = "2025-03-17T16:53:47.322Z" }, - { url = "https://files.pythonhosted.org/packages/d4/7a/f98e1c5d14c1bbbb83025a69da9a37344f7556c09fef39979cf62b464d60/multidict-6.2.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:0dc25a3293c50744796e87048de5e68996104d86d940bb24bc3ec31df281b191", size = 140691, upload-time = "2025-03-17T16:53:48.634Z" }, - { url = "https://files.pythonhosted.org/packages/dd/c9/af0ab78b53d5b769bc1fa751e53cc7356cef422bd1cf38ed653985a46ddf/multidict-6.2.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:a49994481b99cd7dedde07f2e7e93b1d86c01c0fca1c32aded18f10695ae17eb", size = 135254, upload-time = "2025-03-17T16:53:49.866Z" }, - { url = "https://files.pythonhosted.org/packages/c9/53/28cc971b17e25487a089bcf720fe284478f264a6fc619427ddf7145fcb2b/multidict-6.2.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:641cf2e3447c9ecff2f7aa6e9eee9eaa286ea65d57b014543a4911ff2799d08a", size = 133620, upload-time = "2025-03-17T16:53:51.713Z" }, - { url = "https://files.pythonhosted.org/packages/b6/9a/d7637fbe1d5928b9f6a33ce36c2ff37e0aab9aa22f5fc9552fd75fe7f364/multidict-6.2.0-cp310-cp310-win32.whl", hash = "sha256:0c383d28857f66f5aebe3e91d6cf498da73af75fbd51cedbe1adfb85e90c0460", size = 27044, upload-time = "2025-03-17T16:53:53.859Z" }, - { url = "https://files.pythonhosted.org/packages/4e/11/04758cc18a51227dbb350a8a25c7db0620d63fb23db5b8d1f87762f05cbe/multidict-6.2.0-cp310-cp310-win_amd64.whl", hash = "sha256:a33273a541f1e1a8219b2a4ed2de355848ecc0254264915b9290c8d2de1c74e1", size = 29149, upload-time = "2025-03-17T16:53:55.076Z" }, - { url = "https://files.pythonhosted.org/packages/97/aa/879cf5581bd56c19f1bd2682ee4ecfd4085a404668d4ee5138b0a08eaf2a/multidict-6.2.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:84e87a7d75fa36839a3a432286d719975362d230c70ebfa0948549cc38bd5b46", size = 49125, upload-time = "2025-03-17T16:53:56.148Z" }, - { url = "https://files.pythonhosted.org/packages/9e/d8/e6d47c166c13c48be8efb9720afe0f5cdc4da4687547192cbc3c03903041/multidict-6.2.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:8de4d42dffd5ced9117af2ce66ba8722402541a3aa98ffdf78dde92badb68932", size = 29689, upload-time = "2025-03-17T16:53:57.381Z" }, - { url = "https://files.pythonhosted.org/packages/a4/20/f3f0a2ca142c81100b6d4cbf79505961b54181d66157615bba3955304442/multidict-6.2.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:e7d91a230c7f8af86c904a5a992b8c064b66330544693fd6759c3d6162382ecf", size = 29975, upload-time = "2025-03-17T16:53:58.549Z" }, - { url = "https://files.pythonhosted.org/packages/ab/2d/1724972c7aeb7aa1916a3276cb32f9c39e186456ee7ed621504e7a758322/multidict-6.2.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9f6cad071960ba1914fa231677d21b1b4a3acdcce463cee41ea30bc82e6040cf", size = 135688, upload-time = "2025-03-17T16:53:59.653Z" }, - { url = "https://files.pythonhosted.org/packages/1a/08/ea54e7e245aaf0bb1c758578e5afba394ffccb8bd80d229a499b9b83f2b1/multidict-6.2.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0f74f2fc51555f4b037ef278efc29a870d327053aba5cb7d86ae572426c7cccc", size = 142703, upload-time = "2025-03-17T16:54:01.552Z" }, - { url = "https://files.pythonhosted.org/packages/97/76/960dee0424f38c71eda54101ee1ca7bb47c5250ed02f7b3e8e50b1ce0603/multidict-6.2.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:14ed9ed1bfedd72a877807c71113deac292bf485159a29025dfdc524c326f3e1", size = 138559, upload-time = "2025-03-17T16:54:02.973Z" }, - { url = "https://files.pythonhosted.org/packages/d0/35/969fd792e2e72801d80307f0a14f5b19c066d4a51d34dded22c71401527d/multidict-6.2.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4ac3fcf9a2d369bd075b2c2965544036a27ccd277fc3c04f708338cc57533081", size = 133312, upload-time = "2025-03-17T16:54:04.265Z" }, - { url = "https://files.pythonhosted.org/packages/a4/b8/f96657a2f744d577cfda5a7edf9da04a731b80d3239eafbfe7ca4d944695/multidict-6.2.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2fc6af8e39f7496047c7876314f4317736eac82bf85b54c7c76cf1a6f8e35d98", size = 125652, upload-time = "2025-03-17T16:54:05.814Z" }, - { url = "https://files.pythonhosted.org/packages/35/9d/97696d052297d8e2e08195a25c7aae873a6186c147b7635f979edbe3acde/multidict-6.2.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5f8cb1329f42fadfb40d6211e5ff568d71ab49be36e759345f91c69d1033d633", size = 139015, upload-time = "2025-03-17T16:54:07.791Z" }, - { url = "https://files.pythonhosted.org/packages/31/a0/5c106e28d42f20288c10049bc6647364287ba049dc00d6ae4f1584eb1bd1/multidict-6.2.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:5389445f0173c197f4a3613713b5fb3f3879df1ded2a1a2e4bc4b5b9c5441b7e", size = 132437, upload-time = "2025-03-17T16:54:09.491Z" }, - { url = "https://files.pythonhosted.org/packages/55/57/d5c60c075fef73422ae3b8f914221485b9ff15000b2db657c03bd190aee0/multidict-6.2.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:94a7bb972178a8bfc4055db80c51efd24baefaced5e51c59b0d598a004e8305d", size = 144037, upload-time = "2025-03-17T16:54:11.189Z" }, - { url = "https://files.pythonhosted.org/packages/eb/56/a23f599c697a455bf65ecb0f69a5b052d6442c567d380ed423f816246824/multidict-6.2.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:da51d8928ad8b4244926fe862ba1795f0b6e68ed8c42cd2f822d435db9c2a8f4", size = 138535, upload-time = "2025-03-17T16:54:12.453Z" }, - { url = "https://files.pythonhosted.org/packages/34/3a/a06ff9b5899090f4bbdbf09e237964c76cecfe75d2aa921e801356314017/multidict-6.2.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:063be88bd684782a0715641de853e1e58a2f25b76388538bd62d974777ce9bc2", size = 136885, upload-time = "2025-03-17T16:54:13.648Z" }, - { url = "https://files.pythonhosted.org/packages/d6/28/489c0eca1df3800cb5d0a66278d5dd2a4deae747a41d1cf553e6a4c0a984/multidict-6.2.0-cp311-cp311-win32.whl", hash = "sha256:52b05e21ff05729fbea9bc20b3a791c3c11da61649ff64cce8257c82a020466d", size = 27044, upload-time = "2025-03-17T16:54:16.495Z" }, - { url = "https://files.pythonhosted.org/packages/d0/b5/c7cd5ba9581add40bc743980f82426b90d9f42db0b56502011f1b3c929df/multidict-6.2.0-cp311-cp311-win_amd64.whl", hash = "sha256:1e2a2193d3aa5cbf5758f6d5680a52aa848e0cf611da324f71e5e48a9695cc86", size = 29145, upload-time = "2025-03-17T16:54:18.009Z" }, - { url = "https://files.pythonhosted.org/packages/a4/e2/0153a8db878aef9b2397be81e62cbc3b32ca9b94e0f700b103027db9d506/multidict-6.2.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:437c33561edb6eb504b5a30203daf81d4a9b727e167e78b0854d9a4e18e8950b", size = 49204, upload-time = "2025-03-17T16:54:19.193Z" }, - { url = "https://files.pythonhosted.org/packages/bb/9d/5ccb3224a976d1286f360bb4e89e67b7cdfb87336257fc99be3c17f565d7/multidict-6.2.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:9f49585f4abadd2283034fc605961f40c638635bc60f5162276fec075f2e37a4", size = 29807, upload-time = "2025-03-17T16:54:20.398Z" }, - { url = "https://files.pythonhosted.org/packages/62/32/ef20037f51b84b074a89bab5af46d4565381c3f825fc7cbfc19c1ee156be/multidict-6.2.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:5dd7106d064d05896ce28c97da3f46caa442fe5a43bc26dfb258e90853b39b44", size = 30000, upload-time = "2025-03-17T16:54:21.845Z" }, - { url = "https://files.pythonhosted.org/packages/97/81/b0a7560bfc3ec72606232cd7e60159e09b9cf29e66014d770c1315868fa2/multidict-6.2.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e25b11a0417475f093d0f0809a149aff3943c2c56da50fdf2c3c88d57fe3dfbd", size = 131820, upload-time = "2025-03-17T16:54:23.404Z" }, - { url = "https://files.pythonhosted.org/packages/49/3b/768bfc0e41179fbccd3a22925329a11755b7fdd53bec66dbf6b8772f0bce/multidict-6.2.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ac380cacdd3b183338ba63a144a34e9044520a6fb30c58aa14077157a033c13e", size = 136272, upload-time = "2025-03-17T16:54:24.636Z" }, - { url = "https://files.pythonhosted.org/packages/71/ac/fd2be3fe98ff54e7739448f771ba730d42036de0870737db9ae34bb8efe9/multidict-6.2.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:61d5541f27533f803a941d3a3f8a3d10ed48c12cf918f557efcbf3cd04ef265c", size = 135233, upload-time = "2025-03-17T16:54:25.884Z" }, - { url = "https://files.pythonhosted.org/packages/93/76/1657047da771315911a927b364a32dafce4135b79b64208ce4ac69525c56/multidict-6.2.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:facaf11f21f3a4c51b62931feb13310e6fe3475f85e20d9c9fdce0d2ea561b87", size = 132861, upload-time = "2025-03-17T16:54:27.154Z" }, - { url = "https://files.pythonhosted.org/packages/19/a5/9f07ffb9bf68b8aaa406c2abee27ad87e8b62a60551587b8e59ee91aea84/multidict-6.2.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:095a2eabe8c43041d3e6c2cb8287a257b5f1801c2d6ebd1dd877424f1e89cf29", size = 122166, upload-time = "2025-03-17T16:54:28.417Z" }, - { url = "https://files.pythonhosted.org/packages/95/23/b5ce3318d9d6c8f105c3679510f9d7202980545aad8eb4426313bd8da3ee/multidict-6.2.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:a0cc398350ef31167e03f3ca7c19313d4e40a662adcb98a88755e4e861170bdd", size = 136052, upload-time = "2025-03-17T16:54:29.689Z" }, - { url = "https://files.pythonhosted.org/packages/ce/5c/02cffec58ffe120873dce520af593415b91cc324be0345f534ad3637da4e/multidict-6.2.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:7c611345bbe7cb44aabb877cb94b63e86f2d0db03e382667dbd037866d44b4f8", size = 130094, upload-time = "2025-03-17T16:54:31.137Z" }, - { url = "https://files.pythonhosted.org/packages/49/f3/3b19a83f4ebf53a3a2a0435f3e447aa227b242ba3fd96a92404b31fb3543/multidict-6.2.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:8cd1a0644ccaf27e9d2f6d9c9474faabee21f0578fe85225cc5af9a61e1653df", size = 140962, upload-time = "2025-03-17T16:54:32.415Z" }, - { url = "https://files.pythonhosted.org/packages/cc/1a/c916b54fb53168c24cb6a3a0795fd99d0a59a0ea93fa9f6edeff5565cb20/multidict-6.2.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:89b3857652183b8206a891168af47bac10b970d275bba1f6ee46565a758c078d", size = 138082, upload-time = "2025-03-17T16:54:33.655Z" }, - { url = "https://files.pythonhosted.org/packages/ef/1a/dcb7fb18f64b3727c61f432c1e1a0d52b3924016124e4bbc8a7d2e4fa57b/multidict-6.2.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:125dd82b40f8c06d08d87b3510beaccb88afac94e9ed4a6f6c71362dc7dbb04b", size = 136019, upload-time = "2025-03-17T16:54:35.086Z" }, - { url = "https://files.pythonhosted.org/packages/fb/02/7695485375106f5c542574f70e1968c391f86fa3efc9f1fd76aac0af7237/multidict-6.2.0-cp312-cp312-win32.whl", hash = "sha256:76b34c12b013d813e6cb325e6bd4f9c984db27758b16085926bbe7ceeaace626", size = 26676, upload-time = "2025-03-17T16:54:36.32Z" }, - { url = "https://files.pythonhosted.org/packages/3c/f5/f147000fe1f4078160157b15b0790fff0513646b0f9b7404bf34007a9b44/multidict-6.2.0-cp312-cp312-win_amd64.whl", hash = "sha256:0b183a959fb88ad1be201de2c4bdf52fa8e46e6c185d76201286a97b6f5ee65c", size = 28899, upload-time = "2025-03-17T16:54:37.583Z" }, - { url = "https://files.pythonhosted.org/packages/a4/6c/5df5590b1f9a821154589df62ceae247537b01ab26b0aa85997c35ca3d9e/multidict-6.2.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:5c5e7d2e300d5cb3b2693b6d60d3e8c8e7dd4ebe27cd17c9cb57020cac0acb80", size = 49151, upload-time = "2025-03-17T16:54:38.756Z" }, - { url = "https://files.pythonhosted.org/packages/d5/ca/c917fbf1be989cd7ea9caa6f87e9c33844ba8d5fbb29cd515d4d2833b84c/multidict-6.2.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:256d431fe4583c5f1e0f2e9c4d9c22f3a04ae96009b8cfa096da3a8723db0a16", size = 29803, upload-time = "2025-03-17T16:54:40.256Z" }, - { url = "https://files.pythonhosted.org/packages/22/19/d97086fc96f73acf36d4dbe65c2c4175911969df49c4e94ef082be59d94e/multidict-6.2.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:a3c0ff89fe40a152e77b191b83282c9664357dce3004032d42e68c514ceff27e", size = 29947, upload-time = "2025-03-17T16:54:41.545Z" }, - { url = "https://files.pythonhosted.org/packages/e3/3b/203476b6e915c3f51616d5f87230c556e2f24b168c14818a3d8dae242b1b/multidict-6.2.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ef7d48207926edbf8b16b336f779c557dd8f5a33035a85db9c4b0febb0706817", size = 130369, upload-time = "2025-03-17T16:54:43.166Z" }, - { url = "https://files.pythonhosted.org/packages/c6/4f/67470007cf03b2bb6df8ae6d716a8eeb0a7d19e0c8dba4e53fa338883bca/multidict-6.2.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1f3c099d3899b14e1ce52262eb82a5f5cb92157bb5106bf627b618c090a0eadc", size = 135231, upload-time = "2025-03-17T16:54:44.572Z" }, - { url = "https://files.pythonhosted.org/packages/6d/f5/7a5ce64dc9a3fecc7d67d0b5cb9c262c67e0b660639e5742c13af63fd80f/multidict-6.2.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e16e7297f29a544f49340012d6fc08cf14de0ab361c9eb7529f6a57a30cbfda1", size = 133634, upload-time = "2025-03-17T16:54:45.998Z" }, - { url = "https://files.pythonhosted.org/packages/05/93/ab2931907e318c0437a4cd156c9cfff317ffb33d99ebbfe2d64200a870f7/multidict-6.2.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:042028348dc5a1f2be6c666437042a98a5d24cee50380f4c0902215e5ec41844", size = 131349, upload-time = "2025-03-17T16:54:47.837Z" }, - { url = "https://files.pythonhosted.org/packages/54/aa/ab8eda83a6a85f5b4bb0b1c28e62b18129b14519ef2e0d4cfd5f360da73c/multidict-6.2.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:08549895e6a799bd551cf276f6e59820aa084f0f90665c0f03dd3a50db5d3c48", size = 120861, upload-time = "2025-03-17T16:54:49.201Z" }, - { url = "https://files.pythonhosted.org/packages/15/2f/7d08ea7c5d9f45786893b4848fad59ec8ea567367d4234691a721e4049a1/multidict-6.2.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:4ccfd74957ef53fa7380aaa1c961f523d582cd5e85a620880ffabd407f8202c0", size = 134611, upload-time = "2025-03-17T16:54:50.811Z" }, - { url = "https://files.pythonhosted.org/packages/8b/07/387047bb1eac563981d397a7f85c75b306df1fff3c20b90da5a6cf6e487e/multidict-6.2.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:83b78c680d4b15d33042d330c2fa31813ca3974197bddb3836a5c635a5fd013f", size = 128955, upload-time = "2025-03-17T16:54:52.48Z" }, - { url = "https://files.pythonhosted.org/packages/8d/6e/7ae18f764a5282c2d682f1c90c6b2a0f6490327730170139a7a63bf3bb20/multidict-6.2.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:b4c153863dd6569f6511845922c53e39c8d61f6e81f228ad5443e690fca403de", size = 139759, upload-time = "2025-03-17T16:54:53.877Z" }, - { url = "https://files.pythonhosted.org/packages/b6/f4/c1b3b087b9379b9e56229bcf6570b9a963975c205a5811ac717284890598/multidict-6.2.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:98aa8325c7f47183b45588af9c434533196e241be0a4e4ae2190b06d17675c02", size = 136426, upload-time = "2025-03-17T16:54:56.506Z" }, - { url = "https://files.pythonhosted.org/packages/a2/0e/ef7b39b161ffd40f9e25dd62e59644b2ccaa814c64e9573f9bc721578419/multidict-6.2.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9e658d1373c424457ddf6d55ec1db93c280b8579276bebd1f72f113072df8a5d", size = 134648, upload-time = "2025-03-17T16:54:57.896Z" }, - { url = "https://files.pythonhosted.org/packages/37/5c/7905acd0ca411c97bcae62ab167d9922f0c5a1d316b6d3af875d4bda3551/multidict-6.2.0-cp313-cp313-win32.whl", hash = "sha256:3157126b028c074951839233647bd0e30df77ef1fedd801b48bdcad242a60f4e", size = 26680, upload-time = "2025-03-17T16:54:59.399Z" }, - { url = "https://files.pythonhosted.org/packages/89/36/96b071d1dad6ac44fe517e4250329e753787bb7a63967ef44bb9b3a659f6/multidict-6.2.0-cp313-cp313-win_amd64.whl", hash = "sha256:2e87f1926e91855ae61769ba3e3f7315120788c099677e0842e697b0bfb659f2", size = 28942, upload-time = "2025-03-17T16:55:00.813Z" }, - { url = "https://files.pythonhosted.org/packages/f5/05/d686cd2a12d648ecd434675ee8daa2901a80f477817e89ab3b160de5b398/multidict-6.2.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:2529ddbdaa424b2c6c2eb668ea684dd6b75b839d0ad4b21aad60c168269478d7", size = 50807, upload-time = "2025-03-17T16:55:02.162Z" }, - { url = "https://files.pythonhosted.org/packages/4c/1f/c7db5aac8fea129fa4c5a119e3d279da48d769138ae9624d1234aa01a06f/multidict-6.2.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:13551d0e2d7201f0959725a6a769b6f7b9019a168ed96006479c9ac33fe4096b", size = 30474, upload-time = "2025-03-17T16:55:04.097Z" }, - { url = "https://files.pythonhosted.org/packages/e5/f1/1fb27514f4d73cea165429dcb7d90cdc4a45445865832caa0c50dd545420/multidict-6.2.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:d1996ee1330e245cd3aeda0887b4409e3930524c27642b046e4fae88ffa66c5e", size = 30841, upload-time = "2025-03-17T16:55:06.098Z" }, - { url = "https://files.pythonhosted.org/packages/d6/6b/9487169e549a23c8958edbb332afaf1ab55d61f0c03cb758ee07ff8f74fb/multidict-6.2.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c537da54ce4ff7c15e78ab1292e5799d0d43a2108e006578a57f531866f64025", size = 148658, upload-time = "2025-03-17T16:55:07.556Z" }, - { url = "https://files.pythonhosted.org/packages/d7/22/79ebb2e4f70857c94999ce195db76886ae287b1b6102da73df24dcad4903/multidict-6.2.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0f249badb360b0b4d694307ad40f811f83df4da8cef7b68e429e4eea939e49dd", size = 151988, upload-time = "2025-03-17T16:55:09.141Z" }, - { url = "https://files.pythonhosted.org/packages/49/5d/63b17f3c1a2861587d26705923a94eb6b2600e5222d6b0d513bce5a78720/multidict-6.2.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:48d39b1824b8d6ea7de878ef6226efbe0773f9c64333e1125e0efcfdd18a24c7", size = 148432, upload-time = "2025-03-17T16:55:11.089Z" }, - { url = "https://files.pythonhosted.org/packages/a3/22/55204eec45c4280fa431c11494ad64d6da0dc89af76282fc6467432360a0/multidict-6.2.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b99aac6bb2c37db336fa03a39b40ed4ef2818bf2dfb9441458165ebe88b793af", size = 143161, upload-time = "2025-03-17T16:55:12.625Z" }, - { url = "https://files.pythonhosted.org/packages/97/e6/202b2cf5af161228767acab8bc49e73a91f4a7de088c9c71f3c02950a030/multidict-6.2.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:07bfa8bc649783e703263f783f73e27fef8cd37baaad4389816cf6a133141331", size = 136820, upload-time = "2025-03-17T16:55:14.073Z" }, - { url = "https://files.pythonhosted.org/packages/7d/16/dbedae0e94c7edc48fddef0c39483f2313205d9bc566fd7f11777b168616/multidict-6.2.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:b2c00ad31fbc2cbac85d7d0fcf90853b2ca2e69d825a2d3f3edb842ef1544a2c", size = 150875, upload-time = "2025-03-17T16:55:15.625Z" }, - { url = "https://files.pythonhosted.org/packages/f3/04/38ccf25d4bf8beef76a22bad7d9833fd088b4594c9765fe6fede39aa6c89/multidict-6.2.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:0d57a01a2a9fa00234aace434d8c131f0ac6e0ac6ef131eda5962d7e79edfb5b", size = 142050, upload-time = "2025-03-17T16:55:17.186Z" }, - { url = "https://files.pythonhosted.org/packages/9e/89/4f6b43386e7b79a4aad560d751981a0a282a1943c312ac72f940d7cf8f9f/multidict-6.2.0-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:abf5b17bc0cf626a8a497d89ac691308dbd825d2ac372aa990b1ca114e470151", size = 154117, upload-time = "2025-03-17T16:55:19.115Z" }, - { url = "https://files.pythonhosted.org/packages/24/e3/3dde5b193f86d30ad6400bd50e116b0df1da3f0c7d419661e3bd79e5ad86/multidict-6.2.0-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:f7716f7e7138252d88607228ce40be22660d6608d20fd365d596e7ca0738e019", size = 149408, upload-time = "2025-03-17T16:55:20.689Z" }, - { url = "https://files.pythonhosted.org/packages/df/b2/ec1e27e8e3da12fcc9053e1eae2f6b50faa8708064d83ea25aa7fb77ffd2/multidict-6.2.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:d5a36953389f35f0a4e88dc796048829a2f467c9197265504593f0e420571547", size = 145767, upload-time = "2025-03-17T16:55:22.271Z" }, - { url = "https://files.pythonhosted.org/packages/3a/8e/c07a648a9d592fa9f3a19d1c7e1c7738ba95aff90db967a5a09cff1e1f37/multidict-6.2.0-cp313-cp313t-win32.whl", hash = "sha256:e653d36b1bf48fa78c7fcebb5fa679342e025121ace8c87ab05c1cefd33b34fc", size = 28950, upload-time = "2025-03-17T16:55:23.807Z" }, - { url = "https://files.pythonhosted.org/packages/dc/a9/bebb5485b94d7c09831638a4df9a1a924c32431a750723f0bf39cd16a787/multidict-6.2.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ca23db5fb195b5ef4fd1f77ce26cadefdf13dba71dab14dadd29b34d457d7c44", size = 32001, upload-time = "2025-03-17T16:55:25.184Z" }, - { url = "https://files.pythonhosted.org/packages/9c/fd/b247aec6add5601956d440488b7f23151d8343747e82c038af37b28d6098/multidict-6.2.0-py3-none-any.whl", hash = "sha256:5d26547423e5e71dcc562c4acdc134b900640a39abd9066d7326a7cc2324c530", size = 10266, upload-time = "2025-03-17T16:55:52.771Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/3d/2c/5dad12e82fbdf7470f29bff2171484bf07cb3b16ada60a6589af8f376440/multidict-6.6.3.tar.gz", hash = "sha256:798a9eb12dab0a6c2e29c1de6f3468af5cb2da6053a20dfa3344907eed0937cc", size = 101006, upload-time = "2025-06-30T15:53:46.929Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/0b/67/414933982bce2efce7cbcb3169eaaf901e0f25baec69432b4874dfb1f297/multidict-6.6.3-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:a2be5b7b35271f7fff1397204ba6708365e3d773579fe2a30625e16c4b4ce817", size = 77017, upload-time = "2025-06-30T15:50:58.931Z" }, + { url = "https://files.pythonhosted.org/packages/8a/fe/d8a3ee1fad37dc2ef4f75488b0d9d4f25bf204aad8306cbab63d97bff64a/multidict-6.6.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:12f4581d2930840295c461764b9a65732ec01250b46c6b2c510d7ee68872b140", size = 44897, upload-time = "2025-06-30T15:51:00.999Z" }, + { url = "https://files.pythonhosted.org/packages/1f/e0/265d89af8c98240265d82b8cbcf35897f83b76cd59ee3ab3879050fd8c45/multidict-6.6.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:dd7793bab517e706c9ed9d7310b06c8672fd0aeee5781bfad612f56b8e0f7d14", size = 44574, upload-time = "2025-06-30T15:51:02.449Z" }, + { url = "https://files.pythonhosted.org/packages/e6/05/6b759379f7e8e04ccc97cfb2a5dcc5cdbd44a97f072b2272dc51281e6a40/multidict-6.6.3-cp310-cp310-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:72d8815f2cd3cf3df0f83cac3f3ef801d908b2d90409ae28102e0553af85545a", size = 225729, upload-time = "2025-06-30T15:51:03.794Z" }, + { url = "https://files.pythonhosted.org/packages/4e/f5/8d5a15488edd9a91fa4aad97228d785df208ed6298580883aa3d9def1959/multidict-6.6.3-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:531e331a2ee53543ab32b16334e2deb26f4e6b9b28e41f8e0c87e99a6c8e2d69", size = 242515, upload-time = "2025-06-30T15:51:05.002Z" }, + { url = "https://files.pythonhosted.org/packages/6e/b5/a8f317d47d0ac5bb746d6d8325885c8967c2a8ce0bb57be5399e3642cccb/multidict-6.6.3-cp310-cp310-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:42ca5aa9329a63be8dc49040f63817d1ac980e02eeddba763a9ae5b4027b9c9c", size = 222224, upload-time = "2025-06-30T15:51:06.148Z" }, + { url = "https://files.pythonhosted.org/packages/76/88/18b2a0d5e80515fa22716556061189c2853ecf2aa2133081ebbe85ebea38/multidict-6.6.3-cp310-cp310-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:208b9b9757060b9faa6f11ab4bc52846e4f3c2fb8b14d5680c8aac80af3dc751", size = 253124, upload-time = "2025-06-30T15:51:07.375Z" }, + { url = "https://files.pythonhosted.org/packages/62/bf/ebfcfd6b55a1b05ef16d0775ae34c0fe15e8dab570d69ca9941073b969e7/multidict-6.6.3-cp310-cp310-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:acf6b97bd0884891af6a8b43d0f586ab2fcf8e717cbd47ab4bdddc09e20652d8", size = 251529, upload-time = "2025-06-30T15:51:08.691Z" }, + { url = "https://files.pythonhosted.org/packages/44/11/780615a98fd3775fc309d0234d563941af69ade2df0bb82c91dda6ddaea1/multidict-6.6.3-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:68e9e12ed00e2089725669bdc88602b0b6f8d23c0c95e52b95f0bc69f7fe9b55", size = 241627, upload-time = "2025-06-30T15:51:10.605Z" }, + { url = "https://files.pythonhosted.org/packages/28/3d/35f33045e21034b388686213752cabc3a1b9d03e20969e6fa8f1b1d82db1/multidict-6.6.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:05db2f66c9addb10cfa226e1acb363450fab2ff8a6df73c622fefe2f5af6d4e7", size = 239351, upload-time = "2025-06-30T15:51:12.18Z" }, + { url = "https://files.pythonhosted.org/packages/6e/cc/ff84c03b95b430015d2166d9aae775a3985d757b94f6635010d0038d9241/multidict-6.6.3-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:0db58da8eafb514db832a1b44f8fa7906fdd102f7d982025f816a93ba45e3dcb", size = 233429, upload-time = "2025-06-30T15:51:13.533Z" }, + { url = "https://files.pythonhosted.org/packages/2e/f0/8cd49a0b37bdea673a4b793c2093f2f4ba8e7c9d6d7c9bd672fd6d38cd11/multidict-6.6.3-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:14117a41c8fdb3ee19c743b1c027da0736fdb79584d61a766da53d399b71176c", size = 243094, upload-time = "2025-06-30T15:51:14.815Z" }, + { url = "https://files.pythonhosted.org/packages/96/19/5d9a0cfdafe65d82b616a45ae950975820289069f885328e8185e64283c2/multidict-6.6.3-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:877443eaaabcd0b74ff32ebeed6f6176c71850feb7d6a1d2db65945256ea535c", size = 248957, upload-time = "2025-06-30T15:51:16.076Z" }, + { url = "https://files.pythonhosted.org/packages/e6/dc/c90066151da87d1e489f147b9b4327927241e65f1876702fafec6729c014/multidict-6.6.3-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:70b72e749a4f6e7ed8fb334fa8d8496384840319512746a5f42fa0aec79f4d61", size = 243590, upload-time = "2025-06-30T15:51:17.413Z" }, + { url = "https://files.pythonhosted.org/packages/ec/39/458afb0cccbb0ee9164365273be3e039efddcfcb94ef35924b7dbdb05db0/multidict-6.6.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:43571f785b86afd02b3855c5ac8e86ec921b760298d6f82ff2a61daf5a35330b", size = 237487, upload-time = "2025-06-30T15:51:19.039Z" }, + { url = "https://files.pythonhosted.org/packages/35/38/0016adac3990426610a081787011177e661875546b434f50a26319dc8372/multidict-6.6.3-cp310-cp310-win32.whl", hash = "sha256:20c5a0c3c13a15fd5ea86c42311859f970070e4e24de5a550e99d7c271d76318", size = 41390, upload-time = "2025-06-30T15:51:20.362Z" }, + { url = "https://files.pythonhosted.org/packages/f3/d2/17897a8f3f2c5363d969b4c635aa40375fe1f09168dc09a7826780bfb2a4/multidict-6.6.3-cp310-cp310-win_amd64.whl", hash = "sha256:ab0a34a007704c625e25a9116c6770b4d3617a071c8a7c30cd338dfbadfe6485", size = 45954, upload-time = "2025-06-30T15:51:21.383Z" }, + { url = "https://files.pythonhosted.org/packages/2d/5f/d4a717c1e457fe44072e33fa400d2b93eb0f2819c4d669381f925b7cba1f/multidict-6.6.3-cp310-cp310-win_arm64.whl", hash = "sha256:769841d70ca8bdd140a715746199fc6473414bd02efd678d75681d2d6a8986c5", size = 42981, upload-time = "2025-06-30T15:51:22.809Z" }, + { url = "https://files.pythonhosted.org/packages/08/f0/1a39863ced51f639c81a5463fbfa9eb4df59c20d1a8769ab9ef4ca57ae04/multidict-6.6.3-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:18f4eba0cbac3546b8ae31e0bbc55b02c801ae3cbaf80c247fcdd89b456ff58c", size = 76445, upload-time = "2025-06-30T15:51:24.01Z" }, + { url = "https://files.pythonhosted.org/packages/c9/0e/a7cfa451c7b0365cd844e90b41e21fab32edaa1e42fc0c9f68461ce44ed7/multidict-6.6.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ef43b5dd842382329e4797c46f10748d8c2b6e0614f46b4afe4aee9ac33159df", size = 44610, upload-time = "2025-06-30T15:51:25.158Z" }, + { url = "https://files.pythonhosted.org/packages/c6/bb/a14a4efc5ee748cc1904b0748be278c31b9295ce5f4d2ef66526f410b94d/multidict-6.6.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:bf9bd1fd5eec01494e0f2e8e446a74a85d5e49afb63d75a9934e4a5423dba21d", size = 44267, upload-time = "2025-06-30T15:51:26.326Z" }, + { url = "https://files.pythonhosted.org/packages/c2/f8/410677d563c2d55e063ef74fe578f9d53fe6b0a51649597a5861f83ffa15/multidict-6.6.3-cp311-cp311-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:5bd8d6f793a787153956cd35e24f60485bf0651c238e207b9a54f7458b16d539", size = 230004, upload-time = "2025-06-30T15:51:27.491Z" }, + { url = "https://files.pythonhosted.org/packages/fd/df/2b787f80059314a98e1ec6a4cc7576244986df3e56b3c755e6fc7c99e038/multidict-6.6.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1bf99b4daf908c73856bd87ee0a2499c3c9a3d19bb04b9c6025e66af3fd07462", size = 247196, upload-time = "2025-06-30T15:51:28.762Z" }, + { url = "https://files.pythonhosted.org/packages/05/f2/f9117089151b9a8ab39f9019620d10d9718eec2ac89e7ca9d30f3ec78e96/multidict-6.6.3-cp311-cp311-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:0b9e59946b49dafaf990fd9c17ceafa62976e8471a14952163d10a7a630413a9", size = 225337, upload-time = "2025-06-30T15:51:30.025Z" }, + { url = "https://files.pythonhosted.org/packages/93/2d/7115300ec5b699faa152c56799b089a53ed69e399c3c2d528251f0aeda1a/multidict-6.6.3-cp311-cp311-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:e2db616467070d0533832d204c54eea6836a5e628f2cb1e6dfd8cd6ba7277cb7", size = 257079, upload-time = "2025-06-30T15:51:31.716Z" }, + { url = "https://files.pythonhosted.org/packages/15/ea/ff4bab367623e39c20d3b07637225c7688d79e4f3cc1f3b9f89867677f9a/multidict-6.6.3-cp311-cp311-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:7394888236621f61dcdd25189b2768ae5cc280f041029a5bcf1122ac63df79f9", size = 255461, upload-time = "2025-06-30T15:51:33.029Z" }, + { url = "https://files.pythonhosted.org/packages/74/07/2c9246cda322dfe08be85f1b8739646f2c4c5113a1422d7a407763422ec4/multidict-6.6.3-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f114d8478733ca7388e7c7e0ab34b72547476b97009d643644ac33d4d3fe1821", size = 246611, upload-time = "2025-06-30T15:51:34.47Z" }, + { url = "https://files.pythonhosted.org/packages/a8/62/279c13d584207d5697a752a66ffc9bb19355a95f7659140cb1b3cf82180e/multidict-6.6.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cdf22e4db76d323bcdc733514bf732e9fb349707c98d341d40ebcc6e9318ef3d", size = 243102, upload-time = "2025-06-30T15:51:36.525Z" }, + { url = "https://files.pythonhosted.org/packages/69/cc/e06636f48c6d51e724a8bc8d9e1db5f136fe1df066d7cafe37ef4000f86a/multidict-6.6.3-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:e995a34c3d44ab511bfc11aa26869b9d66c2d8c799fa0e74b28a473a692532d6", size = 238693, upload-time = "2025-06-30T15:51:38.278Z" }, + { url = "https://files.pythonhosted.org/packages/89/a4/66c9d8fb9acf3b226cdd468ed009537ac65b520aebdc1703dd6908b19d33/multidict-6.6.3-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:766a4a5996f54361d8d5a9050140aa5362fe48ce51c755a50c0bc3706460c430", size = 246582, upload-time = "2025-06-30T15:51:39.709Z" }, + { url = "https://files.pythonhosted.org/packages/cf/01/c69e0317be556e46257826d5449feb4e6aa0d18573e567a48a2c14156f1f/multidict-6.6.3-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:3893a0d7d28a7fe6ca7a1f760593bc13038d1d35daf52199d431b61d2660602b", size = 253355, upload-time = "2025-06-30T15:51:41.013Z" }, + { url = "https://files.pythonhosted.org/packages/c0/da/9cc1da0299762d20e626fe0042e71b5694f9f72d7d3f9678397cbaa71b2b/multidict-6.6.3-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:934796c81ea996e61914ba58064920d6cad5d99140ac3167901eb932150e2e56", size = 247774, upload-time = "2025-06-30T15:51:42.291Z" }, + { url = "https://files.pythonhosted.org/packages/e6/91/b22756afec99cc31105ddd4a52f95ab32b1a4a58f4d417979c570c4a922e/multidict-6.6.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:9ed948328aec2072bc00f05d961ceadfd3e9bfc2966c1319aeaf7b7c21219183", size = 242275, upload-time = "2025-06-30T15:51:43.642Z" }, + { url = "https://files.pythonhosted.org/packages/be/f1/adcc185b878036a20399d5be5228f3cbe7f823d78985d101d425af35c800/multidict-6.6.3-cp311-cp311-win32.whl", hash = "sha256:9f5b28c074c76afc3e4c610c488e3493976fe0e596dd3db6c8ddfbb0134dcac5", size = 41290, upload-time = "2025-06-30T15:51:45.264Z" }, + { url = "https://files.pythonhosted.org/packages/e0/d4/27652c1c6526ea6b4f5ddd397e93f4232ff5de42bea71d339bc6a6cc497f/multidict-6.6.3-cp311-cp311-win_amd64.whl", hash = "sha256:bc7f6fbc61b1c16050a389c630da0b32fc6d4a3d191394ab78972bf5edc568c2", size = 45942, upload-time = "2025-06-30T15:51:46.377Z" }, + { url = "https://files.pythonhosted.org/packages/16/18/23f4932019804e56d3c2413e237f866444b774b0263bcb81df2fdecaf593/multidict-6.6.3-cp311-cp311-win_arm64.whl", hash = "sha256:d4e47d8faffaae822fb5cba20937c048d4f734f43572e7079298a6c39fb172cb", size = 42880, upload-time = "2025-06-30T15:51:47.561Z" }, + { url = "https://files.pythonhosted.org/packages/0e/a0/6b57988ea102da0623ea814160ed78d45a2645e4bbb499c2896d12833a70/multidict-6.6.3-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:056bebbeda16b2e38642d75e9e5310c484b7c24e3841dc0fb943206a72ec89d6", size = 76514, upload-time = "2025-06-30T15:51:48.728Z" }, + { url = "https://files.pythonhosted.org/packages/07/7a/d1e92665b0850c6c0508f101f9cf0410c1afa24973e1115fe9c6a185ebf7/multidict-6.6.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:e5f481cccb3c5c5e5de5d00b5141dc589c1047e60d07e85bbd7dea3d4580d63f", size = 45394, upload-time = "2025-06-30T15:51:49.986Z" }, + { url = "https://files.pythonhosted.org/packages/52/6f/dd104490e01be6ef8bf9573705d8572f8c2d2c561f06e3826b081d9e6591/multidict-6.6.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:10bea2ee839a759ee368b5a6e47787f399b41e70cf0c20d90dfaf4158dfb4e55", size = 43590, upload-time = "2025-06-30T15:51:51.331Z" }, + { url = "https://files.pythonhosted.org/packages/44/fe/06e0e01b1b0611e6581b7fd5a85b43dacc08b6cea3034f902f383b0873e5/multidict-6.6.3-cp312-cp312-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:2334cfb0fa9549d6ce2c21af2bfbcd3ac4ec3646b1b1581c88e3e2b1779ec92b", size = 237292, upload-time = "2025-06-30T15:51:52.584Z" }, + { url = "https://files.pythonhosted.org/packages/ce/71/4f0e558fb77696b89c233c1ee2d92f3e1d5459070a0e89153c9e9e804186/multidict-6.6.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b8fee016722550a2276ca2cb5bb624480e0ed2bd49125b2b73b7010b9090e888", size = 258385, upload-time = "2025-06-30T15:51:53.913Z" }, + { url = "https://files.pythonhosted.org/packages/e3/25/cca0e68228addad24903801ed1ab42e21307a1b4b6dd2cf63da5d3ae082a/multidict-6.6.3-cp312-cp312-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:e5511cb35f5c50a2db21047c875eb42f308c5583edf96bd8ebf7d770a9d68f6d", size = 242328, upload-time = "2025-06-30T15:51:55.672Z" }, + { url = "https://files.pythonhosted.org/packages/6e/a3/46f2d420d86bbcb8fe660b26a10a219871a0fbf4d43cb846a4031533f3e0/multidict-6.6.3-cp312-cp312-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:712b348f7f449948e0a6c4564a21c7db965af900973a67db432d724619b3c680", size = 268057, upload-time = "2025-06-30T15:51:57.037Z" }, + { url = "https://files.pythonhosted.org/packages/9e/73/1c743542fe00794a2ec7466abd3f312ccb8fad8dff9f36d42e18fb1ec33e/multidict-6.6.3-cp312-cp312-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:e4e15d2138ee2694e038e33b7c3da70e6b0ad8868b9f8094a72e1414aeda9c1a", size = 269341, upload-time = "2025-06-30T15:51:59.111Z" }, + { url = "https://files.pythonhosted.org/packages/a4/11/6ec9dcbe2264b92778eeb85407d1df18812248bf3506a5a1754bc035db0c/multidict-6.6.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8df25594989aebff8a130f7899fa03cbfcc5d2b5f4a461cf2518236fe6f15961", size = 256081, upload-time = "2025-06-30T15:52:00.533Z" }, + { url = "https://files.pythonhosted.org/packages/9b/2b/631b1e2afeb5f1696846d747d36cda075bfdc0bc7245d6ba5c319278d6c4/multidict-6.6.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:159ca68bfd284a8860f8d8112cf0521113bffd9c17568579e4d13d1f1dc76b65", size = 253581, upload-time = "2025-06-30T15:52:02.43Z" }, + { url = "https://files.pythonhosted.org/packages/bf/0e/7e3b93f79efeb6111d3bf9a1a69e555ba1d07ad1c11bceb56b7310d0d7ee/multidict-6.6.3-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:e098c17856a8c9ade81b4810888c5ad1914099657226283cab3062c0540b0643", size = 250750, upload-time = "2025-06-30T15:52:04.26Z" }, + { url = "https://files.pythonhosted.org/packages/ad/9e/086846c1d6601948e7de556ee464a2d4c85e33883e749f46b9547d7b0704/multidict-6.6.3-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:67c92ed673049dec52d7ed39f8cf9ebbadf5032c774058b4406d18c8f8fe7063", size = 251548, upload-time = "2025-06-30T15:52:06.002Z" }, + { url = "https://files.pythonhosted.org/packages/8c/7b/86ec260118e522f1a31550e87b23542294880c97cfbf6fb18cc67b044c66/multidict-6.6.3-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:bd0578596e3a835ef451784053cfd327d607fc39ea1a14812139339a18a0dbc3", size = 262718, upload-time = "2025-06-30T15:52:07.707Z" }, + { url = "https://files.pythonhosted.org/packages/8c/bd/22ce8f47abb0be04692c9fc4638508b8340987b18691aa7775d927b73f72/multidict-6.6.3-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:346055630a2df2115cd23ae271910b4cae40f4e336773550dca4889b12916e75", size = 259603, upload-time = "2025-06-30T15:52:09.58Z" }, + { url = "https://files.pythonhosted.org/packages/07/9c/91b7ac1691be95cd1f4a26e36a74b97cda6aa9820632d31aab4410f46ebd/multidict-6.6.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:555ff55a359302b79de97e0468e9ee80637b0de1fce77721639f7cd9440b3a10", size = 251351, upload-time = "2025-06-30T15:52:10.947Z" }, + { url = "https://files.pythonhosted.org/packages/6f/5c/4d7adc739884f7a9fbe00d1eac8c034023ef8bad71f2ebe12823ca2e3649/multidict-6.6.3-cp312-cp312-win32.whl", hash = "sha256:73ab034fb8d58ff85c2bcbadc470efc3fafeea8affcf8722855fb94557f14cc5", size = 41860, upload-time = "2025-06-30T15:52:12.334Z" }, + { url = "https://files.pythonhosted.org/packages/6a/a3/0fbc7afdf7cb1aa12a086b02959307848eb6bcc8f66fcb66c0cb57e2a2c1/multidict-6.6.3-cp312-cp312-win_amd64.whl", hash = "sha256:04cbcce84f63b9af41bad04a54d4cc4e60e90c35b9e6ccb130be2d75b71f8c17", size = 45982, upload-time = "2025-06-30T15:52:13.6Z" }, + { url = "https://files.pythonhosted.org/packages/b8/95/8c825bd70ff9b02462dc18d1295dd08d3e9e4eb66856d292ffa62cfe1920/multidict-6.6.3-cp312-cp312-win_arm64.whl", hash = "sha256:0f1130b896ecb52d2a1e615260f3ea2af55fa7dc3d7c3003ba0c3121a759b18b", size = 43210, upload-time = "2025-06-30T15:52:14.893Z" }, + { url = "https://files.pythonhosted.org/packages/52/1d/0bebcbbb4f000751fbd09957257903d6e002943fc668d841a4cf2fb7f872/multidict-6.6.3-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:540d3c06d48507357a7d57721e5094b4f7093399a0106c211f33540fdc374d55", size = 75843, upload-time = "2025-06-30T15:52:16.155Z" }, + { url = "https://files.pythonhosted.org/packages/07/8f/cbe241b0434cfe257f65c2b1bcf9e8d5fb52bc708c5061fb29b0fed22bdf/multidict-6.6.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9c19cea2a690f04247d43f366d03e4eb110a0dc4cd1bbeee4d445435428ed35b", size = 45053, upload-time = "2025-06-30T15:52:17.429Z" }, + { url = "https://files.pythonhosted.org/packages/32/d2/0b3b23f9dbad5b270b22a3ac3ea73ed0a50ef2d9a390447061178ed6bdb8/multidict-6.6.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:7af039820cfd00effec86bda5d8debef711a3e86a1d3772e85bea0f243a4bd65", size = 43273, upload-time = "2025-06-30T15:52:19.346Z" }, + { url = "https://files.pythonhosted.org/packages/fd/fe/6eb68927e823999e3683bc49678eb20374ba9615097d085298fd5b386564/multidict-6.6.3-cp313-cp313-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:500b84f51654fdc3944e936f2922114349bf8fdcac77c3092b03449f0e5bc2b3", size = 237124, upload-time = "2025-06-30T15:52:20.773Z" }, + { url = "https://files.pythonhosted.org/packages/e7/ab/320d8507e7726c460cb77117848b3834ea0d59e769f36fdae495f7669929/multidict-6.6.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f3fc723ab8a5c5ed6c50418e9bfcd8e6dceba6c271cee6728a10a4ed8561520c", size = 256892, upload-time = "2025-06-30T15:52:22.242Z" }, + { url = "https://files.pythonhosted.org/packages/76/60/38ee422db515ac69834e60142a1a69111ac96026e76e8e9aa347fd2e4591/multidict-6.6.3-cp313-cp313-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:94c47ea3ade005b5976789baaed66d4de4480d0a0bf31cef6edaa41c1e7b56a6", size = 240547, upload-time = "2025-06-30T15:52:23.736Z" }, + { url = "https://files.pythonhosted.org/packages/27/fb/905224fde2dff042b030c27ad95a7ae744325cf54b890b443d30a789b80e/multidict-6.6.3-cp313-cp313-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:dbc7cf464cc6d67e83e136c9f55726da3a30176f020a36ead246eceed87f1cd8", size = 266223, upload-time = "2025-06-30T15:52:25.185Z" }, + { url = "https://files.pythonhosted.org/packages/76/35/dc38ab361051beae08d1a53965e3e1a418752fc5be4d3fb983c5582d8784/multidict-6.6.3-cp313-cp313-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:900eb9f9da25ada070f8ee4a23f884e0ee66fe4e1a38c3af644256a508ad81ca", size = 267262, upload-time = "2025-06-30T15:52:26.969Z" }, + { url = "https://files.pythonhosted.org/packages/1f/a3/0a485b7f36e422421b17e2bbb5a81c1af10eac1d4476f2ff92927c730479/multidict-6.6.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7c6df517cf177da5d47ab15407143a89cd1a23f8b335f3a28d57e8b0a3dbb884", size = 254345, upload-time = "2025-06-30T15:52:28.467Z" }, + { url = "https://files.pythonhosted.org/packages/b4/59/bcdd52c1dab7c0e0d75ff19cac751fbd5f850d1fc39172ce809a74aa9ea4/multidict-6.6.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:4ef421045f13879e21c994b36e728d8e7d126c91a64b9185810ab51d474f27e7", size = 252248, upload-time = "2025-06-30T15:52:29.938Z" }, + { url = "https://files.pythonhosted.org/packages/bb/a4/2d96aaa6eae8067ce108d4acee6f45ced5728beda55c0f02ae1072c730d1/multidict-6.6.3-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:6c1e61bb4f80895c081790b6b09fa49e13566df8fbff817da3f85b3a8192e36b", size = 250115, upload-time = "2025-06-30T15:52:31.416Z" }, + { url = "https://files.pythonhosted.org/packages/25/d2/ed9f847fa5c7d0677d4f02ea2c163d5e48573de3f57bacf5670e43a5ffaa/multidict-6.6.3-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:e5e8523bb12d7623cd8300dbd91b9e439a46a028cd078ca695eb66ba31adee3c", size = 249649, upload-time = "2025-06-30T15:52:32.996Z" }, + { url = "https://files.pythonhosted.org/packages/1f/af/9155850372563fc550803d3f25373308aa70f59b52cff25854086ecb4a79/multidict-6.6.3-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:ef58340cc896219e4e653dade08fea5c55c6df41bcc68122e3be3e9d873d9a7b", size = 261203, upload-time = "2025-06-30T15:52:34.521Z" }, + { url = "https://files.pythonhosted.org/packages/36/2f/c6a728f699896252cf309769089568a33c6439626648843f78743660709d/multidict-6.6.3-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:fc9dc435ec8699e7b602b94fe0cd4703e69273a01cbc34409af29e7820f777f1", size = 258051, upload-time = "2025-06-30T15:52:35.999Z" }, + { url = "https://files.pythonhosted.org/packages/d0/60/689880776d6b18fa2b70f6cc74ff87dd6c6b9b47bd9cf74c16fecfaa6ad9/multidict-6.6.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9e864486ef4ab07db5e9cb997bad2b681514158d6954dd1958dfb163b83d53e6", size = 249601, upload-time = "2025-06-30T15:52:37.473Z" }, + { url = "https://files.pythonhosted.org/packages/75/5e/325b11f2222a549019cf2ef879c1f81f94a0d40ace3ef55cf529915ba6cc/multidict-6.6.3-cp313-cp313-win32.whl", hash = "sha256:5633a82fba8e841bc5c5c06b16e21529573cd654f67fd833650a215520a6210e", size = 41683, upload-time = "2025-06-30T15:52:38.927Z" }, + { url = "https://files.pythonhosted.org/packages/b1/ad/cf46e73f5d6e3c775cabd2a05976547f3f18b39bee06260369a42501f053/multidict-6.6.3-cp313-cp313-win_amd64.whl", hash = "sha256:e93089c1570a4ad54c3714a12c2cef549dc9d58e97bcded193d928649cab78e9", size = 45811, upload-time = "2025-06-30T15:52:40.207Z" }, + { url = "https://files.pythonhosted.org/packages/c5/c9/2e3fe950db28fb7c62e1a5f46e1e38759b072e2089209bc033c2798bb5ec/multidict-6.6.3-cp313-cp313-win_arm64.whl", hash = "sha256:c60b401f192e79caec61f166da9c924e9f8bc65548d4246842df91651e83d600", size = 43056, upload-time = "2025-06-30T15:52:41.575Z" }, + { url = "https://files.pythonhosted.org/packages/3a/58/aaf8114cf34966e084a8cc9517771288adb53465188843d5a19862cb6dc3/multidict-6.6.3-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:02fd8f32d403a6ff13864b0851f1f523d4c988051eea0471d4f1fd8010f11134", size = 82811, upload-time = "2025-06-30T15:52:43.281Z" }, + { url = "https://files.pythonhosted.org/packages/71/af/5402e7b58a1f5b987a07ad98f2501fdba2a4f4b4c30cf114e3ce8db64c87/multidict-6.6.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:f3aa090106b1543f3f87b2041eef3c156c8da2aed90c63a2fbed62d875c49c37", size = 48304, upload-time = "2025-06-30T15:52:45.026Z" }, + { url = "https://files.pythonhosted.org/packages/39/65/ab3c8cafe21adb45b24a50266fd747147dec7847425bc2a0f6934b3ae9ce/multidict-6.6.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:e924fb978615a5e33ff644cc42e6aa241effcf4f3322c09d4f8cebde95aff5f8", size = 46775, upload-time = "2025-06-30T15:52:46.459Z" }, + { url = "https://files.pythonhosted.org/packages/49/ba/9fcc1b332f67cc0c0c8079e263bfab6660f87fe4e28a35921771ff3eea0d/multidict-6.6.3-cp313-cp313t-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:b9fe5a0e57c6dbd0e2ce81ca66272282c32cd11d31658ee9553849d91289e1c1", size = 229773, upload-time = "2025-06-30T15:52:47.88Z" }, + { url = "https://files.pythonhosted.org/packages/a4/14/0145a251f555f7c754ce2dcbcd012939bbd1f34f066fa5d28a50e722a054/multidict-6.6.3-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b24576f208793ebae00280c59927c3b7c2a3b1655e443a25f753c4611bc1c373", size = 250083, upload-time = "2025-06-30T15:52:49.366Z" }, + { url = "https://files.pythonhosted.org/packages/9e/d4/d5c0bd2bbb173b586c249a151a26d2fb3ec7d53c96e42091c9fef4e1f10c/multidict-6.6.3-cp313-cp313t-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:135631cb6c58eac37d7ac0df380294fecdc026b28837fa07c02e459c7fb9c54e", size = 228980, upload-time = "2025-06-30T15:52:50.903Z" }, + { url = "https://files.pythonhosted.org/packages/21/32/c9a2d8444a50ec48c4733ccc67254100c10e1c8ae8e40c7a2d2183b59b97/multidict-6.6.3-cp313-cp313t-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:274d416b0df887aef98f19f21578653982cfb8a05b4e187d4a17103322eeaf8f", size = 257776, upload-time = "2025-06-30T15:52:52.764Z" }, + { url = "https://files.pythonhosted.org/packages/68/d0/14fa1699f4ef629eae08ad6201c6b476098f5efb051b296f4c26be7a9fdf/multidict-6.6.3-cp313-cp313t-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:e252017a817fad7ce05cafbe5711ed40faeb580e63b16755a3a24e66fa1d87c0", size = 256882, upload-time = "2025-06-30T15:52:54.596Z" }, + { url = "https://files.pythonhosted.org/packages/da/88/84a27570fbe303c65607d517a5f147cd2fc046c2d1da02b84b17b9bdc2aa/multidict-6.6.3-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2e4cc8d848cd4fe1cdee28c13ea79ab0ed37fc2e89dd77bac86a2e7959a8c3bc", size = 247816, upload-time = "2025-06-30T15:52:56.175Z" }, + { url = "https://files.pythonhosted.org/packages/1c/60/dca352a0c999ce96a5d8b8ee0b2b9f729dcad2e0b0c195f8286269a2074c/multidict-6.6.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:9e236a7094b9c4c1b7585f6b9cca34b9d833cf079f7e4c49e6a4a6ec9bfdc68f", size = 245341, upload-time = "2025-06-30T15:52:57.752Z" }, + { url = "https://files.pythonhosted.org/packages/50/ef/433fa3ed06028f03946f3993223dada70fb700f763f70c00079533c34578/multidict-6.6.3-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:e0cb0ab69915c55627c933f0b555a943d98ba71b4d1c57bc0d0a66e2567c7471", size = 235854, upload-time = "2025-06-30T15:52:59.74Z" }, + { url = "https://files.pythonhosted.org/packages/1b/1f/487612ab56fbe35715320905215a57fede20de7db40a261759690dc80471/multidict-6.6.3-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:81ef2f64593aba09c5212a3d0f8c906a0d38d710a011f2f42759704d4557d3f2", size = 243432, upload-time = "2025-06-30T15:53:01.602Z" }, + { url = "https://files.pythonhosted.org/packages/da/6f/ce8b79de16cd885c6f9052c96a3671373d00c59b3ee635ea93e6e81b8ccf/multidict-6.6.3-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:b9cbc60010de3562545fa198bfc6d3825df430ea96d2cc509c39bd71e2e7d648", size = 252731, upload-time = "2025-06-30T15:53:03.517Z" }, + { url = "https://files.pythonhosted.org/packages/bb/fe/a2514a6aba78e5abefa1624ca85ae18f542d95ac5cde2e3815a9fbf369aa/multidict-6.6.3-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:70d974eaaa37211390cd02ef93b7e938de564bbffa866f0b08d07e5e65da783d", size = 247086, upload-time = "2025-06-30T15:53:05.48Z" }, + { url = "https://files.pythonhosted.org/packages/8c/22/b788718d63bb3cce752d107a57c85fcd1a212c6c778628567c9713f9345a/multidict-6.6.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:3713303e4a6663c6d01d648a68f2848701001f3390a030edaaf3fc949c90bf7c", size = 243338, upload-time = "2025-06-30T15:53:07.522Z" }, + { url = "https://files.pythonhosted.org/packages/22/d6/fdb3d0670819f2228f3f7d9af613d5e652c15d170c83e5f1c94fbc55a25b/multidict-6.6.3-cp313-cp313t-win32.whl", hash = "sha256:639ecc9fe7cd73f2495f62c213e964843826f44505a3e5d82805aa85cac6f89e", size = 47812, upload-time = "2025-06-30T15:53:09.263Z" }, + { url = "https://files.pythonhosted.org/packages/b6/d6/a9d2c808f2c489ad199723197419207ecbfbc1776f6e155e1ecea9c883aa/multidict-6.6.3-cp313-cp313t-win_amd64.whl", hash = "sha256:9f97e181f344a0ef3881b573d31de8542cc0dbc559ec68c8f8b5ce2c2e91646d", size = 53011, upload-time = "2025-06-30T15:53:11.038Z" }, + { url = "https://files.pythonhosted.org/packages/f2/40/b68001cba8188dd267590a111f9661b6256debc327137667e832bf5d66e8/multidict-6.6.3-cp313-cp313t-win_arm64.whl", hash = "sha256:ce8b7693da41a3c4fde5871c738a81490cea5496c671d74374c8ab889e1834fb", size = 45254, upload-time = "2025-06-30T15:53:12.421Z" }, + { url = "https://files.pythonhosted.org/packages/d8/30/9aec301e9772b098c1f5c0ca0279237c9766d94b97802e9888010c64b0ed/multidict-6.6.3-py3-none-any.whl", hash = "sha256:8db10f29c7541fc5da4defd8cd697e1ca429db743fa716325f236079b96f775a", size = 12313, upload-time = "2025-06-30T15:53:45.437Z" }, ] [[package]] name = "narwhals" -version = "1.31.0" +version = "2.0.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/36/fa/c2b6a4d5dbc4af15aa58c86920d5275a9c65142318179b246685069f57da/narwhals-1.31.0.tar.gz", hash = "sha256:333472e2562343dfdd27407ec9b5114a07c81d0416794e4ac6b703dd925c6a1a", size = 253463, upload-time = "2025-03-17T15:26:26.065Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e2/8f/51d14e402c4f9d281a2e153a6a805cad5460088027a999faf264b54e7641/narwhals-2.0.1.tar.gz", hash = "sha256:235e61ca807bc21110ca36a4d53888ecc22c42dcdf50a7c886e10dde3fd7f38c", size = 525541, upload-time = "2025-07-29T08:39:04.81Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/f9/c0/fb39bd876ea2fd9509343d643690cd2f9715e6a77271e7c7b26f1eea70c1/narwhals-1.31.0-py3-none-any.whl", hash = "sha256:2a7b79bb5f511055c4c0142121fc0d4171ea171458e12d44dbd9c8fc6488e997", size = 313124, upload-time = "2025-03-17T15:26:23.87Z" }, + { url = "https://files.pythonhosted.org/packages/7f/26/43caf834e47c63883a5eddc02893b7fdbe6a0a4508ff6dc401907f3cc085/narwhals-2.0.1-py3-none-any.whl", hash = "sha256:837457e36a2ba1710c881fb69e1f79ce44fb81728c92ac378f70892a53af8ddb", size = 385436, upload-time = "2025-07-29T08:39:03.163Z" }, ] [[package]] @@ -1551,9 +1984,9 @@ name = "netcdf4" version = "1.7.2" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "certifi" }, - { name = "cftime" }, - { name = "numpy" }, + { name = "certifi", marker = "python_full_version < '3.11'" }, + { name = "cftime", marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/71/ed/4d27fcfa40ebfdad3d2088a3de7ee48dbff7f35163e815ec1870d2a7398c/netcdf4-1.7.2.tar.gz", hash = "sha256:a4c6375540b19989896136943abb6d44850ff6f1fa7d3f063253b1ad3f8b7fce", size = 835064, upload-time = "2024-10-22T19:01:25.521Z" } wheels = [ @@ -1583,11 +2016,32 @@ wheels = [ name = "networkx" version = "3.4.2" source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'linux'", + "python_full_version < '3.11' and sys_platform != 'linux'", +] sdist = { url = "https://files.pythonhosted.org/packages/fd/1d/06475e1cd5264c0b870ea2cc6fdb3e37177c1e565c43f56ff17a10e3937f/networkx-3.4.2.tar.gz", hash = "sha256:307c3669428c5362aab27c8a1260aa8f47c4e91d3891f48be0141738d8d053e1", size = 2151368, upload-time = "2024-10-21T12:39:38.695Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/b9/54/dd730b32ea14ea797530a4479b2ed46a6fb250f682a9cfb997e968bf0261/networkx-3.4.2-py3-none-any.whl", hash = "sha256:df5d4365b724cf81b8c6a7312509d0c22386097011ad1abe274afd5e9d3bbc5f", size = 1723263, upload-time = "2024-10-21T12:39:36.247Z" }, ] +[[package]] +name = "networkx" +version = "3.5" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", + "python_full_version == '3.11.*' and sys_platform == 'linux'", + "python_full_version == '3.11.*' and sys_platform != 'linux'", +] +sdist = { url = "https://files.pythonhosted.org/packages/6c/4f/ccdb8ad3a38e583f214547fd2f7ff1fc160c43a75af88e6aec213404b96a/networkx-3.5.tar.gz", hash = "sha256:d4c6f9cf81f52d69230866796b82afbccdec3db7ae4fbd1b65ea750feed50037", size = 2471065, upload-time = "2025-05-29T11:35:07.804Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/eb/8d/776adee7bbf76365fdd7f2552710282c79a4ead5d2a46408c9043a2b70ba/networkx-3.5-py3-none-any.whl", hash = "sha256:0030d386a9a06dee3565298b4a734b68589749a544acbb6c412dc9e2489ec6ec", size = 2034406, upload-time = "2025-05-29T11:35:04.961Z" }, +] + [[package]] name = "ninja" version = "1.11.1.4" @@ -1623,64 +2077,157 @@ wheels = [ [[package]] name = "numpy" -version = "2.2.4" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/e1/78/31103410a57bc2c2b93a3597340a8119588571f6a4539067546cb9a0bfac/numpy-2.2.4.tar.gz", hash = "sha256:9ba03692a45d3eef66559efe1d1096c4b9b75c0986b5dff5530c378fb8331d4f", size = 20270701, upload-time = "2025-03-16T18:27:00.648Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/04/89/a79e86e5c1433926ed7d60cb267fb64aa578b6101ab645800fd43b4801de/numpy-2.2.4-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:8146f3550d627252269ac42ae660281d673eb6f8b32f113538e0cc2a9aed42b9", size = 21250661, upload-time = "2025-03-16T18:02:13.017Z" }, - { url = "https://files.pythonhosted.org/packages/79/c2/f50921beb8afd60ed9589ad880332cfefdb805422210d327fb48f12b7a81/numpy-2.2.4-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:e642d86b8f956098b564a45e6f6ce68a22c2c97a04f5acd3f221f57b8cb850ae", size = 14389926, upload-time = "2025-03-16T18:02:39.022Z" }, - { url = "https://files.pythonhosted.org/packages/c7/b9/2c4e96130b0b0f97b0ef4a06d6dae3b39d058b21a5e2fa2decd7fd6b1c8f/numpy-2.2.4-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:a84eda42bd12edc36eb5b53bbcc9b406820d3353f1994b6cfe453a33ff101775", size = 5428329, upload-time = "2025-03-16T18:02:50.032Z" }, - { url = "https://files.pythonhosted.org/packages/7f/a5/3d7094aa898f4fc5c84cdfb26beeae780352d43f5d8bdec966c4393d644c/numpy-2.2.4-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:4ba5054787e89c59c593a4169830ab362ac2bee8a969249dc56e5d7d20ff8df9", size = 6963559, upload-time = "2025-03-16T18:03:02.523Z" }, - { url = "https://files.pythonhosted.org/packages/4c/22/fb1be710a14434c09080dd4a0acc08939f612ec02efcb04b9e210474782d/numpy-2.2.4-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7716e4a9b7af82c06a2543c53ca476fa0b57e4d760481273e09da04b74ee6ee2", size = 14368066, upload-time = "2025-03-16T18:03:27.146Z" }, - { url = "https://files.pythonhosted.org/packages/c2/07/2e5cc71193e3ef3a219ffcf6ca4858e46ea2be09c026ddd480d596b32867/numpy-2.2.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:adf8c1d66f432ce577d0197dceaac2ac00c0759f573f28516246351c58a85020", size = 16417040, upload-time = "2025-03-16T18:03:55.999Z" }, - { url = "https://files.pythonhosted.org/packages/1a/97/3b1537776ad9a6d1a41813818343745e8dd928a2916d4c9edcd9a8af1dac/numpy-2.2.4-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:218f061d2faa73621fa23d6359442b0fc658d5b9a70801373625d958259eaca3", size = 15879862, upload-time = "2025-03-16T18:04:23.56Z" }, - { url = "https://files.pythonhosted.org/packages/b0/b7/4472f603dd45ef36ff3d8e84e84fe02d9467c78f92cc121633dce6da307b/numpy-2.2.4-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:df2f57871a96bbc1b69733cd4c51dc33bea66146b8c63cacbfed73eec0883017", size = 18206032, upload-time = "2025-03-16T18:04:53.694Z" }, - { url = "https://files.pythonhosted.org/packages/0d/bd/6a092963fb82e6c5aa0d0440635827bbb2910da229545473bbb58c537ed3/numpy-2.2.4-cp310-cp310-win32.whl", hash = "sha256:a0258ad1f44f138b791327961caedffbf9612bfa504ab9597157806faa95194a", size = 6608517, upload-time = "2025-03-16T18:05:06.647Z" }, - { url = "https://files.pythonhosted.org/packages/01/e3/cb04627bc2a1638948bc13e818df26495aa18e20d5be1ed95ab2b10b6847/numpy-2.2.4-cp310-cp310-win_amd64.whl", hash = "sha256:0d54974f9cf14acf49c60f0f7f4084b6579d24d439453d5fc5805d46a165b542", size = 12943498, upload-time = "2025-03-16T18:05:28.591Z" }, - { url = "https://files.pythonhosted.org/packages/16/fb/09e778ee3a8ea0d4dc8329cca0a9c9e65fed847d08e37eba74cb7ed4b252/numpy-2.2.4-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:e9e0a277bb2eb5d8a7407e14688b85fd8ad628ee4e0c7930415687b6564207a4", size = 21254989, upload-time = "2025-03-16T18:06:04.092Z" }, - { url = "https://files.pythonhosted.org/packages/a2/0a/1212befdbecab5d80eca3cde47d304cad986ad4eec7d85a42e0b6d2cc2ef/numpy-2.2.4-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:9eeea959168ea555e556b8188da5fa7831e21d91ce031e95ce23747b7609f8a4", size = 14425910, upload-time = "2025-03-16T18:06:29.062Z" }, - { url = "https://files.pythonhosted.org/packages/2b/3e/e7247c1d4f15086bb106c8d43c925b0b2ea20270224f5186fa48d4fb5cbd/numpy-2.2.4-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:bd3ad3b0a40e713fc68f99ecfd07124195333f1e689387c180813f0e94309d6f", size = 5426490, upload-time = "2025-03-16T18:06:39.901Z" }, - { url = "https://files.pythonhosted.org/packages/5d/fa/aa7cd6be51419b894c5787a8a93c3302a1ed4f82d35beb0613ec15bdd0e2/numpy-2.2.4-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:cf28633d64294969c019c6df4ff37f5698e8326db68cc2b66576a51fad634880", size = 6967754, upload-time = "2025-03-16T18:06:52.658Z" }, - { url = "https://files.pythonhosted.org/packages/d5/ee/96457c943265de9fadeb3d2ffdbab003f7fba13d971084a9876affcda095/numpy-2.2.4-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2fa8fa7697ad1646b5c93de1719965844e004fcad23c91228aca1cf0800044a1", size = 14373079, upload-time = "2025-03-16T18:07:16.297Z" }, - { url = "https://files.pythonhosted.org/packages/c5/5c/ceefca458559f0ccc7a982319f37ed07b0d7b526964ae6cc61f8ad1b6119/numpy-2.2.4-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f4162988a360a29af158aeb4a2f4f09ffed6a969c9776f8f3bdee9b06a8ab7e5", size = 16428819, upload-time = "2025-03-16T18:07:44.188Z" }, - { url = "https://files.pythonhosted.org/packages/22/31/9b2ac8eee99e001eb6add9fa27514ef5e9faf176169057a12860af52704c/numpy-2.2.4-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:892c10d6a73e0f14935c31229e03325a7b3093fafd6ce0af704be7f894d95687", size = 15881470, upload-time = "2025-03-16T18:08:11.545Z" }, - { url = "https://files.pythonhosted.org/packages/f0/dc/8569b5f25ff30484b555ad8a3f537e0225d091abec386c9420cf5f7a2976/numpy-2.2.4-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:db1f1c22173ac1c58db249ae48aa7ead29f534b9a948bc56828337aa84a32ed6", size = 18218144, upload-time = "2025-03-16T18:08:42.042Z" }, - { url = "https://files.pythonhosted.org/packages/5e/05/463c023a39bdeb9bb43a99e7dee2c664cb68d5bb87d14f92482b9f6011cc/numpy-2.2.4-cp311-cp311-win32.whl", hash = "sha256:ea2bb7e2ae9e37d96835b3576a4fa4b3a97592fbea8ef7c3587078b0068b8f09", size = 6606368, upload-time = "2025-03-16T18:08:55.074Z" }, - { url = "https://files.pythonhosted.org/packages/8b/72/10c1d2d82101c468a28adc35de6c77b308f288cfd0b88e1070f15b98e00c/numpy-2.2.4-cp311-cp311-win_amd64.whl", hash = "sha256:f7de08cbe5551911886d1ab60de58448c6df0f67d9feb7d1fb21e9875ef95e91", size = 12947526, upload-time = "2025-03-16T18:09:16.844Z" }, - { url = "https://files.pythonhosted.org/packages/a2/30/182db21d4f2a95904cec1a6f779479ea1ac07c0647f064dea454ec650c42/numpy-2.2.4-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:a7b9084668aa0f64e64bd00d27ba5146ef1c3a8835f3bd912e7a9e01326804c4", size = 20947156, upload-time = "2025-03-16T18:09:51.975Z" }, - { url = "https://files.pythonhosted.org/packages/24/6d/9483566acfbda6c62c6bc74b6e981c777229d2af93c8eb2469b26ac1b7bc/numpy-2.2.4-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:dbe512c511956b893d2dacd007d955a3f03d555ae05cfa3ff1c1ff6df8851854", size = 14133092, upload-time = "2025-03-16T18:10:16.329Z" }, - { url = "https://files.pythonhosted.org/packages/27/f6/dba8a258acbf9d2bed2525cdcbb9493ef9bae5199d7a9cb92ee7e9b2aea6/numpy-2.2.4-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:bb649f8b207ab07caebba230d851b579a3c8711a851d29efe15008e31bb4de24", size = 5163515, upload-time = "2025-03-16T18:10:26.19Z" }, - { url = "https://files.pythonhosted.org/packages/62/30/82116199d1c249446723c68f2c9da40d7f062551036f50b8c4caa42ae252/numpy-2.2.4-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:f34dc300df798742b3d06515aa2a0aee20941c13579d7a2f2e10af01ae4901ee", size = 6696558, upload-time = "2025-03-16T18:10:38.996Z" }, - { url = "https://files.pythonhosted.org/packages/0e/b2/54122b3c6df5df3e87582b2e9430f1bdb63af4023c739ba300164c9ae503/numpy-2.2.4-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c3f7ac96b16955634e223b579a3e5798df59007ca43e8d451a0e6a50f6bfdfba", size = 14084742, upload-time = "2025-03-16T18:11:02.76Z" }, - { url = "https://files.pythonhosted.org/packages/02/e2/e2cbb8d634151aab9528ef7b8bab52ee4ab10e076509285602c2a3a686e0/numpy-2.2.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4f92084defa704deadd4e0a5ab1dc52d8ac9e8a8ef617f3fbb853e79b0ea3592", size = 16134051, upload-time = "2025-03-16T18:11:32.767Z" }, - { url = "https://files.pythonhosted.org/packages/8e/21/efd47800e4affc993e8be50c1b768de038363dd88865920439ef7b422c60/numpy-2.2.4-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7a4e84a6283b36632e2a5b56e121961f6542ab886bc9e12f8f9818b3c266bfbb", size = 15578972, upload-time = "2025-03-16T18:11:59.877Z" }, - { url = "https://files.pythonhosted.org/packages/04/1e/f8bb88f6157045dd5d9b27ccf433d016981032690969aa5c19e332b138c0/numpy-2.2.4-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:11c43995255eb4127115956495f43e9343736edb7fcdb0d973defd9de14cd84f", size = 17898106, upload-time = "2025-03-16T18:12:31.487Z" }, - { url = "https://files.pythonhosted.org/packages/2b/93/df59a5a3897c1f036ae8ff845e45f4081bb06943039ae28a3c1c7c780f22/numpy-2.2.4-cp312-cp312-win32.whl", hash = "sha256:65ef3468b53269eb5fdb3a5c09508c032b793da03251d5f8722b1194f1790c00", size = 6311190, upload-time = "2025-03-16T18:12:44.46Z" }, - { url = "https://files.pythonhosted.org/packages/46/69/8c4f928741c2a8efa255fdc7e9097527c6dc4e4df147e3cadc5d9357ce85/numpy-2.2.4-cp312-cp312-win_amd64.whl", hash = "sha256:2aad3c17ed2ff455b8eaafe06bcdae0062a1db77cb99f4b9cbb5f4ecb13c5146", size = 12644305, upload-time = "2025-03-16T18:13:06.864Z" }, - { url = "https://files.pythonhosted.org/packages/2a/d0/bd5ad792e78017f5decfb2ecc947422a3669a34f775679a76317af671ffc/numpy-2.2.4-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:1cf4e5c6a278d620dee9ddeb487dc6a860f9b199eadeecc567f777daace1e9e7", size = 20933623, upload-time = "2025-03-16T18:13:43.231Z" }, - { url = "https://files.pythonhosted.org/packages/c3/bc/2b3545766337b95409868f8e62053135bdc7fa2ce630aba983a2aa60b559/numpy-2.2.4-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1974afec0b479e50438fc3648974268f972e2d908ddb6d7fb634598cdb8260a0", size = 14148681, upload-time = "2025-03-16T18:14:08.031Z" }, - { url = "https://files.pythonhosted.org/packages/6a/70/67b24d68a56551d43a6ec9fe8c5f91b526d4c1a46a6387b956bf2d64744e/numpy-2.2.4-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:79bd5f0a02aa16808fcbc79a9a376a147cc1045f7dfe44c6e7d53fa8b8a79392", size = 5148759, upload-time = "2025-03-16T18:14:18.613Z" }, - { url = "https://files.pythonhosted.org/packages/1c/8b/e2fc8a75fcb7be12d90b31477c9356c0cbb44abce7ffb36be39a0017afad/numpy-2.2.4-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:3387dd7232804b341165cedcb90694565a6015433ee076c6754775e85d86f1fc", size = 6683092, upload-time = "2025-03-16T18:14:31.386Z" }, - { url = "https://files.pythonhosted.org/packages/13/73/41b7b27f169ecf368b52533edb72e56a133f9e86256e809e169362553b49/numpy-2.2.4-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6f527d8fdb0286fd2fd97a2a96c6be17ba4232da346931d967a0630050dfd298", size = 14081422, upload-time = "2025-03-16T18:14:54.83Z" }, - { url = "https://files.pythonhosted.org/packages/4b/04/e208ff3ae3ddfbafc05910f89546382f15a3f10186b1f56bd99f159689c2/numpy-2.2.4-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bce43e386c16898b91e162e5baaad90c4b06f9dcbe36282490032cec98dc8ae7", size = 16132202, upload-time = "2025-03-16T18:15:22.035Z" }, - { url = "https://files.pythonhosted.org/packages/fe/bc/2218160574d862d5e55f803d88ddcad88beff94791f9c5f86d67bd8fbf1c/numpy-2.2.4-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:31504f970f563d99f71a3512d0c01a645b692b12a63630d6aafa0939e52361e6", size = 15573131, upload-time = "2025-03-16T18:15:48.546Z" }, - { url = "https://files.pythonhosted.org/packages/a5/78/97c775bc4f05abc8a8426436b7cb1be806a02a2994b195945600855e3a25/numpy-2.2.4-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:81413336ef121a6ba746892fad881a83351ee3e1e4011f52e97fba79233611fd", size = 17894270, upload-time = "2025-03-16T18:16:20.274Z" }, - { url = "https://files.pythonhosted.org/packages/b9/eb/38c06217a5f6de27dcb41524ca95a44e395e6a1decdc0c99fec0832ce6ae/numpy-2.2.4-cp313-cp313-win32.whl", hash = "sha256:f486038e44caa08dbd97275a9a35a283a8f1d2f0ee60ac260a1790e76660833c", size = 6308141, upload-time = "2025-03-16T18:20:15.297Z" }, - { url = "https://files.pythonhosted.org/packages/52/17/d0dd10ab6d125c6d11ffb6dfa3423c3571befab8358d4f85cd4471964fcd/numpy-2.2.4-cp313-cp313-win_amd64.whl", hash = "sha256:207a2b8441cc8b6a2a78c9ddc64d00d20c303d79fba08c577752f080c4007ee3", size = 12636885, upload-time = "2025-03-16T18:20:36.982Z" }, - { url = "https://files.pythonhosted.org/packages/fa/e2/793288ede17a0fdc921172916efb40f3cbc2aa97e76c5c84aba6dc7e8747/numpy-2.2.4-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:8120575cb4882318c791f839a4fd66161a6fa46f3f0a5e613071aae35b5dd8f8", size = 20961829, upload-time = "2025-03-16T18:16:56.191Z" }, - { url = "https://files.pythonhosted.org/packages/3a/75/bb4573f6c462afd1ea5cbedcc362fe3e9bdbcc57aefd37c681be1155fbaa/numpy-2.2.4-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:a761ba0fa886a7bb33c6c8f6f20213735cb19642c580a931c625ee377ee8bd39", size = 14161419, upload-time = "2025-03-16T18:17:22.811Z" }, - { url = "https://files.pythonhosted.org/packages/03/68/07b4cd01090ca46c7a336958b413cdbe75002286295f2addea767b7f16c9/numpy-2.2.4-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:ac0280f1ba4a4bfff363a99a6aceed4f8e123f8a9b234c89140f5e894e452ecd", size = 5196414, upload-time = "2025-03-16T18:17:34.066Z" }, - { url = "https://files.pythonhosted.org/packages/a5/fd/d4a29478d622fedff5c4b4b4cedfc37a00691079623c0575978d2446db9e/numpy-2.2.4-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:879cf3a9a2b53a4672a168c21375166171bc3932b7e21f622201811c43cdd3b0", size = 6709379, upload-time = "2025-03-16T18:17:47.466Z" }, - { url = "https://files.pythonhosted.org/packages/41/78/96dddb75bb9be730b87c72f30ffdd62611aba234e4e460576a068c98eff6/numpy-2.2.4-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f05d4198c1bacc9124018109c5fba2f3201dbe7ab6e92ff100494f236209c960", size = 14051725, upload-time = "2025-03-16T18:18:11.904Z" }, - { url = "https://files.pythonhosted.org/packages/00/06/5306b8199bffac2a29d9119c11f457f6c7d41115a335b78d3f86fad4dbe8/numpy-2.2.4-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e2f085ce2e813a50dfd0e01fbfc0c12bbe5d2063d99f8b29da30e544fb6483b8", size = 16101638, upload-time = "2025-03-16T18:18:40.749Z" }, - { url = "https://files.pythonhosted.org/packages/fa/03/74c5b631ee1ded596945c12027649e6344614144369fd3ec1aaced782882/numpy-2.2.4-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:92bda934a791c01d6d9d8e038363c50918ef7c40601552a58ac84c9613a665bc", size = 15571717, upload-time = "2025-03-16T18:19:04.512Z" }, - { url = "https://files.pythonhosted.org/packages/cb/dc/4fc7c0283abe0981e3b89f9b332a134e237dd476b0c018e1e21083310c31/numpy-2.2.4-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:ee4d528022f4c5ff67332469e10efe06a267e32f4067dc76bb7e2cddf3cd25ff", size = 17879998, upload-time = "2025-03-16T18:19:32.52Z" }, - { url = "https://files.pythonhosted.org/packages/e5/2b/878576190c5cfa29ed896b518cc516aecc7c98a919e20706c12480465f43/numpy-2.2.4-cp313-cp313t-win32.whl", hash = "sha256:05c076d531e9998e7e694c36e8b349969c56eadd2cdcd07242958489d79a7286", size = 6366896, upload-time = "2025-03-16T18:19:43.55Z" }, - { url = "https://files.pythonhosted.org/packages/3e/05/eb7eec66b95cf697f08c754ef26c3549d03ebd682819f794cb039574a0a6/numpy-2.2.4-cp313-cp313t-win_amd64.whl", hash = "sha256:188dcbca89834cc2e14eb2f106c96d6d46f200fe0200310fc29089657379c58d", size = 12739119, upload-time = "2025-03-16T18:20:03.94Z" }, - { url = "https://files.pythonhosted.org/packages/b2/5c/f09c33a511aff41a098e6ef3498465d95f6360621034a3d95f47edbc9119/numpy-2.2.4-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7051ee569db5fbac144335e0f3b9c2337e0c8d5c9fee015f259a5bd70772b7e8", size = 21081956, upload-time = "2025-03-16T18:21:12.955Z" }, - { url = "https://files.pythonhosted.org/packages/ba/30/74c48b3b6494c4b820b7fa1781d441e94d87a08daa5b35d222f06ba41a6f/numpy-2.2.4-pp310-pypy310_pp73-macosx_14_0_x86_64.whl", hash = "sha256:ab2939cd5bec30a7430cbdb2287b63151b77cf9624de0532d629c9a1c59b1d5c", size = 6827143, upload-time = "2025-03-16T18:21:26.748Z" }, - { url = "https://files.pythonhosted.org/packages/54/f5/ab0d2f48b490535c7a80e05da4a98902b632369efc04f0e47bb31ca97d8f/numpy-2.2.4-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d0f35b19894a9e08639fd60a1ec1978cb7f5f7f1eace62f38dd36be8aecdef4d", size = 16233350, upload-time = "2025-03-16T18:21:53.902Z" }, - { url = "https://files.pythonhosted.org/packages/3b/3a/2f6d8c1f8e45d496bca6baaec93208035faeb40d5735c25afac092ec9a12/numpy-2.2.4-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:b4adfbbc64014976d2f91084915ca4e626fbf2057fb81af209c1a6d776d23e3d", size = 12857565, upload-time = "2025-03-16T18:22:17.631Z" }, +version = "2.2.6" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'linux'", + "python_full_version < '3.11' and sys_platform != 'linux'", +] +sdist = { url = "https://files.pythonhosted.org/packages/76/21/7d2a95e4bba9dc13d043ee156a356c0a8f0c6309dff6b21b4d71a073b8a8/numpy-2.2.6.tar.gz", hash = "sha256:e29554e2bef54a90aa5cc07da6ce955accb83f21ab5de01a62c8478897b264fd", size = 20276440, upload-time = "2025-05-17T22:38:04.611Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/9a/3e/ed6db5be21ce87955c0cbd3009f2803f59fa08df21b5df06862e2d8e2bdd/numpy-2.2.6-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:b412caa66f72040e6d268491a59f2c43bf03eb6c96dd8f0307829feb7fa2b6fb", size = 21165245, upload-time = "2025-05-17T21:27:58.555Z" }, + { url = "https://files.pythonhosted.org/packages/22/c2/4b9221495b2a132cc9d2eb862e21d42a009f5a60e45fc44b00118c174bff/numpy-2.2.6-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:8e41fd67c52b86603a91c1a505ebaef50b3314de0213461c7a6e99c9a3beff90", size = 14360048, upload-time = "2025-05-17T21:28:21.406Z" }, + { url = "https://files.pythonhosted.org/packages/fd/77/dc2fcfc66943c6410e2bf598062f5959372735ffda175b39906d54f02349/numpy-2.2.6-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:37e990a01ae6ec7fe7fa1c26c55ecb672dd98b19c3d0e1d1f326fa13cb38d163", size = 5340542, upload-time = "2025-05-17T21:28:30.931Z" }, + { url = "https://files.pythonhosted.org/packages/7a/4f/1cb5fdc353a5f5cc7feb692db9b8ec2c3d6405453f982435efc52561df58/numpy-2.2.6-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:5a6429d4be8ca66d889b7cf70f536a397dc45ba6faeb5f8c5427935d9592e9cf", size = 6878301, upload-time = "2025-05-17T21:28:41.613Z" }, + { url = "https://files.pythonhosted.org/packages/eb/17/96a3acd228cec142fcb8723bd3cc39c2a474f7dcf0a5d16731980bcafa95/numpy-2.2.6-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:efd28d4e9cd7d7a8d39074a4d44c63eda73401580c5c76acda2ce969e0a38e83", size = 14297320, upload-time = "2025-05-17T21:29:02.78Z" }, + { url = "https://files.pythonhosted.org/packages/b4/63/3de6a34ad7ad6646ac7d2f55ebc6ad439dbbf9c4370017c50cf403fb19b5/numpy-2.2.6-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc7b73d02efb0e18c000e9ad8b83480dfcd5dfd11065997ed4c6747470ae8915", size = 16801050, upload-time = "2025-05-17T21:29:27.675Z" }, + { url = "https://files.pythonhosted.org/packages/07/b6/89d837eddef52b3d0cec5c6ba0456c1bf1b9ef6a6672fc2b7873c3ec4e2e/numpy-2.2.6-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:74d4531beb257d2c3f4b261bfb0fc09e0f9ebb8842d82a7b4209415896adc680", size = 15807034, upload-time = "2025-05-17T21:29:51.102Z" }, + { url = "https://files.pythonhosted.org/packages/01/c8/dc6ae86e3c61cfec1f178e5c9f7858584049b6093f843bca541f94120920/numpy-2.2.6-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:8fc377d995680230e83241d8a96def29f204b5782f371c532579b4f20607a289", size = 18614185, upload-time = "2025-05-17T21:30:18.703Z" }, + { url = "https://files.pythonhosted.org/packages/5b/c5/0064b1b7e7c89137b471ccec1fd2282fceaae0ab3a9550f2568782d80357/numpy-2.2.6-cp310-cp310-win32.whl", hash = "sha256:b093dd74e50a8cba3e873868d9e93a85b78e0daf2e98c6797566ad8044e8363d", size = 6527149, upload-time = "2025-05-17T21:30:29.788Z" }, + { url = "https://files.pythonhosted.org/packages/a3/dd/4b822569d6b96c39d1215dbae0582fd99954dcbcf0c1a13c61783feaca3f/numpy-2.2.6-cp310-cp310-win_amd64.whl", hash = "sha256:f0fd6321b839904e15c46e0d257fdd101dd7f530fe03fd6359c1ea63738703f3", size = 12904620, upload-time = "2025-05-17T21:30:48.994Z" }, + { url = "https://files.pythonhosted.org/packages/da/a8/4f83e2aa666a9fbf56d6118faaaf5f1974d456b1823fda0a176eff722839/numpy-2.2.6-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:f9f1adb22318e121c5c69a09142811a201ef17ab257a1e66ca3025065b7f53ae", size = 21176963, upload-time = "2025-05-17T21:31:19.36Z" }, + { url = "https://files.pythonhosted.org/packages/b3/2b/64e1affc7972decb74c9e29e5649fac940514910960ba25cd9af4488b66c/numpy-2.2.6-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c820a93b0255bc360f53eca31a0e676fd1101f673dda8da93454a12e23fc5f7a", size = 14406743, upload-time = "2025-05-17T21:31:41.087Z" }, + { url = "https://files.pythonhosted.org/packages/4a/9f/0121e375000b5e50ffdd8b25bf78d8e1a5aa4cca3f185d41265198c7b834/numpy-2.2.6-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:3d70692235e759f260c3d837193090014aebdf026dfd167834bcba43e30c2a42", size = 5352616, upload-time = "2025-05-17T21:31:50.072Z" }, + { url = "https://files.pythonhosted.org/packages/31/0d/b48c405c91693635fbe2dcd7bc84a33a602add5f63286e024d3b6741411c/numpy-2.2.6-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:481b49095335f8eed42e39e8041327c05b0f6f4780488f61286ed3c01368d491", size = 6889579, upload-time = "2025-05-17T21:32:01.712Z" }, + { url = "https://files.pythonhosted.org/packages/52/b8/7f0554d49b565d0171eab6e99001846882000883998e7b7d9f0d98b1f934/numpy-2.2.6-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b64d8d4d17135e00c8e346e0a738deb17e754230d7e0810ac5012750bbd85a5a", size = 14312005, upload-time = "2025-05-17T21:32:23.332Z" }, + { url = "https://files.pythonhosted.org/packages/b3/dd/2238b898e51bd6d389b7389ffb20d7f4c10066d80351187ec8e303a5a475/numpy-2.2.6-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ba10f8411898fc418a521833e014a77d3ca01c15b0c6cdcce6a0d2897e6dbbdf", size = 16821570, upload-time = "2025-05-17T21:32:47.991Z" }, + { url = "https://files.pythonhosted.org/packages/83/6c/44d0325722cf644f191042bf47eedad61c1e6df2432ed65cbe28509d404e/numpy-2.2.6-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:bd48227a919f1bafbdda0583705e547892342c26fb127219d60a5c36882609d1", size = 15818548, upload-time = "2025-05-17T21:33:11.728Z" }, + { url = "https://files.pythonhosted.org/packages/ae/9d/81e8216030ce66be25279098789b665d49ff19eef08bfa8cb96d4957f422/numpy-2.2.6-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:9551a499bf125c1d4f9e250377c1ee2eddd02e01eac6644c080162c0c51778ab", size = 18620521, upload-time = "2025-05-17T21:33:39.139Z" }, + { url = "https://files.pythonhosted.org/packages/6a/fd/e19617b9530b031db51b0926eed5345ce8ddc669bb3bc0044b23e275ebe8/numpy-2.2.6-cp311-cp311-win32.whl", hash = "sha256:0678000bb9ac1475cd454c6b8c799206af8107e310843532b04d49649c717a47", size = 6525866, upload-time = "2025-05-17T21:33:50.273Z" }, + { url = "https://files.pythonhosted.org/packages/31/0a/f354fb7176b81747d870f7991dc763e157a934c717b67b58456bc63da3df/numpy-2.2.6-cp311-cp311-win_amd64.whl", hash = "sha256:e8213002e427c69c45a52bbd94163084025f533a55a59d6f9c5b820774ef3303", size = 12907455, upload-time = "2025-05-17T21:34:09.135Z" }, + { url = "https://files.pythonhosted.org/packages/82/5d/c00588b6cf18e1da539b45d3598d3557084990dcc4331960c15ee776ee41/numpy-2.2.6-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:41c5a21f4a04fa86436124d388f6ed60a9343a6f767fced1a8a71c3fbca038ff", size = 20875348, upload-time = "2025-05-17T21:34:39.648Z" }, + { url = "https://files.pythonhosted.org/packages/66/ee/560deadcdde6c2f90200450d5938f63a34b37e27ebff162810f716f6a230/numpy-2.2.6-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:de749064336d37e340f640b05f24e9e3dd678c57318c7289d222a8a2f543e90c", size = 14119362, upload-time = "2025-05-17T21:35:01.241Z" }, + { url = "https://files.pythonhosted.org/packages/3c/65/4baa99f1c53b30adf0acd9a5519078871ddde8d2339dc5a7fde80d9d87da/numpy-2.2.6-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:894b3a42502226a1cac872f840030665f33326fc3dac8e57c607905773cdcde3", size = 5084103, upload-time = "2025-05-17T21:35:10.622Z" }, + { url = "https://files.pythonhosted.org/packages/cc/89/e5a34c071a0570cc40c9a54eb472d113eea6d002e9ae12bb3a8407fb912e/numpy-2.2.6-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:71594f7c51a18e728451bb50cc60a3ce4e6538822731b2933209a1f3614e9282", size = 6625382, upload-time = "2025-05-17T21:35:21.414Z" }, + { url = "https://files.pythonhosted.org/packages/f8/35/8c80729f1ff76b3921d5c9487c7ac3de9b2a103b1cd05e905b3090513510/numpy-2.2.6-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f2618db89be1b4e05f7a1a847a9c1c0abd63e63a1607d892dd54668dd92faf87", size = 14018462, upload-time = "2025-05-17T21:35:42.174Z" }, + { url = "https://files.pythonhosted.org/packages/8c/3d/1e1db36cfd41f895d266b103df00ca5b3cbe965184df824dec5c08c6b803/numpy-2.2.6-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd83c01228a688733f1ded5201c678f0c53ecc1006ffbc404db9f7a899ac6249", size = 16527618, upload-time = "2025-05-17T21:36:06.711Z" }, + { url = "https://files.pythonhosted.org/packages/61/c6/03ed30992602c85aa3cd95b9070a514f8b3c33e31124694438d88809ae36/numpy-2.2.6-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:37c0ca431f82cd5fa716eca9506aefcabc247fb27ba69c5062a6d3ade8cf8f49", size = 15505511, upload-time = "2025-05-17T21:36:29.965Z" }, + { url = "https://files.pythonhosted.org/packages/b7/25/5761d832a81df431e260719ec45de696414266613c9ee268394dd5ad8236/numpy-2.2.6-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:fe27749d33bb772c80dcd84ae7e8df2adc920ae8297400dabec45f0dedb3f6de", size = 18313783, upload-time = "2025-05-17T21:36:56.883Z" }, + { url = "https://files.pythonhosted.org/packages/57/0a/72d5a3527c5ebffcd47bde9162c39fae1f90138c961e5296491ce778e682/numpy-2.2.6-cp312-cp312-win32.whl", hash = "sha256:4eeaae00d789f66c7a25ac5f34b71a7035bb474e679f410e5e1a94deb24cf2d4", size = 6246506, upload-time = "2025-05-17T21:37:07.368Z" }, + { url = "https://files.pythonhosted.org/packages/36/fa/8c9210162ca1b88529ab76b41ba02d433fd54fecaf6feb70ef9f124683f1/numpy-2.2.6-cp312-cp312-win_amd64.whl", hash = "sha256:c1f9540be57940698ed329904db803cf7a402f3fc200bfe599334c9bd84a40b2", size = 12614190, upload-time = "2025-05-17T21:37:26.213Z" }, + { url = "https://files.pythonhosted.org/packages/f9/5c/6657823f4f594f72b5471f1db1ab12e26e890bb2e41897522d134d2a3e81/numpy-2.2.6-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:0811bb762109d9708cca4d0b13c4f67146e3c3b7cf8d34018c722adb2d957c84", size = 20867828, upload-time = "2025-05-17T21:37:56.699Z" }, + { url = "https://files.pythonhosted.org/packages/dc/9e/14520dc3dadf3c803473bd07e9b2bd1b69bc583cb2497b47000fed2fa92f/numpy-2.2.6-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:287cc3162b6f01463ccd86be154f284d0893d2b3ed7292439ea97eafa8170e0b", size = 14143006, upload-time = "2025-05-17T21:38:18.291Z" }, + { url = "https://files.pythonhosted.org/packages/4f/06/7e96c57d90bebdce9918412087fc22ca9851cceaf5567a45c1f404480e9e/numpy-2.2.6-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:f1372f041402e37e5e633e586f62aa53de2eac8d98cbfb822806ce4bbefcb74d", size = 5076765, upload-time = "2025-05-17T21:38:27.319Z" }, + { url = "https://files.pythonhosted.org/packages/73/ed/63d920c23b4289fdac96ddbdd6132e9427790977d5457cd132f18e76eae0/numpy-2.2.6-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:55a4d33fa519660d69614a9fad433be87e5252f4b03850642f88993f7b2ca566", size = 6617736, upload-time = "2025-05-17T21:38:38.141Z" }, + { url = "https://files.pythonhosted.org/packages/85/c5/e19c8f99d83fd377ec8c7e0cf627a8049746da54afc24ef0a0cb73d5dfb5/numpy-2.2.6-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f92729c95468a2f4f15e9bb94c432a9229d0d50de67304399627a943201baa2f", size = 14010719, upload-time = "2025-05-17T21:38:58.433Z" }, + { url = "https://files.pythonhosted.org/packages/19/49/4df9123aafa7b539317bf6d342cb6d227e49f7a35b99c287a6109b13dd93/numpy-2.2.6-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1bc23a79bfabc5d056d106f9befb8d50c31ced2fbc70eedb8155aec74a45798f", size = 16526072, upload-time = "2025-05-17T21:39:22.638Z" }, + { url = "https://files.pythonhosted.org/packages/b2/6c/04b5f47f4f32f7c2b0e7260442a8cbcf8168b0e1a41ff1495da42f42a14f/numpy-2.2.6-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:e3143e4451880bed956e706a3220b4e5cf6172ef05fcc397f6f36a550b1dd868", size = 15503213, upload-time = "2025-05-17T21:39:45.865Z" }, + { url = "https://files.pythonhosted.org/packages/17/0a/5cd92e352c1307640d5b6fec1b2ffb06cd0dabe7d7b8227f97933d378422/numpy-2.2.6-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:b4f13750ce79751586ae2eb824ba7e1e8dba64784086c98cdbbcc6a42112ce0d", size = 18316632, upload-time = "2025-05-17T21:40:13.331Z" }, + { url = "https://files.pythonhosted.org/packages/f0/3b/5cba2b1d88760ef86596ad0f3d484b1cbff7c115ae2429678465057c5155/numpy-2.2.6-cp313-cp313-win32.whl", hash = "sha256:5beb72339d9d4fa36522fc63802f469b13cdbe4fdab4a288f0c441b74272ebfd", size = 6244532, upload-time = "2025-05-17T21:43:46.099Z" }, + { url = "https://files.pythonhosted.org/packages/cb/3b/d58c12eafcb298d4e6d0d40216866ab15f59e55d148a5658bb3132311fcf/numpy-2.2.6-cp313-cp313-win_amd64.whl", hash = "sha256:b0544343a702fa80c95ad5d3d608ea3599dd54d4632df855e4c8d24eb6ecfa1c", size = 12610885, upload-time = "2025-05-17T21:44:05.145Z" }, + { url = "https://files.pythonhosted.org/packages/6b/9e/4bf918b818e516322db999ac25d00c75788ddfd2d2ade4fa66f1f38097e1/numpy-2.2.6-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:0bca768cd85ae743b2affdc762d617eddf3bcf8724435498a1e80132d04879e6", size = 20963467, upload-time = "2025-05-17T21:40:44Z" }, + { url = "https://files.pythonhosted.org/packages/61/66/d2de6b291507517ff2e438e13ff7b1e2cdbdb7cb40b3ed475377aece69f9/numpy-2.2.6-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:fc0c5673685c508a142ca65209b4e79ed6740a4ed6b2267dbba90f34b0b3cfda", size = 14225144, upload-time = "2025-05-17T21:41:05.695Z" }, + { url = "https://files.pythonhosted.org/packages/e4/25/480387655407ead912e28ba3a820bc69af9adf13bcbe40b299d454ec011f/numpy-2.2.6-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:5bd4fc3ac8926b3819797a7c0e2631eb889b4118a9898c84f585a54d475b7e40", size = 5200217, upload-time = "2025-05-17T21:41:15.903Z" }, + { url = "https://files.pythonhosted.org/packages/aa/4a/6e313b5108f53dcbf3aca0c0f3e9c92f4c10ce57a0a721851f9785872895/numpy-2.2.6-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:fee4236c876c4e8369388054d02d0e9bb84821feb1a64dd59e137e6511a551f8", size = 6712014, upload-time = "2025-05-17T21:41:27.321Z" }, + { url = "https://files.pythonhosted.org/packages/b7/30/172c2d5c4be71fdf476e9de553443cf8e25feddbe185e0bd88b096915bcc/numpy-2.2.6-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e1dda9c7e08dc141e0247a5b8f49cf05984955246a327d4c48bda16821947b2f", size = 14077935, upload-time = "2025-05-17T21:41:49.738Z" }, + { url = "https://files.pythonhosted.org/packages/12/fb/9e743f8d4e4d3c710902cf87af3512082ae3d43b945d5d16563f26ec251d/numpy-2.2.6-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f447e6acb680fd307f40d3da4852208af94afdfab89cf850986c3ca00562f4fa", size = 16600122, upload-time = "2025-05-17T21:42:14.046Z" }, + { url = "https://files.pythonhosted.org/packages/12/75/ee20da0e58d3a66f204f38916757e01e33a9737d0b22373b3eb5a27358f9/numpy-2.2.6-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:389d771b1623ec92636b0786bc4ae56abafad4a4c513d36a55dce14bd9ce8571", size = 15586143, upload-time = "2025-05-17T21:42:37.464Z" }, + { url = "https://files.pythonhosted.org/packages/76/95/bef5b37f29fc5e739947e9ce5179ad402875633308504a52d188302319c8/numpy-2.2.6-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:8e9ace4a37db23421249ed236fdcdd457d671e25146786dfc96835cd951aa7c1", size = 18385260, upload-time = "2025-05-17T21:43:05.189Z" }, + { url = "https://files.pythonhosted.org/packages/09/04/f2f83279d287407cf36a7a8053a5abe7be3622a4363337338f2585e4afda/numpy-2.2.6-cp313-cp313t-win32.whl", hash = "sha256:038613e9fb8c72b0a41f025a7e4c3f0b7a1b5d768ece4796b674c8f3fe13efff", size = 6377225, upload-time = "2025-05-17T21:43:16.254Z" }, + { url = "https://files.pythonhosted.org/packages/67/0e/35082d13c09c02c011cf21570543d202ad929d961c02a147493cb0c2bdf5/numpy-2.2.6-cp313-cp313t-win_amd64.whl", hash = "sha256:6031dd6dfecc0cf9f668681a37648373bddd6421fff6c66ec1624eed0180ee06", size = 12771374, upload-time = "2025-05-17T21:43:35.479Z" }, + { url = "https://files.pythonhosted.org/packages/9e/3b/d94a75f4dbf1ef5d321523ecac21ef23a3cd2ac8b78ae2aac40873590229/numpy-2.2.6-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:0b605b275d7bd0c640cad4e5d30fa701a8d59302e127e5f79138ad62762c3e3d", size = 21040391, upload-time = "2025-05-17T21:44:35.948Z" }, + { url = "https://files.pythonhosted.org/packages/17/f4/09b2fa1b58f0fb4f7c7963a1649c64c4d315752240377ed74d9cd878f7b5/numpy-2.2.6-pp310-pypy310_pp73-macosx_14_0_x86_64.whl", hash = "sha256:7befc596a7dc9da8a337f79802ee8adb30a552a94f792b9c9d18c840055907db", size = 6786754, upload-time = "2025-05-17T21:44:47.446Z" }, + { url = "https://files.pythonhosted.org/packages/af/30/feba75f143bdc868a1cc3f44ccfa6c4b9ec522b36458e738cd00f67b573f/numpy-2.2.6-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ce47521a4754c8f4593837384bd3424880629f718d87c5d44f8ed763edd63543", size = 16643476, upload-time = "2025-05-17T21:45:11.871Z" }, + { url = "https://files.pythonhosted.org/packages/37/48/ac2a9584402fb6c0cd5b5d1a91dcf176b15760130dd386bbafdbfe3640bf/numpy-2.2.6-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:d042d24c90c41b54fd506da306759e06e568864df8ec17ccc17e9e884634fd00", size = 12812666, upload-time = "2025-05-17T21:45:31.426Z" }, +] + +[[package]] +name = "numpy" +version = "2.3.2" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", + "python_full_version == '3.11.*' and sys_platform == 'linux'", + "python_full_version == '3.11.*' and sys_platform != 'linux'", +] +sdist = { url = "https://files.pythonhosted.org/packages/37/7d/3fec4199c5ffb892bed55cff901e4f39a58c81df9c44c280499e92cad264/numpy-2.3.2.tar.gz", hash = "sha256:e0486a11ec30cdecb53f184d496d1c6a20786c81e55e41640270130056f8ee48", size = 20489306, upload-time = "2025-07-24T21:32:07.553Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/96/26/1320083986108998bd487e2931eed2aeedf914b6e8905431487543ec911d/numpy-2.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:852ae5bed3478b92f093e30f785c98e0cb62fa0a939ed057c31716e18a7a22b9", size = 21259016, upload-time = "2025-07-24T20:24:35.214Z" }, + { url = "https://files.pythonhosted.org/packages/c4/2b/792b341463fa93fc7e55abbdbe87dac316c5b8cb5e94fb7a59fb6fa0cda5/numpy-2.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7a0e27186e781a69959d0230dd9909b5e26024f8da10683bd6344baea1885168", size = 14451158, upload-time = "2025-07-24T20:24:58.397Z" }, + { url = "https://files.pythonhosted.org/packages/b7/13/e792d7209261afb0c9f4759ffef6135b35c77c6349a151f488f531d13595/numpy-2.3.2-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:f0a1a8476ad77a228e41619af2fa9505cf69df928e9aaa165746584ea17fed2b", size = 5379817, upload-time = "2025-07-24T20:25:07.746Z" }, + { url = "https://files.pythonhosted.org/packages/49/ce/055274fcba4107c022b2113a213c7287346563f48d62e8d2a5176ad93217/numpy-2.3.2-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:cbc95b3813920145032412f7e33d12080f11dc776262df1712e1638207dde9e8", size = 6913606, upload-time = "2025-07-24T20:25:18.84Z" }, + { url = "https://files.pythonhosted.org/packages/17/f2/e4d72e6bc5ff01e2ab613dc198d560714971900c03674b41947e38606502/numpy-2.3.2-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f75018be4980a7324edc5930fe39aa391d5734531b1926968605416ff58c332d", size = 14589652, upload-time = "2025-07-24T20:25:40.356Z" }, + { url = "https://files.pythonhosted.org/packages/c8/b0/fbeee3000a51ebf7222016e2939b5c5ecf8000a19555d04a18f1e02521b8/numpy-2.3.2-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:20b8200721840f5621b7bd03f8dcd78de33ec522fc40dc2641aa09537df010c3", size = 16938816, upload-time = "2025-07-24T20:26:05.721Z" }, + { url = "https://files.pythonhosted.org/packages/a9/ec/2f6c45c3484cc159621ea8fc000ac5a86f1575f090cac78ac27193ce82cd/numpy-2.3.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:1f91e5c028504660d606340a084db4b216567ded1056ea2b4be4f9d10b67197f", size = 16370512, upload-time = "2025-07-24T20:26:30.545Z" }, + { url = "https://files.pythonhosted.org/packages/b5/01/dd67cf511850bd7aefd6347aaae0956ed415abea741ae107834aae7d6d4e/numpy-2.3.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:fb1752a3bb9a3ad2d6b090b88a9a0ae1cd6f004ef95f75825e2f382c183b2097", size = 18884947, upload-time = "2025-07-24T20:26:58.24Z" }, + { url = "https://files.pythonhosted.org/packages/a7/17/2cf60fd3e6a61d006778735edf67a222787a8c1a7842aed43ef96d777446/numpy-2.3.2-cp311-cp311-win32.whl", hash = "sha256:4ae6863868aaee2f57503c7a5052b3a2807cf7a3914475e637a0ecd366ced220", size = 6599494, upload-time = "2025-07-24T20:27:09.786Z" }, + { url = "https://files.pythonhosted.org/packages/d5/03/0eade211c504bda872a594f045f98ddcc6caef2b7c63610946845e304d3f/numpy-2.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:240259d6564f1c65424bcd10f435145a7644a65a6811cfc3201c4a429ba79170", size = 13087889, upload-time = "2025-07-24T20:27:29.558Z" }, + { url = "https://files.pythonhosted.org/packages/13/32/2c7979d39dafb2a25087e12310fc7f3b9d3c7d960df4f4bc97955ae0ce1d/numpy-2.3.2-cp311-cp311-win_arm64.whl", hash = "sha256:4209f874d45f921bde2cff1ffcd8a3695f545ad2ffbef6d3d3c6768162efab89", size = 10459560, upload-time = "2025-07-24T20:27:46.803Z" }, + { url = "https://files.pythonhosted.org/packages/00/6d/745dd1c1c5c284d17725e5c802ca4d45cfc6803519d777f087b71c9f4069/numpy-2.3.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:bc3186bea41fae9d8e90c2b4fb5f0a1f5a690682da79b92574d63f56b529080b", size = 20956420, upload-time = "2025-07-24T20:28:18.002Z" }, + { url = "https://files.pythonhosted.org/packages/bc/96/e7b533ea5740641dd62b07a790af5d9d8fec36000b8e2d0472bd7574105f/numpy-2.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:2f4f0215edb189048a3c03bd5b19345bdfa7b45a7a6f72ae5945d2a28272727f", size = 14184660, upload-time = "2025-07-24T20:28:39.522Z" }, + { url = "https://files.pythonhosted.org/packages/2b/53/102c6122db45a62aa20d1b18c9986f67e6b97e0d6fbc1ae13e3e4c84430c/numpy-2.3.2-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:8b1224a734cd509f70816455c3cffe13a4f599b1bf7130f913ba0e2c0b2006c0", size = 5113382, upload-time = "2025-07-24T20:28:48.544Z" }, + { url = "https://files.pythonhosted.org/packages/2b/21/376257efcbf63e624250717e82b4fae93d60178f09eb03ed766dbb48ec9c/numpy-2.3.2-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:3dcf02866b977a38ba3ec10215220609ab9667378a9e2150615673f3ffd6c73b", size = 6647258, upload-time = "2025-07-24T20:28:59.104Z" }, + { url = "https://files.pythonhosted.org/packages/91/ba/f4ebf257f08affa464fe6036e13f2bf9d4642a40228781dc1235da81be9f/numpy-2.3.2-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:572d5512df5470f50ada8d1972c5f1082d9a0b7aa5944db8084077570cf98370", size = 14281409, upload-time = "2025-07-24T20:40:30.298Z" }, + { url = "https://files.pythonhosted.org/packages/59/ef/f96536f1df42c668cbacb727a8c6da7afc9c05ece6d558927fb1722693e1/numpy-2.3.2-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8145dd6d10df13c559d1e4314df29695613575183fa2e2d11fac4c208c8a1f73", size = 16641317, upload-time = "2025-07-24T20:40:56.625Z" }, + { url = "https://files.pythonhosted.org/packages/f6/a7/af813a7b4f9a42f498dde8a4c6fcbff8100eed00182cc91dbaf095645f38/numpy-2.3.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:103ea7063fa624af04a791c39f97070bf93b96d7af7eb23530cd087dc8dbe9dc", size = 16056262, upload-time = "2025-07-24T20:41:20.797Z" }, + { url = "https://files.pythonhosted.org/packages/8b/5d/41c4ef8404caaa7f05ed1cfb06afe16a25895260eacbd29b4d84dff2920b/numpy-2.3.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:fc927d7f289d14f5e037be917539620603294454130b6de200091e23d27dc9be", size = 18579342, upload-time = "2025-07-24T20:41:50.753Z" }, + { url = "https://files.pythonhosted.org/packages/a1/4f/9950e44c5a11636f4a3af6e825ec23003475cc9a466edb7a759ed3ea63bd/numpy-2.3.2-cp312-cp312-win32.whl", hash = "sha256:d95f59afe7f808c103be692175008bab926b59309ade3e6d25009e9a171f7036", size = 6320610, upload-time = "2025-07-24T20:42:01.551Z" }, + { url = "https://files.pythonhosted.org/packages/7c/2f/244643a5ce54a94f0a9a2ab578189c061e4a87c002e037b0829dd77293b6/numpy-2.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:9e196ade2400c0c737d93465327d1ae7c06c7cb8a1756121ebf54b06ca183c7f", size = 12786292, upload-time = "2025-07-24T20:42:20.738Z" }, + { url = "https://files.pythonhosted.org/packages/54/cd/7b5f49d5d78db7badab22d8323c1b6ae458fbf86c4fdfa194ab3cd4eb39b/numpy-2.3.2-cp312-cp312-win_arm64.whl", hash = "sha256:ee807923782faaf60d0d7331f5e86da7d5e3079e28b291973c545476c2b00d07", size = 10194071, upload-time = "2025-07-24T20:42:36.657Z" }, + { url = "https://files.pythonhosted.org/packages/1c/c0/c6bb172c916b00700ed3bf71cb56175fd1f7dbecebf8353545d0b5519f6c/numpy-2.3.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:c8d9727f5316a256425892b043736d63e89ed15bbfe6556c5ff4d9d4448ff3b3", size = 20949074, upload-time = "2025-07-24T20:43:07.813Z" }, + { url = "https://files.pythonhosted.org/packages/20/4e/c116466d22acaf4573e58421c956c6076dc526e24a6be0903219775d862e/numpy-2.3.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:efc81393f25f14d11c9d161e46e6ee348637c0a1e8a54bf9dedc472a3fae993b", size = 14177311, upload-time = "2025-07-24T20:43:29.335Z" }, + { url = "https://files.pythonhosted.org/packages/78/45/d4698c182895af189c463fc91d70805d455a227261d950e4e0f1310c2550/numpy-2.3.2-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:dd937f088a2df683cbb79dda9a772b62a3e5a8a7e76690612c2737f38c6ef1b6", size = 5106022, upload-time = "2025-07-24T20:43:37.999Z" }, + { url = "https://files.pythonhosted.org/packages/9f/76/3e6880fef4420179309dba72a8c11f6166c431cf6dee54c577af8906f914/numpy-2.3.2-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:11e58218c0c46c80509186e460d79fbdc9ca1eb8d8aee39d8f2dc768eb781089", size = 6640135, upload-time = "2025-07-24T20:43:49.28Z" }, + { url = "https://files.pythonhosted.org/packages/34/fa/87ff7f25b3c4ce9085a62554460b7db686fef1e0207e8977795c7b7d7ba1/numpy-2.3.2-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5ad4ebcb683a1f99f4f392cc522ee20a18b2bb12a2c1c42c3d48d5a1adc9d3d2", size = 14278147, upload-time = "2025-07-24T20:44:10.328Z" }, + { url = "https://files.pythonhosted.org/packages/1d/0f/571b2c7a3833ae419fe69ff7b479a78d313581785203cc70a8db90121b9a/numpy-2.3.2-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:938065908d1d869c7d75d8ec45f735a034771c6ea07088867f713d1cd3bbbe4f", size = 16635989, upload-time = "2025-07-24T20:44:34.88Z" }, + { url = "https://files.pythonhosted.org/packages/24/5a/84ae8dca9c9a4c592fe11340b36a86ffa9fd3e40513198daf8a97839345c/numpy-2.3.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:66459dccc65d8ec98cc7df61307b64bf9e08101f9598755d42d8ae65d9a7a6ee", size = 16053052, upload-time = "2025-07-24T20:44:58.872Z" }, + { url = "https://files.pythonhosted.org/packages/57/7c/e5725d99a9133b9813fcf148d3f858df98511686e853169dbaf63aec6097/numpy-2.3.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:a7af9ed2aa9ec5950daf05bb11abc4076a108bd3c7db9aa7251d5f107079b6a6", size = 18577955, upload-time = "2025-07-24T20:45:26.714Z" }, + { url = "https://files.pythonhosted.org/packages/ae/11/7c546fcf42145f29b71e4d6f429e96d8d68e5a7ba1830b2e68d7418f0bbd/numpy-2.3.2-cp313-cp313-win32.whl", hash = "sha256:906a30249315f9c8e17b085cc5f87d3f369b35fedd0051d4a84686967bdbbd0b", size = 6311843, upload-time = "2025-07-24T20:49:24.444Z" }, + { url = "https://files.pythonhosted.org/packages/aa/6f/a428fd1cb7ed39b4280d057720fed5121b0d7754fd2a9768640160f5517b/numpy-2.3.2-cp313-cp313-win_amd64.whl", hash = "sha256:c63d95dc9d67b676e9108fe0d2182987ccb0f11933c1e8959f42fa0da8d4fa56", size = 12782876, upload-time = "2025-07-24T20:49:43.227Z" }, + { url = "https://files.pythonhosted.org/packages/65/85/4ea455c9040a12595fb6c43f2c217257c7b52dd0ba332c6a6c1d28b289fe/numpy-2.3.2-cp313-cp313-win_arm64.whl", hash = "sha256:b05a89f2fb84d21235f93de47129dd4f11c16f64c87c33f5e284e6a3a54e43f2", size = 10192786, upload-time = "2025-07-24T20:49:59.443Z" }, + { url = "https://files.pythonhosted.org/packages/80/23/8278f40282d10c3f258ec3ff1b103d4994bcad78b0cba9208317f6bb73da/numpy-2.3.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:4e6ecfeddfa83b02318f4d84acf15fbdbf9ded18e46989a15a8b6995dfbf85ab", size = 21047395, upload-time = "2025-07-24T20:45:58.821Z" }, + { url = "https://files.pythonhosted.org/packages/1f/2d/624f2ce4a5df52628b4ccd16a4f9437b37c35f4f8a50d00e962aae6efd7a/numpy-2.3.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:508b0eada3eded10a3b55725b40806a4b855961040180028f52580c4729916a2", size = 14300374, upload-time = "2025-07-24T20:46:20.207Z" }, + { url = "https://files.pythonhosted.org/packages/f6/62/ff1e512cdbb829b80a6bd08318a58698867bca0ca2499d101b4af063ee97/numpy-2.3.2-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:754d6755d9a7588bdc6ac47dc4ee97867271b17cee39cb87aef079574366db0a", size = 5228864, upload-time = "2025-07-24T20:46:30.58Z" }, + { url = "https://files.pythonhosted.org/packages/7d/8e/74bc18078fff03192d4032cfa99d5a5ca937807136d6f5790ce07ca53515/numpy-2.3.2-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:a9f66e7d2b2d7712410d3bc5684149040ef5f19856f20277cd17ea83e5006286", size = 6737533, upload-time = "2025-07-24T20:46:46.111Z" }, + { url = "https://files.pythonhosted.org/packages/19/ea/0731efe2c9073ccca5698ef6a8c3667c4cf4eea53fcdcd0b50140aba03bc/numpy-2.3.2-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:de6ea4e5a65d5a90c7d286ddff2b87f3f4ad61faa3db8dabe936b34c2275b6f8", size = 14352007, upload-time = "2025-07-24T20:47:07.1Z" }, + { url = "https://files.pythonhosted.org/packages/cf/90/36be0865f16dfed20f4bc7f75235b963d5939707d4b591f086777412ff7b/numpy-2.3.2-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a3ef07ec8cbc8fc9e369c8dcd52019510c12da4de81367d8b20bc692aa07573a", size = 16701914, upload-time = "2025-07-24T20:47:32.459Z" }, + { url = "https://files.pythonhosted.org/packages/94/30/06cd055e24cb6c38e5989a9e747042b4e723535758e6153f11afea88c01b/numpy-2.3.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:27c9f90e7481275c7800dc9c24b7cc40ace3fdb970ae4d21eaff983a32f70c91", size = 16132708, upload-time = "2025-07-24T20:47:58.129Z" }, + { url = "https://files.pythonhosted.org/packages/9a/14/ecede608ea73e58267fd7cb78f42341b3b37ba576e778a1a06baffbe585c/numpy-2.3.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:07b62978075b67eee4065b166d000d457c82a1efe726cce608b9db9dd66a73a5", size = 18651678, upload-time = "2025-07-24T20:48:25.402Z" }, + { url = "https://files.pythonhosted.org/packages/40/f3/2fe6066b8d07c3685509bc24d56386534c008b462a488b7f503ba82b8923/numpy-2.3.2-cp313-cp313t-win32.whl", hash = "sha256:c771cfac34a4f2c0de8e8c97312d07d64fd8f8ed45bc9f5726a7e947270152b5", size = 6441832, upload-time = "2025-07-24T20:48:37.181Z" }, + { url = "https://files.pythonhosted.org/packages/0b/ba/0937d66d05204d8f28630c9c60bc3eda68824abde4cf756c4d6aad03b0c6/numpy-2.3.2-cp313-cp313t-win_amd64.whl", hash = "sha256:72dbebb2dcc8305c431b2836bcc66af967df91be793d63a24e3d9b741374c450", size = 12927049, upload-time = "2025-07-24T20:48:56.24Z" }, + { url = "https://files.pythonhosted.org/packages/e9/ed/13542dd59c104d5e654dfa2ac282c199ba64846a74c2c4bcdbc3a0f75df1/numpy-2.3.2-cp313-cp313t-win_arm64.whl", hash = "sha256:72c6df2267e926a6d5286b0a6d556ebe49eae261062059317837fda12ddf0c1a", size = 10262935, upload-time = "2025-07-24T20:49:13.136Z" }, + { url = "https://files.pythonhosted.org/packages/c9/7c/7659048aaf498f7611b783e000c7268fcc4dcf0ce21cd10aad7b2e8f9591/numpy-2.3.2-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:448a66d052d0cf14ce9865d159bfc403282c9bc7bb2a31b03cc18b651eca8b1a", size = 20950906, upload-time = "2025-07-24T20:50:30.346Z" }, + { url = "https://files.pythonhosted.org/packages/80/db/984bea9d4ddf7112a04cfdfb22b1050af5757864cfffe8e09e44b7f11a10/numpy-2.3.2-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:546aaf78e81b4081b2eba1d105c3b34064783027a06b3ab20b6eba21fb64132b", size = 14185607, upload-time = "2025-07-24T20:50:51.923Z" }, + { url = "https://files.pythonhosted.org/packages/e4/76/b3d6f414f4eca568f469ac112a3b510938d892bc5a6c190cb883af080b77/numpy-2.3.2-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:87c930d52f45df092f7578889711a0768094debf73cfcde105e2d66954358125", size = 5114110, upload-time = "2025-07-24T20:51:01.041Z" }, + { url = "https://files.pythonhosted.org/packages/9e/d2/6f5e6826abd6bca52392ed88fe44a4b52aacb60567ac3bc86c67834c3a56/numpy-2.3.2-cp314-cp314-macosx_14_0_x86_64.whl", hash = "sha256:8dc082ea901a62edb8f59713c6a7e28a85daddcb67454c839de57656478f5b19", size = 6642050, upload-time = "2025-07-24T20:51:11.64Z" }, + { url = "https://files.pythonhosted.org/packages/c4/43/f12b2ade99199e39c73ad182f103f9d9791f48d885c600c8e05927865baf/numpy-2.3.2-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:af58de8745f7fa9ca1c0c7c943616c6fe28e75d0c81f5c295810e3c83b5be92f", size = 14296292, upload-time = "2025-07-24T20:51:33.488Z" }, + { url = "https://files.pythonhosted.org/packages/5d/f9/77c07d94bf110a916b17210fac38680ed8734c236bfed9982fd8524a7b47/numpy-2.3.2-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fed5527c4cf10f16c6d0b6bee1f89958bccb0ad2522c8cadc2efd318bcd545f5", size = 16638913, upload-time = "2025-07-24T20:51:58.517Z" }, + { url = "https://files.pythonhosted.org/packages/9b/d1/9d9f2c8ea399cc05cfff8a7437453bd4e7d894373a93cdc46361bbb49a7d/numpy-2.3.2-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:095737ed986e00393ec18ec0b21b47c22889ae4b0cd2d5e88342e08b01141f58", size = 16071180, upload-time = "2025-07-24T20:52:22.827Z" }, + { url = "https://files.pythonhosted.org/packages/4c/41/82e2c68aff2a0c9bf315e47d61951099fed65d8cb2c8d9dc388cb87e947e/numpy-2.3.2-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:b5e40e80299607f597e1a8a247ff8d71d79c5b52baa11cc1cce30aa92d2da6e0", size = 18576809, upload-time = "2025-07-24T20:52:51.015Z" }, + { url = "https://files.pythonhosted.org/packages/14/14/4b4fd3efb0837ed252d0f583c5c35a75121038a8c4e065f2c259be06d2d8/numpy-2.3.2-cp314-cp314-win32.whl", hash = "sha256:7d6e390423cc1f76e1b8108c9b6889d20a7a1f59d9a60cac4a050fa734d6c1e2", size = 6366410, upload-time = "2025-07-24T20:56:44.949Z" }, + { url = "https://files.pythonhosted.org/packages/11/9e/b4c24a6b8467b61aced5c8dc7dcfce23621baa2e17f661edb2444a418040/numpy-2.3.2-cp314-cp314-win_amd64.whl", hash = "sha256:b9d0878b21e3918d76d2209c924ebb272340da1fb51abc00f986c258cd5e957b", size = 12918821, upload-time = "2025-07-24T20:57:06.479Z" }, + { url = "https://files.pythonhosted.org/packages/0e/0f/0dc44007c70b1007c1cef86b06986a3812dd7106d8f946c09cfa75782556/numpy-2.3.2-cp314-cp314-win_arm64.whl", hash = "sha256:2738534837c6a1d0c39340a190177d7d66fdf432894f469728da901f8f6dc910", size = 10477303, upload-time = "2025-07-24T20:57:22.879Z" }, + { url = "https://files.pythonhosted.org/packages/8b/3e/075752b79140b78ddfc9c0a1634d234cfdbc6f9bbbfa6b7504e445ad7d19/numpy-2.3.2-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:4d002ecf7c9b53240be3bb69d80f86ddbd34078bae04d87be81c1f58466f264e", size = 21047524, upload-time = "2025-07-24T20:53:22.086Z" }, + { url = "https://files.pythonhosted.org/packages/fe/6d/60e8247564a72426570d0e0ea1151b95ce5bd2f1597bb878a18d32aec855/numpy-2.3.2-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:293b2192c6bcce487dbc6326de5853787f870aeb6c43f8f9c6496db5b1781e45", size = 14300519, upload-time = "2025-07-24T20:53:44.053Z" }, + { url = "https://files.pythonhosted.org/packages/4d/73/d8326c442cd428d47a067070c3ac6cc3b651a6e53613a1668342a12d4479/numpy-2.3.2-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:0a4f2021a6da53a0d580d6ef5db29947025ae8b35b3250141805ea9a32bbe86b", size = 5228972, upload-time = "2025-07-24T20:53:53.81Z" }, + { url = "https://files.pythonhosted.org/packages/34/2e/e71b2d6dad075271e7079db776196829019b90ce3ece5c69639e4f6fdc44/numpy-2.3.2-cp314-cp314t-macosx_14_0_x86_64.whl", hash = "sha256:9c144440db4bf3bb6372d2c3e49834cc0ff7bb4c24975ab33e01199e645416f2", size = 6737439, upload-time = "2025-07-24T20:54:04.742Z" }, + { url = "https://files.pythonhosted.org/packages/15/b0/d004bcd56c2c5e0500ffc65385eb6d569ffd3363cb5e593ae742749b2daa/numpy-2.3.2-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f92d6c2a8535dc4fe4419562294ff957f83a16ebdec66df0805e473ffaad8bd0", size = 14352479, upload-time = "2025-07-24T20:54:25.819Z" }, + { url = "https://files.pythonhosted.org/packages/11/e3/285142fcff8721e0c99b51686426165059874c150ea9ab898e12a492e291/numpy-2.3.2-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:cefc2219baa48e468e3db7e706305fcd0c095534a192a08f31e98d83a7d45fb0", size = 16702805, upload-time = "2025-07-24T20:54:50.814Z" }, + { url = "https://files.pythonhosted.org/packages/33/c3/33b56b0e47e604af2c7cd065edca892d180f5899599b76830652875249a3/numpy-2.3.2-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:76c3e9501ceb50b2ff3824c3589d5d1ab4ac857b0ee3f8f49629d0de55ecf7c2", size = 16133830, upload-time = "2025-07-24T20:55:17.306Z" }, + { url = "https://files.pythonhosted.org/packages/6e/ae/7b1476a1f4d6a48bc669b8deb09939c56dd2a439db1ab03017844374fb67/numpy-2.3.2-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:122bf5ed9a0221b3419672493878ba4967121514b1d7d4656a7580cd11dddcbf", size = 18652665, upload-time = "2025-07-24T20:55:46.665Z" }, + { url = "https://files.pythonhosted.org/packages/14/ba/5b5c9978c4bb161034148ade2de9db44ec316fab89ce8c400db0e0c81f86/numpy-2.3.2-cp314-cp314t-win32.whl", hash = "sha256:6f1ae3dcb840edccc45af496f312528c15b1f79ac318169d094e85e4bb35fdf1", size = 6514777, upload-time = "2025-07-24T20:55:57.66Z" }, + { url = "https://files.pythonhosted.org/packages/eb/46/3dbaf0ae7c17cdc46b9f662c56da2054887b8d9e737c1476f335c83d33db/numpy-2.3.2-cp314-cp314t-win_amd64.whl", hash = "sha256:087ffc25890d89a43536f75c5fe8770922008758e8eeeef61733957041ed2f9b", size = 13111856, upload-time = "2025-07-24T20:56:17.318Z" }, + { url = "https://files.pythonhosted.org/packages/c1/9e/1652778bce745a67b5fe05adde60ed362d38eb17d919a540e813d30f6874/numpy-2.3.2-cp314-cp314t-win_arm64.whl", hash = "sha256:092aeb3449833ea9c0bf0089d70c29ae480685dd2377ec9cdbbb620257f84631", size = 10544226, upload-time = "2025-07-24T20:56:34.509Z" }, + { url = "https://files.pythonhosted.org/packages/cf/ea/50ebc91d28b275b23b7128ef25c3d08152bc4068f42742867e07a870a42a/numpy-2.3.2-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:14a91ebac98813a49bc6aa1a0dfc09513dcec1d97eaf31ca21a87221a1cdcb15", size = 21130338, upload-time = "2025-07-24T20:57:54.37Z" }, + { url = "https://files.pythonhosted.org/packages/9f/57/cdd5eac00dd5f137277355c318a955c0d8fb8aa486020c22afd305f8b88f/numpy-2.3.2-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:71669b5daae692189540cffc4c439468d35a3f84f0c88b078ecd94337f6cb0ec", size = 14375776, upload-time = "2025-07-24T20:58:16.303Z" }, + { url = "https://files.pythonhosted.org/packages/83/85/27280c7f34fcd305c2209c0cdca4d70775e4859a9eaa92f850087f8dea50/numpy-2.3.2-pp311-pypy311_pp73-macosx_14_0_arm64.whl", hash = "sha256:69779198d9caee6e547adb933941ed7520f896fd9656834c300bdf4dd8642712", size = 5304882, upload-time = "2025-07-24T20:58:26.199Z" }, + { url = "https://files.pythonhosted.org/packages/48/b4/6500b24d278e15dd796f43824e69939d00981d37d9779e32499e823aa0aa/numpy-2.3.2-pp311-pypy311_pp73-macosx_14_0_x86_64.whl", hash = "sha256:2c3271cc4097beb5a60f010bcc1cc204b300bb3eafb4399376418a83a1c6373c", size = 6818405, upload-time = "2025-07-24T20:58:37.341Z" }, + { url = "https://files.pythonhosted.org/packages/9b/c9/142c1e03f199d202da8e980c2496213509291b6024fd2735ad28ae7065c7/numpy-2.3.2-pp311-pypy311_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8446acd11fe3dc1830568c941d44449fd5cb83068e5c70bd5a470d323d448296", size = 14419651, upload-time = "2025-07-24T20:58:59.048Z" }, + { url = "https://files.pythonhosted.org/packages/8b/95/8023e87cbea31a750a6c00ff9427d65ebc5fef104a136bfa69f76266d614/numpy-2.3.2-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:aa098a5ab53fa407fded5870865c6275a5cd4101cfdef8d6fafc48286a96e981", size = 16760166, upload-time = "2025-07-24T21:28:56.38Z" }, + { url = "https://files.pythonhosted.org/packages/78/e3/6690b3f85a05506733c7e90b577e4762517404ea78bab2ca3a5cb1aeb78d/numpy-2.3.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:6936aff90dda378c09bea075af0d9c675fe3a977a9d2402f95a87f440f59f619", size = 12977811, upload-time = "2025-07-24T21:29:18.234Z" }, ] [[package]] @@ -1852,61 +2399,82 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/8d/4c/e0370709aaf9d7ceb68f975cac559751e75954429a77e83202e680606560/opt_einsum_fx-0.1.4-py3-none-any.whl", hash = "sha256:85f489f4c7c31fd88d5faf9669c09e61ec37a30098809fdcfe2a08a9e42f23c9", size = 13213, upload-time = "2021-11-07T20:49:32.395Z" }, ] +[[package]] +name = "orb-models" +version = "0.5.4" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "ase" }, + { name = "cached-path" }, + { name = "dm-tree" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "torch" }, + { name = "tqdm" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/ec/63/385c78f164a8062fac89ef631a414463a6029d310d78cff9ee949ef2a9cd/orb_models-0.5.4.tar.gz", hash = "sha256:bc4e7b11eac16e9b1681cb667ccbdd263edf9702433a1eb106969dcc29ce7916", size = 87763, upload-time = "2025-04-30T09:08:39.804Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c3/f8/6c6d12caf9a6fd25e7cb4aaeedf01ebda88cbc5fdc4a9d1db48e2683d15d/orb_models-0.5.4-py3-none-any.whl", hash = "sha256:af096f30c39cb11965aee792092b00b8cc350fd7dfbc13d53528f223f1953fe7", size = 92237, upload-time = "2025-04-30T09:08:38.328Z" }, +] + [[package]] name = "packaging" -version = "24.2" +version = "25.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/d0/63/68dbb6eb2de9cb10ee4c9c14a0148804425e13c4fb20d61cce69f53106da/packaging-24.2.tar.gz", hash = "sha256:c228a6dc5e932d346bc5739379109d49e8853dd8223571c7c5b55260edc0b97f", size = 163950, upload-time = "2024-11-08T09:47:47.202Z" } +sdist = { url = "https://files.pythonhosted.org/packages/a1/d4/1fc4078c65507b51b96ca8f8c3ba19e6a61c8253c72794544580a7b6c24d/packaging-25.0.tar.gz", hash = "sha256:d443872c98d677bf60f6a1f2f8c1cb748e8fe762d2bf9d3148b5599295b0fc4f", size = 165727, upload-time = "2025-04-19T11:48:59.673Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/88/ef/eb23f262cca3c0c4eb7ab1933c3b1f03d021f2c48f54763065b6f0e321be/packaging-24.2-py3-none-any.whl", hash = "sha256:09abb1bccd265c01f4a3aa3f7a7db064b36514d2cba19a2f694fe6150451a759", size = 65451, upload-time = "2024-11-08T09:47:44.722Z" }, + { url = "https://files.pythonhosted.org/packages/20/12/38679034af332785aac8774540895e234f4d07f7545804097de4b666afd8/packaging-25.0-py3-none-any.whl", hash = "sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484", size = 66469, upload-time = "2025-04-19T11:48:57.875Z" }, ] [[package]] name = "pandas" -version = "2.2.3" +version = "2.3.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "python-dateutil" }, { name = "pytz" }, { name = "tzdata" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/9c/d6/9f8431bacc2e19dca897724cd097b1bb224a6ad5433784a44b587c7c13af/pandas-2.2.3.tar.gz", hash = "sha256:4f18ba62b61d7e192368b84517265a99b4d7ee8912f8708660fb4a366cc82667", size = 4399213, upload-time = "2024-09-20T13:10:04.827Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/aa/70/c853aec59839bceed032d52010ff5f1b8d87dc3114b762e4ba2727661a3b/pandas-2.2.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:1948ddde24197a0f7add2bdc4ca83bf2b1ef84a1bc8ccffd95eda17fd836ecb5", size = 12580827, upload-time = "2024-09-20T13:08:42.347Z" }, - { url = "https://files.pythonhosted.org/packages/99/f2/c4527768739ffa4469b2b4fff05aa3768a478aed89a2f271a79a40eee984/pandas-2.2.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:381175499d3802cde0eabbaf6324cce0c4f5d52ca6f8c377c29ad442f50f6348", size = 11303897, upload-time = "2024-09-20T13:08:45.807Z" }, - { url = "https://files.pythonhosted.org/packages/ed/12/86c1747ea27989d7a4064f806ce2bae2c6d575b950be087837bdfcabacc9/pandas-2.2.3-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:d9c45366def9a3dd85a6454c0e7908f2b3b8e9c138f5dc38fed7ce720d8453ed", size = 66480908, upload-time = "2024-09-20T18:37:13.513Z" }, - { url = "https://files.pythonhosted.org/packages/44/50/7db2cd5e6373ae796f0ddad3675268c8d59fb6076e66f0c339d61cea886b/pandas-2.2.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:86976a1c5b25ae3f8ccae3a5306e443569ee3c3faf444dfd0f41cda24667ad57", size = 13064210, upload-time = "2024-09-20T13:08:48.325Z" }, - { url = "https://files.pythonhosted.org/packages/61/61/a89015a6d5536cb0d6c3ba02cebed51a95538cf83472975275e28ebf7d0c/pandas-2.2.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b8661b0238a69d7aafe156b7fa86c44b881387509653fdf857bebc5e4008ad42", size = 16754292, upload-time = "2024-09-20T19:01:54.443Z" }, - { url = "https://files.pythonhosted.org/packages/ce/0d/4cc7b69ce37fac07645a94e1d4b0880b15999494372c1523508511b09e40/pandas-2.2.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:37e0aced3e8f539eccf2e099f65cdb9c8aa85109b0be6e93e2baff94264bdc6f", size = 14416379, upload-time = "2024-09-20T13:08:50.882Z" }, - { url = "https://files.pythonhosted.org/packages/31/9e/6ebb433de864a6cd45716af52a4d7a8c3c9aaf3a98368e61db9e69e69a9c/pandas-2.2.3-cp310-cp310-win_amd64.whl", hash = "sha256:56534ce0746a58afaf7942ba4863e0ef81c9c50d3f0ae93e9497d6a41a057645", size = 11598471, upload-time = "2024-09-20T13:08:53.332Z" }, - { url = "https://files.pythonhosted.org/packages/a8/44/d9502bf0ed197ba9bf1103c9867d5904ddcaf869e52329787fc54ed70cc8/pandas-2.2.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:66108071e1b935240e74525006034333f98bcdb87ea116de573a6a0dccb6c039", size = 12602222, upload-time = "2024-09-20T13:08:56.254Z" }, - { url = "https://files.pythonhosted.org/packages/52/11/9eac327a38834f162b8250aab32a6781339c69afe7574368fffe46387edf/pandas-2.2.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7c2875855b0ff77b2a64a0365e24455d9990730d6431b9e0ee18ad8acee13dbd", size = 11321274, upload-time = "2024-09-20T13:08:58.645Z" }, - { url = "https://files.pythonhosted.org/packages/45/fb/c4beeb084718598ba19aa9f5abbc8aed8b42f90930da861fcb1acdb54c3a/pandas-2.2.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:cd8d0c3be0515c12fed0bdbae072551c8b54b7192c7b1fda0ba56059a0179698", size = 15579836, upload-time = "2024-09-20T19:01:57.571Z" }, - { url = "https://files.pythonhosted.org/packages/cd/5f/4dba1d39bb9c38d574a9a22548c540177f78ea47b32f99c0ff2ec499fac5/pandas-2.2.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c124333816c3a9b03fbeef3a9f230ba9a737e9e5bb4060aa2107a86cc0a497fc", size = 13058505, upload-time = "2024-09-20T13:09:01.501Z" }, - { url = "https://files.pythonhosted.org/packages/b9/57/708135b90391995361636634df1f1130d03ba456e95bcf576fada459115a/pandas-2.2.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:63cc132e40a2e084cf01adf0775b15ac515ba905d7dcca47e9a251819c575ef3", size = 16744420, upload-time = "2024-09-20T19:02:00.678Z" }, - { url = "https://files.pythonhosted.org/packages/86/4a/03ed6b7ee323cf30404265c284cee9c65c56a212e0a08d9ee06984ba2240/pandas-2.2.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:29401dbfa9ad77319367d36940cd8a0b3a11aba16063e39632d98b0e931ddf32", size = 14440457, upload-time = "2024-09-20T13:09:04.105Z" }, - { url = "https://files.pythonhosted.org/packages/ed/8c/87ddf1fcb55d11f9f847e3c69bb1c6f8e46e2f40ab1a2d2abadb2401b007/pandas-2.2.3-cp311-cp311-win_amd64.whl", hash = "sha256:3fc6873a41186404dad67245896a6e440baacc92f5b716ccd1bc9ed2995ab2c5", size = 11617166, upload-time = "2024-09-20T13:09:06.917Z" }, - { url = "https://files.pythonhosted.org/packages/17/a3/fb2734118db0af37ea7433f57f722c0a56687e14b14690edff0cdb4b7e58/pandas-2.2.3-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:b1d432e8d08679a40e2a6d8b2f9770a5c21793a6f9f47fdd52c5ce1948a5a8a9", size = 12529893, upload-time = "2024-09-20T13:09:09.655Z" }, - { url = "https://files.pythonhosted.org/packages/e1/0c/ad295fd74bfac85358fd579e271cded3ac969de81f62dd0142c426b9da91/pandas-2.2.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:a5a1595fe639f5988ba6a8e5bc9649af3baf26df3998a0abe56c02609392e0a4", size = 11363475, upload-time = "2024-09-20T13:09:14.718Z" }, - { url = "https://files.pythonhosted.org/packages/c6/2a/4bba3f03f7d07207481fed47f5b35f556c7441acddc368ec43d6643c5777/pandas-2.2.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:5de54125a92bb4d1c051c0659e6fcb75256bf799a732a87184e5ea503965bce3", size = 15188645, upload-time = "2024-09-20T19:02:03.88Z" }, - { url = "https://files.pythonhosted.org/packages/38/f8/d8fddee9ed0d0c0f4a2132c1dfcf0e3e53265055da8df952a53e7eaf178c/pandas-2.2.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fffb8ae78d8af97f849404f21411c95062db1496aeb3e56f146f0355c9989319", size = 12739445, upload-time = "2024-09-20T13:09:17.621Z" }, - { url = "https://files.pythonhosted.org/packages/20/e8/45a05d9c39d2cea61ab175dbe6a2de1d05b679e8de2011da4ee190d7e748/pandas-2.2.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6dfcb5ee8d4d50c06a51c2fffa6cff6272098ad6540aed1a76d15fb9318194d8", size = 16359235, upload-time = "2024-09-20T19:02:07.094Z" }, - { url = "https://files.pythonhosted.org/packages/1d/99/617d07a6a5e429ff90c90da64d428516605a1ec7d7bea494235e1c3882de/pandas-2.2.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:062309c1b9ea12a50e8ce661145c6aab431b1e99530d3cd60640e255778bd43a", size = 14056756, upload-time = "2024-09-20T13:09:20.474Z" }, - { url = "https://files.pythonhosted.org/packages/29/d4/1244ab8edf173a10fd601f7e13b9566c1b525c4f365d6bee918e68381889/pandas-2.2.3-cp312-cp312-win_amd64.whl", hash = "sha256:59ef3764d0fe818125a5097d2ae867ca3fa64df032331b7e0917cf5d7bf66b13", size = 11504248, upload-time = "2024-09-20T13:09:23.137Z" }, - { url = "https://files.pythonhosted.org/packages/64/22/3b8f4e0ed70644e85cfdcd57454686b9057c6c38d2f74fe4b8bc2527214a/pandas-2.2.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:f00d1345d84d8c86a63e476bb4955e46458b304b9575dcf71102b5c705320015", size = 12477643, upload-time = "2024-09-20T13:09:25.522Z" }, - { url = "https://files.pythonhosted.org/packages/e4/93/b3f5d1838500e22c8d793625da672f3eec046b1a99257666c94446969282/pandas-2.2.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3508d914817e153ad359d7e069d752cdd736a247c322d932eb89e6bc84217f28", size = 11281573, upload-time = "2024-09-20T13:09:28.012Z" }, - { url = "https://files.pythonhosted.org/packages/f5/94/6c79b07f0e5aab1dcfa35a75f4817f5c4f677931d4234afcd75f0e6a66ca/pandas-2.2.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:22a9d949bfc9a502d320aa04e5d02feab689d61da4e7764b62c30b991c42c5f0", size = 15196085, upload-time = "2024-09-20T19:02:10.451Z" }, - { url = "https://files.pythonhosted.org/packages/e8/31/aa8da88ca0eadbabd0a639788a6da13bb2ff6edbbb9f29aa786450a30a91/pandas-2.2.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f3a255b2c19987fbbe62a9dfd6cff7ff2aa9ccab3fc75218fd4b7530f01efa24", size = 12711809, upload-time = "2024-09-20T13:09:30.814Z" }, - { url = "https://files.pythonhosted.org/packages/ee/7c/c6dbdb0cb2a4344cacfb8de1c5808ca885b2e4dcfde8008266608f9372af/pandas-2.2.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:800250ecdadb6d9c78eae4990da62743b857b470883fa27f652db8bdde7f6659", size = 16356316, upload-time = "2024-09-20T19:02:13.825Z" }, - { url = "https://files.pythonhosted.org/packages/57/b7/8b757e7d92023b832869fa8881a992696a0bfe2e26f72c9ae9f255988d42/pandas-2.2.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6374c452ff3ec675a8f46fd9ab25c4ad0ba590b71cf0656f8b6daa5202bca3fb", size = 14022055, upload-time = "2024-09-20T13:09:33.462Z" }, - { url = "https://files.pythonhosted.org/packages/3b/bc/4b18e2b8c002572c5a441a64826252ce5da2aa738855747247a971988043/pandas-2.2.3-cp313-cp313-win_amd64.whl", hash = "sha256:61c5ad4043f791b61dd4752191d9f07f0ae412515d59ba8f005832a532f8736d", size = 11481175, upload-time = "2024-09-20T13:09:35.871Z" }, - { url = "https://files.pythonhosted.org/packages/76/a3/a5d88146815e972d40d19247b2c162e88213ef51c7c25993942c39dbf41d/pandas-2.2.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3b71f27954685ee685317063bf13c7709a7ba74fc996b84fc6821c59b0f06468", size = 12615650, upload-time = "2024-09-20T13:09:38.685Z" }, - { url = "https://files.pythonhosted.org/packages/9c/8c/f0fd18f6140ddafc0c24122c8a964e48294acc579d47def376fef12bcb4a/pandas-2.2.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:38cf8125c40dae9d5acc10fa66af8ea6fdf760b2714ee482ca691fc66e6fcb18", size = 11290177, upload-time = "2024-09-20T13:09:41.141Z" }, - { url = "https://files.pythonhosted.org/packages/ed/f9/e995754eab9c0f14c6777401f7eece0943840b7a9fc932221c19d1abee9f/pandas-2.2.3-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:ba96630bc17c875161df3818780af30e43be9b166ce51c9a18c1feae342906c2", size = 14651526, upload-time = "2024-09-20T19:02:16.905Z" }, - { url = "https://files.pythonhosted.org/packages/25/b0/98d6ae2e1abac4f35230aa756005e8654649d305df9a28b16b9ae4353bff/pandas-2.2.3-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1db71525a1538b30142094edb9adc10be3f3e176748cd7acc2240c2f2e5aa3a4", size = 11871013, upload-time = "2024-09-20T13:09:44.39Z" }, - { url = "https://files.pythonhosted.org/packages/cc/57/0f72a10f9db6a4628744c8e8f0df4e6e21de01212c7c981d31e50ffc8328/pandas-2.2.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:15c0e1e02e93116177d29ff83e8b1619c93ddc9c49083f237d4312337a61165d", size = 15711620, upload-time = "2024-09-20T19:02:20.639Z" }, - { url = "https://files.pythonhosted.org/packages/ab/5f/b38085618b950b79d2d9164a711c52b10aefc0ae6833b96f626b7021b2ed/pandas-2.2.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:ad5b65698ab28ed8d7f18790a0dc58005c7629f227be9ecc1072aa74c0c1d43a", size = 13098436, upload-time = "2024-09-20T13:09:48.112Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/d1/6f/75aa71f8a14267117adeeed5d21b204770189c0a0025acbdc03c337b28fc/pandas-2.3.1.tar.gz", hash = "sha256:0a95b9ac964fe83ce317827f80304d37388ea77616b1425f0ae41c9d2d0d7bb2", size = 4487493, upload-time = "2025-07-07T19:20:04.079Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c4/ca/aa97b47287221fa37a49634532e520300088e290b20d690b21ce3e448143/pandas-2.3.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:22c2e866f7209ebc3a8f08d75766566aae02bcc91d196935a1d9e59c7b990ac9", size = 11542731, upload-time = "2025-07-07T19:18:12.619Z" }, + { url = "https://files.pythonhosted.org/packages/80/bf/7938dddc5f01e18e573dcfb0f1b8c9357d9b5fa6ffdee6e605b92efbdff2/pandas-2.3.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3583d348546201aff730c8c47e49bc159833f971c2899d6097bce68b9112a4f1", size = 10790031, upload-time = "2025-07-07T19:18:16.611Z" }, + { url = "https://files.pythonhosted.org/packages/ee/2f/9af748366763b2a494fed477f88051dbf06f56053d5c00eba652697e3f94/pandas-2.3.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0f951fbb702dacd390561e0ea45cdd8ecfa7fb56935eb3dd78e306c19104b9b0", size = 11724083, upload-time = "2025-07-07T19:18:20.512Z" }, + { url = "https://files.pythonhosted.org/packages/2c/95/79ab37aa4c25d1e7df953dde407bb9c3e4ae47d154bc0dd1692f3a6dcf8c/pandas-2.3.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cd05b72ec02ebfb993569b4931b2e16fbb4d6ad6ce80224a3ee838387d83a191", size = 12342360, upload-time = "2025-07-07T19:18:23.194Z" }, + { url = "https://files.pythonhosted.org/packages/75/a7/d65e5d8665c12c3c6ff5edd9709d5836ec9b6f80071b7f4a718c6106e86e/pandas-2.3.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:1b916a627919a247d865aed068eb65eb91a344b13f5b57ab9f610b7716c92de1", size = 13202098, upload-time = "2025-07-07T19:18:25.558Z" }, + { url = "https://files.pythonhosted.org/packages/65/f3/4c1dbd754dbaa79dbf8b537800cb2fa1a6e534764fef50ab1f7533226c5c/pandas-2.3.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:fe67dc676818c186d5a3d5425250e40f179c2a89145df477dd82945eaea89e97", size = 13837228, upload-time = "2025-07-07T19:18:28.344Z" }, + { url = "https://files.pythonhosted.org/packages/3f/d6/d7f5777162aa9b48ec3910bca5a58c9b5927cfd9cfde3aa64322f5ba4b9f/pandas-2.3.1-cp310-cp310-win_amd64.whl", hash = "sha256:2eb789ae0274672acbd3c575b0598d213345660120a257b47b5dafdc618aec83", size = 11336561, upload-time = "2025-07-07T19:18:31.211Z" }, + { url = "https://files.pythonhosted.org/packages/76/1c/ccf70029e927e473a4476c00e0d5b32e623bff27f0402d0a92b7fc29bb9f/pandas-2.3.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2b0540963d83431f5ce8870ea02a7430adca100cec8a050f0811f8e31035541b", size = 11566608, upload-time = "2025-07-07T19:18:33.86Z" }, + { url = "https://files.pythonhosted.org/packages/ec/d3/3c37cb724d76a841f14b8f5fe57e5e3645207cc67370e4f84717e8bb7657/pandas-2.3.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:fe7317f578c6a153912bd2292f02e40c1d8f253e93c599e82620c7f69755c74f", size = 10823181, upload-time = "2025-07-07T19:18:36.151Z" }, + { url = "https://files.pythonhosted.org/packages/8a/4c/367c98854a1251940edf54a4df0826dcacfb987f9068abf3e3064081a382/pandas-2.3.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e6723a27ad7b244c0c79d8e7007092d7c8f0f11305770e2f4cd778b3ad5f9f85", size = 11793570, upload-time = "2025-07-07T19:18:38.385Z" }, + { url = "https://files.pythonhosted.org/packages/07/5f/63760ff107bcf5146eee41b38b3985f9055e710a72fdd637b791dea3495c/pandas-2.3.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3462c3735fe19f2638f2c3a40bd94ec2dc5ba13abbb032dd2fa1f540a075509d", size = 12378887, upload-time = "2025-07-07T19:18:41.284Z" }, + { url = "https://files.pythonhosted.org/packages/15/53/f31a9b4dfe73fe4711c3a609bd8e60238022f48eacedc257cd13ae9327a7/pandas-2.3.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:98bcc8b5bf7afed22cc753a28bc4d9e26e078e777066bc53fac7904ddef9a678", size = 13230957, upload-time = "2025-07-07T19:18:44.187Z" }, + { url = "https://files.pythonhosted.org/packages/e0/94/6fce6bf85b5056d065e0a7933cba2616dcb48596f7ba3c6341ec4bcc529d/pandas-2.3.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:4d544806b485ddf29e52d75b1f559142514e60ef58a832f74fb38e48d757b299", size = 13883883, upload-time = "2025-07-07T19:18:46.498Z" }, + { url = "https://files.pythonhosted.org/packages/c8/7b/bdcb1ed8fccb63d04bdb7635161d0ec26596d92c9d7a6cce964e7876b6c1/pandas-2.3.1-cp311-cp311-win_amd64.whl", hash = "sha256:b3cd4273d3cb3707b6fffd217204c52ed92859533e31dc03b7c5008aa933aaab", size = 11340212, upload-time = "2025-07-07T19:18:49.293Z" }, + { url = "https://files.pythonhosted.org/packages/46/de/b8445e0f5d217a99fe0eeb2f4988070908979bec3587c0633e5428ab596c/pandas-2.3.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:689968e841136f9e542020698ee1c4fbe9caa2ed2213ae2388dc7b81721510d3", size = 11588172, upload-time = "2025-07-07T19:18:52.054Z" }, + { url = "https://files.pythonhosted.org/packages/1e/e0/801cdb3564e65a5ac041ab99ea6f1d802a6c325bb6e58c79c06a3f1cd010/pandas-2.3.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:025e92411c16cbe5bb2a4abc99732a6b132f439b8aab23a59fa593eb00704232", size = 10717365, upload-time = "2025-07-07T19:18:54.785Z" }, + { url = "https://files.pythonhosted.org/packages/51/a5/c76a8311833c24ae61a376dbf360eb1b1c9247a5d9c1e8b356563b31b80c/pandas-2.3.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9b7ff55f31c4fcb3e316e8f7fa194566b286d6ac430afec0d461163312c5841e", size = 11280411, upload-time = "2025-07-07T19:18:57.045Z" }, + { url = "https://files.pythonhosted.org/packages/da/01/e383018feba0a1ead6cf5fe8728e5d767fee02f06a3d800e82c489e5daaf/pandas-2.3.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7dcb79bf373a47d2a40cf7232928eb7540155abbc460925c2c96d2d30b006eb4", size = 11988013, upload-time = "2025-07-07T19:18:59.771Z" }, + { url = "https://files.pythonhosted.org/packages/5b/14/cec7760d7c9507f11c97d64f29022e12a6cc4fc03ac694535e89f88ad2ec/pandas-2.3.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:56a342b231e8862c96bdb6ab97170e203ce511f4d0429589c8ede1ee8ece48b8", size = 12767210, upload-time = "2025-07-07T19:19:02.944Z" }, + { url = "https://files.pythonhosted.org/packages/50/b9/6e2d2c6728ed29fb3d4d4d302504fb66f1a543e37eb2e43f352a86365cdf/pandas-2.3.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ca7ed14832bce68baef331f4d7f294411bed8efd032f8109d690df45e00c4679", size = 13440571, upload-time = "2025-07-07T19:19:06.82Z" }, + { url = "https://files.pythonhosted.org/packages/80/a5/3a92893e7399a691bad7664d977cb5e7c81cf666c81f89ea76ba2bff483d/pandas-2.3.1-cp312-cp312-win_amd64.whl", hash = "sha256:ac942bfd0aca577bef61f2bc8da8147c4ef6879965ef883d8e8d5d2dc3e744b8", size = 10987601, upload-time = "2025-07-07T19:19:09.589Z" }, + { url = "https://files.pythonhosted.org/packages/32/ed/ff0a67a2c5505e1854e6715586ac6693dd860fbf52ef9f81edee200266e7/pandas-2.3.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9026bd4a80108fac2239294a15ef9003c4ee191a0f64b90f170b40cfb7cf2d22", size = 11531393, upload-time = "2025-07-07T19:19:12.245Z" }, + { url = "https://files.pythonhosted.org/packages/c7/db/d8f24a7cc9fb0972adab0cc80b6817e8bef888cfd0024eeb5a21c0bb5c4a/pandas-2.3.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:6de8547d4fdb12421e2d047a2c446c623ff4c11f47fddb6b9169eb98ffba485a", size = 10668750, upload-time = "2025-07-07T19:19:14.612Z" }, + { url = "https://files.pythonhosted.org/packages/0f/b0/80f6ec783313f1e2356b28b4fd8d2148c378370045da918c73145e6aab50/pandas-2.3.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:782647ddc63c83133b2506912cc6b108140a38a37292102aaa19c81c83db2928", size = 11342004, upload-time = "2025-07-07T19:19:16.857Z" }, + { url = "https://files.pythonhosted.org/packages/e9/e2/20a317688435470872885e7fc8f95109ae9683dec7c50be29b56911515a5/pandas-2.3.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2ba6aff74075311fc88504b1db890187a3cd0f887a5b10f5525f8e2ef55bfdb9", size = 12050869, upload-time = "2025-07-07T19:19:19.265Z" }, + { url = "https://files.pythonhosted.org/packages/55/79/20d746b0a96c67203a5bee5fb4e00ac49c3e8009a39e1f78de264ecc5729/pandas-2.3.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:e5635178b387bd2ba4ac040f82bc2ef6e6b500483975c4ebacd34bec945fda12", size = 12750218, upload-time = "2025-07-07T19:19:21.547Z" }, + { url = "https://files.pythonhosted.org/packages/7c/0f/145c8b41e48dbf03dd18fdd7f24f8ba95b8254a97a3379048378f33e7838/pandas-2.3.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6f3bf5ec947526106399a9e1d26d40ee2b259c66422efdf4de63c848492d91bb", size = 13416763, upload-time = "2025-07-07T19:19:23.939Z" }, + { url = "https://files.pythonhosted.org/packages/b2/c0/54415af59db5cdd86a3d3bf79863e8cc3fa9ed265f0745254061ac09d5f2/pandas-2.3.1-cp313-cp313-win_amd64.whl", hash = "sha256:1c78cf43c8fde236342a1cb2c34bcff89564a7bfed7e474ed2fffa6aed03a956", size = 10987482, upload-time = "2025-07-07T19:19:42.699Z" }, + { url = "https://files.pythonhosted.org/packages/48/64/2fd2e400073a1230e13b8cd604c9bc95d9e3b962e5d44088ead2e8f0cfec/pandas-2.3.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:8dfc17328e8da77be3cf9f47509e5637ba8f137148ed0e9b5241e1baf526e20a", size = 12029159, upload-time = "2025-07-07T19:19:26.362Z" }, + { url = "https://files.pythonhosted.org/packages/d8/0a/d84fd79b0293b7ef88c760d7dca69828d867c89b6d9bc52d6a27e4d87316/pandas-2.3.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:ec6c851509364c59a5344458ab935e6451b31b818be467eb24b0fe89bd05b6b9", size = 11393287, upload-time = "2025-07-07T19:19:29.157Z" }, + { url = "https://files.pythonhosted.org/packages/50/ae/ff885d2b6e88f3c7520bb74ba319268b42f05d7e583b5dded9837da2723f/pandas-2.3.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:911580460fc4884d9b05254b38a6bfadddfcc6aaef856fb5859e7ca202e45275", size = 11309381, upload-time = "2025-07-07T19:19:31.436Z" }, + { url = "https://files.pythonhosted.org/packages/85/86/1fa345fc17caf5d7780d2699985c03dbe186c68fee00b526813939062bb0/pandas-2.3.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2f4d6feeba91744872a600e6edbbd5b033005b431d5ae8379abee5bcfa479fab", size = 11883998, upload-time = "2025-07-07T19:19:34.267Z" }, + { url = "https://files.pythonhosted.org/packages/81/aa/e58541a49b5e6310d89474333e994ee57fea97c8aaa8fc7f00b873059bbf/pandas-2.3.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:fe37e757f462d31a9cd7580236a82f353f5713a80e059a29753cf938c6775d96", size = 12704705, upload-time = "2025-07-07T19:19:36.856Z" }, + { url = "https://files.pythonhosted.org/packages/d5/f9/07086f5b0f2a19872554abeea7658200824f5835c58a106fa8f2ae96a46c/pandas-2.3.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:5db9637dbc24b631ff3707269ae4559bce4b7fd75c1c4d7e13f40edc42df4444", size = 13189044, upload-time = "2025-07-07T19:19:39.999Z" }, ] [[package]] @@ -1923,7 +2491,8 @@ name = "patsy" version = "1.0.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/d1/81/74f6a65b848ffd16c18f920620ce999fe45fe27f01ab3911260ce4ed85e4/patsy-1.0.1.tar.gz", hash = "sha256:e786a9391eec818c054e359b737bbce692f051aee4c661f4141cc88fb459c0c4", size = 396010, upload-time = "2024-11-12T14:10:54.642Z" } wheels = [ @@ -1944,115 +2513,151 @@ wheels = [ [[package]] name = "pillow" -version = "11.1.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/f3/af/c097e544e7bd278333db77933e535098c259609c4eb3b85381109602fb5b/pillow-11.1.0.tar.gz", hash = "sha256:368da70808b36d73b4b390a8ffac11069f8a5c85f29eff1f1b01bcf3ef5b2a20", size = 46742715, upload-time = "2025-01-02T08:13:58.407Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/50/1c/2dcea34ac3d7bc96a1fd1bd0a6e06a57c67167fec2cff8d95d88229a8817/pillow-11.1.0-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:e1abe69aca89514737465752b4bcaf8016de61b3be1397a8fc260ba33321b3a8", size = 3229983, upload-time = "2025-01-02T08:10:16.008Z" }, - { url = "https://files.pythonhosted.org/packages/14/ca/6bec3df25e4c88432681de94a3531cc738bd85dea6c7aa6ab6f81ad8bd11/pillow-11.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c640e5a06869c75994624551f45e5506e4256562ead981cce820d5ab39ae2192", size = 3101831, upload-time = "2025-01-02T08:10:18.774Z" }, - { url = "https://files.pythonhosted.org/packages/d4/2c/668e18e5521e46eb9667b09e501d8e07049eb5bfe39d56be0724a43117e6/pillow-11.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a07dba04c5e22824816b2615ad7a7484432d7f540e6fa86af60d2de57b0fcee2", size = 4314074, upload-time = "2025-01-02T08:10:21.114Z" }, - { url = "https://files.pythonhosted.org/packages/02/80/79f99b714f0fc25f6a8499ecfd1f810df12aec170ea1e32a4f75746051ce/pillow-11.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e267b0ed063341f3e60acd25c05200df4193e15a4a5807075cd71225a2386e26", size = 4394933, upload-time = "2025-01-02T08:10:23.982Z" }, - { url = "https://files.pythonhosted.org/packages/81/aa/8d4ad25dc11fd10a2001d5b8a80fdc0e564ac33b293bdfe04ed387e0fd95/pillow-11.1.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:bd165131fd51697e22421d0e467997ad31621b74bfc0b75956608cb2906dda07", size = 4353349, upload-time = "2025-01-02T08:10:25.887Z" }, - { url = "https://files.pythonhosted.org/packages/84/7a/cd0c3eaf4a28cb2a74bdd19129f7726277a7f30c4f8424cd27a62987d864/pillow-11.1.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:abc56501c3fd148d60659aae0af6ddc149660469082859fa7b066a298bde9482", size = 4476532, upload-time = "2025-01-02T08:10:28.129Z" }, - { url = "https://files.pythonhosted.org/packages/8f/8b/a907fdd3ae8f01c7670dfb1499c53c28e217c338b47a813af8d815e7ce97/pillow-11.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:54ce1c9a16a9561b6d6d8cb30089ab1e5eb66918cb47d457bd996ef34182922e", size = 4279789, upload-time = "2025-01-02T08:10:32.976Z" }, - { url = "https://files.pythonhosted.org/packages/6f/9a/9f139d9e8cccd661c3efbf6898967a9a337eb2e9be2b454ba0a09533100d/pillow-11.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:73ddde795ee9b06257dac5ad42fcb07f3b9b813f8c1f7f870f402f4dc54b5269", size = 4413131, upload-time = "2025-01-02T08:10:36.912Z" }, - { url = "https://files.pythonhosted.org/packages/a8/68/0d8d461f42a3f37432203c8e6df94da10ac8081b6d35af1c203bf3111088/pillow-11.1.0-cp310-cp310-win32.whl", hash = "sha256:3a5fe20a7b66e8135d7fd617b13272626a28278d0e578c98720d9ba4b2439d49", size = 2291213, upload-time = "2025-01-02T08:10:40.186Z" }, - { url = "https://files.pythonhosted.org/packages/14/81/d0dff759a74ba87715509af9f6cb21fa21d93b02b3316ed43bda83664db9/pillow-11.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:b6123aa4a59d75f06e9dd3dac5bf8bc9aa383121bb3dd9a7a612e05eabc9961a", size = 2625725, upload-time = "2025-01-02T08:10:42.404Z" }, - { url = "https://files.pythonhosted.org/packages/ce/1f/8d50c096a1d58ef0584ddc37e6f602828515219e9d2428e14ce50f5ecad1/pillow-11.1.0-cp310-cp310-win_arm64.whl", hash = "sha256:a76da0a31da6fcae4210aa94fd779c65c75786bc9af06289cd1c184451ef7a65", size = 2375213, upload-time = "2025-01-02T08:10:44.173Z" }, - { url = "https://files.pythonhosted.org/packages/dd/d6/2000bfd8d5414fb70cbbe52c8332f2283ff30ed66a9cde42716c8ecbe22c/pillow-11.1.0-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:e06695e0326d05b06833b40b7ef477e475d0b1ba3a6d27da1bb48c23209bf457", size = 3229968, upload-time = "2025-01-02T08:10:48.172Z" }, - { url = "https://files.pythonhosted.org/packages/d9/45/3fe487010dd9ce0a06adf9b8ff4f273cc0a44536e234b0fad3532a42c15b/pillow-11.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:96f82000e12f23e4f29346e42702b6ed9a2f2fea34a740dd5ffffcc8c539eb35", size = 3101806, upload-time = "2025-01-02T08:10:50.981Z" }, - { url = "https://files.pythonhosted.org/packages/e3/72/776b3629c47d9d5f1c160113158a7a7ad177688d3a1159cd3b62ded5a33a/pillow-11.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a3cd561ded2cf2bbae44d4605837221b987c216cff94f49dfeed63488bb228d2", size = 4322283, upload-time = "2025-01-02T08:10:54.724Z" }, - { url = "https://files.pythonhosted.org/packages/e4/c2/e25199e7e4e71d64eeb869f5b72c7ddec70e0a87926398785ab944d92375/pillow-11.1.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f189805c8be5ca5add39e6f899e6ce2ed824e65fb45f3c28cb2841911da19070", size = 4402945, upload-time = "2025-01-02T08:10:57.376Z" }, - { url = "https://files.pythonhosted.org/packages/c1/ed/51d6136c9d5911f78632b1b86c45241c712c5a80ed7fa7f9120a5dff1eba/pillow-11.1.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:dd0052e9db3474df30433f83a71b9b23bd9e4ef1de13d92df21a52c0303b8ab6", size = 4361228, upload-time = "2025-01-02T08:11:02.374Z" }, - { url = "https://files.pythonhosted.org/packages/48/a4/fbfe9d5581d7b111b28f1d8c2762dee92e9821bb209af9fa83c940e507a0/pillow-11.1.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:837060a8599b8f5d402e97197d4924f05a2e0d68756998345c829c33186217b1", size = 4484021, upload-time = "2025-01-02T08:11:04.431Z" }, - { url = "https://files.pythonhosted.org/packages/39/db/0b3c1a5018117f3c1d4df671fb8e47d08937f27519e8614bbe86153b65a5/pillow-11.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:aa8dd43daa836b9a8128dbe7d923423e5ad86f50a7a14dc688194b7be5c0dea2", size = 4287449, upload-time = "2025-01-02T08:11:07.412Z" }, - { url = "https://files.pythonhosted.org/packages/d9/58/bc128da7fea8c89fc85e09f773c4901e95b5936000e6f303222490c052f3/pillow-11.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:0a2f91f8a8b367e7a57c6e91cd25af510168091fb89ec5146003e424e1558a96", size = 4419972, upload-time = "2025-01-02T08:11:09.508Z" }, - { url = "https://files.pythonhosted.org/packages/5f/bb/58f34379bde9fe197f51841c5bbe8830c28bbb6d3801f16a83b8f2ad37df/pillow-11.1.0-cp311-cp311-win32.whl", hash = "sha256:c12fc111ef090845de2bb15009372175d76ac99969bdf31e2ce9b42e4b8cd88f", size = 2291201, upload-time = "2025-01-02T08:11:13.056Z" }, - { url = "https://files.pythonhosted.org/packages/3a/c6/fce9255272bcf0c39e15abd2f8fd8429a954cf344469eaceb9d0d1366913/pillow-11.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:fbd43429d0d7ed6533b25fc993861b8fd512c42d04514a0dd6337fb3ccf22761", size = 2625686, upload-time = "2025-01-02T08:11:16.547Z" }, - { url = "https://files.pythonhosted.org/packages/c8/52/8ba066d569d932365509054859f74f2a9abee273edcef5cd75e4bc3e831e/pillow-11.1.0-cp311-cp311-win_arm64.whl", hash = "sha256:f7955ecf5609dee9442cbface754f2c6e541d9e6eda87fad7f7a989b0bdb9d71", size = 2375194, upload-time = "2025-01-02T08:11:19.897Z" }, - { url = "https://files.pythonhosted.org/packages/95/20/9ce6ed62c91c073fcaa23d216e68289e19d95fb8188b9fb7a63d36771db8/pillow-11.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:2062ffb1d36544d42fcaa277b069c88b01bb7298f4efa06731a7fd6cc290b81a", size = 3226818, upload-time = "2025-01-02T08:11:22.518Z" }, - { url = "https://files.pythonhosted.org/packages/b9/d8/f6004d98579a2596c098d1e30d10b248798cceff82d2b77aa914875bfea1/pillow-11.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:a85b653980faad27e88b141348707ceeef8a1186f75ecc600c395dcac19f385b", size = 3101662, upload-time = "2025-01-02T08:11:25.19Z" }, - { url = "https://files.pythonhosted.org/packages/08/d9/892e705f90051c7a2574d9f24579c9e100c828700d78a63239676f960b74/pillow-11.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9409c080586d1f683df3f184f20e36fb647f2e0bc3988094d4fd8c9f4eb1b3b3", size = 4329317, upload-time = "2025-01-02T08:11:30.371Z" }, - { url = "https://files.pythonhosted.org/packages/8c/aa/7f29711f26680eab0bcd3ecdd6d23ed6bce180d82e3f6380fb7ae35fcf3b/pillow-11.1.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7fdadc077553621911f27ce206ffcbec7d3f8d7b50e0da39f10997e8e2bb7f6a", size = 4412999, upload-time = "2025-01-02T08:11:33.499Z" }, - { url = "https://files.pythonhosted.org/packages/c8/c4/8f0fe3b9e0f7196f6d0bbb151f9fba323d72a41da068610c4c960b16632a/pillow-11.1.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:93a18841d09bcdd774dcdc308e4537e1f867b3dec059c131fde0327899734aa1", size = 4368819, upload-time = "2025-01-02T08:11:37.304Z" }, - { url = "https://files.pythonhosted.org/packages/38/0d/84200ed6a871ce386ddc82904bfadc0c6b28b0c0ec78176871a4679e40b3/pillow-11.1.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:9aa9aeddeed452b2f616ff5507459e7bab436916ccb10961c4a382cd3e03f47f", size = 4496081, upload-time = "2025-01-02T08:11:39.598Z" }, - { url = "https://files.pythonhosted.org/packages/84/9c/9bcd66f714d7e25b64118e3952d52841a4babc6d97b6d28e2261c52045d4/pillow-11.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:3cdcdb0b896e981678eee140d882b70092dac83ac1cdf6b3a60e2216a73f2b91", size = 4296513, upload-time = "2025-01-02T08:11:43.083Z" }, - { url = "https://files.pythonhosted.org/packages/db/61/ada2a226e22da011b45f7104c95ebda1b63dcbb0c378ad0f7c2a710f8fd2/pillow-11.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:36ba10b9cb413e7c7dfa3e189aba252deee0602c86c309799da5a74009ac7a1c", size = 4431298, upload-time = "2025-01-02T08:11:46.626Z" }, - { url = "https://files.pythonhosted.org/packages/e7/c4/fc6e86750523f367923522014b821c11ebc5ad402e659d8c9d09b3c9d70c/pillow-11.1.0-cp312-cp312-win32.whl", hash = "sha256:cfd5cd998c2e36a862d0e27b2df63237e67273f2fc78f47445b14e73a810e7e6", size = 2291630, upload-time = "2025-01-02T08:11:49.401Z" }, - { url = "https://files.pythonhosted.org/packages/08/5c/2104299949b9d504baf3f4d35f73dbd14ef31bbd1ddc2c1b66a5b7dfda44/pillow-11.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:a697cd8ba0383bba3d2d3ada02b34ed268cb548b369943cd349007730c92bddf", size = 2626369, upload-time = "2025-01-02T08:11:52.02Z" }, - { url = "https://files.pythonhosted.org/packages/37/f3/9b18362206b244167c958984b57c7f70a0289bfb59a530dd8af5f699b910/pillow-11.1.0-cp312-cp312-win_arm64.whl", hash = "sha256:4dd43a78897793f60766563969442020e90eb7847463eca901e41ba186a7d4a5", size = 2375240, upload-time = "2025-01-02T08:11:56.193Z" }, - { url = "https://files.pythonhosted.org/packages/b3/31/9ca79cafdce364fd5c980cd3416c20ce1bebd235b470d262f9d24d810184/pillow-11.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:ae98e14432d458fc3de11a77ccb3ae65ddce70f730e7c76140653048c71bfcbc", size = 3226640, upload-time = "2025-01-02T08:11:58.329Z" }, - { url = "https://files.pythonhosted.org/packages/ac/0f/ff07ad45a1f172a497aa393b13a9d81a32e1477ef0e869d030e3c1532521/pillow-11.1.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:cc1331b6d5a6e144aeb5e626f4375f5b7ae9934ba620c0ac6b3e43d5e683a0f0", size = 3101437, upload-time = "2025-01-02T08:12:01.797Z" }, - { url = "https://files.pythonhosted.org/packages/08/2f/9906fca87a68d29ec4530be1f893149e0cb64a86d1f9f70a7cfcdfe8ae44/pillow-11.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:758e9d4ef15d3560214cddbc97b8ef3ef86ce04d62ddac17ad39ba87e89bd3b1", size = 4326605, upload-time = "2025-01-02T08:12:05.224Z" }, - { url = "https://files.pythonhosted.org/packages/b0/0f/f3547ee15b145bc5c8b336401b2d4c9d9da67da9dcb572d7c0d4103d2c69/pillow-11.1.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b523466b1a31d0dcef7c5be1f20b942919b62fd6e9a9be199d035509cbefc0ec", size = 4411173, upload-time = "2025-01-02T08:12:08.281Z" }, - { url = "https://files.pythonhosted.org/packages/b1/df/bf8176aa5db515c5de584c5e00df9bab0713548fd780c82a86cba2c2fedb/pillow-11.1.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:9044b5e4f7083f209c4e35aa5dd54b1dd5b112b108648f5c902ad586d4f945c5", size = 4369145, upload-time = "2025-01-02T08:12:11.411Z" }, - { url = "https://files.pythonhosted.org/packages/de/7c/7433122d1cfadc740f577cb55526fdc39129a648ac65ce64db2eb7209277/pillow-11.1.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:3764d53e09cdedd91bee65c2527815d315c6b90d7b8b79759cc48d7bf5d4f114", size = 4496340, upload-time = "2025-01-02T08:12:15.29Z" }, - { url = "https://files.pythonhosted.org/packages/25/46/dd94b93ca6bd555588835f2504bd90c00d5438fe131cf01cfa0c5131a19d/pillow-11.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:31eba6bbdd27dde97b0174ddf0297d7a9c3a507a8a1480e1e60ef914fe23d352", size = 4296906, upload-time = "2025-01-02T08:12:17.485Z" }, - { url = "https://files.pythonhosted.org/packages/a8/28/2f9d32014dfc7753e586db9add35b8a41b7a3b46540e965cb6d6bc607bd2/pillow-11.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:b5d658fbd9f0d6eea113aea286b21d3cd4d3fd978157cbf2447a6035916506d3", size = 4431759, upload-time = "2025-01-02T08:12:20.382Z" }, - { url = "https://files.pythonhosted.org/packages/33/48/19c2cbe7403870fbe8b7737d19eb013f46299cdfe4501573367f6396c775/pillow-11.1.0-cp313-cp313-win32.whl", hash = "sha256:f86d3a7a9af5d826744fabf4afd15b9dfef44fe69a98541f666f66fbb8d3fef9", size = 2291657, upload-time = "2025-01-02T08:12:23.922Z" }, - { url = "https://files.pythonhosted.org/packages/3b/ad/285c556747d34c399f332ba7c1a595ba245796ef3e22eae190f5364bb62b/pillow-11.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:593c5fd6be85da83656b93ffcccc2312d2d149d251e98588b14fbc288fd8909c", size = 2626304, upload-time = "2025-01-02T08:12:28.069Z" }, - { url = "https://files.pythonhosted.org/packages/e5/7b/ef35a71163bf36db06e9c8729608f78dedf032fc8313d19bd4be5c2588f3/pillow-11.1.0-cp313-cp313-win_arm64.whl", hash = "sha256:11633d58b6ee5733bde153a8dafd25e505ea3d32e261accd388827ee987baf65", size = 2375117, upload-time = "2025-01-02T08:12:30.064Z" }, - { url = "https://files.pythonhosted.org/packages/79/30/77f54228401e84d6791354888549b45824ab0ffde659bafa67956303a09f/pillow-11.1.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:70ca5ef3b3b1c4a0812b5c63c57c23b63e53bc38e758b37a951e5bc466449861", size = 3230060, upload-time = "2025-01-02T08:12:32.362Z" }, - { url = "https://files.pythonhosted.org/packages/ce/b1/56723b74b07dd64c1010fee011951ea9c35a43d8020acd03111f14298225/pillow-11.1.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:8000376f139d4d38d6851eb149b321a52bb8893a88dae8ee7d95840431977081", size = 3106192, upload-time = "2025-01-02T08:12:34.361Z" }, - { url = "https://files.pythonhosted.org/packages/e1/cd/7bf7180e08f80a4dcc6b4c3a0aa9e0b0ae57168562726a05dc8aa8fa66b0/pillow-11.1.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9ee85f0696a17dd28fbcfceb59f9510aa71934b483d1f5601d1030c3c8304f3c", size = 4446805, upload-time = "2025-01-02T08:12:36.99Z" }, - { url = "https://files.pythonhosted.org/packages/97/42/87c856ea30c8ed97e8efbe672b58c8304dee0573f8c7cab62ae9e31db6ae/pillow-11.1.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:dd0e081319328928531df7a0e63621caf67652c8464303fd102141b785ef9547", size = 4530623, upload-time = "2025-01-02T08:12:41.912Z" }, - { url = "https://files.pythonhosted.org/packages/ff/41/026879e90c84a88e33fb00cc6bd915ac2743c67e87a18f80270dfe3c2041/pillow-11.1.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:e63e4e5081de46517099dc30abe418122f54531a6ae2ebc8680bcd7096860eab", size = 4465191, upload-time = "2025-01-02T08:12:45.186Z" }, - { url = "https://files.pythonhosted.org/packages/e5/fb/a7960e838bc5df57a2ce23183bfd2290d97c33028b96bde332a9057834d3/pillow-11.1.0-cp313-cp313t-win32.whl", hash = "sha256:dda60aa465b861324e65a78c9f5cf0f4bc713e4309f83bc387be158b077963d9", size = 2295494, upload-time = "2025-01-02T08:12:47.098Z" }, - { url = "https://files.pythonhosted.org/packages/d7/6c/6ec83ee2f6f0fda8d4cf89045c6be4b0373ebfc363ba8538f8c999f63fcd/pillow-11.1.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ad5db5781c774ab9a9b2c4302bbf0c1014960a0a7be63278d13ae6fdf88126fe", size = 2631595, upload-time = "2025-01-02T08:12:50.47Z" }, - { url = "https://files.pythonhosted.org/packages/cf/6c/41c21c6c8af92b9fea313aa47c75de49e2f9a467964ee33eb0135d47eb64/pillow-11.1.0-cp313-cp313t-win_arm64.whl", hash = "sha256:67cd427c68926108778a9005f2a04adbd5e67c442ed21d95389fe1d595458756", size = 2377651, upload-time = "2025-01-02T08:12:53.356Z" }, - { url = "https://files.pythonhosted.org/packages/fa/c5/389961578fb677b8b3244fcd934f720ed25a148b9a5cc81c91bdf59d8588/pillow-11.1.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:8c730dc3a83e5ac137fbc92dfcfe1511ce3b2b5d7578315b63dbbb76f7f51d90", size = 3198345, upload-time = "2025-01-02T08:13:34.091Z" }, - { url = "https://files.pythonhosted.org/packages/c4/fa/803c0e50ffee74d4b965229e816af55276eac1d5806712de86f9371858fd/pillow-11.1.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:7d33d2fae0e8b170b6a6c57400e077412240f6f5bb2a342cf1ee512a787942bb", size = 3072938, upload-time = "2025-01-02T08:13:37.272Z" }, - { url = "https://files.pythonhosted.org/packages/dc/67/2a3a5f8012b5d8c63fe53958ba906c1b1d0482ebed5618057ef4d22f8076/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a8d65b38173085f24bc07f8b6c505cbb7418009fa1a1fcb111b1f4961814a442", size = 3400049, upload-time = "2025-01-02T08:13:41.565Z" }, - { url = "https://files.pythonhosted.org/packages/e5/a0/514f0d317446c98c478d1872497eb92e7cde67003fed74f696441e647446/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:015c6e863faa4779251436db398ae75051469f7c903b043a48f078e437656f83", size = 3422431, upload-time = "2025-01-02T08:13:43.609Z" }, - { url = "https://files.pythonhosted.org/packages/cd/00/20f40a935514037b7d3f87adfc87d2c538430ea625b63b3af8c3f5578e72/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:d44ff19eea13ae4acdaaab0179fa68c0c6f2f45d66a4d8ec1eda7d6cecbcc15f", size = 3446208, upload-time = "2025-01-02T08:13:46.817Z" }, - { url = "https://files.pythonhosted.org/packages/28/3c/7de681727963043e093c72e6c3348411b0185eab3263100d4490234ba2f6/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:d3d8da4a631471dfaf94c10c85f5277b1f8e42ac42bade1ac67da4b4a7359b73", size = 3509746, upload-time = "2025-01-02T08:13:50.6Z" }, - { url = "https://files.pythonhosted.org/packages/41/67/936f9814bdd74b2dfd4822f1f7725ab5d8ff4103919a1664eb4874c58b2f/pillow-11.1.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:4637b88343166249fe8aa94e7c4a62a180c4b3898283bb5d3d2fd5fe10d8e4e0", size = 2626353, upload-time = "2025-01-02T08:13:52.725Z" }, +version = "11.3.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/f3/0d/d0d6dea55cd152ce3d6767bb38a8fc10e33796ba4ba210cbab9354b6d238/pillow-11.3.0.tar.gz", hash = "sha256:3828ee7586cd0b2091b6209e5ad53e20d0649bbe87164a459d0676e035e8f523", size = 47113069, upload-time = "2025-07-01T09:16:30.666Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/4c/5d/45a3553a253ac8763f3561371432a90bdbe6000fbdcf1397ffe502aa206c/pillow-11.3.0-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:1b9c17fd4ace828b3003dfd1e30bff24863e0eb59b535e8f80194d9cc7ecf860", size = 5316554, upload-time = "2025-07-01T09:13:39.342Z" }, + { url = "https://files.pythonhosted.org/packages/7c/c8/67c12ab069ef586a25a4a79ced553586748fad100c77c0ce59bb4983ac98/pillow-11.3.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:65dc69160114cdd0ca0f35cb434633c75e8e7fad4cf855177a05bf38678f73ad", size = 4686548, upload-time = "2025-07-01T09:13:41.835Z" }, + { url = "https://files.pythonhosted.org/packages/2f/bd/6741ebd56263390b382ae4c5de02979af7f8bd9807346d068700dd6d5cf9/pillow-11.3.0-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:7107195ddc914f656c7fc8e4a5e1c25f32e9236ea3ea860f257b0436011fddd0", size = 5859742, upload-time = "2025-07-03T13:09:47.439Z" }, + { url = "https://files.pythonhosted.org/packages/ca/0b/c412a9e27e1e6a829e6ab6c2dca52dd563efbedf4c9c6aa453d9a9b77359/pillow-11.3.0-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:cc3e831b563b3114baac7ec2ee86819eb03caa1a2cef0b481a5675b59c4fe23b", size = 7633087, upload-time = "2025-07-03T13:09:51.796Z" }, + { url = "https://files.pythonhosted.org/packages/59/9d/9b7076aaf30f5dd17e5e5589b2d2f5a5d7e30ff67a171eb686e4eecc2adf/pillow-11.3.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f1f182ebd2303acf8c380a54f615ec883322593320a9b00438eb842c1f37ae50", size = 5963350, upload-time = "2025-07-01T09:13:43.865Z" }, + { url = "https://files.pythonhosted.org/packages/f0/16/1a6bf01fb622fb9cf5c91683823f073f053005c849b1f52ed613afcf8dae/pillow-11.3.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4445fa62e15936a028672fd48c4c11a66d641d2c05726c7ec1f8ba6a572036ae", size = 6631840, upload-time = "2025-07-01T09:13:46.161Z" }, + { url = "https://files.pythonhosted.org/packages/7b/e6/6ff7077077eb47fde78739e7d570bdcd7c10495666b6afcd23ab56b19a43/pillow-11.3.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:71f511f6b3b91dd543282477be45a033e4845a40278fa8dcdbfdb07109bf18f9", size = 6074005, upload-time = "2025-07-01T09:13:47.829Z" }, + { url = "https://files.pythonhosted.org/packages/c3/3a/b13f36832ea6d279a697231658199e0a03cd87ef12048016bdcc84131601/pillow-11.3.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:040a5b691b0713e1f6cbe222e0f4f74cd233421e105850ae3b3c0ceda520f42e", size = 6708372, upload-time = "2025-07-01T09:13:52.145Z" }, + { url = "https://files.pythonhosted.org/packages/6c/e4/61b2e1a7528740efbc70b3d581f33937e38e98ef3d50b05007267a55bcb2/pillow-11.3.0-cp310-cp310-win32.whl", hash = "sha256:89bd777bc6624fe4115e9fac3352c79ed60f3bb18651420635f26e643e3dd1f6", size = 6277090, upload-time = "2025-07-01T09:13:53.915Z" }, + { url = "https://files.pythonhosted.org/packages/a9/d3/60c781c83a785d6afbd6a326ed4d759d141de43aa7365725cbcd65ce5e54/pillow-11.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:19d2ff547c75b8e3ff46f4d9ef969a06c30ab2d4263a9e287733aa8b2429ce8f", size = 6985988, upload-time = "2025-07-01T09:13:55.699Z" }, + { url = "https://files.pythonhosted.org/packages/9f/28/4f4a0203165eefb3763939c6789ba31013a2e90adffb456610f30f613850/pillow-11.3.0-cp310-cp310-win_arm64.whl", hash = "sha256:819931d25e57b513242859ce1876c58c59dc31587847bf74cfe06b2e0cb22d2f", size = 2422899, upload-time = "2025-07-01T09:13:57.497Z" }, + { url = "https://files.pythonhosted.org/packages/db/26/77f8ed17ca4ffd60e1dcd220a6ec6d71210ba398cfa33a13a1cd614c5613/pillow-11.3.0-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:1cd110edf822773368b396281a2293aeb91c90a2db00d78ea43e7e861631b722", size = 5316531, upload-time = "2025-07-01T09:13:59.203Z" }, + { url = "https://files.pythonhosted.org/packages/cb/39/ee475903197ce709322a17a866892efb560f57900d9af2e55f86db51b0a5/pillow-11.3.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:9c412fddd1b77a75aa904615ebaa6001f169b26fd467b4be93aded278266b288", size = 4686560, upload-time = "2025-07-01T09:14:01.101Z" }, + { url = "https://files.pythonhosted.org/packages/d5/90/442068a160fd179938ba55ec8c97050a612426fae5ec0a764e345839f76d/pillow-11.3.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:7d1aa4de119a0ecac0a34a9c8bde33f34022e2e8f99104e47a3ca392fd60e37d", size = 5870978, upload-time = "2025-07-03T13:09:55.638Z" }, + { url = "https://files.pythonhosted.org/packages/13/92/dcdd147ab02daf405387f0218dcf792dc6dd5b14d2573d40b4caeef01059/pillow-11.3.0-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:91da1d88226663594e3f6b4b8c3c8d85bd504117d043740a8e0ec449087cc494", size = 7641168, upload-time = "2025-07-03T13:10:00.37Z" }, + { url = "https://files.pythonhosted.org/packages/6e/db/839d6ba7fd38b51af641aa904e2960e7a5644d60ec754c046b7d2aee00e5/pillow-11.3.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:643f189248837533073c405ec2f0bb250ba54598cf80e8c1e043381a60632f58", size = 5973053, upload-time = "2025-07-01T09:14:04.491Z" }, + { url = "https://files.pythonhosted.org/packages/f2/2f/d7675ecae6c43e9f12aa8d58b6012683b20b6edfbdac7abcb4e6af7a3784/pillow-11.3.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:106064daa23a745510dabce1d84f29137a37224831d88eb4ce94bb187b1d7e5f", size = 6640273, upload-time = "2025-07-01T09:14:06.235Z" }, + { url = "https://files.pythonhosted.org/packages/45/ad/931694675ede172e15b2ff03c8144a0ddaea1d87adb72bb07655eaffb654/pillow-11.3.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cd8ff254faf15591e724dc7c4ddb6bf4793efcbe13802a4ae3e863cd300b493e", size = 6082043, upload-time = "2025-07-01T09:14:07.978Z" }, + { url = "https://files.pythonhosted.org/packages/3a/04/ba8f2b11fc80d2dd462d7abec16351b45ec99cbbaea4387648a44190351a/pillow-11.3.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:932c754c2d51ad2b2271fd01c3d121daaa35e27efae2a616f77bf164bc0b3e94", size = 6715516, upload-time = "2025-07-01T09:14:10.233Z" }, + { url = "https://files.pythonhosted.org/packages/48/59/8cd06d7f3944cc7d892e8533c56b0acb68399f640786313275faec1e3b6f/pillow-11.3.0-cp311-cp311-win32.whl", hash = "sha256:b4b8f3efc8d530a1544e5962bd6b403d5f7fe8b9e08227c6b255f98ad82b4ba0", size = 6274768, upload-time = "2025-07-01T09:14:11.921Z" }, + { url = "https://files.pythonhosted.org/packages/f1/cc/29c0f5d64ab8eae20f3232da8f8571660aa0ab4b8f1331da5c2f5f9a938e/pillow-11.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:1a992e86b0dd7aeb1f053cd506508c0999d710a8f07b4c791c63843fc6a807ac", size = 6986055, upload-time = "2025-07-01T09:14:13.623Z" }, + { url = "https://files.pythonhosted.org/packages/c6/df/90bd886fabd544c25addd63e5ca6932c86f2b701d5da6c7839387a076b4a/pillow-11.3.0-cp311-cp311-win_arm64.whl", hash = "sha256:30807c931ff7c095620fe04448e2c2fc673fcbb1ffe2a7da3fb39613489b1ddd", size = 2423079, upload-time = "2025-07-01T09:14:15.268Z" }, + { url = "https://files.pythonhosted.org/packages/40/fe/1bc9b3ee13f68487a99ac9529968035cca2f0a51ec36892060edcc51d06a/pillow-11.3.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:fdae223722da47b024b867c1ea0be64e0df702c5e0a60e27daad39bf960dd1e4", size = 5278800, upload-time = "2025-07-01T09:14:17.648Z" }, + { url = "https://files.pythonhosted.org/packages/2c/32/7e2ac19b5713657384cec55f89065fb306b06af008cfd87e572035b27119/pillow-11.3.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:921bd305b10e82b4d1f5e802b6850677f965d8394203d182f078873851dada69", size = 4686296, upload-time = "2025-07-01T09:14:19.828Z" }, + { url = "https://files.pythonhosted.org/packages/8e/1e/b9e12bbe6e4c2220effebc09ea0923a07a6da1e1f1bfbc8d7d29a01ce32b/pillow-11.3.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:eb76541cba2f958032d79d143b98a3a6b3ea87f0959bbe256c0b5e416599fd5d", size = 5871726, upload-time = "2025-07-03T13:10:04.448Z" }, + { url = "https://files.pythonhosted.org/packages/8d/33/e9200d2bd7ba00dc3ddb78df1198a6e80d7669cce6c2bdbeb2530a74ec58/pillow-11.3.0-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:67172f2944ebba3d4a7b54f2e95c786a3a50c21b88456329314caaa28cda70f6", size = 7644652, upload-time = "2025-07-03T13:10:10.391Z" }, + { url = "https://files.pythonhosted.org/packages/41/f1/6f2427a26fc683e00d985bc391bdd76d8dd4e92fac33d841127eb8fb2313/pillow-11.3.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:97f07ed9f56a3b9b5f49d3661dc9607484e85c67e27f3e8be2c7d28ca032fec7", size = 5977787, upload-time = "2025-07-01T09:14:21.63Z" }, + { url = "https://files.pythonhosted.org/packages/e4/c9/06dd4a38974e24f932ff5f98ea3c546ce3f8c995d3f0985f8e5ba48bba19/pillow-11.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:676b2815362456b5b3216b4fd5bd89d362100dc6f4945154ff172e206a22c024", size = 6645236, upload-time = "2025-07-01T09:14:23.321Z" }, + { url = "https://files.pythonhosted.org/packages/40/e7/848f69fb79843b3d91241bad658e9c14f39a32f71a301bcd1d139416d1be/pillow-11.3.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:3e184b2f26ff146363dd07bde8b711833d7b0202e27d13540bfe2e35a323a809", size = 6086950, upload-time = "2025-07-01T09:14:25.237Z" }, + { url = "https://files.pythonhosted.org/packages/0b/1a/7cff92e695a2a29ac1958c2a0fe4c0b2393b60aac13b04a4fe2735cad52d/pillow-11.3.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:6be31e3fc9a621e071bc17bb7de63b85cbe0bfae91bb0363c893cbe67247780d", size = 6723358, upload-time = "2025-07-01T09:14:27.053Z" }, + { url = "https://files.pythonhosted.org/packages/26/7d/73699ad77895f69edff76b0f332acc3d497f22f5d75e5360f78cbcaff248/pillow-11.3.0-cp312-cp312-win32.whl", hash = "sha256:7b161756381f0918e05e7cb8a371fff367e807770f8fe92ecb20d905d0e1c149", size = 6275079, upload-time = "2025-07-01T09:14:30.104Z" }, + { url = "https://files.pythonhosted.org/packages/8c/ce/e7dfc873bdd9828f3b6e5c2bbb74e47a98ec23cc5c74fc4e54462f0d9204/pillow-11.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:a6444696fce635783440b7f7a9fc24b3ad10a9ea3f0ab66c5905be1c19ccf17d", size = 6986324, upload-time = "2025-07-01T09:14:31.899Z" }, + { url = "https://files.pythonhosted.org/packages/16/8f/b13447d1bf0b1f7467ce7d86f6e6edf66c0ad7cf44cf5c87a37f9bed9936/pillow-11.3.0-cp312-cp312-win_arm64.whl", hash = "sha256:2aceea54f957dd4448264f9bf40875da0415c83eb85f55069d89c0ed436e3542", size = 2423067, upload-time = "2025-07-01T09:14:33.709Z" }, + { url = "https://files.pythonhosted.org/packages/1e/93/0952f2ed8db3a5a4c7a11f91965d6184ebc8cd7cbb7941a260d5f018cd2d/pillow-11.3.0-cp313-cp313-ios_13_0_arm64_iphoneos.whl", hash = "sha256:1c627742b539bba4309df89171356fcb3cc5a9178355b2727d1b74a6cf155fbd", size = 2128328, upload-time = "2025-07-01T09:14:35.276Z" }, + { url = "https://files.pythonhosted.org/packages/4b/e8/100c3d114b1a0bf4042f27e0f87d2f25e857e838034e98ca98fe7b8c0a9c/pillow-11.3.0-cp313-cp313-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:30b7c02f3899d10f13d7a48163c8969e4e653f8b43416d23d13d1bbfdc93b9f8", size = 2170652, upload-time = "2025-07-01T09:14:37.203Z" }, + { url = "https://files.pythonhosted.org/packages/aa/86/3f758a28a6e381758545f7cdb4942e1cb79abd271bea932998fc0db93cb6/pillow-11.3.0-cp313-cp313-ios_13_0_x86_64_iphonesimulator.whl", hash = "sha256:7859a4cc7c9295f5838015d8cc0a9c215b77e43d07a25e460f35cf516df8626f", size = 2227443, upload-time = "2025-07-01T09:14:39.344Z" }, + { url = "https://files.pythonhosted.org/packages/01/f4/91d5b3ffa718df2f53b0dc109877993e511f4fd055d7e9508682e8aba092/pillow-11.3.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:ec1ee50470b0d050984394423d96325b744d55c701a439d2bd66089bff963d3c", size = 5278474, upload-time = "2025-07-01T09:14:41.843Z" }, + { url = "https://files.pythonhosted.org/packages/f9/0e/37d7d3eca6c879fbd9dba21268427dffda1ab00d4eb05b32923d4fbe3b12/pillow-11.3.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:7db51d222548ccfd274e4572fdbf3e810a5e66b00608862f947b163e613b67dd", size = 4686038, upload-time = "2025-07-01T09:14:44.008Z" }, + { url = "https://files.pythonhosted.org/packages/ff/b0/3426e5c7f6565e752d81221af9d3676fdbb4f352317ceafd42899aaf5d8a/pillow-11.3.0-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2d6fcc902a24ac74495df63faad1884282239265c6839a0a6416d33faedfae7e", size = 5864407, upload-time = "2025-07-03T13:10:15.628Z" }, + { url = "https://files.pythonhosted.org/packages/fc/c1/c6c423134229f2a221ee53f838d4be9d82bab86f7e2f8e75e47b6bf6cd77/pillow-11.3.0-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f0f5d8f4a08090c6d6d578351a2b91acf519a54986c055af27e7a93feae6d3f1", size = 7639094, upload-time = "2025-07-03T13:10:21.857Z" }, + { url = "https://files.pythonhosted.org/packages/ba/c9/09e6746630fe6372c67c648ff9deae52a2bc20897d51fa293571977ceb5d/pillow-11.3.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c37d8ba9411d6003bba9e518db0db0c58a680ab9fe5179f040b0463644bc9805", size = 5973503, upload-time = "2025-07-01T09:14:45.698Z" }, + { url = "https://files.pythonhosted.org/packages/d5/1c/a2a29649c0b1983d3ef57ee87a66487fdeb45132df66ab30dd37f7dbe162/pillow-11.3.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:13f87d581e71d9189ab21fe0efb5a23e9f28552d5be6979e84001d3b8505abe8", size = 6642574, upload-time = "2025-07-01T09:14:47.415Z" }, + { url = "https://files.pythonhosted.org/packages/36/de/d5cc31cc4b055b6c6fd990e3e7f0f8aaf36229a2698501bcb0cdf67c7146/pillow-11.3.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:023f6d2d11784a465f09fd09a34b150ea4672e85fb3d05931d89f373ab14abb2", size = 6084060, upload-time = "2025-07-01T09:14:49.636Z" }, + { url = "https://files.pythonhosted.org/packages/d5/ea/502d938cbaeec836ac28a9b730193716f0114c41325db428e6b280513f09/pillow-11.3.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:45dfc51ac5975b938e9809451c51734124e73b04d0f0ac621649821a63852e7b", size = 6721407, upload-time = "2025-07-01T09:14:51.962Z" }, + { url = "https://files.pythonhosted.org/packages/45/9c/9c5e2a73f125f6cbc59cc7087c8f2d649a7ae453f83bd0362ff7c9e2aee2/pillow-11.3.0-cp313-cp313-win32.whl", hash = "sha256:a4d336baed65d50d37b88ca5b60c0fa9d81e3a87d4a7930d3880d1624d5b31f3", size = 6273841, upload-time = "2025-07-01T09:14:54.142Z" }, + { url = "https://files.pythonhosted.org/packages/23/85/397c73524e0cd212067e0c969aa245b01d50183439550d24d9f55781b776/pillow-11.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:0bce5c4fd0921f99d2e858dc4d4d64193407e1b99478bc5cacecba2311abde51", size = 6978450, upload-time = "2025-07-01T09:14:56.436Z" }, + { url = "https://files.pythonhosted.org/packages/17/d2/622f4547f69cd173955194b78e4d19ca4935a1b0f03a302d655c9f6aae65/pillow-11.3.0-cp313-cp313-win_arm64.whl", hash = "sha256:1904e1264881f682f02b7f8167935cce37bc97db457f8e7849dc3a6a52b99580", size = 2423055, upload-time = "2025-07-01T09:14:58.072Z" }, + { url = "https://files.pythonhosted.org/packages/dd/80/a8a2ac21dda2e82480852978416cfacd439a4b490a501a288ecf4fe2532d/pillow-11.3.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:4c834a3921375c48ee6b9624061076bc0a32a60b5532b322cc0ea64e639dd50e", size = 5281110, upload-time = "2025-07-01T09:14:59.79Z" }, + { url = "https://files.pythonhosted.org/packages/44/d6/b79754ca790f315918732e18f82a8146d33bcd7f4494380457ea89eb883d/pillow-11.3.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:5e05688ccef30ea69b9317a9ead994b93975104a677a36a8ed8106be9260aa6d", size = 4689547, upload-time = "2025-07-01T09:15:01.648Z" }, + { url = "https://files.pythonhosted.org/packages/49/20/716b8717d331150cb00f7fdd78169c01e8e0c219732a78b0e59b6bdb2fd6/pillow-11.3.0-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:1019b04af07fc0163e2810167918cb5add8d74674b6267616021ab558dc98ced", size = 5901554, upload-time = "2025-07-03T13:10:27.018Z" }, + { url = "https://files.pythonhosted.org/packages/74/cf/a9f3a2514a65bb071075063a96f0a5cf949c2f2fce683c15ccc83b1c1cab/pillow-11.3.0-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f944255db153ebb2b19c51fe85dd99ef0ce494123f21b9db4877ffdfc5590c7c", size = 7669132, upload-time = "2025-07-03T13:10:33.01Z" }, + { url = "https://files.pythonhosted.org/packages/98/3c/da78805cbdbee9cb43efe8261dd7cc0b4b93f2ac79b676c03159e9db2187/pillow-11.3.0-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1f85acb69adf2aaee8b7da124efebbdb959a104db34d3a2cb0f3793dbae422a8", size = 6005001, upload-time = "2025-07-01T09:15:03.365Z" }, + { url = "https://files.pythonhosted.org/packages/6c/fa/ce044b91faecf30e635321351bba32bab5a7e034c60187fe9698191aef4f/pillow-11.3.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:05f6ecbeff5005399bb48d198f098a9b4b6bdf27b8487c7f38ca16eeb070cd59", size = 6668814, upload-time = "2025-07-01T09:15:05.655Z" }, + { url = "https://files.pythonhosted.org/packages/7b/51/90f9291406d09bf93686434f9183aba27b831c10c87746ff49f127ee80cb/pillow-11.3.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:a7bc6e6fd0395bc052f16b1a8670859964dbd7003bd0af2ff08342eb6e442cfe", size = 6113124, upload-time = "2025-07-01T09:15:07.358Z" }, + { url = "https://files.pythonhosted.org/packages/cd/5a/6fec59b1dfb619234f7636d4157d11fb4e196caeee220232a8d2ec48488d/pillow-11.3.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:83e1b0161c9d148125083a35c1c5a89db5b7054834fd4387499e06552035236c", size = 6747186, upload-time = "2025-07-01T09:15:09.317Z" }, + { url = "https://files.pythonhosted.org/packages/49/6b/00187a044f98255225f172de653941e61da37104a9ea60e4f6887717e2b5/pillow-11.3.0-cp313-cp313t-win32.whl", hash = "sha256:2a3117c06b8fb646639dce83694f2f9eac405472713fcb1ae887469c0d4f6788", size = 6277546, upload-time = "2025-07-01T09:15:11.311Z" }, + { url = "https://files.pythonhosted.org/packages/e8/5c/6caaba7e261c0d75bab23be79f1d06b5ad2a2ae49f028ccec801b0e853d6/pillow-11.3.0-cp313-cp313t-win_amd64.whl", hash = "sha256:857844335c95bea93fb39e0fa2726b4d9d758850b34075a7e3ff4f4fa3aa3b31", size = 6985102, upload-time = "2025-07-01T09:15:13.164Z" }, + { url = "https://files.pythonhosted.org/packages/f3/7e/b623008460c09a0cb38263c93b828c666493caee2eb34ff67f778b87e58c/pillow-11.3.0-cp313-cp313t-win_arm64.whl", hash = "sha256:8797edc41f3e8536ae4b10897ee2f637235c94f27404cac7297f7b607dd0716e", size = 2424803, upload-time = "2025-07-01T09:15:15.695Z" }, + { url = "https://files.pythonhosted.org/packages/73/f4/04905af42837292ed86cb1b1dabe03dce1edc008ef14c473c5c7e1443c5d/pillow-11.3.0-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:d9da3df5f9ea2a89b81bb6087177fb1f4d1c7146d583a3fe5c672c0d94e55e12", size = 5278520, upload-time = "2025-07-01T09:15:17.429Z" }, + { url = "https://files.pythonhosted.org/packages/41/b0/33d79e377a336247df6348a54e6d2a2b85d644ca202555e3faa0cf811ecc/pillow-11.3.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:0b275ff9b04df7b640c59ec5a3cb113eefd3795a8df80bac69646ef699c6981a", size = 4686116, upload-time = "2025-07-01T09:15:19.423Z" }, + { url = "https://files.pythonhosted.org/packages/49/2d/ed8bc0ab219ae8768f529597d9509d184fe8a6c4741a6864fea334d25f3f/pillow-11.3.0-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:0743841cabd3dba6a83f38a92672cccbd69af56e3e91777b0ee7f4dba4385632", size = 5864597, upload-time = "2025-07-03T13:10:38.404Z" }, + { url = "https://files.pythonhosted.org/packages/b5/3d/b932bb4225c80b58dfadaca9d42d08d0b7064d2d1791b6a237f87f661834/pillow-11.3.0-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:2465a69cf967b8b49ee1b96d76718cd98c4e925414ead59fdf75cf0fd07df673", size = 7638246, upload-time = "2025-07-03T13:10:44.987Z" }, + { url = "https://files.pythonhosted.org/packages/09/b5/0487044b7c096f1b48f0d7ad416472c02e0e4bf6919541b111efd3cae690/pillow-11.3.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:41742638139424703b4d01665b807c6468e23e699e8e90cffefe291c5832b027", size = 5973336, upload-time = "2025-07-01T09:15:21.237Z" }, + { url = "https://files.pythonhosted.org/packages/a8/2d/524f9318f6cbfcc79fbc004801ea6b607ec3f843977652fdee4857a7568b/pillow-11.3.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:93efb0b4de7e340d99057415c749175e24c8864302369e05914682ba642e5d77", size = 6642699, upload-time = "2025-07-01T09:15:23.186Z" }, + { url = "https://files.pythonhosted.org/packages/6f/d2/a9a4f280c6aefedce1e8f615baaa5474e0701d86dd6f1dede66726462bbd/pillow-11.3.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:7966e38dcd0fa11ca390aed7c6f20454443581d758242023cf36fcb319b1a874", size = 6083789, upload-time = "2025-07-01T09:15:25.1Z" }, + { url = "https://files.pythonhosted.org/packages/fe/54/86b0cd9dbb683a9d5e960b66c7379e821a19be4ac5810e2e5a715c09a0c0/pillow-11.3.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:98a9afa7b9007c67ed84c57c9e0ad86a6000da96eaa638e4f8abe5b65ff83f0a", size = 6720386, upload-time = "2025-07-01T09:15:27.378Z" }, + { url = "https://files.pythonhosted.org/packages/e7/95/88efcaf384c3588e24259c4203b909cbe3e3c2d887af9e938c2022c9dd48/pillow-11.3.0-cp314-cp314-win32.whl", hash = "sha256:02a723e6bf909e7cea0dac1b0e0310be9d7650cd66222a5f1c571455c0a45214", size = 6370911, upload-time = "2025-07-01T09:15:29.294Z" }, + { url = "https://files.pythonhosted.org/packages/2e/cc/934e5820850ec5eb107e7b1a72dd278140731c669f396110ebc326f2a503/pillow-11.3.0-cp314-cp314-win_amd64.whl", hash = "sha256:a418486160228f64dd9e9efcd132679b7a02a5f22c982c78b6fc7dab3fefb635", size = 7117383, upload-time = "2025-07-01T09:15:31.128Z" }, + { url = "https://files.pythonhosted.org/packages/d6/e9/9c0a616a71da2a5d163aa37405e8aced9a906d574b4a214bede134e731bc/pillow-11.3.0-cp314-cp314-win_arm64.whl", hash = "sha256:155658efb5e044669c08896c0c44231c5e9abcaadbc5cd3648df2f7c0b96b9a6", size = 2511385, upload-time = "2025-07-01T09:15:33.328Z" }, + { url = "https://files.pythonhosted.org/packages/1a/33/c88376898aff369658b225262cd4f2659b13e8178e7534df9e6e1fa289f6/pillow-11.3.0-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:59a03cdf019efbfeeed910bf79c7c93255c3d54bc45898ac2a4140071b02b4ae", size = 5281129, upload-time = "2025-07-01T09:15:35.194Z" }, + { url = "https://files.pythonhosted.org/packages/1f/70/d376247fb36f1844b42910911c83a02d5544ebd2a8bad9efcc0f707ea774/pillow-11.3.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:f8a5827f84d973d8636e9dc5764af4f0cf2318d26744b3d902931701b0d46653", size = 4689580, upload-time = "2025-07-01T09:15:37.114Z" }, + { url = "https://files.pythonhosted.org/packages/eb/1c/537e930496149fbac69efd2fc4329035bbe2e5475b4165439e3be9cb183b/pillow-11.3.0-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:ee92f2fd10f4adc4b43d07ec5e779932b4eb3dbfbc34790ada5a6669bc095aa6", size = 5902860, upload-time = "2025-07-03T13:10:50.248Z" }, + { url = "https://files.pythonhosted.org/packages/bd/57/80f53264954dcefeebcf9dae6e3eb1daea1b488f0be8b8fef12f79a3eb10/pillow-11.3.0-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:c96d333dcf42d01f47b37e0979b6bd73ec91eae18614864622d9b87bbd5bbf36", size = 7670694, upload-time = "2025-07-03T13:10:56.432Z" }, + { url = "https://files.pythonhosted.org/packages/70/ff/4727d3b71a8578b4587d9c276e90efad2d6fe0335fd76742a6da08132e8c/pillow-11.3.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4c96f993ab8c98460cd0c001447bff6194403e8b1d7e149ade5f00594918128b", size = 6005888, upload-time = "2025-07-01T09:15:39.436Z" }, + { url = "https://files.pythonhosted.org/packages/05/ae/716592277934f85d3be51d7256f3636672d7b1abfafdc42cf3f8cbd4b4c8/pillow-11.3.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:41342b64afeba938edb034d122b2dda5db2139b9a4af999729ba8818e0056477", size = 6670330, upload-time = "2025-07-01T09:15:41.269Z" }, + { url = "https://files.pythonhosted.org/packages/e7/bb/7fe6cddcc8827b01b1a9766f5fdeb7418680744f9082035bdbabecf1d57f/pillow-11.3.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:068d9c39a2d1b358eb9f245ce7ab1b5c3246c7c8c7d9ba58cfa5b43146c06e50", size = 6114089, upload-time = "2025-07-01T09:15:43.13Z" }, + { url = "https://files.pythonhosted.org/packages/8b/f5/06bfaa444c8e80f1a8e4bff98da9c83b37b5be3b1deaa43d27a0db37ef84/pillow-11.3.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:a1bc6ba083b145187f648b667e05a2534ecc4b9f2784c2cbe3089e44868f2b9b", size = 6748206, upload-time = "2025-07-01T09:15:44.937Z" }, + { url = "https://files.pythonhosted.org/packages/f0/77/bc6f92a3e8e6e46c0ca78abfffec0037845800ea38c73483760362804c41/pillow-11.3.0-cp314-cp314t-win32.whl", hash = "sha256:118ca10c0d60b06d006be10a501fd6bbdfef559251ed31b794668ed569c87e12", size = 6377370, upload-time = "2025-07-01T09:15:46.673Z" }, + { url = "https://files.pythonhosted.org/packages/4a/82/3a721f7d69dca802befb8af08b7c79ebcab461007ce1c18bd91a5d5896f9/pillow-11.3.0-cp314-cp314t-win_amd64.whl", hash = "sha256:8924748b688aa210d79883357d102cd64690e56b923a186f35a82cbc10f997db", size = 7121500, upload-time = "2025-07-01T09:15:48.512Z" }, + { url = "https://files.pythonhosted.org/packages/89/c7/5572fa4a3f45740eaab6ae86fcdf7195b55beac1371ac8c619d880cfe948/pillow-11.3.0-cp314-cp314t-win_arm64.whl", hash = "sha256:79ea0d14d3ebad43ec77ad5272e6ff9bba5b679ef73375ea760261207fa8e0aa", size = 2512835, upload-time = "2025-07-01T09:15:50.399Z" }, + { url = "https://files.pythonhosted.org/packages/6f/8b/209bd6b62ce8367f47e68a218bffac88888fdf2c9fcf1ecadc6c3ec1ebc7/pillow-11.3.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:3cee80663f29e3843b68199b9d6f4f54bd1d4a6b59bdd91bceefc51238bcb967", size = 5270556, upload-time = "2025-07-01T09:16:09.961Z" }, + { url = "https://files.pythonhosted.org/packages/2e/e6/231a0b76070c2cfd9e260a7a5b504fb72da0a95279410fa7afd99d9751d6/pillow-11.3.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:b5f56c3f344f2ccaf0dd875d3e180f631dc60a51b314295a3e681fe8cf851fbe", size = 4654625, upload-time = "2025-07-01T09:16:11.913Z" }, + { url = "https://files.pythonhosted.org/packages/13/f4/10cf94fda33cb12765f2397fc285fa6d8eb9c29de7f3185165b702fc7386/pillow-11.3.0-pp310-pypy310_pp73-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e67d793d180c9df62f1f40aee3accca4829d3794c95098887edc18af4b8b780c", size = 4874207, upload-time = "2025-07-03T13:11:10.201Z" }, + { url = "https://files.pythonhosted.org/packages/72/c9/583821097dc691880c92892e8e2d41fe0a5a3d6021f4963371d2f6d57250/pillow-11.3.0-pp310-pypy310_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:d000f46e2917c705e9fb93a3606ee4a819d1e3aa7a9b442f6444f07e77cf5e25", size = 6583939, upload-time = "2025-07-03T13:11:15.68Z" }, + { url = "https://files.pythonhosted.org/packages/3b/8e/5c9d410f9217b12320efc7c413e72693f48468979a013ad17fd690397b9a/pillow-11.3.0-pp310-pypy310_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:527b37216b6ac3a12d7838dc3bd75208ec57c1c6d11ef01902266a5a0c14fc27", size = 4957166, upload-time = "2025-07-01T09:16:13.74Z" }, + { url = "https://files.pythonhosted.org/packages/62/bb/78347dbe13219991877ffb3a91bf09da8317fbfcd4b5f9140aeae020ad71/pillow-11.3.0-pp310-pypy310_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:be5463ac478b623b9dd3937afd7fb7ab3d79dd290a28e2b6df292dc75063eb8a", size = 5581482, upload-time = "2025-07-01T09:16:16.107Z" }, + { url = "https://files.pythonhosted.org/packages/d9/28/1000353d5e61498aaeaaf7f1e4b49ddb05f2c6575f9d4f9f914a3538b6e1/pillow-11.3.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:8dc70ca24c110503e16918a658b869019126ecfe03109b754c402daff12b3d9f", size = 6984596, upload-time = "2025-07-01T09:16:18.07Z" }, + { url = "https://files.pythonhosted.org/packages/9e/e3/6fa84033758276fb31da12e5fb66ad747ae83b93c67af17f8c6ff4cc8f34/pillow-11.3.0-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7c8ec7a017ad1bd562f93dbd8505763e688d388cde6e4a010ae1486916e713e6", size = 5270566, upload-time = "2025-07-01T09:16:19.801Z" }, + { url = "https://files.pythonhosted.org/packages/5b/ee/e8d2e1ab4892970b561e1ba96cbd59c0d28cf66737fc44abb2aec3795a4e/pillow-11.3.0-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:9ab6ae226de48019caa8074894544af5b53a117ccb9d3b3dcb2871464c829438", size = 4654618, upload-time = "2025-07-01T09:16:21.818Z" }, + { url = "https://files.pythonhosted.org/packages/f2/6d/17f80f4e1f0761f02160fc433abd4109fa1548dcfdca46cfdadaf9efa565/pillow-11.3.0-pp311-pypy311_pp73-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:fe27fb049cdcca11f11a7bfda64043c37b30e6b91f10cb5bab275806c32f6ab3", size = 4874248, upload-time = "2025-07-03T13:11:20.738Z" }, + { url = "https://files.pythonhosted.org/packages/de/5f/c22340acd61cef960130585bbe2120e2fd8434c214802f07e8c03596b17e/pillow-11.3.0-pp311-pypy311_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:465b9e8844e3c3519a983d58b80be3f668e2a7a5db97f2784e7079fbc9f9822c", size = 6583963, upload-time = "2025-07-03T13:11:26.283Z" }, + { url = "https://files.pythonhosted.org/packages/31/5e/03966aedfbfcbb4d5f8aa042452d3361f325b963ebbadddac05b122e47dd/pillow-11.3.0-pp311-pypy311_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5418b53c0d59b3824d05e029669efa023bbef0f3e92e75ec8428f3799487f361", size = 4957170, upload-time = "2025-07-01T09:16:23.762Z" }, + { url = "https://files.pythonhosted.org/packages/cc/2d/e082982aacc927fc2cab48e1e731bdb1643a1406acace8bed0900a61464e/pillow-11.3.0-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:504b6f59505f08ae014f724b6207ff6222662aab5cc9542577fb084ed0676ac7", size = 5581505, upload-time = "2025-07-01T09:16:25.593Z" }, + { url = "https://files.pythonhosted.org/packages/34/e7/ae39f538fd6844e982063c3a5e4598b8ced43b9633baa3a85ef33af8c05c/pillow-11.3.0-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:c84d689db21a1c397d001aa08241044aa2069e7587b398c8cc63020390b1c1b8", size = 6984598, upload-time = "2025-07-01T09:16:27.732Z" }, ] [[package]] name = "platformdirs" -version = "4.3.7" +version = "4.3.8" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/b6/2d/7d512a3913d60623e7eb945c6d1b4f0bddf1d0b7ada5225274c87e5b53d1/platformdirs-4.3.7.tar.gz", hash = "sha256:eb437d586b6a0986388f0d6f74aa0cde27b48d0e3d66843640bfb6bdcdb6e351", size = 21291, upload-time = "2025-03-19T20:36:10.989Z" } +sdist = { url = "https://files.pythonhosted.org/packages/fe/8b/3c73abc9c759ecd3f1f7ceff6685840859e8070c4d947c93fae71f6a0bf2/platformdirs-4.3.8.tar.gz", hash = "sha256:3d512d96e16bcb959a814c9f348431070822a6496326a4be0911c40b5a74c2bc", size = 21362, upload-time = "2025-05-07T22:47:42.121Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/6d/45/59578566b3275b8fd9157885918fcd0c4d74162928a5310926887b856a51/platformdirs-4.3.7-py3-none-any.whl", hash = "sha256:a03875334331946f13c549dbd8f4bac7a13a50a895a0eb1e8c6a8ace80d40a94", size = 18499, upload-time = "2025-03-19T20:36:09.038Z" }, + { url = "https://files.pythonhosted.org/packages/fe/39/979e8e21520d4e47a0bbe349e2713c0aac6f3d853d0e5b34d76206c439aa/platformdirs-4.3.8-py3-none-any.whl", hash = "sha256:ff7059bb7eb1179e2685604f4aaf157cfd9535242bd23742eadc3c13542139b4", size = 18567, upload-time = "2025-05-07T22:47:40.376Z" }, ] [[package]] name = "plotly" -version = "6.0.1" +version = "6.2.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "narwhals" }, { name = "packaging" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/c7/cc/e41b5f697ae403f0b50e47b7af2e36642a193085f553bf7cc1169362873a/plotly-6.0.1.tar.gz", hash = "sha256:dd8400229872b6e3c964b099be699f8d00c489a974f2cfccfad5e8240873366b", size = 8094643, upload-time = "2025-03-17T15:02:23.994Z" } +sdist = { url = "https://files.pythonhosted.org/packages/6e/5c/0efc297df362b88b74957a230af61cd6929f531f72f48063e8408702ffba/plotly-6.2.0.tar.gz", hash = "sha256:9dfa23c328000f16c928beb68927444c1ab9eae837d1fe648dbcda5360c7953d", size = 6801941, upload-time = "2025-06-26T16:20:45.765Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/02/65/ad2bc85f7377f5cfba5d4466d5474423a3fb7f6a97fd807c06f92dd3e721/plotly-6.0.1-py3-none-any.whl", hash = "sha256:4714db20fea57a435692c548a4eb4fae454f7daddf15f8d8ba7e1045681d7768", size = 14805757, upload-time = "2025-03-17T15:02:18.73Z" }, + { url = "https://files.pythonhosted.org/packages/ed/20/f2b7ac96a91cc5f70d81320adad24cc41bf52013508d649b1481db225780/plotly-6.2.0-py3-none-any.whl", hash = "sha256:32c444d4c940887219cb80738317040363deefdfee4f354498cc0b6dab8978bd", size = 9635469, upload-time = "2025-06-26T16:20:40.76Z" }, ] [[package]] name = "pluggy" -version = "1.5.0" +version = "1.6.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/96/2d/02d4312c973c6050a18b314a5ad0b3210edb65a906f868e31c111dede4a6/pluggy-1.5.0.tar.gz", hash = "sha256:2cffa88e94fdc978c4c574f15f9e59b7f4201d439195c3715ca9e2486f1d0cf1", size = 67955, upload-time = "2024-04-20T21:34:42.531Z" } +sdist = { url = "https://files.pythonhosted.org/packages/f9/e2/3e91f31a7d2b083fe6ef3fa267035b518369d9511ffab804f839851d2779/pluggy-1.6.0.tar.gz", hash = "sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3", size = 69412, upload-time = "2025-05-15T12:30:07.975Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/88/5f/e351af9a41f866ac3f1fac4ca0613908d9a41741cfcf2228f4ad853b697d/pluggy-1.5.0-py3-none-any.whl", hash = "sha256:44e1ad92c8ca002de6377e165f3e0f1be63266ab4d554740532335b9d75ea669", size = 20556, upload-time = "2024-04-20T21:34:40.434Z" }, + { url = "https://files.pythonhosted.org/packages/54/20/4d324d65cc6d9205fabedc306948156824eb9f0ee1633355a8f7ec5c66bf/pluggy-1.6.0-py3-none-any.whl", hash = "sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746", size = 20538, upload-time = "2025-05-15T12:30:06.134Z" }, ] [[package]] name = "posebusters" -version = "0.3.6" +version = "0.4.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "pandas" }, { name = "pyyaml" }, { name = "rdkit" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/a4/40/44c4e544798d92afc30d43af869dc0eca555768793f85bf492e632a8df39/posebusters-0.3.6.tar.gz", hash = "sha256:d60cbc0134431a31dfc11d81dacead8b731bcbfe2be10993878649d34462f1a2", size = 4991179, upload-time = "2025-03-22T00:24:42.385Z" } +sdist = { url = "https://files.pythonhosted.org/packages/f9/e1/7646585f7620c7689c8542eceb93a09567469b5230cfdf1231d38a948eff/posebusters-0.4.5.tar.gz", hash = "sha256:d90f22d32f7ce3551d28d23e899059070211c91d7104ae4fbc7229b4696b9146", size = 5022845, upload-time = "2025-07-13T15:16:21.846Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/1f/7b/6f549f8aca269c2aafcd4a0ba453a84df027b81d3f860addf67a32ddb5f1/posebusters-0.3.6-py3-none-any.whl", hash = "sha256:4d33780765abd2ac3c27edcf2963a6322530343034ca18898c64b1a3b45c8e21", size = 554304, upload-time = "2025-03-22T00:24:40.989Z" }, + { url = "https://files.pythonhosted.org/packages/b6/03/6989d7d9d65aa19bcc91ea3f78d791141d4be0957a67f90fe82e6495832a/posebusters-0.4.5-py3-none-any.whl", hash = "sha256:ec8f2b1634b95000efe6c9da0392eefdc681dda07646ef0ed1f4a5efca099ab9", size = 556376, upload-time = "2025-07-13T15:16:20.308Z" }, ] [[package]] @@ -2060,8 +2665,10 @@ name = "pot" version = "0.9.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, - { name = "scipy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c1/40/3e0c8dd88328d944f9d82b30cafd2a1c911bddff0b8bccc8dc9dd5e45b7c/pot-0.9.5.tar.gz", hash = "sha256:9644ee7ff51c3cffa3c2632b9dd9dff4f3520266f9fb771450935ffb646d6042", size = 440808, upload-time = "2024-11-07T10:05:05.567Z" } wheels = [ @@ -2106,117 +2713,129 @@ wheels = [ [[package]] name = "prompt-toolkit" -version = "3.0.50" +version = "3.0.51" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "wcwidth" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/a1/e1/bd15cb8ffdcfeeb2bdc215de3c3cffca11408d829e4b8416dcfe71ba8854/prompt_toolkit-3.0.50.tar.gz", hash = "sha256:544748f3860a2623ca5cd6d2795e7a14f3d0e1c3c9728359013f79877fc89bab", size = 429087, upload-time = "2025-01-20T15:55:35.072Z" } +sdist = { url = "https://files.pythonhosted.org/packages/bb/6e/9d084c929dfe9e3bfe0c6a47e31f78a25c54627d64a66e884a8bf5474f1c/prompt_toolkit-3.0.51.tar.gz", hash = "sha256:931a162e3b27fc90c86f1b48bb1fb2c528c2761475e57c9c06de13311c7b54ed", size = 428940, upload-time = "2025-04-15T09:18:47.731Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e4/ea/d836f008d33151c7a1f62caf3d8dd782e4d15f6a43897f64480c2b8de2ad/prompt_toolkit-3.0.50-py3-none-any.whl", hash = "sha256:9b6427eb19e479d98acff65196a307c555eb567989e6d88ebbb1b509d9779198", size = 387816, upload-time = "2025-01-20T15:55:29.98Z" }, + { url = "https://files.pythonhosted.org/packages/ce/4f/5249960887b1fbe561d9ff265496d170b55a735b76724f10ef19f9e40716/prompt_toolkit-3.0.51-py3-none-any.whl", hash = "sha256:52742911fde84e2d423e2f9a4cf1de7d7ac4e51958f648d9540e0fb8db077b07", size = 387810, upload-time = "2025-04-15T09:18:44.753Z" }, ] [[package]] name = "propcache" -version = "0.3.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/92/76/f941e63d55c0293ff7829dd21e7cf1147e90a526756869a9070f287a68c9/propcache-0.3.0.tar.gz", hash = "sha256:a8fd93de4e1d278046345f49e2238cdb298589325849b2645d4a94c53faeffc5", size = 42722, upload-time = "2025-02-20T19:03:29.191Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/8d/f0/dc9ec44d2e63c13f816a16398c039329736712440ff82b682dd9a78d2258/propcache-0.3.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:efa44f64c37cc30c9f05932c740a8b40ce359f51882c70883cc95feac842da4d", size = 79574, upload-time = "2025-02-20T18:59:44.353Z" }, - { url = "https://files.pythonhosted.org/packages/99/3a/33a207dfcb3ee1131ea23a2aeb726c3c4994f89546d7eadf8c50627c8b63/propcache-0.3.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2383a17385d9800b6eb5855c2f05ee550f803878f344f58b6e194de08b96352c", size = 45898, upload-time = "2025-02-20T18:59:46.783Z" }, - { url = "https://files.pythonhosted.org/packages/af/68/0bde765c9f5dc02b4466d2838600af38c81b184c26c6d3cd44643ac668e3/propcache-0.3.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:d3e7420211f5a65a54675fd860ea04173cde60a7cc20ccfbafcccd155225f8bc", size = 45418, upload-time = "2025-02-20T18:59:49.082Z" }, - { url = "https://files.pythonhosted.org/packages/06/a6/c682669bae41199358e16cc7b1c818f91c5f9e925cc863dabd98ce32716a/propcache-0.3.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3302c5287e504d23bb0e64d2a921d1eb4a03fb93a0a0aa3b53de059f5a5d737d", size = 205116, upload-time = "2025-02-20T18:59:50.606Z" }, - { url = "https://files.pythonhosted.org/packages/fb/ae/82cfb50267d9a1baa0340728eb9e32245a68538fef929d7bb786d01c11a8/propcache-0.3.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:7e2e068a83552ddf7a39a99488bcba05ac13454fb205c847674da0352602082f", size = 219405, upload-time = "2025-02-20T18:59:54.016Z" }, - { url = "https://files.pythonhosted.org/packages/ab/16/7b6b2bf8c207cfd0e5ca3d41aea397392de9899867ec024f88c94f9ae2ab/propcache-0.3.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2d913d36bdaf368637b4f88d554fb9cb9d53d6920b9c5563846555938d5450bf", size = 217656, upload-time = "2025-02-20T18:59:55.747Z" }, - { url = "https://files.pythonhosted.org/packages/f4/eb/41447de61eb5454891658d0fb9b1d7d35d49a4a5dd2e0c86f2c332e8b7e1/propcache-0.3.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8ee1983728964d6070ab443399c476de93d5d741f71e8f6e7880a065f878e0b9", size = 205414, upload-time = "2025-02-20T18:59:59.907Z" }, - { url = "https://files.pythonhosted.org/packages/03/b6/9719878f8b5b20d37ee663a40f8dcbf888559e4d3be2ba2fe5c790fc28d2/propcache-0.3.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:36ca5e9a21822cc1746023e88f5c0af6fce3af3b85d4520efb1ce4221bed75cc", size = 195746, upload-time = "2025-02-20T19:00:03.124Z" }, - { url = "https://files.pythonhosted.org/packages/bb/ec/b79c3210ba459800d1a8f1afeb81d7b503893555a7b79c24082ff26d3314/propcache-0.3.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:9ecde3671e62eeb99e977f5221abcf40c208f69b5eb986b061ccec317c82ebd0", size = 198651, upload-time = "2025-02-20T19:00:04.747Z" }, - { url = "https://files.pythonhosted.org/packages/48/f6/2b0140bc47013e43575973068e72ad51ee9f22f2dad42e6d6e362d715125/propcache-0.3.0-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:d383bf5e045d7f9d239b38e6acadd7b7fdf6c0087259a84ae3475d18e9a2ae8b", size = 195858, upload-time = "2025-02-20T19:00:06.723Z" }, - { url = "https://files.pythonhosted.org/packages/97/3d/2fa19303d87aa21f9a42dcd870d6088a2a776ff5518e394d50412c3679a6/propcache-0.3.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:8cb625bcb5add899cb8ba7bf716ec1d3e8f7cdea9b0713fa99eadf73b6d4986f", size = 197181, upload-time = "2025-02-20T19:00:08.31Z" }, - { url = "https://files.pythonhosted.org/packages/09/f3/a2170ffc9fa774c1dfd52294113c0fa6cdc5b71dbfd7129bb9378fdd8b42/propcache-0.3.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:5fa159dcee5dba00c1def3231c249cf261185189205073bde13797e57dd7540a", size = 207411, upload-time = "2025-02-20T19:00:10.546Z" }, - { url = "https://files.pythonhosted.org/packages/d6/1e/cb8a6c82178efffa0b00dc463f36cd086f747345585140aeb95d5cb93666/propcache-0.3.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:a7080b0159ce05f179cfac592cda1a82898ca9cd097dacf8ea20ae33474fbb25", size = 210724, upload-time = "2025-02-20T19:00:12.207Z" }, - { url = "https://files.pythonhosted.org/packages/2b/72/6e273543337a3e22cf462eb836f065a9830b4d41baeb1f58db2695c934f3/propcache-0.3.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:ed7161bccab7696a473fe7ddb619c1d75963732b37da4618ba12e60899fefe4f", size = 203511, upload-time = "2025-02-20T19:00:14.689Z" }, - { url = "https://files.pythonhosted.org/packages/f3/ea/7412c79bcec06597c967d49789f5a1f7fd76a8654908feeaefafb7447c9a/propcache-0.3.0-cp310-cp310-win32.whl", hash = "sha256:bf0d9a171908f32d54f651648c7290397b8792f4303821c42a74e7805bfb813c", size = 40600, upload-time = "2025-02-20T19:00:16.423Z" }, - { url = "https://files.pythonhosted.org/packages/a3/42/488c90190491f3e61bd2c2fb0b3d91c1c78778270dde2f0b6633fc9ff723/propcache-0.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:42924dc0c9d73e49908e35bbdec87adedd651ea24c53c29cac103ede0ea1d340", size = 44714, upload-time = "2025-02-20T19:00:18.709Z" }, - { url = "https://files.pythonhosted.org/packages/45/c9/cf09ff7e6d09f14149094f7cd50d2dec032b24e61af21fc4540da2b17bfb/propcache-0.3.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:9ddd49258610499aab83b4f5b61b32e11fce873586282a0e972e5ab3bcadee51", size = 79568, upload-time = "2025-02-20T19:00:21.457Z" }, - { url = "https://files.pythonhosted.org/packages/c8/32/2424d89da88cd81b7d148e0d2b3131461b570a02aa9d84a2e567509adb0d/propcache-0.3.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2578541776769b500bada3f8a4eeaf944530516b6e90c089aa368266ed70c49e", size = 45895, upload-time = "2025-02-20T19:00:23.035Z" }, - { url = "https://files.pythonhosted.org/packages/f6/91/ee5b6aa7aa31754fefcf0c5180e09223cac380ef195c4ddc8c266eb641ea/propcache-0.3.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:d8074c5dd61c8a3e915fa8fc04754fa55cfa5978200d2daa1e2d4294c1f136aa", size = 45427, upload-time = "2025-02-20T19:00:25.07Z" }, - { url = "https://files.pythonhosted.org/packages/bf/73/38f0128462b8b616181d8c53bd5d04eac41c50c449b07615c65d56ba0a9b/propcache-0.3.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b58229a844931bca61b3a20efd2be2a2acb4ad1622fc026504309a6883686fbf", size = 232427, upload-time = "2025-02-20T19:00:26.587Z" }, - { url = "https://files.pythonhosted.org/packages/59/82/f3d4e84f4539dcfc9c3d338282b9e915f5b63c921986ecfdf7af2d12f87c/propcache-0.3.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e45377d5d6fefe1677da2a2c07b024a6dac782088e37c0b1efea4cfe2b1be19b", size = 239985, upload-time = "2025-02-20T19:00:28.204Z" }, - { url = "https://files.pythonhosted.org/packages/42/e8/029f58cccbae83c9969a7ee7a06558d5b83a93dfc54e0f4f70234bbaea1b/propcache-0.3.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ec5060592d83454e8063e487696ac3783cc48c9a329498bafae0d972bc7816c9", size = 238827, upload-time = "2025-02-20T19:00:30.147Z" }, - { url = "https://files.pythonhosted.org/packages/8b/a2/c373561777c0cb9b9e7b9b9a10b9b3a7b6bde75a2535b962231cecc8fdb8/propcache-0.3.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:15010f29fbed80e711db272909a074dc79858c6d28e2915704cfc487a8ac89c6", size = 231348, upload-time = "2025-02-20T19:00:32.05Z" }, - { url = "https://files.pythonhosted.org/packages/d7/d2/4673f715beedf6038b485bcd976813149231d9df5bb6196cb69a09c185c9/propcache-0.3.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a254537b9b696ede293bfdbc0a65200e8e4507bc9f37831e2a0318a9b333c85c", size = 220426, upload-time = "2025-02-20T19:00:34.756Z" }, - { url = "https://files.pythonhosted.org/packages/e0/f6/1da65f900927bafd4675a16e890618ec7643f2f922bf0e4d84bb38645618/propcache-0.3.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:2b975528998de037dfbc10144b8aed9b8dd5a99ec547f14d1cb7c5665a43f075", size = 220294, upload-time = "2025-02-20T19:00:38.63Z" }, - { url = "https://files.pythonhosted.org/packages/ff/86/620451bdc02e91b1712cd71890c17077ee97e2a28493836a87e47b8e70ff/propcache-0.3.0-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:19d36bb351ad5554ff20f2ae75f88ce205b0748c38b146c75628577020351e3c", size = 212492, upload-time = "2025-02-20T19:00:41.077Z" }, - { url = "https://files.pythonhosted.org/packages/6e/1b/e8f86921ed4016da80faf3b8f515f7829decabdbff106736bfff353bceba/propcache-0.3.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:6032231d4a5abd67c7f71168fd64a47b6b451fbcb91c8397c2f7610e67683810", size = 215113, upload-time = "2025-02-20T19:00:43.577Z" }, - { url = "https://files.pythonhosted.org/packages/1a/95/a61d86cc49aa0945f6c06f3a4614fc543e311a50558c92861f5e9691a37c/propcache-0.3.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:6985a593417cdbc94c7f9c3403747335e450c1599da1647a5af76539672464d3", size = 228330, upload-time = "2025-02-20T19:00:45.163Z" }, - { url = "https://files.pythonhosted.org/packages/8f/7d/10dbae48ff2bb189e92c2b3487a48f3229146a25941ad0d485934d1104d4/propcache-0.3.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:6a1948df1bb1d56b5e7b0553c0fa04fd0e320997ae99689488201f19fa90d2e7", size = 231942, upload-time = "2025-02-20T19:00:46.771Z" }, - { url = "https://files.pythonhosted.org/packages/39/ce/82d16aec96c5513ae7db13ab901a65a1e54c915292fb5b2390e33275b61d/propcache-0.3.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:8319293e85feadbbfe2150a5659dbc2ebc4afdeaf7d98936fb9a2f2ba0d4c35c", size = 223077, upload-time = "2025-02-20T19:00:53.044Z" }, - { url = "https://files.pythonhosted.org/packages/c8/e0/cb077e8e7a583c733df7f53327fcbdb92e42be59b976ce60bf1d904a0efe/propcache-0.3.0-cp311-cp311-win32.whl", hash = "sha256:63f26258a163c34542c24808f03d734b338da66ba91f410a703e505c8485791d", size = 40455, upload-time = "2025-02-20T19:00:55.338Z" }, - { url = "https://files.pythonhosted.org/packages/d8/35/57abeb6146fe3c19081eeaf3d9d4cfea256f87f1e5101acf80d3332c1820/propcache-0.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:cacea77ef7a2195f04f9279297684955e3d1ae4241092ff0cfcef532bb7a1c32", size = 44705, upload-time = "2025-02-20T19:00:56.947Z" }, - { url = "https://files.pythonhosted.org/packages/8d/2c/921f15dc365796ec23975b322b0078eae72995c7b4d49eba554c6a308d70/propcache-0.3.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:e53d19c2bf7d0d1e6998a7e693c7e87300dd971808e6618964621ccd0e01fe4e", size = 79867, upload-time = "2025-02-20T19:00:59.948Z" }, - { url = "https://files.pythonhosted.org/packages/11/a5/4a6cc1a559d1f2fb57ea22edc4245158cdffae92f7f92afcee2913f84417/propcache-0.3.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:a61a68d630e812b67b5bf097ab84e2cd79b48c792857dc10ba8a223f5b06a2af", size = 46109, upload-time = "2025-02-20T19:01:04.447Z" }, - { url = "https://files.pythonhosted.org/packages/e1/6d/28bfd3af3a567ad7d667348e7f46a520bda958229c4d545ba138a044232f/propcache-0.3.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:fb91d20fa2d3b13deea98a690534697742029f4fb83673a3501ae6e3746508b5", size = 45635, upload-time = "2025-02-20T19:01:07.024Z" }, - { url = "https://files.pythonhosted.org/packages/73/20/d75b42eaffe5075eac2f4e168f6393d21c664c91225288811d85451b2578/propcache-0.3.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:67054e47c01b7b349b94ed0840ccae075449503cf1fdd0a1fdd98ab5ddc2667b", size = 242159, upload-time = "2025-02-20T19:01:10.047Z" }, - { url = "https://files.pythonhosted.org/packages/a5/fb/4b537dd92f9fd4be68042ec51c9d23885ca5fafe51ec24c58d9401034e5f/propcache-0.3.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:997e7b8f173a391987df40f3b52c423e5850be6f6df0dcfb5376365440b56667", size = 248163, upload-time = "2025-02-20T19:01:12.883Z" }, - { url = "https://files.pythonhosted.org/packages/e7/af/8a9db04ac596d531ca0ef7dde518feaadfcdabef7b17d6a5ec59ee3effc2/propcache-0.3.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8d663fd71491dde7dfdfc899d13a067a94198e90695b4321084c6e450743b8c7", size = 248794, upload-time = "2025-02-20T19:01:15.291Z" }, - { url = "https://files.pythonhosted.org/packages/9d/c4/ecfc988879c0fd9db03228725b662d76cf484b6b46f7e92fee94e4b52490/propcache-0.3.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8884ba1a0fe7210b775106b25850f5e5a9dc3c840d1ae9924ee6ea2eb3acbfe7", size = 243912, upload-time = "2025-02-20T19:01:16.95Z" }, - { url = "https://files.pythonhosted.org/packages/04/a2/298dd27184faa8b7d91cc43488b578db218b3cc85b54d912ed27b8c5597a/propcache-0.3.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:aa806bbc13eac1ab6291ed21ecd2dd426063ca5417dd507e6be58de20e58dfcf", size = 229402, upload-time = "2025-02-20T19:01:20.913Z" }, - { url = "https://files.pythonhosted.org/packages/be/0d/efe7fec316ca92dbf4bc4a9ba49ca889c43ca6d48ab1d6fa99fc94e5bb98/propcache-0.3.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6f4d7a7c0aff92e8354cceca6fe223973ddf08401047920df0fcb24be2bd5138", size = 226896, upload-time = "2025-02-20T19:01:23.57Z" }, - { url = "https://files.pythonhosted.org/packages/60/63/72404380ae1d9c96d96e165aa02c66c2aae6072d067fc4713da5cde96762/propcache-0.3.0-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:9be90eebc9842a93ef8335291f57b3b7488ac24f70df96a6034a13cb58e6ff86", size = 221447, upload-time = "2025-02-20T19:01:26.142Z" }, - { url = "https://files.pythonhosted.org/packages/9d/18/b8392cab6e0964b67a30a8f4dadeaff64dc7022b5a34bb1d004ea99646f4/propcache-0.3.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:bf15fc0b45914d9d1b706f7c9c4f66f2b7b053e9517e40123e137e8ca8958b3d", size = 222440, upload-time = "2025-02-20T19:01:28.438Z" }, - { url = "https://files.pythonhosted.org/packages/6f/be/105d9ceda0f97eff8c06bac1673448b2db2a497444de3646464d3f5dc881/propcache-0.3.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5a16167118677d94bb48bfcd91e420088854eb0737b76ec374b91498fb77a70e", size = 234104, upload-time = "2025-02-20T19:01:31.256Z" }, - { url = "https://files.pythonhosted.org/packages/cb/c9/f09a4ec394cfcce4053d8b2a04d622b5f22d21ba9bb70edd0cad061fa77b/propcache-0.3.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:41de3da5458edd5678b0f6ff66691507f9885f5fe6a0fb99a5d10d10c0fd2d64", size = 239086, upload-time = "2025-02-20T19:01:33.753Z" }, - { url = "https://files.pythonhosted.org/packages/ea/aa/96f7f9ed6def82db67c972bdb7bd9f28b95d7d98f7e2abaf144c284bf609/propcache-0.3.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:728af36011bb5d344c4fe4af79cfe186729efb649d2f8b395d1572fb088a996c", size = 230991, upload-time = "2025-02-20T19:01:35.433Z" }, - { url = "https://files.pythonhosted.org/packages/5a/11/bee5439de1307d06fad176f7143fec906e499c33d7aff863ea8428b8e98b/propcache-0.3.0-cp312-cp312-win32.whl", hash = "sha256:6b5b7fd6ee7b54e01759f2044f936dcf7dea6e7585f35490f7ca0420fe723c0d", size = 40337, upload-time = "2025-02-20T19:01:37.655Z" }, - { url = "https://files.pythonhosted.org/packages/e4/17/e5789a54a0455a61cb9efc4ca6071829d992220c2998a27c59aeba749f6f/propcache-0.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:2d15bc27163cd4df433e75f546b9ac31c1ba7b0b128bfb1b90df19082466ff57", size = 44404, upload-time = "2025-02-20T19:01:38.946Z" }, - { url = "https://files.pythonhosted.org/packages/3a/0f/a79dd23a0efd6ee01ab0dc9750d8479b343bfd0c73560d59d271eb6a99d4/propcache-0.3.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:a2b9bf8c79b660d0ca1ad95e587818c30ccdb11f787657458d6f26a1ea18c568", size = 77287, upload-time = "2025-02-20T19:01:40.897Z" }, - { url = "https://files.pythonhosted.org/packages/b8/51/76675703c90de38ac75adb8deceb3f3ad99b67ff02a0fa5d067757971ab8/propcache-0.3.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b0c1a133d42c6fc1f5fbcf5c91331657a1ff822e87989bf4a6e2e39b818d0ee9", size = 44923, upload-time = "2025-02-20T19:01:42.397Z" }, - { url = "https://files.pythonhosted.org/packages/01/9b/fd5ddbee66cf7686e73c516227c2fd9bf471dbfed0f48329d095ea1228d3/propcache-0.3.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:bb2f144c6d98bb5cbc94adeb0447cfd4c0f991341baa68eee3f3b0c9c0e83767", size = 44325, upload-time = "2025-02-20T19:01:43.976Z" }, - { url = "https://files.pythonhosted.org/packages/13/1c/6961f11eb215a683b34b903b82bde486c606516c1466bf1fa67f26906d51/propcache-0.3.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d1323cd04d6e92150bcc79d0174ce347ed4b349d748b9358fd2e497b121e03c8", size = 225116, upload-time = "2025-02-20T19:01:45.488Z" }, - { url = "https://files.pythonhosted.org/packages/ef/ea/f8410c40abcb2e40dffe9adeed017898c930974650a63e5c79b886aa9f73/propcache-0.3.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3b812b3cb6caacd072276ac0492d249f210006c57726b6484a1e1805b3cfeea0", size = 229905, upload-time = "2025-02-20T19:01:49.454Z" }, - { url = "https://files.pythonhosted.org/packages/ef/5a/a9bf90894001468bf8e6ea293bb00626cc9ef10f8eb7996e9ec29345c7ed/propcache-0.3.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:742840d1d0438eb7ea4280f3347598f507a199a35a08294afdcc560c3739989d", size = 233221, upload-time = "2025-02-20T19:01:51.142Z" }, - { url = "https://files.pythonhosted.org/packages/dd/ce/fffdddd9725b690b01d345c1156b4c2cc6dca09ab5c23a6d07b8f37d6e2f/propcache-0.3.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7c6e7e4f9167fddc438cd653d826f2222222564daed4116a02a184b464d3ef05", size = 227627, upload-time = "2025-02-20T19:01:53.695Z" }, - { url = "https://files.pythonhosted.org/packages/58/ae/45c89a5994a334735a3032b48e8e4a98c05d9536ddee0719913dc27da548/propcache-0.3.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a94ffc66738da99232ddffcf7910e0f69e2bbe3a0802e54426dbf0714e1c2ffe", size = 214217, upload-time = "2025-02-20T19:01:55.309Z" }, - { url = "https://files.pythonhosted.org/packages/01/84/bc60188c3290ff8f5f4a92b9ca2d93a62e449c8daf6fd11ad517ad136926/propcache-0.3.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:3c6ec957025bf32b15cbc6b67afe233c65b30005e4c55fe5768e4bb518d712f1", size = 212921, upload-time = "2025-02-20T19:01:57.893Z" }, - { url = "https://files.pythonhosted.org/packages/14/b3/39d60224048feef7a96edabb8217dc3f75415457e5ebbef6814f8b2a27b5/propcache-0.3.0-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:549722908de62aa0b47a78b90531c022fa6e139f9166be634f667ff45632cc92", size = 208200, upload-time = "2025-02-20T19:02:00.026Z" }, - { url = "https://files.pythonhosted.org/packages/9d/b3/0a6720b86791251273fff8a01bc8e628bc70903513bd456f86cde1e1ef84/propcache-0.3.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:5d62c4f6706bff5d8a52fd51fec6069bef69e7202ed481486c0bc3874912c787", size = 208400, upload-time = "2025-02-20T19:02:03.997Z" }, - { url = "https://files.pythonhosted.org/packages/e9/4f/bb470f3e687790547e2e78105fb411f54e0cdde0d74106ccadd2521c6572/propcache-0.3.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:24c04f8fbf60094c531667b8207acbae54146661657a1b1be6d3ca7773b7a545", size = 218116, upload-time = "2025-02-20T19:02:06.042Z" }, - { url = "https://files.pythonhosted.org/packages/34/71/277f7f9add469698ac9724c199bfe06f85b199542121a71f65a80423d62a/propcache-0.3.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:7c5f5290799a3f6539cc5e6f474c3e5c5fbeba74a5e1e5be75587746a940d51e", size = 222911, upload-time = "2025-02-20T19:02:08.748Z" }, - { url = "https://files.pythonhosted.org/packages/92/e3/a7b9782aef5a2fc765b1d97da9ec7aed2f25a4e985703608e73232205e3f/propcache-0.3.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4fa0e7c9c3cf7c276d4f6ab9af8adddc127d04e0fcabede315904d2ff76db626", size = 216563, upload-time = "2025-02-20T19:02:11.322Z" }, - { url = "https://files.pythonhosted.org/packages/ab/76/0583ca2c551aa08ffcff87b2c6849c8f01c1f6fb815a5226f0c5c202173e/propcache-0.3.0-cp313-cp313-win32.whl", hash = "sha256:ee0bd3a7b2e184e88d25c9baa6a9dc609ba25b76daae942edfb14499ac7ec374", size = 39763, upload-time = "2025-02-20T19:02:12.977Z" }, - { url = "https://files.pythonhosted.org/packages/80/ec/c6a84f9a36f608379b95f0e786c111d5465926f8c62f12be8cdadb02b15c/propcache-0.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:1c8f7d896a16da9455f882870a507567d4f58c53504dc2d4b1e1d386dfe4588a", size = 43650, upload-time = "2025-02-20T19:02:15.041Z" }, - { url = "https://files.pythonhosted.org/packages/ee/95/7d32e3560f5bf83fc2f2a4c1b0c181d327d53d5f85ebd045ab89d4d97763/propcache-0.3.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:e560fd75aaf3e5693b91bcaddd8b314f4d57e99aef8a6c6dc692f935cc1e6bbf", size = 82140, upload-time = "2025-02-20T19:02:16.562Z" }, - { url = "https://files.pythonhosted.org/packages/86/89/752388f12e6027a5e63f5d075f15291ded48e2d8311314fff039da5a9b11/propcache-0.3.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:65a37714b8ad9aba5780325228598a5b16c47ba0f8aeb3dc0514701e4413d7c0", size = 47296, upload-time = "2025-02-20T19:02:17.974Z" }, - { url = "https://files.pythonhosted.org/packages/1b/4c/b55c98d586c69180d3048984a57a5ea238bdeeccf82dbfcd598e935e10bb/propcache-0.3.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:07700939b2cbd67bfb3b76a12e1412405d71019df00ca5697ce75e5ef789d829", size = 46724, upload-time = "2025-02-20T19:02:19.588Z" }, - { url = "https://files.pythonhosted.org/packages/0f/b6/67451a437aed90c4e951e320b5b3d7eb584ade1d5592f6e5e8f678030989/propcache-0.3.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7c0fdbdf6983526e269e5a8d53b7ae3622dd6998468821d660d0daf72779aefa", size = 291499, upload-time = "2025-02-20T19:02:21.1Z" }, - { url = "https://files.pythonhosted.org/packages/ee/ff/e4179facd21515b24737e1e26e02615dfb5ed29416eed4cf5bc6ac5ce5fb/propcache-0.3.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:794c3dd744fad478b6232289c866c25406ecdfc47e294618bdf1697e69bd64a6", size = 293911, upload-time = "2025-02-20T19:02:24.248Z" }, - { url = "https://files.pythonhosted.org/packages/76/8d/94a8585992a064a23bd54f56c5e58c3b8bf0c0a06ae10e56f2353ae16c3d/propcache-0.3.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4544699674faf66fb6b4473a1518ae4999c1b614f0b8297b1cef96bac25381db", size = 293301, upload-time = "2025-02-20T19:02:26.034Z" }, - { url = "https://files.pythonhosted.org/packages/b0/b8/2c860c92b4134f68c7716c6f30a0d723973f881c32a6d7a24c4ddca05fdf/propcache-0.3.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fddb8870bdb83456a489ab67c6b3040a8d5a55069aa6f72f9d872235fbc52f54", size = 281947, upload-time = "2025-02-20T19:02:27.838Z" }, - { url = "https://files.pythonhosted.org/packages/cd/72/b564be7411b525d11757b713c757c21cd4dc13b6569c3b2b8f6d3c96fd5e/propcache-0.3.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f857034dc68d5ceb30fb60afb6ff2103087aea10a01b613985610e007053a121", size = 268072, upload-time = "2025-02-20T19:02:29.594Z" }, - { url = "https://files.pythonhosted.org/packages/37/68/d94649e399e8d7fc051e5a4f2334efc567993525af083db145a70690a121/propcache-0.3.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:02df07041e0820cacc8f739510078f2aadcfd3fc57eaeeb16d5ded85c872c89e", size = 275190, upload-time = "2025-02-20T19:02:32.255Z" }, - { url = "https://files.pythonhosted.org/packages/d8/3c/446e125f5bbbc1922964dd67cb541c01cdb678d811297b79a4ff6accc843/propcache-0.3.0-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:f47d52fd9b2ac418c4890aad2f6d21a6b96183c98021f0a48497a904199f006e", size = 254145, upload-time = "2025-02-20T19:02:33.932Z" }, - { url = "https://files.pythonhosted.org/packages/f4/80/fd3f741483dc8e59f7ba7e05eaa0f4e11677d7db2077522b92ff80117a2a/propcache-0.3.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:9ff4e9ecb6e4b363430edf2c6e50173a63e0820e549918adef70515f87ced19a", size = 257163, upload-time = "2025-02-20T19:02:35.675Z" }, - { url = "https://files.pythonhosted.org/packages/dc/cf/6292b5ce6ed0017e6a89024a827292122cc41b6259b30ada0c6732288513/propcache-0.3.0-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:ecc2920630283e0783c22e2ac94427f8cca29a04cfdf331467d4f661f4072dac", size = 280249, upload-time = "2025-02-20T19:02:38.406Z" }, - { url = "https://files.pythonhosted.org/packages/e8/f0/fd9b8247b449fe02a4f96538b979997e229af516d7462b006392badc59a1/propcache-0.3.0-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:c441c841e82c5ba7a85ad25986014be8d7849c3cfbdb6004541873505929a74e", size = 288741, upload-time = "2025-02-20T19:02:40.149Z" }, - { url = "https://files.pythonhosted.org/packages/64/71/cf831fdc2617f86cfd7f414cfc487d018e722dac8acc098366ce9bba0941/propcache-0.3.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:6c929916cbdb540d3407c66f19f73387f43e7c12fa318a66f64ac99da601bcdf", size = 277061, upload-time = "2025-02-20T19:02:42.309Z" }, - { url = "https://files.pythonhosted.org/packages/42/78/9432542a35d944abeca9e02927a0de38cd7a298466d8ffa171536e2381c3/propcache-0.3.0-cp313-cp313t-win32.whl", hash = "sha256:0c3e893c4464ebd751b44ae76c12c5f5c1e4f6cbd6fbf67e3783cd93ad221863", size = 42252, upload-time = "2025-02-20T19:02:44.447Z" }, - { url = "https://files.pythonhosted.org/packages/6f/45/960365f4f8978f48ebb56b1127adf33a49f2e69ecd46ac1f46d6cf78a79d/propcache-0.3.0-cp313-cp313t-win_amd64.whl", hash = "sha256:75e872573220d1ee2305b35c9813626e620768248425f58798413e9c39741f46", size = 46425, upload-time = "2025-02-20T19:02:48.071Z" }, - { url = "https://files.pythonhosted.org/packages/b5/35/6c4c6fc8774a9e3629cd750dc24a7a4fb090a25ccd5c3246d127b70f9e22/propcache-0.3.0-py3-none-any.whl", hash = "sha256:67dda3c7325691c2081510e92c561f465ba61b975f481735aefdfc845d2cd043", size = 12101, upload-time = "2025-02-20T19:03:27.202Z" }, +version = "0.3.2" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/a6/16/43264e4a779dd8588c21a70f0709665ee8f611211bdd2c87d952cfa7c776/propcache-0.3.2.tar.gz", hash = "sha256:20d7d62e4e7ef05f221e0db2856b979540686342e7dd9973b815599c7057e168", size = 44139, upload-time = "2025-06-09T22:56:06.081Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/ab/14/510deed325e262afeb8b360043c5d7c960da7d3ecd6d6f9496c9c56dc7f4/propcache-0.3.2-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:22d9962a358aedbb7a2e36187ff273adeaab9743373a272976d2e348d08c7770", size = 73178, upload-time = "2025-06-09T22:53:40.126Z" }, + { url = "https://files.pythonhosted.org/packages/cd/4e/ad52a7925ff01c1325653a730c7ec3175a23f948f08626a534133427dcff/propcache-0.3.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:0d0fda578d1dc3f77b6b5a5dce3b9ad69a8250a891760a548df850a5e8da87f3", size = 43133, upload-time = "2025-06-09T22:53:41.965Z" }, + { url = "https://files.pythonhosted.org/packages/63/7c/e9399ba5da7780871db4eac178e9c2e204c23dd3e7d32df202092a1ed400/propcache-0.3.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3def3da3ac3ce41562d85db655d18ebac740cb3fa4367f11a52b3da9d03a5cc3", size = 43039, upload-time = "2025-06-09T22:53:43.268Z" }, + { url = "https://files.pythonhosted.org/packages/22/e1/58da211eb8fdc6fc854002387d38f415a6ca5f5c67c1315b204a5d3e9d7a/propcache-0.3.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9bec58347a5a6cebf239daba9bda37dffec5b8d2ce004d9fe4edef3d2815137e", size = 201903, upload-time = "2025-06-09T22:53:44.872Z" }, + { url = "https://files.pythonhosted.org/packages/c4/0a/550ea0f52aac455cb90111c8bab995208443e46d925e51e2f6ebdf869525/propcache-0.3.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:55ffda449a507e9fbd4aca1a7d9aa6753b07d6166140e5a18d2ac9bc49eac220", size = 213362, upload-time = "2025-06-09T22:53:46.707Z" }, + { url = "https://files.pythonhosted.org/packages/5a/af/9893b7d878deda9bb69fcf54600b247fba7317761b7db11fede6e0f28bd0/propcache-0.3.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:64a67fb39229a8a8491dd42f864e5e263155e729c2e7ff723d6e25f596b1e8cb", size = 210525, upload-time = "2025-06-09T22:53:48.547Z" }, + { url = "https://files.pythonhosted.org/packages/7c/bb/38fd08b278ca85cde36d848091ad2b45954bc5f15cce494bb300b9285831/propcache-0.3.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9da1cf97b92b51253d5b68cf5a2b9e0dafca095e36b7f2da335e27dc6172a614", size = 198283, upload-time = "2025-06-09T22:53:50.067Z" }, + { url = "https://files.pythonhosted.org/packages/78/8c/9fe55bd01d362bafb413dfe508c48753111a1e269737fa143ba85693592c/propcache-0.3.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5f559e127134b07425134b4065be45b166183fdcb433cb6c24c8e4149056ad50", size = 191872, upload-time = "2025-06-09T22:53:51.438Z" }, + { url = "https://files.pythonhosted.org/packages/54/14/4701c33852937a22584e08abb531d654c8bcf7948a8f87ad0a4822394147/propcache-0.3.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:aff2e4e06435d61f11a428360a932138d0ec288b0a31dd9bd78d200bd4a2b339", size = 199452, upload-time = "2025-06-09T22:53:53.229Z" }, + { url = "https://files.pythonhosted.org/packages/16/44/447f2253d859602095356007657ee535e0093215ea0b3d1d6a41d16e5201/propcache-0.3.2-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:4927842833830942a5d0a56e6f4839bc484785b8e1ce8d287359794818633ba0", size = 191567, upload-time = "2025-06-09T22:53:54.541Z" }, + { url = "https://files.pythonhosted.org/packages/f2/b3/e4756258749bb2d3b46defcff606a2f47410bab82be5824a67e84015b267/propcache-0.3.2-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:6107ddd08b02654a30fb8ad7a132021759d750a82578b94cd55ee2772b6ebea2", size = 193015, upload-time = "2025-06-09T22:53:56.44Z" }, + { url = "https://files.pythonhosted.org/packages/1e/df/e6d3c7574233164b6330b9fd697beeac402afd367280e6dc377bb99b43d9/propcache-0.3.2-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:70bd8b9cd6b519e12859c99f3fc9a93f375ebd22a50296c3a295028bea73b9e7", size = 204660, upload-time = "2025-06-09T22:53:57.839Z" }, + { url = "https://files.pythonhosted.org/packages/b2/53/e4d31dd5170b4a0e2e6b730f2385a96410633b4833dc25fe5dffd1f73294/propcache-0.3.2-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:2183111651d710d3097338dd1893fcf09c9f54e27ff1a8795495a16a469cc90b", size = 206105, upload-time = "2025-06-09T22:53:59.638Z" }, + { url = "https://files.pythonhosted.org/packages/7f/fe/74d54cf9fbe2a20ff786e5f7afcfde446588f0cf15fb2daacfbc267b866c/propcache-0.3.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:fb075ad271405dcad8e2a7ffc9a750a3bf70e533bd86e89f0603e607b93aa64c", size = 196980, upload-time = "2025-06-09T22:54:01.071Z" }, + { url = "https://files.pythonhosted.org/packages/22/ec/c469c9d59dada8a7679625e0440b544fe72e99311a4679c279562051f6fc/propcache-0.3.2-cp310-cp310-win32.whl", hash = "sha256:404d70768080d3d3bdb41d0771037da19d8340d50b08e104ca0e7f9ce55fce70", size = 37679, upload-time = "2025-06-09T22:54:03.003Z" }, + { url = "https://files.pythonhosted.org/packages/38/35/07a471371ac89d418f8d0b699c75ea6dca2041fbda360823de21f6a9ce0a/propcache-0.3.2-cp310-cp310-win_amd64.whl", hash = "sha256:7435d766f978b4ede777002e6b3b6641dd229cd1da8d3d3106a45770365f9ad9", size = 41459, upload-time = "2025-06-09T22:54:04.134Z" }, + { url = "https://files.pythonhosted.org/packages/80/8d/e8b436717ab9c2cfc23b116d2c297305aa4cd8339172a456d61ebf5669b8/propcache-0.3.2-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:0b8d2f607bd8f80ddc04088bc2a037fdd17884a6fcadc47a96e334d72f3717be", size = 74207, upload-time = "2025-06-09T22:54:05.399Z" }, + { url = "https://files.pythonhosted.org/packages/d6/29/1e34000e9766d112171764b9fa3226fa0153ab565d0c242c70e9945318a7/propcache-0.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:06766d8f34733416e2e34f46fea488ad5d60726bb9481d3cddf89a6fa2d9603f", size = 43648, upload-time = "2025-06-09T22:54:08.023Z" }, + { url = "https://files.pythonhosted.org/packages/46/92/1ad5af0df781e76988897da39b5f086c2bf0f028b7f9bd1f409bb05b6874/propcache-0.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a2dc1f4a1df4fecf4e6f68013575ff4af84ef6f478fe5344317a65d38a8e6dc9", size = 43496, upload-time = "2025-06-09T22:54:09.228Z" }, + { url = "https://files.pythonhosted.org/packages/b3/ce/e96392460f9fb68461fabab3e095cb00c8ddf901205be4eae5ce246e5b7e/propcache-0.3.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:be29c4f4810c5789cf10ddf6af80b041c724e629fa51e308a7a0fb19ed1ef7bf", size = 217288, upload-time = "2025-06-09T22:54:10.466Z" }, + { url = "https://files.pythonhosted.org/packages/c5/2a/866726ea345299f7ceefc861a5e782b045545ae6940851930a6adaf1fca6/propcache-0.3.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:59d61f6970ecbd8ff2e9360304d5c8876a6abd4530cb752c06586849ac8a9dc9", size = 227456, upload-time = "2025-06-09T22:54:11.828Z" }, + { url = "https://files.pythonhosted.org/packages/de/03/07d992ccb6d930398689187e1b3c718339a1c06b8b145a8d9650e4726166/propcache-0.3.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:62180e0b8dbb6b004baec00a7983e4cc52f5ada9cd11f48c3528d8cfa7b96a66", size = 225429, upload-time = "2025-06-09T22:54:13.823Z" }, + { url = "https://files.pythonhosted.org/packages/5d/e6/116ba39448753b1330f48ab8ba927dcd6cf0baea8a0ccbc512dfb49ba670/propcache-0.3.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c144ca294a204c470f18cf4c9d78887810d04a3e2fbb30eea903575a779159df", size = 213472, upload-time = "2025-06-09T22:54:15.232Z" }, + { url = "https://files.pythonhosted.org/packages/a6/85/f01f5d97e54e428885a5497ccf7f54404cbb4f906688a1690cd51bf597dc/propcache-0.3.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c5c2a784234c28854878d68978265617aa6dc0780e53d44b4d67f3651a17a9a2", size = 204480, upload-time = "2025-06-09T22:54:17.104Z" }, + { url = "https://files.pythonhosted.org/packages/e3/79/7bf5ab9033b8b8194cc3f7cf1aaa0e9c3256320726f64a3e1f113a812dce/propcache-0.3.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5745bc7acdafa978ca1642891b82c19238eadc78ba2aaa293c6863b304e552d7", size = 214530, upload-time = "2025-06-09T22:54:18.512Z" }, + { url = "https://files.pythonhosted.org/packages/31/0b/bd3e0c00509b609317df4a18e6b05a450ef2d9a963e1d8bc9c9415d86f30/propcache-0.3.2-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:c0075bf773d66fa8c9d41f66cc132ecc75e5bb9dd7cce3cfd14adc5ca184cb95", size = 205230, upload-time = "2025-06-09T22:54:19.947Z" }, + { url = "https://files.pythonhosted.org/packages/7a/23/fae0ff9b54b0de4e819bbe559508da132d5683c32d84d0dc2ccce3563ed4/propcache-0.3.2-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:5f57aa0847730daceff0497f417c9de353c575d8da3579162cc74ac294c5369e", size = 206754, upload-time = "2025-06-09T22:54:21.716Z" }, + { url = "https://files.pythonhosted.org/packages/b7/7f/ad6a3c22630aaa5f618b4dc3c3598974a72abb4c18e45a50b3cdd091eb2f/propcache-0.3.2-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:eef914c014bf72d18efb55619447e0aecd5fb7c2e3fa7441e2e5d6099bddff7e", size = 218430, upload-time = "2025-06-09T22:54:23.17Z" }, + { url = "https://files.pythonhosted.org/packages/5b/2c/ba4f1c0e8a4b4c75910742f0d333759d441f65a1c7f34683b4a74c0ee015/propcache-0.3.2-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:2a4092e8549031e82facf3decdbc0883755d5bbcc62d3aea9d9e185549936dcf", size = 223884, upload-time = "2025-06-09T22:54:25.539Z" }, + { url = "https://files.pythonhosted.org/packages/88/e4/ebe30fc399e98572019eee82ad0caf512401661985cbd3da5e3140ffa1b0/propcache-0.3.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:85871b050f174bc0bfb437efbdb68aaf860611953ed12418e4361bc9c392749e", size = 211480, upload-time = "2025-06-09T22:54:26.892Z" }, + { url = "https://files.pythonhosted.org/packages/96/0a/7d5260b914e01d1d0906f7f38af101f8d8ed0dc47426219eeaf05e8ea7c2/propcache-0.3.2-cp311-cp311-win32.whl", hash = "sha256:36c8d9b673ec57900c3554264e630d45980fd302458e4ac801802a7fd2ef7897", size = 37757, upload-time = "2025-06-09T22:54:28.241Z" }, + { url = "https://files.pythonhosted.org/packages/e1/2d/89fe4489a884bc0da0c3278c552bd4ffe06a1ace559db5ef02ef24ab446b/propcache-0.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:e53af8cb6a781b02d2ea079b5b853ba9430fcbe18a8e3ce647d5982a3ff69f39", size = 41500, upload-time = "2025-06-09T22:54:29.4Z" }, + { url = "https://files.pythonhosted.org/packages/a8/42/9ca01b0a6f48e81615dca4765a8f1dd2c057e0540f6116a27dc5ee01dfb6/propcache-0.3.2-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:8de106b6c84506b31c27168582cd3cb3000a6412c16df14a8628e5871ff83c10", size = 73674, upload-time = "2025-06-09T22:54:30.551Z" }, + { url = "https://files.pythonhosted.org/packages/af/6e/21293133beb550f9c901bbece755d582bfaf2176bee4774000bd4dd41884/propcache-0.3.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:28710b0d3975117239c76600ea351934ac7b5ff56e60953474342608dbbb6154", size = 43570, upload-time = "2025-06-09T22:54:32.296Z" }, + { url = "https://files.pythonhosted.org/packages/0c/c8/0393a0a3a2b8760eb3bde3c147f62b20044f0ddac81e9d6ed7318ec0d852/propcache-0.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ce26862344bdf836650ed2487c3d724b00fbfec4233a1013f597b78c1cb73615", size = 43094, upload-time = "2025-06-09T22:54:33.929Z" }, + { url = "https://files.pythonhosted.org/packages/37/2c/489afe311a690399d04a3e03b069225670c1d489eb7b044a566511c1c498/propcache-0.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bca54bd347a253af2cf4544bbec232ab982f4868de0dd684246b67a51bc6b1db", size = 226958, upload-time = "2025-06-09T22:54:35.186Z" }, + { url = "https://files.pythonhosted.org/packages/9d/ca/63b520d2f3d418c968bf596839ae26cf7f87bead026b6192d4da6a08c467/propcache-0.3.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:55780d5e9a2ddc59711d727226bb1ba83a22dd32f64ee15594b9392b1f544eb1", size = 234894, upload-time = "2025-06-09T22:54:36.708Z" }, + { url = "https://files.pythonhosted.org/packages/11/60/1d0ed6fff455a028d678df30cc28dcee7af77fa2b0e6962ce1df95c9a2a9/propcache-0.3.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:035e631be25d6975ed87ab23153db6a73426a48db688070d925aa27e996fe93c", size = 233672, upload-time = "2025-06-09T22:54:38.062Z" }, + { url = "https://files.pythonhosted.org/packages/37/7c/54fd5301ef38505ab235d98827207176a5c9b2aa61939b10a460ca53e123/propcache-0.3.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ee6f22b6eaa39297c751d0e80c0d3a454f112f5c6481214fcf4c092074cecd67", size = 224395, upload-time = "2025-06-09T22:54:39.634Z" }, + { url = "https://files.pythonhosted.org/packages/ee/1a/89a40e0846f5de05fdc6779883bf46ba980e6df4d2ff8fb02643de126592/propcache-0.3.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7ca3aee1aa955438c4dba34fc20a9f390e4c79967257d830f137bd5a8a32ed3b", size = 212510, upload-time = "2025-06-09T22:54:41.565Z" }, + { url = "https://files.pythonhosted.org/packages/5e/33/ca98368586c9566a6b8d5ef66e30484f8da84c0aac3f2d9aec6d31a11bd5/propcache-0.3.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7a4f30862869fa2b68380d677cc1c5fcf1e0f2b9ea0cf665812895c75d0ca3b8", size = 222949, upload-time = "2025-06-09T22:54:43.038Z" }, + { url = "https://files.pythonhosted.org/packages/ba/11/ace870d0aafe443b33b2f0b7efdb872b7c3abd505bfb4890716ad7865e9d/propcache-0.3.2-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:b77ec3c257d7816d9f3700013639db7491a434644c906a2578a11daf13176251", size = 217258, upload-time = "2025-06-09T22:54:44.376Z" }, + { url = "https://files.pythonhosted.org/packages/5b/d2/86fd6f7adffcfc74b42c10a6b7db721d1d9ca1055c45d39a1a8f2a740a21/propcache-0.3.2-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:cab90ac9d3f14b2d5050928483d3d3b8fb6b4018893fc75710e6aa361ecb2474", size = 213036, upload-time = "2025-06-09T22:54:46.243Z" }, + { url = "https://files.pythonhosted.org/packages/07/94/2d7d1e328f45ff34a0a284cf5a2847013701e24c2a53117e7c280a4316b3/propcache-0.3.2-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:0b504d29f3c47cf6b9e936c1852246c83d450e8e063d50562115a6be6d3a2535", size = 227684, upload-time = "2025-06-09T22:54:47.63Z" }, + { url = "https://files.pythonhosted.org/packages/b7/05/37ae63a0087677e90b1d14710e532ff104d44bc1efa3b3970fff99b891dc/propcache-0.3.2-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:ce2ac2675a6aa41ddb2a0c9cbff53780a617ac3d43e620f8fd77ba1c84dcfc06", size = 234562, upload-time = "2025-06-09T22:54:48.982Z" }, + { url = "https://files.pythonhosted.org/packages/a4/7c/3f539fcae630408d0bd8bf3208b9a647ccad10976eda62402a80adf8fc34/propcache-0.3.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:62b4239611205294cc433845b914131b2a1f03500ff3c1ed093ed216b82621e1", size = 222142, upload-time = "2025-06-09T22:54:50.424Z" }, + { url = "https://files.pythonhosted.org/packages/7c/d2/34b9eac8c35f79f8a962546b3e97e9d4b990c420ee66ac8255d5d9611648/propcache-0.3.2-cp312-cp312-win32.whl", hash = "sha256:df4a81b9b53449ebc90cc4deefb052c1dd934ba85012aa912c7ea7b7e38b60c1", size = 37711, upload-time = "2025-06-09T22:54:52.072Z" }, + { url = "https://files.pythonhosted.org/packages/19/61/d582be5d226cf79071681d1b46b848d6cb03d7b70af7063e33a2787eaa03/propcache-0.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:7046e79b989d7fe457bb755844019e10f693752d169076138abf17f31380800c", size = 41479, upload-time = "2025-06-09T22:54:53.234Z" }, + { url = "https://files.pythonhosted.org/packages/dc/d1/8c747fafa558c603c4ca19d8e20b288aa0c7cda74e9402f50f31eb65267e/propcache-0.3.2-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:ca592ed634a73ca002967458187109265e980422116c0a107cf93d81f95af945", size = 71286, upload-time = "2025-06-09T22:54:54.369Z" }, + { url = "https://files.pythonhosted.org/packages/61/99/d606cb7986b60d89c36de8a85d58764323b3a5ff07770a99d8e993b3fa73/propcache-0.3.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9ecb0aad4020e275652ba3975740f241bd12a61f1a784df044cf7477a02bc252", size = 42425, upload-time = "2025-06-09T22:54:55.642Z" }, + { url = "https://files.pythonhosted.org/packages/8c/96/ef98f91bbb42b79e9bb82bdd348b255eb9d65f14dbbe3b1594644c4073f7/propcache-0.3.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:7f08f1cc28bd2eade7a8a3d2954ccc673bb02062e3e7da09bc75d843386b342f", size = 41846, upload-time = "2025-06-09T22:54:57.246Z" }, + { url = "https://files.pythonhosted.org/packages/5b/ad/3f0f9a705fb630d175146cd7b1d2bf5555c9beaed54e94132b21aac098a6/propcache-0.3.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d1a342c834734edb4be5ecb1e9fb48cb64b1e2320fccbd8c54bf8da8f2a84c33", size = 208871, upload-time = "2025-06-09T22:54:58.975Z" }, + { url = "https://files.pythonhosted.org/packages/3a/38/2085cda93d2c8b6ec3e92af2c89489a36a5886b712a34ab25de9fbca7992/propcache-0.3.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8a544caaae1ac73f1fecfae70ded3e93728831affebd017d53449e3ac052ac1e", size = 215720, upload-time = "2025-06-09T22:55:00.471Z" }, + { url = "https://files.pythonhosted.org/packages/61/c1/d72ea2dc83ac7f2c8e182786ab0fc2c7bd123a1ff9b7975bee671866fe5f/propcache-0.3.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:310d11aa44635298397db47a3ebce7db99a4cc4b9bbdfcf6c98a60c8d5261cf1", size = 215203, upload-time = "2025-06-09T22:55:01.834Z" }, + { url = "https://files.pythonhosted.org/packages/af/81/b324c44ae60c56ef12007105f1460d5c304b0626ab0cc6b07c8f2a9aa0b8/propcache-0.3.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4c1396592321ac83157ac03a2023aa6cc4a3cc3cfdecb71090054c09e5a7cce3", size = 206365, upload-time = "2025-06-09T22:55:03.199Z" }, + { url = "https://files.pythonhosted.org/packages/09/73/88549128bb89e66d2aff242488f62869014ae092db63ccea53c1cc75a81d/propcache-0.3.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8cabf5b5902272565e78197edb682017d21cf3b550ba0460ee473753f28d23c1", size = 196016, upload-time = "2025-06-09T22:55:04.518Z" }, + { url = "https://files.pythonhosted.org/packages/b9/3f/3bdd14e737d145114a5eb83cb172903afba7242f67c5877f9909a20d948d/propcache-0.3.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:0a2f2235ac46a7aa25bdeb03a9e7060f6ecbd213b1f9101c43b3090ffb971ef6", size = 205596, upload-time = "2025-06-09T22:55:05.942Z" }, + { url = "https://files.pythonhosted.org/packages/0f/ca/2f4aa819c357d3107c3763d7ef42c03980f9ed5c48c82e01e25945d437c1/propcache-0.3.2-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:92b69e12e34869a6970fd2f3da91669899994b47c98f5d430b781c26f1d9f387", size = 200977, upload-time = "2025-06-09T22:55:07.792Z" }, + { url = "https://files.pythonhosted.org/packages/cd/4a/e65276c7477533c59085251ae88505caf6831c0e85ff8b2e31ebcbb949b1/propcache-0.3.2-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:54e02207c79968ebbdffc169591009f4474dde3b4679e16634d34c9363ff56b4", size = 197220, upload-time = "2025-06-09T22:55:09.173Z" }, + { url = "https://files.pythonhosted.org/packages/7c/54/fc7152e517cf5578278b242396ce4d4b36795423988ef39bb8cd5bf274c8/propcache-0.3.2-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:4adfb44cb588001f68c5466579d3f1157ca07f7504fc91ec87862e2b8e556b88", size = 210642, upload-time = "2025-06-09T22:55:10.62Z" }, + { url = "https://files.pythonhosted.org/packages/b9/80/abeb4a896d2767bf5f1ea7b92eb7be6a5330645bd7fb844049c0e4045d9d/propcache-0.3.2-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:fd3e6019dc1261cd0291ee8919dd91fbab7b169bb76aeef6c716833a3f65d206", size = 212789, upload-time = "2025-06-09T22:55:12.029Z" }, + { url = "https://files.pythonhosted.org/packages/b3/db/ea12a49aa7b2b6d68a5da8293dcf50068d48d088100ac016ad92a6a780e6/propcache-0.3.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4c181cad81158d71c41a2bce88edce078458e2dd5ffee7eddd6b05da85079f43", size = 205880, upload-time = "2025-06-09T22:55:13.45Z" }, + { url = "https://files.pythonhosted.org/packages/d1/e5/9076a0bbbfb65d1198007059c65639dfd56266cf8e477a9707e4b1999ff4/propcache-0.3.2-cp313-cp313-win32.whl", hash = "sha256:8a08154613f2249519e549de2330cf8e2071c2887309a7b07fb56098f5170a02", size = 37220, upload-time = "2025-06-09T22:55:15.284Z" }, + { url = "https://files.pythonhosted.org/packages/d3/f5/b369e026b09a26cd77aa88d8fffd69141d2ae00a2abaaf5380d2603f4b7f/propcache-0.3.2-cp313-cp313-win_amd64.whl", hash = "sha256:e41671f1594fc4ab0a6dec1351864713cb3a279910ae8b58f884a88a0a632c05", size = 40678, upload-time = "2025-06-09T22:55:16.445Z" }, + { url = "https://files.pythonhosted.org/packages/a4/3a/6ece377b55544941a08d03581c7bc400a3c8cd3c2865900a68d5de79e21f/propcache-0.3.2-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:9a3cf035bbaf035f109987d9d55dc90e4b0e36e04bbbb95af3055ef17194057b", size = 76560, upload-time = "2025-06-09T22:55:17.598Z" }, + { url = "https://files.pythonhosted.org/packages/0c/da/64a2bb16418740fa634b0e9c3d29edff1db07f56d3546ca2d86ddf0305e1/propcache-0.3.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:156c03d07dc1323d8dacaa221fbe028c5c70d16709cdd63502778e6c3ccca1b0", size = 44676, upload-time = "2025-06-09T22:55:18.922Z" }, + { url = "https://files.pythonhosted.org/packages/36/7b/f025e06ea51cb72c52fb87e9b395cced02786610b60a3ed51da8af017170/propcache-0.3.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:74413c0ba02ba86f55cf60d18daab219f7e531620c15f1e23d95563f505efe7e", size = 44701, upload-time = "2025-06-09T22:55:20.106Z" }, + { url = "https://files.pythonhosted.org/packages/a4/00/faa1b1b7c3b74fc277f8642f32a4c72ba1d7b2de36d7cdfb676db7f4303e/propcache-0.3.2-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f066b437bb3fa39c58ff97ab2ca351db465157d68ed0440abecb21715eb24b28", size = 276934, upload-time = "2025-06-09T22:55:21.5Z" }, + { url = "https://files.pythonhosted.org/packages/74/ab/935beb6f1756e0476a4d5938ff44bf0d13a055fed880caf93859b4f1baf4/propcache-0.3.2-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f1304b085c83067914721e7e9d9917d41ad87696bf70f0bc7dee450e9c71ad0a", size = 278316, upload-time = "2025-06-09T22:55:22.918Z" }, + { url = "https://files.pythonhosted.org/packages/f8/9d/994a5c1ce4389610838d1caec74bdf0e98b306c70314d46dbe4fcf21a3e2/propcache-0.3.2-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ab50cef01b372763a13333b4e54021bdcb291fc9a8e2ccb9c2df98be51bcde6c", size = 282619, upload-time = "2025-06-09T22:55:24.651Z" }, + { url = "https://files.pythonhosted.org/packages/2b/00/a10afce3d1ed0287cef2e09506d3be9822513f2c1e96457ee369adb9a6cd/propcache-0.3.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fad3b2a085ec259ad2c2842666b2a0a49dea8463579c606426128925af1ed725", size = 265896, upload-time = "2025-06-09T22:55:26.049Z" }, + { url = "https://files.pythonhosted.org/packages/2e/a8/2aa6716ffa566ca57c749edb909ad27884680887d68517e4be41b02299f3/propcache-0.3.2-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:261fa020c1c14deafd54c76b014956e2f86991af198c51139faf41c4d5e83892", size = 252111, upload-time = "2025-06-09T22:55:27.381Z" }, + { url = "https://files.pythonhosted.org/packages/36/4f/345ca9183b85ac29c8694b0941f7484bf419c7f0fea2d1e386b4f7893eed/propcache-0.3.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:46d7f8aa79c927e5f987ee3a80205c987717d3659f035c85cf0c3680526bdb44", size = 268334, upload-time = "2025-06-09T22:55:28.747Z" }, + { url = "https://files.pythonhosted.org/packages/3e/ca/fcd54f78b59e3f97b3b9715501e3147f5340167733d27db423aa321e7148/propcache-0.3.2-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:6d8f3f0eebf73e3c0ff0e7853f68be638b4043c65a70517bb575eff54edd8dbe", size = 255026, upload-time = "2025-06-09T22:55:30.184Z" }, + { url = "https://files.pythonhosted.org/packages/8b/95/8e6a6bbbd78ac89c30c225210a5c687790e532ba4088afb8c0445b77ef37/propcache-0.3.2-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:03c89c1b14a5452cf15403e291c0ccd7751d5b9736ecb2c5bab977ad6c5bcd81", size = 250724, upload-time = "2025-06-09T22:55:31.646Z" }, + { url = "https://files.pythonhosted.org/packages/ee/b0/0dd03616142baba28e8b2d14ce5df6631b4673850a3d4f9c0f9dd714a404/propcache-0.3.2-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:0cc17efde71e12bbaad086d679ce575268d70bc123a5a71ea7ad76f70ba30bba", size = 268868, upload-time = "2025-06-09T22:55:33.209Z" }, + { url = "https://files.pythonhosted.org/packages/c5/98/2c12407a7e4fbacd94ddd32f3b1e3d5231e77c30ef7162b12a60e2dd5ce3/propcache-0.3.2-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:acdf05d00696bc0447e278bb53cb04ca72354e562cf88ea6f9107df8e7fd9770", size = 271322, upload-time = "2025-06-09T22:55:35.065Z" }, + { url = "https://files.pythonhosted.org/packages/35/91/9cb56efbb428b006bb85db28591e40b7736847b8331d43fe335acf95f6c8/propcache-0.3.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:4445542398bd0b5d32df908031cb1b30d43ac848e20470a878b770ec2dcc6330", size = 265778, upload-time = "2025-06-09T22:55:36.45Z" }, + { url = "https://files.pythonhosted.org/packages/9a/4c/b0fe775a2bdd01e176b14b574be679d84fc83958335790f7c9a686c1f468/propcache-0.3.2-cp313-cp313t-win32.whl", hash = "sha256:f86e5d7cd03afb3a1db8e9f9f6eff15794e79e791350ac48a8c924e6f439f394", size = 41175, upload-time = "2025-06-09T22:55:38.436Z" }, + { url = "https://files.pythonhosted.org/packages/a4/ff/47f08595e3d9b5e149c150f88d9714574f1a7cbd89fe2817158a952674bf/propcache-0.3.2-cp313-cp313t-win_amd64.whl", hash = "sha256:9704bedf6e7cbe3c65eca4379a9b53ee6a83749f047808cbb5044d40d7d72198", size = 44857, upload-time = "2025-06-09T22:55:39.687Z" }, + { url = "https://files.pythonhosted.org/packages/cc/35/cc0aaecf278bb4575b8555f2b137de5ab821595ddae9da9d3cd1da4072c7/propcache-0.3.2-py3-none-any.whl", hash = "sha256:98f1ec44fb675f5052cccc8e609c46ed23a35a1cfd18545ad4e29002d858a43f", size = 12663, upload-time = "2025-06-09T22:56:04.484Z" }, +] + +[[package]] +name = "proto-plus" +version = "1.26.1" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "protobuf" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/f4/ac/87285f15f7cce6d4a008f33f1757fb5a13611ea8914eb58c3d0d26243468/proto_plus-1.26.1.tar.gz", hash = "sha256:21a515a4c4c0088a773899e23c7bbade3d18f9c66c73edd4c7ee3816bc96a012", size = 56142, upload-time = "2025-03-10T15:54:38.843Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/4e/6d/280c4c2ce28b1593a19ad5239c8b826871fc6ec275c21afc8e1820108039/proto_plus-1.26.1-py3-none-any.whl", hash = "sha256:13285478c2dcf2abb829db158e1047e2f1e8d63a077d94263c2b88b043c75a66", size = 50163, upload-time = "2025-03-10T15:54:37.335Z" }, ] [[package]] name = "protobuf" -version = "5.29.4" +version = "6.31.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/17/7d/b9dca7365f0e2c4fa7c193ff795427cfa6290147e5185ab11ece280a18e7/protobuf-5.29.4.tar.gz", hash = "sha256:4f1dfcd7997b31ef8f53ec82781ff434a28bf71d9102ddde14d076adcfc78c99", size = 424902, upload-time = "2025-03-19T21:23:24.25Z" } +sdist = { url = "https://files.pythonhosted.org/packages/52/f3/b9655a711b32c19720253f6f06326faf90580834e2e83f840472d752bc8b/protobuf-6.31.1.tar.gz", hash = "sha256:d8cac4c982f0b957a4dc73a80e2ea24fab08e679c0de9deb835f4a12d69aca9a", size = 441797, upload-time = "2025-05-28T19:25:54.947Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/9a/b2/043a1a1a20edd134563699b0e91862726a0dc9146c090743b6c44d798e75/protobuf-5.29.4-cp310-abi3-win32.whl", hash = "sha256:13eb236f8eb9ec34e63fc8b1d6efd2777d062fa6aaa68268fb67cf77f6839ad7", size = 422709, upload-time = "2025-03-19T21:23:08.293Z" }, - { url = "https://files.pythonhosted.org/packages/79/fc/2474b59570daa818de6124c0a15741ee3e5d6302e9d6ce0bdfd12e98119f/protobuf-5.29.4-cp310-abi3-win_amd64.whl", hash = "sha256:bcefcdf3976233f8a502d265eb65ea740c989bacc6c30a58290ed0e519eb4b8d", size = 434506, upload-time = "2025-03-19T21:23:11.253Z" }, - { url = "https://files.pythonhosted.org/packages/46/de/7c126bbb06aa0f8a7b38aaf8bd746c514d70e6a2a3f6dd460b3b7aad7aae/protobuf-5.29.4-cp38-abi3-macosx_10_9_universal2.whl", hash = "sha256:307ecba1d852ec237e9ba668e087326a67564ef83e45a0189a772ede9e854dd0", size = 417826, upload-time = "2025-03-19T21:23:13.132Z" }, - { url = "https://files.pythonhosted.org/packages/a2/b5/bade14ae31ba871a139aa45e7a8183d869efe87c34a4850c87b936963261/protobuf-5.29.4-cp38-abi3-manylinux2014_aarch64.whl", hash = "sha256:aec4962f9ea93c431d5714ed1be1c93f13e1a8618e70035ba2b0564d9e633f2e", size = 319574, upload-time = "2025-03-19T21:23:14.531Z" }, - { url = "https://files.pythonhosted.org/packages/46/88/b01ed2291aae68b708f7d334288ad5fb3e7aa769a9c309c91a0d55cb91b0/protobuf-5.29.4-cp38-abi3-manylinux2014_x86_64.whl", hash = "sha256:d7d3f7d1d5a66ed4942d4fefb12ac4b14a29028b209d4bfb25c68ae172059922", size = 319672, upload-time = "2025-03-19T21:23:15.839Z" }, - { url = "https://files.pythonhosted.org/packages/12/fb/a586e0c973c95502e054ac5f81f88394f24ccc7982dac19c515acd9e2c93/protobuf-5.29.4-py3-none-any.whl", hash = "sha256:3fde11b505e1597f71b875ef2fc52062b6a9740e5f7c8997ce878b6009145862", size = 172551, upload-time = "2025-03-19T21:23:22.682Z" }, + { url = "https://files.pythonhosted.org/packages/f3/6f/6ab8e4bf962fd5570d3deaa2d5c38f0a363f57b4501047b5ebeb83ab1125/protobuf-6.31.1-cp310-abi3-win32.whl", hash = "sha256:7fa17d5a29c2e04b7d90e5e32388b8bfd0e7107cd8e616feef7ed3fa6bdab5c9", size = 423603, upload-time = "2025-05-28T19:25:41.198Z" }, + { url = "https://files.pythonhosted.org/packages/44/3a/b15c4347dd4bf3a1b0ee882f384623e2063bb5cf9fa9d57990a4f7df2fb6/protobuf-6.31.1-cp310-abi3-win_amd64.whl", hash = "sha256:426f59d2964864a1a366254fa703b8632dcec0790d8862d30034d8245e1cd447", size = 435283, upload-time = "2025-05-28T19:25:44.275Z" }, + { url = "https://files.pythonhosted.org/packages/6a/c9/b9689a2a250264a84e66c46d8862ba788ee7a641cdca39bccf64f59284b7/protobuf-6.31.1-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:6f1227473dc43d44ed644425268eb7c2e488ae245d51c6866d19fe158e207402", size = 425604, upload-time = "2025-05-28T19:25:45.702Z" }, + { url = "https://files.pythonhosted.org/packages/76/a1/7a5a94032c83375e4fe7e7f56e3976ea6ac90c5e85fac8576409e25c39c3/protobuf-6.31.1-cp39-abi3-manylinux2014_aarch64.whl", hash = "sha256:a40fc12b84c154884d7d4c4ebd675d5b3b5283e155f324049ae396b95ddebc39", size = 322115, upload-time = "2025-05-28T19:25:47.128Z" }, + { url = "https://files.pythonhosted.org/packages/fa/b1/b59d405d64d31999244643d88c45c8241c58f17cc887e73bcb90602327f8/protobuf-6.31.1-cp39-abi3-manylinux2014_x86_64.whl", hash = "sha256:4ee898bf66f7a8b0bd21bce523814e6fbd8c6add948045ce958b73af7e8878c6", size = 321070, upload-time = "2025-05-28T19:25:50.036Z" }, + { url = "https://files.pythonhosted.org/packages/f7/af/ab3c51ab7507a7325e98ffe691d9495ee3d3aa5f589afad65ec920d39821/protobuf-6.31.1-py3-none-any.whl", hash = "sha256:720a6c7e6b77288b85063569baae8536671b39f15cc22037ec7045658d80489e", size = 168724, upload-time = "2025-05-28T19:25:53.926Z" }, ] [[package]] @@ -2254,11 +2873,32 @@ wheels = [ [[package]] name = "py3dmol" -version = "2.4.2" +version = "2.5.2" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/8d/ba/78bc5b451f314c06e6c2a3ca0f0ba18ee751f10e99fed94fc09175b16031/py3Dmol-2.4.2.tar.gz", hash = "sha256:990ed67b2dda5493d21192fef53a52be6128b8afbdc13da8a40a009060120749", size = 7724, upload-time = "2024-11-08T22:19:23.15Z" } +sdist = { url = "https://files.pythonhosted.org/packages/33/e9/fa4fce24843554d4e025fcdb2fcebfd0ec14db88a4a57545eb38b15825be/py3dmol-2.5.2.tar.gz", hash = "sha256:9ef1c72c786b22e33541e05b27e97bb99ea1d3c962819ed55c2203d10a515198", size = 7857, upload-time = "2025-07-31T19:35:41.833Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/04/20/923885064f4e4d4392eb2be798532d91b315f9e60ef44f49f4800ba3c57a/py3Dmol-2.4.2-py2.py3-none-any.whl", hash = "sha256:bec23d9a015d692279a5f7d4db92803e4e82ba3bdcc1434a5b6a2be98a347856", size = 7046, upload-time = "2024-11-08T22:19:21.631Z" }, + { url = "https://files.pythonhosted.org/packages/a3/17/c476487ba903c7d793db633b4e8ca4a420ae272890302189d9402ba8ff85/py3dmol-2.5.2-py2.py3-none-any.whl", hash = "sha256:b921940ff046cf7ca008a249cbd5debec561dcf337f1e5f3df7ac5d4a1954e8e", size = 7154, upload-time = "2025-07-31T19:35:41.022Z" }, +] + +[[package]] +name = "pyasn1" +version = "0.6.1" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/ba/e9/01f1a64245b89f039897cb0130016d79f77d52669aae6ee7b159a6c4c018/pyasn1-0.6.1.tar.gz", hash = "sha256:6f580d2bdd84365380830acf45550f2511469f673cb4a5ae3857a3170128b034", size = 145322, upload-time = "2024-09-10T22:41:42.55Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c8/f1/d6a797abb14f6283c0ddff96bbdd46937f64122b8c925cab503dd37f8214/pyasn1-0.6.1-py3-none-any.whl", hash = "sha256:0d632f46f2ba09143da3a8afe9e33fb6f92fa2320ab7e886e2d0f7672af84629", size = 83135, upload-time = "2024-09-11T16:00:36.122Z" }, +] + +[[package]] +name = "pyasn1-modules" +version = "0.4.2" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "pyasn1" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/e9/e6/78ebbb10a8c8e4b61a59249394a4a594c1a7af95593dc933a349c8d00964/pyasn1_modules-0.4.2.tar.gz", hash = "sha256:677091de870a80aae844b1ca6134f54652fa2c8c5a52aa396440ac3106e941e6", size = 307892, upload-time = "2025-03-28T02:41:22.17Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/47/8d/d529b5d697919ba8c11ad626e835d4039be708a35b0d22de83a269a6682c/pyasn1_modules-0.4.2-py3-none-any.whl", hash = "sha256:29253a9207ce32b64c3ac6600edc75368f98473906e8fd1043bd6b5b1de2c14a", size = 181259, upload-time = "2025-03-28T02:41:19.028Z" }, ] [[package]] @@ -2272,114 +2912,127 @@ wheels = [ [[package]] name = "pydantic" -version = "2.10.6" +version = "2.11.7" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "annotated-types" }, { name = "pydantic-core" }, { name = "typing-extensions" }, + { name = "typing-inspection" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/b7/ae/d5220c5c52b158b1de7ca89fc5edb72f304a70a4c540c84c8844bf4008de/pydantic-2.10.6.tar.gz", hash = "sha256:ca5daa827cce33de7a42be142548b0096bf05a7e7b365aebfa5f8eeec7128236", size = 761681, upload-time = "2025-01-24T01:42:12.693Z" } +sdist = { url = "https://files.pythonhosted.org/packages/00/dd/4325abf92c39ba8623b5af936ddb36ffcfe0beae70405d456ab1fb2f5b8c/pydantic-2.11.7.tar.gz", hash = "sha256:d989c3c6cb79469287b1569f7447a17848c998458d49ebe294e975b9baf0f0db", size = 788350, upload-time = "2025-06-14T08:33:17.137Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/f4/3c/8cc1cc84deffa6e25d2d0c688ebb80635dfdbf1dbea3e30c541c8cf4d860/pydantic-2.10.6-py3-none-any.whl", hash = "sha256:427d664bf0b8a2b34ff5dd0f5a18df00591adcee7198fbd71981054cef37b584", size = 431696, upload-time = "2025-01-24T01:42:10.371Z" }, + { url = "https://files.pythonhosted.org/packages/6a/c0/ec2b1c8712ca690e5d61979dee872603e92b8a32f94cc1b72d53beab008a/pydantic-2.11.7-py3-none-any.whl", hash = "sha256:dde5df002701f6de26248661f6835bbe296a47bf73990135c7d07ce741b9623b", size = 444782, upload-time = "2025-06-14T08:33:14.905Z" }, ] [[package]] name = "pydantic-core" -version = "2.27.2" +version = "2.33.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/fc/01/f3e5ac5e7c25833db5eb555f7b7ab24cd6f8c322d3a3ad2d67a952dc0abc/pydantic_core-2.27.2.tar.gz", hash = "sha256:eb026e5a4c1fee05726072337ff51d1efb6f59090b7da90d30ea58625b1ffb39", size = 413443, upload-time = "2024-12-18T11:31:54.917Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/3a/bc/fed5f74b5d802cf9a03e83f60f18864e90e3aed7223adaca5ffb7a8d8d64/pydantic_core-2.27.2-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:2d367ca20b2f14095a8f4fa1210f5a7b78b8a20009ecced6b12818f455b1e9fa", size = 1895938, upload-time = "2024-12-18T11:27:14.406Z" }, - { url = "https://files.pythonhosted.org/packages/71/2a/185aff24ce844e39abb8dd680f4e959f0006944f4a8a0ea372d9f9ae2e53/pydantic_core-2.27.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:491a2b73db93fab69731eaee494f320faa4e093dbed776be1a829c2eb222c34c", size = 1815684, upload-time = "2024-12-18T11:27:16.489Z" }, - { url = "https://files.pythonhosted.org/packages/c3/43/fafabd3d94d159d4f1ed62e383e264f146a17dd4d48453319fd782e7979e/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7969e133a6f183be60e9f6f56bfae753585680f3b7307a8e555a948d443cc05a", size = 1829169, upload-time = "2024-12-18T11:27:22.16Z" }, - { url = "https://files.pythonhosted.org/packages/a2/d1/f2dfe1a2a637ce6800b799aa086d079998959f6f1215eb4497966efd2274/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:3de9961f2a346257caf0aa508a4da705467f53778e9ef6fe744c038119737ef5", size = 1867227, upload-time = "2024-12-18T11:27:25.097Z" }, - { url = "https://files.pythonhosted.org/packages/7d/39/e06fcbcc1c785daa3160ccf6c1c38fea31f5754b756e34b65f74e99780b5/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e2bb4d3e5873c37bb3dd58714d4cd0b0e6238cebc4177ac8fe878f8b3aa8e74c", size = 2037695, upload-time = "2024-12-18T11:27:28.656Z" }, - { url = "https://files.pythonhosted.org/packages/7a/67/61291ee98e07f0650eb756d44998214231f50751ba7e13f4f325d95249ab/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:280d219beebb0752699480fe8f1dc61ab6615c2046d76b7ab7ee38858de0a4e7", size = 2741662, upload-time = "2024-12-18T11:27:30.798Z" }, - { url = "https://files.pythonhosted.org/packages/32/90/3b15e31b88ca39e9e626630b4c4a1f5a0dfd09076366f4219429e6786076/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:47956ae78b6422cbd46f772f1746799cbb862de838fd8d1fbd34a82e05b0983a", size = 1993370, upload-time = "2024-12-18T11:27:33.692Z" }, - { url = "https://files.pythonhosted.org/packages/ff/83/c06d333ee3a67e2e13e07794995c1535565132940715931c1c43bfc85b11/pydantic_core-2.27.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:14d4a5c49d2f009d62a2a7140d3064f686d17a5d1a268bc641954ba181880236", size = 1996813, upload-time = "2024-12-18T11:27:37.111Z" }, - { url = "https://files.pythonhosted.org/packages/7c/f7/89be1c8deb6e22618a74f0ca0d933fdcb8baa254753b26b25ad3acff8f74/pydantic_core-2.27.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:337b443af21d488716f8d0b6164de833e788aa6bd7e3a39c005febc1284f4962", size = 2005287, upload-time = "2024-12-18T11:27:40.566Z" }, - { url = "https://files.pythonhosted.org/packages/b7/7d/8eb3e23206c00ef7feee17b83a4ffa0a623eb1a9d382e56e4aa46fd15ff2/pydantic_core-2.27.2-cp310-cp310-musllinux_1_1_armv7l.whl", hash = "sha256:03d0f86ea3184a12f41a2d23f7ccb79cdb5a18e06993f8a45baa8dfec746f0e9", size = 2128414, upload-time = "2024-12-18T11:27:43.757Z" }, - { url = "https://files.pythonhosted.org/packages/4e/99/fe80f3ff8dd71a3ea15763878d464476e6cb0a2db95ff1c5c554133b6b83/pydantic_core-2.27.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:7041c36f5680c6e0f08d922aed302e98b3745d97fe1589db0a3eebf6624523af", size = 2155301, upload-time = "2024-12-18T11:27:47.36Z" }, - { url = "https://files.pythonhosted.org/packages/2b/a3/e50460b9a5789ca1451b70d4f52546fa9e2b420ba3bfa6100105c0559238/pydantic_core-2.27.2-cp310-cp310-win32.whl", hash = "sha256:50a68f3e3819077be2c98110c1f9dcb3817e93f267ba80a2c05bb4f8799e2ff4", size = 1816685, upload-time = "2024-12-18T11:27:50.508Z" }, - { url = "https://files.pythonhosted.org/packages/57/4c/a8838731cb0f2c2a39d3535376466de6049034d7b239c0202a64aaa05533/pydantic_core-2.27.2-cp310-cp310-win_amd64.whl", hash = "sha256:e0fd26b16394ead34a424eecf8a31a1f5137094cabe84a1bcb10fa6ba39d3d31", size = 1982876, upload-time = "2024-12-18T11:27:53.54Z" }, - { url = "https://files.pythonhosted.org/packages/c2/89/f3450af9d09d44eea1f2c369f49e8f181d742f28220f88cc4dfaae91ea6e/pydantic_core-2.27.2-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:8e10c99ef58cfdf2a66fc15d66b16c4a04f62bca39db589ae8cba08bc55331bc", size = 1893421, upload-time = "2024-12-18T11:27:55.409Z" }, - { url = "https://files.pythonhosted.org/packages/9e/e3/71fe85af2021f3f386da42d291412e5baf6ce7716bd7101ea49c810eda90/pydantic_core-2.27.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:26f32e0adf166a84d0cb63be85c562ca8a6fa8de28e5f0d92250c6b7e9e2aff7", size = 1814998, upload-time = "2024-12-18T11:27:57.252Z" }, - { url = "https://files.pythonhosted.org/packages/a6/3c/724039e0d848fd69dbf5806894e26479577316c6f0f112bacaf67aa889ac/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8c19d1ea0673cd13cc2f872f6c9ab42acc4e4f492a7ca9d3795ce2b112dd7e15", size = 1826167, upload-time = "2024-12-18T11:27:59.146Z" }, - { url = "https://files.pythonhosted.org/packages/2b/5b/1b29e8c1fb5f3199a9a57c1452004ff39f494bbe9bdbe9a81e18172e40d3/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:5e68c4446fe0810e959cdff46ab0a41ce2f2c86d227d96dc3847af0ba7def306", size = 1865071, upload-time = "2024-12-18T11:28:02.625Z" }, - { url = "https://files.pythonhosted.org/packages/89/6c/3985203863d76bb7d7266e36970d7e3b6385148c18a68cc8915fd8c84d57/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d9640b0059ff4f14d1f37321b94061c6db164fbe49b334b31643e0528d100d99", size = 2036244, upload-time = "2024-12-18T11:28:04.442Z" }, - { url = "https://files.pythonhosted.org/packages/0e/41/f15316858a246b5d723f7d7f599f79e37493b2e84bfc789e58d88c209f8a/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:40d02e7d45c9f8af700f3452f329ead92da4c5f4317ca9b896de7ce7199ea459", size = 2737470, upload-time = "2024-12-18T11:28:07.679Z" }, - { url = "https://files.pythonhosted.org/packages/a8/7c/b860618c25678bbd6d1d99dbdfdf0510ccb50790099b963ff78a124b754f/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1c1fd185014191700554795c99b347d64f2bb637966c4cfc16998a0ca700d048", size = 1992291, upload-time = "2024-12-18T11:28:10.297Z" }, - { url = "https://files.pythonhosted.org/packages/bf/73/42c3742a391eccbeab39f15213ecda3104ae8682ba3c0c28069fbcb8c10d/pydantic_core-2.27.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:d81d2068e1c1228a565af076598f9e7451712700b673de8f502f0334f281387d", size = 1994613, upload-time = "2024-12-18T11:28:13.362Z" }, - { url = "https://files.pythonhosted.org/packages/94/7a/941e89096d1175d56f59340f3a8ebaf20762fef222c298ea96d36a6328c5/pydantic_core-2.27.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:1a4207639fb02ec2dbb76227d7c751a20b1a6b4bc52850568e52260cae64ca3b", size = 2002355, upload-time = "2024-12-18T11:28:16.587Z" }, - { url = "https://files.pythonhosted.org/packages/6e/95/2359937a73d49e336a5a19848713555605d4d8d6940c3ec6c6c0ca4dcf25/pydantic_core-2.27.2-cp311-cp311-musllinux_1_1_armv7l.whl", hash = "sha256:3de3ce3c9ddc8bbd88f6e0e304dea0e66d843ec9de1b0042b0911c1663ffd474", size = 2126661, upload-time = "2024-12-18T11:28:18.407Z" }, - { url = "https://files.pythonhosted.org/packages/2b/4c/ca02b7bdb6012a1adef21a50625b14f43ed4d11f1fc237f9d7490aa5078c/pydantic_core-2.27.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:30c5f68ded0c36466acede341551106821043e9afaad516adfb6e8fa80a4e6a6", size = 2153261, upload-time = "2024-12-18T11:28:21.471Z" }, - { url = "https://files.pythonhosted.org/packages/72/9d/a241db83f973049a1092a079272ffe2e3e82e98561ef6214ab53fe53b1c7/pydantic_core-2.27.2-cp311-cp311-win32.whl", hash = "sha256:c70c26d2c99f78b125a3459f8afe1aed4d9687c24fd677c6a4436bc042e50d6c", size = 1812361, upload-time = "2024-12-18T11:28:23.53Z" }, - { url = "https://files.pythonhosted.org/packages/e8/ef/013f07248041b74abd48a385e2110aa3a9bbfef0fbd97d4e6d07d2f5b89a/pydantic_core-2.27.2-cp311-cp311-win_amd64.whl", hash = "sha256:08e125dbdc505fa69ca7d9c499639ab6407cfa909214d500897d02afb816e7cc", size = 1982484, upload-time = "2024-12-18T11:28:25.391Z" }, - { url = "https://files.pythonhosted.org/packages/10/1c/16b3a3e3398fd29dca77cea0a1d998d6bde3902fa2706985191e2313cc76/pydantic_core-2.27.2-cp311-cp311-win_arm64.whl", hash = "sha256:26f0d68d4b235a2bae0c3fc585c585b4ecc51382db0e3ba402a22cbc440915e4", size = 1867102, upload-time = "2024-12-18T11:28:28.593Z" }, - { url = "https://files.pythonhosted.org/packages/d6/74/51c8a5482ca447871c93e142d9d4a92ead74de6c8dc5e66733e22c9bba89/pydantic_core-2.27.2-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:9e0c8cfefa0ef83b4da9588448b6d8d2a2bf1a53c3f1ae5fca39eb3061e2f0b0", size = 1893127, upload-time = "2024-12-18T11:28:30.346Z" }, - { url = "https://files.pythonhosted.org/packages/d3/f3/c97e80721735868313c58b89d2de85fa80fe8dfeeed84dc51598b92a135e/pydantic_core-2.27.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:83097677b8e3bd7eaa6775720ec8e0405f1575015a463285a92bfdfe254529ef", size = 1811340, upload-time = "2024-12-18T11:28:32.521Z" }, - { url = "https://files.pythonhosted.org/packages/9e/91/840ec1375e686dbae1bd80a9e46c26a1e0083e1186abc610efa3d9a36180/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:172fce187655fece0c90d90a678424b013f8fbb0ca8b036ac266749c09438cb7", size = 1822900, upload-time = "2024-12-18T11:28:34.507Z" }, - { url = "https://files.pythonhosted.org/packages/f6/31/4240bc96025035500c18adc149aa6ffdf1a0062a4b525c932065ceb4d868/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:519f29f5213271eeeeb3093f662ba2fd512b91c5f188f3bb7b27bc5973816934", size = 1869177, upload-time = "2024-12-18T11:28:36.488Z" }, - { url = "https://files.pythonhosted.org/packages/fa/20/02fbaadb7808be578317015c462655c317a77a7c8f0ef274bc016a784c54/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:05e3a55d124407fffba0dd6b0c0cd056d10e983ceb4e5dbd10dda135c31071d6", size = 2038046, upload-time = "2024-12-18T11:28:39.409Z" }, - { url = "https://files.pythonhosted.org/packages/06/86/7f306b904e6c9eccf0668248b3f272090e49c275bc488a7b88b0823444a4/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9c3ed807c7b91de05e63930188f19e921d1fe90de6b4f5cd43ee7fcc3525cb8c", size = 2685386, upload-time = "2024-12-18T11:28:41.221Z" }, - { url = "https://files.pythonhosted.org/packages/8d/f0/49129b27c43396581a635d8710dae54a791b17dfc50c70164866bbf865e3/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6fb4aadc0b9a0c063206846d603b92030eb6f03069151a625667f982887153e2", size = 1997060, upload-time = "2024-12-18T11:28:44.709Z" }, - { url = "https://files.pythonhosted.org/packages/0d/0f/943b4af7cd416c477fd40b187036c4f89b416a33d3cc0ab7b82708a667aa/pydantic_core-2.27.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:28ccb213807e037460326424ceb8b5245acb88f32f3d2777427476e1b32c48c4", size = 2004870, upload-time = "2024-12-18T11:28:46.839Z" }, - { url = "https://files.pythonhosted.org/packages/35/40/aea70b5b1a63911c53a4c8117c0a828d6790483f858041f47bab0b779f44/pydantic_core-2.27.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:de3cd1899e2c279b140adde9357c4495ed9d47131b4a4eaff9052f23398076b3", size = 1999822, upload-time = "2024-12-18T11:28:48.896Z" }, - { url = "https://files.pythonhosted.org/packages/f2/b3/807b94fd337d58effc5498fd1a7a4d9d59af4133e83e32ae39a96fddec9d/pydantic_core-2.27.2-cp312-cp312-musllinux_1_1_armv7l.whl", hash = "sha256:220f892729375e2d736b97d0e51466252ad84c51857d4d15f5e9692f9ef12be4", size = 2130364, upload-time = "2024-12-18T11:28:50.755Z" }, - { url = "https://files.pythonhosted.org/packages/fc/df/791c827cd4ee6efd59248dca9369fb35e80a9484462c33c6649a8d02b565/pydantic_core-2.27.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:a0fcd29cd6b4e74fe8ddd2c90330fd8edf2e30cb52acda47f06dd615ae72da57", size = 2158303, upload-time = "2024-12-18T11:28:54.122Z" }, - { url = "https://files.pythonhosted.org/packages/9b/67/4e197c300976af185b7cef4c02203e175fb127e414125916bf1128b639a9/pydantic_core-2.27.2-cp312-cp312-win32.whl", hash = "sha256:1e2cb691ed9834cd6a8be61228471d0a503731abfb42f82458ff27be7b2186fc", size = 1834064, upload-time = "2024-12-18T11:28:56.074Z" }, - { url = "https://files.pythonhosted.org/packages/1f/ea/cd7209a889163b8dcca139fe32b9687dd05249161a3edda62860430457a5/pydantic_core-2.27.2-cp312-cp312-win_amd64.whl", hash = "sha256:cc3f1a99a4f4f9dd1de4fe0312c114e740b5ddead65bb4102884b384c15d8bc9", size = 1989046, upload-time = "2024-12-18T11:28:58.107Z" }, - { url = "https://files.pythonhosted.org/packages/bc/49/c54baab2f4658c26ac633d798dab66b4c3a9bbf47cff5284e9c182f4137a/pydantic_core-2.27.2-cp312-cp312-win_arm64.whl", hash = "sha256:3911ac9284cd8a1792d3cb26a2da18f3ca26c6908cc434a18f730dc0db7bfa3b", size = 1885092, upload-time = "2024-12-18T11:29:01.335Z" }, - { url = "https://files.pythonhosted.org/packages/41/b1/9bc383f48f8002f99104e3acff6cba1231b29ef76cfa45d1506a5cad1f84/pydantic_core-2.27.2-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:7d14bd329640e63852364c306f4d23eb744e0f8193148d4044dd3dacdaacbd8b", size = 1892709, upload-time = "2024-12-18T11:29:03.193Z" }, - { url = "https://files.pythonhosted.org/packages/10/6c/e62b8657b834f3eb2961b49ec8e301eb99946245e70bf42c8817350cbefc/pydantic_core-2.27.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:82f91663004eb8ed30ff478d77c4d1179b3563df6cdb15c0817cd1cdaf34d154", size = 1811273, upload-time = "2024-12-18T11:29:05.306Z" }, - { url = "https://files.pythonhosted.org/packages/ba/15/52cfe49c8c986e081b863b102d6b859d9defc63446b642ccbbb3742bf371/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:71b24c7d61131bb83df10cc7e687433609963a944ccf45190cfc21e0887b08c9", size = 1823027, upload-time = "2024-12-18T11:29:07.294Z" }, - { url = "https://files.pythonhosted.org/packages/b1/1c/b6f402cfc18ec0024120602bdbcebc7bdd5b856528c013bd4d13865ca473/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:fa8e459d4954f608fa26116118bb67f56b93b209c39b008277ace29937453dc9", size = 1868888, upload-time = "2024-12-18T11:29:09.249Z" }, - { url = "https://files.pythonhosted.org/packages/bd/7b/8cb75b66ac37bc2975a3b7de99f3c6f355fcc4d89820b61dffa8f1e81677/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ce8918cbebc8da707ba805b7fd0b382816858728ae7fe19a942080c24e5b7cd1", size = 2037738, upload-time = "2024-12-18T11:29:11.23Z" }, - { url = "https://files.pythonhosted.org/packages/c8/f1/786d8fe78970a06f61df22cba58e365ce304bf9b9f46cc71c8c424e0c334/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:eda3f5c2a021bbc5d976107bb302e0131351c2ba54343f8a496dc8783d3d3a6a", size = 2685138, upload-time = "2024-12-18T11:29:16.396Z" }, - { url = "https://files.pythonhosted.org/packages/a6/74/d12b2cd841d8724dc8ffb13fc5cef86566a53ed358103150209ecd5d1999/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bd8086fa684c4775c27f03f062cbb9eaa6e17f064307e86b21b9e0abc9c0f02e", size = 1997025, upload-time = "2024-12-18T11:29:20.25Z" }, - { url = "https://files.pythonhosted.org/packages/a0/6e/940bcd631bc4d9a06c9539b51f070b66e8f370ed0933f392db6ff350d873/pydantic_core-2.27.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:8d9b3388db186ba0c099a6d20f0604a44eabdeef1777ddd94786cdae158729e4", size = 2004633, upload-time = "2024-12-18T11:29:23.877Z" }, - { url = "https://files.pythonhosted.org/packages/50/cc/a46b34f1708d82498c227d5d80ce615b2dd502ddcfd8376fc14a36655af1/pydantic_core-2.27.2-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:7a66efda2387de898c8f38c0cf7f14fca0b51a8ef0b24bfea5849f1b3c95af27", size = 1999404, upload-time = "2024-12-18T11:29:25.872Z" }, - { url = "https://files.pythonhosted.org/packages/ca/2d/c365cfa930ed23bc58c41463bae347d1005537dc8db79e998af8ba28d35e/pydantic_core-2.27.2-cp313-cp313-musllinux_1_1_armv7l.whl", hash = "sha256:18a101c168e4e092ab40dbc2503bdc0f62010e95d292b27827871dc85450d7ee", size = 2130130, upload-time = "2024-12-18T11:29:29.252Z" }, - { url = "https://files.pythonhosted.org/packages/f4/d7/eb64d015c350b7cdb371145b54d96c919d4db516817f31cd1c650cae3b21/pydantic_core-2.27.2-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:ba5dd002f88b78a4215ed2f8ddbdf85e8513382820ba15ad5ad8955ce0ca19a1", size = 2157946, upload-time = "2024-12-18T11:29:31.338Z" }, - { url = "https://files.pythonhosted.org/packages/a4/99/bddde3ddde76c03b65dfd5a66ab436c4e58ffc42927d4ff1198ffbf96f5f/pydantic_core-2.27.2-cp313-cp313-win32.whl", hash = "sha256:1ebaf1d0481914d004a573394f4be3a7616334be70261007e47c2a6fe7e50130", size = 1834387, upload-time = "2024-12-18T11:29:33.481Z" }, - { url = "https://files.pythonhosted.org/packages/71/47/82b5e846e01b26ac6f1893d3c5f9f3a2eb6ba79be26eef0b759b4fe72946/pydantic_core-2.27.2-cp313-cp313-win_amd64.whl", hash = "sha256:953101387ecf2f5652883208769a79e48db18c6df442568a0b5ccd8c2723abee", size = 1990453, upload-time = "2024-12-18T11:29:35.533Z" }, - { url = "https://files.pythonhosted.org/packages/51/b2/b2b50d5ecf21acf870190ae5d093602d95f66c9c31f9d5de6062eb329ad1/pydantic_core-2.27.2-cp313-cp313-win_arm64.whl", hash = "sha256:ac4dbfd1691affb8f48c2c13241a2e3b60ff23247cbcf981759c768b6633cf8b", size = 1885186, upload-time = "2024-12-18T11:29:37.649Z" }, - { url = "https://files.pythonhosted.org/packages/46/72/af70981a341500419e67d5cb45abe552a7c74b66326ac8877588488da1ac/pydantic_core-2.27.2-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:2bf14caea37e91198329b828eae1618c068dfb8ef17bb33287a7ad4b61ac314e", size = 1891159, upload-time = "2024-12-18T11:30:54.382Z" }, - { url = "https://files.pythonhosted.org/packages/ad/3d/c5913cccdef93e0a6a95c2d057d2c2cba347815c845cda79ddd3c0f5e17d/pydantic_core-2.27.2-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:b0cb791f5b45307caae8810c2023a184c74605ec3bcbb67d13846c28ff731ff8", size = 1768331, upload-time = "2024-12-18T11:30:58.178Z" }, - { url = "https://files.pythonhosted.org/packages/f6/f0/a3ae8fbee269e4934f14e2e0e00928f9346c5943174f2811193113e58252/pydantic_core-2.27.2-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:688d3fd9fcb71f41c4c015c023d12a79d1c4c0732ec9eb35d96e3388a120dcf3", size = 1822467, upload-time = "2024-12-18T11:31:00.6Z" }, - { url = "https://files.pythonhosted.org/packages/d7/7a/7bbf241a04e9f9ea24cd5874354a83526d639b02674648af3f350554276c/pydantic_core-2.27.2-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3d591580c34f4d731592f0e9fe40f9cc1b430d297eecc70b962e93c5c668f15f", size = 1979797, upload-time = "2024-12-18T11:31:07.243Z" }, - { url = "https://files.pythonhosted.org/packages/4f/5f/4784c6107731f89e0005a92ecb8a2efeafdb55eb992b8e9d0a2be5199335/pydantic_core-2.27.2-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:82f986faf4e644ffc189a7f1aafc86e46ef70372bb153e7001e8afccc6e54133", size = 1987839, upload-time = "2024-12-18T11:31:09.775Z" }, - { url = "https://files.pythonhosted.org/packages/6d/a7/61246562b651dff00de86a5f01b6e4befb518df314c54dec187a78d81c84/pydantic_core-2.27.2-pp310-pypy310_pp73-musllinux_1_1_aarch64.whl", hash = "sha256:bec317a27290e2537f922639cafd54990551725fc844249e64c523301d0822fc", size = 1998861, upload-time = "2024-12-18T11:31:13.469Z" }, - { url = "https://files.pythonhosted.org/packages/86/aa/837821ecf0c022bbb74ca132e117c358321e72e7f9702d1b6a03758545e2/pydantic_core-2.27.2-pp310-pypy310_pp73-musllinux_1_1_armv7l.whl", hash = "sha256:0296abcb83a797db256b773f45773da397da75a08f5fcaef41f2044adec05f50", size = 2116582, upload-time = "2024-12-18T11:31:17.423Z" }, - { url = "https://files.pythonhosted.org/packages/81/b0/5e74656e95623cbaa0a6278d16cf15e10a51f6002e3ec126541e95c29ea3/pydantic_core-2.27.2-pp310-pypy310_pp73-musllinux_1_1_x86_64.whl", hash = "sha256:0d75070718e369e452075a6017fbf187f788e17ed67a3abd47fa934d001863d9", size = 2151985, upload-time = "2024-12-18T11:31:19.901Z" }, - { url = "https://files.pythonhosted.org/packages/63/37/3e32eeb2a451fddaa3898e2163746b0cffbbdbb4740d38372db0490d67f3/pydantic_core-2.27.2-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:7e17b560be3c98a8e3aa66ce828bdebb9e9ac6ad5466fba92eb74c4c95cb1151", size = 2004715, upload-time = "2024-12-18T11:31:22.821Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/ad/88/5f2260bdfae97aabf98f1778d43f69574390ad787afb646292a638c923d4/pydantic_core-2.33.2.tar.gz", hash = "sha256:7cb8bc3605c29176e1b105350d2e6474142d7c1bd1d9327c4a9bdb46bf827acc", size = 435195, upload-time = "2025-04-23T18:33:52.104Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/e5/92/b31726561b5dae176c2d2c2dc43a9c5bfba5d32f96f8b4c0a600dd492447/pydantic_core-2.33.2-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:2b3d326aaef0c0399d9afffeb6367d5e26ddc24d351dbc9c636840ac355dc5d8", size = 2028817, upload-time = "2025-04-23T18:30:43.919Z" }, + { url = "https://files.pythonhosted.org/packages/a3/44/3f0b95fafdaca04a483c4e685fe437c6891001bf3ce8b2fded82b9ea3aa1/pydantic_core-2.33.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:0e5b2671f05ba48b94cb90ce55d8bdcaaedb8ba00cc5359f6810fc918713983d", size = 1861357, upload-time = "2025-04-23T18:30:46.372Z" }, + { url = "https://files.pythonhosted.org/packages/30/97/e8f13b55766234caae05372826e8e4b3b96e7b248be3157f53237682e43c/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0069c9acc3f3981b9ff4cdfaf088e98d83440a4c7ea1bc07460af3d4dc22e72d", size = 1898011, upload-time = "2025-04-23T18:30:47.591Z" }, + { url = "https://files.pythonhosted.org/packages/9b/a3/99c48cf7bafc991cc3ee66fd544c0aae8dc907b752f1dad2d79b1b5a471f/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:d53b22f2032c42eaaf025f7c40c2e3b94568ae077a606f006d206a463bc69572", size = 1982730, upload-time = "2025-04-23T18:30:49.328Z" }, + { url = "https://files.pythonhosted.org/packages/de/8e/a5b882ec4307010a840fb8b58bd9bf65d1840c92eae7534c7441709bf54b/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0405262705a123b7ce9f0b92f123334d67b70fd1f20a9372b907ce1080c7ba02", size = 2136178, upload-time = "2025-04-23T18:30:50.907Z" }, + { url = "https://files.pythonhosted.org/packages/e4/bb/71e35fc3ed05af6834e890edb75968e2802fe98778971ab5cba20a162315/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4b25d91e288e2c4e0662b8038a28c6a07eaac3e196cfc4ff69de4ea3db992a1b", size = 2736462, upload-time = "2025-04-23T18:30:52.083Z" }, + { url = "https://files.pythonhosted.org/packages/31/0d/c8f7593e6bc7066289bbc366f2235701dcbebcd1ff0ef8e64f6f239fb47d/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6bdfe4b3789761f3bcb4b1ddf33355a71079858958e3a552f16d5af19768fef2", size = 2005652, upload-time = "2025-04-23T18:30:53.389Z" }, + { url = "https://files.pythonhosted.org/packages/d2/7a/996d8bd75f3eda405e3dd219ff5ff0a283cd8e34add39d8ef9157e722867/pydantic_core-2.33.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:efec8db3266b76ef9607c2c4c419bdb06bf335ae433b80816089ea7585816f6a", size = 2113306, upload-time = "2025-04-23T18:30:54.661Z" }, + { url = "https://files.pythonhosted.org/packages/ff/84/daf2a6fb2db40ffda6578a7e8c5a6e9c8affb251a05c233ae37098118788/pydantic_core-2.33.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:031c57d67ca86902726e0fae2214ce6770bbe2f710dc33063187a68744a5ecac", size = 2073720, upload-time = "2025-04-23T18:30:56.11Z" }, + { url = "https://files.pythonhosted.org/packages/77/fb/2258da019f4825128445ae79456a5499c032b55849dbd5bed78c95ccf163/pydantic_core-2.33.2-cp310-cp310-musllinux_1_1_armv7l.whl", hash = "sha256:f8de619080e944347f5f20de29a975c2d815d9ddd8be9b9b7268e2e3ef68605a", size = 2244915, upload-time = "2025-04-23T18:30:57.501Z" }, + { url = "https://files.pythonhosted.org/packages/d8/7a/925ff73756031289468326e355b6fa8316960d0d65f8b5d6b3a3e7866de7/pydantic_core-2.33.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:73662edf539e72a9440129f231ed3757faab89630d291b784ca99237fb94db2b", size = 2241884, upload-time = "2025-04-23T18:30:58.867Z" }, + { url = "https://files.pythonhosted.org/packages/0b/b0/249ee6d2646f1cdadcb813805fe76265745c4010cf20a8eba7b0e639d9b2/pydantic_core-2.33.2-cp310-cp310-win32.whl", hash = "sha256:0a39979dcbb70998b0e505fb1556a1d550a0781463ce84ebf915ba293ccb7e22", size = 1910496, upload-time = "2025-04-23T18:31:00.078Z" }, + { url = "https://files.pythonhosted.org/packages/66/ff/172ba8f12a42d4b552917aa65d1f2328990d3ccfc01d5b7c943ec084299f/pydantic_core-2.33.2-cp310-cp310-win_amd64.whl", hash = "sha256:b0379a2b24882fef529ec3b4987cb5d003b9cda32256024e6fe1586ac45fc640", size = 1955019, upload-time = "2025-04-23T18:31:01.335Z" }, + { url = "https://files.pythonhosted.org/packages/3f/8d/71db63483d518cbbf290261a1fc2839d17ff89fce7089e08cad07ccfce67/pydantic_core-2.33.2-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:4c5b0a576fb381edd6d27f0a85915c6daf2f8138dc5c267a57c08a62900758c7", size = 2028584, upload-time = "2025-04-23T18:31:03.106Z" }, + { url = "https://files.pythonhosted.org/packages/24/2f/3cfa7244ae292dd850989f328722d2aef313f74ffc471184dc509e1e4e5a/pydantic_core-2.33.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:e799c050df38a639db758c617ec771fd8fb7a5f8eaaa4b27b101f266b216a246", size = 1855071, upload-time = "2025-04-23T18:31:04.621Z" }, + { url = "https://files.pythonhosted.org/packages/b3/d3/4ae42d33f5e3f50dd467761304be2fa0a9417fbf09735bc2cce003480f2a/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dc46a01bf8d62f227d5ecee74178ffc448ff4e5197c756331f71efcc66dc980f", size = 1897823, upload-time = "2025-04-23T18:31:06.377Z" }, + { url = "https://files.pythonhosted.org/packages/f4/f3/aa5976e8352b7695ff808599794b1fba2a9ae2ee954a3426855935799488/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:a144d4f717285c6d9234a66778059f33a89096dfb9b39117663fd8413d582dcc", size = 1983792, upload-time = "2025-04-23T18:31:07.93Z" }, + { url = "https://files.pythonhosted.org/packages/d5/7a/cda9b5a23c552037717f2b2a5257e9b2bfe45e687386df9591eff7b46d28/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:73cf6373c21bc80b2e0dc88444f41ae60b2f070ed02095754eb5a01df12256de", size = 2136338, upload-time = "2025-04-23T18:31:09.283Z" }, + { url = "https://files.pythonhosted.org/packages/2b/9f/b8f9ec8dd1417eb9da784e91e1667d58a2a4a7b7b34cf4af765ef663a7e5/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3dc625f4aa79713512d1976fe9f0bc99f706a9dee21dfd1810b4bbbf228d0e8a", size = 2730998, upload-time = "2025-04-23T18:31:11.7Z" }, + { url = "https://files.pythonhosted.org/packages/47/bc/cd720e078576bdb8255d5032c5d63ee5c0bf4b7173dd955185a1d658c456/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:881b21b5549499972441da4758d662aeea93f1923f953e9cbaff14b8b9565aef", size = 2003200, upload-time = "2025-04-23T18:31:13.536Z" }, + { url = "https://files.pythonhosted.org/packages/ca/22/3602b895ee2cd29d11a2b349372446ae9727c32e78a94b3d588a40fdf187/pydantic_core-2.33.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:bdc25f3681f7b78572699569514036afe3c243bc3059d3942624e936ec93450e", size = 2113890, upload-time = "2025-04-23T18:31:15.011Z" }, + { url = "https://files.pythonhosted.org/packages/ff/e6/e3c5908c03cf00d629eb38393a98fccc38ee0ce8ecce32f69fc7d7b558a7/pydantic_core-2.33.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:fe5b32187cbc0c862ee201ad66c30cf218e5ed468ec8dc1cf49dec66e160cc4d", size = 2073359, upload-time = "2025-04-23T18:31:16.393Z" }, + { url = "https://files.pythonhosted.org/packages/12/e7/6a36a07c59ebefc8777d1ffdaf5ae71b06b21952582e4b07eba88a421c79/pydantic_core-2.33.2-cp311-cp311-musllinux_1_1_armv7l.whl", hash = "sha256:bc7aee6f634a6f4a95676fcb5d6559a2c2a390330098dba5e5a5f28a2e4ada30", size = 2245883, upload-time = "2025-04-23T18:31:17.892Z" }, + { url = "https://files.pythonhosted.org/packages/16/3f/59b3187aaa6cc0c1e6616e8045b284de2b6a87b027cce2ffcea073adf1d2/pydantic_core-2.33.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:235f45e5dbcccf6bd99f9f472858849f73d11120d76ea8707115415f8e5ebebf", size = 2241074, upload-time = "2025-04-23T18:31:19.205Z" }, + { url = "https://files.pythonhosted.org/packages/e0/ed/55532bb88f674d5d8f67ab121a2a13c385df382de2a1677f30ad385f7438/pydantic_core-2.33.2-cp311-cp311-win32.whl", hash = "sha256:6368900c2d3ef09b69cb0b913f9f8263b03786e5b2a387706c5afb66800efd51", size = 1910538, upload-time = "2025-04-23T18:31:20.541Z" }, + { url = "https://files.pythonhosted.org/packages/fe/1b/25b7cccd4519c0b23c2dd636ad39d381abf113085ce4f7bec2b0dc755eb1/pydantic_core-2.33.2-cp311-cp311-win_amd64.whl", hash = "sha256:1e063337ef9e9820c77acc768546325ebe04ee38b08703244c1309cccc4f1bab", size = 1952909, upload-time = "2025-04-23T18:31:22.371Z" }, + { url = "https://files.pythonhosted.org/packages/49/a9/d809358e49126438055884c4366a1f6227f0f84f635a9014e2deb9b9de54/pydantic_core-2.33.2-cp311-cp311-win_arm64.whl", hash = "sha256:6b99022f1d19bc32a4c2a0d544fc9a76e3be90f0b3f4af413f87d38749300e65", size = 1897786, upload-time = "2025-04-23T18:31:24.161Z" }, + { url = "https://files.pythonhosted.org/packages/18/8a/2b41c97f554ec8c71f2a8a5f85cb56a8b0956addfe8b0efb5b3d77e8bdc3/pydantic_core-2.33.2-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:a7ec89dc587667f22b6a0b6579c249fca9026ce7c333fc142ba42411fa243cdc", size = 2009000, upload-time = "2025-04-23T18:31:25.863Z" }, + { url = "https://files.pythonhosted.org/packages/a1/02/6224312aacb3c8ecbaa959897af57181fb6cf3a3d7917fd44d0f2917e6f2/pydantic_core-2.33.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3c6db6e52c6d70aa0d00d45cdb9b40f0433b96380071ea80b09277dba021ddf7", size = 1847996, upload-time = "2025-04-23T18:31:27.341Z" }, + { url = "https://files.pythonhosted.org/packages/d6/46/6dcdf084a523dbe0a0be59d054734b86a981726f221f4562aed313dbcb49/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4e61206137cbc65e6d5256e1166f88331d3b6238e082d9f74613b9b765fb9025", size = 1880957, upload-time = "2025-04-23T18:31:28.956Z" }, + { url = "https://files.pythonhosted.org/packages/ec/6b/1ec2c03837ac00886ba8160ce041ce4e325b41d06a034adbef11339ae422/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:eb8c529b2819c37140eb51b914153063d27ed88e3bdc31b71198a198e921e011", size = 1964199, upload-time = "2025-04-23T18:31:31.025Z" }, + { url = "https://files.pythonhosted.org/packages/2d/1d/6bf34d6adb9debd9136bd197ca72642203ce9aaaa85cfcbfcf20f9696e83/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c52b02ad8b4e2cf14ca7b3d918f3eb0ee91e63b3167c32591e57c4317e134f8f", size = 2120296, upload-time = "2025-04-23T18:31:32.514Z" }, + { url = "https://files.pythonhosted.org/packages/e0/94/2bd0aaf5a591e974b32a9f7123f16637776c304471a0ab33cf263cf5591a/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:96081f1605125ba0855dfda83f6f3df5ec90c61195421ba72223de35ccfb2f88", size = 2676109, upload-time = "2025-04-23T18:31:33.958Z" }, + { url = "https://files.pythonhosted.org/packages/f9/41/4b043778cf9c4285d59742281a769eac371b9e47e35f98ad321349cc5d61/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8f57a69461af2a5fa6e6bbd7a5f60d3b7e6cebb687f55106933188e79ad155c1", size = 2002028, upload-time = "2025-04-23T18:31:39.095Z" }, + { url = "https://files.pythonhosted.org/packages/cb/d5/7bb781bf2748ce3d03af04d5c969fa1308880e1dca35a9bd94e1a96a922e/pydantic_core-2.33.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:572c7e6c8bb4774d2ac88929e3d1f12bc45714ae5ee6d9a788a9fb35e60bb04b", size = 2100044, upload-time = "2025-04-23T18:31:41.034Z" }, + { url = "https://files.pythonhosted.org/packages/fe/36/def5e53e1eb0ad896785702a5bbfd25eed546cdcf4087ad285021a90ed53/pydantic_core-2.33.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:db4b41f9bd95fbe5acd76d89920336ba96f03e149097365afe1cb092fceb89a1", size = 2058881, upload-time = "2025-04-23T18:31:42.757Z" }, + { url = "https://files.pythonhosted.org/packages/01/6c/57f8d70b2ee57fc3dc8b9610315949837fa8c11d86927b9bb044f8705419/pydantic_core-2.33.2-cp312-cp312-musllinux_1_1_armv7l.whl", hash = "sha256:fa854f5cf7e33842a892e5c73f45327760bc7bc516339fda888c75ae60edaeb6", size = 2227034, upload-time = "2025-04-23T18:31:44.304Z" }, + { url = "https://files.pythonhosted.org/packages/27/b9/9c17f0396a82b3d5cbea4c24d742083422639e7bb1d5bf600e12cb176a13/pydantic_core-2.33.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:5f483cfb75ff703095c59e365360cb73e00185e01aaea067cd19acffd2ab20ea", size = 2234187, upload-time = "2025-04-23T18:31:45.891Z" }, + { url = "https://files.pythonhosted.org/packages/b0/6a/adf5734ffd52bf86d865093ad70b2ce543415e0e356f6cacabbc0d9ad910/pydantic_core-2.33.2-cp312-cp312-win32.whl", hash = "sha256:9cb1da0f5a471435a7bc7e439b8a728e8b61e59784b2af70d7c169f8dd8ae290", size = 1892628, upload-time = "2025-04-23T18:31:47.819Z" }, + { url = "https://files.pythonhosted.org/packages/43/e4/5479fecb3606c1368d496a825d8411e126133c41224c1e7238be58b87d7e/pydantic_core-2.33.2-cp312-cp312-win_amd64.whl", hash = "sha256:f941635f2a3d96b2973e867144fde513665c87f13fe0e193c158ac51bfaaa7b2", size = 1955866, upload-time = "2025-04-23T18:31:49.635Z" }, + { url = "https://files.pythonhosted.org/packages/0d/24/8b11e8b3e2be9dd82df4b11408a67c61bb4dc4f8e11b5b0fc888b38118b5/pydantic_core-2.33.2-cp312-cp312-win_arm64.whl", hash = "sha256:cca3868ddfaccfbc4bfb1d608e2ccaaebe0ae628e1416aeb9c4d88c001bb45ab", size = 1888894, upload-time = "2025-04-23T18:31:51.609Z" }, + { url = "https://files.pythonhosted.org/packages/46/8c/99040727b41f56616573a28771b1bfa08a3d3fe74d3d513f01251f79f172/pydantic_core-2.33.2-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:1082dd3e2d7109ad8b7da48e1d4710c8d06c253cbc4a27c1cff4fbcaa97a9e3f", size = 2015688, upload-time = "2025-04-23T18:31:53.175Z" }, + { url = "https://files.pythonhosted.org/packages/3a/cc/5999d1eb705a6cefc31f0b4a90e9f7fc400539b1a1030529700cc1b51838/pydantic_core-2.33.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:f517ca031dfc037a9c07e748cefd8d96235088b83b4f4ba8939105d20fa1dcd6", size = 1844808, upload-time = "2025-04-23T18:31:54.79Z" }, + { url = "https://files.pythonhosted.org/packages/6f/5e/a0a7b8885c98889a18b6e376f344da1ef323d270b44edf8174d6bce4d622/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0a9f2c9dd19656823cb8250b0724ee9c60a82f3cdf68a080979d13092a3b0fef", size = 1885580, upload-time = "2025-04-23T18:31:57.393Z" }, + { url = "https://files.pythonhosted.org/packages/3b/2a/953581f343c7d11a304581156618c3f592435523dd9d79865903272c256a/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:2b0a451c263b01acebe51895bfb0e1cc842a5c666efe06cdf13846c7418caa9a", size = 1973859, upload-time = "2025-04-23T18:31:59.065Z" }, + { url = "https://files.pythonhosted.org/packages/e6/55/f1a813904771c03a3f97f676c62cca0c0a4138654107c1b61f19c644868b/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1ea40a64d23faa25e62a70ad163571c0b342b8bf66d5fa612ac0dec4f069d916", size = 2120810, upload-time = "2025-04-23T18:32:00.78Z" }, + { url = "https://files.pythonhosted.org/packages/aa/c3/053389835a996e18853ba107a63caae0b9deb4a276c6b472931ea9ae6e48/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0fb2d542b4d66f9470e8065c5469ec676978d625a8b7a363f07d9a501a9cb36a", size = 2676498, upload-time = "2025-04-23T18:32:02.418Z" }, + { url = "https://files.pythonhosted.org/packages/eb/3c/f4abd740877a35abade05e437245b192f9d0ffb48bbbbd708df33d3cda37/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9fdac5d6ffa1b5a83bca06ffe7583f5576555e6c8b3a91fbd25ea7780f825f7d", size = 2000611, upload-time = "2025-04-23T18:32:04.152Z" }, + { url = "https://files.pythonhosted.org/packages/59/a7/63ef2fed1837d1121a894d0ce88439fe3e3b3e48c7543b2a4479eb99c2bd/pydantic_core-2.33.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:04a1a413977ab517154eebb2d326da71638271477d6ad87a769102f7c2488c56", size = 2107924, upload-time = "2025-04-23T18:32:06.129Z" }, + { url = "https://files.pythonhosted.org/packages/04/8f/2551964ef045669801675f1cfc3b0d74147f4901c3ffa42be2ddb1f0efc4/pydantic_core-2.33.2-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:c8e7af2f4e0194c22b5b37205bfb293d166a7344a5b0d0eaccebc376546d77d5", size = 2063196, upload-time = "2025-04-23T18:32:08.178Z" }, + { url = "https://files.pythonhosted.org/packages/26/bd/d9602777e77fc6dbb0c7db9ad356e9a985825547dce5ad1d30ee04903918/pydantic_core-2.33.2-cp313-cp313-musllinux_1_1_armv7l.whl", hash = "sha256:5c92edd15cd58b3c2d34873597a1e20f13094f59cf88068adb18947df5455b4e", size = 2236389, upload-time = "2025-04-23T18:32:10.242Z" }, + { url = "https://files.pythonhosted.org/packages/42/db/0e950daa7e2230423ab342ae918a794964b053bec24ba8af013fc7c94846/pydantic_core-2.33.2-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:65132b7b4a1c0beded5e057324b7e16e10910c106d43675d9bd87d4f38dde162", size = 2239223, upload-time = "2025-04-23T18:32:12.382Z" }, + { url = "https://files.pythonhosted.org/packages/58/4d/4f937099c545a8a17eb52cb67fe0447fd9a373b348ccfa9a87f141eeb00f/pydantic_core-2.33.2-cp313-cp313-win32.whl", hash = "sha256:52fb90784e0a242bb96ec53f42196a17278855b0f31ac7c3cc6f5c1ec4811849", size = 1900473, upload-time = "2025-04-23T18:32:14.034Z" }, + { url = "https://files.pythonhosted.org/packages/a0/75/4a0a9bac998d78d889def5e4ef2b065acba8cae8c93696906c3a91f310ca/pydantic_core-2.33.2-cp313-cp313-win_amd64.whl", hash = "sha256:c083a3bdd5a93dfe480f1125926afcdbf2917ae714bdb80b36d34318b2bec5d9", size = 1955269, upload-time = "2025-04-23T18:32:15.783Z" }, + { url = "https://files.pythonhosted.org/packages/f9/86/1beda0576969592f1497b4ce8e7bc8cbdf614c352426271b1b10d5f0aa64/pydantic_core-2.33.2-cp313-cp313-win_arm64.whl", hash = "sha256:e80b087132752f6b3d714f041ccf74403799d3b23a72722ea2e6ba2e892555b9", size = 1893921, upload-time = "2025-04-23T18:32:18.473Z" }, + { url = "https://files.pythonhosted.org/packages/a4/7d/e09391c2eebeab681df2b74bfe6c43422fffede8dc74187b2b0bf6fd7571/pydantic_core-2.33.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:61c18fba8e5e9db3ab908620af374db0ac1baa69f0f32df4f61ae23f15e586ac", size = 1806162, upload-time = "2025-04-23T18:32:20.188Z" }, + { url = "https://files.pythonhosted.org/packages/f1/3d/847b6b1fed9f8ed3bb95a9ad04fbd0b212e832d4f0f50ff4d9ee5a9f15cf/pydantic_core-2.33.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:95237e53bb015f67b63c91af7518a62a8660376a6a0db19b89acc77a4d6199f5", size = 1981560, upload-time = "2025-04-23T18:32:22.354Z" }, + { url = "https://files.pythonhosted.org/packages/6f/9a/e73262f6c6656262b5fdd723ad90f518f579b7bc8622e43a942eec53c938/pydantic_core-2.33.2-cp313-cp313t-win_amd64.whl", hash = "sha256:c2fc0a768ef76c15ab9238afa6da7f69895bb5d1ee83aeea2e3509af4472d0b9", size = 1935777, upload-time = "2025-04-23T18:32:25.088Z" }, + { url = "https://files.pythonhosted.org/packages/30/68/373d55e58b7e83ce371691f6eaa7175e3a24b956c44628eb25d7da007917/pydantic_core-2.33.2-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:5c4aa4e82353f65e548c476b37e64189783aa5384903bfea4f41580f255fddfa", size = 2023982, upload-time = "2025-04-23T18:32:53.14Z" }, + { url = "https://files.pythonhosted.org/packages/a4/16/145f54ac08c96a63d8ed6442f9dec17b2773d19920b627b18d4f10a061ea/pydantic_core-2.33.2-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:d946c8bf0d5c24bf4fe333af284c59a19358aa3ec18cb3dc4370080da1e8ad29", size = 1858412, upload-time = "2025-04-23T18:32:55.52Z" }, + { url = "https://files.pythonhosted.org/packages/41/b1/c6dc6c3e2de4516c0bb2c46f6a373b91b5660312342a0cf5826e38ad82fa/pydantic_core-2.33.2-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:87b31b6846e361ef83fedb187bb5b4372d0da3f7e28d85415efa92d6125d6e6d", size = 1892749, upload-time = "2025-04-23T18:32:57.546Z" }, + { url = "https://files.pythonhosted.org/packages/12/73/8cd57e20afba760b21b742106f9dbdfa6697f1570b189c7457a1af4cd8a0/pydantic_core-2.33.2-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:aa9d91b338f2df0508606f7009fde642391425189bba6d8c653afd80fd6bb64e", size = 2067527, upload-time = "2025-04-23T18:32:59.771Z" }, + { url = "https://files.pythonhosted.org/packages/e3/d5/0bb5d988cc019b3cba4a78f2d4b3854427fc47ee8ec8e9eaabf787da239c/pydantic_core-2.33.2-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:2058a32994f1fde4ca0480ab9d1e75a0e8c87c22b53a3ae66554f9af78f2fe8c", size = 2108225, upload-time = "2025-04-23T18:33:04.51Z" }, + { url = "https://files.pythonhosted.org/packages/f1/c5/00c02d1571913d496aabf146106ad8239dc132485ee22efe08085084ff7c/pydantic_core-2.33.2-pp310-pypy310_pp73-musllinux_1_1_aarch64.whl", hash = "sha256:0e03262ab796d986f978f79c943fc5f620381be7287148b8010b4097f79a39ec", size = 2069490, upload-time = "2025-04-23T18:33:06.391Z" }, + { url = "https://files.pythonhosted.org/packages/22/a8/dccc38768274d3ed3a59b5d06f59ccb845778687652daa71df0cab4040d7/pydantic_core-2.33.2-pp310-pypy310_pp73-musllinux_1_1_armv7l.whl", hash = "sha256:1a8695a8d00c73e50bff9dfda4d540b7dee29ff9b8053e38380426a85ef10052", size = 2237525, upload-time = "2025-04-23T18:33:08.44Z" }, + { url = "https://files.pythonhosted.org/packages/d4/e7/4f98c0b125dda7cf7ccd14ba936218397b44f50a56dd8c16a3091df116c3/pydantic_core-2.33.2-pp310-pypy310_pp73-musllinux_1_1_x86_64.whl", hash = "sha256:fa754d1850735a0b0e03bcffd9d4b4343eb417e47196e4485d9cca326073a42c", size = 2238446, upload-time = "2025-04-23T18:33:10.313Z" }, + { url = "https://files.pythonhosted.org/packages/ce/91/2ec36480fdb0b783cd9ef6795753c1dea13882f2e68e73bce76ae8c21e6a/pydantic_core-2.33.2-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:a11c8d26a50bfab49002947d3d237abe4d9e4b5bdc8846a63537b6488e197808", size = 2066678, upload-time = "2025-04-23T18:33:12.224Z" }, + { url = "https://files.pythonhosted.org/packages/7b/27/d4ae6487d73948d6f20dddcd94be4ea43e74349b56eba82e9bdee2d7494c/pydantic_core-2.33.2-pp311-pypy311_pp73-macosx_10_12_x86_64.whl", hash = "sha256:dd14041875d09cc0f9308e37a6f8b65f5585cf2598a53aa0123df8b129d481f8", size = 2025200, upload-time = "2025-04-23T18:33:14.199Z" }, + { url = "https://files.pythonhosted.org/packages/f1/b8/b3cb95375f05d33801024079b9392a5ab45267a63400bf1866e7ce0f0de4/pydantic_core-2.33.2-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:d87c561733f66531dced0da6e864f44ebf89a8fba55f31407b00c2f7f9449593", size = 1859123, upload-time = "2025-04-23T18:33:16.555Z" }, + { url = "https://files.pythonhosted.org/packages/05/bc/0d0b5adeda59a261cd30a1235a445bf55c7e46ae44aea28f7bd6ed46e091/pydantic_core-2.33.2-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2f82865531efd18d6e07a04a17331af02cb7a651583c418df8266f17a63c6612", size = 1892852, upload-time = "2025-04-23T18:33:18.513Z" }, + { url = "https://files.pythonhosted.org/packages/3e/11/d37bdebbda2e449cb3f519f6ce950927b56d62f0b84fd9cb9e372a26a3d5/pydantic_core-2.33.2-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2bfb5112df54209d820d7bf9317c7a6c9025ea52e49f46b6a2060104bba37de7", size = 2067484, upload-time = "2025-04-23T18:33:20.475Z" }, + { url = "https://files.pythonhosted.org/packages/8c/55/1f95f0a05ce72ecb02a8a8a1c3be0579bbc29b1d5ab68f1378b7bebc5057/pydantic_core-2.33.2-pp311-pypy311_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:64632ff9d614e5eecfb495796ad51b0ed98c453e447a76bcbeeb69615079fc7e", size = 2108896, upload-time = "2025-04-23T18:33:22.501Z" }, + { url = "https://files.pythonhosted.org/packages/53/89/2b2de6c81fa131f423246a9109d7b2a375e83968ad0800d6e57d0574629b/pydantic_core-2.33.2-pp311-pypy311_pp73-musllinux_1_1_aarch64.whl", hash = "sha256:f889f7a40498cc077332c7ab6b4608d296d852182211787d4f3ee377aaae66e8", size = 2069475, upload-time = "2025-04-23T18:33:24.528Z" }, + { url = "https://files.pythonhosted.org/packages/b8/e9/1f7efbe20d0b2b10f6718944b5d8ece9152390904f29a78e68d4e7961159/pydantic_core-2.33.2-pp311-pypy311_pp73-musllinux_1_1_armv7l.whl", hash = "sha256:de4b83bb311557e439b9e186f733f6c645b9417c84e2eb8203f3f820a4b988bf", size = 2239013, upload-time = "2025-04-23T18:33:26.621Z" }, + { url = "https://files.pythonhosted.org/packages/3c/b2/5309c905a93811524a49b4e031e9851a6b00ff0fb668794472ea7746b448/pydantic_core-2.33.2-pp311-pypy311_pp73-musllinux_1_1_x86_64.whl", hash = "sha256:82f68293f055f51b51ea42fafc74b6aad03e70e191799430b90c13d643059ebb", size = 2238715, upload-time = "2025-04-23T18:33:28.656Z" }, + { url = "https://files.pythonhosted.org/packages/32/56/8a7ca5d2cd2cda1d245d34b1c9a942920a718082ae8e54e5f3e5a58b7add/pydantic_core-2.33.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:329467cecfb529c925cf2bbd4d60d2c509bc2fb52a20c1045bf09bb70971a9c1", size = 2066757, upload-time = "2025-04-23T18:33:30.645Z" }, ] [[package]] name = "pygments" -version = "2.19.1" +version = "2.19.2" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/7c/2d/c3338d48ea6cc0feb8446d8e6937e1408088a72a39937982cc6111d17f84/pygments-2.19.1.tar.gz", hash = "sha256:61c16d2a8576dc0649d9f39e089b5f02bcd27fba10d8fb4dcc28173f7a45151f", size = 4968581, upload-time = "2025-01-06T17:26:30.443Z" } +sdist = { url = "https://files.pythonhosted.org/packages/b0/77/a5b8c569bf593b0140bde72ea885a803b82086995367bf2037de0159d924/pygments-2.19.2.tar.gz", hash = "sha256:636cb2477cec7f8952536970bc533bc43743542f70392ae026374600add5b887", size = 4968631, upload-time = "2025-06-21T13:39:12.283Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/8a/0b/9fcc47d19c48b59121088dd6da2488a49d5f72dacf8262e2790a1d2c7d15/pygments-2.19.1-py3-none-any.whl", hash = "sha256:9ea1544ad55cecf4b8242fab6dd35a93bbce657034b0611ee383099054ab6d8c", size = 1225293, upload-time = "2025-01-06T17:26:25.553Z" }, + { url = "https://files.pythonhosted.org/packages/c7/21/705964c7812476f378728bdf590ca4b771ec72385c533964653c68e86bdc/pygments-2.19.2-py3-none-any.whl", hash = "sha256:86540386c03d588bb81d44bc3928634ff26449851e99741617ecb9037ee5ec0b", size = 1225217, upload-time = "2025-06-21T13:39:07.939Z" }, ] [[package]] name = "pyparsing" -version = "3.2.1" +version = "3.2.3" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/8b/1a/3544f4f299a47911c2ab3710f534e52fea62a633c96806995da5d25be4b2/pyparsing-3.2.1.tar.gz", hash = "sha256:61980854fd66de3a90028d679a954d5f2623e83144b5afe5ee86f43d762e5f0a", size = 1067694, upload-time = "2024-12-31T20:59:46.157Z" } +sdist = { url = "https://files.pythonhosted.org/packages/bb/22/f1129e69d94ffff626bdb5c835506b3a5b4f3d070f17ea295e12c2c6f60f/pyparsing-3.2.3.tar.gz", hash = "sha256:b9c13f1ab8b3b542f72e28f634bad4de758ab3ce4546e4301970ad6fa77c38be", size = 1088608, upload-time = "2025-03-25T05:01:28.114Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/1c/a7/c8a2d361bf89c0d9577c934ebb7421b25dc84bf3a8e3ac0a40aed9acc547/pyparsing-3.2.1-py3-none-any.whl", hash = "sha256:506ff4f4386c4cec0590ec19e6302d3aedb992fdc02c761e90416f158dacf8e1", size = 107716, upload-time = "2024-12-31T20:59:42.738Z" }, + { url = "https://files.pythonhosted.org/packages/05/e7/df2285f3d08fee213f2d041540fa4fc9ca6c2d44cf36d3a035bf2a8d2bcc/pyparsing-3.2.3-py3-none-any.whl", hash = "sha256:a749938e02d6fd0b59b356ca504a24982314bb090c383e3cf201c95ef7e2bfcf", size = 111120, upload-time = "2025-03-25T05:01:24.908Z" }, ] [[package]] name = "pytest" -version = "8.3.5" +version = "8.4.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "colorama", marker = "sys_platform == 'win32'" }, @@ -2387,11 +3040,12 @@ dependencies = [ { name = "iniconfig" }, { name = "packaging" }, { name = "pluggy" }, + { name = "pygments" }, { name = "tomli", marker = "python_full_version < '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/ae/3c/c9d525a414d506893f0cd8a8d0de7706446213181570cdbd766691164e40/pytest-8.3.5.tar.gz", hash = "sha256:f4efe70cc14e511565ac476b57c279e12a855b11f48f212af1080ef2263d3845", size = 1450891, upload-time = "2025-03-02T12:54:54.503Z" } +sdist = { url = "https://files.pythonhosted.org/packages/08/ba/45911d754e8eba3d5a841a5ce61a65a685ff1798421ac054f85aa8747dfb/pytest-8.4.1.tar.gz", hash = "sha256:7c67fd69174877359ed9371ec3af8a3d2b04741818c51e5e99cc1742251fa93c", size = 1517714, upload-time = "2025-06-18T05:48:06.109Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/30/3d/64ad57c803f1fa1e963a7946b6e0fea4a70df53c1a7fed304586539c2bac/pytest-8.3.5-py3-none-any.whl", hash = "sha256:c69214aa47deac29fad6c2a4f590b9c4a9fdb16a403176fe154b79c0b4d4d820", size = 343634, upload-time = "2025-03-02T12:54:52.069Z" }, + { url = "https://files.pythonhosted.org/packages/29/16/c8a903f4c4dffe7a12843191437d7cd8e32751d5de349d45d3fe69544e87/pytest-8.4.1-py3-none-any.whl", hash = "sha256:539c70ba6fcead8e78eebbf1115e8b589e7565830d7d006a8723f19ac8a0afb7", size = 365474, upload-time = "2025-06-18T05:48:03.955Z" }, ] [[package]] @@ -2408,16 +3062,16 @@ wheels = [ [[package]] name = "python-dotenv" -version = "1.0.1" +version = "1.1.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/bc/57/e84d88dfe0aec03b7a2d4327012c1627ab5f03652216c63d49846d7a6c58/python-dotenv-1.0.1.tar.gz", hash = "sha256:e324ee90a023d808f1959c46bcbc04446a10ced277783dc6ee09987c37ec10ca", size = 39115, upload-time = "2024-01-23T06:33:00.505Z" } +sdist = { url = "https://files.pythonhosted.org/packages/f6/b0/4bc07ccd3572a2f9df7e6782f52b0c6c90dcbb803ac4a167702d7d0dfe1e/python_dotenv-1.1.1.tar.gz", hash = "sha256:a8a6399716257f45be6a007360200409fce5cda2661e3dec71d23dc15f6189ab", size = 41978, upload-time = "2025-06-24T04:21:07.341Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/6a/3e/b68c118422ec867fa7ab88444e1274aa40681c606d59ac27de5a5588f082/python_dotenv-1.0.1-py3-none-any.whl", hash = "sha256:f7b63ef50f1b690dddf550d03497b66d609393b40b564ed0d674909a68ebf16a", size = 19863, upload-time = "2024-01-23T06:32:58.246Z" }, + { url = "https://files.pythonhosted.org/packages/5f/ed/539768cf28c661b5b068d66d96a2f155c4971a5d55684a514c1a0e0dec2f/python_dotenv-1.1.1-py3-none-any.whl", hash = "sha256:31f23644fe2602f88ff55e1f5c79ba497e01224ee7737937930c448e4d0e24dc", size = 20556, upload-time = "2025-06-24T04:21:06.073Z" }, ] [[package]] name = "pytorch-lightning" -version = "2.5.1" +version = "2.5.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "fsspec", extra = ["http"] }, @@ -2429,37 +3083,40 @@ dependencies = [ { name = "tqdm" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/d8/75/b8ac01fd3974328ae9239c39550d48899b2fa1b0e064c01615b72a574082/pytorch_lightning-2.5.1.tar.gz", hash = "sha256:27a8adb799c13b8202afad518352248d61303fb230ec1f9fa60e0f81d431d6b1", size = 634309, upload-time = "2025-03-19T20:28:22.444Z" } +sdist = { url = "https://files.pythonhosted.org/packages/01/3e/728fbdc671d07727ad447f9401d98a43570573965beb3cb2060f9a330b4f/pytorch_lightning-2.5.2.tar.gz", hash = "sha256:f817087d611be8d43b777dd4e543d72703e235510936677a13e6c29f7fd790e3", size = 636859, upload-time = "2025-06-20T15:58:27.062Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/82/ff/5701f79317a1a03e5ee8a1bf48e7273a8445162a2774e51fc06411a67c89/pytorch_lightning-2.5.1-py3-none-any.whl", hash = "sha256:0bfbbd3ad80281d3062f5d8029a759093bd969ff8162e7c1fe2918552b269f9e", size = 822982, upload-time = "2025-03-19T20:28:20.1Z" }, + { url = "https://files.pythonhosted.org/packages/e2/42/47c186c8f9e956e559c89e6c764d5d5d0d0af517c04ca0ad39bd0a357d3a/pytorch_lightning-2.5.2-py3-none-any.whl", hash = "sha256:17cfdf89bd98074e389101f097cdf34c486a1f5c6d3fdcefbaf4dea7f97ff0bf", size = 825366, upload-time = "2025-06-20T15:58:25.534Z" }, ] [[package]] name = "pytz" -version = "2025.1" +version = "2025.2" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/5f/57/df1c9157c8d5a05117e455d66fd7cf6dbc46974f832b1058ed4856785d8a/pytz-2025.1.tar.gz", hash = "sha256:c2db42be2a2518b28e65f9207c4d05e6ff547d1efa4086469ef855e4ab70178e", size = 319617, upload-time = "2025-01-31T01:54:48.615Z" } +sdist = { url = "https://files.pythonhosted.org/packages/f8/bf/abbd3cdfb8fbc7fb3d4d38d320f2441b1e7cbe29be4f23797b4a2b5d8aac/pytz-2025.2.tar.gz", hash = "sha256:360b9e3dbb49a209c21ad61809c7fb453643e048b38924c765813546746e81c3", size = 320884, upload-time = "2025-03-25T02:25:00.538Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/eb/38/ac33370d784287baa1c3d538978b5e2ea064d4c1b93ffbd12826c190dd10/pytz-2025.1-py2.py3-none-any.whl", hash = "sha256:89dd22dca55b46eac6eda23b2d72721bf1bdfef212645d81513ef5d03038de57", size = 507930, upload-time = "2025-01-31T01:54:45.634Z" }, + { url = "https://files.pythonhosted.org/packages/81/c4/34e93fe5f5429d7570ec1fa436f1986fb1f00c3e0f43a589fe2bbcd22c3f/pytz-2025.2-py2.py3-none-any.whl", hash = "sha256:5ddf76296dd8c44c26eb8f4b6f35488f3ccbf6fbbd7adee0b7262d43f0ec2f00", size = 509225, upload-time = "2025-03-25T02:24:58.468Z" }, ] [[package]] name = "pywin32" -version = "310" +version = "311" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/95/da/a5f38fffbba2fb99aa4aa905480ac4b8e83ca486659ac8c95bce47fb5276/pywin32-310-cp310-cp310-win32.whl", hash = "sha256:6dd97011efc8bf51d6793a82292419eba2c71cf8e7250cfac03bba284454abc1", size = 8848240, upload-time = "2025-03-17T00:55:46.783Z" }, - { url = "https://files.pythonhosted.org/packages/aa/fe/d873a773324fa565619ba555a82c9dabd677301720f3660a731a5d07e49a/pywin32-310-cp310-cp310-win_amd64.whl", hash = "sha256:c3e78706e4229b915a0821941a84e7ef420bf2b77e08c9dae3c76fd03fd2ae3d", size = 9601854, upload-time = "2025-03-17T00:55:48.783Z" }, - { url = "https://files.pythonhosted.org/packages/3c/84/1a8e3d7a15490d28a5d816efa229ecb4999cdc51a7c30dd8914f669093b8/pywin32-310-cp310-cp310-win_arm64.whl", hash = "sha256:33babed0cf0c92a6f94cc6cc13546ab24ee13e3e800e61ed87609ab91e4c8213", size = 8522963, upload-time = "2025-03-17T00:55:50.969Z" }, - { url = "https://files.pythonhosted.org/packages/f7/b1/68aa2986129fb1011dabbe95f0136f44509afaf072b12b8f815905a39f33/pywin32-310-cp311-cp311-win32.whl", hash = "sha256:1e765f9564e83011a63321bb9d27ec456a0ed90d3732c4b2e312b855365ed8bd", size = 8784284, upload-time = "2025-03-17T00:55:53.124Z" }, - { url = "https://files.pythonhosted.org/packages/b3/bd/d1592635992dd8db5bb8ace0551bc3a769de1ac8850200cfa517e72739fb/pywin32-310-cp311-cp311-win_amd64.whl", hash = "sha256:126298077a9d7c95c53823934f000599f66ec9296b09167810eb24875f32689c", size = 9520748, upload-time = "2025-03-17T00:55:55.203Z" }, - { url = "https://files.pythonhosted.org/packages/90/b1/ac8b1ffce6603849eb45a91cf126c0fa5431f186c2e768bf56889c46f51c/pywin32-310-cp311-cp311-win_arm64.whl", hash = "sha256:19ec5fc9b1d51c4350be7bb00760ffce46e6c95eaf2f0b2f1150657b1a43c582", size = 8455941, upload-time = "2025-03-17T00:55:57.048Z" }, - { url = "https://files.pythonhosted.org/packages/6b/ec/4fdbe47932f671d6e348474ea35ed94227fb5df56a7c30cbbb42cd396ed0/pywin32-310-cp312-cp312-win32.whl", hash = "sha256:8a75a5cc3893e83a108c05d82198880704c44bbaee4d06e442e471d3c9ea4f3d", size = 8796239, upload-time = "2025-03-17T00:55:58.807Z" }, - { url = "https://files.pythonhosted.org/packages/e3/e5/b0627f8bb84e06991bea89ad8153a9e50ace40b2e1195d68e9dff6b03d0f/pywin32-310-cp312-cp312-win_amd64.whl", hash = "sha256:bf5c397c9a9a19a6f62f3fb821fbf36cac08f03770056711f765ec1503972060", size = 9503839, upload-time = "2025-03-17T00:56:00.8Z" }, - { url = "https://files.pythonhosted.org/packages/1f/32/9ccf53748df72301a89713936645a664ec001abd35ecc8578beda593d37d/pywin32-310-cp312-cp312-win_arm64.whl", hash = "sha256:2349cc906eae872d0663d4d6290d13b90621eaf78964bb1578632ff20e152966", size = 8459470, upload-time = "2025-03-17T00:56:02.601Z" }, - { url = "https://files.pythonhosted.org/packages/1c/09/9c1b978ffc4ae53999e89c19c77ba882d9fce476729f23ef55211ea1c034/pywin32-310-cp313-cp313-win32.whl", hash = "sha256:5d241a659c496ada3253cd01cfaa779b048e90ce4b2b38cd44168ad555ce74ab", size = 8794384, upload-time = "2025-03-17T00:56:04.383Z" }, - { url = "https://files.pythonhosted.org/packages/45/3c/b4640f740ffebadd5d34df35fecba0e1cfef8fde9f3e594df91c28ad9b50/pywin32-310-cp313-cp313-win_amd64.whl", hash = "sha256:667827eb3a90208ddbdcc9e860c81bde63a135710e21e4cb3348968e4bd5249e", size = 9503039, upload-time = "2025-03-17T00:56:06.207Z" }, - { url = "https://files.pythonhosted.org/packages/b4/f4/f785020090fb050e7fb6d34b780f2231f302609dc964672f72bfaeb59a28/pywin32-310-cp313-cp313-win_arm64.whl", hash = "sha256:e308f831de771482b7cf692a1f308f8fca701b2d8f9dde6cc440c7da17e47b33", size = 8458152, upload-time = "2025-03-17T00:56:07.819Z" }, + { url = "https://files.pythonhosted.org/packages/7b/40/44efbb0dfbd33aca6a6483191dae0716070ed99e2ecb0c53683f400a0b4f/pywin32-311-cp310-cp310-win32.whl", hash = "sha256:d03ff496d2a0cd4a5893504789d4a15399133fe82517455e78bad62efbb7f0a3", size = 8760432, upload-time = "2025-07-14T20:13:05.9Z" }, + { url = "https://files.pythonhosted.org/packages/5e/bf/360243b1e953bd254a82f12653974be395ba880e7ec23e3731d9f73921cc/pywin32-311-cp310-cp310-win_amd64.whl", hash = "sha256:797c2772017851984b97180b0bebe4b620bb86328e8a884bb626156295a63b3b", size = 9590103, upload-time = "2025-07-14T20:13:07.698Z" }, + { url = "https://files.pythonhosted.org/packages/57/38/d290720e6f138086fb3d5ffe0b6caa019a791dd57866940c82e4eeaf2012/pywin32-311-cp310-cp310-win_arm64.whl", hash = "sha256:0502d1facf1fed4839a9a51ccbcc63d952cf318f78ffc00a7e78528ac27d7a2b", size = 8778557, upload-time = "2025-07-14T20:13:11.11Z" }, + { url = "https://files.pythonhosted.org/packages/7c/af/449a6a91e5d6db51420875c54f6aff7c97a86a3b13a0b4f1a5c13b988de3/pywin32-311-cp311-cp311-win32.whl", hash = "sha256:184eb5e436dea364dcd3d2316d577d625c0351bf237c4e9a5fabbcfa5a58b151", size = 8697031, upload-time = "2025-07-14T20:13:13.266Z" }, + { url = "https://files.pythonhosted.org/packages/51/8f/9bb81dd5bb77d22243d33c8397f09377056d5c687aa6d4042bea7fbf8364/pywin32-311-cp311-cp311-win_amd64.whl", hash = "sha256:3ce80b34b22b17ccbd937a6e78e7225d80c52f5ab9940fe0506a1a16f3dab503", size = 9508308, upload-time = "2025-07-14T20:13:15.147Z" }, + { url = "https://files.pythonhosted.org/packages/44/7b/9c2ab54f74a138c491aba1b1cd0795ba61f144c711daea84a88b63dc0f6c/pywin32-311-cp311-cp311-win_arm64.whl", hash = "sha256:a733f1388e1a842abb67ffa8e7aad0e70ac519e09b0f6a784e65a136ec7cefd2", size = 8703930, upload-time = "2025-07-14T20:13:16.945Z" }, + { url = "https://files.pythonhosted.org/packages/e7/ab/01ea1943d4eba0f850c3c61e78e8dd59757ff815ff3ccd0a84de5f541f42/pywin32-311-cp312-cp312-win32.whl", hash = "sha256:750ec6e621af2b948540032557b10a2d43b0cee2ae9758c54154d711cc852d31", size = 8706543, upload-time = "2025-07-14T20:13:20.765Z" }, + { url = "https://files.pythonhosted.org/packages/d1/a8/a0e8d07d4d051ec7502cd58b291ec98dcc0c3fff027caad0470b72cfcc2f/pywin32-311-cp312-cp312-win_amd64.whl", hash = "sha256:b8c095edad5c211ff31c05223658e71bf7116daa0ecf3ad85f3201ea3190d067", size = 9495040, upload-time = "2025-07-14T20:13:22.543Z" }, + { url = "https://files.pythonhosted.org/packages/ba/3a/2ae996277b4b50f17d61f0603efd8253cb2d79cc7ae159468007b586396d/pywin32-311-cp312-cp312-win_arm64.whl", hash = "sha256:e286f46a9a39c4a18b319c28f59b61de793654af2f395c102b4f819e584b5852", size = 8710102, upload-time = "2025-07-14T20:13:24.682Z" }, + { url = "https://files.pythonhosted.org/packages/a5/be/3fd5de0979fcb3994bfee0d65ed8ca9506a8a1260651b86174f6a86f52b3/pywin32-311-cp313-cp313-win32.whl", hash = "sha256:f95ba5a847cba10dd8c4d8fefa9f2a6cf283b8b88ed6178fa8a6c1ab16054d0d", size = 8705700, upload-time = "2025-07-14T20:13:26.471Z" }, + { url = "https://files.pythonhosted.org/packages/e3/28/e0a1909523c6890208295a29e05c2adb2126364e289826c0a8bc7297bd5c/pywin32-311-cp313-cp313-win_amd64.whl", hash = "sha256:718a38f7e5b058e76aee1c56ddd06908116d35147e133427e59a3983f703a20d", size = 9494700, upload-time = "2025-07-14T20:13:28.243Z" }, + { url = "https://files.pythonhosted.org/packages/04/bf/90339ac0f55726dce7d794e6d79a18a91265bdf3aa70b6b9ca52f35e022a/pywin32-311-cp313-cp313-win_arm64.whl", hash = "sha256:7b4075d959648406202d92a2310cb990fea19b535c7f4a78d3f5e10b926eeb8a", size = 8709318, upload-time = "2025-07-14T20:13:30.348Z" }, + { url = "https://files.pythonhosted.org/packages/c9/31/097f2e132c4f16d99a22bfb777e0fd88bd8e1c634304e102f313af69ace5/pywin32-311-cp314-cp314-win32.whl", hash = "sha256:b7a2c10b93f8986666d0c803ee19b5990885872a7de910fc460f9b0c2fbf92ee", size = 8840714, upload-time = "2025-07-14T20:13:32.449Z" }, + { url = "https://files.pythonhosted.org/packages/90/4b/07c77d8ba0e01349358082713400435347df8426208171ce297da32c313d/pywin32-311-cp314-cp314-win_amd64.whl", hash = "sha256:3aca44c046bd2ed8c90de9cb8427f581c479e594e99b5c0bb19b29c10fd6cb87", size = 9656800, upload-time = "2025-07-14T20:13:34.312Z" }, + { url = "https://files.pythonhosted.org/packages/c0/d2/21af5c535501a7233e734b8af901574572da66fcc254cb35d0609c9080dd/pywin32-311-cp314-cp314-win_arm64.whl", hash = "sha256:a508e2d9025764a8270f93111a970e1d0fbfc33f4153b388bb649b7eec4f9b42", size = 8932540, upload-time = "2025-07-14T20:13:36.379Z" }, ] [[package]] @@ -2508,106 +3165,112 @@ wheels = [ [[package]] name = "pyzmq" -version = "26.3.0" +version = "27.0.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "cffi", marker = "implementation_name == 'pypy'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/3a/ed/c3876f3b3e8beba336214ce44e1efa1792dd537027cef24192ac2b077d7c/pyzmq-26.3.0.tar.gz", hash = "sha256:f1cd68b8236faab78138a8fc703f7ca0ad431b17a3fcac696358600d4e6243b3", size = 276733, upload-time = "2025-03-12T08:04:30.804Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/41/a8/cc21dcd6f0f96dbd636fcaab345f9664cd54e6577a21a74694202479d3fa/pyzmq-26.3.0-cp310-cp310-macosx_10_15_universal2.whl", hash = "sha256:1586944f4736515af5c6d3a5b150c7e8ca2a2d6e46b23057320584d6f2438f4a", size = 1345312, upload-time = "2025-03-12T08:01:58.084Z" }, - { url = "https://files.pythonhosted.org/packages/0b/6d/7e0e52798697536d572a105849c4ab621ca00511674b6ce694cb05e437fc/pyzmq-26.3.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa7efc695d1fc9f72d91bf9b6c6fe2d7e1b4193836ec530a98faf7d7a7577a58", size = 678336, upload-time = "2025-03-12T08:01:59.912Z" }, - { url = "https://files.pythonhosted.org/packages/91/86/8914875e2341a40da460feaa9cace727e50a6b640a20ac36186686bde7d9/pyzmq-26.3.0-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bd84441e4021cec6e4dd040550386cd9c9ea1d9418ea1a8002dbb7b576026b2b", size = 916965, upload-time = "2025-03-12T08:02:01.24Z" }, - { url = "https://files.pythonhosted.org/packages/9a/59/72b390b31ed0cc825881435f21baaae9d57e263aba526fa833863b90d667/pyzmq-26.3.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9176856f36c34a8aa5c0b35ddf52a5d5cd8abeece57c2cd904cfddae3fd9acd3", size = 874003, upload-time = "2025-03-12T08:02:02.994Z" }, - { url = "https://files.pythonhosted.org/packages/97/d4/4dd152dbbaac35d4e1fe8e8fd26d73640fcd84ec9c3915b545692df1ffb7/pyzmq-26.3.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:49334faa749d55b77f084389a80654bf2e68ab5191c0235066f0140c1b670d64", size = 867989, upload-time = "2025-03-12T08:02:04.321Z" }, - { url = "https://files.pythonhosted.org/packages/a4/22/1c5dc761dff13981d27d8225aedb19e70ce9149d16cf0c97c7547570e986/pyzmq-26.3.0-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:fd30fc80fe96efb06bea21667c5793bbd65c0dc793187feb39b8f96990680b00", size = 1207989, upload-time = "2025-03-12T08:02:06.048Z" }, - { url = "https://files.pythonhosted.org/packages/03/89/227ffb9e30b3fbe8196e7c97704345feb750b468e852ab64b0d19fa89e1a/pyzmq-26.3.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:b2eddfbbfb473a62c3a251bb737a6d58d91907f6e1d95791431ebe556f47d916", size = 1520523, upload-time = "2025-03-12T08:02:07.764Z" }, - { url = "https://files.pythonhosted.org/packages/29/d3/e9b99b8404b6a470762cb947bc342e462a853a22ce0b0f2982c65a9b698f/pyzmq-26.3.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:70b3acb9ad729a53d4e751dace35404a024f188aad406013454216aba5485b4e", size = 1419912, upload-time = "2025-03-12T08:02:09.563Z" }, - { url = "https://files.pythonhosted.org/packages/bb/69/074e2cde8135cae9452778e644ea5c91493bc536367d956005fe83072f63/pyzmq-26.3.0-cp310-cp310-win32.whl", hash = "sha256:c1bd75d692cd7c6d862a98013bfdf06702783b75cffbf5dae06d718fecefe8f2", size = 583733, upload-time = "2025-03-12T08:02:11.647Z" }, - { url = "https://files.pythonhosted.org/packages/00/f0/55e57d40f6e21877e96507c0c2dd7e32afffc37b0dde7b834df1170cd749/pyzmq-26.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:d7165bcda0dbf203e5ad04d79955d223d84b2263df4db92f525ba370b03a12ab", size = 647229, upload-time = "2025-03-12T08:02:12.992Z" }, - { url = "https://files.pythonhosted.org/packages/38/1d/6e935b5f06d674c931540b29932a0dd5e1b9d29d047c2764a9c8c6f3ce08/pyzmq-26.3.0-cp310-cp310-win_arm64.whl", hash = "sha256:e34a63f71d2ecffb3c643909ad2d488251afeb5ef3635602b3448e609611a7ed", size = 561038, upload-time = "2025-03-12T08:02:14.305Z" }, - { url = "https://files.pythonhosted.org/packages/22/75/774e9a4a4291864dd37a03a7bfaf46a82d61cd36c16edd33a5739ad49be3/pyzmq-26.3.0-cp311-cp311-macosx_10_15_universal2.whl", hash = "sha256:2833602d9d42c94b9d0d2a44d2b382d3d3a4485be018ba19dddc401a464c617a", size = 1345893, upload-time = "2025-03-12T08:02:15.725Z" }, - { url = "https://files.pythonhosted.org/packages/ca/51/d3eedd2bd46ef851bea528d8a2688a5091183b27fc238801fcac70e80dbb/pyzmq-26.3.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d8270d104ec7caa0bdac246d31d48d94472033ceab5ba142881704350b28159c", size = 678261, upload-time = "2025-03-12T08:02:17.444Z" }, - { url = "https://files.pythonhosted.org/packages/de/5e/521d7c6613769dcc3ed5e44e7082938b6dab27fffe02755784e54e98e17b/pyzmq-26.3.0-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c208a977843d18d3bd185f323e4eaa912eb4869cb230947dc6edd8a27a4e558a", size = 915311, upload-time = "2025-03-12T08:02:18.912Z" }, - { url = "https://files.pythonhosted.org/packages/78/db/3be86dd82adc638a2eb07c3028c1747ead49a71d7d334980b007f593fd9f/pyzmq-26.3.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:eddc2be28a379c218e0d92e4a432805dcb0ca5870156a90b54c03cd9799f9f8a", size = 873193, upload-time = "2025-03-12T08:02:20.311Z" }, - { url = "https://files.pythonhosted.org/packages/63/1a/81a31920d5113113ccd50271649dd2d0cfcfe46925d8f8a196fe560ed0e6/pyzmq-26.3.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:c0b519fa2159c42272f8a244354a0e110d65175647e5185b04008ec00df9f079", size = 867648, upload-time = "2025-03-12T08:02:22.148Z" }, - { url = "https://files.pythonhosted.org/packages/55/79/bbf57979ff2d89b5465d7205db08de7222d2560edc11272eb054c5a68cb5/pyzmq-26.3.0-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:1595533de3a80bf8363372c20bafa963ec4bf9f2b8f539b1d9a5017f430b84c9", size = 1208475, upload-time = "2025-03-12T08:02:23.952Z" }, - { url = "https://files.pythonhosted.org/packages/50/fc/1246dfc4b165e7ff97ac3c4117bdd3747e03ebb62269f71f65e216bfac8b/pyzmq-26.3.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:bbef99eb8d18ba9a40f00e8836b8040cdcf0f2fa649684cf7a66339599919d21", size = 1519428, upload-time = "2025-03-12T08:02:25.706Z" }, - { url = "https://files.pythonhosted.org/packages/5f/9a/143aacb6b372b0e2d812aec73a06fc5df3e169a361d4302226f8563954c6/pyzmq-26.3.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:979486d444ca3c469cd1c7f6a619ce48ff08b3b595d451937db543754bfacb65", size = 1419530, upload-time = "2025-03-12T08:02:27.119Z" }, - { url = "https://files.pythonhosted.org/packages/47/f7/b437e77d496089e17e77866eb126dd97ea47041b58e53892f57e82869198/pyzmq-26.3.0-cp311-cp311-win32.whl", hash = "sha256:4b127cfe10b4c56e4285b69fd4b38ea1d368099ea4273d8fb349163fce3cd598", size = 582538, upload-time = "2025-03-12T08:02:28.576Z" }, - { url = "https://files.pythonhosted.org/packages/a1/2c/99a01a2d7865aaf44e47c2182cbdbc15da1f2e4cfee92dc8e1fb5114f993/pyzmq-26.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:cf736cc1298ef15280d9fcf7a25c09b05af016656856dc6fe5626fd8912658dd", size = 647989, upload-time = "2025-03-12T08:02:29.897Z" }, - { url = "https://files.pythonhosted.org/packages/60/b3/36ac1cb8fafeadff09935f4bdc1232e511af8f8893d6cebc7ceb93c6753a/pyzmq-26.3.0-cp311-cp311-win_arm64.whl", hash = "sha256:2dc46ec09f5d36f606ac8393303149e69d17121beee13c8dac25e2a2078e31c4", size = 561533, upload-time = "2025-03-12T08:02:31.265Z" }, - { url = "https://files.pythonhosted.org/packages/7b/03/7170c3814bb9106c1bca67700c731aaf1cd990fd2f0097c754acb600330e/pyzmq-26.3.0-cp312-cp312-macosx_10_15_universal2.whl", hash = "sha256:c80653332c6136da7f4d4e143975e74ac0fa14f851f716d90583bc19e8945cea", size = 1348354, upload-time = "2025-03-12T08:02:32.699Z" }, - { url = "https://files.pythonhosted.org/packages/74/f3/908b17f9111cdc764aef1de3d36026a2984c46ed90c3c2c85f28b66142f0/pyzmq-26.3.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6e317ee1d4528a03506cb1c282cd9db73660a35b3564096de37de7350e7d87a7", size = 671056, upload-time = "2025-03-12T08:02:34.086Z" }, - { url = "https://files.pythonhosted.org/packages/02/ad/afcb8484b65ceacd1609f709c2caeed31bd6c49261a7507cd5c175cc105f/pyzmq-26.3.0-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:943a22ebb3daacb45f76a9bcca9a7b74e7d94608c0c0505da30af900b998ca8d", size = 908597, upload-time = "2025-03-12T08:02:35.536Z" }, - { url = "https://files.pythonhosted.org/packages/a1/b5/4eeeae0aaaa6ef0c74cfa8b2273b53382bd858df6d99485f2fc8211e7002/pyzmq-26.3.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3fc9e71490d989144981ea21ef4fdfaa7b6aa84aff9632d91c736441ce2f6b00", size = 865260, upload-time = "2025-03-12T08:02:37.562Z" }, - { url = "https://files.pythonhosted.org/packages/74/6a/63db856e93e3a3c3dc98a1de28a902cf1b21c7b0d3856cd5931d7cfd30af/pyzmq-26.3.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:e281a8071a06888575a4eb523c4deeefdcd2f5fe4a2d47e02ac8bf3a5b49f695", size = 859916, upload-time = "2025-03-12T08:02:38.954Z" }, - { url = "https://files.pythonhosted.org/packages/e1/ce/d522c9b46ee3746d4b98c81969c568c2c6296e931a65f2c87104b645654c/pyzmq-26.3.0-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:be77efd735bb1064605be8dec6e721141c1421ef0b115ef54e493a64e50e9a52", size = 1201368, upload-time = "2025-03-12T08:02:40.774Z" }, - { url = "https://files.pythonhosted.org/packages/5a/56/29dcd3647a39e933eb489fda261a1e2700a59d4a9432889a85166e15651c/pyzmq-26.3.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:7a4ac2ffa34f1212dd586af90f4ba894e424f0cabb3a49cdcff944925640f6ac", size = 1512663, upload-time = "2025-03-12T08:02:42.2Z" }, - { url = "https://files.pythonhosted.org/packages/6b/36/7c570698127a43398ed1b1832dada59496e633115016addbce5eda9938a6/pyzmq-26.3.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:ba698c7c252af83b6bba9775035263f0df5f807f0404019916d4b71af8161f66", size = 1411693, upload-time = "2025-03-12T08:02:43.583Z" }, - { url = "https://files.pythonhosted.org/packages/de/54/51d39bef85a7cdbca36227f7defdbfcdc5011b8361a3bfc0e8df431f5a5d/pyzmq-26.3.0-cp312-cp312-win32.whl", hash = "sha256:214038aaa88e801e54c2ef0cfdb2e6df27eb05f67b477380a452b595c5ecfa37", size = 581244, upload-time = "2025-03-12T08:02:45.072Z" }, - { url = "https://files.pythonhosted.org/packages/f2/6a/9512b11a1d0c5648534f03d5ab0c3222f55dc9c192029c1cb00a0ca044e2/pyzmq-26.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:bad7fe0372e505442482ca3ccbc0d6f38dae81b1650f57a0aa6bbee18e7df495", size = 643559, upload-time = "2025-03-12T08:02:46.485Z" }, - { url = "https://files.pythonhosted.org/packages/27/9f/faf5c9cf91b61eeb82a5e919d024d3ac28a795c92cce817be264ccd757d3/pyzmq-26.3.0-cp312-cp312-win_arm64.whl", hash = "sha256:b7b578d604e79e99aa39495becea013fd043fa9f36e4b490efa951f3d847a24d", size = 557664, upload-time = "2025-03-12T08:02:47.896Z" }, - { url = "https://files.pythonhosted.org/packages/37/16/97b8c5107bfccb39120e611671a452c9ff6e8626fb3f8d4c15afd652b6ae/pyzmq-26.3.0-cp313-cp313-macosx_10_15_universal2.whl", hash = "sha256:fa85953df84beb7b8b73cb3ec3f5d92b62687a09a8e71525c6734e020edf56fd", size = 1345691, upload-time = "2025-03-12T08:02:49.508Z" }, - { url = "https://files.pythonhosted.org/packages/a5/61/d5572d95040c0bb5b31eed5b23f3f0f992d94e4e0de0cea62e3c7f3a85c1/pyzmq-26.3.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:209d09f0ab6ddbcebe64630d1e6ca940687e736f443c265ae15bc4bfad833597", size = 670622, upload-time = "2025-03-12T08:02:51.112Z" }, - { url = "https://files.pythonhosted.org/packages/1c/0c/f0235d27388aacf4ed8bcc1d574f6f2f629da0a20610faa0a8e9d363c2b0/pyzmq-26.3.0-cp313-cp313-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d35cc1086f1d4f907df85c6cceb2245cb39a04f69c3f375993363216134d76d4", size = 908683, upload-time = "2025-03-12T08:02:52.659Z" }, - { url = "https://files.pythonhosted.org/packages/cb/52/664828f9586c396b857eec088d208230463e3dc991a24df6adbad98fbaa3/pyzmq-26.3.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b380e9087078ba91e45fb18cdd0c25275ffaa045cf63c947be0ddae6186bc9d9", size = 865212, upload-time = "2025-03-12T08:02:54.187Z" }, - { url = "https://files.pythonhosted.org/packages/2b/14/213b2967030b7d7aecc32dd453830f98799b3cbf2b10a40232e9f22a6520/pyzmq-26.3.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:6d64e74143587efe7c9522bb74d1448128fdf9897cc9b6d8b9927490922fd558", size = 860068, upload-time = "2025-03-12T08:02:55.609Z" }, - { url = "https://files.pythonhosted.org/packages/aa/e5/ff50c8fade69d1c0469652832c626d1910668697642c10cb0e1b6183ef9a/pyzmq-26.3.0-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:efba4f53ac7752eea6d8ca38a4ddac579e6e742fba78d1e99c12c95cd2acfc64", size = 1201303, upload-time = "2025-03-12T08:02:57.073Z" }, - { url = "https://files.pythonhosted.org/packages/9a/e2/fff5e483be95ccc11a05781323e001e63ec15daec1d0f6f08de72ca534db/pyzmq-26.3.0-cp313-cp313-musllinux_1_1_i686.whl", hash = "sha256:9b0137a1c40da3b7989839f9b78a44de642cdd1ce20dcef341de174c8d04aa53", size = 1512892, upload-time = "2025-03-12T08:02:58.68Z" }, - { url = "https://files.pythonhosted.org/packages/21/75/cc44d276e43136e5692e487c3c019f816e11ed445261e434217c28cc98c4/pyzmq-26.3.0-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:a995404bd3982c089e57b428c74edd5bfc3b0616b3dbcd6a8e270f1ee2110f36", size = 1411736, upload-time = "2025-03-12T08:03:00.202Z" }, - { url = "https://files.pythonhosted.org/packages/ee/1c/d070cbc9a7961fe772641c51bb3798d88cb1f8e20ca718407363462624cf/pyzmq-26.3.0-cp313-cp313-win32.whl", hash = "sha256:240b1634b9e530ef6a277d95cbca1a6922f44dfddc5f0a3cd6c722a8de867f14", size = 581214, upload-time = "2025-03-12T08:03:02.412Z" }, - { url = "https://files.pythonhosted.org/packages/38/d3/91082f1151ff5b54e0bed40eb1a26f418530ab07ecaec4dbb83e3d9fa9a9/pyzmq-26.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:fe67291775ea4c2883764ba467eb389c29c308c56b86c1e19e49c9e1ed0cbeca", size = 643412, upload-time = "2025-03-12T08:03:04.007Z" }, - { url = "https://files.pythonhosted.org/packages/e0/cf/dabe68dfdf3e67bea6152eeec4b251cf899ee5b853cfb5c97e4719f9e6e9/pyzmq-26.3.0-cp313-cp313-win_arm64.whl", hash = "sha256:73ca9ae9a9011b714cf7650450cd9c8b61a135180b708904f1f0a05004543dce", size = 557444, upload-time = "2025-03-12T08:03:05.53Z" }, - { url = "https://files.pythonhosted.org/packages/c0/56/e7576ac71c1566da4f4ec586351462a2bb202143fb074bf56df8fe85dcc3/pyzmq-26.3.0-cp313-cp313t-macosx_10_15_universal2.whl", hash = "sha256:fea7efbd7e49af9d7e5ed6c506dfc7de3d1a628790bd3a35fd0e3c904dc7d464", size = 1340288, upload-time = "2025-03-12T08:03:07.638Z" }, - { url = "https://files.pythonhosted.org/packages/f1/ab/0bca97e94d420b5908968bc479e51c3686a9f80d8893450eefcd673b1b1d/pyzmq-26.3.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c4430c7cba23bb0e2ee203eee7851c1654167d956fc6d4b3a87909ccaf3c5825", size = 662462, upload-time = "2025-03-12T08:03:10.039Z" }, - { url = "https://files.pythonhosted.org/packages/ee/be/99e89b55863808da322ac3ab52d8e135dcf2241094aaa468bfe2923d5194/pyzmq-26.3.0-cp313-cp313t-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:016d89bee8c7d566fad75516b4e53ec7c81018c062d4c51cd061badf9539be52", size = 896464, upload-time = "2025-03-12T08:03:11.51Z" }, - { url = "https://files.pythonhosted.org/packages/38/d4/a4be06a313c8d6a5fe1d92975db30aca85f502e867fca392532e06a28c3c/pyzmq-26.3.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:04bfe59852d76d56736bfd10ac1d49d421ab8ed11030b4a0332900691507f557", size = 853432, upload-time = "2025-03-12T08:03:12.948Z" }, - { url = "https://files.pythonhosted.org/packages/12/e6/e608b4c34106bbf5b3b382662ea90a43b2e23df0aa9c1f0fd4e21168d523/pyzmq-26.3.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:1fe05bd0d633a0f672bb28cb8b4743358d196792e1caf04973b7898a0d70b046", size = 845884, upload-time = "2025-03-12T08:03:14.429Z" }, - { url = "https://files.pythonhosted.org/packages/c3/a9/d5e6355308ba529d9cd3576ee8bb3b2e2b726571748f515fbb8559401f5b/pyzmq-26.3.0-cp313-cp313t-musllinux_1_1_aarch64.whl", hash = "sha256:2aa1a9f236d5b835fb8642f27de95f9edcfd276c4bc1b6ffc84f27c6fb2e2981", size = 1191454, upload-time = "2025-03-12T08:03:16.348Z" }, - { url = "https://files.pythonhosted.org/packages/6a/9a/a21dc6c73ac242e425709c1e0049368d8f5db5de7c1102a45f93f5c492b3/pyzmq-26.3.0-cp313-cp313t-musllinux_1_1_i686.whl", hash = "sha256:21399b31753bf321043ea60c360ed5052cc7be20739785b1dff1820f819e35b3", size = 1500397, upload-time = "2025-03-12T08:03:17.918Z" }, - { url = "https://files.pythonhosted.org/packages/87/88/0236056156da0278c9ca2e2562463643597808b5bbd6c34009ba217e7e92/pyzmq-26.3.0-cp313-cp313t-musllinux_1_1_x86_64.whl", hash = "sha256:d015efcd96aca8882057e7e6f06224f79eecd22cad193d3e6a0a91ec67590d1f", size = 1398401, upload-time = "2025-03-12T08:03:19.493Z" }, - { url = "https://files.pythonhosted.org/packages/7e/ec/2e02dde6b1a436b02a6c0e3cb64c779bf6e76cc41c12131f29d9b10a088f/pyzmq-26.3.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:ad03f4252d9041b0635c37528dfa3f44b39f46024ae28c8567f7423676ee409b", size = 835672, upload-time = "2025-03-12T08:03:52.729Z" }, - { url = "https://files.pythonhosted.org/packages/22/ee/30c2c3f162912cff31af2b9d87295533d16f867e7621bd6f9ed62d9cc807/pyzmq-26.3.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0f3dfb68cf7bf4cfdf34283a75848e077c5defa4907506327282afe92780084d", size = 570837, upload-time = "2025-03-12T08:03:54.3Z" }, - { url = "https://files.pythonhosted.org/packages/51/a5/5aead624f5f1033dab9bdaf3e2bc692a8042fcb59355c919a2c042061780/pyzmq-26.3.0-pp310-pypy310_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:356ec0e39c5a9cda872b65aca1fd8a5d296ffdadf8e2442b70ff32e73ef597b1", size = 799508, upload-time = "2025-03-12T08:03:55.841Z" }, - { url = "https://files.pythonhosted.org/packages/ca/8a/dcc0a24cfed80cc004abcba710077147ec9178a12865914e73a60a70cb62/pyzmq-26.3.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:749d671b0eec8e738bbf0b361168369d8c682b94fcd458c20741dc4d69ef5278", size = 758001, upload-time = "2025-03-12T08:03:57.87Z" }, - { url = "https://files.pythonhosted.org/packages/1a/74/f18e63540340f5c740396eb6408d154a84e9f0e9e1ae931b192bf2aa7425/pyzmq-26.3.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:f950f17ae608e0786298340163cac25a4c5543ef25362dd5ddb6dcb10b547be9", size = 556425, upload-time = "2025-03-12T08:03:59.479Z" }, - { url = "https://files.pythonhosted.org/packages/f4/c6/e36b2a2ff6534cb1d1f6b3fb37901ac54675caf7b2e1239613aa40d1d217/pyzmq-26.3.0-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:b4fc9903a73c25be9d5fe45c87faababcf3879445efa16140146b08fccfac017", size = 835670, upload-time = "2025-03-12T08:04:01.037Z" }, - { url = "https://files.pythonhosted.org/packages/1d/b9/8059c5af94b245068e7f7379c08c7e409ec854139d6021aecf2c111d8547/pyzmq-26.3.0-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c15b69af22030960ac63567e98ad8221cddf5d720d9cf03d85021dfd452324ef", size = 570838, upload-time = "2025-03-12T08:04:03.125Z" }, - { url = "https://files.pythonhosted.org/packages/80/a4/f0a4266ff2d94a87f7c32895b1716f9ac0edc0471d518462beeb0a9a94b5/pyzmq-26.3.0-pp311-pypy311_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2cf9ab0dff4dbaa2e893eb608373c97eb908e53b7d9793ad00ccbd082c0ee12f", size = 799507, upload-time = "2025-03-12T08:04:04.704Z" }, - { url = "https://files.pythonhosted.org/packages/78/14/3d7d459f496fab8e487b23423ccba57abf7153a4fde0c3e000500fa02ff8/pyzmq-26.3.0-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3ec332675f6a138db57aad93ae6387953763f85419bdbd18e914cb279ee1c451", size = 758002, upload-time = "2025-03-12T08:04:06.678Z" }, - { url = "https://files.pythonhosted.org/packages/22/65/cc1f0e1db1290770285430e36d51767e620487523e6a04094be637e55698/pyzmq-26.3.0-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:eb96568a22fe070590942cd4780950e2172e00fb033a8b76e47692583b1bd97c", size = 556425, upload-time = "2025-03-12T08:04:08.37Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/30/5f/557d2032a2f471edbcc227da724c24a1c05887b5cda1e3ae53af98b9e0a5/pyzmq-27.0.1.tar.gz", hash = "sha256:45c549204bc20e7484ffd2555f6cf02e572440ecf2f3bdd60d4404b20fddf64b", size = 281158, upload-time = "2025-08-03T05:05:40.352Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/72/0b/ccf4d0b152a6a11f0fc01e73978202fe0e8fe0e91e20941598e83a170bee/pyzmq-27.0.1-cp310-cp310-macosx_10_15_universal2.whl", hash = "sha256:90a4da42aa322de8a3522461e3b5fe999935763b27f69a02fced40f4e3cf9682", size = 1329293, upload-time = "2025-08-03T05:02:56.001Z" }, + { url = "https://files.pythonhosted.org/packages/bc/76/48706d291951b1300d3cf985e503806901164bf1581f27c4b6b22dbab2fa/pyzmq-27.0.1-cp310-cp310-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:e648dca28178fc879c814cf285048dd22fd1f03e1104101106505ec0eea50a4d", size = 905953, upload-time = "2025-08-03T05:02:59.061Z" }, + { url = "https://files.pythonhosted.org/packages/aa/8a/df3135b96712068d184c53120c7dbf3023e5e362a113059a4f85cd36c6a0/pyzmq-27.0.1-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4bca8abc31799a6f3652d13f47e0b0e1cab76f9125f2283d085a3754f669b607", size = 666165, upload-time = "2025-08-03T05:03:00.789Z" }, + { url = "https://files.pythonhosted.org/packages/ee/ed/341a7148e08d2830f480f53ab3d136d88fc5011bb367b516d95d0ebb46dd/pyzmq-27.0.1-cp310-cp310-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:092f4011b26d6b0201002f439bd74b38f23f3aefcb358621bdc3b230afc9b2d5", size = 853756, upload-time = "2025-08-03T05:03:03.347Z" }, + { url = "https://files.pythonhosted.org/packages/c2/bc/d26fe010477c3e901f0f5a3e70446950dde9aa217f1d1a13534eb0fccfe5/pyzmq-27.0.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:6f02f30a4a6b3efe665ab13a3dd47109d80326c8fd286311d1ba9f397dc5f247", size = 1654870, upload-time = "2025-08-03T05:03:05.331Z" }, + { url = "https://files.pythonhosted.org/packages/32/21/9b488086bf3f55b2eb26db09007a3962f62f3b81c5c6295a6ff6aaebd69c/pyzmq-27.0.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:f293a1419266e3bf3557d1f8778f9e1ffe7e6b2c8df5c9dca191caf60831eb74", size = 2033444, upload-time = "2025-08-03T05:03:07.318Z" }, + { url = "https://files.pythonhosted.org/packages/3d/53/85b64a792223cd43393d25e03c8609df41aac817ea5ce6a27eceeed433ee/pyzmq-27.0.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:ce181dd1a7c6c012d0efa8ab603c34b5ee9d86e570c03415bbb1b8772eeb381c", size = 1891289, upload-time = "2025-08-03T05:03:08.96Z" }, + { url = "https://files.pythonhosted.org/packages/23/5b/078aae8fe1c4cdba1a77a598870c548fd52b4d4a11e86b8116bbef47d9f3/pyzmq-27.0.1-cp310-cp310-win32.whl", hash = "sha256:f65741cc06630652e82aa68ddef4986a3ab9073dd46d59f94ce5f005fa72037c", size = 566693, upload-time = "2025-08-03T05:03:10.711Z" }, + { url = "https://files.pythonhosted.org/packages/24/e1/4471fff36416ebf1ffe43577b9c7dcf2ff4798f2171f0d169640a48d2305/pyzmq-27.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:44909aa3ed2234d69fe81e1dade7be336bcfeab106e16bdaa3318dcde4262b93", size = 631649, upload-time = "2025-08-03T05:03:12.232Z" }, + { url = "https://files.pythonhosted.org/packages/e8/4c/8edac8dd56f223124aa40403d2c097bbad9b0e2868a67cad9a2a029863aa/pyzmq-27.0.1-cp310-cp310-win_arm64.whl", hash = "sha256:4401649bfa0a38f0f8777f8faba7cd7eb7b5b8ae2abc7542b830dd09ad4aed0d", size = 559274, upload-time = "2025-08-03T05:03:13.728Z" }, + { url = "https://files.pythonhosted.org/packages/ae/18/a8e0da6ababbe9326116fb1c890bf1920eea880e8da621afb6bc0f39a262/pyzmq-27.0.1-cp311-cp311-macosx_10_15_universal2.whl", hash = "sha256:9729190bd770314f5fbba42476abf6abe79a746eeda11d1d68fd56dd70e5c296", size = 1332721, upload-time = "2025-08-03T05:03:15.237Z" }, + { url = "https://files.pythonhosted.org/packages/75/a4/9431ba598651d60ebd50dc25755402b770322cf8432adcc07d2906e53a54/pyzmq-27.0.1-cp311-cp311-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:696900ef6bc20bef6a242973943574f96c3f97d2183c1bd3da5eea4f559631b1", size = 908249, upload-time = "2025-08-03T05:03:16.933Z" }, + { url = "https://files.pythonhosted.org/packages/f0/7a/e624e1793689e4e685d2ee21c40277dd4024d9d730af20446d88f69be838/pyzmq-27.0.1-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f96a63aecec22d3f7fdea3c6c98df9e42973f5856bb6812c3d8d78c262fee808", size = 668649, upload-time = "2025-08-03T05:03:18.49Z" }, + { url = "https://files.pythonhosted.org/packages/6c/29/0652a39d4e876e0d61379047ecf7752685414ad2e253434348246f7a2a39/pyzmq-27.0.1-cp311-cp311-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c512824360ea7490390566ce00bee880e19b526b312b25cc0bc30a0fe95cb67f", size = 856601, upload-time = "2025-08-03T05:03:20.194Z" }, + { url = "https://files.pythonhosted.org/packages/36/2d/8d5355d7fc55bb6e9c581dd74f58b64fa78c994079e3a0ea09b1b5627cde/pyzmq-27.0.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:dfb2bb5e0f7198eaacfb6796fb0330afd28f36d985a770745fba554a5903595a", size = 1657750, upload-time = "2025-08-03T05:03:22.055Z" }, + { url = "https://files.pythonhosted.org/packages/ab/f4/cd032352d5d252dc6f5ee272a34b59718ba3af1639a8a4ef4654f9535cf5/pyzmq-27.0.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:4f6886c59ba93ffde09b957d3e857e7950c8fe818bd5494d9b4287bc6d5bc7f1", size = 2034312, upload-time = "2025-08-03T05:03:23.578Z" }, + { url = "https://files.pythonhosted.org/packages/e4/1a/c050d8b6597200e97a4bd29b93c769d002fa0b03083858227e0376ad59bc/pyzmq-27.0.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b99ea9d330e86ce1ff7f2456b33f1bf81c43862a5590faf4ef4ed3a63504bdab", size = 1893632, upload-time = "2025-08-03T05:03:25.167Z" }, + { url = "https://files.pythonhosted.org/packages/6a/29/173ce21d5097e7fcf284a090e8beb64fc683c6582b1f00fa52b1b7e867ce/pyzmq-27.0.1-cp311-cp311-win32.whl", hash = "sha256:571f762aed89025ba8cdcbe355fea56889715ec06d0264fd8b6a3f3fa38154ed", size = 566587, upload-time = "2025-08-03T05:03:26.769Z" }, + { url = "https://files.pythonhosted.org/packages/53/ab/22bd33e7086f0a2cc03a5adabff4bde414288bb62a21a7820951ef86ec20/pyzmq-27.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:ee16906c8025fa464bea1e48128c048d02359fb40bebe5333103228528506530", size = 632873, upload-time = "2025-08-03T05:03:28.685Z" }, + { url = "https://files.pythonhosted.org/packages/90/14/3e59b4a28194285ceeff725eba9aa5ba8568d1cb78aed381dec1537c705a/pyzmq-27.0.1-cp311-cp311-win_arm64.whl", hash = "sha256:ba068f28028849da725ff9185c24f832ccf9207a40f9b28ac46ab7c04994bd41", size = 558918, upload-time = "2025-08-03T05:03:30.085Z" }, + { url = "https://files.pythonhosted.org/packages/0e/9b/c0957041067c7724b310f22c398be46399297c12ed834c3bc42200a2756f/pyzmq-27.0.1-cp312-abi3-macosx_10_15_universal2.whl", hash = "sha256:af7ebce2a1e7caf30c0bb64a845f63a69e76a2fadbc1cac47178f7bb6e657bdd", size = 1305432, upload-time = "2025-08-03T05:03:32.177Z" }, + { url = "https://files.pythonhosted.org/packages/8e/55/bd3a312790858f16b7def3897a0c3eb1804e974711bf7b9dcb5f47e7f82c/pyzmq-27.0.1-cp312-abi3-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:8f617f60a8b609a13099b313e7e525e67f84ef4524b6acad396d9ff153f6e4cd", size = 895095, upload-time = "2025-08-03T05:03:33.918Z" }, + { url = "https://files.pythonhosted.org/packages/20/50/fc384631d8282809fb1029a4460d2fe90fa0370a0e866a8318ed75c8d3bb/pyzmq-27.0.1-cp312-abi3-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1d59dad4173dc2a111f03e59315c7bd6e73da1a9d20a84a25cf08325b0582b1a", size = 651826, upload-time = "2025-08-03T05:03:35.818Z" }, + { url = "https://files.pythonhosted.org/packages/7e/0a/2356305c423a975000867de56888b79e44ec2192c690ff93c3109fd78081/pyzmq-27.0.1-cp312-abi3-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f5b6133c8d313bde8bd0d123c169d22525300ff164c2189f849de495e1344577", size = 839751, upload-time = "2025-08-03T05:03:37.265Z" }, + { url = "https://files.pythonhosted.org/packages/d7/1b/81e95ad256ca7e7ccd47f5294c1c6da6e2b64fbace65b84fe8a41470342e/pyzmq-27.0.1-cp312-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:58cca552567423f04d06a075f4b473e78ab5bdb906febe56bf4797633f54aa4e", size = 1641359, upload-time = "2025-08-03T05:03:38.799Z" }, + { url = "https://files.pythonhosted.org/packages/50/63/9f50ec965285f4e92c265c8f18344e46b12803666d8b73b65d254d441435/pyzmq-27.0.1-cp312-abi3-musllinux_1_2_i686.whl", hash = "sha256:4b9d8e26fb600d0d69cc9933e20af08552e97cc868a183d38a5c0d661e40dfbb", size = 2020281, upload-time = "2025-08-03T05:03:40.338Z" }, + { url = "https://files.pythonhosted.org/packages/02/4a/19e3398d0dc66ad2b463e4afa1fc541d697d7bc090305f9dfb948d3dfa29/pyzmq-27.0.1-cp312-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:2329f0c87f0466dce45bba32b63f47018dda5ca40a0085cc5c8558fea7d9fc55", size = 1877112, upload-time = "2025-08-03T05:03:42.012Z" }, + { url = "https://files.pythonhosted.org/packages/bf/42/c562e9151aa90ed1d70aac381ea22a929d6b3a2ce4e1d6e2e135d34fd9c6/pyzmq-27.0.1-cp312-abi3-win32.whl", hash = "sha256:57bb92abdb48467b89c2d21da1ab01a07d0745e536d62afd2e30d5acbd0092eb", size = 558177, upload-time = "2025-08-03T05:03:43.979Z" }, + { url = "https://files.pythonhosted.org/packages/40/96/5c50a7d2d2b05b19994bf7336b97db254299353dd9b49b565bb71b485f03/pyzmq-27.0.1-cp312-abi3-win_amd64.whl", hash = "sha256:ff3f8757570e45da7a5bedaa140489846510014f7a9d5ee9301c61f3f1b8a686", size = 618923, upload-time = "2025-08-03T05:03:45.438Z" }, + { url = "https://files.pythonhosted.org/packages/13/33/1ec89c8f21c89d21a2eaff7def3676e21d8248d2675705e72554fb5a6f3f/pyzmq-27.0.1-cp312-abi3-win_arm64.whl", hash = "sha256:df2c55c958d3766bdb3e9d858b911288acec09a9aab15883f384fc7180df5bed", size = 552358, upload-time = "2025-08-03T05:03:46.887Z" }, + { url = "https://files.pythonhosted.org/packages/6c/a0/f26e276211ec8090a4d11e4ec70eb8a8b15781e591c1d44ce62f372963a0/pyzmq-27.0.1-cp313-cp313-android_24_arm64_v8a.whl", hash = "sha256:497bd8af534ae55dc4ef67eebd1c149ff2a0b0f1e146db73c8b5a53d83c1a5f5", size = 1122287, upload-time = "2025-08-03T05:03:48.838Z" }, + { url = "https://files.pythonhosted.org/packages/9c/d8/af4b507e4f7eeea478cc8ee873995a6fd55582bfb99140593ed460e1db3c/pyzmq-27.0.1-cp313-cp313-android_24_x86_64.whl", hash = "sha256:a066ea6ad6218b4c233906adf0ae67830f451ed238419c0db609310dd781fbe7", size = 1155756, upload-time = "2025-08-03T05:03:50.907Z" }, + { url = "https://files.pythonhosted.org/packages/ac/55/37fae0013e11f88681da42698e550b08a316d608242551f65095cc99232a/pyzmq-27.0.1-cp313-cp313t-macosx_10_15_universal2.whl", hash = "sha256:72d235d6365ca73d8ce92f7425065d70f5c1e19baa458eb3f0d570e425b73a96", size = 1340826, upload-time = "2025-08-03T05:03:52.568Z" }, + { url = "https://files.pythonhosted.org/packages/f2/e4/3a87854c64b26fcf63a9d1b6f4382bd727d4797c772ceb334a97b7489be9/pyzmq-27.0.1-cp313-cp313t-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:313a7b374e3dc64848644ca348a51004b41726f768b02e17e689f1322366a4d9", size = 897283, upload-time = "2025-08-03T05:03:54.167Z" }, + { url = "https://files.pythonhosted.org/packages/17/3e/4296c6b0ad2d07be11ae1395dccf9cae48a0a655cf9be1c3733ad2b591d1/pyzmq-27.0.1-cp313-cp313t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:119ce8590409702394f959c159d048002cbed2f3c0645ec9d6a88087fc70f0f1", size = 660565, upload-time = "2025-08-03T05:03:56.152Z" }, + { url = "https://files.pythonhosted.org/packages/72/41/a33ba3aa48b45b23c4cd4ac49aafde46f3e0f81939f2bfb3b6171a437122/pyzmq-27.0.1-cp313-cp313t-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:45c3e00ce16896ace2cd770ab9057a7cf97d4613ea5f2a13f815141d8b6894b9", size = 847680, upload-time = "2025-08-03T05:03:57.696Z" }, + { url = "https://files.pythonhosted.org/packages/3f/8c/bf2350bb25b3b58d2e5b5d2290ffab0e923f0cc6d02288d3fbf4baa6e4d1/pyzmq-27.0.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:678e50ec112bdc6df5a83ac259a55a4ba97a8b314c325ab26b3b5b071151bc61", size = 1650151, upload-time = "2025-08-03T05:03:59.387Z" }, + { url = "https://files.pythonhosted.org/packages/f7/1a/a5a07c54890891344a8ddc3d5ab320dd3c4e39febb6e4472546e456d5157/pyzmq-27.0.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:d0b96c30be9f9387b18b18b6133c75a7b1b0065da64e150fe1feb5ebf31ece1c", size = 2023766, upload-time = "2025-08-03T05:04:01.883Z" }, + { url = "https://files.pythonhosted.org/packages/62/5e/514dcff08f02c6c8a45a6e23621901139cf853be7ac5ccd0b9407c3aa3de/pyzmq-27.0.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:88dc92d9eb5ea4968123e74db146d770b0c8d48f0e2bfb1dbc6c50a8edb12d64", size = 1885195, upload-time = "2025-08-03T05:04:03.923Z" }, + { url = "https://files.pythonhosted.org/packages/c8/91/87f74f98a487fbef0b115f6025e4a295129fd56b2b633a03ba7d5816ecc2/pyzmq-27.0.1-cp313-cp313t-win32.whl", hash = "sha256:6dcbcb34f5c9b0cefdfc71ff745459241b7d3cda5b27c7ad69d45afc0821d1e1", size = 574213, upload-time = "2025-08-03T05:04:05.905Z" }, + { url = "https://files.pythonhosted.org/packages/e6/d7/07f7d0d7f4c81e08be7b60e52ff2591c557377c017f96204d33d5fca1b07/pyzmq-27.0.1-cp313-cp313t-win_amd64.whl", hash = "sha256:b9fd0fda730461f510cfd9a40fafa5355d65f5e3dbdd8d6dfa342b5b3f5d1949", size = 640202, upload-time = "2025-08-03T05:04:07.439Z" }, + { url = "https://files.pythonhosted.org/packages/ab/83/21d66bcef6fb803647a223cbde95111b099e2176277c0cbc8b099c485510/pyzmq-27.0.1-cp313-cp313t-win_arm64.whl", hash = "sha256:56a3b1853f3954ec1f0e91085f1350cc57d18f11205e4ab6e83e4b7c414120e0", size = 561514, upload-time = "2025-08-03T05:04:09.071Z" }, + { url = "https://files.pythonhosted.org/packages/5a/0b/d5ea75cf46b52cdce85a85200c963cb498932953df443892238be49b1a01/pyzmq-27.0.1-cp314-cp314t-macosx_10_15_universal2.whl", hash = "sha256:f98f6b7787bd2beb1f0dde03f23a0621a0c978edf673b7d8f5e7bc039cbe1b60", size = 1340836, upload-time = "2025-08-03T05:04:10.774Z" }, + { url = "https://files.pythonhosted.org/packages/be/4c/0dbce882550e17db6846b29e9dc242aea7590e7594e1ca5043e8e58fff2d/pyzmq-27.0.1-cp314-cp314t-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:351bf5d8ca0788ca85327fda45843b6927593ff4c807faee368cc5aaf9f809c2", size = 897236, upload-time = "2025-08-03T05:04:13.221Z" }, + { url = "https://files.pythonhosted.org/packages/1b/22/461e131cf16b8814f3c356fa1ea0912697dbc4c64cddf01f7756ec704c1e/pyzmq-27.0.1-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5268a5a9177afff53dc6d70dffe63114ba2a6e7b20d9411cc3adeba09eeda403", size = 660374, upload-time = "2025-08-03T05:04:15.032Z" }, + { url = "https://files.pythonhosted.org/packages/3f/0c/bbd65a814395bf4fc3e57c6c13af27601c07e4009bdfb75ebcf500537bbd/pyzmq-27.0.1-cp314-cp314t-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a4aca06ba295aa78bec9b33ec028d1ca08744c36294338c41432b7171060c808", size = 847497, upload-time = "2025-08-03T05:04:16.967Z" }, + { url = "https://files.pythonhosted.org/packages/1e/df/3d1f4a03b561d824cbd491394f67591957e2f1acf6dc85d96f970312a76a/pyzmq-27.0.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:1c363c6dc66352331d5ad64bb838765c6692766334a6a02fdb05e76bd408ae18", size = 1650028, upload-time = "2025-08-03T05:04:19.398Z" }, + { url = "https://files.pythonhosted.org/packages/41/c9/a3987540f59a412bdaae3f362f78e00e6769557a598c63b7e32956aade5a/pyzmq-27.0.1-cp314-cp314t-musllinux_1_2_i686.whl", hash = "sha256:87aebf4acd7249bdff8d3df03aed4f09e67078e6762cfe0aecf8d0748ff94cde", size = 2023808, upload-time = "2025-08-03T05:04:21.145Z" }, + { url = "https://files.pythonhosted.org/packages/b0/a5/c388f4cd80498a8eaef7535f2a8eaca0a35b82b87a0b47fa1856fc135004/pyzmq-27.0.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:e4f22d67756518d71901edf73b38dc0eb4765cce22c8fe122cc81748d425262b", size = 1884970, upload-time = "2025-08-03T05:04:22.908Z" }, + { url = "https://files.pythonhosted.org/packages/9a/ac/b2a89a1ed90526a1b9a260cdc5cd42f055fd44ee8d2a59902b5ac35ddeb1/pyzmq-27.0.1-cp314-cp314t-win32.whl", hash = "sha256:8c62297bc7aea2147b472ca5ca2b4389377ad82898c87cabab2a94aedd75e337", size = 586905, upload-time = "2025-08-03T05:04:24.492Z" }, + { url = "https://files.pythonhosted.org/packages/68/62/7aa5ea04e836f7a788b2a67405f83011cef59ca76d7bac91d1fc9a0476da/pyzmq-27.0.1-cp314-cp314t-win_amd64.whl", hash = "sha256:bee5248d5ec9223545f8cc4f368c2d571477ae828c99409125c3911511d98245", size = 660503, upload-time = "2025-08-03T05:04:26.382Z" }, + { url = "https://files.pythonhosted.org/packages/89/32/3836ed85947b06f1d67c07ce16c00b0cf8c053ab0b249d234f9f81ff95ff/pyzmq-27.0.1-cp314-cp314t-win_arm64.whl", hash = "sha256:0fc24bf45e4a454e55ef99d7f5c8b8712539200ce98533af25a5bfa954b6b390", size = 575098, upload-time = "2025-08-03T05:04:27.974Z" }, + { url = "https://files.pythonhosted.org/packages/6f/87/fc96f224dd99070fe55d0afc37ac08d7d4635d434e3f9425b232867e01b9/pyzmq-27.0.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:544b995a6a1976fad5d7ff01409b4588f7608ccc41be72147700af91fd44875d", size = 835950, upload-time = "2025-08-03T05:05:04.193Z" }, + { url = "https://files.pythonhosted.org/packages/d1/b6/802d96017f176c3a7285603d9ed2982550095c136c6230d3e0b53f52c7e5/pyzmq-27.0.1-pp310-pypy310_pp73-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:0f772eea55cccce7f45d6ecdd1d5049c12a77ec22404f6b892fae687faa87bee", size = 799876, upload-time = "2025-08-03T05:05:06.263Z" }, + { url = "https://files.pythonhosted.org/packages/4e/52/49045c6528007cce385f218f3a674dc84fc8b3265330d09e57c0a59b41f4/pyzmq-27.0.1-pp310-pypy310_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c9d63d66059114a6756d09169c9209ffceabacb65b9cb0f66e6fc344b20b73e6", size = 567402, upload-time = "2025-08-03T05:05:08.028Z" }, + { url = "https://files.pythonhosted.org/packages/bc/fe/c29ac0d5a817543ecf0cb18f17195805bad0da567a1c64644aacf11b2779/pyzmq-27.0.1-pp310-pypy310_pp73-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1da8e645c655d86f0305fb4c65a0d848f461cd90ee07d21f254667287b5dbe50", size = 747030, upload-time = "2025-08-03T05:05:10.116Z" }, + { url = "https://files.pythonhosted.org/packages/17/d1/cc1fbfb65b4042016e4e035b2548cdfe0945c817345df83aa2d98490e7fc/pyzmq-27.0.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:1843fd0daebcf843fe6d4da53b8bdd3fc906ad3e97d25f51c3fed44436d82a49", size = 544567, upload-time = "2025-08-03T05:05:11.856Z" }, + { url = "https://files.pythonhosted.org/packages/b4/1a/49f66fe0bc2b2568dd4280f1f520ac8fafd73f8d762140e278d48aeaf7b9/pyzmq-27.0.1-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7fb0ee35845bef1e8c4a152d766242164e138c239e3182f558ae15cb4a891f94", size = 835949, upload-time = "2025-08-03T05:05:13.798Z" }, + { url = "https://files.pythonhosted.org/packages/49/94/443c1984b397eab59b14dd7ae8bc2ac7e8f32dbc646474453afcaa6508c4/pyzmq-27.0.1-pp311-pypy311_pp73-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:f379f11e138dfd56c3f24a04164f871a08281194dd9ddf656a278d7d080c8ad0", size = 799875, upload-time = "2025-08-03T05:05:15.632Z" }, + { url = "https://files.pythonhosted.org/packages/30/f1/fd96138a0f152786a2ba517e9c6a8b1b3516719e412a90bb5d8eea6b660c/pyzmq-27.0.1-pp311-pypy311_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b978c0678cffbe8860ec9edc91200e895c29ae1ac8a7085f947f8e8864c489fb", size = 567403, upload-time = "2025-08-03T05:05:17.326Z" }, + { url = "https://files.pythonhosted.org/packages/16/57/34e53ef2b55b1428dac5aabe3a974a16c8bda3bf20549ba500e3ff6cb426/pyzmq-27.0.1-pp311-pypy311_pp73-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7ebccf0d760bc92a4a7c751aeb2fef6626144aace76ee8f5a63abeb100cae87f", size = 747032, upload-time = "2025-08-03T05:05:19.074Z" }, + { url = "https://files.pythonhosted.org/packages/81/b7/769598c5ae336fdb657946950465569cf18803140fe89ce466d7f0a57c11/pyzmq-27.0.1-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:77fed80e30fa65708546c4119840a46691290efc231f6bfb2ac2a39b52e15811", size = 544566, upload-time = "2025-08-03T05:05:20.798Z" }, ] [[package]] name = "rdkit" -version = "2024.9.6" +version = "2025.3.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "pillow" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/a0/44/8908dc5e26248a1e27d5a0f149e17af2b5e85585a227e57d2d4f53f68d07/rdkit-2024.9.6-cp310-cp310-macosx_10_15_x86_64.whl", hash = "sha256:015c7200fffefdae53b59a4658e0323e2d16ac16b2b00174188f769a0cabe52a", size = 29960044, upload-time = "2025-03-12T10:34:47.284Z" }, - { url = "https://files.pythonhosted.org/packages/8f/2e/36a2cee64073df724c7fc635c969de604440d754c65ac1dd2627e5aacc16/rdkit-2024.9.6-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:ffe05be4354178b48bfd60ea40c847e6364e75e71893bd2c659da8819c8afab6", size = 27718856, upload-time = "2025-03-12T10:34:51.256Z" }, - { url = "https://files.pythonhosted.org/packages/13/28/c9abf09ade43a8ed6165bc1e7660cbefdf449f89cb400eb20377e84ceb2c/rdkit-2024.9.6-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:6e6bc9b6edc3e5efa88f796d8a86192b72b5d54ce9488f8ccb043e6b74308b79", size = 33528869, upload-time = "2025-03-12T10:34:54.957Z" }, - { url = "https://files.pythonhosted.org/packages/a3/f6/3a2278acc1d3831fd1d5b81ae355297be1c2ea8d8d0e200291aa2edfbdeb/rdkit-2024.9.6-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:2b5573055c8defbad7ce25db10786a56e0699faf3282daea21100c59f7af7298", size = 34344527, upload-time = "2025-03-12T10:34:59.482Z" }, - { url = "https://files.pythonhosted.org/packages/cb/61/18b1fec82a2c4ab60234eab6673174c7dbcfd82490fc834a4d2b10d42d8b/rdkit-2024.9.6-cp310-cp310-win_amd64.whl", hash = "sha256:9d64c865de57c15bbe7726f20a967016f2ac04f738889c83210091be944b512b", size = 22472377, upload-time = "2025-03-12T10:35:03.297Z" }, - { url = "https://files.pythonhosted.org/packages/cc/c1/b21d793ab572545917af8d00c7519ea1e0fc10d3e74ac936de4a06fc3649/rdkit-2024.9.6-cp311-cp311-macosx_10_15_x86_64.whl", hash = "sha256:bbb3fe860e5ccb8debad592ded991b2aca611c997640fc6e68b0f41beb648639", size = 29960558, upload-time = "2025-03-12T10:35:07.236Z" }, - { url = "https://files.pythonhosted.org/packages/72/64/f96f4a72052be7580ab825279cc71653a476c023c53828765be520db47e6/rdkit-2024.9.6-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a28971024ba0300f7c566874212b8c2c8f7db984e85418eaceec5cb739bcab1d", size = 27719061, upload-time = "2025-03-12T10:35:11.167Z" }, - { url = "https://files.pythonhosted.org/packages/1d/2f/3befa6b3a061665351d5a58aba66e318175777c40cbfbfc511bd57d2aa7b/rdkit-2024.9.6-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:6e236da74308b19f695ed259b12b75e100e56593333f2b47cff2420e9746e571", size = 33524980, upload-time = "2025-03-12T10:35:15.419Z" }, - { url = "https://files.pythonhosted.org/packages/cc/3f/472c33312ca8a55242fc2cf6179809f4a967185e9dc6e76ea28ddd37a097/rdkit-2024.9.6-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:7b473a715711e7f1f747173fc18a16b02d5952eadbeedb09d05d1f0a6879c282", size = 34343341, upload-time = "2025-03-12T10:35:19.528Z" }, - { url = "https://files.pythonhosted.org/packages/94/85/d8d7da7ba8d00ad8977193f30d0062324aa9279e8af67e413674d3f60483/rdkit-2024.9.6-cp311-cp311-win_amd64.whl", hash = "sha256:f302e8a347debe24ffeeb303b8b031ef2e61265872a6450f46aa819b25f5c684", size = 22472536, upload-time = "2025-03-12T10:35:23.174Z" }, - { url = "https://files.pythonhosted.org/packages/b0/b4/54a543114b58bfbb0f659d30c7902d10e7fa908544691455b36e0783fa77/rdkit-2024.9.6-cp312-cp312-macosx_10_15_x86_64.whl", hash = "sha256:f57a2ad6ceff54917586f47a01ecbe43bbdf4a6f00defd746b500228752ccf84", size = 30035515, upload-time = "2025-03-12T10:35:27.512Z" }, - { url = "https://files.pythonhosted.org/packages/d1/5a/83deae41ff455e417f67e91393d3fae9f604e15fc633544412fa16d9bcc6/rdkit-2024.9.6-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:4f3931ff192aeee29ce41cf7c30650883fc10f1504af2a9dcc67d8554a0779d3", size = 27760806, upload-time = "2025-03-12T10:35:31.794Z" }, - { url = "https://files.pythonhosted.org/packages/95/41/c52930491f52d7bdb50e297d5064fdc959a82adea9f735a9ab7c412ac874/rdkit-2024.9.6-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:97661c84a9dc10797d197db1024757676e0ca29a306d5eec289bbfdf4c53c80c", size = 33414832, upload-time = "2025-03-12T10:35:36.122Z" }, - { url = "https://files.pythonhosted.org/packages/e9/f1/2f978f26b2c58716e70ca1e442e51c260b43e55c2eb64bd2ad76584b60da/rdkit-2024.9.6-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:82e91020e700979f6fc94a4722b47d770912d54cd9accb1c85961c57550cb945", size = 34276560, upload-time = "2025-03-12T10:35:40.561Z" }, - { url = "https://files.pythonhosted.org/packages/36/a7/738b3eb302b95ece0efea4d9b9c90d8dd7bb9ef529a56ee3bbbcfd239cde/rdkit-2024.9.6-cp312-cp312-win_amd64.whl", hash = "sha256:995df4e0b09a866fc628b6c9f1c10dca32726964ff1c7506cbf7975af0a9ba60", size = 22487552, upload-time = "2025-03-12T10:35:44.277Z" }, - { url = "https://files.pythonhosted.org/packages/07/be/e0fb0bf1301ce3add94a248a24b29c7784b7d504de4776554fafe3372c54/rdkit-2024.9.6-cp313-cp313-macosx_10_15_x86_64.whl", hash = "sha256:5001786e1f06749fa4cf7d6a9d23b36afc6d9b028d500777a3861c8a4d91238b", size = 30034429, upload-time = "2025-03-12T10:35:47.789Z" }, - { url = "https://files.pythonhosted.org/packages/d4/0e/918410c8975cb365ad51d3e487965c3c5ccfc3dea5715c81f0e89339cec0/rdkit-2024.9.6-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b2578458414600beb3a5691fc900e0640da3c571b30ed3dc76381cfe5caadc9e", size = 27759699, upload-time = "2025-03-12T10:35:52.004Z" }, - { url = "https://files.pythonhosted.org/packages/41/ee/9758cbcde26bacb820579ecbfbd3069ab8bd10b55e08f6731c208ce2eb3f/rdkit-2024.9.6-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:4fbf4d680a1070a15f6d9c7cf93218bed87c7110e4e7a467d1b449e430e20ff5", size = 33413200, upload-time = "2025-03-12T10:35:56.306Z" }, - { url = "https://files.pythonhosted.org/packages/d9/f9/6e5b8f6445450a3b01e28915f749f9d644a8b855f6e7b26c09f193b1978d/rdkit-2024.9.6-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:4b0efe61d9eee664dc19203478e3312e596be0d200c3bbd4ef1b584aef7146d3", size = 34275798, upload-time = "2025-03-12T10:36:00.826Z" }, - { url = "https://files.pythonhosted.org/packages/3f/9b/3c32fd6dbb1210a8e9dc83feee82778c8bf2ad483c9e8394138c25789110/rdkit-2024.9.6-cp313-cp313-win_amd64.whl", hash = "sha256:6c740c543b55f99d8d3bdea92c68617ef5669bcf2cc1652752a65f156f26dbc0", size = 22486705, upload-time = "2025-03-12T10:36:05.44Z" }, + { url = "https://files.pythonhosted.org/packages/f7/dd/325588bac4d71a7cc4acfe9c5565b896de4cf9d46cce9da325858a1075c4/rdkit-2025.3.5-cp310-cp310-macosx_10_15_x86_64.whl", hash = "sha256:323fd21ce62f8b6ca34cebb8d0d7cbd9c4d757f5bff76143e2d2c04c94c1304a", size = 31161247, upload-time = "2025-08-01T08:37:56.364Z" }, + { url = "https://files.pythonhosted.org/packages/e3/5f/7341e6b03e1cf53ee2e82ebd7602e25c0596667c47a3e0bcd6c76ca039ed/rdkit-2025.3.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:1b8aee3c6c9fb0a8a3596a50595e4d88d1e02a70e9cb2d4626ab6134d4a5eceb", size = 28703723, upload-time = "2025-08-01T08:38:01.039Z" }, + { url = "https://files.pythonhosted.org/packages/96/d0/50ce769d29262ca6fa6801dfcc30e9ce6b031a5590aec617f884ac79e209/rdkit-2025.3.5-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:1c70bb2ebcb42125ba6500d692a8ba4738f518f4c334ed1dd0885499617bff53", size = 35372548, upload-time = "2025-08-01T08:38:04.997Z" }, + { url = "https://files.pythonhosted.org/packages/6f/f9/dae8b9785e2cf0361a211166e28f673f283ed2d4829c30b8614537eb66f2/rdkit-2025.3.5-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:07a6d212ce88a91bf8bc187cbd4c98d8a5af3e7c01a7a64b4215f46825456b80", size = 36272060, upload-time = "2025-08-01T08:38:09.233Z" }, + { url = "https://files.pythonhosted.org/packages/e2/bd/af721a3b97d4a59b7d209c71561e9e1c17c08dc5cd2f6f26b69deea94ab1/rdkit-2025.3.5-cp310-cp310-win_amd64.whl", hash = "sha256:1bbc026a8bc873d890b9f79f5df9eec3f322f580dfb1aec9b50aaef8409be6f4", size = 23517085, upload-time = "2025-08-01T08:38:12.942Z" }, + { url = "https://files.pythonhosted.org/packages/9d/04/7724512f61e30cefb6e4af54cd3a1ccf622196acb199bc351017f207747e/rdkit-2025.3.5-cp311-cp311-macosx_10_15_x86_64.whl", hash = "sha256:010c18f84f474f8ff97b5fec6571296ab17475c4378beb5870aa8ea412cc61f2", size = 31161777, upload-time = "2025-08-01T08:38:16.552Z" }, + { url = "https://files.pythonhosted.org/packages/6d/f3/c6c8352008b202351b57191cd9805090a2077d7ebcc301a7e65b38e62ce4/rdkit-2025.3.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:fb10ebcd3f2901a48922bfecb00de64c4589fd121af79b985768952a65436413", size = 28704087, upload-time = "2025-08-01T08:38:20.704Z" }, + { url = "https://files.pythonhosted.org/packages/21/e1/5bec35b13ed5f88d11841bd73dbb32bdae02a317e094ab8737ee1e020920/rdkit-2025.3.5-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:9303de802d03f07936ffab6cd7fe4b9f488314e33fd0eaf3d29edd78cfd9cd88", size = 35367566, upload-time = "2025-08-01T08:38:24.535Z" }, + { url = "https://files.pythonhosted.org/packages/40/17/d3df2ef69504418432364cd42add91da85012985320a571116d2a5492079/rdkit-2025.3.5-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:eb92045567bfda71721f607d52a1d1a7a5dd101b853d62c13c01995f4052b6fb", size = 36271424, upload-time = "2025-08-01T08:38:28.8Z" }, + { url = "https://files.pythonhosted.org/packages/54/7a/98ed651841c61ef761f401e569addf213c28867ab84862fbe4b7d8539f70/rdkit-2025.3.5-cp311-cp311-win_amd64.whl", hash = "sha256:58e7d4498c53205dfb1fa35850daadd7bb18da2791317f251216dc53616e8754", size = 23517947, upload-time = "2025-08-01T08:38:32.187Z" }, + { url = "https://files.pythonhosted.org/packages/42/3c/a377c7117ae5b983c121594e60e8c511a97ebe2c8836548754ca013f5b6c/rdkit-2025.3.5-cp312-cp312-macosx_10_15_x86_64.whl", hash = "sha256:0419a2a19135cfbdd46fc4794dd45a69138e80632883489a499cfc0b750d6c95", size = 31237587, upload-time = "2025-08-01T08:38:35.805Z" }, + { url = "https://files.pythonhosted.org/packages/e7/1d/e293366515c9ec157312db7aa0e4bd5b9f8f00fea526afca033fa8e5df7b/rdkit-2025.3.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:f8a1459a595c2d20928bb477aff155fa3480b09df8a866be9a6c7d334dd2ca2e", size = 28746581, upload-time = "2025-08-01T08:38:40.223Z" }, + { url = "https://files.pythonhosted.org/packages/ba/c7/972600433ab966db58ceb3e1cda4f688bb15ffe4078219323682e6ad9ccd/rdkit-2025.3.5-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:5c5f69325802014aa5a535b8c680636f1c1fb9201fbabf250220bbae429721b7", size = 35243285, upload-time = "2025-08-01T08:38:49.396Z" }, + { url = "https://files.pythonhosted.org/packages/c4/c7/79030fec5fbc72ee46fc35724b707025c0c45c7b0864050d28bff7977baf/rdkit-2025.3.5-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:36b43611c0e455cb5b5f08207fd27d83db405d4f88a5bb8df093aa38c55527a1", size = 36192701, upload-time = "2025-08-01T08:38:53.442Z" }, + { url = "https://files.pythonhosted.org/packages/39/38/0e6ce9905c798abb776a7de1b9582c41359468d9efc82c939146d32264eb/rdkit-2025.3.5-cp312-cp312-win_amd64.whl", hash = "sha256:7f02fb64f0cf5e798357bc49f762eb9854c25c363579b660348557fd1f8f5c00", size = 23537346, upload-time = "2025-08-01T08:38:56.913Z" }, + { url = "https://files.pythonhosted.org/packages/47/2e/c7cc6ea1ce2eca4e03c25d7b7b61e23646d96fff138f472bf7e0015bdd42/rdkit-2025.3.5-cp313-cp313-macosx_10_15_x86_64.whl", hash = "sha256:b34bac98122e0e95c958c96a923cfb8f34b554bff652a571bca1ed85d567d61a", size = 31236417, upload-time = "2025-08-01T08:39:00.325Z" }, + { url = "https://files.pythonhosted.org/packages/c6/1d/4cc7d00c388dde4cb4ce6ff7c4314f0f2c0e342d5b7a7b779fc12c32cf5a/rdkit-2025.3.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b564067066c97136c448c58ac796c66010d041a5fd5dccf8997461a1846e63b7", size = 28745429, upload-time = "2025-08-01T08:39:04.627Z" }, + { url = "https://files.pythonhosted.org/packages/28/ea/c8e6ea43a9f352b9a71e4417ddfc4a66f8bda7813e782a412c206d6bdfbd/rdkit-2025.3.5-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:252288a0d93c5c31344349a85d769d5fa6a7e5ee132efe1fffbc4a0d6d3833c6", size = 35241653, upload-time = "2025-08-01T08:39:08.345Z" }, + { url = "https://files.pythonhosted.org/packages/8b/6a/c6ec8e124bc621ce0706894f877c6f57d7c9c53cbc73510ac046d40b4dc7/rdkit-2025.3.5-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:65c636e4ea0e4bbf80261155e19f96de8e178271513f8823a4364e2939ee5791", size = 36192312, upload-time = "2025-08-01T08:39:12.253Z" }, + { url = "https://files.pythonhosted.org/packages/19/04/6d5c40e8a4b3edd0cef5fa9cda96e36739a7f3e6573d80722b719fd22e13/rdkit-2025.3.5-cp313-cp313-win_amd64.whl", hash = "sha256:8b2c51020f446adb0f56d0fcb38b45a01a665ef4523eea368ea30dc0026cbebd", size = 23535970, upload-time = "2025-08-01T08:39:16.131Z" }, + { url = "https://files.pythonhosted.org/packages/88/c3/548f36267ff114c5a76cea9603eecf51b24bb6a09599a41f19627b6f3662/rdkit-2025.3.5-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:5c84758f67ef8ba09cd8a6310686ffff237337ad65a38c64308ca119132c7914", size = 31238791, upload-time = "2025-08-01T08:39:19.429Z" }, + { url = "https://files.pythonhosted.org/packages/ff/c6/09bc6e443b5e3a526f295f060c743a48c1652fd050d5fead3efb52689102/rdkit-2025.3.5-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:07e2c35d3781bf013c16957b4d8daaa90e4914830ea6490291b276698d2a366d", size = 28763880, upload-time = "2025-08-01T08:39:23.058Z" }, + { url = "https://files.pythonhosted.org/packages/08/80/e90d6fff82972e82e1ff086562d0be7c90dc6367f9f351c8298fbbe88fc8/rdkit-2025.3.5-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:83930691cc76051ca01850d28cba7e21c0f8fcc43605a6355753f9df2fbf855a", size = 35276980, upload-time = "2025-08-01T08:39:27.331Z" }, + { url = "https://files.pythonhosted.org/packages/2b/c2/56cc024b8c9b34c460d0fa17d3180606a96e0e4b1614b9d7e07160ea4f56/rdkit-2025.3.5-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:193a4147598116655b978fe28d9995c864c445a751c9755ba39753a053bbb4a5", size = 36204963, upload-time = "2025-08-01T08:39:31.247Z" }, + { url = "https://files.pythonhosted.org/packages/2c/11/8dfd2d397d45c951571f575e539338f8f16cb85f0997b865b23211ebbc6c/rdkit-2025.3.5-cp314-cp314-win_amd64.whl", hash = "sha256:b715c506f99d3b29ea5eb903fa1fd6787fef567d8a249dfdc14f5f57f372f377", size = 24014579, upload-time = "2025-08-01T08:39:34.935Z" }, ] [[package]] @@ -2626,7 +3289,7 @@ wheels = [ [[package]] name = "requests" -version = "2.32.3" +version = "2.32.4" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "certifi" }, @@ -2634,133 +3297,200 @@ dependencies = [ { name = "idna" }, { name = "urllib3" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/63/70/2bf7780ad2d390a8d301ad0b550f1581eadbd9a20f896afe06353c2a2913/requests-2.32.3.tar.gz", hash = "sha256:55365417734eb18255590a9ff9eb97e9e1da868d4ccd6402399eaf68af20a760", size = 131218, upload-time = "2024-05-29T15:37:49.536Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e1/0a/929373653770d8a0d7ea76c37de6e41f11eb07559b103b1c02cafb3f7cf8/requests-2.32.4.tar.gz", hash = "sha256:27d0316682c8a29834d3264820024b62a36942083d52caf2f14c0591336d3422", size = 135258, upload-time = "2025-06-09T16:43:07.34Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/7c/e4/56027c4a6b4ae70ca9de302488c5ca95ad4a39e190093d6c1a8ace08341b/requests-2.32.4-py3-none-any.whl", hash = "sha256:27babd3cda2a6d50b30443204ee89830707d396671944c998b5975b031ac2b2c", size = 64847, upload-time = "2025-06-09T16:43:05.728Z" }, +] + +[[package]] +name = "rich" +version = "13.9.4" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "markdown-it-py" }, + { name = "pygments" }, + { name = "typing-extensions", marker = "python_full_version < '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/ab/3a/0316b28d0761c6734d6bc14e770d85506c986c85ffb239e688eeaab2c2bc/rich-13.9.4.tar.gz", hash = "sha256:439594978a49a09530cff7ebc4b5c7103ef57baf48d5ea3184f21d9a2befa098", size = 223149, upload-time = "2024-11-01T16:43:57.873Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/f9/9b/335f9764261e915ed497fcdeb11df5dfd6f7bf257d4a6a2a686d80da4d54/requests-2.32.3-py3-none-any.whl", hash = "sha256:70761cfe03c773ceb22aa2f671b4757976145175cdfca038c02654d061d6dcc6", size = 64928, upload-time = "2024-05-29T15:37:47.027Z" }, + { url = "https://files.pythonhosted.org/packages/19/71/39c7c0d87f8d4e6c020a393182060eaefeeae6c01dab6a84ec346f2567df/rich-13.9.4-py3-none-any.whl", hash = "sha256:6049d5e6ec054bf2779ab3358186963bac2ea89175919d699e378b99738c2a90", size = 242424, upload-time = "2024-11-01T16:43:55.817Z" }, ] [[package]] name = "rpds-py" -version = "0.23.1" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/0a/79/2ce611b18c4fd83d9e3aecb5cba93e1917c050f556db39842889fa69b79f/rpds_py-0.23.1.tar.gz", hash = "sha256:7f3240dcfa14d198dba24b8b9cb3b108c06b68d45b7babd9eefc1038fdf7e707", size = 26806, upload-time = "2025-02-21T15:04:23.169Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/34/fe/e5326459863bd525122f4e9c80ac8d7c6cfa171b7518d04cc27c12c209b0/rpds_py-0.23.1-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:2a54027554ce9b129fc3d633c92fa33b30de9f08bc61b32c053dc9b537266fed", size = 372123, upload-time = "2025-02-21T15:01:14.816Z" }, - { url = "https://files.pythonhosted.org/packages/f9/db/f10a3795f7a89fb27594934012d21c61019bbeb516c5bdcfbbe9e9e617a7/rpds_py-0.23.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:b5ef909a37e9738d146519657a1aab4584018746a18f71c692f2f22168ece40c", size = 356778, upload-time = "2025-02-21T15:01:16.788Z" }, - { url = "https://files.pythonhosted.org/packages/21/27/0d3678ad7f432fa86f8fac5f5fc6496a4d2da85682a710d605219be20063/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3ee9d6f0b38efb22ad94c3b68ffebe4c47865cdf4b17f6806d6c674e1feb4246", size = 385775, upload-time = "2025-02-21T15:01:18.192Z" }, - { url = "https://files.pythonhosted.org/packages/99/a0/1786defa125b2ad228027f22dff26312ce7d1fee3c7c3c2682f403db2062/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:f7356a6da0562190558c4fcc14f0281db191cdf4cb96e7604c06acfcee96df15", size = 391181, upload-time = "2025-02-21T15:01:20.214Z" }, - { url = "https://files.pythonhosted.org/packages/f1/5c/1240934050a7ffd020a915486d0cc4c7f6e7a2442a77aedf13664db55d36/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9441af1d25aed96901f97ad83d5c3e35e6cd21a25ca5e4916c82d7dd0490a4fa", size = 444607, upload-time = "2025-02-21T15:01:22.221Z" }, - { url = "https://files.pythonhosted.org/packages/b7/1b/cee6905b47817fd0a377716dbe4df35295de46df46ee2ff704538cc371b0/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3d8abf7896a91fb97e7977d1aadfcc2c80415d6dc2f1d0fca5b8d0df247248f3", size = 445550, upload-time = "2025-02-21T15:01:24.742Z" }, - { url = "https://files.pythonhosted.org/packages/54/f7/f0821ca34032892d7a67fcd5042f50074ff2de64e771e10df01085c88d47/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1b08027489ba8fedde72ddd233a5ea411b85a6ed78175f40285bd401bde7466d", size = 386148, upload-time = "2025-02-21T15:01:26.23Z" }, - { url = "https://files.pythonhosted.org/packages/eb/ef/2afe53bc857c4bcba336acfd2629883a5746e7291023e017ac7fc98d85aa/rpds_py-0.23.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:fee513135b5a58f3bb6d89e48326cd5aa308e4bcdf2f7d59f67c861ada482bf8", size = 416780, upload-time = "2025-02-21T15:01:27.778Z" }, - { url = "https://files.pythonhosted.org/packages/ae/9a/38d2236cf669789b8a3e1a014c9b6a8d7b8925b952c92e7839ae2749f9ac/rpds_py-0.23.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:35d5631ce0af26318dba0ae0ac941c534453e42f569011585cb323b7774502a5", size = 558265, upload-time = "2025-02-21T15:01:28.979Z" }, - { url = "https://files.pythonhosted.org/packages/e6/0a/f2705530c42578f20ed0b5b90135eecb30eef6e2ba73e7ba69087fad2dba/rpds_py-0.23.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:a20cb698c4a59c534c6701b1c24a968ff2768b18ea2991f886bd8985ce17a89f", size = 585270, upload-time = "2025-02-21T15:01:30.879Z" }, - { url = "https://files.pythonhosted.org/packages/29/4e/3b597dc84ed82c3d757ac9aa620de224a94e06d2e102069795ae7e81c015/rpds_py-0.23.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:5e9c206a1abc27e0588cf8b7c8246e51f1a16a103734f7750830a1ccb63f557a", size = 553850, upload-time = "2025-02-21T15:01:32.269Z" }, - { url = "https://files.pythonhosted.org/packages/00/cc/6498b6f79e4375e6737247661e52a2d18f6accf4910e0c8da978674b4241/rpds_py-0.23.1-cp310-cp310-win32.whl", hash = "sha256:d9f75a06ecc68f159d5d7603b734e1ff6daa9497a929150f794013aa9f6e3f12", size = 220660, upload-time = "2025-02-21T15:01:33.643Z" }, - { url = "https://files.pythonhosted.org/packages/17/2b/08db023d23e8c7032c99d8d2a70d32e450a868ab73d16e3ff5290308a665/rpds_py-0.23.1-cp310-cp310-win_amd64.whl", hash = "sha256:f35eff113ad430b5272bbfc18ba111c66ff525828f24898b4e146eb479a2cdda", size = 232551, upload-time = "2025-02-21T15:01:35.529Z" }, - { url = "https://files.pythonhosted.org/packages/1c/67/6e5d4234bb9dee062ffca2a5f3c7cd38716317d6760ec235b175eed4de2c/rpds_py-0.23.1-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:b79f5ced71efd70414a9a80bbbfaa7160da307723166f09b69773153bf17c590", size = 372264, upload-time = "2025-02-21T15:01:37.918Z" }, - { url = "https://files.pythonhosted.org/packages/a7/0a/3dedb2daee8e783622427f5064e2d112751d8276ee73aa5409f000a132f4/rpds_py-0.23.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c9e799dac1ffbe7b10c1fd42fe4cd51371a549c6e108249bde9cd1200e8f59b4", size = 356883, upload-time = "2025-02-21T15:01:39.131Z" }, - { url = "https://files.pythonhosted.org/packages/ed/fc/e1acef44f9c24b05fe5434b235f165a63a52959ac655e3f7a55726cee1a4/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:721f9c4011b443b6e84505fc00cc7aadc9d1743f1c988e4c89353e19c4a968ee", size = 385624, upload-time = "2025-02-21T15:01:40.589Z" }, - { url = "https://files.pythonhosted.org/packages/97/0a/a05951f6465d01622720c03ef6ef31adfbe865653e05ed7c45837492f25e/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:f88626e3f5e57432e6191cd0c5d6d6b319b635e70b40be2ffba713053e5147dd", size = 391500, upload-time = "2025-02-21T15:01:42.584Z" }, - { url = "https://files.pythonhosted.org/packages/ea/2e/cca0583ec0690ea441dceae23c0673b99755710ea22f40bccf1e78f41481/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:285019078537949cecd0190f3690a0b0125ff743d6a53dfeb7a4e6787af154f5", size = 444869, upload-time = "2025-02-21T15:01:44.004Z" }, - { url = "https://files.pythonhosted.org/packages/cc/e6/95cda68b33a6d814d1e96b0e406d231ed16629101460d1740e92f03365e6/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b92f5654157de1379c509b15acec9d12ecf6e3bc1996571b6cb82a4302060447", size = 444930, upload-time = "2025-02-21T15:01:46.191Z" }, - { url = "https://files.pythonhosted.org/packages/5f/a7/e94cdb73411ae9c11414d3c7c9a6ad75d22ad4a8d094fb45a345ba9e3018/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e768267cbe051dd8d1c5305ba690bb153204a09bf2e3de3ae530de955f5b5580", size = 386254, upload-time = "2025-02-21T15:01:48.038Z" }, - { url = "https://files.pythonhosted.org/packages/dd/c5/a4a943d90a39e85efd1e04b1ad5129936786f9a9aa27bb7be8fc5d9d50c9/rpds_py-0.23.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:c5334a71f7dc1160382d45997e29f2637c02f8a26af41073189d79b95d3321f1", size = 417090, upload-time = "2025-02-21T15:01:50.252Z" }, - { url = "https://files.pythonhosted.org/packages/0c/a0/80d0013b12428d1fce0ab4e71829400b0a32caec12733c79e6109f843342/rpds_py-0.23.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d6adb81564af0cd428910f83fa7da46ce9ad47c56c0b22b50872bc4515d91966", size = 557639, upload-time = "2025-02-21T15:01:51.488Z" }, - { url = "https://files.pythonhosted.org/packages/a6/92/ec2e6980afb964a2cd7a99cbdef1f6c01116abe94b42cbe336ac93dd11c2/rpds_py-0.23.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:cafa48f2133d4daa028473ede7d81cd1b9f9e6925e9e4003ebdf77010ee02f35", size = 584572, upload-time = "2025-02-21T15:01:53.13Z" }, - { url = "https://files.pythonhosted.org/packages/3d/ce/75b6054db34a390789a82523790717b27c1bd735e453abb429a87c4f0f26/rpds_py-0.23.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:0fced9fd4a07a1ded1bac7e961ddd9753dd5d8b755ba8e05acba54a21f5f1522", size = 553028, upload-time = "2025-02-21T15:01:54.84Z" }, - { url = "https://files.pythonhosted.org/packages/cc/24/f45abe0418c06a5cba0f846e967aa27bac765acd927aabd857c21319b8cc/rpds_py-0.23.1-cp311-cp311-win32.whl", hash = "sha256:243241c95174b5fb7204c04595852fe3943cc41f47aa14c3828bc18cd9d3b2d6", size = 220862, upload-time = "2025-02-21T15:01:56.966Z" }, - { url = "https://files.pythonhosted.org/packages/2d/a6/3c0880e8bbfc36451ef30dc416266f6d2934705e468db5d21c8ba0ab6400/rpds_py-0.23.1-cp311-cp311-win_amd64.whl", hash = "sha256:11dd60b2ffddba85715d8a66bb39b95ddbe389ad2cfcf42c833f1bcde0878eaf", size = 232953, upload-time = "2025-02-21T15:02:00.297Z" }, - { url = "https://files.pythonhosted.org/packages/f3/8c/d17efccb9f5b9137ddea706664aebae694384ae1d5997c0202093e37185a/rpds_py-0.23.1-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:3902df19540e9af4cc0c3ae75974c65d2c156b9257e91f5101a51f99136d834c", size = 364369, upload-time = "2025-02-21T15:02:02.396Z" }, - { url = "https://files.pythonhosted.org/packages/6e/c0/ab030f696b5c573107115a88d8d73d80f03309e60952b64c584c70c659af/rpds_py-0.23.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:66f8d2a17e5838dd6fb9be6baaba8e75ae2f5fa6b6b755d597184bfcd3cb0eba", size = 349965, upload-time = "2025-02-21T15:02:04.527Z" }, - { url = "https://files.pythonhosted.org/packages/b3/55/b40170f5a079c4fb0b6a82b299689e66e744edca3c3375a8b160fb797660/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:112b8774b0b4ee22368fec42749b94366bd9b536f8f74c3d4175d4395f5cbd31", size = 389064, upload-time = "2025-02-21T15:02:06.633Z" }, - { url = "https://files.pythonhosted.org/packages/ab/1c/b03a912c59ec7c1e16b26e587b9dfa8ddff3b07851e781e8c46e908a365a/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:e0df046f2266e8586cf09d00588302a32923eb6386ced0ca5c9deade6af9a149", size = 397741, upload-time = "2025-02-21T15:02:08.195Z" }, - { url = "https://files.pythonhosted.org/packages/52/6f/151b90792b62fb6f87099bcc9044c626881fdd54e31bf98541f830b15cea/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0f3288930b947cbebe767f84cf618d2cbe0b13be476e749da0e6a009f986248c", size = 448784, upload-time = "2025-02-21T15:02:09.473Z" }, - { url = "https://files.pythonhosted.org/packages/71/2a/6de67c0c97ec7857e0e9e5cd7c52405af931b303eb1e5b9eff6c50fd9a2e/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ce473a2351c018b06dd8d30d5da8ab5a0831056cc53b2006e2a8028172c37ce5", size = 440203, upload-time = "2025-02-21T15:02:11.745Z" }, - { url = "https://files.pythonhosted.org/packages/db/5e/e759cd1c276d98a4b1f464b17a9bf66c65d29f8f85754e27e1467feaa7c3/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d550d7e9e7d8676b183b37d65b5cd8de13676a738973d330b59dc8312df9c5dc", size = 391611, upload-time = "2025-02-21T15:02:13.76Z" }, - { url = "https://files.pythonhosted.org/packages/1c/1e/2900358efcc0d9408c7289769cba4c0974d9db314aa884028ed7f7364f61/rpds_py-0.23.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:e14f86b871ea74c3fddc9a40e947d6a5d09def5adc2076ee61fb910a9014fb35", size = 423306, upload-time = "2025-02-21T15:02:15.096Z" }, - { url = "https://files.pythonhosted.org/packages/23/07/6c177e6d059f5d39689352d6c69a926ee4805ffdb6f06203570234d3d8f7/rpds_py-0.23.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:1bf5be5ba34e19be579ae873da515a2836a2166d8d7ee43be6ff909eda42b72b", size = 562323, upload-time = "2025-02-21T15:02:17.379Z" }, - { url = "https://files.pythonhosted.org/packages/70/e4/f9097fd1c02b516fff9850792161eb9fc20a2fd54762f3c69eae0bdb67cb/rpds_py-0.23.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:d7031d493c4465dbc8d40bd6cafefef4bd472b17db0ab94c53e7909ee781b9ef", size = 588351, upload-time = "2025-02-21T15:02:19.688Z" }, - { url = "https://files.pythonhosted.org/packages/87/39/5db3c6f326bfbe4576ae2af6435bd7555867d20ae690c786ff33659f293b/rpds_py-0.23.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:55ff4151cfd4bc635e51cfb1c59ac9f7196b256b12e3a57deb9e5742e65941ad", size = 557252, upload-time = "2025-02-21T15:02:21.875Z" }, - { url = "https://files.pythonhosted.org/packages/fd/14/2d5ad292f144fa79bafb78d2eb5b8a3a91c358b6065443cb9c49b5d1fedf/rpds_py-0.23.1-cp312-cp312-win32.whl", hash = "sha256:a9d3b728f5a5873d84cba997b9d617c6090ca5721caaa691f3b1a78c60adc057", size = 222181, upload-time = "2025-02-21T15:02:23.353Z" }, - { url = "https://files.pythonhosted.org/packages/a3/4f/0fce63e0f5cdd658e71e21abd17ac1bc9312741ebb8b3f74eeed2ebdf771/rpds_py-0.23.1-cp312-cp312-win_amd64.whl", hash = "sha256:b03a8d50b137ee758e4c73638b10747b7c39988eb8e6cd11abb7084266455165", size = 237426, upload-time = "2025-02-21T15:02:25.326Z" }, - { url = "https://files.pythonhosted.org/packages/13/9d/b8b2c0edffb0bed15be17b6d5ab06216f2f47f9ee49259c7e96a3ad4ca42/rpds_py-0.23.1-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:4caafd1a22e5eaa3732acb7672a497123354bef79a9d7ceed43387d25025e935", size = 363672, upload-time = "2025-02-21T15:02:26.528Z" }, - { url = "https://files.pythonhosted.org/packages/bd/c2/5056fa29e6894144d7ba4c938b9b0445f75836b87d2dd00ed4999dc45a8c/rpds_py-0.23.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:178f8a60fc24511c0eb756af741c476b87b610dba83270fce1e5a430204566a4", size = 349602, upload-time = "2025-02-21T15:02:27.82Z" }, - { url = "https://files.pythonhosted.org/packages/b0/bc/33779a1bb0ee32d8d706b173825aab75c628521d23ce72a7c1e6a6852f86/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c632419c3870507ca20a37c8f8f5352317aca097639e524ad129f58c125c61c6", size = 388746, upload-time = "2025-02-21T15:02:29.914Z" }, - { url = "https://files.pythonhosted.org/packages/62/0b/71db3e36b7780a619698ec82a9c87ab44ad7ca7f5480913e8a59ff76f050/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:698a79d295626ee292d1730bc2ef6e70a3ab135b1d79ada8fde3ed0047b65a10", size = 397076, upload-time = "2025-02-21T15:02:31.255Z" }, - { url = "https://files.pythonhosted.org/packages/bb/2e/494398f613edf77ba10a916b1ddea2acce42ab0e3b62e2c70ffc0757ce00/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:271fa2184cf28bdded86bb6217c8e08d3a169fe0bbe9be5e8d96e8476b707122", size = 448399, upload-time = "2025-02-21T15:02:32.637Z" }, - { url = "https://files.pythonhosted.org/packages/dd/53/4bd7f5779b1f463243ee5fdc83da04dd58a08f86e639dbffa7a35f969a84/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b91cceb5add79ee563bd1f70b30896bd63bc5f78a11c1f00a1e931729ca4f1f4", size = 439764, upload-time = "2025-02-21T15:02:34.372Z" }, - { url = "https://files.pythonhosted.org/packages/f6/55/b3c18c04a460d951bf8e91f2abf46ce5b6426fb69784166a6a25827cb90a/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f3a6cb95074777f1ecda2ca4fa7717caa9ee6e534f42b7575a8f0d4cb0c24013", size = 390662, upload-time = "2025-02-21T15:02:36.616Z" }, - { url = "https://files.pythonhosted.org/packages/2a/65/cc463044a3cbd616029b2aa87a651cdee8288d2fdd7780b2244845e934c1/rpds_py-0.23.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:50fb62f8d8364978478b12d5f03bf028c6bc2af04082479299139dc26edf4c64", size = 422680, upload-time = "2025-02-21T15:02:38.02Z" }, - { url = "https://files.pythonhosted.org/packages/fa/8e/1fa52990c7836d72e8d70cd7753f2362c72fbb0a49c1462e8c60e7176d0b/rpds_py-0.23.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c8f7e90b948dc9dcfff8003f1ea3af08b29c062f681c05fd798e36daa3f7e3e8", size = 561792, upload-time = "2025-02-21T15:02:41.184Z" }, - { url = "https://files.pythonhosted.org/packages/57/b8/fe3b612979b1a29d0c77f8585903d8b3a292604b26d4b300e228b8ac6360/rpds_py-0.23.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:5b98b6c953e5c2bda51ab4d5b4f172617d462eebc7f4bfdc7c7e6b423f6da957", size = 588127, upload-time = "2025-02-21T15:02:42.641Z" }, - { url = "https://files.pythonhosted.org/packages/44/2d/fde474de516bbc4b9b230f43c98e7f8acc5da7fc50ceed8e7af27553d346/rpds_py-0.23.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:2893d778d4671ee627bac4037a075168b2673c57186fb1a57e993465dbd79a93", size = 556981, upload-time = "2025-02-21T15:02:43.969Z" }, - { url = "https://files.pythonhosted.org/packages/18/57/767deeb27b81370bbab8f74ef6e68d26c4ea99018f3c71a570e506fede85/rpds_py-0.23.1-cp313-cp313-win32.whl", hash = "sha256:2cfa07c346a7ad07019c33fb9a63cf3acb1f5363c33bc73014e20d9fe8b01cdd", size = 221936, upload-time = "2025-02-21T15:02:45.339Z" }, - { url = "https://files.pythonhosted.org/packages/7d/6c/3474cfdd3cafe243f97ab8474ea8949236eb2a1a341ca55e75ce00cd03da/rpds_py-0.23.1-cp313-cp313-win_amd64.whl", hash = "sha256:3aaf141d39f45322e44fc2c742e4b8b4098ead5317e5f884770c8df0c332da70", size = 237145, upload-time = "2025-02-21T15:02:47.461Z" }, - { url = "https://files.pythonhosted.org/packages/ec/77/e985064c624230f61efa0423759bb066da56ebe40c654f8b5ba225bd5d63/rpds_py-0.23.1-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:759462b2d0aa5a04be5b3e37fb8183615f47014ae6b116e17036b131985cb731", size = 359623, upload-time = "2025-02-21T15:02:49.02Z" }, - { url = "https://files.pythonhosted.org/packages/62/d9/a33dcbf62b29e40559e012d525bae7d516757cf042cc9234bd34ca4b6aeb/rpds_py-0.23.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:3e9212f52074fc9d72cf242a84063787ab8e21e0950d4d6709886fb62bcb91d5", size = 345900, upload-time = "2025-02-21T15:02:51.268Z" }, - { url = "https://files.pythonhosted.org/packages/92/eb/f81a4be6397861adb2cb868bb6a28a33292c2dcac567d1dc575226055e55/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9e9f3a3ac919406bc0414bbbd76c6af99253c507150191ea79fab42fdb35982a", size = 386426, upload-time = "2025-02-21T15:02:53.489Z" }, - { url = "https://files.pythonhosted.org/packages/09/47/1f810c9b5e83be005341201b5389f1d240dfa440346ea7189f9b3fd6961d/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:c04ca91dda8a61584165825907f5c967ca09e9c65fe8966ee753a3f2b019fe1e", size = 392314, upload-time = "2025-02-21T15:02:55.721Z" }, - { url = "https://files.pythonhosted.org/packages/83/bd/bc95831432fd6c46ed8001f01af26de0763a059d6d7e6d69e3c5bf02917a/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4ab923167cfd945abb9b51a407407cf19f5bee35001221f2911dc85ffd35ff4f", size = 447706, upload-time = "2025-02-21T15:02:59.224Z" }, - { url = "https://files.pythonhosted.org/packages/19/3e/567c04c226b1802dc6dc82cad3d53e1fa0a773258571c74ac5d8fbde97ed/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ed6f011bedca8585787e5082cce081bac3d30f54520097b2411351b3574e1219", size = 437060, upload-time = "2025-02-21T15:03:00.953Z" }, - { url = "https://files.pythonhosted.org/packages/fe/77/a77d2c6afe27ae7d0d55fc32f6841502648070dc8d549fcc1e6d47ff8975/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6959bb9928c5c999aba4a3f5a6799d571ddc2c59ff49917ecf55be2bbb4e3722", size = 389347, upload-time = "2025-02-21T15:03:02.287Z" }, - { url = "https://files.pythonhosted.org/packages/3f/47/6b256ff20a74cfebeac790ab05586e0ac91f88e331125d4740a6c86fc26f/rpds_py-0.23.1-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:1ed7de3c86721b4e83ac440751329ec6a1102229aa18163f84c75b06b525ad7e", size = 415554, upload-time = "2025-02-21T15:03:03.816Z" }, - { url = "https://files.pythonhosted.org/packages/fc/29/d4572469a245bc9fc81e35166dca19fc5298d5c43e1a6dd64bf145045193/rpds_py-0.23.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:5fb89edee2fa237584e532fbf78f0ddd1e49a47c7c8cfa153ab4849dc72a35e6", size = 557418, upload-time = "2025-02-21T15:03:05.489Z" }, - { url = "https://files.pythonhosted.org/packages/9c/0a/68cf7228895b1a3f6f39f51b15830e62456795e61193d2c8b87fd48c60db/rpds_py-0.23.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:7e5413d2e2d86025e73f05510ad23dad5950ab8417b7fc6beaad99be8077138b", size = 583033, upload-time = "2025-02-21T15:03:07.493Z" }, - { url = "https://files.pythonhosted.org/packages/14/18/017ab41dcd6649ad5db7d00155b4c212b31ab05bd857d5ba73a1617984eb/rpds_py-0.23.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:d31ed4987d72aabdf521eddfb6a72988703c091cfc0064330b9e5f8d6a042ff5", size = 554880, upload-time = "2025-02-21T15:03:08.967Z" }, - { url = "https://files.pythonhosted.org/packages/2e/dd/17de89431268da8819d8d51ce67beac28d9b22fccf437bc5d6d2bcd1acdb/rpds_py-0.23.1-cp313-cp313t-win32.whl", hash = "sha256:f3429fb8e15b20961efca8c8b21432623d85db2228cc73fe22756c6637aa39e7", size = 219743, upload-time = "2025-02-21T15:03:10.429Z" }, - { url = "https://files.pythonhosted.org/packages/68/15/6d22d07e063ce5e9bfbd96db9ec2fbb4693591b4503e3a76996639474d02/rpds_py-0.23.1-cp313-cp313t-win_amd64.whl", hash = "sha256:d6f6512a90bd5cd9030a6237f5346f046c6f0e40af98657568fa45695d4de59d", size = 235415, upload-time = "2025-02-21T15:03:12.664Z" }, - { url = "https://files.pythonhosted.org/packages/95/a9/6fafd35fc6bac05f59bcbc800b57cef877911ff1c015397c519fec888642/rpds_py-0.23.1-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:c1f8afa346ccd59e4e5630d5abb67aba6a9812fddf764fd7eb11f382a345f8cc", size = 373463, upload-time = "2025-02-21T15:03:37.242Z" }, - { url = "https://files.pythonhosted.org/packages/5b/ac/44f00029b8fbe0903a19e9a87a9b86063bf8700df2cc58868373d378418c/rpds_py-0.23.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:fad784a31869747df4ac968a351e070c06ca377549e4ace94775aaa3ab33ee06", size = 358400, upload-time = "2025-02-21T15:03:38.752Z" }, - { url = "https://files.pythonhosted.org/packages/5e/9c/3da199346c68d785f10dccab123b74c8c5f73be3f742c9e33d1116e07931/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b5a96fcac2f18e5a0a23a75cd27ce2656c66c11c127b0318e508aab436b77428", size = 386815, upload-time = "2025-02-21T15:03:40.216Z" }, - { url = "https://files.pythonhosted.org/packages/d3/45/8f6533c33c0d33da8c2c8b2fb8f2ee90b23c05c679b86b0ac6aee4653749/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:3e77febf227a1dc3220159355dba68faa13f8dca9335d97504abf428469fb18b", size = 392974, upload-time = "2025-02-21T15:03:42.286Z" }, - { url = "https://files.pythonhosted.org/packages/ca/56/6a9ac1bf0455ba07385d8fe98c571c519b4f2000cff6581487bf9fab9272/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:26bb3e8de93443d55e2e748e9fd87deb5f8075ca7bc0502cfc8be8687d69a2ec", size = 446019, upload-time = "2025-02-21T15:03:43.811Z" }, - { url = "https://files.pythonhosted.org/packages/f4/83/5d9a3f9731cdccf49088bcc4ce821a5cf50bd1737cdad83e9959a7b9054d/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:db7707dde9143a67b8812c7e66aeb2d843fe33cc8e374170f4d2c50bd8f2472d", size = 445811, upload-time = "2025-02-21T15:03:45.218Z" }, - { url = "https://files.pythonhosted.org/packages/44/50/f2e0a98c62fc1fe68b176caca587714dc5c8bb2c3d1dd1eeb2bd4cc787ac/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1eedaaccc9bb66581d4ae7c50e15856e335e57ef2734dbc5fd8ba3e2a4ab3cb6", size = 388070, upload-time = "2025-02-21T15:03:46.905Z" }, - { url = "https://files.pythonhosted.org/packages/f2/d0/4981878f8f157e6dbea01d95e0119bf3d6b4c2c884fe64a9e6987f941104/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:28358c54fffadf0ae893f6c1050e8f8853e45df22483b7fff2f6ab6152f5d8bf", size = 419173, upload-time = "2025-02-21T15:03:48.552Z" }, - { url = "https://files.pythonhosted.org/packages/ce/13/fc971c470da96b270d2f64fedee987351bd935dc3016932a5cdcb1a88a2a/rpds_py-0.23.1-pp310-pypy310_pp73-musllinux_1_2_aarch64.whl", hash = "sha256:633462ef7e61d839171bf206551d5ab42b30b71cac8f10a64a662536e057fdef", size = 559048, upload-time = "2025-02-21T15:03:50.226Z" }, - { url = "https://files.pythonhosted.org/packages/42/02/be91e1de139ec8b4f9fec4192fd779ba48af281cfc762c0ca4c15b945484/rpds_py-0.23.1-pp310-pypy310_pp73-musllinux_1_2_i686.whl", hash = "sha256:a98f510d86f689fcb486dc59e6e363af04151e5260ad1bdddb5625c10f1e95f8", size = 584773, upload-time = "2025-02-21T15:03:52.678Z" }, - { url = "https://files.pythonhosted.org/packages/27/28/3af8a1956df3edc41d884267d766dc096496dafc83f02f764a475eca0b4a/rpds_py-0.23.1-pp310-pypy310_pp73-musllinux_1_2_x86_64.whl", hash = "sha256:e0397dd0b3955c61ef9b22838144aa4bef6f0796ba5cc8edfc64d468b93798b4", size = 555153, upload-time = "2025-02-21T15:03:55.21Z" }, - { url = "https://files.pythonhosted.org/packages/5e/bb/e45f51c4e1327dea3c72b846c6de129eebacb7a6cb309af7af35d0578c80/rpds_py-0.23.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:75307599f0d25bf6937248e5ac4e3bde5ea72ae6618623b86146ccc7845ed00b", size = 233827, upload-time = "2025-02-21T15:03:56.853Z" }, +version = "0.26.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/a5/aa/4456d84bbb54adc6a916fb10c9b374f78ac840337644e4a5eda229c81275/rpds_py-0.26.0.tar.gz", hash = "sha256:20dae58a859b0906f0685642e591056f1e787f3a8b39c8e8749a45dc7d26bdb0", size = 27385, upload-time = "2025-07-01T15:57:13.958Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/b9/31/1459645f036c3dfeacef89e8e5825e430c77dde8489f3b99eaafcd4a60f5/rpds_py-0.26.0-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:4c70c70f9169692b36307a95f3d8c0a9fcd79f7b4a383aad5eaa0e9718b79b37", size = 372466, upload-time = "2025-07-01T15:53:40.55Z" }, + { url = "https://files.pythonhosted.org/packages/dd/ff/3d0727f35836cc8773d3eeb9a46c40cc405854e36a8d2e951f3a8391c976/rpds_py-0.26.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:777c62479d12395bfb932944e61e915741e364c843afc3196b694db3d669fcd0", size = 357825, upload-time = "2025-07-01T15:53:42.247Z" }, + { url = "https://files.pythonhosted.org/packages/bf/ce/badc5e06120a54099ae287fa96d82cbb650a5f85cf247ffe19c7b157fd1f/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec671691e72dff75817386aa02d81e708b5a7ec0dec6669ec05213ff6b77e1bd", size = 381530, upload-time = "2025-07-01T15:53:43.585Z" }, + { url = "https://files.pythonhosted.org/packages/1e/a5/fa5d96a66c95d06c62d7a30707b6a4cfec696ab8ae280ee7be14e961e118/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:6a1cb5d6ce81379401bbb7f6dbe3d56de537fb8235979843f0d53bc2e9815a79", size = 396933, upload-time = "2025-07-01T15:53:45.78Z" }, + { url = "https://files.pythonhosted.org/packages/00/a7/7049d66750f18605c591a9db47d4a059e112a0c9ff8de8daf8fa0f446bba/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4f789e32fa1fb6a7bf890e0124e7b42d1e60d28ebff57fe806719abb75f0e9a3", size = 513973, upload-time = "2025-07-01T15:53:47.085Z" }, + { url = "https://files.pythonhosted.org/packages/0e/f1/528d02c7d6b29d29fac8fd784b354d3571cc2153f33f842599ef0cf20dd2/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9c55b0a669976cf258afd718de3d9ad1b7d1fe0a91cd1ab36f38b03d4d4aeaaf", size = 402293, upload-time = "2025-07-01T15:53:48.117Z" }, + { url = "https://files.pythonhosted.org/packages/15/93/fde36cd6e4685df2cd08508f6c45a841e82f5bb98c8d5ecf05649522acb5/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c70d9ec912802ecfd6cd390dadb34a9578b04f9bcb8e863d0a7598ba5e9e7ccc", size = 383787, upload-time = "2025-07-01T15:53:50.874Z" }, + { url = "https://files.pythonhosted.org/packages/69/f2/5007553aaba1dcae5d663143683c3dfd03d9395289f495f0aebc93e90f24/rpds_py-0.26.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:3021933c2cb7def39d927b9862292e0f4c75a13d7de70eb0ab06efed4c508c19", size = 416312, upload-time = "2025-07-01T15:53:52.046Z" }, + { url = "https://files.pythonhosted.org/packages/8f/a7/ce52c75c1e624a79e48a69e611f1c08844564e44c85db2b6f711d76d10ce/rpds_py-0.26.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:8a7898b6ca3b7d6659e55cdac825a2e58c638cbf335cde41f4619e290dd0ad11", size = 558403, upload-time = "2025-07-01T15:53:53.192Z" }, + { url = "https://files.pythonhosted.org/packages/79/d5/e119db99341cc75b538bf4cb80504129fa22ce216672fb2c28e4a101f4d9/rpds_py-0.26.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:12bff2ad9447188377f1b2794772f91fe68bb4bbfa5a39d7941fbebdbf8c500f", size = 588323, upload-time = "2025-07-01T15:53:54.336Z" }, + { url = "https://files.pythonhosted.org/packages/93/94/d28272a0b02f5fe24c78c20e13bbcb95f03dc1451b68e7830ca040c60bd6/rpds_py-0.26.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:191aa858f7d4902e975d4cf2f2d9243816c91e9605070aeb09c0a800d187e323", size = 554541, upload-time = "2025-07-01T15:53:55.469Z" }, + { url = "https://files.pythonhosted.org/packages/93/e0/8c41166602f1b791da892d976057eba30685486d2e2c061ce234679c922b/rpds_py-0.26.0-cp310-cp310-win32.whl", hash = "sha256:b37a04d9f52cb76b6b78f35109b513f6519efb481d8ca4c321f6a3b9580b3f45", size = 220442, upload-time = "2025-07-01T15:53:56.524Z" }, + { url = "https://files.pythonhosted.org/packages/87/f0/509736bb752a7ab50fb0270c2a4134d671a7b3038030837e5536c3de0e0b/rpds_py-0.26.0-cp310-cp310-win_amd64.whl", hash = "sha256:38721d4c9edd3eb6670437d8d5e2070063f305bfa2d5aa4278c51cedcd508a84", size = 231314, upload-time = "2025-07-01T15:53:57.842Z" }, + { url = "https://files.pythonhosted.org/packages/09/4c/4ee8f7e512030ff79fda1df3243c88d70fc874634e2dbe5df13ba4210078/rpds_py-0.26.0-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:9e8cb77286025bdb21be2941d64ac6ca016130bfdcd228739e8ab137eb4406ed", size = 372610, upload-time = "2025-07-01T15:53:58.844Z" }, + { url = "https://files.pythonhosted.org/packages/fa/9d/3dc16be00f14fc1f03c71b1d67c8df98263ab2710a2fbd65a6193214a527/rpds_py-0.26.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5e09330b21d98adc8ccb2dbb9fc6cb434e8908d4c119aeaa772cb1caab5440a0", size = 358032, upload-time = "2025-07-01T15:53:59.985Z" }, + { url = "https://files.pythonhosted.org/packages/e7/5a/7f1bf8f045da2866324a08ae80af63e64e7bfaf83bd31f865a7b91a58601/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2c9c1b92b774b2e68d11193dc39620d62fd8ab33f0a3c77ecdabe19c179cdbc1", size = 381525, upload-time = "2025-07-01T15:54:01.162Z" }, + { url = "https://files.pythonhosted.org/packages/45/8a/04479398c755a066ace10e3d158866beb600867cacae194c50ffa783abd0/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:824e6d3503ab990d7090768e4dfd9e840837bae057f212ff9f4f05ec6d1975e7", size = 397089, upload-time = "2025-07-01T15:54:02.319Z" }, + { url = "https://files.pythonhosted.org/packages/72/88/9203f47268db488a1b6d469d69c12201ede776bb728b9d9f29dbfd7df406/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8ad7fd2258228bf288f2331f0a6148ad0186b2e3643055ed0db30990e59817a6", size = 514255, upload-time = "2025-07-01T15:54:03.38Z" }, + { url = "https://files.pythonhosted.org/packages/f5/b4/01ce5d1e853ddf81fbbd4311ab1eff0b3cf162d559288d10fd127e2588b5/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0dc23bbb3e06ec1ea72d515fb572c1fea59695aefbffb106501138762e1e915e", size = 402283, upload-time = "2025-07-01T15:54:04.923Z" }, + { url = "https://files.pythonhosted.org/packages/34/a2/004c99936997bfc644d590a9defd9e9c93f8286568f9c16cdaf3e14429a7/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d80bf832ac7b1920ee29a426cdca335f96a2b5caa839811803e999b41ba9030d", size = 383881, upload-time = "2025-07-01T15:54:06.482Z" }, + { url = "https://files.pythonhosted.org/packages/05/1b/ef5fba4a8f81ce04c427bfd96223f92f05e6cd72291ce9d7523db3b03a6c/rpds_py-0.26.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:0919f38f5542c0a87e7b4afcafab6fd2c15386632d249e9a087498571250abe3", size = 415822, upload-time = "2025-07-01T15:54:07.605Z" }, + { url = "https://files.pythonhosted.org/packages/16/80/5c54195aec456b292f7bd8aa61741c8232964063fd8a75fdde9c1e982328/rpds_py-0.26.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d422b945683e409000c888e384546dbab9009bb92f7c0b456e217988cf316107", size = 558347, upload-time = "2025-07-01T15:54:08.591Z" }, + { url = "https://files.pythonhosted.org/packages/f2/1c/1845c1b1fd6d827187c43afe1841d91678d7241cbdb5420a4c6de180a538/rpds_py-0.26.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:77a7711fa562ba2da1aa757e11024ad6d93bad6ad7ede5afb9af144623e5f76a", size = 587956, upload-time = "2025-07-01T15:54:09.963Z" }, + { url = "https://files.pythonhosted.org/packages/2e/ff/9e979329dd131aa73a438c077252ddabd7df6d1a7ad7b9aacf6261f10faa/rpds_py-0.26.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:238e8c8610cb7c29460e37184f6799547f7e09e6a9bdbdab4e8edb90986a2318", size = 554363, upload-time = "2025-07-01T15:54:11.073Z" }, + { url = "https://files.pythonhosted.org/packages/00/8b/d78cfe034b71ffbe72873a136e71acc7a831a03e37771cfe59f33f6de8a2/rpds_py-0.26.0-cp311-cp311-win32.whl", hash = "sha256:893b022bfbdf26d7bedb083efeea624e8550ca6eb98bf7fea30211ce95b9201a", size = 220123, upload-time = "2025-07-01T15:54:12.382Z" }, + { url = "https://files.pythonhosted.org/packages/94/c1/3c8c94c7dd3905dbfde768381ce98778500a80db9924731d87ddcdb117e9/rpds_py-0.26.0-cp311-cp311-win_amd64.whl", hash = "sha256:87a5531de9f71aceb8af041d72fc4cab4943648d91875ed56d2e629bef6d4c03", size = 231732, upload-time = "2025-07-01T15:54:13.434Z" }, + { url = "https://files.pythonhosted.org/packages/67/93/e936fbed1b734eabf36ccb5d93c6a2e9246fbb13c1da011624b7286fae3e/rpds_py-0.26.0-cp311-cp311-win_arm64.whl", hash = "sha256:de2713f48c1ad57f89ac25b3cb7daed2156d8e822cf0eca9b96a6f990718cc41", size = 221917, upload-time = "2025-07-01T15:54:14.559Z" }, + { url = "https://files.pythonhosted.org/packages/ea/86/90eb87c6f87085868bd077c7a9938006eb1ce19ed4d06944a90d3560fce2/rpds_py-0.26.0-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:894514d47e012e794f1350f076c427d2347ebf82f9b958d554d12819849a369d", size = 363933, upload-time = "2025-07-01T15:54:15.734Z" }, + { url = "https://files.pythonhosted.org/packages/63/78/4469f24d34636242c924626082b9586f064ada0b5dbb1e9d096ee7a8e0c6/rpds_py-0.26.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:fc921b96fa95a097add244da36a1d9e4f3039160d1d30f1b35837bf108c21136", size = 350447, upload-time = "2025-07-01T15:54:16.922Z" }, + { url = "https://files.pythonhosted.org/packages/ad/91/c448ed45efdfdade82348d5e7995e15612754826ea640afc20915119734f/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3e1157659470aa42a75448b6e943c895be8c70531c43cb78b9ba990778955582", size = 384711, upload-time = "2025-07-01T15:54:18.101Z" }, + { url = "https://files.pythonhosted.org/packages/ec/43/e5c86fef4be7f49828bdd4ecc8931f0287b1152c0bb0163049b3218740e7/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:521ccf56f45bb3a791182dc6b88ae5f8fa079dd705ee42138c76deb1238e554e", size = 400865, upload-time = "2025-07-01T15:54:19.295Z" }, + { url = "https://files.pythonhosted.org/packages/55/34/e00f726a4d44f22d5c5fe2e5ddd3ac3d7fd3f74a175607781fbdd06fe375/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9def736773fd56b305c0eef698be5192c77bfa30d55a0e5885f80126c4831a15", size = 517763, upload-time = "2025-07-01T15:54:20.858Z" }, + { url = "https://files.pythonhosted.org/packages/52/1c/52dc20c31b147af724b16104500fba13e60123ea0334beba7b40e33354b4/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:cdad4ea3b4513b475e027be79e5a0ceac8ee1c113a1a11e5edc3c30c29f964d8", size = 406651, upload-time = "2025-07-01T15:54:22.508Z" }, + { url = "https://files.pythonhosted.org/packages/2e/77/87d7bfabfc4e821caa35481a2ff6ae0b73e6a391bb6b343db2c91c2b9844/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:82b165b07f416bdccf5c84546a484cc8f15137ca38325403864bfdf2b5b72f6a", size = 386079, upload-time = "2025-07-01T15:54:23.987Z" }, + { url = "https://files.pythonhosted.org/packages/e3/d4/7f2200c2d3ee145b65b3cddc4310d51f7da6a26634f3ac87125fd789152a/rpds_py-0.26.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:d04cab0a54b9dba4d278fe955a1390da3cf71f57feb78ddc7cb67cbe0bd30323", size = 421379, upload-time = "2025-07-01T15:54:25.073Z" }, + { url = "https://files.pythonhosted.org/packages/ae/13/9fdd428b9c820869924ab62236b8688b122baa22d23efdd1c566938a39ba/rpds_py-0.26.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:79061ba1a11b6a12743a2b0f72a46aa2758613d454aa6ba4f5a265cc48850158", size = 562033, upload-time = "2025-07-01T15:54:26.225Z" }, + { url = "https://files.pythonhosted.org/packages/f3/e1/b69686c3bcbe775abac3a4c1c30a164a2076d28df7926041f6c0eb5e8d28/rpds_py-0.26.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:f405c93675d8d4c5ac87364bb38d06c988e11028a64b52a47158a355079661f3", size = 591639, upload-time = "2025-07-01T15:54:27.424Z" }, + { url = "https://files.pythonhosted.org/packages/5c/c9/1e3d8c8863c84a90197ac577bbc3d796a92502124c27092413426f670990/rpds_py-0.26.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:dafd4c44b74aa4bed4b250f1aed165b8ef5de743bcca3b88fc9619b6087093d2", size = 557105, upload-time = "2025-07-01T15:54:29.93Z" }, + { url = "https://files.pythonhosted.org/packages/9f/c5/90c569649057622959f6dcc40f7b516539608a414dfd54b8d77e3b201ac0/rpds_py-0.26.0-cp312-cp312-win32.whl", hash = "sha256:3da5852aad63fa0c6f836f3359647870e21ea96cf433eb393ffa45263a170d44", size = 223272, upload-time = "2025-07-01T15:54:31.128Z" }, + { url = "https://files.pythonhosted.org/packages/7d/16/19f5d9f2a556cfed454eebe4d354c38d51c20f3db69e7b4ce6cff904905d/rpds_py-0.26.0-cp312-cp312-win_amd64.whl", hash = "sha256:cf47cfdabc2194a669dcf7a8dbba62e37a04c5041d2125fae0233b720da6f05c", size = 234995, upload-time = "2025-07-01T15:54:32.195Z" }, + { url = "https://files.pythonhosted.org/packages/83/f0/7935e40b529c0e752dfaa7880224771b51175fce08b41ab4a92eb2fbdc7f/rpds_py-0.26.0-cp312-cp312-win_arm64.whl", hash = "sha256:20ab1ae4fa534f73647aad289003f1104092890849e0266271351922ed5574f8", size = 223198, upload-time = "2025-07-01T15:54:33.271Z" }, + { url = "https://files.pythonhosted.org/packages/6a/67/bb62d0109493b12b1c6ab00de7a5566aa84c0e44217c2d94bee1bd370da9/rpds_py-0.26.0-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:696764a5be111b036256c0b18cd29783fab22154690fc698062fc1b0084b511d", size = 363917, upload-time = "2025-07-01T15:54:34.755Z" }, + { url = "https://files.pythonhosted.org/packages/4b/f3/34e6ae1925a5706c0f002a8d2d7f172373b855768149796af87bd65dcdb9/rpds_py-0.26.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1e6c15d2080a63aaed876e228efe4f814bc7889c63b1e112ad46fdc8b368b9e1", size = 350073, upload-time = "2025-07-01T15:54:36.292Z" }, + { url = "https://files.pythonhosted.org/packages/75/83/1953a9d4f4e4de7fd0533733e041c28135f3c21485faaef56a8aadbd96b5/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:390e3170babf42462739a93321e657444f0862c6d722a291accc46f9d21ed04e", size = 384214, upload-time = "2025-07-01T15:54:37.469Z" }, + { url = "https://files.pythonhosted.org/packages/48/0e/983ed1b792b3322ea1d065e67f4b230f3b96025f5ce3878cc40af09b7533/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:7da84c2c74c0f5bc97d853d9e17bb83e2dcafcff0dc48286916001cc114379a1", size = 400113, upload-time = "2025-07-01T15:54:38.954Z" }, + { url = "https://files.pythonhosted.org/packages/69/7f/36c0925fff6f660a80be259c5b4f5e53a16851f946eb080351d057698528/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4c5fe114a6dd480a510b6d3661d09d67d1622c4bf20660a474507aaee7eeeee9", size = 515189, upload-time = "2025-07-01T15:54:40.57Z" }, + { url = "https://files.pythonhosted.org/packages/13/45/cbf07fc03ba7a9b54662c9badb58294ecfb24f828b9732970bd1a431ed5c/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3100b3090269f3a7ea727b06a6080d4eb7439dca4c0e91a07c5d133bb1727ea7", size = 406998, upload-time = "2025-07-01T15:54:43.025Z" }, + { url = "https://files.pythonhosted.org/packages/6c/b0/8fa5e36e58657997873fd6a1cf621285ca822ca75b4b3434ead047daa307/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2c03c9b0c64afd0320ae57de4c982801271c0c211aa2d37f3003ff5feb75bb04", size = 385903, upload-time = "2025-07-01T15:54:44.752Z" }, + { url = "https://files.pythonhosted.org/packages/4b/f7/b25437772f9f57d7a9fbd73ed86d0dcd76b4c7c6998348c070d90f23e315/rpds_py-0.26.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:5963b72ccd199ade6ee493723d18a3f21ba7d5b957017607f815788cef50eaf1", size = 419785, upload-time = "2025-07-01T15:54:46.043Z" }, + { url = "https://files.pythonhosted.org/packages/a7/6b/63ffa55743dfcb4baf2e9e77a0b11f7f97ed96a54558fcb5717a4b2cd732/rpds_py-0.26.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:9da4e873860ad5bab3291438525cae80169daecbfafe5657f7f5fb4d6b3f96b9", size = 561329, upload-time = "2025-07-01T15:54:47.64Z" }, + { url = "https://files.pythonhosted.org/packages/2f/07/1f4f5e2886c480a2346b1e6759c00278b8a69e697ae952d82ae2e6ee5db0/rpds_py-0.26.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:5afaddaa8e8c7f1f7b4c5c725c0070b6eed0228f705b90a1732a48e84350f4e9", size = 590875, upload-time = "2025-07-01T15:54:48.9Z" }, + { url = "https://files.pythonhosted.org/packages/cc/bc/e6639f1b91c3a55f8c41b47d73e6307051b6e246254a827ede730624c0f8/rpds_py-0.26.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4916dc96489616a6f9667e7526af8fa693c0fdb4f3acb0e5d9f4400eb06a47ba", size = 556636, upload-time = "2025-07-01T15:54:50.619Z" }, + { url = "https://files.pythonhosted.org/packages/05/4c/b3917c45566f9f9a209d38d9b54a1833f2bb1032a3e04c66f75726f28876/rpds_py-0.26.0-cp313-cp313-win32.whl", hash = "sha256:2a343f91b17097c546b93f7999976fd6c9d5900617aa848c81d794e062ab302b", size = 222663, upload-time = "2025-07-01T15:54:52.023Z" }, + { url = "https://files.pythonhosted.org/packages/e0/0b/0851bdd6025775aaa2365bb8de0697ee2558184c800bfef8d7aef5ccde58/rpds_py-0.26.0-cp313-cp313-win_amd64.whl", hash = "sha256:0a0b60701f2300c81b2ac88a5fb893ccfa408e1c4a555a77f908a2596eb875a5", size = 234428, upload-time = "2025-07-01T15:54:53.692Z" }, + { url = "https://files.pythonhosted.org/packages/ed/e8/a47c64ed53149c75fb581e14a237b7b7cd18217e969c30d474d335105622/rpds_py-0.26.0-cp313-cp313-win_arm64.whl", hash = "sha256:257d011919f133a4746958257f2c75238e3ff54255acd5e3e11f3ff41fd14256", size = 222571, upload-time = "2025-07-01T15:54:54.822Z" }, + { url = "https://files.pythonhosted.org/packages/89/bf/3d970ba2e2bcd17d2912cb42874107390f72873e38e79267224110de5e61/rpds_py-0.26.0-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:529c8156d7506fba5740e05da8795688f87119cce330c244519cf706a4a3d618", size = 360475, upload-time = "2025-07-01T15:54:56.228Z" }, + { url = "https://files.pythonhosted.org/packages/82/9f/283e7e2979fc4ec2d8ecee506d5a3675fce5ed9b4b7cb387ea5d37c2f18d/rpds_py-0.26.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:f53ec51f9d24e9638a40cabb95078ade8c99251945dad8d57bf4aabe86ecee35", size = 346692, upload-time = "2025-07-01T15:54:58.561Z" }, + { url = "https://files.pythonhosted.org/packages/e3/03/7e50423c04d78daf391da3cc4330bdb97042fc192a58b186f2d5deb7befd/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7ab504c4d654e4a29558eaa5bb8cea5fdc1703ea60a8099ffd9c758472cf913f", size = 379415, upload-time = "2025-07-01T15:54:59.751Z" }, + { url = "https://files.pythonhosted.org/packages/57/00/d11ee60d4d3b16808432417951c63df803afb0e0fc672b5e8d07e9edaaae/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:fd0641abca296bc1a00183fe44f7fced8807ed49d501f188faa642d0e4975b83", size = 391783, upload-time = "2025-07-01T15:55:00.898Z" }, + { url = "https://files.pythonhosted.org/packages/08/b3/1069c394d9c0d6d23c5b522e1f6546b65793a22950f6e0210adcc6f97c3e/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:69b312fecc1d017b5327afa81d4da1480f51c68810963a7336d92203dbb3d4f1", size = 512844, upload-time = "2025-07-01T15:55:02.201Z" }, + { url = "https://files.pythonhosted.org/packages/08/3b/c4fbf0926800ed70b2c245ceca99c49f066456755f5d6eb8863c2c51e6d0/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c741107203954f6fc34d3066d213d0a0c40f7bb5aafd698fb39888af277c70d8", size = 402105, upload-time = "2025-07-01T15:55:03.698Z" }, + { url = "https://files.pythonhosted.org/packages/1c/b0/db69b52ca07413e568dae9dc674627a22297abb144c4d6022c6d78f1e5cc/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc3e55a7db08dc9a6ed5fb7103019d2c1a38a349ac41901f9f66d7f95750942f", size = 383440, upload-time = "2025-07-01T15:55:05.398Z" }, + { url = "https://files.pythonhosted.org/packages/4c/e1/c65255ad5b63903e56b3bb3ff9dcc3f4f5c3badde5d08c741ee03903e951/rpds_py-0.26.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:9e851920caab2dbcae311fd28f4313c6953993893eb5c1bb367ec69d9a39e7ed", size = 412759, upload-time = "2025-07-01T15:55:08.316Z" }, + { url = "https://files.pythonhosted.org/packages/e4/22/bb731077872377a93c6e93b8a9487d0406c70208985831034ccdeed39c8e/rpds_py-0.26.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:dfbf280da5f876d0b00c81f26bedce274e72a678c28845453885a9b3c22ae632", size = 556032, upload-time = "2025-07-01T15:55:09.52Z" }, + { url = "https://files.pythonhosted.org/packages/e0/8b/393322ce7bac5c4530fb96fc79cc9ea2f83e968ff5f6e873f905c493e1c4/rpds_py-0.26.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:1cc81d14ddfa53d7f3906694d35d54d9d3f850ef8e4e99ee68bc0d1e5fed9a9c", size = 585416, upload-time = "2025-07-01T15:55:11.216Z" }, + { url = "https://files.pythonhosted.org/packages/49/ae/769dc372211835bf759319a7aae70525c6eb523e3371842c65b7ef41c9c6/rpds_py-0.26.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:dca83c498b4650a91efcf7b88d669b170256bf8017a5db6f3e06c2bf031f57e0", size = 554049, upload-time = "2025-07-01T15:55:13.004Z" }, + { url = "https://files.pythonhosted.org/packages/6b/f9/4c43f9cc203d6ba44ce3146246cdc38619d92c7bd7bad4946a3491bd5b70/rpds_py-0.26.0-cp313-cp313t-win32.whl", hash = "sha256:4d11382bcaf12f80b51d790dee295c56a159633a8e81e6323b16e55d81ae37e9", size = 218428, upload-time = "2025-07-01T15:55:14.486Z" }, + { url = "https://files.pythonhosted.org/packages/7e/8b/9286b7e822036a4a977f2f1e851c7345c20528dbd56b687bb67ed68a8ede/rpds_py-0.26.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ff110acded3c22c033e637dd8896e411c7d3a11289b2edf041f86663dbc791e9", size = 231524, upload-time = "2025-07-01T15:55:15.745Z" }, + { url = "https://files.pythonhosted.org/packages/55/07/029b7c45db910c74e182de626dfdae0ad489a949d84a468465cd0ca36355/rpds_py-0.26.0-cp314-cp314-macosx_10_12_x86_64.whl", hash = "sha256:da619979df60a940cd434084355c514c25cf8eb4cf9a508510682f6c851a4f7a", size = 364292, upload-time = "2025-07-01T15:55:17.001Z" }, + { url = "https://files.pythonhosted.org/packages/13/d1/9b3d3f986216b4d1f584878dca15ce4797aaf5d372d738974ba737bf68d6/rpds_py-0.26.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:ea89a2458a1a75f87caabefe789c87539ea4e43b40f18cff526052e35bbb4fdf", size = 350334, upload-time = "2025-07-01T15:55:18.922Z" }, + { url = "https://files.pythonhosted.org/packages/18/98/16d5e7bc9ec715fa9668731d0cf97f6b032724e61696e2db3d47aeb89214/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:feac1045b3327a45944e7dcbeb57530339f6b17baff154df51ef8b0da34c8c12", size = 384875, upload-time = "2025-07-01T15:55:20.399Z" }, + { url = "https://files.pythonhosted.org/packages/f9/13/aa5e2b1ec5ab0e86a5c464d53514c0467bec6ba2507027d35fc81818358e/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:b818a592bd69bfe437ee8368603d4a2d928c34cffcdf77c2e761a759ffd17d20", size = 399993, upload-time = "2025-07-01T15:55:21.729Z" }, + { url = "https://files.pythonhosted.org/packages/17/03/8021810b0e97923abdbab6474c8b77c69bcb4b2c58330777df9ff69dc559/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1a8b0dd8648709b62d9372fc00a57466f5fdeefed666afe3fea5a6c9539a0331", size = 516683, upload-time = "2025-07-01T15:55:22.918Z" }, + { url = "https://files.pythonhosted.org/packages/dc/b1/da8e61c87c2f3d836954239fdbbfb477bb7b54d74974d8f6fcb34342d166/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6d3498ad0df07d81112aa6ec6c95a7e7b1ae00929fb73e7ebee0f3faaeabad2f", size = 408825, upload-time = "2025-07-01T15:55:24.207Z" }, + { url = "https://files.pythonhosted.org/packages/38/bc/1fc173edaaa0e52c94b02a655db20697cb5fa954ad5a8e15a2c784c5cbdd/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:24a4146ccb15be237fdef10f331c568e1b0e505f8c8c9ed5d67759dac58ac246", size = 387292, upload-time = "2025-07-01T15:55:25.554Z" }, + { url = "https://files.pythonhosted.org/packages/7c/eb/3a9bb4bd90867d21916f253caf4f0d0be7098671b6715ad1cead9fe7bab9/rpds_py-0.26.0-cp314-cp314-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:a9a63785467b2d73635957d32a4f6e73d5e4df497a16a6392fa066b753e87387", size = 420435, upload-time = "2025-07-01T15:55:27.798Z" }, + { url = "https://files.pythonhosted.org/packages/cd/16/e066dcdb56f5632713445271a3f8d3d0b426d51ae9c0cca387799df58b02/rpds_py-0.26.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:de4ed93a8c91debfd5a047be327b7cc8b0cc6afe32a716bbbc4aedca9e2a83af", size = 562410, upload-time = "2025-07-01T15:55:29.057Z" }, + { url = "https://files.pythonhosted.org/packages/60/22/ddbdec7eb82a0dc2e455be44c97c71c232983e21349836ce9f272e8a3c29/rpds_py-0.26.0-cp314-cp314-musllinux_1_2_i686.whl", hash = "sha256:caf51943715b12af827696ec395bfa68f090a4c1a1d2509eb4e2cb69abbbdb33", size = 590724, upload-time = "2025-07-01T15:55:30.719Z" }, + { url = "https://files.pythonhosted.org/packages/2c/b4/95744085e65b7187d83f2fcb0bef70716a1ea0a9e5d8f7f39a86e5d83424/rpds_py-0.26.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:4a59e5bc386de021f56337f757301b337d7ab58baa40174fb150accd480bc953", size = 558285, upload-time = "2025-07-01T15:55:31.981Z" }, + { url = "https://files.pythonhosted.org/packages/37/37/6309a75e464d1da2559446f9c811aa4d16343cebe3dbb73701e63f760caa/rpds_py-0.26.0-cp314-cp314-win32.whl", hash = "sha256:92c8db839367ef16a662478f0a2fe13e15f2227da3c1430a782ad0f6ee009ec9", size = 223459, upload-time = "2025-07-01T15:55:33.312Z" }, + { url = "https://files.pythonhosted.org/packages/d9/6f/8e9c11214c46098b1d1391b7e02b70bb689ab963db3b19540cba17315291/rpds_py-0.26.0-cp314-cp314-win_amd64.whl", hash = "sha256:b0afb8cdd034150d4d9f53926226ed27ad15b7f465e93d7468caaf5eafae0d37", size = 236083, upload-time = "2025-07-01T15:55:34.933Z" }, + { url = "https://files.pythonhosted.org/packages/47/af/9c4638994dd623d51c39892edd9d08e8be8220a4b7e874fa02c2d6e91955/rpds_py-0.26.0-cp314-cp314-win_arm64.whl", hash = "sha256:ca3f059f4ba485d90c8dc75cb5ca897e15325e4e609812ce57f896607c1c0867", size = 223291, upload-time = "2025-07-01T15:55:36.202Z" }, + { url = "https://files.pythonhosted.org/packages/4d/db/669a241144460474aab03e254326b32c42def83eb23458a10d163cb9b5ce/rpds_py-0.26.0-cp314-cp314t-macosx_10_12_x86_64.whl", hash = "sha256:5afea17ab3a126006dc2f293b14ffc7ef3c85336cf451564a0515ed7648033da", size = 361445, upload-time = "2025-07-01T15:55:37.483Z" }, + { url = "https://files.pythonhosted.org/packages/3b/2d/133f61cc5807c6c2fd086a46df0eb8f63a23f5df8306ff9f6d0fd168fecc/rpds_py-0.26.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:69f0c0a3df7fd3a7eec50a00396104bb9a843ea6d45fcc31c2d5243446ffd7a7", size = 347206, upload-time = "2025-07-01T15:55:38.828Z" }, + { url = "https://files.pythonhosted.org/packages/05/bf/0e8fb4c05f70273469eecf82f6ccf37248558526a45321644826555db31b/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:801a71f70f9813e82d2513c9a96532551fce1e278ec0c64610992c49c04c2dad", size = 380330, upload-time = "2025-07-01T15:55:40.175Z" }, + { url = "https://files.pythonhosted.org/packages/d4/a8/060d24185d8b24d3923322f8d0ede16df4ade226a74e747b8c7c978e3dd3/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:df52098cde6d5e02fa75c1f6244f07971773adb4a26625edd5c18fee906fa84d", size = 392254, upload-time = "2025-07-01T15:55:42.015Z" }, + { url = "https://files.pythonhosted.org/packages/b9/7b/7c2e8a9ee3e6bc0bae26bf29f5219955ca2fbb761dca996a83f5d2f773fe/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9bc596b30f86dc6f0929499c9e574601679d0341a0108c25b9b358a042f51bca", size = 516094, upload-time = "2025-07-01T15:55:43.603Z" }, + { url = "https://files.pythonhosted.org/packages/75/d6/f61cafbed8ba1499b9af9f1777a2a199cd888f74a96133d8833ce5eaa9c5/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9dfbe56b299cf5875b68eb6f0ebaadc9cac520a1989cac0db0765abfb3709c19", size = 402889, upload-time = "2025-07-01T15:55:45.275Z" }, + { url = "https://files.pythonhosted.org/packages/92/19/c8ac0a8a8df2dd30cdec27f69298a5c13e9029500d6d76718130f5e5be10/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ac64f4b2bdb4ea622175c9ab7cf09444e412e22c0e02e906978b3b488af5fde8", size = 384301, upload-time = "2025-07-01T15:55:47.098Z" }, + { url = "https://files.pythonhosted.org/packages/41/e1/6b1859898bc292a9ce5776016c7312b672da00e25cec74d7beced1027286/rpds_py-0.26.0-cp314-cp314t-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:181ef9b6bbf9845a264f9aa45c31836e9f3c1f13be565d0d010e964c661d1e2b", size = 412891, upload-time = "2025-07-01T15:55:48.412Z" }, + { url = "https://files.pythonhosted.org/packages/ef/b9/ceb39af29913c07966a61367b3c08b4f71fad841e32c6b59a129d5974698/rpds_py-0.26.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:49028aa684c144ea502a8e847d23aed5e4c2ef7cadfa7d5eaafcb40864844b7a", size = 557044, upload-time = "2025-07-01T15:55:49.816Z" }, + { url = "https://files.pythonhosted.org/packages/2f/27/35637b98380731a521f8ec4f3fd94e477964f04f6b2f8f7af8a2d889a4af/rpds_py-0.26.0-cp314-cp314t-musllinux_1_2_i686.whl", hash = "sha256:e5d524d68a474a9688336045bbf76cb0def88549c1b2ad9dbfec1fb7cfbe9170", size = 585774, upload-time = "2025-07-01T15:55:51.192Z" }, + { url = "https://files.pythonhosted.org/packages/52/d9/3f0f105420fecd18551b678c9a6ce60bd23986098b252a56d35781b3e7e9/rpds_py-0.26.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:c1851f429b822831bd2edcbe0cfd12ee9ea77868f8d3daf267b189371671c80e", size = 554886, upload-time = "2025-07-01T15:55:52.541Z" }, + { url = "https://files.pythonhosted.org/packages/6b/c5/347c056a90dc8dd9bc240a08c527315008e1b5042e7a4cf4ac027be9d38a/rpds_py-0.26.0-cp314-cp314t-win32.whl", hash = "sha256:7bdb17009696214c3b66bb3590c6d62e14ac5935e53e929bcdbc5a495987a84f", size = 219027, upload-time = "2025-07-01T15:55:53.874Z" }, + { url = "https://files.pythonhosted.org/packages/75/04/5302cea1aa26d886d34cadbf2dc77d90d7737e576c0065f357b96dc7a1a6/rpds_py-0.26.0-cp314-cp314t-win_amd64.whl", hash = "sha256:f14440b9573a6f76b4ee4770c13f0b5921f71dde3b6fcb8dabbefd13b7fe05d7", size = 232821, upload-time = "2025-07-01T15:55:55.167Z" }, + { url = "https://files.pythonhosted.org/packages/ef/9a/1f033b0b31253d03d785b0cd905bc127e555ab496ea6b4c7c2e1f951f2fd/rpds_py-0.26.0-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:3c0909c5234543ada2515c05dc08595b08d621ba919629e94427e8e03539c958", size = 373226, upload-time = "2025-07-01T15:56:16.578Z" }, + { url = "https://files.pythonhosted.org/packages/58/29/5f88023fd6aaaa8ca3c4a6357ebb23f6f07da6079093ccf27c99efce87db/rpds_py-0.26.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:c1fb0cda2abcc0ac62f64e2ea4b4e64c57dfd6b885e693095460c61bde7bb18e", size = 359230, upload-time = "2025-07-01T15:56:17.978Z" }, + { url = "https://files.pythonhosted.org/packages/6c/6c/13eaebd28b439da6964dde22712b52e53fe2824af0223b8e403249d10405/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:84d142d2d6cf9b31c12aa4878d82ed3b2324226270b89b676ac62ccd7df52d08", size = 382363, upload-time = "2025-07-01T15:56:19.977Z" }, + { url = "https://files.pythonhosted.org/packages/55/fc/3bb9c486b06da19448646f96147796de23c5811ef77cbfc26f17307b6a9d/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:a547e21c5610b7e9093d870be50682a6a6cf180d6da0f42c47c306073bfdbbf6", size = 397146, upload-time = "2025-07-01T15:56:21.39Z" }, + { url = "https://files.pythonhosted.org/packages/15/18/9d1b79eb4d18e64ba8bba9e7dec6f9d6920b639f22f07ee9368ca35d4673/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:35e9a70a0f335371275cdcd08bc5b8051ac494dd58bff3bbfb421038220dc871", size = 514804, upload-time = "2025-07-01T15:56:22.78Z" }, + { url = "https://files.pythonhosted.org/packages/4f/5a/175ad7191bdbcd28785204621b225ad70e85cdfd1e09cc414cb554633b21/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0dfa6115c6def37905344d56fb54c03afc49104e2ca473d5dedec0f6606913b4", size = 402820, upload-time = "2025-07-01T15:56:24.584Z" }, + { url = "https://files.pythonhosted.org/packages/11/45/6a67ecf6d61c4d4aff4bc056e864eec4b2447787e11d1c2c9a0242c6e92a/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:313cfcd6af1a55a286a3c9a25f64af6d0e46cf60bc5798f1db152d97a216ff6f", size = 384567, upload-time = "2025-07-01T15:56:26.064Z" }, + { url = "https://files.pythonhosted.org/packages/a1/ba/16589da828732b46454c61858950a78fe4c931ea4bf95f17432ffe64b241/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:f7bf2496fa563c046d05e4d232d7b7fd61346e2402052064b773e5c378bf6f73", size = 416520, upload-time = "2025-07-01T15:56:27.608Z" }, + { url = "https://files.pythonhosted.org/packages/81/4b/00092999fc7c0c266045e984d56b7314734cc400a6c6dc4d61a35f135a9d/rpds_py-0.26.0-pp310-pypy310_pp73-musllinux_1_2_aarch64.whl", hash = "sha256:aa81873e2c8c5aa616ab8e017a481a96742fdf9313c40f14338ca7dbf50cb55f", size = 559362, upload-time = "2025-07-01T15:56:29.078Z" }, + { url = "https://files.pythonhosted.org/packages/96/0c/43737053cde1f93ac4945157f7be1428724ab943e2132a0d235a7e161d4e/rpds_py-0.26.0-pp310-pypy310_pp73-musllinux_1_2_i686.whl", hash = "sha256:68ffcf982715f5b5b7686bdd349ff75d422e8f22551000c24b30eaa1b7f7ae84", size = 588113, upload-time = "2025-07-01T15:56:30.485Z" }, + { url = "https://files.pythonhosted.org/packages/46/46/8e38f6161466e60a997ed7e9951ae5de131dedc3cf778ad35994b4af823d/rpds_py-0.26.0-pp310-pypy310_pp73-musllinux_1_2_x86_64.whl", hash = "sha256:6188de70e190847bb6db3dc3981cbadff87d27d6fe9b4f0e18726d55795cee9b", size = 555429, upload-time = "2025-07-01T15:56:31.956Z" }, + { url = "https://files.pythonhosted.org/packages/2c/ac/65da605e9f1dd643ebe615d5bbd11b6efa1d69644fc4bf623ea5ae385a82/rpds_py-0.26.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:1c962145c7473723df9722ba4c058de12eb5ebedcb4e27e7d902920aa3831ee8", size = 231950, upload-time = "2025-07-01T15:56:33.337Z" }, + { url = "https://files.pythonhosted.org/packages/51/f2/b5c85b758a00c513bb0389f8fc8e61eb5423050c91c958cdd21843faa3e6/rpds_py-0.26.0-pp311-pypy311_pp73-macosx_10_12_x86_64.whl", hash = "sha256:f61a9326f80ca59214d1cceb0a09bb2ece5b2563d4e0cd37bfd5515c28510674", size = 373505, upload-time = "2025-07-01T15:56:34.716Z" }, + { url = "https://files.pythonhosted.org/packages/23/e0/25db45e391251118e915e541995bb5f5ac5691a3b98fb233020ba53afc9b/rpds_py-0.26.0-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:183f857a53bcf4b1b42ef0f57ca553ab56bdd170e49d8091e96c51c3d69ca696", size = 359468, upload-time = "2025-07-01T15:56:36.219Z" }, + { url = "https://files.pythonhosted.org/packages/0b/73/dd5ee6075bb6491be3a646b301dfd814f9486d924137a5098e61f0487e16/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:941c1cfdf4799d623cf3aa1d326a6b4fdb7a5799ee2687f3516738216d2262fb", size = 382680, upload-time = "2025-07-01T15:56:37.644Z" }, + { url = "https://files.pythonhosted.org/packages/2f/10/84b522ff58763a5c443f5bcedc1820240e454ce4e620e88520f04589e2ea/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:72a8d9564a717ee291f554eeb4bfeafe2309d5ec0aa6c475170bdab0f9ee8e88", size = 397035, upload-time = "2025-07-01T15:56:39.241Z" }, + { url = "https://files.pythonhosted.org/packages/06/ea/8667604229a10a520fcbf78b30ccc278977dcc0627beb7ea2c96b3becef0/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:511d15193cbe013619dd05414c35a7dedf2088fcee93c6bbb7c77859765bd4e8", size = 514922, upload-time = "2025-07-01T15:56:40.645Z" }, + { url = "https://files.pythonhosted.org/packages/24/e6/9ed5b625c0661c4882fc8cdf302bf8e96c73c40de99c31e0b95ed37d508c/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:aea1f9741b603a8d8fedb0ed5502c2bc0accbc51f43e2ad1337fe7259c2b77a5", size = 402822, upload-time = "2025-07-01T15:56:42.137Z" }, + { url = "https://files.pythonhosted.org/packages/8a/58/212c7b6fd51946047fb45d3733da27e2fa8f7384a13457c874186af691b1/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4019a9d473c708cf2f16415688ef0b4639e07abaa569d72f74745bbeffafa2c7", size = 384336, upload-time = "2025-07-01T15:56:44.239Z" }, + { url = "https://files.pythonhosted.org/packages/aa/f5/a40ba78748ae8ebf4934d4b88e77b98497378bc2c24ba55ebe87a4e87057/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:093d63b4b0f52d98ebae33b8c50900d3d67e0666094b1be7a12fffd7f65de74b", size = 416871, upload-time = "2025-07-01T15:56:46.284Z" }, + { url = "https://files.pythonhosted.org/packages/d5/a6/33b1fc0c9f7dcfcfc4a4353daa6308b3ece22496ceece348b3e7a7559a09/rpds_py-0.26.0-pp311-pypy311_pp73-musllinux_1_2_aarch64.whl", hash = "sha256:2abe21d8ba64cded53a2a677e149ceb76dcf44284202d737178afe7ba540c1eb", size = 559439, upload-time = "2025-07-01T15:56:48.549Z" }, + { url = "https://files.pythonhosted.org/packages/71/2d/ceb3f9c12f8cfa56d34995097f6cd99da1325642c60d1b6680dd9df03ed8/rpds_py-0.26.0-pp311-pypy311_pp73-musllinux_1_2_i686.whl", hash = "sha256:4feb7511c29f8442cbbc28149a92093d32e815a28aa2c50d333826ad2a20fdf0", size = 588380, upload-time = "2025-07-01T15:56:50.086Z" }, + { url = "https://files.pythonhosted.org/packages/c8/ed/9de62c2150ca8e2e5858acf3f4f4d0d180a38feef9fdab4078bea63d8dba/rpds_py-0.26.0-pp311-pypy311_pp73-musllinux_1_2_x86_64.whl", hash = "sha256:e99685fc95d386da368013e7fb4269dd39c30d99f812a8372d62f244f662709c", size = 555334, upload-time = "2025-07-01T15:56:51.703Z" }, +] + +[[package]] +name = "rsa" +version = "4.9.1" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "pyasn1" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/da/8a/22b7beea3ee0d44b1916c0c1cb0ee3af23b700b6da9f04991899d0c555d4/rsa-4.9.1.tar.gz", hash = "sha256:e7bdbfdb5497da4c07dfd35530e1a902659db6ff241e39d9953cad06ebd0ae75", size = 29034, upload-time = "2025-04-16T09:51:18.218Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/64/8d/0133e4eb4beed9e425d9a98ed6e081a55d195481b7632472be1af08d2f6b/rsa-4.9.1-py3-none-any.whl", hash = "sha256:68635866661c6836b8d39430f97a996acbd61bfa49406748ea243539fe239762", size = 34696, upload-time = "2025-04-16T09:51:17.142Z" }, ] [[package]] name = "ruff" -version = "0.11.2" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/90/61/fb87430f040e4e577e784e325351186976516faef17d6fcd921fe28edfd7/ruff-0.11.2.tar.gz", hash = "sha256:ec47591497d5a1050175bdf4e1a4e6272cddff7da88a2ad595e1e326041d8d94", size = 3857511, upload-time = "2025-03-21T13:31:17.419Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/62/99/102578506f0f5fa29fd7e0df0a273864f79af044757aef73d1cae0afe6ad/ruff-0.11.2-py3-none-linux_armv6l.whl", hash = "sha256:c69e20ea49e973f3afec2c06376eb56045709f0212615c1adb0eda35e8a4e477", size = 10113146, upload-time = "2025-03-21T13:30:26.68Z" }, - { url = "https://files.pythonhosted.org/packages/74/ad/5cd4ba58ab602a579997a8494b96f10f316e874d7c435bcc1a92e6da1b12/ruff-0.11.2-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:2c5424cc1c4eb1d8ecabe6d4f1b70470b4f24a0c0171356290b1953ad8f0e272", size = 10867092, upload-time = "2025-03-21T13:30:37.949Z" }, - { url = "https://files.pythonhosted.org/packages/fc/3e/d3f13619e1d152c7b600a38c1a035e833e794c6625c9a6cea6f63dbf3af4/ruff-0.11.2-py3-none-macosx_11_0_arm64.whl", hash = "sha256:ecf20854cc73f42171eedb66f006a43d0a21bfb98a2523a809931cda569552d9", size = 10224082, upload-time = "2025-03-21T13:30:39.962Z" }, - { url = "https://files.pythonhosted.org/packages/90/06/f77b3d790d24a93f38e3806216f263974909888fd1e826717c3ec956bbcd/ruff-0.11.2-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0c543bf65d5d27240321604cee0633a70c6c25c9a2f2492efa9f6d4b8e4199bb", size = 10394818, upload-time = "2025-03-21T13:30:42.551Z" }, - { url = "https://files.pythonhosted.org/packages/99/7f/78aa431d3ddebfc2418cd95b786642557ba8b3cb578c075239da9ce97ff9/ruff-0.11.2-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:20967168cc21195db5830b9224be0e964cc9c8ecf3b5a9e3ce19876e8d3a96e3", size = 9952251, upload-time = "2025-03-21T13:30:45.196Z" }, - { url = "https://files.pythonhosted.org/packages/30/3e/f11186d1ddfaca438c3bbff73c6a2fdb5b60e6450cc466129c694b0ab7a2/ruff-0.11.2-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:955a9ce63483999d9f0b8f0b4a3ad669e53484232853054cc8b9d51ab4c5de74", size = 11563566, upload-time = "2025-03-21T13:30:47.516Z" }, - { url = "https://files.pythonhosted.org/packages/22/6c/6ca91befbc0a6539ee133d9a9ce60b1a354db12c3c5d11cfdbf77140f851/ruff-0.11.2-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:86b3a27c38b8fce73bcd262b0de32e9a6801b76d52cdb3ae4c914515f0cef608", size = 12208721, upload-time = "2025-03-21T13:30:49.56Z" }, - { url = "https://files.pythonhosted.org/packages/19/b0/24516a3b850d55b17c03fc399b681c6a549d06ce665915721dc5d6458a5c/ruff-0.11.2-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a3b66a03b248c9fcd9d64d445bafdf1589326bee6fc5c8e92d7562e58883e30f", size = 11662274, upload-time = "2025-03-21T13:30:52.055Z" }, - { url = "https://files.pythonhosted.org/packages/d7/65/76be06d28ecb7c6070280cef2bcb20c98fbf99ff60b1c57d2fb9b8771348/ruff-0.11.2-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0397c2672db015be5aa3d4dac54c69aa012429097ff219392c018e21f5085147", size = 13792284, upload-time = "2025-03-21T13:30:54.24Z" }, - { url = "https://files.pythonhosted.org/packages/ce/d2/4ceed7147e05852876f3b5f3fdc23f878ce2b7e0b90dd6e698bda3d20787/ruff-0.11.2-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:869bcf3f9abf6457fbe39b5a37333aa4eecc52a3b99c98827ccc371a8e5b6f1b", size = 11327861, upload-time = "2025-03-21T13:30:56.757Z" }, - { url = "https://files.pythonhosted.org/packages/c4/78/4935ecba13706fd60ebe0e3dc50371f2bdc3d9bc80e68adc32ff93914534/ruff-0.11.2-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:2a2b50ca35457ba785cd8c93ebbe529467594087b527a08d487cf0ee7b3087e9", size = 10276560, upload-time = "2025-03-21T13:30:58.881Z" }, - { url = "https://files.pythonhosted.org/packages/81/7f/1b2435c3f5245d410bb5dc80f13ec796454c21fbda12b77d7588d5cf4e29/ruff-0.11.2-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:7c69c74bf53ddcfbc22e6eb2f31211df7f65054bfc1f72288fc71e5f82db3eab", size = 9945091, upload-time = "2025-03-21T13:31:01.45Z" }, - { url = "https://files.pythonhosted.org/packages/39/c4/692284c07e6bf2b31d82bb8c32f8840f9d0627d92983edaac991a2b66c0a/ruff-0.11.2-py3-none-musllinux_1_2_i686.whl", hash = "sha256:6e8fb75e14560f7cf53b15bbc55baf5ecbe373dd5f3aab96ff7aa7777edd7630", size = 10977133, upload-time = "2025-03-21T13:31:04.013Z" }, - { url = "https://files.pythonhosted.org/packages/94/cf/8ab81cb7dd7a3b0a3960c2769825038f3adcd75faf46dd6376086df8b128/ruff-0.11.2-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:842a472d7b4d6f5924e9297aa38149e5dcb1e628773b70e6387ae2c97a63c58f", size = 11378514, upload-time = "2025-03-21T13:31:06.166Z" }, - { url = "https://files.pythonhosted.org/packages/d9/3a/a647fa4f316482dacf2fd68e8a386327a33d6eabd8eb2f9a0c3d291ec549/ruff-0.11.2-py3-none-win32.whl", hash = "sha256:aca01ccd0eb5eb7156b324cfaa088586f06a86d9e5314b0eb330cb48415097cc", size = 10319835, upload-time = "2025-03-21T13:31:10.7Z" }, - { url = "https://files.pythonhosted.org/packages/86/54/3c12d3af58012a5e2cd7ebdbe9983f4834af3f8cbea0e8a8c74fa1e23b2b/ruff-0.11.2-py3-none-win_amd64.whl", hash = "sha256:3170150172a8f994136c0c66f494edf199a0bbea7a409f649e4bc8f4d7084080", size = 11373713, upload-time = "2025-03-21T13:31:13.148Z" }, - { url = "https://files.pythonhosted.org/packages/d6/d4/dd813703af8a1e2ac33bf3feb27e8a5ad514c9f219df80c64d69807e7f71/ruff-0.11.2-py3-none-win_arm64.whl", hash = "sha256:52933095158ff328f4c77af3d74f0379e34fd52f175144cefc1b192e7ccd32b4", size = 10441990, upload-time = "2025-03-21T13:31:15.206Z" }, +version = "0.12.7" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/a1/81/0bd3594fa0f690466e41bd033bdcdf86cba8288345ac77ad4afbe5ec743a/ruff-0.12.7.tar.gz", hash = "sha256:1fc3193f238bc2d7968772c82831a4ff69252f673be371fb49663f0068b7ec71", size = 5197814, upload-time = "2025-07-29T22:32:35.877Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/e1/d2/6cb35e9c85e7a91e8d22ab32ae07ac39cc34a71f1009a6f9e4a2a019e602/ruff-0.12.7-py3-none-linux_armv6l.whl", hash = "sha256:76e4f31529899b8c434c3c1dede98c4483b89590e15fb49f2d46183801565303", size = 11852189, upload-time = "2025-07-29T22:31:41.281Z" }, + { url = "https://files.pythonhosted.org/packages/63/5b/a4136b9921aa84638f1a6be7fb086f8cad0fde538ba76bda3682f2599a2f/ruff-0.12.7-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:789b7a03e72507c54fb3ba6209e4bb36517b90f1a3569ea17084e3fd295500fb", size = 12519389, upload-time = "2025-07-29T22:31:54.265Z" }, + { url = "https://files.pythonhosted.org/packages/a8/c9/3e24a8472484269b6b1821794141f879c54645a111ded4b6f58f9ab0705f/ruff-0.12.7-py3-none-macosx_11_0_arm64.whl", hash = "sha256:2e1c2a3b8626339bb6369116e7030a4cf194ea48f49b64bb505732a7fce4f4e3", size = 11743384, upload-time = "2025-07-29T22:31:59.575Z" }, + { url = "https://files.pythonhosted.org/packages/26/7c/458dd25deeb3452c43eaee853c0b17a1e84169f8021a26d500ead77964fd/ruff-0.12.7-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32dec41817623d388e645612ec70d5757a6d9c035f3744a52c7b195a57e03860", size = 11943759, upload-time = "2025-07-29T22:32:01.95Z" }, + { url = "https://files.pythonhosted.org/packages/7f/8b/658798472ef260ca050e400ab96ef7e85c366c39cf3dfbef4d0a46a528b6/ruff-0.12.7-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:47ef751f722053a5df5fa48d412dbb54d41ab9b17875c6840a58ec63ff0c247c", size = 11654028, upload-time = "2025-07-29T22:32:04.367Z" }, + { url = "https://files.pythonhosted.org/packages/a8/86/9c2336f13b2a3326d06d39178fd3448dcc7025f82514d1b15816fe42bfe8/ruff-0.12.7-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a828a5fc25a3efd3e1ff7b241fd392686c9386f20e5ac90aa9234a5faa12c423", size = 13225209, upload-time = "2025-07-29T22:32:06.952Z" }, + { url = "https://files.pythonhosted.org/packages/76/69/df73f65f53d6c463b19b6b312fd2391dc36425d926ec237a7ed028a90fc1/ruff-0.12.7-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:5726f59b171111fa6a69d82aef48f00b56598b03a22f0f4170664ff4d8298efb", size = 14182353, upload-time = "2025-07-29T22:32:10.053Z" }, + { url = "https://files.pythonhosted.org/packages/58/1e/de6cda406d99fea84b66811c189b5ea139814b98125b052424b55d28a41c/ruff-0.12.7-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:74e6f5c04c4dd4aba223f4fe6e7104f79e0eebf7d307e4f9b18c18362124bccd", size = 13631555, upload-time = "2025-07-29T22:32:12.644Z" }, + { url = "https://files.pythonhosted.org/packages/6f/ae/625d46d5164a6cc9261945a5e89df24457dc8262539ace3ac36c40f0b51e/ruff-0.12.7-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:5d0bfe4e77fba61bf2ccadf8cf005d6133e3ce08793bbe870dd1c734f2699a3e", size = 12667556, upload-time = "2025-07-29T22:32:15.312Z" }, + { url = "https://files.pythonhosted.org/packages/55/bf/9cb1ea5e3066779e42ade8d0cd3d3b0582a5720a814ae1586f85014656b6/ruff-0.12.7-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:06bfb01e1623bf7f59ea749a841da56f8f653d641bfd046edee32ede7ff6c606", size = 12939784, upload-time = "2025-07-29T22:32:17.69Z" }, + { url = "https://files.pythonhosted.org/packages/55/7f/7ead2663be5627c04be83754c4f3096603bf5e99ed856c7cd29618c691bd/ruff-0.12.7-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:e41df94a957d50083fd09b916d6e89e497246698c3f3d5c681c8b3e7b9bb4ac8", size = 11771356, upload-time = "2025-07-29T22:32:20.134Z" }, + { url = "https://files.pythonhosted.org/packages/17/40/a95352ea16edf78cd3a938085dccc55df692a4d8ba1b3af7accbe2c806b0/ruff-0.12.7-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:4000623300563c709458d0ce170c3d0d788c23a058912f28bbadc6f905d67afa", size = 11612124, upload-time = "2025-07-29T22:32:22.645Z" }, + { url = "https://files.pythonhosted.org/packages/4d/74/633b04871c669e23b8917877e812376827c06df866e1677f15abfadc95cb/ruff-0.12.7-py3-none-musllinux_1_2_i686.whl", hash = "sha256:69ffe0e5f9b2cf2b8e289a3f8945b402a1b19eff24ec389f45f23c42a3dd6fb5", size = 12479945, upload-time = "2025-07-29T22:32:24.765Z" }, + { url = "https://files.pythonhosted.org/packages/be/34/c3ef2d7799c9778b835a76189c6f53c179d3bdebc8c65288c29032e03613/ruff-0.12.7-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:a07a5c8ffa2611a52732bdc67bf88e243abd84fe2d7f6daef3826b59abbfeda4", size = 12998677, upload-time = "2025-07-29T22:32:27.022Z" }, + { url = "https://files.pythonhosted.org/packages/77/ab/aca2e756ad7b09b3d662a41773f3edcbd262872a4fc81f920dc1ffa44541/ruff-0.12.7-py3-none-win32.whl", hash = "sha256:c928f1b2ec59fb77dfdf70e0419408898b63998789cc98197e15f560b9e77f77", size = 11756687, upload-time = "2025-07-29T22:32:29.381Z" }, + { url = "https://files.pythonhosted.org/packages/b4/71/26d45a5042bc71db22ddd8252ca9d01e9ca454f230e2996bb04f16d72799/ruff-0.12.7-py3-none-win_amd64.whl", hash = "sha256:9c18f3d707ee9edf89da76131956aba1270c6348bfee8f6c647de841eac7194f", size = 12912365, upload-time = "2025-07-29T22:32:31.517Z" }, + { url = "https://files.pythonhosted.org/packages/4c/9b/0b8aa09817b63e78d94b4977f18b1fcaead3165a5ee49251c5d5c245bb2d/ruff-0.12.7-py3-none-win_arm64.whl", hash = "sha256:dfce05101dbd11833a0776716d5d1578641b7fddb537fe7fa956ab85d1769b69", size = 11982083, upload-time = "2025-07-29T22:32:33.881Z" }, ] [[package]] name = "s3fs" -version = "2025.3.0" +version = "2025.7.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "aiobotocore" }, { name = "aiohttp" }, { name = "fsspec" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e4/cd/5dde2fed1699ff48120336249d9857a574e39feb8afaff694568ab1499b3/s3fs-2025.3.0.tar.gz", hash = "sha256:446dd539eb0d0678209723cb7ad1bedbb172185b0d34675b09be1ad81843a644", size = 77153, upload-time = "2025-03-07T21:58:32.114Z" } +sdist = { url = "https://files.pythonhosted.org/packages/bf/13/37438c4672ba1d23ec46df0e4b57e98469e5c5f4f98313cf6842b631652b/s3fs-2025.7.0.tar.gz", hash = "sha256:5e7f9ec0cad7745155e3eb86fae15b1481fa29946bf5b3a4ce3a60701ce6022d", size = 77795, upload-time = "2025-07-15T16:35:22.177Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/3a/3f/35f4041a82a68df89fe4af97c8bb44aa492dad924799cbb02078e9e303e6/s3fs-2025.3.0-py3-none-any.whl", hash = "sha256:88d803615baa04945156ca0e1498009b7acd3132c07198bd81b3e874846e0aa2", size = 30454, upload-time = "2025-03-07T21:58:30.998Z" }, + { url = "https://files.pythonhosted.org/packages/ff/c7/30d13b7fd4f866ca3f30e9a6e7ae038f0c45226f6e26b3cc98d6d197f93b/s3fs-2025.7.0-py3-none-any.whl", hash = "sha256:b6b2d3f84b6aa1c2ba5e62e39dd9410cf54f10a2cce1ea6db1ba0d1a6bcce685", size = 30315, upload-time = "2025-07-15T16:35:20.734Z" }, ] [package.optional-dependencies] @@ -2770,152 +3500,169 @@ boto3 = [ [[package]] name = "s3transfer" -version = "0.11.3" +version = "0.13.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "botocore" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/39/24/1390172471d569e281fcfd29b92f2f73774e95972c965d14b6c802ff2352/s3transfer-0.11.3.tar.gz", hash = "sha256:edae4977e3a122445660c7c114bba949f9d191bae3b34a096f18a1c8c354527a", size = 148042, upload-time = "2025-02-26T20:44:57.459Z" } +sdist = { url = "https://files.pythonhosted.org/packages/6d/05/d52bf1e65044b4e5e27d4e63e8d1579dbdec54fce685908ae09bc3720030/s3transfer-0.13.1.tar.gz", hash = "sha256:c3fdba22ba1bd367922f27ec8032d6a1cf5f10c934fb5d68cf60fd5a23d936cf", size = 150589, upload-time = "2025-07-18T19:22:42.31Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e4/81/48c41b554a54d75d4407740abb60e3a102ae416284df04d1dbdcbe3dbf24/s3transfer-0.11.3-py3-none-any.whl", hash = "sha256:ca855bdeb885174b5ffa95b9913622459d4ad8e331fc98eb01e6d5eb6a30655d", size = 84246, upload-time = "2025-02-26T20:44:55.509Z" }, + { url = "https://files.pythonhosted.org/packages/6d/4f/d073e09df851cfa251ef7840007d04db3293a0482ce607d2b993926089be/s3transfer-0.13.1-py3-none-any.whl", hash = "sha256:a981aa7429be23fe6dfc13e80e4020057cbab622b08c0315288758d67cabc724", size = 85308, upload-time = "2025-07-18T19:22:40.947Z" }, ] [[package]] name = "scipy" -version = "1.15.2" -source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "numpy" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/b7/b9/31ba9cd990e626574baf93fbc1ac61cf9ed54faafd04c479117517661637/scipy-1.15.2.tar.gz", hash = "sha256:cd58a314d92838f7e6f755c8a2167ead4f27e1fd5c1251fd54289569ef3495ec", size = 59417316, upload-time = "2025-02-17T00:42:24.791Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/95/df/ef233fff6838fe6f7840d69b5ef9f20d2b5c912a8727b21ebf876cb15d54/scipy-1.15.2-cp310-cp310-macosx_10_13_x86_64.whl", hash = "sha256:a2ec871edaa863e8213ea5df811cd600734f6400b4af272e1c011e69401218e9", size = 38692502, upload-time = "2025-02-17T00:28:56.118Z" }, - { url = "https://files.pythonhosted.org/packages/5c/20/acdd4efb8a68b842968f7bc5611b1aeb819794508771ad104de418701422/scipy-1.15.2-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:6f223753c6ea76983af380787611ae1291e3ceb23917393079dcc746ba60cfb5", size = 30085508, upload-time = "2025-02-17T00:29:06.048Z" }, - { url = "https://files.pythonhosted.org/packages/42/55/39cf96ca7126f1e78ee72a6344ebdc6702fc47d037319ad93221063e6cf4/scipy-1.15.2-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:ecf797d2d798cf7c838c6d98321061eb3e72a74710e6c40540f0e8087e3b499e", size = 22359166, upload-time = "2025-02-17T00:29:13.553Z" }, - { url = "https://files.pythonhosted.org/packages/51/48/708d26a4ab8a1441536bf2dfcad1df0ca14a69f010fba3ccbdfc02df7185/scipy-1.15.2-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:9b18aa747da280664642997e65aab1dd19d0c3d17068a04b3fe34e2559196cb9", size = 25112047, upload-time = "2025-02-17T00:29:23.204Z" }, - { url = "https://files.pythonhosted.org/packages/dd/65/f9c5755b995ad892020381b8ae11f16d18616208e388621dfacc11df6de6/scipy-1.15.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:87994da02e73549dfecaed9e09a4f9d58a045a053865679aeb8d6d43747d4df3", size = 35536214, upload-time = "2025-02-17T00:29:33.215Z" }, - { url = "https://files.pythonhosted.org/packages/de/3c/c96d904b9892beec978562f64d8cc43f9cca0842e65bd3cd1b7f7389b0ba/scipy-1.15.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:69ea6e56d00977f355c0f84eba69877b6df084516c602d93a33812aa04d90a3d", size = 37646981, upload-time = "2025-02-17T00:29:46.188Z" }, - { url = "https://files.pythonhosted.org/packages/3d/74/c2d8a24d18acdeae69ed02e132b9bc1bb67b7bee90feee1afe05a68f9d67/scipy-1.15.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:888307125ea0c4466287191e5606a2c910963405ce9671448ff9c81c53f85f58", size = 37230048, upload-time = "2025-02-17T00:29:56.646Z" }, - { url = "https://files.pythonhosted.org/packages/42/19/0aa4ce80eca82d487987eff0bc754f014dec10d20de2f66754fa4ea70204/scipy-1.15.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:9412f5e408b397ff5641080ed1e798623dbe1ec0d78e72c9eca8992976fa65aa", size = 40010322, upload-time = "2025-02-17T00:30:07.422Z" }, - { url = "https://files.pythonhosted.org/packages/d0/d2/f0683b7e992be44d1475cc144d1f1eeae63c73a14f862974b4db64af635e/scipy-1.15.2-cp310-cp310-win_amd64.whl", hash = "sha256:b5e025e903b4f166ea03b109bb241355b9c42c279ea694d8864d033727205e65", size = 41233385, upload-time = "2025-02-17T00:30:20.268Z" }, - { url = "https://files.pythonhosted.org/packages/40/1f/bf0a5f338bda7c35c08b4ed0df797e7bafe8a78a97275e9f439aceb46193/scipy-1.15.2-cp311-cp311-macosx_10_13_x86_64.whl", hash = "sha256:92233b2df6938147be6fa8824b8136f29a18f016ecde986666be5f4d686a91a4", size = 38703651, upload-time = "2025-02-17T00:30:31.09Z" }, - { url = "https://files.pythonhosted.org/packages/de/54/db126aad3874601048c2c20ae3d8a433dbfd7ba8381551e6f62606d9bd8e/scipy-1.15.2-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:62ca1ff3eb513e09ed17a5736929429189adf16d2d740f44e53270cc800ecff1", size = 30102038, upload-time = "2025-02-17T00:30:40.219Z" }, - { url = "https://files.pythonhosted.org/packages/61/d8/84da3fffefb6c7d5a16968fe5b9f24c98606b165bb801bb0b8bc3985200f/scipy-1.15.2-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:4c6676490ad76d1c2894d77f976144b41bd1a4052107902238047fb6a473e971", size = 22375518, upload-time = "2025-02-17T00:30:47.547Z" }, - { url = "https://files.pythonhosted.org/packages/44/78/25535a6e63d3b9c4c90147371aedb5d04c72f3aee3a34451f2dc27c0c07f/scipy-1.15.2-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:a8bf5cb4a25046ac61d38f8d3c3426ec11ebc350246a4642f2f315fe95bda655", size = 25142523, upload-time = "2025-02-17T00:30:56.002Z" }, - { url = "https://files.pythonhosted.org/packages/e0/22/4b4a26fe1cd9ed0bc2b2cb87b17d57e32ab72c346949eaf9288001f8aa8e/scipy-1.15.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6a8e34cf4c188b6dd004654f88586d78f95639e48a25dfae9c5e34a6dc34547e", size = 35491547, upload-time = "2025-02-17T00:31:07.599Z" }, - { url = "https://files.pythonhosted.org/packages/32/ea/564bacc26b676c06a00266a3f25fdfe91a9d9a2532ccea7ce6dd394541bc/scipy-1.15.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:28a0d2c2075946346e4408b211240764759e0fabaeb08d871639b5f3b1aca8a0", size = 37634077, upload-time = "2025-02-17T00:31:15.191Z" }, - { url = "https://files.pythonhosted.org/packages/43/c2/bfd4e60668897a303b0ffb7191e965a5da4056f0d98acfb6ba529678f0fb/scipy-1.15.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:42dabaaa798e987c425ed76062794e93a243be8f0f20fff6e7a89f4d61cb3d40", size = 37231657, upload-time = "2025-02-17T00:31:22.041Z" }, - { url = "https://files.pythonhosted.org/packages/4a/75/5f13050bf4f84c931bcab4f4e83c212a36876c3c2244475db34e4b5fe1a6/scipy-1.15.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:6f5e296ec63c5da6ba6fa0343ea73fd51b8b3e1a300b0a8cae3ed4b1122c7462", size = 40035857, upload-time = "2025-02-17T00:31:29.836Z" }, - { url = "https://files.pythonhosted.org/packages/b9/8b/7ec1832b09dbc88f3db411f8cdd47db04505c4b72c99b11c920a8f0479c3/scipy-1.15.2-cp311-cp311-win_amd64.whl", hash = "sha256:597a0c7008b21c035831c39927406c6181bcf8f60a73f36219b69d010aa04737", size = 41217654, upload-time = "2025-02-17T00:31:43.65Z" }, - { url = "https://files.pythonhosted.org/packages/4b/5d/3c78815cbab499610f26b5bae6aed33e227225a9fa5290008a733a64f6fc/scipy-1.15.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:c4697a10da8f8765bb7c83e24a470da5797e37041edfd77fd95ba3811a47c4fd", size = 38756184, upload-time = "2025-02-17T00:31:50.623Z" }, - { url = "https://files.pythonhosted.org/packages/37/20/3d04eb066b471b6e171827548b9ddb3c21c6bbea72a4d84fc5989933910b/scipy-1.15.2-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:869269b767d5ee7ea6991ed7e22b3ca1f22de73ab9a49c44bad338b725603301", size = 30163558, upload-time = "2025-02-17T00:31:56.721Z" }, - { url = "https://files.pythonhosted.org/packages/a4/98/e5c964526c929ef1f795d4c343b2ff98634ad2051bd2bbadfef9e772e413/scipy-1.15.2-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:bad78d580270a4d32470563ea86c6590b465cb98f83d760ff5b0990cb5518a93", size = 22437211, upload-time = "2025-02-17T00:32:03.042Z" }, - { url = "https://files.pythonhosted.org/packages/1d/cd/1dc7371e29195ecbf5222f9afeedb210e0a75057d8afbd942aa6cf8c8eca/scipy-1.15.2-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:b09ae80010f52efddb15551025f9016c910296cf70adbf03ce2a8704f3a5ad20", size = 25232260, upload-time = "2025-02-17T00:32:07.847Z" }, - { url = "https://files.pythonhosted.org/packages/f0/24/1a181a9e5050090e0b5138c5f496fee33293c342b788d02586bc410c6477/scipy-1.15.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5a6fd6eac1ce74a9f77a7fc724080d507c5812d61e72bd5e4c489b042455865e", size = 35198095, upload-time = "2025-02-17T00:32:14.565Z" }, - { url = "https://files.pythonhosted.org/packages/c0/53/eaada1a414c026673eb983f8b4a55fe5eb172725d33d62c1b21f63ff6ca4/scipy-1.15.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2b871df1fe1a3ba85d90e22742b93584f8d2b8e6124f8372ab15c71b73e428b8", size = 37297371, upload-time = "2025-02-17T00:32:21.411Z" }, - { url = "https://files.pythonhosted.org/packages/e9/06/0449b744892ed22b7e7b9a1994a866e64895363572677a316a9042af1fe5/scipy-1.15.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:03205d57a28e18dfd39f0377d5002725bf1f19a46f444108c29bdb246b6c8a11", size = 36872390, upload-time = "2025-02-17T00:32:29.421Z" }, - { url = "https://files.pythonhosted.org/packages/6a/6f/a8ac3cfd9505ec695c1bc35edc034d13afbd2fc1882a7c6b473e280397bb/scipy-1.15.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:601881dfb761311045b03114c5fe718a12634e5608c3b403737ae463c9885d53", size = 39700276, upload-time = "2025-02-17T00:32:37.431Z" }, - { url = "https://files.pythonhosted.org/packages/f5/6f/e6e5aff77ea2a48dd96808bb51d7450875af154ee7cbe72188afb0b37929/scipy-1.15.2-cp312-cp312-win_amd64.whl", hash = "sha256:e7c68b6a43259ba0aab737237876e5c2c549a031ddb7abc28c7b47f22e202ded", size = 40942317, upload-time = "2025-02-17T00:32:45.47Z" }, - { url = "https://files.pythonhosted.org/packages/53/40/09319f6e0f276ea2754196185f95cd191cb852288440ce035d5c3a931ea2/scipy-1.15.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:01edfac9f0798ad6b46d9c4c9ca0e0ad23dbf0b1eb70e96adb9fa7f525eff0bf", size = 38717587, upload-time = "2025-02-17T00:32:53.196Z" }, - { url = "https://files.pythonhosted.org/packages/fe/c3/2854f40ecd19585d65afaef601e5e1f8dbf6758b2f95b5ea93d38655a2c6/scipy-1.15.2-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:08b57a9336b8e79b305a143c3655cc5bdbe6d5ece3378578888d2afbb51c4e37", size = 30100266, upload-time = "2025-02-17T00:32:59.318Z" }, - { url = "https://files.pythonhosted.org/packages/dd/b1/f9fe6e3c828cb5930b5fe74cb479de5f3d66d682fa8adb77249acaf545b8/scipy-1.15.2-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:54c462098484e7466362a9f1672d20888f724911a74c22ae35b61f9c5919183d", size = 22373768, upload-time = "2025-02-17T00:33:04.091Z" }, - { url = "https://files.pythonhosted.org/packages/15/9d/a60db8c795700414c3f681908a2b911e031e024d93214f2d23c6dae174ab/scipy-1.15.2-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:cf72ff559a53a6a6d77bd8eefd12a17995ffa44ad86c77a5df96f533d4e6c6bb", size = 25154719, upload-time = "2025-02-17T00:33:08.909Z" }, - { url = "https://files.pythonhosted.org/packages/37/3b/9bda92a85cd93f19f9ed90ade84aa1e51657e29988317fabdd44544f1dd4/scipy-1.15.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9de9d1416b3d9e7df9923ab23cd2fe714244af10b763975bea9e4f2e81cebd27", size = 35163195, upload-time = "2025-02-17T00:33:15.352Z" }, - { url = "https://files.pythonhosted.org/packages/03/5a/fc34bf1aa14dc7c0e701691fa8685f3faec80e57d816615e3625f28feb43/scipy-1.15.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fb530e4794fc8ea76a4a21ccb67dea33e5e0e60f07fc38a49e821e1eae3b71a0", size = 37255404, upload-time = "2025-02-17T00:33:22.21Z" }, - { url = "https://files.pythonhosted.org/packages/4a/71/472eac45440cee134c8a180dbe4c01b3ec247e0338b7c759e6cd71f199a7/scipy-1.15.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:5ea7ed46d437fc52350b028b1d44e002646e28f3e8ddc714011aaf87330f2f32", size = 36860011, upload-time = "2025-02-17T00:33:29.446Z" }, - { url = "https://files.pythonhosted.org/packages/01/b3/21f890f4f42daf20e4d3aaa18182dddb9192771cd47445aaae2e318f6738/scipy-1.15.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:11e7ad32cf184b74380f43d3c0a706f49358b904fa7d5345f16ddf993609184d", size = 39657406, upload-time = "2025-02-17T00:33:39.019Z" }, - { url = "https://files.pythonhosted.org/packages/0d/76/77cf2ac1f2a9cc00c073d49e1e16244e389dd88e2490c91d84e1e3e4d126/scipy-1.15.2-cp313-cp313-win_amd64.whl", hash = "sha256:a5080a79dfb9b78b768cebf3c9dcbc7b665c5875793569f48bf0e2b1d7f68f6f", size = 40961243, upload-time = "2025-02-17T00:34:51.024Z" }, - { url = "https://files.pythonhosted.org/packages/4c/4b/a57f8ddcf48e129e6054fa9899a2a86d1fc6b07a0e15c7eebff7ca94533f/scipy-1.15.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:447ce30cee6a9d5d1379087c9e474628dab3db4a67484be1b7dc3196bfb2fac9", size = 38870286, upload-time = "2025-02-17T00:33:47.62Z" }, - { url = "https://files.pythonhosted.org/packages/0c/43/c304d69a56c91ad5f188c0714f6a97b9c1fed93128c691148621274a3a68/scipy-1.15.2-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:c90ebe8aaa4397eaefa8455a8182b164a6cc1d59ad53f79943f266d99f68687f", size = 30141634, upload-time = "2025-02-17T00:33:54.131Z" }, - { url = "https://files.pythonhosted.org/packages/44/1a/6c21b45d2548eb73be9b9bff421aaaa7e85e22c1f9b3bc44b23485dfce0a/scipy-1.15.2-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:def751dd08243934c884a3221156d63e15234a3155cf25978b0a668409d45eb6", size = 22415179, upload-time = "2025-02-17T00:33:59.948Z" }, - { url = "https://files.pythonhosted.org/packages/74/4b/aefac4bba80ef815b64f55da06f62f92be5d03b467f2ce3668071799429a/scipy-1.15.2-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:302093e7dfb120e55515936cb55618ee0b895f8bcaf18ff81eca086c17bd80af", size = 25126412, upload-time = "2025-02-17T00:34:06.328Z" }, - { url = "https://files.pythonhosted.org/packages/b1/53/1cbb148e6e8f1660aacd9f0a9dfa2b05e9ff1cb54b4386fe868477972ac2/scipy-1.15.2-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7cd5b77413e1855351cdde594eca99c1f4a588c2d63711388b6a1f1c01f62274", size = 34952867, upload-time = "2025-02-17T00:34:12.928Z" }, - { url = "https://files.pythonhosted.org/packages/2c/23/e0eb7f31a9c13cf2dca083828b97992dd22f8184c6ce4fec5deec0c81fcf/scipy-1.15.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6d0194c37037707b2afa7a2f2a924cf7bac3dc292d51b6a925e5fcb89bc5c776", size = 36890009, upload-time = "2025-02-17T00:34:19.55Z" }, - { url = "https://files.pythonhosted.org/packages/03/f3/e699e19cabe96bbac5189c04aaa970718f0105cff03d458dc5e2b6bd1e8c/scipy-1.15.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:bae43364d600fdc3ac327db99659dcb79e6e7ecd279a75fe1266669d9a652828", size = 36545159, upload-time = "2025-02-17T00:34:26.724Z" }, - { url = "https://files.pythonhosted.org/packages/af/f5/ab3838e56fe5cc22383d6fcf2336e48c8fe33e944b9037fbf6cbdf5a11f8/scipy-1.15.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:f031846580d9acccd0044efd1a90e6f4df3a6e12b4b6bd694a7bc03a89892b28", size = 39136566, upload-time = "2025-02-17T00:34:34.512Z" }, - { url = "https://files.pythonhosted.org/packages/0a/c8/b3f566db71461cabd4b2d5b39bcc24a7e1c119535c8361f81426be39bb47/scipy-1.15.2-cp313-cp313t-win_amd64.whl", hash = "sha256:fe8a9eb875d430d81755472c5ba75e84acc980e4a8f6204d402849234d3017db", size = 40477705, upload-time = "2025-02-17T00:34:43.619Z" }, +version = "1.15.3" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version < '3.11' and sys_platform == 'linux'", + "python_full_version < '3.11' and sys_platform != 'linux'", +] +dependencies = [ + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/0f/37/6964b830433e654ec7485e45a00fc9a27cf868d622838f6b6d9c5ec0d532/scipy-1.15.3.tar.gz", hash = "sha256:eae3cf522bc7df64b42cad3925c876e1b0b6c35c1337c93e12c0f366f55b0eaf", size = 59419214, upload-time = "2025-05-08T16:13:05.955Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/78/2f/4966032c5f8cc7e6a60f1b2e0ad686293b9474b65246b0c642e3ef3badd0/scipy-1.15.3-cp310-cp310-macosx_10_13_x86_64.whl", hash = "sha256:a345928c86d535060c9c2b25e71e87c39ab2f22fc96e9636bd74d1dbf9de448c", size = 38702770, upload-time = "2025-05-08T16:04:20.849Z" }, + { url = "https://files.pythonhosted.org/packages/a0/6e/0c3bf90fae0e910c274db43304ebe25a6b391327f3f10b5dcc638c090795/scipy-1.15.3-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:ad3432cb0f9ed87477a8d97f03b763fd1d57709f1bbde3c9369b1dff5503b253", size = 30094511, upload-time = "2025-05-08T16:04:27.103Z" }, + { url = "https://files.pythonhosted.org/packages/ea/b1/4deb37252311c1acff7f101f6453f0440794f51b6eacb1aad4459a134081/scipy-1.15.3-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:aef683a9ae6eb00728a542b796f52a5477b78252edede72b8327a886ab63293f", size = 22368151, upload-time = "2025-05-08T16:04:31.731Z" }, + { url = "https://files.pythonhosted.org/packages/38/7d/f457626e3cd3c29b3a49ca115a304cebb8cc6f31b04678f03b216899d3c6/scipy-1.15.3-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:1c832e1bd78dea67d5c16f786681b28dd695a8cb1fb90af2e27580d3d0967e92", size = 25121732, upload-time = "2025-05-08T16:04:36.596Z" }, + { url = "https://files.pythonhosted.org/packages/db/0a/92b1de4a7adc7a15dcf5bddc6e191f6f29ee663b30511ce20467ef9b82e4/scipy-1.15.3-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:263961f658ce2165bbd7b99fa5135195c3a12d9bef045345016b8b50c315cb82", size = 35547617, upload-time = "2025-05-08T16:04:43.546Z" }, + { url = "https://files.pythonhosted.org/packages/8e/6d/41991e503e51fc1134502694c5fa7a1671501a17ffa12716a4a9151af3df/scipy-1.15.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9e2abc762b0811e09a0d3258abee2d98e0c703eee49464ce0069590846f31d40", size = 37662964, upload-time = "2025-05-08T16:04:49.431Z" }, + { url = "https://files.pythonhosted.org/packages/25/e1/3df8f83cb15f3500478c889be8fb18700813b95e9e087328230b98d547ff/scipy-1.15.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:ed7284b21a7a0c8f1b6e5977ac05396c0d008b89e05498c8b7e8f4a1423bba0e", size = 37238749, upload-time = "2025-05-08T16:04:55.215Z" }, + { url = "https://files.pythonhosted.org/packages/93/3e/b3257cf446f2a3533ed7809757039016b74cd6f38271de91682aa844cfc5/scipy-1.15.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:5380741e53df2c566f4d234b100a484b420af85deb39ea35a1cc1be84ff53a5c", size = 40022383, upload-time = "2025-05-08T16:05:01.914Z" }, + { url = "https://files.pythonhosted.org/packages/d1/84/55bc4881973d3f79b479a5a2e2df61c8c9a04fcb986a213ac9c02cfb659b/scipy-1.15.3-cp310-cp310-win_amd64.whl", hash = "sha256:9d61e97b186a57350f6d6fd72640f9e99d5a4a2b8fbf4b9ee9a841eab327dc13", size = 41259201, upload-time = "2025-05-08T16:05:08.166Z" }, + { url = "https://files.pythonhosted.org/packages/96/ab/5cc9f80f28f6a7dff646c5756e559823614a42b1939d86dd0ed550470210/scipy-1.15.3-cp311-cp311-macosx_10_13_x86_64.whl", hash = "sha256:993439ce220d25e3696d1b23b233dd010169b62f6456488567e830654ee37a6b", size = 38714255, upload-time = "2025-05-08T16:05:14.596Z" }, + { url = "https://files.pythonhosted.org/packages/4a/4a/66ba30abe5ad1a3ad15bfb0b59d22174012e8056ff448cb1644deccbfed2/scipy-1.15.3-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:34716e281f181a02341ddeaad584205bd2fd3c242063bd3423d61ac259ca7eba", size = 30111035, upload-time = "2025-05-08T16:05:20.152Z" }, + { url = "https://files.pythonhosted.org/packages/4b/fa/a7e5b95afd80d24313307f03624acc65801846fa75599034f8ceb9e2cbf6/scipy-1.15.3-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:3b0334816afb8b91dab859281b1b9786934392aa3d527cd847e41bb6f45bee65", size = 22384499, upload-time = "2025-05-08T16:05:24.494Z" }, + { url = "https://files.pythonhosted.org/packages/17/99/f3aaddccf3588bb4aea70ba35328c204cadd89517a1612ecfda5b2dd9d7a/scipy-1.15.3-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:6db907c7368e3092e24919b5e31c76998b0ce1684d51a90943cb0ed1b4ffd6c1", size = 25152602, upload-time = "2025-05-08T16:05:29.313Z" }, + { url = "https://files.pythonhosted.org/packages/56/c5/1032cdb565f146109212153339f9cb8b993701e9fe56b1c97699eee12586/scipy-1.15.3-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:721d6b4ef5dc82ca8968c25b111e307083d7ca9091bc38163fb89243e85e3889", size = 35503415, upload-time = "2025-05-08T16:05:34.699Z" }, + { url = "https://files.pythonhosted.org/packages/bd/37/89f19c8c05505d0601ed5650156e50eb881ae3918786c8fd7262b4ee66d3/scipy-1.15.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:39cb9c62e471b1bb3750066ecc3a3f3052b37751c7c3dfd0fd7e48900ed52982", size = 37652622, upload-time = "2025-05-08T16:05:40.762Z" }, + { url = "https://files.pythonhosted.org/packages/7e/31/be59513aa9695519b18e1851bb9e487de66f2d31f835201f1b42f5d4d475/scipy-1.15.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:795c46999bae845966368a3c013e0e00947932d68e235702b5c3f6ea799aa8c9", size = 37244796, upload-time = "2025-05-08T16:05:48.119Z" }, + { url = "https://files.pythonhosted.org/packages/10/c0/4f5f3eeccc235632aab79b27a74a9130c6c35df358129f7ac8b29f562ac7/scipy-1.15.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:18aaacb735ab38b38db42cb01f6b92a2d0d4b6aabefeb07f02849e47f8fb3594", size = 40047684, upload-time = "2025-05-08T16:05:54.22Z" }, + { url = "https://files.pythonhosted.org/packages/ab/a7/0ddaf514ce8a8714f6ed243a2b391b41dbb65251affe21ee3077ec45ea9a/scipy-1.15.3-cp311-cp311-win_amd64.whl", hash = "sha256:ae48a786a28412d744c62fd7816a4118ef97e5be0bee968ce8f0a2fba7acf3bb", size = 41246504, upload-time = "2025-05-08T16:06:00.437Z" }, + { url = "https://files.pythonhosted.org/packages/37/4b/683aa044c4162e10ed7a7ea30527f2cbd92e6999c10a8ed8edb253836e9c/scipy-1.15.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:6ac6310fdbfb7aa6612408bd2f07295bcbd3fda00d2d702178434751fe48e019", size = 38766735, upload-time = "2025-05-08T16:06:06.471Z" }, + { url = "https://files.pythonhosted.org/packages/7b/7e/f30be3d03de07f25dc0ec926d1681fed5c732d759ac8f51079708c79e680/scipy-1.15.3-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:185cd3d6d05ca4b44a8f1595af87f9c372bb6acf9c808e99aa3e9aa03bd98cf6", size = 30173284, upload-time = "2025-05-08T16:06:11.686Z" }, + { url = "https://files.pythonhosted.org/packages/07/9c/0ddb0d0abdabe0d181c1793db51f02cd59e4901da6f9f7848e1f96759f0d/scipy-1.15.3-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:05dc6abcd105e1a29f95eada46d4a3f251743cfd7d3ae8ddb4088047f24ea477", size = 22446958, upload-time = "2025-05-08T16:06:15.97Z" }, + { url = "https://files.pythonhosted.org/packages/af/43/0bce905a965f36c58ff80d8bea33f1f9351b05fad4beaad4eae34699b7a1/scipy-1.15.3-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:06efcba926324df1696931a57a176c80848ccd67ce6ad020c810736bfd58eb1c", size = 25242454, upload-time = "2025-05-08T16:06:20.394Z" }, + { url = "https://files.pythonhosted.org/packages/56/30/a6f08f84ee5b7b28b4c597aca4cbe545535c39fe911845a96414700b64ba/scipy-1.15.3-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c05045d8b9bfd807ee1b9f38761993297b10b245f012b11b13b91ba8945f7e45", size = 35210199, upload-time = "2025-05-08T16:06:26.159Z" }, + { url = "https://files.pythonhosted.org/packages/0b/1f/03f52c282437a168ee2c7c14a1a0d0781a9a4a8962d84ac05c06b4c5b555/scipy-1.15.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:271e3713e645149ea5ea3e97b57fdab61ce61333f97cfae392c28ba786f9bb49", size = 37309455, upload-time = "2025-05-08T16:06:32.778Z" }, + { url = "https://files.pythonhosted.org/packages/89/b1/fbb53137f42c4bf630b1ffdfc2151a62d1d1b903b249f030d2b1c0280af8/scipy-1.15.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6cfd56fc1a8e53f6e89ba3a7a7251f7396412d655bca2aa5611c8ec9a6784a1e", size = 36885140, upload-time = "2025-05-08T16:06:39.249Z" }, + { url = "https://files.pythonhosted.org/packages/2e/2e/025e39e339f5090df1ff266d021892694dbb7e63568edcfe43f892fa381d/scipy-1.15.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:0ff17c0bb1cb32952c09217d8d1eed9b53d1463e5f1dd6052c7857f83127d539", size = 39710549, upload-time = "2025-05-08T16:06:45.729Z" }, + { url = "https://files.pythonhosted.org/packages/e6/eb/3bf6ea8ab7f1503dca3a10df2e4b9c3f6b3316df07f6c0ded94b281c7101/scipy-1.15.3-cp312-cp312-win_amd64.whl", hash = "sha256:52092bc0472cfd17df49ff17e70624345efece4e1a12b23783a1ac59a1b728ed", size = 40966184, upload-time = "2025-05-08T16:06:52.623Z" }, + { url = "https://files.pythonhosted.org/packages/73/18/ec27848c9baae6e0d6573eda6e01a602e5649ee72c27c3a8aad673ebecfd/scipy-1.15.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:2c620736bcc334782e24d173c0fdbb7590a0a436d2fdf39310a8902505008759", size = 38728256, upload-time = "2025-05-08T16:06:58.696Z" }, + { url = "https://files.pythonhosted.org/packages/74/cd/1aef2184948728b4b6e21267d53b3339762c285a46a274ebb7863c9e4742/scipy-1.15.3-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:7e11270a000969409d37ed399585ee530b9ef6aa99d50c019de4cb01e8e54e62", size = 30109540, upload-time = "2025-05-08T16:07:04.209Z" }, + { url = "https://files.pythonhosted.org/packages/5b/d8/59e452c0a255ec352bd0a833537a3bc1bfb679944c4938ab375b0a6b3a3e/scipy-1.15.3-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:8c9ed3ba2c8a2ce098163a9bdb26f891746d02136995df25227a20e71c396ebb", size = 22383115, upload-time = "2025-05-08T16:07:08.998Z" }, + { url = "https://files.pythonhosted.org/packages/08/f5/456f56bbbfccf696263b47095291040655e3cbaf05d063bdc7c7517f32ac/scipy-1.15.3-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:0bdd905264c0c9cfa74a4772cdb2070171790381a5c4d312c973382fc6eaf730", size = 25163884, upload-time = "2025-05-08T16:07:14.091Z" }, + { url = "https://files.pythonhosted.org/packages/a2/66/a9618b6a435a0f0c0b8a6d0a2efb32d4ec5a85f023c2b79d39512040355b/scipy-1.15.3-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:79167bba085c31f38603e11a267d862957cbb3ce018d8b38f79ac043bc92d825", size = 35174018, upload-time = "2025-05-08T16:07:19.427Z" }, + { url = "https://files.pythonhosted.org/packages/b5/09/c5b6734a50ad4882432b6bb7c02baf757f5b2f256041da5df242e2d7e6b6/scipy-1.15.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c9deabd6d547aee2c9a81dee6cc96c6d7e9a9b1953f74850c179f91fdc729cb7", size = 37269716, upload-time = "2025-05-08T16:07:25.712Z" }, + { url = "https://files.pythonhosted.org/packages/77/0a/eac00ff741f23bcabd352731ed9b8995a0a60ef57f5fd788d611d43d69a1/scipy-1.15.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:dde4fc32993071ac0c7dd2d82569e544f0bdaff66269cb475e0f369adad13f11", size = 36872342, upload-time = "2025-05-08T16:07:31.468Z" }, + { url = "https://files.pythonhosted.org/packages/fe/54/4379be86dd74b6ad81551689107360d9a3e18f24d20767a2d5b9253a3f0a/scipy-1.15.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:f77f853d584e72e874d87357ad70f44b437331507d1c311457bed8ed2b956126", size = 39670869, upload-time = "2025-05-08T16:07:38.002Z" }, + { url = "https://files.pythonhosted.org/packages/87/2e/892ad2862ba54f084ffe8cc4a22667eaf9c2bcec6d2bff1d15713c6c0703/scipy-1.15.3-cp313-cp313-win_amd64.whl", hash = "sha256:b90ab29d0c37ec9bf55424c064312930ca5f4bde15ee8619ee44e69319aab163", size = 40988851, upload-time = "2025-05-08T16:08:33.671Z" }, + { url = "https://files.pythonhosted.org/packages/1b/e9/7a879c137f7e55b30d75d90ce3eb468197646bc7b443ac036ae3fe109055/scipy-1.15.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3ac07623267feb3ae308487c260ac684b32ea35fd81e12845039952f558047b8", size = 38863011, upload-time = "2025-05-08T16:07:44.039Z" }, + { url = "https://files.pythonhosted.org/packages/51/d1/226a806bbd69f62ce5ef5f3ffadc35286e9fbc802f606a07eb83bf2359de/scipy-1.15.3-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:6487aa99c2a3d509a5227d9a5e889ff05830a06b2ce08ec30df6d79db5fcd5c5", size = 30266407, upload-time = "2025-05-08T16:07:49.891Z" }, + { url = "https://files.pythonhosted.org/packages/e5/9b/f32d1d6093ab9eeabbd839b0f7619c62e46cc4b7b6dbf05b6e615bbd4400/scipy-1.15.3-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:50f9e62461c95d933d5c5ef4a1f2ebf9a2b4e83b0db374cb3f1de104d935922e", size = 22540030, upload-time = "2025-05-08T16:07:54.121Z" }, + { url = "https://files.pythonhosted.org/packages/e7/29/c278f699b095c1a884f29fda126340fcc201461ee8bfea5c8bdb1c7c958b/scipy-1.15.3-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:14ed70039d182f411ffc74789a16df3835e05dc469b898233a245cdfd7f162cb", size = 25218709, upload-time = "2025-05-08T16:07:58.506Z" }, + { url = "https://files.pythonhosted.org/packages/24/18/9e5374b617aba742a990581373cd6b68a2945d65cc588482749ef2e64467/scipy-1.15.3-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0a769105537aa07a69468a0eefcd121be52006db61cdd8cac8a0e68980bbb723", size = 34809045, upload-time = "2025-05-08T16:08:03.929Z" }, + { url = "https://files.pythonhosted.org/packages/e1/fe/9c4361e7ba2927074360856db6135ef4904d505e9b3afbbcb073c4008328/scipy-1.15.3-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9db984639887e3dffb3928d118145ffe40eff2fa40cb241a306ec57c219ebbbb", size = 36703062, upload-time = "2025-05-08T16:08:09.558Z" }, + { url = "https://files.pythonhosted.org/packages/b7/8e/038ccfe29d272b30086b25a4960f757f97122cb2ec42e62b460d02fe98e9/scipy-1.15.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:40e54d5c7e7ebf1aa596c374c49fa3135f04648a0caabcb66c52884b943f02b4", size = 36393132, upload-time = "2025-05-08T16:08:15.34Z" }, + { url = "https://files.pythonhosted.org/packages/10/7e/5c12285452970be5bdbe8352c619250b97ebf7917d7a9a9e96b8a8140f17/scipy-1.15.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:5e721fed53187e71d0ccf382b6bf977644c533e506c4d33c3fb24de89f5c3ed5", size = 38979503, upload-time = "2025-05-08T16:08:21.513Z" }, + { url = "https://files.pythonhosted.org/packages/81/06/0a5e5349474e1cbc5757975b21bd4fad0e72ebf138c5592f191646154e06/scipy-1.15.3-cp313-cp313t-win_amd64.whl", hash = "sha256:76ad1fb5f8752eabf0fa02e4cc0336b4e8f021e2d5f061ed37d6d264db35e3ca", size = 40308097, upload-time = "2025-05-08T16:08:27.627Z" }, +] + +[[package]] +name = "scipy" +version = "1.16.1" +source = { registry = "https://pypi.org/simple" } +resolution-markers = [ + "python_full_version >= '3.13' and sys_platform == 'linux'", + "python_full_version == '3.12.*' and sys_platform == 'linux'", + "python_full_version >= '3.13' and sys_platform != 'linux'", + "python_full_version == '3.12.*' and sys_platform != 'linux'", + "python_full_version == '3.11.*' and sys_platform == 'linux'", + "python_full_version == '3.11.*' and sys_platform != 'linux'", +] +dependencies = [ + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/f5/4a/b927028464795439faec8eaf0b03b011005c487bb2d07409f28bf30879c4/scipy-1.16.1.tar.gz", hash = "sha256:44c76f9e8b6e8e488a586190ab38016e4ed2f8a038af7cd3defa903c0a2238b3", size = 30580861, upload-time = "2025-07-27T16:33:30.834Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/da/91/812adc6f74409b461e3a5fa97f4f74c769016919203138a3bf6fc24ba4c5/scipy-1.16.1-cp311-cp311-macosx_10_14_x86_64.whl", hash = "sha256:c033fa32bab91dc98ca59d0cf23bb876454e2bb02cbe592d5023138778f70030", size = 36552519, upload-time = "2025-07-27T16:26:29.658Z" }, + { url = "https://files.pythonhosted.org/packages/47/18/8e355edcf3b71418d9e9f9acd2708cc3a6c27e8f98fde0ac34b8a0b45407/scipy-1.16.1-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:6e5c2f74e5df33479b5cd4e97a9104c511518fbd979aa9b8f6aec18b2e9ecae7", size = 28638010, upload-time = "2025-07-27T16:26:38.196Z" }, + { url = "https://files.pythonhosted.org/packages/d9/eb/e931853058607bdfbc11b86df19ae7a08686121c203483f62f1ecae5989c/scipy-1.16.1-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:0a55ffe0ba0f59666e90951971a884d1ff6f4ec3275a48f472cfb64175570f77", size = 20909790, upload-time = "2025-07-27T16:26:43.93Z" }, + { url = "https://files.pythonhosted.org/packages/45/0c/be83a271d6e96750cd0be2e000f35ff18880a46f05ce8b5d3465dc0f7a2a/scipy-1.16.1-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:f8a5d6cd147acecc2603fbd382fed6c46f474cccfcf69ea32582e033fb54dcfe", size = 23513352, upload-time = "2025-07-27T16:26:50.017Z" }, + { url = "https://files.pythonhosted.org/packages/7c/bf/fe6eb47e74f762f933cca962db7f2c7183acfdc4483bd1c3813cfe83e538/scipy-1.16.1-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:cb18899127278058bcc09e7b9966d41a5a43740b5bb8dcba401bd983f82e885b", size = 33534643, upload-time = "2025-07-27T16:26:57.503Z" }, + { url = "https://files.pythonhosted.org/packages/bb/ba/63f402e74875486b87ec6506a4f93f6d8a0d94d10467280f3d9d7837ce3a/scipy-1.16.1-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:adccd93a2fa937a27aae826d33e3bfa5edf9aa672376a4852d23a7cd67a2e5b7", size = 35376776, upload-time = "2025-07-27T16:27:06.639Z" }, + { url = "https://files.pythonhosted.org/packages/c3/b4/04eb9d39ec26a1b939689102da23d505ea16cdae3dbb18ffc53d1f831044/scipy-1.16.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:18aca1646a29ee9a0625a1be5637fa798d4d81fdf426481f06d69af828f16958", size = 35698906, upload-time = "2025-07-27T16:27:14.943Z" }, + { url = "https://files.pythonhosted.org/packages/04/d6/bb5468da53321baeb001f6e4e0d9049eadd175a4a497709939128556e3ec/scipy-1.16.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d85495cef541729a70cdddbbf3e6b903421bc1af3e8e3a9a72a06751f33b7c39", size = 38129275, upload-time = "2025-07-27T16:27:23.873Z" }, + { url = "https://files.pythonhosted.org/packages/c4/94/994369978509f227cba7dfb9e623254d0d5559506fe994aef4bea3ed469c/scipy-1.16.1-cp311-cp311-win_amd64.whl", hash = "sha256:226652fca853008119c03a8ce71ffe1b3f6d2844cc1686e8f9806edafae68596", size = 38644572, upload-time = "2025-07-27T16:27:32.637Z" }, + { url = "https://files.pythonhosted.org/packages/f8/d9/ec4864f5896232133f51382b54a08de91a9d1af7a76dfa372894026dfee2/scipy-1.16.1-cp312-cp312-macosx_10_14_x86_64.whl", hash = "sha256:81b433bbeaf35728dad619afc002db9b189e45eebe2cd676effe1fb93fef2b9c", size = 36575194, upload-time = "2025-07-27T16:27:41.321Z" }, + { url = "https://files.pythonhosted.org/packages/5c/6d/40e81ecfb688e9d25d34a847dca361982a6addf8e31f0957b1a54fbfa994/scipy-1.16.1-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:886cc81fdb4c6903a3bb0464047c25a6d1016fef77bb97949817d0c0d79f9e04", size = 28594590, upload-time = "2025-07-27T16:27:49.204Z" }, + { url = "https://files.pythonhosted.org/packages/0e/37/9f65178edfcc629377ce9a64fc09baebea18c80a9e57ae09a52edf84880b/scipy-1.16.1-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:15240c3aac087a522b4eaedb09f0ad061753c5eebf1ea430859e5bf8640d5919", size = 20866458, upload-time = "2025-07-27T16:27:54.98Z" }, + { url = "https://files.pythonhosted.org/packages/2c/7b/749a66766871ea4cb1d1ea10f27004db63023074c22abed51f22f09770e0/scipy-1.16.1-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:65f81a25805f3659b48126b5053d9e823d3215e4a63730b5e1671852a1705921", size = 23539318, upload-time = "2025-07-27T16:28:01.604Z" }, + { url = "https://files.pythonhosted.org/packages/c4/db/8d4afec60eb833a666434d4541a3151eedbf2494ea6d4d468cbe877f00cd/scipy-1.16.1-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:6c62eea7f607f122069b9bad3f99489ddca1a5173bef8a0c75555d7488b6f725", size = 33292899, upload-time = "2025-07-27T16:28:09.147Z" }, + { url = "https://files.pythonhosted.org/packages/51/1e/79023ca3bbb13a015d7d2757ecca3b81293c663694c35d6541b4dca53e98/scipy-1.16.1-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f965bbf3235b01c776115ab18f092a95aa74c271a52577bcb0563e85738fd618", size = 35162637, upload-time = "2025-07-27T16:28:17.535Z" }, + { url = "https://files.pythonhosted.org/packages/b6/49/0648665f9c29fdaca4c679182eb972935b3b4f5ace41d323c32352f29816/scipy-1.16.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:f006e323874ffd0b0b816d8c6a8e7f9a73d55ab3b8c3f72b752b226d0e3ac83d", size = 35490507, upload-time = "2025-07-27T16:28:25.705Z" }, + { url = "https://files.pythonhosted.org/packages/62/8f/66cbb9d6bbb18d8c658f774904f42a92078707a7c71e5347e8bf2f52bb89/scipy-1.16.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e8fd15fc5085ab4cca74cb91fe0a4263b1f32e4420761ddae531ad60934c2119", size = 37923998, upload-time = "2025-07-27T16:28:34.339Z" }, + { url = "https://files.pythonhosted.org/packages/14/c3/61f273ae550fbf1667675701112e380881905e28448c080b23b5a181df7c/scipy-1.16.1-cp312-cp312-win_amd64.whl", hash = "sha256:f7b8013c6c066609577d910d1a2a077021727af07b6fab0ee22c2f901f22352a", size = 38508060, upload-time = "2025-07-27T16:28:43.242Z" }, + { url = "https://files.pythonhosted.org/packages/93/0b/b5c99382b839854a71ca9482c684e3472badc62620287cbbdab499b75ce6/scipy-1.16.1-cp313-cp313-macosx_10_14_x86_64.whl", hash = "sha256:5451606823a5e73dfa621a89948096c6528e2896e40b39248295d3a0138d594f", size = 36533717, upload-time = "2025-07-27T16:28:51.706Z" }, + { url = "https://files.pythonhosted.org/packages/eb/e5/69ab2771062c91e23e07c12e7d5033a6b9b80b0903ee709c3c36b3eb520c/scipy-1.16.1-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:89728678c5ca5abd610aee148c199ac1afb16e19844401ca97d43dc548a354eb", size = 28570009, upload-time = "2025-07-27T16:28:57.017Z" }, + { url = "https://files.pythonhosted.org/packages/f4/69/bd75dbfdd3cf524f4d753484d723594aed62cfaac510123e91a6686d520b/scipy-1.16.1-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:e756d688cb03fd07de0fffad475649b03cb89bee696c98ce508b17c11a03f95c", size = 20841942, upload-time = "2025-07-27T16:29:01.152Z" }, + { url = "https://files.pythonhosted.org/packages/ea/74/add181c87663f178ba7d6144b370243a87af8476664d5435e57d599e6874/scipy-1.16.1-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:5aa2687b9935da3ed89c5dbed5234576589dd28d0bf7cd237501ccfbdf1ad608", size = 23498507, upload-time = "2025-07-27T16:29:05.202Z" }, + { url = "https://files.pythonhosted.org/packages/1d/74/ece2e582a0d9550cee33e2e416cc96737dce423a994d12bbe59716f47ff1/scipy-1.16.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:0851f6a1e537fe9399f35986897e395a1aa61c574b178c0d456be5b1a0f5ca1f", size = 33286040, upload-time = "2025-07-27T16:29:10.201Z" }, + { url = "https://files.pythonhosted.org/packages/e4/82/08e4076df538fb56caa1d489588d880ec7c52d8273a606bb54d660528f7c/scipy-1.16.1-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fedc2cbd1baed37474b1924c331b97bdff611d762c196fac1a9b71e67b813b1b", size = 35176096, upload-time = "2025-07-27T16:29:17.091Z" }, + { url = "https://files.pythonhosted.org/packages/fa/79/cd710aab8c921375711a8321c6be696e705a120e3011a643efbbcdeeabcc/scipy-1.16.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:2ef500e72f9623a6735769e4b93e9dcb158d40752cdbb077f305487e3e2d1f45", size = 35490328, upload-time = "2025-07-27T16:29:22.928Z" }, + { url = "https://files.pythonhosted.org/packages/71/73/e9cc3d35ee4526d784520d4494a3e1ca969b071fb5ae5910c036a375ceec/scipy-1.16.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:978d8311674b05a8f7ff2ea6c6bce5d8b45a0cb09d4c5793e0318f448613ea65", size = 37939921, upload-time = "2025-07-27T16:29:29.108Z" }, + { url = "https://files.pythonhosted.org/packages/21/12/c0efd2941f01940119b5305c375ae5c0fcb7ec193f806bd8f158b73a1782/scipy-1.16.1-cp313-cp313-win_amd64.whl", hash = "sha256:81929ed0fa7a5713fcdd8b2e6f73697d3b4c4816d090dd34ff937c20fa90e8ab", size = 38479462, upload-time = "2025-07-27T16:30:24.078Z" }, + { url = "https://files.pythonhosted.org/packages/7a/19/c3d08b675260046a991040e1ea5d65f91f40c7df1045fffff412dcfc6765/scipy-1.16.1-cp313-cp313t-macosx_10_14_x86_64.whl", hash = "sha256:bcc12db731858abda693cecdb3bdc9e6d4bd200213f49d224fe22df82687bdd6", size = 36938832, upload-time = "2025-07-27T16:29:35.057Z" }, + { url = "https://files.pythonhosted.org/packages/81/f2/ce53db652c033a414a5b34598dba6b95f3d38153a2417c5a3883da429029/scipy-1.16.1-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:744d977daa4becb9fc59135e75c069f8d301a87d64f88f1e602a9ecf51e77b27", size = 29093084, upload-time = "2025-07-27T16:29:40.201Z" }, + { url = "https://files.pythonhosted.org/packages/a9/ae/7a10ff04a7dc15f9057d05b33737ade244e4bd195caa3f7cc04d77b9e214/scipy-1.16.1-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:dc54f76ac18073bcecffb98d93f03ed6b81a92ef91b5d3b135dcc81d55a724c7", size = 21365098, upload-time = "2025-07-27T16:29:44.295Z" }, + { url = "https://files.pythonhosted.org/packages/36/ac/029ff710959932ad3c2a98721b20b405f05f752f07344622fd61a47c5197/scipy-1.16.1-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:367d567ee9fc1e9e2047d31f39d9d6a7a04e0710c86e701e053f237d14a9b4f6", size = 23896858, upload-time = "2025-07-27T16:29:48.784Z" }, + { url = "https://files.pythonhosted.org/packages/71/13/d1ef77b6bd7898720e1f0b6b3743cb945f6c3cafa7718eaac8841035ab60/scipy-1.16.1-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:4cf5785e44e19dcd32a0e4807555e1e9a9b8d475c6afff3d21c3c543a6aa84f4", size = 33438311, upload-time = "2025-07-27T16:29:54.164Z" }, + { url = "https://files.pythonhosted.org/packages/2d/e0/e64a6821ffbb00b4c5b05169f1c1fddb4800e9307efe3db3788995a82a2c/scipy-1.16.1-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:3d0b80fb26d3e13a794c71d4b837e2a589d839fd574a6bbb4ee1288c213ad4a3", size = 35279542, upload-time = "2025-07-27T16:30:00.249Z" }, + { url = "https://files.pythonhosted.org/packages/57/59/0dc3c8b43e118f1e4ee2b798dcc96ac21bb20014e5f1f7a8e85cc0653bdb/scipy-1.16.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:8503517c44c18d1030d666cb70aaac1cc8913608816e06742498833b128488b7", size = 35667665, upload-time = "2025-07-27T16:30:05.916Z" }, + { url = "https://files.pythonhosted.org/packages/45/5f/844ee26e34e2f3f9f8febb9343748e72daeaec64fe0c70e9bf1ff84ec955/scipy-1.16.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:30cc4bb81c41831ecfd6dc450baf48ffd80ef5aed0f5cf3ea775740e80f16ecc", size = 38045210, upload-time = "2025-07-27T16:30:11.655Z" }, + { url = "https://files.pythonhosted.org/packages/8d/d7/210f2b45290f444f1de64bc7353aa598ece9f0e90c384b4a156f9b1a5063/scipy-1.16.1-cp313-cp313t-win_amd64.whl", hash = "sha256:c24fa02f7ed23ae514460a22c57eca8f530dbfa50b1cfdbf4f37c05b5309cc39", size = 38593661, upload-time = "2025-07-27T16:30:17.825Z" }, + { url = "https://files.pythonhosted.org/packages/81/ea/84d481a5237ed223bd3d32d6e82d7a6a96e34756492666c260cef16011d1/scipy-1.16.1-cp314-cp314-macosx_10_14_x86_64.whl", hash = "sha256:796a5a9ad36fa3a782375db8f4241ab02a091308eb079746bc0f874c9b998318", size = 36525921, upload-time = "2025-07-27T16:30:30.081Z" }, + { url = "https://files.pythonhosted.org/packages/4e/9f/d9edbdeff9f3a664807ae3aea383e10afaa247e8e6255e6d2aa4515e8863/scipy-1.16.1-cp314-cp314-macosx_12_0_arm64.whl", hash = "sha256:3ea0733a2ff73fd6fdc5fecca54ee9b459f4d74f00b99aced7d9a3adb43fb1cc", size = 28564152, upload-time = "2025-07-27T16:30:35.336Z" }, + { url = "https://files.pythonhosted.org/packages/3b/95/8125bcb1fe04bc267d103e76516243e8d5e11229e6b306bda1024a5423d1/scipy-1.16.1-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:85764fb15a2ad994e708258bb4ed8290d1305c62a4e1ef07c414356a24fcfbf8", size = 20836028, upload-time = "2025-07-27T16:30:39.421Z" }, + { url = "https://files.pythonhosted.org/packages/77/9c/bf92e215701fc70bbcd3d14d86337cf56a9b912a804b9c776a269524a9e9/scipy-1.16.1-cp314-cp314-macosx_14_0_x86_64.whl", hash = "sha256:ca66d980469cb623b1759bdd6e9fd97d4e33a9fad5b33771ced24d0cb24df67e", size = 23489666, upload-time = "2025-07-27T16:30:43.663Z" }, + { url = "https://files.pythonhosted.org/packages/5e/00/5e941d397d9adac41b02839011594620d54d99488d1be5be755c00cde9ee/scipy-1.16.1-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e7cc1ffcc230f568549fc56670bcf3df1884c30bd652c5da8138199c8c76dae0", size = 33358318, upload-time = "2025-07-27T16:30:48.982Z" }, + { url = "https://files.pythonhosted.org/packages/0e/87/8db3aa10dde6e3e8e7eb0133f24baa011377d543f5b19c71469cf2648026/scipy-1.16.1-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:3ddfb1e8d0b540cb4ee9c53fc3dea3186f97711248fb94b4142a1b27178d8b4b", size = 35185724, upload-time = "2025-07-27T16:30:54.26Z" }, + { url = "https://files.pythonhosted.org/packages/89/b4/6ab9ae443216807622bcff02690262d8184078ea467efee2f8c93288a3b1/scipy-1.16.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:4dc0e7be79e95d8ba3435d193e0d8ce372f47f774cffd882f88ea4e1e1ddc731", size = 35554335, upload-time = "2025-07-27T16:30:59.765Z" }, + { url = "https://files.pythonhosted.org/packages/9c/9a/d0e9dc03c5269a1afb60661118296a32ed5d2c24298af61b676c11e05e56/scipy-1.16.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:f23634f9e5adb51b2a77766dac217063e764337fbc816aa8ad9aaebcd4397fd3", size = 37960310, upload-time = "2025-07-27T16:31:06.151Z" }, + { url = "https://files.pythonhosted.org/packages/5e/00/c8f3130a50521a7977874817ca89e0599b1b4ee8e938bad8ae798a0e1f0d/scipy-1.16.1-cp314-cp314-win_amd64.whl", hash = "sha256:57d75524cb1c5a374958a2eae3d84e1929bb971204cc9d52213fb8589183fc19", size = 39319239, upload-time = "2025-07-27T16:31:59.942Z" }, + { url = "https://files.pythonhosted.org/packages/f2/f2/1ca3eda54c3a7e4c92f6acef7db7b3a057deb135540d23aa6343ef8ad333/scipy-1.16.1-cp314-cp314t-macosx_10_14_x86_64.whl", hash = "sha256:d8da7c3dd67bcd93f15618938f43ed0995982eb38973023d46d4646c4283ad65", size = 36939460, upload-time = "2025-07-27T16:31:11.865Z" }, + { url = "https://files.pythonhosted.org/packages/80/30/98c2840b293a132400c0940bb9e140171dcb8189588619048f42b2ce7b4f/scipy-1.16.1-cp314-cp314t-macosx_12_0_arm64.whl", hash = "sha256:cc1d2f2fd48ba1e0620554fe5bc44d3e8f5d4185c8c109c7fbdf5af2792cfad2", size = 29093322, upload-time = "2025-07-27T16:31:17.045Z" }, + { url = "https://files.pythonhosted.org/packages/c1/e6/1e6e006e850622cf2a039b62d1a6ddc4497d4851e58b68008526f04a9a00/scipy-1.16.1-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:21a611ced9275cb861bacadbada0b8c0623bc00b05b09eb97f23b370fc2ae56d", size = 21365329, upload-time = "2025-07-27T16:31:21.188Z" }, + { url = "https://files.pythonhosted.org/packages/8e/02/72a5aa5b820589dda9a25e329ca752842bfbbaf635e36bc7065a9b42216e/scipy-1.16.1-cp314-cp314t-macosx_14_0_x86_64.whl", hash = "sha256:8dfbb25dffc4c3dd9371d8ab456ca81beeaf6f9e1c2119f179392f0dc1ab7695", size = 23897544, upload-time = "2025-07-27T16:31:25.408Z" }, + { url = "https://files.pythonhosted.org/packages/2b/dc/7122d806a6f9eb8a33532982234bed91f90272e990f414f2830cfe656e0b/scipy-1.16.1-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:f0ebb7204f063fad87fc0a0e4ff4a2ff40b2a226e4ba1b7e34bf4b79bf97cd86", size = 33442112, upload-time = "2025-07-27T16:31:30.62Z" }, + { url = "https://files.pythonhosted.org/packages/24/39/e383af23564daa1021a5b3afbe0d8d6a68ec639b943661841f44ac92de85/scipy-1.16.1-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f1b9e5962656f2734c2b285a8745358ecb4e4efbadd00208c80a389227ec61ff", size = 35286594, upload-time = "2025-07-27T16:31:36.112Z" }, + { url = "https://files.pythonhosted.org/packages/95/47/1a0b0aff40c3056d955f38b0df5d178350c3d74734ec54f9c68d23910be5/scipy-1.16.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:5e1a106f8c023d57a2a903e771228bf5c5b27b5d692088f457acacd3b54511e4", size = 35665080, upload-time = "2025-07-27T16:31:42.025Z" }, + { url = "https://files.pythonhosted.org/packages/64/df/ce88803e9ed6e27fe9b9abefa157cf2c80e4fa527cf17ee14be41f790ad4/scipy-1.16.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:709559a1db68a9abc3b2c8672c4badf1614f3b440b3ab326d86a5c0491eafae3", size = 38050306, upload-time = "2025-07-27T16:31:48.109Z" }, + { url = "https://files.pythonhosted.org/packages/6e/6c/a76329897a7cae4937d403e623aa6aaea616a0bb5b36588f0b9d1c9a3739/scipy-1.16.1-cp314-cp314t-win_amd64.whl", hash = "sha256:c0c804d60492a0aad7f5b2bb1862f4548b990049e27e828391ff2bf6f7199998", size = 39427705, upload-time = "2025-07-27T16:31:53.96Z" }, ] [[package]] name = "sentry-sdk" -version = "2.24.0" +version = "2.34.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "certifi" }, { name = "urllib3" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/fe/c6/61d3d5eecca9f0237251a89043d1bfa3fbeb4e35d48c1cbba2f6dde72664/sentry_sdk-2.24.0.tar.gz", hash = "sha256:4a4a8de31573c8ab14c9b866fd44cf783df062ca7b4a56ed0a108453abbc2a24", size = 317058, upload-time = "2025-03-21T12:33:59.537Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/c4/5d/dbfbda95daa8fdb22276513f930269149b338b67efb9ebf24ebf6d62ca9a/sentry_sdk-2.24.0-py2.py3-none-any.whl", hash = "sha256:7150cfe61dfd37d30b33d8d6b153d25e14c69bbcf6f4a98ffc97e92de88be215", size = 336890, upload-time = "2025-03-21T12:33:57.754Z" }, -] - -[[package]] -name = "setproctitle" -version = "1.3.5" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/c4/4d/6a840c8d2baa07b57329490e7094f90aac177a1d5226bc919046f1106860/setproctitle-1.3.5.tar.gz", hash = "sha256:1e6eaeaf8a734d428a95d8c104643b39af7d247d604f40a7bebcf3960a853c5e", size = 26737, upload-time = "2025-02-22T21:52:43.276Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/9d/e1/9ccff2682c38061baa07e128b60712bc18e3398aa7d5471c51a704f9d24c/setproctitle-1.3.5-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:02870e0cb0de7f68a7a8a5b23c2bc0ce63821cab3d9b126f9be80bb6cd674c80", size = 17256, upload-time = "2025-02-22T21:50:22.744Z" }, - { url = "https://files.pythonhosted.org/packages/ed/64/936c1f92d60052f11a8de9f90a4b7ec4996b8ebd6d67ba425ed214c80771/setproctitle-1.3.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:55b278135be742b8901067479626d909f6613bd2d2c4fd0de6bb46f80e07a919", size = 11893, upload-time = "2025-02-22T21:50:25.255Z" }, - { url = "https://files.pythonhosted.org/packages/01/2d/abc817b3778d9b1f7675020030379a0c39e0bf74b36af211b26191a63da3/setproctitle-1.3.5-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:53fc971f7bf7a674f571a23cdec70f2f0ac88152c59c06aa0808d0be6d834046", size = 31295, upload-time = "2025-02-22T21:50:27.389Z" }, - { url = "https://files.pythonhosted.org/packages/03/4d/e2055dfb1b492fd3a3b27deeaa642d81c580d48a16bc9b07afc3504af677/setproctitle-1.3.5-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fb0500e1bc6f00b8ba696c3743ddff14c8679e3c2ca9d292c008ac51488d17cf", size = 32637, upload-time = "2025-02-22T21:50:29.47Z" }, - { url = "https://files.pythonhosted.org/packages/89/28/a1f23d7d127dff59fe75ad671d1d5c83ab8cba10d0e343820b96d5d8a2f7/setproctitle-1.3.5-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:995b3ac1b5fe510f4e1d1c19ebf19f4bceb448f2d6e8d99ea23f33cb6f1a277e", size = 29772, upload-time = "2025-02-22T21:50:30.597Z" }, - { url = "https://files.pythonhosted.org/packages/df/46/2ea4d436c7d664d41df7e60fbd3103f1139a931638e998f478e870e72255/setproctitle-1.3.5-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a5a05e2c3fdfbda32b9c9da72d0506398d1efb5bd2c5981b9e12d3622eb3d4f9", size = 30811, upload-time = "2025-02-22T21:50:32.802Z" }, - { url = "https://files.pythonhosted.org/packages/45/60/4c17211c2d80e6fe9fa486fa3214d565d0cd9a6eff0b67e6219ddb2ba49c/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:310c7f4ca4c8476a9840b2cd4b22ee602a49a3c902fdcd2dd8284685abd10a9a", size = 30442, upload-time = "2025-02-22T21:50:35.057Z" }, - { url = "https://files.pythonhosted.org/packages/7e/bf/65a8f8f2d03cd9a9429cfa0d6b22282ff7a609a4d08602bcb8351a271bec/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:867af4a5c3d85484fbcc50ea88bcd375acf709cff88a3259575361849c0da351", size = 29492, upload-time = "2025-02-22T21:50:37.23Z" }, - { url = "https://files.pythonhosted.org/packages/c6/96/56f45f0b81fcc776f925c34e2699040df39cfc6b3cc7520d9b378314435b/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:8ec0a7fe9f1ba90900144489bc93ce7dd4dec3f3df1e7f188c9e58364fe4a4c5", size = 31947, upload-time = "2025-02-22T21:50:38.65Z" }, - { url = "https://files.pythonhosted.org/packages/ec/9d/6b697c1562b21368e579d820bca2a607e565638fd332247841eb65dec4b2/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:aaee7acba2733a14a886488b7495bfec4a8d6407124c04a0946dbde1684230a3", size = 29863, upload-time = "2025-02-22T21:50:40.775Z" }, - { url = "https://files.pythonhosted.org/packages/ba/0f/4551cbb120d003fa1284ee35d559366e09b513a87dfee02f804da1936054/setproctitle-1.3.5-cp310-cp310-win32.whl", hash = "sha256:bd2cccd972e4282af4ce2c13cd9ebdf07be157eabafd8ce648fffdc8ae6fbe28", size = 11471, upload-time = "2025-02-22T21:50:42.749Z" }, - { url = "https://files.pythonhosted.org/packages/a6/f4/2dd926687b7a3bdaa83533e2898f929e1ff3bdeb6aa271bdb1d4d5923c7e/setproctitle-1.3.5-cp310-cp310-win_amd64.whl", hash = "sha256:81f2328ac34c9584e1e5f87eea916c0bc48476a06606a07debae07acdd7ab5ea", size = 12196, upload-time = "2025-02-22T21:50:43.852Z" }, - { url = "https://files.pythonhosted.org/packages/ec/4a/9e0243c5df221102fb834a947f5753d9da06ad5f84e36b0e2e93f7865edb/setproctitle-1.3.5-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:1c8dcc250872385f2780a5ea58050b58cbc8b6a7e8444952a5a65c359886c593", size = 17256, upload-time = "2025-02-22T21:50:45.928Z" }, - { url = "https://files.pythonhosted.org/packages/c7/a1/76ad2ba6f5bd00609238e3d64eeded4598e742a5f25b5cc1a0efdae5f674/setproctitle-1.3.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:ca82fae9eb4800231dd20229f06e8919787135a5581da245b8b05e864f34cc8b", size = 11893, upload-time = "2025-02-22T21:50:47.167Z" }, - { url = "https://files.pythonhosted.org/packages/47/3a/75d11fedff5b21ba9a4c5fe3dfa5e596f831d094ef1896713a72e9e38833/setproctitle-1.3.5-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0424e1d33232322541cb36fb279ea5242203cd6f20de7b4fb2a11973d8e8c2ce", size = 31631, upload-time = "2025-02-22T21:50:50.863Z" }, - { url = "https://files.pythonhosted.org/packages/5a/12/58220de5600e0ed2e5562297173187d863db49babb03491ffe9c101299bc/setproctitle-1.3.5-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fec8340ab543144d04a9d805d80a0aad73fdeb54bea6ff94e70d39a676ea4ec0", size = 32975, upload-time = "2025-02-22T21:50:52.188Z" }, - { url = "https://files.pythonhosted.org/packages/fa/c4/fbb308680d83c1c7aa626950308318c6e6381a8273779163a31741f3c752/setproctitle-1.3.5-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:eab441c89f181271ab749077dcc94045a423e51f2fb0b120a1463ef9820a08d0", size = 30126, upload-time = "2025-02-22T21:50:53.496Z" }, - { url = "https://files.pythonhosted.org/packages/31/6e/baaf70bd9a881dd8c12cbccdd7ca0ff291024a37044a8245e942e12e7135/setproctitle-1.3.5-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d2c371550a2288901a0dcd84192691ebd3197a43c95f3e0b396ed6d1cedf5c6c", size = 31135, upload-time = "2025-02-22T21:50:54.931Z" }, - { url = "https://files.pythonhosted.org/packages/a6/dc/d8ab6b1c3d844dc14f596e3cce76604570848f8a67ba6a3812775ed2c015/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:78288ff5f9c415c56595b2257ad218936dd9fa726b36341b373b31ca958590fe", size = 30874, upload-time = "2025-02-22T21:50:57.042Z" }, - { url = "https://files.pythonhosted.org/packages/d4/84/62a359b3aa51228bd88f78b44ebb0256a5b96dd2487881c1e984a59b617d/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:f1f13a25fc46731acab518602bb1149bfd8b5fabedf8290a7c0926d61414769d", size = 29893, upload-time = "2025-02-22T21:50:59.644Z" }, - { url = "https://files.pythonhosted.org/packages/e2/d6/b3c52c03ee41e7f006e1a737e0db1c58d1dc28e258b83548e653d0c34f1c/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:1534d6cd3854d035e40bf4c091984cbdd4d555d7579676d406c53c8f187c006f", size = 32293, upload-time = "2025-02-22T21:51:01.777Z" }, - { url = "https://files.pythonhosted.org/packages/55/09/c0ba311879d9c05860503a7e2708ace85913b9a816786402a92c664fe930/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:62a01c76708daac78b9688ffb95268c57cb57fa90b543043cda01358912fe2db", size = 30247, upload-time = "2025-02-22T21:51:03.779Z" }, - { url = "https://files.pythonhosted.org/packages/9e/43/cc7155461f0b5a48aebdb87d78239ff3a51ebda0905de478d9fa6ab92d9c/setproctitle-1.3.5-cp311-cp311-win32.whl", hash = "sha256:ea07f29735d839eaed985990a0ec42c8aecefe8050da89fec35533d146a7826d", size = 11476, upload-time = "2025-02-22T21:51:05.746Z" }, - { url = "https://files.pythonhosted.org/packages/e7/57/6e937ac7aa52db69225f02db2cfdcb66ba1db6fdc65a4ddbdf78e214f72a/setproctitle-1.3.5-cp311-cp311-win_amd64.whl", hash = "sha256:ab3ae11e10d13d514d4a5a15b4f619341142ba3e18da48c40e8614c5a1b5e3c3", size = 12189, upload-time = "2025-02-22T21:51:07.837Z" }, - { url = "https://files.pythonhosted.org/packages/2b/19/04755958495de57e4891de50f03e77b3fe9ca6716a86de00faa00ad0ee5a/setproctitle-1.3.5-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:523424b9be4dea97d95b8a584b183f35c7bab2d0a3d995b01febf5b8a8de90e4", size = 17250, upload-time = "2025-02-22T21:51:09.785Z" }, - { url = "https://files.pythonhosted.org/packages/b9/3d/2ca9df5aa49b975296411dcbbe272cdb1c5e514c43b8be7d61751bb71a46/setproctitle-1.3.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:b6ec1d86c1b4d7b5f2bdceadf213310cf24696b82480a2a702194b8a0bfbcb47", size = 11878, upload-time = "2025-02-22T21:51:11.679Z" }, - { url = "https://files.pythonhosted.org/packages/36/d6/e90e23b4627e016a4f862d4f892be92c9765dd6bf1e27a48e52cd166d4a3/setproctitle-1.3.5-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ea6c505264275a43e9b2acd2acfc11ac33caf52bc3167c9fced4418a810f6b1c", size = 31940, upload-time = "2025-02-22T21:51:12.977Z" }, - { url = "https://files.pythonhosted.org/packages/15/13/167cdd55e00a8e10b36aad79646c3bf3c23fba0c08a9b8db9b74622c1b13/setproctitle-1.3.5-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0b91e68e6685998e6353f296100ecabc313a6cb3e413d66a03d74b988b61f5ff", size = 33370, upload-time = "2025-02-22T21:51:15.115Z" }, - { url = "https://files.pythonhosted.org/packages/9b/22/574a110527df133409a75053b7d6ff740993ccf30b8713d042f26840d351/setproctitle-1.3.5-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bc1fda208ae3a2285ad27aeab44c41daf2328abe58fa3270157a739866779199", size = 30628, upload-time = "2025-02-22T21:51:16.324Z" }, - { url = "https://files.pythonhosted.org/packages/52/79/78b05c7d792c9167b917acdab1773b1ff73b016560f45d8155be2baa1a82/setproctitle-1.3.5-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:828727d220e46f048b82289018300a64547b46aaed96bf8810c05fe105426b41", size = 31672, upload-time = "2025-02-22T21:51:17.791Z" }, - { url = "https://files.pythonhosted.org/packages/b0/62/4509735be062129694751ac55d5e1fbb6d86fa46a8689b7d5e2c23dae5b0/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:83b016221cf80028b2947be20630faa14e3e72a403e35f0ba29550b4e856767b", size = 31378, upload-time = "2025-02-22T21:51:19.404Z" }, - { url = "https://files.pythonhosted.org/packages/72/e7/b394c55934b89f00c2ef7d5e6f18cca5d8dfa26ef628700c4de0c85e3f3d/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:6d8a411e752e794d052434139ca4234ffeceeb8d8d8ddc390a9051d7942b2726", size = 30370, upload-time = "2025-02-22T21:51:21.218Z" }, - { url = "https://files.pythonhosted.org/packages/13/ee/e1f27bf52d2bec7060bb6311ab0ccede8de98ed5394e3a59e7a14a453fb5/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:50cfbf86b9c63a2c2903f1231f0a58edeb775e651ae1af84eec8430b0571f29b", size = 32875, upload-time = "2025-02-22T21:51:22.505Z" }, - { url = "https://files.pythonhosted.org/packages/6e/08/13b561085d2de53b9becfa5578545d99114e9ff2aa3dc151bcaadf80b17e/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:f3b5e2eacd572444770026c9dd3ddc7543ce427cdf452d40a408d1e95beefb30", size = 30903, upload-time = "2025-02-22T21:51:23.732Z" }, - { url = "https://files.pythonhosted.org/packages/65/f0/6cd06fffff2553be7b0571447d0c0ef8b727ef44cc2d6a33452677a311c8/setproctitle-1.3.5-cp312-cp312-win32.whl", hash = "sha256:cf4e3ded98027de2596c6cc5bbd3302adfb3ca315c848f56516bb0b7e88de1e9", size = 11468, upload-time = "2025-02-22T21:51:25.45Z" }, - { url = "https://files.pythonhosted.org/packages/c1/8c/e8a7cb568c4552618838941b332203bfc77ab0f2d67c1cb8f24dee0370ec/setproctitle-1.3.5-cp312-cp312-win_amd64.whl", hash = "sha256:f7a8c01ffd013dda2bed6e7d5cb59fbb609e72f805abf3ee98360f38f7758d9b", size = 12190, upload-time = "2025-02-22T21:51:26.78Z" }, - { url = "https://files.pythonhosted.org/packages/ab/78/d6b5aa3af2dd64f6c32e78fb85797b9725a3cdcbdf17dffc5838019918c3/setproctitle-1.3.5-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:162fd76781f57f42ddf27c475e5fef6a8df4fdd69b28dd554e53e2eb2bfe0f95", size = 17238, upload-time = "2025-02-22T21:51:28.451Z" }, - { url = "https://files.pythonhosted.org/packages/3d/00/14781f0ac28c7a37fe2ba321c276188ddd5ca73d69dab8a0f739d57b776b/setproctitle-1.3.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:4969d996bdfbe23bbd023cd0bae6c73a27371615c4ec5296a60cecce268659ef", size = 11867, upload-time = "2025-02-22T21:51:29.649Z" }, - { url = "https://files.pythonhosted.org/packages/f0/22/8430c879a8e3201508924a6cf45dba92b9a7b105fac8eebd0ef62e60fba9/setproctitle-1.3.5-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bd70c95a94473216e7c7a7a1f7d8ecbaca5b16d4ba93ddbfd32050fc485a8451", size = 32001, upload-time = "2025-02-22T21:51:32.21Z" }, - { url = "https://files.pythonhosted.org/packages/01/f2/b00fe72c20897695f85932d193a5c57ecf94cbf825c0fd4082e3fa3e00bd/setproctitle-1.3.5-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:7a887582bfdb6dcbc482db0ef9e630ad23ca95875806ef2b444bf6fbd7b7d7ca", size = 33415, upload-time = "2025-02-22T21:51:33.427Z" }, - { url = "https://files.pythonhosted.org/packages/11/5b/e497bf702ea5d553a331ca879e73a18bbd8f7d66d18d275cb2324e4144c4/setproctitle-1.3.5-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:755671c39a9e70834eeec6dc6b61e344399c49881d2e7ea3534a1c69669dd9cc", size = 30606, upload-time = "2025-02-22T21:51:34.729Z" }, - { url = "https://files.pythonhosted.org/packages/16/99/1bcb837134c71f332bfeaf923e68279566362b7d1504aa106af8046696e8/setproctitle-1.3.5-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9ab52b4c2ce056a1b60d439991a81ca90f019488d4b4f64b2779e6badd3677e6", size = 31679, upload-time = "2025-02-22T21:51:37.018Z" }, - { url = "https://files.pythonhosted.org/packages/77/55/72af3dbb0b1304bad54ea3b7cf1b524a8a2868da0b4c38bc18290f0097f7/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:36178b944019ec7fc52bb967ffeee296a11d373734a7be276755bedb3db5c141", size = 31388, upload-time = "2025-02-22T21:51:38.377Z" }, - { url = "https://files.pythonhosted.org/packages/f3/08/fa13f2da6bd10ca756a45f8fed2888f439e9ce7d6402258e87ceef2d4c71/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:269d41cd4f085b69821d1ee6599124f02dbbc79962b256e260b6c9021d037994", size = 30370, upload-time = "2025-02-22T21:51:39.879Z" }, - { url = "https://files.pythonhosted.org/packages/25/4b/83575bb403967f1069b68a8799979fe7979b5a7c17703d2984965d8f4e92/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d880630fd81d1b3bde121c352ca7ea2f2ff507ef40c3c011d0928ed491f912c9", size = 32897, upload-time = "2025-02-22T21:51:42.376Z" }, - { url = "https://files.pythonhosted.org/packages/1a/71/0c1e151ef6899260da4009e7170f56261486d3149e9bad40990b52bdd620/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8a7fed67ab49f60bd51f3b4cffff3f8d754d1bb0a40e42869911301ec6519b65", size = 30944, upload-time = "2025-02-22T21:51:43.698Z" }, - { url = "https://files.pythonhosted.org/packages/38/34/a3bdaeaee03e11aef82b45014738f1210f90e37359c41eda3e49b4ce891c/setproctitle-1.3.5-cp313-cp313-win32.whl", hash = "sha256:e9c0d0cfcf715631b10d5950d04a9978f63bc46535724ef7c2eaf1dca9988642", size = 11463, upload-time = "2025-02-22T21:51:44.869Z" }, - { url = "https://files.pythonhosted.org/packages/ef/f1/a19cde9f3f4054aed7c6077e7fc3420a5151ec6173cf3235fe000722ccb8/setproctitle-1.3.5-cp313-cp313-win_amd64.whl", hash = "sha256:e1d28eb98c91fbebd3e443a45c7da5d84974959851ef304c330eabd654a386f1", size = 12182, upload-time = "2025-02-22T21:51:46.033Z" }, - { url = "https://files.pythonhosted.org/packages/4a/ba/2524329ce958599069f0d0e4cfd3d6fbb7c58a4408b9e5609698e47353ec/setproctitle-1.3.5-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:dc66b84beb0d5eb03abf0c3140c6d2cbe3d67ae9f0824a09dfa8c6ff164319a6", size = 11418, upload-time = "2025-02-22T21:52:24.881Z" }, - { url = "https://files.pythonhosted.org/packages/a6/5f/a049640b05c609585ad0f471e667be0fd9ab533219127b455826d31587d5/setproctitle-1.3.5-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:31dc9b330e7cac7685bdef790747c07914081c11ee1066eb0c597303dfb52010", size = 13425, upload-time = "2025-02-22T21:52:26.833Z" }, - { url = "https://files.pythonhosted.org/packages/a9/15/caa47039e267ea67316b285e2e308ae529872ad6a143edf03a7d8edf6175/setproctitle-1.3.5-pp310-pypy310_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4028639b511f5e641d116b3b54ad70c637ebd1b4baac0948283daf11b104119f", size = 13026, upload-time = "2025-02-22T21:52:28.783Z" }, - { url = "https://files.pythonhosted.org/packages/c1/a2/1fb0647a251f4c788b94f751cf23171b2a905758fd13ef8d126222d41428/setproctitle-1.3.5-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:6bddef4e27d0ed74e44b58bf050bc3108591bf17d20d461fc59cd141282f849c", size = 12222, upload-time = "2025-02-22T21:52:31.088Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/3a/38/10d6bfe23df1bfc65ac2262ed10b45823f47f810b0057d3feeea1ca5c7ed/sentry_sdk-2.34.1.tar.gz", hash = "sha256:69274eb8c5c38562a544c3e9f68b5be0a43be4b697f5fd385bf98e4fbe672687", size = 336969, upload-time = "2025-07-30T11:13:37.93Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/2d/3e/bb34de65a5787f76848a533afbb6610e01fbcdd59e76d8679c254e02255c/sentry_sdk-2.34.1-py2.py3-none-any.whl", hash = "sha256:b7a072e1cdc5abc48101d5146e1ae680fa81fe886d8d95aaa25a0b450c818d32", size = 357743, upload-time = "2025-07-30T11:13:36.145Z" }, ] [[package]] name = "setuptools" -version = "77.0.3" +version = "80.9.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/81/ed/7101d53811fd359333583330ff976e5177c5e871ca8b909d1d6c30553aa3/setuptools-77.0.3.tar.gz", hash = "sha256:583b361c8da8de57403743e756609670de6fb2345920e36dc5c2d914c319c945", size = 1367236, upload-time = "2025-03-20T14:38:08.777Z" } +sdist = { url = "https://files.pythonhosted.org/packages/18/5d/3bf57dcd21979b887f014ea83c24ae194cfcd12b9e0fda66b957c69d1fca/setuptools-80.9.0.tar.gz", hash = "sha256:f36b47402ecde768dbfafc46e8e4207b4360c654f1f3bb84475f0a28628fb19c", size = 1319958, upload-time = "2025-05-27T00:56:51.443Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/a9/07/99f2cefae815c66eb23148f15d79ec055429c38fa8986edcc712ab5f3223/setuptools-77.0.3-py3-none-any.whl", hash = "sha256:67122e78221da5cf550ddd04cf8742c8fe12094483749a792d56cd669d6cf58c", size = 1255678, upload-time = "2025-03-20T14:38:06.621Z" }, + { url = "https://files.pythonhosted.org/packages/a3/dc/17031897dae0efacfea57dfd3a82fdd2a2aeb58e0ff71b77b87e44edc772/setuptools-80.9.0-py3-none-any.whl", hash = "sha256:062d34222ad13e0cc312a4c02d73f059e86a4acbfbdea8f8f76b28c99f306922", size = 1201486, upload-time = "2025-05-27T00:56:49.664Z" }, ] [[package]] @@ -2952,41 +3699,43 @@ wheels = [ [[package]] name = "statsmodels" -version = "0.14.4" +version = "0.14.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "packaging" }, { name = "pandas" }, { name = "patsy" }, - { name = "scipy" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/1f/3b/963a015dd8ea17e10c7b0e2f14d7c4daec903baf60a017e756b57953a4bf/statsmodels-0.14.4.tar.gz", hash = "sha256:5d69e0f39060dc72c067f9bb6e8033b6dccdb0bae101d76a7ef0bcc94e898b67", size = 20354802, upload-time = "2024-10-03T16:15:36.273Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/af/2c/23bf5ad9e8a77c0c8d9750512bff89e32154dea91998114118e0e147ae67/statsmodels-0.14.4-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:7a62f1fc9086e4b7ee789a6f66b3c0fc82dd8de1edda1522d30901a0aa45e42b", size = 10216574, upload-time = "2024-10-03T16:13:31.472Z" }, - { url = "https://files.pythonhosted.org/packages/ba/a5/2f09ab918296e534ea5d132e90efac51ae12ff15992d77539bbfca1158fa/statsmodels-0.14.4-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:46ac7ddefac0c9b7b607eed1d47d11e26fe92a1bc1f4d9af48aeed4e21e87981", size = 9912430, upload-time = "2024-10-03T16:13:44.683Z" }, - { url = "https://files.pythonhosted.org/packages/93/6a/b86f8c9b799dc93e5b4a3267eb809843e6328e34248a53496b96f50d732e/statsmodels-0.14.4-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2a337b731aa365d09bb0eab6da81446c04fde6c31976b1d8e3d3a911f0f1e07b", size = 10444673, upload-time = "2024-10-03T17:09:04.647Z" }, - { url = "https://files.pythonhosted.org/packages/78/44/d72c634211797ed07dd8c63ced4ae11debd7a40b24ee80e79346a526194f/statsmodels-0.14.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:631bb52159117c5da42ba94bd94859276b68cab25dc4cac86475bc24671143bc", size = 10811248, upload-time = "2024-10-03T17:09:20.337Z" }, - { url = "https://files.pythonhosted.org/packages/35/64/df81426924fcc48a0402534efa96cde13275629ae52f123189d16c4b75ff/statsmodels-0.14.4-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:3bb2e580d382545a65f298589809af29daeb15f9da2eb252af8f79693e618abc", size = 10946447, upload-time = "2024-10-03T17:09:35.135Z" }, - { url = "https://files.pythonhosted.org/packages/5c/f9/205130cceeda0eebd5a1a58c04e060c2f87a1d63cbbe37a9caa0fcb50c68/statsmodels-0.14.4-cp310-cp310-win_amd64.whl", hash = "sha256:9729642884147ee9db67b5a06a355890663d21f76ed608a56ac2ad98b94d201a", size = 9845796, upload-time = "2024-10-03T16:13:58.307Z" }, - { url = "https://files.pythonhosted.org/packages/48/88/326f5f689e69d9c47a68a22ffdd20a6ea6410b53918f9a8e63380dfc181c/statsmodels-0.14.4-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:5ed7e118e6e3e02d6723a079b8c97eaadeed943fa1f7f619f7148dfc7862670f", size = 10221032, upload-time = "2024-10-03T16:22:48.191Z" }, - { url = "https://files.pythonhosted.org/packages/07/0b/9a0818be42f6689ebdc7a2277ea984d6299f0809d0e0277128df4f7dc606/statsmodels-0.14.4-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:f5f537f7d000de4a1708c63400755152b862cd4926bb81a86568e347c19c364b", size = 9912219, upload-time = "2024-10-03T17:17:03.799Z" }, - { url = "https://files.pythonhosted.org/packages/b1/f2/91c70a3b4a3e416f76ead61b04c87bc60080d634d7fa2ab893976bdd86fa/statsmodels-0.14.4-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa74aaa26eaa5012b0a01deeaa8a777595d0835d3d6c7175f2ac65435a7324d2", size = 10424053, upload-time = "2024-10-03T17:09:49.325Z" }, - { url = "https://files.pythonhosted.org/packages/9d/4f/a96e682f82b675e4a6f3de8ad990587d8b1fde500a630a2aabcaabee11d8/statsmodels-0.14.4-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e332c2d9b806083d1797231280602340c5c913f90d4caa0213a6a54679ce9331", size = 10752529, upload-time = "2024-10-03T17:10:03.489Z" }, - { url = "https://files.pythonhosted.org/packages/4b/c6/47549345d32da1530a819a3699f6f34f9f70733a245eeb29f5e05e53f362/statsmodels-0.14.4-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d9c8fa28dfd75753d9cf62769ba1fecd7e73a0be187f35cc6f54076f98aa3f3f", size = 10959003, upload-time = "2024-10-03T17:10:17.477Z" }, - { url = "https://files.pythonhosted.org/packages/4b/e4/f9e96896278308e17dfd4f60a84826c48117674c980234ee38f59ab28a12/statsmodels-0.14.4-cp311-cp311-win_amd64.whl", hash = "sha256:a6087ecb0714f7c59eb24c22781491e6f1cfffb660b4740e167625ca4f052056", size = 9853281, upload-time = "2024-10-03T16:14:11.019Z" }, - { url = "https://files.pythonhosted.org/packages/f5/99/654fd41a9024643ee70b239e5ebc987bf98ce9fc2693bd550bee58136564/statsmodels-0.14.4-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:5221dba7424cf4f2561b22e9081de85f5bb871228581124a0d1b572708545199", size = 10220508, upload-time = "2024-10-03T17:10:31.183Z" }, - { url = "https://files.pythonhosted.org/packages/67/d8/ac30cf4cf97adaa48548be57e7cf02e894f31b45fd55bf9213358d9781c9/statsmodels-0.14.4-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:17672b30c6b98afe2b095591e32d1d66d4372f2651428e433f16a3667f19eabb", size = 9912317, upload-time = "2024-10-03T16:22:29.504Z" }, - { url = "https://files.pythonhosted.org/packages/e0/77/2440d551eaf27f9c1d3650e13b3821a35ad5b21d3a19f62fb302af9203e8/statsmodels-0.14.4-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ab5e6312213b8cfb9dca93dd46a0f4dccb856541f91d3306227c3d92f7659245", size = 10301662, upload-time = "2024-10-03T17:13:04.537Z" }, - { url = "https://files.pythonhosted.org/packages/fa/e1/60a652f18996a40a7410aeb7eb476c18da8a39792c7effe67f06883e9852/statsmodels-0.14.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4bbb150620b53133d6cd1c5d14c28a4f85701e6c781d9b689b53681effaa655f", size = 10741763, upload-time = "2024-10-03T17:13:17.594Z" }, - { url = "https://files.pythonhosted.org/packages/81/0c/2453eec3ac25e300847d9ed97f41156de145e507391ecb5ac989e111e525/statsmodels-0.14.4-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:bb695c2025d122a101c2aca66d2b78813c321b60d3a7c86bb8ec4467bb53b0f9", size = 10879534, upload-time = "2024-10-03T17:13:31.19Z" }, - { url = "https://files.pythonhosted.org/packages/59/9a/e466a1b887a1441141e52dbcc98152f013d85076576da6eed2357f2016ae/statsmodels-0.14.4-cp312-cp312-win_amd64.whl", hash = "sha256:7f7917a51766b4e074da283c507a25048ad29a18e527207883d73535e0dc6184", size = 9823866, upload-time = "2024-10-03T16:14:23.828Z" }, - { url = "https://files.pythonhosted.org/packages/31/f8/2662e6a101315ad336f75168fa9bac71f913ebcb92a6be84031d84a0f21f/statsmodels-0.14.4-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b5a24f5d2c22852d807d2b42daf3a61740820b28d8381daaf59dcb7055bf1a79", size = 10186886, upload-time = "2024-10-03T17:10:44.074Z" }, - { url = "https://files.pythonhosted.org/packages/fa/c0/ee6e8ed35fc1ca9c7538c592f4974547bf72274bc98db1ae4a6e87481a83/statsmodels-0.14.4-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:df4f7864606fa843d7e7c0e6af288f034a2160dba14e6ccc09020a3cf67cb092", size = 9880066, upload-time = "2024-10-03T17:10:56.972Z" }, - { url = "https://files.pythonhosted.org/packages/d1/97/3380ca6d8fd66cfb3d12941e472642f26e781a311c355a4e97aab2ed0216/statsmodels-0.14.4-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:91341cbde9e8bea5fb419a76e09114e221567d03f34ca26e6d67ae2c27d8fe3c", size = 10283521, upload-time = "2024-10-03T17:14:06.216Z" }, - { url = "https://files.pythonhosted.org/packages/fe/2a/55c5b5c5e5124a202ea3fe0bcdbdeceaf91b4ec6164b8434acb9dd97409c/statsmodels-0.14.4-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1322286a7bfdde2790bf72d29698a1b76c20b8423a55bdcd0d457969d0041f72", size = 10723228, upload-time = "2024-10-03T17:14:19.587Z" }, - { url = "https://files.pythonhosted.org/packages/4f/76/67747e49dc758daae06f33aad8247b718cd7d224f091d2cd552681215bb2/statsmodels-0.14.4-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e31b95ac603415887c9f0d344cb523889cf779bc52d68e27e2d23c358958fec7", size = 10859503, upload-time = "2024-10-03T17:14:32.798Z" }, - { url = "https://files.pythonhosted.org/packages/1d/eb/cb8b01f5edf8f135eb3d0553d159db113a35b2948d0e51eeb735e7ae09ea/statsmodels-0.14.4-cp313-cp313-win_amd64.whl", hash = "sha256:81030108d27aecc7995cac05aa280cf8c6025f6a6119894eef648997936c2dd0", size = 9817574, upload-time = "2024-10-03T16:14:37.461Z" }, + { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/64/cc/8c1bf59bf8203dea1bf2ea811cfe667d7bcc6909c83d8afb02b08e30f50b/statsmodels-0.14.5.tar.gz", hash = "sha256:de260e58cccfd2ceddf835b55a357233d6ca853a1aa4f90f7553a52cc71c6ddf", size = 20525016, upload-time = "2025-07-07T12:14:23.195Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/3a/2c/55b2a5d10c1a211ecab3f792021d2581bbe1c5ca0a1059f6715dddc6899d/statsmodels-0.14.5-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:9fc2b5cdc0c95cba894849651fec1fa1511d365e3eb72b0cc75caac44077cd48", size = 10058241, upload-time = "2025-07-07T12:13:16.286Z" }, + { url = "https://files.pythonhosted.org/packages/66/d9/6967475805de06691e951072d05e40e3f1c71b6221bb92401193ee19bd2a/statsmodels-0.14.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:b8d96b0bbaeabd3a557c35cc7249baa9cfbc6dd305c32a9f2cbdd7f46c037e7f", size = 9734017, upload-time = "2025-07-07T12:05:08.498Z" }, + { url = "https://files.pythonhosted.org/packages/df/a8/803c280419a7312e2472969fe72cf461c1210a27770a662cbe3b5cd7c6fe/statsmodels-0.14.5-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:145bc39b2cb201efb6c83cc3f2163c269e63b0d4809801853dec6f440bd3bc37", size = 10459677, upload-time = "2025-07-07T14:21:51.809Z" }, + { url = "https://files.pythonhosted.org/packages/a1/25/edf20acbd670934b02cd9344e29c9a03ce040122324b3491bb075ae76b2d/statsmodels-0.14.5-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d7c14fb2617bb819fb2532e1424e1da2b98a3419a80e95f33365a72d437d474e", size = 10678631, upload-time = "2025-07-07T14:22:05.496Z" }, + { url = "https://files.pythonhosted.org/packages/64/22/8b1e38310272e766abd6093607000a81827420a3348f09eff08a9e54cbaf/statsmodels-0.14.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:1e9742d8a5ac38a3bfc4b7f4b0681903920f20cbbf466d72b1fd642033846108", size = 10699273, upload-time = "2025-07-07T14:22:19.487Z" }, + { url = "https://files.pythonhosted.org/packages/d1/6f/6de51f1077b7cef34611f1d6721392ea170153251b4d977efcf6d100f779/statsmodels-0.14.5-cp310-cp310-win_amd64.whl", hash = "sha256:1cab9e6fce97caf4239cdb2df375806937da5d0b7ba2699b13af33a07f438464", size = 9644785, upload-time = "2025-07-07T12:05:20.927Z" }, + { url = "https://files.pythonhosted.org/packages/14/30/fd49902b30416b828de763e161c0d6e2cc04d119ae4fbdd3f3b43dc8f1be/statsmodels-0.14.5-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:4b7091a8442076c708c926de3603653a160955e80a2b6d931475b7bb8ddc02e5", size = 10053330, upload-time = "2025-07-07T12:07:39.689Z" }, + { url = "https://files.pythonhosted.org/packages/ca/c1/2654541ff6f5790d01d1e5ba36405fde873f4a854f473e90b4fe56b37333/statsmodels-0.14.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:128872be8f3208f4446d91ea9e4261823902fc7997fee7e1a983eb62fd3b7c6e", size = 9735555, upload-time = "2025-07-07T12:13:28.935Z" }, + { url = "https://files.pythonhosted.org/packages/ce/da/6ebb64d0db4e86c0d2d9cde89e03247702da0ab191789f7813d4f9a348da/statsmodels-0.14.5-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f2ad5aee04ae7196c429df2174df232c057e478c5fa63193d01c8ec9aae04d31", size = 10307522, upload-time = "2025-07-07T14:22:32.853Z" }, + { url = "https://files.pythonhosted.org/packages/67/49/ac803ca093ec3845184a752a91cd84511245e1f97103b15cfe32794a3bb0/statsmodels-0.14.5-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f402fc793458dd6d96e099acb44cd1de1428565bf7ef3030878a8daff091f08a", size = 10474665, upload-time = "2025-07-07T14:22:46.011Z" }, + { url = "https://files.pythonhosted.org/packages/f0/c8/ae82feb00582f4814fac5d2cb3ec32f93866b413cf5878b2fe93688ec63c/statsmodels-0.14.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:26c028832730aebfbfd4e7501694e1f9ad31ec8536e776716673f4e7afd4059a", size = 10713120, upload-time = "2025-07-07T14:23:00.067Z" }, + { url = "https://files.pythonhosted.org/packages/05/ac/4276459ea71aa46e2967ea283fc88ee5631c11f29a06787e16cf4aece1b8/statsmodels-0.14.5-cp311-cp311-win_amd64.whl", hash = "sha256:ec56f771d9529cdc17ed2fb2a950d100b6e83a7c5372aae8ac5bb065c474b856", size = 9640980, upload-time = "2025-07-07T12:05:33.085Z" }, + { url = "https://files.pythonhosted.org/packages/5f/a5/fcc4f5f16355660ce7a1742e28a43e3a9391b492fc4ff29fdd6893e81c05/statsmodels-0.14.5-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:37e7364a39f9aa3b51d15a208c2868b90aadb8412f868530f5cba9197cb00eaa", size = 10042891, upload-time = "2025-07-07T12:13:41.671Z" }, + { url = "https://files.pythonhosted.org/packages/1c/6f/db0cf5efa48277ac6218d9b981c8fd5e63c4c43e0d9d65015fdc38eed0ef/statsmodels-0.14.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:4263d7f4d0f1d5ac6eb4db22e1ee34264a14d634b9332c975c9d9109b6b46e12", size = 9698912, upload-time = "2025-07-07T12:07:54.674Z" }, + { url = "https://files.pythonhosted.org/packages/4a/93/4ddc3bc4a59c51e6a57c49df1b889882c40d9e141e855b3517f6a8de3232/statsmodels-0.14.5-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:86224f6e36f38486e471e75759d241fe2912d8bc25ab157d54ee074c6aedbf45", size = 10237801, upload-time = "2025-07-07T14:23:12.593Z" }, + { url = "https://files.pythonhosted.org/packages/66/de/dc6bf2f6e8c8eb4c5815560ebdbdf2d69a767bc0f65fde34bc086cf5b36d/statsmodels-0.14.5-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c3dd760a6fa80cd5e0371685c697bb9c2c0e6e1f394d975e596a1e6d0bbb9372", size = 10424154, upload-time = "2025-07-07T14:23:25.365Z" }, + { url = "https://files.pythonhosted.org/packages/16/4f/2d5a8d14bebdf2b03b3ea89b8c6a2c837bb406ba5b7a41add8bd303bce29/statsmodels-0.14.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:6264fb00e02f858b86bd01ef2dc05055a71d4a0cc7551b9976b07b0f0e6cf24f", size = 10652915, upload-time = "2025-07-07T14:23:39.337Z" }, + { url = "https://files.pythonhosted.org/packages/df/4c/2feda3a9f0e17444a84ba5398ada6a4d2e1b8f832760048f04e2b8ea0c41/statsmodels-0.14.5-cp312-cp312-win_amd64.whl", hash = "sha256:b2ed065bfbaf8bb214c7201656df840457c2c8c65e1689e3eb09dc7440f9c61c", size = 9611236, upload-time = "2025-07-07T12:08:06.794Z" }, + { url = "https://files.pythonhosted.org/packages/84/fd/4c374108cf108b3130240a5b45847a61f70ddf973429044a81a05189b046/statsmodels-0.14.5-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:906263134dd1a640e55ecb01fda4a9be7b9e08558dba9e4c4943a486fdb0c9c8", size = 10013958, upload-time = "2025-07-07T14:35:01.04Z" }, + { url = "https://files.pythonhosted.org/packages/5a/36/bf3d7f0e36acd3ba9ec0babd79ace25506b6872780cbd710fb7cd31f0fa2/statsmodels-0.14.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:9118f76344f77cffbb3a9cbcff8682b325be5eed54a4b3253e09da77a74263d3", size = 9674243, upload-time = "2025-07-07T12:08:22.571Z" }, + { url = "https://files.pythonhosted.org/packages/90/ce/a55a6f37b5277683ceccd965a5828b24672bbc427db6b3969ae0b0fc29fb/statsmodels-0.14.5-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:9dc4ee159070557c9a6c000625d85f653de437772fe7086857cff68f501afe45", size = 10219521, upload-time = "2025-07-07T14:23:52.646Z" }, + { url = "https://files.pythonhosted.org/packages/1e/48/973da1ee8bc0743519759e74c3615b39acdc3faf00e0a0710f8c856d8c9d/statsmodels-0.14.5-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5a085d47c8ef5387279a991633883d0e700de2b0acc812d7032d165888627bef", size = 10453538, upload-time = "2025-07-07T14:24:06.959Z" }, + { url = "https://files.pythonhosted.org/packages/c7/d6/18903fb707afd31cf1edaec5201964dbdacb2bfae9a22558274647a7c88f/statsmodels-0.14.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9f866b2ebb2904b47c342d00def83c526ef2eb1df6a9a3c94ba5fe63d0005aec", size = 10681584, upload-time = "2025-07-07T14:24:21.038Z" }, + { url = "https://files.pythonhosted.org/packages/44/d6/80df1bbbfcdc50bff4152f43274420fa9856d56e234d160d6206eb1f5827/statsmodels-0.14.5-cp313-cp313-win_amd64.whl", hash = "sha256:2a06bca03b7a492f88c8106103ab75f1a5ced25de90103a89f3a287518017939", size = 9604641, upload-time = "2025-07-07T12:08:36.23Z" }, ] [[package]] @@ -3051,13 +3800,14 @@ wheels = [ [[package]] name = "torch" -version = "2.7.0" +version = "2.7.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "filelock" }, { name = "fsspec" }, { name = "jinja2" }, - { name = "networkx" }, + { name = "networkx", version = "3.4.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "networkx", version = "3.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "nvidia-cublas-cu12", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" }, { name = "nvidia-cuda-cupti-cu12", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" }, { name = "nvidia-cuda-nvrtc-cu12", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" }, @@ -3078,26 +3828,26 @@ dependencies = [ { name = "typing-extensions" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/46/c2/3fb87940fa160d956ee94d644d37b99a24b9c05a4222bf34f94c71880e28/torch-2.7.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:c9afea41b11e1a1ab1b258a5c31afbd646d6319042bfe4f231b408034b51128b", size = 99158447, upload-time = "2025-04-23T14:35:10.557Z" }, - { url = "https://files.pythonhosted.org/packages/cc/2c/91d1de65573fce563f5284e69d9c56b57289625cffbbb6d533d5d56c36a5/torch-2.7.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:0b9960183b6e5b71239a3e6c883d8852c304e691c0b2955f7045e8a6d05b9183", size = 865164221, upload-time = "2025-04-23T14:33:27.864Z" }, - { url = "https://files.pythonhosted.org/packages/7f/7e/1b1cc4e0e7cc2666cceb3d250eef47a205f0821c330392cf45eb08156ce5/torch-2.7.0-cp310-cp310-win_amd64.whl", hash = "sha256:2ad79d0d8c2a20a37c5df6052ec67c2078a2c4e9a96dd3a8b55daaff6d28ea29", size = 212521189, upload-time = "2025-04-23T14:34:53.898Z" }, - { url = "https://files.pythonhosted.org/packages/dc/0b/b2b83f30b8e84a51bf4f96aa3f5f65fdf7c31c591cc519310942339977e2/torch-2.7.0-cp310-none-macosx_11_0_arm64.whl", hash = "sha256:34e0168ed6de99121612d72224e59b2a58a83dae64999990eada7260c5dd582d", size = 68559462, upload-time = "2025-04-23T14:35:39.889Z" }, - { url = "https://files.pythonhosted.org/packages/40/da/7378d16cc636697f2a94f791cb496939b60fb8580ddbbef22367db2c2274/torch-2.7.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:2b7813e904757b125faf1a9a3154e1d50381d539ced34da1992f52440567c156", size = 99159397, upload-time = "2025-04-23T14:35:35.304Z" }, - { url = "https://files.pythonhosted.org/packages/0e/6b/87fcddd34df9f53880fa1f0c23af7b6b96c935856473faf3914323588c40/torch-2.7.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:fd5cfbb4c3bbadd57ad1b27d56a28008f8d8753733411a140fcfb84d7f933a25", size = 865183681, upload-time = "2025-04-23T14:34:21.802Z" }, - { url = "https://files.pythonhosted.org/packages/13/85/6c1092d4b06c3db1ed23d4106488750917156af0b24ab0a2d9951830b0e9/torch-2.7.0-cp311-cp311-win_amd64.whl", hash = "sha256:58df8d5c2eeb81305760282b5069ea4442791a6bbf0c74d9069b7b3304ff8a37", size = 212520100, upload-time = "2025-04-23T14:35:27.473Z" }, - { url = "https://files.pythonhosted.org/packages/aa/3f/85b56f7e2abcfa558c5fbf7b11eb02d78a4a63e6aeee2bbae3bb552abea5/torch-2.7.0-cp311-none-macosx_11_0_arm64.whl", hash = "sha256:0a8d43caa342b9986101ec5feb5bbf1d86570b5caa01e9cb426378311258fdde", size = 68569377, upload-time = "2025-04-23T14:35:20.361Z" }, - { url = "https://files.pythonhosted.org/packages/aa/5e/ac759f4c0ab7c01feffa777bd68b43d2ac61560a9770eeac074b450f81d4/torch-2.7.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:36a6368c7ace41ad1c0f69f18056020b6a5ca47bedaca9a2f3b578f5a104c26c", size = 99013250, upload-time = "2025-04-23T14:35:15.589Z" }, - { url = "https://files.pythonhosted.org/packages/9c/58/2d245b6f1ef61cf11dfc4aceeaacbb40fea706ccebac3f863890c720ab73/torch-2.7.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:15aab3e31c16feb12ae0a88dba3434a458874636f360c567caa6a91f6bfba481", size = 865042157, upload-time = "2025-04-23T14:32:56.011Z" }, - { url = "https://files.pythonhosted.org/packages/44/80/b353c024e6b624cd9ce1d66dcb9d24e0294680f95b369f19280e241a0159/torch-2.7.0-cp312-cp312-win_amd64.whl", hash = "sha256:f56d4b2510934e072bab3ab8987e00e60e1262fb238176168f5e0c43a1320c6d", size = 212482262, upload-time = "2025-04-23T14:35:03.527Z" }, - { url = "https://files.pythonhosted.org/packages/ee/8d/b2939e5254be932db1a34b2bd099070c509e8887e0c5a90c498a917e4032/torch-2.7.0-cp312-none-macosx_11_0_arm64.whl", hash = "sha256:30b7688a87239a7de83f269333651d8e582afffce6f591fff08c046f7787296e", size = 68574294, upload-time = "2025-04-23T14:34:47.098Z" }, - { url = "https://files.pythonhosted.org/packages/14/24/720ea9a66c29151b315ea6ba6f404650834af57a26b2a04af23ec246b2d5/torch-2.7.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:868ccdc11798535b5727509480cd1d86d74220cfdc42842c4617338c1109a205", size = 99015553, upload-time = "2025-04-23T14:34:41.075Z" }, - { url = "https://files.pythonhosted.org/packages/4b/27/285a8cf12bd7cd71f9f211a968516b07dcffed3ef0be585c6e823675ab91/torch-2.7.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:9b52347118116cf3dff2ab5a3c3dd97c719eb924ac658ca2a7335652076df708", size = 865046389, upload-time = "2025-04-23T14:32:01.16Z" }, - { url = "https://files.pythonhosted.org/packages/74/c8/2ab2b6eadc45554af8768ae99668c5a8a8552e2012c7238ded7e9e4395e1/torch-2.7.0-cp313-cp313-win_amd64.whl", hash = "sha256:434cf3b378340efc87c758f250e884f34460624c0523fe5c9b518d205c91dd1b", size = 212490304, upload-time = "2025-04-23T14:33:57.108Z" }, - { url = "https://files.pythonhosted.org/packages/28/fd/74ba6fde80e2b9eef4237fe668ffae302c76f0e4221759949a632ca13afa/torch-2.7.0-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:edad98dddd82220465b106506bb91ee5ce32bd075cddbcf2b443dfaa2cbd83bf", size = 68856166, upload-time = "2025-04-23T14:34:04.012Z" }, - { url = "https://files.pythonhosted.org/packages/cb/b4/8df3f9fe6bdf59e56a0e538592c308d18638eb5f5dc4b08d02abb173c9f0/torch-2.7.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:2a885fc25afefb6e6eb18a7d1e8bfa01cc153e92271d980a49243b250d5ab6d9", size = 99091348, upload-time = "2025-04-23T14:33:48.975Z" }, - { url = "https://files.pythonhosted.org/packages/9d/f5/0bd30e9da04c3036614aa1b935a9f7e505a9e4f1f731b15e165faf8a4c74/torch-2.7.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:176300ff5bc11a5f5b0784e40bde9e10a35c4ae9609beed96b4aeb46a27f5fae", size = 865104023, upload-time = "2025-04-23T14:30:40.537Z" }, - { url = "https://files.pythonhosted.org/packages/d1/b7/2235d0c3012c596df1c8d39a3f4afc1ee1b6e318d469eda4c8bb68566448/torch-2.7.0-cp313-cp313t-win_amd64.whl", hash = "sha256:d0ca446a93f474985d81dc866fcc8dccefb9460a29a456f79d99c29a78a66993", size = 212750916, upload-time = "2025-04-23T14:32:22.91Z" }, - { url = "https://files.pythonhosted.org/packages/90/48/7e6477cf40d48cc0a61fa0d41ee9582b9a316b12772fcac17bc1a40178e7/torch-2.7.0-cp313-none-macosx_11_0_arm64.whl", hash = "sha256:27f5007bdf45f7bb7af7f11d1828d5c2487e030690afb3d89a651fd7036a390e", size = 68575074, upload-time = "2025-04-23T14:32:38.136Z" }, + { url = "https://files.pythonhosted.org/packages/6a/27/2e06cb52adf89fe6e020963529d17ed51532fc73c1e6d1b18420ef03338c/torch-2.7.1-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:a103b5d782af5bd119b81dbcc7ffc6fa09904c423ff8db397a1e6ea8fd71508f", size = 99089441, upload-time = "2025-06-04T17:38:48.268Z" }, + { url = "https://files.pythonhosted.org/packages/0a/7c/0a5b3aee977596459ec45be2220370fde8e017f651fecc40522fd478cb1e/torch-2.7.1-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:fe955951bdf32d182ee8ead6c3186ad54781492bf03d547d31771a01b3d6fb7d", size = 821154516, upload-time = "2025-06-04T17:36:28.556Z" }, + { url = "https://files.pythonhosted.org/packages/f9/91/3d709cfc5e15995fb3fe7a6b564ce42280d3a55676dad672205e94f34ac9/torch-2.7.1-cp310-cp310-win_amd64.whl", hash = "sha256:885453d6fba67d9991132143bf7fa06b79b24352f4506fd4d10b309f53454162", size = 216093147, upload-time = "2025-06-04T17:39:38.132Z" }, + { url = "https://files.pythonhosted.org/packages/92/f6/5da3918414e07da9866ecb9330fe6ffdebe15cb9a4c5ada7d4b6e0a6654d/torch-2.7.1-cp310-none-macosx_11_0_arm64.whl", hash = "sha256:d72acfdb86cee2a32c0ce0101606f3758f0d8bb5f8f31e7920dc2809e963aa7c", size = 68630914, upload-time = "2025-06-04T17:39:31.162Z" }, + { url = "https://files.pythonhosted.org/packages/11/56/2eae3494e3d375533034a8e8cf0ba163363e996d85f0629441fa9d9843fe/torch-2.7.1-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:236f501f2e383f1cb861337bdf057712182f910f10aeaf509065d54d339e49b2", size = 99093039, upload-time = "2025-06-04T17:39:06.963Z" }, + { url = "https://files.pythonhosted.org/packages/e5/94/34b80bd172d0072c9979708ccd279c2da2f55c3ef318eceec276ab9544a4/torch-2.7.1-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:06eea61f859436622e78dd0cdd51dbc8f8c6d76917a9cf0555a333f9eac31ec1", size = 821174704, upload-time = "2025-06-04T17:37:03.799Z" }, + { url = "https://files.pythonhosted.org/packages/50/9e/acf04ff375b0b49a45511c55d188bcea5c942da2aaf293096676110086d1/torch-2.7.1-cp311-cp311-win_amd64.whl", hash = "sha256:8273145a2e0a3c6f9fd2ac36762d6ee89c26d430e612b95a99885df083b04e52", size = 216095937, upload-time = "2025-06-04T17:39:24.83Z" }, + { url = "https://files.pythonhosted.org/packages/5b/2b/d36d57c66ff031f93b4fa432e86802f84991477e522adcdffd314454326b/torch-2.7.1-cp311-none-macosx_11_0_arm64.whl", hash = "sha256:aea4fc1bf433d12843eb2c6b2204861f43d8364597697074c8d38ae2507f8730", size = 68640034, upload-time = "2025-06-04T17:39:17.989Z" }, + { url = "https://files.pythonhosted.org/packages/87/93/fb505a5022a2e908d81fe9a5e0aa84c86c0d5f408173be71c6018836f34e/torch-2.7.1-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:27ea1e518df4c9de73af7e8a720770f3628e7f667280bce2be7a16292697e3fa", size = 98948276, upload-time = "2025-06-04T17:39:12.852Z" }, + { url = "https://files.pythonhosted.org/packages/56/7e/67c3fe2b8c33f40af06326a3d6ae7776b3e3a01daa8f71d125d78594d874/torch-2.7.1-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:c33360cfc2edd976c2633b3b66c769bdcbbf0e0b6550606d188431c81e7dd1fc", size = 821025792, upload-time = "2025-06-04T17:34:58.747Z" }, + { url = "https://files.pythonhosted.org/packages/a1/37/a37495502bc7a23bf34f89584fa5a78e25bae7b8da513bc1b8f97afb7009/torch-2.7.1-cp312-cp312-win_amd64.whl", hash = "sha256:d8bf6e1856ddd1807e79dc57e54d3335f2b62e6f316ed13ed3ecfe1fc1df3d8b", size = 216050349, upload-time = "2025-06-04T17:38:59.709Z" }, + { url = "https://files.pythonhosted.org/packages/3a/60/04b77281c730bb13460628e518c52721257814ac6c298acd25757f6a175c/torch-2.7.1-cp312-none-macosx_11_0_arm64.whl", hash = "sha256:787687087412c4bd68d315e39bc1223f08aae1d16a9e9771d95eabbb04ae98fb", size = 68645146, upload-time = "2025-06-04T17:38:52.97Z" }, + { url = "https://files.pythonhosted.org/packages/66/81/e48c9edb655ee8eb8c2a6026abdb6f8d2146abd1f150979ede807bb75dcb/torch-2.7.1-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:03563603d931e70722dce0e11999d53aa80a375a3d78e6b39b9f6805ea0a8d28", size = 98946649, upload-time = "2025-06-04T17:38:43.031Z" }, + { url = "https://files.pythonhosted.org/packages/3a/24/efe2f520d75274fc06b695c616415a1e8a1021d87a13c68ff9dce733d088/torch-2.7.1-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:d632f5417b6980f61404a125b999ca6ebd0b8b4bbdbb5fbbba44374ab619a412", size = 821033192, upload-time = "2025-06-04T17:38:09.146Z" }, + { url = "https://files.pythonhosted.org/packages/dd/d9/9c24d230333ff4e9b6807274f6f8d52a864210b52ec794c5def7925f4495/torch-2.7.1-cp313-cp313-win_amd64.whl", hash = "sha256:23660443e13995ee93e3d844786701ea4ca69f337027b05182f5ba053ce43b38", size = 216055668, upload-time = "2025-06-04T17:38:36.253Z" }, + { url = "https://files.pythonhosted.org/packages/95/bf/e086ee36ddcef9299f6e708d3b6c8487c1651787bb9ee2939eb2a7f74911/torch-2.7.1-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:0da4f4dba9f65d0d203794e619fe7ca3247a55ffdcbd17ae8fb83c8b2dc9b585", size = 68925988, upload-time = "2025-06-04T17:38:29.273Z" }, + { url = "https://files.pythonhosted.org/packages/69/6a/67090dcfe1cf9048448b31555af6efb149f7afa0a310a366adbdada32105/torch-2.7.1-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:e08d7e6f21a617fe38eeb46dd2213ded43f27c072e9165dc27300c9ef9570934", size = 99028857, upload-time = "2025-06-04T17:37:50.956Z" }, + { url = "https://files.pythonhosted.org/packages/90/1c/48b988870823d1cc381f15ec4e70ed3d65e043f43f919329b0045ae83529/torch-2.7.1-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:30207f672328a42df4f2174b8f426f354b2baa0b7cca3a0adb3d6ab5daf00dc8", size = 821098066, upload-time = "2025-06-04T17:37:33.939Z" }, + { url = "https://files.pythonhosted.org/packages/7b/eb/10050d61c9d5140c5dc04a89ed3257ef1a6b93e49dd91b95363d757071e0/torch-2.7.1-cp313-cp313t-win_amd64.whl", hash = "sha256:79042feca1c634aaf6603fe6feea8c6b30dfa140a6bbc0b973e2260c7e79a22e", size = 216336310, upload-time = "2025-06-04T17:36:09.862Z" }, + { url = "https://files.pythonhosted.org/packages/b1/29/beb45cdf5c4fc3ebe282bf5eafc8dfd925ead7299b3c97491900fe5ed844/torch-2.7.1-cp313-none-macosx_11_0_arm64.whl", hash = "sha256:988b0cbc4333618a1056d2ebad9eb10089637b659eb645434d0809d8d937b946", size = 68645708, upload-time = "2025-06-04T17:34:39.852Z" }, ] [[package]] @@ -3108,7 +3858,8 @@ dependencies = [ { name = "aiohttp" }, { name = "fsspec" }, { name = "jinja2" }, - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "psutil" }, { name = "pyparsing" }, { name = "requests" }, @@ -3121,67 +3872,70 @@ wheels = [ [[package]] name = "torchmetrics" -version = "1.7.0" +version = "1.8.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "lightning-utilities" }, - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "packaging" }, { name = "torch" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/47/99/3a28bc2be4f562b9aae5b3557ec4c6cbcfcdd0a7d2a6fcb1ab44e7c55334/torchmetrics-1.7.0.tar.gz", hash = "sha256:7a26d5cb73a6ae51ab5cb514aa4dc0543af7287a507719986a06e15df12ea68b", size = 564173, upload-time = "2025-03-20T19:12:01.289Z" } +sdist = { url = "https://files.pythonhosted.org/packages/b0/1d/01fdf2595ac344dcfb2d837e7596c71cd8659e99f0b1fd72a93c76ea2c05/torchmetrics-1.8.0.tar.gz", hash = "sha256:8b4d011963a602109fb8255018c2386391e8c4c7f48a09669fbf7bb7889fda8c", size = 578941, upload-time = "2025-07-23T17:39:11.719Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/71/1d/dbbb54743865bad58a9adc7ac2c9bedd28bc76253cb8ec4b69765ccc8190/torchmetrics-1.7.0-py3-none-any.whl", hash = "sha256:39a72cf40c8452e041f5315b997ef811c2baaae01478131cf6ed9813f553a668", size = 960916, upload-time = "2025-03-20T19:11:58.841Z" }, + { url = "https://files.pythonhosted.org/packages/0b/cc/f41157b446d555bf1446915960f92928dc5e8393fc09b6de6189aef2d9dd/torchmetrics-1.8.0-py3-none-any.whl", hash = "sha256:009832f0df5be9aca72f51a1e6c6a555a131ba53c1ce46a38348a202250c22df", size = 981886, upload-time = "2025-07-23T17:39:09.246Z" }, ] [[package]] name = "torchvision" -version = "0.22.0" +version = "0.22.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy" }, + { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "pillow" }, { name = "torch" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/eb/03/a514766f068b088180f273913e539d08e830be3ae46ef8577ea62584a27c/torchvision-0.22.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:72256f1d7ff510b16c9fb4dd488584d0693f40c792f286a9620674438a81ccca", size = 1947829, upload-time = "2025-04-23T14:42:04.652Z" }, - { url = "https://files.pythonhosted.org/packages/a3/e5/ec4b52041cd8c440521b75864376605756bd2d112d6351ea6a1ab25008c1/torchvision-0.22.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:810ea4af3bc63cf39e834f91f4218ff5999271caaffe2456247df905002bd6c0", size = 2512604, upload-time = "2025-04-23T14:41:56.515Z" }, - { url = "https://files.pythonhosted.org/packages/e7/9e/e898a377e674da47e95227f3d7be2c49550ce381eebd8c7831c1f8bb7d39/torchvision-0.22.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:6fbca169c690fa2b9b8c39c0ad76d5b8992296d0d03df01e11df97ce12b4e0ac", size = 7446399, upload-time = "2025-04-23T14:41:49.793Z" }, - { url = "https://files.pythonhosted.org/packages/c7/ec/2cdb90c6d9d61410b3df9ca67c210b60bf9b07aac31f800380b20b90386c/torchvision-0.22.0-cp310-cp310-win_amd64.whl", hash = "sha256:8c869df2e8e00f7b1d80a34439e6d4609b50fe3141032f50b38341ec2b59404e", size = 1716700, upload-time = "2025-04-23T14:42:03.562Z" }, - { url = "https://files.pythonhosted.org/packages/b1/43/28bc858b022f6337326d75f4027d2073aad5432328f01ee1236d847f1b82/torchvision-0.22.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:191ea28321fc262d8aa1a7fe79c41ff2848864bf382f9f6ea45c41dde8313792", size = 1947828, upload-time = "2025-04-23T14:42:00.439Z" }, - { url = "https://files.pythonhosted.org/packages/7e/71/ce9a303b94e64fe25d534593522ffc76848c4e64c11e4cbe9f6b8d537210/torchvision-0.22.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:6c5620e10ffe388eb6f4744962106ed7cf1508d26e6fdfa0c10522d3249aea24", size = 2514016, upload-time = "2025-04-23T14:41:48.566Z" }, - { url = "https://files.pythonhosted.org/packages/09/42/6908bff012a1dcc4fc515e52339652d7f488e208986542765c02ea775c2f/torchvision-0.22.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:ce292701c77c64dd3935e3e31c722c3b8b176a75f76dc09b804342efc1db5494", size = 7447546, upload-time = "2025-04-23T14:41:47.297Z" }, - { url = "https://files.pythonhosted.org/packages/e4/cf/8f9305cc0ea26badbbb3558ecae54c04a245429f03168f7fad502f8a5b25/torchvision-0.22.0-cp311-cp311-win_amd64.whl", hash = "sha256:e4017b5685dbab4250df58084f07d95e677b2f3ed6c2e507a1afb8eb23b580ca", size = 1716472, upload-time = "2025-04-23T14:42:01.999Z" }, - { url = "https://files.pythonhosted.org/packages/cb/ea/887d1d61cf4431a46280972de665f350af1898ce5006cd046326e5d0a2f2/torchvision-0.22.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:31c3165418fe21c3d81fe3459e51077c2f948801b8933ed18169f54652796a0f", size = 1947826, upload-time = "2025-04-23T14:41:59.188Z" }, - { url = "https://files.pythonhosted.org/packages/72/ef/21f8b6122e13ae045b8e49658029c695fd774cd21083b3fa5c3f9c5d3e35/torchvision-0.22.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:8f116bc82e0c076e70ba7776e611ed392b9666aa443662e687808b08993d26af", size = 2514571, upload-time = "2025-04-23T14:41:53.458Z" }, - { url = "https://files.pythonhosted.org/packages/7c/48/5f7617f6c60d135f86277c53f9d5682dfa4e66f4697f505f1530e8b69fb1/torchvision-0.22.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:ce4dc334ebd508de2c534817c9388e928bc2500cf981906ae8d6e2ca3bf4727a", size = 7446522, upload-time = "2025-04-23T14:41:34.9Z" }, - { url = "https://files.pythonhosted.org/packages/99/94/a015e93955f5d3a68689cc7c385a3cfcd2d62b84655d18b61f32fb04eb67/torchvision-0.22.0-cp312-cp312-win_amd64.whl", hash = "sha256:24b8c9255c209ca419cc7174906da2791c8b557b75c23496663ec7d73b55bebf", size = 1716664, upload-time = "2025-04-23T14:41:58.019Z" }, - { url = "https://files.pythonhosted.org/packages/e1/2a/9b34685599dcb341d12fc2730055155623db7a619d2415a8d31f17050952/torchvision-0.22.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:ece17995857dd328485c9c027c0b20ffc52db232e30c84ff6c95ab77201112c5", size = 1947823, upload-time = "2025-04-23T14:41:39.956Z" }, - { url = "https://files.pythonhosted.org/packages/77/77/88f64879483d66daf84f1d1c4d5c31ebb08e640411139042a258d5f7dbfe/torchvision-0.22.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:471c6dd75bb984c6ebe4f60322894a290bf3d4b195e769d80754f3689cd7f238", size = 2471592, upload-time = "2025-04-23T14:41:54.991Z" }, - { url = "https://files.pythonhosted.org/packages/f7/82/2f813eaae7c1fae1f9d9e7829578f5a91f39ef48d6c1c588a8900533dd3d/torchvision-0.22.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:2b839ac0610a38f56bef115ee5b9eaca5f9c2da3c3569a68cc62dbcc179c157f", size = 7446333, upload-time = "2025-04-23T14:41:36.603Z" }, - { url = "https://files.pythonhosted.org/packages/58/19/ca7a4f8907a56351dfe6ae0a708f4e6b3569b5c61d282e3e7f61cf42a4ce/torchvision-0.22.0-cp313-cp313-win_amd64.whl", hash = "sha256:4ada1c08b2f761443cd65b7c7b4aec9e2fc28f75b0d4e1b1ebc9d3953ebccc4d", size = 1716693, upload-time = "2025-04-23T14:41:41.031Z" }, - { url = "https://files.pythonhosted.org/packages/6f/a7/f43e9c8d13118b4ffbaebea664c9338ab20fa115a908125afd2238ff16e7/torchvision-0.22.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:cdc96daa4658b47ce9384154c86ed1e70cba9d972a19f5de6e33f8f94a626790", size = 2137621, upload-time = "2025-04-23T14:41:51.427Z" }, - { url = "https://files.pythonhosted.org/packages/6a/9a/2b59f5758ba7e3f23bc84e16947493bbce97392ec6d18efba7bdf0a3b10e/torchvision-0.22.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:753d3c84eeadd5979a33b3b73a25ecd0aa4af44d6b45ed2c70d44f5e0ac68312", size = 2476555, upload-time = "2025-04-23T14:41:38.357Z" }, - { url = "https://files.pythonhosted.org/packages/7d/40/a7bc2ab9b1e56d10a7fd9ae83191bb425fa308caa23d148f1c568006e02c/torchvision-0.22.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:b30e3ed29e4a61f7499bca50f57d8ebd23dfc52b14608efa17a534a55ee59a03", size = 7617924, upload-time = "2025-04-23T14:41:42.709Z" }, - { url = "https://files.pythonhosted.org/packages/c1/7b/30d423bdb2546250d719d7821aaf9058cc093d165565b245b159c788a9dd/torchvision-0.22.0-cp313-cp313t-win_amd64.whl", hash = "sha256:e5d680162694fac4c8a374954e261ddfb4eb0ce103287b0f693e4e9c579ef957", size = 1638621, upload-time = "2025-04-23T14:41:46.06Z" }, + { url = "https://files.pythonhosted.org/packages/15/2c/7b67117b14c6cc84ae3126ca6981abfa3af2ac54eb5252b80d9475fb40df/torchvision-0.22.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3b47d8369ee568c067795c0da0b4078f39a9dfea6f3bc1f3ac87530dfda1dd56", size = 1947825, upload-time = "2025-06-04T17:43:15.523Z" }, + { url = "https://files.pythonhosted.org/packages/6c/9f/c4dcf1d232b75e28bc37e21209ab2458d6d60235e16163544ed693de54cb/torchvision-0.22.1-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:990de4d657a41ed71680cd8be2e98ebcab55371f30993dc9bd2e676441f7180e", size = 2512611, upload-time = "2025-06-04T17:43:03.951Z" }, + { url = "https://files.pythonhosted.org/packages/e2/99/db71d62d12628111d59147095527a0ab492bdfecfba718d174c04ae6c505/torchvision-0.22.1-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:3347f690c2eed6d02aa0edfb9b01d321e7f7cf1051992d96d8d196c39b881d49", size = 7485668, upload-time = "2025-06-04T17:43:09.453Z" }, + { url = "https://files.pythonhosted.org/packages/32/ff/4a93a4623c3e5f97e8552af0f9f81d289dcf7f2ac71f1493f1c93a6b973d/torchvision-0.22.1-cp310-cp310-win_amd64.whl", hash = "sha256:86ad938f5a6ca645f0d5fb19484b1762492c2188c0ffb05c602e9e9945b7b371", size = 1707961, upload-time = "2025-06-04T17:43:13.038Z" }, + { url = "https://files.pythonhosted.org/packages/f6/00/bdab236ef19da050290abc2b5203ff9945c84a1f2c7aab73e8e9c8c85669/torchvision-0.22.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:4addf626e2b57fc22fd6d329cf1346d474497672e6af8383b7b5b636fba94a53", size = 1947827, upload-time = "2025-06-04T17:43:10.84Z" }, + { url = "https://files.pythonhosted.org/packages/ac/d0/18f951b2be3cfe48c0027b349dcc6fde950e3dc95dd83e037e86f284f6fd/torchvision-0.22.1-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:8b4a53a6067d63adba0c52f2b8dd2290db649d642021674ee43c0c922f0c6a69", size = 2514021, upload-time = "2025-06-04T17:43:07.608Z" }, + { url = "https://files.pythonhosted.org/packages/c3/1a/63eb241598b36d37a0221e10af357da34bd33402ccf5c0765e389642218a/torchvision-0.22.1-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:b7866a3b326413e67724ac46f1ee594996735e10521ba9e6cdbe0fa3cd98c2f2", size = 7487300, upload-time = "2025-06-04T17:42:58.349Z" }, + { url = "https://files.pythonhosted.org/packages/e5/73/1b009b42fe4a7774ba19c23c26bb0f020d68525c417a348b166f1c56044f/torchvision-0.22.1-cp311-cp311-win_amd64.whl", hash = "sha256:bb3f6df6f8fd415ce38ec4fd338376ad40c62e86052d7fc706a0dd51efac1718", size = 1707989, upload-time = "2025-06-04T17:43:14.332Z" }, + { url = "https://files.pythonhosted.org/packages/02/90/f4e99a5112dc221cf68a485e853cc3d9f3f1787cb950b895f3ea26d1ea98/torchvision-0.22.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:153f1790e505bd6da123e21eee6e83e2e155df05c0fe7d56347303067d8543c5", size = 1947827, upload-time = "2025-06-04T17:43:11.945Z" }, + { url = "https://files.pythonhosted.org/packages/25/f6/53e65384cdbbe732cc2106bb04f7fb908487e4fb02ae4a1613ce6904a122/torchvision-0.22.1-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:964414eef19459d55a10e886e2fca50677550e243586d1678f65e3f6f6bac47a", size = 2514576, upload-time = "2025-06-04T17:43:02.707Z" }, + { url = "https://files.pythonhosted.org/packages/17/8b/155f99042f9319bd7759536779b2a5b67cbd4f89c380854670850f89a2f4/torchvision-0.22.1-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:699c2d70d33951187f6ed910ea05720b9b4aaac1dcc1135f53162ce7d42481d3", size = 7485962, upload-time = "2025-06-04T17:42:43.606Z" }, + { url = "https://files.pythonhosted.org/packages/05/17/e45d5cd3627efdb47587a0634179a3533593436219de3f20c743672d2a79/torchvision-0.22.1-cp312-cp312-win_amd64.whl", hash = "sha256:75e0897da7a8e43d78632f66f2bdc4f6e26da8d3f021a7c0fa83746073c2597b", size = 1707992, upload-time = "2025-06-04T17:42:53.207Z" }, + { url = "https://files.pythonhosted.org/packages/7a/30/fecdd09fb973e963da68207fe9f3d03ec6f39a935516dc2a98397bf495c6/torchvision-0.22.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:9c3ae3319624c43cc8127020f46c14aa878406781f0899bb6283ae474afeafbf", size = 1947818, upload-time = "2025-06-04T17:42:51.954Z" }, + { url = "https://files.pythonhosted.org/packages/55/f4/b45f6cd92fa0acfac5e31b8e9258232f25bcdb0709a604e8b8a39d76e411/torchvision-0.22.1-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:4a614a6a408d2ed74208d0ea6c28a2fbb68290e9a7df206c5fef3f0b6865d307", size = 2471597, upload-time = "2025-06-04T17:42:48.838Z" }, + { url = "https://files.pythonhosted.org/packages/8d/b0/3cffd6a285b5ffee3fe4a31caff49e350c98c5963854474d1c4f7a51dea5/torchvision-0.22.1-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:7ee682be589bb1a002b7704f06b8ec0b89e4b9068f48e79307d2c6e937a9fdf4", size = 7485894, upload-time = "2025-06-04T17:43:01.371Z" }, + { url = "https://files.pythonhosted.org/packages/fd/1d/0ede596fedc2080d18108149921278b59f220fbb398f29619495337b0f86/torchvision-0.22.1-cp313-cp313-win_amd64.whl", hash = "sha256:2566cafcfa47ecfdbeed04bab8cef1307c8d4ef75046f7624b9e55f384880dfe", size = 1708020, upload-time = "2025-06-04T17:43:06.085Z" }, + { url = "https://files.pythonhosted.org/packages/0f/ca/e9a06bd61ee8e04fb4962a3fb524fe6ee4051662db07840b702a9f339b24/torchvision-0.22.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:043d9e35ed69c2e586aff6eb9e2887382e7863707115668ac9d140da58f42cba", size = 2137623, upload-time = "2025-06-04T17:43:05.028Z" }, + { url = "https://files.pythonhosted.org/packages/ab/c8/2ebe90f18e7ffa2120f5c3eab62aa86923185f78d2d051a455ea91461608/torchvision-0.22.1-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:27142bcc8a984227a6dcf560985e83f52b82a7d3f5fe9051af586a2ccc46ef26", size = 2476561, upload-time = "2025-06-04T17:42:59.691Z" }, + { url = "https://files.pythonhosted.org/packages/94/8b/04c6b15f8c29b39f0679589753091cec8b192ab296d4fdaf9055544c4ec9/torchvision-0.22.1-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:ef46e065502f7300ad6abc98554131c35dc4c837b978d91306658f1a65c00baa", size = 7658543, upload-time = "2025-06-04T17:42:46.064Z" }, + { url = "https://files.pythonhosted.org/packages/ab/c0/131628e6d42682b0502c63fd7f647b8b5ca4bd94088f6c85ca7225db8ac4/torchvision-0.22.1-cp313-cp313t-win_amd64.whl", hash = "sha256:7414eeacfb941fa21acddcd725f1617da5630ec822e498660a4b864d7d998075", size = 1629892, upload-time = "2025-06-04T17:42:57.156Z" }, ] [[package]] name = "tornado" -version = "6.4.2" +version = "6.5.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/59/45/a0daf161f7d6f36c3ea5fc0c2de619746cc3dd4c76402e9db545bd920f63/tornado-6.4.2.tar.gz", hash = "sha256:92bad5b4746e9879fd7bf1eb21dce4e3fc5128d71601f80005afa39237ad620b", size = 501135, upload-time = "2024-11-22T03:06:38.036Z" } +sdist = { url = "https://files.pythonhosted.org/packages/51/89/c72771c81d25d53fe33e3dca61c233b665b2780f21820ba6fd2c6793c12b/tornado-6.5.1.tar.gz", hash = "sha256:84ceece391e8eb9b2b95578db65e920d2a61070260594819589609ba9bc6308c", size = 509934, upload-time = "2025-05-22T18:15:38.788Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/26/7e/71f604d8cea1b58f82ba3590290b66da1e72d840aeb37e0d5f7291bd30db/tornado-6.4.2-cp38-abi3-macosx_10_9_universal2.whl", hash = "sha256:e828cce1123e9e44ae2a50a9de3055497ab1d0aeb440c5ac23064d9e44880da1", size = 436299, upload-time = "2024-11-22T03:06:20.162Z" }, - { url = "https://files.pythonhosted.org/packages/96/44/87543a3b99016d0bf54fdaab30d24bf0af2e848f1d13d34a3a5380aabe16/tornado-6.4.2-cp38-abi3-macosx_10_9_x86_64.whl", hash = "sha256:072ce12ada169c5b00b7d92a99ba089447ccc993ea2143c9ede887e0937aa803", size = 434253, upload-time = "2024-11-22T03:06:22.39Z" }, - { url = "https://files.pythonhosted.org/packages/cb/fb/fdf679b4ce51bcb7210801ef4f11fdac96e9885daa402861751353beea6e/tornado-6.4.2-cp38-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1a017d239bd1bb0919f72af256a970624241f070496635784d9bf0db640d3fec", size = 437602, upload-time = "2024-11-22T03:06:24.214Z" }, - { url = "https://files.pythonhosted.org/packages/4f/3b/e31aeffffc22b475a64dbeb273026a21b5b566f74dee48742817626c47dc/tornado-6.4.2-cp38-abi3-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c36e62ce8f63409301537222faffcef7dfc5284f27eec227389f2ad11b09d946", size = 436972, upload-time = "2024-11-22T03:06:25.559Z" }, - { url = "https://files.pythonhosted.org/packages/22/55/b78a464de78051a30599ceb6983b01d8f732e6f69bf37b4ed07f642ac0fc/tornado-6.4.2-cp38-abi3-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bca9eb02196e789c9cb5c3c7c0f04fb447dc2adffd95265b2c7223a8a615ccbf", size = 437173, upload-time = "2024-11-22T03:06:27.584Z" }, - { url = "https://files.pythonhosted.org/packages/79/5e/be4fb0d1684eb822c9a62fb18a3e44a06188f78aa466b2ad991d2ee31104/tornado-6.4.2-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:304463bd0772442ff4d0f5149c6f1c2135a1fae045adf070821c6cdc76980634", size = 437892, upload-time = "2024-11-22T03:06:28.933Z" }, - { url = "https://files.pythonhosted.org/packages/f5/33/4f91fdd94ea36e1d796147003b490fe60a0215ac5737b6f9c65e160d4fe0/tornado-6.4.2-cp38-abi3-musllinux_1_2_i686.whl", hash = "sha256:c82c46813ba483a385ab2a99caeaedf92585a1f90defb5693351fa7e4ea0bf73", size = 437334, upload-time = "2024-11-22T03:06:30.428Z" }, - { url = "https://files.pythonhosted.org/packages/2b/ae/c1b22d4524b0e10da2f29a176fb2890386f7bd1f63aacf186444873a88a0/tornado-6.4.2-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:932d195ca9015956fa502c6b56af9eb06106140d844a335590c1ec7f5277d10c", size = 437261, upload-time = "2024-11-22T03:06:32.458Z" }, - { url = "https://files.pythonhosted.org/packages/b5/25/36dbd49ab6d179bcfc4c6c093a51795a4f3bed380543a8242ac3517a1751/tornado-6.4.2-cp38-abi3-win32.whl", hash = "sha256:2876cef82e6c5978fde1e0d5b1f919d756968d5b4282418f3146b79b58556482", size = 438463, upload-time = "2024-11-22T03:06:34.71Z" }, - { url = "https://files.pythonhosted.org/packages/61/cc/58b1adeb1bb46228442081e746fcdbc4540905c87e8add7c277540934edb/tornado-6.4.2-cp38-abi3-win_amd64.whl", hash = "sha256:908b71bf3ff37d81073356a5fadcc660eb10c1476ee6e2725588626ce7e5ca38", size = 438907, upload-time = "2024-11-22T03:06:36.71Z" }, + { url = "https://files.pythonhosted.org/packages/77/89/f4532dee6843c9e0ebc4e28d4be04c67f54f60813e4bf73d595fe7567452/tornado-6.5.1-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:d50065ba7fd11d3bd41bcad0825227cc9a95154bad83239357094c36708001f7", size = 441948, upload-time = "2025-05-22T18:15:20.862Z" }, + { url = "https://files.pythonhosted.org/packages/15/9a/557406b62cffa395d18772e0cdcf03bed2fff03b374677348eef9f6a3792/tornado-6.5.1-cp39-abi3-macosx_10_9_x86_64.whl", hash = "sha256:9e9ca370f717997cb85606d074b0e5b247282cf5e2e1611568b8821afe0342d6", size = 440112, upload-time = "2025-05-22T18:15:22.591Z" }, + { url = "https://files.pythonhosted.org/packages/55/82/7721b7319013a3cf881f4dffa4f60ceff07b31b394e459984e7a36dc99ec/tornado-6.5.1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b77e9dfa7ed69754a54c89d82ef746398be82f749df69c4d3abe75c4d1ff4888", size = 443672, upload-time = "2025-05-22T18:15:24.027Z" }, + { url = "https://files.pythonhosted.org/packages/7d/42/d11c4376e7d101171b94e03cef0cbce43e823ed6567ceda571f54cf6e3ce/tornado-6.5.1-cp39-abi3-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:253b76040ee3bab8bcf7ba9feb136436a3787208717a1fb9f2c16b744fba7331", size = 443019, upload-time = "2025-05-22T18:15:25.735Z" }, + { url = "https://files.pythonhosted.org/packages/7d/f7/0c48ba992d875521ac761e6e04b0a1750f8150ae42ea26df1852d6a98942/tornado-6.5.1-cp39-abi3-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:308473f4cc5a76227157cdf904de33ac268af770b2c5f05ca6c1161d82fdd95e", size = 443252, upload-time = "2025-05-22T18:15:27.499Z" }, + { url = "https://files.pythonhosted.org/packages/89/46/d8d7413d11987e316df4ad42e16023cd62666a3c0dfa1518ffa30b8df06c/tornado-6.5.1-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:caec6314ce8a81cf69bd89909f4b633b9f523834dc1a352021775d45e51d9401", size = 443930, upload-time = "2025-05-22T18:15:29.299Z" }, + { url = "https://files.pythonhosted.org/packages/78/b2/f8049221c96a06df89bed68260e8ca94beca5ea532ffc63b1175ad31f9cc/tornado-6.5.1-cp39-abi3-musllinux_1_2_i686.whl", hash = "sha256:13ce6e3396c24e2808774741331638ee6c2f50b114b97a55c5b442df65fd9692", size = 443351, upload-time = "2025-05-22T18:15:31.038Z" }, + { url = "https://files.pythonhosted.org/packages/76/ff/6a0079e65b326cc222a54720a748e04a4db246870c4da54ece4577bfa702/tornado-6.5.1-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:5cae6145f4cdf5ab24744526cc0f55a17d76f02c98f4cff9daa08ae9a217448a", size = 443328, upload-time = "2025-05-22T18:15:32.426Z" }, + { url = "https://files.pythonhosted.org/packages/49/18/e3f902a1d21f14035b5bc6246a8c0f51e0eef562ace3a2cea403c1fb7021/tornado-6.5.1-cp39-abi3-win32.whl", hash = "sha256:e0a36e1bc684dca10b1aa75a31df8bdfed656831489bc1e6a6ebed05dc1ec365", size = 444396, upload-time = "2025-05-22T18:15:34.205Z" }, + { url = "https://files.pythonhosted.org/packages/7b/09/6526e32bf1049ee7de3bebba81572673b19a2a8541f795d887e92af1a8bc/tornado-6.5.1-cp39-abi3-win_amd64.whl", hash = "sha256:908e7d64567cecd4c2b458075589a775063453aeb1d2a1853eedb806922f568b", size = 444840, upload-time = "2025-05-22T18:15:36.1Z" }, + { url = "https://files.pythonhosted.org/packages/55/a7/535c44c7bea4578e48281d83c615219f3ab19e6abc67625ef637c73987be/tornado-6.5.1-cp39-abi3-win_arm64.whl", hash = "sha256:02420a0eb7bf617257b9935e2b754d1b63897525d8a289c9d65690d580b4dcf7", size = 443596, upload-time = "2025-05-22T18:15:37.433Z" }, ] [[package]] @@ -3207,26 +3961,38 @@ wheels = [ [[package]] name = "triton" -version = "3.3.0" +version = "3.3.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "setuptools", marker = "sys_platform == 'linux'" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/76/04/d54d3a6d077c646624dc9461b0059e23fd5d30e0dbe67471e3654aec81f9/triton-3.3.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fad99beafc860501d7fcc1fb7045d9496cbe2c882b1674640304949165a916e7", size = 156441993, upload-time = "2025-04-09T20:27:25.107Z" }, - { url = "https://files.pythonhosted.org/packages/3c/c5/4874a81131cc9e934d88377fbc9d24319ae1fb540f3333b4e9c696ebc607/triton-3.3.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:3161a2bf073d6b22c4e2f33f951f3e5e3001462b2570e6df9cd57565bdec2984", size = 156528461, upload-time = "2025-04-09T20:27:32.599Z" }, - { url = "https://files.pythonhosted.org/packages/11/53/ce18470914ab6cfbec9384ee565d23c4d1c55f0548160b1c7b33000b11fd/triton-3.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b68c778f6c4218403a6bd01be7484f6dc9e20fe2083d22dd8aef33e3b87a10a3", size = 156504509, upload-time = "2025-04-09T20:27:40.413Z" }, - { url = "https://files.pythonhosted.org/packages/7d/74/4bf2702b65e93accaa20397b74da46fb7a0356452c1bb94dbabaf0582930/triton-3.3.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:47bc87ad66fa4ef17968299acacecaab71ce40a238890acc6ad197c3abe2b8f1", size = 156516468, upload-time = "2025-04-09T20:27:48.196Z" }, - { url = "https://files.pythonhosted.org/packages/0a/93/f28a696fa750b9b608baa236f8225dd3290e5aff27433b06143adc025961/triton-3.3.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ce4700fc14032af1e049005ae94ba908e71cd6c2df682239aed08e49bc71b742", size = 156580729, upload-time = "2025-04-09T20:27:55.424Z" }, + { url = "https://files.pythonhosted.org/packages/8d/a9/549e51e9b1b2c9b854fd761a1d23df0ba2fbc60bd0c13b489ffa518cfcb7/triton-3.3.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b74db445b1c562844d3cfad6e9679c72e93fdfb1a90a24052b03bb5c49d1242e", size = 155600257, upload-time = "2025-05-29T23:39:36.085Z" }, + { url = "https://files.pythonhosted.org/packages/21/2f/3e56ea7b58f80ff68899b1dbe810ff257c9d177d288c6b0f55bf2fe4eb50/triton-3.3.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b31e3aa26f8cb3cc5bf4e187bf737cbacf17311e1112b781d4a059353dfd731b", size = 155689937, upload-time = "2025-05-29T23:39:44.182Z" }, + { url = "https://files.pythonhosted.org/packages/24/5f/950fb373bf9c01ad4eb5a8cd5eaf32cdf9e238c02f9293557a2129b9c4ac/triton-3.3.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9999e83aba21e1a78c1f36f21bce621b77bcaa530277a50484a7cb4a822f6e43", size = 155669138, upload-time = "2025-05-29T23:39:51.771Z" }, + { url = "https://files.pythonhosted.org/packages/74/1f/dfb531f90a2d367d914adfee771babbd3f1a5b26c3f5fbc458dee21daa78/triton-3.3.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b89d846b5a4198317fec27a5d3a609ea96b6d557ff44b56c23176546023c4240", size = 155673035, upload-time = "2025-05-29T23:40:02.468Z" }, + { url = "https://files.pythonhosted.org/packages/28/71/bd20ffcb7a64c753dc2463489a61bf69d531f308e390ad06390268c4ea04/triton-3.3.1-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a3198adb9d78b77818a5388bff89fa72ff36f9da0bc689db2f0a651a67ce6a42", size = 155735832, upload-time = "2025-05-29T23:40:10.522Z" }, ] [[package]] name = "typing-extensions" -version = "4.12.2" +version = "4.14.1" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/98/5a/da40306b885cc8c09109dc2e1abd358d5684b1425678151cdaed4731c822/typing_extensions-4.14.1.tar.gz", hash = "sha256:38b39f4aeeab64884ce9f74c94263ef78f3c22467c8724005483154c26648d36", size = 107673, upload-time = "2025-07-04T13:28:34.16Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/b5/00/d631e67a838026495268c2f6884f3711a15a9a2a96cd244fdaea53b823fb/typing_extensions-4.14.1-py3-none-any.whl", hash = "sha256:d1e1e3b58374dc93031d6eda2420a48ea44a36c2b4766a4fdeb3710755731d76", size = 43906, upload-time = "2025-07-04T13:28:32.743Z" }, +] + +[[package]] +name = "typing-inspection" +version = "0.4.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/df/db/f35a00659bc03fec321ba8bce9420de607a1d37f8342eee1863174c69557/typing_extensions-4.12.2.tar.gz", hash = "sha256:1a7ead55c7e559dd4dee8856e3a88b41225abfe1ce8df57b7c13915fe121ffb8", size = 85321, upload-time = "2024-06-07T18:52:15.995Z" } +dependencies = [ + { name = "typing-extensions" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/f8/b1/0c11f5058406b3af7609f121aaa6b609744687f1d158b3c3a5bf4cc94238/typing_inspection-0.4.1.tar.gz", hash = "sha256:6ae134cc0203c33377d43188d4064e9b357dba58cff3185f22924610e70a9d28", size = 75726, upload-time = "2025-05-21T18:55:23.885Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/26/9f/ad63fc0248c5379346306f8668cda6e2e2e9c95e01216d2b8ffd9ff037d0/typing_extensions-4.12.2-py3-none-any.whl", hash = "sha256:04e5ca0351e0f3f85c6853954072df659d0d13fac324d0072316b67d7794700d", size = 37438, upload-time = "2024-06-07T18:52:13.582Z" }, + { url = "https://files.pythonhosted.org/packages/17/69/cd203477f944c353c31bade965f880aa1061fd6bf05ded0726ca845b6ff7/typing_inspection-0.4.1-py3-none-any.whl", hash = "sha256:389055682238f53b04f7badcb49b989835495a96700ced5dab2d8feae4b26f51", size = 14552, upload-time = "2025-05-21T18:55:22.152Z" }, ] [[package]] @@ -3252,67 +4018,64 @@ wheels = [ [[package]] name = "urllib3" -version = "2.3.0" +version = "2.5.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/aa/63/e53da845320b757bf29ef6a9062f5c669fe997973f966045cb019c3f4b66/urllib3-2.3.0.tar.gz", hash = "sha256:f8c5449b3cf0861679ce7e0503c7b44b5ec981bec0d1d3795a07f1ba96f0204d", size = 307268, upload-time = "2024-12-22T07:47:30.032Z" } +sdist = { url = "https://files.pythonhosted.org/packages/15/22/9ee70a2574a4f4599c47dd506532914ce044817c7752a79b6a51286319bc/urllib3-2.5.0.tar.gz", hash = "sha256:3fc47733c7e419d4bc3f6b3dc2b4f890bb743906a30d56ba4a5bfa4bbff92760", size = 393185, upload-time = "2025-06-18T14:07:41.644Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c8/19/4ec628951a74043532ca2cf5d97b7b14863931476d117c471e8e2b1eb39f/urllib3-2.3.0-py3-none-any.whl", hash = "sha256:1cee9ad369867bfdbbb48b7dd50374c0967a0bb7710050facf0dd6911440e3df", size = 128369, upload-time = "2024-12-22T07:47:28.074Z" }, + { url = "https://files.pythonhosted.org/packages/a7/c2/fe1e52489ae3122415c51f387e221dd0773709bad6c6cdaa599e8a2c5185/urllib3-2.5.0-py3-none-any.whl", hash = "sha256:e6b01673c0fa6a13e374b50871808eb3bf7046c4b125b216f6bf1cc604cff0dc", size = 129795, upload-time = "2025-06-18T14:07:40.39Z" }, ] [[package]] name = "virtualenv" -version = "20.29.3" +version = "20.33.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "distlib" }, { name = "filelock" }, { name = "platformdirs" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/c7/9c/57d19fa093bcf5ac61a48087dd44d00655f85421d1aa9722f8befbf3f40a/virtualenv-20.29.3.tar.gz", hash = "sha256:95e39403fcf3940ac45bc717597dba16110b74506131845d9b687d5e73d947ac", size = 4320280, upload-time = "2025-03-06T19:54:19.055Z" } +sdist = { url = "https://files.pythonhosted.org/packages/db/2e/8a70dcbe8bf15213a08f9b0325ede04faca5d362922ae0d62ef0fa4b069d/virtualenv-20.33.0.tar.gz", hash = "sha256:47e0c0d2ef1801fce721708ccdf2a28b9403fa2307c3268aebd03225976f61d2", size = 6082069, upload-time = "2025-08-03T08:09:19.014Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c2/eb/c6db6e3001d58c6a9e67c74bb7b4206767caa3ccc28c6b9eaf4c23fb4e34/virtualenv-20.29.3-py3-none-any.whl", hash = "sha256:3e3d00f5807e83b234dfb6122bf37cfadf4be216c53a49ac059d02414f819170", size = 4301458, upload-time = "2025-03-06T19:54:16.923Z" }, + { url = "https://files.pythonhosted.org/packages/43/87/b22cf40cdf7e2b2bf83f38a94d2c90c5ad6c304896e5a12d0c08a602eb59/virtualenv-20.33.0-py3-none-any.whl", hash = "sha256:106b6baa8ab1b526d5a9b71165c85c456fbd49b16976c88e2bc9352ee3bc5d3f", size = 6060205, upload-time = "2025-08-03T08:09:16.674Z" }, ] [[package]] name = "wadler-lindig" -version = "0.1.4" +version = "0.1.7" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/39/7a/fea25d7985211556bbe2511d42e07453b484bf8e0d5d6109aabb08f52784/wadler_lindig-0.1.4.tar.gz", hash = "sha256:75aa3ddd384573c41d5c910fd990e655c2a641e5093cf5081650d0229daf87ad", size = 15356, upload-time = "2025-03-15T21:53:42.05Z" } +sdist = { url = "https://files.pythonhosted.org/packages/1e/67/cbae4bf7683a64755c2c1778c418fea96d00e34395bb91743f08bd951571/wadler_lindig-0.1.7.tar.gz", hash = "sha256:81d14d3fe77d441acf3ebd7f4aefac20c74128bf460e84b512806dccf7b2cd55", size = 15842, upload-time = "2025-06-18T07:00:42.843Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/7b/69/cfb1af44622044d4db0cad65721d283a921a4795f0ad121616b9eaa6ccd7/wadler_lindig-0.1.4-py3-none-any.whl", hash = "sha256:5c463aeb1f4ddc4acc12c3708d22ae21bcfc3e19e7c4d7aeef6642ea57b1a8b8", size = 20126, upload-time = "2025-03-15T21:53:40.835Z" }, + { url = "https://files.pythonhosted.org/packages/8d/96/04e7b441807b26b794da5b11e59ed7f83b2cf8af202bd7eba8ad2fa6046e/wadler_lindig-0.1.7-py3-none-any.whl", hash = "sha256:e3ec83835570fd0a9509f969162aeb9c65618f998b1f42918cfc8d45122fe953", size = 20516, upload-time = "2025-06-18T07:00:41.684Z" }, ] [[package]] name = "wandb" -version = "0.19.11" +version = "0.21.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "click" }, - { name = "docker-pycreds" }, { name = "gitpython" }, + { name = "packaging" }, { name = "platformdirs" }, { name = "protobuf" }, - { name = "psutil" }, { name = "pydantic" }, { name = "pyyaml" }, { name = "requests" }, { name = "sentry-sdk" }, - { name = "setproctitle" }, - { name = "setuptools" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/39/98/0ff2925a21b998d4b84731429f4554ca3d9b5cad42c09c075e7306c3aca0/wandb-0.19.11.tar.gz", hash = "sha256:3f50a27dfadbb25946a513ffe856c0e8e538b5626ef207aa50b00c3b0356bff8", size = 39511477, upload-time = "2025-05-07T20:50:01.341Z" } +sdist = { url = "https://files.pythonhosted.org/packages/73/09/c84264a219e20efd615e4d5d150cc7d359d57d51328d3fa94ee02d70ed9c/wandb-0.21.0.tar.gz", hash = "sha256:473e01ef200b59d780416062991effa7349a34e51425d4be5ff482af2dc39e02", size = 40085784, upload-time = "2025-07-02T00:24:15.516Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/4f/2c/f8bab58c73fdde4442f1baffd9ea5d1bb3113906a97a27e8d9ab72db7a69/wandb-0.19.11-py3-none-any.whl", hash = "sha256:ff3bf050ba25ebae7aedc9a775ffab90c28068832edfe5458423f488c2558f82", size = 6481327, upload-time = "2025-05-07T20:49:33.461Z" }, - { url = "https://files.pythonhosted.org/packages/45/4a/34b364280f690f4c6d7660f528fba9f13bdecabc4c869d266a4632cf836e/wandb-0.19.11-py3-none-macosx_10_14_x86_64.whl", hash = "sha256:0823fd9aa6343f40c04e01959997ca8c6d6adf1bd81c8d45261fa4915f1c6b67", size = 20555751, upload-time = "2025-05-07T20:49:36.392Z" }, - { url = "https://files.pythonhosted.org/packages/d8/e6/a27868fdb83a60df37b9d15e52c3353dd88d74442f27ae48cf765c6b9554/wandb-0.19.11-py3-none-macosx_11_0_arm64.whl", hash = "sha256:c758ef5439599d9023db5b3cf1698477055d82f9fae48af2779f63f1d289167c", size = 20377587, upload-time = "2025-05-07T20:49:39.126Z" }, - { url = "https://files.pythonhosted.org/packages/21/f7/d5cf5b58c2b3015364c7b2b6af6a440cbeda4103b67332e1e64b30f6252d/wandb-0.19.11-py3-none-macosx_11_0_x86_64.whl", hash = "sha256:de2dfd4911e7691735e271654c735e7b90cdee9d29a3796fbf06e9e92d48f3d7", size = 20985041, upload-time = "2025-05-07T20:49:41.571Z" }, - { url = "https://files.pythonhosted.org/packages/68/06/8b827f16a0b8f18002d2fffa7c5a7fd447946e0d0c68aeec0dd7eb18cdd3/wandb-0.19.11-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cfff738850770d26b13f8f3fe400a6456f1e39e87f3f29d5aa241b249476df95", size = 20017696, upload-time = "2025-05-07T20:49:44.04Z" }, - { url = "https://files.pythonhosted.org/packages/f9/31/eeb2878b26566c04c3e9b8b20b3ec3c54a2be50535088d36a37c008e07a3/wandb-0.19.11-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e8ff673007448df11cc69379ae0df28ead866800dc1ec7bc151b402db0bbcf40", size = 21425857, upload-time = "2025-05-07T20:49:46.347Z" }, - { url = "https://files.pythonhosted.org/packages/10/30/08988360678ae78334bb16625c28260fcaba49f500b89f8766807cb74d71/wandb-0.19.11-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:858bc5023fa1b3285d89d15f62be78afdb28301064daa49ea3f4ebde5dcedad2", size = 20023145, upload-time = "2025-05-07T20:49:48.965Z" }, - { url = "https://files.pythonhosted.org/packages/c8/e9/a639c42c8ca517c4d25e8970d64d0c5a9bd35b784faed5f47d9cca3dcd12/wandb-0.19.11-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:90e4b57649896acb16c3dd41b3093df1a169c2f1d94ff15d76af86b8a60dcdac", size = 21504842, upload-time = "2025-05-07T20:49:51.628Z" }, - { url = "https://files.pythonhosted.org/packages/44/74/dbe9277dd935b77dd16939cdf15357766fec0813a6e336cf5f1d07eb016e/wandb-0.19.11-py3-none-win32.whl", hash = "sha256:38dea43c7926d8800405a73b80b9adfe81eb315fc6f2ac6885c77eb966634421", size = 20767584, upload-time = "2025-05-07T20:49:56.629Z" }, - { url = "https://files.pythonhosted.org/packages/36/d5/215cac3edec5c5ac6e7231beb9d22466d5d4e4a132fa3a1d044f7d682c15/wandb-0.19.11-py3-none-win_amd64.whl", hash = "sha256:73402003c56ddc2198878492ab2bff55bb49bce5587eae5960e737d27c0c48f7", size = 20767588, upload-time = "2025-05-07T20:49:58.85Z" }, + { url = "https://files.pythonhosted.org/packages/38/dd/65eac086e1bc337bb5f0eed65ba1fe4a6dbc62c97f094e8e9df1ef83ffed/wandb-0.21.0-py3-none-any.whl", hash = "sha256:316e8cd4329738f7562f7369e6eabeeb28ef9d473203f7ead0d03e5dba01c90d", size = 6504284, upload-time = "2025-07-02T00:23:46.671Z" }, + { url = "https://files.pythonhosted.org/packages/17/a7/80556ce9097f59e10807aa68f4a9b29d736a90dca60852a9e2af1641baf8/wandb-0.21.0-py3-none-macosx_10_14_x86_64.whl", hash = "sha256:701d9cbdfcc8550a330c1b54a26f1585519180e0f19247867446593d34ace46b", size = 21717388, upload-time = "2025-07-02T00:23:49.348Z" }, + { url = "https://files.pythonhosted.org/packages/23/ae/660bc75aa37bd23409822ea5ed616177d94873172d34271693c80405c820/wandb-0.21.0-py3-none-macosx_11_0_arm64.whl", hash = "sha256:01689faa6b691df23ba2367e0a1ecf6e4d0be44474905840098eedd1fbcb8bdf", size = 21141465, upload-time = "2025-07-02T00:23:52.602Z" }, + { url = "https://files.pythonhosted.org/packages/23/ab/9861929530be56557c74002868c85d0d8ac57050cc21863afe909ae3d46f/wandb-0.21.0-py3-none-macosx_11_0_x86_64.whl", hash = "sha256:55d3f42ddb7971d1699752dff2b85bcb5906ad098d18ab62846c82e9ce5a238d", size = 21793511, upload-time = "2025-07-02T00:23:55.447Z" }, + { url = "https://files.pythonhosted.org/packages/de/52/e5cad2eff6fbed1ac06f4a5b718457fa2fd437f84f5c8f0d31995a2ef046/wandb-0.21.0-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:893508f0c7da48917448daa5cd622c27ce7ce15119adaa861185034c2bd7b14c", size = 20704643, upload-time = "2025-07-02T00:23:58.255Z" }, + { url = "https://files.pythonhosted.org/packages/83/8f/6bed9358cc33767c877b221d4f565e1ddf00caf4bbbe54d2e3bbc932c6a7/wandb-0.21.0-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a4e8245a8912247ddf7654f7b5330f583a6c56ab88fee65589158490d583c57d", size = 22243012, upload-time = "2025-07-02T00:24:01.423Z" }, + { url = "https://files.pythonhosted.org/packages/be/61/9048015412ea5ca916844af55add4fed7c21fe1ad70bb137951e70b550c5/wandb-0.21.0-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:2e4c4f951e0d02755e315679bfdcb5bc38c1b02e2e5abc5432b91a91bb0cf246", size = 20716440, upload-time = "2025-07-02T00:24:04.198Z" }, + { url = "https://files.pythonhosted.org/packages/02/d9/fcd2273d8ec3f79323e40a031aba5d32d6fa9065702010eb428b5ffbab62/wandb-0.21.0-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:873749966eeac0069e0e742e6210641b6227d454fb1dae2cf5c437c6ed42d3ca", size = 22320652, upload-time = "2025-07-02T00:24:07.175Z" }, + { url = "https://files.pythonhosted.org/packages/80/68/b8308db6b9c3c96dcd03be17c019aee105e1d7dc1e74d70756cdfb9241c6/wandb-0.21.0-py3-none-win32.whl", hash = "sha256:9d3cccfba658fa011d6cab9045fa4f070a444885e8902ae863802549106a5dab", size = 21484296, upload-time = "2025-07-02T00:24:10.147Z" }, + { url = "https://files.pythonhosted.org/packages/cf/96/71cc033e8abd00e54465e68764709ed945e2da2d66d764f72f4660262b22/wandb-0.21.0-py3-none-win_amd64.whl", hash = "sha256:28a0b2dad09d7c7344ac62b0276be18a2492a5578e4d7c84937a3e1991edaac7", size = 21484301, upload-time = "2025-07-02T00:24:12.658Z" }, ] [[package]] @@ -3390,78 +4153,99 @@ wheels = [ [[package]] name = "yarl" -version = "1.18.3" +version = "1.20.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "idna" }, { name = "multidict" }, { name = "propcache" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/b7/9d/4b94a8e6d2b51b599516a5cb88e5bc99b4d8d4583e468057eaa29d5f0918/yarl-1.18.3.tar.gz", hash = "sha256:ac1801c45cbf77b6c99242eeff4fffb5e4e73a800b5c4ad4fc0be5def634d2e1", size = 181062, upload-time = "2024-12-01T20:35:23.292Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/d2/98/e005bc608765a8a5569f58e650961314873c8469c333616eb40bff19ae97/yarl-1.18.3-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:7df647e8edd71f000a5208fe6ff8c382a1de8edfbccdbbfe649d263de07d8c34", size = 141458, upload-time = "2024-12-01T20:32:32.604Z" }, - { url = "https://files.pythonhosted.org/packages/df/5d/f8106b263b8ae8a866b46d9be869ac01f9b3fb7f2325f3ecb3df8003f796/yarl-1.18.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:c69697d3adff5aa4f874b19c0e4ed65180ceed6318ec856ebc423aa5850d84f7", size = 94365, upload-time = "2024-12-01T20:32:35.736Z" }, - { url = "https://files.pythonhosted.org/packages/56/3e/d8637ddb9ba69bf851f765a3ee288676f7cf64fb3be13760c18cbc9d10bd/yarl-1.18.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:602d98f2c2d929f8e697ed274fbadc09902c4025c5a9963bf4e9edfc3ab6f7ed", size = 92181, upload-time = "2024-12-01T20:32:37.944Z" }, - { url = "https://files.pythonhosted.org/packages/76/f9/d616a5c2daae281171de10fba41e1c0e2d8207166fc3547252f7d469b4e1/yarl-1.18.3-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c654d5207c78e0bd6d749f6dae1dcbbfde3403ad3a4b11f3c5544d9906969dde", size = 315349, upload-time = "2024-12-01T20:32:40.126Z" }, - { url = "https://files.pythonhosted.org/packages/bb/b4/3ea5e7b6f08f698b3769a06054783e434f6d59857181b5c4e145de83f59b/yarl-1.18.3-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5094d9206c64181d0f6e76ebd8fb2f8fe274950a63890ee9e0ebfd58bf9d787b", size = 330494, upload-time = "2024-12-01T20:32:41.833Z" }, - { url = "https://files.pythonhosted.org/packages/55/f1/e0fc810554877b1b67420568afff51b967baed5b53bcc983ab164eebf9c9/yarl-1.18.3-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:35098b24e0327fc4ebdc8ffe336cee0a87a700c24ffed13161af80124b7dc8e5", size = 326927, upload-time = "2024-12-01T20:32:43.73Z" }, - { url = "https://files.pythonhosted.org/packages/a9/42/b1753949b327b36f210899f2dd0a0947c0c74e42a32de3f8eb5c7d93edca/yarl-1.18.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3236da9272872443f81fedc389bace88408f64f89f75d1bdb2256069a8730ccc", size = 319703, upload-time = "2024-12-01T20:32:46.131Z" }, - { url = "https://files.pythonhosted.org/packages/f0/6d/e87c62dc9635daefb064b56f5c97df55a2e9cc947a2b3afd4fd2f3b841c7/yarl-1.18.3-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e2c08cc9b16f4f4bc522771d96734c7901e7ebef70c6c5c35dd0f10845270bcd", size = 310246, upload-time = "2024-12-01T20:32:48.577Z" }, - { url = "https://files.pythonhosted.org/packages/e3/ef/e2e8d1785cdcbd986f7622d7f0098205f3644546da7919c24b95790ec65a/yarl-1.18.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:80316a8bd5109320d38eef8833ccf5f89608c9107d02d2a7f985f98ed6876990", size = 319730, upload-time = "2024-12-01T20:32:50.209Z" }, - { url = "https://files.pythonhosted.org/packages/fc/15/8723e22345bc160dfde68c4b3ae8b236e868f9963c74015f1bc8a614101c/yarl-1.18.3-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:c1e1cc06da1491e6734f0ea1e6294ce00792193c463350626571c287c9a704db", size = 321681, upload-time = "2024-12-01T20:32:52.498Z" }, - { url = "https://files.pythonhosted.org/packages/86/09/bf764e974f1516efa0ae2801494a5951e959f1610dd41edbfc07e5e0f978/yarl-1.18.3-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:fea09ca13323376a2fdfb353a5fa2e59f90cd18d7ca4eaa1fd31f0a8b4f91e62", size = 324812, upload-time = "2024-12-01T20:32:54.947Z" }, - { url = "https://files.pythonhosted.org/packages/f6/4c/20a0187e3b903c97d857cf0272d687c1b08b03438968ae8ffc50fe78b0d6/yarl-1.18.3-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:e3b9fd71836999aad54084906f8663dffcd2a7fb5cdafd6c37713b2e72be1760", size = 337011, upload-time = "2024-12-01T20:32:57.692Z" }, - { url = "https://files.pythonhosted.org/packages/c9/71/6244599a6e1cc4c9f73254a627234e0dad3883ece40cc33dce6265977461/yarl-1.18.3-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:757e81cae69244257d125ff31663249b3013b5dc0a8520d73694aed497fb195b", size = 338132, upload-time = "2024-12-01T20:33:00.247Z" }, - { url = "https://files.pythonhosted.org/packages/af/f5/e0c3efaf74566c4b4a41cb76d27097df424052a064216beccae8d303c90f/yarl-1.18.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:b1771de9944d875f1b98a745bc547e684b863abf8f8287da8466cf470ef52690", size = 331849, upload-time = "2024-12-01T20:33:02.492Z" }, - { url = "https://files.pythonhosted.org/packages/8a/b8/3d16209c2014c2f98a8f658850a57b716efb97930aebf1ca0d9325933731/yarl-1.18.3-cp310-cp310-win32.whl", hash = "sha256:8874027a53e3aea659a6d62751800cf6e63314c160fd607489ba5c2edd753cf6", size = 84309, upload-time = "2024-12-01T20:33:04.832Z" }, - { url = "https://files.pythonhosted.org/packages/fd/b7/2e9a5b18eb0fe24c3a0e8bae994e812ed9852ab4fd067c0107fadde0d5f0/yarl-1.18.3-cp310-cp310-win_amd64.whl", hash = "sha256:93b2e109287f93db79210f86deb6b9bbb81ac32fc97236b16f7433db7fc437d8", size = 90484, upload-time = "2024-12-01T20:33:06.615Z" }, - { url = "https://files.pythonhosted.org/packages/40/93/282b5f4898d8e8efaf0790ba6d10e2245d2c9f30e199d1a85cae9356098c/yarl-1.18.3-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:8503ad47387b8ebd39cbbbdf0bf113e17330ffd339ba1144074da24c545f0069", size = 141555, upload-time = "2024-12-01T20:33:08.819Z" }, - { url = "https://files.pythonhosted.org/packages/6d/9c/0a49af78df099c283ca3444560f10718fadb8a18dc8b3edf8c7bd9fd7d89/yarl-1.18.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:02ddb6756f8f4517a2d5e99d8b2f272488e18dd0bfbc802f31c16c6c20f22193", size = 94351, upload-time = "2024-12-01T20:33:10.609Z" }, - { url = "https://files.pythonhosted.org/packages/5a/a1/205ab51e148fdcedad189ca8dd587794c6f119882437d04c33c01a75dece/yarl-1.18.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:67a283dd2882ac98cc6318384f565bffc751ab564605959df4752d42483ad889", size = 92286, upload-time = "2024-12-01T20:33:12.322Z" }, - { url = "https://files.pythonhosted.org/packages/ed/fe/88b690b30f3f59275fb674f5f93ddd4a3ae796c2b62e5bb9ece8a4914b83/yarl-1.18.3-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d980e0325b6eddc81331d3f4551e2a333999fb176fd153e075c6d1c2530aa8a8", size = 340649, upload-time = "2024-12-01T20:33:13.842Z" }, - { url = "https://files.pythonhosted.org/packages/07/eb/3b65499b568e01f36e847cebdc8d7ccb51fff716dbda1ae83c3cbb8ca1c9/yarl-1.18.3-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:b643562c12680b01e17239be267bc306bbc6aac1f34f6444d1bded0c5ce438ca", size = 356623, upload-time = "2024-12-01T20:33:15.535Z" }, - { url = "https://files.pythonhosted.org/packages/33/46/f559dc184280b745fc76ec6b1954de2c55595f0ec0a7614238b9ebf69618/yarl-1.18.3-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c017a3b6df3a1bd45b9fa49a0f54005e53fbcad16633870104b66fa1a30a29d8", size = 354007, upload-time = "2024-12-01T20:33:17.518Z" }, - { url = "https://files.pythonhosted.org/packages/af/ba/1865d85212351ad160f19fb99808acf23aab9a0f8ff31c8c9f1b4d671fc9/yarl-1.18.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:75674776d96d7b851b6498f17824ba17849d790a44d282929c42dbb77d4f17ae", size = 344145, upload-time = "2024-12-01T20:33:20.071Z" }, - { url = "https://files.pythonhosted.org/packages/94/cb/5c3e975d77755d7b3d5193e92056b19d83752ea2da7ab394e22260a7b824/yarl-1.18.3-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ccaa3a4b521b780a7e771cc336a2dba389a0861592bbce09a476190bb0c8b4b3", size = 336133, upload-time = "2024-12-01T20:33:22.515Z" }, - { url = "https://files.pythonhosted.org/packages/19/89/b77d3fd249ab52a5c40859815765d35c91425b6bb82e7427ab2f78f5ff55/yarl-1.18.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:2d06d3005e668744e11ed80812e61efd77d70bb7f03e33c1598c301eea20efbb", size = 347967, upload-time = "2024-12-01T20:33:24.139Z" }, - { url = "https://files.pythonhosted.org/packages/35/bd/f6b7630ba2cc06c319c3235634c582a6ab014d52311e7d7c22f9518189b5/yarl-1.18.3-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:9d41beda9dc97ca9ab0b9888cb71f7539124bc05df02c0cff6e5acc5a19dcc6e", size = 346397, upload-time = "2024-12-01T20:33:26.205Z" }, - { url = "https://files.pythonhosted.org/packages/18/1a/0b4e367d5a72d1f095318344848e93ea70da728118221f84f1bf6c1e39e7/yarl-1.18.3-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:ba23302c0c61a9999784e73809427c9dbedd79f66a13d84ad1b1943802eaaf59", size = 350206, upload-time = "2024-12-01T20:33:27.83Z" }, - { url = "https://files.pythonhosted.org/packages/b5/cf/320fff4367341fb77809a2d8d7fe75b5d323a8e1b35710aafe41fdbf327b/yarl-1.18.3-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:6748dbf9bfa5ba1afcc7556b71cda0d7ce5f24768043a02a58846e4a443d808d", size = 362089, upload-time = "2024-12-01T20:33:29.565Z" }, - { url = "https://files.pythonhosted.org/packages/57/cf/aadba261d8b920253204085268bad5e8cdd86b50162fcb1b10c10834885a/yarl-1.18.3-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:0b0cad37311123211dc91eadcb322ef4d4a66008d3e1bdc404808992260e1a0e", size = 366267, upload-time = "2024-12-01T20:33:31.449Z" }, - { url = "https://files.pythonhosted.org/packages/54/58/fb4cadd81acdee6dafe14abeb258f876e4dd410518099ae9a35c88d8097c/yarl-1.18.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:0fb2171a4486bb075316ee754c6d8382ea6eb8b399d4ec62fde2b591f879778a", size = 359141, upload-time = "2024-12-01T20:33:33.79Z" }, - { url = "https://files.pythonhosted.org/packages/9a/7a/4c571597589da4cd5c14ed2a0b17ac56ec9ee7ee615013f74653169e702d/yarl-1.18.3-cp311-cp311-win32.whl", hash = "sha256:61b1a825a13bef4a5f10b1885245377d3cd0bf87cba068e1d9a88c2ae36880e1", size = 84402, upload-time = "2024-12-01T20:33:35.689Z" }, - { url = "https://files.pythonhosted.org/packages/ae/7b/8600250b3d89b625f1121d897062f629883c2f45339623b69b1747ec65fa/yarl-1.18.3-cp311-cp311-win_amd64.whl", hash = "sha256:b9d60031cf568c627d028239693fd718025719c02c9f55df0a53e587aab951b5", size = 91030, upload-time = "2024-12-01T20:33:37.511Z" }, - { url = "https://files.pythonhosted.org/packages/33/85/bd2e2729752ff4c77338e0102914897512e92496375e079ce0150a6dc306/yarl-1.18.3-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:1dd4bdd05407ced96fed3d7f25dbbf88d2ffb045a0db60dbc247f5b3c5c25d50", size = 142644, upload-time = "2024-12-01T20:33:39.204Z" }, - { url = "https://files.pythonhosted.org/packages/ff/74/1178322cc0f10288d7eefa6e4a85d8d2e28187ccab13d5b844e8b5d7c88d/yarl-1.18.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:7c33dd1931a95e5d9a772d0ac5e44cac8957eaf58e3c8da8c1414de7dd27c576", size = 94962, upload-time = "2024-12-01T20:33:40.808Z" }, - { url = "https://files.pythonhosted.org/packages/be/75/79c6acc0261e2c2ae8a1c41cf12265e91628c8c58ae91f5ff59e29c0787f/yarl-1.18.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:25b411eddcfd56a2f0cd6a384e9f4f7aa3efee14b188de13048c25b5e91f1640", size = 92795, upload-time = "2024-12-01T20:33:42.322Z" }, - { url = "https://files.pythonhosted.org/packages/6b/32/927b2d67a412c31199e83fefdce6e645247b4fb164aa1ecb35a0f9eb2058/yarl-1.18.3-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:436c4fc0a4d66b2badc6c5fc5ef4e47bb10e4fd9bf0c79524ac719a01f3607c2", size = 332368, upload-time = "2024-12-01T20:33:43.956Z" }, - { url = "https://files.pythonhosted.org/packages/19/e5/859fca07169d6eceeaa4fde1997c91d8abde4e9a7c018e371640c2da2b71/yarl-1.18.3-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e35ef8683211db69ffe129a25d5634319a677570ab6b2eba4afa860f54eeaf75", size = 342314, upload-time = "2024-12-01T20:33:46.046Z" }, - { url = "https://files.pythonhosted.org/packages/08/75/76b63ccd91c9e03ab213ef27ae6add2e3400e77e5cdddf8ed2dbc36e3f21/yarl-1.18.3-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:84b2deecba4a3f1a398df819151eb72d29bfeb3b69abb145a00ddc8d30094512", size = 341987, upload-time = "2024-12-01T20:33:48.352Z" }, - { url = "https://files.pythonhosted.org/packages/1a/e1/a097d5755d3ea8479a42856f51d97eeff7a3a7160593332d98f2709b3580/yarl-1.18.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:00e5a1fea0fd4f5bfa7440a47eff01d9822a65b4488f7cff83155a0f31a2ecba", size = 336914, upload-time = "2024-12-01T20:33:50.875Z" }, - { url = "https://files.pythonhosted.org/packages/0b/42/e1b4d0e396b7987feceebe565286c27bc085bf07d61a59508cdaf2d45e63/yarl-1.18.3-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d0e883008013c0e4aef84dcfe2a0b172c4d23c2669412cf5b3371003941f72bb", size = 325765, upload-time = "2024-12-01T20:33:52.641Z" }, - { url = "https://files.pythonhosted.org/packages/7e/18/03a5834ccc9177f97ca1bbb245b93c13e58e8225276f01eedc4cc98ab820/yarl-1.18.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:5a3f356548e34a70b0172d8890006c37be92995f62d95a07b4a42e90fba54272", size = 344444, upload-time = "2024-12-01T20:33:54.395Z" }, - { url = "https://files.pythonhosted.org/packages/c8/03/a713633bdde0640b0472aa197b5b86e90fbc4c5bc05b727b714cd8a40e6d/yarl-1.18.3-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:ccd17349166b1bee6e529b4add61727d3f55edb7babbe4069b5764c9587a8cc6", size = 340760, upload-time = "2024-12-01T20:33:56.286Z" }, - { url = "https://files.pythonhosted.org/packages/eb/99/f6567e3f3bbad8fd101886ea0276c68ecb86a2b58be0f64077396cd4b95e/yarl-1.18.3-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:b958ddd075ddba5b09bb0be8a6d9906d2ce933aee81100db289badbeb966f54e", size = 346484, upload-time = "2024-12-01T20:33:58.375Z" }, - { url = "https://files.pythonhosted.org/packages/8e/a9/84717c896b2fc6cb15bd4eecd64e34a2f0a9fd6669e69170c73a8b46795a/yarl-1.18.3-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:c7d79f7d9aabd6011004e33b22bc13056a3e3fb54794d138af57f5ee9d9032cb", size = 359864, upload-time = "2024-12-01T20:34:00.22Z" }, - { url = "https://files.pythonhosted.org/packages/1e/2e/d0f5f1bef7ee93ed17e739ec8dbcb47794af891f7d165fa6014517b48169/yarl-1.18.3-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:4891ed92157e5430874dad17b15eb1fda57627710756c27422200c52d8a4e393", size = 364537, upload-time = "2024-12-01T20:34:03.54Z" }, - { url = "https://files.pythonhosted.org/packages/97/8a/568d07c5d4964da5b02621a517532adb8ec5ba181ad1687191fffeda0ab6/yarl-1.18.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ce1af883b94304f493698b00d0f006d56aea98aeb49d75ec7d98cd4a777e9285", size = 357861, upload-time = "2024-12-01T20:34:05.73Z" }, - { url = "https://files.pythonhosted.org/packages/7d/e3/924c3f64b6b3077889df9a1ece1ed8947e7b61b0a933f2ec93041990a677/yarl-1.18.3-cp312-cp312-win32.whl", hash = "sha256:f91c4803173928a25e1a55b943c81f55b8872f0018be83e3ad4938adffb77dd2", size = 84097, upload-time = "2024-12-01T20:34:07.664Z" }, - { url = "https://files.pythonhosted.org/packages/34/45/0e055320daaabfc169b21ff6174567b2c910c45617b0d79c68d7ab349b02/yarl-1.18.3-cp312-cp312-win_amd64.whl", hash = "sha256:7e2ee16578af3b52ac2f334c3b1f92262f47e02cc6193c598502bd46f5cd1477", size = 90399, upload-time = "2024-12-01T20:34:09.61Z" }, - { url = "https://files.pythonhosted.org/packages/30/c7/c790513d5328a8390be8f47be5d52e141f78b66c6c48f48d241ca6bd5265/yarl-1.18.3-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:90adb47ad432332d4f0bc28f83a5963f426ce9a1a8809f5e584e704b82685dcb", size = 140789, upload-time = "2024-12-01T20:34:11.414Z" }, - { url = "https://files.pythonhosted.org/packages/30/aa/a2f84e93554a578463e2edaaf2300faa61c8701f0898725842c704ba5444/yarl-1.18.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:913829534200eb0f789d45349e55203a091f45c37a2674678744ae52fae23efa", size = 94144, upload-time = "2024-12-01T20:34:13.485Z" }, - { url = "https://files.pythonhosted.org/packages/c6/fc/d68d8f83714b221a85ce7866832cba36d7c04a68fa6a960b908c2c84f325/yarl-1.18.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:ef9f7768395923c3039055c14334ba4d926f3baf7b776c923c93d80195624782", size = 91974, upload-time = "2024-12-01T20:34:15.234Z" }, - { url = "https://files.pythonhosted.org/packages/56/4e/d2563d8323a7e9a414b5b25341b3942af5902a2263d36d20fb17c40411e2/yarl-1.18.3-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:88a19f62ff30117e706ebc9090b8ecc79aeb77d0b1f5ec10d2d27a12bc9f66d0", size = 333587, upload-time = "2024-12-01T20:34:17.358Z" }, - { url = "https://files.pythonhosted.org/packages/25/c9/cfec0bc0cac8d054be223e9f2c7909d3e8442a856af9dbce7e3442a8ec8d/yarl-1.18.3-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e17c9361d46a4d5addf777c6dd5eab0715a7684c2f11b88c67ac37edfba6c482", size = 344386, upload-time = "2024-12-01T20:34:19.842Z" }, - { url = "https://files.pythonhosted.org/packages/ab/5d/4c532190113b25f1364d25f4c319322e86232d69175b91f27e3ebc2caf9a/yarl-1.18.3-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1a74a13a4c857a84a845505fd2d68e54826a2cd01935a96efb1e9d86c728e186", size = 345421, upload-time = "2024-12-01T20:34:21.975Z" }, - { url = "https://files.pythonhosted.org/packages/23/d1/6cdd1632da013aa6ba18cee4d750d953104a5e7aac44e249d9410a972bf5/yarl-1.18.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:41f7ce59d6ee7741af71d82020346af364949314ed3d87553763a2df1829cc58", size = 339384, upload-time = "2024-12-01T20:34:24.717Z" }, - { url = "https://files.pythonhosted.org/packages/9a/c4/6b3c39bec352e441bd30f432cda6ba51681ab19bb8abe023f0d19777aad1/yarl-1.18.3-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f52a265001d830bc425f82ca9eabda94a64a4d753b07d623a9f2863fde532b53", size = 326689, upload-time = "2024-12-01T20:34:26.886Z" }, - { url = "https://files.pythonhosted.org/packages/23/30/07fb088f2eefdc0aa4fc1af4e3ca4eb1a3aadd1ce7d866d74c0f124e6a85/yarl-1.18.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:82123d0c954dc58db301f5021a01854a85bf1f3bb7d12ae0c01afc414a882ca2", size = 345453, upload-time = "2024-12-01T20:34:29.605Z" }, - { url = "https://files.pythonhosted.org/packages/63/09/d54befb48f9cd8eec43797f624ec37783a0266855f4930a91e3d5c7717f8/yarl-1.18.3-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:2ec9bbba33b2d00999af4631a3397d1fd78290c48e2a3e52d8dd72db3a067ac8", size = 341872, upload-time = "2024-12-01T20:34:31.454Z" }, - { url = "https://files.pythonhosted.org/packages/91/26/fd0ef9bf29dd906a84b59f0cd1281e65b0c3e08c6aa94b57f7d11f593518/yarl-1.18.3-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:fbd6748e8ab9b41171bb95c6142faf068f5ef1511935a0aa07025438dd9a9bc1", size = 347497, upload-time = "2024-12-01T20:34:34.004Z" }, - { url = "https://files.pythonhosted.org/packages/d9/b5/14ac7a256d0511b2ac168d50d4b7d744aea1c1aa20c79f620d1059aab8b2/yarl-1.18.3-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:877d209b6aebeb5b16c42cbb377f5f94d9e556626b1bfff66d7b0d115be88d0a", size = 359981, upload-time = "2024-12-01T20:34:36.624Z" }, - { url = "https://files.pythonhosted.org/packages/ca/b3/d493221ad5cbd18bc07e642894030437e405e1413c4236dd5db6e46bcec9/yarl-1.18.3-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:b464c4ab4bfcb41e3bfd3f1c26600d038376c2de3297760dfe064d2cb7ea8e10", size = 366229, upload-time = "2024-12-01T20:34:38.657Z" }, - { url = "https://files.pythonhosted.org/packages/04/56/6a3e2a5d9152c56c346df9b8fb8edd2c8888b1e03f96324d457e5cf06d34/yarl-1.18.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8d39d351e7faf01483cc7ff7c0213c412e38e5a340238826be7e0e4da450fdc8", size = 360383, upload-time = "2024-12-01T20:34:40.501Z" }, - { url = "https://files.pythonhosted.org/packages/fd/b7/4b3c7c7913a278d445cc6284e59b2e62fa25e72758f888b7a7a39eb8423f/yarl-1.18.3-cp313-cp313-win32.whl", hash = "sha256:61ee62ead9b68b9123ec24bc866cbef297dd266175d53296e2db5e7f797f902d", size = 310152, upload-time = "2024-12-01T20:34:42.814Z" }, - { url = "https://files.pythonhosted.org/packages/f5/d5/688db678e987c3e0fb17867970700b92603cadf36c56e5fb08f23e822a0c/yarl-1.18.3-cp313-cp313-win_amd64.whl", hash = "sha256:578e281c393af575879990861823ef19d66e2b1d0098414855dd367e234f5b3c", size = 315723, upload-time = "2024-12-01T20:34:44.699Z" }, - { url = "https://files.pythonhosted.org/packages/f5/4b/a06e0ec3d155924f77835ed2d167ebd3b211a7b0853da1cf8d8414d784ef/yarl-1.18.3-py3-none-any.whl", hash = "sha256:b57f4f58099328dfb26c6a771d09fb20dbbae81d20cfb66141251ea063bd101b", size = 45109, upload-time = "2024-12-01T20:35:20.834Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/3c/fb/efaa23fa4e45537b827620f04cf8f3cd658b76642205162e072703a5b963/yarl-1.20.1.tar.gz", hash = "sha256:d017a4997ee50c91fd5466cef416231bb82177b93b029906cefc542ce14c35ac", size = 186428, upload-time = "2025-06-10T00:46:09.923Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/cb/65/7fed0d774abf47487c64be14e9223749468922817b5e8792b8a64792a1bb/yarl-1.20.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:6032e6da6abd41e4acda34d75a816012717000fa6839f37124a47fcefc49bec4", size = 132910, upload-time = "2025-06-10T00:42:31.108Z" }, + { url = "https://files.pythonhosted.org/packages/8a/7b/988f55a52da99df9e56dc733b8e4e5a6ae2090081dc2754fc8fd34e60aa0/yarl-1.20.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2c7b34d804b8cf9b214f05015c4fee2ebe7ed05cf581e7192c06555c71f4446a", size = 90644, upload-time = "2025-06-10T00:42:33.851Z" }, + { url = "https://files.pythonhosted.org/packages/f7/de/30d98f03e95d30c7e3cc093759982d038c8833ec2451001d45ef4854edc1/yarl-1.20.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:0c869f2651cc77465f6cd01d938d91a11d9ea5d798738c1dc077f3de0b5e5fed", size = 89322, upload-time = "2025-06-10T00:42:35.688Z" }, + { url = "https://files.pythonhosted.org/packages/e0/7a/f2f314f5ebfe9200724b0b748de2186b927acb334cf964fd312eb86fc286/yarl-1.20.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:62915e6688eb4d180d93840cda4110995ad50c459bf931b8b3775b37c264af1e", size = 323786, upload-time = "2025-06-10T00:42:37.817Z" }, + { url = "https://files.pythonhosted.org/packages/15/3f/718d26f189db96d993d14b984ce91de52e76309d0fd1d4296f34039856aa/yarl-1.20.1-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:41ebd28167bc6af8abb97fec1a399f412eec5fd61a3ccbe2305a18b84fb4ca73", size = 319627, upload-time = "2025-06-10T00:42:39.937Z" }, + { url = "https://files.pythonhosted.org/packages/a5/76/8fcfbf5fa2369157b9898962a4a7d96764b287b085b5b3d9ffae69cdefd1/yarl-1.20.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:21242b4288a6d56f04ea193adde174b7e347ac46ce6bc84989ff7c1b1ecea84e", size = 339149, upload-time = "2025-06-10T00:42:42.627Z" }, + { url = "https://files.pythonhosted.org/packages/3c/95/d7fc301cc4661785967acc04f54a4a42d5124905e27db27bb578aac49b5c/yarl-1.20.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:bea21cdae6c7eb02ba02a475f37463abfe0a01f5d7200121b03e605d6a0439f8", size = 333327, upload-time = "2025-06-10T00:42:44.842Z" }, + { url = "https://files.pythonhosted.org/packages/65/94/e21269718349582eee81efc5c1c08ee71c816bfc1585b77d0ec3f58089eb/yarl-1.20.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1f8a891e4a22a89f5dde7862994485e19db246b70bb288d3ce73a34422e55b23", size = 326054, upload-time = "2025-06-10T00:42:47.149Z" }, + { url = "https://files.pythonhosted.org/packages/32/ae/8616d1f07853704523519f6131d21f092e567c5af93de7e3e94b38d7f065/yarl-1.20.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:dd803820d44c8853a109a34e3660e5a61beae12970da479cf44aa2954019bf70", size = 315035, upload-time = "2025-06-10T00:42:48.852Z" }, + { url = "https://files.pythonhosted.org/packages/48/aa/0ace06280861ef055855333707db5e49c6e3a08840a7ce62682259d0a6c0/yarl-1.20.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b982fa7f74c80d5c0c7b5b38f908971e513380a10fecea528091405f519b9ebb", size = 338962, upload-time = "2025-06-10T00:42:51.024Z" }, + { url = "https://files.pythonhosted.org/packages/20/52/1e9d0e6916f45a8fb50e6844f01cb34692455f1acd548606cbda8134cd1e/yarl-1.20.1-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:33f29ecfe0330c570d997bcf1afd304377f2e48f61447f37e846a6058a4d33b2", size = 335399, upload-time = "2025-06-10T00:42:53.007Z" }, + { url = "https://files.pythonhosted.org/packages/f2/65/60452df742952c630e82f394cd409de10610481d9043aa14c61bf846b7b1/yarl-1.20.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:835ab2cfc74d5eb4a6a528c57f05688099da41cf4957cf08cad38647e4a83b30", size = 338649, upload-time = "2025-06-10T00:42:54.964Z" }, + { url = "https://files.pythonhosted.org/packages/7b/f5/6cd4ff38dcde57a70f23719a838665ee17079640c77087404c3d34da6727/yarl-1.20.1-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:46b5e0ccf1943a9a6e766b2c2b8c732c55b34e28be57d8daa2b3c1d1d4009309", size = 358563, upload-time = "2025-06-10T00:42:57.28Z" }, + { url = "https://files.pythonhosted.org/packages/d1/90/c42eefd79d0d8222cb3227bdd51b640c0c1d0aa33fe4cc86c36eccba77d3/yarl-1.20.1-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:df47c55f7d74127d1b11251fe6397d84afdde0d53b90bedb46a23c0e534f9d24", size = 357609, upload-time = "2025-06-10T00:42:59.055Z" }, + { url = "https://files.pythonhosted.org/packages/03/c8/cea6b232cb4617514232e0f8a718153a95b5d82b5290711b201545825532/yarl-1.20.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:76d12524d05841276b0e22573f28d5fbcb67589836772ae9244d90dd7d66aa13", size = 350224, upload-time = "2025-06-10T00:43:01.248Z" }, + { url = "https://files.pythonhosted.org/packages/ce/a3/eaa0ab9712f1f3d01faf43cf6f1f7210ce4ea4a7e9b28b489a2261ca8db9/yarl-1.20.1-cp310-cp310-win32.whl", hash = "sha256:6c4fbf6b02d70e512d7ade4b1f998f237137f1417ab07ec06358ea04f69134f8", size = 81753, upload-time = "2025-06-10T00:43:03.486Z" }, + { url = "https://files.pythonhosted.org/packages/8f/34/e4abde70a9256465fe31c88ed02c3f8502b7b5dead693a4f350a06413f28/yarl-1.20.1-cp310-cp310-win_amd64.whl", hash = "sha256:aef6c4d69554d44b7f9d923245f8ad9a707d971e6209d51279196d8e8fe1ae16", size = 86817, upload-time = "2025-06-10T00:43:05.231Z" }, + { url = "https://files.pythonhosted.org/packages/b1/18/893b50efc2350e47a874c5c2d67e55a0ea5df91186b2a6f5ac52eff887cd/yarl-1.20.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:47ee6188fea634bdfaeb2cc420f5b3b17332e6225ce88149a17c413c77ff269e", size = 133833, upload-time = "2025-06-10T00:43:07.393Z" }, + { url = "https://files.pythonhosted.org/packages/89/ed/b8773448030e6fc47fa797f099ab9eab151a43a25717f9ac043844ad5ea3/yarl-1.20.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:d0f6500f69e8402d513e5eedb77a4e1818691e8f45e6b687147963514d84b44b", size = 91070, upload-time = "2025-06-10T00:43:09.538Z" }, + { url = "https://files.pythonhosted.org/packages/e3/e3/409bd17b1e42619bf69f60e4f031ce1ccb29bd7380117a55529e76933464/yarl-1.20.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7a8900a42fcdaad568de58887c7b2f602962356908eedb7628eaf6021a6e435b", size = 89818, upload-time = "2025-06-10T00:43:11.575Z" }, + { url = "https://files.pythonhosted.org/packages/f8/77/64d8431a4d77c856eb2d82aa3de2ad6741365245a29b3a9543cd598ed8c5/yarl-1.20.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bad6d131fda8ef508b36be3ece16d0902e80b88ea7200f030a0f6c11d9e508d4", size = 347003, upload-time = "2025-06-10T00:43:14.088Z" }, + { url = "https://files.pythonhosted.org/packages/8d/d2/0c7e4def093dcef0bd9fa22d4d24b023788b0a33b8d0088b51aa51e21e99/yarl-1.20.1-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:df018d92fe22aaebb679a7f89fe0c0f368ec497e3dda6cb81a567610f04501f1", size = 336537, upload-time = "2025-06-10T00:43:16.431Z" }, + { url = "https://files.pythonhosted.org/packages/f0/f3/fc514f4b2cf02cb59d10cbfe228691d25929ce8f72a38db07d3febc3f706/yarl-1.20.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8f969afbb0a9b63c18d0feecf0db09d164b7a44a053e78a7d05f5df163e43833", size = 362358, upload-time = "2025-06-10T00:43:18.704Z" }, + { url = "https://files.pythonhosted.org/packages/ea/6d/a313ac8d8391381ff9006ac05f1d4331cee3b1efaa833a53d12253733255/yarl-1.20.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:812303eb4aa98e302886ccda58d6b099e3576b1b9276161469c25803a8db277d", size = 357362, upload-time = "2025-06-10T00:43:20.888Z" }, + { url = "https://files.pythonhosted.org/packages/00/70/8f78a95d6935a70263d46caa3dd18e1f223cf2f2ff2037baa01a22bc5b22/yarl-1.20.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:98c4a7d166635147924aa0bf9bfe8d8abad6fffa6102de9c99ea04a1376f91e8", size = 348979, upload-time = "2025-06-10T00:43:23.169Z" }, + { url = "https://files.pythonhosted.org/packages/cb/05/42773027968968f4f15143553970ee36ead27038d627f457cc44bbbeecf3/yarl-1.20.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:12e768f966538e81e6e7550f9086a6236b16e26cd964cf4df35349970f3551cf", size = 337274, upload-time = "2025-06-10T00:43:27.111Z" }, + { url = "https://files.pythonhosted.org/packages/05/be/665634aa196954156741ea591d2f946f1b78ceee8bb8f28488bf28c0dd62/yarl-1.20.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:fe41919b9d899661c5c28a8b4b0acf704510b88f27f0934ac7a7bebdd8938d5e", size = 363294, upload-time = "2025-06-10T00:43:28.96Z" }, + { url = "https://files.pythonhosted.org/packages/eb/90/73448401d36fa4e210ece5579895731f190d5119c4b66b43b52182e88cd5/yarl-1.20.1-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:8601bc010d1d7780592f3fc1bdc6c72e2b6466ea34569778422943e1a1f3c389", size = 358169, upload-time = "2025-06-10T00:43:30.701Z" }, + { url = "https://files.pythonhosted.org/packages/c3/b0/fce922d46dc1eb43c811f1889f7daa6001b27a4005587e94878570300881/yarl-1.20.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:daadbdc1f2a9033a2399c42646fbd46da7992e868a5fe9513860122d7fe7a73f", size = 362776, upload-time = "2025-06-10T00:43:32.51Z" }, + { url = "https://files.pythonhosted.org/packages/f1/0d/b172628fce039dae8977fd22caeff3eeebffd52e86060413f5673767c427/yarl-1.20.1-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:03aa1e041727cb438ca762628109ef1333498b122e4c76dd858d186a37cec845", size = 381341, upload-time = "2025-06-10T00:43:34.543Z" }, + { url = "https://files.pythonhosted.org/packages/6b/9b/5b886d7671f4580209e855974fe1cecec409aa4a89ea58b8f0560dc529b1/yarl-1.20.1-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:642980ef5e0fa1de5fa96d905c7e00cb2c47cb468bfcac5a18c58e27dbf8d8d1", size = 379988, upload-time = "2025-06-10T00:43:36.489Z" }, + { url = "https://files.pythonhosted.org/packages/73/be/75ef5fd0fcd8f083a5d13f78fd3f009528132a1f2a1d7c925c39fa20aa79/yarl-1.20.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:86971e2795584fe8c002356d3b97ef6c61862720eeff03db2a7c86b678d85b3e", size = 371113, upload-time = "2025-06-10T00:43:38.592Z" }, + { url = "https://files.pythonhosted.org/packages/50/4f/62faab3b479dfdcb741fe9e3f0323e2a7d5cd1ab2edc73221d57ad4834b2/yarl-1.20.1-cp311-cp311-win32.whl", hash = "sha256:597f40615b8d25812f14562699e287f0dcc035d25eb74da72cae043bb884d773", size = 81485, upload-time = "2025-06-10T00:43:41.038Z" }, + { url = "https://files.pythonhosted.org/packages/f0/09/d9c7942f8f05c32ec72cd5c8e041c8b29b5807328b68b4801ff2511d4d5e/yarl-1.20.1-cp311-cp311-win_amd64.whl", hash = "sha256:26ef53a9e726e61e9cd1cda6b478f17e350fb5800b4bd1cd9fe81c4d91cfeb2e", size = 86686, upload-time = "2025-06-10T00:43:42.692Z" }, + { url = "https://files.pythonhosted.org/packages/5f/9a/cb7fad7d73c69f296eda6815e4a2c7ed53fc70c2f136479a91c8e5fbdb6d/yarl-1.20.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:bdcc4cd244e58593a4379fe60fdee5ac0331f8eb70320a24d591a3be197b94a9", size = 133667, upload-time = "2025-06-10T00:43:44.369Z" }, + { url = "https://files.pythonhosted.org/packages/67/38/688577a1cb1e656e3971fb66a3492501c5a5df56d99722e57c98249e5b8a/yarl-1.20.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b29a2c385a5f5b9c7d9347e5812b6f7ab267193c62d282a540b4fc528c8a9d2a", size = 91025, upload-time = "2025-06-10T00:43:46.295Z" }, + { url = "https://files.pythonhosted.org/packages/50/ec/72991ae51febeb11a42813fc259f0d4c8e0507f2b74b5514618d8b640365/yarl-1.20.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:1112ae8154186dfe2de4732197f59c05a83dc814849a5ced892b708033f40dc2", size = 89709, upload-time = "2025-06-10T00:43:48.22Z" }, + { url = "https://files.pythonhosted.org/packages/99/da/4d798025490e89426e9f976702e5f9482005c548c579bdae792a4c37769e/yarl-1.20.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:90bbd29c4fe234233f7fa2b9b121fb63c321830e5d05b45153a2ca68f7d310ee", size = 352287, upload-time = "2025-06-10T00:43:49.924Z" }, + { url = "https://files.pythonhosted.org/packages/1a/26/54a15c6a567aac1c61b18aa0f4b8aa2e285a52d547d1be8bf48abe2b3991/yarl-1.20.1-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:680e19c7ce3710ac4cd964e90dad99bf9b5029372ba0c7cbfcd55e54d90ea819", size = 345429, upload-time = "2025-06-10T00:43:51.7Z" }, + { url = "https://files.pythonhosted.org/packages/d6/95/9dcf2386cb875b234353b93ec43e40219e14900e046bf6ac118f94b1e353/yarl-1.20.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4a979218c1fdb4246a05efc2cc23859d47c89af463a90b99b7c56094daf25a16", size = 365429, upload-time = "2025-06-10T00:43:53.494Z" }, + { url = "https://files.pythonhosted.org/packages/91/b2/33a8750f6a4bc224242a635f5f2cff6d6ad5ba651f6edcccf721992c21a0/yarl-1.20.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:255b468adf57b4a7b65d8aad5b5138dce6a0752c139965711bdcb81bc370e1b6", size = 363862, upload-time = "2025-06-10T00:43:55.766Z" }, + { url = "https://files.pythonhosted.org/packages/98/28/3ab7acc5b51f4434b181b0cee8f1f4b77a65919700a355fb3617f9488874/yarl-1.20.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a97d67108e79cfe22e2b430d80d7571ae57d19f17cda8bb967057ca8a7bf5bfd", size = 355616, upload-time = "2025-06-10T00:43:58.056Z" }, + { url = "https://files.pythonhosted.org/packages/36/a3/f666894aa947a371724ec7cd2e5daa78ee8a777b21509b4252dd7bd15e29/yarl-1.20.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8570d998db4ddbfb9a590b185a0a33dbf8aafb831d07a5257b4ec9948df9cb0a", size = 339954, upload-time = "2025-06-10T00:43:59.773Z" }, + { url = "https://files.pythonhosted.org/packages/f1/81/5f466427e09773c04219d3450d7a1256138a010b6c9f0af2d48565e9ad13/yarl-1.20.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:97c75596019baae7c71ccf1d8cc4738bc08134060d0adfcbe5642f778d1dca38", size = 365575, upload-time = "2025-06-10T00:44:02.051Z" }, + { url = "https://files.pythonhosted.org/packages/2e/e3/e4b0ad8403e97e6c9972dd587388940a032f030ebec196ab81a3b8e94d31/yarl-1.20.1-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:1c48912653e63aef91ff988c5432832692ac5a1d8f0fb8a33091520b5bbe19ef", size = 365061, upload-time = "2025-06-10T00:44:04.196Z" }, + { url = "https://files.pythonhosted.org/packages/ac/99/b8a142e79eb86c926f9f06452eb13ecb1bb5713bd01dc0038faf5452e544/yarl-1.20.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:4c3ae28f3ae1563c50f3d37f064ddb1511ecc1d5584e88c6b7c63cf7702a6d5f", size = 364142, upload-time = "2025-06-10T00:44:06.527Z" }, + { url = "https://files.pythonhosted.org/packages/34/f2/08ed34a4a506d82a1a3e5bab99ccd930a040f9b6449e9fd050320e45845c/yarl-1.20.1-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:c5e9642f27036283550f5f57dc6156c51084b458570b9d0d96100c8bebb186a8", size = 381894, upload-time = "2025-06-10T00:44:08.379Z" }, + { url = "https://files.pythonhosted.org/packages/92/f8/9a3fbf0968eac704f681726eff595dce9b49c8a25cd92bf83df209668285/yarl-1.20.1-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:2c26b0c49220d5799f7b22c6838409ee9bc58ee5c95361a4d7831f03cc225b5a", size = 383378, upload-time = "2025-06-10T00:44:10.51Z" }, + { url = "https://files.pythonhosted.org/packages/af/85/9363f77bdfa1e4d690957cd39d192c4cacd1c58965df0470a4905253b54f/yarl-1.20.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:564ab3d517e3d01c408c67f2e5247aad4019dcf1969982aba3974b4093279004", size = 374069, upload-time = "2025-06-10T00:44:12.834Z" }, + { url = "https://files.pythonhosted.org/packages/35/99/9918c8739ba271dcd935400cff8b32e3cd319eaf02fcd023d5dcd487a7c8/yarl-1.20.1-cp312-cp312-win32.whl", hash = "sha256:daea0d313868da1cf2fac6b2d3a25c6e3a9e879483244be38c8e6a41f1d876a5", size = 81249, upload-time = "2025-06-10T00:44:14.731Z" }, + { url = "https://files.pythonhosted.org/packages/eb/83/5d9092950565481b413b31a23e75dd3418ff0a277d6e0abf3729d4d1ce25/yarl-1.20.1-cp312-cp312-win_amd64.whl", hash = "sha256:48ea7d7f9be0487339828a4de0360d7ce0efc06524a48e1810f945c45b813698", size = 86710, upload-time = "2025-06-10T00:44:16.716Z" }, + { url = "https://files.pythonhosted.org/packages/8a/e1/2411b6d7f769a07687acee88a062af5833cf1966b7266f3d8dfb3d3dc7d3/yarl-1.20.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:0b5ff0fbb7c9f1b1b5ab53330acbfc5247893069e7716840c8e7d5bb7355038a", size = 131811, upload-time = "2025-06-10T00:44:18.933Z" }, + { url = "https://files.pythonhosted.org/packages/b2/27/584394e1cb76fb771371770eccad35de400e7b434ce3142c2dd27392c968/yarl-1.20.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:14f326acd845c2b2e2eb38fb1346c94f7f3b01a4f5c788f8144f9b630bfff9a3", size = 90078, upload-time = "2025-06-10T00:44:20.635Z" }, + { url = "https://files.pythonhosted.org/packages/bf/9a/3246ae92d4049099f52d9b0fe3486e3b500e29b7ea872d0f152966fc209d/yarl-1.20.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:f60e4ad5db23f0b96e49c018596707c3ae89f5d0bd97f0ad3684bcbad899f1e7", size = 88748, upload-time = "2025-06-10T00:44:22.34Z" }, + { url = "https://files.pythonhosted.org/packages/a3/25/35afe384e31115a1a801fbcf84012d7a066d89035befae7c5d4284df1e03/yarl-1.20.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:49bdd1b8e00ce57e68ba51916e4bb04461746e794e7c4d4bbc42ba2f18297691", size = 349595, upload-time = "2025-06-10T00:44:24.314Z" }, + { url = "https://files.pythonhosted.org/packages/28/2d/8aca6cb2cabc8f12efcb82749b9cefecbccfc7b0384e56cd71058ccee433/yarl-1.20.1-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:66252d780b45189975abfed839616e8fd2dbacbdc262105ad7742c6ae58f3e31", size = 342616, upload-time = "2025-06-10T00:44:26.167Z" }, + { url = "https://files.pythonhosted.org/packages/0b/e9/1312633d16b31acf0098d30440ca855e3492d66623dafb8e25b03d00c3da/yarl-1.20.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:59174e7332f5d153d8f7452a102b103e2e74035ad085f404df2e40e663a22b28", size = 361324, upload-time = "2025-06-10T00:44:27.915Z" }, + { url = "https://files.pythonhosted.org/packages/bc/a0/688cc99463f12f7669eec7c8acc71ef56a1521b99eab7cd3abb75af887b0/yarl-1.20.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e3968ec7d92a0c0f9ac34d5ecfd03869ec0cab0697c91a45db3fbbd95fe1b653", size = 359676, upload-time = "2025-06-10T00:44:30.041Z" }, + { url = "https://files.pythonhosted.org/packages/af/44/46407d7f7a56e9a85a4c207724c9f2c545c060380718eea9088f222ba697/yarl-1.20.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d1a4fbb50e14396ba3d375f68bfe02215d8e7bc3ec49da8341fe3157f59d2ff5", size = 352614, upload-time = "2025-06-10T00:44:32.171Z" }, + { url = "https://files.pythonhosted.org/packages/b1/91/31163295e82b8d5485d31d9cf7754d973d41915cadce070491778d9c9825/yarl-1.20.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:11a62c839c3a8eac2410e951301309426f368388ff2f33799052787035793b02", size = 336766, upload-time = "2025-06-10T00:44:34.494Z" }, + { url = "https://files.pythonhosted.org/packages/b4/8e/c41a5bc482121f51c083c4c2bcd16b9e01e1cf8729e380273a952513a21f/yarl-1.20.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:041eaa14f73ff5a8986b4388ac6bb43a77f2ea09bf1913df7a35d4646db69e53", size = 364615, upload-time = "2025-06-10T00:44:36.856Z" }, + { url = "https://files.pythonhosted.org/packages/e3/5b/61a3b054238d33d70ea06ebba7e58597891b71c699e247df35cc984ab393/yarl-1.20.1-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:377fae2fef158e8fd9d60b4c8751387b8d1fb121d3d0b8e9b0be07d1b41e83dc", size = 360982, upload-time = "2025-06-10T00:44:39.141Z" }, + { url = "https://files.pythonhosted.org/packages/df/a3/6a72fb83f8d478cb201d14927bc8040af901811a88e0ff2da7842dd0ed19/yarl-1.20.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:1c92f4390e407513f619d49319023664643d3339bd5e5a56a3bebe01bc67ec04", size = 369792, upload-time = "2025-06-10T00:44:40.934Z" }, + { url = "https://files.pythonhosted.org/packages/7c/af/4cc3c36dfc7c077f8dedb561eb21f69e1e9f2456b91b593882b0b18c19dc/yarl-1.20.1-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d25ddcf954df1754ab0f86bb696af765c5bfaba39b74095f27eececa049ef9a4", size = 382049, upload-time = "2025-06-10T00:44:42.854Z" }, + { url = "https://files.pythonhosted.org/packages/19/3a/e54e2c4752160115183a66dc9ee75a153f81f3ab2ba4bf79c3c53b33de34/yarl-1.20.1-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:909313577e9619dcff8c31a0ea2aa0a2a828341d92673015456b3ae492e7317b", size = 384774, upload-time = "2025-06-10T00:44:45.275Z" }, + { url = "https://files.pythonhosted.org/packages/9c/20/200ae86dabfca89060ec6447649f219b4cbd94531e425e50d57e5f5ac330/yarl-1.20.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:793fd0580cb9664548c6b83c63b43c477212c0260891ddf86809e1c06c8b08f1", size = 374252, upload-time = "2025-06-10T00:44:47.31Z" }, + { url = "https://files.pythonhosted.org/packages/83/75/11ee332f2f516b3d094e89448da73d557687f7d137d5a0f48c40ff211487/yarl-1.20.1-cp313-cp313-win32.whl", hash = "sha256:468f6e40285de5a5b3c44981ca3a319a4b208ccc07d526b20b12aeedcfa654b7", size = 81198, upload-time = "2025-06-10T00:44:49.164Z" }, + { url = "https://files.pythonhosted.org/packages/ba/ba/39b1ecbf51620b40ab402b0fc817f0ff750f6d92712b44689c2c215be89d/yarl-1.20.1-cp313-cp313-win_amd64.whl", hash = "sha256:495b4ef2fea40596bfc0affe3837411d6aa3371abcf31aac0ccc4bdd64d4ef5c", size = 86346, upload-time = "2025-06-10T00:44:51.182Z" }, + { url = "https://files.pythonhosted.org/packages/43/c7/669c52519dca4c95153c8ad96dd123c79f354a376346b198f438e56ffeb4/yarl-1.20.1-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:f60233b98423aab21d249a30eb27c389c14929f47be8430efa7dbd91493a729d", size = 138826, upload-time = "2025-06-10T00:44:52.883Z" }, + { url = "https://files.pythonhosted.org/packages/6a/42/fc0053719b44f6ad04a75d7f05e0e9674d45ef62f2d9ad2c1163e5c05827/yarl-1.20.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:6f3eff4cc3f03d650d8755c6eefc844edde99d641d0dcf4da3ab27141a5f8ddf", size = 93217, upload-time = "2025-06-10T00:44:54.658Z" }, + { url = "https://files.pythonhosted.org/packages/4f/7f/fa59c4c27e2a076bba0d959386e26eba77eb52ea4a0aac48e3515c186b4c/yarl-1.20.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:69ff8439d8ba832d6bed88af2c2b3445977eba9a4588b787b32945871c2444e3", size = 92700, upload-time = "2025-06-10T00:44:56.784Z" }, + { url = "https://files.pythonhosted.org/packages/2f/d4/062b2f48e7c93481e88eff97a6312dca15ea200e959f23e96d8ab898c5b8/yarl-1.20.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3cf34efa60eb81dd2645a2e13e00bb98b76c35ab5061a3989c7a70f78c85006d", size = 347644, upload-time = "2025-06-10T00:44:59.071Z" }, + { url = "https://files.pythonhosted.org/packages/89/47/78b7f40d13c8f62b499cc702fdf69e090455518ae544c00a3bf4afc9fc77/yarl-1.20.1-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:8e0fe9364ad0fddab2688ce72cb7a8e61ea42eff3c7caeeb83874a5d479c896c", size = 323452, upload-time = "2025-06-10T00:45:01.605Z" }, + { url = "https://files.pythonhosted.org/packages/eb/2b/490d3b2dc66f52987d4ee0d3090a147ea67732ce6b4d61e362c1846d0d32/yarl-1.20.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8f64fbf81878ba914562c672024089e3401974a39767747691c65080a67b18c1", size = 346378, upload-time = "2025-06-10T00:45:03.946Z" }, + { url = "https://files.pythonhosted.org/packages/66/ad/775da9c8a94ce925d1537f939a4f17d782efef1f973039d821cbe4bcc211/yarl-1.20.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f6342d643bf9a1de97e512e45e4b9560a043347e779a173250824f8b254bd5ce", size = 353261, upload-time = "2025-06-10T00:45:05.992Z" }, + { url = "https://files.pythonhosted.org/packages/4b/23/0ed0922b47a4f5c6eb9065d5ff1e459747226ddce5c6a4c111e728c9f701/yarl-1.20.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:56dac5f452ed25eef0f6e3c6a066c6ab68971d96a9fb441791cad0efba6140d3", size = 335987, upload-time = "2025-06-10T00:45:08.227Z" }, + { url = "https://files.pythonhosted.org/packages/3e/49/bc728a7fe7d0e9336e2b78f0958a2d6b288ba89f25a1762407a222bf53c3/yarl-1.20.1-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c7d7f497126d65e2cad8dc5f97d34c27b19199b6414a40cb36b52f41b79014be", size = 329361, upload-time = "2025-06-10T00:45:10.11Z" }, + { url = "https://files.pythonhosted.org/packages/93/8f/b811b9d1f617c83c907e7082a76e2b92b655400e61730cd61a1f67178393/yarl-1.20.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:67e708dfb8e78d8a19169818eeb5c7a80717562de9051bf2413aca8e3696bf16", size = 346460, upload-time = "2025-06-10T00:45:12.055Z" }, + { url = "https://files.pythonhosted.org/packages/70/fd/af94f04f275f95da2c3b8b5e1d49e3e79f1ed8b6ceb0f1664cbd902773ff/yarl-1.20.1-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:595c07bc79af2494365cc96ddeb772f76272364ef7c80fb892ef9d0649586513", size = 334486, upload-time = "2025-06-10T00:45:13.995Z" }, + { url = "https://files.pythonhosted.org/packages/84/65/04c62e82704e7dd0a9b3f61dbaa8447f8507655fd16c51da0637b39b2910/yarl-1.20.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:7bdd2f80f4a7df852ab9ab49484a4dee8030023aa536df41f2d922fd57bf023f", size = 342219, upload-time = "2025-06-10T00:45:16.479Z" }, + { url = "https://files.pythonhosted.org/packages/91/95/459ca62eb958381b342d94ab9a4b6aec1ddec1f7057c487e926f03c06d30/yarl-1.20.1-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:c03bfebc4ae8d862f853a9757199677ab74ec25424d0ebd68a0027e9c639a390", size = 350693, upload-time = "2025-06-10T00:45:18.399Z" }, + { url = "https://files.pythonhosted.org/packages/a6/00/d393e82dd955ad20617abc546a8f1aee40534d599ff555ea053d0ec9bf03/yarl-1.20.1-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:344d1103e9c1523f32a5ed704d576172d2cabed3122ea90b1d4e11fe17c66458", size = 355803, upload-time = "2025-06-10T00:45:20.677Z" }, + { url = "https://files.pythonhosted.org/packages/9e/ed/c5fb04869b99b717985e244fd93029c7a8e8febdfcffa06093e32d7d44e7/yarl-1.20.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:88cab98aa4e13e1ade8c141daeedd300a4603b7132819c484841bb7af3edce9e", size = 341709, upload-time = "2025-06-10T00:45:23.221Z" }, + { url = "https://files.pythonhosted.org/packages/24/fd/725b8e73ac2a50e78a4534ac43c6addf5c1c2d65380dd48a9169cc6739a9/yarl-1.20.1-cp313-cp313t-win32.whl", hash = "sha256:b121ff6a7cbd4abc28985b6028235491941b9fe8fe226e6fdc539c977ea1739d", size = 86591, upload-time = "2025-06-10T00:45:25.793Z" }, + { url = "https://files.pythonhosted.org/packages/94/c3/b2e9f38bc3e11191981d57ea08cab2166e74ea770024a646617c9cddd9f6/yarl-1.20.1-cp313-cp313t-win_amd64.whl", hash = "sha256:541d050a355bbbc27e55d906bc91cb6fe42f96c01413dd0f4ed5a5240513874f", size = 93003, upload-time = "2025-06-10T00:45:27.752Z" }, + { url = "https://files.pythonhosted.org/packages/b4/2d/2345fce04cfd4bee161bf1e7d9cdc702e3e16109021035dbb24db654a622/yarl-1.20.1-py3-none-any.whl", hash = "sha256:83b8eb083fe4683c6115795d9fc1cfaf2cbbefb19b3a1cb68f6527460f483a77", size = 46542, upload-time = "2025-06-10T00:46:07.521Z" }, ] From 4824862675d4abf93a9b673c59938bff9f528c10 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Tue, 12 Aug 2025 16:05:33 +0000 Subject: [PATCH 21/32] New configs, sampling sweep, modified sampler 1. New configs/slurm scripts for: high noise level, noise check experiment, and full swarm data. 2. Included equilibration mechanism in the baoab and aboba memory samplers. 3. A new wandb sweep script for sweeping over delta and friction. --- .../sample_enhanced_conditioning.yaml | 6 +- .../sample_enhanced_conditioning_sweep.yaml | 6 +- .../train_enhanced_position_conditioner.yaml | 4 +- ...temporal_conditioner_multimeasurement.yaml | 55 ++++++ .../train_enhanced_standard_jamun.yaml | 2 +- configs/sweep_delta_friction.yaml | 18 ++ debug_model3_simple.sbatch | 56 ------ scripts/slurm/sweep.sh | 6 +- scripts/slurm/train_model_comparison.sh | 81 +++++++++ .../train_model_comparison_full_swarm.sh | 83 +++++++++ .../train_model_comparison_high_noise.sh | 82 +++++++++ scripts/slurm/train_noise_check.sh | 165 ++++++++++++++++++ .../spatiotemporal_pretrained.yaml | 1 + src/jamun/model/arch/e3conv_conditional.py | 4 +- src/jamun/model/arch/spatiotemporal.py | 6 +- src/jamun/model/denoiser_multimeasurement.py | 2 +- .../sampling/mcmc/functional/_splitting.py | 12 +- 17 files changed, 511 insertions(+), 78 deletions(-) create mode 100644 configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner_multimeasurement.yaml create mode 100644 configs/sweep_delta_friction.yaml delete mode 100644 debug_model3_simple.sbatch create mode 100644 scripts/slurm/train_model_comparison.sh create mode 100644 scripts/slurm/train_model_comparison_full_swarm.sh create mode 100644 scripts/slurm/train_model_comparison_high_noise.sh create mode 100755 scripts/slurm/train_noise_check.sh diff --git a/configs/experiment/sample_enhanced_conditioning.yaml b/configs/experiment/sample_enhanced_conditioning.yaml index aa39cfe..14625de 100644 --- a/configs/experiment/sample_enhanced_conditioning.yaml +++ b/configs/experiment/sample_enhanced_conditioning.yaml @@ -33,14 +33,14 @@ init_datasets: # conditioner: # N_structures: ${init_datasets.total_lag_time} -num_sampling_steps_per_batch: 1000 +num_sampling_steps_per_batch: 10000 num_batches: 1 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: "sule-shashank/jamun/i0rd0uoa" +wandb_train_run_path: "sule-shashank/jamun/l1n5tg48" # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last @@ -67,7 +67,7 @@ eval_dataset: subsample: 10 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 1 + max_datasets: 10 label_override: "ALA_ALA" # Override ALL callbacks to use eval_dataset for metrics computation diff --git a/configs/experiment/sample_enhanced_conditioning_sweep.yaml b/configs/experiment/sample_enhanced_conditioning_sweep.yaml index 223d0a9..a537caa 100644 --- a/configs/experiment/sample_enhanced_conditioning_sweep.yaml +++ b/configs/experiment/sample_enhanced_conditioning_sweep.yaml @@ -26,15 +26,15 @@ init_datasets: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 1 + max_datasets: 10 label_override: "ALA_ALA" # model: # conditioner: # N_structures: ${init_datasets.total_lag_time} -num_sampling_steps_per_batch: 1 -num_batches: 1 +num_sampling_steps_per_batch: 1000 +num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true diff --git a/configs/experiment/train_enhanced_position_conditioner.yaml b/configs/experiment/train_enhanced_position_conditioner.yaml index 1e02bfe..3a38362 100644 --- a/configs/experiment/train_enhanced_position_conditioner.yaml +++ b/configs/experiment/train_enhanced_position_conditioner.yaml @@ -20,7 +20,7 @@ data: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 1 + # max_datasets: 1 val: _target_: jamun.data.parse_datasets_from_directory @@ -30,7 +30,7 @@ data: subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} - max_datasets: ${data.datamodule.datasets.train.max_datasets} + # max_datasets: ${data.datamodule.datasets.train.max_datasets} model: sigma_distribution: diff --git a/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner_multimeasurement.yaml b/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner_multimeasurement.yaml new file mode 100644 index 0000000..bfcaef3 --- /dev/null +++ b/configs/experiment/train_enhanced_pretrained_spatiotemporal_conditioner_multimeasurement.yaml @@ -0,0 +1,55 @@ +# @package _global_ +defaults: + - override /model: denoiser_multimeasurement + - override /model/arch: e3conv_conditional_spatiotemporal + - override /model/conditioner: spatiotemporal_pretrained + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 4 # Reduced batch size due to increased model complexity + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: 5 + lag_subsample_rate: 1 + # max_datasets: 2 # Increased for more training data + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + # max_datasets: 1 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 1 + max_radius: 1.0 + optim: + lr: 0.002 # Slightly reduced learning rate for stability + +trainer: + val_check_interval: 0.5 + max_epochs: 50 # Increased due to model complexity + # devices: 1 + # gradient_clip_val: 1.0 # Add gradient clipping for stability + +logger: + wandb: + group: enhanced_sampling_conditioner_comparison + notes: "SpatioTemporalConditioner on enhanced sampling data - processes temporal sequences through spatial and temporal modules" + tags: ["spatiotemporal_conditioner", "enhanced_sampling", "transformer", "e3conv"] \ No newline at end of file diff --git a/configs/experiment/train_enhanced_standard_jamun.yaml b/configs/experiment/train_enhanced_standard_jamun.yaml index f05a21d..cc8fd01 100644 --- a/configs/experiment/train_enhanced_standard_jamun.yaml +++ b/configs/experiment/train_enhanced_standard_jamun.yaml @@ -35,7 +35,7 @@ model: _target_: jamun.distributions.ConstantSigma sigma: 0.04 arch: - n_layers: 2 + n_layers: 4 max_radius: 1000.0 optim: lr: 0.002 diff --git a/configs/sweep_delta_friction.yaml b/configs/sweep_delta_friction.yaml new file mode 100644 index 0000000..d4e077d --- /dev/null +++ b/configs/sweep_delta_friction.yaml @@ -0,0 +1,18 @@ +program: jamun_sample +method: grid +project: jamun +name: conditional_vs_self +metric: + name: Jenson-Shannon Divergence vs. Number of Samples for Predicted Trajectory joined + goal: minimize +parameters: + delta: + values: [0.017889, 0.026833, 0.040000, 0.059665, 0.089443] + friction: + values: [2.52572864, 1.2552661, 0.71334989, 0.36384343, 0.10536052] + +command: + - ${program} + - "--config-dir=configs" + - "experiment=sample_enhanced_conditioning_sweep" + - ${args_no_hyphens} \ No newline at end of file diff --git a/debug_model3_simple.sbatch b/debug_model3_simple.sbatch deleted file mode 100644 index 15f07d5..0000000 --- a/debug_model3_simple.sbatch +++ /dev/null @@ -1,56 +0,0 @@ -#!/usr/bin/env bash - -#SBATCH --job-name=debug_model3 -#SBATCH --array=1-2 -#SBATCH --output=logs/debug_model3_%A_%a.out -#SBATCH --error=logs/debug_model3_%A_%a.err -#SBATCH --time=1:00:00 -#SBATCH --partition=gpu -#SBATCH --gres=gpu:1 -#SBATCH --cpus-per-task=2 -#SBATCH --mem=16G - -mkdir -p logs - -eval "$(conda shell.bash hook)" -conda activate jamun - -set -eux - -echo "SLURM_JOB_ID = ${SLURM_JOB_ID}" -echo "SLURM_ARRAY_TASK_ID = ${SLURM_ARRAY_TASK_ID}" -echo "hostname = $(hostname)" - -export HYDRA_FULL_ERROR=1 - -cd /homefs/home/sules/jamun - -# Test basic command first -echo "Testing jamun_train command..." -jamun_train --help - -echo "Testing config validation..." -jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - --cfg job - -# Now test with minimal overrides -if [ $SLURM_ARRAY_TASK_ID -eq 1 ]; then - echo "Testing with betas=[0.9,0.9]" - jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - trainer.max_epochs=1 \ - data.datamodule.datasets.train.num_frames=10 \ - 'model.optim.betas=[0.9,0.9]' \ - --dry-run -else - echo "Testing with betas=[0.9,0.999]" - jamun_train --config-dir=src/jamun/hydra_config \ - experiment=ala_ala_denoiser_experiment_model3 \ - trainer.max_epochs=1 \ - data.datamodule.datasets.train.num_frames=10 \ - 'model.optim.betas=[0.9,0.999]' \ - --dry-run -fi - -echo "Debug test completed for task $SLURM_ARRAY_TASK_ID" \ No newline at end of file diff --git a/scripts/slurm/sweep.sh b/scripts/slurm/sweep.sh index 907db55..0e272e5 100644 --- a/scripts/slurm/sweep.sh +++ b/scripts/slurm/sweep.sh @@ -1,13 +1,13 @@ #!/usr/bin/env bash -#SBATCH --partition gpu2 -#SBATCH --nodes 1 +#SBATCH --partition=gpu2 +#SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node #SBATCH --gpus-per-node=1 # Assign one GPU to each agent #SBATCH --cpus-per-task=12 #SBATCH --time 3-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array 0-15 +#SBATCH --array 0-4 # Check if a Sweep ID is provided as an argument export JAMUN_ROOT_PATH=/data2/sules/jamun-conditional-runs diff --git a/scripts/slurm/train_model_comparison.sh b/scripts/slurm/train_model_comparison.sh new file mode 100644 index 0000000..ce660f8 --- /dev/null +++ b/scripts/slurm/train_model_comparison.sh @@ -0,0 +1,81 @@ +#!/usr/bin/env bash + +#SBATCH --partition=gpu3 +#SBATCH --job-name=model_comparison +#SBATCH --qos=preempt +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-1 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# Define configurations for each job +case ${SLURM_ARRAY_TASK_ID} in + 0) + echo "Job 0: Standard JAMUN" + CONFIG="train_enhanced_standard_jamun" + OVERRIDES="" + ;; + 1) + echo "Job 1: Position conditioner" + CONFIG="train_enhanced_position_conditioner" + OVERRIDES="" + ;; + 2) + echo "Job 2: Self conditioner" + CONFIG="train_enhanced_self_conditioner" + OVERRIDES="" + ;; + 3) + echo "Job 3: Spatiotemporal conditioner with mean pooling and trainable pretrained denoiser" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + ;; + 4) + echo "Job 4: Spatiotemporal conditioner with mean pooling, trainable pretrained denoiser, and ones temporal encoding" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + ;; + *) + echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" + exit 1 + ;; +esac + +# Build the command with base config +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add overrides if any +if [ -n "$OVERRIDES" ]; then + CMD="$CMD $OVERRIDES" +fi + +# Add common training overrides +CMD="$CMD ++trainer.max_epochs=100" +CMD="$CMD ++logger.wandb.group=model_comparison" + +# Add dataset overrides for debugging (quick completion) +# CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" +# CMD="$CMD ++data.datamodule.datasets.val.max_datasets=1" + +# Add job-specific wandb tags +WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison]" + +echo "Running command: $CMD" +exec $CMD diff --git a/scripts/slurm/train_model_comparison_full_swarm.sh b/scripts/slurm/train_model_comparison_full_swarm.sh new file mode 100644 index 0000000..a841ea7 --- /dev/null +++ b/scripts/slurm/train_model_comparison_full_swarm.sh @@ -0,0 +1,83 @@ +#!/usr/bin/env bash + +#SBATCH --partition=gpu3 +#SBATCH --job-name=model_comparison +#SBATCH --qos=preempt +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=2-4 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# Define configurations for each job +case ${SLURM_ARRAY_TASK_ID} in + 0) + echo "Job 0: Standard JAMUN" + CONFIG="train_enhanced_standard_jamun" + OVERRIDES="" + ;; + 1) + echo "Job 1: Position conditioner" + CONFIG="train_enhanced_position_conditioner" + OVERRIDES="" + ;; + 2) + echo "Job 2: Self conditioner" + CONFIG="train_enhanced_self_conditioner" + OVERRIDES="" + ;; + 3) + echo "Job 3: Spatiotemporal conditioner with mean pooling and trainable pretrained denoiser" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + ;; + 4) + echo "Job 4: Spatiotemporal conditioner with mean pooling, trainable pretrained denoiser, and ones temporal encoding" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + ;; + *) + echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" + exit 1 + ;; +esac + +# Build the command with base config +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add overrides if any +if [ -n "$OVERRIDES" ]; then + CMD="$CMD $OVERRIDES" +fi + +# Add common training overrides +CMD="$CMD ++trainer.max_epochs=100" +CMD="$CMD ++logger.wandb.group=model_comparison_full_swarm" +CMD="$CMD ++data.datamodule.datasets.train.root=/data2/sules/ALA_ALA_enhanced_full_swarm/train" +CMD="$CMD ++data.datamodule.datasets.val.root=/data2/sules/ALA_ALA_enhanced_full_swarm/val" + +# Add dataset overrides for debugging (quick completion) +# CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" +# CMD="$CMD ++data.datamodule.datasets.val.max_datasets=1" + +# Add job-specific wandb tags +WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison]" + +echo "Running command: $CMD" +exec $CMD diff --git a/scripts/slurm/train_model_comparison_high_noise.sh b/scripts/slurm/train_model_comparison_high_noise.sh new file mode 100644 index 0000000..689b242 --- /dev/null +++ b/scripts/slurm/train_model_comparison_high_noise.sh @@ -0,0 +1,82 @@ +#!/usr/bin/env bash + +#SBATCH --partition=gpu3 +#SBATCH --job-name=model_comparison +#SBATCH --qos=preempt +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=2-4 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# Define configurations for each job +case ${SLURM_ARRAY_TASK_ID} in + 0) + echo "Job 0: Standard JAMUN" + CONFIG="train_enhanced_standard_jamun" + OVERRIDES="" + ;; + 1) + echo "Job 1: Position conditioner" + CONFIG="train_enhanced_position_conditioner" + OVERRIDES="" + ;; + 2) + echo "Job 2: Self conditioner" + CONFIG="train_enhanced_self_conditioner" + OVERRIDES="" + ;; + 3) + echo "Job 3: Spatiotemporal conditioner with mean pooling and trainable pretrained denoiser" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + ;; + 4) + echo "Job 4: Spatiotemporal conditioner with mean pooling, trainable pretrained denoiser, and ones temporal encoding" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + ;; + *) + echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" + exit 1 + ;; +esac + +# Build the command with base config +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add overrides if any +if [ -n "$OVERRIDES" ]; then + CMD="$CMD $OVERRIDES" +fi + +# Add common training overrides +CMD="$CMD ++trainer.max_epochs=100" +CMD="$CMD ++logger.wandb.group=model_comparison_high_noise" +CMD="$CMD ++model.sigma_distribution.sigma=0.08" + +# Add dataset overrides for debugging (quick completion) +# CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" +# CMD="$CMD ++data.datamodule.datasets.val.max_datasets=1" + +# Add job-specific wandb tags +WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison]" + +echo "Running command: $CMD" +exec $CMD diff --git a/scripts/slurm/train_noise_check.sh b/scripts/slurm/train_noise_check.sh new file mode 100755 index 0000000..428f96a --- /dev/null +++ b/scripts/slurm/train_noise_check.sh @@ -0,0 +1,165 @@ +#!/usr/bin/env bash +# +# Noise check experiment script +# +# Implements 5 out of 7 requested model configurations for m=2,3,4,5,6,7,8,9,10: +# 1. Standard JAMUN with noise level sigma/sqrt(m) +# 2. Spatiotemporal with temporal embedding (pretrained denoiser, trainable=true) +# 3. Spatiotemporal with ones embedding (pretrained denoiser, trainable=true) +# 4. Spatiotemporal with temporal embedding, repeated position dataset +# 5. Spatiotemporal with ones embedding, repeated position dataset +# +# NOT IMPLEMENTED (requires custom dataset class): +# 6. Spatiotemporal with temporal embedding, random lag times +# 7. Spatiotemporal with ones embedding, random lag times +# +# Total jobs: 9 m-values Ɨ 5 models = 45 jobs (array 0-44) + +#SBATCH --partition=gpu3 +#SBATCH --job-name=noise_check +#SBATCH --qos=preempt +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-44 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# Define experiment parameters +# m values: 2, 3, 4, 5, 6, 7, 8, 9, 10 (9 values) +# Model types: 5 types (random lag times require custom implementation) +# Total combinations: 9 * 5 = 45 jobs (0-44) + +M_VALUES=(2 3 4 5 6 7 8 9 10) +MODEL_TYPES=( + "standard_jamun" + "spatiotemporal_temporal_embedding" + "spatiotemporal_ones_embedding" + "spatiotemporal_temporal_embedding_repeated_pos" + "spatiotemporal_ones_embedding_repeated_pos" +) + +# Calculate m and model indices +NUM_MODELS=5 +M_INDEX=$((SLURM_ARRAY_TASK_ID / NUM_MODELS)) +MODEL_INDEX=$((SLURM_ARRAY_TASK_ID % NUM_MODELS)) + +M=${M_VALUES[$M_INDEX]} +MODEL_TYPE=${MODEL_TYPES[$MODEL_INDEX]} + +echo "Job ${SLURM_ARRAY_TASK_ID}: M=${M}, Model=${MODEL_TYPE}" + +# Calculate noise level: sigma / sqrt(m) where base sigma = 0.04 +BASE_SIGMA=0.04 +NOISE_LEVEL=$(python3 -c "import math; print(${BASE_SIGMA} / math.sqrt(${M}))") + +echo "Noise level: ${NOISE_LEVEL}" + +# Configure base parameters based on model type +case ${MODEL_TYPE} in + "standard_jamun") + echo "Model 1: Standard JAMUN with noise level sigma/sqrt(m)" + CONFIG="train_enhanced_standard_jamun" + OVERRIDES="++model.sigma_distribution.sigma=${NOISE_LEVEL}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" + WANDB_TAG="standard_jamun_m${M}" + ;; + "spatiotemporal_temporal_embedding") + echo "Model 2: Spatiotemporal with temporal embedding (pretrained denoiser, trainable=true)" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" + + WANDB_TAG="spatiotemporal_temporal_m${M}" + ;; + "spatiotemporal_ones_embedding") + echo "Model 3: Spatiotemporal with ones embedding (pretrained denoiser, trainable=true)" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones" + OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" + + WANDB_TAG="spatiotemporal_ones_m${M}" + ;; + "spatiotemporal_temporal_embedding_repeated_pos") + echo "Model 4: Spatiotemporal with temporal embedding, repeated position dataset" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train._target_=jamun.data.parse_repeated_position_datasets_from_directory" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val._target_=jamun.data.parse_repeated_position_datasets_from_directory" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" + + WANDB_TAG="spatiotemporal_temporal_repeated_m${M}" + ;; + "spatiotemporal_ones_embedding_repeated_pos") + echo "Model 5: Spatiotemporal with ones embedding, repeated position dataset" + CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones" + OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train._target_=jamun.data.parse_repeated_position_datasets_from_directory" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val._target_=jamun.data.parse_repeated_position_datasets_from_directory" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" + + WANDB_TAG="spatiotemporal_ones_repeated_m${M}" + ;; + *) + echo "Unknown model type: ${MODEL_TYPE}" + exit 1 + ;; +esac + +# Build the command with base config +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add overrides +if [ -n "$OVERRIDES" ]; then + CMD="$CMD $OVERRIDES" +fi + +# Add common training overrides +CMD="$CMD ++trainer.max_epochs=1" +CMD="$CMD ++logger.wandb.group=noise_check_experiment" + +# Add job-specific wandb tags and run name +WANDB_RUN_NAME="noise_check_${WANDB_TAG}" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},noise_check,m_${M}]" +CMD="$CMD ++logger.wandb.name=${WANDB_RUN_NAME}" + +# Add notes about the experiment +WANDB_NOTES="Noise check experiment: ${MODEL_TYPE} with m=${M}" +if [[ ${MODEL_TYPE} == "standard_jamun" ]]; then + WANDB_NOTES="${WANDB_NOTES}, noise_level=${NOISE_LEVEL}" +fi +CMD="$CMD ++logger.wandb.notes=\"${WANDB_NOTES}\"" + +echo "Running command: $CMD" +exec $CMD diff --git a/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml index 391f84d..920b565 100644 --- a/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml +++ b/src/jamun/hydra_config/model/conditioner/spatiotemporal_pretrained.yaml @@ -25,6 +25,7 @@ spatiotemporal_model: irreps_hidden: "120x0e + 32x1e" irreps_sh: "1x0e + 1x1e" irreps_node_attr: "1x1e" # Match spatial module output + irreps_node_attr_temporal: "3x0e" num_layers: 2 edge_attr_dim: 24 num_attention_heads: 1 diff --git a/src/jamun/model/arch/e3conv_conditional.py b/src/jamun/model/arch/e3conv_conditional.py index dcbc82d..757df8a 100644 --- a/src/jamun/model/arch/e3conv_conditional.py +++ b/src/jamun/model/arch/e3conv_conditional.py @@ -381,7 +381,7 @@ def __init__( # Set up input attribute handling self.input_attr_irreps = o3.Irreps(input_attr_irreps) - + self.input_attr_irreps_dim = self.input_attr_irreps.dim # Create input irrep aggregator to combine node_attr with input_attr # Combined irreps: node_attr irreps + input_attr irreps combined_irreps = self.irreps_hidden + self.input_attr_irreps @@ -445,7 +445,7 @@ def forward( # Combine node_attr with spatial features (input_attr) # Validate spatial features shape - expected_dim = self.input_attr_irreps.dim + expected_dim = self.input_attr_irreps_dim if pos_features.shape[-1] != expected_dim: raise ValueError( f"Expected spatial features to have dimension {expected_dim}, " diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index 7e5612b..959f0f1 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -243,7 +243,7 @@ def forward( edge_sh = self.sh(edge_vec) # Compute edge attributes: radial and temporal - if self.radial_edge_attr_encoding_function is not "ones": + if self.radial_edge_attr_encoding_function != "ones": radial_edge_attr = e3nn.math.soft_one_hot_linspace( edge_vec.norm(dim=1), 0.0, @@ -265,7 +265,7 @@ def forward( # Temporal edge attributes from temporal_position differences temporal_edge_vec = temporal_position[src] - temporal_position[dst] - if self.edge_attr_temporal_encoding_function is not "ones": + if self.edge_attr_temporal_encoding_function != "ones": temporal_edge_attr = e3nn.math.soft_one_hot_linspace( temporal_edge_vec.abs(), # Use absolute difference 0.0, @@ -297,7 +297,7 @@ def forward( # Process node attributes with temporal gating # Concatenate node_attr with temporal_position (scalar) - if self.node_attr_temporal_encoding_function is not "ones": + if self.node_attr_temporal_encoding_function != "ones": temporal_position = e3nn.math.soft_one_hot_linspace( temporal_position, # Use absolute difference 0.0, # time always starts at 0 diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py index 0c694bb..681d035 100644 --- a/src/jamun/model/denoiser_multimeasurement.py +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -713,7 +713,7 @@ def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): if self.automatic_optimization: return self._automatic_step(batch, "train") else: - print(f"Manual optimization enabled for training step {batch_idx}.") + # print(f"Manual optimization enabled for training step {batch_idx}.") return self._manual_step(batch, "train") def validation_step(self, batch: torch_geometric.data.Batch, batch_idx: int): diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index 7c5f130..19a4877 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -187,6 +187,7 @@ def aboba_memory( save_trajectory=False, save_every_n_steps=1, burn_in_steps=0, + history_update_frequency=1, verbose=False, cpu_offload=False, delta: float = 1.0, @@ -232,8 +233,9 @@ def aboba_memory( score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) y_hist_traj.append(list(y_hist)) - y_hist.pop(-1) - y_hist.insert(0, y_current) + if i % history_update_frequency == 0: + y_hist.pop(-1) + y_hist.insert(0, y_current) return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj @@ -247,6 +249,7 @@ def baoab_memory( save_trajectory=False, save_every_n_steps=1, burn_in_steps=0, + history_update_frequency=1, verbose=False, cpu_offload=False, delta: float = 1.0, @@ -286,8 +289,9 @@ def baoab_memory( R = torch.randn_like(y) vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R y = y + (delta / 2) * vhat - y_hist.pop(-1) # remove the last element of the history - y_hist.insert(0, y_current) # present point is the first element of the history + if i % history_update_frequency == 0: + y_hist.pop(-1) # remove the last element of the history + y_hist.insert(0, y_current) # present point is the first element of the history psi, orig_score = score_fn_processed(y, y_hist=y_hist) v = vhat + (delta / 2) * psi From c14a900315731012c96544ef5f3fa5165733ed16 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Mon, 18 Aug 2025 17:36:29 +0000 Subject: [PATCH 22/32] conditional-gen --- bond_length_issues_conditional_traj.png | Bin 0 -> 79531 bytes .../sample_enhanced_conditioning.yaml | 14 +- .../sample_enhanced_conditioning_sweep.yaml | 14 +- .../experiment/sample_enhanced_standard.yaml | 12 +- configs/experiment/train_capped_2AA.yaml | 45 ++++-- ...train_capped_2AA_position_conditioner.yaml | 75 +++++++++ .../train_capped_2AA_self_conditioner.yaml | 75 +++++++++ ...capped_2AA_spatiotemporal_conditioner.yaml | 71 ++++++++ .../train_enhanced_position_conditioner.yaml | 6 +- ...n_enhanced_spatiotemporal_conditioner.yaml | 6 +- .../experiment/train_test_single_shape.yaml | 25 +-- .../train_test_single_shape_conditional.yaml | 36 +++-- ...ngle_shape_spatiotemporal_conditioner.yaml | 73 +++++++++ configs/sweep_delta_friction.yaml | 5 +- scripts/slurm/run_train_noise_check.sh | 63 ++++++++ scripts/slurm/sweep.sh | 6 +- scripts/slurm/train_capped_2AA_comparison.sh | 85 ++++++++++ scripts/slurm/train_graph_type_comparison.sh | 138 ++++++++++++++++ scripts/slurm/train_model_comparison.sh | 21 ++- .../train_model_comparison_full_swarm.sh | 21 ++- .../train_model_comparison_high_noise.sh | 26 +-- scripts/slurm/train_noise_check.sh | 153 ++++++++---------- .../callbacks/model_checkpoint.yaml | 3 +- src/jamun/hydra_config/model/arch/e3conv.yaml | 2 +- .../model/arch/e3conv_conditional.yaml | 2 +- .../e3conv_conditional_spatiotemporal.yaml | 2 +- .../model/arch/spatiotemporal.yaml | 2 +- .../model/conditioner/spatiotemporal.yaml | 5 +- src/jamun/model/arch/spatiotemporal.py | 123 ++++++++++---- src/jamun/sampling/mcmc/_splitting.py | 4 + .../sampling/mcmc/functional/_splitting.py | 48 +++--- 31 files changed, 922 insertions(+), 239 deletions(-) create mode 100644 bond_length_issues_conditional_traj.png create mode 100644 configs/experiment/train_capped_2AA_position_conditioner.yaml create mode 100644 configs/experiment/train_capped_2AA_self_conditioner.yaml create mode 100644 configs/experiment/train_capped_2AA_spatiotemporal_conditioner.yaml create mode 100644 configs/experiment/train_test_single_shape_spatiotemporal_conditioner.yaml create mode 100755 scripts/slurm/run_train_noise_check.sh create mode 100755 scripts/slurm/train_capped_2AA_comparison.sh create mode 100755 scripts/slurm/train_graph_type_comparison.sh diff --git a/bond_length_issues_conditional_traj.png b/bond_length_issues_conditional_traj.png new file mode 100644 index 0000000000000000000000000000000000000000..1063ab390a3dad19ffd63ec95303474e432a707b GIT binary patch literal 79531 zcmeFZ_g~KKA3uDyNNGt%(n6(>P^c8ij)s=B2Z?rR545Kwl@yiH(4xIHj8LMbJwzHx zX|MZvxUTDSe}A|i_djso9*^tuiOxFT=lgveuh(U_Ng^EHsC4WTK z@xgG1gZ6oi<&|TKCwYgNsD%$$Tk_;3q;X8A=jVm0r`k7~`6lG17pCT@vt^q#7Ur1c z1Z=WaZP!|9aks=cd}6**Z+t^I?-6<`+0kYZ(T@c`I_FD#zipk4ajA`<=cD!e?@!R( zQwA#!%m4f5F1tYHf8TkVBzEK2f8RJBylMykzkl|aDd@BP`zK7e`3sy(`Z@Rd~fByYT3qmD#69R2ChDbeAvxkn^7%s895M(e>c&jtgt@ zEIMw+#_AaO-nqPznd|VA(-E1M>$pprQZ+KTC9ef+M0E?v3eUlSp+ zY4>S+T&QTa+mNA|nO2hd9iBK@kC1gcwFXLb7w0C$b&pzGTPGC!_N{f24O0}COziX@HS#}+t{?&mOJ(uA5Qsgt9cH!lU z#uT+%1qBDC4;>OSZ{7F!$&=*9cUQPG@_xKjk@5I5z{Y*Z`SH3uo5cs;%bgcy#C(`G z`P|udrsBhg)wo?1nFf^&t*WOE7Vi-eVSIVH?)cXiCVY|adqzfrA9^fZ_#t*RJR-t- z+nIZh)YG&wGc!G7V>fH%n0d=CFPu;7?0$8NFYF|*SGhmC;JzgRr=IR^PrLy&B_T0U z{N89+8Bdy44xZ5F9Xoa?6({SllbzR^SF&(Fx?C4$Wo6|V5y7FAXAz9m zW!ka)^UJfIzCIce5s?E255B6aQ@rZ(!@QY6LqkJTQ}dB>;&I6z)f!mW9hESolMYWey#-fey=-+i%gaP*n0 z$7*ir4M%-eQ&CoL+#^3SIhVWO7<+}CjqQeL=?pc8h`N`zHyzjJ&GcC<=$^L2Zhz2 z`&pJPuG_wSJEy(Fa0?X=4^Mt!VMTj8*ON1m$J=tvvs|WsOsTu>H;%OEZ7H&Uy5QK%(`|P26MZ&Q_G}%iZ z4r80)S$l2Vdnu_YL#HUWbqc@1_joX8JRaDFeWt92Mn(pgE=|=&t>3BT)%PZpywPT< z>s2>sR_?a!l-4eF-H3!>DUXf|j^Nsvl zw$vw|7gSAEXWDPkUXBHu>9Oqo;Gpda4q??#ni+}@W$2lhjul-W;f*+*W!@$cv~ll> zr%#`55HqZjGFm5BjyINa_~n(Cw;xsHmPUrI0dB@XOV*ZqJ1_cJ_r0c~6kZ<**ke+o z|NTSs*!Z~0w)D=Wn1L-aGTcAv9{OUv2&<($&b@}kip!J6ld8|N)YH6h!Jwi0~kV4GeyKDEEx2|R3 zk&%y)F#nWxalf&#F&6(ydIpAyyW7v}U%kr6N9*e5#>UA>E;+%xEqAw=*sDrkmaMDI zyH<1VqsLmSJ-?bV_FlA1|Ia`@Hr(N7=VH5idv7wX-_`BQ<9^u6YWK}sx76HfQq)p+ zJO44wFDgpJFmLZeW z)YP;TG5vCi+661CObYL%4^i>L7fW3WD7<*2PqTuPl7w=8$jHdZHMc2mTu=Gk;m~y@ zt*^g-TXAu5Lk(WeN1?a1wY6THvsZJvbm{5Lw{S&TN|r&T4_9c<$ObI`KedacS7SqE z=1x*rQQy@?^h%s_D7-RDGe$W!CT;t!wJ2@fx;3fBva58b@t)`;)#N)(i)H4OK1|e} zW50j1p~0k#3o6FZqed8*nF(`CJ3h({dEsucy)9F>#Muxlk^I+NG+N5O<#-V5uV23^ znwr$FwPv3(kDZ?$WsFu3(ALqZ#>P?WC%?>Sc_gn(wcoOo`}@b3XSr$E>gp6$+aI-R z5hj-}2ckHv@|WkwN_--$A+$CD*Og-a)7Ev%iI5Ljy2p?E2%Uc#l$n*qW?*1o_8u?W z_E;qudmvG(U{6SW;)%fF;cIHMtvOe7e!aZ#@)ETt+OljzO+J|#Vjib`@mgiobLVeKfi|B@7W@1DS&J9kv;Rx)uu zwhlU7BSFv1tRB&kZ+-Nw@gC)8=QgvkvnQ&1{Aw?t_SDR`icfithKnYlAb^7 zNi7GOzj%Rzm6>nr=WuKi%1C8-c2;KQC!zQ6jYlmkEM}(KEGKm>%RQ-*)YI+{7AOc* zq9#x%L(Lh@E-s^e@4^KSTp!+m0-9ni{@#ymi_(+R99)N-Rfj(1qG^)5Jz1>s_NMuC?r@ zlzTI-S2cO2`0(|ul`NmKj85#k+N7YiE9~U%%e7o5pM0K{>C78JBf}1?!hYMA?qh~6 zTWI(5Ce2FbB&Jvz3Odz?hq62Z0?YHJoG7zce63=Kod3|{mmWtq`1JC!Vve~&^d7p9 zkPxSkMK`_62kV>Gx3sjFYM)9@i__lao0DT7H(la9X=ZFr+!LvXj)!ph>$%U;l@B$a zL6tzW2?Pr01v<-joqZv^>d`fJ^P{4h#LXU0jG|NQ%#mk#)X9Ci=3rKK2{U%RQlZ_I zvh0h}_?A8G@cw>X4v3yZ@lw0PE7yyk&4d8dSd9eD0uAhHlAj`$1LJ4BgEoF4M=+ z`s$xw*xS<9R;I!Am#jHe)2<7PL)WHIcJ11Aut@~q4Q+rdMJ{PaR*!|>D?OVv>^{UD z+Vs$UE?IZp&Gy_DH?;o29UfD|7vgfGHEKjG>^29+#BB0xX1`GOv9|VHqJn@+hj+=L z`hs+~q5HBz&lXEXf%#cjSb{jk8R{O&z8U`dGRvm_w8BFf9gg*niq6_C#@e)T*BE`C zloV1;C%ka=z1qfckSyDnUWP4^d(?jO%tNHGY|1Ou6>5s;Y`>S>QCb44Ukl>cI7E930*{ zMc3m2=30szL?!J8ZvQam{%4m$5P0nCvAnn*0OywWt`lG^HYIV%$=f-_&Ag)bUN*#T zXO+_j1xo3!Nfl=3eLU8Xe4f(;8$sU8ETy%h<*q~Gm+Lv9Hc@SDXb`$$BGV4z-%bU` zN6FEvJs-8rlhqk)NWRt3p!z`8L)Yjl_8`NxGD{U*xg`c6x06p)8q#$(frlw? zOUM3BRaWE=dvrQNIEC-6Wiz=M3o$~zYX&Rdy<3IPUkzsUyZ!`E5B@*jVKlC@jgJ6Z zXC7!D{htY}9~uh$^T$phMuG{1;zQOFN^iFAWJ0mzw~zk!Bf^Dc-RAc3Y;i|hb*~KG zv|qLE<%<_%;7l}*hmPc;F+}K_n+Ib{rmWdn<5;%5o+qbbS;Aj+VJEj=Pne zMoZVsG2@_2&L+)GjiB56;L-Ns&ZUf_4Uz&&+3h&iEj^ltZcOOf)M!`emuHHpHQuZ< zb3+-7TjRexIkVo;+1c=SZ}kDE@yD;p59Vn8>F@9N1~ltMy=^IVD{6hgm3n)1?hJyve+keL0Jr7yWDjqL`OieqJN^Viknrp_jw2*s6FwqzOQ2<+_Z z>uX%2=dm~$`MOaD%`hdmQKDY@?{YiR%cUuAS%8~w*84LkXf>KQ|Gs@|n=|x$RfVf@ zTT}0s2?+v7R@BKZ2LYKS$aynl=HyhPI$C#^Qyq|!G6Ka&Ddchc!C?d3*0OL_(d{Llxx=O%~7CMSfB7EErLS{k=LWWD7dx_{JIbLegO&JP`glLNW8{~4s(PB-%S#`3 z8&=Vi9|2A$7nkKQs*84*bVqvD9QdN8z*cZ-`0Fh+`~=h|$H_qs!n~uRHtf*Krthzb zcpvGeQ|zcb$+U5kRh4|>!iAcx21n-RN6W;neBMeJL4=5gPk1;xP^>3* zL`6jfHTe2L&*tzi4ns{$sJ0$TCC>Kw1qD4QW-TQy`%yg1&h4O(23T`%XW*kpywXnN zwDyAyuSnM)ZV}72?9vslyQeZa*!X**PPXr5oy<|}Ia1c|it2HL17rD9oqKTL2Ixns zg-rL4A1lwCIYY39uC6Y2RTa>O)x`I+Gqba=zI=IE-u74OIT_5$&*#P3)6+xk*;-jq zAqRF52rPldy#|UP~odruc$piIDJxv+Y4+Bzr77haGo5j=eK!UAY$ z<$adVBjYOMvw9P`f}R4xw?9#r$eARLhT7Y)cjS#vgX5wwP03j z#myR6Z3?u3^XLJxvLH;QO;X{@huY>B{gz9ZJKZ(4v;wds86f0^J$Ufkz5~pZycTNO zmTxu%5R`m0vrWTD!AhM@Z@!ftr5Go@T`Sv^4=BznF)?xG&lkmN6bGAYUG3amF`-Z| zdi{(QAKmJhUUeKjF3p`({-e* zM_R?}0Vk;rS5;RhH!fotRC_E<@PMMk2WV_B)4g0H!FkZ;amp|#&)@gJ`eDPuCifVG z#~&*r5EU$C@Q~BEJZh%eKC_0C06q-lKL7l=-Fa$wyWG*E;!~5|uV~0dBOmqo^XFA> zcszFO($0T0)#=fOntKej#oO1{cel7WM{R9w0?3FV+FUOxdWti4%gcfGf|n2m)_ni| zz2e=wQ-u9J*DcZN6F)@|nlb$vzRvr8K7M{lel@kV$I)cN?%y}N8(sj72jIEWiW%Cx zu+57XFIYEkPE(eF{2>j*jxHwds`MA*sp>n@L4oHu`1|*7petIL*>CI+XSVF!%N`#e zpKsMejpbnd^Yb<==k?d4{{GBeMWWy*@C$C)LIaxkCd=r}*utzWG`UwD9i%Q)I&`^B z*A{uL`Fj=h75}HkU;F=J9n`xh3#}!X7jl+f_g@54LonEUd~XKemT5~S-+_QQza%Rr z=vn|KAS*kDm6GTWx-T4n^t!OP$SAdze-AID-hmQYI=UVx5-Fm7mdbMx)0s9@Vx)c{YXoif7afac^p*t*^H_(0No=OpJ|P@SIm<^0+A1h$1Oh-P=@Oc}_VQWC z4s_U~AR0a4iZVwY%D55AwBy1HTGX5aKrpyfs*?~Jf>5Fpplw-MTOZNa=asY{^g>xE zulnXu_6DqXZQ0_~Cfv`J5EIkpEgDle@KAiQcM}~(JIP}Si1teo64EP`o|_(Jp|HyF z3JWtppYtj!lf|Wol9r;9oP9g-@w7c!_n%)Kn@*iNW!jL$c06bU72r!|=5A~WPRKUZ zMAy02l3Bqle!^+?SIeU#H0_1c>jK=B=^W*x?b5Xs99V0?Ae3W?nU8sb#=9CS_VTURS|R(@|CW+6xdmU>+Co` zZmPz7oQzu`g%<@oA;)#@&(D+>cr+*`e1M)$h1)U>Zc&JSrfLF-l>AUdUEOHxyxt3A ze+q91*TFlfj{bf@I*yL{cr!OQ@q166Jn_eUq^I-(3VE}}UWe(hh9d9iI3qRm-OiQg zzQ@w+iMRbW!91||?qAz&lpLPT9S8FOd{iFmw3;wG&zW$8yXXmGUffqVC`Sp5l(xgvvGc zaer4ih@-hwjRee&7OA`~Uggk>3;>d`j8b<3lW=lzaj~+pKDwVMEGfwe4IxRdtTZKV zd>I$59`O;IBr7XR6Wb8DuQ4@pFv`y_K&n7H1pS$cl8}(F@7~aa({nBhZcrTX>Waoj zj&sL0X>zr?NnRvNqUNb6lE?q*&ufU@xVwLj#0|zz`tc9D45+p4x{pp3tnddr15*lx$E!qz}e)h}= zb${UVlfc*GxXVVUWMEpuEm<&yR!~0N7iYwy$0m$v8nI~2UWLCvLsAz3;eIpLQ;BZ1 zsz96B!NGxe1)=zP@et#lkov&n6 z0+PXoiiAeFm-Mc;#u-JG5Ooc(4NXyOAb_lZa@Ahywh0aUHNc}kHqT1zch0tp1vcB5 zxuyNE+esH7=-GYlryz*0J~VDldpvPQ{ILVjL`8R;8s2#Bp-eEKb0VmblZ-^a(gEee z;o%}>uDMsf95;%F4DLSLzy84iYq_EE{T(GP>mg_OW@Kcv?mj}z<}K%#NE-JVER{n+2KivI-PQXdV6z+<|!RDAH_PTnn{@^76-YAgK;M zULN7Q;4~{85*n(pnEyFLPZku&aHuH_?4%lMA;H9I9S|fFuY7rmRz&PHm17WVkKn8)Kuucgn0mnMMYH=X{l9^=y&63-rc3cNg-ccYs+OKL;&j=TlP0_ zKJChtmCeoT0sY?EazPI`1}zC5NPU6rW%5(|Ejz0KXUEWJ`M}GVnVBh`Kr`&bGb$-L z1gb=h1tfp$7y#aCLeucfZQGlnzFW`zxk4OIm{#={MoHs=uuU}o?rwd;$H;w0Y}~jJys+`vsfir`GY?(ISAk*!@%u)K>KQ`IAdH9Y9=6Y|8c5`(86q9sV=aJQ{1Cqq8C|DoSX0WW-y0 z*Lsa~ZRH4&wG70`|8DGyBOy5`1W72+|7h^|k*xgM}2;rd7`zlg{gW61yLZ~>^ICAM9bgH35M6kb9vnw^1l zP9CgEN3Z+Oa4^J%Gz1a++yC0zwHKiCn+<*IpU|AWuoq@bSZr*XuwK#iGzu>y*JSms zH*Zc*SRreuiaeAszi)QYX|o&%uo~52?8mc-?pj*+Po6%#My`0tHS*-v1h`S;lG1e5 z)$jQQ2A)5p;hwv4<;ttq_U_)T)F*Y;&Mq4dngc{{4T_vfYMIl-rQo_EIDL#1vysK` zxt-gDgoK*AUX?D*UB%dTy)mJ_sbDgl&+C7jPpcp@GIFr!ZA@1zDw-gL-o?d5 z%%FlcUP(yxz0CaZUaG@jg%c$qF@6l(zN~AM3Tjp9D$({2h07m4@?6cvMQw zwl%BpJ61+3Je?e0$V=1AqE=T|*UU2X9uyX)pb_oKnC*rRcZr{HBNWyOHY>l2e`tIp|K129ryOCb-DKA58aM8<9I z5k8;RsB`MhOM-&W9Da>LS)n{Ce6d5YMiPW3rwCE;#?9^0c;xkb) zd@l&aqA7Sz@i|)73}vaMYae_wvd#qPNArd8s(U*wB&aX_`FY-Tek5<^SJ9{vnh1eE zs4&&g`iMb~l0>Y25%uSvTAuE?5FJw_x&~4)KmZ!*?LjAFOzH=FSqE~M_eOGC|{k#C(-O9c$W8gZ#yRzo?lWHldN1(uEVy_Yr2ekDV z(2X(xV$L`Jw6vYAx#odzK?uwo`1(?i7#^rcgA3Q{(jYS>FEvU;s%U9(BBjB$c{9`H z%a{KQHZnr;mq$~nf2PRe`ltScX=m|4_?=xM%;r_girdF7RlG)D&~yDwCt=a;uwNKkHiR_}eh}U`;2;%LLXhG z6U`f#Pn0&*y$;KXX^QDu0D zr@_L+PM)k4bt$k%4^T~3Ikv@m9LiGKDpt4cirmaRG7pSOZXU>8c(jo4bmvmp(%hM} zbLWshf+WcT5B_*x^Lucv2Z$$uktQj{->fx9SVF%&-oX0G=9s_$F>9mA_J}UE z{lTno_H5wfr!)Ng+e|*|#w#V&jQ33(5UiQK1*D&rwd3Hy3nRZ?jGp<&58ANz(rVCk z>h8%7s+{exMo2?OzbS8R)k=#SXL!8d_wM%XzS^ayW;d=~vj&n+O?)seQUr0upto;N zdinT#&p%fkz)}DB6uUH>8Il;dFbXS~#E9^O>QmH2a!fW$+Iil(MFX#r)?1Dkx|N{n zU>$XNMj|yb4SfQ8uET0lj)PO1h)$R9_&dSOozBhf=V!X>*RRvBUR^~@rG7bj;y|I^ zZ;*BzlgismxZ>l-`c-~xTB3c-V1GczwwnP;)?-~hn+mi-r6f2ct*mPBYD1Fp?bA<4 z(m~v3OR1ZQx^Bn_L>#Rj3pJv^<~QKdkj&zg$RE5H_}B{4n&75{wMv3=zsA)(n6_ zO2o5--0*`{j$e3cEdw24W|YGy|7!iSlTOe7nqb+c@%LlK&Z43?jq+TnsI0UGEey8l z|B-J~a|^y_qx4geDc)XW;zdFriA1otlVPSS+pe;`nY9qOz;CBrXwmOGg zXINpco^W(LfDQaaj7JmtTldeO{=`}P_AZ=r{@sTU4fh}H82snyJ)ps!cn-m8Y^~kBYfp(zW_qAD;9s77+ydD`4QoCVVO6}RZmlU_0q9sas`3)ND-l|ogjj3nX zNF(zx+)=b1UUNBIA>|C+Lj)H>^CO{KF?=`H`XN3W**V0HmGFmTupiLR6STVrks%P; z*C1qL)$8?4KX};Vav_3{+*W~^2!BFCFLVVncEuB*myk~2M9cmB{Ny*>)|7kw%|EQq zsi$lE1G^X^1ls~-l*AWFLJUuqU;!kQB7T_b;ew!2RT5W#xLXiHNMnZB#+0m@a-V$& zeVxeBP#k?AF(eqjyW{8Y|0?c~%WVLKspv0B%3JnXv~PffaJ27raDBotMk2fus*F%t zCL(J&8??2x$KZ7JeNmJpgq#FgRMWLLUb%9`5qVYupass}PZDT)h(1b?E%7md#eu|$ z{7T|#`2Hgh{88y&AaX`1zyko1M1aLZO$6Y&{mz&kdN-gxt~vmq)U@46an**Fd#mdSkopg1C+LaAd&Ruz67 zvN^`61R+d%3O^W@(N}~{ND|Tu7zhl6=%jxmX7Ap;r-!B*O?uIRy22j7xklzZ#10;^bxk4juDm;HZM3SA_~NUpx6eLPU__ z2hp!zbM*h+u1jhB_;WJs;z2iuwY?r)dSr2`$9-`)renQbPXp!ev*yz}f+xc1!+f)3 z<>i+%AV?f=o!L!71=ik-VOas8p}r8%z$=pr+1G=pA@naw+@{ff68^NY8JDu$eAZY2 zoaQERPT~#~Wgh(--AK^Z2TJKYgASa9x~l7T>{tM4z}H$l5g}i9UFaXzi;+w2_R=`RPp9zo56MnWX?8DN^?D-|ss9P7tbyO5;Yvg`v`T;{kR9yAufzrHE)O zDPN#J5m)C^{86HK!cuRrh>VGi4F-B9*=tf6UKir6kJ};r$_$1 zDSn;2BEyt{9Oxrx{xmiPT=;EO=!;4~m0rHSeQ3at4T4kB5k5p#fhc(}?|X42f6eAT z1RYTA`oPkN5_<9^J)Vi`I}r*|{_&+1Qf#Uzs;eL#Ch3+G3OmtJKxbB}~2Ie|A_ge&!&g+)qLXe|^59|+nn2MVerzVNgucwF1NnDy~b_K=~I zU!5g^D2~*QZuWOk#>+U`ZWY**zEX)hxhKS07xs{Z{z>JR~g0p%;0SedN^6VzDw=6(4j+{$*2zO zE<{flaba8sZRB&Lp)7+=DQ$a(wYCy%FNerTWJV+99I!zcY&0y3)Irb#em=epgEzI9m>9+ZdWUvIVhwzG z_7VKB(LLMgqPx3<@~^hN``viDHqDWzo}c@*Xs&en)#zRFS5etEvIuxGkc$vLEJsiF zE%1c*=F_EXT5P;}#_%`GfK$Sf7Cde7wKD{*;8)i( zC6Z|4*?L4L=S2^=+VtEhV}v!#TE8n&>hEn-B7q`hzN}0kCXv)}y3&V#rt?2t1$fnj z-AZI~gqu=#GUAr_GH)gTh{)-@ycFas8?MgRuxF4w2_WOx)D&TPyN%znmKZTHGatu9 z5gDHY$d5Ou^nshd5x1-XHC2NJ?rn9*W~onQ9v(y}Mk=KWq8FL4ESXL<#;_8=E+^ua z6}WhP6B7_THV#4C@071!3nLk+gw_h|7NcUY*=3fz77^F>#*fLPu#h*;m+sFT-NMg* zw1)%J%a0p@@5MZpT!9u&^wFL}ejiCw604<9;0YcFdLTkdvkZyZLuo08K}t~-yye6O z2@aI>sBlCAbTFt=)VUTNDLKDjr;@k@B_rdeX;V|3q6hqe z#2xUwZvpqeLNUa}dqqcY+-muuupRrHmX`Jr?pMkSP|hnD@j^U|{1Zk~3AAn91BB*^ z`%6q;1QsJ5Uaw!b&WA)L;LppU;x*ozga<$(V~CQaTDweRZ}KCm1#l#phZv+WNtD6J zISlB^iq;Mya8w{Xndz(3MeOrWkX}hj3omf|A;uc^yS{PSZSLo@-D1kiOi^xKS4rte z;w7`8{LX^?x9?*)VkC{^?Hdo=pz#FDBf|rZv*S-vo`dEloqrb6+J`EhZ`04<{AVCO ziwW7zY?%FTRt0k~&`}Nxg!7@OAXokO=gHI|;5j4`5SOr%pjfMcp=0NxrJeSaB4Lj# z2`geca;U|8Ugc0eY@}ytNV;ju&Ye5oNXqk2CpS29&p8*3Vgbd}|8YZThg|Z&Eq*pR zzY6A?0g=crV^|hcp(j?)rt=t@{y$Jdt8Dij0f0(|l#f`bB>Votl-Lbc1=AyYc|OK_?c)0QliHY>X|Sv`mgyX>)bFw8voiWz$JueuZ_9=P&3H8 zp#0b}=BINV)$IIN<7?;Qr2Wa(g!)JZ8VUcy=l-5w3+aL67ReHWED*KFgaP|LV#w`# zL)^i|&DlLJTc+>2>4B@XDJWLY+CLq2*_AgFt3n|B@Bw$aOY1|sWglkWxEee-+Ht9R z_GO%IA7oj6nwL2_-r?a~Vq$@}<8;SJvs@p;@*AudGxa>g5X{G%k1wgvB-Dv)al&Lx zTpj|ii7-GhB}+>zHi$KU=QYt{8|RE2aaWV_ofi1HijkM zo23I~A)+URjCVjd=U=x=`z`IgFhdluYJdKVHyQ&M_}nJBsLK~W2z}6!=XqqHvE=VvNfUSPXC+VQH2ZAaX{yQpTfMy(QE?BB6LP(h! z^@LC=5S`bOuq;k`7=3s-&Yfpe(6;ett5$ zEfo#Zq~n{#3nmdvCpK0v5xIh;YE3R(lEX^Id?Bu9H0T)p4bWn)X~jXnW06|(dl3ybjIeu?Wjyk244p=7)IvE=BpN`x`$V8e0pTgn%+jH@wys%rqu{@{ftN@SFd-Uh;4}q6QXP^mhC?s<`MLtrLNAQIA7oW{RZ&_9GPTkt$;!}CRRzCB*Q4yL5syXF%r>3*#o&#jlh z6}&+U94rx>rJ+}(L`9cJRny;iYjV%82xA$<}etWWfPWhBU- zV7O}wja1~Kr{EPa$<(N5NrM@$58215!uGXp8xIJtdR9J)k^UVj6jt(mxkZD zlL;pTft87p+@tc3GxpDyCZ8Ag%_q?DGruIKRNixWdFPZ`VVjtM4N}gupH=#>kU6f9 zmA_FFEBmvN;qFtrPF!IOr0!@AZ&2M4;EzB+ugkVf;sF2}kPIg|DFGrdaSn4N5t%>% zt%mWSDrgO|T@CS&WQi7_12}aq!}bA!G$f}Hfa*GRGP8xrWKX(b@>MCd68B|bXLNM_ z=4|Q{y*xwD@0gg~;#(~BP7B`n+3rWaU&CMKc=Gf`%}2!%RIlfsJu90Qz`FZ{yabHu z1iT$qV(f8hMhe)w4-N;>hi*y)1{};l#Dg{4P8T;X&Hwp%3tp8zMj@e3R=b!7`mMfJ z%f%~oaoUDYKS7BiAmAyVP}Dhp=k$$?vZkvMdGnfz?f54r7tz@CQ zxBptTA$Om3OeMTF`Fr{5+YYP*(CimNG@lGv0v1qFNH7mdHbjsl;|n zeq-sJHdZsNgd3&@ZYdka24X>7A=kogovP|F9S zKfCmDPsrb8*KzB@mdS3qj7Gy#40p3*T;$x=;!hX)BZAlaOl@G=boD9!Ne|aXRyO63 zf9LrCVUoE?@>pO0QPUg{b0v(w?$Oauq>AJOzHp^kfY^9>(FTtLEKzV`NoL??jaSw^QFmQ)?*?*FyZiq`VYF8ejHOt2W=W2bJf&!TSj%7#! zZZW}ctipQ2$Xk+dXkZ|j@j@g%1cRJOX?>W3tBsX1LNam&eq^>q9*CMF7&#Fop?9gx z!p_t;G`xvpA53*JP;*H5qDkrop4WOEH$8;dFxGQ}1XJOZnhjx(B(Ov~j!`y+F8M?? zD2%+{Ca`;ZPoQFa-nWkHoHIu~L(PJp3Jzkd!Q4tT@6PNVDj$ z08S-?_GCanc43SLva1#1f(mic8xxNQ-wX~8u8ne^=~;;x?XyFTO!D%xM>M)tQc&`b zli!D+Q`TPv=C;atYl{gRZNJd^3){YSZeo{Y@hmy?Qsb_}0_|3Q;jM})j7xE^oH~5v z;ywNSR}lmQ{rmLQSVR{LUfra_sW&%X??K)Ss?eA{qF`=)IoNUulQ6 zdZB>kVw}z`xi?I7U-~6x6hAH}TsbsS`Pkp)48Iqv_hwoSQD_61FJB&&lcT<7Zr*kE zBBU+!T+~T#|5eYfy=VG=#G@qrUN<*&%sJ?nUh;T&`@2=C5|%!JfoIa2ue9av#DO4> zCUQsqL!?J4hRSWx{$$u?e!md`ELr`bx|$B}-{dSZ zbrZ*bAT*6(_*YdV>;(kyU|o}r(=(y zZu|K$+z1Q|)YXvRSVJfn_?E)`eXEep+YWT-xB4bFM%WU6E!XRGUky9v%?deI@?V^d zl*{ftF`!8#Cnxdqt9;G5s$t-JSl-x*fc`(>5pedE3kgsC?*Oy}Z70W{8<-y)q>VQk z(7%_wO~!0Db52UkMv2}3=c^oQ{DLdycAmL?|7p>=Q@-f~T_4|D9jRY=>wibD)kL$l zlo>E{zux-ux#1O+H4xZtczaXzC9tgc-^put6J4H7^}a2Ov}fMD@5ZY8`{OV#ca~|> z2Djb+yCD8pmBcNX*2`v(pAMv-uSw=9oXRlBe>8B5AYV3)C`d<%@CQkl9o?2uip%*u zI4Xk2Kr!kq@tDOMo)6jtmFm#;wJk4AxVhg9)at(LZ~Wx3)|Xl36-fgnu_E~012JQT z6TbMlx$EE4m2kjP_&buCsFRsr4U5d#{4(-tFwQDSR1LYKlD^&7|B(f6fIENw+;DmC zA=tMmOcCglRA9|e_Xq^U6b2dhMEk}`9`Uxau;p+fi8Dk&rO_FL8ZfR&4^Llb)IO)@ z6dwn4HJcM%rHU6&B9+;;(pp&P+&G_?aX2h3F~on{{oEPOPL62ZM>n^@Y64c?r>v|z zAoW<(@Q&f?7qWtz1g|l-uLdwJX}XF%f=&>IgQO6`FvNhdUuftXMBUUk z&QbaBurN3{eWgq}X_fmY-sEwo@O$Oky= z!RhBBLM3Est8rp^;fL%D623&TPpau$nE#&p5vO_qP20FV@VN^^H#$h~{M)gX>3;6{ z(uGL;!};^IJo2x%Y#)|hqHpuncvm=d*M>VNuH!V2(VV{6vD5!bi>Cr^6nk?`bqqsV zwB$RXpt2g$;$+DAU6GYYEkgg0L&YUgc`Sg`MgS&x$QSz1FQL5sjq2@0`X}G+=Nc$R zbbn8E^PrR?o=)az@rMk};bEASJwj4#0~${?s0XHJsgNisX}Xwi#fQPe9-O`a#g!MC zI5k1IE11zgf-nb8?jq9+&@Ynibrn0Y!#2d(BI3j%gLKJD2HecWF}U#jV&1t)Y4{0* zI79l&ZgM~c#Ay`8B_;KkOD1A0@_sYJFYB_K@{q5DkvoXQH!QkDXBnuaaJP;oek3&> zY#?iP=zcP*~6gF=}Lvj^m^vGWbXWx-hEWXSuCd zy}^@+8Hkhlz}6@8LZAU8nq6R~hxSTN3&9~khhc&szB{?Jsifl)OlUS7Xcr9bK#r!u zx64sn$i#nI+uO67NFEM)^(&bApK#I%C{j1>DMgoMGdX#W$hHI=OgbRWyAin73%W%j zOg2KiBprSQA@2CdusR@R(F2SC1IujK>{yAI-vJPoz+qpCYJ$6Wvl876Xrt3y-rJ}) z`QY28gTH85Suur?;oB^i@Dhmk{>k6=6B(0QCoXp7At8lxHxiZROHl}VAZVqeu7xsL zG`eYzF+I}ZiT0abH%;pl*c@wY-kvqO(_J5Lc+lTQ}5M5n?uG)9^O4w@eVkk^mpi$axro4LzZI?hUvs}f2T{3L-L@fWn#Dh zGkbu&C$PRL3zkBm;o;$}V-J8wds8z?$srC@6cWvuG-fRFB5bG4Nqzt3RXeESB`j4+zgV|T=ar<6#vm_^ zg`zs()&-LOS9Bi+@tH>|NIcgd$W0GgD4C1~z0Mp>lw)JJS?glC>^(y}^4`0Oc2{~| zeU7gb!;2vSXLd|{u!UieCPxl307l2jk$5$}j#=HNeWsqs z$r&QnNb&4)pJzg)G({pVeOmExM*AN({JY4X(ou*3DbB<50eW2;$P-PushX};4r_5T zrqitwqQsYkMC(Y|jeO?3e%&WHSms(Wkc3}0#zS$k2M2`$vF`)~^S`HL;W#3O>Y+0m z-4`b{$Xvd8$$rIzPNemfkTfLcoEd$MVgIdlA(hlWJ z2ruMbh>n8^bL0|CJ6;#Hc7J5Qnx}o%E z`wwz_Vmoi#k2s!aaI{q5UFP56eC=oi!VxF)C1?s02|^cMY$b; z>yqE2^J26j*lFT>JOUdOPn2@#%|)%GP-6189BEB4*BPr|9Qwoi;^p#w?GI=d(4Y}A z%5H6>f%|X_;jT_6E~01P24=d>+CZj0`s3RtdAk0OrFE5+M;oVq+CEWspLYApJo?Qx zVUb+wYKGSb>lGHx#`cMr_yY^wTYC+2MC&BNUVWTB^Jl=2jy>cDKTY465I;YgC+NxK zur{-h+u-EPa78OoG|2QuFYXZ{)2cBd-_1d>Z^1JoC%NE|RJ19Qf-hgM=XwL7$M5sb zmgGn!(kRKq0*OH(L=C1)PJX16BY@AkX;UDe97&Rr-9$u0yfP*B?l)spa!6h`{~jX9 zzhw(jTa4|P4+Sncf>Bd)3*l7so5>cCd0aRX3z*3xzk?h%2Iuw%xu@lrvO#2;1HSc5 zG@Ncc$^#C+9t>JsLJ=i58*es-5AzAiA>HtlZ#QN(f~NWoiqNIrA8{F_DZ(<{?sKN@itFBtjXJ%=46)G82kS88c*-DH7$k9-Y(q zT)*%4&#&urUFV#(z4!aR&$FJj?sebys{Jwf=pI8L-TkSU#Y8RS2qN|!V)qCDzGc?jZkR|tW(rLMC%KX1c>umZzu+uK_v`LT0c10(f0s) zi7}+hKn&?z%krzv+b)B<@e&pX4fI2ik^{0^L`?%X>;s-ZKFC`TVGB4aYd``<89!^q zUhqfI2Lspc8{Jn#B7>lt3{CA=VLJxYHb$5O5K1egqVGX*Ap_9C?;lggfz(NbR>M30 zV~z%F!naq!u+aouX%8GwwK8#~rKS4}!~t)E0Mi1E3`kIYK`xAggJ{zrXF{(<^cmpJ z`9TvnYt;Wd8ysL@(iJckDHnP7$qS`3&!$&-?@vW&)@*ym&M{1GbK7lsLeAU*D-TY3 zUr(dA2Qd(ur~#1?zI6}NrcBI&XhS#Nno z=55nGsY|*X<>>x(^WKMMn(@=kVE}#$4=(|~Sa1P23b2x4dGU;8e{*NtQ_HNw3|q=h z{7y?c;cCT}D2eA6T~$UDOw2Ok-JJ1d%ymj3?HkuWyqVd0pf&J96$QmqC=@*pm%97= zg^VI~wImvwn@yo@2jWfbP6{(yFz&M7Ub*oea4v|-@7E581)UQPQ5mO(zS_aMWqznc zA{u<--ZnFdf1KJ}-hmpjTrdV9ipoT#=I z4oxNM12lxjZcD|S=6?t~J#i<>3*1>&ESnYp7qrx5V98Z{DR-1GCx`enLj*U%si{r? zZ4GoktXS+3^&Laa+U*Q`!0u490pa0F z)M}GzKtzI?6p|ufMib%yR4ni_*v#<~fiE=s)$($W0vP20ON9VP6R2kjuDqZ20gtPPSfx^FfeO?MhbI3MxNxLHR@?Lyo16?ht8}}#q2b}|F-NU($kTp|eLzgbri#wN?Fgx^ zWrMIl>xiJ>zFLWnEKK{0ZJCy^i4QXTVh6W?32Io#q=0M*o*%^p4Q;}I`C<=(2^C+> z;v;rf$#W2_PF)EaY85&_l@P2y_31Om#laDxUd<&EzAl=_|U{eUSMD&;g3 z!Be=~Ew9u;uB7_Gu%)Y#V#GEp`Gz>G#Hi~_h~SBSLb7u>1Ez@L z1Orl7cL!OTvx7P);UR9e4WcyN3#*-CjE1f*Pd&zU7hs-qQbsNRg+M zX;CB@JqdxoB^AXZ4*Q}N{}3u)x9`M^7!AX}L^Undk>jF1$JAxv@ua3(bL%Q`eOX%P z%Hgt@>+z>}qxtdezaB;!d{4GG4!NuCJ=i2>k}`Csci(+mAiy5nm2l|Rt%m`Mn}&bE zG>!$o`nR{GnWVS~KWgCkSyCPARk1Ia53Y3fZGT3>aOR`?-;J9QRCV8mEl)x$yj+jz zf~I2y)|8Yj#%^Nz=7X}^B#ew7cM)`U@RLDsDJ;D4x8@-w5 z7s5|_Ty~Y$+vSYmpx{GA#ey-HOb)pZ*l83+0Z=IH;ep-p*Zp_41C>m0O!t@i&QT>b z{Hs?>2G$Br5o=!Ofp4Jn8ojObe+!^7D3)ePvfyyrwV|NLeTFR>`o~zil2UG8hej=W zSi+!m{`2`9DwIS9Kh0=YENih&luVQ_v{g;;t`FpEWEK%F^b2|fVR!ycnn=N%-P$Ch z^SX;qxOUO4@~Z>1h|0P&O=k3P6U6bzgWt-XZNK-It*?bW#~}S~WvY&eUGnrA;pfHJ zVdsFy7MkEf0eCFju?K$*sx&m7JoA4deR9ggg)SxK5CNAqXM26qN?nE#s~`4A zz`u}d3T9%autp@>iRy0QMV+*->#EO1;}g>BJKJz?f2NUwcWwN4$wWkSs01~HePBBK zn&>Qxb-tO!I7L{R_Qx7=kh=~32qOA#I!3c90o|w&BN=UQP*}eqOGhP#ir#U*@Z+#u zOd6Q^5C7@E!a4{YY|z}b-x?Cdw%~-=cFxdR3fbR%ft!ET^5)nGY*t#_zuQ28LvoMX zh0(}=`Ye;6ygR-4@R)r>jLi0m4`KIuX6)ppxIYW4m8QnczpiZ0Cu38qvDqfylq3Hz z_2Ssg;gj}G0=BN$fd4Lt?B_i@ChpggSShjJ0`$63!_L!g-WLPksZ-)FPV ztXy8IDW+aqyQomPB^Kt=FfljbuayYh| zlza>~pi?GO%pAmj!xU}AtroAA@3G!w^tqn*1MU-+i>S*z+*i6%m;y;D2j_n;3~tOw z7dcVHPzX;b`e2oWg_%uUdw1y16MFaexkr$JWzIHaxytPq9g4QZzD8gEVlzf3>4uY* z1}i#)_wR}}Kg?|)?&sZ*k5Op(Ad?wPVnD~5?0kz*C5qS97VH1t5|usqor5S){V;TZ zube9xU^fv49GUJzc_Dm9e*C}Rkt&5#{$$zFe5|eO*`BMV19zi;A1!0~{PLFWMGW6f zSFFMT(GP8r6j`UuEg6Xl$_M2l%N<-rohGyVgmCU16vVB1|83=z<9!bYcSYQtBvX=J z%MmO)SHDh5aq<-YlwmSB7!*o~NXiF49EYv1Sr2+nvo(%Cf`oNT?Ag<5ri~SUQhMUk zry~jvSdRxG*MGuzP62CJ@xTv|3;GK=Uv5g%C2f{LhXGxT&_OMZ)=+Ex9r>|FX)70I zv8%h~GMT(GxuHVmdW+fHpIxxG$0`KzE3r3e$#! zOMkHzC;J=0BHn@})&8a!*09EZ+I-o0M3bKBJKe)B&)hRpjnY^gAe*ZIG1woAc?(FD zwuLNGs5F2S1i3d26|vkIDSugJ=;kH{a&*V2LL9k)&<6+i0ULEo_gm82QcZU%TLmbZPJ0f&8j9me1G1 z9F#ad(9WZRx<~805fsiRc?=KCcA>OZAjx1%M zvdI#IVi*-`sHlKq0-5CC;S*os>IREHK-*Tq;U-?n^ArNL0W$`oVd?{E*MldTV#;|& zkiq+R1u-8F|H~^k^EJgQNT<2PUfnf=w7`4`?iU6Wd0>vg2tWfKpbOfdTLKs!SsyeYEkLYj?S7jP~@2N8Y>K%j@9VboWF$_L4h_5caS@c}pw35ftS3$MnHIUXL3YJa_tOO_tCtJnd3hhT`N_ATuQ9Vr9Yirr)9Z*xhr+ z4>msK@|T`WE}Ks%h{;0+r7!__KPu+zL4y(-7q^(hc7O5I)iZtyL&IZvuX4Q%7pkrQ zFZ3*sXMLp^uL1=pG~>Wvan+9pDJeiA@sG4-y|~lirw0p&9F&M&*5|!Cf5Q3u8K=?H z9p^ib^@t_Clv5-OBqg7(KO9YldPjhjAbqg*H5V5lL~jJ!tcF{=GE=ncZh&n|$c_C>ceH6d?>sdi45e-%KcrcZg; ztjUAVj+cFZY5ey)!WZty7cz@OPg3;$;)&lRqJ%dziLP6Ae6f{2MK2{7co6tFRDRs< zBJcqBhjN$YQ^4fX3WqKcYV(6WPOidXX5#Bt1chhxx3#t&AZu-zu)W?Q8$31OX9BAt zFRK^4sGyWbm{FD(7`1?K?n<^JjE8`Ik+DE0tgJ2NwdabMxp3pK03Pyz+ds%k zked?@4xk9cr-g_Cq<#PvyD(h@i8-OMjtBBQd>pV}ID%S1Fu6yy zA1)38JCSKIgw0hzfDySC)OyYE;xTs-ri1?X$3V=b2>*n)*(aPY;0}qc>y9bEX3CeQ z%$5F9E|YT41UEYfQBgryzz6(_DyM#ldOLC_b{=!5JgBS_rE`8Y$o z)FY_dJ8oa#GXBO#wj5X?*L=Y-<=*6QqY}Nem%(xcxC4Q)2XM`t^UR=^3BLA((!Vw% zyTo9q!5}=Vu-&cA%|?LUz7xj$Nt=F;PsJV;@v@hkOeryPWAQ!QS1kT=CAsp0!z))z z&jZJQv*ww|&?Vxfkvf`>UxH7_x+g3IQ%ww3@`!ZuYOMTCFQU{*lZJzi?i(n=c{~$F z+i$M&ZcZf zy?7oXr>Jvg>uyn(nD*6MuWY|bWn77fF(jKmpf;dmBW2rZVvkj#ZjIpD=h@qVw<;y$ zZPXiEmfsVfrqSU0Q~79+Z9}+v1?m~p`v>EFdr)-HEg~vBvNwGFnguo_AlWCtJ&)#` zJ|<;ywt0g$Ub5tvHT&n74nInZT*c&u)vHh1Ah*!C`|q2U0X!MXD9wO)<=jTKq0pE@ zQgqk7ABFut0d;K#!6+ZhuRzpgG;aW%u7HD|7IA=2IKwVUIDCBX!H-$7B3&FC)&bd? zx3?!2Z0hTSLTLYF${_{N|1tx8GZO4%X=OfuqX9_^!2fZU+ci4cHP zjq`TUEQ2#8+c^Owb`v0Z7O|he0aKrybr$3^4ktPd#vJ&SGY*Z?1Urj7=4sa5hHX^* zTJ|&#)B|FC{Wi)dYN*3%AwXg{Qoy(fB)gA+QJ3*71~|cR5`3ssgZ)|nM0e10oZ;dk zfsKy550M*1!mhHqx+0ie91v!4Kz9cJR1fe$G$uGR<}S`iegGsuG_JXBU_)27eB)tM zdf}C{{Kox#^&PwKQ3^LDO6fVdsEjpj+Pep6jk22<|K#WL@zc3U-~CUVvQRS0fmm>2gf2I~u2dAh_)pxmF zjXsn1%zP)fJI^aB?n}wbH-{YTwG<54VG{!$-GD}u!1yBV9rQ2Y{G7W8u-3(`e#w#7 zJp)!ovU9Nw^4x(2+ym!?`S{vmD6R)%pV|%oyi)M52DU1uji=CT-_l%xP3FP=jb#od zt?tvTeKN)0#4mDN4cA=qs_FV6#`XILH7i;q?^xC1gMe0G3JNha>WNh~)(2QJh#-Yy zXe23UKXFm}7vx?hu-K9DKWYLY%028Ta8Cu=Y7sSDVfRFZU)^F&yM?^*W!Es{@ z&^Q=(v?wxq`hk#wy8p}xGAi1ZZ}cst4_ka}mnj0STfKOtv*VR9TKvG5nVAE$&~QN* zj!EQq42`2(H%?MgrfPsys2-^3%Y0tJ-x(&dNw#d|E+#9f_E#EU&fbhQINq7=flI&r zktt=n9vqzhIVfANnxDBjtCBk6o>iUT9`?9F*wI?-F=8Zt=ifADkeXdw5>+nNir zPTX%gO0ZKeB(E&I;tx0$GKY6ScS5IF1ZlYGirIeu#dG{crQ9(WA6VWD`)O1rY;TTb zJo&d8FKKPSN$R=N?^s%A1j@yQjSOOX+1_dRAz~CKq1k`E*)uIQcG_XU1P-$xTj0 zpS<7B(65A&e&{?-U-3dp=RHrR2pV4Al;bpiYwCS<0^eov=m~3Ve3uyaJn2nwQ;aRW z1%7DgT*KeNn;-Y9+-3;65(&HR&KVb1nq7JQI?x1<`Dgv#L+LQnyf&TJcp0uWyR21$ z%1Qb!;$@|W;Bi1Ei@r1b5S&@vWqh>8kI}HAkFn)(q7^0udoJS8Ae_Gto8Kksttr!1 zb^>~8qc*Cvo1Zt&Yh|_H&krKP`}YfP$~^MkA-GsF|UKK^% zvSoVDm8Ph;gom1(mP7j&K4WPapRF-A^?+cEmhI--*08X5c8a(Wu{mb_E;wajUNi#b zq^VcYCyM?B0Hug!CtCZP%BdnVszS7Z@G43JPQRhxVDWbq{bZ7;qztVYkV|guua)Ti zMC81vDj%s3Uz*T5`1gS{8M@i<x1-@h&d4F#} zQ-5~U;loF`unI3$sz{?Y1m{uwI$fv{^>?~c8L-Uy@3TT=&lR{dge!%reYn=59o3Q^sR)U=4}9nUc~Sc~_5QogOS8`0!W1W0 zmxb_HB<~N%q@*C}^T7{?5Qdz2wWSOvh||AI%vl0IMcYapPXRnf{w7x(CyOYr=~e7l znJdUI=4c9aoUUQ>BBDqnp`c>-==if3H3|!6drTzSP3Dp>ryUX7-XS98rV~@rI^g0T z1T=X=*Cn&1kwU7T_fEMTEI1x2TiE5T8Fbw<$Nr5k8A=@xhKg?cKB@7@DOP*zWLL7E z1y)sP`Rx1!E;)M2=!X4ckENd>lU;OZA-?7G{WRevwqtmT@GAemS5|aKex&K6WSz{5 zgiPm(;3vtEi!J8{UgtriPTBj51X3wyKF`#n{XucItxarEK$Yu}W#msEv!lTwf0|(s zXET(DMvBRl@8!=M7-B|A!Qd^x+FBUhq$7W^Ee;3nv7MtY70VXmw^d-WTj^D+@}H$b z5M;GZ{N4H(7CtV*TX@kIpC;xNtC>ZL-4waaWc;qzFlEx`oB2Nvgx~LkUZEFD4<>0Y zN0Wb)OtSg0Z}7+`p9nGynWt1PIM}Fta&mI;tt1X6Q`54Bv4YgpG4Dtp2DK(beYw*m z2F)T!4Umr07@U}VGT>-Tdlh(ZV*Ep$6SuHt>xbe#`@sQrtNM(fjAU3(TJN!nie_Kl zX@e2=(Gy}{@y@BLsX5q=Rfi~ZVtyu<*beazNOJ%$bpqUs--b*7Mz%&&aAjEej*coA zYBpM)f@gG;jQ% z|7MJW>~adqMr7iGdgOq;yut>Pqf`?p`hkGVyo5{|2_b&(%rer6IT3(L?T`KDJ(ECn ztT|=TlxgAD+2=}^yFXn^w=FwbsebxcaPZn=EF7h90BZ1Am}s1ej-ELtPD~>DzE~Nn zW$OvmE*P&$A-)!57LRA3t=0Fsq+enH@D9}3)L%R?(o z^1w}gemkI6vgKWSXFu$rldsea>Glu`pwQGZgd+rjKL;9{YX&K@-uzDYP&K{|YZ8}gZK=&~R*fmH- zTWH=!4$cjkTpjK0$Q=M#D?%cwpzQ@Nb*^rw20BFmiBTaK87;sMWc>ZAI7F61vE|N% zE0s|rC=i)UAbBsmp)JS?M4T5Wk=q8MM1ns^Adrl}x85E=D&#c*1Q7I8aL^E-a*rK; z@bN2_4T}%TSWPvO2rr%8!d!n@752vJxRZ@$KZ=Ae|8$-*eD2@B{{_C=kA>{~0+_-C zyf>p#RqmKq01$v9E;nFfM{oePlyh=&GV#BJ)%x+Sq44&UO7(jsxcr`KT&Z1;bE+bF zAcXw`+D04#-o=419#jl^5R0*GJz`?K(|~zT720&L&yEBzR-+> zFP4$q(Dq;w3cT#h*~6zB)G(~#m#wHW-Oq>8eZHS-&#Y)O6#zJ*%Ls)w8q1MKrzl&gO95=eV@_^2;04pGv_kGixD5 z3i_8<>N`4)0aL=VC*5bg?LbIndjt1!+~a!hEYZ`vycqtIdtcGDXw@VB zGZ0q6HsB~YBx;xh%7LKU&u6ga^kSlcd=WBTBJW=EWEe>|?;E&@3shF|F4{Q?Y`FR)p`99YLBP!?-0!O#Ds#`(OL zrwhq^h;6R+)7;SsjV)&4t=!wtJoq;;(5gqGX|NzdBW{7A%Z;QIL`FI%!O(8-9vkuY z1&R`#Hlc_!O|S85m-}e1vO^;5IVTb<4k@+9p=`q%F|Y?j{sO=^X&4)001*DvcQ93i zxwS?FWaMPDgD>S=U4Myf)esF_W9cfCQ1JLf@%ob9!JXCl_iFuRVE_m|kmj|YpOAf( zSjG>P_3TGxbHBuve2q5|B+I`HQ*JA3t*`jxb#l<6JNxgi790!;CqT=F8E>$PdKkkn zCHk~PM>%Hr;ujsp@gF;~?;;*zH}?;;W~dOLk~rfd>2AM#MD)?-d=j6|D)XUd`ZTPHo+nC2)+aT zG$zGE2x=)nN|96#0H#>MTRM6DK$}Na%S0k#(DPZ{WN&;~w>n&G-}oZ+W-hsv5q_7N z1@FsS=kbmzV&xD0Nmh5>1J?xT=4h0HlM{?i@y_c6Ox6qTA^wHr!qC zAbVbOxCw8`IP)BDXo1iDr@OfGzNLk&WfqHN3{6*$@LU(bH5zL=f333F^bphM*F+qy z(G5?+U!0bb!f^2adH8TAXX;{Uj5jRlJWmD36~*XhHx9@6HoiPuyF5g~pBddz92E2$ zV22$h;QIr(;ZKhbzeAe|%t~XR>?oc{iP!-9i}BZ!EZJmJd32LT^P~s6DB8{@gvMJfO$u98(;V1w0OZQ9AtZPAlk5> z?V?AVK%`iJ33kW$RA2@0D*#}wH5zdO2hegIG9#a!uAJsz-tain8%Ec8$S?sd4qqZf~o|rsU7pB zr@4h94WGz0xcfd7xoMYV=@*~Cg#{b}SIG~UCU6upk9Ff!#e^n zO9=|q1C;Ls9bJwasJmSd)CcoFvWkFU2pD-L?>lBpJ0rAUX1mQ6Hxyl2FMrVE#Ay`K zeNbFO!s;^D`tl(*6}VFlnZPwh*;dbfc+GBDeo^(l^zX`U(INN1i{!X|e(VJN*fe4) zi;Pk(t`%*%Qk&a-NheEX`zmev7T)NT=)BE6rztNKMEs0Q$S>lF%$C&bNS$4afThpyR?0cTxp4Nc7k&_p9|a}YfzzJQ((IH_={vcvP@`Ut*$ zJ8*2nXm$b+aTvq+roEp7bz50Okb?nF_aqA?nW%hI)9TMf+fJUz3)^qqE)g+b%^)P0 z>UzXVG0FJp{UQ0(Qcb*U-@K|7F<_|5fd~VPhm66R+2H=nGebd0?$Y5V^&h4LAmrrY z*EqYc)OA0OKw@VM^ks0z^mSjUXt_`a{=Y!`Zw3l1Cw#Z{3!(|U_qb_soFdt&nPcQsExrgC*rhIwu% z(`*!%ECT;3UUtNo`!`8oL6X3nDY$sJ&5}647heSl(R$qYUO*WAcWMZQO78stA`&;C z>fsCvi$A2a0&|PR4aOj7NEb~AQliMn^gVKn_pd_O5X=F057J3Q_>8O-K7_cBmo1A22HUxu$Luqllh1`enmw`qWg*{#Hhtz zl%-JQK*l=$zzH@)&r>j~c$$!K+nJawLeuKS>eg_%O2?0TTX))R`KVyp=r939!Yxy!@?Vn9bZ>C~@B z53ZYDOmO?p=T|?CF?4HNIETTcTMqGyI0dmNIu4q2WL~(jd>VUK4|B*%YQ%ls+_o&q zP>KuFa{?CKm5Uea)rB~KL9YbK@y22Dj+w7t`Iob-r(;TT(kw1tCW-L9Tb{i^NW5g} zOb5Cs2=&3Cp(SDgL9LmPHR|L{*x#l4ru{ANo}{{oE0MANuh=!GkWibgAWV--COlk_ zzgq#PTuWOUjH7j}eghc-22VjVPC@U9TFw;%1640&DKF+df*_&-u`tDt5g3E(HuhC2 zxnXeQ#!o?S1Z-`J(dTvRY9IMl7v5{J{%s^dYdFDKr5*1T`|;4poV?Lk{+r21&d8^7 zp_p-}=DSXsT0;Bl7{m6u-`{!iDmR4i#n=u;=*+P|zS~xOyWpr9rGdT0z zkC~B?74}0KObZd~xWp>X_Lw^lX`tbQV5RfLE?iO3>t9~lk`9IsJa-Y>YkNE|<9dbZ zo_fMavUq+-kU)T;q~HgTrR{7a?Y|%Yy(J72%Ls_FZtDR-_-Rs7xb=$>m^uiFRP$si zS`9sDkAoOI?kJogP*gtkGF&PdB=E?vaA#I}zYRLPnJ~w3*~M|}{_New+&BE-yO7Ns z0z37X05n&->%f2?B%s=M38(ZhJ(E#U^}}|W+Fz!nTQ1oDEMJJC3`oB+L)~6cv&&m# zrG88{M9DoFUxyhO5@FzScnz&ItbPPM(ChRv*wnSG>@UBz`NFAJ`(HBCt+D4^pFEA# z(8JvQa%RNq%aP6`4eVzU!2?{F^<83mn_!{OQm@5}PXiuN*A z2AA=@<0XQb>Y^RZn8~c&7b?~=<3op+Sub3Og!VxZ2sYgIt#e?;LLwZJvtM$ZS4_<4lW3n6qJ;WFbxs9w*|VFWn>OP zip9ps849!cz?w`A#%k(bwzOoWzTrd$QX_c1Biy;GH*Sp1CH*Pi^a}wTDRdXMpkmFB zHlN@hfeqUbnYn&!D#h1|pO>W%#hAo7q8xnw%aL`zmhB{VrXyjyCYAKhWL0tRnL`56 zh*kxf_f-$O^c^gfq3Zh=pi?;s!@4>elyw( zK+1EdcOeLj+-;#ZaQ)(MJ5o+oHA-3!B|Bl+&tlOgAMFGpG>-cQqzjzp)4_O)wmpyUy8+dK#N>3hEc%m(X>SHVsz*6xv;b)HhC z_)6b39cy~ZomIfDO!tN}708Eg7F{TMp?p4E9TmA^Sc!%fqs3P3GnvFDTQ#1ct33VlHtC1@bpyv3qQZZZrdyHe{aR| zn&5&`k%g1@wvtQ1hvrZ{&_}nsws(R6OZ;&Bq@r>X%H6KYis_4Rz!GUUv01>#L?; z!nfPHi=$O$bJm=?PUW+AKku9U1zC7VJ~xdlNS1qE&L61ct~xGy*YoZ$-QMJq%;rL@sfjU}wfd)e zo8+FJt6p%jCYF8e1IktdaIYX7ofQ#@g&pYT1pK-b;b`MRiRJ>6?F z(xGaUl8VaY1*_2Yj?YE8si}USqO>4mT79#7oxX*cQ-t_>5#GJmk3M{$%bs%0%;9{Q z!wG{Q$Q>N?)yksxByMHGZBd82^I|>zSyYsOMpn{nW_lv9MWP|*kkin#(%&1A0Sm<+ zy-B|tlYVcq8F~ZrqI{AFj%3zLjv-3oI+}R-b)?|<@M3lR)h>I1uJP{MvbHo6X1229 zYJt+$${Gfu2KTI}Bfh@E*TDm5EF471iDfP7uzDdS2D|h1F7a>nO;ox_Th?7|;%f9| zW3*42ZDZ=i&joL_&hSxKx+CCJod`(^A8ZZS)bZ;@a&Lwoo6x|&my`>8fTm_?tS)Z# z!aG7Dk|H=ky{x6m9!iyWEMevQZtLsAGl+WkP7W+)KMh|gRk`GqSgIvv^*G(J-}UnS zuDxHmzmmVehWuL8!OhIe`@wN3(A`okPUx|&k>@jXK0&P|_;W1@5*r5hB|cC!!jwvC zb;<9Ppc9fd@zmC)Ixh8t32j|?nlRli*!^2g49Hq5KaO9=eh#wem}n)#seOOmTvBl% zYFZ&3GM$jRy2sM$wKR-X>W=A@5iY7EIru-l_rLAmZDWZ+{DCfsjrXkH9w1ZfW?cb= zwmDSS62CTk_vY2Tac%O$7CyXum@dV}*5X*I1->nPeSMUal&&}YpG8JC3|S>DXNp2& zHsKpc<`=((*gf18 zn7P3N!R-~JJ9zkvxD4Nx*dB2N_EIRdp{p-|tx2FnjEsuX`Zi$TGGsv~?h*=VS}>@^ zBO)Uo_RiaGZl8=0N=e44^<=!FH$UC?ls;DI+rS(DFA40&vwq>M4_F;%W63frZlR3W zx(v5U>|6DQ>&!D;u}`+t&p;kVQQNIv{G z>-CE%-nwGCQG3bewKt8wNh!(NCsM+_t#~q>vs4i-YH%B}va%*09+zU`;R%QPNhq>b zRNE|6uu?J0&?UX^K~8aN8y~SR&`fqXAm8y4->F&_f_w=EB5*(r3eHuJKuHd2D*TH; zXdN(cB)K=QlgT4kDTXkG=^8~o} zKHmR=O?4iwIlZ-V#CWGrqL#mUyEb=?5W`j`Oe3`CNxxV3h^P5Z@%0ET>}OHIf>U!a zFlQFthY_mRnVw5Lui^NTUexUQi54XEYa*XLL&RSATMLlThMv}~VxPQmZxoimkqegZ zlarHu&?y0Z15jd1gs)9Us!!UBlrI%^WMpJaj5u~%KngRsJihO+Grs@31+wggitAq> z@A?6$>HKpj+z=ZPh_VlXNe8yGD0lg>JH9`5RyrM(lHVU0?LB~!JrK702QbrKsolK; zX)HZT9nc@u-m`FQI?g!B&CJYvaj00R8R`hWQ(pw*F5yo412OJf%{BwH4A4iwbAVKQ zFMTZEI{zNsoj;L&>2SQ7UL|FNKpT6urQwcR{_BUVtQ8;U%WX;P&3}r%QVI?EbYJ3- zWcR6juF<=CYSQSY!F3>&G0e0rYdfVozOxdNN5Mz@5Qf3>gE1UMKXAG?-e2fmg+K8H zto-~>2MpbpQ^(JPsdl*3&$(1Vl<$jx!qoyeX^D4LRf&Nb{RQrjFbD^kMMa;(o$oE2 z>K@ii;}dq@x#PHpvLTp(aHzE71SCgS6Fs+v-16Vq{C6EW3~9mREsRse-qPG}+Kydw zKj*CV?#*@#FCRNW(CbgD4H-+~mi8Rc5m*6iTh zfwKgzK5?B2Fr;ST*NZK;mM_t&2g!|>@wk*;)k)%q5B;Bc=oM2peD&8jiRW%nD_O#$>% zey;{f15dl{vG5qdCqv2a4X59OMaN@sFwCo5-~hf{7x%vTmswwht?pcFy0Jq)<}6iHjpw_4=;Yer+d%CU&|7)%9FADx=_3MT}OU6J3TB zW1+@;jF%XIb*W?Is9ba3>@fQMbZ^gt!DCeh9l4`b!Z4DENmP{92ZV{xpOE|xcLW9~ zMqR&C;SZ*3vt#*uB*5phYp`8`OnO=+*5KvG)`JAc zHfQqMs1xkfCCMVbrb1%!%O3ALNby7APtKUwLB*C@(FxlKM(BJJ7q;q(7i$65Ih-f+ z)@m#mt}YmQopu?=R#K-qz0A`Px~t4=dB+8%d zNrE41iJ0uN`!i0HE)A!j2exkTtZ`&+5vGcyGUL#{lB@W>x53VSt_+hKq*PN9Tm|(_ zki2TIIf&qaqC5y_NRQ5QR*yYCbBjX-QpJ9yzmHa5B7QhBXlOjRy5=jb{#@24yzYpJ zJXTFr*Z3DPs2C=pq6WO8Ph8rxG00L4b=kOvAQ!4ff1u7#Ykc!2MH={JaKN7dpFf-l z-MXJtXyZDcwkNaJzct{tRnsoIVrl4%pO!zb{gnEd!ejMiL)|(tD*{h3=eNr)1NYC7 zk)4MPNdOw=*>LYYW*8JAWu4G5W0QtOJM+}uqR~DYK#=&AsvcE=Ug%}e0%%n?0lahe%E4S9u_GoLzeq| zk`lXyBn!0NQC^jcl|d!82v@s*Uli)s&Cp+ad*`^fYIB#sQXPD4oE;S@5Ma$MDVYGw zLo@~zCV`-KHY(2t67?W?+WeWUik-IF-E%y5zF|4wwfnI_GLKKUD$kzvU9YFXF-=(> z5i;R$i5q>83|Ey7SJl+g5(&93EljwK0ABq4KFD!z;oXD1lR4FY-42jsF-1B=C208krA%HwJyo`k22@BWJ|k6ER# zosMMIJE+J*P!(|S-I}44u>^Nbv>1zu_(4hZ9<>98-{kwV!=aMOQaSzfk}DYa&qKKY zom7B6__H;9V{4w>@|}OJVKnFY*Y2k~zz_Y_hwQr3|D!h=k+@ z1cHaSlH+$2jQyzbIp zfbmElD6yuY3tz&gIefI?`)n992X8xCfYJ|7>7^$F03ZPr2%Q!PZV@{5^G@q1e`r2! z_aq`h23Cz)I)x8dbFG5F#bNmNn~yc__Z~&UZGs~6OW+mlrk8RBjb(uj?xniQcW=Vj zt_KA}#_uL516UI*H1x=`NJ-=%oJcLpcyrs|bclkKGyUgrX`SPt}Mq}O;nB%xPZCm`j7Rg84fI_-- z+$lX0O2FS>C~O}e)aR*G0Bf8ClNn2Rya0qaE|(6tb%F+k&tX~#)`YEwESYub`i4mx zdA!ueWst z@TfQpF-A?nk-7VX@i`Ix=C`Tt*Pg7_^T|Oa`kxzXe4z$<2hJMhKNqs^e|>I6BXIKs z)aTA#hiC%cLcwmOp@L4WU`Uif?JsHF2@@8ecDIJW-m_3K%IhsfMYY;j%hBKXEk_g1 zhJfLfOS9%b`sz0bz8@=vNU79&1tKL(+MVCo1_>F=iBEG|{B#hX7?JlD92w~u8A0Hy zBDMi!0ZZWE0E+W^ry*ErakGsrRPU9vbhVljHmV4n zFRLbc#R2bC2j|+q+8fuR#ZHNJdi%x4tLE49c&~_PQdq6&`2EUY#PtBMIw*Y`LAsC| z#tiE*_Y)jAC@fTb@4rqf#J$iSA+Gkf%qbfz>~jP9e4TVZ#tzBKBA6HTC`NW zBYK*7tB;+X;G61WBLYdc`cDkhWOEcZWncmK{__j*mS9|&8)L{cM2RW>EWmJq-Ab3LpYz- z8U?I55 zM?g5%>+y~Ys56mS(~sP7{U*t=-rhR6XlgM+s#A5Bf;h7lEhr*pNFP>RJRTi38{VIAHVMs0pGUQZ1oZ8xK1`Tf23O zG;kq`E`0961%VKTc_Hoaqz%^lTVQDs7_FlR`b_iM~ox<$w_C>qN zkF@z{rS(M!EM)j716iIXlFquER%`hX7LBDMv*xbEy#&j6TA<_o`<(6a-g!uenc>zT z%dAk4KMFhYlUYYd{ie~?Asv3x5D-)oCB{*cmO$#g{;D$aiSdVYM_lHoih7h<*wmeW z7lHH2$WH36&5JZ;7*_-RQ1c6E6%_~dme*H zHyRegXq{D5WW}#8Gb-#e@a7HEm_9V+LK}DQAeNDdb^^F=<_yKke0zt=#47lbam3* zhYr#^3uxWM*)Yq2znjH<5s(SGkAr3@_OlPb^vIR3y`w|2 z1%UK$iFE=%7|lR;03 z7gNy_&UuC3PYCYnk&wWSTqqk}me~9t*M9`%kvfaDyDMhk!v@K!YnbHTYQqSe3;-iU zqx1(vplfg&^<%GMIls6x?0(bKSTNmYhat z!CKX%ET1~WWfb=Oeh|I5@%0IfB3z>Cv~16ZZQ*Ln9PLajd_HZ~k?L-`nq(1WDskmW z7h#f%DR^&s6 z`d#2~t5@q$0l8r6(DE2bM~veH4jl)Nm03%Tv&wo(<u@${&I|So|5e+PdNcxC|gO&z}4X#CYkw_j^vEbJfUzoBGb1;R}W^vnjxY*YZ70 zcOg)4QF~opf~-*1g2m&<8GrwR{p7YRQd$v`H@eE!_H=z-FE^C~FNl9nyJu^63kZ@W zI5x%!^>*I7*DGC?+hHgbnn4Vh&P$+|ZY~aVFWiJx*3*8Lg$4O|xi$k6Q40E8z!>Uu z&=S|kvY{Z*2tXIe4UCxx z%TlwWUjulCN{@xTV6hDonGn2v3crF+a_zXohtl(^%1ICE8pctg$>j&h6k>OSHK^?Y z#A5BgQoW&u@)pPg@h-xcC8W)^@`4~}3l$OIki#i{m>ye>^CrJNii*#OcXx*P(Z{-u zb1t}=3)}3;&h=6Ru#%))@_R9i!JrNx#G{&K3kwTMYHA8d2(MguBA9*70Io?pkBd1+sGNL@dXDaN zQ#tn(I=y;|06-vs%tK;p&V&yUrJ1EC3lqKGsw?zz1NJ=!%m>yHx;pkuKF`WyIPScCt9iRbj4_GN9u z-elzN7hNGx&xasbfyD_DJbT={`>6NGiQ8JBM1;CesFq! zbbwA>DuuBe;99X*YWEd7bMr8S5weqsPT_Ij9&8MngT<$fH+wz!CNW!AfqC> z+(^a%bCY=UIz&t#m>wjJMF<3i2Q~fmYhnqjuy9O$eZ8d$f_QQP>h>RaY>41w6Bc+s6`B#UkN~d9;*fEgfkQ-RJQ}{LUkRE=WV`R)43v33SW!TiYjNb1mQI?O zFg`yS+&X_%PiaRn2}lP^$J@UcKiNJynP&3jbZ=x-o={2yWZ6Qr4ZMFA@I1F4i4z3s zBAE5nCAP?!SMl|fLKvMg_|e1Wo`Uw%ESxmZ`U*gsd?5n`wRiC_JPQ?}du!p|jpO4< zKmg?f(QLkO7A-9;L_G8WUqp5I?>^jF`jPeG>Z_3N8dZzH4`HF8q?S!rv-f(lLAY_e z7}Y8Vm2v{0)YY)~P+E=N7!5C;ZW zZ@{79`W0*?uYu|Bn=!X+Z8G`POs*U4zUtdo4NQx~?ywzHF^|=mWhePjraI07@6t$<_-L#@(r zz!3PshYwe}J(QIRf%hc>Zp;N7DM13S8v@z&N)jQnql#~e@R@?6in%Ictbs}5+>_u@ zHXEn_Th*a9ivW|_on<@k4s?8Ezd{+}c?b{n&=%7BW&q<9p+~BFfsH((H>D(ML3g&u ztV%if)dLa@_r>VLEfjrRPczQN#)?H48!daYVb(SOfh~iSG4S{08rebtB8$2LRe<@5Cf^8pQ!{ZmK?6hf!NR4Df=BI+Bk3Wb0t z1dKj)@1GR#eh_ujWsK7Cqm|CPOcPkxdBJ4rE7O`?X+ggb3Fhfs6oafiuc}&{YO;DV z%82yY{t$j%4F3LAYG(i&dQK3Uo9#32zONiz8eDicFLaG6#<#=X{aMsetf{(knwg?V zal`Xeh1dF!zNb20#Y1^-YA_2MTapte=Jxv#swJZ&Ics*?W8&C6qr!obj+ILqvd8f^ zd4mWDtYmdZ??MPI{Pza7c0WtF{ClsDPKx-5l!Yy6zI)TmS`fF>xA}+pJm`c>YUoC& zSuWb8=5K~~I_#>U?K)4?FKCRC-^cojZWDwMQf^Mh3)(xZ+%|EW#V#jK~5y|1%n zv@TMbRfU$K9Ci`?HdsZY*&OW)G_WBt7*oMsziz3ixd&X+vvPTMd`A9^IhrmKLJ15t zwu86vHG9$zKvekcVFXP5ACLw-yYsn5>O(2dmPU^W20WtT6X#ts6w{;o9Nd`r^6-(j zL%~*&G--Wwkig5{J7b&7aw!^drSfWj+9clpN((4279vJ~iEAJ2|6&+~IyrhWaH>MZ zP$UMhL1_xc;F6Tdu8Vz@iQFd!6%USTm8?Bk-b!bdGlxuZ&1e$%Kc^I6Ft?$6A`yUL zUEOtH!w`(7ie86h1RDj_>velGJR?v~;s7Ns0MQm9_b4vOdab@}suy@e8T(m==$=V>9V&4T%Qd*ww; zkAO}t#=BDu?B6-AlVrcT0Ot>ra2i{tC+^yJ|A;nbWdh{_^e|5&jOa1l|3}$-z;pe+ zZR5%=DcM^pMG@IDA_}3bP`1#=CVNELJK4KrQ+CK683|=%Wkps-C?o4RFLdAc^S|%s z|9}2o-@d=TKcDe_U+?QWuk$+2<2X*it~mk}8)ZSmRtOR+I2Q0dScrNpHe!EW@_y{X zP}cnoR_f8U+<)hj7nI%MGmBx8Hzo8RF*k(!K0tf9`M>$QGpAOg4@7>L%+OyJw-zX} zt#FsFD3$eu#Deb@AlE@D%%9f9bf<~K<7oOBV;Xi}(&fEasl6Cfrmmz^+qc-(!=PM1 z{oXhV0>hWl%-}IwAt%7!sR3^S+zL{U^*!jkSP!GvUt${VO}LM{Zb0d%9+iRqt}O)W zQBnCk#l(qih7K*f6MFBr9fPTNS+cfuCvOhD;awGH_hlq+Q@-J)0GoG`NS%=XjvBv{m^!S)_os}5R#I)Qxg*n zBJ1DKA8e0yqluI58=l6>4kQ>U04vOfz%Lk=G`vKyBD2t~ra6&tQYtO1=WI;vb?IQa zAD@Vg??~OQQ;{` z0ND*5+V7>&WOTrJ+#%RJ=g7Qr1Nhi=Q`mGuDiyX?cE3?kTEUmmNfGl~9-onMhP*p) zq@aqO2$;_AP~rkI$^FE|2&iJf|KS1L67o@bl-wt|uE@xB_o6Cjca6q?tC zmm5hbGRR1<*e+#h!yKkAY0#?Epd%Dmgu+y!Li0I%5|UFhq%F8vl#kE`he;IyY_R3| z8K4sGd8w(>i@L^u{VeBxEewG)?8C6>d?dXiEDjV1%gSXuV37zUDnO4@L{o%@fCxdX z1aS1;Epn1t%_^VaQ57Xs-18yF_#_wp#9bkr^$=lu#-U}>Osah0^d#*=aV3(lwHt^8 zgbJ!)K8DzUs2jp;Zk20UWn}_7GsFK4Mz0M}sI{4#>0yLt$%oqN zYk@E58Fh;9^Zx)*F0cgu?-(UK-At&1AgYIxHvhIdJa<4JxlmW~c)?gdN7jokw|h$m zirP-pzRSNcx#5ym+Y&mbAj^+0E`I7eZGB^pKQLKHFb^3ogBdcFwbMqv2gU3@sY?)x$5{^8;#9oUW zl6%qRLuSrxJ98I(%*x6NR!dvk1U#xKXkRh~Ma=C*hngaREa_`@V_4pd9MfZUgX2wp zZe*JgE|Fw6Q=fP7lj-RA{R-7}?OR4zA*Kcjb zm6Wx+bD6*8;;%okrKxfmC>T0(=G;@o#9JDtRe`tF&7gkbH(*47mezyccRAdKUvMxk zT0*z4R7F7O0x9|@pgigBuQw8sO_{Vb6%Ng&w^}6)H;!!16Lzezu?EdZH2=OUAyKRM z*vdLIuwZDO!V9hlac2PP&~#n?0G-6YUSoukFI@GfS2uj)V0oY)o~W(x&X$e$G#R99 z=DfN7BeT3Z6l8&65z_kl3yx;d1(NtTA3TSx2GT-hvebX}0;^Buqv_XnomY>N+PU(G1;4Y#lxRgK#)9W5G4eLq%W;4yu#= z4#y|m&t19{2wptaP_*I$Ko$L9f}S}AE2R zqQjXol8yR@Jd$Gz~lhyFI5rw$(L0RJFxZeYd1j5$_`|;ttq_8Nx~Qb7RTYJ*|yPQxs0x)%V%CIN~TV&J6|J&LK(*gJY1OZ~=OAr{?FMif&W2Y==M@Cm1Qcb@!fr z;g(MFX#%oqZ2hOo`%Fj7`w3>lp?qyDI_x~;6}Lku|<6cLdB0n3v+A|fK8 zcug{xon&)Uj+9Sqk5{qW*NQfuO$a+*jG}>VHNwOn%ouV55EhUvjNrj_yYF_|tRk9q4>>D@_&J&KCo$k9gpx%pYq<|QBubBj+w(^cH zeQa;!Lv3npMo>G+*3TZV^b4m>&0*J0{%*zji^Jz22TO$zVH&1+RZZk-nvj5jt(UjB zC*EwhGMZ!0L1~BtWZ6#VyFMk5^Uk|&m|F>1j~l7V;c+x9s7_EW-jUpUh=s>Ol=|Qy z|D}`wbnL`Wf6CxIa|WKCI&`2iPPB{awT4NMN)8@Wugc2$;V7}tkv{!ZG5i5REq|DA z%lf6dv8}YH_V|*?;UVKzr^XrUHr`IK%d$Zs_k$5s)4?n}Tn92$6KyC;NxvQj?8+5j zsL9Dmb$Y4PVS#+EGN@Gb(tod=L$StToK)nF`O+8?d@Xh>$Trap` z9Jwio{oD#DR{3GbLLJ)PJZ1l#M>98|wYKqh_VPlU^rO6USUxPo=LLla0szM7x_)1x z|2nF~BlsRjlNYXD?V1xnlRE&G4ge&edB_2_mY1E_v=JvDFcTo)n<$@7LY`|Vz9JL^ z;cI|slRnXc{0z+@C_{1paAMlCPKIffBfYn0aCUb9AFH3NAenj8#%kx~MV^?Aq6(~2 zq*M(t5nSzz5TIzJQ0@Zs9}vp9psw#gW!@3%5SSoBTCg`F1A$WLvAd%MNYt<~Tc11* znvbzK)3@W0+~SS_?OY!uQo_LJg#-4MmyScj34TuVL7sX0YXCd%Vp>`By)C1D=;;)m z@-W+zu=iqE7;r~4+qpZ18-S<;hVdt%lAI3oQUN+4H+UZUiItVY z0NMkfpxq4A|CWvpAotlXjkE(t)A-(7&4IDLtDr*x%9T^o=&<4!Byy;5H*yvZ)r*^A z6;p6Yd3;YSx_dGZUzCMY3_OV#OX5%8@-`l6Y`oI%>&G3=B_B?FhF+kedq61k!pmum ziP;2RZRI5H#PxS~=wtAl;n0LoN=PYR!QKJO7*%LY;7|jef!--8%!_b?y8J~CQhhPE1+^9ZURefW5nWq9u+l}1-A^B;sH5l80bXt zPBMG~rOR!Ao`iXk^a^MhD^F)%&UCZhgIuT(9C|IPH;u-&?$po|Hkg8V0GuS z51d)3e8lDCf<}JvKJU=b^0kksf7T1dv%7kEMmM&#&Fq&RllF*&e2!26k_CjZ*@3JK zDa-@>{~W^2tu21V!yyE6y`s9h<9!AfFn&RWlmxr4=!}Ax#7&`$oxtNEJa+8p*#DdS z>h#fl-rfIS^%Z00&O`hEgZ^r&czYWbg!wSks{rl34D}NvbOJIc2_qxqY|R_M&(nbR zZ2(B>LJ-q$2_R#Ei1r=SyFVbk#2W(u_zVX}NS}^-BCSQ)Q&?|^IJgVs@2eow zz(U0aIL3i!?hX=2p~43cbD`Z@2oTP(f4T3iyA#n1MSvvkMO0K1DEIpTPzRPKGc%3?cKxgfcjeL7_cIqtcOUqA7Sat^&^b8|c#oteKHk1vV>~xM1P4 zfOz|;v;hYfZP#Tfv<_g=JRQu*2%!_Khf{7qyU}Zhr4el%M(Cde))k2p%x^vfMB9_v z>e6DlJS!3;o2k0|dC~ZS$wNyv^D*Of14s~g^WboJ25pPL^$MaBd;wpv0(YAPEyoh` zW4YlVt^8LTq6QWy4UXo!?>&K|Gqu>0%8aJY&QRDmkx&XE03I}VguX+;YVzQ84VV#V z7kJThLGmODiYbit5hCiVUZ7(HoC#^v;M>yRVi92!@bJv+Y<;+Wz}x;tJmG_Yw}$vW zn$ELnc5!N>=!A`GjzG7>yk^{ruGu~h%u?ZvF;B1sLj5NxE~}4i*0Mu}3}s*Ck5{^s z%y&L|ufCSrI0^^(ueTPKVF$9g%Rhfc{O-M`&Ebv<-^9M5ec2`e#W0xCdQ^Dn7#l$4 zlYcAIMz{SbszE|n7XX;aXSrE zO=!kOEy_XYlLX}Q&VA^D{(bXQPJz*iFVcxJA-)o`F>Q#GN!KIGEN$7RHJe!INxEuL=ro)w_R9p<1MB z{PvXtm5_$K+aIxvDc`}LiT<*(veNQ0=Pxo5xVy9saB}|NtI-C#68*si3>rHQ%a}so zU9!V1_DhHvIuWQNBqd?!gFA;lUVge@hYvA;c3GOyrBLcptvX&NzVW0M?stXDVCP4WnDGsvI z-py)Y+-ipwJ=t;2+k}k;DF1U+@g&8g*GMCyq6%I1Eb2Q6E?neknG;~xOxx@y{yWA| zL`tr@9&qgvoX?ch9sR&j)j1EZ__>aQ&y@cB@SPzGq4y%ysf|}f;8p_CW>975C^OtK&Wuna(B(yIAtlS-fcZ2 zjg!mtaV%c?Kw*c5IZOBIXz_VW76>m7vox?ny4ynmg7|1WFUtmp)z~4>qB&h|G%XO%C8Rum|OaNzmP9o1Ci;7^VDuWPJ7ItmWJTP-| zK8HH7lSv-b2^Y&Ih{34~T+k+gjhhOh(4>mfYcpRwp|&0hK{EvS?MFP|qmxiZ4(uR{1SWc!+Sy{bs{ zoObB&@F+^ERb&85m3|C?m(G1v_EKs?KSR71cGOzuid?W zEPk)x#YW+vTQ1+t`PYpZ^*Jo(I6P{Ph}3eUtk(S!l|ca2{mqj>iqVoIjS{Iy+|as&B`~LR z{pJiw9*vHeAKf>4I$}CLd}SB84;Dpz>sD~kkOdIj3tH6ghPSPvv+u$d+yUsZfKT38 zc7~>KG+o-IX!G@19F8X^Jxm%}=VplVoEq>HcV2KZpT(jYcH$`zFKAB!GZBR=aLsou zEI2?4vCqXO3EXjPcx2ARzOp)L(U~@uyw|Y!_K2>*r07RFRc+pJ!lxT)c6dx&JYlc= zZ^U0qht-D-AK@{yFojfgaK1Dct9^Q}8S@w(=|y&SbL(VTMq$VNFAfP%_$CSSv5rzJ zTD_8Ike6NcVp+O&pKO;+NOYAz1_z-;IFR2} zR`zTd*M$Jmo{(z}%-o5kJ9~-pL;Som7QyXV*(uS^`IAHJ?QTg z)Qdp6figp+ZQ?Pw8Wv@`+I`7tiURAdUXD?3XcX}v{V72K^0T>XCH-Y@C;X0qdj~qz z3ByLtR63-!M1EaVR22DDKq6xU31>It2gkcgv|t59s6>qMe>=H%(H%tAto>RZt0Eb1 z>!Dw4GIE>%{*h}J-dCgshozc=7e!!Y)lMQCmHPoNCMGv{M9RBrFo&g5?G~L3A(k!i z(kBfHnCy-(k8TT1%_@K<5v?;0AO2YZe<{|L%^Me79) z1+g4hYdFT?)G^bl6UkKa(lr)X8-s4K5qw!6%=rN`k-wey0w#=PT%G@jW_)@pV5%C> z00kI4;Cqm2$TiFo&+wPvfw(AET8M15-?s@ZG8)eT9>0En%^ruEf%yE2WDn=4JdvV8GzNKl<|=1fDN5c`%v5vvQ3m70sEi44!@E*-ujo;lmI zaPpO{jk-N6Mo<^MBJj@;ft9yx&9_FNcb`m-Xp5^LgAN9mBmj`+QK*};$K(`0OI)w= zc~fW~`ei~i%`BiF#?9?H!boF>DJWRXzs%?EgJ$V3WUrI>_;D?41l&;$B0 zKLm{?C|n9Qu>;g+W(pJO_RIo}D+OJ}aH|cqdG$QImQVb_XZj|Vu=+l+F98T4skon6 z7R7GBG(u)Th#Zc1ATYpFV3A-5c4{a|z1K_%4}oF|e2=EGbKWvDF})nw1kEs7U68m8 ze`=M?RV^0_+zGAuNo~AEUdTTGTrJpApNJ&u3l;y5KY zqcU~!@Vb)}#NUZeoV9R#Z!CNnL)ALo3S+;83ffhByIat(JP;kYz?T8p!~iyhq}T-m zwlpwNQs6dECJnNiT7W}=E=g-`5lWyqTTukdx+;sDAzd7&6(C@Z@ z)rjBBoxPelr2y#{yVn|!xuLD`cSIEdhn|MMzG$)S@sr;`{txRzon9XPeV-7NodJP? zckU}|YUak*NmnTTopmO5%WvftJ#Cecc;;4>$x&yAvjXq!M=w?A8Jql7q2sG^g9w&q zJNsEiE{?|^mBpZNuy2y!vTm^p^|c%)L~0N2Dz(&^>P)@^a7WcDjJ9+aPZIbLUr3b865T4uQ9V2H;zH{BA9RlM>;cEWCEfUWb{z`GYeA+*Z^+OMBrb7C%N@I||LFVYgW(GTYJDA#{MyoJ zCp7G!Rmr=(haO5Z@|Qq19PyUbGSC}}m6a6wR{*lhlmX_8fP zhTa&e)={VsD&fQL=eo7n91AQ8??+MWMmKR zE{ApJrTYv&z6U`1H6AVv`l@iY2f-Ps!W`bkyNXHHeSU9=W;kD#hok3`w$Bxj<~~kD zbdHnrSH^LvK|jyOeJ_ggK5>s6aI=!ErF73Fb|};@FI14a`-+8kUodqchT6$zeH)0K zB*1-=JxOWv%l(qb#OYhOv?*)OO<>CLbf=r}_*S(BFP9zjt0%AaBiVhvL)KCqB>aH9 z(V}<&oYnq?H6aM=?*+66DZ0Jdizvt6=4!FcG(9x5#Zzebk(H$|FRR3_(^-a!6U}QD zYrwtjDg?bJ8evTsENh>qWlbKn`cVD9jym^FhK$fuqotTlv1q%^k z`wI;4@rmiiQUWmM8T9^O_OYRe3A~oZVjW=i$vD+_c-~3? z%;eUQB9e;Vf)4r3y6JFq&E>|2i`RyLLjpPqY`X3x@VnC6QhohUWXpM#=>_313lFNO zV5I(u2B7l@>!nMpG>igPwK+}WR(I9a`zz}t6+At!U1D`#-7+$J=YD3Wf!L{6G9mWa z9o>bd1-^!+4Fs0{%M%e3>n*f9Hb@yo_^KgZL#;q@AU{ zw!I*koY>@Gihm8qL059c&!`jJ6yJaf2f`f)FOZ-Sj=7JaDuDI+X0Sfg2hYLdg*G){ zssow#ZXI}37+WiaC0jQ0@$)XZ`nB60u@YcP9zLUy`fJo@@&l5dG=CE{CrK37K27=4yZ((H?_U5xVXQ;_r z=?ffTT~1mqXPrJxOE#1X0>0EgK;1dGPVibRgYD-{$QhgaAPRQg5h$p*=M?fL^xKW{ zszL6xIw2MGdXc%%I)+D5UNO3NXU( z5GNNN&zzP7FF>Op@8=nb^Oo{@t1(HvbR4+7ci$SDmEL&RNlZ1ahqZBS8&Xyo1m#Q` z|C>;>U^eZeryIutJJN>CXl;e&TO&D^l~v$43g%rt_fZx1dU+Kh=vJ99U;8^D{d0~w zd33>5Vn-sL7pygk>HjV6^-=elw?bijuiw*%Zmu5@nl{2Jxs9)OYbkYbZ72BHvxWtn z6LTj*21{u%z&{H9X)2=)>yGx{x&u8T#ttWpWwrI0iGCkQq}OqbA4BEXczyz#rVYdr(Ae!RVF_If2t+Wr!OnDYk*)lYcR+C2Ob%KVXXETs||S zhmy;A9m&Pb>ROK$2J8L_ePLgY`6p-Us_WVj)qRlaAUZ2x?~7i3_CvSA%?8F@U6`H9 z%33GdbVQ_m=i6)7Z{wSqjV=vQP+%^+TwmYP{u8nw-x-tCo`gPx;ow6gCoiz;s4LV{ zF1a*ayAkq&{O%a_)~2}Q=)+%saGCn~V~#6bd9)YxCHutD+n4AhGt;? z+yKvmTqjcaGO|)5dg)5`(0cA{yi;# zJbFKXh}+l{_L4|G;hTQz_OQCBj=N*!3ZgG}_Vbu|C3aIjkUoV45xwU(l5xa|p67sP(>p*M1<6zA*GndGP z*GG?x_xCjm`+mK(nEpxe2X!Bg&kIh@I&K95nf)*v3}JZTFbp#v^)UUSw061Qdvif_ z0srXZ0orXh?l$ux(*7$abZQ3Whv@n6Dc8%WD%vfiKrMo~b+CR&NI*FHE8g@X`y2YG zp&WJEs_vXr7Eyn%9;VJivJ1a4=!P*Fhp)#hv0Y+~Unjvo`IT5+?jEetk5j)RRJL%t zt)-c7E__f+^~yk`!mrtXa%?apFM|3#TO8U5^<>3$%q zd2RagIsoqu)BNxH^5jUT>(9>*C||QRD7Raj`L2RI~ zHMxUj5we-TLcFua(%30T~7*#s}Zy--ZijGhY|4 z%F3FYbR$h_wI1zSx;MU?} zmU`vYpL7PD3)2;r#2yA>B5Wd}^wc|beT$~Rs70ix{W&@qS}*a*FgZL-(dA53Tc(!K zBYkIC&~1`-i@X>c8(jX}`0AYZc?~jX7oQsS@8Uju|3EjOCWUxTw>OY#RiyjJ(=P(A zCu9`f(l_HYak3c7Asrpb=6`&RCya}qZf;qQOT7vJW9$Z=I!&T`O?>7 z0ve|BFXqEcSpWA_f6B^oPP;dDB#znN2p-orv~7Yzri8XFjWxZSUqCkf$vZR(64KAk zTf=eX$-l??&Z6~n;aZeS6!8=4^61ma$2x5>R>DT=I9e57Lk^RfDU{6E96i{K{E7{4 zh1@M42luKs^bEu;#>Op*oy9{_FPTz~&(4lnc?{=s3%Xr}93=%ML48B0Nx{QN;)8BsDk*3-s*9F_EXX4L8{I8C+qKibc(V1pg|nU^(K5Gpdd z*jvqeIb5?wKko%t$lA(39<#dc<&HA}y+lBkn3c8`TH;!|4O`|xmVx9;?2*Gu(} z`pVMsZ|F}h3dD4bC%$RD#0;k5hnq>){aBXDbq3YgW?={VZJ})@s9}j9`A+le_uW+$ zPeV^i;!1M8Yg0$OaldWhe{11hTa=!~P{-mhq+^fW0yl+(yK^Q%cbPIl-`ME!v@YT| z>%L$SqSeakzwe^?F;)LDa&q#zHiN(kO2{vCzgO9QrY4zy(FZKZU(cTc7qK_MwFPPf z5V(V**k8Z@2);;z<324_WLqkZNC4ats&Ik*js2{y+1<;pg+dEK5&>n-f8i~%fL=7H zz`)mdb`0cD^F)Gku?_=n;I1Ptmk^wJ5d{&5Z+ef_WA4L}n^xY>!a)Se%F(pD#pmVC2MlCKTo81S`HECg zV71oR^enX9H_$~hsQq)>6BvTw-|@ffBAyP#dT_oK-XsQAyGpmvv+ z9s`$cY-y>};|R7obonJT1-^|ISS{Ux(Ezpx;!!{w9)5%SnBu&erp^~`( zm~rgKUxDKA>9?LD2F_jA-j#Ykg>VoDHAb{QzUc>lpR$E_LuF9BMM67Zd{0C5!4ZSn4OWLoL4!NaVF+Bwaub8-uZ2@y=*h&RN>{!AdKl0Ci z$J(qupM)U|-)>QnV0QBh{&2=(uESJ>#sU|*lCjTEx@!#7~FB}LZ+vtlr%I3HfqJj>?>H=*y?%Y&G^`fvGR?L zzV9tos6+}46gB%Ah@N*PFJxRDavPGD3)p_mx_tb#jy^T{4?N2A+S1F*`T?x(_*{>t z9T4yiF9oa+H)okKy@!{R2JlS*xcI+)_YwMAJXZZ0+8)mzYWlX^2N0`3pg<)RuYp8# z3LHSzg8~Bsq1D*$Z>_&n;As#wFtM=QwA{e5d8{|9L?kA*wzSf)U{%1OchYCL$$?!a zV>o$2F)P(>{S4d3=vRq8S6}|bx@}Nu3(kM&Vh%}x?(Tamvgg?&52soL^i&6GfG#8h za*0B{k6qDUpxv66&N_&xtpNI6h5v;Xt&Pds|0AvHO&q5F7NfZQ38ziG=Fo;{MHklP ztj3LZZiX~`Hs75|Pd&;|zTjClCeUP0pYym6>%U%2B6s5IyE_L5rhD^n&&a|XWb)5n z0rLQH^Z(O=&~CDY1lcvceCK2brPWoy{&v67h029Xqtn8E9N9L-nT zZ%8uP_4LJa9HGV=k1|_ZI?no^3@HU|gP(NKL7Y#GjOQ}I1L!={VX;|03yM(Nzwuu* zZB3G1WPO!OL`?tekD|-P!|2}{Iv3KQ_c<+4rr7dm!)ad)DwGvF9q#S=Y1sI>4R~^- z_6yIjkCe|J7o9=|{Vk!gxJ2*my_wu~IHaqw;$Q7(!<%+KTZ*8KkR(UtUj6na+?}|+ z46o^Hd#6b7FZ#eU@8vD-Djc*Vlw?(7wV&myAE;tMTUn<0YbyRZ)Eda=hm8yQoGByM zXzs{(y928)E)L?(Z6pBn|kLimaB8}`V-NI40%RwJq&p-p8e_&pg*#)DK2|V^SSd8 zfC28MLb6(raCkIm&ni4Ex`CCcB~|t~x^*y9OJ$&1M@an#r7ZY^VZC?o)wZ7f?Z03k z1EVj zabP?GZKE5Z{8nWGUTQJG00HxIn0H7x4xN0e_F1GeDS-sXjxodQ!M8JVZq8(XweIz1uC-4#(7uhINGxWf@8YRNXFpP0QA_6O>~ zr34^DACx)3^&9410!lBi0D*0h3+8v=cCc^24TQ#oYDSnNaGuks0aqogm$^jt57qKH zOiCedF{2Co8Q8WM*q~Tpy#*?{Q}eWDs`@aP6m@!bSo;poAhWL-0lm8u>{DoeEQiA0 z5#-VA7Z^AJN`KTp=4EE4AJXI@i5N5urtko(*kG{a6WyEyx@Sx+(1Fn9LM_eU*H|6a z_UPUYBXDdDfy9=YmWDd^A=^<14ARolSYyl0jMG4q{4aaEG?O5Fh{| z+cv1ZVxvtQJd&5QWu&EFB2Ho4eMKYZUFf+>1#4-(~>IxSO^lL zCoXo|1qS&(Abz8Cl4aX&5ZTi9_7*RlWMWBI=;;!Nl?WoQ!05!p6W2906Z)2c5(Rxo zX0pK60}6a83hfR$$%%>bW@cuI&7r};f3i21*P$L)+t=5ptegzC7vKOb4198ZEFfW9 zlvap~P5T28TaU_hJ>xtVXa0a0J4gIeMVAiX2?~^~j`k2W=i?HkJA7BYLP@99eqrtSGB8Nb!# z*)LxvdT@o7wT{(m@IlAQ$NVu)vsU77ksJ~?9_NNkoWK0ZyZvioM8pb3VVg$?YUo9Py2c<%~#!y5=a{7UX`;#)uROzCjVnRN%UqlWIc14$ku}k%i>K+Ae z7^?xdBQyOZH`9rkM1+O|iXo9>e*ue36KvD(y4a}zk5;%Dx>^aV-Tpt50-d&j@1EBT zLd&W?RB#7R7Q45f{=Nc7iFhcM40A)zauzPGFz6(zJi?y?R{XgJ8k0>wPE*l1%7*s# ztOTrVI1lxe-FokDFNMtcYj_xk#2LRC+NY$6Gl{V|0ogQab^$g>cZ$rQ9R)D1xpZqiNR584t~PY$E%qw%#V%*Hc8wJ2_-ZcAh`eTkL z5ZaH|5ALPeZ{FF=x;yyyl2{49w2t5Dm3rg=$?uC4pz}pyJT@Kto zD58f>zx6`ic9p9g6Tt490A~v<)M8{B0_-_xn&tr!qZ_(=Kns-)gHVws_)qlC^2PT* zZSf`GYQmcCzb7MRTqAN=e7r1UF zXbN6{&6-QoD0_1Sh$oRt5-?oAdzKt|dLODBfS^V|NVx3U4vKpJhVe0we7=UR$Dk{U z1{**qHHh%;0oNK4N<|zlC4gZ%+DAlILA?w0lNhkjBk|HhL_=bW0mPR~H1$UVYzj0|JmAtRLk6$k|SMMI3l+ z{~u=H8o(^rn8DxTK_OcDEZFo`EY+9hErA>PyPD_pVYh_OOXYhoz}8e(ZEz?%#!)aL#}DJ05jDlT2lG-L_ts zTIgBJ=Ud7)a@v3r%%vR(*1vW1cxRc%0>%yeiGwWGfE3RBCiw8mGE9En>o0EQ7Itl% zG}uyLTj0ft@ev1ssxDTE13jI~y{q8$hRmG(q;p;$z_~41nYmuTBW>tXQg^)^f8*(9 zlEH^p{yiHwPE8`+uCZU;kNN2Fn5w{}`zpIW3%L6NiY%LKAXlZ?O zANR+P>kEA9KOI!_MfF3Ck=tWE<7N6F|EHHa1w;Kj{ER{os&;=r!|V!yZdfZ7nC{Gv zoqMz2jf|-9v^s>tT^OFI%U{Nn2Lx6v3Y8frx4x()f&=tRI7utWU6FgR&<4z%vPX}E zjO}ME_czlQRo?MXJa?u4`}n)KxWE4Nk&)H< ze3N+R%di&5aWW{SDPxVpv&y-DvzMCEpmgBP{uTY_TP~?7DgT(tA-aGFwQ`g`=dnL$ zk!oJm-KUc#-bFdjKb+{y+EVkk;@HG%dUn6i($nrX6faQO@;*(5)6hB72u3Uqzz`0o ztA|+6WQK}Lt;MRuJcH{aH5 zKP^DNtDr#fl2h2bMVe}mdtV)Wa6N}NFewS_>~RS=(6LW(UBzC7wqTOB-nG~N{$4RO z`AY^+)jqsBbXYoS@ps)k@9legz&`Z_pmsk|IY)_ZFzmN&OxLO_^HJu@&{L7QYwhsF zro7)Ndj=^;{CXli#`#3w>YLdc#+pu&Y_=lhmMIHZvOs*#eZ3{zjt=SNiPDQeC> z+@ky%+ZihN;YwrZ2xh;|yuTU}ZX&ovx|>@*Gy7vou328MVx`JGh7UZ6_nM#cae&1a ztgak`9%vQB5gpjyuMUPy(l^S6=@KiRY2g~0$@QxlzaQj}S$Qsi_uDH=>ue;f$0rBdphH2T= zRHkJ?)ZJuhF8bTf3lgL@61DT8aU7a&ujag#rd5R=d8$uD9X2+@sKA1b?n-QAWTlpp zgtc`6!~yn%ij;t$>F9#<0#Go1A+!qqA=UH-GsEl&h)4e%O-?X@xSbjtaCLiT@{xv` zZg0dROCe>0#G(X-?mHA4{W z<2?DSDfj2!*DrZCts0dfP?(5`v73Uu=ro?Ku1eo4VDNt1K7tk|s}>Z`yPjcg*g%f~ zWmQ}qvFZqOJ^YKm-?*1o0Nvd=weczxthm8;l>{@2pdrkw;xpx7@Ol5wcl&@4b6r#4 zPtMa?uf|h}{PaLmeQoW-GvWE4 z6h%Tt##O6)w2~(jetaul^*bJ7I)yqvuF$iefubCQ{@EFl@Pug5$)A3S5kon1ZwT)Y zpDpRsAd;sW{4@*WJL!AiS?gDPt+zM-Z|jwd+21}Tjm+Eo@WGFTv((2y93;)g>LpdY~hW}&Rc>FDjyXsX*z$b*u^#5~VC5S;z zAvs8osyzW_ljPjtzFd4fo$%6b3H=_nsXAp<#R!cUkoI3RZ7fccf~$8Wnh29#2e4n5 zz24CK(}2(D+Vv=8oAKIBoxo;xv3vQT?`guoxZHl%ix&S!Cl9V+H6TE@xh6_^#@(ag z>E@rWUd>x1UrR4QjS6{iBiiPre7Ft>y4>>W8}~(bZ>g;ZhOvqB@M^>i=>~L zV*L6E86r()9L9a~!S|{PM=7h?=Q8Pg`ZevQ%t1%hl{lj1^BJIS+=JIXB1CqQe#IJs zg855nhGaB4&lhymE#)2i9x+$g8-ka?K2F^wF?%>%;hG%k+g*A`Loe$oH|{!{K9gYj z9Yn@4_k;wfFbBg$v%mMl(1u63E8W-#a8TdJkKb}zI$VF)iSyS%#1|pQ`__s#Fozc) z7#nLf6RbM9X*pSDiK*gM^cR<0o9%?kL>zQhh1~!QsZm)cj4$my4MkO)r{&r$vlDGF zsd+vzJ-U7;eQg(amw3-e0b&eu@Bz09dX*+$wk9nZr@wd0vLV-aq?xjQ%U1tUs5M%+ z-G@$|?S|EG=X)Svb>?`Rf0p+IXKe$1bB6DIE& zGrlvhEWCCuXC%&#z_XUy<*{|%ZR`Hu8gJo&nZ7#6}CD&mDs4Xt` z`f|ywP_*R-w>qf6k{_fQ(m4*>x`gLD+w}}u@DAroiqPNg@=Lu{qw4XS;3qDyfkHxQue;kfIX~3ulPU72 zW9zTPqos2ljK68?cN*gP6VRv<9)4tRc9xaRm{@Wli}MAA1}#f@$!G%S6G<)9*2X;h z7m8u~ccCQ<{!(TRmKTO<@<%p3`cLWm5Pz>P4&)?y?F7_HKk4ur)Nm}_d`P} zE80C5wC(If)5^CTf4>|0^=G*NTD@tgk~b66r?6fAGu#I6Q*%c5_k8YkDYhkxy=`4#(X6ub;Rkxqf=V=w#J&#)(5th z3ivR1sRLvciDj9u{??imh6uF4<*$pJ9{Cd_vsw(%N;=T@i!{;#mNhD9NpYo7Q$_o|?KY~Y3Z!faw#ip7 zNb&a{Pk3rm>NVOyO43VK+G+fKa+z^CwAUbb=*JYrqlwWYRo|RdA3ucqQ;7`vO^<~F zuc8wfC1ImmDMR!d3zG$dz7g^UVc*)Gek+Tp>+r(PP+SQ7@HHeMxniDT$JVZ|dV6~G zG|}zoPzSMmmCxqoK)-c{i-4c7Iv1-_ssen9i2|9YO$(<9$G%i?Z+|;_c&WsPuM#Ic z+Fx+_Sm~!1#cKr?U#TqD+)lFG)cg({@foScqiBmEp)NM48Eu-qCWaZd(ANtXZ)mO? z>8hrA4d0eB_wO~ix2UBy?7TuXgt4&rj>|}jt8^P*jF>cCWI8hFu=LboakrNS zD1dANg)t_=BWBva&NBF!-rwry&HvC+7Y@DAv$^(1(&FFc{WTcH-FC99_`1-XOcNhgX;I?b~UyzlLW5&Y!=(WxFzeG>^kAn+uK;%w3ct>E3>RY^j-L zR3N6kn%QO#t^nSU3%dD=OGb`0_dfV_aWDPrg1x;xl_1Bbmm%Wlf!L{|kdbIeM!|fZ z{lsdzf_G1k%INq-5sr(so|6yB`uk_Dkuf_H-|8H$Yh8Wta$$yrR-3%4Ey^B_?V z4L(DG2VA328!PKm|5F5!vR}G90;HPvHt}^RC|`RS>yh-ThZY9ryLKzZP#Bsnbhj5> z(00x;<-LjS>~9mgGgWm`Zf>5U;vNSo+0vDj-}FaGL^5y{x0tRfm=0T9DXPmXP|oIp z3uV5FquGcy1V18)Pp#+9B@GdUe?J;L6nyS<&@4ZNQjSnB9W`9kEbVb9oWPdpaFUT+ z_eTh$UZk6vX|mc+mzS!^)xA{cWI=uG;OR)`d0|y+l@b<4kbbtmBYuJOQl+Rqoo+oj zCwz^`*?%u0rmF~=tVz*5cjZgju!_Eyd&vn8nToYuS}OEeM29c4RN8C(V7c|g!`gq! zE)Prhr5(3GFx?3%>p58%?FZHrXm(3FukHk-ael0NZC6p@&;FO#9Y0eoZ-0_8-)Pu> z)QTrc!O*&_yj+|mysoY7I4HN$)X6p&Arxoy2`+U2(#kp#=$ z@t;2fK_@;2?Vg&TUoQzgz1AD+3{ClZcZW~-OvP)|pM|l4wcY8TKYB2T#^2=l!3~?_ zdmpFN2PqQbxC;vlqYkzLf`WSBLqW5Zr(SThqoV`C{O~JOFb6?89(4A~d>8X$y2D^+ zJM{ULm6A0K<|{$fX!0k!Josb+tlF(pkSCM$-UT^@fZZG!sCm-57^tZHz-Pte+cYr$ z0nSsw96xrfey%r1QAGs@obpQCpiFc3?ynlDuSJE0SXegYbT;Lhr7yX!TIu#Gouf#< z7~i;;?47oehbzbRcY*af{iR{X8j9y_KTb(-@(by9*fuB3h6r|R8BPT`xi9a}X-SnE zX&D)>Q&Vw>iHX@3;9t@*GGEH>$jSyZ?SdLm`p%tOz)bI1L;4Z`6MF}{ySkny=j~i( zlV^BnWKf=p6?Oj3?&nh*ql_GPmVMGjkU_*5=L}AeqtE?v zPUH6m`&aSy>f@-yL@gP`uvhv=&`zG|6E0kIGvbuQn>V?;af9Ub_4S-khWb$hcRDsX z=>>MCwS$BAGt|w0e)YYx{$+#Po-1{e`lG#pfdSU_9HDkD$1X{#SQ$I#Z-bH?(KndG zpyMUp_Sy1;817-N{jXJv6%22E7)eG>&4wQAt`6Yd3}R1mmpCJ!rE;XZ3LIIKYM!`J z(bDE5nn%XPW!pzZN`Qk5?;1ZBmpCXscoq$ejPStOxWxYK*|UjzW6xu$$34e)1%^IS zi?MH>nw6sd$eovJ^oPr0Dw>mPn(>TbnDjaFkM8D8`vXt^IKATIN9xDB-woiVO+?NF zoY4(>z`rDME@Kupr<%aypnUKHHw`ym2}wLzw#a%|7F=9hR3JJn6k%b(gHDP)Wuc{| z{PqijXsD+&MB%bF|1>c0IMx^ltI(1ztcj1O1*U^1=-&09xNl`^+xY!MAQuP= zcPc&JYK?#4A%C*AcF}NPe+&Jwp~j(z3!w8Wa5e}`-~?+_7m=% zIh%#_f{Q7Dv@{QYcP=l#K7B_$tz-rtZ= zsAp#O+}|9~cR+;YIS4rDvszvrEYJNn;RPS03g``2A06za=nY}wXdWH@5YZg$2k0FS zKg&T9qNo9Em*>&(_{-|>e93WfasLuOJ$QBIJ>4ULME61S-le2$T{6iW^x_YFaCYUS z1i=}mXxe*vs22|W%t%43rY=rMp!0iiT7XtY*T6)bvG1V_m(6c_9CQgVv)Faj$&WCz ze+mlL!&|3Kfau-S_jHbAD(2 z-ap>6-ame8{jAkfkHlxV@9WyvzOH-kP(*#GWc=8i@FbySbi0s{#5L{;%fhKEgv!T< zv2fv84Hsw|?1W+t&Zw@#0cITR;C=;9R|x#py?bXiFvPyD!cvu%M@5DE`t^o|4O&7f zHXGJb)QT^WJd6+VRU3=7TXEVz0x~sl2*5ZDg8Ap(hx9E;;!S-O-m_dcc@yfMJv%h8 zM`sv2r3p{IOn%AdkXae*ha*|P0G1LeZaQU&-ZKJ*~^zN*VOai zq+5Ln(#fu3ZT=|r+0UDJ!_1SRCAbOAro!J3EybEJdS*~$tz58p#pLkejVUQ9P^w(# z>N@!;Aiw9F{A*m1;zu3d7uUZE?{ZY5g)dosdwWSjM&5g_&70#>QZB%u#3R3(=I^ih zJLHd;>Q`cw8+8`haKnHcdCkCCUspFgG&Hm^5E3{?jvdSDIOghVk>hD*Vd2g2it%4R z1mD2;@JqC*g?TxsDg6Am_b;DHuI2nk@jP98pf`g8>{Nl_1F{iQ;A zk?q?b1E&oQ4^!i|pbfna<9Rt*PP6*mXWaVA8Nc+DhK7f)Ovjb!kKrQi>o zT4>P%jktZAUw~;1gz%Cie-6#mejGL}!mFqp`Nwa5HbGpIIUQ^Gl;I~Avr}4{6^F@L zVAf^_>z+CsobIdiIbgrcDSLu;?b@{x4$UlRa&L!Zosf>MZUgKca^=02+NB*0vve2- z&U}xBKBvmlf#BsGKqO%pppZk>lh!I!sr;zJyK9%Blv|GxXSKS|Y!wio1$LOhne-@JLnOv?+Ou>yC7fQRyTn-r zBdPIY;9v(v>Q+)PIv00n@%y}uw`YF$=B9x6|NO1Gdwo+t?s3s0@z!vFGQ(2JeNgK1 zWsH!80Gu+;{1^ z__V%ruU*=PB@YEFO`kAFkCUz4n7`D35s+~YeE&pmNI4&Jkl?uYO~=LmcTt60*JPai zqp$z`t>1(Tm$a@r?9z`cWLY@+svS9!{%=7N_mVS94?GL^?kI$D;g~lEjRYLq5bSYC z`#iv)ahQ{b@QwQzOBr66ous9wFPdOjuczTv=KV>JY#!)w?1lUll*HDOBYn^#&WpP= zo6GlWqN{)tCnrOdPtU2+o&_0m1fx!4-9+KJ zlGv&~Ky~m-BJ^Hxrc+K%F8hI^oE*>Atr{b{aZHELc3vuW>C;>oTkT&0Y%j8-W>(C!g!a(ncA%Xwy09poX#FHrj zrl{uHa0VAQhuQAVKYXvH9qBNYtEsNO0+02GySt6?5c{2PAM=6faUfh33Nz*poD8gn z*afV1JECS_idun{N0hYMao%3kvYY_}N_o-)FD4R-Fua6GNp@n{3P+Q0NR$mjUL6KN z1#t&mMYk-wiv8PGBRlS=aZFRNuN^L-t(rQL9@p2Ip8zzIH9$ z`cZ)K#=KkE+Amp13P<++jX`5FPL(UE6Pn@r5_)MfaJ(6YKLxSO|ir85x;o8G3qpT3(M&FE^K0Rhc-ySTKq|(|S{p4Zr~B zRaa34Gn1^(WoB+Y6G%=e+Dvn5uK%B&t^1urLWpn-%*?ktwlFh?;zc?4HTCvd6yW6B zBEiMiZrnI*mVX}moF?@Qr((blr{g^ja=NU%i(Fsk7Zi97)vbgJ(&1ka){n-t+A!iA zg24fw@IqPEUn#%l=La!k_3Zg`gR-Tip94E3CDkr()6JE#4$RZ8L5JJ+=pBU>oSd9A z^8>Q6cXG?PmtBSb+Y>Lsr*pV+y&L>89EDD3RKeST#IP*Ao5G?yc4!?1Fa-u*1!lQphxgu+;$rA;+t}kWWdIWTIbe`hU8t{Y@Re%r{0@l3I>rPS5lj*;% z-JRm#=oo|fz(~BqDERC}!}VsK+1k=-pyI#_`40Ey{v1XfcVjV=#bs+2o3Up)4< zJJAVYeQPgB52kR#pw)MqBda?Ur6|b%Z)1bx2Pe zW;k+HJd`_Tz2_eE$mOuA{Yk<%)l>OnAnC zcVN`Y66)iuvM`sSUy1R5GRDIMjsyN;+v@7;hbAU+OKZ!1zUb&+f}K4RNP&0=g$9Zb zt~9htlhrVonx7dl{)4HK#C@;j1QPcTp?qiL*BnGPS=dn06T3y3BU#2xO;wdb<|Kh? zl6sak#3eTkvzdrTvt_>0{K2=g-It ztr-x&g+cWvm~Se9q$y5@Z!Re*8S5%YPUX+vLiiZifK$`Z%s@a(YhcLf0gQsNUM3XR z;W^Tw#tP`D_V)(p9m8cgzsXIk+3P0;A%MLg2u=!HQ{5Q2kAi^I&tP%nPL45PGD%pX zgDoZ}$2oTSGXvxJ?8GRn31)b|3uoX|C>lItX|D8Az@H3_4;n5I7+Q8Tee(!e$>O$a}Scq2d@gfU*@=|`!PZmjboxShX&E`$CqBV;HBB5#u}tADAcGSciB2R zt|m8DS!oBc2R!G4IYowIN%b#ZZUE9Wan{F=EmZcJpSDeO1Jpt3tMhZjuc_-KgyZ{DW=Zy7Jf0Y4S@?k&H`qlMi?9$qs7L0GXsUjrlu}Fd!|iMRaFh~O1pcpDSE^kt|u_r zS|=zdh|y0XSW5{3tqQk;0WE+kTYGyNZ*OnEfPi6W2x=M_#5aEGJOwyd`zIPH{uc5j zdA+@5J4oCBsES=e5g3I@^UgdIdXy4{g@v2sz3zBb`}+78*Z7_|al-tQEbH~lN{j5# zv9aNxbwGDUVE@F4OxDG7X<(K`}?!Lq+!e%W;az|bpqaI82YZq39M;Iv(!EP@Vw{?^c|<+PXPfb z3IgzXtRXR*nhS76vPCFBjqt$h+-fBu>+rBJxO>u}c-Dj|G5|vSl`CNw%Y^hACqgPf zZWB)WdrWriNSqZ&-C?OD!p~1*RduQXC>W={Er+9m{+yf3oQ zS5j1UAeEs^BR8r(KD3x*= zI5uuP52J=#fGL|K>{j3^LaThIhyXAN#_hKwy1Jc@9wq9WzcHhVsznvym7?^kUXzw; za360rgw2}+het++hK8tCuU=hn(W1Z20)lD~ayo|%W{>$&U=oF(rCAf`Y!3qGWbW2- zCeIK+j;tS0sp?4hP~JWd^J1c_OXz*dAOUQsCoaUsO29U;yu6%e*Dhu@aqFkxIwVl1 z(kZJT?y_x(}OG}0h>m$ zg`u|Eh!8}Q7T4}Gn*{Y!^QS#G`FU}Dev`)*$E^~#VY32xoy{}uK&z3_(RG)2_u`RJ z=OZD`;{?{LLMufw?@Z~8mew~sBgDkbeG$4QLQR1d78a#)0dt?DKnKCki2G2E(XUF> z#8Fs^w-7qUdLN1doq7U(;e5ytx+jrwabn%BOL(W`@_?&W;9x^|X{j_WA`BzBgP>C9 zakkHRp)Ez88TRH8h*2Q%S0KG0+>PJ~kv>mPnOjVb;axif8adyuTo5?SIAf*X<}NHL z*#3k4UTJ9qCemXdV8O4Tz=3M!;JZ&)y>kBFi1RAl>C=l|qaV!Zxp%80xO&VCYoPDL zw|zS;{_jxSB90;5{}vc*j-98MB6oIo;KJj#Z#RM7ZGfWLYBW>`d)*u)FE0;px!2B) zuhMra4vOd__=Sn&Jt$zC?Que53%CG-#cJhb8;WV=Ng?PR%i=&612jVPlmetzXWL=BH6H z0g^Fc1sXcrRG!SC@C3|GY{la?F2k&9uF}=Cw6yJCrYWizMn_T4zDtwBfP*^1U$6&Q zDPdJrRbWdE$}Vz9@24;zq7GwiHqSxZ)z;Nj1nYqxU84t$=^kURc&SwV{QM%x20jh) z3+X43!UNLPXbwXDj2QWM*j;um>*r?}3$dbNWq~?1v*dLhxJKDHq+uqWd_yF~r#NME zd(L6PtrR!_7nK(3@Pz+%i~yvQ$&C{$)YhzJiC;+{k$*?bE5 zs2zxGW`4##@ejVoh~qeskiSXc0q2qEhNCA8Wo2a%?UhkTq1QB!Xnl zF{^!_ia3|m){nM^C0M9DLjz*%Ab3S^E04N zEk?dG8MJ1}ZCtf-WsY02^64@P51@0RTlC0Bm@e_7fHz*4R5;Pt&>+O^4xeA)i>&PI zLX=1e`$nbiNvIySt)R>Rt>tp9Zo8={h)1Z71+Q`V@YQ=(y?%!{8mfiG#Pmi^AvaU0 zf=$Y8y&T5CCnU6QM-rZp{IWbMt7aKp`(DJZm6Ucc-ialqn0$ocC6(&&-ZI^HH4xIn z%MA#~jg`j920m~YzuX%({rpj0G_B;#q?>jNjN)2mX68gs8{2Omo>PYqu-sXue4uwR zJJK01TKEenkGkj2_enXuQib)iP%mr6SI}5~KgQzkEuqk=12+H%;zOgJEZ-w=AWA=T zX}oO#HDJpH!#y+a?v<1i2($6z1iuxjB^Y~QZ!rj~!r zNm)_eKOi})?)`lh<|d@kQa3Ok8!1nLw2ga2%pZ5m?E;g6QaAfv8-O&Hhi%=8(xxi( zjCM*$Fyb~1olRIo({`vKV6VaBXuq|0wL>plzwm-!1(l;5bm~-DacSur)5q9=A||#6 zFGE{CM|U1*YU`>ET-za;By?2rx|iB7TcTaAl=` zqUBvR=uE*c<#f3Lu6CA-Gmf7}i8H?M;6aADukS66)xNx`==%N?QVpEjX~CYVLNyUv zEU_1qz_(?b=DvWHDCOSI1x(S#R=x_WPSe_Y3wES*|IJqe1Kc2=>ya{y#UyA#kdCrv z-8=5A2@yvFB zSx5&0LLxtp9%A*QM;Ie9E9{;cACCg9DZ;i!Z{NNrqb*5js}hFD4kZ)mprT4-U}O8) zHyNW0_xDpfB@bDZ86>qu$$&O&g1JR|<^i^_u&{J+Vbim-##Nt*11}jnKvE6E8P1V6 zCN@NmF70%DKN#;pL}k=4x4ij>_+88XR@E81;3(`uwPiztMJ4(Wl7z7h(N2h;TwU zFl#k>rHVZ|W&kf64jed;Z~e%8xHrVDigv8b5!7o2CG1dZwind0O4iGby6>lN76gMyhac^9{qU z9K+R!_uBjStAT+)*+^6R2Z6Z0CU8-hW%2t<=csIu>Wu7iZm#e!+G<4L z>60f^VAMI%9!jDb;TI5Cf#QNVg%GO&9NW@lPvTpp%OP_-5^s%1MC5?Oy=)FhYaYWQh6>y=dFwc=_eNo#G0IzlVF2B?DEXy}9$mR& zMLMDZg5Dm86kuz>jikuO%apNgQNDc}I_wyMCyOU&erPqCsXAtU7UdqdympAWySqb3 zfCv#n9OAy0r-`_@IP!mOop0Ki`Dxw|!I&AFYCi+lS_s9{_1~!KMlKHI>`1`x?Dev{zrH6#62XmZ)JEVbaY9dM#E|7<40nl zK7_Nn5hW!OXnL$fZv_K5S#>4CjSXl#4PuSG6i1XfGTydDpfhsf9#g znF>9$5iP}v5_ zf+nO;w`?5F_X9vT!VQH$YVypmFYyk?;E4xm!D#>%(s%%i#TwxDhUAn@l1FLp-Mi-j z9%WRvdC#6b=Ni;M|dX zVqCq9%}#n;&8+U=>SvI`>~YAk9Kiu0pPWYqMy*E#0K9CAU(Xj{;FPR_O)u(N-9EQc z$LxoEgx}S&l$%QaOjW18s3Bsbpep#LLAW&1(Mp`0DJUr5jn}@EnrdD62RIJ*d)87c zxRCF#hIjrB?r1mSeMB_*i9Rjv8K<~3(sl=?P9Ecn-b50gsA+L}-9^K$5!Xs@sdQy6 z^`o0oi&FH}1H2KQ0A90Mi-67}z;5L|z*fG6*I(}cJ?mTNWXJYx$3~aRv^`rUQ7O4p zvwEXl8FFvxs3{`|Oi7AUb)u*7?y1)-0~?!1zj-zIfL^hz4cd&F-S}6QAzO6SI+x0u zClUCsC+R9EDDYvc0%3<&RVfCcBjNS_qnYQgJpk+9j*B#`c!)SmvIX(H$^Ee-rZDYc zuAQ4(EU630+zd)sSTsZfMQ6|Ml9i39tyQU6nzK0bd!`AEt9qQ!FYL2^c)}{SVHjL( z=>b|yj~&6kN6lh5;`a6cf4uhK(9l`nEP4z8QBwO{ZboKHv2K>V$R(fQ^_+_Wh&F_@ zv{P2r#9FOL#MRXmtwnMN@ zR20Im7n-8gf}R;xN*v0G+~bQ_j5CO4w*Zi)XJ&+cPKp$vIHjV1c}Tn)PNT2KXp)jq zG-k+(kuua=)D)5i%ji*vfk}g1V?lGKtr{)L(_7iu+Wr<((H^HxX_B(=&u{-yi(SEp z7cSumpq3Eda+hXE0Yb(IIy)a&CZ(1-zg{&rHwPTKy3qPjuGhFF=xONjPQzBh!oR}S ze{8`&PX{ujszE_D=2gQwga)JN|55c{_@0?QzhwA>CIk2M0lJtvUi9 z5Hg<7UTnJ;EP&0YCVz9pEQvGo9$d-+EGXhlk{?nYM==XZ1=*Tg6XFM@WvAGvlHYWC zmdIxL6y41lbLW;(%fV37F*1y?V_k}WFLul?F52T6Qqm7SCM|80iG-a3U3LrH>S#c9^qGF)7FfcB^geX22hIhV zikyMA+fD+OPEK~|nw?yN;tZbgZOf?x+b$1b^n`RXZpgY5=PeO#ZyXROeG>rDh)b6k zOifLpzq1N)+Nm=;-h3TcW>A+>=!soG`m1kiqepr17=Qe$t*Pj{0m9;;AnI}B5%_+$ zQBM%Amw*@4&cpC;IScmAOBZwPZo&#DR;^lP2ivr_I30%aDI96i9%H!Ry$*>D7ffTn z^Mm#Pfyco%9Mpy9U4kO^H$-r-TA5J#g@eLS*U-Rrrsb45o;}UpLWjJGJH7@FJ+Q6o z&>4WO)DbKqhp)XA-j2j0lCV2yXp%lcUritc(H~U6IaLbj0+ z8B{``km_LGn!0#QO#KeO^82Mr*mQ5d;v<6NOXQlU5R(L9n7!@ zfr~R;t6S&mSwR4*N&Ls%+`^57S$CK}V1SHy6e$@d^L7+v zHk?75oeVG7!tH@DT;z+%&%HW;&^v?B5D8C*_B;ESc@y`Oc`smNX4K3=zG7hvc9`3y z4uMJb!QL)}2P##a>yjF7$z{C~7#=Jk+At*#)_qTbyP18c45AM+F1QY_gK|)QOOmuK-0D~nrd;Z=vjrp9+ zp%KMHi3I1K{i7C;8BBH&_#IZJD0PQa^PXxS%tuj&5csg+IxnA_fSQw>_+R3mkuHzW z*Z~Kb-sfo8&P;kT*t*R=zI8W%T~DJKdKNNcfX%a;UwWUOEO=P#IiiU{RN=NTe!luA z(tn+b!C@72Z`PdXua3ef#hBt1;fI(vg`+r{u|@;8#y)@^a&eoPnR$E9VXb-D@#n*= z@n?rQ{yNuz{yCH}Yc-LQ&|aBAT%S}-5uU+}Db@%VffVRSkcJo}0F4H0$M3ig)Bp`? zVSR#eh{9Hdc?%nw;LXY38RCa9zWUdwLN?QjWLbK|k827(!o*=RFv4)_pI@K2ft;S8 ztcNwFX^EXi#n+?9wKZco!clb*ttB#}RZy^F;pfyvSdqc}SYu!aa!Hm%9q;dt|J|J7 ze>MPtZaew-e=<1@K>N>+|9D#Rpa0-L9M1m#_DA$AQ;o9-2vDcRs8R5-PeV_=Q0>UM F{{n@sL_Gii literal 0 HcmV?d00001 diff --git a/configs/experiment/sample_enhanced_conditioning.yaml b/configs/experiment/sample_enhanced_conditioning.yaml index 14625de..cd787f6 100644 --- a/configs/experiment/sample_enhanced_conditioning.yaml +++ b/configs/experiment/sample_enhanced_conditioning.yaml @@ -19,7 +19,7 @@ defaults: init_datasets: _target_: jamun.data.parse_datasets_from_directory - root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + root: "/data2/sules/ALA_ALA_enhanced_full_swarm/val" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" as_iterable: false @@ -33,22 +33,22 @@ init_datasets: # conditioner: # N_structures: ${init_datasets.total_lag_time} -num_sampling_steps_per_batch: 10000 -num_batches: 1 +num_sampling_steps_per_batch: 2000 +num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: "sule-shashank/jamun/l1n5tg48" +wandb_train_run_path: sule-shashank/jamun/qiutegoj # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last -sigma: 0.04 +sigma: 0.06 M: 1.0 -delta: 0.04 -friction: 1.0 +delta: 0.066 +friction: 0.7 inverse_temperature: 1.0 score_fn_clip: null diff --git a/configs/experiment/sample_enhanced_conditioning_sweep.yaml b/configs/experiment/sample_enhanced_conditioning_sweep.yaml index a537caa..90968db 100644 --- a/configs/experiment/sample_enhanced_conditioning_sweep.yaml +++ b/configs/experiment/sample_enhanced_conditioning_sweep.yaml @@ -19,7 +19,7 @@ defaults: init_datasets: _target_: jamun.data.parse_datasets_from_directory - root: "/data2/sules/ALA_ALA_enhanced_full_grid/val" + root: "/data2/sules/ALA_ALA_enhanced_full_swarm/val" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" as_iterable: false @@ -33,19 +33,19 @@ init_datasets: # conditioner: # N_structures: ${init_datasets.total_lag_time} -num_sampling_steps_per_batch: 1000 -num_batches: 10 -num_init_samples_per_dataset: 1 +num_sampling_steps_per_batch: 100 +num_batches: 1 +num_init_samples_per_dataset: 2 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: "sule-shashank/jamun/i0rd0uoa" +wandb_train_run_path: "sule-shashank/jamun/qiutegoj" # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" -checkpoint_type: last +checkpoint_type: epoch=83-step=88872.ckpt -sigma: 0.04 +sigma: 0.06 M: 1.0 delta: 0.04 friction: 1.0 diff --git a/configs/experiment/sample_enhanced_standard.yaml b/configs/experiment/sample_enhanced_standard.yaml index ed6bcd8..027aa23 100644 --- a/configs/experiment/sample_enhanced_standard.yaml +++ b/configs/experiment/sample_enhanced_standard.yaml @@ -24,7 +24,7 @@ init_datasets: pdb_pattern: "^(.*).pdb" as_iterable: false subsample: 1 - total_lag_time: 2 + total_lag_time: 5 lag_subsample_rate: 1 max_datasets: 10 label_override: "ALA_ALA" @@ -34,21 +34,21 @@ init_datasets: # N_structures: ${init_datasets.total_lag_time} num_sampling_steps_per_batch: 1000 -num_batches: 10 +num_batches: 1 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/cz5a6xo8 +wandb_train_run_path: sule-shashank/jamun/9ewlap6x # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last -sigma: 0.04 +sigma: 0.06 M: 1.0 -delta: 0.04 -friction: 1.0 +delta: 0.066 +friction: 0.36384343 inverse_temperature: 1.0 score_fn_clip: null diff --git a/configs/experiment/train_capped_2AA.yaml b/configs/experiment/train_capped_2AA.yaml index fd2f985..c9a6e14 100644 --- a/configs/experiment/train_capped_2AA.yaml +++ b/configs/experiment/train_capped_2AA.yaml @@ -1,4 +1,9 @@ # @package _global_ +defaults: + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml model: sigma_distribution: @@ -8,9 +13,9 @@ model: optim: lr: 0.002 -callbacks: - viz: - sigma_list: ["${model.sigma_distribution.sigma}"] +# callbacks: +# viz: +# sigma_list: ["${model.sigma_distribution.sigma}"] data: datamodule: @@ -21,30 +26,44 @@ data: root: "${paths.data_path}/capped_diamines/timewarp_splits/train" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_GLU', 'GLU_ALA', 'ALA_ALA'] + as_iterable: false + subsample: 5 + total_lag_time: 5 + lag_subsample_rate: 5 num_frames: 320000 val: _target_: jamun.data.parse_datasets_from_directory - root: "${paths.data_path}/capped_diamines/timewarp_splits/val/" + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" - subsample: 100 - max_datasets: 20 - num_frames: 320000 + filter_codes: ['GLU_GLU'] + as_iterable: false + subsample: 5 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 5 + num_frames: 10000 test: _target_: jamun.data.parse_datasets_from_directory - root: "${paths.data_path}/capped_diamines/timewarp_splits/test/" + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" - subsample: 100 - max_datasets: 20 - num_frames: 320000 + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 1000 trainer: val_check_interval: 0.1 - max_epochs: 20 + max_epochs: 100 logger: wandb: - group: train_capped_2AA + group: 2AA_capped_diamines_conditioner_comparison + notes: "Standard JAMUN on ALA_GLU, GLU_ALA, ALA_ALA" + tags: ["standard_jamun", "2AA", "capped_diamines"] + diff --git a/configs/experiment/train_capped_2AA_position_conditioner.yaml b/configs/experiment/train_capped_2AA_position_conditioner.yaml new file mode 100644 index 0000000..59bd863 --- /dev/null +++ b/configs/experiment/train_capped_2AA_position_conditioner.yaml @@ -0,0 +1,75 @@ +# @package _global_ + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_GLU', 'GLU_ALA', 'ALA_ALA'] + as_iterable: false + subsample: 5 + total_lag_time: 5 + lag_subsample_rate: 5 + num_frames: 320000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['GLU_GLU'] + as_iterable: false + subsample: 5 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 5 + num_frames: 10000 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 1000 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 4 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.PositionConditioner + N_structures: ${model.arch.N_structures} + pretrained_model_path: null + c_in: null + +trainer: + val_check_interval: 0.5 + max_epochs: 100 + +logger: + wandb: + group: 2AA_capped_diamines_conditional_denoiser + notes: "Running on 2AA capped diamines, conditional denoiser with position conditioner" + tags: ["position_conditioner", "2AA", "capped_diamines", "generalization"] + diff --git a/configs/experiment/train_capped_2AA_self_conditioner.yaml b/configs/experiment/train_capped_2AA_self_conditioner.yaml new file mode 100644 index 0000000..7078f82 --- /dev/null +++ b/configs/experiment/train_capped_2AA_self_conditioner.yaml @@ -0,0 +1,75 @@ +# @package _global_ +# Training SelfConditioner on 2AA capped diamines data + +defaults: + - override /model: denoiser_conditional + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_GLU', 'GLU_ALA', 'ALA_ALA'] + as_iterable: false + subsample: 5 + total_lag_time: 5 + lag_subsample_rate: 5 + num_frames: 320000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['GLU_GLU'] + as_iterable: false + subsample: 5 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 5 + num_frames: 10000 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 1000 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 4 + N_structures: ${data.datamodule.datasets.train.total_lag_time} + max_radius: 1000.0 + optim: + lr: 0.002 + conditioner: + _target_: jamun.model.conditioners.SelfConditioner + N_structures: ${model.arch.N_structures} + +trainer: + val_check_interval: 0.5 + max_epochs: 100 + devices: 1 + +logger: + wandb: + group: 2AA_capped_diamines_conditioner_comparison + notes: "SelfConditioner on 2AA capped diamines data" + tags: ["self_conditioner", "2AA", "capped_diamines", "generalization"] + diff --git a/configs/experiment/train_capped_2AA_spatiotemporal_conditioner.yaml b/configs/experiment/train_capped_2AA_spatiotemporal_conditioner.yaml new file mode 100644 index 0000000..68319b3 --- /dev/null +++ b/configs/experiment/train_capped_2AA_spatiotemporal_conditioner.yaml @@ -0,0 +1,71 @@ +# @package _global_ +# Training SpatioTemporalConditioner on 2AA capped diamines data + +defaults: + - override /model: denoiser_conditional_spatiotemporal + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_GLU', 'GLU_ALA', 'ALA_ALA'] + as_iterable: false + subsample: 5 + total_lag_time: 5 + lag_subsample_rate: 5 + num_frames: 320000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['GLU_GLU'] + as_iterable: false + subsample: 5 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 5 + num_frames: 10000 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: false + subsample: 1 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: 1 + num_frames: 1000 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 1 + max_radius: 1.0 + optim: + lr: 0.002 # Slightly reduced learning rate for stability + +trainer: + val_check_interval: 0.5 + max_epochs: 100 + devices: 1 + +logger: + wandb: + group: 2AA_capped_diamines_conditioner_comparison + notes: "SpatioTemporalConditioner on 2AA capped diamines data - processes temporal sequences through spatial and temporal modules" + tags: ["spatiotemporal_conditioner", "2AA", "capped_diamines", "transformer", "generalization"] + diff --git a/configs/experiment/train_enhanced_position_conditioner.yaml b/configs/experiment/train_enhanced_position_conditioner.yaml index 3a38362..86fa303 100644 --- a/configs/experiment/train_enhanced_position_conditioner.yaml +++ b/configs/experiment/train_enhanced_position_conditioner.yaml @@ -20,7 +20,7 @@ data: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - # max_datasets: 1 + max_datasets: 250 val: _target_: jamun.data.parse_datasets_from_directory @@ -30,7 +30,7 @@ data: subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} - # max_datasets: ${data.datamodule.datasets.train.max_datasets} + max_datasets: 50 model: sigma_distribution: @@ -48,7 +48,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 1 + max_epochs: 50 devices: 1 logger: diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml index 45ef62f..d8d471e 100644 --- a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml @@ -18,7 +18,7 @@ data: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - # max_datasets: 2 # Increased for more training data + max_datasets: 500 # Increased for more training data val: _target_: jamun.data.parse_datasets_from_directory @@ -28,7 +28,7 @@ data: subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} - # max_datasets: 1 + max_datasets: 100 model: sigma_distribution: @@ -42,7 +42,7 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 1 # Increased due to model complexity + max_epochs: 50 # Increased due to model complexity # devices: 1 # gradient_clip_val: 1.0 # Add gradient clipping for stability diff --git a/configs/experiment/train_test_single_shape.yaml b/configs/experiment/train_test_single_shape.yaml index 9f15a0f..d972fea 100644 --- a/configs/experiment/train_test_single_shape.yaml +++ b/configs/experiment/train_test_single_shape.yaml @@ -25,9 +25,10 @@ data: traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] - as_iterable: false - subsample: 10 - + as_iterable: False + subsample: 80 + start_frame: 0 + num_frames: 800000 val: _target_: jamun.data.parse_datasets_from_directory @@ -35,8 +36,10 @@ data: traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] - as_iterable: false - subsample: 10 + as_iterable: False + subsample: 80 + start_frame: 800000 + num_frames: 100000 test: _target_: jamun.data.parse_datasets_from_directory @@ -44,8 +47,10 @@ data: traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] - as_iterable: false - subsample: 10 + as_iterable: False + subsample: 80 + start_frame: 900000 + num_frames: 200000 trainer: @@ -54,6 +59,6 @@ trainer: logger: wandb: - group: train_test - notes: "alanine dipeptide" - tags: ["single_shape", "ALA_ALA", "standard JAMUN"] \ No newline at end of file + group: model_comparison_delta_t_T_models_graphs + notes: "Standard JAMUN on 2AA capped diamines data" + tags: ["standard_jamun", "2AA", "capped_diamines", "denoiser", "generalization"] \ No newline at end of file diff --git a/configs/experiment/train_test_single_shape_conditional.yaml b/configs/experiment/train_test_single_shape_conditional.yaml index a1732de..a6c9663 100644 --- a/configs/experiment/train_test_single_shape_conditional.yaml +++ b/configs/experiment/train_test_single_shape_conditional.yaml @@ -17,11 +17,12 @@ data: traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] - as_iterable: false - subsample: 1 - total_lag_time: 5 - lag_subsample_rate: 1 - num_frames: 10 + as_iterable: False + subsample: 80 + total_lag_time: 8 + lag_subsample_rate: 10 + start_frame: 0 + num_frames: 800000 val: _target_: jamun.data.parse_datasets_from_directory @@ -29,11 +30,12 @@ data: traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] - as_iterable: false - subsample: 1 + as_iterable: False + subsample: 80 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} - lag_subsample_rate: 1 - num_frames: 10 + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + start_frame: 800000 + num_frames: 100000 test: _target_: jamun.data.parse_datasets_from_directory @@ -41,11 +43,12 @@ data: traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" filter_codes: ['ALA_ALA'] - as_iterable: false - subsample: 1 + as_iterable: False + subsample: 80 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} - lag_subsample_rate: 1 - num_frames: 10 + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + start_frame: 900000 + num_frames: 200000 model: sigma_distribution: _target_: jamun.distributions.ConstantSigma @@ -65,10 +68,11 @@ model: trainer: val_check_interval: 0.5 - max_epochs: 1 + max_epochs: 100 logger: wandb: - group: ALA_ALA, capped diamines, conditional denoiser - notes: "Running on ALA_ALA capped diamine, conditional denoiser" \ No newline at end of file + group: model_comparison_delta_t_T_models_graphs + notes: "PositionConditioner on 2AA capped diamines data" + tags: ["position_conditioner", "2AA", "capped_diamines", "conditional", "generalization"] \ No newline at end of file diff --git a/configs/experiment/train_test_single_shape_spatiotemporal_conditioner.yaml b/configs/experiment/train_test_single_shape_spatiotemporal_conditioner.yaml new file mode 100644 index 0000000..99eebd5 --- /dev/null +++ b/configs/experiment/train_test_single_shape_spatiotemporal_conditioner.yaml @@ -0,0 +1,73 @@ +# @package _global_ +# Training SpatioTemporalConditioner on 2AA capped diamines data + +defaults: + - override /model: denoiser_conditional_spatiotemporal + - override /callbacks: + - timing.yaml + - lr_monitor.yaml + - model_checkpoint.yaml + +data: + datamodule: + batch_size: 32 + datasets: + train: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: False + subsample: 80 + total_lag_time: 8 + lag_subsample_rate: 10 + start_frame: 0 + num_frames: 800000 + + val: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: False + subsample: 80 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + start_frame: 800000 + num_frames: 100000 + + test: + _target_: jamun.data.parse_datasets_from_directory + root: "${paths.data_path}/capped_diamines/timewarp_splits/train" + traj_pattern: "^(.*).xtc" + pdb_pattern: "^(.*).pdb" + filter_codes: ['ALA_ALA'] + as_iterable: False + subsample: 80 + total_lag_time: ${data.datamodule.datasets.train.total_lag_time} + lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} + start_frame: 900000 + num_frames: 200000 + +model: + sigma_distribution: + _target_: jamun.distributions.ConstantSigma + sigma: 0.04 + arch: + n_layers: 1 + max_radius: 1.0 + optim: + lr: 0.002 # Slightly reduced learning rate for stability + +trainer: + val_check_interval: 0.5 + max_epochs: 100 + +logger: + wandb: + group: model_comparison_delta_t_T_models_graphs + notes: "SpatioTemporalConditioner on 2AA capped diamines data" + tags: ["spatiotemporal_conditioner", "2AA", "capped_diamines", "transformer", "generalization"] + diff --git a/configs/sweep_delta_friction.yaml b/configs/sweep_delta_friction.yaml index d4e077d..cd3e24c 100644 --- a/configs/sweep_delta_friction.yaml +++ b/configs/sweep_delta_friction.yaml @@ -1,13 +1,13 @@ program: jamun_sample method: grid project: jamun -name: conditional_vs_self +name: sweep_delta_friction_large_noise metric: name: Jenson-Shannon Divergence vs. Number of Samples for Predicted Trajectory joined goal: minimize parameters: delta: - values: [0.017889, 0.026833, 0.040000, 0.059665, 0.089443] + values: [0.012, 0.039, 0.066, 0.093, 0.12] friction: values: [2.52572864, 1.2552661, 0.71334989, 0.36384343, 0.10536052] @@ -15,4 +15,5 @@ command: - ${program} - "--config-dir=configs" - "experiment=sample_enhanced_conditioning_sweep" + - "++batch_sampler.mcmc.history_update_frequency=100" - ${args_no_hyphens} \ No newline at end of file diff --git a/scripts/slurm/run_train_noise_check.sh b/scripts/slurm/run_train_noise_check.sh new file mode 100755 index 0000000..617fc39 --- /dev/null +++ b/scripts/slurm/run_train_noise_check.sh @@ -0,0 +1,63 @@ +#!/usr/bin/env bash +# +# Wrapper script to run train_noise_check.sh for multiple m values +# +# This script loops over m values from 2 to 10 and submits the train_noise_check.sh +# SLURM script for each value. This ensures only 4 parallel jobs are submitted at a time +# (one for each model configuration) rather than submitting all 36 jobs at once. +# +# Usage: ./run_train_noise_check.sh +# + +# Set script directory +# SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" +TRAIN_SCRIPT="scripts/slurm/train_noise_check.sh" + +# Check if train_noise_check.sh exists +if [ ! -f "$TRAIN_SCRIPT" ]; then + echo "Error: train_noise_check.sh not found at $TRAIN_SCRIPT" + exit 1 +fi + +# Make sure the script is executable +chmod +x "$TRAIN_SCRIPT" + +echo "Starting noise check experiments for m values 2-10" +echo "Each submission will run 4 parallel jobs (one for each model configuration)" +echo "" + +# Loop over m values from 2 to 10 +for M in {2..10}; do + echo "Submitting jobs for M=$M..." + + # Submit the SLURM script with the current m value + JOB_ID=$(sbatch --parsable scripts/slurm/train_noise_check.sh $M) + + if [ $? -eq 0 ]; then + echo " āœ“ Successfully submitted job ID: $JOB_ID for M=$M" + echo " This will run 4 parallel jobs (array 0-3) for the 4 model configurations" + + # Wait for all array jobs to complete before submitting next batch + # echo " ā³ Waiting for job ID $JOB_ID to complete before submitting next batch..." + echo " ā³ Submitted ID $JOB_ID..." + # # Wait for the job to finish (all array tasks) + # while squeue -j "$JOB_ID" 2>/dev/null | grep -q "$JOB_ID"; do + # sleep 30 # Check every 30 seconds + # done + + echo " āœ… All jobs for M=$M (Job ID: $JOB_ID) completed!" + echo "" + + else + echo " āœ— Failed to submit job for M=$M" + exit 1 + fi +done + +echo "" +echo "All jobs submitted successfully!" +echo "Total submissions: 9 (one for each m value from 2-10)" +echo "Total jobs: 36 (4 models Ɨ 9 m values)" +echo "" +echo "Monitor job status with: squeue -u \$USER" +echo "View job outputs in the current directory with pattern: slurm-_.out" diff --git a/scripts/slurm/sweep.sh b/scripts/slurm/sweep.sh index 0e272e5..a33d1c1 100644 --- a/scripts/slurm/sweep.sh +++ b/scripts/slurm/sweep.sh @@ -1,13 +1,12 @@ #!/usr/bin/env bash - #SBATCH --partition=gpu2 #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node #SBATCH --gpus-per-node=1 # Assign one GPU to each agent #SBATCH --cpus-per-task=12 -#SBATCH --time 3-0 +#SBATCH --time 2-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array 0-4 +#SBATCH --array 0-10 # Check if a Sweep ID is provided as an argument export JAMUN_ROOT_PATH=/data2/sules/jamun-conditional-runs @@ -20,6 +19,7 @@ fi SWEEP_ID=$1 # Set up the environment +source ~/.bashrc eval "$(conda shell.bash hook)" conda activate jamun diff --git a/scripts/slurm/train_capped_2AA_comparison.sh b/scripts/slurm/train_capped_2AA_comparison.sh new file mode 100755 index 0000000..fc976f5 --- /dev/null +++ b/scripts/slurm/train_capped_2AA_comparison.sh @@ -0,0 +1,85 @@ +#!/usr/bin/env bash + +#SBATCH --partition=gpu2 +#SBATCH --job-name=capped_2AA_comparison +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-4 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +# Define configurations for each job +case ${SLURM_ARRAY_TASK_ID} in + 0) + echo "Job 0: Standard JAMUN on 2AA capped diamines" + CONFIG="train_capped_2AA" + OVERRIDES="" + ;; + 1) + echo "Job 1: Position conditioner on 2AA capped diamines" + CONFIG="train_capped_2AA_position_conditioner" + OVERRIDES="" + ;; + 2) + echo "Job 2: Self conditioner on 2AA capped diamines" + CONFIG="train_capped_2AA_self_conditioner" + OVERRIDES="" + ;; + 3) + echo "Job 3: Spatiotemporal conditioner with temporal embedding and mean pooling on 2AA capped diamines" + CONFIG="train_capped_2AA_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean" + ;; + 4) + echo "Job 4: Spatiotemporal conditioner with ones temporal encoding and mean pooling on 2AA capped diamines" + CONFIG="train_capped_2AA_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + ;; + *) + echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" + exit 1 + ;; +esac + +# Build the command with base config +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add overrides if any +if [ -n "$OVERRIDES" ]; then + CMD="$CMD $OVERRIDES" +fi + +# Add common training overrides +CMD="$CMD ++trainer.max_epochs=100" +CMD="$CMD ++logger.wandb.group=capped_2AA_model_comparison" +CMD="$CMD ++run_key=${RUN_KEY}" + +# Add dataset overrides for debugging (quick completion) +# CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" +# CMD="$CMD ++data.datamodule.datasets.val.max_datasets=1" + +# Add job-specific wandb tags +WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},capped_2AA_comparison,generalization_test]" + +echo "Running command: $CMD" +exec $CMD diff --git a/scripts/slurm/train_graph_type_comparison.sh b/scripts/slurm/train_graph_type_comparison.sh new file mode 100755 index 0000000..e2f13ed --- /dev/null +++ b/scripts/slurm/train_graph_type_comparison.sh @@ -0,0 +1,138 @@ +#!/usr/bin/env bash +# +# Graph type comparison experiment script +# +# Iterates over: +# - Lag subsample rates: 1, 2, 3, 4 +# - Total lag times: 2, 4, 6, 8 +# - Configs: train_test_single_shape.yaml, train_test_single_shape_conditional.yaml, train_test_single_shape_spatiotemporal_conditioner.yaml +# - For spatiotemporal: hub_n_spoke vs complete graph types (both with ones encoding) +# +# Total jobs: 4 lag_subsample_rates Ɨ 4 total_lag_times Ɨ (1 conditional + 2 spatiotemporal variants) = 48 jobs +# +#SBATCH --partition=gpu3 +#SBATCH --job-name=graph_type_comparison +#SBATCH --qos=preempt +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-task=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-15 + +# Set up the environment +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Define experiment parameters +LAG_SUBSAMPLE_RATES=(1 2 3 4) +TOTAL_LAG_TIMES=(2 4 6 8) +CONFIGS=( + # "train_test_single_shape" + "train_test_single_shape_conditional" + "train_test_single_shape_spatiotemporal_conditioner_default" + "train_test_single_shape_spatiotemporal_conditioner_hub_spoke" + # "train_test_single_shape_spatiotemporal_conditioner_complete" +) + +NUM_CONFIGS=1 +NUM_LAG_TIMES=4 +NUM_LAG_SUBSAMPLE_RATES=4 + +# Calculate indices +LAG_SUBSAMPLE_INDEX=$((SLURM_ARRAY_TASK_ID / (NUM_LAG_TIMES * NUM_CONFIGS))) +REMAINDER=$((SLURM_ARRAY_TASK_ID % (NUM_LAG_TIMES * NUM_CONFIGS))) +LAG_TIME_INDEX=$((REMAINDER / NUM_CONFIGS)) +CONFIG_INDEX=$((REMAINDER % NUM_CONFIGS)) + +LAG_SUBSAMPLE_RATE=${LAG_SUBSAMPLE_RATES[$LAG_SUBSAMPLE_INDEX]} +TOTAL_LAG_TIME=${TOTAL_LAG_TIMES[$LAG_TIME_INDEX]} +CONFIG_TYPE=${CONFIGS[$CONFIG_INDEX]} + +echo "Job ${SLURM_ARRAY_TASK_ID}: lag_subsample_rate=${LAG_SUBSAMPLE_RATE}, total_lag_time=${TOTAL_LAG_TIME}, config=${CONFIG_TYPE}" + +# Generate unique run key +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +# Calculate max_datasets using the formula: floor(250 * (49 / (lag_subsample_rate * (total_lag_time - 1) - 1))) +DENOMINATOR=$((LAG_SUBSAMPLE_RATE * (TOTAL_LAG_TIME - 1) - 1)) +if [ $DENOMINATOR -le 0 ]; then + echo "Warning: Invalid denominator ($DENOMINATOR) for max_datasets calculation. Setting to 250." + MAX_DATASETS=250 +else + # Using bc for floating point calculation and floor function + MAX_DATASETS=$(echo "scale=0; 250 * 49 / $DENOMINATOR" | bc) +fi +echo "Calculated max_datasets = $MAX_DATASETS (lag_subsample_rate=$LAG_SUBSAMPLE_RATE, total_lag_time=$TOTAL_LAG_TIME)" + +# Set base configuration and overrides based on config type +case $CONFIG_INDEX in + # 0) # Standard JAMUN + # CONFIG="train_test_single_shape" + # OVERRIDES="" + # WANDB_TAG="standard_jamun_lag_${LAG_SUBSAMPLE_RATE}_time_${TOTAL_LAG_TIME}" + # ;; + # 0) # Position Conditioner + # CONFIG="train_enhanced_position_conditioner" + # OVERRIDES="++data.datamodule.datasets.train.total_lag_time=${TOTAL_LAG_TIME}" + # OVERRIDES="$OVERRIDES ++data.datamodule.datasets.train.lag_subsample_rate=${LAG_SUBSAMPLE_RATE}" + # OVERRIDES="$OVERRIDES ++model.arch.N_structures=${TOTAL_LAG_TIME}" + # OVERRIDES="$OVERRIDES ++model.conditioner.N_structures=${TOTAL_LAG_TIME}" + # WANDB_TAG="position_conditioner_lag_${LAG_SUBSAMPLE_RATE}_time_${TOTAL_LAG_TIME}" + # ;; + 0) # SpatioTemporal Conditioner - Default (fan graph, temporal encoding) + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++data.datamodule.datasets.train.total_lag_time=${TOTAL_LAG_TIME}" + OVERRIDES="$OVERRIDES ++data.datamodule.datasets.train.lag_subsample_rate=${LAG_SUBSAMPLE_RATE}" + WANDB_TAG="spatiotemporal_default_fan_temporal_lag_${LAG_SUBSAMPLE_RATE}_time_${TOTAL_LAG_TIME}" + ;; + # 2) # SpatioTemporal Conditioner - Hub & Spoke + # CONFIG="train_enhanced_spatiotemporal_conditioner" + # OVERRIDES="++data.datamodule.datasets.train.total_lag_time=${TOTAL_LAG_TIME}" + # OVERRIDES="$OVERRIDES ++data.datamodule.datasets.train.lag_subsample_rate=${LAG_SUBSAMPLE_RATE}" + # OVERRIDES="$OVERRIDES ++model.conditioner.spatiotemporal_model.graph_type=hub_n_spoke" + # OVERRIDES="$OVERRIDES ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones" + # OVERRIDES="$OVERRIDES ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + # WANDB_TAG="spatiotemporal_hub_spoke_ones_lag_${LAG_SUBSAMPLE_RATE}_time_${TOTAL_LAG_TIME}" + # ;; + # 4) # SpatioTemporal Conditioner - Complete + # CONFIG="train_test_single_shape_spatiotemporal_conditioner" + # OVERRIDES="++data.datamodule.datasets.train.total_lag_time=${TOTAL_LAG_TIME}" + # OVERRIDES="$OVERRIDES ++data.datamodule.datasets.train.lag_subsample_rate=${LAG_SUBSAMPLE_RATE}" + # OVERRIDES="$OVERRIDES ++model.conditioner.spatiotemporal_model.graph_type=complete" + # OVERRIDES="$OVERRIDES ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones" + # OVERRIDES="$OVERRIDES ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + # WANDB_TAG="spatiotemporal_complete_ones_lag_${LAG_SUBSAMPLE_RATE}_time_${TOTAL_LAG_TIME}" + # ;; +esac + +# Build command +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add overrides if any +if [ -n "$OVERRIDES" ]; then + CMD="$CMD $OVERRIDES" +fi + +# Calculate validation max_datasets using multiplier of 50 +VAL_MAX_DATASETS=$(echo "scale=0; 50 * 49 / $DENOMINATOR" | bc) +if [ $DENOMINATOR -le 0 ]; then + VAL_MAX_DATASETS=50 +fi + +# Add common overrides +CMD="$CMD ++run_key=${RUN_KEY}" +CMD="$CMD ++data.datamodule.datasets.train.max_datasets=${MAX_DATASETS}" +CMD="$CMD ++data.datamodule.datasets.val.max_datasets=${VAL_MAX_DATASETS}" +CMD="$CMD ++logger.wandb.group=graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},graph_comparison,lag_subsample_${LAG_SUBSAMPLE_RATE},total_lag_${TOTAL_LAG_TIME}]" + +# Set wandb run name +WANDB_RUN_NAME="graph_comparison_${WANDB_TAG}_enhanced_sampling_data" +CMD="$CMD ++logger.wandb.name=${WANDB_RUN_NAME}" + +echo "Running command: $CMD" +exec $CMD diff --git a/scripts/slurm/train_model_comparison.sh b/scripts/slurm/train_model_comparison.sh index ce660f8..ac4da0f 100644 --- a/scripts/slurm/train_model_comparison.sh +++ b/scripts/slurm/train_model_comparison.sh @@ -9,7 +9,7 @@ #SBATCH --cpus-per-task=8 #SBATCH --time=3-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array=0-1 +#SBATCH --array=0-4 # Initialize conda source ~/.bashrc @@ -24,6 +24,10 @@ nvidia-smi echo "Running array job ${SLURM_ARRAY_TASK_ID}" +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + # Define configurations for each job case ${SLURM_ARRAY_TASK_ID} in 0) @@ -42,14 +46,14 @@ case ${SLURM_ARRAY_TASK_ID} in OVERRIDES="" ;; 3) - echo "Job 3: Spatiotemporal conditioner with mean pooling and trainable pretrained denoiser" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + echo "Job 3: Spatiotemporal conditioner with temporal embedding and mean pooling" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean" ;; 4) - echo "Job 4: Spatiotemporal conditioner with mean pooling, trainable pretrained denoiser, and ones temporal encoding" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + echo "Job 4: Spatiotemporal conditioner with ones temporal encoding and mean pooling" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" ;; *) echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" @@ -68,6 +72,7 @@ fi # Add common training overrides CMD="$CMD ++trainer.max_epochs=100" CMD="$CMD ++logger.wandb.group=model_comparison" +CMD="$CMD ++run_key=${RUN_KEY}" # Add dataset overrides for debugging (quick completion) # CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" @@ -75,7 +80,7 @@ CMD="$CMD ++logger.wandb.group=model_comparison" # Add job-specific wandb tags WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" -CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison]" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison,separable_conv]" echo "Running command: $CMD" exec $CMD diff --git a/scripts/slurm/train_model_comparison_full_swarm.sh b/scripts/slurm/train_model_comparison_full_swarm.sh index a841ea7..0ce02c7 100644 --- a/scripts/slurm/train_model_comparison_full_swarm.sh +++ b/scripts/slurm/train_model_comparison_full_swarm.sh @@ -9,7 +9,7 @@ #SBATCH --cpus-per-task=8 #SBATCH --time=3-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array=2-4 +#SBATCH --array=0-4 # Initialize conda source ~/.bashrc @@ -24,6 +24,10 @@ nvidia-smi echo "Running array job ${SLURM_ARRAY_TASK_ID}" +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + # Define configurations for each job case ${SLURM_ARRAY_TASK_ID} in 0) @@ -42,14 +46,14 @@ case ${SLURM_ARRAY_TASK_ID} in OVERRIDES="" ;; 3) - echo "Job 3: Spatiotemporal conditioner with mean pooling and trainable pretrained denoiser" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + echo "Job 3: Spatiotemporal conditioner with temporal embedding and mean pooling" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean" ;; 4) - echo "Job 4: Spatiotemporal conditioner with mean pooling, trainable pretrained denoiser, and ones temporal encoding" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + echo "Job 4: Spatiotemporal conditioner with ones temporal encoding and mean pooling" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" ;; *) echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" @@ -70,6 +74,7 @@ CMD="$CMD ++trainer.max_epochs=100" CMD="$CMD ++logger.wandb.group=model_comparison_full_swarm" CMD="$CMD ++data.datamodule.datasets.train.root=/data2/sules/ALA_ALA_enhanced_full_swarm/train" CMD="$CMD ++data.datamodule.datasets.val.root=/data2/sules/ALA_ALA_enhanced_full_swarm/val" +CMD="$CMD ++run_key=${RUN_KEY}" # Add dataset overrides for debugging (quick completion) # CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" @@ -77,7 +82,7 @@ CMD="$CMD ++data.datamodule.datasets.val.root=/data2/sules/ALA_ALA_enhanced_full # Add job-specific wandb tags WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" -CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison]" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison,separable_conv]" echo "Running command: $CMD" exec $CMD diff --git a/scripts/slurm/train_model_comparison_high_noise.sh b/scripts/slurm/train_model_comparison_high_noise.sh index 689b242..ad1e943 100644 --- a/scripts/slurm/train_model_comparison_high_noise.sh +++ b/scripts/slurm/train_model_comparison_high_noise.sh @@ -1,15 +1,14 @@ #!/usr/bin/env bash -#SBATCH --partition=gpu3 +#SBATCH --partition=gpu2 #SBATCH --job-name=model_comparison -#SBATCH --qos=preempt #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --gpus-per-node=1 #SBATCH --cpus-per-task=8 #SBATCH --time=3-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array=2-4 +#SBATCH --array=0-4 # Initialize conda source ~/.bashrc @@ -24,6 +23,10 @@ nvidia-smi echo "Running array job ${SLURM_ARRAY_TASK_ID}" +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + # Define configurations for each job case ${SLURM_ARRAY_TASK_ID} in 0) @@ -42,14 +45,14 @@ case ${SLURM_ARRAY_TASK_ID} in OVERRIDES="" ;; 3) - echo "Job 3: Spatiotemporal conditioner with mean pooling and trainable pretrained denoiser" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + echo "Job 3: Spatiotemporal conditioner with temporal embedding and mean pooling" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean" ;; 4) - echo "Job 4: Spatiotemporal conditioner with mean pooling, trainable pretrained denoiser, and ones temporal encoding" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.spatial_module.trainable=true ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" + echo "Job 4: Spatiotemporal conditioner with ones temporal encoding and mean pooling" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++model.conditioner.spatiotemporal_model.temporal_to_spatial_pooler._target_=jamun.model.pooling.TemporalToSpatialNodeAttrMean ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" ;; *) echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" @@ -68,7 +71,8 @@ fi # Add common training overrides CMD="$CMD ++trainer.max_epochs=100" CMD="$CMD ++logger.wandb.group=model_comparison_high_noise" -CMD="$CMD ++model.sigma_distribution.sigma=0.08" +CMD="$CMD ++model.sigma_distribution.sigma=0.06" +CMD="$CMD ++run_key=${RUN_KEY}" # Add dataset overrides for debugging (quick completion) # CMD="$CMD ++data.datamodule.datasets.train.max_datasets=1" @@ -76,7 +80,7 @@ CMD="$CMD ++model.sigma_distribution.sigma=0.08" # Add job-specific wandb tags WANDB_TAG="job_${SLURM_ARRAY_TASK_ID}" -CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison]" +CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},model_comparison,separable_conv]" echo "Running command: $CMD" exec $CMD diff --git a/scripts/slurm/train_noise_check.sh b/scripts/slurm/train_noise_check.sh index 428f96a..7be084c 100755 --- a/scripts/slurm/train_noise_check.sh +++ b/scripts/slurm/train_noise_check.sh @@ -2,29 +2,24 @@ # # Noise check experiment script # -# Implements 5 out of 7 requested model configurations for m=2,3,4,5,6,7,8,9,10: -# 1. Standard JAMUN with noise level sigma/sqrt(m) -# 2. Spatiotemporal with temporal embedding (pretrained denoiser, trainable=true) -# 3. Spatiotemporal with ones embedding (pretrained denoiser, trainable=true) -# 4. Spatiotemporal with temporal embedding, repeated position dataset -# 5. Spatiotemporal with ones embedding, repeated position dataset +# Implements 4 model configurations for a given m value (passed as command line argument): +# 1. Standard JAMUN with repeated position dataset and noise level sigma/sqrt(m) +# 2. Spatiotemporal JAMUN with repeated position dataset and total lag time m +# 3. Spatiotemporal JAMUN with total lag time m +# 4. Spatiotemporal JAMUN with total lag time m, hub_n_spoke graph type, ones encoding # -# NOT IMPLEMENTED (requires custom dataset class): -# 6. Spatiotemporal with temporal embedding, random lag times -# 7. Spatiotemporal with ones embedding, random lag times -# -# Total jobs: 9 m-values Ɨ 5 models = 45 jobs (array 0-44) +# Usage: sbatch train_noise_check.sh +# Total jobs: 4 models (array 0-3) -#SBATCH --partition=gpu3 +#SBATCH --partition=gpu2 #SBATCH --job-name=noise_check -#SBATCH --qos=preempt #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --gpus-per-node=1 #SBATCH --cpus-per-task=8 #SBATCH --time=3-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array=0-44 +#SBATCH --array=1-3 # Initialize conda source ~/.bashrc @@ -37,32 +32,37 @@ echo "Python path: $(which python)" echo "Conda environment: $CONDA_DEFAULT_ENV" nvidia-smi -echo "Running array job ${SLURM_ARRAY_TASK_ID}" +# Get m value from command line argument +if [ $# -eq 0 ]; then + echo "Error: Please provide m value as command line argument" + echo "Usage: sbatch train_noise_check.sh " + exit 1 +fi + +M=$1 +echo "Running array job ${SLURM_ARRAY_TASK_ID} with M=${M}" # Define experiment parameters -# m values: 2, 3, 4, 5, 6, 7, 8, 9, 10 (9 values) -# Model types: 5 types (random lag times require custom implementation) -# Total combinations: 9 * 5 = 45 jobs (0-44) +# Model types: 4 types +# Total jobs: 4 models (array 0-3) -M_VALUES=(2 3 4 5 6 7 8 9 10) MODEL_TYPES=( - "standard_jamun" - "spatiotemporal_temporal_embedding" - "spatiotemporal_ones_embedding" - "spatiotemporal_temporal_embedding_repeated_pos" - "spatiotemporal_ones_embedding_repeated_pos" + "standard_jamun_repeated_pos" + "spatiotemporal_repeated_pos" + "spatiotemporal_default" + "spatiotemporal_hub_spoke_ones" ) -# Calculate m and model indices -NUM_MODELS=5 -M_INDEX=$((SLURM_ARRAY_TASK_ID / NUM_MODELS)) -MODEL_INDEX=$((SLURM_ARRAY_TASK_ID % NUM_MODELS)) - -M=${M_VALUES[$M_INDEX]} +# Calculate model index directly from array task ID +MODEL_INDEX=${SLURM_ARRAY_TASK_ID} MODEL_TYPE=${MODEL_TYPES[$MODEL_INDEX]} echo "Job ${SLURM_ARRAY_TASK_ID}: M=${M}, Model=${MODEL_TYPE}" +# Generate unique run key to prevent checkpoint overwrites +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + # Calculate noise level: sigma / sqrt(m) where base sigma = 0.04 BASE_SIGMA=0.04 NOISE_LEVEL=$(python3 -c "import math; print(${BASE_SIGMA} / math.sqrt(${M}))") @@ -71,65 +71,49 @@ echo "Noise level: ${NOISE_LEVEL}" # Configure base parameters based on model type case ${MODEL_TYPE} in - "standard_jamun") - echo "Model 1: Standard JAMUN with noise level sigma/sqrt(m)" - CONFIG="train_enhanced_standard_jamun" - OVERRIDES="++model.sigma_distribution.sigma=${NOISE_LEVEL}" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" - WANDB_TAG="standard_jamun_m${M}" - ;; - "spatiotemporal_temporal_embedding") - echo "Model 2: Spatiotemporal with temporal embedding (pretrained denoiser, trainable=true)" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + # "standard_jamun_repeated_pos") + # echo "Model 1: Standard JAMUN with repeated position dataset and noise level sigma/sqrt(m)" + # CONFIG="train_enhanced_standard_jamun" + # OVERRIDES="++model.sigma_distribution.sigma=${NOISE_LEVEL}" + # OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train._target_=jamun.data.parse_repeated_position_datasets_from_directory" + # OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val._target_=jamun.data.parse_repeated_position_datasets_from_directory" + # OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + # OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" + # OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=500" + # OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=100" + # WANDB_TAG="standard_jamun_repeated_pos_m${M}" + # ;; + "spatiotemporal_repeated_pos") + echo "Model 2: Spatiotemporal JAMUN with repeated position dataset and total lag time m" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++data.datamodule.datasets.train._target_=jamun.data.parse_repeated_position_datasets_from_directory" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val._target_=jamun.data.parse_repeated_position_datasets_from_directory" OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" - - WANDB_TAG="spatiotemporal_temporal_m${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=500" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=100" + WANDB_TAG="spatiotemporal_repeated_pos_m${M}" ;; - "spatiotemporal_ones_embedding") - echo "Model 3: Spatiotemporal with ones embedding (pretrained denoiser, trainable=true)" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" - OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones" - OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + "spatiotemporal_default") + echo "Model 3: Spatiotemporal JAMUN with total lag time m" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++data.datamodule.datasets.train.total_lag_time=${M}" OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" - - WANDB_TAG="spatiotemporal_ones_m${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=500" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=100" + WANDB_TAG="spatiotemporal_default_m${M}" ;; - "spatiotemporal_temporal_embedding_repeated_pos") - echo "Model 4: Spatiotemporal with temporal embedding, repeated position dataset" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train._target_=jamun.data.parse_repeated_position_datasets_from_directory" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val._target_=jamun.data.parse_repeated_position_datasets_from_directory" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" + "spatiotemporal_hub_spoke_ones") + echo "Model 4: Spatiotemporal JAMUN with total lag time m, hub_n_spoke graph type, ones encoding" + CONFIG="train_enhanced_spatiotemporal_conditioner" + OVERRIDES="++data.datamodule.datasets.train.total_lag_time=${M}" OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" - - WANDB_TAG="spatiotemporal_temporal_repeated_m${M}" - ;; - "spatiotemporal_ones_embedding_repeated_pos") - echo "Model 5: Spatiotemporal with ones embedding, repeated position dataset" - CONFIG="train_enhanced_pretrained_spatiotemporal_conditioner" - OVERRIDES="++model.conditioner.spatiotemporal_model.spatial_module.trainable=true" + OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.graph_type=hub_n_spoke" OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.node_attr_temporal_encoding_function=ones" OVERRIDES="${OVERRIDES} ++model.conditioner.spatiotemporal_model.temporal_module.edge_attr_temporal_encoding_function=ones" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train._target_=jamun.data.parse_repeated_position_datasets_from_directory" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val._target_=jamun.data.parse_repeated_position_datasets_from_directory" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.total_lag_time=${M}" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.total_lag_time=${M}" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=1" - OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=1" - - WANDB_TAG="spatiotemporal_ones_repeated_m${M}" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.train.max_datasets=500" + OVERRIDES="${OVERRIDES} ++data.datamodule.datasets.val.max_datasets=100" + WANDB_TAG="spatiotemporal_hub_spoke_ones_m${M}" ;; *) echo "Unknown model type: ${MODEL_TYPE}" @@ -146,8 +130,9 @@ if [ -n "$OVERRIDES" ]; then fi # Add common training overrides -CMD="$CMD ++trainer.max_epochs=1" -CMD="$CMD ++logger.wandb.group=noise_check_experiment" +CMD="$CMD ++trainer.max_epochs=50" +CMD="$CMD ++run_key=${RUN_KEY}" +CMD="$CMD ++logger.wandb.group=noise_check_experiment_multimeasurement_vs_correlation" # Add job-specific wandb tags and run name WANDB_RUN_NAME="noise_check_${WANDB_TAG}" @@ -155,7 +140,7 @@ CMD="$CMD ++logger.wandb.tags=[${WANDB_TAG},noise_check,m_${M}]" CMD="$CMD ++logger.wandb.name=${WANDB_RUN_NAME}" # Add notes about the experiment -WANDB_NOTES="Noise check experiment: ${MODEL_TYPE} with m=${M}" +WANDB_NOTES="Noise_check_experiment:_${MODEL_TYPE}_with_m=${M}" if [[ ${MODEL_TYPE} == "standard_jamun" ]]; then WANDB_NOTES="${WANDB_NOTES}, noise_level=${NOISE_LEVEL}" fi diff --git a/src/jamun/hydra_config/callbacks/model_checkpoint.yaml b/src/jamun/hydra_config/callbacks/model_checkpoint.yaml index ddad6cf..0a59fcd 100644 --- a/src/jamun/hydra_config/callbacks/model_checkpoint.yaml +++ b/src/jamun/hydra_config/callbacks/model_checkpoint.yaml @@ -1,7 +1,8 @@ model_checkpoint: _target_: "lightning.pytorch.callbacks.ModelCheckpoint" dirpath: "${hydra:runtime.output_dir}/checkpoints" - save_top_k: 5 + save_top_k: -1 save_last: true monitor: "val/loss" + save_on_train_epoch_end: true every_n_epochs: 1 diff --git a/src/jamun/hydra_config/model/arch/e3conv.yaml b/src/jamun/hydra_config/model/arch/e3conv.yaml index 4bfd8ee..3e04b19 100644 --- a/src/jamun/hydra_config/model/arch/e3conv.yaml +++ b/src/jamun/hydra_config/model/arch/e3conv.yaml @@ -19,7 +19,7 @@ hidden_layer_factory: _target_: e3tools.nn.ConvBlock _partial_: true conv: - _target_: e3tools.nn.Conv + _target_: e3tools.nn.SeparableConv _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml index c53a440..3a272e7 100644 --- a/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml +++ b/src/jamun/hydra_config/model/arch/e3conv_conditional.yaml @@ -19,7 +19,7 @@ hidden_layer_factory: _target_: e3tools.nn.ConvBlock _partial_: true conv: - _target_: e3tools.nn.Conv + _target_: e3tools.nn.SeparableConv _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP diff --git a/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml index b125f54..3833eae 100644 --- a/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/arch/e3conv_conditional_spatiotemporal.yaml @@ -21,7 +21,7 @@ hidden_layer_factory: _target_: e3tools.nn.ConvBlock _partial_: true conv: - _target_: e3tools.nn.Conv # replace with Conv for non-separable case + _target_: e3tools.nn.SeparableConv # replace with Conv for non-separable case _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP diff --git a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml index ef23d79..c4384ee 100644 --- a/src/jamun/hydra_config/model/arch/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/arch/spatiotemporal.yaml @@ -28,7 +28,7 @@ spatial_module: _target_: e3tools.nn.ConvBlock _partial_: true conv: - _target_: e3tools.nn.Conv + _target_: e3tools.nn.SeparableConv _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP diff --git a/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml index b34a68a..5799ca2 100644 --- a/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/conditioner/spatiotemporal.yaml @@ -32,7 +32,7 @@ spatiotemporal_model: _target_: e3tools.nn.ConvBlock _partial_: true conv: - _target_: e3tools.nn.Conv # replace with Conv for non-separable case + _target_: e3tools.nn.SeparableConv # replace with Conv for non-separable case _partial_: true output_head_factory: _target_: e3tools.nn.EquivariantMLP @@ -48,6 +48,9 @@ spatiotemporal_model: irreps_sh: "1x0e + 1x1e" irreps_node_attr: "1x1e" # Match spatial module output irreps_node_attr_temporal: "3x0e" + conv: + _target_: e3tools.nn.SeparableConv + _partial_: true num_layers: 2 edge_attr_dim: 24 num_attention_heads: 1 diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index 959f0f1..a13bd53 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -7,7 +7,7 @@ - Spatial-temporal graph conversion utilities """ -from typing import Dict, Union +from typing import Dict, Union, Optional import e3nn import torch @@ -47,15 +47,30 @@ def calculate_temporal_positions(temporal_length, mode="linear", device=None): return normalized_positions -def spatial_to_temporal_graphs(batch, temporal_position_mode="linear"): +def spatial_to_temporal_graphs(batch, graph_type="fan"): """ - Convert a batch of spatial graphs to temporal graphs. + Convert a batch of spatial graphs to temporal graphs with configurable connectivity. For each spatial node with position + hidden states, create a temporal graph where: - Node 0: current position - - Nodes 1-T: hidden state positions - - Connectivity: Node 0 connects to all others, sequential connections 1->2->3->... + - Nodes 1-T: hidden state positions + - Connectivity depends on graph_type parameter + + Args: + batch: Input spatial graph batch + graph_type: Type of connectivity to use + - "fan": Hub connects to all + sequential connections (0->all, i->(i+1)) + - "hub_n_spoke": Only hub-spoke connections (0->all, no sequential) + - "complete": Complete graph with self-loops (all-to-all including self) + - "complete_no_self": Complete graph without self-loops (all-to-all excluding self) """ + import torch_geometric + + # Validate graph_type + valid_types = ["fan", "hub_n_spoke", "complete", "complete_no_self"] + if graph_type not in valid_types: + raise ValueError(f"graph_type must be one of {valid_types}, got {graph_type}") + # Get device from input batch device = batch.pos.device @@ -71,8 +86,18 @@ def spatial_to_temporal_graphs(batch, temporal_position_mode="linear"): num_hidden_states = 0 temporal_length = 1 + # print(f"Creating {graph_type} temporal graphs: {num_spatial_nodes} spatial nodes -> {num_spatial_nodes} temporal graphs of length {temporal_length}") + # Store reference to spatial graph - # spatial_graph = batch.clone() + spatial_graph = batch.clone() + + # Set connectivity type code for tracking + connectivity_type_map = { + "fan": 0, + "hub_n_spoke": 1, + "complete": 2, + "complete_no_self": 3 + } temporal_graphs = [] @@ -88,34 +113,64 @@ def spatial_to_temporal_graphs(batch, temporal_position_mode="linear"): temporal_pos = torch.stack(temporal_positions) # Shape: [T, 3] # Calculate temporal positions for this sequence - temporal_position = calculate_temporal_positions(temporal_length, mode=temporal_position_mode, device=device) + temporal_position = calculate_temporal_positions(temporal_length, device=device) - # Create edge connectivity + # Create edge connectivity based on graph_type if temporal_length > 1: - # Node 0 connects to all others: 0->1, 0->2, 0->3, ..., 0->T-1 - hub_src = [0] * (temporal_length - 1) - hub_dst = list(range(1, temporal_length)) - - # Sequential connections: 1->2, 2->3, ..., (T-2)->(T-1) - seq_src = list(range(1, temporal_length - 1)) - seq_dst = list(range(2, temporal_length)) - - # Combine all edges - all_src = hub_src + seq_src - all_dst = hub_dst + seq_dst - - edge_index = torch.tensor([all_src, all_dst], dtype=torch.long, device=device) + if graph_type == "fan": + # Original fan system: hub-spoke + sequential + # Hub connections: 0->1, 0->2, 0->3, ..., 0->T-1 + hub_src = [0] * (temporal_length - 1) + hub_dst = list(range(1, temporal_length)) + + # Sequential connections: 1->2, 2->3, ..., (T-2)->(T-1) + seq_src = list(range(1, temporal_length - 1)) + seq_dst = list(range(2, temporal_length)) + + # Combine all edges + all_src = hub_src + seq_src + all_dst = hub_dst + seq_dst + + edge_index = torch.tensor([all_src, all_dst], dtype=torch.long, device=device) + + elif graph_type == "hub_n_spoke": + # Hub-and-spoke only: 0 connects to all others, no sequential + hub_src = [0] * (temporal_length - 1) + hub_dst = list(range(1, temporal_length)) + + edge_index = torch.tensor([hub_src, hub_dst], dtype=torch.long, device=device) + + elif graph_type == "complete": + # Complete graph without self-loops: all-to-all excluding self + src_nodes = [] + dst_nodes = [] + + for i in range(temporal_length): + for j in range(temporal_length): + if i != j: # Exclude self-loops + src_nodes.append(i) + dst_nodes.append(j) + + edge_index = torch.tensor([src_nodes, dst_nodes], dtype=torch.long, device=device) + else: - # Single node, no edges - edge_index = torch.tensor([[], []], dtype=torch.long, device=device) + # Single node case + if graph_type == "complete": + # Single node with self-loop + edge_index = torch.tensor([[0], [0]], dtype=torch.long, device=device) + else: + # Single node, no edges for other types + edge_index = torch.tensor([[], []], dtype=torch.long, device=device) # Create temporal graph for this spatial node temporal_graph = torch_geometric.data.Data( pos=temporal_pos, edge_index=edge_index, - spatial_node_idx=torch.tensor([node_idx], dtype=torch.long, device=device), # Track which spatial node this came from - temporal_length=torch.tensor([temporal_length], dtype=torch.long, device=device), - temporal_position=temporal_position # Normalized position in sequence [0, 1/T, 2/T, ...] + spatial_node_idx=torch.tensor([node_idx], device=device), # Track which spatial node this came from + temporal_length=torch.tensor([temporal_length], device=device), + temporal_position=temporal_position, # Normalized position in sequence [0, 1/T, 2/T, ...] + connectivity_type=torch.tensor([connectivity_type_map[graph_type]], device=device), + # Note: Removed graph_type string to avoid batching issues with PyTorch Geometric ) temporal_graphs.append(temporal_graph) @@ -123,7 +178,9 @@ def spatial_to_temporal_graphs(batch, temporal_position_mode="linear"): temporal_batch = torch_geometric.data.Batch.from_data_list(temporal_graphs) # Store spatial graph reference - # temporal_batch.spatial_graph = spatial_graph + temporal_batch.spatial_graph = spatial_graph + # Note: Removed graph_type string to avoid batching issues with PyTorch Geometric + # Graph type can be inferred from connectivity_type tensor attribute return temporal_batch @@ -170,6 +227,7 @@ def __init__( edge_attr_dim: int, num_attention_heads: int, reduce: str | None = None, + conv = e3tools.nn.Conv, irreps_node_attr_temporal: Union[str, e3nn.o3.Irreps] = "1x1e", radial_edge_attr_encoding_function: str = "gaussian", node_attr_temporal_encoding_function: str = "gaussian", @@ -207,6 +265,7 @@ def __init__( irreps_gate=irreps_with_temporal,) self.layers = nn.ModuleList() + self.conv = conv for _ in range(num_layers): self.layers.append( e3tools.nn.TransformerBlock( @@ -215,6 +274,7 @@ def __init__( irreps_sh=self.irreps_sh, edge_attr_dim=self.edge_attr_dim, num_heads=self.num_attention_heads, + conv=self.conv, ) ) self.output_head = e3tools.nn.EquivariantMLP( @@ -360,6 +420,7 @@ def __init__( temporal_to_spatial_pooler: nn.Module, radial_cutoff: float, temporal_cutoff: float = 1.0, + graph_type: str | None = "fan", ): """ Initialize the E3SpatioTemporal model. @@ -380,7 +441,7 @@ def __init__( self.temporal_to_spatial_pooler = temporal_to_spatial_pooler self.radial_cutoff = radial_cutoff self.temporal_cutoff = temporal_cutoff - + self.graph_type = graph_type def forward( self, @@ -404,13 +465,17 @@ def forward( - 'spatial_graph': Output spatial graph - 'temporal_features': Temporal features (if requested) - 'temporal_graph': Temporal graph (if requested) + - 'graph_type': Graph type used for conversion Otherwise returns just the final spatial features tensor """ # Store original device device = batch.pos.device # Step 1: Convert spatial graph to temporal graphs - temporal_batch = spatial_to_temporal_graphs(batch) + if self.graph_type is not None: + temporal_batch = spatial_to_temporal_graphs(batch, graph_type=self.graph_type) + else: + temporal_batch = spatial_to_temporal_graphs(batch) # default to fan graph type # Step 2: Process all positions (current + hidden states) with spatial module # Create topology for spatial processing (without positions) diff --git a/src/jamun/sampling/mcmc/_splitting.py b/src/jamun/sampling/mcmc/_splitting.py index 561fd6e..b08e627 100644 --- a/src/jamun/sampling/mcmc/_splitting.py +++ b/src/jamun/sampling/mcmc/_splitting.py @@ -60,6 +60,8 @@ def __call__(self, y: torch.Tensor, score_fn: Callable, **kwargs): @dataclass class ABOBA_memory(ABOBA): + history_update_frequency: int = 1 + def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): kwargs = dataclasses.asdict(self) | kwargs return aboba_memory(y=y, y_hist=y_hist, score_fn=score_fn, **kwargs) @@ -67,6 +69,8 @@ def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): @dataclass class BAOAB_memory(BAOAB): + history_update_frequency: int = 1 + def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): kwargs = dataclasses.asdict(self) | kwargs return baoab_memory(y=y, y_hist=y_hist, score_fn=score_fn, **kwargs) diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index 19a4877..06fb008 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -219,23 +219,24 @@ def aboba_memory( steps_iter = tqdm(steps_iter, leave=False, desc="ABOBA Memory") for i in steps_iter: - y_current = y.clone().detach() - y = y + (delta / 2) * v - psi, orig_score = score_fn_processed(y, y_hist=y_hist) - v = v + u * (delta / 2) * psi - R = torch.randn_like(y) - vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R - v = vhat + (delta / 2) * psi - y = y + (delta / 2) * v + for j in range(1,history_update_frequency): + # inner aboba loop for equilibration to conditional density p(y_t | y_hist) + y_current = y.clone().detach() + y = y + (delta / 2) * v + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + v = v + u * (delta / 2) * psi + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R + v = vhat + (delta / 2) * psi + y = y + (delta / 2) * v if save_trajectory and ((i % save_every_n_steps) == 0) and (i >= burn_in_steps): y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) y_hist_traj.append(list(y_hist)) - if i % history_update_frequency == 0: - y_hist.pop(-1) - y_hist.insert(0, y_current) + y_hist.pop(-1) + y_hist.insert(0, y_current) return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj @@ -283,18 +284,19 @@ def baoab_memory( y_hist_traj.append(list(y_hist)) for i in steps_iter: - y_current = y.clone().detach() - v = v + u * (delta / 2) * psi # update with previous psi - y = y + (delta / 2) * v # update with previous v - R = torch.randn_like(y) - vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R - y = y + (delta / 2) * vhat - if i % history_update_frequency == 0: - y_hist.pop(-1) # remove the last element of the history - y_hist.insert(0, y_current) # present point is the first element of the history - psi, orig_score = score_fn_processed(y, y_hist=y_hist) - v = vhat + (delta / 2) * psi - + # print(f"Equilibrating to conditional density p(y_t | y_hist) for {history_update_frequency} steps...") + for j in range(1,history_update_frequency): + # inner baoab loop for equilibration to conditional density p(y_t | y_hist) + y_current = y.clone().detach() + v = v + u * (delta / 2) * psi # update with previous psi + y = y + (delta / 2) * v # update with previous v + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R + y = y + (delta / 2) * vhat + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + v = vhat + (delta / 2) * psi + y_hist.pop(-1) # remove the last element of the history + y_hist.insert(0, y_current) # present point is the first element of the history if save_trajectory and ((i % save_every_n_steps) == 0) and (i >= burn_in_steps): y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) From 20dfc1e7c7861412a9fe547fdba3cacfc2ee773c Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Tue, 19 Aug 2025 22:58:02 +0000 Subject: [PATCH 23/32] commits for cluster migration --- analyze_validation_errors.py | 452 ++++++++++++++++++ bond_length_issues_conditional_traj.png | Bin 79531 -> 77585 bytes .../sample_enhanced_conditioning.yaml | 10 +- .../experiment/sample_enhanced_standard.yaml | 8 +- .../train_enhanced_standard_jamun.yaml | 4 +- scripts/scrape_grid_points.py | 244 ++++++++++ .../slurm/train_enhanced_long_comparison.sh | 119 +++++ src/jamun/cmdline/sample.py | 7 + src/jamun/model/arch/spatiotemporal.py | 2 +- .../sampling/mcmc/functional/_splitting.py | 30 +- .../sampling/walkjump/_single_measurement.py | 2 +- test_wandb_scraping.py | 65 +++ validation_errors_plot.csv | 10 + validation_errors_plot.png | Bin 0 -> 125597 bytes 14 files changed, 936 insertions(+), 17 deletions(-) create mode 100644 analyze_validation_errors.py create mode 100644 scripts/scrape_grid_points.py create mode 100644 scripts/slurm/train_enhanced_long_comparison.sh create mode 100644 test_wandb_scraping.py create mode 100644 validation_errors_plot.csv create mode 100644 validation_errors_plot.png diff --git a/analyze_validation_errors.py b/analyze_validation_errors.py new file mode 100644 index 0000000..7ec3aa3 --- /dev/null +++ b/analyze_validation_errors.py @@ -0,0 +1,452 @@ +#!/usr/bin/env python3 +""" +Script to analyze validation errors for Denoiser models from WandB runs. + +This script: +1. Scrapes all runs from the WandB group "noise_check_experiment_multimeasurement_vs_correlation" +2. Filters runs by model target "jamun.model.Denoiser" +3. Loads the models and computes validation errors +4. Plots validation errors from 2 to 10 (assuming this refers to some parameter range) +""" + +import os +import sys +import logging +import numpy as np +import matplotlib.pyplot as plt +import torch +import wandb +from pathlib import Path +from typing import List, Dict, Any, Optional +import pandas as pd +from tqdm import tqdm + +# Add jamun to path +sys.path.insert(0, '/homefs/home/sules/jamun/src') + +import jamun +from jamun.model.denoiser_conditional import Denoiser +from jamun.utils.checkpoint import find_checkpoint, get_wandb_run_config +from jamun.data import parse_datasets_from_directory, parse_repeated_position_datasets_from_directory +from jamun.data._dloader import MDtrajDataModule +import torch_geometric + +# Setup logging +logging.basicConfig(level=logging.INFO) +logger = logging.getLogger(__name__) + +def get_data_path(): + """Get the data path from environment or common locations.""" + data_path = os.getenv("JAMUN_DATA_PATH") + if data_path is None: + # Try common locations + possible_paths = [ + "/data/bucket/kleinhej/", + "/data2/sules/", + "/homefs/home/sules/data/" + ] + for path in possible_paths: + if Path(path).exists(): + data_path = path + break + + if data_path is None: + raise ValueError("JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable.") + + logger.info(f"Using data path: {data_path}") + return data_path + +def scrape_wandb_runs(group_name: str, project: str = "sule-shashank/jamun") -> List[wandb.Api.run]: + """Scrape all runs from the specified WandB group.""" + logger.info(f"Scraping runs from group: {group_name}") + + api = wandb.Api() + runs = api.runs(project, filters={'group': group_name}) + runs_list = list(runs) + + logger.info(f"Found {len(runs_list)} runs in group '{group_name}'") + + return runs_list + +def filter_denoiser_runs(runs: List[wandb.Api.run]) -> List[Dict[str, Any]]: + """Filter runs by specific criteria for spatiotemporal multimeasurement analysis.""" + logger.info("Filtering runs by specific criteria:") + logger.info("- cfg.model._target_ = 'jamun.model.denoiser_conditional.Denoiser'") + logger.info("- cfg.data.datamodule.datasets.train.subsample = 1") + logger.info("- cfg.data.datamodule.datasets.train._target_ = 'jamun.data.parse_repeated_position_datasets_from_directory'") + + denoiser_runs = [] + + for run in tqdm(runs, desc="Filtering runs", unit="run"): + try: + config = run.config + if 'cfg' not in config: + logger.warning(f"Run {run.name} missing 'cfg' in config") + continue + + cfg = config['cfg'] + + # Check model target + model_target = cfg.get('model', {}).get('_target_') + if model_target != 'jamun.model.denoiser_conditional.Denoiser': + continue + + # Check subsample = 1 + try: + subsample = cfg['data']['datamodule']['datasets']['train']['subsample'] + if subsample != 1: + continue + except (KeyError, TypeError): + logger.warning(f"Could not extract subsample for run {run.name}") + continue + + # Check data target + try: + data_target = cfg['data']['datamodule']['datasets']['train']['_target_'] + if data_target != 'jamun.data.parse_repeated_position_datasets_from_directory': + continue + except (KeyError, TypeError): + logger.warning(f"Could not extract data target for run {run.name}") + continue + + # If we get here, all criteria are met + # Extract additional parameters that might be useful for plotting + run_info = { + 'run': run, + 'run_path': '/'.join(run.path), + 'run_name': run.name, + 'model_target': model_target, + 'data_target': data_target, + 'subsample': subsample, + 'cfg': cfg + } + + # Try to extract parameters that might be varied (for plotting from 2 to 10) + sigma = cfg.get('model', {}).get('sigma_distribution', {}).get('sigma') + if sigma is not None: + run_info['sigma'] = sigma + + # Extract total_lag_time specifically for data loading + total_lag_time = None + try: + # Try the specific path you mentioned + total_lag_time = cfg['data']['datamodule']['datasets']['train']['total_lag_time'] + run_info['total_lag_time'] = total_lag_time + except (KeyError, TypeError): + try: + # Fallback to the general datasets path + total_lag_time = cfg['data']['datamodule']['datasets']['total_lag_time'] + run_info['total_lag_time'] = total_lag_time + except (KeyError, TypeError): + logger.warning(f"Could not extract total_lag_time for run {run.name}") + logger.warning(f"Available config keys: {list(cfg.get('data', {}).get('datamodule', {}).get('datasets', {}).keys())}") + + # Look for other potential varying parameters + for param_path in [ + ['model', 'arch', 'num_layers'], + ['model', 'arch', 'hidden_dim'], + ['data', 'datamodule', 'batch_size'] + ]: + value = cfg + for key in param_path: + if isinstance(value, dict) and key in value: + value = value[key] + else: + value = None + break + if value is not None: + param_name = '_'.join(param_path) + run_info[param_name] = value + + denoiser_runs.append(run_info) + logger.info(f"Added run: {run.name} (sigma={sigma}, total_lag_time={total_lag_time})") + + except Exception as e: + logger.warning(f"Error processing run {run.name}: {e}") + continue + + logger.info(f"Found {len(denoiser_runs)} Denoiser runs") + return denoiser_runs + +def load_validation_data(total_lag_time: int, val_root: str = "/data2/sules/ALA_ALA_enhanced_full_grid/val/", num_frames: int = 100) -> torch_geometric.loader.DataLoader: + """Load validation data for error computation.""" + logger.info(f"Loading validation data from: {val_root} with total_lag_time={total_lag_time}") + + if not os.path.exists(val_root): + raise ValueError(f"Validation directory not found: {val_root}") + + try: + datasets = parse_repeated_position_datasets_from_directory( + root=val_root, + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + filter_codes=None, # Don't filter by codes, use all available data + as_iterable=False, + subsample=1, + total_lag_time=total_lag_time, + lag_subsample_rate=1, + max_datasets=5 # Use a few datasets for validation + ) + + if not datasets: + raise ValueError("No validation datasets found") + + # Create data module + data_module = MDtrajDataModule( + datasets={'val': datasets}, + batch_size=32, + num_workers=0, # Use 0 for debugging + persistent_workers=False + ) + data_module.setup('val') + + val_loader = data_module.val_dataloader() + logger.info(f"Loaded validation data with {len(datasets)} datasets") + + return val_loader + + except Exception as e: + logger.error(f"Error loading validation data: {e}") + raise + +def load_model_and_compute_rmsd(run_info: Dict[str, Any], val_loader: torch_geometric.loader.DataLoader) -> float: + """Load a model from checkpoint and compute validation RMSDĀ².""" + run_path = run_info['run_path'] + run_name = run_info['run_name'] + + logger.info(f"Loading model for run: {run_name}") + + try: + # Find checkpoint + checkpoint_path = find_checkpoint( + wandb_train_run_path=run_path, + checkpoint_type="best_so_far" + ) + + # Load model + model = Denoiser.load_from_checkpoint(checkpoint_path, strict=False) + model.eval() + + # Move to device + device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') + model = model.to(device) + + logger.info(f"Model loaded successfully for run: {run_name}") + + # Compute validation squared RMSD + total_rmsd_squared = 0.0 + total_batches = 0 + + with torch.no_grad(): + for batch in tqdm(val_loader, desc="Computing validation", leave=False, unit="batch"): + batch = batch.to(device) + + # Use the model's validation logic + sigma = model.sigma_distribution.sample().to(device) + loss, aux = model.noise_and_compute_loss( + batch.pos, + batch, + batch.batch, + batch.num_graphs, + sigma, + align_noisy_input=model.align_noisy_input_during_evaluation + ) + + # Extract RMSD from aux dictionary and square it + if 'rmsd' in aux: + rmsd_value = aux['rmsd'].mean().item() + total_rmsd_squared += rmsd_value ** 2 # Square the RMSD + else: + logger.warning(f"RMSD not found in aux dictionary for {run_name}. Available keys: {list(aux.keys())}") + # Fallback to loss if RMSD not available + total_rmsd_squared += loss.mean().item() + + total_batches += 1 + print(f"total_rmsd_squared: {total_rmsd_squared}") + print(f"total_batches: {total_batches}") + breakpoint() + avg_rmsd_squared = total_rmsd_squared / total_batches if total_batches > 0 else float('inf') + logger.info(f"Validation RMSDĀ² for {run_name}: {avg_rmsd_squared:.6f}") + + return avg_rmsd_squared + + except Exception as e: + logger.error(f"Error computing validation error for run {run_name}: {e}") + return float('inf') + +def plot_validation_errors(results: List[Dict[str, Any]], output_path: str = "validation_errors_plot.png"): + """Plot validation RMSDĀ² values.""" + logger.info("Creating validation RMSDĀ² plot...") + + # Convert to DataFrame for easier plotting + df = pd.DataFrame(results) + + # Debug: Print available columns + logger.info(f"Available DataFrame columns: {list(df.columns)}") + + # Determine the parameter to plot on x-axis (2 to 10 range) + x_param = None + for param in ['sigma', 'model_arch_num_layers', 'model_arch_hidden_dim']: + if param in df.columns: + values = df[param].dropna() + if len(values) > 1 and values.min() >= 2 and values.max() <= 10: + x_param = param + break + + if x_param is None: + # If no parameter in 2-10 range, just use indices + logger.warning("No parameter found in range 2-10, using run indices") + df['index'] = range(len(df)) + x_param = 'index' + + # Sort by the x parameter + df = df.sort_values(x_param) + + # Find the correct validation column name for plotting + validation_col = None + for col in ['validation_rmsd_squared', 'validation_rmsd', 'validation_error']: + if col in df.columns: + validation_col = col + break + + if validation_col is None: + logger.error(f"No validation column found for plotting. Available columns: {list(df.columns)}") + return + + # Create plot + plt.figure(figsize=(12, 8)) + plt.scatter(df[x_param], df[validation_col], alpha=0.7, s=60) + + # Add labels for each point + for i, row in df.iterrows(): + plt.annotate(row['run_name'][:8], + (row[x_param], row[validation_col]), + xytext=(5, 5), textcoords='offset points', + fontsize=8, alpha=0.7) + + plt.xlabel(x_param.replace('_', ' ').title()) + plt.ylabel('Validation RMSDĀ²') + plt.title('Validation RMSDĀ² for Spatiotemporal Multimeasurement Models\nGroup: noise_check_experiment_multimeasurement_vs_correlation') + plt.grid(True, alpha=0.3) + + # Set x-axis limits to 2-10 if appropriate + if x_param != 'index' and df[x_param].min() >= 2 and df[x_param].max() <= 10: + plt.xlim(2, 10) + + plt.tight_layout() + plt.savefig(output_path, dpi=300, bbox_inches='tight') + logger.info(f"Plot saved to: {output_path}") + + # Also save data to CSV + csv_path = output_path.replace('.png', '.csv') + df.to_csv(csv_path, index=False) + logger.info(f"Data saved to: {csv_path}") + + # Save validation RMSDĀ² paired with total_lag_time as npy file + # Find the correct validation column name + validation_col = None + for col in ['validation_rmsd_squared', 'validation_rmsd', 'validation_error']: + if col in df.columns: + validation_col = col + break + + if validation_col is None: + logger.error(f"No validation column found in DataFrame. Available columns: {list(df.columns)}") + return + + if 'total_lag_time' in df.columns: + # Create structured array with total_lag_time and validation values + data = np.array(list(zip(df['total_lag_time'].values, df[validation_col].values)), + dtype=[('total_lag_time', 'i4'), ('validation_rmsd_squared', 'f8')]) + npy_path = "validation_errors_spatiotemporal_multimeasurement.npy" + np.save(npy_path, data) + logger.info(f"Validation data with total_lag_time saved to: {npy_path}") + logger.info(f"Data format: structured array with fields 'total_lag_time' and 'validation_rmsd_squared'") + logger.info(f"Used validation column: {validation_col}") + else: + # Fallback to just validation values if total_lag_time not available + validation_values = df[validation_col].values + npy_path = "validation_errors_spatiotemporal_multimeasurement.npy" + np.save(npy_path, validation_values) + logger.info(f"Validation data saved to: {npy_path}") + logger.info(f"Used validation column: {validation_col}") + logger.warning("total_lag_time not available, saved only validation values") + +def main(): + """Main function to execute the spatiotemporal multimeasurement analysis.""" + logger.info("Starting spatiotemporal multimeasurement validation analysis...") + logger.info("Filtering criteria:") + logger.info("- model target = jamun.model.denoiser_conditional.Denoiser") + logger.info("- subsample = 1") + logger.info("- data target = jamun.data.parse_repeated_position_datasets_from_directory") + + # Configuration + group_name = "noise_check_experiment_multimeasurement_vs_correlation" + project = "sule-shashank/jamun" + + try: + # Step 1: Scrape WandB runs + runs = scrape_wandb_runs(group_name, project) + + # Step 2: Filter by model target + denoiser_runs = filter_denoiser_runs(runs) + + if not denoiser_runs: + logger.error("No Denoiser runs found in the specified group") + return + + # Step 3: Group runs by total_lag_time and load validation data + runs_by_lag_time = {} + for run_info in denoiser_runs: + lag_time = run_info.get('total_lag_time') + if lag_time is None: + logger.warning(f"Skipping run {run_info['run_name']} - no total_lag_time found") + continue + if lag_time not in runs_by_lag_time: + runs_by_lag_time[lag_time] = [] + runs_by_lag_time[lag_time].append(run_info) + + logger.info(f"Found runs with {len(runs_by_lag_time)} different total_lag_time values: {list(runs_by_lag_time.keys())}") + + # Step 4: Compute validation errors for each model + results = [] + total_runs = sum(len(runs_for_lag_time) for runs_for_lag_time in runs_by_lag_time.values()) + + with tqdm(total=total_runs, desc="Processing models", unit="model") as pbar: + for lag_time, runs_for_lag_time in runs_by_lag_time.items(): + logger.info(f"Loading validation data for total_lag_time={lag_time}") + val_loader = load_validation_data(total_lag_time=lag_time) + + for run_info in runs_for_lag_time: + pbar.set_description(f"Processing {run_info['run_name'][:20]}...") + + validation_rmsd_squared = load_model_and_compute_rmsd(run_info, val_loader) + + result = { + 'run_name': run_info['run_name'], + 'run_path': run_info['run_path'], + 'validation_rmsd_squared': validation_rmsd_squared, + **{k: v for k, v in run_info.items() if k not in ['run', 'cfg']} + } + results.append(result) + pbar.update(1) + + # Step 5: Plot results + plot_validation_errors(results) + + # Print summary + logger.info("\n" + "="*50) + logger.info("SUMMARY") + logger.info("="*50) + logger.info(f"Total runs processed: {len(results)}") + logger.info("Top 5 best performing models (lowest RMSDĀ²):") + sorted_results = sorted(results, key=lambda x: x['validation_rmsd_squared']) + for i, result in enumerate(sorted_results[:5]): + logger.info(f"{i+1}. {result['run_name']}: {result['validation_rmsd_squared']:.6f}") + + except Exception as e: + logger.error(f"Error in main execution: {e}") + raise + +if __name__ == "__main__": + main() diff --git a/bond_length_issues_conditional_traj.png b/bond_length_issues_conditional_traj.png index 1063ab390a3dad19ffd63ec95303474e432a707b..eaae7ce0811ab7c59133fb560876e2cf3f1710c9 100644 GIT binary patch literal 77585 zcmeFZ_dnKc{5O7*A{vAg$|{K>EkzLt6_uGSGzrM^(WR{&Rl!jzx@4Z*H z`+2yo>;Bxoe82y|eS2J^!Fj&V_xm_rujhIl-sjH9ucc?Fr%)(sPb$c&P$)Es6v_&2 zx>fj-eJ#P%_@B7V2~8VSi_13lhE|s-N`^K!u3Ok#zh=a3cgf27nuYm3VbMLp`*v|( zwXwNjEiNKr_P@VF*uv_H$mQ)R_wg#LZzyP4Qz+{V$-h)lQjymvR20fdS!p$gkby>f zUDfkTv?FH~xYzvQSlQ6_%3nv>v?yE1Jj7q)eYLWV1WSr-RaJGhHtp3a?f#}JLGvPS zoqo!WFXx_Kw{mrxKE2g=!?p*EBjc{w4m-O;oQD5yeH7dgZ2dEkj)%tUzpn>RPAbzL zll}MW2{V7_+kSV-K<-w$G2dqr%#`bjEv;bayh&ARz$WhF3fpa4YbX+YU#AaZkrkS zST-_VCKYk!rTFyNHxW_e;x)B@{`6i?P`sG4U~FdQdv`5|(bq@YzkL6`zhO1Yp3WKH ztw%GXj+g7)xUpy3hSj(4++mnJRFvVry!IbFkBwTvz4suipjuwutrb+1-rim>KfmG( z>+U^HQ`f&g-$$z&d8Wv|*(LJn4U^hL9`nY`l{Mpdu&ivFR-Pn*PT1WE&iXsxZS4;?-n zrl0K?bmZoT7+*ZkB;yjbrSppdHGYoPyQAyVEF!Lyy}Ds#rK02CtEW1n5wGjrSM|QI zp@CChUq9%e8T;?wzborHlKpDx`?YquxlNDoY?XA5+L&fF7gC+)IMKs&cRiQi@3*H5 zzkgqIsq{%cZf4c2+R4M&dxeA;K3*x4%dqO?e)df{!|KvJp0aqER&Z?Wu3cImXi`jT z?s|Hzyl!Szo@r}3(D>{^CX7+>Ez0N1-hha~Q{QWC$Wwm#xDcdL1b#}GS#@*DE??Iv~^7Hd?M|duO4+^Wx z!6r4C`(!`7m0isI0agefH#bE@Kg<33^FI>C`pH+BNJYUL*$g#o#?C3sx;cJhX6({H zbFS1`Q;R=;{umZN*sPgpvy;+NaaKybD#fHqu*7T_@7eVum`_}sT`$u{da|#|vb}h1 ztk%a>PE!N<(W;T>l8jFH4$e;Y-~09JoK4R!DwSv!vL$Mhj15gp?g#M6=j^-8Lo>gy zkYrYWP%U0}Gj2iA`|}BjH^(oNYmu5CSyAlIwVu+`lr2&Ebhpp*eOC(V>vaOt9p@~< z{imlLLL}_hX=rFfsl2_f-MC(A-uw0kMv9YmWf%{;%VO1)|do1o`c*a}5O#g^gW_4wkH zbl&vi#eqB7*)FfjcJ2rZ3mX|9FT{2J!qY{$Af353(Qw7oaI^ouD?hJ3kn!>Nm;ZSA z#|2~KJ5(!HapMlTEzazipP#=m+IFHcO8I!z6t?(P6af_#m0usPoF%WqCZaDp-t|NI zlbu=00&3Kj(a}+IahbOLPurVCbN%<3k?)Sx+6RpATJMuxPk99G-o1NKrB=|;(i-Vw;8%`|Q&3P4H!m+Ws)hWqW0WIDkB+mfpv+?>ZrXQw&A{N` z@)DJhXrG@MpC0-&%(!90?S~H;s}5Y68S5n9($Lsg9wEQ}khnPS-n|>Tf4(gA+amER z+v)F-5@`Zfp6s~o<-sW9aqk`jrJ$@#q{I$)W+UbEYw0^`m(Y5s-LYrx#tf8jxv@|_ z;~Bk9@(&8yl+x_t;^G+}8L?NG(iUuyQ(jk$C+pzk^gi%Mkmzv=V^>#K8`e(D*!S;> zqpkP2=qW~xnYM>t71oQ=(9mR!rw`Y?Qwlyv?XH<*A z-?8TyC~XP3iz-6rC5NcIMn-O^-#mBjA%*+)?c2%4y63P^T#`*|gsF~U?JGyP7#W54 zjXF&BsZ$-B=qW!Qp{K3=LgCaYm4&U0jIS;IPL)d1tzNAh*qCm493M+L&0LR(jV-d; zb*Lc&cQ^e;SWGOsPywEUth~JGfYiZ*8uqln5KK(08_w`-`~CRdJzC0OMP71JV&X4> zZ{LiD&CSh8t($Bz|MglrN`-{Dytjh#GgK-;P~zs;wyj4T+<(55h`2f46=Q{_ zf=fRp7}z!E+?;x)Y%5m%1?)EVh_Dx87Jk4YhRr!HCaT?(JoK*d?y`H{UzhY>P*YQL zKmJ}KZhWyGm~jc|l&w$6tEJSQY};DBdzbaZhJ`r$YyEqMyc`j+%O0k1q5 zGfZkStgkqa7O`!Tu-mn6i?|2c3i%9Cx*1ll{B%Cq4eBawcGf|c*k8)N|0;!nfx*Mm zvvcRutW4*by_CF~n)Ak|LS%WRzdhY;*9ik;3^*#6V*^)f1O+v%X91 z#lE(SFHLIVni$GbN_@l;1X!m3 zHoKL-mR_NL6L)6+qy2lIuZfS3*Di~`R`HH~E;Tt>cCapmX~Tv`k+tJ!uQpV$AnHu+k` zp3C1iRayFR9tO)ecD+7T$@U}djvYH5aEPrlaG6wp#jI7D`xS8f+50Jg5YLDRKG%iG z^Sku3HeqeuR^&epER$r@tMWwXyzp7e-k8Y> zAd-i^zQ<+O=7`%O=i=hxA=|zu`XxWU8TC^bw`_^~IR9gPe`)y#bP z^E1|Qlci%~kvCXbI5>iN4-W-DiO3b!%Y3u3CRj9Ux45{tf#uGn;>@kInSQ}sr4IJP zO>22PesUQYG&>JU@$<`>x)_G2x(4pn*=Sl5e?n1ljj%z^gPHN}<8-$GUt^RSt*xvI zK3w`XGSRErkad$8JK_#LCIHnfQi1Av=bY82w2X|5{M?wgqe3w8$+Z-CGI)7==VQ~A zqo)tlr7*k{w=K2ayG(Kf62Pr~7v?T4`d|qi2O^aB-zQ71U`u5|k-po7pmkR5qyH{4KrF=@-IV=}xmh5l!qQ3r(O0>SHkiY%>w54x! zkPMxA>`X>|sIYB+^$8ObLDa5DW&45JBpC^EXD+bNdur?EEp3Rf01*l)Za!&HdLSzK z=sH3NY_pT*O@N6gEx3l2ITkH$-8Qom*L*XdJYlI~DHLPvvR32*bhGWN%&&}ARrh`m z>~SsmvBEZ#>=+aJvHHOYmmI+*{?cfV*&8-CvVVsfN4rDa!qJ`6ZTmNZ(?tGtU1mRH zcfhc8#Ze4_g&w|o<%dliX4CuoaH^EsVpg@XNNG8bg>Yr8uG-Ak$2(ST8a2i3k}z*b z51Ty-Mq=PPv)#Rh;i8aww8~19g96kUHSdj=vYK7z;;LOF4jtk?cyQyjYu7UF7k&H2 z5T}z?j9Z&{`3HMSCct|Oes{Ivo$t?R*O{;*HF$5Hefw0^7x9G2%0*>av1*OLNW0+M z$Vg+LA4+9SL?3|U#l4$1is(N;ZHg}4w{IWHs{w@q&hV(a{h&n?CmK!RtD{ag#yU=_ zXE0LZA}@t)vjJ*b1%g6e7`3J*s_grBHJ=j(O-)Utqmjk7TPHQD%4KQcuZh3G+X835 z>~BCGqa&@yeaF%gR<^dbrlWF6)Ki^{+cGspy`+4b4?zv<2T?mHHNcSE?y)c6v#E*bwM0_HfM3xv}*Rm z;Xmk$t}~r%q_D9HX?Xbht~GDU@^+b<%K!RxHPDV~idn_GwW-(t-1e88r6c?)Bjd11 zb*wikYv-JEU86sbBmb_arc4)NI9goivNL2a${CeQG z<0>i}fNj?r7#I*3W!!95mz?j(z($~S^3`%`{KVC!tc<$Gp{7*l8LR!czkKq3pRtCN zI^Qk6Su%Mr@&PSy#^;RaogeY zJSeCot=W|nwenOmo$0x`!otE8@_wAs7cXu@Q}G1i?Mo>GY<_t%#cr@J%x!70_QNFx zaho0rKB)6ZJ}nyLDoP80^dZYO8no`JjcaRaKvq0JOTuyYdaB;D393c8pMEAxbP1>l z+x@tt;#0r4Jmy!rU}$*j?dfN{qN2(vu21)#)C^jj>E=>=b!0Q4CI0^F@qz@wl8S@I zOl~-gCBn1Gf)E8+6B?$~JRWrG*RxS7T|vw}K7IOhwJ|eg zx+~b?GZt#HImd<#ycDm^`z}Y7{(N;*2;#wZTm;X+zyM&1XOvp>qfJt-Sq>9D0d-T) zpFcMOkSh50?KJiS)iIEQsG_&?dK?xO7OjJWaT*7qNl_j1@bK`$%J#xzTAiUQT>4}u zU(R$J1EJwbp@3MT=sO_bJHLPbE{mO91cJg1^!NI}-8XNxiCT3ivCKWeGMTbtLJE-HebaY+QVp4K43s&TYTeoiU9zLwqpDJNL#FCSf zQvhHHi9`A0Wu>I6<-7sUpGQ{@>G{*@iE>G~>Ly?RqXKDyj+T~ov|OIc``&u4uG-v; zqq$;k{E9)ZO(32yiy9WtC_}>FdG^iI*VmU+>`kIZ9*3N#Z^XB4l$mSSq2oa@+J$;7 z8?Evd>PQO6@11nJ!52LZ86PR!`Z>N%xyF9x7}@BM|627q#iIJ~MD#EhQxd(s7(O;I;aDjo}7!fOjB%9;hl8pzEO)v;poo za!HAbGNDA>28i#lIt+O3z0c+m=`|8uf=qU!y5Xm}?yH8)Alj~t)quUUUSUN`^I_R~Z{dA#FnL22o! zDpwSo>6z}(^2oDXxSf0g0(20|@&PmrjmC0^jlp;yG`lTrfjTFL{g=G;6A)Y!1ZyCB zKeSUB-hd9l%e>)juTgn79lF6yIYxI2@(Hv#E>w$Y$PPqQPB*YDAoDR^j4eVWLl zM~{f=`GE8A2A~>4LqljJD{(_TJHI_Mo*rpEvBku1Y^*5>cZDFdkdTly%-fhY(@M)e zkw<^sK=}+_ZgpL3xq{NHT-n+00N{dcCSfqgZCjIIzy`4jq-QfEAr%TEDwTNs5-vrC z|FJiOEgvXzDGPWw_vu*$LD9eKZgHBO=!G64ZW~iQ2Ho_1DV_~Vy5}ZI=kyrR09$JJ zL)KjnVs>rcenKtkoHTlD?)dcSr@NRapP{N+v5GJSBr%ZI;Sm}dDn}#?bgXy^H#mw4 zP5Cu8Mn-N5q>vH&)E~X`O=8c4^)`^2tEuS)&~mjhRMJHoLhNd9Sza`ERT`$Qv(nl2 z+rl4ts?vE2lwm9I1StGMA3rkOKG9pzQyI<7$;o-a+(?(_yXFl9LjHu^EU>WBf9uivkj2AY7iJTalhs`2h(?SDU074Ywsq^)J^S|E zfl#t){niH{hTZA8U;z=(RZvphKYfyb&gl=3A14}m$ddY~%iHeGDI(U5(DmeiqfPU>@q}u*&P=1MGTU}k9n38e=K*@4p zc9QyVnuiCCVyNVnsu=a-xSk89rVO~BXqqg<@<=(252x zRgbRAo~c_r*y$A7(sIE!6PHh#0jli1=g&8jLSx===yOrgwdA#{S4-bpx22`El`2Db z9Ryo~W^o5fs}AE`w!%xA9xdT2kJAzJo#j`0aRHSK53(WM5`NGs%J&xs7&ds*fM|TG zUNEmsq{44teP{^bg-X1b_*yzUSK?*u?d^efXx4K{R@UAd8yy8{AisywwMSI6e5Upb zgP!Syoi(SG6cq9xb*{npYodxlj!3uZ-QZqRA-sZ`x}~j+8lM^XxD`r}wtqy~O=jg* zI6}Oo%sezig~#22kJfpO3k~iBcC+E-%c}sdAXB!W?&X7_S&p@74-h zBYYfEwPybguLnEf6t|%(ytlptt1Ib7tBmhp&eEb2zW+7u==4b8dY;GGIh;+IwhR@K zXXT-X5^Wb3+KK{On{LTBHPq+>44RLQYV1Z71|n($7lxq;9Jhmft!xPv6kJe z#Fv8>5)wqk6@YmTvlBbdygK3yUh^7I@TR0#r`jQ91)s-4&CaQ&HD@AzmX)o~U7Qs{ z?|7H8s02!0gd5$`-5q&QiID;$r2NCB(*q3|%s{5cLGM~RI%wd60s2MP>E$?YrjSR` zGD&;ygwIC9qWi2jY;9e-GaMWo@=&9pG!I`tEvwdW>EgvG$%)*f+l53DN+6-fh@4-w zW|MnQ&((RsMd34Mn4sye~IU;56fYu7)z zE;vGVu3}N6z!C5a4dp_?@q|qB+NcN~u~!~k+C;Y{SM<77XbtD<&+6)OzSw__8jund zn2(Qd1yt`@n`^IdS6d3#OUVMkz^@sF{&Kx3YcuibP|eF>GeEy0oG%?N6Y*CeKww>R zn5deLTQ$k5eYtPa!^XsFA!gpdNcrqewW8^dM*W_V&!1%tuUy#y(3o!3c|S%s?3@=I z83u|GYP88*`9aSS&@N`mXW~n~iw7=oOIGNR*BrbNd>@ygP# zv)+9E(PXyIZnKputucQ!YojtAzPYbPRN2qgn3hC+8?zNxn6JoBzYb(H~m>6z0 za31@*g8+7DB=Q$8UGj#tNzf=tXuti|IVIF|d~^_}i8?>8FPnBo)~ch=;pdN&e^Dg$ zAj8R#-Ule_1B{_>7Mzu}Rmi~f|J47{nn5Ki>*i)>VTho{nsfp?sSXMysM0~{6Qp>- zlDL|D_Vj7ylP6Egc-15fZ}65S)gG4=o${@;bc2XNj@m-Jv)TI9tIKPzz&I#&dQn@e z3B{d{up=S~RlDNhu_xF=iejNu&d&XC$vzE!AK!e(Wv25P`uG9Ur!jE)F3Si8-v9DN zAtkjeREllyb4BoVUG=8{0X%T)gkt03fR2K+=isH&MXL#dpT}y%&j*7YszoGN_v@+B z9hZ@zp^)%EiWO*El~aINvji&FSqh__ot;tJXK$IXb5VVgi_^wbM26gL(5)ov=}%hr z*;hwyQd2V6d&7UAo-lH8`D3dP(73GEqy9+qKfbo*)lw=da6V*t=y$5V;nWnANaw$e z_7r!i#p&b5pOzN?=4$6&1MTs=bLWoSO;kkXY)4xn18cVj?J=YV@eQACEGVGHy}rXp zjW($kxCCzly`vb5(+3+lW)LVjudtBMpc}^S?uJa;)20td>O$JAuO;|Nb1eV1Z5<|3 zl!>|;EL>67c?SYPISyW54p%gOvOg{uXappWJwUGk;U7DHVyKY=0Su6$Vk~BxGo5OZ zu7Unnc*7sHV1xW;bKn8A0;=c3O)8Jqld$QlJQ-)EZ{Qyn8@tDC(K#jfD$3I_{>QCQ zIAT7aFtx&ZC8Y`l@pY=dNbdaJvWWEN%Z7$w%KXp1sf&$+GBC=ZwV3X@AQE_LoV5gt z7k;-Ff*3J!D}}Hj*2CwmaIDcW6O|Ab&lz`sN>6#suJx4pQZN>R`X9tM0$#q18{OIl zyW*IJ#?!i<_gF^6f`n3BwP|-v)=g`h{_3@;rxoc=iRk4UcWIuxD8?|oI5%9?Br!SM z%td(^80ZPf{C#8ATZ3Fzio4MH_^8q`M^uvZL!J-4#$+>I-LX29mGh(`H@Wdcfzb2k^u55yrsMtEi3KtMxqilgI`pRyMQ-v{LuJ zzvxNCA^}B~$0jD%0`R=z%Ur)D@j8Yk1RsBG>FbM`%8XQ`7lQ|kwn4oi2Dx_f)mx5}ebL*`0-O@(m-EmK3ZB++koWTN6Ab%JDiyj-oy=bzb<&tj zNKJWybOBJ@FmrtZ(5YrFoU zt*vXBY@Zr@%kab&@24KvIho}hb%W&OZidfSpR4)u)Lz-u_^m0;v*2J3V*8=CzFZv( z6A8k2b{K72>0V>1+tq>Qg~ZS8+*~O|B_(39#A+r{=gyD$ zn6!=q*qFvDPaPB!6Qi8dd9BaZNoU-7Ft1;POTxC#W3WCAx4wR4bTl6ug(MLsz=Q_B zo$SoW&VCh{YVD-zx)eVXP~o)KtpCQ&{wD_OsBcPH@CxdN$SBC(Y#SI@)8ZyFGOLnv z%VBXoquTM(%`*(Hg8h9p3G0cqV&-F$JEPPQ43WXm+}zKHO@x@#1y~~f@ONTL$G`pj zxlV>IC2W4Bi@7J*z$pMP4J8AI3Y0)~GpmoGVrgIdbBNZAf^h*WXU~BH>b?q2*0Q=- za76r?6n<1%k}x)^D695pr#zR?aPcaZ@BNl%o_zJ_>FMG88_IR&%o)wp>ziLmy2NN7 zYAi2jf{UMbdnJR~AcTvNM*HSg$PW%9EmRr_`fCBU=K+aL_yV3(@HTVaihDEY!*;1z z_mkc3l9~ZA`IVCcf0&`(m*BGynDk(iC^e|f9Wc;3x`Spf4w})!+Zz>o>{&+~=?+sE*m5FGypvJO8R4+O_o&F(`j2gErY^g2MllF_=N&N22w!#Q|7s__>?j z0%TKO+t%d)ms>6^x{$;`caKoZWZFAG?;>~^O*5B8(uSII{lU#mww;0H@a_OrxJc){ zdjJ((wefp?#&HWYM_uz-0(j6cef${ic|LUN$TKg0PfzXLy?ZS_$~{m}L54A)=q zWe)T`di=O*NI2L4REb<1x{77bFCI$5@VY1^7@&Ku*X|)L4pmIWG?MeyMAwtO+dH3p z6WeKU2kSY){(EV(Z|=er2MJ!G2zuPT+m)__@BWp1P4$S&EDN|w)t@1tK9Xa>3;qHP zP057iKzJ+~r@Ks89?)fLON%?2|Hj%3)+@vPH<>p}5MApmoro0SdxhHNm191GhNQEu(`0S~m};!6-1*8-M?ruv(}s%`Sgm)}{Rd(#^v= zp9_;&2|rcNESR)kY(MlR<-n8MHjuRh35*~q?K!m9HsJ4Npx@l$3Uxn5^GS?G9F~{Z zyHn7H?xM&R$HfD@SR$AOIOh%w6!lpZ{&;U>D21|HJNaB7TV;QeFo+n5BAS>Li2ZoJ zZ%4DiZzOZtkr`p{EUmHZd_)TdA7d-fOfrgZA|v1Lj*8cHqmtLwIwCOnq zt-AevLXJxR@wdbUd~piC@T0MBfnq!5qp#^SESQmK78>RraB$ca@)3*Zlq7Tok&#eD z-=fMuQ$0wa7Z&GdAcxCCf7xHX0_ch(D=RC}Vbq8d0(Sc$d@Sw9kBb`T(AtarOfN7g zzp_R03Ch+X02n`?a8qMss35n70|pSX1M<;2Fz^5jw;foD2$G0ze6%0Bf>hNZr>R|t zm-wJVhJl$7yBsy!5)mLQFhY~r?dK~lf%k6`Gw1v_jll<75!%6RMrwcsjZepw^ht0N zOsr^nw4ES9D8XBxeIZIxfIeFut-52EUgrL??stkU9S5Tw@ZwdxqotFt3AC6^Z0qRo zfZG&P2Ptqf(dU2^m9Z8Ho&!8DAmtq^FIG20WiVcfvd3ZM0@*0QiY7iZXJ3l5LVYVh zxgznq$^-*0q9~#lDMy~!qWDr={YnEibAd(f!d+Of(Ytw(&$tBCw;EBe0^E`SU=E;X z;{uP>?EMB?q+B_neE=Fb#n0aNib@AJdsSCZfdzoPPd+M23xbCz(c0?g&n1)`o*^Gt)`6Xo(z zdY=x~rCF>;rGE4FZ7WIA*bQt#;f>MObQ|mV3M+^S5ll~@a44lMked(L4{_q}??b4~ z9?#;~f7PAn5m-Is^+9cK6%eDQ`V`b4*s@tqMe` zbk6#J5^T`t*lkgB+3+HS&yWNaA)?OE{)x>K5to*h2FI-0@d1<;)nv1J4dh=L5Jywa zw7FIuqb@{DA0}qz2&U5t3SQL<0xa~SfIRn+SYd!L(%RL?^IM(57e!NOBnS0xi>2P%Mv&#XJF{0=zr_9${@MT zp#9e#75CqL1buNinpeHh;~1Ai{HNNa2xM!Y5NC+oL97wx|LnE0$r*e_mu{7E-9_xY za;V)TkB5@v@$A{Np2kmUXdUV)PBUX~>DH`a!5Uy08X6)`0D6%Z>aY(4Kb>|>Yz8A_ z)C%(RrHNLI20{KpqNHXDW!(qsjRCwAA{;-Gk|!Vok+M*w&nIS9w+*5WA1qt$UAuRC zLTdmp_tQ+gaI1Rp=U4Rbq9;35Ww8qoT@KtMCdLL2k_FvkBXIGsL>f4F7KOX?nMkPd zryXZsc>z>=pfd}DzkiimQoaG<^ldx>DprqthGzD@?`p)PsyIvQw(pPO{ z>imWP&%KV04*MqY7hi%%svA+?prcN9%J*cJzb_*XV<(-b(SQ@}2DSZi!i|d1e^JDb z_x^rMjHKlR)JkLkmw$!-L=-b=%5m@6!wAt~u;j-NC3O*xV;3%PA31V_=*6Jt*BZjb zjvn2DYJBJMV`hM|Un!>Nk6`n}{rL7wMD*$}ZYaS82#McWwf1+q?I`XU$p%2|A&CS+ zmE`2)!d^-&8)r;R<)8N4y0k6ke1ad?XH^-BxSB1)#$9>vL zJb7a>j&5lL=823GGZE7|IxZ5~8lOAnAa0yI%$#+chul}tuH)XdOL5RH1Ajxj03w^g z(lG)}tIf21-#1mbe1q_x6Sq|9e91?&8>qzw+T$zELxWE{f;Vb|h(J5ym>dWxzpE_N zAEF})J=M>jKc^S|HY3cwp5lNsH#Ss~MYALkg^`cPPzF1^Zy)R)Dhl}psg_u;*ediy zizUJu8WEZsg>t7GeD?lBY~L2xW}q70!f8u7wLo#6eVrm=0FTe)EP+ z_~jv&+5IFC0KD)TuT1`i2&H(ekVj;(K2nNbA>1=DVN*2>DoCQ{h)!+Wa8lZi8HVaB zIvzsPp|1x8-Jm$S?qs1q;xx4e-hp%64y>WPn%PJGt9SAJf_g@>?`Z$UXyl`VY0csA zG7y7bg(L@sf*hbpJ<@J%*d`SBs3-vx830862w$r?{0^G=RnGdNq7!jH>PwQ+(g;zF z@jVE;u?RBy*FvZHX;G+Ns+EQ;PNe}&xPirY6DGH_F>A`bK7Maq%u)^DfYOWo z(w_9JWD}4(K-r(u$Rd+9&QX?Df7LYf*+Gk7%v#_%y_r3V=ZnpT=)GZ1#iwPtq11Kb^G7M?E-n6`ml5ccn06A=M#S=}r=B0@dlZ9_xD z!+?NWAkHzq1^4~@ z&v)TRSmi=TQ*t`mK9~1jzIdk>{Wv5Dg620Fw(vSjt4EQHxPq4YGaf?c43dHKD+=DgxOcAt1!u z(F>?l-ziF?+LN>o5~{%j_pKS*4Ns;s5J7ZpUS6h*cx z3y%Zp(7ysYB|S3f1s@V_YKG5;*Vk7=6fy(0Yg) zj)k#`I5W`FkU-gg?(A{|6IXNX5gm`LGOhIX(?&+Th?)jGe0Yw#-w46(!lEM2@h{fV z+x}|V*&PI0V{p#J<^_5~rf-*}i-zVu>k+!mK0K|t-(6-aXX<1#hRYgs(`Fiek>#;H zrL4(z@SFdm`Bxtj6U!#fKXfT3o0%`s>r2mS9!Aui>FH^i`dow<5Cmg$`M&+1{ZNrs zhUJE}QM`D7diH-_FQd$1kT3l2O3X!GoOr@@oA{2`u4%Pc{PJLg_lcxwdvNZ*TPwF= z>Nz9z&Ye5)>{n4zK252V|LV^C(S3J$!+Ei?pq+Yf$jWB`wxO)u%G=%*t8UhWeDi^;nkTA5W^KR zZARwf^H_Qab6LVH(JhKJZH6R%3v#NnvvacfjsS%VaH}wQlQM=njAc#mF@AJe&Vm&J zZ|T}VBqGoxs?+xuof)Hm7f>Zk2FY9&0CiVih?mz`#5RMp?Hjy{Egc&95kv6sNoQ70 zVqs>!v$!B7r>pCDYv3&1=ue($KR>JMfWByMH^qygQ0%cA(42>iPSQp(&e%(2jXxDJ z@PkOzeXTKw&|#Eel2uD|oHRjQJ# zPgioYo!#m2nU6H7si!YoNt3<8k>R|zt?hRJMr0x#*mh~UpOu`U;Se)dn`2~QV!8zs z*N*VFI|6nPr5;5t(5_q=hW0UyR6Yr~wn2-#SqHZrK!4NztBj~6{`ezPGcnLQVvixG zAM*j}FK&FDEm4P1#)VRrh{THX;5{#a+*5V75*Woo_h_>X_VT*F+Yh3U(T{=xW(^J2 zu?#jUPno>Wved`^3dBwvSvNJ?<|9IeE{$PUUae1!({98z{{==0#6+w=&1}a5h!@1x zECzDvLqxcOz4jV!R-1KmAI!-Z`&hsa=nvA6AeE8khwMW~csyEARb3U>CIr+XRB8gk zj=0W~+-gag)Y1b~seTvT?Hn8&&TxOASZu^-!-Y$i0=r$90xW~}(SG?^ttUeB`dm=n zXBvU;DN%1#TE2-`*QCDDPx}@qbPr6ETpsoZlX6la6VgqdSCknM|l*& z&T$_)vKZz?S9Z0n<#WZsEMt~0f>@LJt1oZEi%>6Uk z2C76$+2c5ImAHnJ5vQGWG|)D8#Ysw(1kY&y1^M zwTS7##KED%J-3awH13ge<5pyRmKX1JqKd0|oq8thi##R|X0IUNo}b&3j5u?a!zeZE zA!HEoPi>k77m+;%bl0&GotOYo#5PZ)_`~3UzHxez>N4JmOcpKle6Vbn1CtH|7L9Rn zfzX81%^F00qsyBhmO>XUMk5S(`t;rGvA;wrryP@!xhYN) z$FwLqDTxaTomEmQLQ6e|o*XxKKvXmzPbCQnVj#MX7!h3SDa#D=1g(#9!~IiRKrk%XypbKoKhVjRmYyCj5_sn4?`z&p7y=%<99q8mC9u#6&4+2w zBFT1h{5ynZZ4y;=)5)nFTeAdW3xXzT_Qh=YxR8;Ee-Fi7`8|LQSI%so5OQYc+c);& zS;}MR0I=O=$O(5MXy27}x_Wx$5I_k6AVWlo{HnoIW2t0>1y744xbO|lH`mF^uJo=k z(@*pE`jSoxT)-`X$hV(MN{sPVDDd}+?{QN|oEUB44sOl+QX;ZIii}y%&2nIZq9l!y zxqr$K1_MYyXSzV5Oc;fN3@zDQz&aJvj5HKbhh^0pV;mffN zzIASNV)kjyf3;4!#zoc;<33o0*8iKEO#XQ7&L$Ixk!v}VAHUP)1l(`IE8*3E+e`#f*2%JLhB zf6=(m`N1L}T45#S$eAo3t!6*j$ZEdwOUKHVLRqEXc;omB!PJnKFB!IlsXdBhy;_m& zWS>%Bz+67piQ0IzCjO=NbL3NC%lLwW`ku&*?#Xs(?4svJ8#i0uEu_g?-dJ)12LR=9tNu^S;>1iJ~$MnKy*81Uy!uBd^|;aPQr z2{vv#uOh(0C==ysb^MQ?_wxo&tb{jH&&M*}4bB8{m3CW64}E-N9GYFsZyM7 ze)wk8e6IU3hws}PjyTcBAN|ymZ87EawFG+)0)$i7VEMb!^%I$LVc_O$;KY>z{@&i@ z>Fv6ZjacYcqO8Hrp8Tl^$O?EgI{n7a z+o>23msci5!+UhxErEl2>GbqTZs`-RJRo?JO;T4jJ#qf;bA#*wTEErUxMj=Q z=ZdS4wvDS-3_iFPxm=ZnS3yB^Ku~zfrr8XJSg{yqx%io zb+j7o~0ty8Z2YOo}_9rE4>(K+iUg~o|%I#1EmbWm58qn=8051gX zy>v?1o9u(od68c|mtvysn+XSi zq_&}6$B;uoNE|7`444Gg3`rPLa078T$SsuH;zlh@P~5>cW&cF~enU`bRLh$yms)B9 z8Myb_p4}4LTgRd#zMtHh5OQl+PlK*Zr@OhmDJf$8ve=~Cpw5;N!S~dnf$qwc?ju|~ z0K!n0iU3SU5YxX}siUREgUKTxI%%v%wNBIO*jvw@oqy%ekd>1%LgFwW;sg!%xD@ms zWKwR5T{7L$#y8v>oA$WY@bl+rh2xo25#eo6Ja`44LIDZb%69K7alJh;5F)v*LH9XP;BY!NiR8<>$Jvl4N>7)(T!O9Fzbq78%YN(d2x z47{)#{Euwmpg{h;2YzCV!MPijmS4v)wbsN&5KiZhA6`0#Tqi_c4#z2Z(71>zx%>L; z{v95%mYsvu08dvrsq32IEqV=&nL2s%4Vo19iFp@^WDpyY3&%KV586L!TimEL9fcga z0e07hxUD-R2rRAT09|HgSN}9rjY8y|_n21i+NG1a%U}*8rYVQ5-3P8WPRiuThEWt1 z-R&QE{9ub_nI1cxeXtdeOu_A%)-}x}6P|5{v+T_~YBwl3rrdTPI>71OR7}Pj)^MZW zB2QYhe6$Kc+-JNyImM!!fD4ki1-+r>q4@`#COVhk6b~}8H9u8v0cz@+^H*N>iO`=v z#p4mQ1a^$^U$ti3@Tw>0@?jy)Uv8zc3X1fjrx^qTNG;gw4ipydJTpenfqM;?9Yk(_ zgh{TMNdfp06BezNBraj3Gi+N2eUZ#~Va{IZ)`0)NYf!=M+0fnAf$~A|{!0Zk+4a7j zDfeEs-VWb@@j+hR6`CI}uO*H7x#GLo42Xf`zNRc7RCr@*K#N2ya3aFPuluh!*Voo^ zBZU=l1QIyqg6aDOyU)mI6e39L4XwF*!2`SX5cxzCkDt7Hpl-VH_@l77_&dJWb;1po z4z7A||I~(f0>Zx_xOo>k3Ry0Q!JP{%DJh8nSt7%9m`@5z^}TavMW)^0eQdpUc*9kh z1~9B3YSSPcg=u8fz)!%r1c)5D7 zw?7?x;LoqrQYl@!Z<(6>f!$M{Xm~u1Vd)@>EI>gqh79)i;c<Zwg0moUGpsh6c>uJ?-$NCKa+m+Jo=-O_1wWIvAi00W#yCGm}EwB0@Yrl z3>cc4`XF0gTU*;>!f?>Px+NsxmFfOd zee%a?Z~UF#`EqjOpUXW}#<#0GSOapf-NKE&{rPjaCL+vA^?H%b&b~)b@GIG}f){{$Mfw=MycE`xx9i<; zI>56t-{;|I+uNT91NVK481Lp|uy;A`n2^QAZ2hm~8g*>oaP}7e1pVJy7LTvp+Pc7B zP+C+<9@9U-h)AEErG;!vwI3(#y{-@bz4MrIwS9&u#xDL^ZC=#p|2g-pX(yk=3ZZJ@ z6V<01#Z+Qmsc&&LShMGX+CM@X+7-TOv*I>|MPIt>F-JpEFA@DGVH{E96cLa}d;BdmpP0?$X*RW_Y>honhJB zLIcJ>iu`w;J!2#BR**iGgayF)6&Qj`x9(m?j>4!(Ey~AAKxZc-SINoyQ!*hxMbC~5 zP9b@NdO~I`5Zw^P0THrta(ze6T)6PsL_$vrAJnmA816nyF>>6@uvv=e-SryKeF~A* z@EMqvJ>gk1*wAXL^HNkd&6~dBN0jNI)GP0QM&H?Mcw2*&HayEht8kUv@D?p{xZM*r zwACJ|WU01CZS)u2sHrz7SY2p?uqo4*7i35a`99Z#XQj$cL zNOT9&LpAkTyOAl!oYl>yCPq}HB$qimbjY*nsp!cLkPq;2*$0OkGFE_?Y~$x|gM!4qj&XG?ZEYMW>YM8Oy!TvW?@5DFNR)QQ(h5 zLHLCPE;94&E%|p$%y1G50b^jVg(W4lB%T*@2Kp;Gq6G5UKa+S=6iDfw(2`&+V|3CB z0kCKnhNBn1Fp|s(9yItKEe<&3cggAPboFKCc zpzqr1ZkWd;DO7|vZo_sY6Vj-W_EYM>un>VFW{(0FfgsBqJLX6`EQkeXQGZ)jG3!5E7e5WJI2>HYdN7ccP%a@ILHJh4CzD^3SNB$r7}*34Qm$oSH)}}zzt-CRrn@yrT}JG_O}s@55<~E2Y*>6CSPg`r00?+@b7Z#RVtzL z@8<_)DfD6P-k)g>uPrzw@0hc09D~_N;vO3eu7Z(iL{8E|K#|0cfI*C56Onhu)8Xag zBX(0Vb`o))kdJ)$=+SM+u@M+|IpjKj2)r$4VQTpsz%|re=A7X&;^DSHqhyrnaQuY% z^Yl-j_%M8en|22a3!gY*w%7@9h^Tzz3SXJf1BO$1QoWFxi2Yl9^1u@4@W$rbrpF%5+ zz~dqI8$u$YSj;a^vgo;xlrC}=9F#p~b^DQi+JMPEX<#vMDsPnB6%I!i-a#sxhN^xJ zp{uIqPJrV)jD{s*mY1v@@CIo}Z{&nAxOXRQ=OIjU;;a#wbLSW_CKs=par^aa?vyu= z$Lvi_`|YQ`B89aFN&9)|t5@ricM@9;V+c_{QxF9r6$Pi?-NJ--$RSIaD*FskWZs|@ z690q5#()M0Bu5Z*Epi*=7`TVXufUWbhtpwvhm1;FWxEMFTA%qYoyx0oCEFd5L`YI! zCFBdb?p4a(7twK!t~prj*f@lv#9*w`Js+6VnQ72ZbDJNPuZ-8@0Bap?v~|KH55iHA z)h-?H*59wN4UT%P>&SB7n)WY`kg8MmbD6uBSyx4j-Mh4woh>_exmn9%5@mUHz{Aqs zadEV*C}tLS-53$?o83{Zp*T#Y0WwcmW_3OBWW)PIr9Hhaz{t5b*#K%BU=uEnjw5Vf%C zUHUHFbVf1{JUcbWu&h4H$jBsuR1@bZ#ttXqMfZ);Q268?9K(|(2j?LDL(T{xC)9yv zSyk=8)aGrR_(0Bw#M}U|Y6J=5>bvz|RE8WLrLOumS!u=ZPKfxZi&7&<{Npt+9dM}! zbc&9izJR1?;AVwkQ^qX7;gYPG+tL0o?u%UN7oJ<&^>-xFc+_Q{TzMwo)8(ShOc!|G zUC7Y^f5mYMb%T!WhY*T4VbSX$ zpoYz!H+%UVnQ@d36Kgn%10Qm1n4>3&MvU;2I3^{bX4Da>2EsCtv5s~nMWX>{ZFEZ_ zh~WjtFCRlnDTK0+;~`Vule3XHW*f}UW8)#gRyBShF_ zLON{DoC1134wnvRB)CKCx{&8iHQdmpRBF*YKFc#bD7nHNWsMx1fneKRjMbQ!c%@G2 zxrfZ5FVd1>BF|GQM7*J;nex6n8Qo zjc$Ti>!oL7~#4-oHIwp{0cE~bpu6TcQ{l)1c-hPml$K0WzN=QJBR08G?c0I z#DG35FSSD2D~Fw`XOr6C%eMPVYvyTE{WzJA-z}8qU{dCRz9)mdMn%Cy5g$285NUTZ zz;;eRe?6^chShc$1aDFsgSljz-klRTm+R^R?_N4$=tt4@xY)`_jV`-SX>9(13fp2d zckkL|Q~dUb9Hy5^C-oJ#xBttTwvvRLjez4YEPW``rk5J&Lwoaa{?2>1p%wR~y!>77 zGKG-{ocXp6{zV}e8p+6FO_HgI&6_uC{|$S;-h(k_aJ+ADy@SkEKh2wSobENvT_V`! z2F{}^0ax~5QR=#BCS&3V_l(p9LgQe8$|Ku^bJ5mINJx;`QVe1IFSg!1p6a!Y8sE|= zluF8&sSGK}kXeR`RES81M1v_qL?kMtOc5GL2$_dUq70>EsLYhPWX_mbc-O6SI?wO- zzWa~o^PF?Gz4!O~z3=P3u63=ou43G#5zr~Vfg!92dIkc{3IrDaUQe5}^2Zu{q$dq^ zf&Yis_6|*WK%|F8yroD>jdW!|YSdA@nh?c!qyNR)B{+MN+Ag)sKf3n%JazhP`zN#L z{9E72oqxj4TI7G5yQDJH#ryP&%}K6-%6S1p`TN!~@}N=5B@zXrkdx>`5E2zyBZM?5 z<2+13#MYkG+_5tG5Kb%G$w5Ox1hP77X?dc(DK8PcP=Kyrk>2}k3Mw!)PPMu{-Y+D2 zZ0@MV5&RqFSJI?t2u?(jh!8Hdk<+b@rD)ioPv;9QQ8dnD($YDxQZL+Td@nLBLTv;6 z*0RM&GI)#htozY4^mRxH4#r1P7AUcz;BwD={zdfrI8Z$SuQPxjUYS$)bO&N6QBe}k zFT5CV1|`&c1s&ia8gnAc#%3bstT-cA49{G@WM&p?ArDR(kcZs9Nnl%CFttDm#_TrU zCCUnz*N zLfW`wGoaGL#dtbJ-pI?7b(#2ilZb*yUl2*Yeai}RbdX~TTtAu1Q)}@R*TH6i!Q@)# zw}8nK7cfFgMjGmkYpJ{O3_t%QcsUS^?gUHRA9BYn@NLQ(!oD#@_}nDpEKJ@EA4fu0 z1e~HP6f6%e(r>T?$;~7C5qYA4+Z49|A8%)_EXpe&l-C=qg}@*P=SvrPwmWnDosG|{ ziA}kKAG9PBX_Lm-R|Sfw^WOe8qPbuBy}pHkfRU`bPLF(|T`p5bZ9}Z;9ncWeNS7an zVz9T;PUid5OlG2#y%&m(@E04inoFk5lA)_ zHUWXt@Jqt_FI-|m2Z9(Aft{`helbBKanuWcu(q#R0j!V?onb#>zY8A_IN<8~uSFl6 zi2RB53y$X~oIex~&?HU45hrRZB9+5)KZz95!s%&QDQR*C1w~J;T!pX(VJRJKsP^I# z3B7tzxQJ5$f>^>sAuTteuP2sS+x4Fx$jTT)Ok0cfjW9^O9dNiPVHMO(aBSG%iGN4F zldv@jAPD}Avy00TXW5uALaKxL1wgScLj`7RNLfO5$WNFU4;|a>= zQGDcTxV}^I)W5v0sD^MsYsj1XUX)?vQEK`%{h!_i!8zpuEBpmGt7jYJP`aEfI*!Pr|BIOO}In?Ou`IDK)h6tE9Pxszj#o5?MH1Z`OI}YZOo4t zQ7QO)j`Fp*>RsKM?lbnDbT&?)^j#LL@m=C;h%PZuKp%kc2TD3eW9auMgND`qYA9a` ziRdn@l^FUPJ_!f>H|@Fx^haVPXGVA9MrCg!#ajW6O%JP6zjetjNp+FlEWRfud(jCf zWPjPKfz8j*X#qaH(Q%Wq@;(!U>3s{Q8Pl804m3m>zYpz@9uW~xvdn{Z)3GxvM46j5 z?R)SmZgS;^1p-{*m3MYs^}3A7DT~)r5*umQj;`o0*7Y*Xrb^H&eYw71>;#N8bBPE;P-@%r#4rg>Q(+Upf! zWxl_TD2@!3&N**>n3`pjChQd$v%4=KVTGU|^@{zm)TnZUG{O8^Hars`qpU2cX|ye* zflh!Kh#{f9#ilbQ?zHXETWh4u(-R>z^=PUwFHWDUL&hL3?btz&#;NeGZy)=b(flVIU$>c$wY4FS*ZBF!HCW3@5b~_C z>cDQJ*#jI^=`nVD%*573@^VWbbPZpAlTNd&V0DttZm|Ay2u#VyqLuhuqcs+!Xn0<| zZEk+8oR=u1{x1txaNhnxOIO=*!dA?z=yAK7T2dTvG&$&^{F0li8k@9L1^ma7xZ4|a z&VOn5u_s;k#aGO_pFi@roiC2?8Rb)+x4e9+9VXeleJ1)MhfRpU>8U9n`|HN?|8PV@ zHG))w_FQxp)B0W$J+M4;9cA2DOzGmUiG&l7Q!{k_H>84BBaCacv{D%NZV?*0Onm#H zv3*uv3o%Men}=}A{Gn0%vmn)?9!pGcjh30ed%xuC*9YS-?uTV$aZqJ%Oye_b93Ell zy$;OnpSRxoFcN6ncR<*P*R1!R6yM&|v5U^fGbI~eh+V#{JVS^^i?4*Gr6r#WC#VnJ zX85(vZYP#Pm^MQdzvwCOiS`fdPB}R! zwUJ)TTC*VkTL*X9YLZVbz63Z2KQzCYHg(^^X|zT5(6d(M1MSz14hWrC(UN<0b@tKB zzfiK~s@q|q-Va?ja?W)MzWZQlb96w5?p?@GV4C4otvFg{cFp|7t&fyWS$oM(c?QoZ zA9n^8YK3$4(odQcwa%=w5EH!-82F}$yNrpnnHFE#A6qV8UZwll>O(n`!kPFpBpO7> zP>y$&N4vQeeb^sj%=-7Oe!ORSOe^x^4ycwWXa!G**HtrfrCCXg{akhLKDhzch*=RD8@JNJU_>)a$m&TKQJN6mgXO>vgR&FvEQ`kjeoP) z*(qFnaRwN~a*6O=87V)nZ4{#A;LzN=Xop5_R?I2BT|!i|6FXGwFK^V+dY(N}YUOX1 zUf=U|{cSZm7QFSpr*)4(@npkQj@!Rwn0FeTv+20sb@lnvM^SDoRC3vOK%cw=@mv4&=y741yGFn3Y^R@}TfNSR?{;Kb z-l5Y~tB!XvM_cMDaP6|Xtn795?*=%f$hqN5=2_0*7~?H+dtF)5pFce?!tQmxg(4_e zRJ>nXkI-}etgL%`E}UM&b>53>>qd{tlkOo?Wqj9k7ZFro6l$4V;`=zfcP zA?029o~tS91J(lfE-!tiCRUmD(8QEmz>4Um7hh&So^Ck)C3n)5=RFT6zp88a(_Wq1 zeQX|)vQ)cwlW+gB=ihZ*Qo<&;u#IcTV6RomgiF8rYQ0?dfPEbIq8-eOiuakV{M+u> zoayt$b&s1B^(oV38?dL=Q96HIKY27&M*q36v$OK@Hwg28Ui$=O1=*x0b|{Yi+7KIm z&gLj*o_KZn?BlaX3OyfRXLCBmh+@~0%D=ND8z4!KUEnn?G-x6J2@{zTFzdOO_JqV2#f{Wd@_aV_7aEk=-?bF=au;xm z3$Q5SiF4PiN#hWQfX&slESedGFlc zsY99!UpR8vt0Zi=IQP2dDedFe%ej2(uv;oHqAy>3{ODjc`_D7XI8l+-UOkfP<6kdl zqamQK@%|mh=~wm2)1%Aj14>r@-MT?kd1cOCva?~|J8QJfPMhN=#?dP)xnCZ>N-pKt zf6Lrw?Go33G=;H(ZwC(7w-(EOq8g6$8Fc%&rQ3a$BgTag3wbwQgbbD-WDyY&1m{FI|2t&t7S`~#yb0Ck8>r1dMWt)+iFklc;zJ=1))*STkIPZxgZK&zc^qIYJs>M8*X&iZUz)iG?i+C!KJ>PXyXq zLC*@5am8AYN%2Dy`(S)aB*?E_#%b^u?;vE)aO^(@C)k`}s0hg`Vd=pG{vTRp53Xej zT}9TNMeiq*x)|zz)?EOUL&1;M-02PIx zNW}08pcI)b0FmoP!1ixZ`+|gcZt5`F5qJqud9M9(=_LOtj(yx>I(EB24XM&eN+mkGS$Gji&d+2^@-y0v1~^RKFlPzkrp`pv;xRAu0||Miz{y?c!ED;L8= z=@psAF?)~2ohm?ouTW$ip{bJ69vW-^T4Eb&q3FYSTI?bo;edQZh!FWz8p7P zBd`n6iJ|Y`W9qvGI>?k9kR}Pr9I(E;flVX+(E8dGbFAW9tBG0(r7tmdCC?AF zqEwS=E3=Y-XC4cOtq>+Yd~^U`1R|3jw#Rx|5d&H zxj);VH0UCEcpJNx#;Q$>qbhKHRDY|Uk@;!P)Y)B9j$yWaTY~7ZudfnPEb_U6k{cmgO5S%cCntv!U1j2H#p1=t#^#T0k<^?opi;l3^zBByLcEeTI-+ul2J4>c-Zw)#?V`V(F8kOo_#a+R*zTFd&7;q8V zaaR0WTmRiCOI4F{D@iK@`TgC2+X48I`?NkkeD@NL%fnQJ9|Sz8?g&Jy!Y2zU5Z6S5 z0)^fxA#-UrnZZ4cpwB$!I9a|$1Qnx9Y=El_q>!oWcJJ;VN!o2k&k zApQQ-kLXjODoGk8^KihhBPbK#V-HbbiRrsooI>G25k|ByU5{rQqeM6f7g70Hl z$KkPrjwcZVk>(^Uv{%y8!+S#~i+Gm9Ei*m$gbCmrqVm<247{uisJ+_os#j}ni7ZLxxkHX$6edA8|!oE}2CS2BYUQM3c;4qSn z6AOo@z~JA!rtC>=i1P2bigX{}s$Di~J)Wdc_c>ZYVL`WPljf@bN*TF`taELL{YJdw$`6qLHi@|TIR#F6hbyY3j=*qt`8Gv(-1Ra6La%b#2*s| zD3A*-zZK?_@FU00MEPEP0-67y8>15$3YA0?&0B(1kdbg5#|6yfq5uQB1w=eR_hrfFb zipO1TOlAN=p;XWTOzz9KqwMXE&=%i7(mNvqhLmmD*fa#|SK3KieFwY;mhE6j*{n z#ls2n5*UHJ$108ewFgPgfg=lTPoAN#Mj`EJ#psd-Nq9!AOtw zo)>I15cYW34~LGrm$|nH-H5U@-l?ZAP?%P$eLHZs+_8glf3p2slBFk%`&V;{mW7bX z00dLTo`j9261?xp=p52<0M)QQ_Y(kgVz=y1U`718PQi;_idR9>TgiFy)Z0?3oHu_o z#6O=!-27?voAEbeq|8B;ixd6*shCW$Yu$m9J1wHso?Xm%=DW|qL35wwYnoBJu#gaK z1EfM%4idchUrP3W#mn2zZgZORYdXTsxy9?IcHUuuP(^x6 z?UY^AdGb$hB$`*f(_`K6qvIoB8IKq)Qvw{4GW7JBGu}vWI?(AN3Y?_h*TD9F|@LOaIjb1PDL5rKQ6|` z9vJS@)1{(1ee=fLp0Uk6ON8zob~(6L_Vd&pq3J26t%zy@#n7C4Vp*G@mq}Wygph$S zW&%T2icw%Bvy5{_k{f!785pm&-8i3h4ZBPs=CYBDR@CatTBJHEj?7r;*SKWq!X+w{ zKCW)*7m_U1FH8^QXl|z6w5iI~;}0-%YZq1=@qGn{$j8r58KNh$w$U!9Cd8GIo+@sg zSxzPIE&Tke0FjdR6D&Y(kj@7TfK(i5@GZqvLI-WgjUGllQN`XtsOU*dObVaNE?F8Y! zfjCyB*mqM(Qxfb?`{^|5i`m7cPOmQflJI+9aq$E z5z=vD(S5?XNi=FZj)wOtV{(7llsyYKF%s*9 z2A9e0)|D@PeEuz6xqNE1t>#VHV~osttxQd-HeFxg=!G#x)ZC;lTg(9%7`}b4f5b2| zw#pM}a8To8o&KE~aWuNt+XQI-eM^6*p+olwk_y+9;}ElO0P`w zzt8f!(ft;VJLb&{d0zkiZkL@f0`Rle4dmPnj?G(TxRl4o+<6pdf*(%z->=%wz;d1L z)!lEWgB8!RScdP|B%Z;peh4QggZZC*k1NbGe)`PLVQz_m@6Cp<%3PC***hL<=05-S zmW!+9-%~hdQms~AI-{d?nb}$Mbln4=G_v|cFISc|XIy0nI{e@FanG78y|S=v)6LS{ zW7aED*k(?BDNF*8hA-^=d(U0uW1nS^=q~7S!X^WdKj&l2JiVd92(ksFq#9aPr*YmaN4Q{+<U+j8j?YoQmL1AsJ2YJ_jzv_nK)!SLCZF$8HpS@${E7DVQX}j9puA~9w_r#rG z@uhuA+56|4iKesBmAj?{PFURP)7jQhbS#dR1&?s`-*<7D$+}*P$`cHVNHVRvNU5in z6}HeYyf<7UteN@t$@V`Rhtd7`@h{ouMs9ar(yi}BzEvg{ME)|$~ zJ8AW{bw$G#2ip9y;Nwx6k9Q6~ayjEo=jTJS^zZkjoNj(~J&K#IjPha=O9}+%)VB5V1%)&dk1UCG;Cp1U6mcA9={^4fNeyMP?2h!2HcUAe0 zKHzY@skTJ!&YzHsw`SHJ#!ZQMcW#ctMm3s2+k=M$+-emT0%r)ZH|Ko?*0iNI(dX?)Kgal!LuM{m8*LA5Y zzo(>|Aek&IE?y3Hi6`&Sufxi*h8M*HmI>|EyX^OQ>-~-Fn_ACeVu0UJo-jiv8}1md zx|P*-lpp~I@3JebWSd;KL-CJCLWJseXko#h6KF6Nz^4*l?RP|k3c?eUy%|o*@?Ql` zCcj#qRgQ~G9JKJ38ZUl)N$JGm6#>7AS+^B>9#z%*x68MoKa`wMqD)1bV5<(?X=k)ACS`4 zlR42y&bguE2dM)iEFkD6#CD=XB*Z9Wtsa1|z9BDr4-DznmQ!3@+;<`a-1hno+t|l^ z-nuzXz@)DoY}iv1w-R{12VW#x%IhaagYE@WBx`Fr1uhB{G&Tp7k zv`IuAo*v`p`j`K7aZx(!=6#>BF)Px~B2TvBS$-THN`LLDBnXn13bWi0axV8fj53u8y#KJ_7{{p8FHi9mGptr248V%_tNvv-eRG{gX@697koH z=`>_SytbhJv~a}#=)HE{MU0(@Ayb)6>+4lJcXs4%UcnHwch%KbEcNXQKV=^ts{X__ zta-O|c)NAdGdhM)+K-wrQet3yO4`!k;6T_&Lwx>W3`s$LONP(^yJBEuT!uk|A0Nqc zP>A&kxHBD~Fz)zXa?|J5;hAZTl-%DEPaT^bG)%aZE+l#f zIh%4F1>sB?t-;q0J0CDxwg9?8owa*oCfS(m$n^`DiCNStKqO7V4{SRM;SUTPo_PTJ2#7|Emoltl*nDs8%YyT= z6UA*+UwP`@vb(*yvgAmE)^b`uAHRPRV=r}}jZiRS5SY$Z&&qR8QvG}qc+{kwsFHX7 z+%Nvzcyhy55og_B`zJTbEf4sgzeCq6CP8Tn*4)jTUs%UjA~{vh-r3%;eCm|#nge?- z-1eF0_}kbhTD#((uyB)y^+aNow|Vl4ah(uOAXdFlgM z4QEw94V%nzCVU$(`BDhQ(m(i&$Mnoh*OW9v)hti{aZz!rIlA=NUa6)%Q_l-;PgpPh zDhY#k5IvI@4Kk!TU-$Py=CPl*9-oLC`8r&6Th)=a0@m;A^_FqT%CaHcK{$mivpEsf z-AS-}52&kahsMRne@9^22Op3{A&WuI)0NyjTmr0XUi`kES8=2#r6RZ#a~K}CPrLtY zadOtKYb^eDwjy?tUCNWIdas3Y(gudu#GW_T`NWkww}f$z_EN3>vX=3^y>gW)6{F_f zqyr3zTH3J(*6?z2dakX$^RS(>eJ` zP{jMu`Z9MnsorQd3k#F@EVExM(mO)v@oZph|s?`NnmSjDy?lV4_B$vKP$W)g>4!dayE2&y%Ku((lLTNH z6Bi(qz`)R0!S7TG_shOGDRb;})xESYpWMSchZ_?2m%sQ-w>(PmF&-FxmfFFC&p6<3 zqVH|8p>ZSJl@wqImgcj9LTQFy%lLD&re;TQFQW-HGxbEnl|kV0G;7v4smM_Hq$MRE z%$E5|otHJVy174VAG7pjo}(+MLTKnx)>&ZMG4cFIcdoOm%K{%)AngGR}B~aMDrtEZf_uqE2Q;=wkBhN|%HE)OdD$3oh*0VM_~Mc4wu zy`wK0dMBEiw(h3llU<{`B7tx4We5F0%PU=+sapH?c@O4vvqQuESuP(piw(*pXzeZ< zf-Kt}22*S?r$K{XH+_4HWbLK@emf}Y~4aJNE&l+``zX6P*miW&`9!E z#mQEg=9|$J?X992Y9+$7vhdt>LEn%q0L{VUcr>Gv-N*Fs(R$caMZXXkFqX#Fo|(eO z9iKRE^E%o8>+CI|(qD8{)K2nTHfwfnTTZQUzp_w(8-jq*(eLiX6ugqcLh4w=y0hPl z_#VI0awap$bx6}fACC6eT-vyy)#%8M17)^=E~>1g1hWyAJrenefpKKK`}v&Cjbv@U z=cl502*L)-*xRHBRKKIRGLTb^W4bqcM}tIs_kt8wNLQ(u+t?fRIOgN;>zCH~)>&Vl zk>17VQRcJ!*s8Bq;&UAw3zcNtq3Y=PxDGmaZZe{x+7KRTU%8SGwq$$KJ#1mRFIF!U zn~E03U+#(Tb^c~bu{}6FtuHnwWWB2cS4-zpJ@u-%tjNOglrSZf0gLZ}oLes8G?Tqo zAv+migc-B4b7M6VBhX(t_ExK40Im?-cHD`Va4#|(-b4|&?wNkK-^fmb|#=T(?r!3!wk5vx0wnW^cQ8-=9rpF7xnjDnnZcHhw)~wOBw|{zP-5WL@ zUfy2xn$iL^OTYpTy?eL6ZxcqYz=^HOI=dwvgG+LxXeb7`zb^~9bRHUb8o~F1XWR1t zc6P_??O_G*9i{N@MUaB4@xlvDJLudH2KQ`RYp{c#U$w}FBkhOiq=i1(k^&!h>c5CE zo1PYPaAc);#G(lY6?%@;a!SZe>PI`P-oA5u|GB$#e5}S_qB&XLWk|x7 zM?EFq3_fukM2QYXbP_$O*{uSiFW|8NKvfN$FwCJy@m=>;=W>x-C&$20-=>S^iRMo` zPYaGuJ_$9qdt?x7_Uv?jN>BnSRk3rGg>`n6Li6~;50k*SHV5zc0j6@u$V@TPDl7|m zl%5{{sgl!jWcS*SCEhM(@2q(nFELWouD3<1!R>i3Ikm=`FH|JrZ$~G0o}*P(QQ?R1 zD&D;B=C();^tcQoHt0MQYTH`$yywxzBUxiB?HnT%zyH!2Q#&63Yuf%kK1#bbgm|wZ z%1V0@>7q9D`d0vovN#N5;IbK}EjsydJK|Ks8Yat2eftvcwQ8r`eKa`VO--|UrNen) zip~6cMxkhYhElVC z@r~m7*~;=~erx}%eIsMz-Xh;;JUV&tk-tAhTkgHpg@Bka z6+HQ{_O;l8PQJ;7J?ELdr7jli9bgu-vlpT0&^et9y>0%$1#w;!^8tCcF%|TC^K^Dg zM=}{j;OX@u-8c%B7q2Q>+gHgM55h$vwyeuHFQzfH0WBeJ%KiDYeC*@yiUdq?YeZeR zgnISH1FRC~BQVS-3MvRb85x5$vSYW&q#>AZmsn(}1*4CbhO@+?G?+Iy77_LYqz9D{ zzWmI=K*V%N)l=c}h?xk?lJ>FcHMJ0Dsg2`uJYt}*dJgpn8w_%xeWABJpErJN+mj`f z59;BGM~oxmc>-(R*H1O%8#uKHH($bOa~!A5Ik@1HU%Ua%NH`1)Aqn33OeiBA5^lv0 zNO)3O;W=UrB^G*hb%N=yW8iUa0Xb$c4YL6hH8U`#bU>`-gV!kz#%c_Tl1J5gFc_}% z9Em8_XCQ8_g=!in??I?ps~}sf%{1MwTJ63tYfPRsHjqX0%S_0awF7!Uh(rH=e&p_}ob}Z+35jftLkK*z=zJnMIgGtlhb|UYnU#%^sum*R1}r%I zRU*DbA0F}8=I6!%by**WqHiS;90d`xn);h|9t0N zgoFexV+jF)B#lqT1&7a_OTZM1pJD-I zCIi~MJM86fE}^(_pG-Br{prk6Fg(<2Lf(HjI`{$ui4rx{xu<7m%0KjV^zFGg|7L#F zC$mIX{Isigj?`h!eMT8OD#l5EfDGP{?ZfAX?w^|(8&bwzo{7^9Yk<3as-Ko&gs#yg z719F6V}Q_etWs%BBA(0)tcX*w;M&M?6$Mp;oWwgmt)ghXNxROG|2yLFiJH^=XTJU@ z$@O>a-x5#1#Fm{|kGB@AChVry%|-=e{p7p7g4cH{nUe*bk$#>_mRqwhjeDQn;NYXx z)0X~^XM7W2!J<;?9KT4Zk6}QR1x%n~TPUFM8gU6EVOqA+$ny0%z8rpqQ3&avQ zed*KX%LcQ1i=go*1A9v=Dk`oFR#={a{MY~`L3E@`A3w%}evZ%D>EL-mdSYX z7dn#sgyh|QB=26BYg_2W#Eu3m3CH)0x*^m3V0(OJva^9bow!YF3T8Fwkcn};jLmF1 zkjuKdojMgWcRmp^B>rvN_9Iez4ZekQL~W?9k$JkL-FM}n^l)^rdmK?b>765!2tig- z8T>f28~gE+E7QZt^O<^OKMp-G*>rkB|8%J^?fv+~3+_3qVF6GRViivtPBzO|^;s4z zr{9yk0oW{l|7@0IY!=H%*drov&Ss{k+XOxIIPt+N z!9~i~Mrh4ZnRE1~Rj<@^ZoPhUFU55mT4a^S=y7tNV7eaUW96 zn`utE$2|U!g;pVSKQ3>-8C3oivYc&$oCGkFmatFgd{uF`pjo zw5W;q=OK575CjmIcB)#yw-$!?Sy0x{LG%~AfrBF*&`*ySlhJJ+L`?yNTM*;4AdTix zCFuaRQeG+9v}Yf=9m6=!i8D|amjA$jca$AQqmxT13OYJq_+d&#Qy`*SG7c-oBuRK9 zs93N4h104(Kzgw9p;Ie=;>lcMCo+y%Z5Fr(Dt&iR&f!$($4N`GTJWBQ5$Z-QcSA~O z;Qc$vYtH0Hi5?s&_fW32NOh-W?|P?A(eOZ)4<=eaRATjH5E5>l>iZ-8g4Wdi5TEH5 zT$$5`Un91>IUp?@rJ<-8xr&5|x$&X+AWb$j zKh8{aG+n?Ctpg8cHq8@+$||u2gU2|t4#F|%1y3&yoqFgmYccOaRSQ!$YEceChNQSN ze<}ufTpVl@^qg88nWWb)q1dj%=#+Fs_Qt?;JhQx+zGQOfldQoUr`8N^&Vc**yV{2j z2S7j-^MTzprE(erJiatk{Z6*YpWcTz)vVt6I*M4V{F9O(lRi@3(2zaW#_FMFTIILq z`EFO6k!5mMnOzr5pX#T4UcHRs9)&CyteOVQ!A-&O*XPZ0UIS<6&c35s`++7zBZuLG z0uq;Q0eLN{s*#r+3}$y5(*%>k0xmden34p8uq=?ffD$ws6VA!#@NgxRcOjSq2DXqs z62P!Jm-#h!A;#20PMtMG2GlC#&Cm2t*d^B`;eH)6E%wvI>hJ9DPevKiA9vSC;Lgx` zibqmkF^f$tAg459MUUYQ4}?vwp=y3YasW)76ghMhKW;+$$-zD_8OkauoL+uCXvdO4 zd^~^(g$4!&VxU7hLO=PjQjUEgt*%3|H!2d7r0HJ?rXBCK`!sLV+|{B#^{ZL&=!y!V zJ0@izw7zLM6Nl>pZhe>9c>U3{>ST3k`p3V?IgS+C0LzvfzU3))j-PX+b8ne$@+K1n zRPStMJkin7@fcd?bKpn3uSe&h_i>F}3eZS~U8abmrwi<7wH zfuWv}OW2uDm-Wwn+TI&D$E*I(Mqrc3hT~6CbF+uqk4dY(V!85JEnQ?_kKZT>3P{mv zk=u#b7$IRx;z!XNC^K5Zo3~F{*)!#sWzehC`PW~|jdjwE46$pTOQgqMi(8H)55iz1 z!jHNB_tx$B`Kz>s@ht1z_xfXvda1{b&%GThnYwvvzQ$XV@78~J&YwRTF~9=ZNLm20 z-jm%pvdO2Y-a{D|wsgiT*6kOk!srjPYb#Eq^vn_xMHDebXInwyVf~M%riM z!bYLk3X;{1j5Zlq+!0tGS6^Qvhj~kTNWHH0rsb%6>&Y`1yAGVG0q9^FevkzyJ1#$h z_QQjlZ=>CcqPsXd7&C4=>FrKK%u+tqMXl+3%8?~t!n+cs8U0@fe__mDqR4HLg zUvw^iZL-Pb`^{|zx7SiA+A0z)4tFq?ZM^pA5q|6~U_uSBPtgYDT2%`r5~_utlYitb+IrLN7oZmTHDJ*%#j|rL@1*dxr;6sDdShSc1Yy zQ+Gd=hDwUFdD@FFFPNmyyUd;&Ygq_Oe-ijhkxF5e?TUJpJ@!9*0yyN;0O{T602JbA z&xNn~OoYB}80rZl+y{zEO1!E#i>eWGVXv@5wHh^D3Wfk>|#>rdG9f+{xA z70N@7X=$fHZPW{)uq;OLlmOt490~^aHYl79jl7pOJ(QPHW7PY;URlg4`owZNr&9tH zK1sRnf@CbWIYtIZxlW`E7A#yr_5dz&A5Mv5=qH_i5D!oKbU0pA*J)y3hEHSl)xwQo z)(*!iSaPx9=W|ubz&a4MVi9h@q&p78FzTs@b$poYugHtej0<~TGi*wj9lWF$dF;rx zt<)aD1j8U86iSZgr^`2zV zbjVpr1xUv(qlemT;zDTU1Cgrd2KCr_SK z*Z4kaq^2-&n>VNm%}*SPyL4vbiKCOZ0+?%0Mh7~NHiAu% zS|BP7@l;*{yjbrB2L1K@kdk}J@lksm488j>eJIlPpzH6B zTd5;{x2o@*Sl~^5-}lapqTzyqGyw^Lnwr{E4vdsJT7ZUG^aG$c8*PD2tM~!xfETqc zqJj=G2vX@Hn!X6^?(QbgF4Cz`z{?3%;)hQpWlhU(FmCURGs$j_v)Ee~!9+vZv5p|r zcoV$eP2>kbK_o-XNEwcyj>wDaK+UYfaCN450LOzpB*kU1c^wTejNYf^J8E!I=*02O zPQiLT1nmIJw47pK`r2Dps6WMa96+B8sd! zHo41sH7KtIGxH**&doHszA>l-AB=tsRWrL1vR7~AK!s^^TrR*WsCu*@5jF; z?G(nAk+zDy?uxdCd5`rt%U@WfxEr}8(-zsD5}~|wz$)RxA0jwkbfm?FQIS&y?1{$& z;zK#pa;f{FS24_f(YlZjGR&=2Sw*Go@b!4S_@QaV@M~9gyeP&w+MR0--A zf8UN#N|vkLyYTLmWs;*sJd0HGyV|`J#*Mx&?^Ys%uZ4#(p064iXodj0fsM`Z#~MnJ zFe&+8^6}zy*^kyH_Dk-`PY^>fZr8Bg}F_yjfA(=8VHd!4ymyR*3sSjD`DP@~*6ZLcKdhOJ59CZE5#q_lx6*pP8nJK zOS3zlx{fQ!>rM@o!gM=*%4*?5&q{JID1ekh7x$7Rvzf|Y5L);V8&=ITqa zzh&kk$em|!I*H8{@(ej+$jcYFFP-<6DyW1-?y!FJLOYyVLZB@}GG&$Dgl>1_^}0PV znTx+I<_~a-(F8`{&1I&i0~(%DX=EWr%%mDhx%3@;NUM55ucOmcGmo9k@lz*tLg?&O^RlSfzdVS}B;fs|A=`1fztift0KN+9STF2!jg;Ek-_ z(DpZG%T$tX2r5bGZEHPfI&1mOEBtJp%b6vsGq#*l3%=IBJlQq7pK(1i`RI05(Tp8Y zSSgF=U0DmAL~|4zuN2mARj)){Ysj|3gC+B8Y-svIe9>{=5ax()8O7TQla{xn)z*!> zN=mX(_N<{xS-r=h=l0@47Kp*maOqFYHejRn2tQsh=J!9NJ~ zItL4JimmuyA@V!Evk-*PSKv1EqR;w3*7XIUDPhf(l$KInCP%QE50r;LHx_iL=Poz# zclWu$?9`u>HlDlqYdJ_q%flYtpLwg(lFc8)!wH=r*61p?e?I1ZQBhL5``nQ+yBi(s zky_eMwlX*OojEQ9WSE-L{rJxsAV~2Iy6wL2V?b{sIWM?rVW{3#S5w(axH@$UIu4%I zXBr9bSfWJpN=83VdO=BZ9$m5(lpTuHtFb;KK1IqlApzB1+uK`RQIlU|BPY*S^q?bS zD_`}AyH34{hqI)}k9|om=tc5b4;JA=y|a__`~y=BQaz7)GK!l{+{%5LvpoBy=Em#U z7br>e^MSg4q*(&$huvb7Uxg8{>i8#3G+*Z;KAounQj(5Y0?EME^`DyET)lpY?TRLe zPClm*R_m1uE#F?61ijU-JN9Tog8!q$$I@O(Vb-~mjHcDGsswYG`(lyRhi>OZ!!2aP z%Z|KTPm};aT?V=%-6oD7x+lCB(?}X{*o&fIMjP=6a=yaQ!w-WeUT3P4UDXuaov}66 zyWqk539VV?frr>227$|u?E=Rs^!|NwdGSL>Cy)*LB<29X3|~_O3jWCly3Bg>Y+%PdXl>|KasPe83*vG-@hI$XhRzK9L{F|yVOyL z9z$P$u$8-b_4-r>+A@b1`i%6f%PXY0D7ibwaRX<6eTVAKEIIrWQIt^=c8IZ zn{S9wz1djz@@d|3(frTf_E5AL5gOoYcq}tBGdLD)0ZY^bE~BN}3?pbz9Ezg1mPa0a zi*twn=*UPF)Jl#>`x%qIW}4mrjyDwJNs0I*8fR1R`OZ)}hq+-7=M6Dzs&h><%$MKW zPCHNSk=R0cxtXNlAO!$4tEk(BXs~PyH^mZw96AG$Jc&3`F=02@?J{(EBx55DZ}Q33 zvlsaXnueasw(q}hpK&~dQ>AvzF{>w>NgQeySSZLUG_udP(6Av{??(@ThK44%prF84 z3l+;POm#W{m02a2c|58(+k@&XvI1vkW-JJJ4^$S!5!;v_X1yPdbc6)WXKGy+VXgP{ zKCb=DG{r@tXEAI>Zcn4v2>Joi1_+C!Z?ysaIBXPtd^9qUYFN>R+_}@y=Y#suU(BY} z0JZbZ+SkFcIGD|mIws*^`vz0lhW)fS=CehRPue8mm>BMKBMSzHLO1(6AenG1q=hh5O8!fK*hO_v?#E? z6QO?`9p1Ch-+**|9Mx1wMa2=6+XeWsoo_=Z5his4rMVZQ#g|#zBbiWRaHOssFALWZ zObEMBMCHNKL7%dYbi6RPLTZtS0OdF-T2nD?8UA%z`B8!S1sGLw;#Hz&#O=q2>1aJOuRDp>y^(OsvRkpv4r9Pw&dirTs4Vi z+v^0j#v*Ou828+EVq_me+y$|KPk3sE$^R`;rD;MQ26gS{ff)}2sS@FXx6 zj44G`O)dD*qesYbu{gPq^BWl%O?2R7DHEn0@4axvDH9+bahe3@C@R9bY3oCyd`(w> zUgVV+7RnZtGZCr9jmIK9Me5 z1c>%>qJsdMu==#9it(B=K&6ZN8v#f`Qtfb9{=(>Uzx6n63Od%jQ&Ng+?>Z#bbL*Q3 zZBgkes;E;Zncl7=%mA2^rh!aE5OK2OkaMS^IdTpKvkNr5q39)`qJAkQfFx))mTwKI z5%;rSQU2V;h=ej=LpV(C+O8VI(g8?3Paj{GcK*Web zq7KJi^Wa$0L;$B@Vg#oNDaA>T8~dGag`lol{UP~2-64{$Gj7h-i}PVNvS_>%rGLz_ z^O`3`!w&b=3t6cVX=##>iqw|yQ4m_RNU{qY4X|>q6!0EfC0u?D8AkwS|AAma*+Q>= z))@>w$jIXXPUH%3d5Bybj(Tow9<$_*Tvg`qQF_ddjg6|et}U3>Ku@i(4N#97(6yOv z`GrK%Y)lJ4G9a+l*w~nElW^fTOsx1+i_z+&=rs}%iJdn7kzC&zn8o$e>4w^dCD|ENz{&mss} zD;IZM)G_NikfxbkWabWt=jHDzdMcY7>B=lup5M+2&Nn^x;Hfpr2{7%=WJAFTA zj>#`Oua$2isMXOgxa`=k2F8cp37XE_OtJWG-#_1-d?I^rSs+?;cns48gWC38io9Nt z{2~Z+-!#1yPdI6}70($0>=Y3H=l&eba4&f7Shrb)eM-AYk+sckGXv)(QJ<=I`s}i% zq9(HRzZ0q)= z_ic*CDzY)L4iLAR$VrDlxvC~4e_o&F>od&{A4An94&z#SdA*cf_+Ebs?W>ClT&=Z4 zp;G~?9omX0!ixOo?{0nyk27sr$LKtEaE{gfWywpE13^K-StF;Jn`byipS$F4;iYKP z6T$-4`V6pwanL|tf*pky(?Ru+u$`s4MJyJ)Vkk#rM z3-n*TzM}o$?3o~DV|jkzwG9!(hQT>n#Qia_MEvvKM9 z!F{ardkbdLj(UqdIdDiL;|I&P^v2ZOgo7r?f7o_JmBY>v41&<>IFo3zYg%@TEW-A=Sc3AYK8 zeN^Tv4fF^n5+U+JVS@>#WCdgTa?`&klptoBlkZr42VbfCYiw+>1%@siXl0ptW9`@1 zG%`7^(8%QBRV@!w%?kLPz=mq4DOfjl>vr`~+7aEy^}7AH`gC8CC3~JUCeWs?LN6%+ zs$v05+5viSHhQGJ{Up%9L%p_M#g8)un2bs48-&OW^a6SrD_l zJpY!G+cv-InZK9`TeeauNCH_{_dwby4dr_py7=0`9bS4USJDx}C8Yr}*j`weu5dSC zqbgLHnRo4dH*(of2xZI|AyFYRPv3qnJ@5Ox@Atjm|8Lv>+wN_9>SG--|W><1tLr*_F3bsg#d(Z9c8 zdhU`LQ>%8&+d4Uh)j=xB&!=A$9iP{ouU^AIQ6)(U}{NS%m8)(B@m$mxgBZVj;; z3>@H|RnVSm<*Cub5JXXcGr8{Uf_UJ}J{f#JeR+n+6}7Wu!tND3#Sll!TL zeg&&g?UjEu#9A9w*W%M3wxF09%>6DhGql=`aPDBeG5|h|El6)}fM~NEGMU4!9=cq8 zE$+>S$Gdmn3=VCnQEJfB&thzh78b9wQODOLi2OHm3Oda__Wl z9?~;9I=rp5VP6Z3|Gvor1 zKR-dR%QmVz0YRWpmdq+jAG~^H&~4!N}%?T;JKh_iVI)CN*cs&y?rdgoT4kZwfM$LD#uWxp-u6)6c{G!TS39Q(LR-@Lhk)H)X| z1r`<-UKxB)EotN-U?##)+X+li+$9|Shh$>nWEt6HuN#&1T79wp$#@EPiernTvkZ60 z-i8l*N({0!$j(B>5rdH(kFnX&si^64z+e~BKEX0b>|@6px!0n*JvAEJ_M5Zx$Gpn- z5_`79U%&i4Ja|>p({#BOqW>!TtB;OLyx}N=|z!`eh#R?pt<~t}^N*7ZX$R#)Eu&)x-9ld1@G; z`luk%RoM%f}qx$*q83I`Ze5Wy~QNF;gSq0P%3+Uh9&Vfc0+{P)blG(u7|Ao3PNknN)C`}yeO6P{fHmPA)9BEt=WDCR3*iwYkGU+mo*%gU#%DDLlQB((oP@-X z8;o-CRhGd>WRH69R}sAwuV5&wdQrnT+Nwke&-L(#%K6df18&U}oK@}*yY|K_X}mCh zq1?=9b3=(_%lqzs|zjW#&j8I=+^TEgq`n(64B(@MnWEp@;EJDu^b@PbjKyuP>|3 zwWV5o74YPoIwy6+kD2U`6xQ;bAG=nY+dj)r(G@wxK;g$0;Z;f3J?IIcfEyLl;QO+tuKS0 zITwi}A{MdUJDY9V03B$R130RIpia)b_8CT^^B;Ua87CZlS#UNinELuK4ATcQYx4Rc zoW)=VrJ1Vcf_70ab+x7E&(_#SQoNBVbtC)cW9^P_5=I_Er%b>4=+ko@$Bt^Z3$0X=+scUU4>=s>5ANyfPd8M+_J&pmpQbpA@DT5v{2;JmF zBt3Q5SoZYyCj!$;Kv!~PxN`(Iq-h@x`jOF5SRv_ryJlmPg_bnu!OI}X(xe>-k%5!; zpYr~X_iaA_Dqc{pxSX8a_9rTq>I7a# zl`_$0edeWDI>FRIS5d;$;^6f3bkd|E@-)KPFL)~^PhbV2XMuwg;q?>#0Hz3p7HpG} z(ncPgad7UpGT4f=osHHKqk)2q%tti-NaOYtm2?Azd- z4P@ww+7^J=5FCy`xH5gxbK@XJ5=0%a!j_tcU@`)Iwvaw({q-(7L#^a3P|D+qK=$)abL_Nh&)DS!_v6uip_On$TJq2|nI5uD2ti)!O1$*x-xP#f4 zAS7<4HL|w>tz1G+CS6ikS$(R3q2xKBHiAkR>=1)+oBiP>M*9p95**v2g4A2DthH6U zqod<^wYE_3F2?}F4JSei)Kp)IONnlmlYFt}@SE~SEfx>cO-&+JJt|_{`A^3A>8SjC zX6xf;bTnrwuC{7K@9OGX{r`xD*hlfIm10h92^fKXOP;A+sj z#;MWFotp*kW#b_{Ld51&9xO~P;`1?VN4h;$moXW&3Km`&;6{9+=o-JN^) z8juV=`8RSTc)oQw(D(t;q9kH-a)p)_N7Cr=BXz<+UM#0~N0>U==w#8r^-RFn4+092M3JZ?O7~c!pX(%}w;h3m%FN2TO-9mT zI0|91KK{%m=X)<+ydb3>UfCbvJoph*3+v+ocV)<*N3=u@Ku;Yc}Y+5*6? zvU4r}XS(K(O*!DPZx<06q>d~bG%jQ6?~CLOjWZscQITFg#DDJmskQ#p3;w3j(F09W zdpsv^=4=rJ6N!9%EPo|UJ%IZ^;t|;bDUjapCSzWV1eg%TP~}w-3G~PlNkkX z;g$d};ka@20cnIp)ab_x4OLWBqQi<@9|0?ObaZ6Ndw5p0J@i{!3-2HKPUIP%&2?W6 zzT}Y!_GrD){_}I9)I#{nMZPoP0-p{<$j1YxlYyBFZ|66w8IdLzNZ3{qd$6k*584Bw z20~bS+(oB2NJ%-?laZ18WZ6Q@XCcO*%gpl!_KO`G8K<63_P*HiTdne&8tFptuSeMa zHWa32(2ED4i4Y6{x3IEr?`Jy7mSm;kwo}|qP1SxIl;6rv9-J7N7z^=haA95bd(C*r z$-Bo3p9-|@e&1lluye%2)-f#}=Vj1?2j3vX8JxBLng)?gnfUgCimcsalV_oZWo$IX z3GjjI1Le-;XELLGY$gZbuZ(&^L(qUq=bF=-E!~N#DT%nHn6Guc_2r9bLx?vO+7=n@G@n;%9$u{ z$yq9&g~f!^R{gKx^@BfjfHpD>t*~LSpWkP_Na7CCHE0Q?Nj-mNHxXjh6>!@#1<%n9(A#0pPkp7dCJ}3kgTu$A~3F#~%ijD|A~V`E>rtW#5KV&gm2g z*bIz}bx71wK*s%K&ua>`Katl!p`k-s0lYE@)-Oorj~qdcO8RX8rHpX`I{hkT!|+sr z1GB>=x!=?BK3)Ht;_ibG7l(A5Fw{PRqDuG%GY>#_F)y04Be~7^>T?u{;DD_zQwFb? zU%NkBDH$ZG9e<;iv0;+_)Vy08=cbm&H>US+eIv=|(-S!6g;?coe|->k| z(8`RSly*ix+_u&1g^b{*i&AJh6QzZ$>>bp437DuRuw%yuU1=nTjt&l2K4n~}c>lg0 zI9w{236;j#EhfYm3hLg*LTzGEH3o#XwJcRFnYhT1h=FvGWMo7y8lvlzsPcI|Dk)5N zJxzp|D7Iq>>1+|ss|DLc1ar^S%a#e=G^mUT;NvX`PS032)bAbjTFTVP;Z&E~@9X!% z#KmL2w~veu+fAKd0%k&b!Z;cRGOm*>6)a>y5M_}RD8V{AQLj~O!8I)8!UnfIx%*f> z!}kxOk;^e_ugCJ@6#eF*myE(l%fVcZ4V7mcm^DPNqh3Z43_su;6X`1GaIX8EpxNkL z^t&xgYyIiI>y^Ry*YhuL4s0E4t#1mPkuW7mT-0Rb|_3u{F|22(7g&pIEl{NIdS7)n}u|Ow`2#n!WX1PZc|fU ztW*(yxJPDYoMnHS?Ulsmlj-Q@YB0nxaZr2u;bCb^=k( zQ*u7(vE4OXV$}JCXPazg2-|)pg?q|J-tx(=c`u@}o^_j})2UrCEoRPw&vn^Imnu+- zL&KmLECFzh8wsl!!~*rQL>p3PB2d1VQJF@<7~5gB4(rGpKpFw+!)N4N)0VC&OpZM2 z%a-(OG-T;g25#p@lZ029FDfr;oV(9TQS2Zr6hs29h?>euWfbdIt*v{zx(K2wu}`zt z4m_iR-ve?jxz;AP?M!Uqe!Ms3+UI)g)xK8*>b0~cz#BxW;;_QMXhN9v5M&)0Dwv2^ zAFyQ5q?Dom&25h|aVC-a;NsPuZ&T4pn%tyJzie4UUf&~z2}XM-1&X^hK*!AwtG*61 z{evA%m?%Vmh59iac~Ra!I=!~=sE;*>E6?+JaqYFsIj;Eofdv8;4uL0X{PsXz)CDB_c zHX>URO~i9t(XykfCxZ0Aa=OG!b=);NoiqmDjtC0y6H(FSrE?ifs=~>cnh&-mMizLl!i1%d3LL z1rOq~&R-7aq67Q)lm3WVN4_SG+-E+@%hIyV9%h-Szv{mAhHJV#JpPN6<|IE%3!JZ##MQX_mgst4l*ac31dBCf@9X82?BKk7$iY*?fq2UGekNT+7{) zJXSQ)<&QOL^QO&>ZVD$~gG|}`;U~=MLyN{!JppNq%X2b8DykHX4d*Q1VV3l3Mktl3 zMK$4VlRCqy+3tVDRFMjgSfDlCn{ z8k_Zup6gUBK^k;y8&uPL@u6++N%N5i_F;~|R5ph*6Wj**N0JTbA2(H02tJU!5=K3( zyqW%UQ(Ey+PvfP_c#EXx>vleVA|6xbLPETg8pYiNIWb+G*cg;B#B|4L_E%YNUtc`o z_ej4$M<*IXK2_MIXfvz|bJw*>VsL?}hxcN)zFKf>oi;JAY7l2Bd@GA-GcP!9b< zmG#w<{Ug#?KmgcB%7_9+`l^1#qvHh|JJa6BHAxZ}>=>mR%6Qp%-{A131ZU(-k)a zF-t=$XV`C9oyvz-3&sI%=c2t;1bs?I&}kB92bxq0LicHiXzOw20euXwG2E^(A^6?zVe0GdCL`~r zkt5;<4=O1RZ=&z$7-+r}_;}glM;nUrgyZ%5Q78S62eA9llSvMnnMZ_0~C^^tOOa(-8w~^kLw^3ajBWR3K0>B ztBN>ha!QplzIVQyS^NFmN1etlN%3nT0|@cq{{SbvwMP$rYfcZ40s zG;A1brg}rRBZnnKG>kH}fo}p7A^mX=&1;b|4h|$7dMW984%`fBrqjxW-iF40dR^I{ zkr(<&cf>k0B*ebb8S*bZ|K%@=iY?)2$H;F77Epv+!TB%f672Ck}ss)T6;gbM2CfFNL+Y!OUa-xcnKFYb_5e#>=z)oE=VGa2rq+ zztSpa_Fbo+wp))AVsmjhvQgBCeFGcZ>wa_yqnY*UQc19AdP(}X9KG5p0%8eiEXY!? zB{elhnNMWwr>7MAmiHi}Q!1*OpMjVZF6}EZlIl9zu92{fM`PsIe0OoB+}zAP)%*7M zM~^I}v>`>ttNDn`j(AEzy?6-EG|H6Y^W(#vUma9L;sjjAf&%@|T5?)$Q}BM)p&a3F zk^XBu*yHHD#Euxl%p3Ljc2bv5(ou5uxyk{F;=hc6IErW(1W~Dv!A}%GFAmi^(c{8G zlXRcZ-;v$Cd2^aCLtvu(!x)zt25#*UYXd1s`P;7XKfAqR&kuCIyzQ~XmLqXN^~G-6 zzbT5#|7Z+A!zX>2%vtR~EVa5 zanG#P`7nlk!x63%pVevU!~8FWH|v;FqMHs1i6)0w-&jf!WL^Z2QV?!iirk9kh!vGq zu1z8nI252w%GQ#m8M-t%f?E`~J$)8CHOgqWDKI8uy|ht$*<)#W&Ol(2dXdshLaes^ zj(NtO^;UK&h-ya{iTF_0oY}6Pmi769IOO!iRs)3w$$3Tce(WN_1l6~tnPTTUbk-l` z`G43hB_DsL>Vu}A&`eH92s6jk!a}D}g@Kk4$#}!%6qVD9nME_~^O8};mJ@>-LOCJw zOTa9IO91x5Sc^$|KiKA1t&}>y{A}Sh`#fKE-AHiwgPU!!X!Sub7LMt^lIAVnDyVgI zdR9)&r=eeS35CV4>~jdVey0-XCW&C8dR7zWbuL=SI)Q)jYa1EWdeHUx9i=|M5>iGz zBEC7A8q&9kJ|ZGFcjC%%x?@tJJh;`}G-}_~#%Gt$93a_(?~*`9aL}esNWRF7J86#g zwUa!1Bd7B{(vh&4-#6CsTQT0Noy1zj=s$apI4J1qMu7U(d$E$$XsBbftnj3vKy}|4 zWqM&G2J<_u6?H#~i{m4;v5nloBpZp;-1LtrrBIY zd54@HEAy^(XU>fMbdDrSv*5K9#eE+Zw>xPK6i$8Jh~8mc!&&!T0W^X{;6rUo)2gZu zi237$igv~L^}S$!-(81UuF6D=qM^HguXDLo~nk-gQ&w$-pY{El=PJ$2Jlu+ zAQhX(%|rcj!#8Yxe44s6wAI3>XkzV5W`maHM8nI-tC7R}tMv!|p>%Hf^ZijMqK<={ zAI;0W&W)3^v)7wiBJG7PwO^}nwP(y4mb7||^3gJc@b9XcsT>~`4%0c%ON#U1 zpj*6NCK!hpg?UyB>*i~KTpQ)HZb`^CzLUrUW;0Kk=PGO~N;p?JW!rk>2#-bjsCbq3 zxzxCwT;XgeOO0r&Mj3vgmKLZGiFhv7Yvqtj-QWGQyQ4#CII{7|n!oNfW;e}OCL*B; zVxz@(3Y*Pr4jO0JAL~DxM28zakyQ^1wuz~X%+`%7rl#kp-TO|9%XTn8!b8Lld_?KU>rM1=0EyR3epy{-MhVlIPbL?IF;^q{ z7bL?FXm+Qbk`Gc#m2~?NnQ=-g^?P;Xj+o(LX?d>SBehYR{mRSgnr_1fP!*LEPVlQw z;ce$!H&6qMiZ2-*J63u3Em7Be8gqW)u*`^pblTQ_i}K$RMR|sn89lv@g@u>;E*L95 z^9)di$?DR@JL)28TDlQv;!IsycoJ#s-6ZO1{JJF8dhcqj`~q?P&L?1xGEm${ink16 zohS^N1zNHSA`U<7o&1=3uBlvUlRh-^YoA+NY-?%~d&7s@;NGjQUc6^^D_7R~Nwe;M z{?%Z+@GnOL2mR(UVO9a5-`8C3O$#?<*HtMftfV9uFW!Rta0ipqa}y?eztC#v+pVVS zN5o%krZ#8|e=3nyJW4Mu0-bXkjuSy|a-2{oG61<@JMgNzIN)Knx*i+H#7MA*2M+aB zV*|Uj))zjmVoz7Ffd*&{c8ZXdFwJ$8bzvXqTbfcO2>ALEHPa|6K zeb_+qv<>7gSS`kA$MwE)yTtIp{W>ct09jgTx6G6K3=t3uP`o=9v!LuJZ~DTetdcQ^!IVCFz}3s%aFAc-!p^FCzVS%HqC2!?k)ECOpe`Os65xw1 zYpzA{i6%$!QY5>I8c z$4qzEPi4_m9X}o$E6`^*24`SdTuv>1lH6v(-pOhN%dpcO_EwnG^#>S!x4QCYQybDWZ^2U>bKz|W-=?_9hT-jFb@jPFqwD%C%{vd`MvdBs2M z)?eip{ZkrDt|(4?gtmncK-@1aZg$xpZD}`}o0aN64rNzanE8hlA1ZG;N%ix~)pPMq zqEoSURRo0WzIYwC&6>AxN&GVv$2Xuc?G|KIB_L)#b{{-PcBP-=>dDqOxYJfLP>Shj zCpASdF_0N`orey6!571C2^JS|A;TVtH*{?`UU?T}|L)>2>kvei4eMAM%vb5(FFPVHpk!@65assBi^HM! zCzLN>6jJ!1Sfhl$6&vMcr^b5iKadPdvPZn4s?V<}qcv(v9#oM~BSNza4OP-_Yib#I z7JEwmP$yb*!0Eq*SuqSVSdtZ%y^a-*f2NVIpFN!?>Gi1e{Z(6ZSS#L(7M*e)KXpCP z+h)hivR>y_hI`rvVlt%dxJ>OEcIc=Tyd)$sMFv`CLs40W%)olJoRM>Nr7m46mayR9s96vW z_Mf}YA(9_Zf@zQmF_Fn=>`Nw^DYQk+|c?(l^M}!_Xe!Ue<;}hbAN7`FgS%Hd55ur&9 zpc6;)u;Tp=7MpKHEkpKE?{bQYdiu6&dK?wY`Fvv}!+}uJDIqjl8sNcm;}5f}E31EwifXQM z{Yg(zF`bnGV~e8lPtn}ApY7kHZ*^}UdpI$$rJGgmRlEAgZH2D85~c zIV{Icc!89vvOgj_1wBUY%?z89W>Tn5F`;K>c^iozIEIu=Dm&b;a?X0 zyJMEN;p0lgg}0i%b4iKG~gv+cBfLNq4u} z^^WVbEjP7TRlceU-=_?-4Jbs^-_oap&dum*X%S)a`P(I(6Si@Q@CXlJPhS-0%-h+AlMS~ zIFRx=8`ELwh{i}G=2)4wCm+dj7~Af{YNq1JvF`@d-!3 zt-=}^#pUH^mFSsHX}^$tb&}2a#Y7zLQsv^~Kipz9GWwekqA3-$RM(wB5LR}CegWe( zhDSqn@qVBPs6TGXXf`G2GbTAW9(%(JKZpmC-<-#Dm@XG4Un^0IA-V1FeJ%S!-FwO6 z_O$nlB!LnI;+OFKj!GnLfc>P+K}feC`Q>%G2S&0004apJ?av9{2(G6z+r?9YS6rJ- zk8c!=MaNg4M&U9T>-JN+N)_*piJ%gc}eh}LkvmQFQ;ubRND%zfO8d4 z?muj(?p$?O+U5|P>ocxW1wB4SEQe2_c~WOpMCpBx8>SUw!>S>^TW!DA zvp@RC-sEMc{D(XC@1~ZobZGlIAquk~;BMj5W(H(lqbDt;i0D6vE4d4-!l8z4qBG%x zFc7l`L1P2c8xqr2`(63W|FWK>?!D#)IRf+j$f)I`TiLl;=dyX(xy^&pBfT%C7s{K7 z4)sv)ve5mPRT%yCOJI=`MuIKBsHkYRS*se1wwj2ET3-lAKp0oUGLsK_s0u-MKvzvd z!+qgr)35RI;1@6K{YP1SVo!I(uzQ^~Ojp(X(Y*ENT)WEP?LL?p=)8?=QI9+A=qP*m zu#(Lj>p(;QrgM4gIv1%vgmLgR*YMnEg&#S%57F(xL_Xr;2fT+cnrTFFSP8@a7X*R` zLZCf3JA71zt|}Smrz$bnfTM|iku#k5)PXpQ%pA1TRq)Na;@{&hamHJyNFrtB>g@6X zHuPr>2(Do)s)YnfdOizVE}_^ww;ydnKH`#z44r1(4Ba*%pHpKdCmnz-G=GPkxEbky z7S0UTxlFWQmX-!TjW{7R|LQ6Z6R&Mv|4TA>KBDA6$+BfU78Z%xP1>r$S<=FSVtGU6 zF>VJjQ80S7R6x|KaGmP=`d2&3%cKvvINb{>?NoO}-u)`liT0{bxvFf@5z1Keu(zE zKUts1bN$4bw5-i*RFJ!jwFr>7twFlOf6 zt+qyhM0O*RJFQx!wX*D6hJ?0*m!04GU%!siP=wF}8Mn^{* zirm~FNh79KVGxT0*o+4gB4M`B5w~4XaOYXAf^E7o@tSq}X zY7*bgeMaRR~`SCo~$Fqa1pW+2oXuWdh8mAl%SwjI{k zpMUuho+E-%zZ1?14F5N(nTSdq+?kz6WvS$n0l%05i8)tv z2CJ3S6QW0GiKm;Ji*Pf?{U6_Osb6?oW}q<{1TVb7;;Y0!f*J!e_9J~yo;BpL9O*HEUF1#)Ajh^+DPQ=94AQv|LBn!jYG-j8e2;)zRSuvBO*gB-QrWaMN?=ns7lGQ>U3RxPh@k_Yr|(& z1+upn-F}8Og`afz}6w^l-=`>z;#V>vauz$%B=$~GU=*~@JBZ=GCX_w&|Qy| z!QBP5(x3FWcd|}Z#F^HBzb^1*Z{<<>r*ryaiSedGr$~`Kazi!wx%YqY!B{;0W!x-u zMfQsYJc-kV(wifj?kzuOR{r!_DVOu%;MvpO-G{F$nLgT+n_qShj{X$)na9_AzXI9a zme@I%FwjsV+7XJm_C?xBc#JBLU7+zDotHs|vtYBfn85z@Y|mH#mC8_14X&4cF^<=> z=H}-9QqSg1;aJWASLV4%z;`5j!DLlY$MzHqT`7&crLJT{o%NU%>CdicDX$Nh?tP)B z|J_jenCs2t?oV824u}lh;2s#3i^wl9nVnl7&cA!k-7a=W&Gk@P{+vkaOU>Z-`TAT{ zV}T9Ewhr-I?m!=ldkL>lu}iYI{$;*=t%9G$(uj#WFnB>ieTrR*v!lo?TXt;A(t7(( z-Rtt0mQv22Tvk~H5fd@9BVLja>Ti|7_xy-sdJJL@IVzaR+d*Ft26?u`i$!RJ{r{80 zFYCjPG=}Z*A>XIr4@K-&R?B~voCuNHsk$SzwEb!5F3@20cN=_a;|t;GZ_848SXJ;8 zcFB*t#8_nq+s3|x1#uF_`aqafRTmS2@`8(_qLIKRmKiw*hX*!C{XUK`SVu|)EumP_ zBm5O3iXllnIP(i=VKzj%7$!lyx&CJ|mkgb9Gs#pAz`z&wp6VJJ54c??FJxc&__}cW z1=2*X5^ws3Kwk?;TUaI+&>sntfG{+ugoG+?)}lfpH@u({u+RJ9jM0ekFQrU35nMv9 zls`Q;=V}?Ye`FV@@65DJ!Nr!uYSlT6G@wu>a3@yJB5xaj-QV0~n@v7JG$Gk{fV>4jD*DnAQxfQ_~o*8=d4lBH{eS=(~6C`FK2W{ZV9=zpG6mH7b{sCO2tPHi)S{I!zzwF zeG&+N!D&85@y6}+G}Rjm{bHdT4rnv=-01T_+fQ&!I+eQRMuWj8huc-C&-hv0WJ#gj z`;V~v^yjunn>{}tUNqTqRG6dDIZLP^NeWJC6c$9Bkp;~*(S)E8Roof4S2dgE!cmR<5E%pb!VY)f_`Yofy5dzYKA9iOBuwci#(6hN>DBF+Ap+~|K06!BxgB`DT@KmUiIlK3RLQuLXi?Nu0}wCd$WF~ ze+cFxPE3Cs6DW!&f`wlgO+@tR_rywa0^Cz?OiSMGj!FGnq>_U5!DHshi& zl8EF{A;n^`qD2*TU0an6%4;lPpe8(a%P8q>ldmN?>ICU&A6U#g+0SP**XU6^8@yy-PDEwa)uM51H6~JPhcRAOQvOdt%949sXU%U<(j-6oz!5kr6`Io7yAPu zM|2~8H?zHz^#A=hWeQ^-W@lpxSGOiBOdM8Fpr;(_U98}9{`%}D`t#FFbl|VN602lY zV-BvAh?04o9e#f!p%hzO1_89igtfDkFDv)4YOm+Ia7ClW*9i zreU=~JsI2nQB2y%5*u;!?c$3TZ_^>P_GR~ERwJPn^mw|@4=N-Ns)n9O zDha#)ccQ{r%Yf6*4Simyo#_)`$?aj}zusQdTtI=WR)P)NX98Pl!h2ycC zWv6Xim3@^Va*0+aZH2mBVA4nAv-NppTnvoqy|njBCol?rEaO~eVj}l62eK@r>`DzK zyE1mgn0B&D{C5)YC=#!^h&DB07c`9NMZrS9HVJg(R$hP_76VG64j~&XNkxuQ0wlLEPnKF+zxczW@FCSRJ6hUPv|tmDyG`vSAL`oMcGT^ zC{TjJfMK}s8br3FPySwmxIHbqWwisW=3r! zf3f#&+;{w~Fw~?+j5~IyRcVi3C6(~7eGYAZ2B|?Bd(ks6WH?yidlV@o1>Q(2&e2O( zkVCeO=2${ek)4{V4bhXGihthkFdo8WvUDkrn{y)5-lC%<1 z<_m}nsLFtKk`s!oj;~MCjISNH0G$u*_Y;est8gwn%Puw0?LVEr4iZEJ~vnf@@tHYiHqrfxHR%>wT<~9tgtaH-YX_vn|Nd+ zx%2qS#PoMII|gB#<8>-Cbwxsl0arLl{Ld{%ZHI-SRLJ6UjRFGKE`;y8SO1vq1j5j_ z46T8Eol9t!$L2{h4poG!iVQHCW(C(L$68N#i$`y!rl)&*=`OfoOu$w0(Uu?oeC5es zC;NLcyYEQP#FbcCL;`AbdfV~XI9tS?gmp`S#h1O7vX#|())H5LGd?K!{g!U8&Bt%b zjXmDD{dr$1Z6}D(pjnEFgIw^b=3fiScV1R@Ih?x1+}ZbJ^WND=A=YG=umP4>TtAkU zp(42=iy56?p$pa$gT1O~nXdXW|3m`1Kr_OaSj1 z+O{OA{!k<_k7D3Si|v9b;wCZ5d7o>IxUNcYZ{YW&r#C9Tx87ib7wsc=E$%)RC2SS* zU!sy>qp=8h#~D@ztr*+LS5&-X@x5psPZU8;;0kksa(n@C?+iZ?vpiX@hvO-1v37bfKF4CMNl&va>HYKd?v{ARj`_s* zr}W0)1oy6w(p|07zOZERHGeMqX8!(uAA9wIWAhX6O+Kk%J2$o-lGvYb^s&417oU62 zjElbnejeDRb?SLAuUv{oCXt>gA`<T ztbTDm0*U95Ik9U!>vsvwYB5p{8GP*;z=tuJ#wDUB_w#en=Z3Qk#9vuoKL%tTcRAB( z^Z4gn5nI`P%Tl>(rM-OBEFI)A7z&JQz&3}1?B50_|JeV~@W5Qw*Xfx&%o*zF>T((l zICJIVl7y?fs`X6UQVGA>PI{*30Dt>@6m}L>Gt7T!?^O}EvZcb7nrS`(a0pj7MB!nN^*8izW%WN zdh8EVbpU$tty$krn(fce`UFTS*X51F6#Y%foe;-c@~Y(_;<&e-lHdQtujs)xWy4JI zM_LnSeV*$)dcc5=4+rfS$+F@N`l5 zWb`Mil8X&o_6mU^FZ^h?0>oSbj)~Y0e3Ef_KAFvUnlmuhF6XSgwwS%vHvH;A+OuCA zi#cNUdM>*C&!53!->aeNmei2VMM1nk;m->iMKlt1_}m2T+6KB}cV3!5!_93PwVHBB zl}4tdhy*x@ot0~Hv3wQQ(QdAKx@1~G7T>IG?c()&85uX4C%kJ)VzZ+1nWyHDpwW=nYFPeW0-d()L@qd%nC!$OF4I_@uqKiky z=}4xyLE}$Iy##k5BTEPKM-_my$ch&Hrq=Ch9Wit2%f2isVVj|=&$v2&YDK&O`TF-# zVSh;xOb8F>L|f~)-+^(B=utwzhQ1O)b~M8VATf@%CN6%kfPp9uodsx|%d4x^?ChS? z`Xp2y2%8^PsSUL0xy^<$h0m6{npOz?J|mi3G??q=q5ti%Zqd_cXm-d4m?$&Q=6w9S zh@BSgSn9$3g!cw$W{a(|p!@ebhxeg-#`Jx3WW)yI4~#D;cxO!{<9I3CzR1A97!1f( zGB{{8d$VwAUZ;74vE#}E$%VJ2im$ZP+b>xJH?%&6WE*pSRe{w&7%5mMW{|lf`Z3R< z?SGKn1 zy>s*_RT|jKd2s$j;n0&f$$mnd)LUgW#BJqzRQYNLyMY-2Z6Mk zyjHPalrAE}qhKN3Wlhk?^RGOP^r4d)2a5H+a|=oNSp#ia%B}AkXVMFj#-xgZcvcTu zr*APiAt_Ao4+#>%Moy3V2?zyqjQqB_R_3u?xB?7y zi5C$wB>WgvBPaM^-u}m@oh~r;qQ>C=_>PHkO?Wx*A)gqn*)1Z-tHNMF{%umET+mx$lg^-^Yi+KzFWWHEX;!kah@fUx} zSxgJ?CrgtbNB%muv`#DtY7q$Q)lX(tR+k6wT`7vnK4h z&RmUL2)wE^!kl2Zt7<={vXGZWozAJrGmr8puc-_R8ot4GD_RVUU>q1uLyy7dh#Zn`+Ug{fBSwD&dZ47hX63i&VpSfYFhpDxz+vNLe?3XpX%ucuzMR@mpbRKF0L-=ZqxAk6tD*Ne(`6Qk?Hw={er*DrDdR9FaDh7*r*ZH zgg#wofB(u@rInjDZF*Bywo!^F0DnOb`SjP5u}bVxyno(PyyEqzCHMO^Uzl^nor&}X z6Z%z6%^Hj|UJc`kTaXs#ynM+nwLjqOxpS-59@f=8`1C2mJuct3kE8@YE&cn#)<2)i zy~^~wzc1jw3urw%Kn1acg7J!PVcqZxa?K0y60_xo@9oNw(UHwBVyi#r?VqwA*-T;1 z@CpsRmsujfyNPmMtNG7&OOo7^aZ2vOh0VU)2kxMeY9c@Mxkwu9@2ClJ;3XJr@nLoW zfH}n-k<(bO+HPvZpe-lDBj&EMyH2*6YBz&9^Wi0QBTwaq)05tP>bla?&MLv2Z24Sb z<8k{m4k=FIJ|Bx{#U-5ED_&K0Tcf|78dwmX-rhq7rO5s#5qehQ7rjNKSh5VGkV1m)p;kgpz{YfRlpZkH2qmZ@{x+X=5X1w_&0HXCcLY<6Ye6xSQGYuS!?2_S+8mPmcIjdwY3Z z#00Rgg=7`MwtUC+2!RRXYv=GaKoMOEiwDmeH;!gEj%A#$UR4&(?z?X{r?p{D*!$31 z|M2nA1HDR({>E*N|1E{^>MXQ#jZRzt{iPC(PP>+47sjGHykqa)rhTn~aHQ)?5fV2s zG11W0Hj%f!dNt`|4kst4#`0Bf{`CoDoy3E7bKgtdXJ}}61&eVH*yBd8OaFEicEaOXR_yKf{U@Tc~}bKqNh-V6W7fA-H~Z}ytvEnKj>DF*Q94nX=o1BleWP2mKF|(_9R32m6~z? zc3%FEi*c41cHVWUpuV%fA_fl$8XerD&F!0v=DX(fumIf zyYK}kCnv9l`1$zU@bg;}D9Z~|&)Wn<_1z-Q<<(;fmA?bb**}7VA|WABjBg2h?v#=% z@;9PeLt8*Hz{Oujj#h8(zU|$IaCjM>2Z7Vl-rhSqTO7cPC&p(rp3QoD`AQ)W&Kr1K zOBx$lp^SVbCt7e5+g|ccWPI3(MG3|%*Ktr*&UWC zV_ugE()Z)$SRK(n1{9YCPC1%5W_h-4tAHQ5xv3``zb_)JE^IurvZTi$})hK z|DutR0F2oay5HR^`w&jB(SuVx>oIpRV-_E^j)OxmI5_y~k58GBzHJVxszS9e*BLIg zN*JXf3VnCl!}IdTRIl!_`KH(pZ6!uem&eM5@Bk)-kVnt^>_)%nDGc{y+r4}D6kw&^ zFQq=~Hf<6Gm=TCceoaaKTFCcI|IXSt}MqOPU6H*3T3`5BA%5eGJ6jAurM!=zn+ zYI$?14_hV>H~t`5)0V82_y977!3x$4^z$7XRLj+&Y05OL;#f{k-;I6FFCy|5D&2f}8fv9X7mDUGF+S zb@`^Z_a(S`LKh=MACkMaa@}Z2H$>2~^mkJ6`1cA8!56FY}QV zqr+V0{{(!@xN!6VTs}B6Psz#E_2(xONN{L`m6eqkcBGaX3no!_XG1eaj^;^_@P^@0 zQE&KQTj_^z@HskR?^X;#)Gv4^EX$EPzkV&)XXjz*ZV5^21 zfsf&Rs}uKrcw{80;laH`!71!hZfuPms6Hvh7*e-XN=gaz88xA1Ys;}ZIvNlZWRT*g*te>xxq$mE-+K12XLvG=;2CrmMK5CYR;24+ zdV6`{R`WFT;E_kW#PpKYQfcq;+%)uDj^$yZPQexofTi1p?_29%;`jHt&gO7RyPl)G z0~xXhRYllIf9r*X&h!7dx&LcOGw-jB{W3TxqNb)sr0chReCpnBgh~G7%#4Zfri~l7 z!>l3!cj>mbxA!$XR<@j(zu2j}-Q3)kM2+r-3rp^rwp**(t7V-x&}VP)>ihAd`Pm4z z*{yr`?uo&E02u9N6koQS*}pb~rIXMGZ*zP4_l#GnO5w(wF z-On2s(0LxY4sOpSZ`CJgj?m+8xbcwq`G<$E!e20MrD4G2)~#=^?B}L!_aidGrkNcl zX_O-WT`p>{yE`Sor@VeYzxcuZ`)99n%B>6Cf^>i2L9Rx~#3j=;(@`c}uRV$MsPpSr zFU@C$n}vi{!Oz@4M6c1_@bYDEj8RyQ#Q$AYRchXvVfV!U)!vncQ`xrb#}JZCna9$g zNFqwBM5W1)%vu>j2$eZmhE#+!@=9b;3Lzq`qEJaHS~6sK-%Jf;iZUc5`@Fy7+xzd{ z`^WybzvKA6qxbzXtmnD!>pHLVIX2Gm}P}Xh5TSOjN_Yod-xp z^LR|)F`Ttuu;Ub#l8V8pl>{p2Su=NIgS7FFXU%6mZS{B{qRY4Xygb$L0s<846A(*Y zkl&dI4-$!o47kXsC1>*z=!p3rKqZfOpKg$t(sbpiR>J38>bIT8(B z_f`Ds`wuiP`1$i^;iE^pKqY-AEA-g?fIL2^g4phPb+cu6^YgV$@=#JUr!eLG41;4I zhlT_)&0|4qEWv8%5wrb)tFc2&N-un~ot@o57nd`DEQ!HyUH}Gt$KKbQ_(zlxRoJ`v z9q((v-a6CV30^OxdJKVM|mA z)D0ZCLtme=zH57v)s9j6wV38dDf%4$S3yAmdDCPkG|Mjd1Gu5+&RQ!()grhQ1$_HQ z{P^+XcWmw{z~tx*2+6_KA%iVyWf+t1tG~&GwVWUO`eN`TK4vC4RHA^%OXXF1tcU%G z^C%yfi?FpO!l8DV2WY(=|8;V6%d#?3vDmavXT=!ZFnd-U4|IIJ`ECG?p?cfwY9Z9`|v?l@#Jp} zD67XS-1jHFnuDsQul1fHd7gN|P#8HOd;nTCJoydcM?_345?4e&$FlOp3oiAt@29gm zL*S%Tu?h62=Bo`>(h2QP7GlA}i>4-02I_C!daCn8$ao)zs;cVk%*=kKqcloNz6A@m z1C%(qxNs>eD+@@i)px(k@mswW65n6&`q%ng53h%VII@h^bYQo+nWuAS;^Gz@IeN4K zl+P&)={a1i4{6X#s30Wvo5ST7jR&EO=da0TQ z*ub0u7`=tyeQm81rx^b{$nP{TY7U9MD1aX)#nwLt!w4z@sD(?vbn+=315OE;Bj>pSMKB2uh9TJ=cP0_p~ch*{k0YltsdUDOxx|-w`%~|Kk#hkhk4Cb zxba?&k3Wt6kio89x8OhM8on3h5|aOCcv)|Eb6X*!b(rcNR$Q;d*?nwjU(qXtrm_`l z)(8?Pn3w03JeWW9^J;49dAyno#CQ@{NI8Ludm;82=)g{>bi^zT578Oh#C-hxDzSaN z9-04kcJ}A0*hNqjJYc)Tz60Er5c~U>URv6lqqBlikDnL%Dba z^p6}_O(rowY;)|`v7_yApz7I$vS;Cb)r}Xz>oYm2vvMULi1+VfV=v$WdeF`7ED|+2 zi!QGSQKgnEw?L%hWQ}M1Avd>5RL;1se1d{IK|A5WQkE6Rt{%tO`rGRpgdn@E#?kZz zU;7)utnHDvB8<|_duG@A1nIRwMl44Q;KK`g3CycK;f-Jb*J^nzIpXLZ<|st@-Hq3= z3+v-GVA@92d$jzOwPmHKq+~SWn&&`8LPEk`IiPSISo*C!^ja7=%VSF`Ebxjj#zL|Q zTyC}jqvFts$jF?7ANyiW&HLu*6U?B_=VifT_0aX6m)j^RC@4rGlUrSCEH=FUMw^9R z`z}R$0X)egii_7lOD!lVDQQ>E5QkCiS?~#nE0H*YF4)L`AiRuhDVnm&{8v=y`*1cj zHT63zq6uLTbN+l4$lnJJ{?p?gavS})@jFDu#TjM#`Q1aMn1Lgse(YuKJ$_w%@SILASt)V7MSP2@!bM}ZRHg7xZFoKKYB7MXU! z4Z4vr1MiUwxg&744J#`v1HuIrHm_7t!e!^w9~v6^g7-)p#083f`_`D*j}_3)_;LcS zTgLHq{8T^2iO8N6c5Q8K6Fmvi_rD&6P>YfB=}mCMRi~Cg+w3(B2OjQ-6T!iVJh5)m zFzcNcae~39fu3BOWWbki-t0knrwmZyT7Bao013xpT$UtwU$MvWKOXQFr7q*b<+qkR zc@p>JNl4V7b<-ph9ugs=aJV)E)-=emtbv>D1BV1AQw@N^`RGwKWz)3x#x^;Ynh;WQ zEGBkrEnXBK6%0tu@?$x_FGyX+dS)xBC$HxoZk!mo7V|nrWGQzeO#?z@rBlSJLzC|sgXIt3F5A5lG;e-O@RF1`HbKPtCN^Aw^ zMQn!c1AXRsg0ks37K;VNzas?c+D)~&$~|V6@?Z}!UKbP;BuO=5n1JaUAF~_V(Dtsu z!*Tmv3@#x*c-r#=@)LK2fSk_0JPGin$yiHBU0x`={&1Fe$73rh2G*~8^H-^;#KV4) zJ=d&V5YSV-vxh4r~Jp7?a))P2Z+nZYnP3bA}emLKN#!&$cxNZsT8 zohxwA@yg2c?|?@~DU5pE80O~jD8zT8xyov}rJz>^`vH2Bk@p}@M(*0Z+7l0Z`ubMl z*`@TOuxf-^rrAod6xv>c<`+B?X7Z}DyrmWC@3zbs8~m{m=t7IkF~Xtg557>^UW+R zmw}NW#U=XzW}_BR!DJmwOL{`WQk-putZiCaTB4`GfInKKOb*8<$G;sy2GGR%FEf?I z6b{f9;j5@aq%r22*Ys=@8;3jx64N#~SH&2PCY}l}YnUa60VnhR-Exfi&$EV>8$1yB7eZ~`GR9kaB2NoAC7a$d0)xZUM5NCq35!ba0a8H8 zZuZd+YsOdlJ0O!vs5_iT&Ny`F(2bH3+jZ~o1ja@{xh(}Vg|vM(DamGnvvujJRr4`o z$j8T5KNC<%B5->tK6+~X<0=7)?7J%t1>MHcUY=T@L=3jdqAMon4Lkg?S~NUc*Vx!t z=>AT@m`M}Il;+ND+c=^4w{WsU8&V%<+Gn_ZgsxCnvtLD5`X{U66%_cuhRNs?J@A-79^{(C<>l#l{iZpfdiJz z6#M;f^xlf59_s2_lND|!=VmuVJPG_k3DFkSlIW?1L7rBEsiI`)Il$=ywV?E0Q|ITbK~5=D0nWfMh5qD zvC>-(O7H+}Oio@tFe_(d%&LWLJQC&&$Xb3z7;N&jZO^f;iiO?7?%-f|o;aaH-V5w17g6<~e~nSR`uDy?H@1(%`tLK2`hQPk z=LQyMFf5ws2bdZLCjP#@tH#otF@0Kz-?ARrf~|WSoa&Xt#3-_zg?hQdgaYd(a!T*x0@V?(P6Kj*l1LVG7%MAHGJhMe}DhGwndzvQi1nF6J4xY zU&5^2W$w?c>lz!Hsb`|J>K%WxmS4ANlYV*mMf@VOLbA37D+=zOvc&HXVguDmuD^9N zm_OE^du#5jTD6KLdTTnXa3GuAVWewcZfElJ5qhc6mDF|9P~@m2=2~|?(wf-sM~vRj zEx1#jdgp1={j~45M#9w8)HJ|8zfIk^2sPy{J3BcT%SM5J=0J{3q8_2(a2Xib|2wMm1isw6aCEPl7-yhB5n~ZOv zN}4HHHFE=UDH=NuJM%sxT+y4roiw1IW%oE>kN-em@O!S}8U;0Tz0v8{SC%pqU&={J z^4|EQEN%S~=KcH>L($dl`pAKl>BBb-GjlEN@P0v)aUyZ%1m12E9>^9L2uTgV#9CC6 z3Frec6zL?ym6eqhf>koq?cEQ{w;iV%B^v3NG(!be+z62`K}jcRfwA?He_#lj{u zSzAH9kIB_z9^bb)MHo{24)PD-a?@GmAV?x8SfFOE0a_u-K`_=n7~!@YCA zzv|I0#GgLR2~J_~_YYr`9mwy`8XNDMC#3}rwTY}*v!?BV$0E|o%QyOpngbXyk|0+9 zQYTlb$K}npK?-*xO4FK}-DN<2Q6)3HhW+1+OlD`Y9Rw*z`$$EfuO|lUbF|z9n5_rU zxx(_7lF%v5PR5{GxyfDNRf|RMQ2$(pAwZ5!Ow4$WXUFEC;nZHuU>5cRFV@wS$HAjd zcD}{+N9mdQP7B+CSA$>^M(Jdg{pa_X>0LUDA0Jw3OX!JAy<1zFbF?AQpaO(sm=M#s zw_BuP=8UuC!>>0mCr?gKNjV37l;@v0IR0@^q^Q>3%N6nIlXJet!Qk|J5382}(n!$) zR)SMMi?M78j0ji5DnP)&)PwvHB&#og=xWC ziL?t2IL;tfZ=JV@*)0pc`mrgYFtJASId0!MSSQ{rDUm`)O;{uV6Lf%hQb3sjO4_;5%+EAhwgz+yMhcE7gK z-;MPN-LqBTvc6*};U~LB#*g>g#=H()| zVwm5LTqqeKicu7s2nA3`I9YKwvW zaR*3O029=zp+_$%=#pmkUB{;iL^DHp)1DgfBb{Z^pvwxzmv@7td-i0cA`Tiu3Q?vS zBwY9K@I~-02wn>4+!JWkP!SeFJtKVvtb-)fHJZwmSUb89Ox5k%w+{k}pMWBGEqbg} zkXJv_kOs!|NxCj9EZX@~@IRpS!E=;lD9Tw%<3y|n@md5snp9dKKOvz|^q&Qp;{v|24K-HFwdT%F z1%&8V=vn;$`|!NBmNb#2*0~w86nnQB@MtzR3B@YV2X_77nKv)-iQSG%A=WvTqu?T8 zq%FTA$I=i#&J{Rx6nLvIETDK=8~(7D0g+=t%(VUU&pFJ~8>eXP8N5;lK9dx_zW)A; z(A8k>;^h)XeZ!o{&d#m{fAzAX!yed;sH-&571N|LwsUm4c64;$>&hDktxvj7{=s240<++Xr{}uGje*0g34+eo&rcL-0y>`y zWoTK)V~u<9k^RAH>b($8*Wx8rb#^)sS_^Z%f8k#ZetA?_xR0oH`n#S!C4FAB#Ht`^ z1Ec2(ZA@ZaRLpCP1}~?`zjSv%c$gPKlt1nc`!C&G;~kRM*U;S4GvqLkN~9|5j^b-H z8RLgNnsXD$6M@x0K!4*7vL0kY{##qob_Xm1^uS`J_jxv^03(^;nl-UE?hRHS1mA#2PM;#44MP@5_5zQFK3W9n77n-555;U)=-qo~o7>YY?Sh z&?{(`(Yuo~cmkwN<|lEHH*9tyVyX=pYe!cduwxl{YM}g>uBy2UUH6wWSP($RftI4` z|53EjS`YO`0(2NgFc`&gIfsw3+WxWS8i%qw8?N53c2-hCpwR?9OxPk+6XZO-gt^C- zU)@~gQ|Hg0{|kFVd;19y~vFRMgN}87qEmTRX7w$NCG>+BCNDbKA28zA0%}M$YL^Xv) zMO(4M;vm}bP{Ly|j*>FtFtL&tDe82VcJeZUOF?Z$*Mn}NYnT1*lSR*?cnME09R!u(EtDd literal 79531 zcmeFZ_g~KKA3uDyNNGt%(n6(>P^c8ij)s=B2Z?rR545Kwl@yiH(4xIHj8LMbJwzHx zX|MZvxUTDSe}A|i_djso9*^tuiOxFT=lgveuh(U_Ng^EHsC4WTK z@xgG1gZ6oi<&|TKCwYgNsD%$$Tk_;3q;X8A=jVm0r`k7~`6lG17pCT@vt^q#7Ur1c z1Z=WaZP!|9aks=cd}6**Z+t^I?-6<`+0kYZ(T@c`I_FD#zipk4ajA`<=cD!e?@!R( zQwA#!%m4f5F1tYHf8TkVBzEK2f8RJBylMykzkl|aDd@BP`zK7e`3sy(`Z@Rd~fByYT3qmD#69R2ChDbeAvxkn^7%s895M(e>c&jtgt@ zEIMw+#_AaO-nqPznd|VA(-E1M>$pprQZ+KTC9ef+M0E?v3eUlSp+ zY4>S+T&QTa+mNA|nO2hd9iBK@kC1gcwFXLb7w0C$b&pzGTPGC!_N{f24O0}COziX@HS#}+t{?&mOJ(uA5Qsgt9cH!lU z#uT+%1qBDC4;>OSZ{7F!$&=*9cUQPG@_xKjk@5I5z{Y*Z`SH3uo5cs;%bgcy#C(`G z`P|udrsBhg)wo?1nFf^&t*WOE7Vi-eVSIVH?)cXiCVY|adqzfrA9^fZ_#t*RJR-t- z+nIZh)YG&wGc!G7V>fH%n0d=CFPu;7?0$8NFYF|*SGhmC;JzgRr=IR^PrLy&B_T0U z{N89+8Bdy44xZ5F9Xoa?6({SllbzR^SF&(Fx?C4$Wo6|V5y7FAXAz9m zW!ka)^UJfIzCIce5s?E255B6aQ@rZ(!@QY6LqkJTQ}dB>;&I6z)f!mW9hESolMYWey#-fey=-+i%gaP*n0 z$7*ir4M%-eQ&CoL+#^3SIhVWO7<+}CjqQeL=?pc8h`N`zHyzjJ&GcC<=$^L2Zhz2 z`&pJPuG_wSJEy(Fa0?X=4^Mt!VMTj8*ON1m$J=tvvs|WsOsTu>H;%OEZ7H&Uy5QK%(`|P26MZ&Q_G}%iZ z4r80)S$l2Vdnu_YL#HUWbqc@1_joX8JRaDFeWt92Mn(pgE=|=&t>3BT)%PZpywPT< z>s2>sR_?a!l-4eF-H3!>DUXf|j^Nsvl zw$vw|7gSAEXWDPkUXBHu>9Oqo;Gpda4q??#ni+}@W$2lhjul-W;f*+*W!@$cv~ll> zr%#`55HqZjGFm5BjyINa_~n(Cw;xsHmPUrI0dB@XOV*ZqJ1_cJ_r0c~6kZ<**ke+o z|NTSs*!Z~0w)D=Wn1L-aGTcAv9{OUv2&<($&b@}kip!J6ld8|N)YH6h!Jwi0~kV4GeyKDEEx2|R3 zk&%y)F#nWxalf&#F&6(ydIpAyyW7v}U%kr6N9*e5#>UA>E;+%xEqAw=*sDrkmaMDI zyH<1VqsLmSJ-?bV_FlA1|Ia`@Hr(N7=VH5idv7wX-_`BQ<9^u6YWK}sx76HfQq)p+ zJO44wFDgpJFmLZeW z)YP;TG5vCi+661CObYL%4^i>L7fW3WD7<*2PqTuPl7w=8$jHdZHMc2mTu=Gk;m~y@ zt*^g-TXAu5Lk(WeN1?a1wY6THvsZJvbm{5Lw{S&TN|r&T4_9c<$ObI`KedacS7SqE z=1x*rQQy@?^h%s_D7-RDGe$W!CT;t!wJ2@fx;3fBva58b@t)`;)#N)(i)H4OK1|e} zW50j1p~0k#3o6FZqed8*nF(`CJ3h({dEsucy)9F>#Muxlk^I+NG+N5O<#-V5uV23^ znwr$FwPv3(kDZ?$WsFu3(ALqZ#>P?WC%?>Sc_gn(wcoOo`}@b3XSr$E>gp6$+aI-R z5hj-}2ckHv@|WkwN_--$A+$CD*Og-a)7Ev%iI5Ljy2p?E2%Uc#l$n*qW?*1o_8u?W z_E;qudmvG(U{6SW;)%fF;cIHMtvOe7e!aZ#@)ETt+OljzO+J|#Vjib`@mgiobLVeKfi|B@7W@1DS&J9kv;Rx)uu zwhlU7BSFv1tRB&kZ+-Nw@gC)8=QgvkvnQ&1{Aw?t_SDR`icfithKnYlAb^7 zNi7GOzj%Rzm6>nr=WuKi%1C8-c2;KQC!zQ6jYlmkEM}(KEGKm>%RQ-*)YI+{7AOc* zq9#x%L(Lh@E-s^e@4^KSTp!+m0-9ni{@#ymi_(+R99)N-Rfj(1qG^)5Jz1>s_NMuC?r@ zlzTI-S2cO2`0(|ul`NmKj85#k+N7YiE9~U%%e7o5pM0K{>C78JBf}1?!hYMA?qh~6 zTWI(5Ce2FbB&Jvz3Odz?hq62Z0?YHJoG7zce63=Kod3|{mmWtq`1JC!Vve~&^d7p9 zkPxSkMK`_62kV>Gx3sjFYM)9@i__lao0DT7H(la9X=ZFr+!LvXj)!ph>$%U;l@B$a zL6tzW2?Pr01v<-joqZv^>d`fJ^P{4h#LXU0jG|NQ%#mk#)X9Ci=3rKK2{U%RQlZ_I zvh0h}_?A8G@cw>X4v3yZ@lw0PE7yyk&4d8dSd9eD0uAhHlAj`$1LJ4BgEoF4M=+ z`s$xw*xS<9R;I!Am#jHe)2<7PL)WHIcJ11Aut@~q4Q+rdMJ{PaR*!|>D?OVv>^{UD z+Vs$UE?IZp&Gy_DH?;o29UfD|7vgfGHEKjG>^29+#BB0xX1`GOv9|VHqJn@+hj+=L z`hs+~q5HBz&lXEXf%#cjSb{jk8R{O&z8U`dGRvm_w8BFf9gg*niq6_C#@e)T*BE`C zloV1;C%ka=z1qfckSyDnUWP4^d(?jO%tNHGY|1Ou6>5s;Y`>S>QCb44Ukl>cI7E930*{ zMc3m2=30szL?!J8ZvQam{%4m$5P0nCvAnn*0OywWt`lG^HYIV%$=f-_&Ag)bUN*#T zXO+_j1xo3!Nfl=3eLU8Xe4f(;8$sU8ETy%h<*q~Gm+Lv9Hc@SDXb`$$BGV4z-%bU` zN6FEvJs-8rlhqk)NWRt3p!z`8L)Yjl_8`NxGD{U*xg`c6x06p)8q#$(frlw? zOUM3BRaWE=dvrQNIEC-6Wiz=M3o$~zYX&Rdy<3IPUkzsUyZ!`E5B@*jVKlC@jgJ6Z zXC7!D{htY}9~uh$^T$phMuG{1;zQOFN^iFAWJ0mzw~zk!Bf^Dc-RAc3Y;i|hb*~KG zv|qLE<%<_%;7l}*hmPc;F+}K_n+Ib{rmWdn<5;%5o+qbbS;Aj+VJEj=Pne zMoZVsG2@_2&L+)GjiB56;L-Ns&ZUf_4Uz&&+3h&iEj^ltZcOOf)M!`emuHHpHQuZ< zb3+-7TjRexIkVo;+1c=SZ}kDE@yD;p59Vn8>F@9N1~ltMy=^IVD{6hgm3n)1?hJyve+keL0Jr7yWDjqL`OieqJN^Viknrp_jw2*s6FwqzOQ2<+_Z z>uX%2=dm~$`MOaD%`hdmQKDY@?{YiR%cUuAS%8~w*84LkXf>KQ|Gs@|n=|x$RfVf@ zTT}0s2?+v7R@BKZ2LYKS$aynl=HyhPI$C#^Qyq|!G6Ka&Ddchc!C?d3*0OL_(d{Llxx=O%~7CMSfB7EErLS{k=LWWD7dx_{JIbLegO&JP`glLNW8{~4s(PB-%S#`3 z8&=Vi9|2A$7nkKQs*84*bVqvD9QdN8z*cZ-`0Fh+`~=h|$H_qs!n~uRHtf*Krthzb zcpvGeQ|zcb$+U5kRh4|>!iAcx21n-RN6W;neBMeJL4=5gPk1;xP^>3* zL`6jfHTe2L&*tzi4ns{$sJ0$TCC>Kw1qD4QW-TQy`%yg1&h4O(23T`%XW*kpywXnN zwDyAyuSnM)ZV}72?9vslyQeZa*!X**PPXr5oy<|}Ia1c|it2HL17rD9oqKTL2Ixns zg-rL4A1lwCIYY39uC6Y2RTa>O)x`I+Gqba=zI=IE-u74OIT_5$&*#P3)6+xk*;-jq zAqRF52rPldy#|UP~odruc$piIDJxv+Y4+Bzr77haGo5j=eK!UAY$ z<$adVBjYOMvw9P`f}R4xw?9#r$eARLhT7Y)cjS#vgX5wwP03j z#myR6Z3?u3^XLJxvLH;QO;X{@huY>B{gz9ZJKZ(4v;wds86f0^J$Ufkz5~pZycTNO zmTxu%5R`m0vrWTD!AhM@Z@!ftr5Go@T`Sv^4=BznF)?xG&lkmN6bGAYUG3amF`-Z| zdi{(QAKmJhUUeKjF3p`({-e* zM_R?}0Vk;rS5;RhH!fotRC_E<@PMMk2WV_B)4g0H!FkZ;amp|#&)@gJ`eDPuCifVG z#~&*r5EU$C@Q~BEJZh%eKC_0C06q-lKL7l=-Fa$wyWG*E;!~5|uV~0dBOmqo^XFA> zcszFO($0T0)#=fOntKej#oO1{cel7WM{R9w0?3FV+FUOxdWti4%gcfGf|n2m)_ni| zz2e=wQ-u9J*DcZN6F)@|nlb$vzRvr8K7M{lel@kV$I)cN?%y}N8(sj72jIEWiW%Cx zu+57XFIYEkPE(eF{2>j*jxHwds`MA*sp>n@L4oHu`1|*7petIL*>CI+XSVF!%N`#e zpKsMejpbnd^Yb<==k?d4{{GBeMWWy*@C$C)LIaxkCd=r}*utzWG`UwD9i%Q)I&`^B z*A{uL`Fj=h75}HkU;F=J9n`xh3#}!X7jl+f_g@54LonEUd~XKemT5~S-+_QQza%Rr z=vn|KAS*kDm6GTWx-T4n^t!OP$SAdze-AID-hmQYI=UVx5-Fm7mdbMx)0s9@Vx)c{YXoif7afac^p*t*^H_(0No=OpJ|P@SIm<^0+A1h$1Oh-P=@Oc}_VQWC z4s_U~AR0a4iZVwY%D55AwBy1HTGX5aKrpyfs*?~Jf>5Fpplw-MTOZNa=asY{^g>xE zulnXu_6DqXZQ0_~Cfv`J5EIkpEgDle@KAiQcM}~(JIP}Si1teo64EP`o|_(Jp|HyF z3JWtppYtj!lf|Wol9r;9oP9g-@w7c!_n%)Kn@*iNW!jL$c06bU72r!|=5A~WPRKUZ zMAy02l3Bqle!^+?SIeU#H0_1c>jK=B=^W*x?b5Xs99V0?Ae3W?nU8sb#=9CS_VTURS|R(@|CW+6xdmU>+Co` zZmPz7oQzu`g%<@oA;)#@&(D+>cr+*`e1M)$h1)U>Zc&JSrfLF-l>AUdUEOHxyxt3A ze+q91*TFlfj{bf@I*yL{cr!OQ@q166Jn_eUq^I-(3VE}}UWe(hh9d9iI3qRm-OiQg zzQ@w+iMRbW!91||?qAz&lpLPT9S8FOd{iFmw3;wG&zW$8yXXmGUffqVC`Sp5l(xgvvGc zaer4ih@-hwjRee&7OA`~Uggk>3;>d`j8b<3lW=lzaj~+pKDwVMEGfwe4IxRdtTZKV zd>I$59`O;IBr7XR6Wb8DuQ4@pFv`y_K&n7H1pS$cl8}(F@7~aa({nBhZcrTX>Waoj zj&sL0X>zr?NnRvNqUNb6lE?q*&ufU@xVwLj#0|zz`tc9D45+p4x{pp3tnddr15*lx$E!qz}e)h}= zb${UVlfc*GxXVVUWMEpuEm<&yR!~0N7iYwy$0m$v8nI~2UWLCvLsAz3;eIpLQ;BZ1 zsz96B!NGxe1)=zP@et#lkov&n6 z0+PXoiiAeFm-Mc;#u-JG5Ooc(4NXyOAb_lZa@Ahywh0aUHNc}kHqT1zch0tp1vcB5 zxuyNE+esH7=-GYlryz*0J~VDldpvPQ{ILVjL`8R;8s2#Bp-eEKb0VmblZ-^a(gEee z;o%}>uDMsf95;%F4DLSLzy84iYq_EE{T(GP>mg_OW@Kcv?mj}z<}K%#NE-JVER{n+2KivI-PQXdV6z+<|!RDAH_PTnn{@^76-YAgK;M zULN7Q;4~{85*n(pnEyFLPZku&aHuH_?4%lMA;H9I9S|fFuY7rmRz&PHm17WVkKn8)Kuucgn0mnMMYH=X{l9^=y&63-rc3cNg-ccYs+OKL;&j=TlP0_ zKJChtmCeoT0sY?EazPI`1}zC5NPU6rW%5(|Ejz0KXUEWJ`M}GVnVBh`Kr`&bGb$-L z1gb=h1tfp$7y#aCLeucfZQGlnzFW`zxk4OIm{#={MoHs=uuU}o?rwd;$H;w0Y}~jJys+`vsfir`GY?(ISAk*!@%u)K>KQ`IAdH9Y9=6Y|8c5`(86q9sV=aJQ{1Cqq8C|DoSX0WW-y0 z*Lsa~ZRH4&wG70`|8DGyBOy5`1W72+|7h^|k*xgM}2;rd7`zlg{gW61yLZ~>^ICAM9bgH35M6kb9vnw^1l zP9CgEN3Z+Oa4^J%Gz1a++yC0zwHKiCn+<*IpU|AWuoq@bSZr*XuwK#iGzu>y*JSms zH*Zc*SRreuiaeAszi)QYX|o&%uo~52?8mc-?pj*+Po6%#My`0tHS*-v1h`S;lG1e5 z)$jQQ2A)5p;hwv4<;ttq_U_)T)F*Y;&Mq4dngc{{4T_vfYMIl-rQo_EIDL#1vysK` zxt-gDgoK*AUX?D*UB%dTy)mJ_sbDgl&+C7jPpcp@GIFr!ZA@1zDw-gL-o?d5 z%%FlcUP(yxz0CaZUaG@jg%c$qF@6l(zN~AM3Tjp9D$({2h07m4@?6cvMQw zwl%BpJ61+3Je?e0$V=1AqE=T|*UU2X9uyX)pb_oKnC*rRcZr{HBNWyOHY>l2e`tIp|K129ryOCb-DKA58aM8<9I z5k8;RsB`MhOM-&W9Da>LS)n{Ce6d5YMiPW3rwCE;#?9^0c;xkb) zd@l&aqA7Sz@i|)73}vaMYae_wvd#qPNArd8s(U*wB&aX_`FY-Tek5<^SJ9{vnh1eE zs4&&g`iMb~l0>Y25%uSvTAuE?5FJw_x&~4)KmZ!*?LjAFOzH=FSqE~M_eOGC|{k#C(-O9c$W8gZ#yRzo?lWHldN1(uEVy_Yr2ekDV z(2X(xV$L`Jw6vYAx#odzK?uwo`1(?i7#^rcgA3Q{(jYS>FEvU;s%U9(BBjB$c{9`H z%a{KQHZnr;mq$~nf2PRe`ltScX=m|4_?=xM%;r_girdF7RlG)D&~yDwCt=a;uwNKkHiR_}eh}U`;2;%LLXhG z6U`f#Pn0&*y$;KXX^QDu0D zr@_L+PM)k4bt$k%4^T~3Ikv@m9LiGKDpt4cirmaRG7pSOZXU>8c(jo4bmvmp(%hM} zbLWshf+WcT5B_*x^Lucv2Z$$uktQj{->fx9SVF%&-oX0G=9s_$F>9mA_J}UE z{lTno_H5wfr!)Ng+e|*|#w#V&jQ33(5UiQK1*D&rwd3Hy3nRZ?jGp<&58ANz(rVCk z>h8%7s+{exMo2?OzbS8R)k=#SXL!8d_wM%XzS^ayW;d=~vj&n+O?)seQUr0upto;N zdinT#&p%fkz)}DB6uUH>8Il;dFbXS~#E9^O>QmH2a!fW$+Iil(MFX#r)?1Dkx|N{n zU>$XNMj|yb4SfQ8uET0lj)PO1h)$R9_&dSOozBhf=V!X>*RRvBUR^~@rG7bj;y|I^ zZ;*BzlgismxZ>l-`c-~xTB3c-V1GczwwnP;)?-~hn+mi-r6f2ct*mPBYD1Fp?bA<4 z(m~v3OR1ZQx^Bn_L>#Rj3pJv^<~QKdkj&zg$RE5H_}B{4n&75{wMv3=zsA)(n6_ zO2o5--0*`{j$e3cEdw24W|YGy|7!iSlTOe7nqb+c@%LlK&Z43?jq+TnsI0UGEey8l z|B-J~a|^y_qx4geDc)XW;zdFriA1otlVPSS+pe;`nY9qOz;CBrXwmOGg zXINpco^W(LfDQaaj7JmtTldeO{=`}P_AZ=r{@sTU4fh}H82snyJ)ps!cn-m8Y^~kBYfp(zW_qAD;9s77+ydD`4QoCVVO6}RZmlU_0q9sas`3)ND-l|ogjj3nX zNF(zx+)=b1UUNBIA>|C+Lj)H>^CO{KF?=`H`XN3W**V0HmGFmTupiLR6STVrks%P; z*C1qL)$8?4KX};Vav_3{+*W~^2!BFCFLVVncEuB*myk~2M9cmB{Ny*>)|7kw%|EQq zsi$lE1G^X^1ls~-l*AWFLJUuqU;!kQB7T_b;ew!2RT5W#xLXiHNMnZB#+0m@a-V$& zeVxeBP#k?AF(eqjyW{8Y|0?c~%WVLKspv0B%3JnXv~PffaJ27raDBotMk2fus*F%t zCL(J&8??2x$KZ7JeNmJpgq#FgRMWLLUb%9`5qVYupass}PZDT)h(1b?E%7md#eu|$ z{7T|#`2Hgh{88y&AaX`1zyko1M1aLZO$6Y&{mz&kdN-gxt~vmq)U@46an**Fd#mdSkopg1C+LaAd&Ruz67 zvN^`61R+d%3O^W@(N}~{ND|Tu7zhl6=%jxmX7Ap;r-!B*O?uIRy22j7xklzZ#10;^bxk4juDm;HZM3SA_~NUpx6eLPU__ z2hp!zbM*h+u1jhB_;WJs;z2iuwY?r)dSr2`$9-`)renQbPXp!ev*yz}f+xc1!+f)3 z<>i+%AV?f=o!L!71=ik-VOas8p}r8%z$=pr+1G=pA@naw+@{ff68^NY8JDu$eAZY2 zoaQERPT~#~Wgh(--AK^Z2TJKYgASa9x~l7T>{tM4z}H$l5g}i9UFaXzi;+w2_R=`RPp9zo56MnWX?8DN^?D-|ss9P7tbyO5;Yvg`v`T;{kR9yAufzrHE)O zDPN#J5m)C^{86HK!cuRrh>VGi4F-B9*=tf6UKir6kJ};r$_$1 zDSn;2BEyt{9Oxrx{xmiPT=;EO=!;4~m0rHSeQ3at4T4kB5k5p#fhc(}?|X42f6eAT z1RYTA`oPkN5_<9^J)Vi`I}r*|{_&+1Qf#Uzs;eL#Ch3+G3OmtJKxbB}~2Ie|A_ge&!&g+)qLXe|^59|+nn2MVerzVNgucwF1NnDy~b_K=~I zU!5g^D2~*QZuWOk#>+U`ZWY**zEX)hxhKS07xs{Z{z>JR~g0p%;0SedN^6VzDw=6(4j+{$*2zO zE<{flaba8sZRB&Lp)7+=DQ$a(wYCy%FNerTWJV+99I!zcY&0y3)Irb#em=epgEzI9m>9+ZdWUvIVhwzG z_7VKB(LLMgqPx3<@~^hN``viDHqDWzo}c@*Xs&en)#zRFS5etEvIuxGkc$vLEJsiF zE%1c*=F_EXT5P;}#_%`GfK$Sf7Cde7wKD{*;8)i( zC6Z|4*?L4L=S2^=+VtEhV}v!#TE8n&>hEn-B7q`hzN}0kCXv)}y3&V#rt?2t1$fnj z-AZI~gqu=#GUAr_GH)gTh{)-@ycFas8?MgRuxF4w2_WOx)D&TPyN%znmKZTHGatu9 z5gDHY$d5Ou^nshd5x1-XHC2NJ?rn9*W~onQ9v(y}Mk=KWq8FL4ESXL<#;_8=E+^ua z6}WhP6B7_THV#4C@071!3nLk+gw_h|7NcUY*=3fz77^F>#*fLPu#h*;m+sFT-NMg* zw1)%J%a0p@@5MZpT!9u&^wFL}ejiCw604<9;0YcFdLTkdvkZyZLuo08K}t~-yye6O z2@aI>sBlCAbTFt=)VUTNDLKDjr;@k@B_rdeX;V|3q6hqe z#2xUwZvpqeLNUa}dqqcY+-muuupRrHmX`Jr?pMkSP|hnD@j^U|{1Zk~3AAn91BB*^ z`%6q;1QsJ5Uaw!b&WA)L;LppU;x*ozga<$(V~CQaTDweRZ}KCm1#l#phZv+WNtD6J zISlB^iq;Mya8w{Xndz(3MeOrWkX}hj3omf|A;uc^yS{PSZSLo@-D1kiOi^xKS4rte z;w7`8{LX^?x9?*)VkC{^?Hdo=pz#FDBf|rZv*S-vo`dEloqrb6+J`EhZ`04<{AVCO ziwW7zY?%FTRt0k~&`}Nxg!7@OAXokO=gHI|;5j4`5SOr%pjfMcp=0NxrJeSaB4Lj# z2`geca;U|8Ugc0eY@}ytNV;ju&Ye5oNXqk2CpS29&p8*3Vgbd}|8YZThg|Z&Eq*pR zzY6A?0g=crV^|hcp(j?)rt=t@{y$Jdt8Dij0f0(|l#f`bB>Votl-Lbc1=AyYc|OK_?c)0QliHY>X|Sv`mgyX>)bFw8voiWz$JueuZ_9=P&3H8 zp#0b}=BINV)$IIN<7?;Qr2Wa(g!)JZ8VUcy=l-5w3+aL67ReHWED*KFgaP|LV#w`# zL)^i|&DlLJTc+>2>4B@XDJWLY+CLq2*_AgFt3n|B@Bw$aOY1|sWglkWxEee-+Ht9R z_GO%IA7oj6nwL2_-r?a~Vq$@}<8;SJvs@p;@*AudGxa>g5X{G%k1wgvB-Dv)al&Lx zTpj|ii7-GhB}+>zHi$KU=QYt{8|RE2aaWV_ofi1HijkM zo23I~A)+URjCVjd=U=x=`z`IgFhdluYJdKVHyQ&M_}nJBsLK~W2z}6!=XqqHvE=VvNfUSPXC+VQH2ZAaX{yQpTfMy(QE?BB6LP(h! z^@LC=5S`bOuq;k`7=3s-&Yfpe(6;ett5$ zEfo#Zq~n{#3nmdvCpK0v5xIh;YE3R(lEX^Id?Bu9H0T)p4bWn)X~jXnW06|(dl3ybjIeu?Wjyk244p=7)IvE=BpN`x`$V8e0pTgn%+jH@wys%rqu{@{ftN@SFd-Uh;4}q6QXP^mhC?s<`MLtrLNAQIA7oW{RZ&_9GPTkt$;!}CRRzCB*Q4yL5syXF%r>3*#o&#jlh z6}&+U94rx>rJ+}(L`9cJRny;iYjV%82xA$<}etWWfPWhBU- zV7O}wja1~Kr{EPa$<(N5NrM@$58215!uGXp8xIJtdR9J)k^UVj6jt(mxkZD zlL;pTft87p+@tc3GxpDyCZ8Ag%_q?DGruIKRNixWdFPZ`VVjtM4N}gupH=#>kU6f9 zmA_FFEBmvN;qFtrPF!IOr0!@AZ&2M4;EzB+ugkVf;sF2}kPIg|DFGrdaSn4N5t%>% zt%mWSDrgO|T@CS&WQi7_12}aq!}bA!G$f}Hfa*GRGP8xrWKX(b@>MCd68B|bXLNM_ z=4|Q{y*xwD@0gg~;#(~BP7B`n+3rWaU&CMKc=Gf`%}2!%RIlfsJu90Qz`FZ{yabHu z1iT$qV(f8hMhe)w4-N;>hi*y)1{};l#Dg{4P8T;X&Hwp%3tp8zMj@e3R=b!7`mMfJ z%f%~oaoUDYKS7BiAmAyVP}Dhp=k$$?vZkvMdGnfz?f54r7tz@CQ zxBptTA$Om3OeMTF`Fr{5+YYP*(CimNG@lGv0v1qFNH7mdHbjsl;|n zeq-sJHdZsNgd3&@ZYdka24X>7A=kogovP|F9S zKfCmDPsrb8*KzB@mdS3qj7Gy#40p3*T;$x=;!hX)BZAlaOl@G=boD9!Ne|aXRyO63 zf9LrCVUoE?@>pO0QPUg{b0v(w?$Oauq>AJOzHp^kfY^9>(FTtLEKzV`NoL??jaSw^QFmQ)?*?*FyZiq`VYF8ejHOt2W=W2bJf&!TSj%7#! zZZW}ctipQ2$Xk+dXkZ|j@j@g%1cRJOX?>W3tBsX1LNam&eq^>q9*CMF7&#Fop?9gx z!p_t;G`xvpA53*JP;*H5qDkrop4WOEH$8;dFxGQ}1XJOZnhjx(B(Ov~j!`y+F8M?? zD2%+{Ca`;ZPoQFa-nWkHoHIu~L(PJp3Jzkd!Q4tT@6PNVDj$ z08S-?_GCanc43SLva1#1f(mic8xxNQ-wX~8u8ne^=~;;x?XyFTO!D%xM>M)tQc&`b zli!D+Q`TPv=C;atYl{gRZNJd^3){YSZeo{Y@hmy?Qsb_}0_|3Q;jM})j7xE^oH~5v z;ywNSR}lmQ{rmLQSVR{LUfra_sW&%X??K)Ss?eA{qF`=)IoNUulQ6 zdZB>kVw}z`xi?I7U-~6x6hAH}TsbsS`Pkp)48Iqv_hwoSQD_61FJB&&lcT<7Zr*kE zBBU+!T+~T#|5eYfy=VG=#G@qrUN<*&%sJ?nUh;T&`@2=C5|%!JfoIa2ue9av#DO4> zCUQsqL!?J4hRSWx{$$u?e!md`ELr`bx|$B}-{dSZ zbrZ*bAT*6(_*YdV>;(kyU|o}r(=(y zZu|K$+z1Q|)YXvRSVJfn_?E)`eXEep+YWT-xB4bFM%WU6E!XRGUky9v%?deI@?V^d zl*{ftF`!8#Cnxdqt9;G5s$t-JSl-x*fc`(>5pedE3kgsC?*Oy}Z70W{8<-y)q>VQk z(7%_wO~!0Db52UkMv2}3=c^oQ{DLdycAmL?|7p>=Q@-f~T_4|D9jRY=>wibD)kL$l zlo>E{zux-ux#1O+H4xZtczaXzC9tgc-^put6J4H7^}a2Ov}fMD@5ZY8`{OV#ca~|> z2Djb+yCD8pmBcNX*2`v(pAMv-uSw=9oXRlBe>8B5AYV3)C`d<%@CQkl9o?2uip%*u zI4Xk2Kr!kq@tDOMo)6jtmFm#;wJk4AxVhg9)at(LZ~Wx3)|Xl36-fgnu_E~012JQT z6TbMlx$EE4m2kjP_&buCsFRsr4U5d#{4(-tFwQDSR1LYKlD^&7|B(f6fIENw+;DmC zA=tMmOcCglRA9|e_Xq^U6b2dhMEk}`9`Uxau;p+fi8Dk&rO_FL8ZfR&4^Llb)IO)@ z6dwn4HJcM%rHU6&B9+;;(pp&P+&G_?aX2h3F~on{{oEPOPL62ZM>n^@Y64c?r>v|z zAoW<(@Q&f?7qWtz1g|l-uLdwJX}XF%f=&>IgQO6`FvNhdUuftXMBUUk z&QbaBurN3{eWgq}X_fmY-sEwo@O$Oky= z!RhBBLM3Est8rp^;fL%D623&TPpau$nE#&p5vO_qP20FV@VN^^H#$h~{M)gX>3;6{ z(uGL;!};^IJo2x%Y#)|hqHpuncvm=d*M>VNuH!V2(VV{6vD5!bi>Cr^6nk?`bqqsV zwB$RXpt2g$;$+DAU6GYYEkgg0L&YUgc`Sg`MgS&x$QSz1FQL5sjq2@0`X}G+=Nc$R zbbn8E^PrR?o=)az@rMk};bEASJwj4#0~${?s0XHJsgNisX}Xwi#fQPe9-O`a#g!MC zI5k1IE11zgf-nb8?jq9+&@Ynibrn0Y!#2d(BI3j%gLKJD2HecWF}U#jV&1t)Y4{0* zI79l&ZgM~c#Ay`8B_;KkOD1A0@_sYJFYB_K@{q5DkvoXQH!QkDXBnuaaJP;oek3&> zY#?iP=zcP*~6gF=}Lvj^m^vGWbXWx-hEWXSuCd zy}^@+8Hkhlz}6@8LZAU8nq6R~hxSTN3&9~khhc&szB{?Jsifl)OlUS7Xcr9bK#r!u zx64sn$i#nI+uO67NFEM)^(&bApK#I%C{j1>DMgoMGdX#W$hHI=OgbRWyAin73%W%j zOg2KiBprSQA@2CdusR@R(F2SC1IujK>{yAI-vJPoz+qpCYJ$6Wvl876Xrt3y-rJ}) z`QY28gTH85Suur?;oB^i@Dhmk{>k6=6B(0QCoXp7At8lxHxiZROHl}VAZVqeu7xsL zG`eYzF+I}ZiT0abH%;pl*c@wY-kvqO(_J5Lc+lTQ}5M5n?uG)9^O4w@eVkk^mpi$axro4LzZI?hUvs}f2T{3L-L@fWn#Dh zGkbu&C$PRL3zkBm;o;$}V-J8wds8z?$srC@6cWvuG-fRFB5bG4Nqzt3RXeESB`j4+zgV|T=ar<6#vm_^ zg`zs()&-LOS9Bi+@tH>|NIcgd$W0GgD4C1~z0Mp>lw)JJS?glC>^(y}^4`0Oc2{~| zeU7gb!;2vSXLd|{u!UieCPxl307l2jk$5$}j#=HNeWsqs z$r&QnNb&4)pJzg)G({pVeOmExM*AN({JY4X(ou*3DbB<50eW2;$P-PushX};4r_5T zrqitwqQsYkMC(Y|jeO?3e%&WHSms(Wkc3}0#zS$k2M2`$vF`)~^S`HL;W#3O>Y+0m z-4`b{$Xvd8$$rIzPNemfkTfLcoEd$MVgIdlA(hlWJ z2ruMbh>n8^bL0|CJ6;#Hc7J5Qnx}o%E z`wwz_Vmoi#k2s!aaI{q5UFP56eC=oi!VxF)C1?s02|^cMY$b; z>yqE2^J26j*lFT>JOUdOPn2@#%|)%GP-6189BEB4*BPr|9Qwoi;^p#w?GI=d(4Y}A z%5H6>f%|X_;jT_6E~01P24=d>+CZj0`s3RtdAk0OrFE5+M;oVq+CEWspLYApJo?Qx zVUb+wYKGSb>lGHx#`cMr_yY^wTYC+2MC&BNUVWTB^Jl=2jy>cDKTY465I;YgC+NxK zur{-h+u-EPa78OoG|2QuFYXZ{)2cBd-_1d>Z^1JoC%NE|RJ19Qf-hgM=XwL7$M5sb zmgGn!(kRKq0*OH(L=C1)PJX16BY@AkX;UDe97&Rr-9$u0yfP*B?l)spa!6h`{~jX9 zzhw(jTa4|P4+Sncf>Bd)3*l7so5>cCd0aRX3z*3xzk?h%2Iuw%xu@lrvO#2;1HSc5 zG@Ncc$^#C+9t>JsLJ=i58*es-5AzAiA>HtlZ#QN(f~NWoiqNIrA8{F_DZ(<{?sKN@itFBtjXJ%=46)G82kS88c*-DH7$k9-Y(q zT)*%4&#&urUFV#(z4!aR&$FJj?sebys{Jwf=pI8L-TkSU#Y8RS2qN|!V)qCDzGc?jZkR|tW(rLMC%KX1c>umZzu+uK_v`LT0c10(f0s) zi7}+hKn&?z%krzv+b)B<@e&pX4fI2ik^{0^L`?%X>;s-ZKFC`TVGB4aYd``<89!^q zUhqfI2Lspc8{Jn#B7>lt3{CA=VLJxYHb$5O5K1egqVGX*Ap_9C?;lggfz(NbR>M30 zV~z%F!naq!u+aouX%8GwwK8#~rKS4}!~t)E0Mi1E3`kIYK`xAggJ{zrXF{(<^cmpJ z`9TvnYt;Wd8ysL@(iJckDHnP7$qS`3&!$&-?@vW&)@*ym&M{1GbK7lsLeAU*D-TY3 zUr(dA2Qd(ur~#1?zI6}NrcBI&XhS#Nno z=55nGsY|*X<>>x(^WKMMn(@=kVE}#$4=(|~Sa1P23b2x4dGU;8e{*NtQ_HNw3|q=h z{7y?c;cCT}D2eA6T~$UDOw2Ok-JJ1d%ymj3?HkuWyqVd0pf&J96$QmqC=@*pm%97= zg^VI~wImvwn@yo@2jWfbP6{(yFz&M7Ub*oea4v|-@7E581)UQPQ5mO(zS_aMWqznc zA{u<--ZnFdf1KJ}-hmpjTrdV9ipoT#=I z4oxNM12lxjZcD|S=6?t~J#i<>3*1>&ESnYp7qrx5V98Z{DR-1GCx`enLj*U%si{r? zZ4GoktXS+3^&Laa+U*Q`!0u490pa0F z)M}GzKtzI?6p|ufMib%yR4ni_*v#<~fiE=s)$($W0vP20ON9VP6R2kjuDqZ20gtPPSfx^FfeO?MhbI3MxNxLHR@?Lyo16?ht8}}#q2b}|F-NU($kTp|eLzgbri#wN?Fgx^ zWrMIl>xiJ>zFLWnEKK{0ZJCy^i4QXTVh6W?32Io#q=0M*o*%^p4Q;}I`C<=(2^C+> z;v;rf$#W2_PF)EaY85&_l@P2y_31Om#laDxUd<&EzAl=_|U{eUSMD&;g3 z!Be=~Ew9u;uB7_Gu%)Y#V#GEp`Gz>G#Hi~_h~SBSLb7u>1Ez@L z1Orl7cL!OTvx7P);UR9e4WcyN3#*-CjE1f*Pd&zU7hs-qQbsNRg+M zX;CB@JqdxoB^AXZ4*Q}N{}3u)x9`M^7!AX}L^Undk>jF1$JAxv@ua3(bL%Q`eOX%P z%Hgt@>+z>}qxtdezaB;!d{4GG4!NuCJ=i2>k}`Csci(+mAiy5nm2l|Rt%m`Mn}&bE zG>!$o`nR{GnWVS~KWgCkSyCPARk1Ia53Y3fZGT3>aOR`?-;J9QRCV8mEl)x$yj+jz zf~I2y)|8Yj#%^Nz=7X}^B#ew7cM)`U@RLDsDJ;D4x8@-w5 z7s5|_Ty~Y$+vSYmpx{GA#ey-HOb)pZ*l83+0Z=IH;ep-p*Zp_41C>m0O!t@i&QT>b z{Hs?>2G$Br5o=!Ofp4Jn8ojObe+!^7D3)ePvfyyrwV|NLeTFR>`o~zil2UG8hej=W zSi+!m{`2`9DwIS9Kh0=YENih&luVQ_v{g;;t`FpEWEK%F^b2|fVR!ycnn=N%-P$Ch z^SX;qxOUO4@~Z>1h|0P&O=k3P6U6bzgWt-XZNK-It*?bW#~}S~WvY&eUGnrA;pfHJ zVdsFy7MkEf0eCFju?K$*sx&m7JoA4deR9ggg)SxK5CNAqXM26qN?nE#s~`4A zz`u}d3T9%autp@>iRy0QMV+*->#EO1;}g>BJKJz?f2NUwcWwN4$wWkSs01~HePBBK zn&>Qxb-tO!I7L{R_Qx7=kh=~32qOA#I!3c90o|w&BN=UQP*}eqOGhP#ir#U*@Z+#u zOd6Q^5C7@E!a4{YY|z}b-x?Cdw%~-=cFxdR3fbR%ft!ET^5)nGY*t#_zuQ28LvoMX zh0(}=`Ye;6ygR-4@R)r>jLi0m4`KIuX6)ppxIYW4m8QnczpiZ0Cu38qvDqfylq3Hz z_2Ssg;gj}G0=BN$fd4Lt?B_i@ChpggSShjJ0`$63!_L!g-WLPksZ-)FPV ztXy8IDW+aqyQomPB^Kt=FfljbuayYh| zlza>~pi?GO%pAmj!xU}AtroAA@3G!w^tqn*1MU-+i>S*z+*i6%m;y;D2j_n;3~tOw z7dcVHPzX;b`e2oWg_%uUdw1y16MFaexkr$JWzIHaxytPq9g4QZzD8gEVlzf3>4uY* z1}i#)_wR}}Kg?|)?&sZ*k5Op(Ad?wPVnD~5?0kz*C5qS97VH1t5|usqor5S){V;TZ zube9xU^fv49GUJzc_Dm9e*C}Rkt&5#{$$zFe5|eO*`BMV19zi;A1!0~{PLFWMGW6f zSFFMT(GP8r6j`UuEg6Xl$_M2l%N<-rohGyVgmCU16vVB1|83=z<9!bYcSYQtBvX=J z%MmO)SHDh5aq<-YlwmSB7!*o~NXiF49EYv1Sr2+nvo(%Cf`oNT?Ag<5ri~SUQhMUk zry~jvSdRxG*MGuzP62CJ@xTv|3;GK=Uv5g%C2f{LhXGxT&_OMZ)=+Ex9r>|FX)70I zv8%h~GMT(GxuHVmdW+fHpIxxG$0`KzE3r3e$#! zOMkHzC;J=0BHn@})&8a!*09EZ+I-o0M3bKBJKe)B&)hRpjnY^gAe*ZIG1woAc?(FD zwuLNGs5F2S1i3d26|vkIDSugJ=;kH{a&*V2LL9k)&<6+i0ULEo_gm82QcZU%TLmbZPJ0f&8j9me1G1 z9F#ad(9WZRx<~805fsiRc?=KCcA>OZAjx1%M zvdI#IVi*-`sHlKq0-5CC;S*os>IREHK-*Tq;U-?n^ArNL0W$`oVd?{E*MldTV#;|& zkiq+R1u-8F|H~^k^EJgQNT<2PUfnf=w7`4`?iU6Wd0>vg2tWfKpbOfdTLKs!SsyeYEkLYj?S7jP~@2N8Y>K%j@9VboWF$_L4h_5caS@c}pw35ftS3$MnHIUXL3YJa_tOO_tCtJnd3hhT`N_ATuQ9Vr9Yirr)9Z*xhr+ z4>msK@|T`WE}Ks%h{;0+r7!__KPu+zL4y(-7q^(hc7O5I)iZtyL&IZvuX4Q%7pkrQ zFZ3*sXMLp^uL1=pG~>Wvan+9pDJeiA@sG4-y|~lirw0p&9F&M&*5|!Cf5Q3u8K=?H z9p^ib^@t_Clv5-OBqg7(KO9YldPjhjAbqg*H5V5lL~jJ!tcF{=GE=ncZh&n|$c_C>ceH6d?>sdi45e-%KcrcZg; ztjUAVj+cFZY5ey)!WZty7cz@OPg3;$;)&lRqJ%dziLP6Ae6f{2MK2{7co6tFRDRs< zBJcqBhjN$YQ^4fX3WqKcYV(6WPOidXX5#Bt1chhxx3#t&AZu-zu)W?Q8$31OX9BAt zFRK^4sGyWbm{FD(7`1?K?n<^JjE8`Ik+DE0tgJ2NwdabMxp3pK03Pyz+ds%k zked?@4xk9cr-g_Cq<#PvyD(h@i8-OMjtBBQd>pV}ID%S1Fu6yy zA1)38JCSKIgw0hzfDySC)OyYE;xTs-ri1?X$3V=b2>*n)*(aPY;0}qc>y9bEX3CeQ z%$5F9E|YT41UEYfQBgryzz6(_DyM#ldOLC_b{=!5JgBS_rE`8Y$o z)FY_dJ8oa#GXBO#wj5X?*L=Y-<=*6QqY}Nem%(xcxC4Q)2XM`t^UR=^3BLA((!Vw% zyTo9q!5}=Vu-&cA%|?LUz7xj$Nt=F;PsJV;@v@hkOeryPWAQ!QS1kT=CAsp0!z))z z&jZJQv*ww|&?Vxfkvf`>UxH7_x+g3IQ%ww3@`!ZuYOMTCFQU{*lZJzi?i(n=c{~$F z+i$M&ZcZf zy?7oXr>Jvg>uyn(nD*6MuWY|bWn77fF(jKmpf;dmBW2rZVvkj#ZjIpD=h@qVw<;y$ zZPXiEmfsVfrqSU0Q~79+Z9}+v1?m~p`v>EFdr)-HEg~vBvNwGFnguo_AlWCtJ&)#` zJ|<;ywt0g$Ub5tvHT&n74nInZT*c&u)vHh1Ah*!C`|q2U0X!MXD9wO)<=jTKq0pE@ zQgqk7ABFut0d;K#!6+ZhuRzpgG;aW%u7HD|7IA=2IKwVUIDCBX!H-$7B3&FC)&bd? zx3?!2Z0hTSLTLYF${_{N|1tx8GZO4%X=OfuqX9_^!2fZU+ci4cHP zjq`TUEQ2#8+c^Owb`v0Z7O|he0aKrybr$3^4ktPd#vJ&SGY*Z?1Urj7=4sa5hHX^* zTJ|&#)B|FC{Wi)dYN*3%AwXg{Qoy(fB)gA+QJ3*71~|cR5`3ssgZ)|nM0e10oZ;dk zfsKy550M*1!mhHqx+0ie91v!4Kz9cJR1fe$G$uGR<}S`iegGsuG_JXBU_)27eB)tM zdf}C{{Kox#^&PwKQ3^LDO6fVdsEjpj+Pep6jk22<|K#WL@zc3U-~CUVvQRS0fmm>2gf2I~u2dAh_)pxmF zjXsn1%zP)fJI^aB?n}wbH-{YTwG<54VG{!$-GD}u!1yBV9rQ2Y{G7W8u-3(`e#w#7 zJp)!ovU9Nw^4x(2+ym!?`S{vmD6R)%pV|%oyi)M52DU1uji=CT-_l%xP3FP=jb#od zt?tvTeKN)0#4mDN4cA=qs_FV6#`XILH7i;q?^xC1gMe0G3JNha>WNh~)(2QJh#-Yy zXe23UKXFm}7vx?hu-K9DKWYLY%028Ta8Cu=Y7sSDVfRFZU)^F&yM?^*W!Es{@ z&^Q=(v?wxq`hk#wy8p}xGAi1ZZ}cst4_ka}mnj0STfKOtv*VR9TKvG5nVAE$&~QN* zj!EQq42`2(H%?MgrfPsys2-^3%Y0tJ-x(&dNw#d|E+#9f_E#EU&fbhQINq7=flI&r zktt=n9vqzhIVfANnxDBjtCBk6o>iUT9`?9F*wI?-F=8Zt=ifADkeXdw5>+nNir zPTX%gO0ZKeB(E&I;tx0$GKY6ScS5IF1ZlYGirIeu#dG{crQ9(WA6VWD`)O1rY;TTb zJo&d8FKKPSN$R=N?^s%A1j@yQjSOOX+1_dRAz~CKq1k`E*)uIQcG_XU1P-$xTj0 zpS<7B(65A&e&{?-U-3dp=RHrR2pV4Al;bpiYwCS<0^eov=m~3Ve3uyaJn2nwQ;aRW z1%7DgT*KeNn;-Y9+-3;65(&HR&KVb1nq7JQI?x1<`Dgv#L+LQnyf&TJcp0uWyR21$ z%1Qb!;$@|W;Bi1Ei@r1b5S&@vWqh>8kI}HAkFn)(q7^0udoJS8Ae_Gto8Kksttr!1 zb^>~8qc*Cvo1Zt&Yh|_H&krKP`}YfP$~^MkA-GsF|UKK^% zvSoVDm8Ph;gom1(mP7j&K4WPapRF-A^?+cEmhI--*08X5c8a(Wu{mb_E;wajUNi#b zq^VcYCyM?B0Hug!CtCZP%BdnVszS7Z@G43JPQRhxVDWbq{bZ7;qztVYkV|guua)Ti zMC81vDj%s3Uz*T5`1gS{8M@i<x1-@h&d4F#} zQ-5~U;loF`unI3$sz{?Y1m{uwI$fv{^>?~c8L-Uy@3TT=&lR{dge!%reYn=59o3Q^sR)U=4}9nUc~Sc~_5QogOS8`0!W1W0 zmxb_HB<~N%q@*C}^T7{?5Qdz2wWSOvh||AI%vl0IMcYapPXRnf{w7x(CyOYr=~e7l znJdUI=4c9aoUUQ>BBDqnp`c>-==if3H3|!6drTzSP3Dp>ryUX7-XS98rV~@rI^g0T z1T=X=*Cn&1kwU7T_fEMTEI1x2TiE5T8Fbw<$Nr5k8A=@xhKg?cKB@7@DOP*zWLL7E z1y)sP`Rx1!E;)M2=!X4ckENd>lU;OZA-?7G{WRevwqtmT@GAemS5|aKex&K6WSz{5 zgiPm(;3vtEi!J8{UgtriPTBj51X3wyKF`#n{XucItxarEK$Yu}W#msEv!lTwf0|(s zXET(DMvBRl@8!=M7-B|A!Qd^x+FBUhq$7W^Ee;3nv7MtY70VXmw^d-WTj^D+@}H$b z5M;GZ{N4H(7CtV*TX@kIpC;xNtC>ZL-4waaWc;qzFlEx`oB2Nvgx~LkUZEFD4<>0Y zN0Wb)OtSg0Z}7+`p9nGynWt1PIM}Fta&mI;tt1X6Q`54Bv4YgpG4Dtp2DK(beYw*m z2F)T!4Umr07@U}VGT>-Tdlh(ZV*Ep$6SuHt>xbe#`@sQrtNM(fjAU3(TJN!nie_Kl zX@e2=(Gy}{@y@BLsX5q=Rfi~ZVtyu<*beazNOJ%$bpqUs--b*7Mz%&&aAjEej*coA zYBpM)f@gG;jQ% z|7MJW>~adqMr7iGdgOq;yut>Pqf`?p`hkGVyo5{|2_b&(%rer6IT3(L?T`KDJ(ECn ztT|=TlxgAD+2=}^yFXn^w=FwbsebxcaPZn=EF7h90BZ1Am}s1ej-ELtPD~>DzE~Nn zW$OvmE*P&$A-)!57LRA3t=0Fsq+enH@D9}3)L%R?(o z^1w}gemkI6vgKWSXFu$rldsea>Glu`pwQGZgd+rjKL;9{YX&K@-uzDYP&K{|YZ8}gZK=&~R*fmH- zTWH=!4$cjkTpjK0$Q=M#D?%cwpzQ@Nb*^rw20BFmiBTaK87;sMWc>ZAI7F61vE|N% zE0s|rC=i)UAbBsmp)JS?M4T5Wk=q8MM1ns^Adrl}x85E=D&#c*1Q7I8aL^E-a*rK; z@bN2_4T}%TSWPvO2rr%8!d!n@752vJxRZ@$KZ=Ae|8$-*eD2@B{{_C=kA>{~0+_-C zyf>p#RqmKq01$v9E;nFfM{oePlyh=&GV#BJ)%x+Sq44&UO7(jsxcr`KT&Z1;bE+bF zAcXw`+D04#-o=419#jl^5R0*GJz`?K(|~zT720&L&yEBzR-+> zFP4$q(Dq;w3cT#h*~6zB)G(~#m#wHW-Oq>8eZHS-&#Y)O6#zJ*%Ls)w8q1MKrzl&gO95=eV@_^2;04pGv_kGixD5 z3i_8<>N`4)0aL=VC*5bg?LbIndjt1!+~a!hEYZ`vycqtIdtcGDXw@VB zGZ0q6HsB~YBx;xh%7LKU&u6ga^kSlcd=WBTBJW=EWEe>|?;E&@3shF|F4{Q?Y`FR)p`99YLBP!?-0!O#Ds#`(OL zrwhq^h;6R+)7;SsjV)&4t=!wtJoq;;(5gqGX|NzdBW{7A%Z;QIL`FI%!O(8-9vkuY z1&R`#Hlc_!O|S85m-}e1vO^;5IVTb<4k@+9p=`q%F|Y?j{sO=^X&4)001*DvcQ93i zxwS?FWaMPDgD>S=U4Myf)esF_W9cfCQ1JLf@%ob9!JXCl_iFuRVE_m|kmj|YpOAf( zSjG>P_3TGxbHBuve2q5|B+I`HQ*JA3t*`jxb#l<6JNxgi790!;CqT=F8E>$PdKkkn zCHk~PM>%Hr;ujsp@gF;~?;;*zH}?;;W~dOLk~rfd>2AM#MD)?-d=j6|D)XUd`ZTPHo+nC2)+aT zG$zGE2x=)nN|96#0H#>MTRM6DK$}Na%S0k#(DPZ{WN&;~w>n&G-}oZ+W-hsv5q_7N z1@FsS=kbmzV&xD0Nmh5>1J?xT=4h0HlM{?i@y_c6Ox6qTA^wHr!qC zAbVbOxCw8`IP)BDXo1iDr@OfGzNLk&WfqHN3{6*$@LU(bH5zL=f333F^bphM*F+qy z(G5?+U!0bb!f^2adH8TAXX;{Uj5jRlJWmD36~*XhHx9@6HoiPuyF5g~pBddz92E2$ zV22$h;QIr(;ZKhbzeAe|%t~XR>?oc{iP!-9i}BZ!EZJmJd32LT^P~s6DB8{@gvMJfO$u98(;V1w0OZQ9AtZPAlk5> z?V?AVK%`iJ33kW$RA2@0D*#}wH5zdO2hegIG9#a!uAJsz-tain8%Ec8$S?sd4qqZf~o|rsU7pB zr@4h94WGz0xcfd7xoMYV=@*~Cg#{b}SIG~UCU6upk9Ff!#e^n zO9=|q1C;Ls9bJwasJmSd)CcoFvWkFU2pD-L?>lBpJ0rAUX1mQ6Hxyl2FMrVE#Ay`K zeNbFO!s;^D`tl(*6}VFlnZPwh*;dbfc+GBDeo^(l^zX`U(INN1i{!X|e(VJN*fe4) zi;Pk(t`%*%Qk&a-NheEX`zmev7T)NT=)BE6rztNKMEs0Q$S>lF%$C&bNS$4afThpyR?0cTxp4Nc7k&_p9|a}YfzzJQ((IH_={vcvP@`Ut*$ zJ8*2nXm$b+aTvq+roEp7bz50Okb?nF_aqA?nW%hI)9TMf+fJUz3)^qqE)g+b%^)P0 z>UzXVG0FJp{UQ0(Qcb*U-@K|7F<_|5fd~VPhm66R+2H=nGebd0?$Y5V^&h4LAmrrY z*EqYc)OA0OKw@VM^ks0z^mSjUXt_`a{=Y!`Zw3l1Cw#Z{3!(|U_qb_soFdt&nPcQsExrgC*rhIwu% z(`*!%ECT;3UUtNo`!`8oL6X3nDY$sJ&5}647heSl(R$qYUO*WAcWMZQO78stA`&;C z>fsCvi$A2a0&|PR4aOj7NEb~AQliMn^gVKn_pd_O5X=F057J3Q_>8O-K7_cBmo1A22HUxu$Luqllh1`enmw`qWg*{#Hhtz zl%-JQK*l=$zzH@)&r>j~c$$!K+nJawLeuKS>eg_%O2?0TTX))R`KVyp=r939!Yxy!@?Vn9bZ>C~@B z53ZYDOmO?p=T|?CF?4HNIETTcTMqGyI0dmNIu4q2WL~(jd>VUK4|B*%YQ%ls+_o&q zP>KuFa{?CKm5Uea)rB~KL9YbK@y22Dj+w7t`Iob-r(;TT(kw1tCW-L9Tb{i^NW5g} zOb5Cs2=&3Cp(SDgL9LmPHR|L{*x#l4ru{ANo}{{oE0MANuh=!GkWibgAWV--COlk_ zzgq#PTuWOUjH7j}eghc-22VjVPC@U9TFw;%1640&DKF+df*_&-u`tDt5g3E(HuhC2 zxnXeQ#!o?S1Z-`J(dTvRY9IMl7v5{J{%s^dYdFDKr5*1T`|;4poV?Lk{+r21&d8^7 zp_p-}=DSXsT0;Bl7{m6u-`{!iDmR4i#n=u;=*+P|zS~xOyWpr9rGdT0z zkC~B?74}0KObZd~xWp>X_Lw^lX`tbQV5RfLE?iO3>t9~lk`9IsJa-Y>YkNE|<9dbZ zo_fMavUq+-kU)T;q~HgTrR{7a?Y|%Yy(J72%Ls_FZtDR-_-Rs7xb=$>m^uiFRP$si zS`9sDkAoOI?kJogP*gtkGF&PdB=E?vaA#I}zYRLPnJ~w3*~M|}{_New+&BE-yO7Ns z0z37X05n&->%f2?B%s=M38(ZhJ(E#U^}}|W+Fz!nTQ1oDEMJJC3`oB+L)~6cv&&m# zrG88{M9DoFUxyhO5@FzScnz&ItbPPM(ChRv*wnSG>@UBz`NFAJ`(HBCt+D4^pFEA# z(8JvQa%RNq%aP6`4eVzU!2?{F^<83mn_!{OQm@5}PXiuN*A z2AA=@<0XQb>Y^RZn8~c&7b?~=<3op+Sub3Og!VxZ2sYgIt#e?;LLwZJvtM$ZS4_<4lW3n6qJ;WFbxs9w*|VFWn>OP zip9ps849!cz?w`A#%k(bwzOoWzTrd$QX_c1Biy;GH*Sp1CH*Pi^a}wTDRdXMpkmFB zHlN@hfeqUbnYn&!D#h1|pO>W%#hAo7q8xnw%aL`zmhB{VrXyjyCYAKhWL0tRnL`56 zh*kxf_f-$O^c^gfq3Zh=pi?;s!@4>elyw( zK+1EdcOeLj+-;#ZaQ)(MJ5o+oHA-3!B|Bl+&tlOgAMFGpG>-cQqzjzp)4_O)wmpyUy8+dK#N>3hEc%m(X>SHVsz*6xv;b)HhC z_)6b39cy~ZomIfDO!tN}708Eg7F{TMp?p4E9TmA^Sc!%fqs3P3GnvFDTQ#1ct33VlHtC1@bpyv3qQZZZrdyHe{aR| zn&5&`k%g1@wvtQ1hvrZ{&_}nsws(R6OZ;&Bq@r>X%H6KYis_4Rz!GUUv01>#L?; z!nfPHi=$O$bJm=?PUW+AKku9U1zC7VJ~xdlNS1qE&L61ct~xGy*YoZ$-QMJq%;rL@sfjU}wfd)e zo8+FJt6p%jCYF8e1IktdaIYX7ofQ#@g&pYT1pK-b;b`MRiRJ>6?F z(xGaUl8VaY1*_2Yj?YE8si}USqO>4mT79#7oxX*cQ-t_>5#GJmk3M{$%bs%0%;9{Q z!wG{Q$Q>N?)yksxByMHGZBd82^I|>zSyYsOMpn{nW_lv9MWP|*kkin#(%&1A0Sm<+ zy-B|tlYVcq8F~ZrqI{AFj%3zLjv-3oI+}R-b)?|<@M3lR)h>I1uJP{MvbHo6X1229 zYJt+$${Gfu2KTI}Bfh@E*TDm5EF471iDfP7uzDdS2D|h1F7a>nO;ox_Th?7|;%f9| zW3*42ZDZ=i&joL_&hSxKx+CCJod`(^A8ZZS)bZ;@a&Lwoo6x|&my`>8fTm_?tS)Z# z!aG7Dk|H=ky{x6m9!iyWEMevQZtLsAGl+WkP7W+)KMh|gRk`GqSgIvv^*G(J-}UnS zuDxHmzmmVehWuL8!OhIe`@wN3(A`okPUx|&k>@jXK0&P|_;W1@5*r5hB|cC!!jwvC zb;<9Ppc9fd@zmC)Ixh8t32j|?nlRli*!^2g49Hq5KaO9=eh#wem}n)#seOOmTvBl% zYFZ&3GM$jRy2sM$wKR-X>W=A@5iY7EIru-l_rLAmZDWZ+{DCfsjrXkH9w1ZfW?cb= zwmDSS62CTk_vY2Tac%O$7CyXum@dV}*5X*I1->nPeSMUal&&}YpG8JC3|S>DXNp2& zHsKpc<`=((*gf18 zn7P3N!R-~JJ9zkvxD4Nx*dB2N_EIRdp{p-|tx2FnjEsuX`Zi$TGGsv~?h*=VS}>@^ zBO)Uo_RiaGZl8=0N=e44^<=!FH$UC?ls;DI+rS(DFA40&vwq>M4_F;%W63frZlR3W zx(v5U>|6DQ>&!D;u}`+t&p;kVQQNIv{G z>-CE%-nwGCQG3bewKt8wNh!(NCsM+_t#~q>vs4i-YH%B}va%*09+zU`;R%QPNhq>b zRNE|6uu?J0&?UX^K~8aN8y~SR&`fqXAm8y4->F&_f_w=EB5*(r3eHuJKuHd2D*TH; zXdN(cB)K=QlgT4kDTXkG=^8~o} zKHmR=O?4iwIlZ-V#CWGrqL#mUyEb=?5W`j`Oe3`CNxxV3h^P5Z@%0ET>}OHIf>U!a zFlQFthY_mRnVw5Lui^NTUexUQi54XEYa*XLL&RSATMLlThMv}~VxPQmZxoimkqegZ zlarHu&?y0Z15jd1gs)9Us!!UBlrI%^WMpJaj5u~%KngRsJihO+Grs@31+wggitAq> z@A?6$>HKpj+z=ZPh_VlXNe8yGD0lg>JH9`5RyrM(lHVU0?LB~!JrK702QbrKsolK; zX)HZT9nc@u-m`FQI?g!B&CJYvaj00R8R`hWQ(pw*F5yo412OJf%{BwH4A4iwbAVKQ zFMTZEI{zNsoj;L&>2SQ7UL|FNKpT6urQwcR{_BUVtQ8;U%WX;P&3}r%QVI?EbYJ3- zWcR6juF<=CYSQSY!F3>&G0e0rYdfVozOxdNN5Mz@5Qf3>gE1UMKXAG?-e2fmg+K8H zto-~>2MpbpQ^(JPsdl*3&$(1Vl<$jx!qoyeX^D4LRf&Nb{RQrjFbD^kMMa;(o$oE2 z>K@ii;}dq@x#PHpvLTp(aHzE71SCgS6Fs+v-16Vq{C6EW3~9mREsRse-qPG}+Kydw zKj*CV?#*@#FCRNW(CbgD4H-+~mi8Rc5m*6iTh zfwKgzK5?B2Fr;ST*NZK;mM_t&2g!|>@wk*;)k)%q5B;Bc=oM2peD&8jiRW%nD_O#$>% zey;{f15dl{vG5qdCqv2a4X59OMaN@sFwCo5-~hf{7x%vTmswwht?pcFy0Jq)<}6iHjpw_4=;Yer+d%CU&|7)%9FADx=_3MT}OU6J3TB zW1+@;jF%XIb*W?Is9ba3>@fQMbZ^gt!DCeh9l4`b!Z4DENmP{92ZV{xpOE|xcLW9~ zMqR&C;SZ*3vt#*uB*5phYp`8`OnO=+*5KvG)`JAc zHfQqMs1xkfCCMVbrb1%!%O3ALNby7APtKUwLB*C@(FxlKM(BJJ7q;q(7i$65Ih-f+ z)@m#mt}YmQopu?=R#K-qz0A`Px~t4=dB+8%d zNrE41iJ0uN`!i0HE)A!j2exkTtZ`&+5vGcyGUL#{lB@W>x53VSt_+hKq*PN9Tm|(_ zki2TIIf&qaqC5y_NRQ5QR*yYCbBjX-QpJ9yzmHa5B7QhBXlOjRy5=jb{#@24yzYpJ zJXTFr*Z3DPs2C=pq6WO8Ph8rxG00L4b=kOvAQ!4ff1u7#Ykc!2MH={JaKN7dpFf-l z-MXJtXyZDcwkNaJzct{tRnsoIVrl4%pO!zb{gnEd!ejMiL)|(tD*{h3=eNr)1NYC7 zk)4MPNdOw=*>LYYW*8JAWu4G5W0QtOJM+}uqR~DYK#=&AsvcE=Ug%}e0%%n?0lahe%E4S9u_GoLzeq| zk`lXyBn!0NQC^jcl|d!82v@s*Uli)s&Cp+ad*`^fYIB#sQXPD4oE;S@5Ma$MDVYGw zLo@~zCV`-KHY(2t67?W?+WeWUik-IF-E%y5zF|4wwfnI_GLKKUD$kzvU9YFXF-=(> z5i;R$i5q>83|Ey7SJl+g5(&93EljwK0ABq4KFD!z;oXD1lR4FY-42jsF-1B=C208krA%HwJyo`k22@BWJ|k6ER# zosMMIJE+J*P!(|S-I}44u>^Nbv>1zu_(4hZ9<>98-{kwV!=aMOQaSzfk}DYa&qKKY zom7B6__H;9V{4w>@|}OJVKnFY*Y2k~zz_Y_hwQr3|D!h=k+@ z1cHaSlH+$2jQyzbIp zfbmElD6yuY3tz&gIefI?`)n992X8xCfYJ|7>7^$F03ZPr2%Q!PZV@{5^G@q1e`r2! z_aq`h23Cz)I)x8dbFG5F#bNmNn~yc__Z~&UZGs~6OW+mlrk8RBjb(uj?xniQcW=Vj zt_KA}#_uL516UI*H1x=`NJ-=%oJcLpcyrs|bclkKGyUgrX`SPt}Mq}O;nB%xPZCm`j7Rg84fI_-- z+$lX0O2FS>C~O}e)aR*G0Bf8ClNn2Rya0qaE|(6tb%F+k&tX~#)`YEwESYub`i4mx zdA!ueWst z@TfQpF-A?nk-7VX@i`Ix=C`Tt*Pg7_^T|Oa`kxzXe4z$<2hJMhKNqs^e|>I6BXIKs z)aTA#hiC%cLcwmOp@L4WU`Uif?JsHF2@@8ecDIJW-m_3K%IhsfMYY;j%hBKXEk_g1 zhJfLfOS9%b`sz0bz8@=vNU79&1tKL(+MVCo1_>F=iBEG|{B#hX7?JlD92w~u8A0Hy zBDMi!0ZZWE0E+W^ry*ErakGsrRPU9vbhVljHmV4n zFRLbc#R2bC2j|+q+8fuR#ZHNJdi%x4tLE49c&~_PQdq6&`2EUY#PtBMIw*Y`LAsC| z#tiE*_Y)jAC@fTb@4rqf#J$iSA+Gkf%qbfz>~jP9e4TVZ#tzBKBA6HTC`NW zBYK*7tB;+X;G61WBLYdc`cDkhWOEcZWncmK{__j*mS9|&8)L{cM2RW>EWmJq-Ab3LpYz- z8U?I55 zM?g5%>+y~Ys56mS(~sP7{U*t=-rhR6XlgM+s#A5Bf;h7lEhr*pNFP>RJRTi38{VIAHVMs0pGUQZ1oZ8xK1`Tf23O zG;kq`E`0961%VKTc_Hoaqz%^lTVQDs7_FlR`b_iM~ox<$w_C>qN zkF@z{rS(M!EM)j716iIXlFquER%`hX7LBDMv*xbEy#&j6TA<_o`<(6a-g!uenc>zT z%dAk4KMFhYlUYYd{ie~?Asv3x5D-)oCB{*cmO$#g{;D$aiSdVYM_lHoih7h<*wmeW z7lHH2$WH36&5JZ;7*_-RQ1c6E6%_~dme*H zHyRegXq{D5WW}#8Gb-#e@a7HEm_9V+LK}DQAeNDdb^^F=<_yKke0zt=#47lbam3* zhYr#^3uxWM*)Yq2znjH<5s(SGkAr3@_OlPb^vIR3y`w|2 z1%UK$iFE=%7|lR;03 z7gNy_&UuC3PYCYnk&wWSTqqk}me~9t*M9`%kvfaDyDMhk!v@K!YnbHTYQqSe3;-iU zqx1(vplfg&^<%GMIls6x?0(bKSTNmYhat z!CKX%ET1~WWfb=Oeh|I5@%0IfB3z>Cv~16ZZQ*Ln9PLajd_HZ~k?L-`nq(1WDskmW z7h#f%DR^&s6 z`d#2~t5@q$0l8r6(DE2bM~veH4jl)Nm03%Tv&wo(<u@${&I|So|5e+PdNcxC|gO&z}4X#CYkw_j^vEbJfUzoBGb1;R}W^vnjxY*YZ70 zcOg)4QF~opf~-*1g2m&<8GrwR{p7YRQd$v`H@eE!_H=z-FE^C~FNl9nyJu^63kZ@W zI5x%!^>*I7*DGC?+hHgbnn4Vh&P$+|ZY~aVFWiJx*3*8Lg$4O|xi$k6Q40E8z!>Uu z&=S|kvY{Z*2tXIe4UCxx z%TlwWUjulCN{@xTV6hDonGn2v3crF+a_zXohtl(^%1ICE8pctg$>j&h6k>OSHK^?Y z#A5BgQoW&u@)pPg@h-xcC8W)^@`4~}3l$OIki#i{m>ye>^CrJNii*#OcXx*P(Z{-u zb1t}=3)}3;&h=6Ru#%))@_R9i!JrNx#G{&K3kwTMYHA8d2(MguBA9*70Io?pkBd1+sGNL@dXDaN zQ#tn(I=y;|06-vs%tK;p&V&yUrJ1EC3lqKGsw?zz1NJ=!%m>yHx;pkuKF`WyIPScCt9iRbj4_GN9u z-elzN7hNGx&xasbfyD_DJbT={`>6NGiQ8JBM1;CesFq! zbbwA>DuuBe;99X*YWEd7bMr8S5weqsPT_Ij9&8MngT<$fH+wz!CNW!AfqC> z+(^a%bCY=UIz&t#m>wjJMF<3i2Q~fmYhnqjuy9O$eZ8d$f_QQP>h>RaY>41w6Bc+s6`B#UkN~d9;*fEgfkQ-RJQ}{LUkRE=WV`R)43v33SW!TiYjNb1mQI?O zFg`yS+&X_%PiaRn2}lP^$J@UcKiNJynP&3jbZ=x-o={2yWZ6Qr4ZMFA@I1F4i4z3s zBAE5nCAP?!SMl|fLKvMg_|e1Wo`Uw%ESxmZ`U*gsd?5n`wRiC_JPQ?}du!p|jpO4< zKmg?f(QLkO7A-9;L_G8WUqp5I?>^jF`jPeG>Z_3N8dZzH4`HF8q?S!rv-f(lLAY_e z7}Y8Vm2v{0)YY)~P+E=N7!5C;ZW zZ@{79`W0*?uYu|Bn=!X+Z8G`POs*U4zUtdo4NQx~?ywzHF^|=mWhePjraI07@6t$<_-L#@(r zz!3PshYwe}J(QIRf%hc>Zp;N7DM13S8v@z&N)jQnql#~e@R@?6in%Ictbs}5+>_u@ zHXEn_Th*a9ivW|_on<@k4s?8Ezd{+}c?b{n&=%7BW&q<9p+~BFfsH((H>D(ML3g&u ztV%if)dLa@_r>VLEfjrRPczQN#)?H48!daYVb(SOfh~iSG4S{08rebtB8$2LRe<@5Cf^8pQ!{ZmK?6hf!NR4Df=BI+Bk3Wb0t z1dKj)@1GR#eh_ujWsK7Cqm|CPOcPkxdBJ4rE7O`?X+ggb3Fhfs6oafiuc}&{YO;DV z%82yY{t$j%4F3LAYG(i&dQK3Uo9#32zONiz8eDicFLaG6#<#=X{aMsetf{(knwg?V zal`Xeh1dF!zNb20#Y1^-YA_2MTapte=Jxv#swJZ&Ics*?W8&C6qr!obj+ILqvd8f^ zd4mWDtYmdZ??MPI{Pza7c0WtF{ClsDPKx-5l!Yy6zI)TmS`fF>xA}+pJm`c>YUoC& zSuWb8=5K~~I_#>U?K)4?FKCRC-^cojZWDwMQf^Mh3)(xZ+%|EW#V#jK~5y|1%n zv@TMbRfU$K9Ci`?HdsZY*&OW)G_WBt7*oMsziz3ixd&X+vvPTMd`A9^IhrmKLJ15t zwu86vHG9$zKvekcVFXP5ACLw-yYsn5>O(2dmPU^W20WtT6X#ts6w{;o9Nd`r^6-(j zL%~*&G--Wwkig5{J7b&7aw!^drSfWj+9clpN((4279vJ~iEAJ2|6&+~IyrhWaH>MZ zP$UMhL1_xc;F6Tdu8Vz@iQFd!6%USTm8?Bk-b!bdGlxuZ&1e$%Kc^I6Ft?$6A`yUL zUEOtH!w`(7ie86h1RDj_>velGJR?v~;s7Ns0MQm9_b4vOdab@}suy@e8T(m==$=V>9V&4T%Qd*ww; zkAO}t#=BDu?B6-AlVrcT0Ot>ra2i{tC+^yJ|A;nbWdh{_^e|5&jOa1l|3}$-z;pe+ zZR5%=DcM^pMG@IDA_}3bP`1#=CVNELJK4KrQ+CK683|=%Wkps-C?o4RFLdAc^S|%s z|9}2o-@d=TKcDe_U+?QWuk$+2<2X*it~mk}8)ZSmRtOR+I2Q0dScrNpHe!EW@_y{X zP}cnoR_f8U+<)hj7nI%MGmBx8Hzo8RF*k(!K0tf9`M>$QGpAOg4@7>L%+OyJw-zX} zt#FsFD3$eu#Deb@AlE@D%%9f9bf<~K<7oOBV;Xi}(&fEasl6Cfrmmz^+qc-(!=PM1 z{oXhV0>hWl%-}IwAt%7!sR3^S+zL{U^*!jkSP!GvUt${VO}LM{Zb0d%9+iRqt}O)W zQBnCk#l(qih7K*f6MFBr9fPTNS+cfuCvOhD;awGH_hlq+Q@-J)0GoG`NS%=XjvBv{m^!S)_os}5R#I)Qxg*n zBJ1DKA8e0yqluI58=l6>4kQ>U04vOfz%Lk=G`vKyBD2t~ra6&tQYtO1=WI;vb?IQa zAD@Vg??~OQQ;{` z0ND*5+V7>&WOTrJ+#%RJ=g7Qr1Nhi=Q`mGuDiyX?cE3?kTEUmmNfGl~9-onMhP*p) zq@aqO2$;_AP~rkI$^FE|2&iJf|KS1L67o@bl-wt|uE@xB_o6Cjca6q?tC zmm5hbGRR1<*e+#h!yKkAY0#?Epd%Dmgu+y!Li0I%5|UFhq%F8vl#kE`he;IyY_R3| z8K4sGd8w(>i@L^u{VeBxEewG)?8C6>d?dXiEDjV1%gSXuV37zUDnO4@L{o%@fCxdX z1aS1;Epn1t%_^VaQ57Xs-18yF_#_wp#9bkr^$=lu#-U}>Osah0^d#*=aV3(lwHt^8 zgbJ!)K8DzUs2jp;Zk20UWn}_7GsFK4Mz0M}sI{4#>0yLt$%oqN zYk@E58Fh;9^Zx)*F0cgu?-(UK-At&1AgYIxHvhIdJa<4JxlmW~c)?gdN7jokw|h$m zirP-pzRSNcx#5ym+Y&mbAj^+0E`I7eZGB^pKQLKHFb^3ogBdcFwbMqv2gU3@sY?)x$5{^8;#9oUW zl6%qRLuSrxJ98I(%*x6NR!dvk1U#xKXkRh~Ma=C*hngaREa_`@V_4pd9MfZUgX2wp zZe*JgE|Fw6Q=fP7lj-RA{R-7}?OR4zA*Kcjb zm6Wx+bD6*8;;%okrKxfmC>T0(=G;@o#9JDtRe`tF&7gkbH(*47mezyccRAdKUvMxk zT0*z4R7F7O0x9|@pgigBuQw8sO_{Vb6%Ng&w^}6)H;!!16Lzezu?EdZH2=OUAyKRM z*vdLIuwZDO!V9hlac2PP&~#n?0G-6YUSoukFI@GfS2uj)V0oY)o~W(x&X$e$G#R99 z=DfN7BeT3Z6l8&65z_kl3yx;d1(NtTA3TSx2GT-hvebX}0;^Buqv_XnomY>N+PU(G1;4Y#lxRgK#)9W5G4eLq%W;4yu#= z4#y|m&t19{2wptaP_*I$Ko$L9f}S}AE2R zqQjXol8yR@Jd$Gz~lhyFI5rw$(L0RJFxZeYd1j5$_`|;ttq_8Nx~Qb7RTYJ*|yPQxs0x)%V%CIN~TV&J6|J&LK(*gJY1OZ~=OAr{?FMif&W2Y==M@Cm1Qcb@!fr z;g(MFX#%oqZ2hOo`%Fj7`w3>lp?qyDI_x~;6}Lku|<6cLdB0n3v+A|fK8 zcug{xon&)Uj+9Sqk5{qW*NQfuO$a+*jG}>VHNwOn%ouV55EhUvjNrj_yYF_|tRk9q4>>D@_&J&KCo$k9gpx%pYq<|QBubBj+w(^cH zeQa;!Lv3npMo>G+*3TZV^b4m>&0*J0{%*zji^Jz22TO$zVH&1+RZZk-nvj5jt(UjB zC*EwhGMZ!0L1~BtWZ6#VyFMk5^Uk|&m|F>1j~l7V;c+x9s7_EW-jUpUh=s>Ol=|Qy z|D}`wbnL`Wf6CxIa|WKCI&`2iPPB{awT4NMN)8@Wugc2$;V7}tkv{!ZG5i5REq|DA z%lf6dv8}YH_V|*?;UVKzr^XrUHr`IK%d$Zs_k$5s)4?n}Tn92$6KyC;NxvQj?8+5j zsL9Dmb$Y4PVS#+EGN@Gb(tod=L$StToK)nF`O+8?d@Xh>$Trap` z9Jwio{oD#DR{3GbLLJ)PJZ1l#M>98|wYKqh_VPlU^rO6USUxPo=LLla0szM7x_)1x z|2nF~BlsRjlNYXD?V1xnlRE&G4ge&edB_2_mY1E_v=JvDFcTo)n<$@7LY`|Vz9JL^ z;cI|slRnXc{0z+@C_{1paAMlCPKIffBfYn0aCUb9AFH3NAenj8#%kx~MV^?Aq6(~2 zq*M(t5nSzz5TIzJQ0@Zs9}vp9psw#gW!@3%5SSoBTCg`F1A$WLvAd%MNYt<~Tc11* znvbzK)3@W0+~SS_?OY!uQo_LJg#-4MmyScj34TuVL7sX0YXCd%Vp>`By)C1D=;;)m z@-W+zu=iqE7;r~4+qpZ18-S<;hVdt%lAI3oQUN+4H+UZUiItVY z0NMkfpxq4A|CWvpAotlXjkE(t)A-(7&4IDLtDr*x%9T^o=&<4!Byy;5H*yvZ)r*^A z6;p6Yd3;YSx_dGZUzCMY3_OV#OX5%8@-`l6Y`oI%>&G3=B_B?FhF+kedq61k!pmum ziP;2RZRI5H#PxS~=wtAl;n0LoN=PYR!QKJO7*%LY;7|jef!--8%!_b?y8J~CQhhPE1+^9ZURefW5nWq9u+l}1-A^B;sH5l80bXt zPBMG~rOR!Ao`iXk^a^MhD^F)%&UCZhgIuT(9C|IPH;u-&?$po|Hkg8V0GuS z51d)3e8lDCf<}JvKJU=b^0kksf7T1dv%7kEMmM&#&Fq&RllF*&e2!26k_CjZ*@3JK zDa-@>{~W^2tu21V!yyE6y`s9h<9!AfFn&RWlmxr4=!}Ax#7&`$oxtNEJa+8p*#DdS z>h#fl-rfIS^%Z00&O`hEgZ^r&czYWbg!wSks{rl34D}NvbOJIc2_qxqY|R_M&(nbR zZ2(B>LJ-q$2_R#Ei1r=SyFVbk#2W(u_zVX}NS}^-BCSQ)Q&?|^IJgVs@2eow zz(U0aIL3i!?hX=2p~43cbD`Z@2oTP(f4T3iyA#n1MSvvkMO0K1DEIpTPzRPKGc%3?cKxgfcjeL7_cIqtcOUqA7Sat^&^b8|c#oteKHk1vV>~xM1P4 zfOz|;v;hYfZP#Tfv<_g=JRQu*2%!_Khf{7qyU}Zhr4el%M(Cde))k2p%x^vfMB9_v z>e6DlJS!3;o2k0|dC~ZS$wNyv^D*Of14s~g^WboJ25pPL^$MaBd;wpv0(YAPEyoh` zW4YlVt^8LTq6QWy4UXo!?>&K|Gqu>0%8aJY&QRDmkx&XE03I}VguX+;YVzQ84VV#V z7kJThLGmODiYbit5hCiVUZ7(HoC#^v;M>yRVi92!@bJv+Y<;+Wz}x;tJmG_Yw}$vW zn$ELnc5!N>=!A`GjzG7>yk^{ruGu~h%u?ZvF;B1sLj5NxE~}4i*0Mu}3}s*Ck5{^s z%y&L|ufCSrI0^^(ueTPKVF$9g%Rhfc{O-M`&Ebv<-^9M5ec2`e#W0xCdQ^Dn7#l$4 zlYcAIMz{SbszE|n7XX;aXSrE zO=!kOEy_XYlLX}Q&VA^D{(bXQPJz*iFVcxJA-)o`F>Q#GN!KIGEN$7RHJe!INxEuL=ro)w_R9p<1MB z{PvXtm5_$K+aIxvDc`}LiT<*(veNQ0=Pxo5xVy9saB}|NtI-C#68*si3>rHQ%a}so zU9!V1_DhHvIuWQNBqd?!gFA;lUVge@hYvA;c3GOyrBLcptvX&NzVW0M?stXDVCP4WnDGsvI z-py)Y+-ipwJ=t;2+k}k;DF1U+@g&8g*GMCyq6%I1Eb2Q6E?neknG;~xOxx@y{yWA| zL`tr@9&qgvoX?ch9sR&j)j1EZ__>aQ&y@cB@SPzGq4y%ysf|}f;8p_CW>975C^OtK&Wuna(B(yIAtlS-fcZ2 zjg!mtaV%c?Kw*c5IZOBIXz_VW76>m7vox?ny4ynmg7|1WFUtmp)z~4>qB&h|G%XO%C8Rum|OaNzmP9o1Ci;7^VDuWPJ7ItmWJTP-| zK8HH7lSv-b2^Y&Ih{34~T+k+gjhhOh(4>mfYcpRwp|&0hK{EvS?MFP|qmxiZ4(uR{1SWc!+Sy{bs{ zoObB&@F+^ERb&85m3|C?m(G1v_EKs?KSR71cGOzuid?W zEPk)x#YW+vTQ1+t`PYpZ^*Jo(I6P{Ph}3eUtk(S!l|ca2{mqj>iqVoIjS{Iy+|as&B`~LR z{pJiw9*vHeAKf>4I$}CLd}SB84;Dpz>sD~kkOdIj3tH6ghPSPvv+u$d+yUsZfKT38 zc7~>KG+o-IX!G@19F8X^Jxm%}=VplVoEq>HcV2KZpT(jYcH$`zFKAB!GZBR=aLsou zEI2?4vCqXO3EXjPcx2ARzOp)L(U~@uyw|Y!_K2>*r07RFRc+pJ!lxT)c6dx&JYlc= zZ^U0qht-D-AK@{yFojfgaK1Dct9^Q}8S@w(=|y&SbL(VTMq$VNFAfP%_$CSSv5rzJ zTD_8Ike6NcVp+O&pKO;+NOYAz1_z-;IFR2} zR`zTd*M$Jmo{(z}%-o5kJ9~-pL;Som7QyXV*(uS^`IAHJ?QTg z)Qdp6figp+ZQ?Pw8Wv@`+I`7tiURAdUXD?3XcX}v{V72K^0T>XCH-Y@C;X0qdj~qz z3ByLtR63-!M1EaVR22DDKq6xU31>It2gkcgv|t59s6>qMe>=H%(H%tAto>RZt0Eb1 z>!Dw4GIE>%{*h}J-dCgshozc=7e!!Y)lMQCmHPoNCMGv{M9RBrFo&g5?G~L3A(k!i z(kBfHnCy-(k8TT1%_@K<5v?;0AO2YZe<{|L%^Me79) z1+g4hYdFT?)G^bl6UkKa(lr)X8-s4K5qw!6%=rN`k-wey0w#=PT%G@jW_)@pV5%C> z00kI4;Cqm2$TiFo&+wPvfw(AET8M15-?s@ZG8)eT9>0En%^ruEf%yE2WDn=4JdvV8GzNKl<|=1fDN5c`%v5vvQ3m70sEi44!@E*-ujo;lmI zaPpO{jk-N6Mo<^MBJj@;ft9yx&9_FNcb`m-Xp5^LgAN9mBmj`+QK*};$K(`0OI)w= zc~fW~`ei~i%`BiF#?9?H!boF>DJWRXzs%?EgJ$V3WUrI>_;D?41l&;$B0 zKLm{?C|n9Qu>;g+W(pJO_RIo}D+OJ}aH|cqdG$QImQVb_XZj|Vu=+l+F98T4skon6 z7R7GBG(u)Th#Zc1ATYpFV3A-5c4{a|z1K_%4}oF|e2=EGbKWvDF})nw1kEs7U68m8 ze`=M?RV^0_+zGAuNo~AEUdTTGTrJpApNJ&u3l;y5KY zqcU~!@Vb)}#NUZeoV9R#Z!CNnL)ALo3S+;83ffhByIat(JP;kYz?T8p!~iyhq}T-m zwlpwNQs6dECJnNiT7W}=E=g-`5lWyqTTukdx+;sDAzd7&6(C@Z@ z)rjBBoxPelr2y#{yVn|!xuLD`cSIEdhn|MMzG$)S@sr;`{txRzon9XPeV-7NodJP? zckU}|YUak*NmnTTopmO5%WvftJ#Cecc;;4>$x&yAvjXq!M=w?A8Jql7q2sG^g9w&q zJNsEiE{?|^mBpZNuy2y!vTm^p^|c%)L~0N2Dz(&^>P)@^a7WcDjJ9+aPZIbLUr3b865T4uQ9V2H;zH{BA9RlM>;cEWCEfUWb{z`GYeA+*Z^+OMBrb7C%N@I||LFVYgW(GTYJDA#{MyoJ zCp7G!Rmr=(haO5Z@|Qq19PyUbGSC}}m6a6wR{*lhlmX_8fP zhTa&e)={VsD&fQL=eo7n91AQ8??+MWMmKR zE{ApJrTYv&z6U`1H6AVv`l@iY2f-Ps!W`bkyNXHHeSU9=W;kD#hok3`w$Bxj<~~kD zbdHnrSH^LvK|jyOeJ_ggK5>s6aI=!ErF73Fb|};@FI14a`-+8kUodqchT6$zeH)0K zB*1-=JxOWv%l(qb#OYhOv?*)OO<>CLbf=r}_*S(BFP9zjt0%AaBiVhvL)KCqB>aH9 z(V}<&oYnq?H6aM=?*+66DZ0Jdizvt6=4!FcG(9x5#Zzebk(H$|FRR3_(^-a!6U}QD zYrwtjDg?bJ8evTsENh>qWlbKn`cVD9jym^FhK$fuqotTlv1q%^k z`wI;4@rmiiQUWmM8T9^O_OYRe3A~oZVjW=i$vD+_c-~3? z%;eUQB9e;Vf)4r3y6JFq&E>|2i`RyLLjpPqY`X3x@VnC6QhohUWXpM#=>_313lFNO zV5I(u2B7l@>!nMpG>igPwK+}WR(I9a`zz}t6+At!U1D`#-7+$J=YD3Wf!L{6G9mWa z9o>bd1-^!+4Fs0{%M%e3>n*f9Hb@yo_^KgZL#;q@AU{ zw!I*koY>@Gihm8qL059c&!`jJ6yJaf2f`f)FOZ-Sj=7JaDuDI+X0Sfg2hYLdg*G){ zssow#ZXI}37+WiaC0jQ0@$)XZ`nB60u@YcP9zLUy`fJo@@&l5dG=CE{CrK37K27=4yZ((H?_U5xVXQ;_r z=?ffTT~1mqXPrJxOE#1X0>0EgK;1dGPVibRgYD-{$QhgaAPRQg5h$p*=M?fL^xKW{ zszL6xIw2MGdXc%%I)+D5UNO3NXU( z5GNNN&zzP7FF>Op@8=nb^Oo{@t1(HvbR4+7ci$SDmEL&RNlZ1ahqZBS8&Xyo1m#Q` z|C>;>U^eZeryIutJJN>CXl;e&TO&D^l~v$43g%rt_fZx1dU+Kh=vJ99U;8^D{d0~w zd33>5Vn-sL7pygk>HjV6^-=elw?bijuiw*%Zmu5@nl{2Jxs9)OYbkYbZ72BHvxWtn z6LTj*21{u%z&{H9X)2=)>yGx{x&u8T#ttWpWwrI0iGCkQq}OqbA4BEXczyz#rVYdr(Ae!RVF_If2t+Wr!OnDYk*)lYcR+C2Ob%KVXXETs||S zhmy;A9m&Pb>ROK$2J8L_ePLgY`6p-Us_WVj)qRlaAUZ2x?~7i3_CvSA%?8F@U6`H9 z%33GdbVQ_m=i6)7Z{wSqjV=vQP+%^+TwmYP{u8nw-x-tCo`gPx;ow6gCoiz;s4LV{ zF1a*ayAkq&{O%a_)~2}Q=)+%saGCn~V~#6bd9)YxCHutD+n4AhGt;? z+yKvmTqjcaGO|)5dg)5`(0cA{yi;# zJbFKXh}+l{_L4|G;hTQz_OQCBj=N*!3ZgG}_Vbu|C3aIjkUoV45xwU(l5xa|p67sP(>p*M1<6zA*GndGP z*GG?x_xCjm`+mK(nEpxe2X!Bg&kIh@I&K95nf)*v3}JZTFbp#v^)UUSw061Qdvif_ z0srXZ0orXh?l$ux(*7$abZQ3Whv@n6Dc8%WD%vfiKrMo~b+CR&NI*FHE8g@X`y2YG zp&WJEs_vXr7Eyn%9;VJivJ1a4=!P*Fhp)#hv0Y+~Unjvo`IT5+?jEetk5j)RRJL%t zt)-c7E__f+^~yk`!mrtXa%?apFM|3#TO8U5^<>3$%q zd2RagIsoqu)BNxH^5jUT>(9>*C||QRD7Raj`L2RI~ zHMxUj5we-TLcFua(%30T~7*#s}Zy--ZijGhY|4 z%F3FYbR$h_wI1zSx;MU?} zmU`vYpL7PD3)2;r#2yA>B5Wd}^wc|beT$~Rs70ix{W&@qS}*a*FgZL-(dA53Tc(!K zBYkIC&~1`-i@X>c8(jX}`0AYZc?~jX7oQsS@8Uju|3EjOCWUxTw>OY#RiyjJ(=P(A zCu9`f(l_HYak3c7Asrpb=6`&RCya}qZf;qQOT7vJW9$Z=I!&T`O?>7 z0ve|BFXqEcSpWA_f6B^oPP;dDB#znN2p-orv~7Yzri8XFjWxZSUqCkf$vZR(64KAk zTf=eX$-l??&Z6~n;aZeS6!8=4^61ma$2x5>R>DT=I9e57Lk^RfDU{6E96i{K{E7{4 zh1@M42luKs^bEu;#>Op*oy9{_FPTz~&(4lnc?{=s3%Xr}93=%ML48B0Nx{QN;)8BsDk*3-s*9F_EXX4L8{I8C+qKibc(V1pg|nU^(K5Gpdd z*jvqeIb5?wKko%t$lA(39<#dc<&HA}y+lBkn3c8`TH;!|4O`|xmVx9;?2*Gu(} z`pVMsZ|F}h3dD4bC%$RD#0;k5hnq>){aBXDbq3YgW?={VZJ})@s9}j9`A+le_uW+$ zPeV^i;!1M8Yg0$OaldWhe{11hTa=!~P{-mhq+^fW0yl+(yK^Q%cbPIl-`ME!v@YT| z>%L$SqSeakzwe^?F;)LDa&q#zHiN(kO2{vCzgO9QrY4zy(FZKZU(cTc7qK_MwFPPf z5V(V**k8Z@2);;z<324_WLqkZNC4ats&Ik*js2{y+1<;pg+dEK5&>n-f8i~%fL=7H zz`)mdb`0cD^F)Gku?_=n;I1Ptmk^wJ5d{&5Z+ef_WA4L}n^xY>!a)Se%F(pD#pmVC2MlCKTo81S`HECg zV71oR^enX9H_$~hsQq)>6BvTw-|@ffBAyP#dT_oK-XsQAyGpmvv+ z9s`$cY-y>};|R7obonJT1-^|ISS{Ux(Ezpx;!!{w9)5%SnBu&erp^~`( zm~rgKUxDKA>9?LD2F_jA-j#Ykg>VoDHAb{QzUc>lpR$E_LuF9BMM67Zd{0C5!4ZSn4OWLoL4!NaVF+Bwaub8-uZ2@y=*h&RN>{!AdKl0Ci z$J(qupM)U|-)>QnV0QBh{&2=(uESJ>#sU|*lCjTEx@!#7~FB}LZ+vtlr%I3HfqJj>?>H=*y?%Y&G^`fvGR?L zzV9tos6+}46gB%Ah@N*PFJxRDavPGD3)p_mx_tb#jy^T{4?N2A+S1F*`T?x(_*{>t z9T4yiF9oa+H)okKy@!{R2JlS*xcI+)_YwMAJXZZ0+8)mzYWlX^2N0`3pg<)RuYp8# z3LHSzg8~Bsq1D*$Z>_&n;As#wFtM=QwA{e5d8{|9L?kA*wzSf)U{%1OchYCL$$?!a zV>o$2F)P(>{S4d3=vRq8S6}|bx@}Nu3(kM&Vh%}x?(Tamvgg?&52soL^i&6GfG#8h za*0B{k6qDUpxv66&N_&xtpNI6h5v;Xt&Pds|0AvHO&q5F7NfZQ38ziG=Fo;{MHklP ztj3LZZiX~`Hs75|Pd&;|zTjClCeUP0pYym6>%U%2B6s5IyE_L5rhD^n&&a|XWb)5n z0rLQH^Z(O=&~CDY1lcvceCK2brPWoy{&v67h029Xqtn8E9N9L-nT zZ%8uP_4LJa9HGV=k1|_ZI?no^3@HU|gP(NKL7Y#GjOQ}I1L!={VX;|03yM(Nzwuu* zZB3G1WPO!OL`?tekD|-P!|2}{Iv3KQ_c<+4rr7dm!)ad)DwGvF9q#S=Y1sI>4R~^- z_6yIjkCe|J7o9=|{Vk!gxJ2*my_wu~IHaqw;$Q7(!<%+KTZ*8KkR(UtUj6na+?}|+ z46o^Hd#6b7FZ#eU@8vD-Djc*Vlw?(7wV&myAE;tMTUn<0YbyRZ)Eda=hm8yQoGByM zXzs{(y928)E)L?(Z6pBn|kLimaB8}`V-NI40%RwJq&p-p8e_&pg*#)DK2|V^SSd8 zfC28MLb6(raCkIm&ni4Ex`CCcB~|t~x^*y9OJ$&1M@an#r7ZY^VZC?o)wZ7f?Z03k z1EVj zabP?GZKE5Z{8nWGUTQJG00HxIn0H7x4xN0e_F1GeDS-sXjxodQ!M8JVZq8(XweIz1uC-4#(7uhINGxWf@8YRNXFpP0QA_6O>~ zr34^DACx)3^&9410!lBi0D*0h3+8v=cCc^24TQ#oYDSnNaGuks0aqogm$^jt57qKH zOiCedF{2Co8Q8WM*q~Tpy#*?{Q}eWDs`@aP6m@!bSo;poAhWL-0lm8u>{DoeEQiA0 z5#-VA7Z^AJN`KTp=4EE4AJXI@i5N5urtko(*kG{a6WyEyx@Sx+(1Fn9LM_eU*H|6a z_UPUYBXDdDfy9=YmWDd^A=^<14ARolSYyl0jMG4q{4aaEG?O5Fh{| z+cv1ZVxvtQJd&5QWu&EFB2Ho4eMKYZUFf+>1#4-(~>IxSO^lL zCoXo|1qS&(Abz8Cl4aX&5ZTi9_7*RlWMWBI=;;!Nl?WoQ!05!p6W2906Z)2c5(Rxo zX0pK60}6a83hfR$$%%>bW@cuI&7r};f3i21*P$L)+t=5ptegzC7vKOb4198ZEFfW9 zlvap~P5T28TaU_hJ>xtVXa0a0J4gIeMVAiX2?~^~j`k2W=i?HkJA7BYLP@99eqrtSGB8Nb!# z*)LxvdT@o7wT{(m@IlAQ$NVu)vsU77ksJ~?9_NNkoWK0ZyZvioM8pb3VVg$?YUo9Py2c<%~#!y5=a{7UX`;#)uROzCjVnRN%UqlWIc14$ku}k%i>K+Ae z7^?xdBQyOZH`9rkM1+O|iXo9>e*ue36KvD(y4a}zk5;%Dx>^aV-Tpt50-d&j@1EBT zLd&W?RB#7R7Q45f{=Nc7iFhcM40A)zauzPGFz6(zJi?y?R{XgJ8k0>wPE*l1%7*s# ztOTrVI1lxe-FokDFNMtcYj_xk#2LRC+NY$6Gl{V|0ogQab^$g>cZ$rQ9R)D1xpZqiNR584t~PY$E%qw%#V%*Hc8wJ2_-ZcAh`eTkL z5ZaH|5ALPeZ{FF=x;yyyl2{49w2t5Dm3rg=$?uC4pz}pyJT@Kto zD58f>zx6`ic9p9g6Tt490A~v<)M8{B0_-_xn&tr!qZ_(=Kns-)gHVws_)qlC^2PT* zZSf`GYQmcCzb7MRTqAN=e7r1UF zXbN6{&6-QoD0_1Sh$oRt5-?oAdzKt|dLODBfS^V|NVx3U4vKpJhVe0we7=UR$Dk{U z1{**qHHh%;0oNK4N<|zlC4gZ%+DAlILA?w0lNhkjBk|HhL_=bW0mPR~H1$UVYzj0|JmAtRLk6$k|SMMI3l+ z{~u=H8o(^rn8DxTK_OcDEZFo`EY+9hErA>PyPD_pVYh_OOXYhoz}8e(ZEz?%#!)aL#}DJ05jDlT2lG-L_ts zTIgBJ=Ud7)a@v3r%%vR(*1vW1cxRc%0>%yeiGwWGfE3RBCiw8mGE9En>o0EQ7Itl% zG}uyLTj0ft@ev1ssxDTE13jI~y{q8$hRmG(q;p;$z_~41nYmuTBW>tXQg^)^f8*(9 zlEH^p{yiHwPE8`+uCZU;kNN2Fn5w{}`zpIW3%L6NiY%LKAXlZ?O zANR+P>kEA9KOI!_MfF3Ck=tWE<7N6F|EHHa1w;Kj{ER{os&;=r!|V!yZdfZ7nC{Gv zoqMz2jf|-9v^s>tT^OFI%U{Nn2Lx6v3Y8frx4x()f&=tRI7utWU6FgR&<4z%vPX}E zjO}ME_czlQRo?MXJa?u4`}n)KxWE4Nk&)H< ze3N+R%di&5aWW{SDPxVpv&y-DvzMCEpmgBP{uTY_TP~?7DgT(tA-aGFwQ`g`=dnL$ zk!oJm-KUc#-bFdjKb+{y+EVkk;@HG%dUn6i($nrX6faQO@;*(5)6hB72u3Uqzz`0o ztA|+6WQK}Lt;MRuJcH{aH5 zKP^DNtDr#fl2h2bMVe}mdtV)Wa6N}NFewS_>~RS=(6LW(UBzC7wqTOB-nG~N{$4RO z`AY^+)jqsBbXYoS@ps)k@9legz&`Z_pmsk|IY)_ZFzmN&OxLO_^HJu@&{L7QYwhsF zro7)Ndj=^;{CXli#`#3w>YLdc#+pu&Y_=lhmMIHZvOs*#eZ3{zjt=SNiPDQeC> z+@ky%+ZihN;YwrZ2xh;|yuTU}ZX&ovx|>@*Gy7vou328MVx`JGh7UZ6_nM#cae&1a ztgak`9%vQB5gpjyuMUPy(l^S6=@KiRY2g~0$@QxlzaQj}S$Qsi_uDH=>ue;f$0rBdphH2T= zRHkJ?)ZJuhF8bTf3lgL@61DT8aU7a&ujag#rd5R=d8$uD9X2+@sKA1b?n-QAWTlpp zgtc`6!~yn%ij;t$>F9#<0#Go1A+!qqA=UH-GsEl&h)4e%O-?X@xSbjtaCLiT@{xv` zZg0dROCe>0#G(X-?mHA4{W z<2?DSDfj2!*DrZCts0dfP?(5`v73Uu=ro?Ku1eo4VDNt1K7tk|s}>Z`yPjcg*g%f~ zWmQ}qvFZqOJ^YKm-?*1o0Nvd=weczxthm8;l>{@2pdrkw;xpx7@Ol5wcl&@4b6r#4 zPtMa?uf|h}{PaLmeQoW-GvWE4 z6h%Tt##O6)w2~(jetaul^*bJ7I)yqvuF$iefubCQ{@EFl@Pug5$)A3S5kon1ZwT)Y zpDpRsAd;sW{4@*WJL!AiS?gDPt+zM-Z|jwd+21}Tjm+Eo@WGFTv((2y93;)g>LpdY~hW}&Rc>FDjyXsX*z$b*u^#5~VC5S;z zAvs8osyzW_ljPjtzFd4fo$%6b3H=_nsXAp<#R!cUkoI3RZ7fccf~$8Wnh29#2e4n5 zz24CK(}2(D+Vv=8oAKIBoxo;xv3vQT?`guoxZHl%ix&S!Cl9V+H6TE@xh6_^#@(ag z>E@rWUd>x1UrR4QjS6{iBiiPre7Ft>y4>>W8}~(bZ>g;ZhOvqB@M^>i=>~L zV*L6E86r()9L9a~!S|{PM=7h?=Q8Pg`ZevQ%t1%hl{lj1^BJIS+=JIXB1CqQe#IJs zg855nhGaB4&lhymE#)2i9x+$g8-ka?K2F^wF?%>%;hG%k+g*A`Loe$oH|{!{K9gYj z9Yn@4_k;wfFbBg$v%mMl(1u63E8W-#a8TdJkKb}zI$VF)iSyS%#1|pQ`__s#Fozc) z7#nLf6RbM9X*pSDiK*gM^cR<0o9%?kL>zQhh1~!QsZm)cj4$my4MkO)r{&r$vlDGF zsd+vzJ-U7;eQg(amw3-e0b&eu@Bz09dX*+$wk9nZr@wd0vLV-aq?xjQ%U1tUs5M%+ z-G@$|?S|EG=X)Svb>?`Rf0p+IXKe$1bB6DIE& zGrlvhEWCCuXC%&#z_XUy<*{|%ZR`Hu8gJo&nZ7#6}CD&mDs4Xt` z`f|ywP_*R-w>qf6k{_fQ(m4*>x`gLD+w}}u@DAroiqPNg@=Lu{qw4XS;3qDyfkHxQue;kfIX~3ulPU72 zW9zTPqos2ljK68?cN*gP6VRv<9)4tRc9xaRm{@Wli}MAA1}#f@$!G%S6G<)9*2X;h z7m8u~ccCQ<{!(TRmKTO<@<%p3`cLWm5Pz>P4&)?y?F7_HKk4ur)Nm}_d`P} zE80C5wC(If)5^CTf4>|0^=G*NTD@tgk~b66r?6fAGu#I6Q*%c5_k8YkDYhkxy=`4#(X6ub;Rkxqf=V=w#J&#)(5th z3ivR1sRLvciDj9u{??imh6uF4<*$pJ9{Cd_vsw(%N;=T@i!{;#mNhD9NpYo7Q$_o|?KY~Y3Z!faw#ip7 zNb&a{Pk3rm>NVOyO43VK+G+fKa+z^CwAUbb=*JYrqlwWYRo|RdA3ucqQ;7`vO^<~F zuc8wfC1ImmDMR!d3zG$dz7g^UVc*)Gek+Tp>+r(PP+SQ7@HHeMxniDT$JVZ|dV6~G zG|}zoPzSMmmCxqoK)-c{i-4c7Iv1-_ssen9i2|9YO$(<9$G%i?Z+|;_c&WsPuM#Ic z+Fx+_Sm~!1#cKr?U#TqD+)lFG)cg({@foScqiBmEp)NM48Eu-qCWaZd(ANtXZ)mO? z>8hrA4d0eB_wO~ix2UBy?7TuXgt4&rj>|}jt8^P*jF>cCWI8hFu=LboakrNS zD1dANg)t_=BWBva&NBF!-rwry&HvC+7Y@DAv$^(1(&FFc{WTcH-FC99_`1-XOcNhgX;I?b~UyzlLW5&Y!=(WxFzeG>^kAn+uK;%w3ct>E3>RY^j-L zR3N6kn%QO#t^nSU3%dD=OGb`0_dfV_aWDPrg1x;xl_1Bbmm%Wlf!L{|kdbIeM!|fZ z{lsdzf_G1k%INq-5sr(so|6yB`uk_Dkuf_H-|8H$Yh8Wta$$yrR-3%4Ey^B_?V z4L(DG2VA328!PKm|5F5!vR}G90;HPvHt}^RC|`RS>yh-ThZY9ryLKzZP#Bsnbhj5> z(00x;<-LjS>~9mgGgWm`Zf>5U;vNSo+0vDj-}FaGL^5y{x0tRfm=0T9DXPmXP|oIp z3uV5FquGcy1V18)Pp#+9B@GdUe?J;L6nyS<&@4ZNQjSnB9W`9kEbVb9oWPdpaFUT+ z_eTh$UZk6vX|mc+mzS!^)xA{cWI=uG;OR)`d0|y+l@b<4kbbtmBYuJOQl+Rqoo+oj zCwz^`*?%u0rmF~=tVz*5cjZgju!_Eyd&vn8nToYuS}OEeM29c4RN8C(V7c|g!`gq! zE)Prhr5(3GFx?3%>p58%?FZHrXm(3FukHk-ael0NZC6p@&;FO#9Y0eoZ-0_8-)Pu> z)QTrc!O*&_yj+|mysoY7I4HN$)X6p&Arxoy2`+U2(#kp#=$ z@t;2fK_@;2?Vg&TUoQzgz1AD+3{ClZcZW~-OvP)|pM|l4wcY8TKYB2T#^2=l!3~?_ zdmpFN2PqQbxC;vlqYkzLf`WSBLqW5Zr(SThqoV`C{O~JOFb6?89(4A~d>8X$y2D^+ zJM{ULm6A0K<|{$fX!0k!Josb+tlF(pkSCM$-UT^@fZZG!sCm-57^tZHz-Pte+cYr$ z0nSsw96xrfey%r1QAGs@obpQCpiFc3?ynlDuSJE0SXegYbT;Lhr7yX!TIu#Gouf#< z7~i;;?47oehbzbRcY*af{iR{X8j9y_KTb(-@(by9*fuB3h6r|R8BPT`xi9a}X-SnE zX&D)>Q&Vw>iHX@3;9t@*GGEH>$jSyZ?SdLm`p%tOz)bI1L;4Z`6MF}{ySkny=j~i( zlV^BnWKf=p6?Oj3?&nh*ql_GPmVMGjkU_*5=L}AeqtE?v zPUH6m`&aSy>f@-yL@gP`uvhv=&`zG|6E0kIGvbuQn>V?;af9Ub_4S-khWb$hcRDsX z=>>MCwS$BAGt|w0e)YYx{$+#Po-1{e`lG#pfdSU_9HDkD$1X{#SQ$I#Z-bH?(KndG zpyMUp_Sy1;817-N{jXJv6%22E7)eG>&4wQAt`6Yd3}R1mmpCJ!rE;XZ3LIIKYM!`J z(bDE5nn%XPW!pzZN`Qk5?;1ZBmpCXscoq$ejPStOxWxYK*|UjzW6xu$$34e)1%^IS zi?MH>nw6sd$eovJ^oPr0Dw>mPn(>TbnDjaFkM8D8`vXt^IKATIN9xDB-woiVO+?NF zoY4(>z`rDME@Kupr<%aypnUKHHw`ym2}wLzw#a%|7F=9hR3JJn6k%b(gHDP)Wuc{| z{PqijXsD+&MB%bF|1>c0IMx^ltI(1ztcj1O1*U^1=-&09xNl`^+xY!MAQuP= zcPc&JYK?#4A%C*AcF}NPe+&Jwp~j(z3!w8Wa5e}`-~?+_7m=% zIh%#_f{Q7Dv@{QYcP=l#K7B_$tz-rtZ= zsAp#O+}|9~cR+;YIS4rDvszvrEYJNn;RPS03g``2A06za=nY}wXdWH@5YZg$2k0FS zKg&T9qNo9Em*>&(_{-|>e93WfasLuOJ$QBIJ>4ULME61S-le2$T{6iW^x_YFaCYUS z1i=}mXxe*vs22|W%t%43rY=rMp!0iiT7XtY*T6)bvG1V_m(6c_9CQgVv)Faj$&WCz ze+mlL!&|3Kfau-S_jHbAD(2 z-ap>6-ame8{jAkfkHlxV@9WyvzOH-kP(*#GWc=8i@FbySbi0s{#5L{;%fhKEgv!T< zv2fv84Hsw|?1W+t&Zw@#0cITR;C=;9R|x#py?bXiFvPyD!cvu%M@5DE`t^o|4O&7f zHXGJb)QT^WJd6+VRU3=7TXEVz0x~sl2*5ZDg8Ap(hx9E;;!S-O-m_dcc@yfMJv%h8 zM`sv2r3p{IOn%AdkXae*ha*|P0G1LeZaQU&-ZKJ*~^zN*VOai zq+5Ln(#fu3ZT=|r+0UDJ!_1SRCAbOAro!J3EybEJdS*~$tz58p#pLkejVUQ9P^w(# z>N@!;Aiw9F{A*m1;zu3d7uUZE?{ZY5g)dosdwWSjM&5g_&70#>QZB%u#3R3(=I^ih zJLHd;>Q`cw8+8`haKnHcdCkCCUspFgG&Hm^5E3{?jvdSDIOghVk>hD*Vd2g2it%4R z1mD2;@JqC*g?TxsDg6Am_b;DHuI2nk@jP98pf`g8>{Nl_1F{iQ;A zk?q?b1E&oQ4^!i|pbfna<9Rt*PP6*mXWaVA8Nc+DhK7f)Ovjb!kKrQi>o zT4>P%jktZAUw~;1gz%Cie-6#mejGL}!mFqp`Nwa5HbGpIIUQ^Gl;I~Avr}4{6^F@L zVAf^_>z+CsobIdiIbgrcDSLu;?b@{x4$UlRa&L!Zosf>MZUgKca^=02+NB*0vve2- z&U}xBKBvmlf#BsGKqO%pppZk>lh!I!sr;zJyK9%Blv|GxXSKS|Y!wio1$LOhne-@JLnOv?+Ou>yC7fQRyTn-r zBdPIY;9v(v>Q+)PIv00n@%y}uw`YF$=B9x6|NO1Gdwo+t?s3s0@z!vFGQ(2JeNgK1 zWsH!80Gu+;{1^ z__V%ruU*=PB@YEFO`kAFkCUz4n7`D35s+~YeE&pmNI4&Jkl?uYO~=LmcTt60*JPai zqp$z`t>1(Tm$a@r?9z`cWLY@+svS9!{%=7N_mVS94?GL^?kI$D;g~lEjRYLq5bSYC z`#iv)ahQ{b@QwQzOBr66ous9wFPdOjuczTv=KV>JY#!)w?1lUll*HDOBYn^#&WpP= zo6GlWqN{)tCnrOdPtU2+o&_0m1fx!4-9+KJ zlGv&~Ky~m-BJ^Hxrc+K%F8hI^oE*>Atr{b{aZHELc3vuW>C;>oTkT&0Y%j8-W>(C!g!a(ncA%Xwy09poX#FHrj zrl{uHa0VAQhuQAVKYXvH9qBNYtEsNO0+02GySt6?5c{2PAM=6faUfh33Nz*poD8gn z*afV1JECS_idun{N0hYMao%3kvYY_}N_o-)FD4R-Fua6GNp@n{3P+Q0NR$mjUL6KN z1#t&mMYk-wiv8PGBRlS=aZFRNuN^L-t(rQL9@p2Ip8zzIH9$ z`cZ)K#=KkE+Amp13P<++jX`5FPL(UE6Pn@r5_)MfaJ(6YKLxSO|ir85x;o8G3qpT3(M&FE^K0Rhc-ySTKq|(|S{p4Zr~B zRaa34Gn1^(WoB+Y6G%=e+Dvn5uK%B&t^1urLWpn-%*?ktwlFh?;zc?4HTCvd6yW6B zBEiMiZrnI*mVX}moF?@Qr((blr{g^ja=NU%i(Fsk7Zi97)vbgJ(&1ka){n-t+A!iA zg24fw@IqPEUn#%l=La!k_3Zg`gR-Tip94E3CDkr()6JE#4$RZ8L5JJ+=pBU>oSd9A z^8>Q6cXG?PmtBSb+Y>Lsr*pV+y&L>89EDD3RKeST#IP*Ao5G?yc4!?1Fa-u*1!lQphxgu+;$rA;+t}kWWdIWTIbe`hU8t{Y@Re%r{0@l3I>rPS5lj*;% z-JRm#=oo|fz(~BqDERC}!}VsK+1k=-pyI#_`40Ey{v1XfcVjV=#bs+2o3Up)4< zJJAVYeQPgB52kR#pw)MqBda?Ur6|b%Z)1bx2Pe zW;k+HJd`_Tz2_eE$mOuA{Yk<%)l>OnAnC zcVN`Y66)iuvM`sSUy1R5GRDIMjsyN;+v@7;hbAU+OKZ!1zUb&+f}K4RNP&0=g$9Zb zt~9htlhrVonx7dl{)4HK#C@;j1QPcTp?qiL*BnGPS=dn06T3y3BU#2xO;wdb<|Kh? zl6sak#3eTkvzdrTvt_>0{K2=g-It ztr-x&g+cWvm~Se9q$y5@Z!Re*8S5%YPUX+vLiiZifK$`Z%s@a(YhcLf0gQsNUM3XR z;W^Tw#tP`D_V)(p9m8cgzsXIk+3P0;A%MLg2u=!HQ{5Q2kAi^I&tP%nPL45PGD%pX zgDoZ}$2oTSGXvxJ?8GRn31)b|3uoX|C>lItX|D8Az@H3_4;n5I7+Q8Tee(!e$>O$a}Scq2d@gfU*@=|`!PZmjboxShX&E`$CqBV;HBB5#u}tADAcGSciB2R zt|m8DS!oBc2R!G4IYowIN%b#ZZUE9Wan{F=EmZcJpSDeO1Jpt3tMhZjuc_-KgyZ{DW=Zy7Jf0Y4S@?k&H`qlMi?9$qs7L0GXsUjrlu}Fd!|iMRaFh~O1pcpDSE^kt|u_r zS|=zdh|y0XSW5{3tqQk;0WE+kTYGyNZ*OnEfPi6W2x=M_#5aEGJOwyd`zIPH{uc5j zdA+@5J4oCBsES=e5g3I@^UgdIdXy4{g@v2sz3zBb`}+78*Z7_|al-tQEbH~lN{j5# zv9aNxbwGDUVE@F4OxDG7X<(K`}?!Lq+!e%W;az|bpqaI82YZq39M;Iv(!EP@Vw{?^c|<+PXPfb z3IgzXtRXR*nhS76vPCFBjqt$h+-fBu>+rBJxO>u}c-Dj|G5|vSl`CNw%Y^hACqgPf zZWB)WdrWriNSqZ&-C?OD!p~1*RduQXC>W={Er+9m{+yf3oQ zS5j1UAeEs^BR8r(KD3x*= zI5uuP52J=#fGL|K>{j3^LaThIhyXAN#_hKwy1Jc@9wq9WzcHhVsznvym7?^kUXzw; za360rgw2}+het++hK8tCuU=hn(W1Z20)lD~ayo|%W{>$&U=oF(rCAf`Y!3qGWbW2- zCeIK+j;tS0sp?4hP~JWd^J1c_OXz*dAOUQsCoaUsO29U;yu6%e*Dhu@aqFkxIwVl1 z(kZJT?y_x(}OG}0h>m$ zg`u|Eh!8}Q7T4}Gn*{Y!^QS#G`FU}Dev`)*$E^~#VY32xoy{}uK&z3_(RG)2_u`RJ z=OZD`;{?{LLMufw?@Z~8mew~sBgDkbeG$4QLQR1d78a#)0dt?DKnKCki2G2E(XUF> z#8Fs^w-7qUdLN1doq7U(;e5ytx+jrwabn%BOL(W`@_?&W;9x^|X{j_WA`BzBgP>C9 zakkHRp)Ez88TRH8h*2Q%S0KG0+>PJ~kv>mPnOjVb;axif8adyuTo5?SIAf*X<}NHL z*#3k4UTJ9qCemXdV8O4Tz=3M!;JZ&)y>kBFi1RAl>C=l|qaV!Zxp%80xO&VCYoPDL zw|zS;{_jxSB90;5{}vc*j-98MB6oIo;KJj#Z#RM7ZGfWLYBW>`d)*u)FE0;px!2B) zuhMra4vOd__=Sn&Jt$zC?Que53%CG-#cJhb8;WV=Ng?PR%i=&612jVPlmetzXWL=BH6H z0g^Fc1sXcrRG!SC@C3|GY{la?F2k&9uF}=Cw6yJCrYWizMn_T4zDtwBfP*^1U$6&Q zDPdJrRbWdE$}Vz9@24;zq7GwiHqSxZ)z;Nj1nYqxU84t$=^kURc&SwV{QM%x20jh) z3+X43!UNLPXbwXDj2QWM*j;um>*r?}3$dbNWq~?1v*dLhxJKDHq+uqWd_yF~r#NME zd(L6PtrR!_7nK(3@Pz+%i~yvQ$&C{$)YhzJiC;+{k$*?bE5 zs2zxGW`4##@ejVoh~qeskiSXc0q2qEhNCA8Wo2a%?UhkTq1QB!Xnl zF{^!_ia3|m){nM^C0M9DLjz*%Ab3S^E04N zEk?dG8MJ1}ZCtf-WsY02^64@P51@0RTlC0Bm@e_7fHz*4R5;Pt&>+O^4xeA)i>&PI zLX=1e`$nbiNvIySt)R>Rt>tp9Zo8={h)1Z71+Q`V@YQ=(y?%!{8mfiG#Pmi^AvaU0 zf=$Y8y&T5CCnU6QM-rZp{IWbMt7aKp`(DJZm6Ucc-ialqn0$ocC6(&&-ZI^HH4xIn z%MA#~jg`j920m~YzuX%({rpj0G_B;#q?>jNjN)2mX68gs8{2Omo>PYqu-sXue4uwR zJJK01TKEenkGkj2_enXuQib)iP%mr6SI}5~KgQzkEuqk=12+H%;zOgJEZ-w=AWA=T zX}oO#HDJpH!#y+a?v<1i2($6z1iuxjB^Y~QZ!rj~!r zNm)_eKOi})?)`lh<|d@kQa3Ok8!1nLw2ga2%pZ5m?E;g6QaAfv8-O&Hhi%=8(xxi( zjCM*$Fyb~1olRIo({`vKV6VaBXuq|0wL>plzwm-!1(l;5bm~-DacSur)5q9=A||#6 zFGE{CM|U1*YU`>ET-za;By?2rx|iB7TcTaAl=` zqUBvR=uE*c<#f3Lu6CA-Gmf7}i8H?M;6aADukS66)xNx`==%N?QVpEjX~CYVLNyUv zEU_1qz_(?b=DvWHDCOSI1x(S#R=x_WPSe_Y3wES*|IJqe1Kc2=>ya{y#UyA#kdCrv z-8=5A2@yvFB zSx5&0LLxtp9%A*QM;Ie9E9{;cACCg9DZ;i!Z{NNrqb*5js}hFD4kZ)mprT4-U}O8) zHyNW0_xDpfB@bDZ86>qu$$&O&g1JR|<^i^_u&{J+Vbim-##Nt*11}jnKvE6E8P1V6 zCN@NmF70%DKN#;pL}k=4x4ij>_+88XR@E81;3(`uwPiztMJ4(Wl7z7h(N2h;TwU zFl#k>rHVZ|W&kf64jed;Z~e%8xHrVDigv8b5!7o2CG1dZwind0O4iGby6>lN76gMyhac^9{qU z9K+R!_uBjStAT+)*+^6R2Z6Z0CU8-hW%2t<=csIu>Wu7iZm#e!+G<4L z>60f^VAMI%9!jDb;TI5Cf#QNVg%GO&9NW@lPvTpp%OP_-5^s%1MC5?Oy=)FhYaYWQh6>y=dFwc=_eNo#G0IzlVF2B?DEXy}9$mR& zMLMDZg5Dm86kuz>jikuO%apNgQNDc}I_wyMCyOU&erPqCsXAtU7UdqdympAWySqb3 zfCv#n9OAy0r-`_@IP!mOop0Ki`Dxw|!I&AFYCi+lS_s9{_1~!KMlKHI>`1`x?Dev{zrH6#62XmZ)JEVbaY9dM#E|7<40nl zK7_Nn5hW!OXnL$fZv_K5S#>4CjSXl#4PuSG6i1XfGTydDpfhsf9#g znF>9$5iP}v5_ zf+nO;w`?5F_X9vT!VQH$YVypmFYyk?;E4xm!D#>%(s%%i#TwxDhUAn@l1FLp-Mi-j z9%WRvdC#6b=Ni;M|dX zVqCq9%}#n;&8+U=>SvI`>~YAk9Kiu0pPWYqMy*E#0K9CAU(Xj{;FPR_O)u(N-9EQc z$LxoEgx}S&l$%QaOjW18s3Bsbpep#LLAW&1(Mp`0DJUr5jn}@EnrdD62RIJ*d)87c zxRCF#hIjrB?r1mSeMB_*i9Rjv8K<~3(sl=?P9Ecn-b50gsA+L}-9^K$5!Xs@sdQy6 z^`o0oi&FH}1H2KQ0A90Mi-67}z;5L|z*fG6*I(}cJ?mTNWXJYx$3~aRv^`rUQ7O4p zvwEXl8FFvxs3{`|Oi7AUb)u*7?y1)-0~?!1zj-zIfL^hz4cd&F-S}6QAzO6SI+x0u zClUCsC+R9EDDYvc0%3<&RVfCcBjNS_qnYQgJpk+9j*B#`c!)SmvIX(H$^Ee-rZDYc zuAQ4(EU630+zd)sSTsZfMQ6|Ml9i39tyQU6nzK0bd!`AEt9qQ!FYL2^c)}{SVHjL( z=>b|yj~&6kN6lh5;`a6cf4uhK(9l`nEP4z8QBwO{ZboKHv2K>V$R(fQ^_+_Wh&F_@ zv{P2r#9FOL#MRXmtwnMN@ zR20Im7n-8gf}R;xN*v0G+~bQ_j5CO4w*Zi)XJ&+cPKp$vIHjV1c}Tn)PNT2KXp)jq zG-k+(kuua=)D)5i%ji*vfk}g1V?lGKtr{)L(_7iu+Wr<((H^HxX_B(=&u{-yi(SEp z7cSumpq3Eda+hXE0Yb(IIy)a&CZ(1-zg{&rHwPTKy3qPjuGhFF=xONjPQzBh!oR}S ze{8`&PX{ujszE_D=2gQwga)JN|55c{_@0?QzhwA>CIk2M0lJtvUi9 z5Hg<7UTnJ;EP&0YCVz9pEQvGo9$d-+EGXhlk{?nYM==XZ1=*Tg6XFM@WvAGvlHYWC zmdIxL6y41lbLW;(%fV37F*1y?V_k}WFLul?F52T6Qqm7SCM|80iG-a3U3LrH>S#c9^qGF)7FfcB^geX22hIhV zikyMA+fD+OPEK~|nw?yN;tZbgZOf?x+b$1b^n`RXZpgY5=PeO#ZyXROeG>rDh)b6k zOifLpzq1N)+Nm=;-h3TcW>A+>=!soG`m1kiqepr17=Qe$t*Pj{0m9;;AnI}B5%_+$ zQBM%Amw*@4&cpC;IScmAOBZwPZo&#DR;^lP2ivr_I30%aDI96i9%H!Ry$*>D7ffTn z^Mm#Pfyco%9Mpy9U4kO^H$-r-TA5J#g@eLS*U-Rrrsb45o;}UpLWjJGJH7@FJ+Q6o z&>4WO)DbKqhp)XA-j2j0lCV2yXp%lcUritc(H~U6IaLbj0+ z8B{``km_LGn!0#QO#KeO^82Mr*mQ5d;v<6NOXQlU5R(L9n7!@ zfr~R;t6S&mSwR4*N&Ls%+`^57S$CK}V1SHy6e$@d^L7+v zHk?75oeVG7!tH@DT;z+%&%HW;&^v?B5D8C*_B;ESc@y`Oc`smNX4K3=zG7hvc9`3y z4uMJb!QL)}2P##a>yjF7$z{C~7#=Jk+At*#)_qTbyP18c45AM+F1QY_gK|)QOOmuK-0D~nrd;Z=vjrp9+ zp%KMHi3I1K{i7C;8BBH&_#IZJD0PQa^PXxS%tuj&5csg+IxnA_fSQw>_+R3mkuHzW z*Z~Kb-sfo8&P;kT*t*R=zI8W%T~DJKdKNNcfX%a;UwWUOEO=P#IiiU{RN=NTe!luA z(tn+b!C@72Z`PdXua3ef#hBt1;fI(vg`+r{u|@;8#y)@^a&eoPnR$E9VXb-D@#n*= z@n?rQ{yNuz{yCH}Yc-LQ&|aBAT%S}-5uU+}Db@%VffVRSkcJo}0F4H0$M3ig)Bp`? zVSR#eh{9Hdc?%nw;LXY38RCa9zWUdwLN?QjWLbK|k827(!o*=RFv4)_pI@K2ft;S8 ztcNwFX^EXi#n+?9wKZco!clb*ttB#}RZy^F;pfyvSdqc}SYu!aa!Hm%9q;dt|J|J7 ze>MPtZaew-e=<1@K>N>+|9D#Rpa0-L9M1m#_DA$AQ;o9-2vDcRs8R5-PeV_=Q0>UM F{{n@sL_Gii diff --git a/configs/experiment/sample_enhanced_conditioning.yaml b/configs/experiment/sample_enhanced_conditioning.yaml index cd787f6..173d580 100644 --- a/configs/experiment/sample_enhanced_conditioning.yaml +++ b/configs/experiment/sample_enhanced_conditioning.yaml @@ -26,21 +26,21 @@ init_datasets: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 10 + max_datasets: 20 label_override: "ALA_ALA" # model: # conditioner: # N_structures: ${init_datasets.total_lag_time} -num_sampling_steps_per_batch: 2000 -num_batches: 10 +num_sampling_steps_per_batch: 1000 +num_batches: 5 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/qiutegoj +wandb_train_run_path: "sule-shashank/jamun/qiutegoj" # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last @@ -48,7 +48,7 @@ checkpoint_type: last sigma: 0.06 M: 1.0 delta: 0.066 -friction: 0.7 +friction: 1.2 inverse_temperature: 1.0 score_fn_clip: null diff --git a/configs/experiment/sample_enhanced_standard.yaml b/configs/experiment/sample_enhanced_standard.yaml index 027aa23..51d83b9 100644 --- a/configs/experiment/sample_enhanced_standard.yaml +++ b/configs/experiment/sample_enhanced_standard.yaml @@ -34,21 +34,21 @@ init_datasets: # N_structures: ${init_datasets.total_lag_time} num_sampling_steps_per_batch: 1000 -num_batches: 1 +num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: sule-shashank/jamun/9ewlap6x +wandb_train_run_path: "sule-shashank/jamun/fh9o4mme" # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" checkpoint_type: last sigma: 0.06 M: 1.0 -delta: 0.066 -friction: 0.36384343 +delta: 0.06 +friction: 1.0 inverse_temperature: 1.0 score_fn_clip: null diff --git a/configs/experiment/train_enhanced_standard_jamun.yaml b/configs/experiment/train_enhanced_standard_jamun.yaml index cc8fd01..83aaade 100644 --- a/configs/experiment/train_enhanced_standard_jamun.yaml +++ b/configs/experiment/train_enhanced_standard_jamun.yaml @@ -36,13 +36,13 @@ model: sigma: 0.04 arch: n_layers: 4 - max_radius: 1000.0 + max_radius: 1.0 optim: lr: 0.002 trainer: val_check_interval: 0.5 - max_epochs: 100 + max_epochs: 500 devices: 1 logger: diff --git a/scripts/scrape_grid_points.py b/scripts/scrape_grid_points.py new file mode 100644 index 0000000..5a58bbe --- /dev/null +++ b/scripts/scrape_grid_points.py @@ -0,0 +1,244 @@ +#!/usr/bin/env python3 +""" +Script to scrape grid point information from trajectory filenames in the enhanced_full_swarm dataset. + +This script analyzes the filenames in /data2/sules/ALA_ALA_enhanced_full_swarm/train to extract +grid codes and provide statistics about the dataset structure. +""" + +import os +import re +from pathlib import Path +from collections import defaultdict, Counter +import pandas as pd +import argparse + +def extract_grid_code_from_filename(filename): + """ + Extract grid code from trajectory filename. + + Expected format: swarm_1ps_{grid_code}_{traj_code}.xtc + Example: swarm_1ps_042_003.xtc -> grid_code = 042 + + Args: + filename: Name of the trajectory file + + Returns: + Grid code as string, or None if pattern doesn't match + """ + # Pattern to match swarm trajectory files + pattern = r'swarm_1ps_(\d{3})_(\d{3})\.xtc' + match = re.match(pattern, filename) + + if match: + grid_code = match.group(1) + traj_code = match.group(2) + return grid_code, traj_code + else: + return None, None + +def scrape_grid_points(data_dir, output_file=None): + """ + Scrape grid point information from trajectory directory. + + Args: + data_dir: Path to directory containing trajectory files + output_file: Optional path to save results as CSV + + Returns: + Dictionary with grid point statistics + """ + data_path = Path(data_dir) + + if not data_path.exists(): + raise FileNotFoundError(f"Directory does not exist: {data_dir}") + + print(f"Scraping grid points from: {data_dir}") + print("=" * 60) + + # Collect grid point information + grid_data = [] + grid_codes = set() + traj_codes = set() + grid_traj_counts = defaultdict(list) + + # Scan all files in directory + all_files = list(data_path.iterdir()) + xtc_files = [f for f in all_files if f.suffix == '.xtc'] + + print(f"Total files in directory: {len(all_files)}") + print(f"XTC trajectory files: {len(xtc_files)}") + print() + + # Process each XTC file + for file_path in xtc_files: + filename = file_path.name + grid_code, traj_code = extract_grid_code_from_filename(filename) + + if grid_code is not None and traj_code is not None: + grid_data.append({ + 'filename': filename, + 'grid_code': grid_code, + 'traj_code': traj_code, + 'grid_point': int(grid_code), + 'trajectory': int(traj_code) + }) + + grid_codes.add(grid_code) + traj_codes.add(traj_code) + grid_traj_counts[grid_code].append(traj_code) + else: + print(f"Warning: Could not parse filename: {filename}") + + # Create DataFrame for analysis + df = pd.DataFrame(grid_data) + + # Print statistics + print("GRID POINT STATISTICS") + print("=" * 60) + print(f"Total valid trajectory files: {len(grid_data)}") + print(f"Unique grid codes: {len(grid_codes)}") + print(f"Unique trajectory codes: {len(traj_codes)}") + print() + + print("Grid code range:") + if grid_codes: + grid_nums = sorted([int(gc) for gc in grid_codes]) + print(f" Min: {min(grid_nums):03d}") + print(f" Max: {max(grid_nums):03d}") + print(f" Grid codes: {', '.join(sorted(grid_codes))}") + print() + + print("Trajectory code range:") + if traj_codes: + traj_nums = sorted([int(tc) for tc in traj_codes]) + print(f" Min: {min(traj_nums):03d}") + print(f" Max: {max(traj_nums):03d}") + print(f" Trajectory codes: {', '.join(sorted(traj_codes))}") + print() + + # Trajectories per grid point + trajs_per_grid = [len(trajs) for trajs in grid_traj_counts.values()] + if trajs_per_grid: + print("Trajectories per grid point:") + print(f" Min: {min(trajs_per_grid)}") + print(f" Max: {max(trajs_per_grid)}") + print(f" Mean: {sum(trajs_per_grid) / len(trajs_per_grid):.1f}") + print() + + # Count distribution + traj_count_dist = Counter(trajs_per_grid) + print("Distribution of trajectories per grid:") + for count, freq in sorted(traj_count_dist.items()): + print(f" {count} trajectories: {freq} grid points") + print() + + # Check for missing trajectories + if grid_codes and traj_codes: + expected_total = len(grid_codes) * len(traj_codes) + actual_total = len(grid_data) + print(f"Expected files (grid Ɨ traj): {expected_total}") + print(f"Actual files found: {actual_total}") + if actual_total != expected_total: + print(f"Missing files: {expected_total - actual_total}") + + # Find missing combinations + missing = [] + for gc in grid_codes: + for tc in traj_codes: + if tc not in grid_traj_counts[gc]: + missing.append(f"swarm_1ps_{gc}_{tc}.xtc") + + if missing and len(missing) <= 20: # Only print if not too many + print("Missing files:") + for mf in missing: + print(f" {mf}") + elif missing: + print(f" (too many to list - {len(missing)} missing files)") + print() + + # Sample of grid points + if len(grid_data) > 0: + print("Sample trajectories:") + sample_size = min(10, len(grid_data)) + sample_df = df.sample(n=sample_size, random_state=42) + for _, row in sample_df.iterrows(): + print(f" {row['filename']} -> Grid: {row['grid_code']}, Traj: {row['traj_code']}") + + # Save to file if requested + if output_file: + output_path = Path(output_file) + df.to_csv(output_path, index=False) + print(f"\nResults saved to: {output_path}") + + # Also save summary statistics + summary_file = output_path.with_suffix('.summary.txt') + with open(summary_file, 'w') as f: + f.write(f"Grid Point Analysis Summary\n") + f.write(f"Directory: {data_dir}\n") + f.write(f"Total trajectory files: {len(grid_data)}\n") + f.write(f"Unique grid codes: {len(grid_codes)}\n") + f.write(f"Unique trajectory codes: {len(traj_codes)}\n") + f.write(f"Grid codes: {', '.join(sorted(grid_codes))}\n") + f.write(f"Trajectory codes: {', '.join(sorted(traj_codes))}\n") + + print(f"Summary saved to: {summary_file}") + + return { + 'data': df, + 'grid_codes': sorted(grid_codes), + 'traj_codes': sorted(traj_codes), + 'grid_traj_counts': dict(grid_traj_counts), + 'total_files': len(grid_data), + 'unique_grids': len(grid_codes), + 'unique_trajs': len(traj_codes) + } + +def main(): + parser = argparse.ArgumentParser(description='Scrape grid point information from trajectory files') + parser.add_argument('--data-dir', '-d', + default='/data2/sules/ALA_ALA_enhanced_full_swarm/train', + help='Directory containing trajectory files') + parser.add_argument('--output', '-o', + help='Output CSV file for results') + parser.add_argument('--also-val', action='store_true', + help='Also analyze validation set') + + args = parser.parse_args() + + # Analyze training set + print("ANALYZING TRAINING SET") + print("=" * 80) + train_results = scrape_grid_points(args.data_dir, args.output) + + # Optionally analyze validation set + if args.also_val: + val_dir = args.data_dir.replace('/train', '/val') + if os.path.exists(val_dir): + print("\n" + "=" * 80) + print("ANALYZING VALIDATION SET") + print("=" * 80) + val_output = args.output.replace('.csv', '_val.csv') if args.output else None + val_results = scrape_grid_points(val_dir, val_output) + + # Compare train vs val + print("\n" + "=" * 80) + print("TRAIN vs VAL COMPARISON") + print("=" * 80) + train_grids = set(train_results['grid_codes']) + val_grids = set(val_results['grid_codes']) + + print(f"Train grid codes: {len(train_grids)}") + print(f"Val grid codes: {len(val_grids)}") + print(f"Overlap: {len(train_grids & val_grids)}") + + if train_grids & val_grids: + print("Warning: Train and validation sets have overlapping grid codes!") + print(f"Overlapping codes: {sorted(train_grids & val_grids)}") + else: + print("Good: No overlap between train and validation grid codes") + else: + print(f"\nValidation directory not found: {val_dir}") + +if __name__ == "__main__": + main() diff --git a/scripts/slurm/train_enhanced_long_comparison.sh b/scripts/slurm/train_enhanced_long_comparison.sh new file mode 100644 index 0000000..1baa66e --- /dev/null +++ b/scripts/slurm/train_enhanced_long_comparison.sh @@ -0,0 +1,119 @@ +#!/usr/bin/env bash + +#SBATCH --partition=gpu3 +#SBATCH --job-name=enhanced_long_comparison +#SBATCH --qos=preempt +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --gpus-per-node=1 +#SBATCH --cpus-per-task=8 +#SBATCH --time=3-0 +#SBATCH --mem-per-cpu=32G +#SBATCH --array=0-5 + +# Initialize conda +source ~/.bashrc +eval "$(conda shell.bash hook)" +conda activate jamun + +# Verify conda activation worked +which python +echo "Python path: $(which python)" +echo "Conda environment: $CONDA_DEFAULT_ENV" +nvidia-smi + +echo "Running array job ${SLURM_ARRAY_TASK_ID}" + +# NOTE: We generate this in submit script instead of using time-based default to ensure consistency across ranks. +RUN_KEY=$(openssl rand -hex 12) +echo "RUN_KEY = ${RUN_KEY}" + +# Define configurations for each job +case ${SLURM_ARRAY_TASK_ID} in + 0) + echo "Job 0: Standard JAMUN on enhanced_long, noise 0.04" + CONFIG="train_enhanced_standard_jamun" + DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long" + NOISE_LEVEL="0.04" + RUN_NAME="enhanced_long_standard_noise0.04" + ;; + 1) + echo "Job 1: Spatiotemporal JAMUN on enhanced_long, noise 0.04" + CONFIG="train_enhanced_spatiotemporal_conditioner" + DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long" + NOISE_LEVEL="0.04" + RUN_NAME="enhanced_long_spatiotemporal_noise0.04" + ;; + 2) + echo "Job 2: Standard JAMUN on enhanced_long, noise 0.06" + CONFIG="train_enhanced_standard_jamun" + DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long" + NOISE_LEVEL="0.06" + RUN_NAME="enhanced_long_standard_noise0.06" + ;; + 3) + echo "Job 3: Spatiotemporal JAMUN on enhanced_long, noise 0.06" + CONFIG="train_enhanced_spatiotemporal_conditioner" + DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long" + NOISE_LEVEL="0.06" + RUN_NAME="enhanced_long_spatiotemporal_noise0.06" + ;; + 4) + echo "Job 4: Standard JAMUN on enhanced_long_state_split, noise 0.04" + CONFIG="train_enhanced_standard_jamun" + DATA_PATH="/data2/sules/ALA_ALA_enhanced_long_state_split" + WANDB_GROUP="withheld_state" + NOISE_LEVEL="0.04" + RUN_NAME="enhanced_long_state_split_standard_noise0.04" + ;; + 5) + echo "Job 5: Spatiotemporal JAMUN on enhanced_long_state_split, noise 0.04" + CONFIG="train_enhanced_spatiotemporal_conditioner" + DATA_PATH="/data2/sules/ALA_ALA_enhanced_long_state_split" + WANDB_GROUP="withheld_state" + NOISE_LEVEL="0.04" + RUN_NAME="enhanced_long_state_split_spatiotemporal_noise0.04" + ;; + *) + echo "Unknown job ID: ${SLURM_ARRAY_TASK_ID}" + exit 1 + ;; +esac + +# Build the command with base config +CMD="jamun_train --config-dir=configs experiment=${CONFIG}.yaml" + +# Add common training parameters +CMD="$CMD ++data.datamodule.datasets.train.root=${DATA_PATH}/train" +CMD="$CMD ++data.datamodule.datasets.val.root=${DATA_PATH}/val" +CMD="$CMD ++data.datamodule.datasets.train.num_frames=10000" +CMD="$CMD ++data.datamodule.datasets.val.num_frames=10000" +CMD="$CMD ++data.datamodule.datasets.train.lag_subsample_rate=1" +CMD="$CMD ++data.datamodule.datasets.val.lag_subsample_rate=1" +CMD="$CMD ++data.datamodule.datasets.train.subsample=10" +CMD="$CMD ++data.datamodule.datasets.val.subsample=10" +CMD="$CMD ++data.datamodule.datasets.train.total_lag_time=5" +CMD="$CMD ++data.datamodule.datasets.val.total_lag_time=5" + +# Add test run parameters (change for full training) +CMD="$CMD ++data.datamodule.datasets.train.max_datasets=200" +CMD="$CMD ++data.datamodule.datasets.val.max_datasets=50" +CMD="$CMD ++trainer.max_epochs=100" + +# Add noise level +CMD="$CMD ++model.sigma_distribution.sigma=${NOISE_LEVEL}" + +# Add wandb configuration +CMD="$CMD ++logger.wandb.group=${WANDB_GROUP}" +CMD="$CMD ++logger.wandb.tags=[${RUN_NAME},enhanced_long_comparison,job_${SLURM_ARRAY_TASK_ID}]" +CMD="$CMD ++run_key=${RUN_KEY}" + +# # Add experiment name +CMD="$CMD ++logger.wandb.name=${RUN_NAME}" + +echo "Running command: $CMD" +exec $CMD diff --git a/src/jamun/cmdline/sample.py b/src/jamun/cmdline/sample.py index 096748e..ac4bdf6 100644 --- a/src/jamun/cmdline/sample.py +++ b/src/jamun/cmdline/sample.py @@ -86,6 +86,13 @@ def run(cfg): # Overwrite the checkpoint path in the config. cfg.model.checkpoint_path = checkpoint_path model = hydra.utils.instantiate(cfg.model) + + # Set default graph_type to "fan" if spatiotemporal model exists but doesn't have graph_type + if (hasattr(model, 'conditioner') and hasattr(model.conditioner, 'spatiotemporal_model') and + not hasattr(model.conditioner.spatiotemporal_model, 'graph_type')): + model.conditioner.spatiotemporal_model.graph_type = "fan" + + print(f'Checkpoint path at: {checkpoint_path}') init_datasets = hydra.utils.instantiate(cfg.init_datasets) # breakpoint() diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index a13bd53..1904fdf 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -472,7 +472,7 @@ def forward( device = batch.pos.device # Step 1: Convert spatial graph to temporal graphs - if self.graph_type is not None: + if hasattr(self, 'graph_type') and self.graph_type is not None: temporal_batch = spatial_to_temporal_graphs(batch, graph_type=self.graph_type) else: temporal_batch = spatial_to_temporal_graphs(batch) # default to fan graph type diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index 06fb008..4b9707f 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -195,6 +195,8 @@ def aboba_memory( M: float = 1.0, inverse_temperature: float = 1.0, score_fn_clip: Optional[float] = None, + cleanup: Optional[bool] = None, + sigma: Optional[float] = None, **_, ): """ABOBA splitting scheme that updates a state history.""" @@ -235,8 +237,17 @@ def aboba_memory( score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) y_hist_traj.append(list(y_hist)) - y_hist.pop(-1) - y_hist.insert(0, y_current) + if cleanup is not None and cleanup and sigma is not None: + y_current = y.clone().detach() + _, orig_score = score_fn_processed(y_current, y_hist=y_hist) + y_denoised_and_noised = y_current + (sigma**2)*orig_score + y_hist.pop(-1) + y_hist.insert(0, y_denoised_and_noised) + y = y_denoised_and_noised + else: + y_current = y.clone().detach() + y_hist.pop(-1) + y_hist.insert(0, y_current) return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj @@ -258,6 +269,8 @@ def baoab_memory( M: float = 1.0, inverse_temperature: float = 1.0, score_fn_clip: Optional[float] = None, + cleanup: Optional[bool] = None, + sigma: Optional[float] = None, **_, ): """BAOAB splitting scheme that updates a state history.""" @@ -295,8 +308,17 @@ def baoab_memory( y = y + (delta / 2) * vhat psi, orig_score = score_fn_processed(y, y_hist=y_hist) v = vhat + (delta / 2) * psi - y_hist.pop(-1) # remove the last element of the history - y_hist.insert(0, y_current) # present point is the first element of the history + + if cleanup is not None and cleanup and sigma is not None: + y_current = y.clone().detach() + _, orig_score = score_fn_processed(y_current, y_hist=y_hist) + y_denoised_and_noised = y_current + (sigma**2)*orig_score + sigma*torch.randn_like(y_current) # clean and add noise + y_hist.pop(-1) + y_hist.insert(0, y_denoised_and_noised) + y = y_denoised_and_noised + else: + y_hist.pop(-1) # remove the last element of the history + y_hist.insert(0, y_current) # present point is the first element of the history if save_trajectory and ((i % save_every_n_steps) == 0) and (i >= burn_in_steps): y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) diff --git a/src/jamun/sampling/walkjump/_single_measurement.py b/src/jamun/sampling/walkjump/_single_measurement.py index c3c3c6b..d846cab 100644 --- a/src/jamun/sampling/walkjump/_single_measurement.py +++ b/src/jamun/sampling/walkjump/_single_measurement.py @@ -116,7 +116,7 @@ def walk( if y_hist_init is None: raise RuntimeError("y_hist_init must be supplied") y, v, y_hist,y_traj, score_traj, y_hist_traj = self.mcmc(y_init, y_hist_init, lambda y, y_hist: model.score(y, y_hist, self.sigma), \ - v_init=v_init) + v_init=v_init, cleanup=True, sigma=self.sigma) if y_traj is not None: t_traj = torch.ones(y_traj.size(0), device=y_traj.device, dtype=int) diff --git a/test_wandb_scraping.py b/test_wandb_scraping.py new file mode 100644 index 0000000..dff198a --- /dev/null +++ b/test_wandb_scraping.py @@ -0,0 +1,65 @@ +#!/usr/bin/env python3 +""" +Test script to check WandB run scraping functionality. +""" + +import wandb +import logging + +# Setup logging +logging.basicConfig(level=logging.INFO) +logger = logging.getLogger(__name__) + +def main(): + """Test WandB scraping.""" + group_name = "noise_check_experiment_multimeasurement_vs_correlation" + project = "sule-shashank/jamun" + + logger.info(f"Testing WandB scraping for group: {group_name}") + + try: + api = wandb.Api() + runs = api.runs(project, filters={'group': group_name}) + runs_list = list(runs) + + logger.info(f"Found {len(runs_list)} total runs in group") + + denoiser_count = 0 + for i, run in enumerate(runs_list): + try: + config = run.config + if 'cfg' in config: + cfg = config['cfg'] + model_target = cfg.get('model', {}).get('_target_') + + logger.info(f"Run {i+1}: {run.name}") + logger.info(f" Model target: {model_target}") + logger.info(f" State: {run.state}") + + if model_target == 'jamun.model.Denoiser': + denoiser_count += 1 + logger.info(f" āœ“ This is a Denoiser run!") + + # Extract sigma value + sigma = cfg.get('model', {}).get('sigma_distribution', {}).get('sigma') + if sigma is not None: + logger.info(f" Sigma: {sigma}") + + logger.info("") + + if i >= 10: # Limit to first 10 runs for testing + logger.info("... (limiting to first 10 runs for testing)") + break + + except Exception as e: + logger.warning(f"Error processing run {run.name}: {e}") + continue + + logger.info(f"Found {denoiser_count} Denoiser runs out of {min(len(runs_list), 10)} examined") + + except Exception as e: + logger.error(f"Error in WandB scraping: {e}") + raise + +if __name__ == "__main__": + main() diff --git a/validation_errors_plot.csv b/validation_errors_plot.csv new file mode 100644 index 0000000..db13c9b --- /dev/null +++ b/validation_errors_plot.csv @@ -0,0 +1,10 @@ +run_name,run_path,validation_rmsd_squared,model_target,data_target,subsample,sigma,total_lag_time,data_datamodule_batch_size,index +noise_check_spatiotemporal_repeated_pos_m2,sule-shashank/jamun/jjx6l8fg,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,2,32,0 +noise_check_spatiotemporal_repeated_pos_m4,sule-shashank/jamun/psxh4cl6,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,4,32,1 +noise_check_spatiotemporal_repeated_pos_m5,sule-shashank/jamun/ss5qjtb3,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,5,32,2 +noise_check_spatiotemporal_repeated_pos_m6,sule-shashank/jamun/y8qzwcfy,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,6,32,3 +noise_check_spatiotemporal_repeated_pos_m3,sule-shashank/jamun/bp5p8gbq,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,3,32,4 +noise_check_spatiotemporal_repeated_pos_m7,sule-shashank/jamun/do28iggc,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,7,32,5 +noise_check_spatiotemporal_repeated_pos_m8,sule-shashank/jamun/feo5op22,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,8,32,6 +noise_check_spatiotemporal_repeated_pos_m9,sule-shashank/jamun/4f0aw2ad,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,9,32,7 +noise_check_spatiotemporal_repeated_pos_m10,sule-shashank/jamun/lik3brd0,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,10,32,8 diff --git a/validation_errors_plot.png b/validation_errors_plot.png new file mode 100644 index 0000000000000000000000000000000000000000..9aba92c1d4c33d9e43f5a17813e5e9ef1337ba39 GIT binary patch literal 125597 zcmeFZcU+JC`#+p0B+6(h7iX-gW~p+YIq)=(ORl2S^0Xh+%_DkY_) zy@%iPbdAsVzVGknzx&Uh$K!fj7rNf#Jdf9LJlAphD_@Y^PQge)LPD}#?%Www{A+@Q zWb4T-y#>;E_?@t`HyJpUNnLQ@}ud?;(Ic*0LlA{NSe>S+JNI8>iAR&=Eb4uMM z_UD)DE_*KiK0fK*;w~*FaU=Va`-sr&4$|fmsR=a)xwsV7l|6hnpP~-m-|Rj??KZ*g zxJ%43dEr{GluIbfn>S|8Z!Ev9>vb5=xjSFzto`^%pknEx%2w7*cmMmZ8qW> z`>zQ`@`;`Q{nsOUpN0Sa>UR|a%iRC_YoFn$>X$h1-+%2kncVl^-}<3HpG)w6fA5<2 z^o6qj{k=_%DSHq8_qXQf`@diN|9k2G@7@02>h1rxB3IezI6I<}s?JqAHqw;zE}~I= zx?p{kcGs>0E9=}EX*CmK+ zjdZq>C5?2Qq3fY{4^h#~jRlooTAchEQW6{-tW=y`QC%(DeE)03o#FE*@LhVz{Ol>K ztY!Z?2>iY>Ucxv2^A!Vb&(1n-A|v}cKi+-!{{6F#j)MIB{Fm(Ps5Woj9L9B#LQ+z4 za%$>kPL7k8aEIxwTeq(M{LH6U=(1DTuJ_Ejb6XOVlAfqO?;jlvuxfiz*4M{!`pyo& z+}!Jrb9vd>$>Kfd86tReZwCejif5D@I;WuUKAc;VnwdE;DvJL7`}d_iJmMBSX-+v`(Gb|E&-VW~xNe%q{%r&Sv6c7eOv-(?hGf6B?H>5_t@*$INUqIm&VP@Zrz-PTUc^ zhWA4`l@1&}EN8dx{Q2|h83uc|GYX!pkKk!4_H@rHDByl!_r2j!l#q2qqHTIw+9g}t z?YNLGE-t0r-Apvx8fS*;!kb%wGr3p3T%GJM6My+8^=@F``?_%Msfm)`1Iu$`0i*}_GBYdZ zUeOn{`Aluw{Zf7-nQr4Fe9OD<>$_kglOFDTYf3$2K>&TGR(TP@ZG97x8 znR#lWw1FRxB`|8Akr zGdDByTj+JC%CTw-DU^>m^2pqlJFU5;WvWF<>WSlz<|uJj{#SbM5?y2}78hNZrM#F+ zR;Ra#IgM`8)YMGVFE}2ltCzQ(MN;qDV;-whoU+0D!h?kF?~{n``*7gEZthf*qVi-M z+>qktJ!KEhoH5n$d!EsoZ5i5rLIhsmj`M?IXNsI8SUD=JJYP5mXiJ$2@ZRd zly`C2DGzq}u+m1U-+`JL21YaAKhO(bVP+Jv@(rNjPCVs9zWDP^meOCyzRwB zpF4Lp{~RCR$-)wxntG_G#G57ZsD62lb;pB{5aW=yBqx6~r+gh9>*LqkInyIVAx={kO>rb90( zw9o`1$YegoouES}{m|ytcM;V;xv)@;t|#PWKpgZdJ@gk}sNcMIZ<~#c4K)po9}eE& z_YX07jz4z&>Mgq0(9j@$MRey*8b>nQGy))pCyVD}$mJGc#!G%%ayeZKFN%DKEb^&wlVzwxyc-D?L%!@k1JE^ofayS$${& zJtJGcJbLtKaX8+a&GOZ&S1xPIIb~+(UWz9D24C`>rcgUciZRFCe3EygJni}P!x^_G$v7J0}_+OWffb5ltx#s+{b}2dVhDPR+j13$KtNH(B|octXdz~#>B=3ZrDg(Lz?Yy zKvHraA0J=m#Q69)aREnP{(ATBUD<~ZJh*~%5-yV@_h@-SQW|4~shE!2D2qR8sjuIL z`%poA!v3<29_vfPjG94g2NJIvo?;$D4bqtX`t=Ld9huDF&MSKG;AvVdsjK}ble*9T zJ%PxW{$(GjHH?xTA`K16xw3CGBkZZtA6F}W)GOesDhU0OBER4z>`leE3P{cO&M&Q69<4#o09 z*O|aNyCdA(+^BMTI$neKI4x%J&z?Oq!Tb%O(Q`2VTJCpFQSpOUUyikAXxPhp%S%Uj zNy+;z5;N%R+1qUCxaf;$enCgy=)JoZC2Y6v_Kv;LE#J}BSkMO|4Pc(0QENc4?C!$gQAE)!B`g7HhPfvWbmR;Digy}t5u0b_a}IVJsr z!E1AJ7cN8;YG!dAuzdO>H{AB|EnBvfqCZ?yfA#d@!fh&+9S#l-MXpJSiC2d|?j<^` zii*nJ;NUxPaeMXj^}h}c`2oL_x3vk$j&mRKrT)|F`bvUviwNw19+Ik>Lcw_4ChFzO zQ^`-C4wU(9&T|;vVNm22UPKZ7@+CK3MJ4I%cA^<|jPf1=%)PoaZMJFi=7Zwm%=qe8 zot*s8=D9Au@WrV6CU>H|qQZXur#3M}0VIJ@Z$Ep+NlZaZhwpMs*Ukg7Dq|A&z;kZo+8)2Z2>wlQGb`2nK{z(j4C7~goNbs<;%o$H?p*R zGc?%v%J_2|(Pm#Kk6ny;8wS6BB0sZma?E;;nP zbJ&bFch}$ll&b2k*w|PC=<}P^B>)7-7!N(;RC?SGB)x@-O4h|i1k`lf%U* zd?n3=OoesREB%7o)*Y`{j~t<7U|^V>p1zfl!Taa#mNmF~GLw>-7e+c6Y}v8ntbzii zIu}jPmoG%W1OgxXU>No)G;0rl#86N}gMwbM2ZQjHo|<3QUwV3!Uuy8QW*8dUOFcwa z+eUYE|Di)$G7L*@gWt?&s%9Ej>RGzun|_R!3dSHlfVU@Zy@LY*0MH=)C}_B%vWo)) zDT%5e=FZG;<4z`~d-#(*+!~aX0W_z>xh^WB&Tv8J-?dv>TIPAK9KX_A=ue>Ia z!YOcNuG-t5BFd-O!*y|SF)F+FR;l{n7%s-wt{m&m!Gq!QWK>jCfY0KMuV}9H%j_T+8#t7S8j_)WcEL1LVwrx$*7SQa9i-@?5 zwleUeFbV6!-TETJ!hGMDHx2-5+zkl0m7dNcE-p^IaO;bUV{u-` z0>i^;$agT`?ekjUIbpjrJv7pl^KyV6ry|o~_<~#)TG+z}4=!an7mxQAxo@GMkk-|sFqwC7i`I%WHdcH6tDIdehb)h>hA`x?wj)TDFwdl zhLh9zPGb>GW*qG7Tk&b-T)D#K#XT04$L4k4+Uzu`zE6Wo85UC& zJQvM%<--oKRs}K~+$ZjGNnL_R_ib2MN0!<7+ToiwZ+WKTX>w$OJ#LgGvRMB;?U zac)KoRgJ>$v%%QV9x-nzpWe%gD%csbHf(|Hcx0R`}f_)uHLX;*?SQ|Y^P_Zcm% zsHSn8dO9%Q-D}HpyA2Es+`QE^G&Df`XpJjwZ$}}VwwDlH<9Yepp!exSVT!uCdf4K8 zx1Ap@&qwFapFiJ)(8%^0>|ec3hCY%e)Yt->@sojEhTN(;NnZQbXJc z&1}0qgU{XFV{1%eCr;ce_FM_TiTu7;uut^bnc0ax!_Qq^T+3_dYWL6(1I+3peyn|b zTG%W6UF5^PDJKhy)Wy-48N^v|{3X0?+qNcv?MzT2pQ^jst@SZ{CTHi(<>cgyfAzde z1dG)fyYAqy8+eco6(m+t|E)IEFD54DW1l;Uz4n)xO<%FT<3z7M1gFs+Z(MXbNDxn4 zRu?nNoGcj4i^nljjiEG^fBbmCdJVVSxIIl9n9x8a?p>$maG0B+;S*^xR$pqAOn9=c zjt--s*9r(Wcda3s68@a?LhPrd6`}WNnw5|}xS(S^bLbNl6AO4O;T~;M7;3oS(HEiV zsd%CK%sbZ$tbZUw;;lRXD$MHN6d#Lg&&PiSpL*9clx|R@qhSCjBy^?VQW@#jZ{JL3 zzlEWa{XsC9da70_drvryJovbkd6f6Wi4$m?0nBaQEAzH8yS;LX)VadxCET*lUAS=E ziiKgh?n8eO##B>@w|89Vgiz5D9-d0hc#lL2mO{nuGZA=gKHz_!u1`3Um7ygHmFi{g zP$R;^1@Om}OHPU-#a?UqY+rE7z-PI*7SK9E3UMUxEu^;?_4r;_Xpd+mIH=~o)$GhNs|3N? z-euQCYV=sj`zU};sWytZ#oy~bHyTc~ zSCwiLqk;={&&&DSci$>=*H+kDczxjgR+T~*`^s@E`~X4u((a?A5APr)*N!Jt)J zftGw0;=;qjqr{wEoH`|!e`|r3M<-eMExD;iJ0D9H^pjQz)tx2L+1?C0X5@LZX%1pW#-dXRU# zD`#8WiR)DuAA5%7Afirwtst6U(wOJmC=CFoxuX~BruF49-fWAe1G3&3tMemG2cgbv z0x7veU7k<~9HNzXm34J(4UH^3$(fClGfC~)g+<6>ly%qpOE)H*MV5Pcyfbw*aU zz8P3gLDe~kQAjz%py=lUz~Ul4F3pgBvB$e_p12!dzkk2%s9RW6WM&DjcPzi~76c~A zTYi3iY&_uP;52G_x2Z`@sshWv{#Uy%^mhNha0Z4!B5(z5n%GF*FdqO(4 zZcnSVKLH`*4)oZAm|xXq42cQr^T1jDz}Ec`gaT&HoIPvYUwTvPt<9rTPH~c+T_c&h zNqUMsnY_HbpqxHvGW(VbG#1Dtyd&OY+Fxb3PQd(Q0_Fm6^|_ABm;onCj=DhjvbIZ)MJP7mm_U2 z>6)9Hvsx}UJ`Tz>sV1akd(rceM=Qe)t5I6nMS#PS>E}D1QdWLY<~leqa10V#;64c* zdnqP5x;v;4dcBTrkhz;+Ljw>ha6x2bp5PckpSHd~dNl2t7~0UX5b?3`*9Tcy%lcOd ztsX^6;I^buP2TDU?x%4I2!DZXILUi`X>*a=9REmRG5(&CS>#i~{AFCP?K;^OH!+SF zV}&8XI(%<9g}V*~%uanxyLf%sS|RG#23*U48DC#t;`*M|KS)b{Q1*UUVTYrXRw#Nr zp-FX4q^GCzLDCJ)KFr6r6K_-wMieA=dEsZLPgM}pAT(_9;SU+^#o83|hDk~)QqTU< zNodSymxceD00$1Z{(TT!X+1q=Leap*L*Z-;sdbUw;;}Z@0o{N+-|>gBvinT)BLVYI&y-j| z9OauYa?t=Ee0t|#PGDGCT3Q#B!TdyDFqA>_%~p!|-csv}NpfAPX!94%!1)Lz=+xoQfb`>Lnw(16&2;Z;CteC;%*(doOkoAW|~eu1+Hl%}-w z^dPHx;P9RF^igZY*M{n#{_K9C{+tx5G~AVb%-7F_t^(4hPqVv%ZQh6IGYAqX{NmSY zqUC@ZFukG%A?@2kzWliW@?Pz|KEo2{8WI|s={&BZEh^fSEVo-bq3p@&JCogZB}N19 z{YAM760TmqE{pHMw!-AHJ1^N(`m#oUZ+GYq%-S0mycH<$Z{waZU~CYuv-+f5r_bz) z1#{gBwKIK5UteG7Da0!#0Fg-ITPHwuwCH%X7rMJY(4DyFGI4{;P`&J@*q=XtPA~3q z3`6WwhN0BK;x94%vAI6PS0@acY^yek-rnBM;YhUeA(!bF`8b3%9Hfnp1UB1tWuJof zq^J|AT2@vTC1P(p@9P5P4L^~9pFVW)#U2pH$C91_MeYj&t66B#uymWy@2>v8nm_kNd3OS62^W5Qj8!tEVO`tb`v{J3D5Mc5cyEEO8-LgmooN z+KNQ> zCJV8ze|$WMV8b4Z)G^0x%hPpp*@j|IT;GAyFD&+&JtMdRn1FDwO1#!4;duVchRU9V zen+S^AniLD82tPDFWGcu%c(q?n05-;MP`KPkhv^cQy&gEd(X9LF8+8HZ~6mgj<_7?RsBDH zYz1({L>z2S*ZrJp#{zYiyK^1p+il42UQ0uCl}^Mxt|@+%p_O?6vhCNAk-M#}>gdX> z&=x=p!T;Z~^?||>$P$9C!o2bhMGmFMX;W^m0H;v-`tw@VU0sY&8eqXzp&AKq1y}Lt zik&V!;LToWb$~YnQ_pIAp_y?XmTIC-j#cMiK6Id1Q3rCOA>ypfzr|~~KvpEmQtsFr zRm@PZcI)}5rsn2d#DTy~QcgUx6-1R4PTdLTpARIC_2YE`fVgOR4FbWkjb=xhItMk) z&CMN0Tem`8BxEmDRn@D&&A4uo?yDhYu!KJ6*&pon{_R;gZUDtta(U#Sc79Qj1Q13O z)S;ga#t>$HL7?;5Of`Ue9x{G?qBjVI%q=X`Y**q;mg_U$8j7NejA2_yiZ zl0k3`jIx($@f2u#5C~secN|S0qPtzJ6AIDkE8171WR)HC6sR|Z3?(-6+>jR^`Se)M zTULGn0d!ibAga^$j*f(D`UzgEi@UqIVW=$(1gd@F4{TNjDz^b9kwQ6C&z`usS~9Gw zyv)cRFf9rl48Kxq>LcgY|B@;0-ld3=`n{?~@}SUdcEsqaD6m9W_{`TkELLBfUIo8y6TnJ6F5l~o8KV;5}F z=I9CB9fF{Dc6Xx!zFff(oodrG^hbm8$GIwU#W$)1z>|TpWab1RNX2gb$f&D%L(~(I zjC`CN?LXkL_&i?C)YaD3wlbq+oeR#T(D(BS3VU!2cA`6F6++uqZT*hAk)Q7T{Mqz( zcXzk$aB(0$FuGBA+xWmh;$rS&HQMMw9Qf%LjEnAS&9KJg#O1>L2npj7PVAi3DTQM(8j zOeM^4OUvs0zNcQjR-IWXi5>(UTVz5PyW_IH`mWvM#Pwf$*w06V4b4nVsbsSR?Tz^{ zUCS9?Q#U&|N3*gpK?jRVXNH}FBS6&Q`xZE7ovtfWgVkYj*s@kP6JCC#Yf_Ky!q&*NPRGRFE! zN)V9xz`0d=dEA-Jqbh6@ACO=CHG9fd5*euj5#hgK)oPfx>#2;f*J*{{b#8c@qX4 zwSfvMsj!?R^jNp<$3wXLSu>9SMxzAG??cdTdbEH$nTX+;wTxkznroog(V{)fg4@Jr z{*gJo+7(w7oaspwUnS7t9#1>9g*MTY-VV)F6K{59CFZTN}SGNEotCXJqirE<{HT@A=Cc z;d1#;o{gbAHrI|vGmE<@AAd3e_gy(mIKyWF z{y%%YCXlbp?uWR#GC1;r7f!0^c*8P1cm8}!K^zB^?g)it@-VU=rhLxLK^#>VzN^LO z#Rx5pB1IE0x`wuFx^$5%Jn!n)TfK#&yS1|nkVnznlMsq`7jqohU;i7EkITya!jPZ~ zuY`m?@Rz*CZ76!DhDKsO@%TO zzQWrHD85M}xr<=bK^r`Fb0~a@>1e%q7MLwGyaU5VrrI$AI#3Pi;;f=6t@Mm2-3>wj z!7ylv@rJ6KRcN>QF2raVl@+u+9&U5EKW;Na3-(&n)$aJ#12_f&qnetZUxi+KaV%(# zzHs68^IHp92839wKJhuzBxl|UNHpM9Kc=uRE_O1dfPg^UO>J#dhS^O{WO~$@O}XF4 z^!uUKTPr<*moU|lv?HlCpkz|e?$%3Vext}!~!)w zvCxml#STX(q;XDS)(DxXEU%IA&4CO|g>=l!N0_`fQulfPE=rtmof?pVs>fFUyPmgX zr~g2&98|M#J3IQh4gaE9Pl0Ge!s1^P#>hy*jbizz9mL$JaO1BrB3K(y`FBf6No zcc5y^H4~~yQhhxSp^#*m`m=oiQi7RsyP%+;cNN*R&sk>M0IdSA_`r!Iw6r>1(7rUi zcd8-I-y0ZUurYjjTsRLQrLNg)eaSrAwp&AG9n{`Eqhk+D16>Fwt5H(m|~V?|!8lDjFk%F}ImvQw0Hpcfh0fuM|;}qE8Cv7Z$!XuA~5j zmTLwwypeJ~l53eKf?IP3-Y(3RiJE#pw{WiK!k6nI3kcP`2e+Zw+oBEVi69T9oDMPi z)(Y@eI9VX%n~S==kTpxyNRMXYful$A37!p6-zsrPS>Af|XtqQd)8ysr%4@Dp%(U54 z`zF74z3=GgNG+84DpsouMk4#PqtMl%Ax2oK=`H@BkRX;f%lSZBjJReUy0_t!JynW7 zDhHzT9-M#n`&!@c^*d2fQ9l=)ot+8V0MwU~yhL&X@kSHmTSBrE688VG3JTh)gX96p z`XN8vjEmEE?gA~I^BP90kn7b<)7lA#1U8aghkg{jK=B17rD`<%LvmxEGY{JOX)2~3<^J|9>)ja?IS$^h2RD{2|+Oe z-U6`MZUBP1I%Tp~&pnqvtaCsqA<$iO%^(hiuM^3516uW{x=bL4hIBVj## z`&=_cy({e}k0ZoeB>3ln2v*_dvRQ83x|KXg#$goh5k#nLHe0kJG&sS$Gyt`VrY61> zPn4m`hO0#Q7b3?imlce3YlYu%8EJk?z(c(W0)MXg54+etIz1qAh)6`r1?FU<80HFpX>%djoWjEOE6HAhO+MI$?%Q|V~>8X$2>GgZFHjtgrf9#cU3kBy6Dojmsat#QA0@FA;l5mX7iX2!sZG&5`i^23jft4TyMKvyjqd zxJiT!8zKm^^@v{Hr-WkpxrcH4{%leRS;{%gR={3i=_cfACP zKpjU?~frD7?y%t{IQ?Yy&x~I--k-_L2^n;ilyGg3oIt<>p)du zg(cS~9+#O#1<0o~FJ@(DBNMop(1<6BR|tVSY$#6LHOr>_<>782kHxEU(NE<}{>}J} za=-f)2E_#9V#(&)A^9``9Cs5P^s#uokTl($;phiT5UaSsNm1;C7_Rt=)iXf{+*EQpoK^W99 zDIPFq|mXU{OQ!s2d>}z#OU9h!vR4 z9e~^fy`LdWw#EMY#00OlnW8_i&H3K36Ca5X4QS~3=2p#30=~=k&hRaKGVC%tRd>08ZCxs%$obEV7Z(~LB$X}@4`Wfi|enk}o1D*V{28(Yc5 z5^Z+z+4w}SzpYuW(Ic~V(*FcG90Qp}cNuz2!2sI7|KwnDXl}ZGfd*}M(;IW8y!?Dd zb?>uh$v?)5`n=T47{JUXQn>>I189ABpa{sAiMvj}PY`p6SvJjs+DbpW>H85VaTbWH z%#U6aotyl6Iu?O$#AvD^_Yt9+&!4qG2hMS{n5y(;kl9LEJ3#i*)`^$&4CHzV+?rMT z1|C=K;aezMW5)mo!!i0j_3FvHZ(QuL^l<3)y>jqUdAzbjfR@0{e~O>>a2}T+5kUse1Uxi z!Kk$I9S@_8DE3}3fktel;O$;+Ks)DfT3WjN1fS&;nnzanPIoZvZRKj^*be|%2;Fq2 zr^uZSIeT&me*@3(8>_~4cEJn+m;2u|N><>a5xP9`=RFB0vr)21C~K|Pmu{AEo14Kj z2(VEt-UV$*8iwap%#xTGx=;xi5(CcneOPl;Qwi*pp3)22NWNR1ienXf$ zo*@ywI(d5j=95`m+7a-4nAFKhkZL;Di`XQS(65@8gRc{U7D#Gt2|JAH7hd2$dx~-F zp(Xupdb7IXU-(X)6Ky_HtV*gn&D>6fN zb8&ergMIf&NWS;r;?tm;=**C9+Emj_=}NkWV{z>IFTFI~+#OGzJh?5T6IASf(v@({ zV1PbMM*6F9WE#LGGO51^8VoT%2ncWgwVW)eegj3jSNjklv@WE`hu_N0%~dhO|3L-y zFC<~x-1OXG-M49Fq2Ea?^vbw6IG!qRnm`}k0eiM~u2iD`!@6zD)9pK0Bq{rUa|EFf zx!sxOXmlv`Q!137_ZTYn$WId03g1|)9*Z<&yLRmYM7XVI z{fsqx@9+GNyC)bGtI?h2f{tcDstxIvK(bHo-7Tcx4<9A)+1ZYGCOFN}BqdzzG4MiTm6k8i1@vsT76UW)%+Bk? zJ3Bfy`c&l(>~KgW(N|^erMqTEXy8S}cEbe7o@cbsX(jC|{bQle_`3B= zPJY*W!scQJzO#tCK}0mtsoIsL4NGwPH^YI{%X+blR1XpJ!y?TVbPgHNhrs+Q9!k_F z^dcrgl1W~}ZTy^L9oaOl%py$pozL${^&|O|Tk-MnWpkJ16cxi$81jF&YNn}6cT2=x zZv#Voc;OgW$Ky2;VkxGdNcqK|zzAAPAai=yIS|15QtOc_2)d~y@B&1>O1RFZH!YynpCsbykW>kd z`L5WlnfT_MjvH&d1_~ZMV!$>5+nR3sylW-a$Z+mM6-ML{j(~Mf4cd=xUdQ2W)(M}& zYh_$mAy$MhBl_h<#L$i?A`VA6N>r>yn%2Y3>2D9Du)HWa!SlelYF>u@)u@3`Twz`-8aFM^UF zcb}uFB>h;d^~|VN&3Jzi;-q!!SN;m@3TW>oI=rnRlmwr#Pl5)=E{awKp7;w2OTe*8 z8uRWh@Q4fBP3D>=dz<^5XI!z@FW8MC3pgTq&g0Q#bECT$7@p6>F2K#8WZ>T+Zc4AHl-V&mYnu}I!5zL>woPZV=}q^OWnGP3 zNc`22)`q1SC)`{1-8ok~uw1xJDY)2e?!`~1wPhy|TVl`y!N>q$nJ&>D)w_$#D1S$H z2i*1Az9kTr&fSHOB7?%h!hSZwIS=9FCLJ|4H@^cuLWnIDK;i^vBDBK6x44WFSeKf~ z^Sc(Uu?~OMi_@!5-jd={gWT^#OaAC5YxltSn&yT#U%!5RcC>9<$?x?O#QZ|ZTlG*u zm5P$|%r6^?S@w8oZ5EAUA?!!3LYK^1(uJ%#o)LKg?G$igG9O0AmOW5srjp43m_a_!7w{6Q7lKoW2kKX0ZbR7A# z0Up3f!bFQlf~l6HXy^{~l=|knNZxoVo?0o_rV^&E5_s$M`rck|Am@R85cztTRvPn5 zu&Jg-+qiw|8tO}`VL$nzYU-QMBhL^lxx59pqZ}W4a`uNBOs0dPqT0@nw^LFI<`*r= zd6uNAzhF*JGzu^&qmRBteR%k9J{&t=4e{wPC&h|ZFvYX?@6X6xvhbqE;s!ib-t!rd z{E$WsfZtA-tI$Dc5QK*B!Zk(*&B93i3;OnZh*MLLbboBOAcQH@;3TZYCvZH7*ke0Q z#5OULWrzERuiakBm{gS-#yqCMv?Q|2$-iM?$$<4LXk;vld}tj!spw>JzP{eQ*l$Ve z@QeR#dRP&z-Ax_oE!e4%emjm!V#a+8v6Tyh@7*L)kh* z#AXHtNYS*2o#^@7uP%jjK+7~ERED#W<_f*IVZ?57NH(B*KNyLK>ogmo${_Qe0LNo; za`J|c&qm_ChrU&{L^#AeUnqIH8aJ=j-y;c|XyruKNM*-tP+BhKs9y6_Z zpl5q0)5g{|^U4>oyn=#y$$A}a4yia%hqDnHaE?7Jh>Xc9_H;%%-*6%wxe@aMiCrr_o+F%tL9k?vu^&rR1ccUGTX%CL zKviIzU@?_LP67=q2;YRLnX^z7Eg02oy5D|CE(wJfQ25)}^^E85wpaSy->M&lY`*{S zD;6&9!Z9@~QIeN02di|Dcx~VrdwJA8;i*`5%FvKRUwLoGWb9uIP<{44JKE=b9v+mf z*h+-5z9&~8cjJv~@-5df+xYUuO{T$H zHKu#Pv8h_pLaMS5mt|P)7M@$u=w38MN6EUBgC@Ea{u;kUPDHuOW3t{j zlT&K19brGb>R5SIe{XP7pJ{?7d@MQEjjqZK?F)N?{5Jq@e$qG8&-gAo6=7osY}_BW zm0tb+;VLGyULqPu02KmY1H=+ggD7;&>e^e(8~Op}N5J|Y&nYgsde|#G6c}9zUU_wu zfzR0Y+ZsH1#9DHGE&!`%%{X0nZWs)KY3m_0Zp_~ac|$e&kAI6uri}k-*vX(W&b_sFa*~0HITztQ*?&sA3kYEQm0_rhlpjMQVW69ydLX*B z5K!y>sqX6P>UxbD2(*VhJUl306(mepD-t_7Jw0reqc5o$53sUsPFxfKt$m!qJ z0Bb<^;d?hDdB}-A5&6>Tr4pn2)PUc&qUZ7RkC|FGvTPu-vMf5&-pJ$nf_2D6r=+lp zIsc4O;;CanXe28eaP8EoQ(4)GJta{Cx$&WrXItKNVJl+r?WjpK?x7Jl19^uwedC+b z0*515D6ee>_kBDzS3HIm>(No=E;A%nyJy-u?OAd3qs&v%u~!WiQ{%buf$jheh(FgY zvLp1vs;LmkcTK_~A|gcWE=tsa<0BR?u~lt?v-dhP^Pvr7VQym4LZ`kdTlJEWc(KUS z8iu(wI@stZc-6{;97mLEGt<1T{Ay0QS7pM+*_t7%s-`vtJtL`Jtz;iG!WMXNn5=zm z#Z~J6n;D$!_mpOql$12(0c)$WmJZBE-oZpmt6{-O0*6U+dL5W@%l7R;FVb*FYUj-n zc?!IutAYT+USjb-B;w&iqt-KyrwN3KR12dNTF@B`UxyqoH7vwoeJv{_@;MR?#>kFk z&0>OB+Kd$_6xnUquY0(b8Z-4lq2_aS#<4?O3RB&}Lrd*A)etC}^U{~2p2BR%P23OcJpQ7-~mzJ7jk&4KS;VI#W~xktseBk9n^{RlL)c?a`@ z^uT-s#?qHsgLz8x2Dy27cnp1t5TQG~jlVP3&QLPv>15NMuzVO89u`{}Dk>^AA`#NG z?0RKWd(5;O8+LAJNi>P5`C!m$@vyG?GNZ{qGOhaL|7e87Qb!#}BLYhdn5`tqH$Ak& zE|e=9O31eC4h%=|#YbgnIs{Db+ZqzylwQ&NB=Ehy!u2H|YRZ;>$hYE#4oE&+n<$wn z?OFM%r=3V-c0v3N@13#OgPk&hR9+yj_ z&sv!XKaz1JoOq6ir^&Q78nD^!Av{6jvN{VLBVzz^rRSQtG+)`wzMIm&6wtR03=^cY z)DM=OhqWxWbd0xS0xheC@=xhoAo-os5I*kfW*E^i+({_W&4Ch)2KkO#Ar*xPB3IuT zKx^018p%{Zg}K4$8Xq04bEqnsOxs}Zorz|bj6?dIL+V~ZbGYYh6A;Aq`yR_nqG%g1-KU+3gh68jW#M42Jdeidpe++R6K=r;Q>B~wyu z5LT>R!8$}I1>G`|en(`L1V3^^A`VGe1E(R61kL}wpPjujL40(D;gk53QY0tfiXJud&kZQ?G;c zybXIdfYoratKZ|NCy$$Wt$44^MV7e=#w7{G)xSxNx85$%NLq??l3D1P@Vbc^8wiX< zuv)GQ13ZW^X$^LDmY`?#t>uNyfqk1nF&?C>O;%80Ia=t2xZ6SA`xVzM3pKHfslX~# z1@{quRImoQZ-&w2-Vj&avXIyMPtyp3x4aUw@K; zbteHBPe;!;rk5_m%&sFHBW^MRfAo40PdW}_ACc~mn`aQ~|JrMb=Suvs9<{LQkZ7Bu z68|1$#ir|YE)p=Dh__75FFi*{Hp`z#!17z%GQ{p{@yf3gK4a~I+TW38A`j=YZa+d~ z(E9V=5FwK3-YCxD;WOBaQz`NClXzV+PA<<`%jqDufV3%=dueHijZ17F zv0GyAu?!IUTKY#+>mWR)5KZK9Jv==fQi{NJA*+o6REy^2n<^?THVMPUN)_2Z!> z;hx{Isf!}`Zs|@D0?Sw{Vjd;PH2ZCV7+ttRA%fxZDVaZS*&sp%%|%E~8}?iHURfXr zMjwaSrSaT8N{ag~ll?c3T>64+BHpJXT6Rs(eR^(HNA*G$ zGjnvU`ZVhvcg*P4DQ=ZGDerwyOdlj8-DrX}OGGujy-OD}&%)v$)>P$mu3mjzwH_Sy z2Qe{D(4>s9snv{S!he4vG}`JT=|o^wJSW7gj^l7*uOVV)OgYAvRKEZEb@W6rT;yX{ z`%Croe-O+2P&YyZPiKt%Mw|nbveh?T^*eFF7O-zi^ut_Dsbazzf*5f4)dYZ)R)KT6 zQBVUJNDiTvvW>&fB;vo*=oNST{K$$|r_XCIqZWt=H5E^t?u4|VEgq8q@w{^hiy@uw zRO54Ugz#7m9=Y>!a%IVKp_E+aPy(|yA0lKjA{h|PDK)V~DnrD){%tAS{?k$%vE?X9 z%p5r+31e;M*cD&S{z$Y8e~KqPkOMWk&2S(P3t>^%f>hHBj@x@#7Ed<0^NNZx@HnWU z{|I}7pNGYAEeW~6WAW>qtXa5r zR+P=nob2o}Q=)=ck3vj3BHaZQa58Q;XbX!7mm)5X+P#@;qmA zZ1G4b5mbnZD{ec%`Q@8obaJ^T)OQP}8_(ffbHA8?mPLppiR=8ion2i6F?PidQ_#^n zh|E^g5Ror+_Wq$ii|8`p@q}bU!%qOf0wI5-Vs`t^ok8q(RDvogHoe!Ic{STcgcWDr zU@!hvn=o>n+I?B}gJ+2ZjLJF|X=+GGHV}Qr56}N-_xRILJvv=(Xn33R|GLVajdpCHftl(WKuMZj2t5^O{=fRQ6| z1x{nMV63-s*Q`EVC-gtKG5GA8CNHbregaPj*aCa*D34K(Q7=4thoJr7nq2OQ`I#&E zKszA|L<+vIL<$Cvt-5Q~B3wISxkpY1i~iXExrKCvWn3V^6rZa-Gj53D%i@eviWZa` ze8H654L-RI9=%)_W+kyE<6-si2Og?ng04wCumFofud#C1*;1B6?B$`a5ZJjV&;Ilu zy$KKEG8yIA=pyiF75_gWI7_Zho+j3cPT!_%67m21@#79S%=}jD+1LG{kzr!<#5Ck0 zR7-3I56i1Tbw7=oFfD5(?gz~?ihCm?77ekYyD@_i5t4Xq*}{Ph5$-xBKseswMN+8q zG71WT-vscXR~Jft3*|{@SJy!c|Ac1(#NFJbJ-BUd@MA^A>4_I)p3Ad0iS$zp;h(@s z{6RUOSK!QphcLtz$;W06YXiq=ul-(X^nQ(VA|6wSczQ4bKsC*HILviIx&(JwI84yTb!Lcl1R`*U>zp<5mA*?DV##)_8BWg3RA3iW z3WoI=m14q%g?5k?{s&+i;KiQuS%(-?FvJP(7{T`#<&I+6nf1bf6I ztbf8{uoRl`*osQ3ixfL{R1wM{YCRFhGNAf^-=m+bL@%M3BBJSV!e{iupe>W_IL8Du+fAJ?%ZOh!!M3Sldd}Q2@6-l^E`(2Nx5a*K;kKV zlAg=NBR()#k@2j?TzQR0lf_DT8}L7OM#7XHSuG>OLo7%_VGxnP+L5-hvT|MqX364o z0v2NcHN6c7htS3e^9lhrY|N8PdawL^nNiH2`m0fLg%aQjp$JFkF2L4L#IhHbHn6C9 zBRH4}Pq>-{>yDN5WJIeE-=v4sB`qqGcrxDh^yvMNBxZK3KQ2 zk=F3v#ja3nr3~%Ks^b;U)R@AO>D! zp$N5CphXfV3lBd_!tM}uo_0dqrV=cASX1UlT<2qgQPrC%^oE$omqvoY$`HY&*7jOp-!E6CoOC z*ik4cg}570Nt%S_$u>4enkb@@sKnho8ydJ%5t;{;28}As(|4}+eunowzQ5r6p5r~< zy`N{>b>F|=b**)-bDis4`peQXGMbQ3^WeAE7hA+=`j|3veh;`Pv>RVx-i}yhwAw`K zAuYoMm%Fs7@hO^Eb%F|(nWK*>!yEItuAshLO}7oP4Q6LEP>L1?K3SF_Q znsvBE0>BjOx#VcN5CC&O*Z|T7)2JP^eDL>JNv1LgwMv>UeP{)s=3dUO5L1}`{O%5u zrO3=)DC+^8SAl?k=W^=^5-$;F_%`s{VQB8ded`67ZjU8K8xgVq@R7zT3xV$An4X z3hf44U1tHTi(p{DgNlN+@PNoUK+OLheY@=$Z8&VrhNF!`t#V8v%WO(!W zB9Ed!_zf956|OL49UWIZoRfNGr^YX0uwpUtmR}?iTo2yePOv~evlgs+1v4`+3n+T41d}jgMnk+b@O1aXx@)fhQqDu?bbrJZ zWHCmBzkYMTNNvKVr-p~-zP|uzW0sl*%=mVRThzEAjDGl*{+EP2JJpo1`4k)z0d|J9 z7V}ri%ms(g>Rtvgr2tmjXxT-Yr`q;H=??lTr-(=6kzJ=W4^MMCaK3U5-j~-2a7VwA$c?pY9yBLIK%$~ zWer_Dk>~@<5R{-`KdW=>@zz3+bf_Apk7vKui4)%tz29P7-uqbw#8>nVP~^~{=^xXQ zo>0u#Sn5H*i8Ke*6CY9MmlGKmi=()wpF8YHu!D6X-nH z*L=`BCJd2T+teOI8IuczX3D6G|S6`n-~qQ zcE~I@F!bg`5Zf~HS-RB%?ES+7O)%cotIv$ovi&{s#25bN&vKRn*#^Y5gdhFKr}8i__-rbs@Uybity`WDZiGa?QrKJI zl&PsF+UnKOU+tJ=S^;_FH8kDjM@XN9uR)H!`^pz&&M_N7t%Es+fQSnho;i*5Nm{or zN4;SD)3OXB9`r;2ATMUhd8E^GBG`XvAfRw=1hmg5KR%SNRGVA!6c5(`gFXb@(PBKL z5w#A~;U^pjvx*lpKy{K0?aI_Ac_{8km=~eB8^bqLMZxjlq0U)G9XTF-0>gD@pdYd8 z4Q{vB!kWTTm_bqF?V-VT;WlTaD35+9?nyF;7$p|%Ozx`~xu+`SYTwhs(FI)FU+<70 z+fvM$4r8{!_)A(T@E2WOgvA9c<1=S&0E3=_zrn2Xr{3t5P$$U@&a%+~YwBBm{Gdkk zJB7`{PJQi?ANrnq4S7^qYn*~*@K)%|YaYG?3EgakVjHGn0-V}Wx+qG>ZvbtR=7SSG z_t4F21l4#J7s7{y82B1eyI;?>FT*5-b;#IOW=_{VmhS z?@|1F-~SxH|1`ifsHd}PlLK0Fr@+PU3o^z0I|hE2f$u2Gc;kUSE)+gDK~9kM!M^gl z^+<0zVQkl?OYmnpjeieffLSO!;Gqoro9h4s2B-jcZF+$*>qo#G71_?a(d{$w zTg#8ZEwY}#{HPRlwoqzB;A=sgf79FhkWZlmX{yPQmIAHo)TG0vGhi)OlZgrmQO-BK zk+$r~Xh%>E>?#CU>u_FWFJBJVC!$@@md4CJX~KL6G`xN-Z^ecAGU%`tQ_{s(3J0P| zE7L>Z1)6#b#VLBEmveIDs3@iDhWw<6Oil_XptdytIB8`*wt0BjfO0db#|1lt=f&?I zgCwMikI)!pUj)q$Zv*eu!OWTD7GxHSlQInHwmN36@@6am2}e$n6e;0I$G3IsVze_Q zl&a}H($fT1-F3ABW^7wQHGp6O&;IkT*8~0{jHRF}3t%cn61w}5+UMLC9Stu?6U+y@2@NzI|g8OURt)d7&md@;jOZxyUGwsHfpBEPKDmeTZ01~}U>NX-45NYGl$AQBD7@3bFdKIYl6neft zfBpIr5)ZzN1Topejq|d~+lhdqKLRu?Xc3f-#6lX1;pQNHx_l5j2(wx{JH3#}snLZ0 zyZQPrezc$|o8k1ui&ZE;)_8=US&?A82sq`~+EGh$vVj2Q&;X37fbw&B+)9{8-+*Jq zNlXEucT8^Zll@^+-{no`1%gRDh4hNf(F*$aAl?vQwStjG;dl}(3dullBknfc-2R;d znM4dVCiSqVCY_1$a+n!}whl3aWG0M5`<44Z+(DPxerIj@ZqvSwf0L#3UV5$wzbIHwC?wRNZM9cj~;K|>_*O-sF6rqCM~@h!4O`Ke*stGa2*wRz*1!o zTl^a|Sp~SBz2Dz2BLKc-%R<}$P+Np}q0XQwt2@%}p>uB_rR$`gpFy&~MT>5dkEgUW zJT(>|mgdR-d7T00jw@HM(sP0F_fq__0|yRV2JMV^2^quhYu2p6$(aeCn|CI(GQ{oJn>#tGwkaaiiGv;M~LjxFxO@VG#YwMG&hkXV3TR78S zgK9M*Vm(={zC5jnBb)c|VVA9qcEogB@f^5KULm3B|{QyEZ$9v7oPc15a0y)}s)cTcdo1okXhVVI_jIw-!$qbGDFw zT)uN>8E(T>oQ18gZZd08U=K^^jsF(yndFbh2kSAIw3+&;^~c*pGaxeoew*8Hm# z#ai>9>_iYrDA#}1f+gzR4OR#Kb;{|l_kO+<`S1V!zyALJ*{v%o@Nplh{ezy&|NTGV zwsVgEf3WpsA8KR2azL{F3?=`AklpaRcA&y&wY^iPBpu~8Sppq|Q9&;Le}4u;!SFPD zLruZt4Van;v_kT{j{69@B;Rcl`i{!4tX?{;#Z`Z8_+KobLpAa=IWdFd{2k>YHg`Z! z!H(hShR)9UIK{O3qOPuPr1t3lcK3`w5%Xbd?TIl) zN|HqXppg4*ox+ynd($p17vADAU`YvoJmgC)wh7T}bLY&tJ)ue3Z`#igp7J7oWB42V z*!`tySmGd^6Y|e>>Jj+&{D%62-IxbO+r>lhNYu*ZFP?ZW{*B-_Ix>fWg25Sl{5|e} zE-jgUFxwye*TNN#Ku~H12oNAUKMtqCa+mlD$}I>kAT-WjxUdHeJ^Tk=N$bhr8LIJp z=gvJ7nL-=*CmcIo)z{0VH~R zC$zOYHVZ=!4Ir40!pa*gsr(A0DZFu_R1=nY9NL{i{AdSMVv zBUu`kZh!mc4P||@3x_1*@1`;-_TXVP(89Ur8~#)(EUn>=+zWOS)!bWr3`$L4)=r%n zw2(7U`r|JU*|f6-hHiT?3P(&Qd1<3Z1vBE}o{cT(DE_I00z)njVhrL`T3JA|%m86% zV{D8U>qvoq>js7mPz}05Zd@c?$Cf7p`oE9&l-ji zABwy01AubN$z=>6;dXHFAK*_>Ez+E2T50RdKM8xF-40Tk4=4#8I6_G%ZXjMzMl|Z<<{qV1u^;`uQ(h2oiw-I?!*SwF2#~oS7X3;o^|6y+V_aHU}ZC6PE@Yf*N}U z#))s64#;Bl0Mvm0_>e;v*U;VVgR-9VxWKh(ni42m`{7)Le|Q*-3jB&zf;l{A*El0)w1AW%^ntlzMK9akIUM86YlP9=rI#Pvg9SfS~-35Gg1 zFYh(lfcQ0J*)o6E8t_8KVfA2?j`Tz{&J;YPNtJr2=JF*!^gkB&h z_$6={!$=uDS86`FhOwWL4hjx-gAr5%XjwV|i2k_nw7Up1aSSUd91Kame@$hN?t!bXHyID855)%}eQd~(m54rGAEBYS^%vi}KOM6&hUw`Tx^ z(T_kwfgI?8xDU5I4r zAv^l-bPYh)(9Qr9*ur&|0B7`Bsn3X;1J3&`#=3D4qL0O_!0$SFKH~yiNO+81Lk0K^ zLAtVF+GZW|IY=q&D3b*)6 zwOa5D+l8imebLfAe|&=Dgi;2nt|k~zVp2l@w{{r}&Ifte+`yXOZ-&bIIBBG5YRq6V zHVFPHq>@N=8eyu}N5HhhuW!0%Bn3mQTSN>>Jz_nVgwUOkxoU@ex=c zRuOH}D&ai#-8=p$RGLza#_*n2!bZGDPQ9%9m5%;zX^p(1J|wN}cu9XpOTNs+&oR{wl%aTTUH zc(x5;7@RwX2ymifXeZ(&QYk0&mEx&cRWScQ0Q*yUG975mPJa1v2?zyhzRfCM0{pGd)R+p9W;>C-P>7qCnf}kQI z5*~oMa)X-koJ_eEgF#jTA+bFWsB`E$4+M~ZNpVgD##3mL{6xYTZ=i^+sl2)qhhY## zVzp>E6PJy`Pg)VA;6Qhd$h}Fu8`)kwWB0CP8%*Mx?kd;?&YD;tJ^_IPraTzjY=rN~ z5{wI%rY}MPm_z zwR$+Z1-s()nR*;TM;C+jS%#D)HTnSOf;OelwkQu5IlLjI%VP=AlfymXMGT?%WL5AyTfkl!S>w*l-U^Hv$31 zJb({`+FP;OO#z=K57jd{q`bCn{+e~-$k5AlpRZ^z^kRvSCwQf^6SAY-PtO<{y5ZN% zg{JlX#DTz2cbs z%tZD2;*Vpy7>GV2xJcCScvD)cVaknezc5-00#K8CpPM<2Y}D0X{|6H7gwkr&)LvU8 zsF7MY5e?3Xf|>JDIF)=H3;R{I@zO?qem6|w9B$=5bVw8z5S@P!y&Ms?8&YT3x5c^zS#D zb{In$M*fZSpPAwfJ8atBaRI!{{bediSx9HOfVAEZbhcbuj2Wv>yRDSVO>y#e6-o>S zFWuvYS?|JPQ#J*VV|n0Fy(f#K0yzsBA!IhXvj-gG{HjGb__VK~%m>)($A>58joL6t zBMV{q`ZR!6skthEBTK05QmkZCr3X?`OyCcN<0oGu`3|VImv~N3VD~~JDYS4U1n#Hn zGVPgt&_FI2)vl_p_5?>KR2S5af+ZKdujPmIj=3Jgasa4niXM(84(!D%{K{qb?8g9* zbVpo|?t(f<2$!Irsxi(3^pXdpVAw3@I;r!?TB)oM>hp&=QdfuSPP+E*(;JdNdJkf+ z;qGH5BrGEZYJ0}4S+hvI*W5Og2qDIO>*Zg*Wg#n5ry{Ff?KqFUz4>F|#fG4s-c(g- zSxr#<1-s?%9|q@iVt@0DFxy7QRLCMHoeciSP9F~6KPQDTH%?Zt@&nBS);g!=( zAp`z1AYwbd$L29Th_$Ylk<(2cs6>u1i5*!u{QeQ*tuU|peVcusA$m5ZUu>Dl$(l|J zg357Md2XXE_g3u^~q*#ljmoL#P-5 z#tjFi0nNIiyujxBWB>(~-=C;D8l`d9rbSR>XzolLHCS^m7A*WXV>yQbe&npyonLUB z$b1T%xlDNn>%#EQQ)sySfv>ky(h?)HlaphXm~UMQqoKdbqWs%k#M zXfHZqv|$pXR2#NzDQ?S5CwmKgB@%@4XCbJPMFc@=;ek~`@jiKI*PMoc!7mX5&ra~r zLdH3Klx!l&)>XNbW{HtMDqq}~NsDQ~v0}h-#nnDI?emZjZ8|h!{B9*VszVWaJ23Ff z`AIz5a8S?tcuXy|@gN|E`U_XKHG1`tm2xi9*gNG2w+x{wy+8g!seKZexY7jkP-m-? z{VEyD0T`S5Lni47E8D?_IsWTWKxdLohRMG5`GwPKq> z{P6loUX2v9Lx6M`A72{m#{=ZnwLKMp);$=G(=f)}7o>jpd7R65?|y1);JK)WQdnD) z4XthyDGvNRow^< zDOjO-tjC7ESua*_v?8Sz?G^OJ^MPl)w=^La$)iSIynMOg`8tFQm3uqppl%np%UN~g zxT=p78?N+hngc|cbT|5#{q*${6WfNNCMWn=dGtJDdxfwu-Hqu<`@6d4;8^jW2CsuH z`>a{4QW*5i5KFi+BK-)_|mh5!62<`trUb zg!9>^bT=JOmpliqrwWcHQ?vhU-~2atHm0%JAxqB$N#b(mWh7PHJ*M}0I$J@@mGgIfRB%m3)~EADw8imB@0M* zy6O@GlS>Fo5ts!gV=xr!Nc-OC#8ESXPM4kQnfG&5>@sp3T0nLgf%a&b*V5SK{h$!R$ zU>WoTb%SAx;@9MUZ+ao^o!|2?SH1$w(Oytj4QQJ|dqFBgvOi=OgEn~%6kN1(blb^S zY!IFajwOIQdxZE+)iXVAsDYuWI0+Sx-}`c4@s(m{=7W!k9uD?GuLxkt3fyj`F}kq( z(WV0OI`uIRJI|8vS(i{C9qm%(~64O5RY@j&qn~NNY@_Jn_EO7|7 zf)MGychSdm(WR~`4yXkI!o|RFkR-vZE|YQLgAq2%LbgZD2W|$mYAK!H`0y!EebDL< zB-UbC7ovf|YAUt+;>>_8^};D@E#fR*kCL=9CjijACfixN&mK>1KyNo@UmerRvPC*A zy?QK3Ig9T``_V{hhb9)-H`@1#RwAvqtH^Q9G1E!IJOCPQdz&QHb+J?t1^LE#!^{^i z{savBh`~G%EU(a#dKEHT2XuJ@DquS8`OZg!QTnU^pe}LIhy6NqD^g>!ihm13V|Q+i zd#Iz0e0gZ;wN`Kz5W9q<#YGnr4;=puzefCB{$7wVG?7ab3}qI-wYdnJhC}6yaxpN` zxrmzlId>6Vz1GO_f?~S@Ko&0ktP*YtgbwNDuZ*FWM!t`)Z*wM&R(N_)Tyz_W@n)u#f}X zSS{;xA`nA>CJ?Xl0?pCdK4_T1^HC&&I{F*(GY*)S`i%0lE;Q@V@_v-xu>=|r0GidP zOD+cmaiYfnVvE*}H2~g7LeAPL`*aKY4?Q#0v5FLQ52=3j`?H>CQJlEw!;gN$0$`#< zQT#8$75Q=^A2UPOfpL`ryW88T@cIHLB;<}4Sbz$B1N9H7Ck|%IuL~kSOPu|0NQCO? z!WTZkpoUg@qi38@@DYtwG>H|##bWb0Ol1T*49L?Sg?SV5qE>-CdZLvR%YU|5P7T_} zK_)bE!CC#5OJTiJ{{rXAx8fNHF9^`anYF8W{(Enh+gepf~{-bR&};`*@u+KD_)*raky&97#JxX)2{mmW|vJ zC_6ax%Q+6Jz^YcC_?e!1)rpmkW0LVOccD6_EECQ4a6}2FjV3-;udNJ7IxLs)`<{}N zL~CxAfUMrt4=(U1SV(!zDSSbWuM-eco7PaduK7i+Hpf+;bhmG?V-5;NNJ>5$oI~Jy zQ&y(ww;wy}ur^KplWn2!h_y&IFp8an18Wes%rUXWex&d5&%wdKuW%6+SFE?jtMGGz zOaW_==CHlN<#ifq4$>-_gOE+2wKm6E1wuef54Nm_svyc=j7gZh zc+g&}${uJ^MUmo)cwoygp5td%!dX`X3>daa&r+m#!68 zCJwm_NZ=4J20gnNbk1TRy(3@BFw0BPifp5+X-P6%xRKD9J?(HugbqJ8dJ?2N?SuC~ z*r44n?GrBT)?9BZ>N6asehu1VyNMp! zkhd*!97vQ%!D5%5hK6+*v3QuY038HQgv8Pcec=Jx2GBP)I_d#2h-p?2SU|AYr@{3R zf#X-6m#6rtM_ewxNu~C9YckT$fvR@gZqNp>Aw>jY_GFn=ZuhRHjhIN8?~^g*cp6GQ zWX8Rgy14g5UpB(4b1`D)*3`H(tWssre$(Vd#F9X1kc&rc-BN-mfeLm7S~Z-w?4XBF zv<{9~iJd})>Y%-pO>OxK3Wu1zxzKo}Xw!fG8 zbpgqo=9GA2pp!g6qC_8K6sdU#)Q?}*D(4$fJ5NDDalyZ*BL$A5a(TD0cLr?@eAg3Y zbR_KLy@9^IzEY{zI(JY*vRKfcF2gL^QGZvwdFJ~QY?K3kwQN3%eRJ{HLhA;}ZhKJ} zhIau!HWb#Q37TgvihuelWX|>UjaAuQ|8^v=b3zPTwaFQ=kMW5=0jQfaI*dtVU!ij; zkCthS4t5Dw*45VP;!_GdJE#l3sLuS$%)Ow-E~%;>1(xSS-*STy6ek|Y#w)*`Ig)PG zbR+*^5z1@9d*6qWQS4>aD}CNJD|1dec6)r$CIeuiY}H$@;mk6&ro!ni6)OZAdsg!B zY^J$TLG&pOA!l@aNCBI*^aQ1Jp+cOvpt{n`y`0uSG zQJ1fDxqzDnf+oN`1gix>bF3fRy>e+{jBIaLSCHRBgLT&^7Kp^d<&~y<>oW1;0-6!> zpEnN9do+@n3{vmD{fOjVX4|X5bKC$4Cx;Jf3_w z+f`dw?sqhUi4}M?hJL-@pB}OOM>KPA3xu;;h1TtuJr_x1pQY@qo~+lC8(l|3fQj?0 zqY)49iCne>N#W4{RPARxOfWkiNh^av@W8b8{p#h3-?CzZK&=wR@w)3>d~ayfpSNN! z0`+SDDnn5wl(7(63aEQ@*%pHVKE7Of6k12dc?IBl4(B1@g!fhTP@-M^?Je5cYcq^@ zxQyFnjr8^Jdj5jFXR}V~;j%g$c5WS2d=$e1kL20K_=X2HTM=En&gJ+y^4m8+wBYZ& z36rH(@Wr}XI2TiMBUyd@da+N(PM8;*LqCOx$^}Bgc!p>l7mf-aDt}c)<7KE2lr?C< zLS64)FJC)iE2~3EOchp)xzmhHyNGUjqR2z_kSmKbk1uvD1k7zjX7p%ZBX-71WD1oh zNArehDojZE-!A$7^DuX^aNRYL+RvZG)@T_7W7>Wmac|9cpZ2UVMR02gEI)3AiYo_LyY^Cv2Iuh4dw zQfPeRCF+5+tcJREm9)^5hePBnl&(LDdQjLPe!fM#wM>g5#l!nC6dLa*yOk>|D}RiS zqgW)g$WP0MAwBkP+=Zz20%NGD85za^wvQXYa_o$Knf47>AC9Is_{|}J4#ryh4Lyu3-ZMe^G@q?G&IPM{?j6$`Ra* z-B*O&GVVjTNW&~jT4;5{mlKnyS5Za?tSS))$ACVQe@;Zlk!U>x(^&N zlm-qRMcW}>r8OLW;;xfUh{Xb1)?>-pA}A5qq0Y~$PESuijVU<4M4%RuzkG9;mjFkf z*Ctic zc*2=OjA{bqStBsY!BEq0AdjzhfMjQyM!%s|wzOSC-oDRK`pJ93`8YB^YKtN12UrM- zgq8Y|B{SG)`vXZ3vzh_oX22u=RwPi{8VG^EgbZP(^G+rtvl}rY6Ww-C#uLx7b9~j} z#k8+I&&cq7Oh@XQGnfHjPNI?b>LGdh@?|`rUF;A+?68Qh+k_G8)A$3kzTQ==m+{M_ zi8c|v%oBc58c*qbqgz__tq~=P5|;aW&Nmgde4R3a+*N` zQt$IqGOEJ>UEyrc);ZG`K0NgFzbCMpJ%HJ~e@NYZ7{p0rHj>&Z|5lJ;70Koyq?b7& zE`Kl#^;qFXZ}^|hLZ3@KI`qmzBoDYR2Q+qJbNW_DiWsA~P8`jP=FEW+aNqVwmm|dA z$zrEKUNsgkQ3u#B5yTs|wyPNb$FyGYPG$lWmZ~c%)?K@GD0TaF(2VaL4Q3Aclivb= z5Af5L>H=j1J{o`vh{UF@o%_uTpJKWQV;1s$Ef^@2cv0|$Cc6P?_Vq%`oN^QY;qSud z+13BYx~&-ARDxOYRjR?87feiHvQac-II_di9tOae64mmBa*Pstq}|Jj%H8nOHt|2&aT zkMUereZD!L>-5K&yo4+4!$3=;aBcfQjWlUc@MBqUJYhEy+KG>&#HM<-ZOQ;%l(Es> za%`?ab%D&aS2%v-3s*n9L}B?-N0G~XE9WiQ;hrD7a_169PrqORB^4MyxGz-@saSXJ z@XNn0Z#X@}f6Jj=2aH`^zgxN1Wfd}qy3JHc6RdvQh7by4NU z8{8J}+E!0~5Gx^-MU=mJhh5Hksg_`a#?&Ss;oh;7wzWGu2Czs&f*SOV zqx^4sTU?nkb%u?h7mJllE@BWE{qHZQJN#P*4jAE-#&|tHlB3R5+?pK+{@)57??J;8 zSh0NJ_PQbiUzz0GMK$gR>Pz@ehYXq4*)@45ni6AbZ0%xSX%wUg5zXs9PEn z@0;Y_w`E303N>X_nEAfczSy1Cx(~{P?azooL*6@}TnU7w7k)E_!zOph{z#tn!lNz~ zWjnA}qDy|3WoGX__JNIJ0an^1TdV4QA!%=K-+Oh<_}a&vn4?tN&HEc0TW9CBAvQL) z%iA#%RK-__^|-;k=?@&c)7}BHx9K-YRCf;@1dMS5)gZynbo4^VS*gTqbH6C5DTwPi?w?NrTk~3RGj(uAmfz64|cc ziw)N~Ee1>`jW&Fq$V5ZW>fO4Vk^bhUHwq|!0ZRvJ1b(-b0__=h^Fh6)m}Ftq-s>j8FTdMQZS^RpnIP9PcpSInEd4g!N{`Nocod85)clO;++LgEx^mCwaU3Ihi*X1v4mi*0uQeZIHn?mpPGVrPYoc0tygdIQu1 z5lA^)Nn_qJLp5UNfu$-r1sF)mXBgiPP`?kw|KkO#H+yn93&-#$-*Gz-a$7WHpi?Ex z(Wt?$e`-6<@iXw(COPpvnf>@J zhZ#rn&80%dMz8mnf`s3NM;+AbfKz8<_{I@``Sb3(l9KqgErySO{u%9OhjFFPEDr}h zMNA+i7x>N*>EWh4#4>-R*@o$!>vs?c>t>qrD;gRqztYF^KwbH*x z#k?N6Fe9x>tVanN7*+0o3+W+YVd)4RhADncC#;hgU&oRglY8BM{dEc?&gSiHS{m1L z3i0-61G~m|$9b;BpsIbvN~Fa)k`FfBQImB?C%yoQJVc`o?W&c#ZOrmtzv$DrJD5jd z3GnH`cJ33<^rO`zs=B)?G4ZY{tbh9?oBfY4yE{@i7B&QYF?n)7P%}Ach8v23gCVD$ z?g+D{!u^WpUEU$$xy2n%D!xCX$$BGJ3c`GRN3hg~ZVz?m0P9OsB7?yq1+u#BV4OY`N~yj9n33gr z`>9#C?a`?IG)u#TFphygZ>;x~-X3B7y}9`@Pkf4X=w>IGBcH|?_PPb;@jkn|Qxi*C zU(a46qFxKYECX2B@6YTn-W^}D^>~AU2A)!=eal@mj6;_WruJc{pi$MJ9rRYlJag*< zS$3l0AqJggIO@Ae&t{eNrXaM|wm$3VqDVof4|k<5o=Ll}UyKbjn1JId*R|D!d__W4 zvf&zCSvc!HdUx>(-S@|n$;ESoJXxumg9@C@5gzZW=SlOe#96IqFgJfOh{Gmi{*G4P zzO1&$Fw(Ka?jpfL91zc|3wH^`$aaFS2WR!!f3bYSz;T<|JeLkDOY$wlsMvXFd?P_4 zs3$#^sYM$K`Hd-$1}3QfEC7kd}U>}ayd`2?=Z{{?+rlDab3x23dX_7{8R zj5mu4UX`=V=t>?g&za-^RpSdx`M9X!ws-ifs8E;DAMUaMLMKwBiF z*yOa*cgb0Asz;J%^(~*=auMbHs#k;FK-rN290R&=o0K*Gb_#4s1SATKl>2c&t2>cp z0>uC9=FAyZoX;yQ*~E}7JYrPMNj5hUwU>_QICdqmQwTgXFPFzP`|*6k>k-^G+x!_LZ#vur zCR2yNL`AZz-&n}Aszhc;XJ40#c7uG(T^HYG>&$v}dbcqh?kMLlZ-4Cdg{9SIxyAiy z$2a;&OvOhQgrFgl%VS%!1X`ijjs_@HE@8mt zYEDyQ_3=9vST>^K)i}Vb?Y{lRsMX%=^g$0b)6>x;yoy9- zeDbj*&sV^H5Q`1w?wES#wKnB+9%iEG$e_MX&EO3nfYF*CEa%-A)KxxKhTv0U-jS zy-Uxc!#du`nd62Uv_!BO6`&WlMd~_>)P@^SJeZ(uwiEWKD`8UUj?<=yXd`yc7{!u_ z(e>Gf0yRq$mYxa!DSKPt+qNT>W=%_e;MAVc6&csHs7e&X#?z~@Rdx5mvkRvm$G208 zPQ{cCAfFW2aOngZV#7^qrdH;RevM~k$=!YCHP(n2&I850CpNAvA8~*RU3#=FLXr1Q z0;#+C!v54>T0c^38;$@dSU*+fvdnWdHfOleGq$Sso{C%t>(`4Larc8(JozbpD|Z*4 z$CSBfyCXWB7Pp%#S-%`r+U@jf>a*6rKeO9CI=IW5Nf}xBLZ%ap*C<*4!qs)I%k;pc9`&r|sZYc7`fcii5)P>MgMvPd`tZ`){Ktq03a z_PF{OL%l=LLM}wk)kW(ktweOMYl@DpA23I}EWskCu)(#yxNa{he%3)+a2kC_K_e>J z4(@2Gq+^b{^(7AK6((gUdLhHDa#LIg{cfnB_rVK>8dFwZ)b|@XPccO}tXGQ`nzsGx z92z=qJ~kYTAz=NW_cH#>VD(-U^~E-jHH07{qG%QfzV>U+z0{M^#W(%16V4|kVA(C%BfB|`-1YBdNJkI$;RH}g7nzVts|G`O+4zM1tR zZuuFxF;|q@g>4fq`pM?dfGCA*>qhq7Y%>Mkvf%R>0q(6Q|8%eF06QZ+RCDnYf(@U3 zJ%MiMc&rL|Y-#i-oUjluU@9h|o8Y1+tj^v1C^P<%aCKHiWV|y|SD#T+ARH^weG?3s zD11a7oH#FfdTb8Y@nrMLp5Fn}Dwq5dPUwTK9+x#Kjhx$d=)K|Cj)^yag#&g4K3k*s zRn+9{)9nI9^*LEnUCmFUON<z!hqg{x+P}>g)9Sl9e=k`;c(6zLH$Al>Ly}52zoNOTpUn@`Jjm^ zB|FxcfXzoHs5LfX5Nj#&gW#0%_LyU*Pz<`$o5}&92=()r_})RNc6^>Za1cTn%#Fa| zfutqT2M+{`lyIZ544qHaticb)DHB4gGvUXc%5=Ch zp``Iy%=4;qZavkqBj4|&HV0~-&Y2kWL??;NtCVXr8G=23-@zZIVP<7RxPF!?>-upz zMXAuytV3E!ZGRTi`1HvG4??^sC{KeNc!>|3c<~Bi?W5lu^c(`rVK&D0u=9_e8Mu%} zEw}bAR1)Tvb8%w-6=~Mo(h?;CUMcEmhww9(263T;((#Gvv3k7+9-}(RPfD^O(C9^3 zoeMjX#=mCsx?L=m+63`8!|Xov3;7gMA*k4Fvo9ZVpC=*`xOL`-)l8;pQ47y;ntNBc z{YS_3u%9{W1;=%PyNnpYV`wwVtE>w>YEeQ> z*>JdFNj7eY{MtAskecSfq;*~g_pn0xkOO~J&>k=L|0Fa~gEa$qpaJwUr z?9uT0CMY=yrxwi=y6+)dH-EL{)~=z!Ys*yKaPISf!nvk@t@E!>7%&S#|4$t_`N_Lm zx+Z6)q!6nNU~Pc};WgZ$Kefs@)M{U6a!{m(6Dx1q9u4SqvT#)Y z;xdTXF4$Y*`+eZY=S`d0jz~Tn^2Y9~TzuvS8a6F=jaQ6JaBf1~Zvn6dp4SEZK1IKgXHh&R7` zQG*Qr^6&#a={Oi|SYai0AZ~7@37!J$g4LsNKLMQ_L^_&0*N98tM`90t=Rh-uXKNvy z2n?6l{rpX}uELMxtnl`h=4PX;Y`=5JYj+?>j!N*fHTatoq zFkUu@H(_4;@3)8FtGO)2|F-|OHT|op_>Ko_ZIq7w%09+b(wfZaQl6p@6hyc7iGS9w z(t~hi%S;@S{q=FLN$c2keaRr8^Fb`~ zA_o_x*Om(5-t#o;nfcLr7n`h9nbjAmJY?$NN7-RlPZNGz$yph~+)VA}{N(undHsw3 zj>;+xFXR<+A$B;8OnRy-Zg8UD?-<}f)=kdp3b;;ns;bDdjgm zY^Z3y7wvklO%Lq&%;hi)V!s;Cxi@TT{K=ltV@m_f^;d)2lE$4BGy+3$kUMxPC1+wY zFLIqiRSn8@{5DR;*ZbQH_Ep8oYL6X?H4EEW*(;+_=W#D7GCY2Rz#0zwOnab~&(DM6 z0Sow23dj-WiOLmolq|Y*S9-G-U9j8d0AN0#M?IqN}* z#h9k#qnqk17jR-<(aMx4z2ycUjviCTs>~ z@09+Bmir9V4f2np)(F5HD@V+z4LL=XQHRGsrk0F2QwRoLB7yod0UskPc42r(jj=`g zpqdhg#4`1bb*@trun)PYX^j<7{+h{gIj+XX>@qyvf+ej^mvuNoBXA@TV8_EOB9jNQ zX-KqxrVf;8f7Wwsbl>C=d*sX7usHUKYWlkRK>J8z6?0lU@(Z}*o0ge7hLTB z3X!x_cFe4JjyMEl(BdC>rrpf?A6NK`sOv~!B9jP6J>(n;(L_vWF_Ogwv9OY<&@qME zTV33u^cF;|8-0W(2SpOdX7;g!xyrSP4w(tzW7Toir<|1nR&?aEy$pe-bMEcP=lT#asx7 ze+b(c7QjiDX@nO#w9-Y@iVcFmp?M z*)N}S`z1BQ@Q?)yDdeTpImDbC0X3y!p9Kwp6<&Mb{_@kPkG;{;lzZHX%&*!xga0lI zLWp_mvH;h%!SZsuNql%uo~9;;{h|%rO81v#{(Ljo zw+J~};0OfDDj^#{ud9XYrJg^<{!73fs)S2bYqN0yC7ZZ#W~-Ld1J@b>us?2=h1KE` zM)uGYK>y<5RmBV9EnG~QW5Cv|5brKjylp6VnVQP8gS$iVqzZh=!MyM@pQus+ z&l!Jmge4rWGFEg~+WvwL)P2JGzsS4?t!4GcnlCXPh$v<8m2|K#(c0*>*~8?2`F5H5!=Y$mOqf$$j9WpcJTf8@81}0+v{_Hc+?x@>PfVSN_mr~+bLb6Ng_$m$i<0#9{Wp4$Z5fS zqzTvYrZt6;aUR*6+6*3~E|W#0t7M(*XyYEomx#slI>=E*LQ|{zJ`GnOnsQ zEg%K7hdknr+pQ%cfBUL$DL4~}TsM_FAPMvZpPH51qYn&!d#0l6=;yGF6)`*|Fy9T; zw8`T*o{N93+h|Sr$7Lv^{;@yC3YVvKd(fTpZ*Dwel-Y96w3I~^Oi3X-zq<>g&2*Yg_U9{wY568arMqwL`i4kcJ z59+vrfv2MNLKv4ZbmK44BJpQ7KHtA^SKC8Kq#Tqhb0!Dky{a;QM1=_D&DiafPGyHlIa+>3LbeYm%_ZXXQ3N&=lNg{|odvWl3{n?_a7uZuK-) z_tGatEMl1OXO4Aeh(;JVDINQjRl#}&1lD^eY@GkG8RmGb${5B09U|{He5EcTqLnfu zs=5$OzBM4p&XqX}I4#il-=XH>?A7Q0oIX3kQ@ASJ7&;Rgn6G^I_W%|v-en*rkqI(> zJueETrC*`ZQMo;{QF(h)38GHWApcJsy{8Il)q0p(MF&-iDyO9bV#}y!2G;drNDs!P zoT>9=qNX`nHs~te=!mM+2J;nxrCP{uI#zic+kmHH+krwqn^-Tx!;Z_&K|OEP_ULAmKlIFes?xo= z!P||LHJynWM$~b2+s|vyA2X^13>f3FdcG62W!`x110vk?zI#J^xaSt-c}I&L z$n2i}0U-7?J78yGMf7%W=|NGBv1~+k**G(3C>SzdoiP!EiNh;i9p**gvpc8{i?FlY zzZoE-1WnS!KP?oWy>qsL&ITX)c%=VVnAK8s5siYmYAghvv><)dxD9o_d~m)yS|B&6 z-pDZNL4Rl%e}7v;6Ep7Svo1ktplA5g;nn3Cy$Im`nF;kw9Kw~`PAlAQ74Y{4u;3iA z$E@{NlSiLbsJ3(&Ob82zGBvUQX-+;Zk^WgoH5^|;?(bI@)8&-4u_%H^+LO3HKx`{< zAI-r@mINEul+SJHg+|MAh9%upg#3M^^sW00K56NMIUb7fYUBq&Bm7fOK;rC**R!a) zg$}I%Oe^P-+v3`4$=Qq&uzOAK*86(lu9*#F5hRK5mEEOh-yD!=jb0o7?a9t}7eHH5 zAu}z8psPZ~4OxDsXxT!rMTH9O6wqJ+Kd!W`wLaeHTnjXZKg9D1{|;PYj>A+SR$<0x zh-!1uj9T6rdQr8?R58H08iwG&jiQe`Q7*fIxX+7Asncp4MB9U#Q_t>HXHvOdZ%F<4 z#!qaT$!M^1QwIFZ!YIXRLNM4HVHMIrLK}_vF9mOPen~hI>r7;S;745}(}TG{PCS>WE9(hX6wb zaJ7@`#x3b>;VXAtI$O2v8wfWK931ga$;uIYgL!BY5hYscq7BJQcLh#kvQ|85I+K~P zm?-0dkaBac6f1?QwzVfmiqbxol_h{gPelf7d^1l}#j)$2>2r5559SG3^+;V-6{l`K z_IF0fNuMY6+m(6BtM`oKAT?sb!55G%ce+^Cmaj|)k3e2S09wgr?gzJ|@lR8OJ&2!N zy~z2p3#EU)lEjryn497M1QISuc;+8r>EWvq<#!sVoa0LjPd)X6ZMsSzj#f|EWD)8%k(@iRUj32?h=+|h79{w0YnlIO%ER&%b} z6ncJ74GP62y*XU>=9QEk3B@@m0PVPDx?VxWPYqn6?|vvcfW(BELR|mJOZOj zr(|J3CN#T`k!n?@9x9bT>mt5~WvG;wyPSs`BL3J}{m ztIxlhVFg5#2Z4uHxC1=o-3@%CHg;X?g!wNJ<@yL&{;n$1c-o=$#D6n1VALF-V%`0X z0g5oUu!&A%JXFuqB{PKkP@9+GoztOVU3KA`nmxiil4$bMTLrAwRnjM-Ocp z55A+m0U&_;Pzs`|i4l&CH`JX@G1llqa1vC-L5%p8lKC^%cf+w^$s~#hpf;5+>2DTH zHlnuj-X!xJb_AHUTsae6mNbO84SRwTHy^kc{Jqb%*#Bne5sCEU4>hog$#+}%AWElgf zewAG3yHgfff{S#c*%xn8)F?xjAP5qcRjqY7S=xu=S(2n68j2!%@$?C(vr0@H(ZKk`?^C-_ zVawFvCX>iSd1RTI^b$3hfupOzWLRO?wSY=%h`_K$Hl5lYGyqGj>p=UqA0S@6iIiF7 zI{enZYaCD5n}Psz&7Ws=1*CXXaL%5;^6+ATk$i&FI4s@b9h7_MkQN{#oHz`2FESPM zwAcgpXmIXUfV_Ditxf|#Oy5wPBu?B?2WoDAf_V!h{nHDRdSOhxLP~M~{RwxF`soT? zRPR4^FQpOO!b#&cVo%27Pn+1nS+ z0C}ofEZFE_K zQ3%Obq9Wu08ub8?AWgCR3gbJ`Nc4e#IMdhu5ri8Eg2ec7u$1-woMa(WYOJBH5+zg9 zhV#>>|3r#YhE`0Ao%5M^xQ$s`sOtV1P`t4cEvHKlL}9|g`Pn=eWNcc~6l~xlXntHw69vIA%9v%Ob)|M~a}r{99=IRv@Bz|! zOrzc>Rh+GKUFk89ti!(#z;eZWm706d_E9ugN%9-_YGGBJHv^>M-mD(y__YFJ_-_6) zJoC~<+habTNe{Op4c2#fYB^6qQ|yJX9CS80o`@=dz!Tq>!8HWYQmjfmo2ICjmVB_I zD24-16Ky6lUr|N|auT$$dB?JV(&iJAH=l9C82l2NaA;#M>jjWXN0Ckf0^A`-ttprT zG@;|-LxRv=n;5vNiDv%X6N+hoUOXq^<|G2XqzCXBY3@q=tm=siyb)(MVEfe%Xi>Dn z56yZyOrv@_dU8>rt^PPQMTo#uuADJ@_2EW_B8dm^XN{pKTW7Skh7e*K2x;u>*&Q)i zW8Zqe7YYptd^64^Q}gbx;LXk4Nm-c)>j;K4y-N(K?ICTN`Nkk1wgUJA(7eTlshQD> zsp*6QP7&K)n%p-%IN&&eXH~JzSV0?7~A(ks$Zc!1rC7nW%8GV{j~4B4wn9QP@joCD~I?r(KJ4s z?k6d_D9gdOMyNT-FY*7d_nl!?omtx_lSwi$iN+p9G*L84(Fg(pf@UHrASfceS?Ebq`R2Mx=ILmZVJStsIo^i#R{g zCZ>SzY`Qp{?`6>Wwhcd4TbuLqPSg|Vv$)g2F)vt!z+Pm)BhD{| za>Ev+DSQYwboEe1G8gM4V(YH+Ujg<$A9I!Srhqmdn#f5#-8@b1XY$mT@O6_PjHzjuqXv;Sv@6`5$jgA(A6osn67lKSZ$_FZ zM>~⋙U=K?EWlq8}Ff#6uXcNFD#XKb^Z(CB)EdFfHhJDds)3TEC?I*NjrGIpzo`u zq7k@%B<5%oPjM3bPSNGfMSplrTsLBTlL$j zsV}K!Lq0g(5x~bi1f_mfC8hTjz71PM=^G%h01Jt1By~lIa8}U9KFi@P8ldTB$9jE= zViFp$2`qx=hD#$0If^hDyqb^$Q-vSSy9XZ~CaR_^zu|(%0PfGH73G!<#)!`-s@z2d z!RB4~o+F^um=|;`)jRfmaoT6d)WoO?iFkYWu3@6uyL1F-rb;`ZgGiXdl(9bBle#^9 zHXvFE)gO|a-UOpb6~)^?)x@Nu1xw+d)#H{OE>)sOMQ~|arF48`ZmJp-`4qPem5bS7 zgP9)W0wwoF25xlRUAYFA$Irv^wq4C%6LY4m!aA#ENxgxa^{%etL=Z$YBaKaeWgzzTAxL5!^PypQXL%jwxryC zZ@P5=+lm=U2QhEFSNoH&AQEdA{r$mHDl)(@Y_ztUO@5hNyOW?&0 zyqyC!gD(|~E3u%8o4<+yieQIgMhO6dMIV0r`IFAp0H`WZ#o#F5v(0F!A)-c|wIQI2 z&6tw@HfN2Pgfs93*Fgw+{i_HTsOwuC4M1cwLCkP1=f3#3Hra5y=d0_his++>`4Rle z*whZRjADQx1U+AXrMvy)gG55CY1IfUrEm}=iNw7VL_Xrhl>^0m)cipZ_F^2g4XQ10 zg)*K)A+7h|Ar}y-N#nzUD**M-46*$P;bVlQBXBT`{Gb3D2fQ#CQ8Xp^o1I=ffux*T zabHjdP^)76+J=za2jPc2I8~Y2T{r;eg;;GfE)AHJ@lwRGV&^tf0Mn99M0V-4RGhTPT9+Wf?)S z*L|f0dB9Oh9pIGZ6qRFIPyXJfO=nY^m!hX?B>;l6F#b3ETY*7ULoVw41VYJrA}bvz zMV(|mUhrTnEde+fO|T?BHC#w@&jb*tNZ3vSE$^X4iXVd#AKY$jrPdZbhYE25af!W0 zuURWCL2iL21|rh*plAaPT5T8&YFPK?>MOckVwOBZ1bGfg{mq>P5%9TZ(fH597bwRe zhVMpj=MmW6?Q$RmKqR7 zd}HTNJ!Ogr{@V7Pe62>GNp@l@+&{0p2S*+B++}yp;5H5!Jdv#C1=L(auCpjyvleuF zDuhtT+up-8;V3=Sj*zx zq-r359V;`OT7VvxbB5@hqO8Fx4F}cn86N3*Dnek~aM^-Zp(Jvr`baK#u84}!Jd_TU znX^6OIQ4|p5Rw9cg0G3iGW^?lMc&px)^pDmN65Oe(3M~fuq)f(=!6Ga(?n4F_Jl^p zrj_6#5PC|93pot?lY36_5ZRId;H?pc)nTwVtg&JOMX}9%Q2(cirblrbsz_NVB_Jj! zB^=0^qQbwpx{UyOgdbngol#7SNRW)-Ezo3kB1dg|_`^JrOAnud+kXT(BGGv$L!}++ zgn}rA>PN9FloG=3{`881-rKMZYNbRQ1ewHu2AKLl_IVe8N#a!Bf*+8F>rL~VBUz-2 zQZ5@yM??)RB=#7ECQdmYRQR+Zu#SI≦Ifa}BwG^AuK>d(&qXIV0{8 zZEB?%JLG1m>IMqW5qJnlhqm0WsPG}~Hvz9?5o1kU;@1<3fh3^>LL+VPMc&v9^1!QT zELt@(NHbhstm)qq_R_auip}@#uynVK9{3Ts1R0u5=(gb@fEX#3xi^!>3KjuH;-k_d zeLc?8b(pS;Fxe<{pGD!_c#BedrN!>XrBo`VXMbmqAO~!{y>6}asK}vX-cd2&_|-_R z2~s)D@1zwR-bSN_D1#pxEZcMxsSo9d1S5bhMjBAWL-#?JWUKjMtn)5{g$ekekE9!; zq7ld^tKrni6w}oJbUpaQ%Iw8ggpiSoyECkWO&p1rR>PZ}HVqotYn zU>Eo@-fpDTfhi)(LoOLO#*mYnHG551RX`TS^N8oZgc`^h$gkbQL!^orxFiHO@E{7X z?XmpHk-3Uel&oTqrYz>+Eu(HTl!SL8LYd>6OR)nUj6|Duwjj-*F2aZIKbAmbAFNq1 zqK4(zakli5i>wKmYXgVO|OR&p-V7-Mi0!`^P<68Q%T8UkgWi0IZ)yeJ;Pe zNea0DCFdK5oSxWuK%0h&^HSnK}Ug zxd%I`*4amVPdlJDfvz%(pZ@sS#xp1m<98yi6Ov}knTszk98uv>HdtRP(Q^Tsg0C`1 z=Rmp8W+pGObLKpmjM zzXP9ELc!qoc^`NU(tJ$vDR=AvNVM$z3+uk0gAmKBeQVZVSbKc`_|;f&HU7B*|AA0_*0s&pEwhe(#$wEdJu??$b~R>Kj^AIbM8u^K*LeMmckf7wRY1WE??<=S@0G7S{&U;kzCm0#Urh{e$ zGoJu!&O#d6%PSs(s@I0WN*f1<(W&$tjMHY+T1kGvzPBDl71X)<&j6?0PBKQTTOR>) zCo(Q6R|sA5dP-H9rTWKJb;iN`AZp~~piJ9zw)7%jTqRZ}ldqnpCiO8oZH z`9%Mr!gZV1hjXb&>#qd8;E*K{LG?xP&?5d$-kCFy&ME<~s|{3`3;6xp#?w?H0cjxk zx%5nY&5J)Gc+}|#5h$r737-3}2lA;TLi93BP*~@J>yNKYyXrh9N^6$J~LvE9J?} zyqH5>->|bTy{76T$WRW74WNXyfw1aIl;S8hN=gIhU%T|G0T1T%xooFa3}!$f;2}t7d+Eio<+P zlS0e!N6OTz@nZ*3LP7t#RXd83jSR0b^-$}F^R|77aaQt?Z>XK6`Pgq?Y%nG=9;g{d z;vK1QO#f1BAi@jC<_~gKp*m~_8Qv|cSCiZbZN0i)23+qXch^nk8lEfl=0e5*915E> z0}u})A$Rvt8Gi95(lfM@d?Yz%Aa}4xmb^Ud=Q?0>>#EgN?5v)Igqy6r|pS%FYIdq0W%=A z8Tx`TIvG6@y~Gp}X%A9vB?>d$)15)=>t)o>RNOmrZ?l|i=zg$YqlG~j162k+toNxD zeP@t3vnXWkjkApE(S16ak4p2htm$~eH-gl`J%2jH{3U4P6)yZUFPB9hM|;qC9Or72 z9g1na#Z;K{PsMH*s7XfmbqNRtYNc3U$V^fvh<3PywGNmJ*X|8N-eV!^nLcOEfY?9w z&kjFfJCB)P;br?bZr{%&48jACO%E2J=(M4XnA))ui{=l29#aJ4^;sYu#&Ue!CMr_( z?#AJm6DWzsV$dK&R9lF;yARFJKX!b!@%GoaN{wfkjFrh(cCr074R(Zhz={pP(gUic z=MrGU@=FS@xIYH(H-f?O*`TAhbQ19WHGgPf?Mh$ye)*m$p-i5#X;%Q!3IjX5@ zfxTcNfvHnr#t$_CWo-kf+M4Gsp5N^u)U2Xl^e79Pxdg=O?x_F~o!`%v8vSOO(~{F;)cgEcvop&xZu=jH)uY;FjWZdl)&uN=B{ zb3aVvj~&pH6CwWE`6fv2{Q0wfFP@hgSSkJ;IG!eK-@A@;lErj88nSSwG{0@mO%D{3 z9VTCxZZEq{dLQq8ydp2RIeoi^J-VHwLU?uRGiUxhkO$oZhfTLnXj8M^vf5yj2%#pD zb|{-FgJ4?z?DjD$0Ddv#fC;CftnA$X^yVyJ%1&kU1v!1I=5llbr8XKSN9E}S$wY14 zP=>PO-bhPx4_y?&i+k5Aw90|rf4eWUX-;4=IH_BI_xe!OCSI5%OwsAA6X%H8?Xq#2 z$>IAiu&xeLm-3yHn(-PH+^<mBlHRUsBC@uhRf!Gq|7*gQSbyFB~ZschUmH~BW zzxzpPgNgnW&fk-|5f%%h14Q=vOAgG@#mszx7qEk>*iBT81>1Z{9+lof4nNz2i6ElH z4f(MHT)rjCIK;-ibCP;sVE01QPm+c5?|_ypk6~V$^X?1rix zN_x&%;Q_d@CAs-l#Dc>T1vp}+-JbJftqE?_L43n@=4Y^izr=Ni9lQJKhB9gnJ=^OG zNfN4-r!+(J^9cxgmwI$!Uu%rzL7?NvUt}2$;(q8$j>>ZqH9_%zsWA8uO(ZHF7}Nj5 zJP3p}BUBN{4b0>aAg05tHNX?Mf^4Wsn-)vPcx+% z3rpQ93;GzgWrxD2+`bz&?15sc<=q*?1uv=s4^5-B03-&Cfzx=$0#tVtjV3G!bfuA+y1l*Sn**I7IT@ec+eSW`9G4x|9V#Sg#B%AlI^N<1Yixk;DDx<102Ibu& zF_H99U${sLe6-NM{k=BYWFl=_^Nnj5-{|ERqfyO&_1R>m=;$ayXZoU59Rq5rb9qXmu2K?Ds0FAv}S49u?$>U?X}8+i?2@7C>!>)*H+c`OysGQ9hD zU8Hsl`_Fn9-rIi{*#XAS3$uTzZ2vEL6#w6Xvy-dbMpyPa>Tf?b>1fmZ(`|m4>Vrud zrJq0fe7{CnRC?|0zpfDx@m?-bacbkOBG#i6y~|}WB0_tuBmb%^e#*NjQ0BSf{lPu< zb5|TXv0(4U(zXBiZTwA#a{>3M)7-9)J{gx=$rr+vY#p5#dM&*o;$%3{EBwxq&i0=_ zUkATOvY36r+r^n(huL-b^O(#YoIloL_FT-Ki$C@P{r~KKnB5Pv`++qZN4?zyv+FRs z4l~Wc?0%Tt53~DWHca|s56p&(Z&zY=9sXnMP;&mmhb&h28-<|-S7xG>D<(Pl0!^1%y_&?e zNKcLWvzw>Er`V~luWyBknrN3>IDbCLsi6poc6z<}AAT}-uPe3Pz56*fwUCQ3wU{b! zfx0bG;`W`GK$-3wA@oEYfu6utO9y^AcU~WLF_n8zi<0X%r5_TJ#bqzir5oa<~m z31Qa@)Z1w|H5g=OVIke!MFZGwL&5dgIm5{#qzRzk%IGp%FVYbv%L=nZ!yhShk;LM@ zulnV8VxF>)C$x2oF@JCgs$p_MX|?tBB@ixYtCZnxI}G+t^}~l>60e0sTz>fBhmd|x zNI_lC$&sd}Vycv=EW-zpPBfL>zmh$1ob+aFhnV=W;)Mvd=G`kHAw1CRM z1v<W{aBtbQ?7?qcEck`}Xxk9W3J;({jN9aTAl{)t6 zhCorUvN}(dFHGNCXl!h3eS-6e+{=GZr}j~_;1689;4UC2DCq48g_Ab}X%TOPsbY9$ z4AZi|qy+*yObU(pPcA0Y45N2Hp{Gk_+*r~js5g?j=K4lA@KSiKrgKJDua<*cMG4Jb zLdE%19?cb8vQf`vmkA_aOP+xfBa|=!ZN!X>jFO+-=FcZZsK&-dB8weP&dij+XF8R3 zQO$}vA@MZMk*)*iY7Tp8ub?fN%Hv1S*Ite7bax7ZEY#=U_l5>2zWdqn>6hGq7_^@k zLt*B#-c26eN>p}Q=$TLRpFs17OySZTH;5Ilbxi8)?4-WpY7TUM+Q4y=dm2myXZ-8j z2&PkV?}0J1&!v;Eb!-=H!-0k>Pl?j9rjdv{bQZz25NrgYNV~dV_qz+gFJ6BeFi?}PbedJToRB>y!Ll3I9-`?W5h zoGmwYXI^Ia3rPH)s{=`6Pnb2vmNV%pIR~VhJJ5RW^4Y4TCjTh6f_w)ooOT(Sz20p_XMab(EImIsFvQM+gn9)Z=WvY z@0~t5edY`ZK4BNTW_}KfYw#>@!7D8ElEe}pzWUBIsJ!!pllp3FYop4CAxE+smY|vg zGtuM@4%vWw@9WpEU!^4B_(EXdK)R-y+NIZ+_VMS>?=AaqaS&M7N+-?t+{$_U(&=Vz zAPDE^a`Vg?+Q~2upY`_i^gM?)Sx6OTjLJanWxp26za>0AhZ}F#;_s&%Np}&A_xn81 z-~H#$%l>aULYtuo<_*Hz;y!)!Mfc@gAMWby?UmW4xpT#k+GwLp{XnHr#fUHJ!UblK zBi76k`i{X|QT)jz`5x-X(Pk|;!ql*v+Yi&RQHR@-n|1xD{JmRg8J;S-p|K6CS9?x0 zwHAZzrJn8#azk_rgK-hMT@a_RG6{ZRaUw)4IQ=p@5w0NGNpE>1Q`yn5oR6;!{4I-W z4oSQ$j2PQl`iA=?+rg9#YFVH3qu1XqSqX12ca>bI)Dl)`n(qp!1r-$)@|G-?$^~ci z;cnxgsdvx?&A9j5TIP5is0|MD!L_H)ZO87b!?%56pUB;&-_>Jfn?G*i-YFJw7Aw1S z1a6MR+W7NxGpL~-vzld&mq70e>iC>JSN_JTw8q@|4c@~vcRIC3Xd|nz`=!lCtY1rV zK@eKW)nKt)*}5O>j~nT4X)>?c$W8y&L1n#=kc^P2N=f#2O<-jlf$>t+)YN>^3yllN zTrVU6^2er(nsC>=0fV#MU`|!PhTKeF{`KMDI_K~t2kT56caq~j0;b5RJZxVV)biV) zh%4RaHuDp_h9V;DbdOYK?C3WdJ>F>rm0F%w0|YVVxS^Cj0LmMc{Z)D^Gsj?-S3+yg zwm$fmUw$Db!QmDfZ3q^{pb_hsjkJCHp6Y#^K1jJ7)HqS}5o;?yV{utu(~2vqLxfkl zuVLz@$mYc96#lKpz8>MWuEOcC8sf)@g&HYHIY9;9trz>Zi^~h3jz*>g&!1VCr3s`zN`ue2$)*=SJrdZAIr^Z zANb=LM8$|r=L=!kA8o*pk5$-TB8^S?9&Wk)7&@+7eF92T{NcSWqv_bDM&(Y>zrII8 z-;sIrv|(eKJeK}>WCe91KCV#Bxd8DZ5>d+SpZSPbie%~_9A+%vi!GTk{sNk9 zZ6r&lz?IP_aY$^I@DVFG8~I*%g5BWIHMvjylZISiMr6%%(1*&XaGrU^tes~@6>FeH zdae$#FxEpn1$#S;61;l?d{-LhV&qeItwHDHh+4$!gCSLw5NILk_CAw>Vm$Ov zZp3CnfQQIWY$zl={MP3YZTctv-KAt_}R8j?JwnVPxzu38w^_1x>@2 z5Gm)MxI1&36uL&IrhRgCs2SW@Sfi!;w!b;PbLR@!RH7D*fpx?O3-3G1?vkYmV>E(3 zJw1I$VZlFAcl%L%D9U{(wqPS=T9`1U6{?u(#@Rax= z4s;GQ+TYJtGWNzXL0$@gsLX&j_>5-EAtsQ0+66)`Pv?mZmKCX~q=`0x`9!fGPk0ET zxGH1W*e}H({XYYt*9W2&VF3*@*P5kLypMQCL}RC}gyQ02TF))w;ze}iJ-BX$dYBAh zht0hU&b!gW7KZV3XIxyIH%N;-StJnmFW_9rBR=1`lMNQx1q#NX_Sfja1$$3s3&Uy^ zJ~%VFUKEpCNZZi#f}I|=hE(Eyh6n!LN{IV{u{Bx0yRYG?jJ~^eFQaD+_9<)6y4e@d z>ddag+dVS74zug#v{4rcem?0)#K-Va@S*;&AYw;ua_ zM~GZ5a+N1&`CS(Nx0&C=QrQ<>;#TmwHkn4b63&DX&iiOc29On%6P*|Zb{x~D+J%*w zZ|i#$JHSEH1|OAHkgb@Ro1X>n#FKqSDsdJ%;J@E61q4=E<3IEJrb{?syDin1SbbO^ zvEobP1s{ETaFM*7i_$l0j~}P5DD~T`c75lyr^?nT4c9^iGrkw&lY28~MUDM>H8-`i z&;x({?DSuicYeR{*r9_TEE^p?y{fA0;++q-o%U#-^t8NHmp|$-uin8zB;sU}Jl;>SCktHt}=q;y}CRDy!5I)pzX>9s9J$cJ`KQv zDHg>QDCZD{mYAT4M1iw@02!k&jkv;enS<(kXWkr3%Qzrlv>&L;BKeEdiAm#uq@7+s zi1T$j5V%r6Bq10QvrYl+-P?`uc|JB>j6~KeFd=o2L-Ppms3757A~9Y(JUy%zjFiC z7&V&0xcCHiJyLmpYlEoL%#bCzQW(bJtcC#V5 zDM8)GnNNn;e#3+nqsRKXiU>6*HzFoEPDXKH&Irr2xY+Jth&9%9Xyp}z+R5X=EE?a zM8aGd^&bad9nnDRD?%CA6-&eL2lY?5Xx*!q1b3o4=QqfDz;cOfLFqCIoPEpP>)tn zU8KI7fjnFdu)NigM|C0+mnKvQ5J$X%C1C2nc^=Py7%w-KyM%V@tl))=Yc3qjh?Oo3 zk%h>%E7z?x1mLR2IP~L#Ajx}ExNO~<5OjZ%wz(1J+Ek%GdFni$WKvHuV@K`~dK_6w zqZ{-&80pvr5rxFXgWi!^ovjPXCeURLI;2i8p=41!3#No?|-uT zx%FIGbWKY={dr+3#>npK$VG9o^XG*s_o)O()9LW}P4@NsAij-aR5?auU8D)}=sQ)k zjzHoZC!WZzopcGO;fHOV0?L>j_W1`s!_T>{mf8{&ydme$co#- z?AJ$0SL$WqhxIP4P^W&AjVTb86YCz(lxcAe1N~uf6r(!wAbkOScN=WD4c@C)O*|>F zFW+49*m@f33lj9t1A)e+spT}20%o(``XpnCw89n3+-PAj2P806_>zF(c&TQOXR#)H zN#hXx5ia0rzbWHQg*AdAT#*emi{&qdIW=y;`e?RP@kTt7 zx5B9sb~U$IjA0zEN*&bUxV`RL#4TyS&Zx^t(2<`m5fXGy76p~Da4y_&VQ6~h?Q&ognRf8X4IMljV#FyBBfvZ$h%d6<~T0oL5GV3*%nXMQwWrIjCg~)+T$3)9`^M zn+({^Qsj$BO`irGSxWIU-jk>vA~^3P^tm5Fh@!Cold=*YSgt}D>J_(jI}QDS`}a>3 z4w|5ezGVdCPvY>~7TYpRFl#!#CS;UMHu}8RDELLLAH2TmO^iY><}h9%^$XgG)_n_% zwKqa8qhPEE&G=$qza;Eq-htLXa2Ci>_PIhp{r&_-?5<0s@l?S6CryKYG(fj;8d^=G zAl>eYdY>3LATU^w>n%RYr@2{tRw>`N6fjs0EaB>#rN2Q3{*Hicv)|~g3mypymg`hP zF9V&f^>#DXzKNFPaSVX@6?~p9RP66l^n%fL!c@XatB`1VT&IFKQB&Vhy`_MmWWy3p zL=*#hhd%_?Yl~?J(4{(k$69r5%x?3O^mdb1ZKn@J$81n=E+!7Nqhz%QxVD9H0rcHZ zRQ$%+WQB@|0QfNAr_U7&sf6jsSJL}wn3s2C%;HhvW6&TCy{D5X!gdjhp)_@`cAzV! zqYXlnAlc57@nE$sA_@U5W3L|;h=%fD$vn!$NdW+cukLwE^!7L1S$fWQKmGQx4Tq;( zaDG`@&p#Nu`RZ9gAuiDN#1R_bBWYd%z#rz)x*1^`NyOV`+hB#tQ0X`1+YUDs|LY>g zMQA7^OhIkDo@Z^=C63cvV7HClt4Rk1GZsS9Godh$(FOSeaS;K=TN8f4^((od1{$)3 zO9OQM8vES9gt8o5hKgXKYys2;;(uk@EZ)~e**3YM2meZXC%sE`QN}==|0o5qZUzUY z7z^&zxnJ8g87>;(ZVlYR1#!LsQ6YD61|~@WVGb zpK`zmgiv1Nz4vNt%JtqWuw|9ypzWi*OwvZrng%5$dmdw4hs$ROQT^I6Fxv=w!H>oe z!8_&qA=KC;#yu}qX?8GQ$#0+gvxUlXrZA-$4&IUXX;|FHv5d!jsAGBDwn+@aD5nZA z4XQBK?gX);5<(E+6qn=Pj3GA^PNUi33=>)&X5DaizW>eXE#sklHn*>3I>B-8jmBZ` zm|$GtjsteCwwU}?LBT?(=j(2`V0<R|Yxe9((&@Aaj^y6@kVNkrTEK%bq`N7On61EHzmGnoN@KD9v$6n^VUKK#Gc81}NUhkdZne?7xuZTqi z0vn%{kruc9GlqkBn3EnWy<>3KHI-P#Bk^*SKbP&A406Kw@1JbH94_mo!OvI{R^1+i zzzC!$r9~9?0HgiMaTgHU%g$1%*%(|G)i zwN;NWK1D0Lw9HNEcweJ936&7U& z14=AOTg4fRlvv$#lbw|glekG!&rbg4-eAA!p_5blD0i!u(>>4*br3z8oHmDv0X)>n z6QR{bDnb*I#lZJ$wl5Q*6b(*RdI|GO^l-@Q?$I&*?F&gG@MW?wmlx@I?M24xe(%%f zxY3MKoKiiE^dcbTk^pb>>HNbMhOyiv#9G6c38Co3W+}UJF~8$JiPwQ6U<#h4LZHYC z8pnkZh9zP~_kYFk)~ltr8DExuL?nlu1-&}bzBU*-}u{qIXk-;ft>HNbnlRXN$O9v6XXhc1+myNa8Dd0(L zUBkFfc@-u7m9~1qeg=?qjH92>hOec^}wfWNq7bZ-1?dn1d6Op4HC$fGcVt zaUTlP)G1S* z4$KCV&Hl0K?%KLmSfvKUg_cEmQC%b5kG|D3a5WCcXg2pAwTJ`FjE!2 zfd>* zf`1~ci!dh`ZP!Fk#!+Coq)Di+l!Fx)u=VHCUFrc@D_b-g=u|Dg)cAtz$R5O(ApW-l z99N-XbV5wx;ltI0t6=)>B>{O=BaGVZet7LS#jd6D7cZ$u$5!W}v7;j)t zVKwp~7MjsTc%*`5V7TTY`CmR?c3@PG`NDO0AqcARf8qF zT}<=D>e>7L9}=$hu;f(*6nv#RFJrjx;eh(&o*bm`v~>cQq&=^jacp)FoSO+Klok0& z!eu(zUox)kiW1_rM2l-GA26+AV8Y-a_LKT<#;d(@;;`Z1IRC}S5fFo2^uG&m`8P0{ z|A*f~@RIsKxmZ;@2C9_;pb4eAbEeU^r0(A*4zGT~;3Q(HoHbc*&@r5zymtbh88lk`xETIHB`k1f5&tF-oDhmvi@-f#3O6+vQqc*JX*K z@`~R*Gb*Ja9-W3VQRf8=YPUb}auu*Z)8NYm}Rjy*k%-Raj}wA8R1Hy%zr+XhtFLPVNK z5PF{b?;pAX`Q*#PHTR`K0ufL$<`!qF#@fQeNj<;%lPFw^RqBIv%5uG^8+od1fAeY= zm%suvL9J{+1TKaU24T_keUvN1+kK976Au~>j=ZtuOOctemY1XYxjffo9^M7KQ2nw| z%W$HtN4%;XW{W!ko>dxzNMjeUv?`?hL1Ufzy6IqU>zj=f^vFb^Xp=sJ3}D19kpv^ zKx=#zeORX{?(~}i%Io#qGy{$EQRFwl%i{-!XI{FO##LOpU;B+qDvE=}%C@`-*D7f? zQivJTaB>>Q$PQtf4ji|#Q3h7*oeP}qvS zlv155R=ITzU^jzqZCpY2TUj;YGyv8K*;V5hhHzKGtmM@>I%zk0Q(5 zqC@Q!@%#%AoO0c}WN3_RoNv0yWi%F27v(N){&?H~wHBUbfGpN39$Gw4TIZWmA;Zg!C!7mslPF1FlU<);T8FC}lzU zuc89_8J6%Nt_EQfJ;#-9FD{~~v>ocK_fro4#u?ltuF+4s6oI-$+Xo+hd;@8fvW^>o z;`3}RH*p~fezV#I*r2jO6MF3?>rcMXTyc|L1s3fdA`O{ickqz9nwv)`n(Lh+iu4L~a+_?2*3j%I<`;Q6{s8|jqR-Yf_{LU!)Vz%2Gn~`@fmfi-)4Qj9Z2I*vocNSEi^btgX8wg0v#J1Ha6KiqXo!0~&@{)ZF0M3O z`P@eC7igyG`CBi!^~7oyR2vLCWI60koKXGidzBI@tSnDxIK24+*`s0mi?k|(>#~?P zA3z1vwewi%$Dy@u^{Vx0s_>(S4yLI=swXSu zYU~avRk^vqc4ML=McJb+4t9MNZb?ZQ!Pd%E7D)v&y#*`9WyGO~E~ZnZRAr#KN9G$#H#J1$Gz)%L}cTiMSe&kcf^Ac3;>YsD_7 z=MTT3peBE^W0NV-K$Je!&BqFD#bsC??`=jd9z+4q12^g&_bEmc8LN~j$sOu+xDpxL zfJ>%CK4C<=VpwvfL15K=fuhC3MQ{z3aJ?5>oSH;_J3X%os3KyM(o{W@e!o#7_}~ms zWwQgOx{AWsJsi=z>=;S3&8q86!?PwFFR0A`G%b@?( z=2l3~mjjNiY#@VP{mEX#X`OD*cuVpXplW0_Htah8Rkc|&7$pga9?QbWA_-(qPfcj% zteScA4|90LpwQzr3fV9Wm)NSMx(8}{(LOl>v-S$X0U{q|Kk8&6tz~@bMYwZG9UfFj z{eKRaCaylBujBp0T&Ww?txw_QYz7+vR&%wnXRT@YI91USEi=)eFJ9{O@cU21&HF+M zO0Jz=nR7RL)mY&tHJR2;I@cT!&=D^C2nQu=TNjRZ1w!_Ne3DzXw12p2Y{w;0My8Ya z!?}D{sydb%2kC3%5|?L%)04vxVc0u`ebq_8pb9!x3E%nY<1b3-gn@domA0@bYSQlQ zYgWa!JVrD^po1xVhWg3GQhrqQw9)Zt*#^$Y*VH-~l0Jg|WZNxMAphu8nU%e8hK_-F zckw+ur4EFW%9qj{xg1&Crmgk1lT(ufzABfx;6_aE{0NVQxHSk^SB!$?OxxjAcMVRc zQo1)mKtVfD!H=e_&{*g+1Y!jpjRce-=3X^~xb?Qi?@E54o#ZwQ=IdJlb&x{+9uW@a zxp3@TAe3*9f@WPYcy+)0kckuMzK2(l6YR?6~M7I^l8#}K$2Ad z?Y8AUn+OI(>wA=G5J4u8Z}6D?fyr3Fm4x)Ei{u7uyR z?b`@<(P*rfVtwd|o>*7tBUrXkPKIZ$)4hD=R=+xI{?F8r{IqtUIYJ&Pj?{8`k)bJ| z58lrRkBkt$p2b7f@o4lONBQva^OlRdR|Zot!e`D2g+bI{+fn2%N1fe*OO&fn1fLat z)+h(^;J`L`SyL!;SCXcOaVLl7Rv<$_Qm*Qjg;=5M`SlB)Wey~|zv|8N-5I*eF13R> zQ{E*vnoW}9)y?n>osfvIB8;{%7l}T0!B$vn{U|O2TT7Uru#2)*#lciYFnEB@>D3YiEbQqgioOuO}>t1eV3*t zB$BuY>D621k9Su1#e-dt4Izs(IN%OsBT9Zqa@LYAI+;}BEdXnRxu&exC$ zbMJXMs$%cGMe=a%CwxYmEaDR_svuRb-}a{NBdkjWC1Yr#e3VVlCD#y$Gx62R^r^_( z?k$wWE9;t6Wa5BzQBb(KWW^;ECvT5TQC5bw|BPmF!lH>*qKu(B0hQ9gpiv$mA{>!H z6-04~R0b;xYegJP>o#62nSV>MYvu-HCGO@W;!xJtbPEUIx|}GR9Q1%U_~0OilS>q>_eaWk7VI*1Hv5#~shb(kut z$(O%90>PuBk_Nw%JpVT6>yLLz7G&$ZcS%1Wqovp}Ru&O&n( z8~K6|MDPhceS%Mrj+i2MQo_uSR1^>sQXXfb2$`gsgwUZwxAnC~K9XW%SZg9Z&>I9A zj+_%o{cWzy4XB$~9xq-%vv^3j!USp}%a9;q6q7&VF=K2?6~NDFI0ext6FEvTXh2$7 z8BYcY`GymkIsqn~81Q8pj~<^ApdMilPqh*qfLW()8IdvjY5K9RwSnsIyoY% zUOzm2@HC-ck#whIg0qbS}J{bpCjP&a2Fab>h zRN)6`uDXbz3J1_B2&%9ed`(~+eIo(v0^lvu>l`3AFyqHdK!nY9#(hL1cu?>|nnC;n z)UkL>i%Fmr22XRyd=V?WCd;;!HW;ERg&G)uP^U(7%e72OL!u_Cof;-|xFLsfam|Ho zsUtQZpNzvzMG)(iY8^x(rP4IO@_+JDP$>=;$k3g@ZNTCYQQ;Q$4h2r`Gnov1s(3GG17%`(#R<73qSZoK!NU^6iL7ZK%Jq%?79z; zJ*<$a2beH$`Kyj@O#!2UnLzLanIjJ#J_9de#$&~=36}rDhWuauVR@Xi3ktVB#Yo!g zfpaN#J^pR*JtF3bxCFK{yd}6B6c?0<)QZSPuhclsK-pN2skp|tqQab;sUf3&k~Vka zOBg4BWu`#%FrW?_8JHnb3iKq@qlG-De5SQ!c?I75cj%IJ*gUYr%Gh8?vwj6+Ws@<= zL+<(*XM$A`D=*nlnrd~Cp#Z`vEKPro8@RFF`@Pa@p`iIdhpc2L<04-=J$i=f%6AC2 z8Xb^u3+-iuhC3`D3YPL}%zcO@D;KY0y!ooQ+2W2oQskimFc7#9i{BWFnNn$qO$6Y$ z>4q!OiXl${!V%$na9GGaPCg)));IyyNV3IF##^=4g6>La#UlAzHBzXIVV;tHpBQ|q zEQNE1eIAULU)>;x9R1euJ2sAThM-8|dQbqI@*I+OWm51lQw<;VaA(;RC+s2A4m(=^26rXUHkstJc-pezBb$IBFE5_X+>=VQLd zYnU_ycGMC>ju5O7;=5=g@NErrZ!s?ZPzG`Ob;0NDVK(HXD58jNYmBanq{boQRUBTx zc%xa)5`9h+Xe{x*jK8zu6aKBO$19XjU7NT^h{`BJGlJ&{7Pr&@ySBMq7G;7*`?j~7 z-@w=iW__F`9eJpHv?F&br@AHvK?!Ay!q*KN$g4gUV?>WE;p;^~qePZPR@lAs-=?98?}gBKNysWp^& z(^#N!(1}PQ2S6oS3_3~AWZmA&ze0E7;45p!m70m@9Czg+1l{Y07b{CN2#mI#Jn4%` zM(Lo9JH>NCfQ5X+h4KSIlbur&+$xDCt4laqAP|l3^_&`tNUj)Td}R6okxX8ZnnZFk zbVu{CosDtHdQq~R;Uwn{nnnc23Fys%pkl<==HZ2mPk1?vV1N^&dfkt0*ULc^6+s-y z_>6sUT|?*|*`$z)l$i*Dj%Cd4SN7;gnHyMsDpZ|uX1DRbEf|CMe&|#LdN1s{QR2Zd zi7f1`zgD;s}JsOizh`4w>61J1GSM{2`*M4S?#^HCk+Iu z0mj;H2Ao7UjKs9r;yF;XR5PdbW{8cBn}K~iBk^W!IlV-eL$XSJ;{;`O4goK@3E3Wr zaCHADj)p)dt&d;?ogpvA%B1`S7f6=7acHpA2{8?oD+UlC-L83qU9aY5gPe>1y|Uoo zAE6cM!kq*x_J_jFCXP%ZORzcR=7YGC(tB%m(GuOq-wH6kP>DrvxQf+EB56NAy|R#dDDhhA{&Oqfmcx99zgk98U;*0 zowyHprZp}o8w!scdk$tq`BR0`G|W+hs3VON)W1Sl0#sjqByL$lD#NuEE+B51Al3C& zI6hXTlq(!*FEM!tjH3oaO4`(!Jf*9PwN1g)CKg}=($RQd#qJshcamUyWN$vm;?#wRQgpjjhb-eiUQq+zdM%(fp)MTSRAC{Z&l$J<@B~O8 zy0yw+Y}-Dxwh+^eXJZxws@i z6?A|m#9culQQ`rgd_GZ;03qm5^8ETFlM&`}36~9lNt+rrh3b^`Cf;^Ns@6d#hcn>o zoN&B_+UcmcG(EI^`VtVRb^@QMHV6DHYvllAwGW+w7e8c75H&fHD%nvzzjjcaOToFD z3@pkMaP5+54Tj}eyVSttuwOg!pE5vYSKuQgOpO?;X4cS=M_z#d61X_3=8^tlJiCN+ zl_=2xYyM>zs?Wo|%tj=ibQo8#py^l%D4^;>b(+Kx12DRs2qeZS1j=O$nBGy5`OdOG z%I`>M!HlO={t?8bYB}*p=o-*$y+h*MWl@E@>sZdMhs0## z5FuN|6O1J)^rp*0lZ-_bUSiPpCXmyHtV^%$Jr667I(#yUA-voQkyhe>C#iN?U@)IE z-o^O#L-b;DxJk0xfevTmjA^o(=#?ky=3MGu|NG)?5Fwtci z?MF?Jl;vmjlK|FIRL*qD7)+lXy^qNfH&~@r*u$}PL0_+ms^e$_MfHscve}5<`~;|1 zhl*;TdZEby{fE(+o)4_4AqU;7*+`XAVa7CunVlDAYjs{#xG4<6YH#Nq2#~2GHHwuu z>&gBSxB%EE7{YysRP@~8^Lj4 zO{!A1uG4;!r5^k;PUW=mvCJmu9Zen-V!qvb%R&CDu{o%vkiX|;%0x0DkZ%y36k;@; znVzKp5sayonBA+-Ek$M-LWA|Nk&?E)_dF_y*Sr&r6r@>2LT^3m8DE*AhiU@jFRb3> z@7cQ;vH$!Z(te4X3!<6xbvu9&KovBlqgMtxGr#FE+0nBLzeR5fLb-OI11-80v8OxU z>n|KWsO@Td_NM<(kNWL+(@&Y?FpV2%n8IMuH-S;?oBxs<;=dss>J?>1s5)1W^fPYR z%hBh(&V$Fn{{?^0MuW5=nn%?6!%$~QN=;+f4Xsgi4UP;@F11DSYfXiiEzu7{kqC>Z z26pI3A-e{tU;8lQ#=YvUD+ioTPJG5`lw%xYuewC!)^IKcwbhJ!e3*bwq~c`Ub;e~} z)kb2(c;&wb82m>dg}Et$)F{_FHIhX@8^sVPDz1T(Q`8Pr=3HOzEv$DtW!ZZa`^G$x zQa5TdqoN2QlDvLsbwQ@YT4Cd3WZA?{p(ej=prY`M*kd`}*W(otBFfbao8pxbi;4zS z5$NEMQBAyz;tFxFpmf_#&4GTBitc3Z4DbXALZ*(xJ`X>jQL;dWmBbAhKHp0t8tI@0 z^%=N={%N_EO~4Q;PTR#mFd!r4)5!tyy9_$jghJB1b<t)23 z%8X>hbq~EMSwZgxbxR1?E*h=`mJMs|7%=i2jlRVHhpCmo@@$Gh-Zd4K2QDwBE$*-p1HfTGz;-qK}T3B2nmvlCDz}iW?CH zcdO)HS6{dZpNuc1W&&D|91{i>;}u0$jgj>hRixDLmS^Tu48%nLsR1a_4~SMGKvle5 zzALn&{TW}Ll7KJQeSI?z{Gm-luDE1AA49mJfc|=bsf|PM-gV#cGv1P*1y2T#6g4sl zhO+RGE4&?FOaAEF$|?9OkHZZ?7myZU0CQd)Bsl=OLw&d=ydB_S%PJ(iTHT)Wwoe}G zBPiGKx8sa2N;v|@n}xm08ZGfcN7ovOBL>Qj2~^WiD8ELm&&(6_Ywc*dxZ? zD4eDshZ+rZtuxg9jL>Q0V^gUMw9}VI`GbM&1S!MdX5lCZNv7Z=Z$MuMNqSm$zG7TS zx(Ti%NC4Yr!h-ttB7(*zul_W*lsHL%WkxsVy{dTuBFp&QrTM>71W9;2L>z-4RQk5j z{{rfi$c;*LCYY8!U5>b!Xq6Ryz*oc;O1ga{xT*<}8!??U%7w~ANh6a-i9l{vhAR1G z>YVR%M(fJ|I3SQh5n<@mCiic6_Fx<6e+?kH^Hor}!GV+pEDa}g67OZDuZ_xA36rKa zKqQ|Ybvo~1z|r77g9y3@6b6kkS8C8hB3#fRM-?$K&W}P)3g0z9e$99Sk9E;J zjcYu?>#l}hZ$RK=k3-XmJQkD)$tQ1##p;f(ft8m)_^PkJgK(xVl~M0d7=*V41wfmEMstl4JY5}^!hSLF{mJyciNcs|jZ48^ZV~t28`oD-8LOeb( zV5e03qu#qqM!`*EIGI$`sq!32!$3&nY)`>j*Fm;u^cQtPS?gIgO1ZKm-}_dc8}4PY*KNLA^9gpgu$W zO0@qm3L+2^MO(nuj^bl#N@(?R>Zt3C=&`2Mx`(4K21p;V*#vvNccTkEb~NUTVQa*J zw>9JEp&}big3uSP0w;&p4LR8U003*ZNVYkMAc%d*R4jwsGVv_Q4F(u9Y|GVQmnQ#F zXuVtlNq!@>ykUJJZrB~ReIP>l4S-K6beq#xLuH|$^vR4bXR*@%4$2PMdt&M&lcj?m?@??B&Hy%H`_Tivbk(s4{s^#L4XN`MwHwbEYug^H_ zr7HZ>mqrUV-#;U4;`Ma;+1ak?P}M`fWNVd! zb(md;nRb14Kg{lj+5Ip(nfjl5V0IGm+l822hyUn0Xct{v>2DI@w`bgNZ#<#UBZs_Et}Djm!p2wHGr!%d=2_6F;$uL{ZJ;Ba2R}E@tN_9jH1zQ~ z*ii`>wMSF&#&R*FjxJDGJ5>O{p#7e%`1ts9H0|@~(WBK<$Z(H>YKbrL#e5bdt7Qfl{+v4bbn}28{6>$-9=8O;67g-Kd|lg*Vx- zr#61g>LZxwLl=S3fM@VuR`@m0BqGe>lYh#=N4KHS?+qf%(h(X&N{qoaAjbyozy{~f zzl9Nsw?S}kLrbIe@WjXSR;~cc$(P6{G+318KY@Nd!7V{I-ZN;Qs(&+vUQRU{)^p3g zqm0MRZ6; zm%6+?#u7C;{5h(rv61+_vwT@u2{7(X~FR~2ekylgY@kjT8ih+9#R zF)SjKY%B0s9oZ&KMq41+>h4g@LC0xx&(O@}+0l*BmX`s1lz`ZDP|F#$bf{R_7dikR zQ>ot6DlQ>W3?1G}a;+d2>M#0;#!K3vxn|YNW;78jLhX2C0+mFtoqm&x0E#sIeT6g5 z=%Pf-Mm&P!`HADwncsKp#R7aESSG1vvS@sv%29l!O8B;Hvo~Ng-l!P96*!dP-j)BV zr&b75zLeHnQEI-yez_2t2P^Q339n7JV97d(X;=Lk{VnFh4S`j1DB@K&H-~;#J8fZO z;|uWa#c0^XqrnD(Xg^i+I3O=CU!4;h+x;sO78v8M#XLqK%_L!+2*9U_ax!?@G~MxV z9{B89SvLB0X1CBD0M^0o!}v3oW;w-I(V*2t%|x95iKY?^NjwUPZh1GnLHQ)3bojCr z<1YW6i7~HeW)*Ja+j{I!NdH#=TX0Pm7}u zH_{OWq3bXavD}$U$(out``_C8&Zw%+wcBNkF~%5!iilDSc2E!$6ah66B%&xFRX|jv zC@p{j(oamj&;%Pc5DW-NZ_))k9*Tt$dXW;@0s;zd*)*lx`E2wA@|_>|jyuL3_ZtI# zL=C&F^{)4Q>U`!LQh5vu<5No6aP7kay%BR7_Pvl zmSF@0S}_3oZkVeM$d!f6BXLImlV+=0#J&TvTI>k+U_`0oeffS2!KYh9PV|_E^Z`70 z0mBuOigW=4m~Wa$CYT!mGc$Vp7w#NhDPz+Typ@L$Xm;ppG34}(p`8{U_DIKo@jU;6 zs|7jzn^m+r0?wytjqtc^`OWaz_5`#9t@6T61nS4FiJHbC4iL`FtFY9WC! zQh3ejn^s}#mfW26El>h=)YKgIna}*R z*_A85Ic+s|M|Y$3#^aLVacckLilxPAXSS9O!Tc727Qc-;+Q$ig#3rUa8s$5mohpr@ z>SDB7Pt^$mnc_zb`>LNa%K4BlUNW8&7WRmSZQuMk?OUM7Ns^Qmmqv#UK5$$Hu>N;w z0XjbBn$1JaHhSK9*gIBSkFL~1al?qeo9vMei_@olNWwT7n4L#0W)Ko^!gx>{0tc&p zx2bEmyNjn!a4*_^-_sDi^6N3uzV)fajM`Dy=~V#7G39`UN+ECg}{3? zMgA=~fc~N+`VJCDB=l15)|i39ur$mly$66bvV*5m7XUB$N{7LGz5p>N#Fk@bHq*TM zU25+k$UM42;?*DIrM!b^_02NM=tsR!VIGcr(d4gTJJ*6Q6uSz+;sC>j1Et zix3U%_%jPlwBoYca`Sr-D{do*)$EWQQm#hhTgipMjL%W-KR*qA}gFt>r8BYCB4 zR99=(hl0?XI)oqGifN({O%^$yLAd=zgcXj{&qz@9z0i2@rj3<-w%mO&CzjJnsa8q<2d3j&7Hnhik0UIgcu zL7vOQ!zPVZbYqOc3PNUW7`>AdBk$VT7ztc*TLBZ5^nB*J76V4g@g@rAVu&3-m+zas zQ2twLy#-C@1uGVZOm8>bT}kM!T?rUkWAZB4yh%A5sEOe4&?;f9Kt9M&m74Dj~-EJoa`-{Q%&<4p;Lomgh z2rVDLAUWXO%b#7B2msE3l~hVBRl0eu=ZGl;jrW}LPRy3`fWOs}> zyy5oTMLOErOOa@(A3nSodhrz4QgQ-EbT2N>5AEmz93TGncB?-54^Y$ve#WZK6>Vb! zLPdcBLx`sFr{rzwdV2gYDFIl4w2>B4wzUwS`d~fBc>2^gV=c~q_2Qe~kQ>m>f!5_J zg|DW*?BT4mvCiqo!*guWbjz|^iY+wtJS8=CT#w+2WG|iclnOOK6=4rXA3KDUmX;<%xllL9!D{Q) ztu#{J2I_MkjQMOjs4!5yYR5EswB7sLau;QQ3@sCl78YjUyc-)E54lCAq)3t0O(rPu z!q7zc?Q9;_5uyP}e6VZx7SLGvK?LwSNJ(U`7Zckb`U3PR-k_9lKj8sFovyxV z(>4AJlm|Pjw}<^-!+^TKCPCWK@`=$y-M6ROqKC=l5`BU9z>ZN2eXV8mstR}Lww(Zs zsREre-CXChaYxc(-Z~Bp3`kv}#zG6x{IDK)PkaA|Q2=(G1@s^BfqZ@a*jD^-U`<>$4FcAMH3@9^e~%F=kne z_AT{CXqJvI1kEu$1I)*5u%{lxH3=gj9(B&N$ve;4Jbi(8h^CIze06C?GX!Kbq%C{( zI^TqZB(+L_oFe_iBpo9oVaz_)eHYXRDoZc4EO=|db?O5_$zpVid_#HwB%Crq% zAZfyKu~&-FJE}6Xu(b38h=%t~Ddeb$r?PwtOUtl#qbxLX_J#IASx(&N zvvTloZpR~8=+p}`B_H%E`D$r;9F3V5cT6)eN@@&b1P@OI07f?yTn4{yo;Hhkm0Dj! z4eMJAxm7>F@Y1?pV_e&}7SJ!nMgw6BS)*r;y~KEV0`r9*a$}{tdwN);lXV`jeMbeSQx6(#x~$2d$Mq_$r-_UvVTRthV{|rr!A( z+P`MkW^2c%73!-eAFEM$b9PVsAKJ6h)_>*oSk#wy{x8xgtL-P=Za2C7t?6Tl&mVW% zA1r(Fp)5!K3QJbrEVyq{@2cAJk&AQ*&0R_o?4;IZf1)6$Fs@#`x>~J9M@2=Y4f{9# zZJE#2qaT3EXg}j28t5}$JaZ-ZAFu4}10JBXYx0vm9zoA9U*s+)77l(E&;+Kr8R$z^ zfZvYegr1}%P+NCZe9=yfI&aUz0}wy}PzJZ6=`&q z4JI6Mw8rdqOk4DoKR6}s)k{911O|f5#tDG^$lk61^3IP9g#l5+y2!F_?9_uF_V2n* zX9fo?wJ`^jlZ0`rVv)6q`g7$pqfzwB(CLfVh^Bt}_sw~LOnBVo2manI#-jjLYGx-S z1|}qO41=nGxAL?3Vq;&4Dlv$oZOMzB0qQl{mUh|GjOsCCtnH7-j~_3}GT#Cuso2O4Qf@7A&}Y zeXd#i)4;%(Q^qywt>AxSVLbIvz}yAPmjG;1KGJfRXr%I>Vj4Vnb=={fgfGKExUz9R zldeZt(YCfWKTNCX@3oo3TolL!J-r=An_-?*01R|k%dBj#Q{opJ1pS`_ZECkO{LJ*kHAcIi&JD_~z+SSW!paFyH1t|RX__MdqQHUljv_2_-H zrs(OqT8vd%&YSif%xYzuIlEQU@bc3+a`}zc3IF&4*M9EYIjwP;xQpw4^6c5O5n0P8 zA)B;n1}=aVJ_Mq%oMFN)0utKpx^C(M?LI{d+;!}Ty1IIq+0~XSC+n>&YBLDSh$X0E zT{*N<)*nN{!F|&<&-#GA%=qtr85-{aRe&GZBsPkO0L`_Okg;WsTE7zN(H#(uj{9a9 z;;evD7bg?5&&g1TS#7}2SO&>h)?Z|o=Fs+SOiRnedj9aA)>9h5lg8dbi8tD32uz3} zUiJ1N0PJF$Mk*(Mcf;7aY-lbR1N>}Z6zz8!A5PbH!+5SY98Xj+kzv0EX@qR67VnXC zOa4vz>2cNkI3nIV&UUZXfiFNY zTF=K|=nfSP;}*T7Eov2E+*5WM1H8){qQ@UmF6vg-0fgsTJnPGnbr_SB>Y>s2@iCC1 zd66Uu9pJCZk3vW8W>%#OuIpx1>ofw!dfRz@H7fy%^@l+(29bTGJ@`Th6l>~M1URKs z5ul3)(qFsUd8TS zn2j66)nqJ>DgN~vHp_9A+!Vugi;4P#tBx5157bitSScZxBidVz*d+3)d*!dUEs z0qYLj_!hr^4&K_n{9QgiK9ww@N*RQKF6f!I`yc<;+oKE4fXyI~KOaO@K?eB<6A7^@ z)j_ad)mvMm--aMe+I#fq(V}tyd&QcnU{o_{9=(IOM|2UatQ`U@M$ROp3ar^jUMNgm z5=Ltuj0kqQ4k+eUDgZoZNKF3(=w7pC4Zp>@gTc$@%$Xwz=p4rYoQEv9mH3qR1R601 z4$)a2%d&p`dM_M+b;gGK_U$8vYJfOa^wh@hFJCCIeKU9l^h^e(;-neNgKxUUJ-a2r zKYRnsoDYC$1E?9+c^UAY6d>6`pmi^BLO5Ui1lsX)!~XN7>G$OzS2jtLwyfVwY*+Nn z?a!_&D;99Ns#qZB^TpQPix7|nAL;kx$~wZHNDT^sOPr03p{X~W?y}+Ame}8y%UCWW zP&|Qy*`TqA>I7?wFsM@`97(~&7;$JaD5A8_xdBe(4Y?pLWEr7_gY!6$!hQ&?lbRHW zQ9&M>>RQV;PU9?jychd^_W-dUlW)ENuJP?a*=0XDi_Pi-h-NW@4;E~5^XUPKIze9} z0FF6-Bnwt{hM=iqagj6pTh`+EX4Q00+Z^>vS|8~AtG&IwkV0o7i9_tQi6BY@ZAUz) zgr`6Uiy&B!%gYry8?aG-jL-;-Nd&_T`~T|$F{6#TSoe~w?e1zy9UwkGdEod4Hd=)QCCw_YmS)qGE41Vqo`>>P)74dr)h_W zS+sUKx|o@__-~4czs&m^H0IGS)4|Ovk3Vvm_PrN3|80x^c;oG-cy029s}Iav#$QV| zvl@S`1!|=;dt=IS%&f+g#hBR}Qx{`qHD*@h(+bRZ*iS1#f6lDN%xZjGftgrz>QxB@d-+tjTvvl=t2@o@!q|M&VvT5Jw;$F4TPFi0{4qPx;spE1y08|X2{2w0ab zH2=cDz*?ir!#o_^|DP(`fn|+X%i1*8yilkwE5*c@aTAK<-5aLWsnTsx@d|(}vsGZu zyNsvP$N?C&^I!hG2=fj#(U@#sCfxuX6DSXhD_El}UX7;Pa@LFCqcN1<)iDEJ59ShMaHRU8MXY0i z!f$)@lUyzILqRWHfD$>)h^y=BE<-oaA9J9EfKl)hZ{}7s&E6gQoT;~l%E7u8Xf){F zH&>^w-jTG^yeLlVELpNd5tNrPY3b-Q6@sN<5nY#;SF>ZQ!VZVm%Sag{Y-q{SrNk)B zOTQzc0bG_ZT~qj-JG*T@XI|c2JERR}WG@hH6rpY(o|LqCvav;yiKK(WzF_0J>XHlN zc`F<*Wz?!@6DMii)rcmBl2Ir;dSl4+PLnhSM*YD)v4tkx&Cv$ej<>0a(Xs{8EXX3{ zc~A=$bu67VSKwO+?HywMJYU@F!Gi|{7Q>+vBcU|a`m`iUBf0>ELKL_VJ+5V_R*UsJ{0Nxaf z!PJZZl*J3b`%OTjFwm_Xz!`r~_R^+GJ<^Jyc+9UV-nVR!N?r~bOX}eUELb6P_1uAH z-!PVOct%vX$J3MZ@%6P&h&3w@A>4@aZ{G;owajNMa|i%LqxVGY8#@5}jfFgjWo@g~ zm9=b5ctTy@?IYb<8i0?h^#FH-A1=m9J^vlIOpn_I8fDD14#oLx))4lg=D0od!RnK1 zs$05g)I2sJJEm#j?)qTt_+oUN@@kH{%x2PsrD08LM@RaN!AlXqE2fPrc3S8uc!8?o zTBp(IXM^lGw!~KfGDXTHuaOMgQ6U6+1b_x8b$WmY$7S;YBS!4~JP=G2f^$OX*I$3- z?&T#!gVRL(JAlBtpSZv z8#L)QZE3~9cm?q*SJsMFuhG#_Viq9I9nt=d0nCLLfsL8QrS%6tpXF(#Vry$lGwbX& zUkTF^4|)q)klSxXj^?Nkw0q3>1}qI zANn1PG8v znY=yReHuBWO%&3VJs@_tk&Tr&47|<0KwwDpa?pIOCK_%{eD>2(K{O^8xwJ652e|as z7`k!ZevAj*m2oq~(+&e+Q=*%LUi)1%k2Keu<>ljx%69Avu{8o~mIjSj-%OLp z8CSs)CeNmcby=huwo z#A!Plu!ZYP_MW4OqwWxx_hIF8{NEk-}jsc2i~1E8fUKxwck;W%OzTTJa2 zXP@m65oyC}dF>a^T5Wvcnjqu#hrW{QK&eZp$M`woN=BbFNXrmv$F8K%q;6mk91{B! zI$=p?sT&96z}1TVWQq@NfANc59Um{R5-4trM~B&6S;V_81ktqu3zs(Ub!{-j2Ld48 zEkvqEG9#y6pe^eSPOl)MisCdqzO;%30}+MRzc^qnTP&y*fB&R0C|YDz!xkc7M+)!oi^zSSR*;=_2N;hlxx+r zB5=v8=Cqu)2N9G2kOThsti6f_IH1eWz8F7rnAV2!1Bo}>;&|J__gAHD#wk7fd#y4p?VB@sjlG5C-721x2U+dKUf>q5qEgV zik+@b#1?@|qrU;Y)@4iWW4KY?Jlj!x(gmW_P-uY_H3mhwIuK^6y?Q_;3ON1w5B`7T z6Ke%aRiD{idbr9Z7McxPv6B|#D9F|f;dz+4t~Wz$Yzz)Ab-+V|As6egQ93uHF_;A7 z+1LO8C&9?IfxCTU|35X*6n$-;{@O?t55*5$CngWF9^j1FLhr!Z!3Z)4ZD^P8512oH zK5L{i^w5Jm;(g+`5$^-djcKca5RWZ%0}8>;byX-I1R#=NW8W0>0*q$+=MH^@-%Kop zn$YWEj(8~+)pPV41OEWM#M$ri0fbHkq~+^OkGql*>k9T$39Y6kw0%^TdYk@?0Vvt!@_00 zPI)FOKbN_uherw_F|)FN=6ZS2`m)mEA3b{HglBHI6TZ>*RlMjRk@AcUADyBtrS<7ac8RJ93E8k>t_ zsBDwgPHu$j27>E?nLDOc`c{o8CeNgH{$&65&wnNZThr-^P$Ymlm1_a zvSo>&J5lMK#l##naSjJ%lo@9ZaR#{}lG_(}&Ae4x-H6j2o46O=*?AOXf?FI~&UME1 zo+57uL2m9Nzv=!Bb5E9|9mtYi#~d?uj#bTR^AT%nofvmibRAGV4~{>#GBYHxD(KF8 zigIxuwxki>@*L039M0SM0USFKlW0T~A}HIkvpohoy8`5XaYKAdmYnp%!MIg5MwBa1 z#%}TQ@}hVUvUsgZh+hwyF^sftON`xD^XD5sxx9wo9$rpmRERm&bMJsnHK2m8z#mj( zpfOgAKaFv!`Rzdz_1M{?EAJJ+X$>!f^2PeMt?x^SdG$}m^Ue8|)6XI_q6rC7^yzs7 z_93qIrlTw(0vsflt_Ml6M9&zLluOWtm9|9q2k{Qmk~Evvol#pzV^1;!YON7Hjn#Sx zP9C4~=<1ZhansGRGtqY8+o=03!c!|}SMbHXD8;hXLqkaU{AVvJUj$v0S}HZobOB5ZH!tf3`i${Z5N0;O##>KU;v6l@7oI* zOT!440N^1N@pAhe;W5qu@eO=hg9$OO^RYR8!{$&R z%GGu|5hgWW`*6Mc-Wx(p^_TA=u$!Wy0K{=6NuH-~j;G*1C{geh8~eGO+ySo3=PH$s~uKZNH2tz8@Vj7moi!Y)GqeOMMP|!-4&@4x`5FPwMrXW z-xnMc7oXy&W#aKAW4Vi?ouvMKulEh>4;yY%jK_gtV-V_(OF6oWX=fQ*A?SP2Z6Y4 zcWFe%(^7!*ST~IQy#X}atWXCtC}@En2R`%z;q9w zNYa%B8^a}?(n&nZeGjX?_WgD4dqDLjObYEw`1q8$;e%9*+#E*w689pXUo-L(?Td?v z>-$g0+gXfsP>ZMwbh=EaMn`N`5uM$2iNd5Y7LMds|IAukUc)u#fJ!(d&o>uZMCE`f zw1t#f0i{?R17NAS889MP>x!H3ac276Mk>#5r9s}!s%v3G+@#Ci_^w}RDRZykG`F<1 zx(^=N3pnBMr+p38Q9sI%=N(01;-%{#FxFd&xChElwc*IX1&UY~3Wo@_LM%HOnL~#w zAETDb(dK=w-A&JN6k2MJhdKR}FJfYX1zE4hBl66#=bG6=x~ZnB#4-zf;Ee%(t(p~9 zrFr*I@Zm;6d9k8g)A@Tw;{#1mUSb3TbL`@-EF?8f5V39p82%Z3QgK9v&kZRPl_t6) zHbh-02`?WJXX-1+fWG!b9hZ+3$QDQ_7Ou=rHIO3!>GwR$NE^sx5S7L zo}3WYq=>{ZK>4Z-^A!H187qrW&GI?PM42P8X;CurViH z%H#Z(i8iyCiHHo*R%~DvF0=GT4gkfhlxO14-ha*YM6;N7Dl^|v;R+)z3xJ+-Pi{n= z5R|FmQVn{Jz@l%5exlfeR1y4d%vR0QoWmsHI#XzD1^JTqrB%w2o$vbljiIDu3oZDQ zZoA;Qq7T?`PzPB>D1ISf@Zr%pUaNH^%uQd(O&t|6N>sbpR8imtOyo1AP@cOgge@V5 z({=sqZ7d@q+1dr$n!csljZ_E8jT1yVb?0p6-c8I@>;{UPVrI&1i>POT%J7Yv7GV2z zoU|9IgpQyfX_{t(z=4gLp;^fQIDDJZ`tm2-QH~9E$2L9O5nO2@b5@v1VpZy!s6=Sk zLM5*VX?92mh9rD(L<;%Mf-luHW9fS!@auz1iz=q^HNwHCWyH9IS9U7EhG4rrXmBem zER;M^;1y?KfvAmWe!0+W%9=`Iib#GMmOE6)qes8h2W3~YG(M)D{VHAxs2jGm$5bk9|-EcD;%Ps0Qq~o`zkQf%z@`?-co!e67UGReO zSWVBnuj2FUP_-&V3Gu4r5I8*xfYHuupiIsfT40pJbDXf#P4X%njk!(~B?B^-oF;q{ zA8OHwZ$m2GDv&gb8Bj587}GnKBpurWb+qdbl|oaH7`FYgh(r5bzu34@7AeP5Ew+n| z;Qv8`%`YPjTRrPLTei)G^rsI{`TFZ;=(NlvDOXnw6rl+~@#x$jD+)!2A9D^UqEQ z&ZKI|{Gjd>PKsg@R~Jl~1h70{!z&4GGm4I+I_N{TBRn$3(#npZ&l2;eDlKZHHVqrm`SxB^ggFMt8Jn;p z)Uo)%A#|{#QMRUPi&XgpP3K}0w~yXBj0R|X$Yv&~8C7kpq4rVHKou=V9)@|HshxOK zka*3hENz^Jz~sLTHWb6xErk^*ALhJOy@}NJHNKkP9w+pPQ~z)z;rYM%>MJ7q)fJv|BCuX;#l=U>q_UcBHUj zzlsoF+3=Dl(r<> zZWb45qnvuD+j~ogR;a@-B@$&SQyII%k$L}BQkMn=`%ShJrbCP|K}qJ?deA?SLOOPT z=!Cp+0?JSoSe1I3iGb$2FXb>OHA~e(O;gY$&+rT8UKWbJH6Z>3vwxFm&tobNp1`T` z4o0YwkKd)OQQ39fa-wxLV<|Bi7QU|k-x%c3%guU_0#)EeRG$fk=$j1i;*sG9_60_4t`}JqdRVcqWAp=iw zo~>ze|8$M$QJ9781J#GO>~wm8aLb#L;U)!S3BL5s&TYG~_0C?pew}a6e|_zZ76YPg zjKlc}%B43A(70HTO`hIQ3z^4vK6T3Lbw36IHk8hi*aDS(QK{%zTL+&m`icA>cIHfM z4Jq#4fS@|W$hTvor#(4Ko2GTF*O4ZN%^OnMwPP1j@de%YiPJdP)Qh5_Jw(67YwN(H zF7N*CN@xKzd)uHl(4rk3k3>Tf9P}hIL8Bx=26YhEq<_)VrPltY>>gtDrm8Cys_$Xr zWLbb+cE!5jpMD^|XQHg#luezgu6Tyqy=L;RU2h!DahSy%6T5-h2T4HF@ClMdU-Y+> z;RV9bvAGm#lp87m+p`GL(_XYUQECW-jjTX%y&j#}aByzd99aSPO>>bGROF{o>5WJ3 zucxDj(H8yJAQiTSYJ!bab$S|C%7Nr7@HrwPVK8u1JQ_wT_2w%D2r~Gh-Y5nivuyvh zjoyY14Qm|W5@^T>?X5M=eB8M9BkVx>IyucLKda-McR5d+C}kXW&y1%0wL<^@ft=&s*~a){ z;xXT-o3xT>129UzpQ_4mZZu>XUEKg5#Z^*)^!akCu^}ktA{wXz%4ul7i6SNOPt&k5 zCIa=STn&Sq@d}5E(k}Lg$(#G}+NWKe)NBq5UV=uKhzQ(3B9&bg6=_D&7L)^p=s$gn zDwRZPA_|MP6zHIyw-^SkysM06bjd8iIh=8YnKTiyK@dyL5nD)f6T-*d{oo3H5IZ#C z46~ur`yE_H)|0by1<$CFnmdJtkDonz{xbXULD@CqoTyz9#mWMIQo>U`r zmtG;nqZGdViyG{;h|7ORlJ2&9)uF3!E<>iFPD+~sZZeME|q;Q;&nXw@L%}t_d zrdl-t!X{&_uh18D9);XN0TPa+h9ii$xpC@0bHf96?jR`!GxmYW3-$UIITQCM-I!L3 z!cJTn?nMk}I%?NkKd z8`NPvy}U@pMF80s(LDr*7Ncy`MnxnvSj#N5Fnlw)Wo80ht8 zU05#d1OF~Vg^!RAAY`q{|*|gP$&(*dv`TleH-v=+Qdt5`|9`KOyeX(MSX&-vp2YpJTo#a zAx?n+^cZYXBGJkrGj9x3L01-qKJao&Kp#-5ORef<=mE+j`Nv2$QniuJe)YTq1%ppq2zqmbbh6Zy(@ zLkGR0(LOXgzbel|FKx@uE;LF?y{!X3Nr;@l@4e>=Ijy9q)62{6&z25E!J;ILb4E=Z zS;sh*e;X*#)vfXBFf?=sjX@{Z*%}_4O_BgthQJ82nF<3!e5h!WL>1y^{vL9wsC<-s zxrl%>OA-FlPyJ_~{L({*Sk#heHdYa@7c+1~g_Q62m%!8ZZSX^ zPrr<|dX2F50;=W_#p`aL(GWQE$M2V51XGTU7v+uIUcrdM3TY$}4|a$Q>Ow zj8l$2NBPK)9r?(G(uXZYwA}CgQQ6--_}nHxuGSs>Vm9h&k;UkA4TS^Kqu0vIV_U$Z zeAmMW?O%$JcoSNoMNgDBmqI66C$wBEl=uqK$Q7Wvk#hn}EVsdna=Ts3W|5D8Z*3ZN*9OV}rQjnc)yn5~h9#-$M=?5>^RKG;0ULu8V&X(hY6!(yoLH zdz^T9sLzVyrjyx;ZQy5wu6mcv2&&UQ&_0tiE~SKr$YTzsO@7;yry?Hs4a*8Ecvh;2 ze1T&#(&3QtTx22?pau6nk>SXok>$sxe2Y*T2OR^`(btKq9m7%uK%}b}urJ+?Wav*w zL6kvQGv*^3j>x-h`5*DA$DSX4uitq&g@6own4T{?=K;uT$`N7;2rh344iwWdO`~H% z$49E45TcYv?7Kd>I%=z^MZo>B`~ewV;O$%lkOk!jA~f{k#n1``I^%B1D9w*!!4G6_ zwxu~nEo-#?KD10OtOmJD0`I;u|O!NwD+@Zbltk{g%$6|!E(#!<3Lo{W=;{D{;uwkeF` zXbX~AM|s{R(3$&TjDzN>*W*lp-d|lkI(w;7k1-6RVcb`7<23qKZ*cvqhbt7*I2@Ce(I^3gN6L+RxNM zQ7@4F#)W8X2)DsTU996w8AVb+E>*4-_8=`C0PKLUk+E710-^)Nb;I97eq*@nkjJf0 zHrFm;@^CVrB1BIqdZOl7J!j&}A)8Gd8IxMlf}_GmCKOy&iG-oDuh&|IES_Y0!oa7| zm4#ZZBC_%}n5($WTd6lti11-K%#-VVi#qWTAO-E}R6}{!#%xQxF2MwCk0TC?P8iyH6i{CKNvr; zZ?^yGNzalfdmu4J!Cruu7a36SG%c?)uve{hbnBFBCH|9Qs#pObbHU4tkh&1{@Y{|W zJnPQ+Z+s}73*UR;1X=*ZaE5To8LT%>sb=sDJLt?80K!$tEw!N}VSUAgGDnJq2sbdv zViO?$?neqwXnVzlqAHG`-Ky(Qxa0n)`6EOtK z;u1mpky7X=NbxA378FErAa`}TE@3!Tx4*xig+iA82Whjr2PK>jGrl#iF$CE(thmq|OdE++}KVFW^DG(4L+L+?u+zI2T;n404w!tSzE;&zasP+2z};gNJh`KXCP)PcHf67c9DF z^lvvb`M)0e@BRC0{wI(;3{MQc(WN^$b}@fowZk=K{B`pcGzF&o99#Rzl}!H6hX+1| z@00)by!mNQetgrGbJOWAOy5nf8I-*I7(sq~{f#HnsVrtLVDtQo=7$FNocIO+eumkn MbWkzjhhsngF9xniQUCw| literal 0 HcmV?d00001 From e0b210d65186a429254c6106095775e21dcef78e Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 22 Aug 2025 19:55:31 +0000 Subject: [PATCH 24/32] pre commit before documentation commit --- analyze_graph_type_comparison.py | 462 +++++++ analyze_validation_errors.py | 62 +- .../sample_enhanced_conditioning.yaml | 4 +- .../sample_enhanced_conditioning_sweep.yaml | 16 +- docs/conditional-gen-branch-documentation.md | 566 ++++++++ docs/pretrained_spatiotemporal.md | 1166 +++++++++++++---- graph_type_comparison_3d_histogram.csv | 17 + graph_type_comparison_3d_histogram.png | Bin 0 -> 646402 bytes scripts/slurm/sweep.sh | 8 +- .../slurm/train_enhanced_long_comparison.sh | 18 +- .../denoiser_conditional_spatiotemporal.yaml | 2 +- validation_errors_plot.csv | 18 +- validation_errors_plot.png | Bin 125597 -> 169982 bytes 13 files changed, 2030 insertions(+), 309 deletions(-) create mode 100644 analyze_graph_type_comparison.py create mode 100644 docs/conditional-gen-branch-documentation.md create mode 100644 graph_type_comparison_3d_histogram.csv create mode 100644 graph_type_comparison_3d_histogram.png diff --git a/analyze_graph_type_comparison.py b/analyze_graph_type_comparison.py new file mode 100644 index 0000000..cdfd7ce --- /dev/null +++ b/analyze_graph_type_comparison.py @@ -0,0 +1,462 @@ +#!/usr/bin/env python3 +""" +Script to analyze validation errors for runs from the graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17 group. + +This script: +1. Scrapes all runs from the specified WandB group +2. For each run, extracts lag_subsample_rate and total_lag_time +3. Loads validation data with the same parameters and max_datasets=1 +4. Computes validation errors for each model +5. Creates a 3D histogram plot with lag_subsample_rate (x), total_lag_time (y), and validation error (height) +""" + +import os +import sys +import logging +import numpy as np +import matplotlib.pyplot as plt +from mpl_toolkits.mplot3d import Axes3D +import torch +import wandb +from pathlib import Path +from typing import List, Dict, Any, Optional, Tuple +import pandas as pd +from tqdm import tqdm + +# Add jamun to path +sys.path.insert(0, '/homefs/home/sules/jamun/src') + +import jamun +from jamun.model.denoiser_conditional import Denoiser +from jamun.utils.checkpoint import find_checkpoint, get_wandb_run_config +from jamun.data import parse_datasets_from_directory, parse_repeated_position_datasets_from_directory +from jamun.data._dloader import MDtrajDataModule +import torch_geometric + +# Setup logging +logging.basicConfig(level=logging.INFO) +logger = logging.getLogger(__name__) + +def get_data_path(): + """Get the data path from environment or common locations.""" + data_path = os.getenv("JAMUN_DATA_PATH") + if data_path is None: + # Try common locations + possible_paths = [ + "/data/bucket/kleinhej/", + "/data2/sules/", + "/homefs/home/sules/data/" + ] + for path in possible_paths: + if Path(path).exists(): + data_path = path + break + + if data_path is None: + raise ValueError("JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable.") + + logger.info(f"Using data path: {data_path}") + return data_path + +def scrape_wandb_runs(group_name: str, project: str = "sule-shashank/jamun") -> List[wandb.Api.run]: + """Scrape all runs from the specified WandB group.""" + logger.info(f"Scraping runs from group: {group_name}") + + api = wandb.Api() + runs = api.runs(project, filters={'group': group_name}) + runs_list = list(runs) + + logger.info(f"Found {len(runs_list)} runs in group '{group_name}'") + + return runs_list + +def extract_run_parameters(runs: List[wandb.Api.run]) -> List[Dict[str, Any]]: + """Extract lag_subsample_rate and total_lag_time from each run.""" + logger.info("Extracting lag_subsample_rate and total_lag_time from runs...") + + run_params = [] + + for run in tqdm(runs, desc="Processing runs", unit="run"): + try: + config = run.config + if 'cfg' not in config: + logger.warning(f"Run {run.name} missing 'cfg' in config") + continue + + cfg = config['cfg'] + + # Extract lag_subsample_rate + lag_subsample_rate = None + try: + lag_subsample_rate = cfg['data']['datamodule']['datasets']['train']['lag_subsample_rate'] + except (KeyError, TypeError): + try: + # Try alternative path + lag_subsample_rate = cfg['data']['datamodule']['datasets']['lag_subsample_rate'] + except (KeyError, TypeError): + logger.warning(f"Could not extract lag_subsample_rate for run {run.name}") + continue + + # Extract total_lag_time + total_lag_time = None + try: + total_lag_time = cfg['data']['datamodule']['datasets']['train']['total_lag_time'] + except (KeyError, TypeError): + try: + # Try alternative path + total_lag_time = cfg['data']['datamodule']['datasets']['total_lag_time'] + except (KeyError, TypeError): + logger.warning(f"Could not extract total_lag_time for run {run.name}") + continue + + # Extract model target for filtering + model_target = cfg.get('model', {}).get('_target_') + + # Extract data target for filtering + data_target = None + try: + data_target = cfg['data']['datamodule']['datasets']['train']['_target_'] + except (KeyError, TypeError): + try: + data_target = cfg['data']['datamodule']['datasets']['_target_'] + except (KeyError, TypeError): + logger.warning(f"Could not extract data target for run {run.name}") + + run_info = { + 'run': run, + 'run_path': '/'.join(run.path), + 'run_name': run.name, + 'lag_subsample_rate': lag_subsample_rate, + 'total_lag_time': total_lag_time, + 'model_target': model_target, + 'data_target': data_target, + 'cfg': cfg + } + + run_params.append(run_info) + logger.info(f"Added run: {run.name} (lag_subsample_rate={lag_subsample_rate}, total_lag_time={total_lag_time})") + + except Exception as e: + logger.warning(f"Error processing run {run.name}: {e}") + continue + + logger.info(f"Successfully extracted parameters from {len(run_params)} runs") + return run_params + +def load_validation_data(total_lag_time: int, lag_subsample_rate: int, val_root: str = "/data2/sules/ALA_ALA_enhanced_full_grid/val/") -> torch_geometric.loader.DataLoader: + """Load validation data for error computation with specific lag parameters.""" + logger.info(f"Loading validation data with total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + + if not os.path.exists(val_root): + raise ValueError(f"Validation directory not found: {val_root}") + + try: + datasets = parse_repeated_position_datasets_from_directory( + root=val_root, + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + filter_codes=None, + as_iterable=False, + subsample=1, + total_lag_time=total_lag_time, + lag_subsample_rate=lag_subsample_rate, + max_datasets=10 # Use 10 datasets for validation + ) + + if not datasets: + raise ValueError("No validation datasets found") + + # Create data module + data_module = MDtrajDataModule( + datasets={'val': datasets}, + batch_size=32, + num_workers=0, + persistent_workers=False + ) + data_module.setup('val') + + val_loader = data_module.val_dataloader() + logger.info(f"Loaded validation data with {len(datasets)} datasets") + + return val_loader + + except Exception as e: + logger.error(f"Error loading validation data: {e}") + raise + +def load_model_and_compute_scaled_rmse(run_info: Dict[str, Any], val_loader: torch_geometric.loader.DataLoader) -> float: + """Load a model from checkpoint and compute validation scaled RMSE.""" + run_path = run_info['run_path'] + run_name = run_info['run_name'] + + logger.info(f"Loading model for run: {run_name}") + + try: + # Find checkpoint + checkpoint_path = find_checkpoint( + wandb_train_run_path=run_path, + checkpoint_type="last" + ) + + # Load model + model = Denoiser.load_from_checkpoint(checkpoint_path, strict=False) + model.eval() + + # Move to device + device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') + model = model.to(device) + + logger.info(f"Model loaded successfully for run: {run_name}") + + # Compute validation scaled RMSE + total_scaled_rmse = 0.0 + total_batches = 0 + + with torch.no_grad(): + for batch in tqdm(val_loader, desc="Computing validation", leave=False, unit="batch"): + batch = batch.to(device) + + # Use the model's validation logic + sigma = model.sigma_distribution.sample().to(device) + loss, aux = model.noise_and_compute_loss( + batch, + sigma, + align_noisy_input=model.align_noisy_input_during_evaluation + ) + + # Extract scaled RMSE from aux dictionary + if 'scaled_rmsd' in aux: + scaled_rmse_value = aux['scaled_rmsd'].mean().item() + total_scaled_rmse += scaled_rmse_value + elif 'rmsd' in aux and 'mse' in aux: + # Compute scaled RMSE manually if not available + rmsd_value = aux['rmsd'].mean().item() + # scaled_rmsd = rmsd / (sigma * sqrt(D)) - but we'll use rmsd as fallback + total_scaled_rmse += rmsd_value + logger.warning(f"Using RMSD instead of scaled RMSD for {run_name}") + else: + logger.warning(f"Scaled RMSD not found in aux dictionary for {run_name}. Available keys: {list(aux.keys())}") + # Fallback to loss if scaled RMSD not available + total_scaled_rmse += loss.mean().item() + + total_batches += 1 + + avg_scaled_rmse = total_scaled_rmse / total_batches if total_batches > 0 else float('inf') + logger.info(f"Validation scaled RMSE for {run_name}: {avg_scaled_rmse:.6f}") + + return avg_scaled_rmse + + except Exception as e: + logger.error(f"Error computing validation error for run {run_name}: {e}") + return float('inf') + +def create_3d_histogram(results: List[Dict[str, Any]], output_path: str = "graph_type_comparison_3d_histogram.png"): + """Create a 3D histogram plot of validation scaled RMSE errors.""" + logger.info("Creating 3D histogram plot...") + + # Convert to DataFrame for easier manipulation + df = pd.DataFrame(results) + + # Debug: Print available columns and data ranges + logger.info(f"Available DataFrame columns: {list(df.columns)}") + logger.info(f"Lag subsample rate range: {df['lag_subsample_rate'].min()} - {df['lag_subsample_rate'].max()}") + logger.info(f"Total lag time range: {df['total_lag_time'].min()} - {df['total_lag_time'].max()}") + logger.info(f"Validation scaled RMSE range: {df['validation_scaled_rmse'].min()} - {df['validation_scaled_rmse'].max()}") + + # Get unique values for x and y axes + unique_lag_subsample_rates = sorted(df['lag_subsample_rate'].unique()) + unique_total_lag_times = sorted(df['total_lag_time'].unique(), reverse=True) # Reverse order: 8 to 2 + + logger.info(f"Unique lag subsample rates: {unique_lag_subsample_rates}") + logger.info(f"Unique total lag times: {unique_total_lag_times}") + + # Create meshgrid for positioning bars + x_positions = np.arange(len(unique_lag_subsample_rates)) + y_positions = np.arange(len(unique_total_lag_times)) + X, Y = np.meshgrid(x_positions, y_positions, indexing='ij') + + # Initialize heights array + heights = np.zeros((len(unique_lag_subsample_rates), len(unique_total_lag_times))) + + # Fill heights array with validation scaled RMSE values + for _, row in df.iterrows(): + x_idx = unique_lag_subsample_rates.index(row['lag_subsample_rate']) + y_idx = unique_total_lag_times.index(row['total_lag_time']) + heights[x_idx, y_idx] = row['validation_scaled_rmse'] + + # Create 3D plot + fig = plt.figure(figsize=(10, 10)) + ax = fig.add_subplot(111, projection='3d') + + # Flatten arrays for bar3d + x_flat = X.flatten() + y_flat = Y.flatten() + z_bottom = np.zeros_like(x_flat) + heights_flat = heights.flatten() + + # Create bars with proper color normalization + dx = dy = 0.8 # Width of bars + + # Create proper normalization for colors + from matplotlib.colors import Normalize + norm = Normalize(vmin=heights_flat.min(), vmax=heights_flat.max()) + colors = plt.cm.viridis(norm(heights_flat)) + + ax.bar3d(x_flat, y_flat, z_bottom, dx, dy, heights_flat, color=colors, alpha=0.8, edgecolor='black', linewidth=0.5) + + # Set labels and title with font size 16 for axis labels, 14 for tick labels + ax.set_xlabel('Lag Subsample Rate', fontsize=16) + ax.set_ylabel('Total Lag Time', fontsize=16) + ax.set_zlabel('Test Scaled RMSE', fontsize=16) + ax.set_title('3D Histogram: Test Scaled RMSE vs Lag Parameters\nGroup: graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17', fontsize=14) + + # Set custom tick labels with font size 14 + ax.set_xticks(x_positions) + ax.set_xticklabels([str(rate) for rate in reversed(unique_lag_subsample_rates)], fontsize=14) + ax.set_yticks(y_positions) + # Reverse the tick labels while keeping positions from 2 to 8 + ax.set_yticklabels([str(time) for time in reversed(unique_total_lag_times)], fontsize=14) + ax.tick_params(axis='z', labelsize=14) + + # Add colorbar with the same normalization + mappable = plt.cm.ScalarMappable(cmap='viridis', norm=norm) + mappable.set_array(heights_flat) + cbar = plt.colorbar(mappable, ax=ax, shrink=0.5, aspect=20) + cbar.set_label('Test Scaled RMSE', fontsize=16) + cbar.ax.tick_params(labelsize=14) + + # Adjust view angle for better visualization + ax.view_init(elev=20, azim=45) + + plt.tight_layout() + plt.savefig(output_path, dpi=300, bbox_inches='tight') + logger.info(f"3D histogram saved to: {output_path}") + + # Also save data to CSV + csv_path = output_path.replace('.png', '.csv') + df.to_csv(csv_path, index=False) + logger.info(f"Data saved to: {csv_path}") + + # Save as structured numpy array + npy_path = "graph_type_comparison_validation_errors.npy" + data = np.array(list(zip(df['lag_subsample_rate'].values, + df['total_lag_time'].values, + df['validation_scaled_rmse'].values)), + dtype=[('lag_subsample_rate', 'i4'), + ('total_lag_time', 'i4'), + ('validation_scaled_rmse', 'f8')]) + np.save(npy_path, data) + logger.info(f"Structured array saved to: {npy_path}") + +def main(): + """Main function to execute the graph type comparison analysis.""" + logger.info("Starting graph type comparison validation analysis...") + + # Configuration + group_name = "graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17" + project = "sule-shashank/jamun" + + try: + # Step 1: Scrape WandB runs + runs = scrape_wandb_runs(group_name, project) + + # Step 2: Extract parameters from runs + run_params = extract_run_parameters(runs) + + if not run_params: + logger.error("No runs with valid parameters found in the specified group") + return + + # Step 3: Compute validation errors for each run + results = [] + + # Create a cache for validation loaders to avoid reloading same data + validation_cache = {} + + # Create progress bar with more detailed description + total_runs = len(run_params) + pbar = tqdm(total=total_runs, desc="Processing runs", unit="run") + + for i, run_info in enumerate(run_params): + try: + lag_subsample_rate = run_info['lag_subsample_rate'] + total_lag_time = run_info['total_lag_time'] + + # Update progress bar with current parameters + pbar.set_description(f"Processing lag_sub={lag_subsample_rate}, lag_time={total_lag_time}") + + # Create cache key + cache_key = (total_lag_time, lag_subsample_rate) + + # Load validation data (use cache if available) + if cache_key not in validation_cache: + logger.info(f"Loading validation data for lag_time={total_lag_time}, lag_subsample={lag_subsample_rate}") + val_loader = load_validation_data(total_lag_time, lag_subsample_rate) + validation_cache[cache_key] = val_loader + else: + val_loader = validation_cache[cache_key] + logger.info(f"Using cached validation data for lag_time={total_lag_time}, lag_subsample={lag_subsample_rate}") + + # Compute validation scaled RMSE + validation_scaled_rmse = load_model_and_compute_scaled_rmse(run_info, val_loader) + + result = { + 'run_name': run_info['run_name'], + 'run_path': run_info['run_path'], + 'lag_subsample_rate': lag_subsample_rate, + 'total_lag_time': total_lag_time, + 'validation_scaled_rmse': validation_scaled_rmse, + 'model_target': run_info['model_target'], + 'data_target': run_info['data_target'] + } + results.append(result) + + logger.info(f"Processed {run_info['run_name']}: Scaled RMSE={validation_scaled_rmse:.6f}") + + except Exception as e: + logger.error(f"Error processing run {run_info['run_name']}: {e}") + continue + finally: + # Update progress bar + pbar.update(1) + + # Close progress bar + pbar.close() + + if not results: + logger.error("No successful results obtained") + return + + # Step 4: Create 3D histogram + create_3d_histogram(results) + + # Print summary + logger.info("\n" + "="*50) + logger.info("SUMMARY") + logger.info("="*50) + logger.info(f"Total runs processed: {len(results)}") + + # Group by lag parameters and show statistics + df = pd.DataFrame(results) + param_groups = df.groupby(['lag_subsample_rate', 'total_lag_time'])['validation_scaled_rmse'] + + logger.info("\nValidation Scaled RMSE by lag parameters:") + for (lag_subsample, lag_time), group in param_groups: + mean_scaled_rmse = group.mean() + std_scaled_rmse = group.std() if len(group) > 1 else 0 + logger.info(f"lag_subsample={lag_subsample}, lag_time={lag_time}: Scaled RMSE={mean_scaled_rmse:.6f} Ā± {std_scaled_rmse:.6f} (n={len(group)})") + + # Find best performing combination + best_result = min(results, key=lambda x: x['validation_scaled_rmse']) + logger.info(f"\nBest performing combination:") + logger.info(f"Run: {best_result['run_name']}") + logger.info(f"Lag subsample rate: {best_result['lag_subsample_rate']}") + logger.info(f"Total lag time: {best_result['total_lag_time']}") + logger.info(f"Validation Scaled RMSE: {best_result['validation_scaled_rmse']:.6f}") + + except Exception as e: + logger.error(f"Error in main execution: {e}") + raise + +if __name__ == "__main__": + main() diff --git a/analyze_validation_errors.py b/analyze_validation_errors.py index 7ec3aa3..541a7f7 100644 --- a/analyze_validation_errors.py +++ b/analyze_validation_errors.py @@ -185,7 +185,7 @@ def load_validation_data(total_lag_time: int, val_root: str = "/data2/sules/ALA_ subsample=1, total_lag_time=total_lag_time, lag_subsample_rate=1, - max_datasets=5 # Use a few datasets for validation + max_datasets=100 ) if not datasets: @@ -210,7 +210,7 @@ def load_validation_data(total_lag_time: int, val_root: str = "/data2/sules/ALA_ raise def load_model_and_compute_rmsd(run_info: Dict[str, Any], val_loader: torch_geometric.loader.DataLoader) -> float: - """Load a model from checkpoint and compute validation RMSDĀ².""" + """Load a model from checkpoint and compute validation MSE (RMSDĀ²).""" run_path = run_info['run_path'] run_name = run_info['run_name'] @@ -220,7 +220,7 @@ def load_model_and_compute_rmsd(run_info: Dict[str, Any], val_loader: torch_geom # Find checkpoint checkpoint_path = find_checkpoint( wandb_train_run_path=run_path, - checkpoint_type="best_so_far" + checkpoint_type="last" ) # Load model @@ -233,8 +233,8 @@ def load_model_and_compute_rmsd(run_info: Dict[str, Any], val_loader: torch_geom logger.info(f"Model loaded successfully for run: {run_name}") - # Compute validation squared RMSD - total_rmsd_squared = 0.0 + # Compute validation MSE (which equals RMSDĀ²) + total_mse = 0.0 total_batches = 0 with torch.no_grad(): @@ -244,31 +244,31 @@ def load_model_and_compute_rmsd(run_info: Dict[str, Any], val_loader: torch_geom # Use the model's validation logic sigma = model.sigma_distribution.sample().to(device) loss, aux = model.noise_and_compute_loss( - batch.pos, batch, - batch.batch, - batch.num_graphs, sigma, align_noisy_input=model.align_noisy_input_during_evaluation ) - # Extract RMSD from aux dictionary and square it - if 'rmsd' in aux: + # Extract MSE from aux dictionary (since rmsd = sqrt(mse), we want mse for squared error) + if 'mse' in aux: + mse_value = aux['mse'].mean().item() + total_mse += mse_value # This is actually MSE (RMSDĀ²) + elif 'rmsd' in aux: rmsd_value = aux['rmsd'].mean().item() - total_rmsd_squared += rmsd_value ** 2 # Square the RMSD + total_mse += rmsd_value ** 2 # Square the RMSD to get MSE else: - logger.warning(f"RMSD not found in aux dictionary for {run_name}. Available keys: {list(aux.keys())}") - # Fallback to loss if RMSD not available - total_rmsd_squared += loss.mean().item() + logger.warning(f"Neither MSE nor RMSD found in aux dictionary for {run_name}. Available keys: {list(aux.keys())}") + # Fallback to loss if neither MSE nor RMSD available + total_mse += loss.mean().item() total_batches += 1 - print(f"total_rmsd_squared: {total_rmsd_squared}") + print(f"total_mse: {total_mse}") print(f"total_batches: {total_batches}") breakpoint() - avg_rmsd_squared = total_rmsd_squared / total_batches if total_batches > 0 else float('inf') - logger.info(f"Validation RMSDĀ² for {run_name}: {avg_rmsd_squared:.6f}") + avg_mse = total_mse / total_batches if total_batches > 0 else float('inf') + logger.info(f"Validation MSE (RMSDĀ²) for {run_name}: {avg_mse:.6f}") - return avg_rmsd_squared + return avg_mse except Exception as e: logger.error(f"Error computing validation error for run {run_name}: {e}") @@ -304,7 +304,7 @@ def plot_validation_errors(results: List[Dict[str, Any]], output_path: str = "va # Find the correct validation column name for plotting validation_col = None - for col in ['validation_rmsd_squared', 'validation_rmsd', 'validation_error']: + for col in ['validation_mse', 'validation_rmsd_squared', 'validation_rmsd', 'validation_error']: if col in df.columns: validation_col = col break @@ -325,8 +325,8 @@ def plot_validation_errors(results: List[Dict[str, Any]], output_path: str = "va fontsize=8, alpha=0.7) plt.xlabel(x_param.replace('_', ' ').title()) - plt.ylabel('Validation RMSDĀ²') - plt.title('Validation RMSDĀ² for Spatiotemporal Multimeasurement Models\nGroup: noise_check_experiment_multimeasurement_vs_correlation') + plt.ylabel('Validation MSE (RMSDĀ²)') + plt.title('Validation MSE for Spatiotemporal Multimeasurement Models\nGroup: noise_check_experiment_multimeasurement_vs_correlation') plt.grid(True, alpha=0.3) # Set x-axis limits to 2-10 if appropriate @@ -342,10 +342,10 @@ def plot_validation_errors(results: List[Dict[str, Any]], output_path: str = "va df.to_csv(csv_path, index=False) logger.info(f"Data saved to: {csv_path}") - # Save validation RMSDĀ² paired with total_lag_time as npy file + # Save validation MSE paired with total_lag_time as npy file # Find the correct validation column name validation_col = None - for col in ['validation_rmsd_squared', 'validation_rmsd', 'validation_error']: + for col in ['validation_mse', 'validation_rmsd_squared', 'validation_rmsd', 'validation_error']: if col in df.columns: validation_col = col break @@ -357,11 +357,11 @@ def plot_validation_errors(results: List[Dict[str, Any]], output_path: str = "va if 'total_lag_time' in df.columns: # Create structured array with total_lag_time and validation values data = np.array(list(zip(df['total_lag_time'].values, df[validation_col].values)), - dtype=[('total_lag_time', 'i4'), ('validation_rmsd_squared', 'f8')]) + dtype=[('total_lag_time', 'i4'), ('validation_mse', 'f8')]) npy_path = "validation_errors_spatiotemporal_multimeasurement.npy" np.save(npy_path, data) logger.info(f"Validation data with total_lag_time saved to: {npy_path}") - logger.info(f"Data format: structured array with fields 'total_lag_time' and 'validation_rmsd_squared'") + logger.info(f"Data format: structured array with fields 'total_lag_time' and 'validation_mse'") logger.info(f"Used validation column: {validation_col}") else: # Fallback to just validation values if total_lag_time not available @@ -411,7 +411,7 @@ def main(): # Step 4: Compute validation errors for each model results = [] total_runs = sum(len(runs_for_lag_time) for runs_for_lag_time in runs_by_lag_time.values()) - + breakpoint() with tqdm(total=total_runs, desc="Processing models", unit="model") as pbar: for lag_time, runs_for_lag_time in runs_by_lag_time.items(): logger.info(f"Loading validation data for total_lag_time={lag_time}") @@ -420,12 +420,12 @@ def main(): for run_info in runs_for_lag_time: pbar.set_description(f"Processing {run_info['run_name'][:20]}...") - validation_rmsd_squared = load_model_and_compute_rmsd(run_info, val_loader) + validation_mse = load_model_and_compute_rmsd(run_info, val_loader) result = { 'run_name': run_info['run_name'], 'run_path': run_info['run_path'], - 'validation_rmsd_squared': validation_rmsd_squared, + 'validation_mse': validation_mse, **{k: v for k, v in run_info.items() if k not in ['run', 'cfg']} } results.append(result) @@ -439,10 +439,10 @@ def main(): logger.info("SUMMARY") logger.info("="*50) logger.info(f"Total runs processed: {len(results)}") - logger.info("Top 5 best performing models (lowest RMSDĀ²):") - sorted_results = sorted(results, key=lambda x: x['validation_rmsd_squared']) + logger.info("Top 5 best performing models (lowest MSE):") + sorted_results = sorted(results, key=lambda x: x['validation_mse']) for i, result in enumerate(sorted_results[:5]): - logger.info(f"{i+1}. {result['run_name']}: {result['validation_rmsd_squared']:.6f}") + logger.info(f"{i+1}. {result['run_name']}: {result['validation_mse']:.6f}") except Exception as e: logger.error(f"Error in main execution: {e}") diff --git a/configs/experiment/sample_enhanced_conditioning.yaml b/configs/experiment/sample_enhanced_conditioning.yaml index 173d580..95aa3df 100644 --- a/configs/experiment/sample_enhanced_conditioning.yaml +++ b/configs/experiment/sample_enhanced_conditioning.yaml @@ -26,7 +26,7 @@ init_datasets: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 20 + max_datasets: 10 label_override: "ALA_ALA" # model: @@ -34,7 +34,7 @@ init_datasets: # N_structures: ${init_datasets.total_lag_time} num_sampling_steps_per_batch: 1000 -num_batches: 5 +num_batches: 10 num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true diff --git a/configs/experiment/sample_enhanced_conditioning_sweep.yaml b/configs/experiment/sample_enhanced_conditioning_sweep.yaml index 90968db..2037d9c 100644 --- a/configs/experiment/sample_enhanced_conditioning_sweep.yaml +++ b/configs/experiment/sample_enhanced_conditioning_sweep.yaml @@ -19,31 +19,35 @@ defaults: init_datasets: _target_: jamun.data.parse_datasets_from_directory - root: "/data2/sules/ALA_ALA_enhanced_full_swarm/val" + root: "/data2/sules/ALA_ALA_enhanced_long/val" traj_pattern: "^(.*).xtc" pdb_pattern: "^(.*).pdb" as_iterable: false subsample: 1 + num_frames: 6 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 10 + max_datasets: 20 label_override: "ALA_ALA" # model: # conditioner: # N_structures: ${init_datasets.total_lag_time} -num_sampling_steps_per_batch: 100 +num_sampling_steps_per_batch: 1000 num_batches: 1 -num_init_samples_per_dataset: 2 +num_init_samples_per_dataset: 1 repeat_init_samples: 1 continue_chain: true +batch_sampler: + mcmc: + history_update_frequency: 10 # Add your checkpoint path here - update with actual trained model path -wandb_train_run_path: "sule-shashank/jamun/qiutegoj" +wandb_train_run_path: sule-shashank/jamun/k20xo7qb # checkpoint_type: last # checkpoint_dir: "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/" -checkpoint_type: epoch=83-step=88872.ckpt +checkpoint_type: last sigma: 0.06 M: 1.0 diff --git a/docs/conditional-gen-branch-documentation.md b/docs/conditional-gen-branch-documentation.md new file mode 100644 index 0000000..2e45370 --- /dev/null +++ b/docs/conditional-gen-branch-documentation.md @@ -0,0 +1,566 @@ +# Conditional Generation Branch Documentation + +This document provides a comprehensive walkthrough of the conditional-gen branch, covering the spatiotemporal model architecture, conditional denoising, and memory-based sampling mechanisms. + +## Table of Contents + +1. [Spatiotemporal Model Architecture](#spatiotemporal-model-architecture) +2. [Conditional Denoiser Architecture](#conditional-denoiser-architecture) +3. [Model.g Parameterization and E3Conv Conditional](#modelg-parameterization-and-e3conv-conditional) +4. [Memory-Based Sampling: BAOAB/ABOBA Subroutines](#memory-based-sampling-baoababoba-subroutines) +5. [Sampling Memory Wrapper](#sampling-memory-wrapper) + +--- + +## Spatiotemporal Model Architecture + +### Overview + +The spatiotemporal model (`E3SpatioTemporal`) implements a complete workflow that processes molecular structures with temporal dependencies by converting between spatial and temporal graph representations. + +### Core Components + +#### 1. Spatial to Temporal Graph Conversion + +The model begins by converting spatial graphs to temporal graphs using the `spatial_to_temporal_graphs()` function: + +```python +def spatial_to_temporal_graphs(batch, graph_type="fan"): + """ + Convert a batch of spatial graphs to temporal graphs with configurable connectivity. + + For each spatial node with position + hidden states, create a temporal graph where: + - Node 0: current position + - Nodes 1-T: hidden state positions + - Connectivity depends on graph_type parameter + """ +``` + +**Graph Type Options:** +- **"fan"**: Hub-spoke + sequential connections (0ā†’all, iā†’(i+1)) +- **"hub_n_spoke"**: Only hub-spoke connections (0ā†’all, no sequential) +- **"complete"**: Complete graph without self-loops (all-to-all excluding self) +- **"complete_no_self"**: Complete graph with self-loops (all-to-all including self) + +#### 2. Temporal Position Calculation + +Temporal positions are normalized to create consistent temporal embeddings: + +```python +def calculate_temporal_positions(temporal_length, mode="linear", device=None): + """Calculate normalized temporal positions [0, 1/T, 2/T, ..., (T-1)/T]""" + positions = torch.linspace(0, 1, temporal_length + 1, device=device)[:-1] + return positions +``` + +### Complete Spatiotemporal Workflow + +The `E3SpatioTemporal.forward()` method implements the following pipeline: + +#### Step 1: Spatial Graph ā†’ Temporal Graph Conversion +```python +temporal_batch = spatial_to_temporal_graphs(batch, graph_type=self.graph_type) +``` + +#### Step 2: Spatial Processing of All Time Steps +For each position (current + hidden states), the spatial module processes: +```python +# Current positions +node_attr_current = self.spatial_module( + pos=batch.pos, + topology=topology, + batch=batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff +) + +# Hidden state positions +for hidden_pos in batch.hidden_state: + node_attr_hidden = self.spatial_module( + pos=hidden_pos, + topology=topology, + batch=batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff + ) +``` + +#### Step 3: Spatial-Temporal Feature Assembly +```python +node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) # [N, T, features] +``` + +#### Step 4: Spatialā†’Temporal Pooling +```python +temporal_node_attr = self.spatial_to_temporal_pooler(node_attr_spatial_temporal, temporal_batch) +``` + +#### Step 5: Temporal Processing +The temporal module processes the temporal graph using an E3 Transformer: +```python +temporal_output = self.temporal_module( + temporal_node_attr, + temporal_batch, + self.radial_cutoff, + self.temporal_cutoff +) +``` + +#### Step 6: Temporalā†’Spatial Pooling +```python +spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) +``` + +#### Step 7: Return to Spatial Representation +The final spatial features are returned for use by the conditional architecture. + +--- + +## Conditional Denoiser Architecture + +### Denoiser_Conditional Overview + +The `Denoiser` class in `denoiser_conditional.py` extends the standard JAMUN denoiser with conditional generation capabilities through the integration of conditioner modules. + +### Key Components + +#### 1. Conditioner Integration + +The denoiser accepts a `conditioner` parameter that processes the input batch to generate conditioning structures: + +```python +def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + if self.conditioning_module is None: + return self.conditioner_default(y) # Returns [y.pos] + elif callable(self.conditioning_module): + return self.conditioning_module(y) + else: + raise ValueError("Conditioner must be a callable or None") +``` + +#### 2. Hidden State Management + +The denoiser properly handles hidden states throughout the pipeline: + +**Adding Noise to Hidden States:** +```python +def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]): + # Add noise to current positions + y.pos = x.pos + sigma * noise + # Add noise to hidden states + for i in range(len(y.hidden_state)): + y.hidden_state[i] = x.hidden_state[i] + sigma * hidden_noise[i] +``` + +**Hidden State Alignment:** +```python +def _align_A_to_B_batched_with_hidden_states(self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch): + # Align positions + A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) + # Align hidden states + if hasattr(A, "hidden_state") and A.hidden_state is not None: + A_aligned.hidden_state = [] + for i in range(len(A.hidden_state)): + A_aligned.hidden_state.append(kabsch_algorithm( + A.hidden_state[i], B.pos, A.batch, A.num_graphs + )) +``` + +### The `xhat_normalized` Method: Core Conditional Processing + +The `xhat_normalized` method is the heart of conditional generation: + +#### Step 1: Normalization Factor Computation +```python +c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) +radial_cutoff = self.effective_radial_cutoff(sigma) / c_in +``` + +#### Step 2: Input Scaling +```python +y_scaled = y.clone() +y_scaled.pos = y.pos * c_in +# Scale hidden states +if hasattr(y, "hidden_state") and y.hidden_state is not None: + y_scaled.hidden_state = [] + for positions in y.hidden_state: + y_scaled.hidden_state.append(positions * c_in) +``` + +#### Step 3: Conditioning Structure Generation +```python +with torch.cuda.nvtx.range("conditioning"): + conditioned_structures = self.conditioner(y_scaled) +``` + +The conditioner returns a list of tensors that will be concatenated for model input. + +#### Step 4: Model Prediction via model.g +```python +with torch.cuda.nvtx.range("g"): + g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), + topology=y_scaled, + c_noise=c_noise, + effective_radial_cutoff=radial_cutoff) +``` + +**Key Point**: `model.g` receives the concatenated conditioned structures as input positions. + +#### Step 5: Output Construction and Hidden State Update +```python +xhat.pos = c_skip * y.pos + c_out * g_pred +if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] # Hidden state shifts forward +``` + +**Hidden State Evolution**: The hidden state list is updated by: +1. Moving current position (`y.pos`) to the front +2. Keeping all previous hidden states except the oldest one +3. This creates a sliding window of historical positions + +--- + +## Model.g Parameterization and E3Conv Conditional + +### Architecture Variants + +The conditional-gen branch introduces several E3Conv variants designed for different conditioning scenarios: + +#### 1. E3ConvConditional + +**Purpose**: Basic conditional model that handles multiple structure inputs +**Key Parameters**: +- `N_structures`: Number of input structures to concatenate +- `irreps_sh`: Extended to `N_structures * self.irreps_sh` for spherical harmonics applied in parallel to N_structures-many 3D vectors. + +**Forward Pass**: +```python +def forward(self, pos: Tensor, topology, c_noise, effective_radial_cutoff): + # pos should be [batch_size*N, 3T] where T is number of time-steps + positions = torch.split(pos, 3, dim=-1) + edge_sh = [] + for block in positions: + edge_vec = block[src] - block[dst] + edge_sh.append(self.sh(edge_vec)) + edge_sh = torch.cat(edge_sh, dim=-1) # Concatenate spherical harmonics +``` + +#### 2. E3ConvConditionalSpatioTemporal + +**Purpose**: Specialized for spatiotemporal conditioning where input combines physical positions with spatial features + +**Key Design**: +- Expects input: `[y.pos, spatial_features]` concatenated along feature dimension +- `N_structures = 1` (processes combined input as single structure) +- `input_attr_irreps`: Defines irreps for spatial features component + +**Forward Pass Logic**: +```python +def forward(self, pos: Tensor, topology, c_noise, effective_radial_cutoff): + # Split positions: first 3 coords are physical, rest are spatial features + pos_physical = pos[:, :3] # [N, 3] - physical coordinates + pos_features = pos[:, 3:] # [N, spatial_features_dim] - spatial features + + # Compute edge spherical harmonics ONLY for physical positions + edge_vec_physical = pos_physical[src] - pos_physical[dst] + edge_sh = self.sh(edge_vec_physical) + + # Combine node_attr with spatial features as input attributes + # ... +``` + +### Configuration Examples + +**Standard Conditional Model**: +```yaml +arch: + _target_: jamun.model.arch.E3ConvConditional + N_structures: 2 # [current_pos, conditioned_structure] + irreps_sh: "1x0e + 1x1e" +``` + +**Spatiotemporal Conditional Model**: +```yaml +arch: + _target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalSpatioTemporal + N_structures: 1 # Combined [y.pos, spatial_features] + input_attr_irreps: "120x0e + 32x1e" # Match spatiotemporal output +``` + +--- + +## Memory-Based Sampling: BAOAB/ABOBA Subroutines + +### Overview + +The conditional-gen branch introduces memory-enhanced versions of BAOAB and ABOBA splitting schemes that maintain and update a history of molecular states during sampling. + +### Core Memory Concepts + +#### 1. Historical State Management +- `y_hist`: List of previous molecular configurations +- States are maintained in chronological order: `[newest, ..., oldest]` +- History updates occur at configurable intervals via `history_update_frequency` + +#### 2. Conditional Density Equilibration +The key innovation is equilibration to conditional densities `p(y_t | y_hist)` rather than marginal densities. + +### BAOAB_Memory Algorithm + +#### Algorithm Structure +```python +def baoab_memory(y, y_hist, score_fn, steps, history_update_frequency=1, ...): + """BAOAB splitting scheme that updates a state history.""" +``` + +#### Key Parameters +- `y`: Current molecular state +- `y_hist`: History of previous states +- `score_fn`: Score function that accepts both `y` and `y_hist` +- `history_update_frequency`: How often to update the history (inner loop iterations) + +#### Main Algorithm Loop + +**Outer Loop** (Main sampling steps): +```python +for i in steps_iter: + # Inner equilibration loop + for j in range(1, history_update_frequency): + # BAOAB step for conditional density p(y_t | y_hist) + y_current = y.clone().detach() + v = v + u * (delta / 2) * psi # B step + y = y + (delta / 2) * v # A step + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R # O step + y = y + (delta / 2) * vhat # A step + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + v = vhat + (delta / 2) * psi # B step + + # Update history + y_hist.pop(-1) # Remove oldest state + y_hist.insert(0, y_current) # Add current state as newest +``` + +### Inner Loop Mechanics: The Role of Equilibration + +#### Purpose of the Inner Loop + +Rather than taking a single MCMC step, the inner loop allows the system to equilibrate to the conditional distribution `p(y_t | y_hist)` given the fixed history. + + +#### Mathematical Justification +``` +Standard MCMC: y_{t+1} ~ p(y | y_t) +Memory MCMC: y_{t+1} ~ p(y | y_hist) after equilibration to p(y | y_hist) +``` + +### ABOBA_Memory Algorithm + +Similar structure to BAOAB_memory but with ABOBA splitting: + +```python +def aboba_memory(y, y_hist, score_fn, steps, history_update_frequency=1, ...): + """ABOBA splitting scheme that updates a state history.""" + for i in steps_iter: + for j in range(1, history_update_frequency): + # ABOBA inner loop for equilibration + y_current = y.clone().detach() + y = y + (delta / 2) * v # A step + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + v = v + u * (delta / 2) * psi # B step + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R # O step + v = vhat + (delta / 2) * psi # B step + y = y + (delta / 2) * v # A step +``` + +### Cleanup Option + +Both algorithms support a `cleanup` option for denoising: + +```python +if cleanup is not None and cleanup and sigma is not None: + y_current = y.clone().detach() + _, orig_score = score_fn_processed(y_current, y_hist=y_hist) + y_denoised_and_noised = y_current + (sigma**2)*orig_score + sigma*torch.randn_like(y_current) + y_hist.pop(-1) + y_hist.insert(0, y_denoised_and_noised) + y = y_denoised_and_noised +``` + +This performs a denoising step followed by re-noising before adding to history. + +### Configuration + +```yaml +batch_sampler: + _target_: jamun.sampling.mcmc.BAOAB_memory + delta: ${delta} + friction: ${friction} + steps: ${num_sampling_steps_per_batch} + history_update_frequency: 10 # Inner loop iterations + save_trajectory: true + cpu_offload: true + verbose: true +``` + +--- + +## Sampling Memory Wrapper + +### ModelSamplingWrapperMemory + +The `ModelSamplingWrapperMemory` class extends the standard sampling wrapper to handle models that depend on historical states. + +#### Key Features + +#### 1. History-Aware Initialization +```python +def __init__(self, model, init_graphs, sigma, recenter_on_init=True): + # Apply mean centering to positions + self.init_graphs = mean_center(self.init_graphs) + + # Mean center hidden states if they exist + if hasattr(self.init_graphs, 'hidden_state') and self.init_graphs.hidden_state: + for i in range(len(self.init_graphs.hidden_state)): + mean = scatter(self.init_graphs.hidden_state[i], self.init_graphs.batch, dim=0, reduce="mean") + self.init_graphs.hidden_state[i] = self.init_graphs.hidden_state[i] - mean[self.init_graphs.batch] +``` + +#### 2. History Sampling +```python +def sample_initial_noisy_history(self) -> list: + """Sample initial noisy history from hidden states.""" + noisy_history = [] + for hidden_state in self.init_graphs.hidden_state: + noisy_history.append(hidden_state + torch.randn_like(hidden_state) * self.sigma) + return noisy_history +``` + +#### 3. Memory-Aware Score and Prediction Functions +```python +def score(self, y, y_hist, sigma): + """Score function that includes history.""" + graph = self.positions_to_graph(y, y_hist).to(self.device) + return self._model.score(graph, sigma) + +def xhat(self, y, y_hist, sigma): + """Prediction function that includes history.""" + graph = self.positions_to_graph(y, y_hist).to(self.device) + xhat_graph = self._model.xhat(graph, sigma) + return xhat_graph.pos +``` + +#### 4. Graph Construction with History +```python +def positions_to_graph(self, positions: torch.Tensor, y_hist: list) -> torch_geometric.data.Data: + """Wraps positions to a graph and attaches the historical states.""" + input_graph = self.init_graphs.clone() + input_graph.pos = positions + input_graph.hidden_state = y_hist # Attach history as hidden_state + return input_graph.to(positions.device) +``` + +#### 5. History-Aware Sample Unbatching +The wrapper handles complex unbatching of trajectory data that includes historical information: + +```python +def unbatch_samples(self, samples: dict[str, torch.Tensor]): + """Unbatch samples including history trajectories.""" + for key, value in samples.items(): + if key == "y_hist" or key == "y_hist_traj": + # Special handling for history data + if key == "y_hist": + value = [value] + value = torch.stack([torch.stack(traj, dim=1) for traj in value], dim=1) + # ... complex unbatching logic for history +``` + +### Integration with SamplerMemory + +The `SamplerMemory` class uses `ModelSamplingWrapperMemory`: + +```python +class SamplerMemory(Sampler): + def sample(self, model, batch_sampler, num_batches, init_graphs, continue_chain=False): + model_wrapped = utils.ModelSamplingWrapperMemory( + model=model, + init_graphs=init_graphs, + sigma=batch_sampler.sigma, + ) + + y_init = model_wrapped.sample_initial_noisy_positions() + y_hist_init = model_wrapped.sample_initial_noisy_history() + + # Memory-aware sampling + out = batch_sampler.sample(model=model_wrapped, y_init=y_init, + v_init=v_init, y_hist_init=y_hist_init) +``` + +### Chain Continuation + +The memory wrapper supports continuing chains across batches: + +```python +if continue_chain: + y_init = out["y"] + v_init = out["v"] + y_hist_init = out["y_hist"] # Continue with updated history +else: + y_init = model_wrapped.sample_initial_noisy_positions() + y_hist_init = model_wrapped.sample_initial_noisy_history() + v_init = "gaussian" +``` + +--- + +## Usage Examples + +### Training a Spatiotemporal Conditional Model + +```yaml +model: + _target_: jamun.model.denoiser_conditional.Denoiser + arch: + _target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalSpatioTemporal + N_structures: 1 + input_attr_irreps: "120x0e + 32x1e" + conditioner: + _target_: jamun.model.conditioners.SpatioTemporalConditioner + spatiotemporal_model: + _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal + spatial_module: # ... spatial E3Conv config + temporal_module: # ... temporal E3Transformer config +``` + +### Memory-Based Sampling Configuration + +```yaml +sampler: + _target_: jamun.sampling.SamplerMemory + devices: 1 + +batch_sampler: + _target_: jamun.sampling.mcmc.BAOAB_memory + delta: 0.04 + friction: 1.0 + steps: 1000 + history_update_frequency: 10 + save_trajectory: true +``` + +### Data Loading for Memory Models + +```yaml +init_datasets: + _target_: jamun.data.parse_repeated_position_datasets_from_directory + root: "/path/to/data" + total_lag_time: 5 # Number of historical states + lag_subsample_rate: 1 # Temporal subsampling + subsample: 1 # Spatial subsampling +``` + +This comprehensive documentation covers the key innovations in the conditional-gen branch, providing both conceptual understanding and practical implementation details for spatiotemporal modeling and memory-based sampling in molecular dynamics. + diff --git a/docs/pretrained_spatiotemporal.md b/docs/pretrained_spatiotemporal.md index 86700cb..7521b17 100644 --- a/docs/pretrained_spatiotemporal.md +++ b/docs/pretrained_spatiotemporal.md @@ -1,337 +1,1009 @@ -# Pretrained Spatiotemporal Models +# KALA-JAMUN: Spatiotemporal Conditional Generation Documentation -This document describes how to use pretrained denoiser models as spatial and temporal modules in the spatiotemporal transformer architecture. +## Introduction -## Overview +KALA-JAMUN introduces conditioning into the JAMUN workflow to enable temporal-aware molecular generation. The key innovation is conditioning the denoising process on past noisy states, allowing the model to learn and maintain temporal correlations in molecular dynamics. This enhancement necessitated significant changes across three core components of the system: -The spatiotemporal transformer consists of four main components: -- **Spatial Module**: Processes spatial positions (can be a pretrained denoiser) -- **Temporal Module**: Processes temporal graphs (can be a pretrained denoiser) -- **Spatialā†’Temporal Pooler**: Converts spatial features to temporal representation -- **Temporalā†’Spatial Pooler**: Converts temporal features back to spatial representation +1. **Modified Dataset Infrastructure**: To support conditioning on historical states +2. **Enhanced Model Architectures**: To process both current and historical information +3. **Memory-Aware Sampling**: To maintain temporal consistency during generation -You can now use pretrained denoiser models directly as spatial/temporal modules using the wrapper approach, with control over: -- Whether to freeze each module (trainable: true/false) -- Loading from WandB runs or direct checkpoint paths -- **No architecture matching required** - pretrained denoisers are wrapped to expose only their `xhat` function +The conditioning mechanism works by feeding past noisy states directly to the model as part of the input data. This enables the model to learn temporal dependencies and generate more realistic molecular trajectories that respect the underlying dynamics. -## Quick Start +This document provides a comprehensive guide covering the complete KALA-JAMUN workflow from data preparation through model architecture to sampling procedures. -### 1. Basic Usage +## Table of Contents -The simplest way to use a pretrained denoiser is to specify it directly in your config: +1. [Chapter 1: Datasets](#chapter-1-datasets) +2. [Chapter 2: Architecture](#chapter-2-architecture) +3. [Chapter 3: Sampling](#chapter-3-sampling) -```yaml -# configs/experiment/my_experiment.yaml -defaults: - - override /model/conditioner: spatiotemporal_pretrained_wrapper_example +--- + +## Chapter 1: Datasets + +### Overview + +In KALA-JAMUN, the conditioning is based on past noisy states that are fed directly to the model as part of the input data. This required fundamental modifications to the data structure itself. -model: - conditioner: - spatiotemporal_model: - spatial_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "your_entity/your_project/spatial_run_id" - trainable: false # Freeze the pretrained spatial module - - temporal_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "your_entity/your_project/temporal_run_id" - trainable: true # Fine-tune the temporal module +### Data Structure Modifications + +The core data class `DataWithResidueInformation` has been enhanced with a new field to support temporal conditioning: + +**Source:** [`src/jamun/utils/data_with_residue_info.py`](src/jamun/utils/data_with_residue_info.py), lines 5-16 + +```python +class DataWithResidueInformation(torch_geometric.data.Data): + """Graph with residue-level information.""" + + pos: torch.Tensor + atom_type_index: torch.Tensor + atom_code_index: torch.Tensor + residue_code_index: torch.Tensor + residue_sequence_index: torch.Tensor + residue_index: torch.Tensor + num_residues: int + loss_weight: float + hidden_state: Any # NEW: Stores past trajectory states ``` -### 2. Mixed Approach +**Key Addition:** +- **`hidden_state`**: This new field stores the historical molecular configurations that enable temporal conditioning. It keeps the past trajectories that the model will condition on during the denoising process. -You can also mix pretrained modules with modules trained from scratch: +This modification allows the data graph to carry both current molecular state (`pos`) and historical context (`hidden_state`), enabling the model to learn temporal dependencies in molecular dynamics. -```yaml -model: - conditioner: - spatiotemporal_model: - # Use pretrained spatial module - spatial_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "entity/project/spatial_run_id" - trainable: false - - # Train temporal module from scratch - temporal_module: - _target_: jamun.model.arch.spatiotemporal.E3Transformer - irreps_out: "3x1e" - irreps_hidden: "8x0e + 4x1e" - # ... other E3Transformer parameters +### 1.1 MDTrajDataset with Subsampling + +The core dataset class `MDTrajDataset` has been enhanced to support KALA-JAMUN's conditioning mechanism through historical state management. + +#### Enhanced MDTrajDataset Structure + +When a data graph is processed in KALA-JAMUN: +- **`graph.pos`**: Contains the current molecular state (positions) +- **`graph.hidden_state`**: Contains `total_lag_time - 1` past states + +The historical states are selected using two key parameters: +- **`lag_subsample_rate`**: Temporal difference between consecutive stored states +- **`total_lag_time`**: Total number of states stored (including the present state) + +#### Subsampling Implementation + +**Source:** [`src/jamun/data/_mdtraj.py`](src/jamun/data/_mdtraj.py), lines 249-261 (in MDTrajDataset.__init__) + +```python +if total_lag_time is not None and lag_subsample_rate is not None: + lagged_indices = get_subsampled_indices( + self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate + ) + # Extract subsampled indices (first element of each list) + subsampled_indices = [indices[0] for indices in lagged_indices] + # Extract lagged indices (all except first element) + self.lagged_indices = [indices[1:] for indices in lagged_indices] + # Subsample the trajectory using the subsampled indices + self.hidden_state = [self.traj[indices] for indices in self.lagged_indices] + self.traj = self.traj[subsampled_indices] # self.traj is permanently modified. ``` -### 3. Available Config Templates +**Example**: With `total_lag_time=5` and `lag_subsample_rate=10`: +- Present state: frame 100 +- Historical states: frames 90, 80, 70, 60 +- Total stored: 5 states (1 current + 4 historical) -Use one of the provided config templates: +### 1.2 Loading with parse_datasets_from_directory -- `spatiotemporal_pretrained_wrapper_example.yaml`: Both spatial and temporal modules from pretrained denoisers -- `spatiotemporal_mixed_example.yaml`: Pretrained spatial module + temporal module trained from scratch +Such datasets can be loaded using the `parse_datasets_from_directory` function: -## Configuration Options +**Source:** [`src/jamun/data/_utils.py`](src/jamun/data/_utils.py), lines 38-49 (function definition) -### Wrapper Parameters +```python +datasets = parse_datasets_from_directory( + root="/data/trajectories", + traj_pattern=r"traj_(\w+)\.dcd", + pdb_pattern=r"(\w+)\.pdb", + total_lag_time=5, + lag_subsample_rate=10, + max_datasets=100 +) +``` -The pretrained wrapper accepts these parameters: +This function automatically: +- Discovers trajectory files using regex patterns +- Matches trajectory files with corresponding PDB topology files +- Creates `MDTrajDataset` objects with proper historical state management +- Applies the index subsampling procedure to populate `hidden_state` -```yaml -spatial_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "username/project/run_id" # Load from WandB run - checkpoint_path: "/path/to/checkpoint.ckpt" # OR load from direct path - checkpoint_type: "best_so_far" # "last", "best_so_far", or specific .ckpt filename - trainable: true # Whether to keep trainable (false = freeze) +### 1.3 RepeatedPositionDataset (Multimeasurement) + +Multimeasurement refers to collecting T independent noisy copies of the same present state. This enables the model to learn from multiple noise realizations applied to identical molecular configurations, improving robustness and sample diversity. + +#### Concept + +Instead of using historical states from different time points, multimeasurement uses: +- **`batch.pos`**: The current molecular state +- **`batch.hidden_state`**: Contains `T-1` copies of the same `pos` + +When noise is independently added in the denoiser, independent realizations of the noise get added to the exact same underlying state, allowing the model to learn the noise distribution more effectively. + +#### MDTrajRepeatedDataset + +To enable multimeasurement, KALA-JAMUN provides a specialized dataset: + +**Source:** [`src/jamun/data/noisy_position_dataset.py`](src/jamun/data/noisy_position_dataset.py), lines 5-37 + +```python +class RepeatedPositionDataset(MDtrajDataset): + def __getitem__(self, idx: int) -> torch_geometric.data.Data: + graph = super().__getitem__(idx) + # Hidden state contains repeated copies of the current position + # instead of historical states from different time points + return graph ``` -### Key Parameters +#### Loading with parse_repeated_position_datasets_from_directory -- **`wandb_run_path`**: Load from WandB run (format: "username/project/run_id") -- **`checkpoint_path`**: Direct path to checkpoint file (mutually exclusive with wandb_run_path) -- **`checkpoint_type`**: Which checkpoint to load ("best_so_far", "last", or specific filename) -- **`trainable`**: Whether the module should be trainable (default: true, false = freeze) +Multimeasurement datasets are created using: -## Common Use Cases +**Source:** [`src/jamun/data/_utils.py`](src/jamun/data/_utils.py), lines 362-373 (function definition) -### 1. Load Pretrained Spatial, Train Temporal from Scratch +```python +datasets = parse_repeated_position_datasets_from_directory( + root="/data/trajectories", + traj_pattern=r"traj_(\w+)\.dcd", + pdb_pattern=r"(\w+)\.pdb", + # No temporal parameters - using repeated states instead + max_datasets=100 +) +``` -```yaml -# Use: spatiotemporal_mixed_example.yaml -model: - conditioner: - spatiotemporal_model: - spatial_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "username/project/spatial_pretrained_run" - trainable: false # Freeze pretrained spatial - - temporal_module: - _target_: jamun.model.arch.spatiotemporal.E3Transformer - irreps_out: "3x1e" - irreps_hidden: "8x0e + 4x1e" - # ... standard E3Transformer configuration -``` - -### 2. Fine-tune Both Modules +**Key Difference**: +- **Temporal Conditioning**: `hidden_state` contains past states from different time points +- **Multimeasurement**: `hidden_state` contains repeated copies of the current state -```yaml -# Use: spatiotemporal_pretrained_wrapper_example.yaml -model: - conditioner: - spatiotemporal_model: - spatial_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "username/project/spatial_run" - trainable: true # Fine-tune pretrained spatial - - temporal_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "username/project/temporal_run" - trainable: true # Fine-tune pretrained temporal -``` - -### 3. Load from Different Checkpoint Sources +This approach allows the model to: +1. Learn from multiple noise realizations on the same state +2. Improve denoising performance through ensemble-like training +3. Generate more diverse samples from the same initial condition -```yaml -model: - conditioner: - spatiotemporal_model: - spatial_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - checkpoint_path: "/shared/models/spatial_best.ckpt" - trainable: true - - temporal_module: - _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser - wandb_run_path: "username/project/temporal_run" - trainable: true +--- + +## Chapter 2: Architecture + +### Overview + +KALA-JAMUN implements a new denoiser class, `denoiser_conditional.Denoiser`, which handles the internal training process for temporal conditioning. This model consists of two main operating submodules, the conditioning module and the architecture module, called according to the following workflow: + +1. Just like `jamun.model.Denoiser`, the modules within `jamun.model.denoiser_conditional.Denoiser` are called from within `xhat_normalized`. +2. The **Conditioning Module**: `denoiser_conditional.Denoiser.conditioner` - first calculates features based on historical states +3. Next, the **Architecture Module**: `denoiser_conditional.Denoiser.g` - processes features from the conditioning module into the final output, which is combined with the noisy coordinates with the normalization factors to construct xhat. + + +**Key Components:** +1. **Conditioning Module**: Implements various conditioning strategies, with the spatiotemporal conditioner being the most sophisticated +2. **Architecture Module (model.g)**: Enhanced E3Conv variants that handle conditional inputs +3. **Training Process**: Integrated mean centering, alignment, scaling, propagation, and loss computation + +**āš ļø Important**: The input signatures for `denoiser_conditional.Denoiser` do not match those of the original `Denoiser`. + + + +### 2.1 Conditioning Module + +Every conditioning module in KALA-JAMUN is of the class `Conditioner`, which defines the interface for calculating features based on historical states. + +**Source:** [`src/jamun/model/conditioners/conditioners.py`](src/jamun/model/conditioners/conditioners.py) + +#### Available Conditioning Modules + +KALA-JAMUN provides several conditioning strategies: + +1. **PositionConditioner**: Returns just the input positions (baseline) +2. **MeanConditioner**: Provides mean-centered positions and repeated structures +3. **SpatioTemporalConditioner**: Uses spatiotemporal processing for feature extraction (most sophisticated) + +#### SpatioTemporalConditioner + +The most sophisticated conditioning module uses a spatiotemporal GNN to output features based on both current and historical states. This conditioner processes the input through a complete spatiotemporal architecture before passing features to the main denoiser. + +**Source:** [`src/jamun/model/conditioners/conditioners.py`](src/jamun/model/conditioners/conditioners.py), SpatioTemporalConditioner class + +```python +class SpatioTemporalConditioner(pl.LightningModule): + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: + # Process through spatiotemporal model + spatial_features = self.spatiotemporal_model(y, c_noise=self.c_noise) + + # Return [positions, features] for concatenation + return [y.pos, spatial_features] +``` + +### 2.2 Spatiotemporal Model (E3SpatioTemporal) + +The spatiotemporal model is the core of the `SpatioTemporalConditioner`. It processes molecular data through several stages: + +**Source:** [`src/jamun/model/arch/spatiotemporal.py`](src/jamun/model/arch/spatiotemporal.py), lines 403+ + +#### Architecture Components + +1. **Spatial Module**: Processes individual molecular graphs (current and historical states) +2. **Temporal Graph Construction**: Converts spatial graphs into temporal graphs with temporal connections +3. **Temporal Module (E3Transformer)**: Applies transformer architecture on temporal graphs +4. **Reconversion**: Converts temporal graph features back to spatial representation + +#### Spatial Module Processing + +The spatial module processes each molecular configuration (current and historical) independently: + +```python +# Process current positions +node_attr_current = self.spatial_module( + pos=batch.pos, + topology=topology, + batch=batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff +) + +# Process historical positions +for hidden_pos in batch.hidden_state: + node_attr_hidden = self.spatial_module( + pos=hidden_pos, + topology=topology, + batch=batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff + ) ``` -## How It Works +#### Temporal Graph Construction + +After spatial processing, the individual graphs are converted into temporal graphs where nodes across different time steps are connected: + +**Source:** [`src/jamun/model/arch/spatiotemporal.py`](src/jamun/model/arch/spatiotemporal.py), create_temporal_graph function + +The temporal graph construction creates edges between the intertemporal copies of the same atom. This gives some freedom as to how to define the connectivity structure of the temporal graph. Three stratgies that have been explored are: + +1. Fan graph--the present node connects to all nodes, and the ith historical node connects to the (i+1)th and (i-1)th historical node (whenever such nodes are available). +2. Hub and spoke--the present node connects to all nodes, and no historical nodes are mutually connected. +3. Complete graph--all nodes are mutually connected. -### Wrapper Mechanism +In the temporal graph we also need to define what features the nodes and the edges have. This will be discussed below. -The `DenoiserWrapper` class replicates the full denoiser logic including normalization: +#### E3Transformer (Temporal Module) + +The temporal module applies a transformer architecture specifically designed for temporal graphs: + +**Source:** [`src/jamun/model/arch/spatiotemporal.py`](src/jamun/model/arch/spatiotemporal.py), lines 217-284 + +##### Temporal Embeddings and Encoding Functions + +The E3Transformer uses several specialized encoding functions to handle temporal information: + +**Key Parameters:** +- **`irreps_node_attr_temporal`**: Irreducible representations for temporal node attributes (default: "1x1e") +- **`node_attr_temporal_encoding_function`**: Encodes temporal position information (default: "gaussian") +- **`edge_attr_temporal_encoding_function`**: Encodes temporal edge attributes (default: "gaussian") +- **`radial_edge_attr_encoding_function`**: Encodes radial distances (default: "gaussian") + +##### Temporal Attribute Processing + +```python +# Split edge attribute dimensions: radial and temporal +self.radial_edge_attr_dim = self.edge_attr_dim // 2 +self.temporal_edge_attr_dim = self.edge_attr_dim - self.radial_edge_attr_dim + +# Temporal gate for combining node attributes with temporal position +irreps_with_temporal = self.irreps_node_attr + self.irreps_node_attr_temporal +self.temporal_gate = e3tools.nn.GateWrapper( + irreps_in=irreps_with_temporal, + irreps_out=self.irreps_hidden, + irreps_gate=irreps_with_temporal, +) +``` + +The temporal gate combines: +- **Node attributes**: From spatial processing of individual timesteps +- **Temporal position**: Encoded position in the temporal sequence +- **Temporal edges**: Connections between atoms across different timesteps + +##### Transformer Layers + +The temporal transformer processes the combined spatial-temporal information through multiple attention layers: + +**Source:** [`src/jamun/model/arch/spatiotemporal.py`](src/jamun/model/arch/spatiotemporal.py), lines 267-284 + +```python +for _ in range(num_layers): + self.layers.append( + e3tools.nn.TransformerBlock( + irreps_in=self.irreps_hidden, + irreps_out=self.irreps_hidden, + irreps_sh=self.irreps_sh, + edge_attr_dim=self.edge_attr_dim, + num_heads=self.num_attention_heads, + conv=self.conv, + ) + ) +``` + +#### Integrating Pretrained Spatial Modules + +A powerful feature of KALA-JAMUN is the ability to use a pretrained unconditional JAMUN model as the spatial module within the spatiotemporal architecture. This enables leveraging existing trained models as building blocks for more sophisticated temporal conditioning. + +##### Architecture Overview + +In this setup, we have: +- **Overlying Conditional Denoiser**: `jamun.model.denoiser_conditional.Denoiser` - the main KALA-JAMUN model +- **Sub-Denoiser**: `jamun.model.Denoiser` - the pretrained unconditional JAMUN model used as spatial module + +The overlying conditional denoiser has its own normalization factor `c_in`, but when scaled data `y_scaled = c_in * y` goes into the sub-denoiser, this scaling must be divided out since the sub-denoiser has its own internal normalization. + +##### DenoiserWrapper: Input Signature Unification + +The core challenge is that `Denoiser.xhat` and `E3Conv` have different input signatures, requiring a wrapper to make them compatible: + +**Source:** [`src/jamun/utils/pretrained_wrapper.py`](src/jamun/utils/pretrained_wrapper.py), lines 56-85 ```python class DenoiserWrapper(nn.Module): - def __init__(self, denoiser_model: nn.Module, trainable: bool = True): + """ + Wrapper around a denoiser model that matches the spatial module interface. + + This allows pretrained denoiser models to be used as spatial/temporal modules + in the spatiotemporal architecture by replicating the full denoiser logic + including normalization factors computed from the denoiser's own parameters. + """ + + def __init__(self, denoiser_model: nn.Module, c_in: float = 1.0, trainable: bool = True): + """ + Args: + denoiser_model: The pretrained denoiser model + c_in: Rescaling factor to convert positions from overlaying model scale + trainable: Whether to keep the model trainable (default: True) + """ super().__init__() self.denoiser = denoiser_model + self.c_in = c_in # Rescaling factor from overlying denoiser +``` + +##### Rescaling Mechanism + +The `DenoiserWrapper` handles the critical `c_in` rescaling between the overlying and sub-denoiser: + +**Source:** [`src/jamun/utils/pretrained_wrapper.py`](src/jamun/utils/pretrained_wrapper.py), lines 98-118 + +```python +def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cutoff): + # Sample sigma from the denoiser's own sigma distribution + sigma = self.denoiser.sigma_distribution.sample().to(pos.device) + + # CRITICAL: Rescale positions from overlaying model scale + y = pos / self.c_in # Divide out overlying denoiser's c_in + + # Apply sub-denoiser's own normalization + c_in, c_skip, c_out, _ = compute_normalization_factors( + sigma, + average_squared_distance=self.denoiser.average_squared_distance, + normalization_type=self.denoiser.normalization_type, + sigma_data=self.denoiser.sigma_data, + D=y.shape[-1], + device=y.device, + ) +``` + +**Key Steps:** +1. **Input Rescaling**: `y = pos / self.c_in` - divides out the overlying denoiser's scaling +2. **Internal Normalization**: The sub-denoiser computes its own `c_in`, `c_skip`, `c_out` +3. **Denoiser Processing**: Full `xhat_normalized` logic is replicated internally +4. **Output**: Features compatible with the spatiotemporal architecture + +This design allows seamless integration of pretrained models while maintaining proper normalization hierarchies between the overlying conditional denoiser and the embedded unconditional denoiser. + +### 2.3 Architecture Module (model.g) + +The architecture module `model.g` is the main processing unit that receives features from the conditioning module. Unlike standard E3Conv models, these variants are designed to handle conditional inputs. + +#### E3ConvConditional vs Standard E3Conv + +The key difference from standard E3Conv models is the ability to process multiple input structures and conditional information: + +**Source:** [`src/jamun/model/arch/e3conv_conditional.py`](src/jamun/model/arch/e3conv_conditional.py), lines 15-40 + +```python +class E3ConvConditional(torch.nn.Module): + def __init__( + self, + # Standard E3Conv parameters... + N_structures: int = 1, # NEW: Number of input structures + # ... + ): +``` + +**Key Features:** +- **Multi-Structure Support**: Processes multiple molecular structures simultaneously via `N_structures` +- **Noise Conditioning**: Integrates noise level information throughout the network +- **Skip Connections**: Uses noise-conditional skip connections for stable training + +#### E3ConvConditionalSpatioTemporal + +A specialized variant designed specifically for spatiotemporal conditioning: + +**Source:** [`src/jamun/model/arch/e3conv_conditional.py`](src/jamun/model/arch/e3conv_conditional.py), lines 312+ + +This variant handles concatenated position and feature data from the spatiotemporal model: + +```python +def forward( + self, + pos: Tensor, # [N, 3 + spatial_features_dim] from [y.pos, spatial_features] + topology: torch_geometric.data.Batch, + c_noise: Tensor, + effective_radial_cutoff: float, +) -> Tensor: + # Split positions: first 3 coords are physical, rest are features + pos_physical = pos[:, :3] # [N, 3] - physical coordinates + pos_features = pos[:, 3:] # [N, spatial_features_dim] - spatial features + + # Compute edge attributes using ONLY physical positions + edge_vec_physical = pos_physical[src] - pos_physical[dst] + edge_sh = self.sh(edge_vec_physical) + + # Combine node attributes with spatial features + combined_attr = torch.cat([node_attr, pos_features], dim=-1) + node_attr = self.spatial_feature_aggregator(combined_attr) +``` + +**Design Principle**: Separates physical coordinates (for geometric operations) from feature coordinates (for conditioning). + +### 2.4 Noising and Denoising Process + +KALA-JAMUN implements a comprehensive training pipeline that handles mean centering, alignment, scaling, model propagation, and loss computation. + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), xhat and xhat_normalized methods + +#### Complete Denoising Workflow + +The denoising process follows these steps: + +##### 1. Mean Centering (Input Preparation) + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), lines 308-311 + +```python +if self.mean_center: + y = mean_center(y) + y = self._mean_center_hidden_states(y) +``` + +Both current and historical states are mean-centered to ensure translational invariance. + +##### 2. Alignment (if enabled) + +Molecular configurations are aligned to a reference to handle rotational variance. This is done after adding noise but before denoising during training. + +##### 3. Scaling (Normalization) + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), lines 264-286 + +```python +# Compute normalization factors +c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) + +# Scale input positions and hidden states +y_scaled = y.clone() +y_scaled.pos = y.pos * c_in +if hasattr(y, "hidden_state") and y.hidden_state is not None: + y_scaled.hidden_state = [] + for positions in y.hidden_state: + y_scaled.hidden_state.append(positions * c_in) +``` + +**Key scaling factors:** +- **`c_in`**: Scales input coordinates and hidden states +- **`c_skip`**: Skip connection scaling +- **`c_out`**: Output scaling +- **`c_noise`**: Noise conditioning scaling + +##### 4. Model Propagation + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), lines 294-299 + +```python +# Step 1: Conditioning +conditioned_structures = self.conditioner(y_scaled) + +# Step 2: Architecture processing +g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), + topology=y_scaled, + c_noise=c_noise, + effective_radial_cutoff=radial_cutoff) +``` + +The conditioner processes scaled inputs and hidden states, then the architecture module processes the concatenated conditioned structures. + +##### 5. Skip Connection and Output Scaling + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), lines 301-304 + +```python +# Apply skip connection and output scaling +xhat.pos = c_skip * y.pos + c_out * g_pred + +# Update hidden state for next iteration +if hasattr(y, "hidden_state") and y.hidden_state is not None: + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] +``` + +##### 6. Mean Centering (Output) + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), lines 317-319 + +```python +# Mean center the prediction +if self.mean_center: + xhat = mean_center(xhat) +``` + +##### 7. Loss Computation + +The loss is computed between the denoised prediction and the clean target, typically using MSE loss on the positions while maintaining proper handling of the hidden states for temporal consistency. + +#### Key Differences from Standard Denoiser + +1. **Hidden State Handling**: All scaling operations are applied to both current and hidden states +2. **Conditioning Integration**: The conditioner processes scaled inputs before the main architecture +3. **Temporal Consistency**: Hidden states are properly updated to maintain temporal sequence +4. **Multi-Structure Processing**: Conditioned structures are concatenated before processing + +This comprehensive pipeline ensures that KALA-JAMUN properly handles temporal information throughout the entire denoising process, maintaining consistency between current and historical states while applying the appropriate transformations for effective learning. + +--- + +## Chapter 3: Sampling + +### Overview + +Once the KALA-JAMUN model is trained, it is time to sample using the score function obtained via the Miyasawa-Tweedie formula from the denoiser. The score function relates the denoised prediction to the true score of the data distribution: + +``` +score(y, Ļƒ) = (xĢ‚(y, Ļƒ) - y) / ĻƒĀ² +``` + +Where `xĢ‚(y, Ļƒ)` is the denoised prediction from the trained model. + +Since KALA-JAMUN conditions on historical states, the sampling process must be modified to properly handle memory. This requires changes to the standard ABOBA and BAOAB samplers to account for the temporal dependencies and ensure that historical information is correctly propagated through the sampling chain. + +### 3.1 Model Loading + +Loading the trained conditional model is straightforward: + +**Source:** [`src/jamun/model/denoiser_conditional.py`](src/jamun/model/denoiser_conditional.py), load_from_checkpoint method + +```python +model = denoiser_conditional.Denoiser.load_from_checkpoint(checkpoint_path) +``` + +The loaded model retains its conditional structure with both the conditioning module and architecture module properly initialized. + +### 3.2 Memory-Aware Sampling Modifications + +#### Changes to ABOBA and BAOAB Samplers + +The standard ABOBA and BAOAB sampling algorithms must be modified to handle historical states. The key changes involve: + +1. **State Representation**: Instead of just current positions `y`, we now track `(y, y_hist)` where `y_hist` is a list of historical states +2. **Score Function Interface**: The score function now takes both current and historical states as input +3. **Memory Updates**: Historical states must be updated periodically during sampling to maintain temporal consistency + +#### Modified BAOAB with Memory + +The memory-aware BAOAB sampler uses a two-loop structure to handle temporal conditioning: + +**Source:** [`src/jamun/sampling/mcmc/functional/_splitting.py`](src/jamun/sampling/mcmc/functional/_splitting.py), lines 255-327 + +```python +def baoab_memory( + y: torch.Tensor, # Current positions + y_hist: list, # Historical states list + score_fn: Callable, # Score function accepting (y, y_hist) + steps: int, + history_update_frequency=1, # Inner loop length + **kwargs +): + """BAOAB splitting scheme with two-loop structure for memory updates.""" + + # Initialize velocity and score processing + v = initialize_velocity(v_init=v_init, y=y, u=u) + score_fn_processed = create_score_fn(score_fn, inverse_temperature, score_fn_clip) + psi, orig_score = score_fn_processed(y, y_hist=y_hist) + + # OUTER LOOP: Iterate over memory updates + for i in range(1, steps): + + # INNER LOOP: Equilibrate to conditional density p(y_t | y_hist) + for j in range(1, history_update_frequency): + y_current = y.clone().detach() + + # Standard BAOAB steps with FIXED history + v = v + u * (delta / 2) * psi # B: velocity update + y = y + (delta / 2) * v # A: position update + R = torch.randn_like(y) + vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R # O: Ornstein-Uhlenbeck + y = y + (delta / 2) * vhat # A: position update + psi, orig_score = score_fn_processed(y, y_hist=y_hist) # B: score update + v = vhat + (delta / 2) * psi + + # MEMORY UPDATE: Shift history after equilibration + y_hist.pop(-1) # Remove oldest state + y_hist.insert(0, y_current) # Add equilibrated state to history +``` + +##### Two-Loop Structure and Conditional Equilibration + +**Outer Loop (Memory Updates):** +- Iterates `steps` times over the complete sampling process +- Each iteration updates the historical memory `y_hist` +- Represents the temporal progression of the molecular system + +**Inner Loop (Conditional Equilibration):** +- Runs for `history_update_frequency` steps with **fixed historical context** +- Equilibrates the current state `y` to the conditional density `p(y_t | y_hist)` +- The historical states `y_hist` remain constant during this equilibration + +**Conditional Density Equilibration:** +The inner loop is crucial because it allows the sampler to properly explore the conditional distribution given the current historical context. Without sufficient equilibration: +- The sampler might not fully explore `p(y_t | y_hist)` before updating history +- This could lead to poor mixing and biased samples +- The temporal correlations learned during training might not be properly respected + +**Key Parameters:** +- **`history_update_frequency`**: Controls the balance between: + - **Computational cost**: Higher values require more inner loop steps + - **Sampling quality**: Longer equilibration ensures better conditional sampling + - **Temporal accuracy**: More frequent updates maintain tighter temporal consistency + +### 3.3 Score Function Wrapper Modifications + +The score function wrapper has been modified to handle current and historical states differently: + +**Source:** [`src/jamun/utils/sampling_wrapper.py`](src/jamun/utils/sampling_wrapper.py), lines 132-140 + +```python +def score(self, y, y_hist, sigma): + """Score function that handles current and historical states.""" + graph = self.positions_to_graph(y, y_hist).to(self.device) + return self._model.score(graph, sigma) + +def positions_to_graph(self, y, y_hist): + """Convert positions and history to graph format.""" + graph = self.init_graphs.clone() + graph.pos = y + graph.hidden_state = y_hist # Assign historical states + return graph +``` + +**Key Changes:** +1. **Dual Input**: Score function now accepts both `y` (current) and `y_hist` (historical) positions +2. **Graph Construction**: `positions_to_graph` method creates data graphs with `hidden_state` populated from `y_hist` +3. **Model Interface**: The wrapped model's score method processes the complete graph with temporal information + +### 3.4 ModelSamplingWrapperMemory + +The sampling wrapper provides the interface between the memory-aware samplers and the conditional model: + +**Source:** [`src/jamun/utils/sampling_wrapper.py`](src/jamun/utils/sampling_wrapper.py), lines 95-127 + +```python +class ModelSamplingWrapperMemory: + """Wrapper for models that depend on a memory of states.""" + + def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True): + self._model = model + self.init_graphs = init_graphs + self.sigma = sigma + + # Mean center both positions and hidden states + if recenter_on_init: + self.init_graphs = mean_center(self.init_graphs) + if hasattr(self.init_graphs, 'hidden_state') and self.init_graphs.hidden_state: + for i in range(len(self.init_graphs.hidden_state)): + mean = scatter(self.init_graphs.hidden_state[i], self.init_graphs.batch, dim=0, reduce="mean") + self.init_graphs.hidden_state[i] = self.init_graphs.hidden_state[i] - mean[self.init_graphs.batch] + + def sample_initial_noisy_positions(self) -> torch.Tensor: + """Sample initial noisy current positions.""" + pos = self.init_graphs.pos + return pos + torch.randn_like(pos) * self.sigma + + def sample_initial_noisy_history(self) -> list: + """Sample initial noisy historical states.""" + noisy_history = [] + for hidden_state in self.init_graphs.hidden_state: + noisy_history.append(hidden_state + torch.randn_like(hidden_state) * self.sigma) + return noisy_history +``` + +**Key Features:** +- **Dual Initialization**: Separately initializes current positions and historical states with noise +- **Mean Centering**: Applies mean centering to both current and historical states +- **Memory Management**: Handles the `hidden_state` list structure properly + +### 3.5 SamplerMemory Module + +The `SamplerMemory` class wraps the memory-aware sampling functionality: + +**Source:** [`src/jamun/sampling/_sampler.py`](src/jamun/sampling/_sampler.py), lines 101-130 + +```python +class SamplerMemory(Sampler): + """A sampler for molecular dynamics simulations that uses memory.""" + + def sample( + self, + model, + batch_sampler, + num_batches: int, + init_graphs: torch_geometric.data.Data, + continue_chain: bool = False, + ): + # Setup model and device + self.fabric.launch() + self.fabric.setup(model) + model.eval() + + # Create memory-aware wrapper + model_wrapped = utils.ModelSamplingWrapperMemory( + model=model, + init_graphs=init_graphs, + sigma=batch_sampler.sigma, + ) - # Control trainability - if not trainable: - for param in self.denoiser.parameters(): - param.requires_grad = False - - def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cutoff): - # Replicates xhat and xhat_normalized logic from the denoiser - # Uses denoiser's own normalization parameters - # Returns properly normalized denoised positions + # Initialize with memory + y_init = model_wrapped.sample_initial_noisy_positions() + y_hist_init = model_wrapped.sample_initial_noisy_history() ``` -### Loading Process +**Responsibilities:** +- **Model Wrapping**: Creates `ModelSamplingWrapperMemory` instance +- **Memory Initialization**: Sets up both current and historical initial states +- **Sampling Coordination**: Manages the overall sampling process with memory -The `return_wrapped_denoiser` function: -1. Loads the full pretrained PyTorch Lightning model from checkpoint -2. Wraps it with `DenoiserWrapper` -3. Sets trainability based on the `trainable` parameter -4. Returns the wrapped model ready for use as a spatial/temporal module +### 3.6 Memory Loop Mechanics -### Advantages +The memory loop in KALA-JAMUN sampling works as follows: -- **No architecture matching required**: Works with any pretrained denoiser -- **Simple configuration**: Just specify wandb run path and trainability -- **Authentic denoiser behavior**: Replicates exact `xhat` and `xhat_normalized` logic -- **Automatic normalization**: Uses the denoiser's own training parameters -- **Flexible**: Mix pretrained and from-scratch modules easily +1. **Initialization**: + - Current positions: `y_init` (noisy version of initial state) + - Historical states: `y_hist_init` (noisy versions of `hidden_state`) -### Normalization Handling +2. **Sampling Step**: + - Compute score using both current and historical states + - Update current positions using BAOAB/ABOBA dynamics + - Periodically update historical memory -The wrapper automatically uses the denoiser's own normalization parameters: +3. **Memory Update**: + - Shift historical states: `y_hist = [y_current, y_hist[:-1]]` + - Maintain constant memory length matching training configuration -- **`normalization_type`**: The type used during training ("JAMUN", "EDM", or None) -- **`average_squared_distance`**: For JAMUN normalization -- **`sigma_data`**: For EDM normalization -- **`mean_center`**: Whether to apply mean centering +4. **Temporal Consistency**: + - Memory update frequency controls temporal coherence + - Balance between computational cost and temporal accuracy -This ensures the pretrained denoiser behaves exactly as it was trained, without any external rescaling parameters needed. +This design ensures that the sampling process respects the temporal dependencies learned during training while maintaining computational efficiency through configurable memory update frequencies. -## Utility Commands (Optional) +--- -### Inspect Checkpoint (if needed) +## Chapter 4: Experiments +### Overview + +This chapter describes the experimental setup for KALA-JAMUN, covering the enhanced dataset generation, training procedures, and sampling protocols. The experiments focus on alanine dipeptide (ALA_ALA) systems with enhanced sampling data to evaluate temporal conditioning performance. + +### 4.1 Enhanced Sampled Data + +#### Dataset Overview + +KALA-JAMUN experiments utilize enhanced sampling data from two main series: + +1. **ALA_ALA_enhanced**: 5 swarms of 50 frames each from 184 different grid points +2. **ALA_ALA_enhanced_long**: 2 swarms of 100,000 frames each from 184 different grid points + +Both datasets consist of swarms sampled at **20 fs intervals**, providing high temporal resolution for learning molecular dynamics. + +#### Data Source and Organization + +**Source Location**: `/data/bucket/vanib/ALA_ALA/swarms/swarm_results` + +The raw swarm data has been reorganized using the script: `reorganize_swarm_data.py` which sorts the trajectories into training and validation buckets according to different splitting strategies. + +#### Data Splitting Strategies + +The reorganization script implements four distinct splitting strategies: + +##### 1. Grid Split (`grid_split`) +- **Training Set**: 172 randomly selected grid codes, all trajectories (001-005) +- **Validation Set**: Remaining 12 grid codes, all trajectories +- **Principle**: Complete separation by spatial location in conformational space +- **Use Case**: Tests generalization to unseen regions of conformational space +- **Output**: `/data2/sules/ALA_ALA_enhanced_full_swarm` + +##### 2. Trajectory Split (`trajectory_split`) +- **Training Set**: All grid points, trajectories 001-004 +- **Validation Set**: All grid points, trajectory 005 +- **Principle**: Ensures both train/val cover all conformational regions +- **Use Case**: Tests temporal generalization within known conformational regions +- **Output**: `/data2/sules/ALA_ALA_enhanced_full_grid` + +##### 3. Long Grid Split (`long_grid_split`) +- **Training Set**: 172 grid codes, 2000ps trajectories (001, 003) +- **Validation Set**: Remaining 12 grid codes, 2000ps trajectories +- **Principle**: Grid-based splitting with extended trajectories +- **Use Case**: Tests generalization with longer temporal context +- **Output**: `/data2/sules/ALA_ALA_enhanced_long` + +##### 4. State Split (`state_split`) +- **Criterion**: Conformational state based on phi/psi angles of first residue +- **Training Set**: Trajectories outside phi āˆˆ (0,100Ā°), psi āˆˆ (-50,100Ā°) +- **Validation Set**: Trajectories with first residue in specified phi/psi range +- **Principle**: Complete withholding of specific conformational states +- **Use Case**: Tests ability to generate unseen metastable conformations +- **Output**: `/data2/sules/ALA_ALA_enhanced_long_state_split` + +**Script Usage:** ```bash -# Basic inspection to verify model can be loaded -python -c " -from jamun.utils.pretrained_wrapper import return_wrapped_denoiser -model = return_wrapped_denoiser(wandb_run_path='username/project/run_id') -print('āœ“ Model loaded successfully') -print(f'Model type: {type(model.denoiser)}') -" +python reorganize_swarm_data.py SPLITTING_STRATEGY ``` -## Module Paths +**Available Strategies:** +- `grid_split`: For standard enhanced data with grid-based splitting +- `trajectory_split`: For full grid coverage with trajectory-based splitting +- `long_grid_split`: For long trajectories with grid-based splitting +- `state_split`: For conformational state-based splitting -Common module paths for extracting from loaded models: +#### Using 2000ps Trajectories -| Module | Typical Path | -|--------|-------------| -| Spatial module from spatiotemporal model | `conditioner.spatiotemporal_model.spatial_module` | -| Temporal module from spatiotemporal model | `conditioner.spatiotemporal_model.temporal_module` | -| Entire spatiotemporal model | `conditioner.spatiotemporal_model` | -| Architecture from denoiser models | `arch` | -| Conditioner from denoiser models | `conditioner` | +For experiments requiring the 2000ps trajectory data (strategies `long_grid_split` and `state_split`), the script automatically uses the longer trajectories. However, if you need to modify this behavior, update the following locations in `reorganize_swarm_data.py`: -## Troubleshooting +**Source:** [`scratch/reorganize_swarm_data.py`](scratch/reorganize_swarm_data.py), lines 468 and 480 -### Model Loading Issues +```python +# Line 468: In reorganize_with_long_grid_split function +copy_files_for_grid_split( + SOURCE_DIR, + os.path.join(target_dir, 'train'), + train_codes, + trajectory_codes, + SINGLE_PDB_FILE, + 'TRAIN', + use_2000ps=True # Set to True for 2000ps trajectories +) + +# Line 480: In the same function for validation split +copy_files_for_grid_split( + SOURCE_DIR, + os.path.join(target_dir, 'val'), + val_codes, + trajectory_codes, + SINGLE_PDB_FILE, + 'VAL', + use_2000ps=True # Set to True for 2000ps trajectories +) +``` -If the checkpoint cannot be loaded as a complete model: -- Check that the checkpoint is a valid PyTorch Lightning checkpoint -- Ensure the model class can be auto-detected or specify `model_class` explicitly -- Verify that all required dependencies are available +**Function Parameter:** [`scratch/reorganize_swarm_data.py`](scratch/reorganize_swarm_data.py), line 151 -### Module Path Errors +```python +def copy_files_for_grid_split( + source_dir: str, + target_dir: str, + grid_codes: List[str], + trajectory_codes: List[str], + single_pdb_file: str, + split_name: str, + use_2000ps: bool = False # Set to True for 2000ps trajectories +): +``` -If a module path cannot be found: -- Use the `inspect` command to see the actual model structure -- Check that the path uses dot notation (e.g., `module.submodule`) -- Verify the checkpoint contains the expected model architecture +### 4.2 Training -### Learning Rate Issues +#### Training Configuration -- Use standard learning rates - freezing/unfreezing is now the main control mechanism -- Use gradient clipping when fine-tuning: `trainer.gradient_clip_val: 1.0` -- Monitor training closely in the first few epochs +Once data selection is completed, models can be trained using the `train_enhanced_*` configuration series: -### Memory Issues +**Command Syntax:** +```bash +jamun_train --config-dir=configs experiment={experimental_config_name} +``` -- Reduce batch size when using complex spatiotemporal models -- Consider freezing larger modules if GPU memory is limited +#### Conditional Denoiser Configuration -## Example Workflows +KALA-JAMUN uses a specialized model configuration for conditional denoisers: -### Workflow 1: Progressive Training +**Configuration Name**: `denoiser_conditional` -1. Train spatial module alone -2. Freeze spatial, train temporal module -3. Fine-tune both together with low learning rates +This configuration differs from the standard denoiser in several key aspects: +- **Conditioning Module**: Includes spatiotemporal conditioning components +- **Architecture Module**: Uses E3ConvConditional variants +- **Training Parameters**: Optimized for temporal conditioning tasks +- **Memory Management**: Handles hidden_state processing during training -### Workflow 2: Transfer Learning +#### Parameter Tuning -1. Use spatial module pretrained on molecular dynamics -2. Train temporal module for your specific task -3. Fine-tune both on your target dataset +All training parameters are preset in the experimental configurations but can be tuned through the Hydra configuration system: -### Workflow 3: Architecture Search +**Key Configuration Sections:** +- `model/denoiser_conditional`: Model architecture and conditioning parameters +- `data`: Dataset loading and preprocessing settings +- `training`: Optimization parameters (learning rate, batch size, etc.) +- `callbacks`: Training monitoring and checkpointing -1. Extract and test different pretrained modules -2. Mix and match spatial/temporal components -3. Find optimal combination for your task +**Example Parameter Modifications:** +```yaml +# In configs/model/denoiser_conditional/spatiotemporal.yaml +conditioning_module: + total_lag_time: 5 + lag_subsample_rate: 10 + +architecture_module: + hidden_dim: 128 + num_layers: 4 +``` -## Configuration Templates +#### Training Process -The following configuration templates are available: +1. **Data Loading**: Enhanced datasets are loaded with proper temporal subsampling +2. **Model Initialization**: Conditional denoiser with spatiotemporal components +3. **Training Loop**: Includes hidden_state processing and temporal loss computation +4. **Validation**: Tests on held-out swarms/states based on splitting strategy +5. **Checkpointing**: Regular model saves for sampling and analysis -- `spatiotemporal_pretrained_advanced.yaml`: Full configuration with all options -- `spatiotemporal_pretrained_spatial_only.yaml`: Load spatial only, train temporal -- `spatiotemporal_pretrained_finetune.yaml`: Fine-tune both with different learning rates -- `spatiotemporal.yaml`: Standard configuration without pretrained loading +### 4.3 Sampling -Choose the template that best matches your use case and customize the pretrained paths. +#### Sampling Configuration -## New Approach: Complete Model Loading +Once training is completed, sampling requires the memory-aware configuration: -This implementation uses a **complete model loading** approach rather than state dict matching: +**Critical Configuration**: Set `config="sample_memory"` in the sampling script to enable memory-aware sampling with historical state management. -### Benefits: -- **No Architecture Matching**: The exact architecture from the checkpoint is loaded directly -- **Eliminates Parameter Mismatches**: No missing/unexpected keys issues -- **Preserves Model Structure**: The complete model hierarchy is maintained -- **Automatic Class Detection**: Model classes are auto-detected from checkpoints -- **Simpler Configuration**: No need to specify complex parameter prefixes or strict loading +#### Sampling Command -### How It Works: -1. **Simple Logic**: If `checkpoint_path` or `wandb_run_path` is provided ā†’ load pretrained module, otherwise use config -2. **Load Complete Model**: `Model.load_from_checkpoint()` loads the entire saved model -3. **Extract Module**: Use `module_path` to extract the specific module (defaults to `"arch"`) -4. **Replace & Configure**: Module replaces the config version, with `trainable` flag applied +**Standard Syntax:** +```bash +jamun_sample --config-dir=configs experiment={experimental_config_name} +``` -This approach is much more robust and eliminates the common issues with architecture mismatches that plague state dict loading approaches. +**Example:** +```bash +jamun_sample --config-dir=configs experiment=train_enhanced_full_grid +``` -## Simplified Configuration +#### Memory-Aware Sampling Setup -The configuration has been simplified to follow a clear principle: +The `sample_memory` configuration automatically handles: -**Rule: If you provide a checkpoint path ā†’ load from checkpoint, otherwise use the config architecture** +1. **Model Loading**: Loads conditional denoiser from checkpoint +2. **Memory Initialization**: Sets up initial historical states from validation data +3. **Sampler Selection**: Uses `SamplerMemory` with `baoab_memory` algorithm +4. **Wrapper Configuration**: Employs `ModelSamplingWrapperMemory` for proper interface -### Simple Parameters: -- **`wandb_run_path`** or **`checkpoint_path`**: Specify where to load from (if neither provided, uses config) -- **`trainable`**: `true` = trainable, `false` = freeze (default: `true`) -- **`module_path`**: What to extract from checkpoint (default: `"arch"`) +#### Sampling Parameters -### No More: -- āŒ Complex `freeze` vs `trainable` logic -- āŒ `strict` loading parameters -- āŒ `model_class` specifications -- āŒ Complex module prefix matching -- āŒ Architecture compatibility checking -- āŒ Learning rate multipliers and parameter groups +Key parameters in the memory sampling configuration: -### Examples: ```yaml -# Load pretrained spatial, freeze it -spatial_module: - wandb_run_path: "user/project/run_id" - trainable: false +sampler: + _target_: jamun.sampling.SamplerMemory -# Load pretrained temporal, fine-tune it -temporal_module: - checkpoint_path: "/path/to/model.ckpt" - trainable: true +batch_sampler: + _target_: jamun.sampling.mcmc.splitting.BAOABMemory + history_update_frequency: 10 # Inner loop equilibration steps + steps: 10000 # Outer loop iterations -# No checkpoint = use config architecture -spatial_module: - trainable: true # Just normal training -``` \ No newline at end of file +model_wrapper: + _target_: jamun.utils.ModelSamplingWrapperMemory + sigma: 0.1 + recenter_on_init: true +``` + +This experimental framework enables comprehensive evaluation of KALA-JAMUN's temporal conditioning capabilities across different data splitting strategies and conformational scenarios. \ No newline at end of file diff --git a/graph_type_comparison_3d_histogram.csv b/graph_type_comparison_3d_histogram.csv new file mode 100644 index 0000000..fe32ff9 --- /dev/null +++ b/graph_type_comparison_3d_histogram.csv @@ -0,0 +1,17 @@ +run_name,run_path,lag_subsample_rate,total_lag_time,validation_scaled_rmse,model_target,data_target +graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_2_enhanced_sampling_data,sule-shashank/jamun/gxomxwtq,1,2,0.27083016000688076,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_8_enhanced_sampling_data,sule-shashank/jamun/g1512wlf,4,8,0.3578928094357252,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_6_enhanced_sampling_data,sule-shashank/jamun/3dz5xmax,4,6,0.30962056666612625,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_8_enhanced_sampling_data,sule-shashank/jamun/9kb3i03b,3,8,0.3430485036224127,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_6_enhanced_sampling_data,sule-shashank/jamun/2wp82cfo,3,6,0.33505777828395367,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_2_enhanced_sampling_data,sule-shashank/jamun/6g919rvm,3,2,0.26015421841293573,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_4_enhanced_sampling_data,sule-shashank/jamun/deipiaj1,3,4,0.27313617803156376,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_2_enhanced_sampling_data,sule-shashank/jamun/gz30rece,4,2,0.26252674777060747,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_4_enhanced_sampling_data,sule-shashank/jamun/c2fvvmrw,1,4,0.2634944459423423,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_8_enhanced_sampling_data,sule-shashank/jamun/ecq1lzck,2,8,0.3460829760879278,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_4_enhanced_sampling_data,sule-shashank/jamun/h1nxa4ed,2,4,0.2966764625161886,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_4_enhanced_sampling_data,sule-shashank/jamun/y05s6wsy,4,4,0.28305633924901485,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_6_enhanced_sampling_data,sule-shashank/jamun/jmaz97j4,2,6,0.2979964315891266,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_2_enhanced_sampling_data,sule-shashank/jamun/6bv6r9ct,2,2,0.26125847920775414,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_6_enhanced_sampling_data,sule-shashank/jamun/rac4bws1,1,6,0.2733909171074629,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory +graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_8_enhanced_sampling_data,sule-shashank/jamun/tjwcsf4g,1,8,0.3070009406656027,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory diff --git a/graph_type_comparison_3d_histogram.png b/graph_type_comparison_3d_histogram.png new file mode 100644 index 0000000000000000000000000000000000000000..3042c5d853f398a46061fa1472398caea59b771b GIT binary patch literal 646402 zcmeFZcRZK<`!}v#Dv8iSWY36(j55m1-YY~_8IeuNR`zNjD5r0+!M}tYq_iAVZOj~;jO!szTGU1TEjXLD;lmJe|?habuaw0&`#raH8jG>B1F z^-wrVgIWL3Oec4smwWMsH0_SKs= zu~tVsJUk8`KK%CWTawS8Kffw0lp1b~R<82&5U~D2{_)9K|NHmvx93Is1yZsak9>-! z80Ix7zxYz8P_2|FkBLH4Q*+T)M$+`t)VH#O)YQ+XPLV!1c9u-g>a&{DVWyLp$Zcna zq$c`GDBs-rChqK9EZ=nG@Zo1Nku2st`OG?nmcBgtr5Wpb3aP)(*z`Y0O=T*HY3uIx zfAr``^T&@y78Ze*A0AD@*D+Hh`up!Hb@;9FQ@CneFeX z>h0@G175N(SvfdJzrL}ldvyB7pFe{VRBd)u~cCEZw% zmns)FNxhXg#u>UIKY!4y6}|YluBGKb`qlUdt0OmW-i(Wjds|(-_1CXo8Tw^sj?>a6 zrKG5eEB{x>D&GII=ql&2T@ItITP!RrNV{idhvnAiTEvz|61`eeWhmEY>gY$mBGGmRD5F;3>9}96x@%wam$8eZE7nwYyu>u~_ox2>w7rMm*u+!~MMl7Wer~ zKD_<-Q6cH)MXAZY64l{7kBoV>GF7f$XZqEW7<%$b$ceLOHK&G8GBT1MJa`XBXe-Hs z2M=1)Aug!>$g^3tqixc$_`t>P37#}(7f`H0>{{!?xUu~+m)ihq4DG;58eelN* z4iX1k?UR)JDJdy@j$>+1Mcg#f7IyB~!NA708=H?76MmfdzQ3A}JUxGVr-p`xR+06Q zU(JuU-Ml$TVfZF7k)BS}JtM!ycYWP`qUZG$X2-yu3b(UnEeYICEvZtt`fqz)-%x&Y zOCLWo{Iay7qJMfitFVYj9*(JeybzUDcaCp(czCJA$#%A?jm0N5O?-uxUH|pA@CUMg zr%s|qgp=;}zRRs!bZvgLt-hf_+38%eii*k~oT3zi3RkTP*TUxauPF~6RPLdnrQJ_M zLlezvLJ@As#Kd&1zuX0<>tu3rvaD%*)GAp#%f{X*jj6D)!(3*A9XJTjzk7X zcbHzk{vG$@%4@Tawwr%F9Q`H!tscVxzZ`g_@ZZBRF)`FQ%&jJdj0zSbA=~r6i;5iKJB;f2;;)s z8=HZl$**q$9zK-9`PsU4D~^ov_4r4R9%+@>Gl$S#@H43kNxC!H%Mloj-4{@`wB#st z9QO+gBX4WrRLi(4>aqGvF;UdKub9uWD{F_Us_N+Y_(=waEqnItaYiwDr!?H1Yp7{q z^Fwmp3%BDw7HROq!(+^%YLopHZh4*?uF>;m+3CXqyf!yz=0;j#S6wz%7x>(l zZ3`^BNUvVK`YJ!)8%2iy^y!22^r2tAXbHP7lZ%Rq7JICnS>M=@W4gL`Z!qHzCPo!C zY3bdiPSYVz&e~KrG{}4##0SDJ<@0aNlRB(E?!en@c`~X zhF%H3p1ywGwRcHfz3UZk|!Iu3KdN?n|CY+jE6uX+NynQ}@Jh>#}t7+u7OK zIyu$-?nt+rA6347{d&NsrO~#N#1m`g=-qq$`gMDrr|kAiN=~~aCI6s>IrpGU1T8*T z#Jzub-114(+~k{qA^d%cVKkIw6rw!W1xvMTEoBF7U0vNTy}d-WBp&vz6R(e6vF(;C zA;;h4nbd_ndZe8^Lm9mTb-Si1mN(5a0Xtq(BU$b|v(>=B;7g$ujk|}3)QHFWG9T93 zzaw4og}?{gKWbMO&NFAIM_Ll=?rzx@m30;!4n>L0UOy-xD5zS~(1UqZ`&+4Fp3(cA z^V0){^S7~~agQG-qavGkWl`bu?EJ28+%fpxKhJB^Gn(bb<0ns=zTG;RJZpn87l%$f zdc*s~33m29e0+Sxw@1i`BlP-4OHjWJ(H#>-J%}=|rL7%p+xhR2R_nK`w?{ivDZw{q z#KYmdzQth;mj4j-+-+%!*y5vStl4I?Jv~J$H~+e+`(%Zow;H1i6CafEiXZcX^P>}($X(wPP|3dJ&b}!R3DR4aa{ZJ{oR+> zHx5+4f1hXe@u*skp0+vzWwabxKG9dd7Tx4?n$iS{n0F0rR`)uY>cyJO0efT*BH*jT|^|psu&jO76Vuzo3RA z)cNcK+&;BWt*zu}qO#p4R#w3kZcE}oJq$2g(k*%q(?>MJM_T8!~F(Jn> zCLu+4cVWN%$5;wk6$rplc4xZ(;nSzb17}ffj;34*)hP9}va)iY`*_xQWtJJBC@Lp( zRF&WQ%PGA|_am1tUrtWU)qZ_BZTz;QWAx~b6+EQ0w6s!|LEc~^t9;xuwfG}Pj+}EC z4U;S>EEL36KaUvfD?@oWedE)2e|k@)QXWP|b)vTbZ}_7f276Yz%$Z3_Nk!YM{r}s4 ziXoDyERRu>T$iS}l@p?)eRbd5%4A@)6SVAn8_lVC3WcDpB~jSL8z?CDpTDWrelXtQ6DQd{k3mA4tFq!rs-Km9q^F!ZJFo1w9;7qUy_Qj)1M{QUgiM@Cqc`(NKs z%oLeDsQi0vVDm5Q<3n+gk@v^Db6+^<)3aXR+PGiWfTF@=>)qJcSo4+d@9szEJP8m` zbZ`*(lpqo;_IG(3{yFKnHhGqKh68lMK)X4m?>QY69t+%i^5n_rjaXhoo?&yWIDTc6 zx2Zdk5F1gOX;4vWzJjVLdp=_5_xQNU($qH;^*l+N3q3tO%fHhTUpbCz2udyeCf&ck zHd8hI`}lb9l~6hc9v(6RS2s6H`jS#zv$o#H$;tWU>(}>yz8@b8N&xG8A05@YZgbtp z$mk0y?z@4?%1VkkCRWxa;p1i<&lUKty(KAmGhG`@bIxU!=~jPv;M=zs@kl4Qx%UeR z37rxYl)^V^E4DK!E-f`~eRk;GhObIMC73(mG6hIf$O={p5*m z(zH`gzUiN(C0>Ee4CRzlmYr)YXFwqG!-oUFAAbJ)`L?~Cf{cvJ*3q#R2vIrxD($1Orj{0eK;sa4F*0C& zVXwagD>)?~;1eHDOZ4+!zx*!U-)ZMOOmXPw(IDa;eSWFCd*8kqoPqF|n4Oo%uUT3K z;rUbSeta11D>;oucFoi@t$pDJFauFVW@o)~b9t+)tHUB9yz$(lV`J6z^}AVESg;H` zCnhJ`UKvRg-26%nR^S~NC=>2z(_elB^hN5^YqzCaedR8y&$VLF;>OJyxP^pt3f6xx z1Pivk_dj5Ve%@E+bf6_cg!7!!&ykV$ot=jOL^2GkMDgrxNfNspCVQ!Xz)5p?#l+~b z(SPRWYp_AYnrP-45QGZA`xHOFcXTvWc6K(7#%`=4HddBaMXjT+!i^@DN8dX(mL@^O z&2MznJT*OCLQ|85nwlEy=Sna&H_^25DR-d^GE-c^xnGpcs%lARuM% zCjqO^`!?2=u2gw$Jj=~BWaKkv1cEQH>L#0-nxaVDv13P)VU;I`u4cTDV=KCHErntjXE)qp1-!-C+4=k5jWwJfNn6`fLqC2HdguMZ7biXd7*_Z89>r4Be2TxIoO1c0@%Wzo z`+dvGF08Mw6Z_fPs`MX6#w{fP8d`wm=j_0%@j~xvYj@yuadgr3zs}FM9qVKVoTm<@ z6A{G!Frkh)O%A7HZeAVy$3AcH|z zZKE%*jL=A20&i-fjPn>)?sIZ->Z|giPfAKcPr7Jk#G6>nh~17qp>G~Rx#o?{ehbZLUHz1-`gK`p#&vNO-(&< z;leTWzYLW$Qi71>nOrF@DM`X}*yl8dO+6*KXxv+Dm(uJAY$sk20)7@H-HaYjuAd055*8Py#W>E@ z=(C4X@`uaJkZ)r*a}ga_e%d&&Y3v1REK(8*o;Pl(d3>$HnKNgq9~?RD13)AD(lFdL zO=0#g9j-q+#0LEwM?m+zkt_--zc)n%1=Zt(UGn<|ZUfi(0x;5a{umqM;A;VAF%3y6 z9#AkgPA1yryZZW!`eHOvCP&)5yu50FV;`{O-MFCko1qKy^FDw#doEqNMC#tLd8OB8 zS^_&({r28A)72_moIQZqGYvr|LLpkq)HcxK5{ z49swFVZ2+`MpIt?!6}mu``%&ul72kR$;oMbp~-86FIDMQ2Cs~!)0JzFzu+X=_H zp`V+apP%o8Q@0xxnRH{?b#a27=<4;KKE+sgVE?MyQm*&{*X~8FA>G(4y)((7SK+Gu z`=;jN@sqzcmb6j6ONttx6=}UTdk6@gIe({uhMu0+eWsSiJ0T%Ko)6NH`j?`cx=BJc zA-c?tFZGV+{kVwU6`eH)dPuXu6T_v==%nBN={>QY+b@w9-TcZj)6;F}#l7U3qhH3N z-15rkJ*g;7@TJe2+PMZ5lChe@(<;i$Gn<;R68>a?o-Jo>`qg@bAQnh<+kJm~&p1{= zL4k^z*MJK|hSTA}XLKBBCT!My#me;`KdSefX-F>ky4((dN>;1Nv!WGGcL|_zYPz+) zUOMiAQx51+QhV{#Ko##fA)!X9;E<4zSJxXT7+6`akZ=DzU?zuEyxsKI)aMsqy9nz)Nc=I9rteUSN6lQfu&-V z)AobaJCYzG)PDFmKF(Ph0*d~wqeFFhD45shsGudA!vpk?%2yt1cOK=^rm;^6@ra2T ze#titzINjVtEwKTEj5o`vC>Z=RB&HtMUCz^1px{f`1w^bC)d9Kw)g-)GAUfbwI^B4pawSPG-VO$pzntL8KkF#Ux_WH72 z=6FM6BgN+&eNMF(8VZBc1uk>Hs)53NK`eaV#1e2x_5daSKX7kjZMrJ?Y;p*l2*+*O zw2fCRe0<92;fM4`u#2+SQ2e=0m7CyTAlw*%KCWg zG~GJdgUqjKwa3_sBy@AqIwKR4lg@5#d%czN_dIf%?x*6=%;8d%Z>QitKr6(AMxOQ4 z>t#yH$#rl=Qm^S+=IanE^&l_CT*!e6uxD^^kYc>)(xIppdxvC;X)#*pFk>DPU% z|2)73S&czzyog&-<01`^Yns71b`FNqIQ?UCyNx3Ye41Gl9%op)JvnoequhN(WpI*v zOo#_;YxG;$n?62(nLSm1S6_^-q8YHOWvWO`32|c2AiZ#w$J6&#dX%-_%mtPqMQJ4U zs%wbpgl4j|*KZEn& zjwNphjuaqk`1>IhfsIq2VQ&IsoJL!dqE3d#`}*!cBjsA{!PSsG8>c9L^(q-EsUPGN z&2LFj0jQA#b{_xyl6?E6L052%Q&yk3*4njZBH#jU1w)bR{Jv9D(e-ya)DDbG`l95E;EH7IfZWl$=lv=jR_r@o#ZIX!)Dz7W+yi`eGdt}d!ATeiq{%}w;2Cg=eTuYp=|$g#7x zh$Y;8;DDz23IJ*{&KjZSo%72l$V7qOX9l@pBQvv|XV1P7H$^utc3sG7b_5)i+m{B; z9E|q#g6XnFT`oS%i&Rh7E4}1%;QiGQv!ipwSXpOX^I@gmr{ZHxjg>FOH?zF~94>$=>-gns8*hF{jDZlUr^@Em% zM=#7x`x5ynn||Sm{)#fC6_nCeP)y0ommBqO!0jN!JT#U1iU~K^5pid39%WHX$aeTT z4%GsH)G@H%M)?CC7^u|S&~rs9vq%*>jOX8#8~cu*Q_3_Ms3W9ZDo%|PsMv!-5aZe) zG^c&tPXHOU#nG$_geRGt3=8JR^2`s4Vd&3yU%q^KlriEPBq&Mx5dA7o{rzq3LIfuU zRFT=}Yim0Q%@5T>2v!Qi?yZ0a;bCES(-jjLPn@{OeBg>4VY8r)-UncpiBE@0M)(s1 zQiQsJpZoUx`)zG)ZOL;sUyJqwh?Abnh|Xt&r2y0}k~p!e)THg~L+^ zoH}gaA0B=XT@06gmzI_m@uyfy>8n>SLyIIjtKdCzfGAsg`)V`-s@}@bmF4AY>&r8~ zbcfBYtWxyL?$|mw_`H0{jf?qbedRWkA}Ilj4wxGQ?sNMIGEedrblwi1SbPpY;TZ}k zu{!WvAcr2njxR4S2Z%-)6k1Xdx+Z#HwDm~(PjtG`p4aJ}HD~no^#yGDkAt21;bY9t z&y$1G5;{D{LGorOw(l7TNF1MupFE$BpP%vU&juCx2=#2OgV0khf*)MFb&Cu}OzQLJ zjOWiE0fKXew1{e9Ixxzg3M>ms7a=_#6cF$hbd~S+FL^BqZ;+xWmq_u0sF<6muZM@JRN9#axCS z@H2>+iD~=&T?a1f=sX@B!HF1cPumNkW_0r=@z)5YyrbiFtIH4?KIqjwy>Xu4ROT2$ zr>tLPk&nCUv9?IX@wMbOtLywIL&q$ZF6NwK=#kSqOG~p#;&pQkMA~N2flMa8`@u>?0lzBO8}ZH!M=!l^5i`h60!^-sb0Hw0%t4~D5cF5 zu7$3^t-ixpmuR;NRITNS*DaDQUr}{O2dcy_htfTy=GJ)zp&sCn5c~=E!)e-}4-Dhl zU#yR;$kiws6d4t<$dSHlSRAvDkCT8bKyL2-tG|cG#qIX;@`C=ZB=GJu7%C(RNhl`- zi}5WQG--SVE~0^9ov|b;gE!Y(Q9PKZLLrJ%7c4w{_;BXe$77HxDB(R&y)=Oh zMC>ROZolu}uYsz`9H)nyO z`SRs$P0eYz-uLgr z@;b?9gFcPbWiI5_gW5Mb%rr{ol zaJ8SqpFH6tWFD10d!4=6t44^bok`gQ z-03TIl$(DI<@5Ek1bD%F2RxSIA$Mv{{7G3^S*iBjOXGv49{3>x%9Uf>Ig8Z&(PQ7g zD~+x_J!_Nj2V31H1lhoMWo{(+a|+0ZFhJos=Reny?BU1{LIxGTdNss6-qX`l90yQp zeB$eyT)nT$2?c{My7<(Rh1Kl(JYck4v$EpUQN$+u7qv!JAuC_sq9hb;~U3^P!k5C^q(}b?%z*ajeaNXtMQ9;S7 zT%7ZIeN8)wn7!U7fVXAz{0|>FvWK0Wo&RMD@CL<#nCpB-tFSR&{i;~m?U9hq9}aJGEHI~~zu!PwPA-!3P98(kj! zg3@N*j0BL}jY~`O&;o#sxD!c6yl|Nli=J z1N=^jo-Z~2rQ~*cb6x2KDZz-!M_~6B3djpDDDCTG4q*AE&b~(?n8V z7ZygZp7><~(}N0?nMJ2%aq_Ea|8q3uAQpYEbL>yTRtD#5X$++mYNBdUEq%cnNa5;| zcITNkQMYaW2sn81XVw)R#`?7egc2-`f7cQGT&FIIU3L3~3uWF_ZcuSL?pH&Deb>{Y z?O>?(^WALv(=hn_ifuwW$&Ryiyl-8)nkdTsIJ-YtD%d!b76;xN8f;BIVShP$xvtJ< z>W$b-wd_e72FK;L_j0haJ4g_FRmNlvXv9>)1#8Zh1CYMh8jmH)ISIfrK!m2KyyLRf=AVO)(vcdW}~Y4 z$yvo;LVs6A=-Y&VCz%laX;7<6lZ*9%EqQ2VjhlDRBYeN8(ueuwujcr)Dfo8`ZblS} zJ|%X+c-(7c)-=gSn?xMYhUu5;FRvK*9e-xS3#8#lB+%LK4`B1WRl>tjBs=_0hcg-jb7_C52gjaA@BQ{iYTW|X3c4Av zRhFMYb>cG5iOu_^xDT=$`i+lI6atF14UeEbEZt#dW!;PCiE*9NRp0Yyc5x@e<%H`` zk^>eZo}vF~wvOZl9EmADZ)V~ubY}S=%u4sjJ{8W9Mhmz=ay(G z{tGutGL~M*AwzuR9Y|K-=Ow$@*18(HDZpr&p3~`B?5 zZ@aPT=u3aADGOvW88L@8iNVxkE*WESzwwz-M&+#QU^)zEUFq>@=B->7pI;-OU^)crpW)<`dzF_LuGJ5%>(An13}*->>(v0d+yTX5hJWN1B6c;p=s4CX zlO$xSQ16a=StHED!&Cct6Yjn&*Bc}f0#~9o5~;d^r~D(45+WoFqv`$%iVCr4pV_!D zRDJf|4;yeQ59QnjBBfX$sud(LmDri3rC6=Y=!o!QA646B1hX(*&l=NX{cM#P9UZMF zlLJypz99lT!Enl%j9yglCts2<({-~OH{>SIYxnLxlC|tLGBUCk9vF?tr^b{FA9=|5 zC6siv8OCmu6cmzfJLAvaey+E0C_3!CdFzRi-YYOMco)l^AS{*)jSo~+Nzy`B=6{nk zhq?*K7ie&-v7^X~_re8T^N!~aB|V@8hL?myK;&8&iyqzU1%yR8G&J-x2k2Fva6B_? zW`=5!3{(?;4j^ee*(+FLLgC}YdnQQV#1KMt(kh`4WlXf~1nyU6-?ML@54vg1^=sEi zu*a|@D3AiF1JVkMjny>QhK6AcGa8`A7hm?c=%S~3mfAt?(0~B8Vdo( zWEU3JOU{N*W*Hz^h)jvy$S2x=0PPRh`(ZD5zzHXM3-|i@`B5a^#P@@cUi*0lx{GvF z&zCRX5%Y@Xq$W9a`t*CSo1X?(>2Q^5ZK^h&JBal5W?6`OtPICu0Thz*`}*{bpR>Kl zyuhnps+RmW8O+!O;8R7+Yx4;*hOs4|;4CEL>{1NFstGFzQG9e_w^pHLv}DQs``eIg zc-fkF=k47skT4!gt^jfe>s_5F?#)zXMGJ4Bm2gQ*PxlALzo@8~TffrY(P0N$ip!w< zFsy@9XU|^2iQ#jZy-vgdVG(F(%OglrY&+;J=>n$=TKg8jrA%x6b{zhmWU0M~(p^b* zhZD+?X-Eypj9O&rAwmZAKW->{uFqUKcm6yTM^kfi4bfHg7eQWNPDZmCmO8NO6j`%t z9>2eHe>Hx0Sm!wx81Xx3Lw0k&6xv#1xwKW1$8Em6VgM!7&uH%GphSX`&AtSfoN84U zUZ}~32g)w);DX{f{K4_l@Ux(AP@+=ahe`4lrX3L42S{eNXcG{uUI-V3`uh52mVcWC zHl+w%tAK@*a-$`IVi*RKTJhC9QzEhfQYfQ;d7^yk`}e!}K5WP9(&g1hN}RfF=~SfQsq`y*6rsOyDha$?}*i@TP82f#gv z$Ac-mH(JbN^$;=%N~NCPzfCu^wJHC0Rm-cu`Afbz1BT52O#6Ek=2igkbVHZuQDNs( zcr50HIY1*LAVT|EJ_uQX`Jm23%Pd%jM))22zVhEXdawGfA}^S%2MFQK{QUsDRurx1 zGEsG%vOC-`e@LSdYarxh$P6)Ml@M&MA)P@nj0!D*EMv9Z*6FEl1f@zcMRfASZt*03 zTyo!W{2#{TeE)xIej9c#5IWcnUc<_oRGG+5+K@h?gX<(jM5r`|yC0>XpkQj`;^yum z547AsoHThDz3ndOSKmSxG6n>qVpEisehc9xSnmuY<5oI4I*MVKV0R&~_T`bhg$Scm zXsPZS6I}~q=%sRl309D>;v`Fu`Dk~15*K&v@A`^8?^yus%maBSCcaSioAgJXUAoWI zhzul~lSpDRQ1p*~qGe(m^O5HKh1rDnWY-K%#t4#-#-nNw`XcMs*jNyjn(9f^N>X{b z@bcOk(tfmr2nf?tsl?tC+=LeGNrNK+&Y3j#oBVv0tXq!cax5CPe5x9o>vIQ?k=+}O z(mRM9z*&Nqi#w&245@ySzjiJUxR1kOeOJH8;asI8Z}BEe4rGiGh#_qF3wI{S<8MIp z97Sactm8*0IPi<;)#CLq|I_7EtD&r!KHbzy#@ zhZ2RAh*d&t@rKYHU8il}zOGAhmsK%=@V@Q4hjBt4mkZhr)iB2XaM3>vM*#S<$u;L| znNzO0{S&PxiW;B;D$d#pOQ>7{mFwHf-G0M0o?=q& zMTH1N2To<4rb06r89BN4(CC;O5s{XI2F9j0jfjR4NQ}5CJ|p5Vsnw2S$l{;^k@E81 zsT2ebTMM2bK!}>_zrz(r<#6Odj)<&q!;`tO6O)5QTovRI>+pP-rL8OJVReF8&&)4DZVqQ6D|@-#VRCpl(>LIWNvgc035PO z{|Z!*t544{C>s#6n&j{NX8@-Z6(MyWK2U#cdX^P1842bgiA9Oda;gb9NYNbxI_W(3 z>tUxUR5PMA2F^l*;&`g2U1C4X-0EiLY7vEw*(CH4z(L9Apm z+pN=iCJNOn^Vt-}it9n#CjXiSjIop#av=zxB(m6tr3<4h=mRS1+p^R-kx$SZbVf#% zXJJQ2{uo@Z!*B_J8d(}Y^lYN79Ek$b6zjX@gi>*p85 z^8SXI*$~(vMN{xX>T_9KqGdS6b|C%)bj~bA7a~%0L_!5Zc6tya7d=8Nl1p94DV>Dp z_{$Yg{uNk2;OxPZmk+I#QD4ZdtE)2sHsQ}rl=b5uIiGh%2W>w(2qylO-#;!{Ac#`X zDY-$kX~LK*w)=tlAtGSWz8zO8LZ=%zCE9a~tOm2Aeo7{uvHXsQ z!gD)3`ki@6XK1+YtYnudT3YgZ4tA53*SwDE3@S^jJKik)s!Q{`Rl82&FUUMbkh6)GBx8!xg*_ z`fLn!u-JY$sG8SAuga4aYgmn(n(0xEW%!55!^|S<&&!e2`!oMr6$1h%p!wqQDBgQ} zNl)CnV<9YGiTZ5rf)JRRidTi}0+XwhE2cIGuUDu_tNkHNW%JycS*0$%%c|&VL^u+f zu2H7`F?VwmS&G03AiOAPIzox8ak0lCR56@YDFovlguREU$G2!d1h(~cOGc0cQjGf+ z%*R+G4~T{ zGKHu_1h_i)9AOP|sMlJh4w)(<-ff!=pFVv+lb?aPMOaAx#sF2^%np?uV3cc_CGhiP zxCYhD0=Vq~ESr`Qz#)SOS<7~xzFtp`3n()RSa>FH2d5#B}DTwO~$H{X28Q|}o+ zihS?63l})m{j))s{y@<~89KCHi8jPpxO5|L(J%k&i?JeJk^=KK@j;>5-m3fpkJ=nT z=&0)UxjEv#q(o|uz#B>|ySIYlv_WP~8;_0sQWSIj!V>B;VGNHZ_~my81F8~Hmzq}# zsF;`0RS4-d?Xt!{)0H3po9hxpL=9Z=*J5H~LMc-v&Hh5O1H`9OVA_RDCR?Xnw{9-4iMTJHBAR1{{)sdD7I0MM zPS%KH2?FRnm$!(m|Jf!kA%VNR1r!0c;t3eZIW$-Qu8#8_K6-Qrr&^tZ<@W74 zmk{8)OdDp_1sv{MB;A;WNJSw z3mn3vaFLHz`Ear(zLn8nKu!GO#k=h%-66X(paNk6MP=lmRpOQ`i!%nf*;aL4rxIa>nk`Y*7tSJ!3p|OT27(UN2%_{7p*4!( zGNTv0o9)c9iA@gB-O_iYw+v|$n-iVJ9 z>fO68nk)$_5G7xyA0izdj$LG7@~dX@)R1dBkjleO|1yZYUn)H=;KGnV^dYzgcvlT7 zuN<>>PIiV;(pJR#f*?as!CR{=J}TaL7P*86^Bro&pzM=^)KJI+isO z--qZFmto~`@PIgeQ&NObvm>s{UtBSH&Q-J603SJeA{6+5LPq&w+ghrYVv@1QR(5gq z1@mM~p&)~H6!4ZepY?M@@{1S2nBqIcAtxilKEJgWWR2OC$H~D#Eo;{)b;8m{2;%|7 zitngC(h3T`2t<3X4AT&ii{}OrX9>ua|1DiMo3#nKk;bTuL^M{kn?|*C0Wg_}Ue0df zBH~gJ_Yzy4K3Gu_)8LW1o#@4t1FX3(F>-J)rKPIZb=seme$6-XbNMMppNHpDBGG$A#nO zGTm2dVZHJ!I8rZ#W#Q@0A)1(1VF+s*S?aySyay3;hT=vzyl@#r3@Y6-s{@++2@L{P z<1$qJnV$`m!mjhXLDsnJ&5aEW4UdCV)VtPFxOOfuN(0)$V|#@$rOXOHP6_n}9;vnh z50wew?5w#PG7DL09GC_3Ik7ug6?GbI&=y4)vK<{9Y3i56EXU!~W)Ea3Zv149IkjY@ z=#QHp6CjtsRMu{6ZXSTTNem|v^B2a(#;e>r>)eTxGOl{WsFYd#490 zCVO15u%O@+2+E^w8>9rF%29na(Lyi?@1G1SBG(1dE3VuoygBZEn`pYX~vN7gGB{8!E z97aRyt*hNwS^Rvg$ObV9-tGM?roW*pJS;QB%JO2~P4%uy zh@f|&H`mX%t0FFeohQbllW`$3bl>oR5AG*IQ#IlQ`AG}?Pw06XA1~!bw)HN=o-}06 z3{k?(OZX)yRA-2$LCA!JpzZPq-qTZI7Yf)(ei-2(&|p+l6vYFTGsb5_)fcYG_bU8K zzNHnsu-yV;L4>oKdW*kW+lzN@Fg-tj#$cx7dA?AfXiRvx^4NG(+M8kk-60~m0}Hx& z;TMK1>gsmE)Rvg+E#!mbgF_K=}euUB(qykY_6^FA^_U}63{_ELXG^CdWn6y;M`soqd$qg@{; zJUJr;O?GdzK1SAn=t-FTlYnVH071xci%K0icFY7aKal`JW{bD*Y(G-)Z3(nuQuymq zNcl50*7`BVnkIldKEop0*M6UvcmP>01tB|o?j@Xbv^#1_D$9Fb8Eqq& zIRadTwI?Q@CIuC#%M`Sz-Gm^e61lMB(nUl>45~c$1MfT)awLV3b_GO88NIWlTkyuG z;{YZ(k_OGHJJgFjz&nT;Y7DVyn>t3MqHGtgC#VTl`wMs4DePw}dYRQ1ON=T2k`ZDA zfGE`*peaH0FqL%~pWS7@6d|(gBt26`6&RtlM`je3;oq@ zG)A$*SvZ2#IDmUn5~Cm^^A6;s4Itv7dp%o=tz{NV%sgfbYvge>{MuD}}e*{~_}q+*vj7Szl8s8x-E+lx-@27wa5 z870)F#Rv_;>mGy*sN(+y1KFI?-y_s563>E763R~I`>z_ADmw`9f;46zOk!f{l$e%5 z=n?{+ROlm!;Ay4%r!8hhGcAXI_rX5%#h3g5j7qN1_t@N6g6yG@S+`&p8jHXNF*OPI zn*7iqwb9pfBCgMn?Mge6pi%(!3kE4dP!eOph~jhB1_(USnobAut3wY-<(WeW6T?-y z$?kB=No$EK|5ViDD7?;^8;Kq}B_t$DVwNy60E-I|W*x?g6IMdFn?zc6Y=mG>2+$FO zr$q7x3aq5NyD-jkcw!>2U4rWTj-3`VD_Q2c7WQk|1w^r~+WcEKdIMaIIA~xzcFTW^ zoLyWP@Gb~+0_2y;H@pxljh&^g9~sZBig8U4;NyD-KL4jfQEU)<{%2u93L!~|Tz?iP zb=$^^2}^F4I%9}>P4P@)`zv5axWV;F`nQg2x-XzA!sA5nz9 zk;JbqlF{KqCzGa)oR4`Z88I3OUu1YpkUZ?=74cPpz=h8ysEnpfF_{cU?T~1az|0Ir zN_K;^6ql9#z@nIY=0$w57p51N`76hbu7`Agp>s~srSz_Lq`BzuXV52GDoe8bu7w+tX z)fT%f-$5BB8wvQ0NCb0E9VVdI9esA^_46sn%bx;=Of&x7nzT*GVe~G-%BWu3x^wj3 z1DGOm#@P{8jl>FU7Uz3}K0o$1_0}e)FC|;rMBEl@API9V$;AaFjHCh{;|@C<=oy|V ze-=RawhEOSh}aQ*xh*yzIJsf+9iHX~T2%yFcUM;}oEARVlqHmn0xoV{GCGPciP$N$ zaXIS>$fC>|Zv@(l7J_*zFG9`4@6Ap(pSa5Jm6@tphgnz?`reqTqOt17FImuhR9{_VN#JmMb`LQ#Y ze;OygEok})!Shf!X&~@kl$H+AlX&(bNXtG|h?2inRJGONml`?80KI!Kn_ zWe+?CXA4X*FhVWjXu!(~-U!?aYicSNtfokgSgBt-$>DDP?ygT_{Wm+!`86fXXDL_%Wed zL6ecV7M5}m)N(mS;%05c+b&Sk-eF!%$wf(t+!IHX(XCju5cenvxj;f&A`WTfwsb{$pwe9=VF?K#VDm)HJ?Aua9Jq(gu?N7(2$`j244kV|1V48} zBpaQaRaq1U8mORVj*-QH>bx-P%gM#DE>5*v1N|hSfxXC}TdvFwp$4mt$gk3`K!CuLHQ>;X?6%(~l_EDz61^re*LVf`oeuz<;jt$FG z0TysSE_jVsN%%(cZCRi_TWd_Vf4{l}(=_pVhbUhkpN!O7>Rl-oM1LZZMNDrY3S2Vf zEv=W$9%%}7BRj@t?9=!Did6)xfB-V8MC574_6$9ix6YWs@h5OZavw6~NKaA>GxtPh z<>m1az!L?K4nwXOFxx?BBbg#UHWt6yps{l2ISVxS%XCdYdh&#*1}bbvD3#WQHIX`L z+fbmoDUjEnBPM0MISo<47&1C(H@5py-XyJoV0Zab)qc}RWL3r z^B$trgla+*cRV|_N5CDhUO~y7z}GN@@gySAPJ=h*P(E=(6TrA`6p6&-bYIEsH0n^H!1!?@Hnw!? z6TM1-2a%oaHB+lCG|?{q%e4pl1-L8$1!+)7uuvo}DvGJi>hvET2Gt{vODi56 zk$45eKW0h~asft`mO70G!-a`qsb1ZWfNVjuD!N?Zsy*8S(SieG++1Bb)N>1jAc;v* zruSQYD=n1N#tT5$$sfKkw`ILf%dLWp=wtX@(9Z&yFc^AClfPRby7`T zg^EG({=nE#p2|NE+mq!sq|P>|&?%rP!7GFIqw1$o`&U+q63OO;Ib;c*!QM-wR+Lfx zxxBzmAy+A4Hzb*qpxe>WAw|iHbp5lK!2!I$i9ly1AdeImCR%Rk8mr$fe%Fk?y?5;GB&TYy zYL4Se9&V?*T7JXWm{Iw;bxSx_fl$!c2Hw~tsw`ZVk{ZO(q8M(cQ0%wr%8E7*y@LLe zSe;i6fwz;w_YlKpPF>~qr8FbBiqW#G(;a)f-e?0I~4Nb zV#9IWrXaj3CpbnE^M_;npnH@GV`v3dnN0avt8sfsz}Yn{T}+9{#k_F&jTSmV$!*B> zs0eJ4>!CRoRguHC7o@`bCbUM6t#Ev_u!1pLte(1XV_| zy@`;Bz%sq7K-q!{3JS#M2r=!ONjyI$Bcu$EM-G=8&}FRZ(GbC^Gm+6+2Q0C-#0xO0 z9jAH=gJo1U6SFo4`uqDa(XPUhN^$b#6?@Yn&G>FR(@Rj_aZ(&#t{BJ^ysIe^#EV^s z$*ZjX*=Ecx0Bc6FZK{2GLr`whBma$6${bsjM`G9^L%@+R3_v&2SVH0E(Gd6cI2F6N zPO==V?(n9-1<%J(QK|Fyx$=yrpd(_=I>$C1?O4jdAVG2k06s|Xs-Yo0e6vV43=n2l z&k&ao?ky>|?$H`odlj?MK480yo1^+1$7(42KqB&PeG|gH5Hha&jP1C)oF~Y@aEY1E zq7;xY0JTvrJRyM&!)_ePbQBccU^7`B? zTUO(z{mRZr*Xbin1LoG8xas>p3YK-~x znKMPrhE7=#*cwQ5YAjDFm|xPENWD1=1qZJ(Qf5iDZ^48Wtg>*nO>_|=UN0y)e--N`&nXY}uBgoQwp)X5%T|sbn`PUPhpIMv`?7`H}gLg&+WeO_~ zZ?55HAN}^1>Z``ZnaK^9tAFsG0%pclf#w7e-DEBi5n4m{8QqxGFmT~BY60b5NTBCqv=We9z z$m7$Il7O6f3IJKDvCTt=c%m`BSqXZwgUbn+TY@yJ_tPG4iSR*(lPt-vxOyHWglrF9 z&Llw`^PE!X3w%iS;N>{d?q?wF!fv6~^+)j@eKm!*0^m)0iOmh{n9qq}-=g*n^PoJ+ zi@>ME3|x*Y408^6bN-@sqZD!byXM7SLrD!_tBr^qHchCAQ;B+^$v1Hz(R~8 z823*eoe?)c&Ww0{mn88fv0(iooG%K~$L&uHFJIn?mgEDOmY4+RcyShbxaF)Hn6IQi z)20%pF(uuEL+~;zL~+z+#iB1y#65z(Ab9Bo8~&^slu`X)jLKMdF7M-Yq>oM$(B=2w zv4JMMUk(;$aJ&s((E;ZpS#kwIYo!U%jXV^Kpv6kf1etN>+t`mpwF93KT)&4b5D^t5 zB6`{V#BAl5r!IBQTBi}dL2}&!0)K&I$8O>cNAUmH3wZD%iRk&A$Ec{7A3tMm!to)$ z(5+wYd=Lw7-xY}$+hCYfo>^uFF@U>BwMwplxIUowT33Efyatn)$pcyb|5$tTxSrQ8 z{1=;To-z-aqZ!GZkTO&@%}T{a6bdOcvXNwpG$<-X(V$30MP|`}W=e>XN}*CJao%_P zdCqyx@AW$W9k183A4A{a^ZDHOy4SkabzMula!>kzYrk)dyD&3^jiP{#`9>nl^0!`;Sobe?|J^ek%_Rr1LxCqdgro@vtM}i2r!u=!>s`N*obWNvp7x)b@n>HZGtlFk5;{57}sT+{ykliGwDUE+8G$dS_gV{I$R zs6{J;RC;OPB@{h;=pprH;_^Wf&h+>oGy=|T-wQ)Mo>mE7=ly~L!_c2ZBEiYvqr$hB zOr1Iv4(=e)_z?uvxp)5gr=u`w^Qb#^tXD%#@esUHl&&;^gK);hFP?PByYuoYDZLV& z>NRX8FpByhfVa8+pBF(s7mkh?JxI4WU=*i*dhVx^=9STDSDO$EjDWj-f4qip4HdTr z(Xrfj6}@wSIrPa48%)+@t7pe8j&?Z-VgZlq>Feu6oC(~Q{hUw{6T->J|0WWI1}hlH zb*9lNQRC2hnW59=?5%s-q!m6TN?Lqgkh=zJXB~bpA;{8!2lAEA)9Xa z^=v*;kE6uwBPW!#__CM&$0gfeZvv_o5xLVF>qT|+HN*oB1ZQN;* z5C4e;;I(vuV_VB~5otKS)sXVqqAuSS!$d-BExtL&Vd`{be7 z2tGZ!-uU+In=oZ$G5kXZMToH>P07Wj(f}5~qG&*J;`el1G{EUa<24FeilgCM3Znl3 zIa_>Uqz9Y-0@O4OMZZ(0c=UbNQq_mvKA;_-ye#~n7tO`>0R0kue~Il+^Kbw_;bp8F zL7y~6uZxT`@_#Sf`8Y9!SekbEKdeD)4ia9DQs}bqF5Ne{k9{!l)q6j*Su`#zT7o-% zIbnvtuq*d;!narSI0Q*de$~|udg0nKbtY2d$NvFXbf4a>V8!FYI~a1Vv~3C9$cJ81 zT0MNq%gWZ0Xo~bibb`SJwdA2AyA%7?@yzy+0WM5JA-vVvIxOgc<^iI!i0`(Fg ztO@8|ihiN!u#;S~#5u7qwffY-&PfAzYxjDhrOTF3CtwNANOt`vCt(q0Uo{Z9!o+}EEWtHAgC!9! z^+RsXNFB%X(z!e{j@9%4{(Gp=2f~v;Ap}RB{`UCf-xDTI6rXo*_gfbu>3fl!5IEiG z$UcL<@PU#jbe_kWWV`Od&ktd}SibPj2|+{PThSy6fA}Sf0*jlV#uJ@Tm2+^0iPi9_ z>pBaQps6|cU^QlA0wh{pl5_0(na!9pM|AQZ3KT`!L2uCSFJ5(I%4=yEOO*3;7Xz%q z80-lpBijNb=Gx|b*^z|um9vZQJi-?85e)CE>mK#+crN0LSPl?v1$>VigUJRn?uFq?RWM;_s@85TPeYX$oA+CZ3nMP7_)!g2X) z*37qSk(-A;_`0$ZslL#p_V8MaD((RL@Gen|Vpfn*s6K@w8Ip!l97t@s5?PYslaW@%sS zbkNKQ&p*rsyHpq%sGLdv*Zq+9>d--O>2L;R{rt&K5>dv>$@L_*0w&xiIlI`W+!S!c z-~D+!99y;!Vv4;>ods|)>t~YXI!!79@~Gp~6Bo!F;P=T;d182RqO5EK z&Dc^vP#{*nxyIXB+U(QQ50j=)ON(lNyiT|t5>W25>Dn(d%JZ~jhj*Uyf88Z^Unn@J z{jTIdB`bX0&mZ3;=Bc80{=k9s=+5t7N8w9_(*6^^!zf2Xpq257!cz~!0$jzti&B!K z!DYSz)rvGl=+zsHS%XM9M5kVmvIJKM*DX=Q17`|Pt@nH9&~bw7x>u>YKY2?U=@OLU zH)`cLGT8S1vU(J<$=G)$xqxH5sT)-YMLJ05K|=KYMl9+^(>y$)vRDYPXp(7q4*d@s z8iwT=`k$NC)iw$c7@5L4&z;at1};CT=l!pRx#<21A9nCr2ztWgQC?mill%ju=a_7K zfc2{DX$ArA2fJB+u%D7<=<&CFLACDwWDLSs?bBsRl$OF@UF4MPm{2l&p5y~TQJ0Vq zHi{lQm(nd0%i+n-91R^TUH5UM9-a8D!(CBnZEfwe52!|jBai|Q*BBY=_#6^~{o7KT z|6Jqp-RI#ZFnf`6AaZ$u1)^Z`Q&=AM@VLk8w>Q0MK(iO@PQ=j@QD+LI1fKf-3k{#@JRWKnjC2=h5gKNHHhW^hrP4A+QZqpFX|g z==y}p{bA3nyD6hA(6$_-q7C7HXwmz#dcbxMk#dXGgCi^NC~Rl_Kh?AWFm(BS2VL=b zepv8bST1;6c#zw(NIsCMpvN3+J)q;4Uz0hkhAiJNoF-uZ{p*)K9jca6lcJ1BpoMLi z=I8|xUgUopGtc}3?fJZId3seJcU3ITcDfxZA3@@qxoH0%UmPV~0g&M5dnNB=!a7Lq zLN4D>mERN0Cba%Z-M*Sdq#invX`v_*h3m=njaj;uPE?8>EXs}x7J9mjmZ)B)FDVW< z%o+ym{FvFlO=aT5@F!!P1-_p~yG|Uq2JTDCy*UU4*PmocbGdgqg}CV8&5qmM_q`CR zHQACCK0sZKbE<5;5@fqH_9LhPa|JHh;8x8eKtR%?FdNQ@@T^D8Xgj&JWcS;9gg^St zens`+ckbL#n^&=PJ>@u$=0SeGLBnf$faYJvE!-gzHF7S^hLM*P1rLa{Oz4*FR~2Cy z0$C;U;d;?PBaVo}s2vFrCdZU!LsqX~i+zJkw4T&Qs(<)nlT_!~tmrYPv^mT-3gSad z(1VO9P9?l;?z6u`%IW1m(NesbloT;5uFi_M@@QfphO(&-9VcmPY3Y6XZB6hZbvrN6 zW(mJ=%Kr!Gu-(XW54--N^C;{wUOLh#-% zlqQIS+KC;H)_70&_iYP~bYQz1G(PtE; zpb(1)b8|xSlMUL_b-o{1MmH^b)wNEd|G7*0ohR~|`!sGonolnmGb~sja)HXVb?qs;Gqe2LM(`c=KRX!YrxmbfOX;83PLdh{2N{Xj85~I$OLZiIzGF#D zzP9rYd9=u9C<4u6Iv03?MGmC9v4obPfrOuY#c^IPz6&xGT}w&&!J_{rQZzT;$;&ue z9+BXRmbFj8YSes!03Tx^YGU%V!E^1xl!XeF+`25aK&CIzP?#ip)KODI9W?LV-zNrg z@O>(ahmNaQLyKsopPpJi1Jn9!9t8_GMx%ZoaTRWo(n!)|WX za6!7zuTEF5i}Ud@xHqNg$ng2DKPNdI4~Zrx76^^(XanJbEKrNgdgztnPW%Am@Y#HH zY5gy$7*Xw(&NwiX-jdMYgjsKLh-h&Jku@_VgC>hGu(~Z4kiPA|G*a2`A}<0z89Lup zg+tLhGK%m9b+mtws>_k~+Ii(nDGfBZl%NP576T|`V-0(3=+sVxi8M~P{*=vFrhI;T zXY`^&y|OQNKD9!Jht;2Al(SeEt@P-7LAaI^41{H|__~AsGTcfKCl?j7WSm=TZfxJV zldi@eB=#3q_T4lZ8T~^ZbZKR!2TSH@LNv`a=wrf%Nx;7mvsBB_&Jj*N@U8+wVT!@u zn>KYtjQOdu@*av{QT+I=arGcogZbWpqEmjAdB4utWBsQq+;afq683hc^CUodJAxoA zfb6)qm&Y=T(>R+6|1o@MpFXL$-Pf?!w+wkXeeTm?^NmT+Q*)hvW{J+2Js zzyosU9+$GuZRd6Vc@lt3Amcbu2*^*A2qLzV{^6>vU0G(UAwHZPC(`ll)ddQoV}mMQ zr{&&G4op#NkwL5{|CRL+0M?$C1{qgu=-ebaHhljsOvHtYFX^j~L!jWBvA!%^r6Y(4 z0_w%3Ky0qpJJ7mM-}0VruYeg^n!dzt*Tt_;l4{PKha;F_g0Ylx0<7V^;6_69wD?tH z0DX?UQG`bCR7IH#Vdv3urxi?0Jyt9IwAd#9TF!|{GE-Q-!U{5CPF7adREu=){pMv{ zW_YOO{N5y)lj}4*L|((A)V#tw{JP6Jm}*_59OiXj_@t2;j^LEkl-lZJj|%4p_6OSm zV9|#@xdl)-EcI8*7P{v{L8+t$zXhZ1hn|()xSq_zlIawLbtd2Z0mdUt%}I%ckBuks zno{fHi!bBm%rOc5YxU}4iRv(eT%p{>K1twK<#SR2Si7Sa7S1#SUH?&&qz@oMVIJ)Q{m%6C4xj;W3uz0?4P_a*!UBNii zSNL_X`;_5ww2rw6AkDa7RbyzdN7oVi$0~A4_b=6dPHvys1^YrKb zz^hbWvP9^=xZ)T2`EH!iVhEIlMHd*D0+JF&cdRq5=(NGIzv(-Oxhp_drJNiP(9-5^0@I9x%5C+cVEiZr%8>$Oe#JKkk7=z;l?W?tBV*2L1eb9!j4z93xP3pFzJbjQz5+FaoWxGqle3E~?`sFP1JCM;^ zLcSAv7$V69Y9>voo~llby~uALv*N$;1Tg*+^EW6Hg!MKQ7jc6)Lq)MlLeFucL6xq5 z^LxN#Kz&LIak>a`9sM+#36bwcHk9tRz@45CYtKxnqT&0mT6r-6wv;v>E;I)Z9(+gQEEu$; zlt2Pv!%dqu4X25FxIdC-snh2vh4l=d=&MwM(nR8lxYO(kvDSs{KX(94F3tmg0u=~* zMGi3G-S_O-vrl-N2F$RItbH;4FE9(SBt(_ZgCmbob9Q!Km~tI-Ma}Y~xFS_({WWXj>*HrN9vs?9nFc@{qUB`F>RVXAbq=`KQj~6pY z)R3P(eY#5(OYST@6NO7ClrQ1I3OYzK<8|5Sjd_`tmVlB+s|x#dp;rJMSh{T4LNWUR zApi+y2>TLvG}h8Hi3fN1=utWSMpc7d;Dig0{Awl+Y)XEQUMbA9`H+rZ6sX_0?#1`!>rak$y9x; zs042gf4u{&OL3>M^6Ju**1K0PahHJ_M77R(7N`@-$~?(Af{-I~xVFWPZd1|m~B`c6h7el+qo&WpS z>iYlDW%Cgq2^gH9N7R!5yU{#{9{TsK538E7?H_ZXNTs{?9`Wb;_3Nkq`z^?F{_mF$ zlwi{`Hf4;gtY%0#R?Y6vOqtoV@cngqh9lV{>4%oKzpUXYy%9Nq+z5os(?W>M&VwuLKjpJgX8g-3;J~yVp}?4u!Qe%zb+O; z&`)NIi2y4lsoH))I!Q_Z-B6gXt@J1Kh?AN1DMoq_wZu{ctrQf+nwsO$(ee@YGiaG3 zc^nxdc$EBFx;VM2&%XBGrh8IsQ!O#dY-z0p=9PlNK5_lOlj$Osf>DXVSa5Zp+}=ve z*-q#3zyI3*=|78nMRTYBsUl2y1SNaul(qYk7maPdVJ&6HN6wZ2@q?LUuv_{+H^QPv zJE4Z}FjjPqZoG9p3KvVpbP3&2p+D!V2*_FX*opS9_bFEiWS;gzch);qFo?C7RKXl3 zPM1?XE(ik@ezhoDcHckd^4xCif08XG;}KF%A=`vZAlPfL^A1ovh(Re7&b%4HU5)Hl z3@Iasi79d~3O+im`}b=`deO=fVt)P!o;~vO$zM700qNj-{Q+={D-T|n(*ZXGB@t(q zE`rnw^ppReN7P@(|8YA?(f*D8_dinC{1zdQ23Kltze z>q`9RM*h#wir;!a{Qvma{~zBzbj2AE(uZZGr6ILf`Ap4BI>eK9)J7RZ`HE?mx-z4x z4=?Rc6!V?6OL*{?{M|o7vZ?pDX7Pu0!IiI*u4iUA?b&nq&5em(htjOZ!KNF>$-aNgz?5_J zamX{pVM)%t4ni6S^klcgf_L=4zb-2W(g~$mBg}eO|Dj?T-7PLoSzE{4EbI**IuDqr zEyNg?;=bd?E+*Q@M#W0gJrf3{KK7?tydw@~A}JH|zzAzmYIFCG>DH}xZw>q)aQr(0 z5+(ZaLqW_*c@C1XMygtl8<(?({Uxy|p^#`M_!3$w|6XU+<6Z4LSf_Hp2pbpRqaOhUQ6R2u23_bzS2^iX#7$&E2I3boa+a4y!du zPWMZ71HwQGIN+~3e%~xwW&h!FXcjJ1R}aXd#DHLF9iqe&Q?-QXOQO4CaPhw*e8awc z*s{HYm+J2V@TH2|7Br|>iPko&Gq?yRTIT6Hg6_g7X$Y|iPNpR}n zQ07h?p;CMgvkrKY&z&h04+y9|3j7#hgMwS4=>NS}`IGE2fkU9q4N)S4qy`V>AzU0R z(y{h~{!gDy$Cp33#b7YeX|q2p%wy$oDxi3BXO(|ap@UCd1AAr=%FKhL79Bf?3YTat z`eoqKQi<)p=IR%XslT6u70vrlEBK-l-?K`O(97m~uI6$hBq+>cDD)6AJOvo~l1TnHUB zT82khW`s>#Kll+|lGLdhO!E=L+(4+K#%=vSZ-nl3I`c#P&hR8v_;yOUbsy-`hXnS+IUR6(uiBaTUUpjPnsibm@dnRH$As|NjArWjq zbuMskNLB`MijtC&@Gb^Z-#?o9*7*R-^7yure-nZK=f6IG(@mAYoPXoy!v_zm7SK#P zH+Od*P9Zf~-uaRzG$jGksW&9bY2CWFyTXsNe{pk^5};^LdTAcE+yT0c`W6ft6(|T@ z$|KBI@J^8o{NEMG??35z1qCVUgkQO`Lq?DdlAM@ww!%;L{I@Y!pk9O<43 zI72u1uZHWS`k}BgrILTaf(HH@f#6Omd6cdvnGO}Lm;?2n!h_kr1Smj~o-bd$ikPdO z@H8vSEjD)Bz1`@IdNLr9MwvLUQ2cq@D>*DY{5@@aY~&W}>PjHOTZ@mze^TT2 zS-^+YgDD|8la%*^E3^m|Ugw4FtB!s)ZDaHs3rfZ^7nSi_!^L=6z$Hjeow(`3ZyEv; zMFLjRI=c_IhoW*dmIIa${f7ur>VHXNoFR%bGy*$Y`ljNAkNgNWf4l)vC!ydK2 z#ds`yG*^oEu#i}?L`v2iV>eMQ<2ugnBQ0+qN(C}SW9U5N4{Q5ncq8HwUe2Mn0LHNG z^5m4Y&qnM4&x}tWCQ;626N~{(sY5JtpQX3=kGWe?A_->cPGqC8guLu9%Stp=IY4Fs z8?oUxL_VjI2mBYAyQnB(J&C#*Yb(m^E*^H`;D@27TKUi)``PzES@(dt@FJI<;(GLw&BgO?sO2vN}<|1vg=i!Tdg*@3;zSoxek7Y;A;H%3?!09Zqy^|Ed{Co z(ZY6YL(GG-hG`(BQXk=RtH3+Evi6}{ zIhZ?fWr2;2O-LP;inPnv-8e}L6&^%1ja|DEKdCS#k;m4T^2O+CiPYGsQ~Qg%PI$#P z<-`67$N#OzLRdcAw{I7at{6xMuI;-@zKVRUCsiL+x$spJJz;_)J&-<37hp!{eQ<9P zYH1;exgqgvt{2x#h6$X2C@v~z5{gnt9b^+sRqKdJUQe%w!g>&ZZXvr}s@g{bc{UEr z`?=~d7+cc~#(U)-Oh{PB?=Y(qvTa}jrF=goY4L01eAkgIYyr^F=Kb5!vfJFbb47iJ z_dhkh?6gm4j`q#-bAw1Du8J=ucCGAjrU*$0btPI)K3=R*Z*P5R8Qji# zh(}KsL=3+G=JcuBwjjT-@CaS($uCzxlonx>maUi<24ffNAEC-Z>4K@cCpS$rZ7quv zuV}i5zbI>)4#pt3K?EM5YXUYOL&IB)6Xetr3zQ{ro|{%4Q{u%m1NQs_b`DVH67myl z1+ggT&aoy^FG_4No)j|Rd~UOO$ymPtY7k+>N%}JryZSNT|4gDuCk+3&%nWC3BX!&H5!QzaGy*8T)3fNY^D)oi#I&rDJQWKiXZej9}a zHIzGhm$Dtwt%9<3(Lfl4k3iqqM!M^+IWO(k83Mo{TAc@P+}M3jYXogMB=-yH%)Wrc zgDc-?!Jv<-_1Zegr%Ukr-?!u+d$TN9Nn#8HH85lE9ts4AIfx=Gf-{&#fvDcczX{Y( zOq|mEAbleI^^*RbJBm_JJQ>P4utJZwjjabqPnvXT|CbE@C{_LZ`SUa&xt;~TX>=li z6D5k66H9?Zek5$&gu@LNpm1%-RCrCI!b4>>o>PQH+^=&-;+W{wi}N44)_Lll^13(M z*+BPHtC@O0S}Il%_$2QNNmEQaMxoxFV@v>lyfIvfq@0T4xEHCJ85weRZ35NgUSq=h zfQ`qtTEfXJCc^l~rf|Xw{7iH?NU+qHlE`EcfzOb3KH%zUP00A{w7~d;FmVt*zz2$I z-HPisZU}2_jCG?|Y@B6ef5V7ZAK|Up(j%Tce?BBIjWLIyZVxE@Ux44L1@|7MvTEHg zVw0HPA?ulN(JE)h9MC}lc2SxL43xu+F1YZv!W@>0LAVdd;_*og8ls{Ffy7I#x|jqL zYWb0Ko8TP@7L)dO3@5ce+WWb!x3hBiEuT|Ta{FCaHk%FT>H*(^ERP4M3Ju?r`{~`{ z`jG%b{2vr=#2cNzbg3%<5d+y0#!Na7+8H){NQa+*-0<)x>l=!Y+}L^dhsV7=Hsz3+ z(c;m$mrzYW#aG7AGRX6{>l&jSHARF6MwZ-qd6NXUN6_2OfBq@j^*m5KS;VACLfGFH z7H-^CF~kKSs~A(KU)8jO!&s=9giM_ig878|Sm}aQ0^lKFPPRFj`p%@AgB9g|53x_j zD3&I-8();XV?mWOR|j8KnOo!zm^#QCCh5_jZI~kJNR;}@8CRu{UT zChu(@3;}GWvuEz^&o9U7Y$IPB6!YaajS|hg6u5-`zAlUw{M3lhQ4O0%M&=C`EM4%z zC*NIo_S(>6j2w%V8L*r9m3c%x$)jyuW^%sYC$H;Sb*j8+jraRszj^**?tN!1t&DLW zHshm}XAcFTCm~@VxW**LB4#Aq?$8-OuY1h57p%8;XMfEPXmJ`8l`EcqJ*inKXMcZ| zkF=f&_ImfhIYk5xOO5rKxJz4aqX})!g(YMnp?6+EPq-}qi)`8Bv6x(%l-T^>!yY`` zh22ro4HqN|#QOc|iOc(Srjf9Q8y?TPnPeo-h8TiSIVxvWo)qeURdE|U*gOhNzgfrX zXQuqUf4fkl6$By+Am(ZTtTQksAu(~VP^seZWKdco=7(Hq{qLy5E?koTiBdXf(v4^A z-aqUvEWMPhFOl&Iw?DTVK=(cp_T09?&Z9<-l!t9%81uy(?O^+aJm71$4-%ah3OlW< z(BDE(qV$mLNf|GTJ<37?sgdjnuhdk-e0ys{Lti>_CLv^BJzUmwILH#}394nV=T3G* zn~*e#j#d}WF8`)jO0gbfy*<*xWPonQ({Nk0d4y3>9uWJ#GxVxq5{ITeWKm1oFUP!3 zUN9LOUYg`hcONtG&S;@@s(pI}HqBL9&SRlLms(5ggbW}?PTHm^{byKXk65EzPy0Ce zFK7emcVB|GE*#F(z=uv%)2`m@MirBxzlk_{v`rx$?aP{%NbFuvEhM=PpzkD1AAFJ( zciG$fTyA=>>(H26DU)b;O7Ao!j`fF)HgRV!saJpH=m)oO20Wb5?MrQ6Ns5M{s?IlG zR}?hqQBAmU4#*C2ALAX^Dvdi-k2O&^KXwzZM+1NUtGzLUyW8&@d9Cf06z<-AQreK5 zdY3(WZi=?}LLNr=P$BzZ>)l!@8TWZafUL5zKE`a`YjJjYk-DA}w%0Hvbs>;5tya*g z$pC`vNs{R+6GJ}LUCb)54W!z+4_(h9bw~-Br$7Wtt>tS+wNiwXsqQjl_Owvh-=8{Y z_m>N^;^Jren34f0ifq)EusvSX=?tqKJMHl>{=C2<1hYr zlb4SJ9$HX(W?)jVxEa|l!YrzEmg+>`(aefXQL8^J*>hmvoB4MJ{qNvM?dLM8LH`;3 zDATrexo>?9LKx9@eLC_iQxJtAYuJ0;AaW5ge3iDw zDc?Mr1@~Q{PomcZ1Lw&W4 zY+>QI5vDj2pwNwMO(s6jLSgK*@#7Z>!qok@Ov*VWEQiBxtQZh>@nSbVTc;{zHnUa7 zbdx%Jav1xqevye)cYn^T!RVaM=egFGBC=F*+tzJM)fa>4)TWtRu{bRItB2$K*>mRf z5TNKI{o$G|3qR0up78M+A8K9s8BfL&ioT%a3Q>H-}}XD0NP5)!!)-E zTP`uhCL{37&(e#`MS_F9p#16ZFhPphk3Lb;^(?+7-djpU!)2~a`G*LQ+a7O!D8zIW zUh-~U)ljS^V5y)LExhB!YEUaqf_3MXcQ^e)+6|K{tM~yT&&4biJFh(_B~A>tPG&yBi0&bi-c3usruR z^(Mq_x8nIs^~Oy+$>?T?^Ewbx+;buP2?_sC9Iz!ld4r?O%j6Ea&Ybns|d!70=Gzx$)j@ zx43|lNpP;hVp3GkiE%Xbjv1P*O5!fh+7YU@BP(#Ok(~=zj8E&M@g4S~1Kdt;T-2c7q>! z-Sq7uc&wXqrkV_Z2EQtz^;!pkJ9Q_=5Eopp@9LT=sDu8eedZ%}X&6}hRr6G9cbbQx z-D{yy=t&>)MC!BDZtnbNGSbHnW^zSQTP2CDnzxk=UpF^W2YWR?I?#6#M-YcjZ-RZF zfQG%<4F@x!p_+vp%x}si=Q{|~YT^b>T%oP9R{=&{u6?nzxAMfXJMmWzou3IfA2m0A z+_IFFuE@BiC(FHE2r=y`MZN5U z2hvd2x1G8fNeTaMY|DdxC!7mx_cpa3`WYo9W+ZGz zvJ3m}DU>_?@`2&a4V*5rpAhrDMtw$5d z@%;x>@3gf&DBLFs>NyfKZY%6<`|f9H#@!f#eD;W$d6BC#KEgf|1dCEDeAdNpb#=P0 zR69!-JT7QhZ>Ez$MQ0xIC5nCOW*)Q*htH}xD-p#+you^kT9QSNTF>X_C_6j54XLr; z##+t%Okf^V_4L<+r*IU)gUaSJ{FROLmv&B3VY)WSxBSu#r~FpA1lW$2nml=hAkS#t zk#jih2uvH6_uzKlN%X739Cy!wo|p4cMF^3OgumyVzt+_W z>KsMObPq>Xj1XrDfQpD7Hptb(ByK}@{KXVojCr}3ZHZq-2YxFk+M;bIpc%-Q| zYUt2`3|bT*h|pr{hGNtppa#m`!Ll=_PuCEi3?l`cYdlsb03l^AxiFH{WeYcTT=iwIyBTp{ppr2XAs$$)`s9? zqqU*)E?g;hfV6@8FUu1SZ33+a!7z{win~mh6x|NA1U-DjoKGy3nS#2-$fO^93Our!~_HVB#{<>BY27WEZA4w|C_1Q-|}WJ`c7tY2z&BBElt= zGU^2c{0S08kcz`N`yh%GL2m_w_y$R2Tq?D@-znI4LK7TeMfAkLcX>H&)-WA%5jXUs(HgsVl}wVDA2Tn zfq`cCzSR+N$@2Zq%tz-#Ztcd$AL=_gX1nQ^=6^?tQk_qhZr%I2WcQwdkAC0%wqp3$ zixP9X9em@|ze_|X+2s-Y$7sw64Sm_aU;knw`NR9hXgKQiAGmVCN)yX}R)#h8kvZ6T z>8rP+RZ9IfY&xA%obp=I;Z9J+C=1)UGk2%Voja}d)iBaZhtJp2E|MYZUS2uY!_(7` zcGweeE^mt?$Buauw$=!4TL1nIEjFbkmS;uNFA<6CNqi5&FjpW%f+?) zAqNcU6-o$iS%Y*%i)%)$MaS57;`8ily!C|_0EqoE?8v3oq&j;|S}{soFcDtp%jry| z11Ec@oenc?bDl>CRB8Qu-Ije}SIJ8DgD9{R;m;$8AA*2RZhlH6ruD|m&=MIR(!sxz zwO=jFRTh)$FgJVhbsh{FFTVLQ`V3MO;}RoA$nk!sVFR`$57h( z(R%1=1uT&RxumivAC*CvOgzNQUdj@w5ytdFz(HfS(M_pR3odIXyzza*P-m z=GTMcl1D%PJCr^%G(h10u0aEoz=nBNW zAz!ma0P%{!qm&GblWM4@E&>A8QT_NUx*y(h?83!H88oolevAUvj$pO}9g=0@sW7}+ z0?j{%1tY+_&89ev*j2^L6gUu*O5gIJ-Yt(x74{tm`tStv;h(o7YK6Nzj3XnWw4scEE0z)<9%f`{;b13C(D#N0^Ax()uXn42WU%~#r)pePy_rW! zUaollfK{68OV~o{oZ1s+&u(7mlH}B6^Wd95(BLwV{?{MPqvi6mUzLinjzFNpe$AUT zYn9M$lQR2>@MBl$MJIX(9ViEC0?Y>n*9Go zwv88Cl_S}QTPT<2Lv3UM?=c6(;?uJ;V=I*`EKsi`@8GtccnUNdhO%cy#o3c5&u{&} z$SgfH=j(T-HbPQXjWE6*?k%muw^y40*1*|50n(Qcm=HLZ&z$HhDYsl1=1EF4>rRVe{`zDtp-gVb}*e zysC(rQ^{0Srlyy_;GMp4X);q#QF$FIrNI!QP~34Ief@cTm&bOadFBQVq%*Ti?hu2{ zTfe+|@2k}_h;7or&I6L>uRcEA{`mDXE}K_d5cSjFqCp=&qr%_IE7XU{*?tu}pU?9v zPmFYw`7q!KhhVzxuq2MCng@O7D@U&&x!Ib(TT@#*531ewv=5`&K-HLE#3^Uaj# zosex~6u=L6e};Pea(iL|RmBxzX1YqS!^dGq?*xZORG&Rov5sWm`njp!IDgY89k9Or zSD32g7oP7tzH$-16BltDmddE1u8XIGI!y8V_fHo*O;koKkX>>kHY}i>sh(RXMV!a&($*+20 zy>YJaiR}G6ts=F8NG}&1ab9I`Klz_Fk8b&9(U!&-YY$h9?ca> z4EZ#Cg4XSrtzWU@adHg1xQLyilvTklIduIg$MF^z#Mf_X2ZUPz*UkJqeu#pJ{#l@d z!AxkaZ2aTTUMmw1&{>Y>m|AT**eIs@SYqGay_e5ibd`gwk^bZs!fLt-M?*N}bjHL4CrhIBnE^;{!Lg3S;$4i%4{ZTs-+ zPz9dV-%c7D(+{XUP8Gv>gz4olUm5bt)~^rx`ncHl&xXTdng8X(VOOu~*PfTeYFll3 zBG27ztcr@=Ob0g8ZhjbvT^?z(<*eM)gf^q)J4<$WH@$IM;lE~cTN|5wGeW0l%ASxa ztmzY)P@{}Cy`d0jVU7OnQ1IJm<7qVj5IF=epW^U~7cD_7az(fLbt-|^@Vg8zT7Oew zDXAfiDCGycvT5+zwe6()eFX(Y@-0(`L)sS_?Rnn>w&E)9G-wqGr0|~?uF3Wtrev0| z*BB0so%Kg32b*!ePq)(+!=g};VJ(rXyxgg!@=|Wt?oTi7ZBNV~kFW-tG9k<;J%|3f z>bs=D`p?6P-|#@5$Ib7@d_My$IgP`wUY&BZWk7_f{UL2m5+tJr7UqCyh&ren!`EFM` zV-}03G!mNivLZ|YHqH9jlr9&$%3$KO58WfDhLLZr{@y$+3lc-~ZF$K7aZOv@BBo7k z`SHV&1Ry6|+dniAL3u_EiAeh*DA+54Nu9aN;>c=i_^`>U+o*5i#pxB(k7jA8g6Yi{ z+iyif_&4-i~FBmYf_}cCFma8hmuLZq%L9nftK+Yl}mf!bi*a znV-4i(^nb?PTy2tu<`NP_Ysq~%xQiWx>-8Rg^;S4aaMb=*s}~a%+)(P&nZ#EfgCRD zYq&K9gTnAzdCE;ao@l%92d!2#!t#n*>gopv4;*+gvl^+))hef?bqm6E#&}Hcy$_#t z?YYAXT+^a`{}yiPkJ2uu)3b`76zr@^krcVx zj@Dy|a8St;KK$x)?$>9%U;KeY$y@Ra6?W0N3SiE#pT~)Df#r$D3iDUp+Os37(~e^{ ziDXMw*eb-$8hycH09~ld5SU)PccyZS>}A-yM~@s_>$mMteJ1>)ht>Ltw;%Yk;z&Hy1~us6Z8moY1C}+~%r^ zicR0*mQwhd;E3bhEUJ@Gm2;Kk<_3Q*{IhGe8?l4xkQb0HU!HDg{+URtpcZla&jZH@_clyyu=~YFoaX7nu z3wc!`>5)zS@?U^2a-+8@kmLT{!S__A>*)R!$1cE`$=>e&6=NC06uQv{%~k>7+#%f& zaIZDmV#&M0Ql|Q~)P4V+Ap2?h)Tuhv92|8{Lc2G(VjArti&113H$H_&hg%hq@%Q?} zL2pgW@RL@`&5y_wt#$?hG$fZNP5bayTLuZ zdU*hM%_4ty=H_E-O452kM5o14=bu;&Lc z<|Acqw91q|7cXoU@R4PB|txHp&hQrPE z3&O+RDqLZk8~7VlosRLfP1u(S<|&<#ieap^$TH#6 zGA+E%or~^uUc1$iGwI5l(ePU$Ze=%?|B_`K+y>b{ic<{y@ zQ&vQh&OR{qIJPn4x#d$Ysq`9_?FPMlo;suNd3Ci=&a$sg&P-h4bURM@q-k#L^Qv?C z%Ri8|J&SC+%Gd)7RU`w6vpm5oO#Rb1ENpg+vT|VyrvE&)F)dyC`huGj@T&$rjJ67x zfd#yUIY{~F#C_V0Nlv~`&DYPRq||+OaKrN3AyzRnPk5a>w{@uHR;%+V#*ycHfp$f0 zEc`2qQX%>0D2KBCM;~{m#}xhgIR#;yWfzb5W6!?j`9=M?A<`!Ckei!D##iw?cgQ+6Y3T7at`rD{E}r_L z1?3t{C9vw8`|SfCKJVy5S5(@_3U`V)gXRspVnY`E)5);wq~Xh%kkZz7(~hvV_eVPV z6VD2*8G64qUwvMLkVk39-Z}BEPe*?u*`dNAWu}V#32(DDWewQ<%*%L@70B6)idJOo za`o{EQ@7+TyxtIDx4Y+)Xu~T3SqHS?9vFVKidVUHB&y4X$e>TK?-HsenwS6$DE+Ld z>3?z~2Pvhnxi~{e%4{XN$UR*7bzQS@Q z=WWi9(|q;Ra&X$EAV=BNs(Zn4)1Qx4+HiPzGm`tfyr!pH?>eI)H}soy{llBW!kM`i z5yw#Lt@1PC3(LPzy}rK96ERIEM1)nQYP)Eji*K)`02*)yPtekN!-=_vAzJtKW=(@;2Z>pfeK{o#53g?EUpKxrQs*yt&_+ z={=*urj-oqa=%NLF2*W)D@};aD;qU_HQo7bd@k+5z>`l^Zyb~JeH!$}xqi8sG1dDf zgHem^{jWdPIhIJwc1%SZ1c!hcQSqC{B}WJipuade&-0Pb)_~%v~THR7^dtiC21PA-4?7AZTA|_IS`X=&ooP|Vmwar%YlDS3?4Z$I^zqk zPMp>EI+?apPQjn?VYa%Wf4lw)-mZSRdfLTT)9~c%{j2#AvsP!@8~$U7VTn< zrZ-UO`OA3aqRd)>RZytbmz=Pp0_F}^p{kQ-g4LAO%8LYpLPAP&mi*rK0mpucF0H## z%4&|JRnrH0R9swKRRg(hGC@7L0pL-ue1{SC_jP%CD=KryT5asjv1B<#EfA)~OnWaV zNv}**Pqh~Mic>1w`oRAEWw4wa6$fR0y9qlUx zNX=grdtW-tn5=!bzC9jcxUDeN#p&DEuPcAmhz|gXT^+cIz-MR07fzTHe2nc>=3``j z&8nDU&w*ih_LTyq2R#Z<2rvNmo9L<9Vwp@l*#O8HUO%Hab%w~KljC8)KUkj`B#y%R z`qLD{&%>>OVjav(aMxzV-v3xPtB89Ls}%)`OGcdbQ3VuD3gvrV7zW3=?^S4tm65gN zg{3`uw2TD2oFlVu4z=)j0h!?$Q_^~galowUgdaPv>!Ml&``loC4qr|jqa@@u|T^+yNE zO=Q$wVwM95(i2ReqqS~aRJiE4f^766uk92iS1l5sz?Bm91cc88SI)w9 z&aSFa+LG)1^f&ZvR$D#)h7R@%_7Tb78ZpQ}68fg@0zF<^lEj=gST3bKwd2DKn zro-M9_Q7w5gEV-HZ1;9xY;3G(>?AjQ-nzFh)GohWsGKmKIcm_1Lej{gZv@}}{g*G> zb~c-8&_%e&o=>)O${RwEISs~D9Dk+C(eL!*{v(stoU`ECl9horEP^v6AgwDwv$_7r z@iOT06Yi^B<(oHe(w%lcnQc05%ruC4{z2l5Q1oa!)BPkE)M;@2svN$yjo!NjvIc)F zRTnL`85NaBI(6!F5&G$jZNF(;Xwk1wbOG@mTp^5KvTL6&I9ofN2XZ#UB8m$Z?Jz}Z z-P-WFgoFcIyI=3P@x+UZ%LbI}E22uxDalN9lN%X3%tzi>H^{Zc>TF^bdr)*U{9GYL zoXz-9%PhNSt<`s2Z!`8RC(>EJqa!XIr+{T&na=2cihZjIm}&jy&1JF>n|^Mvi}|dotcr`geHpuq_@G`&Bul0%Pc;&M5bbMQm z>GvhaquzL>GAnR~=dpjX0>YmXFsz^DzOpwn1SH&~_tgY6z1JT;pB z;G@;?NkMzmn%)T;vE?PVrVP)xrcMjg-YUlA(wyO~uSXimpH`q&zEu6&oNgBfgN5uG z_L}g=_T9ibgvGeb);6wm21?$QswWHX?A?2| zcq0&k7lh82@mH>2Zc&Krs`uLMLW^Hw;(Re@M|<+>#AtCohpS}8TPdyI`18&4=PN{u z)~tnq+xtsJ(jS|?8{M!eiPVxay?He&?bzX^B*YhaKr^;|=`}knXqW!&@*}TnQihlG zH7a>;>E|fq~S{f7d5K7 ze%|r>LEz2RCBB0U=86p^F?6Wq&LCKO{o#g8W5WnvaftJFm8`oR)yum59i-KH!<(d? zOZ;sbW!{j%X{xBGY<2YSeNG(iS;Yiaz1cS#?)RL0M#f9o^2LJLDIRl|G(@R1mz@sp z+NQrE85^mku+So$1OEmrPFBXGLfqE~XKP8zXPWFNF#K0z9nLZ=Dyrw6o4daRDb2b+ zJZ8@=P|H01ehb0V?5rX$)$csQUwjH$T&SECIjoF#DaAIFor*=RZI3x7v1Q*m=Ww znO<{u+mzqkFQxSCZb16hd7B#B_35?+r=y=~ERiPvSIio9z;$C$VdF|5tCH_iXU{eU z+^KF~@CMV@TFC6(ySM!F>X@Pvt5SAdu?+B4q*;^J&@6Ae_v^IvCmVsk=nthwI*V*i zrt7^++WzB6ar`)np3*R|-y4fjxIj4y1{HPMxOc>bVIEtpc4MRYO5a%bRZ2 zMDx=-InlRP_{iuz{@7JVzYwgqs{KKJrs_Go*{8|QrVQBePMHf&w++2RTg zl4&Ee4ebkN9f?#J=jGy(-j2(}$EPJ3gsOKhS(Q$~umOd`q2kLM%aGJ5nT=Ya>RsC{ zvbqK&qc(BoDajqrC^U;d`oB0fF%$8dM^?E`f>oyKk0Fq3Xpt)UxNkm`fDz%1AgK)F zHUopbVIpgc4e&|VX|fzCbtx}5(yNjCNZR`$6K?zL*6?AkzUyxSlom-f6032&J`Rs4K zFZum96O4iN%5}PQjtJ`UR0b$2Z13-1$>rsnvI3+?x2{_*o0fn2fXD_i&Y!(~;@+`o zTUfPnxz>U_r8RBZlas!i$2jqzAX-?HUJnIy9R|-6bwCq zxF-0z>8(9F&=AVWD>AsOAxGA;0GvH(oQ}>WaUO(Y_axJ}VTR8A$$>WP9;e20{ znQU~qI&R&UEph(4JhP`a^QHZR+-ZZR`}8sWcIN!HjiW#B+}ZkiP=@(?b7^94PU=bY3r95rf(fW8k#cC(oFOgr|*Rna>!+pfzx>y z4l|w~@=G_j*Vv^qPvsO{TR{S#W1`IoRFT)(A?C#$_-{HWclbjbnl9 zg&pPGg5(m1Q|ZIn@=;WjAL$H*a;Cvxr}9&QuM>j)($w@KJ5GlVf+phQz1tU~M|IAS zrkOnZ=ry?=ZgYxs1PauoWzg5iI73=`FKDD_$4AwJUI#=t-Hc2O3%7#Rvc+$zoR+X- zgFozF>*!0Tn=0IoO=gkQMn~IQZhlhB=Q9vsZJ9QPh9mPp@_TZrLsKV5`nO~SEh2X{ ziix(AVNbn(v%Y)GRZ=T@Dd}IVVs|~EMWc=LAb!JFOxtd*3{k7goc4Uf-hbZdl2zq? zTI~5MY+vA`Pft5)NzU57z>mhlhAOKD{m70{)Ntv1{6*m6@cP|dNHeDex@T*Aq87Iq za?$~xYO{cOjR_QvFf(K_&dY7_pSM?O=l3cDW8C#w5b89*KTBwD-gRV`k&1cNUx8PK zpUE8B;bbmKwKa6<{cH+yh@}}67v3w*UK#|9XWA*VpnTjk+q|hmp?HYy--l8zwX&18 zF5RDOJ7aZqGW|(hHT2c zyPiD6TJdOA{pY~o!%2>c($Y^9JI>-!7IxVXG<)acDrZJ~h&1PvprJ$FD&wW$85C=pPu2O%@t}1UC5?cilq{QNi}Y>%Q4@n@a>-*`wk*?HRC{k@ij_1XGO>8)lBN2mN<9o zs}R&gm;iG+7qQGJ?|7|BBR;^l-7U2~IKb=iJ`_ zvJ_{TJ?_=D_3maiy#lPG=2XwrP3R9Yy=BjU{5NqI6ufmdR`^D$G!9L>!r~+GOwKCn zPfR$;CaCVT3s)QqW^5=iShA8T!|v0Ih4i&H_^E1aG`+UB@^GX3?Rw?ayNlP~nk+A` z-RX}%>@BWYoV#Z*>q2TORcf92rFpg4H;d=Z+Z}m{if%kE4JWyy#u(s1AkHVUrW+du zJ{?qTxn#CO-Nm|~^YbP~OPk-iS(f3qqUoJyDKKeg<)~RxyGwIEB^AIzc*EQq+#@FldNd_DAnU|>v5@@MXoyr z8=O15^x0PTS?R1*pT3GH5sO2Y`G4KM(_{SI%j1Lj8U5bcn@k;x4I9%ZDJ-%3Q%D_dH@!K{b6nk)x6UxL&(DGmC6@?EtFl)1B~rHRvXv!jTC|8{>&jA%C~Mhs*=kg{ zY(&0!Lob_AzfTf{1=9#O9 z|DL^jbJTWN!WE^hYrn+7y)MDF@Kv?Z8%Z0-w_n#kW_l{?2-V9gE^--)Wb(=wbU>MO z-K{u-rT}z&fqOTZhc}dD^K@$NWymCHLh5UN6PD2u*jmUklPeZpLQq zUbp$^rDBeYE6Mot@qFt_a(^5n`~I8zi!BCjYDh}J4NA=sH8>Glz|5Vww0_B0Q`7d7 zp^^f~59mB`Hpz^Z+?DSt;r*Ze63S1w`_-Gr`R?Bae7VucHZGZp414kY^?S%0B z8e9=ZU4;DvcJFzMzFK-tEwT1r`8I2qcv1hjISy!IbJWS8o#!;qLaz) zTH9M0)#dL$-McTyPk#IO9bp$h<;H_q#g0)y`2j=Ppa(sKX)5ik-SJSgfc zcO?#>b1$k%0fS)!=H56u?#Fd-eknO)?Rz0MHv*9q>|09r*OEzT38)hQlbGmB89t zAXs7-hKuNFAa}NOnJ_qEZX9lqTgTJKWi%3;_}Wcr<%x0E7hCG10xm~t{zwTo)PM;Q zi~Nxt+t&Nk&ZyR)@D++QZr1hn_gynxRBQPkS&bD%O_!vzgKn&F>a95^snv`$N!p?5 zAJfYU$fi6{8;S*QnV6svtN2}qCOV3$<4XP08(LB-d4@7y4Oj$vNll%i6q5tU4b?zt z76FduzEMlbvPazO#1QQA0`lA#7zI|8;A39%-E!6(-AW9V)60{aw2}t#}@%(t+vn3F|udC&Vu8&0KnI zU06nHvE;>gf*L$JnXqkzC-OR1g1>(q-n0x6%&L1Rd6d9fT7Xo_o(f$LzxmUI?|@gu zY_q=2+^beOhF4{VDTIZkf-O7K!qdHFTTK2txwN*!<#tY+L79W&v>cI6Rj5xtZK{jX zl?Cp}V1w<$QrJPU=y%;vIa^@)2dWh;e_k04q;aONvLYX*2(4N3bj>+0gUv2t7}Fh$ zY>|7GIof((Im~bt4oK>ev>lO)X+D8mJ|fE{meqf>)8FjR(1yFRi^iJr`OqK)`ydFS z?FsiohXWFhFr!YZ=*efV1;e2&19Msor##S81tR#;dL2 z>jHPac+>yzO4#PrE`W9h0qgQqy2l$f6IE0bYPxaG$EsEaHD z&d0MKVWV1+P$GE-V?)qi@8PDnZw$_Nh}QR-$La|-uYVe(nri`->hr@F6=Zu3s|}%V znV0KXCwM0m>LAq#CVf9*s}t|>!Gi~!)2w-pKp0VS-1x9DW+jtq2tqz{;4QifzG0XX zJH2YV2$G?652vutwK+|n471QqHj0_c^Cl+%fo%QK8N3b^&5yDtoAjJSahv^4_ZJ{) zjzyq<@##%4g+kHm2pLx`bHZXZBjgMML>BiJ<96 ziicxWmhnj2-TVggm}XrgNCC@XOqM`*1Ng261}AfbKOR?jQMWP?cH)Mj?}V#0?tdD{ zl^tVfb}Ri<3Jkg!*F zI{xi+tdP4b{>0k)HOL~!j{anTZW)_9GrWBOzwq@1?Q`F5;jk$|vax5nbzp15p%;OxHsL^5$A-gRVfEG@ zP#6lSU2nNT(CF7~2a@5193otR17N`pF!H~l(& zA^+ zUwksSK=-{C%F2(0B)IgX7V`&NMSHwBdK;s%<`)(I!pYeamhJlGv7bB*rbno%INJe3 zN%Fs;se9@uLBl<)k($I!d`3Ie#scEIjY(t4Lt9?<)~=Rf7aLsLwXgiijnv&%Roj|QJldJ*i0`ls z&qHox74F?DhWkM>m&TqzK+VBXlwLIu!2P=ORD}T)KtQS&N4U`Bia7cJk}ojBj>s%* z2Y0K^x)R|8;cQ>FJZUg{+4Y%*y@36i5uf(>D$etj?Ck7*xd(=<;vfqr?K9ztnKFl? z6jFXs8U_E%;wuNK9)$4GSP&CXw~uMbLhjDG#=XBB@d!Bx9O5NGW2$HEpnL@vvz{ts zxQM*s_JrZ{AQVSaS*wI^rCwbdniJ3p%;GUaTH1NztPd?S@^P!t0z8m5{WkDoGQ4PPJPHhBB-Rl7G4U-8!wHC`tM0K~&At|E-ZNb? zJ@)kW(L4E-xb|7=;*%k zKMoPva@JEf80#|ZYJ&i*ZYW-G`+ijo4c$ss1B7jI#l#$6AaD$I8!>Xs7r1?Tn2Gvo zjcj{@GutP1OWcd&+uDKeP$z1rF;aoqVE=~hF!h%ZZHvBUeS4D>k(jj;(aKi zx-3vAz1+g&zbo_TF>E8fHbLnh>`m(owXdq56y5A zXN%xGws#M@1RXwSEg<b?kxxk5d?X>(azyNIYX3YEC=Iy4Q@VRW+ zLkp@L4jy)d5Jo%FFJn-kt08n=6nnH{t8@D$Xb0V@EN);`b~&Q1Eor>`u~6;i6F^Qn zZbY8i`lzb823Q1PIIJ&F@%r=isp6v7x;o=rQ?YE;L1Oq?hm~_MLs5kh(R=5?y;N9x zKg+#O82qvPNYs}hMM{~;MUiy#Wp=8Q)oyRi@&Qcp09V>oL>UXj#;_^7UtcUe(d;R5zwT8+>Xt=g*B+861xpGsj|i$ER3%`2@n3KR zK$^o?pJyUhaNL0T=E7!UNZC z!ah7L?YA`@7F?YUBhCH-7^Dn~n3f||sRQ>rRXH|x;Etaq3EW%&S$dgkE?P&74POX7 zN_GQ31Kpw$p?+6B${VV>O)HI(&xSO0g=yg}_iNQeH2oTFbO)!{!ce(MoqExezbHEgx(E@gj&IjSulANY(t z5u};p+Ovh8Kg3jB6w5QpW}LD{%v$AoeZRJL$YA_P`A4~0pT(sfeO%uZzXwmz+I_F# zyK(J|?J7SLbkBLJIWz%%if>cxB3|uvb0Cy2r`Oi~;I(u0ZAxumB=)t4JL@1-3P-0r z^+bi96~V)lyQ(oKVtX*sQ%a%0+%;Gs@x-jIcprvW#_pzMv9>Ei83`>mZy((a5jSq= z9DHn(A6nDXg=*hMz0ya;<3WgxHa6Z&$THN3z4c&yoA8MQWfdT+;vf+`*j;A1u~wF* zpa9cMl|67$&hK!jNdIs9p4C3E6wBTN3^^n{qA_OvJDaw@>XkD!6EE$DW>#9xdM7vW zCD#qmX&VpHhqwZ0rUl;yRfR%IJP1X%_=?)7t(>C{KhWvrM~<|rWouP9W2#-Emvi}w zfRCbR#jbR{^8vN@I8{=OJ>}&U!LH-c{t$WzJV0D%I~%WkniX9!znVXKft66VcRb%# zyX8T&uCw{U$b`?MxF*~GQPU|-N7qy}1|{W&?{)j8--f`EJlcv;rE=uoJPQLo})r=_L+eE25Pp)sZ=i;$Gru>Z}gzYmCOm}OBbZbaAv(^o8knXURUb3M@f~eCG+Kj=cYXauv_x20P z{yNsA5dXp*aIXLyAnNE3z+l+SlFbJaZ8z$zW7LI$Cz6e?cB9|?wfoKjZK(P(Pi8^m z3A4+9i^|{wlfOm08rV5mok{(r=t8s;4kSuxIHafquqV=?jvDt8ZckObMV5dtQ1{5U zJpuMdw{J%JPS@eb4i6eqEF0dvezg{*e@di`S@kTM z*(?S5Ch0AqWAMlJEz%4G{c#Ro>@k>8me(51>g7wztg*~--h)m6c%)DMMLIH+0~ks9 z9a6Y#>27@nEi%l8mf6LLwg)SwcV6Q0)Ly-A95~UO3$!y>fFF0|=2;W9}`a#v5lyrzIW$T|~8`oKa~rH-7qQBzx+C zM{n7?8sou^?LW$=*R(0&@h$M;8zXPWSySj9k*$}x8wJG& zkq@3~BKOMjFgiH`hOKa5?)}?7hDaiMr;d~M4omE2$c%evRU!}xDN?`^mC$ieQbJTk zNRoo!YL^0<^%Z9%*=7jc*U+7RfHEo_3U1LMXt*D($qFM)t%~#{8~_bw3s@^l1e{aQ z`=A`|juKGBrZC7IICTp|Yo`m`+&Dlr3up{mUJ2P4p!Y#bebqpTKmkB{YVl;jHr=PT z=yuq%?rMc?+d`srywEulgrwa!y}%ewKvukH{{!kC z>TsmW2hx70e?CBk;SiB8RF5DbbdKAr8sESRiW8ed>RTa*_&{iHs5e6I+EDh=LjqcD_c9)f^Nt))!n<6$o_3H9`c~ zINTi}LdT815+3d1XV`%dM(qdl`sM@9>yh52j=Ze|;$?v&WqGCD89QgU_2SA!i$H=| zz{RE8(qzPPjn}fA_s}piAwli}<|l#QXE#@OxNa(D=SFd`@L4R5b4JdLbQ(pcZ=obZ!F!2*<

7htuSR4%X_e z{dj8Sli2B!I11*_6k+o`Aw4|uB#27o?wZ#-f(6UUz*1vU=6Ur|ib#CEb_*GMg4A&j zCd|Nr8Wi*HI`*X1;ur^kbKa1VEC2z>`7;hO(lPq9qt^~6Hw>NWZ)1*Yyh7khA)5F& zs870;DL0uAV;%(oyiZ5UZv$Ea#L+5x7!g=fC1iPbA~m+~T3WWSIPr6Qv_5bv|J|!* zawA0(x5rT;tgX8m%|OZqu)e0DIeC@-IT9*{dfqs3#r-u-aIxRaxg9 zsgrFGjo-!vvIXDrGq$B4Aj{FXl6#7aZS@40S*G)oW!=|SDYnE7zrv+XNqUHNzuxVR zW_5PJ<7CyLa?lj})4LUtwA0z7GUEo=vT}UmN!*nj{h8DRN2!N%lqKDC@8X*?er~)aRS(DsDMo1u|0p@zN zU6V;_=|@^BdGybjQ;^f(G){e`EpL8Xw#-7S^;6WzmLOE)8WNA!zks;~ZQ+e)4@|%y z9Of9~_Zhp>tA8PU!7sC;tHV!uvL_zG$Mz*9iK&&5C630=C-d3NG zhoLu69fV615Q5t5-9D>%-J+A@C4)E!61&y_4;sEPb zu(x7%=_nq@ts4RrBpz{fkQ=>h&a1m%C&Cpc{waE&W{v@syRw@uByyt>*7=M*VRk&_ zD@jz74Q)73mhdU5!Vd{E{QDh*zg=8S$BIbG4sz!)5dty{l8O;urwfi0>c2)|jfs}W zX_`u{jN?dOqt)kgyZhIoom)J;P7oD603%bX9#yXL6qG`vs^gplN?6&p3O=CNUY(u= z5G59g9V@mRU?tVNn6oAdfc25ag7H^GN$m(#G7DbL0(VaUG>fc{N6V0jh&TqkPBzk$ zW#}tma0r{(fecs#RE;DMBOCCVN;MfP<*$b2qN_J3yTucfZ*=row-)3#Tb_lCUq<&+ zA-#A0J~v{JK?mB{2?+>1kF%`??4U08sgsSE;NCn@Bs3w&)#dUwHQ`|_#%NNmBwFTj zPx~OL0vBj)*~k#Iv2y&s0y;Z7G~@$$r9tGsMw!<{s1Mp$FjF6lTIYFN(|zWStJbBv zc59BJDO=RX*RpLfNaUK84V-0Eh8_L84Gbuf0LXe=Kq86GYHy4m8!9B*IIi4w>R^uC zZcB)2iqM^C?mv>)@Q$dR@!Q}Kw3>gb!;MC138i!0jxA;{=2@JXl5zo~Gi&*$yHoYZ zTM3ie$d)OKO^?~XT&woCtnw=$*cjk+SM45sbd&KImkOAw^Tnn)O9b_s6<0tB&X`|N zhP7%8#;XKZUW~mC4bTJ~U>z97{3nSSYf+B2-~5nZuIb{x;XX-%{9MFaf;Qi9Y&{zu zPi;XL#)6hu+};kACG`EG4{?Z} z6vfcqx7Ahp!fQ6(-mCiVZq894u4zr3XiEs9R@Iz&`zQoC(EB)I!S;W4$eOxC=r11b z*c(37nOu7o#2Z5Z_m)DQ;_IDhucI@Ah-61U2g0$R0doT5+;l|JU|C-HA=olkYtmla37Z=;NwYA8eA4LqzGd)Xt#!9Y|AYM1S4rvg37klDf|axri!lxp8Y zulaJmlYqs8@v%MC2qNMSYf6cs2aGA>7)@x?NH&mhEr@HFc|2WK{l<+TRq7Qg<17OB zx)H`J6ux`IXJY?7Lk;SWssSkbe8-LuF{mJFQx`LXM8|5_uo%!unFbO~Q;FDNrH?Rz zw3=EoTA&jndE)(^jaYl{!BQd1E8W_aAE$EFd~o3pq3!&x>Ll7)LlZANfL?To{6L=B zTT35qXtI}}Y!qT2gs0dKI3$Tnd|^(j5cgK6J9f>iEoCDsh7?GB10a@WC6{A7FzK>O zi`Lx(YmLp^FDVQzuXA@kgaq_o5AWZC9LA$Bvo1+2MHOmTj{$lZLMvbc25CumoPd07 z$Be_kal_#?pxJYr};8USXYr*7% zmo>CZ;@tYNU-xj12A^`S+O|It=e@bC`*@1(fy9quL1SlQ46Y@{smwRe`mkG-N62|$ zj$-!{OfI)8kkQ~feFW2M?hz^Q3rd|TXyX(0IcSSi^5V!zXag`T{@UFCuv`l`zi8`+ zE@c(FyK;UY4&8&lTFxoG3T4#jQ}5@<`cCBKUkx{dE2d_JZCF><7ck^I&i$c;c+?{M z(*<~a5scJD`apF#>J->I<_=0VO_0$9^x?Wh{jr(+vw7+@^V*bp^Z#7RzxLKu7IGl+ z!0LPE9RcOH0mAy3cud2{0iaHq*q`Ho2J4Vc&i@<@RlU8VLnzy3+r#+Vk5&JQle9&& z4Ig!fPtoUaPBS$+11q$_-?|PxZEZ1qE%}Wf2Qk`lH}eR<<4k1IWlp((>}_#&>mYZ+ zZ>NPClNgwzovjlDf~Ha3iIsufvPE3;h8*wnXU~o!qi=v*{CyG*u*F?`=eTn5;;F{j zkFC~d@nnDN^{{(;48fft^3-{SjwBKQ+}s*zb=oj??Ks}*O*GD_>zLROg@+pgC&6IS zKO7A|(2zHoPX4>&0=I@I3oJufa6$(~YXfeT5-IH&F`5f~`lJ3G$$^`Fw|Al*Zw z(0xj9&aXP^(UPF$Vd)1#vas(0T*PF^6EuJtnMpQs3=YWkA*u!SUzh;Zs9lzk|AzWy z-A8>AJmUd~8$w6Qt#t(buC-y}#tBhiyZqd5aGtPCSQJC-sZIwX z#cE*P);FyX`ey)lvcYvU=v)Rw4Q*v6sI}hIy|+Kebrc;NZ@<=9%+)!4>#OGvS(W3s z9L!D{!3b8mwIP76Er>?d_hbOXw-~KQ1k_*_s9<^O8#Na+U^hb=Up#+qjf~IBevcK_U?6NJW95SNsgi&UBj7&PTvgm zA4`GrYQW@(2RK_s;6a+iH3|v|*`lTB{xDWr*tZ8VJ`}_mguOeecw~!u4H@b(7w~H0 zKD%YWGx0K$#AJS`T-8P(+Kk0e_5j zvT^S4Q@8bfiFWmA)IPP><3VK`9ES~trYeUo=jMJeANumtYl1qwY#&@i11x<8DpJQC z^FCcn8;|;-|Z4jHi!H*w1#5)2oxdDG?60JEm%N=y?Khn1T zF_FH{xDO%MI_RY&s*y;8vk`bS_Q0DBA%f+c3wk-R_zj74X8#-dlHjN-U!T+tEnfEG zkhaIY0#pVwNhc#0qONfsz)Wp%!D=Wl_3-jC8fwOHRI`6ZA<)%8nHgGI(1*Yea@Ml0CL{2)0S>HxC7kQ=r9QLHDOqnM#T2hEOhHWb#1JT z*D7{Z!10=2fyyTH04#haoes(>hv8_28{Y}1=h&Pr?=1_Os> zCefFO5K=ev`J%dw>8CCB`E%YILVLTV;Q}~ABWgKp-d<&LIug;lHSvnWa)8@4fUGb- zJ0HSb!-Xc921cyqMA6!-%~9dH&y+Y;A~n0fp6f4C268<^7Ue*PT@ z1J%R3v2cc1EJ=k)F!}E1f2me&z^#6m4@C>GOPVvl&hKu~Lt{f9VB{^vg6cIHux1+{ zT38>UgZ*xMrHxTc&mWoo4dHziAFPZLLErcqqIOB6GEbb<1f>*i+ z3Em@+pWB2>N}65DkV#vQNEz=T1iM%AWn=V#?W22SFSzJJcS;m+#~bzMUYJTu)Hehk zN-zj(VbsD9AQEQ z5>nmGNw=Zi9R^NZ8RshR)da`Drd@`v5<&Cy7E{~Py&aze!yd3ZEb_xCH$IB^;L6= zLEKcs&;wM#!dtNoWF^~B06>OFVuYn&___HiV!9UPNWfWE6Au{{Q3Yj9l4Qfn3CPJA zk~k~+w63F*HMdW40sKcWd^%<2Zocw}RER{5>aQrecdtXMO67;VB$0Jd1xd52;xD>i zm-ZeXOl(~0a?19{V5uu4wNF%1>(>v0FkpKpVMU}yh`fn7ckg=H)UXNO@ka2I)jQbA zl)@5$2yro|DldfB^l%-72*4A5h?7wYgjLMP_wr%^b7Vn-B>ypy1xgkrUKEr-YiASK zNFgwUC~RFpC&h;FU^*qP@Cf_1Ejk5O6QNg_`m6-1s2HsVy$?#}?HEb~!YFV#C7{OQ za1~OAv)Jevv8;^Thz!}lNAN>p&)2`sEubQ^q{pCnb1Rk&+IE|E5d1>n0;h|YyP$xIYI3A?KV343CS*hkFZ;Jz0O$qko`Kv48@=$_@ z0(j}UdJm$R!HtLFs2GOU1s~YH5$ffswH6x;GajITtVAlgk?~Vf+f>(p{N6g+CYWH- ziAGg_AF4RC$rykvN}ve}mImXaP70v-ghD`W3M_In$8L#&Vu{*=Afx~WshJa$2>}Q| zVJa65qWB{wrUTwJ4IVl%yvX^t$}S<<#UKXGTebZ#MOYE_@&Hy)#StM-A!YaWAR*2` z^%nE-i4ltp_d>YQ1d%`(wqRS=UX)xEyyz7GIb7&{R*&k-u*poMoTbAgTy9X>i{*og%V<Q=4T2DXx}=A#(Qqg4V|XH(($aKzvMJ~Hh=0TbZRJO9iS z=fD@25^HO4M@i?SyPt``LfsQ{9mr^J`}AyyCR#&2f!K-kuoS1o95g~!Ee8b4%@Lv! zCoh71gQiB%+{Kk<7&VrL#*y_keL5mZ2lq`Dq6?t}nnQ7iJ;!mWNuido9{>nI6t}Gf zBIb-X6j`Ug&J1#^0YbIhx%#$Bzg&1V&&u zxC$aP>bwHvchBhP9T-^R^G2AgseyNfsS`lb7lgGDX-!A2FRHg-&b)acINlrdjy;N< zpLVopR)p2Dz)P2CSi5NJeU`YPEutN^>&ZN>wY;*76|Xg6Wy+jqZT89|)kcVC*`hT) z`>|a>bnUL1e#KJFEIWhT(^U%;zrEDpyNqN$2paYxBvq9E$A9G+T)Sd!bBPJ?Z-f1R* zfBwM~r~bYvLBqJpx^k10?fB)*zH&`1g=e=g_E3a9vL`!v69ket-R^I3{y2{r?`Do; zPHLSIz9YzLt>N`Dzm5S{-e(I&P4_rzQf?Rp*ziT`b&8zrVas%``meZo;RsbF6AO!q zJPT}Yja6P}egryF|3T^%HP`x zRfG~z7obT1w zdbqwBFnWf@7bCY$Z8ab)$A65F#231ci+W|X!RusT0d8$KA*G)_L>_Su^~$e%+)nHZ z*M_%bFI$o15xU{fpD7mA8DohP=7&=&s`YT4GGG884Q35^s(M@jy;dIwx4tj9s^D`! z113Kc>Wp23Ln9;Er}l1aS9_yw*0Fw5-rv8|L`z`g!ccv3uTsRoClX+#PGR&6?8U{m z8mW?4hcJ$4O^B=G2ItJ59}2*w3^|^9$7ckv!nfVC&Lg^kzbIeBm>-AHe?oW$vZ97k zjRy@;WemslwysNwI;Zl%8;r70ARAOne4bv9T~$Njbxq;7!K;{bdHE9YYx++kG|aX& zs5k{()eaOAJP(@KLMyclx%|or^BKj7J@|A#;&#@{sXBw*a}P2j>_S;j_i}BdYeD#o z`Y+G)z=O9y-bYte%K~$ufO3xRjnjw{hBt^CR4Idy33BX>EiK~m8~>RBKRbxcC_}LR zytTDe+{2JzW{v!bEH_ymt=nj>J3)}6hj95UV(kz)4#(W4)*TF6Nt_UpV}O%-^0A7d zV_b9<#7@aA(mF_UJE+eM&OYL^lN-Vu_lk!|G^rAy0=u5n0+zNLB4!n7ySJvE(*FL4 z$^@Yh1kn&crU$pr*#^-l8l$X!`vWwO(WRqZxnIKjA17JYuIuWfSDFT{Vl1pOWB`xX zx+B=RLC^%mqZd$zXvOVK&XeC7A^4n!gx}qM9m0U(N5-6Wx`jTc>oMQxOK*Km%FtSmx}s*hQv1lE2@7aB{Y(xG47v-ldrm)uhO&3{#WKN zLO*uvnRx6a$SrZ8>Ygx)n^exFw$I#?BY^F_6w!VPdnAZ}nE z4KAekA6QG8Rw=(vf{w|aJ&^P>_spva($r&pgfe-B2k@(XS&prhDtq1A6UGD(FB6?6 zo;0K^s7%}(7>Zn~6rD2^fh;fH3qAb($Td0$Q$+}~=*0NeMP7!W6*&v|5#<6cL`zg= zo_x1_8XCq9v|y{skkDFyEv$)Q3R}jCNZ$c5^_D+Is38k2a@4R+Dn7X(YebNR52CV0@)4x` zZQ%Zq#!;X~6$Y}~TkoT;LIkNIpeX^Zumd(U41pj?cRz!71($V0i7fseSM3qTpb+uvT{eYZ715wBDnxmhj5iwVT(n^Q5LMAo&`KOO^Q?Mu0>TwJ9 zw)jq-25@kVvbVtxRRh=krtj(J++(Q?UW-S6&WDwQ9>SE-A<{A?h#k~)Q$vA^ZNFxI5 zY)br~2!owMLGMqAQ^os{RZO*}5>mmpVF?x;U|Q}F?;PuV*|b^miiB-?$EDLD-1Ad!gbta_RUV}bYGeQD5s8thvg<`G< z$Lkg$cn2zYcF>;zFGVZtqm8#Q`-t`ib$Bg!xIc_O^7XHS%s#cB)r04WT&lS<%a?2i zBCRy@(#7BrX|iH>H8UAsl>=UEKe-C%!wq1^sxWk-T+*L0{s@)MGQil*P1`-*UHGjX z)wf_6tS522$KuERfj%fBSWRvz!26~_6JQjfT`Zs%Y0lYDx1$RC)Xv~t-@bNtz);g( zpr|OM*_@C?C?Z6@1m5epLtjAeQht7M%xMULM{QWy9`ev=@>Nhw z6(-OE-o^533-pPorbtP3y|j3AFc2*zu#Ywn0*N+5H4rz}0Yq2^aebmz043N1;I3Wnmc2W!Yy|Uh8g0KI(~iScJH@w zwwAiU$*3|LiUJKALFH;(ryG<;qGC%Kn6Bd1HM{tt4xyaVjyf{w^&}Sm0<8iO)Y>J6 z{ywOogT5pS);htB7{*hbZzp?FvjfQ<2?kL9zK`GI+50}y@W+wM0U5u!)9Z*tXTn%4 zSnssP9s~I!q6NM*B!f~4xHg*p019K!*oWt+&Jfdv&<#4{G!6i)s%P+@NXKFUpA#c< z6jIIYu#rR^wVCxXNL*-c9 zs>RSG+6ddd6nfRvK87}}AVh4+5TRkrmpk<8jBTy$*F9?PzqlUJsVea`D>I)-1d(22;*L>x1%PYBmD+lcWb~}>#nA5o zjr~~5dvM6ooMQUBYov((nuNLmwHgR%L|#YyP=a8a#Lcjg>m?)(5?n!ZDRACYa9=U+ zLJ(3`?SBtiz%I^2cy$drw1-GnA6pDf2z6xX0f^X?u^>yu59`wyOo9;LWy>fg%6>{8 zP)4KGz_4@11lSzUp#)Sbh-gUpu_n$d(N3r?X)tT>eF2a-Hh}+|;>q~zv=ePg9b@>m zjd1iqxaml={Vz2wh)YB-K9oFL*GyXoI57=b2MN!S)ddnJhy))CNB1+7fE=iLwS%j9 zgG$>nXO4JH()2geCrpzXt6qPy-2QahsEOnstlU1b=j2mr8k)^$VSdT{3;0)~&4 zStYj>CwI#S?}KO&j@ZP`#tDRl&Y1m#PJx9e6VXQ?fmg+`k42tdjphyl0FB3ZAWiT4 zQiDd;M4~aD;6otU?~y*>t!TCeh4OSBaY^sN{Kx6NFX=*bPG}fyI!vcN6})f7DIYv| zoni;FhltpN0RSjSeaOIk74#U@|6uTeTuWUobaqL?4A_@40x)zj5f(wCYpCvR4R6xy zFi~CTng?S7->=5XAhG%s&0z!|rC`c*K*V>!*A!}y1Vo}g{%BQX50W<$4T)71LAwL# z?kNjM79%3XrDxL(Ksh%sQqk806~8sePHE#Pz(RycosyJSv8A6OUvjlZ$S;xKa4!GQ zzWXOZx4?v~P-?{ka4Q7w2{4fxA~GS!<@TchK^AL}z!aKC0xMZ5p4PYmWEWDX>#hb+ zN3%@Cvu{{iVqJwOBBa3=2w}yk!3n+Qa1}R@l5E6NzfLmHg1Z%<)~}Rnd^lmK?$vgb z4ZU5d*Sv`iDVbdychZDe@Ue*=ftp!?_>&H!GM=vK*qi6LN^FQ^*2Kcb9-xLaq0re* z3J!eNFGp51aLBx^!|`=1mRGp7oi~`Xev;f^BTu8cb3_7 zd=&k~&qYf##h~xMWRX7YN%X|G&j*Ypz*t9&7z)7zMd01?fb9hc+8^u1-9hADuVC)B`PyG!XijnGZ`lx zl&2*^A+ye8qR?gzl}^?da%zWSE+SKSEIwUwnm(bgC|3!he-*4^Fu`HQ@UnkJL+pB; zu_3&6N6v?6mtEFBK8_>B4%p}*7@!}kNpllPj`Oa|98V;119UY0p$UNWfno$b(@-Gs zakv^+{CJ(x4#Df^$Qlwr07yF9dW>uy@&R!;k(cAiaBP}!YCP$CXFB_nlg&vg2bL+? z($Z3YWc81aKT|{*0&y*14<1>M0p(PI72Z=L)!a(BS^D!C9Dbk0cLvNp*?@y-9-EKo|Ky&avwqb`X-M*s^##VM#`Kr(k9)@NB^;YMO5B=A z_K8D8ZWHr`7|mF4+^U+HR zcmLOU&1te#xWKXS+nR6(RB?bY-c)8!ST#UzVDrle)g4p-Gb#+7uWxf7pJ_DvReUw4 z3f2YqNf~a<-aSMO(oDBh-w1IvxcEzO6;x4zKV3^M7JfqQa|DL4B0jO$T{>-K5X@pq zuh*?z3loWwFlLdUE5aybz!$+DoW6wo)$Vn3{5WthmG|t_p8aFTN|b2&7Ny%*R?7G= z$P1=QI|zG2Q5L+%kB_2l|`6VibbZhc`@WTU@$Lw zC?HB0e8KBzZq_|*Wakb+4q3DrM9R6bKlULXwLqC;5WVekceKxpHNrJ`W0=wM{A>h3 z&}ezM_3@q`ug0Nw0)+5-sFL*(b;TYBXC3+x{1CysohC4YqvteC=kmgn2=XDNIq;;K zqXhcv{sgTMxpXLtwwh*BL*aJiLCGiMi>}XkBKXU%ORwY9sK84{883YtkAJ70#-%bk>*H~oH4)q0_m<{( z90Qwcm@zXfzZTUjM6}_3x96_5GE17f!`-GOFNYok>GJIkukPJNOt8Q^K%y-hu9^>^ z9$1%bqvCI29)N7R({oZO)J6y&HCT^K#hLx#{UiZ6bXyTzwYNxLpEF(Z?8@~30HLQn z*8s!ez$4f#rH;`FXaHcNbJPNJEw*LO)FP`M5uLT!7L*I|E&GiDL69C0b)(hqE!_+^ zp5&=g6OB@YGWO~1+4PT};3Q5k=5QoE@*Z`@!f0gqI-ru7lf${Xar2vBR*R5mCmy2& z2bKnEWo!(j`EgVaLMWPciudbFr#K)JTRXVP_D^Gqm*7+)Y!{;Hd-~PsOHfnk$grI| zjhW?zWPUyBE`#vNs?cMuZ6sfVOTTFq=2C#T_G+5IOn&j1=F?>_fr#f1t|AFiP$)=n z9vx`PM(N-Jz-q_;r}gqB0B_+ z+F%&6&y905T>%V7#Yu{{NkxP#5n`my-ec2bsb)bVy{PGd&N$&ObSfxf2WN$57NYE> ztf3Kx%FI-i6OkrH;D{mP4n@&!Kgz(7<6Ahgalx8&AvLMa*!I2v+phxr=pIaPrMODK zlTWu8*Dx!Qv>bbnUAnZN_-{~OfY`?Mwj9lT4vqlQ+@;_){VKV2_Uq4E{NsP) zBz6n_&p$e&jE()De{^#XPT~Lf>%VW}_-_gRTLO;%mf*i7fMNX47W}sa|DA&WE&|0! z|Jj27mf*ir@ZUxF&lb=M{AUaPTY~>i!T;4oXz%O&iN~A*3xJ9i)Cq#><$H7vuf5Sf z`J*cG*U|rl(`aeVkRVk;Xc|hXLy;NPeN?JE|8_(d8)i*eQvcz^Zer%QL1i6F4f%!^96%gx52Etk7d&RDY&^_{>U`nhegT z;i#IhR+lDOK2@6^Aa(1xQ+EZGBFr z4Lrj1&MLX{JS@vD^QNv}&4@{|tWq-)x2*cHMt`Zyo~qm7fzgNxS($N3)K7jhXRh#O z4i{S2Gw!?!J5%m2Im@y^(#z7aEMdlvb;)0N{l{uO&han1o`|>OJR$ah zh3S$tY3?2_wY^N985%zp=ZQ&Pl=1i9z`^lGIqDb2mf1DOmzTd@t+uc7cV^T?V7{Vn zK>MSCZ373IWey6lc&{n7a&8I=d7i44tp0Gl+@G&KpZl$QwK((2wGXTG)zzM*D<=6K zd{*9jV9B6koAFqE>QL_sf3ZU$Qi*)b_xlZvgajULxNOINTIOZ*ajB2uOFx#gy?reFSk>&lp?eAwB@5Rh$M)$sEiG%c9U{Qk2~W~Ygk%{j8-;^u|o zO%Hyl;ab*jIo75zOLALle z4*e4Mz4*pX+^=2w{h8h>&i)BAFL=V$nzQ@nc=m{&L~~P2%IWQ`#=aV-97l!}XSqLi z)(?)$_ntJf3XDJb@W9Xe_-N$cRqYSrdy=vHX8(@p)6HC_!!qWq8D1HJ0jtJN1kWmt z$Sv%(=u;gzb4;(dL&+O0(+0Y_ul&1cIe$||y~xQw{IKjxFJ zbwBUYyjO}B^9p}SS3KV>oir`jsk!BYPo#`ZmFuQL6EDFvCf|R+)cfBD%UBWV61Zky zfkDGN)sa}BrAxEc$Uoo)nn-y_;9$WGLa++)ro3 zNWxm;nbI5Es@}qWla>i*&Wo`NUPOFO=luSY{#-u~@>`i+j=1LE>sy_>`FFOgx*jYg zvCC8boYVNEw0o7V;_{*fMqk-b1Eaal@0!g0pO)#Z@z4ExWK};+RsMXA$=ps)vy$fU z#>=alyBY(EOZc+#>W0;L`+i=S{+|nSM&Y@X`u%C?bv@0?%+nt>PAeaJaecwu{ArDS zCXI|fi{>L4RzDn9?zx{!{^FK}frk8@quevrsOCrX*bR8`_A+5oonekyXM+*HGjrPRnK^ql}h3o^%0k|5|ZD)|L?zk zo+;Dzda;}Cvt5#xtKpq==L#mDd8_&cdv&DsfkF+RVW-EBC)>vTa{;+!-S?OC{n=ZX zj}2D(yedm`va0TZ;X{S=NtTz#SQ*#$+~SDOD}Vf)^uISs>36)*Kh+ZcXBRdp;X{x6 zyDQq=KQWrK;!U*mk(65egH;?+^Yg~o@t^%+O?rQoH?{VY7`RVIDkt1uRb<_ZbMj6( zbVi)tp8A}L&GJ?J{$`JE|9e6Lx>xP)=i+1Wjx|+yOOJ359zB|xZ@Q`Jw#NOKW2p}` z;6MHlwV(s>;^K$Amv{U$<++oZ~OrPVm<(k`TR@K;Z^ zG6+j=0W*0imtT~VrF2BBg3F%AyF+ZR$zZFr4Ch-9E8h3m{@t~iZZcaS^!@4iFE#&; z)vL34io2g-8nCKHN+(%2N}_|!W?$@ic5JhSBo8w#e%IYQx`|sJsU_A{qaN-?9dXe z#$3zEwx+6q5s^@pjlrcG@U`^U(zZ_<7S1k76RquWTlVWO_zG2tJ$oD0TU$I_M?BmO z-Ch@Vt+_ADz_)YHbNJBwTu@=ZS6qcdRcL!2*WZTUKm1JonV+{s+T0*)NrBWnv*R}% zaFkE3zN(w|sC&GE|IGUN{%cP9y=;0ad@?4=UVnz*yZBc%SPSv+sD+)6EZIJoEK|unxMyu@ zM_^)c+ab~ATB6HqK#6?aGZZ;oj*}1?8hX}^)tmy=vt*EpP2bLg&ZED^<4V zn#{lgTUPjyhYG>hWxjvD63>t8dge4cP4xb?O5F47v3v!tM_8LpPJOBG3&tAYD^$gz zD6801qD#s&uEzS8>1X*G8g#bbK9F-*17FoPSMT#xiY~u8!#??jc495QTp=wnIzExx zz$kk6cYuT9MWaN8n$!8dd83bp)r@?%2G_0n{sjKhf1bs$^L*}Uyb|+^xIJ^_&v$n^ zJk8*pJ!hwVa`HbuD=xmvFkOCiZ)RiP{7Lnbb1k>=uZl0;vf`SJySr|Bdb*^#R)@2} zg(vgRmFz+BQawj8u;vxa%V)Hw`#ch3M|SHU3aN z=F8{rk<6L#3qQ`~6MoAnU!Q3x(YMgj*K4hqYfRPDvA3G%k!jfc=oiM)Uz=-x?;Z&( zcG&f>u4m>x$QCE+CU%@>Tz&q2M~VK8u<-EAyu2i8to-nqZcs#paCq6q#g}{k%->N6 z_c+{vSZciIOQ6gLw;z#|y#2P}jPLf>RIB&k4f*r+zFu$$6?_HvtQ}n4a*c7?h0fP^ zyS_1Q!Chpk$xSQQ7tIplZ$!AnWb#y$mzPMX>m)*)!~?gY)+Y08hgW`dSXfv_M#g&O zE6GqktI zox>%0b2|)?bXi(PUS6;NnpMlCdw+vblP)!vl$92{Ia1a$1k57MINk{02E_VrQd~jlE;|Cu&@BeoU^JvNNu!;WE z`8cCc;_x_wSHk*fQ-oU0>-X>fsQF!1^kA1I=iPL+coA1_DPwBsR-~~Ozj`se%tZRFD@b1Y*yD)@N+u)F+@Vy7uT_i-aSgI zPjb$Qlw_7iui{ip`5M9S%hzBXSylJz@j_KN37i%sUwV@4+z=Qn8o-`{w5 z7X8lb`9;c4;o@+Q$ivj{E2uyBrxh_i_QRsDUfIsSb9+*AvTug?w7aif+ezyPhiZ#1 zho9j5Gt2h!p0%E6w^_#Jm+dh(yK3t!$!5_q&3u!(1HNBt)E_SS^+HI6WhlN%&6m!b zm3?CFi9+cWjxDW`%0DAy>2n5h-#=ac8J9jH^C3ftI^Ny=1<$&=hddLsg*9fUzdCOB zSYvI_LAj_mZ1$C3FZ_P5*0go2-{z6&fyHf7CJj}^L{^kSlj8sGKW>!XW3+f5JNEmB zzDrJiVs!|v3TG`q(o`7mZ?N+-(teWlk%E?g9!J#oA`0gYdLZ~Yjwsvv(Q`@>m?bhPr+y5W`p^d%SuoPbzX=##=G_o6O!OWkM za!*uT0s{|E791Qr=$S8nT1Mj8qcfc_;$xFCHPQ8}QhprmmOVdW*}taY3axtcd-wPF zzE5KL(WqBu*=u0EmtEb9Q*c(ozVLr|5<|PpYy7s#gQMQ;4rhGb&^6NPbFxg;cm31D zA=6MDUqrQS9nh1Ga4X=}Q4=H zo0_3Saqv-Y1U0Gp+PTdpCq1zJ8K#z!JkJs5H7XoW544tAQQO;_WNBq(aVw79k%qG} zIGXoy=Aa7M z82t+^C;P_NsL8B&_2Qg`0n%D{Z<|ckQO2#L>R@3;ADshYTUim>{K5XNxlNb9%z&%hj2@{K^U?5* zbfm?)HKRD9_}FHXY$dvMhHKQlau7RYO*jb#GNyL@N&o5}9%VRV0`F-k(To#!5+8AF zpHXhJ`w9uw*c8JW{x0D;QN0Vd-HfI*R~BnUNz_$M8u`15F%+B67Q`k14rU(Y`rm3U=t zbQNXY+m&}eKKZQoL^0wi7Tp8BtNb#p!tY!94KH0ff~J-j`NHbF^YJALcY_&wG}$=G z5=f>Y(~fM^RD>e|l+TtGn=P6~GN@NdyR{=+G7t8dFn@x&8KBB(V%_oKHb zJ2~#6e6f04Nd=d-Gn&_1dCgq4%H3P1V%LQ3TsFT0lZ>L6FJY7)n_NG1``?|_e8R4$ zq0B~O)(&k9s-LK~+aVp_-USm(M04BcXJ%7>5cBMd8=l7HuJHhxOhEJ1kN`BNZY9T+ zq0m0(C<1ns35{n0CG7!Qar|<4M4cy77w4UdJbOJqK;*w@=65x! z`n-H^^a67B1+>%ImfpXq4jqRXUS@MCqa0F=-Y4d56LPB@~8|DWg5q{WfM3?alS>Y@MupM)xGRzzwKY%1*^OZtS|sJ zY%hTx*8m`4Tdi426hI7xu-B>t->^PlasB$EZMDnaHQEl+;=&!qLG~Qz<+~vj5OWj?maRg4wTO*B)En@n9@toO;tVC(uu&g>HX6}e2 zHphiM#N)|re;ohf0s|jFTbWC8Zz_!Oq$3(=aI#200V{E#XC83~6={ItqU7ZV)MaYL z%9AZ=nIH(LyP2DE!_2!-t=_l{G?0K-ND4n6UnjDfuhAwEh>#ssT)6b&iNrl`v%4`%C&Rp)#|Dq7JV1$W%~Aki*E8>^BC|an#n|=nm3C! zUsc0y?zGfE0dtnh2zR~KI&A+GXPsh%6~I7J-4!N7f#%05qruHeMMVW&)=LET76KXu zDvGrf>&92_sb4fik=mQGDAZ!Zn#$xZRLqDRqUJ;SmwS5)ypitD-8A~~NhyVKqkY5C6@ZD2 zQjM8(3)iVNx!HmJZOlsdJieU?vlhqYEoU@&Bl~(c+`6)BUF>{aQ+_5-w@g1vtLbW> z+U_>5jY``XopW3ml~t;*nu5ZQxJocnkrPFe44L&HaqXkHJYSECixc(mp^QAYCZ|`t zq4$Ld|9YRK5tE@IKxIvta+C}vyZR|c0D##To}ATf<#y;0LYBTE#RwR>YdC#n$!_05 z>888gRd0XSKmrGX%2+nPyIi7XUGjj0HHc9sg3xtBdoZIw|GUSZ(FeG5$bWarth~bWzx*B(9Q_Z!_q!q%P`Dc? zp9UP7mtFuH-I=|9|L;~*plf=~#CZ>)D6#R^E1c_<*+~GjxCbAsLpBc5&}64dA3c(0kdnzN0e5X>BKq4%c|n!__RhyhQ~AU3aJG>Gdyy0SLEZUzMd%&{_e zedU%a^Gla<7hEjg!lv^CR4$4D9^`7uXK5*gDSuvL-e@Q^ud>rpR6|YeYi_Rl7NhIe z%NH1Idn~6l=uutbQ8c`Gk>N;+!H74(FDKBqtYEBK*T6%}rrcE`+y%cqdJ}vSP|V)1 z8+(@JnbE1C;Qs(c!El8s*MQ+#HvhnqXE+g(53uMG2Nk}(Z((PMC%1BF8Td5(EDPGw zz%}8~OuPQkP9l-jzxk@|>fVnXYo0#l20}~94>120v?Inth$>b0D~;JMoAOQ{BVGqH z08K8wtsrRbf{?l7dk7<{#dU4)y*l>vyJ?} z)ESY^@XlHupUXy|t(fd*g|SuMA>-|TUC)0Ip}d2^egd;jt(Tm~eY1*U<=*j#;5Qfm z8WhK6&iouvs0tcdT9t?{_>KysZcn)bTCv$TkfU$ z-Z_olI9YuzA8aPnPPC*Mez(7TE(7_{6E^B>F zi2v&CZ=2S#Tw{(*uao6^jLka@-u!348<;h-RK)}9Rq#!~;{|#B4Sdsk>ZaJK?`GHEL0s-j`kJF<`U!ZJ5G>|fsIh!11_KFJbYqdBP`UxP zHcPbviu1pyLB?{){pyOo;EZIjs$_6gMqhRm)cKo&-ehqv?dCyfvV2r?^32jRV>!$ZSuJssSwWMK18JxP3tP z6EJSIr=x4uB~oT(uPn~=&w9#uB9wwvBFNWd0#+}aYb6@!W(BgSt-W1XN{TWcK9E)FVfx7?oC)tuP@t%3|c>*4d)7om5R_vgG zLqlp~fpvxxfdN_?N|z3#o)^GW6))#vV|#lIGMTIx*SORGU(#*m95k}-qvgibaITUf zy?p~zSr?nhRPWuS)&HT*gUd!HMwt?F^p&AeVmy?EMIgU>T8;#1+|6$DHCxE=?7Dpl zSfl%I=NOm`3yE2O<^dm6f_3O%@mi}NsCi7C;Gm;}3qjC~sreBF% z473p^!k+iH_1iul9a?^!G{c<_GHF{X`*aJL{J=?|@iz+0D}QmC#+|6a$83+p8;QDy zzkg-!V$VEu1}v^If6bp`i39*HT={^Hrxgnrd_7J47tKuFXHe@JB9+kh@;>(i4LN^~ zb!Slh6)9oDa8WCGB}=Px{DO|l+m-|_#)oj=nf@8LPZ2hVS@zU)uAB=dix2WW3rGoA z;>T=DHx@yq0Toct0ygOLXE_alSvubR|Hgo+Q4$Fi#!I5fW+}*RuYjc&g|7-XZ&>06tN+wFx6$S3m?F>OL5$1Rxi=2uXl>ucxwyAn;=KTL zOdB@g2H$`y5Wv`_&mR z`b6MT>O@?Bo8wZ(C5J)^KfxMv?t3zxG%+E) z-un(H{JvH$WlXW4$tq+BzsM;;7g=pWXp;Y@Af< zL6P0G91Y?=Kg%P$EG*q~nFs~TV!E2;`hhbJR5Z{DlnD)iTMIPiv?%CX*pRe4*i*)n zI@md5Y%$PQlw|z6~@9$kWI)7t*AI4`s39yDWyyD9$e%cpdR!RIPtsxqeYuGm9$@C zY>4Eg4$6#w*jy@U*uC`PNK+$_ZRi5S%xB?|#z|dX8^IMb2}~gR&w6OOzDQcr?3cQ z&vtVq?mr}E#u-92?{oD%zMjsF@0tSmwK89y+WzyVAokB}x12oS(rf}ny4DRc)hg4Rg>rCf|QA-;v=F}+T4Y;yTm#=PrqS1R0 zmhrxz;hQcTW<1uvI71_9ObMXO@YS%3hZA}l3b}IR6WpZmebR0>$7ZhuEIAy5)Tui% zp(S*@Y^dP?)$C49$-7Vf6R4dfL?rk3-xZtYMz{PEr+sgn`3X2-Nc7=CBgI{vQpaCq z^8RN|pKZsmJyStSsHtc8QeaBKb957y$gaY7y$j*@bM0xSb+?XE2`=-0KwHnhLE9R0 zNw`zmlb1u2X4;HGu{0isF_vSs_w;&5KKCz&QQ6;w69G7l30kq-aWb*Gx#C^y^Ybe2 z915>ZoVbyWFdT<*M@d+K@W}l^Y^{F~+c6-1Klyx|O$%>XPwdfQ?OoQY0S()jX^aSN z9-m}1c-2ZXKgpASJCa0Zn9F}vuKkW)<7@EO8w|1aR<5`wV};axE(FNpSYjG9a6%9U zjd9SsaA)xC*>ghU%7^4UbXesF`{#vT+=cqdd%l01n|Kd#!PB2ad>K_IYDUPuH!rXs zIx~ngly3r?<$r;m2%;rSdEuPdqmzC1;z1eiEIH0Or4xqVud!6N4~{eC1v`T5jkKvkihY1tE2dVqC-sFcx zWqy!oNV= ziUwMPxD731YusiU10+A{Vs_q~Fz?I|f&xpY*M$e6eM=tdll0)|}g^vHU++=>8L*-p=${B(A|L`!GUCo4+qY(odF1ErL z2l?%z(@NBdFl(VX?@QrZuQQLrcOkIri7=oHO9oRJ2|GWX7VxfE)QTk3 zM8kc7CM0;Z3{wCA1Lp|@J(1@p-bnl?uBWGE@R1$V?-32IHB$S^VMo>`oURNs#;Mt3 zU1yZD8wyLo&GeeX3&d_7+^KuShD^cCDq9%kd%SA}<66w2vlhCgNI#M;oAhhPpo<+bj zA_JZwqeCGlJQYfOg(!!3$_aPn!dpzU1QeYx2lpg#_zq*LC5V;daFc* zAug(WM87)~%#jh?JVV@w2g}7iKNO#H`vWRIe3(;8!M^^lkf%evkaqnTuFdN(Gta+IT%>`7?}rVI zi&-{LbBLhL2v7H`xfXpU`vah2rCn%@gO^}bY{@fp=^>{KJ(DddsV;j8HtwFyl9Rg5 zxAdknNgH={dQ;@Y7uC+3HwVINIv zz(6$a7lT7Y1MhQia3sdXZIHvFBi_*-U-HkNKh>OhM$(6I_;?CK45$kF{Mm?D(!WWn zeAmnBHgg52(mvK>C`^H)g z=hCSoX6Ex;_ewLHD47Rb(ZA<32$#G48g}a&kGXP9lQh=76(p!Tft9!C5GuayhL*+$ zX%E$QhvPxK7mOm27xq9E0!a???hn*60yDhkN|5|EqYXTVrwJKzmm1#OVbF*+JMmtI z@gsfRY;e$R0m~NV&J)-NaFEMT16a~!AB^%Gev>l<%5 zxD78hDNRJX1RFXJNcg4{;$yK;HZmOw8BCiN+|tl|N>aPv;RRE9ASENuS18%b6aB#T zwCep7HRrOt}s z{16Q+VZ6*4N${Yn*axk(>e+E|FNieoQi;*$;o;%d4i=~r^KzTx^!4?H1O;233tjdW zLA_)x9JQAKgNsKFRD@NoA7oBA6C?U&?~hYCbmkG+S`<857tYC%>ONTz~ zNjV?=M{1B%GvGu7w+Qc}h)&WKZ>& zQxMHw1?P`^R2046Ws=_U^YfXBsV2G;aXpni61A-z&g`8%-_%Qzdvo_?Aw9FA-}Y17 z`9s)GIqsI&T4bCni-FQ;W~PWLu1JLeALlRXS&M^&`vj{WewpI_wnrY*vle}}ULpj4 zHEg{-%eLw6>T5_9)@#H~ z7~0g@s-aqw*iPbB?S!OFE;)~+42?GF?)YoL>jt5uI7>NtvpGM-h2BJWIT2Pq5!f`& z8>&Hjt+rYU6a;WZ+DA+|OywSSSwbNHqk$%8_`-9{zg@sW)un<@zIr4o{8znV5#5 zR)Pli+8F%&If{$7pu!pYFLs2vIu8cePYqWwMhl$6Cl8;Yy?7#TYdLmQt$^qL4vE%t z0^JYG9&)s2q)#YHr?hJrTMbMU1ZTeCnfxU@Rp>I#Km1jiy_2}u@Fag0;j`)+(VWB5 z%_^iP$%*y>j2Ct~!t~CR$caPMWoV$L`>p<|fqnr<@h-4iZ~fW6^cL+Z)W#ZE&TdKq zJ`+<GX26Ms&|GO)&~TjLp!m{02xd##C1U&26QDuU zuQt0M33;%RAm)K`GAxxwd@s6vNvckXSw##N{c{Y*fsIK;bLRd78#H>KUL4YM?!PoN z(~-6@EinFk>0GkVus40od!jNWJ1lsr+rTD8XWj+AXY9wUCgN}IW0MY`WrQvuanrM2 z5r2w;BF76AH0>?as~WGowv33nWTn_NIg*)88)(|W9q<)bPem#@b)A_gGLQ_l700GHemiM{xj^!7!f*TsW>JJ(+9r&e}p zZ%!-WyD7ZbQVc_GW)(yP{HJ?*O6*rQad3b2&8Yd%pmT}A9GvQLUjC_Gt62Zc?^%uw zwspi{3yc>Vi%uC1x`~NDYY-iKUfMpK6zilAn>FI!h+7(y&I|C|?;MlfJNHNcrvmA{ zX8%N{sP+DET?5@mudZjuEF#5Fy6SeNTg8MaM!l?}chW1BoJYJT5~Qet$do?8Lmw*` zEcB*^3tE-ChJSj*Si(P{LdWC3}110 zQQQ3Ll6H3^|A~HYmuTFHak-eIsI%o_pzZcIXLwF*cVOpxw%>b8L!RdG)*}-gywdn1 z*d^qUJ~?t{PjIYB?-A)?Pzurd-&r~af@AeygmuDj1{gK%6@u~!2kds|51%K(HaOV6-edI&5_dH&g)r}bf z2n>$DQb!gs++U)B;a3cgnzma&2k9`;rut`(X7$C?xZ4exUf|`q-T6WPyg*fJnZ(w8 zn(oVkxoMwn5`T7+zxAZ~?3qk{Ytr}Duv0}gsm+~6OERa3(W_jP5BybuVDmRpv=*t# zU|MJJ&VXXc@7B5z{psvFBFlSgCD4guXvp8+|Irqr`D%tM@W9X@dct!_8Wl+Lgvdm| z)52EJ6-HJh+4B_-(w5u&eI42;F+)q3otw_t(FoQwr67imRugXway)#-%~i(|zluJjQz=?Z1lMJIiSTqZpWC%RPWGLdIB zNP}XeP+~}7VkkmNVI>TIy1DIx2b)jG83T?H-tDE6w9RtUu60C)!nMvY!FlF2!vwe4 zTx7Q26!$9$vcA%-2R|pB#TJrU7@t#%Nx$Y)Crq2Sud-ap!mzPh)M-=pjSmh!hk>01 zCa=f_v)GKxf|jzMKVPlP^(tuDJba*~(X=g1$3NT|9Yoi8Ciq+pJ&D zwfH-5tZdO7C($XjBN03&KkZX!U0s_#R+s)ZJL6@Y#g!ycyQvl>tqOMMF@IVsY0}re zBgAtEcMf-j`DzjVu@o29SuC4xdldD9KoJrXi>j$n$Mhi@b}tcaEiG=fMuEi*XnTm; z?6$zL0x)?Z%&_#^cZ1>vTF_#XT)+ZjU<26InWh-mK<4OcPIWx14^yk`xuF!=xa`OV zi>EVYd5xIl=F!ct-!fc6D1)E-+Eu?_OZICw3OV%^!L-q;+D_+?Zd-LiS3TP`EQiYX zY^4$De&9XI&rmVU3GEce`qkg$?|-|4G;vmOraC)nTC5H*t_-m68(1ewFNqy~8tZrr z&UgEX&WPcfQfCz&YIT9N#7&$s2j}c72!O>yl8(DbRG{(@EGKoSPHG>3Mg2!Zr#X*xbp(q zcCD*Wp{3Ts5xxa1Fesc5?@n_$XGlQ!B=|gSr z-e%RE>Qq#{^~K6*u+Cy)qR=J6*7OL%#$QPeM=Es+29JCF=pXAf6=GpYHGp{nZv`R~ zA+o7NP4^`Kfd!c2$`6eo8;v_sTl9QRv+Iq3C$w`>BWF#*s@8IKU=w~#7O$O5Jj#$a zOgsUBHlAcSpz3z%R*#(7K!$2zEX;fJVE2Y=q~lwDFvwhL+lGE;m8goL79P49i8dTN zjQdd_r_)@+43+)msa8Ys@)6wkC`6Uc+dL+?_XxkN7D=_BB;T^zRQ}P~OmbzIh_BUf zpof{JNhJAA$FR!;Y4~yQR4A>xlqL_+s9QM@vyOB1G50qcL7LZk$ed1^&RG&Ych6d2z-4Jp~pqGg8PjNZ#{dz`4# z%Zhgw10A=#UCHikxTsAuzxC-(&i>A3 zZ}|@V!J)=?!MyVawa0J#)_vo9bEl?EnksE*vGiA0x4$Xic918~t|Zb025i?`a^LvFjwE>)p2wLGg#5%on!9 z5wbIc;>fdKAntB@NDL=%pb^I{OO#e$?Mf7zGYmVZxl;Cv9dw*=p7%Yhx=oKpaNpe} z!nJl#r*|yN+AdRzOLBQd^UBjxRi!7BuFI}&il%gpuWCpwKbxUmWnN0ZK(cRHJ)uhP z((Z(}!oJPVU?{EBNfEcvd@wVoRj;k8a*`|AUVkF&B_^HGE1K@lh*qF|GvX2h(PD zh-bu=IQ^l6>Y`Caxz1?lw8r+xMBF96LbSH%HkkJYKmoyqDzN-#Vn8Y-3Iq@WY~A0>;Bwjd(vi+ zer2e%th8wVRrv^>yE`1VZ}Hystd6HTzhnMCdjLMZM63Q3D@_NRV=3*m1a9}_p=V{& zZ}fF)Pu_Kb)zkhs?jMcTvhUo=CFy9JBErM|qOM1@nTk(smXgss&^s>)CxQ~TFO2nM zqujBa`8s+B_y73+332@7^|vT}>madH`=^Z54keHRLve(1(~V?qZ^fX`Lkc+>Iwg=J3 zvYD9(t%Y1f26qLw8C2PoU*R^Zv%fGxKRU#ybh*;93@JHPRPGfmIl@_HE5v7i8f-sq8 zWQ~6{W+8#cj-}_xkkcVXw%?EP!^^>2XC4~kMqSa_=(I{ed^9>5c?wtj*t&4f+NQCR?{z}N=kGTsMi z*+jBP6}2E>l-_9GCm3gLK*xS+DSfi_j6>gOdzNi5i>VZIlkL~z*?o>qhp?{#1P;UQHek92-`%X(P6SGoWiaA>yRQ zR6{N|;X_E%Rj8e_udjHzm=n_l!RTXZY>1;s=K_;Z6M+84jDoqLOjwiIlnm2s1;o~T zN|Y_b+HU1w7QLJOr@~Ql;}fB_PC1$UDu=@2HH1mR^VAgG;D$PUL5k# zd=?vC6(wwR8d8sJ=YMv7&9K5)LqkIue%~0Cb!yo- z!dk9KFvc%eeg%MX%x6hnP|C+pJA|}Pz%4--FzK(PI&WO1{c#6eb%x53d}ZqcJ37NT z4Fx7Wc64R}FDLb4pD^*V2kz)-9!Jpa3>1w_?*t?47`(c51>&3SIJ9DFE zf#Dv0e)iqwoF4yi-i0ZjuMLvhU_nV`R{v1OI8#n4leXyAok9!~+mZU>RlOby850>d zwpFh^&qWvzF?Y70`5PPz^b$Ci5?@@oJfAJJXEkZ;mX`DQ}O2+Cd$_QKR`@>yWX-be@u2c0l7NJb^j0aOx z%_oo0C$#Ci;MKX{@{7&oMJf0TrnoSe@=e9#S5K8o#%)uzm6xR)&UP!#`z0OnU|J)Y zW7&-{__h)hcU(ciZ`6cPWhr$^y&TKG4o13qWz2%UX z$?jz*jeX=Fx@YbvEj87tWHmLB^q!+Av@b=+`EQP*oR`5i2G{<%GO+A;t%YX? zLjibDE|?ipJJEh)611*oX>?AAF}QOSUCAx%s{Hjo{gl0lx^HQ$Ax-&AvHh7ot%;|N z_S7X)buLtMVL`!AG*TM;*<&w*oF_UTF6C}qgz-5d{K+}A3^VK0;ijb*J&1Yj@)&ru zrsuHQkIu>9N&xa84xQihAvn%lKa3VTHqLV0yL{m-Aqj~X_(O4;5hVg4DYSs_PA^H~ zUaP-g0wR);wjt8!Xu|mi2{qk8j#JF> zC^mgGeatMQ6l_s)cO zZO}_Qy(rx2tDE)TWmWWQTp=Y39~ra^kBiY6Z%f-PG`_-~xar|1YqOm)H-S|E?YpYL z=QMz>Qx@+emkBNibUg58rT4bKZuiFMaBPs11dCJp(Q2~Z`0G2{h?tY+dnjM(c4?$m zszhj(mAZ^ePmM*;U(h>UCW0-yzU^?DY~iUQjGgjrTch{20OY&~0+df|V>^9Tuz*X-`XB2p1ty zv$R8X=hG09NL^pd>OB`uVf1N#Ko-Ryxcz;{n*T=N9 z?~hhG9vnrQ2BbL&QZ@y8OK(20+yBa2GT>ZB4Y_sDO7kG;C1v$mjmRX+E3{17Ky2f0 zuE%L8l&8W#a&V2)z~kWhimKCS%}T`j5L>kLik&3{krIjvrnwm7@awOP{5^t$2G@bT zt}%rn?_$r(-m4`JeOVERQo>bh89vPxf3bK;+Bgj0!p!Nx?2(@;$e~kY=Fs&`2vTj{ z9o_VvLjU_Z+a6HAmfB0#nM+1XSX5sp)==CU26k~>J`>a-nHs=F1a+)G%E=G04kQpb zHvojZy@}4GCZqR+=y~f1Ms}(du(k-`0&o*`&glL6%|G|rsBZs3-mIk1^)O=F%X z4HQsF5co_c$Z6D}shmC@XufO3{m%FTZZZ$OpNBHtBc)u)K=eas84 zLW8bq(Rp!HcV#sSDLB!oO;y@c7205;J6y(iXl%q~a~WCcAjA9v1Bv7qVu+vw$y!hW zSPMngEKB2-YNZHTPXVG5iDgL<|jF7C_*So)IJP$E#+o z(u^Bx#8c7(W-LPAW%1-Fnrg_ePg0NaC{*Qp%KdV zU6Bz(xI&h4Yqb_(9T<=LKljX6Z(PgEI zI_qn)L}FU97HLksFBmML2jlKlZ}EXod(4)*USslh)mukOT&B|-G6>%yrrVx`oohpmpj}Z*i`|0&N=07a1Bwv1A2XO~?`vp`!iA&+?8! zd)0K4xBEGK@7jihugXc^Ha$4_BRwz(bktX33p=EQXw=wBJ?}gnOM$t&5_d<2G1MmT zQI8co7oCss!QZ>^R&IDBD>2-WgH~_zY8e`Fa*(JQB(k^5q%P41JkYbxdgmxE#77GF^43o9crCB#{&vL zS_}!B{Yvjj-B~QvN^siHf8B^K>3x5w?G|2nGw;F@It;)w=KvQoT$gA>zMrNSSk-do zpJ4=7sl=<*pJnw=tWAeO8wSgvwg7WDGMd?4)IRH7Uzfs>@!9PCBq3uHc?xGSAv?tO zir~JS_oUal+il5ghajr!7hVL)c6T<p{n1#d)G@MiX>W~swk<+JgNGU4C_*j7@zF% zV)HJsrt$lYwk=DZMxFEHj@{j4JfFP!?k;hOm`~l&?lsZ=VUNUTkzPKQ6E>?qHJHC7 zX|;KwX;N$FngRdtOrNk$Y0se{vJwRstA68^?7#hRVNj*9Fd^{N4Gn;whCSY>lmM&x zd1`Q_5;lcF`ka07+8Dk*2GhW!EpSpxOmu|8HWFaZ1O7Plh_xzxHDqvx_gUSK65)5dvAiV*96bM&@1!>scOaxH9`Bb6)-U~FHSq8mC=k^x`LX>``Bb}{UE^Z}yJoxP9Jv+DCILk6{7)*Yq)$VH+ z&m%s2v3><6A^U%PtZ=b8+Bh1kmGy=`#zPNv2I+AR!b^7@+q6*s-JPQK zDnC0;&waducQ;0r5B)l}>BRGcp&$GN-^rhe`FKpYt-V4cHOwsR&A8)Xk?di&6QdiB1VR>*?QtR^k1rgXdD0s22;G#C+--@GO-9i}BsxJX9Q&}y0S=uyoKcVbN zF<`^NjYZZOS-r#p!KcQMPRvL&O=|OlgEZK-nhx+iLktV`#azJ+TlYorJu4`YvFa+r z@!6ow_l!D^pQ6j}|BKQeamIeK^9jN2-Cd$-9|dQF@L8>{H)?@o7iTL)dZ(1r$>ytP zaeH&QtklG5iK!tndfrEphDS>D6Kt~BcJJQZ-KP0CvG&pERG*7>Z05inEte@PzQ(l8 z+bEa?0B(BR)3Ym^`qGii+yb~wT^{Htnv%;*;oMWts*`$=*K-#2G~97~znO~?heIF{!3Sk&@r-3)Y6(*pmhB~6a>}rH z6Ir3e)J3aV?vr_}{=Km@3~P-_GSk&(LPCGJ{^{vX|KxOH%nlR5zYyp#A_B1Q<-WaZZM*95l)J&;vxMF1kHob= z{rs(8`gR`W%SkRxN8)=3C0>Qi$mH1S*hv+05;dIGW=@kK?pG#!HdCYYDVJEp;r8`9 zUET55!in0d>~KODXP6ygDvepl?T=S2EDV<4>1?aztRrSnm_M*_B#(pX?OO~dixksHJqmVw(#EUy!~;NzrV7r zzD#_mY^%~1YrOz*H`hxA_jkZ6gPt?utg^0|V$6AwiXDGhMQByokm#&g6SXE2<-Xak zEZR#Lrb&=q$1p!-DVuyv0zo}6&J&TpdYm3;16Jn1_t&$uZO&{rOsm*zGF_Nj_FzG@ zM(jol38r?hc;h-qax8Q#cLw+Q54Ig{bEeY@&gJ0J(eF0%>MmvVw@qDC7##q&dHvBI z7M*&(K9?E80<&W$Je5X*|7^SH;4ktPzI?hN=ZrkGkrIO{%y0xa8PvU;s1M&~Y>K^Y;gVdvhlcCfR|#phcnefN$8gRaV9 zpFLS=(V(WDXm`A>RQqZ}8s(b`ZW>SC)clOwp?fcElsshIru?MZgGCSMgR#Zh#hS|2 znP-MvPufXtdD=SoW#PQmyB}8t2C3*-t31r}Y<0>RjblS8|Lcn$o>%B5vh>d&XW(1r zHR+y@bs9!V!}(%;9U|?elwB^gc7I9BEZXm&Bt^GBOd(uhcC?_rCa=yRZsg|7ZB#eo^nl@|+nO2cBA55x78x1WrPF}e5&s-CB-@1Wx~bz ztyXF14Ys9S(mWFnK*&oDg|Ih_Hl*1PzoE6bOnOhCb$gAj@r^F$zAQ;U$=JVcZ-4fA z=ROOyhoQ2o7_W@TiG$$9@AYe{LdJpO#Fax2(5`#;fme1ppP3Br4h!D>n&Gc7blZB9J=*K46 z%v!zJ+4H?~5|~p!lz#^Z4@zRU5CM?n1?DG|VNfTG@BX81%agS*J8d;O&^sb8#|n8Q zg}BPZi``k5uvj-5unF{RE1FJ?K7NDa29bgYsz~sTqn({y-tY8aGs^aZv;9Kz&`S~L z)6)}n&?j$0dBbS&em8@`j#Lsy(WrK4-{6$$%`_Cc-^R876zE$5*w-U z`BakgNvTQCxj?7%W3Z;poYGS%^d6TCv-x?`z;#z_=F2K=7tOmiy`P3`Rb(Ghrv_wn z7jlWd#u&vNZojAe>$78fnyzrisy^&ALS5b}*Px+rZ~UOJdE?pkl1DpbRHY_b`G+6F zknc;Hyo`9`3&bVo*}rb_cYNjzu5XR;{hhAWtA<2S~qj+@L$q97-x)3F*8PU%ZNl#47{$mFj0DfF+r}M5Xqb%7wcn%a)oTLL93%n z{xx7SrHHqlW7yKM4F5aAhwBEsH)_Yuwa6ky=)I7qjLehpi=2o%g+mMRA&ag(24{qRH&B=P2#8|&rdps zA`Xq~>y+fx=G?jHZh4=}-5oHr@Z{@Df>|;6=x!pifRj%`y{qRv*?aVd)9kpZu^8e; z1nLPF7IIc--9P5$;sYW;z{RyiIzxSAhDYEq%vfu-;X#@O=Cib5ULzWz3XB)5ds-Og z#!*y5Mu!SF*Dn9>RmSyVjiT1Kqh?tognsIoR}|<-8nw;x_g`RQye-;$>Y7ZCN0Mmm zMqLdjN#z0NQzz(y^hqfs4q$}pGi9`Jh}ybf&d>i=VN7+JC8*sA6F*jyUSl{5#-Uv+ zZW}T~=cn^Y4k5{8>Fy>&At4cwh_tkW(7Jh_JjO5`H+kOK?xkz@xt$Z^`ecbo`5A)< z_*#rR5mSQv%qc-4m+i^p&#wk0A0Yf=kN0b;5!{kYtE9|2+533QaS<3jPGsZsh;Ku> zOxQZUXDv~n=X5_6whkO(HHIg|1(vYwb2;m!Z077sDW$}c-V)~f0b7N+d0KPxlsC0p z(-2UST3Z!@(tBx9zh$GAkKX56%SGWoMRT$yr0O!=0#+)F4Sy(5w*S%jO7x0!6@Mj0 z89dtdBocUlyuZRs!#Jh((N?apmiE66I;v^a#&3y{3=hHBp}fq*FV6~-{T0ZqD@GmF z%8Yrc0vcfOzOw}_Gkc)j?j!|MoDgoJ@?LPQ6I#2A6X`aKIx za6kl(b@cg|j)~)_dg>Bus@<~xjqOp66iSr9WGKwz6o5j4NMi(cf=J6~TzD^O2_qK` z*DsKRJ}>x4oE*J6r9@!o$a;*_%qzlR+EnPL1`lDKcVv6dIJPy`W|&1~d-u7CL#NpEbNxlvZ}Pew)4cI++La4S zo`u%Cr>O%S)b+e|4%(~W=M#47!ywI9mAkKt^0bt9e3E;OgmnCJL)-P2vyUIWEokuM zfQ5{Y(VL{y*ZGr~8jDiC#heum=I_YIM24@n-LZbdvjhbuN$dAE!798x8X#jsG58ak z178U;NQgNzwK~NJ!8pYjl`}CaWWrm%cwV1zDU2Y9+(LXwkG^!9Jw)>rw{POP&m*;m z$)=6EU*!7&&tr;NLwU^klr3^Z=!oHpZ#-9er%uC!p6+Wy#BNhh)Kk+B)NtbcRXa0U zLxGXsPQL~smvLLntQLekV!{wBG)K3I9}E#==;E3Y+r|fJf@ltoG`mi|4|3la{O03Z zpIs797kM<=XN8~o5VbjCljyjiyaA8*u7~TrgL%BTs|4*=r(Ej0x>lesZ0;GJIeTRl zKoFG*=w-VEd)*JCtPV)0DW0)deg2KAeODr%Rc(0sm6L7{WMK@;dK41msp&MD+cLa! zALy8VBMqayc8MW8fw)vGIi?C6puO@`qLuHZsle$HTCl1AWw z!0-4#nb78&Lxjb%$$Qt+JIGbpqn#a%3jz+cmHa~Ti};;wjuWA! z+{!7sT@$xf6>A&6wv7nYs5up5*imk-I&{N#&|h@Lw@BGbqJu||oFQ+tlNrpjf+qanw<-;}s8mWyc5k`^Zy%k34U2A+U zJ(etfe#da+!qLbJrz$SVN>NE)1~@aE%=6FdTsi`<8U9(8=W#{Pn`;ywf04Ed4YddH z?_LIcd}%iI%X8vAyE86XiHQiZKT#@0&(yMQzNIPkId7>N@(l?uZj1`$EfNTVeS01z z6JiWnGnoJ+Lyb1AjBkPtta8toxvG&fhUY$)Sh!Yfdm`c4hGlngG`Yu`NVRXsb~<0) z)mLJDcK}ny$30JvW)N{Wu^GN8r0$ggaH-eG&SSAX6~1+NrFtkv7#e{W$bBitN4B5~EbaO%{&l zRhwc=I#tdmy^Cx;p(L^=Y3MPDe;20A#-vX}e102haA{!tpzUp|U)+|Z_05IF_co12 zI|s-PNi+#5tnt{GVDKeT-Y-1tR@4FTE#WY6fAi#ZC=7W&=lHvsIRQinaI$+`1i6J*4#B*kRf4-z6BBy2j-?Wfz0`y< zHS2nmLsD&!$QBB+c!iyIzwFeF3VCupK6ICPg_T2g*nqNX;bmzXGwY#z?>BcoIEmWt ze@Qff6VeNAZA$dIxa4P4mA;%0VshMf6h7*wwXdOCoJ=t5s?AuU=@1(h;HzQ$_8E`Y zrarYhA2PpiEj4bFKYB*+u*MtfhgB`71s>xm`?*r-c;NlDB4)4cT0Cm+;?^Kz$~~LL zGV_w)DplHKsjySW1IXS+Sz|x$PJG~%jHh@9a>YUDB8D$e4u3F|A=cnfSg*}|G1B}N zQ&ET^T}vK0K=c$28z0p0|E9DQPc@k-zpbLM7Ge&1I#}fq4t7Rdf97g00h{kcQ2%Hv z0=S6u9lA3P6&4dlRIBSlT>Vz>wCsbE42fFg#4zIoe&*C<2IA>{imK@H=rO=mMsQDL z)Lf0d&eQ#JqXl)$QBwWU{>OF-m)$2m@q}-ci}48A)fe5-_B4Fwt+&RlK|9CaHjYQ6 zA8FpIz{B48GxS=Z!=?~DYa7|x`y$>WC)^*}L+4cD43YNN8T!Re&OaQ@Xpm5fPS( zbSx2R0g+hJA}t`X=#=hm5V-Th-us+=&-dK>`;9vWjMa zOU)3tjFmGUQ+dDnJ#S@(AF^Ypf$C4yZE)oEXIxxK+3yIYtQv=LAFx8eZ)MZ+kMUci z%<9zX=6UQpUuhT=rhbK~#4OwJ&h^kz$8^GvpEYWf?yc{!o0CmN^1C1nd^T z2%&hMYH|W?3po}s@V(6Q=HbFXtC8CoD4Ch%R1HX?5W)+AG$SDQ{q0x%{y(g|v(Fwt zJ_vY^w$}T5%X1xP`XJ$Y#|BEtCvf(ob`N{B@=2GkcrX4bs}jikw-kNIw=w^`mxuD=1{eY=B^#dhhkAsQG?@Gco&d$o#^DJQLfmG@`q{ zeCon6%%#T`=r=yi?(B-dsDm;C*nItS1xnXh1(>9Ss-K#FPphVst)ClV=OaV^t7@A&J-#sZ43{_GrMnQ?QMG<^!&#*wRf}V(Z7|p z6OG~E8w+hA@^U9Vxz-=8^e;!pza1{OHjOYgwbRl%5pLQwo&(dr{=ZQwzQwrX*}MpU z`()fCp*+ar;Si)vxIXq(wpvyS9n(j7wq62XL`S`kDLU|q($24MfhMtFUub^_W_o~E z4;2a{vkZ22unPdJZ=s0XN)0jlyZHtewY`t4zeFIw`z2yzECW_9l!Q`mBkyVg?vxWS z&jfDbqmIgz2dXlo>B{-gGj6JA7*WYFa98Ed&y#H%`5e5W+TPDo-R|+-t>Rsi*F$UQ zZb`>P-o_RbSQlnzEpu{t<)w@zv_lvL~3Be$Os!sggR-#F$}sCoMS@eGW zz+vd?;4_nqukb&?# z%c{L)VZe_{WPSCkiO1()}?yJ5}$#m8=ob;m1)z%llhtuRjk}j*FA&L) z@fBEAx4^SSw9kAi2mA_7TYz7|)+TZlD*rr>b6_WOoEGxyULun5N?>;82doZH`rrr{ z)<66!y``f~J`%_j|BwfDtOB#o2wagK39Jhdo-OC!ukrsW%tnLdbeLCD=ov9%sBBZp z)-^5ez;14X#@mbecEtL84REeqZYK39TypX7S7G&q*))z)+>%EpY;3>6AXXMCEHa&b zU(Cr#v@%^1m4fc5)544Bdx7-rH(T>78R=m{#3_>`*Vq2bKDER_YRFV^eb?90eO}#l zH$s^hC9%0Pj+!r?PQ*@&9Jzpr$GWdRdDaD4u(2@ra8s%c={JonfcdfMN{ zws`QKm@p3F9u;?m;H(4iuqa{tI^Ht$oky&SllB7_J|7~!PY^)2?I%qq<~D^3imX1b zQa3NOn+yD;kMnRpYp)C-F~dtS;F{!7RJxByhU3b|yF9wf_PJj%|7F%c_1Pxk0Oo~b z3%!E3jkiJ{Zu?Bv{@ek$D*p(&s;)QTY?ORt;kT|X7l5Ad>QJIpW?5&KZEah^ZrCB( zE1e;BWwy63T^1NMSfH`JYP%y4Cg_&*qwOuDYMf=&JB@St%11_VJS)Qxak5DuR=zD# zdW}*F=GBrdS%vkeST*MD&RLbYA|=0Qm|8PVod7jg`x}MlE(c|mqlf>``%(}SZ2Dgp z(AWhN4#eFUjI~el{R3XfE;8+#2TQB9Vk1&ycce>QTmD~wDuly|#72A?`LEGD?#q%Y zM-_tTR{Go?H|6lu{SM0J0>Ek!M_?WI=(ThV>6XhT+x9O6Gz1U_u7Cofdvz-7KXGhC zKODnvJ`JdygX6cY+@SV0b8e7)0sZTS_dqMCs}Uo5wNcjbNNVB>F#^?=`hAPb)>HQ$ z#+=Hscph4kbl0i}`PGlmRc?w%{2`@Te$ZrahE^(;?^1W8q>^NuoS|=N*<^2l%@Atd z^xL30eyYUI`%DAXtP{;za;YYbZ&TD7+qy7uC0qZcAkT(hqC^Vv%V~-t$^XL)`^3yZ ze7U7%_PcUrrWZH?WZU0K`Nt4@=8RwIzTNCUUBV#E@hV<`Lwz5xJU|JFus8sd1xptM zN6=^g*?1Y}bNC?IjyGxL@V^+KLD+{yDkVz?crcAQmvEvhw+BH|j*n z^moVk(B?}Xen`j9f{-?C{?t^70hSmKYwEgtDt`vzatnCAc;ELwlJl%ZqXB=TDCpN+ znm%7Oxj4RFKEll>Ix=QFm|O43lL3EjSF>_CB_-`jUARgzkQs5I>v146+G-|fgpt9C zcgK^LBgaqj``U_N4awiw%_~}JF2`z2ABTZjsx-FaJ7`nO^Z8$vIj3b=)<4BqWMJ7R zS6$yV?>da-eGW?uy>pGKYr1R3BR7xh|9yXo;ei0BUpY2Kknd&E`Uw#+JN?Uc0#*>9 zz65?vt*$?N5gGKNKuyO)3ViYc{sH*8Jm~@B566ImtmvWVxZTbKseFBE0D3 z;xysc%NoPNTdk|H4u-e)9W8dT!2wQ~~T{@HJaat(4pM3Nb)5 z_w$F^|CLH5@$%19s&I*y3F6@k$EX=sELpK9D7?=Z(em?Ne{-wr2PHyX8nBazSl47y zCYZEx*oZ?vyBg-o$TWYaMkzhMk?%e@g>#ri?sbufP;1gN~uDs zL56kd-=l5&-=pn={@iM0w0-g{?|dTdwTu4^3WmmCJ=;`RyDbzp=`59NFUQ)`sl0yM z2yGAA?fsq>0ZC~2KwR!WqhP^*Gfl`&Ma=Er9PeYVf&_156Bgixqmz3R7BYYW|Np_u zD)0p2`9LZ*Pk&>9bjjmIoJZoI73@R#K_~{R3h#CuwJdG77Oi8kid!^1H(Qt7(*7Oj zL|HMI$OueV2=|aSsX4elPMLe$ZZ3)e)wzq_o7jnb6z1WkQecCLo_JlD~ ztwvH`^esxxDA;m!eL8IAw)#~L0^~WeK-Xn2PqvasSZmlH45dyn7Zqr5|M%KkUoP%= zn)=EDBdRA0U}Knt-+r{rPObYLAIoEvX=n84YH}9!!CKT_BWSfjQTYX6^ci^~TRhTOMlVsuY&i&c7k6Jy0M(fgMrE}r;2q-gtUgJToC#E9Kw%#+c;rqm3V70rai6dv+_kDzNDbUcu+=^ z*c4|Jhg_H>?az#mVTy8t5m!uPPgZ{R)Cpub=0MeR+1%RssUI83sG1o;py^SXF~7`JXkDJ{h<%o3k%|4%$-!7l=N>Ce*4}P+ zDJ6Al%DBnK1NE&NUs~FDt-T%@lfRT`a#;>NU9W@P3BGCeronA%qZ87*b)`D)NIHf?g~jN7(n?S?>!&tqg|dB4E`Xf6sMmLA6kg3BEZ^2!i+i zXN)Ru(g7s#;BV_L_byF0{-;L^j(MSwwexu>9th!59lS-OEy0z=$u{lQg_w(}$EaNf zko`9WF)x*_PmaxK;%vN z+%G|X{I>ibGHCCsz4jJ{7U_TPdDl%LX+FYH^8dEy4c7FM?F(IkHWJBrf$13=4>1;j zfi|%ILN;K)%7Kyv?-@YbvJ&LJfzmoxFT{*P9_>8PQ}}HsUPvFiLDa=T48uS3%uNbI z`v{RdKOqjb=V|7(5CN9=k021IIemLL#RZWC8<;n)iA5R=aRru%AVP_wqlE*b*ZSKY zQ*NHEkCrtR9d>i(Os!6|g@33aVk394VFl@4vse_nF-6r+sdS0}cn ziOrLtKP zduS)W5!-kW_eZK_)FYGKjmjLu~^40BLHwcU8=#G_*}bXwd6B)HYF0}dpG<|Vu`{8$av?bS00H}y`UKP zccvgjquN%-VbNUAE>jS2=7)lUaejWM9bal>xU|>)fctucGVb1i##d_};t%d=^_mo7kIhTx{pL>L>K*5fV%rqmRW#BN$>8U!-9TL8KZc|1b~ZAQhR~L{|h(Te-&zxkVzXg)x?+H5Ld4893}! z3S|`0*|knnpX@Crs{a|lDia=%ac8?Ogt0}b(MTFk)o*ryywa6R?ADV0#yj+t*3*3M zf}fX+7^#M?EVeK)MGK1J+Bx9US&)6_BseA3!=^M`KjRS67)O3({3@N52{Ai=VXU~O z*=aUSs(@Y^(o>S|1QeTN@21}q4T#Dsx*r_u?CqaOpI(2LMwijt{6t;jW>$~OeEIgy zq`>kD#%mw*DlBYGy>e6C9axCdU=TVqo~=z;$7eHkdeSyVpS^KyjU7l!Lq@l-oQv;& z!Ve4jenO)uP95NYW{l8??(%t) z*1>xMi}Z1YMmYx_^NVqnYAo(S=f+zGEG{VG(!fF56Q0+{Lp~JTZu+vYl9Dq&FRGGF z+rUFEQ{uQGTgrUnF)Wl*P5POlffX*A^m)?bx2MRe^B`b4;b-=nw9ZC3XikUn4F6Ie z*&~6-6=pCuzNp4pP8620L3y6d{!YzIAMP9G)+w)~4)R6~rjE(c;%PURrCqIpw?E78 z^3ml?M3-xRP`hI{yXWR^zv=-b;=?dt8m)mUY8`B*E1&D@>#I{`4GghI?&C~XD=MaV z)kq$_xacjNSvQm=K1I*PF6DqnmXOj%PJMJUh-K#2%6aPgpJ(}8h5cJua)nS6HWwS2 zE<~eX#P0L%3#<2P`a*s}y80U@m@dL)ck8($J($kQBcH6smXXc*oZKPgUhn;*XjC=5 zM~&;C`Vs9pnf-(4TH#-KLBa7tk`oJjpIj5x78oZpZfCCB8K+$4p`$O}9XY&yv`KyR z#+k7$O?dNm#acw}3aFACHrX6dFxdtM1(8uv)v*!)?SrzYozr4yF;2z=sse9_({&|45$q26OY)X_fxZoCp zaI=J@`5`7K>B@aV^mz_v!57%g9h$Oisz9D!6B|ZX;t$PFmRquV(LQYQN~)|xkL-kw{F>K3;z#c`8zrvD=UO&!Hv8?n7AP@}PK8%?U6aa+ za9Gk}VY9=Ca;ws#*&3L8iVr<6VT|7R$g}7for-s+d+=#r$6q% z<0Q>fZWLHnHl{1sg6LB1$-~V$hP6&z%5uZLJBKK~9uMjd6kaTp=LgGB8w8 zeg~O>`Y$LnD>a_^T(S}EtHe+pH99GUt4p5o26L`H8@>|kA}viLE!@Y*-)tvrpNllFvo-EKfAk)5h=WpT=^=lqo)6QwpAN zj}Wc<@S5wZQ+adT$fGwic5IF%mwql#uYPDS>ppdT@f>nx`+Xk?P#-bjdra0CtczaG z*0;4PYp(fP*Gw5|+g@-gdNVelC%2<(*>L77Dv<_5G3{@x@bbg2N4HjPU0xZqyB2{r zIcstUtMF`k`nIB(yZr$_X|C3!*U^4uT4gS(6?oI`i)Py6FuJZKE)Tr?K!g{78z4%- zXY21;Zi(X7*3tt0Om3;1Fq$6SYPY?lIcSL)itp>{D$Moe?Rb5=#`O#Jf))IC!4F~Sz&!DcjcG!__A$R<1Vuku_Kz{d`Q#r*_|EZ3+o?!_WZ71w;dK^X2{1Iomf2GFQIa(4W%tBt)-WdDXz)6>eH-kvJwus6@31*iBR?1uIo0%8XKFN?L9p~$AwZLKBjfL zXQ)Cci%meFX<>m8&M%~P!P_tml{d7EuJ?B)zAP==tHDfKUS7VjwbdW`qs1E1(p9&A ze`XAUT?<|KXm_?c0D~nZB@5cjQD!Kr#-Cm?8H}paF=h&GAcc?Nuz9wncDL=)zPZRAth{R2W;Ii5gwLK%K+)8S1S7A99)3-mG2?}ff_~M#?fyQ6EVx15=B(J8K83yQ(Dr-Tz z?U3lP)h9IUx{#W0%=rUEKy92Y$McT^=`+QWMnrcPNU^qV*SMk$p z7bG}AUOVUZ_V;&zpWUE+IH|oB0}f6EPO+QU4eqJPd$9aDD32qweQM}n|MU5;eVO0)rn@9szjU8%c67g}qpC#fZE`I- zH6TD-INNU7H6VBu&+ZCW=GE~2$5&FdFu4Ad8>l^A_YTR0*KWDgzC^mD#owRbG-|!X z=SG;Nl_n@4+Rjc`b^1NH`TQKGK+>X?ud!%@=MAc#v-TZW zteBkK#Yh9NFMLQB4#x`G)xdd?KJ!R(?j}&@Sr=$?1u3>RV1hzHWBq zNG8lXWBYzZ!*`MMv}t>EL*e*{Z*!6MsEI|J9l2ZJ=6Bt%)GgUub<{;1tYAM%`F${* z=6SX?uXzybwz`;aFBys)(uTtkSgLI8$KA4nt${3cR!t(j3<~lK?C$IhGrKckg*w77 z6T$@eH3=nJ#jgpGRuxD`(%psU)?m%mrx^+7g=drNf(*|AA^le#?!^El^OR|22Ye|!Z!$pj?TqR8_ ze?flY<;iwSQpumLWc*9j^UDi?yfigiPc)dJK{_oV0dJogzl@9RNejnFHTGKFyoQU~ zN4J6bLGHEA<5dlUBaC^`uuNuVKdrG`0zW8nzEp#3$aTn#vzk>!O-ENS|zs~NwN~A#$3lz)|=R$FOL%TegoR?bV z#slX=?cq0sIckw@qFBSS=|)D=ye5-_;>~5bc9yvhaV0*Cy9P#a7@RMk7YsfvxPAWBL zsiB3(`nXuq8lPp1qiLo$snWRqo%n(r|4SF1*%Dd0aGHq6Z)DS8hQbcqeJRRILHFh% zbhx-@{nj{0HuaK;)T7T|4lCYrZgOe!hPe4=h%Ai8 zdwDXlKehfeKBr7|0r|24()Va)vD%2~*_3qSYUdQ_#M1fe$)2tIO(8rPbs&bj`t1MO z^FTK+dwt|vQs(c&kYl6LUA(r|U0c_EH$gqu*wU0ztfKv9 zob!8!_99J*NcBiWsRu2CqDtrZYOh{{Ufwn4p5dT?n!`6OhP#K`m96{Bm93Bu=1^{k z(Q&0EdRYpE@z=^MYjs-n*R8K->^K&;)MoH``3{Otr-?+}%e_8W9Iw0Q8XR-|kH>ro zYF!gs5tqlQNmkBLQjo8)JJQ{ALZ8;e#EHyI3>tli-Znk`=O4I}0U0kNUX0BM_btDY8 z((>oR+#J~RFK&sc7$rMAX*E<_dCE_*+9lT2>6^ zdz9FWyJ*RXBqEj0OtSVbTVfj7RbMr2&@jeq% zrd^L04tSr|?#p+K{;&oMA8A>w>6z6_Xrwh&0fbwf=9`{b+hHHkcvn2NK1847`l2_- z_T9ls`R|KLvrF67JO1b-whbB7uGBcmqLyBMTUVi}@7&p9w^ z)@Sy}$ICu>2Fa;fR&%sK5+WcVU}a_XHi?BZDiyLzj^vcVMkgj3y1J4BG*dtBcn`Zg z(zgw2DPqx+F}l;Biwg`4yhcv`1%(eUqA*(9$qg!_hK9b4%_VOgR^uQJxDAB$OWLn* zUx-0j!YolO8Vgt*;IReV@wap3k8j%Iqmwbn2i~^sFq#pt2Ie{jU=<`91dGc}!9FrX zG)V_VGh+Pimzq{?)PtM%1*st8aM4sfwD$;lUoeki7XgEZPmE9MzaR{PH`fw5$#ntG$d^teIW(BQjg=yIb}Vms57Wg}(p?haNwoj7!knQ#wvEpPW# z&5mWjdr%VpDsh#~eh}4z9n`^>#(x*N9P;5k!<~Y8Tt}J*l5I@}Kf=SGrKY7bGwUj5 zIpSd12`GGS4Eu=f1O>8Ya@mrKEcO19a0-TAXST7EE~e5EWXZHr*uhCh%E@IHBHKaF(d2#bhao`p1aqI`RJ93<0I zkMXB&Me6>tR^=TeT6j9cN1-#L|Drvg=0lL+*18o7Icrf!`ME;v`%|W7YWTE6mdw8$ zeeo+3-aqY(DO1v4Qw!b|Fvv9U)4f04k!$DvyNPi$I^@l;DihN~Cm6hqXyK4@{R?on*X_VqzIr%# zHTV}LS*-}G_nhOvHdp4Odwo1r7b{0vA*+MWeTOyh`Rm6i3*G>eKqjL@ze4~2~vSZ;9>R0~_aLZKk42x#_#R3mwX zr~2fI>cfkSa>`r#SEs&M9xp-iwkMijBQHpo}K_S$#Nw7Xy(1BL+UU5I^~{_GD($_tByif4GTk&Vmi2SeigN z@s%C&RuSx%&yIQb9VNNs$QM=(1utl~H8!?B;Mt&JO+;MpAVkfVM_lvP}QCoS=U-CQ4QstLPC+V_cU_9i>UHL`H`I-N0z z7|h{q@!zjQXwyznR1)dybOwD`d&m^s*73Mw2bDIw)a6yZCCSc&)pyDHk=-P|q-j+U z%v~K-m>FQA@?eZP5aj}s$3xz!ljONv*Uxw_SXh!NT3QhOtbyjWI)=U9UAfUot$?G6 zHNCl2Au;}5TtY1a2{k12&Zk@QEU6G^AK@e>CI+2j7TCQ|L-L=nYM(xPj02{}!J`Xz z;NdOZAdc$Ml`9^O5zr%E=De~fCQl3qdAspDz4|cutgeOb38;eEO6pE%+u~-J8G%ih zp81%OB)POSE+>Z>?4gT<&?z)_MuIYKh6;9g#5%b~O|eDC{pJ!}_~X;0{CgEbg5@!* zGS_T)n-%O-@5rMMSsXCVm*iNv?su;~34U-Ad_y`1Zam0x8u5q%T`bUx-3@)X2D*o<6%MCNq;lnCjQcDh^K5x+2uDFRa<}DyFZv z(dbWW!*U#_zY|Kx?(w-Ye|v*HF__pT64I625V0s%QdoiMO-(4(;`l0}HTARW<`teF z6CEX#-##hIXfQ|QXz^6n|0bZ`9=8`BT9@l%i0|9ExIdrSK9zZARGNnHni!TrPDVZ^ z_CpCPN;MnUs!gocI|t7Dd#e`+$HE?Lx!M70$^d~GqDe%OVH8PPxdXK32!q~yuXWNE zwo+h<>{maYcfA9!GHd=xPH!TVsYRP~xBr7u>(Zf(?QQJHh?TbYAi*(yZe|LrCQ^n| zKvUoU3O#+|T<;B^<^+uV6~p!U>VA$mZT0#%DYrl8gW=Z^*N%IbGQ)3Q0El@D_`mKu z4YsoV#G?NJfkUeaV<#g;hy^71x*(H5JIb@BHiYyx@fytQ$;@ZN&XDn4*WGz`4YjP} z!KmPPW1#3t;>Y|aidS7F4>rQ~b{D9{EQ{;PrOu?*OEJnQrS~xYRRNWcDnLuk3K78Nn(RCU9O^if zC{7oeN3PYJj));MZ8S)qk9MW?KmUs$VC4dJ9S`S=>VuCh?cF|ZM@zQFGxPJwxw#jc zzBCx-JM!8k-^S@Ojer~@*vQHg*IGWDUs@V2pphYG4|p9Y-A)E$`Q>21wXa>jo~)G{ z4$tX|XQ7&=6ZPLsPQNfWXZ3!S&Prb5{1@~>1;p7~T`(>3^yu7b4z;8|#>f(ftN7QE zk1ss*kg*`(Tlbla?z=v5+@{~&iKmsX1F0l7wGZJpyK3S(F$GmW=h3Mfw<*(FRg57KUS9K@M{d1~9)mbmbpL()OHtV>)uzo(#L->A5 zI)_bRLucXTb*CTC?w$?s{h6XYy|WXg`kjsRRI+g0&$H(PCzJP*RVm{v&YoZ@jajV< zZ>F}jUCO`o(}M)405#d{SSQYo5Uw>cD0tI9BiOCV-hMP{Xz88BfuYXE_**`0RaLhW zB`W`JbYNq1IUX`5dX-DtR!Edv%J)c{D(2hh77oey$v{XasM(EU4TCro zGCBiE8MsAwAWZp6<%m@f$N1__Cc4p4nvv)O_EDR({mrN*aVskpK&)Y_l&4Zq9RukS z)wB0o^r}JQs>F4d{_=SZ=AP7}LfvRy-zXR*Eu`=9XT10z%Q7`QV%~(P@+% zzh_QdZpm(!++0(5GN0$iJ6rz3%*O1y;yGT?_0;xl0t=HfF*kyT=`Ah_Tw|mCX{h z>|~fNVPcAB;fiNrNu`kW3GX?-yRcvo;V3o5cgm2(&8a0f=&zHKXrQYHc7Z*=#W_xp z`j-quH5q^#XH0y2&8RgzVz61byW7l2c;~$SOW8CAyrd^Do-ot^vR#ss%g=2bTeKvc z-SPtmQbee%s0f347SCO7FlV`F)@pum0L=!n)XlzN^ke@0ag`bm&-X@KUr`qp1G15qk_%eGCL?9RtxzLD2xK?k*o4 zRIYkZ@}mnvk7t+==U{w zp41RgOxZO1l#Gv5e?o%@Tc8sAiPa_rjCFz+bVFTR^!Ky^0Q34PbZ6**N7NB+$onN2Hn}f2q21mq@kJn?+}WL_kPv& z9uxxCCuRbOge%8`F&64N|jkh-CDQHmP-e=VZKTLt0&5M!|RW z7R5kWd1`{WQ2$^J`vSPqv$XC?(gvN*tKJax=|CPy6H_Uv&=rI7Fx6Z?gHd=UH!l97 zCjUOWzPPj1uxhQBnc30RC9bSYN-vlA@!6fe6Y-(s9R$@nCMLltDSvni?GpU)pj5RK z-7p_IV=tQ)o|s7G>go!}=+le`>@l}uq~FS>@n~2z{cCO?SS&&@X=-W$iRU~srnse_ zqvx<=3B|?R7c`%>QyvjU~mAc!vkKnmw!9!-{Wyi!qdLGN6 z64-A5)yv#vKf6H90{4Mg#b$@R+bY$m31w@9e z2e7!`mmRZz4|px6^1#A##gBRgjHAMv48TH#8Q%%__jCGF^?n3sLJnL0S3EbUbEyJ8 zwBZsq*2v>LQ%rK|lv_?mw}Q8Pq}gEUGac8mnhyN{*A`uuyt+aCGm45YoB(xP(%aA{ z!tCpmEa?0IW?X8+r`QAeBb(hdv2{4xOn&u(AKK%sux*Pc#(|hN{2NT85 zN|e7!xL`dNLzK5vPw@E@pH`RIS!jQX?NSoDv!B$F>%~#>*;t~T1drQJyrVd@pESe?yJRggPmWBr1X|`USZmIF? zb~V5DHzw`j-kYvk3>!lpp8CN-3Y<3&PlLjKj{D10SH&@;IdsgPx*iHnU`|L=@}i}Y zObq$35Ch!_mw)kAYfm7ndSM{TB;L?ZJynupj!*I25l=m>Ma`rw)IbpM%gN&XY;9c! zr+Ggl55@ zXt@=roOi?%f&q)kiHNRLPgP@Mm~db|yi;FqWpKV;gW%O3uPgJz8wHhG%8Z{P%oA&q z{jG?pPHz>>4|)G?x)6R`U} zP0_3bkwc1_ToA82ov)l4D_U#*t8U(Hm@G3|#rZaik;&4vta+mFBnMB^JG+0DWZO1@ ziOHG18^K};q{C)mxkU?YL&*WGPV?p7nVGK=30Pg(alUnYsvDaBBhv6soUUr{;LdVI zBF!}icGr&9y!HEG@r1u@c_iZE0vi_&@G#7zG1pk~In`5?3ftyGNub7o(=ttD8JB z@jf>aD};ahiSeHDDIeCzlli-k<#pX)fQA=%bqD(hh;x0IKD_aDqyATic@l(l1B{XL z6omE6A}47a!%nl;-k;ju2ga76v-349XQMtl_#Hqr6Sb>5&R5R^{+?WIdSbkGC>Bmm zJm5EUZJvsr+JIMgECSKGfCqZ~X+RW@iS~oGzlC?NuDwIcuH!oY}K6XL}wJUg@mGH1| zyZp(4xOnrS9Bx22eb|RPD51CRr5gm!vt-7n=kWXZopMQIN^|>?Tf)~Jn9CFD=atC> z$7*MlqKmWSnXh$bk`!i&7wZ1nV%Q+uza~WX-m7MyNF=NLqc>%d`B^6ng+$j9ddr_A`) z;=sLRdF=S=6Yv-x(!z44rlrLsCQ2wOf*yskO+%ng>4hw=Z5_8IcOInkWU;URwG$nLZ{+Si=8LWnyN-H1GE+>x!ne~w`rUmN|G6nwc!+XS z*}i;?RL`9WJcFg-LvOG`+<$Sj>2;tD7vLI_-dnAGw5iR(%K905asL}bRFgZng^mV^ z{HCGy+@YD3Ca-^tpUQ!o^gg_5{6wH!5E&n36{b}}B;-KSfdR(gS<21zViBK(r=j~% z_|lpzd4h^bTKOtk>1;fz_Z6>>30Ve)!81-)aB9L0z-0|G$qI)%m`fOQ3n|W)NPL0* zTO8M4mN-wXvnB2dcAC;{*M~@tWW3h;B;PReF={S7(^o5>RVKuGu(nN}#S;C9VeI3w zHEvYCrf;;x%7r2Sla4c&xydXng})B@kXd-*P~&7VokPNEFn!GY^i1gNH{LyjP;+mI{>S*p8&9-CLbt^AM$!c*pC5$XjuZ&hx2)kd??0J-ib-bs5EpWg<-UPhD z2!hX_ur<>Uu%IASE0y3$8md+aQbK$hB{R)!J6;O+)e!Al5s&lA%B10z9@7BxR6eRO zva_-IRFB>O;(a}?iL#~FzXmn4p53_Vv$0;x)2HCjar4lB5lT?m#m*h%-!=B9&BUU! zCtmjcgfyXt@Xx`uOAhE(%FMhBNPlC3C^e&g@l535V8=$s1iQfZ57#aq&yv)x(W$wQOm%LbO*@g zV@Jk0rsxy60r4Bu_|`kmqTBOfKatk|o6F0>%qng5OhWvg#ybUc$cG-8)H*3?Ehcui z{0Drw3R!A{B~K^zf?R&nob#v#wB__eiv9|3e*@KvzR{?am+33-Yd@ExDN&M&ih*?} z)5C}@Q9s(EM+o(NiODT7b>f%QUjf2p1giyIbY?2{9X(x}p;7uD;qT|8@5)5u+rpaGI!qdq+CJq}<`r=p^Q zbK`Pw?_v!2czK3>U;-W`E{#&|P)6p(#q%8wpR7r z-6{UUTX+eB^mwSloKOCnO&@#aB2fLBiciw;$NBi?UF=ca8rQolx**7he>~4g>Z(Rk zR&+XeXsV<*oZjs}L;6ii`c0OW0N8F$th=}oxFkL4T@w~YqhL#S^UV$NP+@yEhef;R z0)9`nmxjh63}FJ{)`L&qHHoXJftp1CzzDba_?&*U=1>2Q7oM%%>|_eQi~k1Y#v(pC zla5MG7C!7kS1pK!YFBDcR7+i@OPx24Q1x6nhyvr_3gB*z=c5l8kJhvgr98KtyteP5 zoWY%z4l<~2A{0Z6)re(*P6=bYC2kGOhko?Fui``71a2}#FN$g%I-#&BwiE6zXs5-Z z18c|h1CXJiUTPxJw9|}E!Kn$b&YE$>*|}|2==8p&FXw5g1FHx%;M-Wm+sKstxpa9n zxkcUc>^X;m;$CIO>i6s-at1*CTVRw(@=k%~ui2HdYq5?pboBbQZ&+-8?l8|%T75?b z^l-#@kZHwhJAqrM7OKx~a?YBC_G-!x5rwVNiSLK!K#rDAWV-Fe_bd(k#`dt?r-cz)XL>BdDU{Nn)HKM_4VNvByY=*urhkdP@ z)wPr6wMTzAi#`m#D7@pXZ%c-0s#)Yde~xBGnb%>kI)FnsvGiA(z>@2jml^Grn9Gmm zmV3EzBKyg_UR#o{f6n~gtG-FC!Du&cc|k^Mt-ICho1E7EB`PcL=|s0MShZ*U-bEkr zEy-H&yFMsXYk>@qUw#)Kro2rDRn`4-2lc=T#ba~jekD>?SxKFc<0r)+@OJLWjKEmh z@(Xn9!FhdBv%$)zI&-dXCT|03qw7*pH~0XV=8;wIl+lDf4#w>u*cjjJkCCkzrs$To zf%_hkE1#?57VriBUK3JTEi$ACrm z_V))iw|o*3hWi3{y2Y?3`*vKai8c`-*fYU)?&h1i%7 zklDKzHsn^C*G$Ua7H+=mO6_KwxX$$9#g#LF?(+S!~zFtSmM?yxeiIb?K9n|h#Po*qS zsOz$O1>m+#LGaIKuL_6c{%hh(XT3CNhL@kuJb3ZO?)jd?$qwJX^Ao*oBp-qX1>G2` zpyH+z+tcL`$CPR5A6f>e_1!>YcXyOkg$p^yp3rm_hM0U@iv!`+0N@toiNKDmLx@ggH`tARbKzb;(x zpJ*6`4uU?%0s73W-$$Sga)0 zF>`&7iOMmVz&EfQaYr5Pd+m82ZJ-Zn>rv=mP7utplU2EQcW-ac(a_KFUbfQs{>YIq zk_of}7f(f&dE8o!caVfo-jpLkMlfdLGE@%6{X)%?c=zTgHdvf_ z?@xR~9p*UI>e2Y!CXMm+DQrS_8ZDTm)GsL2R}OUU;i57d@rk{a@cKU#ykWOZX}HW( zAa&Y=<%GY;`p(wm{372w?LyDpD86Vo;b49Rof@Z(I$mnTC*&DrJ`$)L($SkFYXbtwVVv_O`p7m>Z-|=#~1j`Sv(_qD)a) za3g|uR!0Mu^3)t-n(<#92Fv)+XMx%ftkL>LDNG50{fAStsM&9#m!jd;J&T-)unn4k z3MVClOURj6#Pj1q9hy{UBJSGKJt4YaCz;+#CykYvl`*SNt}C+F1$Q#=eYCBnMZFz_ z7_HXDz2(wc+kfaEwH+@q9PbGoQ{muI&42}j4MOBRIsh!Ux422w4VdF@;_;@(1~e{N zZ_aeUK*YS(MBTlr52!_*VeK!f5Tf=VI^V#e8Osc~6SF!qd+Dn%O36i)JlES7eKYj@ zL=Dew1RZQhBYedjFNN;+FRJzNcb4rwOX(F717nL+lr+q|L5pJc%D#l{)P zCFHA|@fm8pCz~k*sX$?N74~%0S+5)!68JSXRq@lk{PH6hPwh*-=P^P77e(g~y(%MEK3JsDKbvXer9xVgDKkPx!Fb){;wVFgZ?e~W!So9l}rEg<~?5ZAO{fgH>&*XVKh$cUgjfXAUZ9*2vr zGwg{2y~{7Y+YOu7xE>$wD643w^uHe9Cei}jv)cxy*U^HN^$7jKRjcWjUbCO=s-5o3 zTy+lKofnvx&G$GLEdpp04ZvuO(#AhbiD1V*L(A|n^Wnt1*zpv1QE7C`GMc!(uS8wz zXd7ya9I?S%if0JgzF07(bTkljV_y&_AdoXi%V1*Stc~27p|{31r1whqICK0BUnoldH8)Wb|A8K>M(dpxOY_MBXs{+0oc$(+sImj7j>Gak2V-9ysKMjya4j(w zB^U&9?BNSD?rnnfo)$=}=dqo8M=>V-59e+=bjrCe4;EXM58K+=4O9=AE329~oZd_~*dX>;C+0u?y~)SBZ>6L28al0CagR89!oudW zY0cXmC&UTy>qS)6%?IRCn1xHK)@lPbvED3H;;)o)A{}!gXOEO$TWc6jn1r;LiV&?t zWGF~jE5`@BhfmiPvjLKV9<M0;5?~B`7z0*5h#K{vxJhZvOU!%aSdxVqkX1?4~^r zJ905_12o;-JP4BR;g|zJ>uiS>3w;;C`Xd!#jSccEI=Xz_?@FWl&Kx&4Ujz^daIW6fpaf_K?QCQ_wbN$Wr;|+7g3G2&;AXdNbOpHDyQc4 z?vn^|X9#Q42Y?U6!xvoT`q*iI81~%+H5%wWK%fREs9A6VhY)>a07!-!c=fgTfgj`U zjZgoE)2{-bn$UJ>pG%NgWcTJK*yb72W1tNM|LV7nX%KU ze!F8FilLpDk^~>pGe};*kWuBzfOh>i4Da}koJUP`j~*%ZQl_KE=CVA)XT=MKCMv-AP9+kJzU1x_iVq2b2^X7DJZINqXu-co7yC|J%`3sR1MA<6D+IsN>BKMtI8wR2RFmw|PsJEP0mt9bW?8 z?#tc-c>C*fFV&dEm?I})xw2hm1p7j8dh!WUJm#+qL?9I^u;Bhi@&0{<$W=X^Xl#@I z0aXgGOLQySs)Gdm^E%8dMHSMwuH*_wy?FeJq$$v4I5zRz%-F1wwihPP2yVeQqXc3iB`ay zq5`{v$L^(4l?H51eVDHuY$fsee+@c|h#Kiwr^v|3q`kp*!!P|mQKhq{^<57Z$H#SD?>FF7j1`|mx zT1@J5j#G}YZo7-BYpx^(_v}UIQzd8p&kxUgup$mlUuT~ zz2zJaSrecXoB{jiHtP-2z@i4-E3u}R)S$C9e``^6vV+^CYiQ2?p`VyxvR2-)hH%GW zo))9mPx(`8HE%tNbt(&6GfVMhk;eSM4_+TNqN}{fBP&h=*L>zkm$|tfT5$ftEkEQj z*bTvzR9;Ys_Xh;jaNCkc_A|LrP0MU+B}@}Ceq!=4wOWsVjx_At#k?R-m}(~y6IBA{;sK*vNG~lfr2Ar59eucdHLijp&m$#cYn%-w-3dE2nJu0Dqwu%l=yz_zmY8vg` z^~B@gJHHP{j|Jcd9++beNnJQLO|tUJE^NN@ts%I6;>4FwVQD~D@oY0INu)O;|Da*Y zt;&)pSJ5S0(N$-2{5b$QF~v_$%<#WY%;=L7^9nP1@Gc1-K|R%wN843AGG*rV(!(cb z`D`(>5V>X|qSsM_13@_}Qk2T1^vxqSeh{KSv&rz1@6Y`udQsM{e7$~h7aiTo!LCqU z>Amce8An4q6a1{gYmz5+OPEb>g|=Kue*(iGgvU};7#AhmIXFz(m+y>S_Ot(P-r@1E zAf=L0`H06=kP7^NIF~MI$IuSE=%cjIpeH<)@ra3))zr>Yli;An+o7R52b)6ce+c=3 zm97)RZGrEhgAX$D%xPumHiH>#%P{LwR7Xyh3INyG!g)O3lz}ULMh{Hj=HBlJrYP9| z!A%1m%9RI$Qf$#v(~+7pO+rVkirH%KM?j@xlm)j1vC75VCy zzU4PC`?k_`f-tamCe7GtGrMtc_3GZPxOc-eL(BA_gtW4pDLi8S+*4X}k6!9_HIUpR zyB_zBNMfCb*G6dIX*_9k&mWE<8UH#I*MqUu-0uE46ua{u=nThG zymd4pNW0gcn>(C)pYKpM{@XjrNCZc?4&3lyvRK&$Fo{%^?9C&%?emaF%%aZz$f(h~ zZZDgT9x0DR&OnnoLeK-b*MXEpI#*V$2zKBfgKGyonvVSD7+@+orsT(3FZpjEzN@C$ z3fYR`;~~2>7)0?n{~Lin{$aui4nQzj8ja^bd<+C&fawK5TYxR-n4+rpdvhG8!xbG5 zmWyFs*Z6(`_F|OWUM{fCChl}}T5)|@=edHmH6*7%8*FbxNpPl)<#xE8KMh53&sad> z)*4yj)|Y}8mWTF%st(BVbdHGRBcr6;#|TNRXkOhe(fMoJ?-}CrsYfw+GYYNl_$S-e z>kZov6iDG}nKw==Tr0;nSb9j(^F^sdar~(F20OR^ROBLPpP|KVL!~N+_IYiM#hKx8 z*Sq`_sok@96w1uF*^MsVS)#FI-%tsq4Mv;xBcpuk5H`WV`T)ciXNCb1kt((z@~>BM zCk0$dj>qsbITn{+H@bhfYI=>EJ9;9(oTL&U=>o7fphez*qKVZuC;mD#?`Gld27tf7|?iwuMkV0o9~QHA2`%(h^?Mz$IOF;so3f4gJD)m{zk zo!HC(5QUY-)aD+b?wflGg<2~O6C?1VzFYb03Y7JdLic3q^V<_AD2qDlod2gdfvaQD z`GU-G>jDq3(`r3G`Zu`yU5zp@6tUotOjqi=9cfpP`9_5O*13;Zup|i4WtK|VoU~;4 zz>Xf7BUfUZ+B)pQxnLZ)o~T&*J752|3~^n$FnJ}u{+JSeZ+B_HrjDUm?>X0J%l-H> zbsurPk96JlgZExheg}M7^EQiHhVgi@h(g^zoBN9o{d{` zp4(Kn_e7HvjV-1+Mb`m>X5;geq5pYS;**A$RjFmu1SI^dlk@)QzU~!tD<_)Iq`I2h z=5MGu*f<98XFjF;?w=@=!q)b79B4uZD371A7%k@2eJwo7S3fVo@StOQU?n6&=&i&k z3cx4yli2%J?|p*2mvZS+H3Mtwzy2N86ua8tzRj^5E$O5EbJi(l)@*7h5TOiC=gTFQ zwmgNv>zXdFQ#iqN$U4=5pQffi5Dn#j!!M%&X$efxd@3&w@C!U(0qfil0s`X31~gL! zIE*;?4n$ugmqG(YW%0Gr#h^eZWYkPgSZDv{@C7lq8y)AP%J%TSvPg>4#6MR~SL0%l zFqOa9GydGlMV?>dYNVKY#(* zSB2S$yPK}R6@nhuHjG`S0$13s=W=QNbziY{o*3;m`Htw0(HvKGPL!Jb^pm1XZ%xAC zqzXTY>51w!x%sm6yU+aEo+TBpmxC6*!>!i~mP_URBSG3vglEhLRufla63fAX)f7yX zelj*dpy>F80(im6F*HCq=8{6e%8EGv0{nD4A%`R!1Ulu*kswYN`0U0K|KwJU$RAiL z8w7FMm;1fdD>qPenA+mCQ)t^~`hAgLo}wV`$kPm$mmspiTG*@~M+vMy(jfExe#fl% z+&;~qN0i+QA}@<8q^a@0G`$nM5h7q*6PK8HB71H1YtHU(EgGp9>*JJ)D3UvuQN#&c z8ym1eo}x*s)vgQu>4kZO{e7?6r}d&Gq%5ZRV~%Z7k(2^k9;(AaIug0PAwx9RmHe{| zF)e{)^33YpC0rV*avAJ)taZb|c^{cEJe$zoCInkfxtL5W40)UJ_TPXL z3KLqIYA4o-+k5!dxmVAmqIs(KMt*T?7&dQ=f%ReCxyX^uZlQ_k;^T(*l_OKWdF{Tx zJ)Y18T>5O<=@lpzo8#K|h4*uGImM3?xw7X+Wayo{M$VNZ0*>c( z%E_k=@B4}M?pMuGFVX$&r~D_9ne?7l>pZ{D_6TceI$tXR$*84WxcxQC|}`@VZOT0)!sC|naO|fHrOz6{e3@^`EZ8SytwzB{>-TjUhK(x zX*tt-dr#Sa1zvz8f$?VVi#2}y)s(&>hQ1pygV613bfwmnmpND**7rJPQulIF!g0u% zucm8jlVY{(ZQYzg)e#9*dKQxT(G^S^ZH9M|L}arcOOMuZ^LQPvX%DUMj=ZoW%ab55 z(|&a_=;1gs8**!@Z1MWny}CnM2dN#!ITrGRJ?>PKj9y9|P);(eJ~lYEiB1k``u-Ob`c=r_T_bmU4ufT?&$07V*co znw`8Eo%RF{JmDy~6`mxjz!eMve?u>{!ItTSbJz$?WPz)0r9w?5Jbm#mM<_82rOzXx zmQA{AN$Q2o+<^!AT+A~5P2{^3SXVn9^)TWPWa@jG4BVFq7k!#0-7smF#NQzMb^bYz zTWj6CEag$Ev6V-1=ky)Y$Q%dTL~KklFJ#_M23z_(bVNsA$HSXkTofMgB-*aj>oW;UPtTB$eBaTft-t^l z$nSP?Im8wYcvbHo@Uvp(*rQ`#zE?(S~+EF&<#{V-X)d$DD9 z6EPD$`v z{rijqi*QpccsZPw=g<6uNBqsP+uXNhsC&-(`Orb_XK?V7i@BSprx*Y08DB&7{$--} zz@v&kp4M!~CADx6>={IWnG9`taj)nbM5`dEEO~HK{!`Wcm)VQIcbkM}%0~~Gq0oa8 z8yg#D^{*JOylUUDLVfY>ZoQL`o9~CzHRId%YiVww-2Ug>6Cbkt+~s-J&Dg3RYTZ!J z!f0}5T!OHYCN3kHH{o~TQ~9fnK_%DtCT8w%uY;;^4J(Cr1L0FOh&V<9$F$g zWJvz?#)LQBoO2>>F>)V`OwCGT`KPa|JdBdgdHxM?S(;ST{~pFtk+JY%;UuUuAHT#i zN4fFcqg}dUv|~dr{8M#v!ScZ!K93h>zmL5gRs55acoc%dGyTicF$Cfpn`c?u^cONX zQr8zhQsO04xx3cY?fUv!E!DX63?~s|h0)S|l#rHHRWb3pySU#C;Y$J+z@LYdm&n~$)PZUkSV}^4QMpT{i|IqTU5@p8OZj`X}mf-U^%@#TPTtz029>uHV)fRGdFjrMcu&fxsx**pu`4<&t67SSrSx&)tP%^)P z#ES8`C{uSAt>`onX9e8}l{XMJ@E1GS^?HqJevq!_*J2VMf{!j9XAC=Aqt(cuXprk- zwaATp$zl~-dH`DE1bj0y=VE-g*aM3Syb`)}!^BCv&xIfCH1nvJl$DdFbcl&sirsLj zoR@CTW6W%8`vblh<|#FbV;fWkC1?K4vmeOB=aH9DHP0@_Uq2h$ltPaMs^;TcwXnBB zlKgC--FawDXT)G`2Fa+Z)BtpODWot6txsntTMyI`Q;1(RQc;Z%6BUK`G?-SX=(to& zvwKbK-%$Z0^X(LA1vigIB=1+qvg*hI-lgcsJ!O5yI^hX&sbE^JgeF!y=js7jndE7G>TvF`94&JRkU= zmJ7zaU*VT)ykD@#4O}$hRhQl&SQO{k+lYMay7%vvbn3K;!0Jh<%m{QK*@UkjX}^JFw>0Hl`6r-k(bino>d zrY+@)z*L$?yV&aG(KPKB2QH|zu5{mNR_~w2QaO=h-MZ(4DWPK;G9cw=^!#{Y0Du(2%{=%Kq* zfmOfQn3A@U`?ONa$V(&f$Eq2SE^$ z9pCZaoeZSOvM2dD^{>*5Ng+rA%q*p9Gmt*KedGD2uqFG+T>_5Df&Vwb8rpuwk-cmO z_vq9=tkPuB+>@{OHB&BZ4%oO$CH|WDOQA-+$CWqA%D6_3F;Q%wL7pN87XBG94cAQeebiOdk<1sjs)oZ)EU>Dh5ad6DyZ! zzE2als6|@lEwuE(qP=PC&6$vurV_^_-4PG7wW(aLqtS;*ss;j%l98NGiV>HIj(;B^ zEo5m$RHa1E69bLL8$n_gMrdMZFPCx#EZt zVarqV%X$*Jw0-gXIY=ddChfP{BzkPu_>$mu0NHgKIr(3{e66iLcLv5UN;Bpr|IRf? zqwmN8OA+s@H?K?a6d&%KvS=p(A3WdOe@{|vhUd689IZlfF#KBuS}VFIsaZ>`Qv)yT z+qTh)2h9>6L#F5-5Z$DZbo6k=VGl}4zhY3i-e)U)3q*xgtd}P~7JVG=%=n+B(&flv zCl?VJg^8d8My*L}1{lgArpqVydz>;?cP5d+aI#we?b!!*^vI{0A}BA_IxD|dZ{sew z9+&9cG0_*#5SeYHR;oaw`7OFSLVrO#D3OnFzwzALgaFH5(Ii#x_>TwGZG?;$DH>GZ zGJIz9O|(x#Z6#J{e*iKyDSwv9frUl?C)&(R{CWFs8|LP(>u1bY2@?2Cb{c!baemLl z4?bLCmcH?dF67MpyO9}iZmMfMyi3F?uK!joVI!W8o|$+hL3q~(9h020<{H*AiVM8^ zpwYM<3x*FUFU^ID^@{GFh#0g%HguHnA>kp_JK&DKb=FD&UE9fVW&pXojWITuN59y@vd9BXL_Gh$r$f!r9Cv@n9i<6{Y zLA7%X?2I&B4H%~&`6|TrI{lOBpjfP`&jWpXa1s4qL~O@{gLT`T$7jZ5X>zAUaEm_D=CDIOL+A!Q&3dWj zkjxukJc*Wgg5_OXHn>>UIWch~ARyp=igD?+ncxPUjjB@Nf&>o~F!_w^R5c+cAHn^(o8RZd zAnfYW1Nx7a3njjWo%@J5ojWJ!cBg+lC90n}X4wH({yK>zJ`N8{mH|)xdM^u8!C$Ww z*zU{29h6`2YO$h-eBy84&zV?aO+(}qHk-XUO6{EsY8A;=)BcT16;j%1&AzfU&~18Q z!c9Z8)92p)i?Hmo$6SKexlG%R+_|5_cl8%vZz~MmI@>NU(Q`vUQ~wlrib&^NkZ;4) zR!I|~pGkIThkS%HM{CL^500a@{7;FRw{|z|!;Rz02JNVmA zWHzL~J^t2%f?*f~$1S5QdD?&eBsoP%ea#oP_R9`Htb^kch`dI~rRs=&^XOqrZqchH zjE66>?1Vy+MmuHY!yR8|o)tO+#V$`CJ&@y<;-{Rge?tHU-am0o##~I&5HkW1%1x6X ztM>HvdXL^+)QulI$s4pd`fD;A0ZQGcdP-qPM$0O^L_jL(&$gez4yoxNdmn=AXXXsA zQa{lYh+tmbS#7CpfRxCS^yS5@;u4n~<_P7wNiM5$CS1FkwWNT}^mF_!BV8>iBk5SV zZ-%uoJW+dU>$u^08U7Eyu2PR#FyffV2W+$&*$jXA7=9^JAC2%ru~^Nd&DeUzN9pdB zOnNL&PiAxs_21*G-bxyMrM4oI-}M9EWbmS_y*~-{?(Wh(^}1q??3cDB!bWbo%ca&y zkuIgMRbrei8}_)r{kl*f^!Me*7j3R0HtziPAB7+YwGn=|`(>hwPXVnXs#0x3xn{iu6ZruFQ9Jr#7P?qh2jyBTMze zFePHQrXat?!uz_1>gTW3>{LL)=`z|SWS`PL=x0@I3_&ZY)vl=XUB7f`!t&{0p<~is z%x$`O2cPrZ+X;iqX^=Jzz^OzJ5Zz!j3TT)p8&v!xZ(v{$6&?LRhLF#)XLwkW`pcea z?$0RXvz8Q%SY`T0H{Z@pzJ?eF7tNvRWVjw@>`x#})41i*wlKc5mYGYMamreJPJ>#xrBK0fgbX~{x?`A#t9y1$S?qxQDB2m2Z}K3UML z`@|t|u>JMcZ!OZ1hx(_ zqH51=so{##yw!Qt`r{V7iyN*=kHNc`o8_D1GN(jqKD%66+BXA(Z`c<;ho4DHU8I+y zLo#7Jgc5`+t#vf{hgz}I`NirVBxN!{#fTDPBXqialw+_@q74_%!g};z^3dGoc{8HU zew73(-nygOcC6T~lqm0?L#VD^$fd~Y(`|se(NfyL$rAZ`FL zb@}4?DMQrRDZ^W+@Hek(j$FEoTd{KSCor3@xx-bwnV^r^#&Gic^I}%ttLwh+wRA(>)0OWrHSJB~0`} zi*|z@;rm#W!LOTAt^_~GGSB8PRQX{NG5lGv{dG_3#CC?G8-rBjzTDZ+$MrSD)%arp zniroj$EsD1Tr>}8hf2p01RE;@nud(TW6u&p9O2$HE ztSNW@eK!Yu+kCVOUQE~MkrD+ETusZ4jH;5kk^f92N5H5?j?nvYWWYH4S@hrotAlNG zCvBSGfxhuZ`;nqT)S=65l;?sy0m(HQ7B+w57`ey@K)hun(^;R=v|C5R2kZSb@|l|; zqcHd`EijR6go?__T^}Dyv1FIScaMBSLcj3|$DAFE?V8nfbc`Aw*Zwd5H0u?aZ`}h$ z#mHg5B6|S|erniT#j28_M}Bc|F*Z{%CnCKS`UJ8HAOjB`KX7P`cs1sxIaR>Bl_&lR z4YKv{a=fy^ERGC9aR1phh)ElM9W9>9Mc!4YK#dPiS}4SxKf2-a{Wau0J0WXBpriJn zm9u5m#&HOpvRH%>r`xY}AXEe7^UWbAq4!B%)!U&MDZ1N^!lECk zlE+Sgdhyt>%`r=c62%yHb@TA_d&Dz4V7Ur)7d(6-v=!5^+ra?RpIT{}4C8#edIy`+ zy{0a5hQlTpWU=A(iKBU6kZsw|47S|2n4$s|L{dZEM-XJ4G1B*(U%w&S|FnIWS@!na zk55Dp=viqnv7S-C&kwvskHC^~B7ue3(WI$hB^bSC)5z~WN|%+ts%dezYDDAAf;T^e zoSU>g`Sp-8Rq2b^{JlEu3Qx~nbYDw*OOlTrn)chJjR9%o<;Y8s*B6Y=gLj$QlI0k{ z>(UOQ5X}C=E;|1Sq60(nsL+)+&0Xc*$Q#kzj9PX(bUXELodM2Gm@#1^0?R$a~u&yKqyCHMz;KAf#vLJGMo1=nA5P(R?)?3fD2pV8m zpm?=<1iOUNWv^ec$H(Mj!3nTV?M_hcW|SCCe*P_k@V*^=i5sLEn|Jki!-wU)Ia?7Y zG~4ivX=NnGYTo*0Ty!f5<7g|pN6Oc`8VQr%{1Tx{CP@zVc;VyU4uf>fOuPxM30Ef0{vAmmDkir4 zwd4Eu@4`*JGn-Bud;AWoBMr;;IVN4UrKnATQFG(N7uTDZuJ@@P?5S>tc8)`$&+>9= zaWOGy?-06jqP-ssz{TaCBy51Q-9A6Qbp_+d5ng1lyS?`3F4*U@^WnP9&i|<7HSl4$ z~eg&X;>pha(HrjL*tX|oMc@Ut^nVFeq zgx*FR;*cs0TZr@LUZL%%y@TCSM@#d@Aq))=bp;VFH^qXI7I~pIlhPvF81P-l5kDO( za)XKg94xrL)10ywUm0hR^d0drN1Y}FB>Wi@D7zMl*fI51g2YpGxV1pF1HCL>iVmt( zaJ91!OPA4mI1rL^iKu;k{@+if=C*(BxHhq`ulnzGs@jR?H{+b<#lu@b znQCB*oVnw=J1O!|of3vUtEFgqffV;2=}){V^;DPWq)BVTh}wI_OL++p83hId)%Pl_ zOGFYNYh&M=ySC2e$bbK{g+;7ZV!b^__R0qgr*^L4ufDl+ev{rt!hNmzZXKq@#w1KR z-5Xd;3Ex^}#RW|<*NtLcMgJ$teGm{KyG7Or{g4*DW<9ncNIFXFG7RQ|8Y3R^gJT_8 z#9wj`zGKo5ydvz2YX!B)pg2NGOH%-BY60-tbc9pPtlX@0uWi|Wwfp8}vl){pWySXd?efw?o3IsSEfw!y!i1&5A03D&RjQ{4zWO?{(5iXf(pLY99p#5oejTK~{(H(}IswjA&Yqqg z9&%o=L^oX;r{v+Cn5n%+)s?VgaB(|$XDuLExo3sAV>|BD2v?bFWJ%Xk6lf( za-zUVz_FTo=+(`x9cPYdV)25eq+Pa1D-Vs%gZGzyr3z~_NO;@aI5ju6$ zFU!B#a+(ubNT2;;=usF|aF|;XQlWOR6v3Z1XLLt;B>moHBjr9%F3I~cKN?pU&q}Ue zoU-}Z`HtWK<0-oqaL7D%3xui9TPV${WXNI=))5V|Yrt<6D7w>e@bu(yFmC2|1cm9) z84t{@Al=F_6poj9Fh|~uY_kG%UyV7EVh%`0ROq}S#w+st(*yZrIfIeVH2MA+{h1_V z6s3jatck|$K^$$t)*UwR$zE@NMqfY|fDW=n#abbKQKffcA4VrojSv>0FS;^LdqRh(A@#clM{~%gc95E%y({*( z)E5cto}c_*MhjF^n)z;{hbez%bt?S&EAYzG(3s(u$~iGnWg^rIdSj3s9P0i$_mGeI z^RigP;Df`!!g+%LseH1#>Uh^gs-pWko)hZ#94JQ^S9|=`Ju@**Dt!gh13bGK9%~@A zo_qCj*e3$w?#|4J60;B^z~YUa)zRWQE~xvV5Xw|Q3>oHTyHeYsp6N~%Jf~@+EqY7q zGbar4;ubF~AsHDN&p_|ba`()^?QeIH!?#F zIt+^WSr_fgm(c-j41%q9YL?qq4#^S6$xBt6@vCc{e5)*c?4CWwV~_37gn6o;hBlmrF?x$0sTcX%A;=lqu^ZQi-2@%~M?ZG{6 zEK(-666{v779-9?2XtSnkC%V8ebDcw&vPwUbJbNX4`vU=ixX8ff#2ybI8>3;JVGDs&wsb~z|9Q)PC!N|;s28@%1?Sv;f=jcvQFM{ z5ca`365jLYXm-rH$)=~bH?y$C8!gd@t8Kn@_#AFZ5I)M{Su>Wvh~M7Xxq(D9^=o>> zoAzhRe2TCtm*h7O$maQFTU?~IzfS({%v06e*`gJ+4y|W=ih7hU?Ra5SJrz>lMp&j9 zXfF_jCqGCvO5&mPVjmt!$2R1R!%CNOa7g%%vL3nUUq1(to@P)da4D%Xt`)nkUiM*@6Cp_;H2Ydd%xn3nkdzOBl zlUem{!491!jIBDvkEu7u;hT`rKlx}zCr#!96QDQ8j!(N8s$o<8fO(ets zLD?X(N$mt>TlLN+A%}r;ydYSbe=_~}Ta8I30pC-ZGpUu*L&Z+^#)JPET9xG(Xs4|w zKWQr&(V%7Ny+q*{W}b|#N}KZX-rbkEQ*N49ZHQ^l5HF~eeX!b#Y5eh4%IST3YAF`g z5op1|zF-S{D?pZN_|tcv<9LnI)W;wuFbXf<-YYBlF}6Xdu?nNKDKz)!KhN)wnSyH` zrji%9m88#;@Q*iRBo8T*#%u2%-f=r8^T!H*nhySUS= zD0VP{zT?V{zwv-$oqoui8;ZmrMpgF&mp1Q$$}fmK&Qgeuj@C9XKnE=m2_k1liXLzN zj6*JTQlxARDbMA51*I(~>M=Pa^XsDVuz#Jt=#V}-a@>O`2sZK_T^;olJGnQk>rJ%- zX6T0|DMN^^?Y`D|ROWkP~C5#LCk->vfaa zguSXQT~ba{%YZDH{bpx}o=}+^*AVjQ(ZsB+(Q@Rb%ZXecbHHEYmbzgSgD6`WQFA^b zSudV3ktNNQ2w+iGqXjF9*zpf9s%Yi}RA~@bA9M--|HD8>a^x4qLM*bN_wU8y&`|DY z{Ss;BgwbUhIx7m%&!s@Q=$BaFBq$c6K`o{oe+(8*5!_42de0TRz3s?H*@#d-IDx20UpgDs`2VCob9zc#rF^83m|cTkYEv1)?*~&0&u)fDp>hfm3ya`(Ctt3){kthr zH5*R$W;oeFu`-D<4!*%{8*sn=$uTRHFddrL6>FTq=X_L;i!Qz`S~Li7o52(f@qC>u zh@yv1VfO+j_WLG+zXU%~soD&N50BVh-wy2DSG$Sw{vR1iJLdmkgWirpt*orb6F*x` z>zn*=zQapN8lDyVXY%3yz`1bCgZQq~b&}7W z`5^gvfWt#SI_lGLD<+x#(>bu+E|2#uB|r{E<^`LS3qcAU7ydH1m`~B@>AGT*I5{$` zx_ihVnDfwq9DQFcPGEi(#KX?GwF&E%*GWI!Itr_|JENDYMMBb%xt88hfMGpKB&*M; zlTndzffL(}hF-Cmt1de5Wm?ac?P#0UvuT^eOKYWF=-)U(*H3v*$0x#{ev};A`v>GO z5j-y0gW934ft`GN*4nKprl?l&%QLj2DfUCs_C`CPs_d2P)8>RnRVqWw7ib;X{Kay< z4?hfcyqBRmY&!ft^_wJ&>sVnxI-`)F&A~;uzGn0M_4au1;64vrK4+2NRvi7Btg|LzmKT%f8;@deR5Lb^Bls5fQCW;h6uHu(1|MFRpZQk z0}hlIW7KaMxg~AnvB3ufnJdtcdSWg|nV$!wzNWU8BP)C?_Vr*EK5hp2g1tCN{-m9` z&7>zZFw%A?*v=H&6LV@vLvWE2;ljS>OuI=(?Z@aaAQs|+>QZ4}#2mR_KRT$=F)`D@ z(4TEQR3$I+Dawd5Ak#pbgh=JdXrEX~c$0a}2hOGM7fa0D2g5yPjk#?xUKq-yKmP&D zTPfewf(f|6IS<#8S$-3kN|fsG^F`mYtQP_+>) zLDP3F#&1V6H#%0QNhmGXTMU1@j`y5JKiDy759j%a@_3bVSp223Bn~U&i|6B?ue3R0 z*(EtTYCo%nE%L%Bi`!t{Y3uZZN$dk{q#ckxmdnoJgF7Z+cZgyPlI58e>;f!*){G?u zG>A|rIvPsx6q)$Q$2@4yD)70RsHd&^YFk22@Ip_)<1P~kjXSJvDJM>VeDILOywO+o zy21=Tfor_iz7mH8PYu>|@)gZ*&Oe~cmt|lZbUG^k8J4%9Vs75+m%(oox3oqO3LdFM z4j_>XT8%cPx?P#VLl+R;yTSZ_Y6o2koRG?o1C^{sND#HLvHA3x3-qNy-n5e|g&F9q ze(^K*<%$&S5DZ4KF#6F7dI?1R3fA{rK{llf7V7mPMsCxEeTN^G z1zp~_0UHbLt*|HlN8}SUcp+Ev-k20T98l@`vnkHtK07(C!<4C;sKNy0|oLpJBNk4-zRNV7dEWYrZDM5;P6tTq^C0Z~u}gJrm>c8#7T0 z0+|L-FH7TdBC#lPQ~2csa2Usipd^7Fql>4i*cdyW#LZYxJ&8yd0V2V>SG#&IvELFIdfU(c*D68G$fd4 zz@hn}$GBVoc+Z@2x76p_CMN#Sa}UKPK)9UkXo^2VQud1aQ15Ox>E`zT*PZY0{M5C zY1|=W>UOPXxfDAG=p65 z(I%VH7G}Q5Wbi6ie;bF*X376@uv{ppkV4qZtl6v5{Zq;!ZI1(k#0Wz!ORDx9kw4b- z>#B8%DaY$Y;jc3@VV{p%L?y6nWYg`w-8PG|zj-M-dH7?dNpVy>uU#uH3XWL|k^)xA zo+BG037`3u6+_>6IORNT=7oZ`2McSJSmnc9WM?0=e>32(T&SU zj~EQ4SKipR+f>LFA;pP2k54%6iTkpI7jpVdj~!rXT9>7q(_PYMR3mYMgxUqvnU<$L1gfCe}cp@^|1Nr2MBJ92^hqgrX=2Xd0Uv^j< zx(BZ)xtFp{ZXYulb(XRZHxJE;WD2)1e;;lxn`5W?*EHM#KCrjMK3|l};<>tU&1VIs z3XSjtuh#q3u}8uTnG0XfWnPH4?X91bUEh}@CwrS`{yCj1;c1%jd)}g>@j@=Gm3{gR zF}>M{K`UKkrtiNR~-bKO?v_*8e7wZjt2qDhTKnJ8FKB{nH31 zX{7fTCnF&_C0+Ix3}16?xl866jI4H_9(KL-y_xQ*q*{?qFk!{9<2h4ewr+_>W@a8u z6dd!8OY&3q*W06u!~GE!lwSDdH24=Gmka`^l{wQOs@kcJvr;GbF*H*j9Ug#rqWnVE zr%$GJTWrRNg$$~QdA*_cEj`^AfTS}gu`uv0CrdDd^~Pxnhb|R&Of|0)DhiQa?GsP8 z5RS`4%|DRL6!^oz9HIh$v_D&FqPMZwvt*olf)XrqQ%^N7Zl>P7pYdX?dzx*sn3E{d z?)kOQ=zH<~?i#l!?I3lBD)CV7T_+$z_B;;)!?HRIOBFdgOzY z2s^-lF~~Ty9_4A1&JSyhG`{sQUBxT|8RpoqJG8wq-ZJdBdnLs0S(Xs;3t~_v`uC*% zOsS2P(6!S|s}zv_`Femi$aS@#D@aF$HbXhkB{E+^qboG74{NRN%6^Em&+eS7`J;7X&RCVS?u9JAR zh9i~KS%%4>pDAQj=?RrIGz=go9?A7Z@&rJFn3_FX_cFwX?&m4lNevz(o;4M;=-?~vL*}O? zr;r?pJ)`i7w0IDZk)CPmkGmqPl_h99^srO*km_47LD8*bEa${%aZj6q#qwlwC zb$Ui2elyDWSb!+4)Vb>m<#m?tCeR;O{4=VtU%(xAfzvZ1_5yA8R8*Dnhp22>@Vjt? zI``I3EBI|Y&hafwe!HA_eg^f!oKtV1tmVftn{09GEn<@&i?q1K1ih2>gVQyMYk%>O zrYy?E)rVXJ7Ey^&_Zm@?yG59DhSY|r|1_C#9)Z&@f2k+zCHw9nrmMt$v_nX+I3K~+ z#mH&W>)W+GL0plRGb_*Dz55{*azAZO4kQd*UrgV+Q-|%+4CDp80YlMBXomsN2SZhE zWbi%aJ8U=jOkb%X1PKwbk7>JW9{;5%^dwXf&5%Kiqo5kSQ~rN zP?{patifcnoYq*Y6ED}O-$Ab zSb;ka-@>=IGOBL__I6$Hnpp2;m6*?D2Mya4iBG*D-eKS%{8-4vfs#&Hee~eiPHOo8 zb-*UTF3=-TUI!(5`1h0XMP6Nef+ODEQCB~;(96h8+o<{Y@z9mYh+dbggkfZ?e~}Pc zUXXiq&0sV#Az>sEwZkhv8#hYTtKK6pEYRbksOXXC@g**scPnefMk;VL~(*gK$LXwMS+FG*t)b?`t6Ig-(yKSf7ntSnk1I1X=JH9{Hy7X#@*@1 z{ZW|VPcA&bH_+jVH8wFx&d9)&?19w0>6eg~7qhVoaWR^jRBmo=p%ID?=_nq2VC3I`-It}*WF)$Fnmm3cy+;|>aNkuh*W%)9YWQ_78u^W7 zWS=y`JXyZq?s=_RKdAtfdv>;VlA0eiu}IJUXRLfX)@{fH>wf_mh$!iK<2rFOP{ z&B>{K(>OwZQr}?1d+rQvU!T3Qe5CLAXgW1f-;!6~-n=l+A~;^4oLuPzddWE?!(}J=eFlTKHH#z5)qZ^}^^bT>k&G?3*nEnr2ZvhqM8h(qb2+|-W4GPlD zP(w*~ch}G%F?0zM($Wpm-5r8}bdGcjNGV98D9U}u^Sk$d?^^%0W|l`ihjrHazV~^b zXYc*&o$Q~Xr%gP;YA_^STIzu9Rdy(r^===QG-4iKSeP!Ak@PT+9E-wB(C!d!)jIm6 zQK%um`WfMQ2G&L~@V$V)DEJ1;O6;LLOo{G;ec;-NUbE=$MM%}ty-+E`OAZXu8H*#K zyAlqaJ3%jWgd)BR(lPdjsSS_FdVC>H^ewLub zmFsBfKwm}=wKL$*7B@&6U$llop(DHQyRLi6I_ZFZ0K3CjPdLoENP&!h7cC?k6U(hZ@3SNexGFG93z_r+y6F{s-fx_z2X?J<^vIj_ z`MlN5wH-fS<|4DMUElq@VuBl8IdrBa6rVFPBd%YWBSfHibI8dfcx=)7vcY|w54U#J zdDw=}IB-{TK1o;NnHHr1N7SGWF7>3_PoKPGTcRZUz-r=-h{6LwDxPnKu};TaO+EAN zY)>{}WI0}B!}_1Kl6ViG#a1#rZmH7cA39LnCUjA7D2}T)w@L%s@VXp>`sz~~nvEU$xEnOJuyIUH%u3%DkqOyioj%f|Ae@68pfSTyJTKaGa&JeAshW6dOzmlnn^v)LvK5-Hz&qQ z77cq+4-1l(om$K@|KlhfuEF%PWs3Yu-E?LFE6B*vijD0ezp#wT&RnKQ zWPc8kO7Xrr5A-w4qC6P=s*M*S_eJEWGBEo z5<)%LIM7Jt`S8osDo|6|pXG~}g)ks%{zaZmG1ll460dxP$m#@If~ z51;GJTS-rwJ270YHJauV=me)7NG_IA=naO}5Bk>+3jd@pCdTcO3Q;q_StO4vlPr5i zM&68>zYNR2yB8oq$ew-Vw6^`xiTH`sBE;>$kb5?gSE5%^^7h=-hB<@TFdHjXW{sAm z8Wrtz$jXYLgQMeeqcs*C9UY~JkL%EIRVB1SrUpoVOG_ibU$42eExiomlW1mpvcO}I z21b4$&F6PskrvBmq{QWgLKT^afEA3g@03sd-S%M2Zsp|8R**)D6h1hjmFeg(Oz?0W zHzZkLEYX^^ugIdLOaJtXB7Q#c>fJP}4P~BC6M9?0apy>3oq3$g-|D?G1Bsj6Kc^JS zD=6zFn!=Xhx!pET;}j(;@JJ5-R@>k)K+VWehHhJnYPDYYgwKwFO_`z0{q#-$mX(ld zj>emz7I)%_0MlWcRC@=mnT};;+b;VX4yJ=9)?${?5_Uz&Qv0Tu(n$`83Ef)Vj0cZI zxwsNkmzIk4lQ(p5x`$8t-KgR!zd&~F8G&8(=jkleBZk3H#gjv#b>sP$*T9a_d+IM}u!obyr1{rvRVE2nlfex4wjBjNFihJ0?F5><+j1#tJ+}!ZqAGQI;HG`P z6At;Di=J_R@24)t?+Ln#s4WgYTh5W?s7oF{B+|)!HT@x;oy5zDPm55$=Z{-!W^l!A zLc`m-hrhz^b+V1l)p@A*PThau!!$m)gCXn3TWpE?>Fk-5iGitq&_lJFaL-j^r-0n_ zUsqnRiU$mXdmY)u+n7&qhY!UP%})ce0%Td+o^g^uC|w9~zEL%Mbz(qPYI`ILSd_cb z1>I=0h~i?(K3(RFu@INEG^3lg9a(h=I)R1OmGZm%c$5XRbwmxqVVWMao)lf;h~9!L8y! zshx9=;83tY;j?sFbfvBH9S>(`c4uejgHEtXVGD=Dhe%qY3uiVr)q;b?ljyfDZpQAb zL=HR+b(EaPC+0!5@B@E%=5y*J(%jaT9^)JPh^y?|RNABsNHi}fD|@7nG&axsx{p@1 z%wuRezJrMPF*w2xEv(dM{$d3^=reauigi^4`#n^JMeuFE{r^m)aX2v7WAMVXwdU?bpba*WR6J5J?CY z9V*4Kx$c>oiu2EXc)OiaofX?=8KIe~>~<62%M?9?NcP309{0E{?zfYsstT=09(T-8 zA;9rOgRZbUtvmILj5|bO@sG1)a0!J8+21TRvZBtQT=74+Dw)oPX&3yqi!F#Dvn;|q z^hrc)utf$F^&j#zxKzd9)LndPT#o7kcuo`XU)oTjb`a`W)lOJ`>5M zJF7eWDH?nA=H5Hc7>cUE$MI#B=_a{De_cj>LN122uuUft->BNjd~XIN`>eGS-Q`*f zlx)#@zi9-lC?sp@9$)T#xC$L2-SVe&*y$m9MIyG0Oi3B*ba=A;1W9I=EY`QSC40k{ zgilneQUu{TJ%>iw)9g7ye*?~bkPvsIMVI?ARC^M>wsEx=e@l;1w-^K5h-XVxK(dm% zAV5%yOc#1&7O&g|azQ>#I_tanxjC?8F*v!>U&-$#{y3jFl#91%Sy@U71qTNQyePn% zp9qs1RBbkr>~qlE{S9e2PVWH#c04%i@N%sZ*KP zBT`2-s{0PdWGaCiqc8&?qLebLZ;3-%PLZsm?+q1AX1?iCf-X zY_~~T>2Trb--3!B;`OqE_{N4t_8X7tYqr?}og{0K8W-?sZ;FF64UC*|6=-Q}2P({U zkIkru{Ec~x&GXZ7ve~7vsbDt5t)u_2zcK50KfKjjo}gp`E?G#4O}lqP);MAOtBD39>|myzYAiy6k_K z_6qYwQmFbN!*|B*3w?YWxg>fZa{$)yI28QqOhjY(R6u@f%ptHV)L|SC>{@{-_m}`0 zkj6PUIMlBmYb|ghXC+DN$05qd^FP_7&(P+ER3wx+)7*Fc{Neuhj(zG_S-PO!UdsON z0IL2^6nD9GaL`>lIJeHM7?XS4z3cxwK-S5z7bVSC*g!d~wto|RM^;obII*#~+xSx=th4xq zK$A8QVUjmk0r5rMeS5fcCl~1fr;0*&JR3b7#8y--db`N{MF&EhBSskaEw|a_>hm?{ zRfF5M>I9$;D-ARd2g1n~?W|HF+Ap*u_$j!%lq8gGb8|n~h_~d5iLl3PMhB?U+n6zU zCklpoE!-vLUiHp}CUJ1NKl^wv%7nmji(YF5Ydl=zkEirwiO)l9NpJ?#4R2J3gfA^gV8o8N*L3`WEgtxn65&|Hp4Ub@7Q z6wk^^!@oCRlLxq9?taZQt-XEjA9~@q$%AGLAZo|`! z(yFkqXfxQ+_>Zp}?A!2K$`SrhiOppq#1>@*6a4_{6iu{MkYZ`j``4{hE-T*tZhj=< z_bMu(1wE7lPa1CzqJp-I#`NzsX=BWqr<~3k*sXmoDj9fRt#i9am_?js%2GMw&mvVW zAjhIB&7WeJm~BiFt4o)sEAUrE>}wlp{f$g&ftyWa)ZTj>$cgg{_VTbPX)@8rPXqX) zleFdcwN)vVJ&vAWP`c3JfK$t#?fQEVo=97rGHB;)I=?l6@l@NwZL*c;9j$iL;1V9N zV*=)Xt*!h(FS+H~PWd6nWr`aFVk$ELqM#qflUkLJ`nu2En^dBlp700ZnZZ+ZEJg(a z)iR{BHE^{8k!-*X#NGeGiC=hED7ugfutx)1`2WtZM7g?UgbA{q69VJmqva-%A8Q^W zzmJ-b>eSiRkc^RIFOLBn0oP$F@6)ul?r_ljTa(vtQ2+#l3d5O?frqiYiu)qXj6f)sS_4$0woaShu`gsI3R^a_ z)!B5V)fWMYE5Ne%Q)R%wvr&+VDK3=ZpjJCSpM5>ox}x8Wm-i z@DWkCTP6Z*i-VNu3^NJjNR{r&8P^bM{E)n6dngiF$YuS{v2!Nz$;OIceAqT9XfgMV zK12*U_w^!_RLb%{*l;FE$S(GoSFi1?D)MGv&pBghwz0pdb@=F)6ibQs7)nS-I92#X zr*jG>MGLV5d1vwFCpwce!JQ}^RaVuv;q-U zj$fy~f-R+lpvxk6^o#y*tR}yWz#r?w$m}bRSrvh(7!l$L2lnt#U0-UCq^ws_T^pb8 z(-N>Y-l=CbnWXn;?sG8RE?!`Kb&NY&w%#v&?*$TN2EzhS$??AEI8l3qDwfM+jaE=X zD_sxt;|a@%3*BmrQF-$h8w{+o$g$xr{IknFv3>8Xt3!RCpZoZqJYRVSGbnW{iW!RI zfT++x-j?i3Cn0*U5A{Rf-SV56klvry_<#LaorRH&e1drsu|d|&%Ys94I#B}-%BWd? z7KL(7ISB;Q9$!jMMCsv@3EE1Vzd_sJYn5gt(~FcG*}F`lfhX=CrBLBG*d*Y+z{nnW zRka}jK+i1pptVWWVvN;Ihh7P*4h1x_O+UPzD{APgIoGT$D~tXzkWsH$9zAqwo*43- z6zz3b@07GhS87@u39P97fCZhkMEehcdFwo{CtL;O-Dyi*O`Lht^`NkeX*|`!PxC@H zcL~EtZO;4mxdCEkwLv79;b1LED-W!@0vt3))SskkUXhEmxV4o8NChkRYQw+saAPxw zu|BjojMFnUZF8aemxKk(6hI0Lu!gwDM9Uf_v8&7z6-I#2xRS0FPH2Nv@_zzZE|Vgw@Tx4=-}m3z94<2WHnM)#mahm@<7mT zW?7zJW%m(~mZ%Ns-B+l;|0|v3T%AciffY;(SY4;VcSmal9KC! z{-M(!GvId@hC`_A^-;6ifa#;$<7YVr+++7UQ>kQ2KK&aAQRF-*g%bAzyc3;NS@rtZ zDxpDjA;-BSUM{@IX|RsHsU{RdiK4&z){{Ps@iXfMa(X_|l4 z+oxDR7>%g3;v%)%_-yh^fI+(&reE+wJ4@Xx58nO8DTO-&Cu3bVRu^T1AFmIWrp4%w z{#sQv@eRu_d9MW_*9D=apAow5Q6hB117WH%Z@d`QJ!A?Wry^(=g)TK8sC|$h@Iy`r z+(t5jTKKxrcFr7Zd}r&L`Y~y|8BPuQcKg9NPiLrT{0uyt;Zi&J)y|pOr-p z`Z`4qSZKX6=m-lXOKI6SU-!M>BK^33QShx((2NjSIxrcK(0DaQlys>0pIPx;EVG~e z#lnKNKziZ39mVK_{C*C3^mA=rOwPn)ohSBkOzb={;S`Y&1lK*WfJPVA2j2x z-42?af>4+9AHjD&!1fnvf+MrY<&0DlF0AK#j8)|H1Ip#~wdt9gY*AviOJ|huNp~{x zYtrTQ7&np;m8WiD%(xtTXWn#CQi3gEb%esj>mnk1z4%j!>mFz(B4lHvEy+a#QS*LV zB)!`Fz$fgq=34W*hKe6%pgdAdq>%J>2C9|S=3;HT77|@ z>E3txN&s^&79Rf4!G`D)Q4H!P?s#nUA$HzssCQJZen8H(32#|&*kLzS{n|UVsdJ+F zyW=}gGnB#Zv4cbOB1SstsX_HmsG0_;T92A?0{G$vtlss^PyDL!%Aw{S1kI!M-ohSl zuz_Jq&E)V9Tc-hA4{NwDszr@{g<`|beWl(_H95IQnaZV-p3yV1F(Vp{!S#_56(1{m707>}^D__B3qQ+NWBjnn#rgR`qm6SC)jScQ zx_uB8Vz!TTcSfe{c6pzeenCu59vmEW4`igXlEjEyFIkCPOz;LADw${XG{JooK_8b;I z_7LMQEdxUFlzM_5rcZO{NB8*}g>oe&hIi8JH_a`FNq8sYN=wK-!dZ%mZ=B*zdRa{# zFCB6>{V64H-PD%)hr{Re)k31`C3y_Z!TOHj3^%EDw3KGniN)lI{XyLh(K7rv#`a% z5q6w`G55@zY5?DRSUS;FI~Qc$wrbGkV$vb%EJ(@m2vXLS2x2&aZ5U9b4B0^l z&>xQ7pBMYSjZSUi~Eke_zw;}N2V1(jFV*Y2y%Z3qw!_wEYR2sdQM2NG;SV-PPx zZek7W``5rGalloT*>1=qYW>5l2_q8|szqZDoJY8nADs9-#@X9k!Cs|1xG@^6$J1C< zapUw1DH$JRRt@+Fm!{$})zC+-f0+Djyw*J}!%^n;*)D?6wz0(3M%v1}d<4%_am#JI z+a)qCa8YonWgucyy+9%z1NBfhFr?`$I-=@XVuuf)W)7P$A0;qT3TKuD*UOud)rD5j zjjKO_oOV(%c)2u~w{dQ^3a*L|D(dug6DT~OpP5Ug9(}Icl{oeOKJ*6~6AL+KU{j=Y z`wPdi?8S9mk1H0ko!*_6mY8bhGmY>bEqlp7zC2L?a3`U&Nx(^-eA70mQa`IUNg__7 z*J4WFGu8)7CALhpjxz5`Fz>=sSyaF&Arm8E2!JZOGsTsbV_y;-ic$>PC85t{+ik!~ z>|e`hPZ{763bgTg{e5bj`Eb|Ony-cb$$fka{Xxq_u-YFH8F?Q7bAS+8J+968@x0m;@SC*N)h+y+Y{+T6Diw*OtG5uzY;E|T zoL{J7L2f+hXi ziVdsBN2_fK6fe6Ir>0Z^*JfvD_u<&fs%;g7KmcYZo>HhkhO=j6vFDmkvS23V3i!=$ zoW*<3EG!Ix^V{~|1m-4kdpaPW^wGI(9WZ$rR~ImYA%~W<6Rs!h<9d&>C(-KnSYT5h z9a1~o>XIpbKu1{tUurGr1s$Go4RO`Bwv*ls&o6cBf|<5Wj(IK$w%bGo-mFO0aqpxH zEc9QKsg+X-X6>9dh?GsS&}%|(klak79y|v(bAg0z6*^JPVd^lLb(;KTpt9NLHw85ouV99h&4dZs-c$)~AXf9W5Y%73onMzn88 z8kOM`3I4zjBj#;rxd|A^+?h6Opj%l|mw9J*7KND^s(vHEtB?eya-cBccUl5b$FxM4 zsVG_L)3zvp2Ls&|Xrcz3_|pe6`^@o+G>o4S?Wuy39hS;CwTe8b1?N#OjY4U)YkGfG z^=`K;nr1h4gVr6H(X{6GttL6`_n#n_-alT*?*`aL@BpF5z;cl9koyUT@#jA$!Q#mM z|8(~Qy}6BzsrNrN#z83Zz{Y!j?o1n(v#(cvKnV-bUyseeA7`U#K}iXLM^*usMKrlQ z5@nSvC%)7EOTEE&g~~&J4&=nM&V@}Ku5x~Sk7I;p@oHyIIPFX$g5R_q}_`H$XbuA{H zzKL`e3kT@jPwQ5vRxl2d)!E+kxL5D7Wa#k(1r{Z>m7O^UE$GDPAx_LP3_zW}gGX$A zXB%}bn2TPsqy+rRhxsq`?aOkLBs5dMzS_%hC|h3`*!c$%I@i4{e9O+|&eMM|3X*69 z!<7@IE9K$A5mt4@t~arBzyvO+invQlSqY8jya&Ghi%)6o(N zIo;1)8R)Or#M~^*@L7HfFTw%B0>D%V1OoQg8}zFwr~Zi!>X81J$W0Wod5|Ymw+FwxlGf#Ge%>fk zlWqEIpjaT{i`!vDw$KH^^)Hwzd5~my6&d9;4)Rj5=J=L$&p`u8nk^J4ME2%qPVB#&3+i0AE06hOQpm7SY1TWk0O`1Vq% ze^oNX>-0;evbZamXDD4M*97!*1=LJ3YJaQ72h#lM!0#S8P|s+P4IH=zpH*QMrI*0p ztN6qUGadE$@M-^|!S}+(9dKniWVn$UI-Z|)JWUu0qVS`n1p|LU__J$3=k)szDYoOg z!uMR9$c7qOp+jknbKd|yOg&ZX?^%MH(eCNV-QSZs@*&E_PoMgW62!*9h{b2$QTt8g ze1tCG_v)r;O2gEkWPb+XI=FGWPXbSNX7m4ErsRuW&1pGuyag3rx2om#?%CT8sIrri znYL%_&{zkpUPWxD2PzbceSTiPkg=*z zj8uSVwthKF+a>N6Rj>-2w_Vr$qdDzf&~Gmj*9R`m28-~}!qiH0!`P(?C2MfdL_%L82C)(_+Mh}Msqq#-4gdN`5O9_w>>Y2cd!Ifp@LK3Lh{pf=XshS< z_Z{%fhHR>jN{1^;l-G_0>sFq0c^qlPoQz1IZ*Yapk!6; z3No-IM1@Su_vEAZJ1|~LAX}qy<7WV0k(>wH^#QxAW|@t^LA$1VU-Tq%w)1=Uh3dv- zsleIo#k88?-`tacNI;d$A`lJl8fuI20uQ7DK)`q|B;>*A0L@?5aM|X@i9ID^Um_WU zzC`pU>$X*;y8@;1RAatcNwWWk;a1w3$ouK414zjOdfl3S)zj|9}M=4x_#7V3eiDmm)Zf<>8-EoRE3Jaz_?8Vu%dS`KSTE+Z! za0cM`j>WK|*g`5)L&?MCOt0g_&Uw5f<}z=BYbE*lV~$slh3GYxq3?X3sb9+U@V#o9 za=Rj~)zR@0F+?{=I&J9$pQV6(zfT2h1H(Z0BlqSChrLl3fIJp0<4H&W4^dw+ec68Zac2MI2M7PFv&;DCSt&541A zp(rkgfRSAF;?K$7H#7lvkpZW#X<*}N|Aqq*5CjV0<75BbZtK|6;y@BsqOJmPK$ro; zKRq>4qi_iC{BMxC4%X56AJfQt>Ek-Mw!p)_^4B^D-OKu~*n;!e@BWX!sGQPL6XkRE zoIqx`=PW+#qq#eyQNw^e5IOruh&+iG46_<9)fBu4XzTAQBICGbw%SsTLy7bVxVe)# z1Y_k1pO;JfLew;!>T}y?HKEZF_lgpqMaDBDA<$pllF*{$X-Aj`G(0qZn0>Zq<6}I# zO%_G^nTG{O&B)>h&v$3D>C3GsW*CH0~kyobU`pGBFNZ zMgS5HSZ&aH{wEs(s|NP_GRdR#KuHgLFu?5#OxFRb5!&~;ID0MptM`p<4``12$13)w zng1PVv~8A7T67(xTjx-UHAzhP?~~qO({EiMQ_9}S31)6?{^=ZmQ4Ttfcnapy{+z!k zt^EeNUvhGCT{~^o0n3JzJELYcGtnW&KiZz*k2Gaq1~#*3tboQwc;K(hUoJL?I67= zmW0;1@5!fWJ^CcLhW0gUve~&|3*(KeYY75^V+{{m`*80@{vmt5ToMi~1?e9I9`}-y zeyD^KJ9~!cMk;k}{R>u5k^| zk5xCO>Q+RC?BNFxc(Y@m7H2D(Rhu{wOQ3T7cp5v2lfJ{p;#D0yI2E)o59$^Nz#uV< zY)k=|+_qi7fCL{FRB1h)6(QsU14XZ{t^iH_ERjiXI(B#mXgQgP{#iNK!`Ci=yx=Q{ zVey!}-B_p`7iX2G)xu|aEqINYDv8p}lJmIgGZX{#&RlN0NaOT%v8)Y@RLw(I+LjM?pW@ot$~ekssFfS3i}nl=Jpbb8`i{eB^tkQKJ)E(AAuVrHp# zACRi5%-YY`#1xYBHytbj9hvLbX7l%32FN%rnZk+y|1k!x(w<*58vL!99oL>-E-T;O zwaRMCWpKY}d$2wF>^EMC5F9dm#w(Ph5NB-4m_0kf3RR6%R!T+^TKv_#V3_GF4;6B3 zG{$7s_6nc))Vpu|)MM8>k+tgG%r{FwV_)Ve^ZIei@fhAS7&7NKch#@Yz-c!!dB}fL z4umBdj%J5jn2sq3HmzP!`9IgRS*3*EQ~~#>e4w1-Bq%4L(cz<7@C|5LRSHWH6P>se zeX_~zwB3@;mWfWmG#v#`qt668OP`XdjGz&srQ;OMqGTy@=SpkEz#c`8TwXuRp3LYL zb5c>!cW@lse3=TEjc#pXt=@i&iIj!C#;`DWsz>jjxGw+>khkYnrfuHV&$At-n+Kf? zC_HGR5wp6lr=+fhQU9o?2-KbGwA{Qyx{BK0Y7d| z&O-14Uey@2otFx{kBX%BdjrU6nc#fTT&){o!&K=N2&fYtB@9dX`wN2&Z_TR-%Lc(+ zq5sIuOkUf1!R!~E%a87Xs;X3ag=7|3blPrJB-{AAkP$16garny z8kLejYjb&vy8>*JCl}I%8WTrz3BU(Bk!EjKZSll=uEp zFw^+fjzOnCRtZZ+hbFq#VgZABant(8R#o1h{6y<<^Qo;LPxmf4nIoxM##A^J-7BPR zo+r4me}cpvyp;sa@>YmQM{~3jk9j?*V$BE@CTA?lZA{Z!cG49at z{^C&IOOgWV8W-@CFE224G|0Z{9iBv6KooyEs6bvee){C{c&q2R?u5X9wA7)tM`RYsqZ#mP5@9I115Yz3w0qrcZL#*a&)p}Mco|K-ZE zs%m&)E!K%hE3L|;;@9&RhYZm7DO26j0=3-w% zhYCW*WuBDVZgTZAXVQl%jucyOliJn;FLHf*$Yohp%avl#GEcrTU*F*s%qzoKcI(4m zI!Zy;H)=uK$<4KnyJ5(E;cve;x0L57X+uq2I z&m+S|Q-=55h5n05kx8?@4GI=|XWte>(No#C8=~ws#g5LZ7GMg>1`s?DG-ZSGk&5yN zSGg--I!_e59AE+$K0bf_2(K2nShA#UKk5BBn6fLd4xh!0wb@mve#;DiuSkSlak+Y@ z4+C4(kRR{M_f15tl+-%Nhkf&%GO=6{lD0FqEVB7iHz$pOaPfPi$E+u=VHTU7LDk2S zXo2}N0h8KW6^1JJQ@aQhxkgIiR624kCFhZ0!^Bl5o=F^rR+rWW>n7A94Tki|f$+9AX0ZlU8gFYG!myyWB0+992nxbs+6P3Db>^1Hog>1QlOoX?T$!^;oS z1Ae^Dxn2{x4Y{0X&#CEFXI~TG7TGk(qtPWH!us|;nzvZg*nc!!jIF7~uxI;FN_=jo zcvz~CJRUQ;JNVxIk%O3=OL;nYd2U9yvI^goexkQ5Q^tIU_Sc^mL_F#ICW_`ol;E@0OUIWYYAzN;5}rPl#sCAm2?%njB= zXweHL_fkh1mH$coXO@=%^2Pk0s^VQ!#Itm85y&`r2#4hBvxU*)Ztx0D)TL{XB3B%PfQ6r`-^x6zeHK zPE2=_+h*a)1Zf&3ahU>ljbx|z(L~F^T$9UKzG#s;&$#27Ofc1g8C9*`M>%>gx1T{7 zo!R#)^O~vfX4rRSCVN6e>Zm7N4}8)t)-ji>#TKHC<#ar6#ZZJ>^?ld9C_HCd67w^H zO_U3XZwe`TRdq(^8K3)Bvm7}M3z}YeL=819zzAgg)Jn*XM-$0k%D8Zw!}aU+V|@%w z9t!*}U;a8wjo(*lO^55QH1@T1#!m{RMgXEdQW2&aNhb98sh?PU67$=M@{hPMp|co> zYC^~_Pm-0^C0D^& zezo9&bAN*iVW%OzFwYB@_Jp;OdxjjZb(xYnd=lY{zOX7{CE7ROm1vr73p>%*$0YkR@%u%IsKF78=9 zdHboON%!lTke^q|Vycj3B{orc=#!@+SzkJAdxtNk1JjYiztd(sL`nwCL@GMw?lK|=w3 zZjO-0US(Yy`O#)?ek4th&)7MDX218|{gv^J%fGwbN7}*MUY3cMbai#DjSZdrNMD^l zTJK6Xa_I;MtSZ?irkOq=7->iCAW3rCBh~a0y@|q`pSM&27k|D|0I}`l2;h0 z#-#wvvEw$sTITAS_{r~w?)4MH>nFFt;*{Rq^XeN`ZLZ|`7X!71gQjM|`-j07+zKfAM z+6;cx`)9S+dcD(g&1=;eIp!V#JXWHP)i4bUNIQhg&|Vh=h%9&lPgpR?$`&D*6F5m? zK!Pi* zw70f`9vs6n^@{YBh;^gStD7q7gpfwtGC=RFEO#TiKa4?AEw*KL!%1;ik*ctQA_LX@ z@#Lw4UEMFyg*#`YOWlb6+8X{W3YClBB3aXjpe7hUhP|<@i9(xWi~8Xa{VE<hMVcMx#tHgV#6q(J>^fB& z#V1FixC&+;YYvb>WJ>NV6E}D@Vr`;cypCG%V#WE!UiH;~-IB$I2j=ImE-TAaI!ik` zB1?!?MYRm;BPbux1@~G3@p5-FaW%a{|90Krc3pUlKT|OQEYfcT=SR^;$%DZM;rclN zObV=!71rOpffa;4IZXP13I9_T63BOJB(tw&b>KAQiv6nf~WQ4@RhV*R2PG zq0zOYuVN({cN!PBJ@~gg_^o(qYgwFZic$?vWWm)oY?zHlsB%T9hF3tO3D5mj^6l0N zFslg!=s!Dv8uJFEQtNE|T4>`iaP*KUw*fezD-d|QcL4gqX{jFRj2kW)8y`;qgd{|m zKuD4>95uWHAT`{WL7ob~Qo%GGlK=Z+3rKzpN>prYtg?ZwnOQH>w>%foaLEvTwusFW zvy?@BC$NkK{KH3+8GV5|h?o+irOwGSAQG2l#;zHvmzTso*O3Yj#Qw3pV6W~t*9+$B`8~2EX4IqC%3v0*nm5{bC!a<)+;oXUqE0lj z3vU#{F#t<-KxSyL+%g$G+YnH8QHZ%WmN*ntk-F9xaQm`$5=zU-pR_DLjD?)4R(CyQ$( zht$c7jW2HLt5N*|gqBR#<=4y?xRt|Zd{>}wRuZG|oRU|2LVz85&09GYTZKF!rSmyv zY*ogEEb=3Ec2X)O^&544UEUUcnsj-3pG$-sQk^p(HDCvo@4SzJ!tffatqC{3iL9>a z@qoM2z_@y0$J|QyWANKoa1L3>vK*V}3kI6!5Q6>HX^~;?oP>fdo8L=uNjW(=DeLW}QAk2c{l8}J z%)bp!_3#Z>2IMlJu;OEoIJkNyaxM?6EMUR}DdS`A3AS!4<&({M^7M_{Art>F(`~~1 zj75)^?OSNnftMP9ha~aUzPR{EsCEiQ^H2cIMCS^7bF6-g#>}ir?lXV!NL#1tt5x{e z7M!%>N>-q413fc#$WiW74m8^e)+|q7`AI!`+DPwmonY~AUT4LdMx#Oik7Bh`-FO*OM)LsulRF_7iJ`OQ7!O7kg)@cw{N9@-Gy zzDTG++K}W(Q$T;%v5cHtWjl_YbtSt)mTWoa^oM1zt`(pw#ZSSNg`@w6e?pZv#EY{} zRXk);ESBUfwP!odd4-s=%v9@0jMZ5{0O939sMz}9APT*x?3wiRFf%NMK1rrfeA@NW zmf6^p4#ycZQ~C|Utifj40gqk2l9bt#-xNG;MEOHh$bD(~c+^gsKZ^^xMrvYO-BN;I zx~!*__~Z)KgevDK9KYZNuaS3wzd`ociTCH6&}!6aXJ?gu^`ri4s9x1nkSWpDf0BAa zBL4tlQidO=89O*b6n+wX_nQ>Pd%%+f>$8DxH!SXWw0v$xV{{&fmKnTsc(x3@bg)^>UZ352U;+yIyZ7>M#MQS&50I*HUhDnd(O}>3p+(E%kckpL z>Uy9naoXt4H0IDlu6tT&d!AD*cGeO`Ndq_e9I-(1u60wel=~;>F`DsxB+dOFeKMiC zJI#a@N;~lh?mhQFFgHev%&Igxj`zLGbI^w&^&_`aC%3r3o^Zdg&Mb-E(8y@H!RpcE zmMTqNS_`<%6c4n&^mL_VIx}Ql&mo%2 za9heBi&0pCEwR%~(i`c#oI>~S`th@mW|wk#JlSnlpvOLr%B-T72D3!r%4VE5TNap6 z4-aSos>8sWG#Ro?@evAu|1bJIxTrA_NpU6oUp*(m*?vmG04k45d%h%3bS=p!PD0W# z%?loKF5kc<3W^2@Wk zzDe5+RTuiWpQ0Y64b!eXX(}M!Oa{AAYX62baxjv0tvT^?j{O>bZI*5nw{P zI_GJ~MuNOK-9f^v2dIE1EmXQMUvl1^bKdp$-j4U;#|)O<0{3?h6?^ToPHed8+0=^oZhKAG^I{B9~L!`+rpq>9g%7JYB7Z9YG1N9le&g z^~{wJ?yp>&3Zcn5LU6W$5ITeS+_A?|mAe27%88$3L9E%?jN7mp2^evcr~eS0WRPO4 zG_R#JpdPm{y^gH>NV!vOw5pZ?S-5I>a;hRj?5Ac;3g)@^Cy8-9XdZwJKsbEtnoKaTaJ^bjO3?$5iP{QYgW18C9ew~=cf@FP`iRk=U z_Q&-at4@{eF*BW|iO}FYNw)Yb)exr_p!Db;k1$}RnTxjmvT^ccAF&yq*K5a=PYPRp z^6?03d$}>&5-Q_W9=Ya9b92Dt*YJWnkCp^?prMG=gu}`6YRSm(t*x*bt=E=y5pfLc zm)|H34sg{>gismqSR_r1pwZzf-Ak{De^o^s8R)=B_@1&SXG@jty>_nuZFeQ*L<&ij z*WOtE+(^>>-YN|6QCS?_vyx7F|8e#;?3`OR5=hQAr_YaVdtOhvSP#cjb>B?B+ykN9 z--@r^#N759LK=SMFFi_22Q=V6D(Ud(h(1>pNH7p~@Ub5j|A$Iqn)FVHcP}B;5T>@J(=TW!R7r z)oRx(7uosUh7QxEX)@RY;tFH$xMWFd00$^=oeed9Jsej`kC8xLqDpm)iZ8V1LvYWT^&lI-t=u(Fs1 z>}iI-tHNL8WO;3j`?K~)uvVIyqF5}U^=k>1Y|mX?@o70ue>s0RTGJlh&g; z>93L89s{Nq0FO8}1B!SC`pWa~ef6je(oSnUc8vB*ck~vLsV8{?M5GVkDbi+0(2<&r zXNGFn{GM4{)SegXWH#G)kQ8c4JZC}315w+?VnUXKw>0Fk+4{(jYQW@IRi$Xh z$86IC^Ya@?h*wuGc?-vQ4eMt!;3j7mtl?`v!f6LEBiF0iX+b&OhV$Kz9-En}#X~4L zJJZ%jHy5ICy_KhCY4nu3G`Lj>X<4~UgKP+xc!Re1=n+$9O8<25H92VCt8PKdP0Yu~ zcLem#0^M6UtfFuU{l|rpHr7DCT{asvdkFjU3n`kCEKnDEC$zri4EF6Zl|KPD*ijfu zBri(HheBc}{Qpmum%;u|mgg`57uefPet5w*>&LRiF(dH)P({nI$uSE=+S3t4x&A!0 z0D=(N2K}7#HDIZB8X?dff*1UA;P%Sw?yB4LPVMeeZU0K-6=21XTY5pi+k&%)sr{YC zrhCCz09nnqVZikL>LMs^IYa*`Z~pXj=WbV2Y46be8-- zrUCk)jjvs;ofmV}9Zy?6rE1hD#)~b5Ozd2-j8+E;2r12TT_dyS9UPC%dh^S{i(eGn zBb^|v=AbWnoV*&@(u!dug~gy|u42ZNu3|$uN#P24`C<{`UeKp0=mEROmo3XmGID*D z+_4nr)L?%7yMa%@>iQAOc2^Le76C@%d(kg)avDtrFzk_}$1$*BSo2o%{FWPJyMEsC zEl2L+H;;@R<`#xW^c8Sy>bgANa90K}3YP50w+Ymd*c8>Qtr*Y8eAF6{)(WKOks`t* zL;Ed-jT?||n;#h8`@5@mXNcnEm%qPH_JMJ^aWh38YC%%e3ma=5^0>LhMJ+S4gxYVw zO#y6xQYWbS1{A0RO7m4n$o+d@b3o|bAsjlm>2)9ZmX(WaXD}e!t5k)FLeQG?*>ui& zcH(abOf=^ywhvsluDa}iZ`ZaJ2yIXkIeQ2MC}l!GXHlFFN&sDgh$Jw~DgxC@6W7x) ziT*5{Ga%$*fBIsF=FVqqTKuX)jop{l!!leM(HL;WMILtmJU<+xG}6HQH3c*(;B~6t z^~I%`o)J)|r9@za^LC#bl4Pq`8bzieclrPD;pW7bVA#e-5MR@bA@vPf`{%{H;j}eG^yNpNEqy7nj!9VqvXe8+ zd^pH-4@6TL=F)qS(y)(5t(MJz9AAY0Dqr3n-X;OudRWJI3Q3tJ!O3Q0Gqv%nE1*f4 zIblz?6RyN}Y{t%vsAHI^JT)1$+<8{n(E$W@QvYun%78{Ba*cbtN@-%Cj{88BU zMgE-#kdf!POMDUDyuk(DvmllnRwwTE8~l$cT|@h~$?>GCh>iJF^`Wu`HvVOCz$5~q zvNLfb5*crOaL5Tt9Ohr?65sY;3W}|i&U>ooQ>L#==zFcb$Wt@gB2h)!y?il<+ z!(zhR#V@6oFsX@}^1|Y~kI^sx+lsO<3FKpBY`B=KMVCgC-t(@0|Cu3`C+(%L@zH|h&XfRJuW7)| zC8eEtSWApvGRvdj0y`kcpTOuJ5i(~q^OS4)x67XGVv}sJ+2#L$bSwnH1uCHcB8-5@ zkOf8j=#3JLc7FH3Mv9H`jsw6e+!EZ-s{QdtrS|JYd7Kr@lA z_#Ie;0Vq4Isw#eI$qWoyBq1@E6Oai2;z+5hs{_*q4dO^?Ap2H2vCh18Q#}^7y#8-q zjvP^gc!*aq$A4lSRP6Fqn7k2BSi1g1buN^mUJB2o=XqCY=Bo>zt|-7n$#o$@GV|mF zX87!O?t_US)qyw1(^wma1j2q#4SKG9Ef?^icq|Jv3 z%)m+J7!}eh2bPi}TR)$6PMtNiwx)QJYoQB=I}MUBy6m2wa}Qt zei`0A)|TtJ)@{h3Wm&A6GbeyzOE(r5c|83Vq2MXjc}u4l z?T8u@vUCEFfL|7S10 zbC+q;`*HDr;iT~4W<1MBAB>~mn`n5E{ejk8hNbtv4D%Z0+O?77iAA;^M7J813fS9> zzOia1EPW$t`zaL}AoNqK(!GsAMb&+iTRIUisUf%eg63g`(&SM#q>)qV_Zd;AlG)!~` zuX0`l6H|lrYgtCMPBQG{>(M9^&a_qBUo4j&ikfQYt;PY{z``e5j#T=1&o4y?YM`GR zD(@bvfflG4v_RXc!E&VCI>*(@i`^d@oGkNJ1($0aYhCz}#Xs9`{0?UKDvZnmF}bfD z`|_AIj<(XadPdl5 znwn5`zK+Fpy(D3M^YT%Ww9qGImP=?{k&Umd`HXfBi-vd%?W`LYcHV^_aE8m7t;wwx z@#Ra7IPbtDN~~@{0!@)is?_N9+Sd#5DM1}(TxJcF+A3U22KGH@c!BX)@zWFzckX=7 z#>WNGhp$$#A$P*_am(#dQu{Gdj*cLw-DTY5`51?IyqwOyw6ZGh+3g|Kd)-v9K}m#{ zOny(vEt(bKx$H;9(!pu3InIxMd+$PV%Sv|hevQle9O6?X-*mlZTY=7snH%&uTuT?& zZy?k(hP?_D-2YezsKrbn;b`dy=?rAR4*oI&4$Yg-LSL+Ct z9Hzj3Vh|7vWM1KX>+er!IyIQzzIjlX5VJUVG|_OXZCkTmeiwz5&;^=QyP1i$2`+As z0Uuev0n!0+rI_D)zODE4I@Q!#X?;0L_VNjbD`GAwKR9*MyPuR+Bl$nTi!uOS)DOwZ zeft`}oAJX;ufKCi`jKp|T6$Gu_wdbs9e<%i(aWZ*lSczGi*8-4_=OS)nHWoIDfc#u zu7k|kIq%2tbfm0+FNa1ne?*=IiH|O^Kf6i5O=f&HJMoP&oY4ik%2ow~kI%%}x<#+r zE;sFbabwM^DT;9B<5fmBxqid~#MBPL64bi1@l3ck?tTtLD-O0UHpsPTny#jh6vITz zeU@Am35yPBz{)Z&UZA6(yRTzuqxVkoTaft9EPBJLS!znz%&1>sL^pifKG(?kU!zSJW3rHn7qM^$RW+1f>g!eI1+SVZDc@os0=AH|8C=KV^LMFkSgct{{$4f0T-m;@> z&|Qij`X3u~bwQML2Jl@Cot)llvhXf$!%3f(mbSxlF{~ZVjOwWko|>2Mn2O&hx()x@ ze)9L3>pJ&gVE|&Rw5$x`&(fq&2+Pcbb95cECVXgW>VfGPIO&N>d!t=;d-(9suVkv; zIt2|`vFM#Qz)iM@WmnG_jMve-!Ijv29*x?5sDeQ9v_ny+WSQbla*`Eu4_G-tkW_>B}7w)0nIz97i1#2)%{YmDtOt? zt()9D_FA|ajfw|h? zb36BhJ~-fClRYb~N(+O*-AB3^Gqb=xfV^ok5w9e0Tn(&%ls^PIW*}=cL3gb_8NLvw>q)3w)1p#j+smN_l`P z(5+djn|2iv&e);Uy+Dz9@=fFqGhZmxZ;xBMn|83;O{f}lL= zfArmxkX3_fa7ymai2xQaX^a7mwhX3O*o%y}!d3~H5$l@_eEbio3B@#cux{V$km!~m z;w{0hsJmG4@p^72l#;sYi16c=tXb{e(P$LiC#fO6+}DqIaqTPMrDtiA@{2Bof;P8> zm;`+{Br9b}3#=!~LuBc>*87lgb0i}9aJxpwLi*32iAV#+7a_RosUp?xk47Iy_qrjI z%)s}s+7}a zH70kQ$Qyp$ylb}G)ee`ZQ2E_&PEOZbw3i9JSwn|!R1Vuo@CJTyr3=47zZPOMN;>h) zIoZ?IcTy3%s{dkO_LT(C1o1!4tRNb*Zd=$jw6bmj)W!3Q+DOg0A4yJ;9`}T51-BSZ z94(I9viy0ECif-aB?}TijY>3Gp&OhXXeL^06?~~@m=}%tOrpIs_wsw(bm3{U@H}MV zljclW&LGlTYG*(7MRwBWyGuRD?NYm{yJ-YB4f6v}!yAd3rwUhehw1+=HxbC8-Lx!k zBCAF(*4yUKzZPnmzy7CEqF}Rn0mI7x&vbp!vXVs zS@8QILlxwWm^D+6Czs8-Z?I&A!4>z{bvNV@7n;t9NPkX`4|a*Xw}}va$jRx<`GeOF z4I(7nof{538u*NFt9rzpRU)ccKgPR{{W1mg0o{sf-@n(%^%b8(N6sQL5%+AhQ z#2%lX(Y@4FqIPsH)@1ql1U7767~7U3f?E9wWfDdk+?uYAXlXQ>2;~)R6`|+OSdQM5 zrZ&l@(T-Uj5U&nvC$5g8zoq;wLYi7O@p?`Bt1JuP6_j5ns8;5QlD^r1ksH=>XjYHm zVrJOu^l+qd&C$zY)WHbHk$3Il{Z(5F(<#>cg+YuoybOD@QfB#k+MnWx>D^t!w_eBI z^Ely87@i%kyxOTeMG&u!3AjSE)vo*{%+A|mFy3uGWh}yRFM2|aODDp4DblxdwN+gs zEWe7_&9D9~(cs;I2X)BUZSXEF)@|B5W}Tt^&l*kho7bOxHO+0NN>n!q&h_V}Hn_eh zeK^0~lZALRTZ|=mURugR_dyttaU76|_Qb@Walw&H#64lYU4Hwh&pXB6AI-LQTf%Fe z&&x;Ku1@WWm#-;Z&iVD&k}+!_3lEHs2j=7;Xq~-o1g*0Txt_3yLPE>I(fj)02eNNE z0k*0^a>&jJ44j-mU)n|(Iv}d)T|LqVTGC&=#BfD@+1%A^xVe_e+iG=Su2$4eF^|Zl zW%U2xf&=UzJ?{l?itF6<6z{VmQ;!E}JAZ+wYNjw!&eLy50T8b2(=%v!?H;WrZydSQ zB$L`!8a`UhzcP50%2caub&u`b;1x2!8)UPyveF(HAppZ4Ls09bYUF&lsl`@yQDaE_ zFrAM&EW39uXx{Z|_hQ|TJ8W$Eg!3c(r5HC;Wfo%=9bJ1}6%DV?45txE@O$(7_gjB` z#$&Yq^xlLVNN$o5Eh&vA4SLgL@N$lmjZv9Z_DP}g2Wh?ceUMHdJ2)+ou+N)LsM=)y z(N{hgSD5HdYKKay)h=*kEIAr|acsZ&np~=2{sQZDCWdTKJ6P_VvnPfqc($?12SgU&pS5At{ zQOX04nBUIgu<$0aGi3*1N6@1f86xtaU97`UjkW~MQ?pD~71c)kB~g+I-hKx4@mvl61eUZ6|H6+&$lU9k_`F~u7}m?7SmC6<==cQR z#h^YA1e&nv(M~y2j9Iqgq8r&+G_LKk9+p9mFTyEW-atu5@`xRu!=QyKo}*Jf&kWmE z`+=<_o_a|ZjaXc6@|CDiy!pm%=5k)jbAebUJM}^F7g~|_6ISqgX38f(C(bpaivPo8fV=w zmX2%zH0D>h`XSF;FQ{+%PQ(*?C=gOSccki17THx8@#!Cct8}ZpqAmm;Er6F|R&8`5 zrfkh&EHddjAL8WVGSJ`u2+)nT6{NsY{flmVzR+>@ENxQv_R}DR{9hjNwx#Bu$GL94pRE2`1vu4ROhY+ zwW>MM%|Yk3oDo@Y9eRa0%{f?z`Xx$t%&UHd+L-e_ZmkNll)7TM49SG|lIprc!l_A- z*17~%CXGbUMKz^P2K@KFqQ6Cxc4und0c7dJpu*HX)dpJsFAYsL(O*x;Alu{j3v zmg0JUjH9)~*`M=EvwnWuD|tN`>lR}FRRn+jch6g89}J7*ImTt5G2Lb5UOpWwrL#F8383;u6A_YbJ!W#j6=(h{Qi>z6tCt8ang#g}0= zdX2Gd{%4LqjNl`KJJ3j7f`D#V8rQP?!&44#FogizP)xaFQa%k)XxeUA}x73k!=%3RkbN0<7bL$pkg0a#$wnziuO!iXrP!Zo?mD z(AbsTMOj%|9zvVcn_n?4Xi+t@xx7DN_xFvs|AD)N4e9Ao6pph^R}nzK?jpOl+ugTb z;_K_{lMq`t3FbYJ4<7{b6}r~xL0&cT^9b<+wuB4Dl*hM`Rdv=Cc{R1g2b*r##|NA3 z&vX(hCt+!=3zt)X6bfR5+CigVOk2pH%J#Qv{e^s7t*5G^@fhUcih@5eUgWq$BvEB` z)m5n19Dq{re{E&uHZd_>{w=NGv!`(1TalAl3vU-*Cl;SG_l_;>1OVC9?hn~U+Hdk4_54BzGX88oY5ZniV8 zOQ7_dw{WJGnoB>+QO;g8_EQtT=48{f{D32yKmhMjoXp{DfeD7Mygzy}2Htw0QU|`$ zE>ml5V5X&}5aCJUXKj@KIORG!=L|1ryLDdKfw|x z>Dno#(yf({98T~SIy*aMdJXnbwnC$+2!!Ixe*`)4_wyAom-bTM~nX6X{!Y_p)h!n z67e|MZg&HQK@E*~5@Lb}@bmJiXHaWXGqwoQ3>}r}AM*sZ&V5Ry6!we*frd8n$-^XD zQH7dn=D2Pqu{19~KLn|c5et*@A-WLCwqq3&?>3d$t>PpUC-1 z9>e#upJ8Y3`uqFuqL6$}h3i|oopz_gc5)s>orvuo`EidWIS50HkN4M0yn2|G)S_q3 z)@PPZj&@bN)U~xyU`M>4yQfoWZ>aS2@k({?=IQk2Hu%Bhk9V;_K8EQPl6#K?Y)bw+ zzM7Chw=;rIXzMA0&x-PEr9x$3#z&PiysUlN^PMSfkhlTIGLGXeYr!l!CS7cED6?QN zb7WijQI!dEtno-Vtp$sr+8=Mpum1Cy%2%s&gcN={`MTN#%JRzyr!e%VrY~o`)OF2uSY7kjhw1^x6OjCQ+F#DdvuBs$J_h&&!-j5D_>Gm zlA-QS!oVH9#QR3u=aBD+BcatvuZ&AzDsH@D-gYK5bD{I*UO4#%10{(l^McZ#R-}1v zq>wX)&tSmVy!Z$=#x}nteMs;{El#6|wc8=;N#(GDL`&Ki#EKmCiWzPSBj76MO8_icd7samxBFod(X$?BdH-Q}KKO%#&g=(}0HilP zm+v2Sn+rG3Hos+HJx3gCNV6{t(rtZb<@L?8jm<;wfbpBUsi@#l9)7(>=vY~TsK@{b z&Sm4~q`3g=nm|nu7F1U}BWwK>9cM(FY&;r!g8K(Jc17!eM!Yj|S;gld9gx5Xms; z=f%5jmpTLl$>jP8rMmn;S?A@tx0bQ6S16~_SK94qKC3Tyy|mhv)1MrOyD1fZeYvBR z0F>~5yoBGiS^$7dX1)AEV~?S|%i_9pM^%UA{U^?yP&$Lr*bi>%wTS;5jMlXaaVHhO3+{oWV9 zwy%VNIilQZFze~Y`na$|CgAku&fBnE?DH*bNR`+{W`s!Tg4{ZDu1q=+a74)12pPc1 z^%JCBQJrAinzeWB5{k^USWLalpWN%fKM38?@jbb~AXe^zTg)qE= zuFws6gIXE^@wvif+6H4w4*lVDYr$*S6bsw!l*{L5Gv_ie;M%`e0k$v%(k@bX&t|+d z037hCaQnxCu3FDo7WjbwD#DY#b6IYZ(L~v$C|P&5TJ`rgfs1SQC9?O)DcD!$97};c9RN=8+9kGCji=Ks8ODdnnD?t-QEV zZ`DMg9U&%=SNfe=w(JA001>1jyz3Zagq!CEN6-nS5xFapV6*jIMM2J^VC${-;oIFm!DsU@&Aj_s_7y{l>l@@G z(=*Py0*?l!f{Ea;uo>6^g)@d97$_{1^pMK`e#7qlMfs!)0QUwMLrGAvm!UJRMi?Ar zPvNechb=fi@+lHpG@nmvgK@#@I*$WbjY5DUnV{3VSUq35Ek*MzqargsV*#nDi* z@XG}FlYQZRC91Op{z@R11X3gf2S6AEZIl$G#sg|-ervWFW+Kc5j_~I|Lp2k3@ED;> z9|1P1OH$-d|4tmt)s-otT8*elLO7w=%`L1Kw5jsH^ zy7-w^d)jn)i$tO%qI&(&ukx$?fg}+Hw*mCD($7-mnqtk}{w+cYn9cRPd_3BhzuRIO zxv=yugm8#Ndv)Y56shGh`*XB6KPgtN?=;G0jK7qIM=hGw^tjpxzofMYCle|39uU@& zUj$c(rcMmfTm-HNRu{+#gu~dlm9%9p_tALqTO(dMvn`sc2;rT6rQb#oCifm2IlT+I zBhqZUJ?~tqIB(nj44=BRwN$y?H9+RI19=xTt`)qLtCu)j$nx5TmnbC3;dNom^Ubs3 zFirPxC$`5dk+aukJYG|t{g6EAr#!utoScln6Onl(L}OX_NwTY6g3?<(5}e27A{Y)r0kn^DpK+duY}Ys&T@(z+SM5Jr@@e52`rd!so4_p+U=mRXTz<@wq1a#{U^ zT|;tPn*MaJz z59xXOn#uo8xP3|Q-;yvX--4ByLsuWHTM$$vyiYbJSRVk+bYK}du1#AVGonxC*9el)`1%Yl-C|IoR0<{DfwX(E;g zeT$xY6K@mj)8Z*Ak1lC{l~^yQs&}E3D2YqJ%VfIBv1cK2%&uk`8|V5qq<^EO?m6j3 z$iz1cjImjYVzHMC&0?RWQ)KT+3A*^6Z9U4P=Yv58Pm0`VcgmuNaZ_ac;ptU+D6PEi zlWgL06qUL06l+b6vS?di*^*n?Oe;9$mAL^yXy{g1Zf+2=#TY2IDy)kUIrI_He7}d88ymMra3$^6OYr)9fTA9$qjxREY~T_M zYB69ND2Xj&P|4MyCsGE}ar79PwQa~)MG{PB1xBk0`G+crb3BPoCdbvW#?m4EQT zf~4~`cSOm(NHGhCZJ{IAl&yG1^VFJGHT_Zm1B97F==h>YmewGX4c}cmA7;V8oXnLE zRu(4PtxnF(m2uo4 z*BvxJsg0y{lONJCjUtwGydS#c;-p4LL_u@^yO^yLqA9)06r9=1i7OJ&Y0zq#%kX3T z9ZJ2yy}VCaFgp46kDTz=a<(?Yvu&l!rM#)d!CilN*j2O3j@&JUR@uvV75;-$Sy_R* zK1hE)fT2EcWOcTBwO}5(kzkyt9mD%o)D&mv+*jHZV)U%7KNS@fIh7a)>Dk((NN0dH zBDHUDWCTgl1Eb) zJcH3y5R6U@rLXT+f$gmJcxrPA`|Jny=oiE0+L02R*a{Z?j!g|s;dyVa{^DBco_P#| zD%NbTy`SS5Oq?0UuNpP0J*#hDoEf4R%cgF1p?(qSg($)Vw7Polz@7w|AZ#8>S#>hrNp;tSV9D4sXy zC9_3SsnB;_TGbbHEN5!|Q@4;mSAMD@~2R_c-;k=_s3*@GK8qQ?wUauBWBF zo~n6KV%|ql&~WBHK0Z#k`0@cjsu5821;lm)(^&^l;P)m8h(UV_wKE)~obfO|xd^|( z$ZG5AUa?^_8Zs8U!8iF8nfA6CX}!P4+R}3G-NU6{fa~pdzD{1JI%J>HNNPxs_|4Le zSKy2CvUxUHOHv0_EK|ql@uopTyb;~(*WDP2VmJj4Q*P_zeH&gZHe3@~;^wv3P>BCn z6uP);t0k-*`qZNe*NivejuVq~I=hq^w-_=wNSvFKsndL^4*6nG2La?GqD8m(4F|IZ zmSL!OIcgyCTFS14-?spc*+D}Z-Fty!bDHQX;wn6O6wK417hd;xu@pp~d)-m}V2WYb z`#6%WJLQ6z!Uct8+rW74vDqM#x5tV!AD^{e>NJXL`N=^f=@r_u*1Y%04cW-+sV5qAxxgEdN$L?iwMV3yL@4%>*Oi?qlUs5R}rEG%s*sT-&Fd z{rXzWa|bW=ld~=}AmK5wg%7C!*`d3Vh8ZMVYD}npC$UGu%8D5xK0oJ3)8gK}#?7E^ zMFNx5O4n@~B6htN0}dQ4Gzc<*%x!rlO%C#+FhmE~)OFusp`#R4VKY(wVXwTR!k9Nv z3Ey;J=c8m8+p{|66$EZxI&wmE#{jyiF3Qt>O1Y=Q-@mtj;Jx^VGv?4EWS6Ju*ftUC z4dy|em9*|e;NJsohJa_g=3k-T4qVle0kQZBly=N&wVpE(aZXPQ;e9_E&BsJVIh-nh zu3O(X!&``?hw?RY`&(i)BLOG-X$3kaCZ;KeY{RLlDw|4zWTiuG?iQ{_$9Ju1mrdT( z=onIxS7`6wH~q{;-=3z8w-43Pq2_s?d0>87bJFY$BKyI_d3HcUilw1jVed>(rjL%9 zzM7FU6vbT0YRByGN*2i+Qfe}nJ}yqsP$GynDo_>Zr6`@BT522FXnv+QYf&!EmQfq? zw~Y08^)wV`Df;&MO`X+>xu*>T`5(f`SMGo#P?*~{bD@K9&;(bK_#Ii3k;7~eAI@)U zfE=(MYch{nPVdtDGxO+*HFWh@;4xdt3*FQhtRVl`FPi`2a@q8aXVua)4@3d}M#J@_ zK76j}NSMv(K_ft2_`Mp!pat21X?89$jlLZ+=YXio&lK$`COF!(9V4=ZRm)~)zTTS1N z2Ozd0*dixe?-!BE+57wyQ}6uzkN$a(J-?8}7^EOnU3F$4$O($0Mr2`Eq@(iwV!x^g z$U-=@VQjkAK&$FZahmU^W?V}%Yz^k1Yz&~r7^m^c=>}uO$Ldu&l-x4%)?#hr%j%kn z`JnUO_NFll(V0#YUpbjge__iS_n?!_1G$l5m2QIT$)UJ4aa5$85n;(c$DPF3UB)HsGH1bK1p(Ymd?g7 zyH{w$L2CbfeA{LV#xDg)nEi9cb?HB=(YUTmNKX{%Jby$vX42MQdQox^%}MPJB{2;b zm^-2Ji29_+!kMsQMEe}=qcv0>cYaL2ni$p;IBpD#->N9c6Imsbt5=bTS%igJs#DbD zp`5+i2=K>8#^7}UjN=S&uf$N$!o5Nvw=-EXGk>&Pk3uG6K)`7Ou4yYW%z`9Ii~ z@{PHg*7WXF5ob6P_UrOm!{5GrVQfcaFaJZdm^nqx`u`82#nt@tG4s^LPKMN^(lu5) zH%yqf&Ab_ARwY6<6wqqlg9j|nM@^@ivNI!V&JuoCHCCDVj}~P5X%HpM$@ zvHORoiq}&^a0Q8J9N46V(}K%r*wdLv=-*y{5tInYQlTX6AZ_Vx;i;7V5G!nzc}n%3 zJ}IjpR3yeE9DH&{`4%t|!sulDE1P)97y#P;C$hv;f}=hx#P*t(4wG#TCrLTFP+U{B zf?iR%6D_~u!GOe-ycRO^Py@6v4@Oxj;7nEu6flCHzq7#UrsuK8QCQ~^ju4>{=nY6A zqF=y#5>X^Uf6)zG0F0qt6HX<4eFfn3fuB399SB=^0WI&_nFE_(b_F{uXp2WylT}*` zc>Fk#fAz&Q>NNG)3ZDF?>A{YUj`ya3_X{fCSQ&=(C#0Gjo+6Z+))?swB;rKKEizW$ zE57jMx|ND(h9c9>?(abbN1$b}{oDh!9sqJZ7+4+o$xe_d3G!?y|2ua`~;WYw8 zxT}(Px?@;Si=86iCO(xWP^UjT?+%Opq3lsB{)dePmGLmX9Gw;QdPi+eTzkeS@z5i; zyZbYT1clXQtwuLlTDP<8k_=D%*fupSEc(kl+LyjtUxyZ=sNf1PY6k8QLkm&$40u`8 z^4eIgBURUJuPatOc-3lJ^`{Fyy&(juiNd}DdW%nE!arryeN9wg_LcVq869koRJsY< zMdO$fat!x0w_UEFyIg@xP!0XGFZ;o72Q!bTv{%#@f}|p8yKP0xDk%GwbU%+_#X|$x z{E{)^7D)7SZqpGHXE6N6%rj%xlf;AuGA?94x8dQC)|B^Vyex)!y1t;lewzDI%ka|T zt5b&?uEcR6>VOf+-m@jMvJsYr<5=i{ukw?A=YrURUHzqgNy_HUgXdtx-rcWgNCsSV zetMakAMR^eLsnvJP4_xIWL3l#3!-}@Fpv>nRBcEq3`XqFxCJf7V0Q^5K8SX1q^cGI z*s5k18y5nl2}}={&&pysa1{zFK8qg))4#e<_)!}6%XbQ%b9lS%8%T`=YZ+-t$@Vwx z+q2l-uF}y1&_BijWr@{wO4Rw*%*o%KXLk!me7fn;uY1C?8Y~4{4}Gf!*=yon5Ckb* z-IeZQMG<93P9Nfw1tv2lep64jtaH*KU${QP!!7p_m)%rmzGhMAX$*Vzb+u%sXlMo} z>frS(K_Z<`{7@(Ow_<`0ivn8;gOgy_*rv|cskKj;g9_`RR4xrR$cO#WqU~#x*oX`t zqwF+MYz2Y9L;ClI3x|xrlEHo1=|JkhQ~2h}{s;0+UcP^W08YHjwABokg~7G&L-^yvHjGb`hbi3T})pjvl+20nK_5Um7vZ5)bPU)bG(~Y1{rVbopLga zk65B7B4v!DL;o|D+|I0)7RpL41c#SIZLh?Ifhxub6Uy!1RPmngx1PP(^gMP;Iqh}S zv>`0Fo9Qitk@MC+kVj?HN20cXhnZhuwDQ#x5)&W;QRpmS0rv_TDElO@_Xa2_aQ6uufDtGW^|&n4ZFO1W5>>9XQ&Is#Kj};;Hk% zL`z~M?X$Wiv8ZACcA)0ZKY2}e7lWKp_S7vL>hG+In4X4WS1CIh%nc0uqVB6a)#z2IQJaqPRx!)G?s8jA?~R8J9CC-Uz6%Bkutb9r*g?3hYE!pGZIkNT zY?iI$1&z}=*~;rOpB-FAOz-J&%Z;!VzNxCEmX#Zl`b~HH!9D3i+4sMnp{XgM9?ANQ zz4|!(x?w7)rd9W<4bjM`f4Pp6jWpfCb1OPu0lcwiP1B?PbwxLH2bE~VO#D4ML$DNO z85r?{d`pPX>H3V`v+~=NT}dJZFIBFAca4e5jCD@bmBo31?((vuQ9Idc{EfWwABoMs znhp2*l}j#X#NA2wyEBU?rcP9M-CMpkcK*xi1dMt&R!0iiD|lzhtVX1DS@g_L(EA`8j=_P|7^IP`RAqv2>DzwBQQ>hS%G& zVh-e7-wDvAB7zt>xf)L%J-!!^P%yP2uB{I;G~`LrzqAXd4e$ovFrF@suSGzc=$3AK__*7_5+R&0^><-?>^^h!R1_e9H05np!b%Xmljc3w1E zh-}e&Y9#%hWU{u=TeiyT9SP0MZ5a@2i97`ZFQXs#a<9(93K&~N zt98`kdVG~wmCG2mgK0aY+ip+e!$#UuNFiNByieh{$era1nCbvUN~;v|SW|SKa-y2| z-abq8xf;fBhm2PQW2!oOGWsn+0(y69P>Qlt(2S_w$ZZd-;CfSTo?gG* zc+^M{hL=?-#*%0Jq@Iv1dy3qBCC4$aYFDOunxM-mJt3O^L{r#LIv&=u&^Z4rv#suW zke2Gp0>apR#9L%O-8U?6<=9Yi9bW&1k1)>fF%G`j?z+^Fx;WA2c`b3O3-)QQyNNU1 zKxudf(5sbMjR#O-_~s^7raCz}{n&Qx8oFm(H2S`s2;-{1zds~1T>84Y7n5sVt;1H7 zim-=?qPHa_at6NY2&RWqM+OJ-%&w@MGd< z>$8HLmBQ^A=Q8rOm~nz8L#sNF3vy}NR1Q1;cslZ0k0p|voa+803`cQTW2yDK?ggKA}*qG9TCt%e++xf^E$a?ExL->;sbQYpkyZ^tS2E^nC zN9i({w&uWwTmjM9kCFLgfvwS#W&;k$E`44zWj$GW>mMa8I2Z@SbT`CW{*u}N7S#Hm z5&1*atAxG+IH^i+z7(IqI5KK~e_!B1R;JnW`k7u;8~5tIsiQcH2hoa)fBsKi`jhju zQ@8gfM_MyS4V$>f^$YcC&KfV&_R>>^$8m8#*JK!SaIpB8Vt*ty=q0mW!V11%9CVqC z>kjD`Yf3^7{)9)xw1k*1(7vRTW%^lO*AapGw0ieS1)TQaojG==cwXHxoR1@l zUX=F2cz`3=m@HrvTW-+)qV}bDNy~FJjtccB#E+#Lv}5dUkH#z})w)zMsPTnozvB9U zs-C_%wvNZ`QE613NGVu*@cn`CgRi@@kKBAH==BA<7MLZAu!1I0;(>f9ZigFBTM5go zlcVj-wyrouF3$xEy{CCe9njAH_HyWV9ECTexhF0J-K_Z?VZ6R8qil<cRb4zzLo%Zx(wq)4uFqUvEw57Rb=dXq3g3KA__OXIHZQYr z_1eIA?XHKsgVpZCin6Npr|)yD@NH5T(USVu!rxZN{L(Nc*;1E~t(691ktFosoY;47ql`xD1WrFQPE_@8~ zKgYL3(Wy6vsvMVH4`<(oEtYMTQhs~J{b%Q7&;4cl1gOgOQ_jDL)79{H{#t>@N) zY7_)tiN8hW){&zQo|>wU$(m4*WZc)MoVLQkoyz6ff09rnIlvzycJ?%)X%Dw-53NTz zNFuHCSH(#GsbBz?k_3hHy+8IOb{&qBVj@@39H;B>V)s1F+aD{N=_qyHPPv3;b%ll6FhLf_dLIYn_XTsP(GG#a40Ac*%!K z1fJPS@S4M}BPn^BUNnBwoWzeM)Kr<13!h7du|WD;EbXlXYJw_k%KCW9^Qccv#=^-mO{!z&rI(YUmn+dY8VURdau@rCz=HjDM zUZSX+97x{X!+go1%4S_rcZ4n$6@gl;bXviMC;|{8ZL+i7D*}}&MVuja9}m#|E#vX z@4NC_&$S<~Nflr5yBIb@sU_MNQHhCumr|uxlcgqUeg`YrL@VV1o=777sR6Ft;o$^b z0JZa5n-4C2_4`5}DwXf;es*iRC$vmK=f0toq#pHW{8CSuctVvL*+=myCKYDkcB75N zE1u6IBedG^maqRYYQJx<5je0fQ^P9YVQW*r_@=S<>T9vl@yX}9IUy`ZT|yb-7;+_V z<~8{X{N^=(CWJkC<}1w_)c$dD?^9YOA8rS}(Tj?=j0ap{(B+t!o5v?7x0c)N9C`A6 zs8SIxbg%0L%6z5sDj|5%W$X$6?~uD-!bCrP_Iryw-vciQ!$Hg4@txxs;NaDv`S4AwLd3gLh_(3E+H;K?j+RWvZp<2p<`8ocIYdE3h(97D?_5Q(JHDO=A z%dS>!)3F|N1LMlzqq(r)@Jf-;jc75g^alKlZvd>tp{%*-LJTlU0!igW$YEx)D!3&~d5R6`2?=nbk2gqu? zh98b-lAwx;$^^hDa&uq0Z6v3FEaIjdYA5*w<$c8N2I+VeQ;=qUwp@e8nd-Jge74qi zW{t8z($iAm;f){M3yhN5pM>A&-HnP07RsMFF6p_t#G;MDbnDNs#j-9f>17^?6~1u` zCLk>f!SMF5sqo|*DAxI{Qgh>D(ESgyj%s6!FS6I_9=N_Bc6i4XkTX!Q&&*{Mo0HR$ zc?Vt>zH{;v$6~6+Sd~hnLf6#&I%@Yu*!ennGGjk8kO9_&q{Z4P@ORlt=KPSKT@%_glK?R#_LHXR*hVdq=6&>#jcB(P$+BdJ0 z_wk1d-=U88oKp$*!wy90QDf-4A&_PRaTAs7` zfOJf+Q1zWK*1zwXM#ow^Z#B4cn+=<0Igk>HTdsJFWp#Dcq(lp=d2~O{p-Q$&XKB35 z%C&CVcCXt>kd?p}d0-gky-BkmKbmmRrjJU~E5yU1Si5wb$Y%%sdsrx5bMKwsN8%t< znac75f5z7T&d{YifXqtknsrd{BKzR!gWI-~l|}RGu-73vBs!ndx2`#LF7zNrG{y&~ zf<|ASP^q9(`;LAi?SN#DLGFxSUvYjWy$g?7^cWGa7!nR}J~)2CWWXfZuMZX>?~LU6 zj#CWV8{()P`Ag~;T{nbGv%bu~l6#tr7js+??snT}?1N=8@iC6RNr#_=n6Jb{W3)t? z)V}iT9V5R#?#b%e;%;eO`Rpt|6b|9A<=nRqJd9@yyt1%w79N{_zlSJUy@#}6!yKoP z*T#W;xwHN~-HWn!O16f?H(gLoykxb8*!^R-v#&+^O86HE=JG#x!8Dh)+BKdPWRoyV z{_{(yryCa@qH%m>W`MEwh1(MAqnG06emYG=GOzEcRiK(vv=i;h`9!SCZInKR=oyA*xP`B>-6;W_PVjk z+23t!^w?r6re-*>Vq#;p^z|s56Rde)L4Kl}mmTqV4 zu|H<+25O)k=Mlov{WPXWsdH!Ynfs;)|x+l*?wo5Dqu<5t`>2zq&H0}7K^3&L_n^Ulj zlv^yRKg%l>i*qRnHP)((39!=se0JgN_mv(|G_1>N5B@kSzp%W$U+9V2AK0pm-}f3> z5~`>eUP`uKGOH{}O(RF~cEl1fsuJDhJaIlMX0jm~WE!_a8s;p=YX3v=vVXUNIRTRy*jd z^qB3&Nzm+rL_p_9xv_myU4F=fE}1q#_r-Jz%bF(XhBw;DRGO(N@=3PIYjad@`|cY% z?cc{EpVJ-}3J|PJo%pchUng7Vo|LT-X3WD1hatXQw8cVVsWhu)JerKYelzLsa3Lb| z-?R8DyHebRNed)`80#*->!OnY2291&vZUgE z4+Wo`MBKtXXz#Cc(9B`44ieNFCJ)KqeKirUKT6WuqUK&+sIY>z|GDohDFNWa$;!)p&pGH}LXGIXZGvRilfRcD|gT?E5GQzqkO` zr|n`AUZp#DZQd8z19C?S5AN`aDUE$0&Lvr!c z>cbbrV?S#oiJ^}sg)xMY!v~j1lZ0TLdX;02KJ=IDm054UyEAglSy&wH?Vpa!G@B{* zC2YL@9bdraz*#nqrdL!XJ@T9yLP(7(cBXf|;kXdXNi3>$EfcKAB9^r#{gXLP+z!}x z&1=-_fV3`SfnB-BuJH#Fa!?0&)tCU+L60AM0Y3LmSrCOs4@%iucLb4!wGXh#p&vJM z^YRfqLj+aaydUO9o|u@Zh2<5|{&Bt9M52&J15FF!y|TtItW_^HAPkv`vIb50>^UUa zqwsUcFEstoQr1!&J~L87zD?gXJz8Zp*(|;a3MY!4r71sW-n~5Or-v=IN-B}MRC0>B zME+RUS#GQ%Z}JEl65H*FLnooD<#>fg+o0h~4!}=(Ua5I%*13`WjWLhFgED||nS0I` zZC9WqU-PhbOLlsz+ZPN!bI-ola|Q|`#;rA;3kzX&qldGmNK958ZA7P-DG@R5FF%@6 zx|ZyJ2&NdHo$-|5bg250x5Je(yPDtZo~KkUuha0+zzQqp`+V7Fv!wcXl%h~-Vv0n< zo&>@l9N2`xZp^}ORRKqT(RQ| z5=-SjxIqTd?ZyJnvRBHY8&YIxgn$`#Ac;jaUKq64PncY|x^+sG3zduK>g^ta>Ed?|d2S^c8dYvUUIR9N_?Ar~_N$BkrpFZ+)KFz5|5weim-y2Gt1khN%{h zy8>NvEsT$wyJ^S6qlbIGjT=}ZTf>(S!E-Wv7pt|iw-@}k8wkyryN+22YzHH}mFGok;jch-& zhBz1`tJpp^sp69^tAxmhGJp!Vz%Tid@wBUu02rUbMxDDOpJN@ zQG`?qM>0M8)Rx*5X7O8cLZ?5bB{j#jDtw%&Nr{OBF<4>o<4~Nue_+$eNF5yMOTYC* z@~+5~&rhs}dNj5tiM*3BxWJDXAUAIKWMLD?gg59%o%}N+y%R@bk7ct0OPM=WZ@;bY zc2D}tU?diA=^!`R_AEYFdna#!52_2zxUy-o5pmD9a>s8r(wLvfpvSwfvpdo~H;oZ} zva=NPJ`b~rX!X4@qt@gLA|eAz%V(2Lg^IK#<>j$`hBR^FvwJqlRCIK7B_$>IFr#{! zN(U)og6W@v$@SSc#KOe|e4r#8_L%Ruy}Z5oap~QPkcs}BUlQ6uq~9)#oj>VA7<)*X z^!dPA#vUd@yX1d6wwDahpjaTXa*D%&RW>W!cVP6*p2v~saBDs|C;wbj`$D}&a-_zO zS0|q#_feY=zvIK!hni^h9s(ZaS)+R>2ZDr{nMHG1%vqZpN=hNa!r*tmpyYRNOKN_l zd=40WX={3Q=T|4;u%_6dvA# z!h3&2^*e`!MtOMYqXp@8$tW7q-!KOFP0=H?cSok*eVQ&CVK_hCnB}z$ka(bQt%z{{&Cn7_G;uj@(6) zC~A*-ne-JTxpV(%9i1uUZ#6Xo6e9)rr>0;`LlmLem=Pi&A~^_k3qjy$X(~uMsi3!* zH3;n@2T4eDDrJjD7mZ5;inIY`P~dX(pNd#1n%dKSa*EOZ;^)~hB10CjVC}h%ryUdy zoaaavAehYvsNAYHHfZC-ul9i3sdiRWRP-W?^lz3Q`iW|dI)nF`inmeE!u29hU}8my zq_S78(ID`NXw;;Rz731I_fFEXFC74A$RYjmBi_KnQFmH>_4tRxw$vWTw1(E}I=N;S znLCC_Pu$fe(IWO3V%%^QQnSSa23D|>kg<}G@5>Ie^=i|~Y8Q>r5>@vzgx2-XU0pqB ztIjySpv=L@4;~bWd^uptGe?H9Qy(c|*D7?HOAO}T-wtTI6JSdDKH5yURDl_CxgwND zFqSM@-VU7dBzY_udbH8TnGtjJ*r{d;bWn++#U9mxSUKIXjOO5Oh z-uFB_AL&^`R2F!=R~z9)$!|`(i=?OL`1dKE*IiyM+deaxk_^sarN?7-0hY+5@_s!PWZC=+Ml2+*MpK6hd5eMs%|1);j0wFX)L z3&RNemm+HSg3Sd?(6>9ce}CA+XKJi|5Rs6qiF&V!{ypRpm?KAA;KReYZWqQzJSJF^C_8(xZvTe_AT^{{`ME0yU9L1xT;iO zebL_xFBgXpw)2qvhOaY82<~aiObRL1rTw|l%pjXk?T2MOTS95oGpr*%oU}J&Ti*L74wjFCp zbJ3=4dpmX{IWjWYakM9mpqQN2hT&bmeA<-tn`mwu#iaIj%QBFKJ8S(unkiS_M~bTV z22Q@p&W~qJm(}qj%|7pBJ{CCt75hjuU$&*SiGRskyU9dDApW&JzhzIhs@8mge?zEI zU1J=R)@0iT8V-ic7i5{Noq21GBWYxZl!)uU^9SDqd?uZkBS;aIqoQf{VNwTQ6i_SP z1wY#ZeG_4X2uErXbgi~UXVM9*(pW&eAW?M ztn1CiM%~=a0L$<}6|hlXHTj(F?=`mGOy2(0IRYENZD!#B;Q;=({ml*AvsEOY1EhC8 zH#|RpP}(~)epy07;%dd`D(d3Ahre}Y-v>VF9r6C9?eH*Tr}n>Zs_kW90^(59UY5dw z?!jX0x6*s39qey@=svP*(o`198(;( zKg;axM^1&{GFW@kLx2F9u@{sMjj3w<>@~5KG|0hgK=}j-$0W$;%19UD&Bo8TE)%Db zhCtbA6cYe|QSh5(3QSJPoRI-xVGBfLFH*@M=|T#1(seCqui4Kw&M=4OTrEwsTx1ls zy4h2bIK2^y#G21;P)J&wp(O0w;!N=TF(aH-o%`MM=GI%g^?DXnvfO+WbKnZ4eQ= zx$=d(t{K!Bo?mXS0UCM9>qu6MDAoYkJ}MZe?Zg14PbRa}|ITbXk^@=UdDjzNiznJ)U} z0G|_Tq`Ip8M!s1sFUDeowzj}%r^J)Rk2*6J3~Q@dxLZ#@^4ySdeSQ1B_+Umot>$^9 zA>(rzMx}}f@StYXK9UCzV`OwQzi-jm#p*VEX+ZkBWP6eqS;lV^Ck?2 zDly=O-^+t1g7@?1^H=!96okoBM~<_lo`u!aEqkpuB(>ai#82ykCf4Dijg{U!LM;nv zoxcS=Lmkyd1yV5 zHZn_cDtIQ`DmW?w3oZ6|#tor%DGwRgdx&mAJTg|P`HbiPKc9nYvk58YfFw|8mN(Q2 zE-43`AUJ?G^22Q3HQ1e@-OQG(1V#}?ZF-e=FB*D91$CRd!^&4Td`m-HLpR)T= zB+2dxZzYL(6z!b^ve?a@?SS)k(i{*^@9yWh>t4UV@-N(k@QgmXIi^fkJxr7~Kem?a zJrt;}8Q^L-vZuhGuu6^1{U2ab8wd>nx&`DIJdokH z%}?#H>zg#s$vQJ_H6Qozy6psg@R9ie986+Vu}gqgKC4}5D1#!lRkj|+mG5pf%-@{W z9ogK%f56W$F$P$rfMfT<3O;?dayPQX@?YR!hX|31Qgx<%yh9{6feok0SF$kr9WmFd z?X#5Icl2=j4;!mc*natOs%MNtP|qz^0g;a1c+aQ(m!#C(d&3<GT4_``=fvwo}9H zc3b9uT%TouLX7ldylF~$ToKT(C5dqcNi%D0OO|zANEouM$Rp_Yi6t1ZKl-qEl>soL5eKADEVGhRZ`M z)yV($I??Kb3D#QqxdTVLk#qj^j>wv%lCrMAE(rB>V(&;wyD-kUF9llPLyO-kHlw}t z29y73eOrYH7Z9y)$&Y=w_HoJPxB+rdYUg0Mp4PyGh7oF<8|wJ0?a~L47JA(9mw83o z{#PjaODF=Z9cUP&)irK)Ef?{Z9|HzDs1WR&mW}BJYfuL-1Oo>LM^aL9vGEOf3W2p$kgNHzz5hTF=)>Unu+0}OW z4ejj}$i=~_vjltPmk`|eMaE73?N)sdx%7*FWDQ?xv z4)8bNc>3)luHwk!#tFeZ2moV%>#OiN`Sltm9$!4dDJLMiBX3BPlyimG19+w-+)z0r zRHZCHxr;DpdW6k~Ab~QXYzk?%AuF|GLtJ=oIUdl!Cd$(8Ni?||H(5EzXgDO5M^P=k zc`|9eG~iXu?qU_z{g)X)UdcO9&cVON$YZsmSNE^BP_A;uHho91S{gich;gFxteC(`Iq+>8)#NLv1{LZWKf7d>Yh0uBkwA2*QxwS55CK4+F*i)aeABGGI|NhGT~} z8vel~iH0PpVoj5sCHIBE1NZR&!HhP8Qn!}2?@5iwq^T=uaydfesv$gLJYbJu5Knv4 zB_gWw^bKXF>w$;|7Saa@rFAawV?z|er$7OdfEq=|Jl-AH$`YW0vL?N#yE2{{; z*7LHCe&3PI`IPXG>+@rM8>s!{ec>UiXKY}wloAb=s>8?}|G(VMACVBE;WwCsC|_Em z4R@DtM+-6g?J}5lj~X)@l_`z{A^{Eny5Cwo8-FGSTqQD2?3Z>_B8Qs9Ql!6Pb6dvS z*xQpF>7rfQlfJ7N9UWIo;VwtIcO+R_nbj+Bj=g0_-U(w>UN zJtLixi7R0H1b+U-Pw#-e9<6Ff{N`0!mMB{Ma5M&ilb zncgqu0}4e9W30uTku{@V<^(1^h5T{=PjKF{1g0N+_zwDSYtAZ%(UJ8oq`GO&^~L4o zPb1L5cmSwY=BT-}|AIHkGKO7>O(u3utj=s$2?vIUS@`**^33iO5LF8P?}O8N zd1(CWV-fiWARxmZkUfW^QQlvNeQu5r3r|bR%6^%$9$a1nuhtvw=3az#4U8t<{=gY< zx>myvR$79e=i>X}KVW6|;nN=nPp{e;uO0;I9b6p^A0h59Nb*=%0=V5=vs|uC7g|8AoeBXxvUH z@U-2)CHy|?Anl}bnS8;U5mO~uM?B%P%pmg2qWVaw>3#)ujg^^x6_mf&-?AOTyM|ZJ z&~3*@?=RM)uek?$x~0ooJ?<#7sixrghr@*=1B|bI=S3v#bPJ4SaSvcjSyIr-^SKSa zp59($L{S|xJJw55K}z)>^q)!tye_Gd5=hyFjYyO#?*y_OU0l`523pO+79N6R0c%bc z{IEWgH~s{u?6mCEHg0U;`-NkzSBO_T%7f-?VCUp8d9pchvU#1YhZmP@FJ!myDY)Rg zV8i#B2>`%@2IJSfyq5v$I&GCPy-|@$)pIrNRH<=Y7~Sy4IgoOP{E5a!{pDf}!Tj%} z^)t`h4deUef;p$koy2od(#IcOE}8sw!nvm^6}y|ulY4=Bq8V%4mszQ9uv82^&(@a5 zpd#?%6yDY(R^sojEOhA;M3hna^b-j;fS4o7==X{g_FA(QG>?#skS}~0FB7?h#}%wG zOq`2u(w1%4`@CI(R=<|wl&A-EZ!-{Dgb{RUa(voSr=p=vxBaSXPAp@NYn5YoB|FH$ zyO8hRN!<@1B7BF1JXNjk>Xd>;q^^|@PlHA~|NW)L4&LLf~wc#=)iXnKuhev>%l5C}CSy))-> z|Hl8JHNe%=4Y1OzcG9gZ$5;XBXYA(oLRd~LS045$Pf@YS<&G_E@s&&xJ!SkSksmA# zi8lh0n8y19X+wgFw0N{}sFUS7Ju82J8vs;PvNUOcd_*mgAjmv{;OprBnNi}`0GmW0 zN^#o1f9ChpnN&bLEtces2bvx&?}vehKbi^d&4@tl$F^l&;+$ehjdK)Swr(1|^>Z*sA-ZJ8F)nS%Z#4;Xz8chhKVw9pwuo717(;lt=97@%}l_oJ6NGq$xmUoD>`DN zixJ;$Qo2*lYX>!@M8OKW>PByxafXPa(IQDmV5hY!oL|kMCHRaAGsq~k;3$i8EaaL; zDQ`y2KCd{q=AhhD&vmjTW8xBY{d-??va7dTP9r5hNvpdk{;g%$ct4vmYQll3cXped z$O=aEv*FaDP}N#1aEoKg>E+Ez^rjgsC)@2LH8ro&nff+aD-LiQ-R)(pEYVCVUxm9} z9)osi+A`jf9V)@tB1DOeNKglOQ}$?qf8Obb2Y{~{f*Gk%o`Up$;JzXtBIR+?x zNL8batA#DFN+~NVtE#Epe+h{Kue=C3mYQ)IbIU@V=NV!+S>$REWk8&=hpk5dAe@cz zoqs;|`=Gg~%^mMjB)$8Jb#hi-k}52=aICVKACH;4bmPe+7A_60S2hjY%pqDM-etE% zAkY4MG;-cSsLWv&ZJG7p+zGHzdVrL&9HVln5nFOE)OuGnm7Q+!{jcT|-ptr~#-U9? zfTW8rp`23h)+ZD;e<`fG5r~|FKJJ%Lvh0{g)>Ee`60oIwzib)4dTqz5mAAw_i>yW4 zPtu;0N%N)*ZaJs&LGE|kP$EhN20U6PuzMR!B-qX^GibV8@d4{6OH$wdL`g2A^wjdV ziI|;PpF`nS#VtR6qZpyodgJy0k-bD(Q*~iu_Dv(i{ifv)1)WMyHYhet=I>#?5R zYxF*()mN+i(AxaIwefv!MdS(rFPS8=YmShgLubcI0X3(;8+A$#*~~4dl>>u=6|Jq( zE!qptx6;DEK~S~?5a2i#g+kn9=kU0;pic;31MI}$CkCTXSQo-`1-1cqFaeaHWwbiw0!)Fd^W~RP=DRKmz(>+>XHC>0$V108 z=lDx=;W8Te$46f`iDhcfRF`Zx`kB6w7$rjpiC}_f)pJ~L`n#{H7Cv3+HRoHWh_WL& zl(%`=*_%QlM#El;CpVgs(+_O%Lee=1?*UI zvkWd);wXcuV;}x=leMkl9jlLS)X~C_ z!I98%C%K*ehPE%vd_+|vGhOXloK6=TTEB9-ym7=%v5z1iCSN4!@z`Pg;FwD|b0=0; z_5-z4-zmhLB@CVj=zX~A9fm=TPe@Hj9Cxx97*WsKSmpruN(59;s!S2v2bg1k9nV6> z%xhFNH4W|Tu#}XVeA8lLVidG(|Lm6{HGJ1BE+Fe z;m)rH@*)goTq(4dO|rpkyLZ912Ya2!;octKPtt*aeVrdB3p{(-!)jW>@C*FeKT8r3IK&yRn7I75r+ZAAqPj}fpJJwaMIY??)yE4{G)Wp*r5igxZc@tE z8&{+E3<}siI_K5X&w5RTy%F+A^z1e;N_DK@3*ADtsz&z2z)C%S?O2(AVG-8{om5xT z&$2IHkqx9d*4l9J5di^uPzMlAfvmb#YvCG6Dna&?u2wj$AmF5zq&}(APgq$&(1VOx zfpnz%K4jaD2}AJPAP)&amA2;7)=ttU!nVbNSfuUBH6ts-bBb#6eZ6bx zNq1YEk`{x6P3#F*J}0_7<$iWtLK7JWq-`77;CD|^+1yZPNI|V| zl)lC^P90Fy4S)Y3))*;Ycu)!(>Yw!+mfDBKZLq+B+fk_`e!R?oy4p@N;eETzVJlEP z;heWqmp_L8oDt-#x?aExU_tV9U0&Q(_Co+fHNxSZPkNYV;+Ujxa36!FkLhg?BU_}q zsTHlqMgY*fZDeyiPq7Xdq_DA?Z(^zck@;b)phDd>L~{*4LK#a%NzRFcjX>z+0zBl{ z#2FOP$wKcIO_~;D1V^3zQOV6RY=!OoR+%jwEHy{zsO?oOuTu7W5w-q`kl2LPInvP3 zTKB>88siYpHTs+}|FVGeOkLQ<~w6*%U)t*~3WijG3Doy1PqXYvqz~0$^tD zH#-(Jgf;3BU^L3!;VM)tA{zNB-UsfdKW31`-9_yToYja2nkWO}h^b;-CgeAqWnos_ zG#&#FUdumL*v(W3_@2Q4ef5y)?+2w>>OW`dUWc5Pfg)$_x7Un5Flp0%r@B-o7@Up0 zio)-5I;8D?7&z$vhk!ZxM}$4AB@NFwLt z3u9z^%XHM7O>>T&1AdfmS_!L^vs<}jSgOQ|6T4~|4hQk<$K4cJ+kIKLHeYn>V`T4$ zKL6{Yncz+#)5h~p@ucbk8QkREZz&DQ!mbiSA8Y{ksUtChi@y+-b}OwWO+dhHT0psR88 zGW}eLtBf(7clo(%9mn872U&nG9*J}r~x)JO30C@QP8k#_O?^t3dtAvNNZ8P4u9Zv+sZ&r|;;zWwbS_%Z}t9 zE?vCFNL?At1OR668P}706^oo92e^DT1^eYb=9U90$$9V{z5gN6K_3)pTs2*q+Mv*+ zFL3F8it93|8ClO0MN*3?d9$#Ag=^^SoRcP zC6FsszF#o8<0vhl!F~*3+*bUl*hoDDI<=zOe4>0(JZ27Q>HP}T#Xr#(`lgOUPS4z- z_5ogxNcUxhoqv?}PFI$wI472H=W-*RRpE5BksiIhZbZq`K@x#cdKA7lYn9}k{3d5w z!{QO?$Fcm%n_V189E-}p6D`YIlv+(pOZ1dgxIvH;2e+%MCN1L9`bk^B6vm{}9f%9S zrTFheHP-jWD(G3uF>Lf7fw?AK_SVlaNY7%?zAD>^BXzn(g|s7y+Y*Qn1q*8j%>OV0 zZGIk1VAKg$OY}(eK;+A#Ze!%g;iZsH48A}ZQm>x1Yq2)}(liqCFyAK&HCB5l>jb{3 zaY983TC%cV?($Rm_4JY1d=G=*l; zz48aA&WZx$WFfj?QpkU6UiB+_|NaHR)6PBOeSv!4tE)SI{lW&Ob}`Dv`-8GHlnZB{ zzU&J>BS*puimfAY!d{5QTt4^*%b&`m2?c#IQ=Zuu=vpsedXFwh%Gcqw!}e{pun1Iu z6D!OSDOt#HUKd19TdBQy!8D2ldywQuOgopjc?LE~ZEOIE$r)z*t}^%noEn@-<7ch2 zUJLzj{cktbb3XgW#d;1mNJE&Ar!M=G^i~+`!FppvL*HiO8x&5;$7Y(ZtNdBIU=8t7 zpiyyfN?-qr>RjwQi=sEfp3EQ2$m{FlhzcX+*;|5)SOJt~XF1~J+L*HnGuNb)#Fb?iYiGB~iRf>Li4M=7G*}_XbSfRSOii7xN)4dY#F=zDaO>#o zh^1^-5S^jAuC|4piky1W0WR*pW=`OX1uw=1`IPpFdm5;Nh-VSdGVv4Z)LF^Nceo?A zIz&3{lM#8$BG4J{DvN9D=ujs>^=387#a*-#z?L5n6@9n{obJ8=P)d?Kb>gK2qbai8 z7qnaLO7K6UqCozv4Rlb&J5%H6o<2e zN#s7!xW+WiVLcWTG3yyIp;KGM{+H#sLc9xav_%st=_jz_oGrrI?M$~dCTPecPOai8 z=FsjF)~>?!#3ulGCrlDnp9Y?n8o9?~K-~B=5HV89$Yi_Am`B5FZH;YCruVGUT-WD2 zk|GU8d&_i*q<$_P9&X4vVuTKd?V&75L}C%AQ8Doatv@9m(wMEv(ZZ(*{ml3l`~MO4v3$)@E_7dk*IJok_hx;qh4&#JT$AxK6p4A%Y zHc>?yEG;jeiHJQm2??McQB*^MIG!t3X?&iK+Fds{|&U7u>K^Fupxo@bmsZU z8%D;e=H_sOsqG_hKOrC<0^%ViP{RJ1FkiUSz2c4(oso{0kh?jVIWI9zVl_mQU7_0`3_y0DypkF!O0KY(l__HVRPOJv_ zLAwMU#_H6^#%EOmdEb6^wmIeJ&$dSf&E9kH;qpD_mO6qduM*)^vhTYDxviFg`y{rt z%4?{N8?U7Jg~2|F*F=8>nIew$r(mL19}ap5WRO#$DLE>rK^9I=Cg!@sPhv-Q6Aa-(%! zQy&7q;UwOxWe5DUAddA{gibHkZFLnSoI7$!aiy&TQv0`g&0eKk}8#fW_) z8Vc`OVb&tXx=2Nce3ooyB2Q7k5i|jUvz#q@;?|u@OA1U1WaQY0Iw>k5NfRkUce0dc*xwEoBh%aNP=qxh`GW>*6kv z#=e^N(wM({{NjT$*7En)8u>nU{c{lD$5Xv&iwR`us;D%_+onUPQ&1WhGYw9Vi&!0PGGhfnACVNMNW_o@~IA@q;Lq6 zKD=|fKY6Xt!q+~wfh?2JQNxhjfVXY&6bQSfA_(J+^b?gdq4$2~TznMS5EC|*5wh6U zb0#j;|6dM~6CJpIiF(18P7rvv%R#?_VBth-LAXXeSH0B!`ma;7whhqoi03|O954MB z#OApW$imE$4Q4SJ(p6LD2=~b+M~xE?5YKHt*_r&~$^$@euwzgxf6P_Ja@+;{kp+TV zVsfa#moJ0?W;sqE1qXwHU}BBB)#PPykvI=zKalC6FH8dp4EGj(upsuaWyc-=`M+-& zYBcCrM(%TcG(?XALeuCmZ6X6l;^#e1?5JIEtLD%mYkwvTJ=IWhJZ2fc7438bU$J`V2l~YV00< z<`77TgsYwmdq=-=j<^9pHotaQv&pfc37=&LWt;lGyoq`JV5RC@eue0RsqB*mtJ9x@ zD@4#&!fgrZKNHfYQu8LCo0TydD5KWL2N)!+(sI03evWdk4_GB3KLchdQnE=WInEo* zuRtRk`w^_680NdAC%ricx@h`w$!$K)MRR}1NM6h7hBQX#WGYZE-8bIEB&aSU7tFl{&V4C5je{M9n zW?Ek+L02aZ)N01YC{5ZO6@%7nl7l-PDFono34nwWpVO@%Oa=ND?EpqV zh9(peiA{j04`Cef=157t^F_(a*6e)6(kF+r1)-s!PVm1)#c0I_RKZ;{(%{AK zQra2;-?*z+04|PMr1>-QsPRqQcXTCcN6+x3FwhZQF5G(G0FTaRAQr{SG5TEkDF;)w z9kl&&HEw@q6I}mrGadpu?h`cOj7SR?#%? zwp!n@v~2o1RDd3z0zb@dZ|BU3Wn)s;j!@%vylHE+laJ5Jdd^}Z~>d$Q!$yJ|i2A{0D z#B?W$9OLMCEB>L*{Ts_mJwFnS7&;<@S+j8NYH}6aGqIxH+TxK0Y(?y#Vx0CLf&^|a zsrl6&6Z-+ybt)JAHY z>L#7h^=Q-92GKFbA7W*jkdnKfl39*rRzMNvaOS+3=lKMA=;&8cB9+2Qu-Be@aXzUzhOBtrh&$DA+b2 zTYH#TivN@&UbViV7qNR8xSHL-0_}64@U@0`pX-U|V&}?-6swdk12GB$FaBNO-vXSf zoqOMm8a6@Zs-%?ETQ;_U6;5Ehv|emgFSTy+shT#_)AJ1o2(ZpwlYY-a@C?$yeOJ&q zJpn%i-Z%H-C(LLaKkZT}_@{U!>7nF+6BbMi)Z2CU(V7G$1;!7F%4b{ws{pC4QAR_? zq?Beqgn&+K?w7aQdx>nc#pq5FM-V!oF+>9E97kho_$gwlSerm_<9L!dfEy?12Jf1WBeoLl z!##jS5?nkhog*}^L5>8+gvlNz>GBGYL&Xd{e(7N&#yGI9##95n5InZvv8b0lt9iDKmN@;K^8B< z-{$LA8{!_P0`*s@-Gr~`M8`_xRO$W%oG=~+{J44GtNRHF;`)iBc?r#P(#7lfJpf7o zjva(tZ#gy+jU2m28NZ;rn{NE$yQKdcEri(`16djNYQU+pE|@)21`mHpFalN_FaSd2 zWKbPMeM%+x_x#R1TN!TIpXJzOwZ&#@m&Xno1DFwf0Zp}mPlGyOw27mp$Fk^CPBW=u zPcp0HjN6*mz%ZlJ$e!3`13{ z$w6+9tkS!ZMpDostLx+dfW1iRr;{A7Hml9O@s+j&;tqaxL0iiwgD(3@B5}@NUu+_T zz`of9J;XyitCR3RCYSjnxlNrlS*3C0Z`3H?sO>b7s^m?P5+@Uw=-f1++4}geOUrYz zme0C^1G#tJH!H3l;; z9&`SN({9O1*fjscLCG>7bse6ur3VU{=J!k4aoJ~|DK_;yk2Dt9 znHc;DGsc;dv!!rrN;>;$^2CePdFdmY!k*Q}AeUbb1~KGFWnD4#Ue{z{-gb-#rn^0n zh~Eay#QvNv2xq7y-mGLUyP-B%ZU{J37;qFxWGDmqd3~wV;}6>S-o<}lsS+rTZYu*c zoU9WDm{hm_EXl}V5~Q^8vqwzi)8@cOURi1?R~lXKHx7Y$Re|gfy{$6Kt`5_<4t0I0 zG+I%(>$%iYSY9yy`GPJRP@rvq4&)bY$x0#>TCqrrZPGs#Zs~opkSnTI5p)mM{cUb| zZ6sGB;qFvfhIu;L+aE-(fksN{O5ESy#Q&R1cqXL90J(~{x~V+qN(3aG~|iHt#e>J_U6!93Lsi41OiZ&qZ_w z!6lkY^(U?;!Y^u+^ZMoM9TdMm5qjn9VktcTTF4Cp=Sq)Vd!KK=WoS}#5>NZM<1K>? zH}neb`zXn^^S)Ej^lJq62i=jjZPj`_LjSx#(MgSZ__lw(0FQUP#RH01 znjm!}#pwd%>6=M-R@dOfy0i@EQS|4qvZhf;v(NA(dct1wFYZvxb!$hT;LW5kp<&HV zb~M82F6Z&-Ia=xBk5FbaG z7^MJ+dIHhx9ZccYJBKGHPLc@0seGYOq2luPcrnnR0l_H+1;w|qE$bfoyYb+|D-Alt zco3CFYnc87Kof5gn#cysYZ`a1@r23l+MPDt167PpC-laJR~)pRT24i(r$Ehp;c%IG zvPe<_X~qiSmTNNmeDiJa6y&Ys=69{V(R*NIaMK{@n+HEzV2-tQ#rcLhwfM}MSv~co zOZtt`U!3dja##v-wf-zw*EwV6a-K^&Um5v>sK5N=XBXDhN0*_u+m=agDwA;V3P8jG zl&RRUKS+$U%DT96jiq=bl0WK)jXKL%ESJ+z>o3=@Uzc2v9GUHL;17lc%$w1>VIkI_ zM>(dlg<-7K*0w6{0)S5*bS4lWH-T$TPBX91I?YyK5Dc!o;;|!L3c0k^3{v)v118t=w#>!wvi2OS|0QC!1EO?pQbZr+5V!kJVDI0`SW`MaRHa1pLT6($h z4VWD@c&p?KEt`Cb$Mwz3LUMCo0RSaN%(2)cP#{a8W{yv4UkSX#1jhf?*EvY9C3{K_ z>z59bWGY~_wq%z8#rY5IZ*2fbD$)vbrBQ+Gir-2F@Cq7G2#6E&oE{{3CC4{1j9eZ$ zdQtc62L{^zj9&wQN7_W%rKxc1WTP(Wlj;Obt^sT?A3g7u|fM}1JR{%i*9qf0F!9nNM+6V~p zGD%t|KUc)w(ZV%K7rt?>kORQ_QhV5+DQbzrkoC5gXnaS_9su2nL7%V1)A=SZhaOww zOVX4NJ>CNxm+qlM?6YYCiQ+P^29`2uX?=fFf?wp0u5uXM7Q6^KsXa9eMaChA?)X5@ z%_z>wDNfHC=w2|c7$j&yJ0*eKZ%5pE-o<#>hT$E}2(si2C#&G%i9}=r(S)y-NM5ZG z^_0W;>+7v7y_yeR7X+bf>*dvxc9_XAfx_!Vt9-xM`p;pQird96Jig;BjJJGzLm-`? zdrXFZ5YTNP^#|;{02Kg`urT(93RO#)o!0s9(`ITas1>B(7I*wYQKiC0L98zUb@v}9 zkJG;We`AY&Z3t}fYZw=Jr;*R0rUrZ)9!dx4c@N(qXrO*7VHn^0w6cF6S9u++NG)(mU^Bd?qp1MTibP zcpAzJK2@H*J#B5*8zP5gZe~8o=?Zx%pHQHBrXX(suv<7)=c3f;4MxYK4&nF_<8GimuC})T>wVGmmK#mmnA$2{ds3_v6q!_E zXV=s9*fBxf)OY9-;~tb5Wo@QABeU%G;7B26OU}1J)WrOIOMBCxQPa&~bE9$?nfr&e;)zb!ji|Hu*97TtIQ3cBNzHNehL=97KFMj$9f0d;J13)ZMId@^(E# zZ-C!#Wd?VUN0>seuU~&gO8Gf+3B7kR$P|~sit0gR!z04>T-WGyKK?g!Q)h2X8`rq< zxEoxiVA#nJ1-2_rP~ZL;pb9zwV-UF#J-pF$f6%?96@CK%XN@{*I}l0BDn)VU$gj)T z#}XiZ)1zcN7h4x;t2xb%E3muJ$F)t1KnzAcJ~f;{AeWngO#7$a;2$MPI%~vuw9Hcw zZnbSTM(&7%9@#eSz1k02_0tAuHD?z)K)dtRMBykQUzLlDIKcPZuC_-!#Si53Cc%45AjsCf(uj%ztYv$KB)a13eFklEb7Z)_{z9v5MOP_08>$@$S-bzU)-Np zxL9{l-M~2y$k)++tx{?|UHh>FVh;#Gko@IdQxtIFpWJ~%o3#x}N;O2kL5tO5AS((8 za?7%i3mCx3 z(n?!*bySZ(dWQH5*(nTfBJHGRZ(<^@QSdAE(9jBEQr_oWQG_y3|9@l*+r3ee-cqgvw`Q&!> zS>yg9gC$0++H2s-7e*)!zb-t}Q>QZJninHp=zWy#>P~ZJ6B%ouR&8HgeF)zIEW?TG zX!Cu=C?p;GXSY+&+Wx4q?td)f{AQ`xX?bhGlIo^m;cx~Kh{bN7i_3FEj>sTm^?uUh z#02ci06hf<3w<;DZAv{~R!c;M_tCW-K&J(Ke}|+mc+@2T3+nAxgp&;iD6V}ZO4U1R zaBXoFd4y^nh5V#>y+F{&ociIi2`nOrhERO#NnE)H?M!?ePlCNoiVwn>tq&SJG zh|wKDu?E0G%Aj+?PN1Hphh(mNmVBWtKx%m)7~sS~G~2JO;}P5(1TfKQ`{hn`I!b|w zYsb$ifv> zWscxN_hwgvkDe3?t_C6CDvsFE`xzT2DeI9mX+IYJFwSdw^RlQX$j9*W&&_8y6)#v} z4s#v6_;3&$10R@YT3i2dUuD|pmg~Q){5UhjI5+PsC-Y286qs1Pmhu=EceO{n!nb~Nh{!GkC@%tP zYTfW?4ML=S+c!*b#7y72fA8K$p%03PU?|WIYdkOxa%)S(?ASbYTc$)>>^cDG(4Rkc z0NAxd@a1!8CLk-`>Wo7wHZBeis8H_)=54U;QT1~-IAZ_J;#dboK<_QLfK;fwQPUAi zb~E-ARzkF!>mRpg8tNy=cCvMKw1gm6El|(#Vd8}+1X}6b@&l~rjuS_KyLTQGAqrl? zv|_bfqXteIDXR>DZ40QqQ8b-wzi+o*RuS*G#@TpiDg}4N!aI_x@~vy+nGm|Pmph9? z!OM?g`Xrw{RZ{ws-vfSj#7M>R;Jxmqe{P$?wkwjR>>O~IYm+6q{D^C5K=rU!TCp)w zkfll7!6+R8^U8HpFldf-Rxq&o+|x0~z1_rh-#o5i;G-!X|CXfIPHzl~I{nkh@|c|69_P7nyKUsZp_u?vEGUcA7y&O4+7{H0o-}#u4r6A&g}{ zA&j0&VUPUoLtGc{-M6?H4?F2a4{QNqzIVw~ z{se@Gy{&by(><3TDW=Eg#ZJMoK*X0xBhFF%7OY*-ngsmKmpl0C>;GcL)Yqsoz(l!2 zJM}B_)R41#qRc$s$Snhh?cr?YRXx(n(@v*Z1}HcGzSvgb-0uzLRS)OM5A)M<17v>h zo`=Q?GziwK##b$i5!(tVDB;=A2XMqc_*oJ^`=Lc1y;!ctdg3kWj$)ZN5(8#8VubR1 zqT}AYe@RegHL9xdt)lTgUDanRR*u;iZuUlge-}^G+fCL+oAU?CJ*5hd$7$BSYe*?d ziH`ER3dmKzM+V;lc(^_L-r@?=>c9g&e{NcPK;Qyx?VJ!_5A|od6gztyGLTNq>;dVi zH7#KTyuipnnQk2+LHx(ERDC1>ExyO_7JCsw0(CpeF(`WSs5>`fcnGlSO99X!U;hg9vzUc3T1^-E!?cToS!-@ ztRn(ZtE#G+x0TA5!NMjmu~h!6b^@;{2%5r0P1`iE6xapmf2iKd(M95a(hqP^VF2`Z zBCudjp{J++{{8zW1Q&#eNWH_7r^MedBl0u@5q4}$LC{ah)8mgikq84Ri+W<7+gt}W zb$C(H`L#D;#1q?QJZ1)Hwk!?kuKz`to9=_?i*c(JFG4V6-`gcKeadjU6&%H8rryNh z_(MIM?krt~o6sKJjHzLk`g*7#(hA-eciL!yIiulhPjtrILb}cK>vhcQEHAqPuBTUN zHm${N18;*B`f(ziXTNF;at+qJ3fIT4QtvkZ4VvX5jyG8s4?vIoCu$gzAWdj=56+Zh zP;FW=P$!t!;2hT^jf{gxGXiAg3Tc`^9Qun(V@2`mi3IR(Ej!+iJN*VO9Y`b=K^i$= z_c`LfVj7du6!2c&8;ZB8`+tnR1yGh<7cQ)Tln6*m3esH)ND3lKBi+*7(%k|ADuQ%3 z(v5_4cXxMpJ;2#F@B4ix{{PIJnFmK1kl6d)d#!6-;XoiSu-u$%`a0#1M|c^S9Qn$1 z40JFKsc1LlM%kE$d}NSFsqo3%KybIwO|fb6J}C|iX6#Ou)lB`^KfrDVKuUbUnkA}S z>L8OBej@~^sIdi46bMSkldZZg$}uCnOC=HebgQZTflZ8R#gR%yV<SlZQ3!;<6jm;5WZj9NpuIfJ!a! zq%Zck2{c$-IS>be-u6}8YQ`@u3Sx8)c6Q;{2?ArQx4S8l)!af*Pi$_BBah5>pfN5g zhXba}HpGBhwRbSXluX@H2L=$cLThr$9lyu7fqMcl6>QZ#EKn8*XkOfF^fTskF~tb5 zR}8MZ#9shGbM^qh{_Lv4;O}V*VD)uZ*;l0e0eECb{P2+wj=94h2$1=$|Vs{^z~O1G>Ms{ z@d08P0Vo0NTwLvdkN^}v_YG45>Su*K#oSSPF3TCAar=8agh@Ia2+TsZ0VpRbIieQL zx5>$3Ih+jF)$!!A2s|2zed!Y&8NIM&Sw=j7Zb@UFNWb1?e@!mgpgnccX!v}%BHa18 z?uJPn&7Z5(V>fq~8WHQSBug@Nb<+5#pJiH>Nar2bxY6^tOd0Dns48cYM?YIDl5k>h zX_@5aiSeqE ziNSc0DR49`)k-OL!xlYXk_cOvH#VEoQ3@XUWd$m>7RFT@i@e$38Fo2NfLe!H>BGsu zE*Qq15dD9fC^((kLr(AFXMj+=;eB9(q5S2=DoXnoe=;hEyR?aSzE@Kep*?Wg9R5Owk^QTpSc2=VdLJ=V zk~vbfUM(fsM28o3gLD9%Ru033FN`!q7Z-IRgFQ?~_Zi~dsmvOR-^;dsKkn=IUyZkz zapBPq-R;o>{sJ485DRpe9QmTY+as8BoD^~qa!zqQ)}c@9o*sLmCB_@EXQa)VDG0eX9BJx}t zf4%51q)jJ^;%hk{O~8j*<)9WomVS!^CG4|OjlU)%MO(W! zWaBmiu!nA&Hn24LjW;NW`zjv9G+0T3`Oo;$UvT|7F_kpM|*mo-P((b`n3qJcVIznZn)eXXWFd^@pz{G8Xdv2E+zD4%c7cZ zItGWsuu*OIU=_wL5e)toqWOz=TBo_n7A7{^hfV|}B>X4e!Y~6@&xA6G^YF5~$F~h^ zZGQmsd5!T56-Ed3IA!`sAoKy=^Gr;|Wo3T{GyfO;g%5oi2YDYK z)~-OmPTvdhbzgwEFR{`&B5QRyi`(GT3{M^w4k#+N5=ktDQa#KroJn;!Hc$xv)#7>V zk~89jaX)W*dOvTH`)A(NoLw)(>&GJ?@kt*}H1g&>P{QP9Utp(!*WWx=GlMO&V*qy; z1-Rpn-y|~0FxrG_|f(o`q;;4WPLT)c#lm*x%_HGv_w2PsbYqSFS7)-yB zY^Rs}6uj%YO%)V5vF4tF7x8$Cj&cXlGjYSH8pV_!d57gJcTWK9HGXjI9?$O@1x38? zUncH@=t@hJO^6d2GBdaQO&BeAnkDro_*u*3aQxp=y!nUPnhqGr%`w#)mg@MN?TM7tg=3+j3h08A^6VW; zAd+jIyC{%YBXeK{b#S^pNZ9YG2*0B&+X!5iKqgh{KJrG_53a8o2QAMizx1?;quXDT zu3}=m0tMCDk117SEu0zybqj}SZ$Fp4MVlX?ysdtA^7*K5JDYdrwkJq#UX zGfu0X%I27D9<`{}rgG}XzZf=1k!kphgQaPg_$T6OeABX<=voX(Emg8gA8utZOXG~< zKM;L0$dA=FgOsP00b1Vwy$$+=eAn^cHVE!1yp@XbldNbO94d$T7cG0P2-C(x9gG#~ zaqCJpy?9c2qG@gk0BoLZ&oYt7%yQAZq?=TbFSQ!8|M7cNtvS%3{0@~MwM->UeCw880~L)XfkE! z&(Us!?&hWH^NB_!2|SLGQwdQOPxSg@Uxvi7M5)xtrNwhtYvA&c4v5e`#7G=`>c$ZP zj@Us8SBn&m?FCXln8K^J1R@ui3 zG@I9t?SWqUBVAeI9)d7&)^xbp2TaK?Qn8u#&5;RBBnj4rz?T2j9atmk2=a)|2fN^X z?d{7pF%2!Ub{ggGJ(qqwLHi$&Z7v`jn1HYjsFZmr!PxF5leQ}=d1aTWyAZfE0Ymva zf1^Za#CDqt;GMXtnFMrP4)D>3P8tQI0+b+N9skkvcLG0d0x9_#8gZt_wMdK;<2-J! zCY*ahZ9_g1?-rn^;a+OdnXdDkbFtx$#m>0o^jrxis7)gD!q*_dJymER+h5_XkD&ug zmW>p>Fv}k%@%Rxkznk|!Grl#U6d~7|NiU0TfdMOH#Hw?|su_b*E$U{@13QdXf&_kJ zZh7(1;Itw;W3|o81^{Q<5+jwyXO-9csL}x!|KK$UMy!7@r&6McVh2XN+#WN{xJx6p zHa4@Nd4h_#;AsQQHZa_0T%TU-IbGdz(F^ywmMI}?pZ^LX`x5y> zOk!7Tt1X3wS-)C@1HVyVEKs1`5Z1oD8Ne~T$w?Z%9#;4Qv>9I-sexff%fLjAfpgYL z19%c|MqIs0wF)7B#8@bJ5C#PTfnoj5t@OA(A$TT>tSP>OS&eVwf*ktI$d;Q?7y^6i z!mkj1Y-NU>-^spQpm92UJHzr_j%fbOG5vAEHL_uiad$inHk9X^HW z4TGu7AiQh75>r4RaU~(1x^Q%*+^a6#ykH+ZqE<2@e+8|H3 zv&U}O$vc^lJSF`5SmwZ}>-->bPfm7e0!MTjuXJuI)+yx|7pxxJ#7s>ZtKV|$r3(32 zy(OLV?%R**Z(#F>bLaKHVq%^_wa4LzybMUns8o;z@z!6%!q|aNGB2LTo13tq|w5?-6@wR4+`zK&e2P&QNV446X*B@8QpEB>sdJ zF8&}Z2SgbP2PZwgzk20f;S2tD2rTiq)RnR>e2$?81o*>Hdle*>DKqIs^!-3k{WzvF; z(mJK5`0f3z)yp30G=4cdC2i*j6i$77Z%=wexHI4AxqG`9-YeRSKUS)Z^q4UcvMehL zx0G?dNPGq}-S{>)IQX4=ZSJ!C`G0Ym*0I-*>=H^M+e@BUY9*|%m8-+nfqfO^Tm>OD zAajmmy+0{SfnJV2vdVT96Odvwi7>j|${`S7Wr%t= zgR_M`EqhZk{bAEJ!=(;^tTt)E_k5WEG5?Y(wi1suO1T)#w(P zVT^!%C#;e`VNnD1+>*1w9lU2?R8z#QP{P9QMBEKf`b@=+Pvd6E_ZA@)qk)*yr%D|PIHDDT{u}T~ z{p}|woWsMzKz;}2&xo96^cTyauhwh3imvwc;9D*kJ#e~!p}gKErc#eAj9!n@Ok zqx+TfIEnynY5OjB*4#bY+IKFu(|2bLs||Picdo!FDa<%smj&h_$c!51XCdup5%_sF z!ICe1TxlwQ9Y=$4dm&4|h9{n!`;EOH*|8A8Pf#_k)kJ6dS~PQba?9iyR>DVJfi9T|%qEbJRv3oA-1`m$Dq?$5hLi zwV`L=m>5ygx=@$}cXqINPaMUX)I-_IJRqvN=W|Nt4=luaUK7v5uDF~bXKMh66#@@P zr>&eU;4uXZtV(KXrYx+ca}uASKJk2{#6|!-;Ze!2Bos$2@hkopfOI|qX?U}&$5%JG z+RF}(3)P@f8rg>BzVLWohxE>3`rY^`34{LGi*K7X%q;a66fCs|a3J{xSdhYbKLFe@ zjasMgz#jB-wwkOmJ?`52x~huGCjj>V>Pfy5NNE6p9mOhNfk5RTSh;pdr6Ah!RU9AS z!$EpTppXNKV}Qx>@N~Wi=W7BsZPvzx?yFsqmJf~#@2(2j(3T6rp_Ap@^X?pq&sjx1 zeu#&L_^^m%k_ihbE7uDSIHeWPi(ki&0Dm-PZYm@k#4_AVOUCX+|8DAE1!XE)#X03T zIEBBWwnc-sHi8i5fa6^0@W$qfHQBdW5ek(4kQbD@2{b`K z0)dBS;m5juKqV%JV?x8watN&M_dnb8>G-kkzm>zO9mJV44!(c~$rHIF9FM_g6ZqQ` zcc<$r(#5gZb~|}{fy*-V`&iasqz?Bho|fwumAq?}SR+jczW2N$3pw+AdCqY<^+&&J zI!7hOtFMk;-a=DwFs9Tx3|2k>nbjOzT;mp2u7Hp)oFdB=|17*c&^1YpIB#xQ&9QKW z!H+xpzlttkdjBfUz}-Co`0oh$3_^!2D?vb=6M4(i?b+&jexZ_ip$|q`3ug`ISKP+G zb_35MwG-y;+Xhke=OhMJK@+la?y-MH~khgkAKL|EDni{Wnm@l-8 z1_XbIZ~?_sHQDv^w^ugYSOA2#JgXg~Jl}uUdV*(quof8@!t(NR;Oj9^)CK^1{rPp^ zA%%A8auaqQmyobZ2J`j&dw1vwBkSqwBi(rUqk}SJSAoEJ`db z9{juf$O@BxrE{u5 zrgyHZESGR};(IhOaPux!tu0RXWA*2zW@bivk8H=~@?7%s9>rjF2)LVH{CTTq;ca*? z#3(}B8ncfgR6d}41E3s~RWvq!=ikfqn(OjWQbEm&if&)8vYwsFELXjv;*e?Vkgc#K zaPS@VM`H*{T9i+^;w)DA3B2p>A$H}uEk*F~g2bXmKqx6b}!H)chwO7V&THIfnm2i2hB!WN*K|aH*QVJ)xhn&*AHwW-cy1^ zq}l;@4I_&x4ueE`V;^nCJjp9F^rnkKf)x? zU|qGujc!OM!|>;;Djk@<1dpNXm_m&b7I5!x4C3KioB+T+8nG%Haik84&Ro#eAyzE6 zgz*w%f!T0RJ}^E7_Ck82dC^fH#?*A`>b)u#Uu0%IJMP6A9)`yZ`>_ej&JKRp#4FtN ze=wXBY+v54G297)AxuLf2xo`LsOL4~hihC7K)T)mY@-2p`Q9TQIBB9Gz_s6uwRmmU z9Y}`#LA7zyzr_XzNI+n`H`dZmpppU?K%ExwCSdaJP|6c6ot!zc+fRCz26D1`4Y8Y9 zEBGq7tW5|IGff;oD;#Q|I#E%#8otau-4qa<3W*t$E9ZsQfQ<3do9yDDmdj>_>s2k3 zm-e;LZIuBAI$o-9!P^Udez67+1Tk+%aO?nLMj;(oxF>8G7I=3+cF{-%YpmznhXfP1 zi-NGgpW+$rf@7>`bEgH4had|oG2G)M?HWB85zd>WIT~a{0jfYTAmKP2E>$EeL^ zPQy~zM))nQCkP*C0^a2n;GISS$Im=20zIPv`k1^91!Q+XSJn}@Bf^Y1Yu4M{@6J61 z8;%$n;Qj)Ersw|t4gdWqQCWYQ-?DtDNjB}Oc=4v4CG6e4qoFE5VUW`9tBx8XD@1lA z9`Zq^?>u|~MYztTn+LnU)N0*E_3dZdY!wFh7Y;>6d}$JzK*#54Nhkwj-v0B3-taXo z*#SC^Xd~ClFe5j#cITH$h6!UZIB3X_FI{R`a^w~kX=QNG?%o291{TUgM38Yv+%lFi zoM0OZpoV1?y!;tdFu==GLi}loE2((C0NkR0!5ix<-?dY+bL&=C!kOO86x)yYOO85c zjB3UC2m{77(Mwof4$OB$%1}YTQM!I4-77nde9yRVcvPFujG6=kmHAT*6wmXxpCw%u zLFiW*PaEuC^vk$FzQh)&_#`ToRAcYBp;Ty~6@pMNxBhcCjBvgq`zyia1u^C{5~(Wn+-4O)JjUTYp7G|;6(E! zyZlZz$@Vpe7WMv)2uiv@f!Od7?XO#`T+IOeg$7FZQ zo|JLDmac$;bpdWA8(_lP40!xgu7us-&~cugo~zJ{X2U@}48J=RT=TpjyG4R1WZ+AH z>euZwPkR-FkYHQ@3*gdSwt0XkU34bFKHS3k@7((JoLU*Q8m(^Etd0O~ak!=2;G(U( zcc#=~UakgDnvV5g(b{R5=Fzmn10EH|J8-K29WTcV8Nfj$zCvsx)R|)KShW zBc#rorz^U0L-YhLI?TdXZdg_wt}`O|BEpj@j%ck%TF8S2M2Pu46z}K)H4A-GD@3Ai zF=NNWS@_?4GA9Hq&Db@by~SW1%5DGp34Xtv6Bz$u5; z-*f!&88-b`89UOG*Tcp9bU?V}gmbx`ha#~hB{b(a|NT2NYvbT|U0l*#Tw!4TpU*C8 z7y2V6LuZWEu*;3^+5n=~KSe`?$XC$9LGkfV${u5Z(Bjlw`g=Ly4!YRNKKi!ePkwA7iF< z9BavNp3t&v16rE(zJz5c2EqCK&wTIwLPG)9({Nc87RSSaH^Xi>+uHN&?mPb5H87UD z?z{7eOPWweM*)^L06uu7p*&{}_PnpgyK|S3CDQi8p zvOG*>mVuPV5hSNv6S@Y@H>kc50!Bbch2zs&uc~%Iyc_Y`^{NsuZ%g3rSl25H?4km^cFDDFSNn6!HA9@^vXAx{DV?Q$)cs4Z zne$*+se!Is%swKy5Ix*4;YEk8Z8bQ2(hVo>@R zv<^{k8wD?rR4OwjAOP|9h~@4F+07u??D+9*``rcE<Y0S=r6ZdSY z3ofy7k)`ggd)B1160mr%^iSKXT51v_{s%Ra)aOfBqGXv!G=O412Gz$G5oU+2B1!7g zpWhDD*MeMdr*}i(<)?kjtyCa%yvFqPl4{5HSufLUAShjsI#Q>F zWX8nD!PS%CCyxs9VvV%xb#uiVMH?n;Amm>=Om%X*STey86-M%vepN1m4oXwr@mEEJ zkbbJ&cUhC-xpE(b{f~C7E0W>j=JwuGNBsT1A7qbpzs@@lTaSAcbTO~P z?ApJr=qU8<%-Zb3YKP|&1cBQGL0gpUnpQ6U?HE|Ju#_6lW2@U<&E{wlAi}7@hw9p6?XMwq!uOFN14OCcwKpeqttGhERtLxc7LD-`EBf#!l)rN(5 z7VkP}jzi3dvh}mQ`Rls*ZkC%Nmczo^(S|8oSFknt_nk2sP9z%2*4bu5=FWI4Ec8_I zraolrIyB|v21|J65eC1pDOdo!^>@4Prri1>G;mEcXw5qWhbvi$K zCnti@YmoTfAUhw1g)YP;wg_^ClA@BHTZeZX$Sen^E_|RlOpiF&YG*rsq{hWXZ-gfi zo!NQSjXS2@vy8I&HtWS#4N0u`7yGv>OE^`{aHc|0>PtA^Om-7#b5V1{KyX{GA{a;S z7?Oa*UfX3+PESlQi2Ax`(c>V`Y%uo za3=#ZaU{=zzAKaD@EDS2)l&XSiv8RD3#NP(x9e8CN<2w%TqNaVuf4!;VOYouam4L{ z0Kss<8np`s$*<`xtA_`a*iJuCnhM>}@>Eit^+@$Rp61E^WtXxY&s|~0KR~cg9fnf7 z5#+s>Nh`RS6CT_Ou>eZR%jer&IvKR6DM36ZpK!X3hcvXCXn*UCGN^>U21-2pTLaD{ zGHp|zg$qU?!o%6c$=?^%exGeE4T)lIrR$*eKAkk6bJW&yYtzraaTP4v>11%D8paIlkgGvtHt zi>!>-xf@i@lH>t0`-B?jS12;G81m zo-XO5nu6+?RemOXv4XZ@j4pQjXFi0HrgOcC_!*&KqV=T*rG- zsT^5sK?JBoP{tiuGpmF9T)5lHi)qW_OE=gJ5M=C|#6#5s(9nYUgH`oSGjITsMPg5x zjk*EUrtKPFZVLtwdsj})tQQmQ0u{Sw5Nieq>t~0fAymrhvA*#2u|SuQVHQ>%fq6}V zp2|=OES_m5v-FBL{B;qP(CEdcJpFM#3qGI!aLS(z;Z80P}l zD#gMfl>~^=zijd!m0d|FD}hVHir&m8p3x8Pm02X04nT*MFq-{QyH3>I;$hm)-5fuY z)D1{W@BN*w_IpcWqKsk~1p>|d>W9_;1$8vwkeT}jb&M!)wHK&6c_wh0BZD0xS`En| z-Ij@*Avt87);dtDgX0#sx;6jD#B|e@e(HG`l;d%KhwFOkTZth*G)E+MeEq&=Jde!k zZq}n$+T-WUy;vn=0n37pk-u=aZ{pAneZ1R5$89izIaj@v#?dLo0 z;TK%j0;;;j^zGM7U=V^XJv`-ko*`^#ll!@fWGkyrxE=kG{#wd`9+y;p zEDZ=WF%+(6e3RbfzdPxOyLAm&MRs<0J}13P=aB3keAh%mUhP$mzdC*{@z`^8$(G4a zMOsnOhbaL;{k4rIzL`U8&j{@p3{C5GP=xf41Y(`DASAY}IOK{m%$=QSYgzG~GI|uF zy~kkppfkJ_Gv!o98IF%@E8X?S%v@|%o|i1DL(;C&S>~8+)fvx z&bdk&{ZzY@Ut%J9)FRS%5P~)-&*$6)LATH1?Um$Ps(;E$Xal5wc?o#Dg1z3d<+e;9^g4Fm}Zb`U7I5~XfAIabSH^7-@U$+D}f zt0#fV08Jz-dQzC? zE-hBPx_Tr`GtV0j{b|jPuJA9-uMy%A{J-+zO+sZrUgY6-w8LX~;{ahzvFNoIp>J!N zvHE7rq6%XHc%JZ`;)6r%$d%NcTJxfViw{6#I)CL|I&-{fl*PV$>Rc{AGF!H%?V*1_IzWEKDS_cLf z>aT8i`kng%){BR(^Gyh$OS>o`1^UQCD5G@pWVtZuwcA<5bM z*PLPYilvU6zguZidZO-Lp2B#=xN{sHFhSQvuN@*aaYeKq^7c?PRj8oV+QpWDLm)IE zzPXc<%_!?bygM2bCdI&%R8d{HkOWe2ZluYMu0e+9aF;(vT|%6_|7u^ab6#yFR+C0t zOf>n+s60OtswJneNDQ4RV*$ZiW1^HX%M;mj^ZuOY*D-PA5_@A8c zs0?h~u)h(I*}?lU z!|D0XOW1a>&s!j2HzElO^-=YL&*Q4Snr$c0a6JACjwbLpfdE)gJr$Q<4iRnU&9<~4 z++BO#St%&qBbFN*8`&q{D$-B5#VmDkQ;$~cK!mbxDd1Y2GsT@jSiGVm2@o|8O-xi7 z$EMOwIP5CwL)v9s2~Is+3tV*@!Op-@0&zc01QcY%Cz zzr=kV;FW@!Y`&$WfJyonNubF5>d*Ym9#eSXh!&m8b7gXWOro_K%fnBu96A!?_T5;0 zoL@CszYO6PK|?!Dm}<`PdOu6t*yN;YR0JnD)Vtwgz~Y9?rKueN-=e6>x?KxkS}8gV z@BN^1-s3uWe_K2tFi^RcLFdn+ve(&}1E*B8>JaY*z+0yi&rN+ATO)^*a!KiPWh|W) zi@F93)po3MYv7r>&EDqwgD~XPpkX{gK?8Pn!-zh(8jODAl(%SQ0WPrF1cWQsgYt5P zRsYynWl56th73YIq){IF(e)Qj7nqIBfd`BsJ%3pzkKt8b=vApkT$)5&7Rw>-k&@34 zrXTts#-fYLolZ@T&&b$r;~Sj9ndLG0j8;3Ll6Ogt+6$B1_`SXYN^o=Y_?;6_4KYf; z4oZln;eX|oN_3IRgfQ)bY=4%5)IDCH@>?IjDduHW*!crK$5u?kTRd|Vor0!gzMc!p z&ckDU>VP8o<5tYf^$B24HfXHmEDhqzt|cQQ~TPN8#F>LJs6qk{(tjqnqRQZw(2{Pi}R)tL!-Vhdg zoaPOlZq@0l6L$QxgN&e*!02r*~)#=@0xK=*mdjy3~bHJ81(e?h+O(?zeTN<=M<4z@2FX!Geu49 z?)9kM!s^tcP-7Hb{w}QM&NAjY)U{#e<={HguO4masY(l4unBB_a-}B_81R)X)GMpN z+}NbBPDZNpC=4LT%`v*{%JuXc6=g&7!6$z1)cBBZnwpneat-!;tb^8)x*PVj71rPO z>}rxWCF*%-Afa)6kz7-ZV)oV;^y>iyT4R(nsHm&$2up|cGv5`j866CLi2wX!X9c-5 zi?k`*ieh3)q^J}_JKg3cW?$Jr^^|$mZt}g^ab0O;sS=eW*g?Hu2gRWL&|Ci?gT9!) z4hSjzuMUZU#+!Nod}=vT@4B{U!0JjA@%HXJFL6AK;zu>9!%O?=t@A0nde(mYd#64D0l}wFpB_jJ8=IKWA#zrJ z+%zf$#Gbp0@VgG+cw!GsRu6w*W(I4A3=nLlvd#EfOgU8{!0Q-(Y;KX)qwIZ_Nwan_ zra8Yh-*8HX;N!vE0JGIPEBSQ@DLkX1qDaW@41bcYk9>Rs<^BMaCOVbYfIm<(%xhii zo+g-!`bJVzE*j__zKfAAdLxl8EC!OT0SU0ax%pQW#sHJp5Ylx3;a%S#z&!*kG+=W3=Y^a-Bzo3e0#n#vBoKfNSW_78 z8zSm1JNDz6?G}}@ZG&lIT&Gnkj1@~Kw)OK8@G*J(Mm!reT4;AZ3LU?319ZHQ3Rc)! zU*(wYC8zLt_gM6&n)8J(iH;5F<&OraNRjt+n#74o$MU()=^BnT%0J<24#eV=h4XMD z*}ZW6gK}b@cIkMVDv_a}M_xy-`t}S*bR30vhsL@{5WkO@w72{v8EIp${`)>P%Kf-c z7xipDy>AFHUJZQKfv=H4-)ATw6igTX@+7whPfAe{)cO}h>JM45ZP+84Re}i4+)!yE zD2RxR_IlDb7Lh?Y0Y~C~k;X*(an~}?BHd-pQ0Z#4a=gp|S)T5;lCWLNBys(FXy|u( zKUR($4Jt`ciI9|iiWX8QOBo|oNc+)FL-mmUlDFW4!(y&)ZZ;=DFL*sjk9RgK0fh)C z^X<}?o`ne>Ah(n8_0b4ken2PY{tI1Fr{mf(^77b#%Jnro93JFXhvfG*FIrDBV*&uj z^3U$8`m52xHDSQfR7TSo8zk7wtS*YdC*+3lejKp~kQYzy6!LWwpN%-|{nqXV(sZ6--WVH2CBc99(ZB z=^X#sv3zHEB@ALgf`$1=*8yxxrwwRmRbaP7Cnjp0I+A-TSDEO>uBE=3fI@j0v@cOX zi*nCw0OP!d;vD#MVeXOr^9boIeExNDo$3sxs>fSfTd!d8W4!J~nuQ&aJ% zz?bxQ^ZX!G?I;%hh-(pNm@1BFm%I$7bv3A-e8Hl}f^1^%Cv5-O?qP)yR<|@}CH031 zjtP^H;qauzpmyndS`IW63+W>V`vr_zWdtrMQ)IJMcuI?lZHB#@00a!}BIpt%b=@fz zC6`Q6lHnW@wbLZCzrP5w=EiF6FW?+G(lLUIz=dY9?FXTf=wWVR#`!bGDK4tX%ChV8e zhK7aO=5nQ32HvJ@Tw@(DW^{MIsaTT>I(N;Q-ySN4N?%!XRjq*BfK)<-J-E!=wZwYH z?vEaJ<#C1jI=pqa`)fz3qc(8;oSM=CI<{M1J2$F)H))fv4^CLc--=mTHq8%<7BwK% z6~B?@QzBA5-QyM=J~($+-rwz03?h5>!p^~K=YZ9cUW(d;-PSGkI}#1@tE(VV zMPZ-RF&ft_M^F*c$4W2uYlpdtJ8fS$MEL|{ZqR105PIRf_NrR&2GoD7px;i|p7fDR zpeXz&ClJ6r>S}6A{QZJpHe@7tt$MpVe;b-~2U}NB{X@{iS>{5x{n{k#L~h+mi`=jJ z(~EUO1<^FRnGf<+*yG3V<7{+0pDu06eKd>@8ho|~yQ-|(?z6%|dsMX)%mBHyb@p=2 zOL7v;AN8A2D2MA#ALR%VoOFaB)bO4bV0p7(jG4k(0qM7|oRW{_!^i6h=hcz(|JJxg> zE-XZxlH<}f1Hbs$`6?9jLPJwmlFzBD?Fha1cbAKqdT2)L10 zAh^^ZpPVQA``{8~sfaKk=!Lwg4h~EG4fsgovq6P_IKVv=_uOueZ*759nq8FIfbFV- zGp(Hl`Rxto$>^}H9CrPH2hXFT>r>gsWd0f3i~MC1LAgrs+q%U}u`EKG=j0m#erD64 zyy<9b1FdnqBr0#r^Ot}k%XPrzi1}~t^!T@+UGWp5${FCvh^|A)s3Gp>QKNSlj1|hH*v>6+Ma#s&!h5FP%k{J5?1Y$M2__L@YfI?NT- zFW?mAkgthQu8B5RKT7iUN}Bq`tRM&NaOU{6_ItRHozJ+l>;p!A71Wk51`t)PMyXe= z{E&Ecsu4%d)a{>*ZtGhEY92Y>T-aA{%v7ws+GA-*kM=`61K=C&HboYH2znH7G6;GY zWpNDTE=sJu3QVdiR5Ng2K5od%rEj>IZs-L+1jIuqqkT{r8|h(`Q8Z&p^JvC4-E1sc z=4JfXiwy0n41@e85ONL@R6<&vKpn;He6&*1K>inEG=MhpZ_>?V(qa91omF$c>c6MZ zK_LRWnMkp4CQz_Vr<-0T>pr8`rx)*B?Pk~VoRwubY9gR5lU7#h>DxVXu1$FwBIc2} z3i~FdJJ&pn#>el4^-9+FjoWOs(u((*uWl~P?scfX;;L-EbO7Vcn3Cbl0Ou!8QE7mS8wcmSWNVl~x?9+fbFkeqNYoo7aZa-t>()3DQSWjsS&)A8!$ z`R*zHfR^XOuyxirLOp7pE^|X-hPb5>JsF#=pkW|qN=mP*3pFdOi-MfiYgedP=(aEI zfb0w^naNw$MusNg$y=B&ZqFO~rMPAl(DbG%LQ`p53m-Qvvd+%V5_4O7fl$Y|%gakq zMYeDf-Zo%33j7*AAF}ylKk)Z;9Je2|Pk`d=K_UfUM0!|_1_%4I8qwmw*)}pr4HsFs zwmMAam3L{;`BmZO^=4@>)l@EDZ5OVCH|=9btRRFtKz8d96F7PbZoj$>BOu2D-wgQ8 zTy6s%%oBbis=56Hk3xy&hM|($R|Emx>vZ5;HIK_(laBZW0KHBg({XrvqV?a8EjoTF zIjZbB+v&(=ER9Qm4*!sbiPUvS-)V$e?}wSFrJpPXqDvV$e$&o&( zi0aU|Y*+l_NI|AUZW^1L3nm;W6`z4=%h(QQOpe^|%V)Kpv1k$~czK$3aspYQA%_F@ z&Q3eN$?uba^M?k#R(ABy`ZP?s71W;CUIJ-z@ez5C+o$gA zJenD2WKVf9#Tlj$T4Z1NeEB9@6*hRbG+RkLDN~W(*z>>*D8S9`djnpFP z7JPp7x_pHk1NR*pWBb%cdCu6TSx)~yHvHiV4go5*(UT_xbY@Y-izf9B?Fnu-X7kTA z@WHKDumE}ERc~g-Md4acoOFYh`5qV&HV=keuS5ZVa=qcE^6njb3}XPvV~b9U>9+lI zJAkS~$hysHr1H@4a5GSW^gt&HK=kXgUDl8U4jzV)@V*1}Hq^a=q6km>q6r^>0E&ts zNC#od$=k23Jg=vD+%8xFzMH2+tiE`T9fVL+rMtMZWBxzrdJCW`!>(=HK&3&EkQ6~c zTDnC_X_4;k?gml15s6KQbhm(jG{~mA-5}j;y8N%3=Xt;Hop0u!*~~b@j5B)Q`?}Uz z=W(9vz!jY`a=g$wcXXrjr)q(cMdwDVo^`cigxHI#fxRniTR{D~t?@7RE`sZy`7#y9 zXa9@9<@Dfc91!*zkVPGrO|g45)r@7bBb8n(v-4-Lq?S^(U11X&&DxMj#0tw7kFB~F zgMyPwDu0uSR%m>Ey$ineQGsQL*WY?OzJ=`+=$#4Uhz$KQvmU5<@n+}?d5nc@jn(kd zD${v`OWEKa8lcmwYr*G(>+$RD4I-H1{lxpCJdS8GeFjS*>NIk4%SG&F>wU@tm>g%5 z>Gf*4M7rv>0Zfgr3?6OQxwIJ=7ZxtrT6Hc?p?`t}$HGXvq&mvi6nS7zr>A%vP31Cz zs8BRLQJqc@QMDO#Pl~6=EaM*P@_E+@O%@3bY9Qd9qu+?pO&j7RJshIP&hEepUTgKi zRolRR0@gMx|3%w=q6VP+J$MtQRah zzD%-;D5M*Ekd%+XaW^5hX0&59C?xM4Y*6m?o8*MmO2_c|p$d#4twd0tbYrO)9 z26H62V1w074)VHW@q#JMWAhbI4N(I`rfT~U?LmH>kDebNANLh{+ynZB_36$%g3p=K zoLmzsaUsIbe?jNkix|{B-+z}0*H^ASYp%D~ssY3Lc=1Vt>*4jn#;Jxz{A5~N>OOR= zCFJ7Q%CDHK>-=fRC>izhtSadVBfMYuhji|5b$>gp%mB6;9M0!f=x2JS@cE(C^G{_} zslvA0H_gn|^!ap5zm3Mla`S1Rmy5%&nxkF0yWD`izIGM7S_dFy)ZnleB{r&afBczP z!C)C=R@mhzKkMa~*7UD5$9VwS^re|ExuJ}dR{k`_w>1r={A zD@Jg0AB2V~ObY)vXT33-O+(d>r=O#(MEc^?lSo^^FDT(!{D%*--9^*8M@G3A@G)6s zV+Vb2kX{?ox=TQ$Gh(MnE6Q*K7vVq+RoPaK&{SYsTTOHvBcH0G0AW>yd_wm?P~MeC z^uW44Uv`kaK(d=fIH;){H_(b@lDg&7Ru4c|I2`5kEiw{2?Q(AI5THDMTV*XiqI)&B z)Ii1K(!3?0`SMo~>c-+4tU++FcPVby6$Jas1D(Y9&3(7=BArf9xxZzu|NG=QVaAaF zPPaWk)fsBUg-H+z^b;{TF*Vim_wU~7ucNbpNb;q8N#L{i$cf8Ygb$X0AA-My~IRsR`Rs zXiOs6EM*N1nf+N0P%ZR{!%c|8OT5Sl<>sCatR&_35;pHYxB-y~aS(kc!xxrN=5L;_ zPb@-KJ$9@>VWq2thDUI+A;?0(d8*eL|ML8g33CcecNCxAjQMhW{>`V!r>U!UIQONAn>yTUsbDu@lGBcdg&Ik1>TZ7c|-GIVR2)l zr}}e8D=Vui+x}3MBrrKs?5}3j?S-z<(w3VGMN&(uTsz~>$r4ZrNBJla(0#GAD7o8BXGW^!ss5d3eJisaE8DCrYFrUVRU@b?J4)* zvz~}Bvp=t?EPE<=NA#CGxa8DD5;4#z-%dJ1O2Hsu%b7Sx&u_&SO_W9GWMxJgQ^MoE zEud~GJbAD6I;Gw&GH-vXq?&8-Rbdt`8V4TQUvzq-;-?jValyRk3GCIA1I#y9H}Yr0 zIKFWH#dA4u7qv<296nh%%H_7-|MOPQJ>O}uSTrF2qeC;aULhX~ID`x+7*NPtkg0oi zn_~q_)gr=&NtUIEY>I0-aNl%FB`~NLkiF+i)j_&10xoeM?0Pr^N5o`eZb)GvSpEk^ z=dEW3L30Tefqb=BxD847A7StBx8txRalL;17DQt`ommk73rIAkWEGyb3PnB#guXFA zX)#u2N>Gv0mJ}7e54Jt4J*)Wxs>NCINvS=~?W@bCYLVQ0A27wvdmkWEVV#upWXanF(l%U z>GK9jkkdeoP!2l^^w=!@feAjIK55&dYrWXm`towiQ(P5(9Ha87ov3DF4Qt0!*O#P%LeN#Pa{(@>p{5DmBZ@hw=`(=3=C~#H7GET{q@FiruP3cmuse!S8&z|c))7Sn5MU+`#B8i&5 zLIDBM#*`@`bq5oxI?dNo&K>NOdj9L7_n|bJ&eOt{W zt+}TNc_%TH{xIl6f?2h;iaXvPpr>i3nQ{(sM1bIo6v)34mMbqO_qEqZ1#=;u&*O;h zbis2k%REm$ec|MXN^-Yc^I~Y$(^=gwpyLZyL7P_URs&`PPl}sB-#KM+$Iqa=XKoUAURq@Yem-k(DCxN`NJf{WGJO}L z0=r(Rn$ts@?1G*7sN*Uen6)7zFG&?Ruf2CNQ4NjTR z^=A~-7M8#g9*j{dCtOaAK7zKCDnp=ad{3n&Cr{cGRm9@p78Wi4vf7a5g<5sBMjs!@ zPN97^moEZWa6OQnnrS7SMQV~CQdjq!1r9W`7D5C z3Fi!IbS8B?nG$q;+;f7MJDMwJXB{bP+mD}WmGRX&u2lc8s-@WO$#+~g1~1!Lt}@^2 zt@@K0OFyj-woC!SCL1CCHS|XJK+FaZSD-Bw=e_biV7Wr)&{_x{Ezs;T>g(OrXAk>% zT(+k|M=5;H5Y`XaqA1=p7pxV%+}YV-kOOO4ot&LZ&Zi8!BbhLa2Hr8 zK~Y{k#pAbu5uC=5K7+(f0`;Xq#yb)qR3P@Yx+Z6|I0kcDY`#b>VVv)&85IJa^QP#i z*3_Pd{dBQ&oL#yd7Uuh=TZ;C51k?*@SxKUf29-6#^rUf`-?!WR%pk6=>Jv6!QLH#! z5lMWV+dr%y<;IfE7VcMMtZ+@VE$eO$2(N#0) zLgU#6u9La1$J&Sr<*}KcLPH|n84r>u4PS=53<}Qc0cmpQ z9!5BF#eKo>gpTgjuaPfYt#aF%r`>r2I%KY6 zuNtG8KL`!{j~J!(tHeU!1!7M285BIVVu-FExg$ zJwS))lZuxVpEQc`K`cVAZv2Mb$rx1NyS zW74K8RQkB6Uh`}Ke(icavFhTG1}H{0Jki9e3ecN)p(=8Eosnob?DVvQN2Xk%T!}lo z?7@u204%N&b7^gdl|}xn$qM5;x-M`q01MVslom<6>(nXVu_(ncB-^p5fyudRLge_e z{&xbSg>^GfI({>hq``1Z>6M=hpr8`zH)Orr2qo`yU^GCs!+{=VL|ehGPA+&u6eZ;F z^&isr1_uD6S73$d3+_cPc1Yb~5On(Q)Bb$>v>!fvNJK(%-vNO;Tg&UTj!UTHrf)yv zp@MFd`>}wNLNRgAI6+$l(}+Iwb73K{B`+n`Du(|1)B1%}z%9DUbdWjeW`57dwGXMa zbMPNI+y(%44}l2Vms?OM#X+onTERVM*B&|wT}Ou(lPx~d%e_WyU?1cgb*^jRhTd&p z8mkjk_$H1_0OW*zrg6Rh=Z+Jzb9i*!?@A?&1HGo4lJaN$sj;eIJ3|b{PIG?|c;;!u zJ#2Wc`y6CW_dip96TaiC`p9p==3bL_zEF0SAwFjh^1C%sC>vt!_(ayTsD4Z{PCI@v zdic;0Av0=ZGiP99Mx^}3&NFcpBb(!B&wWPNC{Bm0lY2-D;rwGk#~2k$h|YECkOUf`2C?@XPq zlpW=(VLv#D*Ftrqp=e9ddhI`L!TR602P#x9uBN;dH-g?qoX>ArJ=lY3NNp=YV&NeHL@Nmw-U(=6o zR^|vnHQFM~G4pF&RTsn}BZ(mw%OB)9J^0q2q*$nPLv4qFHB>XCgM>4%@{}pb1weEj z9v+|iAF*NF{VfszKfV_S3dpb-qoTslt%km?YGoGB>Ao{C6>K(}QIr?#JXr3o8$0v> zQ)!Cp^>=-_E$iYeh3BytP*fp>kK6Q7b8|CRQ*Cpz5Yz_Ws;8qZpm2dAwC=S}U3Dfv zV@8HRPu-it1X~ZCs4gdvW}-LsBF^Q&iDLb3Fp-(F^O+~#!Kw4|o|)+M7s^+Zg}EI) zqC?)OleN7vQPB|*hO~HKZ05rD*5WT?q0Jv#3)-Ot>HKV*{yIl)wb0KIDucxW?5`Dz z0T63)K#DP039)#DzZRz?G60K5^w8n498TTTf>yEq%kFSuT?m#z`YrG%sX9Mkm)~ntZg;HosgZK-IfidZh|L$`W_lmFoU%pa(N2MHAKzZ z;H(EU`Ny+s5oJDch4=^H&t~&8BO{eoX#nQzT%3+fwrMa|e~$MjHvNhk#3A{%Y@B-o z9lK4Pa>sd8mfPCgYaGINH7bnHYbJFX5${ffuXnB$>gtt?((muK7hsv(^zw>68{BK3oRTUp3Ho44;bE7q#8sy_PN!Bsx7EP2ijeq9oV2`SZl~Ew{qM3mBvo;3`9~Ws`@oJ0ufukDZ zH`qV1-I$40N-B~3gBJ}XENw0_1^`y+eY{#3YZObgjuWajb#)t1fI?nPD{l_)jj%OqboaBT@4t>-4Y+aJ6ZLy?=S*|4Fi1; z`0z@TEnfp#C6J;e9thym3@X-Fb8AA~P-6<{n{tGE|G>Z@Bp&zW*45M;^tD{|&5~J& z8&|1A16ixibl^ZQo7wmB0n^@8@XMBt%3|@+e-KVH*|h6yHq*6d1Zeuk^Fc65H6GOk ziBd!3<28tLt2h8a34%yb9J?Hm+^H;)2QmS3wLSf+om!HJzwVU5_3`tlMskBwq`n-T zu&QUdCZt;8_vatoJP#cN05+DdF1KH^O*MY)tcX<~diq`4m0kmikC!1P0qvQeOzVq#bpzy3kh$ z_)m{WX4E{IZ#iah!GZ5S8U$OTeO%3&$rlMFIH}*^W@HLF0u;LI!;T(ZngY zv>;uK5u@tIpj#ksli^331L(<@;IgAAdTDgmgMD0FDFu z1Rz0|p&2Gdt)T1pe!yFUvAYKEI_`J2D;Tmor3bl1tHO1w%+>2r%Rtp&oGW(tie7UHFr z=k%u11UcaT;4WotIb=#M-E#2hS&B)|Hw16OmuYQ4zKPuW6gJBe@jea74%j=OEy1 zz5xBjh?=!U@6O=YRIo%`C;ZkPX98nSV?O={3)HID6m4TryGfN+=D)F&MgK5~_*WkB z*ttgV!v-_d-Bw(akO-9_mZk&3!?mKGMY^9lmPju;$8L>eTYb%3di3EB{NQt^V4JjJ z4EJD+NYXKIBZT+nI(ZTC=H`0#YGLo1JVD3f71SfvA5Ul!Ix9iJO#Zl8D&s@SSd<-d zbk%1}HPsCWk6J}aM?ei%R8%aTLgP$&P+lYxw`)Mtt%EwSR1LD;hvoW$kc*Z733_YT zMQexf7w+_PxM{rX-tO)yNSo7~?rKngW ztFI?fybA^MzC2E*nA1_$!>f5Pz4Cmmk@VLfaY|h3rs`!!*Q*A(jFtppUzW6Gy~yww zL2h9eHx%wT8bQ+-552F5%WhPz2WIOf**ld49NAJV6E=Er4C#+4{qN|0zY9LkC%e2H zP=Q$|?!RUo*qc)xjpyGmOlwbZ%{D|7ikz-nO0R29P6corO+JIBWCX9B2WC|IB-a>G z57wWCx0ED&u*4}uFtoGNC!?A&`*WN|XnhS!_3HO@L|f$V>*)e@hgd`{TOcc5M}LSG z>ZqfFGMVf#ApjHnG0JP+GJV%Exd_TBg7e>$ zGH~iBkN_iAhtT9uG(XyIgN;3{kBbX|p6K{%ef#HCc0(;X&qt90PsR1k%z{1lnp}qs zvxn)xP)gVxn~_6y87fqO1G**HdN#}lfN|cpF_4gy1H$6$rVZaoNC3C)LMwC=74z)< z0~r_;AQ@-YaE@^hc?L!G0JX&$jJ74TUoG19gz8XWV^+BVJSCc9w&#KQHiD_(>|g5U`R6_?n0 zGn_8^{;N#qKUN$y6kXaQYb-`*G5J$^|olv`2*v*E>TZ&mBEg z(Qj53P24tr1}vRs6X|~ojvOAA%itz#8{2a6^`!17EIe*YJo;EB6`G6K1H|TR#F{(s z8yF3wQ6Z1q0X3e=DoLJhaN(5Fr{h{Z(ypy8`7&8Nme89PBLi3D+sGIF7vAqOaY)&v zffdZrtYLFz5Dfqv0|J0m8lK*H9M9cGw~G}f89;#W|V`^^h_5~LM;~y3cD)*pwboZ zw`#7kG`n1R7Cj?F4ZqkAFY;j#pEpc6U^>h4et9Tp>eZPMd^}6T4j~=({NlAd0y~;gqtH&FqwB@^8+qYTD8ax>4j`X4SP zlwI}Q?XUosE`H*6di59;`8msf_JRrD1iQcSO3fuPqUrOwrBxYH!k;;Q{hs>kIK=>U z^uw_i|Dhc5RVqJe)?&oa!HHX^fQft84~~@jp$4zGTKq7LWQI;hy>W+|_sU9Hnb_cS z=j*h(K`*7rD6k^%#2VV~UUan}`x~3ZS)cwn>|PA|e6~hd*t14H5gS#whXbU3cKB z0xhz zAd;~c3bX=$4PN>g3ipEBa(s6cJ{aDC7O0Z3{(jx>IFNWZr==siO+{iRu2VP^_;cu{ z;G4=>zNcOz^92`SZrj1<#(styL`I+mU49HI54Ot-zhoIihpD9C4Yh@KqOdMbtLP)J z5h5Qqq$arEar7B*($(Xxo}uVWz=K$s_65e?lWBMhx>!9_E?Y2`nW)doGCIKX3HpGp)Ja)jmd-*#@2}QQ5tdc}+5>)8LP0gvjm$msw5{kg(q84Z6ynz2+yg6; zNv>Apkm%#j)`BQWjr32UF>9lPT0xDwIgg`#PCv1#neNY6dCKlP_*J-Vq*OMtj2QYC zW6u?I>qQgMOyFo=Pqe^%HvvbVYi%axl7F~5wK!1I(q)%DOSs~m2?91@W1Yrts-m@q zPu|!IwgA@40iF|3o$CZ5lr#f&8@a6x6opEEa0G;PA9B5Nw)kBdJ9h=5GdAh-S2M`5GDIRgxNS=rf!qP<8F zyArqZL6NYvoG2zZOPtgJNEt~?zd6bSJDQhyz#dk8A_0Jp-gltvcTnzOJcCv(f$JGW z>YjaJs>z(-h zBC%$73RQVZn&$66mhH1vJq9WtbHkm<%3e%t!$?`Z!IvMt0*RKVpj$3eVw0c^)*s<% zll-z!MwUrugsp*i$`J|LeLZY!Ls>b5I#-%CB8echFoSc<(CNs~sXBX;Z(@pPKqf^4 z^_p38vTun-EFv1mvgV_5oSrl-ut|ZX4@qG$FydCJjsxJ4k_m-8Qu3tT`L@&$?U@Zc zQZt1nTz>nz1CGba}jvJ>kf#~n_O(A5*21nMr-S4o|O8JhDeR; zOQ!tegT0^Po_5MtF4lDZ=G8;%mg1^?c%d>gXiKXo$S{$_hk?W|tT-NHxk#cqYLjoJ zh0}il@-WLyF|lUy*4>#cmW-yGd%!o%#W^73;6DxLQQ-n+DRt+&2#&TLY)qtp=aX5t(0PQYltRNZ%N;Wl13bdZhFAQCCKRP`{YPK9}3ZKu! zu^dA=w?9Fg-l{yXHl2e99=c^ye2fUwIeGzd2yX&`=CjlDgKHTK3$R`1oe38~IPw{g zkScTR%v+)B5!!sb0QBgmno|j3gKNtmH~1+eLb*>iE_6bP37OsJ*FHA&D+JKK^P*Jw%hq3PtGxwceN_(a#a43AGtGAHF~|qNcgWv>xUwIe z&k6$N)<3+-6ukz!wM_I-90e!_WOD2b*uVSSpn$XF=Eg=O2wRWCy6K`K|6U(Pg9@Sg zai}ury~)C2q59m6$do3{@|}&wkN5zC_4Q#oz#1RL=dN_$4gEM{UrqmtWW!0LbD2;i zSOfGkH)a^%bH~IW9zMo8VTH*b+t!=gGa$q2x{BgKCz4UZPdE@>n?)aIk@>-s!FRA$ zQWkrKK|gXyV;a=jm3MFi=Y;c@fvbCfjNi4+lM8P^3)OrgT)QB;mmIA`rA!Nzy|tS6 zcQ~iyg!GOZD!Vc5;ZBdKy_9I8RB(COD^)c1p#Ww|xU+lXKCf19VM)I#KW|NOP5sA| zcbrze9FRRhT=546uqVLmA??7V3;cmb!u0Gyb#IA8wdynlK>DH#+;g}6GS99U{*z6a2; z0JT|jfINdbiG0f;gHUF6+pj0%I^{mf?8n$Tk3;LpdT@I z*S2&Y4%6v?-d?D12Th@^-RZomBdE{7ftqXn1UqO0be@%c%*%l!Hh;iW;(JMnb$53S zLCa#x`sI$lPUkLIANgYe1jQqYoTcjoO4pXGpTB8QVK|H!K~JW5VC89vR1?Ubd$WRO zRO%(1ec|PgP4V36Ojsy|H=I`}L@?o(WKMaW;deQK$cYc>MJevBcf-5ZPaaZ>#iXDT z^tq=6aq_)&(fhPLL?`z`0aM?jHe&NMyN+vlYgfe9YHD4^)n#b5B6>yb*i>1-1E3+2 z&F}AnPuiM6UaS#~1nXLEMXVzajCysB1Gz^24#OX{+6m8MuXS-GpwOa`yN6i)yfIST z!;4iF&C%oOBpZ@a2Jh)hz(8;RS{!SBB$8O9P7EZk#^f0bT`Y!Z ze5*Ev{f>N6N7D+2k4 z5!!E+)H(d^i2?5=D&+|Wp3t3Yohi0| zBX>~1-U&!b0YBOsnZ-paU1*G!^33?az(5H?yz$9{r^s>UUefwc{{HiHCOTp3xrlPy zRSr)XqaQo~9X1QRR#P0Mu`>Mu>^M*Oyl}Yv)ppo*6AI3;p;(T_ zC%Q+rB;uGkac4Szrn#mwAA=$&kqStePNih^<_KPF3%cOpdB2uC5+>4_vv3STY*(|n z6-BafLx0@l=M*!7qB#EfTaCENyrGsRlCoH_sjv>Ym*u09#T|^uFY;MnZErF^n#&Qc z)t2~ZxD-y40cjsK|7#fwt4~%|n3I+JR_v3JS_-V2S^kYIzX+{o+d*6Rt(IPw^tYB4 ze-%fXy7y%z2ADN**IG2|<`TTzQO^Y=gIcW7PovlV8y?s`oWfs*Wn#-ptSI5tsYYF- zrVDB92gN|qNCk=pE62YJ_M%To8drs8_zG6iIwLb)<90NR$-ApzZc2U9hg)y?h_Oe9 z?TEUoLh`y`lGyACX}^Bwx-YGnx-A{hLe@ArPfrq%zTyaCcNcyE6uca#&NeGv@)MBn zJ6t3m^vy%AUX%na2Q(%oCXoL^=(^tB)pZ6IJh0dRlZ-Qku7+fJ`4b9q@>Kw(Lk$+# z-1#C#?Wq}qpxwSe6qpu7L-Al=@9QaW#;f>n%2tU`JxS_n|$u_ARTYw;^?BQ^F zHbg9(KH(dSTELxN5blgDY7kv*bKWc@Nu0v6p%Hb_3*P?43hEUbE>(p-b|!&A(XcVZ z)^6M6XvpU5X29}h%X$VKiTb;~ATfX|eAz@QPC5rSl#2eSl{NA3m$ACX&lwaer=Rx( zYN3ds7V5M0Z42eq$N%e_Y@F3^Qg%@bujnzbmkWlb#X$J$$O_{rKGl<)(t5n?O!0wX z{J2_%pkuICQIUt8f_Q%TPSEzEmFh^E@eHY9mlmpi0}szjO1OkBiG;dnA|8m3m`V%C zB{nOJEJZ;*PclPw!o_k@wMKOmA3u!IY+c%SIi7a1bo}Kw-&xWhx9RSklr>7ao69>F zvE@!(_1ffXd&lVV_Y(E4z`XN17T*Z|o1LOD@iuS`UEPBM0^FBL7T<%IhnygmThhs} z`QP=Pst4*N1TNqLH;K;S*Mxl#uiotT*~Z5w4H`rO=^1OQmg#hbrI-MF3||~9r2>FM z#rmmAQqY<^sgPG9hy}519O2Upj?kE?)S@?*i1`@hNC50-{%stL zGgP1XAYpnx6rXLrr2jnz3Mz%5*x5Q|ic=u`7w&2YU_)S1`m%B{UVo`Q3^@n~er8Yz zzM#$SypTdBj@zS2Y;$<}kR#`wPq*q9hxBSOIP|q~GCfs(TU6G~GZi&F9a)%wsBYKT z`Bl^@L2zS1-hz9XYW3@cQF#)r=Jc_}Uc0k)WraOk1whu+Q)He%)>)w*xZx?n7V2xD z@>KALtIbCTIA%eGHs_?F?6cmQ`uOs z;guvms3j^v*PHjyjL{L5-7qg0woJXF&NwKd=&UsUJOeyUVslidCsu%aj5O2t{E5cBY&^HIx;CQIbMXk~Z#Fx`a;MN|VU{AdNSO z`K!O(Z&tMTJP3g|?o};U3oxufVl;T7!Id7M4%OYD+~xgJ4+IRnzDZ zVpbBA(HlHp-SVY$ZXA?_g6p4!X6>qR!sGHScEzdJ7pf z{aKkX#3zTdu}~WLhBRF>vQ(G=P|NP6OGZfDx$rV|%YTYUSs8eKo>*;kIBOTz6Kw#} ztm8J@Ofm*WO6l`|NHbudrvjweMeg+U(}D^BH$Uv{qCE%&_ptcr>&+}M;$VFdkzc5? zftV=o2jsj+Qe*dhThmi90-Y+W@QGLq{YW-L*=Xlj9x!5)74zOX`%TowSWQ(Y;AoFJ z8a#r#{k?TDKprQ4<-EfJJ$_;zry+tX?0m^GrB2V^aP$F@MgUDyM2jZNLuFTSV@`#ygNbb(Feq{Ohan>z9>E zkAEiNyqPgZH6jq@n+<ninWD4S+jFgn^=e z{MqkvKnrHCNk-vBN5~_1U&gU=QW3IQ?{w)d`fsgh+*Oj7F=0{rzeSeo$@OAf`S>LPUy%sxGA-=Xg8#}CiG#J3@xYI)z#7cz{VOH@Z*PqpOF-?6j zo?uBWyT<)s?Yx^#nP6Z{DRs4hJZTF=2S%N5JAojyt|UV@z0ba6SKQ54WxA9fb-STB zg$VA29;i$gq2#5dx%K`{fax?PN;F3LH55$`MNN3>6FRs^<9my#P;M<~tbv*dp#b-ztOOd}hK$fBbHwy*$u#X(R)--- zx=>?3E^-{SL^OFuwEbsObK7(L-?IXjLu_K!TJ#j!N;WV)#O^kAZ( zThIr*Pl2qbHo?V72`x*eEe|8jxuoo9e@7iN8+Fv|qY-<#-) zuN~n2RLq8(=E0IBkze+%6De-~yt`d|138gd+LJvpfp7oVDf^;tRLH*1LZ4UDJY2b<1Za;HY=XEiMqK?- z-`Vvu78K}Lvd7*A3WPv`0{a|EET3BHWA!|70NkrIuO?JQHsz$=TMy7Z@G$Gm1!kE8 zS(V#^$}$5$hGDav=y|%9)mvyH*iz$u-Ue$Nmhoy#eSQ-9jeXF`g({?>X`I!YCy?X=(g|vUMgMIbvtLdt+6mM(LQNt_)hhPmgmMJp4=7A6renNFz0- z;cU|~xyO*YGQ1phuuesL4niS8#fk&?I7dkFhI$yLl77&8gRybU?PCM|>zv0{NwN=! zmxopmf}Lj|lcrTK8e|^vkkWuYS5@^L-BXd#iBjU{Vw?)f_$`aNm+{5iDEkbG>>@Jd zvd^OrYwXxndmr=|znPhj0h%x!)%aRo=@zcD_N`Up2o*Mk;$ndfmX^r9gsd8zn2=Rt z*5lTy@!K=|E9V<&hiy{uEi3Mj0o}z6!G8lHXTCg+x8WGKHMd08w1M`@dUr>H+VS7- z0M>303;A6YRiA5}iQ{~BzhFh3BGD#L&E>N{S+`47erROj=GigdXNmh*H*#66%vTq? z%9Ulr<3EE5-z<#_^f+5BaVVJ|!$o+i;=u&2@U8UssQC93rE4)gGOuGvO}GT*x};d` zT>J;qFP}h0Q**QYH4qd@14iOhU=vh}#PjFeOkq2kB_x^9Lo3w9*HV8hqLiG&e@aqd z0u9PM@&c*EI=?++lN~MIH|vmCZ(wHXbN#SOYTg?%!l`5lCI%msh3|GVzkBX@96ez2 z_a3ZBiD*jPL)5<^FzgO6$h^nFG?X9w0|ue(+0?t9sM$>sfBbarJ@R;&Qm^}sIk}R! zluyMfaJ&93JX3#LjkyP*PEcx`X~ySacwfTwk%y)qG|QU;2p;oh-t^ z*QF0nRXcYTfdk%?BJ3%F2#;Lm#{+GklMbcGE$;&2&w zG_jz(hXF3b<8H@dFVxWiunqz!0-2=UGW z9u)~neWoJ8aXhea#95zq_S68lj$aQj19&d9W(%{EzgMs)UKnV z+Jb(oa&1jP2k(+pQo#YP_L5S@(+7L9SGmFkLeicOjlC@|AryRW9ekPwzR+_mVJ*mC ziUcBZTRYxQ@5A`>l57)SC4__MD*gsWsEzN6^k=Pt8+A*9ES18nY??xpqR8)I%Cpw#}aHI*LQVL-THr=OYO9`3XRflF3&pxG^oOA`Q+P zx357_Z+-^}9Z`CCdb)#hAIcEu0N9`=Q2%e60PW{cl&89qEbJY5}wT z4hS*wSUg0h6`!;f#}2^D;>Y_{W>hTQA4IXk=Ip3VI z=A1k-$?A82$Al=A?98KKXv?A?PTz+_u^&TSKUldTl0c+5PlI_UOym3iOnTALSFad> z5)Q8?a%pzReU@%;bM@-zgTJsQcjk$XSA_K)HOJSW;(ITT2{g?gwE?DtB8hK7DFM#7 zq3c%kmX8&l1QrI&SjGe-$q4{0amrJ(;Vx9&?e|Ux&w`#G8uiY|{_((;69U#DA>tzw zVo=rBI0UlNzMosTf&Q$jA9pnEqnZ?dWNDSXP(A-D4hFzjrtvqV zg;h26CPTyM~b3T`EMyH5YAm136ZTHNQsrh+2=9+Mp_RP9{%MS)q6FNYC=2}n>uysX5E}K+gP|8e-7ZtUygBj#(fJUXQqG#bXi|WoYo~MTW zL6IN6{rK0xeVq>*Ee42lZ)dEtO=3I2AQ`XF?J+tz3v&Md*pX$~eRu6Sd{LF~X;Xoc zC@awY7Mgy#;F)c`J&mK?ercipU&XJ~VO2yL8G!C|&b!^#da1+nCfa_q1Fo=iLT&Eq zqmEjoW1Cx0OequlgLfdM<|8V*h_{gcGu9Wz?m`N$Dt&9yc7U7E%vJ|WNj#HsOhS~- zby@s(3W(c~5#W><34fd{vrn;j!Mu1AwYLw7=SwZHXYS{=G0QSAa^{qjl>8UirE39D z+#xXPv2k*WGZKI+o|u@J2XOiz*f_hu{MhD2jZCQrI%WJT!J%V;GRAS1FjwonfGS6W zMY@vcSAOsrka|G==SLj5aKQ^Q3Uy)wDoHeNIu1nE-}gqj^!R{jJ%1ib|KifMaX>P+CynGeKTkO4>v~*} zdGX2utpwo@a`zTJ$>{+vvK&!hofKK z!Xj}SeoNLTJ^F>zI?^NatJM)H5zKGaX!aA?u3oC(Jzu_fiI6$8{)GQ@|B&|G&-p3c zvVOzQ;jspM{8I3w=PT59>Spgs8H6^EX_BtMm*G$;zeX$enIq^T;2HG9Nfw|bza|s= zGLHf>a7`JXIv4j3_Sk4n%m}okAsvi465eERq)0l~a^akBqMdJU*!(1#eI4@hU7}<} zG@zAalLXxUQ3;&=vKKmQ#Ti*#9K_JszjtfE+@7kC!qG;h4da0#BSfjD zMgfb_RmG^)Co;@&WOiRdl@=5N>Nfd>#&K%WyQBR%U=$byk8*VA&uTqMZ6X(4Zqmbyrhfqv<*&d4(ng60tQ1&mRvPDD?#$N{ zf2U9+WsAFLWr;LktqN^}HLPZr+}yD~eq<+@P=AEN-W#K3>FAjDzr#f9>tVNHq5vtT zzYP;TZR?LW==<$F53PZoG6#l~s}tLQT+qU_PsbSd-P5{(X~7+OU_~t#3^$YiE_82K zf!bg?clRzN<7Uy_Mi}*gkmf3Nuu~f3f>OPUv|As_l|Q;WjpCpZ-9RU>8CDBOR_ z$RhoNTbI+i(UE@jiDj|Ks+5v)=j6oavXTb?Fs3mCd3qHXaqMery(xlGo2NkEp>`{e z61Gk7XYXtVKJJM#>S^Q`D0Oe`N#~th(`h;%3J*uT)?K;qTtV`ssqMgm7a?#dp?X~R z$}?-fpXLU2j0^Dv4Ut)e8Fcv#3Yo7%n!ejLOTFz~`*t|ac2HjNjD7I6K9)hWWz+n+ zCDo>-xMHn)xH47IaCj_uTmvhjiP)v4*$hH2G^iAG8GJZ{3kzq3{+Tw?fD|l870l88 zjBkOgHLS;eS})r~;9{aN^i@zk^d``T1 z=Q%9(WjwI~7j!PNa|-ow?_J?EANTvx;vPAkuOuMB`sUA&1PKy6(W7)AptNQkVo!jn zREO&tD2Fduybu_5g8u*sm;i0K0pVIsieye0+~eS}Ad_3plfA$QX}K!>+UlUe_ZF+U z*KH3>t93tDn4pDex+lme?`=aU=LJN$tD+j-{CV`|547$tO6YDKF5SYIVbKS1N%|1!TTp0cpqPSt0R6p}_-qggb^R%{x zQ6AY$~DVmMBa)il7Wwx6B@% z43s}Eg~#)Zf*jE;?(T8mR6x9FPOyAjtpysgtMeg?t!;u6Zrl^@p;@J{9_Pxo|23OB7@R~ifCb<4 zTH}|beiD{jRQP-EW_QoV^9JHDfdczHPU>!0ow}uRuU3Qnsyn$Z2;Y#*`8a3aO~%=@ z1C{jmWJllA#NI}NqM`s%$BL_ub5g4vV89Nau#_J$`tyT)#V7ztA-ORzF^kVH3kwgy z7dQq&5N@Nb!9y2y*ZGEP3?&@vNus0XF!ofL-T|nCR z@%6jv{`2*jg@4Ily;e&$mhU@FaHR^x6{S2&ZT&xX3)#W@xX=Pv9n8y#@+xZI_B*PQ>juHSWFRDK94nACQUJEjD}{!^RoHDJt)mj6zK-<_4&Em!35 z_VtwcSRVX~TkqD3^^s^n-`MiM1`CgxT*JSl3PJq@%HJSPa5(>LV0hN>*X!F0Upzgc z5+~N+F4I~o{jRycK6KlQ7)K=sm3EX|3Xf?3TLh)=4hX?G)N(p1DIn9J zL3&sK4C+ML#C^rVrRWWqs4kdbJ9JVy)eN~L*A!IhJT7~L*4}8RTLYk%1(N%^7cnh6 zkRGLYczsLY9q!pfargVDv+TOd{W=6r*RdD-zc}8VbbBg-D0Qz}kucxB3z*DyE@yl5 z_!pGoIve&S?yZ&)O$Fs{XXmjeNW*^m5EvLpAHoVPPk`+8rTMs(vUI@M$Uk;CtJmji zjhBdxDF+gXNUtGX(#L?sY-V9m!n4w3E0T+$q34QSgOQ;dS!9{P>l^}`VFHZ@fgaS6 zu=th`bhc9oZ7}QsfHuQI017)hvv+|c?OoXbZI3)ikZ3GD&wm?BB zm;mYR-aeH>4Jee`jwFtn==3@0wRmcOW`)P|a)$1Va^GPvkUDY+2{8TIgeq={}+Kjt%gE=_4K>Q5NvG(FB!(n>wn4>+9dY zi66NCAQ3{%`qE6QG+(PQzBUXgxpBupb$i1`u+7szR(2)G48hQAG+ti%AJBk(0Z<7E z^lLL_TdOIRTyQJz+7~1bvmA5O&LE!u{}Cs70A>Jo0mCdGu<0;Z=GDHs-|mC0^!QI< zX@zx`|Ni~^aTd(HJm+)TcKQJ(E3L%uyA+6lv6SPIe*_^jwR{N)fd?fsuk#Ly`>H23 zU>Ze!O$qK_Xd2~TJxA*RUw1p}Z;538@o|8j2WU-Q^q8wCvVpt^7W(kTl^F>aE*(bB zft+^SD*|g{-wX$8zoP@8d2c^(FK_Yy!U@c)3X);QKp+8PC^wWqBFqV7I5rmJ4d{Bh zsyHJ(zXorCMCt0q1gF?aK+^U$&|7t&m?Qy#;71Ucb26sOtlSQx?5=p27fF6w0)|0F>^u!5^F zGMF4z=eMg>-My~Ws|&}>OY;0fio;JsGO+x!Lsc>cV~YG47*?hpQGj{!w45=m7Mod| zlrf?nd ziahz-o|@n&38P4JHt&5R139+F;wMBJH)8GfM$gAxhM=X$<573}dFwy>!h8qJIr%V^ zq1pdZ8IEEK33X$}2n_1z&7y%kp>S&ya?baiyV<>?K(c(o|`7p7yx^* zjrS@n#u7&modM0Lsk!;$KddaE9qXK*gS7G+;3541AT=g_ett?cz{deB6Q{8MJgp** z7Chf;UNtpVxp+Xp`UG}>>!6@!fM?@o%ZBO0Hc6gZZrg2_fNa%viICSR2t?Nkt;6c= z57{yOB>ovd1mEog)*L;qD@HV!zrg~F0Ok&YadsiOLm47R31D!MDl1;rH1{lfJQShD z6me3m3xs@B>mD1~m*~pzs|~jo+u?V$2Mdk68H88SLsC5;rn+i32W!5YKoL2y^eLwq z$jw1529_0@Bcq!|%Jbm3>9G5X@b}!c;b?taws7EH>-YaqvB46#o1CX~MugvKNf;HL zg>Uiv$|YCgfKG6EQ3*Lx^E@vLeEpvv+P#sc7e6|a?D?>R{a6nv909v{^y?{k<|b>* zW-3WdM>ct+NCqu>xeBQuwqikFPqtKI<_-fxP>%EXMAQxNi zV_XWWx>x3W#xwc*Z#tj;eY#*oH6{ieP?&-Ws70&!(B-B*fJbHs)ZXP0ek)`cZtV?)ok`L(pO0Y8<(G+jO8O z{g289GotqPHvqdS;4Ni${w?p^fRPAT_ssC%K~Tv95F&4b=AbZ$^;0i{t(IMy*((Wh zE;<65!QJfieE1vjOr{Rp%5kW=oh77qI-bN#BvPSclH1&jBF(oH&7 zWM^)_Og^$R62~96>X6O6Kc%rqivPmH4rf}9y;zpr68&x6fhC8wfwzp)3e7oJ#I! zFvF{#p-k58f`wTU?45fc_f&&R1vNegi0`<(bUF8iy23cO`S~w2pWfGD%=rL| z&7fI5V7S^YwNxShWNcopqxq(F06RlG4@!$~feksdtgQFNG6J#37Z{C1G!#urO+<+m zpkJ?VeLg%RZ0>rYB>~3vWz<*FlJtLu6dRY9juUc4%xWW#xyxoMq>5B5_p_C0=zuXzE~+P_BA=>V|Xuh@rM3kFexj0d({Y@ z((hI#Df8&k5gL5%@Zj+6lo5V*6!Yw^>Yy+JMvlLWrPJ>ep|6S4?c4x_TE#-b4mj^z zy^w6U?2F=&E+VH^!GlgdC?8RN%E96pyV*2XdTt9% z{Xf#ThQ4Ebg+L%i@9S%f4)^wQ0B&xwp%DcY^&AW{;UI}NxyN@f9VK9z#cf)N5l&^1f&B^H>FIs3?5Wg1UJ&_|fwlO>Zw0px6s$}yQD zQ8WKZdTP+~KXs+1=U1G~t{V574`z_`5EAxfrS<*ax=rmvC!n=Zu-2Max(<6?{fpF6nF zM9Op*k9~Yo&?T0EVH#OQ?(=IhKW0T0-mN`zJXN;RD-P=wLUu>LLiuNtSCnC@XK` z9IvfqmCxjBtl*>Cx()zIv+lN*9sU6E220cHu=y)^vrG=#%^XWn#lt`$+EQHH-1aoU zg9L$jkW`B0pF?s?*U2uolMTjz=d9I2TiCDzyb|~T41_B5cQPhI=pE2nUhYr-0DzE1 z#aFzfZm@(C!QgnwSw&K37`l{tcZ=|t&W&3$XU99K%~Uf9=cHt9)d$B`e3-SySCZ4^ zE5iM2;-h{}Mz%omlub;CgBe{gW;*HB1@*)M`R%;} z=-y5dBX^`D0R0W#$!F8`3|QF#t22@{<(QSUbfhChYAodIYa}m7&mt)UeibKo7D`&IjYENk?-NnSSk_P#EmhG)nS+kDjB?6*Fyt%(jT@F zc{yM?6S%?w-v$DkJgp;4+#2N)`PwU(v(L4oDjYp4C9zbX~URJ3DH ziGp<#Ric6FIZHIUtPD9|rA}jD*BUyW`Oo z^#!Y!R|qTw@8|@N4pb{1n=`kaLN$WkF^KN{Pe}#bgzIlIOv>gXmg|XTK0eED$2PFX z^%7&MJ4uV{H7=duA$6GxLN9%kp!R~mDKK#!wM{G2ex28+EW7C= zx@`cl4b%)I5ssyiZc?$fqu_JD{t_8kW)XDdMLWX>E*x9%1DSvJabWz?W-eh3WMh_B4 zlWawidKa3k3O}0t0#4*s+tK?%W?x!#!OMv}%R*bIbudf_()->lH^X87@@g>{mnj4d z9UZsRJ!AQ~p1nP@258Ll)^w!w^j7kMd<+Re0)aKNWEdN)|1Cb_UTPF5Y9L#y=U~iU zqcaYc{k)`xid-Lu+_GKxe_aXB?B*!gS%P?38)-698G_jFy%C9zk{x!4H?(k4fw<~L zw`RU^ORDULAj?f&G;KY+{Cpz7_582e`LsZ5g=)hcDZ1>9s~0?fVf7iC+-kh230}=3 zKVbeH&e+Ata)e#GGihtt^wDX~!dk##t88k%tZAsy|WGge2H| z&qF2Wp@<**vdlk#b1@>`b*DCi!_qora7zWx>IPE;R1jwXhN732RiMt);2*h@FW8wA8(D{ZAp83`>l8I zT4q`r-Iz+4G}zc#!m3m+t6#MBAN{0e<8~M>l?%*Ahy>f3Z6wruUIOVuk|CvSpIskw6-at^4e6J58eaL>^9{nW?nQOr=2{Q7M);gpjvrBc#rJ@-Y5zPT-vsjII4x zmw&52I4Ga3FeeO-@E*S!<@fb~N;rJME*B5hG+54pg<3g)k8FgyQrQalcaa9T*WqB= z(9Ah5{heM$Buw|q>j*>Hd0oz*`}>waApV>IO@WJxYie<^wSvg2ROE|u;mb$qckrZc zDwe^r(;SX^?z5&^s-c;Hlx!oa&%@dTPr2p<|Jra4e3#M~BwE~+0%YyISzC=#ROAm% zkvK7ELfA81LSO^laM}zXp6RpKO0c9^egWGoavxa-L$c$ zvm+Lkz_iKV01~4a?hX6nE(?n|*UEWZ^(R&J{#~^r$6z{(yXtnIPh8SpQ%w(l*RC&M zl_{uW+7G(xGs!mgDjk$%7b!B!U2e2;y>^Z$;FX(`GKwWTEC@3fS5(XVJ}KGWnQ4*p zBAb}0+xpFHfdfu$YtqeVY~}1VBdR%51JjMVPasGDNsMC6B%sJ5=yjQD$qL?t0Rskb zNeTEQeQRuG?x%z4ewLLDJ+{DxUkYHvYdYPef^|G8LDE}SVWe4ixmI4)YQ-=5O1)|| zRg$7faBYPOR(sSk3R}7kgW#jj$LgEGYYG%^1#W)__F~(T&U^lPgDY#o?~&S^;ZW^xV%6|hc-HxZBQ7g zpdd^Zo73TEt%pqP@jF1x-wi}H*jZ*3lF?&*DDAZ?f*p;D|EUtmGUaj6yFkiTcXu}k zZ>|2@(j~n4^XIctRU|!Vq~VAKhqjGIN9>C>ZQX!~c6)FGlKx~UQ(3ofN-P7O>Q7A6 zYVNA9kTbC(){_cNJqMu>Q|H6sZ+SG3?FPaZIvmCYN581{a^L^`6#Cl01&=>f8*9u( zg%+H3J265Lt5ae%)yY4>BWzVH3~hHVv*6r=FzxJt4S67JUNflLfeq5#mP(!riR&PD zWSvAbq~YqQzF1m(s5CKc|H9l8qP9oAu}GDAZdh0QIzzP%uj{gu)MZwIVqTnLKA;L- z5o=)aVx&+!bzFA4>|umTvd1`wZV%{EbokDjB@>Nm4F;}7j^M^ce~GHiE=OI`s~OWn zNU=3%=-kQElSi2c!7=@;_UqJ5UWTe#;nqw6l-fN7xz9sB!pT&7(B zaArcAvEO-cSx-^p3z(&X;E5+n0Tk@Q>+e4*(N*}SC0k| zS#W3qf8=^UaxEsvE}!0a2^I&p93BeXJM>8@`acA9!@D({j2g!cgp9*7F|&mtJFEGj zqIs3MUG}F5Qm-lm*JtrzbtI2Ce|d*6iKvcTCS)Q&W_l!Ok?p0 zN%=vs!WNP~2(%wW2$f7hlXKzK<;a|bVoIiz`olwY`m3*dd#$3*)%GWX=TV>|al|BD z-{gQZvLqs+2=8_lqZQ}<+oK%P89BWg>`#*pgz*);p=AdgpU@_;c!M^z!81*fiDJo4 z?QxXL5R02x-lP~83?PpFZEx4VHWyxYvRo?Dyv*Ga=4%y5e^b~-vPwlv-=m)BLP{63 zg6sN1Yrmy#UOHJWl%1P{0;f;4-HBT{=~S(n!9om?0FvG#%P-fJ@!qs|aPdM@KCOYX z-{pWX?SOp4y70qq34P<0x?W{fhTZ55ffsePGID;}?s^ejs`2{2r zKN!1u?Q|CmYISGeeK`TIT>u;ws}}+;dc~EMs3q5!E=J5cr2qnMm#c^<@T&hxUGE8t zNx-IHB@~pOoed9lDpku0hDt8Y_yr6*0~8+>6E7W7T4kc2eWuvf2uT~;M< zcFE`oL}w0S|3Av`MtCef&eM%qD(*y5c9xWKdA<(;=4rhcR9T_w2BDKn!JTa{W_PHW3_}oTi$0kST7_xOYZIR3OFbbIRtxB%EQMA`t znP@)Y|0=)VGwKcDEXFh2CAQT%fV`o&%IH!8d#1}IKb@2s?`(x^;sUMH;#5Bys1Yb- zry>mbq#UMF+~QI%G19Cst~r%X?ZN3Co_idyIe4sGvf=D@m!o@y=5TR6A+sP?WXzU{ z_dbaLw1=4>Y`rBbs_M=Xv2j`#8mM64^THljIt+b(^zk)Z?)HA(K??ASS?2xp+GGv~ zB2Ndy*damEm0`jL*oK>1uzGM%3Kp5YaZ|Wdva@3b1Ns%RCZ&O#N-jCra4roCEffL< z1h)Xwrtu#7p|_3!4WT&FHg*>+Ls|t6%a2T`Mnrqn4~dfG!{Ex;{ZyyGK^HkJs0Q z8J6%%o}~C+a_I#K*;Kq9@PeA5)uo9bVX$mmqfdjr=Y&28eABdoS>}fh_jP$N6%%~O z;q5aW%>4X|I8yd)nH+XzKYpy7y0>Yyf^FhZ-2kve$~wZ+xmA;Q+C_Q9v=1-!o>w?O zz{{WiDqzg0ar9p)UvhD~=RpzQ+M(4}3(oX~C>Rk)nG`t)Aotfh>LJTYpSN)noVW-Q-%kNV z>lSc)!Pd`h{inT$M+z(v=jVpH!Ma0=!Q-}p{$-mr7=r**pUtGRb(*+=!CJ^J;J|?Z z-N=sQk7ifMq20aJwo9Kz4OBHF;V5le+?-sD_}%Uao2sxuF3E0cWH4hhZ<|r@2wV1S zGSfW2Om&Q+yA}{u^GldH;8|P2H)DVRTS~_#$8XnMu<v0a~r)saGpKxFo zcdg2R+eJw5D|{TZtad;>O91a>jAh`C;N_eTBue3Bj2J)Wc5BlAx=HbB0sqsZ4)zTL}U zU*x^~5jse4V;356arzE=s%YztEuFaXGWpET*dhmRQvdFxY-1%JOnXw6PIYVNY7<{Z zkOBw_HQ!w@&t>w{LsWbOdz0O*!FXM_1=I92C@kQw@^3v5N(Ac=8<&D*n$P4-MBc>O znn9B1DHZHGc*f(jg$UJiCxT(UBM@AYOwacf{L{dp0Rx@!@$&;%1I63d#J-nGAFfIj z&L>3CbIT6$+_pmu&#ZwUAaHX3 zqs3lt0v}piyxCF;G*uQUko%ZT>;H7#KLhkR{1Y{v({BcU7X?Kt%a_t;dY@)BUB2I0$U*F9UJRh+sHhaAPLJ-+L; z59?dHmssVM(UBUwAT?kbNNkL!7 ze-`s(rPXRdUyT`h5mqb?rih=p1f!45fZjYpdVr)53^^z!RIkAbeMz5&tpjeF+yxaO%pXzP2@95`m-mN3SztC#)bZW@;_Mxv z$4w#i72n1F!i9CGxJRlA-buA*H3ijFKcs#y{OW-VZgucSsUH{J5^ zXV`tC5UJMi?LdVcQ0A4E(t^G|1>{j&J1^-!v^S99rsrBhrnk3;1$Q<+FSVbfEvZ&b ziOA0o=GQ*YXLV;CO$(35$NLpM`nyQk1f)cWc8{#wzwd~41Gbsh#CxwsOU_E5HNCx4 z^)_|guyr1*gt+5$su8`@xZ^|@sn1I0tykNG=&wJ~S~F;r2A-Bt%H7zN*E)x7xN&N$ zm|S(iZ1ZX1Tl4bw-+b*5KM>922=YGLU%xVMK?XzqBYV;#``qrf zjBohnT%04iR^KWTe{$N6oGY;ziiC*`VrnjFFxP1XG{3$@DK*vys`OJ&I?1PAj1vrj* z>+)oeNqWaSwMNz#l*mS;c-1sWAWsVHa&j#RK$H|%!A`>FB+e1J;?qdf%){(Lg_@4e>DX+uK4g|{?EqHGN|28irSxoMAHOb;T(cHmlFE5~ z?$B+7Vv`|mH`N$_v36}&;uh!5ML>}yRpSAkqFH?2GsDK!clQapV72BA`YXCT*3Zjf`CU1pWMWEW(%;@8=I4(MjGKI;ipQfK z#>N8N_H`N zL1`7=Ri(E==C=3h?nSQdC`VUneL^|@a+;j2Vb;(CO4{}>Ou66h_`hJUdg@4$w&uC& zl6j*~`eq79Pg(@|Hh8rtTN9aAUl1P#vAb(GI&(Mn@q7e<>^B%UB=1jzz8o%Yl}r7Z zHEDOEqAZgSko8QmJYIY4u_5-uSj-pG<9=%=vfciRI#n&H1}#z|R^%OKK_t8Cv5{a< zsTED(nRnN>m(dtkF~?Sj&g9lzsZj}j7ktZqB2uC^SwJlF!-K%HNsENZNhLtX159z{ z-ydP!&I>R!hn{znJl0&Tjk0V5g=vf--osy@`YWn_0xN12pe22@`-JFxR|n(byO?Bv z^xfY;Xr}M{a2`Z0kV`(iCKm_FrUt+)0#YVJSUy{AL&J;}sofG+YgIK*xQkvhHcV;M z%~E5V{&dy(2yGgAg^Tp+1BhdNVJ`nNQCZh@R*mSvMC9n3>cElUQWMyoF)%U3$x_v7 zB)98Q;-PSLEvxC`y690w_Kb1tQY_iBuDTStDtZr3>J0BcMQR%D3Wz#_51^CDGtPFW z?o-lW(e5Iv%4bmcLu1Zlqc{1T5Z7{PIm+GFN4Br*xQuuIs77ZN`24}T7OBJj>0-^v zq1PQIekM=&D{U_(5;t$0gPG#W`n|7~T1jT?+eD+?IUQ$RXki_XgsWqc_JJ1;B>A<8Ulw9hR`>8_1M0pU#+T(J`H`E z1XZU;`>JcCq6r&55cGpB3I@Pls`0YNN)Gr$fwPM>M<0O4&zhRsqind4Gk8hLXRUq! zaQOyMs66%i-`hV$$uPHLDdbCvSE_F=tvzzVyq-^>52`cr93Cfn@DqX>`1?xq-N9hk z*2L^6fw08@k=CQ#!CJm3shGllC=-&q0f}3f>Lp&5O2=?`CrFF^S)L_9YgN7_LGWTNhJygaDQ{kRo!|L@*AN+)P3?RHNSk^&MzLp&z^LSeCQ-VQ zE=5L0CLMm9LyjCLl2t%p1jGV^KOXUb%dZp7a?4theWI%G3o~}nCA_=97xWJVl?ime z7U&N5;$crM@^Ql9Zx5owazuVb*DwG5%4;0-*d!ly)VDut*f=;!L6G=*o z_EE~rHs`ELv{6^sST|T7FgH`W!~1HtPtf#GvO3UB?5@lG2wPQSR#d<%#lbDd$r+Qgk*YacKv=T4~)H7M{rNO7E`$NT9oNddO*#*b+ zptg<7P=e~?-W`v4tztdGPwsoI@)Pmz_oSXFHaTz6T_jj((aDFWsT66-up4;196rN5 zkH)x;20wEw1K+k11a%e*C)GH=NE6~ktaZ7k#BYhCF$y z+t8ni?{kWffTkQXr`dLUw2n$M}3EKkQR8#Ax zyFdM~V~mx9VZTIatdZVg`cD>4G23(_JSv_mjqrH(oN!iN{qV`3ze>uNXw2Fh5h+Gse zoRqM?SE{gNbf!D?YADr{E{2$T_i7_G?dMWO1|)rbMEtP);K;I!T3#tLwRdz^sVPdQ zGS}wa%5E<^ThPtVmu3~KAJlh}dXp}_wcYF2ukI##Ki8LY$xzp+Mshmjx;(r=S?S5Y zV_{?_eWIECNwNO(Wowgv(Q`rZFY9{{6p8V|orFQ01KUyL#9lA!LD{4(Wu#}t+C_0` z{oDey*#13CaWCH2`YwGau&6X^^JY%}8AE>?5*&T|)Y157YL`n6d zE0UX4;f!&VP4@0*s>f`Q1A$)cw1~{&odg z+RoI{@7GS^*6^jI{CvX>=n~?g=TR}BKyJMIR5o2dFN42q^oH6X=LbW^zR)g9)(cEM zWOKUyT&_;VCv(F$&8k+7hhrZ)xb)8&MqYlXz*4jAJ{ z2;lTr#xah4ryWr&Vff9%NvM|Z@F0vtP11!Teuf&gNtE{8^Yx@wd37dp`CA&X<|DJV zt*BK|>xVj5VL5x%p5>YmKlUF|r(LPiwcML2^q#ULqC2D^JVA$xCiE9rR&+0h2!1?DR9UH*)G9ek&n(-5ADhkE_`x)nLB_d{vHpdNjtsY zh?O@9ydk82XMjkq3rurm!OocrMXSpE>LNiITpSWYf0Z`QJtX?BV`F?wR{EbVKpX?? zt@zLQ9<>wV4x}G2ZAXDJZ3hVFVddwS;Y#n_JZAxCSuDB4El70n1gZqS>kcRP?(VNWihtPstW^EU}e*O;;?mi8~V;MR*Y=V*X3B8P14o;p!5;3{=J|1YY z8$wPP;{j3)Tu{qq-CUmstVGP>lCja;p+qml=_Vyx&;hP-aBOMbJ98$49QGid?p>j##fCg^qJa2+Z-cG`R z`cBHEU&LC(+wC8T|C*Ul;rbh;b`)Y6GAK0ke=0SqbIOV~Bi@JL zArbxDeSPIsqBptRS9!wMm}Z#<2$`&0Tp`L4g@u$jI5;WU+4k6WT2+cPXF%lh2mE8C zg+oqGUOBg8?#=Az*1kH21d3vQnAs}1^TCv zetNG$XJQwl9O3g}4asoF2_lzmj|UHW#Y&7K)8h)FxXF@d)QrOj2F%5&F8@ZJ5xQ%c z4rNXes0}QF+9Y*u*XWC_@|Zl-^#`W02X)`U`pAMO9XlAV;V+B4p-#?!I8$=TMnC;U z>mz}-o=ULh#4|m%T}sF=wY|mB7~CuVtUg4i@;QrM;`5rm>UP!o|h0}_Zi@dYx z0rcZvoH^G8x5#5U^I;1ABrt-frD^46gE1~F~5 zdy7i^9!KCyIqjc9Pqtu{E$NGSyI0MsMIIY(z(LV>hH%~|aULU|o9+bkEa9U=2q#Pn z?@()qs+rqr?dKPNYpbcHJy{2_V-L4dXL4pNwA5zXtu1#7DVVwFQ6CXe9TD}p>e0zZ z>zZc8Wr8cL1Y(LbOW}Nna8B^NYgH+)!|`jD;cJviCo?tTWufHux^ag@(N-22aK*@a zR!v6y2-%#Q;0(-R7AE8QG_jf%6~ps8Lt4r)B{ownAW4GUJ@O7~w!QypyNV zX=%L%ElFAMvAriza__F*?u} zS$zX5F*;aXM)yvB!yBL!aZw@Am*W>TdJ{Zc;XXwvy*soFKe5tYL&F2Eu+@or56 z+LBOa&!22?GhYRwphs(+krI_5l0mZbs>=rUN=1a+^g^RU0&_f6Cz z6*H$TTrX?CX;=3?V}H&1;lo%b$i_GWzPuzDzxM%_uA^nJd2@eP4SL)^Fy=FGRJz+R zfPL>f`yX#Kbz(2T{_kl1{)h`UDQ*&543GRL2~GpKMs3Pe*}9e%NxH-?M=k{sak5x- zM0PwBH=m^`8?6(b^ky+!BX;qN)&Tg%w8%B6-SD3=)utI&3*AgBYhEM&(olBlGkG~s zT29xin(aEUK|se6UwHYOd@tY`cojVQBuW{nN;A@WN4FJ0VOgGQ#nlRp%{9BFY%f$Wk8m*P)@(p;!WEC^%O^0ohJ^eHheR?id(ary)}<1^>vH`@Ec|6( zcu3Ki_qAVdmS2t~{Q6HG@R5dK(G7}kk4(=Tk;(TKv3bXBjIg2@JhQjKTibgsTdjJ( z5aae^A^BhMyxIynfvvlld3|Zl?D`1%EGrapE;B2dRq1@JZhhglk?`l$N?1P%iw*Jf zpCA{O?gFdalFlu*Og#@I$S2DiS>TY4V;G*|DXqOss?E0ltTIOIB8vg_U)ZZf?THR-XX}Dj<|R z96P-2u@x(E2CMgXVCOL$s}UuHiymb2r0WC6>;*`yPl2t4VEkp6jT%rV;>LGD!wIH* zFnbkK$E^`l>s}uALbyES>G9rk_Q)N=((!qZ>x+z0q z$bLE$Yn+vLU98mTTFj9S!_WZ?i`76C4|MtUMXuRY-m>~)$2w0~;TWHLS&yI;tx_Fg z$&UvPhCxRG1S;0RfJS6nUslt0td)GroP+eA9Ss#tM`oQ^d6eAorWbM>*Cw zt!C?-aJPc2WC5@9a%p|kOQ51FP!2ijt#_nb?}TZSl{w{#6st3|O1B-$xe_6k5yc9h zJ$ya#L~kX~KmXl6mCJ~VYqiWf9BV*0J!q@Rm7xZla?=x>sHmSR=K@?ywRkF5lD0>y z?s2z3TFdp~sb_Oe8>)@xxf4jUT<71#Sbc%HH*6=5k+)0noe4hSuysp* zTT(Eih(+UPDV>Wf%4noz%F7?&Ui^P1u!x;A&)W8Q<)&YIgd%!`ru>E*nx#xaDjvXA zP^mIK#Lmavp&3`m_x|BQCVDGgT=O+nA6@TqDx{5ZAv=5}=UH$QGB#VYAKaQg{~4HO zzNpl~zmWC5XtZzC=W(V#1^csR8K^7h`M?KEHcZSxMMty5TPB~HrbDKoXTtTw?qe2K zXW7_ojSrQ@w-gait0~tq~$>wTvR=;n}5F`getdKI-3t{3FC1;PjGZ z)`vL*yRw+ZufWAkgGGjgbe_^L5MtH1Se5~4X=y1M2GeT>28NoNnx6{(e^v4(E$B1w zk%F(cIq_iva1<_HBy5q(U^7bDYG`EDVS+TJ%@X_pwbAhDSADs~(nz~W4%^<$lx-(S zX#RpPxDJ4Hdt-xa(4_7vb6JuG%PN?qd);BYBiWqJ1Z}FPVoze!`#E1N6K5ytaWez- znFgHHy=F7g%Y&lur-|pPh0>tj66g5(#2VVnMa7GT*=87j5x_paP*_`WGpFNezu@yiEei)pk=>h1R7_NP3{+X8)J*#u^48(qaMqJ>R^lg5x<&?> z;Qt1Svy{3^xoH{2KX?Moyw1cnZ^etwf;QWUoAz+cTj`M}UEAI;Eb9~2tDI_~YjWwN z(-$`3jR{~L>7MEqsdcJ%*|(=T~CC3;FXqF@(`jnXPrFxD!1yp*}Y zypF~=j{-mQti2m(0xP76RI0GGMo7rjQi@b{y8fH@D52?3D~$Mm5wm!!lKf7FJ5SMK zBGD~B{oF$Ucpq3fIfD;d|EC*kKLl@oh%8k%5ZDKe?}9{d%&?5{~%6<~4-Mr-{C-BYr?gMRgFJcg&!D?huCRl<1BbHBZPB$dE&9rc6EVY;xH9BivxKYEUMg=zl{|W+| zaO2XwZdepYOru*}wCx0^Vwn8FlEB}(>4aJA*=-ave@ENo%Spf-yI#xM>COva@=(^T z8)%zXkQV-caY{slUO`bJ?GkTXA!bV2rvY}^9&Y&=5W|EJ9iO18a~xKtXU6(}1}SB$ z(aVjFiBYDvVXvqpUlyMYpvvpysroKK0cbmY(ls+S_6=xDJvK12C7F_q0r(lnL+cif z(sq-FO@Ws87YMij<1~;72X+Ja`1t$#yRb!7(EK;pA(yW0$Le_1$)CFr{X=WA>Mjdp zFI-|oMn(d5$Yv`DN)5JZlrB`xRW4O5pQXV<0zKC`h?a_X;90J9eEr;--sGhBW07)% z4c8KqAr4Y7XnC}7VCyk-^ms`M^_NBW!Io(>1tLWj*;UZu5&kzPwkLf2b9F!LF{O;4 zE|%Cl&bb!v?B8>bgsn_0e!q6Zd#8OZs7W-i{5DwklH_yMahwlL?^Hg%^x)eT0zRA~ zoSz(mK@a($r6Og-yjR`r_(DuSAbBYup0}%1FJlrpZ4yzBMIYHx6;zotN6BwVtNO+9l_IA#+0Eky;Fm(8oTmb$hwZW!j)NVmn8Y~Q;I=7 zi%uxt@JN0QY^joKs#8Qa=X+(9P{Q{qX^4HVRcayv^E~R!mHcnk7wyD<#~RcbwUV2L z-K;2(sU<=8TFo%G6x+tBYYJv%d5>@2uD|d=a9!CW{nU;z^yiScy}*(nRx(JEqHB}* zhG%HosoV)*DQ&RrES(5LArM|Lxp)@Vzgg*EMuT;2 z@%aE-`hfCNmP)i%vp|`%_=V6NZ-&>f=NtlOlb)i>*l&kiK$d10H7TECYyR>3=Rm@tY ze-OezY|xIC$e-+Zisn3xa4qD?9}+b%jN&wAF{1iu$GRl~2> zKQlyfZyqyQX?AMVd}ST{8#`2@Jak^j;N_&STQAk!=d_V?O6gLzw5fNbBkz;;DYxEH z5@b*i5_V6R{I%_@lNAz;?PtaEB#c&}Qpq6I$)GgQuV{vjm<^gL=U8jz(yZpb{t?dl z+WdspJWo&LQkr|$vB`VHzlg{~BcwiL&ZB-OTbMnil&g%m>?3XSxyS;Jdwf-T&jTj- zv{cFDCgmyW71-=LysI3oPxLfvxIfc*)0M@V_vrXw=4aHHib|H2K&n^HW3GplN3xmL3$&|2;p7>FqjtL^aVXC zX{J1Qq_wv1psq^NjgRerNj~?Ek3Q+2n#)aVKUG>1fQc!%6m57qb}s4{j#{aWL|nT5 zgog|3j}L9L0ck*Q@29E#%QSwOU7{Y}o*AAZzB%O)R)F(; z_O%k-)CBJ7$;azA&9(k}(2qM# zA%+cbNz$}zsj~DLspGp)flf^Zr=xk+Xd?a)Q!UUnfK3K{>m}?6uv;WNsbEs(;hDB&m%{NC%F?8~*uV|f@?EJMVkw0XDFrrSQna7-<`JOD~ zf7ee{IvH{o0=Wj$k;8MW+pQHEN|xik)z{%x9YY)AaV838r6%PTczBu~4H!w-?>9)O z3DU^+-iecZxp+BA0$y8UK3M@5JdK5|hS`0A)s*sU%K}A?V<<3`4pQBwtdqUOuLc+f<&`(N+GY`_#nIZ}j+5rcleKc*;tu z%921>;hTrU*BowAY_v9!04}_Mk?sGscyA(mJsvhm6o6~1kxQXa*?O$mx{;v1z8)wU zz``(GEN^zvX97u)1GUSQmQ?pu)uQ*IRW{er8<9)i)L!s}VxwUw+QfEk{qU zZ&xsli?5OWnc>O4n?*Z7whhw~pLDL)$0$#^^wNRCe3u6eoOx=mFKjli?tDPA=3N>r z!7g^_mRFX$2hqntv-YFRKr8TeS9Wm_6tAC83i1+bwphIG^mr$Sc%w1y#QHm}d@4yh z>E<)0=NvNmA3Z#pZQ0abFmTBnQW}L$u&}b6TNnzc*MJ$U2j<%@EZb?$c#E_5lfQ+* zpe1o<0LL8sEa5d>R!-a)%*T=!{I-t19u>1ek|ko`c9FB^o=!+Vh$iAP9O2yen0jj2 z$Z)#k+UdzwFbJ(z@BoMA&YO^DOOi4>8LO$Kc9Qa;LMaZ1@>vfoh{7eXOn{wCuiN;mYuA}<~u zAMc`%#NK24K3newxe<6vOeO2>k4%*XU8Bi}Lj5uiRhUo{Zr zh%o76GRu^6%?i;hdMchU1ji0aW{RC?r|8-xPE>a+`dBndX+A`?V(a|*pYft;x}@i@ z&q*I0J6b#n?MJs9fWpiG6sGRXNhYM=_wG}&u{ClU4L!N+2p7x{9k3@zvY9b583GD9 zJfg++Fqlp&-2|;1SuU#w_9$Kp7TIrZ+ef4P>NsI=1=CU}BQAXmANC|FO!(a-)M}NJ zF~isQvsPExIHQ9IXx0fjte(}s?tIR~-=ytnHT|-pb>2#LPIOjSd{)?hg%8^_1UsLO zUjA`*@ZaE9anj|6;v)oGl9u>3xiLXB6Ejyu&f?T+axYur#)|4Sw1b3#|^cOsD+5AliT&tD0c!A z{yE8%Mv$c${O?a{6p2Q>8xA`0g(;bvtM7TR=ZLAPNgHUhyT||hFbhe5Nws`OuM9vU zTp|suxLVPcC{&nmXa$(1H=96QpvbEM08n9%#N@Zj!>xkV4gA{%|hg(KssuwnW&3 zDCRdf1?N_*?u}z`AORuQz;VGCJt#SZVn_hN1>DFC>9uP2-=Q-~0ct;W6ZETsvZS_7 z(W~Ub0MgBH10eBrAbM4g6E9K~1;-RR2Vtl!;Mc(42CC&ZFNcw(neUBUjBPKySnj*Z z7yj{uVQkYStZGIX;16V({?w@Hgw`uc~uQKr)R9j6uXs z#Qx9GI7wrE`*cG=IYFzy437Y*0>f8=@Gdq@!Lakd^|N#=a=y}LE)@9G?1P=EiXx%ba*API)w`aatQe85mJl+mkO^j3mZXdBiatcFzJWyL@_c?J=mU8DEW%3po#SYn%bmf6k<%Cxs zrIf0Ul~dR|OkuSX zC0c}%Et>T-3XFtS&U4Swub#OvDASA8j0^c7+nyGCIoWP7L$j#t*RSC#BZq3X2?woa zU)SnVL$hOR9!7XQzsV=roH5gG3$cwCiS2iGyOk3!PMTz)J=i3$Yp{rOb1`h4JL3~+c|Dfx`Mb z;baP7HIVA`zu_(qQL&p-K&b77heYZBGJ%CHB(?0zu?}7WxgHQ@f$Mtj7;3SFl1_3~ z18bt3<^v70a#NV|84?TlU(-c zQ!Es$=m=a&5g$Yw^MTHsVDF9^PdF`=o(QV#9J+`Ol^0*?-~AqBD*Md^ z8Me6!2z&ADSqpF7j@()=1W%JN^Bohjml2KzJNq58rXaTvnBOS$3}|umUbk%P?9iY*d~DWe!SMG$LusXX)7Ao&bH>nFF`| z{}+7$rcnT>p9X&d0j*_IjA+6)tU_s4;%3LX%GPnWQGnHw$8Ue(?pucz1H`16U%*j~ zK<19lg9UbBkXNT>J!4VraM~E3b^|=s{vwZmR?Am278LBWv`aXae*x!&xyo-42AD*O zZUXUXR{pH#D{#FM4Z17pi8Lb^{KD3e{c=<5zdY%T3F5%rcVk@5ppAYu>jPd1q+g!^pxRPz>X_Jmz!RPrdN^ zY>V!xv_c9orISM+^41{Q3CEC5HQt&eNz*h^`iK)|q9?w55tw@*hSf-V)|B!0nWRKi z7q(6{KLQo9zb($?G^Pf>hLzO(Pd^iVy-%3b63S#0vJIZ(a(P!SUHRlNSRgqyK0CE_ zj9eI}2Df-BYE}LRhh9b=nnaegcC!n_Bpdxy493QC@;VM9&ZFmZC;~a*Bi+2;B?f=n zSg?2SmSww0j|r&cijC1lFw1|Y+Na)>LPwEh#fG&SIb`^h5yoK17QK7e<<~^=nlR;AVJ}J|VhAr!%27p_+QpP6Ox$ttH?BC||!^vrPo1 zZR_QBB-Az_Evdm7K5_nWQ~)06Z#Ltl&bO!$ucZL#I0F6^UYljH?Pi^cFQ%r!AX#Ya zk#d0wW8c3`iMrat(`ic1{dYtaDL#{0aR?;*B;dJ6_85TESHHAo>zJueAGF5`8P&!0Z9d z8FdMcAyB#j!CSg1UJq_gPn;pGsvElJX7qHoPuhYYo))&xJq%s-5|ga^Rr(9n~&lP7j^^X;AG9# zoZ#I_s+3ViWy6C_T@k~W1TLSU9qpEP(UdiUUBhyi8FZK#t_}_EuX+@1!t}XB!L`Lf z+JM0>S@R0b-Ox|{Q?i(Ka5}elldlB< zL=X<#C>lxgC!dJlT1I|)FH$ZC%fWaNh!of+a8??gt<7Am8PCGZyj-^zyulyV22wsj z5Bj5a{ZjDmeq!Z0Sa1OYm)p!scwz*PGGh(@;xm3Hb;)G1#SBoa;Nsx{Uh}VMrCb^r>N5gDLVW`R z05Gv1AduE))C$_nmJ56cDjz;|WgXaqAe-hHov5kV8`NQ>XnfY9Sci@61c?QS^>*Fi zl4~c6DKw*cIBI|oEANzX@z}S_Ka+Y4Ll0i@nCjP#ynG*R`IAHUL z1>=Dd?QgQJV6sMX_Ir)OKw5T!G8w61hr;*R3kPti7e9e6`rEXx%BWcLl;(j=J+fnf z@|V=iXrL_QdxlGjis}<8o$z6Kb9()H&QRyBXqeNN6{%J8Nw#+IYR^w9-b%4p#YT2V zE~$LJ@N5mZenhq*#J6@vI4_cbzc-?`daF*pU09@D_{p1$D7`#ZnT)8M<<+0OuYjVU zJQ3j_;c@02+p;&EzBik_Mg4W}Z>@OkwKC=8FKkXgOz@(sX!$2$m|2z?-(#!q;L%60 z%x`A|TUf`+9Q)ZfQqW}%0vXc5;@cOb0zoT0;Fhz#v7w=^u3l~21%x6%dJ5U_;hsDO zKrB893DB!nJD3Z;^M?2dUzsw5Im+@8k)XKsZ{Ee*MNd8Y$`X>3cdpz@CZNW>V9#H5 zIL@`&x+9(L3=kNIY{-`)1nN(WIeWXi5doNb-Q>_M@7$=1O=yWAVxQT(X=5Ien8Q#wsS~0hu^*nzx_mBc5b8PTt zv|>D9!sNunL8rQaW~xBc>QT00p4mQ4yJ`_$_jie`q$UYEaprh4f7av68-Vx7MQ3DW zj0@Z9&jUAcVg)1Ku4xs56W<5&^0Bgp#xIG~U!$<7M!1YbzU-bb;$@;dGoDxKVSCr5 zf}#RlRFk_Edz6P+hf5EP$r(o#cv$5*d8E39>#`i9M0RKb_Q)}WL@3W&e%gm9VvW5;3LmN-WGV%lTKbmr zaIohTDI6Kn>sqfJ2%y|q$T-iIu4-L=DrK8%t6Zj1A%Xh;39p*Sk@S#DMe8gvP)I0n z`TPpZk2)Sq2H7s=)jnE`O5MV;cdMfKcj)sMw09#?_M@ZQ8EENaGjxCmt1EO7Xbx2= zG{AeOvOzmZkSu{%J|8u^fT;HZ;|rpwmM7)mYc%URXMxobcT|p^Uu@s9Oby;eI*elf z^hyoZS+4N#8C}D=`2L8TL1LCr{If4_twnVJr8%iS`|nYF!JAjE1h_06!BSQ5=! z3`7fO4=Ug0E7O0SJ$>bBcG}*4l)jdylzRbm!ftyF$MTSu+w=b1D~(2QONxk@fiM)( zt^@ag_nSB4yUZr@I{(gWK8c!3?z@t0yWY-`)ZczFA(1v99LREdxWPE$7slwoHp>h{ zGrfPJGj_e?SCx)`Clmn*?26utcA0}o=>Jt%fyX?eS7(TcB#(h4pAjS4#Q7htL^wmY z5>pV9Lzc~Rix?mrL?a~q$pCq7Mew?{qtif%x^7i=)#*4EJY&LGR*uR1fDuGJ*F&tA zvS7mPJjh2k%yqa6`DBcX)cFqob4@R~B&<+6QjNg$?@wM85Y-m4X2LipIUIyp9&qr_ z6DCKOEs>)xH{qcF8|eOkw;jo&Aw&RlL~&NE?@JU@0&IQr$J%x5n%(Y{c^p(o;J)*j zYz0Zd`3dB?g<35LNJw@&SkK{jr-3Dj^J!uqcf-JeMm5?+eBj$87%&33nw-{Mf{@$< z1%OSspB8*vySv{1c<52PU0kmU$Ns$SW#l_Z&I{bzq~XdKQ_pov@caY%@q!(JV|1uy8BaW!rR?U`iw;3-g7uF9Yvy?R8ecKjox ze6^3R*U>v)DN|i8`j6Z?G8!hJF41H$(d0fE*E^Y1>Bp->;tT=t(mG|3w)dZCpJO$N z3Pa#4-B`&veEgMPtscH0Y0vz)xdLf;7w>+8s|JKI@FW6M z9%oQ{e>xA~zNJ_|EO)17*?j8=E~xJpJ{Ao=;yV9uzThQnh*BM4tSwgTwB zrV1;>rzM=cDGJ2BW_W0WZY-dDs$QOG_5T`b9N~inl2q zF(reGgnsWo+XX3-9~at|^D3g{^rQ`bjwiMs5Sd_Xka*NgBSFRbmXp(e2pOo3MRwd^ zIao)b=trUHF*12DYNZ5fEm*cD!=kDMz1=9bLR!ng)qz;xK{#pq34)v8$z9IiUzVNU zN#~1#9eY}r=h`;XKj`gNcl(3U4I42HgMx&_zLPUr@c#}6lPJ(!iG7d2EwCD;jj-_J z=)<;&4e|aa8mpA()nz4)=pu2DG{R-)mnnSGcbPkcrQd4D{D`YwB%gIaQ7;UjnT`xG zy8=C}Hs61Uck{~vc1wWOziY>RW;0*uRU-e}eCH_lS5t@}Dxp<0b=miL1V4zt?OVJqkKV)Y*v8|5{}uK}eU7gRm%0x-nMf^fF?kXhQ;B!GgX%OoX%OJ>W7AA72hc=D!tg{SBc zh-`PNz=uz0g^y{nG5XIk zbp}doY4VHqDo{>LK}`vs+?*!-$%}w_11=nroXLtwr6H4)>B;@K0RBlDjCQ3J*DL<2 zq_UZS55p_b!#tq=x7#hMccE2WGT2Pmuu~QXsCR5mS~y9}74%PC@f~4v0Y{ zbn0|}f(>*W73!?0Rb?2a3-Q|fVsIe_86w^js2E46SQP;VdnF|&3zbj%PT0XIoDKhr zglP0nPuvEbiXE7hRC2o&O`hr02d2}=3$V%y*hc_=>Y*E-WvIa&c2BF6?V32TikV6q zr+r9}SArOzH6$&ar-x=ckt(qxH%Y>~+E!PoKw325!}(GyRfd3q%`7#oZ_S!8depW^ z8FX#dgG0K+Wx&jNz)ZI-D_p3zOZd*lvhh?^K2-oId1|6?V%!}R#{&^;B=iNj)e+kz z&8Z)+D?d^a&tJ0b=t8ZzVlo`4Q>Ywt*_yHIbxc6uwQzq83T_ZTY2e0v*`0Pnd3*rE z5)Pp!3rX!kQ^JBt%4#vdpHt5&a&4x*#2-RA`JjPl2UNvtT)Lm4AOT${i{j3@Sbgz? zZwyIZ)pQ0D$EdOP$%nI!S#&G zNzSX-Bd%I|3HLMb4WPDvkjX2l(=Lx=^Wg@pxBX@EX(G@3EtTmlRmiOT3{%8DK4H*Oj=jdMYkN zU)-lX#WA!4DVfQ9q4_GTtS#9Xt*DqI0YRsrMeSo)e!NhyM=jZk2W3 zN_#RP4p~lCc?##R5!~LMakW&Eyg!adsZese<2I~k>E$W?^0ukTt&#x-oz7i{LZ*)D zW0(%uImi$BKiWy?z`b;D!WjhMU(p7zRGnML01odg-)@jM5Xw7P55L*SJhNe>g1xur3qxRY z0=MqTD!B)xKL;0Ahf?ZMjDyzr{+B9?W2^f!D|gVYxM7!eO^haR&=RM6-pSn*xo|tT z@Cq{$nd}w>p@Xs}5vfmd71{Ho4IE>t@zr*gfhJ`;KIMMBVtB>_(kwH(8qBee#pAiu zN=d7qR`RFFXw1GGR;A#tiJ!sbyn{zb>NQT);oJ29VR6hztk5Kj9kU`_s??vj36yi zzELKk$bY^vRyCDc(|5(2U~WPhP=OTi9ZCC3^eee>xkwt{B2ykm{|b*N*u4$AXB8JU>I76*@WWkA18twtM zm23wO+h(15sFk>$Q*PyQ$V+HQaLO$G(c?#tq{QATgSTEPcN8l0g;=FUV97JN^}GaO z1C@jNrQNwgr*u%HiVZlO4yGrX;wAvC1++HS6T*_hgvnx&=$YH#uR;O%C4r?Bg=I-t zqvvx@WMmgD2p54~7YG2Z;da;MBwPb=UJrHLc)HF@#i3RO(%G1x1LwXM%4Ki5J1wyT zG4I7Q|32}8y-_LcO8G$^UN_Mg25#2HN$m4paW*-j@|E1zfSE2o|W|bqj84Luc^WG=EEu&C$_r# zD@{y|a98Lj5hhAH_Dq}C$}~4<$e(Fds((H-qKlJXrDah8cYX(yKGe?%o=-9`1ey?I z^7uD2mZAu#21RbVd>0$aGK|+;>ev5*F;Qs~_s3xL*f`)DK`;;;v0=5_A6g`L7@6Gm`J$XxSA+ckRVO0bXSi%) zTZ8U=Y~D&gIh?M%NC!g=dWFybgr20*>J8EA+F6lapC&a8RC2#QOnN@F|84R7??z&Q zW1_uYfIF{L2lIP7nV4NhluJoA2Ns}Ps56XW8u|bL9mmyl3?Ti7r$Kb9Z)Dk>5H>GW zv0N!WeoN0&_R*zL7tqnuO|>6`=GHTp_I}@X=-$YR6ap`hUE6bRYv>D>#qnXdS2C3M z^D0IDX_DH208n2wgHoO${)J2V|HcU*35Q0CBId^HNMai#a#eK?>ynu1Q>4SW;uj!J z^PyQ6I0{F`k+y(zPnKcJ>vT-**c;9LYiVhz4*pgrSXfXka?S18%w;Pc5J;)dj&FlW z8IpKi0H0YG+0AbU8{&v5s$|C*F=+%|hOLXI#A4@!nZ{P1EXG~0h zeQRtI#@6e&F)V90bv3&}Z=)Y>N^-XQ={8D5(O&6zZ{~$E(|oXP)q{LUem&2C!5B6M zZXUk`-ujG4QGD}-6UkBz*3h1ye(lZ-vsCPsi>ROl zyex+1kHF__pk@&ThH2uy_tLS$F=1i;dGMvOEaVHcBa2i^i}(T5N(v%7MsZt4JV!hM zl;LeQy#~P?;Dg;O1$^c`hCPP=kG`+;bycK?>x-JDh%4*DOsm>T`i@H_7IP5PakLNmqewJVYqi%Y$u1EemfBjJ zmh+xn+DzoynIc8xB8~86Pn}-Bv%p}c=cMEQY1s)$GgNSnY7{1JVu^h6iKG{`vr zCroQil30PxpUa)E=}2)tnKzd=!sn{?b58ozH;sVu4}kL_!E%r9HU{ zk#cFq{v0T|W~+w(VmU-t-UbYYV(_9923hCmRBe-wtq}AkBE#5+*$$1#Wk)4>SXI&V zt|YaXaQxBrdD{*m5rqSQ#E=JgRc!*m$KB;I!0lI&vCGb*h0i$pKjAM7^UN>V*n89~ zP4=DR+}dIfv#g-|c2*VuZsC|G`$^rO;aG1x7E*Vy3Ud7`R!^OLw!PUwd3~Tad*_#q ztM`PqfPx^FD)&!JHxNH(o5I{3FmyN$5`i#O??%=AW!SA)t~p4;_(v_e`UtIOquD+Q4Lz;lo4CZ1ijdC7fbjN6rEbATe?SYI&eh zNL_^%35W!OioRObnFdC;ILYn~{wq)g1J4lD2M8?{%x~?*USYaW@v!~^A1IlCtcMWs zr`5;n@DCg04AMcZo;>$A$NZdiyIT@tolt4+Ln{ol7X4Y=^}6Dllk9ai;8~m>bI;jh zJqmZ5U2z@>(QOBLb`+@~nJTSPJP>#m;o1AvY@tPwwe7mi{88&kqHjr{VBhmB&*kB7 zpZ0^1%3lYcCG@dOgA19xcg~nhn`JdjQfpu8*la!*r>H$OE{P$+{iU2GwL;Nqa-N?F z{K3^{KmBMi%)kTq7Xg!3Z6YzcCo?MN~>0v@KVzx3PC4PSLWvw5QFg}$k1EB|H zDx9dOi~^d7eaPsC;aJb*07snF*FjF+{u;kdT);lTd6Tu+ESg~;3h)ywL!aw$W<|b< z+>syAkhY`x9a2jiK3B-Fs_!riZ%ZU1ndh`(jgZS#4>X&J!3Efvj0ZvmcUIlK`?6@ptttWDAW^hO@0pR0Er%vOT z^l#V^c#Qy+WcUZ|(<;T=b1d@ zTR<%mU_2iE*0lmC|GnA`(mPP;3RY0<7rE>Xa65(?srG>0JFMbz54(9`2r|Jl*REnh zrGt=*4ax=zYy;#Ksp}>=H0&IMNXv<20TWHT=d=gQNy)1Tf@4Pz=ja~ZI@jT-1Qie;W%HJG5ZGq(n#vPYD`z+#(fTW) zeZCNNUdaM7f?>f_hq$UKHVa|ZLF~-IN{il*(U?&?o{m^@S(N=m$JLn^dcAc`loofT zE}SfFvh3!v`UQl%iUxC>F6=zLomDbayPrWM2Jq3r3D^M7vaCnrE^Jcvo;1wS)ib+E zz=FGZlwG`HqefaY4lT2hSQ!K4?9@OpN{7H#3YBr@3g4PQ@qwNfZDdb~)CPb=E>9T> zhl9R=CR{DD@?f}MxIe!?UFEP(g3#^jKhW%OkVSTDJg@=`?x3}N7m8wGAO(5gMJc_{ zxo2CT#tV=vrY5ifGR4ZX{~DF0Z`be45(;PF4=!c!dhC zKmqK&^-3uBQJ7Q1Q#Tc#B5mdeaBej14!dMxCi2h&s*61~nzg{wvBv=4kucr*8#|f4 zqu;=!DDA!b4*qSaokZjNmzD^sS!>6J#RZyU+%p<+0DaLZ72s`nikXkCGYYGjx|0DJ z>2?B=M@z4c?8IkbA`?vZbAer-m(W7dgXJ*3k+mKbd+p3h4fH0CEKzQQc9x;>Zuxyn zp7I37NOxK}OFo4nbHlk9ri zYlPWbE7&>V`d2>A%WW3{G^_$mUt~0zQvNyc*!gBE|K)+-Svp3Jt~iC=8nT`H!U-Cgj?^sjX?ec~U~tv_^((B%r_ z?E4a%htbEH^pDk9MpZ{7q-o>AV<#PpA80v&6$U-i4fvw+h0h0eCqWoako-3>3MdF? zHPp(TQWykhU2ZqunL!C!fO|qT*X6PU=_E6j^-vrwVyPOjplb}kT5JYZK!^^+e+XRm zD9oRnZlrk$K_@}YK`RP~P&o-+m;=|Thd-P+^JI{w zQU^nTAb%k?2=BG{!3>p)naAZlmg&&niFgl_n?I~7eYUsw5N#>X;D0s_`C#sc0; ze|RaQ`l=_7Q-!`JOl~ldQYp3Lui!-yE$|}?as2K0@U8fLepYVkz<$WQ(E=XZCqix~ zng6!@xq}}xg5Mx~asy<`vOE+qUalwAg8kGcPUqJIr^SM@SKPxH?s5*=FD+x+K<%H0 z(eux*Nq}brDtPW=X?E{e0VzhC;den$r6dW#wyKWj<=V~M!Yzq{>!0Xp%|Xx(+!@L} z-wodJU+Yd}*OOgmUtTH!%NY(ki0eX=q^!!v>8?N^C!o4gk<^M~xvnKWFfSL$$-Fhl z#eb(GEN7uQ^Zqp7r_nfmmg#y0+Y4>9VxWHbiQ^dcWecVZ@OVVMiYefx*Lzm#uIsa8 z(|Y(6oZt9{M)b|x&n7OM+(3$6Ajd?784FP2=r#m=Tfm~hOn8Y4Yc+$PxncwnJthDs zKM6%Y2~GbCBzZCh$fE`_<;tcmAWH9xXTYCXDf@6V1^b|uZfwJs4eoS(o_iKNN?eu( zQ#2DV&jcqN@~Diq$@@hhKHAq`Tpd8Bfk=D4=sT(B1+ox*#G^>q|oLHt8U$0}nis0oLH@gJON92TmloQHwhR`Vx znz;t`XoDlrYXBgd)2?BPrbGKP5D-172DFTjf_6gS&QAE;t^pjD3*%)_VhB{;W(Q8g zV&@xh-ckT?=@IxRU9Hb^VbIgxpDdd{$>Ba!^Ey?7QU(}Ey@2!l0>TX})D~t@etfxH z1nrKR@aB?F%-nh`?=4^wD<1qK1nyJ3;@3FrL@3e7iJ2%BDY}m=8)e63Msx}$SnJKp zFW%IVw#vr7CBU!q&j@5|6!M8g0?!66D;mFZQP$g>KiS3d`{q(;TOv&p|G@H76KgK} zxz^!^gaD})ez6K5DwMrv_L2s+KLWYYLpQ98Mx;*OG+Qpw9P427EzUL5XA@O<%!40b z!u~nierUoN_Xq9&vKg;y5tjPLGtbXD0&n|4SVtajZ%?8MI2=*X$eHA2$*A8cizEOn z`v9x8#2g(XhRe3_eB%xo}JZRAgy=mThs>0bU`mJfaS2nXj0j`WfmB@ub=Eoq`M^jhGojY& z|0cTMj_R%g*X4=3pkLxpW-YTTCD1WqSXNM?Va#Ns8`2O_hs>>x-F!Djz+ZaAr9vqa zm7GMOfUCv}Ho*|bIPu*N0umAFu`icN#3$CGJ}wxWuk*9{H)M+B^h<@jhCH)=o9x+~ z+Lh;LDyC?j)Z}$0;B0|<>>$Qe;HaCjs7obFf54MBC$^k}yuHb`i^QqWrGNYMs4TwG zT(Zv%v`BXW(tv}5gMsuOJQj{$-m+ge1U7iLg*>lLwsc6Pywr8c%e9%$t*P`2P}?5C zhmCGW2Aw$wx>S+3{Y*pu8TN{OtNX~F#5y8w?OJ={?kqkH(Td~5EoWG1zo}Y(-?o5D zd4Sx5QRd1RSB_%@AYZ|sTtvVwLCLxL-Ct6-rZZTuYV}}r; zb9g~Ir?*Q9GKzFyxyb@@OwE4loT3TT)VSnK5`{RevkS4HUsxTa>aU;@tkm>U%fJf3 zPyS46%z6jetisG2J_f(vT{`1 zS|&+dvG@3I~>V>j~L(yEVIGF(XN7F|k?PWqiq~#pE`MEiz(R8!=dGn@3D05{8 z7ZiT_2tn;=%P8$bh4t6QjdSumqj|N)_^6VkfhI5mHQJ6|&RC*|Pg3E}h5C1wv?Yd@_unKim3tnUvAo@+xm7l1PfLD`g( zQHvndU)|mW)fhX^b)^Anud;F$T!6N~-T`n~H1n6ok&>hu=ZB#D7}`i8kQ04#h4Ow# z_1TW#W@(rgea8B9iW#%ZEWiJL`)B{nO>upwCaJBS#LoBaI-U@^hiE;S z_vK*kP^unoe0<{#mofQPWaO!H!`tCE+*GT-61kT8vy}{@X=vBit&@>ea@$4L_1?4@ z`6D4n5PVBzSOT2o!o%(%AIW41*jsb;%OWj=nwg^HTGee8Tg0&m=*O1%pqrcllkuHH z)g#1rMlu<3MN}gCw}75g`1#r=#{`l)Qpi5~qC)i_!3(cFk)f=_{HfbG9U zI{kdRu0D@r1^^1%X>Yenj16g4I*ovX@C@s0G_7PZ?+R8+hGGRR?My*19Ik!5^M|HI zh$mGp7-I@#_5J-4A|PsklBot1kULuRH$q)e?M)-H8zpC8EX^xg6WX&0Z(BKhU!a;pWvGA^wPn0OPiR|XXc?48tbk?#YHZ})!LQfdizzxtlp7#hKwhlMEwX`i(#)uzmg9N3*EYqI#}s-DZ^YGW+9Ic31<{9_;}=l_L<*Ttdw6 z_#K!lk^P^16Y~+H_zkqcL(ZKXHQ-2qisnH;0Ex>&$}p!KNG*9e%J~+0OGw*>`C(FK zsnPa1BI~3nQ|PxDm1UOAUuqSN!YRyiW>kd5_iL);EF8n8w_K8CpJ}tAqR50;)kp_YZ&d>j2-=ClyX_pEgGbzsr%q6u02& zp6(w_4q>RZA)1dY$&PwhfvXC6dGnA}`wLEL3ut5qW8wgIh`43_?rY1sE$XK0x&0?v zXnOLpvIqavnqfcy76$0*!+MTIp;r%Y2UNMv6)q<$Yi}T8XlrXL_wu;?8V=g1hfoCA!y(%RH2jc+Y2j+Ty5_Kx2u$-I?@vE&RC{h$KY*}D zcc54@LW4Qpjs01vVBD9CH=1$#kIkxv>zz*}xCwuky%-X%*0NC=t-MO}RsM>jGkOu# z!$&>6I{ms*i9=aoXl4EAYXCScB+%QoU*QvgSm$uL3?kK%z?rJIy*=9jZ|yhVNwy}| zep%`_wvI~hPwZ9)Tsv6SyaP4U7qsVl>L@#E|3Fq6#wO`yj43cBuIrGha4p1 zSb|=bF#f&t_q2}+c~W)56_NYOoFTUvXv*$_hzV3}p&Ts(>bKxD2EK<97jt*_2FgBN zD_#)P1A-kvP@m-dXKo-MJalV5ciX<(le2OiWe(37tg?V~D9~5;G+*s7-s>_RWV-2) z_(E_+oC=!5JAc`cFhiirRq6gq*k@+t?OK4TXQUNp_9w$N-Tstaw&2aWqiRr*vM=Vf zw^orNFI#Biub}9mtgNhUQbvv-lS?EC8pN&_nBQg73D{qapz)CmCQHrB40DTd^1K2^ zL_4w;iXM~!LgL!xiOgqO$fK*+N4C+wQg~jo`9qUj)Wj2J`r1EVWn_dXmem8U0uom% z6LtPYxHrTgf1PFa?8Co|&)L}UWdfOLo>y^BuHrW*ERXY-4qo{He;psl zcEQ1pJ=L-g9GB=j!ZE;^)iuVy>D$2MybElTey=&+=>X{xM}TOSfwFmq}+}sC}RBt;mk7U1Ln#nn;l4hp&omR$i4K!!bJM zRSL!~Kp1(S9QWKqc&PTooGl{o&K^A^34@NWQ55To4`6VL<6B+gV%ZyB=`f6RfaxRUr*1wey=V> zlFY-tZ9*70OQ5LY2RW;Wf}5sfr~PXnemY3bil~i3P3{W1kqB-RJaN4oY``z=x$?EA_-vgUwQ+eABR5XSN5kK zRr)D11fd@bK~fhph+^^6=~V)ch^stEt0?_5@OFDLiB;gV#?b548==eNKGXX?2Mnm5 zJ{OfQSBNw7rGbpKP+Ey1aD`8KmN8qCkdq|sf3Sk{k38j5W!ibqbKk25a*Po683X9) zw@Od8Bo6!9_OA_48{E9GuoC$&7vVu<2sD3!f&VT*nI7J}=v8fwiNJa7&>FP-q47w~ zfTq3k0ows<|6g6WQ$Cj{$4%vQy0X#@&;#@n0qE4h>pZFXGD!d&=%!|ux+U=s%nUF~ zLRT-4Xau~i5VZzjJ2zt=Z*hPE$;>mY#{eSZYe0hw0TGdznYmW!P1uNOCB4elv#(t% zZ|UT~_75-EFh7|efizjwITs{I)(_H9*Vm{i}|F1UG8tH%c)qB7Y7UB0mnh(>JON z$-r~Paq1Nh1}L}wMD%skY||Vg@cvw_cN>hk$lbDE1{W$n6T!zPvt6uvDr&)Gl_{QppQzNNg|e8sd}<)3c%7ZEP^p|e&!KsM6J$p!UMu11LQeZ zlt1XweBg@UF!pKd=CAw02{joEG!}L&=cgM;=gpNd{th;H8olja`+fbIizTYh@neP5 zk`FZwJ3?dyGqhI~gZn}rK-YvHSItM=3}=W!p*0c?j|gOdZx>)gp}V*H{?|5At47qe_`9~LcL496}`q^UtsMA~hx zR(>(y$ZjcZlLHgAg#B+)4`)@1Ul(NO&`tYI@N^avSdwf zeBRp!TQ8%NdR0uc?PlDb!bd>YwmgE?erzM!^S6k0=+&J>fo4@igW9*S+bq&n0q~Ce zqW9bCDDekrmW_3O3RfW#<*WC1+I9%h*&3*9gAr_D1?FYoG?qG||KIAZVC5|;(l*H! zX%OG0@3?(k9U3d?do>VPpp|5$&S$6LFE^+>KWj>x&lNc8%{8ksU}tw(6?DeQpe^Mu z3oa1)_fVOTrmz)m>xpY4ByiD&b}rYgY+I!_ME|`UuTA5ZEoqE@eG82nMYGU+9X()p z8T;`PM?QtGa@`3WjgZ3Eq&|2P6hNGjG&=%Q3^;hbmY@L`z^U!91yGZCF9^YJ&VzQT zqn(IDmJWTkV6VCd3jJJ;k2a+WTdrAzPmn>xjmv=4MY)COIEr2|@x*u>Wqa6IiL!+k ziPF1Qse&yIIy?U2rX0DYqcWfBH|zZKv^M7--7T_aOg?Ts^i`jK4<;hAv%};VFab?^ z(N_}7fHQ$8kd@{z;nq6C>$pm*o;Tn**}|XUDopUh*URBHAcRfqm?j?0O3^36C)*Ny z7Dwc$Vf3TB#vrc~pODo@#TpIxARS2~`oIEk<%5gLc;c5lfzUiRBy~2x?==|ynh-@$ zt$Z!4#bY;R7uW1qea0ePYkcDBMST`<#Yc>|W7%N~!4M~bYhBR;L|m}08U>eLr5bQM za|UW2DMsAw5yW#`C?)nv*RCupMJg;*#NUwy^K}wqt`gHoeFos%fL5chhA5Mc_{A}x z@4eds(f03JR7W^Np?q@aba%g=SawC3~vG83;G9o*Jo4LesGv^8UxM^Xb8`H@QV=LvjwNCu>4sqoaFSrg6 z`xk4e2cXTFfm0dtJL$_Po*=zb(H4o5th3g{TsoCFRez5CSDPS8EeGA>(=U;PThmkb zxaq~iUB=u->-F>wSLoNhp`({U;fbXPUA z9u32EcFdEa<`S)Me6<)KQRyEN53UOy5)a1u=Qwr1kqISxLgd)9MKIYX^VnpnwapB} z<7p3Xlba{~kn%$(()4U;zS)>7zN*nhw>u46HdjUX2LpL`TA$4{OXj$TtBc4$yaB|F zUjWP*yZ}OJ;lP$uE&#~6{omc=MjE(~fE7I8)UtK46NnK2VTi^RZWionCBkdA8haGG zcEDr;&-3SxMupUbNd@D6OKY6bY+Pj%-^|VQQI#fa{Aa|}&vd%8+~N&nr@v(lF51-F zYw6Y&Nz?r%rnwt7Wg6DMSq0SV$l8uD!D9&~qax-Hvh3}E6-Ayln;0$i+X3ge?U!%B z;jtb=l}slVz7vXW*lZrpo!D%;T{ka5J~URP&RdRFJ4WB(AAO2$SR0gKkDC2?sjUD$ zXcI|)npH?=-Pmsv4_KF9io6M7{f*I^{|>wY`JAtiYl=l{>%=l(ViTwhh6xcy3F)z$ zA(l!EpeiD6EvbL5@P$$4*!^6O;Cn%_Tqgc~Tzw;K1^tpykkTdkKm|a=Ds;LT<*YIb}oWyjBo@Y8_H( z9s^^^r-2$sc?bSCT$WHiPF9}lvP95bB>@Xx_7cyn?PGcSFnw+oihrS|FvXnLhWbt-p_x1{|QXt>6@?%_*U)&-I+goiGwrWi!T6w8?SDs2s~3)m zmkq$npN=bpMc{IFE0xdgq_96^?1%1=r_KnrlyFeJ1L7WSYh_E-+Z}XL?@n@oEMR`* zoFAx+s%LFcK3@Hae<(qnb?~7Nd_l6@io&wG9Kx~=U6Pv*P?{f^H)2-Vt z`FE5nAGNi$&kw~YMk3dof)#@y@6N2_jD?wJ?lxcbyeos6wv9NNL>1RIJo{rv0I_HK z5=7=X?ly`Uan2OKlQbG(ZY=dGb;92(Qft%Z)!4VRw$0+or!_WClD8)ccvh2c0FO8K zr)IGtg}!b2necg*M_jC3-xvB^!(SVCHb?OG18*jn1Pl>JkFbK}F@s4c@JS>`Lv_)d zUhZa%{V&4aJDlr1{vR%tY!VV78nX9ZAw;NT&+NVTPBz(lhlA|wO&NV;CVM9{BYTVc z`S$(Z_x=0pzOU=J&UKwj$2s2b&+GMkK4!kmlily=Ka+F}e69-9*fQFiZ=> zm;QzG7pQ|=2$mM*cWX%qeBaqxrh&}rnlfL(`q~+WV!C@f@KVB^9&6_vo zXh~47^0#wY)tV3dTAT6~*0po!fSTlTMVKdlLy9uA&vz0I-zsVx!p`G9jf$WYnixLZ zcWYue*Pu5z0DA4EdVHvpX;`6nA2+?hNP`-ex;m522tP$y4Oy#Z$?#Y6cQ3`B89Okk z)D{U%a%&H{gst3^FMAHjxd&o|HL4J~+my~4Py3veU;2qx&{GqlV*Cj-yyo-t%2gH1 zK#sz@1yv9q4jv3TfSPPcjPZ5D3Gifb3@|EMEao~*RG2;XjClvEEM-Cm6&37xb37RQ z{%Dc3e)u0rS9$4l_ky*vH#=7hZ+3gwYt6+-{b0ln`C{nP96uL(&fT9mZ{VV9A%5`t z?plgxO!ctMQo%B#_l7cySK$(Fou^ls<{pnqU6@&6GQbJ<|4~NY(&JJwt%Wq(8YInj zV5r{*B|3{QgU|?)WvUCv(f^cz^iMr=yrC_ajBSs$<}9nQXP>htEf})6-$j%e5PA=R zAGdG!j{NSP%|5;-H+TdEqFo42VTb6FL8%@H*=k2co6ZZJw_vGBCywat?|;KcGAVpj z$Z35n83h9P)av+gQ)}Se^!j;hA2EHe~=_^((uzqM9Zp$n2Pj^CoE@?A@ADy5?Me0YS&U^Vh2p&MD3B_QBR0C&_exI=FW z#NSU($yYU*QHX>B=BNjpavI)*dc&ujzRAhCx}@kbbO@FNK^k=+q$Q=KoY@oV7uam% zilEDJS;-7^*@?+$obQkRZZM)aa~(-0Y?(K7nm{z;z=<9foi>OZ$B4`!{1S77iSh64 zBUmyU^cGj2=0fqE_kIfp?r7ArT$2dM!EM+WCT3S^a(l}WRF|}9S)3fRys<8&LE(RT zRKmqJ)~9GcZPjfzocGnYsaxLoll;p!8nW31|L-~|BKd@0g4dS7`u@OLCRUjt1-^0a zxmIhE-YopGUxN4lZyUj5PJG#uZN0Vw`Y3$v(K0%gJ_Y%-Z*)roM5J9`K-5IpiYTW) zWPJ8x6)=k&VLNueV+;Y@K~Jce;HYMoU2^b9n4>uv5!^MpA9;qrDNQ>c=Gx} zwZAQ?_YENg$~$JC&!$vy2O93mVR5SS9Hi$*)=g$eFQpw2I*-rhfsGmYU}ON0R5>a> z=8SB3s-koqqtI_O(k}<0rv>`eYw!!46NgIL5c3V+u9wS$N? zcyP=P3hV0X9I0MDtk>9=2C9NGqyHw>J<~spJbRxDwK<-t71r?@o6Z(7r7-HSq-Hxl zks5D&OXpl3%>0;r?ZvDUfVj9NYkK`G3Mok)+FkE9veroA=Ohl%w{J>ebl?+&P_*uJ zNyMcq6rMBj7sSlPo|$~JVcq4ZF5(WdmhekiOOp-U z1={$!vr<>ju>L0vjqE?O-+-@naQNPfhe=?WiaE=V|b>be>Qy*C^_3 zs-JK;c;BUC5V8VOWB8L&H=}YyS)@Br#NGSry`Z2-+)S; zA&%8cn$9u9esAUVrPm|EW8p#Z!Ga8(gO;(AQ#?eQ`ofd7YCCL!p626j=#U|f5Q5Qq z3Lc$+8JjRPwbnH67js4E0OTUn)V&3|T292pK9bg6cQ9doycH-i<@>s76Dhf@j}X}~ zeYv)JT5{fS8BiiJ(CikZ+~&BzC~YBE!8EkGVPqfvQ@fhkG>m&eApOXy#;|VSNR<9d zh?Qx{#H1Q(WQnbVyAIesqxgw+L3!qzB>}pss%}@>=BWuNY43?Ved2ayI@XNjp9Hht zfdXx75*Ma%$W>4Z2>s_e!}hL$0G=sh?;GvZjy~fbns=IRwQ`mLZ6w|fM^*Lnk=p~5 zT0B8)vLAF~3!g-qR&wiP<O!zWcPz;{r#o zV)ePcxPG;5b-5Vp=&@KcMRhLq+o5nwA!RI=4`B}SDIwgBD*}qRSpIhIfQ{wlZ$<`= zYQY{t9CSoc$#*zqZ>#0p0oWZ*glmuf5e!1l!102p>@>Pf0@x7~UaAV1Zji>)*$ZN5 zeoTJ)_=}YA&MjcS|Fuk>3nJ1t`doL!edD%w%G`>uU7d zW4ZeZ>xXZ4k;Iw~!6QXS^nbdn>nx@;;qa3S7-OXt~{)# z&fEO=vYDf4M4p#JMNrPv*T#sGSCrA^X=_pzErhnU;w&cnvDS#E6K!oTrrX|0Jn!~@?!bgHYj-b)yCziEY@;Ah63}T^X+(NBO5`Wq2?=tfLr7Y-$ zBzq$@Pm^YzY2OWwLR*ePxgsLM!<`tv)N3KHR}TYRgpT4_h>RXq*|n$gA114N5n1kE z=@k}e#fAEVdSP_L=?v8!2wM^jMP1rD{b?!$~^nR_cVO}M}!q!F=Y3L7}73d zY##XCrSV~VXD^pk6Z^2Ko%x4D*cDwgzq`vA#5K<)AAOI`{ST`)Ts$-Jp^E0yQQWh1 zjIuAhy0bQ?_A3Rhf9rNP8b@%H&BDGsax3KRsJ9+Z-c+3U6IKpQAeWIX*td&A1aLSqIFC3#hGn{O@SjhCUE%fF2QKV%r8QZ-V>Z zx=$Rr7#8yTCPq06`6uP&u-5DPPNcgXmW@4pMK@_y%cJ>D?m^~ zOKe%rp=mnn`S*-7OvJMZRS698l8CGTdF6`S^}vnsw5_Bm#CRl9HVqO zy@cqZjaG^RpLlaB8g4qTt*tQ@xkJ_~!$P#&8M48o5*G6HmqSTdDf^>l3Ipk)@9FG? z7wAngid@cMc4J2#^)aXtT}Fa~;hUQ-6`#9DLqkMJtU~n}AO9l!31L9aZb0j3@8+eK z2szpsl#?6&JYsRh)YsQP161H|xJC$Vj5Jd0zSY6vWjj5q11u&&_H5wTzkucUR(af0 zql?|se$D?XDLyX*8SA_blQymI8W#_+fYP;=iMw30#niQuzM#cAb$7@z z&712fNTbz&KI!^O##BD<$EkM{ACe?Te@-2XFZO4peY7`OO?fZP!Zh#4tUI1kVdiMc zLN(>!N3Rc0ZhB-rFpwZ@(*Hv#W<``juSBnCJ76>L!zRvmLwS2eVD_T=eI$ii|7f*{ z&fvruBUAWjHFtu|lGl8_0<}HC_d`XDG9{@9%0*yOs6n`M5w7ihx>2)>oG^%$lH7@A znhDA@1n~4rj3&8(!lJ#bp+;0hCBW=jvdkA0mi&hB8V}Srd-#NIgMl*%8T}g94O!MY z?IgUTUEUMu`9r^0F{T`BQ+2BGRr7}2g@XVCX)&PrcE3DfY1;d#96DEQ()<_s&k#BK zijTy2H;AVIDT8G|7Vs6BO`u|q#$r-?y8++(s7>%}tIcD}mudmRnzeQ(l9KMu98}kA z17PGfZRousPBk&IyyEA;-wS7)yl{rju0VG=V*PQ2cY8GIFr-88EaGBYq)TF;!;>a^;3-XDb~L*(7LT5NDCAGpunx7Bb+#eg_pM ze0;uXGT8;tn9>TYi>&p1d0u{rC$<=+B9JJCMLlN9yK74x$T3LRM`Wy0V)VZnCR|-I zVX@``s3O4BtuAz|oVtmkGrz`HoW9r~9M3e!$(f`7Rv}v~j9y7eSDH!hEuq5$aP-nl z5acq?g?kMrv8pcb8^?{Ew_JT*S-PsMs(@{v7!7)md-v{X-eYVusU1r3bzaS_J81FU zuPX_=77IXdiqIgB>ddOmWAzp*re}GaplC9>!v$k$!#oV6h?YQW0Mn5cc+ifV1nn-^ zrEvR1F3#}6N<<(Z37P;K{SbGirsw|fq}ztuwDcq(aJ46^g~x?0Ck!)N0Xz;K$L?~O zuYQ5yR}hprtNX92CBp0d2p9}1#+BcdyAq0UbB~N@+4nBOy60J20$pYd@<&$@BtF}8 zB3=WVCu8@6v7j&Pd^8_;VWxsifxZ2HAY>6sQ7vgIS+kxXDbs4=rcQqs87RT17a$ez>_kl6uRM6j9v4R0EqYYVKo*Qxn8Gm!GP z)~4GXq&Mg@V`#={UB)~BFEraOsV1Buqic|zzySjrzSq_@O4o%bxr94-fyanK#^>5i zr<2#!65vVl>9M-!65NM%9c`bAXp92mXYjd0F4~u+{=tvJWm6b>R#JtAuC!L)%t;bV z?c+Fm@g52Ik3Idz`(jOQ$@zG?`+ zk1G~b{7Jd+Y_;@!wO&^imVS5@AlSj&BZ11QftEv{Q|$xI-K&Aq)rA4%|W!@Pm30=nkiPN#6N5*qxEY+&l@gh`JX$ z7ww7-d3+P9E$vr+WGcJQZKS^yJl~F>g(O7|jzPClXsF73#k*#H0of=SAn^mRw9&YC zGgiR^-r7avwTu6@ji@vbHd{k~wiFmwWak^H{4iSpZ*v4P5Kiyg{pr4z^s~J2A}M!# zBHbaYJadH>Yjnd3!=5jj?c%S~e_~5XnA=XI0m?1U9}=@8wGjqZ`V<;kXioTd68?Wx z$}9Ue{saOa$dRNmorZuz#;B4AjwCBxHX~{3ExuSe5oX0ehK6{^ zqpIhkhLBoEk?IzwvnpyA;bqh}`ud>!%|zNw_O+{BeN~LsVbYT`iIqm5%h|%()znim z>i(z|Pdjy18~ozBm;pOmA`JO{W1asWE?~h;*Zz5!T_|~`Y_1_I3S59%+o|w^$C6vs z`hejlCq;aA2&o_Q+E?+8V<>HmHWr+}o8n|2)}hy~$3}Tri@MC*y6nG2hG66%7`ZM_ zftrI+Mo8pr4&S-)r3bvseQP=wZrVT&5=Nj#J^SDV1CQbnv6Plr5OMM_RRY!28bvMn9{AbK z{l}7N|Mf1Wd!HX07mQmD!jFi0ZqT7>IjLRadFXb&_wphZvca~pgez*X z(3w{G@vv8|{c%|LzJwKJpXM&HrQ&4l@z&QI8&Unir=j4QdGYUJecCmZ(OB*9;-%7l zvva@G!6O4&(biv4%cyO9yG(`n(dCx&*xMxaB%W<@(Eg z^nG`?^)eI*6DcuXU>*2n5vfs}p^o=blCj-dam`MVu~29>WLln1AaY@3n;SRI*jVK* zZAW(d60CsZe_oeo6e3tf03Yo7j2ZY=k9qRhUoZ@RFNk*s>}DG|!l3iQS*a94gb^4t z)p0}(X{cG)J$BXUH(*CXbe1c?wp$S2a7&4v7;#@;<^&}rCEd_+4h%z#_q&zOMri_7 zPVP%p*MCfux&kY-cG;4i#8F=u}I82sLnK1u1 zDU+h-vfp_0Ukpl9+`k96om-CVCjJ}fJ<)oCMi%h5!`zSU%&WDGsn=;Ccw;3%^T$eE z=BN7lDXnYW-fB33T>c#Tgnd&hAlkWp;+a^!esbDm%TTt*y;?6Hi6KgJMRDV2qfKu9 zDv?sS^ug`Qwr5ojZ7pNpF)Ow6WY1^JVVAjeDHDB=OdLu;uRWCryKHWml^s(yLZccW>-tnMu6mQLcyH|#e90VkM9~PP(`>A#z4_cyB z4J7s#`?hU~mIBUy7l3h?%L-*qKCD#;L5?oCIR@e^wZg0#YO0$S9iuWhrcR<~0JIq?VE?f8OG{Zr* zkv(vqB$V$_$4lP1>wUtf7D$q3I!u5)*y0B-maj%$XAkt%C4GX;SU9UeqntHOV*2Q` z!ijumKk) zhSGWc_=J3w7_&epFWi{UN@*QwRA((@i(!_=Q6zZ zzpBI&>vnX_!dhvldocCBGPwXLqvI*4Ej*K)CP)4Qd!3<4?O64yknB67m3$+WC!cWQ z!n=|l#TkeA9&He?MYr$YI`bRY|9Zw3G!ymU!2oWKMHkWc0@EFOIp6xh@OxigMkaEL zMs^m^Yw>?YmGFF?-?Um^PQ(5=?aZ#xv)YzqHX~t&X^a4wBTRsWFPcl^zEb8H=z;>Vef{g+Y(< zMag|k>Ls-zg~gtpp5%nl8B^hdF?~)lT%5Kq)f=v+2bPo7SM7;eR0{r0^-v_ zQl0$YAUuEjw=Kw#bsRtwL)@~X4vc#!?=uuOTGACnt5mzw#V}eK=KB_a z&dftnF(+AAvj=lC6l19obY4H2!}qQ=MN0xGv9Po`x#gs=D8b(-KBB1~jpC<5(`X}MuH3GMazcnUM)M=)V)rnz?$3V7(7sV2_;n@UV4f_w|<>)-0f4WMkA5N|e z{`>Ix#+AX6H?QtO8v~B9LSB|kk9y$tZ(1|j;6D%if&oz$x6dcc9NRK0_^sGlq~Jua zJ27^&b$j?*P}I}Wu#(Zm5zg+$_{9>`=42>X<1+#Sz!;{Lh;Hc|8!hAeZ@@KpGXf<}>`YG{QuU_bSPCQ_XZBphj-utBQl87~C;xf= z>yN`qV%9R-AuU!+lI;to)EK_LMWCs}X76ft6JI}<WN8we_hwd%13k>n}VH z{M+UnXQFeNG^!a-TpGTNyEVX&gov8jA#Cc8;E3yoQ4`KjiuA}ni{wTl(8@zX%o&jS z??SMaeFusJ616VAHq7iZQ+l6m6W?HqQvmI#*XE&D1cIH)k521eT8;%_Xm}VW)60PC zf#|H@-nK=aYC&i#NYMadA>ov3t4`sl-=O#-O`FzUqSde@R86=+u>&;xok~S2 zD0U$XE^id27w1zHJrtcMLY}oJ=K;)vL(c=Jy~S94%4Ahhgqj+*Tv{)wl|)= zY1q0k|8GWciz9OCc*Wx7&Um~?9ZtWyu>?zz{O9AH zS4hYDBlqhn2D06svr9&^F;4#Xu{>7EhhK?o=u;w=j497skxBF=44 zcM_rCvorXr52a$gbHi3uyX+{e>m$y6bP3(t$)k+{@G-un>awvj)}PTgUk4me#>H{Q zE}+&?2-UVA4dFM@YkVF@tYH01If7?jQE-1LWoGX&qP)T}B&a~5k%C;EHYPeoiZ#-n z=A!23pjP5+FHd922b2)D0Nr-#;Vsgkr`-gSHhu8QV^eogo- zVz8Y}jECwAAp&=^I>c~H;&}!JPSS~|Mj&(ahN_G^g6FBrf_eBFgL^C45zHq?Zq+X{ z($kmDcKer+z6H*Qb%`Bl$(PYMFm*#J&P5wtUTXO-E(DFRhTZJ@q8)aK786N3&UG|- zf(ojQh3QLXPLMX;g$DU@p0d4}oXaQh3pwELfDj*SAW{4fqbQyV8_KI6|`oF5(^ z)U@ETmhu~2PQFYj@jCME4RElxE-WlGZb;`SK!H|~X`L9O1X@xXYJU48^W3LJN#=916?gkmd%z-^@Q`vs zUad`z$@W>-C#BW!?%&`QEp` zs4Y-IrgVA}&nrC32=qSE<#g7FCEh4kO)NG`d`!kK+S>k{@W<(ZSb38*abUhVf4<9p>S24^iIRKlXh7x!g0_u%$z8UQt8pb_FCAP=d-ESf;6AyO*$;cnDMN zyTLRBVh5H6bt@6zK{7P<>b*p+t)DA6^=6oLiS2wK{zJwninJ} zku-WaI1HKkGW%(STdFamXN<1h;AFh>-XTk3PWC^m1<1z9GpxfdPMM~^zq0Kx^o(>T z{-@qkCR?=T=+UA(lbrcf>Hm2-U@>IeB*$qBRKeGC`RA84|14S{__5PRclA5u3mY_U zMnNJg?N8Orc`HVqM6*!jsQ3T;T)^3b)v?McRdMLx$5AuSCHC)}6Y~@0UuQj~1Qc{k#ml%$vxv9DDsWA*7W1qe;YN=XP;^daa_gLl2jR z--ZcvpMcm*RGdX8LwST z7e4#bXSXEGj$T=?L&zy-e$F_F%NyHd;ha`?EQwuNH~Qj9FNF-3$mAm!WL6&e2PlwR zw-lqVH*S=nC^fcJSxF5}4>q)zMpaNjNhB@f(RvO{U1lxx;K)*-874pBNBAE89>z96!3$TdFfb?;LU z?LqKjNI2Zog>lNtGxVA`atT{<-SLtVXnt$u`i=!})sGk(h`YPs7z@Y20O{Ob&ekF5^B&KCRtLvoPI^@}l zUqFq`8x)ruIBd9qnNj8WgCh67+))>6%>5Pk`y2A zEW|-S#Tb8`r5@tP9io!36-=-uM z@86{n1nY)wPe*GK$TAe^{S`uFAN!gn_l@$S)En0RB6%;TQG=O4;(CPwA@6y z>V-5dALg&mVW8(XK2}B2t}v# zhEZcX|D|V0srSFNCpkYHr!i&Ck^@YWk#`6<51Sb`m4c?WnHp=VnrWw$ikKQkYzVI0 zpQaLH%(c(T(k?A6F-u$tJ@{G)r=ybnUu6|kM#xCzv2)|0$D~_Ji0cHEz10F`ib&q$ zoDs?&oc!UxCY@`r=L)JS8FlhPY2pGYzm_O&yZDj@P;rL2n6b?nqmuQtnTxVYP&N*_-$Cs@*l>r+}KuD9PQJI(aXA-b>LF9nwPzz5oIl9 z4v|K+#oW&qEQ^QY4+HAi4qvGYp z8WoPl5mV=n$DSrDs$pdDYADu517phENYD2+77POmRy4sICU*Cp_>DbX9Y#Cu@d_FD zRb;qp(cojr^||;U?GHaXwsrj8Y1W5hIEtnu1ES7nq(b`TB6KXFoF*4fo8;yM+f5&;n*2L^=te5pkw(uzwOevX2?>L>cBsJ_;d+k<|xmBIyI9k6$0e87I;(M#n?BR zGX-%C0P7rs1f%ZPHb7*>8uxd>dTbw_!XMF+HSh#^SD<}iJi65LXH0rK;(xka2a)%4 z5kNkUI2pqkdq`ive@!2^%e>>*m;n1yiETlI2c+QayQI3 zYpJ}?Y>}j8fwi4u4=CrYwis^(2197m{T94GKC-52!uKYWR2hR-=+2uI#S9Ov&nSBZ+gnm| zy;~A4vQqh8h@+(R-w2j*yi&R;-tQXUXVSMur7#HouHKU?3Oj2LOvooQNP2ACuaG;2 zrJFo9=CVcjNczJ=$|jMG`=j5|?&m1{^$*_&$E1^+JmR7JKc`+m2KO8%W^{kz9ms&! z{n<;p6E7Xjm-lvsnC_@+bP=m)euZ`PYYETd7|E1nRFFU#N!Gwh!H7U790o|3@g-x3 z91wesz>Y4G2=7EiFrHK)+_U4xVNEJbMG5PZHP)gG+izJrnQEq52SL|G%(4bKam4P2 ze-~l<$r~`??Ow;hc2=H43JSYs&o^e7Cp?`W#309$k5gQabK^V(a=^*T$?;{%=0IOM z+;;)GA7fiN<$(6CFUel3-%WK&*>Re*>n`6q_PC7ig)zIiB6Zkgvj;B7ixebW_UW#C zWHoU0ot;5%2Hs%QILIqjUh882hOQwHNypSJ@FX?3XAVo(Y;TQGn9fx3I zt3vL!q!l>D|BKr!%uy~IzzEY72X(CF`#Dw&eIUn-Zy5;Hq_ z=Z^3^n*Bw;4|}-v;~SIE#YVi!H#Sur_Q5pP9tj=Z5*E}Bp&Pm zExQ$Dxi|53_1d4$4F*=AWw?ua9Uf$I+xzr1f5SQbY$i9DJf~W`e_QFGeA;tqNr36c zk~Yik6*(C$R5aqO37Q*83SAj$emucc>b#v|8c?SCnv?7i$(r|QFzm4JXHSY}?f9FW z-2S}K>4GHxF30h3M1{C}J`H;Nfe0@LCEcEC#JYiS;Ag(*iXc2j4wK~3NHhj3o9RIq|;Jb19 z$3&s$#cr|~gXH&#lI|1EhC2g_O@7vl;WaDWYpEv6%Rw^kW1ZHT#DJ}9)iECx+V$JT z&D~CaxKJ?{Pocl>ty1@jjFbw$HX58p52Q88AAIcZW%%?+X6w{ZMBhUHgC*xWe;w@A zvbP$nbaRY5bMk%&rbF-#`HX7UzAI~<&|mb1KVR}r@9JV-$<@lcgv-bPF3bN!STS;| z@hE6QTJHoy_cvkRda$T4Z}hA)t%lmdcLXkrZfmvpXPl`|KqLEs$V+H!Q*I-<)oBrD zWZ2VWdAZ~9r^ve4rLg&W4qBV3ktMT|*2mAy?Gqb}Xh%8GXQSuUF{n+JI7sHVYj6Kk zr;n@~a=SSWnW{*@Kk|hb+b;V_`@nwaI2`~5NF8@{dWMFE-j{BSR|F!_e{0oDz+NxE zM}7PDt==r*9MTEmAh0Q9e0*G@EfEos@>r7IMi6b@@Px*NN0)6_X`3Fj5EJ!u*SEHU zm=CIzOg2p%wVWqXYD7MQg7hL1u@`As$INanBc~ctT(F$>%voO8dg-?L0D0&Q_5}J# zW^~{FWpO%O^V{ensSO=*{C^+e$)$Ls>Js~ZTI=M;k{xuiutO2R8%$Fx}st>cuaqby) z=qOy@w`t)SRBu<|0)c-kN-k5RQ*_X?-_L2wh314<-cG}evSNJH(%6m5V zr`KpVOorQBUE*i?Bh{GjVqKM zTcXRG8($U*Ve@(}6AF?UwVPAsbvgL^>fbFEaTc zvSb9i>(CoGuGb~SejizFDL~}_mgt|-wA<&0Eg6WvrleB*2dAYduY8cUfYZ8i;jpk} zuxkFnX3xaYXwD4h(ek&u&^;eJnZf z?@g#{;6oMZFTy*8Kd^w8^_^lYSDMW^NX9nAt?F)HQ5=WC<4x3^pxbim54p>4>afmq z7vCX|^+*nh`xM8qz5`& z(3XTPo`}mh8B36B;}aT8`h|NfHV^N0x2&F$%nCFS-4PK!lNZ==dYbS0WU((QO6rik z_4)(Zw-b+#;rMw*Ia&FcU2SJvjn$zs>bg%<@lnjYd9z71+nPRe@BhN}@L6~o9-Wpa zABWCmRv`8*x!pG*DD8epUQlE!>ATSA(7Pq42i)CH$US7}x)#0{-2K0KZxf}nwENEu z?lX5D2B0a9?i!+6gxnS_hOPc(XC9V`E!}T<^C-2B9m_ue5wNu4BH$_#Z+HrC5IqG8 ziI`?+PCN$&aeTkXg$JT%HOznvj>~pGTH$R>O0T0iSDgyW6i`_ZLQW;XGC#ZJ0Dr#R zz(7Ln{B;!2SR~09(KXHtTJny7g7nV4-Bi#)Yq1=HM#-OOh!coyO=8;culF1K>YDK`(^a> zPq$LO>JhVdyfx3bH#YK0D_AwIP8w<%ovq%N9Zpt7kxK5~TvGS%8M$!Qu#p_Ss-Zp> zMER%wQFeo6*Rn>l*ph-nMBrGbmEMB}wqE|6by50`Kgt@rSi?Jv6;hiWR5Ao|>ooBb zzB~kc=kNVZ{w#jF()*r^Wz6%2ItyPH!zB8i)-pjhPZy|903;+Bb$X~rRw!p0g4_4S=QuB@71D=j_ipuIkO3=y<5E9jz|94`K2bd9J$)3jfZgC7=?zTJbt3 zk?(m9r+8B`$|4qKNWg+tWR zsan^I{O;awTqRigJ<^OL&Xd{|9&h>PzMQt*YCI%h}q-J&IfSx zm@lS`#!4M&578$!4t6!zM#kz~O?$ClPBZuanP7x0P5eq0RR@k&?oPosy)i0u~}> zB;vnLO}5|Xv_L|s1Cl5K*ss9^z9Dkr@R^e(b^OWeCsb+#eM4OMSFLe0l7@QA5nr>$ zpOSHIxq^zo{65N#BBB9{C4bhGI94h3Vz+6x!MfA0x-0yclG3ap!v*V!DyAXyp{7Zy z$22*u;zam@ezZhiGw#u{mQzeu3_W^}>!9!lO=~nb`o0VrtZ%PAezF@j>uU*D+-M22F0LY~Z#V7n5otBJ^N5N4vGKBrM*CxqzYW8nUKKLHT zpVE~Rx$@X~-Q&-;dWv%&sukVTa^^%k2TE+ja9Rx5zatVk6oC0!Y6p5sL{}B^V;Q=2 ztdT^x@5p%$alF;85meA{f)Izxpw_7>Tss9fCH6VAZn_PWYB>%oH7+t)+d6s{zAhcT zBS;c~YS6Y~B+=MIQotpsW@Q(8WH8F;HyZQ=UaAVbl(clWdjHYWj(>{_c1R^Iyhd#|s{?0m(^I2eywkr7R;$Tt(y!S*xb3e|{i1oI zj*n^{y+gxi%YJ3w%t!Cw{;(Zc#p`cH33Q7Qte#tU5N;@@A?UGlruza~13^kuIq|Sv zvPU7IQMPt_HJPgNHA4*-k4$9@`_o)ZvXds=v+a(yB;3L&v%99w;M4sb<@sraW0E^x zo+>jLKFQ1ueymL{+qVXz+exXP4RsJKov=Ej@zVV+?h2Q%TTn`V-=(LHyDpYwx6LMV zld1zlVvF`0X1HaFKy}pa2g5)0@`=21ZLbe*RX%91Bx189 zYH8*Zz8x;5@q5$P(5A+MOLq7^3Wv`lMcdOWLrTBv|@#U$rk(%dBB(YfywT|?w(&+$?3i)qE`Y1ab zp!Ed0QwlA{7wz^A>@=HS(WlK{cfk2Pj295j?$bo}!mY5&BI0HN3&EqRMb0Z`arZjb zsm&HS!z6E1yv|J&XkYM|#FVNbklUb%9t*bW{8L3ho>Wd8w}a!~bDPLMkOWs(a$W!f zRde{rg!ehsY#NlIx`vFOWsh~~Ess@z*L@^mklo3uq^sJ}tU~Hj59L?JMGb1%X-r+o z=LDmY7`Msm36>O+Eo)-hpX*JAC`j>EWUh{N(8v9pkHr@WxP>$S;)DKCPTC(Cc#06O zl66IK9fr&-nPMBGw5%@MSXzt;UE0*UKX{z)8$}A&aFfBBnfM4$u!n*Ee-zs|Sut)? zOJdDFrt<9`sBd~P&~ddy-EwHOiMa2vTZ|QC@hMzB7>S@(+4${*PX&XbKAGW zEgvOM$vrHiko&(Fdkdg0+wJ{V1(B3S5JZqp=?39Tr?fPZ(kMuWlG5EB0@9L7iL|sx zgP=%DgOqvJ!`}aMe&?M3oH=`R_RKpv@7_GmbFX!;bzPrJaLP}#hkJ^*?e3(V0J{|% zYY_$ZEEB*6l+j2#vv}WHe6&aZ$d4AnVWu)B*yyO0E{t+p{xBQ=}$+;+Oidyy6We+;?52lLt|Y3C^B z&^&w*);O1O?f7x-n`XnHU&8hJ%vSMC${?VSBm)73x!ndj7BQWyWGcYllTd&R&)%qHl3Ub8au(Tk?ebO5atYpDrabyz5|Kr3zLzxu%6 zYItxMD5pJ_FeGwIiv`EGr zrnq;JOPA4`nT4QC9!YwEB=@(~Ab$y}_PgZhp&$NH2O69`Phk~mLQ&1@YBMJ=K_g?M z?q)!FyZ&~E{Vz;A*m)9F2lGB)&wGLS5%>M%4y#vkep$qXK{D@P*?Ycg2lG+|^x<2D z=$Ou*xl74@^XMeR-;Wt9AFe&r9<5JS{Z-ZT*mK7D3IPi93FyN)ex88{ZB~co z2(&MHlRrcIySuv!VoSztVQt#Dw(5UAXu5`w$P4^0-*8pBNqEX(Wf9&&}Dx?Ea{j#ybHe-~188~ts{m}>Sh zL8YNd3$FPS7Gvdb#^%hQe(=a-7=M~^^okQa@Oqq=>QAnIy-bBWXXG{W&gn(f7J6YJ z{2DGZ?6DnaqMIlier(;QOCxwh4x(%j8r4iV1;#(YdgP=ny`RqU>vWaHP=iygfz^z< zvm$l<1=k-l^Rdcc3qx*g)8*9k>}agYQm2yK!# zR}nHQr7;tqaIz*bztTUnEY<4iEqGPCGRaOoFRva+Fqv@n^xwe^&P)57m*r-$B^Ro+ zv~5|-6rz=QvlQbgwzwp&$n2nV`vntlr*@|IQD4RF=!n!y(j4A@s;dgqm+uob+%>N+ zh2#=wz-~MF&*zN#MK&~T+Hu`o?H!T@VfS|@o|VI*>p}VdDu>qt+Z1SA-*X670Xt+z zf7E-bAod@kF@bo)g^bi+NX&@N7A51Po!eNSxUtcXUq7eYN5HRdz$E%{6NzC()Me~_ zk2AB7%d!0@36_s;ev2z3LqoU_)IbaNRdzCNRbp7tzvzOgOAW}a&1>{-zcC7HfGL(F zjbxJ%MvBB^;-Es~$%$QQn(-z>gVN>D+jEW|<~@ueJItS=^H3J6=FPKWagv_89=h1Y zDZP)tleSL{HC~W(D@af}cJi2Vx;0h4S{+ce6>LEazo5noKMQs-OYm-ZK5O%W_jboJ z4q^Bneaf6Fjo6)~Aw1hSxniN5CW1A8reNDyM)yaUyp72UPuYIHhQ_hSIe3NyCb63L zwJKCJ99KFT$GG>u^}MG+WxW+ZC9Oz?CSn{^mTldnC1sR}!)i!mwK}x?keDxbl`;~9 zNb0s?jj7McXp0#RHWwPa=2d)~V@V93TPvRh4rn4yBv>{hLSgxXuXG?3-U6X;3!p`h z(yR#Nm%MMrEWasZP1G2N_ra|sureIA#3Y&J-t(nneiSdU2|L~R-CZLp-LeHazw!15 z*9HTp*7~r?4O;Ndm3AMxq0)HLdG@1D8gTJc%73le8i=kvs?B0yp`q&)!5U=dgJol8 zZa~ydy?ttpj~Ba3*Ds{l%k7w0Fd(!vB!Y~f<}$h?FGf(+VxA>#zngwsjtJ(xremyw zh=p094RvP=wY6C8kq;y^fBDg^>EjmJx=5(u-qjltJ`VE|Yi;Zp!;gb`*BL*cU2X~h z=(niSh3Bgd1mMkZrzflL^8Ya=-vlYAQxoRCc6JdU3j1fWe+EL3uu(xvLDt)a+N@82 zr@`szJ6&aL|8wyjc(jf_rM^q@ZhQCD9+`rlL$n8Yk&Oq7Bk<&`Q$q;Yre^JaFZ%2O zXv1mA=gTkZH(&fkui;aJ0uEXCJ^HqXbhc3p!-O`{riXALf)Lo0vIt~@CS&+e8x`HIu~t8La? zb*0WP4mpmg7;^mrHFpWD3wdsIY73Z{1}n6Zc22DNYKKYbmRd1uGnP)dGS8#>-UkK5 z*Xv ztiT}fIcKHDg8`T?amfX8JIn)Z5yp*D$#QI2={3dsUm48%K=v=AXjbux)qGuY>zR;i z717eA@5^%h+Cfy~E8F@5B+&HMk#_z-Rolz4$L@GCVPv-Mg%AvSO_rN9(8OWErePl+k$9AwEUw@C=f!Ss zrjziX?YoZs+&ig5Skk+WZT)vvL?>IS@pC!pVFZ*D~vSIZD&`&dNidJrP#*c1_K z*5hhua}bOi*oL);af^FXW9u=^BS&zM+P>f}DJf}!d=>-^%mmo%+rp-TxM;@H>)Dc= z4M93eX7i18uJ4vYDbJ9q04^~*L=}Ly5TosIJ)?M7^mziLPu(TJ%=`Fyp(3vIG2^Eg z58P5I*DIBz-6UP*;`cX|XcA9eI;0NTj%I_qW(>!E!B*Hz@l?PnV2aY^aUp5sxHvTo z8N;Kud-=AKYOlel$k1_^O?$9zh`Q~@GiE8inn*o6{6)!^UotraUJh&IYt0H2b!tH8 z%e;2oLV$Hf7j51NQWP{0Rk*n!>@&KbmruBmu*$!ha)l9OKukbDK+f`xe|*LbXcv}0&)mD0 z)&IG*7C^y<{AdD)6zm|H>I2B>;ATng2;hMP_YUl<+foqaK0MDyJ%a6i&OqVLxm|==Uewf?!MG!IG84 zy`L`g&+Y*e!N^eH{e=<$FhmbhXYIR$v>1Xn&{W4g_dBj#4I83GrHHVEk+lGLT}+bLHkM8i1U~PZ*Mu?DZLr#gh1ga>ftC6zuQgJ{vEY0HWbi*hI2-bQh za348YMOE@?`6BaEb_;W-iN+r^X(=ZI1IvAG9=up%w%g@Dh|; zk9kGCzC~AD6jr*zx=7boo-;c*kRHVERSV&#Y>{FVE+@FMwUuW@;j(o6V8r1 z+Vffn-EMu@etZ2BN6Uhp0praV2J5~xfv|?%oLb918mou84kM+iIXiMTtQ{QMU~-I% z;)7j}_DsRnCA=f*>Pfp+2*g@L{CE-f1QAgox43xQ*Mm20JB~-q^33>U!NDa09wCUEdW|d?BDq)Q|chR zHX!EUdk1;P#NoK60apdt^iEu6y8umRT?u>+muX{<12`pez$rRtYyVI`MN3#~`uOS& z`J7edys7g&|LySNVp~Lym$Wvbh83+9Q}IBV9wCF#Purq8qYLIXHs~)s?>C5`y&sq?z;K>&D-L8bT*wJ zL`ErnS8HIQk$P|Syr%KV-3cE*+5}=IH1%pSX4rUgl)$kvMhql2f-Qfg!cM zK6}e93^vg*2^rKPgy=YfxvWM%-=!WN#9WTziYaBse%YUxuRu+i*j&a_sPsJ4^lG@x z6I3_Y&rpClV1ytD=wXXwG^|R?c<{dSz0!JJM`d+e>P~k(*vUz98$VI{#@ra${+-b} zN>-RuppsLOL0kQa@0IAzhs#&c!+E6Q!lHf!eATrYbbmb_avBmjDKM?J_@(WsLw=%F zSNg5dP2Zh^-WTBIadN=}*DnchgaHnGEeDc5%`?_otNmbWxrPYYkw8DD?5Cdsr>Vh5 zc!6?8@fmrzaHfc zso>D2ytKUuGL2SP$!Fi_g+v5+t)9cmiq$ ziJSfn?6H;1FTWWZ61&Pqo^J$eqxIdA^kcp~xZG)w#r0&ix*E21fc*Y;>@ED~$GGtEgq9*a*5xOG9W)`R0AGnwP|J3L+h0~B+r&NnQPvX-pbbvz zx>3D*_U|u4`w2+l?Rp0F!;dj7ug|;fn$0h&xvVe*>uHPHoG?8*WI{yLM+QkgV$|zs zKW9h%1=`Ou+xr1YRP6$Sj2iIOTBZJ~FADCX~r zo4Tph$HLBa(ftbCY+T5ifcV@l`KoYiTHOoA7GyFQ!5$3f9UL)@^G?g~+i~ABqaQI@i&+Ro$NB?6+n3 z{&(I>+YI<42zhKDX@JJ>-}77Cn&0V<71yx&*K#YtyO%eo1N1mnNOaobL5v4NVHNf` zNG}eb-dgRYbY4NuJ4DRP#zbAP7gHI>f?-Ji>0Ml&Sq7O&wf}E|;;9VwWP-J?n^rz+ z%QmDdu0<$oAI^7MDXXCvGY<#*6E^~4#EKo~gWNjQO)d1E24Mzkp?4y-xBu@+Gv22? zXWlq5>jw^l1!55dD)`vUraq;qVQM*taCNCM zsN}BQ``g(aT5RuKF78)GIx-Z`q#94fh{tY~prK`%)@^?ca|@o;$5o6ZQk)H`E6Cr6 z2{RA$EbuD8fH_GdIQIGYl$e4isi1TNzJJUEVb`VQF8`Scb+=l(h^&kvRKKmflf_cc z{U{QR1ka~UB!akEG)4A;SKRM)>cBNdaGm1x>-Vf`-H+IhI2VNFS+Xdt$JPSd5WP!S z^cM2z&D1@@zrZ*=#hd5CMLbgwG55gXtxE6A4EY6rF#F4Sw!gEIDX`%9|11Vqi5-{T z-#KNh$jG`1daI`Z59Q(KUk&EM>w3?p3Ipk0=ui9+Z_VvAoQI8f!zxcYXTTXNhCJ8{ zEGZyIMXR0fo+_LEBtoEv!r)*pQYzlav|zKG+B%kSiq?y{5RXAnbLKC?BAc{`_e8wm z;APkmjp-rKTYnunW~aYs^Y&wH>#^w@kNk#vcB@VTW07y9kMznLwC8s*rK%oBPI0l% zr*AF$-5Z$}IR7~9mi_-rNz9Nn%jej!+~P2sw*cR z(&;sFiBWH;ZeURvUx9>RYaq4R@Uc$MVZ;p`N0?0y1UU^vMF?e?Gi|(I+d262K{Bu>5r9N4_LNF2))(VV^%V2zduv<3Y?8b&P9bQ02 z`hYCsEyzRb7Lo9(-$3mHbdz-!3_KIt=m#XXE%>?4kq(9+$ z0q%n?gCTZF8B=!r6RQPL?z3tQwZwL`x<#&Egly_mD{H<~v(g4B{Qpw~KQ7*2gGX+O z4+_taLU5Cy&D(tq=8{*VA*W6Rpat3ppFNB^n5RGS4GjP`hW?4%zI}(ka~U}g1Xa;V zMWHLP0CugRGac{yj2?`f_4$LQPY6GmgVei;)nuU$llVP)@j{+tDAC2>y0C0u=v5)? z;3tW?8wl@o8{DvPUSqN_}$L{Owiv%>XbTZtf0Xg!Ef#j#+x;f;F(MbzAwU7G4K%mI90Pv&=9gO` z+rd_3AB(7VkG->zL{Q`bf@?|~8H$4$A$E3FU#W-)$R&@3RM^SrsAKEEpEM$}#g4Eq zj^Mdwiu;_rJ@ix5GpR@(Qp2QOv%Nr&KKu9IoAy6DYUElhs|Jy1gSA3S z!NhNYUElQY@2~Wn9MIN0YJ+(8`(BV-7v4wL!Z8K{aNpMxTW;3Y)@eC8jEVj1-X*D3 zg7Zm+DR!!Az>ddz2s_;Rbvi@gFIDv~Os4}>x{9u1)ufF?-Y>`c(`fkU_;5hgzMYc{ zBA`G73|6O^Mnx8nB$p*-@ECyLrY(U^^bd{6GsGu<5HEfvUFz?AQik`L;a9+1T8EV( z9`J3y9BL|*v<2VD+3aq$Cj=FZoOV5=GGh6Eu7<-2=oBSW{TAfL-T0Mmc>~R2p<;l;@66Iu-u^WAj>@ZT{A^VkQ#3@QUWW@Pq z+qKQa**-zBOHO)v)q`QGSo|sbU5YsjHGj!Dd*LQ+IRs&v$uPDXY$nYP4we?|gj5d12o6TFM0M;p{c zQSlFP=Np;hie%owlho28<*~bohOfaUrE;H;E~Ki7^y{Xn(70|y?LfR$O5ni5PwzWB zdu62QB*T~$QO)&YpTp){`4>h{n;WRIXtIZ-A}2rbRqE^OxKh&)U_KJm$>deb_e#M3vj-g zsCymk@19Qk?@XTo2UPIzyw$aD(IQs7cRBVRC&bS@$R_0}vO52%uB5?Cl0{%$s%>y` z-tjQ5riL2?|6x7m;OD52>rWmCH9z_kZSAHbcqM(LBvqMXFu9)n%Y?T(--N`-r}vGd z_M+eIuQDyHQK7L4Ga6l8Sf&-I%>?nQN}U>uPn^<|=Et{SJ=@nGUcBACma~i@hlENE z6gQ)>MKXvxe=IR6C7>6gH-OUJ?N$HS&>X})hGEj)anNXarRmhbKdN3B5iP5+3LaIjp-Q~=5}`2&njm&bR(dIEyPq%%tfBYF{muy{5Wf%^ zUV$t7($|ilk#^N_O9zK25NgzwY&n{$Kl`qgJAfFM5d!YIePUwb5vl<7BM5B5w;LTH zH8q=H?ZAI(%q9t*84#6Wg?!X09BTHE->A%i`DcavQWdF|rL%hSRJ z0>j?M$6s3GmRT{%i``n4eUjMYaPk(8Jbhsp7YP_yz zPJv5=clwRtu6^)xrC{(dltyWMSrx>|_@FtXJd#f95DK3WxYCi-8XH3XxY04;?(HI@ z%=|@W(5d-(VhQbh2nt^9#pG7!Q>9Z#qcLlIofyHI=p4dZuKl_Ob_Z9JJ6^Y;H*jfi zgqz}@_>8{@t=a+#1zotKVG^_YQ-}@uKJj6`ZbM8hjr@tYY!)HzaLb2JWcFPZ9EZm} zCa&AA)5T>u`MxL@d#EFhY~kGoD~{;-!cszl8Xyij%&X*Ktv>8S4pIdl1Xh>gC;he<%b_fBf&4u=& z`^t{pQ!bB>&)|Ohw5&cUa~XT+4sD)nm>t%YpK?uDMW^fG zFkquqOA=~asNmA4QI5urxQS=CvcvbnD~n*v+R!z*yS%dg*wMK&-A#v~ytuzqzbq+R zDNR5?!dHfzR*pzIRryh-R$aw|=}G2fpFLF3XB>}+;qqJC;*Eda7TdnmWC6ZMNtO#X zcs)w9K;wIZy+oZ^oteZFhNM{NS5#C~pevHPL%ZY#XCWAQ`t^&z1nJ!tW2LXw^7#4K zn1(y-_c^!l(XT+5gXq60w_EU0W@lf8xXO*&rE!~8-fg?+T;rR3Cp`JLySNkp=mM@` z9-hGI7CpC+9@A?NA${*D3_vNk5cuWJf8duKauUR{)wQNOk>cgpcdEOt-f6SA)rc3W za@Nwbg%h78MDg6aL6X}i&Q3Ubmf2!zFzHY}P2vm(HTI+czM{mIkK$}?d7J{+gB2Pd zsPf`OC7ZZH96883KZo8+OTuy4vfQ0P#b9-V2SBxkN3m2uwNk?&gv$-o{NG$z%|he8 z%bTB_H(XkFc$Bp>p%YD2x6-ef9~P~uLkh^AW=B}c%haPIbM9ekFns=>Lr|3(Qh!JF z5~=LK>{Uz{nj;zv;?6po)uEoJH%9n$ve|hrwy>{jYPf1@4!$wSB`~u@S+hkXKU+5M zvi*1BmMU>7pc0}Yj7#nfovF@O<{GV;emi7YGty^wu zEAahb(hBe*(lRoeTz1=xwJb}DP${mNUt|(WEPSc}pEmChAi)mFsWuH;r81<|3REwL z5K3qQt-ZZ7{@X{~1*aiABg9C=7Y~+h$$*BZ9jdf3%7PiZqfY2P>jo(vZ`iwl!&A_g zCAr4RYFX7Mn-TLoHR|u;!PSD7jae9yl?jDu;xd(4#bd7?3JiciGfnKV=+XvRL8-9} zk?2_TuxQ)s44ld?y?f(!6CWPs=EaFC6wTn88MvdhkapstC&KUu*_0B1HNxC2LDMb6 zyATt*gsAw&Z4IJs3R)Ev$YE9TpW5m#Ij69ov~VcQ{|+|*re*ZwvN9RP)iKBH=VW)v z81{)%M&FWta?2(lM9ExwHOElGW6PVRU--&pW8+LGE-mFwKe@kS(WK%7vXS~ESwMNQ z$2;qEIOukP(pCkOwyn+8%gHXGwAx#qZS$usk+e#aB0(FCb|o_tNt$`X?eCt5K}ClK z{q>_!YQP)-nGkHj$aX95H43^5fKpM3JjMJ^cL6ubnwpxcI>J5QKX3!f1CzT|V2_lT zwmTt3l%>wm$C!~uv@4g$n>YvsKhzw}2=khaU@qmNPqm}?7(r^7ybzZwz666}^{-0h zw%iI+f+L^)&eTnPZ1M^~I=vn~v~;B_=>?DG_PBb!U-P%mX+KTnwfbF+^NcIz9{IIA z1CRz%7Z~&xe23LRIO+roH#YbGq7^i2cagh;Z`>by7EL>XcSiSCxelJzwD?(&;++~t zkUn6&!sPx6b06)@V!d=o|0Ko0F!2R@w<{`bA@^B0o@FAHwHN>GLI<-X4nFN5U-^s0 zBx{n+7#6-&;VH67rSO4azIMWJ%L?Utkiz(^9mdJHx7cZS~{# z+_ zQT{~luc5iHFAyPidZ9C-xfy-sh>4apd7!zXMg08B^%ijF?%3RSAGmsD>~40zkR7P+ zmKfmh;7o=8^L;qga)pR^_4*5AzgQo!Vd-0j5Ut8d?(K@b5dGw(?b`(HM8L+a^HuWi zKR7O!`gi)+XrzUUars%>)lQZxOJE__Vk=@<#n&ib;o7pg*+6vu!(S}t(o(lTLH`1L z3oY-`Sf*kpUnPugm&A!G6bUU<1UD&Q@)9VW_-OSMSmrXX!T*>h0qGhZ*-DKKOf&C; zuMA3o1c^}`zGQ*hwwsQX+$|aDu4rZKm@{A( z<{QC0jrH5nJ;f?stpskmaGk{%H)@uL!UlvvXtBv}-O4_7bu&CVEgXC8-H2U-wc*X1 zwkB7kGhB(C?p-brh|VKhZ8OYWoJ3hYT%QbF?bu0ZyKn=vx2d{8DmE zLAga3h~q2ei_)s#$Llf-Sz~IQ^r6`M3cVCBoMKkO$-J4B)x+R4M7M!+=`|UXib`jC z`PVVu?)|0dK<0Ty`B3buc5tngO-|F0-~Yx0mNR=0?iJXID9t7@$7%6`{Q$$Qa&Z;K{Ai!^3z2@wSc5Ulxy!Z$!!> zp1+5?D|95-!dodVYclOrlvEISCyqyaz1h1B5o1DI27i-DtPQzpQBascA08hc&;LeV zzw*SJ-SzbqS@YwmqCW>jPutF$ESdHa!aXHqVyY69)q0!pP z>~>4UR=@GVCWF<-h}p-OK{|UV57UIw2RLcb?m|a|V($pz&(r`?jowKIFSm!QPqbX6 z3hgO+2v`M+D7JMkA#MsHOh&A}H^ep#Wj9c`oM5(@7$;%V(xw!C4Fm)X>O0-xiQuO3 zOAO_D;A=E?!-bs}7k8MZpIou`b@F6cR(k#n_9Fr9?0ONU47xMzxG;EZMu80Q*!-fn zE+gL25Z$u|Ik1D4l?_#hMBKP5bw`_JJa4U@{u)qnv3i?vdmFzb2DMBRw5_NO9(o;# zI~nyohEyn5qX8fD85WY8yVrO`;qnDYf}7DU-@=(RIa9NzRyUt|cO0=LV*?865lkZh zYQmHy3_aD7iZ3?Or~OA>PNqOll;!y?{5PQ5s*m(Uvey8SV+{u*bCI{mWp&O}knQ+9 zZ^DWK7RK&EOMw4vpd}S)XzYHmva&%`noEP2Q}qBNbI&$E;kJ>UXn==A4q5c7x%N4AFdzW1{9R z^J=HY_l7Ic-s#>9^6o#STK_V~cVGWT=d=Gp^Uvb8YQDzZ#QS?Cz!WITG9mB8WeD$X zTuw!%h}9IGLz)#-MN=ON9iTz@@$f4HszL|L^(d4pWP{Y0SAk0@P{YAWi$*zykHKYF zceYH*Ks8~FS%Q=XqCshN+I71MeX=f9*k+RSYu>rvc5CD%v^c0GSo;tb9P*NJz83Z7 zdOBU(N;k8KiW>uxOU$&)ii!`KCU$+uRVHs166_i%@IYLXZZmlVy%ONF;J6_pakt^S z2?ah(`o(ofA4WPS__d;{*H*ZTkGT@;SHaWzp#6B^#k;X{DcM@4cr zzgiDK5PHaJ&T9*RAz~03iME0+^J5e|vVtN-B+#B?nUO{ISmoZvL?@?&hQ}ewQ*80{ zw0wF?c1g`jy2gt_j`uIWI>a34>Zw^@5opWN@%vQ2mhgbI%BaNe-3s@Mz$zcGLJ_F$ zz86r9Xp+{B(3&V8er#QtMa%1T@110pjX6^djOui6Vo+_nJVGnfV2g)`5r?=WBONpQ zg1NM0zRmAn120W?9+-*k=;Ui$lI+CSgSo`X$H^xdS&4pbCRokutis6?)Ph zfG@xW9@39_(w~++!U*bk=g^Kc#N7rTD9Yi|0-zN{rv{=P96q!-7~o!n&V5C@z2Wu# zJA4_%+N>by$P!(hqa8DOAr+!GQbSBe*?@x8*NTfUQ?OU-S6kpD!tJ1IDZo2cgk!KM z;_LpF#i)z%w4`h4r?ROm)L!HBylJ%I!)l&J+1UEi)pAGucQMkMii3yw`95OZa_+>G zv(dIwELQlMENXA&G9J82d6X{hefO_=ppE&BnhEOG0EKozADuOwu$|?6^)>_gT>n7S z3G7$B%`kb>!2AUBo2@T$uym}@9MrT4;c}eslkTklD?1;^*2*Pe#R0kqsp>%RF^h zfTsYv&PDw@U3EFf6*-k!VaanJNO?F0?_{e%!5rsH^{u+7=uE-0)uQ$f3dH*@ z^?4HC=2sk5*BM8Da-oge)HQ2A5jcLwUq4f^R_r&;)$1+BzRqda6rVu0HN1R4cw0OG zba?=CgUdV||KBI!GOto3G(5D_f)y$sM(;tWy3@ObbL>-h&7jS|CX|c!T4@ScEAV1t zA6f=KcAzvdMaTR2(mRI$thkPt6(cNdma7{ZL?cAT zHnD?JnSHQw(`J=rP5w5ab-zfRnM+u>AClt$>DO^|bS`%8nBXg^Lt>; zz2x#BZKFC`m$uHNniobWKYs9SKAkI7X!&oUe)=>CuK6X0D|tuWZFvA8+=yA^B-sG9 ztcOIOSg0>`1hsv%PD9zJ*IU!IMzmfl(eIb>;>hpCqceSam5Qh!IfgZ>*SRT4l|mGY z@fQ@C#gyoqGHraXQ9J)*^TrJX7sTbVRoun}c%4`n=Nga;Ul-s9~8zs}?L|IOzn50`X2dBNgUm zfJJwKj!8X|XdITjrtN}l*Cqm=Rh|%6w4tA`5Vvtn2jtI#pTT(`um|QspS2mL)EY#D zaZ#XafMW!9F=UYl@GivEWWTCbDuLL?Y@-G);i34G9rc&OY<=Gc6v~TU+&xnTU0sjvN=#a1T7iHkwCZ z1MCR|kgq>~2J2Nu6MU%9W`*;OOI%#_f7Rg9z;qSH7cO_YwvG!vBF`bhjIYktsh)j0 zt21RGTCr<<8;^q{Q+D)|j=@&@WK>O4&hW1D=p8oAPx6k^2R~8tOllYzW?klu`a5F| zN!(jjO6PPclqRTnEV!3`hH*xevif**LO2tPY}1go;i@V6R2^af0L z@$cWRU(WukLDWwozA_cx`fI%uQiM5|O)9cG!ue_voj zM+H9%3M<6l0IR145sjiPNu#JjKkJI-2Pg|f; zfbn)uhtE&1EYSz36BV(5ySnJJ891%2+f)j5@8mAV_7T%$ z&B(DgWh(gM%r)!$+I6~T#PO)VUbUY~<$%1`?S2BX^_inLX_EXt^#p zK-EfWQa4K=?W0tC0L#o9lEpuN`U`_@=J${&L-`!Xieb9hx~=J_l%4+WG1uv`y1>U= zyT|5e5QtJ?{Z(|irEhQ29YH?6v)#;4V-`_j7Ks44bi_EwYhOJ$OpE3@_P&7Z`AOY5 zxCIR#VMJ85g^;V&8Y*o^DQ)dHfwcp4+M?80X4u>$H?5Ofi90I_-k7I>4=|m6gjoXq zc^}P-!4)BUg0ufIf3`ie1aw|6>|m0#sS!I6h_(~*sCQ4TtAjdxr^DO;MPPG!bFUUv zJScd*F|~N`?G0mfX1)Z>h z{A>c~|GmR!>E zepKv*aJ+OWxfu2$vqb__Mc|h`k2&KC`9YT=5(v-R+_Wt&DPbXt9F(|6fGMQIwR|A- zGc?1~R6KX|8WK0-U?OM#X+B!adi}STM}W#B7S5}(Js&sS#%4~w8cqHkk5>Ff$|>E! zDyZgg1$GgN4s;1Kp^lj(b@D7Q5t+v3`)>6;nJT^X#s1I2;m2zm)Peag2MxLZMVBTS zkayM(=_tgw7YJ@q2Eq=_^OD zp6UN|6Q~B@GGOcJO}tXi4+{A2JM)Q<@ z{{7u4BFc1dWiWDhlYW3jBMcXn()tYV(-XnV8LuF-_~O=F``y8W_ZfPk(#~q)EkxTp zp3mkQQKl9lXoarSCPaM&OmjM*O{#63HkcH(n&wQ&-=a1zg%?irwQ8dIF2QOi$+5xZ zI+J4Ek+_`J+s|m!=$kIUb{%iowi+EX4Ql9u_1Wsn9tym6EGO;!ti)HNic0I`uT|dW zxX?l58)f5|eTUJ#WE9C0IZT^;kUkNZCmz_C$fZ3D#sZY3Ga$#~`+5u1l5MYFzUeQj zkc3WKfCg!U!s*9EZFvSueNI71N=-K?8%3|Ad*5F7Srr%j>|phV^}Sti*OI@p+7~Y| zYDPxJv@{#{*$@9MGhNI@Y1!FU?(P?CXSn(P=dK{$ew}XG6W+o9u35`w8^xVJ?amMf zxi|eY%b6iIXW5B}{ODk{L^fCpcRQp3<>jsWIx=LzK!n8<%;I6q$pLOUBm+g%y9HSN zn|F88u+!~}wtjw&dK=|Dh6A0a4&KB`UY3vQeYuh&iG-TFx^kzUVbR-+0=1Om6<~J( ziAk{u$3p>D?Ex^+HN{GTiLMDubb{|BEp@ikmBwC?^$$y^v3R7(wY{YYrf|~P`{DVE zpfkx{7*&yBI}K)8mPu>ai5;QR0NRGh`>H==CxGdfWR0RHA3jU3B6mzth1Ze_P9g@s zwUCNv76q+Zn%?cT$D*;z5B_54*j1I*6NrzyVGgt%O^h@S%r)zNNOIkYe(5T~OOlZJ zLS(af1dcRdYamn7l_M{vtR}sZ+W&pUxsrrb96d>-?TCdrtT9m|yOcd6zZXOd22-H+yDCv8) z!dl-&l)9^Z=N2EEm3bjTDWc1R!$;bav$aMsqTL1!jC}m6(CN#M z^FcoigO)@!k#Lu$AsE}h5o*bkes%P2>L4PBgQ~Q2be`3?22X%IbnL~RWSG?0yTNCL zvllCx>5CN&%>LmGUj^ke0}_gjWMIH)L7&+FFoV&gI#q$b6CNP~;k>YTfN}urx~R8c z%=!JdDRQ&Hz;SXA>=r8b@55YL8g5vP-*Wnv&ET#JN)I@2vOZJ?Kp|tel`Rm~+;*z`7WUp*`q3K`T5k4|* z@Vvm<-T`MG#Hwc@ewh^jxM1e)#+)wfuW9Rvi>=z(n@8?oiWH)kuGuSV;QBzrfo3#ul<$CjUD?(O~a zZRIdsd>C;j%Hj=RZQal_yv5xkW}I@TYL5Yx^SVf z{GXL@wMTaBI8mWZ`x9Z=krpE zWi{#hG#?&hGO8kQpX6zPtRio=5l_;r1Vt+npaV36?!s7!2s3h~=zy8Zh(llZ10fVFPxxd?~Lbiyzts3XVe$a?CxpDlcp zXy0|-+jGm6KWZ}U`haEW-XsYS>kh$b@6yQpiNb}gvp;-BWZ=2mS1tC5R`rxT>lauT_ZU~&U9f{31Pf$Qro!Y)lz`1gKu<*R=H zaghQ&S=$NAv~0og%lrwrh0iuE`m=zY0#>~s)VGHJ$|5Z2UUzy3ANOTl zEz;T{fG0&`>y`^QMR&f)C0D#t`To2@L@8)hwC!p6yakcp*Vxu>uJ0R{*9KgS`ZXE? zpK6AiJ}tcU0nA5jfb4}b3a>N--N}?{yUb-g`fn9Qq~*wO5SYSVo#OUwChAyadoSeF z4vM?~l(xG7?u6NDu9G?sV08}jF*ui*qGqp~_bxwi-~Alkx14WKdH5a;|Dz4x^R-(Z z*#X1IICB}OElWrsbFB`~-FIi^@4(dUg4m5{69Km$f>1o`MkVHLojJR`$N7d!ujA~4 z?)?Rl>lrCA&8NECdZcD@n2^c^gCEx9gckK*&^C4AiQDS$qIy)-x^r%}&+yZuE@D4ggUVM1vsW`A*Ml;Ff`6gQZ+NH- zSwfW|2a!iQhEHbzj6{?;cm2&PA2#I=x64da?M%;K(tDGuf0#oEli%}l(K7ySV;lY5 zIf3xu3cqMol>ZhR)%7}$Jre+|DOc)R{CCUu=ux&>kRxdKgRbGb?0=PlMF?~Wydy%= z5#GqGp$8J4Kklh+WE9(qLGsC;#VDP`!n?E}VSK#;%I!Yj)L?N{%Qji*Z_hPE-e}v9 zMX0*li*1y-qw=s!gdY7i(BcucH&eD6=&j)J!FuWsV*Mp>30T2_hBGyvqbCF?XfMeA z8iIHC8edR$ti;b(#2-k^#aW8po-X<{0R!VL^83%pe5&7HIOLF(OR9-C3s7*DT450M z=W38t(x=zSosFc;oWyD7$GBaJB5%F5B;O=+tRK37;kYd?pmE?y*>Kwf@3IE!((7zf z{|8g=9nNL{{*RYCLP%s~3uW(>Rc3bCd+$AxnNi5zJ5og1o9vL)%LpMmn}qDm_q=+4 zj^FV+j{EhWboY8**YiBj$N3nut|K~2q>`agpUCB>>5x(Uigv{c=6ehMR|{9r{2ZPwip{JgiJG=6{oPr5L8YG3WgD+e4#U0u6n$4L&~D-(dEK7B0g|iEkILDB(r%36e7Vgge2M>P-5tHu3iBSaQ`e@!??F@=m-|vQ_H~{8(*AivnDrgOgl!ZeUz{q zQy_Sg=AaPV`v{Q6pA{B2AUFr?K!8>Q4K-k8V2Bi8IS60hq6jjh6o%}(S){)D;JOFe z(jQ0yHRxIb>THdZEkHx6DYhM^@)xWh*MHVTsoEMILAE7m*y9!0P2dUjXgQsRMc*@n zuJ!$R<&Vb@eDD>Cx0aT)+qn4>k>$e!{pS4iP}ryeo<^^;?XK|#kF9LB0Uv3m4tQj{ znhs}BaKwOjz~%~c6NqjGg5{Q=k#T$CVO|i>(kIBlmbV|+*(0^z zqDO3#5bX>S=lbk{QnBV;g0M1&Nq(@0z=2Bp;cf9D5;TUq0Y@N{Yz0SC7o$qB5Jt>` z-kvSqJ2ZbIb@;N?c-`qvUjAS2kX>W7JuNE>4UhlQrZc4kQRN8o7+6jJn$O|IOWSmM zZs{yR>g(JGLE5;env8EE-{SBGjGX5b(B0Ft}V zl0r-bO?nX9ohk42Vtr7T_AgvPOygx^s0?Zxj$!Tr6PAI-hs#LH__I}{+uvyQX#9>G z$>n*=WeOrS&Ckzg$Rdg9LJSoC`79udSoiUo5y)*lTa^``|7XH9<$42oOCWn^d<+T; z1gCs-0V%{zo1m-&_pb;Slyot!fT?_cuBAemE@oV1;N}a_X7VI=3vL@7GYHn~YNY$yha7QRrA~ zz;IDQVl2Ys_0dvc@-8G92ONEf@J!|U#tGRO;U4~u0K9wSY+~|E!(s* zy)brsIwR)I!Vk?2?6emYyF1Uxz2)U-+@LR%5uv(wC+eHmOQ`&THQMEA)qcWUuG z*J@zV5+|XSV3>9PHvcBCr^KMa6O=zbQ)7^cW}yRC*(=8z$xBxt@$qY8*-CjS`p-JI zdq}U}SV{>l2VKeNF9`iZoZq zQB66wi-V6C5W*ly?CEX6Z=^G%EQ~uk1#&aPGkw}^Wy&o>v5#J3vBR$?oSYCSd z*%`K9p88+Lq@|HHZ$~-(h394X*t=TSV86cO1mJpQZ<&v!>VR|QS^i*vJTP1)V= zcqq%;r*s{33XR>GcJ3UrH5@*o*3Z`Oi7n`Vove{5KuD2mzkPuURd)wGea?+3r7Sp9rM28 z$a-d6_ly-5{6GZ+;TqMA_lMtwkcj8lTU@`#U9#a! z8;P}XH**2~%hb4A@xfxHj;oxqjx>imrQfq{Hv=Z}s3`ZRxs9JJ9mY>O>i+T|K;#_s|K`+MIb7;)2ly3xk^ko9>B(lV$a#g$Pf#t`8=-sMe z<%PurmF4%Zsay@+pbMe$TD;XScKU}wFXnZ-3BZ5LY#eX;Tz`GJ`p$#o7d5c)5Sw#NFW#6Np+^MhwHdSiVJ+UA)DKy zRZ7~5;yZO^~%Z2|mg(`*X^T6>nQoYdCGMT{ZqpM{U zwoTsO`MbZX_+om{wP~*rX)t!NdJX-^2Lo*)Z1JmB{_hXpMQzhif9rM7b2?0@AekxB zz1v$nU_t^J3kI}{=bUHu_TQ1|1&EQF8!xS6A}&(P zI#Gu*ru%j={E&cCjJ znR9ccgPl7(CG3rVB^;afHTy+&6;o;)-Mw$!7J;AO0HVP^!?Xx^NJ>&XhLr@6vwK=N zQm@;O*lGTKNzxdA(4)SZmJ>BCLlZCuJ~|OR7sxMy)(G)d^|@Na6Uziqov%vP?|~-zN4JPpcI?|8mWbf8*DCsw(7q@QfuxryT?Wyax$Pf7(}~@bMyV zVJTA_$~>v3PeI6`f&*-jk7lk7<^y8DUiHRBF?H~&Q zuJNK+ChR3Cr&*in$qIZIwu3Ke2cNk8I^xMm^v&fF8uyHYH^rkX@;7| z-S$@0Cu6DVRT5}Cdd8{>E%Q|K(I>ydPMlIKe{8)Gd@l`oBl97t1ukz-(2jTUyxqo$ z@1X}y5TpzxaQGji^nEs}Z%kN%$1Qu)o+=W1IKTb5aB?CF+AhRa4GHT&XmD_|lfqnh zCyE;=KkYZgdyzCW@rwoV2P2P#`R?EvLR8u=WP_v_osSq9>^>Jc_-WQSPgJVS8woUb zFO=R)8x$8+q_-l(4o=O;SnLwN;6h?(>(=t#5H~I#OggRXm->quU7n0-Rb4dgI=o~c;+7_vp!P{#>hx)L|EN`~U~`Y# zz)@$WRQ`i$lDBASb@IbyZv#^IXeEme;rmOGB4pmCgxnkOKn;S;3@bHZ8xFy+?Bi9l zYx~_5sKIAZm{@c&ZCF;88~MPlj@BrR&)>SM=1GwyAWzRJBPiJH2;qj#_JWlU^E8Cq z*fWGs^srM{QpB>tF6TmQ6Dt=VtGy4ME(o$Q<11VN(mR#IIr~CYySX>mA~xW*DbT3x z$=LmAAoly?mof3*Es5mA$ueO3BQ5u8C)kJGxq$Iwq~|Ye{uvVxoMZkFr2AtcRzN&C zY8O9~M4lofmd6zgBtUwV;qJS5m)Zo_7KpB#-CuTH_8+OGS~4he4&_>hQ?yx91o-yB z#$={{Y2J`UL!Yyi-mb4$%e;GOr_ou5k@M+BEr+KskXMlAD+8{XT`HAaRek4HbmY^x z;o!7Tfnr~PMBz7w8PGsvAVzG+lOsC5JR%4td!EB@X1-da7|7het-B`#a|Hsdi5%7q zEP>x92VlDX?TA~2;&8@cBUBIa*&~@v=4QGLV7Y05uLAqv)cpkElNK; zg8yHyag<>v;-u9K0cbm5#x_G1;{YZKqDW#Ek~gAR_0PpzvuUG@6O2lXgkx>$2|f_M z(Wtbjiy-1&yI4lcc@aJAkgr6WoL%gg@qC%yr@m_>oh9xK-Pd|I_xa~CVA~8k zJygB8jBEAIJNyXM>DSbgN&M5+Iy328I+qXgGKTHxH-48Q-XTDju{8jnkTX=v8LF|( zPS|5}G4GAze4?7Niv?wU$?oO3|0VUMxQ_?bJspkN%S-po;sQgvwHH9B2{1AG5gH!e ziB6P&FBgfm!)Mepz`!0hcr-%#^&J7>*RKpGQ8ElN=XWHfq#s2QpiAHSj#<&o_+L8? z&#;e{hSr$(!14K0cPA|mEe;KLyN!W_E>6*rk;9oyG57Tux7ov8FUWA|K3>Th088B_ zcLVVDr0_r6Zd^Cm+OaU`8x`eAmB2KL0=VzNgK#xXRYwz(Kzn|@ee<@XqtA!sr`533 zdbkvEw(0h*P`Q3!Shv(Oni>ur?u7;M5-(q4;{XVPcTpmBt45sC-#9hbnF=4mSROI; zeA2B7Q#o%)oL!yRv z!Sl-vcC0ek_Pf%Lg&_LU7t}k~zd{5R`0o6D7r91)AA~>v0Mj^(yrnrS+fD z33kLH?{dWISVa%9O#E_#$jN}EtC}N!;U}{vdzSnM&#TzW$~ek&3B3~dlJr!woI z2jf`Y)W)r2Y4X08X%;PBq1TVluaC=S=y2zbE;ivf-}VvI&HNIYZk@nsDO|&PXiIuh zO)KFUNq$_sTmLF+t9RzqH^J#^Y)*uunusM{@Wfmb1h)p^XU+psiVn_t{wqtakzByS za#=5~Jt6L7_2?wCaz?S$dIm2{Jz0i{&kzaJogEv)ab0E(`eY5D7qUI*FoR>ZD|qPG z(cRsh-o;<|Y)!B~ykfKM1_1#^a1=!>Z?$+P%&OK}Curw2-uO3@DWaaT<48gm`eG7=`z7PaO73zY8nL+33>S}tI z_(6FL-=68zZ~hH}ZhQa0hgTxSA3Z34;l+k}ksq6XWvN=PFv+1V^NmG?qpmDUC{aSs-QIZ5JUa(2g6F?o7FH@zW5Q<+cvHT;uE zBrp);HRo6k>zhxaZCtHtf_(o@)$HfQqpUkLk3)T*$E5|o7UFy%oZ!|%m-Ke69wdoq zUw^C?IOowRBr^~wZeKh7tksrgcSI6;%d{|(i#7e4N~)|^fbjT$WG|O=2*b`I)?oX6 zSr^`WoF&f(Kh}))kcz@>47Wz=1D1_TadIi``7*SpJFcj zK$0SV`_e*c~L~{gnhm=$sk^jv6wIe1WIbKuo$7X#k?&IUr^tH(-0UwzT znK6D}>~LIGm@53~JIlJ<^=B5lxV)IXEcDjdfVB(~pR9Q{g(OSD`2m#b(g$G}GZzQ7 zT1^J8UcEY7%eq)g@5=I=#{%ppuOOvXqmaeCe$I5T8N3azL|3^JyF1SkP>Aa#j~UP= z%+uHPVuZ?kvUpeG=DU1}$yVJRTPT&i#xHn1PCvjpt@x%`F<;mZ6TE?$zQBCpRFyx6 z5UIf2oBD#c>v}(Ofa63{r~pa_Imil0wjivQ72{E&_rJVw%Le5U}@YGcADPwrPig`)AgT57xf2TP6X5K!YxII`)f^G zCr@eB9NF>r-MS8bOAuzC^NB{X!F-qq4t%rP+fKPZ>b>G*J_Ej{)oVp&HfpbS&w7Q& zpH?a&(`z}caaEcFCPSaVyezQ%@ja-_>ZoAF4lYi5qep4{s!!4$-YgoBN(R&#dI&iZ zNw+9VA!fOJG5dRaf5WvJv-6gkAbRW9r<`cr-6O<5COvtK9ZA#*kD(TFHAkVkmt2uV zcX)dN8wU!ylhWjVfmTz&{6f^37tosfVcE`kMEHSlBWJut38nhZIKt@#`7TSWjyK2N zDP2RO>a{D5S3ag+9sHBLM8J7VNy}6zNBRk_&$)mu44W@x?1CZCso-ch@4bXTUz_*wI*EDhmceJ7nImATZ-;)`yykRPX*H z(qUYi{j17bLF!(m{6_2fr>X*Z|KGT(-cBtB#O(MR*~wp=#$7UPjNeEKL{6pVMks&J zHo6F*L}PevQpUFyh{2i`Zcj8xI+ zQS^Ds_h*#&NcM_+D=l_$d^p9=5HC| z&E*pA$)+;^Co5)+p%?$D%&MAQLmFne?n*QBox4u1telcBy@oz~7ZeneHyN8#OHAC7nkEn`!xhfFGQx|NG|K*a@Gh~SL}Y%YDPe43w2*mK5Dzh@?}4p-C=BNk83;+n-oCe1-dOx}e1JJ*?d@UX%1HH;eAtTsBo5l0 ztu|@2kX|+aWX&6b$CJZj1m;(oKq7)iC89m%I(CP1*;G%8#SyR zze{D-w32%(rm#cY9ltYk_i+R)@kkqCq;7@vcCu=P_SvtANOk5~Qx)?COJEdJ60F@z zRocvn)nmS4vT3gJi&y!ESf<%aI4jwD?x4E20Yt{#Gp#P~dBWMoHxfd1%=Qw4dG3*H z-b>K?()#{k$;g{O4W@!_83wf$@2A3Myx94%<6-nj$;u|G8f9fG&ZI8LbF*#A%gY-Tsrmf*XVv~`?y%F? zZ3*5-H2A-ewL%w#8x9aRE?Y0oPvK>P?f4Ff|2B#5<>~CdyjMu_sYH1DVI8I{%`8w1utcBtNSE$ZKBb*1CKv zeG1fm(l4(mGJkf;OS3`$xJEL0qqpZ5ag;F;$MKKb-MmqKA9{kStoic$nROrADLvqm zWcEtce-J&0cNOjDHOx;muWUn84eVtdb)FXr)H+-%q5+&!=BM1Fm9TZPdGCgQcneGpKXYb`ZK4`E0hSs~c1LCzak^M(S=Nv-zlK^VH~yn0hy_ zeCs(kDs6<5*C)B#JC`h?6af*K~aV`HvPmS+Cda%xV)jW#}Aa^q~B>Vg2c4H;= z$##?ca6EIcub6oDA`||8FhLzF`Zv&JlQdZ?pqwc_&XcdeesX=Lc|UpZQ&uffyMWSn z)GFG4Xhy-N)m`MIlWd?Hn_SLzrX|_6GekmsY3oTf{NwTtLAEY=6TN)9YG}xJ-2|n9 z@>D@7e#+|`#wLXHp~WYp>Rj2Q_n+EHNu=tPJ;IlMRJRUTo){8k4^v+hqPo_-z#{B8!r$HZ3RJ-Q8uO zYJi&-A`qa6x6LSsnv~*{Q|T*X5k|% zk#m3^FTuy4Jo2IOl4Gan@v!bL9M|xvhu4+Lv%2?BLXn7!#=TsQ{qxhw7XQoiF5vF0 zDg50<`ar^NUs%F7C2wUeB9|kwJxuR{7i5ZWdvJU56^jvYc3W0PYAI?Z@`8q>B()7)KX;<@{^c%t)17$q z#=4%_kR(clJCt1u$D}_+BJ~F8hOrqTNL5lE5#6Xy9%%9B2 zpTgbiB3L{M`U7R=@q&3}bQ$AaG`B)9^Ec~LDWA}WkiD*Ow)9%+Wr$bY4hrIt203pm zV+R@!ed;|Gyr@^N{v-8Xz1vIr0>|{2xc9fz#=+a6RqQlZ$MR{fjn?%>{$Cnt`>@VG zaL_S5<-Xk&_rDYQstLyMatPFPKlh2eN#q;?O93Lma_6nPOGI$e%Iiec4EPCL1E;(H zP6zxTyubxZ>zZgZg%H}W4bS(aqv}ZFOs&}uU9-q-3D2bgqw{Us%@qs`jL$#>{Jisx zg?#5+NlB^L%f2g`vYEAE$aUvf)6TOv8La<`0Z!lj^paY{Eayk9(MR)t9$gmPL4!@DRBovGz&hye8*%+fB$Vj zBAN{#icwF-9_E&FJl$&h9Qk69IjQsQkM7$S-K>3;r2a1zz5O&WD?e&u4O20gMj02* zC%)zK7J5kYo1jV}()D)E!P73&e12&d7`XLWDaGmJFTN*AvKXC(@>zIqE~6c8E{sk%P8~A6Wt? zKn%1)2hczcNMT390K_F5$*~szh&!)-f6v$TrLUlW?H;217$w5I2CoT2rgG`y`ce6rZnwv900CmJM`^WhU^?1}0*qzu>#mM9|v-&qJ_4f6-tJCbl zIt!Kl?(P?4i-W8>oV8sB->-uB{)&g9Z zI)o=xEPINBgkqj!vmk@4F1ATD!4vMNG1f;cUr8HCD?Q@Ibcpw!l@|-iRgv`T(-^

m(OuWALUq!FW95V9B5Ow3~z{YXA*R1$JajL+M zJ-$w-gOp(WJ+{+(OIwU_c9F|>mA<-38rvGlfCu6m*f0_UD{PTJu&qGR5#tV z2_LZ#pM&i};IvND_TR2PqfK9m^F5XGOZ{;-zDa=oz2FmuzLvbAZG?T;=c((*aX-a;Y=ZyKjgpZH#-mC$2n#a^Z;~DX@NZkGdnks zwGN~asGE(?v+Rw%r+#UCu^YINdh*d@Hil+hmf^Feb7SwKx65lMx@)AgC577~2OSn# zHxJgqA;hE9t5I}tWAfHrBY<+duYu{t!rpn`%-c3Gwb^s*paAP~5fS&<$Sel^OWfW} z>Nq@-sA2n&LhqB|Qk%+I`v>nvvjdCA3rF;Mh6&j9xo!*)gonSlNXO0@9!~JZ{mP%$ z-OK-E`4c+WekxguC#0~H=Eo*Z|Ln05Pdn|ljM16Ny!(eY@P~6P+0}A7__`PEbnwtW znseZ$)7~{(jtTu(C*HLp;w8Gn1R>hm3)rme(V@JP%sJcssr5HP=3nL=5l<*2_)Rf>D6siaq8jo$HmHuXCGdH1`*4<95qXTA=A?cVUnh}k@NoYg+b&kuD8 zcUQQ`gqLqnNd9{hH5`%|c*J3+_vVgKYJj1y|Gvw%pGmhF$qZAvH6PkK=JKeEuSYn1v_Lu&nNDLRw8+6mEWk#~VfZWFtQ!l&a-$6#O_FS4Id%h0{;&Vc_9td^&3$kv zGtGYy$F4@Ighl*QtG&DB0rSiorc5ENY5AtcSAA20>~9n*`Tyj4ViOm8*EWwwP4CTO9Cso=F<^61{bo|Nx{-Pk3<7CxUpc8xSPp9iVA0Ox|L;zra|D*7WXR_uf zbQ#?P-(TNx>3<-yVdPuuN~9Gc&XH-!f3rcc;2%$Pdh{2Yc$(v8*ab$^=up7$=;&`y zWFW3(QDI@1F_t-=n9&2E7+;lHv~iwKrgp;aCTpMAMA1vok+CU$gSI4kVqzjh(%Q+XIKCm_sfQ6Rmgt8Vkfv>Cg@Q+a zmX^^z<-M}z@fZ5FSFV1L{#q0o5^@3iuG9z8;(i5(SoPwmWn(fDwS7%m3rjlvuGmlP zixO|if5;AU`7#pN!2Ew#VULn2(Pq#V}w=CYUJ`Tvi%lxDdijSVA6Y*rwFImHqx* zXQ7YL*8)z5l>TCluIaxo*;${wvj#_`?Qf~iM5Hy66V1f7I(TS&QFdCkx7FTPhL}-~ z+c~ys*WzC$^cXqLGMv}+N1?ta>0xoFB_~$|&e+vN;TkW{a-DqOAdo1;Q%fc1r;=uS znIp*XjUeS=TW2x~bzQSqzfys~(iu+fZT$+{9k<&CRUv5pxo8=api|FL43I}E`lB@x zhyR`S$}4wjuz>3z+5lr)Zrx86N5Ju^+ckp&7c}=l=_y0t|Y%wnF zT@OhsD^^ZUPAn{gYUdV+OUMiR4Sm_Sa;vT_uv*m9Kslr)@fJ5NbR?9)!di>gD5?*o zrLu5-?HwG7V_Qu`jE!}4ULGC!I!@P-gBR-ik=Q){ zi^XNRgl#m&$owo;#*CvcB_)vDLhc7yO38{ZcXBI1N=wJbAKAq|3|*cJmYwc1$(+V? zL`RA^e_QdZuM!X|4XMU(+LF+c#+xY;q`$*8dNCux$$hx}=$13t0&R~~sA6KGp&nzx z6n~qlon1lSgZE2<*R0&X8k`Sm5}a&!}K zU2R65?rWX2xY}!FeKMrk`3;tcZL)ObLLmh ze6~^f{Ja|;HiQXrD@qN#=P>R~3qAFnFhZ&qRZf%UARX2%Gsl4vMeIK83%0ejmGkaS zx^2Mg^+%W6q=5!55+bw?y9e5bYie3z9zVF9m z6A~Us$(hF;dy1*zFp3p2cQs7F4Z%WrJ)ccCynBPGZvB$bby=L!)&3Wa^Za#ou9FN* z3OwV-?WX@?2#km=7?y;;$MO3x(gk)qSCv*=4=V_nv>Pt4O&N2Ap^xzJ@bqz)#?vhX zf#HVKxyft#?HL=ri0d!O`lGdvqhsm{-1-&ln7@s1@Z`HDs+`h1?*-1krX&&S;w7|A z4cne?YnJ0P$`!ME!b19-_#%XNyyp1v2{R!>mT8^cPD0SgK({?wuV*Gbj5z%Av|&|t+K|1QZURM-Ldmtu z$)|jUZTWfcU1!~uzTfb})|)6>-b}VseICwE9k?klRGir+vhdRT51;K$w(qUQ(iNXv zV&~@AwLGdfam1#%_uYF5EtAmbgDEm3Ov~cZxJnL|Cvxfh29Vfc$8Gb}-0Gh+-lMX`YJ`?-oy{4>QwU7okWsfY`k8_1{d+qL!9c zvWScE=ZD{)W8<2iogA)A)jk>gml1em$E{bM)6upp)s=Lc&-@1-B5LqHJ6taFTGu~; zdvyQ{27?PU^7H=qF+TBidyMxSq5|0A@dxurT; zM^^{g5^oP%X!@(PM;tTl(&&5jd9QYJO{F7t_l$N=v=GBly7qYuihrVsq_Fd`o@8I*d?q zY@&>ej0SfQ3uI9^e**eCv2+i%;ImqEC-dQV<)V}MtrVmRmJEHNb#nuKsQLNnzh(Rk zrhS5xg4U}zkNqWDznCsWmB^OLQxoLjR_`R=ndJNE<-TUU6EgJcQ=2@l01d4PRjdQM zNJ!+{>NnEbc+S^QD5@(<*Ais^F1(zIz9)h!^?k~kW42@bgQiyFBePIN6Nfcdzb+S2 zVnRz&5|Y;fvUENBP!&AUp3?ttKF(-MqF)l36}KmnTpGX?>Ptw36GQpPL)f!_xH!`xFMk(yg*%x465&}R;mpNB7L0R*^M)cP(`NB zWV$pmA)0I0?HfzgVga>>X38dX1flORcOkwP1pz!oCf?cESrpXO)m0os9b{~591|HC ziJs-+;*yq;!3iNoV&dTblcnwy4rx!MViClpu!6NNce~%-6R|toTOH!F>O>b!X8$VF zXBD2+=1{*m3SIY+&>F+W&rGijKF__7mPJXaf?Vh{h*azv$m zw6_|(zdo4_n{ zIwi~{7b!hE85Qk0J+OBlQufX!Y!hEpZ=hLr`vUo_uc%5E#kTshEI85&3Wa(z`hTgf zB&~n5>u510SY)IlFP(|W-JvN2{{R4i%M9X9hme$`!5gu*gBvwj<;3T?rirYN#+R_E zDub0_V8VO_KbCJ(BG(~0Zb>a54N%0Y1 zPAKJ|j#C@l$&R^8099e9{WP1>5OkH4_hF(Bze}goI2w-4jR(GauETe5f{B?p;^e>I zxN2E0W!t0bJGgj)+>Hel0xl%Vo`p%A=fXNH7C zZK#IgmIMwD7F+Q-ZmA1;)BMy><$?jp_s! z(L>CSqFi@8^x)?vqq``$b$=Ba5g_#<<_Tnm(wv-2Sj5&02{eNqBxzsuzM~f-6)e~4 z!i~j_DNg}9Il&wuz)pip9rR}O+dw6D~~Kv#lydi+lo zJ`$hlH>|zQS^Cn3Bm2*>XSmqQnrnVd`c z+_&29r<6}w-K)Fd5K4j5&NblvOz+eiy~iyJ@PpG`PaIYUj`M?xJ`3V!;UD4-8dY2h zanJt6L`FQ-NXk#@IY{Ck`PBd_olJXX;(g z!g)Q9$qKt*oGUjFy;~dXp2UB*YT@k1`xz?S1K14lLZDZ)d7v(sNz~<#iD~DjEPVjl z{qWfZY{N<@nrG1((*7sgZ6y=Arlz6YC+5ztE<<5o{iqF7d*TpgzWG@7^*L!%`aj|0 zHU$ndO48TiW<68fTHV>o!*X>TEUAISacQ>`_;7YqOf4>qT9fpjk-%AfhV%D;QJW*Q z?K~&>t9iK@s0hDB>$i42GsB}n3($ds?2mMjW#E2&?`0BO7 zZ2Q2UBla~YZkk`UK&Q-i*X_#z>faczSLuiB|pTTXYuj#YMt!9$Yfo?gnUa+ zx_k&Ze+1l4ys&GukD&n`x~S`ELZs0y@V19I1$~7``1^clD(`) zl$liUaG%c_ekIg5+Lh3J*z)7y$awm1H+B-Wycb{E2-h44$@%d;$d6$uxxl#qRwgD+4_J>(Def;CMnu2yRLA`V2Ln@9^4i*3 zj@8?HPqhcV3cUbF`UGK^b)Dj~t7sHJd^d4vJ%~ZzN%Lo_Rv;haLerjO%60p8DEcA^ z>*Cb{%2j=jR5x;I_Bp1s;s?hb3t@~yVSyo7f{d!_S$eU|>dL6|WdLJb+AL4PxFL6_ zUep2Mlbf>(tAc!N98k#pHe7~M0da#R}>mr(+wNji-1J94)^d%Ab-oR}Mj zL~cz0l(#y7mmK*gn|RydFA4J(;=B^Ofi#>{N%;sY1~DfAAF1$h4hvxp%a4V5M`tW0 zZS{1mWh%A9{>0<6-bBHjDAeJd42+p6+Ff=L-Idbe|Dd)ouH$yefvyFY_M!Ta;7Z+w zXnKJ;p?veR@Y8(!nI={n7M4rTA|nYJzerJ;TETCf)%1g-D>j-qH?2b_As~oHhNMwbqzC(hYf~y3}f^tH-;GadAF^ zqGJxNq5K`%|0q6-R@2h2`}RA_ji>tt2rK9f>j%`8j~f6PO!wdE&~5M+yX(2OpjMjw zXOn+%R^Xu93BaB!(WsI;nwV>mI1@5F!_2GZfqT0LoBf7(0`FCP>uKLzWPU~XhPgs=*h^L;WT&8IycqZKcGFM& zlNtSjxRyscVRLO2L#dAq46M8C@`PMszGLwh5L3Gp#if{tdWw->ydc?r@nq{bf;FrD z2h>$0SW1t--V>Z1Sktca8Z53(Mlf6|*s!_U{~jAWYR`Ycf`!M2rx4jmUfsCuKJaDf zCy%{r=gu!E%dvBw%~jMh;)I#p`j$%9l_#{j6L%*hrCpnZO#j3@Fsac{Zp;y9QnX5* z4|7>#-=b?e?K=NY;!9bq{9+p%Z7A2^3j<2RAHFg(5u`qw(Z14MrX-n3zY{LP_-QX& zEbZ??iSI3P3K8;z^|@}_aLxc~y-)!Qiu4nj&8fkiOufnx@h=pR>_`@GAg5C5sGy^H z?o<5!P-V8g!X8{N#MSMEki+ON!0wcf&LAtjVvZ4Q`uukilpubBnfAYXC{OVP+M{4_Zq5dLKBH(-+zS-C z@a19j&(dhY_`?|#>-6z53H$OuJas`Ql z$S7Y6kPaMr`kX_84NotGyw4N3{VJ!)0YD4=CWvsC=? z)K9ST?uu=H#2&dWV@Hgivju; z0Y+d9jPF$Zs+4*7pe?;V1c@uhidw{36Dc8}ENtUozJ3pc+ZuOFzIYYeU>s@MHHCT} z_%z>_6V#RY^vQc~Sx`FpwOl=Ep0l#vol8w+mM&F{ieC-{wKtCJjZ0rOT*;7oe~`YI zB9ZeGC$c78`E7d&;r=8|JT&Gu8y|vwh74ZturA!bsN1$c&XW&rP4dkKDEMk>TNApb}8f7l#_RZ!7>fWH$fX zxj*7r9}o7*QClK7Ps`rM^gTO0PfsDRD_O5whr7gj8#9z2i}B)6%U6D@en&@ZRghe} zy_<)eoctNoDeC$KP_*qAS+V@)4oN{_4O+>K*Ln4Qa@*cnQ`uyQ!*}4^p29oZi_}bh zad$&kEGhg0$Ns!uP?k~3pUhxd%G@UOD+~K-2UCUw&4g`VbcuocgzCI&-t|=z?E6=3 zcJ|&s`@!Qc((v@*_aFaJAc8LS+hLwXc4x5SC5XoL+l7qMFD8wFqATx$Nn<>l=XlSK zXU}~boJWwL5xF{6YBI>r!P;=L^H^uydA5O8NJvP&>))h{gwDEpx#mzWwj|JL@Uje( z(w=vl3h+o#kr-R$)EwZf4>Kwz5I-w%AlBa#R{arIviZ7|%b?B?H#~80Zk&xM{F(Oq zil(eHui(h2%2}e7Z}R~(Oj%Jl0dJ`G&eY5Z%vaR(q2L~loHqA*u+cMQRc#@7vooF; zD@0DB*vI2-$6WWbjQ6TBDK1B&<}OS~bJZj_F6cEd4?N{!&^wA>+I@cWHkLGW&P02{kxc0NXpT& zx{_=OKGsEBTh}3s_wm44jnggPfsEiZv;fbHjR7N>+_R(Qrb2kv|Gg5tePpF8pQ~l|wsXH`V;gVs_Mc~%1n*<(O4Ppktj`&H{St~}J z?_)9j{|?bmJ`tc#2^Ib`^*$vNN`DXF7bK(?pVIn*{sX0vd~N;H&6$kigy+!4D(Uqy zy^xXC!W&Y4^)@7=d4tF9WLYSh^bbxU%J%8(Ua1{Kxc60JlBJ&tt>lR0Jh}e3$xPjuqr_my2PF zKC9meI+%M&Tlo)GS;LAe%X&KZ4J3F`wG)O)iSD#BKL93DPp@eVdSyqbzT=4F;uM-Z zuG1L5h`Vz zWWB+1M2X;dT8N;OmQTjbUcLu9RlY0Je@7lcfJ_+#W<#X3JnpJ?R~1xV4{ta;1imhpx5<4P5D%3FBR_(3J!`DaH5Bo$ehR zU9V!7|FL9qwXa=PDv`3gQ=Ww?!TFoV`h)snr#+dcDEYs%S!V1EGz5( z5`JjxcF(Fd)ho-XmgwnSd>i#S-YDDW#Ml4OcsS}`>>c^8~yRWqL ztjB7>8xy?xO!azulxL?Zzg6ueG{h5X;}&Jz0<)J_z4e&{gvF^)c{Bt^#smIFbJ`C9 zm>bg=_Wd#O68BW*WqX+;?o*GC{WWTF?q&99=UL4UyqZ?Vo9cgWCMt;z&EP6KWHfd+ z-UOUQRwmI1ULyHKgs`r?TRgE?je8Ij!?L@qFpEjtfYw%1fEP(N;i^0#vROD>oKaYHF& zkDlKbup$CzX9ocfD(^VLJ3eb)gl?xFCAdKvQRC88=(xH=P_0*)ARv<9K1!86M38dr z^3&GI;=|y28C31?NlY4Ze0ccq7AEh$Nk`=CLmeV#EA5!a>OA5bBcb%u@o@v|{lQ8D z>lI5=1iQf`exJSMSeJU=zN&q)Y!3gc7GQAr0#CeYJo+E@ zB&E-MGQ*~W6{+_Dd>XIwbm+`n9WP^s)k?3ekLQwzncn~A;Ih1&cH_$01<7Rck&8X) zRjdi`vLyFh(mozPDq{@!^nxITYQdL3%PxFI{+FhROxv*;{YU9+Q%8CBjZz^iv9>GBUS0^z)+A3pNhPmt)CT6FJjzgdnfW&#(Ph*+kC#uDo%ZMl!=0n^DA)PP7G(u|JcjG$ml#Yw@k7gqqtqC9ieP&z+}yuQGmXs85U9Ud90{bI8EMwS%+^3 zad z)jhV8Rk!u4oukXk`4OST$jFFpod?Hgk=~G{AIL&}bn!Gs!7!4|2vNSMl~3S+;8oIR z@88ui79`r?2^17sIzC!7%;(NGzm@K2Qd<{SRQQ8iO_=}Y>{)b& z_rDq=Vv^h>iDTJEMjt-m)l8p>PE!};mGBpd_0*&iq)0kr>qn2VqK<|f!vum=6tFJ> zDu)t(-JY*e()~ZR1bD4n2MxOf7Ap2I2aWui;-`BWgp|DBhCGA7L=Zt<<73o_9_*>A zjA3vPJ)P;#=G#kGWP~aBOW^rx>gXWr0+fI5|899z=DqzpYT_;kTioF9S=VovkM{RJ z13KG$u`@A-Zno4!0wEYE1#RPgXXM(H#zE{1RvGjBhINyPumtC!Fe5A*7jjAv%yAj% z5LA-HV<<8ujL=OTcFIuvL7EbuX!0uI*e`9>?&nK7vY4uzH=cwt-;L+9-;rIQ^R6dX zKnre8R?U@5DS-h*MY>s1p z)En>crBf*`odG8Y0m$8cq5ZXd7h&@^ko{95(uRzZV;CHNW=1Jrn*lpYT8IAnKDYUV zTEIW>@Sf z5HA}L))I9k*jR(tKj+f$=1WDQ2q_;O@#YZ^Y-N|o=BLIQkM}s4Xl&<6<$VENFQqmC zw|>UyV)4VZ5rT9JR>>Wmoti7(9$$w#Fj)2~s%QHwoGnf0;O`?v&KveG+P5V?`;ji8 z7i-bJ_sFCq5kQcx@y?r)(D6!fX42L0)R6g~-JKRjinD(~=M=z9 z@-Ove#sVKW<8}{9juu!79((@p{EU8o#?K}6LAN33eg66c1bpQQc-IYjPC0Dih^=zd zJ61;cpKHfRc(^VKGt;$u%wFKKSQC!36E~u2ctm2=OJjHqx4?0)w$q{O!nIfGbNnWm zfYFEGInRcNUu0}}vp{VV5$~fn*}Kjc*?Pp3lN@!LpH-e)NH*i`_R2K+>2YP!H}!M6o~+z|tB^9}E93*|B>D1;8)Rx2O(kgRnN>qwd5wqJBigNLX7So-{q& zS=6m{XKgt6eFYJNjYASZ$|ESDvVnL222(0c*r`8HiTV$~Rg}VxlmG|CfJ9BS#I#Tf zcz+-~^Hld@tuDo=jag||Qbs8fp2@j$b#KQIWhrTGZHZnjEI(BrmkO zG{>13D?*#gvB0xKa~01sd0*`_3&C!dX2z-VNA2RK3cDd*zkNr@5BaCHz>OftSy}b$*PEBp?-!S)veWz}P>Em5C;((sQ($Y3q zZDVkPF*$DnMe0J0{*puK*Dz#1tT&k96rU*12jwfKynwq@a5CZ^hTPS05O0TqJlJe^ zsVAc(-%T=qJ3@}_7F)E+(PL1CtYx4Ie*#`RGIBUh)`^0JW-P;+A}m-fBuy2c^SE_X z(S4j-%U+%CSIb_}`gmDAd%@%Tk|mkqZIO{Y@jcnlE(K~_jljkqh*0fy zEGg5kU;CUw6(@V_s(+8z2UiOTzntqfjZ?pPv7wjq5ejt%&>gOG&D`bt&}H#PExDLF z0cwJ%nL%gbcC?yr0oC)$aDm1(--!y4G*CfVBKxROniCo$K=G)6;|WW1uER%0Z_f7e zCEi24S2zZ7h{s#RgbQtrfYOK*BoC)KdW&sMLU$cXy!Ig9_`T6C0vSEScBE(G@D9k@T6Awl?EzyqWaz zXkchC+P$OxJ7xd#H?bnl^5mlx%$)b9@;IL6{5>~~{&K*l^XW(?lvS>+e>D8*gXp?O z%HF+L0MM4OXR=wCMn+4FUV)DcerCXaN_JacG9UIU3Z7Od%6SNH3d*@{mcv}|ylB)< zT~V4=n5uss1XEK`QJp;p-d#`r3zA4(Td8bJL4&9#Q3YaijXcHgCx^Rya61+ss*NDh zI)E%_fV@>||COBin4`?P=jIf|i%|Ye9#TsJG5xq*ZQSIF#dhxo;_|mHnSkiGo08Ha3e%QOB`6z?;SV8}HytzR+-G zOfhErv^xKEr{b^WD0)$W)3VUV+2g@}#CJC(dp(ULTsK?2@1jB!HsQnU!?h_eS0XJQ z`sKx}G$;3r3$?leZt?DTWWD8$r>;*w3O4*@9}nz)L58!(*?-WWtTEPAq_Ppl}PrR2hTP zI*MAPSB-LL0|IaBb{l;xa1}Kh32$ILnWsBVga|}8a_r&OMAj3{aEOt}4pI65V)r&! zizCw}kfW6jQd(n$RcdLv&k;$k+tnoh@NksS)2Yu+l^hBfxU3#@=%{hCYjlcGCWHu9P-!J#Wkeod>i9=pkx< ziPJr_fHe#s%z?{93(_C}(JCk?XbRCHS?mcLz$`WB^#CZEMM0uT^Y4}pl6F}iCz-DF zzsDCCy(o8ZcSN>sQz)?L@UPiR+AxqV5J9N#Ufo>5QPxvNQGs3R;8TTn2E7= z7)Y&MZ@)cBj+tugsQ9bfBB=#hv=0Q9buKU2Bc3()^l0l1eKbO4CwOu4i=Y0)rdM~R zMT!+eTYH|K2cn0JZ}@&W#^e{drs}+XpJl!7GHpB-zf@4#M3CYmZso&Po;Esl|6i=` z7_(Thk3Ii>tFE$Mh6#aCh!Se%%X-x$JO_G~+nYeR!3ffG66QxN+%7N`^eBcV0_+cd zcXE~ffKK!i6e}T_hrq-mqUYa0^z4;ml-8g+L~1^_`BpP{cuEZ=DH<04)+Z^Tdgnx@ z3{GI8`3LROsn=+`2ow{5M*Ilp-}51vgM$@W^P5Zf@fzKIN}h%TbM^2~AfXWj!!Fke zyL>aC?~vCMPY~Xd@qilHquMKReo}Ol-~cL(R&aI+hP5!XU^OWMYVCv#5|L}V!e{l{ z|JK`H01|aDy&3ot>VH1Q*@Qko5woHI)wu{RQVh>lt{AXiS0F&8zLz_0ENIBLyNRd* zZ)N(ELZ&G<-U2)hQF@%fLLMHZA}ud3&mlz8+v@pqU*s0dT(zEWLRQ?=64O%Zxp`nb;LZ3a2NJl`&x>ZpaL z;L%jDG>;ik0O_KFz_HUcS>UKz-U!@jh$AmE)wW)YV{vJL7?-N33gZV7)iU(jJ+66D z>KB=|GjK>q$nz>4#t^F-C4FO9Q0KF|Gza*U^UgLnSp0wx?*{S$B*5U$`aya|9Iwow zx>h(&k3jrzoF*#VP7dH;PWfVr5p)K*nV`%L*(^1zB}E2fT%muo9@zlB&npD7utOwT zTX0#k;DalP2ZET=cBGJ8S#<6aWwf*UEI5E7FRdTu6^f3~4luS+JsWuMPR*~2h+~mA zOq&)tqS*e$XoLy?RgUa_{*{;JELH71U5}|J`tRMWH=dszK&cc77mAC)?lRLl{ZjAf zna})AGhA=T&<9Y)@j8qah4e2V?Z)Z;WM8WLQ~^{^7vze*b$p`n`5x#NO#VM=j3BtI=dGA#jU$`Ge7of zsR0&R+>cgWyOc|hig=-Ob*kQ*tRhg2(aj)Q-HBC)%Au`~Amz;2h0vzX;>Ug(q2&^; zvCAzc5m-Roi~g5;4#kZSb*tdtl>uzzxPmq!;gg|41Pn9%f|^-taLlm=-(un|Bo1V{ z6EQ$z4RC>Du7|@9n(n)v_Jcdn%zzN9gTQBxVv*|`BYtF??lCBz%xm@{5yK()U)h78MFkqS zSe;VH`2gFGZ?IA3+Ng-Mg4u`?4E~Uuh;y18D-s<*1WJN4q_oT&#ykD0=UPb~NQOsokY?>=uHxuLYqiA=A{ga$J2`Do|Eg%OV$VGh8KM+Pj zuiIN#jDV@gbAa-XEU6E&2ed1X0CO|^JEDSkuRytYmJ^8vd2G;U`(-e*X$B|cjEWP; zDn9^H_-v1XllI?(f>Ap@v%}MWv$2p?N436}8#(}7h7|0E9s@F1&){g!K}G)qNj07r zZLTy8HMQqJr)0^-(v3rq8ThEfBk|>}0FvK$x)YW_ke#zxdsQj*)(Rx7&Pg_fP>AOg zyXKmFIQ&#inq5VY>^|U0M6SG?O!C;3dI0Kk`_T=CXBLl9B6RsKzZue!@P=b{kJ!w)9l+oUp;%Y?0Z#`|T z3QRCHNE`RRP(T09?8~&1_#nb1|;pB*cc{7j*6VTqrS zS*u>zZ{|}Gv;L}l`M=pjXhHCd$6qg-T&Ed0N}P%9_%ol3n&EAGlg81NUCOogqM4MG zxb_<3@@x`Y$TO%lJ-`0Mq^Li#LGv+PCS<-See=S7I#t=u6-slEl{Pu-pwPv7pRHMM zLo{jzB$~ET1mx-->k~qDh$@`6l7a=P*Yw$P6$@2Ul%aSshDsx65vm7i@Wi71y9(8> za&0|vhD)Rc6leZs8HU~U=CdKB{vVHb&u9 z4f`+HtFz1$M4oqt45{Iy#U4Hiy8W{4?_uDh9#?B7eC&$_n_ofdY6ewgOP0*5_#}GI zVKs=+iP^$JO3hHE#wWon+RfR(*+4!(qXG55ELtwXblkLP-f=5+Ei#WQ-yJxsjd+uK`1Sxle z?G{LmsVu63XXfYFE8gMGe|VQm>6KOBvV>={77?DBAklVpDmQ^%P({{we=^gjD06Z( zt-u>4oBV7e5C2sv##!CWprzQmIb|SF07yxHjR)kMFi65ICk~nrbQdlW-Jil~0JL%p z8~hL5b=DzQX9hOGE6A*XZy@TciPr3^F2Nuc5OQK1G>Fuyf>k zUNndJy_XeyD0++^edr@f<253z#u?4y%zHrXxO9D&v&loOnb`WXyN|e|cD5!T=Ih^O zxBrK%#hCdtJtIyQ!4oT6^{{&chCLpSvzje7JfkY@rhE@b3PLu9kW#5}Uti8QHXes&?Z=Cc3B*X` z03`!X5J)~nhLW>cjox4;YyQ2n9up7c@VLX?@Q0VhgC49a_tjyAZl4x# zt+QUc)9CBd;;eEV!c1&D;2S_@Xmc<4LXemMIaSNi0k|ry`ZR4y!{=snxfVUnguL~H zpvYO2qVsPHQ$_d!7?ANm2C&m%99n32>Gj|NU1s0^gZijPP^1++-B&E#g1~<0i zp04`ewQBzmKJ41-Ev$t&B9%1vDbe_*mD ze>|_xGhq^wtU8!5+mK@cWAGrYL@*l@TYf=7OuYN|X+z~tc~dsf^g=y>R4118Ti?*D z4*&Hh>i>_N6eGY};@oKUm zqxMkj65(-04OhnAV1U6voJ(iHuLP%Ly7;tZ0v+8%7LKbDCJaorw|(h^_w~tnq|z>L zX>hg34}5I0pwsI0_laDw>e>1#rE=r#k$O~N)M|nEIqQyshlCJqVe(U=SjH&$`{qJ& z`Q%3OTL1Vit|GnY6^t(T>AR33;tX20=%Wo7nr_1DlL;m-LrF)d8Zik4-VL;8_(tAo z2SG9q4~Eeh=YAoRpyG2`)$%&Z9c<$y|#c9|g&0fh~z_^n>7AMb~Q z2D{adys_Y{48+I;lA!w(M9xhxkj*+{twQzK2^n7Umb988>mw%B*wBtAeN$7n(;vwu zg&fy=Czho$_65AL!@^;p)G++sAe$p?A6Wf^Vu?4BEJ=@npFKOsqGF=TqxKZ(y`bWL zkh^j9xoN%o-%OXzu=aM}=JS&mFlu}EWVF1U*I3PvzVZBmZ!ArSLHc&GnbV>+SQog_ zN3}J8ARjvxVs2mkfsI;Y;<13KQiu4z(~J;S1ti7`cJK`0DT4mvDVh?&l>g!Av&(~( z34=Qjvqz15))Go6q7F{q;RodKGH-;pfbFl)BD$lqA`d%xH~pp7T_^!SletO{kcz{t zghsb9wnpT06uqZc2c~tlP3SpUbvp&nz`qI<5@(ozTTOp3rvj}H4YR|AXYaooZ2100 zF1Y3|O9+iQ%Jdx9>*xsEUyw@_jjatecb9xbt*Z>Hg2}P4W+d{hgVeTe=~>LuOq)a~ zf9LMLsUGfdd9Pg}f&t76S6QYh&VJ>6$n#j?ppQta=JGa(Pn8-qxB`tFmM#`{TAy2! zN_m+BrFC`VJk}YPJa785H-aq3wcd$lqE$N|?-&=?0tRhq3xu?Bhttx)Z|PY&@cdDG zuQwhQO9euQo(=Vnki*3eMk1^>2hhH*h`k`dLhDs#efEs8ltfI*RCgX(6F*3vBMxre z-zoNkbn2teZ!;rYW`r9{mL*A-0?pfQ;`{)rn32!U7?1{)*$S=9%Hfe~{4?|8Cq!cm zBH#ag5li;WoZ0*4hyWb%1SrA@H%NSi+yiK=&xjT+-=@A?^34_OjWm}WqzBt*zIz|CP-*xmZGpl4-Y3SxT zg{as_H>KwlVIKcSv`uL&;yh)eL`8a<^vd0o+eX(;Z>|fS$fSuNHfnJZ2aS$?QN7ru zRyoT6QWrUTW?oS9PaNQCg( z*KQFGHT7`uzowrY65e}n;F4iy%YN8wULqdx!ns&?J^vyT933DW42Up zaagO|F{jX#A!W)aeCyRH_Y>!2kWl6whd;;)P%@2 zC2p%)cGcDVl(Hz-R`oUGxf2X&uaJc@<6ux|@ep>Aw!q~T4zG6xXtYsU8AV0skqrop z&!zBxt^@5GO^)1*Z|0{0z19gPJ}di%ukUK=>Bav9!(y|ZrsgSvrjROo_5zZel7Pgl z2!E-SW{&5YQ^A4eU7jU;szZcH>P8zZE^=*-XDvxB2fKG|()u3P(=(+M-faqx*%h7O z*(fMCt%Ku%ojYGM??NQL^JR@{(#3B9pH6$RT)$WS=^=@oqD)@bM z4)kjHZ>T}6*d12oe)C|i!B8>>=o=~@UpeM)=rGE&4MT1s%uyJ^qf}@Dpnlvp@JgH) z`sryXZ}uVFNZw#gg_i^4I~b;en0zJE{r9H#;~gWABmNL{vdiMXfXq5rql%`W!r=e| zk1Z~T`mQ@lvZKvI5DotsC>)MRQ)qAxicg7UcXXgo-{4<3XiMu?HzptXC6 z84F40$T0){AX!vAVtp}J!>|B|11@i4kC`CN%p3Da5o8y)KkkFSWrF6GBn2TOd|$ud zOGCp8WPe#RO~99SQ%i`C&wz;#*6Adfi?8mQh!jM1ldztG=3EwJ=PA`+wW~1!H`x8W zE^U?N;0=B;PS>;TuZ?K0ynJG=Z7kaEpoI|d)@l{w*J4o}kuRTumglQnKkO`x2&o>l z(HJ@we^Oqx?TBg1x%))TbnmbCH%-j=yqYV~>c=2i!VK2VQ>05lkbb(`lGo`+NQ)u{br zbgD>=m-9C4SDI0h#{u55K>UWovgWKwS@naVqk_C?6oj6Oogkh-!slr86My znfuDlA_NX94xGUITDw+o*)uyWB(@}h_=2n%R7h3(U!I@H-B6mffu@HKl$pSieEsD4 zZX}c}bVe#dI?@RQ@ri`Eex1O79N$aD!tEE~?2 zK>7((RcnUXYMG#3U+E5MtCFNMWz_mHk;X*)M5jqx{mrw8EeIK3+W@UZW<=zg*PCHP z6so*;nekanl#?2JR*5~EriCmPNt9u@=BKZ_wx;)S73yu$-YVzqR(el@aXh1LW$Bdq z?wRkydVWz#?-&9vlXn}WyO=V?u#Hud)H@c+jJw}Y{`kw?fbn8@j;PRw_N+j|09XWIq9KL|o@>t!Q>d(CE>Xl+;w%lh)Lfw!iGJ!lnllPXK=(rlVi7=9`LU}pE1MVqIj<3A8;)8 zz|VNI7+mE~C{ukbv?U2NxbsAmKK*rV?jF)!xn`UA76<6Uf;5h}{m~XKa>2^rDKz^I7CB#C4XMS#D0nXBN zm+F5%RTxg(^>MUv!aGM32&ASe+dZ(Fs_ajh%((pg+*~#PjJC}7!P;n^2$(fvrv5-> z28NJ@{9<1;Jftf>!wQoV*J%KYRs9-Pbq3npN%KUh%1=cge-r0K3~54k1;m;|o!Qi= zz&Y;vUu5U6@0;0Ol0KXCavljdAJcML|%jIBLVI&r>C(zhy1yleA)C75E99m;2}o6&doMGd=2(x&{P z#&3lK?#m`FM|9Byoy^lhAt76aQTZP<-;}Exrp4sjPQGw}GN71w*28J?Mf9G7vdzG8 z`=*JS&NKdgjsh6G`2B3arlcbX!9K*FVu91PhUMC)D(B4T58CfXpmD8BDwBvq&I?RJ zBfQZ4*M-Jh&%O0(0PW!wq=G@bG_ITh<$OIlK1pFX7238|n4*j{#3KNZVP&2T4cgU^ zC7celGBA6JtmUSohN?4=#Gumg((&>kK03{rjBbz()TmaGk%M~4qCpXa>)PhV_xm*z9KYrfSdb0s^!73L=!^G+`w zK3{khaD_?{4=FdKqMezv86q9L-JJOxck)w9oA2wS282a^f=kfbAEmzu?|=XU`E`AL z3-}z0K50-+{~PZ)a)T-$r_inX{!oV1=4;bU zQH{SU10t!vC!?@7WIz#eLh)M~U@l=GC^x*NGg5Ts5)yT~a@120eDH)0fM8)}%>k@W z!Dh6T@tmur&TE^;`Ab_M)*TCU_N#K&2l_$mQe>k9xjPgqY{v)@o6qLGwT3R5dcj!P z9Do{C5F7akYMBL0rQjnAA>%la0IZs5v1ry>iOjV=y4_(YXWVF@$(~G3p}Th|RuE_z z)R(bvkftmzHLM;QcPxX&P`{qebJLpo&g*GXbv>*8ss=j^oJ@GnWn!crE9!M+8EJ|x zxww2{&2`|9R-V1}$Cl)k?X(Z)(7LGT+~1W!RSQk$3j;42T#Zz^AA`FXw-9<$<=(7V zR4W+2uLZ|T?1=0Sc*~2Xc2bJgksWM!D>*s&g;l-Uj(4q3QHpKr^OeV730CeT%kwR^ zhrm3KO%ioqs(?-O!b2#2p_GdGuIK|ALBywDXXlvY|Ft%D z7WtB+8089gY1uYI>zD$ikqJ_e`ytLCyh^u>jlGLOeEh7`8#+czV$h8sAZmuT^ukKRDorFw@oCJY#Q@U-6tG*FR;pVs~JQ3YP z=$DUMX&w6`Ht2HywkB@y{4?{mCpj zHxK>2s=)^s@gUb1_}3B^pd&n(ZSZ+KTn3$SwQYGG23_@@`VPb}m`1jmcg5MUHk@I2 z)bJF?#zAp)GTnZ-F=4L$_#Yh-*tz@$705qfJ$%dN+2~j@kYc2E=luS@@STNxHMvfP zQHg$6;k$SD7(}pqf3LEEJ1sLO2M-SaviMJPnsgCux!8B{WB9Udu!sIKGm%nO2IFMJ zU7Pu({8Z2zC^aS~N2{0A3?09zO9I7*{Cd>oyTZ-h-xVH7#KUT`us^mwwuukeRp*`? zH7vbg&p#k#<3ALQ>vMN`%n(^scY&&SKL73u07}@JXwR$hn-#B%`QJCN1c7BAWY|I{ zD+uzDeTVEWhJ|$TxzzyfhKCqDQxID$vm_Xh+%xL@hBhS`zT-~MDhQ+|6ug`y4*&Zm zymoBY;Ln5>NL%Hv)YjEW+!MT6xv|%+4rW^Sv1K$6wjlenX$SQW+-^Mj4xiFbGvZ{$ z!gKd}xyD@A^B+3g>$r?H!cZjfg# z-QtQ|wW6^~+R|nzobhNX3ehwtU;eB0qLyJ=SY4!W{sZX38YB-VS0B$Sehm*ZjN-K1 z@_bg8$+5v1cTO8~P8uIC3S$x8nzU@DM#*I^!e2$z!kS^fvpRCd=l8E>VC}vt`QO7} zG9^%cX5CB9LH5r6>(E+7s#l{w7LmJL^gm`b&TXV=bwar{)^MyXRn@~8O9L>6(w6Au z#uI#N?eV1=x7Fwhd-%eNrhc71c?zML<^0K>t%iRWEWIZ6kLJNNjqEnO_7gno&K&u; zWosfqPx_``YO$&*Hb8A%Ayr^Q10_{3eApd1iN#OfWYsVpu&Fb!o)I`#3gZu4 zzL!gG|6pjt-UGxZE16ol*UElrG4Wcw#6nbVYESMqsUw}de7EQH$5J%{-L zx*uSNEU7qD1&Ex3lk?Jhc=^jJ)-`#T9>Eb89iLs` zLW6>>ntD)>3(mmQ8_HL3*@J%(78uUsAWtFaWgkvn+|?(cgt`krixbz7ON4+&!p?Uy z<8kxACrZrv^*OScHbc+9adbv~qRY0T&1?5X)DNqvSiq!l>ffL4ImrS#BWF0y5x4Fk-y{SF+}v zJI}#mZACY*0e>|}PWL&nl4>@+a{p+k4Do1B2*YhwHXdDD-$k2R<+Y}Nia|<_Ziib5 zSpjnEJRWd!eM~JM5IYOtdF_@mh*?`e6hvgZkH7r;2$*4GE7V2=Iw^%ype{+t!Zdnp zOgw?dLd^C7aD;6dJ-Qf!MQlg7bZ-fa?GuS1U?nGf)*r=jLGVF>`wJ>wY!I=<&<4Ww zl|Ba6T;kljO_<*L1PABFN;3{puuM_hoKw3dvR(^&G0+VM*QgcyQf6JMipk0 z-&4hr6WZh+N@#lc%pt$|V&Q8nVPd9?92~oJhl_K5XDTCg3<7-mK5y36vHP*lvZ!qT zmEm9}6>r{ouX^f{82jTJ_C%AM?r{L<3FH@cmz)h)uX2kC+p`ovs(CLDVfDt_(DI=J1~uR}UQj(jR)09()ep+8cwB+hXaJW47_B># znnT$6%dn)dmGGYculTx+7Z6}sZ>FW{?*9QFo`Sl%`tTOa8Tblr9eb_3I>RVG+zH3~ zOJ`b?kL-~8l&ZuWfVLeOlv}9}?;(Bw&^xWC^}r~krFYi&y)7l`{;j4p;3EVRdvxqv zEi^uSj%;_p(xgm7u93bDattg^YvK^HNXJwnKlf$tlI@5NRIb<p8Ch8P!ToeU%?E#l!a39Gs98q*ZJmT(au|k11 z7YTzHnhBGFSStNg-w*(5u+y4G{}vjyQe)x$dPlNs7@ z$Obx5WQwIbSJHySME|Wp;)6OZ$+hVR&RQJ{j`AFaS9%>pRf4#$-mvc=zow)sjww#0 zK@qJVYfg`pZNCy&(^2v)a(e-Bx9a}JLYwi-0g!l^p&P-*Mg)FGY~bcTWe+4=nl z?IS`a*O&%j$7vBd;I|m=u+jdD2L-N|h^-xDI47X)^xyYU-<|^&g~%JNN(As#x*Z?+ zL=BlE4OhVvnD%H5WA|uTi5c5?yqu#|P3JzuJ%5mViCvoHl}iKG4QbIiIgn45p@&!*;1+pAp+fN1t4Qxk_VX5~A#^5pa^3kr1b>(s zF*SeHqc8)oPF+79b$7to`c5VL1fLCtOgO+U!2m;7&!HYhgP4S~j7~rltNzB}|J7}c z6~hupg0c+JA6yZ1qbKa(yl5-WYa5!UiSoQemR6ngErmrzA7+(f!tzOSUQONz)E1AC zyIpwK%s}R?e%q9bC zAF$+S5sh?U7tRV+6v|U9jVb;1ZMW&&yZyGzNJqddVyG1O9^q{~crW32?x1fvJq1@c zwIrB-eQmAKX&J=*XkQ^+k#d+=8tT(QDgQ!Qp5s5MNoy@75`RitI>hV(R2h7O{~cls zyqH9aF9OEUFjEOTrglm`n1u-)&@U;#`h*J?*4Md{Qrr(kEdL!0;3DQS^x>ajk!wp& z9H(}J$>b$KcY=*Be!C<{gq*qHXnp z`uYlrr!u3{hV0pno{@4JtF~w6wH4_n*O|AZU;rZCX>l*(my7GwnEu;U?a}(9V9jxi zRiO%Luom`Mp`7j@RJ))ZZ=C_<2I!^o@HZ?cvu{ zR!18NUj359uw39adC~s-qV?!Ur2NBVMc@6Ru^kMy;bSa?X>_Lp8u|=MY`@U;?;~Ku zN%K4QLi;WtH0C?Jx3@9*6egFHXnb(PK@vI$Dk?I@TYO zi87Mbv%cksvPn=EafO`4M1zmGUNNE?kZMMtB*oX6THzx2Z58+O+>JXCnEsAqb@peD z{fB_U1bf|OU)avfVN;xufsM|PzkmGx5~HKy(t9`nXc3%SWGPlV_E(-Lm}(n;j}<#X zzLoXR#}9f)G+_V^l`J%o$@UiT_M^y5Oh+(i2H^ONHsnGdH)8JDNFGpiSrU*!zm2;w z;txXy%AMF|Efu&F$-z_P1?rZo|I6*x|ij_3^f&ljL6$;0U)34E{3=jdZz z@s?l{IGv=@5k>jXb^L`D?q`uAsLVm%;?UV@oV2cfBAZeTmr@ zPV}5lb785J`MmK#k^pbCi}vG$3$@6t>YYUMn(1c@o*!>1^*tur{X%NO)0zZDJv$n_ z03B1xA*7YaiVI#w5P7&7Ac4jDT!+0RCW#l{u?Wzc551s z@pzux2YNi=k22$%&;C*v;8;lRw~$y>&Io+2LYEm!OMMO*#51I~%ls=J?_MOm!G{?S zsJ-WD%E({UDDKsJG@+&6K32!YEPsCQuY?DNUF$|}{a%Kwl~^G(TNGE}er<<&c8TMR zSRXj)w^s%`LEA3@7ZO77f>$0TO(3Lr<}+M?ZO}s^xH7xp9!eEK)~ZF97AAj}i{M-| za_tv{`m__h?&y($j~}7=O5rD?Bo#c0vQYDZv+$pMUCs)}85#gYHUp!E$MdJz;o*vu zA~ZH}hbHROfI^_We+2tbg5Wr@*jXE!D7$wStWlaE?16kc50712X(#=2C;&6zaZKCe zjDs`bB0@0%w!S(G(+My%vGoE5(q*`BPp~lnEbN0{}Nr z+{+)|`XOtuTY+rOpAQ+u%ChP~YSi`nD(&stnKau~-I_saKfZs5Fw^BR!n~%R5|Gu^ zu|2G5v7p(KtI&I0Ny_tQV&K>_OwlY0JTxOaKG@j`Mq@$Zc_PY6(oOCPyt!IIC%p@_ zXlzIJ`6qJBEknaqxj~R;BSn1!FzQD&CUTx0Ux|5CKi+3 zhSXK45BTSKcd@L*V8oOg4ZB50ZhC_MpX@htU=tw6in5N3kl)Nj8%(CVxnW@sVsyrg)~s}F~6FIJmr zCh_dYaBW9QkP=+P1zooTc6wymIckh?${$_@Q(CcsaJ+HEwN>RHwFnOvR(-gtTIdG~ zi;F8ctoc%vbw2mC_Q-07G?G(hgSl`4t#8Rvj=^HqtF8!U__ooRlb3H{0@%TQ$iNk4 z`N4fe1X@Czn}j!XOzbvN2p^YGi11ec*Og|cXaOoh3X%L&LQ4Ah*w`Cr0?(a0r`J{X z=1uc&m^r2G&Ph-&g+JEH+f;xpEMj1E+aINUPskEzV}0xZKPy+@8cmi!EO2pVZo3#W zjx<*c%O{Z}N6W1B(+hWY-(=!^Gd((pb5CLm_esdxQihE;DK+$zfybJ>+gv!y5A7e1 zm78_k!U*f!BralzLP)vj7q3n)gYvHisUqY+hmg0TtcWXgFv^_@yG~Z1Wfa=V{)Gjk30HJWj+{(VPOBu^qgq9!f54GgZRzYj; z3XyzSxDkfn1Z;vA-M_!Bo3i5#X#0?WrD_s*+a{&0rlKHsk2-b&GsmYIs9V3&csfuT}t zSh$x|VMa#FN?I1WVBsH%`~;8}e;#no8oN#(K;zXO{et7KVRm1&QP!)amt{}aq_%T} ztb|F{X_~t?q5zda4GBua;hNvTlZTBairr!^$mrE`8$Qfa$gv6nkkTL5@SABHrouat zG3HbRq25@SM#ubfe=Pc2iqW!dS4SAnGd~#H^y@`*ElpJ>yi_qz2$p6%OnPS2J=!Q>SYUz=y zc7DpzZf8z@;A9n(|IxaRxECW15zL#ZFmkT}>0@(Qby4R8q6Fc#>pRTi?G!B)&w(Hx z&gF6=vj4L^JC)pW@5@gTNjWSH7qm$c(CvOy&E?1>y1e!WI;DK}nxsFk6$~HtYkq$U zk$BNc-R;oSTh#2mYC`Lwx|c1b!zs#J8XOBWZ_`EvNawgq1px$<8gmZ+rI7^FWeN+C zm_7D95(4TYc7dRMane&wzz%*27w-Zp)sYboB?e(1Oa~H+(*a{Yn3h7EF6a94mItp~ z29HRI^0oD4YoSg{(YgfA`Ro6v#frK*<eq@DI-|N$2$NO?&Rgq&01EDe31D_LKjQ~(*-yDI<`0K_@8KPZcMEGUK2 zh;VBRxqfm9q6Qs=nuF2D;Al7;MKy)(SkDi~MM)$_+97)rpwx$8bOT4SL30;q z;t17!n6mbGmm!kPV5p>a;jyv^p=_) zC7&)%IDZuZ%Uu>eYKcrE(o%muIO6gc==igKJpKECZ}agHtYMikpkKJP=`8+zp;-f0 zJN^p(6SePWwvqw^v9=nZO4DdbLZmtHFA|xPlaf-hp8R$&hHnAs<}7lYK&XY_l=2@% zu8rAxT2$qt^dn*^f&3QIrV^hdqgI}-rNYVH8AGRqVSmzRFP8`+VTml1^UIg!4_?>R z(M4gu#D5_4lD~$&Uj7}3gQxbMs2NAbmBm8Bg5voleo@*?(fkjs%zR5qde$Vq=Ty^{ zf$~FHcLnczMJV6A@9kBcfJ(*J)4Z*b$xa73ZZ-orZyaCzrjA%H&}w)du`YYga!T}^ zHJ~bnlRrv?%p?l71AAJUc`}|* zz;BveA)74G&BQIx(Df1`_0%nWCN?%0EQXe88nlXzBKwUQQTthfnp$e0S?lSX8G&(w zsPeg$@8R6bcx{?0)#NjD0TRoev~eMCMSZ_xQhadioz9>7YXAAhqJtNt1y*a&K<6xGlERv{dl_H38`Th4}l*0jVL7*dSk{ zRLOaW=T;u}rp0D76l)^Kq}RoM(r>!Jve=L+UcWs|>%7OG_`dux9197&laY11)5VvV zDrYru$MURrL002umls}|Q--P$NI#8lO{(`CxMjf1%n(+~4*pkyR>TdYz(*4i6WDmj z26H}fV3nOlcn9d!aj`zStyl&u>k9$gE<+y@{HTYZ4Jq=+u#r;MZ6sEv+!{o!dnUC6*ihYmE zz06)8@nzZp9M4M$WiTCHX21KIC*N6QB=w^Nq)}j+`A>$z*=!%eU3PJv^RY; zvcNiuP&Xh*U?Int%G5pl5!#b@7f)grbFkZ`9@v3p?BMY#b5u}cMvdN?qDaj*il zlZv&lCuKTaI|6kE?1Q+E&(}BU94+@6YZCNTsWO_>$Bb7gU9+>tB-0ypAK_t6os(E$ zP|SK^ke(11cQ6rS#sLO0kih=|^{TKX2@bsy4(ohEYax@%KlB0R@$YdFg|rh1DEHBV zk!O&m#>RquGJzyKM5kdT)9>jD;Pjb82w3Fz*G#lUiHKfe;&+-?z?Hzy-RHb=3Jd!n zZ=MG>z?{84Om9->)vML1HF~f%Nf(FjlI~@-=EA6MkJ~XW%jY=QDETBW8j}V~E@}L3 zzLZ-2;`6V-khkpjT#Z_cg>VB}FbOG!!6ThsWBiT*?R|BopAT>6R|FmE+Dlo9+}5Tb z>vzi5VIcY>$up*V>8!fYd>@)~nELEcBE|J)OOiY;+r$q)*CG->Hl{5)`05r5I>tEW zyjCt9!j{EW=;GdyBp0nmoe(Ch;eba2352`{!17s8Ma16Fz)H=31i~5BdInnv63Pi z6`5IB0sJ+_oGmf7*Orid%!h=z;;GxKgKQ#V@c-M+nM zadYeJ$;-7!iH`G?KAVpI*NaFtuuX*OT^2UD1(N(DaE&p<=K6UoSX}lT-0`X*OafyR zsP8CIMAh<36Fg(gfzhiuc_IYo{IOny1bpZlmrVF}za}|v5OY((HI)ASH%I#=OL>=R z?Fa#%R%QcXZ7LmB*t#?4q~EkR=-nzudiE|ezt|6!(@9w-_V(2TQmhbA69MXzm|3@Ts64Kso->0eJ3Rc?}ogGAKsW-ZmbNhHEAx-rY zCZ!9YnYAGF5d0wnshG94Rw5@9tlozqM<}e|x--T@8!QnT&_dre4V*QERfqWOJWP_> zU?%bI$2DN`fHIRKxC{4nJ?u`wg2gB`gykk+V#F%pxUGVDH+ZhqKb8y*@{=14Eb11JG1Ytg%-l)Eg;3xz z!!{7i5IjG6YXvkVe?Z02!%<_7Kf(sOnLkAc)$kOtCS%vk^rM;!#xc=%W5YVqi^{ba zyxNqh^}#q6Fg}{N;0d34#vy5>1>U!*kV;BEuQt89jWK;e;C$3tPGBd|6l+JHCfk%{ zjim*Aya(7G0M&0W->30P>GWVrx2rN{c%mvor@BE z3|7o!J}a|e0uM)1)T++`XCdpKoAuC(Nm+&>R316U(bW#WL&N)N@W$tX!52+6&|Xc@ z7uaw_FqObXDzwWXJ&nH^-WtoI9%(|S&v(xYqW=!7HAzVe4%v5?m9I`JQ;(-E@p@&3 zxm??~;G)Ti>chaU37wIbzTfM*iIZS^T;@o(MRw;CH4dxYypA!QE;ml|9+yT4>hSm4 zE$=kl3aLCkisDGE22Ba3DOp6;a2i|ysKFQaYw%fNcVzN0X$|lY!O|p1tdXJMvhElk zT+(dxyCq%lorECz7H|zkAhSNNUm41G;e2o@`UOFsh1-t_Je(uPh14w)0H)x1^y_}Z z0xttxY^NCL;_+AOe#%S^?c8sd&VxD_S1w9BKE|3(-O!PCly6h}!EP0y(2Wc=0k3jq zrA~6P!atDHQsK(AddRfPy41BtR9KL?rlHl`l!;~U1F z)3P!P?zarH_6xn>RW~UOzAPw0T%1=Jg*kVcX*a#VX4LqzUi@6&rh%5$7uTCzT+p7v zU*hX<_H$j)ZTJ?Y!9;)fo#v&7XN{aZjz9WOvXDLtk{XOzo zd-WLZ&z65VZyT(f%rLj20e>`K5p2~)JaN&DH~jEvwQMc*B9VF>i3chN<<&$|h{Oc; z00snXf=){O+u+W|Bsu5?EdV2nfZDF+wl&bSfKptjlBgcUvj_Ah_~&#yDnf%lZ4J>N zZ4De9Sao?UG|w}59`o2rP=y7?b5rVHYeM3L<`Y@l1B`sey4}{~i?Zfy&aAxU7D3=$ zCYP|ChHlQa&iFqIKRlI22Q9qP6c_mcS8^J9pEtpo(DKypJ~ZZIJC z$lOfPH#Ln!L=r~tYFH=^+X@?Ntq#DN>J0M>bo%#vT>&h9Bq-}=z-S_cGP=_+4>GLh z6rekA0ZnEiP?NySKVyks4q}@KWF~o?zsS=8WuR&?TD1@@t2XXfXZB0Hn2RkijfqSg z7t&oEiN)m;#EL4UFleiLXMKfkwlq34Qv14fbhMEC?7-<{M#&~!B21& zrpB^)rzDh!o%WCar9P*4g;**yWH{_c=F{r(Uq^z0VLDtl(qe3esuiMy`?eJ+-LL5a zz89?68J}Z@{UO|Q@N6#Z1r zn25j@HP9m?UikWMOq443qN-b!s~YD4?9YS&?&JRq)-7eJ$V2pk1WRSDe%d7CDKP+k z=Ab{)99^dnnGk(7Fk8V2=cUl%cC?ry?CeQlxq1g~*le`um z;e%#ngW<=+$p&HuM;WrOt%hB)$L{qLCW+Gj@E*g^Em!P?{{Ywu@DHUxwr5)EgW4Om zbV|VIQU&8=1Y`$&47&9u9wN>c1Z^Bo2~M^0Fl4tOS{dS*Le5hbBqtOeGJ}2}rdFz2 zBK$F!R$Vj}`+uXI9YnN2p=Ac2M^WpuAX|{$M4)feentZ>|C4+pSPRf38I4@bTk+71 z2I)qlo?W1~oKA(g4)wdoFldC?awTQUA;_K2QB{Ia$vq@FV6;DIKfH6pfT7%J@h3;w z6sC9K;-{GkYf!Uz!eh$^Xf8!uyBVDI`5bc(@8htvO!R30;j)lip$H6yV-*YCG9XM@ z9HkbmE6HQ_5GsX_?r@OiQHfa2hwe63y4-qn0Ljx|6drgtU*Uu zd4TG&X0TrXCF*LSNxc}`dYl&e2kQ2etihthO<+xCfER<{1lW$|CAqI2(pU|oqTnPk zP#iS&laSI3iccqCfaex}2a)Wf4;QWlSz&U0>M~%^OI!T}3NVkqQlPa~tfXYO=sEaSAL3KGP|a?J+8Hza;vmk#Y)x9Q*g$4SQhr!g75x>+s?#vaw8^EYpwp< zz=WY$0Q&1AmMm(*JOLfD34iO;O1_GT)WZS=HXsLdR>8ph6c%bic;LaNbLnFI3VSPH z(U_&=EIi-;W;*NXG=Q2@gUMti~s`~O+NECLP<@=SboCg(cW+1Wq$noQLYC@ras$2O0{`2mDZAVq`$U zDKZ~O9?kJ~_P;>GSU;8g{XY5j?c0*;Fn)C(#aoJi_8zOn+{EiDMRjVI+IZ0WGAAl0 z_je;7ZB{ex+DX8(t8Pp1<`k3f2*)iPFdQUaNKN^$n&h*H_#PTjvLbo7&@w6YEqM30 zI3ec0_B)~<|`6A|VJI{$vYB>8Wc&!5i0+&Dv1rng8UYSy%Ux`R);#`0VKF`?#&4z#C%wg4axn zM4MR&18ZSNvzEa7F$gC7a9s>y{w$>XkeisnclCp%U6Mw({lG;JBokVc(TA(n$zveuM^@kLJe_T*Y za*lqITysF+dMR`%3gCg+giUF5XcT!J5@&~djepyzty2+5i;03zA@|J-bww^Di|UQ_ z!?Xf{Fu@9rCRonUC(lDk(rJ3UbrDv6<6sr?b)xVDi=(M;nvy1@$W=Fh-nHw2zztFe#Bb_2#BU>w2c`s4mp z%Z$3Z>E8%o1jW}WGJM6u?*vO%IM=x;bPWZ|W4XwZ7%y$s);g*_#OkCvtUs3<Wdlv_pX<-gC59u&r|2Iqh4>osoT)T6$6q z5Wsjqw9Z;WGA_`e%N&dULE@ltZy-q(YQyOdwP3EUGi}JEzLhS&?10%#er^vC4J-*m6zu@ z1lXX){%!mxfDUil*{J9z09-=Q5LoKqx05g~z6iH;hBR~lEim6lm_ig5LBA7n|09tT zauUsOKo_P9=89C_v^fUh<_lC{?fiohPtm&w4+U0)#Fi4;vIy#rUB?l){Pc!_=*S07 z|DmBDg8!1gC?Txl)3TMsZ+*dsp&X*o$o)6G1%)Guheu4I3LyqNN&45pu2>X=*&&oK zDaFpX()V1O6sBfpK+2OBFwe=51(<${+Oy{j_uM?lAJ?T+4~Ek4=c>gFg`JhDky#Yr zRLQz=Yi^V@)gd;7{r;C=qG^4yB;WzRcDP8y9wSWjX(CLI(#;@Yxb9h|S%}NWGcLWU zWp$v=uB6@4W}cJ32~;KsE2ouw$4%M`qJQ+2aj={ij+v#JZC5nN#nyJ zg%#KrJQB*Po#=`W*OuC?Twz7IS5NrYU6M;4PcFIlT?JMiC zSoIOyh5kYP&V@8!7y&Z9u^HNgf56U3^~^kozh5QkL>+`r4z*Y9VU#FocMNFIL0kS; z*{TUSM5EzUye(rtocnme7dQ2Scyxq3aI^xk zEb-#5WiqOTDz*8WD{;2v+1#!(GD}O(y48fg?k7#&snLLc>%9httn#0!*DpF>|ALuF zRX}a-qO-(-;qeQeyfPis<6VBIspZ8bkVQ~O&IN`Pn=c!YYL;Tfv66;IE8)Bsk;LEl||{cjKt#X)w) z7_dP%&;i)rW9*?tc<$-Ef&%TS>v^y&6&nNLEc5Pf6n3}mRSNJFwBt{&y2kPe>AJZJ z;}p)Nfik1A8#C(<+^?fo(XrNjQ7$nHTp=UAWTcu#B)dc1KKLD@Lo%L@>e)8& zr2%$9-j2_&Aw^6AF%{42?(0IiaAKg#5T*?+ThMh+=++)^jcvoP{{T4gHMW(0e~&Q< zaDR1*fWg!F{fxoF^Ghf_T_?*rm<5}PFXWZ9@LCRPTG6#ol+xmmd@!?al68$0r=M=T zxNA{t|EyM!A?{)B5QhO}{NX=sH_Jxj6>;CER7d9f_Fcw(_{M)&<1)>CRO9HHxE=%J_h9eheFZp?`;j?}&W`Sd zjdJ0x(XlEq&4m`Dy=OUSEyg1)7~l((|XhKS}OFE?3I=`IeP-U(JPVu*R((M1T1DojP(;}H2=zSRJG&BUN&oyF8!$|V=dW|mS7?>i(#vOz^pucUe z6KRJ%C@KDa!ZsjH&ggZ$7+|MxfF0TuY_zfrdbrwI{htW72{5n8FA!dkqv={xYF~dt zRf0uFd4>~jy0775<=v!dF1T;0A{Q`wF8ph$#X+Y8wfgg;ChyzBPG150@)&`JzfnHv zL{|`!>7Ku8I)$^ly+purXwhO&{tw+z=!^a9kO9)sy@9gPp{YVm_unrACDG z9gh_=&Yg{mH~U^2X*<9?m+jg*q#Q0ZDm?mMXl7k(d1-di{oC(sXplaMVrvx|!Ul}| zjo{0WR(W#bpX@vI7D+t{XK@vCDIm*nZYZb=(MNNx~II-~J` zOPfR-4IMsoO+RZn;mwtg@BufjaxWszUw-t39(6S#)Rrog_V#pX z_H@zqhx~a04L_sSz9`f@C+|0|;3=>cx(DAnCZQ7U`0mbosH-!4SNYEYkZd-Mgc~M? zjD98@-uvF!qLE*(>9n8NK3ws2?TUK%8-u0m1e2CdY!}>5GrV3kRL6fc$|m0wOOm&e z1qF~5+2nKj_K)U{+|7hxTBIb>3#FIpJWrv@aF}5By13>)e-wi{{`7bcM5tsuuv&~= z%6fzkDF< z^rFavA)2u+29TlKX{!n^+$aUK)w$EY^E!gaDbMsWw()e-btP-Os%e;_D_2jas?<;H zpZtC>TxanHX{g!ixc&uE0j(B&*Jh9<979=zNo1v1R@;nf6Wm>A_e50+hFPC>w0EvM zl^x{JoU!11L_S9Q@eNzB#KD_Q4RI&!aOYNl-l;9Sku7eRUaVSSm0EPkShW9WuEBY! z;>uMHobQVDC6Y#H@l$f$dCMUFplhi%cN*ordt8~UGoD8N`)KTCF(Y5YXW0JF=3LanCic536Q`#1vlUC8A_E_6TyU# zkS^l&@#0R1a?nW~qlMHduHEtGR1+{;6nZfGVyr;l0(|+fI>b;m=)V#6mQh))QM)jy zluCD_NH+)qk_sZ-9STS{2!en}ql6&cEgjM+NSA_?bgLjKAOiBu#eUCs&NzS082jJe zZlC8_>z?`v@ejM*FNGiPO(IrX1iIg=K0V32l z?$;4cjQ^>r1j8CoB_G+f7(YfX=u^n~KxMrJZr+4wF_u`P|IBVk3;hG4z!??B4`uzK z{t$;e+X7vc`14CcyMg|GVBQpP+^B`QM-pwYlqAwHjVOg|Y$yRj1K5f!QTStVZ6{Rf z)2e2DDAt-v9|{t*XuTU3J<+yLzce>9fy62AEfgkTJ$SqO|y8Ln$sKt|n5X?m;^EpB%bX(U!~uSuT_c0n*T5wcOm=Rk8hx= zBRxu%!80ZPjA)9U$8igv0Mp>0oUSD7LBnlVf$h|iS2O7EC)_Ls1`V7il>42Smr(}; z-9dy;!IMisC6jFx;R|@U+P`W#rep9jK%`m58&45%TnFc5wGD`-Ak=)QTkCaTdCUF+ zl%a54*iI@xJUz(=g#>GY_X%Qq#vh#|0h?p`%c$XWB2|hLKJ>&Pts-Ky&o$p~2WKrl zuBYgAn+(^g(NJK16Y=I7C&L_Nv=IvD)sV0+_EB19eRO<<&Zf3AxKW&as8G#R*`W0V| zD0OgZ12~$JAK2DG#N&+c!@+3h;Lz(baa;hZPCRk%?Em2ol>OrPF(NVM(%JI?eTt|> z&>(QWn;e=dyEWnw<7Qw7pk$9_!;s2TB6#P%7^tw z_qOH2JrAGqUDb&PUH_+!x3q7vrsCFKUh7tSY2!O9DELIgF^Q1iMD7KaoFMI(s>6Nu zgQVseuGBYMnaX*}C&M^`(ylXvgVxzv1j+too7VY)b*0l0PRw^|nr~TPNV;*XdOKy2 zPt}bt(#FF4!|)T-Wq_uCf9>l5s$6Um_8@%WfJkO>4baz;0J*PXf?u?NFz^)u zq;f6P6?}lgQTA;PD8cG$N+?qhv}Rb7vZF4(;E=JI|7~~q#KDlEy0%v6c3x8-1;4I6 z1h9UI=KFgr1WxgtR+`v1m9Zp=&>HhW0m05Ook%PCR+~3CFq=Bh-S^WAl`s3-+;q6t zuG(i*Q=)Ymm!qM)jRKuq|}y`A#lJT z7v~QBCXQCPD=LiD`uWGi16C#PT1q)i|`R*-H4^g5v|$YaWG zw)A^eNK&TeCgTe#N!ElrWywjU|JmC-j`?}K(CE$%a|jh+4u4N-FHL3+&1rHU=;=`a ztLkSku^~kW55YAv8%r#WM>N1MEmmW@X5QGGO!w<8MgVDY&KVhZOj6gA$5FdZ!rePE z&V#WxlBs8@5mgB_x>X$eQISnUgHhbc)cO}D~%+joHC)*zS0s^^2!$t zWt;bo1FJs%=!cJuLj_$+ZlE~%dp_bM2P5WFV(N8Ftb?1(Gtc~mJRq1P->}uZkFduc ztsU6SDdT^E+nKSmzwd%aE&N%l>BrQ6j#mINqmgjBq9g&;#^?O=Z!0YRlugq{QhO#s zwJ~>;vhtu51~u)CTWf>wM_=2iZ(Wy8)vL;PIq@nMn~_cS;{yMcwh+zw zfM?aApfsS`vxzHu&=SBpKocnHswEN|pwj=nCRtM>f{j$`pgHsh-McERUL2N43+ zn1N?5Nt@q5mj(oba2RcuLC7sFO#J|k6^MIZ!E$*9r#p$8aQdiHCF35lT0s9p?a?J@fBVy=HkIZGDAxt|n)*av=si1+cf?!Na~f z+1NXEB`MdMStJ{Xw^UJz?Eocm7>m=&qYLTPN+em){;%t;~XyzRrwp+O|up8Y{)M8K+71w^F-u2;_QAnn7X22yG z|4TbH_il{MdUvc6ir!7%XNH2L-Ei{JJn$KWHdmVcD(|6>LqmXDcYS27LOZ0ulu!EotWeANQF*n--UyQ7xqWltFaqQB3eGU zY>|lDZm7TN{{{hc^9RJcW4 zS)7ZC0qPr?EANMR#^}{)g%01{L=5I*k5W_P2X8rtw8Ohju$_uv?)-gB1 zU--8oCVuv=&z()=Z>;&hZkZ02Y3(hPuV5z#Rn8^ZnkOJQxi#h8?kil^!<*C?(P<3s#_Hsp$WM&E8`GvGi(c zfYgpa+Ybjo925$q2Fr2nth-=NuddyZ3ha-5dRq-dACJV+Sf3yiHGs$NZa>ig;|mhL zA%8y1SUm);a&i7-j4NQ3{LD;|J{g@zm0z^p{pE~>6Q_QS2AcX_ z+*=e@d9~zd@4Fgz!b+%gYWw}6Jz$elZtXf?99sdt;ie~@EV>m*yELJ znZd8>XDidjA9a899vjO_`c>Keor2K(`9!XqB}FEP)nkkv#riWt(c65r2M4CiJCI8O zz_I~y=$Y2B=YdZZ031X~WZYyZzPku}l}O+}Vfr87Qdxd?ovyM#jD+@}5afV|fV#qe zl!?twpJxSegfa3Mpq~ttWz@EE1ruY0hw-G?fNnXetZEZPcSo@@9l`r){QPRvi$RBPYT$mSGXFWFW zckVIx3u6E^ZJnx!_xX3XiZj!u+*1uT2Z7#AXFxpmoetg z3-S?f`RTU^{Rl22sVgNG(LVA@e~1}#e~+iUxD~j_0CKV z4Qr%6u&)P*xQ6X|G8ry%<4b>e=8j_qX6mP8VMZ+Qntnto08sf!aY;EU{-%Vv4vSm` zGaIMTeI)HGoV~1ETvLCyz(Ris%uaMhN8A9R^V9WX$CmAAjw2^d3g*dq$AjrcSD-m|_`UHR#`7mj z%5^$5hq^)NebLT)dw>bX@94ceK$%4Ee+8vH!UCosx!+nxT|-+rP58Q1-NLv`6@0lXs93zxT)&e)i}Sp+$@DmKKY`-Noz^$5_8DD~*Zn>t z(By4R&TjF)&3k;2{(_WyKP2u7fT{?adGR^y3lYy`#Cc+Z@4#~(S-T?aFL-eH;0$sH zu_XhKu#STCqRva~27cB>VC{Zr6?&F<$&tvByHV_7Sv%OihQxY+yEe)fg@hxX{Dv}c z+g?GOClJRsXyxTE6-9{K0OB>p7=r~5wS;vTbrEQr-Jt${4c3t3$rH^YATR3#u2B*( zA%;K5;}9JJ`Txr^#&BXJN&&G=2d3cGpS>ipS}@9kL^jk#h|_|(3U?O*5N~oMs4v6Q z??m1W2t`CT?*P3*Sjjb{=7URJt{q5k;e=5GiMhXsP8kxZO`y-mh_>Xvbt4IN&!5Z4zo4|V94;I`OBzrRO&n0^|`CYK-AlgSKOJ82|-}5>wpJx8S4Vl{w zgkWSDd6TmZY%`o7eh)|xzzEktDui`$$xH&IUtK{GA1=-00k4>TCMM}={s82{GBIGy zgs@(+U*=D_ks$uHz(I>E*+8h2AWjKHZp_Z5T%L6#788iIl#ugiGm8BWG2H_M&{C3Jy<^%5XL;sj>4GgzP8exafehL}G9+l`{e9kz!nz+4~l zoa=gk^%C+HXm6)Fp&`)$zt7>%BKdh;@Yf})8354@A0#q6QYx{6=~ZHd?_yaLCf(an zItVatU4O5q7sA7UMx*`ZgVlwm;*D)Lq3_*pbIrQLyh^mr`;F^A;?^o9nN1v73pBgF zw_G+RYHSm6HB$Eo82dv{nPO&X2AsUvwUZ-ol8y+99%pdcm!n8^I$7>Ml*fW$u@#&m zFa#G!8Ud?O;ffG5$9o za7{Lp#dckJANoBM{A&V914#5J$Vn(ssGfy*YUU*7G8?o{Ol|Zl9$P`@*tK?RSGnuYMccrE8 zK{J6+_u%5EsH?P}5kjbu*m@cHk=Vbolvr;$Kw?Z>SJOV`m?T&aL*cD}cw>~XTY_VJs(6I zNoE|Ht^pu4(F8W{J<#DH3^0`Gt8(C${lD#KqG+H&2iLA}C`|8md7hT)fb}7L(JnmSl{gLMy^4^&t;;MKV=EkCqbOx`V(f#RT=gP+%c7^c07Jr6?)x=?uSe_u>D9iTky|Lc3U zQ&1Gu@YKV@BXUjObd8t*Q;iryRti%JQy^Ua^|Dg)8)n~eE4;8wR6#e0hw`o=6sv6iFY^VF*k zWwMsaq2;PwIAJal%3lZyeNkV^QfH(4fa%~noA?(`Ru=9TihnvGiHUj=eDFN9XOB`a zK+yHp8R&AZ@V!uWdbojYX~LkkbJj{aHPQTKkXLithoVb{KQ)30W?+A~&g)idVGLh- zSMFmmG@xr}_zAHMp{&x>BH8im5xvaog_RuQX6f8{?apGnfkWla6Y*`NH$n`sn!!}_ z=;9M&cOVvlS5ACHo1Uy4k1LSks?D~2FBW|a#aC=*+Z15vNoyQW zyy80q_UBJ-CVeZO;f7~30X@$K3cZr#_QNxn15mfiuE_-r zc4HE~GAN&mBpoM|^SPEg5|MD{sNrqtdhO+I-OG2)wsDYE+?L1jW zM{XqO8lf{{9=Ab7a2nUrSt`HyTmC5WDw4w~WI-RxIS|%Mq%?-)lA@dxH{QK_*J<(P zRkYnZ&OCSSgdkJVQB|5*AM$|Q%e`s*zv;g*h=OqBb0^@6f&Bcm7Jl6+N9Ja}Hvw;z z3vgSF_4KSi$E$L}J4TOc#kBd>sxc&3kDk|&`Gg1{DLW}#Fhg2rns%Od=O?Za$!i={Q=n_{n>gd zVc_&JGL8^Pv19<2XixrK$#jq03S&`MS5Q<@A@ebnze$MLIKJQ(jqO_mK5w;NLs_AV zg^39rkOO-gC0-vN9U1M1HGB4wxuo72tD~<~Vq&FkRpI)w9P`KfQWlJiHrkb* zGFd{O@#JuQ1XXoqQGj|%o)P?&!Q~=C+szw=KlT08lA@Z$FpnRvCs*n;v3GHIjNK(e z64hnkGeWZ$l<>8%zywm6cVE|4CMG6NU{#syRc4&dZt(E%s5+&@5h95UItPDO0-lzK z_QuLGzFb?g5Pf$&QM4l~i(z4)cJ$Y<1y#d`0l4sfxsT^mR}*nR{}QoPUYd6C`}c>P zU$~Ha6EY-zk zO^ny{#OsAt9Yg}e?-aD&JRu+t?RtPu&=s4*rH@`0z;Z8JB-o=YCw-EZ-CUh~f8~qH zz`hfk`GaD$j7-eI2zHYDLD_8h;Vc(?b$1WH<2}`W`~qJ0YLN;)*bj8g!LwADFE3GU zsDX3vPgV6pKU>Gq_Omvsq9Sb{J2*MRVh4k78;$|LDqKJ8!9>>O4;f=5nPXX}2%c8! z<*Kuqd?fDV*#)OJG|7kVgKl@pEA6lV9*sJs&q5Uf3W956Qc4PYLuq+=`Fan?-P6kIG61;~6#pk!?xoac56GU= z&}LMGY|b5jNFSZK+N8+5;@yy(PlCahrm58xBS-tgYuaHPe1T?vczMY>=rU9{=$`V; zS?&A~WwqE=Sv$oxyTU!=95AIMt%X8hM?c1`b*SDIUD&8Vf;w=od#$!z&Jw))vkO!wp+=VedYU>~Y{L<06_kBKA+KXR6IrRRMW{3bn@^CYs*yuEv;P+t{@pt7Hq06-%h}{L0D1-hHO@Rbkmc zJXT-hYpGvLFyE`5<6)0A>I36Ory!}0pGWO4+PXx@^u$w>F47O{Yg2kIr(-)I`#Z{E zp@r7E`4DAyM^#C@*Zjnw)(fZ*I*h<{08yX&-IQvr{+EZ|i(dO@2h1N*^GNB zxNeWZoWX;aW53>0lIgtHfDNcNBlj%VzOJrr;g8BFvHeinfdfEeTh|YtJ$sfgi!e57 z-k9@+0V+l|3-UKWowsZo5O>fPes|*&mP<#vZQE&AS67dRPukDuGZdL$aK;vrvzrT% z+0td!IRz=Qsu6d%`IQY(ci0g(MF(IAbQXFp*aW`4@S+Mn-s_EDBc@!ku)`$N=aN$+ zJ{8mdI{u?HH^e_2vo3c zpJi;lqt)q@FPdcQJ%V-TxXU>IP);EXTzS8~2R1a!{IpZ)zED!%>O@BBY9w^-l`J^O1q%=5j3$-TR^@k{If8}#<v^He)4gp)RY;88hmV9nQx^7BJ;Kfd7F8Bv;~=5 z)QR)x6eHe*Rb|2P4z^-ohj5Hgt&-Q!T*@hng^p@Rzf# z9h*~ZSTx=~hVInI#BY+jEo4HN7N#IvSqFVV(vxL;eR--%tUms;Y{+0T%4@Fci zCEkXh+Ec&A>%-3xLQc)CTU~Js&=zBnQsyo>KxA|<1 z`%Y8ZcrR!4^w^lj!OBViqA=bHWs3B-xg}xh`~7_3;@o7C#+=jg=~D!pm=U#n`c4rA zq0PUr-vP|WE`+<(0{LH`;mf{sRrUcSpgB*SRi4#Uz%Lut2`4qm(oT;nkK%5TVBmWe9Ad4Ps^lGb$=O(;hUWn#r0wMW z=e~G!v27mX!^~-t`Q!)Vm8mN+`if(z0!y}0p)AJrCi#EDo?#`HQ{c#`=G^PWNkR>h zJrCVbZm+AwehQiqw6wIeH}iy0ZER4vz4{()`5E-6K)JyJ_w1was3s5(897P-m3mYE z0w~(LyEXV|h#1IdAQU}6I<&W1r*;3`_nn=qrrQ|c_K?l;r(dTv07$B65bp{GuW;1c z{mv)-7+`J`buYdPYJ3DY-|)bd#^T`OLSj%c2Qd9*Zis(%&{HJFV37T&YeCOHza?oo zYYwlfMsjjT&%GmFSl)JC zxB1!VpzZ8>_Jsz!@VNE3#S>6OXcceNYajOCY5Ael`i=8aDsgt!d|q?-pu+e3`N+Uc zw%hL%rZ0xg`L4|p7IptM>f(#Gh9ZFeoQ6R$k5eY`4HL#-{2J=(EFU?mo!DkM zeCtd%7(%F*&#?uukQOtHBu)(0e$aQLw9jI1Hb%k~m*UWfjl4r|APln80ozO~|HHq+ zqHuW0mcgZ@2bou<2ApcY1CJ9qrsB8fOEbSC(sZ~YX6hi`8G0XMUip{oWzh}5R%$QJ zvVkd&o**ZFZB(Q$0oN>kS|FHbMM`Th%uqZueyu?t%QyH5U)Jgfrqf&hfYmI?{DPi1 zBl*Kqr%xJgTf|OJ?&P3h2gikl&URTZ=>8B6q1$C))2?Y}={9WH40V|)G+L#KKc&)< z61;bZY~l}2(yza4{GCZS#P>_*?F$!|PYi|mnR`ekJSr@_&MNOhKQ0)d^5Un42Qyah zh+tGy)CvT7eL}oOps6NnUc^Lv05Yxh`}p8k_6)JFs^9%tC*FGu;EF`*oE$_l02XN+k_ehqLNMgyEl4EPedAR@g-PEOvwyvzum zaslweqPKxFKCXV4f!`z6EEXKR&h~Bal z%1S(JQwAr;@x3NaG+N{QRA05v;bQ;sK*+b8=`u@(&-G(P(!72Y)AZ2p6xshRingYk zd*s)roK4v9`H9g=wLlDO3HPhnsA*BuNUw11lW*g1Xd>`90aYPW^0v%gC)i;6&SO&U z>^o0fwut7&cPARnN{{xHB=Z8sZQWbf1wXx2u-=FOium=hS&|_@?yL z(?}4Gtg{F55<*~$L@OX9q^q7N#|N_7-90DAa`*NLKxI-3KO+dZpi%z)dohUHCoAOX z7v0O^9c(74M(neQPS=k3G*3_K6&B$bO~Q)|Lw!0po+6U9KftLB{`gVqX6!@2od!on z5ElDURjR0Edr3RQ+8OVK>nsKmDer!GF|m)3!^{B7K60h50&khRl)aU-BG zv|4%{I65Dv@;xi{ zY`;TvDoD6AYr@}5CZsMU%um)_ax2w(WGTGx*l@j(;&S&ye?HK5h6@fwzDOJ^=EnPh zv}%ueU&OAQ?qZK_(|7}t_*C)FAY~W=mrA`woZI8qWLJ~2S)$dK18}+8VViv&<+J-7 zmr;o|CGmYJc>88&slixU1O`fe5cNmUbZZjlWhyoBD`m*v-0kA_B|tE+f?w6~?m#Fs zb#WP;&sStdZ}%aRFBr31lY2=hO|02#KCtgu&HVm#C>!>-U%l?=!D^*64gNNDOdEY6 z_3AE3WJSten(^FlNqtL7%dQ%kLcbuY+&5((OS(N>4qq;w4H1(;F8%?m;YC!Um zORe(G&!0G6_Qb@)$FNo9|Ni60N5sehiLn)YYLgp>l1jIwf z<@t|HS@;cWs#L4Y|xN7e9bNw%>VU^6o$k6Vv~b>gK4!9aQvP}U zMP49@@l#mRQx1Q-RgH=y zi@^E$`Nm22zFP_PwA?LwZyZa!^_<(GBR*jagdqsAp}8zzu^DH%H2sRqu^udAy@%*r zT-;l^MQpioQ02)?e0cBezP#~gzIAw5*=GsY1Z<{YYzvb`D=Nx{AFEiWoNJ0XI}KcddhOvhQof~WjS zuj+C*mgHrK2x8c2#qssBL(QLyF!xCW+t0JeAN7W71u60?-j)r_Nuwk2iFhPDR#w4= z(b&|4u6N12jD;p>^|nM<1Xm*#qEv*%`E|R6Rxp&=f_{MP!Gi}#K=u-J7M-vm7b{&s zR-dpfY%83wk_%ynN|=l6J6| zX<-Q$Ii!^hQnbCk@Qi*ex$RibIyPCu%9}Lr-z~Yuj93)>#|=*yb?ihQ$$s@&Zt=U` zR>`WCWyNkjQBAsG`!YrmCo(Xm_||uCsh2U)_@Rv!BAD-EmJ~aA$-SRH7vm1;=qlp^ z$_LKdzMIOTzdoqyhtk3M@QA6?MkIomj4|?(V*+rgaU<^7!LiDBvFm1CaJQQel)+pk z@8=4#CjhXz({n>~Aq+Zd$+{s<3=EV(awtOIeU16(M8kHVb=pD0KHiW@av50d=_~`tbKEA!P6x@ z_z87vKpfqkR1L=s5ElzBul(DVtqHxG%KGLk*PQX=7_#w|b^gf2&xoVQ7z0ku@_FjQ zAisC+rg*xDJT>5vAh0n|kyZ5Zj*bONZ1#Fd zbVGA99_y_hGevg~4{5Z*jR2}9l74hKBNj>gOcU(!_uz%}~tG6j5WBB{Cd&yVvW%oevMofy`eqm_eYT<>y|5D6` zQ!S9c6elUXV89R(1BVO+?ZG>oB*74#r9`WhR0SE?^9Eftj6f6tPJhXK zqqUJ8Aygh+&WcjZQNPmA6%6TLQP@^<yM)oUmduB!&nOD_bk8;(oS6eg~^+p$z zvj}&vT{W1jaM>cZSdx&LHgXx%;2<9F#j|gp3@FED%vN_^xN}9%`AuQcnn`hUVn%>P z5Z#MyYZ^0=wt&*whePzePM~zCet!v^s+}UVBg|c)X~>cSRrd+|&)~ zK^7n2YfMWyNw2LocyHxp4M)g@Vp0n=tdi)lF%Q#e*_xk)z85h2g}zK%kr}UH9WwWM ziTnU3T~-z&Q~$oS@8?{WK74j_oCLq|069KluC^!ldI^$mEbkdAh*qJ+6ZVo_Eo4fy zMv4QBrHBv$Ts8(FRywv%!0Az8GHlW0vB|*}*G*GFhcED{h^(*DD$gy28{~JS;vGaf0!Wtt=3Vi>48t(u9)3{n#bUiyY!4?6 z7;k-Ebcaz4YtjdvG>nzK{;^M>sQ~(kAxi`%G;?z!26!J1af}21cu48Mdy5(yq8#Mn zLItwpjmx;QtPt7`*d>J_f^l+)+z|c&X^@SnyF$jnM-DyPi#CQNN7td3A zc#e554WywgXlQ68rlx*^O}|2`qX{d=Q2bapqo+ID_(g=Y^ zgY#C^hxuNb1ur^ROrK2i?hrNMRo5I$MiaY4S{>M8Y(bT!F;-U!pqXbA7~ z_L~-|T^X$syF>6v8^ zZ(L?s7Q<;fQuvd|mm3^ZoIb^7;1LtUyRU|VMgEp3gOlmIM=uH7nAnu^1N*ZX0Np-< zrh*E0EOv46I`mg7USuQjn1h#$$z^m%$ZthM!C6MtM6P+63fAG^394-)x~R8*axKjV)e zc3Nsfh<8FWfTcZvl4v^7^7p!K`kmJKMF2sgJTrS;#qY0%&kEIoReyv+vSUP8VX2;| zaTDjOl?8PL@?f?7x`U{)&7%eQoKLE`nbheR+bkQ=T~T}R8wV1dP<(15Ke69(N3yX>_3@z9N0IF zM_=7=48+Q}d8)Me+D?gB2pb(AX^kR9w}m-4I7}ZuMq61~NzBOLt}id!OiWB{{{%UI zVHohKc!h+50rK7mCsoxwGxk^c$`L4V@!_)hJgf(61;i=g2^>T9{vrm&a*{>F+sEhc zSVs4!Pk}W3K$yCXP(+)5C^_yopTD^@kOAx&CDVp}lp@&e-xq<>RTZI>nhTidP8~N# z`ui28+e#%I>I?=(1+KyJ=Mxf$U~>Lbr^HnLr9#0Nrinp00Zu&ht&pCPpPyghx}i+| z;AB_N&rbsRxME_4`=@Hen1ufM{LK#zt2?4zYWCt4Sv1sK-U(d|0Trff zvnpw!q4*c^Yf?3%w~CBB%dvx1S(LY8JAHe6{5!uDMGg*&uLi|k!7$Z%^Ai9U#Lft@ zqJ-W9pn#q~e}qZN$Z%;3tSQqW8%*rt>_DD57U@TzJ-l-Dst74ldj>B8HX!r<(^Tgx z@d*j2ajhYkCtlpZ^MK|NxAh=xhqz<|P)g{vChxJXNDr(PkK2Ark7*0 zW2s!7=SFZ^nE{L|9sQSd>JYB)_N1&8#ZWR!B%5%3O5H~%2k)M2-=4CuE^KV?x*@YX zCbuUlV&B!!AOesTCL0I5{0Mg)mY&fHc|3RThF4bd;^E=(2+KpK+6_FdKmbIj1_lN! zUL|tFK(UuR*?a|z;i$nwKLTcaL|l;`#Q-}|MIUHOQBj%{)a%w zQVdME)AjX&gaWn(Z};4e3Cpk#4NaywUL{U@{wnIrYh~JL4PEnXzmM)Ou`(@0avS2X z4B`*&PuwkuKZZMO>qQ{`_svZiWR3(HFw*)V2QD#@MdVpbpLvDF04ZYhNJB&8AzWQv z9<}iWtv!PQ-kjLBlaw}Ojb=-J0BoR7zyiA84AE)V-O}KCWc+)2$}1{ryMQPfk<2Ox zRmF8PGDCbXd;lT5VtVDXn^Cz4zTU{B2UUzATC1>+o~@qw#!AU4u?e<3ap1BfbC2G6 zZd@X8P~g>r0|l_x^r?rtyP2maIb7aH_ukWL7s^kH5KVue6pqSU9gUK5YaVG+IWdTA zP1d$SS-J%5sR4{?! zaA-N42ykBjYydhYHSfxA^j68*LeYB7v9$s39FCo@aHU1L&TBs!SFD?Xf!Dod(R*Ut z?`&6I>T!@FY=0pmG>WE!BOS%%=lAawGKs#o{h>O>6DVdzev8t&$rFu}#F*$AmAxAO z#7R}dKhXaL;dr#a{z|6{fO|4qy4zL}%EXwy9-yp8b)HJ-F{27qT#9^veT9RS%Y07D zY=DcULM^Q=m#+gS=__A}^Y@9WW=aMOC?UKsS;8$u3^zpw)MTJ;Dd61X;N&zzqE&!@ z58W_Qp}%}l`~CYj#x6%_Vcu><9l_KfLoa?#cy+HM|(d2hF+0ul8BWL|D zAbj0$PcuVWYoCV&V<NpIUMdKr*OwKhBiMA8vOQ< zl^6Iq$U#z4(!wjxS7&Xa@rveDFDVWINTgX+i$8%#_StT`emr}F1yAY&DHrV<*pJ(< zhF?mCinN-e$Vy5NXowk$)J&H?6#X>tb(^{W>vns8JeJLC1<4943@@fUR4m7rinn2h zBOnlnY|uTX%#cD4-cu@-gtT)78W21l{mwBJPirAj)_swgR2PQQ`?5yko~@zDfxrGNMCr} z;cbC`kmH)6$!2k3|APFyPsLY#SE>hy6`eOQ!$|npW7QvOKsvo1S7kdU3%`73nC!CG5v$L**@=aPsB)h*_L!0jDBMyXC)E`Ys4%8PR2~6ix%h!(uX8|VkiElw>}Py@Ns7{d*7c?M z(DNt8uSxQ74xlF5=Y<74zNWe)AwVtsahF2v<)8f50Jt&@Hz^cz(W@2j2^F2_XJ%p& z<>E0dkAzpO29cQF;EVnNu+;-OIo`W>ag@e*1Oc;J!-*yyuNqO#?H%ggLx_OD-WX5Y2WVOdWO=)HI49jk^GYP;#_x`#Qb z+_j)(VePkvcz-hAcE2&cs`e5`bW8WO+?PK}^o-d>Y<`cI#7OSnM~kkca(?^894Ab$ zmU!vcnRKP8GJ7vQnYKjh=VX0y625_l@Y{o5Fmlh@2}K~s8D1NhWo~U?J+-t%?nMXycfrzfaH3#NfZj4T!_EFEC{0DvAFI2;y@tcMDGL*uGsH zaU4y6e6m?QP9r^vk-bsCf29UyyxOgE;`<+=KS#P`uxcBTvTZ9d6#@4nvTz(pjIob; zSgv~i*v+qu{cV;FHIuL*WIU!*LBH4hF;?RD3QFD|t1YWl3N1#)J3*)GN#0d!#P?f=0c#YW*VHfgoLxxe~UHPry?20W1*ytG*S` z&%ZqdX%7+frh)iq;495j(WMwL3m4_u9(bt*6 zae@d|Yu(D=2WFG^059XOD@%X_7E|hHW|p;DYuQ|%wa1FZXrr$^#Rsp&NB&KjK>o}h z&!$Oj^^A+I=??&^o(QzDsWKpgSyP3v!QFJUnXSccZdt8|%uFz_!BMo>fu8_EKaukT zmqZhzh2c#?YSdLoFGf-di+e>;G~|+3CdYN}UUzmYKTqb*YzkVd5sG zQ~JI2u)!%uI?(NRy}|n7-@mDDvF%qvCT~X2G4kC7gYQL~M`}hn75_2?uC0SQrN5+I zcqG!2gLYE4&VeRAKAwk{_vPN+-dkn;U^CdD5W!^Xn^GPp!Q$pCF4d!CQ{Pu&!{8Hkp1+;{y7HAD;!Tktra_hQ3Vn{L8VS;3e0J-J(4MhOtV4 zJhO6wR-u$6!TcNtJ>v*KAh+5s{vpeO%8Ck8c-+Mf;Cr^W=lnVZC0z zwQ&a2#E7G6EXzBJHo;4K>LtA8kmJigsnBBtoicJgh>20B6uT5xLaYRO((22(DzVU; ziTE!y|7yN7)z2Wa)g)05Ctw(Ssk-!{eNC$3=S-l`njLo{Q(U(CfR)3nr+Fy!pN)Sq zzHPLoX5W(b(ee~wdmHJ=r&i(eQc>cG@f970F`j)+?=!7mC1es3bvQ{)3S!%?dvkMh zm#M+POn&U(Fns-~zEL+va=sbxP7pV1fZCBy2b~ujhPlvPIO__zKCYJXt@I8iA*e6H~9%`j*;qWuDE6m?X^VX1CrgP77Z`Xx}nvGIPn#lH;|M4YPA<-PFO=Z+^9}H zrDj66B=^)p?_z_9I-{IP?9Rs#oYKy_38bK*UHg4>lns9$;%v0FRYEh&aWCrakR>8a zv74#A_QUPl^{3bI@OBMp0WS$XTJZBj3|TRE>ZpY5+8^FdBkuc)SS2RRdO+BLY}GbB z!U=7~q)B#jImQ3+6d{wV;Y$<0p37wG4j$scLwbEUM6j`ea5#&dnYlT8osO-aKf|5a zUnM9PFy#3U9}Sl>b^NO+TFx>xrI1p(sZYYULQ9;DA9PqeqUC&l?(|U<)lJ+By(1Dr zbuHcOs>goIx3_2DJF^Tqm3+pVkuUw^uwc0NM~d`Z^z6x-Z?zG%2YGM1LD! zA)LY!ZB~IFbw6`ws?Ie(fWJ@nCtcomBali>Fko|=~n5}AzV~R$>;A`49&zyLwJU`Ej~VE z-L#eOF)DS+S2fiVVF*vdDi$)>@3Hk!!Y3e!m`tzyeT~t0M($VM?w>4A4pKU!)i%c< z{#W9QeOKr89cjZUmDaC+IoIjMd+bJ8CG{lDQJ7Ok0h`!;%f>?*3-9`Mge8es5kORL z5L8Y0)bVbI`XDvwenSN|_LN z1GCJ1lFUH`5FY67L^JwMI#JR(4~6};o|6Q7Dl55kz7=+)2xE3W@rm(FPsrO~O@);i z{S=o6?2oQWIZur?8up*aZ|J=1OgemRyRjb{u+*HW{YN;A+A3(!@M^erhF{*UJh?v# zVvf}_D}ucC4&%{J%~uAg=Mw#;^tR(njJunc=`D?c8ysB+ZWD;7BE^{ zSuVG{4Z!ALTh(Qi{E@e8nmD7d3h1nMkt9X!$%$7Jp=Dkr^nOTxwX|dgpw)-*T*

z!5Ep~MDCTzh9-?MTQ$4kvQ?C3TtrT&wOI2ree?fg>%GIV{`>!737Oe5yJYWVCW*}G zls%Ib%HAtvm(f5cqJ+e0WRnpxvbRL`${tx+zvtWMd;gC6{{8ObxQ^@k>*{cx=llJ7 zJ)e(R^Sm)W+g#pI|EYm_2UhzlGF%Rtncf@tcOiu|*w;BqqxO255#?4}eVg4>$AjT*c1jT5+7y(Vb|WYIUttP_UZy?20}DgT9%{93!4xzK${|1G_;32V8A)o4)t;&-=NLV z#omAK7shVk>9~t5T7_5UDQI+UI8yPqCnhAUc+CFa%U&6Yxs*X&Zu&*e0XK%%9jJKt z{u+h+YO}U>$TI#i_o)9=+D}>e7)Y?K$xr!1S8dD1lLSLiEazlvoo-05d=tP~pImD- zpvk{o@1)8Dz$ic|bw>0_y>mpzDTm#2L)AMX+*Idqx6d=CT6XZ-`)Hw_vlbZmojRE7=X#-df zA33HuQB9D^PjU!)6UgpQ6X(KAv*P_;0 z87o9gptxDqt(-lV)I@$@8Lje1COYbsYo@NW%O@dC0fJ?#)=1$fL*sD9?s*FG(!e__ z9W(!$?@UD%Qe<*rgiY08`Tb@Te>c61LLK}KuDE8;ezMzwNPqItEYSyB`o2Ih2KrXM zc-7280OcM6DcrQRRm9XaR&1kLkKoR!sm*Z8!FrDzFfdQU9vMP`O@Of{hQkdS*A(?Y>RE)JYd}x2N2)GszuF%Q&a)d*FK7Obr%_}|@MDA2PHSY&0=NN^jJ~eu zQ_DYq7zhN`gTLvg^~XCfe8rK$$eQ}o)djo9?R)?aAC+F~-}1Y!xGe}%+(onvoy}Uv zT*HKjWaSdAC2o(1M5dFl6QcNagxUi*gEbb+FVMmpL>F(5?P8{ZdUcPT&NY3}F)?et zHC)uH-aj&!b=7&!(-PNH&QT9%`Vl8-$^yZ!N-J-Eke?Pp#xe#Rp)e5qN6aaxEC>Us z0sA5R#URUtOX@{*wE3e)t$K0E5fRzwcTfz!xU|$DJ-&IdqqCDb<-E&@k*dM%@^0*?L;b8KM?x)CSe>$xT4^r1%mMxXmV4QFQcL#VT zE`9^35qU#}h0V(~N+A{L|7T?_{p56&*8;X;QQyW(AN2S~Pc^pa&;wgu%Nd@Ce(g^1}A-N1Bi4tkE0>>Vz9wg+sp`gH*}{)eIa zn&<7zMhgyHL=y>0Di|038?KrrCt05*C|tx2d{obESlbq4yz%KBOGu=X!eYUOAmgMZ_Ha|A{#$$+G?An(hZs_yWrRj;^j> zQytUW(r`T@K`PMq^HSr}s?$8sQjO3rJ+K9Z9zdu(ClaVQt0O|&(7Oa`*TxnI z_#@=3v3@sG4mDZ8cg@rmDik^>b2s7$E6*bQ55|r17w@3dpKolMbT{LzQ7$Q_`HIZcTVzALyk_xr5^oF`y8enjlAIh>xi`p_`Dr*x zx>2C~k1tBhe*cgSx7p61OUSwqc+K^7B_Q9=cuh!xqokxf<2BPN0Kx_9$wg2EFv2;8 z9PS8BwX>u1q*E2X#}1&-x?~R@r@(Qg4-O|p#(D&;LJi=$ivbV(LmwXoPy+#K3bg3r z>S_}JD9;e~9jO&I-O152x3tXY)x4~IHp+IDR8dQ9UurFpjnh9@0;(*0HpEV@z>acC z^U;rWzxo#fLRamLBF7|hzrbq?lE^Wi<-7_gmvI_>QVuBd(8&q>)=1aKPcu&s570b= zypsL*nuq0T_1?TZh+uJ-`8CFiuXqrMoXKvEe}2ZYbh~hfs)mnA29Atgk2YS3t(32*RR6 zAMuCW-1dt$9-f}!kCp*`Nbl|K4MIeoobTMUK8kbbo*i459?LU9We?~ro+;O0c>f_q zGDmf%d#A{d0Frvh`2xRD#iVR$AvP|OAY}?G#6#$F16UJydI{*hTQ+4ziFL`%!&Y%o zcK-o%47cypB0hD2ib-tgkde{R_S`F%Esyhg+@623?)gT69vThL$Nh2*j=Vn~AKJ^t z5zs;WsMAw<@1j{29r?E!rm7WpUdES%%iN}+iuX!7nm-Nv!h8sC3I{Z(5&-GCe)h$2*_}N|L*X?|O^SzQ z@C4!&tsu{v*}uL-KfxA4+G;2tYeo>A%kipgpsK}3^)#on3UYHW`)%Ah0_Q$&#c%!> zl0exG*Q~iqfE>gbKz8Af5mq>WD<=&S&rX=b7CI{9EALjO$CBVy z%N!J1)qO%HR(JYyZ(CHZ5mf%GdFPnwx0iXD+z#$e1zur~O17@@dM2t(6+Nx`*0dmt zN0yE^pxXP}{AyfY-$%km^qf&rqQEEe@Hnaqw?bLYgP7}l&M$-5d2i_q9t=cNSl-b@ zQDGo231IsA|6}ET&T|uy0s*g#XaOJ`X7oM()hJ~r#f+Iua5$5Po;AI8NO2=MOw)q} z0b+<5E2H0*E-dI{j1-~Vkt}{zzb!%%M2{2O-HMg{$CY2RYj^}hS`*B&Cc6RP)I@Sb zkPxS7B-(+3pq5(P5Ohcx2CvDg1U}@Xy!!Nyt0Cn&DQKsEn62E0P4^<#5{?V zd?C@Fa8JTUj1WVc>9jT3Pk3TwL9%^zup-|PqLv#ubO-Xp+k52X-=&FtJh&|DxURok z8tCu0IJr%@C^ag5|M-oujriRHpW&RXAocZ61ll&jOpQ;K-z43a4WSt*EM81OrVCk=^aLh8#9i*90#nUd0laeyACR~?^9 z!?x(ur@}chVPs&iS=vdJ3UY+?#u6+ejr%hS{=teU5Er%%xf@!2md3 z-o2PkjFk_a-iH{4Ty=iU0xe6-U%vxKu~!FLbIU;y+1k~6(VOE^c8c+%!Fp2J&H)9# zT6e^Y;=f(Yhr`3Krc^R5Epgj^}o8opn|1jyT}d__TGN9ByY1Fo&3**ox)t zr$li1$;e-3-$L}aX~!!zn>FkHq<2VkT9HZ}`}|70VPjVkB;eXO;9w5)aFGwxmOW38 z_h$UpeOaLA>qFAoDMXP@pB~qph9eECh`?|2Wvt)#y$cs zw3|QXMF> z5F;jTwfl*&_gs9%MgtMZyOsi00KpTeT4YBoNYG6$221MgpAC=x?(JECOpe1`T}Q_m zu<#k=#~%kTDw0FU5xcs$$hDsYJjY!Wl`_^HvLGot=_qVD9BsD#uCE8i2|BGkLht_> zQJ_-AWS2FF!!HpP3t(}3ZsMSBe}5>^{1Af1v-!%e_kf(ljZn#m%(k$N)&k8|@yqaK zaTrHvJ+uF_0`;dL@2ehZM7AVP7MTJqod*D4*NTaKkA2qbQ)^{(r9_H8#BRaoK~=|4SCnV%vPPYF zTt*qPCFq>z_Nu(!eJ<<+cBJKe!go*HoV3CE5J+f(S`#8}E6nLNY32{)>v$m1=XoC+s*K|)Llcam(^%~L-=#H$ZwAkK?E=G_aBc#w<~gd-;* zZqYkPX;8}0yM=^gA*9?=fU$F~2eeFy>*##2dG^^!c~uF8gLUpkV8Y%(b#mPot`Mw@ zkZbxorFkKKC(X=h?l-tz@N=i@e{!rO{S+~9Enkuua!s_Kv>Xu6-5W?}bRAS(Y!5ya z$+%Nbm|c`f{88Tea$}3%J276~IJ9aeLp6tjf@IV1&ypy_zYy_saNe5KYHLfLJzeVY zSOs~asgLcCRhaWCR?9|n@Ni?HQg{qvZ(xw%KNts+A{w&(IkWHPjY2@7Aq5p~Ao3a^ z2Zn0H+a}q6y|HkK3*!s7YxHzQlICLF68wTgw zvu%uVg|c(=gIVlHjn417HV{v@=yq`S!SBW-DDWXWRC0bx>GIqV8{^Y4{UGsL0WOJy z-i&>zO8c84fin(YkG;>wP96OD;Qhr)65e=}-)09jIf@C99uKRACmOpDAX!pL3CKD$ z39XvYY6oQ09Ix4CO|&V@iIZ@T-}$i$4Cy|klo@YRC#=KSzFAgYlQ+8y3mO%YiR$NQ zBMNdOTuL@f9KEAQy9=JkkMp@!l_)*s&UeFL9iIQ^{yWfI0vr8=ChSw8POf!7tNd(! zZOn)Vsy)Z8;gx|h$opejL0nQN(HmR#(4NNl8@xvl}?QQg}8@~o@P98^k9S?;G; z*l4YW$NJ237YJ$-U}r*Xn)`O~3Ate0(Y{&6$H|`43kwtyNftjAfV2dH=^JBjfK(4_ zKqo6w-ziXW3}J8eh#v)(p`XKLL(2$A3;#D&htshI*#*PS;#c56UET*Gz_jn*5gZR! zyCfGF%|=Ha!;-s_24#$!36Rx8uO50kiojJ?nnIUVG(xVRRmUXPDqW*(?0XGgo_#vb zz`v9vwZ+@xIfVM%6)q#$-;^>RzRTcSvd^Kpa}o({0Tb*AcCE=&;9+I|h8VDn{~~HQU;M~W#1V#ds<_ky+)a~k3*Gt65N>fZ{U+EHX6$bOu-7K z+}l7Z2~_3*ZGcQNOQu7}{tXgZ+EWM^F>yLX_H{%y2*^^+!s_GeV0M7)8!U)8S%Z|K z12Go{HB?7R61l?razjrW^i6|%wk5xqQs`SVu;xTfnGF-dTj^zsuTn%bXiK(Q1PBWF znjxz5t)CPr+4rs$MEVu>jgqTmD&bOgW$dEzZg_P;xHQM*5WFaRO0>DK;@Mh$42*}F1t+(kDfW2Ty)%>yQ z=+mFvGcmDli(dB`{%Y*#c}QQz1ECo4pOI%aGQ>MN3Wy5t_w#M2rCTL@o28m)$kTOkR3-`G3b?19Jr--FUu1L8_eXy$XVZt5Y2GS2<2?8VEN=y-GbE?x%|jt zPMkg6Wx8rXfNIuMX5aIs9fxy(qJElBe6Yv-U<10eR4jITJBvMN{oB@*k|K)7wBk)# zc(1wgN7ipEy0-m?G&&a}DRdroc7q|tb#gfcZuc5wD$#xh;b>zpgK(6m!KDg1^X(FK zMbrj7&$Y9=@YyK<$J!}T7j*c#dVP2ls-fJv+hzV_yzUts)Y+~wq&|=CY^;vbXKhcI zuT&!+zjxOCdX{OTXjo8#`Eu=kqfsND8d}R3H>$364>L6V9L{B6K$)xZpI9R`nLSzGwWBxbU2J3=?sWP zKA-;+BIPw}zX2b?*b&K?Etngc@vJ+1x4asVYQz)Y0A*8P0*B5*lDqvfg5@J1-THZI z4SseQ`vY;@q6cx3bzIF*HNDL1@n?9+bI##yL_{0mnnWa<`qt$!*0&R7)A7vrpUe)MioO2z+_hst z&`pO(z|H+K2k43e@~;Mdm?Ms)7*vy&|0ct78!@l5ZWo03W%#)MasRc_x5Yq5JX zA5t5Vf9W71)<{g9?L8N^9U21t(!V*lYfw=gX6TGH$4?*WUn+j`@hXMei_9R4-pyj?$S0jN z>t;}ke0EnZ*2*mw!fAT#`m|wa4nE2%Tkj$XsM_(6z^vs}i2150n9Db-V}^_hU=1dB zZ=N1>*CrV8nhbA3dO`fy+Mf5hZ;-Y;?{k=TdSrv70oEo?5T{VGL2qH4)>fm3ItNJk zu%r;RS58vFvJa1?DlZ?m2?J%6-}qg8*H`)D{^9YOAhE{Q;TAUD6($dnfSl2b-dmlO zb^PM*c~)M}IAJ)WZnp{lgi5QN+jGa)_m-$q_L*gjjIJ)_b8EGKVJzkS7_&RsRY}L4 zdE64c?K?DM_paq-AI2%03zK$xM-)j|pb{4epxwTZeujM@!*4nR-1A>%%U~m3ft0A) zgxnk+01U%np2=DI*H8$<@GBv%2Y8CdJmti7-K!87(gezl4G3jnMr1NTb3w|h(3w^I z7P3=AU=cAwa3MsDeyLr2!Hzda$btlEUubwO`c3E5!7OJY?Lay0S<4&$-o@g8U_Ir*F!)a%MuM)v1!jLD2S* zG@`XTjC?mhs_!|VEzm0qq+&W7y001p$+5-=*faQhSKh<%!LFW>lh?UDhSa%(P;EzVKpBK~%*?D)f# za?&3~1yt|HRC2doJ}hsE=;`+7iP1-H$N=ccgD9Hz5sLz=#1LRZ=2#XZZMMIL!g`y+Xu~b^J4p-jt7cR@h z&mbt`-%>6Da^NF(1;ER+BUM3t@LQMxg%Y<@nvzxVd<{oU;I=`W{~N6aA>zfbbB#`< z7W22?o<&J9Qq%f5yf#@0f;u#1_PsRyi@8|2o)kJ)wxmZy4tHH9h@SYa+*BNuQ&7Nd z)BUXUZ1PP7eUbIOt5YAvT zU&JH69scIdNqnN3qG*PqseX_d(j^9J85ZOe%>f?`#M^OI*3{MAjGId|>Vmlj+aQ2r zZixK5D@bqiDbJx~1jUi%9h$fv!2o#qB}Nk;wQ)q>`l4K$2bW;VbSf^JP)Cfj)BQ4S z!$5nuP*4L|R(iy%Z7l2!e~k`I33t1$E$h~I(2yQh6t$G^#1(c6J9pgAz$9r!XDb`9 z z-8NPICUk|^B;1_NBed~-*=Gm8NHH;j^Vo7WHZ~u{L)#|D_)0Ia8SRmhqiTK@K~7$j ziw-rQ_weCBfo-x(qD9r6ClRr@;7!1W@X*jJ7V`~>`#CuY00a|&Y$8zH?9Z^AMgX85 z?|+2U(^G0loI~OyOacD&8ILgpy}5v(V0f+URt)&cc>i!=9?;DdKd`pXS)wL|@Mc-g zW$oN&u9L3=tb1Nh^JBKTt$oD?B=gzrwjbLlO3Hplh0N4V<$shtlbGdNIWm~K&NO6F zi|2Dhy<);4^5%EH*)S3iY%5IoMv|F&Vy2A{l_1cv682D|O&Oj0!(blUX?r`jWUD1k zt4QnWpyK@1t6$G_RLG^G9fBNDJR;~9%d4NYdE(|cU#kp{e$3AP8L^W zTZNj7p|VFQ;#xVp8Y2uam{R)q=t%!uhmYx=$2s}5(QK~DP;mX6qe%CQkG(^Q!o3-J zkKdT(@AE5fUpIl<-s?AKnEM-404YoVc>#7(zo%t@Nnu-t#!VG7Smn+9s?J7X*#$6v zUQrCu`qp}v&%jj$)p2>7dT6C%lWO3Ak&vaE^z4P*{Q}22dva&hc&akaYg<9uR=)Ca zR!@nU9~T9*dE!%X1kLFB+Vx|s{4sSe_?y~IQr8#cg$7`q4P!m z&B5FCl;@&z{|S5(GfK+MF%&Yhd>W>+)&A=2xX?A)zPI49&tPv2g$jl8NV*A%7&)c2 zL(?nPdDJv&eY71U=Y}q2+>t5wNgcLtpn0Jb-I3fcnYi|ZAHK-lz>PAu_d6{ zo{tQ7!0w)h%ry*f`u_QT`wDF%m z#`k`H)syb%IF?YXR}1YhWS0q$f4hbZ4laeBQ73z)R*mSs+`uP@z#hxq2g9t??PCw;qR&7U#Sh$ zl6tqOL>9ON#FyHBn;sEv7?Bhc){>Huhvq-f6d-gIY_1V$QWHp_v0MCdf?D@ekl=4W zPHco;&oD@0Y^Ms`y{+puohod;r6bkV?%C5yGa^$ zu&LXywvR)b>GCQ_-{4t%1vKKe-1`Z8`c;aL-LoTEQ6$)ctgivrV~tW)hJ`bUb! z97EN$Px;tbk(ey4+iKf}jc8#X=_~6gbrXn!*nt2JiY61Ocg>KGLwJIWslCaATdHcc zaDGJA?EXS-H{9~O>gqT}kIUykoqWedXU5cbb98j)--ubMMI zhMc_Z)^fgu7nZqkP91seV^=aU##XoSw|VA#3bbkL1Qh5MAf933B!T1oEH;`r`2PD4 ztRsc6%XBd?jN&eQ4GC)CH91awL!x0^_ADY|%=zZsXxh$~uE+M9S~{<@X@)Ry?Alnm zgDSTEqAYDC;i*nCWgF{qWi5#C{Jz4@#uwMFwj2lTHPJ$3Jl97TG%1Vr@SZ)6QKX0) zE%gbw(Vo~qYNq|)t0hdxpp#P%^?j-i$&mdxJ)OR8xkw23_zdz8ZsNMGNo zd0f7rWBbnh5DcQdp0@TeYqU+WAp z)IPcIGS+OI8P=;1P!WAFW*p^If?KX;Q8m!&rrS!4vGP7{CzprVyKKATc+t;d5=P^q zaMVK($o~xFlg~&6D(|XBLQ~_fVVAv?k^5TSlvs$)!mK?{$IL9v{@&*2PpMlqD=s&a zvr~WCODZqtd4l>?%3oEjEcSh>HlZB}Y?&b_fe->Ya(GZ#To(b90@HGg z>IE@b>@L=}!Kb-yd~rUfJu7vT`Fn?*f2 zxKX_GC$wdZ>fK^9V~bc*5Op4(S#%$H2OnjZTi3P9vTFr4zQn(cbGskd)J&<{k@U6d zwW?qFNhX@Vp9BRAc4q&r8`&RYTxyOX{bj3k4EEl_{GHJHCniWgMbQYRV6J(pQnIt7 zqhg%d+TP11RGoEfgY%$&WQ2f#Rh(4yZgW*%UEfiHTtKiPiWU8bD*=RYSCR2ragvm9k!4xUM`&aS=8j=hxvV{JuVLPof*ZAf?)FDris7=wftZzEeX+Pd+H_rf`!bI? zg@|U}mWGL1Ik2mQZ2W6AdyCPn5i)?GnwYLKVLnC8?&8osN~g5mNPcI-h#Wf2Ba$uv zE36=P=Kq2d58`S-By(VxKRN>==SGxXbpzlO2xawkd~Kof$w*OC>FNQlOj}!XzzQUndmqzap+P2%WJO1W3i0ZNQ8&0186{jO zLH}~!vJ3rmMxnmN`0{5IYTq%j4c(+P86S-HU=^K>U!sll2NR zQhm8%wS27|qOO6hbMC%EUayz=MD+v;Y_)l4eBTh&DxXqQL^gRe1~z zJA2g~P_0tGvnyUbMUtBMXBg?*@;Ouktw_(%4?+Oc{+S5J(cjd?@9&9@qO6yH&*eq` zJ2E*WiFD$T?1h&UJ=UBQl_JS0T<=*8(wHqIF_>XrfqqlCUG?{~3={&c@@9AkDZK&g7y0V-+FRAe-HIxCkMmp%(`;y9*M-6z_Sosd#TO zkzjSt&Mrd8xeG3Tx_Qifmis*lT~mG;{eP+w7ZPF3htOPO&Pda}s$p{{C!rE|+M8y0 zA~WrF1{blUh*ZSF+|<_?@Ls@0j}-;uS*UB#*IU|325NV6ZRD#;I4WJ&Ml}=q#;o+`#w|Bc~&G#hRCB$gRDm_lw#ov2ZTa@|RO)7P?z8jHbXqeo^;uo|F! zWhuy9NvQa-M1OT}QUMJa$>^a`GFJ##qjy3@NtnW}d6`>2(^tfW^u=f8AxrhD@cb&Q zK08CGAeA2!bXBvvz%m+pO@%_2{Xd8>pn&_IqudM_ISw6a-byHDVGZFr*aj&>_PiC= zH?X5U(*r7sqkIC9#6fAHooN@qRk8;^G!w>BAu}{m(=UB~?(wgZ&+~`6?3v#$$Znhi zF1vo`kOgGv=M6L^;rx;TR@?QI!X0m^wIG-*i_K*F8TLV>Im}^in!|O{mr*EAwOewUMyUcNNQ{`uJ6%)Sve&P z1nCkKMee6$u~3`ZlbwPYz~};gtop8&(jx8z z+%M=%J}VYGB!`D_kTwE49cx!2-##$xbg6@- z@jJ$alN{ALP5prl(nQ%;t31dV3x-QCGWSzrvm}JxQUW#i@(eU{WHe-c=I@?(!X9Rr>r zm2hW#%O0ZrsWM7a<320WLu&Dmf#wJ(r5i{FuB8Y%ClF>V6_ z*onjOy`^7Q1^+9vF;mSy7Q=cZ#Yx=w;BLndO#{_Edt&qEATn&-DY?!yZ!=%~Vkd)x zr-$fjUCMdO06NM^U9tJB{kf8lLQnZ+DH}hmnQZ?a_-rz8ZK!7T`Lcn(vF0_YC4O-y zULaOvr&fy_eU&oi6C4v6IS=&YtIM9VQ~c^`BX9ZJTQv66*Fj05|A2u{*^DOta_fVv z;W>OP1U0YBL93GRchYwe$5XdLmMY0v)H}ONW4kqMg@2(SlkqG}b?38Yyx%%vo(4XCNTcZSLNno<(Fvy+Cq#CnM2jexOxrMBGlG z`a}X&&ZLrdgFK){ff!08gb%JiBtvBbQmb{1O+gzSaL{h{5P}ePtM8UVk^!z+v6XkV z!K#TH049C%8P_dP<;;7r`U7jrvzd3G}=yu1J4D-AAKO8u!Hmj_>xv+<+ ztg}u%eNK5<{0_8W!92s2>hiCXpxL1sJiQYwZ0C~AqbgaZ>S3d6C%*Ih`n*WaJAjK*lT|(9C3(& z5D2O7M5+gn`U&jOELDhOR#FN#p@04VhL~Wr(1%j!HBR`N$TD1UoRbn&gXcXpwA|c0^5%I(Zx6 z&{VL0B@pJ_o(nhkZ$Y!b(|SmM8LEwellDkK3Cd4o5qPoRx!8ye)sgG%j(eN6GNz=BAaw*Q@T19J0cK+q-t zP8C1?ZX%me&_c{~f`3~fw9&&{GnE04nA@cIDfR(@fg*~Ps`tCiep{LtS(JA%^Y{H9 z{No#6xOBvp$*#M|f2B6;6)zGJvEnEXx)N^G2Xq+L}d0uiwk2;fQ_; z1SLTjxGB1Ev1H(X5F{FTwPs+?Z9vrd5I}&0r@<`)DWSl0H-akzl9$fhKk(y8o@Ky8 zL)NZw{WW~~h+snWnPw`7{dpj^J3rq*j46tG?_8`*NgYKngA!SFJ?C|DM{<89>a4EHuQ;04d=GGX8s@nDd&B zl-T@Ed8z0GQB57Lma}w8En~tk6M{CZZ{URC2TeMNB7W;rW)22xI1`%xJLDfa#i45Q zdy#+WIFsUNu;x#HkT4f$uIzira94kjx_;8QKlhV-R?U6(ma@9m}JO>n;eT7=$4r9FPsNr+Xpkh7nv>ny5v9O9IW-IfBJ=+7gd(b+L+l z@QZbRX&sIRG&vYBcMQqBy`@?(5aOi%sAXGQx7N;@QG@7SCuZ2EX6AlE0R3V%)d86Z zib`Rn#rSCV5Jul7|C@}X#K3b+h91F@2XCC!~%+)-@hu!GJ~xAJ5M*cEpK~+DD8N?1V4D z^=*{UV#Gc?W3WdaZ13Sgtp)Ib4-!Jrnp-lAc!h8H}DqS-E^=H{}XmtsV&-4_v z^SigLFNg;HEYbh8g?VXJs++6!w$Ai0bf!*~U+luUFhje>G{7N?BlkY>*>S*c;(Y(! zeC%|-ZNS}1=Q&pa-qR`q5WxrHZ`dpv>)J{b8!vw8P27}PSflI0yc44_fHt&}WiJlE zQ#?fDOta>9te?7Gt<`A(lF*Dr-7B#Na-@Xiwzyd+KIaOyK9h4$U4ZL+cAF zKX92#v3M+z1lTFBn7b4r*Pn_J+cc+&l{^&B?-+hMRuAM~ybORhGRQCV< zaSHe-AvTa%GgOFaM;ox+$Dzo6e#2HnA(=99ZvARtI;%ys!H3@INW)0KInC_@{mQz@ z)Hl?v#8-Qk>*CApSmx9mTmT zy1PVw@8GAwJ+>hm6kst*Mz~?{7DgbCZMqO}e&7{Cf}R=dw{qfnE$2A}+b3EG8w~23 ztUq#L$2AbAFk}i-E!#W@ze^p&nw3c1bmVPY+oD;U_GaoMwz<5sw%(%#xvbZsGRvG9 zs^`9V5Z}2m)23k=)|@&pPiA|`+19|ciX?sZ@b12cs=`3}^hOuzk3SnTL{HqFKe=T6 z=*MXGygi$);p~H6Re?s18hjh!WMn@5%iIv#_f6HBo_B=fkq-?8MN&dj z>H(87hg6(~yF5ypbo7%ti6?&XBP;AeGuJMsEPvmV6K5~0dBH1sOuwIw31W#Y z;N+z_FPC0R>#1L5JGawXRV{&DD+f332y~G zD15Un#8T^honYiiEfq!NFAe6OxaX6r<+Coq*K(!c=3d-WO|?d-GYx@#b4Kg_L3}YFE6NV|a8LHFt?jO+*?><&4wDVjhlNvOS};)EFy-g|BQYH!P|zQl#J2arwiQm#QAD@k`foMTUy*?$%c&#yyI@ zrOq*2^>unI&FP5ad5*Pe3u)&;j)0ftZTwkY|Dmxy#yrZdHTGN;@jas`k{{8My zq1P(7H8)iL?q2=CgZ*4hYJOfs|HopFTO=fyz56P%xTe)Rk*1)jS2P_`#nk$EGgsU| zO!T3tVZx{)PUGpZQ&BRzo_PmaN64i1rDjK>?ECle4|%KgD9M|Y?kN#~LyC#<35qCI z^ZBVjf!L*7EygDCeYy#LdzB~K7G0f*I%v|@$fHAmUZHTi-CkaY7pI(oJT*F$Lc~nG zXb#I#Vq#m?T$?ztB7I}DlKfq^cPnx{BPeHo59N{}NvXO)A$#^Vz0Mb|68*x%BLY+E z_DBDt``7&cp0aIODU-GrKpv<+pdF;HR9KirmUZLvuZ|m?5$qaYd6Ev{*}PUB`Yh&!pjsZ9BZ&0 zYnAC3Wh(P57F9zry@R>*zB2%;sLeMp%c}5|fx;al;q~O)u+LmPQm@`|OsOiOdPjPn zX<${QHo?1@ln``|HJAT(zg}U5Ot>X`sCE(M4~3lBitzxJ?A`Y4PYhYvb<1k#Ov#fR znyzc>txnb@m>3C&)&z{X)GpzRB3usjNcgHUn_m!VFBmpL3?MvHEMZ09ylJak#bNegX zs=20TE8dSUJ;|EG&|C_>e)6)-IH`v|;#J?bwl)gg5)L|nK_O|aclQRCKB08PBJ;bh zuyPa`;-ObWu9=*F@U3;J^-(*S{4Ji|Ah#k)y9UA@o$NSz6|oLO9U`0bY%#gl;i632 zPo2j}4Z5L?rWn39h1s5*VZ3t47)$RtQ*KeQ120m>7hCQoMZ&EpdyiIhjz>}oEeWqO z-C#l}M;qro%|Ug7m&xBeHN0}8Dw{@L1bzO)(a~w&bIrE=%7wrfH*9!Y-Q2QM3nJE5 zl;0O$KGzuc_Fek}bK&RecjWuU`1qETuXN;~&mVKU_GdhXF(tmx&Ke&>N$=hvk1oK( z&8c~Y)m&wl$gRhzwI{&XS@*NxlaY(5Hzmxv;|cS+GILPl9dT4}e%ff-X_d?$9UQ2* z{e=Mk(J#UTQMJ}zgF+j{XC0!l}i#>>L@_c%F7;JJmbwCjmNa@)&g;{<|dlsA+^umDudg{DKy&SFwPC zXQMLSc&&VCL0WT>xOXBVBEYb}L$HGrEG&o)sPOBB>*pJV3L%pT(XkL?Lo5|~?%ux-fIfeb1r;}xuj=>qcwb(%1T=0PoP z>SGfXxL2{?`Q4`q3Oc!l_pe{yv{y(vxoInk9=p@B`0dEMPRDX{oU(CiYdr<5UqyD0 z9$=h}yDOs7y)xNXUSCz1;u?M>gIS92A{=&@>i9L(t_L*-rlmxxOHWkRo?K#7yY%2(<7-q z0ym6su60+ZjZobK-Oe|al8%jq&6Jw+0DXG~PTdg12+ywC(9U;DzmiHB@BnXqXJ8-} zD=tq88Ht^`?ju{%dt&`3ntl_ekr59+aM5WAL~y4+e>S4nrTQyLTCitkQwCRXPeu6q zT+6EKv-?D~?+rJ&53Cn(GLOsI+_P1&|7c%$MGy<$JZng>TNTkCH(KJ`N>__RFbP3# z=C5RA*?h4vzd)f%_(Or=hl1422~I2s*J(bWHKyv~%TD^^=jv+AX9Pc%(9JIHN_)1B zIEP9N(XNAE5+T4Bl7ebOLBSIJ?p+gTjST$YYkr2heJd2{w@$?6m9j;xS)go^#?=KxBwEP9NC zb{#ARvVmY6Damg^x?J~%%4CQkXc@#H!nnfbmD_1JM*_T|czjvoeGZ_RpyWWR)T8YKvo zUVe6f%Y=Rr@+$d~<|l1pQqT7V;^#P~FF#QvUtCwSDEPws;)UDXu;BKfED5hZ!)jKO zF5}gn<)q`5mG{X`z?E?GViQiisk|tJ>1HbqIb8Yme_3Gy>Rz6K+l;($tNQ! zzsfJXraxZE<{nslzw?9qD4;F_gTQrb9W&)aQ+|3=gYYT51g$o7C)E%QXxZhw_BcE zH9yl^^f;I)%^-zkFt3|e@kW=Lqb&%is6NQ)=x{}8Ip-^@5gkSV!+{HX;v@0-K9Q1ZYam zT4vwYw{j`lKdexd=*sE)$v}C%bB{6OjlO=zLyzmvLphkL9Ad~Dh0e#lW8(3tqVx)o z4<}1Pqh1HTOn%^?FsZ2;X(`Y8fl%&$QT5*8T>kC>c1D>u{O@@xXvAYJSm^7lIbAza(y<#USKf^cNpE6Uajad|5PQ+*J=f1SON|r)k zx=*01lOP#9JLeGc&dt@zY(QUnwW)YD03;dLeTaFIO&8=^<6iWq;hN}*X*~2=F6GAa zy`@HLNh8usBB9P-AUqT(JsXwiAW^@F@3)d?{>o+4NzT_)UO453^lV&YPJ(XxcQXT5 zS1Y}XX8mP8rS|*+@5DH)GDO%|v0lO_Ldm0$^T14Vxo`$REpHIG*^XxNb!9dHyp;zI zfQa%BnDHuDjsAxt4#Hsqz90wuyH+_U{6h~AT3fB-{SB5YJ>7p!x-T?3BC(RIupLpD z_wN857QwAnN)ljt!Tik(d2xKB>(#&jgv`jxA2@jMRQo!@7X=9$%!)==u*L$EP{jtF z%vaAN*R>mxP0U`|`M8nMNlei=R@yr%E6lS^S3neq^z);|%Y@bYO49kmhP-@4 zKNgM0=~*ogE=1YZ1`-mf|{F2>Q8K-8`{zVMzAOVQ@d&8(}FfMn--d%?)eyAr3y0; zezac`^5c-Oxbi7oJgPSt$FIVctv$ANz;Z#P8}s2!-zk5-ygedA&-a zFLOJ;s^f9a8FhlLw7e6D8y%p-3?Q<{XZi}VHXLD=+r z78v!eflVQj)>FK2ngdB=A4<}+dXY-E9I)hZ_w?Hk@=cI&0zRVaU>(la1*guw@cvKl z`@!IqCbg9p@$UR*b3yak(!2A3w}Pd`oPVW{hqK1(BOHbO`z!4!|6HoT<&fkZD1=9& zEuN*{MYfZGvJM?I>B>Au_@cV8d3>W{nh`ZSI)*R8%#}Wt=y9QiT0LObD|Cn&Y_3KP8%k1=Dva0aI3wP9-y zzZHPpRwS>EdGAJ_dH^~qRSr680~AI4^ex}L;#I)X#ZA$RqnD zkObK$g&!Qf$O4)gb=Ozy`7fPFHlKFLcZv=W1A$r;aXJSF0K|XF{>lkW$k&J~?~$LR zw^sr2OhcHRkqzPSn^b)oUEWr@=Hc_y=4lJ7d#Z zTH_{o6jP#@?M|7`14m`VJxpIFTAK3m(G9zi$1@8~6%JktI<#?UAy}~-sps&lZoTcQ z%BlX5XY*lgW-W&o3zO||uz!@g@gmP$N|(FWge;|-@wljo*i4+E2t$?I#rlFDZ4H;y zTCOV|we=~<$%|=7{EVAdw>Nn5uJ@W@-iRESGUdHj*>k}+Wv@=5W(9qhvU4v`s)t;C zHTVNs9TbJ^3g-6CWhL%ebS5mq3dI=_ld&=T$y3qI%{#V^jzaU<&)ynTBPJW#2!w&+-oD20b_?Neg;y7>7NqF`n!d#yq@X-OC97uVbo`a+v z1KE6l80%)&9XN{;Rd!E{Iq$+|HWAqcB8DJe6Yq39y+zYt0&7=96tjl3o)$czu-Rxh zZid*Xw&qXQ&)2UbhJ2|@VD(^)@Fv6(K5ZiJ0+Le8l)_n^52g_$AE3J}4_e~v`Ug*u zA+o%-oCWPD4W3~mcU*l(+y*9y-`bbuS!naP3^m%`O*$e<85$YW&iZDKsLsJQA@S)) zU8@F~OfrThe52#F1P_MP-@bd^ICS$Fp6Mr-GRG|A1!D&DO$RSeHmYV>>*ozF)#L$! zX$LWZj@tK{iAW}|hEI(AN;9Re+AyeQYaRac?K%Eq?Dpk~#RWp?NUq&y09vn)2p6S3 zse2~%i(=O~moeJRoMe78&kfIbU&r)SFHPxeszo}H?L~eJLCow~{u2Lap1sUX&j?{^ zVlmX%j$}lTf-ZgG_3s%{m1ObZiGC9If{coq-O*s~8@6oBoi(nj6MnejJCkLdCC}dQ zreuW$p!?AEdoQL5=Iv&t=@Y%5ag(b5`ex#l(v;*?dVp~oVA~4E)I9h0hu95=P2CmEep6c89r>HYdjz1uLRSaTQc!d|6By ziWn=FYE=w7gNv^{34fFop$-|?&8dWMKGcn0?hm>TD+`uLQ@%RWOXp)^+vL9-=c5fp zc3#&QibPTUsDA1gc+RnLS7S<1&BWs>rWLuikgyt!0~Q_t;oZ_``#)QX}t*^G0EYQ5)Ous+|~DE7C{wb7sI z%PpQv65qy0FOuBS<({0Uj23x_vdyG5#-0_ly!Jpf+uS|BP?J(DY$EY8a@m$OOjgyg z64B_X+%O;P{eF43Aa}6$!)1pCp6A4=>BdEFPV$oa7ejb{6Y()3RA#pqQv#<^BG2CJ zJ!pV6v~JKMaII_FKBFL)_3EzRU;Yx`4iei6+dng(6ND-cYS!rk@q6EZR6+r2!@G>5 zzmaMP5jl+;KM`}M2T<05y#xU$V9L}3Cl091dH!>z1x12(&*F~E|+^BGLTHia2IL(VTMUF0TTUd1|IGxSBuF5gJauRt*zk6`U zU>=Q>tbl6mvtm{VaHiu`6jjd!fir&%fy5!XALEdE* z;l>QN__crS&jr-~*Z%x^M!Q?^E^OeL>Gr+3gL{|C%H5pQBpWs!9!H80v@KxK!8Q4g zTH(!dZ^osv)=4~4T?F?e_`L2Dw^{RC-E2<0ZR*{)7=-@B)`y{m(nPpEI;`@5ql zA*XIgtxG}EW2Rq$?3X&}Y|}{FO@<7@Q4by+6IcCB_Ks(b-!;9jM$OYqyt>F|uK6Y; zB58R2J1vkddxCY^riGf3#d%^YS>j|p=-<+Hb1WCTOji4wKp z?i=_5Sv+0AHM)5WMyPRY>F&qfHf=4rnju>38R}7uv(U^t!%ALp!~-e}%w&mWAM?R{ z1IcoM@VQ^%OSLegMK|)3ysET1v-I$fgWUGlgqBeP1UFxQ61d5{U%FyG^tV=)4@NFO-vx1SD$YJb8E7OH81;1!>o}q zGtYC5#V~dclFf;lI?DDkhqDg*W?M=zmG`VasPH1cn&EO2_vH)0N(ldtNolZT+G5;+ zh-F=m%AfU$I(9n(*XLyvvGq(EkzU@04C}A>yR5YdTwSk>%wJ-DQP7IpmR&@u`b;`S z-CyBSH0{J2qp_x)>6d!!4?m8{7_hlUi=;7-#$VbMkE|K*&1U{1iEevAm`eGZ0rn%H zd@O+zt$WlEqVoe?OF?D@C%)C6-}8UTfIWNK*M6}gZa3ggI5gu!2Vp;N4u6h}jO08f za}@i0etteYr@V#)t;KRG{fGR>#AHJ=F;e&asF^)xWl>t95Z>iB&jqHjkB)sck*+jy z!WB#l<47&X9#8$OxKmFj)*~syuQtR_$ykF@SY_;jNqIR-h1i8gF`=S%$r3VZ} zXeY3AEW{{Wep9UeO077&jw76Rn&V3~ety7qo0-5illwjwmJm*aj3K0ZfSb`ELy6IO zqfMOWN!5A)mt$H8cL`&p^_7|2%<|;77}`vf(`R;Vg6-oAYeVW3RQ^T-yeD@h74Pg% zj?~p`Q#nbb*WoJ5ORq!$k`!HJ&(?sYIr=hLx`e-o)|9szHcafWVPekLz|26!4(esV z;;nsMVZB#HJV8ur>=X0T)8Iz2LR>tgAHx(hf3A06v*Y7-LE+-ZHFKrqf|>hAD_i%v zenF7kj~C0}e^SAkjE9CqMu1VumiyaBcIg$xdco&isi0XIJ^cCMbLOE>&?$&Ig89-u zME%3-gp%LVi)!}4#7o!*L8iCyj$U z1KquDN5Rd0Jl~ag8;_sE@anb=%!q()`zUYYAKHn6=-|&+`v+ z0d1!T37fxo;^W~8MGLKQ2S0?1ID`&jm^lSKAL`{0IN@!W12e;?FYGkRYh&ToNq(p0 z43?VsSKg9t?#P9a&bEJ6w!(e@9n3Sw;uqL zCRaT97f|}zR}-5C!+b9dCC@lP@wtMCRQ7rq`s zEN4VhWISzj7>v5M5Elwb9WQ}pSmHNXgQsGiwK*R-DX-4U>jv8cyAa)o8W!q2u}Zms z0M^$_Y0#m!RQ4Jw3-t-Kj{N+sm=(J8Ww1-H%S^DrCG{Pu0PnbbSBIk+m2EUjU^LE~ zaa$W3>k={U$(DP5P`u@nFwBR=nYE{8`(k;uJ$kOe(%r`9D~FbtF0gkD|v*U36@)O^1|8h`2(DD3q%8NGdJs9P6x`L zNmU6MhlITTmD2_!%#?rWmO%F^?(XhZ-ULUz@;u0!>p%Pn`3%eRK1jH_0e6%JyS~?7 z#{=t8j}o7v5fCMH*L=+LosdFcPy$bI0i#1~vfhJ-5-O;jzx&i@ehCcF?#y@c{y zoHn)a?^Y<$er|Z^p_6W={W`8rE+$W8!*oYH{cW}n$4w;z=6w~sk1Xd(YkGFl&Y&Xm zc`ZtJsVK{!k2js2B>igGb^i6Rcm z>SS?#?nM<6c#jzTJ#T3-@j7{TzW43}tRt!`KWrl0@IS{%e?aJg7+N8O@%aGV4zVR& z5x_9?!O%ztB=;`Q6`X!RLtj`pw6VboW{IQc#Gt77PT_|S*ShnYT;CHBT85xaVOSp| zkK`B%60PL*1K13!o!2WFv6?z_ZY&O(vqt!v>F#`10_|?>tR=eFvbdBy16tLWss*3B zIo%I0&!Ig}5`I@u{WIpxCEx#wUXnOPHR^iisZk$W6fnvuB7Do;ojG` zC5Gq~YRU`}C})PXu8SC^+}5T0c6vlJntV^yaR+KNwoM84-dV@m3g?jYK0^TfWVk7zk}1_g~59^H?`_}8Kz?eGeA=|_F00#iE(QjG2bB{Uaof} zfTCTjWIel-R9Vx8ztntNp!LkK_8t{*_&%G>A(DI~S(@6?Pq$|eF~LV1kpDV}-n9Cj ziW>d3M)DvQ*$!c2N1)k2ytuK2x3cAu(zhYd+XDC%NeZcHhy^$rG-=Zy0)*7gF1ryf zpfy8a6(Tb`vsIX3C3h2S8qrIU7CrJsPELlc9Kai>+7pepotoM!rLD~0(`oZiF@g{EHZzH&V+(TlqG@g{0 zh*)F~4G+hb?G+#YBLg~A7`!5Oa1zeOxvl9APM1h@0sP1iXlwp61ROPxuq#?xdGMBw zB>jcQ!}VlmoYKl=B90!G)VZTJ0oNWrV-tz2pmS0lb5M}KtxXQg>~gnYc3++s_{ML7 zska0{WJYN)p6>KfqjI=X^n+NzsPT9scMwKZPWILdHz${F|Hn2S<0SYg!_SJ>FWNkF zGFYcU)urLz?hkSWb{eo@BP`t-ProUga<62?l`mik+t<@Rr&xyvSH)-OQ`94evYmTX zutiEb-8M`;mr)V5^eO8Oz8J6Bg`)Cp)jLmFR)`emVSBA$G}@a?Zt8(;DjPA7A;mgN zsW+lklKLt7w(r?L2IXXBihp(WycNJm;Hr+K9R4@^6uNtWSGD^D@vm+o_UXXSPr80C z2?!x#sSVO5R`LXVSh_<}g?*uKz~?VtNF08B%xnM;mr{HOD-iu4e%!5RAh_&Bk}W4k zJ9q0x|1xG93eRrItK8@l{0jbl27JdM>8rmXx`df91?I4y83wF zLou{%BbAPqJ9!@@o_-?jiZ_nkZj{MgXu_qR<(i z^8MYZ{bzR*7MbIAEIwb$nP;jarZ4@IXbyXbEr~FnX}s=PgJUZS1#e^z#G=5B0?UA} z;Eibyp8wd;OComV@X7oO0Ys)UqgpG5zR(c&_si_MJD@&u{v0v}yaNopmGSZM*45RS znf0KzXj4l|=JWT~w^5Zfz2FjP>C;}!dHl1vd++3IAIZ1&hHw0Uz)|0T2uuTcGXAn5 zaEkw$k3un|0lVI+T92Z;|0WR!NM<{4K5(P`t~KwE*qVSF`vQpDN2t0l{>u{vgc?#K z!_~MY=8pVjziJYzifFB>^76Z8<&WEi$Vb%l5!>a2eZ=~GPfxXsPLkHDWR8?wDMUdg zMF%`iyoTK!*jFBuLM1^c93X)asH&<$5~e`G1sOq*KjLTE z-2E7$p5Q})4$jWThEcqh3M9S2s&0U%g@kP*^#AXj`#$+J`W5M495Fw>Gtc>W`V;!* zP!a&IE^bfv;C-lf*o^$0Y98>7`3Jryp(Ed^D?4!nP zAUyMbx#U~s3GGO_8oXHGM?D&H*Xt=KJ}a80FUei;8@^g0)m1J?W9@0BD9&T6x={Uo z#+3?T?gX&fiL^R5|IV!cEXC5CxnVRXq4icp`9)BsM1bOFW_oNIOJMwMaI5v(FN@|h z;K!NY^i`DTt5U%Gy2aL!P-(`E_q*hciER3?3hvD0bgwb+g_w&2jD+H6b(Vwl8++-z z&#i53Z{<>P_?ZUC+`)Tf+;qw5AX1RPRIM$`0Il)5zWFSTudk-u-kQdV5Yc%)ZkU7c zBY+fCBBt1YVLP^l|9WTyzMITIFzhVtYKrV_wW;XHP*h$sQdpnaS_5(O%gY5;9^#K`3 zyBXG@rit&|Mn9Zp&nxOuzKn&m*3G<)S5Ai(xFQcT5+YM7Be~d;)lu6!nor8E(mqv_ zaSw057dy%q`Kd%HS)%gsTYq_s@$?}-uj(AUg>C_ue;m$a@CWO8PXn{5L{DpmXwj+< z<;C|S3&FRb?_2yOU93LlLuc)!yfLY0<(smFw zWN9#)@ReB(wTfMEw9YeQ&vV6=SS!zYyZtWhf0zl5QT^z!8xOD}^(5*X9G>T8$Dsa_ zNb`3iUE0LoJ|b~`wd`Qsxt5D$6oG}UBxJ0B@s+dj5HCM71i>bNUy?2C#Sk$EN-m>rs-t3Pn&(3gfwya6B>3qdEbUXJ;-T1`B2cp||F`GP=jJ|im^@aFg zhZ&Hq)Y)PJbw_6Y5bbZu;N#9+N4SD+fIU2tp@$?%AW;JlQiZ=?>3Zq!QVdcS2Z$t= zU;g~a_m8lh{2W{rz-4WhbaQ%I@1N~Bbcv8byS=~vvvU>b&U~1nrE*TbTEp9IVXdva zPFoEoR^m1^p`KZ`V-9-q*3d?Da&gKVZ6B~*dHI8`*H6!IVD|&>I7yZFu~$wV&}4;p z4M^aENmkmS497;HFBSblc*{nOO$5N|Tw%Ifh#HinP?K;pkl*3k!E4|jsJ_58`|J}#CHEZ(R z)p&7hiMkj+Hk=^-=FlB_^Be+Dl(i8seB!5xPRtl12TEbIDB~faD9giV8G)>CD5pn= zOHvt#-fjM#rI_G7OV|t)czj|xKSTRU4L7W-3t-{HwI>n8y<=m9|6+1<_0T;+LJ)nY zC?XC7hfw5zK}V0hB?DmexIoA^0!T9Lsu=>$I`(A$#{^VHpSewvd`9t5Z1-poMhIA% zzXvT4_b)^$=-KtlX%~F6y5@anG$q!+=Ehy^8zcBzgB|i;xfHSFd$jo>buMp&C4~42 zgYNS@{0rFU5h2#Vp512w$R6~tGoSLxKhkZLWU{Lfw|mpbv#-@vRe-0pwzfiuI0PoM zjdgW($vwzKp%fImUIZNH_|ijN{l`fL{_zhBHVw-Q8rAS5E>!c;3#z` zvBfw%4WVRv0AwwczIY~8c*eLH^9(jjzc0Tnr;V^GabsJ~7rC0!w58PMxNh`jt9M?n zUA9RwjniOCUzX}W#^U32aEvjm;~Tjzo70D!McmMNmfCxl4ZALtqEM==X%KE`CO#@x z`s@}ig(gjWWfYZBLQf>oSmulz!&4avV>TaS%=tId^xyEDx&Hz{ei&1SfV7To%3u>Z zyl`~+jejxIIYpx+MJs?gTU9JLcgL75&y;d3^1$4`X5`X!`AGXnQW|>teu(WMmi?gv z33!-D+YgP2I;vz3toad&B9e8EWMaTWg<}x*VuX#^?IM!v3i%JK@Hlr!O)6S6WtSK4 zVQBzK1geKTK^erXef4c zTTfx8eoCwX25&}xJXgc>4zEK*!Rbu~CCQ14>qdSwcK9%W@tE4;`YJM~y><%NW_0 zL$-g?U(6BCt3avw`Lm;Q6@4xYagEdqm0MFEeR)JHM4LG)OO2Irn8OHxud3ki- zjKpRxOPNWi7)J$+1)ZUT4CB478g2U?F&y`7`&63x1$;LVUvEipRmCu! zMBMjN`Lm#TLROvMK>66V=?GsE4EYu>VZiGGN+xyiZ2)7AV;&q|K#cyonzKJJ;grRL zhQKWMz>Tem_ky1#{D^ zjg2bY*mu!v+Kt9V>CQ|eVeO)?%8sr$pE4^-6I}qu_l0P!ftbjj^%_O2>%q!*$7ThS zXD;{BMSNqB^=O@WiZLNX`Yz=XUl;#kzGW2m=7ne^B!NoTVtB)!=vTb=$7{xqE26ou z3=EZpQ$lSR%XO9v`*I3W+=5(}+a9 z0lxb8wsWSzVP|JY0>|M?fL?;6y@4&o&mjx&k$^K|2qAj>Egmq3gz*N!=l$1N2JE?> zKpZHzsF8qR*JI=eBK8m$axeg7N%n`hg>D~@m~)sQT51GKXFfi*PR#Xe!vkL$>}Djn zc?h^mA5CyVeP3b32m>_jUv3%zR02t{6kU#)+Q3ffeC z82_S(2rKuV+Tz?+d5){$6}epFao#cei#i5Wu7=o*MUG-FZB9)>wPs>7ip(OZSMJK9 zZ#2=Od+=|(^VR&qLfE5gaaVn&8F&Ang4TOk&##xk<8UsxvP-6!v7(q0e9JbYI-X~j z#o~Rs*mvv9Eo-1;?Uc5krHW?_yxyhe@*c~Q=s}lcJ+_N~=Dl-=sXA$$QgBf`6eD8<*70juBl!BuwX4d=HIcdC%9pZl1^M|>BEDPuW37|-6E`DC z>BB-v(ECUxq$1yc{Y`X?k)O;)mGdD|(KlS}XT@|Hy=;ak$yu)QQEuEl-AW8KaqR3A z2yPOxAG8_E68d_tv%IF}hQGgm!owU}nuYc|0zdax2x9ueP$+H#@X2p!S-26nu{`BI zz#h&Rdri=1W0*$XNmEnvMiayBt3ax5;xP?DZtXaZf>_fDnm#0o3jjGu`xCQz2esuh zg!bloduA_S9f>JJ)XwD}FDG57?fT7K)A9XPW3%SM1#$;+EO}C;%s9VcdJVSGC{h=9 zcSR2mJ~wywqMbNHH)Unz@~W!-uWNlzv8q~c=u+aMmA<7$H~*u-pUqD4Q1E`pWlCEg z^d5A>3>|~wqB^-3kJ3Wf>TZ^@1~w^WrKqM-UOp0Pl_mGAr}rVGkHmX9-196eb(5J@ zoW8Lpd&L%Bmv~gTFR|jh?V#_t1EGdOi|AdmoPt~8%IrOtMo1OKi()vBz&`g@fTt-6UqEDZ}Fp&uSm7qBP zHhGtMxN()G-dnRbLm~XbSdF^y^KOvh~B2AV|{X=!MW|6NQ5 z9S5>))<{>ikm(TLo zaH9cIdvJX>$3_;qB)9HD{)&C(1EhS%lCk`5+(RnC4cNbZ>Hc#KPxt~Z>o6-WycN76 z3-rPetli5zR)AIR*-S@8MO9v2zI{NTV7-8 zFt4=Dy;k~hAB*g980SPuNT@1tdoyMGkP0Cob0L|M33k8m&fH^mON=oSm9efiTjF%< zj+TD!K8LwKn}6yzsV)axQ8JE8lVQP9nR+a|G;ux`X^6cOzwY8sThjq!40ByDEZsk9ZL|AO1b;Mwj`_a9l$j&Fkr`?rtwSKuT)Gz|f}iKDqD$c9R{he7OGHu`&jZ&(DJ$(hiy#_@S)tmbsX!pFw35m>!Wah_y)OHq^>rhgK`))^H z_9NcctHZhA(}Ihi;+sH<))EUwx%Wp27M>9@2zkYk2MvY9Z=Xzh*63rs&Of)9&fDi> z4&0sXt=JpLvn)wE^r9xx z`gR;Nt#E!nlhn_oYOzrELKjRxJ8dH=EsDUz-L~3}CLz?*$*X96GEBQI;-e+u?Ue3b z8av|dnu9YVYLCMFD7^)BZmLL8$ZuOmBU>2z(W6I55%uu!NEj}8@q((lx_Y6}mEzT_ zS2~7orOgH|4lZ zFXLjt5uBTwi`?Og5C8t%$H2(QZaA4KklGfXsrSz#4Qb)Q@VF~sR#@3*LV?Whzg=>1 zpMU+@^`$83MLXfa_kNt5+ZhOt)wI z+y#G$Ef!tf_>Xcox)S#fj(LsG+i_hslRp(}29-QNk()`;6NB^hWu~0CfN`wp_S>oM zGZ-sZzDKCmW5jHRvMxFImU}QtjHSoQ=%;$n3!@9l0ba($=k25=H^k zMRXmb?Mk#O+WYryaiMEb{Tp@gj8C2}2z(Doyy$Fp)GfA8sx6l*;9bh9#wj465*R3@ zF{5N@$$+Y?)U^iE!{8-4y4MQ}u@L5MfBba~Ifj9Ofi3Y=a16-`p+rjWe}||uTx5x~ zxO4_S6Z(tR{DE-3DDe4jcOK3~DvWp8W&d_~cz4qJ$5lzMy2L)?B}$=$-YyK*m;+|AMU z7;yY7Lseom8Hj0>vAHey*}f~zK5V}@_Q$D1BwRVrzHdC&P~MdJ@f7O8oS zoB>ZJ9(+p8y-RoS9?KBqS0~%vxHCW2IBBIo=2Umfw8el4?J}-Xu;z~$t6bIGj##hu zWxMr|^rM=2@&XE_wguh1>_>Df{6jZ3$WN|wEL>M8{xjn!I>|U9OkYoki{2+_ljGF? z(!Rvq(YinTIhj(M?I1arTIx07u$-*jzu}p6|7A$`QK0v*Mr4M4!=j0Tx0CTi4G+g- zXEH&19YYR-Sv*|y?)`|G6I&ll1+Of0%#amKHN7WXT9Ts17f&*7+2wEPOw#7i(|t@$jOppL_S9YW8;mySyx9_7inPNv@G6G z%I8nlXT@cvOSp3V^7>HH&_w>#rl7 z)NX={yu*uKqpx0l;Md;B%f`?16_t<};+>2I1I{+Z?A_=CR?AMy}~kT%JE*zeyMoqnve)c7a67 z79ZUOmGFTXI|>{1C{>wiW1@7L`;fCOyS|G>`^#`ZN|NO%2iHMw*#zIQdx)N>3`q1UAeyd zyK8vA$}l`O`nQRlVDe<@=Gk`4oOQkrQB=3;g66r(`)<8Bh>&pZQyzxY~ z@g5rb90KBT>>jDpHtM#^Xw>YHBWoXVg2RY8Q*ku?*Jr}PKD^YPX+JMmW^eVc`_w?6YSMt)v*Rye_YFq7V&BG%UeEhZJS~{jYS?OWfNnc<`Y-#cvZ_ zRV?l%yXeiq-d+8w>ptb{!jWh0>@RHXzS{BhkP6{$+fJiO<*^whV`_E;rAV&3t7Ot@ z^G+n(EmYVo;Toi&RL0Y4!Iw4nUOiM8w<)B*F~KA;mN4Bw=N=N9C0#JYc`p?Ei^s*9 zgcJ=m-;-F8Y}TGnoS`$iXcL;pm{v&7io0EUMVjR0`j=2{_bCGo(b?>yEPZCbE=zuD z=r~?A40c{#tBoty+x$iDES*15*KV$&qGFQw>D%4nIRY$VMn#uLDA zqN1XtlCl|NSwC5Qa(QCYXP+bG>y*Gqfp(GKB3Aj~#b+Iz0}rW(s4h(8Nr_Y+3-|EU zc7*fPv@qA`P7P37Cp^3P*y-#}RY?if(;JU^S385?$iuGX{dKi~a1KWZ86 z{`Tm0NvCG7)&H*2qtuT9{*@F8PoVj+xLm>^F&Q3?ft+S2(fp5jVsr<|xq7jRu?^xb z_p_z!RUj1xTLhNR=VYwiZ#B5ra*iv z+E$4uJ|IryDTg%MJ?accS`;e$;~MSi+@0*Yb4Aar?9!dQ#VuAJTgCYaC6iv+C)V*I zY--V?rnuB-GjG;Nj%{erz0lD6F0B0MLz-EaDFbQ)<S`kRoK;{*99xK--x>J1xAYu>!zrn#Z2M!$V3Hd< zSx~B92(*)sCp}ApMHn%x1NtENq&V|B&~J|4PKHFo#KP8DJvV101F5`1qcJ2`5}`e6 zt}QAZ$i3Yamo4HvNEJ7cC4@QpSYYyP)=R`mQ3g`wbVnmeTY$m+4b%~(FJN<^C3E_< z5AO)N{r}eN(ssrf&$}`Ob+`>Dyl+(Li7#+-bL*Iz#)XH6cd(Q3UeMT)mwB7Ik;*Es zn;uE>DW_CE^zGo9AP&>iC$TB$bsJ_vT?3?Sp=0vl=H#A!KeZ98a)}-07Mnk*8=bad zJnfQIuY^f!m^6RHSy_rGKTf{;(%!#Nwvk6Ib1m6X4i^dexKu^OG_T=Kk#dAjK%B}M zPiU(3MNG^BH)T;K<>tLxo`JU>Mm)L7ysKn=Ge$C z>g#72DoMs|GOvAh-nB*wMW?2cr>Cdeyh`VStFQO`>;!IJI{nVd&Q1m-x@L1c7io_Z zXoN6e8VsymRfd9nlg7RUe5kxZ4RwR>=;RFX4MiZ^l9TVh4z`}cuEFV-ro;-R*x(9X zgY)DdQo$Qe5j7h)c%`jQfa~5vWYZQV$GF)50l^Fc9bXlu9IvY&bbX<44Ya?y{fH~4 zXh@RvtEG~^S58&%ws=^F{7D!*+$S`e$vq~DDVB0?vdXA zC@*T|Am_|7k~WU=63?ehmN7c=4Bkn6OAoQO`vkAGEksT1mNr#xZ_QmhgQqvwe(qt(1wm*G`##~nNAxFyQ0Vkfh{g-R6L(V<=U$WOP z-@PP=qik4I&W*KY>s?C9e7a4+>rZbnpYlVlrt`~8g1l#z6kBkC$B$zEsv{%1fplRb zEdd!*u@uP$9u*oxGLFKS2^xc0bVH)HKRABSE0eHG^kBsZOyPC!hxJpspfHJNvAM=a zp1ytQHWRnZ)c%<50%?Z+7m^QS_4UH(85y>Vjjk>(E-#VC4ehdiQ&VdzEpQVSU=s)D zHySD4(DjrlFaRtN?ETj9@f!FW_9F?xFM=UryXD&;-8a|={8)_n10jI@)6Z_&XC~20 zk%Lt7CVQV|JkV5VK$9WfVRd`V8OdNXs6jhh zyqIT*c%%~Ye_;{No{XT8`jJG!9;8GEPlZk3J)4xma})mNpPjqRakwQ8y}yN z4v$vZBzZ1K?RywYZ$hM+QMkyTy$iS*ycZPMv*{CF-?VW0L~cpUKCi(gKk~i6hU|fN zn+2>&6>!T@nQsq7>a@gOYsVf6f3~U;v7RE5-^jh|k-W&3V*UM{#2xuNn6?+Nl2ZnU zS@DqT`lTY2uv8e|E57_1;Y*?5b@xmEwJmc%=1(&ELWCG)MHMq=&y*(dD<;HDx%DUBRauFZDj(nS&g z5f0#^_C1j7Q(|Ldo5o!wgI5P(f=vO>^dK9Bk_6td_q+Q?jt}v3&yl0(~ebPS}4% zFy11;6UKvo;#Ab8;?;ii;19F}|=#C~9nk`-uvMLXynn zS1fr4{LWFV%2^3Q!}Ww}snc5ZCTdl|8N3pjiA_KM;#X(%%o~whl21_N_z%S>sZf>^ zG2)DHYgoozwRw^#tV)iCLnF==SgW);s6hB;#*<+8b53@~o7&m#(o~Z~NiQ5Z!v?Vk zv6d_9_b8O8{Ul(V=(qbaWM_MD92=(am{qWWPb<~Yr6@18g5&$hmpJs}AE`Q#w7;#} zJ~brz|J-8@q1Oo?Q#JXT6;9kbk}2ApB$+0JM;t>3fAVs1u{x$6p|uDyX`n|y3wrg* zu;mAh4Mz;WsfP=D0cAg`RdpR7P&%Vf@w!iM*uE*({d#it4+s`BGdF)aC!#5G3J(N9 zU7=kPbQn%Xm+;PEyrqI6AJRQEIP?RSv;@G1z2XHX?*qY2V0`j-*ALCOH>4TcZch~f z^3(@Bd?e4BhOBG=kz&z`I5|jtH-Z1dYfxJqi9o~~1cfr$pIo{caBCXswjy}c?}Pa)8iE?v5IP!Qz-vCOQ_ITb!%mh5K=BcjL%; zzS+ZMV^><~Up+>j$-8Coq?3iS_s6k@mq<&3ubdGLJ^}Rc%FmNTl)f{Mc|ME9kt8Q5 zeZBV~nQN5HBl=;D`i#=W=eh1q7Z>fe1GHR6S5l8nub$WO&s{!Z6sof=x^_Jzh#=5G z4@S}OI<9~R63>Hzu~O*CX57=x6A!P+GVWve^feN#x!dqC8R_Ua*JcSPuq|UEC{?P1 zWPX$vhM*?qc_?*sMFauy%6q{vG?}h|@_4A&c!u5^yiQD#FO%yTMN+$BMcN-LO~wqr zUh}e9Zt^nye}d6DIC8xFAG~$?9IyF& ze3KokO2$sgXY+Z4NI&yqLYyFrw<6oD&#RNXtquF8cf3)5s!Y)=;zaYBivw zs+f9nDxGm7wF;MlriE=JsLw62mGZ|=r4q$b(!9YSp9DS&k%rfl?UZF)!5jqM8+TN& zY#-pfy)6ivhCeJ#g^3v+WJ*6rYOWU61*HDI^P5Mu)pq*8bd?mB{r9dPw$d_?-X2^P{|EwU&#`*#Xx8bjm^z5D?;r_eK6!AmOu~+tpu&}KSX1e z!WCBUZZZ8ILZk>Gn(|;=i&NY zD=t!>c*fXUeZLifHp$g$yGTT5nGj}H zFGE{JZzm~`s+z~un`YKS!&-@z8GCWPU7_IP6%Kip%FXm7}3<^X+_)s z9Z@?-9kf~j9-{GhG~@LiJGrIVuhYBFMk)A`maS@_(zmqGfWu6dOrS872INox)U7ao zSndxij5;8U-CAj*g6ju5H7c0&*SsW!FH^ufQ_}+ZSYMIL2f5+k9e-6vt93i?-scKr z-tYzp&{?WCtOq$FNZ$&JuEOVczu5tjQP%MYNp42maUrw%G$EIS>}QV_%Q<`qSVD}S z&dJ{y%a^|U0*kDrr6r1*GL+i_UPDWZWUas-qv_J1?>MSjk*;U5OMcK&E{kKLQ&fAP zH{PL{dh~>#^y8E5DF@1nD6e)(IyUFF)cnrJtQ=%ctf`bT)G`r!Wdf?HZ!(^il20@- z`Uvuumb8+R^0p*)-mc)^noRFb^`N;aNvclzHhhsz{>SZYV)k8JOrk;3R+e15I=_On z`tw^EBX{h5`gSF7>F~c-XW{r4Ua6rcr*w`_mK52)&yCeGuAUPiRF0xRea6Cyu|86q zKND<4XObKXMekn4GeyummmGrBbgrji@3H=95D2q;vdW~FGw{4lh4dRk2|@tU6~pBV zBcP(9BIeI8?~=e6fhJ7v=1l@KGc!o&WrxPI2&xpkwK|lkx}-`d!La1_AvUiI6J0;_ z$KCz)@$qxeDOOpul7j;|KVoAJH~ZaAE6IEs1-DAK+sU(Ytu zPVO3-PyRMQ4nIHf*5eNCxVnG|EqO9IaW-&SnXL6c` z^ZB_j3UT#5{})VGu`{%LDGW^U*~a%+C(_?fS$Mhd<4&y2cTOGt?J9^P=)U4eKT*9T zI*X_28~*&q_to$NofBJADu)|(AHxCjF`tmxyl3)!Q3 z(=FWpUhpM6!=joL$o8e8Lf+oqzL%BKo)B;ZMY!Q%`BHB-eU(;nYwh|7RN|STE%!`r%}Y z_lKOjh=+$X;V}K|VA8qHx6DJKt()z$~q?2d--bmM=MBZ@MJ6I0(eWB{9 zW$~1+el0$8kRla&BNsjvB0mT>F{j?mNhXoh-Dc(K9jG!zi!!l(|m4_1t0~8s!8T?2nkjk7w`d?bYCoAcac189rdZfI7ADI*o zC!P)|slyf*L5|=r{@Q#2X5}oo#s+h3Prwb9)By=#8{Ux3uEax6nwK+gPfXOpERCxp zO%uW?)0g)y!iiz&)_RtLHDGDPT(#w|7MXp`C~Z@m{JQYXtaV`mmSKS z0?l%XRdU9cSa>hKn3Po|wCO#v4)1xntehJ9W8V-4?XaZ#L4*8%#P3nF)N3o@a3tyW zGb?x=K@Rz{;06AswvdhFY+<8`={A)#GG}3&5!7+omOdX3_i{WwF8T^CsS-`0V0_14h|QjZZW0gQS7kFEEP=X!tRhixBw@4aU>*_k0D5+#&f zGO|}>W+x+iWF?IwQ8H5Y$cTtUwve5SknZc{obT^<-;c+A9_O!f9_i!re!XAUbv@T* z3)lR_IW&PqRG$7nJ&CovGm+GpKfg?oFKB;!>NpzCEtdLd50V(%3;`S%KpPm&=T*Ax z6l|{a=fgXEN}xA5({zLE@eon$euU!tDK!^MISDPdM=zoqrC=HTxE-Sk2xzoK9W!Ts+2+!RKLt9 zs|g>`(k`Zdj9jveQ-mnJatRWoA?qr@z!;`y90aE~A&zt$9XZBNaP7_P~f?@_$IWQ&#xs(Pue z&H>4~KF;?<+C;1iZUoNndg=XtBmsqRMZ!Bh;Hz@85g<$?3IMyOrk)-VY7{Ljm`o+_ zaT3tQ9n2l~t0{vrr;7Z4mmo`r+mJxEepUH8#bg@Iqkw3kOV09cdXv1VxSA z_#6yscmpqakrbwY8DaUJ+hTIuVXcy+Z+4$?8w)NjVwp0B$S2RFXKRpp>i z0@(wQJG0z$+IN|GyH%YhpnQCDilkxECML3S6X+^UUBFG6ggY>4~T_(|2w0n(h+W5;I!Gj80*5MQAh~XWV|{Wl<&CdFnd5w~EFW zZB&6#^Ktbm`H+msrAskCe%u@$9qoKoT~@{qG&EA<%R`)yq+J`vl1UK}n|LWK|Em^- z)zvhx{`cAz=o4%~Iw(QPvksz=nJ5~gCa}b69!(RT!6?Xlr|1vdKQGK}&_h=Jo-d+) z&O@GZlaSH_>#yaaw%OeXE!s$CP!3Q=}?K9=s*OCgy$VV6DU{gAe&l9Ye0YLgnce*SA2buPoSLQMacH*o*m z9ttGbSY^&l9Ge=WxEK%d;2@4 zdET@P42#ODCnskW`@fT;xm2I@W-^Eb5-VA+hU}ODy->R%+2H|L$P75XI<*U@ax3vbn#=^^(w<4~E#&6R*rXUwee=^o48= z$`%wn&PPS;53bVA*sp@KUm{nDhwB|Yp4{QFl+cU7WMlJnB}1j^Cw3Uz9T< z?XlaAxr&;eP9D>-(J}rw4(pu~Hn-!mSrZi4*w{2BI_JA?*)SE0&-SkK4Q>zKcgcw! zEYY$5wKTZ-JNQ5B=MYz0|Io0f(Z#n3w96gUR^6O24JgVt``In=x2yF1B>zU zy}6E?CgE{I1r1^1XNfhn)6ulg6cW!xv2LUa=^1hN26K03==Q1JT1+7_w6o!g!UEwD zkC=EGqk~0wvEbM;NolNG*>Hu|fcQ@L-J$sV>vY@-S$(;Q|4f9&vTBa1243#694k95 zC5d^dZdoc8)y26CbDq19Z%iw8vw}}6*6F@je?pN(>aWrZW(0z}4HXYsdnC=*FJgsZ z-4~Unj^iWHDy5Y0pEYP!0Gj<((8>(iD~&MCRHi(4vQAcSQV@fkMPq6m5r53(#jihd z=i;P!put`7{g|*gH|eFA&v@pJa>qTEn3@{7rTT(xMz|B9T2uTn#n*!O%o|NqketLQ za;XBeGS)3S=gyt;07`uh!mio|bvW`{F+;5~hHqijJ;H&(lHJIDO^`x)Qw`jUB%mce z1-DN%^C-Afw@`rZKW`?_EuAC6f|jfqy%?|-vFLn+2`q5lpCv(0>+(xGd_btuJia|K zeLS10FV_5Xp$B=nu$t#@saCQ|{PUMzq~%o}Jhgta@<449+TH?{6UGQgh)SDgoVUO- z-B`KNd5H!0OYP^F4_SF7H>XkpHMwi;VI-^nSx#m=Lw%X1<-fvS*JyRY*f@DR1j%RdY{gHm>=bo)3pWvn3S>&am%vH^ zz+1P#)HV$^-R8RtS7Xt$gkf4CR>l{i$^P1|Pt<{d&icX%Y`S>qdCIZ6Uy)=xy`W;dDIl;qm449L!LC^n=>hBpIHC`K(f{agHqQ zTBXrymlS@MSA=7y@@kqqP{w@MY|(*In@F6@b7-E+CSXfaU)rGZio#Nc&GO{i5?S3O zpc9X-nCb0~vTFT}dSs`}JuS!inv1Zd*mvAennKE1w!$a7w_3euZ60pLbEOkKLSLi>{rH>_k$ZbR!-lREY@g$5a*uA^yrb# zaVniH2!R}AFgBLF3IW!ur6we&wY|Dl{XM9t?Ql%yvVc)HkKpUxCaA{C7Ls=$(2@Hj zDo`*l#@KUV%)+aWcXqJ#)=S^}^McfH0dbqH9;-U;?PIuQKT(BzS6*voq+LTOadV}<#7D=K?rXQ?3fO{DyZdNQ_|@T7=KH9;1uM96_p@SmO0kgxqP8hZYC@^}{3 zy`i#{TC{Y~QRUd<@!9Osj)mIR2ay*|D?77bkTC2{mtqI)AH7p=7WVt8smE(?A?D@8 z*93BR8ki^>E6#wXT2Q@ge66Eyu^<7|3nzP1YbQ+n8SOn3^JhY@_i&|ovmd9$0}9*9 zOXcLn>e)tx-lM_6!EofK30SXNtn}X){qnlAZSsd4Exm++VpH6cv3apRUt9JcEF~pMeYwOqOiGfx^buP|CCl~-`>UWI1IxFjjT zdroa!H1O3EkTACocL{uh@L2An#9g!#3|5!Y_}@m36~r`^o|jw$6CA_U09z|bCv|n~ z?;juDN7+1}0eOLAOJPlMDr6(aAw^v_g$v9K@D{mVp18D`M zMmy+JVygQUMT!`QxIq^9LgSOJ{2~Z6Kx5#vd#BhTf>#d=emP~$GaeALeeP$eIspI3 zyBG|QIbb0@U4Ev`v`HH~$C5YrXB~t1(QB$_Q!X8oe^fEewsk@t{mq8#qC2_MY!wa0 z8L4uAaP`RfElnlKdgaWSe5vQZWZJyvw8_FLWBMddjKxJ&*18-uv=H(1=I%1dbNNrs zt7EPSGg`r7?ivKC-lSeFdnF-w!JB#xyJ)xG!R^0^y)rQ0?%c zb&0#8>NT%>5=#%;aW_Bdi`zoavKgRkvVx#!6b6S3F6?_B8Cx%13;FzPUy423rz_Eb z6Khc88MK@{G}Ky73gDgdfHd!-N00DPJ-r&V%r`t@Zr|yS+pt$tRI1z3w#CeU-lJtX zb|Xj6DC-Z$*r)O_!9al>ZFwCBvT$RCpYk-(rv~5VeFW_hH7yTufrjPvu4636s|V-d zN?j#VF+}4x_15>iPC;t>>x8H6rc={KGJw+Fn8>E53Y0yO@-m+F(eu7s+*|N%D>m5X zMe2g>o+`b`Skw{+o||W+f)$rWV7|Mfkp_lqdbIJzo?M~y0I3+MM4Y;>E_x5X_hv_e z`VJ@lpY1b$C`8_~`^hGmjwpZPlJ;s`{~g%yiwV_S^L0_ZuNeybC*MV$ojCXVH!8n=&PN-ibu4?R4l1@f`6BRPpNO>*RV zjDbrTK=d%ou0}D5n^4wWv&blNX3jGbP_X?T#-$lO4WX2Dlc9e@CPMZ$ByDG#Hwk5i zWr>gc^C`p|MqVQc+e~NxKCpN0L5klpZIIJNv-2#?V0P zg9HJ;r0%z$^;VcyhdCzSs^|`LPZxc=M$zme|BX56%*gw0|`j(3%uKB#yzWqvseR<3vt(uS^v3XiCN9ozi7tSgmcWAy<6Dh=M#v>x(Jz^kg zT~B*@Dp(o}m%y`cZFO?kgLTI_`!wYf+j}LyE`CzJkmvj-PN1I`Q_RqYb*u?@Hup*Q zsG9~54tlPVJ&z4@qmkz2kz4+MwraZ!aNLG=!Vo)@Bxzj`^g{^1o_5i{ps%DK?W-0z zhTBfV>~q!v8r9B?2ZFo)CFm@&7T~&)C0CeFeo(WMaP0^OD3t_@;{i>IM++0#q~F-2 zXd2&=aDAe-J(r4SW>%t=H6NWF60FLB+-Jxff$n&}^$8}^)H|R9LAyPp;1!>PrLmB@ zt3*)0tl@^+8uKtkEuzY-KKEFFkqow+V^cK^rf7LVC0{lt->LyD$yQk*|_^R*GlNO~e;86NKEl~}d zvEnS)isRLXrl)FRXyuZwk2o$R&IB`F;$=-v-Uu1(N@mHlpJ~Idce6k65=5_3C|(IN zuuO74n<>FdfF&weWBC2%_fC?Ap6oaTM=Xhl*cA}0Jqc(nYzcDezsD$=9-Xcu(2AKc zJRc>XFA&6FPg+&1W=^YyVT`7t-;DC3RsM@pIf)xL!8xXNbmODHIX4}tbePiw*fo?_-PU{m87Se z^q=X`%DL7?W7A>z@yh6dKQ|2-f=oCuL8&|^PZjzJr|i}l1dz7As_};~HG2?JN%mX| zT7LAZCQ}R``BtQZt3HB2N4OU^GqT1vX{6vX#8`&Gpkfx3I9Wu=KRlFTVfdhK6q#mG0GL*mowi!^oM~mwS5d z_>iM*9+x8Wntx%v(e$7DN?S)C2!b~sn#;Uis#DbB#=X2Ay-V%zwNKJxno3lJU0AQb z!AO=wfuUhSOrEv>YFpMf4ADBDOWB<~vnaJ=YiU!Dt7IIS^y7AyuDR^(8ue%6M%)>i zgzqeN{mUGK{5s9mmh%{HONZ*lM?C)|_9iSu4QSOmd>A}iZ_B@Swx5yKjLclmL$OHk z+#CCn)bDkiqG1nK2U~w`mRiuep4WP>=0Se{1)Xf)c;(|6kyioFxGrN=;E+va-=V(x zyGH~C1fk@7?KnwnwTy-GyshCUrXWxr$O`f3NAfV=l@Z&%9N90mo;<+UmtXWF*Zr(8 zbkOixnQigJ5yD9zH;+s|DFmtyDZo4M-ag}@&nG=WF)<^TT+k>cO7}N`oV6qafhf0k zSDWEo9s272^!e3B@Qk6kV9m|wp(-2LW7+QuI9)>`lG}fd;l3Xn8_OQdDl^EAAfn;7 zL1c}jW2CNn*~dkaOb)#fGHEF)wRr+AJosG3K(hh5!>-?fR1;NI)U7cTF;AcBZ9tNc zmEV|SEyT1T#l%GD-#?dM-=zDhIKK6{3`@>!r>NKp26dxEj+H3Up`ACf!jh3lTW02IsI*F43+97Wx7j!!#pHw;t361 z^1MG*_P+f3WYWXEN;I|sL*tkGqB}iQMrl@2P8DO_Dx#+DSwcO>JD?%cI^edT)iD8PjAxfu>i*|F%lA}UdD1*= zQaqyxz1=A;{OH;xb^jM3BS3X=X8X5YMkwStEf4uUKL3!NrwcC(FctBl68o}1a6Zq8 zZS=f!&WT3;gj+DB4gZ9(utkDY@Z85UI%Ba!n1YScIyiw~E z+=sEE3RPb2EsqZ=CH_pegke!m-(kGUma`hB>ASl>>7erQx^_7|eeCY~+3e4ILRhJy znk0sAr8Q>_gq;gi>c2Lq^g9l4N#9;e=pt1ORl1_Gvx#?w%SI!WiB$|A@^5(&GM@*j zcOjE8L(6Too^wb7hxBso+2(gok$^23!HbYtAt)@Yfy7zD!cwVe{M0n(eVCS0#~Qz8 zBdIca^+11x4-p2}j%FDGl?WobWDT%fv2rvk!~J#f)aS8Ic>U65v{&5!H-xp0Jzj%# z{&=MJs=KroP(0ci19ruNT1v_}C4y+6Tr z_WWanflE!PbB=@|4RA#>ACbgh{MLK6OV+-0z&X-i1)kt&aL9FD7X}QVaH|wqEZ~iW z>=Yy_tAyNH^v3Qlp7|R$;zyx6HQjm8clDxKd6rc7-Ipb-!$z)n)wbPUJr_S4RpkF6 z>&KtHH>y{Kr)|!;C-sDrMGI6ZX-!lQW58^ZVI_UaR?zO7Se>)l#;=ympk^|jQtE{; z3}2-tU*1tYdFX8=+Q&3c$I04fy40+i|Ni>BYcv`C2is|HMa#Y+GOOGIH*mtg_~YC+ zCBK9;#_}(79|Y))bcR26nTXLD+_vnHN4IbHFlfY1YkdeN z)M#s}i<3K#`!$4Yz>tHHq~10oovalBWzZJUK|5kG=Kmi(?m4RkF3b{Wdfh*l?e6-) z@vw)+OYqbwZ8#AiGn#z1j6Zn#`SGQU1wnPEQg2XgKrGw)N}uHmM-YbywGrI)wL2r0 zCKoQGF7lidZSK*5Att~Zn1ca9^~~h(8Bh=Ty51H?0|OyIuERgY-yc<-A2*~Ql0m3; z)Q+_+%5+6DFhZt+vfJVO1P zn4j_LqMrQKI&c5PR%(_R_A@|Yfcl~KJq+{^8T@xia39bjbF&L#E!@ z8y!u6AijXvIDtxNk;;87flq<8kZ7G@w@ urW{d@1>%?mnn?pQD?=d;N|eJMx;jy z(P|Ju{j2`HE1Oe!y8PvJEY6cOpUP2B@g zD=6g-ST+<`xLYjo=_gm|xzI{3{S1|BcjjWX zxMO-tsWieY?R`_Y4LUeuw6gr(z47{yd+{`RJC3768>o$xOY~+AtS=D01 z3I0}C2ck0g$(8fJAvZVZX!2diq1^{B2jMc1=yg@GrVJQ}m0h$+R5zhj2JuD!r1fw8 zUVLZ|di($g&YRu|YF&|)k(oX|+~`EvqbQ&^OUYmC-Ql;5=$=fPWuB;|m%aMDe~Nm0 zJUiK9P@e;S>wx(b9TZYqQs#8V?~dE-N{E}F*aL}WEW$Dauhs?L(OxU=o@RAQ?D^7< zMv7(O4vbm0cv_V?e*QzW9h~~qwAyi0v|kI>E3zXx8C!m}%MK0ic+6UCtm19cM%T1+ zx(Lg>VX1n$ThG?PcrlaaX%CTm4#p?Q2Jc9H)PN#pW+z-o>f#%goN+eVTz=2hCN+ON z-cf_Y!6Dy57p-~b==l%PmMdlFZn!dyk1Xb%3(H#m z!8l1vPyb!$XLuwfPJVtq;u=t?LX3dt=(FmwVEaFJ8!u+Kp#W)tf49A>y*BY1#z?FfAHm zbucCYHO%Ti;dx?}1&(A68qxdI)5Kh0$2?hK4AJHj#MZ=Wc-CYm6 z4;y;|RxirSF>v32haif_>~xLt1<-te@eUV$AmQYlLpNl(-3Nb2WQWt4DlbdFQlt2lFwIRzcrddMnEvW)xq=7PFNqyfYOi0V#cp3qEl;J7ltdkgDanu z>A3jSzcAe_(W;=dS>n3HiFXuDai^7{I+vx4`-w-nk>G`(4I@(0oagFkvKTz;)4r`7 z^!?c_TF+iE5tC%d_zNao-ujB=*=oSt!D+eiOruGH6flU#r3z}Zy#aXmZGK@Sw~4d} zaU;Ba_Hzcm{}VH{gqIa2D*b?#Zg$x8eEZSwZArJUuYzN;WsDxvpzj_U3UI}RBc;~x za9gd+0oX-APzkY38c23}GV2gsr>LXjhLmQxTa(D}J`2r@=`Q`Xai@6eMaVk2M!L0aWzB3!w-Db zoj~g}v5+k<2O<=(d-Ejfzxrs^=HT|YfwTHe)+a70q7RIkke=e+xMMyMyxkW70Pf2h zgWL$XFmca^kJ<9#&W;FR=id-3^SwVm0m4(5FW@wjVuLatekYJE^E>%@@PemaS*aER zt752}VvunHC3-S5GXwTSF%735gxxANYciE9E9Rs^DH4GZG-(30GXtwB!R)bJUZWw- zSz%hlfl0LR>KE*I}fd z06Q~eHiq+jUWmB8G(Z~NU8tCEoNwfjrn%>%v3JO2fXjb+X6LSzw6)j`e0-CNY!ka7 z^-NBlQG88c|=ZBMzQ zuv{3rooQvOPs-+b-3R$RH{>wyPkq%&c1{<4eB{16=zHc$qwmr$*=AaTEIdKx*0#7b z5AfO|A;-X~L`gjbTq<^-O7#F4N)J7+X_}egGU9n?W zy)YH;4sKdO>b#(+XemPw9-9LkVx%aB#n4eJPo5%8IuhfRmO7{vG(ZPFgFfP|&_m0& z0E_O!g4;yj6;(~^lB6FXo5v*WxCqf73T9P^42*aM~Q$5=(z2Gc}*N zn3p`{(`rCQpLfKkWL!wcU3jSJ*o%E;kBf3MQawWW$$$glygOeXVIqqYF=+XoIvZne z8tZLW5KR?{u+d+yQa>1CRHjHZBriyCmA^+r4|mRi<-odI*ri*9Sz)}=wDr~~E&d)j z?&n}BbC-V>Q;cLZQq2NbpC#o;3@@MYNpl(G!Tu)?8D-&Bb^ZE^2w5x8ltsou&%@7} zh~cUP^OUG%4Jk}utzZeCtMG>=#D4N^V3w>4eKa^_I6IjUDjKoG^53CRY_-n&ll3v545(}aN+UlMQD0z^i;AqZCrkohpl~~@CGrY z&h=sWn8tJ)N_5OD^5aZ^r|UPD-B_n-n=EXQZ4A!1MG5^XWGVX~5!B1ro2NiCYVB{`VZgqclup9|oiJ&u z+P`U8_QH{#{iq{`_gOHWT27kL6{pq4j4eslWyA(rP^9UnpOYZQ>nueP-Bv)`g3^eF zv*%}_;`x&%rj+1Bc5^{%bf>n5j!)Y9Cbzm%FDX!0{yYxL^Eb0`)}CJekbzy>1C7Yy~CBo&0k=Ti0K zNk7kot$IX-nu1vzOJVf4uR=Xm{5XF5TEvqTiMLm?>am**F|ARy>E>_leQ6O(c5#wh z`t2<8_oE%|PGE8u?{87P$xA^r*ZFx2_z1-PoHi4E;2$6a25V$l!Q4`NL1~B^hLN>R_pa5Jqthr>qCoj_i4q4x8U zfYx|A5!BJTPQF!G4?+wBij+kiEzH@G9lydkKWPB#cl7YWQ#|!j?0e^+tu56p&irTH zk>eBJpr6-PYUcm$hzhtPkQ^U{7o*Ms$wcW-U5P?QhntOiQ=uFvm=gYWPhjTeK%mtI z7(Ahtlo=*GDTy3KNf;qA9YJ=9n}Z{k_yc}4ZaASM9E%;W~A z7%_J1qE6FX`0}1kI&Y=}QG5^4=QV;fUbR>orq{J@0=%E*xdi$|ABexv$}-Uyz&M!Q zeg2eEqX}Z7I%%$)zw8+kSu8hzU+3hMWL&4Cn6CA{_)LNX4b|9-cACPyPJ8aEdn%Ur ze+ntMYJC*9X#>Ak9=t=e3mRAfc$?V@up6;YoI%@LQP%+x z@uW9+NKbgR{~Q_@cuk(H7EBym;2QLkGAnKn`044|$d<`w|3NlU5*ag*))@7g6labo z6X;L1AZ-ty@$)B7n!(GS0M7O4!>PZjNUQ=m5lpGPI}Hxuqbp@b3~aiAHAu54ga#H$RKEDCW@t@#d}tv+&bZhFC<>`iS#!e0WZq{E zbt(iGHOpWgbO`iNqn!Q*=@IIwj^**xw6xT=z0g~9lC&<~l?bfmRltobb~gSd*7c38 zgekK3asZZG71>>DX$6Y&t{C$Eb0O>v32)xlMiE5vbrjOl;b`9-aJgAn(x_h~!mXvq zb%~=T!b)2DZsp2vd$n3oKJ{OMZLESJ_S~ZeDK?P2VpnwAp^#_x=pQu^wUjHYSH5mn z2gtQ$$L)@VJlon8&?lbA!NJCE9iBOT1YTY2rjH?*eJDwEWaRBket%ed|8G~}JtU*~ zFh2GNUXiy-Q~lkE{kOa1Z`OHo+lH=D=CwZdFYGg=IA2JI0|4J+%XSkTw!26!3V>zM zJFpC+E))#9$odWvWYo|Bca`iWUFqxEo$KqsjcBQMP5a+< z@$Lj0L8Spzo|E?)Jw;F$zl0yQ83gGPJEPY+Q9}T!Rdx_0h6J(T^>IDbONxhsati+E z-UUl`t;0D_&zJw)?_L*-kErurw1Z}QA3BqlAU8jP7F5echw$CRQfJ4hs~5>HNjj#| z%$-#d!cND#P_7c)Dg851^ri8s`r}hwR7=7-V)e_yyh4I44;+}S611}Tao8zHUocpm z)({XEEP$c#X%Bw;;NdE40vck@nxpDZONlINR=+6y_72P%8a`xgyCdeUxGz0t$=`$v zZaL~c;CcSu|8E}WXs^#p_o%3c^Gvt3)aTQ}m!h*pdHZ=?b;ZwA{}M59AQ@b{CMw!N zks)~f`HD(BtY;^2vD2ZRHCRznU%ToR;_|WOW1*nLLEZbsQ___csoyCvjCb%7u?%cA z)S9j{ZZcgJGzc$~2UGV4KCb=a(rzhEJOE0npQR>a8ZvhOrhd2noU&@lt-OEL?P0(w z+2YwQvTCaOWn6n8_DEUGlOO?P8o)^~b7;c}d2`u`5(T{gL>b%BaTeXskO5Uzf7#nx z8hP`P_d7GQ1o52FWRb3Od3ls4Yb1eZeHr}Q3#nPGH%>tHy6?p~rUn>t%eA_pKy1PaEYkPC`)(m%dHTn*7adeOtSQF5sV)#D z8WN&%^X5$u5F|?sivQb;{+mOYDj`9_3h3_T8doNe%Am4^!W#D0t@nc-hX0Xe6ZG)6 zn+*WM&#R;|nVpNKnHQ^9l{|YUSn!RG@{<;A-B=>6d%U2yu`nqV9HuXUk!_UZ+g$j6 z;62_qO>2B(YcUC-o#QECYk`=y3>(pk0I3QWqr9oy$#bbXYf5@4*8@P)Tlce^fOhK^ zwUb{7M}yOMo{7?elAr$U-EP+p7S2Ud4aVsNmtJ%`%~(xbTRHku{!%yq!o)9?+9|6jNfoB;kY0 z{tegc!i;gAv7>KO`jQb-2d4Hci5N5+G(a)&xPv%iYz3Y`o~ zTDgzO^N;B3C$th0;qXo;P?#Oa*9sjB?3$evU-3!h5k+x#h1sLV8T&s@#oKw z+U7Bdlt(p7JW(hi4QaD8L@gqa&hXEgkf2}_N(c3v-Y9;kRTBx<-Rm_Q<0uRxaZe0hn`7m9)9@b)TZ*=hG>XJi4<4H_bzF!FP15 z5hQTW&-$o&H%R~Byh+5tWS`6{G*fu!c#akewW6^i5UjfWI|p}rc;kmlmFkqjjI zuTX*n7N@E<|Gq>4F>V4WDjJ!ex<`%vxA*yCWpr1gvwIJ=UVbnqwB9xqcb*IhZ29Ap z+tog>XBFATTYK@em9UKm?wwY`RvFGFs%h3BPAmeV99B_hv7QJ1IFJcT2YI_5EWc6v2h!J*3-5yn|(LRF#j$lZ`2 z#Q(p&Da4_}s|d_Ho4p74a1bSn9Lh5uQ6iR5Zx)oDP6A_*+hh*nA&~liw&FC9LukAN zL2%vLvv?}4&`2Wek^3Fosg9f|1!OE{QUbippKJi!m5k^G<|xrsY0?j9CGQJoZ^M7| z%)CDPh1$&?B9#IYy63jqe7jap@i%bmp^%6u;Qa{EN}{WGemg23@6Q2lPfu1rk&0 z=n%38EpZ`v^q2Ry#QoMKUtSMoi6?nf=~C9?IQ2=UPAhfQyoz;6^~gpi0@sI&=VSTJ zViYZDc2R;F_DquDh8w}*U}%xh)X*U(P&1b;cV~Fux>Gchy72Z{X&{A0j{||tKV<%O zH_JFKwR~rv%W?k4pdvet>vWygFLU`J)d61?gV@S_g33!uuUL2$ITE#069&Gilgb2n zx2`T*KEbqb_oO}M?qBuH9oY^&z1MFr_4xk)^VoQ#XGq;6!dtKBZ04$V8AS6dO&jO^ zIwlI!yC5MhfEN%YFCB}yq=F$KaT~8Y+a>AdW!>81cv$c%Z*ElO#`bW@&=jedfdrE= z8#Qq5(MW%aW=F%3k35j!hei+Bd;(${rv9zl|I5qYn7ju$QYE1L%A5Q6{YwUB#{{Vt z(QSduPM)3Ypvgo`G~5=pkU^2`m<@Ir)PXqA1qJbe)+-VsL~Os+`Mf;7&z)E26Ch&{^{QZNtcmMhm*{{$fs*oUVYxCq zJMl`784YBDGjW{D&CT5cCU6#*z;>a>c@c0fra@;E_Vg)}{@H-WNsoxK>T2nkAc#E3 zD};NktLt4{FS0@Y4~#$2{((%e1GHwb9rS0esP4@skX=n=FS7f=d-|#H=ZdQ4NJ=5A zD&-L>d5$l24s~kftaS^~&dpeQsYQas^Q85P}T7nk(unhYV z^2n4w6eb;hOjQ$}!=1wY1ms-@ck17>WynkK%jZvShrVsJ{P=S$=&`L0AYA}skSB$$ z>ok$VFw2E!&>A(msb(w88jqVZyTl-tY;ji-z>OGD+#;za2*SLmL2QUa7sjTaGyd{6 z+576Rt5yrl%?hCR1Ic>REgu{MwG5_n@x-2AogdHCjJvl&eEv2Y|KL4j_Rf>1qTx4C zb8!(t_lBt{uJvfi2-;-?MZXfF`O!l|tVek{dH`{)Qegq$KYi)YO%2PgRM&&RW7L+X z;EjkN&?^7sh+c4E|tufUmRz9%<5D5{6lD@2O;nUrXxb z2xWzh+_6{o^8F{9!8a3$V%@-wkOgtaFc1%&D-Kd-hqSI@7|Xy5Pc#ZkIpGpS`R35o zu2I)$1(%(Fn6y0fxCHRD6C(m#Q8F^aw&P;B4ZE+3OE0&~6B)6}H{-|HZ^kWWnN>~j zlV``XvQD4#vk_(Z*Wty(gB)J-$#d8{gtk#(>)Yyk0k#K!!ltn($oQerxV(Mf^04`_<2zs}gM*c4 zI~{_Py2+|JniyMQ#)SktN|XlhU)T(|V54l5bpLNe5y}~Yvw?K8tmHZ$4(T*(Agc0m zl>-r4=vYfPfHe)5dY}N_>l)XU@_l+vz6mm!s+Y}u&r%@ z8lD`vsp&3Rg@dH+g{7UWDwrS8$ARrW8z(_}AWWExmvpo^3nMxCnIjK0Cn59~geQVr zpD36>0)f>1cIelrS#@T;)8;u~$nYbVQT4bja`Bp5+k4d;>eh{Cri#AP>eF2R+Lm&` zT%)*3Iff@HtQ()yG%UABQjTghm)0yCTu9(WDlRWnU*dG)#JZ(VqIlC~#^q-8@{SrK zK@&c048h(&k)WVRWJx9GUiIOl2871;pgl!N zePZwS?GT41L0ZzeM#43{{d*HBO%NsmzO_gcc^xZ>VtPP}O?53~hYnW8-UBLyvY+*0 zDi?+uXT?91^Oy_j5>?RClkx<#+yCFr0y&hm3HeG#JU1=Cb_*kWJTwygYik`!HBiw& zQKI}L3-Axd>Ld6GEl@e7nbSa)ha_;ja%<%Fr18ZkN8110HXG8rpI6?(LuC_E1*1lR z=zj6`C-K$51v-#D(1H1}8MWM8n$b}-j6gL3%Fkl{{Bk&wM{Nc*&TMRe9|$gA%trO| z=cGu116Xr9C%cVqYLxUxWg76~^FxB6Cz53XeyFXlPgCPM`qKUq)+0?mgM#Und0Z>w z&*3T@x~UO+G1eW|9fa;u_cQS?r7}LvV6VH~0h=gjwI4~Te?#fobKx^P5i4f}$7p=$ zlA-79!M7N4P1arax+<%%A^$i` z^h$HQ@HidJAd-GdJoY5AG+O?4-z22}CZ>Fl(;Rx4>eY5%RDW{Z|M^G8%UVK$!NoZ(tFoDWwaXwJH9BlIz2qV1?%19Lct?eSl#sN>_Mb*E^V7kd18;Y_5;?CC7XZC~ z+}9vBGQ<4+`E5{4&o#a3c7KrmwDDFY2e%#;qfOQ>qcPc!hu%t&Y2H*y&JRv3=MG&` zjS*Uayo2)+X7UR!L2Uv`mh0FTACnk-2$iaHp@bmRyjr)rgfJL1Vagc@Q13`&iiC|~C@Ka#JbZMH@nqK?&V%Q2JOQ5}4@LyWji}X%O1>K#ee>gZa$S429&aN#)=b%PBC)^0uL^ zQkWL=fg6xUs+E%m+d@&WcX5W>)H}9^K>u9fd$IE42TGRBS^&AgO`~?#PBwp5U>qZW zGA>^bZ~R4#lKe_v_aSHJ`cKfJ9O{LHLT*=$qSg<-E{#TT{-!%Pn_jzx>xJ&>d?=6RDl zeJr>aNv|8dfSCbT z0Q=NkEql+OQGx0v&#C*z*uuVRsBPZ3$4u$dZS24l|1>N*pNC2!7zTH7()Z$DY@))Z zaiGicR4l5uPIjFY6s;!_Xgzy+a7{P`Bh9W(W*==WoxM<=b{BaIouxdir7bQ*KK7G$ zS87v(u?1_FZi{A@lfgY}EAIWnN*4b6;suFcand4aUIfLBd7X7ATHI~p{x~hAAC`7` zDO-t`zxh%9%mr-7`(sgMEW^f5!de`_#SUnSsK}1Ttx}c{C@RCVKUts#JGkht`&oNa z(eq!~h#6}>H}GZI+7R#BGs3KZUJ2YAdoqRJMlq3uO}uL6Oent&*2FQMo($jtgbrO> z0>s3`61FXPXth}_IOtC@D-aU{;Bm>w_-hvic35N;HJ%N4GPt~%QI`gFVn3|m{>gR3 zC{LOx z7FR(FoVqVL83;}v1Hs+Zd&qDv*~6=MZrszmkW-Jk%qSn_-cAE+(EQw@+W7M2FLVCD zboGPP@$K*LxyUwx+;pjoX&7nv6^>wcQwz`qp5TSFoh{Xz9%cJ-O{DUBE4lLDn>-`O zZ|F|BL6%`3#vI>$2x90np4sVcX6QvMJU!{0xZDabkmvP%ihEgNUn>|sljfp;^MWl>TJ^pgU#SueSK^u#h2rZ9qwp5`6 z^fis2WCS6xpGy1mw%@qj;+NQ7o|e#vzbwI5KK!#hu^3%;l;&fb$kwamTCce$)r_uo#+j=lW5cM#zn5aL$u zJej0IcCBd4l&l}equwkp;L{dBzl&Q?#AcI});T5lu{rk-J;8Hzg6qIGQN-Y!O&2<^ z1;LPaPy`0-4rpI{GnE7*7q$I~B-7UA-(tp(E)pc!uRcOy0TtE_7rvUqNC?nMQPu+3 zZ8`xB$aRl(_U@A%ZL_`u73b00e-MofTMv>fV5ey&Kuo-i`D4}BJAGS@P+sh%uCFvG z(gQgHs*g~;h3b6a>xf$G+?EZ*rJNYeVHX@y885*(XV(IEck=+m2cPnO;fDU-diARPjt-#odW=i7U~?>~Eub&cCG){RTAIOm+l zJc2BbS}l>d0CQJED40+6w}g1_gZnYv)m}@#SSYH=j8kGw3A=MH(2(#1y?u@IA>kq2 z%d8AO(UJBl2CB^#D-lZUB0!GfbZ)dfzZ^!PL_{Q6NHw>4fi^8DOnK@-otQy%h}frp zIl;-4N9A9VN1awzhe;@m&B?=z!d{g^W$#UM3OXLZpJ=^I`^;zwVL@bJz+(-lD zT?6>fyf7!imTn3<^HA{n=faJZ+EsPt70z3zq4Muo2DEHKL1&^lIeFyOZmnhJ-!AK_ zxTr_UwB4ja?t@15vv^t+b|J%-p|4NBy!h2|BBrPrRInk?ypy5|Q!CGI11u@nT%n~i zoI}P4jdkpZyQu-=FiAK% z_S1*=hl{WuxlI6zJ852I!i5^9jQ+7?DQ`C|G4&O7b#}}w@=~7xCWTw))``0qxJ~qr z<-dR53WPzE1fHjXNH?0BbR(LXVVK<+mL$KS!OPrngAGH`=e_Thna6XSGD=*SfItu# zfgTJ|$w6Ob1RGEf;ClfBV9*Qtc<-sBvj}$t{h5L%9v6RsFMMJnuWi6>C;^XeDfje4Y`pZN9$(Ea@$9wTEb?;X~6)M67lPx2_y$4E=; zw{*BdkXnQ+B}Yu}5@=RHsOVIcCC2=@+x(mm*iX&| zr0MJQS#g?zf#my})-MQ3Q|n74j7wg#zvz5Kch05Dl&i}8=@t2r@qt&5IogB;uN+SZ zmsU8f?p03ig`?iVu2+!fxp&S4ALW@x3p@!MQ<^SKo#x9MZG~S|ei{ef+ffj>otCUl zPoMUby1AQ*M(kPkS4N7hu0h39XFp@}PI9;$e$QsYVjRQ|q?Vw?g^vZ8D3GDk94_dY zC@QGhI8`?Ex-TQRb+3hP>AGzP*=o^0s;=}SBLNwqyvSEEVnpC))Bw9nq{B4OTjX@N#{SQDv9?P?+CQp84w zf9=LBlyCU&2-#O@klgJApdSchantEM>7BugOXVl~2Ab&L88Mdn02p04pP(l|X|9V} z+(;h3BoX_wlfauWteg7wE!4SN3?KM0{FxIDO{CDB>3`)vUJ$7Z82bgs7yt^$$mz46 z3sUXq+1s|yMP?R_9H$3%4IuVf^84V%P8;F^W zVe^6F5GQ@MRnx$lp+P{N5>Z)^(G(wo+mi+H5$38g@cK0YWe~;+Jg1m8pf1mzdwUOr z%XQ?p6fJGzfkZ50==GWM-){?QxAo0k@!Q|8=$PP-L-l;REArf_-RTM#^f%E=e|r2W zT!~6cJEY@RVAD+S_{VtJ7--1H#si~MbbqJl8F_tfh?|qU?kIZajyMgjS*`;OjSG|I zS1GUkN=%qfqvc*QEnvr6}soV~*Z%{ZD0tM0Bc2A;mjN~1)WD75cj z`v5S4rPLXSI%`5618D03T@HD<&1hiFS~YPzhw9VOJs|O&%-FCOt4B^6U?9Z=4)vWw z*~l3PUOECP1_V}NJxC}4nEI0AmrLkLBd&%K{LYSbHNXf&EcoAb|BmV^Gi3l|xCFSg z<){1;D@1S(i+QlN>EVFz*MG-#|M1@rGws*F6nCu_HmW}WYIu5J1Lh!BFQxJQVWOBu zlU%1$_sLoWslfm>cdR?A`*%zZY^2j*wK$a21CZ4QEK)GEnos11T|u!|*8qfk2XOD6 z%?>*Yv3$0jUceV@K-@x*ShAe+*}y{nTpp9uO9DPgU07Fhu++BU84 z&w^Q6J#+z2A=CIQ@+SiZEummUB{K}fCw5FQ9E`giso#Czddf$dv}VuYo#8+z-NVOU zI3JcZ&3=s7H9h@0_DXYA6UNp zEx+b6_^nZM>eQ=J>c(CyJGpx_mCDL@L~ZDS;U2qn*^rii&^G>(;>dj9#gMdz)VBQY zI}0i8uP2#_omKRQRb-2jrTR1V1-)bV-?IxHwNGDx(zCX7wXSOL2{{h3C-$4*o9!Ez zpg1@it18l|@P%6QMuL#~y)-K^zXiRtd}VGT<`7g)HC^T7UdIXz1o!M@bxj3Az?=`3 z{nz+Q_}q~Dc>;(5$XO9Fou4SQZjJ3$jn6BK`5b0>T#fHqSW{dxb6d!5HGuG{R#>n- zvm-qi>lh;+N5r{WfaN8aZlkUcGE#mNfkzMuu!G*=>aV0j zYB>UC5Fx@;VbW;Kf7wO!8xFFco(*nmyc_~L1zHKy1A&Lyh+LbO} z-)V!t6#6Fc4DOMhGaU^sIQ^bYmWORNB`zL6gNf93GH}3pqRK0N z%`UWnHs?|2bI&Lq)(cx#C%OgXR|PI8{3+-0=w%J#_iwK{mf|~ocJS8&`~ec9ow;?F zqt0jk@3G_F$vV@_v)6Wivlbe*3B}{Q@rK%|+Wx|Rcx-&-M8>mI^*vH=gx#Bh12%uM z9*}X88PvtNQQf9YZl%H7OCB?pYo&KH>Sp5=?`M*jNbIvD>-9zt3Dj0##xLtMve72!(~~IIZaM=Jb75zQA%6AVG?X|ndn7M z@Jd9l0w=2%oq_JU*n#vlVlpr&53>VYo4g8xXQBxV|5bmCT|Drn=Tl)m#Sz*3q!tptfE(_9 z>J&%<9ANQb0>CGV1yDKI)GA+bszMfsvjCWXL4RIIi9AW1;+E-$b_cD1HgF_K#ZhCZ8$KqN5&6RwBrwh&Gp+)kY)>JUh*mM4W1 zlpxPhF_4y5EcS3G3wKeL)qMXBB$4M_(t0Y1_y4j#n+j&HRQiMYiM|>vOJ6fTE23f* zxZD`|H=4%g{x!>mdrHo2m)SgSe-1WKgOQuCz|7>IYGbG9&Z-|bPJM8e?8>DjeWzT@~AX=MmbEZ1c6uo;_^t6#~IHuOma4sj2IZ6#UHb z9WgkBSB7vCe+IUb{w9?HX1M=#|D){fQaeDsa^VzQlVpuZ4z538G;BhkQS-G z-L^7ikY*(xrh>J@cA*QudQB1Jv@Cx)9E<~%{0o$lggmLh&rDh0bN;2{+h9r@id75P zPnSGuCPMZ_X!@<0gf)R9(6na@-WH+w^6%89?|vGfR~Gn{o8KS5pFtL+FkKj>H2hQs z>l>@@t>9$@=k;^KrR8J*)a6kKHi8QMK&=fWIPZ_|g9JeG{+x!+F|RO=A7-2;q-KTY z_i{FmnLQ$NBe*YGYHuX+%bZgc60qq%IYzrTCEYQ!kVDQ2;Ibmd5Ror~>}=5Fh6D=T z(nmIx`bCj;598R7Sl9y^21nGcL*t8x=HNfP{bqL(EWnVTBZ|U8FgYC4RIhHvlxrVx zuFAp=4ppXc&&BkUmmDqYa6?DIIs1#1`Y*mpA9erR8hC z@v_U2MJF7n&WX~+gD{&}rD#BkJ|uqy*6j3#SA*LF?pO)o;XN40sxpYx_9lrXq!b$& zNu?0f3gnCGn9vRwlm5kAY}~#zSsa`=VIX+X{tlIe+r`OxYl?1M`oNsdEG47zo>ZEv zBznfPy#$TddVK|w!p~&Qiv$izci<&FtKOC5d?&`bGya?gvSg2Xs#cy(Nf$joW3V+n zH~ZA(w+gfbbocK((3t*4LqI@}Ap`z!J1M~_zkx9S$O69(eVoYO7nNgy za&nmVpvb!vhxroAQhYfzQ@v8uguZ_8&6UZ;`G?xKlQ{*P!fd#1@NjW8AZJx{*g#9D zbm$FCM6^0623b*L1V68kY7-)edhJ>>yAIw}deQ?aTEGm20cIeJ#^iK8=<+sD<%QqC zxTQJiIbRD}4I_EOZ#cZdkop16BnQzPTxUSSPm+OvvNelFD6l`SVb>tcc?KB=*eI|L zUNB;?lcWo2ez~+$F!DjZmFcbCW}*n=#!w3x&W;EByALt*u?#7Z>F&YPGnBjob9`OB zNpA@zUp>vKUv_F-b^+=(l}RWJ&0y*vzVdTA|83ahBv$qc#dt6qkHZ<-Y$F{8efDNg zIul*?8nMOcaT3FA9Bj2net<pH+}Eg$E^*K0)IuCL?Y`^j>Zm1H-Ds^p&rcY~hCd-g{b`2)_OLAouv`4>7H&M-n-o02LBpE*3b$gy4`sZi==KZK z_i39pUX|Eht+o42wfn1vVw&ZSP-sLvS^fL9KeTA5fqb9m^zq@(C{BZS0n zj{r#DZYj6mKC4Dt}zjKUkl=lWCYgl`O+1NilzPgtsu9Kk23&nFQVuxyY) zGX_ZVAdHfTm>7j!qm$?N0yDKE>5{z z`al?l7s*s;{4>DrMWVGB*nz8$ZHy)wlJICdG>n>wjoG`W=(*U8&NO#V_^rO>uTQ&D zNhU$B>`3{aB5Lj(We7#@XG2&{bo$GP*3&xCz63$t0O8R5QwQTtlQ-9k^}Q05 zo(PLq*x_rMl1NCGQeQa?tjne>T&eZXsk|$;w&W8wAW~gpenYM(*?@qo0e7zV`a9#L zY$cZL9$sUmGx~z06JgcwWkRfe^B>kP_hsVx5d8km+RiO>lb;?tDQ&DFtGirRNZURb zn=JAU=Yq?W!WyNdC}neWql?Ub*0*s|YP^=WdrX6L2`Nm++FFWi&sbenMcnIyqaB^r zocS)W9d0$CX3)u4>)Ey=`Oju)0*A=wFDqDCVjAtE0Z&EJhZ{bGV))Vx<%P{_E1ZuL&TB-Ir#8{9rsIlsyf zcls>|cYXwO56peN^xu*fRAObD=4Dw}94eI`SueSeC_b+td74(Ef7iWy_%9s=$xC^T zta>pa&ZGZp59cK22*#l^33_%7fFH%j>xBOW|Dmx8`&Mdo$pBy%zsk?&TJxGZ{8r%r zVD&WgMZsggFrTDlNgV#`_SU>IB-iR6-(6z#Fezr46n{7IiSN()``SYyd={5GXD?nL z$dS4JUm!Y|0-cCN!BwKMF@$&%v>$=nBVQ^z<{QCuAO=qSABap#b%s)b8IruCj@D|( z(HZxBOpo<89?b(Io%~2_PH}$6PsC5-46+-a$UJW4eCTmUT%jRkap7L<@^11-b<;DJ znqx~F0b?h;ffqBIhYe|0XjmNS^_xQHg)o2pXZG=L#~t~cqB9Oab(4%VrGr$dB+Nsb zjjK`N%lSvws#VloNyRH{u;a~0BpPl6cD8Vo`RyIlQ<&a>5ILnrWWn6DOmg)a2}`K4 z+(!865LpAK5v+Ib7rXiXycs)=kit!t!@*4!PR(ivcz!of#i^S)Jodbx@2yy|*od|D z3%JQ9xm?#1GUBQFKnP8}`$z@{r==}kuvDFrrx7QReN0*9)uHmm(tkaI9Wj z5NqZp@7{<;Y}G4gL6Rl<*_2axFAfT>e10hu^TqCm3dD1+_|cF1Ur!W`mks0qf0K?6 zaqE1)LMYyBYrh(Ohi}0lG=yb`w{DfE5iWP1;c1$DgOfgypI+CKhWa~Vrc^{}qr!p^ z2l-uc7siE#IV?WHShLXQuWtN6jc}L9RAggg_31uksNN$t-e>0w~+;}MyOH5jSfa_8NGNqv9xo5Qb zqZuaMiv=q#gTih+;Hcqux#jj%MNiD9+xemkKS{KKx6ep`*Lv z{ZI-rl_WkaLG#Xf5CQg%^pUl51l@A;XTtkWOaUxbyLe_2|993Pw7I1FcU<5?iE*(CvjHKO>wJ1>G9>AH-n|31BdTcSa|CCRTMR{MYRYB zAp6U2fIL3chyq9&QCK(EfnJxt7pr}QW`1i%zOdUQ9@O33x2>qaWe#z*(8 z5|gXr*8ETP%^a*0#l-CsdL<-yl^w|J|5wFF5h))>pJqqfde-|gf=*O2r9bbJQU;xQ z`*jHRU{RbI9>_E+qGl4gSh%n0>%$UDYyK&!@&q~m33s70e-Bg~rFf1R$iHeZaGwZ4 zfH0b;AYmiu%`S4r_JNJtZ?JZbeo$&|jMOxM+pv3u(*qmDEYTno!PTkPY#Re{i8p{d zkX=~HvLb?CZ1t9$b{-h)5YM2)34GH8CDz%GqM`2@not@Yqh$_|E*MU=d=xV+-YGibk1@Ncy?v9`QPA>RF6+#D5Q zsTbQf(s;j{+gRj@G(CP+zk6BEQCWm$R>8wlb-b?0JUvaYv8H75FHp1*#-5neSMQyF_V$q*9A>6 z+4W*L%|!y11rn)uvD*Ha&0qrV*-L%@9K~08PUSS)OK1;>PW-81L(Lz_oku~4cE8bx zl!n4h9{xML4|s9X?*Bz`o|HMs>pK7Ae-D2fEE5tiE<|Jsv?qf7`DtWFg>09B*s$RR zInHQ$Iw32A)o26IU=b#aY|j5u@T(-lO{cqh19)cexHiG?0wSqwNAzs7Fh$t^4&G5y zw}|n%ejsjE^mTvsTrMZWhTgQiY%K1EULmaB}ry^BqQJ{}&qrk<@NoQQszAsxrCwNI4CBq2Uvm5SLv! z@W%p$<8&GX7{U+f$_D?XhCq)MF&E_|P(OW5BJPNBWxsgz1!aVyG)cf} zKs3975<6gsxp|)Adg}W#wv*zPABb_gHE61ov)hsjvse`@<*s(9TxndizX^kS#&sSX zm0JQqO}Wzuyocy)e;6~Gu6dF-`hkc501!0tpvQi=yE0oO0-YpUydY-Yo4c>xsQwSW zf7;CKw&&<7TIybU}Qoi2fZyB^4~{(jHMPmEr?_aG%8pSqKW`x z78&fsS%CZyh8Pp@lF{joN&WT-I=52caaUNsc~I#jE+Sp)Mbr1zPJ%`sY3eaIuM*#i zbEK+xDw=^-%CV+&WMVsa&>jKi^m{Qug&&v;UfD@(>hJMkW1T2i$Q$SUrzs?LtkoPX z?-4V^PeIo z)F^&k^)r~Q{j;4jsJF3-D?Q`ehk50Db)7sBEAj#r(Z&w8jBylGZgPy4R&6P?7B;!xtZ(1U{I#4#T>G_8Og zBM|~hYJmj0Pm}*kDL%>j1`gMW=ndkAK_hTUN#V;zl)Vo&N+M~UZ9q}oP}6}!=x)4k zNPSc=c1Ne4&+!wrCTGpvX*J%yD@qM6UJsqfFzqFeW5~zc4(}Sgl^@26@0e~3fW}}3 z*5^q58+S#}jL_xFDfc!SOoV?ccGC*(a8&PguZ<*j#gKb*f$yx!m#9|_<38QLoSia3 zbvKe!T*{n=|A&fG*D5JYwjBbf*N-gQjIK0dtUQz=@wPc<#;A@dlEa5P7fW6FNTZtb}^d?y#kiX znZJ*_sO;nvKGn_NO=+*s`pT&FAjtoE{l-Nj9QejhpEex8&Rh;@q_O;fo$|9SN%Ivi zp@G2d;LPy;evLn?7BWpeetUvw4ICr2M?dMc!aF!DQg2b-Fg6+5LdIZMh@)5}AUY)X z=gHohq;$dI>c4+0?*!wcSg0uap580adHYdkzk0fehW75Y5Q=w&yDR-Qg@5027!s^gXYoiGzw9gzB*S7ED46kP$^@*z6pzT;o-@Kf+a5VQapPQ+1C$3$KU_d>cWIz?0CoP6H`Zf2&yjMYQ zjeJs<87FPAnR3T#g@KpWhudy|U91*co4Xm+4&mJ=4pGKSw-b1eUIO1#&# zs&jLQkGJFJRHnPQY=QCKHPv;U*c=(%!yYZsQumKS!@UH_#&YKL4-eSJwQf5nC@lw&WkkYpkX7l=7$VdF~*Foo&% z0+F=ddWoMVO9-wwz8}|@2K)s|!itB&g7GoSHor@XCXIh_I!aH)X4Ou%?g~jxvC({; z4W6xShXz4xcTjWxr?N{EpT^?H!#g1=B4=A8njL002e&o2hvM{yOBz|Lg-?)fRBfp+x4B-k$+9StOVn)kJ@ zX{CD+>RtK6BAoT>lfg$e5&p;tvEYmA`4n=iQChdRDPpTPdR!;IX$xjEH74uUuy$gO zW%;BR@w8VcINJP5*l8=tx43v0X*+xNydf3v`hZIrbA`~baC-R5n_o^3yDL>X>h~zl z#P9f~nNpFmj1?0m&rJMTWzgq7PH+A2PU`R&oU&e`%T%Bpve6Jj+XFG2)7U<4b`7?a z8!>s(1moYOX9zTQ*>Ws1IG3p+&TsunTZqGLnulB!pXc@V{M9GPKvDy)q)LA5Cr+C#m2z`2oyWfPSJ3ym2>ui1UU|9{D>k`|k9C>68GF+!|+!nf0j(X=rIty+POJ zYH5!XAaol*VJZI1A^U5LfARsf$<*Czvt&>2NRC@;!z0JvmJqFbdCS9ROEFEoF1*yW z=Go%1M{P7FL~N&b6SWFJ?lnpLRVL2k?NFO|zio$w(|}Ykdh63pgu^*f(x&p_i=Om* z8Rx-A$$g%;3rl6;=+Dae zc^IYgcIn~#_^yNha(#eeO4h)Jhp>WDD8$&3+U~uL>xtQ$sG2@Anyb;9~`aM)iFCz*`|Ph1~sSJ>wTnA`pNC_oV-h z&us51;gR)YnO-yOmC+135ymF>bgh12-?)xqm3I!(=Aty=u9fN^*`{8) zbSz|<`RJvs5`CWwe60>A?&QASuR8ewy!t5b|P z6S=o8Ybh@%&ASXcj~ejLA$JNCuv+5lE$sL{TE~~gD@Zm&Dz>rBsPUd`>H1sMsf#=2 zFO66@Wo_R*ZSfG!aO!j6vxNv|`VJsvfxFSl2V!@{srb=y#7Bl{F&yqVWlXDH)PQR` zEWg4a5Q{@C?Gl|20c*9P#L!>w&$YkPf+;4{PA8vy%Ys6xud=fI!7(Kb=68heeSp&& zL^`rUj9t+XGr@#nb&z-&o~kLBU!=Kxx1tYa#m`pY0b1-LjP&zJzl?N1z*qwU*(Aog zM&4ER#gjTZvpQd!7M2%bMR^J5WDGt#u3Ss}nAxIJ$FFp`$ibIos)AnBfe%GFL6v8J zc>MEzjqBgh>*n@imRU4vlF`&rKiK2)UEXGj7r$e|GZ&3ar?V>USGeL*&0dAc-d~_F zCr(b7$I`>3@KQ??MO<}hEGjU+cJAUS(t$EQRpOp)OI`IziGlLcF|?xJdLvN_a&d(+vb}M>1tZnVyJIiFgBeUlg=?FC_ial z{(9XUYWLUsl@K1j;hPB?_yh>CLh$~hKKiJ3TOqb`lb_cPvO%;HTE1GAVl>p3`4|)r{hj_U(AVWm*aWKc&-W#*85VHeDF@kO2jQs$+ z%6({JPEKlj_5c-Brgy|rFra}4h1E1zm*Z7TxG_OlH<@V+@qw{@)KQrWbYxwypaLQp z2@G}SKoA8^Wf6M z>}5vtwwA9QE12T`HBD~XPa1#Kq(aAfCG_IlRR+8poC$SxgqFU)OQ{-|Vow?k%rGDq zx;&sT;QAudTCvLcrD^VU&jPdCPO43OnnQdwCP%SSeLQaiDOlpP6sHG$;oH(Ledf(!Y)vh-4TgsQK`CTFjRqdaYNZF;0Zre+d~t3obR z!%7-VQ!2JL#0h8Bcvpz7+2{r3rD+MP+&0;%J#vCoro;Tp`w-mOTewa8IaLF>G=K2e ze1xKrAc|sap^{onGyr|C>KQ7c%xjvh7}RO82411J5JHDCYjp(13LXz4^d*4IJq5LW zAP3&clR4%xnBX=8IXW`s9QRr&1i4KFfb&~uMl%?mH+pGl2G#^98V|YbNkX0(dWjlR zgxMZz4qen&FZrk|KhPM-v`>q0VZ;}1oN-3J?NrB6ko8+=@RV&5WomTg}M3X_sKGV!6L_J18 z`9Y4C=Q^Ih@xW;q7N22jBBl+9CHN5iW4Yn`TYKPIxq)d7cz9 zg!D;-r20Vvh8|{wY|vIREGE&Zh*pE?CLFQ(r!`nLxz2GiHbu2;q?vLDC)+96C2^RG zlJiidz1@=-i~ICGj8Bii=4|OQn|wXyn}pI(C)r2QzVUZMuWiScQZ7*1AC?qZ=%#nD z(z6D)rSFtihn?->BO$ps&_CYbn9_Kdg8As3eNW~$+b9=&CwVw4!nR&hLBj8gUKph8 zP7e3$%U?7Ng?7bfKHfo-x660_)r+9T>Mv+Y86z=fayze^ViEXWilNO0olAdq@?_@G zn~7MS@Oj9f3Q=w=2uL&H%j1!sO^dIJn3Uy$O*}3NaRa7hZca{4mm}y3^WMFSgh(m~ znW?X5zE3T;BdCn~f$xjq%CjCNg1%pKHvZ^)IAk#x9)$pf14y=N0UOgc7*g8-XVx`% z@CVX}5hcYL8clv!&!S!j`k6WZ1j_+nt33GR;aHT{r+h6oc>v&gk&Ii~O)6{}zXr4m z<|~R;l4khs8RAS3+-R;Gbi7zc{r-Wb#FpR32W`@$r_P3dRpX_{nv=d*tSu<)RLzWe z=Hja5%6s+d`^AL;eJ?4}Xlc~w20l(!o)bs zlyLHPvAP<^Cd?)*RAZ=1w!(F$53eYR1kC1&hq4{xs0k?1kaj@-Z>;jldo0BL@@_Tm z%#Sx97EgmO{)59zqrSQ#PMQYM)cygCfGzOGc^)q3IZk{_r-vqTYpxy1q?jSD5^8m8 z;mPrF`ojLy^^+sTNnt?f!gB)9ikp3X!38~FC}4KLqI3d(^U_%`wD^Fp0wBK!phIF{ zNl}8`ulyi9+EiWqx73qI9nNVsx1Ht>V#&m;E$Pdn(&-wCojf05+B3@7)+<(?9KJTD z=y_?buKZQIW8#(ekcuU3_v%e;!7(Yqv;KKgLLv|7f+=!xPuaXyB-S7m6bbx8m@QL# zq{`bBU=S+ktL3kol3E>Mm3oOLdhxYx%Fz{~TON*ahNle#iw9;@4Aw`9`JD||HNz*e zUp#^!M@g7aU|5=4>m#*ZUh_ z-Iaz8ZupCDfKz1|{XUp_ZNsIhuips}X%*OyGDvy3{3A7wsPDiN@V=CVk&H1F4_=nA zD_4!sAVR00ap#>&6f2z>!F)%Gx3>&>W3i6|&4+HK;{S8f5*BQY2>dn{bIOMOG#@T& zFhQJANRO`l>m&KM3k!p4|2T*EoY^hkdlr8i)NjLy)0|PkCGr_^o9{heE>uCvG**1~ z5|!QQR@EO|$;>qV8N8d^8S$aR(3@ z)&=D0uk4BXjR@vn1g~deUWLc^FsIGj%N_0!aWb-3`$~}NmB+q&+oaBRHzN#VrQ=CS z6iUW!58HbO(M9}7!NN?Ro~C^XOVuJP%j#%5dPoV6D^KDUw+r;804Yqj9oN2TVlb7{mJ)~`w>s47rTOwgeMCwf*T~oM z_P{68WPeOSa%ho7T}(OWHp}p~l+XO)AE|G>1gXYEO^P%j&x)h8==)n+?<(b}VGEa7 zGyZTDwfm5I^}KeL6&%?CuU=eWjDivvyOw2Ri`;WE*TYwg83) zO9(bOtg7Az=-rVkQ|C4%I{Ag{fRM_-9B?cbVLY0JMr;u}`PYDOK_tv&iSBJl#5_Y6 zm=kqCGgJxmMFs)WaAtbyfY|)lQ5Gy=k>|HQvc?E3N+D$~6kZ1)aw8lsDlo#Q&PGcgT7u~nGKTQ<^xR@mC+L&-z0=k7!FN-P zkQEfD-D!mAUf{%GtZ3-ywqKn)4u}0xENG^lLRd+SHo-ES#WRo=&JV0Thw?`Au%Z3n zC9DLqY^<;ADf7MV20vU~g=|LUY{uNiTP}TdZP@2idDd2sMMJ`(T=|A?(YRB~vDQUc zWlH|Quf|wKb)eV1Ae8HSHtz4ob84e9&o4^p3kwe0S6FKYRC=D*hDS8Q;=3sLMyrYB zLuZIUgO`kx!BC0FE#J;>y&tuPR15~j4BpcaJv_)?Z(1%$uJQ{}owy*Qb^WgCdCz-i z!J7d7L%IKGklp&tvtQ>kA~Cgf?g5?73D+@g?71}2vm1bG33_gsBF_wz8dOGPvaq-a z=Aq2UYzKrke~^+V^3NY1u^a3spBk}J$Vn3(uGZhr8;vy3-Qp{krH@tx!3FLMaBkb6 zRVbZ(R6-9Rx&^3ZkiGy;o-purxKm_)M(lZhY?TOa{fqV4FONj~F4fMyUh7cTo`5(V zp13du|9f)`$xqMM5XoD-HB#o;nuznt9C)9!E`N&0bm6botG1u^+Je|}<<@cpl=`-| z*+*YF>0qp>!TFWmes7JWZU1Z5z^(grgJm`%;Rf1W!R9QiPL(*8mLui_VGO|hoj zq6IV=E0oT!bW%cWLrYY>ub%!w6gAjfAh#owOQTnAV+^9`%MU&T1GrlzW;b5x2rgQQ zFiqSVW=HQwxU}4Vb0&p1EZ)%ffb8LstC=52{YwEgLV9pGd|`~-gv+ssR0L&l!yM8C z?tqu@zbVADXSj^~`bkB^c?Ph&jUR+}@_)vCy}uM1%|K`}KaYzsa4n$rTwv3@nOFAF zoVVTE+RayGd${mfL4u1gp($SMx78;f?0eK^R0@ZtuMd1DK9!f3IdFPZcl*>kbl9v+ zjIAWds^kYVxC9)(yF8(|r?b5`mribBy(DXC_5VBW;fGwvj}wiSAM7i3NsXm#5puh# zlixoj_*xFL#FebZid> ztClG@3=Pl1dt~~(%<2=+XF4-wG68Z84h;=mt%Nwjwa@#8=t617s2;;)rTOdiyiw}X z4H%U<)$Kgn+0=xdGR3@-Jof?}CK%)I*X|jf4W*5IPnl0Lyr$jsVz-c3$eXmLAWA>6 z@78^Xu3ZI%Fx?QrRD8jKC+n)(f)e9@UuWvYJTA}RW*?J=hBlEf3)lT|Qpfr48B%Tl zUg8|ko+-rT6=nLfTjt&tEU1b5Q{u4p-Ns90-<@v1``y1Zj4kmz-K2>^R>auGaWd9( z{DaSNCf~|3F=EPhz!Wj1*RDg2@uj8<9NaGg(i#8^4X8^lHCIiYy%ELO4Cf@IIiVpX z;N--u4wYMpryJdAUStbivTG2Hhn@=R7e0Mx!*K1MFbF7nkShd|WWwh(`18~_NFmzL z1z%Vb5D!|Qt+~X-^}INE=}hu<^P2&R^PVt+j_Z(2IS$NoXhvL{C@UtuK+h^c(Xj*- zJeS>*ZlIr&xNMrX#L6p={rTu))4;^fAFRy_PLFi2$eu43#YzfCOJ{XAbQ3Ie74i4jN-ghm^prTiwS9STpR#h} zy4!4jgIDY;u8stC0@m+wZws1hUM(&TC%?~B^`5~Hg?@Oy{5fTE%jk0$*3s+XJktVm zMz4YWxjoZbZ|~Z$_kYr!RP>|$$4ME-H%iTars(U#JWt`um>=;gi`6h}mSb$>O>X&d z{|9{+IG14phv5p1T^uS&h!CV3!jBKjDTpv>hGDMW{$*;NOl`qS z9Pp{6a3J^zt>0nBLkAhmeBcui%d%K~1t?RGh6=PJnM9(&fen#7@yM7j;V*qLBo2{V zzS5l^Pi1%Exvf(4dR~F*A(84Ze-rB}Y!_Z_+tyWUatFOE8f4{d$Ce)!?(NUBAImR} zlNeRNEOm(+2n$})ec5KS89%Nra?*IHRneMuFG$=A#S6ykhq0eXvA4_Wh1HaP@+*e(VDQ|--Q zQtf_#Kx6?zOtAsGN>dJ5FP%}yfQUAbfc>=mK3zCt^gvRCt#sw~xWxOev>~$?t&7Bv6j|>0!+olSo z$a&}NKQ6V#Sl`3&><@WaIx|*Ho%twv_sU8T^cT1Fjv}UQ-U9aegsOO7_%kDp%{||H z<`dt9rFK8-LQNm0J*nx-SZ`7Yjec?*xM3+i$C?}Uq_2*SiBT^|_cX`}tsaTq(SHe# z-CKn#WThISIa6R2g(i8sy5l`){9xY!g%=bK_dTHh}bN^jXJ;_ ze5;fjmuMcrY=zL3|8wgBo+tnJ*2C)YMK?Y8Uy4?8gj;UiyZ4BO)_6g&xbYysQZeH> z;n{U=pR~e2uEI8F<$*F6k84S^0_7CK6s0q2XmD8BL`^x#N()Xop_YJUgncU)zgtP0Q)1c6p+7`brOr7Fd&^F_Z$u?$U zwYeA^qu#F+yAubKB_S;~P5bL0vj<|Sua?qZXxNOEMoTR?;U z7{)Lsw4jX7{1Xg~u}b5OoB_n}zie>xS*zpQ^5?%-iQ|}fHW-hvF5rbHKsXg9uS8WT(K#17P9r6$<2GB2UiI>af?l&(z8&w0IQ`7D+DT}JyZ-M-by2XOVi*!C!3k9V&wVy3a>D(gP40&KYS!PNf5o#= z6_&pru^3+HGugK4iyPnXr#&M9C=k7PbrEap#~`+N1>)~}oNuaa8ZS%K`f!pytpZQ zbpnwkyYhptfqD742Sts^g?cIbb4SS=kF=V39-KRR@ym`jI{C|2L9rhz4^93C9ssRD zNjKpc8CNdEaZJPbhy-Yh$eKdnL2b{0 ze@ZbGi1NSR|2w?;cct#{Gc?$utN>)u&6aVC;$hwMkrtwNve*$XXiP|%!zY`E7BtdB zpy+4DO@6y1qr=-_Mm-Yt#zk+-e2j#&*98_r{9UJ5x<|}9wlanfcrD-Gih&RCie-BD zOWa**jP;&H5hOAj)B86%>Pz-Krp*6ZCP@DC@g=v182*>rO_{a08-V1`^a@6Fi5oO3)+Bg34+h*8ivpFlq?IJ#nJh{df2!4Z7S=>>sk@HU+q%=rQ1IA4Y!0$8$>4VMW2VD0fH}D(jYciI*7)s4` zrKPy=v)ARJb_$Lzwf?q8W4BG8t>9j+&$aTm-k-}NsC2o#HTX|x;++*E#@J{tga()xJC# z>v57WR!Nxl<-E*ptGx6UAU{9@Zvne&El5{uWv3zpo>n^uKNw{EBy)3fk;fAa4b4v6 z#?vLMAbj!#SeTt$pm3Xk$i2m`ipdKM=BlG-F?D#0V?(AQA0oc33nvd6GjDZlJo zaBqGSfA@OO#rc7iE({X)I2*d0QBq!-MMSwJx}0Nqv~3<)u5}#SvMdy=W*?m z=G7CXUNRNcl$ReU6&8`$6|LQ}a#_iDe;j_q-z!_;LPSi2R9RJ^P=>#n51kZG-+iuJ zFgcwDFvE1YKNtK$V50l;dlSs!Af!wPdLQ~qU%+SJwk@&DIh{S3?+oRD{xN1RgF?19 z;U;c@6wV3c2_G3j{qi3I3$kZN5T$Ox@4H$pm%TKM1b7*q3k;LKi!2M>8LFGp4LTVy zHLg6a>z~zk_5R&-ndR^qw%y(rWL25k-;}ya+yB@i8T73N!+oUk+*NyCi~&3f$wjeM zZ~2M|PT_vNf2rr@K|UYm2FGt>$_NGMmK#V-`Y&@+ln5wJ3DJVoT0`PL?Ph{jY*i~LvU1((b_vl=_Bi#DA1t}OR=N;!F#&$!#jifzN+>j zHu<;CNo@G~%)3`kIB3P$M)1Ci@%?%)FE03OD+yEN@fOzjpDCFc=$g#o)={<4{L)aK z@jQzRrv~!W7kM5Y5Y_#ezp59)Z#!oGb1?rCKtg|jU%3F5omYbJNaIZIomA9EG5-Tq z<|OtBMcra+1YNf$q&j+lKJj49y{^L_i}x-4x1k|)Bu0z>ywnZWFldyHzs1Iu#`Rqo zy#M@;S02Kt_b2RtT(bUd*psMm%q!t*RL85t1^;$ATp!D{t{FV+gB>@lTC(pDnB+as7Pv{Lhl zK+svGVf(X<=)iqxQadTyl3kl{zZ>aw*q0Rz?*>O$q_qe>@e4(;BwEHeh>i#KsO)ek zbAbb76>K-fe!h7n3TU(^1hP7~SND1V){B^Rv%r>Jbgw&N0Y>|y0Y&_i8!5MnYxUsD z$(ZyEs0_ep1OFy(L{motRtqc%P;{MMq&FBhdBVkY6qN6#1BckQZSMitCc-YQQ*hE znfuprD2}<8lx3hy7gx+HHC->E;*arKVp~yWK$#GooJ*cvK_J7ME^R&aV)t@uyUG4> zR|EZFTgT`3G;eR83jKFja=nnO>sCl-Ehlxi!tJn+5|&nL9%Q?AyitBEXRoA?-54gO~|q3_md7%Sw|g1Ud?t2tIa|NG5F(C;wuO_Zgz*58hOJa3~>@^ITlLfail*^*q^rbytKb83L z@t79WYp<<_>1s9E5Sd^Kj2TB*nFh)Jqn{Tf%TmH=`am{4afAsL38~^1mmX3&GvzC5 z6?1Lke*Ehu+0+c*sf>>#WK~K$0fSrQHp3Rl1^g$31;b|;4Stj%(8@+bqGCCb2^ov{ z3&cZ2s3zJ4f(C+;CgViSq3M?a?{0&mMvP%{{A8@^S}4BE2`6*<9!IlBlllrUH5G}W^B>s}P55BiPK9t9_>RYz_u+Mk zhb1?P_S)Lo^6J`GJ?UfmYX0}+i|H4Nf~Y1$CUin|1Nce;pdJ1T_DLeHa+iN0U>Piw zrY&TS7{J#L3--J7NQ0)GP2cDGV^b7vBh>Exk&%&wq`p@-ouT^?1g-90z~)L}#U*{P zqd8tJ&bac^>W>dgwy}J~i^wmX#QPh96jH3kvO=nI#$wsuEtQIL3xrZI5ynYfq2*Hr ze9kqP0L=1dlXI_SZ+7VMC49+93n>$Xk2v&ORhMF-+a_84H7-rw>ZWt}Po{>;&e^XE z+g8`h4L1mBW@76PH*b%-gqJSYYCMzQ$aH_$zq5w>%T>LnK*R@+#+z+HG2hvZVwdqw zwVa5s9Bb;uXwg8pN3byN0do;u?Q4VL02P4Jj6Sd)g<^Ol01>gd4 zF0a+;AzT`*JiAg`QAA1M*AwWBRYCo^J+=0XX{d!`d`xL*oGZaDWo@b!hh3f6)_6a_-)}+h?bUIMX|Rv1H#A0?ZonkA+*{^izVab9eGGpyaU>L9h$(~%P_7(&tI+8gb2 ze^}GSd8jj^bj8-#C^JLjH%llVme9_iHm&X7K_OL@fzD7X$)2uBz3m=5s_s=nEZYYe zq`dexhL7HRe0?>UpFLUFd{;>bdndMgnpGZwYsf*DLD*W$?GF|{^L1#t%k{!ms(IJ! zVYvvSa1a7@!FMJNvV8?SHl~f9Jf!ee-a6?m z=U)8Brei&%G#F-poGj#hRS7(;1F&oBfm;sNk$?Lhjf25aGrB;%QIBZVf)->T0Lo%+ zTa|yIf&K^(Y>2eYi=MX8f+RsOOCd!tG-M-d0uEqzyA6sB>Zlu4Ks7>goCg(tQiiIe zyTTurSW|m9W)yCmF$tENI2cq=Z@Y}|*DH4qt8jeuzRO_TcrU*yvB9u}r-g1c6{dtq zTTrl}(|`4u_&6t?QFQ+J5 ziV8+wNE#D^m#~r=tCDmK3 zcy(cuuV5u&vC^&_Vf+gnJO0&Z_|64@XchG3@Q2F@(=6nn=^zXllBezh+Ulqcl}|9S z-arpR1Yy)i6(g{sV&8Pjf<_oi*$rpG(UnrjvQ|=2`Mfq>$AOpDy_59nRRf{t+7S$} z2K(0&Fu|-t1s#zrebD0pY0GW~v~!YPFv6##SL1TyVb%J=&fYbA^_F}wW_aH_<`0^+FZ09L7|}=%rt=I?V0^JG zXwDE9enfs+?_j7Ob+({MSOYq7z)EHe`RA-(s~|KN^_Rd&f4e1~=Q$$MA|fI_oir3b zplUFH=>raQ2wWUqDgE{9*HX}P8@|f^P`4J21{-*wvY`$|e9=Hz0y8s;hYd-#VGdj!VgR0>SwJvH-{u-pLBCqJUmFk)ubHkdsmwmKe*4~ z(SDUCG$M$<-U#3K$|>>;%Sp}1>^RY%u!jAyZ_wG9$&L$oRlgpQOiDvz%P9E`w*;4R z=E?quGQOP?_hYjaK7#618QD#1vdVtWQPI;&s_Nj5lf#hT*VAi#S3Jm(es!) zaAHkesM$S*WPyO^(!e{9uJpNbHf)g-^*bKWWuySw9O$ip2{+-P#$cbWpP%2z1`Ppx zyGfA8jUu^G)Xp@32jif>xl9~NtXJYS;4eSmgH24M@j{LJPPWPRX&8#FGe2_4cXz4c z4%EAzjoweoHIMZ-STd2_sCsKQhxH+TFzU~iOnn-Cq_W`6IFCb8nnvR&V@`hqOxYr%VcG$8A3WX=EJPNMMo?W<1>Qt3v1w{LUYyxa29V4jSY ztH-C%Jc}{s=%^aZc0`+US65|DmgO2p?3Bkcwfl7UT%Yo6U&NYGvX8+c>3Py8kWFVh z;?Tby{lPt5i>4w!KMd>+00qiIjXRnt)YOWH)GbHKqjR-N3^mBENWMUAKp<#uoLxVr zd^cR0-Wi;*i2j3!X)C+$$s#_GQN#~#uMzN~B>Ct6{y2>PewW7ZDi4+smbJ#FlsnxYCO{*Po%g zdZlBhz}p|#!{c#S78)Uvyh=>`A97=+o@Fc!nyZx+X{m8x$*%&Wko7B@jx1M-i|$#d z3a(D~$2wwedB(UGlLs`3ljj_$4Le=03&=6O4lsk*jxVDWCSmAXArK<}yrhq-VP(bX zm>oYD$yys}9r5-;!}^<__f)*3#BP4XyV&kzD~P=&vwvHGGbk+%m)&T*ZRA(uC6YIOPZc!KHMd^FG}UTR!In)T44YzLS1R^ zOeq?25Ksg&6Rc3jOGZeJeyLQU_y{hwqkn05AxeclQU-w*&>s5}Pc)pdYu`wgQbEn7x@6o?u{vovtYF+Dja%ccpO+a~Jem8aE;lPv{}a<`yhD|| z%yX8@#q|B$cwB1uT;GOQw?XdQPp!t~F?NMGbb{V!X<++WqcdyxFN3KIA)mNemuzTe z6qXx>+ONWY%r8TVE8mTUUzeXRMl3pB`O`{AW_|sj-Ef!x>dfy4uBFdcYZx01a&(mE zp+2jIYPrhy(!=K^1ca{+bJxof6O%YG(66%h%@r64D|&^f+CFywfjT>(H}6|NzU1C5CM7??M6epU zI3H-@K7DSB@|~S|;DK_7$V~@27XyTu=T)rmgG>IKwjvh`MNGUVb=3Qh2#bJlcw zwIQVXg{Ye;!J{6bg~(y&hjn6uH8lW7pjN-KgfODi`(cmC|Gp)!j(jm3pb~-7uX;Gc zH9(n#1lI@K%Y(Kc-bMr?@GK~zQWcC)&}#|IOddJqh4 zd0<(k(=b`7)%D!z$G_ez-tX4H-vLcS3s<`|t9Z^KXei^HC~F1^SW-o=o3GUwRFQdw3i0 z0Sf?43p=O^x5u77y_H-J=K+-W87ndr-9P`8{l!wFuZYj-t@8fwJ01{wNeo^pV9)SF zHr6{oaRWj`%UN)e<=J@G@3%9Dgwoa{cS}oSQ*HIyw(JYcf3H%%UeLr1mgLkQz=F@r zhAFzKjrU-G490U)wa$+v3k%tCiHl*<;xY_I^SHiQ2mRdC%>`M5+qc#q;Ge(_HR;Gd zbh5}xoDG-eo?vjwi#Ib2*f`_2(eblQUEm&*Fm=79-(FYx@W zD=RA{CB+N)KhP$(K=~8jueCag+k$+{slIc8pmQT7wE?7%3j#S&rZIXwz^@-DnKoz( zHknB{eSKLkd zvv&zh81gz|ClsblND@l)$;e39Xhcj0l(ORE&T8(5KGPk`q<`Pr6CtQ-%_85;4LJb? z-Xjt2irgQ7C64BD>Pt1x0D%Up57XT}`iU%}j9LD-qOn>1op1pt(Z)n)H>pK>9&S(Z z->+l;*32R*L8`zR?kd=BrOeY^Bz zBW_vfpf+8tj$gRReZBc}!FBhs28V9e*1|8{33^do{fP&O(gA@dwl%qXm+|^4_s=Sp z_3-m>mll`unevkx_)JU?LwbZC^o0l4`e4tx#=Et!bI z1&Sdo02Z1lSSXu9RPUlBIpnPYfE0DHA$=Dlk9n;5{us5R3V?G}A-K&kLV zi}iq6zB^ycU`dI5Y)862jZfkDn3HV;OkyC@9dM*cv7w32aLT^wvRSr$C}cT9u|2t^ z8(7tc?Gr1|h!Iu5CQ(gL?tLP@ZRnrTkvi)}{~ZnpFY|4Xrue=fqGkCmG5#@Tm>vIR z7zLfN|5~&n2vwYAqEz^pgBJP3-@qDer^BT=BK6#@+gn4TL}1Q!GiHV@Z?U+$OWW{i zt;Z>E{twmHW_VNYDv?6e#34P*8;AxbM8F^P{DuZk;fsx6FFXZ+JX94)k+Uyy|#iP(S| z)F`nYvHtTL0&1Gj1G2VTKvtK5mJePK(VH#z0`tH={(m58%{z77Ks)hYz8zFRgN_ei zz(H+QbDMne;5I~iWfl}fz}kV~vmJQ-Py-aVsxVmsZc^ji3i;`vPXM$Kg>#@(N)(hx zX~Yq6{yaVNO(Oy(Iu-2~LiT6MM|bohwY$72-Fu3E$2B&clpeB@vF_4Sw0x9fg6ZrV z-l?)b!xggX_ADtcF$*+=DM~c{oxcqtoW%3BF=x1ynKcEktK-4gec0g=4Pwdt zgH)Kccnar7?x}=x*t83rdrdads!Fq^fDd#CfebhG&-^wxv*RFq2<4E#cJ^(H+f5w6 z3%%bPuWqe#cYs4+o%s z7V6uKphpj}($Ghvs-Y1_{Ysi1d1;Z+9Z{>Gk%2@$RlvAD5STmN3#Ywu&M`U|Z>{%( zXEg8-B%uv1kTtyvZ6XZ-#dv9{ih0Phx_)Eec1O_03!6ptkkqqtSop@Li=RAl8URaN z;n=;;3$N|no}}B5blw}V&5W5Uw{P#o|4<}t{ z8pTQboi^%@Wuf_+Qk5(y76#Fdu zV_zZ1JQibZkdV@I%6Me^xMO9j3KJg4%_WqWUrb)Lb;s7*W7Uk;ikE<2JFZLj=9S{d zR%z)jnyS4mzlq!KVT}lzOmrTO3w?_&y^z*7DAJQ7Wd;EX7%STWg8&p88~~us@Zt6Y zpS(r1+KMaI$?9)Fi=0ZQ`-KV+!Wfr&+pFtEh;moWm#8F(uZV&c3gE{$w zKDjSNOVugOdINnsnWklrzX?`b`RC)LGv9WT(H=WRDiaOAf;|``Ek#XAzu3*K#+Ps_ zp5oH^Wx{%fYrN$*5iZ^k7amyPOE#mWs#K+_w8T`demZvLEtmd}JBa*8fe-+bM=qM3 z4t}VCP(SpDrzt~aY1X@UmuW&v$omHdz5x*X89hI0eJE}{p$EJ1YUEhs(YS{MtuSRH zM%G5Bq?d0PgI1*b+P0_Qd$cTk-rzeo46+$btG`m&n!n`I94(Toy^U0ufK_}766is= zMT`g2J+#on-x~=4(bw4ym~ukRA25ip>wh}70^1%hZ@71Pd1C9b&Y9H8{0rG|CdKKU^39JjDHCXCAy17H}j!jQ7|nTRnAjlha3XxZgOh zqTB=~_>$}J1!u$!(JWEiTlSjpqm@MeMhDjD#6BdcRAxe z?FdDkYj7Z`Pdyz&O-t1#wwm1Q`lpxh_^@cvW!jb*N8L@G_N(JxZGRa*2i(p?KYYb zA!E={c5~)ns+z3L(!7_Z<*cXM!Orhwuy`%JriQg(ch@w-|6v>w-GYA=xn`vq6BWKS zH%~%D$q!k*eGkWJ5O9ihWefuoVW2fP1hgW@AW*csU9YNp zW&rx8;5H>fhVEEM#(@+ZW~A!?<-uc~Z7?eQfcC*3kbpM7l6AH-7j(36xT8qxUlSK< zen6#x5k#h=@M6Ha!Bf&=~2NQTmklVW$NSh@-N+V^0~mdG zvS-QZLCo|WdVA1Q38N|c3s-LMFVaugPAu{gK}v{4#~r#WviYa+0D^O4{^8YexRnSK9Nm#4LJdbZIc}ER;~Uh^PF@ zXo?=Z`_YYjZIS8ZP~Ue>yFTuhHgBKVygk++lc5za&`)=@iELvTA@=SXfw!Ys@3gV-$Lm>k$75`QSm4X}Mp`;=#)MT0}#G0*HT;q|z|W zz>(sI&16XGSyZ|VFVzfUPdZ=te&d+V{Jm_kKVe5AYA=9uKoCz(Pxic6z#5DIz?QMYl9{E@`2~3g z(1UcQklOg=((g5FjPb6!} znw=Fm8NL=+Mt$gF9h@8l^tj>M`*W#QByFS~e!lDiZ#kbbxeE$To8%byw-8~>oX88h zr31E$d$va(&l+oQdPr({HEw72KRM_gy=~}pJc*y*fJJm5_{V;hd2-%t>YB{1ZA9Zq ze%Eqy6LOu%$+HyYc}4-EcR*7KWbJ6$@1Cy8%!e#aO|lik5uOqJkFfDUFh~P}L4Q2t zI|4UIEbyurtJVDw@k}UVz}NgbY5VWO2564;>z7H}+kbx9q31Ln`5ib*zI^%eK%M=> z`rT__(n4T542r)%5NvMuy(2}eY^k6bEGyHH(h_CVld5+BF^%oeI#=-YrhFu_mBJ9m z6Hti2WO1Tzaa3@?q@I0ABNPuO7l+6G&nny?e0WPy=ODroVR0$m9fS-vXRs~bfy=;!y72U%ct+<<_6@1(<}M}qKbp~lL}t<|TK;zsqi z^`tBhCEwHf1V%;a6EZbz*}LJezG=XF@a$Rz^_hW6$8<=$I_3&79A@!e17A3J^yS;;bg_?y21FdBXl;hI3 z;eYZL!vdwuGzmGa98v}^rAF(P9}*qDKFJ}_W0)}f^ACWSOL(k@t{3A!HZQv}yN}rh z7VLT07d`FZ@%V83S%$UY&?#^~SV#E2`n0Klk~wIli%A;@6wThmr7Y(E=Nw^TCb ziDbt6`>tp%fgS+c*JP8hJ!a+RK3pCu{txQqjxR;D3KC}^fES7B!9va3-~k{O+RdJ` zl?PT<3hsq4?^nTs7$41YA5*%FQn#RJ4ILveixSiS<*Ca%;La&usz#AoM+z9IYs2gh z!a7RpCD-`~90sVC(lzZOKRl9^qev{sXXaN^ ze$TR3t@uAgR;-a-hM5XDozXkN%p7r||5oxMKPcqb&;xO;s8R6c6Yt;4#lnvr|G0Jju1vc%Jl~NHG@vzA zgPjswBQn|zv}b?D~+C|>ob%1w8{kUab}LkaJ}%e07| zs}`nf3u`v2z3fZxDQA+diPxKjEXH*WNIFFApR|J}PlC_g3;T)Xohgs7TM&Cw>w ze7(6zedzl2>r0)J;#tXM%@C%7C_r#hJhmH)N=j-61t)T2L(KBY^1SBt>$K1O{ZB(M zT@zW!OL)5gx4b3Ab#zE{d{aIThXmaxla|;{lgpdXfht8{az(toiyMC!18N;RzQy;2 z=*vfrZ?yD4N6?_{#|JMU^czW=Lii@m00mzLGeX6o`5`z6fk!ar99*&cPa;_jixZ05 z&;kPP7e^#-hOkc;sMwa_>lg#6q;Qs+}y}mJ5=yzui$vl zOKKv0|GOn-AWM#&Ys-j!q?ZQ$%HjY#FX)erjnsxiW{UuD@Ihg>U744c4=#O3HETj%UD!AwH4E%?Xf6b^5QU&~{t~wDKqHXplj@7@Py{9G`Fm7v_lo$3 znCfFy?ejj05LIzFB^Qc5Z+++vF@dnBhw%4dwH~W2hu3^;HLNjiQO&e znkHk@9$6NV!9szA z_L?%`oJ{JrW`qT$e;l~18A~-)?}p2j&^uKbtHz%9GkD-DdMY;NO=N%Go|yUk(|e2| zZ5iIN`_?7&V^;{_OnkIO+kC51369O~{n z70Er{r19)#9r=u_!9*$psPf-*1LQC^df*Db?89$P2{W)xI=zNCy_Vq%Q!z7fWo9SW zy+zLF-)AY&o>rOd>G;t%F_Ap4?~5-Pa@_#(D^RTu(D2i)@u??b9-?vun5gBAjEpep zEM=`3SM-`5*aCA=X+|O4R2akKZ5)Mq2`$v7OoJ{BJR4P>N>=p552>T{4iK zF{bKgdNi0pCFa-!1{xe8S8I04R1%SXXXyS;eOOSm_nkSc4DNQSR>{Ecpkag~A3o$- z=s0VFB(!E2>s)~M(>h6SVQ!8JLdxK|1snjH=}M0ggd(B-~1e)n8X90OmP06$2TfOIr z29r&Rj;qdxNdrzpNnY+{7B^RZ(p^jSFMt>MBD19I_T6jt@(l_l^n{{c##>hEr5s%2 znguV-m6J`orMlQE!jW6?);wblhopW!ybYh#=(0_QO7LG#u?R$zLd~DZ|K=&)&ATeDbStC|ZbYr6*|TTQ?xEzu zIt^s;qYL~C|DV*}ys<+uGFUDdI0|0v>h0nW96;BjRGtDjUE#w-^=g_g+yJt3fd`Z? z9;i_0m5w6?aLm++FSb0A(6kQ4NKw#hZ`Ubt!4_D@63Wa)egH2X0UXrzIg5}(J`(v! z;fqEkL+Bs|;id~^s`6{7C5}$uc#ezn>cwSldvaH$98SkfM=hS&2PV2@c)`rkuFk!c z%p|7eiMgrtnwuV>;(~WMDe88#3s@UF*;&sDLmL;0-e~Kj-Dw5+|d3)&aHp?*H zKhG@LY=km!j-3cpXgl8bT1j=};ffvVt@e0zv1>&~(J7P8Bw!~^#o)d6y(Y`H4!n&e zchY~{y>M^5!1&Q3-zo>`c%t7v55IjVYlq#xbsXaHe$+O2?pF<|(!jwBP%K*`gf_f^ z;S$l{2<`GGyp*_T7R;7xMhRLZ6WCKwmB~M)1_1WYK^lz|fG8NCdUxXNk)RJ&A=&_{ zuJJ>b^antsPESt@ZVwnOJ38NjbYs+KipG=_CFu||oK761Us|pCyMa_cZa98*9^oZWAwK(^WBNOL#;aPNJ(2^c+#~F>)CGP z#%mFJ*EW0uZR=OU1INpPEVZ2cKC{bN3J^9EU{4e`tKS&@!kJ_)Gx~o2q3hrpSF=>J zoBT-L&~6nK+WxO3(>ipXHD&#M#j)TbYoi@4PnDcQ^Y-juPm{I$3d#gpCL;|sZ188f4q zpZ#O2A2*KUMtq+C;Il3j7@CGBKzuqI%#x?y^;SmZV?H4l-j!i1(9u28ioeOrdtB-M z{RWtEf)>lKKgfRuYmp87@gUHf%jhc}T&kzvZc{@<7QlQFz|Wv!Pu@$9JCa4f+~I>H z&@V4CHpdWE7{I{&3xv3jZHGbATnmEZs{i_t86JfqD)*QhEZ3Sk)cE-z&Y0Eb6+M~o zzX+`05P>jyg!RKce_Nol*o7>fQWVCt`EJ@`HG{Kf#>vIyOW zZ(E!$oW@n1j_LC_JF-9Tp;hnV#_ zA0sXjGzoo_%IOvKMl)iu-0Lc9SKS=e z*@NkbjCn6hGn$@OrKe|%U7ONsv(iR$X|qU7PtVc5@d7AirEC^~EO5Whfm-QjKA5J# z@YW7gs?=Ze0xUl#`yIej1G6NWC(9r>6xj<=AQ7xjQ^=zO6RXwe4%C1&BDoE}12alO zE1i=M92LJN0fnFkY z2#AiMx-S%m$IZ=c12519jEo-*pc{%>;N6(A4-Ek^JG8~djQH%4AXIZJwoEfTR z!WTLpG1qqaUI1(3OZ#5!TV=&u(+Bh<$`L{2iL=M#MWnimaQx_CVp=rgEv8>J9@co- zb|2FcGAz+67p?oX{Jr_i=zD^=mIs^JamVgvo2dA6i$&)(iX#ErwbA(^WxmZlf_S&ED2dh--iutL6k zlIej0phOZlFViLXx^1aYTrfO-ysL~<1YF=gLMjVi zoO7D?fyAOqM@iH>HGitUSdAU7IPHOPC^3!V=%StbchUBM+*_m6#vnmh0IWoA&b*>O z72-3%UxRF}FnvuRHzgDtjl;+0^gNheKmhnV2!gGygY6iKh39H1o{zqT*wRafb5g(N zq=YAcE4t`3=5Z4-a_i3CyZbNedDZptDW%`OCAawT>9#sgVN&t+)#R&X(1_j@QAv%< z7at>&_18>|rEv#MP!_fenof`m{4o&6?8K*|LfdsDUkCmA%C-JdP< zO1$Ym)qLn89I=^O;u zrE#0se?fr&`XLcD_>ycYqS>)5hi8ChoP1UD>nhT#U9NtN)0v>AuD(%s>39AYD^PAD zjjsf38|Wq+y%X~62a!&f-A2eZcZ zfX*n}>eCv`H#mVJLgB-Hbz&e@48fV`0qqs4ULJ?$?nkm}=U29ud(`fy1=RUU;xPjA zY@k(^xT4!#&azwjs$kLVzI={P%v6TiU9D_K=R{e$(ZEO~D9y?ay^VrvXhEgmPyDQo zm|8vCvMY5cu5V*|iF!GOOpQyKsitb}WzDb}SMk#Ka%Ul{5_1MqCL9w-X^FI zz;~2ju;eL!jX?q>G1xEV*~NJkPF~wPZ?PQuHSq+?EJ4L_;Qu zc$1LLoB{5m_FCp^QApkf?n49%j1{)PgrjClE|3KSE^uqP2pAew1)5+-d>X;n zq~AX?k3t&J$5wr?G6}^J|EqObupqYM8EDd061!CxfBC3iF1+^XIvPBM*CGa4AsAE_*CbzO zi517t@m)ZU0mgYw;k|{B4u0Vmqs&WBX5B1DyDHQ69*Q20@YIYvTp%;~cGr(%2yt7#Nw@?4roIor= znwBC^kKchM0F3W}(kCRfM!@}ju;HK^?D*Yllou~*0O1FN!3^xm)^^{)*MWAjpoU37 z7#dkaEL;*BP;nk|k8uGRh=ig5C8N;#H6?V=DAMnIz0Ee#f;`wZkIk?6C8orD)X zpe03VuY&9K=PWSZop_c^>`CK38z97l>QI2zXQu3%c2R_N54D6r2 z4hq+(7-cw#zsP*;{*VGmbTLyH>aPS69Vga?(nGKhbS{2w!CYS*+DaD4F>$;I3*rS> z`d1lgTL?~NP?`vwXXw6|r=woCloe)#{XvT1byiWx?05dkq>A!5N=jNTxo>!r*Ft?G z&CLjc-zI%pj})gqO1U?{)5#8fWIDtK5u!OJE29S&r}#`%2a0xV??nt5(?Ca7(q;y`>eoN_=0b$>VN8b!g(tgu~5+?@~gw{@p=4S?>InhFc1jmTMlymhIC8 z<J#xwcEv>$fL&T+68k ztFjZb)>_L$^{^U@0zVS&NfKygW%aEyLPdf`h0{BQ^x+|M&5l%_u(dSHbx`mY)bv&@ zHYQ5@t#{QHkLa>pg-bI8bN|)=!9BNxteqU3dE#xH)mPo-u~vhm2i~#4oZl#$h%B5l zpWZNSPvauR?f)CXe3hE#)Ta(PQOPsIxTg&BM51s_%6B~rcLd)&S!uDDx%LVd{$}jF zPoyuiy>8rnzb;%_x^$63_5q=%bLair${)gk2C@~7&-OQ4@V!DC0*R)1Roj}7e*u!x zVBI_ax7Y}R{;{paeuS(d6cDk<@JogYC!%B=n0m}=uGcG|iV+&M zwx=bzll5>n8dfZp5LNX8Xu$WgL#*(ruwy+pZZ7Su*UCHZoG_bBF=G$^{Dv9Y?i+2u zl#L|1p4-^a6@Z13hKne^7Fi|00wVW0Rh=ayr!UbzGQ>&)KqLpX9gfwadU|@tA(}BY8#gWa*;*j6uN0oxO>Zq)>F(}kmkzeo=9ZkapyJ^O z?K*b3>t-^e4C$aKm82uCvpYA~uav@$cW$_?p+L-Q;Pke(>;_EXZ$PQ4)O5D~;4+{O z{`Xq(*l-LSnb>>T1!nei1BBbHoNae#3PB&Lb+5I>3PLODI4!T&eQjWzrasL}M z3_U7j7>ci)$8Ex;oh5&awmcJFdd5<*8 zvqpi#K!rUKlJ7kG|1B9p8GWct0&I!-s!3aLY3d>252~Zy{P6UUWmw(>?4bWc-%~yP zwGNC;Y8;C8+1s2gDP2aiFQl-v{v~uehfmIKNTKUIsR9)!KrMBGa0QUQ)7h@HL$>ho zlLUzuJiWopoCcf8FgeLov~9zPTF@D#&!5>k+l4k?LQwUSX@G`D#} z?#chn5bBM{@H3KaKsxfenG*sYSv>8F8$o_@`U440)jr_50>{$!_R~8yXum-X<>3 zm}P%odb1`c#e#bE(&ggit;3ANhioOEj_6=z=VN{;qSx!hQ`Xj{cX22q9D2Sw63I$B z4XRhyG!}6a!q_RAq8`0*Z~t>#$msZGwFXPqr&{%;C+grhDDeU-$@@@G$(JS64_$oE zioV{>JqOqSDe&L1!4;h5%OMfK`9=c?C?WdbD!(j4r30jzHt=T;^1eoNTBC3%xxy^n_D-owts>aeJy+Wm1FJ8`$$?btJK0zc zkGv6Rd|CUb;oT4o4EdIM8vX?xQW<$dF6IpH(`}Hj$F9O1CnR`<)-@DNXlUqwfgIf- z5c%%;{MpPX;LCs0LBLdtYti@K0REcDunly^QfYZPKe{u3)9je$JMvR%Z<)pGX84fZZu4j4+>D5s zaueUX9V8xQOc(shN+uC^DApt!7~sLOM27rFxs?^}1UgS&z3GBwiBUJ;wFOO6%li!A z^k6{1=2zw6ilvPil|2rO&(&Ubr}Odc4#nSIAP5mOj4QiRoK$TNE6K?_JmuX(+HCL>3Qdn00Ji$R(8NI@o=rBBws=APU2Mhu_+vK7@PeBp`3xb!Q4hlN@ivCjC0Env3@u=`HBvt2cdx;i;k@**0c~=SkM5W- z36dl<#-S<|z_erZGv>hmh#U>hVDR3X>Q=Ok4h=nxkucq52dNKx-T3P^j%tUCTbF)g zli~@im%eUGFB_Q2?Wx(h9rLDr%_GEeol9sW;ybsz#)aXx(hTS5&nas&dZ~C~u9oFp z9B?ETAjPS-=(l5S6u?+DXL9%4BWhwvHU4?OVw%J8{${_)9yYpiu(FSE2^eP7vR*v^H9@n_&J*WI^up-ayCmPu8#ZP3rPX(qXz6 zQR@{S#zuva|?GgA+7t*wfgLDweK=I&_BG z+U}|tYqyQ{$6iQ3A{Oq<N&b^x1qWUrTv1kd>b=f7jTrR1DYa-ea|7y07bBzp>lw{ z*bE}CkI>XfAoR0iX)Ki5^O|2@7**Rlw;q9Em-L13rLi$U`3Y`K!M`aD7}mE=6yZe| zfIRGM6-x!YCr-W6eu4uD`A-QA zcKTimyPFJj3ZSxQcDyS@DGbDN7^_Xt3%^>0UWucKAvlnxfP|u>42>Kp4cg7P`yNZe zSBlJtkj6W2R$rs(wh4RQs@0&GMxQ{b?j(rWDfhC=S@#p-gl4U{qU*x9rpmts)Eb?`&p|>w!!T>X6m-6{=71b zcdW+7;Cl}%*cxe>KIjC99QL$xbK?meJ1BN^Xt#(Piptm~uE@@Qy{AFs9rxd1_yjnN zUW=L8!VHIejp7jgeug3|J1(9fX@?W_>0CbxDJ2e&jLmp)M(QKS{;g9c#D0Bu@L2g4 zCGtik6kU1*czP-v_?W#+p3!ts2pXS3i~&-Hr%AqS>Rb|koEk+(_WYxHFEf&ng31Ee zUGR~@FT#nwZG^jQzfr4ZIUAlDWeD{Vte<0#!C5(L?Z*LV^BLMZN4du0ZD>t|YP9nJ zd>$^LuwNMcpWKlPd^ zp}aDnC#Bg{?dKrRL?0qJ0F(sai*OB;sQI3Crz_(rK@{34#+x!;7xd!#*E3tji(lm| z)@an(aN@2Tja-wsE3pDgU~U6y%R@eXbFP>WSK>_gPX?QoXMIQ?_1wfEbEdR1NK-6_}l z?@Te)1{V!1yW8Ll{|0tB-gZUTu+gPkCK2XNECR2s#DBWu=3QU%>R|MI|H%wbG(;ol8L<1kByHZZq;jPaX1FAfF7JkDvgB-5na_1CSQ>U)kos{ATtBB`DtB^F_owN`?FvrGIX0$B6<`wzQyHTX?sYaJS~q zfpDVx1df8xycX#YuPdF5d!1zVnDg^^o=aix^1kj=Jk{um&&+TQTV)|BF7C-aj9us- zqu=jjd^e;idUrkDL(*=1s?k@JmOJdwlm)jzrsF}4&k(g%AaCD9Ilz{X*QCTGs@$oB zi;HbInDRQ2^d9q;u3KXN__lI4x4lk72fn>dd+loXOx;dtRuCV%`0Ns#SYMB;tfs$=OobwS6BB z%s1P*gCI{bGckUEAhq%9y*oNZ(7H@X30VT#f)2Q1l6aDY7WGA^G4wY-(r!}>lW0lr z=Z4;Q$V}ADaGzFPT+%mMI?MZ|zw-`tIlhlSzVX7<3r9g9;D`tZ3I1JjReS0GTIEn4 z(F|gw(94j+ZUPiBblr}F$oImHnqQoSCarYB!g{b!ova6K#^X0Q=ueXpc& za^rF673lbn=*5*daTAK=6eDF_IuXlFG7>ed^+{X%boNfBmiAesVn^W)@6tjhw)*wZ zKiJZJuQZ)?V1_HhAwOZ2$3iB=-)zuVlo+otAtrJp9M&~(S-6&~+$U_cy5Rj(j(6?c zQW%`t9+Z?ZW(IL&)vJ@G1c|ec6zP6WXU8q8*Ky-t?~Z7HKt1mRijH5JH|5eOmxT`BL5>@E1t2E9Op(zN*uVtFz1`*wSC0AV#=M( zKnD$GO^{lk^lujkrIco%TKG=g^Tknr(E#@)7U=^Cp^mOfGnd7-W5!*wN{i~mpT&hQK7mr1Z_P|wi2A~k0a{FtjJ_hX3rgjjtFo^`(Xgvm|c|$`y@EPKC zfoEp5J^&211uUE8OB^7_Z-6YBP4$us;HA>k9C{dvi~@oB9Z^ve#BjX3w){C9`(q0$ znX0O4_rfD7Y6nsTt%4L=NSFgMa8>RSKmP&|Rlh&Mn7xN`d}c~fd>UlSAR#Xd{`A-j z5}Aikw|E4uqIisPerHoAu?8CMKrsEkPo=Z3T8zmQmuGcYtq$GL_J6uq&sH%uATmU(^U3J zyL`s+^YcV82C)N<^s{R03GF=ZSR0)ly7kWX^jp>woacPfGLD@5i^6-1%=7F6Oevqp zV;AS`S$p}lqnpT1KiZpz`^2_}8@ct8tqyzpvhG#<-vo57x3AP4 zICj>TxA_k|qQCbdMIGlvZf#j2LD--iQ&i1jaIDCgMVvdTl>L zuJ{Mqr(6QL(mP)0>9Qc4oc4|FYN_Jc+-A{!G8gA z%jRjxN)|spDHM&2#tOy4felNibk9+0X%UDxN1&x zw86Rax!`STXn{7I&vXal^Up-slx5xm=EPO}^ekXgRBxC+T3IsjsvTL$s3@aQ| zm^xA>>sd9aoh7lQloYmSbY#88he< zS;rdWFFEjni1&&!q@T0xb1o%GJ z(-Eu)n_%7GFzraDYLPqB*^O8C!$R+X5^1{7Vu^AJpqUTYg^Vwy=x##~1S{DNK7559 zWfc{=@#CE^`vq)54tseJ1)+~MBgWldAQ)V>6-Bt^CP+k z2X@hU?{I40#hcABxL|_*58J=lx$jQNEX9|;S+2n4{s=@IV3GzXwhc)$!N1`XY5&e4 zC(9ulkY(H!EYf(tIk&+sZ3ku3A%p{a@n`y%UOGi>FUu%=BNT3*aErV$e^8y7-?$|4 z^TOS0NQi|(8$KFJJU?rmfI8IwM`gs$PmDNIxD}qr{+;@79>AoJq(~qUvCB!K1{4+< zb5P0%HrF}PDP%wah7C#D(Dn&SeH>tfn{Z2flxWV8PvBig%6pV<4LO~mC}0n~Cuxi2 zR8tq0yBywt#1K`@AfqGN@u2TlLf(qCEruCxe~yh!oPQJ!_OkYHgwFOhUEs_?@Dygb zJ}2A8+`QBM(g_Lin|#jSYme%26NN4gueisu3-G?fDq~Cg4dC=qO5`M^Yj7yI^a2+1q3t&@zfe!Z;l!aOrru<99| z#bsI##hj3pk*jKXPu@8=f+qA|Zpgul(Eu0y@WAj;jl%K2oc9#cA`{|ExDhdK`- zqz!t){#S)jFFV@lj_G-=&bpH6w68KY^ucVGO?XM;=vRLXdjleTEWek2>}}r$Y#1pZ zBtxj;fbgO97VSoTP$=#k@!>$c{WJU)dLHgC@%1}sK-)REM<#yw5M}_@i(I47Yo}{F zp+40%*Lh?VFGK0FATKLq?R)KiIyH%#clxG-d@jm$B0`=xU4$44@tHJ zhL}^#V`bBLhI%M+H4J?C)Blx}jRsXjHzUag{uaLY2tVaHhu8U)LcAQfKP~%&<1% z*U^zOe=1tK^Zvl&!&R4t_|NpU1f3_{97p6pSVUaM&W?8i3al0qx4POaPt1SM9f;;%R6v`HT0+uH($kwcK^-O)Lgp#w-tPQG9sOsQ%5NT2X-5vb# z8lc}cmR*ED=HGejDD&?;2Dk!jB z=zLQFLq{StavGy+1U(3|hbE%k?UEXKKB}0=6RBd@)b>R1i+^&^K^Fup8<(DL;8^QW zzjWB*_N)#8oE&|mGZLC(A~U|`*t2pTdg0%d<=yShE6C%=&!(lFvEaw1u_uDtoCA`F zu0l^_bK)yV)BCeJ;v(laNiXvihATm8ZRz5}J_qPcWiXyiS(ZreVoS&}r^%a3rqkN& zkXXqa;o%$}e8xbN?q|oP!ebX|8t-qVp|9aw9k5vHT-{$_&Jc%N;mKk9CM#Z;!c(x5>GpINvR9wO;rwqF`DoV7aCA6B$>+ziCtNr;CBFw1jAYR% z-gw(Vd8dAExA3V{VXzB)zvke{xO;ytLz2mO6PnWA{DWgFWD_Ij4xoMnlnKbbV%!+P zdm{sz9P*g}iA5=p@GCajlKs7S#WVkXs)msGfs(9|eKOaR*RW%0RQ~D-A zpaMfcaC=wZ8ZdRw{qCfr94DYcAjx^u@H=u>q6g(4qAwGsH&h=w?dvrcW)_bMGtmQF zi2y3eMTwN+leSO{RL}AsIU26~$6|4gV;o#Z%wCX;!Ayl&hW;e*!tI6C`XF^-+5SRK z#d7M&Zp;1PoHegehl1}r+DG=d_mI!PE!WII+E&Nzz0TvglWnlzb*-iE(pC(SN+_gQ zyVZB#zXw>hj$4 zuckH@W`X^NX1seQ#1egXj0$vSj;+6e4Im{bSmkf%(i3e7_*CbL`Yvc~Ln1`ofGdA5S1dA2q<-DUn%G-VAm&mT)oEtU!eY%;Pc{XolB+v zyRvDT-h1xo;6rZ$^MWw6gdOBhLUY;x*wCmZq z&iMGI^D$F#RJ=D;wXb$F}z=nhI4}9gOdnXQUS4 z+ekndgJrP`FKA9aQD-KO#yml2{MJ3zoHdU;eAE5vrMK3c{pio8_Z~0o5f|;W;bQ~Q z-|zRre0EPAo8>x#hSo4p{RU!}Qww-9W);7@MpMM{?el&UpuPZ4J=)BP_xmdGCwDqa zO#`1Jbr-2b#DxT{G!dr58Ev}-E%<$;L&4wcM{oOc^2(2e|8cyz@_Z_Oe4?;)A&Z(A z51I{0Q34I=Y_%m!jxJ+jb*uKJ$G%eMg!wCfdc*+rzAxx>>IePZnoKojB{GV7qBpxg z5lQ9W$t|{dLLS(}+#q{w@*+CMF1V{e9Ji&MRr3Rn%$m*ven$@6;ya%ANwte&GNI(g zzCm}8I04{_9#HK4&J6yrvdru!zuKAmqqKjIs?~B@->sl#qJ;7>SYDmHpB+8H$$&W{>!f8r-)_8|f<&e?AE&g2VIf z#WB^A!VzFzwWVu4x?G{$4W(VWyThuB(#vA`4j?vhE0xAT+DLVy!`_u~Ym|DmuShDd zxbjB^<2LQ~u2!O$<6@n2Nxe~`q+{n>cW2cFA1uV^YT;N$6l&nZsz_HA5 z{#r;wo#uU7&J+G_bd0=J#Hbe}u1au3dN)pSJOfn>{4XL_B8b5MmJbv2_sv-BpXPC} z15J)UOS}ikM$d(#6^7PO=5`h(C6b8`PlUFc}zjPmG9bSFW%ig zrlI$#VLj|A(o7?~6ILO`T&D>t@-nH*`^_{_)H*`Pcy^ozs6^y2yeUC$k7o(eQ7OsawWv%KG z1lYmVV41-)MS?{&pAz0`v-EywLFw*}PcspAw@kZ3)DH43-@9Ssp7{a%KRpOOe@XR{ zg}2ml>;2j3P5mBg-`1|&BPXev;Y8c@VFfDM+ zMvLQ#DklVNt$)kCy=UJnr|9|c=hwDYl!*Z-D1-WeUn%?fT>E2Wn|yU~!}T2ydO6N) zC7?1;x`_cwN`AhSj5sed!O|x1LOj+=SpDCUa}lt$wRKso7FZsgknr~O`6(Wfq&fl8 z%y!u5z@*Kwo=I>1S2eUB8#hEVIdJdLyDhnTZJ%?PGxFN$)jK43VuqXDX)f?{C2reJ zIV`Y+IkTeTZGrSE)$$7+E=`ukV(aGjnn=zDWe?B!s<6LTVD594n0ZzeF<{-LdY9hu z^wvoFiNa-V-YX@Oeya$MmM>uyu;E-?5-oKoWPG2=g>8*e`5E?J{n}g4Yg6VKcdC@# zxVo>3*}SArL&B_Gu0Zv2mie&CPva#<`cbOeLv9?u-CKAA^$gX|0ItnQ6#>tO9(0Ti z?`K^x!;+L~C8tfEK?(ar)gjyldQ7wiFlxwG_yV#gv7tN)*BL8oYvJ-!5`QYL+IWQUfIFtpX}kSKBV zpcl{iE3iDltd7;2Dr-xlMG8}1oZy*DD&d1I zz5*|Gc1h0s_yrEey>c6BbAx7Xghpp>>17v-xVP?~uYo{M8s%ts)OHxE`E?Q{Op}z=Uc~x<9Bg z=aume1!l9#mu&9I8}fPfRRy`kxBvFr_aC|fD^`}3{{%xqdBpJ@owNSrPiyy<|LmczU~DlN8YeT2m>n!OP*_L}A$Q!TR*3 z#U!-k*+`>Te~!w~MttNOH(mxg^DD4TH1XpTO^LsvcJbFcbYb5E73Z&4sByR85J&ub zLi?-Usx@p>Tm8iS+sHdtKJL$%8TZ@vI=ahyJXv4oy36xc^o{2H^C_vbn^-i0=6Ams zI&(gIaBuXi!^UJ#MvAlV@_1PWCV3CGTMB(H*q>1PWli;b{q&)|Ys^wJW5nMHDTx0t zu^IWs;=AUxsrO*)O4qSR6?CuV>MZoxgOp3v?j$AID+D;foDb2u(0FA>ryM*IQt%?k|0OW`^Pcrbo%~HVj%F(Cp zlY;o}^rc3Hc$CD-K2LYvpxbfsyLEUKXyB0UUs<(|w8kD^y4pWKAzcZf>6*WyjfiBk zI=BqKd-Ara0|SXYX|g>Cv5bk-?_O+=_Dq{o(!y~dNSrXSf+bl-#M14K_v%!O7%A-PRXM_3tN6KT;EU#u{W~JOuzW$?Fi&rsluxQEL&!vrxE5ax( zs}r}e^j^N$AXNQKW8;C&(!_8NQQvK2h8Ejcf0=8~(a#(kv!h0pCO&I{pSp&y>~3yh zZ+9N+5q80e18*E7a?JiqQ!4HaL7(WB>j&*j@yO*xT5gfQd5kH^E8*p)Zvm zV~&eEvJ052m6gwXhjmXOfxv6jfF^rHEqlM`#Q0Y zKi+3im%Z9HPns>isx<6W=DIMM+VywR{qJ~>^!F+!_vz$UdA}^^`Af*e9vfxBc|!vc zhgRWHe`n(uKV`G;gq*a@pN-RPapyYMiwY60==8thu4SkxSTdX8ML4(++A?_6*R zZh|{I!QX!^%-5Xa{`<+_6B+O)aB_p1BqTK?I_1Le*$cpRp-A-f$6avu^%KIkCDRDu zjuG!jA4I}7-V*)bbV&8nb18cDXu*eB;_1a_cwYYaT#;IBhO!eDrR7zz#L!Uk!}^K? z3`+Wa#Xajg?f#HrZOcY|a&{QEe+lvkYpLzg?2x#kaoFU+ny2wuu_5E>ABu!p>L<8V z3>R!ugigMfT)5+R^=cMsqrzsM&G%Nl`|#wgN_T+R5ph;fELz>sX`Lmh0>n*pQFS7b z2IK++#;15bkxjUc7>Po-w@anz?+=(YL2A*L3J<$Usx5bQbv2=T6BGtZu}fqvL0r;| zjM#V%fx!wM1i%5M7r-|cT`mB}M4znoV{I6-V2b6FqkK{DNMVxfLJ*Kzr-bxBJp@~d z`aMg>Tg6uPKk}<0xA`^D3TDMB$K{*P*aUgqxLbE1`R&6^_Gj&_`)3V;8YmfeBdvtO zt%Ul{F>lxu*sma>^y*lcvz&i6*Bs+tOC>L+J>GVL>)9aq)nD%{I(9;WeH%Pj00iBi zpU^PNT?F1878A42$tnBku|+x-I@8d9E`Jv9lbm#$ZaXg{w1D69rN95N(>?sy?qHtF z!_r?0Xw_8#&iQE&9^cC*MT2pHzrA)KexS*AfuBQqaF>9X*2i zeoi$##$%njv|;v8mcvR`M@?5F+3mg=BJAdyc#kw5EfPsx`KPMoO~n1u$o#Zk{nzuV zh>yvNqN&7+5KX+&uepAz-h{*F6+7paq5(Pz<;dK^`R_*wV_{nAjp|s_eRcZmzpGl) zn?Khf%UGH%_rm<_AT$F6IeChpM}lr0HlF$z7aSa1gbR{O#YYaXY9Ct2a(6TQG;T#h zyGvS*+cPNgrW9CjI=J?vRP({Uj-|ZRO`(a`tg(Yk-g@W@H+JvshE$o{Wh(|lT`6*nklKzXY&Q+1>Y+53x4Gh;VHX7 zuT<8fwsA6Hf2-o7l%mNGJqlS#eX%z52_vKUJDOUv znelA@1cps)sz@&CPiUYVTy~L$pv!9{i)fy60;n^zb^b#u&006EXuc`H{s2_874Hsq zu9RthUnHYdG*ql-c!t}Si$+6F^M*I4k?4dH5CU(fG6#n<5i+&qVMD7gT+-LiuUu~; zc{AVI7RLLAu-PF*c)9TuJbmpnzkpT7kJ42>ZYBlBz zo6y!osa`hHKd8F=aT2$KEN@&U?GE)*$jy{vh9zSu*(#7e7VWCjnhHkRWk=oL9efJ; zgAdAf?QQu;Oqu`wblq9+E>rvaz38{41oVy%=GKV!!_A2}2M{$l$Q?$s(89%FYvov4LUoQsr^{xwXze_N^<X;|9Cmg;Y_e%pSmfA-f@8g@u{-&H$*o4YlF!)2WS=*b6rjMtu4)JPE zzK2XYy8jhIA2xoJ{;sT7%QfRzVj{`I^QQA$DCJqD(Xp=I%XlcebGKcYPGMYTyo1^N z{=wznn%|2)Y^QhK5jDar)!}a6P02w(_7Qk1vNRKilYQg&(w0@OOfu!>iXHw56592q2WtAMiz!-wy-9>CNpM>4J-Qvv zuz4;=jvV><`B13^IlQnkKsQ!!v$XKf>0a4TQk`J(cRLdU1g1=a#P|~Ah3F1JBPNHN zS|OTx4p{}npdN92yhOK+HjcHHear(5gdsMZx8*OKEZ<)C^$2&indb)585M-=dPsJV z+>TYyTjIrpwWBj{`fItwx=Q4X>;ZHFMpdrVifhbAu%*R2Mod2M78EeI8i}O#txgH2 zB&eV1c%i3kz)7#0@hl)vS~qxMqgPr7g)b`RJQFvsPxgE->?tAo|`*#zi!IOgaTg6=CqCvha z?Z+;+hO(R;^*ym^ud~@lL&rmgBgTCqJk%xKjl{OLzw*W)Nz24sIGW$d_AKr{Ai}4R z_Gg7^|0hbQL4Sh0VZOD1O*j#zFgnuIR2gYE9o&DtcyE^JOYe93mCui%If1jO4(};J z{~>y-@z6h|NI)FGl9Vp7F5F>V%L#wYTJR&u;|ZZnjtguIz>&B_UPv)S3s+el#HOjO z9VrplmI=BP1w$Mw^Ht)=L!#qB_^V_-X!JZQ;M$~aN(z(UR?C&q+a8^kZRDhMc(WL( zo_3IpjL9Cd%T4yGduLHJYQvLeZb3==IX1F~`vfJtvfXHX=olhm9tauj?ssMD4ju1z zO|tlH=Pf)bbvC@*`@~7bZbh4r(!zkWv2VWS)86rGY8{+MGCVfhIFA@mkjCkw5==N) z2qN79^EdkHAP*z0{YQ`L;0hr>A!ezpP(?8O$)buF3dP38PWCL`L~Fc{eoI<05;BWo z{qbMTN4P|kAvoQ{@rmRQ!bGOw^ONamxBrCib|tH-EKD1bkXXb*s{5+4)ZrQI*Vo## zbz@vg)UTl*J;#mER205@7nNr@!foI@5gybUcBI&Ic~Zyb#_6TgF#)yy>B&o-7I3QsTrFz(wDK@L!)!tgl|J~cMl6>rjzHbU^K~Oy5mo zZV#lJ4vIhaSS#(cwt(UC(twPnhZlB=Kb_b^O7a<3SH%(Wz+-x%R|Y-F6U5HJLH7R$ zFW&xka0-#@=g_1O6N~MPsw%zYEczY|4yetq@j9()O_j`fn2KhsVSVn|eSgkrxWxa8 z*TsmZ!UO34hS#yPdLwX91o~VpN5`$C;m1`X0=KZ2%*MHQlNidT!y{yFHD4DKtLYs* zwQ-YI=3!G4QwraKm>8>VG4}?G_8vbMq_DH0C1QY*W=Ywb!ek`F@6!5I+NiJ1BeJD7 ztVF5vr=aLm4l2Xy3no z*SRgXze7U3L^QRn$aND@kpUIALOIV0Cb*qDuVesDoz+QTvk1Oz~fsj6f1% z*t4^(32q1RLY>O|gLmXgAngw^MZ$Dua49xfmB8ghsRF)*zKB)7CxaB<1w&z5t|PP& zl1n=poKj>1wwIVU3S=iU#cGKLtoPjeG`(rh@TWTZcg7KEv+1R=eB;A`!WNz3*1P{u zeKHR%=krokwYDd`-vY~iPse=gFCjmZhKx<)Lue2BzLK?C&%qfW&J_FX*gg}6e@V??z9Z)SOh= z+>kg392-R*pXn2GHeSC85wQChn?F#MH%W0QQ_a7=$F=c zrH8GG!?E)ngu}sJC;$O;2%5>Pzv6I0;oI7fS() zaHPazYp9XF?rRRaE!ryuPeoAH+ftjQkJuGF?D&?y>GP|NxmKP6HqV&Y+`3;}Pms9d zXSm*mBDTM$OwIHQWA}>*u5$@U$vyWbc6n@liBpx<{yH zBq7*$t9JA%V8@}Hf7c!_G4KP$g3)JfhvyH!qAzG4e9aMI z^dpk8-5+CyZ11Wy+k#5RG^W3QQwe{hIyGUI-`$@7K;y1>iOE`1x^EQPg?uxl4dXri z&d%^L!&r6e_K>=`Y};EQWA0dBMN6H1RJC(UrE5w8^3!hI2y#)5>YIK2)mld(EKEA;Yne9Gqq zPJX6Xv-B((9kWU)gdy8V{#wiQ(bL<;#6O{%(%!_gv8YXdTSax~@LvKNdzesz20i$C!&kF(%SCb<@y2-YReNSM) zANm)$E9{H1Yw!#b|2PtM4HU%Iz3ac>%9k0(?jfNVCmyaxrpg-h;WRf@0~j}K+H^HP zKi}gSGcX>qU4v%c2h2mc?5l&eL&chgpO-XB=6(Hcq%i*_khb37`0IiXH(9Gj@0y4m z%2Mx?&=qSMo#+i$er)~aJdbjQblxVq9d~!Pw@o#0FiGj|Yf|Vb=Cq}{O&l$}*;&u+ z8mrtd|3#ElA#C_cJ8RS~>(ie^Z`?7}e{7_A`Os<-^!{_{P9jPq+)>buJW8^)Mb+V7 z3fFmJ{z@G?aOTVoJduoCJFvd#jXk?h5Su|zRS6(S8ZNJIpN%UC_nd$jc)B9u;3+kb?M=pyEHxP?QYx}E{w$%mm+0B;PiaY5~*A26mEQ@WoGFc#{^VP9i zKp27jBlf1%61 z0sjKxVML1l%pp#@*bLtTtk4Cdo@DU&%Rg^O?(4(@h*}dc=!pSKOG^NcAN0LpQBlnq zuFkwy;f47KJ}p>H-Wc&Zr==<@7Gbm^B0-+EX5CP8Cwuoc)~cgd2Ic9ws>HS&Q&)Y* zOmF3Jsh~qbx2!zBR@^>wbT})PLEZgPL{{Hv7Z>>*-_GfDU!TpV44>n%ZdHDhU?^7* zvL!3lOdy9QcEj^k=YxaG)>NM{cqog}2Iu$Q?B`v;{x7X)C6X!T%yglj|pQY`&_TsSBk$ah>x zcR4E*z}ABYVz~N>$BK@kntq1g*4UTuW zg^Bvr_J|aeJUGH)R95ecZm%$4>@Gh|HxcLe zHP?mj|-1Ol4|%dSqOjwy`ls^hx*cPt__h z6Ckk)*DDbTjCB?kDttj6Q8Mmqm>)+_D4P=xU13QMR@0yBHhQo8!&A55b@Ib7Lcz<| zT_ih`XHrF1R9%ku6O4QYlCG}*0Jz&$dIE{*uOR5zjXRoPKKzP@`>F0`1lp8mrV8Jlo7x}U?;G#$(9J({x^3BRpW$w~JL?p9=`0l$LPooc2f?2DlC zL|uAQeTtsHIGNaV??v~5p72a&$Zx<=po^14PoSYzM;dnHsSN9vyY(qM?Kd>PfcvX! zYg6Fk5A*0yh0h{TMB}dh8h4My<&ii4RDMdZ#yiP=*(7npk%N=x%U1pWDyiDWx~an@ z9*z!0!yq^rSH{D`hg1Q8j;t%o3qNK)3=*$CjF<%+^4TYXmU?>3Xy51y1^Y3z)q!Y! z4wj=U=~U0HzcV}+xkH0mB`jK$m~TC^IC5yR^a3CSIjgq@8XwN+Pn&uBbN=R#wRI31i;G}b0eb4zKL`L)dt zrlh1eC7kR{i?YB$APZwcwnuk~yUL|}4YBryLp>c6w_u_wdD6inR)V91MNHP=oJ+7z zN6(mO1TqzxuDO}c)9}quMQ#t_zMhzy=)3;c&ID4grl+Ud+^c6yrj1E``H%i!g9m;p z5EhA*FIjGa(f?S%>HP~ovh~E;&>3D^Rm?4rxL$29cHe^gt+}P8Kw)o_6bXHCauNqf zrc=do&ABVY?0bWxm5SzjJZM4(@8nnY-&@io0s4~$CmT=$pVmi9^^17J zJsctvSFiEd^CW@Co=VG<_{H9VIz2YXs~Pyh?hup^a?GJ4ca&_Y4iD9stpd9EUniXw z3TvnFf5vjqV>vXfT2u8pOiepD|J2ce^qmW~d30H7R-H0n(UY5R(M}Lca8#ihB+^r^ z>{lBM3h@0k-}^iF#LHzQ(i}}%FqqpCz_$o$0}-eMP@A)u$YtS)O!BH>5cCMF9RA6o z%|Ey>gaYT8k@VYj5t5croG#dF-S_X66)OMrKMK{#hJdsc)dc6l4XXzQYp@F??JF@< z0ObkN)dmJ#s^s1V&h|=tEI(&bF$d@l%5iwUjReUg_!fgiSep)o09`o zij1t$u^~IdHe|)p3pX3ow9qnJIc+(j>-)~bFrSf*QiMbED$rpcWyQM36*4+ax%J)? z1tpoPfz%F(uNJ!vkY``35oVbF{bfbVvRG=iQIY?s2F8hyr2Wf_t>4Si@HtUOw>sV$ z`{*SO2(h4@!u5SygH%|oc5y7%_^=e)86FX4$B13lmnNh3#T)59HdSw1b%{A3HIfqj z2I-_vi1{&MK>L-czF4KAM)Tnb2dp;zkMq#D;q;LSl65AlQ;cV#;8iVVLbbV z3gTO~1pirZS0ZLesD7gt?%_G?!P*cF0BDQk`XH2sw{nBBa(r#A4$e|JRGe^(q4|Im za1$bifo)Mg%)m|}T69Sc8@p0nuw&%itfq!AYpFY23raFjw-R%9vQD&p!~QISbT|LL ztIb}y@RCe*_^}5YmtMfEJO%ZZ$9Rf=+iI{aV3>kIW-*`^-Uz0AXCZBsB>zU(tdVVv zqUr2zSVH6XMX@gEV5JG?>uYas9~c-23lDFwUq1-GAF;nrIRC>In=ii+(Wpj=EJ4Gj z6_J$iE9Qb8Q-;UaFtp^lYYVK{QJVbR4$W86n2gRdTnR(+_)Oj%d8u|qz~Smj`Ipv? zx0VE$@9hx#RuJbi%TeXvrC7iwJOIW>`VsrI)Oo@cYO-@rcvi%e_lzC$cxbR;k5gSy zzV)l@UDnatV)AKMS6=|$M|1P?ys5t9dRn5-fBmp0Y--CJ%6rkMXZF8U7!e~%2H1;Z z8~yB)Yv|CkqpcsYxEO@BWIt@*wjKBRx*UOE2jdbyo+S=6#M2WTc{wP)O3oGR8SRGX znAp%m9-NMCQH#!Kv3Qzo1^38?$=Dsm`JPg@>BD`E8@8h+uOyi;#C+t#nG60yTK`=c z-GT@eq96Ii?s2^_yV8a-q6{IT{f~8sh2oL8$?rZ#F^G{%?2Typ9kjAKoivp$B3riH zO-b=botLSEzu=AAw|Rk3czv4;N*Me~0+RuyOtzoe@`!ImT~oc=<&=T*DJQMjG_-Ea zO3mjx-wSP|CiWMbGxK@q0fl^35b^L{X2)KJVa8l$Oqd@8Sob=tf7M(sp|g?o%Lq;9 zz|7UPyPKvDA8z={i~4CU+m+1Jbn3@f)ALd@JND0ena}(hOkvhMo;{aK({+KAVK?>o4Qb@cRvB_$)#AhNBby_b;@6cTd& zi(SX9SP<2);VOsEMe%Pv`i3hev^!|l&EOrG`0JncB;(y5fGpk#rym@N3xTsz{zI&b zqzA&8Ky6_}`VEm~h?(REqHk<_0yB4F%0XV>NWwnH5;^9%&BxiaL=l%OA}*fx-~l5% zj`#*==RqMNqcq+L-19z&u)+T971^Ny&mbP$7$bJ#%1%jgQvCK<@sW>ciZ!FKw4;RA zHs*AOh7FDd=G=$V+A}YAWd+G>w4)RZF1dS+c7x+4ownfo1DiwI`fa#{&u^hO8vkJy z1TN&L_U`cOV;%mZ{Mly=wAQ!@)TG4~U#Xk0mrT;R3rSze#e1&ak1(%^oHQjZt-_nq z{c3zx66cS|$H=cNDr%15k=m2KCAF=S6_a-QElm3^3S-+FV?>&4to-ZM+wk+XPygNthQCB961+S$(N_d=ZmGb9W_>E6k z5<|7qY8w^oddZK2JCKim=7%fqCXRAmX*+hZRl);_lFY0y#1DF7H8AwISeCs(j{{o> zWlGv>6G^7B6Zmnv;6Vz!)D`2IM@ua1ha|@^0zL7}@>Nb@$ ztJ^RT;QZ8iAh+W8m|Yp;sB6i)mH9v2l9C3J9Cae)XzNE4CIuM0ey26CUdTZONQR{?BOt<$u?yvN64{^N3{p(^P zu|LP zr6vxdM>lvSV-fMFCYq}hFauF*=ZIx$XCD`ghRZM!P>_@#vaiEx5MzYd`Zv=`tIb+H zLCUDEs~be2t&pO&%_w%9A-iW1r7-5+{uQ6@6&L^zIbPx6ild^h36w<`co3xBWC^*m zXLQ?tBVca#u_KL$i79M#L4kjEbI_A0u?UbuB$TU75?m^{Ncl$iOyAUZuLZrsbyIDK z&B4crjKI587a(Q?i;IsB!4FlzH6w=SrXStKCjuu+OFw>+ut${XIDZKCBv`qsway!} zoj4o0%lcC+{ndH6rB2 zT}SNjJMD+0KNt97rt{6q))$?)7`Y;-kQqN2PK&Fmt-U@g#b%7Bo^%&cQ7jBsO!zGS z0T^e3TK4@P$n)5<5c2%Zn?f}a!N<&JcclTbhMt@(i;&TLi9`^gutr_K&e+)42-^?9 zUU>+ThyQ>jku~G2HGcYb{KCRFG8J^Y98$YYq^rc?8F4l%h#nPuod9H*!19e5-g~`$ za(XdWf7;vIb3wQO#`$Lmyrq~Rslr6_1Z=!mgpDH&jD}L^wfDCo_o#oBW)<5ey#Zt% z$YMh03|O;jjD~Fdz$a4fROY6prV`Hp<+y%1AcTZ?25|Cn|2d#jt-29S zbfo33Yn6@m9Czw-rb&%c(SE<{Q){!a)0nSctxU9Fs}$xR0moac74}9}^b>Rr+n(Cf zyRu^i9eZBPcbrL$&o-$mW|wtd>bqHi(SS$s5($$6u7>E!0unFY6in(;HqbIxl*w+| zYcD)~-`N`N_LySW0m+;mdo`L^h8tx-{e;-B-Ra!**lhQ+Lo0xyO8G4Pt9QO<_6ZhZ zau8&p-(Ii2h@;{G!3rab>gWL+_0Krr4>15WZ&ab7)CeYTCxHJdq0P#Y`E5_BKM@g?|-OL7w#}CpM zzdR5PrX?!_c&(2wg`3EBk*&{p9MvxUhFq?)irRJE# zIqBdfzLii<@J%mr#ds^11};4N38Shqm=HBfpVbI-Y8D6^=4th&iS7IN&wc-|CxZC= z!&Wk=tu@US>WM^W$e155J`so1>Ms3;a|%f!gyA^*%M*`Y%auvMtbWxkxQ^mtaJq4< z_-%n>fm62@MimUdP=o?4CGQ)z)XK83`r2PVP@`h8m>}fY$L`@997rHp4)MhH%l<`)#Rs~v|Fw8aZT=J|1Oq< z!8=j{H}29!oA(t)NaUM!zSrPqjhgIuZrIUEX-7xn#^tUrYSJtt^GsJt>e)xfj%+^L zdAInXf%>R`jJP@K5|x^5yStU0&DM3gOeht$7FcWX=bb)tQ#-^=GsN`D-BGmi@$lp# ziv^;@y8Yn6gLC?MmB{uaM;BK&!aO^9WnsqA=KVY2$63?GH7&7ZBwIC%tWNG-FZ^i8 zLB$Nfdw#A@dFB^Gjx129q2kcq@!p3cqFc6DP7gJZHNw{CXHnND(G8EbehV&VbG&*K z#vhxRb zC-l{Mj}?B%Zw*zRIgB-%U)A8ICaE09IKpa&fcC813o*6iMR9I^mc8QhMH;kPVvf)E zNo#YRizL>S*V~~oI-MQ#z(t-ImQo}F;t5KXCpeUZWCMcDjSyF}n7;!gnH5JR$l@AJ zESJQIT{l)*?~VB!Tri*(W6@y-)a`-LCQ;FhxKr&VA-wVtR}pl2LZXU@t$`-IMQ~Cu z-~HsL4=8W=;sml{x!Mmu?`bk%zYye^Fp@oS#NzEkvwveGMgks=1|qzz!Zs^%o;eRh zfVpG+J};u_2^(a?t1b)RW!2LrkpfRoO<2eTo~Djyino zYm@or!3T|AFT;t?xr%OQ1I){rFDiy!2BUmifsIf9_zC^fR0p3d;Q`~jD^91g9jJG_ z(~)tfCrjOyp}TqFyBKlcBV(|5(EuxbmAe4AU@&eg$4d>4aS!-*a@z%UV47C+7{zf6 zVC32ZfC%@e3&p+ie_L9^fr8LKfH|T$UjJ2IFL0sZNaAo_ZZ2`DBV}rF|9s5z<67jF zpOg+c2HYvikk_?#`=Y4<*sXZ596NI+J}S?4m62%4XHMF8Ybg6JWGgRyJ5)!?6{5%J z&01Y}SD3W|o*_1r@O7u5#)zEW*6Kb?SkPD~o$Yd2fpnLgRO(V9@fFs)T?M8@ebNXK zZ>rU8V8UI3{hCGTyOx$S#CQhCTiZz@TZL1NfRIq?yLe+zkbmUbyzj%3$C~7qV~EY7 zD~o>9UmgW33N zj+j!sx2#7;t|TW_5#SQ50E;6W1&);kNm`=B1`0S3+|$~t*%Jm~j*?b1VwFic50dyy zL_T1Vg`*i#?s<-Z9cy@8OB*;?q;&~Lng}V850j&Bu9M2%=ge8!t1rPh(5StQEbr(9 z5?zwDlaG6kJ^qQHtkvKef3E+wDqLfcuht=)?!Ss`VFMRtGLJ9ZKf=lOo=DM0Lbq-B zvH8DP15RC4P>qRf-b}Jea$Ha$$$AkR29h-kY@B#G^!F6Nxf@>xl#oL8;Jk=li;x=L zywQ3>^T78t$!8(R35Z-6l0L%B@_oq%i&so*NAEo>msm(iSuLQvk*93>wRf?ycJf~~ z`V+&C>y3HRdfEbPHF&H|*%>;6#M--anVQN^f6qawad_zKNOe$1&g+oqtZCkAxWK`1 z!=R}}XB+l9CtA(9RF)Y>gb(e>Skg11z5W`R_%r3mVO2Lk*Hu9C6cytz1z_C_5=IjK)BXfDg*fAc zXLEVQYGKUF0AVY?<%Jqv1Er0GGa&RZ*q3iV35LYd5%9Fo7gvttLej*TGFe4OMU|V& zBxkDsR?$a2Aqsok1t2C(DT&nx@G`r?a^4lAMdh;C29BDco!KLkxZfqgd4edI{sq=P zPX;2iIaPe>HWfdGHWa=8#+k?FoN}RMko&D!b<5qVbG_qPX-1mT-qkPg(7i&FV$u3; zIs%yLxO#aJo$xH4g8r9)}l_80u>krqmcH@%l3vLWoYrpaEJUnz`0$FHW{sQ==~ z7nF2L=_>FA3SF9QlJ=7%-Gs@N9ZcbO%=2V)m};)mGLB&D*-SkM+!D|9JiYrw_6;bJUBpb{==v z$Oo0@iot%`!{KPFJ>#1DhxYXk*c?(Xqm7IHXwLBuA)VE_KzXt#jXQ-6t{OQ^)2h-@VQtpj1h*-&tlA9xn@ zi|U0>TU1PJ0L^J-!5BwB{Kg<{AWR=P(}G4dFdo1XKlFme^z_VWuMhW!%#ydpqSX8x z*1vFV`Hs7N6__DN1XcN~xQ5@qf#s4Mi6i+o`^832AK?h?pCF1^oIeRv4&%axhB_V6 ziJ}XHO;o>JPUmeIN0U64v1h9rPBtRG7Gv_FzPR&rsiQ|p!|&PeJ`YR_Ls9BNK$flk zCE4IDvTaiTbQ8}w@{el%A#aRIb{1e1KJCFT1VS2=iL#-2tjVX4WiT*cMM4g+$gKWv zZbwi^CXxdQW`@@ZbXqujgr+1`^Eu+OU!SvFiw#-sf9hJn%&UPCOtv>KPEc&@@JFw& zSEdl8neYI;K8xZ==skLVJD@oJmhr{rX)!~|TnWZ@$NFghopbBsjxDPQo2d`5jWr(K z6>THh7{CQmL9vWY2^Q|u^1)RnxQy48bDLfhH+39Mh5b@%gC<~!lb?mtcCNi5FzUZj zz3f3kh`nNXi%z%x3x*TijiXN$aYN7l6cJdFU{)DKdxl4~@czDhYvG=!2LznZjkK*Y zay(=(XxwkRo{|It1BnPH^2}}9^gATjZl&2ykM|~`YPdG<-?!h`7;9X5fXdX0<(GGp zI|1RDAR_G~9r}8uC)z!KiKO1Ue}6H02-gm*S+H7$QwoVUA_Nft)!(0PR9pA;yUFp# zyN%xuKTGaxB5VJh3&Z;UW#(WPx|@mBBtsidLh6xfv19_0VzT?006I{SU6v~rWhfVV zxpzGM`E&4>rxIa%EKd?>wmM1vIHKak*Raq5tIuMm5|t5IvmMGmdinvre1t)hIe2|J zp!g))a)>%UjR)!V7_kyueB(tvXnhmRQBl^RZ7U-M~B!Dgp2h z@3^ILQBmwTQ=B`jG#uoG8(cvH2_spq#O&!O=7Ry;ZG-rUUl;J)6Z;)~Ki*VBYgpgJ zTE;JB`kcP(Qt6Q@TGpu-J_4_OqU^X{HIt?~hkNybpq8Qp?J6h!fdGRyU{riay4x>_ z&G#-yp$ZNGp3hf|;*y{1Fxy}INh<&F#Q_@nLf%8q0 z$p-I?N>39zpVbi#l4?a>=>E_W7{J?DAS%~&P!GUikS)dP)TzkGNL4SdALoBRM+^H$ zVu&T@11eh`8E*k%Jb{(v*ZGA>@6BrKwrt(%_-(Rfk9+djdUw9MsVXYyyECv8GBN2U5LHU#Hk(?!|BgLp1gPydPcQ1L4YyYt8r!OE>NL+85y}Kz$K2){z z`~s)^8LQFBmh;9<`E3!yTB0P)8qv((2^*~6N9Fi~Vabk!cDK~nnr)@Gi5M48+Ns~` zcadg9P)yf-y4jul<;o2gF^p2k8qz?7y~F`Ntmuk<+cq3+=wN<8+!4_W=Vhjgsy)0D z^~&80HxeJ&M&c`-pSh;NIm5&TEUgkT3PIT;dRVS^0kI^gk2X_(YNx=85%+s{ny-v8#!W5W7@{-Emy1IcGYIrByh$<4$E()^{KbHH^!V) z@42y$XOpz9Jx9+^lB>%k+w}sfkl8kh83f62BRco-=An8ird0L-^Y_K##QmB}+U|fu z_2iIBQCchd2L;q86(lRDVd==yew>5p*j_y)sYi~AGP9_9_?fWJ*a&iEyKkhgeR8@9 z8O!&M{&;YKA{~7|S@JyFXMxAZgvL9!z5XAv7S<7)#l$y|ucHh>L&-eCY1HbDCum#E zb%uuJi&7V4%~SEf=mIGZ=GdD2=usFoPPVcFS>t%hYRC)j0R&`0u4(e|(({gAy0XA$ zKnv6dZo)}KbCEI+GFbkc*%wT2sOR9IR*!pblfklq(7Z|ZAo0?9gq{iqO4}hl^bKEl zOYK0^(v*`{kcEWIy7fP(9zXDb!>VE*k2W#BvQ@iu>nn4w(2iDD`VJ=lZWe}Ztmo9= zpp#W0Q!7{czN19V^MSmP`lXO8_kz6-_o=EjNwX$P8>w?^I?8rXxCD3^lkCI!9iH-N zvW3&-2=OSVIy>;lf(-8#$I{!P(>^IEqwk5>Qq&{bXFq_ngbbtz5_0-^6mQqh8R=N5 zBnM+T^#0{#><-`rxZyb4X7~lsi~ni4RI#~VbMuS z0%BsZU50%X!u7~Ap{c8T)A>3}w*%JOi}{S=JHb{EyxtzHe);ASA}t_pjo594I$pkO z{@P}42X{}iyD~?|^mm`ds7nwDh*@*a$c)lMs(~*kzvyvZWJl|@x`PTdfIWdZzKlyQ zdxYSxmcM}5oOpX%o~lKctesYp&BJqA_W3$GtXR*#AkV4#INy=`E&|^8m;S6>6ZJ>VtqKJpnC97Z? z21OAGktEdmMb;chNnpDtJukRkOmRH?>wsZXo`&vUfjyO`%*sz`Yc2P#{_NI!tiMg> z9^uVNxWV;5AVb+eHB`P>Y9$p|49D+Izk<5%Kb8 zFo_!-JgAPQ$h6SuwY%A|3k&Z2xxSuLR9W&Ok+nsjFOby>-n!uBmLx}qtZJc)wCsyX z!`kx+-ddyyzXy8toYvAmFcKqNnqK@Oa9bm-+m8>&@wT;4v8vS298mcfLSzC?a@gqP zv?SaZ>ZUAeBK8#y5-$ddeYW}M=8T+K`i7i_dTaaSMD#CcV+Z zzn&&|Yef-L3aWTbb#)&u-|J1+{fFXEiNdNd=3~ujg!Cx<8i~%TgbPfNtKjox5|Abs_xPCNej@lDjr*)<1Nx;k%K3^pXclCllJC(JPNsH%veB ztD9zf6p+EeRtxj+b(aQP_EJ6}B+vv37s#KHE$Dm{7IDM9FwljvA%<(2Q( zRN#m|y95GB0TNb(?cm7=Me;w|)0AD=(C(;>|C!3M2-<)%p&CrXPX=Sys(p-|t!J*Q9or`SMjwZtf56JeJYXZK`LYso5cPt6+te=pMbW zFf=El8ge;bU8HzAC8+D$0@}l?Qd;It>$c9Y&K)?X<#+9ZeOmUSuZy5%38--^56S;y zfT0arA58!=47yDdM>r01WDkC5W6mD@Vn?MlFl(i7G5vaz$9f_A47QChcSf#NZ$naDG$B|p;|>&w*y)zaA2G7df=HiY(GJ~xNzZ{qfySOA?+=b zq?dPtP?K(x6}JG1+AH>K%2y-SUk9J~PTXEyB zztRz=PQlPj#P8ti6C;0=gmsv+imbIG#+@+y5FSAnEaJk2727q%I{IMoLg*eC^GFC4 z@qmRnIqZ7Ye(b!TA=0a+)xqDmBR9v#@0@<&$e7Th7W>g8rY0#}M%%Kkjg{^z6_~;U z)Sp)Qsk0Yz*dCrX^?P!4PN5^9oomXmfzsX*@hPa8VmtSh>%7uOX_?yr``>IFy^^Ea z2ht?4d7H*dd5TB^_({ib#ZpnuNwBTMho?y$#J%q->8F1(+5S(#EvZMupg&l=SVJW* z*h3L%%Y*`j4TIvMbtc*iaAIs-eSZI1h}~{UM^4;|_2*8=zC955+!o>lcu@Kzp7#JH zj)WjXZTT$nln`r3_Q>M87q|evg9fRoj5M7E4P>sdn7_wZc4F{tJXxj_^C$xme=K;? zE2?eUrU-xoDbJ}<*S-0oc~76-25F;mCS01yBA~1f^9vRsa)jE^l&-%?Qc}`mVV;$B z5oaIY=q*Ra6nvJeNAo=@9@``W7R*K?U>zd_ng&pvN?YWu1gKl9r&PyH!g!uTFh>m{YPi#gsoxjh`hfwYu|8lz@$_ z!8?Q9?vE~o@st~hGhDGfoI*!@SXdrKeBh`!pti?3VRAz0^o@ruSyF++(FV1qwzSnuGt3toH!t zx^4f*+oe0Av}{FENLC0f38Bo&sLbpld#k9Bj7rE}p%ls<$%;s__ewU|r1-rq_w#&z zzwiHl9M63m_whVE`uM!B_j#S?Yn@YQ+g`vz=ymB(y*&T!Gw;a~7qU8N5sG4VpB8B4 zKC?n(x5@1W??bb;*o1_-LRVLo?-KhcFRlBtSmWxeRq!@?LWEFn5Zy#xapR+_^4WAc zrko}kWK0lyA$E+}f%7Bq?#ZQ=8t4*Art_{`Qoj2l`l9l+iNEkQi%QQ~k{3GqvLgl< z!kO6NdAn}k^EO8 zKG10=cQZU1=bv+3{RI5AQD8q&B?==> zDeaJmIQm`Xkor5V(AFD*&nWxr+dc?~KYs8a#b|?*3?6hq@5R#kzD&Ev#&o9bXQ`F{ zvvq97T%FJ9GH+e8t}wsD$NJg!D}9DyhK5%>pQyNs4$16W;5xa6i3j;Udy?YQTsW+w zT0ir3L~lF`u$@~<9U^H+oPTWABUI-F&iH9260k9g-qB-;aQySOr`H3e;rSxc(?DBj0vWlD{ zh^T?83lpg&9E2qQHT4+o`RAKmfh9mE>AmsL{DYc-K0p~yI;84Wo}`)4ms^13=q0^A zLOObZAM}wdr&HwuI`H-=4f`S2lVC|3a9ymQ z8Kv2L&4(>wMkUXE)8#H?taqf=DiHnQ`Adu=Z9}F~q`Ur=g=)5K;sp~Ts(poRnOxbv z@nhHBxPJ)!(kk8rKZ9gz>XMTJAm7F{e^4|74SCoJTcHy}Cuz4HgGOX{WaQYk^-b23i%z(|3Bhs)Tc`Sjzq;g0 zL6D4`fPeQ-)>@0t*&oQ!`wjIu>F=N=GM(4`rbxCM6XhQQoo@g%O)XPkb-1gVtaePN zZJS9+y8Ig?|-@xlQ)Q1^gnQZ$5szs`JHVU?$W1>Fz{%C zIjoitIA@zD{Vg&ZDpDduA8tt24KfNTBZ4^Oz&ECh5MMypfxX+N2sLC_yC1{atPsp* z!-?b!k`;hiF*&uH&el( z0js-s8J*~=+@RzGw_q*jM{5R#@*?GH-VI+qNLxBRX`-RjV`32AcKz-~zT`1ko0;xY z2F?E{5{$pG`+qQg`weIlZX&K|$157f9pV;qNMDNuYJ|xlwIPESG#Fxw$PNLb)!^Nx z`y@g+!?fiS3H*kzkgPzn_CdN82wZ*OwK!i=67yS9pAb<#Mq&-kYcLz=>$AgI*9}xS z9^+p1$O6IDNUkN+`RyM3yTe=sT+mN=UH>S|6iYhf$WH&z_amCn^X3 z8;&^g%i1x#90xJ@#0O2uCoYs))~d?lpj2td8=KQB^@5?I^JX4{Nt<{lDkcc zt?t@Yo-?ZT-1BZ5Gw_k9*{4Es+-Wm;jf*vQmyzr{{0P2-Jruuk%=KDIk%B70?{*zA z)t`}0!w3X-F)yR^SF;8{SrHUbO z2I7KuSGU9`qTdOCW*1q$6_u64zY6H%XaR~*kwOGUFSekdp!5a8!N}8Cd<_CfNfr*- zTV!SVG~(|vL=8~rNrnqx$imQ@94u)O(wq&W8}do>fJ5m|66X~ai|}0%Y7#-jap-^v zY8R*=`Y2Vm+_G8w!CFR>YM;)Olo6y)9w!QfY8N5(oL_j`2`Djcd%*ytK0%?#q9a|( zU|2S~g{i#v95twX%DLq9S61C%lx9&T4OYGqypZqf6bRe2aeq%`{LSoDxm9Wy7Y0P^`;0)L zw)l2>b7@M#+oLp77;i1`SAl}k4Wh)FE^OknGTe6*GP+uOo`MA(BA8)GJK%sMR8pNV zN-}?GUGY4W~Wx5Meu?VBsd!1bBX4WL2Yj6$+f z=yStJn+Fh`EK-|jG85QE^&lGw1B*z2F7ahNp;LYsp8v>uRK1|A$KJQ?h)!N5rCSdY zb2pkUW3+_2So7>eWZf9inZpN4ZBo`C^A1o!Ie~g1X4sSa`e=s1;W%M5m-g_Zg=iHZ za2WJSd(=n(x#3oabvSGUNiz;F9w%}Sv@B25#;rREB`H}ZiP7>L?_SemUwOCjWOc72 zVhFT@xlM8FJetD>>1A&)J{Y3UmA0R~|4tF0g4pJwtobYc?zHf(aU3 zcgdqWsLeE*K4@pe!%{DAE#ODDvaz~o)o7!T-Y7?RuK+>arPVr@$LmoJ$Iv755gDKpKQGYTrOpMEQW)YGR7-DC?cUd5a9S7*f zr#9w1aNdSqM*@-W(Y$?{!Jr}Y+4JXVXd{@Mew_xTuttYFKt7Z|X05_Q(=KNT=BP_s z+(>aEvZ#pVnFPk1F~bGelOps17GX(?SzEO^(c}Nbn(gL^O9oJryJA_;!pd^t%F*3_ zd^$IM#7t1xo_&YhEpsCocZ*nV<^p!wkPG`U_Rvj-{;oaJmg)ki!Qhj=1Sxv3N4Yu0(HO(o2I{6#ORG%df5(TOZ}s z`31TD6+J92R< zqcp2e^Pf6SNhQn@M4CXd3!uKfL`?e(8&4c9dr8ii7IQzP^v71% zF?@ghrE~JC`>X}qE~T73b%C`_Y%h?1`JKfAeVvb0}^VPyV2WQc#=R}LogKNRwSd(=-2RRY!>sZndn0_ zSU2lVq;i zWXBtk?5U=VO=U|zR^l%_dJ983P(`<(YYvYv0vjydiC+6@&?T~Cy~j68=G#`BVS?miAo;{@quLUJN0FJJAl`EtyY6+jdOtbwxak;lnUPrmbD zj!E+g6Nm0edP=Q~S%%gOMvd|Op{>_EwIP$qFbl-Zh-8fgXguUt{Ps><^e`lS3bNAv zWrQ9|usTSY*D)R-Lmm-L0x9;@GPvqr7HsIJ+%#+OCYS_alNE)qq?HYLb6M{Z&uR;$ zqg8u-r3kayx6w2*$>Nn7t(*NjZd+(#(E_n)u*e+)?=R`;=@pQ-CbZFmD5X*>X7SPy*8v;CD##pmHl!9zG!|e}3%?7C< zmH5mhqHh9ARf3z81RcIJv7a`)F$vf0Pux_-yA9g|4VzOza3Nr(B!#~gbh#1D=O1~U^q+q;G&d2u@%sn=feR7@ zv=rElVU{M{3<(pel<{DdSEo3>NFSo-`NDSw8plo2Y4GtzA1)S1@L{#~tOTlY=4`M@+NJD0d zj-1)7lm`g{R>k{8tXYPF&%Sf&Xgi4MoJwSHLqLv*-;li*WPtT*K{Up%*7Zk?_y=B) zdMxTr{9olvTAn=%!fe)8GQYk_xGf$O*>S3wGU}JkX>mjb$T7R{pKj%S_DT6smuE|e zxaYHR7~P6DOxC-QywclzM~vza?qtNf|@er(Pw3z5#!4`o2lI{?&tA@MX@HiV4n&Wk-ky$M6YNyMIOcC5P% zeU|e^a!7~^*=EXM*X=&Gm02p3<0woQWYprqT5V3ql0W+&Ao{QQ#beDK5PoDsEnx?u zM`zmiYxs zL$D{K={?c8|GU?}TFWRaufe~9hq0<|#J#bGJLu}8aXn*)Q$1vr3JDj+DC4WOo6>x) z{@!JBv`7D`nTDR==;|UN1e3Tow4SH`Z*J=W z!2_XC>kK+H6{hI8SFL|?fHUrTP#2RMBbAc!hiAWA8M74jYAFwTN;P(`Zfz#q9aKQ6 z2=aEZ7%|)HTgsX^X2LS`53Scb#{E!Lh;5GJG;9|%*lVCmZ}6vWlcTz~uEq`#zZR8? zj(IU}QnWdm;f(+8yrF7Myiuuh_@hs%JR|h$gRz@R8%(JaL=c2}{sZYle0R zFHHLz#|9ec+h~{==-Y9e2s7IoQdHDI{TZ=IkG0Od(eQZl>4uy3@C@^}2&|O3k1&pb z)ZK?!TM=7^@kTuavCg%0FmT6L3;{VeXiO0Aa*rPzjgiV4XaaRX+PLBj>6^H zx2TtYo77LAf_Msjm@d}Wn0DsvldlWO-y?{M<84qURZ8AU!uq|$#ymE5_OZ*!ItU9X z{N3<&HUACigQ?r>ZES>a!I5VHBWF^>(r&XNQ_&}!%JIK&q{`RDQMc@g06=7x*-?fV zz;v69vN7&e;E%j|8+ax{Ib>3m^^F?TcHlYW5!c*7BtEN<;oloI)Y)Wfm= zkoGh7Z^mdf-~_s?Wv!ZiOtCio1!eI^6#RKe??`$U%tO^wH^wHuM%od?RL2#P-q){5 zDkhP-zShi6Yp_6m8QE`)wH5w>fd`7@YPDe&C93>B5@`qVYgWo9WjfEH1MJq+iNP?W z6n-%;#H1HKpmbeX!E9FOOc9+x_1UJ66Z0hEfj}(3kjV0`rG}_IG;S?=z_=?Ce)cm} zMyux#08#Ffbd%=ADv=*YPCtI^Sf{)$SW!GvkpJPBr>#k~z^`DJb!tix@LAD))&v`d zyYY!L=_|3Jl1R7!8(Em`sZop+2$-e){#tIGLz9H(v!1E4Q}Hcv?p4IfsVtMPanIPa zSit!pP%buf_?QLG++7o0jKw~h*#Xe5!-2b7Bi_aq32j95jx?d3`K5m6Lh5LF=#v@H zp&5C9A&!_kZDyIJ+EPm?ANbrz6(ZJZytEK$crHheJU{M zb0P*VBNH9K>mrVU&h*3aiLakOcN5;17{Cd|VtVac+Q-Pm!S6l9E)5D-2f8lkTt{+J zBThopMH117dEwHF8M1|%F#RLxT;4#$iKrdf{-V~*c?0&E0KQ3np9;oZwK}iPCxg1w zX3IWb6NiLZ=Io@(!|`SiO#RQ``IBDUCyz>A%y48$Ukw0@7T4SEO*F5+#pLUBw5x6l zZ~F*oY@_KwWy-yj5*y!hHNGFX{;3myw5JC=CkO29q_07fvNN*f>c`p$8Z5|#B!JUl0vWxMxhJkBTQ?A(-Z|^ICRTi zUXV@Oo<}YCtV=OxNV(Dr5iK?x;Z2KglT&Lx6)z&X6416;z z=GpxNWRuRSZ12;HZ+A3uP6R4>MKZ=FMFk9dZbj}_xR&?Q?%tg zJ}LXpg^GI|i{aQf6r%9w1f3%XC8bs!uhn|3rZ%x=2CaamX-%ja^u_(qvzz8aTfz_ZB@=7eLMlqku-qH_82^%oUG|ENoK7=M1VtM&4&4+)w`XZ4~lh!c1pESgmNm+tT8`M%1CYZXAj zJ|e~C7Dfv;*W;2PrV(35FhZOzr`_6BM8)6t_;?!jB6ojBRL?8Rz<7cC&shxff_uKaVvLd zc``$CK9w2tH;JIef_k{rf83Gl`%A7~Y_~Sb9bj-41N7_Q*4(!<{|^)fkgnxiHMDY- zJR1!q6ba7k!yxY6ac#!l=yw6~}H_DKH25WPali`;KwH(LgSPL$QX(~EjtK}7%X zMdsg?W0En0U-2$y5SUIGSqw|z@kVyx=arOp6y4y>C@vu#u=m&%zHJ9~4{so$R)a3jXzpTBL3CZW(9`{%kKZy+gg59v$|q! zE%+ZxS!C>c+M|sj`&ihWsfYHp>@^tRlF8Tq>-uiYA#M9!8Ovceg=~eIJBe>P6|`oB zK1Hbu$VZLPRmiL*Q9kBbJ)0E7XY2E?)sU&uJZJK!M2p{07S-tl*6(|zV~wPuzFDu2 z=qbLnv6t!Q=p%WbrRW4fq*`zY1vyljnBJ!>SKu>g4H?Q@*O9?syZ^TMgIVij-u0ry zZG*YckzvAq)t`;t{q>d9bsVGJ*;~c7406T>?_;Cq3~LOzO2d=zsl_wuD#J@Fy9CXf z;Fkc#4KzjyY_I;t>%8eaqctr7ACr<(j5E&Mryh%Xs)IWLZvIqcO(JZmb2y7YV@U0S zCsf+lI4Lh;&l#)F7&5WCQpWpBm(#TTSBa%j;i^~8Av{N!W-(D(qQ;V&P;H)_Ug`&1 zWc*UVdiOPWVsq94PR%L;e+F`R$c;}{jKCKge9vgrmOa#)tNEj%B#W}mY zw%vaH8uUqo#%bwV7e|Yt|6`T9reLj|PseQrJx2Ll&yhGC?qJ&v>ph@9F1(IP=l3Yq zM%IcQpix3?>#l7cMNtPW~q>@p1Htv*UTw3M7$1 zu&rm8_>wi$?Ie5=ukA!TOi08mIX4ZhL*wXf@dH8*&r>n80TzcQUGpIMb6Q#CqakpNPMKs{HymZMxmqu%-XISE44Ru@*r}a5b!o5W%{_ z(z?FJMv-Z6C0nW7fsV?O?{l}ZE52^k!;+4#Z=zso`yuG||BE>5T$w**1-K3lM98bj z9wpp5?ZyjioScscsE4CgfUkNYRUcZ`RwYFH!OHddYuJfUE%T6r2%=^DP2Z^%-Bx1w z?hfDWJexmrorNHNX`i2Ks-&Ndl#ImFYyS=LJ$U`s;63X#5>z*;F_WI`O&) z!TgE=Xh&`rl*s)^^XvuDK&1Xe$yKLtg4;%1J}7oi^l^$!_pqoCMx8;M2aGm*-A}P} z*1^vr{GR{^0UGhMwzIqwEPrsZe{k8On+Kk*y zbOB!i*AP1~*@|)l8LAytZ+kByCgfowl#00biwXXTY$76mv~C9%BM1Hy%~B%QokMdK z#3Krtij3CaXg8-MT#_w*9{^3boUzny&)Ug%ohhe&Zc+Na$!1|dt0^kOm>$sIyW&~H zkWb$PF)>1Nw2dGjcz;KK&7RR~$)HckUHc|zGv8R*yWlYo8;^bf0z>P)?;g>d>JaRE`V%_rSLej@OQbnCg}O|DWX*b3tGPJTuw^)S54n> z2Ix@%W0Pwn&(6Yh;OZ>`LzjeOfJsj7!^zV_)vUH@KQ0JG3$c2Vh&#+QOC@v4R$vE7 z`}_joBuCVodsU%( zl%_|@HYFdG?{8cdtX4YT9&&K*VN|i=^&29Ya=vs4cW>y@$D0qcv2R`lBXC;3qPS$e zwN;!^l;woxj7T#e`T)r135}x8OYOZr^v6k-CP~F?e)Q_h!-9UeQF?&zjlci4^3Tz{ zkAU+#M`Pz+5nUq^A8aeftc2odiJT2uv`eFDhAcH0CDVS$Pc(s2{6aWONZHzltsTPm zI8N;Cu+2kO9XVrQO~1Y7TvBVmQ19X1;CP1AV>MI}I% zeF+MV$BmASVzl)rliAwG*KR2zU6&*u`7ns2X|Yxg>Q3fQI6r~rd;=kD)!H&ZidS8` zmz(?G@Ru)Nj0_DyrBOBa)r?^qQv9^JL&sS5$yJqG$B!^KGS+M;v3ZUW^i@lLE#oH-UmDV?f9;6AzT7$A@*YzV-!)T_DosU2fRM1!BqJE_J&+IJ0^rA-hcr42xYCSUIW>V`2||;arrCawv3^bQ`}F@mwn7i zleaXEef%cAiR$x>sb?p9aD)GVqR#;E#}Z@N{jz!9d@g$JTy5if4JtyTu|y|f|zE_W5JB~2U`Q8N3dLCKIxan^%(34eq0mRId^BDxZRlj zHMWwL>S@zTKBw5YHmc0zsFr*_u&+k(`%z z{YDQ-NtT@uwsy2cnN_kY9s46d_TDB;-Q6N&vtvFX_ZfJH_jF}qn4gJ0cJ5}9mTa8> z)k!hSGv@$_aIpZ!4vFu{;rW*dkw4y15+Z0~hhpqk?>QFWe97F_baLBTZIp;W9(k_$5%}}YD z&TeZOENX~f>qTsFkQFGYs7NB-dUOid)HEJmZq0k$)eSkK3ai0ZWbYJsm>bZ!BXMvr z@*YtBO@Nek5l07qn`*5+3Je+CP+*|*>&d$IT998vf6dTe&+h4j_rKG-%>Ep;dG-9X z$ay$BvYw5o3DjS{q@gPHDr)Of@05E+A$U7diDw+`Tl590M$m;jHSf#{8uo;D`0X<* z-X?NdtUvnwp!-Mo{*F+b^}a=Kpqt?E#6)J4$~lTPT~Ittur8gV<+NtzvHLCHX>M-% zHQ(JR*Y;0(>iCf=H{?vU<>-ZKfDIvA954&EfLcPjQ+54Q1uuWukYJgQ(#EqlZ9OxU zdqzKgcVc*Ib;Wzau`;IUO+r~>Y*8|HE!F?p{O6@jt5LgEojg(Xhq**ZP9!|-1)^a< zg1rD^kj(|6WIr58Eq7NQ2Rfy0H10QH z(OQMujv?$4M`Wevg|6>VeYR}|Y_UWY4i8a_DW{oy9avR@*ASaQ>SXQwqot|vAJN`C zy(CP}x35Ia&Pzzr7o{DTCx8;Bqf0tlWDBN?l~P_1YG4UZq3j;4J6-ykyY{z_}D0R&e4z>RO#G%Mm;XIjBi zZea)P!5(_11z-Xs=Mcd)O5W^B4*aB1ol|{3*#r3KCK-d^6i*GCmtS;}AQ4K)c@f4w z*16_k7D){#^UBFWaSQ`;^77oP-Q^zX1A@-HoJQ+wL*E~D$(-CddWn1egKY=64;-<{ z{x248smXCao^zX)?zc#HO{>~-ot{On}%`AvMg5F}3_a@zE$9M|Z> zPv7MK#4~@HDHUgEto_Dl5n(YulI&jf>dKX-HVuw=&lDqnYQ1^6H}YGb_%R}@;n&kW z|M(aWV%ac>8SPrx-8GKOJ{#G-JcaqNI+JW-(ufTkkKAV5uK7-Y88o7inp8TAy@tEo zWf7IOXZ2+5gcSd&%xy^jB|gIcq!W%#Egf(O8mweQT<@*j@hw_7V*sY(aR4Enm8}b+ zj`oOmuj*g~4MOf90z#6=!MZxMSM~Mv@2p4E2sMjvhq(ob)`96sG#&ir?O7=~|AY?NF!4WRMIYEOB z2P_4ACY1N!w(tRXFn90f(fka}1aUE1xBr`1+vjy8g+*Lk3(L`p8Oo#nbv!nv{N2v1kU5c!2;P+V9ma3$tsU+1TS_-NSyE)nvmE z?1$khES)cy7YEiX3a!{(^W5#x^Oyd+d=MNTHW$BdumJtQ1^SPK+3M-(%{=U~V3l+l z<@C!g#0?5ehXNecc)-ZSR-)L&6FN+7e=_0eClcbW54FU9BBWd*F3Wi>Yo0r|kx&|B ze?od&8rMH<8H4}_fHd+=Vsy?9?;l!A#{~kCyDP&#CBHLHC4J}3x3~sVfEj)j?(=EZ zM|mTgrt`p-ai2*hZ5p1iP3}*c{Oe>uCaC&#fS{&4r%WsCw3-%IT5=~A4pqK!?<;J1 zzKO#9(pkk*u`)Yj4;@mXP2B%Ilyw8Of67rI$)jG^Has7{a4N_?^rrJJ71o#{hC$SsnF4H zshxU-H?$hy4M|<)*RQ1Y+H*8KokNxkLcO=r-YAlP*)FK6>i#|>{N~q}7XPRWo$zLc zQbzZ-EQi_06MXhBG)jcGLqf{pwmvFmi7G%6 zudbx-hupIN$;o8|g86gr_Vy@nUP^ObvH__t9V+zZSM`L)-Fqbq-kM^$5f^A4)T->E z^P{#6{><ZMof1XNT@<8qc$wuRbq9sEavX>oFv$L#?(_oC9d(nJwy zOSaqGBA)l3j;yemb|y~^^1D?V9AIO+k5l~&x{%uJ3mjW_PnQ3P@{?Q_q*Svr*wRz) z@{Tn;FD2O`dXq_RAG7)>seMA8UMFes6YS z8=`=2;HR9L)A=$q^eAX~dAUACt8HvLHu{iM`9bm1#}c~4XJ0G=OCiAogb;@*oBwMJ zhmI%%)k_5qY&_7zw4j-1=MT{$UBwFrxpoAjK72d!rfT-XMC%7{{F?e%-24vK)jFkK zwGhw)E?ghWt}{I0crXwlJxJ^Bhk8)zg0=W*HP>+ls?8`=cJ}tQk*C*h-F5hB%zzv@ zV*ABqKrC-zEjcTH>mDC2o$iQJ!A#$#e*NmkP%+-y?wWn1332x;!oeCJy%p$ z7pM>>nhN-Vicz$L2ppr#ZaO)Mp#de?faIMc*d~&0WWLh2DCIj?6WjoKa(n+4Ej>Lt ze@M;S^8h_G|X=DVK}aO@HQ@$fx*A&`g)fyjbHE(3o;DfVcF`VD*T+ z>q=GoXs)~ehS!OihW0KuIS<*X=lzJ|rMB}ZKoT(-32ruXt?T^#bz!@+ZtUr=C{Sy9~>iN~o!L2dV_re!FeNqq&}K67EPg|YT6V{-oWaCG18fX-k?DJTnk%9ca?X#~=TU)NmSR<{I3g6qJ@){p;DY|#m1W!8a89CrHEd}-thHQ5AGrL`+Lm!?J&h%EI`btgZ&!fmJJtoxZ5F_ z{YQ1$LB#DEWt~FOfsF~nYtHc>UZY-RAjTDbrK(ZvqG2Ah#)YorSbh6OL)R78MwfB9 zLtE{a?<8E#{BT0qB9Vu7dC~b@h^gI`{C6R5>~o~TI-XSq$DT~J*T}N$m%VnhoO<8B z3Varlq>1Rfh2Ou4agJA~WJ^qIAobO_LYL9D%qQ3?53}_N&|rK%%CsH;_P*iuM9AW~ z*J{a5OuNtg@PaPkok3IF8I>mGPASeyF?-5jf^+kBCbG;?7mvfd z4?WfGo=*(Ix7|==^m7lL!N){iQ{3d8c#cn^48GfR^YinEYLRSOuz2v~lMyJ2`x?stLZ|;4Oo4wCK~^SOMkXX zGx;;rAC ziB}VR1QBY8jPliHk`K$Z1sSrKh z*5n3RRFui7sc5_`d2n=DVQ74HKQUAIVpq#*v#~!@`V2OEZDD1Sb?EI!)`WfUBMuyQ zFO6g@`lHygg+)N%DbbmHagVHuz4d6Fo>BQbyyGwH1Rf)MpUf%XmlrVd@-bdc(G-E6 z{4Da24ynBI%aicl{3*wJWU%prMmW2kv0*b~Vyvk0?IZ<(bN33*UNT7ExZQg|Tv2?& zMb0HRLi)_lv0T?=gVxj&%I!%8Q??Cq7fm`5aNzjBHG&P+PyHBJ}D92@Iwt&mK&0u_$=vCW^G@ zggD4W+v^*^%}lDK-JzkPcR(`l^jGdWAfr?$uSk3Na!M88)ir)&dQ~$NFMqcW-4k9o4;?S=$|!zM`r8o}N1?so zn0%$R^j81RN5LWS(MI+W!mlH{wI@Fo_fFb2RM9f3}a4<=|OgO57}pjk>&W9;iM<{_eqeYH4C`~jkwmx zDj=mK70wt&M>XaL_`^IrUk7hQm+Q^wT@0)!!7YcCXx!ZD@3Y?YZo!ovjXs8KfDZ0V zsj<&Jw`X+#jP&t>}Dq?ryu(>sE?zzBz{#HJ^l)0(QJuMfbokn?L4K@ zRrf6MA4S6%C3Br#m~?0Bc4--@6Z4;ov&x8KkWKk`Rs_I>aOkEx8qBAYzI?&wNXco zpx@u#jcd@WXlC7ByTGw20+I?Mw#3!#JjyWCe%^56&9Wh4-fxQ6A^*0 zur}c$vvK_^Z8qKEqW349C)xg%s^bm2g2@6Ov>bp}B~izqX&^riN6eZ&^H!I{CMt%{ zAoDJuP7T&2Y~OWQ#ji=uqAQDjPw2~}K-%+`>Kn!C-|O)T$h}CtD*o%8!%usq7?on$ z?E#~z)%8B!H4bgR@)lauE}<30e=Z3=Q#sFP6SQ7lIW^gaA<;(8hF{2gbGX{u+-ri$ z4;}G1UL{oma3q#!g#Lf*xvjVN<_C2n7(EbB-G_O$*i`7tfGVDeu6&3;JYzuYdrqlffwC1=lX z|BT1>Wo%tM)3!AAHv-1>4>8(TnK>EEEM3pG^gw5mr1qxr#6(^(yhm~nWhEuixgSez z@QOZEQdVX;*?Z2H`Q%q%CyA=byD8}eZa{Tbwm5DTIoKr=k(orxj+vJ3oZ_#!xp-Wj ze{i*vh6yzy@T99trUjlg|FvX1rfM9!)V8sy9L{ljqloLYmbpf-M#KsSJi6e^E+O(< zxD1XgiaAftP8HV;KTD{1K}VMqADzUbqMEPG7vITL^JjkGXtoN+_@1p> z50>7q&?^|4-z=VZK~?Kj>tokFA( zRzgBzX1L*>1&+?wnS&Cp%VPMw`Y{-^giHg7C$cX9PwPG&;_bdI{Oi}R8+bglN7T5^ zk2H3jrD70X^C!KcMvqVXgZd+HZ~o&=Cr4fw?_Q)H)h71K0I}mq_sB3Okwmgs~gNLuyye;Gkij1x`zG zr5+!uGbPFY!7hNs;pDfk5NqgZ0qidxzpKg{^=IJ7?xSF4pJExm>J{x5+Tx)QbPr>@ z<^|t_75;9@){y~o*_CxaJ>J@`brPG>KGv_rzx(m~nmatI@)thk6-^|ZQQr|?p;s3B zVk*xofm$T0sX*?+*Sy6to&zG&6D3u74x5!Py$d#T<+#1)qrv2wQ>7kS9|Ha8YCgp7 z4N%&7J2B;D#HG5=V?V4|EkhBvAZqmx_5_k=L9(}eu|Ysv6T{P0U{o9mFG;xX)5EHV z%gwaVx&A1C@NyS3GcyC#1AnnZ4&m|U)M9Ces?ihP@kUY%`&jnxzf07;c=!os_tUQ+ z9}<%plGv`IL#FvsQ2X*~R7Cn2kFDm@gS#Ze#l_u6%l&5;mM}6>#j4J7U5Z?T?b$q$U|0QW=X|Ml{RwEs*m1jYyTxu!OS=440TGUF8WRg0`Hsg4 z1b}u#DlH*|hvL+BzoBwg?uY%B**2g%MBU%p&5U>CPX79ZTHZY|QG*7V z!mYT6H+*w)LnW5{JXTQp{dbl17(X5W%$N+tUifiCR%F}<#`1z`W^!^e3XkxGTs*y6 zLj98!d>jJBqn_iX=MTHMoo*M8<_oOeQ-Tj=7+W$jGmld}Sf5a5#I||RJq5-bGI$V@ z04q@2f97lb{#z4vq8y*=(W6H_cvBYnY;}>4!cI22(3O-J?w(E#0|Jv6E^owFnle2Th27l3&oV~za8?GkEKbeti za5y0E;P}3fc#EG^`;c5J54$(MvF)I}KJ%tv+Q-Kx{}a&$0-A`UL*LMTekP;e;m_r+ zrxDmNjGNSFgKTaG)eTdo6PTTeFBYLF39c*4nAfNSEA&(X>N&kWf$bo{jc?RHM59re z8Epw9NZs80X-q}12~5t-34M*iGRu9bsi`d?H9CMyKW)UYENN_f7+vAcIaN1kN?wS9@(MmU*^XyyHeeEDrWaDb;ruZQFA+DT9E-C%4zk``MhQ2p@u z;G>sl%n45_aI5nrRO=5wXRp<>Utb{Mim|~*w#s423gZC2SVBTTEMnewtEAr&n}DU% z-y0T%i#xWPn=~ZtQcHdYkuRkiBH)3Op7Ww539RUZj$AO=O`dMn<%*0QsLW@?1ZMP# zaoOcEHr>0x#Ka_E*7_9XmA>NH&`^1bR&E*THsQqaaeL7@uBobuwiV6g4TRK?7p}}+ zRW(lX(T*}Fk{eT0R7BoD+#v7Jl43xrwIMSPK5XyE$OXVJcV?P*onsXd3C3EUeMQ1} z?_ZA*Ocvdt`4f=)B_{#K78l<`?Fq&!VcAb)#L=P;F$@qD=P&4t^y zc|oY8y?;h)tEG0>(8y5uC6!Vbr?84+D-HJ>sqm}E`m|?18K?H!i^h0vqI2cA(fJ~J zaM!OXhtf0cp-&5wC5I;?%}EPI2$S4S{83%e4EPX`>J32j-w zJ(=Sm$+rpda(u!7fXZuUa#2-Li^ag*me#I76O} z5XhwD?5VdB_T3-X5(q+8)~jtSBwi6?DdrwY9Ce?`;Od=pbd_WwpN`IceD1x;6GG@B z4j>mP)^4&_>Khlo19g>oCSGkiPb$Vr9IeM|o=e!q$v(SM;3R_bm-|Lf#@YGg-&^=@lcc7-Fl|(jE|GM0) zBXgrIXQzK2jf^)IdtMmQW|kBm&?~KQ!SuVPq1SRY+Z+9N2@N{~1LjpBoj5(SJL<^a zb&;{!osLRc6(CK{G-&Q~p(@Sqarq;DCy7dlkyBXf>-+TtJ!{b`E$;PKt2o}_XqLwB zjuj3S6A@2H=x1$GX|d!*-xjngxVwoX0qpX&zA6KLQj#G~6)3|k^zzydU_t7czVkIS)}|lSF(o|Z z(F#Fy>@U>T+zyum4qT3?2t=G)QX+}hs%_xh(W6IkM9!nGOwP}5{w}n9R=xKp;0n^S z;(N~4*mcpFfLXC%j?HTn(cnKW>^fAjDxm3g+5ex``p{re3JLau2#(I3P5K|n;A#T` zWsrSpbxRX4!_yp)dZ$}XRfS%OCv3=^#Df~k*PG3Ze`1pR6Ym)ZCzDtZ*s>K3VF~Z&u?4obr z6{HkM^jtjUcfV@s!KO%~S0V8$+{*LcC&E3Y0Ga20@lDF{xv#i!r&=K8#pFG^(>>U3 zuA`~onB;!Z^=9fRb+&vL5>O^}J{1>>-0nUV--p{o>e3}@)H3{*2Q@6#qe{+4kQN)t zXS6XE!VKoAdeX_UiJ6<4a*qvB_^u%tiWo2U^YT)Amwtlty9|h>@a^tZvmU}Ug#jz? z;-9?(+zY542{OQGIIfi}Apuzm>RcLL=mTYMY#)AaB4!afA=8KGTOqy2i6P;s%YUXX z>Pm!GdjgbVpu(Xc5|VuahL_?88V2^PUe?tAg*@#XScZVQt0XFUJbbw2z@fNIN1TuW z1hhtDf;s-iSOikhg2kWo0fB-18=}bykLO3PD}bs=%a?EtCsZrO@@#g!q$F7wmWVqY z@9dYxc0jM8)n&K|2HP@^k`7d3>Pfs&9^zn#+5iaFO%ro`Rr`!Wq^zt2h*%9ck@=qe z-;ZeJ2%J8Bnlvl>&v{Af#DBI1ml#o#~FL8 zt3Ya3Wfi9VG}J7p-&bzK%doA1ug+Em@9rKFL``I-Mt?Lee#DT;xdP!SdAh6cO)zV) zN>IVNcyz+1LVZ@E9j&kN<&2PF7|oF0Vm1jF#w^nBT%v`4q3)%sTYnOZ@Ywe6|74c% zra+gJ?BnWje;hedTvpKd<4J7Avw3`9M%tvmx4*v>IL{%q*Uj8VHJ3w~?C!n)sB+^* zq(*m@snDId(W~e#^%=hSxQGGBg9G_404DHPYtVMx^QVoCxL7Bk%aP8ei*SSSOSB$t zZfi(U#yd~`TBEK4@!j5~W*}7u77#X;2tFN83TYjbL0_JY%~P|QOWv|h08+r%blXTl zMWE{7P6;N71&75R{pqiDdUBMMs)S?$IMGv6xZHwHO;<^7UUDjkdOP37je%1&6AZ@} z!s?FjCMJbEFR`}`8Y+bX4PU!R2JYLSNhYwQ)g-H%r3s7dVZDzHEP7x>^UII>% zOM?-O4g9yYr=HyP>(_~i8o0_LfbGH74@dBN1o&VK+3t@Kz4rB0{xt`L{FZZ{eCg+t zNFd!4(f#2IBG#!tfB*73U6$>dt?25C!2V{!^r~l?o2kSK;7m)o{uL*>E31*l$3SH} zc=zpGmZ39iNsFr;{?ny;oR!-%<Q^;W}<#01z_ zSIg^mG4YQA*pAa35$Vq)IW;r$gS0rO+}D{G#kqzea2K7kWpYZQ5gmR7%v_=Ha~GL) zj)Li=#dL(ZCb#-6hY6UWsrcy;_mql24t`wMXf5PWmq?%_`c27m=d#6J+;9gdW8m9X zPKJ9Q7ez7^kZTi!!W6Nzqi@c%u*nU>EK!$de}`P^SFY@}u&|g4;j(;<;lSvTua~z; zN^(h}qM^Z_YeErXX7b^U+eEl&rVV6X`0>z8Pq4|!XwmeJD@iYj3xvv*ZJe(%9DEea zqat5jS1-2_d~8 zf`>(}dW2e#NGpv0JZ4wWFR{iQ+KQw1@g)69ooh;3iz!hd02xX*?+`su_WIVlrTKw( zGpR<>-V{MxR|5ds;a9)eewjH?qSJZdmo|o)5=;g@i_3q*xSXr<&@&KO)>E#l->-Nt zpZx3MaISjhNp-*M(&#M7Fhz~4MZcC=zB=?wMg~6bU#+8&Lt~c_Fd>Tzhp$6X6YLL} zWdIQxZ}9uxg=2v4P9<6GF`5DT)13LL_Ne=Nv5A(F9?3o{qu**iAP^2mi?TaoIt%Y6 zoGIb!Hg0XV3J3moXSkj-Qx{D(c4Oy!NCbzagWD4b%~_sp6krk}FDt31>TFz%xdP5> zrHMm11lAfHNE}*ZrPDI)8KBlY-@W5S`j%(ixTLkwp%98&cU=CMZk!Zw?E3IQ4_1%> zH(ZC$`?rS|jvR|h%uITec!RH_qf1{^S;q1p^~6K=bMhO`73&@A8R|ccmUIzn_*d5p zm$lR%+xZPP+#nzeZo{kj4j)ok`NRQ?qf@+yYYZ&8yJ8s8SUmLhj=^r`mlDpcZ~kV+ z8$IRLcn{FYkLlo|kuA{d7L4_ogYi1q&dkGD&UP|=dCQe;HP=%_>)iJOWv ztTgOwX<8XYSt+G5lM2}yc1R`3v9d|X%Ew2t9%BIan zZ_6Q#$j{q+QT)lum|@?0A=h#6H#%R+69V#TXlwcDu|}S1;y+jBEnY3@gF&g`S2J~^ zSJNVjZ`)6gDIW%}bD^gcvzymdD4~{n@;SxeX+fBYar7PDsC~P4U#(ecBv*|KGYW96 zK*jHd_eTIx`Mx{d8Mp$9q43TPf=Ud_qcop|h(I^HA_2U6|>i3j-j_83I(c{zl8 zE0f+x04}fZ^YzVagT0W=H<4`)L*GmY|7=P*1;uOp7Z_hoixkXQ>of> zvQjm=Z9RYM!ge+dYrY}b*nJyJnAwcq1(}R^o>-%BfvVX)v8W+!sn0RXR$SjhlAaC0olO0mQ%k4vEP7jeOF?Piz|AiD zI?;h|$py-Zy!?C+^yWJ_wcN%*YSYqHbPxG4ezAquh6IDbfO1Pog@Du#2nktW^u!>x zFulw~+;{bM6U_8P1Q}#I^?{(LF#%K~rhL-OY&-SgBEC=DGHyS?q(bi_9NcwpGIe_E zT^ie?zYOgmr~G1kc|-$8K78m~GiWCSbr8R-Y&e?P)l5O55a8*Kpp^Ok{+@Xyo)KR_ zt82h(`V6#w#%w|2)pX0o%X-&JG1feTqX!*L?CSbdh$AWVU24ms=@?#sx zZH?t++srGrZfP}I`4S-36R^G`BMvm+oI14;7=D?pKHB9@R4f!EV5j?pe_pR?vX!CC zOQ;-LFtJFj72X!@&Ll3s;+RXqq(exywadL9I=Qz7OXde_7r96o#h;FfycjUh;DmS1 zijURQpYO2RNGA0;xk~7QIiQfj7x=#Ma9*ym+Z5g}E{qeTC4pMPm?L`QLJ)oTfoG5u z806ci3sj!=p19hYp36j6c;%nu-n5CyqBm1_R6zixpa402mnu593GN5egn1GqN-MH+ zkUujoeTAVsAVa7T+HiVz9?8VuHTW3yGX*qQp?nKNffFAeG(R;qi0)Q|#m3!= zKhO)$k!8)AQqbXB6DyitP5>h!yoT!V)h?#AaKV^w4;*ebv}wp+=i=wXR*e zC&Lb~o9&wv!4%<9epkK7&P=_j`?Hh zP8WO;Mj}2O*9NyKxe6=%esrS)4LArb&zkeaIAFi?Nhr3%ZVO&6ycO`P{cCf5a;@}) zBQD58spe~;%n|@h0!3+)r4(_M&kt>(Iz;yju!Y6no<0%Jqzevt2Y4tL0eD0!3JX18 zWDrN%u)ehay`Aayy#}PEJZeohKpOqy0)Y4IPrM2J|dF zJwkaQI>)zoNf&P(2PyDiXs+id)ULf?lTwJpB?yF5%xO%(*hTKyDP!Nsh zKJ?PQhN0bx7+S2_PlXl0EJ`Z~Fewn>1hJIv(=~YOs1rl0s#40Hns1(pRUxXRw?iXj zC9SH8P- z9U#O4z=jci;Tg&2YJz|zPx|r8e2-ZD>DBwE2>GF;3jlqbb*_UBQ3fXt>SCxzZO%m= zW*NdxqM-U+UJ(}DXtqb`X~ zjDDCMBY9_u@}guI)swWJ9J)z0KRn#0`fSkkJ9uC#H~o)pgIEag4Us4`oCepY#=5o= zJd1;-j7aXU+0Lm}ntME-pp7HY0=4G_;gcz5Dmk;{k2AdN^I2|=uUr}M{>C@YU6r5W z?w*3mGcL@jBtm)*>19PW@v9WDH*rVa2s&;y_FLjxl#s~O1tCp|%7oVR{WexY5@7>g z0_h0cGO+e{eXK>xK4oRn z4M95fA&#*bZr5GF)YYRaVz!t)Kjb4A<}~bye_bchKn*}Ghm_@dNNdx1y0VT^X#@@= z{v|=dK-sz8?RWR~i~7!qe>4ZwCn2KAcaF})Kxodg21pFyRltQ%4ifT(?5ivRdQBCL z)Hg7u`_uKED_an6+G(e+VVMj*+q^ae_SMhlT%4SCe*gO1aK@o6D^gpO(6pfnY`?7G z#M+7KqpYgB?XvgSI3y44$azZdT;m;iJgNFwO*bH{Qf1!C*=Jw~Yi_;*kt3EFzfq85 z0hy04fjSRdBAWMi&V&jom&OP4O$kKXho`3?ze^ z;&`1zu;@#`XX+Qmimok#s6?~V7>`SrI4%C%X!#83A{n-wZ?SPr8mAtwXME+t>=xz)wr{{rd;&xixXx_b4Z7WJ7kCK}4h%l(me zdJ!jb_45-G>7QzXX0MTV2_YdPqP^NjYUwGnFpK|{Fwfr0eDbc8)n!1q>5S0N#Gc`{ zT-_}GaDFfL2+ka5n^WOm6{`h~F6_)qWx1U%F()kUf}{+!Sqn$_&b;Qq!hV#a=fK&* z)uTVo+H0XeG0PNOjp1YVqZC24hSNpU^8IJkXK}e8XiE+HJWqfB^4oVy&a=_Jiz0Q2ECOc2PndBWS znY(C?aJ)NhqZrkb&PgGYVF`(qpwjwVROlc;#pz?yI%MC1YZFZBb=+@FIb+QY*l3Tz zJd>wg;!YVYAA}3@03=H5yBna^I{}2>AX^440Hx4!N?x9Rtv!Df*J+eYKl0%jaxs!X z4W1r*WzNextH7@Zp3F~aR)*^bzK3L>n0;$*HmucU{x$P+yl!vMaQXYN<$D@|f{mj& zHWxA+Oi4ZUu_e{X}9W(D96d=Jv53{h|zXJ~%M)j%Ka@|7!O zbw*bbaA^GtG_Pphg@N7^u7;O@YABt5Hnv^fn=i3D&oX(^lFI(N>f!zIvrODCRJjx6 zqK~SzhRKET>dFf1Z{!y_ynJ8K78TN9eR{+MIT}hT<}Zl5Pdyd5=s7^+j%Xl+ z)RLhm4Yg8DtdFrDow-pfYB54|c!A=gr5PSJ1^G$uW^f$h`vp51QAk^_L`Brp3 z0+{MJnf(}(5<~@Fn8q&_gh^Nj)NmMfNWrBiNp(xSBumRxY%iE ziG|)7vbEC6f$5VV>%%UTm^$;1dTA!2IBsp$^}~0W#yDn`t`AwZ7)EFOnigTlW9G-M z+HPJ7(D`!pS;@<8pnx#ic!FhXd*?KCp7uHbwdW`tI{FQ_yCE=yWb?W)cbkqbIf(aw zYhx~|cBmbb1$AtM>k;0M8POG;E51Si)Ju3hfErOhFahs^gi1gK#u{XgsYN6Xrao@C z4J~}rdqA$}X+$mRW$E^UU|xLfN)8U?M1$;(y3FnL5UR{>UWzgUMzxHSQw2dFOEwo$ z(dW7_0v+Sh;~FUsWh6eRI@}p(JR{^8UA-u91C!9K|Dk2T<{UBDy0?}`TaE7{r$!b} zkhTP;>ME1#Oc5^fG*<#LXJuiz(dmjg6jb{|M7iM`s&Bo0{WpFrDPmB_X`n+t7u6gl zd2S!Nl2c$sG4uDQBbnjT$7UxWHb!u$V7lV*>#&qMRhuSCI?qbq8ssq%TMC|UIY{l2 zR_m_Qg8b(vK{e(?(Yas`Lc6Uq7?HLrQ;3zT8i znGZD{01NSl-a%pyDbe?}hodKZU!qZYAE8iA%O`-1)!r-5x^RKoI_@6(zsidF0re+; zAi;L`-o2{}@mqp2IS#0Xebd2~pf`LTx&MG}B96~zZHZ3&a*y0MkaDUiD=R;p%LPIf zARs;CL&3#HZ~q8|X~PYS%@z)B6Y?}^!qf^RyI5K2JU~ZKg805DA0zd5Ox*XEcP7x0 zP>>b)!kgI710=l^9(9nngZ|;HMFflNwJ09NR6i|nN7$}mYK^#0(xELlXkp*J9S4CC z^5EPPhY^^;$X|e>psb=&Z0iF{lZeO-4C>#yyO%s-utwl6##_cCD7dil!GqXUEAn{& z_Lp%b1?*UElQ*HlDz?YTPFJwckApS5FGfnUbfHPS7j8IAKFoNwxUbOG!l$!r*)sZd zghDrDqfoKoFhV-`_;lMY^*?wr+IaldK=pYltnkupyYL0A>c&bD0FUzqjS4A@0oTkb zrcfDBsQA5T2(#f6&H(s`2W%LgYb3^L^%ZF}0XXesgF#IauN`h+$aLH70HfOsl!KR> zd*=20lP*CA7`G|g7%e}FzIAe2xV6k(+P*T5&{3r)hc${8FIey`P(vQ5%E;5FDov@= z^*?>@U+L~n;J|c*$v=RPLO7CoxYW$o6yynf%{bN0z1|6 z?g-8aHN6Nx25)34RdsRZ>=A%u7Ki7Ov4xy*2zPJp>+F69iAf&WeBhA~bNwM4Tq3kd z0IWQ;e+N)a$zyltB{#Nu=AvD}lZQxQKLCN40}p1QGKG|xq%L|iCGU?pvNk$Skc1#; zhhc^QA5pDZ$iUD z@7);TC40mh&%O6gOjvJV(z|2mK%Sf`zLB_CUTHQ@`@9>G{Ffq6ya@T(|BGzZ#*i(M z`gwcStBTGR6sg@xO4bmbsulvgON|TdH~94vO%S2Xpy`S3dJ0cXXBwtx7FJe8#EcgO zZ;crD!R|*%0^Kn(J#2bZd77q8+l~V{n7$X^7EJ2GX`;Cg;Dh1wLze0KO0vly;3-rD zq8bQjtV+@`YT$ao7cTGa;4drt?P!|iCEy8nDbj6h(5&9MSMKzeC!wutYiWK#{DqRG zYC}RTEGN*CIaKV<|3SqK@XJigct1`Kjv@>Kz^~eu^u{X$QO}O^0ZvP_`FZD?Pq-*6 zAnezZZ!3*6F;Q}>MaT3A6=Ua-p(p5=maSa57smpCusk4uC`gz;C+G!Z>Pf3R<`!;$ zJuWUB{V2}+_|vhL80swb54#ij!bxL>GfB)fpR9zmwXpB9+MAav_7 zd5nh^&KKmKj({E)!>QouiPJ4`L9P?+zr;b#>K;;n2IdE39$(j(i-lUs+$DW6dU;LU zg|cc&KHlDuP}gD-`KY|6GNJ(=AsnMp1&9jS!*`|u?jp|rI{mJOnJHMNYX#6MGlD)vaWGBkBYZ zQ~0{gFBL#2CvxuJZE=i}>_v@EH=coC#KgB47LkW#vFl{7B}?{!ipCoahGQSZ1a)=~ z&V+$xw%;bIb*u}2F~H9GVNs$`mSX-3>|XQ+IKZnFDTK?YC&bg8R6k*#)&X7WS6J#l677ZSV`DxV|i9^iQcTnad`> zLSg&~aMYyWADem<0&a|2YD(NTGTnf}A5gjb;i@Q)XkU=pbOTdo)jWU>UHDjz>h9fn zOW6LwcmhEcA8rfvv*c(!|VPI1(M=lp&fp2TMJfR3FJGRc#i-h_FI7WNmcQG zd=yh6EDzrzbiQD+!$pcXd5~4@=X8s+9Qrl@q-}%#t)LbLRLs2?TP#V>wz(l};= zDZ%XDXwNP5mi0IH-2iGD6L*^Trp#S+TtWbFETw$b zCrrAab$_C>88@{m_Q7CY6NlmhsDWu|XwC0ygQM^7$>FyeS3G-m3ywTgB~Uk_U83-6 zef)gg^34#u2x16?nAg6yp4t|axa$ySVX&_&*t2&p9Tbc*?-AVd777X9DhI3HgN)li z!U<(ScV0liDzBlSx8ltWd&t>IxkYG*vL7;ZQKmF-RnW<(u!=aCewuH~J(D$2?2!kz z60)PKC4AD}xL)z-0xOGsc6`>GMrGSsH7v^<=T_=R8+Lzz|J2$&(*7Q2;HDB7vm+ytI``j zho2l0MF!nkG+A%Z>zI9voD1fRRR_RFa7Sk^IaG8*~Hua0>3}^)CF@j^O`!Y_s9msH$ zMwvj(=IO)=A4tzgn1(pyH)kaqY+Yg3h6^hRB9iq&%F(@1OBXI|QkWWL#++0#RQDMR zEninuumzXw*|%>YTt{S9A#lM`O3L}y765exVK_VJ2+yUadU(voJ7@ay+Z%G>R(*-8 z;2MB;klfbk4t=+Jt5;~d&`nXxMJ7t{Fyd=a%>J2KiuRN=fIXVy=v{gFc^}9yKCsKG)wiUs zcwN{8AvV{HgWLmenm_65rxse*IWF-C@;-SGzyTrvcEjvN)>K>vWfckZ0zEu`roaVOi=1Yi z=&jGD^*54t%9L$C%sxR!0F5p>^IrEi&j-6L@Zs=Tjlf%wQ))%CNfIZdcA#cM5&XB} zdaL`Higv_408SmAI)g_RbgFGh2&i_9crRdAYK1=!&1&87* zzkhkZ<~|q$3f;u)zN-7Z&E_HmA?${l7a=A$m$j%($zon;$6qF+`yOtnVexFiVhx3f zJY*5lx{0XbEkGzBg;=&`jVehI=^jCzLCzeuUm*O zBh@5rNHaShTMC5%L^O){AyC7I`lVMLRm01>|bDYJnlD?(DdLY&^-haIf2u(`Q8FyT-=cxMXJpcX43&w1#3KCThpfqKu$ zI*Ex8nmI5Dpu>U7Y4)eMW8d$*RpV#FW5Lx8eBC68kP{i(jvGk=n>+ybspHO&1|N!I z%wPa>$74h_D?olYWCvRJo=#iO6^(FB{-FXsLX}PZKHhXCOHGaMo1tUFfFiJQ;GICEJA0v#Vv@r}2poz>=G~`#WUj2}A zMt6LCyv4>HRt#e&{eBbyD%=T95ATIbpLjb5OQNkyc;5|Ti_=d3b#gvmq`<&|u86#G(-_r1tl#k0 zx8vdC3K)1Vx2f88e8dXX`~B49umWC{%lRio14M`FI`PC|^xIq&X9PaEhVWiYCaG5| zsRyTVLO1c=B7ap`ALtaN5$iz&wv0>|Zdb@@P+K0t5kNH*DM@DKZugzK+@m#L!IJd2~B`XXqU^9)7_)cOiS4#nci^ zW1$ed1IFGU);}u_Gv2-nXJ)#bn>@tscr=>}Q{u#?%G;_y3wN8!3JSAdznp(M3f}C& z?iwHz8Aj68O$HC+@aM$&`S~kEy(3vzSa!n92~^^P>+d5@TR&}E!eER-FMv+p6kwZq zL-ct7h7?Loa6W$b)n3~aYkOf$BIpfZ`ZqLJ@NeC~tcZ9Qq(W(m(bzaEr)#+w)=G;b5w>#$Tj`TkCdfg- z*N;2D?5iwp51shR4!cK%Ylenb!M6v}qYR*sI$mV!flcq8w0 z{Ii%q=ggbe*0*rFx7cm>R~x*01*|N&jYI)ZImZ>eUg&GIfN50XK?XX-gfyW^<^7l63;4 za=7R(ot(+BGgzUe4yT5we{SP_uo6Vi70k0!tEg?O0VRb@Fg_GMNx#EeR+Y!9^qO*+7w??bifn7xG<=ml1*{|>B8l0=_F z2p*O4Sye?;@Tkd`S$l_jjo7pEAw)ThpkjRbah zU(~w!fv3k#5b$^b+6YUPl>GIwoj4#|>*;A?V6jm8{ryd$F&OJLV1e-3?5fh)R4T>mHgTZ-4nIenHgPpUq^sEoN6dLbz* zawD}xX!;H&8pt)vI|bi!?#ph&N65>|BVsePT_jaz!8Bv#nfVvJ=6=ELdOjb$uP+F~ z7l@}d+@4y$V*aFx&$dui^URE)8DaPGV&eoUOqvotpq`zTX8iMD1nIA-D8aK}xHq+= zmEtYHQ&O+9qhGc~d#t??g0|x=MK^aId8X}FPo{m8rj44BdFRTxfKS5?2FTRaeUub@ zz&~xYuW8B3B3cIM$5E*Dclx!Hm>-jV@o41LwD}#3J%BP)Fp1*37p-TjK-OIHsgIWd zWdZ+oG?NQsk}GB;x`!TKqt5CJJ>xk^rvKd zZk5i0tx5I~a}z>gIT{u)#XgU3KfxudsN0|Xmb*RwC#n8BOi<{;hxT{_PK7?91Xo~|Vc2e{0R zZiQ!eKFp3Ei*OnW929z!#J256efAF6NJTEEO_|%@s$2gk=P5;s0+&Q^TkDG2p09$+72fm4Js^*Rv<^o|n9KB5H9nl)?P zSk(!@f55zPv9AYDUm;4aJonD7uHPRNOf!JPffZOpxv{FYi{rHRXuXMZxY6)k15HS5 zSZE;xX!T7We&j_L1boYK-=FAiCaRgCYDR*A*OzAZ{R+BjqWt?k2xPV0yXPQjYx^2q z2lv9$+@>y2Gzr)>G2T5FAT7r6 z-LAXYv~J@jWb$sRSd{8inS7^PzREtj95V)a8M|wSF7}Yv5YAJ}nll%$;Rp8QLi5zj z6b}o)HiW>TRHw?NI!KLU)+3GV_9r*Akq1Yy5a7mDp&9k@j>pu87r-5(`!hsB;;7}~ z5}FCTomoCxtrq?gI$4;U7q6AMh3~j!NUYZkUjX7|kiS!;8ozVbkrM6f(`~u8KF7UkPt;5G11C$O zrB3P$i0h!90H3zETIac9%jYsyF${irP+1jX%gcTw0&PUl9Q3Cce<-ztP&m|TixKBJ zBD5Sp?T%e^(AIS`wqMWhw8{?GHu%gYfC&ia8FzawD2=>W5J$cQnOGQ#u0{~Zj(V)nw1-62kbU!0|`Wyw~!Ey)R#>_#!Ya2 zA6C~khL=o=)k4D&M>P<#E&K>F>zvK`Ev3ND|u$m%LAY`>)T$x>^Q|d)zfg_dtmoSsA}&>_&8~C3*~d553ug? zoL!Ck_HM|->7v7OnC}lC%975T^sT+DR{c!xAQBXcAd9H`_v)E2a%lp9-~pZdvaL)N z+pe(pd^S|nZRj@(-rK(ZdXyQWsa+T&DV+!6L#j?_qc{vL2G`&xdc$~i34lfV+FqD! z(5z=AHb$D^?L~J7+#PHNL`YtvGMlsK#L8W2yl-;?l~MR{IkV(%Ix%P z3egfhJN_nISkv=2{06)N0!gqUkUR!=AplV5Co}B2_knLAl!ePPZ;4t3>_MwYxBzC| zy;5XieB9@B$iXtcb~xS>iwmk**HKw zlaA>m3J@~HO2si)v_^UPIYk=7TLY9+yyuR5!71?OD_U=SIq?9~bQZjo!1>C~J*|z& zasf~|-${IW%DtMF`QLo9RGF*j^A(@;l;$H5(fBG6-O2$QNJZ!Fify1U9AUGt_`0FX z?!65-ad}L&B${8}J?oCaa&0bC$vbtIA9db_$&7q0AU{lF!eWnwz;zP^wRgduxpI!= zmo~xAKD=~sSF^A|mNK3zg#34{6tcg!;kM_LbFzKP|HONKIT}&AipAFSLYfG^;g-=R zZtI;4#?p1N%4j_axyL;e>URd}BUDxTmJTI+`7IopYo5BgSO(7NP(OToAkLvgXlPw7 zFH6zKlpTt9(~HT=O}9+@r-)S)PvF$Uirb!M%R{ zMbHf-Z9wr1s48(;j|LM+16soA^T!Zh#J=&kEn8ZH{NB@a?XAIwrDg=7UAz@mq}srI z4czVkkVKVJ%a|VnoqZT5h(5Bs&$J4)4$O-pa^Ku*P_5rV`X2ajH2+E~0@zMMdFVl%|TA#1Gw++L#=IId}~aWXcN37uNuRh#)Sn1 zy`dN4xO;wZj+X1_W;ahJ_bsMjk4XQx};vm`y-gx9Y3R@WqqGk0#ga`dwYUgYD|Cc#dj({XC^t z1$i^sWr3RY+TBB?(9CghoRJ#TiXvkI7ge=-kv_(JsGOVdc$6%^fDd_g{2$|_V{4{?}tOm_>_P?1}8iT4-XG*ZMF^DJ8ts;ASFJ`1B~rA09tL7-IzDid<1hJ?W)zeB z3<$+2c^Z`qPU4=E7IoE@kf>|A(ZLmuG5V&FM(TqIQLttjx|E<+a z(9vd`-*5ie@G5Zj=Z9(q*K9|g2s#<$9vW-Mlzu+ozLPfW6VS44jfQIdeJl5It(Su!WFv6#!8@gu^RcGIjc{aBxuB ztv(H<$|KurH&DGgKWUm;MB@(B_vEaexPa0{qBnZT5NZ9e_DBYGH>8DYN--L?D!Y^} zhByMlKK|>77K<`aFdc|ORDn-4?=>Xk7nARcKAL-o{DfaF99}YKv-jmXb5mSDln?k? zXvPGlsRMX)HBJUSxA3@Uam+j&Tyj3w?qyR?h{y>=SePjnMOiG0C0$vQ9l|J1G(NZ- z|He`kJCuLeqojmvg@^+{)-bNW#|#7MXX=E!y#V~8)(`?wPlpZ}U!S%6(BT9fjJ9g1 zk{Gp@lIqf^G_w>*)`$!V#Cw@!*L~zjvdc0q1&e2Kd6_3Ym1T;rJrd`YYSNT&+`0md zj1A0){2ThV3iC(st6VQMx}vye;in{f4YW$6H-I@To7b@380-x?y{l+ie?lO05I4r+ zsd)%iLS+RAv=mD}wnMaX!z}nx7Wxt#Py+9OM^Hd1RJx(uoJq&uT!yXaDpUZpru!p& zQ~(=!Iki0Xp@s5S^@o8Z(cX~{0Ii!pWmO+@9a>IYmCXA`y{Bh$td1I5@6f+wi( zq1VFSIJclr?_)cr(Tg=sgTu)nWty$i%`^HADG&tX^ZWZ5iBMJAB}gp`WDRaIgDous zYGXa}7{AD$Prg+)eV87=|v4_w6-YnK9W86~YKc4I5cE zysGrR1Nk)F5c<9@mYIjZx7=?D!j$Ffr{xSd+vsfhskb8hqp`hrEgWr_jrp zfQp?ncP^)w@>p1FWLw?skPu}Q2e85Bg|MdGU$|lYQuO9gRI|!+P z_6XR9jD3&Kt5{hW#-ZB_;*um>F1pK_gh`Rz z4&#^Q4{)uHN$ahuj90K_;`PU&Z^n%w;@A27>0ss7q1~!?a~zK}EO?U2H2>$&kTDQ$ z3TsMrajmMdMJO%Cwl>LNgHAE@a=65dVtG^{p#A#FO#yAuse~R|nN4~xw-B@5U6cB% zvf`FdTpT$A2wsJf2#w!-bT`OzK>2ZZHW&X>b<(E(0_%`^FgsBe5Wq4MWtlPC zEON-au1XWyyzW7mRed^m5L-(`pv8cg-RJuZypo_4`Y>u~ zfz94hW+Tv9K0s+%6uu6_^9>yaI6_aN{*Zgu-Y3YVj{YMWcp0Z328h=QYC+TD?j@OZ z@1bhJBxH>5Ih0>^Y-Hpo_`y%97ROKtO|-Ss0}=b`6O@d19qmc=Swg?*lCHd$pWTb> zrW072NY>2H>^tJL=Bm9Zxy7syuW)Y#m#Q~!mazw>93SX9rJ6lc3r|)ByaA2pHJ&r- zh*U2Qv>Rl3>oWG<0wFI^ zZj*}ndP6gM(C3tc(M@7MTUi*6`3Uubq7j?r$82b&5+(0}Gc_L_>7B=5P#fkesSfQc zzPsjM7^KCq^Ho;AF*gjv(D&`eP9U5vXU58w=>>D)k${{F`qzYn2PB_0DvZDW7y=9s0+)heHDl&@zuLy=U)Rmye@O|ZK=JBz1~Y{>tl9Ks@#pghdc>@EA3?uYt%}k5hRXr*|M7t zmZ!ZxD+!B~C%s?LU23dknrU%&o! zg4pAA%=*ZAASnl61yR3FQE9mgI#H^e*sd`D9=vFJPb>cz0cqC&&oZrJuS!}^qB25? zyM2FO5%sc=h0jPWOf*}Cie-uq#V&MYTgh`ZOMhJhKuQp~|3q!I9PH3+b`^4Ud4Zmq zh-lPc6SE6{PL}!Z|Bja76llGU%J^Wt*K8to`Lm&M_6D&>p(@QGadF$9Y*?{;c|PvI zLr{EJWv)|p1s*X95;PVlebAAmkB3q2L>&8#sJ5+v55WAA#zY8&+3?B`ws3f1@^_|C zh;E?R3d8r$d0P)AbqERtWMyj?7_{@rab$sw`y24$D;SM}cRe4|t02`nV11$dnch3# z`=ag|JaC=gc8mJjgn#2%Hdp|p!{krYK8ikyj(CL?rDNmw#Z9S+go!n}2R_E1h|&nt zC^tz0ze6`BfqY{H0U1F8g!K}ub|CEqOg{VZGlalmG6HT zzJ|O7F$56d@pp@JVP7~lK3Hg%sZOmGCilT=hWLbcy+OUxL=A3!)~&g=bMCDJ?c12_s2a-n&jJ3de!Rqp(7NgdK; zaQ^QBA|z99v%}a^8Im+0aOj2Tk?Ag~WmWnMXb3?Wa&_USLl6^%pRQCQEyodi?1>(j z6Ou#`1%!$KxUwzkTpP*t&>GFG4OlfXF+l;`2z;ASWGxnT;ly(=wkR!^6sE=lvDroo zUKuN+%ry2}LH;feK58L1`!cxf#}AW$N1R_}*i0^6AsOvHMt8FtfFCZ7uxG zl~o8cFU2A98p>aUMG^K4@+0Fh;XiicJ`nfGyf7g{Ed+QRQQS`_F`{fg>o{qFmTO|{ zJWAW1|4se)Ge-w>?fIQZKsTqKCW3a0PfF^m)Sq6BMfyL1<+>R# zX%XRhwl3o~yvh*4KG)KpvXJ9gNJ~xRCr1oH2Z@!YT$peoQ9B5_I##he*qhK&9O%3I z%@2d{-}a365X7v*>MEc=r+hxv+-6?2&68r_N(ON0V{$IUdMc?4<7vxE8I0u!jlvt< zi9Yw9!1nLuHNZ}Yss{RH>%0!4Dcnjx&H6E5m}V}FEeeJ78Xx7=;z9u9W!SK2G+qZE zGQMJTCi!5nX}&-{;>hSI?b85jN2WrvFHbme-~!!tardmBPKm)rfgEmhym)U=74Z|= z(2+jG-wB+Uw_^AiMlUgZgP0mjk}sYG{e}xlvfz(71A}gFks<4_BqT9p3q;mD)~Mu} z*+<2Pg>otA{Ucje4~Ws(UAw3>eA=Xw*5iqg6X?F3oE$$qJFyG(Q3*HvfK%uOe@N%p z-~&U&uM1H8YZadNn*477-aDaPC(VqAN0b?bHv=r9fl+n)+L?ql64L-mrm?eg6BG6a z7*d`W$xuwBaga1urI7jiC>Ifmi-4 zU*~tLZ45>zjAxLK%n`{*?FjY-u~-Duo0c+K9OBdRx8V~IXoj;Fz(CuVcq2dbQ?voF zMD!^{&;Z3|KRWkGJt90d9%>&-076&`sQQC|cGU05s(>Ct4l?z>&Tw7!nCaMx zZc>oIgZ3Hm0VU-Te?e~UzaTfWCwMI}Bwb)+hGmldq9z~Kqpa#~L!mc*ck2Xhe;D;j zv1@I^DNtF^gstw*;}&{7M(cV2#V@{H^rFN(6^P9Uo);S7hs%Cemd}2MdllV5V1gq7 zOgM#r6s^we7zA7)g8Z$AiY5l1W>->LA1!#nG-%%a4u)CqNWca85`)JMtU^`|=5dC5 zGbZvjqzC-NV32(SWdYi{(of-FO`6~VY{%gW%u~hT%icHmzx1EiYLw>OEi-wDo)Di| z3sM3|4yfOtU;xBvIJ=Vrg+^73&*a?Ox$`1#{DNG(Ak)4e6ilnCCQE6?fj3deZ8>sr zTt`9mUBYHpg(6<=io04iEEfK)8O^JN|FR{=k26tJ3YrxRf@PQ^oUfu00X;-)335%L z>kByHQ+b-IH3{pGiiXPsC1C?XTwu5|l1YTW0-yFVS%jbnb<8nSTW~WSnr2K*d-e78 z-D@BO-i5%8{^N`$_5J(yHPNsK&eSq~X>95;jJUECkJ$Gw8suYaYu30_1by2M3wkF) z>;9}t;1CtezETFD7=pHbwVI%=_N9oh$5U0>eBRX$nl&pxFnUk-DD1^qqSzctsNKJ=>{DgcT!hqlrQ>$d=d`Hr?`-kUyY zMM8f?#eEl1Iw<`B^CqbX*Pp9Og_c}GVmUHJ@zNF$QegJX3R`#}FCa_f4}+@N$qT@UkD3@w+6=K3%12X8+20L&v+JLx6nz1m<;4R@p*?v}0Db4p$D-!>qV~ ziK)?T>UYF47Wd@!&+E+Fvxn=O@pcA$4{^MY+x>}t`ls%Y>gf$BAhti^7q|eAbWK<$ zz6?t2hxW0h;NUf^6p0f=_l^=F4!hfG{sX^%{|*cYIH|2RnhIb3Cs^W?d^m2J?J?uF z#XJliL=>+S(dk3KJmEJlC4LXW?cd zffp&`VD#E9Y1*Z7A5{(W*t>JPhJD+l;b_23Vt>7)QzaB%Rx4)0BI~9YoN_2TNxKcD z5y}AEisn|^CN#3+qSw9BW}k}(nQU7Xe>N`e^jRa)aDdu-Q==zXLjnD;^zM53>W>qB zy*6VHg8@Nf5!8&huV)dh2Gt65ZOCSvxnS{P%2kPrn6MOD|Tbd<`=uccFv z%Ya&KJSV84d^J1!tyV#yu{8Wq3UW?ucIt{E=L%bK$XIhtj_VE}4-&I#oj%l>r_TO~ zuWYvR<1pF#0wU460HQ%M!VvZGoPeR?k%P8a*lj3z#|&f&j%zcdcNm4$14Q$MAbun1BrawE^T!+>?{XerI2!qN+tc{x zhyvyn>)Y1ncSu{PJ>p^B#GA+Io@%}uHgs+kejLsdinWUT-g8$NzwVk%WIrBcu#xIb zJ_sJ@jioUe>e*MJI#d1#%=oig5idGiQH@ajn*OidifM%uaSkz>i4?qoB+lqRzNg9j zL$fuOEH+|T6!YusdxYx(qO@gOkz+GEyihbow~#d)oi^I1EbGg++1B{EX(~IrNJ3 zorCwo$ov>|9hq@U#_PW~I+U|z9 zDhO;L{v}Naq!#H)fT@u~me!S_;Wc--1TkO`p0o|#H~nLoh_Hf}X>jlnB#%_j(KXfe z%Y)i~1UeBPNYTI}{~XErp=k++wbfsK($o7kS7DP`f%!gu!J%2OAFYve`zJ-PTEwbWHXlH!h?%U-RGM42ubRb*M1Vld7Xv9f9Ihq-S z^Dv8}@;_cx>ydFfT_Jh1RV`Wr<!o{@t^2lzJ{^h%+Sr>Ts8vmC_{<#RSjPs*#70Wq;^ywRi{P!&!jZ$c(qHu$y> zJ@vHc36a_2OMdD??*I;@f$g^5HHQOye0%R0;{nY36qcjXBkALX>_WHg4*FhM*+;rDCVcjW<;5Qy&(t{FB<=yrB~DT zO0QW;+t_eY#cO{-w{Hm+j=aqCH?vJeSv)H1?qzZs|A8)nBtyWaIg_f-^0Gu(RAfV1 z0RF)=FWg!XKLWarW|xZ}GhCoYL>`pX{`UUiVF4^y2`R+#8)(4*I*y|BAlOEUuS8E} zD#XyxkaX?@i-HE>x|nRvOJEE?|2(MfF}{(K2xtK5$idmK!kIG!n!#89XVrAi!atVf zs}2h~x=WK-f5LZ7Xglud+8BK%Gq26)9*LwRG-woy-=cv(oPePV?!vkSrQb8Fq5i-y zQL3|NzgmR@x};FJbeDSM#;|_3l_3fscupo4u{GHdBZ5f@UXa(LHtkl@41~~eZXVroVZ>>~ zkZ~5Rg#OTd0i!4|B1wfeI=uI4X|pF?t~7aonnPw2v?VGoY^iD$+b~TYM!GR-4LPhr z$sGNSrZR+3W1$x6HO1k9J1Vvvr|}5F0th%Gp#ie*=Y~@~RwF{5wm9TkVWz}u8iV?= z?#aFmvHcyHwwK85jt_a6&0y#>h3voG3q!5tCqUJWBHx992!YG2Q->RT6ceO*#-&5B$J4Gpg$r3$$6TLMoUK@!HBYEq z0W>pcVGvl4mSX@H(G8@xE{Hiu}!JV$u^k|*(mw3Y*STK-k%)7sdMO&>Vw<=U##df?AVF$Xa;-Eu70?sTx3HnIWn@j0Go(25Xq@D3f207KlA zVLbO8F@|^_(SOYzuZUXnDS&n8TMycE2ejNTQoi8Yp|~WVNwsl?I*r=Lt4BI57kEyfqN*3E6|&iiPyRhU4aIHi3VH7iYHZpJC{f2vBC;B zH#aKHaB!KANe%v%Kt?-egOe#od|UP-(g?pm^>agyyc~j#(QRn_r6+brp_w5ThsL6d zdd?ApLLoTB`*6eJ-CaBbzic}i7tHqrpAdYVe5tbLRy4Ct7t<{;p>Jzyeg}1Zy7Cf( zBPRYN;CeisXAswsu1_g(=+Yx_9NX&I0(KFV2EZP2i&g)e%a$<#c!cHAc2DPh-1F-; z>|2oBm^#%!Z%gsFE_wqLDucGzSgIbGKsrwjS?Sp<P+#= zS#&8zAU#FiH4L~uao_mh;Jv{#UPgQw^kIOUtu-}Uh~5d2IWI`tSMQzPd;awOX=_2b z4hwT&m8-7Azy*yWZHe1B447FCRvL+zg$VJx^pAc>Hq#=|4eTR8wfBbBbJmy`{n>QB* z{V_{kfE!j+VB8*S7#*wj`%)@xdp}o2yl@_!7DAzIU>NM#hmq4 zp*v$}Z9h>coboOXL0=xt55ul0Jwm`KGWDGZTt*N=1K)zyPfb{1b#?Uo;9+>8{>GZ> zU7INvl7k!oyjH@{ct63kOKQ-H8X^tPBxc*`4PbEwX`HZcxW)=c9ix!n6|10z;d%hir9#9Un)vu&CrRXNb3(>bO zq^AUIFmBf_EeXgAN4L!VnKEjq)8k*hX!wj`)X@|+3xEDMoc`Wy`A?3&4BH4%1G-=V zUfwwx8s&dF(_$egKdCbXI62cJ_vVRkU}2c&31&s1Uy)QDex`RT4Q{BexOFO&XANP0 z2$c$RvCxn@D6S|z!V9~)1kr#T1&jD&gLZxPx3<$0$))7*Tpff6=E4Lcp9=eM1 zv3j7hFN|<@R80V~ zMSS+vl|n=NZh=+=?FE=gSlAEvJ?LgWKHwskk^9tYjh5^L@A~hzX9Iviw*A3(nMx-J6?16TKVK~An104l)NQn5R_eDiQS&^87ltk?1#;G zcyy+~F@7|>iuj*>qR{$K90Wy?;Z+*=lf5=D?y6iH0cGb~S`P~vjcL)Mhjw)W*z#@_ z+MOq}m8R9;Kptf~zSg0GTpM9a`sUfL_=$5vfI4fh_sn>2|A_5EM{H4fe*pYJ;MzSr)=!v~rxww-2QqK_E9!v-#FrCov~l zJ*@G`nJ1@n9?k!Z)JGpmI}|FMB-AGQz8AJ5$%oX@m;QJcJm=qZxih_~g(dXU>?i`< zVPhS~>!);Wr(DOAI67xaT6x;9#@Kq$zrvEc~12^@YNN0J`( zt5R^CyFbS-X$aVvMk?y;AvdV+Guo0=txDV{QuwIy_I9L$V*UEUco7k1&{c7o1Iy(^ zTOfAiNZ$GLMxn=+# zi+N!N$W-PoPHtg$`1JjvD0}L!qFk}|*1b80@gx6oGxS@K=n_9664@9#k6c7HC+Z0@^>u?sUtP#v_FCO73s(n4kPYWuTO{|4 z_P1jf=ZS9-=Mmu#&IdOsNwa1Q=Y?_EVYVn6bhtEYu61~Ayl%)yDuT7vlx@z-72e9) zyL`_ZY_);(>;-NZNsD?_yp}$`Oc@|ZfBuz|K71Qh(i+gBYNdy7ARmJtg`foHgzruX z(>Ol#;+se`>$z8X>3MAWYUw-UHcKZ4I~~IcXx_r}!u)re^_!TQ^fjl->yLWuwQ}|6 za!cf&8NibD#{1T{o*te1?7@*t|0u9`uZ-1taA4!D75gnW9lU>_wyy5(w{`39rHoFz zv2Z`ix{vj8MAg8cSxEK{8`D`k0NB|q^)B}Fh~B}7*FDh}kWWl9L?<`FWLMQ4?OSm$ zejYjD@Fi-(6ioq^*zmDG6UBx`b4u$$Q%$6&xw&~__ay&Shf%O5yHIbjQteJ>1}+NZ zvM8jbr)Rt0%l`ejx3RqBXUxkQ6l@&Einu(8C@>}ee}w&MT#jq|#t)Y{V?ySPnG2ah zNfA+@On zx~c0r&*RvKZQJ&32hNs_4&tr`v-+4i4%aCW^y}h$@nU18=iq3cl=6UtpWv0!o0tAf zomXDYE)$#S!xd#aEQ7m)HgfOn(C2z=>VB-vxRW@c6vr+t9D&HQ+gsgXo{b}(A40bW zOQ@tVzwY)}pY{cW)F(Io0Ppq6wB2lPR~Iw}frdMZroP9U=BH=?GR@`Kd6W5GjXr7sp($p zfo){06HQwIu}7SwXb^VoDXn*RRg-#$H*J|+xYZ+T_p~v#UxvA;=bW5(t?a?4E%kCn zt=w1J({gw01amY!vJZ|(`E5W4U)9c?nj3ebETA}p*Lk$mdQ@K~n1rkYt~x*d*!Yii zjUu~)Cyb_sMUsB_vjvmhvj{v`lRG!kcXbN1pIv9W;g8E3Eq9IeoH&*-6WW;g)i+nz z>$-e;^-@;W(Mb*Z`0=Af9aCq$rK}~YvwbUp%8m0>##F-MFpIDcN7mo*qJ$N(7_i9% zB3@W=)%~2D(^N0lLFX8x%YXcv;Sa#??NmtJa_+n$w`PY2{Ib_=_w2TVX@dZ>04Zt^ zYS|c`N=rN3K|?0%C_+-t^!~FWMKym__e)WK4n}7~g~`l8N{<)xn*=kmPhG2xt?Qt# zr3mR(W{0Y{C!Sw&f7<-_Xdk<|_cTnH+5PyCl}~;2e+8$%3pII8vKB=-N8N41p+sZP ztGN+QebJy~J`B86xC-x*gsm5k==u4@?RTA#Sl~aPwWjfg2CGY*7aG?qt4M!eF5bh; zjmQ)y9ky(rC0g4hcI5fDmjB4&2UFs+3NC-d!t7sWd*U9rcW(ll9gp6MRU9%U>GZMP zj1>(l7ul+|aXtOd7~9AR%lfs=Eis#uRxwUvw$a`jV}0JXd)sK(vuBS)Ov+9CBoiPX zwfEgTVUgkh>xi1`8WGbuB-q8-S@b?KPsPjq_kH=NtiExsVzeMDad<89EH{au=VemhtJgs!ySKq#Ff-@H+r(|0!D*MV>@{#orVD2 z|3oobRE(|0L~_;D{;XQA*GbLTKRvwG(qT($PTJnsv!*FOM8-x+osFJY=`_V)o2vpB zW3IN&%)uL*4H5oWTxeZbFcZTa`rG&{Jh44yL}T2Qv|_pAzkD)xQ9vvir5YTMmUnlD^PWV_9nL$A<8W6%@ z9JbzV+a_VTU}tZC5BV>8Lt%hve9zYITtnXLCw}z{&$n41OMwPyi7F=@9C(2}w0)T1-v}kbMU_ovm2N2#a=3 zhRbEH$MUUyDnoZ!G?JlJsy69;iw6}x64OFDo5=mShg;A3a~{g3#CbHzduNhGwig=< zFgA8LpdVEFmvX6=LDOLCrcYs3d4-q}c_c1%a&lrmHBjqfbc*rBuwWronD7LD2nxL& z!e6-^+R}Cx&nZ4joo7J#G~Xr~=_DzD8JvL543kN;h<7Ofd%|s=apc8| z<0ipPrs39NGZ{W)Pgh}8#Px(|zu{ZPq_wXT_ohuSzpTA4;#y{S8<#__{cm3Dbv*NV zhDWz}jr+xKSIph3XBtpE*l~Mp`})l^$G2PxPaqHp&EB|v7sja#n)(!ElHk(OHaAjG z6}xTyNm}4W)?^~7Go>{ev2nc$3q$ldeVWS^0kIhG=0_gyJyTmuuTXqgj!Mm)Q1mm2 z^GwL7X0?TziTTd}d6C7d+cKbV)pf(~| zQF~U`VN)Men&MoxNCaU?#3{m;wkff1i>P z8;IMX5u2+ z{aq(F-^uB`Fwt{7#kII8ky;4aPyC+C9%>QQ5qoGAeEI6F4#p2zex~ydX9sanJJ@Cr z6SWp+7}`sUr-4BJvyqHn84$ad+DUfvayoXx7M{&~-~Xf;-}uY=J2SG38y%TeXeA@& ztXxHpG#vBC)2Yf*1eEIb>wj(hKRwPSZ`S|S{uO;Y6_}or<+wY$oA;S@1g>sHlkS&- z-<7X#+)g(JiJaPVgJZ*%iU@*%sJu@Lu2CKG{j5^Ge(kJi@<6}Z$t`k-Rh=o-H+x-C zoHK$AK#V#~I*d3EQ@nS52`}Yo$kHZEtYz1_JxqDtr5RcyfcHHfSH?Ylw~jG-*75hN6D9DUWaj->-7j6{9J<`< z*_XJtox_{C9$Rs=Q03sA?Gs$?yfAg^{6yp4^Al~k&ZT{T4kR1h6g* zy`uO^=_z$`W1${r^gb#XsS==E-*5`;vZsU6lTRg%hzOoVWSwH8jUo#fVjm@M7H$2R z%n@JnYI37Xjg(jCtAyn7}|X=VPOEuUPNHyKb;u84I_CjZrqPC);p`AC{zE5YK#cp zd--p-8_hoFl75ygaaR=Y^OePulX2{ta(>*=NY9Bi2F1wxVWPSov*lABK40WAS0fZv zPt5z&L+|uls=mCs^*v(!3q+Yf+`DyiZm!Hyvpt?v$UPxM#{ICM-%-0`&?zp$&_Des zod;lZ68wk2xzINm55EUelH8R^e8Q0|lLVQFS$)%V4Z&3eUx<9k$?Ue%ICgPboc^)4 zjolCKz7fG|ObQCD1j_h2mefxuHHVU1^ z*((R5Ii~%~c3S2THA<_?m6R9YMVt^H{b}fc=6`jX4njjhP*yFBNJRX&wJ^0wVBuUm zBoJE_U!RbC-K8W~pPg=xIObuMn(HhJ5ts{?hNQu;fnLVDbr#)h!d~5aR`BP&?iknV zU|n5nxdcs6QhHXyWVGS5p1&saDSTW+ zZg_HoQvpi7=$fGQ6gSXpn^!3d+-b#8$71v(AG2B0riK_-A<%0=CD$4BJNs_!Qi~to zhY9hC1U$ay*e4ua!p@f6yT@*;ET=iW&LPv4-4(JuW&W*TqdhlIPxQZMGjIw4Bd5vynpymrYsN1IR*pHjWjiVRrKtCQ~3Js2LKunqR>J8Tn|+IC7WlHipy8 zbjNtMHb2+I(z-*#+d~S5Z@I5wdScK&_2#r{yz*_sVF$X^BW5j}Gk z?ekrOmjeDHR1^blvD_-Mn9O0s6b6V^Y);Oc5xaM)YJ<27q9V1&=TkJxR7YKqg~%Bg z0%jY#V2^u?$H*B_PElLRu>B0R92dR$fPAdi7>@!NkC^TZi$kh}pn2sGhJ84hvem23 zc4f$HpetzyZ@P^!!fHqaqGX6rB@+kB6+A83O~SLdt*2=FVnm*uV+JV6HI~JEd=Xvp z*y)aCDkPW?P#ns$wEEfi9;F!iZ%Im(DiA9tT9aj&7RWuWe3GqyUA?3KC@xh?x_EY8 z)}lHfv>or+@i!fWNCVkdS%W4t$(k(~$&ipQ_QVVh3kJ3Eg}4SdnwoTl-u&~;RXJlA zWg)IFMg}Tu8$$-QH9Wk>AMIQl1K~Ubl)l_BIW+j(5Y$_EESMSb{Nt4Mzvc~Q{SD*o z9z=eevdC+uOlNvU9HzueCpIIMfVoY{I}=&DXsL%bo^uQ9r?QQ)wK$nE5`HV z&OZJ2AY@Jc4?o>K*LOCv-Z=6=m-_4Kyel&2Z|;v54i>#CdVz7i$8~S!6#oOQ z$EuCndaceh=b`%W$;8#@Tv|YKFu_E^DOcKSxcT{tmo)T3lb}$u`L*uj>DwQM<0~dJ zJzQ4rZ~(|}yLq)L z*sxP1D|C$~hWIKxhV>tTm`9LMZer-(mD^*tBwNcu4w{}jPd0fCbEpryM1^`VB4Rl7 zSB(3X$>mGY>d?q|DvvH>8}q9~NJpl1OP$#%ZZk{ny6}kh zREF_p#q!l^_zXzi2een$i%TZ(r1c0i_edAY0rMgT!{QY%jN$`6^b@kgLUkdcp0CkW zKMWwRjA-GMGspK{eq(~?#02U_S&9y^@2%eBoeXe`Ut{SvngR<@%?tB=ook5|xmJkSTmj(0!Ec*&I zMfN7o4qJ95e75CvzJ^hDeQD29OuD@Mja`)g5}ZN(m=XVT<23XFR<&hc6|r?=@#~%$ zDxaYOcr;%L{96*5y=0x=pQVK|KOqW))5#ZV4~ZHNw#QrJng=ggG^u4Lh0gqazBUbN zF0WFSYM&4+9ySaQ1V@$x-y^T|m>;DF|6La4MMnk2Ig!n(rKP2LWH@QwIa2rA?ol~x zdaUF{G!UP+N0vjUxZ$*#gm1B%mP!o4TG@>xP!#IPb6}BHNyjJD*d5ovPPd)lxlB+% z(I3iKH0oQ`>;Y5TM3xTrk4?@i^Zqf;Y~+PecB-rPcYV}md}x;`NxR;Cd^v_(Dzh^5 z9cC$0o0qgMKkLh6kKjsFrF#z@I+Q*!ICJXF)${1y#-RU1_4vu9*N$;V0IT9pxBtB4 zDa+$1gEwj>toP(M%cu+n3OZZV*Rf~T5#Lo4b}&6~(`*35aXNVnT#js*uUD^LOmcE@ z_zyzFWh9aCbSq*Hrf5cyE|Q4Z?yrYGAYt`j&%gIQKz~jpK(`!W`xoyS>*i)$-5aDS zVs20=U&FLS++yf?da%uto9$X0d!qFGVE~oa;IlFAI_B%|Xux77TKV$0m4f}I4b@9= zmhmOP;lYWv?0IC9FYTb#EQOWMvW>VR05Zrb)m%}&4fTL$}6smSCue>&~@^afM>%g5O6s`t86=DEIWwLWz| z+^5N!9*4SiNVxK}-oZQf!XA8ncIb}L0d2MIr@i`Q4M_JdDU`mSoHl1$e`7b6l0u2g zkRepTGz1Qmi18}%05W$f^odASGU|t0zBJc+G9T~+USEIy{CU{OyOi12@Zkl$IRM6C z9m4(u86f9Qm`&=}P0#Cd04|#pM@n`hKBEMANSJefJ(sp`EQnNC!Cm)nT?d6MH!9o@ z)H8;dpd7lwSEp4&=k@EC+Z%u&XSbWgCu_XBCR@Z3tsd(vyO?Fz!uTx3k(glsz^!O9 z;Lp1T-5aY@t$o^abDOy)wbzdF77U{VW;jUnalFN>7d z(}&TSfhH7?*-UcBGVnkbyX&8sEsVOhHJ*ISzOXT&7K!Wwzh^!n)EKfzQ#KcYqg0N0 z*7tr+5^m<80|}a5sb2T2!_GZz@uk(cWnb>*{G%T7(nKjPM8|4a^Sb>jZ0=q@J36b& z$ik7fiiM5?1SypX74rQ2gr57H_P_%hZ1xf<9`>dhjN02qK!K_<4#bxu0v$;;=``BS zWY?otI8~hlgC-}|@v_>uPPd)Ba)KdOyYWa$=qyYf81Gf_>OLihVbDqZCd#`-2EvxJn!xtW}JNl`{qUnp$ z-Z{LTFpM`X$3phYwAQh5e(v8bcTv&Oe1a=hiPpr@vKywydbI#>6Dd0(t=1jOzTSI{ z_O$G#e6nu>5aZh?L2Xeg-nAKs032^T?IuTRi0(=zhtfW47d;zybM)?nxVL=i63a==cT#MYl7Bx@R^oM}G*|r+ zuBJpEYdjcOkXm}oE9w2BkStm8EruQ>MmgkQv%Wp*Q~GFP%863KU0e+r;DE#bd^@7g z4(G#^ofO`q5wc44f_k*<26JPrh(rTdJUKug_t5B)b0TtYJ7i}M}Mu!6>FM%*1Ti5J8dOsLuZk4a;~HMH@Up$rn_$sXsG=NBc;Z`?spU;#?z(@<|1+C^7tA<-tm<1CoU&iIZke~B zjP&@i-M5DbRG?}!lPZsiCoMIySvPm}Nf&$6CzBVFvt*-A?B}rbsXU1cQgal_X9pp4TuBcq+(55gM_gsP_U%VmR)@#~*yX(4qprh33rt0JP&ZBzl zGr}1MF=jXPyqxU#oD`@%WGiezKkV)8-6wKiOL1hDI+O3c_t3uBh4wzW%|ATt*ZR@h zgTto$IDdhhBm6CyFY2TA{+kG0c&o@^4SB0{TKATR#V=c`bJB)xi!?|x(@V{!^A1(p zvs`UP$7A@lp`V_yg$GihEf``aqHN!FXONDG&BUsdQ5*lg!S6uN2>AWIrG5(uBXqf} z!u>bJ*mJ%t;04^?ygF;G!-(bF1!-KU-xl_373zA9jzRo1h;-{$s|J4e8seu`?ewZz zQp4rT+&@WaPs>nRGYP9c6V|nDSchwX9*3^5eqK<6_-WqHdr|u^9zTc*zV4dDJUSlu zqrtcfIQ4A7q(FXo+Pig}yKS$SFv z`TH7zpwP)<#YO*E@fz2arZxXzvvci&$m z(D$=QmuAj-*McT)sflZUjmTrM`%?C|7JZcr-bn&xQZL(g>v~J(lovo^2K#x_DR@nw z?&ViAnP?7@OG$E$FJzZ3Iw5)F$n;B*!N}CV%A@8c$9qlhB|sNRU{*zm&JDa-Q4`u= zl02t!%Rc7;0m>tIDX&pXbO#U?oJknmWpgMPe*gYqE5;JUxd}7`Zhmo>xpFfdBjzQ2 z3`pq^V1VtLM{4f*O~KPD-{V8f1WBmQi}iPMHR}zei=x=<_HC{rY7@?vRL*0QW(AJe z)23OodsI>w!HoaZ|4f+e3yn3Ny&sb=hpyd0Tv0aPHo8h63mISIQJhV3Q6e?&p*B;x z@_Bg5s#l$AS4RNL#yw}Vm6iuGT8%6AWxYOm$PZ)&P)9Fin^FuCLXd%wC%;$uRTx13 z!t{uCd{w@s*2@eOsnoeur(An+V558dF?Hjz$eCgyhMkZ3p#$0+l%2DLm0mNiP$+uT zp3^C=w||fN1mHdx#AA{}uxnxaN0}0}(N*kZK7V$-6z$ZYQ`~8@U)t|}Zn@7L5n=`K ztvhS+XkZo&KAlL9#K5k!vDli>9fob}`r7q1!3lT6?gG==3|zkTBw9ooaC?MAz=fw@ z*-tIh7q4!{Tn&j*l>UI-%4 z<1e62RP0BC6oCjM-55Y*Hvi;M-GLf)^{Vw(iC)va5XjJPMwW?ka0ckkXM!ZjuW z7Hvy?%`i>grQj&3v$MBPqt7OJipd2P%73MxQKZqQfUDKXY47K53ZN@Es*lc#quWRfd!Q40G{Z<0;vRakp=WX4_iCO2Regid6{2~^V3u0r|jYj)rr zEC*klS?-qPSZN&T4#pqZ3*9_=`Gp5?a40LA$o2)E(|uT8?9%Hal8{xs95~vU8k+F1 z=wOijmf{gRnH^s!hPP)dudOy@z8ld-EJ+v}nAF_3f!?&52E4)sUGFr!+2-pkjL!*} zhXL`owaUmOYJYjOyOtVm3S&sblasy#R_J3}L5fyNjUG|-9!S?;q$ML1#b6Z+iGZ{E zsE3SY0G62n0oXG9E^t}MhddRiXS+I^z$agljSJ?)UvMM$&ywF8hAJ7 zMuNAjhNcmmr*lb}Pqv#XYb~4Eh5n?U?BA-P+5(D#;1~heVA*?EM(PS<2J(R#OKa-+ z*aa~Q`uX(OSMhDt%^WHyi%I7at@0VNkfpW!ikd1-yrxCl65_mhjp30OH0WA3KD*q^P_% zTmkQ5Y;fj^%x4SwEuh)(Ubhf|aCZK-Q)gpe zt|=w4KL#7=`1`SC-|hC_0_ni}!!?3<8qS(^!gP4y6eZ_ znKAFcq?XExwX+3fpf@1Tt^&)t>Rh%6P0G)^epeNhmlEuAy`lXQOM9}r?^`u4ASsi| z`UA5=xsF6Zk-7_ePc!>)X^1Mysx6*8QZ?`-#Hy&B52mLq=qLUqqA#K~)>J|vY}vH% z+E=hy(cAKwH3tq`JJM z`tEy%h!gPz9!O=e%q{8oC7O~z;05@zv^QNE)~N}KNRAzV0h1`TA~$D1_p#lgR%zFO zjF)Q`gk&-9bN9uIs_@Ufi{3L1V8xUPLWyim-6shN=y$$uC9y@X+5ER5Rq4|Dze=Ol zps#C$V>H+_x53?Mfz4hD-b}e5l$yn|zIoxAvGaEfGc;hiC(rT`-7a$o_mYz*%+Z_F za+KonLXcxdH$5qcWx$3AH;V7_jqyd|^OYMNm24XHQG&pVOHWVNv_HCgcU_Xc-Mggy zUXWW9_w~V6DMgj%s&BS@Qw|7Id6HU-1R>URD3e22hmIer$w?hQ(+GrVR6MPOA)@W1 znXNtJ()8;k^*ZIF#6ult#x5$kRghF1^OoIPeFoiKz~C9Z-XTEV$smD$tG&41z9|dk z&+|Q58L4M9YLs&5x7x;&fx`%i(%BLQWQ2$k!5`w>+0}K$^g+{>zlkTA;6KqFj!a3l zd+$%1F>u#KZestgdkiKn*tgF+<8mr)t!_ISlQhl>Z0=dlko6k0X?Y|9licVKKwnv_ zQGAQS7FZE}6@HexVJWVz${8#sv*V8c*#ga3tDX2)@-lRgI=*(E& zK{2=}^&)LwI-|!#bSLwZW-NoIum}=l9C1h2w5%Q<93#}f%F^-F>QStd@c0u0V9|72KbsD#d6v6G2yVq?|Mn9PikW8ZlS-aPcyLE#eY&zu!6v0k z85hHX;Z(&21>H`(l@8mi>u^V$zr1bA%M;pEUNQ%1Me9Ks-|BMJ1O{U11qsA#O^3yc z76IHFFbLT(UlFxfhYU^ou{oRe#+VR9 zAf|%aC3;OvYPv6*4`#mvK^TF2?CCDVU1{URDNNey|6Re<|HBkwiN!KZJL~qkz#GX{ zI?9I!m|T3_+_Pmbfey*nHZUQuDKQ22-?dE6^{Lb4%AA{rY@@#IqHrYd%+;~Dn6O@b zkn4rp{TB6Wb#+#s?DryhgB0`S8G>$Vb*C2gJ z!o*?nPyFnFes@Clr6ms(Bin9|9Z5T!1_AgU99iwi5= zVB;c}0E0`j4KmJ(m#qMF3?IeW4D9BtJeF&@AE7bkKnDd=#mpCll!k2tJ)vy9qmcy= zb4y*;DR<%je|o+pt;+37T3B4wiv1S;3-3(NTal)qiOneNS6Z%UhTc`>Y>tgOh5y3~4L#^sY`?fvQHaJ6JNpLb!! z90$&p48;-9l|nx)DH`FKv~B3pEI~c@YtwMaDn{9*8=Ihx*YpO&>@=iAy8hywx$>#1 z?Qrqx2wQHGR;jNcJa z05_j#TzAlIci!xzo7hAG0CzL}uP90FRzANPa2;1;d#q}AS1S-GaO|4mBZ9L5glc`_ z{jnQULe!}6SZzGa^AP3rW3(~|t?!EYyUXVbGwL+hX8)`->-a}nH(7idgK0!$NTQxUT^RD zzUrMpMJDpq#Np~jt6=LWY?HX1zb>za1N_&| zkkaAqx7#-P%#r;_)LzT-~P-7Uk8zKC$L1!=CMr}AW7lE;gYHn5^ZNGm# z4=YROMO)cK)!)19&(5nK6|X@~=#)H^%#ShS35d%Ndwk1oaht&_g;@HUx4p^Ltd&rs z3|z6G*FAAZvULHR;ZnHk>$m#SdL4zX1Wgd{T%4ZCEen_EsTdk@Ml1iXQP0gYuGGKRU?2t>tIlEfEJ&~H7^Mir3Ltz;Q%hShX8?_Qs~ z-LtCvH~JkB2IiGLjRLil&2Fqaw8>GsZBYJlxG`x2CEtK3$SdKqk zIBt&&cu)?5d|}+-IniI=Y&~of!Vu4Z5e5_P%pXK`L#49f$CT!^&!vq6Ri}NxcD0+y z5A)T_H*egqA%K&?;Y+u~LaVlhLAzs$_th_V$2i^p#44Rn1}wRQZPbm@e-R;z;1%pu zr0~$@`|tZ6Q?5Jrx^)e4F@VoFf@nza0};P(96q(BHiFna$=3ENx3HT&Khg6P{6^QJ zc8Wg#Y*BBu9_=8PrRpdR)T*QRh7R{bHjDNPd)CRrH*HXuRPp3Lvw}uQwBnx`#x2Xw z&>M>2_K&c%>emoj6T~Y+Wk!SWDx-FmHpm6t0c4v$Rt{LZ{-5!)hp)q56h(%HP`abFn zge?S!7bbr;2e(&6AGgkbKQKDj_7_Q4_h|EAKG>-Ifb{dYv|allrOq8%H#{)~Od0wb{X z$tzfu6BaXi(Dnv1rdD8dTY$qk43F?;#m|(l z0v|Ub*O6d#9so>SXU>~?Oso;U!}%cX3XPmBaz{Y(8NV~d%N4 zMteM7y<7W22~g@K7Yv+xEk_xWXDdB0c%H1Q5X@h{)*Ko%qqh>3hK%xRFY)I!iA;^A=it;&E07T* z6Yo6xdllm9F^4sh$;w1@i_#t=Niyj8Y6DFEPBq5RDVCDqE6!742fUVBn=u8+P&?K$ z(ym+lSgNHSm_Qm6B?zuHcDM@zgMg%x6YTLV$3g9Z=kS}Rr7LLXWU@&-3#Fwy{{FLG z5tdWVN8zw)Q<45@g|8gn?h#WP4YrZi;&5_sH*>E?#OH$K9j`c>|3`(dOMDE+>Y~_u zj`P|XcL4M{aXH$Q3nDL;f(lsf+lVgoFlp(Ahmiy>AZyFWBpmMA_fNDS+SCd+N;?V# zWL9d21X-Andf(=IY22jF-$Abj1=VNO=`@?s{^|Em7EbE$X_l_UL(e&O2(=5bIWX}wc`)jxA^-a zTm46U#PtT`4Be?5cE$xIwJZf6viezGQs3yVaLlAPxVvGVDVh=6c>yF5N0s<#&3DQi zoSIqyNcR_Wiq5`|FJUS;-?($Y_B5>w8egw3naq_i%3kUU%Iz&+|>j!Hz!rz%Ob zw^JK=rw$Z%7id-qVOZn|^0{*5;D^A@xnS}SrgBTP0kaZR~fLWVF3V}$}Ju9_$=ypAM%9VjTgYa z6gO(>>J>(n-OEo&_uU{KaoJ!_q7VtT)B1Jm_@{m5{pKqD*UH58K7YJ<-B6u*R1Nu1 zJcT?wL2Ss6O?ThhiYjQ^kdP%wW9S06bl`Jj5M4hEkt(rhOW|>Sr;q2r@dVQw zzhhq1hLactPy7*b>R2;rXq#)#{d%nSmw*Q52-F^Lv&W%fG|dZI5g!Wr?DFY9-fr;R zshU6ULgss>kXuUoM#NhLf{1!$=)?%Rl{DZ05UG3OBt?Pr+J4F&E1yl?*|!~f6ToFR zc*C}X4}f!ohq9!1Ahto=*)^E5V(Ce{wybOEC}6YB4hL+|2H`49ReX*iKF){-l;*P7 zQaBUJ_~|}tG^UC|fO(FV)QC&I$=ZPXLM5yoJpjXT^`JVv(rrUMVe1xA}1hA zR0?$={rn+=CjGBq4&Hm}8Y4wE&A#$AWFQbg_mqbIdSEtXF$fhoW~r4j@LnFpHx+Ne z%DPW6c_{7re)OV1Lv8x_#Ym4WyfZS1BItY$r1xFAd_I-XJ+hSZ*V40&DO+Y({v3UOU+es70~{jFiuzNiJ>Xg}KL(kqwPGnCKHd6o zRG3`bHr&HnBj?Nl3~(DXwO*N}R8LH%VDKc+5yF{wh<#9BX&3V?;*iMp1ng|oX5a!6 z37iepQ|F^(*8gADzHp%zQi)H@8_4cJ`jfE~aABW9xCoYw9U8Wx3kRN;^@fx;i^*@a zhX64ZLst%LgRqnkx9VrW?+;LPyaHh*XnjnMM+GjfA{l*>ZSQ0^oOr@6HYwV$_4(?| zsB_bUEj@W;GMoULCVWfJdACDnJe>PH7=vQUcv!`x*TeLJ4d~D0sB8bKd5ir#yr%sj z`u!E}&(JY3Br3o@A0d0z{EcsC{83KS>oiSgd^PU}aH5^MBqWRHDf|w0CEIC;*c4fV z3%VDnq-jsv@+x|~wLO)u)Hx5f;05ecN`=)ud_Hy;a`2In;uZydx;0|uFDdpwA{h5=HDctEJnsu+)#o~@t-(>Qa}oFhyYQR zAfsGxZI)$#d6Av0p<|Wped@4z-!Ii!cv1N{UzK^_t(?TX!b1PBu&kAolHwQXrRPF6 zJ8(Qrts`Cn(14Rknzd4!3W^IV7{ZK7@%p#r>qW`I1`lG0AM@ne_PtE}err_UVb%QE z>-L>*^7kblJZBv?Zd^o0g0b9L!@r&RfV83m0&IINld>jdqIE&X(s@rjD@=r4PY$4y_y)A7Pj@Ex76(pH# zZjJua=~vm2LKA@0=e(m=@MA8JRM|t<oLZCrnCzYhy=x((RS2emH7k8BsgqkJn(PDE`_il}( zpyoW{p_}N`h#z%dVOK_1LAb96A&_P{_0mcKdM}@*7nfZTd{*W#HQ_rLRC(=JUIMNr z+7dckWt+vb3X7Uy)b;hp53wvq-q65jNy@aEeI16ihb=xw@@0*Xk!$JHUz_<{AkN)* zv}Gwn;+7KmWSW!wzPMkjNzJhoG$auSyn~>|b!f|K4K*T}&I6YJEta+_RH@!-U zaSCp&#^IK>iUngC?g2KEshqG+ugy!$=QX#>T|KakhzO~hvpetqe(H8Ta=)xH@NRwJ zo%pfh5?o`A*1cltN}>@VZii!viJ!*&cHFS*YW z3i!DZp%|u#v5AOMzzjr21+(RMCfwPU-Uf<9z6_0I2gZiW>MW-3JB36A<4VB>iR11I zAex*TDN=JqoVWS+hP<*?^ESR^xC&V!e_dRy1vsV%kVAt;gED9Ly|WvuOvYuE>=4k3 zXOaF!IsVO!&aQ$d?w#%%%C4uG_-ty^P{7V@BHL2`Nx}QyFnU+PX>Ftu!`bQ$)e7L zLNjkA0&hMG2|AF89_w7Z7#rf2F&;FCJT5#k_k~5|DLkC~RWl*OWYC6$SB4HfW&WAX;^lDJCA;ouY zdNKAbuc4ghIXtO2r;z}!M=W#0?biA5(OE@zMP`U>zhk#6@4TX?3}i3B_?K=v z1V3;j7zLpNS~V2<7sf_31;vI<2#y~2Ii z^uiSkgC)`*+_FX`I|_TWDmpRNI;1Qwwx$ft7FnU>nU=G$ScFILd*#iOd}_Y%6^qS68IjG)t^uN}+-kiCA9DR+-&-+3RmR zFgpK#D)o|jgLJR6-AJHnl*QCYDqAv^XKecfs9Q}I|BKVLWm1nA+~{ z`wa=(EVik0Do-&6{DEx2GAE7sjO72^SEDWxKVEs_koI1ovovEDxE!_LgAK~GM6pJ$r6lutSSSz&WWIDGPv$FUp&3A$urbTZC zS4tek-3uGpVktLg36(sG%dKeG;U*6a8Qk^^L9-#PXM9pp1Lt<<{SUcx&~uKVT2{`z z_2U=EWbeU)_iYB={u0Ef4r7(=#zmFcFA^A#=^rJJ$1eeDCDNL#e>=Iohwam=u}glW zFL{a{R@nG0JG>)ckN8!%ap;SsCR5BdO#-8qMce-+vLZ~ZUshi;=Y)W*ZoJu-Jo$WZ zII;F8{U7aat^o~Mg9zD)K8QkhZDGm;`oY=w`p9ld9tYX%_CHT{CA^M?VNNfsFq}71 zsJeWQ5NO@-FN21Pe+Q#^9wP#>y3;gGB%5L3W4mEYk3n;K`W)L9vFX=s$3_{(JqmaB z;tek8$G9GS3pS7sMHslm~MSWa8W9=C-wrCyT2mK#=6*e-LzwNsYSto5wQdK)WQ+3f0<3k$04qIu}I$ z5Sh;rVmKwNl3r`4i}cn==% zjOpH_tCv@&GMf!&Viz@4)4B@$r^cSCzVr0QVBognEcIKLyp^b8_ku)(?J%=;phiVZ`Kxz%Cm{ zwMg^-we()fG#3U;Q>)PDlX7QXJC zn`58yHcIe1g*y*_z8I*PIBL`=UiXqSwR=bt*C%lHNY5m1ri8Rcy`@hTD`FXXB-6{D z5+GKaAFgTl$MKrl25kLKpXm#XB5FFRF_?2k*%{`@atJCu*?eBVc@HThK~Ce6l1$ah z?)OIaMlZCOF)*>)qLaU$b47-EklS!F;f}O6n-gNI*p!yqYRKw#h+%%LRJ<=2C zu&W;&UAH9|5fO|hgG1QWsOMd=?m|M9?+v72-4GR!gB#qj3_iFj!|Z`c;-SEbf_FtN z_rCLh(#g@r`OZ#!X{?oO^4Ii(yHK@lv^%Hcu3yYovnNsnHe;oQfnU_Y2=9B&o zVphDW5T)Ae?_6|uf+$WXJW78)I$mUyM$@SrTDzme(QX3=4xGKMzbBt4`SsIkw+D0| z*yo?QQM~G`WhF;SXc?vA%E04u2Aob7l$em?rZs19(2nrzc#+UuLs!S(N;Rl5oyh7=J4PD4tb=3xsS$AaySiWQt z2pF>uYv3GT~(5MY-#L3U;Y{T({AZlV1h1kZf7wLLw&RM(nU zdz%wj$C3AQyTk|G3-2;Xyq1divnw;xNadbma&yfHO?OL|IhBDtr?urT4rowNdsZ4 zicU;TJ|$F&UGVqy9I&@c%}`y|{vB~3JiD&LO124vUORp7r7T~e|G;qJFiaaq!|LHb zdm*kU9$F~L;w_5MmKr_W`4SM72BzUlbF@iEM|B)2-5!E{nLX(r8 zs=uomtTc*T{U5Qk!}$Ugc|QOo5i9`n?+*^P^{vsi1kWu1dsx&{Hx|@BGHAsN-rdgN z;F1QFWxNvdv-H2qXPYZNoWOLE2EBh;*09{>g27N=y&~^tmrrf^4P{;useRt24n_IC z_Awt|PXwT3EAq`7$twrb#D-Q0;YZX6T><+#?0aQND$QV6r5KekOM%U2(#&-gU>)nL zdpAlqnRz9{G8NOz-xF~T?S?2+nX=GodrdVJ42wELS_7iXRj!0{j-R==7mtM!sU`2> zSb$Ct#e|ojWZ&>M2wh<0{Bt&&9|!YicKQ4JC*2y~$S9I!I-u4|2qaR`>AkwJ)^gF* z=9&|&lbFAShwR8K-A`}tf};KunC8G3^D1ERICxaz@SG2zx=7n1v6opMdcOEM*XOo3 z05Cf+X7QPr$wVWW;99U?L1YuNi@&A}IC#-8Xn8a`B)Z+#%_MCMw-HG~19h@#n*Syl0@XXJq6Lt4g;=Fr~?Dt|)f|6w({UMbsJ z<1fW*otozoP`@cuGaPb%tvDcWGB~mb0NLnCRZ?;Sr=9G#&?vIOC^$`F;`Tl)`gS`o z^D8(me~Kj1#$&}5ZC?WNVWbXmbN0<^6uK^V6C|=U&vbU8Rjy!Q_u)>fSLJL` zu6d&0!WvW!G)aiZseU*af2*HTwcsr@0xqyb933fn=-X1f4LaCH{J;GNx^F#scjy1h zHb85CdAiJuKwQAC-@A8j)Nc(Uc>{Rq=9E{G=#SM^^glODdwPfbsqP_etFr^mB1+lRP*c^iw2&ww!O^)1F{{Rdixehd=_d4AsRWL|gP zFoUnC*+Q@RWWVsQIVK}-V%YL++0k?~{ox&!THd2_qMnoCov>>+hYe0QSdox%L0=>UJw! zscupZ#Seayr@JKSGg>r$3YduL+L%|e>rIGn3|((b^3(7KJ0YJI1ofKU*jzo4xPK2e z;?D)Y<*yy{?d#Vy5DB%4*^nM?wE`8Gn41D_{2A^d%zP%!4hiG zmFZV+IhS#+Fi+A^ zsBc!(K1&Y!#FOprJU?Z>Ig>yyx%U|rm!huJvEs;klPnTW8-ALAu2<43m+kTOv&^Vw zt&CrEjh$5;Wk>30&(o)Z?hkF2gnkVnZzppvz3dyHZ=b-jd8r#$HBDW8(|y3wi4;}= zeEaL(!Op|=nd&EIh@1tRgK?0&04Ji$1HhQVbRKcH0`s;wQ(`pMi)siz1PQ0>Ovt%H zzU1l)8ViEc)6lU*1t}|4Bsihhgyl}D9UsjA|eiZP~;H3s9;jqDNY6Ju1qN&$?fi6@5+p_R_d;%jCje!kA#1 zcw~sJrAJg?F}1Nw3{X4Rt9)FioqlDtJqExl2RGBuXsoyC!PMluz&ddRY+2(aAiv7( z|ElAu{|Tt-l|2sFKaeMB4GyXOTA?^AW)?z2G18u&s=7YG&Y#98MzP&+_^Ls!b8S&xIscC3@obo2;B*+`$ zW%?h6gLl$`;@|mB z4Xq(jcgCebxtBeQM*JCRwTw7a6eeR@vKE^%Z-S!aKQK18&i_T=cTzRULQQ(A(EEKI zwvN$C`xagG>o6s2yi2gX`U7@Gb9zON$BdBO!;q6^-GKQxE^pYeyrFl<;u_*BCfZ_# zRkS+FP)V3Y;GCP<2y{2}^vi*@kiL<92B@d7t)+3$B-WaF>7C@|YQ8Uv%V8_ZC9&0{ zgXr$kL1uzvxQY5a@#%n0xqT=MDP#JBuFD6RHEUL6v6%P+F^K(2H+Pe5Ww~o{B`tk9 zN}>(54K38P$l~kP$?}ItSMdk{u5aDz^jHYDZGD@x(=C0Jde37#W$>g+=$E(p#7nl9^AE!^bJPIpM!RF=W@{ zw(sq+m!@Qk z%jM7ezfx)DBKR2*aWd6#yU&e){d1Mzi^P5m9HbUQeMCC|G8voz@!m$rqtw?}Ie5+) zqntU7q;pNV0XW@+9 zDpwe621P?nm)V#7mUUc|*cRcjEi(WosJbej{LE?Ixu-(u4xMj5ZZy|V-F7%iB*3|m z0HE;g$6_Ac-1FU2T%81GMr}f@%MWFryn>ZOAYzXZ-#)Sq`RR3k_Emo<;%6C!rmXfpz3JTl<7J~`wp;D3k z&RcYwWXnb)kinxgXrsNy5)S~Zm16$bcIK}t)U)2O{--W^OQhk5O6mLw_>`O+t)*B` zHt#Vh&)|nB4sfn)iGaNkBWR?$1r%AkyL-ZIVN2yTh07YU?`0RU>60z zqiIFK-XSmkjt;tOwQ9CNa1Z3IxyBZqa$A6*W1YufP9id>EYw_oILM`etFa?z!jNU< z-@bw6$-ocJa=B~I=O_F0s86;0|G5oiqGub7PT;q{6z6!L4Zq51VwQX#gQB$-?{hkJ zf04`cd8+va%oZKwV%NN7b*q+aFCqAK)6;8*Y!I4(%Sw}#Coj@91b`3p7f$#fQ=&Er zQ=1zn9S;Yn7X856%|q}Fj6H!?sI6&2<@0QpM*+VtdnFLXm~Hb;+N2C81=&=&S)+qY zvb@AGXd#yaap0`BIkTu0BL4M`_n~@aktWlO)cLVwARln*R?*f!BRh_Qt

%}p#2=hu(8Q73Z=V1~>)>G! zt%XZEA|cTh4|n>gWUii}xPO-lsYaw?%7>yT(|>xzqNNZcG3*ZAVqfX60|Ws{sCpEW zNEIz|6h=wRY5!Q%-i;guOnsfpEkP18D*x9_6hvlH-6L8oyKX*ePv-v#HlW9ZHIS4A z%Bwv;>T=j$&4%bU$hj{|O3VdZ$7M+=YRC?ZFL&(dN8DwOK%7Dfs!8XXHUG1`c;QV; zw2=E9!#qpa@&*^_5ye)iICcau6BRibsLnV=VR)OLvyg0vZYkWJU0Q=uYC0jxI>&^b z#mL1*hh9F5ADX9gwn>+ZC4E^oB1=yERKn(CRv^YyKt^#n-3twtJ~*K6QRqAnT_-{GhA zw0>C>;vfcAq0!h}@_0Z^4_p@vson8+^Z`aE{-}I(y{|B{7Sm~3Xj+%}w&y?Mjsc4O z4v`v4fyd*c`9Va1CgjqCm4lEl0ISJuXQL!zAj2oNMTo67s(s&X6^4VDcMzn87eIQw z6p1ZT*v1-y_cqYpkXAMm@#x0CH z0oz=jbMs@kW|F6V$#>nE34brip31Fy+p2dlNyy zwqB#UlbQGi&V5hEE;v#{V;-wZU2J)uN#xR@t}2l3F972VHzPwaxR$xws&s}a=KiuU6TI5iQ4u+>DE^E|$vfT#)O9pGfM8){*K-(eA zQEdH_29XnC4HHxFs7tBi5{GHQs6!H-;@<+yXt7Jp*)4F2*8Sf)fv*d?9vRJV>6%+i zcJ`rh-J3lVO&6mR4J8SF&L+SK^7a2v_U2(duig81hRwz{Z1X&3$V^2jnJGn7GL(#^ z5<)1Hac2xsRFt7PiV{l69GYm*rU+3gNh*?t=e1(r&++>`e?9lHkMH)q@4f5udB3k~ zt#h5{xz=)F$;|6N@J7?CCLc#S*&u3U(VA5F7T1}}@Qtn&6lcanbps3b!vy}-?#v;q zs7xH-f2O9C3AFzkn-MUl595yl7liv|jo!OX_LvYXr%}Rd9qav|@U!!lIJL&j>x*U! z*e@a5U0xp7f-MzoPCpKVH(5dCAzZJTxNl&@Br)_6Hio5{+ZumoYQB^sLzFLWi2RHs zv<_#5(Rx#(xKr3$VTevP?p7}SZ-;@T^pQR3FxJ`LVZGyj1zYI<3btdb>EF1EHG9=B z`-HvOINQGyBNL2iWsq^7JWSN}1|h!5X<_FO_Mg2DHamKC{$L8@Mhr;*wCU%fc1Ht; z&bx$DmndwJd1mI74&WsUw)&4dtPX0lOGs~sU2V6)fh$iPTmL~=f|4~x;ZOU`1B`;s zK7iteX3k~_0yDj7XrGnhuy>0Qk%)3y3k)j{n^JKFh$w;g#VNdG0>?+}`2dF|Wt|Y7 zObnDc7aEGk>S%GX6vL+go>xT|$({td(hsrG0#ST(6F?Z=UUU1UhS^TQ&pMI=-FzVaIcS;HxZvmeF(x z$9V3jG^~8?MFsNx?#`ph7rf#L4=e(T+BA$Y0ZeUwyrMT8z^z}+Mrg>*T-+!BQX-~= zL6zdKl4&sgW}W^x*YDGU)A0#`MYP_Tj69)cUUG`oEMg^NBP}QH4js9B#d`hZA_Aoh zszLC#zB$W3wf$ZA&*qA2l{VRy455_=ATCUYgY0WKf8NJ9a$g&@CZb!T)1P_2ohPUI zx`^gcov-8fZpjF6kCo9<3=p81k{)gk=+nrn44@(hi0Kfk@O?l^ghI<>!|9R-z`;*L zuh`KHwK@9EssE3Qp~bHevp%G_Z7?%D-oG&lQmQIYHaBejeZmXO_l>Z7*fyD95a-WP zyZM+yq?zzqa1qFx zAEf*3FPG3=_9kKqA)@Svt2S!;2cMga@ymONJFoHa!W;Bx=i%tTol}?z#u76&*_Kss z^TN8$j6Oqk-@)8dG=iei1;s`U5D!DZVDJ?es12px~CX@I% zG?d4!Pj^CA{QqN2IC^NQxFBh7ZN);5hPX0d4maRk;HWn_`bOai!Cl6|+_wEIu#wVY z6s9CAE#Gf}7-()EFS%%ck=H1leI3RprJn9lb`uw=c$l#MyaBx}++c- zH~#AHdSlPg9luAlOc_ZH_7^-x{hn+d0M0A3C=DslrR(jpL-qDD&77UR^ysxvf7~|k zR)|@gPLe$9h27UuSYiU`wx#vpz@b&PpR+pczVl^nk~PMgg%kXU_OD(s!uuhb&c8Tk z!g|b()&K;TJaO)e3AJJ^B?OJU)D2=cpjLiQgGWC|prJ+NdnRCjHpZS9WF&zRJBQ*< zclBG>@g2H~*+$dRL4YLRFz4zZT%gF_><8W>oe2d0nWJ%BgV7NJIlBC9z?KJDkOB@!?9l9 z$E^#Q@;16h-oQ2k83F_aCC$&AaR#~ zzuCb92bLj6p~Ea2<%!tJ6m1!dOMputlH#tp?Ravyqeqk;y-X^ApnB?yRpy4Tfw_|j zWZ>VO@8f&^7#PsZ4DZ=;EJ$>Vu=;1yVoQZwq4#^*G&@*3^_3%y(J)47?K@qX7yNCU zB@2K6nyD`jeos!_xww^Mr_%fTimQK_0d74dWP1nHn*Byn$xuERxu>1pgiIn^-~sv1 z7`OwaSMz$i7n4X|tud#rmli{r%Lr|M|FJ?5_=r37;{UV-ge_&?+Jb`W3C%xl2+{uH zXVgWyE6Ww~O6j(d1`$a+3wpFUPdMWlHj{-9feLSs4j+HhJN?oh&1Eih+ndKX=0)Ji zvi|946VmTo&BctuXNBc#h?g6JmH06<9iz3{1HXj4#HiC!nCuR;Nl*lnLhqCJSs?vs z>vxlh|LMnW*Uck-tuFer&e}!ab^Rj^ny2p_OtU8`P=>?_=m5E4t+p(-Nt$%N=Hk$Q z0)Ju+`&;NV#JQJ~+@ije0YGT&*fVrOoUn{ zlcaW?JZa?et^fU>I$%;v`y2_eunH!X9P6c#>HFS-fQj-2UGkeDke07?8#Zh;*YpR9 z(|pf^;;FRd^mkZ$lQEUDznkzi1xrNsi$V_T7sV~l##GF${y{)vRwWC47;4s$7rY#> zLSxLWq@kMnL>Mt|ij$HRAk79g$K=dUsAAp0FZJz97$i6qk;BDJz z4}*fA3|@FF&uM(nL~rU2u0x&OHk(yM!oJfA(I1)@B~8R;_#T|0R$Q8ziP&$|p12>9jJ=0B;S?!X1Mj$x^N&A8WpxznaHL;L%lR!iKX1?Mt>d z2%Lg@uN%g_iH7$jI4A{k^1@OX=OC)byjPmuQ-@sN3a$}wOQrh19Rs0@HG~dFgkXJG z@%jdevQC$+}A`#C{5 zrP4D;^Yo>|OVgGMv4hjFGppC%R8u=1;QRc|Oobxx^j?`SK_1TW!J6x_>r@++SD~}a z5U<*yL$nuDhF~8gqOm!}>xToc2A>UO)zIwV^jqT-nGTpy7%@So+J^W3kV;`uWoojuXX&J)9@dI*&H`wpT7!U)ZF^s7tOZ(P zQ(r;@BIMvd^jBDy%;*+HVMa$Z62x+jgKR#FI9+&9uR$kss#Dxhk3|jLt5#l1i7Jh( zkL`_ZC(52#Iz7UuC{9`h4`?bgK15ff2yF7YnAal4FI=Z}OWZKN(f6fjmHi+XZyX^y z2aq;2Q@39maXw-xZ}K{Wqb0jF8MnImg;(F1?a7BzDeGJ=#828+JA25)n@~AoA|Y(= z+rNMRe=8*L`4hz-E^>c0e@+!>085<9#*!XC z+vi@ch5O}=e)awyz(r?5{9}wiY{-(+5wVb<|59y1XvT%qd}D`sjvZt!0nS9kQX1Wl ztr#=xSDM|^%I5K>3%?FQ?-%cr&Uo^ArH!2|?e97b{%WDp8fs@Y5mCjQrx+0jB2Kqb z0y)dZy~TmSC}LTaWZeF_0i_tHV&t}pFV{zQY}0+1{X=@m7|+S?{X<>pu>a7`@Vni% z%)L8wjGB3zx?@dr{HCdMq^ThQG2r~J5j|)ig|Oprqu=b4Sur>(o>5>rgVa}V?-1%X z%U_O|c;JfW1-nxCMK(04Jqf6TB1^|APE#9)Y$mgphwb#R4hG%JNKVeFu**~gMJY*C zv9vu?XilP=TeO24Oq|^nAAM^UfS-xGm^@^77eDjJ=#Jqa6buQIZ#F_^IVSU@UoS{< zA3)uS$dtnb580$A1Urog;ncNIB9pO}%ciHwfV%9$cQ0G(e zf4!GtNNCweH#Q&qdiFstmpo?o!a@DHCVpyL7+Bi-rV*@ z3q6W4Jm@C_s=eSIaX2X)>F{Uz;IYhlG3QbZwa!^|pXncAtt$Q*$jCm>?f&pn-MV+5 zMM71AdXvFP6wAF;RN6w!lW1i$EF#x7;_L~QZ~{%DqO(_w6#t|UK?7$`-}l{x!IPw0 z8Gp>Yzp!Nwnlh}WZQg5XI+j7lG_wV-)Y~+)-p_vUhQaAih&>emfdQ;Fm5G@^{GoMj z1vpnJ(P6%R+kB%L_Pw`N?<;A^k(6wb5oKVN1Uz0?M9?2!D(-Yt9oOfqYyLdtl|h-c z7~xJ<_d8Mf$_79x&zjOmxK5ef!m^cb>3o{WsK{Et!gkDys1>6QyPt#pSa$cg zO$s*4tJ`D$EU#v(9G3Z+nIScc#YX{U?VV{3r6VJU-jv*%TTuNH0p@xpePcd27!6nl z%q0{LIxA_J^*VWm=cr5Z6*Iff{hb2#oxE2OH4ev}P`g_;2OQL4N;O5kK4uCq=L6d9 zupCBQ`%*WU&_BTUYPy0>oG9{83{1&rL9d|`1gmX393UT;bwj5?UaL`MsTKCgWk#enFLcy6+jg4lUgRbUlAkc}4Jgt@_v9uJyU;+@xHGH%nwIJV`TzhUT`2uRUB7~zE@9R7}#Aq^^M3vGJ%AWivGr-U$$iV zHu}`*uSSLFh#JAHfJ#{J!E7McGVtbeU?0~|qmOq}*Qs-4C=s=H zhW6pu0PyBfJXCPE#dB-M0ryA|eGAke@^V!89O9FmC5i3Rw|Xx66f2L}upAGBa{_O8)k;d5ywVm$C7aFai$;F+K<4<~DZWZc|f0j@pa8bm08E7T(9G>IkCUjswknLaH z#-8}!Hi+y&KUU9+ZOQlFjo5=?|4+IlGlc&S`$WW&KHMbJ0B<|2-*v`jrh9_yt#4Mo z%Ay~*W!&~JcN)vGIDuEZ2k4-Se>7;%f+&7v>o#rJEC4In|HiO_Q@O+-Db}`}p2*OV zrOaSvYzsBGAKBoPj$eoIo4Fk*S;>HaOJ_pIdM#u|IJ`n1*yJWH&a*IHN$M~TSjw)x zfF|5`*dQ>0R&xeDL2hNNk=yPO)i2mg-q+towZT- zv_xmWR#2T5sUGvc<1u^MlHcl`(Rn@*opz;QXlA||4@sQ?@lU75)3ET{IXVYahD25U#!)>}(eLjS_TOm^s8o%)nFNr%4Sjur9Ffs4bbV#2VXT$~r| zcFZ8=Gr7Ju6wUsZb$fH6)88ztRL7_&IRLo!g{01A%y}nw=x;eaCa!L3wU`qAYXC(% z(!EyN{;_m0LJZBJ;a+B|gFISSEk>u6(J*9{?3a!=Cq*BG(GvW`aBXs+3_cJL2^2Hd zeli}EB8`l=Y5xz0h0=@NJ=#ruizim94~i*>#NQKg;ZudBWl+MlYcv{+vpl+#CnO~d zDI{b3P1~_&zHMLF{GWzmwkK3FQ*q?)u?I)#ms?SI+#$|h-X z3X$~qwq#mj0ezS0KepZ`ufyqCnvPGC7mn1`vOAQ4f}_qFdgz7jp5?#c$Iahaqt(0g zXG%$fmi-n7OPb1UNAsbY5sCnSAd3t^tFizDoJ2!TUg2l6Fue{BAgpA73=&;^uffJa zFN|c&7xt^tlXGwV-ctf zDaPxL?B0V2)AoDKPQ>G#Ta5u#NPH(fV{BmO1yDr9>|3{P4QqQUaF%ZOahnc7zsUqz z9^dIr(B()WDY$bxtE|G-;I{hIq-bBdjAw}&wc4A(;+)oP<=yNLR%5uN!r6lbgTmSE z?>W`05#okGg!nTtuEaUmtHrwp#v?2GyH(yoVa05!M>6tTd5a;HKDwT6{92ibJ}Iy0 zBL5zzt|h~7G*_FceDLK7gXo7${BO~>DSwO&T7DB^ zlW`Dm)wHFqio;T)DqxEyfc*ub|6TUc$5;GbQ$8tr6uIl!$erplmvy-L+MwXwdjh1) z9{%EU=*HsEj4--)QvR56>(i+M8`+o6|H_c@)b=%LJsZ~# zFPZaYvSlkj<%ktA3xo7b&N8t|Je=?cRA#JjIWAL!+L%Uu{EvKX0Ltf6|enDd%|(vaV?05Zf{1cMDq?XjvOidVZX+vOh^Y)Goqg!6}$6s zmJ~zSksWG&m=CcUpsr+8!-$#Vc&AQ&)${uAeHzW4w&`7M#OCa!a}Dw{>} zuLgh4IPq8g)(!Xf-q-Zv`6Y|nG;;Yfs%D*uiSj?`s>>Do?CZI2-;AU=^N$Yhxa7p? ziH%ORFf8s7?SHD@8W*nZnF86?}-y9H2dxui@PbdJov?y!w;4XoPYT&r~1?NsBxWe zNI#`at}aQ)_5Wesq3K?B?0vBFnyasG$Yy8NRZk}%A{r3*uCj^!a1YJbp*uEhaZO;T zsusl1Q!;%Sn0{cZKlT0?(oEK%T4vq2LO|bq--_ZV0}J2U*j-Pbk^1LDX3_|A@11iG zwRIC(j;}atoKfE3kwVfCznIm)%ya+-??#Jv!8oCqtMIM0+n=p6;Gm|#N*MC&U(qlf zRsAi~72RohO=OXsm&{=hsD>W?l6x>Gdo%9WV17i@2e(Bh*hyJ4PC>!NG;uwocR{NE z5yc<9bdpQk?w@jkAvo1JJ!+ql(?Yl(KfN`&I68Z5VC;_pEomu$)(FD-joUSFySDK| zt_MJ-nZG|R|J%xwCL$vJ_&#ZDMK?Mk##^TNJl1mh(JgKcVy_B-R&yWoWU7j1iRh_6 zTf-L^8qC17{?#$okSzA^_V>R^y%c+MM~>Qo2g}HiH`mdft z>t+r2T+w0Ko4XC0JF>@}=Q+Dmr%pE3^%DJ-Vu(wwh#V51Li>~|mPC1SbGO-^rgRU4 zx2Y*PeuW39OM&WA^~5dL*eg$|Yl}gB`%VqLRT+dg#+PW9&)W&5KD+?DF<}UwyX^Yn zO3#NEF3&tUvvIA)@Gi%A)qynX4T}$}+PW{ikA{XxhVh2PznAL7hm}2FwAKRgGp_k^ z?Z%nO%qugXdu=ZN$+GgwqXTQ!LbTLQ-s5&C$6vElduQbA#s)5--eBRBYi?0vA)o4Bo1>U_9zXWYzTF$M^Dm%4BLbmhkI z+!rSzo7VO{Y5cCo=xjThDU(}j_z+y4*qrWjE4AAia+`foLrv%7xSN}nRwf9mmFefFItdROM_Ffc`r#?9(Y4(7XFeDc)T zdX3>>AB}pccDcTD!UP=jsSG(8od5pbWhAMxMe!bP`}-YI(jxpl1^A`KY~5_K{&8O$ zS4%8FAy_){&mSC`0<|9iNcxCVcOJ@gb@KEJa~8Py!WtGrGM!LdnC7)oUdX!|6IF(9 zouiaJlQMsEmW`A2TlaD1AD0`~zgxpbM>Dit5=7MMtoDv%I&4ZWbhv#DzaDMupWx{=pe!XR~g@uJ&L^@Rm#o7G1tp0?4 zW{G8_1}lo~rj2&HymxA@k*rTfKmUDJ zpBK18^Vo~|nXpELK0opFb{nnpsA#^c;jHPoqw?KHt1GW{Sqv-Iznwzjwr*gdx%T$< z##1|%T0yp?Gd?>##AH#*xubbkl8#}X*Z6X~K~$dCR>cR6)w#_`4l2v;@5@O_Y`Aq~ z<829kDzjcu2bGawde~Hj#SblQ5awSHa`SUacFnizeVm8kGYh@e^i9>$%xkhYt+pyp zbK%WRm*~9<9u#up-M;VJYpcK8f4UM^Mtk?eh_NSI)NFqHu*=#s<_!6h%L}i>b=PO2 z#h**--yU)5%?C_7y|f(uSXTN|Zdpr>GQF&y;RhC^j<$NPL(D7BTmHOI?j9jC`QINx zp)6V(W0-QTcx>)#Hc`$`xe|M%YvP&Bua&e8OzCZtly*PXLi6ay%d7h(W6F zMm5>Kczb!ejdjB#ajRS5BQ%Dza^9MFXc6X*7-XA)bI#vaN>o)9~G=pW$;!rX3o4^b>kbpPqbL{+2I4~3 zB}vEj5%6eH%AHoe90uXOC{12^-+O$cEM;c3(|(-?xe49s^**+5aJ-9k=F4ciaQH)i z8Lck&@wI*75Z%*ib6na20sd@QwIgT@J||^KAfsPxJnnO`d9s{-g(UU*?(+99{YQHG z;<@=2yt~_C)vtFN_c*cf^4?PSWl3U?M@}EOW;`i?q-V+GW-33sl;s3@BY>&F!i5Ws z_j=AB?58)tr^Dhcch)*O3{3FLx4!-=H+jz3tM2O5eJfY2xNjIC?Tyl~9m9M@O1E)059QjCNP8tTJ6`9v+$Lw{cpg4@rgKgJ%x)!5-qp>`Z*=xH<SJZz6xRF>C|DjD$ZkOX_wKoy~4U> zm{sT1-QcP5VFt`Eavi*2u1nH>%|(e>5@d)Fa{~hdZQ}Y`pd-~Zurlv`bUt;sAg7dL z>BE1PX7(F z5thE@UUt(cMUMopZz@mw#=6t)DTCxsJC?s>W%Q(TMP5hQ>cR-amOvdgV)7lNZ-60Uj2j z;-!Q5s&d>`T96+-#%%B#8e6qqhB&@0-8-q*t9AF9^Iw`75850brotz`WE?q{UhJ|w z+ur>1`kV`D(SdxQ+P5Kx4*iuLFlRDO=sDB{)z1raZ=PAXpsCZn8nxV;rqs`6|1Lk0 z&sWv1SSVRsA@ABnM}yY#r=jZznOC~oz5hGr>ytrIMb|RrlxV8_i^VQB>WTYbE3_LA zYRTN*`6=hxzb4fyDJjt>;XX9#V{fhgM#uWKKWIU6u4E)o-v`$x0hDi;zWTaxk>Wx( zU%dhEuXi84oNo+cu?ja+B>>y@Uh~lbq+I#NyhV-z|MypRjvZq?@HjR8b@>I1!wH0% z`_U!P;QW|T;S0Rn11(PkEI*8#;;cM!{EiKvrJ(s&&SO_RhneN?hYfc$Jf(UqP<{3D zuzp!0-e3{?dAiARZKv*|f)V_nt^MD>U*UKwcVb|ZHvK+=8a0;Z!F&2qIr=kS$7mw+ zf?{MClh5(m?ngn`uQnuW)O4ISxP2-VJH5_*hs?h$TUqpRqZuG_83xXs3g*(ME=yyU zPIb{*$5d|Ac|mUyT<)DR_$)|xN3$;N{)0~(t9h4#1QGtzWzADmM-@6`4tE_hceZ-T z9>p`yCirzSM=l4cm}J>x*c$S8-ffQ0sMWfL?j&`qclY-LEgklDOfDsE5HT*0${(Kj zxp6i86_ci(0#&LbBSV6T)E_}6u#_?H?YvBza1F+q9cNX2d!opYi>mR&ienZYBr z;LJFuQc8eb?)VHn~TlY1)U^bcJ*8kSDbl0*|garx4r=A1LbUT4<35+Q`m+u zZ8O9_6+DIn%O{#cwfN34GAx?k?tQF$2^q(bIrKN2wkK4UN`~N3^7!8$zcu}O*7}2s zx0OtVmCvsATE4)ISr(Xk?PUd=C3+_pyN{Ok%$fSy+6Ooo_V!UwLw|1`maw_g^ILLe z*Dbv;sh|Ae^dDw@E+YsCS}mth!V(Hwu4G@WowwcX7ALJ;2h~a!&;dvo&LkG&lDaGbz01ldy|73NKqP!V2;Xdw$<)_J>fvU&jb8v;`5d=R8|>7Pr>NE};bFjwwh71j zIC1gp9wsjzrGa-TfXw`D<{Xs$}P>#Xb=C&=P8fdH>Ec* zl~D$sAlR{xaI;fNZZ=eNzW2R_JjaTIOXGeOOXoE=C?~C^oVqawY?BHl=N!-?mCZHm z4S0Lvp$U>1PS*iiH4pvr?Bpy!1imrKRDWfq+~pL6?~!q97P$3L(ACzSOYl~Kgw-gy z-)0~)iG3z7O@aegq7s&;8(UL8OxJt7iDXdeo=DEvqs-(vpNz#@i>~HY8tFlL_HiZ_ zaLxOweCAsp>0BUp4TPtlVEi$`J`!gn>hZ1~_Y}dm6 zagncTg}&LD>9S%}Au9)h7x8f&bD*%V2Hn?en-yW7YiMNR%WE!BOrRn!n&dt;nN9P2He0&8yd+YH9nQ3 zGh3JOi$_3BojGhk15rhdu38;K*#uVz;lqECgtTZ>d**%w~&?-srNgR7CVeHFaG_)8^x+!{+47nhBsx{tv4 z6B`HoJD}TN2DiT4Z8mVxwTqJ4coj3pmTm93q2eoM`&7>SRmWNf#)BJTx-lYLVXrO@ zeT>1c$Vu_IiwhB4N>Ir70O}ZU*bRFo+ZkbhN;FmWyhpx{2w)>1YdLbJa zm}A*@qazM5U;59|YsZvE6HL{*&SN&%Mp!2AGLKPho92T+4FGj=0DByncA-X99=5{K zPQOCKjR|0#8H(q*N|w|%E-fNPobcfnA~jh0tK_bYt@b08&Ty%}a|hYPID@;7m>(fO zf96;R&s#YoJbkdar3ljbw#)3Z+e!(sDb0pOzcF=cvfOJ1HZCa+rFtggo$2vSM_W4u zMp_lOtOh%I{8fvay->|C&S^Yf`6u?0GM;b&)xY|O5jZ%58KB*@Jc0_->$y6>rp6L1 zv4)~L=KJfd2w&c{mydyi8rG*8iz*DcX}}gn!o30d*Br&x&!DwjW%SehD|cI(u6cR& zvE#nNbPJ`C?Tpw*JC z&E=7=+qi9cL;k*_*^U?QPebjJOy3x-ew-EblkYqaQ_3uhED5os-2>yM4{oo2toKwb zCdz-*-B^;t83-aE*ChL|_o7^Z6D&OE)ICKxS-nrb z-pa08__(~QN~_&3y-ywuS)S6MrTW<;?D^f7;bmW-vnJTuVG8ai_x<+{>W&0pX?yn)%_;c&FHF zVG3nt+qk0g9}M{RrTJq3PY|^KcHR6d^AlEe0cM7xi ztaYw?(DjCUo)!v=$5WL9LhpRI0A@lT36tV+mVE+E4VDx`2>B;tUsx^dJr9&;wAjv9 z@yw=EBibJX>r17H!2GSjN%zS<@(@@e)p7IT{_)YV%(ixE@v51=)}w~HO_q1hcQ^NX z#`AOa^6I`ht5jVo4{E71HPywA!F0aIb(N|8_S~0pXs+(%uRZFP371#1wWJbI+oJ9^c?iH+*l z+k0;*wSnd#yQ^PTDz+}sT)XR(Syv!jiQ(z>#EZA|%dn^>zYUj7q$ zo)vHevoLSrLJ!TxS{@ujX14^x$eu~5%-ZHp53(1|u41RTSSACA)pfY&yMpp}HpzYD zk&nqA?stE4UF^R7xX;l#J~(eo_<|^=!^rL6IF^r|o!H2nQ!KQ+e&@~PF?s*B;hmHz zyrQT=$dbKC#lyRt74Z{V?459VCa_YLucISfjvt?GwM{vcP&i=Bm<6z4M=}j&>FTBv zLr=K4tY4U8`UoZPBA+OL$L%oav3SG%$$)|AdcyLdSDu&409Pyu0PDdow)tlS{Uet! z5>zcX$tek&Kqyg>5=UI-Sa?INMQjjv2ze;(lA=I8LzO36c)&- z2kn;TnPpxRMnn$pfk7dd=nhw%{jgun`2yvgFbKkV)P6CfUjSrnQ(l~%?Kl8lnm3)u z3KY2dIe>*(MY(ktWL@h&SJa`rKbFC8J9Js2-)$tND7=k4;Uk&TzN{mFrjmvt=ogMP zIrn1Q5PiQx+=+f#{;xu^wyoi!dC(7+m3z5~&g=E@Lk8;mB`6))UNJ`4Fa)Weo=^OG zKP6UIYo9h>*xNcQxgDV&yIM>c!$iB6jt8(#q|DXuwPALe6_o19mij!}(=FN@`1JJL zZfxBb#ENVfnNa0*&rxGxMM^_u-J{Qj7ixx&x^De*U-2J)A2w5ap`3*R_=1To)0^vi zBx8=yqo7Qa&yEdQ^NW%eM$B}Q%31Q1*T(WEXnO@T*rdab zhPx-evWe>!1-CT&tMjiXW|>s{s+UBF9VALF2vk7Ts3Tr^KhiBYYd-KE6>+m45*Ab; zc(fXiHc!kogHwhI4yLx8IktS%JEtzB1L--c9lJVn9xgtU(7s;>2RBxeR3%HbtIpIlpUr$ zgm1^(wiw_gh%hY#wQGBfw z%5%M~b^ce~W~@M-7Q+)sb5e24cQ0mnpXEr+bP0*f?F?UPkn%X90|a92&+J+i3?o_x znze4vvmok6I96?}@9sBgC3T_X1V^vj8;T~Z2L&FNaXxRXQ>(+P&bli`Eknl5l4mP`krtZbDSR&M)k1%}?F<`>EFD4%aySGr3MF#O`v~!Bq&s z{_4~C7hv-HBc03wSy#3(Bfx#hSoYjUuw&Wv*lJyBw-m(Zvyz+8no`c$Zp$q=ibRuv zkHo-Fr<3F`o{2t|U_<7+!5-^k*_lVe1cvqa_rLlI1Byk(m$x_>~jCCQVUlC4ZjaHdbBn6f~nKY>WUt zzS!(kIzS#_h-z7b6l81kp`<-fEnRAR=}(H zg=169f58emt_^691)zK+aSI*`;bK30)i|}XWSa~-S_PPp+yyX{v33IJ1Re|L$awYv zq!}rBggU_kmQN3cW^o9m`I~J^pMQ|3h$eLvqtFnOC<96aSZoQsLfdfj;%&O3@0#F) zFIjZPQ`^JceR$s6mP<4$KVB%vj*W$z{b2O+s_?V`iy*|eN_r}06}-I!&o+k@suyd| z^eHW$F_+t8An!lOK+EL>S2YpK9PDygc9*{zRXf+CRpF_eBTwLLK(=Ah^~M+EjVR>x z|3zT0t2d&{dH2zpN9LW$-6GegarZ3~@C)6{e)-haq#Yqreix5_%gdz+37>D|w7}Z& zX`gNuta&hG29+-~|4e~bd>swAvT>gmQoH%_S~xWG`r8h1FMf;ukVc2!cGlqmE{t)#l>@UGUa3dmJZz=AYOxhe6O71gtEzjV5^x8dgPoPM@T@yqE!xuCZ1N3~isg$i zU%>@r;J$tP{^;@~ccIS5RGY4~uRXIv*++taHV^z#Ec_nvAIEl9mU-=Uegj;Elq5BIzLTwTnSIf{T-`l6FDgV1C_z+Ts0_y} z^(ms|&HLyV3sJeC zTxQMhmb$w4Fdi8^l0QRAgm$X*yoZMWRXsUdZB(y$`EUHL2VQiKJYdGlB$K4!b=E+8 zU5J8kH$$lZj%Js<>~zok_`csDigpx9e%-gc+1clzYQ3kNnFonETQ)?03X7K@WwIer zwWg9jpcmJQ96OA#+Elv3)x!02R0xIldB>Vk@2DK-af(VI<#n%JVZqJ`btK?~lsjME zGUV<&KBXBh(DYEsih7K@(=G38Ouh6>jr-XH3lP34>Z1UnG1AQ zm>4Ab^xYKR$&uKz%Rj$b^gV0C5iuO>Z@vO3QF;dtzr1sr2yGW=xP~WPbYyGV8YmW{ z%xk=0`Al+d`>T2(yEDaFm4-$QTFy}P_VkEWX{oX@WaHYGCr%DD)CQ`%vD`F}Afv<) zs_81{U*VDV&Wm1ke|nR(WP#h7nDTk;n6?5sPeX&Qdb)z5)%HOC$(PzBb{!g1Hdhrdz~?F_6vS)RR;0axml46D={qUu zT20aSl}+DK1wp}S3~=FD8iTl=(~({Mz|V%kTXzMp8$x-#OJ( zzsHdU`u?14{L%Iw&pAnE0bRw1UUR3dRon!~QUup$*c}Q&^ZGuxlw21#9$C6=KEJ-F zJgyzFKBT5X(oKdKWX0ryF42E=$@Y`UkFi?#h727Be@i$2Y*Qq13H0~3(Oaf|uQ{M1G+w^mQa2m^C+ zjrJV7G&`(WIxww~kj36#%7J)q5<00`p1=)y5#%+~C41ye$}bNr6&M)8w)|tB4W!wa zcV@FCJh_9FO{o>ATLnfVHENi9?AlMSV;1pY=aM6@G6UAv&u<`_TFq8nSNmP3>b~`c zCNb2O{Kp2$hQCAxq#1o4&WtpT?^(y6p7ML}MhH=Ag&R%B`knc%JNlImc~fy?_Z%if z>4BqwZ`r-G$TOTX*pQn(7mVUmieJIF2&&UOWHIU_R=mQfC*GQ|xCvG^M_T8)8%ia* zwc>`~13Rj1>H91ItjvME%K5z0I@5!B0qi(&m=@B5kJlf$!r4CSVxf+0DS5~e;a!4Y zC1lc&1z}TQ88jB9YWOoLY68$iKWjwzwM%n-5Ac-Uy>#~;Tzc6>k&`nMAZ_HOZnyDq z-;)D)5)Q6*`AgF6?*4uj&42M+6S~IPyjOEpPcB`wotfpjF0;1YOWo5J$8`v7K;?4x zrmk3Ap0+bGws~3fRQV@(yw0lOG%7VQn&5H?^Us65y_A)Uy~o-HL5cLqe={w#Q=j$+ zmEhPAVKP7^pBk&OykDgHFqW5*C16}7N#?T36IsRqp-+z%l%nPWWKYN+DDA*Q#g z_Du+05T6GV`Z-U3<@n84%WZRu>wulY>4U$@Lj~t2SKG`!5xDL+C=ac(!u21W4>R(Y z*e2kWRopgIOUx?Fmh3;qIcM#)tzO(XMZ87Z~O={tRSh!aS%L55~%{snR-LXL|{Vt;0=e%|dO6cc6h%#h;nC1zKyBEOvWsbbD z9m@AEHpg+aS;F@d&(Wc4ZWfIKcIhw+?m21TgqK$c)rpu7VUM1RQpgX@_w!sc7rns{ zN=M77`(Ru|JlIO+LxooqB(>PU>s$MrKu+cxVyDvCkeg!w&>rm3i>Iidu;9qAH27xr z%FJtb-(J4;Be#pMzyHj+#_i7V(*sFY=I@Q=_i<4N z@{>~=4mf`NxNV|pMn#trzxw`m)hpbe=OHEv3!&I;^}#ix!WpV#UQMa!uRfGt>%8K* z192urz64I4N#d9i(Pcu3&4~sKN*Ust#oQ zlLPVN1|`^{7xRJ(ytaOpqNOS$;odx~4qz~CEv?5-@`cM#IO_ZRh1RH;HQ0tRYhmPi z-@)-;gWI?dJjz8C!ZG>L4+v^SEgoad!z+bR;>}5;2-Z4~Ic2m?N7}m{cu?8$Z|y#v z`&~R#^gpE0KTPebopPtAZAb`($3jmC)nSp((uqcGSW!k|BiyJ0E^hr}&c|Eh)Kbq7qBhA1oU& z#`yvzq$OWq!S+DM&f%M0g{?g3wdGwCT`%<)u})^&dmMa_##}x!^T>nd&E%_3c;0kH zx$wjID~GmmU@?LVJj}-)GvUi2zR2{GDGxzAy7z@_pWxSIzDMF$&$YJ}XlZHH`2Dzo zjeJUee!flVh;;RCamNayJk)H?qe2B>g$m}H7C^~}Liio3&2f3{e4Y3*f|?Sw(o0+{ z*_~fqvxX?21T=lsDRLMe!*M0+PoD}zf`z}|?Yi|ME(>4V#en8C!E)k%BEPJE|LB~k z%v97ub7>xr+!Zy3*e->|D|n@vvBi^?JijAOUYeoH5mhU{Kf9<;g}`A4`HSi712oeh z|G(th{}1j~q{a|g#IO6>;it8biT8SfYk;8-X%pEVN1DT=_ z2Kwfosy7{-l-r1sr8V~IkJS!|RgoA;w1ms^vditHQJSDEQZ(_$Po&RQY88g#+8_MU zZnv{9;fb^5(26o(YVW^^*19fLIeaiR4bVe6WqNgEvynzHSc^tyt=la>K_;6RZa?8O zy~lO~5zxhSB`hB>0v6n1K=p)4Q|Z zXL8F9?t0>0EKeF&901dD(X=koIofR-R7bj21?mj{N1wUlbzS*w8W*#(vtsU;-h=nc~NVOZ4=2T@WGN}I*;FWLxM5Y zVF7Ztw4HZU z5v)CXwuG9h#uh%4iyj<=gBD96$TJ3z=v;Ae2SSf7 zB!JH)P=|f0tb00Y8BfFN%7%6GP90u{R&M9jHLhgPtbUew&}+u%^h~uiF|p5==6P4e zGhd9Ui=RUYR9U+Et=B^&63?_!t3OvU$5L26jJK=mVE&-q?gO<^O>s+fZNXHZ#US>R z!mv#5ZZ}0BqGL-S4oDi>j$m4ld}sA~&$E`fL8tjxYIcrt&&1iT9>xd952Sb>IuFp7uXmo>#A*JTY%uBbVg1@tj%5wS!F6<_o&QXeaA5I5n{6D-l#f*S-u!invc$UPmD z-V$o~_IX7iW2pr72echq+F^53s$|UezPO{tIJC))$6=w5*Ip)vBY*i!wldeM z8T;Jeu1aAVH5}}X&_~2MacDvM?SA-H-GQ7e+3y0y3W@mELUf$Y% zla8#e9aL1TE)gzXw6yTD-1qd5Elr$14TmR=AuA1zPl3#%_`0aPWvYMD#r?G$m3mI&Z@<|e-rDk5~8?xOYjk6#^XLu*M?2)tqlt;zZPTp3&a?pGzD@TS!-UT#O} zb6Og(xy0UBZl)hi(Ie)2;|z_rW?KoI2uy89Pg86S=nSG*@!5Sxw=_1v^RsxOAZ7e8 zCdu|@Fb@Lt9D<*(|dPxB?SyRyN)bYxc{4bQd_I*K5{CH z&*lthdmyu8i-z^-AI6U3ddOK?Xcxvh#Q0QfI>=>6=L8*5!`~8HvSy&AXsa|r?Su_I zFSM4Z=e!+L!$$^*Cw$m`biLP_+KAgW)Bv8$IwQ*_ynDA%t4~1%AOB+4dg=31T&@#}jzt#FeR9jmZv2ITcjnN@ zDV%~d+`E2g+)`zVPf7BkrjOEN)5cWf9IHNmKdJ*?r?3A4mrnN_Jg4H$!+p)rF};+> z)*YoSh}Z3^7@K~>Vzb4lqC~&)w_W4*TN%2$xSXdfvl8ezV);Kp&^S7qmA-z|wbkXp zGl@(Bs%(;x)z^)G()FWqi_m3WKPI|PN$%Ut*m31}r$A@-RePt0du@g1IN{+@a3s&p z3vkVT&`>^dzm3jQ+y}Y~NvHTDeHcB_5-*5+(y#&sm`kZ57%B$$wxn_fSzH7nR=O(d zKCt`62`HSN-!u4Eu)CdK*WQ3vk`|{4h$Ll4vZ?o6-d}I3**(Kn{1SEg%$?@nB-N?c z63<{F@|$EX?7mnPOvKSQmEMDK#nBf4rUiIx=b|RpcG7>5x@W?$$oH)^kJ^p3Rv{-Q zsp8{tTlx5!wd18Sh<<^~VS^<}V{tM3MD?3>%T9YKcM#QH3EgiML-g7tIUhWF3_0OJU+kRb(C;BE>VMGw@aFv_0ivdg8gnS{-&0ZDaw02i5K0cI> zBWX7`ua(^4BFv)Lx>oOL0Qu>XXZKZo*qY|_(Wu1G`PNz4>`?DxkN)-^G4bJ&;U8lO z1Gvb}adxLuB8~rGR|bKKjrkcbw5s4V$3u>FqZVDaJj_@~pSGN&9`{u;)+2pdF{#}A zle+n&Tva8jS#nEp_dl`z^@l}(v84&5EdDxiqOF$y>#sLO2Zqn5M&_~7MF74A?!Ts9 z*pp{D>Xu1{Yhjdl>|mN}SMdVG2`mzqMEEz+W_OQ(EeNPb`R_w%_WsC;#Yb^0N6n@v zen|Y+MIN5*bIHdLK4QcH(MW~kSk)%|TGE*X90>IL58I#*H!S^6+>%!zHx74qD-H`!*bEohs@y>fhNA zZDH1#p`QtopW&66pVa4^4aYa^z=9rJ2UBcYsnB{CWXC1tG12kPlBX?eDT*SwN4ufC zGS>Aw+lkz!<}B@q8uowb zhaK>?wC<_zC~f?UPE)9f1wIr)G`lHk_IzXKCFq9_v4PXO=XJC-p6ty&Nhn4tE?ei5iuRXPS3v(`&yWb_4-osZX5nS*0>Cx%7Qhh-o@L20}*B zzgH$>;JCr8mf$u?_xALX&_c11*m9XiiOvVz($~tiMfAPv;U>&Fl^fN#d1s+}y^n3E zyW2Z9^!VRr1^i($8R%G3<&fSl|II(u5Yg$NZ=99Ydx}QRP7l0hL~vF+DWloyyv7(D z3fLL#O;a6b$+YiCnhvlL(X}?0BD7qHO=3bigXmN#GGcyyz8M9{Z9s>4?aq#+>vWJy zIN@ap%E=-u_dk7bg9$o$EBZ&C*FHxf{~Nk<~zh8v)3c zJTF7e#@5N|3Tlp*bLwvLgnr^vk^)u>5zTdDfX4@rY#HmH^rZwN-}ikSnoA=4V!qxa z!_@QiOopbN$Alqg*$t;0K?<|nv3Ph5GALAnXh+LUQO6wDTZV&5(GuZ@0HIK!SiCU!p>c5SE``H9j4jc;qHi8|8!zK z#3vDjvcrtW-f&QfEryZGe%`I_I2r%@o9+2h>#q9)mrOpz8MI$yRr76aqFBxy!f!n* zpPpA?n$@WG*+b6Lj6D{46y>}YkaPC8alzXuMx|K`o6q3d`MA=E{lJZlsk);rU|@uF zmY@p7tUb2vij5|tLoV5gp`VngQTBB56Kv6ruX3xIlyMlwUMc^+)g_Q`5CxLUyUgP( zmz{{e(l$%Ibp$=JSXkpi|FldpuYXlUdmv`=qsY2%b35|zFQQJN*87F}=C0EzIt@u6 zIYD!{i!efr-JHy%vpHb&bfzWCT|nKRGq`#SnGIGEkeu@V!(ImZa)K93&Mn^;JUj=XhMWvyaBcV!A^Yy6s+^jR>ny;h>Yq5YxpWRl&+I)uD5@{7&{V8@+R6;bfOuKLP zO31L9H#paWXd5H~IekiN8?32+N_-_%g`;;qy6~98M^HbD+2`Tm zAzF-n&tQ*BZ>1}??qh110+8bXmYp8WApBkqH*M?Qh=M3QDj|1NbGqZ`b<0Y z6rb-!G_HZzM!BJ0d5=rg`KC^)zPp#oG*ygRE0euU1dsJ>b(BOyA=7@?;)`9@+?Xg1 zLR@X(6$9%j+>W%w_imYeD@M#MT3>_p=>D|x@@LOo0QK5}QL*4w_<+SI>UzUFig%Y# z>(9NP3eo4%V*IJ~BPZ(Q>-j!OsP`D;VZ{s|mv_NOGJ=g}V=qw8ou3;m;e*D#OKg02 zWOMq4c5X-^I)Z2kj{{A&`E6hbp{cayf?45|VPpL$+k|1aNzdRR%MKQb)tVEZ>zGLe zwOGAgxF5juD*mKXy;#%gB!dXk#LF1y1|$Iaa~= z*q1!$0AP_`@cmL%s)ay62I5aAf`6TSAoorytxB3a!q(dpq@v#b)8$uh#KuhP>(V+R z|I|I#{!P1?IRreg>B^155c6?whsAmrWu=ADQP<3J=D*i2QG|56{^7=$V(B|TOBtl9 znr?d2eS=mCWqkm#_Tgd2kveC^E{C`%#bl2P86^g15PV^=BmU`GmA{nNH&s)9wR7&s z8SYtn(s=!T=|d(7<)<5&8e-Y4TH?3yvbKR@>-|>d-6_ML?g?zms3cnx>E&Eb%wCae z=S4K}gTK-ie8X+Ox83ha4s2Nj#l*$ae$3qs08f};iiyRW+sBoS@G1K!dUdz5Crx(-8_MPW*hoC-yi({#Lrknq9i<8S)8`>#q@q3iVA zFT2tB5IUkZ7l&NEsv?dg0c_$UNlP=pt}zG&jB>M`WJ)Y!YF*tFrCuVw9_WG!`DPW( zL?fj+I^|BpzQ89P;dNwCLub{2GxJVTHgT(rXKR$7(2UF3|6mz82{p#1tKb?~90T~M zH2Tzp29n0l_iD3@4QIAVP)S#cW%3Jxz~?=0A99HD_xJbGBTAAHH2*>FQ4^b#s0_)c zN=9m`(+gEwN+nsAU0Y)`TeJEql~r-*Pib@{$_Ik%g+c|V2S*LnLaY7p_?7XmNU*r~ zFN^DKEn0D|S;s{@Ukxz=UJc-oiMKwD=7C^RSDnd|q8?fH1XH_kN^=dsQGv`d5P{=e zF!3aBWa-ms{1d{4;@}xRbjqs5v|okx7?}`c2W{yqLBnUcQLxIyY5H30IgfY`Jw?AI z35xp7y&&czd~@U0*7J(&8yUGKK1|pQsIi`FsxH`=7-4_Qh^1{r`p#pIba=&yt$cU=QmGnI868F z*MwfFa(_D?wuE#Byx17Zx~z*_;qKp0_p4$S?!0Oa`ZMrBcAw`<>MU6U@eB(}k|vD! zp*MA4T++lnf=*@u7V8WY1r)t=97>935!QGSc*tvQwWw7u81@ykl!p*9H9Sf)lkAK6Q96!Z$Y4?cZPdW1TQB zS2^-zPcLb=LC*|e#-jAsIA&dR=iZ>3XBdzI(m>_`lHhR=rox{aGC5JJ#Bd}nQMi=F zX%HKfTf$0vQ}JRF`uk|v)HBj|Pkn?C4=HznOk{;KQpi2W*C-yOub_N3BMZ(m^wn-x zTPmUs%xEY^sE3A+2O9>Rkd_~AwAapYbpi}!d)fWE5vUh z=jkjeBBWQ;yOogMzOMz-j?sMLpICH1PR};K)NA{<1sZhBGFW404||n~QlhO>8=PD! zXI>g_L{pS7OxUML7ODXz)77WXq?6R78U-cy=xd;hQ;NxgtjKVm3a3>G)mN<-k#lMJ zHB#xEE)*}=>c^1uz5ot5g@GOaDxmu#74vAj4i8l_jPMRRuFkMsxxkH@{(@MU^MLpK zr$hVSp{=UP$i^zxo}t-49(G)g3hRw=)t<(*UI^D%}? z=bL0R8FHiP9Oz>4(%&=UH2$P? zb+Bt_hr^C?fEOn^8ei)^JLFog{TSytOwl=PP;&?6QdS|R`_aXclUwE zr6dHSUWA-#+%_?qaWR(kH!P|oXm!gd$TSE{NZ2O3OAv0%vqqE_BCJc})fnduclFG2 zEnSgW(wQNqf+>NQG*_IuR8op4eH=U-=X60XpsK99B24^(P*^FMk_vTWS}B+VFMx;( zHPuI3G@1#F#W3M>gd~r$t)6nMEm4# zXT>urqhER)EmKc!xe}qXNX9M@M17BUjyHCcDvT6aq3Jm<2cH(DL+7y@13pLQb&m{z z$DU=mI<<8SCr_ZZNZ3TuRoG5s>I5SK`X-k`CkDY~1OTCH%z5eAyQ)^<2nZC{OS{Rw zyL%-Au$A!5*kshjc7>r`-QxPxnP=oXoDSEiSciro#t7BTpN)9{G~gL6;X=X%kI6rq zENo_D<%@BIm$;;$>&hXea}UDuNv>S z{YOcC?_XVd-rK(JmwJz49V!4(-p)(vCwvrIYMy1Gv*rGK9X)kuAR$!qj>cfO^@*~Ax^|J+~fag z?_9iU%=a+fah!2-W=hknT*l>~L@o=tgy@WMViV1gE#;DHiKM1D6Qk zI7t3B-!r~V>pX3U)^s57j_H+in`;KrpTd!WvSq%3$XV(8q9?z(c0>(orN3J|M>vGo zh!ktHD7ToWBn)wg_XPA&CR4OyegkqzuJb{QbvD}X`susKxipmf%(&x$C6rJUg;Md% zP}IXI7|%+hvQ)~vO0yV=i4=o$ zS)S9>I3<4{?o-j3RR?;q!u(OiI=-oS=I)Kgx{&+fl)9vLwEbK&XbM+rz>>}XJ@3YS zp%0NH?U<@xnmFi%5CXCKn+il)OutfgE9KbUOrnS2FZwdKprJh!2MVi#_f@MLN;9uS zWGYMKP*Du$vYs@;388yVbc$GgVW!AJ>ShIo5+H|EIlz3F&mvYPcQq6C_-ly~`p94O zM+Q+1Vmr0X_3o$UxgAGGd@DR3)OA4KZ%4srs5UM*n~YzASPjXfz+;5kF!l1X3=}yQ zH{139450F#fHaYt%@Evg9sAO}_zsc8pgRJUr#l=BolSNR^bo2z5uU)kZt+1)5*fW0 zAt(}Ju9P)86jgpgUC`P5?LTO#sdz=78{gXXV%oXKgO8sz-RpQ60+J;ii4;oOFQN77 zzW~9^j9OfAT_aKk7%vTKDmmQyS-!CGD{=N|E~}7a0K_2{Vt`*L zeRv|EQ);x{0HC7fqEn56%xsDY0j80-CJ>>}3GkvVq}*}Tu$>di@)k{1r~{dl437nM zgDTu%w<2-3s%~kQSD}c|l;ijD!L&iT4RXe-lIgihOS<6h$*F>l!_tT>Hqi^9+yT3( zh3fQ9-+iq9viNIJk|PHxqIUbFD{o^s;>Q)CQu-BjjB@b9yKCYMxue z_vDEg0r-Cg+;xE;>*jY(QcmjhN(=c*&aT|rzGBSSv0C5*x!l5o%PN@rf}CvYUL89; z{fgg)~8;aF#0{LJLa}&4!vCJp{NbJ zlWG61+k3mGbI^dn6c@gykPZv_DxPZMAx9x5_!&?ukqV1v4=GQiOpEND$AhhyGK!)D zH+fHyD9yW|@!8(Du)2(fLiQ^4( z(7KagGH5xaHp0>Z3VnQl)@Sgfl~)ngcCw>g#I;%x%V7Vq#w{y2Il4<-WkADK22n0t zp{PwqsIpjWFl`pjGb6>ks_iotf(#tQuo-NV9cXL=x+&aND<6I2M70kP5Xm^a7ZagNS{z1j2B z?W2yMbmN5WOZf;>agj`ngKFJJV&KXc#K?Dhk5rM>~Aml=O`_t~BT614QSgOS5|=)3uNinW)R z975NAnez8p?Wg^w;BHo>lXm@CB1b(H5tZN$>=9a6>Q)LAMErm_lDLDO;HG$K(e1<8 zx2e$zS?@&$gMBW@A-tGITPA{E?p`e(@9@JXTd^LHHGImgdBzoqi6r6Fbe#w-6Phv3uY;3Rri zxzm`IIjq(UZdKgxULF*?w3mYn-vAjgr)3RQ+&OVfe2YXlE%Go&(d-Bfa@&{p#&(Wq z#f5%oObgH@@t|pnxYyc&PEeg%TvS`Ba;Fc9nR4*jA{d%-SFWA`#!D>YrpV|v?Zr2+ z)zbP`<>vm-dMl;_<9N7m;pYpFr^7$!+UjGM&zfrnPpW;+qJ3_R=`nI?eDMSIoQ9vi z8~Xm7!rIrkd+R2!p+xpv-{;zXcSu)+_BPB+**8ZZ$2 z=VD^i`I0m-b(ojOT%4n9KC6@(%S45*32X0Kbpp=+&-8^9n_3Q9+c&?Gcv}0{qzwer z4g-sN2T>spd3yLuaR-n>1B{Od1BmPKh3?fQkevgN8Kd@ny@x==s6IFhWp} zJxih^+> z8cd80n#n|48QfmVt~wk}sspSP2ou0Si_ZND$@yu{=z;gVmAeyrs?)EoYx31^me8>- zWAPwhYgzLCM6Z52cak;tzVzloyJ@VcyLV-#5O!kHJ$dsDY#vqWB66Gjp{-N2IJ{57m)iV_v zcl@-&dgjr+mCgD{(~pGkD=~-RbWCZ^%wNLWS|}e@4INZZD*1xLrF1I_YZVk8QE{fP(J}PW^?*+d~i!OVb z8yViw>m1fcZFVTdjBXqKn^-eK(OHGA>;5SZ4oW0n-%iZ3^efBG-W5`y@(NoHLq@4< zW&A@RAR%Wjc$aPfrO0t&P)dO$Vti1Op<*8+T_PXb$yjqgZsUqR>QYQt(;Tve`kL}D zFTFvUJ?d3Kwoz`C{4lbUUBe0S2^5Buz*o6sImKtNdDEMo>My$ana8TF1n*n6@Nqgo zC>PCuf^8m6QUG}2@6li5D_S9LD(pquP<5e{UcVCX9+HXuhh&vVBw3IJG&;s=yM+Y6 z#S)le56C(3p8qt>N_3lgJFDM6E*0!0xoEBWj=zHvf+wFu{|;x|4`jv%@W^~)Y=5sN zmhL>r#xmgMXX~3xxd9LMprd+;U$1Stf64m=Y*lLHlS4`hDuUI|o^s`Ni5g~k-z)53 z7t1fzSQrWFPs!h-TKxGH>c7S|ZTz9eOVGFqtQs#t<0YUtXp9Aom!NSKG;Rc{qu8c( Z9>WNm?*Fo@Vj2Jd literal 0 HcmV?d00001 diff --git a/scripts/slurm/sweep.sh b/scripts/slurm/sweep.sh index a33d1c1..f319723 100644 --- a/scripts/slurm/sweep.sh +++ b/scripts/slurm/sweep.sh @@ -1,12 +1,12 @@ #!/usr/bin/env bash -#SBATCH --partition=gpu2 +#SBATCH --partition=b200 #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 # Number of agents to run in parallel on this node #SBATCH --gpus-per-node=1 # Assign one GPU to each agent #SBATCH --cpus-per-task=12 -#SBATCH --time 2-0 +#SBATCH --time 1-0 #SBATCH --mem-per-cpu=32G -#SBATCH --array 0-10 +#SBATCH --array 0-24 # Check if a Sweep ID is provided as an argument export JAMUN_ROOT_PATH=/data2/sules/jamun-conditional-runs @@ -32,4 +32,4 @@ echo "Starting ${SLURM_NTASKS} agents for sweep: ${SWEEP_ID}" # Launch multiple wandb agents in parallel using srun. # Each agent will poll the sweep server, get a configuration, and run one training job. # PyTorch Lightning will automatically use the single GPU assigned by Slurm to each task. -srun wandb agent "${SWEEP_ID}" \ No newline at end of file +wandb agent "${SWEEP_ID}" \ No newline at end of file diff --git a/scripts/slurm/train_enhanced_long_comparison.sh b/scripts/slurm/train_enhanced_long_comparison.sh index 1baa66e..9874b5a 100644 --- a/scripts/slurm/train_enhanced_long_comparison.sh +++ b/scripts/slurm/train_enhanced_long_comparison.sh @@ -1,9 +1,8 @@ #!/usr/bin/env bash -#SBATCH --partition=gpu3 +#SBATCH --partition=b200 #SBATCH --job-name=enhanced_long_comparison -#SBATCH --qos=preempt -#SBATCH --nodes=1 + #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --gpus-per-node=1 #SBATCH --cpus-per-task=8 @@ -34,7 +33,7 @@ case ${SLURM_ARRAY_TASK_ID} in echo "Job 0: Standard JAMUN on enhanced_long, noise 0.04" CONFIG="train_enhanced_standard_jamun" DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" - WANDB_GROUP="model_comparison_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long_take2" NOISE_LEVEL="0.04" RUN_NAME="enhanced_long_standard_noise0.04" ;; @@ -42,7 +41,7 @@ case ${SLURM_ARRAY_TASK_ID} in echo "Job 1: Spatiotemporal JAMUN on enhanced_long, noise 0.04" CONFIG="train_enhanced_spatiotemporal_conditioner" DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" - WANDB_GROUP="model_comparison_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long_take2" NOISE_LEVEL="0.04" RUN_NAME="enhanced_long_spatiotemporal_noise0.04" ;; @@ -50,7 +49,7 @@ case ${SLURM_ARRAY_TASK_ID} in echo "Job 2: Standard JAMUN on enhanced_long, noise 0.06" CONFIG="train_enhanced_standard_jamun" DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" - WANDB_GROUP="model_comparison_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long_take2" NOISE_LEVEL="0.06" RUN_NAME="enhanced_long_standard_noise0.06" ;; @@ -58,7 +57,7 @@ case ${SLURM_ARRAY_TASK_ID} in echo "Job 3: Spatiotemporal JAMUN on enhanced_long, noise 0.06" CONFIG="train_enhanced_spatiotemporal_conditioner" DATA_PATH="/data2/sules/ALA_ALA_enhanced_long" - WANDB_GROUP="model_comparison_enhanced_long" + WANDB_GROUP="model_comparison_enhanced_long_take2" NOISE_LEVEL="0.06" RUN_NAME="enhanced_long_spatiotemporal_noise0.06" ;; @@ -66,7 +65,7 @@ case ${SLURM_ARRAY_TASK_ID} in echo "Job 4: Standard JAMUN on enhanced_long_state_split, noise 0.04" CONFIG="train_enhanced_standard_jamun" DATA_PATH="/data2/sules/ALA_ALA_enhanced_long_state_split" - WANDB_GROUP="withheld_state" + WANDB_GROUP="withheld_state_take2" NOISE_LEVEL="0.04" RUN_NAME="enhanced_long_state_split_standard_noise0.04" ;; @@ -74,7 +73,7 @@ case ${SLURM_ARRAY_TASK_ID} in echo "Job 5: Spatiotemporal JAMUN on enhanced_long_state_split, noise 0.04" CONFIG="train_enhanced_spatiotemporal_conditioner" DATA_PATH="/data2/sules/ALA_ALA_enhanced_long_state_split" - WANDB_GROUP="withheld_state" + WANDB_GROUP="withheld_state_take2" NOISE_LEVEL="0.04" RUN_NAME="enhanced_long_state_split_spatiotemporal_noise0.04" ;; @@ -103,6 +102,7 @@ CMD="$CMD ++data.datamodule.datasets.val.total_lag_time=5" CMD="$CMD ++data.datamodule.datasets.train.max_datasets=200" CMD="$CMD ++data.datamodule.datasets.val.max_datasets=50" CMD="$CMD ++trainer.max_epochs=100" +CMD="$CMD ++trainer.val_check_interval=0.2" # Add noise level CMD="$CMD ++model.sigma_distribution.sigma=${NOISE_LEVEL}" diff --git a/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml b/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml index e4f18cd..69102df 100644 --- a/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml +++ b/src/jamun/hydra_config/model/denoiser_conditional_spatiotemporal.yaml @@ -9,7 +9,7 @@ sigma_distribution: _target_: jamun.model.sigma_distribution.ConstantSigma sigma: 0.1 -max_radius: 1000.0 +max_radius: 1.0 average_squared_distance: 10.0 add_fixed_noise: false add_fixed_ones: false diff --git a/validation_errors_plot.csv b/validation_errors_plot.csv index db13c9b..63f43d9 100644 --- a/validation_errors_plot.csv +++ b/validation_errors_plot.csv @@ -1,10 +1,10 @@ run_name,run_path,validation_rmsd_squared,model_target,data_target,subsample,sigma,total_lag_time,data_datamodule_batch_size,index -noise_check_spatiotemporal_repeated_pos_m2,sule-shashank/jamun/jjx6l8fg,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,2,32,0 -noise_check_spatiotemporal_repeated_pos_m4,sule-shashank/jamun/psxh4cl6,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,4,32,1 -noise_check_spatiotemporal_repeated_pos_m5,sule-shashank/jamun/ss5qjtb3,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,5,32,2 -noise_check_spatiotemporal_repeated_pos_m6,sule-shashank/jamun/y8qzwcfy,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,6,32,3 -noise_check_spatiotemporal_repeated_pos_m3,sule-shashank/jamun/bp5p8gbq,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,3,32,4 -noise_check_spatiotemporal_repeated_pos_m7,sule-shashank/jamun/do28iggc,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,7,32,5 -noise_check_spatiotemporal_repeated_pos_m8,sule-shashank/jamun/feo5op22,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,8,32,6 -noise_check_spatiotemporal_repeated_pos_m9,sule-shashank/jamun/4f0aw2ad,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,9,32,7 -noise_check_spatiotemporal_repeated_pos_m10,sule-shashank/jamun/lik3brd0,inf,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,10,32,8 +noise_check_spatiotemporal_repeated_pos_m2,sule-shashank/jamun/jjx6l8fg,1.287444397287128e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,2,32,0 +noise_check_spatiotemporal_repeated_pos_m4,sule-shashank/jamun/psxh4cl6,1.4849490481435916e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,4,32,1 +noise_check_spatiotemporal_repeated_pos_m5,sule-shashank/jamun/ss5qjtb3,1.3529375963257174e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,5,32,2 +noise_check_spatiotemporal_repeated_pos_m6,sule-shashank/jamun/y8qzwcfy,1.4670496489178373e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,6,32,3 +noise_check_spatiotemporal_repeated_pos_m3,sule-shashank/jamun/bp5p8gbq,1.6283258377672115e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,3,32,4 +noise_check_spatiotemporal_repeated_pos_m7,sule-shashank/jamun/do28iggc,1.7884779541672415e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,7,32,5 +noise_check_spatiotemporal_repeated_pos_m8,sule-shashank/jamun/feo5op22,1.7367485100323023e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,8,32,6 +noise_check_spatiotemporal_repeated_pos_m9,sule-shashank/jamun/4f0aw2ad,1.762919576003462e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,9,32,7 +noise_check_spatiotemporal_repeated_pos_m10,sule-shashank/jamun/lik3brd0,1.891950703725839e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,10,32,8 diff --git a/validation_errors_plot.png b/validation_errors_plot.png index 9aba92c1d4c33d9e43f5a17813e5e9ef1337ba39..afc3849950fbe0713311720d5a62a84b13d211dc 100644 GIT binary patch literal 169982 zcmeFZcRZGT|39ulC89)pT=#ul_wV!jeeVB$e||k4*VT22^E{8^c)wq-=X$*ZFPxXBrevZdBO{|$I47%) zKibL2ww>Hcf$y~M8jZ#ul8$mZjv6*s9bJs=&B#=Y9c{1MI9|8Ba=_Wl-oeu5h7hj+ zKd%ta0SiY*TL(!#KI?yeg4f30obT$9ydb>FHrsQ$4rFA!hlzhS+)R>jCfh(prXYJt z^JdIgx6{qt+J8hR-TR->%F{95p{L2H+)uf?{q!XZPE|{#OKG`&5z>BJIKvn=?sjq6 z`kS{>!S>1rgZh;%Hq$1~q9r*`gCDOr7`PjZbx34~#tNDH&%`F|I=z*4=YRd8Kghn+ zO?K$Ne#Mto_euWOzdgid&!{gi`(MB0cKDD2{_9urCu7_F|MipYq?e$|IQ(DVI{whz zN~iz&#pKSoF;lev_jgu%{Qm!5^#AM5|KBJ5|6OeVzb)_P_gN+%pR1>^ zWtx0^FsLVOQZunUbII2B&Dct&PS&A{r*g6rfzO{G`20j*(|A{be3YQsWg8nlk^{T) zlSjQ@*lI5xyObZcI&>-jdQ`8-~v-WS0) ze&Em{H`d+N0*sma1OM70$p-}MCMf<>B_2ClT3T+UqWTbf+=Z5&-Y+@%XyU6^@7vlK zoSmJIUVOc~w5;sM_&BwaQr+r}jHb;L6h9ZI`U@R@?Yhxj?AO>HDioT}K4b~TYYZuR%9@_Q{kSSJqy)Kp3>NJf`}GYB=v&1+mmTr&&NSsS zFfhEy&Gn6q-T(dj_r71h0)yze>X~8?c$^{@%Ux_tU5f)*b({o@$C$QlAnz)&bVrstD%KT z9UU*u$4U|}+nH zq=(EB%_B}~m5Er%eCGpPy_1rXetdZR(@9W!e5XO~oqBTy<~HM|oCH-;WPj zhxNK~`}i&lnsydBaByiR-+Cy$y!Gym{p!1QwveTyq!hZ%aYgVN1l``W<=y-Do8r9J zi`2R($T&GU1A>FAi2KrCP0f7bCha|jWABFJymuHG8Hr0syl|Q5`S!+mle&fm_3Ohz zapB?gZgZo4N0cJ^z85(ZdjFB;)i1myv%a{=(9n>qC*#KaV5HHN+Az+uD-ViSC%pfB zWz;b@Hoi?s&-Gj@T`xqtCgjLWnzi3WJv}C`wK?rDUIQjNj`JT{Q?=D zx6tb?|M9{8A7f+pok(qi_V>0I5L>W5NAB3sfp1wR*;&=t=hG#+vZkiwPHEata<6x& zH)}H;)_EKda6grQuJx&yO}*%W^Q=!rtpyb&{$A~5|8&uFH0RHsA8AdEWK-E{9QKMj zwX48})v(O%h(fAXIun!Vb&J{JfaJ@H=g(JT3998>RasqI=q*zJi04mBM;DqP$Rgu) zAD@*mN$T`gkL6#lMtaJQKa=r();}#MC^$L$OXF~^vdp$^+rAY$aW5}S(4cO7jy|%2ox@t5tY;dMph_7CMm_n8a?xgbzm=c~#=x z1g|-nHh=vpVAFF#A-b$gN-IM@@Y^??6E`Qx+A|D7OW9ST!${*?pX1}>8`E@!JXXdF zf_xP66qJ=^&z%d{sdrXZ)=&4;J`&a)y%W|a9_g6}b)2=gC;okFOg*mt2flFWpMAg_ zh<&Wstd*v78+-C}lKS3s78N?S!fX50erTm?9Si=2MHLVfL~N-8zK=XvSXm!Flv!td z`SK;Bz!hJ$t}s+1T-N2;ktT}BERQAo;BnCpcy?Kiztn@{!Z|f4(k~U9&AZmxH#9_X z>C&YSRaFNC1$WV~oOnA}7iqV)?8L^wu`lTu*@a?9OY!5!>u~*FccO@nj9B1jE!Woh zjLJWJ2&9|BYB3q_%vQZhqH$eQArdv@}@{1zm<=856EWOY`oF4E>@* z+x23k+(l7GLsI9;GqFj33=iK`isZFiEhV;jOY+XwnkgGOI5>24b%_sT(^FcBPing~ zZT7vyg=X*Gy{_Nw4;|tfq3qOoh|N#t?d^S^Nu;)Km4;RNLro}0@VCD^KG*1fK%JXq z>udVjaPGUX49m3>l9EjF4_T^8U1x+db_ibm%qVQxyg8g(yI$=$YM=k7PoL5jPz1zo zzH>TzmI6zJQ^CR6xf)wxme*=!ew=9&35C(8*K36uZ``)`Lku3#+RB1SSHAVPeCvHb z`}^l^d@S)=osy^8C#p;mMPDg&U*IKn`1U=#m+@mv+EgVD!#kKnuK8kBCZIS-Tx&0R zt(8tGCMM?V=eL3EcXx?-Zg!5tk9Xv{dKF%4_GLW_tzn!R!W3tx0Ej5pmy6Gp@H|MC8=$<)uPo6EDD>uU=}Ey-$U@~zd?ccC`$ZGT=LSXd}7D=%+e zR*;t`qSw)(W|UW7RYlIr%UfhPu`+_+)+w}o!Q+Xy{8ng}@Utd@g@uJt=ql9?X7O8h zX;^=zW|lV>jvm^%v+NJ$Phw>zA9ZkY@{5a$WBbzRUV~?u{dQr)CJM&yqJOW?TmJxm z(_5&ol^B(nI?EVB#weSXTv_E+ZmCf8BiwY9=<_4I&&v4RH^{!ot!P42m5E6pK2O<0hsLJ)iK* zGk#Bj-l)`E>_qadzKY8C6tL`kHz_AYkj*_aY6eB1OPS)dTP}}VQP#2@I@B!xceR#sn|+kFxIWs-9o+6yY(eQEHv8i5ekL3BdAel7 zzyD8D9FI#~bi=#eURIOp`!#As(OT*IM6WX{tEe~)1>e8_p}WLos;_dB?Af!2czJh- zii+AgJNw_gyM;;Eg3R1}m^m-=Iqm7b>A||2lYJWu%RK_IcTjl~6vMdye@a74 zvS!MzYc0u-ImWkc-~PVHVR&+J$^@0XvZ-n3@87>IzP&kET3R}tNL`eq93_~HHhiPE z{Fnj_6_xD5?`{W8XJQkNjs{Im+LE0-cW&$a@9ukm$(+0sT?^;@RlZt2E3oN3d-JBava<4p(wlZ+$1X9HmzM`sb1vVU z`bqw~r|ceq@8#tW3kdAov112u6TwCQv@aTAP)}}_Li-K!x?P4 zJ`!oWl@)3AM*Ni@KYr+bbkfmbL}UChIvVYyUva!iynw(2+8O$ls9KN1!ZrhxYo_bb zn>XKn{J4dhN%&Li<^EFX9k=fY_0*l<+e}V= z5bcI5oH8RLW27U81K`_fh)=OyA&O{w(`V6O&`j@Ohg_Y@Ly=0pn94BPn)<3R?wq3H zI{?$2jEn(dHa%~fn*~WaDLPqKGIFJGrF(ew-vTkv;_XrJ3JVM0e)&QT5O8{FdN5ee z$jwcXz*QwK6F>190fB+1@ExMLU8xFsfn}JimO!Dbe1D~M2ul+HbPLwRJA4R0vQ6QN zHhz>BqizcdG#m30pKC z(EcqmFf7&ULhoCCHZtDh{K3C!dc6LDzj!GAMTM<;2+U|RSq*7^yo(|->dBL%F)=Z0 zgDIL^QzK0Y*9&@!ol+MU`ya|BDTbR`t9j!u)YA1trdB2@?qP@cbTZP?9^e)`aNt1u zbjrPZ_vlbm%Z_TiRL<}H+SaCZAJyiB*XlQmIioVSi^@+$RfKJN^mX*~M!&aXyI)zH z{Nc97BAlnz8;P&#Y@a)KOuLp>R=96;mGByt9wi!IOl)jkev)k9t}B3vKhfkUTEE82 zZC7wTrkDH3g9Y!uy1MEc7^skMJvlYC6Dx-S0E{B9~L_p1G%XkNq-i<0mq=G@m|t4<5I8Y-rnBl%b=oMHKE;kp|poSdM4$IT9R_I6H?;> z?%&@X%AwkJ^WmYHi`k|M0xrMx(>+`8?p2>2ANGrlHOL>-%rrD0nymTLaP*cv*hIU$ zyu84|6-nznm-5pVAy3Ha=sc;M>#p!NA~0R9Rr`6A<izO^l zZkn^Cc2%8Am$;YL3X6+_u#E#CXpFAbSC{sP4TyiaH{}Gzv^d{oBZsn-`&&#%NT?}M zfiBB=T(7J;AgZMflv1zF9KGYp{8&f4@rPyHc>MDIOy&i{7 za>e!Q*9oX+ulWZW*KUW)MW)G4?T+he^rlJL+sDk`e7A`>qWbgAG5=j+txTiUYvez~KWkp7cF#1bV1NQ7Hgc2adb@(eMV)H5)BJ_J zp;T5n?U>v`2C0tvT+!?s3r9?hXl1 z`vYjBBTG)SVXUN+;L)Q(8eZRLJhcrMMEoBFPXCNK{jXrE`9o+oTSa89?A;9#g{ zKD9Fo3)-4n$n?wH=mi7>C|Wm>;W9_QzM{F>@XXqFud5|+ksb2s6PR1#_rF$r_apxgOd$1f1cy{Ni4e$7J$7c*!5Qvl9(^8+n|byN);}ckPtw> znez^Fa&Fsy-1!5FZGv{W&lZ|^jaTPUv>MdbmJV`pDMkzd^o3r1W@;pi4rFg1uBP6*Uw+SIF-7<^C}1FW#hb-IX{2?%w~zD{H?;f;-{c5Eu)m1 zFhNwp+pLIikdG+00hf`ScedDGt}i0+uGb^gudOe!{z_xVgSs^z#JPibYdAdA@a5L_+cQ)R+gEy``=`Xs^aYpQ8dEK5Re!2Yt2? zy;Iqq_P zq7>rm6A>#0g4Lqh`;0cscDwW%e=oEHhE|W^5Eo~}{S_Hu%Hk&^wSlsiE0xjug|-_2 zdTQNlEG+{e<}xLz;E56t{c3%TWupKczmbkSu?nA+Uajo?=p=^(h zk3+M%O$d;l%e%ER4XY}Bwj2o^^rc<=T{3YY;ViY|@9(;q^Fv8F<_+->iOwk~JWfp2 z99dbq)FOLVcyhlN@KoO4BtW)$*FmOik zNA1;F_-(oX_-+Ed0}NHn&AoT9zkd@4$E&mj)i~+!&8knHK7If3BS+UNJ{!|l^ZJ-T z$hEaZw`_cUkvrGAiva~q7FUGr>R;x*@Dr>l8gmGO&EuzGJd|&RlV!J$RS<`!F>uvO*caXF20oqD#XlQ7(q-bn`rFABp z>tf%}pBuqaw;lwE!;@tpj0JJeWnswZ+0#{3RWH!J3HL%Vk~h|RX<|KMZf(DLc23UM z7587$Qy*FDHtLeSt*xab{Jfc=FKQs$Z0ziZuzS#$9>uakuxiBfBiO%6v~ZKo!&k4& z+jI^?0eQ$G72*M%!!@=y6P&TbDe|q zKhgW}u>6B_)Be7`m&d+keakVYfmG-Nh-w-fm!6(Zlts`e=4a2I5jskO{KGe{GgrTt zdtOqn$Fm^dvCk&e>1A=AqkqBf+3~JD4_RgE6|qH`C0%~s=eqd%ZEGt%p!-%z%6CIw zp3;XeNgHu+s$Hmd^ZQcLZU&V)OgSGaqnsZly<-F}as=rKd=l6}8Q=v9Uy>M7ev{-Mt_36Ja0VsSzR_q;=KowQRGxzJUQ>@Js`{ z?x?Y`vFr>IJhiRz^72sC4j(x}j-m`dRrcgbGQ7#*ngz&N0Q59y%aKyTXF>tJua!Pa zh0@fSZ~auHEh{%yR$rePwumjd`S)U{?ZXXm4a;2&XMt|y<>X!{gt9~8&KS!<2Ox|_ z$WD4~g}lPTyMczPAUE&ax9@45HC{t{b@EJ%w5Q&#>uH*fj>5%`zd}T=3X6%!LojVB zcCrdPsPAV)aItWlgcoXPkrfZrIxT^2(^#MILzvN{+ zk}9%S-1b2E@@Q(B$tw1C;mxT-qwSfQB5})er-6@h%G%oGq4pHJOk_+M9|3?I>CEFM z6hopwSY@LN{{TD)u{MHUh~?G~01-4V_2|!n$96@d1X(rdSO0Q8)mUmG=cfPNct*+y z*Je9eh&JA0j@KL+fE<8#*0q!Q(n}lbTT?z%~ zC%i3M$k|LYkO=~?YPLZ))a?O-KYGmpmMPm(T3TB6`uMM3DNaT}6hQ~i_gN$q!)7yvqAb?5%u0-R#7H z0)7I^#-2KLO0SNJn!5F7%tl(^2zvA>&DB_H5%|hJ59mAzr(R1QYs+y+Nl7Vm?Irp@ zFZAU5ozhAEpxl=CpuO~%#Yj3S!r7>um&YYmp$cbzaY3gG9wx`u+R4nEmSd>$@k7b@ zNOwsA916?)C0yRzdPe71PuZpZiJtG@^*yGl_WVSjKbx)^`*P-eiRWq5?%erp*id`m zn1_e|7#d=K_wL>9ycf8s@#wQxurm$;^r}4r=$&ekXRR3>|6Y_6zj~T+*kzS>Zbl0? zZuj7&3a_&2-ji4eko@?vzZ3wLszD0Sm4;<1x6;&@6(vg8)oIF**zEe=Za@uqY+Vi= zM!>3##~5_-xQ&eKw5)mA=&(ML6lvZl#+RVLa(0!1#O=w?&(C&_6Sq6vg4#;*1V*ib zOC}QW_}MdFujOAH!D}HQcJf}VcgZbF7jAfufy<)^gpb>B96o%e-}TL#f3%F7GAZ?c zi5b-63&E|wl}A0>5F2nBYHCp}?P`aRsA|PJSI1ZJX&P=(AU5{aL&e?pJt@K@}_5Y-UUMK&sc5Nqp%hX@rR4MAG{4`!{c0g~`Vqv2zsU+qI4do~EQvUa|#pr8OUYtxcJ?tXLY&fJ{~ z6TJnwv7^j}g&@2V4`b*4 z{plhT==HJ?EX?K_;$)1>%c3)02{p(J4GmpsOVjOO878*g`sxzXN20`DoGlCP)*cRX zGeZ4vtr><$8|oqLJf@RANa%4`oeIxjQJQ;tO%Et^lxw)L+Tm5QOUXV!!pt+&<}oVQ zhvZu`-I{4+q$cyUcfrH&Q$_jN-d-!iXBz*iHpwGb{)xv8p1XcM?fU5qT-?@e+tleRXxN-6&{!$jP=F{=DfDu;^=xcw2q_7>nd*(as# z02)qr4G9gEL;i)$6-G_C^_w$>pDZt^sFvX5G2-AS{}Lb-hVS zdtzNh_R~dWPW6AZq9&_729Ka|xhqpG!T$-r@f{s(+MDpx)JScPUY|Tr091VM-X-xI zK5;@HGPp{F_lJO;L7k661%j8~Z^HsDbQpdRUrK(kkXKYx3qcv>{0W4K(nksw;Q|G3 za0Yz*8rA;o_DO8n$nbWAhvYswz9f+T#xX|~VTRK)5_@i0^pqAMuF@Li4Y4S@bb`WG zIa*jjZACtq=?FK#sfgsv;DVhF?MHV)rvz>e8r9bRmJw?Inm>4sQGX#K>kdiK2&oNC z&54h!-Z9oj=GU&hK6>t$goN(m;^MhlegT0dCh>4iHa1xXM)j;Ka#g3^oGzv2MOG^; zkqY%EcK9>WNr`)j!(Ri>51-OpSnro(x|=rBf8&8Em?G zZZP;1!$D%LYE_SQ~`AO+3C$#7C!?|ILftOV-?#odlk4+GpyhZdHa5G z+mq8p8FvMt3n;eHl&z5QpYb8xf&AWYL-rHJNPSj{&!CtMewOKy_n$SFSl5}AXEt8* zBFpQBGt>WzuwwtVzJ3P;>F`7%hy-|cQU5F;7+#Xik}z*TYI45xkBlT6o#}1Lx4r}< zc`9Tcx(TDc;SB|7B)~F9EKvf!Wt(mP%Sr|&ftzW&4;tcSFp&#AXdK311tGcM zjfeR88~f%5@{lb$6M^1c@d9KxXxba(9>6{gxt$kT`bR#0`I2SVfA*%U>#y%xN$Qqq z!*6eIqG*kc{Q2`IKYSAAGrnAdu?=v2B1x0iNJ-GH=6a)czw2)k&mwie5 z@ivDMYX}=3hQ4^KqC!S`ebK~YvT_SlTY2EKsiomK6VMA(4FSdD_Ms&uB|OIf6Hx(} z3RKUXJBg-)^4*q}Yx(u%kBN#u2;W&|%fVd?qR6KoB#@Rqdn zbf<_^O{6~ff$WntxqPOFKR_eMd)9uVr!)|%6TyCt8kL^_Vg2jQMhTKbNS*K zBpd|+#7b^BA0`~4MyX>qgnCQt#m{k}0$;$hfCAUj7EW))^84O5J>7}gPP=z6r$Q}gNarGwV3?x<`&Ju~wj*TrVJzP?T| zH-Wdp{+9wicf7HJzWxH_J0ur8o9rYyI=}n;bvdVcMC{nvoBz0dFR+Pw$51!eYNyj! z?f$mQiVoYzXPUu}#O6sKmc_Ey2%Z{0dYN1oe&>v+*XJ=(`O5F(j-)mf^S$sso=7^q zW;=HH@FsA>mZ$~qKP#-T({p;dzJ0rlW=?F)kfBXvpohU|!Ai~0CbJxd&&Q5yv5Ik- znVI>)jDA;L&CZP{VJ=xQ18?ET1Xx7?!!bR^7JnBKuQJtMqgPni$aOqUbQ)>gfU9Uu zjlse&K};hg7hd%tR0zAhQC#>nC|TUaV!{y2i82-v;Tl8Da)L3=b3UqG5a|=hFq}%- zf-iOQuF>D|@yY3V`s4`_oq;6oUtb!9G(?tlmuSrs7Z}5Uky-Hmt^tb^rcg0Bp08vSpEaKn=70^JjSU3okg!5cp~l${PJ(=B0wa3cpgE zjJGtb;G8d!XoB0X7$s2ut@MZO`A-f>LX^Oj(Rk25SEJc)TXB7Tf-KY_SMRyei)?Ic z1J`1gJIg(f8&-H7Q^0CvCK_KM6`BYLa@98#jWopnk~LmEU>nEMT54fhgOs%c5=TU0 zD-tbDrf*zOhLuE%J~y6BM%u)JTJqv>fB)gGKcl@V2hJ%eoq>h}!8#DTr_jzHrHQomO{&F;1BAnm?=+)59Tg^i0dYWrXSHwXKG8TwYk zpw>`1RO#AgFO)#IMO?fI*E_NaGl|U-yG1Ye({s@;Q9>NO8yWDT55WWo2RbPKA(9lZ z5+}c^%De!QvvqQk!w!1`5}S}`xw<%ooI}XZaqf$+ebG9Vn$dc*u2fMbYNrxiU0*@Zc@ZWd>%$D|m-zd&ClL!!20O0hC%p++pZGn>8pdM4$GOG(fd?Jw4JS zPka09QyZ=sc`Y3w1pk{WNGw_^|3Tz(_a7<6u%oxDgZC*l!}rLLvgLq7R!&`^__+9+_+xmOV{AExG}^Ar7q2 z=8nrG-_F3O`3Dzrl@| zXj++%Ky1s+?@1b&ZRN@W*W(#C~ZO zehggSvD> z`$0v8=w-#D5`&OTU^-kz=X6sYU^v|ev4&B?o{bBtEeIuJo^|zemYa_hzK#kY`5SSuz_TUrswKp z8cNkDdMJ7#D4Q*O<^lN`>z-2G>2hp1i|*pQQ)arCE+qg&h6v9|pV{Wwpb9f7BI#m^ zy4=LJU!#Kqrr#g$`DlEKd17$>%vncHs&<8Cu5%_*g3rLE6TrsZ2Af*`B4C+Yn2{EI z4hc3y*noQrtF#BBni@)Y3^NJhenhF*z=;=po|&Z9O{{3JoVrEusYVv{Je8ZW&6IpL zdEg?CUCL7F?F~Z8c6ok0RhvcSV-!Mkrs0A|q!;@()wmmc>>x5D5MglpgY=d!$kY?A z#=!wv0gDvZw#C{+mc(1}PlAsqHLtF%iJYf&ZHI2tAJ2x@OjkqiK`GG5FH+= zQUuR!l#!+`OXLTrf`3h%Pp(V!x&c$hn~$B|u*hC6Kl?JS%k-4J;!QWVcr=@kzF@Qt z#AHrqaulV?n8{S2IDwHi^_CWoF1 zXUc-{UV!5XG4?q3I1nuyzeX!{LK{3EBYqpr@&M905MGJf2Ygm-a+H_%{Is-$#9ov@ z0;crz9^|u>Y+&kq1?!F#5ns6r7cOXZ0+t-~ZTwcj_;h(d?Tpqr*Y=WOmcH>?<7Fa3 zYM9!#;ZfRXcnXv~x9|;QJv}`k3s9mcf{#y+it||XOOa*w)yBB7W`)ITD5963*9au z+oGX3y+<9|fQa1nnWdGLy5?EeJ1cN`1OQ6Yj4D|0B$h?=f-?doT{#aP+yIR>Jh3(X zQhVR@C=wb8eZ^Q$$dpil7KtIGWMM0Dva++G6J1Ddx*Qe|w0_~_G(y7u@IiLh?iYsT zycm3NQ@yt7xgB}($jHd)o%-5<)Aw1VB-=NUZWXK)fE0+^4{W{l=V^-wa+zt?E!yHu zF5%lfp+OLxd89dMJ8&Bj_k&DF8Odw#4v5uuqDKmuSF2+tmGxv0 z^Rt`17LxO%)>Ly_SJa}&&F$L`+5*4>^dOXfh`fCQwS8~+nHm+*{<(!No-$DMqT-M%^oJv8Ks)#M=^LIL~>;X1v`qM@H-@=2(uFO{Cie z!<0uVosjIShV&N*$Zo#UxBUo~B;oSF-9C~UGSOSX3UBT`2w&4HRaSU>EzhmKzkB;= zY7V7-a_GxGg-*1GEcb;QK3y-QZ}5ILq4azxgqR>kTgpgZ!Vr*C`s(s0C37)Z)_K9U zM;rC~6J^Gxo}TyM9M{@53GQGy5kQ3eJbz=`O#0Ba?45D@@b2A4bk(r$LS=vmKzC*| z$7?6t%gM7co05{_{;fG3wfWE3#|FX3F5_AxVg!9AIasfFKu19wg%AdWN|2Xj|$7q`+T5Rw$}rvo=v(!}NqxmwriG@CO$*3zS}Yuzq|_ zwx==SsHrMCZ!T}Be#jc9{pE`m|224i%9@*l7lRiyIASj~*;d>Zd7#$v8nzk!hCR%&VyJz@;j&DtnUM}Wvy8S-mM z_}@SNFC;7l65~ViZeQ_;4WlD>O_N{dCKjI8Z%CLS3oy|O;U4$QU8CE%^Tk}u!t|gL z6?h#iFcC%1;$}lGgq?%bXaMpfL==!>O^Wgml%LUx z(9gmuzx=td@(fbbUr)Ts&CS)9sGl|dni|R3w?^d%F2$t*^A=^z5IZ82`Y@_lx?Q;Dyp~JXJ@Y^QGIbIwH?c~XnJRH+leT59SFBQ zCU@i1B|o`uZrlFR&>|;S)@S7&s8Agw)FF6WXVLYrobTeFx@p9oSbIHYR5A)O@iamz zX3(m0*e_@5=1{;-)Qtmx_cY}Zh`SbazC$B5#vPI#Yf`y z3QZb7isI;3J-IY@&Z~FkSir}iU2;qRa1x(?IEkq+;5|aJN7yW}9K+Lk9ueVWhUNRP}AeNW)9TX}1~D z=-_{w2!Mw2Q6n8EWWFQ+Mt3ynuUy<;A`#g@xP&JdFmuvCpOS69Ct+pxl)hQ+Kcb z&4C9C_MNPI{78v0N5@R0xMC34WV`)z$r!u}GY_|U)_iEB-x8)RDl~dNIZ}6Lm6fT~ z)zwYc-a!GQdhL!NPoc{M3-JMBH<;tL7Gl1nd~N8Ejz71jepw#*`N|Z^NTfAdf0^6E;iq`Hyhk>_%HS!O*Gky|a*T z;O0@Frx5R5TNC_>epu9G!dqoYr1g}IC!D}1f@VQHo}kl3>%P_?q#HMGWJ?u`j*2p# z?5iy5+;IkL-UNyO)RAz{h$2BHFG*+sJn0y*TWSnh4c)Wmt~O?bIZ+>A{A;Vqy}e$B zYf`7roFQx^1-#DQq54>2LINVmR)_+=n7@GfLfDmv=hW76-XGRIS#IP@E1_pBRPlhQ z&bpn*s+ykKCgDjh;%diS{;SCy3$zY2n>8_&+jLL(4qBfN9a z37wFk-@hNA=Jo^GEJ93aL~0gh;XarML}s_JQ|o;7KE!A~#LL~AL?Z2acQHMx_8HVN z;U1Y!VMs~<%o9P8Li@p>N~b;S(bc;#-zGLX!6q&)K2f2=qSoup>M?n{>7mdOn$3sT zL@XM~K-cTs;+{MS06-jc>nm9n|MIQ>q>7ViWJDwcJa?n_|Evav|EvZzV!lcAlmk#< zwnF779UIwY06fV$@5RZ32s5)OkYLY)CPG9 z_~IUm{J7Xx+7hToO(nlqkxM-AGz{aF;R%ARGkEaB$)yY3Sn-(q*`%O{DHAWm%=Oa? zAV@&TJlnQ{?S=%=xoj0;Q1{obsMPUK+fbx~L)S4J6uA)WDvv}N=5vlH*)%6ANawQu zA>`7C2vl^9p7_oR#QJplB@}ym%d;8^I)YW5e}CuP&b|VgOGF6Ew2l*xxNl8@L6}@< z3px3qTiHZOsrZiGf+<6_qVu&W3&v@YNc>?+Q0mE(C!ZcvmYz^eHfnzw9eoA*pha(a znc*O#NvW~5bugk&S0ED-^q5G3j5Ho37CJ$0$PaFOivrUGaD0OFbb4m-=aWO`Eeos* z20sKn7Hw-*z>j_+c%eIiK$Yn*DMr15rKV<)?t+(82vpYI$nux`F@M{Pw*Q%;-+K|^ znnncu6!dIu`LQ{&d+z%CUzzAB3$c!Yl8*%zJIE^OcrHSeXk1-OFu}@Z297}a(S3B~ z{6-VZ2c@>p!{^DSe@V>z!wN4W@s6zMK|~}+L=@5km)A1E3(#wiSVptwH2x6~5vc;v zGyPZzhb9<4+VHjR(a>T9oH$)=#$F{2Q_Q`UCEH3v({MLpH4f4zF-;J1^~Q|{P^ktq z?oHo)`t&IjoHeD(m!I>bA6?+9UK}4zeAwie;q2nlUYNf8$Nm3gf(RiLuw4_fG1X7aNbF>4ImUtS0Eh=`3$ z>0=T!4H(eTk`V*J)a@4tp&M;S{+5%lB`XJ0j2>cHSWsou0Vqva8JVR0y_@I_{&O5k z!q?o5u@x2GZ57(-dVWZ6)eOUnxdUyXK08jzo!!;O%q(d^@fw-wZ4+`JGHOB#P$=hKH6O`pqBKUyvgxi@jXSR=T?#5`vxgHtsb!Ko)if|{)4R@?VT*#=j=H|NA zo)HWcQL!0>k2?Bj#zW}A@87*U-EU5as)){HwAoNSc<>-w*i-ABU_LSqJi6Jk5y)0+ z2U~l_r7RpHGv%iKMfB8=1+H*Kll^jj0$wZCCwe=fpFmG%D!9n{RTw&CSGoMtJJK&} z&fj3&j!}15NTrp7oP!7#xG%m%Mux~Wk(??Tpdom~RlCa#O4cw=Tcr)74!E+?(V1M3 zKx9bxtiw^7lr1?&*x7%=t|~ILQBnil4`%KsHFam5H(f(GqWjJjLh68_5z!ZY;q_UWQn(( zqr%RITrl!H`<3oiI58u8YH2zI+17zIWS!=`m;}fGg^2StAQtAXc_y>i4!nI}_bbh(6R@R^T>4wH!sn(C0=I(Hq|b5b+<>iv`r{IRBdE#7*0BiWJTy zPX4#wR9)LY;?{`FbJMXae_E*?$iegf#GY;4Ti*R#niqr4NbH0qUd+(%=yT{_dCOR_ z2I$L}YyTDv$8KfbT0swO()ddJQpJ_~0x!4G&_t%~NTZo6=gDlyA*AS}y>8FS05SeD zTQNbHO@FW_DTPnWk(FTk)6z$dBwe^KvOT+0a!2ng)W%}eM(jQ8G=(12<1<@dxJWD; zb`5HJ#qsrQK^#0g7x~p;x3cQujNTF#J{YCnVd4US!t)?s@a-6$h>#X6t>4 z9pCe;_8>3f7!k)QGqJ>whg8P@x(9Re^Q-@*iz_RCK(V!3Uv=F~L7^y}3Tq3I`GN z1GKx41y_kE8nVs;8$P8=$d&a}6yOwrG=mbYzseshlEGw8IXQ$@2rWq=nsy_qYD24A z&eN9geX2?g)CBBK8myu`AN?6z2gBl-JY`rC6zMT)-z+89F*6YiVS;4PhAnKyU73#v z6UP_sSz+?%z0+NPK(uG;6MiwL|I&Y^u%@3n;mSY;^O5c8hZH;&b0zE^2<(c@=-;M| zjN<_BVM)Iv8Coi2E-y;sT!|8%alFJ?Ma7^~gH4wRY@-Tn;}~m7wItAnBuVywCxR(s zojQ@XZrwUzRK#1+uxG-Dpk^!?Z*Fu;`rHYaGrmu9Bg3$t%>5?jvgFLs%?C)gRYnoa zxbCNZMyvC$yW)CZv%2I z3GUa2kET`X={4GvaQFI^MiiF0=v#Xj82lmUsZA7%s9cq}Ir-esj5xXlXGAoX{8k18 z3WcCwGb7OikkXQ>tr~(D2{lnMGwm_a?+!e8Hl(e<#;CBQWVkR`u8Kxxl?>;?D0D)5 zAm%~!Q*#hUC#DDn=y&bf$jTb0pJD^_kbN+KsixjvW)Y{_V~6ndCUD1{@}^xOlV5i5S~}C$3&vCOJ2A zT~l*67(XC8ZM0V=Q0G^xBd-YG6IUN1jD11HDy?rNJvTwaDBVURatV_{K@Sne>h*_V zAU%_wOcV4;BsP;zq%4-n`4@ttEk1&IL?{DikXU_Y=Zx<$LeOk)?oUItq0v#U_Kg4iw$Bj%c4~Sda2-Q%(7acJR zD{hS`YCFylL5LhKA_D1?2mlYER?y%kc+7ke%btZYd5DLHhgC1EpyzIRCkCx=-@cta zGXw42AxRq2FJ@H~Ln086MaFXEVm<1z8IAb#^z>=OI*?+`m>*a`k3YKO>h!`wLzAyJ8R6y+tij-&qm<rl_ajz#>l9!J*AdpmWdB zw;3VQ0P|VdO5juz;&=+PQ+k;612VEc|1<^jAe3N_!SLRV|2j|L7!S`I3?w4F-(T{b#>c?u71AYc%!bKcqTO* zai5xx1BnR)C`KZPQ#IU(5<5HEs+4~PEO%4x`EMGmU*YH&BQPA)xd$%~Jij+r@f7>q z3=bJ%@Q}@^uzvY*NPto^RE@hBYv8Q?gDg4ny7Kvw?U>;-OTCcHje!pY9{?M?ykTP($B7FP-oKz`wmq(kJ7j2}Lze zm{M{8EA(@bMPE8QnTQ!^p0$#R^7klF)Vwaxu(DT_CK6Pomo~#73Q3g0NDMIo2B(^u zO9;dJGOLrD0I*H(vkyftBxS+{4vCG!#&<|B$L*?FjzF%-7{_`9O%rp}49~Za8F@^R zAzo&PC*#Dt|Y8g$rhhmpCI^5?*lN_dJN)XCEu>BJ;Izb%KghLCi z&%kQj2{K<7EenG8ys&OEP{`RGP%T=9_! zsJ~4w&TL6nO@D}_{vIQb+t{^NRoKKFjWAxMGc7GC$%HEU9+CkSR|75K+M)Tw-^Qp8 z0;o8q>fLpYIu1;|Ob^sxVl8(};RjGsW8pP&%Z2weGH-!=L-G-Ig()Esr)TN`TZ<5n ziSuNj(Gy9RdB^cFbcf40HY(YmM3nGsS3GdD-ehMBI)`p3jUDqGuHT=gleLi$K8Oqu z5VcjzfR%}f4^o-?aznnY$R?>9V*_T_IAb8>BsN5DMvEgb1~C=HFg7+Yz;-DsA4EnO8f$k*Ts>K~auOEeI) z9-D{OA>yBPbLxujG9fCXOVQ$8JBF}>`LJ+6OP*+Xm`ZIDZ~AD->wMt za9?Xz@Ozw>m&bEi6w@#W8p&P0d^vxKMcN|(Q1A+@2(m|3CdFgPMv)ud!h^=yCzov} zZPf;F+8Qh>915k-OoQWRV251+Vh>`;ro|~xfDX^`#>p36Y(ywBATcrVdVzLD#+F-{ z05(RJr(+1Cf`4hi#92Yz@0$v2zQI!U@brX-c<;~Jgf|lCDe)edWjcksX2!jb20DXP zIjV8Z#(9_VPCw9RD}()R_)82YeqqucoSRA_TCqP~P^lT&2x8yQj&*2Eop}h@Of(R36^q+x;Cew&sYMc>B ze8n%c^-GgLO5E6>zxP5=m$RFOWq=gtZbmdfUmqQ)t&z zm>DHow~6z_iC}K6YsBGfxLB<9T1EFYIH7sh{LonwpaLUo>K_=m4Z$y>q+Uh=2)Pvi zujjdGO$f4V;XS>cIBbVFYk-Z8YGJIyhv;0`t*A4PRGtvSc1_ydQws|piJA0m_Cb&) zz_kbm5glbavK@Y*p>)Iq6W*8ABkzjB39Fdm&#G4tCVLUKm}n_T|u$q7ZD*23?QUkB&K#CB6(CR zjS+}~7CnlXY6CMsX?DyHmBUY4H7a^xT1RAkv7ZJF_8@`-pmr9Wxtx;DV*&WlO*dC& zSBORhF`DTQ3m&LH#F;iXjCc8BLbay;vCP^m9dSAvXz+6^?l(C(?@_Y|+Zc{59l{T_ zoo(=2pqi&vUtDqI#M~y4SC+8B=-G4dnKy6VoWu?{wzA^hzLA)HzkrNR8L93(o;)#S zL8O&K$`K+!uw)D2KLC;v$1Fnk!|5=_xCrQJ(P=!u(-7kI@&z@Wv6gS+Gb=S?zB(9y zBLN2z{A+Y>wt_gjDZ+ZP2g7~DU^V7GPlkB?AMV~f9P52;8`r8;t9jHUnvkJV$(W%@ zMXAgo6iMcp5L!*9q>L3(hRl>XLW7|QB}0h2=uR0k&%EcQwfD34`~HsOJ&xa>Z^!ZM z_3XV$_x=5TKG$%b=XGAq1#>|={ef6=2!Je^kD!ux4I4CQb}sgPUr58A(B8G+`G>H0 zKi{)RFn2LJAG=WHQ}Pca))?0$+&R|k0^}22;&p_qa9j!XiAFpvgevqxmlKRPsi~>a z=n4R9xd-WD*6T*iSiayx(#RHUAR@r&KY9A}4zmHV77DsM2Ax|MI1Y7rQ`<2(G#@i` zm{0DhR?{I%HsW~@;?IC(fk6t)IcfB1+OG3+)}z{zbX?Gv0O2`X+%V6333mlb$wk?A zou05ie1We8mXg>4yY8lK>SAwknxP+5S@9%cdtdhXSJ&o3jlYY^cZg1BqWDu^tl^2I zgb%wE-v$GIZ%QBt*_1cjJU?*&Rv%}f6eKM^=LE9GOMsm3L4*HGvV86wCwUCf#3!Df zdIQ*+^_T!qd~RiErVy)qD%UZ+XpWHyBaa=0z+oWS?;(IvPB^s`CzqwE3)GPX29ak% z*;zhs^;RQte1lJhE(y^3Gyw;L_VpoY4`${Ngsc(|}@=933xv+dY$w*9a7|gg_#3WYyMFml53Cw31U(t!p@a-^sGw zqkP&YOE#6oG@*w$+NLG)%dTl3DQwr;rXKZZf+C$Ro`d#)03^b2Z9a@dl}*lJJKfTE(G9zF-}R)~Sf z1xaJLN+~Lj`pF;V(!@aRFkG!z<#+JQAqxGmaTY!gx9kb@bCVNe`un8s(?l?Xq01T0puPBH^HDpbhyZ zxQ>UYVLaRJLsLxlf0!zhtAxNfsc&2tcri*I!w}B+DwH=Cu+gjg_7$+*Th4KNUM-cA z>rsnbhxiMF|K_re@InGN=n0v6+g6(725KS$vT;Q%E}S{|h1rL1t-IqqNcLv&ZZ6Q< z*9BrG!xI+v+bcS&DGV4_&T+n*RO_L3*P|2J59F2{WYH~^ z3Ma!77B?(3n%}`d0|Etz(6;1khYeTN>eDQHtvcnVTHV&W>;;|r3l2SRRYT^_o%<5m zB_-EQ*Zp13eqew4{8?I1{OEmonFHG!54<62x5dnkes?$-)~yD!4$5F3E*e1Yi*%K_ zbuYKI$|=+PBIdWy`$a%1?AZ_)41GL3RW#`9Fxh#>8TF5CVZ0ox48p|UUS9KOFXq^q zW4{8_CBo|&&?qBS*`HtM)$f1=gN6eDg>7-Vs08Ma33xHJEy8C357FaH0rseU$Q(BG&YP$+>T9BR)>eqzDMWDm8>+lS(PO3)fTL2;Q7 zAnyMJARL`GjWTXPBB4*9hy!)?BOn=X6Jjq;Obwj3XnfJ|!VkfQWcgxzTcXL&5U>#Q*-jK*7)|9V9*c;90(u2Te6}4`(;121u~y>Zj@gUW-5rSR41q9p9SUC( zeZ8ovs^ZaaBTHn2sv@N0rSOdd0v7VOZ5^^QAE`v%8_hsortm>OVg9yKHtMN>=9#Ct z%Cv9Yup8o_(U?;>`#J2vODX&RlZ-FA%V4Og*L&vUBBxCW35njQ z$)PgT)E4M|(}0*9W+I}~u!p#s05X{{**XH{Z&%3!^NH4a4lzIt-!Wfm&tZ$(Q3%S4 z)QB<-G+%&h4EDfFdtkvj(IxP#yo{=TW7#_)V4t6!I|!~6(7uUs=6g&6e4x+3nCp;{ z`q%f*tPMyFhSR-!xM}mR!I$j7arU8I^!}9n`DI?c1)v|Q%)$K4xjK3UJ?t0M0&sc@ zY;x#pe_;Oj)%)6S#sBV1IsT(F{Y_7w!a-gwZWHs8siC3KjG34sBq^Julg$Y3JN1gb zpt_}|<)!WSOw24|;AaVS*EMrjfwjh;;~_08j_|jvCj7Z(oDGNrZr04jPvyQ)m%>Fy z2v1@S*O4~`1+y5=L+sj+Y=GAa2Ob*(BR5PMoz>S5Xz3jQn~fVTOmX%{ZE{$-$44;U z#TX&9ME3|tSGuBMEN>Js`P^79^km{;-?iQg!Ce&^i$l;t+*&7g1>rm#uMc&W(rGeH z#8Gso(2eX0b0^KD^jai%%$kxL03`iiFtJUT0APS(yebkX;3T!6>%I2unL|=~i}Qzk zS62y0vA}S#A(Kbg_C|yDrhbpiYtdlSFir21s3IV(8#*9%!>B&JOv?Wx%CdYxRWO3V z4#>kX?m8FKO!(H>3iUOZ-UEQ%OpVIgb6xoE02;ocwISc4{-kJB^rDu{v#2`*6kym` zL%@(hIr&49IZ1ANWh9a#=Hj=L$0CNORv_TZAYh&`FqpI4@A-szByQ+B5)M=PLrr?2 zu?DjlAQvd3mOqg6LcOqBIT{r|rZul3d+0(rjuSecOcgNY#f+J1gAWfx@>i){M2xYp zu~GI(RB}z4PG_m=1GoQp??>GY$cy+Xwp5=^8H+3Q^TF#HIf&< zVMu7i_TNaV4A-?KBaMIk?p!y8hrl!FC2E`=Na9fQzQptZSYL8c0i8L~lz5l(+6&Z% zAE&-2s89)X7Wf;97GkIwK zQN?~iw~6!nqN%Criz9S-3efY)Vq!`fA%vTTHwsBE1L_Vy@e4#A{I|n+GarDv&2dnZ#9*=wR*~dd?pQ}|R`OS*}We0bPwI?1( z?8Af==FpI1PQpxv`hf*Ec=i*h^Ww#egua8ol2Ms_K@3K|566e|{Ntz1+l~fmwovtB z#PelS69>|&3`C+Xsavc-}RhW3}_}GBKXTxm%Ox2f3)Dd5i9-zXM z0cC-a&Xag-Kr`BMeRcL>bo~l7awyoKKRAr$1pYtRys#MEnR1-+TpSiY3;rI~6;P;7 z+mmu|VCl$T7}R`)$y5A-?c@S*@2^Y2~PLOdyj^2Ae5_X|Isw`fs$_jzXQl^q#e)8{5UyliVe+g=0- z`c*{H^K-7#D&iAx{-Hd?%=2vM*-)asf(b4Xf*YKPTtou?k?{VjK~%x)cKVZawD`&U3=4%z=cRIcyg*aa_|O+lJ+zyId{{7>ay|Is5Hk%l^V82mZG zYd}|N^xAV~JIe-$7GO&78o-+ZTAkg-H4D(SU0WYoRg476xS& zG_9pB{Ivd}`~1?SANG7wTdr^xadg|yXavnbtWx>?t7PfFi$vBzKmYHwD*y2$+}-C%rH-4i{BuNk1wmiH17I#-&b88+85xJr=fFeR z8Qz}cH-*T7*@wM|h-52_GUF>+7IGtFVq{aCj5mfyMwo+Y{~x!$#QxthA--9pPJ$`t zg>j%e=SiOg0Xw#~91YcnLDgA6gczY-18if##%oW5j~*2Nm9=?O_~%MMDh~|Yc}_gZ zpZ*K?@cB3S`R6CJK-7n+qqnanEd_nWm~l{|dU&g7S4GImH6rD|p6c)a zGH?0+yq+_dRoMr%G-nyin!r|*JM~S~6X-O6;HIyeL%SsY_g}0l|6_|pr3rp7AkVa| z&d82{iFY_Uqe!9}UV$ua3Q;pGL9+)X1_s~p!pMLUdeQ28I)D8QD3i=)V1`SmCXy9d z^I~3XB~JO@9&g7cIyP*aMOhSO-Uj4(yniYINKlGC;Wb>R^MmiUFI&Fcvg@-LT|0Qj zpv47<%0_EtfO})F&YIlvcMLPs_mDK5H8Prqzk&?0(E;Mz97_JU1QT>w4t?BRjC>!! zS^#QaqQ*y8gim=rC}lvurFZF5|DE9V~zzR<*V4f^QW~ z^`I5N1N$2sD}@yh##ZblGqRUVQrW zX{*G^p_gMWThF)RZziKBkRxDh78@8Ckb4!NI2z8eN^HXGrfv)w7BApxR1_CFDGCKd zPSn?n*S5`Wx;2OD1;mA~_*uAPmjNx%SB8A(8)S!=a-kuqD{hi zk%u8?TqE#A))7WZ!bB8Wc-g1{58_9n_`ikA!z@`nh@K6?kJ%)OqLQXr>_Da=mh*tu3x=66Z74-pb_H9p3-y2-AQO4M7@ZCu-DbqLhad- z0!dqXTf}dr1U-0gi`WG;KX_FKz=@!Gqz5nu%b4IhM`au;?fH%AiU4hi@A_SKZpLXr z*^GlMgAbqp)*EB9-?AQDu@b0R?c9+=H8xICGgx){K!Ed5Pf{o<9NYGPv2h{IyL(KME?^^v{ZQsbYcprmK)S~E6!38};E&5Db`gc(I zx$mEXMB(o$7#Ysk+e>rtE+R+&+ISFsF=wX{j!X^d?{$G#kmD?ZNS!^)28eCb0$ouf z?u5|+vANzOf;s@TNzy>pTeF=P<8a72&{#uE2q4+F^^dPPKEVH@@4OOT*952HMy5_o zLWfDs9QJ9wd1%)vqBYiu*Fy+6xk)rQfBE`#+G!X9I;+nQP=Ei+7X18J!=D>a-x6a| zfEpX_i}&~08jdxz`(mI1oXtjrA1(lrX&jsHCF3`EBGP%G`SXAr>T3NZup%_DzHPsWqi`<7(U6#bC&FapA+!y=l0b-%G|L^AN}z-mk!bnRXrgKZp^{*d8FqMQQ-_ z(Ryj!Ibhg7z~}d@xq0${6R@DumwMlBp;vn-Rs^~}QY2yC{PZbI8YODN92_6d1OJvT zM*G%1TZ;Xc?1#9{3&K?h?=i=U8*$@rhN%6wny14MPGI=}AMa~?cBHA3-NN9PE=ahU zL(|(Jfrkt~aVC>% z;w1dxPUT&C1}McbV;~p|A4t$2pIz_mp2+QYUL!szG+V299B(ZBmMugYGLR!3B--2- zm(vUIfG3`we$Ha-T`Ro1QxBG*Yc;QdXA_E!Cpcw%HTECaCx>x2tLdt`S<*8OC>DuX{? z)4MP?Pu1XuWn8>#eU5!HbVG^gd#i_?1FG(TwX^zLACRb_X#i4pcW86gu6q}xhB?Zu)RX_3r2T90>Ej)v_mFZv zKtrC}Mwa=~<`wnd&`SccMpO{>^-yKLsd&sBA&UDbeiA*-fV{$3hCG_|A-+Y@*C?P zEeEDpq3~rd22h4OQY+GqUn%XXPBT=>y|Oz^ew>kR98@klIpSj(<{7ZJ`C&u%D+KDoBDCcJMj6NqT zdlGexCILwe2SFw_Dv4V9gfaV{LesWonAX&69r~zxSX*1uaDLUPvjQ8VxGQQIjzU)W#~yVJ4Zge^ zrR`2ni@Nku_o!kXI(gvb%6kH)GfePymV+{J>~?i>a?0)gxOVO7RfwR#>p~;)B2cwq zF^Qsq)}NADNe{;Z=}rO2y={6q=CK-zdl~rbYY(6%!LK_#xeVjyzyJTGtcrjaeWQTZ z%mZch&aH;0@4pPX(PzY{E!>zW*;s1J8VF|wc$35nllm!!f12q)-SJKEFLVcCSk@Yvp&h18Yg!GabBKMRl=4#*MIaNGA2@>fINX&>9CI zo^svUf`vszYeUq6VQN8kgJI!fD_S^e}A=-^`yB-DLz>2A<|MszKn zZDY`$(oB-92HA|k=@6~Ll{s(_1_GH&yhI>`nxIEGE@qK zx43&JwrXY^vwQ#rDF&ubRZ2sVLWDtedp0NJk6?ZeSy()glIPy6=NfVB%$W@k0SzSa zE5UCcts0Re2a_+6trJ>)^n>Rt8n3Hdx+DQr&Q&a9U48n{zGoSu_#26>-AFcE4@e_M z=2xjr(UFm)Wx-hZ)8<_?1`IJO85SfD{$ZZXi;WLHIAE(K@{E5Q(VOS?gim^>My${CMNU^;YCuLWs28LR?6s4`J~KO%)Zxh7gjtoF%&&JfE;^~ zAVM2h{c-Pp=rL?j;*#Gw%79&2H^E=ALlATb15`~8dU|>^%0@!w?g^tKT4G|~S%|5l zRifAwX%?x<%)Qd9S^nnDi`yFyke?K9p4%TRfTSs@4CJ+9Ya{(C(*GYF*`{aYd#ize ze^ysqOsoe)T4>GKro6tVp!3CG`hi`%sUR{yb+I9YVY8bD z5i|&~7=?B>NqGT6h#E5*(syXIi5P)OY6tW=2}2OF&`E};@M84Jus|ox43VJdAGf__ zg9$sRz%w|tp6$V6*4(s42#;h5_EI5lw18aq73P16@db+N zD^KgfOAj<&#M#$3$etc*<)sj_fQ}&Fa2nw;amCy+9>h%C>C4ZaJ=@2S7QuNJZFu+w zx#)F8#bRpb@pN%3aRJ$gfWx&SZn|kqQ0e@a7daDJh+DKBfD~fjH%Q_{#0aj)CRAI# zuBiUNNzOsV5|LGoC`$BlKvYA0Jxb!m)SiQ=LKW|W=AJrhTo|$n;cV!JL<_;=<0l_u zUlgq1RgD*zK?t~y_V7ZuibWc;(E>q64o{(3M5_XeVB75>zaIOz$hi<+_+jxUz1h&};Jg*;G#>(RIKTYDo=v3Qc{KB$Xu9yPzh1xi8d3&ba<`ogA; zxJU!v!BW*#Sj9FM$Uf+e7K9?$67r*t<#aHI6a}A+tcpC_Vdy%SHf>o}I-*c|k2MM) z%h%v19>BXr&3XM_D#W>M5X1h!WCMK7{y^)Nn?HmH_5|YM1ld%31LM~PK`QHF)}<{X(E9Onr6=6z+FFt3}cC2IiUhp zFkWE0AEE>x_eadJ5@CjWEysNr&SSKz3W?0OS#K8tDFWn66gm1~SdhYw=3_5ZfG{zi z^`&!|_>fPi#yM|`Kwk^GawB|{ywR%g#f@nJZos|JQ~i~P(~7i5?vC`HK+zHE1tgk= z(CZHTNtp*oS2#}Q+R`Dg%Mfd&)Q@634I@H}A5KKmb5K^4aO$I#& zkMb*Mx?f?61HVLV+g9M7ZxZ_sh+wB~!Cz$hgXaf7Z)4S8KKK;4_+&JWD%F5p8d>x; zZYtLRwkm!Bll&SSeDC-9txK^fFo7Y5mjw29^T_jQ`_oVqoVSfTOA988eoYpEPR3OC z4S-H)Pya*(0wjw3_5oEF;9pokU09c8MUrkzQ50E*XM^g2a8r5ZZT<7-Z{WnxiWSnM zf{(kLx%eSCr9u>9mdp}X@3EL2BRH0pP_ZTqLC~=mkatR#5(aI)pk=L_eT&uKI3BSEj(pXw2e8GssHj)>R{K#ca=A<3~q zQGKUzOZDRojb3Gj>&KFSS&+--GIvd=Vr@D;EyZy<4E&3>^P%_G!+BU+O1|m;Km7ZD zCcW8-G4d)HT|s(Rw)Q~7|7PHSIl;#PWG>>c8`=BpS%l4?Z`z_SXS$BrR~RD>lm@kX zUFp|^jI0}C!TSkAbVn(@VZR8aHw3hJmjO3c+RBTTejDTr7fv?*PMPOsjbvWHTF~iV z`o#?mnk=`U<6s3b%1SCxG0Ha3e8ptvZ?A7I)b6{#m%YL0pRBBJ8uHPbyks&Ffgt1{ z`C(b1v?wsopWxoQZJWLWTq$CIIH6sX?S;}uIeFr?A>vPpFwdh0#aIpA&VYPw1%A_8 z;cf&58Dyf}frNw=nyNA8un=_7JTCUgFkiommXz;AmNc1~Bp?fZg=8zGS_Hu#Yc&;i zZ_P#es9mxI!*9|TDrS8MV5v-C#fJ|H5LxoaX-alOLziYY1#wMDFE!{m7DpAF(R(c; z*J;d5WKeNb-nm}dWh<@!8vx|tf%?&de03b~`}h#;_>H>};91j!1gP_>kz%1U-;7>J zyieM7@-*$OM#?(g$~Zo;+G$8M84j07`$sYyF!C&eK%w86fP>rI?TEMv(0Jopb5ql8 zcxPOIfz!WB+I8He`QwC1wAcJ`VItw7rKj(nTKmhML z?fXKk2^5`OGHws7u>j#DwJg~%404RsHfFUp+U4sE>OOsf2po2n8*=W<@Sm%r-FxRZEn>iIeEg8p%0 zS2wrQTgyVxz8Mpu$0hQTIOI!9`K}J0`$V26N-%C`1?mvjA$ ztDnw98O{=EmkNO96hL;21A#6S@+ch;8*c{3P&C+n55orMj-1hv5f4;FJaN)IJV(%` zuBS~3x{!{G=Z86J-PqATl++}h>!b}q)Jhbq%OMJ>`>rM}(pq~F7W&c+E$ zHY5G%)1?q4tU?zi-j0N}3_$t{3^CmyoAD;ovW18Sgnp$sW0&tnWLD0m=ME7GU8^=W zb$!FScb=w?oj}B`XkWu#tQ1-85POa0b*3#J$EZi z7iryfU_SDhXMJM#wL54Lk7U@}+OlGat!JC!s`cxWt=R@%$Yb+NqTifqt}R;#P?BAP z7fbS!+39t6QMJbIjIn}D-jn-lEnxU+>sey9$Wc}*D_9LOQIu9YJl-6U~$v*dw zOqlEO5SOsA#^UH=cQ^?eGC7W^3*#y`iUd?D6P`@G73ak59OLdkNrYp7!M3S~Cq= zagM@~J$-jwx^&6r*5U;Vla3xgUgMx$b{r|542>hpQTdksu_#q($(8nb&E|D=b?T?F za#LXf0|tRi3Vh+u|EVu7asO+u>klTr=aT(zD&cO|_eh8?f$Nu+mI$>60J}>n0f}A) zNY7uwfa>^Nh?r8~B!tFl$#yF^k@)w)2Fu3G00U{jUS7k;-Pfi6hXQgqf9Cz5c4+se z*UQ)2!9j-yTVXgmNu1jEf47u3UVx=gfzQzwMt5{4`NO%SreZO>_%$40xynE+B*XjS zPa-Z8V&*IH@y4neK>ukPX<}{geCqOm=^m||yPyN!sHF<`{(AKzTHf(wQ{-O3&=(&6^4R$SGejLqW1ttJ7{fnL|$j9B5)%OvZO5 zIBvlGgfsd|=@l(No389zgIq}aUoACL$&2|W`ySN~m@mCW+f_LYQPEK?G@W(fT=;iF zf_dJ^q2vrSW$ciy`tCk3^fy?2)uhvNQI}oqW&4OwK+~3?4RF)=8lRl!8;PYP8>FN- zz*14o9^PYu8kuHfWi{Z82(ybcTEp{NAb)(5Y&$&CfIiN`z;kNTF(6CDXm>abfndl0 z^KIU>%fZQsR+l4(R6a{YV>&&^Q3^B&l}vw9<__%d;6hGMu+9NTsEd6XzG-Nba2IL8 z7b+EK3lqz&NQaEDEmw&uC|W1t?S)Sao_2HK7p%39l2C)CWC#GBGmsi+zk_oKWrSk) zs0v>|$J)3jcO{*KF|&}Q_F$Z&!mt;cG+H~W0WH!MxwlRr8(wi!z_V42_BhO3PoQx6lQ zslbcBiD{UUq%9b`SqP5WyCHr^mgL^}4cfNnD{Xx^1@P2oqwMk}b30T4OVRIE=14%C zk;sVO(Bvf)=~H$KN@*Nnw`E*%KLWsV{GTfUns?EF2pYbBF9gV~5BJCJg_o@=n4g5y zSFu;q0jLtf(1nEW%e8FXEiV(@U)W5h1M;AcLxR zE^%q=Gl3(I_(y&D(DZFU7=ZZaQRUHa%=T!&vDLdp7hgH)!lZ-%sxLs@8!c2ht`68`OS^+g`n4AvI)+0>%fE z5z3Ae5x2zSUU8NjSR@u%IkWj-B!|J$r*&gO$+?3XB|=1j#oM}!Aka!+hev53yd8BC z54MeF-7>y8Z{ED-yLz`|`JBF9U)7AYRSEsVUAaR|kuy$i+Qh}%*IqIt3%t?`heIy2 z_Pu{Thll%cxqY5eR9|?Bk^7Ktd_%F>yEO(OOPxBN9u8>3wi!ON+<}_Nl=%YBpD11p z3&;ec^SJ-c&%E^+JAs}ZUvuG0dad?RrQIqYmkcqnAuY1sEm#qv9me=~5x-f{H`BM|-B?}Y{ z9ypv2SQfG=Pwt8M)Oep#30upt^{lx*Wnz8r1eFdwGffaQ!yMN0l2o@&d<0(o;QNwE zP{L><)kUU2c&=>JKiQi%*4X`6;)TkpF|83dwf44sLFnWoZ?4n{a0@Bj)B9$#U!cLe zyLb9GSUEd&J!@4Q4Vk|^>r6z%*V=$Mj#lJ|r>zg1xC2bhZWIhPISELOb)-h7)Q@jx zIJPHxbMve0-ZWN$l?rK+9gqTzW5VTv#6XpjEYqnoT@9`l+Yu8(1G9vq9f%GdX8lUcv%cQAh`&%5{f8wS^rIz)Z7U>)4gxffXYZa}%;97bV9P` z!xhy7qZ3XMwj6jhkxaki;<5D;ukd;Ntjbj|JhpI-2v!gP0 zc#^vPe1U0Nqo3L4O_WW=8@Vc};0PTZab|~n5huA8Ct0O>LNNSm zNQHAFZ*+)a)YvxEdkGf8A8ghcG?->;1)Lur%SMOs_}!#RXiI`rQp?)-UJncCEPGze zT&$cui6a`E_Po)q!PM$Ybgy$Igb{ks1_hWUKjL*=7#a`|uvA%);2{2O6Cj z#?28vQl6M5FND(9uf#`k@DqAklYr%!ZATA?ugv^xRk~WpNV(EP!{gwla*4ioq9_oR z&d-S6!ziSKVjTig?%Q+m zz0Qi8?zk(RkQ74IrSjID?G}2L$POQ;?p*tHzYLM+jZI#O$JgiV#CgAor>^o$dqm2h zJux%s!{wdKu@DpmwGAUo%+xkxr;;}Uvcz#QroZU;uh)9mF64pN6YVa1JIseHW zQG_L>LwDl0hl#PUjOTkx9)rDH@9f}N7zk|rWCWk;)8yDGTMClvX@gmg3P1U3(DH0Q zj$VQMi5q5_jdn-nv|~IWsVcOtkPK25wrnn-&uP3l%~8oE?ORFAg(A-~uG)aweAg+y z%$%2S6Y}--Smtr?d0+XSYB~SUcL4(Kne302=8_C?Ku9i8 zI(1i3^@II0d@}zzGwm@G2s0Fw^oFFQAZ4kZ}38~24_<=E=Uot2HfXXM>EStA0 ze`<2=6kEf0j$vs$Cg1Zjj&m;Sxw?LQ107^~a+J(thx*z^ebG?CpvWnpT_~;ZnR~rh zxlw$6QJPEwsv@&pY(hreUl`?fv(=um8BN~C7~t6gvEc);rMiBcETrrVJ%~cCLc+d#qIs8~El-Mbjz@ zEF5%bTX3B86h{Cpx0LWF-$Yp01@XW-Y%aV-@drtFIsA)nqG48x4F#|t?H}M8^YpQi zlWKk(kL$6jL;H1AR16TZ8!JRajL(jAr$83ngrj1Vp1YIPIS8Ls&SA$6_OQPE$$@10 z;=aCrVCzdu5u5qVu!#WOtP*)Z!`uWbC#{l-P9pdY{KmX2l%JGF$jlCMos$rJ>==YX!-$ceaFlhR#nxeA1G-bO@=j^X(8)I4L$nd6h-zD+kd5`7yt$snKk6(^AK0_>tHvxAmJM2K zry5OSA`xOlCh=RM_bd=@zP9}M5&OSbSl*v8Ni7q;gGyxg0hNQKDbp)vw@;|dpxkm1 z=|g06^r%4YiaF~n_S(KWRtb2-s2UE8hLu+Ghr72;-L3^IrUx;A;k}));(TjkGq`Qp zN{_tOQPY%N6RhZx?-f&*r}p9Xb=#d!D&$2PSK#{gxJi#cdQ}VWiY2huzbJuJy|ky= z{VjS%*AU0HSj%RQEvlmYhaCp)3}>Lp5jT(61WS=)4x&-ni*M8%bna+b`g3eMfBn$M z9x+rC4ZQkDrR$rwNRP^b!gkzfwT#SM0dlm7RlQ&LizaZbHWGaEB%noBDx zl0UrLj($_#akZIVzQ@9z)TAy46|ER=sd%Z;+#AD^HEOO5<*J%irnUKVoE~%Qi83G} z%{K+hkF0tBkejb>$q)|9YT#!s<1Oo_Vtx(aLDSU139ZS z8)~Jh8?~Jmyw_Sa^cuk2a(a1}2I~ycGs=p5XMKJ@sGq+anb=#$E_>!>8)KB}>4*20 z+lVw>)I+bCSTYr`-g!InRHew<87v!?W2!3QmD@)I(z`b<@I_}wd`O1-i{UeT3n9;l z@KI*Ng*QgMlWb#7$tPHyw|jq0t$6#^gEB#r%UGueGdHs(fsmgBCKalYR_mp@wZi-4{1>DlXy?hZmkz(hk`~RL4U1 zXQXXPcRv2)smrsCE;aIJ1H(pZmgFCPXeejz+6P#(SgbX~&-WA-O0R8i%!+nByYJ3T_bQ zhC`U$bq`xgjr6B`3YMQJzye)reQ)An9Rzb(BK9&sP5a)BH^xc0?RcHLT_&#w-t@&+ z-qV?;)ZfjJ1}3@$Ft(4DT^$#l8I60f8$V)$Zmgn6Y$+N@m!PyuFJ{aspkP}+Z$1ml z{cDm#pW(ZB_m1*ljiz}N@BM5)sF(Y1Pa=#Nr{!Wt?fHzP8{ypD`A_qQMwLIbR^F8w zquc?kgdL78?)c!sMYq`5fbfiRub-z0ZSc+2J2ySRwZ4enU0D3SHJ~NGKsfm@&T*I| zT4PJB%E5pjD!3Z^+})v0vi^*Sqf#H;u6rnV^o-7I7M6z@_nzIlX<4Ig0Pcgo zIn=(%1AlmPUSDYUK};WRf-0%;k-k45j-msLCH>WUU^D{L@P-UT1wwaRjNY1zq_~P%^1adU2jNE^d7Q_llh$ ztiAK33@anE&?_1z=7N){xnLLEE4bIBL*!}r%hzjPEm*1S-F>GhAtQ+^X8q3&bNHLG zS4i|7Q&h*pX$H@gq^PTPurnmg7PCAqzpkz?*z6;*3?%mSxtY0GVhLlU++H`>G$Zy& zxnPCFa_Qlhi;X_p=|_%zGIDRV_P)*-D2fLtO_`57z#9v>5VrkNx^Nr)tnzeRzURcf zDJ$~^ys_1c02Xv&;2T({Ybe`VtyQQe5-kp;A~F=r-4e0{o;il;4T-U%c{j}XhIEaU zfKHYS=<5Bjrk&bpA@I7Ok00PC17KG%_-r-X*e(Q+ba^ZGH4*q(5Gs4wxZUZQhGgbQ z;jXSPUhjHo?dCsNF5if~v}^Cd0t}ZnUtgttHM?n`)HsjH~p(bq}47 zSw?5?%p~QZuQ^1m#cMkVIoxgaJ0T^@+l~V$|H?3>q}zV!Qc&0m>6vsGK8@$8t;#?j zFZ%V`@@6VjF9ANxFdWdz3q6g@HYKGPwxqC;k1hTZ9S5T48FBlI~!}4 zAyo~EBtP$!D;enJ3OcRG{QDr}P%2wI4A6?2Qd*ww^YN&3L7iJ7zduf}Zw<9CLQntX zLuXY@Pe7Mmskzd?NXOdD-nmw=^0!3a!WF=^%p(h^oL>pHIDNH#!$0CeY|p`;BVqFbD&@K#9Y8mN?_AT_i?b9CtXCaTkpS}5B(Ifr~A2} zA^ohNeiV(YXvz*l1-CnD99bf9>P#Evm?wM#y}Z0AWGB{SK$pD+>EH-o1Db9T^=Q;u zvZYLeXR!(LW#5!oQSF*xGpBNV7qmP|CBj3Kf+`~u+c#04RNd4>7*AkoN(!GzL>|gy z0i0Ua{E1F;CG9bUKJ~Mqo*b|0PHocGOT$yzvJFF3I}$fn)kdLmKmPOo30|NQ~p zxBfs#%`Q!jw5C_u$>f_>CuX7(!C={Af1xNB0i}-Y_9Wh+d%^V&SM1)(Oy0aH&VHMM zva+(k+wQ6dI2aIn$nm0xo``!MI(q*lB=(PbfO!i2=D(9W?leGb6x=iJB3=FOk4 zBsJC+vK;TC03TIL^K3Q$ze7RQBRUg{B!0c%;o)(I8;}GFQjYx5>g48dukD_QV}ev* zXqr9Kr&IY9fSQe)W-IJl`Y2`jmXt_;Li;hEA2$AN3$H#<;L{TLcZTkd&44UO@#2p3pJ)AD+U}?@FgFY4A6eIT{AyYkuuNVLQA}xK#BB{ zF;sG@{7*46K0;BDI{@q?KWGA@6?&%lgf6TYL^#IR@QLS#+d8Wwa&`Fl>wwGJF4J_5 z=P?|CP%n9EZ`(c@1{2}=q=0GAR0y5kM{XQ(K^gqMp`q-6rw@+<^^xb($u!f#=Bk8O z3?Ca~p5kUok&%AK+BYN!oon>lT>;+#%;JcayLRC2J6!TLQqSCu_8-cq@#|N!c7;KU4Um7|KHrh~Oeq z+D;#Rj_MPGLG|_Xj8i-Kg@Vc_&m%2jL#nWRSg;WTsBH~z7T2a+_tVCW#=|?;|ijGJoCN6e2U2O4AVWQI9^K$@qMYQFn zVB6+25M^p1!0%-BQT)fI&)6Twhdj$A3x8hA9-xTTvQ9tqRyCZA>f?~Jt*Q}w37^FE zh$C^T3z%bZqg~M*RFYN+zqLI%2PD}FdJF3ieJFPE0H*ksO_g>84cB-W$d>`3V)Zx} zZt@IvSWjU%7%i{?#l@Ffs6zbJ12^TZeG6)9#Lzuvklp6}KHkN3jjmG@Di6NK0DwT1 zbIZRMb%~AN0P6aHv~gtQA_&srs|L}eor)LSrP41Y+=_R8!hvVVeeA566~caHR7@y?=*Kk>L6-mJ*MSciv9%}V(=p-(!_`;9XkIzzx8=nD;#DI=>1iJ zWQ?>PJ|5VLl=8&>}mo2aT6~?A~pfSmArvObD=Ef z=ip6P%l3<08vZ2;+NU(obs%0RYWzE)pxdASdbKul3J1IhHA}p*0rbN?C=j^yed_Ug8KQVBo7@NAFa(#n%Z0WaJ++msY0jQx!4eJP@gP|+5Jl5 z1lx-llY^P{^I{b{_|rFQO~}KNSgC~1G0R1XZmd7!ZFUU;i~rv`!nTcCbcWiBYw-|2 zK5(x-iO>}9c`KEEf&y^M8yDKt{y9T$%8$FOGkl$w55Ug<<&M0OGV{l$^J}7DH5To! zGT?H?ln+G55zb~nKnZ>smCgbgkHkBkyjGmu49SHuP4^mlUifI#60%`aGZ66u>3Td2 zwQ755{oi$|Y|>Q?>>a8C>3*iYEO z{hNVs7{5-_gDJh6U(H})iN)v6r$H6Jl9d1*=qy6sA}Fs={U9-;o{u@h7i!O96*VoD zsLs!q`pS40@9DsWK=tWllA=8_r_s~PE54tl36DeRd`E07{`UmAtH~HpN_#a&EFYaz zjjpnM#bT6x{wV|?0k~57X^Q<1n_Y%-7+eAR`uZwaoi+p*RrWqUFGK-Ysf1uA?wwvr zkf$}LeLQ}YPczHn?4@oe#!;n%wp+k{yU?NV55qN7Ij$NPNAw2{ahUQTI?Dnyf9lbl zszva7Wv|t7byie|1`x!d#AI(!ek5<4XVloK=gdG&#Y4^vbX6^&^Q;xsp$4Jq|7n?U z182Y#Q@}NLFs6qM>qUgOqw40Xo^FLkd#%gGwlFWXwbNPjpTkR5t&U-JZuiB|M?|2; zkku#Fb{qoI6G{gMqERy<8r@@h#PParBSfEX;KTAQ$8SRaTy1wM*dIOQuJ@OnquMH~ z85nO6it2!{$Udq)C@`=HctE@J`Fj8p3ZZH8D-pJpOhNr5qphv2SVyQOU%ErJikgu7 zjNO&aQnh#4k-mAarcqhu((4x%x#gi_)r!$J(~LT9{lS+Q&=akJCkti-Ev~n=55$g3 z`nt;x9kuOrVvK2H4UKSY2ck-h;(8tR=>Uea#Y=q-XFvdwAaRAssSDIpF+3O!Oh8s& zrBP=RU$Yn>m~!Xp5}m76|8+Lg&~>gBHaS%(9|8I0bQ6fj<4q|V5orY`=Puk_FD(U^ zVRizG>arp0=s~2}Lk0e=U+q;Opl|o*VZHFZ-W_Z#EVqqNAl|~9uvpYM>~4}R&iKT| z$!${58mX&r2~@QEC@)foAFLcZ3P4o%`r-tY^F<-}A9m|=cE%@V$04 zi7X$hfeIC(JAnw7C1rwZ5d9hHr+A243ZYLoNXsr(Sra;6cLa*aDWl#qGI~8_<}QNz zk-eaJrkT^&y7WMpGtgBi*=f`N;c)ajT}36T80nw`Do{uqJs~-uEZ0ahXGHsAHsP%p zW(!l!Ks_$n?7o?%abr{$0KmJ+W|p;)1k|*p{%VaRcVn ziTlt|Q3~dlIQ+;AKP%-`g%iMSUfD+~J+I~q6re(%xPMKx6^!&sVwd@xP)~kbA^$z* z!s8#k&aFihhmK=76FD)i^NBDo<`7Wy2P+RElpR8Z&n^1!DJnj{4t1nriA!HFLz&Mb z0kRR||5Xh)1TDl6+FGhS(&t>rLYY8|3}y9$vpEcb>QRXv+d+6pzj-fClg?LXf13+! zwW_wfcsQ?K-ZabU%*guYdo-@mJQq~a!bwIUY4s)nS znfj~~vr4UvWJ^W=>A!{(GU!#*?4)O?v0*kR$XE0}V!gl8@z@&4Y%Dc0>3v`>-d!ks zj$U`kO$&G&ctS27#p$6|gX*sNu=J<`q&-o>`;YA+f|&{KE4H$hMd10b{o}*M(n$(q z%x=jO_#I8%cj|%=%VXqy+aIo+XkL|nB%9ZcP6yTRTS6@Mf5RHV1dM6d4mU6+SBf;+ zD5=+>k!_~dDClCl(ukYJaWt1$x99A5Wd_UTL*K8=-yZX8y#r+ErJ_K3bU}pETZ@!P z$k+)U8Qsqk;tP}x*~_~*57upAJ{p4Jo+_l~+Wbk|6)RV&o=Qfpnf&D@U4T3rlu}N} z-#`<+ZZ~$jQ;Ky&kv@67sNvgd7W(dhM~VoNSbAqygS@Tjr*a%^J%X^mb$q9;eI2Ts zow!u~rk_!Eq%S+66bOdQzRAI$WMGu|3*vQQ8E+e5OLYl#gwN3f$udc0gGxNAMsM5p z?Y9Vf&A=c5MJ9>iFINNPSO5rl6CX>FA^N$hLR~G?^L(h7Q2+sMq5L|IESYFChIYam zVv!9QxX{DM5bE)B6|1Yk5@z)ju_={6G`(gHyUodC@bgY>xQmoF{_}*ZCq+NtM(V51 z)_9`eBLu{EmtHOUy9m^D)9j<-<|o*aH2wWa1cTKTxr$>i_RV4U@4c-AJ{qBL$8j`8 z{t7SoccK0rMw~?3-|xOk`{LQucD2HxyrEobjFm<&AKm_pb9+uliiT-CcctZ4xW!1& zHS%ZF5cPy5+hN`{YrkACnWVmM$@3L?psRORZ`YT0{lz9ky2@s?)*H;;c1Tts9F0h^X@xq=En zAXTnV`Dl;Xgy{Ycgv^8hm=Q++;yyh zP-W%Ob-l#S>5Mn8l4TONimzps(uIm4NYM^FYY{axAH<2^G%FNz)T_w#fL7b0qE*HKoY87lwKorJ?* z&Kcl5*1$NvhQ+{|Bu=8}z4a&u8YkxpOMx1as~!U|-;Dn|KCD!E zg>TZ)G*Aq0fnv{x-R{#y%1Onz!bXx!`0`>=9f)Uj-xluPxYHALKUuW77H38c^jH{C1KAh zB9LmeolI%=rMx-u1Y9?6;$Cpab^x4^SK1A7PIThCNWKR#3ixUvyA%gV$w7z{r7aLT zVLur8t|m|!d_ckRrlP`!5L)oax_CrAI6TF$ztP3VU^C@s-z>zi`6xf#EH6Bz2zX0P z%T{H4AlcO$0l5&Y_gaJ%PjKEvsG(L9F?U^109DzxpQ9$$NzcCD$5tGh%GXQHa|cwh zCWD`pG(=rRz=QiK>5+Eo9YRRwCgdVzO%evT*I-kgP5OB~zU~~xgqZ|*AcD=I_LVveUj6>IU>iw=PPAz{7@oO-Hu@9lbB18A2Vye;4Fn^OoL zf3>E0u5H}{eMq%0J;!0<6%0YCT}ZuIbnyw)!xF&Vd(cT8LN!T%4VCl<5qG9hPVnwX zyR>_+0R8b2$~lyd_4h=Fxj2^09A{B3;KT;Oyc8qXeI(4T}6p$Wj`s?HN zVccm94)yXd7ubq#!b*`Ib8BRYZ#1SC!%9`G9<(mJsglrn3lnOR)0J&J-LiO_U)l^q zsTDFZGK8Rzew>Rtqj@Su^j!fd0qveS1(*X1={Vu# zvkYU7$$gQLg5{!{2^3?D)@1q)_EG;wubLDoB?oK=wy_tJV|nQ^DPd7G2FQ!$pG+Ec zCfd|`pf4z(NSu;_g=0lPicc}`UBZy>GH@2>D7PsfR?_K*!i?SW)!Z!wbOHg|+yq`+ zjDeIE9N*+$Uq8@A#j*~C*g>p{Cd}+0-FdE;lTeo3q=bfIb3I(@J>YwG5c5q0WswFQ zyM%hctQTFIyIdDSUlqr&sw7moxbk!Oj$L8=V#`L|QV70+FbKLn?!#>_nUGr);xE_3 zpc`-TUcDXY&;1X6qavUOXLvjIW2}W*fj9LihxNOimIKvlwFYs27yG&zhPsg>&&Us%Ke!+$a0b z055T+H49d1n%wdI_MUV}T6+3dYz9Q=%2J6Ss}F&+7s}I}6pN?|z|r9X-?55F2Y|E1 zh*S4%a-9lU)e0ypGdV-4!J?3}SZeqTg^zzpWJpI(2N<7HAlPykl0`4%+efb$t>D^3 z5e75fW=!98Ai4Bh^a&C2pqN2t^Dw)ua^2PrjNAEO5V{a4{AA$&V((qRu};^vakE-8 zt!dhf_JvH5LehqWqM}4m*=Jh`Aw(e|&CIHlqHIG_NhLh8&%P#mq3rj{HXfTn$o@O8 zr>U6#djIeDe&73k-}@cMIF8k7#Pj@~-|xP!`?}8ayw2+} z)<9emYZZQX0dd4#xm-~CrLGz|23xg<_-2Lp>Mf@@g1fp42Jn+`aSsLWoriI;Q`a0A~* zr6crd%jlEhDS<0*x-L+#+=mXADk5Dp^qp+-E>seyN85&qdy>sojSmqI1muwMwGaj902$O)8W?oF_{M61#!qP!8dg6p-@ zDFUJr%cgDufduEysO&vh`PjP;!aKofkeE;wT_Pvr%G4L}C8YyTQ?oEr`Gm;%|NRCj zCs(b?n_vQ@@lLRU4aORClB+P3$H6(4K0YsF3lfztH34x{3j5T*`za{GL(B1N_i4UG zTV>op14;)UIuSw|>1it0AsCf@Ocm06`co9Pb=mvvP#LYk%Uu&la6TRP(p1YVdl;T1^O2!Mp1Dn@(|wo*#g$LY5vx5IQWUx# zg(WXNjpgZ&6rJp@$aZYBMG~VXSnt0=n1mJgfnLN7>XPcz+Y~dq;;xh^Nk(LKnRnk} zbPV;`(78ZO#{4xDkhng?G%ao@Q`Zkv!eP>I{MC%_WkZn5ZbN5kA%>8Ok9RBU>&E4fgQ96f!!qgfV|+DX7D3y=)XEaj zyr=}=d*Ch5j@L&*p+QT?gF0+@;EA`N-mjzTx)N2)6QRLikU&LWJ_ za5!%ZN+UPI7T>F(4{Ibqic#Jn&9^V-o8#{}5*cj5n^YWFj^z4&y{@fwUb$ z&-nxpQZ*@Btq#>PN5-{F-fw_rQffx!?VE})2RAj2AzAc$b6TeK?&*ZKRA<{n>*#yGtQb9(5%Eqj@_uPwFmk_9HGP<+ZlmaG}vh0$fJl(r9)N{n8E?#2Bq zri=w6aaDVQfgN|3idpEy^;|YTMdcoV*`eh@Yf8PGmZ7Jh`t{kK-*(Q3hVd5M+>TXQ z1FJp;?uqb3g-M!bPjE`r7NQVDQar2CI`Is6!@NA7)7I9eLMHRr*fyzAdW3uH2DyyS z_8r>ZmlY_Fq8nd(>gAk`NLg6x84{s6jYnClUw?9>f~KGW&UTFH34_3$c#}+k2~F5- z*I;;~n%qU43QJ&ygL2ET`|qfIWoSM2A};Kvev`WThoqxK%fqjDVi|l3?5LN-IM7BN zkkjh(BPBuQ^6UH>L9JY;`~dCl+(6H2nrA%1VyFE6BeVi5&`IV-hnvYsqU5MpKw+|& zXU{bkAZpTZs!|J0q6}IexO*?@yP$GZjhIQZ)fW07!#wh2$p_pi&bOi_kuq2V?lWsLP{Ef#IO!0 z@vs>qkuUTl`n>4}G=^g^8qJ1Pc_CBFf}Va$k{*Q~I8c!oBa3h>)S6GMYdM-*TUogq zG-oaZT4BzVPH5Una356N5?axSB*YD9{s!vmFwNdspi6=)l*1W>&JwQ~a$Xu5W5}-t zT?nBdhVhn^(>WdJ%?5rnFNm{jkXnI6kvvG72)5(2DLq7bKDoaBQ31KvskO2}?$->{Iia+^so4@!n8xW>P$lS$T|>iqTuIJNvaEt~6NErc;Pfcu>@pzk z5p;^&8aQ-}BMDNF?Z7$~P_060f%qz;0EIFpF3%Aats4TU7;&!@GUH})Jh=ucJR9F! zl0McV?0HB0$zF6kHlfJ44tguKxX1W(A&Nn2d!h=-3X3)RPeUyX$R$&3$;A5un`9on{YR%y> zg;!BiKH5{W6#=aY`L#n;9|+$?1j^0fK}I3432r&Glnel|V33Um{CLfqTUy9kh6;{D zXc5H!Mf9E+8Q=%`K$R@rE{Yo{OYb8_vuZmODN7*DjP!i)oEnu;n=Gf&4aF}>;7_7Y z0=~-W@8Lxm5vMY-GKsYM=W#Br@%9F1Mv!+;`s8^U2Zc_RlZh=pQ3wiIu;dThepNJJ z3-#=wsRnk*Kf!BgDR4b{S6MnZF}u(Eyh|iw;gTiUprlq%V-OnAB_Xq5$B`fupvD3S zp&1P+bQ>bl1!30!95kl>HWJ1_BfyXm8bHo=1^KNULM=zjs{ynnxFyt|qcUmEh}`4uwD{35IH^{#b%-s;Bl9%G#E?;Gju?t)u-Xi zberjo0azcyI=hB~Ro%oVbAOqaTPE(WB+H^RbaUGjO_4JWay;Dmiwph#d>lUmSt0iJv#F2I#ez_=0| zHYBgWAo06OjAg`KA@{isWMEq>QUD^#AbOHRj(Z>Gz$oV3D-mu@y*l;uKnT%iBUlvU z29JV(*n=Q;AN3%;-5#JRUWIKP+5oo9Eq0g8l@xP7d8A80;i$0{bt zgsh+{gh(VTHFxxoWbN9Z2CrlHerygC}^v!Qud278dJ zSOOR)x;LpibIW4-9EjCV!;ARWt-FSn30Y8P$B^5*(odPJs%s^929OMqf-V4Cc4K%j zUd-3YE^Z?ecK9sP{|Rr!&eCm(762PCcx+PZdMJ=w`Y^Nf9nd(}L?B^Ayv9wEHM?!= z%Q3AWSr_rj)Ig>_NMdO`8I2#_fG-j`?~V4z3;PdNHA_H|LSK&+ub-NmO!+SrT3c(r#R!i~vQNN~S? zd64D-ev&nmG35Dx%2)vu{qS-gg>)pCf+nycdqGV=BD)cfzJBG>+lV6<^6=?FLMZ}& z4wR~&1ciQa)T8|Q>%t{jYpzm{9;Y4$;3L}cj`>0fgn-`2aw%?uc-|ccp0`jNp0WIz zK(FaH;HSf%ux;e{E+{J=oumn5<(UMKe#GIrh+^Ld+j6XFDvz);ligNDI1qtV+()1n z+5nUY+Gg7!^8s?g>_r+nhuXS0Kci{(k|Me!^|l1nX;WXQu5G@E5VOF98Nb^Q@v;p! z5Z@=w1JwigKAT@>e2s-GB1c4FP2$^Meqbd;{kFyLP&~TAQnl1xl8)P8=gA41de&|5 z=7HFm!q8ecad#|`A%T7=um6P`?%3~-a;O}UDgl!qtqi7#9-Me9K?4zx0R zNmBL)ChROxhm*bVXa@pBN6MTFeRz|@!cMgqfQbKO%Q-`&cIp_z*E#Jzzn_GSX{zHT zs(t&n(6fBJE!XC_7oD1@i?$@!Vu(7`K--*4qnz!eEeD9o$l&)K^-z3y z$2ziftaHSMJa?XO?|Ip1i0D2Y%; zmrW@0$wXv`q!Jd{GxdVm6loIOKCam4Pz!+qTIUY}`i`o@2~)3zNYA?}YD zR5FqDFS?~qTxyPYvxL7F4(T;QJwQaG0I7t0esvsfRo2;VT!A~-MA%2pPI)c9aY=fu znJ1E_yrGJIUfV|z)vETX(GZ`c11jjR9|(WjqE&RXOhL4Dw5qdU=x_~mQI%OQZdf~4{WTdyDhwD%k#7JBG8c-HYGjOv+y zS|vQITBKT(A?(J2C^n2@>?(o`eAiZnFufMS8K%-W;rQ)R7mzP=CXHnBNmDCyVpPjB~OT z<0B+kC%b+fBF^t!y%6V`nFQf-DRn8_@;97^qAB4!tLCYF*Q5vF)kH#A3w z@Q!*9q2wWWbRjx1FnSRSmg-g z=a-jv9H0Pdj)JR?3nZAdL@8;2@;AVPGm#&!FFKOObZt&L=K}`y-|B}Yu~PB;5^6pG zMkS~X-%k56V7_kHm6C>a#Hc{}%gxeifOUuTeJMLTJq(grpM(3M{?Vzi_4V^{5F0GH zPSN?nm`V`lo_Ya*v2{)th*EnhK?p!GHvU!X`D&msUYWC4fJ z@zJzH0+&UAGIazooxrsb;`|Y>V1TH^!ZYca3uNmpqgRHyfw3-%K^6qmlYlZ(x!wT9 zIW;68{^_=NQ4!@%dZ^HFfd)ILTqQH+vQ`4jsOyjbDQcK;Y?Exb0(NUPSH^%4;4fmR zA4KP`PY(YrBPo&uc{q6PpR@Ti{Pbl%1mHn(iFwIOC z2a$aRT+P~5tE{c9X|K6K;28be5atWBI`njKoOsty0PyuvR|(x|6F`p-fx{PTVkXPOb?%J30!T z;U#d=vxY=Jp7Tf2^wOi_E04K87y&N#q;JbAlH48~9HMTjs;YnkZPft2@+Kzm+@6$m zKO{60LVDeJN3iAP=Quc2{BWbywk=t?tcf{uSB6H3pn4`H9puqe)ral3g!-%_GN!^S zJDTD{4>Px%K%lDXv|dzjd=L@Wcf1beo9s5Lt}9YftzTPnq<*Oh2V?HXPbuTpA*e%1*n9i_kJ%-6;Xc*3fy$k|mE)(jc{tPM5^Qu~6IewE7AN zVGp45fk;S_YZ2B$oNo$Lyf;B|k_;x$wwj*^gh~MAp7k#)oVME;!zfqn&TdP5B z3YE~;?_myv5p(TEE&IK+nkiI#+JRmGt3hax!Q*a8PB{J=I!WTH>+iLKpN(GxwxD~s@Qm8##|RPnfa7q;*V2R17NQR zE6%ug%#E-XmUjACgw2DwsTRr>4>&}I zN1#oyHRgFPT(qc6VcYR|yelO*THoC@d5v0I_j7$L(%$f;*8ZbD)^Hl=Mfb6xvj$&e z@Ph{r!j*O>5&}c=88w2VBCapN6SJn^0RIwrP}6rBt5{Gb;90?1-V%#iet|-LJ6SlnxX?8a z$3;2Ibqh;;L~4Tc7P3Uj#$Tg@g>g6lwz^dCwsoQgo;~hvI%Q>&;Ti$f9x1h_vGmV@7%~!;{>q+U|fM? z%l2ZVv2jhrmnDe5JmvyxW=9nNs#Vz6i(%5&yr+Wt01$E0PbOj!W1?P#Ktdw;4TYkt z!GYsRz~*^V;CUXLQ6L*>f{bz<4w4Fu_fxN4 z#0mlLs)3vM!;{0fr`Ms-T?Z`Evs@c@b*$Ijd0oKn^=OMBjkRfO$Ib$KZ-T_wzIe7{oC)#( zS8T#Pva(b_kQ{-!FN#DO!+6t4KuZoD5?59SzAD%!D_e`YkjN-vEDu*9Ev823>Jx!5 z+;|2ZR}=nAA@6BKL~|15J!;+0iB|zzrNe^2lszWzO5l@`+~{QD-Mh~}w6ZDw^D`92 z+Me>0bEiRJ*q$9>?nU|ef}_cs?785^w=OIuO6LZ#snk*iWtF>2q>(nNiwf%Ix`M}= zK|L=xSKLuUM4Qh+rb;C9(EyZ)g4&U!V>1W0u!R=NXZIx*HwYS!rI>_6Ar)fOLyN2= zPwFMfjB)4%7OSJuF+g{d)yp|z8H6fgr%_V*o{mZkS0x6IWXw3P=n#d_Z7qLcQr`7M zga?*B7DAyt=<|T59XBtc50vloDIP3wY|)fMhXS{g9gIA*L~ly`F`x=ijm4yv73q z0?Zozc4F7R8wI$X9O0Hx)!S)k;BrA!HkbqJ@<8%GJO;ZKzj+ z^yucX1<1dmw?O!!TuolvcwhJnVGF){IpvDsjh$9baBL%di z4%1#R1{u^zK$$1uDkPp~X9xe|Mm@JAklG3zCk4fTx>lmYsZW}=9N8Gqkxj`B{gS=C zJ;~cEAsP=x&CTM)!C3^ZiD&^u<-*Rx3Oc~R#p|JUWL+K53M{0>)X-q**u+7b>qEuf+F%Zz-lI#U6alKAD@uxKTcE639e7@_Ln0U@OG%+bK1k^$qdzbIPiSD$@uaO;4Ojdsr=850wI0vh4Zcj zuzMOgiZ0NVH(BKq&VMa4tUP<xAKt|=qVGQ57I>b|w%-kuuVT6fg*sXnVrm-AMX#lT3> z*}Fx%5G&L#mIk?FwMCxdAT=442Btb>_Cm-1O-eJpdNSW`o1KGRjfpu0G(x24Y}YWMm6_qQ9i@fBz;4=U7O*1(-4^P$5d zs9wzP9Vz$8)7dokIzsC{hM*kaQ>x{m!8C|r769J@1)DreUlO9{Mxd9WWzSGRFJU#B zz`jY2^f46cL90O%%1x9N0ToCqq1NFolZ4u=3GJmDGuKS{Xgijth?yO(9VsqdyIww{ z`o71l#N&$Lf@8?vTkXZ)geu0ofBoce^Ks+dSfry^q)9c+qOKk}ZXU(5>vUC?x~ebA z8o_U+JL;~(u zmlqtX!q3E(~Swp_B0eXnnnN|W*B&ZVizu@RT;_SjQ3^huK%oSTaT zP-HgCH4uF&lr$+ad~w6bh=wGK4D{Szv6*s@*i4=}njfl_80?SENNS(l7oKnFwH)qZYEBg`!ZohZyfT+a_~=AWHUZwh8f`U>7GUyNV_j$UhR0#wR7BanOYGf zGkR_2Scw7^;^+-*0w{0|o?}$HkbB@pRAf{MlL!XKEA?Sf4;Qsoammh^a#cGF@fv53 z)Lzp}P1{<&*gFGTi)TJNmzJ2T1(%k(cqk*K_I76_!yxQd+!^zAy1Gj@FOGD(qumO0 zVn&lu#?a;NgB+GKij}OnkdKeBgutR-h@qH!_lnbTOtM~UVQU6p3e1;UdA-# ziL&?=O+N3-?-?%i4l&#RkS@A2N$cVO(y0eNe# zsh1-%-#}XLN2CMg3(*9=N~^g=cGSn-rrHR9zS$-WvHpUnGm1x*IbYwJ&NW}yb42%$ ze2SQZy{zmyy@9fD8?G7=SKJZ_b1~qt5^*Mr+1XtuD!LtFe6y48dq|is>F}Sg6Akl?my9)Cd~;D$+GCNc3JSroBJ44~j8NbBdyMJqHZEbL zG18-|eBpUf10C1e+c~+68Xzt?*c*!z#r5MR)0Z(H3WI@e;+xJYE&thXUz&oE_Go%@ zyib93N8I?Fu847c>xOHqJ(wqIr}NG`c%B+LS;@*S^!7sX-d{VSIH&q`{iQ*%j2YL# z7*D;MC;onQA4VtN{Mg})7^oK6WqHCvW|X7x=PZRk-S)pFR_Plij5~X1kMrUGl=ZE* zz1IBm9lop9U+dzMlRqr5`h=5}Nosuhz0^5fmM$}w3+7JSoUi|JQv0>Qs#RL8EhwU^ zfActJ?kwXV6yQ{LerPiDUtA^}&IJ4-$|gN($`|>{%A4HRt6tBqn306r$xsYWM0er@>@Z&_Pg^)n~R*746h@#@dNNeRUJ+W6fF z;+czw^$S*s8CO-WUC$rkIe2xSff0RTgcjudx-+fs7 z*n7)Qb$@;>^0qeatlM54n{(XPFE+I?7tXT}ynuvSKsMjx|6<% z9(vz8x=j4-ntNO1x_~GKPNFvqLp}U-CZg((=Gx6=ZzO*iS@WljcduTzI*Y=t&ykAL z`C}*EtXqQ_a~n_h`hPq=@kBG$)XB@sLL!&(K!n{@J;MD!Yvffw*9w%(tF%P&uTbR1 zwIJ8zlHPjc^82#Y>38avmAcpdImQ-Q)KzB}&8^x<`3m59#O1}|I@~{KFx&@ruij|f zUQH)vY)-+I_<|Uo1V#DdhP!db#%OqFTpb9BVm#w*bRM?DsH@M}N_0&0eMQ&Wj}r&4 z*oeRwV(PBH%g(F|F&{IkEVP;o*nF(`TF9FabJxoIiv60=4>v51)Z(V}2hq~YXJK}6 zxJUH4{PJemqbBOItUW`=_+|BiRcl7WqR-b!W-N2CbB`D5EAC6X%lLCdU?2zlIPtp6 z;J0fbbT%BDpoA)*mHu}~Re8DDRfYS{9&hanShQyq`%UjfZ^R!RZfsrq@oqSsia)yD?f%mn@co3?jb-LCfeX3Zubxrqz9a(?Tu$$}Tr&2y$;}0U zebE!?yAH*yy%Oosfnabf!Mr;2n^Z};28KBR2Q!|C9VLH~F8SZUxlm~mW*b$a-gOK& z;+0+!-$P-o1)G;dUMm(AypN|Xj(aT=79VAvJ@)=hsM^Vrdup*c`SNz}4@UlJxi2U~ z{IPw1n_KcsA)tp}NO)RhMh(5FibeL!Z1?`ZrIVU|JhA=fh_C*7(W!gwuY3AmA@KWa zjs3O8zTAPow%A`=?8{<+O837;Q~bXlawgnU9o(a17(JPj^)beLn>}CcxA^;hm+!6R z|MBDM9jR}>%}_aW*!}KcW#i6>Ss!D%qGH$kT-S=rxySICk^e(|$Gw>mOFo6yTAfWz?%6aw z@rD2S%fPHycQ({A^0GTsKR?|6%a`=mW#A6{cUvcaJ(Ry5%761v_SX{qYl;4~L_hBW z^f&%b;C5w6#w|0o?{`ha9wEiWIM&%0jVmE`FE9 z-iPkVHnKB-qxB6^xEw}9aJ*%MUZND3#c-X(zFQS)>wWt%a!H!3DNcji?n&r@Y$kvV zafhGJl)6!^2JcB6L;4-kTXJ!6@#q$=fXFaNknwC443LU;4(LMx zK#qWpXh<(2Q%)kV zm}L6+`}>o69-Vln&Ye3a;Ism&BOS>56CtYEj89{rJ1qr!zA|iC```A`EE5RNZDh?p z#z>FGZ2!fE8%C@#&^S@kVIp+GedHrN)Zpw4HqsZO|EAE;;V$kV$T!+x)*xPPI90uk z+bUVXWng1tL*7l+aDZUle~40o$PF7V48qpvVRYZChz{*nG{s?M+eDx6g8#OZbo5^W60f^%#TS<&!5NGpUMrbl$jD z)YK2gD>y)d&ZGG~{j9LjZ#LPlN9%{u93dE1l9742*LM>}0?o`KXE`#Po42*wee$Dw zKbZ)^8f?WT-|4yO@4GZM7k{hji9U7U7{IQxih!hLMbBSCI(ZUb)r+@cQ#H%_{p>J^k5{ya_icX zlGCpZCQmo?Hf7J-r}+8hvIFQILg~lRPLpQXF0@r8x}|j6L~>ZCKv4DTictvD=ZU95 zy?iNz@AGf5ell48nH$Z(@thB$)QPgH&}!9U3UI>SuSLn zg8ITcRmtLSH}sjBn)cBT7IHLv%=$1mXoNZDJ?AQt`ywYtBK8SQ6>%su zPeYRZC?uriZksdM(tgXyK}+ zG()di2&3j_MQP~lE`)PNAnD*BDOAKT`3@$tq`KPQ$H#}p9(Lsges&wk$fLUP61k1bsDPL1`s&D<(wpFV>#-K|XW z{Haq0|ln&Ge2d8aPPI*H4i!XmU8HU&DDoLmXm-AxXNfP1=D>;RMN02=%i1=b@$qqQafOg#sq~WHF%V6R{BkZkTF4hyRQMJa z7DDdH2f0GN^mvJM%N)zIGN9nAofyMVIk!ScUjT9P;Is0&x!$^aRi3@AD>aVDUOv8n^`yr2H(? z)YHqBE!=bVLX)rEHo1IP(R3aixy*q24-d&75tEMhP4se4ZluOK3;2GJCTx^*GalX8 zfc~V-`N<(?=SpTG4AA@G7(eLJT8V4C@#9aN46Kxs4QZ*?I_rH^w3&-$o!a(0?(cc~ zY~0_Jrf=L1)c3GuM;Z*cHH_y=N@ElBQCN|;E|y=@VkqozxTB+ke4x-zu>hZ>{r&qK z=+@dDG4P_Ht+Bq>4UjB{?}cnDqH8hk8N1*o?tKdAOcF91Hf5 z{pNK^amvPac0+o&`@)lb!kChkw)uD}xu!wdqj>M&UYM;>@P?X)wymzN?&z3B2FxRF zMPfLDxA3d7va%LP2@7Dp*f_c2lwJ>v`v2+xE8sjf4*!y zAT86W!0Apg)JebJzoL0_(C~gdA~CM6=+Dt>E_E;Xw6?aAl*E2GUtg-Z8v5kZ5OP0C zPL_o5t$g9F-jY8mjL$q@Olgz&-b z>AJyO@2xQGJlQj;<~CX5JZfnVZ}^l7;b%6swnh+BRzWeR*QN!5JzD%NRe!O8AC5^q z1axF?R1 zf7SMGt0PB#BEK^7TA5KOsZ)e`1#A~ORB*)n{u|!Kk%y!!QbUPFhAdy zrw2d+)kZ!>8z2Ap$K84R))9)%2A&h%yqk>+U-e8*^k}%z5R@((M-2@PGUrlv!yXrF z!>$~5R%a0 z962eEiqvi!Kgefr1Vy7Oii|6VF{K3(sl+w%(Qpq+c{j&0hs-ll;Ksob?=RJN(|z`E zcEPV?T0D2yCaS!=oQ&2hP4AzZGkf-fgKJgYvfa5Ync5ec{%N=Ed2jybVCmR*9-Emf z%x~tLr$C3jR;|-JBgGNdpUAjH`%2>)XeIzV> zdEamQJusod7()EvU!_0($~tW@R`#7&ev6kv2ozK2r}oJ`5{L%&x-_&zeuIE0Ue<|#BQh{}5)jCJ<5 z1ff+=y)^|&^iAl|8fQ4eQj`@X(emiWOS1`~rb^3BN^V`Z*wvx*88h46vl*h!eCVb( z!vuDmkt{}|=M3-A8d)Z4TKY{kwRn1Or0jUmI2I`zM7HtKSv3#=@@N@63lcl^acuk7 zqp|J8*{BKe`3#<*j~Q!}*ZT@)BR-7w!Ha+7LK6e#mfMgg4A-_v`SS83RJ@my_J;`- zXPnmO}Eg6Y;tGIyBNHAg@<2FlgFYK%*U3c2Q8C$r&?hCx>V{FL7|jVOW=jY)2C@dg-hx zTUWUKky6ZOgjvIsG#3)v4b8PL{?4==MnAzFvVg0rtCRPjF)m1GX*|N^IHf;g#)V(Y zN=wf?mr9=8HTwMd^A!~d<~8aj2mD3a@0Qo4Hgq@5%fvj1H{|VvvZcMNi^fWMlV=Wg zF%rFVg%Zqgy2uX47DLRzWYHxhn}j4n3zW4yt#8*p|vkW`*ELNks}}X$LBMxQ8;wj zAosTAEU!+E6s?`x*1t{j%eC0z<|R)GXj-35pFJ3q9h;EAEieDaJ!0cwL`6l-S7x@i zi$C6&JY;tAB5w{gWL zto>>@VCJ$mZKu|y{O+La7w7NLo1BXMzCtY#hS_t)rEcr5XLfdST5rRF&%@6uTJRG` zGU8%mRnz3&AoE2hM$-j@>gwuC*QAx>;^O?!297o?Z{B4~m*yam@29cwtmBP*sL1VDFOWIsV)1`))L$^HlZxyUrs*I?qn! z^uFsn|5@4e`z~v0v8_hxCi!?H2J+M~KfY5&@J2sVGkoxh$PccK8~vm!$8qV0OeWiU zY_56LnQ1}ah@~s{bUA(0O8IAuyaIGfQwJF5(1#DrbUg2LpL~P@52>M1PGUvcVBAY} zZu)6_J>N_2_90F%{SJwYM7K^5@^#eW){#w@wUIO<}W2S)O z-pV!_)s)h_T39?h>GKZPTU-D8C9wxlQ5)rp-@Lg=^BG1vB{yui5$V)r=G}x-**MRO z|8j~E&d`+Z6o7YI^!f3!&wVdAc!mq1!?5;(imPesc+mwzpYhJt9h7C_E&f^>V9UMlVmVclu z-P?J#lI?_l6;&IK?fdkrnbP5d0|W4;51RCZ|Es0DE|ZGsqY3^9a%3;$=JCy!I4p9|0O-*kQIW70MDEUkC3!Ub zl~pyzC((P+*PdcJNim8{glv30n$f;~9tImh^l*{tnOIsDVVjjSH0UqumvjCghcY_r zXk;R?5ptYQglnKRT0xHEAf+q-4||Jc{cUWl*~9f&yO-#yejnatfld!r5(L8vsnpa| zEPO?&fzShn=;{;4I>AgFkBNzilQr=|s7hWvoviGyU+9`iQxb+V)=1q%^8PR(K@1>> zhBf@#jL~HVJ7>r5zWa`9WAwjmLbAII?S_XUf50cY(6L%a&Z8ObL`C$p=q%^#%Qwck zKzPuQq%Djuh6B=v1IBWL#&X}R|J)Y(_7?xbSyW9=O!Q1j!IPp0Ng~};s=XXOpfUCg z^>!1Ma5+^^Mn-1V#a;Nk9uFcSgs?Y9!@Gnm?f21-I2GE#;H#M``2LH+jw{jncHHAZ zYN}KE&LqrrFvNU@%TDuVZbD<<7_b!GG+;xm8tEfy9#lFEBV$jkm`Jv5ply61c;_>> zs|oP0rGSfQ5!r&;)rbaSj}JsU_d>tq&YlhEA5%mkA$l%dpVsz0f*n8Z8ubv@VxNwdeQ`U*04?MYrYz?rY zJWW{&5_JP~8=ZSDPSaG#$>A{YJQ{{Vs|Y7&Rg{4o5(0Xxc$4FJGpO~IU`^(?>)t8Z zA*-JX3JT~xqlx?q*&y4ax0cFl)XT^qjj<~1tMc`HDAwYVwmo6(i-UEuT6CptzMafP z4@y7K9%J-QO~E^{2%7;e79eZeQoOqR9py?0Zo1-n7$HsvH+#5C({5*HXSBGfir+!E z1uk*V(^YJjT>nKL{H3Vdzf8l)5+R+XwJ5OkWbs^%hyuduf*@{usajVcbM<< z@)ULLElLJw+nVn)d3yVp;!07)KtG9l;w5WD3_CtP6e(RKm!v~4aaZB{FU0bx46G_2 zMvaeykL4?#umq!1PjB<`8*Ny?91$3=9v$s>eY9nbys7Cd`{CS(#IvmrUY(QQhdA$2 z%A%1~*L)?M=V~YF6#k+X(BHJ%Ec&R~ffh$+lMT8~A+u-C<{fb2_x0ShvZ_);39Hw? z!!>lD;^OVet6lHZc`VT8upUlSv7U&v+GR`E_k82ZFVy(+qgB;)L6*`_1%A-zs=dMx z(NvPtz|_aVhY8Lb-cG(>cH4V@24fYs1syJ;q*_Er|JfYgU zpTk#Bf4h*-9GmQkOG|0i#^I+&Ua%Lusja;|g!1a8Dm%UFY~d*%rK)eB#+tePfT3K^ zj_>kYPS(~3#CN4G<6PI%d2sJ?Q&*E^YMxfX`>=vl)H`4Aoa_&2Y_Ac&M)dh8j z)8p2x5pmCZ(0b_bGt<*68lGtVYupd+53SRnd$4IX4R|*i&J=&(690WV_M< zef|R{EAN;r*z+hZ3gfKWa&<=a3v*t1YPr1#>pJ5=UhWe`$>KWwJh%Fv9Y=pwdwFIFT_#)llt=K5kY=3t z4;to$M-p!vOC||OmMd~qoXtrt>)LnkU{0ykb#JcDJ6-Sd@XwbuED9qvY-i`Q(lEB< zud0ow`T&}wwVv_GM_V~qo*Pl%U;Kl8`-bnBw)_ii3$@oRE-ul?%j!y%_uv-R^0ZFe z7H;^SUW znz*Qey5#;eM#q~N4h{CVw(AZ_>pT_>3i)-6~%_v^ao zR5{?x?;c+7&s$ZRc)B9#dxH*@wF})kk^r18$Sz$YUmc`BbhAvmq(ooh36n+ol*xR( zaHn&OV8LI1ZTaB)T}h4ft%V*|iUZllyYQ#{O6-&U2oCty@cLSt{pBpr3peK}?_H9o zDEacFu9KwQHM7S;4>eDpv5Tk)(Nx(hV{X+pQq-fF5Uu;^y_T++`fD*<__eyC%QefH zuGweCb(uEO+}+w`e*~1-^Se%6F>0RERe7oAJVytI77XPP&2G;B zqU$*lu#BGwV~pe>wH}_)Dbm zi;7+>U1YZD34@P+O+fO9TW&S}^suIH*kt;bmM*mJ+xqF29XUH?T_2cZlbJHsT+W;O z&yB$zya5uU%Imr7>X*=<-J>a-sf))NZNF8`L_GRGZlWdH&W5S_x8=ojr~5jL{sC{ ztxMHoLmG-450hCd;iJ&kdplQQpN#HAx~r0+dDM;atfc)x8U|Z;FI=>X2 zcYjzAP;yXzf2e9G4RT1`~l-kfO(J%+*YU&(2Vl9h?9Y zVZ6`}&qqV}e6)9WD>*qiK}J%78V;zgC@C#q7!nBNf1E)G5Pn*i00XMwFs?%sSye+kd9)?ZI}y(BG2ll4{BH~l3=EBkD1eDOoCzKshux)p zftj8seCoMN9bXQoeDn6y*JBNLVngEuU=?vR4V|z@=odf zgxUjnc?Ih`zZWlFtgL`6hCnM=I5rdB^SN}F(-|f01v!WT)|M%p2UY}pDroO~IMHo_ z!HdqBH@k;j(?7a86%G(j>#9F?47^=JcHDIiF(W{E!=#x2Hp1gP?Cb1@{lsZCJ~JAG zqa#SVLRhB+;p}Rze*Sz1kP1(DLpH;;v4|>bODCrG%-g4*-4SEs{QheU!Kd6)N~AXp zHk7BLnaHREcm(jT2!!QIh|YRrSYIZhmA~IR($O$TGT1x6g@hmWn9?QK<{v=%h_>Tj z0h`TuM!j_1=e6b5C~Et?+L5{5aaOAmO&%dU6_)Du8V2c) z@Q{doC7gSv^Y1eq0^10z2jwZ*FkDEKoIH#2gwZoGq{irem-iq~BBBT3Z$3A%8yqqL z*xpS;(Djh1jIwreo5ld9liUINz5#@49aD$)5%J;*PrETb*r6W2tr+KHzn8GcHu|8j z52sBCPuA5z!>^3zCJz33NWi;K%(-98`-m={nbhw;&uhk9;JT(2Gr#*T8uApHH!}0K z9qR*!`rllSPx|GH=*}+oSOpK7LP3m|U^X^Ok6yN&jZF*9oNxIEf;S(q+{DTkNxaK& zF9|zqp#>7Xd50|kF>dt&Qcb`b$&&>!0^_yo64mtWtlin8Xn9Vh9<+G(zz9LjkgxE` z)7U)6_cGko#9^)pJfkc%T2*8OdCn)^1oIBJhL6l^2Jd~C0DfNn#je9lP|Ycrp;X7* z4r;B{?tGhF-<1Q7jdxq6iowae-1#Nx#wWYLJ^+GV96l8dZ3E;Hk5`cQz(&AMGjo^1 z=d*-P%YzqNrT~Lk6N`p{#q_!!}ul`TDC6+-TmZLpGD?qh$s$$rN3Q$K!5m2fpjWOZTnHJVxv=919ESBcrvou%+D>%Rh2!vhtxGlHyS}CC2MsQNaKHh6-X$k6@L4h_jG&C2S2Mhnh zTm%+3ymIJivYVTmf&*-vE~mKa85_49ElZQujz1Y1YuB|(Oy|;< zz8FMV3ZuY__v!1y23mTcRON7@b&Fp1yac2bDUFLFA9IpJXU+hIyx%Q84~g2PQc>c^ zy)gDd-3L7&=!Ke+Uv z24pR5MtF1@gWMDMa>RkmLR&hV?#;KLje$s?v<1pDsmXxiDch{x$*`G1pmcTmjaJw+jNqj5Dt;lCkwk1_Lkf zb6g#!l?Z_?)|G;#k2}gXaS=y4+idNh2vAGA4uTg z*tj4l2To!oH8lZ|jxB@?;KdZd^u!@2;QE&0Ic!iHOA(a z_W9b@_>r#DX;_UvIgchd2MB{8CqaoPZ&z2>=$v9jSP(-_x3$w@>kZ7oZS9%p?&-tC zu8rvN~|rBQC8WGSzVNg~|=n zUd;Y3*m(909yHd2n%f%aIU(I*5C6x7ca5Kglq{iNP)fsBl?_W_ND`tu!7Z8o?V%7= zPdW40*xqV>g)CVAaJPfF%_%=~E&M-3lw3IawW!Cd9K86yPLxcAvG7t)_WmajCBi>q z)N&5&RNYRv;n`45y6oXqwzn6j71VbX2s=;U*7Uv@X4@^Rv3%X5uTTeXXe9q5t23dS zjGEkDe3;WcyXg9pFhcY;rG-jy#1A%yNjnRiet13KJMs>JLk7+=vwI=PXoFkcw)^4; z!n`^&ca%-be;)rWzrr#wIHA;W$8Gw#U6A={ zYNS1N+u^^ED7nc?L!4-y?alQw*z`oEk{Ec3Zg;-WCjA+gy4u?XLp0Ylhb-{l)ReK& zZ>y0iXp`|m@HEF8ZyoZpYP^+lW|zv<%w3=X?2o-1{DadFZsfdY$MT;N!z#Ss_ zH@2lxU(ruob@n|(W}?$1&&?zX1WX9vUW+qMN`9}ssvB=Q1*XJndaCe=7}mZYd2pME z6!0mX@40fYh(!0>eZzL@!}P1Xyg0o2_QkL*#!qz$W4Ls^xtI^#%(gYxbqwMeAMbdU zl{eQG1ePKG`K6#(TZN$HCBr#jnx_Tlw_A~1_6J`y^Wf!37<%|9((72%4d1y2Pc3|R z9c@d=JhpPXQwXteyu-csMZvkfOj^wL{M1+Y2_J21;X2&}A&!IrP+>|n{YMk{4LDS^ zw@K))j1($Ao>!U0D_=8M4@j%(Y22E973Pux%x~Cc&zzb#Txfc=JztDZULt8^(^gF- z8QqNr989+kfBXqZ{ML&X)?4fb=7i3`HrqS(YG(kThY!9WP}tZm@0p(zsFAx#Ap%lTmdWdcMN@r!SkEjQXE><-ys= zZ+bFVdEVTy;#_ihgW5c{=RaKkif1GFu{wT)804NYeLynwqSpV$&pn7YGEVABSeanJ z0n+5L_mz8Dq@RGfjiP z;WqNWm{USyKNyPWsE8aQ=y2yKph1cizkJz@p3$ttmT|m z`!JO@%Vdhk3nLTED2y)4y;>}1Kaz#TI9wLCD)%>L@;Gv}6>lcEQ3C&X7=kVs9^CJH z(IQ?6)e))5DASH!nRpZPj$S1TcPimrG>Kn%WC8B@3g@dhRjr^Q92d3f5oO04GcVK! zZ)#X3xRBp}X#W1Q>Zwr@@>V|##8N#4Y3Ed(aw{qx&Mg6w!2pipvVBRAFC(Z@a{a=g zM_2c^NM>hBv{*SFt63jtU#2;e>yK=E=t7z6Uw=4$tOjOhPSJ&Jb_Qa+i$jfGBR$_7 z^*ZE;AcK`XK2X19emAz&a8`X-Ps8N8MZQ$@QoM0_Ao@}<(%d{$Qk9!iFQL~Edq}Xk zNKL62RS;Ciwwe9W9(mcc@mZjusI392K|c=gT%A8q5}icfE%K!j_1&avTNkE{_9X|4 z{wD(#gG7G+w6=}0@r7USiB=no{qhXO-2jW{;i!uu*3`ysjmrAQraL()CmRM)gQ+_f zSa)I-l|=4sujKnrR`za?KpNFlj(+w$%8zllzV!1EWhDmZ6CFQOUlqrM#EiOkI1++u~9zL-BJ6_AZ@!{g)j+na&b}dHAF;j2sU8szi)*PTqw2+9= zrs~M&hf4m;`om(uU`@=`up3#ZBk8zVnFxi8t;2di`R~ppRW2@_cbr3V2v!~S66@+_ zGUzEz)@n!7?OSRUS=*kcZHoHc3Oj#jb)|3M!P8}fzbT42YAG-8^jZyGVE!FWXUPh6 zRH~Q6C>kGhpMGarcQ4M$41le;goG8pqayF)^HBm1Uy01MELd9qJ0%`)I`-%PR7r)? zW8Zv=2pniL8^RDnS4eC0x$-$9yIIG=nZ;V)gA|Ro9sp>&MDF}A>m3|f6C6^-xterJ zkk1YUrq_avvwHEbT29Hi|4@|TJi%~_nT4}vloUiq8HG>*rOZC{5bJR%&DJ{pTssobpgH>QS<x$aiAS7bI>R3gABU)Sd)qn^=_{ZI)J3@(6qh~*7 z&(6Q_m@*S>U@z_+8z1)!)kSX`Iwt|!sTrv?IB0J~bZ+0IoCXgmZ!&7So5Z-uk zDacjTLztVA3j=6e-sNGN?=PTwcyBD%ADPL`l#@7sXOpb?&LE(ZD;{f&|Ye0Dg zE%(VGqSK`?!=7Q2BZM$wbyoB;8 zm|aqTSUGQF6R7lb=NLkdCBM=T5_=QvqC*G>{UnVlV8E;^ffhyBTrAQSo~ai+!{SV8 zAcUd<)x#W_&W(PE`HM1ms|MTy$`B(nOcASEgTSbCJ7%(_ zIUb%z`kh`nvR*Zr;V3H-snB1qfk+Au#vA$$A~AB=4Xr4c=q(ir=1X)=juf%|qz8hS zyjDF6ZmFzqv6yPL0A&oRc$XLRRz3kNHLF01`^zu&!A7IvB+t&iGg1~PF1T_g8n~<5 zP(zXW`KUZ)G}=(wv1oaR3JM_mNIVhgErD0puGYX2YW6Y=R6J0@E2b1zsM}bFyxsgK z=IG(@Fy0Y0P%4A@Ql5^Dlh2voql?I9}(R0<7POvc8Ff(rcDh8x;)IvyK&?A6LWL94Iaa< zS5DtA&qcF7%I+^LelWum`RH!$qx>=AoC7KV6^wk5D2Pc+G`l37{fCi!KPYL^;H8d- z4S){0^y@cut`onDqy5~ui|(*{+>SmBz&m$Gxgre}9o?F*JI@t`QXL~1G=K9<>zyK# zfW?!sU2II`$HA%ly{;iuXrp;Or?cB|M0vRDUCDM<~fM7aW^%wL&B zz)vEVeAN4 z4E`WeI6ZmS@X#b;qFG=77P$Y8$ugmAcm4(Ajz}6lI5hVGRc{feED5yQVzjWa%-@Q1 z@1Dmir1)#wd9s8sV?@;y905{wM~ z7OP+JRJ`Xnytfy}Xmj4sgYo98yavP6Yddj<&8ehYh4YQ`vx4a0Tg%NSL`;f3`U7YM7+u{If(VisTGrxc8 zKa5Ur!jVZ=gwdJ?CM3o5!S-D(Y&wuC~}>RvJ7!x6rw# z0+e|6{)r}s>SFfrzWSc^K}SO#R0>~xphQLo;HZGIn^)UbM|cc0s=UGjmaf0^GV?$! zEBP{{(iswaT<9d*;Lb@&S7_G|35)vf$VGjNyfX4z#h7{RFumTK9R!%-!X#lQ*XHF|Pme%MF(}J!9#h(d;UM8j}wr9eXKX9~0 zopr0eVJ5ts*sqVV8qmha3N0^N1|p_ioSdBr`9Nv~MPQU$qoN!c6}1K;0pL?>hQlnu zyI5&gx**}$##WS8iv_R)P%jj)V%Ri)mPMWv&QSF3?Q24CKpmHWzBQJ~eDHEvozOk0t%!53eQNgOL@HAw+*YaP7?Vku%?+wp;@B zkERwv#el7;yGEWIH{yaj_vD$3AYKbiHY$VeRKxySO4Kz+YPJ1$yL#XzJW<`DgUl^j^i|zUy+AM@pe|l=nLQL3s zWsblRH4`)gE=zy@JRUMav_jy%9Rc%}P*AuCL0GV?g`so+hi!G%rFll?=4DiT0ui|x z4nrxas-TvRB@qqJk)E_zXE=q5F4W8Qa3au=FZ31_WxLJ))Mp;K`lP;6^p*fB*g3uR zWO+Bk!DgfE-U8l)Yw!4Z@h4lfNF3okkHq7(--yH<+BKUpjM9qry$8p(NfAqw_=I@JK~vP~11c+gyK8jz$J&kG5rGF&@BU za&DME9EXH|N0wQp;pk(@6ekm8H06Cy%-?o5DY-m3Cr>8N!Pwe2GCKNlU_^Z|GG0Zu z1Mgu9UEJ4$TjAUn`QcAf+L>VUpgOE?8z*4p26BMo`V=xl7kys%IM>S9#H2z+_kgx> zN_*`l^}+6d(C2lR3Dt{S)Hf;4KT#uA2ojz7(HdYi)K46lTT}tL+#)>J$z4TBcoxs? zqCZq6>~C&Z7*O%7Do`;jhal{}l{Hr@c4nz8OSLS}8=G%`u5E8smHr>s4lip`iNJmJ zS4GJ9&q;mU&0BTOVd8CpHEb^bGhXm0+2f-#IQN2HC@Zuer$g+7F9V8e@8|3=V>xfz zd91Pa#^vpm-&P36uYJ8YaHcw$nx<8f-uV0AmscrGoCc(1jjNevg^4om!t*XGY<4mb2{CyH^#$ap>$zfP) zpGjOs$;J3uFO@90!{(%?RPKI?$_Xl6D3uME%BdL?WW~K+&f}Po<%+~Xka*Hvptn&~ zp?`i;R606T6pg3jl zdKjSu@W}F+TKv_-a(kgS#}C@&|DSj!@1|4j`wrVx4NXq#I-3V+om%U+VxPx98816G zee&tuW*ZoRZ-;F}OP-=G?~|*H0Oi@oejUIokAv zW^jnG>#OIC!hYMFxP@2Mv+R_6od(|?XJh)qf15Ll*LP+)2dt4lI%Hawzbk0Cy7d4r zOZ>LUL0(qy*2L0fO9fN23){{vRBr9Q|HUr5Poc=`jJU$T(anGqPOI`3w%I&d3^5VpsF-~e$imn{BqWg%Zubk z?x%mHp6#rqd4!SUIR9~E>pIQgiu&yj zloDC-nQ{A^6$S!^A{*Z>VaXNcUFy)9fTGSk7iLvCGWm`(Dj^ z`)S(eKl%DERv*`{PlpDF1?0z!=gIqS-L!f}1EE%W*}N1RnJ^8DPl zbn0N^*Qwsl-p@;4qw|Yq^KyX5fUOC zowjbBUzJpdoPH1f>nFUdqV8&SYxH)J#kDz_|6$zHGDO=9vhCN%uT@&=$5797vo5l7 zh|p0BFgSX<>Fc+Ojwc!ddm8>SH{ph|3epw^)5z`ho%>!pJ(k#>(lB%m4)HqB^Spmi<{A^x=seLa8@q;T=BMbF zK4uhF`fN2aU!S6&=FIJIxb4nux;FF2xx=rzNcP~hZBp{Wi6;Tp306+7a1bqQqBZMMYzIXxUapqo6|yhZzPz8+CMxU+2uL#as_>?+Xsbgjnil#CNh>Wl_-6 zM&Hy+Uz2vh9qWI6SnFK2<2`jjd5)~znG=s?M|*Bo)uyFNDGOXIj1MnW&&=Hcb5vWY zR=g`wCgFm{T@oH8h+RJ zr|tV^;mZFmtS1q}wWO`)!qK)Sw+x0tjd@Y!XD5Yi|F}jrutH72UDsW_PBdnt-_3o| zA{kLpX=`3^Is~Ie!*$(q z?inlAdwsv4x7vT6eOD0{;<_}4(O z|FETe&b`n#Dn4^giNAXkeKqF=;)_FKbJigpv0iS@;>O>1wal@K^n+O5Irf>p@L&DG ztkd7GTm5|#G!B1{#LOuCJrW>}85hHQbjs(4jze#2uB-%EeC z`mdkTA0#(eHvCr>Iu!Sl_NF)g$zx&Ql6mvPe1G`7Q2(vV?jMA+?al5BU%>BheN|Tz zvV9$WeN8V1*o=YHh)Gjdca%ZJTCbfvXSMyq!LaGaS0$}Djw*;4l=G~C>2ABchr1Ax zx}kxPIvRx@O*oQ1S*y170PoK1c(8Kf#e%cLZAc}<89PC-`ClK8<7<*u?51v;jVOV| zAYu3HmZXIYVAH7ijZ}_=1P5Co&A!9@yT1;>A3XTJ!%-+Dd-ETELMivN!%dJ>(nt2_ z^iz&o<&ubwdp7UQAE{7WaT#Yd@ekGka9w-dXzuB_>_YGR@Yx}N*Pk348kz*)zV-fV zACnh9a{NkSuT}wf_F%cff*ul(LOIav;PFhiPH#QbOtu#*CJreJ>9&zw^RyfJ0H)zH2Rk2j~OHvEld+AN+b`d8oot>z2`|cjc zb*zy1eEQK#2$!(rW3@b!TUkm1OkX;Bx4e)-R9+W*qLzv~xU_TMjd( zbH8fgXP!ZZ*m&^2n7_GCng%Lt&artqwGZJlbeDo4cQcT+7qt`(O-;|!-^V9qW@hF# zl$XD1`Zizi6-NHU+5E;OW?sgJukd~zzWjBpFBWt2&H!j)ZUIf0{zH=>fsvN*!qB;$ zrMX^=epU1K_RQcxf4FxsS(sq(KFxAJ&{M7l2*aCtY%+@`8^RRZXlIw{?QxQvq!6-0y*~Ve#nTX zeZwx)U20wb{n| zKg~T8*STW~w-&?(|N6xe?(yXY>e9&ruzB^NkRjc({k^~%`+)qi0>2NL0{ZH2pZNVq z{yh?ZkHp^*uu4S`X9 zz-f+$b2#-L&*EQNHcX5TK%rt>AA-LGM*+Dx5FoaJWlV^(47i`_%NgOn!RZVyHCw*C zW)kY_u8jlXBh84q52y#kRT;Bs%!FC7ju@}s7Ool4@?VOdy?<)@|NPSxckcH>=DkpN zvPyT>tJ8gCK&Xu3kC1kEA>S{@QNL8S?IpJNL+z{Uj|{f21qT3v6mxE z(z|MWOw0c1?gdb#-VC~ULn?nZR6N>9r3CCbM>+?B))yfZmo%O8tUN(jT&Bkl&>878%t0t))e0e#M!x5Lf*K*W{? zWwg}7%?~QWrZr;FGAbp>W+1D025dMFIkm}K!vSL2rBF_?fA@fZc0kSUk$K-mV&>QW z&+fc01$YQJ7b@uZLN0l`6I&>g%j0k|VqcEa=I{IF<4OVYATMe$Et^U5IWeO#*puH$ z5aAsYfPF;sM88L_<`m#%Y-wAPt(QH#-D&m-^#p7|$ga*vA%b?DWS{QksnMqVi- ziH{wSPev)kJX172zmzTfQHp|H)o{=T7B zMqZu_(9KLjske>fO}i4qA!TDlRh2&=`J3qlaswgx)$~C-f@(vfh`@6Okcc4KI|KxT zq2EujE2#kMOm?anLdJt-n{#8rab{dMn@X5z*#mAl(}m~Z%Bd|-zi3FZ21INCoi{`X zGV8PR)%oM$voFr!#yHesu1$|jPZ7p)eoih_c4$)9Xjv;tL*$5LhnNmw8i@g~#z>d9g%GJv<^qi6HH<5d>XGf2$8A1SWw^5KJP z3b(1J8LQ?_4JEgst%Q@0^jJGdRRu{o51;Y=d}f9|@R&rSh9U@WUuBT)1Hg#>c)1(V zqkyyvu+Weivm!nVkh$d*(#sMz7Q3`=izp=8z?ccdn;mKXL(@th7p=zI&(|F|hg4SwiiXvOx%??m2oFFhh*;~`SPMjfVr z_i_q(DG*_bupE7q{ppEJ&kyCZB#@6|tEQo~e z0uG?2N~E*}NiOMi0nY3946$3Opp1VhY<=Z%NF)Dk2>!*{#%!7+7^$e37y+VofY;v-IgZ!sk_W>2 zTnrB#vqyuM)~m_~<&j*QOoRPvBROpn$!8SOZVB-e1K6I1mzgGM8U(o>o1)YWkkZnd zHYrVw*@H4O3M_08aLUpxX|(fd0S`QMrxz*RV$yd9fP8m>6j1W)_F%BP2J}+UJ73e!o(N@kr?hP!K+-G~vxiX#v9yScyvZJ=Wz}mm!>^BkT`*SI(&sgD#4D8IZfb^IZ*xrmAhW-Dp8knlM}-rCxFia9+hd5b-$OOpunA90Xlm3VQr{s0OQWnLc&Bq{$8QB zYP6JKCw1BFsE`Qr19F@+(TL?B4O(z_gZ5NRmuPNB%be|fSfbom>w1pX?PGEi~eZ*SDw3E@FiXh`?v-M!DK9OtfI8xC4so}~22CmLn!V;3>sWeNw zShQz@w;`-w!|xRVrUyP-vbIYmkvxxY$XO^gu7N6^isnzpFc@QlmrbAD;uZ1~?v#@aWsz0=B z&h3g=y~VnuZlAJ;0a>GdhK>aija&6=+}k9Yf=H_pwHRzP;9?bnr{UQ>HZ$1yU({6V zmXErC0CYFEp|LSnyUo1fU4`6RvUOT$Z=mGp zjonEyCM`qTBvvXgkJ%&gVU`=U`*l(8kui&H8@;|?iI!UDT)c_&Hd=%E&d_wTenrjP z$oJfn%RO^XhonG>Nt81VLE$9HI$g>fz<;J=4+=D-=TeBeXEa{AfL2T1Po*-WLsts{8W~=t37#(AWOx_wp4mmG9t(btKhYBeCcXM~R!eapX+JTY%D`V*5H{_CJ~_=E z^SGaUvViw812s`5b&T7P{6Ls)&sQJ1`eG0TA}c=T2t5033+flcboRM%r{@5wX=e4` zu^)QZn(cUpa0pX#UEjU*Bo`?y)V&9J!8TV5;sU zruN|F%{Jk9G3tPBE*ec!RB1xUC-eF({S?hI-(79vNduJs0jclxB_r5*I_}A9wa;}N zeH8h&tKFjhktnywxlWU*=C~Ra4lS71ww`=mT{~Qd+UxT>iL)-fO0kLVcpdeIlTR_O zig&>V;Z%2We0@129x)pneL= z@?rY>Nw|qfif{q6n({j16WBfcP(?cHqX}9lo1;NnkZ)vWrWpAf5f#BCN=VveJ3v&l zpFjCSg_&971Fn^LpR0pY><9l!rZC`qUorv>5<`>e=(Fh>-_hh^O}Aio?9}WF%?Q{l)W~9KHu0C; zPfAEg$jyP_Ij%Vy9><1%b8zP>^m0)njVngsVl}NR)K`ZcV1WBeBv_)E!jJE43q>uj z7S9$j5KIIMViS|#8MG)Sq zw7*cvO;|blKp}`x#F<9>K1o1!@yM%9a$wjP1Q6=qM};hQt5xWj(QT17s1{?tC~S+SH`WxEbk7X$?MIHg3s~=kx z4Z#DEd8i{0T6~WB`S5{*M|lih>yoWZ#~f+{D|;nnqaTAJMqDhyE}eE@k_ihRcA{C1 zjx>{bqZ4g15!iBMhdYvKd zO88-f7INFSzlWqPT?pCO0YuK5iTBOk4yh$6Le3QMw25C10Z%4@CEiDrIwpMRtW2ZR zpWQSmMDLUzVW}iM)Y?18_xNu5qzR&P8#=qfE3d%A*hkoZcN)moz*H5(FKF zw?;>i=W(ZOX8122l+7HfSthy*OwE8*Uf&Zpc^JK6j;oU@Y^_OhPNzwgaoTZp&&S!I zNLwSXQ8m&Br<)f&Y!-%|e|8gD4)mM4-3K*I$72!qKg)k&$pIMnZ!iS?KbScMFPCDE z^k4tEcrRct&!I)KzHDz1i3m>vZw!p-v7|3%ej{lgGG^=u==!%w-GvV~)EHcEds{ib zX#0-EM$0p#oe%Ur0eCj0Akxyv!pwZ|{>(uag%@7x@`dyM0MQ6xv@E82A!Gg&%Yr4}&ThxxvnIT^F~;8!(kUd^~WTuT$H&AUMu z(YV?qRM*eQ0uWSHDkRm)6*?yKeHmto29JG`tC;HP$G;0%knZBS6Yz0AAZa5l&4_`J zJK?uL*l@4?K`GO{;w(+Msd-!3a()vC8{Bki+Q9=u^DN;XznL*srV-eEu-s!}XHD(i zyUv)UOSDIJi4G1MUf-BH0|ze6-<8?o@x76S(ZRNB(KeToqXfOK8U&>oSo8PsJ&B&u zz|HkP>GXCq?TMbn1NXCJ6RexcZ3)3l2xRZc%Zp*kMksXn9T*i51!q0B{PQNKwK8CV z_X^D{ORg_(X|B}qyWnQ!K4xMdI$YKNG(Oz4?g5;NlNur041UlAOg#!@1nc?2cWtdF z>l-v#M--F^3O!<4<~0;j!BQ(OPYX%0JE==3aru6wn#K?w-W(;Jtnokhy8JjZ@H+X( z$j*!d*R2YvOHw*KT{QSN{Z||bK#+8$elY7cHr z!9B4=wnpl5!H%JYY`{29rnmhsR>dCjDkr05YoD?ryj{~aYGkfj=hyqztk$Ie{=Mxo zdC7v{tfbU5TXx_nt`_1sQ|=}QEXCYw+_R73R|88Mm z`PsW=Gvovz;CX5#18FU++crp!V`ZWVkXp5zQklYwWK z+_tOfI2u3}HN5}YS#EJbB4Sd)^{epNwG!X*-_m-b5=%RJ5(xT#AZW|lnugQ^vsmf; z)aBv2UuA}&vGY}Q^s1bOK^J4W_QLP*r6PdWhNOLoM-|iqlt$Yqtp-?C7$qSezLTGM zJJ;*}ACiJ_*qc9Up5E`JME3g@*X)MYWD>H=G~tX=TtV9wlIyaEE2mApI`>J9<{^(>42U-aG?9*_T~O zn^{;6FUhkv$AX(k+KfK7Ugj05Lhe2v>`&6Rjeb@kK7DQ0LHRK&xZ!z zeLXvs&ei?lubJ!FRr{djH_M()rAFq<8)K@?s_IoDbD|uJ$CM=Rn!o(K>5G~D?oA1n zbp!ga$*2EmL3u5^uv8cqHr()L7V^a|2*U~~Tb7WNwVo6diC)VVr@+?P)h48#!U)uL z8@=p4`<`R7Jdc|FRx2v~Rfe~j3IDU9jHHaWzumWZw*LJLXWl+;*#Kn=QimCvLNO>q zyGOs==;a(d5p!h4hqLxo_Z$=%)68PfKMHp=Y6M#vpq=XI06nh)+ifj+~B-=WGQ zK@K3YSXg=#)IrEps&91mDl+dGRh90eE&FP->Z^w`%4Q<~r5HR4wq-xl(IY)z4J;KFpUGZ+lWFxw7Czf!>WMKzWZJN%sCQjB?a4U$dq7Rk`ZUl2VzXyK(Lh2Xteg7+3OW&-20I>pZHdP(%uCd-EV0=gp zC#|8&%-I}~nq9=~JCV!=7x97VND`y4`LmPY6V_Pp59_+9|2IrB3=DgAP$%rMVQ3MN zCImuK$hYMDDDdMx-vgSO_(0dk1F5(xb=ve@Z*U0paqPDKvwT2vvY1=Sw|? zm7KfLd?6Q2uaQ!L=W$bz6oW0ZYc#U$%m9YY9h(CCe4u{8`{`Ex$>z?Idu8Yfyf~=t zZp%6=g=*6S?q+(&s=JaG>K6v1-B=t*OZ6)Hfl0%>Z*G>&@HT-_ein7yQrmZ;Heyn} zDCK}JyYD7-AOGnm0r$JZQ29S`h&PZyx-sd_VQ}E%at|I}L}iD8Ys+Vz*I~Wv;pSh& zD*PA~;!ssc6id?x@;qvB+g*2uN|9Butx1}2k6CHZ@Q)=U>1qd(r+-$OJ#OtLZ>-s>yz{bL5omqt|oRXG3wyl-uD&XaOOe+Mt^?1f>_G z)o-tKrwqzy+sp%Ur3=pR?nX9g2Qq@TI=XzDRDZc@g=mcX^^)+0lNIIlgQ3Yyx+|XS zQToMpUAmS#R_NM$8O==Yu&pX9XLtF}9eSHI7GAmJKJhwehlP$*C~NsMvq?YwL;FVB z+^5>a5Y}dlk5Pw|e&pGawSXD}Pi^5W2XC9eC8fiRF45tZ z@TjhoEJN$b?ED^8ECINZQc9AaB@6ln`4z-~)29)9A=I@}y2I3hFIwrJ?9=6+u6%P} zNS@{;bq`gNlPmDc{gJi2)*C4l>W9&v$#gZ(PS#AQ>bG|vu^;N9ZXT%C4qPsq-CZPW z;V`y#-x`ivV3%swvCmo4+Bi*^C$YN_oIb-xV~v&b!^Kh(;Lw2bHzK4R;dyj8^5}*N z&;ixq9@2enwSabk+!K>}s!-cg3gStp zt4-`5ap@KGWSAIE zwVEwx>0hy=CF8GqJJiL!R7Fh75A6dJBT-2S}T{cGf(H9`1Ei#dVAX`M;~13 z2d^;_o=HdHs)eju7?v|r6yIhYl)Eh_EZ>(5-x;tUj2YARKlLG^8E zpNwt+12d+-4lWC9$S#P=Nkcc9nX%te^q#ioaqhH*S($W&cnb5L&)|?Esq}R9DTYf@ zFQbK_GxOEsS<=4)08Efsrn=R++v8A!%U(6pYjNgAi3@rfJ|Ng2h#9`pRpVq09x{Vr zNnc-V^=Gy5k0rC#d*;aevM5}Nq51QUXHfB*^I2^W<7lOgF6<=NLcNx~`8f{sIBXts zz6(wlvg@gOzyA#P!r}?luu#Jk!n1hu&pMVf=kz!n8+YXb5~8cwd3hwCQ0oGfVM31J zG2M{88GY(0eHGnv60fYDO=-;>x#e?_?Mcp5bfW4BU!Q{N!rfs|#jd;LI8Hu3~Hg83nx72XcYpXeAoO5)gS%d$4~B&PFTcpeSOd+n_SS z6)I8~sEeLHF=s#^0NTEszs9TzAOvchLsDK=C|0zP3s0qJ#?a*?v@@W9hCODu&CzB7 znINJNT{$^35oXW#<{$SEn6t#&ub;J|w$0F|VJ9hfK>5wcxK7wuJpf?>!6r!f#sCR; z#~u^ikYb_ijmY;d4Ci1tcr5ujOgM=ZkGvKMr4f*WbZn6P)dBA#Vs`T z9r%T81Ok<=<}1$NO=fQ@|BK%b^1&TLX0(_#AF4UXj20?UU;0xhNytyV+n!AsMd(tN z;*NtTSxc$%voW*wYj(H|2Y>#pE8P}P)rw5F23wGk+Ei@-sA$Ytb5Ay%nQ8$qII%_D zw{BCuO7Q7-(4bQ% zG;I2T&4s%gAhXjS<31Kc+ak9__no~yvq`m?Rln3p$V-3ys91JRM#) zV^`W^R8Bz?%@H>ej++-T*5bIX&Ubc2_j^eN3ibQ}6xLT$7yAB!m`XDd*Res8Hl%)3 zDJbRbFjnC_)Wa_LzBxkoQNopglCz!k4t*;lj%27}9iv_M6a&58idUd3%I>=#R2}Ff zrpd@rVI@$3)*7vDTB!{bq#L!|U0G*(ff#Tku`?ol5V36eOsCx0t9u;2dBd^+!ju%5 zF_hf)YvYYk*CYdaPNH^rba340(nJK0n*iBB9Q-f~{oUBMgSu=9xc3X7%Z5p_1IGau zkC6R>kQo|cv}~%yn;0O33bhvVG)ttZ{Ls#f;|M!92}P|MV8qCadziHs>&fefkBp+F zqBs9{K;NAn`E5$#WT5sV zN&2b-@Q8%Kc`_N?$?>!9f9^HY%0L;NP#=WwpqLZpfq~%mcfZ**-OLeBPfHlEe@e~_ z!D6WOOO1|mR6>GDb{QHS6jIioo&JwkO_(ohtC_)Zd~AC3UYK&+$xau?=!8aq*m zM}A#|Y@*TysmS^@ubY=eZPH*r#)&i4hN6KvYBx{ogQtF-={ks5)kpsn zkiJ6GT2WC#QHf(7>^i&s!FEkVKHLBYP-mU9f8385Lp_~(_k`M%%MT3@cup;+Q+4Q$ zIqRe$a8G0_$i+Jdp~&N$D@@+*XMvi@7TqZWLfwq{wfe@mS!w$^Z6ARc=1%J5MYW4& zY5+&@GFiwuWayGuXEU}u&+D^}?9B6f2pA!mi>ibRc7+5_EL;-$9nD`LeVMAc7-&c6 zOF_(YV(s(Rm55(=7V^*m{^@!_5Zzw9`_{ zccLR=%81?aF+H7I+#F;rb|;w|Jg$Nf(-Um~8oFVHS2t}g7N-40=>ChJQ9`-cZutIe z?Eio9YYcY(CYuUH*Aa)6~8gbJ3SWUinWDYq^itOH4yjZnJ(5sGvXQezoZJ{|r_j`uDV`?U~A z?yzXQ%;fM9^4qJQ-Zw`h%AuYmIqkAY8&Fw^fhj02jNY9$R&s2W9)}ZQ@=y!wWT9%x zO0ve{SkR+{>d-zza$cs~hkGUK%2Kwh7|CH?b`1Am{A<^_U;gX+fBj4Lvm**|-<^N) z%h%g|^h}JZbU#(u^)I>8e0RoNzs~pi$I;JKc$fZT@#QT_t2{oF-?D1efMnRq$;q!$ z^LW4c;hbN$YgukmTFO{%pIc*F=18GB)5~P2^{;j3QI>qOC2RQU)2D|zIOZ~!%@)2~ zPV>jb+W(Qfb#TR@4lb2EfP*Ma86@dwMYT|D_)*ZBUM6aKjCH ztM^`fg1uw?k}ZXVn4xMob*H)veaxKB`@Id>ZP!lv6cXHu_P)knIRW44N{@XX@l2It=5EJ2;qn9q9a`mu2Ad=F&zZ}Rfr@?$Z9Y*-AdmQz>ee|KbbdQ@O_11 z=BfC8P`%ofp7!Qw;m`FxK$Zyfqnw^wzBTpw*wJ-JmX7Q$b#^IO!sXqj9HUy+%^Vmg7IH2~0i@tyFq51dN5->9^Wv z=l8&cpCK+ZFMt_~OYI1H3O{#KQy)-knc%1j8&}4vxyHy{;umYVs!WYhcmB>{W!ksk z0uX@KS`*mR0#I?Fl53}O2ktZKmgy^KB^Ob-AfOK}VF2(L))yXSxdWA7rmTiJ*htlH&=@X= zBI?BlIGnmW*6V7nz^A=Y8fOc^KIigm`@YZDa)b@fGp}|%fgmQUKJPVOU+kf;=U&G{ zdrzj1B=9e0y%3jgXB2i;CC~-iwELXYbVjmjqMS#1>7}APkVgQ4(YKe}LxF`+`<`P9 z%I7wUwc%QC0=!8OaEzn5@M;N5V2z|;D^Ujkk!w6PL6YKV%SCVz0!^@0-~rFFnt~{~ zrJ5XvrF(8(WGiBh{n4z~lbbIX62fY1Zy1QY8W(o}!PpIG2{wle)V@7?#wYg!v_g(% zQ?$LP(C2IA(RL2Hzjd3wNlO4&1CSZB-cQA4n;fXLE4a-lMFTpXaG# z?dwm(H#tPlPoM`M zrgR{b3uI{R7WGXcCu9huUKXY7S&CLLj9XkvGS!@}< zDOi_w;PYKA&p%vzL3?O--<&(t8FrvMWR}su?U#El}7RW@Upq@UQez{Rh-K|?d307d7@AQ zr7C~OqAt;P%xr=};(?+|2jD*6YuLsyqwehxsHwWIljCGIJW8;~8p-O^8#h)_aU;a9 z(+#GNEp{xd|AgC60?M~mkYY8hkU+z=QurMjxqav_IiH*N{zLWacrQ^$;#_3(^10-i z{$ca=%)`#U4uNqAdL0YeAIy1Ysi}iKtc>!n(l0bkEziibYB>MZ>i+bM9J7Q*8Pp>P zU z3$HMjST$AIO^_xaC%S)sM zl~{M&R?vUHS+4+o1c_|(w_!xMr|C=pcSa#01|PtZJwf(jGQMvD2B`!_xURL9^4N;i z97ZHol<%h22s9#^$;X2gR9^JRjtOV&(FI-H?S9mo)-v15!dR`jn2M08ecsKR<1|!s zefQZqks2iV{YcfjC8Zt(oq-IrIAn*Emt4C~8kw4!`t=ZPiMj~;7dyyXQJhbN&RfMWlv%bq%4TS790waqQl!i-5prr@%lniY%)Y;PRV1SlY z;TlIMHB&DG=s99nMLZxtdcBl`H^}$+gzOK2fC!{tnhl}nVuBI_5G~Xb&ap*a&r!9_ z?4^E})Oj6&CZkIM+i>^O4%-J=@K?y;ShhuP7YRoDM{Gt-63f~`cPNK&!vs1z$IX|u z;RMHHb>GRt`}C!t6NG~G5h^usB-jN6LS-ZQvj$taf{96&=p=f1#J$OgH`mU#E2TiG z8Mws~iN?1+Xw;F{`5YMYA$FTbtN=mMbNle&+g}x~MO{%r4R9BXIGRCP9&|JqYU&-Q zCG@e$OSpOu@2u@3l>6I)m4sPq18|Dea#E0v8|-b6Af#_2Tz|uNZ(jCXhF0~5k6uke zomB&+H1u5$-$pA0>SW=$x!U%}vkQ322nE=&^uw?5r=Iwv@=tV1p%QdDzmuXns2$EM3A9umK)W+9*WJr`|VUESXqACAMHY(!OoQ$T!J zT$5_ZO!`lU^1+?Mi*$v%N1oknot&(4P7~D5h)+kRk9vyx>ca|+V)%}|+|GHSZ&kIE z;IIsk`P_uGF73n+4!OHuj`LZ*iS4w~h7YT_qv0x)y9va+Ov28T@@ptC92K7}M9rR@ z@7SKGd8UO}%7F~bcWGZz@CF)|&CH?5mSEuA$oLKRHObi8vZ}w51lG=OFK7Bk{CvDn zO&9`R^g3@D#}vIOD2zB&%XO%VC6#Q&5mN3lOX##}m@j+ys)F5$z#gxg&pX9L7Y&HI zaab;oI+dZ8tc9OunXM}~GvenM3OpdCv)1!*Og@i;vGK*Wh-(sTIGipn@f(=>94GM5 zR{S`KxWyS}>?%F{J}WwcbFP8~3NPX$0 z=gM$u0U*Z_r(OTtHrA?vI^?wYtF;2#=!hsj_0e5Sy5a?BgFaCqGpVeDVpQMFUpQ{n z=AX9y+;>EPon)ts(Fdm#kDdZxC!32Rx-9$UD7-*w4a187oyZ5}<>g0yYUVhL%B^^r z%p`C*s5IFR#BsV-^alU^^8N9jW-p8o=(^xYg|LC=Coi?2O6HC>EO{nBbEb=cao4g3 z3j|5JQAOXOsLX}GW>I>C5Ru!gnx9~w-yzS|k4JvS5ln#o_FI?I9oaNE;+eHk|M5q; z9EOgo>IPmJ?OtvZuUBS>#r0By5Hc{tkI{6o2ZeSN4Q{V#TI8f@-5Cc?)%We|g0|RO zK6Q)Va=iW6BQ#EHmT21?#?56hM!|;j$6-Sv7i{GzW|%_j-^z1(YC9SObdf?O>sy6e2Yr*sMR(fH&CLgOM_phjnw=Q8`E+C6Y}vYiZhEMCe}|1@ub z`H0rM1>7b0n|f{py25j}aa`^C>v-5LWL%3ODW6uu&CGxa2}nkM=gz|cj%RynscM#) z1g>C97l?g5Z(OiCYrE??CP`&D*|}xHkuHGJSPI(Yu4%RafX+K(Z7 zSf8Ymwi}h(F^4cY4FEC-Dd8xg>@uU9JnqRC@cY>XF)Hf;41nCix)rx0dnU zXXm$oHDkey<+v%EEwHZ&z5rWq17!jhADae~EvpY*q}t9AQXpe@q6;%JQwMd;aY3tvCRAj zm@@8x`Wy!2x|0ZuLBN;1JGUUUXhucU{#0z_?79J4$qJq)B@HVwESlgcZPd=%jx^Q} zl=EY>DE9%m0H+puS(owDs_iU}CqA?(J`0;cOt_OVtqOSLWfU7_ZVY5lWd)fs73m-$ z`O#EH%U$g(n?mqRCqwP(jIt$VIL=cQgRL+9F{oY7sB;t1CW>1!kHo1=rQY?9SPacl zlIDMI^J>IfXWqeA&-J2Qi>)cehh_xjcm4`!`oMQ1;}6Mzj^UHMXR!$0n4EVaji`A zlaOswj^vo`dLa}7aeTUsi$2J_(!W&lws8K$*oSlb;S!0tapVhkxK3u)y6gSHuA9)m zaI)@UE2k@SJzeFtmOZFH?0JAq%Cy7m_12!3RUiMX^51W1u|%ejy1TEI^F@v|=wWvF zu2loTi7W$-tQ?nVf|q~S-iu4037zD4I6ck(H_v-eHLwe2M;WcO&Ye5=0NQpeDJIZs ztP+kLiMHWr^}~x)#W<=WTzs0U5tx36^U67L99-@C*a6!96yvzXI?bxUuDXY8*9ns` z?j!#vez)S)E-Z1?Tx7)TKlS4{_RvD(n_BjuN(;#Ek2*r|^rCYd19tt@^7VW#h_#tv zTz{5F`Wkv`nGQXG6{D^*l--ghq*xJ?cyMOk9I3-y0?Kc#8Zhsa*ZZgMM9qv6uFrdc z{ub2D1M8@v*OB99+IiROB(d~f|D%4`Wy%=$ga09An~j1HWujOjv+(2DgTmyrqvkPK zx}mJPOiK399#M$=-<^t&=Ev*UcZIg3N}w~8N~E;Hv5GdzpKQhSX2pvsL>j_3_ z!=N(uu-L2ftIKc6T^c^mX}<&x!55-Lr5OCb6quZz_+uP*Z*u~>-y027>OS^;g1jQM zGS-06v{4tacJJZ?9M0k{!y~Yaq%`pH%l@TvJ%`uE#bqb*KUi6?IAHBBHwTEPDV z?buVW-pK%>xUMNBhrJCcM7tCFIqpAR+f%ScL`0+owbuP^?;fr^1q@tEo9toJ0WHi# zsaqWL;^`-pl2YNN-}4D zky@8cOifEXKAA^Zaw|*ow6?Z*ho|&#-bHhi*m=5MuvHV6>2&pUcu&c6BLdT!tkx$Qd;Y||Hobtc27-k zZ7ERbkF%vj<~g2~xk!gVFOS{8(BaJsSYs5SHY-WK{j=;d4%4*D^Ke>Amb~BEHR|rh z+9SOU1L&A_CcBlLEQHj;8hMBoP{Argb1%nGiATei$U&zxd0)(1DwM#}WxaZbU{-Ci z8(ISUlXY_7AHmt25qK|%(CeqhHQ#t zFZjr^n$=QGT(dtJMD`S+6Gd>HbOVsP`U?QuCsI zTO_a~7#xAf^;_f!&ALs^hzJ|Huh9`Mli;1bQf&)HWs$s){>i5|_OJvP95>2$q$;3` zh(8o(04!gdB8k*y$VEzm7uCR1D{BeV%`Ln}JqszTfqNRQB@j&5WsZp0!{_D=EK+>{ zBQgOl@ME9^X;BnYYj&nwUcdkwGBLAZ_^3)sNg_gCm&1#ps}V9e6|7Cn7raztdrslq z_`l=aZ&iMam&z29r^+%z9-zMtCHJhtHniL%{f47LWpF&e@tbBNS=fLe62bg;qsh+SG`gRCjUod4V($yV{y?QkeQ8=8i?Z zqE~zI4*Q`%A&7{qHmZrkdU$jy!{+fzIW-4pnmd+SvVkv1H?&djJJ_ot<^=#tiEuUw z80i&6drmP199!r0*R)R4p^>?blBN3ac}G#$zX~owr2F)g;rfj)QbsDsn_j9&Y1TQYyHtA)^$NI?UaLYD*F#FTD)P9W8@A=p|ZlNKGC4NicP8^Vc5^PigE%qOEaNSiGSe;xcfS0^>MlE4L^3fL_C^0f zpA`CkWup1w049+*xysb{jjauXHOE@bd8d5Guz{NvBA;nZ;uJW`eqSxe7Jp(t=%%9g zC3TQvZ(~HeV9z+(ZBmiT4O_;MnD^kmjv$s4laQ6Cg~5bO2vs|>ORd>qJrUa(U0>Ky zV>agkPLpspuFgsuHNK>r;W`GjcN5BLzu6Psh4AzsEC?1NNu{8}Q-(6ukz%rZoJX9&5i5|vf|D75;RU=K zHicerW2@u0n{|jKBY@nVe0ic~2Kj0|d>kIF2X~}#u`MvzCB#SvG#O6+A75;4hTl+1 zH(O;biQ1?bf|vn(eYH&t%eox7ys&G$bMM2$SPf6fU}uzbUB?6ZB4Flu!Pj8p7YMuo zQb9v=7%_G7rpNQxbWtL*pbMLpbXt;9gf!WcyT$z+d;Xy?__%rsc3{8N17|?yoZ9$e zeNtPm$15z@6b(iP`qRO!@WbUa;Nc$xR645Yk%qpIAS40Jvb; zH`lBwgik#{Df(>@qX(RT--wpX$doA@>(^$6OF=iuWi4$+DAsBgCUYFVYcYzB9*I@h zOkALRQc9J938)LGQB97}I#_*b7=U0jX>`zuM;s(W}V?f=`xo#){*N@rs`7 zYRsO(0lJ8`zwFE(C%i~J_J2#}@1r{VA?xMq)84?DMi~PBAMziXycPVS8JyNTE$T!| z76KN;kH~l+gKv>65-nwF&goSfE5NR;FjaI+%K6-~N4=b5ZXmsLER$SQr_%EJ>W_m- zRRohL`qlkgF!}$@GhglE82ebUh5j1v;D;tumJ_sQeL6LY`tb2r!|hS|0ttsJY5trC z8*9>2FOt2hP)&I@ubU@+<+RT6a@uQ_G^swy($3+?|8rF(o)D-}{J+?H52&iobbXYJ z6P-*f5W9P47lTLzGjIv5ZZe5(QAU}A5^G+I6 z+9@qW>uNg9cO)3DAp^%Yl}yYrbOe-zPFW;O0WCBw*D&sLv*(d{CST9`88^Qz&OY{# zMQIo!FdMl}%0eRor&<9mQ6~f)-u`}x4pu!di8sN7plukG_;5sud#)^Pf)}|tRS_>O z;!%4WGvne1_uyIksRGI#3G1+5S<$ajYB_&Q!q?K4z`*R{8X;P+f^Nk@l1;297J~Yo z8E=05GHq?NqJ6tL>lZ_pYn_M%oCf?@1&>(4{YT**p7`yQvF=xiGKqg7Ooe>#5KTGEU;6~*t!ZrKPi)MWf zlV`2fxhR3+xC1u7<6wVR_>GM=KsOtMS>F+~WO^&Odq!f5!j2}uKBVKg1EDLCek^RR z<>_s0ZDh92mS&>zp@aMc<8u58J7f=IU-D48!vfZ3LpV-6nMEV0ccpEw$0HXoVd7c& zo89!gao8DRM`fXZ2AnVea5wCBB2mojPj9big;>$v!a`XtZP&kMd`tWeB90EM)lDpQ z2{X&ar`MWTz=-Q#>5&4friq0ccj?bO3y3G;D;_I_c?s$S*@eW)trs{Qj+}nh%hU@8 zGNi~8{o(45A?Vjd+6N8?X68L)mGNqC9$hd&SfI$b3!R;!eA`tts@ca%tgf^RJ=^eg zCdz9f5S&noKFsSS1z>@Ljnj;WzoUUjhQWp@4!~&Gs=l!&ccePEh-$_p(u5?bmko** z&1*;BthFI<{>K+eC6u(r7D==+`N%AmPFKO}-_AK!I?lNC^})la*1~{OiP8Hw9g&TL zcgd<%t88IC;d8}yJVBlgX%aTBle||cS=Qi;%rg11X8!PptZg(M&l-t(L+YBO`XEQ8 zw7F}gvx));@=3rAb`cxpv+9ExEXb+Rxe?)A>cfRLC|F z7a3SuvWrD&n|oxazk~%a693Yf>EPQ^_P*L@F!%=`t>RBY8R$Pnw&^iOp@fi$*BU7* z)U1Y*Aae6JkImdEgKwLrTb@Jkhq6*LNnIeNagE$l-6THwzrTi_Dwl&woVJ3o(?URD{B_Yq}n-7 zCr?RugiINT^+0+j@=sKGjXUXh+E8#)b$Nhk0gl%U;*1eb_bvNI-NH<6bsQHikykE0pP0)P|D*Eh^) z@PkI;yHP2iR_=kh%?}uX`9!A1K8!g#o~B@*7%1_EfRdlaIPza%1lK-|oA7eZIPkdh zK7WJ21SxtRwsID1#__6xly%*PJXukgu_)4o?49HGp$IC($1)OTjG^0q;Y1BFfiwX~ zUWjHcJI1B-1>h4_HbLNcr2`hh_lHBu-&u=M?xwT_SW*&cbx!7^2qan<3IcLC2QtPS z88fE?RilgF!;jezP8p;m3RuqJ*6lM^ro({;RM<&0AQ8ix+G661y_KX1H2?6@%yYtt z8@v386Gf~8QnR7k16BrWpMr3I`N(b;V2-F8j3Gzr`-j?^@4H``GDxwNo&)8eZiE0l za_PI%-rUTqbgVJ#iF)ZPAGsPy&lM05qWYX2 zQhCP$XZDq)sz#hMl9z0T7VOpYL3Z@c8r;M_1>Lu9U@vqe&%?W7=5jdy2FgxQgG;By z0&-6+5J}hv?97IrGYnnwpA0LcJQcvu+hN}rDLdz$pvWTO5Qay0u=WHaX0p-60ZlFv;&|?n)x#_k{mAs za63x;$WLm#j6cl|4?^+H_<2?Sr(?+fy#Dv!2lIcze*EbK_~kLC`sJYQm)7{w8tfLh z-uz{zObvzqSGPvS-%mr=(e}|i(MFONC0%x;c#+NW9c!m!vrq{Vfz@|&W8DY!{R&m30cEF_7G7RTO2Rv%)r!m6SL!AC_C;Z6uP?j%*~}pB`)SKgh74(=8>R9 zRkFP7CVB!9kF`Rk&kxAYJhYGcupAQp8#alrr+O;R{_5g0zWU%mgm&7u5XTmvkdJc6 z^0G6-1hWGB%!5ct^8EpC$o;=5*s#e;QN#@y0D+DW4X7xrZ^a=F82bQh{>_vu0TJ)z z>XQ$?^7WjtmEO|{(atk=q3r3S1u`Ls|D&R#H&LkrkxLI96$ZvebH_FV#9BZC+7Opi z6`zsBeIR0M!&x1PQzhYq3B5xYO;AzlUUZs2`%K7yQ$kg94z7(Lr(?yHmF0+^p(rZS z&tW(JKLP#Qy&Odj&4QHpscK^T?$-;(IC&}Lx{i>0Al-K&F6Z{k_nC9x6l*jg*qRT@EcaGrjloU+iZ8Cl7>jQ!<-{ z;L6}8>&hOm-@ijq%C&qeogQq7%}!WiLMr-;z998fI6M^M_~8$2$c{TtIXIG&kkM;I zbgRT5^+Q5h@`l`QfH-AR0h3B@)8Q@ej{`s9-u9j(VZbo*2t~1g*h_@&=uUziL;L$p zwo@ww{*-yyoApY75TsPmA{3^8OePAh1Q@*qeXNeeCnS?8SiIdJxGoT^9U3FuF7-Js zf+t@;aLC_fVQy)Ox*c!*g5_Htl%{cn{v)ED=_YAr3nB{@=T5=^rb-00u-Z(u<@%HZc;rgBl5`eHWVTC zM1_Pt)&Aw(?)DEXVG)sAah4-*Dv%Kq9v`MGsei?vQqS6O zO57y=O()oaI={{=vTMa7`SR`+$U4uPU7ugyxdf}s!CF#!=NGPm^dhnH#9)6$c6Pp^7Dox={$S6k^B%9`F~?p&~pTOzAwrPJ@!zDR_JdkL`mY zP6`FL@97i|YJiLp$V&i_AX898U`EiN8uwc{sQ?+HHYQp12t@-MB6?9)fjIfd7zhao z5km@@Ru{JN3sRi6sO_Yy_N27bi%zwm#DR2A&TTjri}a&X6bp8myJr&tg4vq!K(-$5 zIfFby;8ugZI7-$xQF;j=joinlffFI99XGJO@h_mj-<*WRuo1FBm>4GAz_M>6j<_vS zG+ihASn;{iqIIVX@TDqAD@e~$ddZ|XL=wl|9wdpk0zzB!W;osm`H2DGSAHQKhdqaq2kwGIg4Aa*WN#_q!A8|kxs@4` zls%d=wF#KY6xg3giwbuYSCSdA3pOFyYsKXp?(-XooJ|P_ZsM5|1_TS$dSF@!M}VWR z-)6lqqQelL|I;0w2w)%3$5};g@2{i(#em9uf!-ydZY40UX=MZYc(^)uM2q+%Bz2RS zRBuzrPxiaIIbF!ulJ@xA=bJ0b)cgFci`zlbw$&rd4K7t{-LT@p%cFqOb%#A4nMtWb%|0 z6ck88fe!d!kp}xTiA2q88$Gm*Kh{fE{=0RsQ7zo1y!OmrRq z zLd?<@i5IJBu0b*_{muDR1^uY}6%s}M{Q3Q3CXYeF6vPO{if@->n+>GM05O4p79e6B z%!YEnG07~tPjyTTeYYhl|LTUu7eh<6)dH67Ez%wwkkK!bR= zp*{g4hSZkHcY+y`_J@{0Jg{R3uzHJ8Na#vwtsWyO4FX!X4KR^SAyF#Y1F9s)3e(2rih9)JYiO5`n=sVKkD&4DimmIsj;iC$c!3LP3O#L zPk0Hkngvl8hEu@auE9Vj8|`hNGKhqaP|cbWjSfX$b3E68oXm(0K}6L@hDmDx&*aTb za~CUv$O_`E`i;%{2AIl1EQD{=L*o+?wo*J-&`&dKz%JDy!P*le(XwY~m14NG6n)At ze0A?A@uJhV=RmQ~;(%?n#@ETgRypZsAO@Z*fvwuo|8#Xj*HrSyy)SLRV8bi3|B?$< zBuq&{FiMMwZV6VQFri;4U5y90g`kQSDI$@OD^qtT2>#b-=%fDC5%dkm^(3c4t-8d= zB`7_GAvG_Bw>~sbHhy;AMw5}%VPOT{m`>{U780~47$5*^8*^n&yt-#mAcee< z2x|}7d&pIcAEJgZZ?2AsMZ^Q*I(%}Ec2jnE|rf=WxoyQ zC}OWgc?Kd;UVoN|tDE_}l|maSE+s@L<~}VKL`x!X9wK|R)@2W=74=}^q@<5}jeYWB zoKDXj*r3;j3?4-XRek(8<4v| z5ixA9t`@Ud1~A?`ZETnyf($Tv3Yj!6&*T6|*WHUnGWB?^N1ny-XzC{*cY-koWUegb zFCZQv>337zkd{$uXzENN`4Duk$nn)HoNEf0G@gQ0KBYcsKA=)}2uSx$>`GpTxTEv% zpC-sWo}b$lSB1AF2vZ%WiPmC% zY7}Rf0oLOtR;<*XQw`Rua&!5?zW1X9E^#jjo=Ti$IR-hRpsvS0%OO%B9^wFiq5)8v zKk6+;3>F=8tLIf!8N0fyiu;Bo)71X-z@`R(`K^#Xb1KpRda)}r9J_P{qGK`3W9lkT zX`|qn=G;MxyusR`BZMu3UdOlWU7aYB=xmU3`^0e1tEzCGsea`QCiI)#4#)>>@@J8z1Hfaih3jc11ot~0#yUj2G#Wf|3{ro< zI!W)|4R8v#97To%1|k+lVOKAU8GmL3v3u7^&%S%sx2P&%Gc7WFWi$?Ve&s;nRRNO} zD(^UkfzVczQ?HAxJUc2P?`K+rx_jVRcDv|4s{HrPi~vWJ;_TUxV0{bNQKHT*QJh*4 zAH-cJkR9jq?=P%Y&KZUC35nZVSPugPy$x$3E?-BN#{tQ&%l`+BvgRHXFMQ%*+nz6y z&TE7OEY&~?`zdbX3z#|~yX)*EETH&xfVV{u+$%9iJPYNd5>O6oU+(`iAezRUbD)&VuWS=o#`qOJRPLo_qDqjdK*FAAkMh$>(*f!EPEN8fG>WIB%10$zB6i<9 z`pD+@g|bu*-`7=_?PwZHN?PnP69U2(L=iOc5U)KJaXyg(Dhp*@kVC&h%pK_a1Enmyy=j!8P%b6r|(G(E}Nm zO~@Qsg}Dy4QnU_@iGA38x0yb_PStYy8q@}}jG`udKIldDTQ)6OQLTkD*z`SI4R8W{^~#?81mAJMs8!3}%o=kIj>2)g2( zeGXH7dwpaRbZMSPdd+CO4rB#pyNGu-$Zl% zs#2M?(oVYv?aXU>^)8CA#eIN3ln|1|7_&turK+r(y0Q2_iT5HN+0rafsJbZ1sHv(s zdFh^Je!bv6Mc4l)z;FX?-e1*2fuSRTMcV_bqtv9bByb`eQObMYpL?hD+D#v|j9}F$_ zcGG3VbOr`B`>l%!*6c33p{kK&Inru}Hx#_B?*Fl?bbYUV-nP< z6wI$Zj;S#E?XtZ_^-G4jUtE4!`Tk1t9bRykag+?V^{U={VQ!cDrTA%Z>-O*1FYJpK z=3B<{i?b|v_A{T#V(4&edNgRSl-ucnP&Hmm$wg<=IVby~2VpvihW0APn>9j(S(@EZ z>V77_KL~5B&fY2$uwA$%yTScxO?HXEC*$Fi1+MSUQS@G0GmbS39 zX8IA)hRu&N>S|XB+N9OB)<+boEn$hM5zSiG|7uOFqm!eR;vsIqrE$;jGG>;WY|4K( z(@4E##^b^@)ouDFRK>fl0##ueE_b!7L(2E_2d1AiQsy3UcF32W<$Uv3*Syiz>Vtaz zF0=dgjfLqj*3rMKiKZB$lZRW*30AjiO7DLN_Rf7> zIYm+B*+BpsNDVn|4%hET_n7ftmJ~OzUK5sZTc={$WgO-D& zMYfCF)$12}*%ji}De{tObR@S^jzNPoGUX|zU}L3Np+mtmT-1&s6>6eK!IJ`%C+&e2RPDuk7#lwd{$ZTSi9;W0rU; zQ~&BDI*hH+VJIi2MhoxfkHy`$nSZ#WlQ<1sde4BS$GhjoKWOV2Yo%C@T(YiR*`}hG zW~&rieLW}6$8sLxy|Kbe_JT{Nm!o#bQY^OPLSNa9UZd1dq{*ZQbWZpE6p*sdQ-M!*qT25`>n7n0XY7PF+#bQDTojpLxNteKVeyl9>JhvPe zux?dx7M8V9-1jasyKt~qc6j~2dBnao z_j8_AG;D546)xg8Gvjy2ufo8o5Q0;2?#}Wa@@9AA z6_`&lZthzeubHM$p__GFN0}RoPuUSY|BMXf^Zn}orukwqoa(k@SuqUt*{*s(unoi*15Zn@bX-!X6lj3UGQwp*bY z&5ZK>w3^|w!Txt;8>ZEi4vw(<;{E)p2&MBMTm1$zlxO$rx?QemUmx?LrYu9H!raoV z2!HYBKIiPXA1hd9Z-?KxzIRC~)5QdNP&9~LC;l>_Zp)e-dpSFIMy-jl-}d~?-YbXi{`j|le)oOB_h;}v%i-5$mf5PB%vbUYt#8qL(J_> z`&Nx<-L-rz354GI3Cka!Zti!2Ra#}Gn)yFM_utN&$hQjv9!K= za`xi_&(-f$+E;q)9h@WEjSrd^v6dDGhJ|@nBp;SfcdE>9wX~8_44<~EZP(V^Kx^M} zg)3gI#pQ(uialhUC8aoH#hTv5N!E%6=y0V-W*co;rNtL05y77%Uah>zU$;*@Ww7&h zoMYa;^Rni0jJ{NPY9Y@Gc}>11ZI?H48g`1zQ!YZV2FMPYhukD}ps8VuE$eU74k&g?E9##CBoOZZODmKJ#MT(EST*S4z z^@j@k-Z(j^n&*GJAf3-&LF|sXX0!c%h4jM(SNr_5v)c`)>vXREjXWe2<><{!1SGiX zhnq4a@5RaF*`>AIUh$$#bUTAz;=G`wW=UMBW?RbfL4}m&(^_-*Rz!EX-l$EkuxTz1 zNM)*Y`UwR1QU}&{#4ehYKd1h1Pxu~lTfd;q(y9n8S1`gF?1`8oYS@sfBX;~r@@ioN z&AsNmb3H4RIb(0dSa5snRpwpa!)sw?W+7){zCPC1rbkn@Ke;85SzI#kt%=fW9+Y68 z6hi2;xw#K?VFowOs85kQ4g;Y0yX@?21Q=w;Xj=#S<-|e(1xLqzSG_T+mMFP)dy@iQ zW!7n1(A&#uYeVS-Owu`|+-7H=-TfjF4rgzE(94mOBYIvVik!}s2GAJ@hk}P7g!4@4 zU_W=>yvW#Ce&|xezO@+TEFU7I5bUa~vU1=uOD(3H4dn+l$)AXjm8lk57;jEaPQ+1R zOaD?r21kXF(bHFd24K4xxm04(B%$mIdL`F{>m!|Qi0wvFN+CFRcQJdPbPr9ODXSL3$xAmr9!WboJdzY6>4Hav*c9;TMybU#t*ZaD zm(n9IS-shHEEY=vszww8vvu*Pnt@{9yI)MkNG+O-NEKq$s#(3qQ-Xf`?Kh%F)!oDK#yCWS4xhvJv-a3) z5Z`~N{5bi{l65vtZ29B~PZXqhr=ofM%m7SVX4KVj0+$_0K zLtPqnt>HNEN0piy>g#_;VjhI0_IKH)8BzL_M3In;PaDvL-8JQ+;jF`j(W#McV}yrx z=&WH2luJ0YKc(#Ixv&w^xXHoMQ?x%bGn1&Y&ma|s6Tz%mvozrFPRw8zczw6gds1@X za7;NjJq9uLz}$`07GR%gF{1ro$PU4mMvz!dN#2i?$F8@y2_q+{k#puW8tnNQ3nH2) z8px*AUtFzTnprb0CP@}LSW-@W@p07lq$D8CdHJk3NxDd&BzKDpVBuJGuI|DKc zS8mo{51@?{MM$Us?#kYjaUki$ey1*iZa^wD7@ZGiaucxo2d?CPy5gTzOuv-pC?^6pVCM6KEtu4(gmLGl|0Gv z3{=*+NFFI@2&~F}g)o`z+XGPXkSH1<>>@ITP10l5UKspMq;+}#Og-+*S&1Yublq4N zoeRGsEbgHWn#9q`H9X?q?)qb0_()}VlZBOYwL*>?lrBt8o!TyE17-lF%<83&LSY5v zSvF`Il}{cy4EuV+q9EZp!ecSSoDRa+ZNqQYjvkdB{ovKb zOj*!7`ChM|PFP8}tG_3=6NnJJDa;uo&Wx`-!f_xO=FXqrY5NRig-{@C*2@_ehXnG$ zaNQ`Lr5D|nA|Ui+mN3@E%e zDwS1K^j&6B61V^FS-NLNM?q+f=o24Yd3a5br* z?u;FhhXb(>VtdJtbdDpT^_PPNJNZKdO($^*kbd8Mt@5ty0-cIp03j7lJIOKYS% zl3(t$`r@Cem!h-5;Z472CV>7jxdzz>Ds5)Bf=0ZRWv_MzlV8Q($W-FT-1 z5v8V<{e69F-nYkNHVPMY(53`JC})zn*W>;PJ){TMc48jVPyF?yq#ji-pf!p*j;Cal z3<0KLWl2HklO*suO{&& z?~h?!DC$5bt$_zMGLY5>lX4lzpNgdQOYX6}d-dNw!(m>|ZeeVqLDd}{9YglTSG=4* z49q!({2yC%i@ZeHwbQwC=b{SO>qP7?!ZeUaF|iv^%a$a7jx_TXs;N6;vt_e&Wp4K9 z75p|V+0Q(3X0k}Ldc}mr)yV*RJ3Aq+6pXzNrr2ibxQK{`>&#AwvQu4_=+|uLQ`t4A zxk*Ksbj>gc4f|iTvS|jj8TO>qoXZ^8?8zMM>>3qt`zY&X_dMyU@3Jw;zW$a;u+*6E zgaOL@q|VMxW9>AotwiIBcPd6PyI*|lAnc$CcSWlwbszah;k30#Ic^Pa(9b$`iws`V(j5OX;rz_;df)o-)0F!o zQp-4Q3DI(ylBZ9f@~s&o&_6^1H=iCoVs2)Z4|vXtntG6mbi2Z#9#PbR;)H2@`Ub*w z*kDx=?YF{a|4qn5#Np00)i^REFJ$)1%E~fs0<=?)WlOK`74~^5fu_4vXqb@2&YY}f zPfm$4Q=vZAtD;~K(J_6nExc4wDk42|n#ltXm*xC|{b3}-GDWipzujm1b{5$&h@_1{ zIetCO(xVy^$29k7g^5XULBqtv1Pi>QP`}&zb88zRdaHJ)yiGUHqBcXA>;TBVm|HGV zIiFN`&S~}_-ChLIpV47J(w{-uS+6_b+O=!LpYmjiDYz2SBYDI=vCdg*Kkb0Bxi>e( zKlNZgH`1b27&n6IViRFS>M-0yWD1Xvcf3uNS=9PU&JQktr#wRlnp09@S2Be9nV$4e zWyu;pjq#0wWBaxg#feSh;9p5}sC zL5&de-vo3aw(39O7SQI zH6ZB%3P;M13uB!rhT&is`V8ssk$aPx@(-@tHscfH(kY*f5AmOZ7TQVW%cU zLPALbxgAdiir1OuR#xEM$w!p#W1N7KHZJa3S;-bHhr`Kc=WHsblIT7givyd(Qw$9C z{w2FIFe{Cq%s*#j1zVxBhswSKW&Sxn#Aladz~-izv;Y;>8>H=qD#AYC6&ct&Y#_zoOmd@u4RoUwo2W~Q zLnXZ1=YR4I6xrA$<460x#_r+X_iQ-YjQPyr_4 zLxQtJD!2Y3ofkuHRluS2kVF-Rh$Hzr!n4gd&yy2AA%&o4_RMj6KZm%%WIgame3JJl z>!)l`;+DAdmb>g)7a_7HupOfa3`Za7vx~N>UiI`e#t11IAzETeiv3@B{-S^Pe%>ra zoz4#&D*n!r4_-Eo&fj96KI5aNH6?FsZi+}XU79JR$?+t3(kW_ul)FDP3!GTaUYED8 zjBN_c`R#7)wg8jVX%Ti)F{ha80rjQDt+w@sWrJwwe^~34?jsU3r^lY-k zsOUOjIIchl zdbaHMes{cg@8E3NM5DZu=Ve)Pj8q#g&KKrIwjv#OO>X8$gaub<7Kf~Ou`M%+e^pvX zfx27Y?rA6JhOx7zgG6ry@vP&QxUE)ol9QgG90q$QZ}|RIPql4>j*?-^N#WkTT*0%? zC!N;c-s@!0|5YR$Oc|YKP0xaK71;CZY3knf)v!8MjWs%z9A+|fp?eqy@bSz*;VhiV zGd*Te?t_vNzq{_6Sh)1d{xR+ou;tAfCu4KV;mU^Sij^zSxve?BwXdukvCA-v^?epv z>ZcKE5$&T8eKRaDmaE9t%iKbNzv*S02a+u}#}gox_bt>@=Xo^$R2eGA7_)5&8ZP-( zsaPYgDCDBoky9vFo+itUOB0^P92`Lfi=@V%;&C6|Gx%t2>aV54t_?dJJYFJKWb-hj z_y=*Py~sZJ$)|t3%*5RTB@8r|a!PUWJZn{qR4$0Srsy(+Q7-M*P?N(@S@ z)~|jgWQvWhBEG^1 z9knYWWQ{{QhTQ*)#OE=PIURI%He;ti?|ufYVuxaVOH?_!c3zb>-w7I2V(1J z_9eEZcU&-<=v8DL7EO$;&Q7NjDvX;}C%Jo_xH>SN+D)bInisd57A056w_zU5k?jkb zOc`!mGcgkyjx$a97sS^4aaSf9^s7K?#uP(c{4SLcXhd?ix*V-VBBU^`wCbqNB*h>g zWFnrt8e=jzTP0p}n`Vp&W`qv|Be)^4?{0DGO`o}56@r>EQMy)!E@Q9n-<^~sRTP+f z>GCT6OoPcXrJ7Q5+hUE*xT$8FWJ|HS0w>}i zj*Uwv8{v+o6}Q~V%rPX8h2mIU8kQx4&Y#^;E>kKLKM(F7-_#XXJt}Cks{Ve5Ohit0 z%VeH$>Q~xADj3{mvF)QvdQYZE)3vPr{k6o8I+!qSMx-#fSFDWopx0oTR%eg?xa{BD zS980GX>IAy7_mPZ(A;pk3lZ)wOM{8yEp36w z6y;)M7t8jc_n#PbjQ)1*<~2og{-_k?1L@iY)=3k;&cShT_hhfAI-A4%k?GAHC}^O` zFf8=ECbaTGf2%!4dA5Xu>=|3VBmM0SPGzGH7flw2-H$Dt9E1YelAGmdJ#Mnul)2Zu zQe4RC2(w09Cf+l5S`_vBM$@V`eIeE`o2Gh<&*U_1`N!hclYGiTxJgp1LvE>U2 zoys+(R;A8gFX!17gcZss!nrD`QMmTsj(^B|)MePYG7NQ@$;oM|$}z&-c4G%udOQh; zD>rS=v5c*Nedqcf-HQVP&K<*;;aex1%sDu=+e`#C0sm;>5Acor);Lm|w+V~vCSL|} z?~;P&&*~r@B{;GRiTVA}J|SB&1)G#NwFrdx_AD!2<1I|%13&Iu-V?j;H-~tAZ633h z8SZsf5ZC0>*kah2H+?)@5S^^PEiL6^IM3D8NWa=cI+(YWdA*GfANM3;CLmA0IcajV z9JHJ4Sx9#JyzuJCN2=Bk3p{IRPExAl7A30{`&>>sBdEeeJJ8u){Ecl!;z>&r&*1XF zzKVj~UX!8U>aQji21O_AB$Te?DPa}7d)E%IX5>H}v|r1rs>IK}m|Xv4Z*=dTR_&yb z+pvFp)ucbu&yF{fPNw63=h(RVb1*jk$w?wF$=zm^{=X*jhu>p|E#jDe>`#iLFPN)O@y38d7S1>HVXZ}`_@CKyX}LOtsgzajl5GDTwGji$4RH`FmB}EgI-^n{soTt z9``LnQc_dd+Hq8FgpOAcw9j3=&z?PdhUnMtN_8v9uoXIkymDklM&*ToKVD{k;qCNjKc9&W|JH;`Rp^61+&`3u=@9|=0Oz5jIw+Gl*Y$)E z{3Bm)v`McSI$wdWqh~&F=zUNBUXb{t$*i7X2&rA7yKZp!Zu+meZ(SGHupVrz4NPs@ z{kCJ@>V5hbYMi2HXZj3Y<+SD;y3J&0Ygk;G(bl6~0+%m2wt@Y?J) zP)N0&Yuv4<*cXY`OaAtU>+xH2I%~jB5EbgWnz2S6nLywjPoo?${o$57(^=fRkZMJ! z=4_B7df1spuC;;pG^k{c=Un?-45gM&xrI6=d~{pOah0?GlzVic)EfOZ8?QM5K{fSX-4dAbQASOJ zn0Bf~wMZ$~Y9AapA2hy4QMj1^I<{lG)(##d39D~Hra=yoKzYBAd5t@Luq`GDv@h=b zAHV*#8xqOJ$L7pvr|nlI)2^W+QJ+s7;pS1YZUa(G>GkeR9msPGbYT*XH`FNsFq)fP zm6VjgcNcu(eP}nzc{pyT%Q*L)@XGl*U|mhavFX^&|3^iqlQ&1jc!9N1yi2zr%pRjQY*~zKzo0;~;8@&Icyushw$$}}Obo=XEX(%jq+J};`ymNLgy7bb(j_CuQ zV@U4LNB&#am7|lHdL%))!D)uzgu7g%(TzD9g8*HL(v0ow86@sC(-o@&2xEHPE5Ye~ zidx}_*jvzekQ|GtOt&@3>Gq8$_&wsBSFCJA;OIU>1Qu)wKSE6r zStCZ*_HK^Vs(!3=STLXu(h(tsi?1KW84==f6`}=J5~PM$5)6& zFe@w*ybYq>+l~QIYd}!|Ii*@bmqi2_1lP~!vOq^jYvk-YC5ze9 zC*DagH^SvZ;2|4IN!GqMc~RLdLMLD2SfxMO8Xnh8@ELe1Q>`-%_5UH zz1|Fi-l_4uC`xi1QmwcI>iZ^r-^4N+uL>5k%@UhO0WKe5v@DoEzgePw;DGH|nZo8G zczmU+!&Hf+v3MXf4lc?jIfprh5SnJ~EXsoYAsb^%stm;3cgw#gP&x9~Y?+}XnZoM2 zzEuA+W2Bt_{{2ZtBPpKK`Q7DV1~)9(4Qyc2*~r#1hly8ievq*F{NtnQJuu`FA`uk^ z;RGRfUK|WlB>&;7B*Fd&bOC}$*2kN#ZJR`Aga5$hYjHeS{)w~{p~CXssNRjxfbx+H z?#g>dcL5CEFUF?%#OY)*xXHi>ZhapTndTN2MRPD&n@55B3|2Cc_{Zk3h^VL*_mcVs zl3By2-rNKg#ObC_T{0&-+a*3KoimQ31NN+kY=^*cDkO;C1}5i~N!A92))96=(S0kV zmX>9`9m4ULUYG%X6&FyJbu9r>)|!mr{P1BpD5~jx2!duso9f2q)CmxWWKn!SP!4MDUgF>gDF7kKGc@HhUH)H3i`3 z*pe{Ln&WVfe(f7}EL9e;)1FqF&Ar@7SOjmRy8u1g!OS4q0}c+EB?Cj|oi*uVfYg&k zHP{BtkUTj_=_5jN65Lu`T)dAZPI5xJ)dZ>3>kZi(QCiap0PE{YJz(d}gQ^Ky4^I*a zW9FKF*BsA5(7Ka{er8MTfr|j0PGXvzN&v*ZxetA`;L|22oHDoHuGGti{&{e2cKlvv3(1N>}xabh*il^;wPiVTN+c zW$TuZUN`8V+n^oslmzFXJ5KHkbuB9;96Kd%r}QH`bHQ#@K`bD=Wpw4gd2P8Hlp=`+ z^4pw!JB)@o*$&7AU@r7Gk(3!FrN~H$P^zRki1RadX83`AxKoNj{=AT|)7iyt)>(ay zZvBn}5i)}j=faHcW_@Z&$dyr=T-L8|A1{1+??zt>##Q6cPEwQVQY5bD9O`x(8ghxi z5)%_bPB~VMH0#cgv=2GLa1%FxWPNCc7xiF=sjP8@+Sh{9o(b>Y28vkLY_~o)c_$8U z<B4!t~HnKgnl1(vV;Iz5C zM{NF+D<)N#n85w=oLo71UH!!Inu6j#O+xL}Y==l8Z|~l{x*tAa%&x;O?2_M;rwPDl zF$O^MD0;fz2{IogR+chg7CILRA0o*+K!9`7-JqGY|FzKMl^+<6qXYuDM?Wf!384$* z7#KVx_T^e>fcU~PXP~)x6jC?7AepUwAB54g4?AkS0j6tB4N|CetZdL}gdu}x%(z)< z+DQ32T2@#YNaRq4qvXP7d%NE0&?Dw^JNgJmypa}3*^y`q8C0f@~&RJ+DLpg#4w`~168M6 zLeomD*M531r>w3njGJc8wrsG4Lnf%U+T_oW` zj6fIHTh4X4q(fK2~LAmLBtfK=+^e7PmT@uOXC>8fZf2;P^d}@f}~cd63-Oe zvgH?A7KT{!6z;LDXt^n(Bg41c#t+V6Ig^Mt4n>Rc z4Mpyja&#lHBnYKngPtcDq%tG|fBp4W>IB-U@Toru147>aLUIMk5rOIkVD%?`>O5*U zQe`6!q@m*jGLvFck|?W8ib5c7lJx;o&&`EAi-z0Oa|?jEr??i0V5p`D*VfjOk{2-_ z$5QeQ|ZgBtwJ4 zJ^$C2uX@7QU)c1TgbY`hgp}5}HC~go!FrEes^StXpco#RRqNs;^E&isTE?fE2CSJ*32l_upVmAZ+*uuj?6xPih> z2MW8vNmjP!3eTW7HxP}4i~M$3Ng?2h@E+d(bP@oG$Yng^aB}x?p_S=?6}x_Xef|XS zngfHTv0=Ym(*#?!OoXwQ3xLA?Tpr2E$v2Z?85KS~uU}}V9AS$uX%Rq)5Xbm<+!?tjY2!Q`&9ct2axOdLWMs-St+hpF(8e1*8$;YkE&f57!?k0;V$WjrWIRe+J`oY;>PU zcfUVQApR$rVW#Kp&Gr7q$`IE~`tΞP~qJd!3K5d;utH^r2Lzwas)t#0Y8=1;+vS zHbhhyiYtBeajLczV9M?O$5vTJA6Ec z{#$m6++EHbM3yA8p*WBsPPl&W5R9JT_kGmE=iw(W)??zpEhn;IE+#zbA!zj0^gT+P zqS-7O#{ep7(&UySB?ipyK7efZvJYnrbIWF9bD=c3O^_6ILXZ~@_FBrPn- zu3ChIu2qT+7L4Er#pqs~7O(F%$GuvNoBcoAuhq=u2iW<7z0T)x^~9eYrUv+gwXRak zu6v4A;7utY!01WTlQ> zt$8A9IaG4ns%uJqCt&6HyuSHTEOxzInP%k(>&lzsF51lcq`I)){%4cS1dazU$2rs+ zA{QB>U=Cd%X@~`2HK?iRLG0H&@7bWYF)c}g@4M>99hQRZCFC_ zF;@J%q6snq^{nFhiHULCDz*_Acm}tvv7!RwMCDe--iG(m*>T_zI_7m=wCenp73QL? zRHb+7$Ip=?YN_Lml>9%z$V0v4!o=4gH6hWHQLVDc-{vs0GwP|uD{-IWv93odVk+7y z@7ZsZ!^g>o3nEw7z6e_%-zxnB>+@*dVm;CLs<7Uzi0r{p7?A1-y!X%ra7&-q#qKre zw>MAOTUE^)*#EC*j+N#u*yB7}H9wDO#c}%7mgvwleKq1)*+l!Gk8T~7pM8FB&$H#T z9Wc{NAg7{lDio<~^RT_xM-dijnHuU=Z6iP1t7JA+sAd`m1xDVAHrurHqVeT#FWl0 z&6)pf3u|CxXISCFUsePJp6hXuUb|pzXi3W%f9BEm4MMiMEiYnKd=@vup6|y`RXayC z->H6Xg?riK=0vH^Mvs5~P_!^}&rVm~44=h&$1Fp)Nxa&xb$A8eo#VmTS9HWfJ(sAa zY_&LKWSo~VS}VTCpYd?xQ>Dl2xn}Ryus+I{z3%fxZAf;CgX7cT<*zb-5LmJGJC%FY z+3IGCmo>@m+9fW1L~Y57WwDnc&5z&s^@F#9!#t1HBN?|g+}fz8F;hpt-ZdwE|JS97e{#d2Yfgt(mV--eH5 zW@R1zxUu6OQVJ5gb$OSBUitPdW8vkgNhZfuUFA4)XKuUB(lPt)*Pl9m68kV1vZOTg z?sg};&VLlf95<*83c4L_e%0Q7Ug@iYYx$Jt1@@?^HOX2S@UB}YzPNc$e=+BcEQ3Rh zEwSONs^TS!rSI$cELqmvIHhFvJeB08INX&&6W>0(Z}_|AR_o%+r|imxk0c-0`S`Bp zd%BhVJdZ=F%zyOS_a4$bysFi1H}8^C!Q731|NFN$434^@xs2kiC^P3 zt~}-vEwFj>lA+#?e_mM7zQVcV;L_;B#|0ffPsZ!n<1%N`<#6V0&o=OX<(wyw5Eu7- zbL{QQl?pL$&l&la9!U{jx$?!eaR19+aTXqHv5YLubw>`F;pBEMP)#KNy z@_74jc=_`rq1hjQ@W{xxd<1SD&FLsk?>Tz(=={$(eU5ul`Gr%R#2$X8=F=s|@DaD? zA4+{&4d3`>`-Z09$&`A+pr7ga+1-k`yeA!1n zkA(j(x5d}#^B?pPdSWX~ zR8z+4Kw)#WEC4vSTEG4M6ue#=P)GGDC@6T=1GkYYND8OPkx<+3kgZ#}XoKQKbv+kh zB%tp-wuSBds+JJj1H-J`QS*-wGn%J7))LUYUDUU9Xl(7#3ez!T)Q z8&IopO)4gH4k{p9qYUvFFx=IobIAJXb5u=FY5H}6y~R5Z(YlMMu*Ii)mU1syvLpeC zQ&O{zfKtgJ6_rH*ZC}mz<}vBGmj>CI?GQ#ehD_)O%D!K?G-KBcI*@KD0TuNN@1RKq z)j2_v74QREk+;_E%xaXgt)G5>VK;2tL{4iXGb{<`cWcO*88&4PcNdaR#0O;lQ)Ac7 zc)}CHcu}<@N0y^f?Fc)9AK>HQz8sE8cz5X!55L{~JNapND`vJQDv-Q-cXxN{sS^kt zsfhI9=gnunpD)pV@6=F78rQlTFm~DNwT6=nH_Y>9U_f6PqajRR-ahof!|DvzK~bk?+8ksmL7dzT?HKbkA5nfB$8Wv+`(bYkQ_`WMnjd(N`94FcvQJ z#b<3>w{G@=HLD5z?zzatDfHySZvykk?r^?0b=<_yj&8b+nyTs&*oNc>RYL~f)@M@R zg8e(!TYXrsjzhe}4HnWSPpFy5cHqKP5{`Z?XfH0)O|6M~VD{Z~ug^N9?2DwT(?p2c z1y2L2+-=w&w3w3lHSotMef28f^e8m;FOu$}nBw1(eIbyETG`u6+EYLvT;bj{My|m~alrfK>#*2Xq2e69!ad84@7tL);zfslGTRU&w zJPm_aYODNwVpV^Us5s@`SGaP)b61Uquc~-h5<7;BJm#3@-u=MQ;V-=A`bgrF7n;(s zU6<733}O~9S@Ib5rp5l+HgfF{BaF52BD4z`Pg-O=L0y3cGr!M8?ilnG&CTth2V{4i zUGHO&^hR_+1@BW%bIEDf_wMDS>(5HooAwZ9{3JPM|5NN+D;cfG)Kn2NmA)Mt>j8bd zb!76UE&HRx8#11k%t18`2Ul2qY_+Vpe+uZBV_06U$w%Ih^;(lq?>h7t_@z~mk&*jz zMj^uW7U&Bv*xm4;s81VcKm0)a%| zPgqcvsYD&{0MO(ypjj)(IOi?Q+piFs%kk{5p({l#wTHngsWy$Ng_PA3z?&O@M(}{! zqH_Y=tX$vP!w=+}?)Rx{CRi`IVc3cc` zG_5l%t$MWUWsBn6^{T3>AE5LnPVciX>V-y$u#nL2RBmK7!}4b_K~|Q=S1Rj$#zA3X z=oey!-h@1#bwoN%9Wuze-*zYFEn4y_WLe!jfEC|^E0ZC|HGhBq6jnuQMk7PvEuY?i z{P=VUu`81cw1%z34VY2!`(0|h1JjhMg|Rno+)z!;;s4=>XOdx(r)`EIQ|EPX>qYB*_$+a3ev z<08CD_Uh{Do`EOGR#ZYwO-(Oi75a{78^4n&>N}tt1jjxGTVg_4kMuJ0h6j2C-4@_g z7N46v?=p4&Oz-N&-lk9mfpl(c4=eCcX4jlQwpR+0|KWb-MzlL`{DV}q zTZyE=#*L=Fxmu@BpAO6;i$n}%tYGTz6NU&r_5lf*Nr=Ocu^!>0)L-Nq>1Ux z1lDM5Ir&p*4?%eLQc|%vAKw^+kKf)3w!AFoyr`BUA6({Fg-RI@wYiRP<^CZij?vfq|lwG7gM-QM~Hk0P?C)8Y5ViD5F z&;igiP_@z$H$p+GA- z=%t4C9)_N&Njrd>3lnm7IKb}BQneYy@;z94cJH1GYcYbT6uB~XVpk_OPV0hAQ75_w zBs%YNVM?r9_ZFdsCpoOcVR>E}yagbdKl=z5~nFt+3k(falR+ zRn^Bd>y~dlwwoM=6Ckd+M@~-8JC1Ck@8-d+YAo}yZCR^b^9O(gfe*Rc`!{nQRX5l zY5AxpxPh2T#Gp%n9r5kx=u2b}N)ACZjYda?@<0OQq}}!yOuRJWkM1JFPDxkvC9?B% zz`3RJ4)Ktbbn^$<%%>?nexIDCN2B|rwDUv;un`3b*Fl5amPXZK-jpKd?YOvNNdW-? z%QCT5YyMAr-yId@`E@%+6B0WrAOfP;3o2CvDMk<*8f^3;BZ5*4T{=mADuRGPM8E<9 zVxyxJX+~j25Co)3O&A2F3W$Ip+9yncRu63~z})#O%Zxp)qIafs*-?&Q*=lYYa*e-*|bq>8F02U+WJwz$h$4IyIdQcR|US zM~*)hqoSm==;p53u&P+!>2A}}(P`}sT(Kw2{>53g)#=_U5Y&*#~jI>htUj`|IgBHuQ2*PuIhdGBJ_F#2HJ z=WR8+FODG{WAh^K+*yrr(+MyDB9Y+8{l4!u6GXgTz+e`}%!*y#cglW@xKZRD^rYt} zc6|WT7uu_1yu7?b1p+kQqIu}hgI63(>J_BfobtI37c5wihj>A9kT6_TKTVt?QZX;D z1m?+M5gm^RZWN&S5|8E}K?pH10vUOCG^1=W4D+8?jl5$#a~bevFQ&icfkMYe@eqAh zFLXTf!OGvfwrZfOZZmPN(=jpDE{K0vK_v`dY94GJB8AfQBgr~0-9E4W6=urYXiH!TT?x(|#S%nuz8v4oq z6Ge+|SsNnjD+BH#UO@-*{$4OkkGEnXWNMiWO_hpDPnQ!DgJ!b`jpWP18eYvFL^EnL zG~Bn-B-haDmoEwa5XkbzTo#&<-V3K~{od?jSwH6V?IxLB>&xr^c_NI*RqLKPctz7p z?E6hocNKQ1skXd+o&UqCYjZIz_eWJ#MAs!2WS)-|Tda&8&d8XUmGH8ndFJPPa?7J@i(B&VV7nJdwgF2oaK-@>N#$@s=2e!j0g?9>Gm;;aTNY63C$m`xyiC>-}4vgs%u z5sh{=M2oS;;)IKn8|>>uU(0(L;VpE672c_@!VJ_S+j zBaK0M-~tXDpC3SNKAwFuWZO$>e@MYTPlEepkxw)ldLm!~#&~VBj1xM6$oQDNjNa9Y zMgjU2ilvhryJ~^FFC-@?zd;`~VPXmUH>C^Zel|v;!OBHMj}{Y}{H9s3+`ZUL`9QyX z6xJaZM=_CoPottE+JqL$+nWocxhO7da<0fc=g`NfxHAzVr!gDAu5I0L{{rh?Wm##6|7g~s76<+L3J&Z1P^4$=;}eNW-=tjqMNyY~-ps$l z)0l-oJ|x-L;X-(-+{zL?K0m*=-rmmwSZ@rquWqqEVPKH8$xclbNAQ+OZjM3oyI&NU z2_l7Jmv_Hjt}yA;G5{=i=639HnBNxbt-h)_Kn?TZT00gkin^D&6#ICY&&g%i{de9H z(v!vD{}Gl{d*nFIagZq zI_#=?RI71&uMa;=veeM1sF1;))HIaeEiZHIg-%wg#7UuQBs)=NPeYkKlBu~~_u&Ri zSqYVRVB~zw{pZ}(>o!=92&e|1pTCNgcYc1_TZ(0Dps@Y$r+R(>_;LLpv z=mbnF4R`67@}%_SmBELqRGCbdPgCxCf94XrzP<>np$6k{`0c2+LHWXtcA?0acnKwa zO_3Q|alVUhiA0*6ICt6|HGC&cQ+-JjmCje0(rFtd{~`o7gdz3G8r5%JrQH6iLj6Fh z&_TsS<3^^WwWs5bip|kckK&gqX&dr&oUHEISABDn@*}4WK7IabR8Ai?zYu@sVW`~* zKfl`c7Pak6`6GuaSubTCY96ef>w6^5_i&{{lARjc+dIAYR5|D6Q=YsW_srJj6_Zg& zq9DFeOhYeS{$Q-$VFLlX;5a{fc`;>WKKDn#s?~u{cHTOmm$lEsaP1?f?zs)}eEEg# zk3w>j<*O=X^IUz`Mj55;E4vlF$y!KxS&m)HVu?g29ic2qrEW~ly&zSxRjOKI(uVQs z+G_}E*Mo<+G@Fs3&`^_xPEY=h-Fv1pjT2t3?$NG@HfcBuGC`2jdY@lF0(lU<>E&DR zSI?_h_u3!bIkl7B#W7>GJ2=F>_nGvu4D$`I<(FszrPmL>yr!*pP_)uvmW0`>T5o$* z=i8ArxAfSnpEaeLCM{9k==iouqELZPEqYn&dOab$FDW%fsoV9$z8TqH#0R;boO&61 za5Zgg%u@Ge*@L=bc==V5_3ggE<2^^xw!nZouH9JBX*~Ex-#zoaw>b}At!u7)^<1yz zA5IrO$WjPbCV?y4+V^qP+gn>-XMV>?AzELBG~=Z1Wp^t+cC@zLHZic;zCJ$oyf^n4 z*~>RX4o-eb5yL?pUU_t4*BfSU0G%SKCEPSk$zV_9l3a)2C%=wny(xc4$OIb|`+z@A zSKuIjto7xQ$?5n~d)i!d6NNkNbmnP(o8WaSOb$IF7=+u{W82pc$hkL$=kkeW| zUWM6!^Nzw^nOp8vH`vak6C5p8sKEcQAp5xjzgo1|$@WE2@i!$F36)EnQagOLesbXH zrNRC656PIccdvFpfyF1#qV8eX(XBi9J1pJ;ftG~aab>62tJ#OXGM7uPV9`{EjE6r; z>J2DAw_2e#Y~?Z@(5>~U#J1_$>#vC(8PPcUKOWX_S5>W$k%oOa_iL)bn9`#r^}+Gg z|Ih<*HyIc>xcoM8=1mEaj`r#+ z9r9x-_0_qVUl|{j4H&k1KJlO$``Pn7YDYZ`u_1hSs$)M`U2;vhbI)(5I{oAcSVH*w zCR>_3`S;69AM~AhH{otRbkwatKEk1UE^U>H9r6v2GOKTugXVSI{d~fRa!;mF(+17a z5|Mu>=lGOo7-YFbVzy|Fl$h9Bd44rR!?jhFGI=mIN3z@SZ?LPpLDYg>DEwb+&ZO$u zre!Iya((>=%A)K)ZLsavI->5roq4}LA?Kz9alpL2VUop^H}YjaTEI3`2zprf>F;aD zA1Jx8L;}y!&DlKM+xv}9?NT4#zNni^MMPuo_;b(W?%UNyY4}0D+HhHaxM|L%!JhdT zSF6@D84RfO(;7;~8adDG(pu~cR+N4jKGvIzX5FvfX4=b(*l8V7H+3+GKESJ>Wq%>- z&|0PP+^w_U7?p_)PQD`JI@%k7yKgmj{x+fp_TK6>MFAe5-bITYiH-9cymDC0eLF|i zAQv~-Xmo^cT~!s%)JKPo1uM=Bmt~|c$#MKCm*0hN@@HDE!(G}G`pn>5UFPWf>*yi- zHES*aDVRDYUQN9t3HKrEDnQo=+8?tvvarbC;ZvrA&xRFeG%tqx?AbBX8{D)bE^3oM zQ~C<*fh$i|3?MQ%T8Xx~%66t+#xqw!@jUpBWUq@1 zb@{;FoCHH>zR~7aa3UD0u%oHZUsFFMyyMMR&-6E2k;lgu2(+@jZrqZHFgsCEc{q=7 zatW#%;?rl$unU>|RJ^Qyzw$*I{PV|@R&fcF|9t39;jb(bgE3G2D}TW7%A4T3{{1UU zZ2tWbUl!xvYw=|z7>s|f#lP1AEAW5BUaWcMz%bv1M2`*!ql8eU=+Z!Olym{5x=~SB zz|Y?bE{J&k7y1G>{y*maHGP3w4AXtiR@f@qGh~?%)M2hgy^AtRm=b0?|L=Y`dTR4d zqG_HI9Qy2<$gHpF&J3?q8XP_8rP3UY*hId1X~Zq6iX#`CQFI#l6rb55^5Cd+mUepY@t(G zyu?>Qv0$Ymst9i|aaY*(*-}ky?T`+xfL)g^GPrxM%8H|Nc|N_BXjnK{L6C^ z?9HgF_=EajoWg$H*2YGSq#1P#KkDN)fBG-P-|8#(hRj7Z*lz2G8m6Y6Ugj$f#=Xub z4G)qV@CTzpcJKEYmv-MSf2)6vQFDXV9fNx#F;tKz%-)U24V!a^H7GY5cWtIu@Fo-k ziNMHg&O6`Rem{Xlh5gHyFNd~`40H%ky~aR=5!Oje0+JqL`HKiqSqJqsmlSp@t`Srl z!STM0^;?bbnRX!s{58O~l!Q zLTT}laOIN~chbRQ)ZH+b5!Zh6++rVMoh4zm!^-e-D7I*)|9o^^%r_BRUMLRM4?$3> zT&%j!PROfE3o|Z9>vBKo)q9L|m0kPz6Wt=L-+2_FaDofL(Nz;dbEWDbx9L6$4g*S3^fbnE~Kn|q*j7qT*zFQM|=`@TOgn8kF95J zX-RBOHnP%}2hPSqnJRs_^(qsUkY2R2tD~kc`{2~Xhl|lF??-)0k3Vn;3Aox*(z#IF zLG`JV7ZeR|AnSZ8)DWc*COFl7C!_Wdirw+c7>Pf?3lQz#G|EMrU9A6n{0o)HSFc%f z9(AldBu$wR;tFk3b_o~_AA7&*TZLLh$|GY`;;8zOq%iH;AkPoww0ZuJU=RTqdi1rE~~vVUpe$?eF0gnL@g zWx{}TQ{_f`3uAlzXfRw7D6-z(#Qh_}4QPNuC}b^oFTonm4+Amg$jMLM49a)c4q2Kgd?`fQ3JH>njnU8w>6`n5&DBROO zTK}E#h%Cxy=@7}4R z8=&QfjTt$g>w8_!#$=|b(pE5aB<)BtLLZ#sI6*$!!tsOO4&A0o{e5jG=$KijY6~QJ z2=xQ;tfEIvkI37bXnmMFXFI-%*I41ue?~KgE0RV}y$|GFl*lMy(MR%_BQUMR=;{PN z$p(Ayc+4PA4pM4c4718DXsQw{WCe(dvW4yTTBuHo z;JYBAAbT)&9Y_b{QLc_YUd~Vw!QK}^d9uV|CPQ~8HX9KEL%~JL18K0V1%FswHO12G zkyB63VbPz7`DW%vl6O6!(`sgeH=!rwZrnhiOnhvEBV3OS_-pwzDLvE`Gb({CtNXZ~ z_$LE{?6TlAjp6T%iEal~oi$wgaudWf#6{+KSxLkC5-XGpVRu&l3p)MGH>Ed=J)Ul0 z)CFL6?CLR$nmv`IopzHraWv;THiK$^;v%!Wgx7#c3QmDpaWP~<)K4(XUcAdqGezW8pwtfJGy^^0PCb1CxJBZl$r8!ndaUd8r6T$s!9% zGWoFT)zNm!Iopa+hpRTp;AVB^^6ppuWDm-ZDGNizf1W8Um$<9VW0dDiz}CK<4W&W5@DY8~_>ZwJt<>u)-yoMlRiZnpf*-3@YV-Df|3FPddjRM!W!D%P`&!4Pb9Fo>JXS zxt_vc25i|oO(nh_cc8-okMKE<;Z)1p*j?9U4AZO)E|q*5N)jEvc+mV5(j;Qd7-;)q zktj$~8cJzYgN01uvEpnz4)H)_`^|NPJT|e^Pg+>mU7Y3>{DHV080vgY@b-bbL)Zo3 zPgNPxYG91hZUwGc$jeLQk!AjJV(hz!wPyK=Dl1uRwQYHZvci_#Q&r-rVPFglRkLCP z8lb#;>f>M{5tpH#py)LC3z`N74bgT+VBGn|f7=ymE20uRe*Eb6so%{suNo#MZq@0+ zpfz>ZK>aXxCi&w<0t*(X!f${Q?-fj<>_dfGIrBgz56vJQC2;J*Gk;=87rZHpVgeuW z^1_7%kXx`NCGMegJk}K!qh;rg046O2^Zy!%*jz3K5pA&|WeG;Pa?Re$nz2^{fEU}A zNHF65B5-Cee_e16(Bs*ef`OST47W(oX63-VkmSH8TRoWiHbxr6U^;^`n{lZ4OL*$$ zBURf=OcSDdh1G!*ch$i!Ky(~cZ!0^bty?TQ;~#&1B1o zW+3tNpO1;}Cson|EF$#j36ED-2_ZF-kCx;Rr50k?!FVtaqKU=gkBg@g`2qm}b#Ncw z_cSRhjSrWwp&!IY@D2F5RFx=ui@>Q{O$-oqVH;I$CkKd#c|g!iFx$XP(z=3aR>t~ofuAU0JaD{>VpQJp9XY9ce()fvTKrR;bJf6an2S=H9 z+8TF2;8Nn~lYddUcOKeKwp--Gruzv>5n&`;@Jdk7kPiANRA5&wVE5Bta(48tOT^6g z$DC_93}faHlK*8Y&ko=;VoMN`C+C7$%ZhemM8YQ<1J5m-Gu_->DNJN%7FY|%{PyhL z9p>&A5PnccCqp9?=PP)SKn;zF7If;Wl_6a<;g#P+tbopwkq?({h`58EemsYf_$_%( za?|Ku2X!QOD1Rf)fR}y}URUb*3V!n;08CMMvi5Wzj)|E3E#V>O`$0_#V z5HO+c9>lz0&U$*2Iv8=pYJfL5fBrVd|M`Z@^#@}Soaqsn$RAuXQB6a`8yqc0T*(Uu zm1g%Web8L$J1yL@^apz_;6^eAz(Zzn&)6U+)N*UruC)#N9+nC#tOn*Xc0C(>wcrjp zhTE^;%;gc+-ju~9VSA2`Wz3kS;@gh(5ytY{!0)gkw*5wY9RzXQ{wExIHgUuVB}%bj zB50>Xp~Z+gF@#t07#{>CA+nH#HN^0FdC79*gI`1Rl|mN}_)QzEVm>rC=aM*@WIu8J z!be+H*&xtL^Z1#eTSSK&48vz(nKXYIRxGXz5ef_#*+iJl@S&yZM1CVh&FD|5K1ii$9H^%GfS?4*e&JC`M%YQB*+6fvC>=W&0*(9>K$F~DFIWJ_ zL_7h4SU7z1hKGl(MEJbKDWXCDgS4T`+gsk=#CV^NPLsZs)t1U>p!7QT6|5q<2z5S) zj}@6B$h4u@39kH#%s60dJ{F-SOk#?8%)NW}5V@D{=%@BOQdf&IX<&nyXljGe8Kx|B zuStMG_y&0>U$bu`9NeKrm}!U(r?@*fcs412QeiXtOi3}44(Rw@(q zjp&lO7=w_PZo}cW?clAZ{<&SLb&}F+Mck!O*QH##L-D>{}1_cBzqtJnnt%YS69H$3Av2!Inxw9)aNRnHd5s+K^~_ z(*#+Rw*%s^;DpiTj^$nNj!>i;^nR+T)ZWhb72pQ(B=1%<(A}T_RF|t3uvpPH1o;73 zx1o0nl|9_-QOl8XA`5_^@wRZxH-B;{q(fh$!*Mt8wepdxuod{;>)O3Lcf2WD=+MZ3 zy0C9AhuyY`k8w5m?)455R$LiALc)_4-2m!V=d~FRet|<-gX40&NKRQ1fiJ^5i(9`VMY^6I^G`&m_T(+E_tQTZVAY)=g;y`%jlW}g}dsJDg!tQYxv zyksE~vApqYxIZ5+`7-cVaZb?db4{kZ_F+dV3$`Tn4+Qlb?`8hpuwk2}s*&j-KtN3e zTlRjZVu4#TXr^g5S7h3*Yy}oZO_lg}aDgC7z#uFFvUp*&nqg~A=e|Ny-L#n80#I$Tt9Wth)!YH-l18lp%JSYLubvp>RU3DOi4oFgo^ z2|jQ zh1xmW??h38ps)y7m`zx{28YEC54Fk>hRw{+2pjSr-P^s0~gZk#ohv=A^EoQ^VX9 zp(xjpgHRb9x(-nWOcgPQZ3Sj}$(A8IwHiH~Aa$y$g4M;$f}U*>{%MmFWROO-WB^fm zDKCwFfsy$oI>;earRU^$MA@lqi2q-Frpp;8z#R zk8{`wV6>~*PAPpvTI8T%7F}?~rX%4+yO8g_Vsyb?C3I_H-SX^rJOjW8F75d^8ecc2 z?Ub!q3>JZE-k*Q`mQv_IZU9qCZtiM>mvoHUzh4+h2H#dh8>yBEB)A3xy=Ot89k6m+ z0U5|SwQdq*Slz{DC;W0LYB_R{FwOl##0-Uby!EOYeO6;T+BCwX3wTa2+zQ>@R+Y~h zzdGgdhoepIgT;#y^fQum+B`>xi15i2W7Ix@Qv78ZC4kn*h{t**Q0v^qbn?jfb+YZg!J0Q94z7POJvrlyfmxFCqIV`WIn1^b|lLwXyYsvF3!7sz#o zX84OJu&6gcuI1u(G9jrC@Wp-}cR6i{coN-5h6dk(TU(+z4_1@5KNNcfOTuLC>tvQ> zqd$NfPNizgGIZI||ESxcvZ2726p_|`fc$N`_)v%DsryUJ*L zxjt~-o0b{k#WHW}DI;KkiBhZ5j^VqFXW+nao7j?-mJ|1x@@#UQiBGw^X)_)>yjjr& zC`v9wzt=>dwZ{#byvf(A(Mr zY~)AU^-}07^9JuOCBGj)`05coJt10FW@k-OXaz6^Hfpv$7hw0QTQ~Km z_p(Jx!uppAJ}?6(Yg`)XBkGj9&;zEfU(TAEQu;=u7ktt9P*e}~(76=c-ET$NtBRU* z2^&JJ`fiavl2+V&JjAHmZ~*F>X=V@g))vd(1{<651b&WSnY`SiCkch$(kI+rL23)+ zpgm3wZANrFi)pR^8I{mp?n3_B0O|D2YhNQ&KhdZ+HSxMD_C|`wJ|kqkjD`!fZ1FpB zT(Z5zD4;0|a3C$epSmZhQH;=dtGrWpXSUn}(kB{!K)oDkkiw1SMxE5j68jC3Hxxd> z98A#dam)N}rV$LKDm@>-7Vq_#$4WBv>nMpHMJtvj+?(h(k{7`oGH~?_Uu5a zt_8VzLT=8^UL)XEk!r-QA^nH^!fF{xS*j`H8 zBP1x)T*OGNqW?kHz0vV&`)+g4b8!9?8Ue3OJ6-2dZ- zF9oku0U%r(IP+s&ZivZYzln$)6dX*wB+VqDFzZSQkT^VWY?t+vR3>C~soTM?m3$R) z+ES7nwB+Sc--CuBmun7xedPJ~D&>BQWY%)`B}C18#vAYL-`fc*gGe9^iI&td_zWr; zh1wv;NByzmW1phnwa}Te06vD0CAC3<;caE_x6z~5mz$42H(R7u61_&9lt0GggsoED zSWWm61*cyaHL<+QcI2V)W*+UF<_G2beDMRz55;Mu*?{$vrDS02FkLUwo>5LcjaIyZ zUUBwl3}>a~48+A|JbCg&?oGsR)WU}hhzLP#nj_H>f+aW&HQ_GDDT&uf*qSOT9g4El z!!3+N*_OTEKCTJ_pPwk~N$9SR9mYaa;{+3`0)Bykf!aB4&XmhTkeFU31EZCG8f1My zro=dt?N>p<96|$FRUX=L7t;Q_xJK5eyuq20No}#)GLi8{vzW?YJgbSz)qYya<5%IZa~pH+{uY;kqq8orSwwDBz;b$HMAx}Vc$cb7V5_Ni`Djj&PaviZL&G`aG z7w|;4C;6);+KD{oC)h(tRW|jZU{m`cJFgBgj%L6xKWrs^_SA^;o%%ebIW-xxmk1)+ zXabkS?E*xHHWdi*qezkjL1z**%b>EQT=*QNE$2ZtKG(Se%}XF$~>YQG^<(Fw1Z;m@NYgA#pQQX;q<^=bx8X%fF@P{#lLx+j>9m#_KW%mxw>_ zoAYPW>*!~xziwhmiu>*pG#2ds>1^Dr;Gd=spPR>?`Err`HouwjDPo~=Q5wr*{I}eE zQZsMf%d**F)1P`@EWh|-r?r3R44>Uywz@3~I(q_Q2HV!%c48h)8SD>z6*~ARyd$Tl zWBBUu1gBf&6)z~LlU@;>-=h(S`x@sX;c?i5Cr#K)@~+D#emr*WhtKbLtWfdzm#Mx_ z%-3nfcXs|h@z;NfZ0D+>PW&JK?UlUjKX7yYH{Ut&TR%De>uXZvYK13XYa#|+x4y3U z`726?;Q&2bXod)Nep=O~-a6HkBZyGy>>SY>Xy_6xe-#%OM{UHimNgeK38RfBv3O0H zdO%CdsmM&UkirZktY1@tIo0x2AZ5)76G-7~B3qpMv}aTsN^!Ka1JF?fm>`yv2wV&x)YVi>u8xXg)ecg&TKqfk&BE6s4{WUDO&ORp~#$e zM!eHSB?hI`OO$D0ijHVgAz=B5lP3#rW0;Th=FMp%icp@q2xmj1I@CI*^2}bJRf6o$ z0pzvck3y7Y5B`8Wugx<%gxD$QTPInnL`RzMgxsN0^VYO_+dY(CbHgN%P~^9d_P8LeDv82NXH#x&qAc2pNgN5S_B;cKgC&vlEdkol(!Pp5 zN2FS8YZNHqf;HxAhu1V~c;v7Jt?~i#B@omIqQNXE1S@n7PwXfA#YUe^hnHju{3YEH z%6Ww-8hiYdxM3GMmP1KUiZmO1syuOzi+ej!ZCLPa2bGIx4o`%!K_HzAY7|rykqqgt@ugF?=9&fPSZ_0}5Ja9wO}Agf)`yRHVWU4bJ&78%LPx z%IGp*(_V)`UMu%q(_Zj?1f{l(+7s-+cSd{vKES1}NJrCPsJB&db-W`~XJ9pV8?gnStBK;u}GcwutV_twe*`C8hIO@gDSsUNJB`P`wj$}sK4m-aKcBt(G+aY$`F!6Ufzn1MCV&r^!MuC|-S{)rTC)W!jnp z6jyEmysDEy;*eg~|Mq5uMRiYY)yrX}BVC$u+|PBof*vaJlfrS6q`b@NfKVF-A8!k7(7& z7b2NBJr_bRwb|pyd5hmo$jHcOFcU3CI%#+N{X2J_A-JETtBkBwA7(?`j$%*z{-1L5 zzPNLFeYgr61rjrqfG_j`sR zIliEZMzJ7j<|uvMOYD7LOu(DDI2~#Os6Op(#4|5B4zNL;?QbOXU1WDhMI-8MCHO-A?^zd2dc9z64$>?74qGiabPU zh6^N7o%RT((xrqp@b^DvqpC+WPE2uAuXjSg(FVU| zU2Om|)%uSv4UkmoBfUW259{hJLb8_6C}Z(d$q4){%5VG#3(>*KAQM zyp3pzL?QtnE+bZ#!JaAxYb9_w2x4@-DCHdPGoWcm$xtKFN9#ZG0s8}6P;M$ARSKF- zB#2ZE5i4NeF!?*?)~yxLl7UI@*DPtoV4--V`|)%BDnfg1ty1AzI&Z+c90R7oEo<1> z*#)+M@j#S$h&mkTbDc1@Kbw3pwhWPz$SZIs6@fN+9EHs@12Hi%(lRn46`;^vr~O98 zw9dV{rQlLOe4H{lLf}Ph?WRnpQw41z?Y_>aIX$P^?T?BkdEWInhU$tIkG-bAqdp_+ z46L%eKr5&3GBG~3KuKy(*o2?-njf{t>`Qm(Xpi6)+hYhM?U4U$h3JIRv62(FuLl$D zADn^$WRu1?nK1dmpfdswk%N9))%9$gZH~&6Evi|sp_x()Bqr%OHX7Lmmrf5w|4s>V zSR`ce79^rw{oDbBnEEzj0D0}%(-%!xw9?_VwHV0xrMFv{ha5Aua5Bh`$o(!O<{46w zK=N0xzQR{VpjvvB2rbxqud-d8;48|3#j~KD_Z)SHgNF}S{`I*9`P~b#FJNh7iX-gW~p+YIq)=(ORl2S^0Xh+%_DkY_) zy@%iPbdAsVzVGknzx&Uh$K!fj7rNf#Jdf9LJlAphD_@Y^PQge)LPD}#?%Www{A+@Q zWb4T-y#>;E_?@t`HyJpUNnLQ@}ud?;(Ic*0LlA{NSe>S+JNI8>iAR&=Eb4uMM z_UD)DE_*KiK0fK*;w~*FaU=Va`-sr&4$|fmsR=a)xwsV7l|6hnpP~-m-|Rj??KZ*g zxJ%43dEr{GluIbfn>S|8Z!Ev9>vb5=xjSFzto`^%pknEx%2w7*cmMmZ8qW> z`>zQ`@`;`Q{nsOUpN0Sa>UR|a%iRC_YoFn$>X$h1-+%2kncVl^-}<3HpG)w6fA5<2 z^o6qj{k=_%DSHq8_qXQf`@diN|9k2G@7@02>h1rxB3IezI6I<}s?JqAHqw;zE}~I= zx?p{kcGs>0E9=}EX*CmK+ zjdZq>C5?2Qq3fY{4^h#~jRlooTAchEQW6{-tW=y`QC%(DeE)03o#FE*@LhVz{Ol>K ztY!Z?2>iY>Ucxv2^A!Vb&(1n-A|v}cKi+-!{{6F#j)MIB{Fm(Ps5Woj9L9B#LQ+z4 za%$>kPL7k8aEIxwTeq(M{LH6U=(1DTuJ_Ejb6XOVlAfqO?;jlvuxfiz*4M{!`pyo& z+}!Jrb9vd>$>Kfd86tReZwCejif5D@I;WuUKAc;VnwdE;DvJL7`}d_iJmMBSX-+v`(Gb|E&-VW~xNe%q{%r&Sv6c7eOv-(?hGf6B?H>5_t@*$INUqIm&VP@Zrz-PTUc^ zhWA4`l@1&}EN8dx{Q2|h83uc|GYX!pkKk!4_H@rHDByl!_r2j!l#q2qqHTIw+9g}t z?YNLGE-t0r-Apvx8fS*;!kb%wGr3p3T%GJM6My+8^=@F``?_%Msfm)`1Iu$`0i*}_GBYdZ zUeOn{`Aluw{Zf7-nQr4Fe9OD<>$_kglOFDTYf3$2K>&TGR(TP@ZG97x8 znR#lWw1FRxB`|8Akr zGdDByTj+JC%CTw-DU^>m^2pqlJFU5;WvWF<>WSlz<|uJj{#SbM5?y2}78hNZrM#F+ zR;Ra#IgM`8)YMGVFE}2ltCzQ(MN;qDV;-whoU+0D!h?kF?~{n``*7gEZthf*qVi-M z+>qktJ!KEhoH5n$d!EsoZ5i5rLIhsmj`M?IXNsI8SUD=JJYP5mXiJ$2@ZRd zly`C2DGzq}u+m1U-+`JL21YaAKhO(bVP+Jv@(rNjPCVs9zWDP^meOCyzRwB zpF4Lp{~RCR$-)wxntG_G#G57ZsD62lb;pB{5aW=yBqx6~r+gh9>*LqkInyIVAx={kO>rb90( zw9o`1$YegoouES}{m|ytcM;V;xv)@;t|#PWKpgZdJ@gk}sNcMIZ<~#c4K)po9}eE& z_YX07jz4z&>Mgq0(9j@$MRey*8b>nQGy))pCyVD}$mJGc#!G%%ayeZKFN%DKEb^&wlVzwxyc-D?L%!@k1JE^ofayS$${& zJtJGcJbLtKaX8+a&GOZ&S1xPIIb~+(UWz9D24C`>rcgUciZRFCe3EygJni}P!x^_G$v7J0}_+OWffb5ltx#s+{b}2dVhDPR+j13$KtNH(B|octXdz~#>B=3ZrDg(Lz?Yy zKvHraA0J=m#Q69)aREnP{(ATBUD<~ZJh*~%5-yV@_h@-SQW|4~shE!2D2qR8sjuIL z`%poA!v3<29_vfPjG94g2NJIvo?;$D4bqtX`t=Ld9huDF&MSKG;AvVdsjK}ble*9T zJ%PxW{$(GjHH?xTA`K16xw3CGBkZZtA6F}W)GOesDhU0OBER4z>`leE3P{cO&M&Q69<4#o09 z*O|aNyCdA(+^BMTI$neKI4x%J&z?Oq!Tb%O(Q`2VTJCpFQSpOUUyikAXxPhp%S%Uj zNy+;z5;N%R+1qUCxaf;$enCgy=)JoZC2Y6v_Kv;LE#J}BSkMO|4Pc(0QENc4?C!$gQAE)!B`g7HhPfvWbmR;Digy}t5u0b_a}IVJsr z!E1AJ7cN8;YG!dAuzdO>H{AB|EnBvfqCZ?yfA#d@!fh&+9S#l-MXpJSiC2d|?j<^` zii*nJ;NUxPaeMXj^}h}c`2oL_x3vk$j&mRKrT)|F`bvUviwNw19+Ik>Lcw_4ChFzO zQ^`-C4wU(9&T|;vVNm22UPKZ7@+CK3MJ4I%cA^<|jPf1=%)PoaZMJFi=7Zwm%=qe8 zot*s8=D9Au@WrV6CU>H|qQZXur#3M}0VIJ@Z$Ep+NlZaZhwpMs*Ukg7Dq|A&z;kZo+8)2Z2>wlQGb`2nK{z(j4C7~goNbs<;%o$H?p*R zGc?%v%J_2|(Pm#Kk6ny;8wS6BB0sZma?E;;nP zbJ&bFch}$ll&b2k*w|PC=<}P^B>)7-7!N(;RC?SGB)x@-O4h|i1k`lf%U* zd?n3=OoesREB%7o)*Y`{j~t<7U|^V>p1zfl!Taa#mNmF~GLw>-7e+c6Y}v8ntbzii zIu}jPmoG%W1OgxXU>No)G;0rl#86N}gMwbM2ZQjHo|<3QUwV3!Uuy8QW*8dUOFcwa z+eUYE|Di)$G7L*@gWt?&s%9Ej>RGzun|_R!3dSHlfVU@Zy@LY*0MH=)C}_B%vWo)) zDT%5e=FZG;<4z`~d-#(*+!~aX0W_z>xh^WB&Tv8J-?dv>TIPAK9KX_A=ue>Ia z!YOcNuG-t5BFd-O!*y|SF)F+FR;l{n7%s-wt{m&m!Gq!QWK>jCfY0KMuV}9H%j_T+8#t7S8j_)WcEL1LVwrx$*7SQa9i-@?5 zwleUeFbV6!-TETJ!hGMDHx2-5+zkl0m7dNcE-p^IaO;bUV{u-` z0>i^;$agT`?ekjUIbpjrJv7pl^KyV6ry|o~_<~#)TG+z}4=!an7mxQAxo@GMkk-|sFqwC7i`I%WHdcH6tDIdehb)h>hA`x?wj)TDFwdl zhLh9zPGb>GW*qG7Tk&b-T)D#K#XT04$L4k4+Uzu`zE6Wo85UC& zJQvM%<--oKRs}K~+$ZjGNnL_R_ib2MN0!<7+ToiwZ+WKTX>w$OJ#LgGvRMB;?U zac)KoRgJ>$v%%QV9x-nzpWe%gD%csbHf(|Hcx0R`}f_)uHLX;*?SQ|Y^P_Zcm% zsHSn8dO9%Q-D}HpyA2Es+`QE^G&Df`XpJjwZ$}}VwwDlH<9Yepp!exSVT!uCdf4K8 zx1Ap@&qwFapFiJ)(8%^0>|ec3hCY%e)Yt->@sojEhTN(;NnZQbXJc z&1}0qgU{XFV{1%eCr;ce_FM_TiTu7;uut^bnc0ax!_Qq^T+3_dYWL6(1I+3peyn|b zTG%W6UF5^PDJKhy)Wy-48N^v|{3X0?+qNcv?MzT2pQ^jst@SZ{CTHi(<>cgyfAzde z1dG)fyYAqy8+eco6(m+t|E)IEFD54DW1l;Uz4n)xO<%FT<3z7M1gFs+Z(MXbNDxn4 zRu?nNoGcj4i^nljjiEG^fBbmCdJVVSxIIl9n9x8a?p>$maG0B+;S*^xR$pqAOn9=c zjt--s*9r(Wcda3s68@a?LhPrd6`}WNnw5|}xS(S^bLbNl6AO4O;T~;M7;3oS(HEiV zsd%CK%sbZ$tbZUw;;lRXD$MHN6d#Lg&&PiSpL*9clx|R@qhSCjBy^?VQW@#jZ{JL3 zzlEWa{XsC9da70_drvryJovbkd6f6Wi4$m?0nBaQEAzH8yS;LX)VadxCET*lUAS=E ziiKgh?n8eO##B>@w|89Vgiz5D9-d0hc#lL2mO{nuGZA=gKHz_!u1`3Um7ygHmFi{g zP$R;^1@Om}OHPU-#a?UqY+rE7z-PI*7SK9E3UMUxEu^;?_4r;_Xpd+mIH=~o)$GhNs|3N? z-euQCYV=sj`zU};sWytZ#oy~bHyTc~ zSCwiLqk;={&&&DSci$>=*H+kDczxjgR+T~*`^s@E`~X4u((a?A5APr)*N!Jt)J zftGw0;=;qjqr{wEoH`|!e`|r3M<-eMExD;iJ0D9H^pjQz)tx2L+1?C0X5@LZX%1pW#-dXRU# zD`#8WiR)DuAA5%7Afirwtst6U(wOJmC=CFoxuX~BruF49-fWAe1G3&3tMemG2cgbv z0x7veU7k<~9HNzXm34J(4UH^3$(fClGfC~)g+<6>ly%qpOE)H*MV5Pcyfbw*aU zz8P3gLDe~kQAjz%py=lUz~Ul4F3pgBvB$e_p12!dzkk2%s9RW6WM&DjcPzi~76c~A zTYi3iY&_uP;52G_x2Z`@sshWv{#Uy%^mhNha0Z4!B5(z5n%GF*FdqO(4 zZcnSVKLH`*4)oZAm|xXq42cQr^T1jDz}Ec`gaT&HoIPvYUwTvPt<9rTPH~c+T_c&h zNqUMsnY_HbpqxHvGW(VbG#1Dtyd&OY+Fxb3PQd(Q0_Fm6^|_ABm;onCj=DhjvbIZ)MJP7mm_U2 z>6)9Hvsx}UJ`Tz>sV1akd(rceM=Qe)t5I6nMS#PS>E}D1QdWLY<~leqa10V#;64c* zdnqP5x;v;4dcBTrkhz;+Ljw>ha6x2bp5PckpSHd~dNl2t7~0UX5b?3`*9Tcy%lcOd ztsX^6;I^buP2TDU?x%4I2!DZXILUi`X>*a=9REmRG5(&CS>#i~{AFCP?K;^OH!+SF zV}&8XI(%<9g}V*~%uanxyLf%sS|RG#23*U48DC#t;`*M|KS)b{Q1*UUVTYrXRw#Nr zp-FX4q^GCzLDCJ)KFr6r6K_-wMieA=dEsZLPgM}pAT(_9;SU+^#o83|hDk~)QqTU< zNodSymxceD00$1Z{(TT!X+1q=Leap*L*Z-;sdbUw;;}Z@0o{N+-|>gBvinT)BLVYI&y-j| z9OauYa?t=Ee0t|#PGDGCT3Q#B!TdyDFqA>_%~p!|-csv}NpfAPX!94%!1)Lz=+xoQfb`>Lnw(16&2;Z;CteC;%*(doOkoAW|~eu1+Hl%}-w z^dPHx;P9RF^igZY*M{n#{_K9C{+tx5G~AVb%-7F_t^(4hPqVv%ZQh6IGYAqX{NmSY zqUC@ZFukG%A?@2kzWliW@?Pz|KEo2{8WI|s={&BZEh^fSEVo-bq3p@&JCogZB}N19 z{YAM760TmqE{pHMw!-AHJ1^N(`m#oUZ+GYq%-S0mycH<$Z{waZU~CYuv-+f5r_bz) z1#{gBwKIK5UteG7Da0!#0Fg-ITPHwuwCH%X7rMJY(4DyFGI4{;P`&J@*q=XtPA~3q z3`6WwhN0BK;x94%vAI6PS0@acY^yek-rnBM;YhUeA(!bF`8b3%9Hfnp1UB1tWuJof zq^J|AT2@vTC1P(p@9P5P4L^~9pFVW)#U2pH$C91_MeYj&t66B#uymWy@2>v8nm_kNd3OS62^W5Qj8!tEVO`tb`v{J3D5Mc5cyEEO8-LgmooN z+KNQ> zCJV8ze|$WMV8b4Z)G^0x%hPpp*@j|IT;GAyFD&+&JtMdRn1FDwO1#!4;duVchRU9V zen+S^AniLD82tPDFWGcu%c(q?n05-;MP`KPkhv^cQy&gEd(X9LF8+8HZ~6mgj<_7?RsBDH zYz1({L>z2S*ZrJp#{zYiyK^1p+il42UQ0uCl}^Mxt|@+%p_O?6vhCNAk-M#}>gdX> z&=x=p!T;Z~^?||>$P$9C!o2bhMGmFMX;W^m0H;v-`tw@VU0sY&8eqXzp&AKq1y}Lt zik&V!;LToWb$~YnQ_pIAp_y?XmTIC-j#cMiK6Id1Q3rCOA>ypfzr|~~KvpEmQtsFr zRm@PZcI)}5rsn2d#DTy~QcgUx6-1R4PTdLTpARIC_2YE`fVgOR4FbWkjb=xhItMk) z&CMN0Tem`8BxEmDRn@D&&A4uo?yDhYu!KJ6*&pon{_R;gZUDtta(U#Sc79Qj1Q13O z)S;ga#t>$HL7?;5Of`Ue9x{G?qBjVI%q=X`Y**q;mg_U$8j7NejA2_yiZ zl0k3`jIx($@f2u#5C~secN|S0qPtzJ6AIDkE8171WR)HC6sR|Z3?(-6+>jR^`Se)M zTULGn0d!ibAga^$j*f(D`UzgEi@UqIVW=$(1gd@F4{TNjDz^b9kwQ6C&z`usS~9Gw zyv)cRFf9rl48Kxq>LcgY|B@;0-ld3=`n{?~@}SUdcEsqaD6m9W_{`TkELLBfUIo8y6TnJ6F5l~o8KV;5}F z=I9CB9fF{Dc6Xx!zFff(oodrG^hbm8$GIwU#W$)1z>|TpWab1RNX2gb$f&D%L(~(I zjC`CN?LXkL_&i?C)YaD3wlbq+oeR#T(D(BS3VU!2cA`6F6++uqZT*hAk)Q7T{Mqz( zcXzk$aB(0$FuGBA+xWmh;$rS&HQMMw9Qf%LjEnAS&9KJg#O1>L2npj7PVAi3DTQM(8j zOeM^4OUvs0zNcQjR-IWXi5>(UTVz5PyW_IH`mWvM#Pwf$*w06V4b4nVsbsSR?Tz^{ zUCS9?Q#U&|N3*gpK?jRVXNH}FBS6&Q`xZE7ovtfWgVkYj*s@kP6JCC#Yf_Ky!q&*NPRGRFE! zN)V9xz`0d=dEA-Jqbh6@ACO=CHG9fd5*euj5#hgK)oPfx>#2;f*J*{{b#8c@qX4 zwSfvMsj!?R^jNp<$3wXLSu>9SMxzAG??cdTdbEH$nTX+;wTxkznroog(V{)fg4@Jr z{*gJo+7(w7oaspwUnS7t9#1>9g*MTY-VV)F6K{59CFZTN}SGNEotCXJqirE<{HT@A=Cc z;d1#;o{gbAHrI|vGmE<@AAd3e_gy(mIKyWF z{y%%YCXlbp?uWR#GC1;r7f!0^c*8P1cm8}!K^zB^?g)it@-VU=rhLxLK^#>VzN^LO z#Rx5pB1IE0x`wuFx^$5%Jn!n)TfK#&yS1|nkVnznlMsq`7jqohU;i7EkITya!jPZ~ zuY`m?@Rz*CZ76!DhDKsO@%TO zzQWrHD85M}xr<=bK^r`Fb0~a@>1e%q7MLwGyaU5VrrI$AI#3Pi;;f=6t@Mm2-3>wj z!7ylv@rJ6KRcN>QF2raVl@+u+9&U5EKW;Na3-(&n)$aJ#12_f&qnetZUxi+KaV%(# zzHs68^IHp92839wKJhuzBxl|UNHpM9Kc=uRE_O1dfPg^UO>J#dhS^O{WO~$@O}XF4 z^!uUKTPr<*moU|lv?HlCpkz|e?$%3Vext}!~!)w zvCxml#STX(q;XDS)(DxXEU%IA&4CO|g>=l!N0_`fQulfPE=rtmof?pVs>fFUyPmgX zr~g2&98|M#J3IQh4gaE9Pl0Ge!s1^P#>hy*jbizz9mL$JaO1BrB3K(y`FBf6No zcc5y^H4~~yQhhxSp^#*m`m=oiQi7RsyP%+;cNN*R&sk>M0IdSA_`r!Iw6r>1(7rUi zcd8-I-y0ZUurYjjTsRLQrLNg)eaSrAwp&AG9n{`Eqhk+D16>Fwt5H(m|~V?|!8lDjFk%F}ImvQw0Hpcfh0fuM|;}qE8Cv7Z$!XuA~5j zmTLwwypeJ~l53eKf?IP3-Y(3RiJE#pw{WiK!k6nI3kcP`2e+Zw+oBEVi69T9oDMPi z)(Y@eI9VX%n~S==kTpxyNRMXYful$A37!p6-zsrPS>Af|XtqQd)8ysr%4@Dp%(U54 z`zF74z3=GgNG+84DpsouMk4#PqtMl%Ax2oK=`H@BkRX;f%lSZBjJReUy0_t!JynW7 zDhHzT9-M#n`&!@c^*d2fQ9l=)ot+8V0MwU~yhL&X@kSHmTSBrE688VG3JTh)gX96p z`XN8vjEmEE?gA~I^BP90kn7b<)7lA#1U8aghkg{jK=B17rD`<%LvmxEGY{JOX)2~3<^J|9>)ja?IS$^h2RD{2|+Oe z-U6`MZUBP1I%Tp~&pnqvtaCsqA<$iO%^(hiuM^3516uW{x=bL4hIBVj## z`&=_cy({e}k0ZoeB>3ln2v*_dvRQ83x|KXg#$goh5k#nLHe0kJG&sS$Gyt`VrY61> zPn4m`hO0#Q7b3?imlce3YlYu%8EJk?z(c(W0)MXg54+etIz1qAh)6`r1?FU<80HFpX>%djoWjEOE6HAhO+MI$?%Q|V~>8X$2>GgZFHjtgrf9#cU3kBy6Dojmsat#QA0@FA;l5mX7iX2!sZG&5`i^23jft4TyMKvyjqd zxJiT!8zKm^^@v{Hr-WkpxrcH4{%leRS;{%gR={3i=_cfACP zKpjU?~frD7?y%t{IQ?Yy&x~I--k-_L2^n;ilyGg3oIt<>p)du zg(cS~9+#O#1<0o~FJ@(DBNMop(1<6BR|tVSY$#6LHOr>_<>782kHxEU(NE<}{>}J} za=-f)2E_#9V#(&)A^9``9Cs5P^s#uokTl($;phiT5UaSsNm1;C7_Rt=)iXf{+*EQpoK^W99 zDIPFq|mXU{OQ!s2d>}z#OU9h!vR4 z9e~^fy`LdWw#EMY#00OlnW8_i&H3K36Ca5X4QS~3=2p#30=~=k&hRaKGVC%tRd>08ZCxs%$obEV7Z(~LB$X}@4`Wfi|enk}o1D*V{28(Yc5 z5^Z+z+4w}SzpYuW(Ic~V(*FcG90Qp}cNuz2!2sI7|KwnDXl}ZGfd*}M(;IW8y!?Dd zb?>uh$v?)5`n=T47{JUXQn>>I189ABpa{sAiMvj}PY`p6SvJjs+DbpW>H85VaTbWH z%#U6aotyl6Iu?O$#AvD^_Yt9+&!4qG2hMS{n5y(;kl9LEJ3#i*)`^$&4CHzV+?rMT z1|C=K;aezMW5)mo!!i0j_3FvHZ(QuL^l<3)y>jqUdAzbjfR@0{e~O>>a2}T+5kUse1Uxi z!Kk$I9S@_8DE3}3fktel;O$;+Ks)DfT3WjN1fS&;nnzanPIoZvZRKj^*be|%2;Fq2 zr^uZSIeT&me*@3(8>_~4cEJn+m;2u|N><>a5xP9`=RFB0vr)21C~K|Pmu{AEo14Kj z2(VEt-UV$*8iwap%#xTGx=;xi5(CcneOPl;Qwi*pp3)22NWNR1ienXf$ zo*@ywI(d5j=95`m+7a-4nAFKhkZL;Di`XQS(65@8gRc{U7D#Gt2|JAH7hd2$dx~-F zp(Xupdb7IXU-(X)6Ky_HtV*gn&D>6fN zb8&ergMIf&NWS;r;?tm;=**C9+Emj_=}NkWV{z>IFTFI~+#OGzJh?5T6IASf(v@({ zV1PbMM*6F9WE#LGGO51^8VoT%2ncWgwVW)eegj3jSNjklv@WE`hu_N0%~dhO|3L-y zFC<~x-1OXG-M49Fq2Ea?^vbw6IG!qRnm`}k0eiM~u2iD`!@6zD)9pK0Bq{rUa|EFf zx!sxOXmlv`Q!137_ZTYn$WId03g1|)9*Z<&yLRmYM7XVI z{fsqx@9+GNyC)bGtI?h2f{tcDstxIvK(bHo-7Tcx4<9A)+1ZYGCOFN}BqdzzG4MiTm6k8i1@vsT76UW)%+Bk? zJ3Bfy`c&l(>~KgW(N|^erMqTEXy8S}cEbe7o@cbsX(jC|{bQle_`3B= zPJY*W!scQJzO#tCK}0mtsoIsL4NGwPH^YI{%X+blR1XpJ!y?TVbPgHNhrs+Q9!k_F z^dcrgl1W~}ZTy^L9oaOl%py$pozL${^&|O|Tk-MnWpkJ16cxi$81jF&YNn}6cT2=x zZv#Voc;OgW$Ky2;VkxGdNcqK|zzAAPAai=yIS|15QtOc_2)d~y@B&1>O1RFZH!YynpCsbykW>kd z`L5WlnfT_MjvH&d1_~ZMV!$>5+nR3sylW-a$Z+mM6-ML{j(~Mf4cd=xUdQ2W)(M}& zYh_$mAy$MhBl_h<#L$i?A`VA6N>r>yn%2Y3>2D9Du)HWa!SlelYF>u@)u@3`Twz`-8aFM^UF zcb}uFB>h;d^~|VN&3Jzi;-q!!SN;m@3TW>oI=rnRlmwr#Pl5)=E{awKp7;w2OTe*8 z8uRWh@Q4fBP3D>=dz<^5XI!z@FW8MC3pgTq&g0Q#bECT$7@p6>F2K#8WZ>T+Zc4AHl-V&mYnu}I!5zL>woPZV=}q^OWnGP3 zNc`22)`q1SC)`{1-8ok~uw1xJDY)2e?!`~1wPhy|TVl`y!N>q$nJ&>D)w_$#D1S$H z2i*1Az9kTr&fSHOB7?%h!hSZwIS=9FCLJ|4H@^cuLWnIDK;i^vBDBK6x44WFSeKf~ z^Sc(Uu?~OMi_@!5-jd={gWT^#OaAC5YxltSn&yT#U%!5RcC>9<$?x?O#QZ|ZTlG*u zm5P$|%r6^?S@w8oZ5EAUA?!!3LYK^1(uJ%#o)LKg?G$igG9O0AmOW5srjp43m_a_!7w{6Q7lKoW2kKX0ZbR7A# z0Up3f!bFQlf~l6HXy^{~l=|knNZxoVo?0o_rV^&E5_s$M`rck|Am@R85cztTRvPn5 zu&Jg-+qiw|8tO}`VL$nzYU-QMBhL^lxx59pqZ}W4a`uNBOs0dPqT0@nw^LFI<`*r= zd6uNAzhF*JGzu^&qmRBteR%k9J{&t=4e{wPC&h|ZFvYX?@6X6xvhbqE;s!ib-t!rd z{E$WsfZtA-tI$Dc5QK*B!Zk(*&B93i3;OnZh*MLLbboBOAcQH@;3TZYCvZH7*ke0Q z#5OULWrzERuiakBm{gS-#yqCMv?Q|2$-iM?$$<4LXk;vld}tj!spw>JzP{eQ*l$Ve z@QeR#dRP&z-Ax_oE!e4%emjm!V#a+8v6Tyh@7*L)kh* z#AXHtNYS*2o#^@7uP%jjK+7~ERED#W<_f*IVZ?57NH(B*KNyLK>ogmo${_Qe0LNo; za`J|c&qm_ChrU&{L^#AeUnqIH8aJ=j-y;c|XyruKNM*-tP+BhKs9y6_Z zpl5q0)5g{|^U4>oyn=#y$$A}a4yia%hqDnHaE?7Jh>Xc9_H;%%-*6%wxe@aMiCrr_o+F%tL9k?vu^&rR1ccUGTX%CL zKviIzU@?_LP67=q2;YRLnX^z7Eg02oy5D|CE(wJfQ25)}^^E85wpaSy->M&lY`*{S zD;6&9!Z9@~QIeN02di|Dcx~VrdwJA8;i*`5%FvKRUwLoGWb9uIP<{44JKE=b9v+mf z*h+-5z9&~8cjJv~@-5df+xYUuO{T$H zHKu#Pv8h_pLaMS5mt|P)7M@$u=w38MN6EUBgC@Ea{u;kUPDHuOW3t{j zlT&K19brGb>R5SIe{XP7pJ{?7d@MQEjjqZK?F)N?{5Jq@e$qG8&-gAo6=7osY}_BW zm0tb+;VLGyULqPu02KmY1H=+ggD7;&>e^e(8~Op}N5J|Y&nYgsde|#G6c}9zUU_wu zfzR0Y+ZsH1#9DHGE&!`%%{X0nZWs)KY3m_0Zp_~ac|$e&kAI6uri}k-*vX(W&b_sFa*~0HITztQ*?&sA3kYEQm0_rhlpjMQVW69ydLX*B z5K!y>sqX6P>UxbD2(*VhJUl306(mepD-t_7Jw0reqc5o$53sUsPFxfKt$m!qJ z0Bb<^;d?hDdB}-A5&6>Tr4pn2)PUc&qUZ7RkC|FGvTPu-vMf5&-pJ$nf_2D6r=+lp zIsc4O;;CanXe28eaP8EoQ(4)GJta{Cx$&WrXItKNVJl+r?WjpK?x7Jl19^uwedC+b z0*515D6ee>_kBDzS3HIm>(No=E;A%nyJy-u?OAd3qs&v%u~!WiQ{%buf$jheh(FgY zvLp1vs;LmkcTK_~A|gcWE=tsa<0BR?u~lt?v-dhP^Pvr7VQym4LZ`kdTlJEWc(KUS z8iu(wI@stZc-6{;97mLEGt<1T{Ay0QS7pM+*_t7%s-`vtJtL`Jtz;iG!WMXNn5=zm z#Z~J6n;D$!_mpOql$12(0c)$WmJZBE-oZpmt6{-O0*6U+dL5W@%l7R;FVb*FYUj-n zc?!IutAYT+USjb-B;w&iqt-KyrwN3KR12dNTF@B`UxyqoH7vwoeJv{_@;MR?#>kFk z&0>OB+Kd$_6xnUquY0(b8Z-4lq2_aS#<4?O3RB&}Lrd*A)etC}^U{~2p2BR%P23OcJpQ7-~mzJ7jk&4KS;VI#W~xktseBk9n^{RlL)c?a`@ z^uT-s#?qHsgLz8x2Dy27cnp1t5TQG~jlVP3&QLPv>15NMuzVO89u`{}Dk>^AA`#NG z?0RKWd(5;O8+LAJNi>P5`C!m$@vyG?GNZ{qGOhaL|7e87Qb!#}BLYhdn5`tqH$Ak& zE|e=9O31eC4h%=|#YbgnIs{Db+ZqzylwQ&NB=Ehy!u2H|YRZ;>$hYE#4oE&+n<$wn z?OFM%r=3V-c0v3N@13#OgPk&hR9+yj_ z&sv!XKaz1JoOq6ir^&Q78nD^!Av{6jvN{VLBVzz^rRSQtG+)`wzMIm&6wtR03=^cY z)DM=OhqWxWbd0xS0xheC@=xhoAo-os5I*kfW*E^i+({_W&4Ch)2KkO#Ar*xPB3IuT zKx^018p%{Zg}K4$8Xq04bEqnsOxs}Zorz|bj6?dIL+V~ZbGYYh6A;Aq`yR_nqG%g1-KU+3gh68jW#M42Jdeidpe++R6K=r;Q>B~wyu z5LT>R!8$}I1>G`|en(`L1V3^^A`VGe1E(R61kL}wpPjujL40(D;gk53QY0tfiXJud&kZQ?G;c zybXIdfYoratKZ|NCy$$Wt$44^MV7e=#w7{G)xSxNx85$%NLq??l3D1P@Vbc^8wiX< zuv)GQ13ZW^X$^LDmY`?#t>uNyfqk1nF&?C>O;%80Ia=t2xZ6SA`xVzM3pKHfslX~# z1@{quRImoQZ-&w2-Vj&avXIyMPtyp3x4aUw@K; zbteHBPe;!;rk5_m%&sFHBW^MRfAo40PdW}_ACc~mn`aQ~|JrMb=Suvs9<{LQkZ7Bu z68|1$#ir|YE)p=Dh__75FFi*{Hp`z#!17z%GQ{p{@yf3gK4a~I+TW38A`j=YZa+d~ z(E9V=5FwK3-YCxD;WOBaQz`NClXzV+PA<<`%jqDufV3%=dueHijZ17F zv0GyAu?!IUTKY#+>mWR)5KZK9Jv==fQi{NJA*+o6REy^2n<^?THVMPUN)_2Z!> z;hx{Isf!}`Zs|@D0?Sw{Vjd;PH2ZCV7+ttRA%fxZDVaZS*&sp%%|%E~8}?iHURfXr zMjwaSrSaT8N{ag~ll?c3T>64+BHpJXT6Rs(eR^(HNA*G$ zGjnvU`ZVhvcg*P4DQ=ZGDerwyOdlj8-DrX}OGGujy-OD}&%)v$)>P$mu3mjzwH_Sy z2Qe{D(4>s9snv{S!he4vG}`JT=|o^wJSW7gj^l7*uOVV)OgYAvRKEZEb@W6rT;yX{ z`%Croe-O+2P&YyZPiKt%Mw|nbveh?T^*eFF7O-zi^ut_Dsbazzf*5f4)dYZ)R)KT6 zQBVUJNDiTvvW>&fB;vo*=oNST{K$$|r_XCIqZWt=H5E^t?u4|VEgq8q@w{^hiy@uw zRO54Ugz#7m9=Y>!a%IVKp_E+aPy(|yA0lKjA{h|PDK)V~DnrD){%tAS{?k$%vE?X9 z%p5r+31e;M*cD&S{z$Y8e~KqPkOMWk&2S(P3t>^%f>hHBj@x@#7Ed<0^NNZx@HnWU z{|I}7pNGYAEeW~6WAW>qtXa5r zR+P=nob2o}Q=)=ck3vj3BHaZQa58Q;XbX!7mm)5X+P#@;qmA zZ1G4b5mbnZD{ec%`Q@8obaJ^T)OQP}8_(ffbHA8?mPLppiR=8ion2i6F?PidQ_#^n zh|E^g5Ror+_Wq$ii|8`p@q}bU!%qOf0wI5-Vs`t^ok8q(RDvogHoe!Ic{STcgcWDr zU@!hvn=o>n+I?B}gJ+2ZjLJF|X=+GGHV}Qr56}N-_xRILJvv=(Xn33R|GLVajdpCHftl(WKuMZj2t5^O{=fRQ6| z1x{nMV63-s*Q`EVC-gtKG5GA8CNHbregaPj*aCa*D34K(Q7=4thoJr7nq2OQ`I#&E zKszA|L<+vIL<$Cvt-5Q~B3wISxkpY1i~iXExrKCvWn3V^6rZa-Gj53D%i@eviWZa` ze8H654L-RI9=%)_W+kyE<6-si2Og?ng04wCumFofud#C1*;1B6?B$`a5ZJjV&;Ilu zy$KKEG8yIA=pyiF75_gWI7_Zho+j3cPT!_%67m21@#79S%=}jD+1LG{kzr!<#5Ck0 zR7-3I56i1Tbw7=oFfD5(?gz~?ihCm?77ekYyD@_i5t4Xq*}{Ph5$-xBKseswMN+8q zG71WT-vscXR~Jft3*|{@SJy!c|Ac1(#NFJbJ-BUd@MA^A>4_I)p3Ad0iS$zp;h(@s z{6RUOSK!QphcLtz$;W06YXiq=ul-(X^nQ(VA|6wSczQ4bKsC*HILviIx&(JwI84yTb!Lcl1R`*U>zp<5mA*?DV##)_8BWg3RA3iW z3WoI=m14q%g?5k?{s&+i;KiQuS%(-?FvJP(7{T`#<&I+6nf1bf6I ztbf8{uoRl`*osQ3ixfL{R1wM{YCRFhGNAf^-=m+bL@%M3BBJSV!e{iupe>W_IL8Du+fAJ?%ZOh!!M3Sldd}Q2@6-l^E`(2Nx5a*K;kKV zlAg=NBR()#k@2j?TzQR0lf_DT8}L7OM#7XHSuG>OLo7%_VGxnP+L5-hvT|MqX364o z0v2NcHN6c7htS3e^9lhrY|N8PdawL^nNiH2`m0fLg%aQjp$JFkF2L4L#IhHbHn6C9 zBRH4}Pq>-{>yDN5WJIeE-=v4sB`qqGcrxDh^yvMNBxZK3KQ2 zk=F3v#ja3nr3~%Ks^b;U)R@AO>D! zp$N5CphXfV3lBd_!tM}uo_0dqrV=cASX1UlT<2qgQPrC%^oE$omqvoY$`HY&*7jOp-!E6CoOC z*ik4cg}570Nt%S_$u>4enkb@@sKnho8ydJ%5t;{;28}As(|4}+eunowzQ5r6p5r~< zy`N{>b>F|=b**)-bDis4`peQXGMbQ3^WeAE7hA+=`j|3veh;`Pv>RVx-i}yhwAw`K zAuYoMm%Fs7@hO^Eb%F|(nWK*>!yEItuAshLO}7oP4Q6LEP>L1?K3SF_Q znsvBE0>BjOx#VcN5CC&O*Z|T7)2JP^eDL>JNv1LgwMv>UeP{)s=3dUO5L1}`{O%5u zrO3=)DC+^8SAl?k=W^=^5-$;F_%`s{VQB8ded`67ZjU8K8xgVq@R7zT3xV$An4X z3hf44U1tHTi(p{DgNlN+@PNoUK+OLheY@=$Z8&VrhNF!`t#V8v%WO(!W zB9Ed!_zf956|OL49UWIZoRfNGr^YX0uwpUtmR}?iTo2yePOv~evlgs+1v4`+3n+T41d}jgMnk+b@O1aXx@)fhQqDu?bbrJZ zWHCmBzkYMTNNvKVr-p~-zP|uzW0sl*%=mVRThzEAjDGl*{+EP2JJpo1`4k)z0d|J9 z7V}ri%ms(g>Rtvgr2tmjXxT-Yr`q;H=??lTr-(=6kzJ=W4^MMCaK3U5-j~-2a7VwA$c?pY9yBLIK%$~ zWer_Dk>~@<5R{-`KdW=>@zz3+bf_Apk7vKui4)%tz29P7-uqbw#8>nVP~^~{=^xXQ zo>0u#Sn5H*i8Ke*6CY9MmlGKmi=()wpF8YHu!D6X-nH z*L=`BCJd2T+teOI8IuczX3D6G|S6`n-~qQ zcE~I@F!bg`5Zf~HS-RB%?ES+7O)%cotIv$ovi&{s#25bN&vKRn*#^Y5gdhFKr}8i__-rbs@Uybity`WDZiGa?QrKJI zl&PsF+UnKOU+tJ=S^;_FH8kDjM@XN9uR)H!`^pz&&M_N7t%Es+fQSnho;i*5Nm{or zN4;SD)3OXB9`r;2ATMUhd8E^GBG`XvAfRw=1hmg5KR%SNRGVA!6c5(`gFXb@(PBKL z5w#A~;U^pjvx*lpKy{K0?aI_Ac_{8km=~eB8^bqLMZxjlq0U)G9XTF-0>gD@pdYd8 z4Q{vB!kWTTm_bqF?V-VT;WlTaD35+9?nyF;7$p|%Ozx`~xu+`SYTwhs(FI)FU+<70 z+fvM$4r8{!_)A(T@E2WOgvA9c<1=S&0E3=_zrn2Xr{3t5P$$U@&a%+~YwBBm{Gdkk zJB7`{PJQi?ANrnq4S7^qYn*~*@K)%|YaYG?3EgakVjHGn0-V}Wx+qG>ZvbtR=7SSG z_t4F21l4#J7s7{y82B1eyI;?>FT*5-b;#IOW=_{VmhS z?@|1F-~SxH|1`ifsHd}PlLK0Fr@+PU3o^z0I|hE2f$u2Gc;kUSE)+gDK~9kM!M^gl z^+<0zVQkl?OYmnpjeieffLSO!;Gqoro9h4s2B-jcZF+$*>qo#G71_?a(d{$w zTg#8ZEwY}#{HPRlwoqzB;A=sgf79FhkWZlmX{yPQmIAHo)TG0vGhi)OlZgrmQO-BK zk+$r~Xh%>E>?#CU>u_FWFJBJVC!$@@md4CJX~KL6G`xN-Z^ecAGU%`tQ_{s(3J0P| zE7L>Z1)6#b#VLBEmveIDs3@iDhWw<6Oil_XptdytIB8`*wt0BjfO0db#|1lt=f&?I zgCwMikI)!pUj)q$Zv*eu!OWTD7GxHSlQInHwmN36@@6am2}e$n6e;0I$G3IsVze_Q zl&a}H($fT1-F3ABW^7wQHGp6O&;IkT*8~0{jHRF}3t%cn61w}5+UMLC9Stu?6U+y@2@NzI|g8OURt)d7&md@;jOZxyUGwsHfpBEPKDmeTZ01~}U>NX-45NYGl$AQBD7@3bFdKIYl6neft zfBpIr5)ZzN1Topejq|d~+lhdqKLRu?Xc3f-#6lX1;pQNHx_l5j2(wx{JH3#}snLZ0 zyZQPrezc$|o8k1ui&ZE;)_8=US&?A82sq`~+EGh$vVj2Q&;X37fbw&B+)9{8-+*Jq zNlXEucT8^Zll@^+-{no`1%gRDh4hNf(F*$aAl?vQwStjG;dl}(3dullBknfc-2R;d znM4dVCiSqVCY_1$a+n!}whl3aWG0M5`<44Z+(DPxerIj@ZqvSwf0L#3UV5$wzbIHwC?wRNZM9cj~;K|>_*O-sF6rqCM~@h!4O`Ke*stGa2*wRz*1!o zTl^a|Sp~SBz2Dz2BLKc-%R<}$P+Np}q0XQwt2@%}p>uB_rR$`gpFy&~MT>5dkEgUW zJT(>|mgdR-d7T00jw@HM(sP0F_fq__0|yRV2JMV^2^quhYu2p6$(aeCn|CI(GQ{oJn>#tGwkaaiiGv;M~LjxFxO@VG#YwMG&hkXV3TR78S zgK9M*Vm(={zC5jnBb)c|VVA9qcEogB@f^5KULm3B|{QyEZ$9v7oPc15a0y)}s)cTcdo1okXhVVI_jIw-!$qbGDFw zT)uN>8E(T>oQ18gZZd08U=K^^jsF(yndFbh2kSAIw3+&;^~c*pGaxeoew*8Hm# z#ai>9>_iYrDA#}1f+gzR4OR#Kb;{|l_kO+<`S1V!zyALJ*{v%o@Nplh{ezy&|NTGV zwsVgEf3WpsA8KR2azL{F3?=`AklpaRcA&y&wY^iPBpu~8Sppq|Q9&;Le}4u;!SFPD zLruZt4Van;v_kT{j{69@B;Rcl`i{!4tX?{;#Z`Z8_+KobLpAa=IWdFd{2k>YHg`Z! z!H(hShR)9UIK{O3qOPuPr1t3lcK3`w5%Xbd?TIl) zN|HqXppg4*ox+ynd($p17vADAU`YvoJmgC)wh7T}bLY&tJ)ue3Z`#igp7J7oWB42V z*!`tySmGd^6Y|e>>Jj+&{D%62-IxbO+r>lhNYu*ZFP?ZW{*B-_Ix>fWg25Sl{5|e} zE-jgUFxwye*TNN#Ku~H12oNAUKMtqCa+mlD$}I>kAT-WjxUdHeJ^Tk=N$bhr8LIJp z=gvJ7nL-=*CmcIo)z{0VH~R zC$zOYHVZ=!4Ir40!pa*gsr(A0DZFu_R1=nY9NL{i{AdSMVv zBUu`kZh!mc4P||@3x_1*@1`;-_TXVP(89Ur8~#)(EUn>=+zWOS)!bWr3`$L4)=r%n zw2(7U`r|JU*|f6-hHiT?3P(&Qd1<3Z1vBE}o{cT(DE_I00z)njVhrL`T3JA|%m86% zV{D8U>qvoq>js7mPz}05Zd@c?$Cf7p`oE9&l-ji zABwy01AubN$z=>6;dXHFAK*_>Ez+E2T50RdKM8xF-40Tk4=4#8I6_G%ZXjMzMl|Z<<{qV1u^;`uQ(h2oiw-I?!*SwF2#~oS7X3;o^|6y+V_aHU}ZC6PE@Yf*N}U z#))s64#;Bl0Mvm0_>e;v*U;VVgR-9VxWKh(ni42m`{7)Le|Q*-3jB&zf;l{A*El0)w1AW%^ntlzMK9akIUM86YlP9=rI#Pvg9SfS~-35Gg1 zFYh(lfcQ0J*)o6E8t_8KVfA2?j`Tz{&J;YPNtJr2=JF*!^gkB&h z_$6={!$=uDS86`FhOwWL4hjx-gAr5%XjwV|i2k_nw7Up1aSSUd91Kame@$hN?t!bXHyID855)%}eQd~(m54rGAEBYS^%vi}KOM6&hUw`Tx^ z(T_kwfgI?8xDU5I4r zAv^l-bPYh)(9Qr9*ur&|0B7`Bsn3X;1J3&`#=3D4qL0O_!0$SFKH~yiNO+81Lk0K^ zLAtVF+GZW|IY=q&D3b*)6 zwOa5D+l8imebLfAe|&=Dgi;2nt|k~zVp2l@w{{r}&Ifte+`yXOZ-&bIIBBG5YRq6V zHVFPHq>@N=8eyu}N5HhhuW!0%Bn3mQTSN>>Jz_nVgwUOkxoU@ex=c zRuOH}D&ai#-8=p$RGLza#_*n2!bZGDPQ9%9m5%;zX^p(1J|wN}cu9XpOTNs+&oR{wl%aTTUH zc(x5;7@RwX2ymifXeZ(&QYk0&mEx&cRWScQ0Q*yUG975mPJa1v2?zyhzRfCM0{pGd)R+p9W;>C-P>7qCnf}kQI z5*~oMa)X-koJ_eEgF#jTA+bFWsB`E$4+M~ZNpVgD##3mL{6xYTZ=i^+sl2)qhhY## zVzp>E6PJy`Pg)VA;6Qhd$h}Fu8`)kwWB0CP8%*Mx?kd;?&YD;tJ^_IPraTzjY=rN~ z5{wI%rY}MPm_z zwR$+Z1-s()nR*;TM;C+jS%#D)HTnSOf;OelwkQu5IlLjI%VP=AlfymXMGT?%WL5AyTfkl!S>w*l-U^Hv$31 zJb({`+FP;OO#z=K57jd{q`bCn{+e~-$k5AlpRZ^z^kRvSCwQf^6SAY-PtO<{y5ZN% zg{JlX#DTz2cbs z%tZD2;*Vpy7>GV2xJcCScvD)cVaknezc5-00#K8CpPM<2Y}D0X{|6H7gwkr&)LvU8 zsF7MY5e?3Xf|>JDIF)=H3;R{I@zO?qem6|w9B$=5bVw8z5S@P!y&Ms?8&YT3x5c^zS#D zb{In$M*fZSpPAwfJ8atBaRI!{{bediSx9HOfVAEZbhcbuj2Wv>yRDSVO>y#e6-o>S zFWuvYS?|JPQ#J*VV|n0Fy(f#K0yzsBA!IhXvj-gG{HjGb__VK~%m>)($A>58joL6t zBMV{q`ZR!6skthEBTK05QmkZCr3X?`OyCcN<0oGu`3|VImv~N3VD~~JDYS4U1n#Hn zGVPgt&_FI2)vl_p_5?>KR2S5af+ZKdujPmIj=3Jgasa4niXM(84(!D%{K{qb?8g9* zbVpo|?t(f<2$!Irsxi(3^pXdpVAw3@I;r!?TB)oM>hp&=QdfuSPP+E*(;JdNdJkf+ z;qGH5BrGEZYJ0}4S+hvI*W5Og2qDIO>*Zg*Wg#n5ry{Ff?KqFUz4>F|#fG4s-c(g- zSxr#<1-s?%9|q@iVt@0DFxy7QRLCMHoeciSP9F~6KPQDTH%?Zt@&nBS);g!=( zAp`z1AYwbd$L29Th_$Ylk<(2cs6>u1i5*!u{QeQ*tuU|peVcusA$m5ZUu>Dl$(l|J zg357Md2XXE_g3u^~q*#ljmoL#P-5 z#tjFi0nNIiyujxBWB>(~-=C;D8l`d9rbSR>XzolLHCS^m7A*WXV>yQbe&npyonLUB z$b1T%xlDNn>%#EQQ)sySfv>ky(h?)HlaphXm~UMQqoKdbqWs%k#M zXfHZqv|$pXR2#NzDQ?S5CwmKgB@%@4XCbJPMFc@=;ek~`@jiKI*PMoc!7mX5&ra~r zLdH3Klx!l&)>XNbW{HtMDqq}~NsDQ~v0}h-#nnDI?emZjZ8|h!{B9*VszVWaJ23Ff z`AIz5a8S?tcuXy|@gN|E`U_XKHG1`tm2xi9*gNG2w+x{wy+8g!seKZexY7jkP-m-? z{VEyD0T`S5Lni47E8D?_IsWTWKxdLohRMG5`GwPKq> z{P6loUX2v9Lx6M`A72{m#{=ZnwLKMp);$=G(=f)}7o>jpd7R65?|y1);JK)WQdnD) z4XthyDGvNRow^< zDOjO-tjC7ESua*_v?8Sz?G^OJ^MPl)w=^La$)iSIynMOg`8tFQm3uqppl%np%UN~g zxT=p78?N+hngc|cbT|5#{q*${6WfNNCMWn=dGtJDdxfwu-Hqu<`@6d4;8^jW2CsuH z`>a{4QW*5i5KFi+BK-)_|mh5!62<`trUb zg!9>^bT=JOmpliqrwWcHQ?vhU-~2atHm0%JAxqB$N#b(mWh7PHJ*M}0I$J@@mGgIfRB%m3)~EADw8imB@0M* zy6O@GlS>Fo5ts!gV=xr!Nc-OC#8ESXPM4kQnfG&5>@sp3T0nLgf%a&b*V5SK{h$!R$ zU>WoTb%SAx;@9MUZ+ao^o!|2?SH1$w(Oytj4QQJ|dqFBgvOi=OgEn~%6kN1(blb^S zY!IFajwOIQdxZE+)iXVAsDYuWI0+Sx-}`c4@s(m{=7W!k9uD?GuLxkt3fyj`F}kq( z(WV0OI`uIRJI|8vS(i{C9qm%(~64O5RY@j&qn~NNY@_Jn_EO7|7 zf)MGychSdm(WR~`4yXkI!o|RFkR-vZE|YQLgAq2%LbgZD2W|$mYAK!H`0y!EebDL< zB-UbC7ovf|YAUt+;>>_8^};D@E#fR*kCL=9CjijACfixN&mK>1KyNo@UmerRvPC*A zy?QK3Ig9T``_V{hhb9)-H`@1#RwAvqtH^Q9G1E!IJOCPQdz&QHb+J?t1^LE#!^{^i z{savBh`~G%EU(a#dKEHT2XuJ@DquS8`OZg!QTnU^pe}LIhy6NqD^g>!ihm13V|Q+i zd#Iz0e0gZ;wN`Kz5W9q<#YGnr4;=puzefCB{$7wVG?7ab3}qI-wYdnJhC}6yaxpN` zxrmzlId>6Vz1GO_f?~S@Ko&0ktP*YtgbwNDuZ*FWM!t`)Z*wM&R(N_)Tyz_W@n)u#f}X zSS{;xA`nA>CJ?Xl0?pCdK4_T1^HC&&I{F*(GY*)S`i%0lE;Q@V@_v-xu>=|r0GidP zOD+cmaiYfnVvE*}H2~g7LeAPL`*aKY4?Q#0v5FLQ52=3j`?H>CQJlEw!;gN$0$`#< zQT#8$75Q=^A2UPOfpL`ryW88T@cIHLB;<}4Sbz$B1N9H7Ck|%IuL~kSOPu|0NQCO? z!WTZkpoUg@qi38@@DYtwG>H|##bWb0Ol1T*49L?Sg?SV5qE>-CdZLvR%YU|5P7T_} zK_)bE!CC#5OJTiJ{{rXAx8fNHF9^`anYF8W{(Enh+gepf~{-bR&};`*@u+KD_)*raky&97#JxX)2{mmW|vJ zC_6ax%Q+6Jz^YcC_?e!1)rpmkW0LVOccD6_EECQ4a6}2FjV3-;udNJ7IxLs)`<{}N zL~CxAfUMrt4=(U1SV(!zDSSbWuM-eco7PaduK7i+Hpf+;bhmG?V-5;NNJ>5$oI~Jy zQ&y(ww;wy}ur^KplWn2!h_y&IFp8an18Wes%rUXWex&d5&%wdKuW%6+SFE?jtMGGz zOaW_==CHlN<#ifq4$>-_gOE+2wKm6E1wuef54Nm_svyc=j7gZh zc+g&}${uJ^MUmo)cwoygp5td%!dX`X3>daa&r+m#!68 zCJwm_NZ=4J20gnNbk1TRy(3@BFw0BPifp5+X-P6%xRKD9J?(HugbqJ8dJ?2N?SuC~ z*r44n?GrBT)?9BZ>N6asehu1VyNMp! zkhd*!97vQ%!D5%5hK6+*v3QuY038HQgv8Pcec=Jx2GBP)I_d#2h-p?2SU|AYr@{3R zf#X-6m#6rtM_ewxNu~C9YckT$fvR@gZqNp>Aw>jY_GFn=ZuhRHjhIN8?~^g*cp6GQ zWX8Rgy14g5UpB(4b1`D)*3`H(tWssre$(Vd#F9X1kc&rc-BN-mfeLm7S~Z-w?4XBF zv<{9~iJd})>Y%-pO>OxK3Wu1zxzKo}Xw!fG8 zbpgqo=9GA2pp!g6qC_8K6sdU#)Q?}*D(4$fJ5NDDalyZ*BL$A5a(TD0cLr?@eAg3Y zbR_KLy@9^IzEY{zI(JY*vRKfcF2gL^QGZvwdFJ~QY?K3kwQN3%eRJ{HLhA;}ZhKJ} zhIau!HWb#Q37TgvihuelWX|>UjaAuQ|8^v=b3zPTwaFQ=kMW5=0jQfaI*dtVU!ij; zkCthS4t5Dw*45VP;!_GdJE#l3sLuS$%)Ow-E~%;>1(xSS-*STy6ek|Y#w)*`Ig)PG zbR+*^5z1@9d*6qWQS4>aD}CNJD|1dec6)r$CIeuiY}H$@;mk6&ro!ni6)OZAdsg!B zY^J$TLG&pOA!l@aNCBI*^aQ1Jp+cOvpt{n`y`0uSG zQJ1fDxqzDnf+oN`1gix>bF3fRy>e+{jBIaLSCHRBgLT&^7Kp^d<&~y<>oW1;0-6!> zpEnN9do+@n3{vmD{fOjVX4|X5bKC$4Cx;Jf3_w z+f`dw?sqhUi4}M?hJL-@pB}OOM>KPA3xu;;h1TtuJr_x1pQY@qo~+lC8(l|3fQj?0 zqY)49iCne>N#W4{RPARxOfWkiNh^av@W8b8{p#h3-?CzZK&=wR@w)3>d~ayfpSNN! z0`+SDDnn5wl(7(63aEQ@*%pHVKE7Of6k12dc?IBl4(B1@g!fhTP@-M^?Je5cYcq^@ zxQyFnjr8^Jdj5jFXR}V~;j%g$c5WS2d=$e1kL20K_=X2HTM=En&gJ+y^4m8+wBYZ& z36rH(@Wr}XI2TiMBUyd@da+N(PM8;*LqCOx$^}Bgc!p>l7mf-aDt}c)<7KE2lr?C< zLS64)FJC)iE2~3EOchp)xzmhHyNGUjqR2z_kSmKbk1uvD1k7zjX7p%ZBX-71WD1oh zNArehDojZE-!A$7^DuX^aNRYL+RvZG)@T_7W7>Wmac|9cpZ2UVMR02gEI)3AiYo_LyY^Cv2Iuh4dw zQfPeRCF+5+tcJREm9)^5hePBnl&(LDdQjLPe!fM#wM>g5#l!nC6dLa*yOk>|D}RiS zqgW)g$WP0MAwBkP+=Zz20%NGD85za^wvQXYa_o$Knf47>AC9Is_{|}J4#ryh4Lyu3-ZMe^G@q?G&IPM{?j6$`Ra* z-B*O&GVVjTNW&~jT4;5{mlKnyS5Za?tSS))$ACVQe@;Zlk!U>x(^&N zlm-qRMcW}>r8OLW;;xfUh{Xb1)?>-pA}A5qq0Y~$PESuijVU<4M4%RuzkG9;mjFkf z*Ctic zc*2=OjA{bqStBsY!BEq0AdjzhfMjQyM!%s|wzOSC-oDRK`pJ93`8YB^YKtN12UrM- zgq8Y|B{SG)`vXZ3vzh_oX22u=RwPi{8VG^EgbZP(^G+rtvl}rY6Ww-C#uLx7b9~j} z#k8+I&&cq7Oh@XQGnfHjPNI?b>LGdh@?|`rUF;A+?68Qh+k_G8)A$3kzTQ==m+{M_ zi8c|v%oBc58c*qbqgz__tq~=P5|;aW&Nmgde4R3a+*N` zQt$IqGOEJ>UEyrc);ZG`K0NgFzbCMpJ%HJ~e@NYZ7{p0rHj>&Z|5lJ;70Koyq?b7& zE`Kl#^;qFXZ}^|hLZ3@KI`qmzBoDYR2Q+qJbNW_DiWsA~P8`jP=FEW+aNqVwmm|dA z$zrEKUNsgkQ3u#B5yTs|wyPNb$FyGYPG$lWmZ~c%)?K@GD0TaF(2VaL4Q3Aclivb= z5Af5L>H=j1J{o`vh{UF@o%_uTpJKWQV;1s$Ef^@2cv0|$Cc6P?_Vq%`oN^QY;qSud z+13BYx~&-ARDxOYRjR?87feiHvQac-II_di9tOae64mmBa*Pstq}|Jj%H8nOHt|2&aT zkMUereZD!L>-5K&yo4+4!$3=;aBcfQjWlUc@MBqUJYhEy+KG>&#HM<-ZOQ;%l(Es> za%`?ab%D&aS2%v-3s*n9L}B?-N0G~XE9WiQ;hrD7a_169PrqORB^4MyxGz-@saSXJ z@XNn0Z#X@}f6Jj=2aH`^zgxN1Wfd}qy3JHc6RdvQh7by4NU z8{8J}+E!0~5Gx^-MU=mJhh5Hksg_`a#?&Ss;oh;7wzWGu2Czs&f*SOV zqx^4sTU?nkb%u?h7mJllE@BWE{qHZQJN#P*4jAE-#&|tHlB3R5+?pK+{@)57??J;8 zSh0NJ_PQbiUzz0GMK$gR>Pz@ehYXq4*)@45ni6AbZ0%xSX%wUg5zXs9PEn z@0;Y_w`E303N>X_nEAfczSy1Cx(~{P?azooL*6@}TnU7w7k)E_!zOph{z#tn!lNz~ zWjnA}qDy|3WoGX__JNIJ0an^1TdV4QA!%=K-+Oh<_}a&vn4?tN&HEc0TW9CBAvQL) z%iA#%RK-__^|-;k=?@&c)7}BHx9K-YRCf;@1dMS5)gZynbo4^VS*gTqbH6C5DTwPi?w?NrTk~3RGj(uAmfz64|cc ziw)N~Ee1>`jW&Fq$V5ZW>fO4Vk^bhUHwq|!0ZRvJ1b(-b0__=h^Fh6)m}Ftq-s>j8FTdMQZS^RpnIP9PcpSInEd4g!N{`Nocod85)clO;++LgEx^mCwaU3Ihi*X1v4mi*0uQeZIHn?mpPGVrPYoc0tygdIQu1 z5lA^)Nn_qJLp5UNfu$-r1sF)mXBgiPP`?kw|KkO#H+yn93&-#$-*Gz-a$7WHpi?Ex z(Wt?$e`-6<@iXw(COPpvnf>@J zhZ#rn&80%dMz8mnf`s3NM;+AbfKz8<_{I@``Sb3(l9KqgErySO{u%9OhjFFPEDr}h zMNA+i7x>N*>EWh4#4>-R*@o$!>vs?c>t>qrD;gRqztYF^KwbH*x z#k?N6Fe9x>tVanN7*+0o3+W+YVd)4RhADncC#;hgU&oRglY8BM{dEc?&gSiHS{m1L z3i0-61G~m|$9b;BpsIbvN~Fa)k`FfBQImB?C%yoQJVc`o?W&c#ZOrmtzv$DrJD5jd z3GnH`cJ33<^rO`zs=B)?G4ZY{tbh9?oBfY4yE{@i7B&QYF?n)7P%}Ach8v23gCVD$ z?g+D{!u^WpUEU$$xy2n%D!xCX$$BGJ3c`GRN3hg~ZVz?m0P9OsB7?yq1+u#BV4OY`N~yj9n33gr z`>9#C?a`?IG)u#TFphygZ>;x~-X3B7y}9`@Pkf4X=w>IGBcH|?_PPb;@jkn|Qxi*C zU(a46qFxKYECX2B@6YTn-W^}D^>~AU2A)!=eal@mj6;_WruJc{pi$MJ9rRYlJag*< zS$3l0AqJggIO@Ae&t{eNrXaM|wm$3VqDVof4|k<5o=Ll}UyKbjn1JId*R|D!d__W4 zvf&zCSvc!HdUx>(-S@|n$;ESoJXxumg9@C@5gzZW=SlOe#96IqFgJfOh{Gmi{*G4P zzO1&$Fw(Ka?jpfL91zc|3wH^`$aaFS2WR!!f3bYSz;T<|JeLkDOY$wlsMvXFd?P_4 zs3$#^sYM$K`Hd-$1}3QfEC7kd}U>}ayd`2?=Z{{?+rlDab3x23dX_7{8R zj5mu4UX`=V=t>?g&za-^RpSdx`M9X!ws-ifs8E;DAMUaMLMKwBiF z*yOa*cgb0Asz;J%^(~*=auMbHs#k;FK-rN290R&=o0K*Gb_#4s1SATKl>2c&t2>cp z0>uC9=FAyZoX;yQ*~E}7JYrPMNj5hUwU>_QICdqmQwTgXFPFzP`|*6k>k-^G+x!_LZ#vur zCR2yNL`AZz-&n}Aszhc;XJ40#c7uG(T^HYG>&$v}dbcqh?kMLlZ-4Cdg{9SIxyAiy z$2a;&OvOhQgrFgl%VS%!1X`ijjs_@HE@8mt zYEDyQ_3=9vST>^K)i}Vb?Y{lRsMX%=^g$0b)6>x;yoy9- zeDbj*&sV^H5Q`1w?wES#wKnB+9%iEG$e_MX&EO3nfYF*CEa%-A)KxxKhTv0U-jS zy-Uxc!#du`nd62Uv_!BO6`&WlMd~_>)P@^SJeZ(uwiEWKD`8UUj?<=yXd`yc7{!u_ z(e>Gf0yRq$mYxa!DSKPt+qNT>W=%_e;MAVc6&csHs7e&X#?z~@Rdx5mvkRvm$G208 zPQ{cCAfFW2aOngZV#7^qrdH;RevM~k$=!YCHP(n2&I850CpNAvA8~*RU3#=FLXr1Q z0;#+C!v54>T0c^38;$@dSU*+fvdnWdHfOleGq$Sso{C%t>(`4Larc8(JozbpD|Z*4 z$CSBfyCXWB7Pp%#S-%`r+U@jf>a*6rKeO9CI=IW5Nf}xBLZ%ap*C<*4!qs)I%k;pc9`&r|sZYc7`fcii5)P>MgMvPd`tZ`){Ktq03a z_PF{OL%l=LLM}wk)kW(ktweOMYl@DpA23I}EWskCu)(#yxNa{he%3)+a2kC_K_e>J z4(@2Gq+^b{^(7AK6((gUdLhHDa#LIg{cfnB_rVK>8dFwZ)b|@XPccO}tXGQ`nzsGx z92z=qJ~kYTAz=NW_cH#>VD(-U^~E-jHH07{qG%QfzV>U+z0{M^#W(%16V4|kVA(C%BfB|`-1YBdNJkI$;RH}g7nzVts|G`O+4zM1tR zZuuFxF;|q@g>4fq`pM?dfGCA*>qhq7Y%>Mkvf%R>0q(6Q|8%eF06QZ+RCDnYf(@U3 zJ%MiMc&rL|Y-#i-oUjluU@9h|o8Y1+tj^v1C^P<%aCKHiWV|y|SD#T+ARH^weG?3s zD11a7oH#FfdTb8Y@nrMLp5Fn}Dwq5dPUwTK9+x#Kjhx$d=)K|Cj)^yag#&g4K3k*s zRn+9{)9nI9^*LEnUCmFUON<z!hqg{x+P}>g)9Sl9e=k`;c(6zLH$Al>Ly}52zoNOTpUn@`Jjm^ zB|FxcfXzoHs5LfX5Nj#&gW#0%_LyU*Pz<`$o5}&92=()r_})RNc6^>Za1cTn%#Fa| zfutqT2M+{`lyIZ544qHaticb)DHB4gGvUXc%5=Ch zp``Iy%=4;qZavkqBj4|&HV0~-&Y2kWL??;NtCVXr8G=23-@zZIVP<7RxPF!?>-upz zMXAuytV3E!ZGRTi`1HvG4??^sC{KeNc!>|3c<~Bi?W5lu^c(`rVK&D0u=9_e8Mu%} zEw}bAR1)Tvb8%w-6=~Mo(h?;CUMcEmhww9(263T;((#Gvv3k7+9-}(RPfD^O(C9^3 zoeMjX#=mCsx?L=m+63`8!|Xov3;7gMA*k4Fvo9ZVpC=*`xOL`-)l8;pQ47y;ntNBc z{YS_3u%9{W1;=%PyNnpYV`wwVtE>w>YEeQ> z*>JdFNj7eY{MtAskecSfq;*~g_pn0xkOO~J&>k=L|0Fa~gEa$qpaJwUr z?9uT0CMY=yrxwi=y6+)dH-EL{)~=z!Ys*yKaPISf!nvk@t@E!>7%&S#|4$t_`N_Lm zx+Z6)q!6nNU~Pc};WgZ$Kefs@)M{U6a!{m(6Dx1q9u4SqvT#)Y z;xdTXF4$Y*`+eZY=S`d0jz~Tn^2Y9~TzuvS8a6F=jaQ6JaBf1~Zvn6dp4SEZK1IKgXHh&R7` zQG*Qr^6&#a={Oi|SYai0AZ~7@37!J$g4LsNKLMQ_L^_&0*N98tM`90t=Rh-uXKNvy z2n?6l{rpX}uELMxtnl`h=4PX;Y`=5JYj+?>j!N*fHTatoq zFkUu@H(_4;@3)8FtGO)2|F-|OHT|op_>Ko_ZIq7w%09+b(wfZaQl6p@6hyc7iGS9w z(t~hi%S;@S{q=FLN$c2keaRr8^Fb`~ zA_o_x*Om(5-t#o;nfcLr7n`h9nbjAmJY?$NN7-RlPZNGz$yph~+)VA}{N(undHsw3 zj>;+xFXR<+A$B;8OnRy-Zg8UD?-<}f)=kdp3b;;ns;bDdjgm zY^Z3y7wvklO%Lq&%;hi)V!s;Cxi@TT{K=ltV@m_f^;d)2lE$4BGy+3$kUMxPC1+wY zFLIqiRSn8@{5DR;*ZbQH_Ep8oYL6X?H4EEW*(;+_=W#D7GCY2Rz#0zwOnab~&(DM6 z0Sow23dj-WiOLmolq|Y*S9-G-U9j8d0AN0#M?IqN}* z#h9k#qnqk17jR-<(aMx4z2ycUjviCTs>~ z@09+Bmir9V4f2np)(F5HD@V+z4LL=XQHRGsrk0F2QwRoLB7yod0UskPc42r(jj=`g zpqdhg#4`1bb*@trun)PYX^j<7{+h{gIj+XX>@qyvf+ej^mvuNoBXA@TV8_EOB9jNQ zX-KqxrVf;8f7Wwsbl>C=d*sX7usHUKYWlkRK>J8z6?0lU@(Z}*o0ge7hLTB z3X!x_cFe4JjyMEl(BdC>rrpf?A6NK`sOv~!B9jP6J>(n;(L_vWF_Ogwv9OY<&@qME zTV33u^cF;|8-0W(2SpOdX7;g!xyrSP4w(tzW7Toir<|1nR&?aEy$pe-bMEcP=lT#asx7 ze+b(c7QjiDX@nO#w9-Y@iVcFmp?M z*)N}S`z1BQ@Q?)yDdeTpImDbC0X3y!p9Kwp6<&Mb{_@kPkG;{;lzZHX%&*!xga0lI zLWp_mvH;h%!SZsuNql%uo~9;;{h|%rO81v#{(Ljo zw+J~};0OfDDj^#{ud9XYrJg^<{!73fs)S2bYqN0yC7ZZ#W~-Ld1J@b>us?2=h1KE` zM)uGYK>y<5RmBV9EnG~QW5Cv|5brKjylp6VnVQP8gS$iVqzZh=!MyM@pQus+ z&l!Jmge4rWGFEg~+WvwL)P2JGzsS4?t!4GcnlCXPh$v<8m2|K#(c0*>*~8?2`F5H5!=Y$mOqf$$j9WpcJTf8@81}0+v{_Hc+?x@>PfVSN_mr~+bLb6Ng_$m$i<0#9{Wp4$Z5fS zqzTvYrZt6;aUR*6+6*3~E|W#0t7M(*XyYEomx#slI>=E*LQ|{zJ`GnOnsQ zEg%K7hdknr+pQ%cfBUL$DL4~}TsM_FAPMvZpPH51qYn&!d#0l6=;yGF6)`*|Fy9T; zw8`T*o{N93+h|Sr$7Lv^{;@yC3YVvKd(fTpZ*Dwel-Y96w3I~^Oi3X-zq<>g&2*Yg_U9{wY568arMqwL`i4kcJ z59+vrfv2MNLKv4ZbmK44BJpQ7KHtA^SKC8Kq#Tqhb0!Dky{a;QM1=_D&DiafPGyHlIa+>3LbeYm%_ZXXQ3N&=lNg{|odvWl3{n?_a7uZuK-) z_tGatEMl1OXO4Aeh(;JVDINQjRl#}&1lD^eY@GkG8RmGb${5B09U|{He5EcTqLnfu zs=5$OzBM4p&XqX}I4#il-=XH>?A7Q0oIX3kQ@ASJ7&;Rgn6G^I_W%|v-en*rkqI(> zJueETrC*`ZQMo;{QF(h)38GHWApcJsy{8Il)q0p(MF&-iDyO9bV#}y!2G;drNDs!P zoT>9=qNX`nHs~te=!mM+2J;nxrCP{uI#zic+kmHH+krwqn^-Tx!;Z_&K|OEP_ULAmKlIFes?xo= z!P||LHJynWM$~b2+s|vyA2X^13>f3FdcG62W!`x110vk?zI#J^xaSt-c}I&L z$n2i}0U-7?J78yGMf7%W=|NGBv1~+k**G(3C>SzdoiP!EiNh;i9p**gvpc8{i?FlY zzZoE-1WnS!KP?oWy>qsL&ITX)c%=VVnAK8s5siYmYAghvv><)dxD9o_d~m)yS|B&6 z-pDZNL4Rl%e}7v;6Ep7Svo1ktplA5g;nn3Cy$Im`nF;kw9Kw~`PAlAQ74Y{4u;3iA z$E@{NlSiLbsJ3(&Ob82zGBvUQX-+;Zk^WgoH5^|;?(bI@)8&-4u_%H^+LO3HKx`{< zAI-r@mINEul+SJHg+|MAh9%upg#3M^^sW00K56NMIUb7fYUBq&Bm7fOK;rC**R!a) zg$}I%Oe^P-+v3`4$=Qq&uzOAK*86(lu9*#F5hRK5mEEOh-yD!=jb0o7?a9t}7eHH5 zAu}z8psPZ~4OxDsXxT!rMTH9O6wqJ+Kd!W`wLaeHTnjXZKg9D1{|;PYj>A+SR$<0x zh-!1uj9T6rdQr8?R58H08iwG&jiQe`Q7*fIxX+7Asncp4MB9U#Q_t>HXHvOdZ%F<4 z#!qaT$!M^1QwIFZ!YIXRLNM4HVHMIrLK}_vF9mOPen~hI>r7;S;745}(}TG{PCS>WE9(hX6wb zaJ7@`#x3b>;VXAtI$O2v8wfWK931ga$;uIYgL!BY5hYscq7BJQcLh#kvQ|85I+K~P zm?-0dkaBac6f1?QwzVfmiqbxol_h{gPelf7d^1l}#j)$2>2r5559SG3^+;V-6{l`K z_IF0fNuMY6+m(6BtM`oKAT?sb!55G%ce+^Cmaj|)k3e2S09wgr?gzJ|@lR8OJ&2!N zy~z2p3#EU)lEjryn497M1QISuc;+8r>EWvq<#!sVoa0LjPd)X6ZMsSzj#f|EWD)8%k(@iRUj32?h=+|h79{w0YnlIO%ER&%b} z6ncJ74GP62y*XU>=9QEk3B@@m0PVPDx?VxWPYqn6?|vvcfW(BELR|mJOZOj zr(|J3CN#T`k!n?@9x9bT>mt5~WvG;wyPSs`BL3J}{m ztIxlhVFg5#2Z4uHxC1=o-3@%CHg;X?g!wNJ<@yL&{;n$1c-o=$#D6n1VALF-V%`0X z0g5oUu!&A%JXFuqB{PKkP@9+GoztOVU3KA`nmxiil4$bMTLrAwRnjM-Ocp z55A+m0U&_;Pzs`|i4l&CH`JX@G1llqa1vC-L5%p8lKC^%cf+w^$s~#hpf;5+>2DTH zHlnuj-X!xJb_AHUTsae6mNbO84SRwTHy^kc{Jqb%*#Bne5sCEU4>hog$#+}%AWElgf zewAG3yHgfff{S#c*%xn8)F?xjAP5qcRjqY7S=xu=S(2n68j2!%@$?C(vr0@H(ZKk`?^C-_ zVawFvCX>iSd1RTI^b$3hfupOzWLRO?wSY=%h`_K$Hl5lYGyqGj>p=UqA0S@6iIiF7 zI{enZYaCD5n}Psz&7Ws=1*CXXaL%5;^6+ATk$i&FI4s@b9h7_MkQN{#oHz`2FESPM zwAcgpXmIXUfV_Ditxf|#Oy5wPBu?B?2WoDAf_V!h{nHDRdSOhxLP~M~{RwxF`soT? zRPR4^FQpOO!b#&cVo%27Pn+1nS+ z0C}ofEZFE_K zQ3%Obq9Wu08ub8?AWgCR3gbJ`Nc4e#IMdhu5ri8Eg2ec7u$1-woMa(WYOJBH5+zg9 zhV#>>|3r#YhE`0Ao%5M^xQ$s`sOtV1P`t4cEvHKlL}9|g`Pn=eWNcc~6l~xlXntHw69vIA%9v%Ob)|M~a}r{99=IRv@Bz|! zOrzc>Rh+GKUFk89ti!(#z;eZWm706d_E9ugN%9-_YGGBJHv^>M-mD(y__YFJ_-_6) zJoC~<+habTNe{Op4c2#fYB^6qQ|yJX9CS80o`@=dz!Tq>!8HWYQmjfmo2ICjmVB_I zD24-16Ky6lUr|N|auT$$dB?JV(&iJAH=l9C82l2NaA;#M>jjWXN0Ckf0^A`-ttprT zG@;|-LxRv=n;5vNiDv%X6N+hoUOXq^<|G2XqzCXBY3@q=tm=siyb)(MVEfe%Xi>Dn z56yZyOrv@_dU8>rt^PPQMTo#uuADJ@_2EW_B8dm^XN{pKTW7Skh7e*K2x;u>*&Q)i zW8Zqe7YYptd^64^Q}gbx;LXk4Nm-c)>j;K4y-N(K?ICTN`Nkk1wgUJA(7eTlshQD> zsp*6QP7&K)n%p-%IN&&eXH~JzSV0?7~A(ks$Zc!1rC7nW%8GV{j~4B4wn9QP@joCD~I?r(KJ4s z?k6d_D9gdOMyNT-FY*7d_nl!?omtx_lSwi$iN+p9G*L84(Fg(pf@UHrASfceS?Ebq`R2Mx=ILmZVJStsIo^i#R{g zCZ>SzY`Qp{?`6>Wwhcd4TbuLqPSg|Vv$)g2F)vt!z+Pm)BhD{| za>Ev+DSQYwboEe1G8gM4V(YH+Ujg<$A9I!Srhqmdn#f5#-8@b1XY$mT@O6_PjHzjuqXv;Sv@6`5$jgA(A6osn67lKSZ$_FZ zM>~⋙U=K?EWlq8}Ff#6uXcNFD#XKb^Z(CB)EdFfHhJDds)3TEC?I*NjrGIpzo`u zq7k@%B<5%oPjM3bPSNGfMSplrTsLBTlL$j zsV}K!Lq0g(5x~bi1f_mfC8hTjz71PM=^G%h01Jt1By~lIa8}U9KFi@P8ldTB$9jE= zViFp$2`qx=hD#$0If^hDyqb^$Q-vSSy9XZ~CaR_^zu|(%0PfGH73G!<#)!`-s@z2d z!RB4~o+F^um=|;`)jRfmaoT6d)WoO?iFkYWu3@6uyL1F-rb;`ZgGiXdl(9bBle#^9 zHXvFE)gO|a-UOpb6~)^?)x@Nu1xw+d)#H{OE>)sOMQ~|arF48`ZmJp-`4qPem5bS7 zgP9)W0wwoF25xlRUAYFA$Irv^wq4C%6LY4m!aA#ENxgxa^{%etL=Z$YBaKaeWgzzTAxL5!^PypQXL%jwxryC zZ@P5=+lm=U2QhEFSNoH&AQEdA{r$mHDl)(@Y_ztUO@5hNyOW?&0 zyqyC!gD(|~E3u%8o4<+yieQIgMhO6dMIV0r`IFAp0H`WZ#o#F5v(0F!A)-c|wIQI2 z&6tw@HfN2Pgfs93*Fgw+{i_HTsOwuC4M1cwLCkP1=f3#3Hra5y=d0_his++>`4Rle z*whZRjADQx1U+AXrMvy)gG55CY1IfUrEm}=iNw7VL_Xrhl>^0m)cipZ_F^2g4XQ10 zg)*K)A+7h|Ar}y-N#nzUD**M-46*$P;bVlQBXBT`{Gb3D2fQ#CQ8Xp^o1I=ffux*T zabHjdP^)76+J=za2jPc2I8~Y2T{r;eg;;GfE)AHJ@lwRGV&^tf0Mn99M0V-4RGhTPT9+Wf?)S z*L|f0dB9Oh9pIGZ6qRFIPyXJfO=nY^m!hX?B>;l6F#b3ETY*7ULoVw41VYJrA}bvz zMV(|mUhrTnEde+fO|T?BHC#w@&jb*tNZ3vSE$^X4iXVd#AKY$jrPdZbhYE25af!W0 zuURWCL2iL21|rh*plAaPT5T8&YFPK?>MOckVwOBZ1bGfg{mq>P5%9TZ(fH597bwRe zhVMpj=MmW6?Q$RmKqR7 zd}HTNJ!Ogr{@V7Pe62>GNp@l@+&{0p2S*+B++}yp;5H5!Jdv#C1=L(auCpjyvleuF zDuhtT+up-8;V3=Sj*zx zq-r359V;`OT7VvxbB5@hqO8Fx4F}cn86N3*Dnek~aM^-Zp(Jvr`baK#u84}!Jd_TU znX^6OIQ4|p5Rw9cg0G3iGW^?lMc&px)^pDmN65Oe(3M~fuq)f(=!6Ga(?n4F_Jl^p zrj_6#5PC|93pot?lY36_5ZRId;H?pc)nTwVtg&JOMX}9%Q2(cirblrbsz_NVB_Jj! zB^=0^qQbwpx{UyOgdbngol#7SNRW)-Ezo3kB1dg|_`^JrOAnud+kXT(BGGv$L!}++ zgn}rA>PN9FloG=3{`881-rKMZYNbRQ1ewHu2AKLl_IVe8N#a!Bf*+8F>rL~VBUz-2 zQZ5@yM??)RB=#7ECQdmYRQR+Zu#SI≦Ifa}BwG^AuK>d(&qXIV0{8 zZEB?%JLG1m>IMqW5qJnlhqm0WsPG}~Hvz9?5o1kU;@1<3fh3^>LL+VPMc&v9^1!QT zELt@(NHbhstm)qq_R_auip}@#uynVK9{3Ts1R0u5=(gb@fEX#3xi^!>3KjuH;-k_d zeLc?8b(pS;Fxe<{pGD!_c#BedrN!>XrBo`VXMbmqAO~!{y>6}asK}vX-cd2&_|-_R z2~s)D@1zwR-bSN_D1#pxEZcMxsSo9d1S5bhMjBAWL-#?JWUKjMtn)5{g$ekekE9!; zq7ld^tKrni6w}oJbUpaQ%Iw8ggpiSoyECkWO&p1rR>PZ}HVqotYn zU>Eo@-fpDTfhi)(LoOLO#*mYnHG551RX`TS^N8oZgc`^h$gkbQL!^orxFiHO@E{7X z?XmpHk-3Uel&oTqrYz>+Eu(HTl!SL8LYd>6OR)nUj6|Duwjj-*F2aZIKbAmbAFNq1 zqK4(zakli5i>wKmYXgVO|OR&p-V7-Mi0!`^P<68Q%T8UkgWi0IZ)yeJ;Pe zNea0DCFdK5oSxWuK%0h&^HSnK}Ug zxd%I`*4amVPdlJDfvz%(pZ@sS#xp1m<98yi6Ov}knTszk98uv>HdtRP(Q^Tsg0C`1 z=Rmp8W+pGObLKpmjM zzXP9ELc!qoc^`NU(tJ$vDR=AvNVM$z3+uk0gAmKBeQVZVSbKc`_|;f&HU7B*|AA0_*0s&pEwhe(#$wEdJu??$b~R>Kj^AIbM8u^K*LeMmckf7wRY1WE??<=S@0G7S{&U;kzCm0#Urh{e$ zGoJu!&O#d6%PSs(s@I0WN*f1<(W&$tjMHY+T1kGvzPBDl71X)<&j6?0PBKQTTOR>) zCo(Q6R|sA5dP-H9rTWKJb;iN`AZp~~piJ9zw)7%jTqRZ}ldqnpCiO8oZH z`9%Mr!gZV1hjXb&>#qd8;E*K{LG?xP&?5d$-kCFy&ME<~s|{3`3;6xp#?w?H0cjxk zx%5nY&5J)Gc+}|#5h$r737-3}2lA;TLi93BP*~@J>yNKYyXrh9N^6$J~LvE9J?} zyqH5>->|bTy{76T$WRW74WNXyfw1aIl;S8hN=gIhU%T|G0T1T%xooFa3}!$f;2}t7d+Eio<+P zlS0e!N6OTz@nZ*3LP7t#RXd83jSR0b^-$}F^R|77aaQt?Z>XK6`Pgq?Y%nG=9;g{d z;vK1QO#f1BAi@jC<_~gKp*m~_8Qv|cSCiZbZN0i)23+qXch^nk8lEfl=0e5*915E> z0}u})A$Rvt8Gi95(lfM@d?Yz%Aa}4xmb^Ud=Q?0>>#EgN?5v)Igqy6r|pS%FYIdq0W%=A z8Tx`TIvG6@y~Gp}X%A9vB?>d$)15)=>t)o>RNOmrZ?l|i=zg$YqlG~j162k+toNxD zeP@t3vnXWkjkApE(S16ak4p2htm$~eH-gl`J%2jH{3U4P6)yZUFPB9hM|;qC9Or72 z9g1na#Z;K{PsMH*s7XfmbqNRtYNc3U$V^fvh<3PywGNmJ*X|8N-eV!^nLcOEfY?9w z&kjFfJCB)P;br?bZr{%&48jACO%E2J=(M4XnA))ui{=l29#aJ4^;sYu#&Ue!CMr_( z?#AJm6DWzsV$dK&R9lF;yARFJKX!b!@%GoaN{wfkjFrh(cCr074R(Zhz={pP(gUic z=MrGU@=FS@xIYH(H-f?O*`TAhbQ19WHGgPf?Mh$ye)*m$p-i5#X;%Q!3IjX5@ zfxTcNfvHnr#t$_CWo-kf+M4Gsp5N^u)U2Xl^e79Pxdg=O?x_F~o!`%v8vSOO(~{F;)cgEcvop&xZu=jH)uY;FjWZdl)&uN=B{ zb3aVvj~&pH6CwWE`6fv2{Q0wfFP@hgSSkJ;IG!eK-@A@;lErj88nSSwG{0@mO%D{3 z9VTCxZZEq{dLQq8ydp2RIeoi^J-VHwLU?uRGiUxhkO$oZhfTLnXj8M^vf5yj2%#pD zb|{-FgJ4?z?DjD$0Ddv#fC;CftnA$X^yVyJ%1&kU1v!1I=5llbr8XKSN9E}S$wY14 zP=>PO-bhPx4_y?&i+k5Aw90|rf4eWUX-;4=IH_BI_xe!OCSI5%OwsAA6X%H8?Xq#2 z$>IAiu&xeLm-3yHn(-PH+^<mBlHRUsBC@uhRf!Gq|7*gQSbyFB~ZschUmH~BW zzxzpPgNgnW&fk-|5f%%h14Q=vOAgG@#mszx7qEk>*iBT81>1Z{9+lof4nNz2i6ElH z4f(MHT)rjCIK;-ibCP;sVE01QPm+c5?|_ypk6~V$^X?1rix zN_x&%;Q_d@CAs-l#Dc>T1vp}+-JbJftqE?_L43n@=4Y^izr=Ni9lQJKhB9gnJ=^OG zNfN4-r!+(J^9cxgmwI$!Uu%rzL7?NvUt}2$;(q8$j>>ZqH9_%zsWA8uO(ZHF7}Nj5 zJP3p}BUBN{4b0>aAg05tHNX?Mf^4Wsn-)vPcx+% z3rpQ93;GzgWrxD2+`bz&?15sc<=q*?1uv=s4^5-B03-&Cfzx=$0#tVtjV3G!bfuA+y1l*Sn**I7IT@ec+eSW`9G4x|9V#Sg#B%AlI^N<1Yixk;DDx<102Ibu& zF_H99U${sLe6-NM{k=BYWFl=_^Nnj5-{|ERqfyO&_1R>m=;$ayXZoU59Rq5rb9qXmu2K?Ds0FAv}S49u?$>U?X}8+i?2@7C>!>)*H+c`OysGQ9hD zU8Hsl`_Fn9-rIi{*#XAS3$uTzZ2vEL6#w6Xvy-dbMpyPa>Tf?b>1fmZ(`|m4>Vrud zrJq0fe7{CnRC?|0zpfDx@m?-bacbkOBG#i6y~|}WB0_tuBmb%^e#*NjQ0BSf{lPu< zb5|TXv0(4U(zXBiZTwA#a{>3M)7-9)J{gx=$rr+vY#p5#dM&*o;$%3{EBwxq&i0=_ zUkATOvY36r+r^n(huL-b^O(#YoIloL_FT-Ki$C@P{r~KKnB5Pv`++qZN4?zyv+FRs z4l~Wc?0%Tt53~DWHca|s56p&(Z&zY=9sXnMP;&mmhb&h28-<|-S7xG>D<(Pl0!^1%y_&?e zNKcLWvzw>Er`V~luWyBknrN3>IDbCLsi6poc6z<}AAT}-uPe3Pz56*fwUCQ3wU{b! zfx0bG;`W`GK$-3wA@oEYfu6utO9y^AcU~WLF_n8zi<0X%r5_TJ#bqzir5oa<~m z31Qa@)Z1w|H5g=OVIke!MFZGwL&5dgIm5{#qzRzk%IGp%FVYbv%L=nZ!yhShk;LM@ zulnV8VxF>)C$x2oF@JCgs$p_MX|?tBB@ixYtCZnxI}G+t^}~l>60e0sTz>fBhmd|x zNI_lC$&sd}Vycv=EW-zpPBfL>zmh$1ob+aFhnV=W;)Mvd=G`kHAw1CRM z1v<W{aBtbQ?7?qcEck`}Xxk9W3J;({jN9aTAl{)t6 zhCorUvN}(dFHGNCXl!h3eS-6e+{=GZr}j~_;1689;4UC2DCq48g_Ab}X%TOPsbY9$ z4AZi|qy+*yObU(pPcA0Y45N2Hp{Gk_+*r~js5g?j=K4lA@KSiKrgKJDua<*cMG4Jb zLdE%19?cb8vQf`vmkA_aOP+xfBa|=!ZN!X>jFO+-=FcZZsK&-dB8weP&dij+XF8R3 zQO$}vA@MZMk*)*iY7Tp8ub?fN%Hv1S*Ite7bax7ZEY#=U_l5>2zWdqn>6hGq7_^@k zLt*B#-c26eN>p}Q=$TLRpFs17OySZTH;5Ilbxi8)?4-WpY7TUM+Q4y=dm2myXZ-8j z2&PkV?}0J1&!v;Eb!-=H!-0k>Pl?j9rjdv{bQZz25NrgYNV~dV_qz+gFJ6BeFi?}PbedJToRB>y!Ll3I9-`?W5h zoGmwYXI^Ia3rPH)s{=`6Pnb2vmNV%pIR~VhJJ5RW^4Y4TCjTh6f_w)ooOT(Sz20p_XMab(EImIsFvQM+gn9)Z=WvY z@0~t5edY`ZK4BNTW_}KfYw#>@!7D8ElEe}pzWUBIsJ!!pllp3FYop4CAxE+smY|vg zGtuM@4%vWw@9WpEU!^4B_(EXdK)R-y+NIZ+_VMS>?=AaqaS&M7N+-?t+{$_U(&=Vz zAPDE^a`Vg?+Q~2upY`_i^gM?)Sx6OTjLJanWxp26za>0AhZ}F#;_s&%Np}&A_xn81 z-~H#$%l>aULYtuo<_*Hz;y!)!Mfc@gAMWby?UmW4xpT#k+GwLp{XnHr#fUHJ!UblK zBi76k`i{X|QT)jz`5x-X(Pk|;!ql*v+Yi&RQHR@-n|1xD{JmRg8J;S-p|K6CS9?x0 zwHAZzrJn8#azk_rgK-hMT@a_RG6{ZRaUw)4IQ=p@5w0NGNpE>1Q`yn5oR6;!{4I-W z4oSQ$j2PQl`iA=?+rg9#YFVH3qu1XqSqX12ca>bI)Dl)`n(qp!1r-$)@|G-?$^~ci z;cnxgsdvx?&A9j5TIP5is0|MD!L_H)ZO87b!?%56pUB;&-_>Jfn?G*i-YFJw7Aw1S z1a6MR+W7NxGpL~-vzld&mq70e>iC>JSN_JTw8q@|4c@~vcRIC3Xd|nz`=!lCtY1rV zK@eKW)nKt)*}5O>j~nT4X)>?c$W8y&L1n#=kc^P2N=f#2O<-jlf$>t+)YN>^3yllN zTrVU6^2er(nsC>=0fV#MU`|!PhTKeF{`KMDI_K~t2kT56caq~j0;b5RJZxVV)biV) zh%4RaHuDp_h9V;DbdOYK?C3WdJ>F>rm0F%w0|YVVxS^Cj0LmMc{Z)D^Gsj?-S3+yg zwm$fmUw$Db!QmDfZ3q^{pb_hsjkJCHp6Y#^K1jJ7)HqS}5o;?yV{utu(~2vqLxfkl zuVLz@$mYc96#lKpz8>MWuEOcC8sf)@g&HYHIY9;9trz>Zi^~h3jz*>g&!1VCr3s`zN`ue2$)*=SJrdZAIr^Z zANb=LM8$|r=L=!kA8o*pk5$-TB8^S?9&Wk)7&@+7eF92T{NcSWqv_bDM&(Y>zrII8 z-;sIrv|(eKJeK}>WCe91KCV#Bxd8DZ5>d+SpZSPbie%~_9A+%vi!GTk{sNk9 zZ6r&lz?IP_aY$^I@DVFG8~I*%g5BWIHMvjylZISiMr6%%(1*&XaGrU^tes~@6>FeH zdae$#FxEpn1$#S;61;l?d{-LhV&qeItwHDHh+4$!gCSLw5NILk_CAw>Vm$Ov zZp3CnfQQIWY$zl={MP3YZTctv-KAt_}R8j?JwnVPxzu38w^_1x>@2 z5Gm)MxI1&36uL&IrhRgCs2SW@Sfi!;w!b;PbLR@!RH7D*fpx?O3-3G1?vkYmV>E(3 zJw1I$VZlFAcl%L%D9U{(wqPS=T9`1U6{?u(#@Rax= z4s;GQ+TYJtGWNzXL0$@gsLX&j_>5-EAtsQ0+66)`Pv?mZmKCX~q=`0x`9!fGPk0ET zxGH1W*e}H({XYYt*9W2&VF3*@*P5kLypMQCL}RC}gyQ02TF))w;ze}iJ-BX$dYBAh zht0hU&b!gW7KZV3XIxyIH%N;-StJnmFW_9rBR=1`lMNQx1q#NX_Sfja1$$3s3&Uy^ zJ~%VFUKEpCNZZi#f}I|=hE(Eyh6n!LN{IV{u{Bx0yRYG?jJ~^eFQaD+_9<)6y4e@d z>ddag+dVS74zug#v{4rcem?0)#K-Va@S*;&AYw;ua_ zM~GZ5a+N1&`CS(Nx0&C=QrQ<>;#TmwHkn4b63&DX&iiOc29On%6P*|Zb{x~D+J%*w zZ|i#$JHSEH1|OAHkgb@Ro1X>n#FKqSDsdJ%;J@E61q4=E<3IEJrb{?syDin1SbbO^ zvEobP1s{ETaFM*7i_$l0j~}P5DD~T`c75lyr^?nT4c9^iGrkw&lY28~MUDM>H8-`i z&;x({?DSuicYeR{*r9_TEE^p?y{fA0;++q-o%U#-^t8NHmp|$-uin8zB;sU}Jl;>SCktHt}=q;y}CRDy!5I)pzX>9s9J$cJ`KQv zDHg>QDCZD{mYAT4M1iw@02!k&jkv;enS<(kXWkr3%Qzrlv>&L;BKeEdiAm#uq@7+s zi1T$j5V%r6Bq10QvrYl+-P?`uc|JB>j6~KeFd=o2L-Ppms3757A~9Y(JUy%zjFiC z7&V&0xcCHiJyLmpYlEoL%#bCzQW(bJtcC#V5 zDM8)GnNNn;e#3+nqsRKXiU>6*HzFoEPDXKH&Irr2xY+Jth&9%9Xyp}z+R5X=EE?a zM8aGd^&bad9nnDRD?%CA6-&eL2lY?5Xx*!q1b3o4=QqfDz;cOfLFqCIoPEpP>)tn zU8KI7fjnFdu)NigM|C0+mnKvQ5J$X%C1C2nc^=Py7%w-KyM%V@tl))=Yc3qjh?Oo3 zk%h>%E7z?x1mLR2IP~L#Ajx}ExNO~<5OjZ%wz(1J+Ek%GdFni$WKvHuV@K`~dK_6w zqZ{-&80pvr5rxFXgWi!^ovjPXCeURLI;2i8p=41!3#No?|-uT zx%FIGbWKY={dr+3#>npK$VG9o^XG*s_o)O()9LW}P4@NsAij-aR5?auU8D)}=sQ)k zjzHoZC!WZzopcGO;fHOV0?L>j_W1`s!_T>{mf8{&ydme$co#- z?AJ$0SL$WqhxIP4P^W&AjVTb86YCz(lxcAe1N~uf6r(!wAbkOScN=WD4c@C)O*|>F zFW+49*m@f33lj9t1A)e+spT}20%o(``XpnCw89n3+-PAj2P806_>zF(c&TQOXR#)H zN#hXx5ia0rzbWHQg*AdAT#*emi{&qdIW=y;`e?RP@kTt7 zx5B9sb~U$IjA0zEN*&bUxV`RL#4TyS&Zx^t(2<`m5fXGy76p~Da4y_&VQ6~h?Q&ognRf8X4IMljV#FyBBfvZ$h%d6<~T0oL5GV3*%nXMQwWrIjCg~)+T$3)9`^M zn+({^Qsj$BO`irGSxWIU-jk>vA~^3P^tm5Fh@!Cold=*YSgt}D>J_(jI}QDS`}a>3 z4w|5ezGVdCPvY>~7TYpRFl#!#CS;UMHu}8RDELLLAH2TmO^iY><}h9%^$XgG)_n_% zwKqa8qhPEE&G=$qza;Eq-htLXa2Ci>_PIhp{r&_-?5<0s@l?S6CryKYG(fj;8d^=G zAl>eYdY>3LATU^w>n%RYr@2{tRw>`N6fjs0EaB>#rN2Q3{*Hicv)|~g3mypymg`hP zF9V&f^>#DXzKNFPaSVX@6?~p9RP66l^n%fL!c@XatB`1VT&IFKQB&Vhy`_MmWWy3p zL=*#hhd%_?Yl~?J(4{(k$69r5%x?3O^mdb1ZKn@J$81n=E+!7Nqhz%QxVD9H0rcHZ zRQ$%+WQB@|0QfNAr_U7&sf6jsSJL}wn3s2C%;HhvW6&TCy{D5X!gdjhp)_@`cAzV! zqYXlnAlc57@nE$sA_@U5W3L|;h=%fD$vn!$NdW+cukLwE^!7L1S$fWQKmGQx4Tq;( zaDG`@&p#Nu`RZ9gAuiDN#1R_bBWYd%z#rz)x*1^`NyOV`+hB#tQ0X`1+YUDs|LY>g zMQA7^OhIkDo@Z^=C63cvV7HClt4Rk1GZsS9Godh$(FOSeaS;K=TN8f4^((od1{$)3 zO9OQM8vES9gt8o5hKgXKYys2;;(uk@EZ)~e**3YM2meZXC%sE`QN}==|0o5qZUzUY z7z^&zxnJ8g87>;(ZVlYR1#!LsQ6YD61|~@WVGb zpK`zmgiv1Nz4vNt%JtqWuw|9ypzWi*OwvZrng%5$dmdw4hs$ROQT^I6Fxv=w!H>oe z!8_&qA=KC;#yu}qX?8GQ$#0+gvxUlXrZA-$4&IUXX;|FHv5d!jsAGBDwn+@aD5nZA z4XQBK?gX);5<(E+6qn=Pj3GA^PNUi33=>)&X5DaizW>eXE#sklHn*>3I>B-8jmBZ` zm|$GtjsteCwwU}?LBT?(=j(2`V0<R|Yxe9((&@Aaj^y6@kVNkrTEK%bq`N7On61EHzmGnoN@KD9v$6n^VUKK#Gc81}NUhkdZne?7xuZTqi z0vn%{kruc9GlqkBn3EnWy<>3KHI-P#Bk^*SKbP&A406Kw@1JbH94_mo!OvI{R^1+i zzzC!$r9~9?0HgiMaTgHU%g$1%*%(|G)i zwN;NWK1D0Lw9HNEcweJ936&7U& z14=AOTg4fRlvv$#lbw|glekG!&rbg4-eAA!p_5blD0i!u(>>4*br3z8oHmDv0X)>n z6QR{bDnb*I#lZJ$wl5Q*6b(*RdI|GO^l-@Q?$I&*?F&gG@MW?wmlx@I?M24xe(%%f zxY3MKoKiiE^dcbTk^pb>>HNbMhOyiv#9G6c38Co3W+}UJF~8$JiPwQ6U<#h4LZHYC z8pnkZh9zP~_kYFk)~ltr8DExuL?nlu1-&}bzBU*-}u{qIXk-;ft>HNbnlRXN$O9v6XXhc1+myNa8Dd0(L zUBkFfc@-u7m9~1qeg=?qjH92>hOec^}wfWNq7bZ-1?dn1d6Op4HC$fGcVt zaUTlP)G1S* z4$KCV&Hl0K?%KLmSfvKUg_cEmQC%b5kG|D3a5WCcXg2pAwTJ`FjE!2 zfd>* zf`1~ci!dh`ZP!Fk#!+Coq)Di+l!Fx)u=VHCUFrc@D_b-g=u|Dg)cAtz$R5O(ApW-l z99N-XbV5wx;ltI0t6=)>B>{O=BaGVZet7LS#jd6D7cZ$u$5!W}v7;j)t zVKwp~7MjsTc%*`5V7TTY`CmR?c3@PG`NDO0AqcARf8qF zT}<=D>e>7L9}=$hu;f(*6nv#RFJrjx;eh(&o*bm`v~>cQq&=^jacp)FoSO+Klok0& z!eu(zUox)kiW1_rM2l-GA26+AV8Y-a_LKT<#;d(@;;`Z1IRC}S5fFo2^uG&m`8P0{ z|A*f~@RIsKxmZ;@2C9_;pb4eAbEeU^r0(A*4zGT~;3Q(HoHbc*&@r5zymtbh88lk`xETIHB`k1f5&tF-oDhmvi@-f#3O6+vQqc*JX*K z@`~R*Gb*Ja9-W3VQRf8=YPUb}auu*Z)8NYm}Rjy*k%-Raj}wA8R1Hy%zr+XhtFLPVNK z5PF{b?;pAX`Q*#PHTR`K0ufL$<`!qF#@fQeNj<;%lPFw^RqBIv%5uG^8+od1fAeY= zm%suvL9J{+1TKaU24T_keUvN1+kK976Au~>j=ZtuOOctemY1XYxjffo9^M7KQ2nw| z%W$HtN4%;XW{W!ko>dxzNMjeUv?`?hL1Ufzy6IqU>zj=f^vFb^Xp=sJ3}D19kpv^ zKx=#zeORX{?(~}i%Io#qGy{$EQRFwl%i{-!XI{FO##LOpU;B+qDvE=}%C@`-*D7f? zQivJTaB>>Q$PQtf4ji|#Q3h7*oeP}qvS zlv155R=ITzU^jzqZCpY2TUj;YGyv8K*;V5hhHzKGtmM@>I%zk0Q(5 zqC@Q!@%#%AoO0c}WN3_RoNv0yWi%F27v(N){&?H~wHBUbfGpN39$Gw4TIZWmA;Zg!C!7mslPF1FlU<);T8FC}lzU zuc89_8J6%Nt_EQfJ;#-9FD{~~v>ocK_fro4#u?ltuF+4s6oI-$+Xo+hd;@8fvW^>o z;`3}RH*p~fezV#I*r2jO6MF3?>rcMXTyc|L1s3fdA`O{ickqz9nwv)`n(Lh+iu4L~a+_?2*3j%I<`;Q6{s8|jqR-Yf_{LU!)Vz%2Gn~`@fmfi-)4Qj9Z2I*vocNSEi^btgX8wg0v#J1Ha6KiqXo!0~&@{)ZF0M3O z`P@eC7igyG`CBi!^~7oyR2vLCWI60koKXGidzBI@tSnDxIK24+*`s0mi?k|(>#~?P zA3z1vwewi%$Dy@u^{Vx0s_>(S4yLI=swXSu zYU~avRk^vqc4ML=McJb+4t9MNZb?ZQ!Pd%E7D)v&y#*`9WyGO~E~ZnZRAr#KN9G$#H#J1$Gz)%L}cTiMSe&kcf^Ac3;>YsD_7 z=MTT3peBE^W0NV-K$Je!&BqFD#bsC??`=jd9z+4q12^g&_bEmc8LN~j$sOu+xDpxL zfJ>%CK4C<=VpwvfL15K=fuhC3MQ{z3aJ?5>oSH;_J3X%os3KyM(o{W@e!o#7_}~ms zWwQgOx{AWsJsi=z>=;S3&8q86!?PwFFR0A`G%b@?( z=2l3~mjjNiY#@VP{mEX#X`OD*cuVpXplW0_Htah8Rkc|&7$pga9?QbWA_-(qPfcj% zteScA4|90LpwQzr3fV9Wm)NSMx(8}{(LOl>v-S$X0U{q|Kk8&6tz~@bMYwZG9UfFj z{eKRaCaylBujBp0T&Ww?txw_QYz7+vR&%wnXRT@YI91USEi=)eFJ9{O@cU21&HF+M zO0Jz=nR7RL)mY&tHJR2;I@cT!&=D^C2nQu=TNjRZ1w!_Ne3DzXw12p2Y{w;0My8Ya z!?}D{sydb%2kC3%5|?L%)04vxVc0u`ebq_8pb9!x3E%nY<1b3-gn@domA0@bYSQlQ zYgWa!JVrD^po1xVhWg3GQhrqQw9)Zt*#^$Y*VH-~l0Jg|WZNxMAphu8nU%e8hK_-F zckw+ur4EFW%9qj{xg1&Crmgk1lT(ufzABfx;6_aE{0NVQxHSk^SB!$?OxxjAcMVRc zQo1)mKtVfD!H=e_&{*g+1Y!jpjRce-=3X^~xb?Qi?@E54o#ZwQ=IdJlb&x{+9uW@a zxp3@TAe3*9f@WPYcy+)0kckuMzK2(l6YR?6~M7I^l8#}K$2Ad z?Y8AUn+OI(>wA=G5J4u8Z}6D?fyr3Fm4x)Ei{u7uyR z?b`@<(P*rfVtwd|o>*7tBUrXkPKIZ$)4hD=R=+xI{?F8r{IqtUIYJ&Pj?{8`k)bJ| z58lrRkBkt$p2b7f@o4lONBQva^OlRdR|Zot!e`D2g+bI{+fn2%N1fe*OO&fn1fLat z)+h(^;J`L`SyL!;SCXcOaVLl7Rv<$_Qm*Qjg;=5M`SlB)Wey~|zv|8N-5I*eF13R> zQ{E*vnoW}9)y?n>osfvIB8;{%7l}T0!B$vn{U|O2TT7Uru#2)*#lciYFnEB@>D3YiEbQqgioOuO}>t1eV3*t zB$BuY>D621k9Su1#e-dt4Izs(IN%OsBT9Zqa@LYAI+;}BEdXnRxu&exC$ zbMJXMs$%cGMe=a%CwxYmEaDR_svuRb-}a{NBdkjWC1Yr#e3VVlCD#y$Gx62R^r^_( z?k$wWE9;t6Wa5BzQBb(KWW^;ECvT5TQC5bw|BPmF!lH>*qKu(B0hQ9gpiv$mA{>!H z6-04~R0b;xYegJP>o#62nSV>MYvu-HCGO@W;!xJtbPEUIx|}GR9Q1%U_~0OilS>q>_eaWk7VI*1Hv5#~shb(kut z$(O%90>PuBk_Nw%JpVT6>yLLz7G&$ZcS%1Wqovp}Ru&O&n( z8~K6|MDPhceS%Mrj+i2MQo_uSR1^>sQXXfb2$`gsgwUZwxAnC~K9XW%SZg9Z&>I9A zj+_%o{cWzy4XB$~9xq-%vv^3j!USp}%a9;q6q7&VF=K2?6~NDFI0ext6FEvTXh2$7 z8BYcY`GymkIsqn~81Q8pj~<^ApdMilPqh*qfLW()8IdvjY5K9RwSnsIyoY% zUOzm2@HC-ck#whIg0qbS}J{bpCjP&a2Fab>h zRN)6`uDXbz3J1_B2&%9ed`(~+eIo(v0^lvu>l`3AFyqHdK!nY9#(hL1cu?>|nnC;n z)UkL>i%Fmr22XRyd=V?WCd;;!HW;ERg&G)uP^U(7%e72OL!u_Cof;-|xFLsfam|Ho zsUtQZpNzvzMG)(iY8^x(rP4IO@_+JDP$>=;$k3g@ZNTCYQQ;Q$4h2r`Gnov1s(3GG17%`(#R<73qSZoK!NU^6iL7ZK%Jq%?79z; zJ*<$a2beH$`Kyj@O#!2UnLzLanIjJ#J_9de#$&~=36}rDhWuauVR@Xi3ktVB#Yo!g zfpaN#J^pR*JtF3bxCFK{yd}6B6c?0<)QZSPuhclsK-pN2skp|tqQab;sUf3&k~Vka zOBg4BWu`#%FrW?_8JHnb3iKq@qlG-De5SQ!c?I75cj%IJ*gUYr%Gh8?vwj6+Ws@<= zL+<(*XM$A`D=*nlnrd~Cp#Z`vEKPro8@RFF`@Pa@p`iIdhpc2L<04-=J$i=f%6AC2 z8Xb^u3+-iuhC3`D3YPL}%zcO@D;KY0y!ooQ+2W2oQskimFc7#9i{BWFnNn$qO$6Y$ z>4q!OiXl${!V%$na9GGaPCg)));IyyNV3IF##^=4g6>La#UlAzHBzXIVV;tHpBQ|q zEQNE1eIAULU)>;x9R1euJ2sAThM-8|dQbqI@*I+OWm51lQw<;VaA(;RC+s2A4m(=^26rXUHkstJc-pezBb$IBFE5_X+>=VQLd zYnU_ycGMC>ju5O7;=5=g@NErrZ!s?ZPzG`Ob;0NDVK(HXD58jNYmBanq{boQRUBTx zc%xa)5`9h+Xe{x*jK8zu6aKBO$19XjU7NT^h{`BJGlJ&{7Pr&@ySBMq7G;7*`?j~7 z-@w=iW__F`9eJpHv?F&br@AHvK?!Ay!q*KN$g4gUV?>WE;p;^~qePZPR@lAs-=?98?}gBKNysWp^& z(^#N!(1}PQ2S6oS3_3~AWZmA&ze0E7;45p!m70m@9Czg+1l{Y07b{CN2#mI#Jn4%` zM(Lo9JH>NCfQ5X+h4KSIlbur&+$xDCt4laqAP|l3^_&`tNUj)Td}R6okxX8ZnnZFk zbVu{CosDtHdQq~R;Uwn{nnnc23Fys%pkl<==HZ2mPk1?vV1N^&dfkt0*ULc^6+s-y z_>6sUT|?*|*`$z)l$i*Dj%Cd4SN7;gnHyMsDpZ|uX1DRbEf|CMe&|#LdN1s{QR2Zd zi7f1`zgD;s}JsOizh`4w>61J1GSM{2`*M4S?#^HCk+Iu z0mj;H2Ao7UjKs9r;yF;XR5PdbW{8cBn}K~iBk^W!IlV-eL$XSJ;{;`O4goK@3E3Wr zaCHADj)p)dt&d;?ogpvA%B1`S7f6=7acHpA2{8?oD+UlC-L83qU9aY5gPe>1y|Uoo zAE6cM!kq*x_J_jFCXP%ZORzcR=7YGC(tB%m(GuOq-wH6kP>DrvxQf+EB56NAy|R#dDDhhA{&Oqfmcx99zgk98U;*0 zowyHprZp}o8w!scdk$tq`BR0`G|W+hs3VON)W1Sl0#sjqByL$lD#NuEE+B51Al3C& zI6hXTlq(!*FEM!tjH3oaO4`(!Jf*9PwN1g)CKg}=($RQd#qJshcamUyWN$vm;?#wRQgpjjhb-eiUQq+zdM%(fp)MTSRAC{Z&l$J<@B~O8 zy0yw+Y}-Dxwh+^eXJZxws@i z6?A|m#9culQQ`rgd_GZ;03qm5^8ETFlM&`}36~9lNt+rrh3b^`Cf;^Ns@6d#hcn>o zoN&B_+UcmcG(EI^`VtVRb^@QMHV6DHYvllAwGW+w7e8c75H&fHD%nvzzjjcaOToFD z3@pkMaP5+54Tj}eyVSttuwOg!pE5vYSKuQgOpO?;X4cS=M_z#d61X_3=8^tlJiCN+ zl_=2xYyM>zs?Wo|%tj=ibQo8#py^l%D4^;>b(+Kx12DRs2qeZS1j=O$nBGy5`OdOG z%I`>M!HlO={t?8bYB}*p=o-*$y+h*MWl@E@>sZdMhs0## z5FuN|6O1J)^rp*0lZ-_bUSiPpCXmyHtV^%$Jr667I(#yUA-voQkyhe>C#iN?U@)IE z-o^O#L-b;DxJk0xfevTmjA^o(=#?ky=3MGu|NG)?5Fwtci z?MF?Jl;vmjlK|FIRL*qD7)+lXy^qNfH&~@r*u$}PL0_+ms^e$_MfHscve}5<`~;|1 zhl*;TdZEby{fE(+o)4_4AqU;7*+`XAVa7CunVlDAYjs{#xG4<6YH#Nq2#~2GHHwuu z>&gBSxB%EE7{YysRP@~8^Lj4 zO{!A1uG4;!r5^k;PUW=mvCJmu9Zen-V!qvb%R&CDu{o%vkiX|;%0x0DkZ%y36k;@; znVzKp5sayonBA+-Ek$M-LWA|Nk&?E)_dF_y*Sr&r6r@>2LT^3m8DE*AhiU@jFRb3> z@7cQ;vH$!Z(te4X3!<6xbvu9&KovBlqgMtxGr#FE+0nBLzeR5fLb-OI11-80v8OxU z>n|KWsO@Td_NM<(kNWL+(@&Y?FpV2%n8IMuH-S;?oBxs<;=dss>J?>1s5)1W^fPYR z%hBh(&V$Fn{{?^0MuW5=nn%?6!%$~QN=;+f4Xsgi4UP;@F11DSYfXiiEzu7{kqC>Z z26pI3A-e{tU;8lQ#=YvUD+ioTPJG5`lw%xYuewC!)^IKcwbhJ!e3*bwq~c`Ub;e~} z)kb2(c;&wb82m>dg}Et$)F{_FHIhX@8^sVPDz1T(Q`8Pr=3HOzEv$DtW!ZZa`^G$x zQa5TdqoN2QlDvLsbwQ@YT4Cd3WZA?{p(ej=prY`M*kd`}*W(otBFfbao8pxbi;4zS z5$NEMQBAyz;tFxFpmf_#&4GTBitc3Z4DbXALZ*(xJ`X>jQL;dWmBbAhKHp0t8tI@0 z^%=N={%N_EO~4Q;PTR#mFd!r4)5!tyy9_$jghJB1b<t)23 z%8X>hbq~EMSwZgxbxR1?E*h=`mJMs|7%=i2jlRVHhpCmo@@$Gh-Zd4K2QDwBE$*-p1HfTGz;-qK}T3B2nmvlCDz}iW?CH zcdO)HS6{dZpNuc1W&&D|91{i>;}u0$jgj>hRixDLmS^Tu48%nLsR1a_4~SMGKvle5 zzALn&{TW}Ll7KJQeSI?z{Gm-luDE1AA49mJfc|=bsf|PM-gV#cGv1P*1y2T#6g4sl zhO+RGE4&?FOaAEF$|?9OkHZZ?7myZU0CQd)Bsl=OLw&d=ydB_S%PJ(iTHT)Wwoe}G zBPiGKx8sa2N;v|@n}xm08ZGfcN7ovOBL>Qj2~^WiD8ELm&&(6_Ywc*dxZ? zD4eDshZ+rZtuxg9jL>Q0V^gUMw9}VI`GbM&1S!MdX5lCZNv7Z=Z$MuMNqSm$zG7TS zx(Ti%NC4Yr!h-ttB7(*zul_W*lsHL%WkxsVy{dTuBFp&QrTM>71W9;2L>z-4RQk5j z{{rfi$c;*LCYY8!U5>b!Xq6Ryz*oc;O1ga{xT*<}8!??U%7w~ANh6a-i9l{vhAR1G z>YVR%M(fJ|I3SQh5n<@mCiic6_Fx<6e+?kH^Hor}!GV+pEDa}g67OZDuZ_xA36rKa zKqQ|Ybvo~1z|r77g9y3@6b6kkS8C8hB3#fRM-?$K&W}P)3g0z9e$99Sk9E;J zjcYu?>#l}hZ$RK=k3-XmJQkD)$tQ1##p;f(ft8m)_^PkJgK(xVl~M0d7=*V41wfmEMstl4JY5}^!hSLF{mJyciNcs|jZ48^ZV~t28`oD-8LOeb( zV5e03qu#qqM!`*EIGI$`sq!32!$3&nY)`>j*Fm;u^cQtPS?gIgO1ZKm-}_dc8}4PY*KNLA^9gpgu$W zO0@qm3L+2^MO(nuj^bl#N@(?R>Zt3C=&`2Mx`(4K21p;V*#vvNccTkEb~NUTVQa*J zw>9JEp&}big3uSP0w;&p4LR8U003*ZNVYkMAc%d*R4jwsGVv_Q4F(u9Y|GVQmnQ#F zXuVtlNq!@>ykUJJZrB~ReIP>l4S-K6beq#xLuH|$^vR4bXR*@%4$2PMdt&M&lcj?m?@??B&Hy%H`_Tivbk(s4{s^#L4XN`MwHwbEYug^H_ zr7HZ>mqrUV-#;U4;`Ma;+1ak?P}M`fWNVd! zb(md;nRb14Kg{lj+5Ip(nfjl5V0IGm+l822hyUn0Xct{v>2DI@w`bgNZ#<#UBZs_Et}Djm!p2wHGr!%d=2_6F;$uL{ZJ;Ba2R}E@tN_9jH1zQ~ z*ii`>wMSF&#&R*FjxJDGJ5>O{p#7e%`1ts9H0|@~(WBK<$Z(H>YKbrL#e5bdt7Qfl{+v4bbn}28{6>$-9=8O;67g-Kd|lg*Vx- zr#61g>LZxwLl=S3fM@VuR`@m0BqGe>lYh#=N4KHS?+qf%(h(X&N{qoaAjbyozy{~f zzl9Nsw?S}kLrbIe@WjXSR;~cc$(P6{G+318KY@Nd!7V{I-ZN;Qs(&+vUQRU{)^p3g zqm0MRZ6; zm%6+?#u7C;{5h(rv61+_vwT@u2{7(X~FR~2ekylgY@kjT8ih+9#R zF)SjKY%B0s9oZ&KMq41+>h4g@LC0xx&(O@}+0l*BmX`s1lz`ZDP|F#$bf{R_7dikR zQ>ot6DlQ>W3?1G}a;+d2>M#0;#!K3vxn|YNW;78jLhX2C0+mFtoqm&x0E#sIeT6g5 z=%Pf-Mm&P!`HADwncsKp#R7aESSG1vvS@sv%29l!O8B;Hvo~Ng-l!P96*!dP-j)BV zr&b75zLeHnQEI-yez_2t2P^Q339n7JV97d(X;=Lk{VnFh4S`j1DB@K&H-~;#J8fZO z;|uWa#c0^XqrnD(Xg^i+I3O=CU!4;h+x;sO78v8M#XLqK%_L!+2*9U_ax!?@G~MxV z9{B89SvLB0X1CBD0M^0o!}v3oW;w-I(V*2t%|x95iKY?^NjwUPZh1GnLHQ)3bojCr z<1YW6i7~HeW)*Ja+j{I!NdH#=TX0Pm7}u zH_{OWq3bXavD}$U$(out``_C8&Zw%+wcBNkF~%5!iilDSc2E!$6ah66B%&xFRX|jv zC@p{j(oamj&;%Pc5DW-NZ_))k9*Tt$dXW;@0s;zd*)*lx`E2wA@|_>|jyuL3_ZtI# zL=C&F^{)4Q>U`!LQh5vu<5No6aP7kay%BR7_Pvl zmSF@0S}_3oZkVeM$d!f6BXLImlV+=0#J&TvTI>k+U_`0oeffS2!KYh9PV|_E^Z`70 z0mBuOigW=4m~Wa$CYT!mGc$Vp7w#NhDPz+Typ@L$Xm;ppG34}(p`8{U_DIKo@jU;6 zs|7jzn^m+r0?wytjqtc^`OWaz_5`#9t@6T61nS4FiJHbC4iL`FtFY9WC! zQh3ejn^s}#mfW26El>h=)YKgIna}*R z*_A85Ic+s|M|Y$3#^aLVacckLilxPAXSS9O!Tc727Qc-;+Q$ig#3rUa8s$5mohpr@ z>SDB7Pt^$mnc_zb`>LNa%K4BlUNW8&7WRmSZQuMk?OUM7Ns^Qmmqv#UK5$$Hu>N;w z0XjbBn$1JaHhSK9*gIBSkFL~1al?qeo9vMei_@olNWwT7n4L#0W)Ko^!gx>{0tc&p zx2bEmyNjn!a4*_^-_sDi^6N3uzV)fajM`Dy=~V#7G39`UN+ECg}{3? zMgA=~fc~N+`VJCDB=l15)|i39ur$mly$66bvV*5m7XUB$N{7LGz5p>N#Fk@bHq*TM zU25+k$UM42;?*DIrM!b^_02NM=tsR!VIGcr(d4gTJJ*6Q6uSz+;sC>j1Et zix3U%_%jPlwBoYca`Sr-D{do*)$EWQQm#hhTgipMjL%W-KR*qA}gFt>r8BYCB4 zR99=(hl0?XI)oqGifN({O%^$yLAd=zgcXj{&qz@9z0i2@rj3<-w%mO&CzjJnsa8q<2d3j&7Hnhik0UIgcu zL7vOQ!zPVZbYqOc3PNUW7`>AdBk$VT7ztc*TLBZ5^nB*J76V4g@g@rAVu&3-m+zas zQ2twLy#-C@1uGVZOm8>bT}kM!T?rUkWAZB4yh%A5sEOe4&?;f9Kt9M&m74Dj~-EJoa`-{Q%&<4p;Lomgh z2rVDLAUWXO%b#7B2msE3l~hVBRl0eu=ZGl;jrW}LPRy3`fWOs}> zyy5oTMLOErOOa@(A3nSodhrz4QgQ-EbT2N>5AEmz93TGncB?-54^Y$ve#WZK6>Vb! zLPdcBLx`sFr{rzwdV2gYDFIl4w2>B4wzUwS`d~fBc>2^gV=c~q_2Qe~kQ>m>f!5_J zg|DW*?BT4mvCiqo!*guWbjz|^iY+wtJS8=CT#w+2WG|iclnOOK6=4rXA3KDUmX;<%xllL9!D{Q) ztu#{J2I_MkjQMOjs4!5yYR5EswB7sLau;QQ3@sCl78YjUyc-)E54lCAq)3t0O(rPu z!q7zc?Q9;_5uyP}e6VZx7SLGvK?LwSNJ(U`7Zckb`U3PR-k_9lKj8sFovyxV z(>4AJlm|Pjw}<^-!+^TKCPCWK@`=$y-M6ROqKC=l5`BU9z>ZN2eXV8mstR}Lww(Zs zsREre-CXChaYxc(-Z~Bp3`kv}#zG6x{IDK)PkaA|Q2=(G1@s^BfqZ@a*jD^-U`<>$4FcAMH3@9^e~%F=kne z_AT{CXqJvI1kEu$1I)*5u%{lxH3=gj9(B&N$ve;4Jbi(8h^CIze06C?GX!Kbq%C{( zI^TqZB(+L_oFe_iBpo9oVaz_)eHYXRDoZc4EO=|db?O5_$zpVid_#HwB%Crq% zAZfyKu~&-FJE}6Xu(b38h=%t~Ddeb$r?PwtOUtl#qbxLX_J#IASx(&N zvvTloZpR~8=+p}`B_H%E`D$r;9F3V5cT6)eN@@&b1P@OI07f?yTn4{yo;Hhkm0Dj! z4eMJAxm7>F@Y1?pV_e&}7SJ!nMgw6BS)*r;y~KEV0`r9*a$}{tdwN);lXV`jeMbeSQx6(#x~$2d$Mq_$r-_UvVTRthV{|rr!A( z+P`MkW^2c%73!-eAFEM$b9PVsAKJ6h)_>*oSk#wy{x8xgtL-P=Za2C7t?6Tl&mVW% zA1r(Fp)5!K3QJbrEVyq{@2cAJk&AQ*&0R_o?4;IZf1)6$Fs@#`x>~J9M@2=Y4f{9# zZJE#2qaT3EXg}j28t5}$JaZ-ZAFu4}10JBXYx0vm9zoA9U*s+)77l(E&;+Kr8R$z^ zfZvYegr1}%P+NCZe9=yfI&aUz0}wy}PzJZ6=`&q z4JI6Mw8rdqOk4DoKR6}s)k{911O|f5#tDG^$lk61^3IP9g#l5+y2!F_?9_uF_V2n* zX9fo?wJ`^jlZ0`rVv)6q`g7$pqfzwB(CLfVh^Bt}_sw~LOnBVo2manI#-jjLYGx-S z1|}qO41=nGxAL?3Vq;&4Dlv$oZOMzB0qQl{mUh|GjOsCCtnH7-j~_3}GT#Cuso2O4Qf@7A&}Y zeXd#i)4;%(Q^qywt>AxSVLbIvz}yAPmjG;1KGJfRXr%I>Vj4Vnb=={fgfGKExUz9R zldeZt(YCfWKTNCX@3oo3TolL!J-r=An_-?*01R|k%dBj#Q{opJ1pS`_ZECkO{LJ*kHAcIi&JD_~z+SSW!paFyH1t|RX__MdqQHUljv_2_-H zrs(OqT8vd%&YSif%xYzuIlEQU@bc3+a`}zc3IF&4*M9EYIjwP;xQpw4^6c5O5n0P8 zA)B;n1}=aVJ_Mq%oMFN)0utKpx^C(M?LI{d+;!}Ty1IIq+0~XSC+n>&YBLDSh$X0E zT{*N<)*nN{!F|&<&-#GA%=qtr85-{aRe&GZBsPkO0L`_Okg;WsTE7zN(H#(uj{9a9 z;;evD7bg?5&&g1TS#7}2SO&>h)?Z|o=Fs+SOiRnedj9aA)>9h5lg8dbi8tD32uz3} zUiJ1N0PJF$Mk*(Mcf;7aY-lbR1N>}Z6zz8!A5PbH!+5SY98Xj+kzv0EX@qR67VnXC zOa4vz>2cNkI3nIV&UUZXfiFNY zTF=K|=nfSP;}*T7Eov2E+*5WM1H8){qQ@UmF6vg-0fgsTJnPGnbr_SB>Y>s2@iCC1 zd66Uu9pJCZk3vW8W>%#OuIpx1>ofw!dfRz@H7fy%^@l+(29bTGJ@`Th6l>~M1URKs z5ul3)(qFsUd8TS zn2j66)nqJ>DgN~vHp_9A+!Vugi;4P#tBx5157bitSScZxBidVz*d+3)d*!dUEs z0qYLj_!hr^4&K_n{9QgiK9ww@N*RQKF6f!I`yc<;+oKE4fXyI~KOaO@K?eB<6A7^@ z)j_ad)mvMm--aMe+I#fq(V}tyd&QcnU{o_{9=(IOM|2UatQ`U@M$ROp3ar^jUMNgm z5=Ltuj0kqQ4k+eUDgZoZNKF3(=w7pC4Zp>@gTc$@%$Xwz=p4rYoQEv9mH3qR1R601 z4$)a2%d&p`dM_M+b;gGK_U$8vYJfOa^wh@hFJCCIeKU9l^h^e(;-neNgKxUUJ-a2r zKYRnsoDYC$1E?9+c^UAY6d>6`pmi^BLO5Ui1lsX)!~XN7>G$OzS2jtLwyfVwY*+Nn z?a!_&D;99Ns#qZB^TpQPix7|nAL;kx$~wZHNDT^sOPr03p{X~W?y}+Ame}8y%UCWW zP&|Qy*`TqA>I7?wFsM@`97(~&7;$JaD5A8_xdBe(4Y?pLWEr7_gY!6$!hQ&?lbRHW zQ9&M>>RQV;PU9?jychd^_W-dUlW)ENuJP?a*=0XDi_Pi-h-NW@4;E~5^XUPKIze9} z0FF6-Bnwt{hM=iqagj6pTh`+EX4Q00+Z^>vS|8~AtG&IwkV0o7i9_tQi6BY@ZAUz) zgr`6Uiy&B!%gYry8?aG-jL-;-Nd&_T`~T|$F{6#TSoe~w?e1zy9UwkGdEod4Hd=)QCCw_YmS)qGE41Vqo`>>P)74dr)h_W zS+sUKx|o@__-~4czs&m^H0IGS)4|Ovk3Vvm_PrN3|80x^c;oG-cy029s}Iav#$QV| zvl@S`1!|=;dt=IS%&f+g#hBR}Qx{`qHD*@h(+bRZ*iS1#f6lDN%xZjGftgrz>QxB@d-+tjTvvl=t2@o@!q|M&VvT5Jw;$F4TPFi0{4qPx;spE1y08|X2{2w0ab zH2=cDz*?ir!#o_^|DP(`fn|+X%i1*8yilkwE5*c@aTAK<-5aLWsnTsx@d|(}vsGZu zyNsvP$N?C&^I!hG2=fj#(U@#sCfxuX6DSXhD_El}UX7;Pa@LFCqcN1<)iDEJ59ShMaHRU8MXY0i z!f$)@lUyzILqRWHfD$>)h^y=BE<-oaA9J9EfKl)hZ{}7s&E6gQoT;~l%E7u8Xf){F zH&>^w-jTG^yeLlVELpNd5tNrPY3b-Q6@sN<5nY#;SF>ZQ!VZVm%Sag{Y-q{SrNk)B zOTQzc0bG_ZT~qj-JG*T@XI|c2JERR}WG@hH6rpY(o|LqCvav;yiKK(WzF_0J>XHlN zc`F<*Wz?!@6DMii)rcmBl2Ir;dSl4+PLnhSM*YD)v4tkx&Cv$ej<>0a(Xs{8EXX3{ zc~A=$bu67VSKwO+?HywMJYU@F!Gi|{7Q>+vBcU|a`m`iUBf0>ELKL_VJ+5V_R*UsJ{0Nxaf z!PJZZl*J3b`%OTjFwm_Xz!`r~_R^+GJ<^Jyc+9UV-nVR!N?r~bOX}eUELb6P_1uAH z-!PVOct%vX$J3MZ@%6P&h&3w@A>4@aZ{G;owajNMa|i%LqxVGY8#@5}jfFgjWo@g~ zm9=b5ctTy@?IYb<8i0?h^#FH-A1=m9J^vlIOpn_I8fDD14#oLx))4lg=D0od!RnK1 zs$05g)I2sJJEm#j?)qTt_+oUN@@kH{%x2PsrD08LM@RaN!AlXqE2fPrc3S8uc!8?o zTBp(IXM^lGw!~KfGDXTHuaOMgQ6U6+1b_x8b$WmY$7S;YBS!4~JP=G2f^$OX*I$3- z?&T#!gVRL(JAlBtpSZv z8#L)QZE3~9cm?q*SJsMFuhG#_Viq9I9nt=d0nCLLfsL8QrS%6tpXF(#Vry$lGwbX& zUkTF^4|)q)klSxXj^?Nkw0q3>1}qI zANn1PG8v znY=yReHuBWO%&3VJs@_tk&Tr&47|<0KwwDpa?pIOCK_%{eD>2(K{O^8xwJ652e|as z7`k!ZevAj*m2oq~(+&e+Q=*%LUi)1%k2Keu<>ljx%69Avu{8o~mIjSj-%OLp z8CSs)CeNmcby=huwo z#A!Plu!ZYP_MW4OqwWxx_hIF8{NEk-}jsc2i~1E8fUKxwck;W%OzTTJa2 zXP@m65oyC}dF>a^T5Wvcnjqu#hrW{QK&eZp$M`woN=BbFNXrmv$F8K%q;6mk91{B! zI$=p?sT&96z}1TVWQq@NfANc59Um{R5-4trM~B&6S;V_81ktqu3zs(Ub!{-j2Ld48 zEkvqEG9#y6pe^eSPOl)MisCdqzO;%30}+MRzc^qnTP&y*fB&R0C|YDz!xkc7M+)!oi^zSSR*;=_2N;hlxx+r zB5=v8=Cqu)2N9G2kOThsti6f_IH1eWz8F7rnAV2!1Bo}>;&|J__gAHD#wk7fd#y4p?VB@sjlG5C-721x2U+dKUf>q5qEgV zik+@b#1?@|qrU;Y)@4iWW4KY?Jlj!x(gmW_P-uY_H3mhwIuK^6y?Q_;3ON1w5B`7T z6Ke%aRiD{idbr9Z7McxPv6B|#D9F|f;dz+4t~Wz$Yzz)Ab-+V|As6egQ93uHF_;A7 z+1LO8C&9?IfxCTU|35X*6n$-;{@O?t55*5$CngWF9^j1FLhr!Z!3Z)4ZD^P8512oH zK5L{i^w5Jm;(g+`5$^-djcKca5RWZ%0}8>;byX-I1R#=NW8W0>0*q$+=MH^@-%Kop zn$YWEj(8~+)pPV41OEWM#M$ri0fbHkq~+^OkGql*>k9T$39Y6kw0%^TdYk@?0Vvt!@_00 zPI)FOKbN_uherw_F|)FN=6ZS2`m)mEA3b{HglBHI6TZ>*RlMjRk@AcUADyBtrS<7ac8RJ93E8k>t_ zsBDwgPHu$j27>E?nLDOc`c{o8CeNgH{$&65&wnNZThr-^P$Ymlm1_a zvSo>&J5lMK#l##naSjJ%lo@9ZaR#{}lG_(}&Ae4x-H6j2o46O=*?AOXf?FI~&UME1 zo+57uL2m9Nzv=!Bb5E9|9mtYi#~d?uj#bTR^AT%nofvmibRAGV4~{>#GBYHxD(KF8 zigIxuwxki>@*L039M0SM0USFKlW0T~A}HIkvpohoy8`5XaYKAdmYnp%!MIg5MwBa1 z#%}TQ@}hVUvUsgZh+hwyF^sftON`xD^XD5sxx9wo9$rpmRERm&bMJsnHK2m8z#mj( zpfOgAKaFv!`Rzdz_1M{?EAJJ+X$>!f^2PeMt?x^SdG$}m^Ue8|)6XI_q6rC7^yzs7 z_93qIrlTw(0vsflt_Ml6M9&zLluOWtm9|9q2k{Qmk~Evvol#pzV^1;!YON7Hjn#Sx zP9C4~=<1ZhansGRGtqY8+o=03!c!|}SMbHXD8;hXLqkaU{AVvJUj$v0S}HZobOB5ZH!tf3`i${Z5N0;O##>KU;v6l@7oI* zOT!440N^1N@pAhe;W5qu@eO=hg9$OO^RYR8!{$&R z%GGu|5hgWW`*6Mc-Wx(p^_TA=u$!Wy0K{=6NuH-~j;G*1C{geh8~eGO+ySo3=PH$s~uKZNH2tz8@Vj7moi!Y)GqeOMMP|!-4&@4x`5FPwMrXW z-xnMc7oXy&W#aKAW4Vi?ouvMKulEh>4;yY%jK_gtV-V_(OF6oWX=fQ*A?SP2Z6Y4 zcWFe%(^7!*ST~IQy#X}atWXCtC}@En2R`%z;q9w zNYa%B8^a}?(n&nZeGjX?_WgD4dqDLjObYEw`1q8$;e%9*+#E*w689pXUo-L(?Td?v z>-$g0+gXfsP>ZMwbh=EaMn`N`5uM$2iNd5Y7LMds|IAukUc)u#fJ!(d&o>uZMCE`f zw1t#f0i{?R17NAS889MP>x!H3ac276Mk>#5r9s}!s%v3G+@#Ci_^w}RDRZykG`F<1 zx(^=N3pnBMr+p38Q9sI%=N(01;-%{#FxFd&xChElwc*IX1&UY~3Wo@_LM%HOnL~#w zAETDb(dK=w-A&JN6k2MJhdKR}FJfYX1zE4hBl66#=bG6=x~ZnB#4-zf;Ee%(t(p~9 zrFr*I@Zm;6d9k8g)A@Tw;{#1mUSb3TbL`@-EF?8f5V39p82%Z3QgK9v&kZRPl_t6) zHbh-02`?WJXX-1+fWG!b9hZ+3$QDQ_7Ou=rHIO3!>GwR$NE^sx5S7L zo}3WYq=>{ZK>4Z-^A!H187qrW&GI?PM42P8X;CurViH z%H#Z(i8iyCiHHo*R%~DvF0=GT4gkfhlxO14-ha*YM6;N7Dl^|v;R+)z3xJ+-Pi{n= z5R|FmQVn{Jz@l%5exlfeR1y4d%vR0QoWmsHI#XzD1^JTqrB%w2o$vbljiIDu3oZDQ zZoA;Qq7T?`PzPB>D1ISf@Zr%pUaNH^%uQd(O&t|6N>sbpR8imtOyo1AP@cOgge@V5 z({=sqZ7d@q+1dr$n!csljZ_E8jT1yVb?0p6-c8I@>;{UPVrI&1i>POT%J7Yv7GV2z zoU|9IgpQyfX_{t(z=4gLp;^fQIDDJZ`tm2-QH~9E$2L9O5nO2@b5@v1VpZy!s6=Sk zLM5*VX?92mh9rD(L<;%Mf-luHW9fS!@auz1iz=q^HNwHCWyH9IS9U7EhG4rrXmBem zER;M^;1y?KfvAmWe!0+W%9=`Iib#GMmOE6)qes8h2W3~YG(M)D{VHAxs2jGm$5bk9|-EcD;%Ps0Qq~o`zkQf%z@`?-co!e67UGReO zSWVBnuj2FUP_-&V3Gu4r5I8*xfYHuupiIsfT40pJbDXf#P4X%njk!(~B?B^-oF;q{ zA8OHwZ$m2GDv&gb8Bj587}GnKBpurWb+qdbl|oaH7`FYgh(r5bzu34@7AeP5Ew+n| z;Qv8`%`YPjTRrPLTei)G^rsI{`TFZ;=(NlvDOXnw6rl+~@#x$jD+)!2A9D^UqEQ z&ZKI|{Gjd>PKsg@R~Jl~1h70{!z&4GGm4I+I_N{TBRn$3(#npZ&l2;eDlKZHHVqrm`SxB^ggFMt8Jn;p z)Uo)%A#|{#QMRUPi&XgpP3K}0w~yXBj0R|X$Yv&~8C7kpq4rVHKou=V9)@|HshxOK zka*3hENz^Jz~sLTHWb6xErk^*ALhJOy@}NJHNKkP9w+pPQ~z)z;rYM%>MJ7q)fJv|BCuX;#l=U>q_UcBHUj zzlsoF+3=Dl(r<> zZWb45qnvuD+j~ogR;a@-B@$&SQyII%k$L}BQkMn=`%ShJrbCP|K}qJ?deA?SLOOPT z=!Cp+0?JSoSe1I3iGb$2FXb>OHA~e(O;gY$&+rT8UKWbJH6Z>3vwxFm&tobNp1`T` z4o0YwkKd)OQQ39fa-wxLV<|Bi7QU|k-x%c3%guU_0#)EeRG$fk=$j1i;*sG9_60_4t`}JqdRVcqWAp=iw zo~>ze|8$M$QJ9781J#GO>~wm8aLb#L;U)!S3BL5s&TYG~_0C?pew}a6e|_zZ76YPg zjKlc}%B43A(70HTO`hIQ3z^4vK6T3Lbw36IHk8hi*aDS(QK{%zTL+&m`icA>cIHfM z4Jq#4fS@|W$hTvor#(4Ko2GTF*O4ZN%^OnMwPP1j@de%YiPJdP)Qh5_Jw(67YwN(H zF7N*CN@xKzd)uHl(4rk3k3>Tf9P}hIL8Bx=26YhEq<_)VrPltY>>gtDrm8Cys_$Xr zWLbb+cE!5jpMD^|XQHg#luezgu6Tyqy=L;RU2h!DahSy%6T5-h2T4HF@ClMdU-Y+> z;RV9bvAGm#lp87m+p`GL(_XYUQECW-jjTX%y&j#}aByzd99aSPO>>bGROF{o>5WJ3 zucxDj(H8yJAQiTSYJ!bab$S|C%7Nr7@HrwPVK8u1JQ_wT_2w%D2r~Gh-Y5nivuyvh zjoyY14Qm|W5@^T>?X5M=eB8M9BkVx>IyucLKda-McR5d+C}kXW&y1%0wL<^@ft=&s*~a){ z;xXT-o3xT>129UzpQ_4mZZu>XUEKg5#Z^*)^!akCu^}ktA{wXz%4ul7i6SNOPt&k5 zCIa=STn&Sq@d}5E(k}Lg$(#G}+NWKe)NBq5UV=uKhzQ(3B9&bg6=_D&7L)^p=s$gn zDwRZPA_|MP6zHIyw-^SkysM06bjd8iIh=8YnKTiyK@dyL5nD)f6T-*d{oo3H5IZ#C z46~ur`yE_H)|0by1<$CFnmdJtkDonz{xbXULD@CqoTyz9#mWMIQo>U`r zmtG;nqZGdViyG{;h|7ORlJ2&9)uF3!E<>iFPD+~sZZeME|q;Q;&nXw@L%}t_d zrdl-t!X{&_uh18D9);XN0TPa+h9ii$xpC@0bHf96?jR`!GxmYW3-$UIITQCM-I!L3 z!cJTn?nMk}I%?NkKd z8`NPvy}U@pMF80s(LDr*7Ncy`MnxnvSj#N5Fnlw)Wo80ht8 zU05#d1OF~Vg^!RAAY`q{|*|gP$&(*dv`TleH-v=+Qdt5`|9`KOyeX(MSX&-vp2YpJTo#a zAx?n+^cZYXBGJkrGj9x3L01-qKJao&Kp#-5ORef<=mE+j`Nv2$QniuJe)YTq1%ppq2zqmbbh6Zy(@ zLkGR0(LOXgzbel|FKx@uE;LF?y{!X3Nr;@l@4e>=Ijy9q)62{6&z25E!J;ILb4E=Z zS;sh*e;X*#)vfXBFf?=sjX@{Z*%}_4O_BgthQJ82nF<3!e5h!WL>1y^{vL9wsC<-s zxrl%>OA-FlPyJ_~{L({*Sk#heHdYa@7c+1~g_Q62m%!8ZZSX^ zPrr<|dX2F50;=W_#p`aL(GWQE$M2V51XGTU7v+uIUcrdM3TY$}4|a$Q>Ow zj8l$2NBPK)9r?(G(uXZYwA}CgQQ6--_}nHxuGSs>Vm9h&k;UkA4TS^Kqu0vIV_U$Z zeAmMW?O%$JcoSNoMNgDBmqI66C$wBEl=uqK$Q7Wvk#hn}EVsdna=Ts3W|5D8Z*3ZN*9OV}rQjnc)yn5~h9#-$M=?5>^RKG;0ULu8V&X(hY6!(yoLH zdz^T9sLzVyrjyx;ZQy5wu6mcv2&&UQ&_0tiE~SKr$YTzsO@7;yry?Hs4a*8Ecvh;2 ze1T&#(&3QtTx22?pau6nk>SXok>$sxe2Y*T2OR^`(btKq9m7%uK%}b}urJ+?Wav*w zL6kvQGv*^3j>x-h`5*DA$DSX4uitq&g@6own4T{?=K;uT$`N7;2rh344iwWdO`~H% z$49E45TcYv?7Kd>I%=z^MZo>B`~ewV;O$%lkOk!jA~f{k#n1``I^%B1D9w*!!4G6_ zwxu~nEo-#?KD10OtOmJD0`I;u|O!NwD+@Zbltk{g%$6|!E(#!<3Lo{W=;{D{;uwkeF` zXbX~AM|s{R(3$&TjDzN>*W*lp-d|lkI(w;7k1-6RVcb`7<23qKZ*cvqhbt7*I2@Ce(I^3gN6L+RxNM zQ7@4F#)W8X2)DsTU996w8AVb+E>*4-_8=`C0PKLUk+E710-^)Nb;I97eq*@nkjJf0 zHrFm;@^CVrB1BIqdZOl7J!j&}A)8Gd8IxMlf}_GmCKOy&iG-oDuh&|IES_Y0!oa7| zm4#ZZBC_%}n5($WTd6lti11-K%#-VVi#qWTAO-E}R6}{!#%xQxF2MwCk0TC?P8iyH6i{CKNvr; zZ?^yGNzalfdmu4J!Cruu7a36SG%c?)uve{hbnBFBCH|9Qs#pObbHU4tkh&1{@Y{|W zJnPQ+Z+s}73*UR;1X=*ZaE5To8LT%>sb=sDJLt?80K!$tEw!N}VSUAgGDnJq2sbdv zViO?$?neqwXnVzlqAHG`-Ky(Qxa0n)`6EOtK z;u1mpky7X=NbxA378FErAa`}TE@3!Tx4*xig+iA82Whjr2PK>jGrl#iF$CE(thmq|OdE++}KVFW^DG(4L+L+?u+zI2T;n404w!tSzE;&zasP+2z};gNJh`KXCP)PcHf67c9DF z^lvvb`M)0e@BRC0{wI(;3{MQc(WN^$b}@fowZk=K{B`pcGzF&o99#Rzl}!H6hX+1| z@00)by!mNQetgrGbJOWAOy5nf8I-*I7(sq~{f#HnsVrtLVDtQo=7$FNocIO+eumkn MbWkzjhhsngF9xniQUCw| From 4c350bbfbd4d923ef371d0793e00242e9e07ddb3 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 22 Aug 2025 20:18:48 +0000 Subject: [PATCH 25/32] documentation commit --- analyze_graph_type_comparison.py | 462 ------------------------ analyze_validation_errors.py | 452 ----------------------- bond_length_issues_conditional_traj.png | Bin 77585 -> 0 bytes docs/pretrained_spatiotemporal.md | 73 +++- graph_type_comparison_3d_histogram.csv | 17 - graph_type_comparison_3d_histogram.png | Bin 646402 -> 0 bytes test_wandb_scraping.py | 65 ---- validation_errors_plot.csv | 10 - validation_errors_plot.png | Bin 169982 -> 0 bytes 9 files changed, 71 insertions(+), 1008 deletions(-) delete mode 100644 analyze_graph_type_comparison.py delete mode 100644 analyze_validation_errors.py delete mode 100644 bond_length_issues_conditional_traj.png delete mode 100644 graph_type_comparison_3d_histogram.csv delete mode 100644 graph_type_comparison_3d_histogram.png delete mode 100644 test_wandb_scraping.py delete mode 100644 validation_errors_plot.csv delete mode 100644 validation_errors_plot.png diff --git a/analyze_graph_type_comparison.py b/analyze_graph_type_comparison.py deleted file mode 100644 index cdfd7ce..0000000 --- a/analyze_graph_type_comparison.py +++ /dev/null @@ -1,462 +0,0 @@ -#!/usr/bin/env python3 -""" -Script to analyze validation errors for runs from the graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17 group. - -This script: -1. Scrapes all runs from the specified WandB group -2. For each run, extracts lag_subsample_rate and total_lag_time -3. Loads validation data with the same parameters and max_datasets=1 -4. Computes validation errors for each model -5. Creates a 3D histogram plot with lag_subsample_rate (x), total_lag_time (y), and validation error (height) -""" - -import os -import sys -import logging -import numpy as np -import matplotlib.pyplot as plt -from mpl_toolkits.mplot3d import Axes3D -import torch -import wandb -from pathlib import Path -from typing import List, Dict, Any, Optional, Tuple -import pandas as pd -from tqdm import tqdm - -# Add jamun to path -sys.path.insert(0, '/homefs/home/sules/jamun/src') - -import jamun -from jamun.model.denoiser_conditional import Denoiser -from jamun.utils.checkpoint import find_checkpoint, get_wandb_run_config -from jamun.data import parse_datasets_from_directory, parse_repeated_position_datasets_from_directory -from jamun.data._dloader import MDtrajDataModule -import torch_geometric - -# Setup logging -logging.basicConfig(level=logging.INFO) -logger = logging.getLogger(__name__) - -def get_data_path(): - """Get the data path from environment or common locations.""" - data_path = os.getenv("JAMUN_DATA_PATH") - if data_path is None: - # Try common locations - possible_paths = [ - "/data/bucket/kleinhej/", - "/data2/sules/", - "/homefs/home/sules/data/" - ] - for path in possible_paths: - if Path(path).exists(): - data_path = path - break - - if data_path is None: - raise ValueError("JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable.") - - logger.info(f"Using data path: {data_path}") - return data_path - -def scrape_wandb_runs(group_name: str, project: str = "sule-shashank/jamun") -> List[wandb.Api.run]: - """Scrape all runs from the specified WandB group.""" - logger.info(f"Scraping runs from group: {group_name}") - - api = wandb.Api() - runs = api.runs(project, filters={'group': group_name}) - runs_list = list(runs) - - logger.info(f"Found {len(runs_list)} runs in group '{group_name}'") - - return runs_list - -def extract_run_parameters(runs: List[wandb.Api.run]) -> List[Dict[str, Any]]: - """Extract lag_subsample_rate and total_lag_time from each run.""" - logger.info("Extracting lag_subsample_rate and total_lag_time from runs...") - - run_params = [] - - for run in tqdm(runs, desc="Processing runs", unit="run"): - try: - config = run.config - if 'cfg' not in config: - logger.warning(f"Run {run.name} missing 'cfg' in config") - continue - - cfg = config['cfg'] - - # Extract lag_subsample_rate - lag_subsample_rate = None - try: - lag_subsample_rate = cfg['data']['datamodule']['datasets']['train']['lag_subsample_rate'] - except (KeyError, TypeError): - try: - # Try alternative path - lag_subsample_rate = cfg['data']['datamodule']['datasets']['lag_subsample_rate'] - except (KeyError, TypeError): - logger.warning(f"Could not extract lag_subsample_rate for run {run.name}") - continue - - # Extract total_lag_time - total_lag_time = None - try: - total_lag_time = cfg['data']['datamodule']['datasets']['train']['total_lag_time'] - except (KeyError, TypeError): - try: - # Try alternative path - total_lag_time = cfg['data']['datamodule']['datasets']['total_lag_time'] - except (KeyError, TypeError): - logger.warning(f"Could not extract total_lag_time for run {run.name}") - continue - - # Extract model target for filtering - model_target = cfg.get('model', {}).get('_target_') - - # Extract data target for filtering - data_target = None - try: - data_target = cfg['data']['datamodule']['datasets']['train']['_target_'] - except (KeyError, TypeError): - try: - data_target = cfg['data']['datamodule']['datasets']['_target_'] - except (KeyError, TypeError): - logger.warning(f"Could not extract data target for run {run.name}") - - run_info = { - 'run': run, - 'run_path': '/'.join(run.path), - 'run_name': run.name, - 'lag_subsample_rate': lag_subsample_rate, - 'total_lag_time': total_lag_time, - 'model_target': model_target, - 'data_target': data_target, - 'cfg': cfg - } - - run_params.append(run_info) - logger.info(f"Added run: {run.name} (lag_subsample_rate={lag_subsample_rate}, total_lag_time={total_lag_time})") - - except Exception as e: - logger.warning(f"Error processing run {run.name}: {e}") - continue - - logger.info(f"Successfully extracted parameters from {len(run_params)} runs") - return run_params - -def load_validation_data(total_lag_time: int, lag_subsample_rate: int, val_root: str = "/data2/sules/ALA_ALA_enhanced_full_grid/val/") -> torch_geometric.loader.DataLoader: - """Load validation data for error computation with specific lag parameters.""" - logger.info(f"Loading validation data with total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") - - if not os.path.exists(val_root): - raise ValueError(f"Validation directory not found: {val_root}") - - try: - datasets = parse_repeated_position_datasets_from_directory( - root=val_root, - traj_pattern="^(.*).xtc", - pdb_pattern="^(.*).pdb", - filter_codes=None, - as_iterable=False, - subsample=1, - total_lag_time=total_lag_time, - lag_subsample_rate=lag_subsample_rate, - max_datasets=10 # Use 10 datasets for validation - ) - - if not datasets: - raise ValueError("No validation datasets found") - - # Create data module - data_module = MDtrajDataModule( - datasets={'val': datasets}, - batch_size=32, - num_workers=0, - persistent_workers=False - ) - data_module.setup('val') - - val_loader = data_module.val_dataloader() - logger.info(f"Loaded validation data with {len(datasets)} datasets") - - return val_loader - - except Exception as e: - logger.error(f"Error loading validation data: {e}") - raise - -def load_model_and_compute_scaled_rmse(run_info: Dict[str, Any], val_loader: torch_geometric.loader.DataLoader) -> float: - """Load a model from checkpoint and compute validation scaled RMSE.""" - run_path = run_info['run_path'] - run_name = run_info['run_name'] - - logger.info(f"Loading model for run: {run_name}") - - try: - # Find checkpoint - checkpoint_path = find_checkpoint( - wandb_train_run_path=run_path, - checkpoint_type="last" - ) - - # Load model - model = Denoiser.load_from_checkpoint(checkpoint_path, strict=False) - model.eval() - - # Move to device - device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') - model = model.to(device) - - logger.info(f"Model loaded successfully for run: {run_name}") - - # Compute validation scaled RMSE - total_scaled_rmse = 0.0 - total_batches = 0 - - with torch.no_grad(): - for batch in tqdm(val_loader, desc="Computing validation", leave=False, unit="batch"): - batch = batch.to(device) - - # Use the model's validation logic - sigma = model.sigma_distribution.sample().to(device) - loss, aux = model.noise_and_compute_loss( - batch, - sigma, - align_noisy_input=model.align_noisy_input_during_evaluation - ) - - # Extract scaled RMSE from aux dictionary - if 'scaled_rmsd' in aux: - scaled_rmse_value = aux['scaled_rmsd'].mean().item() - total_scaled_rmse += scaled_rmse_value - elif 'rmsd' in aux and 'mse' in aux: - # Compute scaled RMSE manually if not available - rmsd_value = aux['rmsd'].mean().item() - # scaled_rmsd = rmsd / (sigma * sqrt(D)) - but we'll use rmsd as fallback - total_scaled_rmse += rmsd_value - logger.warning(f"Using RMSD instead of scaled RMSD for {run_name}") - else: - logger.warning(f"Scaled RMSD not found in aux dictionary for {run_name}. Available keys: {list(aux.keys())}") - # Fallback to loss if scaled RMSD not available - total_scaled_rmse += loss.mean().item() - - total_batches += 1 - - avg_scaled_rmse = total_scaled_rmse / total_batches if total_batches > 0 else float('inf') - logger.info(f"Validation scaled RMSE for {run_name}: {avg_scaled_rmse:.6f}") - - return avg_scaled_rmse - - except Exception as e: - logger.error(f"Error computing validation error for run {run_name}: {e}") - return float('inf') - -def create_3d_histogram(results: List[Dict[str, Any]], output_path: str = "graph_type_comparison_3d_histogram.png"): - """Create a 3D histogram plot of validation scaled RMSE errors.""" - logger.info("Creating 3D histogram plot...") - - # Convert to DataFrame for easier manipulation - df = pd.DataFrame(results) - - # Debug: Print available columns and data ranges - logger.info(f"Available DataFrame columns: {list(df.columns)}") - logger.info(f"Lag subsample rate range: {df['lag_subsample_rate'].min()} - {df['lag_subsample_rate'].max()}") - logger.info(f"Total lag time range: {df['total_lag_time'].min()} - {df['total_lag_time'].max()}") - logger.info(f"Validation scaled RMSE range: {df['validation_scaled_rmse'].min()} - {df['validation_scaled_rmse'].max()}") - - # Get unique values for x and y axes - unique_lag_subsample_rates = sorted(df['lag_subsample_rate'].unique()) - unique_total_lag_times = sorted(df['total_lag_time'].unique(), reverse=True) # Reverse order: 8 to 2 - - logger.info(f"Unique lag subsample rates: {unique_lag_subsample_rates}") - logger.info(f"Unique total lag times: {unique_total_lag_times}") - - # Create meshgrid for positioning bars - x_positions = np.arange(len(unique_lag_subsample_rates)) - y_positions = np.arange(len(unique_total_lag_times)) - X, Y = np.meshgrid(x_positions, y_positions, indexing='ij') - - # Initialize heights array - heights = np.zeros((len(unique_lag_subsample_rates), len(unique_total_lag_times))) - - # Fill heights array with validation scaled RMSE values - for _, row in df.iterrows(): - x_idx = unique_lag_subsample_rates.index(row['lag_subsample_rate']) - y_idx = unique_total_lag_times.index(row['total_lag_time']) - heights[x_idx, y_idx] = row['validation_scaled_rmse'] - - # Create 3D plot - fig = plt.figure(figsize=(10, 10)) - ax = fig.add_subplot(111, projection='3d') - - # Flatten arrays for bar3d - x_flat = X.flatten() - y_flat = Y.flatten() - z_bottom = np.zeros_like(x_flat) - heights_flat = heights.flatten() - - # Create bars with proper color normalization - dx = dy = 0.8 # Width of bars - - # Create proper normalization for colors - from matplotlib.colors import Normalize - norm = Normalize(vmin=heights_flat.min(), vmax=heights_flat.max()) - colors = plt.cm.viridis(norm(heights_flat)) - - ax.bar3d(x_flat, y_flat, z_bottom, dx, dy, heights_flat, color=colors, alpha=0.8, edgecolor='black', linewidth=0.5) - - # Set labels and title with font size 16 for axis labels, 14 for tick labels - ax.set_xlabel('Lag Subsample Rate', fontsize=16) - ax.set_ylabel('Total Lag Time', fontsize=16) - ax.set_zlabel('Test Scaled RMSE', fontsize=16) - ax.set_title('3D Histogram: Test Scaled RMSE vs Lag Parameters\nGroup: graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17', fontsize=14) - - # Set custom tick labels with font size 14 - ax.set_xticks(x_positions) - ax.set_xticklabels([str(rate) for rate in reversed(unique_lag_subsample_rates)], fontsize=14) - ax.set_yticks(y_positions) - # Reverse the tick labels while keeping positions from 2 to 8 - ax.set_yticklabels([str(time) for time in reversed(unique_total_lag_times)], fontsize=14) - ax.tick_params(axis='z', labelsize=14) - - # Add colorbar with the same normalization - mappable = plt.cm.ScalarMappable(cmap='viridis', norm=norm) - mappable.set_array(heights_flat) - cbar = plt.colorbar(mappable, ax=ax, shrink=0.5, aspect=20) - cbar.set_label('Test Scaled RMSE', fontsize=16) - cbar.ax.tick_params(labelsize=14) - - # Adjust view angle for better visualization - ax.view_init(elev=20, azim=45) - - plt.tight_layout() - plt.savefig(output_path, dpi=300, bbox_inches='tight') - logger.info(f"3D histogram saved to: {output_path}") - - # Also save data to CSV - csv_path = output_path.replace('.png', '.csv') - df.to_csv(csv_path, index=False) - logger.info(f"Data saved to: {csv_path}") - - # Save as structured numpy array - npy_path = "graph_type_comparison_validation_errors.npy" - data = np.array(list(zip(df['lag_subsample_rate'].values, - df['total_lag_time'].values, - df['validation_scaled_rmse'].values)), - dtype=[('lag_subsample_rate', 'i4'), - ('total_lag_time', 'i4'), - ('validation_scaled_rmse', 'f8')]) - np.save(npy_path, data) - logger.info(f"Structured array saved to: {npy_path}") - -def main(): - """Main function to execute the graph type comparison analysis.""" - logger.info("Starting graph type comparison validation analysis...") - - # Configuration - group_name = "graph_type_comparison_experiment_enhanced_sampling_data_onlyfan_aug17" - project = "sule-shashank/jamun" - - try: - # Step 1: Scrape WandB runs - runs = scrape_wandb_runs(group_name, project) - - # Step 2: Extract parameters from runs - run_params = extract_run_parameters(runs) - - if not run_params: - logger.error("No runs with valid parameters found in the specified group") - return - - # Step 3: Compute validation errors for each run - results = [] - - # Create a cache for validation loaders to avoid reloading same data - validation_cache = {} - - # Create progress bar with more detailed description - total_runs = len(run_params) - pbar = tqdm(total=total_runs, desc="Processing runs", unit="run") - - for i, run_info in enumerate(run_params): - try: - lag_subsample_rate = run_info['lag_subsample_rate'] - total_lag_time = run_info['total_lag_time'] - - # Update progress bar with current parameters - pbar.set_description(f"Processing lag_sub={lag_subsample_rate}, lag_time={total_lag_time}") - - # Create cache key - cache_key = (total_lag_time, lag_subsample_rate) - - # Load validation data (use cache if available) - if cache_key not in validation_cache: - logger.info(f"Loading validation data for lag_time={total_lag_time}, lag_subsample={lag_subsample_rate}") - val_loader = load_validation_data(total_lag_time, lag_subsample_rate) - validation_cache[cache_key] = val_loader - else: - val_loader = validation_cache[cache_key] - logger.info(f"Using cached validation data for lag_time={total_lag_time}, lag_subsample={lag_subsample_rate}") - - # Compute validation scaled RMSE - validation_scaled_rmse = load_model_and_compute_scaled_rmse(run_info, val_loader) - - result = { - 'run_name': run_info['run_name'], - 'run_path': run_info['run_path'], - 'lag_subsample_rate': lag_subsample_rate, - 'total_lag_time': total_lag_time, - 'validation_scaled_rmse': validation_scaled_rmse, - 'model_target': run_info['model_target'], - 'data_target': run_info['data_target'] - } - results.append(result) - - logger.info(f"Processed {run_info['run_name']}: Scaled RMSE={validation_scaled_rmse:.6f}") - - except Exception as e: - logger.error(f"Error processing run {run_info['run_name']}: {e}") - continue - finally: - # Update progress bar - pbar.update(1) - - # Close progress bar - pbar.close() - - if not results: - logger.error("No successful results obtained") - return - - # Step 4: Create 3D histogram - create_3d_histogram(results) - - # Print summary - logger.info("\n" + "="*50) - logger.info("SUMMARY") - logger.info("="*50) - logger.info(f"Total runs processed: {len(results)}") - - # Group by lag parameters and show statistics - df = pd.DataFrame(results) - param_groups = df.groupby(['lag_subsample_rate', 'total_lag_time'])['validation_scaled_rmse'] - - logger.info("\nValidation Scaled RMSE by lag parameters:") - for (lag_subsample, lag_time), group in param_groups: - mean_scaled_rmse = group.mean() - std_scaled_rmse = group.std() if len(group) > 1 else 0 - logger.info(f"lag_subsample={lag_subsample}, lag_time={lag_time}: Scaled RMSE={mean_scaled_rmse:.6f} Ā± {std_scaled_rmse:.6f} (n={len(group)})") - - # Find best performing combination - best_result = min(results, key=lambda x: x['validation_scaled_rmse']) - logger.info(f"\nBest performing combination:") - logger.info(f"Run: {best_result['run_name']}") - logger.info(f"Lag subsample rate: {best_result['lag_subsample_rate']}") - logger.info(f"Total lag time: {best_result['total_lag_time']}") - logger.info(f"Validation Scaled RMSE: {best_result['validation_scaled_rmse']:.6f}") - - except Exception as e: - logger.error(f"Error in main execution: {e}") - raise - -if __name__ == "__main__": - main() diff --git a/analyze_validation_errors.py b/analyze_validation_errors.py deleted file mode 100644 index 541a7f7..0000000 --- a/analyze_validation_errors.py +++ /dev/null @@ -1,452 +0,0 @@ -#!/usr/bin/env python3 -""" -Script to analyze validation errors for Denoiser models from WandB runs. - -This script: -1. Scrapes all runs from the WandB group "noise_check_experiment_multimeasurement_vs_correlation" -2. Filters runs by model target "jamun.model.Denoiser" -3. Loads the models and computes validation errors -4. Plots validation errors from 2 to 10 (assuming this refers to some parameter range) -""" - -import os -import sys -import logging -import numpy as np -import matplotlib.pyplot as plt -import torch -import wandb -from pathlib import Path -from typing import List, Dict, Any, Optional -import pandas as pd -from tqdm import tqdm - -# Add jamun to path -sys.path.insert(0, '/homefs/home/sules/jamun/src') - -import jamun -from jamun.model.denoiser_conditional import Denoiser -from jamun.utils.checkpoint import find_checkpoint, get_wandb_run_config -from jamun.data import parse_datasets_from_directory, parse_repeated_position_datasets_from_directory -from jamun.data._dloader import MDtrajDataModule -import torch_geometric - -# Setup logging -logging.basicConfig(level=logging.INFO) -logger = logging.getLogger(__name__) - -def get_data_path(): - """Get the data path from environment or common locations.""" - data_path = os.getenv("JAMUN_DATA_PATH") - if data_path is None: - # Try common locations - possible_paths = [ - "/data/bucket/kleinhej/", - "/data2/sules/", - "/homefs/home/sules/data/" - ] - for path in possible_paths: - if Path(path).exists(): - data_path = path - break - - if data_path is None: - raise ValueError("JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable.") - - logger.info(f"Using data path: {data_path}") - return data_path - -def scrape_wandb_runs(group_name: str, project: str = "sule-shashank/jamun") -> List[wandb.Api.run]: - """Scrape all runs from the specified WandB group.""" - logger.info(f"Scraping runs from group: {group_name}") - - api = wandb.Api() - runs = api.runs(project, filters={'group': group_name}) - runs_list = list(runs) - - logger.info(f"Found {len(runs_list)} runs in group '{group_name}'") - - return runs_list - -def filter_denoiser_runs(runs: List[wandb.Api.run]) -> List[Dict[str, Any]]: - """Filter runs by specific criteria for spatiotemporal multimeasurement analysis.""" - logger.info("Filtering runs by specific criteria:") - logger.info("- cfg.model._target_ = 'jamun.model.denoiser_conditional.Denoiser'") - logger.info("- cfg.data.datamodule.datasets.train.subsample = 1") - logger.info("- cfg.data.datamodule.datasets.train._target_ = 'jamun.data.parse_repeated_position_datasets_from_directory'") - - denoiser_runs = [] - - for run in tqdm(runs, desc="Filtering runs", unit="run"): - try: - config = run.config - if 'cfg' not in config: - logger.warning(f"Run {run.name} missing 'cfg' in config") - continue - - cfg = config['cfg'] - - # Check model target - model_target = cfg.get('model', {}).get('_target_') - if model_target != 'jamun.model.denoiser_conditional.Denoiser': - continue - - # Check subsample = 1 - try: - subsample = cfg['data']['datamodule']['datasets']['train']['subsample'] - if subsample != 1: - continue - except (KeyError, TypeError): - logger.warning(f"Could not extract subsample for run {run.name}") - continue - - # Check data target - try: - data_target = cfg['data']['datamodule']['datasets']['train']['_target_'] - if data_target != 'jamun.data.parse_repeated_position_datasets_from_directory': - continue - except (KeyError, TypeError): - logger.warning(f"Could not extract data target for run {run.name}") - continue - - # If we get here, all criteria are met - # Extract additional parameters that might be useful for plotting - run_info = { - 'run': run, - 'run_path': '/'.join(run.path), - 'run_name': run.name, - 'model_target': model_target, - 'data_target': data_target, - 'subsample': subsample, - 'cfg': cfg - } - - # Try to extract parameters that might be varied (for plotting from 2 to 10) - sigma = cfg.get('model', {}).get('sigma_distribution', {}).get('sigma') - if sigma is not None: - run_info['sigma'] = sigma - - # Extract total_lag_time specifically for data loading - total_lag_time = None - try: - # Try the specific path you mentioned - total_lag_time = cfg['data']['datamodule']['datasets']['train']['total_lag_time'] - run_info['total_lag_time'] = total_lag_time - except (KeyError, TypeError): - try: - # Fallback to the general datasets path - total_lag_time = cfg['data']['datamodule']['datasets']['total_lag_time'] - run_info['total_lag_time'] = total_lag_time - except (KeyError, TypeError): - logger.warning(f"Could not extract total_lag_time for run {run.name}") - logger.warning(f"Available config keys: {list(cfg.get('data', {}).get('datamodule', {}).get('datasets', {}).keys())}") - - # Look for other potential varying parameters - for param_path in [ - ['model', 'arch', 'num_layers'], - ['model', 'arch', 'hidden_dim'], - ['data', 'datamodule', 'batch_size'] - ]: - value = cfg - for key in param_path: - if isinstance(value, dict) and key in value: - value = value[key] - else: - value = None - break - if value is not None: - param_name = '_'.join(param_path) - run_info[param_name] = value - - denoiser_runs.append(run_info) - logger.info(f"Added run: {run.name} (sigma={sigma}, total_lag_time={total_lag_time})") - - except Exception as e: - logger.warning(f"Error processing run {run.name}: {e}") - continue - - logger.info(f"Found {len(denoiser_runs)} Denoiser runs") - return denoiser_runs - -def load_validation_data(total_lag_time: int, val_root: str = "/data2/sules/ALA_ALA_enhanced_full_grid/val/", num_frames: int = 100) -> torch_geometric.loader.DataLoader: - """Load validation data for error computation.""" - logger.info(f"Loading validation data from: {val_root} with total_lag_time={total_lag_time}") - - if not os.path.exists(val_root): - raise ValueError(f"Validation directory not found: {val_root}") - - try: - datasets = parse_repeated_position_datasets_from_directory( - root=val_root, - traj_pattern="^(.*).xtc", - pdb_pattern="^(.*).pdb", - filter_codes=None, # Don't filter by codes, use all available data - as_iterable=False, - subsample=1, - total_lag_time=total_lag_time, - lag_subsample_rate=1, - max_datasets=100 - ) - - if not datasets: - raise ValueError("No validation datasets found") - - # Create data module - data_module = MDtrajDataModule( - datasets={'val': datasets}, - batch_size=32, - num_workers=0, # Use 0 for debugging - persistent_workers=False - ) - data_module.setup('val') - - val_loader = data_module.val_dataloader() - logger.info(f"Loaded validation data with {len(datasets)} datasets") - - return val_loader - - except Exception as e: - logger.error(f"Error loading validation data: {e}") - raise - -def load_model_and_compute_rmsd(run_info: Dict[str, Any], val_loader: torch_geometric.loader.DataLoader) -> float: - """Load a model from checkpoint and compute validation MSE (RMSDĀ²).""" - run_path = run_info['run_path'] - run_name = run_info['run_name'] - - logger.info(f"Loading model for run: {run_name}") - - try: - # Find checkpoint - checkpoint_path = find_checkpoint( - wandb_train_run_path=run_path, - checkpoint_type="last" - ) - - # Load model - model = Denoiser.load_from_checkpoint(checkpoint_path, strict=False) - model.eval() - - # Move to device - device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') - model = model.to(device) - - logger.info(f"Model loaded successfully for run: {run_name}") - - # Compute validation MSE (which equals RMSDĀ²) - total_mse = 0.0 - total_batches = 0 - - with torch.no_grad(): - for batch in tqdm(val_loader, desc="Computing validation", leave=False, unit="batch"): - batch = batch.to(device) - - # Use the model's validation logic - sigma = model.sigma_distribution.sample().to(device) - loss, aux = model.noise_and_compute_loss( - batch, - sigma, - align_noisy_input=model.align_noisy_input_during_evaluation - ) - - # Extract MSE from aux dictionary (since rmsd = sqrt(mse), we want mse for squared error) - if 'mse' in aux: - mse_value = aux['mse'].mean().item() - total_mse += mse_value # This is actually MSE (RMSDĀ²) - elif 'rmsd' in aux: - rmsd_value = aux['rmsd'].mean().item() - total_mse += rmsd_value ** 2 # Square the RMSD to get MSE - else: - logger.warning(f"Neither MSE nor RMSD found in aux dictionary for {run_name}. Available keys: {list(aux.keys())}") - # Fallback to loss if neither MSE nor RMSD available - total_mse += loss.mean().item() - - total_batches += 1 - print(f"total_mse: {total_mse}") - print(f"total_batches: {total_batches}") - breakpoint() - avg_mse = total_mse / total_batches if total_batches > 0 else float('inf') - logger.info(f"Validation MSE (RMSDĀ²) for {run_name}: {avg_mse:.6f}") - - return avg_mse - - except Exception as e: - logger.error(f"Error computing validation error for run {run_name}: {e}") - return float('inf') - -def plot_validation_errors(results: List[Dict[str, Any]], output_path: str = "validation_errors_plot.png"): - """Plot validation RMSDĀ² values.""" - logger.info("Creating validation RMSDĀ² plot...") - - # Convert to DataFrame for easier plotting - df = pd.DataFrame(results) - - # Debug: Print available columns - logger.info(f"Available DataFrame columns: {list(df.columns)}") - - # Determine the parameter to plot on x-axis (2 to 10 range) - x_param = None - for param in ['sigma', 'model_arch_num_layers', 'model_arch_hidden_dim']: - if param in df.columns: - values = df[param].dropna() - if len(values) > 1 and values.min() >= 2 and values.max() <= 10: - x_param = param - break - - if x_param is None: - # If no parameter in 2-10 range, just use indices - logger.warning("No parameter found in range 2-10, using run indices") - df['index'] = range(len(df)) - x_param = 'index' - - # Sort by the x parameter - df = df.sort_values(x_param) - - # Find the correct validation column name for plotting - validation_col = None - for col in ['validation_mse', 'validation_rmsd_squared', 'validation_rmsd', 'validation_error']: - if col in df.columns: - validation_col = col - break - - if validation_col is None: - logger.error(f"No validation column found for plotting. Available columns: {list(df.columns)}") - return - - # Create plot - plt.figure(figsize=(12, 8)) - plt.scatter(df[x_param], df[validation_col], alpha=0.7, s=60) - - # Add labels for each point - for i, row in df.iterrows(): - plt.annotate(row['run_name'][:8], - (row[x_param], row[validation_col]), - xytext=(5, 5), textcoords='offset points', - fontsize=8, alpha=0.7) - - plt.xlabel(x_param.replace('_', ' ').title()) - plt.ylabel('Validation MSE (RMSDĀ²)') - plt.title('Validation MSE for Spatiotemporal Multimeasurement Models\nGroup: noise_check_experiment_multimeasurement_vs_correlation') - plt.grid(True, alpha=0.3) - - # Set x-axis limits to 2-10 if appropriate - if x_param != 'index' and df[x_param].min() >= 2 and df[x_param].max() <= 10: - plt.xlim(2, 10) - - plt.tight_layout() - plt.savefig(output_path, dpi=300, bbox_inches='tight') - logger.info(f"Plot saved to: {output_path}") - - # Also save data to CSV - csv_path = output_path.replace('.png', '.csv') - df.to_csv(csv_path, index=False) - logger.info(f"Data saved to: {csv_path}") - - # Save validation MSE paired with total_lag_time as npy file - # Find the correct validation column name - validation_col = None - for col in ['validation_mse', 'validation_rmsd_squared', 'validation_rmsd', 'validation_error']: - if col in df.columns: - validation_col = col - break - - if validation_col is None: - logger.error(f"No validation column found in DataFrame. Available columns: {list(df.columns)}") - return - - if 'total_lag_time' in df.columns: - # Create structured array with total_lag_time and validation values - data = np.array(list(zip(df['total_lag_time'].values, df[validation_col].values)), - dtype=[('total_lag_time', 'i4'), ('validation_mse', 'f8')]) - npy_path = "validation_errors_spatiotemporal_multimeasurement.npy" - np.save(npy_path, data) - logger.info(f"Validation data with total_lag_time saved to: {npy_path}") - logger.info(f"Data format: structured array with fields 'total_lag_time' and 'validation_mse'") - logger.info(f"Used validation column: {validation_col}") - else: - # Fallback to just validation values if total_lag_time not available - validation_values = df[validation_col].values - npy_path = "validation_errors_spatiotemporal_multimeasurement.npy" - np.save(npy_path, validation_values) - logger.info(f"Validation data saved to: {npy_path}") - logger.info(f"Used validation column: {validation_col}") - logger.warning("total_lag_time not available, saved only validation values") - -def main(): - """Main function to execute the spatiotemporal multimeasurement analysis.""" - logger.info("Starting spatiotemporal multimeasurement validation analysis...") - logger.info("Filtering criteria:") - logger.info("- model target = jamun.model.denoiser_conditional.Denoiser") - logger.info("- subsample = 1") - logger.info("- data target = jamun.data.parse_repeated_position_datasets_from_directory") - - # Configuration - group_name = "noise_check_experiment_multimeasurement_vs_correlation" - project = "sule-shashank/jamun" - - try: - # Step 1: Scrape WandB runs - runs = scrape_wandb_runs(group_name, project) - - # Step 2: Filter by model target - denoiser_runs = filter_denoiser_runs(runs) - - if not denoiser_runs: - logger.error("No Denoiser runs found in the specified group") - return - - # Step 3: Group runs by total_lag_time and load validation data - runs_by_lag_time = {} - for run_info in denoiser_runs: - lag_time = run_info.get('total_lag_time') - if lag_time is None: - logger.warning(f"Skipping run {run_info['run_name']} - no total_lag_time found") - continue - if lag_time not in runs_by_lag_time: - runs_by_lag_time[lag_time] = [] - runs_by_lag_time[lag_time].append(run_info) - - logger.info(f"Found runs with {len(runs_by_lag_time)} different total_lag_time values: {list(runs_by_lag_time.keys())}") - - # Step 4: Compute validation errors for each model - results = [] - total_runs = sum(len(runs_for_lag_time) for runs_for_lag_time in runs_by_lag_time.values()) - breakpoint() - with tqdm(total=total_runs, desc="Processing models", unit="model") as pbar: - for lag_time, runs_for_lag_time in runs_by_lag_time.items(): - logger.info(f"Loading validation data for total_lag_time={lag_time}") - val_loader = load_validation_data(total_lag_time=lag_time) - - for run_info in runs_for_lag_time: - pbar.set_description(f"Processing {run_info['run_name'][:20]}...") - - validation_mse = load_model_and_compute_rmsd(run_info, val_loader) - - result = { - 'run_name': run_info['run_name'], - 'run_path': run_info['run_path'], - 'validation_mse': validation_mse, - **{k: v for k, v in run_info.items() if k not in ['run', 'cfg']} - } - results.append(result) - pbar.update(1) - - # Step 5: Plot results - plot_validation_errors(results) - - # Print summary - logger.info("\n" + "="*50) - logger.info("SUMMARY") - logger.info("="*50) - logger.info(f"Total runs processed: {len(results)}") - logger.info("Top 5 best performing models (lowest MSE):") - sorted_results = sorted(results, key=lambda x: x['validation_mse']) - for i, result in enumerate(sorted_results[:5]): - logger.info(f"{i+1}. {result['run_name']}: {result['validation_mse']:.6f}") - - except Exception as e: - logger.error(f"Error in main execution: {e}") - raise - -if __name__ == "__main__": - main() diff --git a/bond_length_issues_conditional_traj.png b/bond_length_issues_conditional_traj.png deleted file mode 100644 index eaae7ce0811ab7c59133fb560876e2cf3f1710c9..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 77585 zcmeFZ_dnKc{5O7*A{vAg$|{K>EkzLt6_uGSGzrM^(WR{&Rl!jzx@4Z*H z`+2yo>;Bxoe82y|eS2J^!Fj&V_xm_rujhIl-sjH9ucc?Fr%)(sPb$c&P$)Es6v_&2 zx>fj-eJ#P%_@B7V2~8VSi_13lhE|s-N`^K!u3Ok#zh=a3cgf27nuYm3VbMLp`*v|( zwXwNjEiNKr_P@VF*uv_H$mQ)R_wg#LZzyP4Qz+{V$-h)lQjymvR20fdS!p$gkby>f zUDfkTv?FH~xYzvQSlQ6_%3nv>v?yE1Jj7q)eYLWV1WSr-RaJGhHtp3a?f#}JLGvPS zoqo!WFXx_Kw{mrxKE2g=!?p*EBjc{w4m-O;oQD5yeH7dgZ2dEkj)%tUzpn>RPAbzL zll}MW2{V7_+kSV-K<-w$G2dqr%#`bjEv;bayh&ARz$WhF3fpa4YbX+YU#AaZkrkS zST-_VCKYk!rTFyNHxW_e;x)B@{`6i?P`sG4U~FdQdv`5|(bq@YzkL6`zhO1Yp3WKH ztw%GXj+g7)xUpy3hSj(4++mnJRFvVry!IbFkBwTvz4suipjuwutrb+1-rim>KfmG( z>+U^HQ`f&g-$$z&d8Wv|*(LJn4U^hL9`nY`l{Mpdu&ivFR-Pn*PT1WE&iXsxZS4;?-n zrl0K?bmZoT7+*ZkB;yjbrSppdHGYoPyQAyVEF!Lyy}Ds#rK02CtEW1n5wGjrSM|QI zp@CChUq9%e8T;?wzborHlKpDx`?YquxlNDoY?XA5+L&fF7gC+)IMKs&cRiQi@3*H5 zzkgqIsq{%cZf4c2+R4M&dxeA;K3*x4%dqO?e)df{!|KvJp0aqER&Z?Wu3cImXi`jT z?s|Hzyl!Szo@r}3(D>{^CX7+>Ez0N1-hha~Q{QWC$Wwm#xDcdL1b#}GS#@*DE??Iv~^7Hd?M|duO4+^Wx z!6r4C`(!`7m0isI0agefH#bE@Kg<33^FI>C`pH+BNJYUL*$g#o#?C3sx;cJhX6({H zbFS1`Q;R=;{umZN*sPgpvy;+NaaKybD#fHqu*7T_@7eVum`_}sT`$u{da|#|vb}h1 ztk%a>PE!N<(W;T>l8jFH4$e;Y-~09JoK4R!DwSv!vL$Mhj15gp?g#M6=j^-8Lo>gy zkYrYWP%U0}Gj2iA`|}BjH^(oNYmu5CSyAlIwVu+`lr2&Ebhpp*eOC(V>vaOt9p@~< z{imlLLL}_hX=rFfsl2_f-MC(A-uw0kMv9YmWf%{;%VO1)|do1o`c*a}5O#g^gW_4wkH zbl&vi#eqB7*)FfjcJ2rZ3mX|9FT{2J!qY{$Af353(Qw7oaI^ouD?hJ3kn!>Nm;ZSA z#|2~KJ5(!HapMlTEzazipP#=m+IFHcO8I!z6t?(P6af_#m0usPoF%WqCZaDp-t|NI zlbu=00&3Kj(a}+IahbOLPurVCbN%<3k?)Sx+6RpATJMuxPk99G-o1NKrB=|;(i-Vw;8%`|Q&3P4H!m+Ws)hWqW0WIDkB+mfpv+?>ZrXQw&A{N` z@)DJhXrG@MpC0-&%(!90?S~H;s}5Y68S5n9($Lsg9wEQ}khnPS-n|>Tf4(gA+amER z+v)F-5@`Zfp6s~o<-sW9aqk`jrJ$@#q{I$)W+UbEYw0^`m(Y5s-LYrx#tf8jxv@|_ z;~Bk9@(&8yl+x_t;^G+}8L?NG(iUuyQ(jk$C+pzk^gi%Mkmzv=V^>#K8`e(D*!S;> zqpkP2=qW~xnYM>t71oQ=(9mR!rw`Y?Qwlyv?XH<*A z-?8TyC~XP3iz-6rC5NcIMn-O^-#mBjA%*+)?c2%4y63P^T#`*|gsF~U?JGyP7#W54 zjXF&BsZ$-B=qW!Qp{K3=LgCaYm4&U0jIS;IPL)d1tzNAh*qCm493M+L&0LR(jV-d; zb*Lc&cQ^e;SWGOsPywEUth~JGfYiZ*8uqln5KK(08_w`-`~CRdJzC0OMP71JV&X4> zZ{LiD&CSh8t($Bz|MglrN`-{Dytjh#GgK-;P~zs;wyj4T+<(55h`2f46=Q{_ zf=fRp7}z!E+?;x)Y%5m%1?)EVh_Dx87Jk4YhRr!HCaT?(JoK*d?y`H{UzhY>P*YQL zKmJ}KZhWyGm~jc|l&w$6tEJSQY};DBdzbaZhJ`r$YyEqMyc`j+%O0k1q5 zGfZkStgkqa7O`!Tu-mn6i?|2c3i%9Cx*1ll{B%Cq4eBawcGf|c*k8)N|0;!nfx*Mm zvvcRutW4*by_CF~n)Ak|LS%WRzdhY;*9ik;3^*#6V*^)f1O+v%X91 z#lE(SFHLIVni$GbN_@l;1X!m3 zHoKL-mR_NL6L)6+qy2lIuZfS3*Di~`R`HH~E;Tt>cCapmX~Tv`k+tJ!uQpV$AnHu+k` zp3C1iRayFR9tO)ecD+7T$@U}djvYH5aEPrlaG6wp#jI7D`xS8f+50Jg5YLDRKG%iG z^Sku3HeqeuR^&epER$r@tMWwXyzp7e-k8Y> zAd-i^zQ<+O=7`%O=i=hxA=|zu`XxWU8TC^bw`_^~IR9gPe`)y#bP z^E1|Qlci%~kvCXbI5>iN4-W-DiO3b!%Y3u3CRj9Ux45{tf#uGn;>@kInSQ}sr4IJP zO>22PesUQYG&>JU@$<`>x)_G2x(4pn*=Sl5e?n1ljj%z^gPHN}<8-$GUt^RSt*xvI zK3w`XGSRErkad$8JK_#LCIHnfQi1Av=bY82w2X|5{M?wgqe3w8$+Z-CGI)7==VQ~A zqo)tlr7*k{w=K2ayG(Kf62Pr~7v?T4`d|qi2O^aB-zQ71U`u5|k-po7pmkR5qyH{4KrF=@-IV=}xmh5l!qQ3r(O0>SHkiY%>w54x! zkPMxA>`X>|sIYB+^$8ObLDa5DW&45JBpC^EXD+bNdur?EEp3Rf01*l)Za!&HdLSzK z=sH3NY_pT*O@N6gEx3l2ITkH$-8Qom*L*XdJYlI~DHLPvvR32*bhGWN%&&}ARrh`m z>~SsmvBEZ#>=+aJvHHOYmmI+*{?cfV*&8-CvVVsfN4rDa!qJ`6ZTmNZ(?tGtU1mRH zcfhc8#Ze4_g&w|o<%dliX4CuoaH^EsVpg@XNNG8bg>Yr8uG-Ak$2(ST8a2i3k}z*b z51Ty-Mq=PPv)#Rh;i8aww8~19g96kUHSdj=vYK7z;;LOF4jtk?cyQyjYu7UF7k&H2 z5T}z?j9Z&{`3HMSCct|Oes{Ivo$t?R*O{;*HF$5Hefw0^7x9G2%0*>av1*OLNW0+M z$Vg+LA4+9SL?3|U#l4$1is(N;ZHg}4w{IWHs{w@q&hV(a{h&n?CmK!RtD{ag#yU=_ zXE0LZA}@t)vjJ*b1%g6e7`3J*s_grBHJ=j(O-)Utqmjk7TPHQD%4KQcuZh3G+X835 z>~BCGqa&@yeaF%gR<^dbrlWF6)Ki^{+cGspy`+4b4?zv<2T?mHHNcSE?y)c6v#E*bwM0_HfM3xv}*Rm z;Xmk$t}~r%q_D9HX?Xbht~GDU@^+b<%K!RxHPDV~idn_GwW-(t-1e88r6c?)Bjd11 zb*wikYv-JEU86sbBmb_arc4)NI9goivNL2a${CeQG z<0>i}fNj?r7#I*3W!!95mz?j(z($~S^3`%`{KVC!tc<$Gp{7*l8LR!czkKq3pRtCN zI^Qk6Su%Mr@&PSy#^;RaogeY zJSeCot=W|nwenOmo$0x`!otE8@_wAs7cXu@Q}G1i?Mo>GY<_t%#cr@J%x!70_QNFx zaho0rKB)6ZJ}nyLDoP80^dZYO8no`JjcaRaKvq0JOTuyYdaB;D393c8pMEAxbP1>l z+x@tt;#0r4Jmy!rU}$*j?dfN{qN2(vu21)#)C^jj>E=>=b!0Q4CI0^F@qz@wl8S@I zOl~-gCBn1Gf)E8+6B?$~JRWrG*RxS7T|vw}K7IOhwJ|eg zx+~b?GZt#HImd<#ycDm^`z}Y7{(N;*2;#wZTm;X+zyM&1XOvp>qfJt-Sq>9D0d-T) zpFcMOkSh50?KJiS)iIEQsG_&?dK?xO7OjJWaT*7qNl_j1@bK`$%J#xzTAiUQT>4}u zU(R$J1EJwbp@3MT=sO_bJHLPbE{mO91cJg1^!NI}-8XNxiCT3ivCKWeGMTbtLJE-HebaY+QVp4K43s&TYTeoiU9zLwqpDJNL#FCSf zQvhHHi9`A0Wu>I6<-7sUpGQ{@>G{*@iE>G~>Ly?RqXKDyj+T~ov|OIc``&u4uG-v; zqq$;k{E9)ZO(32yiy9WtC_}>FdG^iI*VmU+>`kIZ9*3N#Z^XB4l$mSSq2oa@+J$;7 z8?Evd>PQO6@11nJ!52LZ86PR!`Z>N%xyF9x7}@BM|627q#iIJ~MD#EhQxd(s7(O;I;aDjo}7!fOjB%9;hl8pzEO)v;poo za!HAbGNDA>28i#lIt+O3z0c+m=`|8uf=qU!y5Xm}?yH8)Alj~t)quUUUSUN`^I_R~Z{dA#FnL22o! zDpwSo>6z}(^2oDXxSf0g0(20|@&PmrjmC0^jlp;yG`lTrfjTFL{g=G;6A)Y!1ZyCB zKeSUB-hd9l%e>)juTgn79lF6yIYxI2@(Hv#E>w$Y$PPqQPB*YDAoDR^j4eVWLl zM~{f=`GE8A2A~>4LqljJD{(_TJHI_Mo*rpEvBku1Y^*5>cZDFdkdTly%-fhY(@M)e zkw<^sK=}+_ZgpL3xq{NHT-n+00N{dcCSfqgZCjIIzy`4jq-QfEAr%TEDwTNs5-vrC z|FJiOEgvXzDGPWw_vu*$LD9eKZgHBO=!G64ZW~iQ2Ho_1DV_~Vy5}ZI=kyrR09$JJ zL)KjnVs>rcenKtkoHTlD?)dcSr@NRapP{N+v5GJSBr%ZI;Sm}dDn}#?bgXy^H#mw4 zP5Cu8Mn-N5q>vH&)E~X`O=8c4^)`^2tEuS)&~mjhRMJHoLhNd9Sza`ERT`$Qv(nl2 z+rl4ts?vE2lwm9I1StGMA3rkOKG9pzQyI<7$;o-a+(?(_yXFl9LjHu^EU>WBf9uivkj2AY7iJTalhs`2h(?SDU074Ywsq^)J^S|E zfl#t){niH{hTZA8U;z=(RZvphKYfyb&gl=3A14}m$ddY~%iHeGDI(U5(DmeiqfPU>@q}u*&P=1MGTU}k9n38e=K*@4p zc9QyVnuiCCVyNVnsu=a-xSk89rVO~BXqqg<@<=(252x zRgbRAo~c_r*y$A7(sIE!6PHh#0jli1=g&8jLSx===yOrgwdA#{S4-bpx22`El`2Db z9Ryo~W^o5fs}AE`w!%xA9xdT2kJAzJo#j`0aRHSK53(WM5`NGs%J&xs7&ds*fM|TG zUNEmsq{44teP{^bg-X1b_*yzUSK?*u?d^efXx4K{R@UAd8yy8{AisywwMSI6e5Upb zgP!Syoi(SG6cq9xb*{npYodxlj!3uZ-QZqRA-sZ`x}~j+8lM^XxD`r}wtqy~O=jg* zI6}Oo%sezig~#22kJfpO3k~iBcC+E-%c}sdAXB!W?&X7_S&p@74-h zBYYfEwPybguLnEf6t|%(ytlptt1Ib7tBmhp&eEb2zW+7u==4b8dY;GGIh;+IwhR@K zXXT-X5^Wb3+KK{On{LTBHPq+>44RLQYV1Z71|n($7lxq;9Jhmft!xPv6kJe z#Fv8>5)wqk6@YmTvlBbdygK3yUh^7I@TR0#r`jQ91)s-4&CaQ&HD@AzmX)o~U7Qs{ z?|7H8s02!0gd5$`-5q&QiID;$r2NCB(*q3|%s{5cLGM~RI%wd60s2MP>E$?YrjSR` zGD&;ygwIC9qWi2jY;9e-GaMWo@=&9pG!I`tEvwdW>EgvG$%)*f+l53DN+6-fh@4-w zW|MnQ&((RsMd34Mn4sye~IU;56fYu7)z zE;vGVu3}N6z!C5a4dp_?@q|qB+NcN~u~!~k+C;Y{SM<77XbtD<&+6)OzSw__8jund zn2(Qd1yt`@n`^IdS6d3#OUVMkz^@sF{&Kx3YcuibP|eF>GeEy0oG%?N6Y*CeKww>R zn5deLTQ$k5eYtPa!^XsFA!gpdNcrqewW8^dM*W_V&!1%tuUy#y(3o!3c|S%s?3@=I z83u|GYP88*`9aSS&@N`mXW~n~iw7=oOIGNR*BrbNd>@ygP# zv)+9E(PXyIZnKputucQ!YojtAzPYbPRN2qgn3hC+8?zNxn6JoBzYb(H~m>6z0 za31@*g8+7DB=Q$8UGj#tNzf=tXuti|IVIF|d~^_}i8?>8FPnBo)~ch=;pdN&e^Dg$ zAj8R#-Ule_1B{_>7Mzu}Rmi~f|J47{nn5Ki>*i)>VTho{nsfp?sSXMysM0~{6Qp>- zlDL|D_Vj7ylP6Egc-15fZ}65S)gG4=o${@;bc2XNj@m-Jv)TI9tIKPzz&I#&dQn@e z3B{d{up=S~RlDNhu_xF=iejNu&d&XC$vzE!AK!e(Wv25P`uG9Ur!jE)F3Si8-v9DN zAtkjeREllyb4BoVUG=8{0X%T)gkt03fR2K+=isH&MXL#dpT}y%&j*7YszoGN_v@+B z9hZ@zp^)%EiWO*El~aINvji&FSqh__ot;tJXK$IXb5VVgi_^wbM26gL(5)ov=}%hr z*;hwyQd2V6d&7UAo-lH8`D3dP(73GEqy9+qKfbo*)lw=da6V*t=y$5V;nWnANaw$e z_7r!i#p&b5pOzN?=4$6&1MTs=bLWoSO;kkXY)4xn18cVj?J=YV@eQACEGVGHy}rXp zjW($kxCCzly`vb5(+3+lW)LVjudtBMpc}^S?uJa;)20td>O$JAuO;|Nb1eV1Z5<|3 zl!>|;EL>67c?SYPISyW54p%gOvOg{uXappWJwUGk;U7DHVyKY=0Su6$Vk~BxGo5OZ zu7Unnc*7sHV1xW;bKn8A0;=c3O)8Jqld$QlJQ-)EZ{Qyn8@tDC(K#jfD$3I_{>QCQ zIAT7aFtx&ZC8Y`l@pY=dNbdaJvWWEN%Z7$w%KXp1sf&$+GBC=ZwV3X@AQE_LoV5gt z7k;-Ff*3J!D}}Hj*2CwmaIDcW6O|Ab&lz`sN>6#suJx4pQZN>R`X9tM0$#q18{OIl zyW*IJ#?!i<_gF^6f`n3BwP|-v)=g`h{_3@;rxoc=iRk4UcWIuxD8?|oI5%9?Br!SM z%td(^80ZPf{C#8ATZ3Fzio4MH_^8q`M^uvZL!J-4#$+>I-LX29mGh(`H@Wdcfzb2k^u55yrsMtEi3KtMxqilgI`pRyMQ-v{LuJ zzvxNCA^}B~$0jD%0`R=z%Ur)D@j8Yk1RsBG>FbM`%8XQ`7lQ|kwn4oi2Dx_f)mx5}ebL*`0-O@(m-EmK3ZB++koWTN6Ab%JDiyj-oy=bzb<&tj zNKJWybOBJ@FmrtZ(5YrFoU zt*vXBY@Zr@%kab&@24KvIho}hb%W&OZidfSpR4)u)Lz-u_^m0;v*2J3V*8=CzFZv( z6A8k2b{K72>0V>1+tq>Qg~ZS8+*~O|B_(39#A+r{=gyD$ zn6!=q*qFvDPaPB!6Qi8dd9BaZNoU-7Ft1;POTxC#W3WCAx4wR4bTl6ug(MLsz=Q_B zo$SoW&VCh{YVD-zx)eVXP~o)KtpCQ&{wD_OsBcPH@CxdN$SBC(Y#SI@)8ZyFGOLnv z%VBXoquTM(%`*(Hg8h9p3G0cqV&-F$JEPPQ43WXm+}zKHO@x@#1y~~f@ONTL$G`pj zxlV>IC2W4Bi@7J*z$pMP4J8AI3Y0)~GpmoGVrgIdbBNZAf^h*WXU~BH>b?q2*0Q=- za76r?6n<1%k}x)^D695pr#zR?aPcaZ@BNl%o_zJ_>FMG88_IR&%o)wp>ziLmy2NN7 zYAi2jf{UMbdnJR~AcTvNM*HSg$PW%9EmRr_`fCBU=K+aL_yV3(@HTVaihDEY!*;1z z_mkc3l9~ZA`IVCcf0&`(m*BGynDk(iC^e|f9Wc;3x`Spf4w})!+Zz>o>{&+~=?+sE*m5FGypvJO8R4+O_o&F(`j2gErY^g2MllF_=N&N22w!#Q|7s__>?j z0%TKO+t%d)ms>6^x{$;`caKoZWZFAG?;>~^O*5B8(uSII{lU#mww;0H@a_OrxJc){ zdjJ((wefp?#&HWYM_uz-0(j6cef${ic|LUN$TKg0PfzXLy?ZS_$~{m}L54A)=q zWe)T`di=O*NI2L4REb<1x{77bFCI$5@VY1^7@&Ku*X|)L4pmIWG?MeyMAwtO+dH3p z6WeKU2kSY){(EV(Z|=er2MJ!G2zuPT+m)__@BWp1P4$S&EDN|w)t@1tK9Xa>3;qHP zP057iKzJ+~r@Ks89?)fLON%?2|Hj%3)+@vPH<>p}5MApmoro0SdxhHNm191GhNQEu(`0S~m};!6-1*8-M?ruv(}s%`Sgm)}{Rd(#^v= zp9_;&2|rcNESR)kY(MlR<-n8MHjuRh35*~q?K!m9HsJ4Npx@l$3Uxn5^GS?G9F~{Z zyHn7H?xM&R$HfD@SR$AOIOh%w6!lpZ{&;U>D21|HJNaB7TV;QeFo+n5BAS>Li2ZoJ zZ%4DiZzOZtkr`p{EUmHZd_)TdA7d-fOfrgZA|v1Lj*8cHqmtLwIwCOnq zt-AevLXJxR@wdbUd~piC@T0MBfnq!5qp#^SESQmK78>RraB$ca@)3*Zlq7Tok&#eD z-=fMuQ$0wa7Z&GdAcxCCf7xHX0_ch(D=RC}Vbq8d0(Sc$d@Sw9kBb`T(AtarOfN7g zzp_R03Ch+X02n`?a8qMss35n70|pSX1M<;2Fz^5jw;foD2$G0ze6%0Bf>hNZr>R|t zm-wJVhJl$7yBsy!5)mLQFhY~r?dK~lf%k6`Gw1v_jll<75!%6RMrwcsjZepw^ht0N zOsr^nw4ES9D8XBxeIZIxfIeFut-52EUgrL??stkU9S5Tw@ZwdxqotFt3AC6^Z0qRo zfZG&P2Ptqf(dU2^m9Z8Ho&!8DAmtq^FIG20WiVcfvd3ZM0@*0QiY7iZXJ3l5LVYVh zxgznq$^-*0q9~#lDMy~!qWDr={YnEibAd(f!d+Of(Ytw(&$tBCw;EBe0^E`SU=E;X z;{uP>?EMB?q+B_neE=Fb#n0aNib@AJdsSCZfdzoPPd+M23xbCz(c0?g&n1)`o*^Gt)`6Xo(z zdY=x~rCF>;rGE4FZ7WIA*bQt#;f>MObQ|mV3M+^S5ll~@a44lMked(L4{_q}??b4~ z9?#;~f7PAn5m-Is^+9cK6%eDQ`V`b4*s@tqMe` zbk6#J5^T`t*lkgB+3+HS&yWNaA)?OE{)x>K5to*h2FI-0@d1<;)nv1J4dh=L5Jywa zw7FIuqb@{DA0}qz2&U5t3SQL<0xa~SfIRn+SYd!L(%RL?^IM(57e!NOBnS0xi>2P%Mv&#XJF{0=zr_9${@MT zp#9e#75CqL1buNinpeHh;~1Ai{HNNa2xM!Y5NC+oL97wx|LnE0$r*e_mu{7E-9_xY za;V)TkB5@v@$A{Np2kmUXdUV)PBUX~>DH`a!5Uy08X6)`0D6%Z>aY(4Kb>|>Yz8A_ z)C%(RrHNLI20{KpqNHXDW!(qsjRCwAA{;-Gk|!Vok+M*w&nIS9w+*5WA1qt$UAuRC zLTdmp_tQ+gaI1Rp=U4Rbq9;35Ww8qoT@KtMCdLL2k_FvkBXIGsL>f4F7KOX?nMkPd zryXZsc>z>=pfd}DzkiimQoaG<^ldx>DprqthGzD@?`p)PsyIvQw(pPO{ z>imWP&%KV04*MqY7hi%%svA+?prcN9%J*cJzb_*XV<(-b(SQ@}2DSZi!i|d1e^JDb z_x^rMjHKlR)JkLkmw$!-L=-b=%5m@6!wAt~u;j-NC3O*xV;3%PA31V_=*6Jt*BZjb zjvn2DYJBJMV`hM|Un!>Nk6`n}{rL7wMD*$}ZYaS82#McWwf1+q?I`XU$p%2|A&CS+ zmE`2)!d^-&8)r;R<)8N4y0k6ke1ad?XH^-BxSB1)#$9>vL zJb7a>j&5lL=823GGZE7|IxZ5~8lOAnAa0yI%$#+chul}tuH)XdOL5RH1Ajxj03w^g z(lG)}tIf21-#1mbe1q_x6Sq|9e91?&8>qzw+T$zELxWE{f;Vb|h(J5ym>dWxzpE_N zAEF})J=M>jKc^S|HY3cwp5lNsH#Ss~MYALkg^`cPPzF1^Zy)R)Dhl}psg_u;*ediy zizUJu8WEZsg>t7GeD?lBY~L2xW}q70!f8u7wLo#6eVrm=0FTe)EP+ z_~jv&+5IFC0KD)TuT1`i2&H(ekVj;(K2nNbA>1=DVN*2>DoCQ{h)!+Wa8lZi8HVaB zIvzsPp|1x8-Jm$S?qs1q;xx4e-hp%64y>WPn%PJGt9SAJf_g@>?`Z$UXyl`VY0csA zG7y7bg(L@sf*hbpJ<@J%*d`SBs3-vx830862w$r?{0^G=RnGdNq7!jH>PwQ+(g;zF z@jVE;u?RBy*FvZHX;G+Ns+EQ;PNe}&xPirY6DGH_F>A`bK7Maq%u)^DfYOWo z(w_9JWD}4(K-r(u$Rd+9&QX?Df7LYf*+Gk7%v#_%y_r3V=ZnpT=)GZ1#iwPtq11Kb^G7M?E-n6`ml5ccn06A=M#S=}r=B0@dlZ9_xD z!+?NWAkHzq1^4~@ z&v)TRSmi=TQ*t`mK9~1jzIdk>{Wv5Dg620Fw(vSjt4EQHxPq4YGaf?c43dHKD+=DgxOcAt1!u z(F>?l-ziF?+LN>o5~{%j_pKS*4Ns;s5J7ZpUS6h*cx z3y%Zp(7ysYB|S3f1s@V_YKG5;*Vk7=6fy(0Yg) zj)k#`I5W`FkU-gg?(A{|6IXNX5gm`LGOhIX(?&+Th?)jGe0Yw#-w46(!lEM2@h{fV z+x}|V*&PI0V{p#J<^_5~rf-*}i-zVu>k+!mK0K|t-(6-aXX<1#hRYgs(`Fiek>#;H zrL4(z@SFdm`Bxtj6U!#fKXfT3o0%`s>r2mS9!Aui>FH^i`dow<5Cmg$`M&+1{ZNrs zhUJE}QM`D7diH-_FQd$1kT3l2O3X!GoOr@@oA{2`u4%Pc{PJLg_lcxwdvNZ*TPwF= z>Nz9z&Ye5)>{n4zK252V|LV^C(S3J$!+Ei?pq+Yf$jWB`wxO)u%G=%*t8UhWeDi^;nkTA5W^KR zZARwf^H_Qab6LVH(JhKJZH6R%3v#NnvvacfjsS%VaH}wQlQM=njAc#mF@AJe&Vm&J zZ|T}VBqGoxs?+xuof)Hm7f>Zk2FY9&0CiVih?mz`#5RMp?Hjy{Egc&95kv6sNoQ70 zVqs>!v$!B7r>pCDYv3&1=ue($KR>JMfWByMH^qygQ0%cA(42>iPSQp(&e%(2jXxDJ z@PkOzeXTKw&|#Eel2uD|oHRjQJ# zPgioYo!#m2nU6H7si!YoNt3<8k>R|zt?hRJMr0x#*mh~UpOu`U;Se)dn`2~QV!8zs z*N*VFI|6nPr5;5t(5_q=hW0UyR6Yr~wn2-#SqHZrK!4NztBj~6{`ezPGcnLQVvixG zAM*j}FK&FDEm4P1#)VRrh{THX;5{#a+*5V75*Woo_h_>X_VT*F+Yh3U(T{=xW(^J2 zu?#jUPno>Wved`^3dBwvSvNJ?<|9IeE{$PUUae1!({98z{{==0#6+w=&1}a5h!@1x zECzDvLqxcOz4jV!R-1KmAI!-Z`&hsa=nvA6AeE8khwMW~csyEARb3U>CIr+XRB8gk zj=0W~+-gag)Y1b~seTvT?Hn8&&TxOASZu^-!-Y$i0=r$90xW~}(SG?^ttUeB`dm=n zXBvU;DN%1#TE2-`*QCDDPx}@qbPr6ETpsoZlX6la6VgqdSCknM|l*& z&T$_)vKZz?S9Z0n<#WZsEMt~0f>@LJt1oZEi%>6Uk z2C76$+2c5ImAHnJ5vQGWG|)D8#Ysw(1kY&y1^M zwTS7##KED%J-3awH13ge<5pyRmKX1JqKd0|oq8thi##R|X0IUNo}b&3j5u?a!zeZE zA!HEoPi>k77m+;%bl0&GotOYo#5PZ)_`~3UzHxez>N4JmOcpKle6Vbn1CtH|7L9Rn zfzX81%^F00qsyBhmO>XUMk5S(`t;rGvA;wrryP@!xhYN) z$FwLqDTxaTomEmQLQ6e|o*XxKKvXmzPbCQnVj#MX7!h3SDa#D=1g(#9!~IiRKrk%XypbKoKhVjRmYyCj5_sn4?`z&p7y=%<99q8mC9u#6&4+2w zBFT1h{5ynZZ4y;=)5)nFTeAdW3xXzT_Qh=YxR8;Ee-Fi7`8|LQSI%so5OQYc+c);& zS;}MR0I=O=$O(5MXy27}x_Wx$5I_k6AVWlo{HnoIW2t0>1y744xbO|lH`mF^uJo=k z(@*pE`jSoxT)-`X$hV(MN{sPVDDd}+?{QN|oEUB44sOl+QX;ZIii}y%&2nIZq9l!y zxqr$K1_MYyXSzV5Oc;fN3@zDQz&aJvj5HKbhh^0pV;mffN zzIASNV)kjyf3;4!#zoc;<33o0*8iKEO#XQ7&L$Ixk!v}VAHUP)1l(`IE8*3E+e`#f*2%JLhB zf6=(m`N1L}T45#S$eAo3t!6*j$ZEdwOUKHVLRqEXc;omB!PJnKFB!IlsXdBhy;_m& zWS>%Bz+67piQ0IzCjO=NbL3NC%lLwW`ku&*?#Xs(?4svJ8#i0uEu_g?-dJ)12LR=9tNu^S;>1iJ~$MnKy*81Uy!uBd^|;aPQr z2{vv#uOh(0C==ysb^MQ?_wxo&tb{jH&&M*}4bB8{m3CW64}E-N9GYFsZyM7 ze)wk8e6IU3hws}PjyTcBAN|ymZ87EawFG+)0)$i7VEMb!^%I$LVc_O$;KY>z{@&i@ z>Fv6ZjacYcqO8Hrp8Tl^$O?EgI{n7a z+o>23msci5!+UhxErEl2>GbqTZs`-RJRo?JO;T4jJ#qf;bA#*wTEErUxMj=Q z=ZdS4wvDS-3_iFPxm=ZnS3yB^Ku~zfrr8XJSg{yqx%io zb+j7o~0ty8Z2YOo}_9rE4>(K+iUg~o|%I#1EmbWm58qn=8051gX zy>v?1o9u(od68c|mtvysn+XSi zq_&}6$B;uoNE|7`444Gg3`rPLa078T$SsuH;zlh@P~5>cW&cF~enU`bRLh$yms)B9 z8Myb_p4}4LTgRd#zMtHh5OQl+PlK*Zr@OhmDJf$8ve=~Cpw5;N!S~dnf$qwc?ju|~ z0K!n0iU3SU5YxX}siUREgUKTxI%%v%wNBIO*jvw@oqy%ekd>1%LgFwW;sg!%xD@ms zWKwR5T{7L$#y8v>oA$WY@bl+rh2xo25#eo6Ja`44LIDZb%69K7alJh;5F)v*LH9XP;BY!NiR8<>$Jvl4N>7)(T!O9Fzbq78%YN(d2x z47{)#{Euwmpg{h;2YzCV!MPijmS4v)wbsN&5KiZhA6`0#Tqi_c4#z2Z(71>zx%>L; z{v95%mYsvu08dvrsq32IEqV=&nL2s%4Vo19iFp@^WDpyY3&%KV586L!TimEL9fcga z0e07hxUD-R2rRAT09|HgSN}9rjY8y|_n21i+NG1a%U}*8rYVQ5-3P8WPRiuThEWt1 z-R&QE{9ub_nI1cxeXtdeOu_A%)-}x}6P|5{v+T_~YBwl3rrdTPI>71OR7}Pj)^MZW zB2QYhe6$Kc+-JNyImM!!fD4ki1-+r>q4@`#COVhk6b~}8H9u8v0cz@+^H*N>iO`=v z#p4mQ1a^$^U$ti3@Tw>0@?jy)Uv8zc3X1fjrx^qTNG;gw4ipydJTpenfqM;?9Yk(_ zgh{TMNdfp06BezNBraj3Gi+N2eUZ#~Va{IZ)`0)NYf!=M+0fnAf$~A|{!0Zk+4a7j zDfeEs-VWb@@j+hR6`CI}uO*H7x#GLo42Xf`zNRc7RCr@*K#N2ya3aFPuluh!*Voo^ zBZU=l1QIyqg6aDOyU)mI6e39L4XwF*!2`SX5cxzCkDt7Hpl-VH_@l77_&dJWb;1po z4z7A||I~(f0>Zx_xOo>k3Ry0Q!JP{%DJh8nSt7%9m`@5z^}TavMW)^0eQdpUc*9kh z1~9B3YSSPcg=u8fz)!%r1c)5D7 zw?7?x;LoqrQYl@!Z<(6>f!$M{Xm~u1Vd)@>EI>gqh79)i;c<Zwg0moUGpsh6c>uJ?-$NCKa+m+Jo=-O_1wWIvAi00W#yCGm}EwB0@Yrl z3>cc4`XF0gTU*;>!f?>Px+NsxmFfOd zee%a?Z~UF#`EqjOpUXW}#<#0GSOapf-NKE&{rPjaCL+vA^?H%b&b~)b@GIG}f){{$Mfw=MycE`xx9i<; zI>56t-{;|I+uNT91NVK481Lp|uy;A`n2^QAZ2hm~8g*>oaP}7e1pVJy7LTvp+Pc7B zP+C+<9@9U-h)AEErG;!vwI3(#y{-@bz4MrIwS9&u#xDL^ZC=#p|2g-pX(yk=3ZZJ@ z6V<01#Z+Qmsc&&LShMGX+CM@X+7-TOv*I>|MPIt>F-JpEFA@DGVH{E96cLa}d;BdmpP0?$X*RW_Y>honhJB zLIcJ>iu`w;J!2#BR**iGgayF)6&Qj`x9(m?j>4!(Ey~AAKxZc-SINoyQ!*hxMbC~5 zP9b@NdO~I`5Zw^P0THrta(ze6T)6PsL_$vrAJnmA816nyF>>6@uvv=e-SryKeF~A* z@EMqvJ>gk1*wAXL^HNkd&6~dBN0jNI)GP0QM&H?Mcw2*&HayEht8kUv@D?p{xZM*r zwACJ|WU01CZS)u2sHrz7SY2p?uqo4*7i35a`99Z#XQj$cL zNOT9&LpAkTyOAl!oYl>yCPq}HB$qimbjY*nsp!cLkPq;2*$0OkGFE_?Y~$x|gM!4qj&XG?ZEYMW>YM8Oy!TvW?@5DFNR)QQ(h5 zLHLCPE;94&E%|p$%y1G50b^jVg(W4lB%T*@2Kp;Gq6G5UKa+S=6iDfw(2`&+V|3CB z0kCKnhNBn1Fp|s(9yItKEe<&3cggAPboFKCc zpzqr1ZkWd;DO7|vZo_sY6Vj-W_EYM>un>VFW{(0FfgsBqJLX6`EQkeXQGZ)jG3!5E7e5WJI2>HYdN7ccP%a@ILHJh4CzD^3SNB$r7}*34Qm$oSH)}}zzt-CRrn@yrT}JG_O}s@55<~E2Y*>6CSPg`r00?+@b7Z#RVtzL z@8<_)DfD6P-k)g>uPrzw@0hc09D~_N;vO3eu7Z(iL{8E|K#|0cfI*C56Onhu)8Xag zBX(0Vb`o))kdJ)$=+SM+u@M+|IpjKj2)r$4VQTpsz%|re=A7X&;^DSHqhyrnaQuY% z^Yl-j_%M8en|22a3!gY*w%7@9h^Tzz3SXJf1BO$1QoWFxi2Yl9^1u@4@W$rbrpF%5+ zz~dqI8$u$YSj;a^vgo;xlrC}=9F#p~b^DQi+JMPEX<#vMDsPnB6%I!i-a#sxhN^xJ zp{uIqPJrV)jD{s*mY1v@@CIo}Z{&nAxOXRQ=OIjU;;a#wbLSW_CKs=par^aa?vyu= z$Lvi_`|YQ`B89aFN&9)|t5@ricM@9;V+c_{QxF9r6$Pi?-NJ--$RSIaD*FskWZs|@ z690q5#()M0Bu5Z*Epi*=7`TVXufUWbhtpwvhm1;FWxEMFTA%qYoyx0oCEFd5L`YI! zCFBdb?p4a(7twK!t~prj*f@lv#9*w`Js+6VnQ72ZbDJNPuZ-8@0Bap?v~|KH55iHA z)h-?H*59wN4UT%P>&SB7n)WY`kg8MmbD6uBSyx4j-Mh4woh>_exmn9%5@mUHz{Aqs zadEV*C}tLS-53$?o83{Zp*T#Y0WwcmW_3OBWW)PIr9Hhaz{t5b*#K%BU=uEnjw5Vf%C zUHUHFbVf1{JUcbWu&h4H$jBsuR1@bZ#ttXqMfZ);Q268?9K(|(2j?LDL(T{xC)9yv zSyk=8)aGrR_(0Bw#M}U|Y6J=5>bvz|RE8WLrLOumS!u=ZPKfxZi&7&<{Npt+9dM}! zbc&9izJR1?;AVwkQ^qX7;gYPG+tL0o?u%UN7oJ<&^>-xFc+_Q{TzMwo)8(ShOc!|G zUC7Y^f5mYMb%T!WhY*T4VbSX$ zpoYz!H+%UVnQ@d36Kgn%10Qm1n4>3&MvU;2I3^{bX4Da>2EsCtv5s~nMWX>{ZFEZ_ zh~WjtFCRlnDTK0+;~`Vule3XHW*f}UW8)#gRyBShF_ zLON{DoC1134wnvRB)CKCx{&8iHQdmpRBF*YKFc#bD7nHNWsMx1fneKRjMbQ!c%@G2 zxrfZ5FVd1>BF|GQM7*J;nex6n8Qo zjc$Ti>!oL7~#4-oHIwp{0cE~bpu6TcQ{l)1c-hPml$K0WzN=QJBR08G?c0I z#DG35FSSD2D~Fw`XOr6C%eMPVYvyTE{WzJA-z}8qU{dCRz9)mdMn%Cy5g$285NUTZ zz;;eRe?6^chShc$1aDFsgSljz-klRTm+R^R?_N4$=tt4@xY)`_jV`-SX>9(13fp2d zckkL|Q~dUb9Hy5^C-oJ#xBttTwvvRLjez4YEPW``rk5J&Lwoaa{?2>1p%wR~y!>77 zGKG-{ocXp6{zV}e8p+6FO_HgI&6_uC{|$S;-h(k_aJ+ADy@SkEKh2wSobENvT_V`! z2F{}^0ax~5QR=#BCS&3V_l(p9LgQe8$|Ku^bJ5mINJx;`QVe1IFSg!1p6a!Y8sE|= zluF8&sSGK}kXeR`RES81M1v_qL?kMtOc5GL2$_dUq70>EsLYhPWX_mbc-O6SI?wO- zzWa~o^PF?Gz4!O~z3=P3u63=ou43G#5zr~Vfg!92dIkc{3IrDaUQe5}^2Zu{q$dq^ zf&Yis_6|*WK%|F8yroD>jdW!|YSdA@nh?c!qyNR)B{+MN+Ag)sKf3n%JazhP`zN#L z{9E72oqxj4TI7G5yQDJH#ryP&%}K6-%6S1p`TN!~@}N=5B@zXrkdx>`5E2zyBZM?5 z<2+13#MYkG+_5tG5Kb%G$w5Ox1hP77X?dc(DK8PcP=Kyrk>2}k3Mw!)PPMu{-Y+D2 zZ0@MV5&RqFSJI?t2u?(jh!8Hdk<+b@rD)ioPv;9QQ8dnD($YDxQZL+Td@nLBLTv;6 z*0RM&GI)#htozY4^mRxH4#r1P7AUcz;BwD={zdfrI8Z$SuQPxjUYS$)bO&N6QBe}k zFT5CV1|`&c1s&ia8gnAc#%3bstT-cA49{G@WM&p?ArDR(kcZs9Nnl%CFttDm#_TrU zCCUnz*N zLfW`wGoaGL#dtbJ-pI?7b(#2ilZb*yUl2*Yeai}RbdX~TTtAu1Q)}@R*TH6i!Q@)# zw}8nK7cfFgMjGmkYpJ{O3_t%QcsUS^?gUHRA9BYn@NLQ(!oD#@_}nDpEKJ@EA4fu0 z1e~HP6f6%e(r>T?$;~7C5qYA4+Z49|A8%)_EXpe&l-C=qg}@*P=SvrPwmWnDosG|{ ziA}kKAG9PBX_Lm-R|Sfw^WOe8qPbuBy}pHkfRU`bPLF(|T`p5bZ9}Z;9ncWeNS7an zVz9T;PUid5OlG2#y%&m(@E04inoFk5lA)_ zHUWXt@Jqt_FI-|m2Z9(Aft{`helbBKanuWcu(q#R0j!V?onb#>zY8A_IN<8~uSFl6 zi2RB53y$X~oIex~&?HU45hrRZB9+5)KZz95!s%&QDQR*C1w~J;T!pX(VJRJKsP^I# z3B7tzxQJ5$f>^>sAuTteuP2sS+x4Fx$jTT)Ok0cfjW9^O9dNiPVHMO(aBSG%iGN4F zldv@jAPD}Avy00TXW5uALaKxL1wgScLj`7RNLfO5$WNFU4;|a>= zQGDcTxV}^I)W5v0sD^MsYsj1XUX)?vQEK`%{h!_i!8zpuEBpmGt7jYJP`aEfI*!Pr|BIOO}In?Ou`IDK)h6tE9Pxszj#o5?MH1Z`OI}YZOo4t zQ7QO)j`Fp*>RsKM?lbnDbT&?)^j#LL@m=C;h%PZuKp%kc2TD3eW9auMgND`qYA9a` ziRdn@l^FUPJ_!f>H|@Fx^haVPXGVA9MrCg!#ajW6O%JP6zjetjNp+FlEWRfud(jCf zWPjPKfz8j*X#qaH(Q%Wq@;(!U>3s{Q8Pl804m3m>zYpz@9uW~xvdn{Z)3GxvM46j5 z?R)SmZgS;^1p-{*m3MYs^}3A7DT~)r5*umQj;`o0*7Y*Xrb^H&eYw71>;#N8bBPE;P-@%r#4rg>Q(+Upf! zWxl_TD2@!3&N**>n3`pjChQd$v%4=KVTGU|^@{zm)TnZUG{O8^Hars`qpU2cX|ye* zflh!Kh#{f9#ilbQ?zHXETWh4u(-R>z^=PUwFHWDUL&hL3?btz&#;NeGZy)=b(flVIU$>c$wY4FS*ZBF!HCW3@5b~_C z>cDQJ*#jI^=`nVD%*573@^VWbbPZpAlTNd&V0DttZm|Ay2u#VyqLuhuqcs+!Xn0<| zZEk+8oR=u1{x1txaNhnxOIO=*!dA?z=yAK7T2dTvG&$&^{F0li8k@9L1^ma7xZ4|a z&VOn5u_s;k#aGO_pFi@roiC2?8Rb)+x4e9+9VXeleJ1)MhfRpU>8U9n`|HN?|8PV@ zHG))w_FQxp)B0W$J+M4;9cA2DOzGmUiG&l7Q!{k_H>84BBaCacv{D%NZV?*0Onm#H zv3*uv3o%Men}=}A{Gn0%vmn)?9!pGcjh30ed%xuC*9YS-?uTV$aZqJ%Oye_b93Ell zy$;OnpSRxoFcN6ncR<*P*R1!R6yM&|v5U^fGbI~eh+V#{JVS^^i?4*Gr6r#WC#VnJ zX85(vZYP#Pm^MQdzvwCOiS`fdPB}R! zwUJ)TTC*VkTL*X9YLZVbz63Z2KQzCYHg(^^X|zT5(6d(M1MSz14hWrC(UN<0b@tKB zzfiK~s@q|q-Va?ja?W)MzWZQlb96w5?p?@GV4C4otvFg{cFp|7t&fyWS$oM(c?QoZ zA9n^8YK3$4(odQcwa%=w5EH!-82F}$yNrpnnHFE#A6qV8UZwll>O(n`!kPFpBpO7> zP>y$&N4vQeeb^sj%=-7Oe!ORSOe^x^4ycwWXa!G**HtrfrCCXg{akhLKDhzch*=RD8@JNJU_>)a$m&TKQJN6mgXO>vgR&FvEQ`kjeoP) z*(qFnaRwN~a*6O=87V)nZ4{#A;LzN=Xop5_R?I2BT|!i|6FXGwFK^V+dY(N}YUOX1 zUf=U|{cSZm7QFSpr*)4(@npkQj@!Rwn0FeTv+20sb@lnvM^SDoRC3vOK%cw=@mv4&=y741yGFn3Y^R@}TfNSR?{;Kb z-l5Y~tB!XvM_cMDaP6|Xtn795?*=%f$hqN5=2_0*7~?H+dtF)5pFce?!tQmxg(4_e zRJ>nXkI-}etgL%`E}UM&b>53>>qd{tlkOo?Wqj9k7ZFro6l$4V;`=zfcP zA?029o~tS91J(lfE-!tiCRUmD(8QEmz>4Um7hh&So^Ck)C3n)5=RFT6zp88a(_Wq1 zeQX|)vQ)cwlW+gB=ihZ*Qo<&;u#IcTV6RomgiF8rYQ0?dfPEbIq8-eOiuakV{M+u> zoayt$b&s1B^(oV38?dL=Q96HIKY27&M*q36v$OK@Hwg28Ui$=O1=*x0b|{Yi+7KIm z&gLj*o_KZn?BlaX3OyfRXLCBmh+@~0%D=ND8z4!KUEnn?G-x6J2@{zTFzdOO_JqV2#f{Wd@_aV_7aEk=-?bF=au;xm z3$Q5SiF4PiN#hWQfX&slESedGFlc zsY99!UpR8vt0Zi=IQP2dDedFe%ej2(uv;oHqAy>3{ODjc`_D7XI8l+-UOkfP<6kdl zqamQK@%|mh=~wm2)1%Aj14>r@-MT?kd1cOCva?~|J8QJfPMhN=#?dP)xnCZ>N-pKt zf6Lrw?Go33G=;H(ZwC(7w-(EOq8g6$8Fc%&rQ3a$BgTag3wbwQgbbD-WDyY&1m{FI|2t&t7S`~#yb0Ck8>r1dMWt)+iFklc;zJ=1))*STkIPZxgZK&zc^qIYJs>M8*X&iZUz)iG?i+C!KJ>PXyXq zLC*@5am8AYN%2Dy`(S)aB*?E_#%b^u?;vE)aO^(@C)k`}s0hg`Vd=pG{vTRp53Xej zT}9TNMeiq*x)|zz)?EOUL&1;M-02PIx zNW}08pcI)b0FmoP!1ixZ`+|gcZt5`F5qJqud9M9(=_LOtj(yx>I(EB24XM&eN+mkGS$Gji&d+2^@-y0v1~^RKFlPzkrp`pv;xRAu0||Miz{y?c!ED;L8= z=@psAF?)~2ohm?ouTW$ip{bJ69vW-^T4Eb&q3FYSTI?bo;edQZh!FWz8p7P zBd`n6iJ|Y`W9qvGI>?k9kR}Pr9I(E;flVX+(E8dGbFAW9tBG0(r7tmdCC?AF zqEwS=E3=Y-XC4cOtq>+Yd~^U`1R|3jw#Rx|5d&H zxj);VH0UCEcpJNx#;Q$>qbhKHRDY|Uk@;!P)Y)B9j$yWaTY~7ZudfnPEb_U6k{cmgO5S%cCntv!U1j2H#p1=t#^#T0k<^?opi;l3^zBByLcEeTI-+ul2J4>c-Zw)#?V`V(F8kOo_#a+R*zTFd&7;q8V zaaR0WTmRiCOI4F{D@iK@`TgC2+X48I`?NkkeD@NL%fnQJ9|Sz8?g&Jy!Y2zU5Z6S5 z0)^fxA#-UrnZZ4cpwB$!I9a|$1Qnx9Y=El_q>!oWcJJ;VN!o2k&k zApQQ-kLXjODoGk8^KihhBPbK#V-HbbiRrsooI>G25k|ByU5{rQqeM6f7g70Hl z$KkPrjwcZVk>(^Uv{%y8!+S#~i+Gm9Ei*m$gbCmrqVm<247{uisJ+_os#j}ni7ZLxxkHX$6edA8|!oE}2CS2BYUQM3c;4qSn z6AOo@z~JA!rtC>=i1P2bigX{}s$Di~J)Wdc_c>ZYVL`WPljf@bN*TF`taELL{YJdw$`6qLHi@|TIR#F6hbyY3j=*qt`8Gv(-1Ra6La%b#2*s| zD3A*-zZK?_@FU00MEPEP0-67y8>15$3YA0?&0B(1kdbg5#|6yfq5uQB1w=eR_hrfFb zipO1TOlAN=p;XWTOzz9KqwMXE&=%i7(mNvqhLmmD*fa#|SK3KieFwY;mhE6j*{n z#ls2n5*UHJ$108ewFgPgfg=lTPoAN#Mj`EJ#psd-Nq9!AOtw zo)>I15cYW34~LGrm$|nH-H5U@-l?ZAP?%P$eLHZs+_8glf3p2slBFk%`&V;{mW7bX z00dLTo`j9261?xp=p52<0M)QQ_Y(kgVz=y1U`718PQi;_idR9>TgiFy)Z0?3oHu_o z#6O=!-27?voAEbeq|8B;ixd6*shCW$Yu$m9J1wHso?Xm%=DW|qL35wwYnoBJu#gaK z1EfM%4idchUrP3W#mn2zZgZORYdXTsxy9?IcHUuuP(^x6 z?UY^AdGb$hB$`*f(_`K6qvIoB8IKq)Qvw{4GW7JBGu}vWI?(AN3Y?_h*TD9F|@LOaIjb1PDL5rKQ6|` z9vJS@)1{(1ee=fLp0Uk6ON8zob~(6L_Vd&pq3J26t%zy@#n7C4Vp*G@mq}Wygph$S zW&%T2icw%Bvy5{_k{f!785pm&-8i3h4ZBPs=CYBDR@CatTBJHEj?7r;*SKWq!X+w{ zKCW)*7m_U1FH8^QXl|z6w5iI~;}0-%YZq1=@qGn{$j8r58KNh$w$U!9Cd8GIo+@sg zSxzPIE&Tke0FjdR6D&Y(kj@7TfK(i5@GZqvLI-WgjUGllQN`XtsOU*dObVaNE?F8Y! zfjCyB*mqM(Qxfb?`{^|5i`m7cPOmQflJI+9aq$E z5z=vD(S5?XNi=FZj)wOtV{(7llsyYKF%s*9 z2A9e0)|D@PeEuz6xqNE1t>#VHV~osttxQd-HeFxg=!G#x)ZC;lTg(9%7`}b4f5b2| zw#pM}a8To8o&KE~aWuNt+XQI-eM^6*p+olwk_y+9;}ElO0P`w zzt8f!(ft;VJLb&{d0zkiZkL@f0`Rle4dmPnj?G(TxRl4o+<6pdf*(%z->=%wz;d1L z)!lEWgB8!RScdP|B%Z;peh4QggZZC*k1NbGe)`PLVQz_m@6Cp<%3PC***hL<=05-S zmW!+9-%~hdQms~AI-{d?nb}$Mbln4=G_v|cFISc|XIy0nI{e@FanG78y|S=v)6LS{ zW7aED*k(?BDNF*8hA-^=d(U0uW1nS^=q~7S!X^WdKj&l2JiVd92(ksFq#9aPr*YmaN4Q{+<U+j8j?YoQmL1AsJ2YJ_jzv_nK)!SLCZF$8HpS@${E7DVQX}j9puA~9w_r#rG z@uhuA+56|4iKesBmAj?{PFURP)7jQhbS#dR1&?s`-*<7D$+}*P$`cHVNHVRvNU5in z6}HeYyf<7UteN@t$@V`Rhtd7`@h{ouMs9ar(yi}BzEvg{ME)|$~ zJ8AW{bw$G#2ip9y;Nwx6k9Q6~ayjEo=jTJS^zZkjoNj(~J&K#IjPha=O9}+%)VB5V1%)&dk1UCG;Cp1U6mcA9={^4fNeyMP?2h!2HcUAe0 zKHzY@skTJ!&YzHsw`SHJ#!ZQMcW#ctMm3s2+k=M$+-emT0%r)ZH|Ko?*0iNI(dX?)Kgal!LuM{m8*LA5Y zzo(>|Aek&IE?y3Hi6`&Sufxi*h8M*HmI>|EyX^OQ>-~-Fn_ACeVu0UJo-jiv8}1md zx|P*-lpp~I@3JebWSd;KL-CJCLWJseXko#h6KF6Nz^4*l?RP|k3c?eUy%|o*@?Ql` zCcj#qRgQ~G9JKJ38ZUl)N$JGm6#>7AS+^B>9#z%*x68MoKa`wMqD)1bV5<(?X=k)ACS`4 zlR42y&bguE2dM)iEFkD6#CD=XB*Z9Wtsa1|z9BDr4-DznmQ!3@+;<`a-1hno+t|l^ z-nuzXz@)DoY}iv1w-R{12VW#x%IhaagYE@WBx`Fr1uhB{G&Tp7k zv`IuAo*v`p`j`K7aZx(!=6#>BF)Px~B2TvBS$-THN`LLDBnXn13bWi0axV8fj53u8y#KJ_7{{p8FHi9mGptr248V%_tNvv-eRG{gX@697koH z=`>_SytbhJv~a}#=)HE{MU0(@Ayb)6>+4lJcXs4%UcnHwch%KbEcNXQKV=^ts{X__ zta-O|c)NAdGdhM)+K-wrQet3yO4`!k;6T_&Lwx>W3`s$LONP(^yJBEuT!uk|A0Nqc zP>A&kxHBD~Fz)zXa?|J5;hAZTl-%DEPaT^bG)%aZE+l#f zIh%4F1>sB?t-;q0J0CDxwg9?8owa*oCfS(m$n^`DiCNStKqO7V4{SRM;SUTPo_PTJ2#7|Emoltl*nDs8%YyT= z6UA*+UwP`@vb(*yvgAmE)^b`uAHRPRV=r}}jZiRS5SY$Z&&qR8QvG}qc+{kwsFHX7 z+%Nvzcyhy55og_B`zJTbEf4sgzeCq6CP8Tn*4)jTUs%UjA~{vh-r3%;eCm|#nge?- z-1eF0_}kbhTD#((uyB)y^+aNow|Vl4ah(uOAXdFlgM z4QEw94V%nzCVU$(`BDhQ(m(i&$Mnoh*OW9v)hti{aZz!rIlA=NUa6)%Q_l-;PgpPh zDhY#k5IvI@4Kk!TU-$Py=CPl*9-oLC`8r&6Th)=a0@m;A^_FqT%CaHcK{$mivpEsf z-AS-}52&kahsMRne@9^22Op3{A&WuI)0NyjTmr0XUi`kES8=2#r6RZ#a~K}CPrLtY zadOtKYb^eDwjy?tUCNWIdas3Y(gudu#GW_T`NWkww}f$z_EN3>vX=3^y>gW)6{F_f zqyr3zTH3J(*6?z2dakX$^RS(>eJ` zP{jMu`Z9MnsorQd3k#F@EVExM(mO)v@oZph|s?`NnmSjDy?lV4_B$vKP$W)g>4!dayE2&y%Ku((lLTNH z6Bi(qz`)R0!S7TG_shOGDRb;})xESYpWMSchZ_?2m%sQ-w>(PmF&-FxmfFFC&p6<3 zqVH|8p>ZSJl@wqImgcj9LTQFy%lLD&re;TQFQW-HGxbEnl|kV0G;7v4smM_Hq$MRE z%$E5|otHJVy174VAG7pjo}(+MLTKnx)>&ZMG4cFIcdoOm%K{%)AngGR}B~aMDrtEZf_uqE2Q;=wkBhN|%HE)OdD$3oh*0VM_~Mc4wu zy`wK0dMBEiw(h3llU<{`B7tx4We5F0%PU=+sapH?c@O4vvqQuESuP(piw(*pXzeZ< zf-Kt}22*S?r$K{XH+_4HWbLK@emf}Y~4aJNE&l+``zX6P*miW&`9!E z#mQEg=9|$J?X992Y9+$7vhdt>LEn%q0L{VUcr>Gv-N*Fs(R$caMZXXkFqX#Fo|(eO z9iKRE^E%o8>+CI|(qD8{)K2nTHfwfnTTZQUzp_w(8-jq*(eLiX6ugqcLh4w=y0hPl z_#VI0awap$bx6}fACC6eT-vyy)#%8M17)^=E~>1g1hWyAJrenefpKKK`}v&Cjbv@U z=cl502*L)-*xRHBRKKIRGLTb^W4bqcM}tIs_kt8wNLQ(u+t?fRIOgN;>zCH~)>&Vl zk>17VQRcJ!*s8Bq;&UAw3zcNtq3Y=PxDGmaZZe{x+7KRTU%8SGwq$$KJ#1mRFIF!U zn~E03U+#(Tb^c~bu{}6FtuHnwWWB2cS4-zpJ@u-%tjNOglrSZf0gLZ}oLes8G?Tqo zAv+migc-B4b7M6VBhX(t_ExK40Im?-cHD`Va4#|(-b4|&?wNkK-^fmb|#=T(?r!3!wk5vx0wnW^cQ8-=9rpF7xnjDnnZcHhw)~wOBw|{zP-5WL@ zUfy2xn$iL^OTYpTy?eL6ZxcqYz=^HOI=dwvgG+LxXeb7`zb^~9bRHUb8o~F1XWR1t zc6P_??O_G*9i{N@MUaB4@xlvDJLudH2KQ`RYp{c#U$w}FBkhOiq=i1(k^&!h>c5CE zo1PYPaAc);#G(lY6?%@;a!SZe>PI`P-oA5u|GB$#e5}S_qB&XLWk|x7 zM?EFq3_fukM2QYXbP_$O*{uSiFW|8NKvfN$FwCJy@m=>;=W>x-C&$20-=>S^iRMo` zPYaGuJ_$9qdt?x7_Uv?jN>BnSRk3rGg>`n6Li6~;50k*SHV5zc0j6@u$V@TPDl7|m zl%5{{sgl!jWcS*SCEhM(@2q(nFELWouD3<1!R>i3Ikm=`FH|JrZ$~G0o}*P(QQ?R1 zD&D;B=C();^tcQoHt0MQYTH`$yywxzBUxiB?HnT%zyH!2Q#&63Yuf%kK1#bbgm|wZ z%1V0@>7q9D`d0vovN#N5;IbK}EjsydJK|Ks8Yat2eftvcwQ8r`eKa`VO--|UrNen) zip~6cMxkhYhElVC z@r~m7*~;=~erx}%eIsMz-Xh;;JUV&tk-tAhTkgHpg@Bka z6+HQ{_O;l8PQJ;7J?ELdr7jli9bgu-vlpT0&^et9y>0%$1#w;!^8tCcF%|TC^K^Dg zM=}{j;OX@u-8c%B7q2Q>+gHgM55h$vwyeuHFQzfH0WBeJ%KiDYeC*@yiUdq?YeZeR zgnISH1FRC~BQVS-3MvRb85x5$vSYW&q#>AZmsn(}1*4CbhO@+?G?+Iy77_LYqz9D{ zzWmI=K*V%N)l=c}h?xk?lJ>FcHMJ0Dsg2`uJYt}*dJgpn8w_%xeWABJpErJN+mj`f z59;BGM~oxmc>-(R*H1O%8#uKHH($bOa~!A5Ik@1HU%Ua%NH`1)Aqn33OeiBA5^lv0 zNO)3O;W=UrB^G*hb%N=yW8iUa0Xb$c4YL6hH8U`#bU>`-gV!kz#%c_Tl1J5gFc_}% z9Em8_XCQ8_g=!in??I?ps~}sf%{1MwTJ63tYfPRsHjqX0%S_0awF7!Uh(rH=e&p_}ob}Z+35jftLkK*z=zJnMIgGtlhb|UYnU#%^sum*R1}r%I zRU*DbA0F}8=I6!%by**WqHiS;90d`xn);h|9t0N zgoFexV+jF)B#lqT1&7a_OTZM1pJD-I zCIi~MJM86fE}^(_pG-Br{prk6Fg(<2Lf(HjI`{$ui4rx{xu<7m%0KjV^zFGg|7L#F zC$mIX{Isigj?`h!eMT8OD#l5EfDGP{?ZfAX?w^|(8&bwzo{7^9Yk<3as-Ko&gs#yg z719F6V}Q_etWs%BBA(0)tcX*w;M&M?6$Mp;oWwgmt)ghXNxROG|2yLFiJH^=XTJU@ z$@O>a-x5#1#Fm{|kGB@AChVry%|-=e{p7p7g4cH{nUe*bk$#>_mRqwhjeDQn;NYXx z)0X~^XM7W2!J<;?9KT4Zk6}QR1x%n~TPUFM8gU6EVOqA+$ny0%z8rpqQ3&avQ zed*KX%LcQ1i=go*1A9v=Dk`oFR#={a{MY~`L3E@`A3w%}evZ%D>EL-mdSYX z7dn#sgyh|QB=26BYg_2W#Eu3m3CH)0x*^m3V0(OJva^9bow!YF3T8Fwkcn};jLmF1 zkjuKdojMgWcRmp^B>rvN_9Iez4ZekQL~W?9k$JkL-FM}n^l)^rdmK?b>765!2tig- z8T>f28~gE+E7QZt^O<^OKMp-G*>rkB|8%J^?fv+~3+_3qVF6GRViivtPBzO|^;s4z zr{9yk0oW{l|7@0IY!=H%*drov&Ss{k+XOxIIPt+N z!9~i~Mrh4ZnRE1~Rj<@^ZoPhUFU55mT4a^S=y7tNV7eaUW96 zn`utE$2|U!g;pVSKQ3>-8C3oivYc&$oCGkFmatFgd{uF`pjo zw5W;q=OK575CjmIcB)#yw-$!?Sy0x{LG%~AfrBF*&`*ySlhJJ+L`?yNTM*;4AdTix zCFuaRQeG+9v}Yf=9m6=!i8D|amjA$jca$AQqmxT13OYJq_+d&#Qy`*SG7c-oBuRK9 zs93N4h104(Kzgw9p;Ie=;>lcMCo+y%Z5Fr(Dt&iR&f!$($4N`GTJWBQ5$Z-QcSA~O z;Qc$vYtH0Hi5?s&_fW32NOh-W?|P?A(eOZ)4<=eaRATjH5E5>l>iZ-8g4Wdi5TEH5 zT$$5`Un91>IUp?@rJ<-8xr&5|x$&X+AWb$j zKh8{aG+n?Ctpg8cHq8@+$||u2gU2|t4#F|%1y3&yoqFgmYccOaRSQ!$YEceChNQSN ze<}ufTpVl@^qg88nWWb)q1dj%=#+Fs_Qt?;JhQx+zGQOfldQoUr`8N^&Vc**yV{2j z2S7j-^MTzprE(erJiatk{Z6*YpWcTz)vVt6I*M4V{F9O(lRi@3(2zaW#_FMFTIILq z`EFO6k!5mMnOzr5pX#T4UcHRs9)&CyteOVQ!A-&O*XPZ0UIS<6&c35s`++7zBZuLG z0uq;Q0eLN{s*#r+3}$y5(*%>k0xmden34p8uq=?ffD$ws6VA!#@NgxRcOjSq2DXqs z62P!Jm-#h!A;#20PMtMG2GlC#&Cm2t*d^B`;eH)6E%wvI>hJ9DPevKiA9vSC;Lgx` zibqmkF^f$tAg459MUUYQ4}?vwp=y3YasW)76ghMhKW;+$$-zD_8OkauoL+uCXvdO4 zd^~^(g$4!&VxU7hLO=PjQjUEgt*%3|H!2d7r0HJ?rXBCK`!sLV+|{B#^{ZL&=!y!V zJ0@izw7zLM6Nl>pZhe>9c>U3{>ST3k`p3V?IgS+C0LzvfzU3))j-PX+b8ne$@+K1n zRPStMJkin7@fcd?bKpn3uSe&h_i>F}3eZS~U8abmrwi<7wH zfuWv}OW2uDm-Wwn+TI&D$E*I(Mqrc3hT~6CbF+uqk4dY(V!85JEnQ?_kKZT>3P{mv zk=u#b7$IRx;z!XNC^K5Zo3~F{*)!#sWzehC`PW~|jdjwE46$pTOQgqMi(8H)55iz1 z!jHNB_tx$B`Kz>s@ht1z_xfXvda1{b&%GThnYwvvzQ$XV@78~J&YwRTF~9=ZNLm20 z-jm%pvdO2Y-a{D|wsgiT*6kOk!srjPYb#Eq^vn_xMHDebXInwyVf~M%riM z!bYLk3X;{1j5Zlq+!0tGS6^Qvhj~kTNWHH0rsb%6>&Y`1yAGVG0q9^FevkzyJ1#$h z_QQjlZ=>CcqPsXd7&C4=>FrKK%u+tqMXl+3%8?~t!n+cs8U0@fe__mDqR4HLg zUvw^iZL-Pb`^{|zx7SiA+A0z)4tFq?ZM^pA5q|6~U_uSBPtgYDT2%`r5~_utlYitb+IrLN7oZmTHDJ*%#j|rL@1*dxr;6sDdShSc1Yy zQ+Gd=hDwUFdD@FFFPNmyyUd;&Ygq_Oe-ijhkxF5e?TUJpJ@!9*0yyN;0O{T602JbA z&xNn~OoYB}80rZl+y{zEO1!E#i>eWGVXv@5wHh^D3Wfk>|#>rdG9f+{xA z70N@7X=$fHZPW{)uq;OLlmOt490~^aHYl79jl7pOJ(QPHW7PY;URlg4`owZNr&9tH zK1sRnf@CbWIYtIZxlW`E7A#yr_5dz&A5Mv5=qH_i5D!oKbU0pA*J)y3hEHSl)xwQo z)(*!iSaPx9=W|ubz&a4MVi9h@q&p78FzTs@b$poYugHtej0<~TGi*wj9lWF$dF;rx zt<)aD1j8U86iSZgr^`2zV zbjVpr1xUv(qlemT;zDTU1Cgrd2KCr_SK z*Z4kaq^2-&n>VNm%}*SPyL4vbiKCOZ0+?%0Mh7~NHiAu% zS|BP7@l;*{yjbrB2L1K@kdk}J@lksm488j>eJIlPpzH6B zTd5;{x2o@*Sl~^5-}lapqTzyqGyw^Lnwr{E4vdsJT7ZUG^aG$c8*PD2tM~!xfETqc zqJj=G2vX@Hn!X6^?(QbgF4Cz`z{?3%;)hQpWlhU(FmCURGs$j_v)Ee~!9+vZv5p|r zcoV$eP2>kbK_o-XNEwcyj>wDaK+UYfaCN450LOzpB*kU1c^wTejNYf^J8E!I=*02O zPQiLT1nmIJw47pK`r2Dps6WMa96+B8sd! zHo41sH7KtIGxH**&doHszA>l-AB=tsRWrL1vR7~AK!s^^TrR*WsCu*@5jF; z?G(nAk+zDy?uxdCd5`rt%U@WfxEr}8(-zsD5}~|wz$)RxA0jwkbfm?FQIS&y?1{$& z;zK#pa;f{FS24_f(YlZjGR&=2Sw*Go@b!4S_@QaV@M~9gyeP&w+MR0--A zf8UN#N|vkLyYTLmWs;*sJd0HGyV|`J#*Mx&?^Ys%uZ4#(p064iXodj0fsM`Z#~MnJ zFe&+8^6}zy*^kyH_Dk-`PY^>fZr8Bg}F_yjfA(=8VHd!4ymyR*3sSjD`DP@~*6ZLcKdhOJ59CZE5#q_lx6*pP8nJK zOS3zlx{fQ!>rM@o!gM=*%4*?5&q{JID1ekh7x$7Rvzf|Y5L);V8&=ITqa zzh&kk$em|!I*H8{@(ej+$jcYFFP-<6DyW1-?y!FJLOYyVLZB@}GG&$Dgl>1_^}0PV znTx+I<_~a-(F8`{&1I&i0~(%DX=EWr%%mDhx%3@;NUM55ucOmcGmo9k@lz*tLg?&O^RlSfzdVS}B;fs|A=`1fztift0KN+9STF2!jg;Ek-_ z(DpZG%T$tX2r5bGZEHPfI&1mOEBtJp%b6vsGq#*l3%=IBJlQq7pK(1i`RI05(Tp8Y zSSgF=U0DmAL~|4zuN2mARj)){Ysj|3gC+B8Y-svIe9>{=5ax()8O7TQla{xn)z*!> zN=mX(_N<{xS-r=h=l0@47Kp*maOqFYHejRn2tQsh=J!9NJ~ zItL4JimmuyA@V!Evk-*PSKv1EqR;w3*7XIUDPhf(l$KInCP%QE50r;LHx_iL=Poz# zclWu$?9`u>HlDlqYdJ_q%flYtpLwg(lFc8)!wH=r*61p?e?I1ZQBhL5``nQ+yBi(s zky_eMwlX*OojEQ9WSE-L{rJxsAV~2Iy6wL2V?b{sIWM?rVW{3#S5w(axH@$UIu4%I zXBr9bSfWJpN=83VdO=BZ9$m5(lpTuHtFb;KK1IqlApzB1+uK`RQIlU|BPY*S^q?bS zD_`}AyH34{hqI)}k9|om=tc5b4;JA=y|a__`~y=BQaz7)GK!l{+{%5LvpoBy=Em#U z7br>e^MSg4q*(&$huvb7Uxg8{>i8#3G+*Z;KAounQj(5Y0?EME^`DyET)lpY?TRLe zPClm*R_m1uE#F?61ijU-JN9Tog8!q$$I@O(Vb-~mjHcDGsswYG`(lyRhi>OZ!!2aP z%Z|KTPm};aT?V=%-6oD7x+lCB(?}X{*o&fIMjP=6a=yaQ!w-WeUT3P4UDXuaov}66 zyWqk539VV?frr>227$|u?E=Rs^!|NwdGSL>Cy)*LB<29X3|~_O3jWCly3Bg>Y+%PdXl>|KasPe83*vG-@hI$XhRzK9L{F|yVOyL z9z$P$u$8-b_4-r>+A@b1`i%6f%PXY0D7ibwaRX<6eTVAKEIIrWQIt^=c8IZ zn{S9wz1djz@@d|3(frTf_E5AL5gOoYcq}tBGdLD)0ZY^bE~BN}3?pbz9Ezg1mPa0a zi*twn=*UPF)Jl#>`x%qIW}4mrjyDwJNs0I*8fR1R`OZ)}hq+-7=M6Dzs&h><%$MKW zPCHNSk=R0cxtXNlAO!$4tEk(BXs~PyH^mZw96AG$Jc&3`F=02@?J{(EBx55DZ}Q33 zvlsaXnueasw(q}hpK&~dQ>AvzF{>w>NgQeySSZLUG_udP(6Av{??(@ThK44%prF84 z3l+;POm#W{m02a2c|58(+k@&XvI1vkW-JJJ4^$S!5!;v_X1yPdbc6)WXKGy+VXgP{ zKCb=DG{r@tXEAI>Zcn4v2>Joi1_+C!Z?ysaIBXPtd^9qUYFN>R+_}@y=Y#suU(BY} z0JZbZ+SkFcIGD|mIws*^`vz0lhW)fS=CehRPue8mm>BMKBMSzHLO1(6AenG1q=hh5O8!fK*hO_v?#E? z6QO?`9p1Ch-+**|9Mx1wMa2=6+XeWsoo_=Z5his4rMVZQ#g|#zBbiWRaHOssFALWZ zObEMBMCHNKL7%dYbi6RPLTZtS0OdF-T2nD?8UA%z`B8!S1sGLw;#Hz&#O=q2>1aJOuRDp>y^(OsvRkpv4r9Pw&dirTs4Vi z+v^0j#v*Ou828+EVq_me+y$|KPk3sE$^R`;rD;MQ26gS{ff)}2sS@FXx6 zj44G`O)dD*qesYbu{gPq^BWl%O?2R7DHEn0@4axvDH9+bahe3@C@R9bY3oCyd`(w> zUgVV+7RnZtGZCr9jmIK9Me5 z1c>%>qJsdMu==#9it(B=K&6ZN8v#f`Qtfb9{=(>Uzx6n63Od%jQ&Ng+?>Z#bbL*Q3 zZBgkes;E;Zncl7=%mA2^rh!aE5OK2OkaMS^IdTpKvkNr5q39)`qJAkQfFx))mTwKI z5%;rSQU2V;h=ej=LpV(C+O8VI(g8?3Paj{GcK*Web zq7KJi^Wa$0L;$B@Vg#oNDaA>T8~dGag`lol{UP~2-64{$Gj7h-i}PVNvS_>%rGLz_ z^O`3`!w&b=3t6cVX=##>iqw|yQ4m_RNU{qY4X|>q6!0EfC0u?D8AkwS|AAma*+Q>= z))@>w$jIXXPUH%3d5Bybj(Tow9<$_*Tvg`qQF_ddjg6|et}U3>Ku@i(4N#97(6yOv z`GrK%Y)lJ4G9a+l*w~nElW^fTOsx1+i_z+&=rs}%iJdn7kzC&zn8o$e>4w^dCD|ENz{&mss} zD;IZM)G_NikfxbkWabWt=jHDzdMcY7>B=lup5M+2&Nn^x;Hfpr2{7%=WJAFTA zj>#`Oua$2isMXOgxa`=k2F8cp37XE_OtJWG-#_1-d?I^rSs+?;cns48gWC38io9Nt z{2~Z+-!#1yPdI6}70($0>=Y3H=l&eba4&f7Shrb)eM-AYk+sckGXv)(QJ<=I`s}i% zq9(HRzZ0q)= z_ic*CDzY)L4iLAR$VrDlxvC~4e_o&F>od&{A4An94&z#SdA*cf_+Ebs?W>ClT&=Z4 zp;G~?9omX0!ixOo?{0nyk27sr$LKtEaE{gfWywpE13^K-StF;Jn`byipS$F4;iYKP z6T$-4`V6pwanL|tf*pky(?Ru+u$`s4MJyJ)Vkk#rM z3-n*TzM}o$?3o~DV|jkzwG9!(hQT>n#Qia_MEvvKM9 z!F{ardkbdLj(UqdIdDiL;|I&P^v2ZOgo7r?f7o_JmBY>v41&<>IFo3zYg%@TEW-A=Sc3AYK8 zeN^Tv4fF^n5+U+JVS@>#WCdgTa?`&klptoBlkZr42VbfCYiw+>1%@siXl0ptW9`@1 zG%`7^(8%QBRV@!w%?kLPz=mq4DOfjl>vr`~+7aEy^}7AH`gC8CC3~JUCeWs?LN6%+ zs$v05+5viSHhQGJ{Up%9L%p_M#g8)un2bs48-&OW^a6SrD_l zJpY!G+cv-InZK9`TeeauNCH_{_dwby4dr_py7=0`9bS4USJDx}C8Yr}*j`weu5dSC zqbgLHnRo4dH*(of2xZI|AyFYRPv3qnJ@5Ox@Atjm|8Lv>+wN_9>SG--|W><1tLr*_F3bsg#d(Z9c8 zdhU`LQ>%8&+d4Uh)j=xB&!=A$9iP{ouU^AIQ6)(U}{NS%m8)(B@m$mxgBZVj;; z3>@H|RnVSm<*Cub5JXXcGr8{Uf_UJ}J{f#JeR+n+6}7Wu!tND3#Sll!TL zeg&&g?UjEu#9A9w*W%M3wxF09%>6DhGql=`aPDBeG5|h|El6)}fM~NEGMU4!9=cq8 zE$+>S$Gdmn3=VCnQEJfB&thzh78b9wQODOLi2OHm3Oda__Wl z9?~;9I=rp5VP6Z3|Gvor1 zKR-dR%QmVz0YRWpmdq+jAG~^H&~4!N}%?T;JKh_iVI)CN*cs&y?rdgoT4kZwfM$LD#uWxp-u6)6c{G!TS39Q(LR-@Lhk)H)X| z1r`<-UKxB)EotN-U?##)+X+li+$9|Shh$>nWEt6HuN#&1T79wp$#@EPiernTvkZ60 z-i8l*N({0!$j(B>5rdH(kFnX&si^64z+e~BKEX0b>|@6px!0n*JvAEJ_M5Zx$Gpn- z5_`79U%&i4Ja|>p({#BOqW>!TtB;OLyx}N=|z!`eh#R?pt<~t}^N*7ZX$R#)Eu&)x-9ld1@G; z`luk%RoM%f}qx$*q83I`Ze5Wy~QNF;gSq0P%3+Uh9&Vfc0+{P)blG(u7|Ao3PNknN)C`}yeO6P{fHmPA)9BEt=WDCR3*iwYkGU+mo*%gU#%DDLlQB((oP@-X z8;o-CRhGd>WRH69R}sAwuV5&wdQrnT+Nwke&-L(#%K6df18&U}oK@}*yY|K_X}mCh zq1?=9b3=(_%lqzs|zjW#&j8I=+^TEgq`n(64B(@MnWEp@;EJDu^b@PbjKyuP>|3 zwWV5o74YPoIwy6+kD2U`6xQ;bAG=nY+dj)r(G@wxK;g$0;Z;f3J?IIcfEyLl;QO+tuKS0 zITwi}A{MdUJDY9V03B$R130RIpia)b_8CT^^B;Ua87CZlS#UNinELuK4ATcQYx4Rc zoW)=VrJ1Vcf_70ab+x7E&(_#SQoNBVbtC)cW9^P_5=I_Er%b>4=+ko@$Bt^Z3$0X=+scUU4>=s>5ANyfPd8M+_J&pmpQbpA@DT5v{2;JmF zBt3Q5SoZYyCj!$;Kv!~PxN`(Iq-h@x`jOF5SRv_ryJlmPg_bnu!OI}X(xe>-k%5!; zpYr~X_iaA_Dqc{pxSX8a_9rTq>I7a# zl`_$0edeWDI>FRIS5d;$;^6f3bkd|E@-)KPFL)~^PhbV2XMuwg;q?>#0Hz3p7HpG} z(ncPgad7UpGT4f=osHHKqk)2q%tti-NaOYtm2?Azd- z4P@ww+7^J=5FCy`xH5gxbK@XJ5=0%a!j_tcU@`)Iwvaw({q-(7L#^a3P|D+qK=$)abL_Nh&)DS!_v6uip_On$TJq2|nI5uD2ti)!O1$*x-xP#f4 zAS7<4HL|w>tz1G+CS6ikS$(R3q2xKBHiAkR>=1)+oBiP>M*9p95**v2g4A2DthH6U zqod<^wYE_3F2?}F4JSei)Kp)IONnlmlYFt}@SE~SEfx>cO-&+JJt|_{`A^3A>8SjC zX6xf;bTnrwuC{7K@9OGX{r`xD*hlfIm10h92^fKXOP;A+sj z#;MWFotp*kW#b_{Ld51&9xO~P;`1?VN4h;$moXW&3Km`&;6{9+=o-JN^) z8juV=`8RSTc)oQw(D(t;q9kH-a)p)_N7Cr=BXz<+UM#0~N0>U==w#8r^-RFn4+092M3JZ?O7~c!pX(%}w;h3m%FN2TO-9mT zI0|91KK{%m=X)<+ydb3>UfCbvJoph*3+v+ocV)<*N3=u@Ku;Yc}Y+5*6? zvU4r}XS(K(O*!DPZx<06q>d~bG%jQ6?~CLOjWZscQITFg#DDJmskQ#p3;w3j(F09W zdpsv^=4=rJ6N!9%EPo|UJ%IZ^;t|;bDUjapCSzWV1eg%TP~}w-3G~PlNkkX z;g$d};ka@20cnIp)ab_x4OLWBqQi<@9|0?ObaZ6Ndw5p0J@i{!3-2HKPUIP%&2?W6 zzT}Y!_GrD){_}I9)I#{nMZPoP0-p{<$j1YxlYyBFZ|66w8IdLzNZ3{qd$6k*584Bw z20~bS+(oB2NJ%-?laZ18WZ6Q@XCcO*%gpl!_KO`G8K<63_P*HiTdne&8tFptuSeMa zHWa32(2ED4i4Y6{x3IEr?`Jy7mSm;kwo}|qP1SxIl;6rv9-J7N7z^=haA95bd(C*r z$-Bo3p9-|@e&1lluye%2)-f#}=Vj1?2j3vX8JxBLng)?gnfUgCimcsalV_oZWo$IX z3GjjI1Le-;XELLGY$gZbuZ(&^L(qUq=bF=-E!~N#DT%nHn6Guc_2r9bLx?vO+7=n@G@n;%9$u{ z$yq9&g~f!^R{gKx^@BfjfHpD>t*~LSpWkP_Na7CCHE0Q?Nj-mNHxXjh6>!@#1<%n9(A#0pPkp7dCJ}3kgTu$A~3F#~%ijD|A~V`E>rtW#5KV&gm2g z*bIz}bx71wK*s%K&ua>`Katl!p`k-s0lYE@)-Oorj~qdcO8RX8rHpX`I{hkT!|+sr z1GB>=x!=?BK3)Ht;_ibG7l(A5Fw{PRqDuG%GY>#_F)y04Be~7^>T?u{;DD_zQwFb? zU%NkBDH$ZG9e<;iv0;+_)Vy08=cbm&H>US+eIv=|(-S!6g;?coe|->k| z(8`RSly*ix+_u&1g^b{*i&AJh6QzZ$>>bp437DuRuw%yuU1=nTjt&l2K4n~}c>lg0 zI9w{236;j#EhfYm3hLg*LTzGEH3o#XwJcRFnYhT1h=FvGWMo7y8lvlzsPcI|Dk)5N zJxzp|D7Iq>>1+|ss|DLc1ar^S%a#e=G^mUT;NvX`PS032)bAbjTFTVP;Z&E~@9X!% z#KmL2w~veu+fAKd0%k&b!Z;cRGOm*>6)a>y5M_}RD8V{AQLj~O!8I)8!UnfIx%*f> z!}kxOk;^e_ugCJ@6#eF*myE(l%fVcZ4V7mcm^DPNqh3Z43_su;6X`1GaIX8EpxNkL z^t&xgYyIiI>y^Ry*YhuL4s0E4t#1mPkuW7mT-0Rb|_3u{F|22(7g&pIEl{NIdS7)n}u|Ow`2#n!WX1PZc|fU ztW*(yxJPDYoMnHS?Ulsmlj-Q@YB0nxaZr2u;bCb^=k( zQ*u7(vE4OXV$}JCXPazg2-|)pg?q|J-tx(=c`u@}o^_j})2UrCEoRPw&vn^Imnu+- zL&KmLECFzh8wsl!!~*rQL>p3PB2d1VQJF@<7~5gB4(rGpKpFw+!)N4N)0VC&OpZM2 z%a-(OG-T;g25#p@lZ029FDfr;oV(9TQS2Zr6hs29h?>euWfbdIt*v{zx(K2wu}`zt z4m_iR-ve?jxz;AP?M!Uqe!Ms3+UI)g)xK8*>b0~cz#BxW;;_QMXhN9v5M&)0Dwv2^ zAFyQ5q?Dom&25h|aVC-a;NsPuZ&T4pn%tyJzie4UUf&~z2}XM-1&X^hK*!AwtG*61 z{evA%m?%Vmh59iac~Ra!I=!~=sE;*>E6?+JaqYFsIj;Eofdv8;4uL0X{PsXz)CDB_c zHX>URO~i9t(XykfCxZ0Aa=OG!b=);NoiqmDjtC0y6H(FSrE?ifs=~>cnh&-mMizLl!i1%d3LL z1rOq~&R-7aq67Q)lm3WVN4_SG+-E+@%hIyV9%h-Szv{mAhHJV#JpPN6<|IE%3!JZ##MQX_mgst4l*ac31dBCf@9X82?BKk7$iY*?fq2UGekNT+7{) zJXSQ)<&QOL^QO&>ZVD$~gG|}`;U~=MLyN{!JppNq%X2b8DykHX4d*Q1VV3l3Mktl3 zMK$4VlRCqy+3tVDRFMjgSfDlCn{ z8k_Zup6gUBK^k;y8&uPL@u6++N%N5i_F;~|R5ph*6Wj**N0JTbA2(H02tJU!5=K3( zyqW%UQ(Ey+PvfP_c#EXx>vleVA|6xbLPETg8pYiNIWb+G*cg;B#B|4L_E%YNUtc`o z_ej4$M<*IXK2_MIXfvz|bJw*>VsL?}hxcN)zFKf>oi;JAY7l2Bd@GA-GcP!9b< zmG#w<{Ug#?KmgcB%7_9+`l^1#qvHh|JJa6BHAxZ}>=>mR%6Qp%-{A131ZU(-k)a zF-t=$XV`C9oyvz-3&sI%=c2t;1bs?I&}kB92bxq0LicHiXzOw20euXwG2E^(A^6?zVe0GdCL`~r zkt5;<4=O1RZ=&z$7-+r}_;}glM;nUrgyZ%5Q78S62eA9llSvMnnMZ_0~C^^tOOa(-8w~^kLw^3ajBWR3K0>B ztBN>ha!QplzIVQyS^NFmN1etlN%3nT0|@cq{{SbvwMP$rYfcZ40s zG;A1brg}rRBZnnKG>kH}fo}p7A^mX=&1;b|4h|$7dMW984%`fBrqjxW-iF40dR^I{ zkr(<&cf>k0B*ebb8S*bZ|K%@=iY?)2$H;F77Epv+!TB%f672Ck}ss)T6;gbM2CfFNL+Y!OUa-xcnKFYb_5e#>=z)oE=VGa2rq+ zztSpa_Fbo+wp))AVsmjhvQgBCeFGcZ>wa_yqnY*UQc19AdP(}X9KG5p0%8eiEXY!? zB{elhnNMWwr>7MAmiHi}Q!1*OpMjVZF6}EZlIl9zu92{fM`PsIe0OoB+}zAP)%*7M zM~^I}v>`>ttNDn`j(AEzy?6-EG|H6Y^W(#vUma9L;sjjAf&%@|T5?)$Q}BM)p&a3F zk^XBu*yHHD#Euxl%p3Ljc2bv5(ou5uxyk{F;=hc6IErW(1W~Dv!A}%GFAmi^(c{8G zlXRcZ-;v$Cd2^aCLtvu(!x)zt25#*UYXd1s`P;7XKfAqR&kuCIyzQ~XmLqXN^~G-6 zzbT5#|7Z+A!zX>2%vtR~EVa5 zanG#P`7nlk!x63%pVevU!~8FWH|v;FqMHs1i6)0w-&jf!WL^Z2QV?!iirk9kh!vGq zu1z8nI252w%GQ#m8M-t%f?E`~J$)8CHOgqWDKI8uy|ht$*<)#W&Ol(2dXdshLaes^ zj(NtO^;UK&h-ya{iTF_0oY}6Pmi769IOO!iRs)3w$$3Tce(WN_1l6~tnPTTUbk-l` z`G43hB_DsL>Vu}A&`eH92s6jk!a}D}g@Kk4$#}!%6qVD9nME_~^O8};mJ@>-LOCJw zOTa9IO91x5Sc^$|KiKA1t&}>y{A}Sh`#fKE-AHiwgPU!!X!Sub7LMt^lIAVnDyVgI zdR9)&r=eeS35CV4>~jdVey0-XCW&C8dR7zWbuL=SI)Q)jYa1EWdeHUx9i=|M5>iGz zBEC7A8q&9kJ|ZGFcjC%%x?@tJJh;`}G-}_~#%Gt$93a_(?~*`9aL}esNWRF7J86#g zwUa!1Bd7B{(vh&4-#6CsTQT0Noy1zj=s$apI4J1qMu7U(d$E$$XsBbftnj3vKy}|4 zWqM&G2J<_u6?H#~i{m4;v5nloBpZp;-1LtrrBIY zd54@HEAy^(XU>fMbdDrSv*5K9#eE+Zw>xPK6i$8Jh~8mc!&&!T0W^X{;6rUo)2gZu zi237$igv~L^}S$!-(81UuF6D=qM^HguXDLo~nk-gQ&w$-pY{El=PJ$2Jlu+ zAQhX(%|rcj!#8Yxe44s6wAI3>XkzV5W`maHM8nI-tC7R}tMv!|p>%Hf^ZijMqK<={ zAI;0W&W)3^v)7wiBJG7PwO^}nwP(y4mb7||^3gJc@b9XcsT>~`4%0c%ON#U1 zpj*6NCK!hpg?UyB>*i~KTpQ)HZb`^CzLUrUW;0Kk=PGO~N;p?JW!rk>2#-bjsCbq3 zxzxCwT;XgeOO0r&Mj3vgmKLZGiFhv7Yvqtj-QWGQyQ4#CII{7|n!oNfW;e}OCL*B; zVxz@(3Y*Pr4jO0JAL~DxM28zakyQ^1wuz~X%+`%7rl#kp-TO|9%XTn8!b8Lld_?KU>rM1=0EyR3epy{-MhVlIPbL?IF;^q{ z7bL?FXm+Qbk`Gc#m2~?NnQ=-g^?P;Xj+o(LX?d>SBehYR{mRSgnr_1fP!*LEPVlQw z;ce$!H&6qMiZ2-*J63u3Em7Be8gqW)u*`^pblTQ_i}K$RMR|sn89lv@g@u>;E*L95 z^9)di$?DR@JL)28TDlQv;!IsycoJ#s-6ZO1{JJF8dhcqj`~q?P&L?1xGEm${ink16 zohS^N1zNHSA`U<7o&1=3uBlvUlRh-^YoA+NY-?%~d&7s@;NGjQUc6^^D_7R~Nwe;M z{?%Z+@GnOL2mR(UVO9a5-`8C3O$#?<*HtMftfV9uFW!Rta0ipqa}y?eztC#v+pVVS zN5o%krZ#8|e=3nyJW4Mu0-bXkjuSy|a-2{oG61<@JMgNzIN)Knx*i+H#7MA*2M+aB zV*|Uj))zjmVoz7Ffd*&{c8ZXdFwJ$8bzvXqTbfcO2>ALEHPa|6K zeb_+qv<>7gSS`kA$MwE)yTtIp{W>ct09jgTx6G6K3=t3uP`o=9v!LuJZ~DTetdcQ^!IVCFz}3s%aFAc-!p^FCzVS%HqC2!?k)ECOpe`Os65xw1 zYpzA{i6%$!QY5>I8c z$4qzEPi4_m9X}o$E6`^*24`SdTuv>1lH6v(-pOhN%dpcO_EwnG^#>S!x4QCYQybDWZ^2U>bKz|W-=?_9hT-jFb@jPFqwD%C%{vd`MvdBs2M z)?eip{ZkrDt|(4?gtmncK-@1aZg$xpZD}`}o0aN64rNzanE8hlA1ZG;N%ix~)pPMq zqEoSURRo0WzIYwC&6>AxN&GVv$2Xuc?G|KIB_L)#b{{-PcBP-=>dDqOxYJfLP>Shj zCpASdF_0N`orey6!571C2^JS|A;TVtH*{?`UU?T}|L)>2>kvei4eMAM%vb5(FFPVHpk!@65assBi^HM! zCzLN>6jJ!1Sfhl$6&vMcr^b5iKadPdvPZn4s?V<}qcv(v9#oM~BSNza4OP-_Yib#I z7JEwmP$yb*!0Eq*SuqSVSdtZ%y^a-*f2NVIpFN!?>Gi1e{Z(6ZSS#L(7M*e)KXpCP z+h)hivR>y_hI`rvVlt%dxJ>OEcIc=Tyd)$sMFv`CLs40W%)olJoRM>Nr7m46mayR9s96vW z_Mf}YA(9_Zf@zQmF_Fn=>`Nw^DYQk+|c?(l^M}!_Xe!Ue<;}hbAN7`FgS%Hd55ur&9 zpc6;)u;Tp=7MpKHEkpKE?{bQYdiu6&dK?wY`Fvv}!+}uJDIqjl8sNcm;}5f}E31EwifXQM z{Yg(zF`bnGV~e8lPtn}ApY7kHZ*^}UdpI$$rJGgmRlEAgZH2D85~c zIV{Icc!89vvOgj_1wBUY%?z89W>Tn5F`;K>c^iozIEIu=Dm&b;a?X0 zyJMEN;p0lgg}0i%b4iKG~gv+cBfLNq4u} z^^WVbEjP7TRlceU-=_?-4Jbs^-_oap&dum*X%S)a`P(I(6Si@Q@CXlJPhS-0%-h+AlMS~ zIFRx=8`ELwh{i}G=2)4wCm+dj7~Af{YNq1JvF`@d-!3 zt-=}^#pUH^mFSsHX}^$tb&}2a#Y7zLQsv^~Kipz9GWwekqA3-$RM(wB5LR}CegWe( zhDSqn@qVBPs6TGXXf`G2GbTAW9(%(JKZpmC-<-#Dm@XG4Un^0IA-V1FeJ%S!-FwO6 z_O$nlB!LnI;+OFKj!GnLfc>P+K}feC`Q>%G2S&0004apJ?av9{2(G6z+r?9YS6rJ- zk8c!=MaNg4M&U9T>-JN+N)_*piJ%gc}eh}LkvmQFQ;ubRND%zfO8d4 z?muj(?p$?O+U5|P>ocxW1wB4SEQe2_c~WOpMCpBx8>SUw!>S>^TW!DA zvp@RC-sEMc{D(XC@1~ZobZGlIAquk~;BMj5W(H(lqbDt;i0D6vE4d4-!l8z4qBG%x zFc7l`L1P2c8xqr2`(63W|FWK>?!D#)IRf+j$f)I`TiLl;=dyX(xy^&pBfT%C7s{K7 z4)sv)ve5mPRT%yCOJI=`MuIKBsHkYRS*se1wwj2ET3-lAKp0oUGLsK_s0u-MKvzvd z!+qgr)35RI;1@6K{YP1SVo!I(uzQ^~Ojp(X(Y*ENT)WEP?LL?p=)8?=QI9+A=qP*m zu#(Lj>p(;QrgM4gIv1%vgmLgR*YMnEg&#S%57F(xL_Xr;2fT+cnrTFFSP8@a7X*R` zLZCf3JA71zt|}Smrz$bnfTM|iku#k5)PXpQ%pA1TRq)Na;@{&hamHJyNFrtB>g@6X zHuPr>2(Do)s)YnfdOizVE}_^ww;ydnKH`#z44r1(4Ba*%pHpKdCmnz-G=GPkxEbky z7S0UTxlFWQmX-!TjW{7R|LQ6Z6R&Mv|4TA>KBDA6$+BfU78Z%xP1>r$S<=FSVtGU6 zF>VJjQ80S7R6x|KaGmP=`d2&3%cKvvINb{>?NoO}-u)`liT0{bxvFf@5z1Keu(zE zKUts1bN$4bw5-i*RFJ!jwFr>7twFlOf6 zt+qyhM0O*RJFQx!wX*D6hJ?0*m!04GU%!siP=wF}8Mn^{* zirm~FNh79KVGxT0*o+4gB4M`B5w~4XaOYXAf^E7o@tSq}X zY7*bgeMaRR~`SCo~$Fqa1pW+2oXuWdh8mAl%SwjI{k zpMUuho+E-%zZ1?14F5N(nTSdq+?kz6WvS$n0l%05i8)tv z2CJ3S6QW0GiKm;Ji*Pf?{U6_Osb6?oW}q<{1TVb7;;Y0!f*J!e_9J~yo;BpL9O*HEUF1#)Ajh^+DPQ=94AQv|LBn!jYG-j8e2;)zRSuvBO*gB-QrWaMN?=ns7lGQ>U3RxPh@k_Yr|(& z1+upn-F}8Og`afz}6w^l-=`>z;#V>vauz$%B=$~GU=*~@JBZ=GCX_w&|Qy| z!QBP5(x3FWcd|}Z#F^HBzb^1*Z{<<>r*ryaiSedGr$~`Kazi!wx%YqY!B{;0W!x-u zMfQsYJc-kV(wifj?kzuOR{r!_DVOu%;MvpO-G{F$nLgT+n_qShj{X$)na9_AzXI9a zme@I%FwjsV+7XJm_C?xBc#JBLU7+zDotHs|vtYBfn85z@Y|mH#mC8_14X&4cF^<=> z=H}-9QqSg1;aJWASLV4%z;`5j!DLlY$MzHqT`7&crLJT{o%NU%>CdicDX$Nh?tP)B z|J_jenCs2t?oV824u}lh;2s#3i^wl9nVnl7&cA!k-7a=W&Gk@P{+vkaOU>Z-`TAT{ zV}T9Ewhr-I?m!=ldkL>lu}iYI{$;*=t%9G$(uj#WFnB>ieTrR*v!lo?TXt;A(t7(( z-Rtt0mQv22Tvk~H5fd@9BVLja>Ti|7_xy-sdJJL@IVzaR+d*Ft26?u`i$!RJ{r{80 zFYCjPG=}Z*A>XIr4@K-&R?B~voCuNHsk$SzwEb!5F3@20cN=_a;|t;GZ_848SXJ;8 zcFB*t#8_nq+s3|x1#uF_`aqafRTmS2@`8(_qLIKRmKiw*hX*!C{XUK`SVu|)EumP_ zBm5O3iXllnIP(i=VKzj%7$!lyx&CJ|mkgb9Gs#pAz`z&wp6VJJ54c??FJxc&__}cW z1=2*X5^ws3Kwk?;TUaI+&>sntfG{+ugoG+?)}lfpH@u({u+RJ9jM0ekFQrU35nMv9 zls`Q;=V}?Ye`FV@@65DJ!Nr!uYSlT6G@wu>a3@yJB5xaj-QV0~n@v7JG$Gk{fV>4jD*DnAQxfQ_~o*8=d4lBH{eS=(~6C`FK2W{ZV9=zpG6mH7b{sCO2tPHi)S{I!zzwF zeG&+N!D&85@y6}+G}Rjm{bHdT4rnv=-01T_+fQ&!I+eQRMuWj8huc-C&-hv0WJ#gj z`;V~v^yjunn>{}tUNqTqRG6dDIZLP^NeWJC6c$9Bkp;~*(S)E8Roof4S2dgE!cmR<5E%pb!VY)f_`Yofy5dzYKA9iOBuwci#(6hN>DBF+Ap+~|K06!BxgB`DT@KmUiIlK3RLQuLXi?Nu0}wCd$WF~ ze+cFxPE3Cs6DW!&f`wlgO+@tR_rywa0^Cz?OiSMGj!FGnq>_U5!DHshi& zl8EF{A;n^`qD2*TU0an6%4;lPpe8(a%P8q>ldmN?>ICU&A6U#g+0SP**XU6^8@yy-PDEwa)uM51H6~JPhcRAOQvOdt%949sXU%U<(j-6oz!5kr6`Io7yAPu zM|2~8H?zHz^#A=hWeQ^-W@lpxSGOiBOdM8Fpr;(_U98}9{`%}D`t#FFbl|VN602lY zV-BvAh?04o9e#f!p%hzO1_89igtfDkFDv)4YOm+Ia7ClW*9i zreU=~JsI2nQB2y%5*u;!?c$3TZ_^>P_GR~ERwJPn^mw|@4=N-Ns)n9O zDha#)ccQ{r%Yf6*4Simyo#_)`$?aj}zusQdTtI=WR)P)NX98Pl!h2ycC zWv6Xim3@^Va*0+aZH2mBVA4nAv-NppTnvoqy|njBCol?rEaO~eVj}l62eK@r>`DzK zyE1mgn0B&D{C5)YC=#!^h&DB07c`9NMZrS9HVJg(R$hP_76VG64j~&XNkxuQ0wlLEPnKF+zxczW@FCSRJ6hUPv|tmDyG`vSAL`oMcGT^ zC{TjJfMK}s8br3FPySwmxIHbqWwisW=3r! zf3f#&+;{w~Fw~?+j5~IyRcVi3C6(~7eGYAZ2B|?Bd(ks6WH?yidlV@o1>Q(2&e2O( zkVCeO=2${ek)4{V4bhXGihthkFdo8WvUDkrn{y)5-lC%<1 z<_m}nsLFtKk`s!oj;~MCjISNH0G$u*_Y;est8gwn%Puw0?LVEr4iZEJ~vnf@@tHYiHqrfxHR%>wT<~9tgtaH-YX_vn|Nd+ zx%2qS#PoMII|gB#<8>-Cbwxsl0arLl{Ld{%ZHI-SRLJ6UjRFGKE`;y8SO1vq1j5j_ z46T8Eol9t!$L2{h4poG!iVQHCW(C(L$68N#i$`y!rl)&*=`OfoOu$w0(Uu?oeC5es zC;NLcyYEQP#FbcCL;`AbdfV~XI9tS?gmp`S#h1O7vX#|())H5LGd?K!{g!U8&Bt%b zjXmDD{dr$1Z6}D(pjnEFgIw^b=3fiScV1R@Ih?x1+}ZbJ^WND=A=YG=umP4>TtAkU zp(42=iy56?p$pa$gT1O~nXdXW|3m`1Kr_OaSj1 z+O{OA{!k<_k7D3Si|v9b;wCZ5d7o>IxUNcYZ{YW&r#C9Tx87ib7wsc=E$%)RC2SS* zU!sy>qp=8h#~D@ztr*+LS5&-X@x5psPZU8;;0kksa(n@C?+iZ?vpiX@hvO-1v37bfKF4CMNl&va>HYKd?v{ARj`_s* zr}W0)1oy6w(p|07zOZERHGeMqX8!(uAA9wIWAhX6O+Kk%J2$o-lGvYb^s&417oU62 zjElbnejeDRb?SLAuUv{oCXt>gA`<T ztbTDm0*U95Ik9U!>vsvwYB5p{8GP*;z=tuJ#wDUB_w#en=Z3Qk#9vuoKL%tTcRAB( z^Z4gn5nI`P%Tl>(rM-OBEFI)A7z&JQz&3}1?B50_|JeV~@W5Qw*Xfx&%o*zF>T((l zICJIVl7y?fs`X6UQVGA>PI{*30Dt>@6m}L>Gt7T!?^O}EvZcb7nrS`(a0pj7MB!nN^*8izW%WN zdh8EVbpU$tty$krn(fce`UFTS*X51F6#Y%foe;-c@~Y(_;<&e-lHdQtujs)xWy4JI zM_LnSeV*$)dcc5=4+rfS$+F@N`l5 zWb`Mil8X&o_6mU^FZ^h?0>oSbj)~Y0e3Ef_KAFvUnlmuhF6XSgwwS%vHvH;A+OuCA zi#cNUdM>*C&!53!->aeNmei2VMM1nk;m->iMKlt1_}m2T+6KB}cV3!5!_93PwVHBB zl}4tdhy*x@ot0~Hv3wQQ(QdAKx@1~G7T>IG?c()&85uX4C%kJ)VzZ+1nWyHDpwW=nYFPeW0-d()L@qd%nC!$OF4I_@uqKiky z=}4xyLE}$Iy##k5BTEPKM-_my$ch&Hrq=Ch9Wit2%f2isVVj|=&$v2&YDK&O`TF-# zVSh;xOb8F>L|f~)-+^(B=utwzhQ1O)b~M8VATf@%CN6%kfPp9uodsx|%d4x^?ChS? z`Xp2y2%8^PsSUL0xy^<$h0m6{npOz?J|mi3G??q=q5ti%Zqd_cXm-d4m?$&Q=6w9S zh@BSgSn9$3g!cw$W{a(|p!@ebhxeg-#`Jx3WW)yI4~#D;cxO!{<9I3CzR1A97!1f( zGB{{8d$VwAUZ;74vE#}E$%VJ2im$ZP+b>xJH?%&6WE*pSRe{w&7%5mMW{|lf`Z3R< z?SGKn1 zy>s*_RT|jKd2s$j;n0&f$$mnd)LUgW#BJqzRQYNLyMY-2Z6Mk zyjHPalrAE}qhKN3Wlhk?^RGOP^r4d)2a5H+a|=oNSp#ia%B}AkXVMFj#-xgZcvcTu zr*APiAt_Ao4+#>%Moy3V2?zyqjQqB_R_3u?xB?7y zi5C$wB>WgvBPaM^-u}m@oh~r;qQ>C=_>PHkO?Wx*A)gqn*)1Z-tHNMF{%umET+mx$lg^-^Yi+KzFWWHEX;!kah@fUx} zSxgJ?CrgtbNB%muv`#DtY7q$Q)lX(tR+k6wT`7vnK4h z&RmUL2)wE^!kl2Zt7<={vXGZWozAJrGmr8puc-_R8ot4GD_RVUU>q1uLyy7dh#Zn`+Ug{fBSwD&dZ47hX63i&VpSfYFhpDxz+vNLe?3XpX%ucuzMR@mpbRKF0L-=ZqxAk6tD*Ne(`6Qk?Hw={er*DrDdR9FaDh7*r*ZH zgg#wofB(u@rInjDZF*Bywo!^F0DnOb`SjP5u}bVxyno(PyyEqzCHMO^Uzl^nor&}X z6Z%z6%^Hj|UJc`kTaXs#ynM+nwLjqOxpS-59@f=8`1C2mJuct3kE8@YE&cn#)<2)i zy~^~wzc1jw3urw%Kn1acg7J!PVcqZxa?K0y60_xo@9oNw(UHwBVyi#r?VqwA*-T;1 z@CpsRmsujfyNPmMtNG7&OOo7^aZ2vOh0VU)2kxMeY9c@Mxkwu9@2ClJ;3XJr@nLoW zfH}n-k<(bO+HPvZpe-lDBj&EMyH2*6YBz&9^Wi0QBTwaq)05tP>bla?&MLv2Z24Sb z<8k{m4k=FIJ|Bx{#U-5ED_&K0Tcf|78dwmX-rhq7rO5s#5qehQ7rjNKSh5VGkV1m)p;kgpz{YfRlpZkH2qmZ@{x+X=5X1w_&0HXCcLY<6Ye6xSQGYuS!?2_S+8mPmcIjdwY3Z z#00Rgg=7`MwtUC+2!RRXYv=GaKoMOEiwDmeH;!gEj%A#$UR4&(?z?X{r?p{D*!$31 z|M2nA1HDR({>E*N|1E{^>MXQ#jZRzt{iPC(PP>+47sjGHykqa)rhTn~aHQ)?5fV2s zG11W0Hj%f!dNt`|4kst4#`0Bf{`CoDoy3E7bKgtdXJ}}61&eVH*yBd8OaFEicEaOXR_yKf{U@Tc~}bKqNh-V6W7fA-H~Z}ytvEnKj>DF*Q94nX=o1BleWP2mKF|(_9R32m6~z? zc3%FEi*c41cHVWUpuV%fA_fl$8XerD&F!0v=DX(fumIf zyYK}kCnv9l`1$zU@bg;}D9Z~|&)Wn<_1z-Q<<(;fmA?bb**}7VA|WABjBg2h?v#=% z@;9PeLt8*Hz{Oujj#h8(zU|$IaCjM>2Z7Vl-rhSqTO7cPC&p(rp3QoD`AQ)W&Kr1K zOBx$lp^SVbCt7e5+g|ccWPI3(MG3|%*Ktr*&UWC zV_ugE()Z)$SRK(n1{9YCPC1%5W_h-4tAHQ5xv3``zb_)JE^IurvZTi$})hK z|DutR0F2oay5HR^`w&jB(SuVx>oIpRV-_E^j)OxmI5_y~k58GBzHJVxszS9e*BLIg zN*JXf3VnCl!}IdTRIl!_`KH(pZ6!uem&eM5@Bk)-kVnt^>_)%nDGc{y+r4}D6kw&^ zFQq=~Hf<6Gm=TCceoaaKTFCcI|IXSt}MqOPU6H*3T3`5BA%5eGJ6jAurM!=zn+ zYI$?14_hV>H~t`5)0V82_y977!3x$4^z$7XRLj+&Y05OL;#f{k-;I6FFCy|5D&2f}8fv9X7mDUGF+S zb@`^Z_a(S`LKh=MACkMaa@}Z2H$>2~^mkJ6`1cA8!56FY}QV zqr+V0{{(!@xN!6VTs}B6Psz#E_2(xONN{L`m6eqkcBGaX3no!_XG1eaj^;^_@P^@0 zQE&KQTj_^z@HskR?^X;#)Gv4^EX$EPzkV&)XXjz*ZV5^21 zfsf&Rs}uKrcw{80;laH`!71!hZfuPms6Hvh7*e-XN=gaz88xA1Ys;}ZIvNlZWRT*g*te>xxq$mE-+K12XLvG=;2CrmMK5CYR;24+ zdV6`{R`WFT;E_kW#PpKYQfcq;+%)uDj^$yZPQexofTi1p?_29%;`jHt&gO7RyPl)G z0~xXhRYllIf9r*X&h!7dx&LcOGw-jB{W3TxqNb)sr0chReCpnBgh~G7%#4Zfri~l7 z!>l3!cj>mbxA!$XR<@j(zu2j}-Q3)kM2+r-3rp^rwp**(t7V-x&}VP)>ihAd`Pm4z z*{yr`?uo&E02u9N6koQS*}pb~rIXMGZ*zP4_l#GnO5w(wF z-On2s(0LxY4sOpSZ`CJgj?m+8xbcwq`G<$E!e20MrD4G2)~#=^?B}L!_aidGrkNcl zX_O-WT`p>{yE`Sor@VeYzxcuZ`)99n%B>6Cf^>i2L9Rx~#3j=;(@`c}uRV$MsPpSr zFU@C$n}vi{!Oz@4M6c1_@bYDEj8RyQ#Q$AYRchXvVfV!U)!vncQ`xrb#}JZCna9$g zNFqwBM5W1)%vu>j2$eZmhE#+!@=9b;3Lzq`qEJaHS~6sK-%Jf;iZUc5`@Fy7+xzd{ z`^WybzvKA6qxbzXtmnD!>pHLVIX2Gm}P}Xh5TSOjN_Yod-xp z^LR|)F`Ttuu;Ub#l8V8pl>{p2Su=NIgS7FFXU%6mZS{B{qRY4Xygb$L0s<846A(*Y zkl&dI4-$!o47kXsC1>*z=!p3rKqZfOpKg$t(sbpiR>J38>bIT8(B z_f`Ds`wuiP`1$i^;iE^pKqY-AEA-g?fIL2^g4phPb+cu6^YgV$@=#JUr!eLG41;4I zhlT_)&0|4qEWv8%5wrb)tFc2&N-un~ot@o57nd`DEQ!HyUH}Gt$KKbQ_(zlxRoJ`v z9q((v-a6CV30^OxdJKVM|mA z)D0ZCLtme=zH57v)s9j6wV38dDf%4$S3yAmdDCPkG|Mjd1Gu5+&RQ!()grhQ1$_HQ z{P^+XcWmw{z~tx*2+6_KA%iVyWf+t1tG~&GwVWUO`eN`TK4vC4RHA^%OXXF1tcU%G z^C%yfi?FpO!l8DV2WY(=|8;V6%d#?3vDmavXT=!ZFnd-U4|IIJ`ECG?p?cfwY9Z9`|v?l@#Jp} zD67XS-1jHFnuDsQul1fHd7gN|P#8HOd;nTCJoydcM?_345?4e&$FlOp3oiAt@29gm zL*S%Tu?h62=Bo`>(h2QP7GlA}i>4-02I_C!daCn8$ao)zs;cVk%*=kKqcloNz6A@m z1C%(qxNs>eD+@@i)px(k@mswW65n6&`q%ng53h%VII@h^bYQo+nWuAS;^Gz@IeN4K zl+P&)={a1i4{6X#s30Wvo5ST7jR&EO=da0TQ z*ub0u7`=tyeQm81rx^b{$nP{TY7U9MD1aX)#nwLt!w4z@sD(?vbn+=315OE;Bj>pSMKB2uh9TJ=cP0_p~ch*{k0YltsdUDOxx|-w`%~|Kk#hkhk4Cb zxba?&k3Wt6kio89x8OhM8on3h5|aOCcv)|Eb6X*!b(rcNR$Q;d*?nwjU(qXtrm_`l z)(8?Pn3w03JeWW9^J;49dAyno#CQ@{NI8Ludm;82=)g{>bi^zT578Oh#C-hxDzSaN z9-04kcJ}A0*hNqjJYc)Tz60Er5c~U>URv6lqqBlikDnL%Dba z^p6}_O(rowY;)|`v7_yApz7I$vS;Cb)r}Xz>oYm2vvMULi1+VfV=v$WdeF`7ED|+2 zi!QGSQKgnEw?L%hWQ}M1Avd>5RL;1se1d{IK|A5WQkE6Rt{%tO`rGRpgdn@E#?kZz zU;7)utnHDvB8<|_duG@A1nIRwMl44Q;KK`g3CycK;f-Jb*J^nzIpXLZ<|st@-Hq3= z3+v-GVA@92d$jzOwPmHKq+~SWn&&`8LPEk`IiPSISo*C!^ja7=%VSF`Ebxjj#zL|Q zTyC}jqvFts$jF?7ANyiW&HLu*6U?B_=VifT_0aX6m)j^RC@4rGlUrSCEH=FUMw^9R z`z}R$0X)egii_7lOD!lVDQQ>E5QkCiS?~#nE0H*YF4)L`AiRuhDVnm&{8v=y`*1cj zHT63zq6uLTbN+l4$lnJJ{?p?gavS})@jFDu#TjM#`Q1aMn1Lgse(YuKJ$_w%@SILASt)V7MSP2@!bM}ZRHg7xZFoKKYB7MXU! z4Z4vr1MiUwxg&744J#`v1HuIrHm_7t!e!^w9~v6^g7-)p#083f`_`D*j}_3)_;LcS zTgLHq{8T^2iO8N6c5Q8K6Fmvi_rD&6P>YfB=}mCMRi~Cg+w3(B2OjQ-6T!iVJh5)m zFzcNcae~39fu3BOWWbki-t0knrwmZyT7Bao013xpT$UtwU$MvWKOXQFr7q*b<+qkR zc@p>JNl4V7b<-ph9ugs=aJV)E)-=emtbv>D1BV1AQw@N^`RGwKWz)3x#x^;Ynh;WQ zEGBkrEnXBK6%0tu@?$x_FGyX+dS)xBC$HxoZk!mo7V|nrWGQzeO#?z@rBlSJLzC|sgXIt3F5A5lG;e-O@RF1`HbKPtCN^Aw^ zMQn!c1AXRsg0ks37K;VNzas?c+D)~&$~|V6@?Z}!UKbP;BuO=5n1JaUAF~_V(Dtsu z!*Tmv3@#x*c-r#=@)LK2fSk_0JPGin$yiHBU0x`={&1Fe$73rh2G*~8^H-^;#KV4) zJ=d&V5YSV-vxh4r~Jp7?a))P2Z+nZYnP3bA}emLKN#!&$cxNZsT8 zohxwA@yg2c?|?@~DU5pE80O~jD8zT8xyov}rJz>^`vH2Bk@p}@M(*0Z+7l0Z`ubMl z*`@TOuxf-^rrAod6xv>c<`+B?X7Z}DyrmWC@3zbs8~m{m=t7IkF~Xtg557>^UW+R zmw}NW#U=XzW}_BR!DJmwOL{`WQk-putZiCaTB4`GfInKKOb*8<$G;sy2GGR%FEf?I z6b{f9;j5@aq%r22*Ys=@8;3jx64N#~SH&2PCY}l}YnUa60VnhR-Exfi&$EV>8$1yB7eZ~`GR9kaB2NoAC7a$d0)xZUM5NCq35!ba0a8H8 zZuZd+YsOdlJ0O!vs5_iT&Ny`F(2bH3+jZ~o1ja@{xh(}Vg|vM(DamGnvvujJRr4`o z$j8T5KNC<%B5->tK6+~X<0=7)?7J%t1>MHcUY=T@L=3jdqAMon4Lkg?S~NUc*Vx!t z=>AT@m`M}Il;+ND+c=^4w{WsU8&V%<+Gn_ZgsxCnvtLD5`X{U66%_cuhRNs?J@A-79^{(C<>l#l{iZpfdiJz z6#M;f^xlf59_s2_lND|!=VmuVJPG_k3DFkSlIW?1L7rBEsiI`)Il$=ywV?E0Q|ITbK~5=D0nWfMh5qD zvC>-(O7H+}Oio@tFe_(d%&LWLJQC&&$Xb3z7;N&jZO^f;iiO?7?%-f|o;aaH-V5w17g6<~e~nSR`uDy?H@1(%`tLK2`hQPk z=LQyMFf5ws2bdZLCjP#@tH#otF@0Kz-?ARrf~|WSoa&Xt#3-_zg?hQdgaYd(a!T*x0@V?(P6Kj*l1LVG7%MAHGJhMe}DhGwndzvQi1nF6J4xY zU&5^2W$w?c>lz!Hsb`|J>K%WxmS4ANlYV*mMf@VOLbA37D+=zOvc&HXVguDmuD^9N zm_OE^du#5jTD6KLdTTnXa3GuAVWewcZfElJ5qhc6mDF|9P~@m2=2~|?(wf-sM~vRj zEx1#jdgp1={j~45M#9w8)HJ|8zfIk^2sPy{J3BcT%SM5J=0J{3q8_2(a2Xib|2wMm1isw6aCEPl7-yhB5n~ZOv zN}4HHHFE=UDH=NuJM%sxT+y4roiw1IW%oE>kN-em@O!S}8U;0Tz0v8{SC%pqU&={J z^4|EQEN%S~=KcH>L($dl`pAKl>BBb-GjlEN@P0v)aUyZ%1m12E9>^9L2uTgV#9CC6 z3Frec6zL?ym6eqhf>koq?cEQ{w;iV%B^v3NG(!be+z62`K}jcRfwA?He_#lj{u zSzAH9kIB_z9^bb)MHo{24)PD-a?@GmAV?x8SfFOE0a_u-K`_=n7~!@YCA zzv|I0#GgLR2~J_~_YYr`9mwy`8XNDMC#3}rwTY}*v!?BV$0E|o%QyOpngbXyk|0+9 zQYTlb$K}npK?-*xO4FK}-DN<2Q6)3HhW+1+OlD`Y9Rw*z`$$EfuO|lUbF|z9n5_rU zxx(_7lF%v5PR5{GxyfDNRf|RMQ2$(pAwZ5!Ow4$WXUFEC;nZHuU>5cRFV@wS$HAjd zcD}{+N9mdQP7B+CSA$>^M(Jdg{pa_X>0LUDA0Jw3OX!JAy<1zFbF?AQpaO(sm=M#s zw_BuP=8UuC!>>0mCr?gKNjV37l;@v0IR0@^q^Q>3%N6nIlXJet!Qk|J5382}(n!$) zR)SMMi?M78j0ji5DnP)&)PwvHB&#og=xWC ziL?t2IL;tfZ=JV@*)0pc`mrgYFtJASId0!MSSQ{rDUm`)O;{uV6Lf%hQb3sjO4_;5%+EAhwgz+yMhcE7gK z-;MPN-LqBTvc6*};U~LB#*g>g#=H()| zVwm5LTqqeKicu7s2nA3`I9YKwvW zaR*3O029=zp+_$%=#pmkUB{;iL^DHp)1DgfBb{Z^pvwxzmv@7td-i0cA`Tiu3Q?vS zBwY9K@I~-02wn>4+!JWkP!SeFJtKVvtb-)fHJZwmSUb89Ox5k%w+{k}pMWBGEqbg} zkXJv_kOs!|NxCj9EZX@~@IRpS!E=;lD9Tw%<3y|n@md5snp9dKKOvz|^q&Qp;{v|24K-HFwdT%F z1%&8V=vn;$`|!NBmNb#2*0~w86nnQB@MtzR3B@YV2X_77nKv)-iQSG%A=WvTqu?T8 zq%FTA$I=i#&J{Rx6nLvIETDK=8~(7D0g+=t%(VUU&pFJ~8>eXP8N5;lK9dx_zW)A; z(A8k>;^h)XeZ!o{&d#m{fAzAX!yed;sH-&571N|LwsUm4c64;$>&hDktxvj7{=s240<++Xr{}uGje*0g34+eo&rcL-0y>`y zWoTK)V~u<9k^RAH>b($8*Wx8rb#^)sS_^Z%f8k#ZetA?_xR0oH`n#S!C4FAB#Ht`^ z1Ec2(ZA@ZaRLpCP1}~?`zjSv%c$gPKlt1nc`!C&G;~kRM*U;S4GvqLkN~9|5j^b-H z8RLgNnsXD$6M@x0K!4*7vL0kY{##qob_Xm1^uS`J_jxv^03(^;nl-UE?hRHS1mA#2PM;#44MP@5_5zQFK3W9n77n-555;U)=-qo~o7>YY?Sh z&?{(`(Yuo~cmkwN<|lEHH*9tyVyX=pYe!cduwxl{YM}g>uBy2UUH6wWSP($RftI4` z|53EjS`YO`0(2NgFc`&gIfsw3+WxWS8i%qw8?N53c2-hCpwR?9OxPk+6XZO-gt^C- zU)@~gQ|Hg0{|kFVd;19y~vFRMgN}87qEmTRX7w$NCG>+BCNDbKA28zA0%}M$YL^Xv) zMO(4M;vm}bP{Ly|j*>FtFtL&tDe82VcJeZUOF?Z$*Mn}NYnT1*lSR*?cnME09R!u(EtDd diff --git a/docs/pretrained_spatiotemporal.md b/docs/pretrained_spatiotemporal.md index 7521b17..fad1628 100644 --- a/docs/pretrained_spatiotemporal.md +++ b/docs/pretrained_spatiotemporal.md @@ -794,7 +794,7 @@ This design ensures that the sampling process respects the temporal dependencies --- -## Chapter 4: Experiments +## Chapter 4: Usage ### Overview @@ -1006,4 +1006,73 @@ model_wrapper: recenter_on_init: true ``` -This experimental framework enables comprehensive evaluation of KALA-JAMUN's temporal conditioning capabilities across different data splitting strategies and conformational scenarios. \ No newline at end of file +This experimental framework enables comprehensive evaluation of KALA-JAMUN's temporal conditioning capabilities across different data splitting strategies and conformational scenarios. + +### 4.4 Experiments + +This section describes key experiments designed to evaluate KALA-JAMUN's performance and validate design choices for temporal conditioning. + +#### 4.4.1 Model Comparison + +**Objective**: Compare different conditioning strategies and temporal graph topologies to establish the effectiveness of spatiotemporal conditioning. + +**Models Compared**: +1. **Standard JAMUN**: Baseline unconditional denoiser without temporal information +2. **Position Conditioner**: Simple conditioning using current positions only +3. **Spatiotemporal Conditioner (Fan Graph)**: Full spatiotemporal model with fan temporal graph topology +4. **Spatiotemporal Conditioner (Hub-and-Spoke)**: Full spatiotemporal model with hub-and-spoke temporal graph topology + +For instance, check out this wandb [run](https://genentech.wandb.io/sule-shashank/jamun/runs/scxc4bt4/overview) and its associated group. + + +#### 4.4.2 Noise Check (Multimeasurement Validation) + +**Objective**: Validate the multimeasurement approach by comparing standard JAMUN with reduced noise against spatiotemporal models using repeated position datasets. + +**Experimental Setup**: + +**Standard JAMUN Configuration**: +- Noise level: `Ļƒ/āˆšT` (reduced noise to account for T measurements) +- Dataset: Standard molecular trajectory data +- Model: Unconditional denoiser + +**Spatiotemporal Model Configuration**: +- **Repeated Position Dataset**: `total_lag_time = T` with repeated copies of current state +- **Standard Temporal Dataset**: `total_lag_time = T` with historical trajectory states +- Noise level: Standard `Ļƒ` +- Model: Spatiotemporal conditioner + +**Experimental Script**: [`scripts/slurm/train_noise_check.sh`](scripts/slurm/train_noise_check.sh) + +**Key Comparisons**: +1. **Standard JAMUN (Ļƒ/āˆšT)** vs **Spatiotemporal + Repeated Dataset (Ļƒ)** +2. **Standard JAMUN (Ļƒ/āˆšT)** vs **Spatiotemporal + Temporal Dataset (Ļƒ)** +3. **Repeated Dataset** vs **Temporal Dataset** (both with spatiotemporal conditioning) + +**Sample wandb run** +Run [here](https://genentech.wandb.io/sule-shashank/jamun/runs/4j8bfj5k/overview) and check out its associated group. + +#### 4.4.3 Total Lag Time vs Lag Subsample Rate Experiment + +**Objective**: Systematically evaluate the impact of temporal parameters (`total_lag_time` and `lag_subsample_rate`) across different temporal graph topologies. + +**Parameter Space**: +- **Total Lag Time**: Number of historical states included (e.g., 2, 4, 6, 8, 10) +- **Lag Subsample Rate**: Temporal spacing between consecutive states (e.g., 5, 10, 20, 50 timesteps) +- **Graph Types**: Fan, Hub-and-Spoke, Complete graph topologies + +**Experimental Design**: +- Grid search across parameter combinations +- Fixed computational budget per configuration +- Consistent evaluation metrics across all runs + +**Experimental Script**: [`scripts/slurm/train_graph_type_comparison.sh`](scripts/slurm/train_graph_type_comparison.sh) + +**Wandb runs** + +Run [here](https://genentech.wandb.io/sule-shashank/jamun/runs/tjwcsf4g/overview) and check out its associated group +#### 4.4.4 Sampling runs + +1. **Bond degradation** The bond degradation of KALA-JAMUN vs Standard JAMUN with a trajectory of 50K steps was compared. We also compared KALA-JAMUN to a standard JAMUN trained for 500 epochs. The run for KALA jamun is [here](https://genentech.wandb.io/sule-shashank/jamun/runs/1j4us3nx?nw=nwusersuleshashank). The run for Standard JAMUN is [here](https://genentech.wandb.io/sule-shashank/jamun/runs/vigqbemt/overview) and the run for the highly trained standard JAMUN is [here](https://genentech.wandb.io/sule-shashank/jamun/runs/9u4qo5ax/overview). + +2. **Comparing ensembles**: We compared KALA JAMUN vs JAMUN in terms of being able to converge the distribution from the short swarm data (1ps). The results for KALA-JAMUN are [here](https://genentech.wandb.io/sule-shashank/jamun/runs/jwk7i45j/overview) and standard JAMUN are [here](https://genentech.wandb.io/sule-shashank/jamun/runs/u2of58jn/overview). \ No newline at end of file diff --git a/graph_type_comparison_3d_histogram.csv b/graph_type_comparison_3d_histogram.csv deleted file mode 100644 index fe32ff9..0000000 --- a/graph_type_comparison_3d_histogram.csv +++ /dev/null @@ -1,17 +0,0 @@ -run_name,run_path,lag_subsample_rate,total_lag_time,validation_scaled_rmse,model_target,data_target -graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_2_enhanced_sampling_data,sule-shashank/jamun/gxomxwtq,1,2,0.27083016000688076,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_8_enhanced_sampling_data,sule-shashank/jamun/g1512wlf,4,8,0.3578928094357252,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_6_enhanced_sampling_data,sule-shashank/jamun/3dz5xmax,4,6,0.30962056666612625,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_8_enhanced_sampling_data,sule-shashank/jamun/9kb3i03b,3,8,0.3430485036224127,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_6_enhanced_sampling_data,sule-shashank/jamun/2wp82cfo,3,6,0.33505777828395367,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_2_enhanced_sampling_data,sule-shashank/jamun/6g919rvm,3,2,0.26015421841293573,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_3_time_4_enhanced_sampling_data,sule-shashank/jamun/deipiaj1,3,4,0.27313617803156376,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_2_enhanced_sampling_data,sule-shashank/jamun/gz30rece,4,2,0.26252674777060747,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_4_enhanced_sampling_data,sule-shashank/jamun/c2fvvmrw,1,4,0.2634944459423423,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_8_enhanced_sampling_data,sule-shashank/jamun/ecq1lzck,2,8,0.3460829760879278,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_4_enhanced_sampling_data,sule-shashank/jamun/h1nxa4ed,2,4,0.2966764625161886,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_4_time_4_enhanced_sampling_data,sule-shashank/jamun/y05s6wsy,4,4,0.28305633924901485,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_6_enhanced_sampling_data,sule-shashank/jamun/jmaz97j4,2,6,0.2979964315891266,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_2_time_2_enhanced_sampling_data,sule-shashank/jamun/6bv6r9ct,2,2,0.26125847920775414,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_6_enhanced_sampling_data,sule-shashank/jamun/rac4bws1,1,6,0.2733909171074629,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory -graph_comparison_spatiotemporal_default_fan_temporal_lag_1_time_8_enhanced_sampling_data,sule-shashank/jamun/tjwcsf4g,1,8,0.3070009406656027,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_datasets_from_directory diff --git a/graph_type_comparison_3d_histogram.png b/graph_type_comparison_3d_histogram.png deleted file mode 100644 index 3042c5d853f398a46061fa1472398caea59b771b..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 646402 zcmeFZcRZK<`!}v#Dv8iSWY36(j55m1-YY~_8IeuNR`zNjD5r0+!M}tYq_iAVZOj~;jO!szTGU1TEjXLD;lmJe|?habuaw0&`#raH8jG>B1F z^-wrVgIWL3Oec4smwWMsH0_SKs= zu~tVsJUk8`KK%CWTawS8Kffw0lp1b~R<82&5U~D2{_)9K|NHmvx93Is1yZsak9>-! z80Ix7zxYz8P_2|FkBLH4Q*+T)M$+`t)VH#O)YQ+XPLV!1c9u-g>a&{DVWyLp$Zcna zq$c`GDBs-rChqK9EZ=nG@Zo1Nku2st`OG?nmcBgtr5Wpb3aP)(*z`Y0O=T*HY3uIx zfAr``^T&@y78Ze*A0AD@*D+Hh`up!Hb@;9FQ@CneFeX z>h0@G175N(SvfdJzrL}ldvyB7pFe{VRBd)u~cCEZw% zmns)FNxhXg#u>UIKY!4y6}|YluBGKb`qlUdt0OmW-i(Wjds|(-_1CXo8Tw^sj?>a6 zrKG5eEB{x>D&GII=ql&2T@ItITP!RrNV{idhvnAiTEvz|61`eeWhmEY>gY$mBGGmRD5F;3>9}96x@%wam$8eZE7nwYyu>u~_ox2>w7rMm*u+!~MMl7Wer~ zKD_<-Q6cH)MXAZY64l{7kBoV>GF7f$XZqEW7<%$b$ceLOHK&G8GBT1MJa`XBXe-Hs z2M=1)Aug!>$g^3tqixc$_`t>P37#}(7f`H0>{{!?xUu~+m)ihq4DG;58eelN* z4iX1k?UR)JDJdy@j$>+1Mcg#f7IyB~!NA708=H?76MmfdzQ3A}JUxGVr-p`xR+06Q zU(JuU-Ml$TVfZF7k)BS}JtM!ycYWP`qUZG$X2-yu3b(UnEeYICEvZtt`fqz)-%x&Y zOCLWo{Iay7qJMfitFVYj9*(JeybzUDcaCp(czCJA$#%A?jm0N5O?-uxUH|pA@CUMg zr%s|qgp=;}zRRs!bZvgLt-hf_+38%eii*k~oT3zi3RkTP*TUxauPF~6RPLdnrQJ_M zLlezvLJ@As#Kd&1zuX0<>tu3rvaD%*)GAp#%f{X*jj6D)!(3*A9XJTjzk7X zcbHzk{vG$@%4@Tawwr%F9Q`H!tscVxzZ`g_@ZZBRF)`FQ%&jJdj0zSbA=~r6i;5iKJB;f2;;)s z8=HZl$**q$9zK-9`PsU4D~^ov_4r4R9%+@>Gl$S#@H43kNxC!H%Mloj-4{@`wB#st z9QO+gBX4WrRLi(4>aqGvF;UdKub9uWD{F_Us_N+Y_(=waEqnItaYiwDr!?H1Yp7{q z^Fwmp3%BDw7HROq!(+^%YLopHZh4*?uF>;m+3CXqyf!yz=0;j#S6wz%7x>(l zZ3`^BNUvVK`YJ!)8%2iy^y!22^r2tAXbHP7lZ%Rq7JICnS>M=@W4gL`Z!qHzCPo!C zY3bdiPSYVz&e~KrG{}4##0SDJ<@0aNlRB(E?!en@c`~X zhF%H3p1ywGwRcHfz3UZk|!Iu3KdN?n|CY+jE6uX+NynQ}@Jh>#}t7+u7OK zIyu$-?nt+rA6347{d&NsrO~#N#1m`g=-qq$`gMDrr|kAiN=~~aCI6s>IrpGU1T8*T z#Jzub-114(+~k{qA^d%cVKkIw6rw!W1xvMTEoBF7U0vNTy}d-WBp&vz6R(e6vF(;C zA;;h4nbd_ndZe8^Lm9mTb-Si1mN(5a0Xtq(BU$b|v(>=B;7g$ujk|}3)QHFWG9T93 zzaw4og}?{gKWbMO&NFAIM_Ll=?rzx@m30;!4n>L0UOy-xD5zS~(1UqZ`&+4Fp3(cA z^V0){^S7~~agQG-qavGkWl`bu?EJ28+%fpxKhJB^Gn(bb<0ns=zTG;RJZpn87l%$f zdc*s~33m29e0+Sxw@1i`BlP-4OHjWJ(H#>-J%}=|rL7%p+xhR2R_nK`w?{ivDZw{q z#KYmdzQth;mj4j-+-+%!*y5vStl4I?Jv~J$H~+e+`(%Zow;H1i6CafEiXZcX^P>}($X(wPP|3dJ&b}!R3DR4aa{ZJ{oR+> zHx5+4f1hXe@u*skp0+vzWwabxKG9dd7Tx4?n$iS{n0F0rR`)uY>cyJO0efT*BH*jT|^|psu&jO76Vuzo3RA z)cNcK+&;BWt*zu}qO#p4R#w3kZcE}oJq$2g(k*%q(?>MJM_T8!~F(Jn> zCLu+4cVWN%$5;wk6$rplc4xZ(;nSzb17}ffj;34*)hP9}va)iY`*_xQWtJJBC@Lp( zRF&WQ%PGA|_am1tUrtWU)qZ_BZTz;QWAx~b6+EQ0w6s!|LEc~^t9;xuwfG}Pj+}EC z4U;S>EEL36KaUvfD?@oWedE)2e|k@)QXWP|b)vTbZ}_7f276Yz%$Z3_Nk!YM{r}s4 ziXoDyERRu>T$iS}l@p?)eRbd5%4A@)6SVAn8_lVC3WcDpB~jSL8z?CDpTDWrelXtQ6DQd{k3mA4tFq!rs-Km9q^F!ZJFo1w9;7qUy_Qj)1M{QUgiM@Cqc`(NKs z%oLeDsQi0vVDm5Q<3n+gk@v^Db6+^<)3aXR+PGiWfTF@=>)qJcSo4+d@9szEJP8m` zbZ`*(lpqo;_IG(3{yFKnHhGqKh68lMK)X4m?>QY69t+%i^5n_rjaXhoo?&yWIDTc6 zx2Zdk5F1gOX;4vWzJjVLdp=_5_xQNU($qH;^*l+N3q3tO%fHhTUpbCz2udyeCf&ck zHd8hI`}lb9l~6hc9v(6RS2s6H`jS#zv$o#H$;tWU>(}>yz8@b8N&xG8A05@YZgbtp z$mk0y?z@4?%1VkkCRWxa;p1i<&lUKty(KAmGhG`@bIxU!=~jPv;M=zs@kl4Qx%UeR z37rxYl)^V^E4DK!E-f`~eRk;GhObIMC73(mG6hIf$O={p5*m z(zH`gzUiN(C0>Ee4CRzlmYr)YXFwqG!-oUFAAbJ)`L?~Cf{cvJ*3q#R2vIrxD($1Orj{0eK;sa4F*0C& zVXwagD>)?~;1eHDOZ4+!zx*!U-)ZMOOmXPw(IDa;eSWFCd*8kqoPqF|n4Oo%uUT3K z;rUbSeta11D>;oucFoi@t$pDJFauFVW@o)~b9t+)tHUB9yz$(lV`J6z^}AVESg;H` zCnhJ`UKvRg-26%nR^S~NC=>2z(_elB^hN5^YqzCaedR8y&$VLF;>OJyxP^pt3f6xx z1Pivk_dj5Ve%@E+bf6_cg!7!!&ykV$ot=jOL^2GkMDgrxNfNspCVQ!Xz)5p?#l+~b z(SPRWYp_AYnrP-45QGZA`xHOFcXTvWc6K(7#%`=4HddBaMXjT+!i^@DN8dX(mL@^O z&2MznJT*OCLQ|85nwlEy=Sna&H_^25DR-d^GE-c^xnGpcs%lARuM% zCjqO^`!?2=u2gw$Jj=~BWaKkv1cEQH>L#0-nxaVDv13P)VU;I`u4cTDV=KCHErntjXE)qp1-!-C+4=k5jWwJfNn6`fLqC2HdguMZ7biXd7*_Z89>r4Be2TxIoO1c0@%Wzo z`+dvGF08Mw6Z_fPs`MX6#w{fP8d`wm=j_0%@j~xvYj@yuadgr3zs}FM9qVKVoTm<@ z6A{G!Frkh)O%A7HZeAVy$3AcH|z zZKE%*jL=A20&i-fjPn>)?sIZ->Z|giPfAKcPr7Jk#G6>nh~17qp>G~Rx#o?{ehbZLUHz1-`gK`p#&vNO-(&< z;leTWzYLW$Qi71>nOrF@DM`X}*yl8dO+6*KXxv+Dm(uJAY$sk20)7@H-HaYjuAd055*8Py#W>E@ z=(C4X@`uaJkZ)r*a}ga_e%d&&Y3v1REK(8*o;Pl(d3>$HnKNgq9~?RD13)AD(lFdL zO=0#g9j-q+#0LEwM?m+zkt_--zc)n%1=Zt(UGn<|ZUfi(0x;5a{umqM;A;VAF%3y6 z9#AkgPA1yryZZW!`eHOvCP&)5yu50FV;`{O-MFCko1qKy^FDw#doEqNMC#tLd8OB8 zS^_&({r28A)72_moIQZqGYvr|LLpkq)HcxK5{ z49swFVZ2+`MpIt?!6}mu``%&ul72kR$;oMbp~-86FIDMQ2Cs~!)0JzFzu+X=_H zp`V+apP%o8Q@0xxnRH{?b#a27=<4;KKE+sgVE?MyQm*&{*X~8FA>G(4y)((7SK+Gu z`=;jN@sqzcmb6j6ONttx6=}UTdk6@gIe({uhMu0+eWsSiJ0T%Ko)6NH`j?`cx=BJc zA-c?tFZGV+{kVwU6`eH)dPuXu6T_v==%nBN={>QY+b@w9-TcZj)6;F}#l7U3qhH3N z-15rkJ*g;7@TJe2+PMZ5lChe@(<;i$Gn<;R68>a?o-Jo>`qg@bAQnh<+kJm~&p1{= zL4k^z*MJK|hSTA}XLKBBCT!My#me;`KdSefX-F>ky4((dN>;1Nv!WGGcL|_zYPz+) zUOMiAQx51+QhV{#Ko##fA)!X9;E<4zSJxXT7+6`akZ=DzU?zuEyxsKI)aMsqy9nz)Nc=I9rteUSN6lQfu&-V z)AobaJCYzG)PDFmKF(Ph0*d~wqeFFhD45shsGudA!vpk?%2yt1cOK=^rm;^6@ra2T ze#titzINjVtEwKTEj5o`vC>Z=RB&HtMUCz^1px{f`1w^bC)d9Kw)g-)GAUfbwI^B4pawSPG-VO$pzntL8KkF#Ux_WH72 z=6FM6BgN+&eNMF(8VZBc1uk>Hs)53NK`eaV#1e2x_5daSKX7kjZMrJ?Y;p*l2*+*O zw2fCRe0<92;fM4`u#2+SQ2e=0m7CyTAlw*%KCWg zG~GJdgUqjKwa3_sBy@AqIwKR4lg@5#d%czN_dIf%?x*6=%;8d%Z>QitKr6(AMxOQ4 z>t#yH$#rl=Qm^S+=IanE^&l_CT*!e6uxD^^kYc>)(xIppdxvC;X)#*pFk>DPU% z|2)73S&czzyog&-<01`^Yns71b`FNqIQ?UCyNx3Ye41Gl9%op)JvnoequhN(WpI*v zOo#_;YxG;$n?62(nLSm1S6_^-q8YHOWvWO`32|c2AiZ#w$J6&#dX%-_%mtPqMQJ4U zs%wbpgl4j|*KZEn& zjwNphjuaqk`1>IhfsIq2VQ&IsoJL!dqE3d#`}*!cBjsA{!PSsG8>c9L^(q-EsUPGN z&2LFj0jQA#b{_xyl6?E6L052%Q&yk3*4njZBH#jU1w)bR{Jv9D(e-ya)DDbG`l95E;EH7IfZWl$=lv=jR_r@o#ZIX!)Dz7W+yi`eGdt}d!ATeiq{%}w;2Cg=eTuYp=|$g#7x zh$Y;8;DDz23IJ*{&KjZSo%72l$V7qOX9l@pBQvv|XV1P7H$^utc3sG7b_5)i+m{B; z9E|q#g6XnFT`oS%i&Rh7E4}1%;QiGQv!ipwSXpOX^I@gmr{ZHxjg>FOH?zF~94>$=>-gns8*hF{jDZlUr^@Em% zM=#7x`x5ynn||Sm{)#fC6_nCeP)y0ommBqO!0jN!JT#U1iU~K^5pid39%WHX$aeTT z4%GsH)G@H%M)?CC7^u|S&~rs9vq%*>jOX8#8~cu*Q_3_Ms3W9ZDo%|PsMv!-5aZe) zG^c&tPXHOU#nG$_geRGt3=8JR^2`s4Vd&3yU%q^KlriEPBq&Mx5dA7o{rzq3LIfuU zRFT=}Yim0Q%@5T>2v!Qi?yZ0a;bCES(-jjLPn@{OeBg>4VY8r)-UncpiBE@0M)(s1 zQiQsJpZoUx`)zG)ZOL;sUyJqwh?Abnh|Xt&r2y0}k~p!e)THg~L+^ zoH}gaA0B=XT@06gmzI_m@uyfy>8n>SLyIIjtKdCzfGAsg`)V`-s@}@bmF4AY>&r8~ zbcfBYtWxyL?$|mw_`H0{jf?qbedRWkA}Ilj4wxGQ?sNMIGEedrblwi1SbPpY;TZ}k zu{!WvAcr2njxR4S2Z%-)6k1Xdx+Z#HwDm~(PjtG`p4aJ}HD~no^#yGDkAt21;bY9t z&y$1G5;{D{LGorOw(l7TNF1MupFE$BpP%vU&juCx2=#2OgV0khf*)MFb&Cu}OzQLJ zjOWiE0fKXew1{e9Ixxzg3M>ms7a=_#6cF$hbd~S+FL^BqZ;+xWmq_u0sF<6muZM@JRN9#axCS z@H2>+iD~=&T?a1f=sX@B!HF1cPumNkW_0r=@z)5YyrbiFtIH4?KIqjwy>Xu4ROT2$ zr>tLPk&nCUv9?IX@wMbOtLywIL&q$ZF6NwK=#kSqOG~p#;&pQkMA~N2flMa8`@u>?0lzBO8}ZH!M=!l^5i`h60!^-sb0Hw0%t4~D5cF5 zu7$3^t-ixpmuR;NRITNS*DaDQUr}{O2dcy_htfTy=GJ)zp&sCn5c~=E!)e-}4-Dhl zU#yR;$kiws6d4t<$dSHlSRAvDkCT8bKyL2-tG|cG#qIX;@`C=ZB=GJu7%C(RNhl`- zi}5WQG--SVE~0^9ov|b;gE!Y(Q9PKZLLrJ%7c4w{_;BXe$77HxDB(R&y)=Oh zMC>ROZolu}uYsz`9H)nyO z`SRs$P0eYz-uLgr z@;b?9gFcPbWiI5_gW5Mb%rr{ol zaJ8SqpFH6tWFD10d!4=6t44^bok`gQ z-03TIl$(DI<@5Ek1bD%F2RxSIA$Mv{{7G3^S*iBjOXGv49{3>x%9Uf>Ig8Z&(PQ7g zD~+x_J!_Nj2V31H1lhoMWo{(+a|+0ZFhJos=Reny?BU1{LIxGTdNss6-qX`l90yQp zeB$eyT)nT$2?c{My7<(Rh1Kl(JYck4v$EpUQN$+u7qv!JAuC_sq9hb;~U3^P!k5C^q(}b?%z*ajeaNXtMQ9;S7 zT%7ZIeN8)wn7!U7fVXAz{0|>FvWK0Wo&RMD@CL<#nCpB-tFSR&{i;~m?U9hq9}aJGEHI~~zu!PwPA-!3P98(kj! zg3@N*j0BL}jY~`O&;o#sxD!c6yl|Nli=J z1N=^jo-Z~2rQ~*cb6x2KDZz-!M_~6B3djpDDDCTG4q*AE&b~(?n8V z7ZygZp7><~(}N0?nMJ2%aq_Ea|8q3uAQpYEbL>yTRtD#5X$++mYNBdUEq%cnNa5;| zcITNkQMYaW2sn81XVw)R#`?7egc2-`f7cQGT&FIIU3L3~3uWF_ZcuSL?pH&Deb>{Y z?O>?(^WALv(=hn_ifuwW$&Ryiyl-8)nkdTsIJ-YtD%d!b76;xN8f;BIVShP$xvtJ< z>W$b-wd_e72FK;L_j0haJ4g_FRmNlvXv9>)1#8Zh1CYMh8jmH)ISIfrK!m2KyyLRf=AVO)(vcdW}~Y4 z$yvo;LVs6A=-Y&VCz%laX;7<6lZ*9%EqQ2VjhlDRBYeN8(ueuwujcr)Dfo8`ZblS} zJ|%X+c-(7c)-=gSn?xMYhUu5;FRvK*9e-xS3#8#lB+%LK4`B1WRl>tjBs=_0hcg-jb7_C52gjaA@BQ{iYTW|X3c4Av zRhFMYb>cG5iOu_^xDT=$`i+lI6atF14UeEbEZt#dW!;PCiE*9NRp0Yyc5x@e<%H`` zk^>eZo}vF~wvOZl9EmADZ)V~ubY}S=%u4sjJ{8W9Mhmz=ay(G z{tGutGL~M*AwzuR9Y|K-=Ow$@*18(HDZpr&p3~`B?5 zZ@aPT=u3aADGOvW88L@8iNVxkE*WESzwwz-M&+#QU^)zEUFq>@=B->7pI;-OU^)crpW)<`dzF_LuGJ5%>(An13}*->>(v0d+yTX5hJWN1B6c;p=s4CX zlO$xSQ16a=StHED!&Cct6Yjn&*Bc}f0#~9o5~;d^r~D(45+WoFqv`$%iVCr4pV_!D zRDJf|4;yeQ59QnjBBfX$sud(LmDri3rC6=Y=!o!QA646B1hX(*&l=NX{cM#P9UZMF zlLJypz99lT!Enl%j9yglCts2<({-~OH{>SIYxnLxlC|tLGBUCk9vF?tr^b{FA9=|5 zC6siv8OCmu6cmzfJLAvaey+E0C_3!CdFzRi-YYOMco)l^AS{*)jSo~+Nzy`B=6{nk zhq?*K7ie&-v7^X~_re8T^N!~aB|V@8hL?myK;&8&iyqzU1%yR8G&J-x2k2Fva6B_? zW`=5!3{(?;4j^ee*(+FLLgC}YdnQQV#1KMt(kh`4WlXf~1nyU6-?ML@54vg1^=sEi zu*a|@D3AiF1JVkMjny>QhK6AcGa8`A7hm?c=%S~3mfAt?(0~B8Vdo( zWEU3JOU{N*W*Hz^h)jvy$S2x=0PPRh`(ZD5zzHXM3-|i@`B5a^#P@@cUi*0lx{GvF z&zCRX5%Y@Xq$W9a`t*CSo1X?(>2Q^5ZK^h&JBal5W?6`OtPICu0Thz*`}*{bpR>Kl zyuhnps+RmW8O+!O;8R7+Yx4;*hOs4|;4CEL>{1NFstGFzQG9e_w^pHLv}DQs``eIg zc-fkF=k47skT4!gt^jfe>s_5F?#)zXMGJ4Bm2gQ*PxlALzo@8~TffrY(P0N$ip!w< zFsy@9XU|^2iQ#jZy-vgdVG(F(%OglrY&+;J=>n$=TKg8jrA%x6b{zhmWU0M~(p^b* zhZD+?X-Eypj9O&rAwmZAKW->{uFqUKcm6yTM^kfi4bfHg7eQWNPDZmCmO8NO6j`%t z9>2eHe>Hx0Sm!wx81Xx3Lw0k&6xv#1xwKW1$8Em6VgM!7&uH%GphSX`&AtSfoN84U zUZ}~32g)w);DX{f{K4_l@Ux(AP@+=ahe`4lrX3L42S{eNXcG{uUI-V3`uh52mVcWC zHl+w%tAK@*a-$`IVi*RKTJhC9QzEhfQYfQ;d7^yk`}e!}K5WP9(&g1hN}RfF=~SfQsq`y*6rsOyDha$?}*i@TP82f#gv z$Ac-mH(JbN^$;=%N~NCPzfCu^wJHC0Rm-cu`Afbz1BT52O#6Ek=2igkbVHZuQDNs( zcr50HIY1*LAVT|EJ_uQX`Jm23%Pd%jM))22zVhEXdawGfA}^S%2MFQK{QUsDRurx1 zGEsG%vOC-`e@LSdYarxh$P6)Ml@M&MA)P@nj0!D*EMv9Z*6FEl1f@zcMRfASZt*03 zTyo!W{2#{TeE)xIej9c#5IWcnUc<_oRGG+5+K@h?gX<(jM5r`|yC0>XpkQj`;^yum z547AsoHThDz3ndOSKmSxG6n>qVpEisehc9xSnmuY<5oI4I*MVKV0R&~_T`bhg$Scm zXsPZS6I}~q=%sRl309D>;v`Fu`Dk~15*K&v@A`^8?^yus%maBSCcaSioAgJXUAoWI zhzul~lSpDRQ1p*~qGe(m^O5HKh1rDnWY-K%#t4#-#-nNw`XcMs*jNyjn(9f^N>X{b z@bcOk(tfmr2nf?tsl?tC+=LeGNrNK+&Y3j#oBVv0tXq!cax5CPe5x9o>vIQ?k=+}O z(mRM9z*&Nqi#w&245@ySzjiJUxR1kOeOJH8;asI8Z}BEe4rGiGh#_qF3wI{S<8MIp z97Sactm8*0IPi<;)#CLq|I_7EtD&r!KHbzy#@ zhZ2RAh*d&t@rKYHU8il}zOGAhmsK%=@V@Q4hjBt4mkZhr)iB2XaM3>vM*#S<$u;L| znNzO0{S&PxiW;B;D$d#pOQ>7{mFwHf-G0M0o?=q& zMTH1N2To<4rb06r89BN4(CC;O5s{XI2F9j0jfjR4NQ}5CJ|p5Vsnw2S$l{;^k@E81 zsT2ebTMM2bK!}>_zrz(r<#6Odj)<&q!;`tO6O)5QTovRI>+pP-rL8OJVReF8&&)4DZVqQ6D|@-#VRCpl(>LIWNvgc035PO z{|Z!*t544{C>s#6n&j{NX8@-Z6(MyWK2U#cdX^P1842bgiA9Oda;gb9NYNbxI_W(3 z>tUxUR5PMA2F^l*;&`g2U1C4X-0EiLY7vEw*(CH4z(L9Apm z+pN=iCJNOn^Vt-}it9n#CjXiSjIop#av=zxB(m6tr3<4h=mRS1+p^R-kx$SZbVf#% zXJJQ2{uo@Z!*B_J8d(}Y^lYN79Ek$b6zjX@gi>*p85 z^8SXI*$~(vMN{xX>T_9KqGdS6b|C%)bj~bA7a~%0L_!5Zc6tya7d=8Nl1p94DV>Dp z_{$Yg{uNk2;OxPZmk+I#QD4ZdtE)2sHsQ}rl=b5uIiGh%2W>w(2qylO-#;!{Ac#`X zDY-$kX~LK*w)=tlAtGSWz8zO8LZ=%zCE9a~tOm2Aeo7{uvHXsQ z!gD)3`ki@6XK1+YtYnudT3YgZ4tA53*SwDE3@S^jJKik)s!Q{`Rl82&FUUMbkh6)GBx8!xg*_ z`fLn!u-JY$sG8SAuga4aYgmn(n(0xEW%!55!^|S<&&!e2`!oMr6$1h%p!wqQDBgQ} zNl)CnV<9YGiTZ5rf)JRRidTi}0+XwhE2cIGuUDu_tNkHNW%JycS*0$%%c|&VL^u+f zu2H7`F?VwmS&G03AiOAPIzox8ak0lCR56@YDFovlguREU$G2!d1h(~cOGc0cQjGf+ z%*R+G4~T{ zGKHu_1h_i)9AOP|sMlJh4w)(<-ff!=pFVv+lb?aPMOaAx#sF2^%np?uV3cc_CGhiP zxCYhD0=Vq~ESr`Qz#)SOS<7~xzFtp`3n()RSa>FH2d5#B}DTwO~$H{X28Q|}o+ zihS?63l})m{j))s{y@<~89KCHi8jPpxO5|L(J%k&i?JeJk^=KK@j;>5-m3fpkJ=nT z=&0)UxjEv#q(o|uz#B>|ySIYlv_WP~8;_0sQWSIj!V>B;VGNHZ_~my81F8~Hmzq}# zsF;`0RS4-d?Xt!{)0H3po9hxpL=9Z=*J5H~LMc-v&Hh5O1H`9OVA_RDCR?Xnw{9-4iMTJHBAR1{{)sdD7I0MM zPS%KH2?FRnm$!(m|Jf!kA%VNR1r!0c;t3eZIW$-Qu8#8_K6-Qrr&^tZ<@W74 zmk{8)OdDp_1sv{MB;A;WNJSw z3mn3vaFLHz`Ear(zLn8nKu!GO#k=h%-66X(paNk6MP=lmRpOQ`i!%nf*;aL4rxIa>nk`Y*7tSJ!3p|OT27(UN2%_{7p*4!( zGNTv0o9)c9iA@gB-O_iYw+v|$n-iVJ9 z>fO68nk)$_5G7xyA0izdj$LG7@~dX@)R1dBkjleO|1yZYUn)H=;KGnV^dYzgcvlT7 zuN<>>PIiV;(pJR#f*?as!CR{=J}TaL7P*86^Bro&pzM=^)KJI+isO z--qZFmto~`@PIgeQ&NObvm>s{UtBSH&Q-J603SJeA{6+5LPq&w+ghrYVv@1QR(5gq z1@mM~p&)~H6!4ZepY?M@@{1S2nBqIcAtxilKEJgWWR2OC$H~D#Eo;{)b;8m{2;%|7 zitngC(h3T`2t<3X4AT&ii{}OrX9>ua|1DiMo3#nKk;bTuL^M{kn?|*C0Wg_}Ue0df zBH~gJ_Yzy4K3Gu_)8LW1o#@4t1FX3(F>-J)rKPIZb=seme$6-XbNMMppNHpDBGG$A#nO zGTm2dVZHJ!I8rZ#W#Q@0A)1(1VF+s*S?aySyay3;hT=vzyl@#r3@Y6-s{@++2@L{P z<1$qJnV$`m!mjhXLDsnJ&5aEW4UdCV)VtPFxOOfuN(0)$V|#@$rOXOHP6_n}9;vnh z50wew?5w#PG7DL09GC_3Ik7ug6?GbI&=y4)vK<{9Y3i56EXU!~W)Ea3Zv149IkjY@ z=#QHp6CjtsRMu{6ZXSTTNem|v^B2a(#;e>r>)eTxGOl{WsFYd#490 zCVO15u%O@+2+E^w8>9rF%29na(Lyi?@1G1SBG(1dE3VuoygBZEn`pYX~vN7gGB{8!E z97aRyt*hNwS^Rvg$ObV9-tGM?roW*pJS;QB%JO2~P4%uy zh@f|&H`mX%t0FFeohQbllW`$3bl>oR5AG*IQ#IlQ`AG}?Pw06XA1~!bw)HN=o-}06 z3{k?(OZX)yRA-2$LCA!JpzZPq-qTZI7Yf)(ei-2(&|p+l6vYFTGsb5_)fcYG_bU8K zzNHnsu-yV;L4>oKdW*kW+lzN@Fg-tj#$cx7dA?AfXiRvx^4NG(+M8kk-60~m0}Hx& z;TMK1>gsmE)Rvg+E#!mbgF_K=}euUB(qykY_6^FA^_U}63{_ELXG^CdWn6y;M`soqd$qg@{; zJUJr;O?GdzK1SAn=t-FTlYnVH071xci%K0icFY7aKal`JW{bD*Y(G-)Z3(nuQuymq zNcl50*7`BVnkIldKEop0*M6UvcmP>01tB|o?j@Xbv^#1_D$9Fb8Eqq& zIRadTwI?Q@CIuC#%M`Sz-Gm^e61lMB(nUl>45~c$1MfT)awLV3b_GO88NIWlTkyuG z;{YZ(k_OGHJJgFjz&nT;Y7DVyn>t3MqHGtgC#VTl`wMs4DePw}dYRQ1ON=T2k`ZDA zfGE`*peaH0FqL%~pWS7@6d|(gBt26`6&RtlM`je3;oq@ zG)A$*SvZ2#IDmUn5~Cm^^A6;s4Itv7dp%o=tz{NV%sgfbYvge>{MuD}}e*{~_}q+*vj7Szl8s8x-E+lx-@27wa5 z870)F#Rv_;>mGy*sN(+y1KFI?-y_s563>E763R~I`>z_ADmw`9f;46zOk!f{l$e%5 z=n?{+ROlm!;Ay4%r!8hhGcAXI_rX5%#h3g5j7qN1_t@N6g6yG@S+`&p8jHXNF*OPI zn*7iqwb9pfBCgMn?Mge6pi%(!3kE4dP!eOph~jhB1_(USnobAut3wY-<(WeW6T?-y z$?kB=No$EK|5ViDD7?;^8;Kq}B_t$DVwNy60E-I|W*x?g6IMdFn?zc6Y=mG>2+$FO zr$q7x3aq5NyD-jkcw!>2U4rWTj-3`VD_Q2c7WQk|1w^r~+WcEKdIMaIIA~xzcFTW^ zoLyWP@Gb~+0_2y;H@pxljh&^g9~sZBig8U4;NyD-KL4jfQEU)<{%2u93L!~|Tz?iP zb=$^^2}^F4I%9}>P4P@)`zv5axWV;F`nQg2x-XzA!sA5nz9 zk;JbqlF{KqCzGa)oR4`Z88I3OUu1YpkUZ?=74cPpz=h8ysEnpfF_{cU?T~1az|0Ir zN_K;^6ql9#z@nIY=0$w57p51N`76hbu7`Agp>s~srSz_Lq`BzuXV52GDoe8bu7w+tX z)fT%f-$5BB8wvQ0NCb0E9VVdI9esA^_46sn%bx;=Of&x7nzT*GVe~G-%BWu3x^wj3 z1DGOm#@P{8jl>FU7Uz3}K0o$1_0}e)FC|;rMBEl@API9V$;AaFjHCh{;|@C<=oy|V ze-=RawhEOSh}aQ*xh*yzIJsf+9iHX~T2%yFcUM;}oEARVlqHmn0xoV{GCGPciP$N$ zaXIS>$fC>|Zv@(l7J_*zFG9`4@6Ap(pSa5Jm6@tphgnz?`reqTqOt17FImuhR9{_VN#JmMb`LQ#Y ze;OygEok})!Shf!X&~@kl$H+AlX&(bNXtG|h?2inRJGONml`?80KI!Kn_ zWe+?CXA4X*FhVWjXu!(~-U!?aYicSNtfokgSgBt-$>DDP?ygT_{Wm+!`86fXXDL_%Wed zL6ecV7M5}m)N(mS;%05c+b&Sk-eF!%$wf(t+!IHX(XCju5cenvxj;f&A`WTfwsb{$pwe9=VF?K#VDm)HJ?Aua9Jq(gu?N7(2$`j244kV|1V48} zBpaQaRaq1U8mORVj*-QH>bx-P%gM#DE>5*v1N|hSfxXC}TdvFwp$4mt$gk3`K!CuLHQ>;X?6%(~l_EDz61^re*LVf`oeuz<;jt$FG z0TysSE_jVsN%%(cZCRi_TWd_Vf4{l}(=_pVhbUhkpN!O7>Rl-oM1LZZMNDrY3S2Vf zEv=W$9%%}7BRj@t?9=!Did6)xfB-V8MC574_6$9ix6YWs@h5OZavw6~NKaA>GxtPh z<>m1az!L?K4nwXOFxx?BBbg#UHWt6yps{l2ISVxS%XCdYdh&#*1}bbvD3#WQHIX`L z+fbmoDUjEnBPM0MISo<47&1C(H@5py-XyJoV0Zab)qc}RWL3r z^B$trgla+*cRV|_N5CDhUO~y7z}GN@@gySAPJ=h*P(E=(6TrA`6p6&-bYIEsH0n^H!1!?@Hnw!? z6TM1-2a%oaHB+lCG|?{q%e4pl1-L8$1!+)7uuvo}DvGJi>hvET2Gt{vODi56 zk$45eKW0h~asft`mO70G!-a`qsb1ZWfNVjuD!N?Zsy*8S(SieG++1Bb)N>1jAc;v* zruSQYD=n1N#tT5$$sfKkw`ILf%dLWp=wtX@(9Z&yFc^AClfPRby7`T zg^EG({=nE#p2|NE+mq!sq|P>|&?%rP!7GFIqw1$o`&U+q63OO;Ib;c*!QM-wR+Lfx zxxBzmAy+A4Hzb*qpxe>WAw|iHbp5lK!2!I$i9ly1AdeImCR%Rk8mr$fe%Fk?y?5;GB&TYy zYL4Se9&V?*T7JXWm{Iw;bxSx_fl$!c2Hw~tsw`ZVk{ZO(q8M(cQ0%wr%8E7*y@LLe zSe;i6fwz;w_YlKpPF>~qr8FbBiqW#G(;a)f-e?0I~4Nb zV#9IWrXaj3CpbnE^M_;npnH@GV`v3dnN0avt8sfsz}Yn{T}+9{#k_F&jTSmV$!*B> zs0eJ4>!CRoRguHC7o@`bCbUM6t#Ev_u!1pLte(1XV_| zy@`;Bz%sq7K-q!{3JS#M2r=!ONjyI$Bcu$EM-G=8&}FRZ(GbC^Gm+6+2Q0C-#0xO0 z9jAH=gJo1U6SFo4`uqDa(XPUhN^$b#6?@Yn&G>FR(@Rj_aZ(&#t{BJ^ysIe^#EV^s z$*ZjX*=Ecx0Bc6FZK{2GLr`whBma$6${bsjM`G9^L%@+R3_v&2SVH0E(Gd6cI2F6N zPO==V?(n9-1<%J(QK|Fyx$=yrpd(_=I>$C1?O4jdAVG2k06s|Xs-Yo0e6vV43=n2l z&k&ao?ky>|?$H`odlj?MK480yo1^+1$7(42KqB&PeG|gH5Hha&jP1C)oF~Y@aEY1E zq7;xY0JTvrJRyM&!)_ePbQBccU^7`B? zTUO(z{mRZr*Xbin1LoG8xas>p3YK-~x znKMPrhE7=#*cwQ5YAjDFm|xPENWD1=1qZJ(Qf5iDZ^48Wtg>*nO>_|=UN0y)e--N`&nXY}uBgoQwp)X5%T|sbn`PUPhpIMv`?7`H}gLg&+WeO_~ zZ?55HAN}^1>Z``ZnaK^9tAFsG0%pclf#w7e-DEBi5n4m{8QqxGFmT~BY60b5NTBCqv=We9z z$m7$Il7O6f3IJKDvCTt=c%m`BSqXZwgUbn+TY@yJ_tPG4iSR*(lPt-vxOyHWglrF9 z&Llw`^PE!X3w%iS;N>{d?q?wF!fv6~^+)j@eKm!*0^m)0iOmh{n9qq}-=g*n^PoJ+ zi@>ME3|x*Y408^6bN-@sqZD!byXM7SLrD!_tBr^qHchCAQ;B+^$v1Hz(R~8 z823*eoe?)c&Ww0{mn88fv0(iooG%K~$L&uHFJIn?mgEDOmY4+RcyShbxaF)Hn6IQi z)20%pF(uuEL+~;zL~+z+#iB1y#65z(Ab9Bo8~&^slu`X)jLKMdF7M-Yq>oM$(B=2w zv4JMMUk(;$aJ&s((E;ZpS#kwIYo!U%jXV^Kpv6kf1etN>+t`mpwF93KT)&4b5D^t5 zB6`{V#BAl5r!IBQTBi}dL2}&!0)K&I$8O>cNAUmH3wZD%iRk&A$Ec{7A3tMm!to)$ z(5+wYd=Lw7-xY}$+hCYfo>^uFF@U>BwMwplxIUowT33Efyatn)$pcyb|5$tTxSrQ8 z{1=;To-z-aqZ!GZkTO&@%}T{a6bdOcvXNwpG$<-X(V$30MP|`}W=e>XN}*CJao%_P zdCqyx@AW$W9k183A4A{a^ZDHOy4SkabzMula!>kzYrk)dyD&3^jiP{#`9>nl^0!`;Sobe?|J^ek%_Rr1LxCqdgro@vtM}i2r!u=!>s`N*obWNvp7x)b@n>HZGtlFk5;{57}sT+{ykliGwDUE+8G$dS_gV{I$R zs6{J;RC;OPB@{h;=pprH;_^Wf&h+>oGy=|T-wQ)Mo>mE7=ly~L!_c2ZBEiYvqr$hB zOr1Iv4(=e)_z?uvxp)5gr=u`w^Qb#^tXD%#@esUHl&&;^gK);hFP?PByYuoYDZLV& z>NRX8FpByhfVa8+pBF(s7mkh?JxI4WU=*i*dhVx^=9STDSDO$EjDWj-f4qip4HdTr z(Xrfj6}@wSIrPa48%)+@t7pe8j&?Z-VgZlq>Feu6oC(~Q{hUw{6T->J|0WWI1}hlH zb*9lNQRC2hnW59=?5%s-q!m6TN?Lqgkh=zJXB~bpA;{8!2lAEA)9Xa z^=v*;kE6uwBPW!#__CM&$0gfeZvv_o5xLVF>qT|+HN*oB1ZQN;* z5C4e;;I(vuV_VB~5otKS)sXVqqAuSS!$d-BExtL&Vd`{be7 z2tGZ!-uU+In=oZ$G5kXZMToH>P07Wj(f}5~qG&*J;`el1G{EUa<24FeilgCM3Znl3 zIa_>Uqz9Y-0@O4OMZZ(0c=UbNQq_mvKA;_-ye#~n7tO`>0R0kue~Il+^Kbw_;bp8F zL7y~6uZxT`@_#Sf`8Y9!SekbEKdeD)4ia9DQs}bqF5Ne{k9{!l)q6j*Su`#zT7o-% zIbnvtuq*d;!narSI0Q*de$~|udg0nKbtY2d$NvFXbf4a>V8!FYI~a1Vv~3C9$cJ81 zT0MNq%gWZ0Xo~bibb`SJwdA2AyA%7?@yzy+0WM5JA-vVvIxOgc<^iI!i0`(Fg ztO@8|ihiN!u#;S~#5u7qwffY-&PfAzYxjDhrOTF3CtwNANOt`vCt(q0Uo{Z9!o+}EEWtHAgC!9! z^+RsXNFB%X(z!e{j@9%4{(Gp=2f~v;Ap}RB{`UCf-xDTI6rXo*_gfbu>3fl!5IEiG z$UcL<@PU#jbe_kWWV`Od&ktd}SibPj2|+{PThSy6fA}Sf0*jlV#uJ@Tm2+^0iPi9_ z>pBaQps6|cU^QlA0wh{pl5_0(na!9pM|AQZ3KT`!L2uCSFJ5(I%4=yEOO*3;7Xz%q z80-lpBijNb=Gx|b*^z|um9vZQJi-?85e)CE>mK#+crN0LSPl?v1$>VigUJRn?uFq?RWM;_s@85TPeYX$oA+CZ3nMP7_)!g2X) z*37qSk(-A;_`0$ZslL#p_V8MaD((RL@Gen|Vpfn*s6K@w8Ip!l97t@s5?PYslaW@%sS zbkNKQ&p*rsyHpq%sGLdv*Zq+9>d--O>2L;R{rt&K5>dv>$@L_*0w&xiIlI`W+!S!c z-~D+!99y;!Vv4;>ods|)>t~YXI!!79@~Gp~6Bo!F;P=T;d182RqO5EK z&Dc^vP#{*nxyIXB+U(QQ50j=)ON(lNyiT|t5>W25>Dn(d%JZ~jhj*Uyf88Z^Unn@J z{jTIdB`bX0&mZ3;=Bc80{=k9s=+5t7N8w9_(*6^^!zf2Xpq257!cz~!0$jzti&B!K z!DYSz)rvGl=+zsHS%XM9M5kVmvIJKM*DX=Q17`|Pt@nH9&~bw7x>u>YKY2?U=@OLU zH)`cLGT8S1vU(J<$=G)$xqxH5sT)-YMLJ05K|=KYMl9+^(>y$)vRDYPXp(7q4*d@s z8iwT=`k$NC)iw$c7@5L4&z;at1};CT=l!pRx#<21A9nCr2ztWgQC?mill%ju=a_7K zfc2{DX$ArA2fJB+u%D7<=<&CFLACDwWDLSs?bBsRl$OF@UF4MPm{2l&p5y~TQJ0Vq zHi{lQm(nd0%i+n-91R^TUH5UM9-a8D!(CBnZEfwe52!|jBai|Q*BBY=_#6^~{o7KT z|6Jqp-RI#ZFnf`6AaZ$u1)^Z`Q&=AM@VLk8w>Q0MK(iO@PQ=j@QD+LI1fKf-3k{#@JRWKnjC2=h5gKNHHhW^hrP4A+QZqpFX|g z==y}p{bA3nyD6hA(6$_-q7C7HXwmz#dcbxMk#dXGgCi^NC~Rl_Kh?AWFm(BS2VL=b zepv8bST1;6c#zw(NIsCMpvN3+J)q;4Uz0hkhAiJNoF-uZ{p*)K9jca6lcJ1BpoMLi z=I8|xUgUopGtc}3?fJZId3seJcU3ITcDfxZA3@@qxoH0%UmPV~0g&M5dnNB=!a7Lq zLN4D>mERN0Cba%Z-M*Sdq#invX`v_*h3m=njaj;uPE?8>EXs}x7J9mjmZ)B)FDVW< z%o+ym{FvFlO=aT5@F!!P1-_p~yG|Uq2JTDCy*UU4*PmocbGdgqg}CV8&5qmM_q`CR zHQACCK0sZKbE<5;5@fqH_9LhPa|JHh;8x8eKtR%?FdNQ@@T^D8Xgj&JWcS;9gg^St zens`+ckbL#n^&=PJ>@u$=0SeGLBnf$faYJvE!-gzHF7S^hLM*P1rLa{Oz4*FR~2Cy z0$C;U;d;?PBaVo}s2vFrCdZU!LsqX~i+zJkw4T&Qs(<)nlT_!~tmrYPv^mT-3gSad z(1VO9P9?l;?z6u`%IW1m(NesbloT;5uFi_M@@QfphO(&-9VcmPY3Y6XZB6hZbvrN6 zW(mJ=%Kr!Gu-(XW54--N^C;{wUOLh#-% zlqQIS+KC;H)_70&_iYP~bYQz1G(PtE; zpb(1)b8|xSlMUL_b-o{1MmH^b)wNEd|G7*0ohR~|`!sGonolnmGb~sja)HXVb?qs;Gqe2LM(`c=KRX!YrxmbfOX;83PLdh{2N{Xj85~I$OLZiIzGF#D zzP9rYd9=u9C<4u6Iv03?MGmC9v4obPfrOuY#c^IPz6&xGT}w&&!J_{rQZzT;$;&ue z9+BXRmbFj8YSes!03Tx^YGU%V!E^1xl!XeF+`25aK&CIzP?#ip)KODI9W?LV-zNrg z@O>(ahmNaQLyKsopPpJi1Jn9!9t8_GMx%ZoaTRWo(n!)|WX za6!7zuTEF5i}Ud@xHqNg$ng2DKPNdI4~Zrx76^^(XanJbEKrNgdgztnPW%Am@Y#HH zY5gy$7*Xw(&NwiX-jdMYgjsKLh-h&Jku@_VgC>hGu(~Z4kiPA|G*a2`A}<0z89Lup zg+tLhGK%m9b+mtws>_k~+Ii(nDGfBZl%NP576T|`V-0(3=+sVxi8M~P{*=vFrhI;T zXY`^&y|OQNKD9!Jht;2Al(SeEt@P-7LAaI^41{H|__~AsGTcfKCl?j7WSm=TZfxJV zldi@eB=#3q_T4lZ8T~^ZbZKR!2TSH@LNv`a=wrf%Nx;7mvsBB_&Jj*N@U8+wVT!@u zn>KYtjQOdu@*av{QT+I=arGcogZbWpqEmjAdB4utWBsQq+;afq683hc^CUodJAxoA zfb6)qm&Y=T(>R+6|1o@MpFXL$-Pf?!w+wkXeeTm?^NmT+Q*)hvW{J+2Js zzyosU9+$GuZRd6Vc@lt3Amcbu2*^*A2qLzV{^6>vU0G(UAwHZPC(`ll)ddQoV}mMQ zr{&&G4op#NkwL5{|CRL+0M?$C1{qgu=-ebaHhljsOvHtYFX^j~L!jWBvA!%^r6Y(4 z0_w%3Ky0qpJJ7mM-}0VruYeg^n!dzt*Tt_;l4{PKha;F_g0Ylx0<7V^;6_69wD?tH z0DX?UQG`bCR7IH#Vdv3urxi?0Jyt9IwAd#9TF!|{GE-Q-!U{5CPF7adREu=){pMv{ zW_YOO{N5y)lj}4*L|((A)V#tw{JP6Jm}*_59OiXj_@t2;j^LEkl-lZJj|%4p_6OSm zV9|#@xdl)-EcI8*7P{v{L8+t$zXhZ1hn|()xSq_zlIawLbtd2Z0mdUt%}I%ckBuks zno{fHi!bBm%rOc5YxU}4iRv(eT%p{>K1twK<#SR2Si7Sa7S1#SUH?&&qz@oMVIJ)Q{m%6C4xj;W3uz0?4P_a*!UBNii zSNL_X`;_5ww2rw6AkDa7RbyzdN7oVi$0~A4_b=6dPHvys1^YrKb zz^hbWvP9^=xZ)T2`EH!iVhEIlMHd*D0+JF&cdRq5=(NGIzv(-Oxhp_drJNiP(9-5^0@I9x%5C+cVEiZr%8>$Oe#JKkk7=z;l?W?tBV*2L1eb9!j4z93xP3pFzJbjQz5+FaoWxGqle3E~?`sFP1JCM;^ zLcSAv7$V69Y9>voo~llby~uALv*N$;1Tg*+^EW6Hg!MKQ7jc6)Lq)MlLeFucL6xq5 z^LxN#Kz&LIak>a`9sM+#36bwcHk9tRz@45CYtKxnqT&0mT6r-6wv;v>E;I)Z9(+gQEEu$; zlt2Pv!%dqu4X25FxIdC-snh2vh4l=d=&MwM(nR8lxYO(kvDSs{KX(94F3tmg0u=~* zMGi3G-S_O-vrl-N2F$RItbH;4FE9(SBt(_ZgCmbob9Q!Km~tI-Ma}Y~xFS_({WWXj>*HrN9vs?9nFc@{qUB`F>RVXAbq=`KQj~6pY z)R3P(eY#5(OYST@6NO7ClrQ1I3OYzK<8|5Sjd_`tmVlB+s|x#dp;rJMSh{T4LNWUR zApi+y2>TLvG}h8Hi3fN1=utWSMpc7d;Dig0{Awl+Y)XEQUMbA9`H+rZ6sX_0?#1`!>rak$y9x; zs042gf4u{&OL3>M^6Ju**1K0PahHJ_M77R(7N`@-$~?(Af{-I~xVFWPZd1|m~B`c6h7el+qo&WpS z>iYlDW%Cgq2^gH9N7R!5yU{#{9{TsK538E7?H_ZXNTs{?9`Wb;_3Nkq`z^?F{_mF$ zlwi{`Hf4;gtY%0#R?Y6vOqtoV@cngqh9lV{>4%oKzpUXYy%9Nq+z5os(?W>M&VwuLKjpJgX8g-3;J~yVp}?4u!Qe%zb+O; z&`)NIi2y4lsoH))I!Q_Z-B6gXt@J1Kh?AN1DMoq_wZu{ctrQf+nwsO$(ee@YGiaG3 zc^nxdc$EBFx;VM2&%XBGrh8IsQ!O#dY-z0p=9PlNK5_lOlj$Osf>DXVSa5Zp+}=ve z*-q#3zyI3*=|78nMRTYBsUl2y1SNaul(qYk7maPdVJ&6HN6wZ2@q?LUuv_{+H^QPv zJE4Z}FjjPqZoG9p3KvVpbP3&2p+D!V2*_FX*opS9_bFEiWS;gzch);qFo?C7RKXl3 zPM1?XE(ik@ezhoDcHckd^4xCif08XG;}KF%A=`vZAlPfL^A1ovh(Re7&b%4HU5)Hl z3@Iasi79d~3O+im`}b=`deO=fVt)P!o;~vO$zM700qNj-{Q+={D-T|n(*ZXGB@t(q zE`rnw^ppReN7P@(|8YA?(f*D8_dinC{1zdQ23Kltze z>q`9RM*h#wir;!a{Qvma{~zBzbj2AE(uZZGr6ILf`Ap4BI>eK9)J7RZ`HE?mx-z4x z4=?Rc6!V?6OL*{?{M|o7vZ?pDX7Pu0!IiI*u4iUA?b&nq&5em(htjOZ!KNF>$-aNgz?5_J zamX{pVM)%t4ni6S^klcgf_L=4zb-2W(g~$mBg}eO|Dj?T-7PLoSzE{4EbI**IuDqr zEyNg?;=bd?E+*Q@M#W0gJrf3{KK7?tydw@~A}JH|zzAzmYIFCG>DH}xZw>q)aQr(0 z5+(ZaLqW_*c@C1XMygtl8<(?({Uxy|p^#`M_!3$w|6XU+<6Z4LSf_Hp2pbpRqaOhUQ6R2u23_bzS2^iX#7$&E2I3boa+a4y!du zPWMZ71HwQGIN+~3e%~xwW&h!FXcjJ1R}aXd#DHLF9iqe&Q?-QXOQO4CaPhw*e8awc z*s{HYm+J2V@TH2|7Br|>iPko&Gq?yRTIT6Hg6_g7X$Y|iPNpR}n zQ07h?p;CMgvkrKY&z&h04+y9|3j7#hgMwS4=>NS}`IGE2fkU9q4N)S4qy`V>AzU0R z(y{h~{!gDy$Cp33#b7YeX|q2p%wy$oDxi3BXO(|ap@UCd1AAr=%FKhL79Bf?3YTat z`eoqKQi<)p=IR%XslT6u70vrlEBK-l-?K`O(97m~uI6$hBq+>cDD)6AJOvo~l1TnHUB zT82khW`s>#Kll+|lGLdhO!E=L+(4+K#%=vSZ-nl3I`c#P&hR8v_;yOUbsy-`hXnS+IUR6(uiBaTUUpjPnsibm@dnRH$As|NjArWjq zbuMskNLB`MijtC&@Gb^Z-#?o9*7*R-^7yure-nZK=f6IG(@mAYoPXoy!v_zm7SK#P zH+Od*P9Zf~-uaRzG$jGksW&9bY2CWFyTXsNe{pk^5};^LdTAcE+yT0c`W6ft6(|T@ z$|KBI@J^8o{NEMG??35z1qCVUgkQO`Lq?DdlAM@ww!%;L{I@Y!pk9O<43 zI72u1uZHWS`k}BgrILTaf(HH@f#6Omd6cdvnGO}Lm;?2n!h_kr1Smj~o-bd$ikPdO z@H8vSEjD)Bz1`@IdNLr9MwvLUQ2cq@D>*DY{5@@aY~&W}>PjHOTZ@mze^TT2 zS-^+YgDD|8la%*^E3^m|Ugw4FtB!s)ZDaHs3rfZ^7nSi_!^L=6z$Hjeow(`3ZyEv; zMFLjRI=c_IhoW*dmIIa${f7ur>VHXNoFR%bGy*$Y`ljNAkNgNWf4l)vC!ydK2 z#ds`yG*^oEu#i}?L`v2iV>eMQ<2ugnBQ0+qN(C}SW9U5N4{Q5ncq8HwUe2Mn0LHNG z^5m4Y&qnM4&x}tWCQ;626N~{(sY5JtpQX3=kGWe?A_->cPGqC8guLu9%Stp=IY4Fs z8?oUxL_VjI2mBYAyQnB(J&C#*Yb(m^E*^H`;D@27TKUi)``PzES@(dt@FJI<;(GLw&BgO?sO2vN}<|1vg=i!Tdg*@3;zSoxek7Y;A;H%3?!09Zqy^|Ed{Co z(ZY6YL(GG-hG`(BQXk=RtH3+Evi6}{ zIhZ?fWr2;2O-LP;inPnv-8e}L6&^%1ja|DEKdCS#k;m4T^2O+CiPYGsQ~Qg%PI$#P z<-`67$N#OzLRdcAw{I7at{6xMuI;-@zKVRUCsiL+x$spJJz;_)J&-<37hp!{eQ<9P zYH1;exgqgvt{2x#h6$X2C@v~z5{gnt9b^+sRqKdJUQe%w!g>&ZZXvr}s@g{bc{UEr z`?=~d7+cc~#(U)-Oh{PB?=Y(qvTa}jrF=goY4L01eAkgIYyr^F=Kb5!vfJFbb47iJ z_dhkh?6gm4j`q#-bAw1Du8J=ucCGAjrU*$0btPI)K3=R*Z*P5R8Qji# zh(}KsL=3+G=JcuBwjjT-@CaS($uCzxlonx>maUi<24ffNAEC-Z>4K@cCpS$rZ7quv zuV}i5zbI>)4#pt3K?EM5YXUYOL&IB)6Xetr3zQ{ro|{%4Q{u%m1NQs_b`DVH67myl z1+ggT&aoy^FG_4No)j|Rd~UOO$ymPtY7k+>N%}JryZSNT|4gDuCk+3&%nWC3BX!&H5!QzaGy*8T)3fNY^D)oi#I&rDJQWKiXZej9}a zHIzGhm$Dtwt%9<3(Lfl4k3iqqM!M^+IWO(k83Mo{TAc@P+}M3jYXogMB=-yH%)Wrc zgDc-?!Jv<-_1Zegr%Ukr-?!u+d$TN9Nn#8HH85lE9ts4AIfx=Gf-{&#fvDcczX{Y( zOq|mEAbleI^^*RbJBm_JJQ>P4utJZwjjabqPnvXT|CbE@C{_LZ`SUa&xt;~TX>=li z6D5k66H9?Zek5$&gu@LNpm1%-RCrCI!b4>>o>PQH+^=&-;+W{wi}N44)_Lll^13(M z*+BPHtC@O0S}Il%_$2QNNmEQaMxoxFV@v>lyfIvfq@0T4xEHCJ85weRZ35NgUSq=h zfQ`qtTEfXJCc^l~rf|Xw{7iH?NU+qHlE`EcfzOb3KH%zUP00A{w7~d;FmVt*zz2$I z-HPisZU}2_jCG?|Y@B6ef5V7ZAK|Up(j%Tce?BBIjWLIyZVxE@Ux44L1@|7MvTEHg zVw0HPA?ulN(JE)h9MC}lc2SxL43xu+F1YZv!W@>0LAVdd;_*og8ls{Ffy7I#x|jqL zYWb0Ko8TP@7L)dO3@5ce+WWb!x3hBiEuT|Ta{FCaHk%FT>H*(^ERP4M3Ju?r`{~`{ z`jG%b{2vr=#2cNzbg3%<5d+y0#!Na7+8H){NQa+*-0<)x>l=!Y+}L^dhsV7=Hsz3+ z(c;m$mrzYW#aG7AGRX6{>l&jSHARF6MwZ-qd6NXUN6_2OfBq@j^*m5KS;VACLfGFH z7H-^CF~kKSs~A(KU)8jO!&s=9giM_ig878|Sm}aQ0^lKFPPRFj`p%@AgB9g|53x_j zD3&I-8();XV?mWOR|j8KnOo!zm^#QCCh5_jZI~kJNR;}@8CRu{UT zChu(@3;}GWvuEz^&o9U7Y$IPB6!YaajS|hg6u5-`zAlUw{M3lhQ4O0%M&=C`EM4%z zC*NIo_S(>6j2w%V8L*r9m3c%x$)jyuW^%sYC$H;Sb*j8+jraRszj^**?tN!1t&DLW zHshm}XAcFTCm~@VxW**LB4#Aq?$8-OuY1h57p%8;XMfEPXmJ`8l`EcqJ*inKXMcZ| zkF=f&_ImfhIYk5xOO5rKxJz4aqX})!g(YMnp?6+EPq-}qi)`8Bv6x(%l-T^>!yY`` zh22ro4HqN|#QOc|iOc(Srjf9Q8y?TPnPeo-h8TiSIVxvWo)qeURdE|U*gOhNzgfrX zXQuqUf4fkl6$By+Am(ZTtTQksAu(~VP^seZWKdco=7(Hq{qLy5E?koTiBdXf(v4^A z-aqUvEWMPhFOl&Iw?DTVK=(cp_T09?&Z9<-l!t9%81uy(?O^+aJm71$4-%ah3OlW< z(BDE(qV$mLNf|GTJ<37?sgdjnuhdk-e0ys{Lti>_CLv^BJzUmwILH#}394nV=T3G* zn~*e#j#d}WF8`)jO0gbfy*<*xWPonQ({Nk0d4y3>9uWJ#GxVxq5{ITeWKm1oFUP!3 zUN9LOUYg`hcONtG&S;@@s(pI}HqBL9&SRlLms(5ggbW}?PTHm^{byKXk65EzPy0Ce zFK7emcVB|GE*#F(z=uv%)2`m@MirBxzlk_{v`rx$?aP{%NbFuvEhM=PpzkD1AAFJ( zciG$fTyA=>>(H26DU)b;O7Ao!j`fF)HgRV!saJpH=m)oO20Wb5?MrQ6Ns5M{s?IlG zR}?hqQBAmU4#*C2ALAX^Dvdi-k2O&^KXwzZM+1NUtGzLUyW8&@d9Cf06z<-AQreK5 zdY3(WZi=?}LLNr=P$BzZ>)l!@8TWZafUL5zKE`a`YjJjYk-DA}w%0Hvbs>;5tya*g z$pC`vNs{R+6GJ}LUCb)54W!z+4_(h9bw~-Br$7Wtt>tS+wNiwXsqQjl_Owvh-=8{Y z_m>N^;^Jren34f0ifq)EusvSX=?tqKJMHl>{=C2<1hYr zlb4SJ9$HX(W?)jVxEa|l!YrzEmg+>`(aefXQL8^J*>hmvoB4MJ{qNvM?dLM8LH`;3 zDATrexo>?9LKx9@eLC_iQxJtAYuJ0;AaW5ge3iDw zDc?Mr1@~Q{PomcZ1Lw&W4 zY+>QI5vDj2pwNwMO(s6jLSgK*@#7Z>!qok@Ov*VWEQiBxtQZh>@nSbVTc;{zHnUa7 zbdx%Jav1xqevye)cYn^T!RVaM=egFGBC=F*+tzJM)fa>4)TWtRu{bRItB2$K*>mRf z5TNKI{o$G|3qR0up78M+A8K9s8BfL&ioT%a3Q>H-}}XD0NP5)!!)-E zTP`uhCL{37&(e#`MS_F9p#16ZFhPphk3Lb;^(?+7-djpU!)2~a`G*LQ+a7O!D8zIW zUh-~U)ljS^V5y)LExhB!YEUaqf_3MXcQ^e)+6|K{tM~yT&&4biJFh(_B~A>tPG&yBi0&bi-c3usruR z^(Mq_x8nIs^~Oy+$>?T?^Ewbx+;buP2?_sC9Iz!ld4r?O%j6Ea&Ybns|d!70=Gzx$)j@ zx43|lNpP;hVp3GkiE%Xbjv1P*O5!fh+7YU@BP(#Ok(~=zj8E&M@g4S~1Kdt;T-2c7q>! z-Sq7uc&wXqrkV_Z2EQtz^;!pkJ9Q_=5Eopp@9LT=sDu8eedZ%}X&6}hRr6G9cbbQx z-D{yy=t&>)MC!BDZtnbNGSbHnW^zSQTP2CDnzxk=UpF^W2YWR?I?#6#M-YcjZ-RZF zfQG%<4F@x!p_+vp%x}si=Q{|~YT^b>T%oP9R{=&{u6?nzxAMfXJMmWzou3IfA2m0A z+_IFFuE@BiC(FHE2r=y`MZN5U z2hvd2x1G8fNeTaMY|DdxC!7mx_cpa3`WYo9W+ZGz zvJ3m}DU>_?@`2&a4V*5rpAhrDMtw$5d z@%;x>@3gf&DBLFs>NyfKZY%6<`|f9H#@!f#eD;W$d6BC#KEgf|1dCEDeAdNpb#=P0 zR69!-JT7QhZ>Ez$MQ0xIC5nCOW*)Q*htH}xD-p#+you^kT9QSNTF>X_C_6j54XLr; z##+t%Okf^V_4L<+r*IU)gUaSJ{FROLmv&B3VY)WSxBSu#r~FpA1lW$2nml=hAkS#t zk#jih2uvH6_uzKlN%X739Cy!wo|p4cMF^3OgumyVzt+_W z>KsMObPq>Xj1XrDfQpD7Hptb(ByK}@{KXVojCr}3ZHZq-2YxFk+M;bIpc%-Q| zYUt2`3|bT*h|pr{hGNtppa#m`!Ll=_PuCEi3?l`cYdlsb03l^AxiFH{WeYcTT=iwIyBTp{ppr2XAs$$)`s9? zqqU*)E?g;hfV6@8FUu1SZ33+a!7z{win~mh6x|NA1U-DjoKGy3nS#2-$fO^93Our!~_HVB#{<>BY27WEZA4w|C_1Q-|}WJ`c7tY2z&BBElt= zGU^2c{0S08kcz`N`yh%GL2m_w_y$R2Tq?D@-znI4LK7TeMfAkLcX>H&)-WA%5jXUs(HgsVl}wVDA2Tn zfq`cCzSR+N$@2Zq%tz-#Ztcd$AL=_gX1nQ^=6^?tQk_qhZr%I2WcQwdkAC0%wqp3$ zixP9X9em@|ze_|X+2s-Y$7sw64Sm_aU;knw`NR9hXgKQiAGmVCN)yX}R)#h8kvZ6T z>8rP+RZ9IfY&xA%obp=I;Z9J+C=1)UGk2%Voja}d)iBaZhtJp2E|MYZUS2uY!_(7` zcGweeE^mt?$Buauw$=!4TL1nIEjFbkmS;uNFA<6CNqi5&FjpW%f+?) zAqNcU6-o$iS%Y*%i)%)$MaS57;`8ily!C|_0EqoE?8v3oq&j;|S}{soFcDtp%jry| z11Ec@oenc?bDl>CRB8Qu-Ije}SIJ8DgD9{R;m;$8AA*2RZhlH6ruD|m&=MIR(!sxz zwO=jFRTh)$FgJVhbsh{FFTVLQ`V3MO;}RoA$nk!sVFR`$57h( z(R%1=1uT&RxumivAC*CvOgzNQUdj@w5ytdFz(HfS(M_pR3odIXyzza*P-m z=GTMcl1D%PJCr^%G(h10u0aEoz=nBNW zAz!ma0P%{!qm&GblWM4@E&>A8QT_NUx*y(h?83!H88oolevAUvj$pO}9g=0@sW7}+ z0?j{%1tY+_&89ev*j2^L6gUu*O5gIJ-Yt(x74{tm`tStv;h(o7YK6Nzj3XnWw4scEE0z)<9%f`{;b13C(D#N0^Ax()uXn42WU%~#r)pePy_rW! zUaollfK{68OV~o{oZ1s+&u(7mlH}B6^Wd95(BLwV{?{MPqvi6mUzLinjzFNpe$AUT zYn9M$lQR2>@MBl$MJIX(9ViEC0?Y>n*9Go zwv88Cl_S}QTPT<2Lv3UM?=c6(;?uJ;V=I*`EKsi`@8GtccnUNdhO%cy#o3c5&u{&} z$SgfH=j(T-HbPQXjWE6*?k%muw^y40*1*|50n(Qcm=HLZ&z$HhDYsl1=1EF4>rRVe{`zDtp-gVb}*e zysC(rQ^{0Srlyy_;GMp4X);q#QF$FIrNI!QP~34Ief@cTm&bOadFBQVq%*Ti?hu2{ zTfe+|@2k}_h;7or&I6L>uRcEA{`mDXE}K_d5cSjFqCp=&qr%_IE7XU{*?tu}pU?9v zPmFYw`7q!KhhVzxuq2MCng@O7D@U&&x!Ib(TT@#*531ewv=5`&K-HLE#3^Uaj# zosex~6u=L6e};Pea(iL|RmBxzX1YqS!^dGq?*xZORG&Rov5sWm`njp!IDgY89k9Or zSD32g7oP7tzH$-16BltDmddE1u8XIGI!y8V_fHo*O;koKkX>>kHY}i>sh(RXMV!a&($*+20 zy>YJaiR}G6ts=F8NG}&1ab9I`Klz_Fk8b&9(U!&-YY$h9?ca> z4EZ#Cg4XSrtzWU@adHg1xQLyilvTklIduIg$MF^z#Mf_X2ZUPz*UkJqeu#pJ{#l@d z!AxkaZ2aTTUMmw1&{>Y>m|AT**eIs@SYqGay_e5ibd`gwk^bZs!fLt-M?*N}bjHL4CrhIBnE^;{!Lg3S;$4i%4{ZTs-+ zPz9dV-%c7D(+{XUP8Gv>gz4olUm5bt)~^rx`ncHl&xXTdng8X(VOOu~*PfTeYFll3 zBG27ztcr@=Ob0g8ZhjbvT^?z(<*eM)gf^q)J4<$WH@$IM;lE~cTN|5wGeW0l%ASxa ztmzY)P@{}Cy`d0jVU7OnQ1IJm<7qVj5IF=epW^U~7cD_7az(fLbt-|^@Vg8zT7Oew zDXAfiDCGycvT5+zwe6()eFX(Y@-0(`L)sS_?Rnn>w&E)9G-wqGr0|~?uF3Wtrev0| z*BB0so%Kg32b*!ePq)(+!=g};VJ(rXyxgg!@=|Wt?oTi7ZBNV~kFW-tG9k<;J%|3f z>bs=D`p?6P-|#@5$Ib7@d_My$IgP`wUY&BZWk7_f{UL2m5+tJr7UqCyh&ren!`EFM` zV-}03G!mNivLZ|YHqH9jlr9&$%3$KO58WfDhLLZr{@y$+3lc-~ZF$K7aZOv@BBo7k z`SHV&1Ry6|+dniAL3u_EiAeh*DA+54Nu9aN;>c=i_^`>U+o*5i#pxB(k7jA8g6Yi{ z+iyif_&4-i~FBmYf_}cCFma8hmuLZq%L9nftK+Yl}mf!bi*a znV-4i(^nb?PTy2tu<`NP_Ysq~%xQiWx>-8Rg^;S4aaMb=*s}~a%+)(P&nZ#EfgCRD zYq&K9gTnAzdCE;ao@l%92d!2#!t#n*>gopv4;*+gvl^+))hef?bqm6E#&}Hcy$_#t z?YYAXT+^a`{}yiPkJ2uu)3b`76zr@^krcVx zj@Dy|a8St;KK$x)?$>9%U;KeY$y@Ra6?W0N3SiE#pT~)Df#r$D3iDUp+Os37(~e^{ ziDXMw*eb-$8hycH09~ld5SU)PccyZS>}A-yM~@s_>$mMteJ1>)ht>Ltw;%Yk;z&Hy1~us6Z8moY1C}+~%r^ zicR0*mQwhd;E3bhEUJ@Gm2;Kk<_3Q*{IhGe8?l4xkQb0HU!HDg{+URtpcZla&jZH@_clyyu=~YFoaX7nu z3wc!`>5)zS@?U^2a-+8@kmLT{!S__A>*)R!$1cE`$=>e&6=NC06uQv{%~k>7+#%f& zaIZDmV#&M0Ql|Q~)P4V+Ap2?h)Tuhv92|8{Lc2G(VjArti&113H$H_&hg%hq@%Q?} zL2pgW@RL@`&5y_wt#$?hG$fZNP5bayTLuZ zdU*hM%_4ty=H_E-O452kM5o14=bu;&Lc z<|Acqw91q|7cXoU@R4PB|txHp&hQrPE z3&O+RDqLZk8~7VlosRLfP1u(S<|&<#ieap^$TH#6 zGA+E%or~^uUc1$iGwI5l(ePU$Ze=%?|B_`K+y>b{ic<{y@ zQ&vQh&OR{qIJPn4x#d$Ysq`9_?FPMlo;suNd3Ci=&a$sg&P-h4bURM@q-k#L^Qv?C z%Ri8|J&SC+%Gd)7RU`w6vpm5oO#Rb1ENpg+vT|VyrvE&)F)dyC`huGj@T&$rjJ67x zfd#yUIY{~F#C_V0Nlv~`&DYPRq||+OaKrN3AyzRnPk5a>w{@uHR;%+V#*ycHfp$f0 zEc`2qQX%>0D2KBCM;~{m#}xhgIR#;yWfzb5W6!?j`9=M?A<`!Ckei!D##iw?cgQ+6Y3T7at`rD{E}r_L z1?3t{C9vw8`|SfCKJVy5S5(@_3U`V)gXRspVnY`E)5);wq~Xh%kkZz7(~hvV_eVPV z6VD2*8G64qUwvMLkVk39-Z}BEPe*?u*`dNAWu}V#32(DDWewQ<%*%L@70B6)idJOo za`o{EQ@7+TyxtIDx4Y+)Xu~T3SqHS?9vFVKidVUHB&y4X$e>TK?-HsenwS6$DE+Ld z>3?z~2Pvhnxi~{e%4{XN$UR*7bzQS@Q z=WWi9(|q;Ra&X$EAV=BNs(Zn4)1Qx4+HiPzGm`tfyr!pH?>eI)H}soy{llBW!kM`i z5yw#Lt@1PC3(LPzy}rK96ERIEM1)nQYP)Eji*K)`02*)yPtekN!-=_vAzJtKW=(@;2Z>pfeK{o#53g?EUpKxrQs*yt&_+ z={=*urj-oqa=%NLF2*W)D@};aD;qU_HQo7bd@k+5z>`l^Zyb~JeH!$}xqi8sG1dDf zgHem^{jWdPIhIJwc1%SZ1c!hcQSqC{B}WJipuade&-0Pb)_~%v~THR7^dtiC21PA-4?7AZTA|_IS`X=&ooP|Vmwar%YlDS3?4Z$I^zqk zPMp>EI+?apPQjn?VYa%Wf4lw)-mZSRdfLTT)9~c%{j2#AvsP!@8~$U7VTn< zrZ-UO`OA3aqRd)>RZytbmz=Pp0_F}^p{kQ-g4LAO%8LYpLPAP&mi*rK0mpucF0H## z%4&|JRnrH0R9swKRRg(hGC@7L0pL-ue1{SC_jP%CD=KryT5asjv1B<#EfA)~OnWaV zNv}**Pqh~Mic>1w`oRAEWw4wa6$fR0y9qlUx zNX=grdtW-tn5=!bzC9jcxUDeN#p&DEuPcAmhz|gXT^+cIz-MR07fzTHe2nc>=3``j z&8nDU&w*ih_LTyq2R#Z<2rvNmo9L<9Vwp@l*#O8HUO%Hab%w~KljC8)KUkj`B#y%R z`qLD{&%>>OVjav(aMxzV-v3xPtB89Ls}%)`OGcdbQ3VuD3gvrV7zW3=?^S4tm65gN zg{3`uw2TD2oFlVu4z=)j0h!?$Q_^~galowUgdaPv>!Ml&``loC4qr|jqa@@u|T^+yNE zO=Q$wVwM95(i2ReqqS~aRJiE4f^766uk92iS1l5sz?Bm91cc88SI)w9 z&aSFa+LG)1^f&ZvR$D#)h7R@%_7Tb78ZpQ}68fg@0zF<^lEj=gST3bKwd2DKn zro-M9_Q7w5gEV-HZ1;9xY;3G(>?AjQ-nzFh)GohWsGKmKIcm_1Lej{gZv@}}{g*G> zb~c-8&_%e&o=>)O${RwEISs~D9Dk+C(eL!*{v(stoU`ECl9horEP^v6AgwDwv$_7r z@iOT06Yi^B<(oHe(w%lcnQc05%ruC4{z2l5Q1oa!)BPkE)M;@2svN$yjo!NjvIc)F zRTnL`85NaBI(6!F5&G$jZNF(;Xwk1wbOG@mTp^5KvTL6&I9ofN2XZ#UB8m$Z?Jz}Z z-P-WFgoFcIyI=3P@x+UZ%LbI}E22uxDalN9lN%X3%tzi>H^{Zc>TF^bdr)*U{9GYL zoXz-9%PhNSt<`s2Z!`8RC(>EJqa!XIr+{T&na=2cihZjIm}&jy&1JF>n|^Mvi}|dotcr`geHpuq_@G`&Bul0%Pc;&M5bbMQm z>GvhaquzL>GAnR~=dpjX0>YmXFsz^DzOpwn1SH&~_tgY6z1JT;pB z;G@;?NkMzmn%)T;vE?PVrVP)xrcMjg-YUlA(wyO~uSXimpH`q&zEu6&oNgBfgN5uG z_L}g=_T9ibgvGeb);6wm21?$QswWHX?A?2| zcq0&k7lh82@mH>2Zc&Krs`uLMLW^Hw;(Re@M|<+>#AtCohpS}8TPdyI`18&4=PN{u z)~tnq+xtsJ(jS|?8{M!eiPVxay?He&?bzX^B*YhaKr^;|=`}knXqW!&@*}TnQihlG zH7a>;>E|fq~S{f7d5K7 ze%|r>LEz2RCBB0U=86p^F?6Wq&LCKO{o#g8W5WnvaftJFm8`oR)yum59i-KH!<(d? zOZ;sbW!{j%X{xBGY<2YSeNG(iS;Yiaz1cS#?)RL0M#f9o^2LJLDIRl|G(@R1mz@sp z+NQrE85^mku+So$1OEmrPFBXGLfqE~XKP8zXPWFNF#K0z9nLZ=Dyrw6o4daRDb2b+ zJZ8@=P|H01ehb0V?5rX$)$csQUwjH$T&SECIjoF#DaAIFor*=RZI3x7v1Q*m=Ww znO<{u+mzqkFQxSCZb16hd7B#B_35?+r=y=~ERiPvSIio9z;$C$VdF|5tCH_iXU{eU z+^KF~@CMV@TFC6(ySM!F>X@Pvt5SAdu?+B4q*;^J&@6Ae_v^IvCmVsk=nthwI*V*i zrt7^++WzB6ar`)np3*R|-y4fjxIj4y1{HPMxOc>bVIEtpc4MRYO5a%bRZ2 zMDx=-InlRP_{iuz{@7JVzYwgqs{KKJrs_Go*{8|QrVQBePMHf&w++2RTg zl4&Ee4ebkN9f?#J=jGy(-j2(}$EPJ3gsOKhS(Q$~umOd`q2kLM%aGJ5nT=Ya>RsC{ zvbqK&qc(BoDajqrC^U;d`oB0fF%$8dM^?E`f>oyKk0Fq3Xpt)UxNkm`fDz%1AgK)F zHUopbVIpgc4e&|VX|fzCbtx}5(yNjCNZR`$6K?zL*6?AkzUyxSlom-f6032&J`Rs4K zFZum96O4iN%5}PQjtJ`UR0b$2Z13-1$>rsnvI3+?x2{_*o0fn2fXD_i&Y!(~;@+`o zTUfPnxz>U_r8RBZlas!i$2jqzAX-?HUJnIy9R|-6bwCq zxF-0z>8(9F&=AVWD>AsOAxGA;0GvH(oQ}>WaUO(Y_axJ}VTR8A$$>WP9;e20{ znQU~qI&R&UEph(4JhP`a^QHZR+-ZZR`}8sWcIN!HjiW#B+}ZkiP=@(?b7^94PU=bY3r95rf(fW8k#cC(oFOgr|*Rna>!+pfzx>y z4l|w~@=G_j*Vv^qPvsO{TR{S#W1`IoRFT)(A?C#$_-{HWclbjbnl9 zg&pPGg5(m1Q|ZIn@=;WjAL$H*a;Cvxr}9&QuM>j)($w@KJ5GlVf+phQz1tU~M|IAS zrkOnZ=ry?=ZgYxs1PauoWzg5iI73=`FKDD_$4AwJUI#=t-Hc2O3%7#Rvc+$zoR+X- zgFozF>*!0Tn=0IoO=gkQMn~IQZhlhB=Q9vsZJ9QPh9mPp@_TZrLsKV5`nO~SEh2X{ ziix(AVNbn(v%Y)GRZ=T@Dd}IVVs|~EMWc=LAb!JFOxtd*3{k7goc4Uf-hbZdl2zq? zTI~5MY+vA`Pft5)NzU57z>mhlhAOKD{m70{)Ntv1{6*m6@cP|dNHeDex@T*Aq87Iq za?$~xYO{cOjR_QvFf(K_&dY7_pSM?O=l3cDW8C#w5b89*KTBwD-gRV`k&1cNUx8PK zpUE8B;bbmKwKa6<{cH+yh@}}67v3w*UK#|9XWA*VpnTjk+q|hmp?HYy--l8zwX&18 zF5RDOJ7aZqGW|(hHT2c zyPiD6TJdOA{pY~o!%2>c($Y^9JI>-!7IxVXG<)acDrZJ~h&1PvprJ$FD&wW$85C=pPu2O%@t}1UC5?cilq{QNi}Y>%Q4@n@a>-*`wk*?HRC{k@ij_1XGO>8)lBN2mN<9o zs}R&gm;iG+7qQGJ?|7|BBR;^l-7U2~IKb=iJ`_ zvJ_{TJ?_=D_3maiy#lPG=2XwrP3R9Yy=BjU{5NqI6ufmdR`^D$G!9L>!r~+GOwKCn zPfR$;CaCVT3s)QqW^5=iShA8T!|v0Ih4i&H_^E1aG`+UB@^GX3?Rw?ayNlP~nk+A` z-RX}%>@BWYoV#Z*>q2TORcf92rFpg4H;d=Z+Z}m{if%kE4JWyy#u(s1AkHVUrW+du zJ{?qTxn#CO-Nm|~^YbP~OPk-iS(f3qqUoJyDKKeg<)~RxyGwIEB^AIzc*EQq+#@FldNd_DAnU|>v5@@MXoyr z8=O15^x0PTS?R1*pT3GH5sO2Y`G4KM(_{SI%j1Lj8U5bcn@k;x4I9%ZDJ-%3Q%D_dH@!K{b6nk)x6UxL&(DGmC6@?EtFl)1B~rHRvXv!jTC|8{>&jA%C~Mhs*=kg{ zY(&0!Lob_AzfTf{1=9#O9 z|DL^jbJTWN!WE^hYrn+7y)MDF@Kv?Z8%Z0-w_n#kW_l{?2-V9gE^--)Wb(=wbU>MO z-K{u-rT}z&fqOTZhc}dD^K@$NWymCHLh5UN6PD2u*jmUklPeZpLQq zUbp$^rDBeYE6Mot@qFt_a(^5n`~I8zi!BCjYDh}J4NA=sH8>Glz|5Vww0_B0Q`7d7 zp^^f~59mB`Hpz^Z+?DSt;r*Ze63S1w`_-Gr`R?Bae7VucHZGZp414kY^?S%0B z8e9=ZU4;DvcJFzMzFK-tEwT1r`8I2qcv1hjISy!IbJWS8o#!;qLaz) zTH9M0)#dL$-McTyPk#IO9bp$h<;H_q#g0)y`2j=Ppa(sKX)5ik-SJSgfc zcO?#>b1$k%0fS)!=H56u?#Fd-eknO)?Rz0MHv*9q>|09r*OEzT38)hQlbGmB89t zAXs7-hKuNFAa}NOnJ_qEZX9lqTgTJKWi%3;_}Wcr<%x0E7hCG10xm~t{zwTo)PM;Q zi~Nxt+t&Nk&ZyR)@D++QZr1hn_gynxRBQPkS&bD%O_!vzgKn&F>a95^snv`$N!p?5 zAJfYU$fi6{8;S*QnV6svtN2}qCOV3$<4XP08(LB-d4@7y4Oj$vNll%i6q5tU4b?zt z76FduzEMlbvPazO#1QQA0`lA#7zI|8;A39%-E!6(-AW9V)60{aw2}t#}@%(t+vn3F|udC&Vu8&0KnI zU06nHvE;>gf*L$JnXqkzC-OR1g1>(q-n0x6%&L1Rd6d9fT7Xo_o(f$LzxmUI?|@gu zY_q=2+^beOhF4{VDTIZkf-O7K!qdHFTTK2txwN*!<#tY+L79W&v>cI6Rj5xtZK{jX zl?Cp}V1w<$QrJPU=y%;vIa^@)2dWh;e_k04q;aONvLYX*2(4N3bj>+0gUv2t7}Fh$ zY>|7GIof((Im~bt4oK>ev>lO)X+D8mJ|fE{meqf>)8FjR(1yFRi^iJr`OqK)`ydFS z?FsiohXWFhFr!YZ=*efV1;e2&19Msor##S81tR#;dL2 z>jHPac+>yzO4#PrE`W9h0qgQqy2l$f6IE0bYPxaG$EsEaHD z&d0MKVWV1+P$GE-V?)qi@8PDnZw$_Nh}QR-$La|-uYVe(nri`->hr@F6=Zu3s|}%V znV0KXCwM0m>LAq#CVf9*s}t|>!Gi~!)2w-pKp0VS-1x9DW+jtq2tqz{;4QifzG0XX zJH2YV2$G?652vutwK+|n471QqHj0_c^Cl+%fo%QK8N3b^&5yDtoAjJSahv^4_ZJ{) zjzyq<@##%4g+kHm2pLx`bHZXZBjgMML>BiJ<96 ziicxWmhnj2-TVggm}XrgNCC@XOqM`*1Ng261}AfbKOR?jQMWP?cH)Mj?}V#0?tdD{ zl^tVfb}Ri<3Jkg!*F zI{xi+tdP4b{>0k)HOL~!j{anTZW)_9GrWBOzwq@1?Q`F5;jk$|vax5nbzp15p%;OxHsL^5$A-gRVfEG@ zP#6lSU2nNT(CF7~2a@5193otR17N`pF!H~l(& zA^+ zUwksSK=-{C%F2(0B)IgX7V`&NMSHwBdK;s%<`)(I!pYeamhJlGv7bB*rbno%INJe3 zN%Fs;se9@uLBl<)k($I!d`3Ie#scEIjY(t4Lt9?<)~=Rf7aLsLwXgiijnv&%Roj|QJldJ*i0`ls z&qHox74F?DhWkM>m&TqzK+VBXlwLIu!2P=ORD}T)KtQS&N4U`Bia7cJk}ojBj>s%* z2Y0K^x)R|8;cQ>FJZUg{+4Y%*y@36i5uf(>D$etj?Ck7*xd(=<;vfqr?K9ztnKFl? z6jFXs8U_E%;wuNK9)$4GSP&CXw~uMbLhjDG#=XBB@d!Bx9O5NGW2$HEpnL@vvz{ts zxQM*s_JrZ{AQVSaS*wI^rCwbdniJ3p%;GUaTH1NztPd?S@^P!t0z8m5{WkDoGQ4PPJPHhBB-Rl7G4U-8!wHC`tM0K~&At|E-ZNb? zJ@)kW(L4E-xb|7=;*%k zKMoPva@JEf80#|ZYJ&i*ZYW-G`+ijo4c$ss1B7jI#l#$6AaD$I8!>Xs7r1?Tn2Gvo zjcj{@GutP1OWcd&+uDKeP$z1rF;aoqVE=~hF!h%ZZHvBUeS4D>k(jj;(aKi zx-3vAz1+g&zbo_TF>E8fHbLnh>`m(owXdq56y5A zXN%xGws#M@1RXwSEg<b?kxxk5d?X>(azyNIYX3YEC=Iy4Q@VRW+ zLkp@L4jy)d5Jo%FFJn-kt08n=6nnH{t8@D$Xb0V@EN);`b~&Q1Eor>`u~6;i6F^Qn zZbY8i`lzb823Q1PIIJ&F@%r=isp6v7x;o=rQ?YE;L1Oq?hm~_MLs5kh(R=5?y;N9x zKg+#O82qvPNYs}hMM{~;MUiy#Wp=8Q)oyRi@&Qcp09V>oL>UXj#;_^7UtcUe(d;R5zwT8+>Xt=g*B+861xpGsj|i$ER3%`2@n3KR zK$^o?pJyUhaNL0T=E7!UNZC z!ah7L?YA`@7F?YUBhCH-7^Dn~n3f||sRQ>rRXH|x;Etaq3EW%&S$dgkE?P&74POX7 zN_GQ31Kpw$p?+6B${VV>O)HI(&xSO0g=yg}_iNQeH2oTFbO)!{!ce(MoqExezbHEgx(E@gj&IjSulANY(t z5u};p+Ovh8Kg3jB6w5QpW}LD{%v$AoeZRJL$YA_P`A4~0pT(sfeO%uZzXwmz+I_F# zyK(J|?J7SLbkBLJIWz%%if>cxB3|uvb0Cy2r`Oi~;I(u0ZAxumB=)t4JL@1-3P-0r z^+bi96~V)lyQ(oKVtX*sQ%a%0+%;Gs@x-jIcprvW#_pzMv9>Ei83`>mZy((a5jSq= z9DHn(A6nDXg=*hMz0ya;<3WgxHa6Z&$THN3z4c&yoA8MQWfdT+;vf+`*j;A1u~wF* zpa9cMl|67$&hK!jNdIs9p4C3E6wBTN3^^n{qA_OvJDaw@>XkD!6EE$DW>#9xdM7vW zCD#qmX&VpHhqwZ0rUl;yRfR%IJP1X%_=?)7t(>C{KhWvrM~<|rWouP9W2#-Emvi}w zfRCbR#jbR{^8vN@I8{=OJ>}&U!LH-c{t$WzJV0D%I~%WkniX9!znVXKft66VcRb%# zyX8T&uCw{U$b`?MxF*~GQPU|-N7qy}1|{W&?{)j8--f`EJlcv;rE=uoJPQLo})r=_L+eE25Pp)sZ=i;$Gru>Z}gzYmCOm}OBbZbaAv(^o8knXURUb3M@f~eCG+Kj=cYXauv_x20P z{yNsA5dXp*aIXLyAnNE3z+l+SlFbJaZ8z$zW7LI$Cz6e?cB9|?wfoKjZK(P(Pi8^m z3A4+9i^|{wlfOm08rV5mok{(r=t8s;4kSuxIHafquqV=?jvDt8ZckObMV5dtQ1{5U zJpuMdw{J%JPS@eb4i6eqEF0dvezg{*e@di`S@kTM z*(?S5Ch0AqWAMlJEz%4G{c#Ro>@k>8me(51>g7wztg*~--h)m6c%)DMMLIH+0~ks9 z9a6Y#>27@nEi%l8mf6LLwg)SwcV6Q0)Ly-A95~UO3$!y>fFF0|=2;W9}`a#v5lyrzIW$T|~8`oKa~rH-7qQBzx+C zM{n7?8sou^?LW$=*R(0&@h$M;8zXPWSySj9k*$}x8wJG& zkq@3~BKOMjFgiH`hOKa5?)}?7hDaiMr;d~M4omE2$c%evRU!}xDN?`^mC$ieQbJTk zNRoo!YL^0<^%Z9%*=7jc*U+7RfHEo_3U1LMXt*D($qFM)t%~#{8~_bw3s@^l1e{aQ z`=A`|juKGBrZC7IICTp|Yo`m`+&Dlr3up{mUJ2P4p!Y#bebqpTKmkB{YVl;jHr=PT z=yuq%?rMc?+d`srywEulgrwa!y}%ewKvukH{{!kC z>TsmW2hx70e?CBk;SiB8RF5DbbdKAr8sESRiW8ed>RTa*_&{iHs5e6I+EDh=LjqcD_c9)f^Nt))!n<6$o_3H9`c~ zINTi}LdT815+3d1XV`%dM(qdl`sM@9>yh52j=Ze|;$?v&WqGCD89QgU_2SA!i$H=| zz{RE8(qzPPjn}fA_s}piAwli}<|l#QXE#@OxNa(D=SFd`@L4R5b4JdLbQ(pcZ=obZ!F!2*<

7htuSR4%X_e z{dj8Sli2B!I11*_6k+o`Aw4|uB#27o?wZ#-f(6UUz*1vU=6Ur|ib#CEb_*GMg4A&j zCd|Nr8Wi*HI`*X1;ur^kbKa1VEC2z>`7;hO(lPq9qt^~6Hw>NWZ)1*Yyh7khA)5F& zs870;DL0uAV;%(oyiZ5UZv$Ea#L+5x7!g=fC1iPbA~m+~T3WWSIPr6Qv_5bv|J|!* zawA0(x5rT;tgX8m%|OZqu)e0DIeC@-IT9*{dfqs3#r-u-aIxRaxg9 zsgrFGjo-!vvIXDrGq$B4Aj{FXl6#7aZS@40S*G)oW!=|SDYnE7zrv+XNqUHNzuxVR zW_5PJ<7CyLa?lj})4LUtwA0z7GUEo=vT}UmN!*nj{h8DRN2!N%lqKDC@8X*?er~)aRS(DsDMo1u|0p@zN zU6V;_=|@^BdGybjQ;^f(G){e`EpL8Xw#-7S^;6WzmLOE)8WNA!zks;~ZQ+e)4@|%y z9Of9~_Zhp>tA8PU!7sC;tHV!uvL_zG$Mz*9iK&&5C630=C-d3NG zhoLu69fV615Q5t5-9D>%-J+A@C4)E!61&y_4;sEPb zu(x7%=_nq@ts4RrBpz{fkQ=>h&a1m%C&Cpc{waE&W{v@syRw@uByyt>*7=M*VRk&_ zD@jz74Q)73mhdU5!Vd{E{QDh*zg=8S$BIbG4sz!)5dty{l8O;urwfi0>c2)|jfs}W zX_`u{jN?dOqt)kgyZhIoom)J;P7oD603%bX9#yXL6qG`vs^gplN?6&p3O=CNUY(u= z5G59g9V@mRU?tVNn6oAdfc25ag7H^GN$m(#G7DbL0(VaUG>fc{N6V0jh&TqkPBzk$ zW#}tma0r{(fecs#RE;DMBOCCVN;MfP<*$b2qN_J3yTucfZ*=row-)3#Tb_lCUq<&+ zA-#A0J~v{JK?mB{2?+>1kF%`??4U08sgsSE;NCn@Bs3w&)#dUwHQ`|_#%NNmBwFTj zPx~OL0vBj)*~k#Iv2y&s0y;Z7G~@$$r9tGsMw!<{s1Mp$FjF6lTIYFN(|zWStJbBv zc59BJDO=RX*RpLfNaUK84V-0Eh8_L84Gbuf0LXe=Kq86GYHy4m8!9B*IIi4w>R^uC zZcB)2iqM^C?mv>)@Q$dR@!Q}Kw3>gb!;MC138i!0jxA;{=2@JXl5zo~Gi&*$yHoYZ zTM3ie$d)OKO^?~XT&woCtnw=$*cjk+SM45sbd&KImkOAw^Tnn)O9b_s6<0tB&X`|N zhP7%8#;XKZUW~mC4bTJ~U>z97{3nSSYf+B2-~5nZuIb{x;XX-%{9MFaf;Qi9Y&{zu zPi;XL#)6hu+};kACG`EG4{?Z} z6vfcqx7Ahp!fQ6(-mCiVZq894u4zr3XiEs9R@Iz&`zQoC(EB)I!S;W4$eOxC=r11b z*c(37nOu7o#2Z5Z_m)DQ;_IDhucI@Ah-61U2g0$R0doT5+;l|JU|C-HA=olkYtmla37Z=;NwYA8eA4LqzGd)Xt#!9Y|AYM1S4rvg37klDf|axri!lxp8Y zulaJmlYqs8@v%MC2qNMSYf6cs2aGA>7)@x?NH&mhEr@HFc|2WK{l<+TRq7Qg<17OB zx)H`J6ux`IXJY?7Lk;SWssSkbe8-LuF{mJFQx`LXM8|5_uo%!unFbO~Q;FDNrH?Rz zw3=EoTA&jndE)(^jaYl{!BQd1E8W_aAE$EFd~o3pq3!&x>Ll7)LlZANfL?To{6L=B zTT35qXtI}}Y!qT2gs0dKI3$Tnd|^(j5cgK6J9f>iEoCDsh7?GB10a@WC6{A7FzK>O zi`Lx(YmLp^FDVQzuXA@kgaq_o5AWZC9LA$Bvo1+2MHOmTj{$lZLMvbc25CumoPd07 z$Be_kal_#?pxJYr};8USXYr*7% zmo>CZ;@tYNU-xj12A^`S+O|It=e@bC`*@1(fy9quL1SlQ46Y@{smwRe`mkG-N62|$ zj$-!{OfI)8kkQ~feFW2M?hz^Q3rd|TXyX(0IcSSi^5V!zXag`T{@UFCuv`l`zi8`+ zE@c(FyK;UY4&8&lTFxoG3T4#jQ}5@<`cCBKUkx{dE2d_JZCF><7ck^I&i$c;c+?{M z(*<~a5scJD`apF#>J->I<_=0VO_0$9^x?Wh{jr(+vw7+@^V*bp^Z#7RzxLKu7IGl+ z!0LPE9RcOH0mAy3cud2{0iaHq*q`Ho2J4Vc&i@<@RlU8VLnzy3+r#+Vk5&JQle9&& z4Ig!fPtoUaPBS$+11q$_-?|PxZEZ1qE%}Wf2Qk`lH}eR<<4k1IWlp((>}_#&>mYZ+ zZ>NPClNgwzovjlDf~Ha3iIsufvPE3;h8*wnXU~o!qi=v*{CyG*u*F?`=eTn5;;F{j zkFC~d@nnDN^{{(;48fft^3-{SjwBKQ+}s*zb=oj??Ks}*O*GD_>zLROg@+pgC&6IS zKO7A|(2zHoPX4>&0=I@I3oJufa6$(~YXfeT5-IH&F`5f~`lJ3G$$^`Fw|Al*Zw z(0xj9&aXP^(UPF$Vd)1#vas(0T*PF^6EuJtnMpQs3=YWkA*u!SUzh;Zs9lzk|AzWy z-A8>AJmUd~8$w6Qt#t(buC-y}#tBhiyZqd5aGtPCSQJC-sZIwX z#cE*P);FyX`ey)lvcYvU=v)Rw4Q*v6sI}hIy|+Kebrc;NZ@<=9%+)!4>#OGvS(W3s z9L!D{!3b8mwIP76Er>?d_hbOXw-~KQ1k_*_s9<^O8#Na+U^hb=Up#+qjf~IBevcK_U?6NJW95SNsgi&UBj7&PTvgm zA4`GrYQW@(2RK_s;6a+iH3|v|*`lTB{xDWr*tZ8VJ`}_mguOeecw~!u4H@b(7w~H0 zKD%YWGx0K$#AJS`T-8P(+Kk0e_5j zvT^S4Q@8bfiFWmA)IPP><3VK`9ES~trYeUo=jMJeANumtYl1qwY#&@i11x<8DpJQC z^FCcn8;|;-|Z4jHi!H*w1#5)2oxdDG?60JEm%N=y?Khn1T zF_FH{xDO%MI_RY&s*y;8vk`bS_Q0DBA%f+c3wk-R_zj74X8#-dlHjN-U!T+tEnfEG zkhaIY0#pVwNhc#0qONfsz)Wp%!D=Wl_3-jC8fwOHRI`6ZA<)%8nHgGI(1*Yea@Ml0CL{2)0S>HxC7kQ=r9QLHDOqnM#T2hEOhHWb#1JT z*D7{Z!10=2fyyTH04#haoes(>hv8_28{Y}1=h&Pr?=1_Os> zCefFO5K=ev`J%dw>8CCB`E%YILVLTV;Q}~ABWgKp-d<&LIug;lHSvnWa)8@4fUGb- zJ0HSb!-Xc921cyqMA6!-%~9dH&y+Y;A~n0fp6f4C268<^7Ue*PT@ z1J%R3v2cc1EJ=k)F!}E1f2me&z^#6m4@C>GOPVvl&hKu~Lt{f9VB{^vg6cIHux1+{ zT38>UgZ*xMrHxTc&mWoo4dHziAFPZLLErcqqIOB6GEbb<1f>*i+ z3Em@+pWB2>N}65DkV#vQNEz=T1iM%AWn=V#?W22SFSzJJcS;m+#~bzMUYJTu)Hehk zN-zj(VbsD9AQEQ z5>nmGNw=Zi9R^NZ8RshR)da`Drd@`v5<&Cy7E{~Py&aze!yd3ZEb_xCH$IB^;L6= zLEKcs&;wM#!dtNoWF^~B06>OFVuYn&___HiV!9UPNWfWE6Au{{Q3Yj9l4Qfn3CPJA zk~k~+w63F*HMdW40sKcWd^%<2Zocw}RER{5>aQrecdtXMO67;VB$0Jd1xd52;xD>i zm-ZeXOl(~0a?19{V5uu4wNF%1>(>v0FkpKpVMU}yh`fn7ckg=H)UXNO@ka2I)jQbA zl)@5$2yro|DldfB^l%-72*4A5h?7wYgjLMP_wr%^b7Vn-B>ypy1xgkrUKEr-YiASK zNFgwUC~RFpC&h;FU^*qP@Cf_1Ejk5O6QNg_`m6-1s2HsVy$?#}?HEb~!YFV#C7{OQ za1~OAv)Jevv8;^Thz!}lNAN>p&)2`sEubQ^q{pCnb1Rk&+IE|E5d1>n0;h|YyP$xIYI3A?KV343CS*hkFZ;Jz0O$qko`Kv48@=$_@ z0(j}UdJm$R!HtLFs2GOU1s~YH5$ffswH6x;GajITtVAlgk?~Vf+f>(p{N6g+CYWH- ziAGg_AF4RC$rykvN}ve}mImXaP70v-ghD`W3M_In$8L#&Vu{*=Afx~WshJa$2>}Q| zVJa65qWB{wrUTwJ4IVl%yvX^t$}S<<#UKXGTebZ#MOYE_@&Hy)#StM-A!YaWAR*2` z^%nE-i4ltp_d>YQ1d%`(wqRS=UX)xEyyz7GIb7&{R*&k-u*poMoTbAgTy9X>i{*og%V<Q=4T2DXx}=A#(Qqg4V|XH(($aKzvMJ~Hh=0TbZRJO9iS z=fD@25^HO4M@i?SyPt``LfsQ{9mr^J`}AyyCR#&2f!K-kuoS1o95g~!Ee8b4%@Lv! zCoh71gQiB%+{Kk<7&VrL#*y_keL5mZ2lq`Dq6?t}nnQ7iJ;!mWNuido9{>nI6t}Gf zBIb-X6j`Ug&J1#^0YbIhx%#$Bzg&1V&&u zxC$aP>bwHvchBhP9T-^R^G2AgseyNfsS`lb7lgGDX-!A2FRHg-&b)acINlrdjy;N< zpLVopR)p2Dz)P2CSi5NJeU`YPEutN^>&ZN>wY;*76|Xg6Wy+jqZT89|)kcVC*`hT) z`>|a>bnUL1e#KJFEIWhT(^U%;zrEDpyNqN$2paYxBvq9E$A9G+T)Sd!bBPJ?Z-f1R* zfBwM~r~bYvLBqJpx^k10?fB)*zH&`1g=e=g_E3a9vL`!v69ket-R^I3{y2{r?`Do; zPHLSIz9YzLt>N`Dzm5S{-e(I&P4_rzQf?Rp*ziT`b&8zrVas%``meZo;RsbF6AO!q zJPT}Yja6P}egryF|3T^%HP`x zRfG~z7obT1w zdbqwBFnWf@7bCY$Z8ab)$A65F#231ci+W|X!RusT0d8$KA*G)_L>_Su^~$e%+)nHZ z*M_%bFI$o15xU{fpD7mA8DohP=7&=&s`YT4GGG884Q35^s(M@jy;dIwx4tj9s^D`! z113Kc>Wp23Ln9;Er}l1aS9_yw*0Fw5-rv8|L`z`g!ccv3uTsRoClX+#PGR&6?8U{m z8mW?4hcJ$4O^B=G2ItJ59}2*w3^|^9$7ckv!nfVC&Lg^kzbIeBm>-AHe?oW$vZ97k zjRy@;WemslwysNwI;Zl%8;r70ARAOne4bv9T~$Njbxq;7!K;{bdHE9YYx++kG|aX& zs5k{()eaOAJP(@KLMyclx%|or^BKj7J@|A#;&#@{sXBw*a}P2j>_S;j_i}BdYeD#o z`Y+G)z=O9y-bYte%K~$ufO3xRjnjw{hBt^CR4Idy33BX>EiK~m8~>RBKRbxcC_}LR zytTDe+{2JzW{v!bEH_ymt=nj>J3)}6hj95UV(kz)4#(W4)*TF6Nt_UpV}O%-^0A7d zV_b9<#7@aA(mF_UJE+eM&OYL^lN-Vu_lk!|G^rAy0=u5n0+zNLB4!n7ySJvE(*FL4 z$^@Yh1kn&crU$pr*#^-l8l$X!`vWwO(WRqZxnIKjA17JYuIuWfSDFT{Vl1pOWB`xX zx+B=RLC^%mqZd$zXvOVK&XeC7A^4n!gx}qM9m0U(N5-6Wx`jTc>oMQxOK*Km%FtSmx}s*hQv1lE2@7aB{Y(xG47v-ldrm)uhO&3{#WKN zLO*uvnRx6a$SrZ8>Ygx)n^exFw$I#?BY^F_6w!VPdnAZ}nE z4KAekA6QG8Rw=(vf{w|aJ&^P>_spva($r&pgfe-B2k@(XS&prhDtq1A6UGD(FB6?6 zo;0K^s7%}(7>Zn~6rD2^fh;fH3qAb($Td0$Q$+}~=*0NeMP7!W6*&v|5#<6cL`zg= zo_x1_8XCq9v|y{skkDFyEv$)Q3R}jCNZ$c5^_D+Is38k2a@4R+Dn7X(YebNR52CV0@)4x` zZQ%Zq#!;X~6$Y}~TkoT;LIkNIpeX^Zumd(U41pj?cRz!71($V0i7fseSM3qTpb+uvT{eYZ715wBDnxmhj5iwVT(n^Q5LMAo&`KOO^Q?Mu0>TwJ9 zw)jq-25@kVvbVtxRRh=krtj(J++(Q?UW-S6&WDwQ9>SE-A<{A?h#k~)Q$vA^ZNFxI5 zY)br~2!owMLGMqAQ^os{RZO*}5>mmpVF?x;U|Q}F?;PuV*|b^miiB-?$EDLD-1Ad!gbta_RUV}bYGeQD5s8thvg<`G< z$Lkg$cn2zYcF>;zFGVZtqm8#Q`-t`ib$Bg!xIc_O^7XHS%s#cB)r04WT&lS<%a?2i zBCRy@(#7BrX|iH>H8UAsl>=UEKe-C%!wq1^sxWk-T+*L0{s@)MGQil*P1`-*UHGjX z)wf_6tS522$KuERfj%fBSWRvz!26~_6JQjfT`Zs%Y0lYDx1$RC)Xv~t-@bNtz);g( zpr|OM*_@C?C?Z6@1m5epLtjAeQht7M%xMULM{QWy9`ev=@>Nhw z6(-OE-o^533-pPorbtP3y|j3AFc2*zu#Ywn0*N+5H4rz}0Yq2^aebmz043N1;I3Wnmc2W!Yy|Uh8g0KI(~iScJH@w zwwAiU$*3|LiUJKALFH;(ryG<;qGC%Kn6Bd1HM{tt4xyaVjyf{w^&}Sm0<8iO)Y>J6 z{ywOogT5pS);htB7{*hbZzp?FvjfQ<2?kL9zK`GI+50}y@W+wM0U5u!)9Z*tXTn%4 zSnssP9s~I!q6NM*B!f~4xHg*p019K!*oWt+&Jfdv&<#4{G!6i)s%P+@NXKFUpA#c< z6jIIYu#rR^wVCxXNL*-c9 zs>RSG+6ddd6nfRvK87}}AVh4+5TRkrmpk<8jBTy$*F9?PzqlUJsVea`D>I)-1d(22;*L>x1%PYBmD+lcWb~}>#nA5o zjr~~5dvM6ooMQUBYov((nuNLmwHgR%L|#YyP=a8a#Lcjg>m?)(5?n!ZDRACYa9=U+ zLJ(3`?SBtiz%I^2cy$drw1-GnA6pDf2z6xX0f^X?u^>yu59`wyOo9;LWy>fg%6>{8 zP)4KGz_4@11lSzUp#)Sbh-gUpu_n$d(N3r?X)tT>eF2a-Hh}+|;>q~zv=ePg9b@>m zjd1iqxaml={Vz2wh)YB-K9oFL*GyXoI57=b2MN!S)ddnJhy))CNB1+7fE=iLwS%j9 zgG$>nXO4JH()2geCrpzXt6qPy-2QahsEOnstlU1b=j2mr8k)^$VSdT{3;0)~&4 zStYj>CwI#S?}KO&j@ZP`#tDRl&Y1m#PJx9e6VXQ?fmg+`k42tdjphyl0FB3ZAWiT4 zQiDd;M4~aD;6otU?~y*>t!TCeh4OSBaY^sN{Kx6NFX=*bPG}fyI!vcN6})f7DIYv| zoni;FhltpN0RSjSeaOIk74#U@|6uTeTuWUobaqL?4A_@40x)zj5f(wCYpCvR4R6xy zFi~CTng?S7->=5XAhG%s&0z!|rC`c*K*V>!*A!}y1Vo}g{%BQX50W<$4T)71LAwL# z?kNjM79%3XrDxL(Ksh%sQqk806~8sePHE#Pz(RycosyJSv8A6OUvjlZ$S;xKa4!GQ zzWXOZx4?v~P-?{ka4Q7w2{4fxA~GS!<@TchK^AL}z!aKC0xMZ5p4PYmWEWDX>#hb+ zN3%@Cvu{{iVqJwOBBa3=2w}yk!3n+Qa1}R@l5E6NzfLmHg1Z%<)~}Rnd^lmK?$vgb z4ZU5d*Sv`iDVbdychZDe@Ue*=ftp!?_>&H!GM=vK*qi6LN^FQ^*2Kcb9-xLaq0re* z3J!eNFGp51aLBx^!|`=1mRGp7oi~`Xev;f^BTu8cb3_7 zd=&k~&qYf##h~xMWRX7YN%X|G&j*Ypz*t9&7z)7zMd01?fb9hc+8^u1-9hADuVC)B`PyG!XijnGZ`lx zl&2*^A+ye8qR?gzl}^?da%zWSE+SKSEIwUwnm(bgC|3!he-*4^Fu`HQ@UnkJL+pB; zu_3&6N6v?6mtEFBK8_>B4%p}*7@!}kNpllPj`Oa|98V;119UY0p$UNWfno$b(@-Gs zakv^+{CJ(x4#Df^$Qlwr07yF9dW>uy@&R!;k(cAiaBP}!YCP$CXFB_nlg&vg2bL+? z($Z3YWc81aKT|{*0&y*14<1>M0p(PI72Z=L)!a(BS^D!C9Dbk0cLvNp*?@y-9-EKo|Ky&avwqb`X-M*s^##VM#`Kr(k9)@NB^;YMO5B=A z_K8D8ZWHr`7|mF4+^U+HR zcmLOU&1te#xWKXS+nR6(RB?bY-c)8!ST#UzVDrle)g4p-Gb#+7uWxf7pJ_DvReUw4 z3f2YqNf~a<-aSMO(oDBh-w1IvxcEzO6;x4zKV3^M7JfqQa|DL4B0jO$T{>-K5X@pq zuh*?z3loWwFlLdUE5aybz!$+DoW6wo)$Vn3{5WthmG|t_p8aFTN|b2&7Ny%*R?7G= z$P1=QI|zG2Q5L+%kB_2l|`6VibbZhc`@WTU@$Lw zC?HB0e8KBzZq_|*Wakb+4q3DrM9R6bKlULXwLqC;5WVekceKxpHNrJ`W0=wM{A>h3 z&}ezM_3@q`ug0Nw0)+5-sFL*(b;TYBXC3+x{1CysohC4YqvteC=kmgn2=XDNIq;;K zqXhcv{sgTMxpXLtwwh*BL*aJiLCGiMi>}XkBKXU%ORwY9sK84{883YtkAJ70#-%bk>*H~oH4)q0_m<{( z90Qwcm@zXfzZTUjM6}_3x96_5GE17f!`-GOFNYok>GJIkukPJNOt8Q^K%y-hu9^>^ z9$1%bqvCI29)N7R({oZO)J6y&HCT^K#hLx#{UiZ6bXyTzwYNxLpEF(Z?8@~30HLQn z*8s!ez$4f#rH;`FXaHcNbJPNJEw*LO)FP`M5uLT!7L*I|E&GiDL69C0b)(hqE!_+^ zp5&=g6OB@YGWO~1+4PT};3Q5k=5QoE@*Z`@!f0gqI-ru7lf${Xar2vBR*R5mCmy2& z2bKnEWo!(j`EgVaLMWPciudbFr#K)JTRXVP_D^Gqm*7+)Y!{;Hd-~PsOHfnk$grI| zjhW?zWPUyBE`#vNs?cMuZ6sfVOTTFq=2C#T_G+5IOn&j1=F?>_fr#f1t|AFiP$)=n z9vx`PM(N-Jz-q_;r}gqB0B_+ z+F%&6&y905T>%V7#Yu{{NkxP#5n`my-ec2bsb)bVy{PGd&N$&ObSfxf2WN$57NYE> ztf3Kx%FI-i6OkrH;D{mP4n@&!Kgz(7<6Ahgalx8&AvLMa*!I2v+phxr=pIaPrMODK zlTWu8*Dx!Qv>bbnUAnZN_-{~OfY`?Mwj9lT4vqlQ+@;_){VKV2_Uq4E{NsP) zBz6n_&p$e&jE()De{^#XPT~Lf>%VW}_-_gRTLO;%mf*i7fMNX47W}sa|DA&WE&|0! z|Jj27mf*ir@ZUxF&lb=M{AUaPTY~>i!T;4oXz%O&iN~A*3xJ9i)Cq#><$H7vuf5Sf z`J*cG*U|rl(`aeVkRVk;Xc|hXLy;NPeN?JE|8_(d8)i*eQvcz^Zer%QL1i6F4f%!^96%gx52Etk7d&RDY&^_{>U`nhegT z;i#IhR+lDOK2@6^Aa(1xQ+EZGBFr z4Lrj1&MLX{JS@vD^QNv}&4@{|tWq-)x2*cHMt`Zyo~qm7fzgNxS($N3)K7jhXRh#O z4i{S2Gw!?!J5%m2Im@y^(#z7aEMdlvb;)0N{l{uO&han1o`|>OJR$ah zh3S$tY3?2_wY^N985%zp=ZQ&Pl=1i9z`^lGIqDb2mf1DOmzTd@t+uc7cV^T?V7{Vn zK>MSCZ373IWey6lc&{n7a&8I=d7i44tp0Gl+@G&KpZl$QwK((2wGXTG)zzM*D<=6K zd{*9jV9B6koAFqE>QL_sf3ZU$Qi*)b_xlZvgajULxNOINTIOZ*ajB2uOFx#gy?reFSk>&lp?eAwB@5Rh$M)$sEiG%c9U{Qk2~W~Ygk%{j8-;^u|o zO%Hyl;ab*jIo75zOLALle z4*e4Mz4*pX+^=2w{h8h>&i)BAFL=V$nzQ@nc=m{&L~~P2%IWQ`#=aV-97l!}XSqLi z)(?)$_ntJf3XDJb@W9Xe_-N$cRqYSrdy=vHX8(@p)6HC_!!qWq8D1HJ0jtJN1kWmt z$Sv%(=u;gzb4;(dL&+O0(+0Y_ul&1cIe$||y~xQw{IKjxFJ zbwBUYyjO}B^9p}SS3KV>oir`jsk!BYPo#`ZmFuQL6EDFvCf|R+)cfBD%UBWV61Zky zfkDGN)sa}BrAxEc$Uoo)nn-y_;9$WGLa++)ro3 zNWxm;nbI5Es@}qWla>i*&Wo`NUPOFO=luSY{#-u~@>`i+j=1LE>sy_>`FFOgx*jYg zvCC8boYVNEw0o7V;_{*fMqk-b1Eaal@0!g0pO)#Z@z4ExWK};+RsMXA$=ps)vy$fU z#>=alyBY(EOZc+#>W0;L`+i=S{+|nSM&Y@X`u%C?bv@0?%+nt>PAeaJaecwu{ArDS zCXI|fi{>L4RzDn9?zx{!{^FK}frk8@quevrsOCrX*bR8`_A+5oonekyXM+*HGjrPRnK^ql}h3o^%0k|5|ZD)|L?zk zo+;Dzda;}Cvt5#xtKpq==L#mDd8_&cdv&DsfkF+RVW-EBC)>vTa{;+!-S?OC{n=ZX zj}2D(yedm`va0TZ;X{S=NtTz#SQ*#$+~SDOD}Vf)^uISs>36)*Kh+ZcXBRdp;X{x6 zyDQq=KQWrK;!U*mk(65egH;?+^Yg~o@t^%+O?rQoH?{VY7`RVIDkt1uRb<_ZbMj6( zbVi)tp8A}L&GJ?J{$`JE|9e6Lx>xP)=i+1Wjx|+yOOJ359zB|xZ@Q`Jw#NOKW2p}` z;6MHlwV(s>;^K$Amv{U$<++oZ~OrPVm<(k`TR@K;Z^ zG6+j=0W*0imtT~VrF2BBg3F%AyF+ZR$zZFr4Ch-9E8h3m{@t~iZZcaS^!@4iFE#&; z)vL34io2g-8nCKHN+(%2N}_|!W?$@ic5JhSBo8w#e%IYQx`|sJsU_A{qaN-?9dXe z#$3zEwx+6q5s^@pjlrcG@U`^U(zZ_<7S1k76RquWTlVWO_zG2tJ$oD0TU$I_M?BmO z-Ch@Vt+_ADz_)YHbNJBwTu@=ZS6qcdRcL!2*WZTUKm1JonV+{s+T0*)NrBWnv*R}% zaFkE3zN(w|sC&GE|IGUN{%cP9y=;0ad@?4=UVnz*yZBc%SPSv+sD+)6EZIJoEK|unxMyu@ zM_^)c+ab~ATB6HqK#6?aGZZ;oj*}1?8hX}^)tmy=vt*EpP2bLg&ZED^<4V zn#{lgTUPjyhYG>hWxjvD63>t8dge4cP4xb?O5F47v3v!tM_8LpPJOBG3&tAYD^$gz zD6801qD#s&uEzS8>1X*G8g#bbK9F-*17FoPSMT#xiY~u8!#??jc495QTp=wnIzExx zz$kk6cYuT9MWaN8n$!8dd83bp)r@?%2G_0n{sjKhf1bs$^L*}Uyb|+^xIJ^_&v$n^ zJk8*pJ!hwVa`HbuD=xmvFkOCiZ)RiP{7Lnbb1k>=uZl0;vf`SJySr|Bdb*^#R)@2} zg(vgRmFz+BQawj8u;vxa%V)Hw`#ch3M|SHU3aN z=F8{rk<6L#3qQ`~6MoAnU!Q3x(YMgj*K4hqYfRPDvA3G%k!jfc=oiM)Uz=-x?;Z&( zcG&f>u4m>x$QCE+CU%@>Tz&q2M~VK8u<-EAyu2i8to-nqZcs#paCq6q#g}{k%->N6 z_c+{vSZciIOQ6gLw;z#|y#2P}jPLf>RIB&k4f*r+zFu$$6?_HvtQ}n4a*c7?h0fP^ zyS_1Q!Chpk$xSQQ7tIplZ$!AnWb#y$mzPMX>m)*)!~?gY)+Y08hgW`dSXfv_M#g&O zE6GqktI zox>%0b2|)?bXi(PUS6;NnpMlCdw+vblP)!vl$92{Ia1a$1k57MINk{02E_VrQd~jlE;|Cu&@BeoU^JvNNu!;WE z`8cCc;_x_wSHk*fQ-oU0>-X>fsQF!1^kA1I=iPL+coA1_DPwBsR-~~Ozj`se%tZRFD@b1Y*yD)@N+u)F+@Vy7uT_i-aSgI zPjb$Qlw_7iui{ip`5M9S%hzBXSylJz@j_KN37i%sUwV@4+z=Qn8o-`{w5 z7X8lb`9;c4;o@+Q$ivj{E2uyBrxh_i_QRsDUfIsSb9+*AvTug?w7aif+ezyPhiZ#1 zho9j5Gt2h!p0%E6w^_#Jm+dh(yK3t!$!5_q&3u!(1HNBt)E_SS^+HI6WhlN%&6m!b zm3?CFi9+cWjxDW`%0DAy>2n5h-#=ac8J9jH^C3ftI^Ny=1<$&=hddLsg*9fUzdCOB zSYvI_LAj_mZ1$C3FZ_P5*0go2-{z6&fyHf7CJj}^L{^kSlj8sGKW>!XW3+f5JNEmB zzDrJiVs!|v3TG`q(o`7mZ?N+-(teWlk%E?g9!J#oA`0gYdLZ~Yjwsvv(Q`@>m?bhPr+y5W`p^d%SuoPbzX=##=G_o6O!OWkM za!*uT0s{|E791Qr=$S8nT1Mj8qcfc_;$xFCHPQ8}QhprmmOVdW*}taY3axtcd-wPF zzE5KL(WqBu*=u0EmtEb9Q*c(ozVLr|5<|PpYy7s#gQMQ;4rhGb&^6NPbFxg;cm31D zA=6MDUqrQS9nh1Ga4X=}Q4=H zo0_3Saqv-Y1U0Gp+PTdpCq1zJ8K#z!JkJs5H7XoW544tAQQO;_WNBq(aVw79k%qG} zIGXoy=Aa7M z82t+^C;P_NsL8B&_2Qg`0n%D{Z<|ckQO2#L>R@3;ADshYTUim>{K5XNxlNb9%z&%hj2@{K^U?5* zbfm?)HKRD9_}FHXY$dvMhHKQlau7RYO*jb#GNyL@N&o5}9%VRV0`F-k(To#!5+8AF zpHXhJ`w9uw*c8JW{x0D;QN0Vd-HfI*R~BnUNz_$M8u`15F%+B67Q`k14rU(Y`rm3U=t zbQNXY+m&}eKKZQoL^0wi7Tp8BtNb#p!tY!94KH0ff~J-j`NHbF^YJALcY_&wG}$=G z5=f>Y(~fM^RD>e|l+TtGn=P6~GN@NdyR{=+G7t8dFn@x&8KBB(V%_oKHb zJ2~#6e6f04Nd=d-Gn&_1dCgq4%H3P1V%LQ3TsFT0lZ>L6FJY7)n_NG1``?|_e8R4$ zq0B~O)(&k9s-LK~+aVp_-USm(M04BcXJ%7>5cBMd8=l7HuJHhxOhEJ1kN`BNZY9T+ zq0m0(C<1ns35{n0CG7!Qar|<4M4cy77w4UdJbOJqK;*w@=65x! z`n-H^^a67B1+>%ImfpXq4jqRXUS@MCqa0F=-Y4d56LPB@~8|DWg5q{WfM3?alS>Y@MupM)xGRzzwKY%1*^OZtS|sJ zY%hTx*8m`4Tdi426hI7xu-B>t->^PlasB$EZMDnaHQEl+;=&!qLG~Qz<+~vj5OWj?maRg4wTO*B)En@n9@toO;tVC(uu&g>HX6}e2 zHphiM#N)|re;ohf0s|jFTbWC8Zz_!Oq$3(=aI#200V{E#XC83~6={ItqU7ZV)MaYL z%9AZ=nIH(LyP2DE!_2!-t=_l{G?0K-ND4n6UnjDfuhAwEh>#ssT)6b&iNrl`v%4`%C&Rp)#|Dq7JV1$W%~Aki*E8>^BC|an#n|=nm3C! zUsc0y?zGfE0dtnh2zR~KI&A+GXPsh%6~I7J-4!N7f#%05qruHeMMVW&)=LET76KXu zDvGrf>&92_sb4fik=mQGDAZ!Zn#$xZRLqDRqUJ;SmwS5)ypitD-8A~~NhyVKqkY5C6@ZD2 zQjM8(3)iVNx!HmJZOlsdJieU?vlhqYEoU@&Bl~(c+`6)BUF>{aQ+_5-w@g1vtLbW> z+U_>5jY``XopW3ml~t;*nu5ZQxJocnkrPFe44L&HaqXkHJYSECixc(mp^QAYCZ|`t zq4$Ld|9YRK5tE@IKxIvta+C}vyZR|c0D##To}ATf<#y;0LYBTE#RwR>YdC#n$!_05 z>888gRd0XSKmrGX%2+nPyIi7XUGjj0HHc9sg3xtBdoZIw|GUSZ(FeG5$bWarth~bWzx*B(9Q_Z!_q!q%P`Dc? zp9UP7mtFuH-I=|9|L;~*plf=~#CZ>)D6#R^E1c_<*+~GjxCbAsLpBc5&}64dA3c(0kdnzN0e5X>BKq4%c|n!__RhyhQ~AU3aJG>Gdyy0SLEZUzMd%&{_e zedU%a^Gla<7hEjg!lv^CR4$4D9^`7uXK5*gDSuvL-e@Q^ud>rpR6|YeYi_Rl7NhIe z%NH1Idn~6l=uutbQ8c`Gk>N;+!H74(FDKBqtYEBK*T6%}rrcE`+y%cqdJ}vSP|V)1 z8+(@JnbE1C;Qs(c!El8s*MQ+#HvhnqXE+g(53uMG2Nk}(Z((PMC%1BF8Td5(EDPGw zz%}8~OuPQkP9l-jzxk@|>fVnXYo0#l20}~94>120v?Inth$>b0D~;JMoAOQ{BVGqH z08K8wtsrRbf{?l7dk7<{#dU4)y*l>vyJ?} z)ESY^@XlHupUXy|t(fd*g|SuMA>-|TUC)0Ip}d2^egd;jt(Tm~eY1*U<=*j#;5Qfm z8WhK6&iouvs0tcdT9t?{_>KysZcn)bTCv$TkfU$ z-Z_olI9YuzA8aPnPPC*Mez(7TE(7_{6E^B>F zi2v&CZ=2S#Tw{(*uao6^jLka@-u!348<;h-RK)}9Rq#!~;{|#B4Sdsk>ZaJK?`GHEL0s-j`kJF<`U!ZJ5G>|fsIh!11_KFJbYqdBP`UxP zHcPbviu1pyLB?{){pyOo;EZIjs$_6gMqhRm)cKo&-ehqv?dCyfvV2r?^32jRV>!$ZSuJssSwWMK18JxP3tP z6EJSIr=x4uB~oT(uPn~=&w9#uB9wwvBFNWd0#+}aYb6@!W(BgSt-W1XN{TWcK9E)FVfx7?oC)tuP@t%3|c>*4d)7om5R_vgG zLqlp~fpvxxfdN_?N|z3#o)^GW6))#vV|#lIGMTIx*SORGU(#*m95k}-qvgibaITUf zy?p~zSr?nhRPWuS)&HT*gUd!HMwt?F^p&AeVmy?EMIgU>T8;#1+|6$DHCxE=?7Dpl zSfl%I=NOm`3yE2O<^dm6f_3O%@mi}NsCi7C;Gm;}3qjC~sreBF% z473p^!k+iH_1iul9a?^!G{c<_GHF{X`*aJL{J=?|@iz+0D}QmC#+|6a$83+p8;QDy zzkg-!V$VEu1}v^If6bp`i39*HT={^Hrxgnrd_7J47tKuFXHe@JB9+kh@;>(i4LN^~ zb!Slh6)9oDa8WCGB}=Px{DO|l+m-|_#)oj=nf@8LPZ2hVS@zU)uAB=dix2WW3rGoA z;>T=DHx@yq0Toct0ygOLXE_alSvubR|Hgo+Q4$Fi#!I5fW+}*RuYjc&g|7-XZ&>06tN+wFx6$S3m?F>OL5$1Rxi=2uXl>ucxwyAn;=KTL zOdB@g2H$`y5Wv`_&mR z`b6MT>O@?Bo8wZ(C5J)^KfxMv?t3zxG%+E) z-un(H{JvH$WlXW4$tq+BzsM;;7g=pWXp;Y@Af< zL6P0G91Y?=Kg%P$EG*q~nFs~TV!E2;`hhbJR5Z{DlnD)iTMIPiv?%CX*pRe4*i*)n zI@md5Y%$PQlw|z6~@9$kWI)7t*AI4`s39yDWyyD9$e%cpdR!RIPtsxqeYuGm9$@C zY>4Eg4$6#w*jy@U*uC`PNK+$_ZRi5S%xB?|#z|dX8^IMb2}~gR&w6OOzDQcr?3cQ z&vtVq?mr}E#u-92?{oD%zMjsF@0tSmwK89y+WzyVAokB}x12oS(rf}ny4DRc)hg4Rg>rCf|QA-;v=F}+T4Y;yTm#=PrqS1R0 zmhrxz;hQcTW<1uvI71_9ObMXO@YS%3hZA}l3b}IR6WpZmebR0>$7ZhuEIAy5)Tui% zp(S*@Y^dP?)$C49$-7Vf6R4dfL?rk3-xZtYMz{PEr+sgn`3X2-Nc7=CBgI{vQpaCq z^8RN|pKZsmJyStSsHtc8QeaBKb957y$gaY7y$j*@bM0xSb+?XE2`=-0KwHnhLE9R0 zNw`zmlb1u2X4;HGu{0isF_vSs_w;&5KKCz&QQ6;w69G7l30kq-aWb*Gx#C^y^Ybe2 z915>ZoVbyWFdT<*M@d+K@W}l^Y^{F~+c6-1Klyx|O$%>XPwdfQ?OoQY0S()jX^aSN z9-m}1c-2ZXKgpASJCa0Zn9F}vuKkW)<7@EO8w|1aR<5`wV};axE(FNpSYjG9a6%9U zjd9SsaA)xC*>ghU%7^4UbXesF`{#vT+=cqdd%l01n|Kd#!PB2ad>K_IYDUPuH!rXs zIx~ngly3r?<$r;m2%;rSdEuPdqmzC1;z1eiEIH0Or4xqVud!6N4~{eC1v`T5jkKvkihY1tE2dVqC-sFcx zWqy!oNV= ziUwMPxD731YusiU10+A{Vs_q~Fz?I|f&xpY*M$e6eM=tdll0)|}g^vHU++=>8L*-p=${B(A|L`!GUCo4+qY(odF1ErL z2l?%z(@NBdFl(VX?@QrZuQQLrcOkIri7=oHO9oRJ2|GWX7VxfE)QTk3 zM8kc7CM0;Z3{wCA1Lp|@J(1@p-bnl?uBWGE@R1$V?-32IHB$S^VMo>`oURNs#;Mt3 zU1yZD8wyLo&GeeX3&d_7+^KuShD^cCDq9%kd%SA}<66w2vlhCgNI#M;oAhhPpo<+bj zA_JZwqeCGlJQYfOg(!!3$_aPn!dpzU1QeYx2lpg#_zq*LC5V;daFc* zAug(WM87)~%#jh?JVV@w2g}7iKNO#H`vWRIe3(;8!M^^lkf%evkaqnTuFdN(Gta+IT%>`7?}rVI zi&-{LbBLhL2v7H`xfXpU`vah2rCn%@gO^}bY{@fp=^>{KJ(DddsV;j8HtwFyl9Rg5 zxAdknNgH={dQ;@Y7uC+3HwVINIv zz(6$a7lT7Y1MhQia3sdXZIHvFBi_*-U-HkNKh>OhM$(6I_;?CK45$kF{Mm?D(!WWn zeAmnBHgg52(mvK>C`^H)g z=hCSoX6Ex;_ewLHD47Rb(ZA<32$#G48g}a&kGXP9lQh=76(p!Tft9!C5GuayhL*+$ zX%E$QhvPxK7mOm27xq9E0!a???hn*60yDhkN|5|EqYXTVrwJKzmm1#OVbF*+JMmtI z@gsfRY;e$R0m~NV&J)-NaFEMT16a~!AB^%Gev>l<%5 zxD78hDNRJX1RFXJNcg4{;$yK;HZmOw8BCiN+|tl|N>aPv;RRE9ASENuS18%b6aB#T zwCep7HRrOt}s z{16Q+VZ6*4N${Yn*axk(>e+E|FNieoQi;*$;o;%d4i=~r^KzTx^!4?H1O;233tjdW zLA_)x9JQAKgNsKFRD@NoA7oBA6C?U&?~hYCbmkG+S`<857tYC%>ONTz~ zNjV?=M{1B%GvGu7w+Qc}h)&WKZ>& zQxMHw1?P`^R2046Ws=_U^YfXBsV2G;aXpni61A-z&g`8%-_%Qzdvo_?Aw9FA-}Y17 z`9s)GIqsI&T4bCni-FQ;W~PWLu1JLeALlRXS&M^&`vj{WewpI_wnrY*vle}}ULpj4 zHEg{-%eLw6>T5_9)@#H~ z7~0g@s-aqw*iPbB?S!OFE;)~+42?GF?)YoL>jt5uI7>NtvpGM-h2BJWIT2Pq5!f`& z8>&Hjt+rYU6a;WZ+DA+|OywSSSwbNHqk$%8_`-9{zg@sW)un<@zIr4o{8znV5#5 zR)Pli+8F%&If{$7pu!pYFLs2vIu8cePYqWwMhl$6Cl8;Yy?7#TYdLmQt$^qL4vE%t z0^JYG9&)s2q)#YHr?hJrTMbMU1ZTeCnfxU@Rp>I#Km1jiy_2}u@Fag0;j`)+(VWB5 z%_^iP$%*y>j2Ct~!t~CR$caPMWoV$L`>p<|fqnr<@h-4iZ~fW6^cL+Z)W#ZE&TdKq zJ`+<GX26Ms&|GO)&~TjLp!m{02xd##C1U&26QDuU zuQt0M33;%RAm)K`GAxxwd@s6vNvckXSw##N{c{Y*fsIK;bLRd78#H>KUL4YM?!PoN z(~-6@EinFk>0GkVus40od!jNWJ1lsr+rTD8XWj+AXY9wUCgN}IW0MY`WrQvuanrM2 z5r2w;BF76AH0>?as~WGowv33nWTn_NIg*)88)(|W9q<)bPem#@b)A_gGLQ_l700GHemiM{xj^!7!f*TsW>JJ(+9r&e}p zZ%!-WyD7ZbQVc_GW)(yP{HJ?*O6*rQad3b2&8Yd%pmT}A9GvQLUjC_Gt62Zc?^%uw zwspi{3yc>Vi%uC1x`~NDYY-iKUfMpK6zilAn>FI!h+7(y&I|C|?;MlfJNHNcrvmA{ zX8%N{sP+DET?5@mudZjuEF#5Fy6SeNTg8MaM!l?}chW1BoJYJT5~Qet$do?8Lmw*` zEcB*^3tE-ChJSj*Si(P{LdWC3}110 zQQQ3Ll6H3^|A~HYmuTFHak-eIsI%o_pzZcIXLwF*cVOpxw%>b8L!RdG)*}-gywdn1 z*d^qUJ~?t{PjIYB?-A)?Pzurd-&r~af@AeygmuDj1{gK%6@u~!2kds|51%K(HaOV6-edI&5_dH&g)r}bf z2n>$DQb!gs++U)B;a3cgnzma&2k9`;rut`(X7$C?xZ4exUf|`q-T6WPyg*fJnZ(w8 zn(oVkxoMwn5`T7+zxAZ~?3qk{Ytr}Duv0}gsm+~6OERa3(W_jP5BybuVDmRpv=*t# zU|MJJ&VXXc@7B5z{psvFBFlSgCD4guXvp8+|Irqr`D%tM@W9X@dct!_8Wl+Lgvdm| z)52EJ6-HJh+4B_-(w5u&eI42;F+)q3otw_t(FoQwr67imRugXway)#-%~i(|zluJjQz=?Z1lMJIiSTqZpWC%RPWGLdIB zNP}XeP+~}7VkkmNVI>TIy1DIx2b)jG83T?H-tDE6w9RtUu60C)!nMvY!FlF2!vwe4 zTx7Q26!$9$vcA%-2R|pB#TJrU7@t#%Nx$Y)Crq2Sud-ap!mzPh)M-=pjSmh!hk>01 zCa=f_v)GKxf|jzMKVPlP^(tuDJba*~(X=g1$3NT|9Yoi8Ciq+pJ&D zwfH-5tZdO7C($XjBN03&KkZX!U0s_#R+s)ZJL6@Y#g!ycyQvl>tqOMMF@IVsY0}re zBgAtEcMf-j`DzjVu@o29SuC4xdldD9KoJrXi>j$n$Mhi@b}tcaEiG=fMuEi*XnTm; z?6$zL0x)?Z%&_#^cZ1>vTF_#XT)+ZjU<26InWh-mK<4OcPIWx14^yk`xuF!=xa`OV zi>EVYd5xIl=F!ct-!fc6D1)E-+Eu?_OZICw3OV%^!L-q;+D_+?Zd-LiS3TP`EQiYX zY^4$De&9XI&rmVU3GEce`qkg$?|-|4G;vmOraC)nTC5H*t_-m68(1ewFNqy~8tZrr z&UgEX&WPcfQfCz&YIT9N#7&$s2j}c72!O>yl8(DbRG{(@EGKoSPHG>3Mg2!Zr#X*xbp(q zcCD*Wp{3Ts5xxa1Fesc5?@n_$XGlQ!B=|gSr z-e%RE>Qq#{^~K6*u+Cy)qR=J6*7OL%#$QPeM=Es+29JCF=pXAf6=GpYHGp{nZv`R~ zA+o7NP4^`Kfd!c2$`6eo8;v_sTl9QRv+Iq3C$w`>BWF#*s@8IKU=w~#7O$O5Jj#$a zOgsUBHlAcSpz3z%R*#(7K!$2zEX;fJVE2Y=q~lwDFvwhL+lGE;m8goL79P49i8dTN zjQdd_r_)@+43+)msa8Ys@)6wkC`6Uc+dL+?_XxkN7D=_BB;T^zRQ}P~OmbzIh_BUf zpof{JNhJAA$FR!;Y4~yQR4A>xlqL_+s9QM@vyOB1G50qcL7LZk$ed1^&RG&Ych6d2z-4Jp~pqGg8PjNZ#{dz`4# z%Zhgw10A=#UCHikxTsAuzxC-(&i>A3 zZ}|@V!J)=?!MyVawa0J#)_vo9bEl?EnksE*vGiA0x4$Xic918~t|Zb025i?`a^LvFjwE>)p2wLGg#5%on!9 z5wbIc;>fdKAntB@NDL=%pb^I{OO#e$?Mf7zGYmVZxl;Cv9dw*=p7%Yhx=oKpaNpe} z!nJl#r*|yN+AdRzOLBQd^UBjxRi!7BuFI}&il%gpuWCpwKbxUmWnN0ZK(cRHJ)uhP z((Z(}!oJPVU?{EBNfEcvd@wVoRj;k8a*`|AUVkF&B_^HGE1K@lh*qF|GvX2h(PD zh-bu=IQ^l6>Y`Caxz1?lw8r+xMBF96LbSH%HkkJYKmoyqDzN-#Vn8Y-3Iq@WY~A0>;Bwjd(vi+ zer2e%th8wVRrv^>yE`1VZ}Hystd6HTzhnMCdjLMZM63Q3D@_NRV=3*m1a9}_p=V{& zZ}fF)Pu_Kb)zkhs?jMcTvhUo=CFy9JBErM|qOM1@nTk(smXgss&^s>)CxQ~TFO2nM zqujBa`8s+B_y73+332@7^|vT}>madH`=^Z54keHRLve(1(~V?qZ^fX`Lkc+>Iwg=J3 zvYD9(t%Y1f26qLw8C2PoU*R^Zv%fGxKRU#ybh*;93@JHPRPGfmIl@_HE5v7i8f-sq8 zWQ~6{W+8#cj-}_xkkcVXw%?EP!^^>2XC4~kMqSa_=(I{ed^9>5c?wtj*t&4f+NQCR?{z}N=kGTsMi z*+jBP6}2E>l-_9GCm3gLK*xS+DSfi_j6>gOdzNi5i>VZIlkL~z*?o>qhp?{#1P;UQHek92-`%X(P6SGoWiaA>yRQ zR6{N|;X_E%Rj8e_udjHzm=n_l!RTXZY>1;s=K_;Z6M+84jDoqLOjwiIlnm2s1;o~T zN|Y_b+HU1w7QLJOr@~Ql;}fB_PC1$UDu=@2HH1mR^VAgG;D$PUL5k# zd=?vC6(wwR8d8sJ=YMv7&9K5)LqkIue%~0Cb!yo- z!dk9KFvc%eeg%MX%x6hnP|C+pJA|}Pz%4--FzK(PI&WO1{c#6eb%x53d}ZqcJ37NT z4Fx7Wc64R}FDLb4pD^*V2kz)-9!Jpa3>1w_?*t?47`(c51>&3SIJ9DFE zf#Dv0e)iqwoF4yi-i0ZjuMLvhU_nV`R{v1OI8#n4leXyAok9!~+mZU>RlOby850>d zwpFh^&qWvzF?Y70`5PPz^b$Ci5?@@oJfAJJXEkZ;mX`DQ}O2+Cd$_QKR`@>yWX-be@u2c0l7NJb^j0aOx z%_oo0C$#Ci;MKX{@{7&oMJf0TrnoSe@=e9#S5K8o#%)uzm6xR)&UP!#`z0OnU|J)Y zW7&-{__h)hcU(ciZ`6cPWhr$^y&TKG4o13qWz2%UX z$?jz*jeX=Fx@YbvEj87tWHmLB^q!+Av@b=+`EQP*oR`5i2G{<%GO+A;t%YX? zLjibDE|?ipJJEh)611*oX>?AAF}QOSUCAx%s{Hjo{gl0lx^HQ$Ax-&AvHh7ot%;|N z_S7X)buLtMVL`!AG*TM;*<&w*oF_UTF6C}qgz-5d{K+}A3^VK0;ijb*J&1Yj@)&ru zrsuHQkIu>9N&xa84xQihAvn%lKa3VTHqLV0yL{m-Aqj~X_(O4;5hVg4DYSs_PA^H~ zUaP-g0wR);wjt8!Xu|mi2{qk8j#JF> zC^mgGeatMQ6l_s)cO zZO}_Qy(rx2tDE)TWmWWQTp=Y39~ra^kBiY6Z%f-PG`_-~xar|1YqOm)H-S|E?YpYL z=QMz>Qx@+emkBNibUg58rT4bKZuiFMaBPs11dCJp(Q2~Z`0G2{h?tY+dnjM(c4?$m zszhj(mAZ^ePmM*;U(h>UCW0-yzU^?DY~iUQjGgjrTch{20OY&~0+df|V>^9Tuz*X-`XB2p1ty zv$R8X=hG09NL^pd>OB`uVf1N#Ko-Ryxcz;{n*T=N9 z?~hhG9vnrQ2BbL&QZ@y8OK(20+yBa2GT>ZB4Y_sDO7kG;C1v$mjmRX+E3{17Ky2f0 zuE%L8l&8W#a&V2)z~kWhimKCS%}T`j5L>kLik&3{krIjvrnwm7@awOP{5^t$2G@bT zt}%rn?_$r(-m4`JeOVERQo>bh89vPxf3bK;+Bgj0!p!Nx?2(@;$e~kY=Fs&`2vTj{ z9o_VvLjU_Z+a6HAmfB0#nM+1XSX5sp)==CU26k~>J`>a-nHs=F1a+)G%E=G04kQpb zHvojZy@}4GCZqR+=y~f1Ms}(du(k-`0&o*`&glL6%|G|rsBZs3-mIk1^)O=F%X z4HQsF5co_c$Z6D}shmC@XufO3{m%FTZZZ$OpNBHtBc)u)K=eas84 zLW8bq(Rp!HcV#sSDLB!oO;y@c7205;J6y(iXl%q~a~WCcAjA9v1Bv7qVu+vw$y!hW zSPMngEKB2-YNZHTPXVG5iDgL<|jF7C_*So)IJP$E#+o z(u^Bx#8c7(W-LPAW%1-Fnrg_ePg0NaC{*Qp%KdV zU6Bz(xI&h4Yqb_(9T<=LKljX6Z(PgEI zI_qn)L}FU97HLksFBmML2jlKlZ}EXod(4)*USslh)mukOT&B|-G6>%yrrVx`oohpmpj}Z*i`|0&N=07a1Bwv1A2XO~?`vp`!iA&+?8! zd)0K4xBEGK@7jihugXc^Ha$4_BRwz(bktX33p=EQXw=wBJ?}gnOM$t&5_d<2G1MmT zQI8co7oCss!QZ>^R&IDBD>2-WgH~_zY8e`Fa*(JQB(k^5q%P41JkYbxdgmxE#77GF^43o9crCB#{&vL zS_}!B{Yvjj-B~QvN^siHf8B^K>3x5w?G|2nGw;F@It;)w=KvQoT$gA>zMrNSSk-do zpJ4=7sl=<*pJnw=tWAeO8wSgvwg7WDGMd?4)IRH7Uzfs>@!9PCBq3uHc?xGSAv?tO zir~JS_oUal+il5ghajr!7hVL)c6T<p{n1#d)G@MiX>W~swk<+JgNGU4C_*j7@zF% zV)HJsrt$lYwk=DZMxFEHj@{j4JfFP!?k;hOm`~l&?lsZ=VUNUTkzPKQ6E>?qHJHC7 zX|;KwX;N$FngRdtOrNk$Y0se{vJwRstA68^?7#hRVNj*9Fd^{N4Gn;whCSY>lmM&x zd1`Q_5;lcF`ka07+8Dk*2GhW!EpSpxOmu|8HWFaZ1O7Plh_xzxHDqvx_gUSK65)5dvAiV*96bM&@1!>scOaxH9`Bb6)-U~FHSq8mC=k^x`LX>``Bb}{UE^Z}yJoxP9Jv+DCILk6{7)*Yq)$VH+ z&m%s2v3><6A^U%PtZ=b8+Bh1kmGy=`#zPNv2I+AR!b^7@+q6*s-JPQK zDnC0;&waducQ;0r5B)l}>BRGcp&$GN-^rhe`FKpYt-V4cHOwsR&A8)Xk?di&6QdiB1VR>*?QtR^k1rgXdD0s22;G#C+--@GO-9i}BsxJX9Q&}y0S=uyoKcVbN zF<`^NjYZZOS-r#p!KcQMPRvL&O=|OlgEZK-nhx+iLktV`#azJ+TlYorJu4`YvFa+r z@!6ow_l!D^pQ6j}|BKQeamIeK^9jN2-Cd$-9|dQF@L8>{H)?@o7iTL)dZ(1r$>ytP zaeH&QtklG5iK!tndfrEphDS>D6Kt~BcJJQZ-KP0CvG&pERG*7>Z05inEte@PzQ(l8 z+bEa?0B(BR)3Ym^`qGii+yb~wT^{Htnv%;*;oMWts*`$=*K-#2G~97~znO~?heIF{!3Sk&@r-3)Y6(*pmhB~6a>}rH z6Ir3e)J3aV?vr_}{=Km@3~P-_GSk&(LPCGJ{^{vX|KxOH%nlR5zYyp#A_B1Q<-WaZZM*95l)J&;vxMF1kHob= z{rs(8`gR`W%SkRxN8)=3C0>Qi$mH1S*hv+05;dIGW=@kK?pG#!HdCYYDVJEp;r8`9 zUET55!in0d>~KODXP6ygDvepl?T=S2EDV<4>1?aztRrSnm_M*_B#(pX?OO~dixksHJqmVw(#EUy!~;NzrV7r zzD#_mY^%~1YrOz*H`hxA_jkZ6gPt?utg^0|V$6AwiXDGhMQByokm#&g6SXE2<-Xak zEZR#Lrb&=q$1p!-DVuyv0zo}6&J&TpdYm3;16Jn1_t&$uZO&{rOsm*zGF_Nj_FzG@ zM(jol38r?hc;h-qax8Q#cLw+Q54Ig{bEeY@&gJ0J(eF0%>MmvVw@qDC7##q&dHvBI z7M*&(K9?E80<&W$Je5X*|7^SH;4ktPzI?hN=ZrkGkrIO{%y0xa8PvU;s1M&~Y>K^Y;gVdvhlcCfR|#phcnefN$8gRaV9 zpFLS=(V(WDXm`A>RQqZ}8s(b`ZW>SC)clOwp?fcElsshIru?MZgGCSMgR#Zh#hS|2 znP-MvPufXtdD=SoW#PQmyB}8t2C3*-t31r}Y<0>RjblS8|Lcn$o>%B5vh>d&XW(1r zHR+y@bs9!V!}(%;9U|?elwB^gc7I9BEZXm&Bt^GBOd(uhcC?_rCa=yRZsg|7ZB#eo^nl@|+nO2cBA55x78x1WrPF}e5&s-CB-@1Wx~bz ztyXF14Ys9S(mWFnK*&oDg|Ih_Hl*1PzoE6bOnOhCb$gAj@r^F$zAQ;U$=JVcZ-4fA z=ROOyhoQ2o7_W@TiG$$9@AYe{LdJpO#Fax2(5`#;fme1ppP3Br4h!D>n&Gc7blZB9J=*K46 z%v!zJ+4H?~5|~p!lz#^Z4@zRU5CM?n1?DG|VNfTG@BX81%agS*J8d;O&^sb8#|n8Q zg}BPZi``k5uvj-5unF{RE1FJ?K7NDa29bgYsz~sTqn({y-tY8aGs^aZv;9Kz&`S~L z)6)}n&?j$0dBbS&em8@`j#Lsy(WrK4-{6$$%`_Cc-^R876zE$5*w-U z`BakgNvTQCxj?7%W3Z;poYGS%^d6TCv-x?`z;#z_=F2K=7tOmiy`P3`Rb(Ghrv_wn z7jlWd#u&vNZojAe>$78fnyzrisy^&ALS5b}*Px+rZ~UOJdE?pkl1DpbRHY_b`G+6F zknc;Hyo`9`3&bVo*}rb_cYNjzu5XR;{hhAWtA<2S~qj+@L$q97-x)3F*8PU%ZNl#47{$mFj0DfF+r}M5Xqb%7wcn%a)oTLL93%n z{xx7SrHHqlW7yKM4F5aAhwBEsH)_Yuwa6ky=)I7qjLehpi=2o%g+mMRA&ag(24{qRH&B=P2#8|&rdps zA`Xq~>y+fx=G?jHZh4=}-5oHr@Z{@Df>|;6=x!pifRj%`y{qRv*?aVd)9kpZu^8e; z1nLPF7IIc--9P5$;sYW;z{RyiIzxSAhDYEq%vfu-;X#@O=Cib5ULzWz3XB)5ds-Og z#!*y5Mu!SF*Dn9>RmSyVjiT1Kqh?tognsIoR}|<-8nw;x_g`RQye-;$>Y7ZCN0Mmm zMqLdjN#z0NQzz(y^hqfs4q$}pGi9`Jh}ybf&d>i=VN7+JC8*sA6F*jyUSl{5#-Uv+ zZW}T~=cn^Y4k5{8>Fy>&At4cwh_tkW(7Jh_JjO5`H+kOK?xkz@xt$Z^`ecbo`5A)< z_*#rR5mSQv%qc-4m+i^p&#wk0A0Yf=kN0b;5!{kYtE9|2+533QaS<3jPGsZsh;Ku> zOxQZUXDv~n=X5_6whkO(HHIg|1(vYwb2;m!Z077sDW$}c-V)~f0b7N+d0KPxlsC0p z(-2UST3Z!@(tBx9zh$GAkKX56%SGWoMRT$yr0O!=0#+)F4Sy(5w*S%jO7x0!6@Mj0 z89dtdBocUlyuZRs!#Jh((N?apmiE66I;v^a#&3y{3=hHBp}fq*FV6~-{T0ZqD@GmF z%8Yrc0vcfOzOw}_Gkc)j?j!|MoDgoJ@?LPQ6I#2A6X`aKIx za6kl(b@cg|j)~)_dg>Bus@<~xjqOp66iSr9WGKwz6o5j4NMi(cf=J6~TzD^O2_qK` z*DsKRJ}>x4oE*J6r9@!o$a;*_%qzlR+EnPL1`lDKcVv6dIJPy`W|&1~d-u7CL#NpEbNxlvZ}Pew)4cI++La4S zo`u%Cr>O%S)b+e|4%(~W=M#47!ywI9mAkKt^0bt9e3E;OgmnCJL)-P2vyUIWEokuM zfQ5{Y(VL{y*ZGr~8jDiC#heum=I_YIM24@n-LZbdvjhbuN$dAE!798x8X#jsG58ak z178U;NQgNzwK~NJ!8pYjl`}CaWWrm%cwV1zDU2Y9+(LXwkG^!9Jw)>rw{POP&m*;m z$)=6EU*!7&&tr;NLwU^klr3^Z=!oHpZ#-9er%uC!p6+Wy#BNhh)Kk+B)NtbcRXa0U zLxGXsPQL~smvLLntQLekV!{wBG)K3I9}E#==;E3Y+r|fJf@ltoG`mi|4|3la{O03Z zpIs797kM<=XN8~o5VbjCljyjiyaA8*u7~TrgL%BTs|4*=r(Ej0x>lesZ0;GJIeTRl zKoFG*=w-VEd)*JCtPV)0DW0)deg2KAeODr%Rc(0sm6L7{WMK@;dK41msp&MD+cLa! zALy8VBMqayc8MW8fw)vGIi?C6puO@`qLuHZsle$HTCl1AWw z!0-4#nb78&Lxjb%$$Qt+JIGbpqn#a%3jz+cmHa~Ti};;wjuWA! z+{!7sT@$xf6>A&6wv7nYs5up5*imk-I&{N#&|h@Lw@BGbqJu||oFQ+tlNrpjf+qanw<-;}s8mWyc5k`^Zy%k34U2A+U zJ(etfe#da+!qLbJrz$SVN>NE)1~@aE%=6FdTsi`<8U9(8=W#{Pn`;ywf04Ed4YddH z?_LIcd}%iI%X8vAyE86XiHQiZKT#@0&(yMQzNIPkId7>N@(l?uZj1`$EfNTVeS01z z6JiWnGnoJ+Lyb1AjBkPtta8toxvG&fhUY$)Sh!Yfdm`c4hGlngG`Yu`NVRXsb~<0) z)mLJDcK}ny$30JvW)N{Wu^GN8r0$ggaH-eG&SSAX6~1+NrFtkv7#e{W$bBitN4B5~EbaO%{&l zRhwc=I#tdmy^Cx;p(L^=Y3MPDe;20A#-vX}e102haA{!tpzUp|U)+|Z_05IF_co12 zI|s-PNi+#5tnt{GVDKeT-Y-1tR@4FTE#WY6fAi#ZC=7W&=lHvsIRQinaI$+`1i6J*4#B*kRf4-z6BBy2j-?Wfz0`y< zHS2nmLsD&!$QBB+c!iyIzwFeF3VCupK6ICPg_T2g*nqNX;bmzXGwY#z?>BcoIEmWt ze@Qff6VeNAZA$dIxa4P4mA;%0VshMf6h7*wwXdOCoJ=t5s?AuU=@1(h;HzQ$_8E`Y zrarYhA2PpiEj4bFKYB*+u*MtfhgB`71s>xm`?*r-c;NlDB4)4cT0Cm+;?^Kz$~~LL zGV_w)DplHKsjySW1IXS+Sz|x$PJG~%jHh@9a>YUDB8D$e4u3F|A=cnfSg*}|G1B}N zQ&ET^T}vK0K=c$28z0p0|E9DQPc@k-zpbLM7Ge&1I#}fq4t7Rdf97g00h{kcQ2%Hv z0=S6u9lA3P6&4dlRIBSlT>Vz>wCsbE42fFg#4zIoe&*C<2IA>{imK@H=rO=mMsQDL z)Lf0d&eQ#JqXl)$QBwWU{>OF-m)$2m@q}-ci}48A)fe5-_B4Fwt+&RlK|9CaHjYQ6 zA8FpIz{B48GxS=Z!=?~DYa7|x`y$>WC)^*}L+4cD43YNN8T!Re&OaQ@Xpm5fPS( zbSx2R0g+hJA}t`X=#=hm5V-Th-us+=&-dK>`;9vWjMa zOU)3tjFmGUQ+dDnJ#S@(AF^Ypf$C4yZE)oEXIxxK+3yIYtQv=LAFx8eZ)MZ+kMUci z%<9zX=6UQpUuhT=rhbK~#4OwJ&h^kz$8^GvpEYWf?yc{!o0CmN^1C1nd^T z2%&hMYH|W?3po}s@V(6Q=HbFXtC8CoD4Ch%R1HX?5W)+AG$SDQ{q0x%{y(g|v(Fwt zJ_vY^w$}T5%X1xP`XJ$Y#|BEtCvf(ob`N{B@=2GkcrX4bs}jikw-kNIw=w^`mxuD=1{eY=B^#dhhkAsQG?@Gco&d$o#^DJQLfmG@`q{ zeCon6%%#T`=r=yi?(B-dsDm;C*nItS1xnXh1(>9Ss-K#FPphVst)ClV=OaV^t7@A&J-#sZ43{_GrMnQ?QMG<^!&#*wRf}V(Z7|p z6OG~E8w+hA@^U9Vxz-=8^e;!pza1{OHjOYgwbRl%5pLQwo&(dr{=ZQwzQwrX*}MpU z`()fCp*+ar;Si)vxIXq(wpvyS9n(j7wq62XL`S`kDLU|q($24MfhMtFUub^_W_o~E z4;2a{vkZ22unPdJZ=s0XN)0jlyZHtewY`t4zeFIw`z2yzECW_9l!Q`mBkyVg?vxWS z&jfDbqmIgz2dXlo>B{-gGj6JA7*WYFa98Ed&y#H%`5e5W+TPDo-R|+-t>Rsi*F$UQ zZb`>P-o_RbSQlnzEpu{t<)w@zv_lvL~3Be$Os!sggR-#F$}sCoMS@eGW zz+vd?;4_nqukb&?# z%c{L)VZe_{WPSCkiO1()}?yJ5}$#m8=ob;m1)z%llhtuRjk}j*FA&L) z@fBEAx4^SSw9kAi2mA_7TYz7|)+TZlD*rr>b6_WOoEGxyULun5N?>;82doZH`rrr{ z)<66!y``f~J`%_j|BwfDtOB#o2wagK39Jhdo-OC!ukrsW%tnLdbeLCD=ov9%sBBZp z)-^5ez;14X#@mbecEtL84REeqZYK39TypX7S7G&q*))z)+>%EpY;3>6AXXMCEHa&b zU(Cr#v@%^1m4fc5)544Bdx7-rH(T>78R=m{#3_>`*Vq2bKDER_YRFV^eb?90eO}#l zH$s^hC9%0Pj+!r?PQ*@&9Jzpr$GWdRdDaD4u(2@ra8s%c={JonfcdfMN{ zws`QKm@p3F9u;?m;H(4iuqa{tI^Ht$oky&SllB7_J|7~!PY^)2?I%qq<~D^3imX1b zQa3NOn+yD;kMnRpYp)C-F~dtS;F{!7RJxByhU3b|yF9wf_PJj%|7F%c_1Pxk0Oo~b z3%!E3jkiJ{Zu?Bv{@ek$D*p(&s;)QTY?ORt;kT|X7l5Ad>QJIpW?5&KZEah^ZrCB( zE1e;BWwy63T^1NMSfH`JYP%y4Cg_&*qwOuDYMf=&JB@St%11_VJS)Qxak5DuR=zD# zdW}*F=GBrdS%vkeST*MD&RLbYA|=0Qm|8PVod7jg`x}MlE(c|mqlf>``%(}SZ2Dgp z(AWhN4#eFUjI~el{R3XfE;8+#2TQB9Vk1&ycce>QTmD~wDuly|#72A?`LEGD?#q%Y zM-_tTR{Go?H|6lu{SM0J0>Ek!M_?WI=(ThV>6XhT+x9O6Gz1U_u7Cofdvz-7KXGhC zKODnvJ`JdygX6cY+@SV0b8e7)0sZTS_dqMCs}Uo5wNcjbNNVB>F#^?=`hAPb)>HQ$ z#+=Hscph4kbl0i}`PGlmRc?w%{2`@Te$ZrahE^(;?^1W8q>^NuoS|=N*<^2l%@Atd z^xL30eyYUI`%DAXtP{;za;YYbZ&TD7+qy7uC0qZcAkT(hqC^Vv%V~-t$^XL)`^3yZ ze7U7%_PcUrrWZH?WZU0K`Nt4@=8RwIzTNCUUBV#E@hV<`Lwz5xJU|JFus8sd1xptM zN6=^g*?1Y}bNC?IjyGxL@V^+KLD+{yDkVz?crcAQmvEvhw+BH|j*n z^moVk(B?}Xen`j9f{-?C{?t^70hSmKYwEgtDt`vzatnCAc;ELwlJl%ZqXB=TDCpN+ znm%7Oxj4RFKEll>Ix=QFm|O43lL3EjSF>_CB_-`jUARgzkQs5I>v146+G-|fgpt9C zcgK^LBgaqj``U_N4awiw%_~}JF2`z2ABTZjsx-FaJ7`nO^Z8$vIj3b=)<4BqWMJ7R zS6$yV?>da-eGW?uy>pGKYr1R3BR7xh|9yXo;ei0BUpY2Kknd&E`Uw#+JN?Uc0#*>9 zz65?vt*$?N5gGKNKuyO)3ViYc{sH*8Jm~@B566ImtmvWVxZTbKseFBE0D3 z;xysc%NoPNTdk|H4u-e)9W8dT!2wQ~~T{@HJaat(4pM3Nb)5 z_w$F^|CLH5@$%19s&I*y3F6@k$EX=sELpK9D7?=Z(em?Ne{-wr2PHyX8nBazSl47y zCYZEx*oZ?vyBg-o$TWYaMkzhMk?%e@g>#ri?sbufP;1gN~uDs zL56kd-=l5&-=pn={@iM0w0-g{?|dTdwTu4^3WmmCJ=;`RyDbzp=`59NFUQ)`sl0yM z2yGAA?fsq>0ZC~2KwR!WqhP^*Gfl`&Ma=Er9PeYVf&_156Bgixqmz3R7BYYW|Np_u zD)0p2`9LZ*Pk&>9bjjmIoJZoI73@R#K_~{R3h#CuwJdG77Oi8kid!^1H(Qt7(*7Oj zL|HMI$OueV2=|aSsX4elPMLe$ZZ3)e)wzq_o7jnb6z1WkQecCLo_JlD~ ztwvH`^esxxDA;m!eL8IAw)#~L0^~WeK-Xn2PqvasSZmlH45dyn7Zqr5|M%KkUoP%= zn)=EDBdRA0U}Knt-+r{rPObYLAIoEvX=n84YH}9!!CKT_BWSfjQTYX6^ci^~TRhTOMlVsuY&i&c7k6Jy0M(fgMrE}r;2q-gtUgJToC#E9Kw%#+c;rqm3V70rai6dv+_kDzNDbUcu+=^ z*c4|Jhg_H>?az#mVTy8t5m!uPPgZ{R)Cpub=0MeR+1%RssUI83sG1o;py^SXF~7`JXkDJ{h<%o3k%|4%$-!7l=N>Ce*4}P+ zDJ6Al%DBnK1NE&NUs~FDt-T%@lfRT`a#;>NU9W@P3BGCeronA%qZ87*b)`D)NIHf?g~jN7(n?S?>!&tqg|dB4E`Xf6sMmLA6kg3BEZ^2!i+i zXN)Ru(g7s#;BV_L_byF0{-;L^j(MSwwexu>9th!59lS-OEy0z=$u{lQg_w(}$EaNf zko`9WF)x*_PmaxK;%vN z+%G|X{I>ibGHCCsz4jJ{7U_TPdDl%LX+FYH^8dEy4c7FM?F(IkHWJBrf$13=4>1;j zfi|%ILN;K)%7Kyv?-@YbvJ&LJfzmoxFT{*P9_>8PQ}}HsUPvFiLDa=T48uS3%uNbI z`v{RdKOqjb=V|7(5CN9=k021IIemLL#RZWC8<;n)iA5R=aRru%AVP_wqlE*b*ZSKY zQ*NHEkCrtR9d>i(Os!6|g@33aVk394VFl@4vse_nF-6r+sdS0}cn ziOrLtKP zduS)W5!-kW_eZK_)FYGKjmjLu~^40BLHwcU8=#G_*}bXwd6B)HYF0}dpG<|Vu`{8$av?bS00H}y`UKP zccvgjquN%-VbNUAE>jS2=7)lUaejWM9bal>xU|>)fctucGVb1i##d_};t%d=^_mo7kIhTx{pL>L>K*5fV%rqmRW#BN$>8U!-9TL8KZc|1b~ZAQhR~L{|h(Te-&zxkVzXg)x?+H5Ld4893}! z3S|`0*|knnpX@Crs{a|lDia=%ac8?Ogt0}b(MTFk)o*ryywa6R?ADV0#yj+t*3*3M zf}fX+7^#M?EVeK)MGK1J+Bx9US&)6_BseA3!=^M`KjRS67)O3({3@N52{Ai=VXU~O z*=aUSs(@Y^(o>S|1QeTN@21}q4T#Dsx*r_u?CqaOpI(2LMwijt{6t;jW>$~OeEIgy zq`>kD#%mw*DlBYGy>e6C9axCdU=TVqo~=z;$7eHkdeSyVpS^KyjU7l!Lq@l-oQv;& z!Ve4jenO)uP95NYW{l8??(%t) z*1>xMi}Z1YMmYx_^NVqnYAo(S=f+zGEG{VG(!fF56Q0+{Lp~JTZu+vYl9Dq&FRGGF z+rUFEQ{uQGTgrUnF)Wl*P5POlffX*A^m)?bx2MRe^B`b4;b-=nw9ZC3XikUn4F6Ie z*&~6-6=pCuzNp4pP8620L3y6d{!YzIAMP9G)+w)~4)R6~rjE(c;%PURrCqIpw?E78 z^3ml?M3-xRP`hI{yXWR^zv=-b;=?dt8m)mUY8`B*E1&D@>#I{`4GghI?&C~XD=MaV z)kq$_xacjNSvQm=K1I*PF6DqnmXOj%PJMJUh-K#2%6aPgpJ(}8h5cJua)nS6HWwS2 zE<~eX#P0L%3#<2P`a*s}y80U@m@dL)ck8($J($kQBcH6smXXc*oZKPgUhn;*XjC=5 zM~&;C`Vs9pnf-(4TH#-KLBa7tk`oJjpIj5x78oZpZfCCB8K+$4p`$O}9XY&yv`KyR z#+k7$O?dNm#acw}3aFACHrX6dFxdtM1(8uv)v*!)?SrzYozr4yF;2z=sse9_({&|45$q26OY)X_fxZoCp zaI=J@`5`7K>B@aV^mz_v!57%g9h$Oisz9D!6B|ZX;t$PFmRquV(LQYQN~)|xkL-kw{F>K3;z#c`8zrvD=UO&!Hv8?n7AP@}PK8%?U6aa+ za9Gk}VY9=Ca;ws#*&3L8iVr<6VT|7R$g}7for-s+d+=#r$6q% z<0Q>fZWLHnHl{1sg6LB1$-~V$hP6&z%5uZLJBKK~9uMjd6kaTp=LgGB8w8 zeg~O>`Y$LnD>a_^T(S}EtHe+pH99GUt4p5o26L`H8@>|kA}viLE!@Y*-)tvrpNllFvo-EKfAk)5h=WpT=^=lqo)6QwpAN zj}Wc<@S5wZQ+adT$fGwic5IF%mwql#uYPDS>ppdT@f>nx`+Xk?P#-bjdra0CtczaG z*0;4PYp(fP*Gw5|+g@-gdNVelC%2<(*>L77Dv<_5G3{@x@bbg2N4HjPU0xZqyB2{r zIcstUtMF`k`nIB(yZr$_X|C3!*U^4uT4gS(6?oI`i)Py6FuJZKE)Tr?K!g{78z4%- zXY21;Zi(X7*3tt0Om3;1Fq$6SYPY?lIcSL)itp>{D$Moe?Rb5=#`O#Jf))IC!4F~Sz&!DcjcG!__A$R<1Vuku_Kz{d`Q#r*_|EZ3+o?!_WZ71w;dK^X2{1Iomf2GFQIa(4W%tBt)-WdDXz)6>eH-kvJwus6@31*iBR?1uIo0%8XKFN?L9p~$AwZLKBjfL zXQ)Cci%meFX<>m8&M%~P!P_tml{d7EuJ?B)zAP==tHDfKUS7VjwbdW`qs1E1(p9&A ze`XAUT?<|KXm_?c0D~nZB@5cjQD!Kr#-Cm?8H}paF=h&GAcc?Nuz9wncDL=)zPZRAth{R2W;Ii5gwLK%K+)8S1S7A99)3-mG2?}ff_~M#?fyQ6EVx15=B(J8K83yQ(Dr-Tz z?U3lP)h9IUx{#W0%=rUEKy92Y$McT^=`+QWMnrcPNU^qV*SMk$p z7bG}AUOVUZ_V;&zpWUE+IH|oB0}f6EPO+QU4eqJPd$9aDD32qweQM}n|MU5;eVO0)rn@9szjU8%c67g}qpC#fZE`I- zH6TD-INNU7H6VBu&+ZCW=GE~2$5&FdFu4Ad8>l^A_YTR0*KWDgzC^mD#owRbG-|!X z=SG;Nl_n@4+Rjc`b^1NH`TQKGK+>X?ud!%@=MAc#v-TZW zteBkK#Yh9NFMLQB4#x`G)xdd?KJ!R(?j}&@Sr=$?1u3>RV1hzHWBq zNG8lXWBYzZ!*`MMv}t>EL*e*{Z*!6MsEI|J9l2ZJ=6Bt%)GgUub<{;1tYAM%`F${* z=6SX?uXzybwz`;aFBys)(uTtkSgLI8$KA4nt${3cR!t(j3<~lK?C$IhGrKckg*w77 z6T$@eH3=nJ#jgpGRuxD`(%psU)?m%mrx^+7g=drNf(*|AA^le#?!^El^OR|22Ye|!Z!$pj?TqR8_ ze?flY<;iwSQpumLWc*9j^UDi?yfigiPc)dJK{_oV0dJogzl@9RNejnFHTGKFyoQU~ zN4J6bLGHEA<5dlUBaC^`uuNuVKdrG`0zW8nzEp#3$aTn#vzk>!O-ENS|zs~NwN~A#$3lz)|=R$FOL%TegoR?bV z#slX=?cq0sIckw@qFBSS=|)D=ye5-_;>~5bc9yvhaV0*Cy9P#a7@RMk7YsfvxPAWBL zsiB3(`nXuq8lPp1qiLo$snWRqo%n(r|4SF1*%Dd0aGHq6Z)DS8hQbcqeJRRILHFh% zbhx-@{nj{0HuaK;)T7T|4lCYrZgOe!hPe4=h%Ai8 zdwDXlKehfeKBr7|0r|24()Va)vD%2~*_3qSYUdQ_#M1fe$)2tIO(8rPbs&bj`t1MO z^FTK+dwt|vQs(c&kYl6LUA(r|U0c_EH$gqu*wU0ztfKv9 zob!8!_99J*NcBiWsRu2CqDtrZYOh{{Ufwn4p5dT?n!`6OhP#K`m96{Bm93Bu=1^{k z(Q&0EdRYpE@z=^MYjs-n*R8K->^K&;)MoH``3{Otr-?+}%e_8W9Iw0Q8XR-|kH>ro zYF!gs5tqlQNmkBLQjo8)JJQ{ALZ8;e#EHyI3>tli-Znk`=O4I}0U0kNUX0BM_btDY8 z((>oR+#J~RFK&sc7$rMAX*E<_dCE_*+9lT2>6^ zdz9FWyJ*RXBqEj0OtSVbTVfj7RbMr2&@jeq% zrd^L04tSr|?#p+K{;&oMA8A>w>6z6_Xrwh&0fbwf=9`{b+hHHkcvn2NK1847`l2_- z_T9ls`R|KLvrF67JO1b-whbB7uGBcmqLyBMTUVi}@7&p9w^ z)@Sy}$ICu>2Fa;fR&%sK5+WcVU}a_XHi?BZDiyLzj^vcVMkgj3y1J4BG*dtBcn`Zg z(zgw2DPqx+F}l;Biwg`4yhcv`1%(eUqA*(9$qg!_hK9b4%_VOgR^uQJxDAB$OWLn* zUx-0j!YolO8Vgt*;IReV@wap3k8j%Iqmwbn2i~^sFq#pt2Ie{jU=<`91dGc}!9FrX zG)V_VGh+Pimzq{?)PtM%1*st8aM4sfwD$;lUoeki7XgEZPmE9MzaR{PH`fw5$#ntG$d^teIW(BQjg=yIb}Vms57Wg}(p?haNwoj7!knQ#wvEpPW# z&5mWjdr%VpDsh#~eh}4z9n`^>#(x*N9P;5k!<~Y8Tt}J*l5I@}Kf=SGrKY7bGwUj5 zIpSd12`GGS4Eu=f1O>8Ya@mrKEcO19a0-TAXST7EE~e5EWXZHr*uhCh%E@IHBHKaF(d2#bhao`p1aqI`RJ93<0I zkMXB&Me6>tR^=TeT6j9cN1-#L|Drvg=0lL+*18o7Icrf!`ME;v`%|W7YWTE6mdw8$ zeeo+3-aqY(DO1v4Qw!b|Fvv9U)4f04k!$DvyNPi$I^@l;DihN~Cm6hqXyK4@{R?on*X_VqzIr%# zHTV}LS*-}G_nhOvHdp4Odwo1r7b{0vA*+MWeTOyh`Rm6i3*G>eKqjL@ze4~2~vSZ;9>R0~_aLZKk42x#_#R3mwX zr~2fI>cfkSa>`r#SEs&M9xp-iwkMijBQHpo}K_S$#Nw7Xy(1BL+UU5I^~{_GD($_tByif4GTk&Vmi2SeigN z@s%C&RuSx%&yIQb9VNNs$QM=(1utl~H8!?B;Mt&JO+;MpAVkfVM_lvP}QCoS=U-CQ4QstLPC+V_cU_9i>UHL`H`I-N0z z7|h{q@!zjQXwyznR1)dybOwD`d&m^s*73Mw2bDIw)a6yZCCSc&)pyDHk=-P|q-j+U z%v~K-m>FQA@?eZP5aj}s$3xz!ljONv*Uxw_SXh!NT3QhOtbyjWI)=U9UAfUot$?G6 zHNCl2Au;}5TtY1a2{k12&Zk@QEU6G^AK@e>CI+2j7TCQ|L-L=nYM(xPj02{}!J`Xz z;NdOZAdc$Ml`9^O5zr%E=De~fCQl3qdAspDz4|cutgeOb38;eEO6pE%+u~-J8G%ih zp81%OB)POSE+>Z>?4gT<&?z)_MuIYKh6;9g#5%b~O|eDC{pJ!}_~X;0{CgEbg5@!* zGS_T)n-%O-@5rMMSsXCVm*iNv?su;~34U-Ad_y`1Zam0x8u5q%T`bUx-3@)X2D*o<6%MCNq;lnCjQcDh^K5x+2uDFRa<}DyFZv z(dbWW!*U#_zY|Kx?(w-Ye|v*HF__pT64I625V0s%QdoiMO-(4(;`l0}HTARW<`teF z6CEX#-##hIXfQ|QXz^6n|0bZ`9=8`BT9@l%i0|9ExIdrSK9zZARGNnHni!TrPDVZ^ z_CpCPN;MnUs!gocI|t7Dd#e`+$HE?Lx!M70$^d~GqDe%OVH8PPxdXK32!q~yuXWNE zwo+h<>{maYcfA9!GHd=xPH!TVsYRP~xBr7u>(Zf(?QQJHh?TbYAi*(yZe|LrCQ^n| zKvUoU3O#+|T<;B^<^+uV6~p!U>VA$mZT0#%DYrl8gW=Z^*N%IbGQ)3Q0El@D_`mKu z4YsoV#G?NJfkUeaV<#g;hy^71x*(H5JIb@BHiYyx@fytQ$;@ZN&XDn4*WGz`4YjP} z!KmPPW1#3t;>Y|aidS7F4>rQ~b{D9{EQ{;PrOu?*OEJnQrS~xYRRNWcDnLuk3K78Nn(RCU9O^if zC{7oeN3PYJj));MZ8S)qk9MW?KmUs$VC4dJ9S`S=>VuCh?cF|ZM@zQFGxPJwxw#jc zzBCx-JM!8k-^S@Ojer~@*vQHg*IGWDUs@V2pphYG4|p9Y-A)E$`Q>21wXa>jo~)G{ z4$tX|XQ7&=6ZPLsPQNfWXZ3!S&Prb5{1@~>1;p7~T`(>3^yu7b4z;8|#>f(ftN7QE zk1ss*kg*`(Tlbla?z=v5+@{~&iKmsX1F0l7wGZJpyK3S(F$GmW=h3Mfw<*(FRg57KUS9K@M{d1~9)mbmbpL()OHtV>)uzo(#L->A5 zI)_bRLucXTb*CTC?w$?s{h6XYy|WXg`kjsRRI+g0&$H(PCzJP*RVm{v&YoZ@jajV< zZ>F}jUCO`o(}M)405#d{SSQYo5Uw>cD0tI9BiOCV-hMP{Xz88BfuYXE_**`0RaLhW zB`W`JbYNq1IUX`5dX-DtR!Edv%J)c{D(2hh77oey$v{XasM(EU4TCro zGCBiE8MsAwAWZp6<%m@f$N1__Cc4p4nvv)O_EDR({mrN*aVskpK&)Y_l&4Zq9RukS z)wB0o^r}JQs>F4d{_=SZ=AP7}LfvRy-zXR*Eu`=9XT10z%Q7`QV%~(P@+% zzh_QdZpm(!++0(5GN0$iJ6rz3%*O1y;yGT?_0;xl0t=HfF*kyT=`Ah_Tw|mCX{h z>|~fNVPcAB;fiNrNu`kW3GX?-yRcvo;V3o5cgm2(&8a0f=&zHKXrQYHc7Z*=#W_xp z`j-quH5q^#XH0y2&8RgzVz61byW7l2c;~$SOW8CAyrd^Do-ot^vR#ss%g=2bTeKvc z-SPtmQbee%s0f347SCO7FlV`F)@pum0L=!n)XlzN^ke@0ag`bm&-X@KUr`qp1G15qk_%eGCL?9RtxzLD2xK?k*o4 zRIYkZ@}mnvk7t+==U{w zp41RgOxZO1l#Gv5e?o%@Tc8sAiPa_rjCFz+bVFTR^!Ky^0Q34PbZ6**N7NB+$onN2Hn}f2q21mq@kJn?+}WL_kPv& z9uxxCCuRbOge%8`F&64N|jkh-CDQHmP-e=VZKTLt0&5M!|RW z7R5kWd1`{WQ2$^J`vSPqv$XC?(gvN*tKJax=|CPy6H_Uv&=rI7Fx6Z?gHd=UH!l97 zCjUOWzPPj1uxhQBnc30RC9bSYN-vlA@!6fe6Y-(s9R$@nCMLltDSvni?GpU)pj5RK z-7p_IV=tQ)o|s7G>go!}=+le`>@l}uq~FS>@n~2z{cCO?SS&&@X=-W$iRU~srnse_ zqvx<=3B|?R7c`%>QyvjU~mAc!vkKnmw!9!-{Wyi!qdLGN6 z64-A5)yv#vKf6H90{4Mg#b$@R+bY$m31w@9e z2e7!`mmRZz4|px6^1#A##gBRgjHAMv48TH#8Q%%__jCGF^?n3sLJnL0S3EbUbEyJ8 zwBZsq*2v>LQ%rK|lv_?mw}Q8Pq}gEUGac8mnhyN{*A`uuyt+aCGm45YoB(xP(%aA{ z!tCpmEa?0IW?X8+r`QAeBb(hdv2{4xOn&u(AKK%sux*Pc#(|hN{2NT85 zN|e7!xL`dNLzK5vPw@E@pH`RIS!jQX?NSoDv!B$F>%~#>*;t~T1drQJyrVd@pESe?yJRggPmWBr1X|`USZmIF? zb~V5DHzw`j-kYvk3>!lpp8CN-3Y<3&PlLjKj{D10SH&@;IdsgPx*iHnU`|L=@}i}Y zObq$35Ch!_mw)kAYfm7ndSM{TB;L?ZJynupj!*I25l=m>Ma`rw)IbpM%gN&XY;9c! zr+Ggl55@ zXt@=roOi?%f&q)kiHNRLPgP@Mm~db|yi;FqWpKV;gW%O3uPgJz8wHhG%8Z{P%oA&q z{jG?pPHz>>4|)G?x)6R`U} zP0_3bkwc1_ToA82ov)l4D_U#*t8U(Hm@G3|#rZaik;&4vta+mFBnMB^JG+0DWZO1@ ziOHG18^K};q{C)mxkU?YL&*WGPV?p7nVGK=30Pg(alUnYsvDaBBhv6soUUr{;LdVI zBF!}icGr&9y!HEG@r1u@c_iZE0vi_&@G#7zG1pk~In`5?3ftyGNub7o(=ttD8JB z@jf>aD};ahiSeHDDIeCzlli-k<#pX)fQA=%bqD(hh;x0IKD_aDqyATic@l(l1B{XL z6omE6A}47a!%nl;-k;ju2ga76v-349XQMtl_#Hqr6Sb>5&R5R^{+?WIdSbkGC>Bmm zJm5EUZJvsr+JIMgECSKGfCqZ~X+RW@iS~oGzlC?NuDwIcuH!oY}K6XL}wJUg@mGH1| zyZp(4xOnrS9Bx22eb|RPD51CRr5gm!vt-7n=kWXZopMQIN^|>?Tf)~Jn9CFD=atC> z$7*MlqKmWSnXh$bk`!i&7wZ1nV%Q+uza~WX-m7MyNF=NLqc>%d`B^6ng+$j9ddr_A`) z;=sLRdF=S=6Yv-x(!z44rlrLsCQ2wOf*yskO+%ng>4hw=Z5_8IcOInkWU;URwG$nLZ{+Si=8LWnyN-H1GE+>x!ne~w`rUmN|G6nwc!+XS z*}i;?RL`9WJcFg-LvOG`+<$Sj>2;tD7vLI_-dnAGw5iR(%K905asL}bRFgZng^mV^ z{HCGy+@YD3Ca-^tpUQ!o^gg_5{6wH!5E&n36{b}}B;-KSfdR(gS<21zViBK(r=j~% z_|lpzd4h^bTKOtk>1;fz_Z6>>30Ve)!81-)aB9L0z-0|G$qI)%m`fOQ3n|W)NPL0* zTO8M4mN-wXvnB2dcAC;{*M~@tWW3h;B;PReF={S7(^o5>RVKuGu(nN}#S;C9VeI3w zHEvYCrf;;x%7r2Sla4c&xydXng})B@kXd-*P~&7VokPNEFn!GY^i1gNH{LyjP;+mI{>S*p8&9-CLbt^AM$!c*pC5$XjuZ&hx2)kd??0J-ib-bs5EpWg<-UPhD z2!hX_ur<>Uu%IASE0y3$8md+aQbK$hB{R)!J6;O+)e!Al5s&lA%B10z9@7BxR6eRO zva_-IRFB>O;(a}?iL#~FzXmn4p53_Vv$0;x)2HCjar4lB5lT?m#m*h%-!=B9&BUU! zCtmjcgfyXt@Xx`uOAhE(%FMhBNPlC3C^e&g@l535V8=$s1iQfZ57#aq&yv)x(W$wQOm%LbO*@g zV@Jk0rsxy60r4Bu_|`kmqTBOfKatk|o6F0>%qng5OhWvg#ybUc$cG-8)H*3?Ehcui z{0Drw3R!A{B~K^zf?R&nob#v#wB__eiv9|3e*@KvzR{?am+33-Yd@ExDN&M&ih*?} z)5C}@Q9s(EM+o(NiODT7b>f%QUjf2p1giyIbY?2{9X(x}p;7uD;qT|8@5)5u+rpaGI!qdq+CJq}<`r=p^Q zbK`Pw?_v!2czK3>U;-W`E{#&|P)6p(#q%8wpR7r z-6{UUTX+eB^mwSloKOCnO&@#aB2fLBiciw;$NBi?UF=ca8rQolx**7he>~4g>Z(Rk zR&+XeXsV<*oZjs}L;6ii`c0OW0N8F$th=}oxFkL4T@w~YqhL#S^UV$NP+@yEhef;R z0)9`nmxjh63}FJ{)`L&qHHoXJftp1CzzDba_?&*U=1>2Q7oM%%>|_eQi~k1Y#v(pC zla5MG7C!7kS1pK!YFBDcR7+i@OPx24Q1x6nhyvr_3gB*z=c5l8kJhvgr98KtyteP5 zoWY%z4l<~2A{0Z6)re(*P6=bYC2kGOhko?Fui``71a2}#FN$g%I-#&BwiE6zXs5-Z z18c|h1CXJiUTPxJw9|}E!Kn$b&YE$>*|}|2==8p&FXw5g1FHx%;M-Wm+sKstxpa9n zxkcUc>^X;m;$CIO>i6s-at1*CTVRw(@=k%~ui2HdYq5?pboBbQZ&+-8?l8|%T75?b z^l-#@kZHwhJAqrM7OKx~a?YBC_G-!x5rwVNiSLK!K#rDAWV-Fe_bd(k#`dt?r-cz)XL>BdDU{Nn)HKM_4VNvByY=*urhkdP@ z)wPr6wMTzAi#`m#D7@pXZ%c-0s#)Yde~xBGnb%>kI)FnsvGiA(z>@2jml^Grn9Gmm zmV3EzBKyg_UR#o{f6n~gtG-FC!Du&cc|k^Mt-ICho1E7EB`PcL=|s0MShZ*U-bEkr zEy-H&yFMsXYk>@qUw#)Kro2rDRn`4-2lc=T#ba~jekD>?SxKFc<0r)+@OJLWjKEmh z@(Xn9!FhdBv%$)zI&-dXCT|03qw7*pH~0XV=8;wIl+lDf4#w>u*cjjJkCCkzrs$To zf%_hkE1#?57VriBUK3JTEi$ACrm z_V))iw|o*3hWi3{y2Y?3`*vKai8c`-*fYU)?&h1i%7 zklDKzHsn^C*G$Ua7H+=mO6_KwxX$$9#g#LF?(+S!~zFtSmM?yxeiIb?K9n|h#Po*qS zsOz$O1>m+#LGaIKuL_6c{%hh(XT3CNhL@kuJb3ZO?)jd?$qwJX^Ao*oBp-qX1>G2` zpyH+z+tcL`$CPR5A6f>e_1!>YcXyOkg$p^yp3rm_hM0U@iv!`+0N@toiNKDmLx@ggH`tARbKzb;(x zpJ*6`4uU?%0s73W-$$Sga)0 zF>`&7iOMmVz&EfQaYr5Pd+m82ZJ-Zn>rv=mP7utplU2EQcW-ac(a_KFUbfQs{>YIq zk_of}7f(f&dE8o!caVfo-jpLkMlfdLGE@%6{X)%?c=zTgHdvf_ z?@xR~9p*UI>e2Y!CXMm+DQrS_8ZDTm)GsL2R}OUU;i57d@rk{a@cKU#ykWOZX}HW( zAa&Y=<%GY;`p(wm{372w?LyDpD86Vo;b49Rof@Z(I$mnTC*&DrJ`$)L($SkFYXbtwVVv_O`p7m>Z-|=#~1j`Sv(_qD)a) za3g|uR!0Mu^3)t-n(<#92Fv)+XMx%ftkL>LDNG50{fAStsM&9#m!jd;J&T-)unn4k z3MVClOURj6#Pj1q9hy{UBJSGKJt4YaCz;+#CykYvl`*SNt}C+F1$Q#=eYCBnMZFz_ z7_HXDz2(wc+kfaEwH+@q9PbGoQ{muI&42}j4MOBRIsh!Ux422w4VdF@;_;@(1~e{N zZ_aeUK*YS(MBTlr52!_*VeK!f5Tf=VI^V#e8Osc~6SF!qd+Dn%O36i)JlES7eKYj@ zL=Dew1RZQhBYedjFNN;+FRJzNcb4rwOX(F717nL+lr+q|L5pJc%D#l{)P zCFHA|@fm8pCz~k*sX$?N74~%0S+5)!68JSXRq@lk{PH6hPwh*-=P^P77e(g~y(%MEK3JsDKbvXer9xVgDKkPx!Fb){;wVFgZ?e~W!So9l}rEg<~?5ZAO{fgH>&*XVKh$cUgjfXAUZ9*2vr zGwg{2y~{7Y+YOu7xE>$wD643w^uHe9Cei}jv)cxy*U^HN^$7jKRjcWjUbCO=s-5o3 zTy+lKofnvx&G$GLEdpp04ZvuO(#AhbiD1V*L(A|n^Wnt1*zpv1QE7C`GMc!(uS8wz zXd7ya9I?S%if0JgzF07(bTkljV_y&_AdoXi%V1*Stc~27p|{31r1whqICK0BUnoldH8)Wb|A8K>M(dpxOY_MBXs{+0oc$(+sImj7j>Gak2V-9ysKMjya4j(w zB^U&9?BNSD?rnnfo)$=}=dqo8M=>V-59e+=bjrCe4;EXM58K+=4O9=AE329~oZd_~*dX>;C+0u?y~)SBZ>6L28al0CagR89!oudW zY0cXmC&UTy>qS)6%?IRCn1xHK)@lPbvED3H;;)o)A{}!gXOEO$TWc6jn1r;LiV&?t zWGF~jE5`@BhfmiPvjLKV9<M0;5?~B`7z0*5h#K{vxJhZvOU!%aSdxVqkX1?4~^r zJ905_12o;-JP4BR;g|zJ>uiS>3w;;C`Xd!#jSccEI=Xz_?@FWl&Kx&4Ujz^daIW6fpaf_K?QCQ_wbN$Wr;|+7g3G2&;AXdNbOpHDyQc4 z?vn^|X9#Q42Y?U6!xvoT`q*iI81~%+H5%wWK%fREs9A6VhY)>a07!-!c=fgTfgj`U zjZgoE)2{-bn$UJ>pG%NgWcTJK*yb72W1tNM|LV7nX%KU ze!F8FilLpDk^~>pGe};*kWuBzfOh>i4Da}koJUP`j~*%ZQl_KE=CVA)XT=MKCMv-AP9+kJzU1x_iVq2b2^X7DJZINqXu-co7yC|J%`3sR1MA<6D+IsN>BKMtI8wR2RFmw|PsJEP0mt9bW?8 z?#tc-c>C*fFV&dEm?I})xw2hm1p7j8dh!WUJm#+qL?9I^u;Bhi@&0{<$W=X^Xl#@I z0aXgGOLQySs)Gdm^E%8dMHSMwuH*_wy?FeJq$$v4I5zRz%-F1wwihPP2yVeQqXc3iB`ay zq5`{v$L^(4l?H51eVDHuY$fsee+@c|h#Kiwr^v|3q`kp*!!P|mQKhq{^<57Z$H#SD?>FF7j1`|mx zT1@J5j#G}YZo7-BYpx^(_v}UIQzd8p&kxUgup$mlUuT~ zz2zJaSrecXoB{jiHtP-2z@i4-E3u}R)S$C9e``^6vV+^CYiQ2?p`VyxvR2-)hH%GW zo))9mPx(`8HE%tNbt(&6GfVMhk;eSM4_+TNqN}{fBP&h=*L>zkm$|tfT5$ftEkEQj z*bTvzR9;Ys_Xh;jaNCkc_A|LrP0MU+B}@}Ceq!=4wOWsVjx_At#k?R-m}(~y6IBA{;sK*vNG~lfr2Ar59eucdHLijp&m$#cYn%-w-3dE2nJu0Dqwu%l=yz_zmY8vg` z^~B@gJHHP{j|Jcd9++beNnJQLO|tUJE^NN@ts%I6;>4FwVQD~D@oY0INu)O;|Da*Y zt;&)pSJ5S0(N$-2{5b$QF~v_$%<#WY%;=L7^9nP1@Gc1-K|R%wN843AGG*rV(!(cb z`D`(>5V>X|qSsM_13@_}Qk2T1^vxqSeh{KSv&rz1@6Y`udQsM{e7$~h7aiTo!LCqU z>Amce8An4q6a1{gYmz5+OPEb>g|=Kue*(iGgvU};7#AhmIXFz(m+y>S_Ot(P-r@1E zAf=L0`H06=kP7^NIF~MI$IuSE=%cjIpeH<)@ra3))zr>Yli;An+o7R52b)6ce+c=3 zm97)RZGrEhgAX$D%xPumHiH>#%P{LwR7Xyh3INyG!g)O3lz}ULMh{Hj=HBlJrYP9| z!A%1m%9RI$Qf$#v(~+7pO+rVkirH%KM?j@xlm)j1vC75VCy zzU4PC`?k_`f-tamCe7GtGrMtc_3GZPxOc-eL(BA_gtW4pDLi8S+*4X}k6!9_HIUpR zyB_zBNMfCb*G6dIX*_9k&mWE<8UH#I*MqUu-0uE46ua{u=nThG zymd4pNW0gcn>(C)pYKpM{@XjrNCZc?4&3lyvRK&$Fo{%^?9C&%?emaF%%aZz$f(h~ zZZDgT9x0DR&OnnoLeK-b*MXEpI#*V$2zKBfgKGyonvVSD7+@+orsT(3FZpjEzN@C$ z3fYR`;~~2>7)0?n{~Lin{$aui4nQzj8ja^bd<+C&fawK5TYxR-n4+rpdvhG8!xbG5 zmWyFs*Z6(`_F|OWUM{fCChl}}T5)|@=edHmH6*7%8*FbxNpPl)<#xE8KMh53&sad> z)*4yj)|Y}8mWTF%st(BVbdHGRBcr6;#|TNRXkOhe(fMoJ?-}CrsYfw+GYYNl_$S-e z>kZov6iDG}nKw==Tr0;nSb9j(^F^sdar~(F20OR^ROBLPpP|KVL!~N+_IYiM#hKx8 z*Sq`_sok@96w1uF*^MsVS)#FI-%tsq4Mv;xBcpuk5H`WV`T)ciXNCb1kt((z@~>BM zCk0$dj>qsbITn{+H@bhfYI=>EJ9;9(oTL&U=>o7fphez*qKVZuC;mD#?`Gld27tf7|?iwuMkV0o9~QHA2`%(h^?Mz$IOF;so3f4gJD)m{zk zo!HC(5QUY-)aD+b?wflGg<2~O6C?1VzFYb03Y7JdLic3q^V<_AD2qDlod2gdfvaQD z`GU-G>jDq3(`r3G`Zu`yU5zp@6tUotOjqi=9cfpP`9_5O*13;Zup|i4WtK|VoU~;4 zz>Xf7BUfUZ+B)pQxnLZ)o~T&*J752|3~^n$FnJ}u{+JSeZ+B_HrjDUm?>X0J%l-H> zbsurPk96JlgZExheg}M7^EQiHhVgi@h(g^zoBN9o{d{` zp4(Kn_e7HvjV-1+Mb`m>X5;geq5pYS;**A$RjFmu1SI^dlk@)QzU~!tD<_)Iq`I2h z=5MGu*f<98XFjF;?w=@=!q)b79B4uZD371A7%k@2eJwo7S3fVo@StOQU?n6&=&i&k z3cx4yli2%J?|p*2mvZS+H3Mtwzy2N86ua8tzRj^5E$O5EbJi(l)@*7h5TOiC=gTFQ zwmgNv>zXdFQ#iqN$U4=5pQffi5Dn#j!!M%&X$efxd@3&w@C!U(0qfil0s`X31~gL! zIE*;?4n$ugmqG(YW%0Gr#h^eZWYkPgSZDv{@C7lq8y)AP%J%TSvPg>4#6MR~SL0%l zFqOa9GydGlMV?>dYNVKY#(* zSB2S$yPK}R6@nhuHjG`S0$13s=W=QNbziY{o*3;m`Htw0(HvKGPL!Jb^pm1XZ%xAC zqzXTY>51w!x%sm6yU+aEo+TBpmxC6*!>!i~mP_URBSG3vglEhLRufla63fAX)f7yX zelj*dpy>F80(im6F*HCq=8{6e%8EGv0{nD4A%`R!1Ulu*kswYN`0U0K|KwJU$RAiL z8w7FMm;1fdD>qPenA+mCQ)t^~`hAgLo}wV`$kPm$mmspiTG*@~M+vMy(jfExe#fl% z+&;~qN0i+QA}@<8q^a@0G`$nM5h7q*6PK8HB71H1YtHU(EgGp9>*JJ)D3UvuQN#&c z8ym1eo}x*s)vgQu>4kZO{e7?6r}d&Gq%5ZRV~%Z7k(2^k9;(AaIug0PAwx9RmHe{| zF)e{)^33YpC0rV*avAJ)taZb|c^{cEJe$zoCInkfxtL5W40)UJ_TPXL z3KLqIYA4o-+k5!dxmVAmqIs(KMt*T?7&dQ=f%ReCxyX^uZlQ_k;^T(*l_OKWdF{Tx zJ)Y18T>5O<=@lpzo8#K|h4*uGImM3?xw7X+Wayo{M$VNZ0*>c( z%E_k=@B4}M?pMuGFVX$&r~D_9ne?7l>pZ{D_6TceI$tXR$*84WxcxQC|}`@VZOT0)!sC|naO|fHrOz6{e3@^`EZ8SytwzB{>-TjUhK(x zX*tt-dr#Sa1zvz8f$?VVi#2}y)s(&>hQ1pygV613bfwmnmpND**7rJPQulIF!g0u% zucm8jlVY{(ZQYzg)e#9*dKQxT(G^S^ZH9M|L}arcOOMuZ^LQPvX%DUMj=ZoW%ab55 z(|&a_=;1gs8**!@Z1MWny}CnM2dN#!ITrGRJ?>PKj9y9|P);(eJ~lYEiB1k``u-Ob`c=r_T_bmU4ufT?&$07V*co znw`8Eo%RF{JmDy~6`mxjz!eMve?u>{!ItTSbJz$?WPz)0r9w?5Jbm#mM<_82rOzXx zmQA{AN$Q2o+<^!AT+A~5P2{^3SXVn9^)TWPWa@jG4BVFq7k!#0-7smF#NQzMb^bYz zTWj6CEag$Ev6V-1=ky)Y$Q%dTL~KklFJ#_M23z_(bVNsA$HSXkTofMgB-*aj>oW;UPtTB$eBaTft-t^l z$nSP?Im8wYcvbHo@Uvp(*rQ`#zE?(S~+EF&<#{V-X)d$DD9 z6EPD$`v z{rijqi*QpccsZPw=g<6uNBqsP+uXNhsC&-(`Orb_XK?V7i@BSprx*Y08DB&7{$--} zz@v&kp4M!~CADx6>={IWnG9`taj)nbM5`dEEO~HK{!`Wcm)VQIcbkM}%0~~Gq0oa8 z8yg#D^{*JOylUUDLVfY>ZoQL`o9~CzHRId%YiVww-2Ug>6Cbkt+~s-J&Dg3RYTZ!J z!f0}5T!OHYCN3kHH{o~TQ~9fnK_%DtCT8w%uY;;^4J(Cr1L0FOh&V<9$F$g zWJvz?#)LQBoO2>>F>)V`OwCGT`KPa|JdBdgdHxM?S(;ST{~pFtk+JY%;UuUuAHT#i zN4fFcqg}dUv|~dr{8M#v!ScZ!K93h>zmL5gRs55acoc%dGyTicF$Cfpn`c?u^cONX zQr8zhQsO04xx3cY?fUv!E!DX63?~s|h0)S|l#rHHRWb3pySU#C;Y$J+z@LYdm&n~$)PZUkSV}^4QMpT{i|IqTU5@p8OZj`X}mf-U^%@#TPTtz029>uHV)fRGdFjrMcu&fxsx**pu`4<&t67SSrSx&)tP%^)P z#ES8`C{uSAt>`onX9e8}l{XMJ@E1GS^?HqJevq!_*J2VMf{!j9XAC=Aqt(cuXprk- zwaATp$zl~-dH`DE1bj0y=VE-g*aM3Syb`)}!^BCv&xIfCH1nvJl$DdFbcl&sirsLj zoR@CTW6W%8`vblh<|#FbV;fWkC1?K4vmeOB=aH9DHP0@_Uq2h$ltPaMs^;TcwXnBB zlKgC--FawDXT)G`2Fa+Z)BtpODWot6txsntTMyI`Q;1(RQc;Z%6BUK`G?-SX=(to& zvwKbK-%$Z0^X(LA1vigIB=1+qvg*hI-lgcsJ!O5yI^hX&sbE^JgeF!y=js7jndE7G>TvF`94&JRkU= zmJ7zaU*VT)ykD@#4O}$hRhQl&SQO{k+lYMay7%vvbn3K;!0Jh<%m{QK*@UkjX}^JFw>0Hl`6r-k(bino>d zrY+@)z*L$?yV&aG(KPKB2QH|zu5{mNR_~w2QaO=h-MZ(4DWPK;G9cw=^!#{Y0Du(2%{=%Kq* zfmOfQn3A@U`?ONa$V(&f$Eq2SE^$ z9pCZaoeZSOvM2dD^{>*5Ng+rA%q*p9Gmt*KedGD2uqFG+T>_5Df&Vwb8rpuwk-cmO z_vq9=tkPuB+>@{OHB&BZ4%oO$CH|WDOQA-+$CWqA%D6_3F;Q%wL7pN87XBG94cAQeebiOdk<1sjs)oZ)EU>Dh5ad6DyZ! zzE2als6|@lEwuE(qP=PC&6$vurV_^_-4PG7wW(aLqtS;*ss;j%l98NGiV>HIj(;B^ zEo5m$RHa1E69bLL8$n_gMrdMZFPCx#EZt zVarqV%X$*Jw0-gXIY=ddChfP{BzkPu_>$mu0NHgKIr(3{e66iLcLv5UN;Bpr|IRf? zqwmN8OA+s@H?K?a6d&%KvS=p(A3WdOe@{|vhUd689IZlfF#KBuS}VFIsaZ>`Qv)yT z+qTh)2h9>6L#F5-5Z$DZbo6k=VGl}4zhY3i-e)U)3q*xgtd}P~7JVG=%=n+B(&flv zCl?VJg^8d8My*L}1{lgArpqVydz>;?cP5d+aI#we?b!!*^vI{0A}BA_IxD|dZ{sew z9+&9cG0_*#5SeYHR;oaw`7OFSLVrO#D3OnFzwzALgaFH5(Ii#x_>TwGZG?;$DH>GZ zGJIz9O|(x#Z6#J{e*iKyDSwv9frUl?C)&(R{CWFs8|LP(>u1bY2@?2Cb{c!baemLl z4?bLCmcH?dF67MpyO9}iZmMfMyi3F?uK!joVI!W8o|$+hL3q~(9h020<{H*AiVM8^ zpwYM<3x*FUFU^ID^@{GFh#0g%HguHnA>kp_JK&DKb=FD&UE9fVW&pXojWITuN59y@vd9BXL_Gh$r$f!r9Cv@n9i<6{Y zLA7%X?2I&B4H%~&`6|TrI{lOBpjfP`&jWpXa1s4qL~O@{gLT`T$7jZ5X>zAUaEm_D=CDIOL+A!Q&3dWj zkjxukJc*Wgg5_OXHn>>UIWch~ARyp=igD?+ncxPUjjB@Nf&>o~F!_w^R5c+cAHn^(o8RZd zAnfYW1Nx7a3njjWo%@J5ojWJ!cBg+lC90n}X4wH({yK>zJ`N8{mH|)xdM^u8!C$Ww z*zU{29h6`2YO$h-eBy84&zV?aO+(}qHk-XUO6{EsY8A;=)BcT16;j%1&AzfU&~18Q z!c9Z8)92p)i?Hmo$6SKexlG%R+_|5_cl8%vZz~MmI@>NU(Q`vUQ~wlrib&^NkZ;4) zR!I|~pGkIThkS%HM{CL^500a@{7;FRw{|z|!;Rz02JNVmA zWHzL~J^t2%f?*f~$1S5QdD?&eBsoP%ea#oP_R9`Htb^kch`dI~rRs=&^XOqrZqchH zjE66>?1Vy+MmuHY!yR8|o)tO+#V$`CJ&@y<;-{Rge?tHU-am0o##~I&5HkW1%1x6X ztM>HvdXL^+)QulI$s4pd`fD;A0ZQGcdP-qPM$0O^L_jL(&$gez4yoxNdmn=AXXXsA zQa{lYh+tmbS#7CpfRxCS^yS5@;u4n~<_P7wNiM5$CS1FkwWNT}^mF_!BV8>iBk5SV zZ-%uoJW+dU>$u^08U7Eyu2PR#FyffV2W+$&*$jXA7=9^JAC2%ru~^Nd&DeUzN9pdB zOnNL&PiAxs_21*G-bxyMrM4oI-}M9EWbmS_y*~-{?(Wh(^}1q??3cDB!bWbo%ca&y zkuIgMRbrei8}_)r{kl*f^!Me*7j3R0HtziPAB7+YwGn=|`(>hwPXVnXs#0x3xn{iu6ZruFQ9Jr#7P?qh2jyBTMze zFePHQrXat?!uz_1>gTW3>{LL)=`z|SWS`PL=x0@I3_&ZY)vl=XUB7f`!t&{0p<~is z%x$`O2cPrZ+X;iqX^=Jzz^OzJ5Zz!j3TT)p8&v!xZ(v{$6&?LRhLF#)XLwkW`pcea z?$0RXvz8Q%SY`T0H{Z@pzJ?eF7tNvRWVjw@>`x#})41i*wlKc5mYGYMamreJPJ>#xrBK0fgbX~{x?`A#t9y1$S?qxQDB2m2Z}K3UML z`@|t|u>JMcZ!OZ1hx(_ zqH51=so{##yw!Qt`r{V7iyN*=kHNc`o8_D1GN(jqKD%66+BXA(Z`c<;ho4DHU8I+y zLo#7Jgc5`+t#vf{hgz}I`NirVBxN!{#fTDPBXqialw+_@q74_%!g};z^3dGoc{8HU zew73(-nygOcC6T~lqm0?L#VD^$fd~Y(`|se(NfyL$rAZ`FL zb@}4?DMQrRDZ^W+@Hek(j$FEoTd{KSCor3@xx-bwnV^r^#&Gic^I}%ttLwh+wRA(>)0OWrHSJB~0`} zi*|z@;rm#W!LOTAt^_~GGSB8PRQX{NG5lGv{dG_3#CC?G8-rBjzTDZ+$MrSD)%arp zniroj$EsD1Tr>}8hf2p01RE;@nud(TW6u&p9O2$HE ztSNW@eK!Yu+kCVOUQE~MkrD+ETusZ4jH;5kk^f92N5H5?j?nvYWWYH4S@hrotAlNG zCvBSGfxhuZ`;nqT)S=65l;?sy0m(HQ7B+w57`ey@K)hun(^;R=v|C5R2kZSb@|l|; zqcHd`EijR6go?__T^}Dyv1FIScaMBSLcj3|$DAFE?V8nfbc`Aw*Zwd5H0u?aZ`}h$ z#mHg5B6|S|erniT#j28_M}Bc|F*Z{%CnCKS`UJ8HAOjB`KX7P`cs1sxIaR>Bl_&lR z4YKv{a=fy^ERGC9aR1phh)ElM9W9>9Mc!4YK#dPiS}4SxKf2-a{Wau0J0WXBpriJn zm9u5m#&HOpvRH%>r`xY}AXEe7^UWbAq4!B%)!U&MDZ1N^!lECk zlE+Sgdhyt>%`r=c62%yHb@TA_d&Dz4V7Ur)7d(6-v=!5^+ra?RpIT{}4C8#edIy`+ zy{0a5hQlTpWU=A(iKBU6kZsw|47S|2n4$s|L{dZEM-XJ4G1B*(U%w&S|FnIWS@!na zk55Dp=viqnv7S-C&kwvskHC^~B7ue3(WI$hB^bSC)5z~WN|%+ts%dezYDDAAf;T^e zoSU>g`Sp-8Rq2b^{JlEu3Qx~nbYDw*OOlTrn)chJjR9%o<;Y8s*B6Y=gLj$QlI0k{ z>(UOQ5X}C=E;|1Sq60(nsL+)+&0Xc*$Q#kzj9PX(bUXELodM2Gm@#1^0?R$a~u&yKqyCHMz;KAf#vLJGMo1=nA5P(R?)?3fD2pV8m zpm?=<1iOUNWv^ec$H(Mj!3nTV?M_hcW|SCCe*P_k@V*^=i5sLEn|Jki!-wU)Ia?7Y zG~4ivX=NnGYTo*0Ty!f5<7g|pN6Oc`8VQr%{1Tx{CP@zVc;VyU4uf>fOuPxM30Ef0{vAmmDkir4 zwd4Eu@4`*JGn-Bud;AWoBMr;;IVN4UrKnATQFG(N7uTDZuJ@@P?5S>tc8)`$&+>9= zaWOGy?-06jqP-ssz{TaCBy51Q-9A6Qbp_+d5ng1lyS?`3F4*U@^WnP9&i|<7HSl4$ z~eg&X;>pha(HrjL*tX|oMc@Ut^nVFeq zgx*FR;*cs0TZr@LUZL%%y@TCSM@#d@Aq))=bp;VFH^qXI7I~pIlhPvF81P-l5kDO( za)XKg94xrL)10ywUm0hR^d0drN1Y}FB>Wi@D7zMl*fI51g2YpGxV1pF1HCL>iVmt( zaJ91!OPA4mI1rL^iKu;k{@+if=C*(BxHhq`ulnzGs@jR?H{+b<#lu@b znQCB*oVnw=J1O!|of3vUtEFgqffV;2=}){V^;DPWq)BVTh}wI_OL++p83hId)%Pl_ zOGFYNYh&M=ySC2e$bbK{g+;7ZV!b^__R0qgr*^L4ufDl+ev{rt!hNmzZXKq@#w1KR z-5Xd;3Ex^}#RW|<*NtLcMgJ$teGm{KyG7Or{g4*DW<9ncNIFXFG7RQ|8Y3R^gJT_8 z#9wj`zGKo5ydvz2YX!B)pg2NGOH%-BY60-tbc9pPtlX@0uWi|Wwfp8}vl){pWySXd?efw?o3IsSEfw!y!i1&5A03D&RjQ{4zWO?{(5iXf(pLY99p#5oejTK~{(H(}IswjA&Yqqg z9&%o=L^oX;r{v+Cn5n%+)s?VgaB(|$XDuLExo3sAV>|BD2v?bFWJ%Xk6lf( za-zUVz_FTo=+(`x9cPYdV)25eq+Pa1D-Vs%gZGzyr3z~_NO;@aI5ju6$ zFU!B#a+(ubNT2;;=usF|aF|;XQlWOR6v3Z1XLLt;B>moHBjr9%F3I~cKN?pU&q}Ue zoU-}Z`HtWK<0-oqaL7D%3xui9TPV${WXNI=))5V|Yrt<6D7w>e@bu(yFmC2|1cm9) z84t{@Al=F_6poj9Fh|~uY_kG%UyV7EVh%`0ROq}S#w+st(*yZrIfIeVH2MA+{h1_V z6s3jatck|$K^$$t)*UwR$zE@NMqfY|fDW=n#abbKQKffcA4VrojSv>0FS;^LdqRh(A@#clM{~%gc95E%y({*( z)E5cto}c_*MhjF^n)z;{hbez%bt?S&EAYzG(3s(u$~iGnWg^rIdSj3s9P0i$_mGeI z^RigP;Df`!!g+%LseH1#>Uh^gs-pWko)hZ#94JQ^S9|=`Ju@**Dt!gh13bGK9%~@A zo_qCj*e3$w?#|4J60;B^z~YUa)zRWQE~xvV5Xw|Q3>oHTyHeYsp6N~%Jf~@+EqY7q zGbar4;ubF~AsHDN&p_|ba`()^?QeIH!?#F zIt+^WSr_fgm(c-j41%q9YL?qq4#^S6$xBt6@vCc{e5)*c?4CWwV~_37gn6o;hBlmrF?x$0sTcX%A;=lqu^ZQi-2@%~M?ZG{6 zEK(-666{v779-9?2XtSnkC%V8ebDcw&vPwUbJbNX4`vU=ixX8ff#2ybI8>3;JVGDs&wsb~z|9Q)PC!N|;s28@%1?Sv;f=jcvQFM{ z5ca`365jLYXm-rH$)=~bH?y$C8!gd@t8Kn@_#AFZ5I)M{Su>Wvh~M7Xxq(D9^=o>> zoAzhRe2TCtm*h7O$maQFTU?~IzfS({%v06e*`gJ+4y|W=ih7hU?Ra5SJrz>lMp&j9 zXfF_jCqGCvO5&mPVjmt!$2R1R!%CNOa7g%%vL3nUUq1(to@P)da4D%Xt`)nkUiM*@6Cp_;H2Ydd%xn3nkdzOBl zlUem{!491!jIBDvkEu7u;hT`rKlx}zCr#!96QDQ8j!(N8s$o<8fO(ets zLD?X(N$mt>TlLN+A%}r;ydYSbe=_~}Ta8I30pC-ZGpUu*L&Z+^#)JPET9xG(Xs4|w zKWQr&(V%7Ny+q*{W}b|#N}KZX-rbkEQ*N49ZHQ^l5HF~eeX!b#Y5eh4%IST3YAF`g z5op1|zF-S{D?pZN_|tcv<9LnI)W;wuFbXf<-YYBlF}6Xdu?nNKDKz)!KhN)wnSyH` zrji%9m88#;@Q*iRBo8T*#%u2%-f=r8^T!H*nhySUS= zD0VP{zT?V{zwv-$oqoui8;ZmrMpgF&mp1Q$$}fmK&Qgeuj@C9XKnE=m2_k1liXLzN zj6*JTQlxARDbMA51*I(~>M=Pa^XsDVuz#Jt=#V}-a@>O`2sZK_T^;olJGnQk>rJ%- zX6T0|DMN^^?Y`D|ROWkP~C5#LCk->vfaa zguSXQT~ba{%YZDH{bpx}o=}+^*AVjQ(ZsB+(Q@Rb%ZXecbHHEYmbzgSgD6`WQFA^b zSudV3ktNNQ2w+iGqXjF9*zpf9s%Yi}RA~@bA9M--|HD8>a^x4qLM*bN_wU8y&`|DY z{Ss;BgwbUhIx7m%&!s@Q=$BaFBq$c6K`o{oe+(8*5!_42de0TRz3s?H*@#d-IDx20UpgDs`2VCob9zc#rF^83m|cTkYEv1)?*~&0&u)fDp>hfm3ya`(Ctt3){kthr zH5*R$W;oeFu`-D<4!*%{8*sn=$uTRHFddrL6>FTq=X_L;i!Qz`S~Li7o52(f@qC>u zh@yv1VfO+j_WLG+zXU%~soD&N50BVh-wy2DSG$Sw{vR1iJLdmkgWirpt*orb6F*x` z>zn*=zQapN8lDyVXY%3yz`1bCgZQq~b&}7W z`5^gvfWt#SI_lGLD<+x#(>bu+E|2#uB|r{E<^`LS3qcAU7ydH1m`~B@>AGT*I5{$` zx_ihVnDfwq9DQFcPGEi(#KX?GwF&E%*GWI!Itr_|JENDYMMBb%xt88hfMGpKB&*M; zlTndzffL(}hF-Cmt1de5Wm?ac?P#0UvuT^eOKYWF=-)U(*H3v*$0x#{ev};A`v>GO z5j-y0gW934ft`GN*4nKprl?l&%QLj2DfUCs_C`CPs_d2P)8>RnRVqWw7ib;X{Kay< z4?hfcyqBRmY&!ft^_wJ&>sVnxI-`)F&A~;uzGn0M_4au1;64vrK4+2NRvi7Btg|LzmKT%f8;@deR5Lb^Bls5fQCW;h6uHu(1|MFRpZQk z0}hlIW7KaMxg~AnvB3ufnJdtcdSWg|nV$!wzNWU8BP)C?_Vr*EK5hp2g1tCN{-m9` z&7>zZFw%A?*v=H&6LV@vLvWE2;ljS>OuI=(?Z@aaAQs|+>QZ4}#2mR_KRT$=F)`D@ z(4TEQR3$I+Dawd5Ak#pbgh=JdXrEX~c$0a}2hOGM7fa0D2g5yPjk#?xUKq-yKmP&D zTPfewf(f|6IS<#8S$-3kN|fsG^F`mYtQP_+>) zLDP3F#&1V6H#%0QNhmGXTMU1@j`y5JKiDy759j%a@_3bVSp223Bn~U&i|6B?ue3R0 z*(EtTYCo%nE%L%Bi`!t{Y3uZZN$dk{q#ckxmdnoJgF7Z+cZgyPlI58e>;f!*){G?u zG>A|rIvPsx6q)$Q$2@4yD)70RsHd&^YFk22@Ip_)<1P~kjXSJvDJM>VeDILOywO+o zy21=Tfor_iz7mH8PYu>|@)gZ*&Oe~cmt|lZbUG^k8J4%9Vs75+m%(oox3oqO3LdFM z4j_>XT8%cPx?P#VLl+R;yTSZ_Y6o2koRG?o1C^{sND#HLvHA3x3-qNy-n5e|g&F9q ze(^K*<%$&S5DZ4KF#6F7dI?1R3fA{rK{llf7V7mPMsCxEeTN^G z1zp~_0UHbLt*|HlN8}SUcp+Ev-k20T98l@`vnkHtK07(C!<4C;sKNy0|oLpJBNk4-zRNV7dEWYrZDM5;P6tTq^C0Z~u}gJrm>c8#7T0 z0+|L-FH7TdBC#lPQ~2csa2Usipd^7Fql>4i*cdyW#LZYxJ&8yd0V2V>SG#&IvELFIdfU(c*D68G$fd4 zz@hn}$GBVoc+Z@2x76p_CMN#Sa}UKPK)9UkXo^2VQud1aQ15Ox>E`zT*PZY0{M5C zY1|=W>UOPXxfDAG=p65 z(I%VH7G}Q5Wbi6ie;bF*X376@uv{ppkV4qZtl6v5{Zq;!ZI1(k#0Wz!ORDx9kw4b- z>#B8%DaY$Y;jc3@VV{p%L?y6nWYg`w-8PG|zj-M-dH7?dNpVy>uU#uH3XWL|k^)xA zo+BG037`3u6+_>6IORNT=7oZ`2McSJSmnc9WM?0=e>32(T&SU zj~EQ4SKipR+f>LFA;pP2k54%6iTkpI7jpVdj~!rXT9>7q(_PYMR3mYMgxUqvnU<$L1gfCe}cp@^|1Nr2MBJ92^hqgrX=2Xd0Uv^j< zx(BZ)xtFp{ZXYulb(XRZHxJE;WD2)1e;;lxn`5W?*EHM#KCrjMK3|l};<>tU&1VIs z3XSjtuh#q3u}8uTnG0XfWnPH4?X91bUEh}@CwrS`{yCj1;c1%jd)}g>@j@=Gm3{gR zF}>M{K`UKkrtiNR~-bKO?v_*8e7wZjt2qDhTKnJ8FKB{nH31 zX{7fTCnF&_C0+Ix3}16?xl866jI4H_9(KL-y_xQ*q*{?qFk!{9<2h4ewr+_>W@a8u z6dd!8OY&3q*W06u!~GE!lwSDdH24=Gmka`^l{wQOs@kcJvr;GbF*H*j9Ug#rqWnVE zr%$GJTWrRNg$$~QdA*_cEj`^AfTS}gu`uv0CrdDd^~Pxnhb|R&Of|0)DhiQa?GsP8 z5RS`4%|DRL6!^oz9HIh$v_D&FqPMZwvt*olf)XrqQ%^N7Zl>P7pYdX?dzx*sn3E{d z?)kOQ=zH<~?i#l!?I3lBD)CV7T_+$z_B;;)!?HRIOBFdgOzY z2s^-lF~~Ty9_4A1&JSyhG`{sQUBxT|8RpoqJG8wq-ZJdBdnLs0S(Xs;3t~_v`uC*% zOsS2P(6!S|s}zv_`Femi$aS@#D@aF$HbXhkB{E+^qboG74{NRN%6^Em&+eS7`J;7X&RCVS?u9JAR zh9i~KS%%4>pDAQj=?RrIGz=go9?A7Z@&rJFn3_FX_cFwX?&m4lNevz(o;4M;=-?~vL*}O? zr;r?pJ)`i7w0IDZk)CPmkGmqPl_h99^srO*km_47LD8*bEa${%aZj6q#qwlwC zb$Ui2elyDWSb!+4)Vb>m<#m?tCeR;O{4=VtU%(xAfzvZ1_5yA8R8*Dnhp22>@Vjt? zI``I3EBI|Y&hafwe!HA_eg^f!oKtV1tmVftn{09GEn<@&i?q1K1ih2>gVQyMYk%>O zrYy?E)rVXJ7Ey^&_Zm@?yG59DhSY|r|1_C#9)Z&@f2k+zCHw9nrmMt$v_nX+I3K~+ z#mH&W>)W+GL0plRGb_*Dz55{*azAZO4kQd*UrgV+Q-|%+4CDp80YlMBXomsN2SZhE zWbi%aJ8U=jOkb%X1PKwbk7>JW9{;5%^dwXf&5%Kiqo5kSQ~rN zP?{patifcnoYq*Y6ED}O-$Ab zSb;ka-@>=IGOBL__I6$Hnpp2;m6*?D2Mya4iBG*D-eKS%{8-4vfs#&Hee~eiPHOo8 zb-*UTF3=-TUI!(5`1h0XMP6Nef+ODEQCB~;(96h8+o<{Y@z9mYh+dbggkfZ?e~}Pc zUXXiq&0sV#Az>sEwZkhv8#hYTtKK6pEYRbksOXXC@g**scPnefMk;VL~(*gK$LXwMS+FG*t)b?`t6Ig-(yKSf7ntSnk1I1X=JH9{Hy7X#@*@1 z{ZW|VPcA&bH_+jVH8wFx&d9)&?19w0>6eg~7qhVoaWR^jRBmo=p%ID?=_nq2VC3I`-It}*WF)$Fnmm3cy+;|>aNkuh*W%)9YWQ_78u^W7 zWS=y`JXyZq?s=_RKdAtfdv>;VlA0eiu}IJUXRLfX)@{fH>wf_mh$!iK<2rFOP{ z&B>{K(>OwZQr}?1d+rQvU!T3Qe5CLAXgW1f-;!6~-n=l+A~;^4oLuPzddWE?!(}J=eFlTKHH#z5)qZ^}^^bT>k&G?3*nEnr2ZvhqM8h(qb2+|-W4GPlD zP(w*~ch}G%F?0zM($Wpm-5r8}bdGcjNGV98D9U}u^Sk$d?^^%0W|l`ihjrHazV~^b zXYc*&o$Q~Xr%gP;YA_^STIzu9Rdy(r^===QG-4iKSeP!Ak@PT+9E-wB(C!d!)jIm6 zQK%um`WfMQ2G&L~@V$V)DEJ1;O6;LLOo{G;ec;-NUbE=$MM%}ty-+E`OAZXu8H*#K zyAlqaJ3%jWgd)BR(lPdjsSS_FdVC>H^ewLub zmFsBfKwm}=wKL$*7B@&6U$llop(DHQyRLi6I_ZFZ0K3CjPdLoENP&!h7cC?k6U(hZ@3SNexGFG93z_r+y6F{s-fx_z2X?J<^vIj_ z`MlN5wH-fS<|4DMUElq@VuBl8IdrBa6rVFPBd%YWBSfHibI8dfcx=)7vcY|w54U#J zdDw=}IB-{TK1o;NnHHr1N7SGWF7>3_PoKPGTcRZUz-r=-h{6LwDxPnKu};TaO+EAN zY)>{}WI0}B!}_1Kl6ViG#a1#rZmH7cA39LnCUjA7D2}T)w@L%s@VXp>`sz~~nvEU$xEnOJuyIUH%u3%DkqOyioj%f|Ae@68pfSTyJTKaGa&JeAshW6dOzmlnn^v)LvK5-Hz&qQ z77cq+4-1l(om$K@|KlhfuEF%PWs3Yu-E?LFE6B*vijD0ezp#wT&RnKQ zWPc8kO7Xrr5A-w4qC6P=s*M*S_eJEWGBEo z5<)%LIM7Jt`S8osDo|6|pXG~}g)ks%{zaZmG1ll460dxP$m#@If~ z51;GJTS-rwJ270YHJauV=me)7NG_IA=naO}5Bk>+3jd@pCdTcO3Q;q_StO4vlPr5i zM&68>zYNR2yB8oq$ew-Vw6^`xiTH`sBE;>$kb5?gSE5%^^7h=-hB<@TFdHjXW{sAm z8Wrtz$jXYLgQMeeqcs*C9UY~JkL%EIRVB1SrUpoVOG_ibU$42eExiomlW1mpvcO}I z21b4$&F6PskrvBmq{QWgLKT^afEA3g@03sd-S%M2Zsp|8R**)D6h1hjmFeg(Oz?0W zHzZkLEYX^^ugIdLOaJtXB7Q#c>fJP}4P~BC6M9?0apy>3oq3$g-|D?G1Bsj6Kc^JS zD=6zFn!=Xhx!pET;}j(;@JJ5-R@>k)K+VWehHhJnYPDYYgwKwFO_`z0{q#-$mX(ld zj>emz7I)%_0MlWcRC@=mnT};;+b;VX4yJ=9)?${?5_Uz&Qv0Tu(n$`83Ef)Vj0cZI zxwsNkmzIk4lQ(p5x`$8t-KgR!zd&~F8G&8(=jkleBZk3H#gjv#b>sP$*T9a_d+IM}u!obyr1{rvRVE2nlfex4wjBjNFihJ0?F5><+j1#tJ+}!ZqAGQI;HG`P z6At;Di=J_R@24)t?+Ln#s4WgYTh5W?s7oF{B+|)!HT@x;oy5zDPm55$=Z{-!W^l!A zLc`m-hrhz^b+V1l)p@A*PThau!!$m)gCXn3TWpE?>Fk-5iGitq&_lJFaL-j^r-0n_ zUsqnRiU$mXdmY)u+n7&qhY!UP%})ce0%Td+o^g^uC|w9~zEL%Mbz(qPYI`ILSd_cb z1>I=0h~i?(K3(RFu@INEG^3lg9a(h=I)R1OmGZm%c$5XRbwmxqVVWMao)lf;h~9!L8y! zshx9=;83tY;j?sFbfvBH9S>(`c4uejgHEtXVGD=Dhe%qY3uiVr)q;b?ljyfDZpQAb zL=HR+b(EaPC+0!5@B@E%=5y*J(%jaT9^)JPh^y?|RNABsNHi}fD|@7nG&axsx{p@1 z%wuRezJrMPF*w2xEv(dM{$d3^=reauigi^4`#n^JMeuFE{r^m)aX2v7WAMVXwdU?bpba*WR6J5J?CY z9V*4Kx$c>oiu2EXc)OiaofX?=8KIe~>~<62%M?9?NcP309{0E{?zfYsstT=09(T-8 zA;9rOgRZbUtvmILj5|bO@sG1)a0!J8+21TRvZBtQT=74+Dw)oPX&3yqi!F#Dvn;|q z^hrc)utf$F^&j#zxKzd9)LndPT#o7kcuo`XU)oTjb`a`W)lOJ`>5M zJF7eWDH?nA=H5Hc7>cUE$MI#B=_a{De_cj>LN122uuUft->BNjd~XIN`>eGS-Q`*f zlx)#@zi9-lC?sp@9$)T#xC$L2-SVe&*y$m9MIyG0Oi3B*ba=A;1W9I=EY`QSC40k{ zgilneQUu{TJ%>iw)9g7ye*?~bkPvsIMVI?ARC^M>wsEx=e@l;1w-^K5h-XVxK(dm% zAV5%yOc#1&7O&g|azQ>#I_tanxjC?8F*v!>U&-$#{y3jFl#91%Sy@U71qTNQyePn% zp9qs1RBbkr>~qlE{S9e2PVWH#c04%i@N%sZ*KP zBT`2-s{0PdWGaCiqc8&?qLebLZ;3-%PLZsm?+q1AX1?iCf-X zY_~~T>2Trb--3!B;`OqE_{N4t_8X7tYqr?}og{0K8W-?sZ;FF64UC*|6=-Q}2P({U zkIkru{Ec~x&GXZ7ve~7vsbDt5t)u_2zcK50KfKjjo}gp`E?G#4O}lqP);MAOtBD39>|myzYAiy6k_K z_6qYwQmFbN!*|B*3w?YWxg>fZa{$)yI28QqOhjY(R6u@f%ptHV)L|SC>{@{-_m}`0 zkj6PUIMlBmYb|ghXC+DN$05qd^FP_7&(P+ER3wx+)7*Fc{Neuhj(zG_S-PO!UdsON z0IL2^6nD9GaL`>lIJeHM7?XS4z3cxwK-S5z7bVSC*g!d~wto|RM^;obII*#~+xSx=th4xq zK$A8QVUjmk0r5rMeS5fcCl~1fr;0*&JR3b7#8y--db`N{MF&EhBSskaEw|a_>hm?{ zRfF5M>I9$;D-ARd2g1n~?W|HF+Ap*u_$j!%lq8gGb8|n~h_~d5iLl3PMhB?U+n6zU zCklpoE!-vLUiHp}CUJ1NKl^wv%7nmji(YF5Ydl=zkEirwiO)l9NpJ?#4R2J3gfA^gV8o8N*L3`WEgtxn65&|Hp4Ub@7Q z6wk^^!@oCRlLxq9?taZQt-XEjA9~@q$%AGLAZo|`! z(yFkqXfxQ+_>Zp}?A!2K$`SrhiOppq#1>@*6a4_{6iu{MkYZ`j``4{hE-T*tZhj=< z_bMu(1wE7lPa1CzqJp-I#`NzsX=BWqr<~3k*sXmoDj9fRt#i9am_?js%2GMw&mvVW zAjhIB&7WeJm~BiFt4o)sEAUrE>}wlp{f$g&ftyWa)ZTj>$cgg{_VTbPX)@8rPXqX) zleFdcwN)vVJ&vAWP`c3JfK$t#?fQEVo=97rGHB;)I=?l6@l@NwZL*c;9j$iL;1V9N zV*=)Xt*!h(FS+H~PWd6nWr`aFVk$ELqM#qflUkLJ`nu2En^dBlp700ZnZZ+ZEJg(a z)iR{BHE^{8k!-*X#NGeGiC=hED7ugfutx)1`2WtZM7g?UgbA{q69VJmqva-%A8Q^W zzmJ-b>eSiRkc^RIFOLBn0oP$F@6)ul?r_ljTa(vtQ2+#l3d5O?frqiYiu)qXj6f)sS_4$0woaShu`gsI3R^a_ z)!B5V)fWMYE5Ne%Q)R%wvr&+VDK3=ZpjJCSpM5>ox}x8Wm-i z@DWkCTP6Z*i-VNu3^NJjNR{r&8P^bM{E)n6dngiF$YuS{v2!Nz$;OIceAqT9XfgMV zK12*U_w^!_RLb%{*l;FE$S(GoSFi1?D)MGv&pBghwz0pdb@=F)6ibQs7)nS-I92#X zr*jG>MGLV5d1vwFCpwce!JQ}^RaVuv;q-U zj$fy~f-R+lpvxk6^o#y*tR}yWz#r?w$m}bRSrvh(7!l$L2lnt#U0-UCq^ws_T^pb8 z(-N>Y-l=CbnWXn;?sG8RE?!`Kb&NY&w%#v&?*$TN2EzhS$??AEI8l3qDwfM+jaE=X zD_sxt;|a@%3*BmrQF-$h8w{+o$g$xr{IknFv3>8Xt3!RCpZoZqJYRVSGbnW{iW!RI zfT++x-j?i3Cn0*U5A{Rf-SV56klvry_<#LaorRH&e1drsu|d|&%Ys94I#B}-%BWd? z7KL(7ISB;Q9$!jMMCsv@3EE1Vzd_sJYn5gt(~FcG*}F`lfhX=CrBLBG*d*Y+z{nnW zRka}jK+i1pptVWWVvN;Ihh7P*4h1x_O+UPzD{APgIoGT$D~tXzkWsH$9zAqwo*43- z6zz3b@07GhS87@u39P97fCZhkMEehcdFwo{CtL;O-Dyi*O`Lht^`NkeX*|`!PxC@H zcL~EtZO;4mxdCEkwLv79;b1LED-W!@0vt3))SskkUXhEmxV4o8NChkRYQw+saAPxw zu|BjojMFnUZF8aemxKk(6hI0Lu!gwDM9Uf_v8&7z6-I#2xRS0FPH2Nv@_zzZE|Vgw@Tx4=-}m3z94<2WHnM)#mahm@<7mT zW?7zJW%m(~mZ%Ns-B+l;|0|v3T%AciffY;(SY4;VcSmal9KC! z{-M(!GvId@hC`_A^-;6ifa#;$<7YVr+++7UQ>kQ2KK&aAQRF-*g%bAzyc3;NS@rtZ zDxpDjA;-BSUM{@IX|RsHsU{RdiK4&z){{Ps@iXfMa(X_|l4 z+oxDR7>%g3;v%)%_-yh^fI+(&reE+wJ4@Xx58nO8DTO-&Cu3bVRu^T1AFmIWrp4%w z{#sQv@eRu_d9MW_*9D=apAow5Q6hB117WH%Z@d`QJ!A?Wry^(=g)TK8sC|$h@Iy`r z+(t5jTKKxrcFr7Zd}r&L`Y~y|8BPuQcKg9NPiLrT{0uyt;Zi&J)y|pOr-p z`Z`4qSZKX6=m-lXOKI6SU-!M>BK^33QShx((2NjSIxrcK(0DaQlys>0pIPx;EVG~e z#lnKNKziZ39mVK_{C*C3^mA=rOwPn)ohSBkOzb={;S`Y&1lK*WfJPVA2j2x z-42?af>4+9AHjD&!1fnvf+MrY<&0DlF0AK#j8)|H1Ip#~wdt9gY*AviOJ|huNp~{x zYtrTQ7&np;m8WiD%(xtTXWn#CQi3gEb%esj>mnk1z4%j!>mFz(B4lHvEy+a#QS*LV zB)!`Fz$fgq=34W*hKe6%pgdAdq>%J>2C9|S=3;HT77|@ z>E3txN&s^&79Rf4!G`D)Q4H!P?s#nUA$HzssCQJZen8H(32#|&*kLzS{n|UVsdJ+F zyW=}gGnB#Zv4cbOB1SstsX_HmsG0_;T92A?0{G$vtlss^PyDL!%Aw{S1kI!M-ohSl zuz_Jq&E)V9Tc-hA4{NwDszr@{g<`|beWl(_H95IQnaZV-p3yV1F(Vp{!S#_56(1{m707>}^D__B3qQ+NWBjnn#rgR`qm6SC)jScQ zx_uB8Vz!TTcSfe{c6pzeenCu59vmEW4`igXlEjEyFIkCPOz;LADw${XG{JooK_8b;I z_7LMQEdxUFlzM_5rcZO{NB8*}g>oe&hIi8JH_a`FNq8sYN=wK-!dZ%mZ=B*zdRa{# zFCB6>{V64H-PD%)hr{Re)k31`C3y_Z!TOHj3^%EDw3KGniN)lI{XyLh(K7rv#`a% z5q6w`G55@zY5?DRSUS;FI~Qc$wrbGkV$vb%EJ(@m2vXLS2x2&aZ5U9b4B0^l z&>xQ7pBMYSjZSUi~Eke_zw;}N2V1(jFV*Y2y%Z3qw!_wEYR2sdQM2NG;SV-PPx zZek7W``5rGalloT*>1=qYW>5l2_q8|szqZDoJY8nADs9-#@X9k!Cs|1xG@^6$J1C< zapUw1DH$JRRt@+Fm!{$})zC+-f0+Djyw*J}!%^n;*)D?6wz0(3M%v1}d<4%_am#JI z+a)qCa8YonWgucyy+9%z1NBfhFr?`$I-=@XVuuf)W)7P$A0;qT3TKuD*UOud)rD5j zjjKO_oOV(%c)2u~w{dQ^3a*L|D(dug6DT~OpP5Ug9(}Icl{oeOKJ*6~6AL+KU{j=Y z`wPdi?8S9mk1H0ko!*_6mY8bhGmY>bEqlp7zC2L?a3`U&Nx(^-eA70mQa`IUNg__7 z*J4WFGu8)7CALhpjxz5`Fz>=sSyaF&Arm8E2!JZOGsTsbV_y;-ic$>PC85t{+ik!~ z>|e`hPZ{763bgTg{e5bj`Eb|Ony-cb$$fka{Xxq_u-YFH8F?Q7bAS+8J+968@x0m;@SC*N)h+y+Y{+T6Diw*OtG5uzY;E|T zoL{J7L2f+hXi ziVdsBN2_fK6fe6Ir>0Z^*JfvD_u<&fs%;g7KmcYZo>HhkhO=j6vFDmkvS23V3i!=$ zoW*<3EG!Ix^V{~|1m-4kdpaPW^wGI(9WZ$rR~ImYA%~W<6Rs!h<9d&>C(-KnSYT5h z9a1~o>XIpbKu1{tUurGr1s$Go4RO`Bwv*ls&o6cBf|<5Wj(IK$w%bGo-mFO0aqpxH zEc9QKsg+X-X6>9dh?GsS&}%|(klak79y|v(bAg0z6*^JPVd^lLb(;KTpt9NLHw85ouV99h&4dZs-c$)~AXf9W5Y%73onMzn88 z8kOM`3I4zjBj#;rxd|A^+?h6Opj%l|mw9J*7KND^s(vHEtB?eya-cBccUl5b$FxM4 zsVG_L)3zvp2Ls&|Xrcz3_|pe6`^@o+G>o4S?Wuy39hS;CwTe8b1?N#OjY4U)YkGfG z^=`K;nr1h4gVr6H(X{6GttL6`_n#n_-alT*?*`aL@BpF5z;cl9koyUT@#jA$!Q#mM z|8(~Qy}6BzsrNrN#z83Zz{Y!j?o1n(v#(cvKnV-bUyseeA7`U#K}iXLM^*usMKrlQ z5@nSvC%)7EOTEE&g~~&J4&=nM&V@}Ku5x~Sk7I;p@oHyIIPFX$g5R_q}_`H$XbuA{H zzKL`e3kT@jPwQ5vRxl2d)!E+kxL5D7Wa#k(1r{Z>m7O^UE$GDPAx_LP3_zW}gGX$A zXB%}bn2TPsqy+rRhxsq`?aOkLBs5dMzS_%hC|h3`*!c$%I@i4{e9O+|&eMM|3X*69 z!<7@IE9K$A5mt4@t~arBzyvO+invQlSqY8jya&Ghi%)6o(N zIo;1)8R)Or#M~^*@L7HfFTw%B0>D%V1OoQg8}zFwr~Zi!>X81J$W0Wod5|Ymw+FwxlGf#Ge%>fk zlWqEIpjaT{i`!vDw$KH^^)Hwzd5~my6&d9;4)Rj5=J=L$&p`u8nk^J4ME2%qPVB#&3+i0AE06hOQpm7SY1TWk0O`1Vq% ze^oNX>-0;evbZamXDD4M*97!*1=LJ3YJaQ72h#lM!0#S8P|s+P4IH=zpH*QMrI*0p ztN6qUGadE$@M-^|!S}+(9dKniWVn$UI-Z|)JWUu0qVS`n1p|LU__J$3=k)szDYoOg z!uMR9$c7qOp+jknbKd|yOg&ZX?^%MH(eCNV-QSZs@*&E_PoMgW62!*9h{b2$QTt8g ze1tCG_v)r;O2gEkWPb+XI=FGWPXbSNX7m4ErsRuW&1pGuyag3rx2om#?%CT8sIrri znYL%_&{zkpUPWxD2PzbceSTiPkg=*z zj8uSVwthKF+a>N6Rj>-2w_Vr$qdDzf&~Gmj*9R`m28-~}!qiH0!`P(?C2MfdL_%L82C)(_+Mh}Msqq#-4gdN`5O9_w>>Y2cd!Ifp@LK3Lh{pf=XshS< z_Z{%fhHR>jN{1^;l-G_0>sFq0c^qlPoQz1IZ*Yapk!6; z3No-IM1@Su_vEAZJ1|~LAX}qy<7WV0k(>wH^#QxAW|@t^LA$1VU-Tq%w)1=Uh3dv- zsleIo#k88?-`tacNI;d$A`lJl8fuI20uQ7DK)`q|B;>*A0L@?5aM|X@i9ID^Um_WU zzC`pU>$X*;y8@;1RAatcNwWWk;a1w3$ouK414zjOdfl3S)zj|9}M=4x_#7V3eiDmm)Zf<>8-EoRE3Jaz_?8Vu%dS`KSTE+Z! za0cM`j>WK|*g`5)L&?MCOt0g_&Uw5f<}z=BYbE*lV~$slh3GYxq3?X3sb9+U@V#o9 za=Rj~)zR@0F+?{=I&J9$pQV6(zfT2h1H(Z0BlqSChrLl3fIJp0<4H&W4^dw+ec68Zac2MI2M7PFv&;DCSt&541A zp(rkgfRSAF;?K$7H#7lvkpZW#X<*}N|Aqq*5CjV0<75BbZtK|6;y@BsqOJmPK$ro; zKRq>4qi_iC{BMxC4%X56AJfQt>Ek-Mw!p)_^4B^D-OKu~*n;!e@BWX!sGQPL6XkRE zoIqx`=PW+#qq#eyQNw^e5IOruh&+iG46_<9)fBu4XzTAQBICGbw%SsTLy7bVxVe)# z1Y_k1pO;JfLew;!>T}y?HKEZF_lgpqMaDBDA<$pllF*{$X-Aj`G(0qZn0>Zq<6}I# zO%_G^nTG{O&B)>h&v$3D>C3GsW*CH0~kyobU`pGBFNZ zMgS5HSZ&aH{wEs(s|NP_GRdR#KuHgLFu?5#OxFRb5!&~;ID0MptM`p<4``12$13)w zng1PVv~8A7T67(xTjx-UHAzhP?~~qO({EiMQ_9}S31)6?{^=ZmQ4Ttfcnapy{+z!k zt^EeNUvhGCT{~^o0n3JzJELYcGtnW&KiZz*k2Gaq1~#*3tboQwc;K(hUoJL?I67= zmW0;1@5!fWJ^CcLhW0gUve~&|3*(KeYY75^V+{{m`*80@{vmt5ToMi~1?e9I9`}-y zeyD^KJ9~!cMk;k}{R>u5k^| zk5xCO>Q+RC?BNFxc(Y@m7H2D(Rhu{wOQ3T7cp5v2lfJ{p;#D0yI2E)o59$^Nz#uV< zY)k=|+_qi7fCL{FRB1h)6(QsU14XZ{t^iH_ERjiXI(B#mXgQgP{#iNK!`Ci=yx=Q{ zVey!}-B_p`7iX2G)xu|aEqINYDv8p}lJmIgGZX{#&RlN0NaOT%v8)Y@RLw(I+LjM?pW@ot$~ekssFfS3i}nl=Jpbb8`i{eB^tkQKJ)E(AAuVrHp# zACRi5%-YY`#1xYBHytbj9hvLbX7l%32FN%rnZk+y|1k!x(w<*58vL!99oL>-E-T;O zwaRMCWpKY}d$2wF>^EMC5F9dm#w(Ph5NB-4m_0kf3RR6%R!T+^TKv_#V3_GF4;6B3 zG{$7s_6nc))Vpu|)MM8>k+tgG%r{FwV_)Ve^ZIei@fhAS7&7NKch#@Yz-c!!dB}fL z4umBdj%J5jn2sq3HmzP!`9IgRS*3*EQ~~#>e4w1-Bq%4L(cz<7@C|5LRSHWH6P>se zeX_~zwB3@;mWfWmG#v#`qt668OP`XdjGz&srQ;OMqGTy@=SpkEz#c`8TwXuRp3LYL zb5c>!cW@lse3=TEjc#pXt=@i&iIj!C#;`DWsz>jjxGw+>khkYnrfuHV&$At-n+Kf? zC_HGR5wp6lr=+fhQU9o?2-KbGwA{Qyx{BK0Y7d| z&O-14Uey@2otFx{kBX%BdjrU6nc#fTT&){o!&K=N2&fYtB@9dX`wN2&Z_TR-%Lc(+ zq5sIuOkUf1!R!~E%a87Xs;X3ag=7|3blPrJB-{AAkP$16garny z8kLejYjb&vy8>*JCl}I%8WTrz3BU(Bk!EjKZSll=uEp zFw^+fjzOnCRtZZ+hbFq#VgZABant(8R#o1h{6y<<^Qo;LPxmf4nIoxM##A^J-7BPR zo+r4me}cpvyp;sa@>YmQM{~3jk9j?*V$BE@CTA?lZA{Z!cG49at z{^C&IOOgWV8W-@CFE224G|0Z{9iBv6KooyEs6bvee){C{c&q2R?u5X9wA7)tM`RYsqZ#mP5@9I115Yz3w0qrcZL#*a&)p}Mco|K-ZE zs%m&)E!K%hE3L|;;@9&RhYZm7DO26j0=3-w% zhYCW*WuBDVZgTZAXVQl%jucyOliJn;FLHf*$Yohp%avl#GEcrTU*F*s%qzoKcI(4m zI!Zy;H)=uK$<4KnyJ5(E;cve;x0L57X+uq2I z&m+S|Q-=55h5n05kx8?@4GI=|XWte>(No#C8=~ws#g5LZ7GMg>1`s?DG-ZSGk&5yN zSGg--I!_e59AE+$K0bf_2(K2nShA#UKk5BBn6fLd4xh!0wb@mve#;DiuSkSlak+Y@ z4+C4(kRR{M_f15tl+-%Nhkf&%GO=6{lD0FqEVB7iHz$pOaPfPi$E+u=VHTU7LDk2S zXo2}N0h8KW6^1JJQ@aQhxkgIiR624kCFhZ0!^Bl5o=F^rR+rWW>n7A94Tki|f$+9AX0ZlU8gFYG!myyWB0+992nxbs+6P3Db>^1Hog>1QlOoX?T$!^;oS z1Ae^Dxn2{x4Y{0X&#CEFXI~TG7TGk(qtPWH!us|;nzvZg*nc!!jIF7~uxI;FN_=jo zcvz~CJRUQ;JNVxIk%O3=OL;nYd2U9yvI^goexkQ5Q^tIU_Sc^mL_F#ICW_`ol;E@0OUIWYYAzN;5}rPl#sCAm2?%njB= zXweHL_fkh1mH$coXO@=%^2Pk0s^VQ!#Itm85y&`r2#4hBvxU*)Ztx0D)TL{XB3B%PfQ6r`-^x6zeHK zPE2=_+h*a)1Zf&3ahU>ljbx|z(L~F^T$9UKzG#s;&$#27Ofc1g8C9*`M>%>gx1T{7 zo!R#)^O~vfX4rRSCVN6e>Zm7N4}8)t)-ji>#TKHC<#ar6#ZZJ>^?ld9C_HCd67w^H zO_U3XZwe`TRdq(^8K3)Bvm7}M3z}YeL=819zzAgg)Jn*XM-$0k%D8Zw!}aU+V|@%w z9t!*}U;a8wjo(*lO^55QH1@T1#!m{RMgXEdQW2&aNhb98sh?PU67$=M@{hPMp|co> zYC^~_Pm-0^C0D^& zezo9&bAN*iVW%OzFwYB@_Jp;OdxjjZb(xYnd=lY{zOX7{CE7ROm1vr73p>%*$0YkR@%u%IsKF78=9 zdHboON%!lTke^q|Vycj3B{orc=#!@+SzkJAdxtNk1JjYiztd(sL`nwCL@GMw?lK|=w3 zZjO-0US(Yy`O#)?ek4th&)7MDX218|{gv^J%fGwbN7}*MUY3cMbai#DjSZdrNMD^l zTJK6Xa_I;MtSZ?irkOq=7->iCAW3rCBh~a0y@|q`pSM&27k|D|0I}`l2;h0 z#-#wvvEw$sTITAS_{r~w?)4MH>nFFt;*{Rq^XeN`ZLZ|`7X!71gQjM|`-j07+zKfAM z+6;cx`)9S+dcD(g&1=;eIp!V#JXWHP)i4bUNIQhg&|Vh=h%9&lPgpR?$`&D*6F5m? zK!Pi* zw70f`9vs6n^@{YBh;^gStD7q7gpfwtGC=RFEO#TiKa4?AEw*KL!%1;ik*ctQA_LX@ z@#Lw4UEMFyg*#`YOWlb6+8X{W3YClBB3aXjpe7hUhP|<@i9(xWi~8Xa{VE<hMVcMx#tHgV#6q(J>^fB& z#V1FixC&+;YYvb>WJ>NV6E}D@Vr`;cypCG%V#WE!UiH;~-IB$I2j=ImE-TAaI!ik` zB1?!?MYRm;BPbux1@~G3@p5-FaW%a{|90Krc3pUlKT|OQEYfcT=SR^;$%DZM;rclN zObV=!71rOpffa;4IZXP13I9_T63BOJB(tw&b>KAQiv6nf~WQ4@RhV*R2PG zq0zOYuVN({cN!PBJ@~gg_^o(qYgwFZic$?vWWm)oY?zHlsB%T9hF3tO3D5mj^6l0N zFslg!=s!Dv8uJFEQtNE|T4>`iaP*KUw*fezD-d|QcL4gqX{jFRj2kW)8y`;qgd{|m zKuD4>95uWHAT`{WL7ob~Qo%GGlK=Z+3rKzpN>prYtg?ZwnOQH>w>%foaLEvTwusFW zvy?@BC$NkK{KH3+8GV5|h?o+irOwGSAQG2l#;zHvmzTso*O3Yj#Qw3pV6W~t*9+$B`8~2EX4IqC%3v0*nm5{bC!a<)+;oXUqE0lj z3vU#{F#t<-KxSyL+%g$G+YnH8QHZ%WmN*ntk-F9xaQm`$5=zU-pR_DLjD?)4R(CyQ$( zht$c7jW2HLt5N*|gqBR#<=4y?xRt|Zd{>}wRuZG|oRU|2LVz85&09GYTZKF!rSmyv zY*ogEEb=3Ec2X)O^&544UEUUcnsj-3pG$-sQk^p(HDCvo@4SzJ!tffatqC{3iL9>a z@qoM2z_@y0$J|QyWANKoa1L3>vK*V}3kI6!5Q6>HX^~;?oP>fdo8L=uNjW(=DeLW}QAk2c{l8}J z%)bp!_3#Z>2IMlJu;OEoIJkNyaxM?6EMUR}DdS`A3AS!4<&({M^7M_{Art>F(`~~1 zj75)^?OSNnftMP9ha~aUzPR{EsCEiQ^H2cIMCS^7bF6-g#>}ir?lXV!NL#1tt5x{e z7M!%>N>-q413fc#$WiW74m8^e)+|q7`AI!`+DPwmonY~AUT4LdMx#Oik7Bh`-FO*OM)LsulRF_7iJ`OQ7!O7kg)@cw{N9@-Gy zzDTG++K}W(Q$T;%v5cHtWjl_YbtSt)mTWoa^oM1zt`(pw#ZSSNg`@w6e?pZv#EY{} zRXk);ESBUfwP!odd4-s=%v9@0jMZ5{0O939sMz}9APT*x?3wiRFf%NMK1rrfeA@NW zmf6^p4#ycZQ~C|Utifj40gqk2l9bt#-xNG;MEOHh$bD(~c+^gsKZ^^xMrvYO-BN;I zx~!*__~Z)KgevDK9KYZNuaS3wzd`ociTCH6&}!6aXJ?gu^`ri4s9x1nkSWpDf0BAa zBL4tlQidO=89O*b6n+wX_nQ>Pd%%+f>$8DxH!SXWw0v$xV{{&fmKnTsc(x3@bg)^>UZ352U;+yIyZ7>M#MQS&50I*HUhDnd(O}>3p+(E%kckpL z>Uy9naoXt4H0IDlu6tT&d!AD*cGeO`Ndq_e9I-(1u60wel=~;>F`DsxB+dOFeKMiC zJI#a@N;~lh?mhQFFgHev%&Igxj`zLGbI^w&^&_`aC%3r3o^Zdg&Mb-E(8y@H!RpcE zmMTqNS_`<%6c4n&^mL_VIx}Ql&mo%2 za9heBi&0pCEwR%~(i`c#oI>~S`th@mW|wk#JlSnlpvOLr%B-T72D3!r%4VE5TNap6 z4-aSos>8sWG#Ro?@evAu|1bJIxTrA_NpU6oUp*(m*?vmG04k45d%h%3bS=p!PD0W# z%?loKF5kc<3W^2@Wk zzDe5+RTuiWpQ0Y64b!eXX(}M!Oa{AAYX62baxjv0tvT^?j{O>bZI*5nw{P zI_GJ~MuNOK-9f^v2dIE1EmXQMUvl1^bKdp$-j4U;#|)O<0{3?h6?^ToPHed8+0=^oZhKAG^I{B9~L!`+rpq>9g%7JYB7Z9YG1N9le&g z^~{wJ?yp>&3Zcn5LU6W$5ITeS+_A?|mAe27%88$3L9E%?jN7mp2^evcr~eS0WRPO4 zG_R#JpdPm{y^gH>NV!vOw5pZ?S-5I>a;hRj?5Ac;3g)@^Cy8-9XdZwJKsbEtnoKaTaJ^bjO3?$5iP{QYgW18C9ew~=cf@FP`iRk=U z_Q&-at4@{eF*BW|iO}FYNw)Yb)exr_p!Db;k1$}RnTxjmvT^ccAF&yq*K5a=PYPRp z^6?03d$}>&5-Q_W9=Ya9b92Dt*YJWnkCp^?prMG=gu}`6YRSm(t*x*bt=E=y5pfLc zm)|H34sg{>gismqSR_r1pwZzf-Ak{De^o^s8R)=B_@1&SXG@jty>_nuZFeQ*L<&ij z*WOtE+(^>>-YN|6QCS?_vyx7F|8e#;?3`OR5=hQAr_YaVdtOhvSP#cjb>B?B+ykN9 z--@r^#N759LK=SMFFi_22Q=V6D(Ud(h(1>pNH7p~@Ub5j|A$Iqn)FVHcP}B;5T>@J(=TW!R7r z)oRx(7uosUh7QxEX)@RY;tFH$xMWFd00$^=oeed9Jsej`kC8xLqDpm)iZ8V1LvYWT^&lI-t=u(Fs1 z>}iI-tHNL8WO;3j`?K~)uvVIyqF5}U^=k>1Y|mX?@o70ue>s0RTGJlh&g; z>93L89s{Nq0FO8}1B!SC`pWa~ef6je(oSnUc8vB*ck~vLsV8{?M5GVkDbi+0(2<&r zXNGFn{GM4{)SegXWH#G)kQ8c4JZC}315w+?VnUXKw>0Fk+4{(jYQW@IRi$Xh z$86IC^Ya@?h*wuGc?-vQ4eMt!;3j7mtl?`v!f6LEBiF0iX+b&OhV$Kz9-En}#X~4L zJJZ%jHy5ICy_KhCY4nu3G`Lj>X<4~UgKP+xc!Re1=n+$9O8<25H92VCt8PKdP0Yu~ zcLem#0^M6UtfFuU{l|rpHr7DCT{asvdkFjU3n`kCEKnDEC$zri4EF6Zl|KPD*ijfu zBri(HheBc}{Qpmum%;u|mgg`57uefPet5w*>&LRiF(dH)P({nI$uSE=+S3t4x&A!0 z0D=(N2K}7#HDIZB8X?dff*1UA;P%Sw?yB4LPVMeeZU0K-6=21XTY5pi+k&%)sr{YC zrhCCz09nnqVZikL>LMs^IYa*`Z~pXj=WbV2Y46be8-- zrUCk)jjvs;ofmV}9Zy?6rE1hD#)~b5Ozd2-j8+E;2r12TT_dyS9UPC%dh^S{i(eGn zBb^|v=AbWnoV*&@(u!dug~gy|u42ZNu3|$uN#P24`C<{`UeKp0=mEROmo3XmGID*D z+_4nr)L?%7yMa%@>iQAOc2^Le76C@%d(kg)avDtrFzk_}$1$*BSo2o%{FWPJyMEsC zEl2L+H;;@R<`#xW^c8Sy>bgANa90K}3YP50w+Ymd*c8>Qtr*Y8eAF6{)(WKOks`t* zL;Ed-jT?||n;#h8`@5@mXNcnEm%qPH_JMJ^aWh38YC%%e3ma=5^0>LhMJ+S4gxYVw zO#y6xQYWbS1{A0RO7m4n$o+d@b3o|bAsjlm>2)9ZmX(WaXD}e!t5k)FLeQG?*>ui& zcH(abOf=^ywhvsluDa}iZ`ZaJ2yIXkIeQ2MC}l!GXHlFFN&sDgh$Jw~DgxC@6W7x) ziT*5{Ga%$*fBIsF=FVqqTKuX)jop{l!!leM(HL;WMILtmJU<+xG}6HQH3c*(;B~6t z^~I%`o)J)|r9@za^LC#bl4Pq`8bzieclrPD;pW7bVA#e-5MR@bA@vPf`{%{H;j}eG^yNpNEqy7nj!9VqvXe8+ zd^pH-4@6TL=F)qS(y)(5t(MJz9AAY0Dqr3n-X;OudRWJI3Q3tJ!O3Q0Gqv%nE1*f4 zIblz?6RyN}Y{t%vsAHI^JT)1$+<8{n(E$W@QvYun%78{Ba*cbtN@-%Cj{88BU zMgE-#kdf!POMDUDyuk(DvmllnRwwTE8~l$cT|@h~$?>GCh>iJF^`Wu`HvVOCz$5~q zvNLfb5*crOaL5Tt9Ohr?65sY;3W}|i&U>ooQ>L#==zFcb$Wt@gB2h)!y?il<+ z!(zhR#V@6oFsX@}^1|Y~kI^sx+lsO<3FKpBY`B=KMVCgC-t(@0|Cu3`C+(%L@zH|h&XfRJuW7)| zC8eEtSWApvGRvdj0y`kcpTOuJ5i(~q^OS4)x67XGVv}sJ+2#L$bSwnH1uCHcB8-5@ zkOf8j=#3JLc7FH3Mv9H`jsw6e+!EZ-s{QdtrS|JYd7Kr@lA z_#Ie;0Vq4Isw#eI$qWoyBq1@E6Oai2;z+5hs{_*q4dO^?Ap2H2vCh18Q#}^7y#8-q zjvP^gc!*aq$A4lSRP6Fqn7k2BSi1g1buN^mUJB2o=XqCY=Bo>zt|-7n$#o$@GV|mF zX87!O?t_US)qyw1(^wma1j2q#4SKG9Ef?^icq|Jv3 z%)m+J7!}eh2bPi}TR)$6PMtNiwx)QJYoQB=I}MUBy6m2wa}Qt zei`0A)|TtJ)@{h3Wm&A6GbeyzOE(r5c|83Vq2MXjc}u4l z?T8u@vUCEFfL|7S10 zbC+q;`*HDr;iT~4W<1MBAB>~mn`n5E{ejk8hNbtv4D%Z0+O?77iAA;^M7J813fS9> zzOia1EPW$t`zaL}AoNqK(!GsAMb&+iTRIUisUf%eg63g`(&SM#q>)qV_Zd;AlG)!~` zuX0`l6H|lrYgtCMPBQG{>(M9^&a_qBUo4j&ikfQYt;PY{z``e5j#T=1&o4y?YM`GR zD(@bvfflG4v_RXc!E&VCI>*(@i`^d@oGkNJ1($0aYhCz}#Xs9`{0?UKDvZnmF}bfD z`|_AIj<(XadPdl5 znwn5`zK+Fpy(D3M^YT%Ww9qGImP=?{k&Umd`HXfBi-vd%?W`LYcHV^_aE8m7t;wwx z@#Ra7IPbtDN~~@{0!@)is?_N9+Sd#5DM1}(TxJcF+A3U22KGH@c!BX)@zWFzckX=7 z#>WNGhp$$#A$P*_am(#dQu{Gdj*cLw-DTY5`51?IyqwOyw6ZGh+3g|Kd)-v9K}m#{ zOny(vEt(bKx$H;9(!pu3InIxMd+$PV%Sv|hevQle9O6?X-*mlZTY=7snH%&uTuT?& zZy?k(hP?_D-2YezsKrbn;b`dy=?rAR4*oI&4$Yg-LSL+Ct z9Hzj3Vh|7vWM1KX>+er!IyIQzzIjlX5VJUVG|_OXZCkTmeiwz5&;^=QyP1i$2`+As z0Uuev0n!0+rI_D)zODE4I@Q!#X?;0L_VNjbD`GAwKR9*MyPuR+Bl$nTi!uOS)DOwZ zeft`}oAJX;ufKCi`jKp|T6$Gu_wdbs9e<%i(aWZ*lSczGi*8-4_=OS)nHWoIDfc#u zu7k|kIq%2tbfm0+FNa1ne?*=IiH|O^Kf6i5O=f&HJMoP&oY4ik%2ow~kI%%}x<#+r zE;sFbabwM^DT;9B<5fmBxqid~#MBPL64bi1@l3ck?tTtLD-O0UHpsPTny#jh6vITz zeU@Am35yPBz{)Z&UZA6(yRTzuqxVkoTaft9EPBJLS!znz%&1>sL^pifKG(?kU!zSJW3rHn7qM^$RW+1f>g!eI1+SVZDc@os0=AH|8C=KV^LMFkSgct{{$4f0T-m;@> z&|Qij`X3u~bwQML2Jl@Cot)llvhXf$!%3f(mbSxlF{~ZVjOwWko|>2Mn2O&hx()x@ ze)9L3>pJ&gVE|&Rw5$x`&(fq&2+Pcbb95cECVXgW>VfGPIO&N>d!t=;d-(9suVkv; zIt2|`vFM#Qz)iM@WmnG_jMve-!Ijv29*x?5sDeQ9v_ny+WSQbla*`Eu4_G-tkW_>B}7w)0nIz97i1#2)%{YmDtOt? zt()9D_FA|ajfw|h? zb36BhJ~-fClRYb~N(+O*-AB3^Gqb=xfV^ok5w9e0Tn(&%ls^PIW*}=cL3gb_8NLvw>q)3w)1p#j+smN_l`P z(5+djn|2iv&e);Uy+Dz9@=fFqGhZmxZ;xBMn|83;O{f}lL= zfArmxkX3_fa7ymai2xQaX^a7mwhX3O*o%y}!d3~H5$l@_eEbio3B@#cux{V$km!~m z;w{0hsJmG4@p^72l#;sYi16c=tXb{e(P$LiC#fO6+}DqIaqTPMrDtiA@{2Bof;P8> zm;`+{Br9b}3#=!~LuBc>*87lgb0i}9aJxpwLi*32iAV#+7a_RosUp?xk47Iy_qrjI z%)s}s+7}a zH70kQ$Qyp$ylb}G)ee`ZQ2E_&PEOZbw3i9JSwn|!R1Vuo@CJTyr3=47zZPOMN;>h) zIoZ?IcTy3%s{dkO_LT(C1o1!4tRNb*Zd=$jw6bmj)W!3Q+DOg0A4yJ;9`}T51-BSZ z94(I9viy0ECif-aB?}TijY>3Gp&OhXXeL^06?~~@m=}%tOrpIs_wsw(bm3{U@H}MV zljclW&LGlTYG*(7MRwBWyGuRD?NYm{yJ-YB4f6v}!yAd3rwUhehw1+=HxbC8-Lx!k zBCAF(*4yUKzZPnmzy7CEqF}Rn0mI7x&vbp!vXVs zS@8QILlxwWm^D+6Czs8-Z?I&A!4>z{bvNV@7n;t9NPkX`4|a*Xw}}va$jRx<`GeOF z4I(7nof{538u*NFt9rzpRU)ccKgPR{{W1mg0o{sf-@n(%^%b8(N6sQL5%+AhQ z#2%lX(Y@4FqIPsH)@1ql1U7767~7U3f?E9wWfDdk+?uYAXlXQ>2;~)R6`|+OSdQM5 zrZ&l@(T-Uj5U&nvC$5g8zoq;wLYi7O@p?`Bt1JuP6_j5ns8;5QlD^r1ksH=>XjYHm zVrJOu^l+qd&C$zY)WHbHk$3Il{Z(5F(<#>cg+YuoybOD@QfB#k+MnWx>D^t!w_eBI z^Ely87@i%kyxOTeMG&u!3AjSE)vo*{%+A|mFy3uGWh}yRFM2|aODDp4DblxdwN+gs zEWe7_&9D9~(cs;I2X)BUZSXEF)@|B5W}Tt^&l*kho7bOxHO+0NN>n!q&h_V}Hn_eh zeK^0~lZALRTZ|=mURugR_dyttaU76|_Qb@Walw&H#64lYU4Hwh&pXB6AI-LQTf%Fe z&&x;Ku1@WWm#-;Z&iVD&k}+!_3lEHs2j=7;Xq~-o1g*0Txt_3yLPE>I(fj)02eNNE z0k*0^a>&jJ44j-mU)n|(Iv}d)T|LqVTGC&=#BfD@+1%A^xVe_e+iG=Su2$4eF^|Zl zW%U2xf&=UzJ?{l?itF6<6z{VmQ;!E}JAZ+wYNjw!&eLy50T8b2(=%v!?H;WrZydSQ zB$L`!8a`UhzcP50%2caub&u`b;1x2!8)UPyveF(HAppZ4Ls09bYUF&lsl`@yQDaE_ zFrAM&EW39uXx{Z|_hQ|TJ8W$Eg!3c(r5HC;Wfo%=9bJ1}6%DV?45txE@O$(7_gjB` z#$&Yq^xlLVNN$o5Eh&vA4SLgL@N$lmjZv9Z_DP}g2Wh?ceUMHdJ2)+ou+N)LsM=)y z(N{hgSD5HdYKKay)h=*kEIAr|acsZ&np~=2{sQZDCWdTKJ6P_VvnPfqc($?12SgU&pS5At{ zQOX04nBUIgu<$0aGi3*1N6@1f86xtaU97`UjkW~MQ?pD~71c)kB~g+I-hKx4@mvl61eUZ6|H6+&$lU9k_`F~u7}m?7SmC6<==cQR z#h^YA1e&nv(M~y2j9Iqgq8r&+G_LKk9+p9mFTyEW-atu5@`xRu!=QyKo}*Jf&kWmE z`+=<_o_a|ZjaXc6@|CDiy!pm%=5k)jbAebUJM}^F7g~|_6ISqgX38f(C(bpaivPo8fV=w zmX2%zH0D>h`XSF;FQ{+%PQ(*?C=gOSccki17THx8@#!Cct8}ZpqAmm;Er6F|R&8`5 zrfkh&EHddjAL8WVGSJ`u2+)nT6{NsY{flmVzR+>@ENxQv_R}DR{9hjNwx#Bu$GL94pRE2`1vu4ROhY+ zwW>MM%|Yk3oDo@Y9eRa0%{f?z`Xx$t%&UHd+L-e_ZmkNll)7TM49SG|lIprc!l_A- z*17~%CXGbUMKz^P2K@KFqQ6Cxc4und0c7dJpu*HX)dpJsFAYsL(O*x;Alu{j3v zmg0JUjH9)~*`M=EvwnWuD|tN`>lR}FRRn+jch6g89}J7*ImTt5G2Lb5UOpWwrL#F8383;u6A_YbJ!W#j6=(h{Qi>z6tCt8ang#g}0= zdX2Gd{%4LqjNl`KJJ3j7f`D#V8rQP?!&44#FogizP)xaFQa%k)XxeUA}x73k!=%3RkbN0<7bL$pkg0a#$wnziuO!iXrP!Zo?mD z(AbsTMOj%|9zvVcn_n?4Xi+t@xx7DN_xFvs|AD)N4e9Ao6pph^R}nzK?jpOl+ugTb z;_K_{lMq`t3FbYJ4<7{b6}r~xL0&cT^9b<+wuB4Dl*hM`Rdv=Cc{R1g2b*r##|NA3 z&vX(hCt+!=3zt)X6bfR5+CigVOk2pH%J#Qv{e^s7t*5G^@fhUcih@5eUgWq$BvEB` z)m5n19Dq{re{E&uHZd_>{w=NGv!`(1TalAl3vU-*Cl;SG_l_;>1OVC9?hn~U+Hdk4_54BzGX88oY5ZniV8 zOQ7_dw{WJGnoB>+QO;g8_EQtT=48{f{D32yKmhMjoXp{DfeD7Mygzy}2Htw0QU|`$ zE>ml5V5X&}5aCJUXKj@KIORG!=L|1ryLDdKfw|x z>Dno#(yf({98T~SIy*aMdJXnbwnC$+2!!Ixe*`)4_wyAom-bTM~nX6X{!Y_p)h!n z67e|MZg&HQK@E*~5@Lb}@bmJiXHaWXGqwoQ3>}r}AM*sZ&V5Ry6!we*frd8n$-^XD zQH7dn=D2Pqu{19~KLn|c5et*@A-WLCwqq3&?>3d$t>PpUC-1 z9>e#upJ8Y3`uqFuqL6$}h3i|oopz_gc5)s>orvuo`EidWIS50HkN4M0yn2|G)S_q3 z)@PPZj&@bN)U~xyU`M>4yQfoWZ>aS2@k({?=IQk2Hu%Bhk9V;_K8EQPl6#K?Y)bw+ zzM7Chw=;rIXzMA0&x-PEr9x$3#z&PiysUlN^PMSfkhlTIGLGXeYr!l!CS7cED6?QN zb7WijQI!dEtno-Vtp$sr+8=Mpum1Cy%2%s&gcN={`MTN#%JRzyr!e%VrY~o`)OF2uSY7kjhw1^x6OjCQ+F#DdvuBs$J_h&&!-j5D_>Gm zlA-QS!oVH9#QR3u=aBD+BcatvuZ&AzDsH@D-gYK5bD{I*UO4#%10{(l^McZ#R-}1v zq>wX)&tSmVy!Z$=#x}nteMs;{El#6|wc8=;N#(GDL`&Ki#EKmCiWzPSBj76MO8_icd7samxBFod(X$?BdH-Q}KKO%#&g=(}0HilP zm+v2Sn+rG3Hos+HJx3gCNV6{t(rtZb<@L?8jm<;wfbpBUsi@#l9)7(>=vY~TsK@{b z&Sm4~q`3g=nm|nu7F1U}BWwK>9cM(FY&;r!g8K(Jc17!eM!Yj|S;gld9gx5Xms; z=f%5jmpTLl$>jP8rMmn;S?A@tx0bQ6S16~_SK94qKC3Tyy|mhv)1MrOyD1fZeYvBR z0F>~5yoBGiS^$7dX1)AEV~?S|%i_9pM^%UA{U^?yP&$Lr*bi>%wTS;5jMlXaaVHhO3+{oWV9 zwy%VNIilQZFze~Y`na$|CgAku&fBnE?DH*bNR`+{W`s!Tg4{ZDu1q=+a74)12pPc1 z^%JCBQJrAinzeWB5{k^USWLalpWN%fKM38?@jbb~AXe^zTg)qE= zuFws6gIXE^@wvif+6H4w4*lVDYr$*S6bsw!l*{L5Gv_ie;M%`e0k$v%(k@bX&t|+d z037hCaQnxCu3FDo7WjbwD#DY#b6IYZ(L~v$C|P&5TJ`rgfs1SQC9?O)DcD!$97};c9RN=8+9kGCji=Ks8ODdnnD?t-QEV zZ`DMg9U&%=SNfe=w(JA001>1jyz3Zagq!CEN6-nS5xFapV6*jIMM2J^VC${-;oIFm!DsU@&Aj_s_7y{l>l@@G z(=*Py0*?l!f{Ea;uo>6^g)@d97$_{1^pMK`e#7qlMfs!)0QUwMLrGAvm!UJRMi?Ar zPvNechb=fi@+lHpG@nmvgK@#@I*$WbjY5DUnV{3VSUq35Ek*MzqargsV*#nDi* z@XG}FlYQZRC91Op{z@R11X3gf2S6AEZIl$G#sg|-ervWFW+Kc5j_~I|Lp2k3@ED;> z9|1P1OH$-d|4tmt)s-otT8*elLO7w=%`L1Kw5jsH^ zy7-w^d)jn)i$tO%qI&(&ukx$?fg}+Hw*mCD($7-mnqtk}{w+cYn9cRPd_3BhzuRIO zxv=yugm8#Ndv)Y56shGh`*XB6KPgtN?=;G0jK7qIM=hGw^tjpxzofMYCle|39uU@& zUj$c(rcMmfTm-HNRu{+#gu~dlm9%9p_tALqTO(dMvn`sc2;rT6rQb#oCifm2IlT+I zBhqZUJ?~tqIB(nj44=BRwN$y?H9+RI19=xTt`)qLtCu)j$nx5TmnbC3;dNom^Ubs3 zFirPxC$`5dk+aukJYG|t{g6EAr#!utoScln6Onl(L}OX_NwTY6g3?<(5}e27A{Y)r0kn^DpK+duY}Ys&T@(z+SM5Jr@@e52`rd!so4_p+U=mRXTz<@wq1a#{U^ zT|;tPn*MaJz z59xXOn#uo8xP3|Q-;yvX--4ByLsuWHTM$$vyiYbJSRVk+bYK}du1#AVGonxC*9el)`1%Yl-C|IoR0<{DfwX(E;g zeT$xY6K@mj)8Z*Ak1lC{l~^yQs&}E3D2YqJ%VfIBv1cK2%&uk`8|V5qq<^EO?m6j3 z$iz1cjImjYVzHMC&0?RWQ)KT+3A*^6Z9U4P=Yv58Pm0`VcgmuNaZ_ac;ptU+D6PEi zlWgL06qUL06l+b6vS?di*^*n?Oe;9$mAL^yXy{g1Zf+2=#TY2IDy)kUIrI_He7}d88ymMra3$^6OYr)9fTA9$qjxREY~T_M zYB69ND2Xj&P|4MyCsGE}ar79PwQa~)MG{PB1xBk0`G+crb3BPoCdbvW#?m4EQT zf~4~`cSOm(NHGhCZJ{IAl&yG1^VFJGHT_Zm1B97F==h>YmewGX4c}cmA7;V8oXnLE zRu(4PtxnF(m2uo4 z*BvxJsg0y{lONJCjUtwGydS#c;-p4LL_u@^yO^yLqA9)06r9=1i7OJ&Y0zq#%kX3T z9ZJ2yy}VCaFgp46kDTz=a<(?Yvu&l!rM#)d!CilN*j2O3j@&JUR@uvV75;-$Sy_R* zK1hE)fT2EcWOcTBwO}5(kzkyt9mD%o)D&mv+*jHZV)U%7KNS@fIh7a)>Dk((NN0dH zBDHUDWCTgl1Eb) zJcH3y5R6U@rLXT+f$gmJcxrPA`|Jny=oiE0+L02R*a{Z?j!g|s;dyVa{^DBco_P#| zD%NbTy`SS5Oq?0UuNpP0J*#hDoEf4R%cgF1p?(qSg($)Vw7Polz@7w|AZ#8>S#>hrNp;tSV9D4sXy zC9_3SsnB;_TGbbHEN5!|Q@4;mSAMD@~2R_c-;k=_s3*@GK8qQ?wUauBWBF zo~n6KV%|ql&~WBHK0Z#k`0@cjsu5821;lm)(^&^l;P)m8h(UV_wKE)~obfO|xd^|( z$ZG5AUa?^_8Zs8U!8iF8nfA6CX}!P4+R}3G-NU6{fa~pdzD{1JI%J>HNNPxs_|4Le zSKy2CvUxUHOHv0_EK|ql@uopTyb;~(*WDP2VmJj4Q*P_zeH&gZHe3@~;^wv3P>BCn z6uP);t0k-*`qZNe*NivejuVq~I=hq^w-_=wNSvFKsndL^4*6nG2La?GqD8m(4F|IZ zmSL!OIcgyCTFS14-?spc*+D}Z-Fty!bDHQX;wn6O6wK417hd;xu@pp~d)-m}V2WYb z`#6%WJLQ6z!Uct8+rW74vDqM#x5tV!AD^{e>NJXL`N=^f=@r_u*1Y%04cW-+sV5qAxxgEdN$L?iwMV3yL@4%>*Oi?qlUs5R}rEG%s*sT-&Fd z{rXzWa|bW=ld~=}AmK5wg%7C!*`d3Vh8ZMVYD}npC$UGu%8D5xK0oJ3)8gK}#?7E^ zMFNx5O4n@~B6htN0}dQ4Gzc<*%x!rlO%C#+FhmE~)OFusp`#R4VKY(wVXwTR!k9Nv z3Ey;J=c8m8+p{|66$EZxI&wmE#{jyiF3Qt>O1Y=Q-@mtj;Jx^VGv?4EWS6Ju*ftUC z4dy|em9*|e;NJsohJa_g=3k-T4qVle0kQZBly=N&wVpE(aZXPQ;e9_E&BsJVIh-nh zu3O(X!&``?hw?RY`&(i)BLOG-X$3kaCZ;KeY{RLlDw|4zWTiuG?iQ{_$9Ju1mrdT( z=onIxS7`6wH~q{;-=3z8w-43Pq2_s?d0>87bJFY$BKyI_d3HcUilw1jVed>(rjL%9 zzM7FU6vbT0YRByGN*2i+Qfe}nJ}yqsP$GynDo_>Zr6`@BT522FXnv+QYf&!EmQfq? zw~Y08^)wV`Df;&MO`X+>xu*>T`5(f`SMGo#P?*~{bD@K9&;(bK_#Ii3k;7~eAI@)U zfE=(MYch{nPVdtDGxO+*HFWh@;4xdt3*FQhtRVl`FPi`2a@q8aXVua)4@3d}M#J@_ zK76j}NSMv(K_ft2_`Mp!pat21X?89$jlLZ+=YXio&lK$`COF!(9V4=ZRm)~)zTTS1N z2Ozd0*dixe?-!BE+57wyQ}6uzkN$a(J-?8}7^EOnU3F$4$O($0Mr2`Eq@(iwV!x^g z$U-=@VQjkAK&$FZahmU^W?V}%Yz^k1Yz&~r7^m^c=>}uO$Ldu&l-x4%)?#hr%j%kn z`JnUO_NFll(V0#YUpbjge__iS_n?!_1G$l5m2QIT$)UJ4aa5$85n;(c$DPF3UB)HsGH1bK1p(Ymd?g7 zyH{w$L2CbfeA{LV#xDg)nEi9cb?HB=(YUTmNKX{%Jby$vX42MQdQox^%}MPJB{2;b zm^-2Ji29_+!kMsQMEe}=qcv0>cYaL2ni$p;IBpD#->N9c6Imsbt5=bTS%igJs#DbD zp`5+i2=K>8#^7}UjN=S&uf$N$!o5Nvw=-EXGk>&Pk3uG6K)`7Ou4yYW%z`9Ii~ z@{PHg*7WXF5ob6P_UrOm!{5GrVQfcaFaJZdm^nqx`u`82#nt@tG4s^LPKMN^(lu5) zH%yqf&Ab_ARwY6<6wqqlg9j|nM@^@ivNI!V&JuoCHCCDVj}~P5X%HpM$@ zvHORoiq}&^a0Q8J9N46V(}K%r*wdLv=-*y{5tInYQlTX6AZ_Vx;i;7V5G!nzc}n%3 zJ}IjpR3yeE9DH&{`4%t|!sulDE1P)97y#P;C$hv;f}=hx#P*t(4wG#TCrLTFP+U{B zf?iR%6D_~u!GOe-ycRO^Py@6v4@Oxj;7nEu6flCHzq7#UrsuK8QCQ~^ju4>{=nY6A zqF=y#5>X^Uf6)zG0F0qt6HX<4eFfn3fuB399SB=^0WI&_nFE_(b_F{uXp2WylT}*` zc>Fk#fAz&Q>NNG)3ZDF?>A{YUj`ya3_X{fCSQ&=(C#0Gjo+6Z+))?swB;rKKEizW$ zE57jMx|ND(h9c9>?(abbN1$b}{oDh!9sqJZ7+4+o$xe_d3G!?y|2ua`~;WYw8 zxT}(Px?@;Si=86iCO(xWP^UjT?+%Opq3lsB{)dePmGLmX9Gw;QdPi+eTzkeS@z5i; zyZbYT1clXQtwuLlTDP<8k_=D%*fupSEc(kl+LyjtUxyZ=sNf1PY6k8QLkm&$40u`8 z^4eIgBURUJuPatOc-3lJ^`{Fyy&(juiNd}DdW%nE!arryeN9wg_LcVq869koRJsY< zMdO$fat!x0w_UEFyIg@xP!0XGFZ;o72Q!bTv{%#@f}|p8yKP0xDk%GwbU%+_#X|$x z{E{)^7D)7SZqpGHXE6N6%rj%xlf;AuGA?94x8dQC)|B^Vyex)!y1t;lewzDI%ka|T zt5b&?uEcR6>VOf+-m@jMvJsYr<5=i{ukw?A=YrURUHzqgNy_HUgXdtx-rcWgNCsSV zetMakAMR^eLsnvJP4_xIWL3l#3!-}@Fpv>nRBcEq3`XqFxCJf7V0Q^5K8SX1q^cGI z*s5k18y5nl2}}={&&pysa1{zFK8qg))4#e<_)!}6%XbQ%b9lS%8%T`=YZ+-t$@Vwx z+q2l-uF}y1&_BijWr@{wO4Rw*%*o%KXLk!me7fn;uY1C?8Y~4{4}Gf!*=yon5Ckb* z-IeZQMG<93P9Nfw1tv2lep64jtaH*KU${QP!!7p_m)%rmzGhMAX$*Vzb+u%sXlMo} z>frS(K_Z<`{7@(Ow_<`0ivn8;gOgy_*rv|cskKj;g9_`RR4xrR$cO#WqU~#x*oX`t zqwF+MYz2Y9L;ClI3x|xrlEHo1=|JkhQ~2h}{s;0+UcP^W08YHjwABokg~7G&L-^yvHjGb`hbi3T})pjvl+20nK_5Um7vZ5)bPU)bG(~Y1{rVbopLga zk65B7B4v!DL;o|D+|I0)7RpL41c#SIZLh?Ifhxub6Uy!1RPmngx1PP(^gMP;Iqh}S zv>`0Fo9Qitk@MC+kVj?HN20cXhnZhuwDQ#x5)&W;QRpmS0rv_TDElO@_Xa2_aQ6uufDtGW^|&n4ZFO1W5>>9XQ&Is#Kj};;Hk% zL`z~M?X$Wiv8ZACcA)0ZKY2}e7lWKp_S7vL>hG+In4X4WS1CIh%nc0uqVB6a)#z2IQJaqPRx!)G?s8jA?~R8J9CC-Uz6%Bkutb9r*g?3hYE!pGZIkNT zY?iI$1&z}=*~;rOpB-FAOz-J&%Z;!VzNxCEmX#Zl`b~HH!9D3i+4sMnp{XgM9?ANQ zz4|!(x?w7)rd9W<4bjM`f4Pp6jWpfCb1OPu0lcwiP1B?PbwxLH2bE~VO#D4ML$DNO z85r?{d`pPX>H3V`v+~=NT}dJZFIBFAca4e5jCD@bmBo31?((vuQ9Idc{EfWwABoMs znhp2*l}j#X#NA2wyEBU?rcP9M-CMpkcK*xi1dMt&R!0iiD|lzhtVX1DS@g_L(EA`8j=_P|7^IP`RAqv2>DzwBQQ>hS%G& zVh-e7-wDvAB7zt>xf)L%J-!!^P%yP2uB{I;G~`LrzqAXd4e$ovFrF@suSGzc=$3AK__*7_5+R&0^><-?>^^h!R1_e9H05np!b%Xmljc3w1E zh-}e&Y9#%hWU{u=TeiyT9SP0MZ5a@2i97`ZFQXs#a<9(93K&~N zt98`kdVG~wmCG2mgK0aY+ip+e!$#UuNFiNByieh{$era1nCbvUN~;v|SW|SKa-y2| z-abq8xf;fBhm2PQW2!oOGWsn+0(y69P>Qlt(2S_w$ZZd-;CfSTo?gG* zc+^M{hL=?-#*%0Jq@Iv1dy3qBCC4$aYFDOunxM-mJt3O^L{r#LIv&=u&^Z4rv#suW zke2Gp0>apR#9L%O-8U?6<=9Yi9bW&1k1)>fF%G`j?z+^Fx;WA2c`b3O3-)QQyNNU1 zKxudf(5sbMjR#O-_~s^7raCz}{n&Qx8oFm(H2S`s2;-{1zds~1T>84Y7n5sVt;1H7 zim-=?qPHa_at6NY2&RWqM+OJ-%&w@MGd< z>$8HLmBQ^A=Q8rOm~nz8L#sNF3vy}NR1Q1;cslZ0k0p|voa+803`cQTW2yDK?ggKA}*qG9TCt%e++xf^E$a?ExL->;sbQYpkyZ^tS2E^nC zN9i({w&uWwTmjM9kCFLgfvwS#W&;k$E`44zWj$GW>mMa8I2Z@SbT`CW{*u}N7S#Hm z5&1*atAxG+IH^i+z7(IqI5KK~e_!B1R;JnW`k7u;8~5tIsiQcH2hoa)fBsKi`jhju zQ@8gfM_MyS4V$>f^$YcC&KfV&_R>>^$8m8#*JK!SaIpB8Vt*ty=q0mW!V11%9CVqC z>kjD`Yf3^7{)9)xw1k*1(7vRTW%^lO*AapGw0ieS1)TQaojG==cwXHxoR1@l zUX=F2cz`3=m@HrvTW-+)qV}bDNy~FJjtccB#E+#Lv}5dUkH#z})w)zMsPTnozvB9U zs-C_%wvNZ`QE613NGVu*@cn`CgRi@@kKBAH==BA<7MLZAu!1I0;(>f9ZigFBTM5go zlcVj-wyrouF3$xEy{CCe9njAH_HyWV9ECTexhF0J-K_Z?VZ6R8qil<cRb4zzLo%Zx(wq)4uFqUvEw57Rb=dXq3g3KA__OXIHZQYr z_1eIA?XHKsgVpZCin6Npr|)yD@NH5T(USVu!rxZN{L(Nc*;1E~t(691ktFosoY;47ql`xD1WrFQPE_@8~ zKgYL3(Wy6vsvMVH4`<(oEtYMTQhs~J{b%Q7&;4cl1gOgOQ_jDL)79{H{#t>@N) zY7_)tiN8hW){&zQo|>wU$(m4*WZc)MoVLQkoyz6ff09rnIlvzycJ?%)X%Dw-53NTz zNFuHCSH(#GsbBz?k_3hHy+8IOb{&qBVj@@39H;B>V)s1F+aD{N=_qyHPPv3;b%ll6FhLf_dLIYn_XTsP(GG#a40Ac*%!K z1fJPS@S4M}BPn^BUNnBwoWzeM)Kr<13!h7du|WD;EbXlXYJw_k%KCW9^Qccv#=^-mO{!z&rI(YUmn+dY8VURdau@rCz=HjDM zUZSX+97x{X!+go1%4S_rcZ4n$6@gl;bXviMC;|{8ZL+i7D*}}&MVuja9}m#|E#vX z@4NC_&$S<~Nflr5yBIb@sU_MNQHhCumr|uxlcgqUeg`YrL@VV1o=777sR6Ft;o$^b z0JZa5n-4C2_4`5}DwXf;es*iRC$vmK=f0toq#pHW{8CSuctVvL*+=myCKYDkcB75N zE1u6IBedG^maqRYYQJx<5je0fQ^P9YVQW*r_@=S<>T9vl@yX}9IUy`ZT|yb-7;+_V z<~8{X{N^=(CWJkC<}1w_)c$dD?^9YOA8rS}(Tj?=j0ap{(B+t!o5v?7x0c)N9C`A6 zs8SIxbg%0L%6z5sDj|5%W$X$6?~uD-!bCrP_Iryw-vciQ!$Hg4@txxs;NaDv`S4AwLd3gLh_(3E+H;K?j+RWvZp<2p<`8ocIYdE3h(97D?_5Q(JHDO=A z%dS>!)3F|N1LMlzqq(r)@Jf-;jc75g^alKlZvd>tp{%*-LJTlU0!igW$YEx)D!3&~d5R6`2?=nbk2gqu? zh98b-lAwx;$^^hDa&uq0Z6v3FEaIjdYA5*w<$c8N2I+VeQ;=qUwp@e8nd-Jge74qi zW{t8z($iAm;f){M3yhN5pM>A&-HnP07RsMFF6p_t#G;MDbnDNs#j-9f>17^?6~1u` zCLk>f!SMF5sqo|*DAxI{Qgh>D(ESgyj%s6!FS6I_9=N_Bc6i4XkTX!Q&&*{Mo0HR$ zc?Vt>zH{;v$6~6+Sd~hnLf6#&I%@Yu*!ennGGjk8kO9_&q{Z4P@ORlt=KPSKT@%_glK?R#_LHXR*hVdq=6&>#jcB(P$+BdJ0 z_wk1d-=U88oKp$*!wy90QDf-4A&_PRaTAs7` zfOJf+Q1zWK*1zwXM#ow^Z#B4cn+=<0Igk>HTdsJFWp#Dcq(lp=d2~O{p-Q$&XKB35 z%C&CVcCXt>kd?p}d0-gky-BkmKbmmRrjJU~E5yU1Si5wb$Y%%sdsrx5bMKwsN8%t< znac75f5z7T&d{YifXqtknsrd{BKzR!gWI-~l|}RGu-73vBs!ndx2`#LF7zNrG{y&~ zf<|ASP^q9(`;LAi?SN#DLGFxSUvYjWy$g?7^cWGa7!nR}J~)2CWWXfZuMZX>?~LU6 zj#CWV8{()P`Ag~;T{nbGv%bu~l6#tr7js+??snT}?1N=8@iC6RNr#_=n6Jb{W3)t? z)V}iT9V5R#?#b%e;%;eO`Rpt|6b|9A<=nRqJd9@yyt1%w79N{_zlSJUy@#}6!yKoP z*T#W;xwHN~-HWn!O16f?H(gLoykxb8*!^R-v#&+^O86HE=JG#x!8Dh)+BKdPWRoyV z{_{(yryCa@qH%m>W`MEwh1(MAqnG06emYG=GOzEcRiK(vv=i;h`9!SCZInKR=oyA*xP`B>-6;W_PVjk z+23t!^w?r6re-*>Vq#;p^z|s56Rde)L4Kl}mmTqV4 zu|H<+25O)k=Mlov{WPXWsdH!Ynfs;)|x+l*?wo5Dqu<5t`>2zq&H0}7K^3&L_n^Ulj zlv^yRKg%l>i*qRnHP)((39!=se0JgN_mv(|G_1>N5B@kSzp%W$U+9V2AK0pm-}f3> z5~`>eUP`uKGOH{}O(RF~cEl1fsuJDhJaIlMX0jm~WE!_a8s;p=YX3v=vVXUNIRTRy*jd z^qB3&Nzm+rL_p_9xv_myU4F=fE}1q#_r-Jz%bF(XhBw;DRGO(N@=3PIYjad@`|cY% z?cc{EpVJ-}3J|PJo%pchUng7Vo|LT-X3WD1hatXQw8cVVsWhu)JerKYelzLsa3Lb| z-?R8DyHebRNed)`80#*->!OnY2291&vZUgE z4+Wo`MBKtXXz#Cc(9B`44ieNFCJ)KqeKirUKT6WuqUK&+sIY>z|GDohDFNWa$;!)p&pGH}LXGIXZGvRilfRcD|gT?E5GQzqkO` zr|n`AUZp#DZQd8z19C?S5AN`aDUE$0&Lvr!c z>cbbrV?S#oiJ^}sg)xMY!v~j1lZ0TLdX;02KJ=IDm054UyEAglSy&wH?Vpa!G@B{* zC2YL@9bdraz*#nqrdL!XJ@T9yLP(7(cBXf|;kXdXNi3>$EfcKAB9^r#{gXLP+z!}x z&1=-_fV3`SfnB-BuJH#Fa!?0&)tCU+L60AM0Y3LmSrCOs4@%iucLb4!wGXh#p&vJM z^YRfqLj+aaydUO9o|u@Zh2<5|{&Bt9M52&J15FF!y|TtItW_^HAPkv`vIb50>^UUa zqwsUcFEstoQr1!&J~L87zD?gXJz8Zp*(|;a3MY!4r71sW-n~5Or-v=IN-B}MRC0>B zME+RUS#GQ%Z}JEl65H*FLnooD<#>fg+o0h~4!}=(Ua5I%*13`WjWLhFgED||nS0I` zZC9WqU-PhbOLlsz+ZPN!bI-ola|Q|`#;rA;3kzX&qldGmNK958ZA7P-DG@R5FF%@6 zx|ZyJ2&NdHo$-|5bg250x5Je(yPDtZo~KkUuha0+zzQqp`+V7Fv!wcXl%h~-Vv0n< zo&>@l9N2`xZp^}ORRKqT(RQ| z5=-SjxIqTd?ZyJnvRBHY8&YIxgn$`#Ac;jaUKq64PncY|x^+sG3zduK>g^ta>Ed?|d2S^c8dYvUUIR9N_?Ar~_N$BkrpFZ+)KFz5|5weim-y2Gt1khN%{h zy8>NvEsT$wyJ^S6qlbIGjT=}ZTf>(S!E-Wv7pt|iw-@}k8wkyryN+22YzHH}mFGok;jch-& zhBz1`tJpp^sp69^tAxmhGJp!Vz%Tid@wBUu02rUbMxDDOpJN@ zQG`?qM>0M8)Rx*5X7O8cLZ?5bB{j#jDtw%&Nr{OBF<4>o<4~Nue_+$eNF5yMOTYC* z@~+5~&rhs}dNj5tiM*3BxWJDXAUAIKWMLD?gg59%o%}N+y%R@bk7ct0OPM=WZ@;bY zc2D}tU?diA=^!`R_AEYFdna#!52_2zxUy-o5pmD9a>s8r(wLvfpvSwfvpdo~H;oZ} zva=NPJ`b~rX!X4@qt@gLA|eAz%V(2Lg^IK#<>j$`hBR^FvwJqlRCIK7B_$>IFr#{! zN(U)og6W@v$@SSc#KOe|e4r#8_L%Ruy}Z5oap~QPkcs}BUlQ6uq~9)#oj>VA7<)*X z^!dPA#vUd@yX1d6wwDahpjaTXa*D%&RW>W!cVP6*p2v~saBDs|C;wbj`$D}&a-_zO zS0|q#_feY=zvIK!hni^h9s(ZaS)+R>2ZDr{nMHG1%vqZpN=hNa!r*tmpyYRNOKN_l zd=40WX={3Q=T|4;u%_6dvA# z!h3&2^*e`!MtOMYqXp@8$tW7q-!KOFP0=H?cSok*eVQ&CVK_hCnB}z$ka(bQt%z{{&Cn7_G;uj@(6) zC~A*-ne-JTxpV(%9i1uUZ#6Xo6e9)rr>0;`LlmLem=Pi&A~^_k3qjy$X(~uMsi3!* zH3;n@2T4eDDrJjD7mZ5;inIY`P~dX(pNd#1n%dKSa*EOZ;^)~hB10CjVC}h%ryUdy zoaaavAehYvsNAYHHfZC-ul9i3sdiRWRP-W?^lz3Q`iW|dI)nF`inmeE!u29hU}8my zq_S78(ID`NXw;;Rz731I_fFEXFC74A$RYjmBi_KnQFmH>_4tRxw$vWTw1(E}I=N;S znLCC_Pu$fe(IWO3V%%^QQnSSa23D|>kg<}G@5>Ie^=i|~Y8Q>r5>@vzgx2-XU0pqB ztIjySpv=L@4;~bWd^uptGe?H9Qy(c|*D7?HOAO}T-wtTI6JSdDKH5yURDl_CxgwND zFqSM@-VU7dBzY_udbH8TnGtjJ*r{d;bWn++#U9mxSUKIXjOO5Oh z-uFB_AL&^`R2F!=R~z9)$!|`(i=?OL`1dKE*IiyM+deaxk_^sarN?7-0hY+5@_s!PWZC=+Ml2+*MpK6hd5eMs%|1);j0wFX)L z3&RNemm+HSg3Sd?(6>9ce}CA+XKJi|5Rs6qiF&V!{ypRpm?KAA;KReYZWqQzJSJF^C_8(xZvTe_AT^{{`ME0yU9L1xT;iO zebL_xFBgXpw)2qvhOaY82<~aiObRL1rTw|l%pjXk?T2MOTS95oGpr*%oU}J&Ti*L74wjFCp zbJ3=4dpmX{IWjWYakM9mpqQN2hT&bmeA<-tn`mwu#iaIj%QBFKJ8S(unkiS_M~bTV z22Q@p&W~qJm(}qj%|7pBJ{CCt75hjuU$&*SiGRskyU9dDApW&JzhzIhs@8mge?zEI zU1J=R)@0iT8V-ic7i5{Noq21GBWYxZl!)uU^9SDqd?uZkBS;aIqoQf{VNwTQ6i_SP z1wY#ZeG_4X2uErXbgi~UXVM9*(pW&eAW?M ztn1CiM%~=a0L$<}6|hlXHTj(F?=`mGOy2(0IRYENZD!#B;Q;=({ml*AvsEOY1EhC8 zH#|RpP}(~)epy07;%dd`D(d3Ahre}Y-v>VF9r6C9?eH*Tr}n>Zs_kW90^(59UY5dw z?!jX0x6*s39qey@=svP*(o`198(;( zKg;axM^1&{GFW@kLx2F9u@{sMjj3w<>@~5KG|0hgK=}j-$0W$;%19UD&Bo8TE)%Db zhCtbA6cYe|QSh5(3QSJPoRI-xVGBfLFH*@M=|T#1(seCqui4Kw&M=4OTrEwsTx1ls zy4h2bIK2^y#G21;P)J&wp(O0w;!N=TF(aH-o%`MM=GI%g^?DXnvfO+WbKnZ4eQ= zx$=d(t{K!Bo?mXS0UCM9>qu6MDAoYkJ}MZe?Zg14PbRa}|ITbXk^@=UdDjzNiznJ)U} z0G|_Tq`Ip8M!s1sFUDeowzj}%r^J)Rk2*6J3~Q@dxLZ#@^4ySdeSQ1B_+Umot>$^9 zA>(rzMx}}f@StYXK9UCzV`OwQzi-jm#p*VEX+ZkBWP6eqS;lV^Ck?2 zDly=O-^+t1g7@?1^H=!96okoBM~<_lo`u!aEqkpuB(>ai#82ykCf4Dijg{U!LM;nv zoxcS=Lmkyd1yV5 zHZn_cDtIQ`DmW?w3oZ6|#tor%DGwRgdx&mAJTg|P`HbiPKc9nYvk58YfFw|8mN(Q2 zE-43`AUJ?G^22Q3HQ1e@-OQG(1V#}?ZF-e=FB*D91$CRd!^&4Td`m-HLpR)T= zB+2dxZzYL(6z!b^ve?a@?SS)k(i{*^@9yWh>t4UV@-N(k@QgmXIi^fkJxr7~Kem?a zJrt;}8Q^L-vZuhGuu6^1{U2ab8wd>nx&`DIJdokH z%}?#H>zg#s$vQJ_H6Qozy6psg@R9ie986+Vu}gqgKC4}5D1#!lRkj|+mG5pf%-@{W z9ogK%f56W$F$P$rfMfT<3O;?dayPQX@?YR!hX|31Qgx<%yh9{6feok0SF$kr9WmFd z?X#5Icl2=j4;!mc*natOs%MNtP|qz^0g;a1c+aQ(m!#C(d&3<GT4_``=fvwo}9H zc3b9uT%TouLX7ldylF~$ToKT(C5dqcNi%D0OO|zANEouM$Rp_Yi6t1ZKl-qEl>soL5eKADEVGhRZ`M z)yV($I??Kb3D#QqxdTVLk#qj^j>wv%lCrMAE(rB>V(&;wyD-kUF9llPLyO-kHlw}t z29y73eOrYH7Z9y)$&Y=w_HoJPxB+rdYUg0Mp4PyGh7oF<8|wJ0?a~L47JA(9mw83o z{#PjaODF=Z9cUP&)irK)Ef?{Z9|HzDs1WR&mW}BJYfuL-1Oo>LM^aL9vGEOf3W2p$kgNHzz5hTF=)>Unu+0}OW z4ejj}$i=~_vjltPmk`|eMaE73?N)sdx%7*FWDQ?xv z4)8bNc>3)luHwk!#tFeZ2moV%>#OiN`Sltm9$!4dDJLMiBX3BPlyimG19+w-+)z0r zRHZCHxr;DpdW6k~Ab~QXYzk?%AuF|GLtJ=oIUdl!Cd$(8Ni?||H(5EzXgDO5M^P=k zc`|9eG~iXu?qU_z{g)X)UdcO9&cVON$YZsmSNE^BP_A;uHho91S{gich;gFxteC(`Iq+>8)#NLv1{LZWKf7d>Yh0uBkwA2*QxwS55CK4+F*i)aeABGGI|NhGT~} z8vel~iH0PpVoj5sCHIBE1NZR&!HhP8Qn!}2?@5iwq^T=uaydfesv$gLJYbJu5Knv4 zB_gWw^bKXF>w$;|7Saa@rFAawV?z|er$7OdfEq=|Jl-AH$`YW0vL?N#yE2{{; z*7LHCe&3PI`IPXG>+@rM8>s!{ec>UiXKY}wloAb=s>8?}|G(VMACVBE;WwCsC|_Em z4R@DtM+-6g?J}5lj~X)@l_`z{A^{Eny5Cwo8-FGSTqQD2?3Z>_B8Qs9Ql!6Pb6dvS z*xQpF>7rfQlfJ7N9UWIo;VwtIcO+R_nbj+Bj=g0_-U(w>UN zJtLixi7R0H1b+U-Pw#-e9<6Ff{N`0!mMB{Ma5M&ilb zncgqu0}4e9W30uTku{@V<^(1^h5T{=PjKF{1g0N+_zwDSYtAZ%(UJ8oq`GO&^~L4o zPb1L5cmSwY=BT-}|AIHkGKO7>O(u3utj=s$2?vIUS@`**^33iO5LF8P?}O8N zd1(CWV-fiWARxmZkUfW^QQlvNeQu5r3r|bR%6^%$9$a1nuhtvw=3az#4U8t<{=gY< zx>myvR$79e=i>X}KVW6|;nN=nPp{e;uO0;I9b6p^A0h59Nb*=%0=V5=vs|uC7g|8AoeBXxvUH z@U-2)CHy|?Anl}bnS8;U5mO~uM?B%P%pmg2qWVaw>3#)ujg^^x6_mf&-?AOTyM|ZJ z&~3*@?=RM)uek?$x~0ooJ?<#7sixrghr@*=1B|bI=S3v#bPJ4SaSvcjSyIr-^SKSa zp59($L{S|xJJw55K}z)>^q)!tye_Gd5=hyFjYyO#?*y_OU0l`523pO+79N6R0c%bc z{IEWgH~s{u?6mCEHg0U;`-NkzSBO_T%7f-?VCUp8d9pchvU#1YhZmP@FJ!myDY)Rg zV8i#B2>`%@2IJSfyq5v$I&GCPy-|@$)pIrNRH<=Y7~Sy4IgoOP{E5a!{pDf}!Tj%} z^)t`h4deUef;p$koy2od(#IcOE}8sw!nvm^6}y|ulY4=Bq8V%4mszQ9uv82^&(@a5 zpd#?%6yDY(R^sojEOhA;M3hna^b-j;fS4o7==X{g_FA(QG>?#skS}~0FB7?h#}%wG zOq`2u(w1%4`@CI(R=<|wl&A-EZ!-{Dgb{RUa(voSr=p=vxBaSXPAp@NYn5YoB|FH$ zyO8hRN!<@1B7BF1JXNjk>Xd>;q^^|@PlHA~|NW)L4&LLf~wc#=)iXnKuhev>%l5C}CSy))-> z|Hl8JHNe%=4Y1OzcG9gZ$5;XBXYA(oLRd~LS045$Pf@YS<&G_E@s&&xJ!SkSksmA# zi8lh0n8y19X+wgFw0N{}sFUS7Ju82J8vs;PvNUOcd_*mgAjmv{;OprBnNi}`0GmW0 zN^#o1f9ChpnN&bLEtces2bvx&?}vehKbi^d&4@tl$F^l&;+$ehjdK)Swr(1|^>Z*sA-ZJ8F)nS%Z#4;Xz8chhKVw9pwuo717(;lt=97@%}l_oJ6NGq$xmUoD>`DN zixJ;$Qo2*lYX>!@M8OKW>PByxafXPa(IQDmV5hY!oL|kMCHRaAGsq~k;3$i8EaaL; zDQ`y2KCd{q=AhhD&vmjTW8xBY{d-??va7dTP9r5hNvpdk{;g%$ct4vmYQll3cXped z$O=aEv*FaDP}N#1aEoKg>E+Ez^rjgsC)@2LH8ro&nff+aD-LiQ-R)(pEYVCVUxm9} z9)osi+A`jf9V)@tB1DOeNKglOQ}$?qf8Obb2Y{~{f*Gk%o`Up$;JzXtBIR+?x zNL8batA#DFN+~NVtE#Epe+h{Kue=C3mYQ)IbIU@V=NV!+S>$REWk8&=hpk5dAe@cz zoqs;|`=Gg~%^mMjB)$8Jb#hi-k}52=aICVKACH;4bmPe+7A_60S2hjY%pqDM-etE% zAkY4MG;-cSsLWv&ZJG7p+zGHzdVrL&9HVln5nFOE)OuGnm7Q+!{jcT|-ptr~#-U9? zfTW8rp`23h)+ZD;e<`fG5r~|FKJJ%Lvh0{g)>Ee`60oIwzib)4dTqz5mAAw_i>yW4 zPtu;0N%N)*ZaJs&LGE|kP$EhN20U6PuzMR!B-qX^GibV8@d4{6OH$wdL`g2A^wjdV ziI|;PpF`nS#VtR6qZpyodgJy0k-bD(Q*~iu_Dv(i{ifv)1)WMyHYhet=I>#?5R zYxF*()mN+i(AxaIwefv!MdS(rFPS8=YmShgLubcI0X3(;8+A$#*~~4dl>>u=6|Jq( zE!qptx6;DEK~S~?5a2i#g+kn9=kU0;pic;31MI}$CkCTXSQo-`1-1cqFaeaHWwbiw0!)Fd^W~RP=DRKmz(>+>XHC>0$V108 z=lDx=;W8Te$46f`iDhcfRF`Zx`kB6w7$rjpiC}_f)pJ~L`n#{H7Cv3+HRoHWh_WL& zl(%`=*_%QlM#El;CpVgs(+_O%Lee=1?*UI zvkWd);wXcuV;}x=leMkl9jlLS)X~C_ z!I98%C%K*ehPE%vd_+|vGhOXloK6=TTEB9-ym7=%v5z1iCSN4!@z`Pg;FwD|b0=0; z_5-z4-zmhLB@CVj=zX~A9fm=TPe@Hj9Cxx97*WsKSmpruN(59;s!S2v2bg1k9nV6> z%xhFNH4W|Tu#}XVeA8lLVidG(|Lm6{HGJ1BE+Fe z;m)rH@*)goTq(4dO|rpkyLZ912Ya2!;octKPtt*aeVrdB3p{(-!)jW>@C*FeKT8r3IK&yRn7I75r+ZAAqPj}fpJJwaMIY??)yE4{G)Wp*r5igxZc@tE z8&{+E3<}siI_K5X&w5RTy%F+A^z1e;N_DK@3*ADtsz&z2z)C%S?O2(AVG-8{om5xT z&$2IHkqx9d*4l9J5di^uPzMlAfvmb#YvCG6Dna&?u2wj$AmF5zq&}(APgq$&(1VOx zfpnz%K4jaD2}AJPAP)&amA2;7)=ttU!nVbNSfuUBH6ts-bBb#6eZ6bx zNq1YEk`{x6P3#F*J}0_7<$iWtLK7JWq-`77;CD|^+1yZPNI|V| zl)lC^P90Fy4S)Y3))*;Ycu)!(>Yw!+mfDBKZLq+B+fk_`e!R?oy4p@N;eETzVJlEP z;heWqmp_L8oDt-#x?aExU_tV9U0&Q(_Co+fHNxSZPkNYV;+Ujxa36!FkLhg?BU_}q zsTHlqMgY*fZDeyiPq7Xdq_DA?Z(^zck@;b)phDd>L~{*4LK#a%NzRFcjX>z+0zBl{ z#2FOP$wKcIO_~;D1V^3zQOV6RY=!OoR+%jwEHy{zsO?oOuTu7W5w-q`kl2LPInvP3 zTKB>88siYpHTs+}|FVGeOkLQ<~w6*%U)t*~3WijG3Doy1PqXYvqz~0$^tD zH#-(Jgf;3BU^L3!;VM)tA{zNB-UsfdKW31`-9_yToYja2nkWO}h^b;-CgeAqWnos_ zG#&#FUdumL*v(W3_@2Q4ef5y)?+2w>>OW`dUWc5Pfg)$_x7Un5Flp0%r@B-o7@Up0 zio)-5I;8D?7&z$vhk!ZxM}$4AB@NFwLt z3u9z^%XHM7O>>T&1AdfmS_!L^vs<}jSgOQ|6T4~|4hQk<$K4cJ+kIKLHeYn>V`T4$ zKL6{Yncz+#)5h~p@ucbk8QkREZz&DQ!mbiSA8Y{ksUtChi@y+-b}OwWO+dhHT0psR88 zGW}eLtBf(7clo(%9mn872U&nG9*J}r~x)JO30C@QP8k#_O?^t3dtAvNNZ8P4u9Zv+sZ&r|;;zWwbS_%Z}t9 zE?vCFNL?At1OR668P}706^oo92e^DT1^eYb=9U90$$9V{z5gN6K_3)pTs2*q+Mv*+ zFL3F8it93|8ClO0MN*3?d9$#Ag=^^SoRcP zC6FsszF#o8<0vhl!F~*3+*bUl*hoDDI<=zOe4>0(JZ27Q>HP}T#Xr#(`lgOUPS4z- z_5ogxNcUxhoqv?}PFI$wI472H=W-*RRpE5BksiIhZbZq`K@x#cdKA7lYn9}k{3d5w z!{QO?$Fcm%n_V189E-}p6D`YIlv+(pOZ1dgxIvH;2e+%MCN1L9`bk^B6vm{}9f%9S zrTFheHP-jWD(G3uF>Lf7fw?AK_SVlaNY7%?zAD>^BXzn(g|s7y+Y*Qn1q*8j%>OV0 zZGIk1VAKg$OY}(eK;+A#Ze!%g;iZsH48A}ZQm>x1Yq2)}(liqCFyAK&HCB5l>jb{3 zaY983TC%cV?($Rm_4JY1d=G=*l; zz48aA&WZx$WFfj?QpkU6UiB+_|NaHR)6PBOeSv!4tE)SI{lW&Ob}`Dv`-8GHlnZB{ zzU&J>BS*puimfAY!d{5QTt4^*%b&`m2?c#IQ=Zuu=vpsedXFwh%Gcqw!}e{pun1Iu z6D!OSDOt#HUKd19TdBQy!8D2ldywQuOgopjc?LE~ZEOIE$r)z*t}^%noEn@-<7ch2 zUJLzj{cktbb3XgW#d;1mNJE&Ar!M=G^i~+`!FppvL*HiO8x&5;$7Y(ZtNdBIU=8t7 zpiyyfN?-qr>RjwQi=sEfp3EQ2$m{FlhzcX+*;|5)SOJt~XF1~J+L*HnGuNb)#Fb?iYiGB~iRf>Li4M=7G*}_XbSfRSOii7xN)4dY#F=zDaO>#o zh^1^-5S^jAuC|4piky1W0WR*pW=`OX1uw=1`IPpFdm5;Nh-VSdGVv4Z)LF^Nceo?A zIz&3{lM#8$BG4J{DvN9D=ujs>^=387#a*-#z?L5n6@9n{obJ8=P)d?Kb>gK2qbai8 z7qnaLO7K6UqCozv4Rlb&J5%H6o<2e zN#s7!xW+WiVLcWTG3yyIp;KGM{+H#sLc9xav_%st=_jz_oGrrI?M$~dCTPecPOai8 z=FsjF)~>?!#3ulGCrlDnp9Y?n8o9?~K-~B=5HV89$Yi_Am`B5FZH;YCruVGUT-WD2 zk|GU8d&_i*q<$_P9&X4vVuTKd?V&75L}C%AQ8Doatv@9m(wMEv(ZZ(*{ml3l`~MO4v3$)@E_7dk*IJok_hx;qh4&#JT$AxK6p4A%Y zHc>?yEG;jeiHJQm2??McQB*^MIG!t3X?&iK+Fds{|&U7u>K^Fupxo@bmsZU z8%D;e=H_sOsqG_hKOrC<0^%ViP{RJ1FkiUSz2c4(oso{0kh?jVIWI9zVl_mQU7_0`3_y0DypkF!O0KY(l__HVRPOJv_ zLAwMU#_H6^#%EOmdEb6^wmIeJ&$dSf&E9kH;qpD_mO6qduM*)^vhTYDxviFg`y{rt z%4?{N8?U7Jg~2|F*F=8>nIew$r(mL19}ap5WRO#$DLE>rK^9I=Cg!@sPhv-Q6Aa-(%! zQy&7q;UwOxWe5DUAddA{gibHkZFLnSoI7$!aiy&TQv0`g&0eKk}8#fW_) z8Vc`OVb&tXx=2Nce3ooyB2Q7k5i|jUvz#q@;?|u@OA1U1WaQY0Iw>k5NfRkUce0dc*xwEoBh%aNP=qxh`GW>*6kv z#=e^N(wM({{NjT$*7En)8u>nU{c{lD$5Xv&iwR`us;D%_+onUPQ&1WhGYw9Vi&!0PGGhfnACVNMNW_o@~IA@q;Lq6 zKD=|fKY6Xt!q+~wfh?2JQNxhjfVXY&6bQSfA_(J+^b?gdq4$2~TznMS5EC|*5wh6U zb0#j;|6dM~6CJpIiF(18P7rvv%R#?_VBth-LAXXeSH0B!`ma;7whhqoi03|O954MB z#OApW$imE$4Q4SJ(p6LD2=~b+M~xE?5YKHt*_r&~$^$@euwzgxf6P_Ja@+;{kp+TV zVsfa#moJ0?W;sqE1qXwHU}BBB)#PPykvI=zKalC6FH8dp4EGj(upsuaWyc-=`M+-& zYBcCrM(%TcG(?XALeuCmZ6X6l;^#e1?5JIEtLD%mYkwvTJ=IWhJZ2fc7438bU$J`V2l~YV00< z<`77TgsYwmdq=-=j<^9pHotaQv&pfc37=&LWt;lGyoq`JV5RC@eue0RsqB*mtJ9x@ zD@4#&!fgrZKNHfYQu8LCo0TydD5KWL2N)!+(sI03evWdk4_GB3KLchdQnE=WInEo* zuRtRk`w^_680NdAC%ricx@h`w$!$K)MRR}1NM6h7hBQX#WGYZE-8bIEB&aSU7tFl{&V4C5je{M9n zW?Ek+L02aZ)N01YC{5ZO6@%7nl7l-PDFono34nwWpVO@%Oa=ND?EpqV zh9(peiA{j04`Cef=157t^F_(a*6e)6(kF+r1)-s!PVm1)#c0I_RKZ;{(%{AK zQra2;-?*z+04|PMr1>-QsPRqQcXTCcN6+x3FwhZQF5G(G0FTaRAQr{SG5TEkDF;)w z9kl&&HEw@q6I}mrGadpu?h`cOj7SR?#%? zwp!n@v~2o1RDd3z0zb@dZ|BU3Wn)s;j!@%vylHE+laJ5Jdd^}Z~>d$Q!$yJ|i2A{0D z#B?W$9OLMCEB>L*{Ts_mJwFnS7&;<@S+j8NYH}6aGqIxH+TxK0Y(?y#Vx0CLf&^|a zsrl6&6Z-+ybt)JAHY z>L#7h^=Q-92GKFbA7W*jkdnKfl39*rRzMNvaOS+3=lKMA=;&8cB9+2Qu-Be@aXzUzhOBtrh&$DA+b2 zTYH#TivN@&UbViV7qNR8xSHL-0_}64@U@0`pX-U|V&}?-6swdk12GB$FaBNO-vXSf zoqOMm8a6@Zs-%?ETQ;_U6;5Ehv|emgFSTy+shT#_)AJ1o2(ZpwlYY-a@C?$yeOJ&q zJpn%i-Z%H-C(LLaKkZT}_@{U!>7nF+6BbMi)Z2CU(V7G$1;!7F%4b{ws{pC4QAR_? zq?Beqgn&+K?w7aQdx>nc#pq5FM-V!oF+>9E97kho_$gwlSerm_<9L!dfEy?12Jf1WBeoLl z!##jS5?nkhog*}^L5>8+gvlNz>GBGYL&Xd{e(7N&#yGI9##95n5InZvv8b0lt9iDKmN@;K^8B< z-{$LA8{!_P0`*s@-Gr~`M8`_xRO$W%oG=~+{J44GtNRHF;`)iBc?r#P(#7lfJpf7o zjva(tZ#gy+jU2m28NZ;rn{NE$yQKdcEri(`16djNYQU+pE|@)21`mHpFalN_FaSd2 zWKbPMeM%+x_x#R1TN!TIpXJzOwZ&#@m&Xno1DFwf0Zp}mPlGyOw27mp$Fk^CPBW=u zPcp0HjN6*mz%ZlJ$e!3`13{ z$w6+9tkS!ZMpDostLx+dfW1iRr;{A7Hml9O@s+j&;tqaxL0iiwgD(3@B5}@NUu+_T zz`of9J;XyitCR3RCYSjnxlNrlS*3C0Z`3H?sO>b7s^m?P5+@Uw=-f1++4}geOUrYz zme0C^1G#tJH!H3l;; z9&`SN({9O1*fjscLCG>7bse6ur3VU{=J!k4aoJ~|DK_;yk2Dt9 znHc;DGsc;dv!!rrN;>;$^2CePdFdmY!k*Q}AeUbb1~KGFWnD4#Ue{z{-gb-#rn^0n zh~Eay#QvNv2xq7y-mGLUyP-B%ZU{J37;qFxWGDmqd3~wV;}6>S-o<}lsS+rTZYu*c zoU9WDm{hm_EXl}V5~Q^8vqwzi)8@cOURi1?R~lXKHx7Y$Re|gfy{$6Kt`5_<4t0I0 zG+I%(>$%iYSY9yy`GPJRP@rvq4&)bY$x0#>TCqrrZPGs#Zs~opkSnTI5p)mM{cUb| zZ6sGB;qFvfhIu;L+aE-(fksN{O5ESy#Q&R1cqXL90J(~{x~V+qN(3aG~|iHt#e>J_U6!93Lsi41OiZ&qZ_w z!6lkY^(U?;!Y^u+^ZMoM9TdMm5qjn9VktcTTF4Cp=Sq)Vd!KK=WoS}#5>NZM<1K>? zH}neb`zXn^^S)Ej^lJq62i=jjZPj`_LjSx#(MgSZ__lw(0FQUP#RH01 znjm!}#pwd%>6=M-R@dOfy0i@EQS|4qvZhf;v(NA(dct1wFYZvxb!$hT;LW5kp<&HV zb~M82F6Z&-Ia=xBk5FbaG z7^MJ+dIHhx9ZccYJBKGHPLc@0seGYOq2luPcrnnR0l_H+1;w|qE$bfoyYb+|D-Alt zco3CFYnc87Kof5gn#cysYZ`a1@r23l+MPDt167PpC-laJR~)pRT24i(r$Ehp;c%IG zvPe<_X~qiSmTNNmeDiJa6y&Ys=69{V(R*NIaMK{@n+HEzV2-tQ#rcLhwfM}MSv~co zOZtt`U!3dja##v-wf-zw*EwV6a-K^&Um5v>sK5N=XBXDhN0*_u+m=agDwA;V3P8jG zl&RRUKS+$U%DT96jiq=bl0WK)jXKL%ESJ+z>o3=@Uzc2v9GUHL;17lc%$w1>VIkI_ zM>(dlg<-7K*0w6{0)S5*bS4lWH-T$TPBX91I?YyK5Dc!o;;|!L3c0k^3{v)v118t=w#>!wvi2OS|0QC!1EO?pQbZr+5V!kJVDI0`SW`MaRHa1pLT6($h z4VWD@c&p?KEt`Cb$Mwz3LUMCo0RSaN%(2)cP#{a8W{yv4UkSX#1jhf?*EvY9C3{K_ z>z59bWGY~_wq%z8#rY5IZ*2fbD$)vbrBQ+Gir-2F@Cq7G2#6E&oE{{3CC4{1j9eZ$ zdQtc62L{^zj9&wQN7_W%rKxc1WTP(Wlj;Obt^sT?A3g7u|fM}1JR{%i*9qf0F!9nNM+6V~p zGD%t|KUc)w(ZV%K7rt?>kORQ_QhV5+DQbzrkoC5gXnaS_9su2nL7%V1)A=SZhaOww zOVX4NJ>CNxm+qlM?6YYCiQ+P^29`2uX?=fFf?wp0u5uXM7Q6^KsXa9eMaChA?)X5@ z%_z>wDNfHC=w2|c7$j&yJ0*eKZ%5pE-o<#>hT$E}2(si2C#&G%i9}=r(S)y-NM5ZG z^_0W;>+7v7y_yeR7X+bf>*dvxc9_XAfx_!Vt9-xM`p;pQird96Jig;BjJJGzLm-`? zdrXFZ5YTNP^#|;{02Kg`urT(93RO#)o!0s9(`ITas1>B(7I*wYQKiC0L98zUb@v}9 zkJG;We`AY&Z3t}fYZw=Jr;*R0rUrZ)9!dx4c@N(qXrO*7VHn^0w6cF6S9u++NG)(mU^Bd?qp1MTibP zcpAzJK2@H*J#B5*8zP5gZe~8o=?Zx%pHQHBrXX(suv<7)=c3f;4MxYK4&nF_<8GimuC})T>wVGmmK#mmnA$2{ds3_v6q!_E zXV=s9*fBxf)OY9-;~tb5Wo@QABeU%G;7B26OU}1J)WrOIOMBCxQPa&~bE9$?nfr&e;)zb!ji|Hu*97TtIQ3cBNzHNehL=97KFMj$9f0d;J13)ZMId@^(E# zZ-C!#Wd?VUN0>seuU~&gO8Gf+3B7kR$P|~sit0gR!z04>T-WGyKK?g!Q)h2X8`rq< zxEoxiVA#nJ1-2_rP~ZL;pb9zwV-UF#J-pF$f6%?96@CK%XN@{*I}l0BDn)VU$gj)T z#}XiZ)1zcN7h4x;t2xb%E3muJ$F)t1KnzAcJ~f;{AeWngO#7$a;2$MPI%~vuw9Hcw zZnbSTM(&7%9@#eSz1k02_0tAuHD?z)K)dtRMBykQUzLlDIKcPZuC_-!#Si53Cc%45AjsCf(uj%ztYv$KB)a13eFklEb7Z)_{z9v5MOP_08>$@$S-bzU)-Np zxL9{l-M~2y$k)++tx{?|UHh>FVh;#Gko@IdQxtIFpWJ~%o3#x}N;O2kL5tO5AS((8 za?7%i3mCx3 z(n?!*bySZ(dWQH5*(nTfBJHGRZ(<^@QSdAE(9jBEQr_oWQG_y3|9@l*+r3ee-cqgvw`Q&!> zS>yg9gC$0++H2s-7e*)!zb-t}Q>QZJninHp=zWy#>P~ZJ6B%ouR&8HgeF)zIEW?TG zX!Cu=C?p;GXSY+&+Wx4q?td)f{AQ`xX?bhGlIo^m;cx~Kh{bN7i_3FEj>sTm^?uUh z#02ci06hf<3w<;DZAv{~R!c;M_tCW-K&J(Ke}|+mc+@2T3+nAxgp&;iD6V}ZO4U1R zaBXoFd4y^nh5V#>y+F{&ociIi2`nOrhERO#NnE)H?M!?ePlCNoiVwn>tq&SJG zh|wKDu?E0G%Aj+?PN1Hphh(mNmVBWtKx%m)7~sS~G~2JO;}P5(1TfKQ`{hn`I!b|w zYsb$ifv> zWscxN_hwgvkDe3?t_C6CDvsFE`xzT2DeI9mX+IYJFwSdw^RlQX$j9*W&&_8y6)#v} z4s#v6_;3&$10R@YT3i2dUuD|pmg~Q){5UhjI5+PsC-Y286qs1Pmhu=EceO{n!nb~Nh{!GkC@%tP zYTfW?4ML=S+c!*b#7y72fA8K$p%03PU?|WIYdkOxa%)S(?ASbYTc$)>>^cDG(4Rkc z0NAxd@a1!8CLk-`>Wo7wHZBeis8H_)=54U;QT1~-IAZ_J;#dboK<_QLfK;fwQPUAi zb~E-ARzkF!>mRpg8tNy=cCvMKw1gm6El|(#Vd8}+1X}6b@&l~rjuS_KyLTQGAqrl? zv|_bfqXteIDXR>DZ40QqQ8b-wzi+o*RuS*G#@TpiDg}4N!aI_x@~vy+nGm|Pmph9? z!OM?g`Xrw{RZ{ws-vfSj#7M>R;Jxmqe{P$?wkwjR>>O~IYm+6q{D^C5K=rU!TCp)w zkfll7!6+R8^U8HpFldf-Rxq&o+|x0~z1_rh-#o5i;G-!X|CXfIPHzl~I{nkh@|c|69_P7nyKUsZp_u?vEGUcA7y&O4+7{H0o-}#u4r6A&g}{ zA&j0&VUPUoLtGc{-M6?H4?F2a4{QNqzIVw~ z{se@Gy{&by(><3TDW=Eg#ZJMoK*X0xBhFF%7OY*-ngsmKmpl0C>;GcL)Yqsoz(l!2 zJM}B_)R41#qRc$s$Snhh?cr?YRXx(n(@v*Z1}HcGzSvgb-0uzLRS)OM5A)M<17v>h zo`=Q?GziwK##b$i5!(tVDB;=A2XMqc_*oJ^`=Lc1y;!ctdg3kWj$)ZN5(8#8VubR1 zqT}AYe@RegHL9xdt)lTgUDanRR*u;iZuUlge-}^G+fCL+oAU?CJ*5hd$7$BSYe*?d ziH`ER3dmKzM+V;lc(^_L-r@?=>c9g&e{NcPK;Qyx?VJ!_5A|od6gztyGLTNq>;dVi zH7#KTyuipnnQk2+LHx(ERDC1>ExyO_7JCsw0(CpeF(`WSs5>`fcnGlSO99X!U;hg9vzUc3T1^-E!?cToS!-@ ztRn(ZtE#G+x0TA5!NMjmu~h!6b^@;{2%5r0P1`iE6xapmf2iKd(M95a(hqP^VF2`Z zBCudjp{J++{{8zW1Q&#eNWH_7r^MedBl0u@5q4}$LC{ah)8mgikq84Ri+W<7+gt}W zb$C(H`L#D;#1q?QJZ1)Hwk!?kuKz`to9=_?i*c(JFG4V6-`gcKeadjU6&%H8rryNh z_(MIM?krt~o6sKJjHzLk`g*7#(hA-eciL!yIiulhPjtrILb}cK>vhcQEHAqPuBTUN zHm${N18;*B`f(ziXTNF;at+qJ3fIT4QtvkZ4VvX5jyG8s4?vIoCu$gzAWdj=56+Zh zP;FW=P$!t!;2hT^jf{gxGXiAg3Tc`^9Qun(V@2`mi3IR(Ej!+iJN*VO9Y`b=K^i$= z_c`LfVj7du6!2c&8;ZB8`+tnR1yGh<7cQ)Tln6*m3esH)ND3lKBi+*7(%k|ADuQ%3 z(v5_4cXxMpJ;2#F@B4ix{{PIJnFmK1kl6d)d#!6-;XoiSu-u$%`a0#1M|c^S9Qn$1 z40JFKsc1LlM%kE$d}NSFsqo3%KybIwO|fb6J}C|iX6#Ou)lB`^KfrDVKuUbUnkA}S z>L8OBej@~^sIdi46bMSkldZZg$}uCnOC=HebgQZTflZ8R#gR%yV<SlZQ3!;<6jm;5WZj9NpuIfJ!a! zq%Zck2{c$-IS>be-u6}8YQ`@u3Sx8)c6Q;{2?ArQx4S8l)!af*Pi$_BBah5>pfN5g zhXba}HpGBhwRbSXluX@H2L=$cLThr$9lyu7fqMcl6>QZ#EKn8*XkOfF^fTskF~tb5 zR}8MZ#9shGbM^qh{_Lv4;O}V*VD)uZ*;l0e0eECb{P2+wj=94h2$1=$|Vs{^z~O1G>Ms{ z@d08P0Vo0NTwLvdkN^}v_YG45>Su*K#oSSPF3TCAar=8agh@Ia2+TsZ0VpRbIieQL zx5>$3Ih+jF)$!!A2s|2zed!Y&8NIM&Sw=j7Zb@UFNWb1?e@!mgpgnccX!v}%BHa18 z?uJPn&7Z5(V>fq~8WHQSBug@Nb<+5#pJiH>Nar2bxY6^tOd0Dns48cYM?YIDl5k>h zX_@5aiSeqE ziNSc0DR49`)k-OL!xlYXk_cOvH#VEoQ3@XUWd$m>7RFT@i@e$38Fo2NfLe!H>BGsu zE*Qq15dD9fC^((kLr(AFXMj+=;eB9(q5S2=DoXnoe=;hEyR?aSzE@Kep*?Wg9R5Owk^QTpSc2=VdLJ=V zk~vbfUM(fsM28o3gLD9%Ru033FN`!q7Z-IRgFQ?~_Zi~dsmvOR-^;dsKkn=IUyZkz zapBPq-R;o>{sJ485DRpe9QmTY+as8BoD^~qa!zqQ)}c@9o*sLmCB_@EXQa)VDG0eX9BJx}t zf4%51q)jJ^;%hk{O~8j*<)9WomVS!^CG4|OjlU)%MO(W! zWaBmiu!nA&Hn24LjW;NW`zjv9G+0T3`Oo;$UvT|7F_kpM|*mo-P((b`n3qJcVIznZn)eXXWFd^@pz{G8Xdv2E+zD4%c7cZ zItGWsuu*OIU=_wL5e)toqWOz=TBo_n7A7{^hfV|}B>X4e!Y~6@&xA6G^YF5~$F~h^ zZGQmsd5!T56-Ed3IA!`sAoKy=^Gr;|Wo3T{GyfO;g%5oi2YDYK z)~-OmPTvdhbzgwEFR{`&B5QRyi`(GT3{M^w4k#+N5=ktDQa#KroJn;!Hc$xv)#7>V zk~89jaX)W*dOvTH`)A(NoLw)(>&GJ?@kt*}H1g&>P{QP9Utp(!*WWx=GlMO&V*qy; z1-Rpn-y|~0FxrG_|f(o`q;;4WPLT)c#lm*x%_HGv_w2PsbYqSFS7)-yB zY^Rs}6uj%YO%)V5vF4tF7x8$Cj&cXlGjYSH8pV_!d57gJcTWK9HGXjI9?$O@1x38? zUncH@=t@hJO^6d2GBdaQO&BeAnkDro_*u*3aQxp=y!nUPnhqGr%`w#)mg@MN?TM7tg=3+j3h08A^6VW; zAd+jIyC{%YBXeK{b#S^pNZ9YG2*0B&+X!5iKqgh{KJrG_53a8o2QAMizx1?;quXDT zu3}=m0tMCDk117SEu0zybqj}SZ$Fp4MVlX?ysdtA^7*K5JDYdrwkJq#UX zGfu0X%I27D9<`{}rgG}XzZf=1k!kphgQaPg_$T6OeABX<=voX(Emg8gA8utZOXG~< zKM;L0$dA=FgOsP00b1Vwy$$+=eAn^cHVE!1yp@XbldNbO94d$T7cG0P2-C(x9gG#~ zaqCJpy?9c2qG@gk0BoLZ&oYt7%yQAZq?=TbFSQ!8|M7cNtvS%3{0@~MwM->UeCw880~L)XfkE! z&(Us!?&hWH^NB_!2|SLGQwdQOPxSg@Uxvi7M5)xtrNwhtYvA&c4v5e`#7G=`>c$ZP zj@Us8SBn&m?FCXln8K^J1R@ui3 zG@I9t?SWqUBVAeI9)d7&)^xbp2TaK?Qn8u#&5;RBBnj4rz?T2j9atmk2=a)|2fN^X z?d{7pF%2!Ub{ggGJ(qqwLHi$&Z7v`jn1HYjsFZmr!PxF5leQ}=d1aTWyAZfE0Ymva zf1^Za#CDqt;GMXtnFMrP4)D>3P8tQI0+b+N9skkvcLG0d0x9_#8gZt_wMdK;<2-J! zCY*ahZ9_g1?-rn^;a+OdnXdDkbFtx$#m>0o^jrxis7)gD!q*_dJymER+h5_XkD&ug zmW>p>Fv}k%@%Rxkznk|!Grl#U6d~7|NiU0TfdMOH#Hw?|su_b*E$U{@13QdXf&_kJ zZh7(1;Itw;W3|o81^{Q<5+jwyXO-9csL}x!|KK$UMy!7@r&6McVh2XN+#WN{xJx6p zHa4@Nd4h_#;AsQQHZa_0T%TU-IbGdz(F^ywmMI}?pZ^LX`x5y> zOk!7Tt1X3wS-)C@1HVyVEKs1`5Z1oD8Ne~T$w?Z%9#;4Qv>9I-sexff%fLjAfpgYL z19%c|MqIs0wF)7B#8@bJ5C#PTfnoj5t@OA(A$TT>tSP>OS&eVwf*ktI$d;Q?7y^6i z!mkj1Y-NU>-^spQpm92UJHzr_j%fbOG5vAEHL_uiad$inHk9X^HW z4TGu7AiQh75>r4RaU~(1x^Q%*+^a6#ykH+ZqE<2@e+8|H3 zv&U}O$vc^lJSF`5SmwZ}>-->bPfm7e0!MTjuXJuI)+yx|7pxxJ#7s>ZtKV|$r3(32 zy(OLV?%R**Z(#F>bLaKHVq%^_wa4LzybMUns8o;z@z!6%!q|aNGB2LTo13tq|w5?-6@wR4+`zK&e2P&QNV446X*B@8QpEB>sdJ zF8&}Z2SgbP2PZwgzk20f;S2tD2rTiq)RnR>e2$?81o*>Hdle*>DKqIs^!-3k{WzvF; z(mJK5`0f3z)yp30G=4cdC2i*j6i$77Z%=wexHI4AxqG`9-YeRSKUS)Z^q4UcvMehL zx0G?dNPGq}-S{>)IQX4=ZSJ!C`G0Ym*0I-*>=H^M+e@BUY9*|%m8-+nfqfO^Tm>OD zAajmmy+0{SfnJV2vdVT96Odvwi7>j|${`S7Wr%t= zgR_M`EqhZk{bAEJ!=(;^tTt)E_k5WEG5?Y(wi1suO1T)#w(P zVT^!%C#;e`VNnD1+>*1w9lU2?R8z#QP{P9QMBEKf`b@=+Pvd6E_ZA@)qk)*yr%D|PIHDDT{u}T~ z{p}|woWsMzKz;}2&xo96^cTyauhwh3imvwc;9D*kJ#e~!p}gKErc#eAj9!n@Ok zqx+TfIEnynY5OjB*4#bY+IKFu(|2bLs||Picdo!FDa<%smj&h_$c!51XCdup5%_sF z!ICe1TxlwQ9Y=$4dm&4|h9{n!`;EOH*|8A8Pf#_k)kJ6dS~PQba?9iyR>DVJfi9T|%qEbJRv3oA-1`m$Dq?$5hLi zwV`L=m>5ygx=@$}cXqINPaMUX)I-_IJRqvN=W|Nt4=luaUK7v5uDF~bXKMh66#@@P zr>&eU;4uXZtV(KXrYx+ca}uASKJk2{#6|!-;Ze!2Bos$2@hkopfOI|qX?U}&$5%JG z+RF}(3)P@f8rg>BzVLWohxE>3`rY^`34{LGi*K7X%q;a66fCs|a3J{xSdhYbKLFe@ zjasMgz#jB-wwkOmJ?`52x~huGCjj>V>Pfy5NNE6p9mOhNfk5RTSh;pdr6Ah!RU9AS z!$EpTppXNKV}Qx>@N~Wi=W7BsZPvzx?yFsqmJf~#@2(2j(3T6rp_Ap@^X?pq&sjx1 zeu#&L_^^m%k_ihbE7uDSIHeWPi(ki&0Dm-PZYm@k#4_AVOUCX+|8DAE1!XE)#X03T zIEBBWwnc-sHi8i5fa6^0@W$qfHQBdW5ek(4kQbD@2{b`K z0)dBS;m5juKqV%JV?x8watN&M_dnb8>G-kkzm>zO9mJV44!(c~$rHIF9FM_g6ZqQ` zcc<$r(#5gZb~|}{fy*-V`&iasqz?Bho|fwumAq?}SR+jczW2N$3pw+AdCqY<^+&&J zI!7hOtFMk;-a=DwFs9Tx3|2k>nbjOzT;mp2u7Hp)oFdB=|17*c&^1YpIB#xQ&9QKW z!H+xpzlttkdjBfUz}-Co`0oh$3_^!2D?vb=6M4(i?b+&jexZ_ip$|q`3ug`ISKP+G zb_35MwG-y;+Xhke=OhMJK@+la?y-MH~khgkAKL|EDni{Wnm@l-8 z1_XbIZ~?_sHQDv^w^ugYSOA2#JgXg~Jl}uUdV*(quof8@!t(NR;Oj9^)CK^1{rPp^ zA%%A8auaqQmyobZ2J`j&dw1vwBkSqwBi(rUqk}SJSAoEJ`db z9{juf$O@BxrE{u5 zrgyHZESGR};(IhOaPux!tu0RXWA*2zW@bivk8H=~@?7%s9>rjF2)LVH{CTTq;ca*? z#3(}B8ncfgR6d}41E3s~RWvq!=ikfqn(OjWQbEm&if&)8vYwsFELXjv;*e?Vkgc#K zaPS@VM`H*{T9i+^;w)DA3B2p>A$H}uEk*F~g2bXmKqx6b}!H)chwO7V&THIfnm2i2hB!WN*K|aH*QVJ)xhn&*AHwW-cy1^ zq}l;@4I_&x4ueE`V;^nCJjp9F^rnkKf)x? zU|qGujc!OM!|>;;Djk@<1dpNXm_m&b7I5!x4C3KioB+T+8nG%Haik84&Ro#eAyzE6 zgz*w%f!T0RJ}^E7_Ck82dC^fH#?*A`>b)u#Uu0%IJMP6A9)`yZ`>_ej&JKRp#4FtN ze=wXBY+v54G297)AxuLf2xo`LsOL4~hihC7K)T)mY@-2p`Q9TQIBB9Gz_s6uwRmmU z9Y}`#LA7zyzr_XzNI+n`H`dZmpppU?K%ExwCSdaJP|6c6ot!zc+fRCz26D1`4Y8Y9 zEBGq7tW5|IGff;oD;#Q|I#E%#8otau-4qa<3W*t$E9ZsQfQ<3do9yDDmdj>_>s2k3 zm-e;LZIuBAI$o-9!P^Udez67+1Tk+%aO?nLMj;(oxF>8G7I=3+cF{-%YpmznhXfP1 zi-NGgpW+$rf@7>`bEgH4had|oG2G)M?HWB85zd>WIT~a{0jfYTAmKP2E>$EeL^ zPQy~zM))nQCkP*C0^a2n;GISS$Im=20zIPv`k1^91!Q+XSJn}@Bf^Y1Yu4M{@6J61 z8;%$n;Qj)Ersw|t4gdWqQCWYQ-?DtDNjB}Oc=4v4CG6e4qoFE5VUW`9tBx8XD@1lA z9`Zq^?>u|~MYztTn+LnU)N0*E_3dZdY!wFh7Y;>6d}$JzK*#54Nhkwj-v0B3-taXo z*#SC^Xd~ClFe5j#cITH$h6!UZIB3X_FI{R`a^w~kX=QNG?%o291{TUgM38Yv+%lFi zoM0OZpoV1?y!;tdFu==GLi}loE2((C0NkR0!5ix<-?dY+bL&=C!kOO86x)yYOO85c zjB3UC2m{77(Mwof4$OB$%1}YTQM!I4-77nde9yRVcvPFujG6=kmHAT*6wmXxpCw%u zLFiW*PaEuC^vk$FzQh)&_#`ToRAcYBp;Ty~6@pMNxBhcCjBvgq`zyia1u^C{5~(Wn+-4O)JjUTYp7G|;6(E! zyZlZz$@Vpe7WMv)2uiv@f!Od7?XO#`T+IOeg$7FZQ zo|JLDmac$;bpdWA8(_lP40!xgu7us-&~cugo~zJ{X2U@}48J=RT=TpjyG4R1WZ+AH z>euZwPkR-FkYHQ@3*gdSwt0XkU34bFKHS3k@7((JoLU*Q8m(^Etd0O~ak!=2;G(U( zcc#=~UakgDnvV5g(b{R5=Fzmn10EH|J8-K29WTcV8Nfj$zCvsx)R|)KShW zBc#rorz^U0L-YhLI?TdXZdg_wt}`O|BEpj@j%ck%TF8S2M2Pu46z}K)H4A-GD@3Ai zF=NNWS@_?4GA9Hq&Db@by~SW1%5DGp34Xtv6Bz$u5; z-*f!&88-b`89UOG*Tcp9bU?V}gmbx`ha#~hB{b(a|NT2NYvbT|U0l*#Tw!4TpU*C8 z7y2V6LuZWEu*;3^+5n=~KSe`?$XC$9LGkfV${u5Z(Bjlw`g=Ly4!YRNKKi!ePkwA7iF< z9BavNp3t&v16rE(zJz5c2EqCK&wTIwLPG)9({Nc87RSSaH^Xi>+uHN&?mPb5H87UD z?z{7eOPWweM*)^L06uu7p*&{}_PnpgyK|S3CDQi8p zvOG*>mVuPV5hSNv6S@Y@H>kc50!Bbch2zs&uc~%Iyc_Y`^{NsuZ%g3rSl25H?4km^cFDDFSNn6!HA9@^vXAx{DV?Q$)cs4Z zne$*+se!Is%swKy5Ix*4;YEk8Z8bQ2(hVo>@R zv<^{k8wD?rR4OwjAOP|9h~@4F+07u??D+9*``rcE<Y0S=r6ZdSY z3ofy7k)`ggd)B1160mr%^iSKXT51v_{s%Ra)aOfBqGXv!G=O412Gz$G5oU+2B1!7g zpWhDD*MeMdr*}i(<)?kjtyCa%yvFqPl4{5HSufLUAShjsI#Q>F zWX8nD!PS%CCyxs9VvV%xb#uiVMH?n;Amm>=Om%X*STey86-M%vepN1m4oXwr@mEEJ zkbbJ&cUhC-xpE(b{f~C7E0W>j=JwuGNBsT1A7qbpzs@@lTaSAcbTO~P z?ApJr=qU8<%-Zb3YKP|&1cBQGL0gpUnpQ6U?HE|Ju#_6lW2@U<&E{wlAi}7@hw9p6?XMwq!uOFN14OCcwKpeqttGhERtLxc7LD-`EBf#!l)rN(5 z7VkP}jzi3dvh}mQ`Rls*ZkC%Nmczo^(S|8oSFknt_nk2sP9z%2*4bu5=FWI4Ec8_I zraolrIyB|v21|J65eC1pDOdo!^>@4Prri1>G;mEcXw5qWhbvi$K zCnti@YmoTfAUhw1g)YP;wg_^ClA@BHTZeZX$Sen^E_|RlOpiF&YG*rsq{hWXZ-gfi zo!NQSjXS2@vy8I&HtWS#4N0u`7yGv>OE^`{aHc|0>PtA^Om-7#b5V1{KyX{GA{a;S z7?Oa*UfX3+PESlQi2Ax`(c>V`Y%uo za3=#ZaU{=zzAKaD@EDS2)l&XSiv8RD3#NP(x9e8CN<2w%TqNaVuf4!;VOYouam4L{ z0Kss<8np`s$*<`xtA_`a*iJuCnhM>}@>Eit^+@$Rp61E^WtXxY&s|~0KR~cg9fnf7 z5#+s>Nh`RS6CT_Ou>eZR%jer&IvKR6DM36ZpK!X3hcvXCXn*UCGN^>U21-2pTLaD{ zGHp|zg$qU?!o%6c$=?^%exGeE4T)lIrR$*eKAkk6bJW&yYtzraaTP4v>11%D8paIlkgGvtHt zi>!>-xf@i@lH>t0`-B?jS12;G81m zo-XO5nu6+?RemOXv4XZ@j4pQjXFi0HrgOcC_!*&KqV=T*rG- zsT^5sK?JBoP{tiuGpmF9T)5lHi)qW_OE=gJ5M=C|#6#5s(9nYUgH`oSGjITsMPg5x zjk*EUrtKPFZVLtwdsj})tQQmQ0u{Sw5Nieq>t~0fAymrhvA*#2u|SuQVHQ>%fq6}V zp2|=OES_m5v-FBL{B;qP(CEdcJpFM#3qGI!aLS(z;Z80P}l zD#gMfl>~^=zijd!m0d|FD}hVHir&m8p3x8Pm02X04nT*MFq-{QyH3>I;$hm)-5fuY z)D1{W@BN*w_IpcWqKsk~1p>|d>W9_;1$8vwkeT}jb&M!)wHK&6c_wh0BZD0xS`En| z-Ij@*Avt87);dtDgX0#sx;6jD#B|e@e(HG`l;d%KhwFOkTZth*G)E+MeEq&=Jde!k zZq}n$+T-WUy;vn=0n37pk-u=aZ{pAneZ1R5$89izIaj@v#?dLo0 z;TK%j0;;;j^zGM7U=V^XJv`-ko*`^#ll!@fWGkyrxE=kG{#wd`9+y;p zEDZ=WF%+(6e3RbfzdPxOyLAm&MRs<0J}13P=aB3keAh%mUhP$mzdC*{@z`^8$(G4a zMOsnOhbaL;{k4rIzL`U8&j{@p3{C5GP=xf41Y(`DASAY}IOK{m%$=QSYgzG~GI|uF zy~kkppfkJ_Gv!o98IF%@E8X?S%v@|%o|i1DL(;C&S>~8+)fvx z&bdk&{ZzY@Ut%J9)FRS%5P~)-&*$6)LATH1?Um$Ps(;E$Xal5wc?o#Dg1z3d<+e;9^g4Fm}Zb`U7I5~XfAIabSH^7-@U$+D}f zt0#fV08Jz-dQzC? zE-hBPx_Tr`GtV0j{b|jPuJA9-uMy%A{J-+zO+sZrUgY6-w8LX~;{ahzvFNoIp>J!N zvHE7rq6%XHc%JZ`;)6r%$d%NcTJxfViw{6#I)CL|I&-{fl*PV$>Rc{AGF!H%?V*1_IzWEKDS_cLf z>aT8i`kng%){BR(^Gyh$OS>o`1^UQCD5G@pWVtZuwcA<5bM z*PLPYilvU6zguZidZO-Lp2B#=xN{sHFhSQvuN@*aaYeKq^7c?PRj8oV+QpWDLm)IE zzPXc<%_!?bygM2bCdI&%R8d{HkOWe2ZluYMu0e+9aF;(vT|%6_|7u^ab6#yFR+C0t zOf>n+s60OtswJneNDQ4RV*$ZiW1^HX%M;mj^ZuOY*D-PA5_@A8c zs0?h~u)h(I*}?lU z!|D0XOW1a>&s!j2HzElO^-=YL&*Q4Snr$c0a6JACjwbLpfdE)gJr$Q<4iRnU&9<~4 z++BO#St%&qBbFN*8`&q{D$-B5#VmDkQ;$~cK!mbxDd1Y2GsT@jSiGVm2@o|8O-xi7 z$EMOwIP5CwL)v9s2~Is+3tV*@!Op-@0&zc01QcY%Cz zzr=kV;FW@!Y`&$WfJyonNubF5>d*Ym9#eSXh!&m8b7gXWOro_K%fnBu96A!?_T5;0 zoL@CszYO6PK|?!Dm}<`PdOu6t*yN;YR0JnD)Vtwgz~Y9?rKueN-=e6>x?KxkS}8gV z@BN^1-s3uWe_K2tFi^RcLFdn+ve(&}1E*B8>JaY*z+0yi&rN+ATO)^*a!KiPWh|W) zi@F93)po3MYv7r>&EDqwgD~XPpkX{gK?8Pn!-zh(8jODAl(%SQ0WPrF1cWQsgYt5P zRsYynWl56th73YIq){IF(e)Qj7nqIBfd`BsJ%3pzkKt8b=vApkT$)5&7Rw>-k&@34 zrXTts#-fYLolZ@T&&b$r;~Sj9ndLG0j8;3Ll6Ogt+6$B1_`SXYN^o=Y_?;6_4KYf; z4oZln;eX|oN_3IRgfQ)bY=4%5)IDCH@>?IjDduHW*!crK$5u?kTRd|Vor0!gzMc!p z&ckDU>VP8o<5tYf^$B24HfXHmEDhqzt|cQQ~TPN8#F>LJs6qk{(tjqnqRQZw(2{Pi}R)tL!-Vhdg zoaPOlZq@0l6L$QxgN&e*!02r*~)#=@0xK=*mdjy3~bHJ81(e?h+O(?zeTN<=M<4z@2FX!Geu49 z?)9kM!s^tcP-7Hb{w}QM&NAjY)U{#e<={HguO4masY(l4unBB_a-}B_81R)X)GMpN z+}NbBPDZNpC=4LT%`v*{%JuXc6=g&7!6$z1)cBBZnwpneat-!;tb^8)x*PVj71rPO z>}rxWCF*%-Afa)6kz7-ZV)oV;^y>iyT4R(nsHm&$2up|cGv5`j866CLi2wX!X9c-5 zi?k`*ieh3)q^J}_JKg3cW?$Jr^^|$mZt}g^ab0O;sS=eW*g?Hu2gRWL&|Ci?gT9!) z4hSjzuMUZU#+!Nod}=vT@4B{U!0JjA@%HXJFL6AK;zu>9!%O?=t@A0nde(mYd#64D0l}wFpB_jJ8=IKWA#zrJ z+%zf$#Gbp0@VgG+cw!GsRu6w*W(I4A3=nLlvd#EfOgU8{!0Q-(Y;KX)qwIZ_Nwan_ zra8Yh-*8HX;N!vE0JGIPEBSQ@DLkX1qDaW@41bcYk9>Rs<^BMaCOVbYfIm<(%xhii zo+g-!`bJVzE*j__zKfAAdLxl8EC!OT0SU0ax%pQW#sHJp5Ylx3;a%S#z&!*kG+=W3=Y^a-Bzo3e0#n#vBoKfNSW_78 z8zSm1JNDz6?G}}@ZG&lIT&Gnkj1@~Kw)OK8@G*J(Mm!reT4;AZ3LU?319ZHQ3Rc)! zU*(wYC8zLt_gM6&n)8J(iH;5F<&OraNRjt+n#74o$MU()=^BnT%0J<24#eV=h4XMD z*}ZW6gK}b@cIkMVDv_a}M_xy-`t}S*bR30vhsL@{5WkO@w72{v8EIp${`)>P%Kf-c z7xipDy>AFHUJZQKfv=H4-)ATw6igTX@+7whPfAe{)cO}h>JM45ZP+84Re}i4+)!yE zD2RxR_IlDb7Lh?Y0Y~C~k;X*(an~}?BHd-pQ0Z#4a=gp|S)T5;lCWLNBys(FXy|u( zKUR($4Jt`ciI9|iiWX8QOBo|oNc+)FL-mmUlDFW4!(y&)ZZ;=DFL*sjk9RgK0fh)C z^X<}?o`ne>Ah(n8_0b4ken2PY{tI1Fr{mf(^77b#%Jnro93JFXhvfG*FIrDBV*&uj z^3U$8`m52xHDSQfR7TSo8zk7wtS*YdC*+3lejKp~kQYzy6!LWwpN%-|{nqXV(sZ6--WVH2CBc99(ZB z=^X#sv3zHEB@ALgf`$1=*8yxxrwwRmRbaP7Cnjp0I+A-TSDEO>uBE=3fI@j0v@cOX zi*nCw0OP!d;vD#MVeXOr^9boIeExNDo$3sxs>fSfTd!d8W4!J~nuQ&aJ% zz?bxQ^ZX!G?I;%hh-(pNm@1BFm%I$7bv3A-e8Hl}f^1^%Cv5-O?qP)yR<|@}CH031 zjtP^H;qauzpmyndS`IW63+W>V`vr_zWdtrMQ)IJMcuI?lZHB#@00a!}BIpt%b=@fz zC6`Q6lHnW@wbLZCzrP5w=EiF6FW?+G(lLUIz=dY9?FXTf=wWVR#`!bGDK4tX%ChV8e zhK7aO=5nQ32HvJ@Tw@(DW^{MIsaTT>I(N;Q-ySN4N?%!XRjq*BfK)<-J-E!=wZwYH z?vEaJ<#C1jI=pqa`)fz3qc(8;oSM=CI<{M1J2$F)H))fv4^CLc--=mTHq8%<7BwK% z6~B?@QzBA5-QyM=J~($+-rwz03?h5>!p^~K=YZ9cUW(d;-PSGkI}#1@tE(VV zMPZ-RF&ft_M^F*c$4W2uYlpdtJ8fS$MEL|{ZqR105PIRf_NrR&2GoD7px;i|p7fDR zpeXz&ClJ6r>S}6A{QZJpHe@7tt$MpVe;b-~2U}NB{X@{iS>{5x{n{k#L~h+mi`=jJ z(~EUO1<^FRnGf<+*yG3V<7{+0pDu06eKd>@8ho|~yQ-|(?z6%|dsMX)%mBHyb@p=2 zOL7v;AN8A2D2MA#ALR%VoOFaB)bO4bV0p7(jG4k(0qM7|oRW{_!^i6h=hcz(|JJxg> zE-XZxlH<}f1Hbs$`6?9jLPJwmlFzBD?Fha1cbAKqdT2)L10 zAh^^ZpPVQA``{8~sfaKk=!Lwg4h~EG4fsgovq6P_IKVv=_uOueZ*759nq8FIfbFV- zGp(Hl`Rxto$>^}H9CrPH2hXFT>r>gsWd0f3i~MC1LAgrs+q%U}u`EKG=j0m#erD64 zyy<9b1FdnqBr0#r^Ot}k%XPrzi1}~t^!T@+UGWp5${FCvh^|A)s3Gp>QKNSlj1|hH*v>6+Ma#s&!h5FP%k{J5?1Y$M2__L@YfI?NT- zFW?mAkgthQu8B5RKT7iUN}Bq`tRM&NaOU{6_ItRHozJ+l>;p!A71Wk51`t)PMyXe= z{E&Ecsu4%d)a{>*ZtGhEY92Y>T-aA{%v7ws+GA-*kM=`61K=C&HboYH2znH7G6;GY zWpNDTE=sJu3QVdiR5Ng2K5od%rEj>IZs-L+1jIuqqkT{r8|h(`Q8Z&p^JvC4-E1sc z=4JfXiwy0n41@e85ONL@R6<&vKpn;He6&*1K>inEG=MhpZ_>?V(qa91omF$c>c6MZ zK_LRWnMkp4CQz_Vr<-0T>pr8`rx)*B?Pk~VoRwubY9gR5lU7#h>DxVXu1$FwBIc2} z3i~FdJJ&pn#>el4^-9+FjoWOs(u((*uWl~P?scfX;;L-EbO7Vcn3Cbl0Ou!8QE7mS8wcmSWNVl~x?9+fbFkeqNYoo7aZa-t>()3DQSWjsS&)A8!$ z`R*zHfR^XOuyxirLOp7pE^|X-hPb5>JsF#=pkW|qN=mP*3pFdOi-MfiYgedP=(aEI zfb0w^naNw$MusNg$y=B&ZqFO~rMPAl(DbG%LQ`p53m-Qvvd+%V5_4O7fl$Y|%gakq zMYeDf-Zo%33j7*AAF}ylKk)Z;9Je2|Pk`d=K_UfUM0!|_1_%4I8qwmw*)}pr4HsFs zwmMAam3L{;`BmZO^=4@>)l@EDZ5OVCH|=9btRRFtKz8d96F7PbZoj$>BOu2D-wgQ8 zTy6s%%oBbis=56Hk3xy&hM|($R|Emx>vZ5;HIK_(laBZW0KHBg({XrvqV?a8EjoTF zIjZbB+v&(=ER9Qm4*!sbiPUvS-)V$e?}wSFrJpPXqDvV$e$&o&( zi0aU|Y*+l_NI|AUZW^1L3nm;W6`z4=%h(QQOpe^|%V)Kpv1k$~czK$3aspYQA%_F@ z&Q3eN$?uba^M?k#R(ABy`ZP?s71W;CUIJ-z@ez5C+o$gA zJenD2WKVf9#Tlj$T4Z1NeEB9@6*hRbG+RkLDN~W(*z>>*D8S9`djnpFP z7JPp7x_pHk1NR*pWBb%cdCu6TSx)~yHvHiV4go5*(UT_xbY@Y-izf9B?Fnu-X7kTA z@WHKDumE}ERc~g-Md4acoOFYh`5qV&HV=keuS5ZVa=qcE^6njb3}XPvV~b9U>9+lI zJAkS~$hysHr1H@4a5GSW^gt&HK=kXgUDl8U4jzV)@V*1}Hq^a=q6km>q6r^>0E&ts zNC#od$=k23Jg=vD+%8xFzMH2+tiE`T9fVL+rMtMZWBxzrdJCW`!>(=HK&3&EkQ6~c zTDnC_X_4;k?gml15s6KQbhm(jG{~mA-5}j;y8N%3=Xt;Hop0u!*~~b@j5B)Q`?}Uz z=W(9vz!jY`a=g$wcXXrjr)q(cMdwDVo^`cigxHI#fxRniTR{D~t?@7RE`sZy`7#y9 zXa9@9<@Dfc91!*zkVPGrO|g45)r@7bBb8n(v-4-Lq?S^(U11X&&DxMj#0tw7kFB~F zgMyPwDu0uSR%m>Ey$ineQGsQL*WY?OzJ=`+=$#4Uhz$KQvmU5<@n+}?d5nc@jn(kd zD${v`OWEKa8lcmwYr*G(>+$RD4I-H1{lxpCJdS8GeFjS*>NIk4%SG&F>wU@tm>g%5 z>Gf*4M7rv>0Zfgr3?6OQxwIJ=7ZxtrT6Hc?p?`t}$HGXvq&mvi6nS7zr>A%vP31Cz zs8BRLQJqc@QMDO#Pl~6=EaM*P@_E+@O%@3bY9Qd9qu+?pO&j7RJshIP&hEepUTgKi zRolRR0@gMx|3%w=q6VP+J$MtQRah zzD%-;D5M*Ekd%+XaW^5hX0&59C?xM4Y*6m?o8*MmO2_c|p$d#4twd0tbYrO)9 z26H62V1w074)VHW@q#JMWAhbI4N(I`rfT~U?LmH>kDebNANLh{+ynZB_36$%g3p=K zoLmzsaUsIbe?jNkix|{B-+z}0*H^ASYp%D~ssY3Lc=1Vt>*4jn#;Jxz{A5~N>OOR= zCFJ7Q%CDHK>-=fRC>izhtSadVBfMYuhji|5b$>gp%mB6;9M0!f=x2JS@cE(C^G{_} zslvA0H_gn|^!ap5zm3Mla`S1Rmy5%&nxkF0yWD`izIGM7S_dFy)ZnleB{r&afBczP z!C)C=R@mhzKkMa~*7UD5$9VwS^re|ExuJ}dR{k`_w>1r={A zD@Jg0AB2V~ObY)vXT33-O+(d>r=O#(MEc^?lSo^^FDT(!{D%*--9^*8M@G3A@G)6s zV+Vb2kX{?ox=TQ$Gh(MnE6Q*K7vVq+RoPaK&{SYsTTOHvBcH0G0AW>yd_wm?P~MeC z^uW44Uv`kaK(d=fIH;){H_(b@lDg&7Ru4c|I2`5kEiw{2?Q(AI5THDMTV*XiqI)&B z)Ii1K(!3?0`SMo~>c-+4tU++FcPVby6$Jas1D(Y9&3(7=BArf9xxZzu|NG=QVaAaF zPPaWk)fsBUg-H+z^b;{TF*Vim_wU~7ucNbpNb;q8N#L{i$cf8Ygb$X0AA-My~IRsR`Rs zXiOs6EM*N1nf+N0P%ZR{!%c|8OT5Sl<>sCatR&_35;pHYxB-y~aS(kc!xxrN=5L;_ zPb@-KJ$9@>VWq2thDUI+A;?0(d8*eL|ML8g33CcecNCxAjQMhW{>`V!r>U!UIQONAn>yTUsbDu@lGBcdg&Ik1>TZ7c|-GIVR2)l zr}}e8D=Vui+x}3MBrrKs?5}3j?S-z<(w3VGMN&(uTsz~>$r4ZrNBJla(0#GAD7o8BXGW^!ss5d3eJisaE8DCrYFrUVRU@b?J4)* zvz~}Bvp=t?EPE<=NA#CGxa8DD5;4#z-%dJ1O2Hsu%b7Sx&u_&SO_W9GWMxJgQ^MoE zEud~GJbAD6I;Gw&GH-vXq?&8-Rbdt`8V4TQUvzq-;-?jValyRk3GCIA1I#y9H}Yr0 zIKFWH#dA4u7qv<296nh%%H_7-|MOPQJ>O}uSTrF2qeC;aULhX~ID`x+7*NPtkg0oi zn_~q_)gr=&NtUIEY>I0-aNl%FB`~NLkiF+i)j_&10xoeM?0Pr^N5o`eZb)GvSpEk^ z=dEW3L30Tefqb=BxD847A7StBx8txRalL;17DQt`ommk73rIAkWEGyb3PnB#guXFA zX)#u2N>Gv0mJ}7e54Jt4J*)Wxs>NCINvS=~?W@bCYLVQ0A27wvdmkWEVV#upWXanF(l%U z>GK9jkkdeoP!2l^^w=!@feAjIK55&dYrWXm`towiQ(P5(9Ha87ov3DF4Qt0!*O#P%LeN#Pa{(@>p{5DmBZ@hw=`(=3=C~#H7GET{q@FiruP3cmuse!S8&z|c))7Sn5MU+`#B8i&5 zLIDBM#*`@`bq5oxI?dNo&K>NOdj9L7_n|bJ&eOt{W zt+}TNc_%TH{xIl6f?2h;iaXvPpr>i3nQ{(sM1bIo6v)34mMbqO_qEqZ1#=;u&*O;h zbis2k%REm$ec|MXN^-Yc^I~Y$(^=gwpyLZyL7P_URs&`PPl}sB-#KM+$Iqa=XKoUAURq@Yem-k(DCxN`NJf{WGJO}L z0=r(Rn$ts@?1G*7sN*Uen6)7zFG&?Ruf2CNQ4NjTR z^=A~-7M8#g9*j{dCtOaAK7zKCDnp=ad{3n&Cr{cGRm9@p78Wi4vf7a5g<5sBMjs!@ zPN97^moEZWa6OQnnrS7SMQV~CQdjq!1r9W`7D5C z3Fi!IbS8B?nG$q;+;f7MJDMwJXB{bP+mD}WmGRX&u2lc8s-@WO$#+~g1~1!Lt}@^2 zt@@K0OFyj-woC!SCL1CCHS|XJK+FaZSD-Bw=e_biV7Wr)&{_x{Ezs;T>g(OrXAk>% zT(+k|M=5;H5Y`XaqA1=p7pxV%+}YV-kOOO4ot&LZ&Zi8!BbhLa2Hr8 zK~Y{k#pAbu5uC=5K7+(f0`;Xq#yb)qR3P@Yx+Z6|I0kcDY`#b>VVv)&85IJa^QP#i z*3_Pd{dBQ&oL#yd7Uuh=TZ;C51k?*@SxKUf29-6#^rUf`-?!WR%pk6=>Jv6!QLH#! z5lMWV+dr%y<;IfE7VcMMtZ+@VE$eO$2(N#0) zLgU#6u9La1$J&Sr<*}KcLPH|n84r>u4PS=53<}Qc0cmpQ z9!5BF#eKo>gpTgjuaPfYt#aF%r`>r2I%KY6 zuNtG8KL`!{j~J!(tHeU!1!7M285BIVVu-FExg$ zJwS))lZuxVpEQc`K`cVAZv2Mb$rx1NyS zW74K8RQkB6Uh`}Ke(icavFhTG1}H{0Jki9e3ecN)p(=8Eosnob?DVvQN2Xk%T!}lo z?7@u204%N&b7^gdl|}xn$qM5;x-M`q01MVslom<6>(nXVu_(ncB-^p5fyudRLge_e z{&xbSg>^GfI({>hq``1Z>6M=hpr8`zH)Orr2qo`yU^GCs!+{=VL|ehGPA+&u6eZ;F z^&isr1_uD6S73$d3+_cPc1Yb~5On(Q)Bb$>v>!fvNJK(%-vNO;Tg&UTj!UTHrf)yv zp@MFd`>}wNLNRgAI6+$l(}+Iwb73K{B`+n`Du(|1)B1%}z%9DUbdWjeW`57dwGXMa zbMPNI+y(%44}l2Vms?OM#X+onTERVM*B&|wT}Ou(lPx~d%e_WyU?1cgb*^jRhTd&p z8mkjk_$H1_0OW*zrg6Rh=Z+Jzb9i*!?@A?&1HGo4lJaN$sj;eIJ3|b{PIG?|c;;!u zJ#2Wc`y6CW_dip96TaiC`p9p==3bL_zEF0SAwFjh^1C%sC>vt!_(ayTsD4Z{PCI@v zdic;0Av0=ZGiP99Mx^}3&NFcpBb(!B&wWPNC{Bm0lY2-D;rwGk#~2k$h|YECkOUf`2C?@XPq zlpW=(VLv#D*Ftrqp=e9ddhI`L!TR602P#x9uBN;dH-g?qoX>ArJ=lY3NNp=YV&NeHL@Nmw-U(=6o zR^|vnHQFM~G4pF&RTsn}BZ(mw%OB)9J^0q2q*$nPLv4qFHB>XCgM>4%@{}pb1weEj z9v+|iAF*NF{VfszKfV_S3dpb-qoTslt%km?YGoGB>Ao{C6>K(}QIr?#JXr3o8$0v> zQ)!Cp^>=-_E$iYeh3BytP*fp>kK6Q7b8|CRQ*Cpz5Yz_Ws;8qZpm2dAwC=S}U3Dfv zV@8HRPu-it1X~ZCs4gdvW}-LsBF^Q&iDLb3Fp-(F^O+~#!Kw4|o|)+M7s^+Zg}EI) zqC?)OleN7vQPB|*hO~HKZ05rD*5WT?q0Jv#3)-Ot>HKV*{yIl)wb0KIDucxW?5`Dz z0T63)K#DP039)#DzZRz?G60K5^w8n498TTTf>yEq%kFSuT?m#z`YrG%sX9Mkm)~ntZg;HosgZK-IfidZh|L$`W_lmFoU%pa(N2MHAKzZ z;H(EU`Ny+s5oJDch4=^H&t~&8BO{eoX#nQzT%3+fwrMa|e~$MjHvNhk#3A{%Y@B-o z9lK4Pa>sd8mfPCgYaGINH7bnHYbJFX5${ffuXnB$>gtt?((muK7hsv(^zw>68{BK3oRTUp3Ho44;bE7q#8sy_PN!Bsx7EP2ijeq9oV2`SZl~Ew{qM3mBvo;3`9~Ws`@oJ0ufukDZ zH`qV1-I$40N-B~3gBJ}XENw0_1^`y+eY{#3YZObgjuWajb#)t1fI?nPD{l_)jj%OqboaBT@4t>-4Y+aJ6ZLy?=S*|4Fi1; z`0z@TEnfp#C6J;e9thym3@X-Fb8AA~P-6<{n{tGE|G>Z@Bp&zW*45M;^tD{|&5~J& z8&|1A16ixibl^ZQo7wmB0n^@8@XMBt%3|@+e-KVH*|h6yHq*6d1Zeuk^Fc65H6GOk ziBd!3<28tLt2h8a34%yb9J?Hm+^H;)2QmS3wLSf+om!HJzwVU5_3`tlMskBwq`n-T zu&QUdCZt;8_vatoJP#cN05+DdF1KH^O*MY)tcX<~diq`4m0kmikC!1P0qvQeOzVq#bpzy3kh$ z_)m{WX4E{IZ#iah!GZ5S8U$OTeO%3&$rlMFIH}*^W@HLF0u;LI!;T(ZngY zv>;uK5u@tIpj#ksli^331L(<@;IgAAdTDgmgMD0FDFu z1Rz0|p&2Gdt)T1pe!yFUvAYKEI_`J2D;Tmor3bl1tHO1w%+>2r%Rtp&oGW(tie7UHFr z=k%u11UcaT;4WotIb=#M-E#2hS&B)|Hw16OmuYQ4zKPuW6gJBe@jea74%j=OEy1 zz5xBjh?=!U@6O=YRIo%`C;ZkPX98nSV?O={3)HID6m4TryGfN+=D)F&MgK5~_*WkB z*ttgV!v-_d-Bw(akO-9_mZk&3!?mKGMY^9lmPju;$8L>eTYb%3di3EB{NQt^V4JjJ z4EJD+NYXKIBZT+nI(ZTC=H`0#YGLo1JVD3f71SfvA5Ul!Ix9iJO#Zl8D&s@SSd<-d zbk%1}HPsCWk6J}aM?ei%R8%aTLgP$&P+lYxw`)Mtt%EwSR1LD;hvoW$kc*Z733_YT zMQexf7w+_PxM{rX-tO)yNSo7~?rKngW ztFI?fybA^MzC2E*nA1_$!>f5Pz4Cmmk@VLfaY|h3rs`!!*Q*A(jFtppUzW6Gy~yww zL2h9eHx%wT8bQ+-552F5%WhPz2WIOf**ld49NAJV6E=Er4C#+4{qN|0zY9LkC%e2H zP=Q$|?!RUo*qc)xjpyGmOlwbZ%{D|7ikz-nO0R29P6corO+JIBWCX9B2WC|IB-a>G z57wWCx0ED&u*4}uFtoGNC!?A&`*WN|XnhS!_3HO@L|f$V>*)e@hgd`{TOcc5M}LSG z>ZqfFGMVf#ApjHnG0JP+GJV%Exd_TBg7e>$ zGH~iBkN_iAhtT9uG(XyIgN;3{kBbX|p6K{%ef#HCc0(;X&qt90PsR1k%z{1lnp}qs zvxn)xP)gVxn~_6y87fqO1G**HdN#}lfN|cpF_4gy1H$6$rVZaoNC3C)LMwC=74z)< z0~r_;AQ@-YaE@^hc?L!G0JX&$jJ74TUoG19gz8XWV^+BVJSCc9w&#KQHiD_(>|g5U`R6_?n0 zGn_8^{;N#qKUN$y6kXaQYb-`*G5J$^|olv`2*v*E>TZ&mBEg z(Qj53P24tr1}vRs6X|~ojvOAA%itz#8{2a6^`!17EIe*YJo;EB6`G6K1H|TR#F{(s z8yF3wQ6Z1q0X3e=DoLJhaN(5Fr{h{Z(ypy8`7&8Nme89PBLi3D+sGIF7vAqOaY)&v zffdZrtYLFz5Dfqv0|J0m8lK*H9M9cGw~G}f89;#W|V`^^h_5~LM;~y3cD)*pwboZ zw`#7kG`n1R7Cj?F4ZqkAFY;j#pEpc6U^>h4et9Tp>eZPMd^}6T4j~=({NlAd0y~;gqtH&FqwB@^8+qYTD8ax>4j`X4SP zlwI}Q?XUosE`H*6di59;`8msf_JRrD1iQcSO3fuPqUrOwrBxYH!k;;Q{hs>kIK=>U z^uw_i|Dhc5RVqJe)?&oa!HHX^fQft84~~@jp$4zGTKq7LWQI;hy>W+|_sU9Hnb_cS z=j*h(K`*7rD6k^%#2VV~UUan}`x~3ZS)cwn>|PA|e6~hd*t14H5gS#whXbU3cKB z0xhz zAd;~c3bX=$4PN>g3ipEBa(s6cJ{aDC7O0Z3{(jx>IFNWZr==siO+{iRu2VP^_;cu{ z;G4=>zNcOz^92`SZrj1<#(styL`I+mU49HI54Ot-zhoIihpD9C4Yh@KqOdMbtLP)J z5h5Qqq$arEar7B*($(Xxo}uVWz=K$s_65e?lWBMhx>!9_E?Y2`nW)doGCIKX3HpGp)Ja)jmd-*#@2}QQ5tdc}+5>)8LP0gvjm$msw5{kg(q84Z6ynz2+yg6; zNv>Apkm%#j)`BQWjr32UF>9lPT0xDwIgg`#PCv1#neNY6dCKlP_*J-Vq*OMtj2QYC zW6u?I>qQgMOyFo=Pqe^%HvvbVYi%axl7F~5wK!1I(q)%DOSs~m2?91@W1Yrts-m@q zPu|!IwgA@40iF|3o$CZ5lr#f&8@a6x6opEEa0G;PA9B5Nw)kBdJ9h=5GdAh-S2M`5GDIRgxNS=rf!qP<8F zyArqZL6NYvoG2zZOPtgJNEt~?zd6bSJDQhyz#dk8A_0Jp-gltvcTnzOJcCv(f$JGW z>YjaJs>z(-h zBC%$73RQVZn&$66mhH1vJq9WtbHkm<%3e%t!$?`Z!IvMt0*RKVpj$3eVw0c^)*s<% zll-z!MwUrugsp*i$`J|LeLZY!Ls>b5I#-%CB8echFoSc<(CNs~sXBX;Z(@pPKqf^4 z^_p38vTun-EFv1mvgV_5oSrl-ut|ZX4@qG$FydCJjsxJ4k_m-8Qu3tT`L@&$?U@Zc zQZt1nTz>nz1CGba}jvJ>kf#~n_O(A5*21nMr-S4o|O8JhDeR; zOQ!tegT0^Po_5MtF4lDZ=G8;%mg1^?c%d>gXiKXo$S{$_hk?W|tT-NHxk#cqYLjoJ zh0}il@-WLyF|lUy*4>#cmW-yGd%!o%#W^73;6DxLQQ-n+DRt+&2#&TLY)qtp=aX5t(0PQYltRNZ%N;Wl13bdZhFAQCCKRP`{YPK9}3ZKu! zu^dA=w?9Fg-l{yXHl2e99=c^ye2fUwIeGzd2yX&`=CjlDgKHTK3$R`1oe38~IPw{g zkScTR%v+)B5!!sb0QBgmno|j3gKNtmH~1+eLb*>iE_6bP37OsJ*FHA&D+JKK^P*Jw%hq3PtGxwceN_(a#a43AGtGAHF~|qNcgWv>xUwIe z&k6$N)<3+-6ukz!wM_I-90e!_WOD2b*uVSSpn$XF=Eg=O2wRWCy6K`K|6U(Pg9@Sg zai}ury~)C2q59m6$do3{@|}&wkN5zC_4Q#oz#1RL=dN_$4gEM{UrqmtWW!0LbD2;i zSOfGkH)a^%bH~IW9zMo8VTH*b+t!=gGa$q2x{BgKCz4UZPdE@>n?)aIk@>-s!FRA$ zQWkrKK|gXyV;a=jm3MFi=Y;c@fvbCfjNi4+lM8P^3)OrgT)QB;mmIA`rA!Nzy|tS6 zcQ~iyg!GOZD!Vc5;ZBdKy_9I8RB(COD^)c1p#Ww|xU+lXKCf19VM)I#KW|NOP5sA| zcbrze9FRRhT=546uqVLmA??7V3;cmb!u0Gyb#IA8wdynlK>DH#+;g}6GS99U{*z6a2; z0JT|jfINdbiG0f;gHUF6+pj0%I^{mf?8n$Tk3;LpdT@I z*S2&Y4%6v?-d?D12Th@^-RZomBdE{7ftqXn1UqO0be@%c%*%l!Hh;iW;(JMnb$53S zLCa#x`sI$lPUkLIANgYe1jQqYoTcjoO4pXGpTB8QVK|H!K~JW5VC89vR1?Ubd$WRO zRO%(1ec|PgP4V36Ojsy|H=I`}L@?o(WKMaW;deQK$cYc>MJevBcf-5ZPaaZ>#iXDT z^tq=6aq_)&(fhPLL?`z`0aM?jHe&NMyN+vlYgfe9YHD4^)n#b5B6>yb*i>1-1E3+2 z&F}AnPuiM6UaS#~1nXLEMXVzajCysB1Gz^24#OX{+6m8MuXS-GpwOa`yN6i)yfIST z!;4iF&C%oOBpZ@a2Jh)hz(8;RS{!SBB$8O9P7EZk#^f0bT`Y!Z ze5*Ev{f>N6N7D+2k4 z5!!E+)H(d^i2?5=D&+|Wp3t3Yohi0| zBX>~1-U&!b0YBOsnZ-paU1*G!^33?az(5H?yz$9{r^s>UUefwc{{HiHCOTp3xrlPy zRSr)XqaQo~9X1QRR#P0Mu`>Mu>^M*Oyl}Yv)ppo*6AI3;p;(T_ zC%Q+rB;uGkac4Szrn#mwAA=$&kqStePNih^<_KPF3%cOpdB2uC5+>4_vv3STY*(|n z6-BafLx0@l=M*!7qB#EfTaCENyrGsRlCoH_sjv>Ym*u09#T|^uFY;MnZErF^n#&Qc z)t2~ZxD-y40cjsK|7#fwt4~%|n3I+JR_v3JS_-V2S^kYIzX+{o+d*6Rt(IPw^tYB4 ze-%fXy7y%z2ADN**IG2|<`TTzQO^Y=gIcW7PovlV8y?s`oWfs*Wn#-ptSI5tsYYF- zrVDB92gN|qNCk=pE62YJ_M%To8drs8_zG6iIwLb)<90NR$-ApzZc2U9hg)y?h_Oe9 z?TEUoLh`y`lGyACX}^Bwx-YGnx-A{hLe@ArPfrq%zTyaCcNcyE6uca#&NeGv@)MBn zJ6t3m^vy%AUX%na2Q(%oCXoL^=(^tB)pZ6IJh0dRlZ-Qku7+fJ`4b9q@>Kw(Lk$+# z-1#C#?Wq}qpxwSe6qpu7L-Al=@9QaW#;f>n%2tU`JxS_n|$u_ARTYw;^?BQ^F zHbg9(KH(dSTELxN5blgDY7kv*bKWc@Nu0v6p%Hb_3*P?43hEUbE>(p-b|!&A(XcVZ z)^6M6XvpU5X29}h%X$VKiTb;~ATfX|eAz@QPC5rSl#2eSl{NA3m$ACX&lwaer=Rx( zYN3ds7V5M0Z42eq$N%e_Y@F3^Qg%@bujnzbmkWlb#X$J$$O_{rKGl<)(t5n?O!0wX z{J2_%pkuICQIUt8f_Q%TPSEzEmFh^E@eHY9mlmpi0}szjO1OkBiG;dnA|8m3m`V%C zB{nOJEJZ;*PclPw!o_k@wMKOmA3u!IY+c%SIi7a1bo}Kw-&xWhx9RSklr>7ao69>F zvE@!(_1ffXd&lVV_Y(E4z`XN17T*Z|o1LOD@iuS`UEPBM0^FBL7T<%IhnygmThhs} z`QP=Pst4*N1TNqLH;K;S*Mxl#uiotT*~Z5w4H`rO=^1OQmg#hbrI-MF3||~9r2>FM z#rmmAQqY<^sgPG9hy}519O2Upj?kE?)S@?*i1`@hNC50-{%stL zGgP1XAYpnx6rXLrr2jnz3Mz%5*x5Q|ic=u`7w&2YU_)S1`m%B{UVo`Q3^@n~er8Yz zzM#$SypTdBj@zS2Y;$<}kR#`wPq*q9hxBSOIP|q~GCfs(TU6G~GZi&F9a)%wsBYKT z`Bl^@L2zS1-hz9XYW3@cQF#)r=Jc_}Uc0k)WraOk1whu+Q)He%)>)w*xZx?n7V2xD z@>KALtIbCTIA%eGHs_?F?6cmQ`uOs z;guvms3j^v*PHjyjL{L5-7qg0woJXF&NwKd=&UsUJOeyUVslidCsu%aj5O2t{E5cBY&^HIx;CQIbMXk~Z#Fx`a;MN|VU{AdNSO z`K!O(Z&tMTJP3g|?o};U3oxufVl;T7!Id7M4%OYD+~xgJ4+IRnzDZ zVpbBA(HlHp-SVY$ZXA?_g6p4!X6>qR!sGHScEzdJ7pf z{aKkX#3zTdu}~WLhBRF>vQ(G=P|NP6OGZfDx$rV|%YTYUSs8eKo>*;kIBOTz6Kw#} ztm8J@Ofm*WO6l`|NHbudrvjweMeg+U(}D^BH$Uv{qCE%&_ptcr>&+}M;$VFdkzc5? zftV=o2jsj+Qe*dhThmi90-Y+W@QGLq{YW-L*=Xlj9x!5)74zOX`%TowSWQ(Y;AoFJ z8a#r#{k?TDKprQ4<-EfJJ$_;zry+tX?0m^GrB2V^aP$F@MgUDyM2jZNLuFTSV@`#ygNbb(Feq{Ohan>z9>E zkAEiNyqPgZH6jq@n+<ninWD4S+jFgn^=e z{MqkvKnrHCNk-vBN5~_1U&gU=QW3IQ?{w)d`fsgh+*Oj7F=0{rzeSeo$@OAf`S>LPUy%sxGA-=Xg8#}CiG#J3@xYI)z#7cz{VOH@Z*PqpOF-?6j zo?uBWyT<)s?Yx^#nP6Z{DRs4hJZTF=2S%N5JAojyt|UV@z0ba6SKQ54WxA9fb-STB zg$VA29;i$gq2#5dx%K`{fax?PN;F3LH55$`MNN3>6FRs^<9my#P;M<~tbv*dp#b-ztOOd}hK$fBbHwy*$u#X(R)--- zx=>?3E^-{SL^OFuwEbsObK7(L-?IXjLu_K!TJ#j!N;WV)#O^kAZ( zThIr*Pl2qbHo?V72`x*eEe|8jxuoo9e@7iN8+Fv|qY-<#-) zuN~n2RLq8(=E0IBkze+%6De-~yt`d|138gd+LJvpfp7oVDf^;tRLH*1LZ4UDJY2b<1Za;HY=XEiMqK?- z-`Vvu78K}Lvd7*A3WPv`0{a|EET3BHWA!|70NkrIuO?JQHsz$=TMy7Z@G$Gm1!kE8 zS(V#^$}$5$hGDav=y|%9)mvyH*iz$u-Ue$Nmhoy#eSQ-9jeXF`g({?>X`I!YCy?X=(g|vUMgMIbvtLdt+6mM(LQNt_)hhPmgmMJp4=7A6renNFz0- z;cU|~xyO*YGQ1phuuesL4niS8#fk&?I7dkFhI$yLl77&8gRybU?PCM|>zv0{NwN=! zmxopmf}Lj|lcrTK8e|^vkkWuYS5@^L-BXd#iBjU{Vw?)f_$`aNm+{5iDEkbG>>@Jd zvd^OrYwXxndmr=|znPhj0h%x!)%aRo=@zcD_N`Up2o*Mk;$ndfmX^r9gsd8zn2=Rt z*5lTy@!K=|E9V<&hiy{uEi3Mj0o}z6!G8lHXTCg+x8WGKHMd08w1M`@dUr>H+VS7- z0M>303;A6YRiA5}iQ{~BzhFh3BGD#L&E>N{S+`47erROj=GigdXNmh*H*#66%vTq? z%9Ulr<3EE5-z<#_^f+5BaVVJ|!$o+i;=u&2@U8UssQC93rE4)gGOuGvO}GT*x};d` zT>J;qFP}h0Q**QYH4qd@14iOhU=vh}#PjFeOkq2kB_x^9Lo3w9*HV8hqLiG&e@aqd z0u9PM@&c*EI=?++lN~MIH|vmCZ(wHXbN#SOYTg?%!l`5lCI%msh3|GVzkBX@96ez2 z_a3ZBiD*jPL)5<^FzgO6$h^nFG?X9w0|ue(+0?t9sM$>sfBbarJ@R;&Qm^}sIk}R! zluyMfaJ&93JX3#LjkyP*PEcx`X~ySacwfTwk%y)qG|QU;2p;oh-t^ z*QF0nRXcYTfdk%?BJ3%F2#;Lm#{+GklMbcGE$;&2&w zG_jz(hXF3b<8H@dFVxWiunqz!0-2=UGW z9u)~neWoJ8aXhea#95zq_S68lj$aQj19&d9W(%{EzgMs)UKnV z+Jb(oa&1jP2k(+pQo#YP_L5S@(+7L9SGmFkLeicOjlC@|AryRW9ekPwzR+_mVJ*mC ziUcBZTRYxQ@5A`>l57)SC4__MD*gsWsEzN6^k=Pt8+A*9ES18nY??xpqR8)I%Cpw#}aHI*LQVL-THr=OYO9`3XRflF3&pxG^oOA`Q+P zx357_Z+-^}9Z`CCdb)#hAIcEu0N9`=Q2%e60PW{cl&89qEbJY5}wT z4hS*wSUg0h6`!;f#}2^D;>Y_{W>hTQA4IXk=Ip3VI z=A1k-$?A82$Al=A?98KKXv?A?PTz+_u^&TSKUldTl0c+5PlI_UOym3iOnTALSFad> z5)Q8?a%pzReU@%;bM@-zgTJsQcjk$XSA_K)HOJSW;(ITT2{g?gwE?DtB8hK7DFM#7 zq3c%kmX8&l1QrI&SjGe-$q4{0amrJ(;Vx9&?e|Ux&w`#G8uiY|{_((;69U#DA>tzw zVo=rBI0UlNzMosTf&Q$jA9pnEqnZ?dWNDSXP(A-D4hFzjrtvqV zg;h26CPTyM~b3T`EMyH5YAm136ZTHNQsrh+2=9+Mp_RP9{%MS)q6FNYC=2}n>uysX5E}K+gP|8e-7ZtUygBj#(fJUXQqG#bXi|WoYo~MTW zL6IN6{rK0xeVq>*Ee42lZ)dEtO=3I2AQ`XF?J+tz3v&Md*pX$~eRu6Sd{LF~X;Xoc zC@awY7Mgy#;F)c`J&mK?ercipU&XJ~VO2yL8G!C|&b!^#da1+nCfa_q1Fo=iLT&Eq zqmEjoW1Cx0OequlgLfdM<|8V*h_{gcGu9Wz?m`N$Dt&9yc7U7E%vJ|WNj#HsOhS~- zby@s(3W(c~5#W><34fd{vrn;j!Mu1AwYLw7=SwZHXYS{=G0QSAa^{qjl>8UirE39D z+#xXPv2k*WGZKI+o|u@J2XOiz*f_hu{MhD2jZCQrI%WJT!J%V;GRAS1FjwonfGS6W zMY@vcSAOsrka|G==SLj5aKQ^Q3Uy)wDoHeNIu1nE-}gqj^!R{jJ%1ib|KifMaX>P+CynGeKTkO4>v~*} zdGX2utpwo@a`zTJ$>{+vvK&!hofKK z!Xj}SeoNLTJ^F>zI?^NatJM)H5zKGaX!aA?u3oC(Jzu_fiI6$8{)GQ@|B&|G&-p3c zvVOzQ;jspM{8I3w=PT59>Spgs8H6^EX_BtMm*G$;zeX$enIq^T;2HG9Nfw|bza|s= zGLHf>a7`JXIv4j3_Sk4n%m}okAsvi465eERq)0l~a^akBqMdJU*!(1#eI4@hU7}<} zG@zAalLXxUQ3;&=vKKmQ#Ti*#9K_JszjtfE+@7kC!qG;h4da0#BSfjD zMgfb_RmG^)Co;@&WOiRdl@=5N>Nfd>#&K%WyQBR%U=$byk8*VA&uTqMZ6X(4Zqmbyrhfqv<*&d4(ng60tQ1&mRvPDD?#$N{ zf2U9+WsAFLWr;LktqN^}HLPZr+}yD~eq<+@P=AEN-W#K3>FAjDzr#f9>tVNHq5vtT zzYP;TZR?LW==<$F53PZoG6#l~s}tLQT+qU_PsbSd-P5{(X~7+OU_~t#3^$YiE_82K zf!bg?clRzN<7Uy_Mi}*gkmf3Nuu~f3f>OPUv|As_l|Q;WjpCpZ-9RU>8CDBOR_ z$RhoNTbI+i(UE@jiDj|Ks+5v)=j6oavXTb?Fs3mCd3qHXaqMery(xlGo2NkEp>`{e z61Gk7XYXtVKJJM#>S^Q`D0Oe`N#~th(`h;%3J*uT)?K;qTtV`ssqMgm7a?#dp?X~R z$}?-fpXLU2j0^Dv4Ut)e8Fcv#3Yo7%n!ejLOTFz~`*t|ac2HjNjD7I6K9)hWWz+n+ zCDo>-xMHn)xH47IaCj_uTmvhjiP)v4*$hH2G^iAG8GJZ{3kzq3{+Tw?fD|l870l88 zjBkOgHLS;eS})r~;9{aN^i@zk^d``T1 z=Q%9(WjwI~7j!PNa|-ow?_J?EANTvx;vPAkuOuMB`sUA&1PKy6(W7)AptNQkVo!jn zREO&tD2Fduybu_5g8u*sm;i0K0pVIsieye0+~eS}Ad_3plfA$QX}K!>+UlUe_ZF+U z*KH3>t93tDn4pDex+lme?`=aU=LJN$tD+j-{CV`|547$tO6YDKF5SYIVbKS1N%|1!TTp0cpqPSt0R6p}_-qggb^R%{x zQ6AY$~DVmMBa)il7Wwx6B@% z43s}Eg~#)Zf*jE;?(T8mR6x9FPOyAjtpysgtMeg?t!;u6Zrl^@p;@J{9_Pxo|23OB7@R~ifCb<4 zTH}|beiD{jRQP-EW_QoV^9JHDfdczHPU>!0ow}uRuU3Qnsyn$Z2;Y#*`8a3aO~%=@ z1C{jmWJllA#NI}NqM`s%$BL_ub5g4vV89Nau#_J$`tyT)#V7ztA-ORzF^kVH3kwgy z7dQq&5N@Nb!9y2y*ZGEP3?&@vNus0XF!ofL-T|nCR z@%6jv{`2*jg@4Ily;e&$mhU@FaHR^x6{S2&ZT&xX3)#W@xX=Pv9n8y#@+xZI_B*PQ>juHSWFRDK94nACQUJEjD}{!^RoHDJt)mj6zK-<_4&Em!35 z_VtwcSRVX~TkqD3^^s^n-`MiM1`CgxT*JSl3PJq@%HJSPa5(>LV0hN>*X!F0Upzgc z5+~N+F4I~o{jRycK6KlQ7)K=sm3EX|3Xf?3TLh)=4hX?G)N(p1DIn9J zL3&sK4C+ML#C^rVrRWWqs4kdbJ9JVy)eN~L*A!IhJT7~L*4}8RTLYk%1(N%^7cnh6 zkRGLYczsLY9q!pfargVDv+TOd{W=6r*RdD-zc}8VbbBg-D0Qz}kucxB3z*DyE@yl5 z_!pGoIve&S?yZ&)O$Fs{XXmjeNW*^m5EvLpAHoVPPk`+8rTMs(vUI@M$Uk;CtJmji zjhBdxDF+gXNUtGX(#L?sY-V9m!n4w3E0T+$q34QSgOQ;dS!9{P>l^}`VFHZ@fgaS6 zu=th`bhc9oZ7}QsfHuQI017)hvv+|c?OoXbZI3)ikZ3GD&wm?BB zm;mYR-aeH>4Jee`jwFtn==3@0wRmcOW`)P|a)$1Va^GPvkUDY+2{8TIgeq={}+Kjt%gE=_4K>Q5NvG(FB!(n>wn4>+9dY zi66NCAQ3{%`qE6QG+(PQzBUXgxpBupb$i1`u+7szR(2)G48hQAG+ti%AJBk(0Z<7E z^lLL_TdOIRTyQJz+7~1bvmA5O&LE!u{}Cs70A>Jo0mCdGu<0;Z=GDHs-|mC0^!QI< zX@zx`|Ni~^aTd(HJm+)TcKQJ(E3L%uyA+6lv6SPIe*_^jwR{N)fd?fsuk#Ly`>H23 zU>Ze!O$qK_Xd2~TJxA*RUw1p}Z;538@o|8j2WU-Q^q8wCvVpt^7W(kTl^F>aE*(bB zft+^SD*|g{-wX$8zoP@8d2c^(FK_Yy!U@c)3X);QKp+8PC^wWqBFqV7I5rmJ4d{Bh zsyHJ(zXorCMCt0q1gF?aK+^U$&|7t&m?Qy#;71Ucb26sOtlSQx?5=p27fF6w0)|0F>^u!5^F zGMF4z=eMg>-My~Ws|&}>OY;0fio;JsGO+x!Lsc>cV~YG47*?hpQGj{!w45=m7Mod| zlrf?nd ziahz-o|@n&38P4JHt&5R139+F;wMBJH)8GfM$gAxhM=X$<573}dFwy>!h8qJIr%V^ zq1pdZ8IEEK33X$}2n_1z&7y%kp>S&ya?baiyV<>?K(c(o|`7p7yx^* zjrS@n#u7&modM0Lsk!;$KddaE9qXK*gS7G+;3541AT=g_ett?cz{deB6Q{8MJgp** z7Chf;UNtpVxp+Xp`UG}>>!6@!fM?@o%ZBO0Hc6gZZrg2_fNa%viICSR2t?Nkt;6c= z57{yOB>ovd1mEog)*L;qD@HV!zrg~F0Ok&YadsiOLm47R31D!MDl1;rH1{lfJQShD z6me3m3xs@B>mD1~m*~pzs|~jo+u?V$2Mdk68H88SLsC5;rn+i32W!5YKoL2y^eLwq z$jw1529_0@Bcq!|%Jbm3>9G5X@b}!c;b?taws7EH>-YaqvB46#o1CX~MugvKNf;HL zg>Uiv$|YCgfKG6EQ3*Lx^E@vLeEpvv+P#sc7e6|a?D?>R{a6nv909v{^y?{k<|b>* zW-3WdM>ct+NCqu>xeBQuwqikFPqtKI<_-fxP>%EXMAQxNi zV_XWWx>x3W#xwc*Z#tj;eY#*oH6{ieP?&-Ws70&!(B-B*fJbHs)ZXP0ek)`cZtV?)ok`L(pO0Y8<(G+jO8O z{g289GotqPHvqdS;4Ni${w?p^fRPAT_ssC%K~Tv95F&4b=AbZ$^;0i{t(IMy*((Wh zE;<65!QJfieE1vjOr{Rp%5kW=oh77qI-bN#BvPSclH1&jBF(oH&7 zWM^)_Og^$R62~96>X6O6Kc%rqivPmH4rf}9y;zpr68&x6fhC8wfwzp)3e7oJ#I! zFvF{#p-k58f`wTU?45fc_f&&R1vNegi0`<(bUF8iy23cO`S~w2pWfGD%=rL| z&7fI5V7S^YwNxShWNcopqxq(F06RlG4@!$~feksdtgQFNG6J#37Z{C1G!#urO+<+m zpkJ?VeLg%RZ0>rYB>~3vWz<*FlJtLu6dRY9juUc4%xWW#xyxoMq>5B5_p_C0=zuXzE~+P_BA=>V|Xuh@rM3kFexj0d({Y@ z((hI#Df8&k5gL5%@Zj+6lo5V*6!Yw^>Yy+JMvlLWrPJ>ep|6S4?c4x_TE#-b4mj^z zy^w6U?2F=&E+VH^!GlgdC?8RN%E96pyV*2XdTt9% z{Xf#ThQ4Ebg+L%i@9S%f4)^wQ0B&xwp%DcY^&AW{;UI}NxyN@f9VK9z#cf)N5l&^1f&B^H>FIs3?5Wg1UJ&_|fwlO>Zw0px6s$}yQD zQ8WKZdTP+~KXs+1=U1G~t{V574`z_`5EAxfrS<*ax=rmvC!n=Zu-2Max(<6?{fpF6nF zM9Op*k9~Yo&?T0EVH#OQ?(=IhKW0T0-mN`zJXN;RD-P=wLUu>LLiuNtSCnC@XK` z9IvfqmCxjBtl*>Cx()zIv+lN*9sU6E220cHu=y)^vrG=#%^XWn#lt`$+EQHH-1aoU zg9L$jkW`B0pF?s?*U2uolMTjz=d9I2TiCDzyb|~T41_B5cQPhI=pE2nUhYr-0DzE1 z#aFzfZm@(C!QgnwSw&K37`l{tcZ=|t&W&3$XU99K%~Uf9=cHt9)d$B`e3-SySCZ4^ zE5iM2;-h{}Mz%omlub;CgBe{gW;*HB1@*)M`R%;} z=-y5dBX^`D0R0W#$!F8`3|QF#t22@{<(QSUbfhChYAodIYa}m7&mt)UeibKo7D`&IjYENk?-NnSSk_P#EmhG)nS+kDjB?6*Fyt%(jT@F zc{yM?6S%?w-v$DkJgp;4+#2N)`PwU(v(L4oDjYp4C9zbX~URJ3DH ziGp<#Ric6FIZHIUtPD9|rA}jD*BUyW`Oo z^#!Y!R|qTw@8|@N4pb{1n=`kaLN$WkF^KN{Pe}#bgzIlIOv>gXmg|XTK0eED$2PFX z^%7&MJ4uV{H7=duA$6GxLN9%kp!R~mDKK#!wM{G2ex28+EW7C= zx@`cl4b%)I5ssyiZc?$fqu_JD{t_8kW)XDdMLWX>E*x9%1DSvJabWz?W-eh3WMh_B4 zlWawidKa3k3O}0t0#4*s+tK?%W?x!#!OMv}%R*bIbudf_()->lH^X87@@g>{mnj4d z9UZsRJ!AQ~p1nP@258Ll)^w!w^j7kMd<+Re0)aKNWEdN)|1Cb_UTPF5Y9L#y=U~iU zqcaYc{k)`xid-Lu+_GKxe_aXB?B*!gS%P?38)-698G_jFy%C9zk{x!4H?(k4fw<~L zw`RU^ORDULAj?f&G;KY+{Cpz7_582e`LsZ5g=)hcDZ1>9s~0?fVf7iC+-kh230}=3 zKVbeH&e+Ata)e#GGihtt^wDX~!dk##t88k%tZAsy|WGge2H| z&qF2Wp@<**vdlk#b1@>`b*DCi!_qora7zWx>IPE;R1jwXhN732RiMt);2*h@FW8wA8(D{ZAp83`>l8I zT4q`r-Iz+4G}zc#!m3m+t6#MBAN{0e<8~M>l?%*Ahy>f3Z6wruUIOVuk|CvSpIskw6-at^4e6J58eaL>^9{nW?nQOr=2{Q7M);gpjvrBc#rJ@-Y5zPT-vsjII4x zmw&52I4Ga3FeeO-@E*S!<@fb~N;rJME*B5hG+54pg<3g)k8FgyQrQalcaa9T*WqB= z(9Ah5{heM$Buw|q>j*>Hd0oz*`}>waApV>IO@WJxYie<^wSvg2ROE|u;mb$qckrZc zDwe^r(;SX^?z5&^s-c;Hlx!oa&%@dTPr2p<|Jra4e3#M~BwE~+0%YyISzC=#ROAm% zkvK7ELfA81LSO^laM}zXp6RpKO0c9^egWGoavxa-L$c$ zvm+Lkz_iKV01~4a?hX6nE(?n|*UEWZ^(R&J{#~^r$6z{(yXtnIPh8SpQ%w(l*RC&M zl_{uW+7G(xGs!mgDjk$%7b!B!U2e2;y>^Z$;FX(`GKwWTEC@3fS5(XVJ}KGWnQ4*p zBAb}0+xpFHfdfu$YtqeVY~}1VBdR%51JjMVPasGDNsMC6B%sJ5=yjQD$qL?t0Rskb zNeTEQeQRuG?x%z4ewLLDJ+{DxUkYHvYdYPef^|G8LDE}SVWe4ixmI4)YQ-=5O1)|| zRg$7faBYPOR(sSk3R}7kgW#jj$LgEGYYG%^1#W)__F~(T&U^lPgDY#o?~&S^;ZW^xV%6|hc-HxZBQ7g zpdd^Zo73TEt%pqP@jF1x-wi}H*jZ*3lF?&*DDAZ?f*p;D|EUtmGUaj6yFkiTcXu}k zZ>|2@(j~n4^XIctRU|!Vq~VAKhqjGIN9>C>ZQX!~c6)FGlKx~UQ(3ofN-P7O>Q7A6 zYVNA9kTbC(){_cNJqMu>Q|H6sZ+SG3?FPaZIvmCYN581{a^L^`6#Cl01&=>f8*9u( zg%+H3J265Lt5ae%)yY4>BWzVH3~hHVv*6r=FzxJt4S67JUNflLfeq5#mP(!riR&PD zWSvAbq~YqQzF1m(s5CKc|H9l8qP9oAu}GDAZdh0QIzzP%uj{gu)MZwIVqTnLKA;L- z5o=)aVx&+!bzFA4>|umTvd1`wZV%{EbokDjB@>Nm4F;}7j^M^ce~GHiE=OI`s~OWn zNU=3%=-kQElSi2c!7=@;_UqJ5UWTe#;nqw6l-fN7xz9sB!pT&7(B zaArcAvEO-cSx-^p3z(&X;E5+n0Tk@Q>+e4*(N*}SC0k| zS#W3qf8=^UaxEsvE}!0a2^I&p93BeXJM>8@`acA9!@D({j2g!cgp9*7F|&mtJFEGj zqIs3MUG}F5Qm-lm*JtrzbtI2Ce|d*6iKvcTCS)Q&W_l!Ok?p0 zN%=vs!WNP~2(%wW2$f7hlXKzK<;a|bVoIiz`olwY`m3*dd#$3*)%GWX=TV>|al|BD z-{gQZvLqs+2=8_lqZQ}<+oK%P89BWg>`#*pgz*);p=AdgpU@_;c!M^z!81*fiDJo4 z?QxXL5R02x-lP~83?PpFZEx4VHWyxYvRo?Dyv*Ga=4%y5e^b~-vPwlv-=m)BLP{63 zg6sN1Yrmy#UOHJWl%1P{0;f;4-HBT{=~S(n!9om?0FvG#%P-fJ@!qs|aPdM@KCOYX z-{pWX?SOp4y70qq34P<0x?W{fhTZ55ffsePGID;}?s^ejs`2{2r zKN!1u?Q|CmYISGeeK`TIT>u;ws}}+;dc~EMs3q5!E=J5cr2qnMm#c^<@T&hxUGE8t zNx-IHB@~pOoed9lDpku0hDt8Y_yr6*0~8+>6E7W7T4kc2eWuvf2uT~;M< zcFE`oL}w0S|3Av`MtCef&eM%qD(*y5c9xWKdA<(;=4rhcR9T_w2BDKn!JTa{W_PHW3_}oTi$0kST7_xOYZIR3OFbbIRtxB%EQMA`t znP@)Y|0=)VGwKcDEXFh2CAQT%fV`o&%IH!8d#1}IKb@2s?`(x^;sUMH;#5Bys1Yb- zry>mbq#UMF+~QI%G19Cst~r%X?ZN3Co_idyIe4sGvf=D@m!o@y=5TR6A+sP?WXzU{ z_dbaLw1=4>Y`rBbs_M=Xv2j`#8mM64^THljIt+b(^zk)Z?)HA(K??ASS?2xp+GGv~ zB2Ndy*damEm0`jL*oK>1uzGM%3Kp5YaZ|Wdva@3b1Ns%RCZ&O#N-jCra4roCEffL< z1h)Xwrtu#7p|_3!4WT&FHg*>+Ls|t6%a2T`Mnrqn4~dfG!{Ex;{ZyyGK^HkJs0Q z8J6%%o}~C+a_I#K*;Kq9@PeA5)uo9bVX$mmqfdjr=Y&28eABdoS>}fh_jP$N6%%~O z;q5aW%>4X|I8yd)nH+XzKYpy7y0>Yyf^FhZ-2kve$~wZ+xmA;Q+C_Q9v=1-!o>w?O zz{{WiDqzg0ar9p)UvhD~=RpzQ+M(4}3(oX~C>Rk)nG`t)Aotfh>LJTYpSN)noVW-Q-%kNV z>lSc)!Pd`h{inT$M+z(v=jVpH!Ma0=!Q-}p{$-mr7=r**pUtGRb(*+=!CJ^J;J|?Z z-N=sQk7ifMq20aJwo9Kz4OBHF;V5le+?-sD_}%Uao2sxuF3E0cWH4hhZ<|r@2wV1S zGSfW2Om&Q+yA}{u^GldH;8|P2H)DVRTS~_#$8XnMu<v0a~r)saGpKxFo zcdg2R+eJw5D|{TZtad;>O91a>jAh`C;N_eTBue3Bj2J)Wc5BlAx=HbB0sqsZ4)zTL}U zU*x^~5jse4V;356arzE=s%YztEuFaXGWpET*dhmRQvdFxY-1%JOnXw6PIYVNY7<{Z zkOBw_HQ!w@&t>w{LsWbOdz0O*!FXM_1=I92C@kQw@^3v5N(Ac=8<&D*n$P4-MBc>O znn9B1DHZHGc*f(jg$UJiCxT(UBM@AYOwacf{L{dp0Rx@!@$&;%1I63d#J-nGAFfIj z&L>3CbIT6$+_pmu&#ZwUAaHX3 zqs3lt0v}piyxCF;G*uQUko%ZT>;H7#KLhkR{1Y{v({BcU7X?Kt%a_t;dY@)BUB2I0$U*F9UJRh+sHhaAPLJ-+L; z59?dHmssVM(UBUwAT?kbNNkL!7 ze-`s(rPXRdUyT`h5mqb?rih=p1f!45fZjYpdVr)53^^z!RIkAbeMz5&tpjeF+yxaO%pXzP2@95`m-mN3SztC#)bZW@;_Mxv z$4w#i72n1F!i9CGxJRlA-buA*H3ijFKcs#y{OW-VZgucSsUH{J5^ zXV`tC5UJMi?LdVcQ0A4E(t^G|1>{j&J1^-!v^S99rsrBhrnk3;1$Q<+FSVbfEvZ&b ziOA0o=GQ*YXLV;CO$(35$NLpM`nyQk1f)cWc8{#wzwd~41Gbsh#CxwsOU_E5HNCx4 z^)_|guyr1*gt+5$su8`@xZ^|@sn1I0tykNG=&wJ~S~F;r2A-Bt%H7zN*E)x7xN&N$ zm|S(iZ1ZX1Tl4bw-+b*5KM>922=YGLU%xVMK?XzqBYV;#``qrf zjBohnT%04iR^KWTe{$N6oGY;ziiC*`VrnjFFxP1XG{3$@DK*vys`OJ&I?1PAj1vrj* z>+)oeNqWaSwMNz#l*mS;c-1sWAWsVHa&j#RK$H|%!A`>FB+e1J;?qdf%){(Lg_@4e>DX+uK4g|{?EqHGN|28irSxoMAHOb;T(cHmlFE5~ z?$B+7Vv`|mH`N$_v36}&;uh!5ML>}yRpSAkqFH?2GsDK!clQapV72BA`YXCT*3Zjf`CU1pWMWEW(%;@8=I4(MjGKI;ipQfK z#>N8N_H`N zL1`7=Ri(E==C=3h?nSQdC`VUneL^|@a+;j2Vb;(CO4{}>Ou66h_`hJUdg@4$w&uC& zl6j*~`eq79Pg(@|Hh8rtTN9aAUl1P#vAb(GI&(Mn@q7e<>^B%UB=1jzz8o%Yl}r7Z zHEDOEqAZgSko8QmJYIY4u_5-uSj-pG<9=%=vfciRI#n&H1}#z|R^%OKK_t8Cv5{a< zsTED(nRnN>m(dtkF~?Sj&g9lzsZj}j7ktZqB2uC^SwJlF!-K%HNsENZNhLtX159z{ z-ydP!&I>R!hn{znJl0&Tjk0V5g=vf--osy@`YWn_0xN12pe22@`-JFxR|n(byO?Bv z^xfY;Xr}M{a2`Z0kV`(iCKm_FrUt+)0#YVJSUy{AL&J;}sofG+YgIK*xQkvhHcV;M z%~E5V{&dy(2yGgAg^Tp+1BhdNVJ`nNQCZh@R*mSvMC9n3>cElUQWMyoF)%U3$x_v7 zB)98Q;-PSLEvxC`y690w_Kb1tQY_iBuDTStDtZr3>J0BcMQR%D3Wz#_51^CDGtPFW z?o-lW(e5Iv%4bmcLu1Zlqc{1T5Z7{PIm+GFN4Br*xQuuIs77ZN`24}T7OBJj>0-^v zq1PQIekM=&D{U_(5;t$0gPG#W`n|7~T1jT?+eD+?IUQ$RXki_XgsWqc_JJ1;B>A<8Ulw9hR`>8_1M0pU#+T(J`H`E z1XZU;`>JcCq6r&55cGpB3I@Pls`0YNN)Gr$fwPM>M<0O4&zhRsqind4Gk8hLXRUq! zaQOyMs66%i-`hV$$uPHLDdbCvSE_F=tvzzVyq-^>52`cr93Cfn@DqX>`1?xq-N9hk z*2L^6fw08@k=CQ#!CJm3shGllC=-&q0f}3f>Lp&5O2=?`CrFF^S)L_9YgN7_LGWTNhJygaDQ{kRo!|L@*AN+)P3?RHNSk^&MzLp&z^LSeCQ-VQ zE=5L0CLMm9LyjCLl2t%p1jGV^KOXUb%dZp7a?4theWI%G3o~}nCA_=97xWJVl?ime z7U&N5;$crM@^Ql9Zx5owazuVb*DwG5%4;0-*d!ly)VDut*f=;!L6G=*o z_EE~rHs`ELv{6^sST|T7FgH`W!~1HtPtf#GvO3UB?5@lG2wPQSR#d<%#lbDd$r+Qgk*YacKv=T4~)H7M{rNO7E`$NT9oNddO*#*b+ zptg<7P=e~?-W`v4tztdGPwsoI@)Pmz_oSXFHaTz6T_jj((aDFWsT66-up4;196rN5 zkH)x;20wEw1K+k11a%e*C)GH=NE6~ktaZ7k#BYhCF$y z+t8ni?{kWffTkQXr`dLUw2n$M}3EKkQR8#Ax zyFdM~V~mx9VZTIatdZVg`cD>4G23(_JSv_mjqrH(oN!iN{qV`3ze>uNXw2Fh5h+Gse zoRqM?SE{gNbf!D?YADr{E{2$T_i7_G?dMWO1|)rbMEtP);K;I!T3#tLwRdz^sVPdQ zGS}wa%5E<^ThPtVmu3~KAJlh}dXp}_wcYF2ukI##Ki8LY$xzp+Mshmjx;(r=S?S5Y zV_{?_eWIECNwNO(Wowgv(Q`rZFY9{{6p8V|orFQ01KUyL#9lA!LD{4(Wu#}t+C_0` z{oDey*#13CaWCH2`YwGau&6X^^JY%}8AE>?5*&T|)Y157YL`n6d zE0UX4;f!&VP4@0*s>f`Q1A$)cw1~{&odg z+RoI{@7GS^*6^jI{CvX>=n~?g=TR}BKyJMIR5o2dFN42q^oH6X=LbW^zR)g9)(cEM zWOKUyT&_;VCv(F$&8k+7hhrZ)xb)8&MqYlXz*4jAJ{ z2;lTr#xah4ryWr&Vff9%NvM|Z@F0vtP11!Teuf&gNtE{8^Yx@wd37dp`CA&X<|DJV zt*BK|>xVj5VL5x%p5>YmKlUF|r(LPiwcML2^q#ULqC2D^JVA$xCiE9rR&+0h2!1?DR9UH*)G9ek&n(-5ADhkE_`x)nLB_d{vHpdNjtsY zh?O@9ydk82XMjkq3rurm!OocrMXSpE>LNiITpSWYf0Z`QJtX?BV`F?wR{EbVKpX?? zt@zLQ9<>wV4x}G2ZAXDJZ3hVFVddwS;Y#n_JZAxCSuDB4El70n1gZqS>kcRP?(VNWihtPstW^EU}e*O;;?mi8~V;MR*Y=V*X3B8P14o;p!5;3{=J|1YY z8$wPP;{j3)Tu{qq-CUmstVGP>lCja;p+qml=_Vyx&;hP-aBOMbJ98$49QGid?p>j##fCg^qJa2+Z-cG`R z`cBHEU&LC(+wC8T|C*Ul;rbh;b`)Y6GAK0ke=0SqbIOV~Bi@JL zArbxDeSPIsqBptRS9!wMm}Z#<2$`&0Tp`L4g@u$jI5;WU+4k6WT2+cPXF%lh2mE8C zg+oqGUOBg8?#=Az*1kH21d3vQnAs}1^TCv zetNG$XJQwl9O3g}4asoF2_lzmj|UHW#Y&7K)8h)FxXF@d)QrOj2F%5&F8@ZJ5xQ%c z4rNXes0}QF+9Y*u*XWC_@|Zl-^#`W02X)`U`pAMO9XlAV;V+B4p-#?!I8$=TMnC;U z>mz}-o=ULh#4|m%T}sF=wY|mB7~CuVtUg4i@;QrM;`5rm>UP!o|h0}_Zi@dYx z0rcZvoH^G8x5#5U^I;1ABrt-frD^46gE1~F~5 zdy7i^9!KCyIqjc9Pqtu{E$NGSyI0MsMIIY(z(LV>hH%~|aULU|o9+bkEa9U=2q#Pn z?@()qs+rqr?dKPNYpbcHJy{2_V-L4dXL4pNwA5zXtu1#7DVVwFQ6CXe9TD}p>e0zZ z>zZc8Wr8cL1Y(LbOW}Nna8B^NYgH+)!|`jD;cJviCo?tTWufHux^ag@(N-22aK*@a zR!v6y2-%#Q;0(-R7AE8QG_jf%6~ps8Lt4r)B{ownAW4GUJ@O7~w!QypyNV zX=%L%ElFAMvAriza__F*?u} zS$zX5F*;aXM)yvB!yBL!aZw@Am*W>TdJ{Zc;XXwvy*soFKe5tYL&F2Eu+@or56 z+LBOa&!22?GhYRwphs(+krI_5l0mZbs>=rUN=1a+^g^RU0&_f6Cz z6*H$TTrX?CX;=3?V}H&1;lo%b$i_GWzPuzDzxM%_uA^nJd2@eP4SL)^Fy=FGRJz+R zfPL>f`yX#Kbz(2T{_kl1{)h`UDQ*&543GRL2~GpKMs3Pe*}9e%NxH-?M=k{sak5x- zM0PwBH=m^`8?6(b^ky+!BX;qN)&Tg%w8%B6-SD3=)utI&3*AgBYhEM&(olBlGkG~s zT29xin(aEUK|se6UwHYOd@tY`cojVQBuW{nN;A@WN4FJ0VOgGQ#nlRp%{9BFY%f$Wk8m*P)@(p;!WEC^%O^0ohJ^eHheR?id(ary)}<1^>vH`@Ec|6( zcu3Ki_qAVdmS2t~{Q6HG@R5dK(G7}kk4(=Tk;(TKv3bXBjIg2@JhQjKTibgsTdjJ( z5aae^A^BhMyxIynfvvlld3|Zl?D`1%EGrapE;B2dRq1@JZhhglk?`l$N?1P%iw*Jf zpCA{O?gFdalFlu*Og#@I$S2DiS>TY4V;G*|DXqOss?E0ltTIOIB8vg_U)ZZf?THR-XX}Dj<|R z96P-2u@x(E2CMgXVCOL$s}UuHiymb2r0WC6>;*`yPl2t4VEkp6jT%rV;>LGD!wIH* zFnbkK$E^`l>s}uALbyES>G9rk_Q)N=((!qZ>x+z0q z$bLE$Yn+vLU98mTTFj9S!_WZ?i`76C4|MtUMXuRY-m>~)$2w0~;TWHLS&yI;tx_Fg z$&UvPhCxRG1S;0RfJS6nUslt0td)GroP+eA9Ss#tM`oQ^d6eAorWbM>*Cw zt!C?-aJPc2WC5@9a%p|kOQ51FP!2ijt#_nb?}TZSl{w{#6st3|O1B-$xe_6k5yc9h zJ$ya#L~kX~KmXl6mCJ~VYqiWf9BV*0J!q@Rm7xZla?=x>sHmSR=K@?ywRkF5lD0>y z?s2z3TFdp~sb_Oe8>)@xxf4jUT<71#Sbc%HH*6=5k+)0noe4hSuysp* zTT(Eih(+UPDV>Wf%4noz%F7?&Ui^P1u!x;A&)W8Q<)&YIgd%!`ru>E*nx#xaDjvXA zP^mIK#Lmavp&3`m_x|BQCVDGgT=O+nA6@TqDx{5ZAv=5}=UH$QGB#VYAKaQg{~4HO zzNpl~zmWC5XtZzC=W(V#1^csR8K^7h`M?KEHcZSxMMty5TPB~HrbDKoXTtTw?qe2K zXW7_ojSrQ@w-gait0~tq~$>wTvR=;n}5F`getdKI-3t{3FC1;PjGZ z)`vL*yRw+ZufWAkgGGjgbe_^L5MtH1Se5~4X=y1M2GeT>28NoNnx6{(e^v4(E$B1w zk%F(cIq_iva1<_HBy5q(U^7bDYG`EDVS+TJ%@X_pwbAhDSADs~(nz~W4%^<$lx-(S zX#RpPxDJ4Hdt-xa(4_7vb6JuG%PN?qd);BYBiWqJ1Z}FPVoze!`#E1N6K5ytaWez- znFgHHy=F7g%Y&lur-|pPh0>tj66g5(#2VVnMa7GT*=87j5x_paP*_`WGpFNezu@yiEei)pk=>h1R7_NP3{+X8)J*#u^48(qaMqJ>R^lg5x<&?> z;Qt1Svy{3^xoH{2KX?Moyw1cnZ^etwf;QWUoAz+cTj`M}UEAI;Eb9~2tDI_~YjWwN z(-$`3jR{~L>7MEqsdcJ%*|(=T~CC3;FXqF@(`jnXPrFxD!1yp*}Y zypF~=j{-mQti2m(0xP76RI0GGMo7rjQi@b{y8fH@D52?3D~$Mm5wm!!lKf7FJ5SMK zBGD~B{oF$Ucpq3fIfD;d|EC*kKLl@oh%8k%5ZDKe?}9{d%&?5{~%6<~4-Mr-{C-BYr?gMRgFJcg&!D?huCRl<1BbHBZPB$dE&9rc6EVY;xH9BivxKYEUMg=zl{|W+| zaO2XwZdepYOru*}wCx0^Vwn8FlEB}(>4aJA*=-ave@ENo%Spf-yI#xM>COva@=(^T z8)%zXkQV-caY{slUO`bJ?GkTXA!bV2rvY}^9&Y&=5W|EJ9iO18a~xKtXU6(}1}SB$ z(aVjFiBYDvVXvqpUlyMYpvvpysroKK0cbmY(ls+S_6=xDJvK12C7F_q0r(lnL+cif z(sq-FO@Ws87YMij<1~;72X+Ja`1t$#yRb!7(EK;pA(yW0$Le_1$)CFr{X=WA>Mjdp zFI-|oMn(d5$Yv`DN)5JZlrB`xRW4O5pQXV<0zKC`h?a_X;90J9eEr;--sGhBW07)% z4c8KqAr4Y7XnC}7VCyk-^ms`M^_NBW!Io(>1tLWj*;UZu5&kzPwkLf2b9F!LF{O;4 zE|%Cl&bb!v?B8>bgsn_0e!q6Zd#8OZs7W-i{5DwklH_yMahwlL?^Hg%^x)eT0zRA~ zoSz(mK@a($r6Og-yjR`r_(DuSAbBYup0}%1FJlrpZ4yzBMIYHx6;zotN6BwVtNO+9l_IA#+0Eky;Fm(8oTmb$hwZW!j)NVmn8Y~Q;I=7 zi%uxt@JN0QY^joKs#8Qa=X+(9P{Q{qX^4HVRcayv^E~R!mHcnk7wyD<#~RcbwUV2L z-K;2(sU<=8TFo%G6x+tBYYJv%d5>@2uD|d=a9!CW{nU;z^yiScy}*(nRx(JEqHB}* zhG%HosoV)*DQ&RrES(5LArM|Lxp)@Vzgg*EMuT;2 z@%aE-`hfCNmP)i%vp|`%_=V6NZ-&>f=NtlOlb)i>*l&kiK$d10H7TECYyR>3=Rm@tY ze-OezY|xIC$e-+Zisn3xa4qD?9}+b%jN&wAF{1iu$GRl~2> zKQlyfZyqyQX?AMVd}ST{8#`2@Jak^j;N_&STQAk!=d_V?O6gLzw5fNbBkz;;DYxEH z5@b*i5_V6R{I%_@lNAz;?PtaEB#c&}Qpq6I$)GgQuV{vjm<^gL=U8jz(yZpb{t?dl z+WdspJWo&LQkr|$vB`VHzlg{~BcwiL&ZB-OTbMnil&g%m>?3XSxyS;Jdwf-T&jTj- zv{cFDCgmyW71-=LysI3oPxLfvxIfc*)0M@V_vrXw=4aHHib|H2K&n^HW3GplN3xmL3$&|2;p7>FqjtL^aVXC zX{J1Qq_wv1psq^NjgRerNj~?Ek3Q+2n#)aVKUG>1fQc!%6m57qb}s4{j#{aWL|nT5 zgog|3j}L9L0ck*Q@29E#%QSwOU7{Y}o*AAZzB%O)R)F(; z_O%k-)CBJ7$;azA&9(k}(2qM# zA%+cbNz$}zsj~DLspGp)flf^Zr=xk+Xd?a)Q!UUnfK3K{>m}?6uv;WNsbEs(;hDB&m%{NC%F?8~*uV|f@?EJMVkw0XDFrrSQna7-<`JOD~ zf7ee{IvH{o0=Wj$k;8MW+pQHEN|xik)z{%x9YY)AaV838r6%PTczBu~4H!w-?>9)O z3DU^+-iecZxp+BA0$y8UK3M@5JdK5|hS`0A)s*sU%K}A?V<<3`4pQBwtdqUOuLc+f<&`(N+GY`_#nIZ}j+5rcleKc*;tu z%921>;hTrU*BowAY_v9!04}_Mk?sGscyA(mJsvhm6o6~1kxQXa*?O$mx{;v1z8)wU zz``(GEN^zvX97u)1GUSQmQ?pu)uQ*IRW{er8<9)i)L!s}VxwUw+QfEk{qU zZ&xsli?5OWnc>O4n?*Z7whhw~pLDL)$0$#^^wNRCe3u6eoOx=mFKjli?tDPA=3N>r z!7g^_mRFX$2hqntv-YFRKr8TeS9Wm_6tAC83i1+bwphIG^mr$Sc%w1y#QHm}d@4yh z>E<)0=NvNmA3Z#pZQ0abFmTBnQW}L$u&}b6TNnzc*MJ$U2j<%@EZb?$c#E_5lfQ+* zpe1o<0LL8sEa5d>R!-a)%*T=!{I-t19u>1ek|ko`c9FB^o=!+Vh$iAP9O2yen0jj2 z$Z)#k+UdzwFbJ(z@BoMA&YO^DOOi4>8LO$Kc9Qa;LMaZ1@>vfoh{7eXOn{wCuiN;mYuA}<~u zAMc`%#NK24K3newxe<6vOeO2>k4%*XU8Bi}Lj5uiRhUo{Zr zh%o76GRu^6%?i;hdMchU1ji0aW{RC?r|8-xPE>a+`dBndX+A`?V(a|*pYft;x}@i@ z&q*I0J6b#n?MJs9fWpiG6sGRXNhYM=_wG}&u{ClU4L!N+2p7x{9k3@zvY9b583GD9 zJfg++Fqlp&-2|;1SuU#w_9$Kp7TIrZ+ef4P>NsI=1=CU}BQAXmANC|FO!(a-)M}NJ zF~isQvsPExIHQ9IXx0fjte(}s?tIR~-=ytnHT|-pb>2#LPIOjSd{)?hg%8^_1UsLO zUjA`*@ZaE9anj|6;v)oGl9u>3xiLXB6Ejyu&f?T+axYur#)|4Sw1b3#|^cOsD+5AliT&tD0c!A z{yE8%Mv$c${O?a{6p2Q>8xA`0g(;bvtM7TR=ZLAPNgHUhyT||hFbhe5Nws`OuM9vU zTp|suxLVPcC{&nmXa$(1H=96QpvbEM08n9%#N@Zj!>xkV4gA{%|hg(KssuwnW&3 zDCRdf1?N_*?u}z`AORuQz;VGCJt#SZVn_hN1>DFC>9uP2-=Q-~0ct;W6ZETsvZS_7 z(W~Ub0MgBH10eBrAbM4g6E9K~1;-RR2Vtl!;Mc(42CC&ZFNcw(neUBUjBPKySnj*Z z7yj{uVQkYStZGIX;16V({?w@Hgw`uc~uQKr)R9j6uXs z#Qx9GI7wrE`*cG=IYFzy437Y*0>f8=@Gdq@!Lakd^|N#=a=y}LE)@9G?1P=EiXx%ba*API)w`aatQe85mJl+mkO^j3mZXdBiatcFzJWyL@_c?J=mU8DEW%3po#SYn%bmf6k<%Cxs zrIf0Ul~dR|OkuSX zC0c}%Et>T-3XFtS&U4Swub#OvDASA8j0^c7+nyGCIoWP7L$j#t*RSC#BZq3X2?woa zU)SnVL$hOR9!7XQzsV=roH5gG3$cwCiS2iGyOk3!PMTz)J=i3$Yp{rOb1`h4JL3~+c|Dfx`Mb z;baP7HIVA`zu_(qQL&p-K&b77heYZBGJ%CHB(?0zu?}7WxgHQ@f$Mtj7;3SFl1_3~ z18bt3<^v70a#NV|84?TlU(-c zQ!Es$=m=a&5g$Yw^MTHsVDF9^PdF`=o(QV#9J+`Ol^0*?-~AqBD*Md^ z8Me6!2z&ADSqpF7j@()=1W%JN^Bohjml2KzJNq58rXaTvnBOS$3}|umUbk%P?9iY*d~DWe!SMG$LusXX)7Ao&bH>nFF`| z{}+7$rcnT>p9X&d0j*_IjA+6)tU_s4;%3LX%GPnWQGnHw$8Ue(?pucz1H`16U%*j~ zK<19lg9UbBkXNT>J!4VraM~E3b^|=s{vwZmR?Am278LBWv`aXae*x!&xyo-42AD*O zZUXUXR{pH#D{#FM4Z17pi8Lb^{KD3e{c=<5zdY%T3F5%rcVk@5ppAYu>jPd1q+g!^pxRPz>X_Jmz!RPrdN^ zY>V!xv_c9orISM+^41{Q3CEC5HQt&eNz*h^`iK)|q9?w55tw@*hSf-V)|B!0nWRKi z7q(6{KLQo9zb($?G^Pf>hLzO(Pd^iVy-%3b63S#0vJIZ(a(P!SUHRlNSRgqyK0CE_ zj9eI}2Df-BYE}LRhh9b=nnaegcC!n_Bpdxy493QC@;VM9&ZFmZC;~a*Bi+2;B?f=n zSg?2SmSww0j|r&cijC1lFw1|Y+Na)>LPwEh#fG&SIb`^h5yoK17QK7e<<~^=nlR;AVJ}J|VhAr!%27p_+QpP6Ox$ttH?BC||!^vrPo1 zZR_QBB-Az_Evdm7K5_nWQ~)06Z#Ltl&bO!$ucZL#I0F6^UYljH?Pi^cFQ%r!AX#Ya zk#d0wW8c3`iMrat(`ic1{dYtaDL#{0aR?;*B;dJ6_85TESHHAo>zJueAGF5`8P&!0Z9d z8FdMcAyB#j!CSg1UJq_gPn;pGsvElJX7qHoPuhYYo))&xJq%s-5|ga^Rr(9n~&lP7j^^X;AG9# zoZ#I_s+3ViWy6C_T@k~W1TLSU9qpEP(UdiUUBhyi8FZK#t_}_EuX+@1!t}XB!L`Lf z+JM0>S@R0b-Ox|{Q?i(Ka5}elldlB< zL=X<#C>lxgC!dJlT1I|)FH$ZC%fWaNh!of+a8??gt<7Am8PCGZyj-^zyulyV22wsj z5Bj5a{ZjDmeq!Z0Sa1OYm)p!scwz*PGGh(@;xm3Hb;)G1#SBoa;Nsx{Uh}VMrCb^r>N5gDLVW`R z05Gv1AduE))C$_nmJ56cDjz;|WgXaqAe-hHov5kV8`NQ>XnfY9Sci@61c?QS^>*Fi zl4~c6DKw*cIBI|oEANzX@z}S_Ka+Y4Ll0i@nCjP#ynG*R`IAHUL z1>=Dd?QgQJV6sMX_Ir)OKw5T!G8w61hr;*R3kPti7e9e6`rEXx%BWcLl;(j=J+fnf z@|V=iXrL_QdxlGjis}<8o$z6Kb9()H&QRyBXqeNN6{%J8Nw#+IYR^w9-b%4p#YT2V zE~$LJ@N5mZenhq*#J6@vI4_cbzc-?`daF*pU09@D_{p1$D7`#ZnT)8M<<+0OuYjVU zJQ3j_;c@02+p;&EzBik_Mg4W}Z>@OkwKC=8FKkXgOz@(sX!$2$m|2z?-(#!q;L%60 z%x`A|TUf`+9Q)ZfQqW}%0vXc5;@cOb0zoT0;Fhz#v7w=^u3l~21%x6%dJ5U_;hsDO zKrB893DB!nJD3Z;^M?2dUzsw5Im+@8k)XKsZ{Ee*MNd8Y$`X>3cdpz@CZNW>V9#H5 zIL@`&x+9(L3=kNIY{-`)1nN(WIeWXi5doNb-Q>_M@7$=1O=yWAVxQT(X=5Ien8Q#wsS~0hu^*nzx_mBc5b8PTt zv|>D9!sNunL8rQaW~xBc>QT00p4mQ4yJ`_$_jie`q$UYEaprh4f7av68-Vx7MQ3DW zj0@Z9&jUAcVg)1Ku4xs56W<5&^0Bgp#xIG~U!$<7M!1YbzU-bb;$@;dGoDxKVSCr5 zf}#RlRFk_Edz6P+hf5EP$r(o#cv$5*d8E39>#`i9M0RKb_Q)}WL@3W&e%gm9VvW5;3LmN-WGV%lTKbmr zaIohTDI6Kn>sqfJ2%y|q$T-iIu4-L=DrK8%t6Zj1A%Xh;39p*Sk@S#DMe8gvP)I0n z`TPpZk2)Sq2H7s=)jnE`O5MV;cdMfKcj)sMw09#?_M@ZQ8EENaGjxCmt1EO7Xbx2= zG{AeOvOzmZkSu{%J|8u^fT;HZ;|rpwmM7)mYc%URXMxobcT|p^Uu@s9Oby;eI*elf z^hyoZS+4N#8C}D=`2L8TL1LCr{If4_twnVJr8%iS`|nYF!JAjE1h_06!BSQ5=! z3`7fO4=Ug0E7O0SJ$>bBcG}*4l)jdylzRbm!ftyF$MTSu+w=b1D~(2QONxk@fiM)( zt^@ag_nSB4yUZr@I{(gWK8c!3?z@t0yWY-`)ZczFA(1v99LREdxWPE$7slwoHp>h{ zGrfPJGj_e?SCx)`Clmn*?26utcA0}o=>Jt%fyX?eS7(TcB#(h4pAjS4#Q7htL^wmY z5>pV9Lzc~Rix?mrL?a~q$pCq7Mew?{qtif%x^7i=)#*4EJY&LGR*uR1fDuGJ*F&tA zvS7mPJjh2k%yqa6`DBcX)cFqob4@R~B&<+6QjNg$?@wM85Y-m4X2LipIUIyp9&qr_ z6DCKOEs>)xH{qcF8|eOkw;jo&Aw&RlL~&NE?@JU@0&IQr$J%x5n%(Y{c^p(o;J)*j zYz0Zd`3dB?g<35LNJw@&SkK{jr-3Dj^J!uqcf-JeMm5?+eBj$87%&33nw-{Mf{@$< z1%OSspB8*vySv{1c<52PU0kmU$Ns$SW#l_Z&I{bzq~XdKQ_pov@caY%@q!(JV|1uy8BaW!rR?U`iw;3-g7uF9Yvy?R8ecKjox ze6^3R*U>v)DN|i8`j6Z?G8!hJF41H$(d0fE*E^Y1>Bp->;tT=t(mG|3w)dZCpJO$N z3Pa#4-B`&veEgMPtscH0Y0vz)xdLf;7w>+8s|JKI@FW6M z9%oQ{e>xA~zNJ_|EO)17*?j8=E~xJpJ{Ao=;yV9uzThQnh*BM4tSwgTwB zrV1;>rzM=cDGJ2BW_W0WZY-dDs$QOG_5T`b9N~inl2q zF(reGgnsWo+XX3-9~at|^D3g{^rQ`bjwiMs5Sd_Xka*NgBSFRbmXp(e2pOo3MRwd^ zIao)b=trUHF*12DYNZ5fEm*cD!=kDMz1=9bLR!ng)qz;xK{#pq34)v8$z9IiUzVNU zN#~1#9eY}r=h`;XKj`gNcl(3U4I42HgMx&_zLPUr@c#}6lPJ(!iG7d2EwCD;jj-_J z=)<;&4e|aa8mpA()nz4)=pu2DG{R-)mnnSGcbPkcrQd4D{D`YwB%gIaQ7;UjnT`xG zy8=C}Hs61Uck{~vc1wWOziY>RW;0*uRU-e}eCH_lS5t@}Dxp<0b=miL1V4zt?OVJqkKV)Y*v8|5{}uK}eU7gRm%0x-nMf^fF?kXhQ;B!GgX%OoX%OJ>W7AA72hc=D!tg{SBc zh-`PNz=uz0g^y{nG5XIk zbp}doY4VHqDo{>LK}`vs+?*!-$%}w_11=nroXLtwr6H4)>B;@K0RBlDjCQ3J*DL<2 zq_UZS55p_b!#tq=x7#hMccE2WGT2Pmuu~QXsCR5mS~y9}74%PC@f~4v0Y{ zbn0|}f(>*W73!?0Rb?2a3-Q|fVsIe_86w^js2E46SQP;VdnF|&3zbj%PT0XIoDKhr zglP0nPuvEbiXE7hRC2o&O`hr02d2}=3$V%y*hc_=>Y*E-WvIa&c2BF6?V32TikV6q zr+r9}SArOzH6$&ar-x=ckt(qxH%Y>~+E!PoKw325!}(GyRfd3q%`7#oZ_S!8depW^ z8FX#dgG0K+Wx&jNz)ZI-D_p3zOZd*lvhh?^K2-oId1|6?V%!}R#{&^;B=iNj)e+kz z&8Z)+D?d^a&tJ0b=t8ZzVlo`4Q>Ywt*_yHIbxc6uwQzq83T_ZTY2e0v*`0Pnd3*rE z5)Pp!3rX!kQ^JBt%4#vdpHt5&a&4x*#2-RA`JjPl2UNvtT)Lm4AOT${i{j3@Sbgz? zZwyIZ)pQ0D$EdOP$%nI!S#&G zNzSX-Bd%I|3HLMb4WPDvkjX2l(=Lx=^Wg@pxBX@EX(G@3EtTmlRmiOT3{%8DK4H*Oj=jdMYkN zU)-lX#WA!4DVfQ9q4_GTtS#9Xt*DqI0YRsrMeSo)e!NhyM=jZk2W3 zN_#RP4p~lCc?##R5!~LMakW&Eyg!adsZese<2I~k>E$W?^0ukTt&#x-oz7i{LZ*)D zW0(%uImi$BKiWy?z`b;D!WjhMU(p7zRGnML01odg-)@jM5Xw7P55L*SJhNe>g1xur3qxRY z0=MqTD!B)xKL;0Ahf?ZMjDyzr{+B9?W2^f!D|gVYxM7!eO^haR&=RM6-pSn*xo|tT z@Cq{$nd}w>p@Xs}5vfmd71{Ho4IE>t@zr*gfhJ`;KIMMBVtB>_(kwH(8qBee#pAiu zN=d7qR`RFFXw1GGR;A#tiJ!sbyn{zb>NQT);oJ29VR6hztk5Kj9kU`_s??vj36yi zzELKk$bY^vRyCDc(|5(2U~WPhP=OTi9ZCC3^eee>xkwt{B2ykm{|b*N*u4$AXB8JU>I76*@WWkA18twtM zm23wO+h(15sFk>$Q*PyQ$V+HQaLO$G(c?#tq{QATgSTEPcN8l0g;=FUV97JN^}GaO z1C@jNrQNwgr*u%HiVZlO4yGrX;wAvC1++HS6T*_hgvnx&=$YH#uR;O%C4r?Bg=I-t zqvvx@WMmgD2p54~7YG2Z;da;MBwPb=UJrHLc)HF@#i3RO(%G1x1LwXM%4Ki5J1wyT zG4I7Q|32}8y-_LcO8G$^UN_Mg25#2HN$m4paW*-j@|E1zfSE2o|W|bqj84Luc^WG=EEu&C$_r# zD@{y|a98Lj5hhAH_Dq}C$}~4<$e(Fds((H-qKlJXrDah8cYX(yKGe?%o=-9`1ey?I z^7uD2mZAu#21RbVd>0$aGK|+;>ev5*F;Qs~_s3xL*f`)DK`;;;v0=5_A6g`L7@6Gm`J$XxSA+ckRVO0bXSi%) zTZ8U=Y~D&gIh?M%NC!g=dWFybgr20*>J8EA+F6lapC&a8RC2#QOnN@F|84R7??z&Q zW1_uYfIF{L2lIP7nV4NhluJoA2Ns}Ps56XW8u|bL9mmyl3?Ti7r$Kb9Z)Dk>5H>GW zv0N!WeoN0&_R*zL7tqnuO|>6`=GHTp_I}@X=-$YR6ap`hUE6bRYv>D>#qnXdS2C3M z^D0IDX_DH208n2wgHoO${)J2V|HcU*35Q0CBId^HNMai#a#eK?>ynu1Q>4SW;uj!J z^PyQ6I0{F`k+y(zPnKcJ>vT-**c;9LYiVhz4*pgrSXfXka?S18%w;Pc5J;)dj&FlW z8IpKi0H0YG+0AbU8{&v5s$|C*F=+%|hOLXI#A4@!nZ{P1EXG~0h zeQRtI#@6e&F)V90bv3&}Z=)Y>N^-XQ={8D5(O&6zZ{~$E(|oXP)q{LUem&2C!5B6M zZXUk`-ujG4QGD}-6UkBz*3h1ye(lZ-vsCPsi>ROl zyex+1kHF__pk@&ThH2uy_tLS$F=1i;dGMvOEaVHcBa2i^i}(T5N(v%7MsZt4JV!hM zl;LeQy#~P?;Dg;O1$^c`hCPP=kG`+;bycK?>x-JDh%4*DOsm>T`i@H_7IP5PakLNmqewJVYqi%Y$u1EemfBjJ zmh+xn+DzoynIc8xB8~86Pn}-Bv%p}c=cMEQY1s)$GgNSnY7{1JVu^h6iKG{`vr zCroQil30PxpUa)E=}2)tnKzd=!sn{?b58ozH;sVu4}kL_!E%r9HU{ zk#cFq{v0T|W~+w(VmU-t-UbYYV(_9923hCmRBe-wtq}AkBE#5+*$$1#Wk)4>SXI&V zt|YaXaQxBrdD{*m5rqSQ#E=JgRc!*m$KB;I!0lI&vCGb*h0i$pKjAM7^UN>V*n89~ zP4=DR+}dIfv#g-|c2*VuZsC|G`$^rO;aG1x7E*Vy3Ud7`R!^OLw!PUwd3~Tad*_#q ztM`PqfPx^FD)&!JHxNH(o5I{3FmyN$5`i#O??%=AW!SA)t~p4;_(v_e`UtIOquD+Q4Lz;lo4CZ1ijdC7fbjN6rEbATe?SYI&eh zNL_^%35W!OioRObnFdC;ILYn~{wq)g1J4lD2M8?{%x~?*USYaW@v!~^A1IlCtcMWs zr`5;n@DCg04AMcZo;>$A$NZdiyIT@tolt4+Ln{ol7X4Y=^}6Dllk9ai;8~m>bI;jh zJqmZ5U2z@>(QOBLb`+@~nJTSPJP>#m;o1AvY@tPwwe7mi{88&kqHjr{VBhmB&*kB7 zpZ0^1%3lYcCG@dOgA19xcg~nhn`JdjQfpu8*la!*r>H$OE{P$+{iU2GwL;Nqa-N?F z{K3^{KmBMi%)kTq7Xg!3Z6YzcCo?MN~>0v@KVzx3PC4PSLWvw5QFg}$k1EB|H zDx9dOi~^d7eaPsC;aJb*07snF*FjF+{u;kdT);lTd6Tu+ESg~;3h)ywL!aw$W<|b< z+>syAkhY`x9a2jiK3B-Fs_!riZ%ZU1ndh`(jgZS#4>X&J!3Efvj0ZvmcUIlK`?6@ptttWDAW^hO@0pR0Er%vOT z^l#V^c#Qy+WcUZ|(<;T=b1d@ zTR<%mU_2iE*0lmC|GnA`(mPP;3RY0<7rE>Xa65(?srG>0JFMbz54(9`2r|Jl*REnh zrGt=*4ax=zYy;#Ksp}>=H0&IMNXv<20TWHT=d=gQNy)1Tf@4Pz=ja~ZI@jT-1Qie;W%HJG5ZGq(n#vPYD`z+#(fTW) zeZCNNUdaM7f?>f_hq$UKHVa|ZLF~-IN{il*(U?&?o{m^@S(N=m$JLn^dcAc`loofT zE}SfFvh3!v`UQl%iUxC>F6=zLomDbayPrWM2Jq3r3D^M7vaCnrE^Jcvo;1wS)ib+E zz=FGZlwG`HqefaY4lT2hSQ!K4?9@OpN{7H#3YBr@3g4PQ@qwNfZDdb~)CPb=E>9T> zhl9R=CR{DD@?f}MxIe!?UFEP(g3#^jKhW%OkVSTDJg@=`?x3}N7m8wGAO(5gMJc_{ zxo2CT#tV=vrY5ifGR4ZX{~DF0Z`be45(;PF4=!c!dhC zKmqK&^-3uBQJ7Q1Q#Tc#B5mdeaBej14!dMxCi2h&s*61~nzg{wvBv=4kucr*8#|f4 zqu;=!DDA!b4*qSaokZjNmzD^sS!>6J#RZyU+%p<+0DaLZ72s`nikXkCGYYGjx|0DJ z>2?B=M@z4c?8IkbA`?vZbAer-m(W7dgXJ*3k+mKbd+p3h4fH0CEKzQQc9x;>Zuxyn zp7I37NOxK}OFo4nbHlk9ri zYlPWbE7&>V`d2>A%WW3{G^_$mUt~0zQvNyc*!gBE|K)+-Svp3Jt~iC=8nT`H!U-Cgj?^sjX?ec~U~tv_^((B%r_ z?E4a%htbEH^pDk9MpZ{7q-o>AV<#PpA80v&6$U-i4fvw+h0h0eCqWoako-3>3MdF? zHPp(TQWykhU2ZqunL!C!fO|qT*X6PU=_E6j^-vrwVyPOjplb}kT5JYZK!^^+e+XRm zD9oRnZlrk$K_@}YK`RP~P&o-+m;=|Thd-P+^JI{w zQU^nTAb%k?2=BG{!3>p)naAZlmg&&niFgl_n?I~7eYUsw5N#>X;D0s_`C#sc0; ze|RaQ`l=_7Q-!`JOl~ldQYp3Lui!-yE$|}?as2K0@U8fLepYVkz<$WQ(E=XZCqix~ zng6!@xq}}xg5Mx~asy<`vOE+qUalwAg8kGcPUqJIr^SM@SKPxH?s5*=FD+x+K<%H0 z(eux*Nq}brDtPW=X?E{e0VzhC;den$r6dW#wyKWj<=V~M!Yzq{>!0Xp%|Xx(+!@L} z-wodJU+Yd}*OOgmUtTH!%NY(ki0eX=q^!!v>8?N^C!o4gk<^M~xvnKWFfSL$$-Fhl z#eb(GEN7uQ^Zqp7r_nfmmg#y0+Y4>9VxWHbiQ^dcWecVZ@OVVMiYefx*Lzm#uIsa8 z(|Y(6oZt9{M)b|x&n7OM+(3$6Ajd?784FP2=r#m=Tfm~hOn8Y4Yc+$PxncwnJthDs zKM6%Y2~GbCBzZCh$fE`_<;tcmAWH9xXTYCXDf@6V1^b|uZfwJs4eoS(o_iKNN?eu( zQ#2DV&jcqN@~Diq$@@hhKHAq`Tpd8Bfk=D4=sT(B1+ox*#G^>q|oLHt8U$0}nis0oLH@gJON92TmloQHwhR`Vx znz;t`XoDlrYXBgd)2?BPrbGKP5D-172DFTjf_6gS&QAE;t^pjD3*%)_VhB{;W(Q8g zV&@xh-ckT?=@IxRU9Hb^VbIgxpDdd{$>Ba!^Ey?7QU(}Ey@2!l0>TX})D~t@etfxH z1nrKR@aB?F%-nh`?=4^wD<1qK1nyJ3;@3FrL@3e7iJ2%BDY}m=8)e63Msx}$SnJKp zFW%IVw#vr7CBU!q&j@5|6!M8g0?!66D;mFZQP$g>KiS3d`{q(;TOv&p|G@H76KgK} zxz^!^gaD})ez6K5DwMrv_L2s+KLWYYLpQ98Mx;*OG+Qpw9P427EzUL5XA@O<%!40b z!u~nierUoN_Xq9&vKg;y5tjPLGtbXD0&n|4SVtajZ%?8MI2=*X$eHA2$*A8cizEOn z`v9x8#2g(XhRe3_eB%xo}JZRAgy=mThs>0bU`mJfaS2nXj0j`WfmB@ub=Eoq`M^jhGojY& z|0cTMj_R%g*X4=3pkLxpW-YTTCD1WqSXNM?Va#Ns8`2O_hs>>x-F!Djz+ZaAr9vqa zm7GMOfUCv}Ho*|bIPu*N0umAFu`icN#3$CGJ}wxWuk*9{H)M+B^h<@jhCH)=o9x+~ z+Lh;LDyC?j)Z}$0;B0|<>>$Qe;HaCjs7obFf54MBC$^k}yuHb`i^QqWrGNYMs4TwG zT(Zv%v`BXW(tv}5gMsuOJQj{$-m+ge1U7iLg*>lLwsc6Pywr8c%e9%$t*P`2P}?5C zhmCGW2Aw$wx>S+3{Y*pu8TN{OtNX~F#5y8w?OJ={?kqkH(Td~5EoWG1zo}Y(-?o5D zd4Sx5QRd1RSB_%@AYZ|sTtvVwLCLxL-Ct6-rZZTuYV}}r; zb9g~Ir?*Q9GKzFyxyb@@OwE4loT3TT)VSnK5`{RevkS4HUsxTa>aU;@tkm>U%fJf3 zPyS46%z6jetisG2J_f(vT{`1 zS|&+dvG@3I~>V>j~L(yEVIGF(XN7F|k?PWqiq~#pE`MEiz(R8!=dGn@3D05{8 z7ZiT_2tn;=%P8$bh4t6QjdSumqj|N)_^6VkfhI5mHQJ6|&RC*|Pg3E}h5C1wv?Yd@_unKim3tnUvAo@+xm7l1PfLD`g( zQHvndU)|mW)fhX^b)^Anud;F$T!6N~-T`n~H1n6ok&>hu=ZB#D7}`i8kQ04#h4Ow# z_1TW#W@(rgea8B9iW#%ZEWiJL`)B{nO>upwCaJBS#LoBaI-U@^hiE;S z_vK*kP^unoe0<{#mofQPWaO!H!`tCE+*GT-61kT8vy}{@X=vBit&@>ea@$4L_1?4@ z`6D4n5PVBzSOT2o!o%(%AIW41*jsb;%OWj=nwg^HTGee8Tg0&m=*O1%pqrcllkuHH z)g#1rMlu<3MN}gCw}75g`1#r=#{`l)Qpi5~qC)i_!3(cFk)f=_{HfbG9U zI{kdRu0D@r1^^1%X>Yenj16g4I*ovX@C@s0G_7PZ?+R8+hGGRR?My*19Ik!5^M|HI zh$mGp7-I@#_5J-4A|PsklBot1kULuRH$q)e?M)-H8zpC8EX^xg6WX&0Z(BKhU!a;pWvGA^wPn0OPiR|XXc?48tbk?#YHZ})!LQfdizzxtlp7#hKwhlMEwX`i(#)uzmg9N3*EYqI#}s-DZ^YGW+9Ic31<{9_;}=l_L<*Ttdw6 z_#K!lk^P^16Y~+H_zkqcL(ZKXHQ-2qisnH;0Ex>&$}p!KNG*9e%J~+0OGw*>`C(FK zsnPa1BI~3nQ|PxDm1UOAUuqSN!YRyiW>kd5_iL);EF8n8w_K8CpJ}tAqR50;)kp_YZ&d>j2-=ClyX_pEgGbzsr%q6u02& zp6(w_4q>RZA)1dY$&PwhfvXC6dGnA}`wLEL3ut5qW8wgIh`43_?rY1sE$XK0x&0?v zXnOLpvIqavnqfcy76$0*!+MTIp;r%Y2UNMv6)q<$Yi}T8XlrXL_wu;?8V=g1hfoCA!y(%RH2jc+Y2j+Ty5_Kx2u$-I?@vE&RC{h$KY*}D zcc54@LW4Qpjs01vVBD9CH=1$#kIkxv>zz*}xCwuky%-X%*0NC=t-MO}RsM>jGkOu# z!$&>6I{ms*i9=aoXl4EAYXCScB+%QoU*QvgSm$uL3?kK%z?rJIy*=9jZ|yhVNwy}| zep%`_wvI~hPwZ9)Tsv6SyaP4U7qsVl>L@#E|3Fq6#wO`yj43cBuIrGha4p1 zSb|=bF#f&t_q2}+c~W)56_NYOoFTUvXv*$_hzV3}p&Ts(>bKxD2EK<97jt*_2FgBN zD_#)P1A-kvP@m-dXKo-MJalV5ciX<(le2OiWe(37tg?V~D9~5;G+*s7-s>_RWV-2) z_(E_+oC=!5JAc`cFhiirRq6gq*k@+t?OK4TXQUNp_9w$N-Tstaw&2aWqiRr*vM=Vf zw^orNFI#Biub}9mtgNhUQbvv-lS?EC8pN&_nBQg73D{qapz)CmCQHrB40DTd^1K2^ zL_4w;iXM~!LgL!xiOgqO$fK*+N4C+wQg~jo`9qUj)Wj2J`r1EVWn_dXmem8U0uom% z6LtPYxHrTgf1PFa?8Co|&)L}UWdfOLo>y^BuHrW*ERXY-4qo{He;psl zcEQ1pJ=L-g9GB=j!ZE;^)iuVy>D$2MybElTey=&+=>X{xM}TOSfwFmq}+}sC}RBt;mk7U1Ln#nn;l4hp&omR$i4K!!bJM zRSL!~Kp1(S9QWKqc&PTooGl{o&K^A^34@NWQ55To4`6VL<6B+gV%ZyB=`f6RfaxRUr*1wey=V> zlFY-tZ9*70OQ5LY2RW;Wf}5sfr~PXnemY3bil~i3P3{W1kqB-RJaN4oY``z=x$?EA_-vgUwQ+eABR5XSN5kK zRr)D11fd@bK~fhph+^^6=~V)ch^stEt0?_5@OFDLiB;gV#?b548==eNKGXX?2Mnm5 zJ{OfQSBNw7rGbpKP+Ey1aD`8KmN8qCkdq|sf3Sk{k38j5W!ibqbKk25a*Po683X9) zw@Od8Bo6!9_OA_48{E9GuoC$&7vVu<2sD3!f&VT*nI7J}=v8fwiNJa7&>FP-q47w~ zfTq3k0ows<|6g6WQ$Cj{$4%vQy0X#@&;#@n0qE4h>pZFXGD!d&=%!|ux+U=s%nUF~ zLRT-4Xau~i5VZzjJ2zt=Z*hPE$;>mY#{eSZYe0hw0TGdznYmW!P1uNOCB4elv#(t% zZ|UT~_75-EFh7|efizjwITs{I)(_H9*Vm{i}|F1UG8tH%c)qB7Y7UB0mnh(>JON z$-r~Paq1Nh1}L}wMD%skY||Vg@cvw_cN>hk$lbDE1{W$n6T!zPvt6uvDr&)Gl_{QppQzNNg|e8sd}<)3c%7ZEP^p|e&!KsM6J$p!UMu11LQeZ zlt1XweBg@UF!pKd=CAw02{joEG!}L&=cgM;=gpNd{th;H8olja`+fbIizTYh@neP5 zk`FZwJ3?dyGqhI~gZn}rK-YvHSItM=3}=W!p*0c?j|gOdZx>)gp}V*H{?|5At47qe_`9~LcL496}`q^UtsMA~hx zR(>(y$ZjcZlLHgAg#B+)4`)@1Ul(NO&`tYI@N^avSdwf zeBRp!TQ8%NdR0uc?PlDb!bd>YwmgE?erzM!^S6k0=+&J>fo4@igW9*S+bq&n0q~Ce zqW9bCDDekrmW_3O3RfW#<*WC1+I9%h*&3*9gAr_D1?FYoG?qG||KIAZVC5|;(l*H! zX%OG0@3?(k9U3d?do>VPpp|5$&S$6LFE^+>KWj>x&lNc8%{8ksU}tw(6?DeQpe^Mu z3oa1)_fVOTrmz)m>xpY4ByiD&b}rYgY+I!_ME|`UuTA5ZEoqE@eG82nMYGU+9X()p z8T;`PM?QtGa@`3WjgZ3Eq&|2P6hNGjG&=%Q3^;hbmY@L`z^U!91yGZCF9^YJ&VzQT zqn(IDmJWTkV6VCd3jJJ;k2a+WTdrAzPmn>xjmv=4MY)COIEr2|@x*u>Wqa6IiL!+k ziPF1Qse&yIIy?U2rX0DYqcWfBH|zZKv^M7--7T_aOg?Ts^i`jK4<;hAv%};VFab?^ z(N_}7fHQ$8kd@{z;nq6C>$pm*o;Tn**}|XUDopUh*URBHAcRfqm?j?0O3^36C)*Ny z7Dwc$Vf3TB#vrc~pODo@#TpIxARS2~`oIEk<%5gLc;c5lfzUiRBy~2x?==|ynh-@$ zt$Z!4#bY;R7uW1qea0ePYkcDBMST`<#Yc>|W7%N~!4M~bYhBR;L|m}08U>eLr5bQM za|UW2DMsAw5yW#`C?)nv*RCupMJg;*#NUwy^K}wqt`gHoeFos%fL5chhA5Mc_{A}x z@4eds(f03JR7W^Np?q@aba%g=SawC3~vG83;G9o*Jo4LesGv^8UxM^Xb8`H@QV=LvjwNCu>4sqoaFSrg6 z`xk4e2cXTFfm0dtJL$_Po*=zb(H4o5th3g{TsoCFRez5CSDPS8EeGA>(=U;PThmkb zxaq~iUB=u->-F>wSLoNhp`({U;fbXPUA z9u32EcFdEa<`S)Me6<)KQRyEN53UOy5)a1u=Qwr1kqISxLgd)9MKIYX^VnpnwapB} z<7p3Xlba{~kn%$(()4U;zS)>7zN*nhw>u46HdjUX2LpL`TA$4{OXj$TtBc4$yaB|F zUjWP*yZ}OJ;lP$uE&#~6{omc=MjE(~fE7I8)UtK46NnK2VTi^RZWionCBkdA8haGG zcEDr;&-3SxMupUbNd@D6OKY6bY+Pj%-^|VQQI#fa{Aa|}&vd%8+~N&nr@v(lF51-F zYw6Y&Nz?r%rnwt7Wg6DMSq0SV$l8uD!D9&~qax-Hvh3}E6-Ayln;0$i+X3ge?U!%B z;jtb=l}slVz7vXW*lZrpo!D%;T{ka5J~URP&RdRFJ4WB(AAO2$SR0gKkDC2?sjUD$ zXcI|)npH?=-Pmsv4_KF9io6M7{f*I^{|>wY`JAtiYl=l{>%=l(ViTwhh6xcy3F)z$ zA(l!EpeiD6EvbL5@P$$4*!^6O;Cn%_Tqgc~Tzw;K1^tpykkTdkKm|a=Ds;LT<*YIb}oWyjBo@Y8_H( z9s^^^r-2$sc?bSCT$WHiPF9}lvP95bB>@Xx_7cyn?PGcSFnw+oihrS|FvXnLhWbt-p_x1{|QXt>6@?%_*U)&-I+goiGwrWi!T6w8?SDs2s~3)m zmkq$npN=bpMc{IFE0xdgq_96^?1%1=r_KnrlyFeJ1L7WSYh_E-+Z}XL?@n@oEMR`* zoFAx+s%LFcK3@Hae<(qnb?~7Nd_l6@io&wG9Kx~=U6Pv*P?{f^H)2-Vt z`FE5nAGNi$&kw~YMk3dof)#@y@6N2_jD?wJ?lxcbyeos6wv9NNL>1RIJo{rv0I_HK z5=7=X?ly`Uan2OKlQbG(ZY=dGb;92(Qft%Z)!4VRw$0+or!_WClD8)ccvh2c0FO8K zr)IGtg}!b2necg*M_jC3-xvB^!(SVCHb?OG18*jn1Pl>JkFbK}F@s4c@JS>`Lv_)d zUhZa%{V&4aJDlr1{vR%tY!VV78nX9ZAw;NT&+NVTPBz(lhlA|wO&NV;CVM9{BYTVc z`S$(Z_x=0pzOU=J&UKwj$2s2b&+GMkK4!kmlily=Ka+F}e69-9*fQFiZ=> zm;QzG7pQ|=2$mM*cWX%qeBaqxrh&}rnlfL(`q~+WV!C@f@KVB^9&6_vo zXh~47^0#wY)tV3dTAT6~*0po!fSTlTMVKdlLy9uA&vz0I-zsVx!p`G9jf$WYnixLZ zcWYue*Pu5z0DA4EdVHvpX;`6nA2+?hNP`-ex;m522tP$y4Oy#Z$?#Y6cQ3`B89Okk z)D{U%a%&H{gst3^FMAHjxd&o|HL4J~+my~4Py3veU;2qx&{GqlV*Cj-yyo-t%2gH1 zK#sz@1yv9q4jv3TfSPPcjPZ5D3Gifb3@|EMEao~*RG2;XjClvEEM-Cm6&37xb37RQ z{%Dc3e)u0rS9$4l_ky*vH#=7hZ+3gwYt6+-{b0ln`C{nP96uL(&fT9mZ{VV9A%5`t z?plgxO!ctMQo%B#_l7cySK$(Fou^ls<{pnqU6@&6GQbJ<|4~NY(&JJwt%Wq(8YInj zV5r{*B|3{QgU|?)WvUCv(f^cz^iMr=yrC_ajBSs$<}9nQXP>htEf})6-$j%e5PA=R zAGdG!j{NSP%|5;-H+TdEqFo42VTb6FL8%@H*=k2co6ZZJw_vGBCywat?|;KcGAVpj z$Z35n83h9P)av+gQ)}Se^!j;hA2EHe~=_^((uzqM9Zp$n2Pj^CoE@?A@ADy5?Me0YS&U^Vh2p&MD3B_QBR0C&_exI=FW z#NSU($yYU*QHX>B=BNjpavI)*dc&ujzRAhCx}@kbbO@FNK^k=+q$Q=KoY@oV7uam% zilEDJS;-7^*@?+$obQkRZZM)aa~(-0Y?(K7nm{z;z=<9foi>OZ$B4`!{1S77iSh64 zBUmyU^cGj2=0fqE_kIfp?r7ArT$2dM!EM+WCT3S^a(l}WRF|}9S)3fRys<8&LE(RT zRKmqJ)~9GcZPjfzocGnYsaxLoll;p!8nW31|L-~|BKd@0g4dS7`u@OLCRUjt1-^0a zxmIhE-YopGUxN4lZyUj5PJG#uZN0Vw`Y3$v(K0%gJ_Y%-Z*)roM5J9`K-5IpiYTW) zWPJ8x6)=k&VLNueV+;Y@K~Jce;HYMoU2^b9n4>uv5!^MpA9;qrDNQ>c=Gx} zwZAQ?_YENg$~$JC&!$vy2O93mVR5SS9Hi$*)=g$eFQpw2I*-rhfsGmYU}ON0R5>a> z=8SB3s-koqqtI_O(k}<0rv>`eYw!!46NgIL5c3V+u9wS$N? zcyP=P3hV0X9I0MDtk>9=2C9NGqyHw>J<~spJbRxDwK<-t71r?@o6Z(7r7-HSq-Hxl zks5D&OXpl3%>0;r?ZvDUfVj9NYkK`G3Mok)+FkE9veroA=Ohl%w{J>ebl?+&P_*uJ zNyMcq6rMBj7sSlPo|$~JVcq4ZF5(WdmhekiOOp-U z1={$!vr<>ju>L0vjqE?O-+-@naQNPfhe=?WiaE=V|b>be>Qy*C^_3 zs-JK;c;BUC5V8VOWB8L&H=}YyS)@Br#NGSry`Z2-+)S; zA&%8cn$9u9esAUVrPm|EW8p#Z!Ga8(gO;(AQ#?eQ`ofd7YCCL!p626j=#U|f5Q5Qq z3Lc$+8JjRPwbnH67js4E0OTUn)V&3|T292pK9bg6cQ9doycH-i<@>s76Dhf@j}X}~ zeYv)JT5{fS8BiiJ(CikZ+~&BzC~YBE!8EkGVPqfvQ@fhkG>m&eApOXy#;|VSNR<9d zh?Qx{#H1Q(WQnbVyAIesqxgw+L3!qzB>}pss%}@>=BWuNY43?Ved2ayI@XNjp9Hht zfdXx75*Ma%$W>4Z2>s_e!}hL$0G=sh?;GvZjy~fbns=IRwQ`mLZ6w|fM^*Lnk=p~5 zT0B8)vLAF~3!g-qR&wiP<O!zWcPz;{r#o zV)ePcxPG;5b-5Vp=&@KcMRhLq+o5nwA!RI=4`B}SDIwgBD*}qRSpIhIfQ{wlZ$<`= zYQY{t9CSoc$#*zqZ>#0p0oWZ*glmuf5e!1l!102p>@>Pf0@x7~UaAV1Zji>)*$ZN5 zeoTJ)_=}YA&MjcS|Fuk>3nJ1t`doL!edD%w%G`>uU7d zW4ZeZ>xXZ4k;Iw~!6QXS^nbdn>nx@;;qa3S7-OXt~{)# z&fEO=vYDf4M4p#JMNrPv*T#sGSCrA^X=_pzErhnU;w&cnvDS#E6K!oTrrX|0Jn!~@?!bgHYj-b)yCziEY@;Ah63}T^X+(NBO5`Wq2?=tfLr7Y-$ zBzq$@Pm^YzY2OWwLR*ePxgsLM!<`tv)N3KHR}TYRgpT4_h>RXq*|n$gA114N5n1kE z=@k}e#fAEVdSP_L=?v8!2wM^jMP1rD{b?!$~^nR_cVO}M}!q!F=Y3L7}73d zY##XCrSV~VXD^pk6Z^2Ko%x4D*cDwgzq`vA#5K<)AAOI`{ST`)Ts$-Jp^E0yQQWh1 zjIuAhy0bQ?_A3Rhf9rNP8b@%H&BDGsax3KRsJ9+Z-c+3U6IKpQAeWIX*td&A1aLSqIFC3#hGn{O@SjhCUE%fF2QKV%r8QZ-V>Z zx=$Rr7#8yTCPq06`6uP&u-5DPPNcgXmW@4pMK@_y%cJ>D?m^~ zOKe%rp=mnn`S*-7OvJMZRS698l8CGTdF6`S^}vnsw5_Bm#CRl9HVqO zy@cqZjaG^RpLlaB8g4qTt*tQ@xkJ_~!$P#&8M48o5*G6HmqSTdDf^>l3Ipk)@9FG? z7wAngid@cMc4J2#^)aXtT}Fa~;hUQ-6`#9DLqkMJtU~n}AO9l!31L9aZb0j3@8+eK z2szpsl#?6&JYsRh)YsQP161H|xJC$Vj5Jd0zSY6vWjj5q11u&&_H5wTzkucUR(af0 zql?|se$D?XDLyX*8SA_blQymI8W#_+fYP;=iMw30#niQuzM#cAb$7@z z&712fNTbz&KI!^O##BD<$EkM{ACe?Te@-2XFZO4peY7`OO?fZP!Zh#4tUI1kVdiMc zLN(>!N3Rc0ZhB-rFpwZ@(*Hv#W<``juSBnCJ76>L!zRvmLwS2eVD_T=eI$ii|7f*{ z&fvruBUAWjHFtu|lGl8_0<}HC_d`XDG9{@9%0*yOs6n`M5w7ihx>2)>oG^%$lH7@A znhDA@1n~4rj3&8(!lJ#bp+;0hCBW=jvdkA0mi&hB8V}Srd-#NIgMl*%8T}g94O!MY z?IgUTUEUMu`9r^0F{T`BQ+2BGRr7}2g@XVCX)&PrcE3DfY1;d#96DEQ()<_s&k#BK zijTy2H;AVIDT8G|7Vs6BO`u|q#$r-?y8++(s7>%}tIcD}mudmRnzeQ(l9KMu98}kA z17PGfZRousPBk&IyyEA;-wS7)yl{rju0VG=V*PQ2cY8GIFr-88EaGBYq)TF;!;>a^;3-XDb~L*(7LT5NDCAGpunx7Bb+#eg_pM ze0;uXGT8;tn9>TYi>&p1d0u{rC$<=+B9JJCMLlN9yK74x$T3LRM`Wy0V)VZnCR|-I zVX@``s3O4BtuAz|oVtmkGrz`HoW9r~9M3e!$(f`7Rv}v~j9y7eSDH!hEuq5$aP-nl z5acq?g?kMrv8pcb8^?{Ew_JT*S-PsMs(@{v7!7)md-v{X-eYVusU1r3bzaS_J81FU zuPX_=77IXdiqIgB>ddOmWAzp*re}GaplC9>!v$k$!#oV6h?YQW0Mn5cc+ifV1nn-^ zrEvR1F3#}6N<<(Z37P;K{SbGirsw|fq}ztuwDcq(aJ46^g~x?0Ck!)N0Xz;K$L?~O zuYQ5yR}hprtNX92CBp0d2p9}1#+BcdyAq0UbB~N@+4nBOy60J20$pYd@<&$@BtF}8 zB3=WVCu8@6v7j&Pd^8_;VWxsifxZ2HAY>6sQ7vgIS+kxXDbs4=rcQqs87RT17a$ez>_kl6uRM6j9v4R0EqYYVKo*Qxn8Gm!GP z)~4GXq&Mg@V`#={UB)~BFEraOsV1Buqic|zzySjrzSq_@O4o%bxr94-fyanK#^>5i zr<2#!65vVl>9M-!65NM%9c`bAXp92mXYjd0F4~u+{=tvJWm6b>R#JtAuC!L)%t;bV z?c+Fm@g52Ik3Idz`(jOQ$@zG?`+ zk1G~b{7Jd+Y_;@!wO&^imVS5@AlSj&BZ11QftEv{Q|$xI-K&Aq)rA4%|W!@Pm30=nkiPN#6N5*qxEY+&l@gh`JX$ z7ww7-d3+P9E$vr+WGcJQZKS^yJl~F>g(O7|jzPClXsF73#k*#H0of=SAn^mRw9&YC zGgiR^-r7avwTu6@ji@vbHd{k~wiFmwWak^H{4iSpZ*v4P5Kiyg{pr4z^s~J2A}M!# zBHbaYJadH>Yjnd3!=5jj?c%S~e_~5XnA=XI0m?1U9}=@8wGjqZ`V<;kXioTd68?Wx z$}9Ue{saOa$dRNmorZuz#;B4AjwCBxHX~{3ExuSe5oX0ehK6{^ zqpIhkhLBoEk?IzwvnpyA;bqh}`ud>!%|zNw_O+{BeN~LsVbYT`iIqm5%h|%()znim z>i(z|Pdjy18~ozBm;pOmA`JO{W1asWE?~h;*Zz5!T_|~`Y_1_I3S59%+o|w^$C6vs z`hejlCq;aA2&o_Q+E?+8V<>HmHWr+}o8n|2)}hy~$3}Tri@MC*y6nG2hG66%7`ZM_ zftrI+Mo8pr4&S-)r3bvseQP=wZrVT&5=Nj#J^SDV1CQbnv6Plr5OMM_RRY!28bvMn9{AbK z{l}7N|Mf1Wd!HX07mQmD!jFi0ZqT7>IjLRadFXb&_wphZvca~pgez*X z(3w{G@vv8|{c%|LzJwKJpXM&HrQ&4l@z&QI8&Unir=j4QdGYUJecCmZ(OB*9;-%7l zvva@G!6O4&(biv4%cyO9yG(`n(dCx&*xMxaB%W<@(Eg z^nG`?^)eI*6DcuXU>*2n5vfs}p^o=blCj-dam`MVu~29>WLln1AaY@3n;SRI*jVK* zZAW(d60CsZe_oeo6e3tf03Yo7j2ZY=k9qRhUoZ@RFNk*s>}DG|!l3iQS*a94gb^4t z)p0}(X{cG)J$BXUH(*CXbe1c?wp$S2a7&4v7;#@;<^&}rCEd_+4h%z#_q&zOMri_7 zPVP%p*MCfux&kY-cG;4i#8F=u}I82sLnK1u1 zDU+h-vfp_0Ukpl9+`k96om-CVCjJ}fJ<)oCMi%h5!`zSU%&WDGsn=;Ccw;3%^T$eE z=BN7lDXnYW-fB33T>c#Tgnd&hAlkWp;+a^!esbDm%TTt*y;?6Hi6KgJMRDV2qfKu9 zDv?sS^ug`Qwr5ojZ7pNpF)Ow6WY1^JVVAjeDHDB=OdLu;uRWCryKHWml^s(yLZccW>-tnMu6mQLcyH|#e90VkM9~PP(`>A#z4_cyB z4J7s#`?hU~mIBUy7l3h?%L-*qKCD#;L5?oCIR@e^wZg0#YO0$S9iuWhrcR<~0JIq?VE?f8OG{Zr* zkv(vqB$V$_$4lP1>wUtf7D$q3I!u5)*y0B-maj%$XAkt%C4GX;SU9UeqntHOV*2Q` z!ijumKk) zhSGWc_=J3w7_&epFWi{UN@*QwRA((@i(!_=Q6zZ zzpBI&>vnX_!dhvldocCBGPwXLqvI*4Ej*K)CP)4Qd!3<4?O64yknB67m3$+WC!cWQ z!n=|l#TkeA9&He?MYr$YI`bRY|9Zw3G!ymU!2oWKMHkWc0@EFOIp6xh@OxigMkaEL zMs^m^Yw>?YmGFF?-?Um^PQ(5=?aZ#xv)YzqHX~t&X^a4wBTRsWFPcl^zEb8H=z;>Vef{g+Y(< zMag|k>Ls-zg~gtpp5%nl8B^hdF?~)lT%5Kq)f=v+2bPo7SM7;eR0{r0^-v_ zQl0$YAUuEjw=Kw#bsRtwL)@~X4vc#!?=uuOTGACnt5mzw#V}eK=KB_a z&dftnF(+AAvj=lC6l19obY4H2!}qQ=MN0xGv9Po`x#gs=D8b(-KBB1~jpC<5(`X}MuH3GMazcnUM)M=)V)rnz?$3V7(7sV2_;n@UV4f_w|<>)-0f4WMkA5N|e z{`>Ix#+AX6H?QtO8v~B9LSB|kk9y$tZ(1|j;6D%if&oz$x6dcc9NRK0_^sGlq~Jua zJ27^&b$j?*P}I}Wu#(Zm5zg+$_{9>`=42>X<1+#Sz!;{Lh;Hc|8!hAeZ@@KpGXf<}>`YG{QuU_bSPCQ_XZBphj-utBQl87~C;xf= z>yN`qV%9R-AuU!+lI;to)EK_LMWCs}X76ft6JI}<WN8we_hwd%13k>n}VH z{M+UnXQFeNG^!a-TpGTNyEVX&gov8jA#Cc8;E3yoQ4`KjiuA}ni{wTl(8@zX%o&jS z??SMaeFusJ616VAHq7iZQ+l6m6W?HqQvmI#*XE&D1cIH)k521eT8;%_Xm}VW)60PC zf#|H@-nK=aYC&i#NYMadA>ov3t4`sl-=O#-O`FzUqSde@R86=+u>&;xok~S2 zD0U$XE^id27w1zHJrtcMLY}oJ=K;)vL(c=Jy~S94%4Ahhgqj+*Tv{)wl|)= zY1q0k|8GWciz9OCc*Wx7&Um~?9ZtWyu>?zz{O9AH zS4hYDBlqhn2D06svr9&^F;4#Xu{>7EhhK?o=u;w=j497skxBF=44 zcM_rCvorXr52a$gbHi3uyX+{e>m$y6bP3(t$)k+{@G-un>awvj)}PTgUk4me#>H{Q zE}+&?2-UVA4dFM@YkVF@tYH01If7?jQE-1LWoGX&qP)T}B&a~5k%C;EHYPeoiZ#-n z=A!23pjP5+FHd922b2)D0Nr-#;Vsgkr`-gSHhu8QV^eogo- zVz8Y}jECwAAp&=^I>c~H;&}!JPSS~|Mj&(ahN_G^g6FBrf_eBFgL^C45zHq?Zq+X{ z($kmDcKer+z6H*Qb%`Bl$(PYMFm*#J&P5wtUTXO-E(DFRhTZJ@q8)aK786N3&UG|- zf(ojQh3QLXPLMX;g$DU@p0d4}oXaQh3pwELfDj*SAW{4fqbQyV8_KI6|`oF5(^ z)U@ETmhu~2PQFYj@jCME4RElxE-WlGZb;`SK!H|~X`L9O1X@xXYJU48^W3LJN#=916?gkmd%z-^@Q`vs zUad`z$@W>-C#BW!?%&`QEp` zs4Y-IrgVA}&nrC32=qSE<#g7FCEh4kO)NG`d`!kK+S>k{@W<(ZSb38*abUhVf4<9p>S24^iIRKlXh7x!g0_u%$z8UQt8pb_FCAP=d-ESf;6AyO*$;cnDMN zyTLRBVh5H6bt@6zK{7P<>b*p+t)DA6^=6oLiS2wK{zJwninJ} zku-WaI1HKkGW%(STdFamXN<1h;AFh>-XTk3PWC^m1<1z9GpxfdPMM~^zq0Kx^o(>T z{-@qkCR?=T=+UA(lbrcf>Hm2-U@>IeB*$qBRKeGC`RA84|14S{__5PRclA5u3mY_U zMnNJg?N8Orc`HVqM6*!jsQ3T;T)^3b)v?McRdMLx$5AuSCHC)}6Y~@0UuQj~1Qc{k#ml%$vxv9DDsWA*7W1qe;YN=XP;^daa_gLl2jR z--ZcvpMcm*RGdX8LwST z7e4#bXSXEGj$T=?L&zy-e$F_F%NyHd;ha`?EQwuNH~Qj9FNF-3$mAm!WL6&e2PlwR zw-lqVH*S=nC^fcJSxF5}4>q)zMpaNjNhB@f(RvO{U1lxx;K)*-874pBNBAE89>z96!3$TdFfb?;LU z?LqKjNI2Zog>lNtGxVA`atT{<-SLtVXnt$u`i=!})sGk(h`YPs7z@Y20O{Ob&ekF5^B&KCRtLvoPI^@}l zUqFq`8x)ruIBd9qnNj8WgCh67+))>6%>5Pk`y2A zEW|-S#Tb8`r5@tP9io!36-=-uM z@86{n1nY)wPe*GK$TAe^{S`uFAN!gn_l@$S)En0RB6%;TQG=O4;(CPwA@6y z>V-5dALg&mVW8(XK2}B2t}v# zhEZcX|D|V0srSFNCpkYHr!i&Ck^@YWk#`6<51Sb`m4c?WnHp=VnrWw$ikKQkYzVI0 zpQaLH%(c(T(k?A6F-u$tJ@{G)r=ybnUu6|kM#xCzv2)|0$D~_Ji0cHEz10F`ib&q$ zoDs?&oc!UxCY@`r=L)JS8FlhPY2pGYzm_O&yZDj@P;rL2n6b?nqmuQtnTxVYP&N*_-$Cs@*l>r+}KuD9PQJI(aXA-b>LF9nwPzz5oIl9 z4v|K+#oW&qEQ^QY4+HAi4qvGYp z8WoPl5mV=n$DSrDs$pdDYADu517phENYD2+77POmRy4sICU*Cp_>DbX9Y#Cu@d_FD zRb;qp(cojr^||;U?GHaXwsrj8Y1W5hIEtnu1ES7nq(b`TB6KXFoF*4fo8;yM+f5&;n*2L^=te5pkw(uzwOevX2?>L>cBsJ_;d+k<|xmBIyI9k6$0e87I;(M#n?BR zGX-%C0P7rs1f%ZPHb7*>8uxd>dTbw_!XMF+HSh#^SD<}iJi65LXH0rK;(xka2a)%4 z5kNkUI2pqkdq`ive@!2^%e>>*m;n1yiETlI2c+QayQI3 zYpJ}?Y>}j8fwi4u4=CrYwis^(2197m{T94GKC-52!uKYWR2hR-=+2uI#S9Ov&nSBZ+gnm| zy;~A4vQqh8h@+(R-w2j*yi&R;-tQXUXVSMur7#HouHKU?3Oj2LOvooQNP2ACuaG;2 zrJFo9=CVcjNczJ=$|jMG`=j5|?&m1{^$*_&$E1^+JmR7JKc`+m2KO8%W^{kz9ms&! z{n<;p6E7Xjm-lvsnC_@+bP=m)euZ`PYYETd7|E1nRFFU#N!Gwh!H7U790o|3@g-x3 z91wesz>Y4G2=7EiFrHK)+_U4xVNEJbMG5PZHP)gG+izJrnQEq52SL|G%(4bKam4P2 ze-~l<$r~`??Ow;hc2=H43JSYs&o^e7Cp?`W#309$k5gQabK^V(a=^*T$?;{%=0IOM z+;;)GA7fiN<$(6CFUel3-%WK&*>Re*>n`6q_PC7ig)zIiB6Zkgvj;B7ixebW_UW#C zWHoU0ot;5%2Hs%QILIqjUh882hOQwHNypSJ@FX?3XAVo(Y;TQGn9fx3I zt3vL!q!l>D|BKr!%uy~IzzEY72X(CF`#Dw&eIUn-Zy5;Hq_ z=Z^3^n*Bw;4|}-v;~SIE#YVi!H#Sur_Q5pP9tj=Z5*E}Bp&Pm zExQ$Dxi|53_1d4$4F*=AWw?ua9Uf$I+xzr1f5SQbY$i9DJf~W`e_QFGeA;tqNr36c zk~Yik6*(C$R5aqO37Q*83SAj$emucc>b#v|8c?SCnv?7i$(r|QFzm4JXHSY}?f9FW z-2S}K>4GHxF30h3M1{C}J`H;Nfe0@LCEcEC#JYiS;Ag(*iXc2j4wK~3NHhj3o9RIq|;Jb19 z$3&s$#cr|~gXH&#lI|1EhC2g_O@7vl;WaDWYpEv6%Rw^kW1ZHT#DJ}9)iECx+V$JT z&D~CaxKJ?{Pocl>ty1@jjFbw$HX58p52Q88AAIcZW%%?+X6w{ZMBhUHgC*xWe;w@A zvbP$nbaRY5bMk%&rbF-#`HX7UzAI~<&|mb1KVR}r@9JV-$<@lcgv-bPF3bN!STS;| z@hE6QTJHoy_cvkRda$T4Z}hA)t%lmdcLXkrZfmvpXPl`|KqLEs$V+H!Q*I-<)oBrD zWZ2VWdAZ~9r^ve4rLg&W4qBV3ktMT|*2mAy?Gqb}Xh%8GXQSuUF{n+JI7sHVYj6Kk zr;n@~a=SSWnW{*@Kk|hb+b;V_`@nwaI2`~5NF8@{dWMFE-j{BSR|F!_e{0oDz+NxE zM}7PDt==r*9MTEmAh0Q9e0*G@EfEos@>r7IMi6b@@Px*NN0)6_X`3Fj5EJ!u*SEHU zm=CIzOg2p%wVWqXYD7MQg7hL1u@`As$INanBc~ctT(F$>%voO8dg-?L0D0&Q_5}J# zW^~{FWpO%O^V{ensSO=*{C^+e$)$Ls>Js~ZTI=M;k{xuiutO2R8%$Fx}st>cuaqby) z=qOy@w`t)SRBu<|0)c-kN-k5RQ*_X?-_L2wh314<-cG}evSNJH(%6m5V zr`KpVOorQBUE*i?Bh{GjVqKM zTcXRG8($U*Ve@(}6AF?UwVPAsbvgL^>fbFEaTc zvSb9i>(CoGuGb~SejizFDL~}_mgt|-wA<&0Eg6WvrleB*2dAYduY8cUfYZ8i;jpk} zuxkFnX3xaYXwD4h(ek&u&^;eJnZf z?@g#{;6oMZFTy*8Kd^w8^_^lYSDMW^NX9nAt?F)HQ5=WC<4x3^pxbim54p>4>afmq z7vCX|^+*nh`xM8qz5`& z(3XTPo`}mh8B36B;}aT8`h|NfHV^N0x2&F$%nCFS-4PK!lNZ==dYbS0WU((QO6rik z_4)(Zw-b+#;rMw*Ia&FcU2SJvjn$zs>bg%<@lnjYd9z71+nPRe@BhN}@L6~o9-Wpa zABWCmRv`8*x!pG*DD8epUQlE!>ATSA(7Pq42i)CH$US7}x)#0{-2K0KZxf}nwENEu z?lX5D2B0a9?i!+6gxnS_hOPc(XC9V`E!}T<^C-2B9m_ue5wNu4BH$_#Z+HrC5IqG8 ziI`?+PCN$&aeTkXg$JT%HOznvj>~pGTH$R>O0T0iSDgyW6i`_ZLQW;XGC#ZJ0Dr#R zz(7Ln{B;!2SR~09(KXHtTJny7g7nV4-Bi#)Yq1=HM#-OOh!coyO=8;culF1K>YDK`(^a> zPq$LO>JhVdyfx3bH#YK0D_AwIP8w<%ovq%N9Zpt7kxK5~TvGS%8M$!Qu#p_Ss-Zp> zMER%wQFeo6*Rn>l*ph-nMBrGbmEMB}wqE|6by50`Kgt@rSi?Jv6;hiWR5Ao|>ooBb zzB~kc=kNVZ{w#jF()*r^Wz6%2ItyPH!zB8i)-pjhPZy|903;+Bb$X~rRw!p0g4_4S=QuB@71D=j_ipuIkO3=y<5E9jz|94`K2bd9J$)3jfZgC7=?zTJbt3 zk?(m9r+8B`$|4qKNWg+tWR zsan^I{O;awTqRigJ<^OL&Xd{|9&h>PzMQt*YCI%h}q-J&IfSx zm@lS`#!4M&578$!4t6!zM#kz~O?$ClPBZuanP7x0P5eq0RR@k&?oPosy)i0u~}> zB;vnLO}5|Xv_L|s1Cl5K*ss9^z9Dkr@R^e(b^OWeCsb+#eM4OMSFLe0l7@QA5nr>$ zpOSHIxq^zo{65N#BBB9{C4bhGI94h3Vz+6x!MfA0x-0yclG3ap!v*V!DyAXyp{7Zy z$22*u;zam@ezZhiGw#u{mQzeu3_W^}>!9!lO=~nb`o0VrtZ%PAezF@j>uU*D+-M22F0LY~Z#V7n5otBJ^N5N4vGKBrM*CxqzYW8nUKKLHT zpVE~Rx$@X~-Q&-;dWv%&sukVTa^^%k2TE+ja9Rx5zatVk6oC0!Y6p5sL{}B^V;Q=2 ztdT^x@5p%$alF;85meA{f)Izxpw_7>Tss9fCH6VAZn_PWYB>%oH7+t)+d6s{zAhcT zBS;c~YS6Y~B+=MIQotpsW@Q(8WH8F;HyZQ=UaAVbl(clWdjHYWj(>{_c1R^Iyhd#|s{?0m(^I2eywkr7R;$Tt(y!S*xb3e|{i1oI zj*n^{y+gxi%YJ3w%t!Cw{;(Zc#p`cH33Q7Qte#tU5N;@@A?UGlruza~13^kuIq|Sv zvPU7IQMPt_HJPgNHA4*-k4$9@`_o)ZvXds=v+a(yB;3L&v%99w;M4sb<@sraW0E^x zo+>jLKFQ1ueymL{+qVXz+exXP4RsJKov=Ej@zVV+?h2Q%TTn`V-=(LHyDpYwx6LMV zld1zlVvF`0X1HaFKy}pa2g5)0@`=21ZLbe*RX%91Bx189 zYH8*Zz8x;5@q5$P(5A+MOLq7^3Wv`lMcdOWLrTBv|@#U$rk(%dBB(YfywT|?w(&+$?3i)qE`Y1ab zp!Ed0QwlA{7wz^A>@=HS(WlK{cfk2Pj295j?$bo}!mY5&BI0HN3&EqRMb0Z`arZjb zsm&HS!z6E1yv|J&XkYM|#FVNbklUb%9t*bW{8L3ho>Wd8w}a!~bDPLMkOWs(a$W!f zRde{rg!ehsY#NlIx`vFOWsh~~Ess@z*L@^mklo3uq^sJ}tU~Hj59L?JMGb1%X-r+o z=LDmY7`Msm36>O+Eo)-hpX*JAC`j>EWUh{N(8v9pkHr@WxP>$S;)DKCPTC(Cc#06O zl66IK9fr&-nPMBGw5%@MSXzt;UE0*UKX{z)8$}A&aFfBBnfM4$u!n*Ee-zs|Sut)? zOJdDFrt<9`sBd~P&~ddy-EwHOiMa2vTZ|QC@hMzB7>S@(+4${*PX&XbKAGW zEgvOM$vrHiko&(Fdkdg0+wJ{V1(B3S5JZqp=?39Tr?fPZ(kMuWlG5EB0@9L7iL|sx zgP=%DgOqvJ!`}aMe&?M3oH=`R_RKpv@7_GmbFX!;bzPrJaLP}#hkJ^*?e3(V0J{|% zYY_$ZEEB*6l+j2#vv}WHe6&aZ$d4AnVWu)B*yyO0E{t+p{xBQ=}$+;+Oidyy6We+;?52lLt|Y3C^B z&^&w*);O1O?f7x-n`XnHU&8hJ%vSMC${?VSBm)73x!ndj7BQWyWGcYllTd&R&)%qHl3Ub8au(Tk?ebO5atYpDrabyz5|Kr3zLzxu%6 zYItxMD5pJ_FeGwIiv`EGr zrnq;JOPA4`nT4QC9!YwEB=@(~Ab$y}_PgZhp&$NH2O69`Phk~mLQ&1@YBMJ=K_g?M z?q)!FyZ&~E{Vz;A*m)9F2lGB)&wGLS5%>M%4y#vkep$qXK{D@P*?Ycg2lG+|^x<2D z=$Ou*xl74@^XMeR-;Wt9AFe&r9<5JS{Z-ZT*mK7D3IPi93FyN)ex88{ZB~co z2(&MHlRrcIySuv!VoSztVQt#Dw(5UAXu5`w$P4^0-*8pBNqEX(Wf9&&}Dx?Ea{j#ybHe-~188~ts{m}>Sh zL8YNd3$FPS7Gvdb#^%hQe(=a-7=M~^^okQa@Oqq=>QAnIy-bBWXXG{W&gn(f7J6YJ z{2DGZ?6DnaqMIlier(;QOCxwh4x(%j8r4iV1;#(YdgP=ny`RqU>vWaHP=iygfz^z< zvm$l<1=k-l^Rdcc3qx*g)8*9k>}agYQm2yK!# zR}nHQr7;tqaIz*bztTUnEY<4iEqGPCGRaOoFRva+Fqv@n^xwe^&P)57m*r-$B^Ro+ zv~5|-6rz=QvlQbgwzwp&$n2nV`vntlr*@|IQD4RF=!n!y(j4A@s;dgqm+uob+%>N+ zh2#=wz-~MF&*zN#MK&~T+Hu`o?H!T@VfS|@o|VI*>p}VdDu>qt+Z1SA-*X670Xt+z zf7E-bAod@kF@bo)g^bi+NX&@N7A51Po!eNSxUtcXUq7eYN5HRdz$E%{6NzC()Me~_ zk2AB7%d!0@36_s;ev2z3LqoU_)IbaNRdzCNRbp7tzvzOgOAW}a&1>{-zcC7HfGL(F zjbxJ%MvBB^;-Es~$%$QQn(-z>gVN>D+jEW|<~@ueJItS=^H3J6=FPKWagv_89=h1Y zDZP)tleSL{HC~W(D@af}cJi2Vx;0h4S{+ce6>LEazo5noKMQs-OYm-ZK5O%W_jboJ z4q^Bneaf6Fjo6)~Aw1hSxniN5CW1A8reNDyM)yaUyp72UPuYIHhQ_hSIe3NyCb63L zwJKCJ99KFT$GG>u^}MG+WxW+ZC9Oz?CSn{^mTldnC1sR}!)i!mwK}x?keDxbl`;~9 zNb0s?jj7McXp0#RHWwPa=2d)~V@V93TPvRh4rn4yBv>{hLSgxXuXG?3-U6X;3!p`h z(yR#Nm%MMrEWasZP1G2N_ra|sureIA#3Y&J-t(nneiSdU2|L~R-CZLp-LeHazw!15 z*9HTp*7~r?4O;Ndm3AMxq0)HLdG@1D8gTJc%73le8i=kvs?B0yp`q&)!5U=dgJol8 zZa~ydy?ttpj~Ba3*Ds{l%k7w0Fd(!vB!Y~f<}$h?FGf(+VxA>#zngwsjtJ(xremyw zh=p094RvP=wY6C8kq;y^fBDg^>EjmJx=5(u-qjltJ`VE|Yi;Zp!;gb`*BL*cU2X~h z=(niSh3Bgd1mMkZrzflL^8Ya=-vlYAQxoRCc6JdU3j1fWe+EL3uu(xvLDt)a+N@82 zr@`szJ6&aL|8wyjc(jf_rM^q@ZhQCD9+`rlL$n8Yk&Oq7Bk<&`Q$q;Yre^JaFZ%2O zXv1mA=gTkZH(&fkui;aJ0uEXCJ^HqXbhc3p!-O`{riXALf)Lo0vIt~@CS&+e8x`HIu~t8La? zb*0WP4mpmg7;^mrHFpWD3wdsIY73Z{1}n6Zc22DNYKKYbmRd1uGnP)dGS8#>-UkK5 z*Xv ztiT}fIcKHDg8`T?amfX8JIn)Z5yp*D$#QI2={3dsUm48%K=v=AXjbux)qGuY>zR;i z717eA@5^%h+Cfy~E8F@5B+&HMk#_z-Rolz4$L@GCVPv-Mg%AvSO_rN9(8OWErePl+k$9AwEUw@C=f!Ss zrjziX?YoZs+&ig5Skk+WZT)vvL?>IS@pC!pVFZ*D~vSIZD&`&dNidJrP#*c1_K z*5hhua}bOi*oL);af^FXW9u=^BS&zM+P>f}DJf}!d=>-^%mmo%+rp-TxM;@H>)Dc= z4M93eX7i18uJ4vYDbJ9q04^~*L=}Ly5TosIJ)?M7^mziLPu(TJ%=`Fyp(3vIG2^Eg z58P5I*DIBz-6UP*;`cX|XcA9eI;0NTj%I_qW(>!E!B*Hz@l?PnV2aY^aUp5sxHvTo z8N;Kud-=AKYOlel$k1_^O?$9zh`Q~@GiE8inn*o6{6)!^UotraUJh&IYt0H2b!tH8 z%e;2oLV$Hf7j51NQWP{0Rk*n!>@&KbmruBmu*$!ha)l9OKukbDK+f`xe|*LbXcv}0&)mD0 z)&IG*7C^y<{AdD)6zm|H>I2B>;ATng2;hMP_YUl<+foqaK0MDyJ%a6i&OqVLxm|==Uewf?!MG!IG84 zy`L`g&+Y*e!N^eH{e=<$FhmbhXYIR$v>1Xn&{W4g_dBj#4I83GrHHVEk+lGLT}+bLHkM8i1U~PZ*Mu?DZLr#gh1ga>ftC6zuQgJ{vEY0HWbi*hI2-bQh za348YMOE@?`6BaEb_;W-iN+r^X(=ZI1IvAG9=up%w%g@Dh|; zk9kGCzC~AD6jr*zx=7boo-;c*kRHVERSV&#Y>{FVE+@FMwUuW@;j(o6V8r1 z+Vffn-EMu@etZ2BN6Uhp0praV2J5~xfv|?%oLb918mou84kM+iIXiMTtQ{QMU~-I% z;)7j}_DsRnCA=f*>Pfp+2*g@L{CE-f1QAgox43xQ*Mm20JB~-q^33>U!NDa09wCUEdW|d?BDq)Q|chR zHX!EUdk1;P#NoK60apdt^iEu6y8umRT?u>+muX{<12`pez$rRtYyVI`MN3#~`uOS& z`J7edys7g&|LySNVp~Lym$Wvbh83+9Q}IBV9wCF#Purq8qYLIXHs~)s?>C5`y&sq?z;K>&D-L8bT*wJ zL`ErnS8HIQk$P|Syr%KV-3cE*+5}=IH1%pSX4rUgl)$kvMhql2f-Qfg!cM zK6}e93^vg*2^rKPgy=YfxvWM%-=!WN#9WTziYaBse%YUxuRu+i*j&a_sPsJ4^lG@x z6I3_Y&rpClV1ytD=wXXwG^|R?c<{dSz0!JJM`d+e>P~k(*vUz98$VI{#@ra${+-b} zN>-RuppsLOL0kQa@0IAzhs#&c!+E6Q!lHf!eATrYbbmb_avBmjDKM?J_@(WsLw=%F zSNg5dP2Zh^-WTBIadN=}*DnchgaHnGEeDc5%`?_otNmbWxrPYYkw8DD?5Cdsr>Vh5 zc!6?8@fmrzaHfc zso>D2ytKUuGL2SP$!Fi_g+v5+t)9cmiq$ ziJSfn?6H;1FTWWZ61&Pqo^J$eqxIdA^kcp~xZG)w#r0&ix*E21fc*Y;>@ED~$GGtEgq9*a*5xOG9W)`R0AGnwP|J3L+h0~B+r&NnQPvX-pbbvz zx>3D*_U|u4`w2+l?Rp0F!;dj7ug|;fn$0h&xvVe*>uHPHoG?8*WI{yLM+QkgV$|zs zKW9h%1=`Ou+xr1YRP6$Sj2iIOTBZJ~FADCX~r zo4Tph$HLBa(ftbCY+T5ifcV@l`KoYiTHOoA7GyFQ!5$3f9UL)@^G?g~+i~ABqaQI@i&+Ro$NB?6+n3 z{&(I>+YI<42zhKDX@JJ>-}77Cn&0V<71yx&*K#YtyO%eo1N1mnNOaobL5v4NVHNf` zNG}eb-dgRYbY4NuJ4DRP#zbAP7gHI>f?-Ji>0Ml&Sq7O&wf}E|;;9VwWP-J?n^rz+ z%QmDdu0<$oAI^7MDXXCvGY<#*6E^~4#EKo~gWNjQO)d1E24Mzkp?4y-xBu@+Gv22? zXWlq5>jw^l1!55dD)`vUraq;qVQM*taCNCM zsN}BQ``g(aT5RuKF78)GIx-Z`q#94fh{tY~prK`%)@^?ca|@o;$5o6ZQk)H`E6Cr6 z2{RA$EbuD8fH_GdIQIGYl$e4isi1TNzJJUEVb`VQF8`Scb+=l(h^&kvRKKmflf_cc z{U{QR1ka~UB!akEG)4A;SKRM)>cBNdaGm1x>-Vf`-H+IhI2VNFS+Xdt$JPSd5WP!S z^cM2z&D1@@zrZ*=#hd5CMLbgwG55gXtxE6A4EY6rF#F4Sw!gEIDX`%9|11Vqi5-{T z-#KNh$jG`1daI`Z59Q(KUk&EM>w3?p3Ipk0=ui9+Z_VvAoQI8f!zxcYXTTXNhCJ8{ zEGZyIMXR0fo+_LEBtoEv!r)*pQYzlav|zKG+B%kSiq?y{5RXAnbLKC?BAc{`_e8wm z;APkmjp-rKTYnunW~aYs^Y&wH>#^w@kNk#vcB@VTW07y9kMznLwC8s*rK%oBPI0l% zr*AF$-5Z$}IR7~9mi_-rNz9Nn%jej!+~P2sw*cR z(&;sFiBWH;ZeURvUx9>RYaq4R@Uc$MVZ;p`N0?0y1UU^vMF?e?Gi|(I+d262K{Bu>5r9N4_LNF2))(VV^%V2zduv<3Y?8b&P9bQ02 z`hYCsEyzRb7Lo9(-$3mHbdz-!3_KIt=m#XXE%>?4kq(9+$ z0q%n?gCTZF8B=!r6RQPL?z3tQwZwL`x<#&Egly_mD{H<~v(g4B{Qpw~KQ7*2gGX+O z4+_taLU5Cy&D(tq=8{*VA*W6Rpat3ppFNB^n5RGS4GjP`hW?4%zI}(ka~U}g1Xa;V zMWHLP0CugRGac{yj2?`f_4$LQPY6GmgVei;)nuU$llVP)@j{+tDAC2>y0C0u=v5)? z;3tW?8wl@o8{DvPUSqN_}$L{Owiv%>XbTZtf0Xg!Ef#j#+x;f;F(MbzAwU7G4K%mI90Pv&=9gO` z+rd_3AB(7VkG->zL{Q`bf@?|~8H$4$A$E3FU#W-)$R&@3RM^SrsAKEEpEM$}#g4Eq zj^Mdwiu;_rJ@ix5GpR@(Qp2QOv%Nr&KKu9IoAy6DYUElhs|Jy1gSA3S z!NhNYUElQY@2~Wn9MIN0YJ+(8`(BV-7v4wL!Z8K{aNpMxTW;3Y)@eC8jEVj1-X*D3 zg7Zm+DR!!Az>ddz2s_;Rbvi@gFIDv~Os4}>x{9u1)ufF?-Y>`c(`fkU_;5hgzMYc{ zBA`G73|6O^Mnx8nB$p*-@ECyLrY(U^^bd{6GsGu<5HEfvUFz?AQik`L;a9+1T8EV( z9`J3y9BL|*v<2VD+3aq$Cj=FZoOV5=GGh6Eu7<-2=oBSW{TAfL-T0Mmc>~R2p<;l;@66Iu-u^WAj>@ZT{A^VkQ#3@QUWW@Pq z+qKQa**-zBOHO)v)q`QGSo|sbU5YsjHGj!Dd*LQ+IRs&v$uPDXY$nYP4we?|gj5d12o6TFM0M;p{c zQSlFP=Np;hie%owlho28<*~bohOfaUrE;H;E~Ki7^y{Xn(70|y?LfR$O5ni5PwzWB zdu62QB*T~$QO)&YpTp){`4>h{n;WRIXtIZ-A}2rbRqE^OxKh&)U_KJm$>deb_e#M3vj-g zsCymk@19Qk?@XTo2UPIzyw$aD(IQs7cRBVRC&bS@$R_0}vO52%uB5?Cl0{%$s%>y` z-tjQ5riL2?|6x7m;OD52>rWmCH9z_kZSAHbcqM(LBvqMXFu9)n%Y?T(--N`-r}vGd z_M+eIuQDyHQK7L4Ga6l8Sf&-I%>?nQN}U>uPn^<|=Et{SJ=@nGUcBACma~i@hlENE z6gQ)>MKXvxe=IR6C7>6gH-OUJ?N$HS&>X})hGEj)anNXarRmhbKdN3B5iP5+3LaIjp-Q~=5}`2&njm&bR(dIEyPq%%tfBYF{muy{5Wf%^ zUV$t7($|ilk#^N_O9zK25NgzwY&n{$Kl`qgJAfFM5d!YIePUwb5vl<7BM5B5w;LTH zH8q=H?ZAI(%q9t*84#6Wg?!X09BTHE->A%i`DcavQWdF|rL%hSRJ z0>j?M$6s3GmRT{%i``n4eUjMYaPk(8Jbhsp7YP_yz zPJv5=clwRtu6^)xrC{(dltyWMSrx>|_@FtXJd#f95DK3WxYCi-8XH3XxY04;?(HI@ z%=|@W(5d-(VhQbh2nt^9#pG7!Q>9Z#qcLlIofyHI=p4dZuKl_Ob_Z9JJ6^Y;H*jfi zgqz}@_>8{@t=a+#1zotKVG^_YQ-}@uKJj6`ZbM8hjr@tYY!)HzaLb2JWcFPZ9EZm} zCa&AA)5T>u`MxL@d#EFhY~kGoD~{;-!cszl8Xyij%&X*Ktv>8S4pIdl1Xh>gC;he<%b_fBf&4u=& z`^t{pQ!bB>&)|Ohw5&cUa~XT+4sD)nm>t%YpK?uDMW^fG zFkquqOA=~asNmA4QI5urxQS=CvcvbnD~n*v+R!z*yS%dg*wMK&-A#v~ytuzqzbq+R zDNR5?!dHfzR*pzIRryh-R$aw|=}G2fpFLF3XB>}+;qqJC;*Eda7TdnmWC6ZMNtO#X zcs)w9K;wIZy+oZ^oteZFhNM{NS5#C~pevHPL%ZY#XCWAQ`t^&z1nJ!tW2LXw^7#4K zn1(y-_c^!l(XT+5gXq60w_EU0W@lf8xXO*&rE!~8-fg?+T;rR3Cp`JLySNkp=mM@` z9-hGI7CpC+9@A?NA${*D3_vNk5cuWJf8duKauUR{)wQNOk>cgpcdEOt-f6SA)rc3W za@Nwbg%h78MDg6aL6X}i&Q3Ubmf2!zFzHY}P2vm(HTI+czM{mIkK$}?d7J{+gB2Pd zsPf`OC7ZZH96883KZo8+OTuy4vfQ0P#b9-V2SBxkN3m2uwNk?&gv$-o{NG$z%|he8 z%bTB_H(XkFc$Bp>p%YD2x6-ef9~P~uLkh^AW=B}c%haPIbM9ekFns=>Lr|3(Qh!JF z5~=LK>{Uz{nj;zv;?6po)uEoJH%9n$ve|hrwy>{jYPf1@4!$wSB`~u@S+hkXKU+5M zvi*1BmMU>7pc0}Yj7#nfovF@O<{GV;emi7YGty^wu zEAahb(hBe*(lRoeTz1=xwJb}DP${mNUt|(WEPSc}pEmChAi)mFsWuH;r81<|3REwL z5K3qQt-ZZ7{@X{~1*aiABg9C=7Y~+h$$*BZ9jdf3%7PiZqfY2P>jo(vZ`iwl!&A_g zCAr4RYFX7Mn-TLoHR|u;!PSD7jae9yl?jDu;xd(4#bd7?3JiciGfnKV=+XvRL8-9} zk?2_TuxQ)s44ld?y?f(!6CWPs=EaFC6wTn88MvdhkapstC&KUu*_0B1HNxC2LDMb6 zyATt*gsAw&Z4IJs3R)Ev$YE9TpW5m#Ij69ov~VcQ{|+|*re*ZwvN9RP)iKBH=VW)v z81{)%M&FWta?2(lM9ExwHOElGW6PVRU--&pW8+LGE-mFwKe@kS(WK%7vXS~ESwMNQ z$2;qEIOukP(pCkOwyn+8%gHXGwAx#qZS$usk+e#aB0(FCb|o_tNt$`X?eCt5K}ClK z{q>_!YQP)-nGkHj$aX95H43^5fKpM3JjMJ^cL6ubnwpxcI>J5QKX3!f1CzT|V2_lT zwmTt3l%>wm$C!~uv@4g$n>YvsKhzw}2=khaU@qmNPqm}?7(r^7ybzZwz666}^{-0h zw%iI+f+L^)&eTnPZ1M^~I=vn~v~;B_=>?DG_PBb!U-P%mX+KTnwfbF+^NcIz9{IIA z1CRz%7Z~&xe23LRIO+roH#YbGq7^i2cagh;Z`>by7EL>XcSiSCxelJzwD?(&;++~t zkUn6&!sPx6b06)@V!d=o|0Ko0F!2R@w<{`bA@^B0o@FAHwHN>GLI<-X4nFN5U-^s0 zBx{n+7#6-&;VH67rSO4azIMWJ%L?Utkiz(^9mdJHx7cZS~{# z+_ zQT{~luc5iHFAyPidZ9C-xfy-sh>4apd7!zXMg08B^%ijF?%3RSAGmsD>~40zkR7P+ zmKfmh;7o=8^L;qga)pR^_4*5AzgQo!Vd-0j5Ut8d?(K@b5dGw(?b`(HM8L+a^HuWi zKR7O!`gi)+XrzUUars%>)lQZxOJE__Vk=@<#n&ib;o7pg*+6vu!(S}t(o(lTLH`1L z3oY-`Sf*kpUnPugm&A!G6bUU<1UD&Q@)9VW_-OSMSmrXX!T*>h0qGhZ*-DKKOf&C; zuMA3o1c^}`zGQ*hwwsQX+$|aDu4rZKm@{A( z<{QC0jrH5nJ;f?stpskmaGk{%H)@uL!UlvvXtBv}-O4_7bu&CVEgXC8-H2U-wc*X1 zwkB7kGhB(C?p-brh|VKhZ8OYWoJ3hYT%QbF?bu0ZyKn=vx2d{8DmE zLAga3h~q2ei_)s#$Llf-Sz~IQ^r6`M3cVCBoMKkO$-J4B)x+R4M7M!+=`|UXib`jC z`PVVu?)|0dK<0Ty`B3buc5tngO-|F0-~Yx0mNR=0?iJXID9t7@$7%6`{Q$$Qa&Z;K{Ai!^3z2@wSc5Ulxy!Z$!!> zp1+5?D|95-!dodVYclOrlvEISCyqyaz1h1B5o1DI27i-DtPQzpQBascA08hc&;LeV zzw*SJ-SzbqS@YwmqCW>jPutF$ESdHa!aXHqVyY69)q0!pP z>~>4UR=@GVCWF<-h}p-OK{|UV57UIw2RLcb?m|a|V($pz&(r`?jowKIFSm!QPqbX6 z3hgO+2v`M+D7JMkA#MsHOh&A}H^ep#Wj9c`oM5(@7$;%V(xw!C4Fm)X>O0-xiQuO3 zOAO_D;A=E?!-bs}7k8MZpIou`b@F6cR(k#n_9Fr9?0ONU47xMzxG;EZMu80Q*!-fn zE+gL25Z$u|Ik1D4l?_#hMBKP5bw`_JJa4U@{u)qnv3i?vdmFzb2DMBRw5_NO9(o;# zI~nyohEyn5qX8fD85WY8yVrO`;qnDYf}7DU-@=(RIa9NzRyUt|cO0=LV*?865lkZh zYQmHy3_aD7iZ3?Or~OA>PNqOll;!y?{5PQ5s*m(Uvey8SV+{u*bCI{mWp&O}knQ+9 zZ^DWK7RK&EOMw4vpd}S)XzYHmva&%`noEP2Q}qBNbI&$E;kJ>UXn==A4q5c7x%N4AFdzW1{9R z^J=HY_l7Ic-s#>9^6o#STK_V~cVGWT=d=Gp^Uvb8YQDzZ#QS?Cz!WITG9mB8WeD$X zTuw!%h}9IGLz)#-MN=ON9iTz@@$f4HszL|L^(d4pWP{Y0SAk0@P{YAWi$*zykHKYF zceYH*Ks8~FS%Q=XqCshN+I71MeX=f9*k+RSYu>rvc5CD%v^c0GSo;tb9P*NJz83Z7 zdOBU(N;k8KiW>uxOU$&)ii!`KCU$+uRVHs166_i%@IYLXZZmlVy%ONF;J6_pakt^S z2?ah(`o(ofA4WPS__d;{*H*ZTkGT@;SHaWzp#6B^#k;X{DcM@4cr zzgiDK5PHaJ&T9*RAz~03iME0+^J5e|vVtN-B+#B?nUO{ISmoZvL?@?&hQ}ewQ*80{ zw0wF?c1g`jy2gt_j`uIWI>a34>Zw^@5opWN@%vQ2mhgbI%BaNe-3s@Mz$zcGLJ_F$ zz86r9Xp+{B(3&V8er#QtMa%1T@110pjX6^djOui6Vo+_nJVGnfV2g)`5r?=WBONpQ zg1NM0zRmAn120W?9+-*k=;Ui$lI+CSgSo`X$H^xdS&4pbCRokutis6?)Ph zfG@xW9@39_(w~++!U*bk=g^Kc#N7rTD9Yi|0-zN{rv{=P96q!-7~o!n&V5C@z2Wu# zJA4_%+N>by$P!(hqa8DOAr+!GQbSBe*?@x8*NTfUQ?OU-S6kpD!tJ1IDZo2cgk!KM z;_LpF#i)z%w4`h4r?ROm)L!HBylJ%I!)l&J+1UEi)pAGucQMkMii3yw`95OZa_+>G zv(dIwELQlMENXA&G9J82d6X{hefO_=ppE&BnhEOG0EKozADuOwu$|?6^)>_gT>n7S z3G7$B%`kb>!2AUBo2@T$uym}@9MrT4;c}eslkTklD?1;^*2*Pe#R0kqsp>%RF^h zfTsYv&PDw@U3EFf6*-k!VaanJNO?F0?_{e%!5rsH^{u+7=uE-0)uQ$f3dH*@ z^?4HC=2sk5*BM8Da-oge)HQ2A5jcLwUq4f^R_r&;)$1+BzRqda6rVu0HN1R4cw0OG zba?=CgUdV||KBI!GOto3G(5D_f)y$sM(;tWy3@ObbL>-h&7jS|CX|c!T4@ScEAV1t zA6f=KcAzvdMaTR2(mRI$thkPt6(cNdma7{ZL?cAT zHnD?JnSHQw(`J=rP5w5ab-zfRnM+u>AClt$>DO^|bS`%8nBXg^Lt>; zz2x#BZKFC`m$uHNniobWKYs9SKAkI7X!&oUe)=>CuK6X0D|tuWZFvA8+=yA^B-sG9 ztcOIOSg0>`1hsv%PD9zJ*IU!IMzmfl(eIb>;>hpCqceSam5Qh!IfgZ>*SRT4l|mGY z@fQ@C#gyoqGHraXQ9J)*^TrJX7sTbVRoun}c%4`n=Nga;Ul-s9~8zs}?L|IOzn50`X2dBNgUm zfJJwKj!8X|XdITjrtN}l*Cqm=Rh|%6w4tA`5Vvtn2jtI#pTT(`um|QspS2mL)EY#D zaZ#XafMW!9F=UYl@GivEWWTCbDuLL?Y@-G);i34G9rc&OY<=Gc6v~TU+&xnTU0sjvN=#a1T7iHkwCZ z1MCR|kgq>~2J2Nu6MU%9W`*;OOI%#_f7Rg9z;qSH7cO_YwvG!vBF`bhjIYktsh)j0 zt21RGTCr<<8;^q{Q+D)|j=@&@WK>O4&hW1D=p8oAPx6k^2R~8tOllYzW?klu`a5F| zN!(jjO6PPclqRTnEV!3`hH*xevif**LO2tPY}1go;i@V6R2^af0L z@$cWRU(WukLDWwozA_cx`fI%uQiM5|O)9cG!ue_voj zM+H9%3M<6l0IR145sjiPNu#JjKkJI-2Pg|f; zfbn)uhtE&1EYSz36BV(5ySnJJ891%2+f)j5@8mAV_7T%$ z&B(DgWh(gM%r)!$+I6~T#PO)VUbUY~<$%1`?S2BX^_inLX_EXt^#p zK-EfWQa4K=?W0tC0L#o9lEpuN`U`_@=J${&L-`!Xieb9hx~=J_l%4+WG1uv`y1>U= zyT|5e5QtJ?{Z(|irEhQ29YH?6v)#;4V-`_j7Ks44bi_EwYhOJ$OpE3@_P&7Z`AOY5 zxCIR#VMJ85g^;V&8Y*o^DQ)dHfwcp4+M?80X4u>$H?5Ofi90I_-k7I>4=|m6gjoXq zc^}P-!4)BUg0ufIf3`ie1aw|6>|m0#sS!I6h_(~*sCQ4TtAjdxr^DO;MPPG!bFUUv zJScd*F|~N`?G0mfX1)Z>h z{A>c~|GmR!>E zepKv*aJ+OWxfu2$vqb__Mc|h`k2&KC`9YT=5(v-R+_Wt&DPbXt9F(|6fGMQIwR|A- zGc?1~R6KX|8WK0-U?OM#X+B!adi}STM}W#B7S5}(Js&sS#%4~w8cqHkk5>Ff$|>E! zDyZgg1$GgN4s;1Kp^lj(b@D7Q5t+v3`)>6;nJT^X#s1I2;m2zm)Peag2MxLZMVBTS zkayM(=_tgw7YJ@q2Eq=_^OD zp6UN|6Q~B@GGOcJO}tXi4+{A2JM)Q<@ z{{7u4BFc1dWiWDhlYW3jBMcXn()tYV(-XnV8LuF-_~O=F``y8W_ZfPk(#~q)EkxTp zp3mkQQKl9lXoarSCPaM&OmjM*O{#63HkcH(n&wQ&-=a1zg%?irwQ8dIF2QOi$+5xZ zI+J4Ek+_`J+s|m!=$kIUb{%iowi+EX4Ql9u_1Wsn9tym6EGO;!ti)HNic0I`uT|dW zxX?l58)f5|eTUJ#WE9C0IZT^;kUkNZCmz_C$fZ3D#sZY3Ga$#~`+5u1l5MYFzUeQj zkc3WKfCg!U!s*9EZFvSueNI71N=-K?8%3|Ad*5F7Srr%j>|phV^}Sti*OI@p+7~Y| zYDPxJv@{#{*$@9MGhNI@Y1!FU?(P?CXSn(P=dK{$ew}XG6W+o9u35`w8^xVJ?amMf zxi|eY%b6iIXW5B}{ODk{L^fCpcRQp3<>jsWIx=LzK!n8<%;I6q$pLOUBm+g%y9HSN zn|F88u+!~}wtjw&dK=|Dh6A0a4&KB`UY3vQeYuh&iG-TFx^kzUVbR-+0=1Om6<~J( ziAk{u$3p>D?Ex^+HN{GTiLMDubb{|BEp@ikmBwC?^$$y^v3R7(wY{YYrf|~P`{DVE zpfkx{7*&yBI}K)8mPu>ai5;QR0NRGh`>H==CxGdfWR0RHA3jU3B6mzth1Ze_P9g@s zwUCNv76q+Zn%?cT$D*;z5B_54*j1I*6NrzyVGgt%O^h@S%r)zNNOIkYe(5T~OOlZJ zLS(af1dcRdYamn7l_M{vtR}sZ+W&pUxsrrb96d>-?TCdrtT9m|yOcd6zZXOd22-H+yDCv8) z!dl-&l)9^Z=N2EEm3bjTDWc1R!$;bav$aMsqTL1!jC}m6(CN#M z^FcoigO)@!k#Lu$AsE}h5o*bkes%P2>L4PBgQ~Q2be`3?22X%IbnL~RWSG?0yTNCL zvllCx>5CN&%>LmGUj^ke0}_gjWMIH)L7&+FFoV&gI#q$b6CNP~;k>YTfN}urx~R8c z%=!JdDRQ&Hz;SXA>=r8b@55YL8g5vP-*Wnv&ET#JN)I@2vOZJ?Kp|tel`Rm~+;*z`7WUp*`q3K`T5k4|* z@Vvm<-T`MG#Hwc@ewh^jxM1e)#+)wfuW9Rvi>=z(n@8?oiWH)kuGuSV;QBzrfo3#ul<$CjUD?(O~a zZRIdsd>C;j%Hj=RZQal_yv5xkW}I@TYL5Yx^SVf z{GXL@wMTaBI8mWZ`x9Z=krpE zWi{#hG#?&hGO8kQpX6zPtRio=5l_;r1Vt+npaV36?!s7!2s3h~=zy8Zh(llZ10fVFPxxd?~Lbiyzts3XVe$a?CxpDlcp zXy0|-+jGm6KWZ}U`haEW-XsYS>kh$b@6yQpiNb}gvp;-BWZ=2mS1tC5R`rxT>lauT_ZU~&U9f{31Pf$Qro!Y)lz`1gKu<*R=H zaghQ&S=$NAv~0og%lrwrh0iuE`m=zY0#>~s)VGHJ$|5Z2UUzy3ANOTl zEz;T{fG0&`>y`^QMR&f)C0D#t`To2@L@8)hwC!p6yakcp*Vxu>uJ0R{*9KgS`ZXE? zpK6AiJ}tcU0nA5jfb4}b3a>N--N}?{yUb-g`fn9Qq~*wO5SYSVo#OUwChAyadoSeF z4vM?~l(xG7?u6NDu9G?sV08}jF*ui*qGqp~_bxwi-~Alkx14WKdH5a;|Dz4x^R-(Z z*#X1IICB}OElWrsbFB`~-FIi^@4(dUg4m5{69Km$f>1o`MkVHLojJR`$N7d!ujA~4 z?)?Rl>lrCA&8NECdZcD@n2^c^gCEx9gckK*&^C4AiQDS$qIy)-x^r%}&+yZuE@D4ggUVM1vsW`A*Ml;Ff`6gQZ+NH- zSwfW|2a!iQhEHbzj6{?;cm2&PA2#I=x64da?M%;K(tDGuf0#oEli%}l(K7ySV;lY5 zIf3xu3cqMol>ZhR)%7}$Jre+|DOc)R{CCUu=ux&>kRxdKgRbGb?0=PlMF?~Wydy%= z5#GqGp$8J4Kklh+WE9(qLGsC;#VDP`!n?E}VSK#;%I!Yj)L?N{%Qji*Z_hPE-e}v9 zMX0*li*1y-qw=s!gdY7i(BcucH&eD6=&j)J!FuWsV*Mp>30T2_hBGyvqbCF?XfMeA z8iIHC8edR$ti;b(#2-k^#aW8po-X<{0R!VL^83%pe5&7HIOLF(OR9-C3s7*DT450M z=W38t(x=zSosFc;oWyD7$GBaJB5%F5B;O=+tRK37;kYd?pmE?y*>Kwf@3IE!((7zf z{|8g=9nNL{{*RYCLP%s~3uW(>Rc3bCd+$AxnNi5zJ5og1o9vL)%LpMmn}qDm_q=+4 zj^FV+j{EhWboY8**YiBj$N3nut|K~2q>`agpUCB>>5x(Uigv{c=6ehMR|{9r{2ZPwip{JgiJG=6{oPr5L8YG3WgD+e4#U0u6n$4L&~D-(dEK7B0g|iEkILDB(r%36e7Vgge2M>P-5tHu3iBSaQ`e@!??F@=m-|vQ_H~{8(*AivnDrgOgl!ZeUz{q zQy_Sg=AaPV`v{Q6pA{B2AUFr?K!8>Q4K-k8V2Bi8IS60hq6jjh6o%}(S){)D;JOFe z(jQ0yHRxIb>THdZEkHx6DYhM^@)xWh*MHVTsoEMILAE7m*y9!0P2dUjXgQsRMc*@n zuJ!$R<&Vb@eDD>Cx0aT)+qn4>k>$e!{pS4iP}ryeo<^^;?XK|#kF9LB0Uv3m4tQj{ znhs}BaKwOjz~%~c6NqjGg5{Q=k#T$CVO|i>(kIBlmbV|+*(0^z zqDO3#5bX>S=lbk{QnBV;g0M1&Nq(@0z=2Bp;cf9D5;TUq0Y@N{Yz0SC7o$qB5Jt>` z-kvSqJ2ZbIb@;N?c-`qvUjAS2kX>W7JuNE>4UhlQrZc4kQRN8o7+6jJn$O|IOWSmM zZs{yR>g(JGLE5;env8EE-{SBGjGX5b(B0Ft}V zl0r-bO?nX9ohk42Vtr7T_AgvPOygx^s0?Zxj$!Tr6PAI-hs#LH__I}{+uvyQX#9>G z$>n*=WeOrS&Ckzg$Rdg9LJSoC`79udSoiUo5y)*lTa^``|7XH9<$42oOCWn^d<+T; z1gCs-0V%{zo1m-&_pb;Slyot!fT?_cuBAemE@oV1;N}a_X7VI=3vL@7GYHn~YNY$yha7QRrA~ zz;IDQVl2Ys_0dvc@-8G92ONEf@J!|U#tGRO;U4~u0K9wSY+~|E!(s* zy)brsIwR)I!Vk?2?6emYyF1Uxz2)U-+@LR%5uv(wC+eHmOQ`&THQMEA)qcWUuG z*J@zV5+|XSV3>9PHvcBCr^KMa6O=zbQ)7^cW}yRC*(=8z$xBxt@$qY8*-CjS`p-JI zdq}U}SV{>l2VKeNF9`iZoZq zQB66wi-V6C5W*ly?CEX6Z=^G%EQ~uk1#&aPGkw}^Wy&o>v5#J3vBR$?oSYCSd z*%`K9p88+Lq@|HHZ$~-(h394X*t=TSV86cO1mJpQZ<&v!>VR|QS^i*vJTP1)V= zcqq%;r*s{33XR>GcJ3UrH5@*o*3Z`Oi7n`Vove{5KuD2mzkPuURd)wGea?+3r7Sp9rM28 z$a-d6_ly-5{6GZ+;TqMA_lMtwkcj8lTU@`#U9#a! z8;P}XH**2~%hb4A@xfxHj;oxqjx>imrQfq{Hv=Z}s3`ZRxs9JJ9mY>O>i+T|K;#_s|K`+MIb7;)2ly3xk^ko9>B(lV$a#g$Pf#t`8=-sMe z<%PurmF4%Zsay@+pbMe$TD;XScKU}wFXnZ-3BZ5LY#eX;Tz`GJ`p$#o7d5c)5Sw#NFW#6Np+^MhwHdSiVJ+UA)DKy zRZ7~5;yZO^~%Z2|mg(`*X^T6>nQoYdCGMT{ZqpM{U zwoTsO`MbZX_+om{wP~*rX)t!NdJX-^2Lo*)Z1JmB{_hXpMQzhif9rM7b2?0@AekxB zz1v$nU_t^J3kI}{=bUHu_TQ1|1&EQF8!xS6A}&(P zI#Gu*ru%j={E&cCjJ znR9ccgPl7(CG3rVB^;afHTy+&6;o;)-Mw$!7J;AO0HVP^!?Xx^NJ>&XhLr@6vwK=N zQm@;O*lGTKNzxdA(4)SZmJ>BCLlZCuJ~|OR7sxMy)(G)d^|@Na6Uziqov%vP?|~-zN4JPpcI?|8mWbf8*DCsw(7q@QfuxryT?Wyax$Pf7(}~@bMyV zVJTA_$~>v3PeI6`f&*-jk7lk7<^y8DUiHRBF?H~&Q zuJNK+ChR3Cr&*in$qIZIwu3Ke2cNk8I^xMm^v&fF8uyHYH^rkX@;7| z-S$@0Cu6DVRT5}Cdd8{>E%Q|K(I>ydPMlIKe{8)Gd@l`oBl97t1ukz-(2jTUyxqo$ z@1X}y5TpzxaQGji^nEs}Z%kN%$1Qu)o+=W1IKTb5aB?CF+AhRa4GHT&XmD_|lfqnh zCyE;=KkYZgdyzCW@rwoV2P2P#`R?EvLR8u=WP_v_osSq9>^>Jc_-WQSPgJVS8woUb zFO=R)8x$8+q_-l(4o=O;SnLwN;6h?(>(=t#5H~I#OggRXm->quU7n0-Rb4dgI=o~c;+7_vp!P{#>hx)L|EN`~U~`Y# zz)@$WRQ`i$lDBASb@IbyZv#^IXeEme;rmOGB4pmCgxnkOKn;S;3@bHZ8xFy+?Bi9l zYx~_5sKIAZm{@c&ZCF;88~MPlj@BrR&)>SM=1GwyAWzRJBPiJH2;qj#_JWlU^E8Cq z*fWGs^srM{QpB>tF6TmQ6Dt=VtGy4ME(o$Q<11VN(mR#IIr~CYySX>mA~xW*DbT3x z$=LmAAoly?mof3*Es5mA$ueO3BQ5u8C)kJGxq$Iwq~|Ye{uvVxoMZkFr2AtcRzN&C zY8O9~M4lofmd6zgBtUwV;qJS5m)Zo_7KpB#-CuTH_8+OGS~4he4&_>hQ?yx91o-yB z#$={{Y2J`UL!Yyi-mb4$%e;GOr_ou5k@M+BEr+KskXMlAD+8{XT`HAaRek4HbmY^x z;o!7Tfnr~PMBz7w8PGsvAVzG+lOsC5JR%4td!EB@X1-da7|7het-B`#a|Hsdi5%7q zEP>x92VlDX?TA~2;&8@cBUBIa*&~@v=4QGLV7Y05uLAqv)cpkElNK; zg8yHyag<>v;-u9K0cbm5#x_G1;{YZKqDW#Ek~gAR_0PpzvuUG@6O2lXgkx>$2|f_M z(Wtbjiy-1&yI4lcc@aJAkgr6WoL%gg@qC%yr@m_>oh9xK-Pd|I_xa~CVA~8k zJygB8jBEAIJNyXM>DSbgN&M5+Iy328I+qXgGKTHxH-48Q-XTDju{8jnkTX=v8LF|( zPS|5}G4GAze4?7Niv?wU$?oO3|0VUMxQ_?bJspkN%S-po;sQgvwHH9B2{1AG5gH!e ziB6P&FBgfm!)Mepz`!0hcr-%#^&J7>*RKpGQ8ElN=XWHfq#s2QpiAHSj#<&o_+L8? z&#;e{hSr$(!14K0cPA|mEe;KLyN!W_E>6*rk;9oyG57Tux7ov8FUWA|K3>Th088B_ zcLVVDr0_r6Zd^Cm+OaU`8x`eAmB2KL0=VzNgK#xXRYwz(Kzn|@ee<@XqtA!sr`533 zdbkvEw(0h*P`Q3!Shv(Oni>ur?u7;M5-(q4;{XVPcTpmBt45sC-#9hbnF=4mSROI; zeA2B7Q#o%)oL!yRv z!Sl-vcC0ek_Pf%Lg&_LU7t}k~zd{5R`0o6D7r91)AA~>v0Mj^(yrnrS+fD z33kLH?{dWISVa%9O#E_#$jN}EtC}N!;U}{vdzSnM&#TzW$~ek&3B3~dlJr!woI z2jf`Y)W)r2Y4X08X%;PBq1TVluaC=S=y2zbE;ivf-}VvI&HNIYZk@nsDO|&PXiIuh zO)KFUNq$_sTmLF+t9RzqH^J#^Y)*uunusM{@Wfmb1h)p^XU+psiVn_t{wqtakzByS za#=5~Jt6L7_2?wCaz?S$dIm2{Jz0i{&kzaJogEv)ab0E(`eY5D7qUI*FoR>ZD|qPG z(cRsh-o;<|Y)!B~ykfKM1_1#^a1=!>Z?$+P%&OK}Curw2-uO3@DWaaT<48gm`eG7=`z7PaO73zY8nL+33>S}tI z_(6FL-=68zZ~hH}ZhQa0hgTxSA3Z34;l+k}ksq6XWvN=PFv+1V^NmG?qpmDUC{aSs-QIZ5JUa(2g6F?o7FH@zW5Q<+cvHT;uE zBrp);HRo6k>zhxaZCtHtf_(o@)$HfQqpUkLk3)T*$E5|o7UFy%oZ!|%m-Ke69wdoq zUw^C?IOowRBr^~wZeKh7tksrgcSI6;%d{|(i#7e4N~)|^fbjT$WG|O=2*b`I)?oX6 zSr^`WoF&f(Kh}))kcz@>47Wz=1D1_TadIi``7*SpJFcj zK$0SV`_e*c~L~{gnhm=$sk^jv6wIe1WIbKuo$7X#k?&IUr^tH(-0UwzT znK6D}>~LIGm@53~JIlJ<^=B5lxV)IXEcDjdfVB(~pR9Q{g(OSD`2m#b(g$G}GZzQ7 zT1^J8UcEY7%eq)g@5=I=#{%ppuOOvXqmaeCe$I5T8N3azL|3^JyF1SkP>Aa#j~UP= z%+uHPVuZ?kvUpeG=DU1}$yVJRTPT&i#xHn1PCvjpt@x%`F<;mZ6TE?$zQBCpRFyx6 z5UIf2oBD#c>v}(Ofa63{r~pa_Imil0wjivQ72{E&_rJVw%Le5U}@YGcADPwrPig`)AgT57xf2TP6X5K!YxII`)f^G zCr@eB9NF>r-MS8bOAuzC^NB{X!F-qq4t%rP+fKPZ>b>G*J_Ej{)oVp&HfpbS&w7Q& zpH?a&(`z}caaEcFCPSaVyezQ%@ja-_>ZoAF4lYi5qep4{s!!4$-YgoBN(R&#dI&iZ zNw+9VA!fOJG5dRaf5WvJv-6gkAbRW9r<`cr-6O<5COvtK9ZA#*kD(TFHAkVkmt2uV zcX)dN8wU!ylhWjVfmTz&{6f^37tosfVcE`kMEHSlBWJut38nhZIKt@#`7TSWjyK2N zDP2RO>a{D5S3ag+9sHBLM8J7VNy}6zNBRk_&$)mu44W@x?1CZCso-ch@4bXTUz_*wI*EDhmceJ7nImATZ-;)`yykRPX*H z(qUYi{j17bLF!(m{6_2fr>X*Z|KGT(-cBtB#O(MR*~wp=#$7UPjNeEKL{6pVMks&J zHo6F*L}PevQpUFyh{2i`Zcj8xI+ zQS^Ds_h*#&NcM_+D=l_$d^p9=5HC| z&E*pA$)+;^Co5)+p%?$D%&MAQLmFne?n*QBox4u1telcBy@oz~7ZeneHyN8#OHAC7nkEn`!xhfFGQx|NG|K*a@Gh~SL}Y%YDPe43w2*mK5Dzh@?}4p-C=BNk83;+n-oCe1-dOx}e1JJ*?d@UX%1HH;eAtTsBo5l0 ztu|@2kX|+aWX&6b$CJZj1m;(oKq7)iC89m%I(CP1*;G%8#SyR zze{D-w32%(rm#cY9ltYk_i+R)@kkqCq;7@vcCu=P_SvtANOk5~Qx)?COJEdJ60F@z zRocvn)nmS4vT3gJi&y!ESf<%aI4jwD?x4E20Yt{#Gp#P~dBWMoHxfd1%=Qw4dG3*H z-b>K?()#{k$;g{O4W@!_83wf$@2A3Myx94%<6-nj$;u|G8f9fG&ZI8LbF*#A%gY-Tsrmf*XVv~`?y%F? zZ3*5-H2A-ewL%w#8x9aRE?Y0oPvK>P?f4Ff|2B#5<>~CdyjMu_sYH1DVI8I{%`8w1utcBtNSE$ZKBb*1CKv zeG1fm(l4(mGJkf;OS3`$xJEL0qqpZ5ag;F;$MKKb-MmqKA9{kStoic$nROrADLvqm zWcEtce-J&0cNOjDHOx;muWUn84eVtdb)FXr)H+-%q5+&!=BM1Fm9TZPdGCgQcneGpKXYb`ZK4`E0hSs~c1LCzak^M(S=Nv-zlK^VH~yn0hy_ zeCs(kDs6<5*C)B#JC`h?6af*K~aV`HvPmS+Cda%xV)jW#}Aa^q~B>Vg2c4H;= z$##?ca6EIcub6oDA`||8FhLzF`Zv&JlQdZ?pqwc_&XcdeesX=Lc|UpZQ&uffyMWSn z)GFG4Xhy-N)m`MIlWd?Hn_SLzrX|_6GekmsY3oTf{NwTtLAEY=6TN)9YG}xJ-2|n9 z@>D@7e#+|`#wLXHp~WYp>Rj2Q_n+EHNu=tPJ;IlMRJRUTo){8k4^v+hqPo_-z#{B8!r$HZ3RJ-Q8uO zYJi&-A`qa6x6LSsnv~*{Q|T*X5k|% zk#m3^FTuy4Jo2IOl4Gan@v!bL9M|xvhu4+Lv%2?BLXn7!#=TsQ{qxhw7XQoiF5vF0 zDg50<`ar^NUs%F7C2wUeB9|kwJxuR{7i5ZWdvJU56^jvYc3W0PYAI?Z@`8q>B()7)KX;<@{^c%t)17$q z#=4%_kR(clJCt1u$D}_+BJ~F8hOrqTNL5lE5#6Xy9%%9B2 zpTgbiB3L{M`U7R=@q&3}bQ$AaG`B)9^Ec~LDWA}WkiD*Ow)9%+Wr$bY4hrIt203pm zV+R@!ed;|Gyr@^N{v-8Xz1vIr0>|{2xc9fz#=+a6RqQlZ$MR{fjn?%>{$Cnt`>@VG zaL_S5<-Xk&_rDYQstLyMatPFPKlh2eN#q;?O93Lma_6nPOGI$e%Iiec4EPCL1E;(H zP6zxTyubxZ>zZgZg%H}W4bS(aqv}ZFOs&}uU9-q-3D2bgqw{Us%@qs`jL$#>{Jisx zg?#5+NlB^L%f2g`vYEAE$aUvf)6TOv8La<`0Z!lj^paY{Eayk9(MR)t9$gmPL4!@DRBovGz&hye8*%+fB$Vj zBAN{#icwF-9_E&FJl$&h9Qk69IjQsQkM7$S-K>3;r2a1zz5O&WD?e&u4O20gMj02* zC%)zK7J5kYo1jV}()D)E!P73&e12&d7`XLWDaGmJFTN*AvKXC(@>zIqE~6c8E{sk%P8~A6Wt? zKn%1)2hczcNMT390K_F5$*~szh&!)-f6v$TrLUlW?H;217$w5I2CoT2rgG`y`ce6rZnwv900CmJM`^WhU^?1}0*qzu>#mM9|v-&qJ_4f6-tJCbl zIt!Kl?(P?4i-W8>oV8sB->-uB{)&g9Z zI)o=xEPINBgkqj!vmk@4F1ATD!4vMNG1f;cUr8HCD?Q@Ibcpw!l@|-iRgv`T(-^

SGE!6d(OzqvvZ=Sp|% z0kerMFOvLm%j7YD)agWBMOMejylf=6zYdJc5;f`%M?4=H*K$AMYke$!u+f2?R*<_z zxPv!>i*p}(=$CyZIf#zD&lPIu-?$jxucx(o&Kuz@CBKlUQSb_JlpH-h-}29bO>o7$ z&#$eE%`JZO%qV)EI*CwsbhN9$CI@E{rEb?8!oYor)RGa0o@xA#4wFIc|C+dep7f9h z%y(6#1&I3LF*ot>PAblv1@%X~CkIKI|Drd;hO5MDs)7|EBdKdd*At0v$y^kn{!3G( ztvs(I;y%Y73LCYt*x*93Uu<*d*^7}xWZy88Jd~ZocZR*Tyzp#T`un@F zqMMV9QXA^BLF{9(>w*sV5~35r#UC+>f^_V^pW1swlBrN7d=dk&YrNHJo8FDIr@Hf+ zQ79=}H9A|SOINu(j-&7KWy&C9Y{PG5zhM=k9(s%M%MGjkI7=4|z1>%PX@RPGZdg{~ zqfDtHH+l&;qK237^CV;OO-zHx;$u%4)&G->six>sfM(v4++&Y<>!^BrKW;xotui^J zVo@r4m@MaMfv1urGJ=}2AR~p#PS3qvuzK~1VakPTpmC9bV=mpUWDXhnd<|9*x|$&j z?sJLN%+D$%fIr0u^L%vax*bNEJHCYa;^tJ<=oOadzJ8Mp9vlb_4vDGo|9jp4p}&92 zVV%tM)nwEXvsiOIrF-$q)^8+z4;UH`p+SBr0Bj5*@k5B)OA%Su87Q#zDn9ZO*s+=( zuJkpLV>Yi5=ZMYMYoyK2(SbO#x|<8uP0HW%)B{aeHi{Efui0behIKyGEbcP4cw`dM zd2Q^2M1Zx>_*VFYIgbnR!fUvZ-e-zDUu%g4Lq?hI-xgC7eS!@=YrMPO=|OK9lnQS5 zM<$p@2Vf1Vz|4ncK-NEX^>*P+VwRhEa2Da$5)1|1QtyM4&u!cl($}8(lwRQu)ld-n^Tb^Fv@3=2Pd#8iaD7iwQvBA00d3h)0_$h7W=PFCh;G{8~L#|vZ ztp6=yr1<`Mwu247*9h12!VEMEF;p}zo+Y#x> z8Dzg z7Z_Ty_{~I}=&|*fsv?$DKh`!X=ms)ta z^jl_@e!}m6cI@)|SLQDNzKB$RWkVDfDDZ~6r(Ug^V?QxC(xfYuFQ+`8qrRX8iXq7n zuq&Dv6Z5N=kw#|B(1#7#bN}P2!NThEQg6(H;1p1feAg6==8<(H+(FWqmA)^4`D!9t z!*cy!*_hk>a{*zzgk3IRaRn4d3!RzS!pr-)jV&zaE-90E_XyXOrs{|utyj;uPP}O2 zjC=f7uVr5kG%(J6D?p+K1hiRIO#`KQ)<}L_nf%bw3YuinGpwO7Vmp%o*Ftr+?*lI8 zk>k$dpRP0>4pwJzU~orWC3RE4uQtSt8kZsdMTa}?z1LEDBeey+qWg+jGyJ%r~*VO!du z>Qe8fKkjW7Rjf-=v&RJY`dK4bpS=5p&P5b%h=EC%qk>!IttOo6!#ccLt~b4-L|CI> zSUjoawQ|ejLY7Gbhmqym%QVaB;qOb%c`;@qxIX;lz<(6OcqIY&Gh9cX{$=|03$0MV z0 z`NBM2n=PgOQZ-NKO8n(;2iFE`xu4p+0bt9}DU!#PXF|f8q)OmLxbZPq05|j?7L}{5 z@A1C+@)0US| zxZuPsiPb?pN|-vLyOZy0UpyUiUtXGVs*WNDwe0yVr9xjx&FOJ*<6Slr9XB>xWl)KE9lBZbF#G$ z+J67WT4F=JugABn3xGB|3MZhL>eGgloV)fQGx_x-3n@qO<0;j0N-taDkFl-H8f`e= z6KyiU{@c^}?~nOSo2KrUZy8Ihdg30vL6ci<`j+;e6e<1a$Xstx zPpm#szy4lZnh6~akTN}o{%;tNtziwa!5mabyoC|^7_;&bJsw>R;g+CbpoPPfWx3#M zv-ge8c{(N0Yi!+~AAb-{rWY2=yp0feARnyw^rjgO@rI9se342U^1x5Dx^z)?ZPd>8 zBSpZWcDiQyJpXKVsjHbhe~(IwsuKa_`(UTc2L#>UQzxUpOUT4PJ(T^JUV;W+TTO>6 z?(tE@nK`%4_V)Ytv_{EW_lo%tNn9_kz*O0?Tj0Y0gjcl)z_Opgy7tzo} z-u8g2k>ti_m+Zp)sLr<(HylZV_0!GIXx4IdGJNyYv4$BRQopc`{lq8HOLgx9^K2qx zO{th&Ec{35eba_m=Q|I*{G4Lf?Z=#5XgooRp?-8yp(gG+f|auSLkqz?!*Huh7O-ZJ zr8qD1r|_kzV!B1Ml2K6bqVtB}i%OB2#-@TN_1Dy$oQzJ&N!Hi1HO+XXs_!n}+`SgE z)I(Jm{#4WN6Bqg_pla-)}5hDWZ^&5OAzD-2Msfo*byR zX=(XKM}Pm8ai9((xIz#TmOg+Dc4AgK#E5LJ4(^lRh6CN*k2|`^k9Qq^4CpddXG`Q& zu{!xX(jWpVZ|z*$L_22l^!cyLmN3T{+Nl-@#%=Zwj3q49I28VvYJU*l}5~=S?ZoJ$7B-VMq*>cL+(py8&Egq??*)VQhC9>bvjF9Pm zL-B9S;$rFChbF5bB`4xFR)!COppFWIY zQy3cYU-DaQJSgfcm8|+$lRv>MimH7jLEApxMmv;mgMG|UO6hg8X)jqKmCGm6aAfwk z5)ow)X@BH*C+VO0F=k;wT0%T3#9e}{ob4jRL5eqDNi|eFF!mDUF-v|7#^ihh35Ae? zP2QF^dF+s5jQur@p`Sv5e$bi%+nNIH8X3KO5b4$Y*a>x(F7pQTM^nXiz0Vz1#&QCP zTzVfcZzsmw4VI{zru;UUu60sB)K}xkAn8HXlc<|jJ6ZVj8J-!%*241!eg;I>*e9fU4e3X2NBH7OT&j|-9s0lY@UDTsv;fZy5CPAZ0BNV_>X3IsuzxxJ0YuDxn`YFfq@_T*_SMq@2T9_H^0KnB z`E3Z%tlG5DI?<6_@SBm(=L21>59ENg-PFu~Q91g{eyDcphDnK=Y^srm;G`pJXGBA| zqeiXB{a2;;%OuMsk`tqUug&P+f|2g;>$<@b(H%#vrkFe_Fn}I>myP*-&o&Cbx@%Q! zXjA1#MJN1fHjehw=%KCzZmfArX!n6+zpKbIaGQ1*btSF@$LgTK&zq^g2R4DIB%a;H z#}!Xtz-9dAeEl^~?HGADfBH4IuW4p+pS943TU$E;+J+8PwdoCqBCnL!6m$Fy8K5}`N)o`%rtcYb^nt6p~^ z{fDkjJ^<)?OvEyB4y?`%qC`3wF#-Xb70Z$1&`bV&o^JW#F#SF>nq!xT1&( zbo2CqirBHaV+&m8F`{RG+ub=lnk;yCR@m53!V3QW;-udRbl1`P*rJv-?sH<)K79F{ zxzr!pn%^tLDMqu01Z7R=Bh?hN%sU~jr%|O#Mb@VdZ&qEo59(Fl$KECKf0_G(H+V`_ z2eUGTYDB9j$R=vo3alA| zmvxnpwvF>v$OLyrT`*4&qrJ2pX^tHJy?%S9ptQr6sn`)>&|xOnwmW<}KVWn;Rt}Z< z`?2M>WEpGIJrxD9lrE-nDKY2S(9v3LrR(h|5A$~lXzqhtl5G(;9HtIDo zIY=XHhiHvHEcsly@Q=WFC(81Hu#*v@?#A7IVHUCGnUnMlP|VUGTx1r%O(b(^%X_Th zPQ?uJ({^ZS0I@ z9j5)@|3 z#H`G4AR*VX!A;hf4XRcBmSURK`p|_ozUy|>yP`W?pH-DGu@p#kszN5-h0X0OaI;S) zJ;!Ecd~0mu67I{dfo(Ejh{v+@%E~6ADR4yhxVrD32Q@f~ zJSS8;xLl4s)ElHav(+)&;!A+(X@}G`*lc(?m}kj-q9>9`Idrmy;=21Cp?gGd4?M_1 z?p8jVA3Xu1cZ&Jo@%3-;#}00T9qtK`^?O!!1*0W_Oam`B0YoKy3wKB8RTS_+^qwqK zEs#&XenQ3k;sw082!U-~*Rcswe+K?S`d9V*3L@6pL4qZ54K87Vh2J|1)9TqE_4!jJ zRIfSr*-@9c;qDQTvKz9)b7z6sW#!^;1e=gVH_6mDO35r&a?2cU+-|;BmNLmUcJCKe zkHUQiJ-eyBB3(mot}y)(Epo+$n&R8UNpT#yG8jf(-|N4a&f_ceFzt+YRN!0QY=y?; z)14ehrO5t|Nx>_LCTbM=zupgTKN|1cZggIz7tM>|e0oyPiL(;IQ#}YDkWAJ+tISe9LIXd@`@AuV|%i-U=b3pKxDr6t*^JxD+r= zzjF1;6*;L#>eUl?xrRq9EWug5g$TD#>~YG=x7xO`4TN| zSp?m`Yh5gtUI*KdKoKEJK{Sf@V}J`sieWS`(_1k`h$XkDktILoR;gjlNaNk5VEpoq zkTx5ZjFazOEnjbhyE39+X(B(TRZEa3Lr5=`c#BS|JW=0!RisZeHs5mfRHOAt=(q5d zMt805y=}e^sQ&hdjl7Ndp{N;W$3g~&KIzG!iL3BD-isS6{Aaynt|^6C{vHw8tALxR zRL)cGwBC@w?h*fQ?QyzxTje|7QXHifi`@;_METuduOA*`uumK+im-&hAK9gK4`>q)Mo<$tADh>-2J zAbA=BW0y(xDCvKi>e1-ZW0kBC_j9Dkr0ta2dI`pO==;#3dlK$;;k>>gUAo2*%yYe- zH)Grg$Etugf>Aky0Y_-ZTrO3pNL7%hAOx#cL9o(+^xNznHsh8?o4#o|YtR}#ja2G= zlMpNzTyv869y`uysxMo8R|*~RB{$f7>FbF__6;pf8x^6US+2{N==j~BoeVB0c{{%>wgTCC7bU~lGaoI$Ve!cRAlx*8?52kOJ6>`LmxV5t4W5)Jb!WBj& zepzHCT4*}biCDp&fg*tq_Os(V_29)refksnM=>zb0|c_7pk|U|_PE~zA*->3NO6FK z2!ZAKs|`9BQuNzS0XF4piD@~XY+1T_`S)gFI_YP~#XAO%@h!fXOC-olmeO5)db0jj z?L6IMno7dG{e3UBuxPc>Wu0dkd|McKL$5S~I5qN5S|Q@i)lajV)ksS;us6WqV8iae zX*XO0b-T@~O6yc3)#>?`Y^=E|-RlIxHN#ueN0JN>z7_*XLgg zhRT6SKxYa)XD4YQe%{YcRuo+)DVI<04J-Cl8)Bu#+x@-V?S%JPOI@6Q9Z$qaYDSzbUFr&JHRe;NrBnP?uBPFWNbp9y z7oJtD-}WKMeiX?C!33bsKK7jv#t>ajHPdah=({It42cKQbhqC9%=T(}vf#~}dFGUh zy{yS%p~>^8!q8e7Hb^tuco&N*7v_Dth@J)!=6Imiy-p`*`vHtg@0#67t}U{5X-ZRI zuM0v>*pBKFlz3fmViikEQ@*=>jbM;;F9lV6JcTg)`Ns5z_XxKEymQ&^ASpe*K!{S? z2<~lHU{!K2p$E2Sjqm4I5C_$Q&5bGWNs2ji;eX>`?XLB-B{1;n*1}6O%xmU5&FiVA zf_drqmgUS}Mz!XiUmp{r=SIzHR`F|VI+zXF>ELaQC>T7WEOy`bnysH97YM6JGH@oy zX0RpmF6b;JWU(&R(aoSw5|I!~2!L0vj8Cffj@9>Y#7JC26VT4DeH1RD|3@-2m-CtH z^=rc!lr-`P=3$HQgA33`^#jy03Sc?PrRvzD;XHUsr-WryGuA9`rxfo++>66Ra2r9V zBTSbDF(vmtHTRJ)K2R2>E+D0TnlFZKo-x-P?_!{rd3)ZejvNtS%I;@_iTZB zcJK83`)jeUBU(mH%gFg`BTd8!Zs{-rPN9?NKZFqJ-8z!SU+BI-!gfNcDF|_Z@>yC| zRaG^=jpAs@0A8TU@aR&Ve)PXALEz5ViFqKFw_sr~Ikr*!LB#k?kNKTuv8P$Bt*!Id z0ZZNR^?IbPK(1yxP_?}qPMIc8h^nr+UK?(-E_+dUCYt+piOuRaZfA_)Z;3=5JT{i^ zdelbP8g2%Z0;M%N7LU>riHoxvFHzf4JT}F09LjO>RJ18!ueg9gLqPYi>_@b3Vi_On z>Gh|X4ri)0MOP#vdCX%{+R`d>oh^~qGj{nQQ-;vPRMHo=#=Gbq_^1j}n%;-N%4ohl z5ShYkg_jH;?jK1>Jk|UkLMx3ewao+3&@)u=)!wh>3U>u)7aqTt^_nH-+wodG@_77f zA->UzN@Kt8)(GT>Ujkv~B1lg9)cJ6|)wrxfj}95a9e1E%5cl27pn zw6dTXE|BOsSjdWBu3VmF5yjXQ9~)h+`)>omClbjBrzMUkKAu1{d5vpIPV-j4T}&U4 z&Uq?!G8m&-%~s8P(_9|4b3g+B47qLz6LKT>4YWdkA9Bi{KD3HiFUh{KE6ZucDd_p@ zu;?a_&HQ-YQEi-m`~IC)$|Am)QVjI5saB2!L*B{c=VRRL=kKZ+Gt(K91Qpp6uH}FYP;0E{qemo-HOov14b0}jD4%*obdURSq#s5@ zfqcO#1_3eQ1yAlI_Ijoyx7c-VqJ6F6BF2oh=kIjtC}kYIe;eyELvslrx-4*^ehmN; zTL*Ejfz@L+ER+I{d>=s2xVFB&P`_THSDvBG$_U3T@h+iL9G z9wTt7wpO>!Hhb=oyHUYvH~eH#XeFaPNz1T~ksk9eB9}~15laIPgft|=!OYqk79*_? z{oe6Uu|%sphA{F$mD^VDe|;ppYoT{wK~xf_t}1N*ng|<87taH{+z0Xbu5ph% zpg!&HyLRMNt`J%rcOx!J9XLI)@>Gl|(XR%ze_q2Pp!xFd4xMETmo=`Lg5cxO5KQf> zA!)`x-@bdd`?U|pMO5M2SC>f15C&CP*v4{}v*TLHkkY7PTQTbJZsWG6nV8N9Sv4rE z-@XA&TC}1L<}iMCO6VPn08C;GMYH~5se9I4vkl`_@qSGx4%@CR1Kx=2{uf?0q{ro3 zKRPevdJEdpHlMF9mD%PjPf0D1sulU(!F){d5Oa{;@ifPAI0`bHH0qbr?GSaWEOXLM zG>jhnQ-5~X|M^;gA5X2RkieAlf7_l@C^mNjAv8i{Yfhmee*R1hiK>-eQXNZBq@$^FNAE6Pv) zNEzD7HutziB(z33nb@D|#|}(Sc>nhng^puk?<9-sl<_CJR2QuUx7cu;Eb#hZ@Hcit z&7;L{{`~x0-}$OW&4=7R%)joooPS5j5oAM1Qxp<9krPyW3w(tK&9Aa_V+xa%d#HYg z%E;>?ii(#@Y)fNOr7w$AqK17-w(NejcjXHW94i?z8eQ6qG=S=^LQUnOv;ExZY-u*j zS@A#(+oV5^0ccox0iF2SugmV3#=cLQECzp$bUz@ML!b`asF;{tNM)|coqVz$_L$A-8j^nvE+;n;dwpA5X|RP! zF?db}uv?wtWBY7)R}f?JM)0ayMm!IcP>{TexGk-CfDG=%Onp?6R(H^R;6p(Q$KU*+ zzqOOm?_j$s*U1mFMgE-NEmoDzp|)vG8aeDXJj7#fk_+MdX~CsFk(fXF;ChqO9|uG9^- zyF9FE^wCV7bBdEcVXh0JMU-FY9n0Ugsj{|3eWnFNnw?WewGk1~0I^_w-B!Pnh&s6he)1+iE&2 zIXXeUM1(r=?4u-|UwH(uD_xAk6Av57A;x5 zhv5G)_SR8Z?d!j&s3=k*-S`?vccX|%mvjn9N_R-BgdnNFOACTDFCCI9-AGHTGzcgN zNP3>hTEBhnzW1Cl?iljda_qI&WX|vRc|N)36!3b>?^+eb+}IbvlN~F!09TCP9hQP8 z9$Q{RXNmi49X2)@De0Eg($CW^-B*~5le2zRMBpVZF-&4~llq@AJ7DL15(dn^i! zp;nCVhL$;cNvf+hX%taXH&EXs%vca;DUX ze>}ee7svHL4UZr!)*xOAr4+!kEutOsF#GAD2U;t$14%brT6)>SS{hgq2wT5;GZ_~x=m#&BN%r6`gO0VYWI*}4$K%fGms8MO!o z%b)bg$;op?bsOJ1ANZ9OuUYdY{vwA%>^(Gv?^ab;dZ||hAr0WjxL~tmVQUQ^K}=_b zr9z9=cvS=zF7_pr3W;M>k6XNAIm5VHC%;@ z&1=+qDdJ892y8g*^0vm(Gy+L_%fETY;NO41NOucJ0YxteXuD8Sy8KvNDnu_69`ITD z{0f`_rRQ3gO-Bd9o%n=Pe?b5e`9Pa-o`x{*qqD8iSMQ&%yPcBA1BFhVaT7^P9b69C zVc!%~X~r_|g& z#UG5naO>67EpuId^#ul&V(*A&V3j>C5(GEcj8a;L)zoS+WKH4>@STu_*>fW3c4;Ue zM*iOhWEc!c+Mwff0fGiDAP_4})kY4)3!Ekkm2Myc2!xG+Piq}9#$G!A^2KdnJw3w( zx$Ayy$La~^O%~ZIc#;e#W1ST<5G|_CC^g3-Bw&*HlFO8886s^?9LBt|%lc7|JOnM2 zZCNyEIl0U~U|$^1*lF>dtu5_pIbW3f)|afm62Hk_e)Fhk+i;aOBvy$<>lVFY4xFt=FStEWq?+x! zSTR{vPjIWd%ON*-OZX!um>0{uv!Lb@SQGzMwN;93y^1QlWJ-vS{}a*9e8aBl6Io*e zGkh`T$&Y6j=`EKE>_sDgN%&%A&xhqcIY$$So!kK;9+ zrs{h~+BcjeRsNCqQkl73?@})aS@A`{OU-*$JeU-(Cz@}2zQb$1;}cOv)IEOryU09~ zRyiA5+rlkny3gA*rT*AxRlF9fjgWFDf7Oxk=_wJu;QDcIJT#m;Jv}EUXC(Pr{&1wR zEi3|)?MTKS^tqROr+mQNikQs>vL;tz-u!;YABEs(1V+WP2*74Zx%aN_Xrq1WL(sBD z!7Z+L=AMTx`e@O}z)>GfJDj1^N0>RgbWJy!hWhxeTVLvEI|?`JnD0^%H+dL#Z&B~2 zM$pEvvY2&!q3EqRp-Er~vvDtfDA4)6iNp-O=$P^_h-eW<)w^(Qf(Q?FG+?>|_(r&) z$y@mNn$?)$hPoi6zmUdfgg*~IfygIW2By2yNBF{60pMo5rS zWFd%`OmPP8>k#nD7CNDNHvCvgXe>8Bn|eE>F55P0b(}W}&J$3!;>OTyR~FXPI$Hel z4>qy)zq_Oz{QWeH-R{=#mbF?TZXnqV%pPrZgG^S5L4WC?Qf7fj+ z98Hvw^^c3rrZu!WMLf)YXl$(d`|PPqH%2*9qoXj5St)lAr~=WF!ohc3!9Fultg)>O zAw@2#_A>Y9IyBY4-PKgMT29sOyYTHC)xo4Z8?z-N1cwfj98d@hq19S$yHfS zUlzq-YH$(|YpA(_71OOj>#k^_>G-G8cWzxEl#|Phxc?gaJB(AUj*BXLLA-FjMWbwJ z!%bdct1&7^vU!$0ZA!m1cwZ-i2o-GZHr6j*k-SJhH>`H0xH*ShGFdP3a8wsHl(X=0 z;u-};b+)$P(qnYLAle`EP(MFvvg{$UJSNFJTWRz45cGQ>QMef?y75!Yy}+~)@^F~k z-`PI5n#i)N_V@;`kVHsr+zTjYyYLl=8B2MR->VAuP&+b=);?-L^2DsyF#ZLT19)nJ^GOM*cg}t7KmrzN4O3iibukC@a zJl$s-5ABgIQOF{P6k4-#-L3tjBoa%eDr+w!P79>UCWs#pF}eha1X*fBJiug?ll+Pp z7}!qG(CBVySD?7k=UPSDq+v)@{!LGZR*89{$lMe3jpObk6Ou8p25fnH_eIhAGj2M2 z*W!rS$FRpPK#s%fU$jcaJlW>QrWcgR&leOfdMEpzJ<_LefH4 zqA-t8w)k^T@`Rga62D29wIY60fY9uuK<#r|8JlgpL}rQTx;F`3zj(u|3ihI`+JIZa zb7T?oy*~T=#S6{O72F#J{bIJCNBQR}8Nc-`a#NDp;3HBAy#<#Y%W@BbKc`~mD=7!R=o&N(6e0%kWV>?JeVfY0qk+M~i7YIWRd(hplkgjkG zBEZCu9SF(4AFT)?&oE%hLQE5J@>uYNk^oi}{tCU2Y&A)==#~E=Se-)f+~OZqFy^Ot zH3YH8Wxc$6F|mRq;_J;91ys2kVPylRg9>K21KCMOjGCDeV)x;ISF{e65K|{_`jPd3 zfT&bDblmheldvF(1Ydj^=`sGG13{VkCO|fLUP=E^Kbb$HlE{96bq#o4wTP(K`~kGz zL$sLs&hvezY%AQ?-m!6HrL6kC`OITuAy&{xUcXf z_k~!HLca>Pfp-ftZ*jE?(%8Y`8#fW-VWKLCz_0m>!Q6;(6j(vdU&Vh9ib2Ht`Ov47 z|K~(q?I@mgZ9m{82v!6C=r4T2Pp}cKHR6gl{Xhl3>!G1Jx^UT`p53L5cv=|_8_Z%b zmmuB{>lufPa^~bw_d*k4`7N>YJjY$Hbq%wW(hAO3l@RCUN(A^nu&#-5!%4d%M>m{-C5S7&-hKYgZ%8Ky(YfC8gWF5fzd%qE2X#g;g(`a%w_Kf} zoj4Snp1NCRPT|dGd3$)3I(~D>24b1$(7AjsJ8^gSCX%TraHsW1^*y>dkt|w!geN}o zftJ(t3%aiD3pGf3XO53bgI=QFm)4ylCdoMUHP5>fB zx8sU^f;K@j24NsZAg$xD9Q0sp8HM+2W#!iiJE(IhaD&b`Rw#ZA5H+sxq%pZ8VRb54>df(9S?I4*CEw6VzgQ`vVAQ3j)-Pz;ld)%~s(IM9V5LE(8KZR_0jcHr&d+V z^uY1a(%mJ>AD@|YgEkP-B`)+4{ z7Blj{MWWFwhOE)z*_P)E4xZ(Y9>4gDAA$LPy@k5#&}h;*g*J3Nzzl7{Y!+XlqH`YX zYeWr_>sL%hg?zPKCc0rNE@x5zj{TkqrJW8R9$Ny{Yx8eCp1iHMW6r0lEhw7I_-Gye zD(*@m(y#O}n@)M9ZE(Bmvv755i1OSU{ZWQ-{h|du&SvLy6*krnuV}lL#Hy+L&c{|p z%oq`X$qGTCGejV9?KXt;_P=aLl5!C^7#<OhRHh;m@ow)KbmxOoeXH*uQ^PQzLwO zJ2dV!M3wi6ggxTIDJ*Q7*Xs6ZrkVK&LA_JMo4O%8Oj%lrCZ|VT72BBD`IQO1ytCP( z)a?V74StmBc_waBlVvYQBG@mP+PB&18Z!Oiij;r*#%q$DB1|x0LQmeI0;&@kMYOGM zEVB&$z)f;49WLf6m(Vaj?kem)>>Z8lY(Ct0X*BFU-G z=5B_Akk+e9EpQ#>eo=XT_~VdppIEBu8k3ji-Fse+9pxWMhiGjF$l+25OSx>Ty)G&M zWp|c^gE-q(COz@!g>0RhH8c6C(|zj`7L!$D4R&{0h1CPxX)D}6OM4tEy=hJ72KE`2 z>=yFaD5FcJNJc~!vTo0`+jCbU&9iHfld?z&J(74kT2Fyszz%K5+851e)X)4;PX7P8>namZI_gy0nePYBA*7|fI#`LKrpFA7+9MlvD zAbTn(XlI&{WJM(22sj*d+@R;xF`DwPz>SUkBT%l~j1*o>SUfosm@BS__8=FG*8M45 zF-orJS^sl;s}rjp0=#Qi+i4uGbcwH*|R9 zT*v)48p}om?y|3aArTySHPuzvOX)F!X{IbTd~{tv+SbMJtM-U4zo+ND9rhOsS5w3H zG3uJoS9Q`K%OYF5oi$DVoVV>`{&!;7yj7<4hU@?Fb0!1%WJ3>@Kb0j;C`9;VMcn?uPs<<<)Lk=Q?aj}U9b zo@cVumpb!dnk5kdWfPN=ZN7Lp)GUw}H6E;z363*m{pp=OQUaMgb=8dYKOq1cp-Yg` zE(c?EgVut5@)puVfOr}fz|+J1@g+EoJ`dWxQ1KfCVwk0<--&hxej!L8B~gTQ!`H`e zRq7x80cBwl60p)*b!os>K9B64DKd0V&Bimt5$LzRj|os_4y?H+r3k!(0?yEbER$F< z%??+*%cIq7-D+8N%Ay{&VY`~@H^;KqZRB_EnJ$lN$QP_`hWuEs+gm9>Q_OwX3fBW@^6oHgp3L~G=_~~Cmudd7dL8#Sl zLN*28k+iG4P4HYI=lSSDFv=f!-vA}|iW;v{f;18&t(he4&w02F3#(0C^+Tem(sckmM7IPOh*xPV6OVC(pjYSdPp{7rbQUS3{YOeusm zy7u_X01njw1o%cj<>5oJCep?-wjebXra!}F5BJqkBsZdJrul!2ZEAKDG8R@=Yr@w{7TU~bj1NvP*vK#YRdM6uRT!NT;>rJFQ%RKq zlk%`?a}wVdsS;~iz>(rZ-+eZ_fSV{*!REngc0f!_Rs?jNtDn%&zu3a?GUv&6&gIR` zy_0pV&U(;IdUh%pHZ?UpcUbd+u7)S002mJ0E(M{syFI;>eU3cse1DD1vd za|FXYC!+9kxV7em0ImSAhKx=W1nbIW)oo5eEhUML;4nhC#VrBo{jp253Bq{DApkcp zLb4m~y7LkLKH872(x7WcDVBp_cOv+(by4|w?H;X=7%B1gnzlKll!)iqGg9Z|;}JH7 zud{IUxPNIS8dwp4;urf-d($-j4#}UCaAg^@Ir5=K6lToiplz`aMbAFfeq1vtoZz1z z_4uBlkwi$u4hx~v}Xqatz+5_-ZXWUiEr&fH)N z%q^tiXR^6l`;&8-vBi4heZ?xl-Ge88emp*OQ5{cm%&v5L@m3lCF}0~m$u&jP#DiPm zH({KZ?1!=sqVRS1x5Kf zq(vm>^%}4wLUtE6z99M^tS4`~CEk8SY^ccbdp0?~zc|~xrrU`Yt|-dWJaHr+(ryq7 zDRP0_xqc)|d_5${vI<^WL7M)IovCA$I;{@AEO((~rADSRn)rrYMaBCszsW2pbUSM| zY}=u|?*=MPg!Q@ffkZ|1Yl)Rb_w>+V(pv?0-uv6N#V_IL?}?S98rsR{nS9kJ+Y6exy)Lrk zv9SHVE4jd~JuLwX``kH5`yhg+E8&`s_I|mJv%bg7#2jUsj7cuH+p__@2&YxXc#AC@&2LRh!s^JIckcS6L)N+IS~Q z%!qrEt%$wQw}mU1zpC7(+*Cc5940)#(Kcd!+e?@_IY%cMVy;xzng4Q$qn&$!?E2AC z1N)y*`l%P<7FGp?Xyps`{bR`)q(y8&yyIqs)dtOp6U&1_YGlO&rx7-`jBo-XJXZMQ z+X0YuM)p>biT)zEPNOdSF5gG;lFvRGD>5L9-$9&;aHNJKt+f>u^S*a98sDxUTrm;> zd%Y2p{m)ewb@xOFxO#CX8676lFVH}Nh_=!3stQ+!rdN=*7$kan%Kz{2Oz4jvKiFg< zX|dkLey}9S8h(ijVcl9ypX`c?i!TRA9O{ghkP|l*8s**dP|(R5mGaZjGZ53wa}cib z=HuLak#S{;PKDjH@tdjhexOmvVjk+;Eqvc#&%p&l$%!OZ^L_jii?RR(S0AHm&vupEBM_aO-6dY9~GWbDjYIE zspO8*I>kC7`GTED*~u+=yU_-A ziNbFL#nu;mSI(!zQ7CBtt+OvnBgK4;sEgRGnJ&Fc2>Oyjr(nvQbeEPLHU$w?=zfx2 z5sK-X`(3~V@}h)0(&E@49g+wkj2lttS+fEsJDU}F=A=n7;!F4y1C2rUE`^!ZcG$@# zw$kCKogBTqnmn#-V0(9FxonlzsO^JnL)|;~2G2*LEFYfHcP;sxE^H$ev$@=n)f*>% zE6#{aA!H)exr=$q?{NKSWAlUqict2Ft}d+lYm10Yh(p^@cI){`U!IR52S`(WkFsr>8ZoG|~STc3Z*UZeUmuK$R`#oBe-rIU} zPmXw-Z|D#V;C0yNYN=A~l$-`#6;O(mWdxC5o|B_HT!DWO)6m=TTX5s(DtjlNZh6&Q z^Mix6o`3WAK5=lEXA{t%tH%%Drxk}kB{ON6=)@82T)_uW`)14y2Xsd=n_R<3WCCN+?pDJEUgjPAnkGZr!qDe5>B2&$jBgT#qdhi z9Ktq0tt6I??xk|zaq>l1SyF#M7)W&!yVOJa<=B^j86#ci;M1)N(!W;KH2ru*8fGDZ zz!l~9owN$RNj}6s9J;r@&+Y_w+9sR0DMD6-^iL56ZQjPCFQs-_Z!DNo*-ef3Ml9ya zQrk>Ts}O3>noN|^>^HF-JhXDz8MDr0S~=c!)b156?u=bZ=oYC9wIdHATF0bsvn80^ z?711!!i~*GNJxnF96z5@m0#lHYGml%MCH(7D3M=*n@^q5*q1rn#O(1R`MH9PIV}%i zzX~=^_WZ<5wHZHaEdu~J`8Cyl$Qvcy+7%6d{M$KKkSJQ{DFV2xB2tMLV!u&_d$nUu zQ1P-oQtUU<-POdouJP+yLws%N3JdE~PjjcZ8TF)K=|sPILyzAzrZv<`8_kT(E$EjY$W<3$ zPT}u4k>{^F(-p9f3;|W`{X?_hijT#B#?SK&npIF^WJ}Y)z~H>fnWLy_3LdxR22#X(q1+&$4ZfK zR$k|wd^UFcuzRnK-OCDGPX~oudNzZMauWTo_!Z_mU%T-1tjS6ICMpdM zg~&$zE5g#m8DUfyFcFad=94RbYPtoXn7B-~U^KqM|2?P{UcR>r?c6lhKB8`=sWoxu zeW9_w*{Mi$jIF&V6$2EgMeAncn;LUdEpovtsrL={;9U}Fg&t@J|HDZ{Ij+nRKL-)t zi^wDbej1Q?$0+A8lj8o~-6bY`fm{|)GIaKFny&!#@M30;f@4nO_ZMaF0d@a^S0kE37EcNAA-NX5SAvAlrTLr}EiH*{VviU#6 zjJhjBH^<^b+Pf@f=o4)ZfHW9Apu~y#V;!ZYW{@Jog*uaYW90xhyHV zyD3}6n8o+vH1K~5d8rm5?jyx_cO3b_%*0lqPBpEPW;n>#WIi(qa zG6`u)LPQDB#r)=TwBRw^3~Gz0(|f>7cJLV|5UkEV+;{+Ci0o3 z$1``Ojv+jGx+`IgtiKSIvLa@LOa|FJ`G0aIUFY_01X>E6>fcFEQ*XfQ;IMW3l`PlY zyZv9jd_kkXT^Q@SbyEP(j3rCR=MUFkByoDt1yL!v^a;;w2U-TfL9}UV+Jlzp%5d}Y z_U4v0=CZT{{na6%U|@~z`zoDr70-K$rM~ay+J2$_E-Y}bJydp5cH4GhtyLd#znA^; z-ln}q!+q!SGdT~6QK3BDG*qG@6HkemJ-%K^@@=Ouq zmn|gi!_R-yOTiXs8Cs!Snw|sMj?VAk(!CgvEwgh)GZUZKrkjYdVv^n&)9Zjono*kM z(xSeuH}8g`HZ_kfd$WfrTln*fh37&EOP}$}jf_~9NUF-IvZ*x3Y#6{ud5T$)RpORh zBo0b|G^|w!dRhK}?|bD=RY}Rm>ruDM7Q`1+%un2P?8CXht1?xN)Ry zFBsgf$hp(&t*Fc>RS=8F6%!1P6D%2EO_^!mKl?lIB?HB+5`nbe(L~v29E)!qs-lL9 zKPTDMj9fJTuFUsT8}l}wljV4NnS13yoGvQAJnq-@1b3cZBX%+#GGJW_PDuZEz`8g3 zjyZvrnQkehJk5^`WgS6W-j0(CJC%Nj9VaxxL&6^w`g4|i0<}^IJxF~*sGG_}jB9Il zyt_sno)N|UnYH&kb{o!Cw~xM(YUZh^=cbYQkh|=-!_+Ixsy~@V(v(U5TbG(pjM8F} zAns#p1AH_2E2oP<$Y^vX7wGIW2AJRgAABjkcyx3d1}#CzvFdxNwYZi?s^)SoG0u#K z3URF>VnSk^^MJ?dpo6%Br=e&7n*XjuiylhaXyU?|Gcr4CbRheUluzUld}yYX9M&e2 ztv-7Et!Ky7Yxf#9EV`@M-|`KJXN8^7Y*QNpx++@!>9T9uve%_>Mma*wnX=MTzyCbDy|8>Xqhj+lQFiXfswvT)H6!#EvkVmugdy^&D4na66Mb+qRB_;}{J^gE z&FQWx_L*?Kz(PZ5OeR*xGU=tqkri^U1$CVn6{dPjNN#}cc-&O7@SDh_JT8ibdHJ>z z?|ldw)~+g9N~ z#^Xzw(?)tO5ly>^t3^)u#agKRi4x||9(0sY(a^+LpT-ptDcz7U1XbP6?yfTc9Z&vN zvEj8ly`PH|eTtdRTYzKd0nzsU?K2P$2*q>ujr$ge$Z-qttQpD<5ty=MakV+_jz4w* z!1Ku@N|F^WEsOGTa=Xyn8>z~?ckcB5{>}I639ZhI5gBelw(LlkJC{V^W8rAszYDiM zY`9jf587?P2tkl;jbnyhx_p4?bj1(974KX-od+$$euoT5t55s`B!;2acl}nJJZP28 zcxo26(dNdA+|haR?EWt?MtUWs8P}N$gdX~In4-z(3muHBuuD$pewid&4i|qwUyI#Q zrY~SG>dzUACK^%Gp!b;Ra7adt3(%-hoZ5n(l1bb2zLtv~b}>^*x~=Q19PD{V*EZEQ zf~dT0rMk%(r4Yi-H4)7P($9b4A;-_8DmozOfN#>!CVeD%g3AtKG0RcYrI#sEI|%L^ z9ItY%`8+%()}SN95BidjUw{+%*L!1-SpUhxNrBB6ll#`GfD^Zp-?e`sg5gKnRh5*W zH61Q?5b-cFGHN|+CV>w$kmv-6MJjt&N|TN zb?do2Gn8W^YvjwNnmDfR0+X{p#>-#$ALRId2l8k)dyP0l7xZ~vZIsa2&_HCt5;yU4 zzqRvLP@c+7pDZE-r5+uuY+rtE`*gD0SDUc}~M@4gn_?F2isBLI|F zV0vO@zWa^!vVoQweM^O1GUCTZ_|5WuF&d1Uwh<9gs`-vVCX|8@n&LxQ9ynkip1Q+~ z2U_5nrP#<#q3;T`(-1;gTM{_O=M*3{N47+s+*IXjuiH2x|C9zHg_$%(Hu9~X#5G){ z5$3S5%;5+$Vu=z7y~r11t}982GKoredlqO&_!wf&ncpS26Cz>_z3cenB!L>$m3rhR zHsJ#AbOFS@GFCp{`JM9Oi?ffGRgYVSiOxxy^=0J=+zo!^K|a8;F1(AjF~j|KYi>si z<6^u2#6G{W4eyCwFU_{I`?TFlgN)!8FFcVJ(+~wQieAGKl ziRaEa2k5iPs#3$l^Ya3z5dv*~w$2{)&Q^GlaHR=umw8N1>)Q^p3fD$p6+NL>BO%E~9o>l7Cx^>qzUAiA%%RH#iG zO?q;BL$hg$>1ee|AI?@&SJ&amzvA5U7`(5fW?rQ+mChnv91RZp17%Y-)eo0UNm#(N1+~wl?FY5 z2~?;-R1r~B`%LLP8yY#X#)wBS%M-=g%d;=`=)zn&^VcqnV;7m*e3J z&aeSHE)t)y4V)MCWL1NFVj6h*kdR-2!q-%|K_zxH-DO%2KMS^{R~r7(;iF!NI$4f- zg7|TPlUTkpL*PK1QCDa`je^Y5tV#<%KPcH{wy?11SlV6_8!KazmX@vu@zr6b_*HQ+ zJZz&uR})b+_q^OJkaVvZoj-$Br(vNkVU>ZU!Tz@0g04L8`j@8&)$pWP<8*Rg_sD#Jo4;TnLC zVgShC{NRvW6q`aD)Z1jz40uI|l;q_oNnUTx*vg;mJHoqmsDp}G*cC&kScW8#-Qw!T zBR}(dZ_foHhMeH%Ri{~fS0d!)+Mr9`zwapaKeq7$GwRUnY(MOmqV7Mg5{Gv1i^7YW z(d_DM9It$5BYIpA)q*}oIuM(in}MSRehVi^0=5~kv{|lbDrC&Pm?M$)6@GgwIca_W zQ&orv{))H0N@IG1Ecu}){%*TdLUOXHwe?HT=YXrSw!hLYd0;>lskx`8Z*%WHZ~qa@ zRG@O3z`Hs-J9AqbCS($e05Zbt!uwy$5Wf-CiWz$bo( zv#AH!SF50=1a{eZXW=tEiHc#>xx8e!*zjgTq?(1hJ2_^I^A08cg$AO<&ni2Z9dP1M z=0+1xq`nJsAV9Me&Bo&i4j%{O!UEa<`BK^s;gu5J{MX+~_902sET&z7rbHZJ`ji zg4BKTS;qugCeF+lav}w&)qSa-6XD_6+uJu^dZEJ90@IeMsi|bUVZ^nKhbghImsgVI2UZu#f#JgwkL1cYGs>?yK~jFKM71ZcyVS3ObHBB3J@+= zb_S{C(bwgjDt6Cc_^rt0L|jhC$J3EV3UY>R5&-zXVrBWJI6z)8YKY@9!1gu4tvf|zIy4+VZ z0yDW1i(X(hh5!dr_=(k4`PT93Isik(D}nLd3qnvGiV0#;krcWQPPlOtvq? z8_ldLeW{vXmVuWl_nF#DczZTy2g`ggt|IF}Tywwv1oy84F)on+vm-&xXM+?UHCT#< zdkB;coNgt;9(2y!D92puQJs~~vmR=aMAoL)5fk!)SJ~mY#0b8ZiiZvk$Ov3FhWd^q z&)0po{{sw(NJgy!%ny6~ismtNQw(H<3$P`ghtd$ZHy#igKtRdo%!ANey5CpO$rd*A zK@NJ9Wsv}N_{8CA>`4c^ly+q$me#Wg3;P+Uk63Y2T_%=QISpXDqL$r^z-I9m2P8ty9@H%AHUcRedB`q8c|-DM$M5-s#^5y%Gsjej7Y z3PZClVevF-CTLp0$wMHY7REuZSEml1=<5NdFAng$04 z^Lu%E`udOxV&A>#cK914BqWN8ihRdj&k*iI^12*|lfSWI_}eILCwI@r6W5bY%Q9D* z@R*8LvI({U1`YlQNW|Eyk!Ln31)Hty0$uT^M??sAJZxKIUURqGWtM);;LRQfD<-)b zBYnIWH`2DA7=O`^wOYbfkj~X`Xp8Fs1+ETCx-d1oBqpWrwW9?{D?`~oVWp>e8b^GBYm`bQoZfM<;r7pu=io1Jri_P2>aD*!63Kt4PDY2(-+Z^ywkd}*;K);cD=;^i z`+_Mx+T1h71WKSUGWbLz3^|nX@Z7nhbnVquB)$zrihBhx$JN|SAs}|4uj~&&#(-tm z2ND1RYb78W+2#Q{f*Z^!5Kb4t^1Rkg2Kn~r=yvpis`a>Z3bp1zzoUA6QqNwusB(8* zLIpem)6`A?pHl+>hB$Exx0 z1h%*|=POqWU7oOcr7FrkJ{Q{WR{%ZCy#svEC#`isL&n$716cZEDAIBxu8YO!9YzJNCM8xT4Hk;&KlLAse?>x>f^J16pWf}cUqZj0q`aW_3UL(r0a;BKFD^XPMlN?ceUHVL zMGB`nlEk(M<{pqD(SHRp176M7ij19RJfyhTn;A7~xIpB_Fk$!+v;(nNASG48P(1|0 zmhD0{V@t4>cE^u4^n#wv;?ffA>|$v~do^%h@$h_y#(`(MqAiZM(x63R>X%+zYw46N z_1ERW4{PaAkFrV_>$&|3ZoOISH$OO7|5^`3Qxr@k_y>%Hm6R0cC{QC<`pQ;g9tC5zdAc^^ktphZJT`Ss6S z)DryMN>v7b+W!7Nf4>rA%#9ei|Ig6xy0fn#=BXv^`=U4V9&I1|#jxr(wM`6aoH4m{{r^J5F+UH4t^ZVSw9%OGp6U3J zv*#5;4#6Go-4d|Xm5s9@5E6XuqM~(+i#u6aH|c$Ut0F(Zc|h{@eIPKBk@IgN9=RPf zU78c}G+u&S`3(}BiDcc+EZI#;GcQ!~yueh)xE9#W0Fv|&qSg=_@f_02@dVr%U!pgq zLcC(D+4{|R*r#3}z*}tb_;CYK_UKwp^Yc&X6iu`c;M=X0x`ssAXTCG9$xVzG)>2`z zJhHqWbk+c9x#Gv3$CKA3YxHhaCabCC6yoL(^_H2~7W*++#)pZom(fioo9iOxjL3{i zivhNh+@;QcMO)*x~dj9J`jFZH0D zhA0NT4{n&*vjf)aZnsg~ssply*rT=g4ce-wbH>aIS27KE|4z50->cLP>iO@99*UcI zd3kM$G=UmGWNFYy2LDBdLghn6AnAlH^bom5RLPKo(&^7nV=<5nZV9 z&w3n%R@CnMvbwD|yK^aeZ*reey@QM^iY@fgNW$D}=kI1p!MBBZf=GE+FL`O)Jr!k1 zgD5L>z0b0lSdRQR>7_d|Z`dN89GToH0f9P|WO@A9kFjK{j<8&)CKqnHf*DBpYRX=D zyugp5F5bXRoUv~>(T*0VtsoWtOXS96kZCi9o49uALcouS$}sTvt1^)$Bqshl^3N4o z`^NnrW`2>37~zZ$aE{gmvXv_E(WtP_#ZjV{LU8q1$iknOmD8jX31H& za^gh$xLMx2XvA1)Z#GEclTFykhR26wxTKsR{}{OD2D>X6ab4${Qi(cxU=M!q072YA z78-o(5SCx$OBWMUzA;FLA)OY7-0sb@{O4PIbTMi=@^883w|5*qI$tesG}k5ndNuBB z(BJycrZiFPRC^hDU6C&;tme)pZq-_9Lt2w5q87SGA-k%pCZY^@Xfm7C{}6yZ2yaTF zv2HYAAu0#yG+r;kwI)+b+T_Y*iD=^ZyRN2slOz+F)VUOF@f$N}wU)-bqE5_shTpnI zu><|qq!q>Pa(;H|m1|_vEz=&mdmbCThG1o(6Ll{!Z6~B{H+FYN!`y=)mP^~LZ<70$ zgk6`3UJ%kjy)I{pu!xBK)vHd(%PPT_M*9p0RnM9Y(i-)Z(56mm`=DvnWP)GqADGQ$4wZyn2;S*zp=f3qB@1bg>#QjtrQ z9WNvp!tM>X@8Tj-$^!7;Lyqe2HQ<+-@3;aJ=kxz2M7-0~-6r{|kR@LeBA?PdoEkf;|e=@WErXA{G&w@h-AMB@{%gap1$H#yy6xNXRQr?Hu zl)op(h)&dJzuRaLiLZH{3Uis=XDp=5sKjH_@Z?ATNg;c)gUO+aJSL$>$9Hg6UO_1Js^Ae}4Iy;$J=VVMThSf^?JpqyeT~ zcO!w#yJPdMhP+I7tL}h?)?u3W6Md!;Q&IVkxZEb|W`U{muEBx4&98O;k$r(~qPZ^< zKOZV!AeQSTMkEv4D5vZHOk7??^$;#-VcB}De+bIZ=lApDn$D8hPOoO7}eCq zEx&N}b<@h;{0?SGZE&zolXS2x1+&S>$RHCDQgntu2g=H?JI%CiYuog~1rC=UQx;$b zkTJ1SWbW3>#Yh-_*P!o zTT6=GqG$_UTx?QYShOkup~bpxZ2C=bld_yvB#)cw51V* z9H_2Yp5YOZYD&CrN6yWNXv86(@Oj_}5j_2pmvNc|7Bc2nY+UNGQA_iZ~@tD3`qDE_Ukh*bY zQ+E8AM=$6DBL#tD02C?T68L|RY#kbLaS{f1Q*T; zMN)*j{*9E*JPnsan%*Jsf2*<2PS8+EZJM%q^QdpramI7$H1%g7%5NCYpmq(ui@F*h z!-Nihl#6m;bJi;C%N?wWa?-#jbnVqoi3BL`Xdi`I0CGyvUS@(?S6bqCOr5^_QVqd>S6i4tSDFYn2z4 zD=$DpI2oBsB3K=Cui^X_NxT zXZKs44LGWDnP(F#>9abVA}kBPKl43qR}l+dnck-pE+TQ`W@@cs=}3>gMjb&MAuYV} zE=K3!!`~B0+N(DNfX;zc4zZUZ>-*!XYd=s+PB7MqAf2iEbsr8zPaqR;y6)GfP5Fdr zNn{rxsImLx59vUTh|(U@#ooQul%W043EO}7vHyjyzRJv+>I?FF?=q{a<1hRFX@>m} zsjf9 zc?Ka`D!GbH5OBCg)=2j1*aD7%1l>EcGCAgC@hr`TIHd%D(cd*bz<^v0t@!-io9iH* zIhCvB$w??*zG4%Yz_0uPHuZl5z+d_;{NJ3as#sdOk)#l%NAg9+D zAQSMW(>kcvI7f;ZXnZlcUT2=M%^p=G0w8@w%J;G<$DI1hArqFF4Nk*Waqo^YVz1E> zl7JKqy3F!4AnT!R?iYW5RQntR^V54Rqkz!*2QJ_e5@H%-fP{IRucu-0z~r!0U3ZxG zHotXI3>?ORU!v7s0z(F>n@F;#I?+bK`fmh=lsb-YCAW;k#kW zcBJL!quKq_dOq6qT(7f}`=Pb@75Q|N&>pmd`u@Q}qX%a3r;o$(quW6ZC}aE^caA&ZE5fycN+MNHgM)_N^N ztTx8-d|FeKm?6_>TXBA>$t2H(z4LB$7kIS)|15O>@MFN@@Ey4^7kk4ExenGlCn!t% z72)91Eb8!Z^EiKSr%%3rdb!q8uz^Be?LJOv9s)f;wK56Q2JQ6jH)L1woSdAwF#NgA znM5A4(V$Z;0P*%hQSQNn2!fo z6*LThH+OV|IJ{l9*EsTJuDHP?MK@Jl`~31;N3rtUh5*%dQ~f^*aSW51yp6HeKCN1t z?;b{;liH5ejxgZe(o!O=V6gqsNb@bDt&wD$y-?~X2+Jm6z@#D6^(3$n8-!D7$VHq$K6}t_W?v3?qI+=z6z^D`m)}eU3LOt)uH6@}HF0_+=7P5rF z5BU}3qc(5L+gn2#XkEW;|K62SUkvT5h~&rln4gXHhLUqnC_>ig_f8^)zL$nt+kH_` zW)n%`pn>Bi0G8ZcL@f>8azwrgfEWO;jJIwjCMNz8*{y-u_;7FYF^p+2wQ<3)*wPaC zF!M@A%73aztGp9`VU1q+?2LM0LR2w)9U&fV;+wdX*!2(Imh$aMRAEP@x`#_DY! zLD?wc*ydkbYst_+h$34Ss9Ki20Sn|zeY9MYI z*?aH33E3(mWW`NZlo=V3ohY*RmYHmKwq$#+@ALb=$ML?;q2qXtMc;GisPv`(;q3xgUF2Uk)nQ_Kx(E z=GDsRjYnVEa&FG;FKPeTwka9O^qp);qs9|`BrpJQg~inf7SW5oV-2aXt_4OdYrutK>=N5_?D^E@#*|7((xn!57>Ky!fJd|)24TT91iWQ#^UyF)D* z6ge9m!C%NOg+Izp_gP9&nndJ+e)L}^Rb6DH7}vCn8Ob6o+&|<0PY_24g({$-j+8&v z*4M3|wTBDE^g28w7Bf6M8+$>x{cB{*@YIz==Y6+erYeJ_s)ph(Qh9!XLsfGO0|jVx ztDhzIZ8{ch*%Y=({VBF}ott9rXtoKhegeKuHB)KreEAhUP<|s*1>Euv+RQ3rVN+n&zl;-G)qrQ)h7;hYYPf*!zyxKDl-8=?3(N004;r zGXobMRt=H&^(^m#{SX=QXS}Lse4G*?x;#8QhA=Yg69LF#(gVbFSiJ!KH2sft=+W_U z?f!&=*!jt>gZ>6E%gE>&@&_>lIs8r0*bYNO%J2r%6y8@ZWkOm-Cg*Fxq!wV=z(50~ z|4D$fU-lZ7={m}gHjqp$#>U2$dJ}4E-@2Yf&WXd1d-?L{{Y`LuVyYR$#RCAdHi+2X==LC`0I?FP#zkdh|L+KXVN-0bj4{o}ajb43fCk z8p$T6*=nH5^`sq*#1UG5`NE*aT!{|1sF*0a6DlF)paxl=Aax zdX2lYubDaK8W$OI>E~qUH9{4Gq_WJZ6C|yhYZ$vg3!;2)T@V4_U9Dla-nQX?f<0rd zQGiK6s`uZz^E$2YlA)59XsPhq$J?<%-AKswT678AO$)eUku(XB3IfV`%VXmmQp82W z=Y2Q*`9WL#x|7TNJpFmiCdTo<*>7QTC{D{PaDnICR{o`4a2+wWVttXHm-lS3BfOzO z9K5|yl5A`!Owl#V8CdTH@=1v~s7~o%&-Hj@-!u}xkRj)Pn{q5o|C0^mc1C9nBsDJr zpsgk%gXvwJ&hkI2;4TfK!m5(l#{Iqf%XzU+uY%-X$d^chA_CiklF>9g+r7vTT_>)4 zo1Nego0WmkpB?d@gWaplmsVL}LD%Qj6^*&Dmjcnk zEIl+Jx`1^sl5~)V5vgP$(jmblt#q}t2iH$og0j&aOqSSf=jtZfCr zQ>OV6Lj0!&+QkdxV;vQj_|v2~TDG=$K|}!~mmHw~M`%7c@qlu0WjI8cSzCvJ{uYK# z!~w7Z8}PxaszCC6Ih-R3s#z+7JOMTtqn|wYVZZ3<-?9Jpi$YD=eZ>lKbXW*F-#qUb z%Gj?6EB>PPpBIm&D9sWWaQp_SVa5Uy!u`E|>HDA!cRDx{NFfxN_8|L2(@uCf7%_6k zhQFb2^EYDWBbcOmX8KJ!Rd1i42AyYuH*dn9CCvJI3uUNq>z4%CaC<%NvVbSwUlOFK za=+Gl+U2elr3_0W2k^R%JNFenQv^$1#@7tbdvfJl)w;M29L#mtIMEGq47~r;mO1dS z`4>7~yQz1tNbmK_4%4w(M-MstJ+2DOzRE#n10&%5pdNg$R7T1sS-N)1aXly2HK9pL zN-*P3W$1^ndh?L;lft;w<)7c(L%Ww~H!Cndqd!D*^vrqK`}Ha3{QrAR1m=P;0z#~s zr9+?%!BQN%>%gU_b4_U6)+5>xqGNHz18 zYRA&Go;Z&&=Lqyhgv4M;x%F0J-43G{6aWW6rS)EwUE^z;g=7;!w2H-SoAA7`Ttxwo ztj}9dGNUzjrnbJ~_y7NH)B#q?w~gJGcr90C_w)`_>J@_^y02T}tJc-8BnZ9uUe&g% z?;P6u0EF4vs9mPxq2B{Y!hl#Qp=+^bH81S|=GZ#H*}M&^TBzS) zObL)KL)Y&($jT3wgSSlPm?VQ{uJjj-l(vnp2+d#Vk3XJ#+cqq&*{&(BrT2MVT_1Lc zO&SK1nih_;^sL-eZx@Q{*r|I%9Vnu{s9PcUS8j6i$)q0!m0$LMn;SIx#Yh+D6cd

m(OuWALUq!FW95V9B5Ow3~z{YXA*R1$JajL+M zJ-$w-gOp(WJ+{+(OIwU_c9F|>mA<-38rvGlfCu6m*f0_UD{PTJu&qGR5#tV z2_LZ#pM&i};IvND_TR2PqfK9m^F5XGOZ{;-zDa=oz2FmuzLvbAZG?T;=c((*aX-a;Y=ZyKjgpZH#-mC$2n#a^Z;~DX@NZkGdnks zwGN~asGE(?v+Rw%r+#UCu^YINdh*d@Hil+hmf^Feb7SwKx65lMx@)AgC577~2OSn# zHxJgqA;hE9t5I}tWAfHrBY<+duYu{t!rpn`%-c3Gwb^s*paAP~5fS&<$Sel^OWfW} z>Nq@-sA2n&LhqB|Qk%+I`v>nvvjdCA3rF;Mh6&j9xo!*)gonSlNXO0@9!~JZ{mP%$ z-OK-E`4c+WekxguC#0~H=Eo*Z|Ln05Pdn|ljM16Ny!(eY@P~6P+0}A7__`PEbnwtW znseZ$)7~{(jtTu(C*HLp;w8Gn1R>hm3)rme(V@JP%sJcssr5HP=3nL=5l<*2_)Rf>D6siaq8jo$HmHuXCGdH1`*4<95qXTA=A?cVUnh}k@NoYg+b&kuD8 zcUQQ`gqLqnNd9{hH5`%|c*J3+_vVgKYJj1y|Gvw%pGmhF$qZAvH6PkK=JKeEuSYn1v_Lu&nNDLRw8+6mEWk#~VfZWFtQ!l&a-$6#O_FS4Id%h0{;&Vc_9td^&3$kv zGtGYy$F4@Ighl*QtG&DB0rSiorc5ENY5AtcSAA20>~9n*`Tyj4ViOm8*EWwwP4CTO9Cso=F<^61{bo|Nx{-Pk3<7CxUpc8xSPp9iVA0Ox|L;zra|D*7WXR_uf zbQ#?P-(TNx>3<-yVdPuuN~9Gc&XH-!f3rcc;2%$Pdh{2Yc$(v8*ab$^=up7$=;&`y zWFW3(QDI@1F_t-=n9&2E7+;lHv~iwKrgp;aCTpMAMA1vok+CU$gSI4kVqzjh(%Q+XIKCm_sfQ6Rmgt8Vkfv>Cg@Q+a zmX^^z<-M}z@fZ5FSFV1L{#q0o5^@3iuG9z8;(i5(SoPwmWn(fDwS7%m3rjlvuGmlP zixO|if5;AU`7#pN!2Ew#VULn2(Pq#V}w=CYUJ`Tvi%lxDdijSVA6Y*rwFImHqx* zXQ7YL*8)z5l>TCluIaxo*;${wvj#_`?Qf~iM5Hy66V1f7I(TS&QFdCkx7FTPhL}-~ z+c~ys*WzC$^cXqLGMv}+N1?ta>0xoFB_~$|&e+vN;TkW{a-DqOAdo1;Q%fc1r;=uS znIp*XjUeS=TW2x~bzQSqzfys~(iu+fZT$+{9k<&CRUv5pxo8=api|FL43I}E`lB@x zhyR`S$}4wjuz>3z+5lr)Zrx86N5Ju^+ckp&7c}=l=_y0t|Y%wnF zT@OhsD^^ZUPAn{gYUdV+OUMiR4Sm_Sa;vT_uv*m9Kslr)@fJ5NbR?9)!di>gD5?*o zrLu5-?HwG7V_Qu`jE!}4ULGC!I!@P-gBR-ik=Q){ zi^XNRgl#m&$owo;#*CvcB_)vDLhc7yO38{ZcXBI1N=wJbAKAq|3|*cJmYwc1$(+V? zL`RA^e_QdZuM!X|4XMU(+LF+c#+xY;q`$*8dNCux$$hx}=$13t0&R~~sA6KGp&nzx z6n~qlon1lSgZE2<*R0&X8k`Sm5}a&!}K zU2R65?rWX2xY}!FeKMrk`3;tcZL)ObLLmh ze6~^f{Ja|;HiQXrD@qN#=P>R~3qAFnFhZ&qRZf%UARX2%Gsl4vMeIK83%0ejmGkaS zx^2Mg^+%W6q=5!55+bw?y9e5bYie3z9zVF9m z6A~Us$(hF;dy1*zFp3p2cQs7F4Z%WrJ)ccCynBPGZvB$bby=L!)&3Wa^Za#ou9FN* z3OwV-?WX@?2#km=7?y;;$MO3x(gk)qSCv*=4=V_nv>Pt4O&N2Ap^xzJ@bqz)#?vhX zf#HVKxyft#?HL=ri0d!O`lGdvqhsm{-1-&ln7@s1@Z`HDs+`h1?*-1krX&&S;w7|A z4cne?YnJ0P$`!ME!b19-_#%XNyyp1v2{R!>mT8^cPD0SgK({?wuV*Gbj5z%Av|&|t+K|1QZURM-Ldmtu z$)|jUZTWfcU1!~uzTfb})|)6>-b}VseICwE9k?klRGir+vhdRT51;K$w(qUQ(iNXv zV&~@AwLGdfam1#%_uYF5EtAmbgDEm3Ov~cZxJnL|Cvxfh29Vfc$8Gb}-0Gh+-lMX`YJ`?-oy{4>QwU7okWsfY`k8_1{d+qL!9c zvWScE=ZD{)W8<2iogA)A)jk>gml1em$E{bM)6upp)s=Lc&-@1-B5LqHJ6taFTGu~; zdvyQ{27?PU^7H=qF+TBidyMxSq5|0A@dxurT; zM^^{g5^oP%X!@(PM;tTl(&&5jd9QYJO{F7t_l$N=v=GBly7qYuihrVsq_Fd`o@8I*d?q zY@&>ej0SfQ3uI9^e**eCv2+i%;ImqEC-dQV<)V}MtrVmRmJEHNb#nuKsQLNnzh(Rk zrhS5xg4U}zkNqWDznCsWmB^OLQxoLjR_`R=ndJNE<-TUU6EgJcQ=2@l01d4PRjdQM zNJ!+{>NnEbc+S^QD5@(<*Ais^F1(zIz9)h!^?k~kW42@bgQiyFBePIN6Nfcdzb+S2 zVnRz&5|Y;fvUENBP!&AUp3?ttKF(-MqF)l36}KmnTpGX?>Ptw36GQpPL)f!_xH!`xFMk(yg*%x465&}R;mpNB7L0R*^M)cP(`NB zWV$pmA)0I0?HfzgVga>>X38dX1flORcOkwP1pz!oCf?cESrpXO)m0os9b{~591|HC ziJs-+;*yq;!3iNoV&dTblcnwy4rx!MViClpu!6NNce~%-6R|toTOH!F>O>b!X8$VF zXBD2+=1{*m3SIY+&>F+W&rGijKF__7mPJXaf?Vh{h*azv$m zw6_|(zdo4_n{ zIwi~{7b!hE85Qk0J+OBlQufX!Y!hEpZ=hLr`vUo_uc%5E#kTshEI85&3Wa(z`hTgf zB&~n5>u510SY)IlFP(|W-JvN2{{R4i%M9X9hme$`!5gu*gBvwj<;3T?rirYN#+R_E zDub0_V8VO_KbCJ(BG(~0Zb>a54N%0Y1 zPAKJ|j#C@l$&R^8099e9{WP1>5OkH4_hF(Bze}goI2w-4jR(GauETe5f{B?p;^e>I zxN2E0W!t0bJGgj)+>Hel0xl%Vo`p%A=fXNH7C zZK#IgmIMwD7F+Q-ZmA1;)BMy><$?jp_s! z(L>CSqFi@8^x)?vqq``$b$=Ba5g_#<<_Tnm(wv-2Sj5&02{eNqBxzsuzM~f-6)e~4 z!i~j_DNg}9Il&wuz)pip9rR}O+dw6D~~Kv#lydi+lo zJ`$hlH>|zQS^Cn3Bm2*>XSmqQnrnVd`c z+_&29r<6}w-K)Fd5K4j5&NblvOz+eiy~iyJ@PpG`PaIYUj`M?xJ`3V!;UD4-8dY2h zanJt6L`FQ-NXk#@IY{Ck`PBd_olJXX;(g z!g)Q9$qKt*oGUjFy;~dXp2UB*YT@k1`xz?S1K14lLZDZ)d7v(sNz~<#iD~DjEPVjl z{qWfZY{N<@nrG1((*7sgZ6y=Arlz6YC+5ztE<<5o{iqF7d*TpgzWG@7^*L!%`aj|0 zHU$ndO48TiW<68fTHV>o!*X>TEUAISacQ>`_;7YqOf4>qT9fpjk-%AfhV%D;QJW*Q z?K~&>t9iK@s0hDB>$i42GsB}n3($ds?2mMjW#E2&?`0BO7 zZ2Q2UBla~YZkk`UK&Q-i*X_#z>faczSLuiB|pTTXYuj#YMt!9$Yfo?gnUa+ zx_k&Ze+1l4ys&GukD&n`x~S`ELZs0y@V19I1$~7``1^clD(`) zl$liUaG%c_ekIg5+Lh3J*z)7y$awm1H+B-Wycb{E2-h44$@%d;$d6$uxxl#qRwgD+4_J>(Def;CMnu2yRLA`V2Ln@9^4i*3 zj@8?HPqhcV3cUbF`UGK^b)Dj~t7sHJd^d4vJ%~ZzN%Lo_Rv;haLerjO%60p8DEcA^ z>*Cb{%2j=jR5x;I_Bp1s;s?hb3t@~yVSyo7f{d!_S$eU|>dL6|WdLJb+AL4PxFL6_ zUep2Mlbf>(tAc!N98k#pHe7~M0da#R}>mr(+wNji-1J94)^d%Ab-oR}Mj zL~cz0l(#y7mmK*gn|RydFA4J(;=B^Ofi#>{N%;sY1~DfAAF1$h4hvxp%a4V5M`tW0 zZS{1mWh%A9{>0<6-bBHjDAeJd42+p6+Ff=L-Idbe|Dd)ouH$yefvyFY_M!Ta;7Z+w zXnKJ;p?veR@Y8(!nI={n7M4rTA|nYJzerJ;TETCf)%1g-D>j-qH?2b_As~oHhNMwbqzC(hYf~y3}f^tH-;GadAF^ zqGJxNq5K`%|0q6-R@2h2`}RA_ji>tt2rK9f>j%`8j~f6PO!wdE&~5M+yX(2OpjMjw zXOn+%R^Xu93BaB!(WsI;nwV>mI1@5F!_2GZfqT0LoBf7(0`FCP>uKLzWPU~XhPgs=*h^L;WT&8IycqZKcGFM& zlNtSjxRyscVRLO2L#dAq46M8C@`PMszGLwh5L3Gp#if{tdWw->ydc?r@nq{bf;FrD z2h>$0SW1t--V>Z1Sktca8Z53(Mlf6|*s!_U{~jAWYR`Ycf`!M2rx4jmUfsCuKJaDf zCy%{r=gu!E%dvBw%~jMh;)I#p`j$%9l_#{j6L%*hrCpnZO#j3@Fsac{Zp;y9QnX5* z4|7>#-=b?e?K=NY;!9bq{9+p%Z7A2^3j<2RAHFg(5u`qw(Z14MrX-n3zY{LP_-QX& zEbZ??iSI3P3K8;z^|@}_aLxc~y-)!Qiu4nj&8fkiOufnx@h=pR>_`@GAg5C5sGy^H z?o<5!P-V8g!X8{N#MSMEki+ON!0wcf&LAtjVvZ4Q`uukilpubBnfAYXC{OVP+M{4_Zq5dLKBH(-+zS-C z@a19j&(dhY_`?|#>-6z53H$OuJas`Ql z$S7Y6kPaMr`kX_84NotGyw4N3{VJ!)0YD4=CWvsC=? z)K9ST?uu=H#2&dWV@Hgivju; z0Y+d9jPF$Zs+4*7pe?;V1c@uhidw{36Dc8}ENtUozJ3pc+ZuOFzIYYeU>s@MHHCT} z_%z>_6V#RY^vQc~Sx`FpwOl=Ep0l#vol8w+mM&F{ieC-{wKtCJjZ0rOT*;7oe~`YI zB9ZeGC$c78`E7d&;r=8|JT&Gu8y|vwh74ZturA!bsN1$c&XW&rP4dkKDEMk>TNApb}8f7l#_RZ!7>fWH$fX zxj*7r9}o7*QClK7Ps`rM^gTO0PfsDRD_O5whr7gj8#9z2i}B)6%U6D@en&@ZRghe} zy_<)eoctNoDeC$KP_*qAS+V@)4oN{_4O+>K*Ln4Qa@*cnQ`uyQ!*}4^p29oZi_}bh zad$&kEGhg0$Ns!uP?k~3pUhxd%G@UOD+~K-2UCUw&4g`VbcuocgzCI&-t|=z?E6=3 zcJ|&s`@!Qc((v@*_aFaJAc8LS+hLwXc4x5SC5XoL+l7qMFD8wFqATx$Nn<>l=XlSK zXU}~boJWwL5xF{6YBI>r!P;=L^H^uydA5O8NJvP&>))h{gwDEpx#mzWwj|JL@Uje( z(w=vl3h+o#kr-R$)EwZf4>Kwz5I-w%AlBa#R{arIviZ7|%b?B?H#~80Zk&xM{F(Oq zil(eHui(h2%2}e7Z}R~(Oj%Jl0dJ`G&eY5Z%vaR(q2L~loHqA*u+cMQRc#@7vooF; zD@0DB*vI2-$6WWbjQ6TBDK1B&<}OS~bJZj_F6cEd4?N{!&^wA>+I@cWHkLGW&P02{kxc0NXpT& zx{_=OKGsEBTh}3s_wm44jnggPfsEiZv;fbHjR7N>+_R(Qrb2kv|Gg5tePpF8pQ~l|wsXH`V;gVs_Mc~%1n*<(O4Ppktj`&H{St~}J z?_)9j{|?bmJ`tc#2^Ib`^*$vNN`DXF7bK(?pVIn*{sX0vd~N;H&6$kigy+!4D(Uqy zy^xXC!W&Y4^)@7=d4tF9WLYSh^bbxU%J%8(Ua1{Kxc60JlBJ&tt>lR0Jh}e3$xPjuqr_my2PF zKC9meI+%M&Tlo)GS;LAe%X&KZ4J3F`wG)O)iSD#BKL93DPp@eVdSyqbzT=4F;uM-Z zuG1L5h`Vz zWWB+1M2X;dT8N;OmQTjbUcLu9RlY0Je@7lcfJ_+#W<#X3JnpJ?R~1xV4{ta;1imhpx5<4P5D%3FBR_(3J!`DaH5Bo$ehR zU9V!7|FL9qwXa=PDv`3gQ=Ww?!TFoV`h)snr#+dcDEYs%S!V1EGz5( z5`JjxcF(Fd)ho-XmgwnSd>i#S-YDDW#Ml4OcsS}`>>c^8~yRWqL ztjB7>8xy?xO!azulxL?Zzg6ueG{h5X;}&Jz0<)J_z4e&{gvF^)c{Bt^#smIFbJ`C9 zm>bg=_Wd#O68BW*WqX+;?o*GC{WWTF?q&99=UL4UyqZ?Vo9cgWCMt;z&EP6KWHfd+ z-UOUQRwmI1ULyHKgs`r?TRgE?je8Ij!?L@qFpEjtfYw%1fEP(N;i^0#vROD>oKaYHF& zkDlKbup$CzX9ocfD(^VLJ3eb)gl?xFCAdKvQRC88=(xH=P_0*)ARv<9K1!86M38dr z^3&GI;=|y28C31?NlY4Ze0ccq7AEh$Nk`=CLmeV#EA5!a>OA5bBcb%u@o@v|{lQ8D z>lI5=1iQf`exJSMSeJU=zN&q)Y!3gc7GQAr0#CeYJo+E@ zB&E-MGQ*~W6{+_Dd>XIwbm+`n9WP^s)k?3ekLQwzncn~A;Ih1&cH_$01<7Rck&8X) zRjdi`vLyFh(mozPDq{@!^nxITYQdL3%PxFI{+FhROxv*;{YU9+Q%8CBjZz^iv9>GBUS0^z)+A3pNhPmt)CT6FJjzgdnfW&#(Ph*+kC#uDo%ZMl!=0n^DA)PP7G(u|JcjG$ml#Yw@k7gqqtqC9ieP&z+}yuQGmXs85U9Ud90{bI8EMwS%+^3 zad z)jhV8Rk!u4oukXk`4OST$jFFpod?Hgk=~G{AIL&}bn!Gs!7!4|2vNSMl~3S+;8oIR z@88ui79`r?2^17sIzC!7%;(NGzm@K2Qd<{SRQQ8iO_=}Y>{)b& z_rDq=Vv^h>iDTJEMjt-m)l8p>PE!};mGBpd_0*&iq)0kr>qn2VqK<|f!vum=6tFJ> zDu)t(-JY*e()~ZR1bD4n2MxOf7Ap2I2aWui;-`BWgp|DBhCGA7L=Zt<<73o_9_*>A zjA3vPJ)P;#=G#kGWP~aBOW^rx>gXWr0+fI5|899z=DqzpYT_;kTioF9S=VovkM{RJ z13KG$u`@A-Zno4!0wEYE1#RPgXXM(H#zE{1RvGjBhINyPumtC!Fe5A*7jjAv%yAj% z5LA-HV<<8ujL=OTcFIuvL7EbuX!0uI*e`9>?&nK7vY4uzH=cwt-;L+9-;rIQ^R6dX zKnre8R?U@5DS-h*MY>s1p z)En>crBf*`odG8Y0m$8cq5ZXd7h&@^ko{95(uRzZV;CHNW=1Jrn*lpYT8IAnKDYUV zTEIW>@Sf z5HA}L))I9k*jR(tKj+f$=1WDQ2q_;O@#YZ^Y-N|o=BLIQkM}s4Xl&<6<$VENFQqmC zw|>UyV)4VZ5rT9JR>>Wmoti7(9$$w#Fj)2~s%QHwoGnf0;O`?v&KveG+P5V?`;ji8 z7i-bJ_sFCq5kQcx@y?r)(D6!fX42L0)R6g~-JKRjinD(~=M=z9 z@-Ove#sVKW<8}{9juu!79((@p{EU8o#?K}6LAN33eg66c1bpQQc-IYjPC0Dih^=zd zJ61;cpKHfRc(^VKGt;$u%wFKKSQC!36E~u2ctm2=OJjHqx4?0)w$q{O!nIfGbNnWm zfYFEGInRcNUu0}}vp{VV5$~fn*}Kjc*?Pp3lN@!LpH-e)NH*i`_R2K+>2YP!H}!M6o~+z|tB^9}E93*|B>D1;8)Rx2O(kgRnN>qwd5wqJBigNLX7So-{q& zS=6m{XKgt6eFYJNjYASZ$|ESDvVnL222(0c*r`8HiTV$~Rg}VxlmG|CfJ9BS#I#Tf zcz+-~^Hld@tuDo=jag||Qbs8fp2@j$b#KQIWhrTGZHZnjEI(BrmkO zG{>13D?*#gvB0xKa~01sd0*`_3&C!dX2z-VNA2RK3cDd*zkNr@5BaCHz>OftSy}b$*PEBp?-!S)veWz}P>Em5C;((sQ($Y3q zZDVkPF*$DnMe0J0{*puK*Dz#1tT&k96rU*12jwfKynwq@a5CZ^hTPS05O0TqJlJe^ zsVAc(-%T=qJ3@}_7F)E+(PL1CtYx4Ie*#`RGIBUh)`^0JW-P;+A}m-fBuy2c^SE_X z(S4j-%U+%CSIb_}`gmDAd%@%Tk|mkqZIO{Y@jcnlE(K~_jljkqh*0fy zEGg5kU;CUw6(@V_s(+8z2UiOTzntqfjZ?pPv7wjq5ejt%&>gOG&D`bt&}H#PExDLF z0cwJ%nL%gbcC?yr0oC)$aDm1(--!y4G*CfVBKxROniCo$K=G)6;|WW1uER%0Z_f7e zCEi24S2zZ7h{s#RgbQtrfYOK*BoC)KdW&sMLU$cXy!Ig9_`T6C0vSEScBE(G@D9k@T6Awl?EzyqWaz zXkchC+P$OxJ7xd#H?bnl^5mlx%$)b9@;IL6{5>~~{&K*l^XW(?lvS>+e>D8*gXp?O z%HF+L0MM4OXR=wCMn+4FUV)DcerCXaN_JacG9UIU3Z7Od%6SNH3d*@{mcv}|ylB)< zT~V4=n5uss1XEK`QJp;p-d#`r3zA4(Td8bJL4&9#Q3YaijXcHgCx^Rya61+ss*NDh zI)E%_fV@>||COBin4`?P=jIf|i%|Ye9#TsJG5xq*ZQSIF#dhxo;_|mHnSkiGo08Ha3e%QOB`6z?;SV8}HytzR+-G zOfhErv^xKEr{b^WD0)$W)3VUV+2g@}#CJC(dp(ULTsK?2@1jB!HsQnU!?h_eS0XJQ z`sKx}G$;3r3$?leZt?DTWWD8$r>;*w3O4*@9}nz)L58!(*?-WWtTEPAq_Ppl}PrR2hTP zI*MAPSB-LL0|IaBb{l;xa1}Kh32$ILnWsBVga|}8a_r&OMAj3{aEOt}4pI65V)r&! zizCw}kfW6jQd(n$RcdLv&k;$k+tnoh@NksS)2Yu+l^hBfxU3#@=%{hCYjlcGCWHu9P-!J#Wkeod>i9=pkx< ziPJr_fHe#s%z?{93(_C}(JCk?XbRCHS?mcLz$`WB^#CZEMM0uT^Y4}pl6F}iCz-DF zzsDCCy(o8ZcSN>sQz)?L@UPiR+AxqV5J9N#Ufo>5QPxvNQGs3R;8TTn2E7= z7)Y&MZ@)cBj+tugsQ9bfBB=#hv=0Q9buKU2Bc3()^l0l1eKbO4CwOu4i=Y0)rdM~R zMT!+eTYH|K2cn0JZ}@&W#^e{drs}+XpJl!7GHpB-zf@4#M3CYmZso&Po;Esl|6i=` z7_(Thk3Ii>tFE$Mh6#aCh!Se%%X-x$JO_G~+nYeR!3ffG66QxN+%7N`^eBcV0_+cd zcXE~ffKK!i6e}T_hrq-mqUYa0^z4;ml-8g+L~1^_`BpP{cuEZ=DH<04)+Z^Tdgnx@ z3{GI8`3LROsn=+`2ow{5M*Ilp-}51vgM$@W^P5Zf@fzKIN}h%TbM^2~AfXWj!!Fke zyL>aC?~vCMPY~Xd@qilHquMKReo}Ol-~cL(R&aI+hP5!XU^OWMYVCv#5|L}V!e{l{ z|JK`H01|aDy&3ot>VH1Q*@Qko5woHI)wu{RQVh>lt{AXiS0F&8zLz_0ENIBLyNRd* zZ)N(ELZ&G<-U2)hQF@%fLLMHZA}ud3&mlz8+v@pqU*s0dT(zEWLRQ?=64O%Zxp`nb;LZ3a2NJl`&x>ZpaL z;L%jDG>;ik0O_KFz_HUcS>UKz-U!@jh$AmE)wW)YV{vJL7?-N33gZV7)iU(jJ+66D z>KB=|GjK>q$nz>4#t^F-C4FO9Q0KF|Gza*U^UgLnSp0wx?*{S$B*5U$`aya|9Iwow zx>h(&k3jrzoF*#VP7dH;PWfVr5p)K*nV`%L*(^1zB}E2fT%muo9@zlB&npD7utOwT zTX0#k;DalP2ZET=cBGJ8S#<6aWwf*UEI5E7FRdTu6^f3~4luS+JsWuMPR*~2h+~mA zOq&)tqS*e$XoLy?RgUa_{*{;JELH71U5}|J`tRMWH=dszK&cc77mAC)?lRLl{ZjAf zna})AGhA=T&<9Y)@j8qah4e2V?Z)Z;WM8WLQ~^{^7vze*b$p`n`5x#NO#VM=j3BtI=dGA#jU$`Ge7of zsR0&R+>cgWyOc|hig=-Ob*kQ*tRhg2(aj)Q-HBC)%Au`~Amz;2h0vzX;>Ug(q2&^; zvCAzc5m-Roi~g5;4#kZSb*tdtl>uzzxPmq!;gg|41Pn9%f|^-taLlm=-(un|Bo1V{ z6EQ$z4RC>Du7|@9n(n)v_Jcdn%zzN9gTQBxVv*|`BYtF??lCBz%xm@{5yK()U)h78MFkqS zSe;VH`2gFGZ?IA3+Ng-Mg4u`?4E~Uuh;y18D-s<*1WJN4q_oT&#ykD0=UPb~NQOsokY?>=uHxuLYqiA=A{ga$J2`Do|Eg%OV$VGh8KM+Pj zuiIN#jDV@gbAa-XEU6E&2ed1X0CO|^JEDSkuRytYmJ^8vd2G;U`(-e*X$B|cjEWP; zDn9^H_-v1XllI?(f>Ap@v%}MWv$2p?N436}8#(}7h7|0E9s@F1&){g!K}G)qNj07r zZLTy8HMQqJr)0^-(v3rq8ThEfBk|>}0FvK$x)YW_ke#zxdsQj*)(Rx7&Pg_fP>AOg zyXKmFIQ&#inq5VY>^|U0M6SG?O!C;3dI0Kk`_T=CXBLl9B6RsKzZue!@P=b{kJ!w)9l+oUp;%Y?0Z#`|T z3QRCHNE`RRP(T09?8~&1_#nb1|;pB*cc{7j*6VTqrS zS*u>zZ{|}Gv;L}l`M=pjXhHCd$6qg-T&Ed0N}P%9_%ol3n&EAGlg81NUCOogqM4MG zxb_<3@@x`Y$TO%lJ-`0Mq^Li#LGv+PCS<-See=S7I#t=u6-slEl{Pu-pwPv7pRHMM zLo{jzB$~ET1mx-->k~qDh$@`6l7a=P*Yw$P6$@2Ul%aSshDsx65vm7i@Wi71y9(8> za&0|vhD)Rc6leZs8HU~U=CdKB{vVHb&u9 z4f`+HtFz1$M4oqt45{Iy#U4Hiy8W{4?_uDh9#?B7eC&$_n_ofdY6ewgOP0*5_#}GI zVKs=+iP^$JO3hHE#wWon+RfR(*+4!(qXG55ELtwXblkLP-f=5+Ei#WQ-yJxsjd+uK`1Sxle z?G{LmsVu63XXfYFE8gMGe|VQm>6KOBvV>={77?DBAklVpDmQ^%P({{we=^gjD06Z( zt-u>4oBV7e5C2sv##!CWprzQmIb|SF07yxHjR)kMFi65ICk~nrbQdlW-Jil~0JL%p z8~hL5b=DzQX9hOGE6A*XZy@TciPr3^F2Nuc5OQK1G>Fuyf>k zUNndJy_XeyD0++^edr@f<253z#u?4y%zHrXxO9D&v&loOnb`WXyN|e|cD5!T=Ih^O zxBrK%#hCdtJtIyQ!4oT6^{{&chCLpSvzje7JfkY@rhE@b3PLu9kW#5}Uti8QHXes&?Z=Cc3B*X` z03`!X5J)~nhLW>cjox4;YyQ2n9up7c@VLX?@Q0VhgC49a_tjyAZl4x# zt+QUc)9CBd;;eEV!c1&D;2S_@Xmc<4LXemMIaSNi0k|ry`ZR4y!{=snxfVUnguL~H zpvYO2qVsPHQ$_d!7?ANm2C&m%99n32>Gj|NU1s0^gZijPP^1++-B&E#g1~<0i zp04`ewQBzmKJ41-Ev$t&B9%1vDbe_*mD ze>|_xGhq^wtU8!5+mK@cWAGrYL@*l@TYf=7OuYN|X+z~tc~dsf^g=y>R4118Ti?*D z4*&Hh>i>_N6eGY};@oKUm zqxMkj65(-04OhnAV1U6voJ(iHuLP%Ly7;tZ0v+8%7LKbDCJaorw|(h^_w~tnq|z>L zX>hg34}5I0pwsI0_laDw>e>1#rE=r#k$O~N)M|nEIqQyshlCJqVe(U=SjH&$`{qJ& z`Q%3OTL1Vit|GnY6^t(T>AR33;tX20=%Wo7nr_1DlL;m-LrF)d8Zik4-VL;8_(tAo z2SG9q4~Eeh=YAoRpyG2`)$%&Z9c<$y|#c9|g&0fh~z_^n>7AMb~Q z2D{adys_Y{48+I;lA!w(M9xhxkj*+{twQzK2^n7Umb988>mw%B*wBtAeN$7n(;vwu zg&fy=Czho$_65AL!@^;p)G++sAe$p?A6Wf^Vu?4BEJ=@npFKOsqGF=TqxKZ(y`bWL zkh^j9xoN%o-%OXzu=aM}=JS&mFlu}EWVF1U*I3PvzVZBmZ!ArSLHc&GnbV>+SQog_ zN3}J8ARjvxVs2mkfsI;Y;<13KQiu4z(~J;S1ti7`cJK`0DT4mvDVh?&l>g!Av&(~( z34=Qjvqz15))Go6q7F{q;RodKGH-;pfbFl)BD$lqA`d%xH~pp7T_^!SletO{kcz{t zghsb9wnpT06uqZc2c~tlP3SpUbvp&nz`qI<5@(ozTTOp3rvj}H4YR|AXYaooZ2100 zF1Y3|O9+iQ%Jdx9>*xsEUyw@_jjatecb9xbt*Z>Hg2}P4W+d{hgVeTe=~>LuOq)a~ zf9LMLsUGfdd9Pg}f&t76S6QYh&VJ>6$n#j?ppQta=JGa(Pn8-qxB`tFmM#`{TAy2! zN_m+BrFC`VJk}YPJa785H-aq3wcd$lqE$N|?-&=?0tRhq3xu?Bhttx)Z|PY&@cdDG zuQwhQO9euQo(=Vnki*3eMk1^>2hhH*h`k`dLhDs#efEs8ltfI*RCgX(6F*3vBMxre z-zoNkbn2teZ!;rYW`r9{mL*A-0?pfQ;`{)rn32!U7?1{)*$S=9%Hfe~{4?|8Cq!cm zBH#ag5li;WoZ0*4hyWb%1SrA@H%NSi+yiK=&xjT+-=@A?^34_OjWm}WqzBt*zIz|CP-*xmZGpl4-Y3SxT zg{as_H>KwlVIKcSv`uL&;yh)eL`8a<^vd0o+eX(;Z>|fS$fSuNHfnJZ2aS$?QN7ru zRyoT6QWrUTW?oS9PaNQCg( z*KQFGHT7`uzowrY65e}n;F4iy%YN8wULqdx!ns&?J^vyT933DW42Up zaagO|F{jX#A!W)aeCyRH_Y>!2kWl6whd;;)P%@2 zC2p%)cGcDVl(Hz-R`oUGxf2X&uaJc@<6ux|@ep>Aw!q~T4zG6xXtYsU8AV0skqrop z&!zBxt^@5GO^)1*Z|0{0z19gPJ}di%ukUK=>Bav9!(y|ZrsgSvrjROo_5zZel7Pgl z2!E-SW{&5YQ^A4eU7jU;szZcH>P8zZE^=*-XDvxB2fKG|()u3P(=(+M-faqx*%h7O z*(fMCt%Ku%ojYGM??NQL^JR@{(#3B9pH6$RT)$WS=^=@oqD)@bM z4)kjHZ>T}6*d12oe)C|i!B8>>=o=~@UpeM)=rGE&4MT1s%uyJ^qf}@Dpnlvp@JgH) z`sryXZ}uVFNZw#gg_i^4I~b;en0zJE{r9H#;~gWABmNL{vdiMXfXq5rql%`W!r=e| zk1Z~T`mQ@lvZKvI5DotsC>)MRQ)qAxicg7UcXXgo-{4<3XiMu?HzptXC6 z84F40$T0){AX!vAVtp}J!>|B|11@i4kC`CN%p3Da5o8y)KkkFSWrF6GBn2TOd|$ud zOGCp8WPe#RO~99SQ%i`C&wz;#*6Adfi?8mQh!jM1ldztG=3EwJ=PA`+wW~1!H`x8W zE^U?N;0=B;PS>;TuZ?K0ynJG=Z7kaEpoI|d)@l{w*J4o}kuRTumglQnKkO`x2&o>l z(HJ@we^Oqx?TBg1x%))TbnmbCH%-j=yqYV~>c=2i!VK2VQ>05lkbb(`lGo`+NQ)u{br zbgD>=m-9C4SDI0h#{u55K>UWovgWKwS@naVqk_C?6oj6Oogkh-!slr86My znfuDlA_NX94xGUITDw+o*)uyWB(@}h_=2n%R7h3(U!I@H-B6mffu@HKl$pSieEsD4 zZX}c}bVe#dI?@RQ@ri`Eex1O79N$aD!tEE~?2 zK>7((RcnUXYMG#3U+E5MtCFNMWz_mHk;X*)M5jqx{mrw8EeIK3+W@UZW<=zg*PCHP z6so*;nekanl#?2JR*5~EriCmPNt9u@=BKZ_wx;)S73yu$-YVzqR(el@aXh1LW$Bdq z?wRkydVWz#?-&9vlXn}WyO=V?u#Hud)H@c+jJw}Y{`kw?fbn8@j;PRw_N+j|09XWIq9KL|o@>t!Q>d(CE>Xl+;w%lh)Lfw!iGJ!lnllPXK=(rlVi7=9`LU}pE1MVqIj<3A8;)8 zz|VNI7+mE~C{ukbv?U2NxbsAmKK*rV?jF)!xn`UA76<6Uf;5h}{m~XKa>2^rDKz^I7CB#C4XMS#D0nXBN zm+F5%RTxg(^>MUv!aGM32&ASe+dZ(Fs_ajh%((pg+*~#PjJC}7!P;n^2$(fvrv5-> z28NJ@{9<1;Jftf>!wQoV*J%KYRs9-Pbq3npN%KUh%1=cge-r0K3~54k1;m;|o!Qi= zz&Y;vUu5U6@0;0Ol0KXCavljdAJcML|%jIBLVI&r>C(zhy1yleA)C75E99m;2}o6&doMGd=2(x&{P z#&3lK?#m`FM|9Byoy^lhAt76aQTZP<-;}Exrp4sjPQGw}GN71w*28J?Mf9G7vdzG8 z`=*JS&NKdgjsh6G`2B3arlcbX!9K*FVu91PhUMC)D(B4T58CfXpmD8BDwBvq&I?RJ zBfQZ4*M-Jh&%O0(0PW!wq=G@bG_ITh<$OIlK1pFX7238|n4*j{#3KNZVP&2T4cgU^ zC7celGBA6JtmUSohN?4=#Gumg((&>kK03{rjBbz()TmaGk%M~4qCpXa>)PhV_xm*z9KYrfSdb0s^!73L=!^G+`w zK3{khaD_?{4=FdKqMezv86q9L-JJOxck)w9oA2wS282a^f=kfbAEmzu?|=XU`E`AL z3-}z0K50-+{~PZ)a)T-$r_inX{!oV1=4;bU zQH{SU10t!vC!?@7WIz#eLh)M~U@l=GC^x*NGg5Ts5)yT~a@120eDH)0fM8)}%>k@W z!Dh6T@tmur&TE^;`Ab_M)*TCU_N#K&2l_$mQe>k9xjPgqY{v)@o6qLGwT3R5dcj!P z9Do{C5F7akYMBL0rQjnAA>%la0IZs5v1ry>iOjV=y4_(YXWVF@$(~G3p}Th|RuE_z z)R(bvkftmzHLM;QcPxX&P`{qebJLpo&g*GXbv>*8ss=j^oJ@GnWn!crE9!M+8EJ|x zxww2{&2`|9R-V1}$Cl)k?X(Z)(7LGT+~1W!RSQk$3j;42T#Zz^AA`FXw-9<$<=(7V zR4W+2uLZ|T?1=0Sc*~2Xc2bJgksWM!D>*s&g;l-Uj(4q3QHpKr^OeV730CeT%kwR^ zhrm3KO%ioqs(?-O!b2#2p_GdGuIK|ALBywDXXlvY|Ft%D z7WtB+8089gY1uYI>zD$ikqJ_e`ytLCyh^u>jlGLOeEh7`8#+czV$h8sAZmuT^ukKRDorFw@oCJY#Q@U-6tG*FR;pVs~JQ3YP z=$DUMX&w6`Ht2HywkB@y{4?{mCpj zHxK>2s=)^s@gUb1_}3B^pd&n(ZSZ+KTn3$SwQYGG23_@@`VPb}m`1jmcg5MUHk@I2 z)bJF?#zAp)GTnZ-F=4L$_#Yh-*tz@$705qfJ$%dN+2~j@kYc2E=luS@@STNxHMvfP zQHg$6;k$SD7(}pqf3LEEJ1sLO2M-SaviMJPnsgCux!8B{WB9Udu!sIKGm%nO2IFMJ zU7Pu({8Z2zC^aS~N2{0A3?09zO9I7*{Cd>oyTZ-h-xVH7#KUT`us^mwwuukeRp*`? zH7vbg&p#k#<3ALQ>vMN`%n(^scY&&SKL73u07}@JXwR$hn-#B%`QJCN1c7BAWY|I{ zD+uzDeTVEWhJ|$TxzzyfhKCqDQxID$vm_Xh+%xL@hBhS`zT-~MDhQ+|6ug`y4*&Zm zymoBY;Ln5>NL%Hv)YjEW+!MT6xv|%+4rW^Sv1K$6wjlenX$SQW+-^Mj4xiFbGvZ{$ z!gKd}xyD@A^B+3g>$r?H!cZjfg# z-QtQ|wW6^~+R|nzobhNX3ehwtU;eB0qLyJ=SY4!W{sZX38YB-VS0B$Sehm*ZjN-K1 z@_bg8$+5v1cTO8~P8uIC3S$x8nzU@DM#*I^!e2$z!kS^fvpRCd=l8E>VC}vt`QO7} zG9^%cX5CB9LH5r6>(E+7s#l{w7LmJL^gm`b&TXV=bwar{)^MyXRn@~8O9L>6(w6Au z#uI#N?eV1=x7Fwhd-%eNrhc71c?zML<^0K>t%iRWEWIZ6kLJNNjqEnO_7gno&K&u; zWosfqPx_``YO$&*Hb8A%Ayr^Q10_{3eApd1iN#OfWYsVpu&Fb!o)I`#3gZu4 zzL!gG|6pjt-UGxZE16ol*UElrG4Wcw#6nbVYESMqsUw}de7EQH$5J%{-L zx*uSNEU7qD1&Ex3lk?Jhc=^jJ)-`#T9>Eb89iLs` zLW6>>ntD)>3(mmQ8_HL3*@J%(78uUsAWtFaWgkvn+|?(cgt`krixbz7ON4+&!p?Uy z<8kxACrZrv^*OScHbc+9adbv~qRY0T&1?5X)DNqvSiq!l>ffL4ImrS#BWF0y5x4Fk-y{SF+}v zJI}#mZACY*0e>|}PWL&nl4>@+a{p+k4Do1B2*YhwHXdDD-$k2R<+Y}Nia|<_Ziib5 zSpjnEJRWd!eM~JM5IYOtdF_@mh*?`e6hvgZkH7r;2$*4GE7V2=Iw^%ype{+t!Zdnp zOgw?dLd^C7aD;6dJ-Qf!MQlg7bZ-fa?GuS1U?nGf)*r=jLGVF>`wJ>wY!I=<&<4Ww zl|Ba6T;kljO_<*L1PABFN;3{puuM_hoKw3dvR(^&G0+VM*QgcyQf6JMipk0 z-&4hr6WZh+N@#lc%pt$|V&Q8nVPd9?92~oJhl_K5XDTCg3<7-mK5y36vHP*lvZ!qT zmEm9}6>r{ouX^f{82jTJ_C%AM?r{L<3FH@cmz)h)uX2kC+p`ovs(CLDVfDt_(DI=J1~uR}UQj(jR)09()ep+8cwB+hXaJW47_B># znnT$6%dn)dmGGYculTx+7Z6}sZ>FW{?*9QFo`Sl%`tTOa8Tblr9eb_3I>RVG+zH3~ zOJ`b?kL-~8l&ZuWfVLeOlv}9}?;(Bw&^xWC^}r~krFYi&y)7l`{;j4p;3EVRdvxqv zEi^uSj%;_p(xgm7u93bDattg^YvK^HNXJwnKlf$tlI@5NRIb<p8Ch8P!ToeU%?E#l!a39Gs98q*ZJmT(au|k11 z7YTzHnhBGFSStNg-w*(5u+y4G{}vjyQe)x$dPlNs7@ z$Obx5WQwIbSJHySME|Wp;)6OZ$+hVR&RQJ{j`AFaS9%>pRf4#$-mvc=zow)sjww#0 zK@qJVYfg`pZNCy&(^2v)a(e-Bx9a}JLYwi-0g!l^p&P-*Mg)FGY~bcTWe+4=nl z?IS`a*O&%j$7vBd;I|m=u+jdD2L-N|h^-xDI47X)^xyYU-<|^&g~%JNN(As#x*Z?+ zL=BlE4OhVvnD%H5WA|uTi5c5?yqu#|P3JzuJ%5mViCvoHl}iKG4QbIiIgn45p@&!*;1+pAp+fN1t4Qxk_VX5~A#^5pa^3kr1b>(s zF*SeHqc8)oPF+79b$7to`c5VL1fLCtOgO+U!2m;7&!HYhgP4S~j7~rltNzB}|J7}c z6~hupg0c+JA6yZ1qbKa(yl5-WYa5!UiSoQemR6ngErmrzA7+(f!tzOSUQONz)E1AC zyIpwK%s}R?e%q9bC zAF$+S5sh?U7tRV+6v|U9jVb;1ZMW&&yZyGzNJqddVyG1O9^q{~crW32?x1fvJq1@c zwIrB-eQmAKX&J=*XkQ^+k#d+=8tT(QDgQ!Qp5s5MNoy@75`RitI>hV(R2h7O{~cls zyqH9aF9OEUFjEOTrglm`n1u-)&@U;#`h*J?*4Md{Qrr(kEdL!0;3DQS^x>ajk!wp& z9H(}J$>b$KcY=*Be!C<{gq*qHXnp z`uYlrr!u3{hV0pno{@4JtF~w6wH4_n*O|AZU;rZCX>l*(my7GwnEu;U?a}(9V9jxi zRiO%Luom`Mp`7j@RJ))ZZ=C_<2I!^o@HZ?cvu{ zR!18NUj359uw39adC~s-qV?!Ur2NBVMc@6Ru^kMy;bSa?X>_Lp8u|=MY`@U;?;~Ku zN%K4QLi;WtH0C?Jx3@9*6egFHXnb(PK@vI$Dk?I@TYO zi87Mbv%cksvPn=EafO`4M1zmGUNNE?kZMMtB*oX6THzx2Z58+O+>JXCnEsAqb@peD z{fB_U1bf|OU)avfVN;xufsM|PzkmGx5~HKy(t9`nXc3%SWGPlV_E(-Lm}(n;j}<#X zzLoXR#}9f)G+_V^l`J%o$@UiT_M^y5Oh+(i2H^ONHsnGdH)8JDNFGpiSrU*!zm2;w z;txXy%AMF|Efu&F$-z_P1?rZo|I6*x|ij_3^f&ljL6$;0U)34E{3=jdZz z@s?l{IGv=@5k>jXb^L`D?q`uAsLVm%;?UV@oV2cfBAZeTmr@ zPV}5lb785J`MmK#k^pbCi}vG$3$@6t>YYUMn(1c@o*!>1^*tur{X%NO)0zZDJv$n_ z03B1xA*7YaiVI#w5P7&7Ac4jDT!+0RCW#l{u?Wzc551s z@pzux2YNi=k22$%&;C*v;8;lRw~$y>&Io+2LYEm!OMMO*#51I~%ls=J?_MOm!G{?S zsJ-WD%E({UDDKsJG@+&6K32!YEPsCQuY?DNUF$|}{a%Kwl~^G(TNGE}er<<&c8TMR zSRXj)w^s%`LEA3@7ZO77f>$0TO(3Lr<}+M?ZO}s^xH7xp9!eEK)~ZF97AAj}i{M-| za_tv{`m__h?&y($j~}7=O5rD?Bo#c0vQYDZv+$pMUCs)}85#gYHUp!E$MdJz;o*vu zA~ZH}hbHROfI^_We+2tbg5Wr@*jXE!D7$wStWlaE?16kc50712X(#=2C;&6zaZKCe zjDs`bB0@0%w!S(G(+My%vGoE5(q*`BPp~lnEbN0{}Nr z+{+)|`XOtuTY+rOpAQ+u%ChP~YSi`nD(&stnKau~-I_saKfZs5Fw^BR!n~%R5|Gu^ zu|2G5v7p(KtI&I0Ny_tQV&K>_OwlY0JTxOaKG@j`Mq@$Zc_PY6(oOCPyt!IIC%p@_ zXlzIJ`6qJBEknaqxj~R;BSn1!FzQD&CUTx0Ux|5CKi+3 zhSXK45BTSKcd@L*V8oOg4ZB50ZhC_MpX@htU=tw6in5N3kl)Nj8%(CVxnW@sVsyrg)~s}F~6FIJmr zCh_dYaBW9QkP=+P1zooTc6wymIckh?${$_@Q(CcsaJ+HEwN>RHwFnOvR(-gtTIdG~ zi;F8ctoc%vbw2mC_Q-07G?G(hgSl`4t#8Rvj=^HqtF8!U__ooRlb3H{0@%TQ$iNk4 z`N4fe1X@Czn}j!XOzbvN2p^YGi11ec*Og|cXaOoh3X%L&LQ4Ah*w`Cr0?(a0r`J{X z=1uc&m^r2G&Ph-&g+JEH+f;xpEMj1E+aINUPskEzV}0xZKPy+@8cmi!EO2pVZo3#W zjx<*c%O{Z}N6W1B(+hWY-(=!^Gd((pb5CLm_esdxQihE;DK+$zfybJ>+gv!y5A7e1 zm78_k!U*f!BralzLP)vj7q3n)gYvHisUqY+hmg0TtcWXgFv^_@yG~Z1Wfa=V{)Gjk30HJWj+{(VPOBu^qgq9!f54GgZRzYj; z3XyzSxDkfn1Z;vA-M_!Bo3i5#X#0?WrD_s*+a{&0rlKHsk2-b&GsmYIs9V3&csfuT}t zSh$x|VMa#FN?I1WVBsH%`~;8}e;#no8oN#(K;zXO{et7KVRm1&QP!)amt{}aq_%T} ztb|F{X_~t?q5zda4GBua;hNvTlZTBairr!^$mrE`8$Qfa$gv6nkkTL5@SABHrouat zG3HbRq25@SM#ubfe=Pc2iqW!dS4SAnGd~#H^y@`*ElpJ>yi_qz2$p6%OnPS2J=!Q>SYUz=y zc7DpzZf8z@;A9n(|IxaRxECW15zL#ZFmkT}>0@(Qby4R8q6Fc#>pRTi?G!B)&w(Hx z&gF6=vj4L^JC)pW@5@gTNjWSH7qm$c(CvOy&E?1>y1e!WI;DK}nxsFk6$~HtYkq$U zk$BNc-R;oSTh#2mYC`Lwx|c1b!zs#J8XOBWZ_`EvNawgq1px$<8gmZ+rI7^FWeN+C zm_7D95(4TYc7dRMane&wzz%*27w-Zp)sYboB?e(1Oa~H+(*a{Yn3h7EF6a94mItp~ z29HRI^0oD4YoSg{(YgfA`Ro6v#frK*<eq@DI-|N$2$NO?&Rgq&01EDe31D_LKjQ~(*-yDI<`0K_@8KPZcMEGUK2 zh;VBRxqfm9q6Qs=nuF2D;Al7;MKy)(SkDi~MM)$_+97)rpwx$8bOT4SL30;q z;t17!n6mbGmm!kPV5p>a;jyv^p=_) zC7&)%IDZuZ%Uu>eYKcrE(o%muIO6gc==igKJpKECZ}agHtYMikpkKJP=`8+zp;-f0 zJN^p(6SePWwvqw^v9=nZO4DdbLZmtHFA|xPlaf-hp8R$&hHnAs<}7lYK&XY_l=2@% zu8rAxT2$qt^dn*^f&3QIrV^hdqgI}-rNYVH8AGRqVSmzRFP8`+VTml1^UIg!4_?>R z(M4gu#D5_4lD~$&Uj7}3gQxbMs2NAbmBm8Bg5voleo@*?(fkjs%zR5qde$Vq=Ty^{ zf$~FHcLnczMJV6A@9kBcfJ(*J)4Z*b$xa73ZZ-orZyaCzrjA%H&}w)du`YYga!T}^ zHJ~bnlRrv?%p?l71AAJUc`}|* zz;BveA)74G&BQIx(Df1`_0%nWCN?%0EQXe88nlXzBKwUQQTthfnp$e0S?lSX8G&(w zsPeg$@8R6bcx{?0)#NjD0TRoev~eMCMSZ_xQhadioz9>7YXAAhqJtNt1y*a&K<6xGlERv{dl_H38`Th4}l*0jVL7*dSk{ zRLOaW=T;u}rp0D76l)^Kq}RoM(r>!Jve=L+UcWs|>%7OG_`dux9197&laY11)5VvV zDrYru$MURrL002umls}|Q--P$NI#8lO{(`CxMjf1%n(+~4*pkyR>TdYz(*4i6WDmj z26H}fV3nOlcn9d!aj`zStyl&u>k9$gE<+y@{HTYZ4Jq=+u#r;MZ6sEv+!{o!dnUC6*ihYmE zz06)8@nzZp9M4M$WiTCHX21KIC*N6QB=w^Nq)}j+`A>$z*=!%eU3PJv^RY; zvcNiuP&Xh*U?Int%G5pl5!#b@7f)grbFkZ`9@v3p?BMY#b5u}cMvdN?qDaj*il zlZv&lCuKTaI|6kE?1Q+E&(}BU94+@6YZCNTsWO_>$Bb7gU9+>tB-0ypAK_t6os(E$ zP|SK^ke(11cQ6rS#sLO0kih=|^{TKX2@bsy4(ohEYax@%KlB0R@$YdFg|rh1DEHBV zk!O&m#>RquGJzyKM5kdT)9>jD;Pjb82w3Fz*G#lUiHKfe;&+-?z?Hzy-RHb=3Jd!n zZ=MG>z?{84Om9->)vML1HF~f%Nf(FjlI~@-=EA6MkJ~XW%jY=QDETBW8j}V~E@}L3 zzLZ-2;`6V-khkpjT#Z_cg>VB}FbOG!!6ThsWBiT*?R|BopAT>6R|FmE+Dlo9+}5Tb z>vzi5VIcY>$up*V>8!fYd>@)~nELEcBE|J)OOiY;+r$q)*CG->Hl{5)`05r5I>tEW zyjCt9!j{EW=;GdyBp0nmoe(Ch;eba2352`{!17s8Ma16Fz)H=31i~5BdInnv63Pi z6`5IB0sJ+_oGmf7*Orid%!h=z;;GxKgKQ#V@c-M+nM zadYeJ$;-7!iH`G?KAVpI*NaFtuuX*OT^2UD1(N(DaE&p<=K6UoSX}lT-0`X*OafyR zsP8CIMAh<36Fg(gfzhiuc_IYo{IOny1bpZlmrVF}za}|v5OY((HI)ASH%I#=OL>=R z?Fa#%R%QcXZ7LmB*t#?4q~EkR=-nzudiE|ezt|6!(@9w-_V(2TQmhbA69MXzm|3@Ts64Kso->0eJ3Rc?}ogGAKsW-ZmbNhHEAx-rY zCZ!9YnYAGF5d0wnshG94Rw5@9tlozqM<}e|x--T@8!QnT&_dre4V*QERfqWOJWP_> zU?%bI$2DN`fHIRKxC{4nJ?u`wg2gB`gykk+V#F%pxUGVDH+ZhqKb8y*@{=14Eb11JG1Ytg%-l)Eg;3xz z!!{7i5IjG6YXvkVe?Z02!%<_7Kf(sOnLkAc)$kOtCS%vk^rM;!#xc=%W5YVqi^{ba zyxNqh^}#q6Fg}{N;0d34#vy5>1>U!*kV;BEuQt89jWK;e;C$3tPGBd|6l+JHCfk%{ zjim*Aya(7G0M&0W->30P>GWVrx2rN{c%mvor@BE z3|7o!J}a|e0uM)1)T++`XCdpKoAuC(Nm+&>R316U(bW#WL&N)N@W$tX!52+6&|Xc@ z7uaw_FqObXDzwWXJ&nH^-WtoI9%(|S&v(xYqW=!7HAzVe4%v5?m9I`JQ;(-E@p@&3 zxm??~;G)Ti>chaU37wIbzTfM*iIZS^T;@o(MRw;CH4dxYypA!QE;ml|9+yT4>hSm4 zE$=kl3aLCkisDGE22Ba3DOp6;a2i|ysKFQaYw%fNcVzN0X$|lY!O|p1tdXJMvhElk zT+(dxyCq%lorECz7H|zkAhSNNUm41G;e2o@`UOFsh1-t_Je(uPh14w)0H)x1^y_}Z z0xttxY^NCL;_+AOe#%S^?c8sd&VxD_S1w9BKE|3(-O!PCly6h}!EP0y(2Wc=0k3jq zrA~6P!atDHQsK(AddRfPy41BtR9KL?rlHl`l!;~U1F z)3P!P?zarH_6xn>RW~UOzAPw0T%1=Jg*kVcX*a#VX4LqzUi@6&rh%5$7uTCzT+p7v zU*hX<_H$j)ZTJ?Y!9;)fo#v&7XN{aZjz9WOvXDLtk{XOzo zd-WLZ&z65VZyT(f%rLj20e>`K5p2~)JaN&DH~jEvwQMc*B9VF>i3chN<<&$|h{Oc; z00snXf=){O+u+W|Bsu5?EdV2nfZDF+wl&bSfKptjlBgcUvj_Ah_~&#yDnf%lZ4J>N zZ4De9Sao?UG|w}59`o2rP=y7?b5rVHYeM3L<`Y@l1B`sey4}{~i?Zfy&aAxU7D3=$ zCYP|ChHlQa&iFqIKRlI22Q9qP6c_mcS8^J9pEtpo(DKypJ~ZZIJC z$lOfPH#Ln!L=r~tYFH=^+X@?Ntq#DN>J0M>bo%#vT>&h9Bq-}=z-S_cGP=_+4>GLh z6rekA0ZnEiP?NySKVyks4q}@KWF~o?zsS=8WuR&?TD1@@t2XXfXZB0Hn2RkijfqSg z7t&oEiN)m;#EL4UFleiLXMKfkwlq34Qv14fbhMEC?7-<{M#&~!B21& zrpB^)rzDh!o%WCar9P*4g;**yWH{_c=F{r(Uq^z0VLDtl(qe3esuiMy`?eJ+-LL5a zz89?68J}Z@{UO|Q@N6#Z1r zn25j@HP9m?UikWMOq443qN-b!s~YD4?9YS&?&JRq)-7eJ$V2pk1WRSDe%d7CDKP+k z=Ab{)99^dnnGk(7Fk8V2=cUl%cC?ry?CeQlxq1g~*le`um z;e%#ngW<=+$p&HuM;WrOt%hB)$L{qLCW+Gj@E*g^Em!P?{{Ywu@DHUxwr5)EgW4Om zbV|VIQU&8=1Y`$&47&9u9wN>c1Z^Bo2~M^0Fl4tOS{dS*Le5hbBqtOeGJ}2}rdFz2 zBK$F!R$Vj}`+uXI9YnN2p=Ac2M^WpuAX|{$M4)feentZ>|C4+pSPRf38I4@bTk+71 z2I)qlo?W1~oKA(g4)wdoFldC?awTQUA;_K2QB{Ia$vq@FV6;DIKfH6pfT7%J@h3;w z6sC9K;-{GkYf!Uz!eh$^Xf8!uyBVDI`5bc(@8htvO!R30;j)lip$H6yV-*YCG9XM@ z9HkbmE6HQ_5GsX_?r@OiQHfa2hwe63y4-qn0Ljx|6drgtU*Uu zd4TG&X0TrXCF*LSNxc}`dYl&e2kQ2etihthO<+xCfER<{1lW$|CAqI2(pU|oqTnPk zP#iS&laSI3iccqCfaex}2a)Wf4;QWlSz&U0>M~%^OI!T}3NVkqQlPa~tfXYO=sEaSAL3KGP|a?J+8Hza;vmk#Y)x9Q*g$4SQhr!g75x>+s?#vaw8^EYpwp< zz=WY$0Q&1AmMm(*JOLfD34iO;O1_GT)WZS=HXsLdR>8ph6c%bic;LaNbLnFI3VSPH z(U_&=EIi-;W;*NXG=Q2@gUMti~s`~O+NECLP<@=SboCg(cW+1Wq$noQLYC@ras$2O0{`2mDZAVq`$U zDKZ~O9?kJ~_P;>GSU;8g{XY5j?c0*;Fn)C(#aoJi_8zOn+{EiDMRjVI+IZ0WGAAl0 z_je;7ZB{ex+DX8(t8Pp1<`k3f2*)iPFdQUaNKN^$n&h*H_#PTjvLbo7&@w6YEqM30 zI3ec0_B)~<|`6A|VJI{$vYB>8Wc&!5i0+&Dv1rng8UYSy%Ux`R);#`0VKF`?#&4z#C%wg4axn zM4MR&18ZSNvzEa7F$gC7a9s>y{w$>XkeisnclCp%U6Mw({lG;JBokVc(TA(n$zveuM^@kLJe_T*Y za*lqITysF+dMR`%3gCg+giUF5XcT!J5@&~djepyzty2+5i;03zA@|J-bww^Di|UQ_ z!?Xf{Fu@9rCRonUC(lDk(rJ3UbrDv6<6sr?b)xVDi=(M;nvy1@$W=Fh-nHw2zztFe#Bb_2#BU>w2c`s4mp z%Z$3Z>E8%o1jW}WGJM6u?*vO%IM=x;bPWZ|W4XwZ7%y$s);g*_#OkCvtUs3<Wdlv_pX<-gC59u&r|2Iqh4>osoT)T6$6q z5Wsjqw9Z;WGA_`e%N&dULE@ltZy-q(YQyOdwP3EUGi}JEzLhS&?10%#er^vC4J-*m6zu@ z1lXX){%!mxfDUil*{J9z09-=Q5LoKqx05g~z6iH;hBR~lEim6lm_ig5LBA7n|09tT zauUsOKo_P9=89C_v^fUh<_lC{?fiohPtm&w4+U0)#Fi4;vIy#rUB?l){Pc!_=*S07 z|DmBDg8!1gC?Txl)3TMsZ+*dsp&X*o$o)6G1%)Guheu4I3LyqNN&45pu2>X=*&&oK zDaFpX()V1O6sBfpK+2OBFwe=51(<${+Oy{j_uM?lAJ?T+4~Ek4=c>gFg`JhDky#Yr zRLQz=Yi^V@)gd;7{r;C=qG^4yB;WzRcDP8y9wSWjX(CLI(#;@Yxb9h|S%}NWGcLWU zWp$v=uB6@4W}cJ32~;KsE2ouw$4%M`qJQ+2aj={ij+v#JZC5nN#nyJ zg%#KrJQB*Po#=`W*OuC?Twz7IS5NrYU6M;4PcFIlT?JMiC zSoIOyh5kYP&V@8!7y&Z9u^HNgf56U3^~^kozh5QkL>+`r4z*Y9VU#FocMNFIL0kS; z*{TUSM5EzUye(rtocnme7dQ2Scyxq3aI^xk zEb-#5WiqOTDz*8WD{;2v+1#!(GD}O(y48fg?k7#&snLLc>%9httn#0!*DpF>|ALuF zRX}a-qO-(-;qeQeyfPis<6VBIspZ8bkVQ~O&IN`Pn=c!YYL;Tfv66;IE8)Bsk;LEl||{cjKt#X)w) z7_dP%&;i)rW9*?tc<$-Ef&%TS>v^y&6&nNLEc5Pf6n3}mRSNJFwBt{&y2kPe>AJZJ z;}p)Nfik1A8#C(<+^?fo(XrNjQ7$nHTp=UAWTcu#B)dc1KKLD@Lo%L@>e)8& zr2%$9-j2_&Aw^6AF%{42?(0IiaAKg#5T*?+ThMh+=++)^jcvoP{{T4gHMW(0e~&Q< zaDR1*fWg!F{fxoF^Ghf_T_?*rm<5}PFXWZ9@LCRPTG6#ol+xmmd@!?al68$0r=M=T zxNA{t|EyM!A?{)B5QhO}{NX=sH_Jxj6>;CER7d9f_Fcw(_{M)&<1)>CRO9HHxE=%J_h9eheFZp?`;j?}&W`Sd zjdJ0x(XlEq&4m`Dy=OUSEyg1)7~l((|XhKS}OFE?3I=`IeP-U(JPVu*R((M1T1DojP(;}H2=zSRJG&BUN&oyF8!$|V=dW|mS7?>i(#vOz^pucUe z6KRJ%C@KDa!ZsjH&ggZ$7+|MxfF0TuY_zfrdbrwI{htW72{5n8FA!dkqv={xYF~dt zRf0uFd4>~jy0775<=v!dF1T;0A{Q`wF8ph$#X+Y8wfgg;ChyzBPG150@)&`JzfnHv zL{|`!>7Ku8I)$^ly+purXwhO&{tw+z=!^a9kO9)sy@9gPp{YVm_unrACDG z9gh_=&Yg{mH~U^2X*<9?m+jg*q#Q0ZDm?mMXl7k(d1-di{oC(sXplaMVrvx|!Ul}| zjo{0WR(W#bpX@vI7D+t{XK@vCDIm*nZYZb=(MNNx~II-~J` zOPfR-4IMsoO+RZn;mwtg@BufjaxWszUw-t39(6S#)Rrog_V#pX z_H@zqhx~a04L_sSz9`f@C+|0|;3=>cx(DAnCZQ7U`0mbosH-!4SNYEYkZd-Mgc~M? zjD98@-uvF!qLE*(>9n8NK3ws2?TUK%8-u0m1e2CdY!}>5GrV3kRL6fc$|m0wOOm&e z1qF~5+2nKj_K)U{+|7hxTBIb>3#FIpJWrv@aF}5By13>)e-wi{{`7bcM5tsuuv&~= z%6fzkDF< z^rFavA)2u+29TlKX{!n^+$aUK)w$EY^E!gaDbMsWw()e-btP-Os%e;_D_2jas?<;H zpZtC>TxanHX{g!ixc&uE0j(B&*Jh9<979=zNo1v1R@;nf6Wm>A_e50+hFPC>w0EvM zl^x{JoU!11L_S9Q@eNzB#KD_Q4RI&!aOYNl-l;9Sku7eRUaVSSm0EPkShW9WuEBY! z;>uMHobQVDC6Y#H@l$f$dCMUFplhi%cN*ordt8~UGoD8N`)KTCF(Y5YXW0JF=3LanCic536Q`#1vlUC8A_E_6TyU# zkS^l&@#0R1a?nW~qlMHduHEtGR1+{;6nZfGVyr;l0(|+fI>b;m=)V#6mQh))QM)jy zluCD_NH+)qk_sZ-9STS{2!en}ql6&cEgjM+NSA_?bgLjKAOiBu#eUCs&NzS082jJe zZlC8_>z?`v@ejM*FNGiPO(IrX1iIg=K0V32l z?$;4cjQ^>r1j8CoB_G+f7(YfX=u^n~KxMrJZr+4wF_u`P|IBVk3;hG4z!??B4`uzK z{t$;e+X7vc`14CcyMg|GVBQpP+^B`QM-pwYlqAwHjVOg|Y$yRj1K5f!QTStVZ6{Rf z)2e2DDAt-v9|{t*XuTU3J<+yLzce>9fy62AEfgkTJ$SqO|y8Ln$sKt|n5X?m;^EpB%bX(U!~uSuT_c0n*T5wcOm=Rk8hx= zBRxu%!80ZPjA)9U$8igv0Mp>0oUSD7LBnlVf$h|iS2O7EC)_Ls1`V7il>42Smr(}; z-9dy;!IMisC6jFx;R|@U+P`W#rep9jK%`m58&45%TnFc5wGD`-Ak=)QTkCaTdCUF+ zl%a54*iI@xJUz(=g#>GY_X%Qq#vh#|0h?p`%c$XWB2|hLKJ>&Pts-Ky&o$p~2WKrl zuBYgAn+(^g(NJK16Y=I7C&L_Nv=IvD)sV0+_EB19eRO<<&Zf3AxKW&as8G#R*`W0V| zD0OgZ12~$JAK2DG#N&+c!@+3h;Lz(baa;hZPCRk%?Em2ol>OrPF(NVM(%JI?eTt|> z&>(QWn;e=dyEWnw<7Qw7pk$9_!;s2TB6#P%7^tw z_qOH2JrAGqUDb&PUH_+!x3q7vrsCFKUh7tSY2!O9DELIgF^Q1iMD7KaoFMI(s>6Nu zgQVseuGBYMnaX*}C&M^`(ylXvgVxzv1j+too7VY)b*0l0PRw^|nr~TPNV;*XdOKy2 zPt}bt(#FF4!|)T-Wq_uCf9>l5s$6Um_8@%WfJkO>4baz;0J*PXf?u?NFz^)u zq;f6P6?}lgQTA;PD8cG$N+?qhv}Rb7vZF4(;E=JI|7~~q#KDlEy0%v6c3x8-1;4I6 z1h9UI=KFgr1WxgtR+`v1m9Zp=&>HhW0m05Ook%PCR+~3CFq=Bh-S^WAl`s3-+;q6t zuG(i*Q=)Ymm!qM)jRKuq|}y`A#lJT z7v~QBCXQCPD=LiD`uWGi16C#PT1q)i|`R*-H4^g5v|$YaWG zw)A^eNK&TeCgTe#N!ElrWywjU|JmC-j`?}K(CE$%a|jh+4u4N-FHL3+&1rHU=;=`a ztLkSku^~kW55YAv8%r#WM>N1MEmmW@X5QGGO!w<8MgVDY&KVhZOj6gA$5FdZ!rePE z&V#WxlBs8@5mgB_x>X$eQISnUgHhbc)cO}D~%+joHC)*zS0s^^2!$t zWt;bo1FJs%=!cJuLj_$+ZlE~%dp_bM2P5WFV(N8Ftb?1(Gtc~mJRq1P->}uZkFduc ztsU6SDdT^E+nKSmzwd%aE&N%l>BrQ6j#mINqmgjBq9g&;#^?O=Z!0YRlugq{QhO#s zwJ~>;vhtu51~u)CTWf>wM_=2iZ(Wy8)vL;PIq@nMn~_cS;{yMcwh+zw zfM?aApfsS`vxzHu&=SBpKocnHswEN|pwj=nCRtM>f{j$`pgHsh-McERUL2N43+ zn1N?5Nt@q5mj(oba2RcuLC7sFO#J|k6^MIZ!E$*9r#p$8aQdiHCF35lT0s9p?a?J@fBVy=HkIZGDAxt|n)*av=si1+cf?!Na~f z+1NXEB`MdMStJ{Xw^UJz?Eocm7>m=&qYLTPN+em){;%t;~XyzRrwp+O|up8Y{)M8K+71w^F-u2;_QAnn7X22yG z|4TbH_il{MdUvc6ir!7%XNH2L-Ei{JJn$KWHdmVcD(|6>LqmXDcYS27LOZ0ulu!EotWeANQF*n--UyQ7xqWltFaqQB3eGU zY>|lDZm7TN{{{hc^9RJcW4 zS)7ZC0qPr?EANMR#^}{)g%01{L=5I*k5W_P2X8rtw8Ohju$_uv?)-gB1 zU--8oCVuv=&z()=Z>;&hZkZ02Y3(hPuV5z#Rn8^ZnkOJQxi#h8?kil^!<*C?(P<3s#_Hsp$WM&E8`GvGi(c zfYgpa+Ybjo925$q2Fr2nth-=NuddyZ3ha-5dRq-dACJV+Sf3yiHGs$NZa>ig;|mhL zA%8y1SUm);a&i7-j4NQ3{LD;|J{g@zm0z^p{pE~>6Q_QS2AcX_ z+*=e@d9~zd@4Fgz!b+%gYWw}6Jz$elZtXf?99sdt;ie~@EV>m*yELJ znZd8>XDidjA9a899vjO_`c>Keor2K(`9!XqB}FEP)nkkv#riWt(c65r2M4CiJCI8O zz_I~y=$Y2B=YdZZ031X~WZYyZzPku}l}O+}Vfr87Qdxd?ovyM#jD+@}5afV|fV#qe zl!?twpJxSegfa3Mpq~ttWz@EE1ruY0hw-G?fNnXetZEZPcSo@@9l`r){QPRvi$RBPYT$mSGXFWFW zckVIx3u6E^ZJnx!_xX3XiZj!u+*1uT2Z7#AXFxpmoetg z3-S?f`RTU^{Rl22sVgNG(LVA@e~1}#e~+iUxD~j_0CKV z4Qr%6u&)P*xQ6X|G8ry%<4b>e=8j_qX6mP8VMZ+Qntnto08sf!aY;EU{-%Vv4vSm` zGaIMTeI)HGoV~1ETvLCyz(Ris%uaMhN8A9R^V9WX$CmAAjw2^d3g*dq$AjrcSD-m|_`UHR#`7mj z%5^$5hq^)NebLT)dw>bX@94ceK$%4Ee+8vH!UCosx!+nxT|-+rP58Q1-NLv`6@0lXs93zxT)&e)i}Sp+$@DmKKY`-Noz^$5_8DD~*Zn>t z(By4R&TjF)&3k;2{(_WyKP2u7fT{?adGR^y3lYy`#Cc+Z@4#~(S-T?aFL-eH;0$sH zu_XhKu#STCqRva~27cB>VC{Zr6?&F<$&tvByHV_7Sv%OihQxY+yEe)fg@hxX{Dv}c z+g?GOClJRsXyxTE6-9{K0OB>p7=r~5wS;vTbrEQr-Jt${4c3t3$rH^YATR3#u2B*( zA%;K5;}9JJ`Txr^#&BXJN&&G=2d3cGpS>ipS}@9kL^jk#h|_|(3U?O*5N~oMs4v6Q z??m1W2t`CT?*P3*Sjjb{=7URJt{q5k;e=5GiMhXsP8kxZO`y-mh_>Xvbt4IN&!5Z4zo4|V94;I`OBzrRO&n0^|`CYK-AlgSKOJ82|-}5>wpJx8S4Vl{w zgkWSDd6TmZY%`o7eh)|xzzEktDui`$$xH&IUtK{GA1=-00k4>TCMM}={s82{GBIGy zgs@(+U*=D_ks$uHz(I>E*+8h2AWjKHZp_Z5T%L6#788iIl#ugiGm8BWG2H_M&{C3Jy<^%5XL;sj>4GgzP8exafehL}G9+l`{e9kz!nz+4~l zoa=gk^%C+HXm6)Fp&`)$zt7>%BKdh;@Yf})8354@A0#q6QYx{6=~ZHd?_yaLCf(an zItVatU4O5q7sA7UMx*`ZgVlwm;*D)Lq3_*pbIrQLyh^mr`;F^A;?^o9nN1v73pBgF zw_G+RYHSm6HB$Eo82dv{nPO&X2AsUvwUZ-ol8y+99%pdcm!n8^I$7>Ml*fW$u@#&m zFa#G!8Ud?O;ffG5$9o za7{Lp#dckJANoBM{A&V914#5J$Vn(ssGfy*YUU*7G8?o{Ol|Zl9$P`@*tK?RSGnuYMccrE8 zK{J6+_u%5EsH?P}5kjbu*m@cHk=Vbolvr;$Kw?Z>SJOV`m?T&aL*cD}cw>~XTY_VJs(6I zNoE|Ht^pu4(F8W{J<#DH3^0`Gt8(C${lD#KqG+H&2iLA}C`|8md7hT)fb}7L(JnmSl{gLMy^4^&t;;MKV=EkCqbOx`V(f#RT=gP+%c7^c07Jr6?)x=?uSe_u>D9iTky|Lc3U zQ&1Gu@YKV@BXUjObd8t*Q;iryRti%JQy^Ua^|Dg)8)n~eE4;8wR6#e0hw`o=6sv6iFY^VF*k zWwMsaq2;PwIAJal%3lZyeNkV^QfH(4fa%~noA?(`Ru=9TihnvGiHUj=eDFN9XOB`a zK+yHp8R&AZ@V!uWdbojYX~LkkbJj{aHPQTKkXLithoVb{KQ)30W?+A~&g)idVGLh- zSMFmmG@xr}_zAHMp{&x>BH8im5xvaog_RuQX6f8{?apGnfkWla6Y*`NH$n`sn!!}_ z=;9M&cOVvlS5ACHo1Uy4k1LSks?D~2FBW|a#aC=*+Z15vNoyQW zyy80q_UBJ-CVeZO;f7~30X@$K3cZr#_QNxn15mfiuE_-r zc4HE~GAN&mBpoM|^SPEg5|MD{sNrqtdhO+I-OG2)wsDYE+?L1jW zM{XqO8lf{{9=Ab7a2nUrSt`HyTmC5WDw4w~WI-RxIS|%Mq%?-)lA@dxH{QK_*J<(P zRkYnZ&OCSSgdkJVQB|5*AM$|Q%e`s*zv;g*h=OqBb0^@6f&Bcm7Jl6+N9Ja}Hvw;z z3vgSF_4KSi$E$L}J4TOc#kBd>sxc&3kDk|&`Gg1{DLW}#Fhg2rns%Od=O?Za$!i={Q=n_{n>gd zVc_&JGL8^Pv19<2XixrK$#jq03S&`MS5Q<@A@ebnze$MLIKJQ(jqO_mK5w;NLs_AV zg^39rkOO-gC0-vN9U1M1HGB4wxuo72tD~<~Vq&FkRpI)w9P`KfQWlJiHrkb* zGFd{O@#JuQ1XXoqQGj|%o)P?&!Q~=C+szw=KlT08lA@Z$FpnRvCs*n;v3GHIjNK(e z64hnkGeWZ$l<>8%zywm6cVE|4CMG6NU{#syRc4&dZt(E%s5+&@5h95UItPDO0-lzK z_QuLGzFb?g5Pf$&QM4l~i(z4)cJ$Y<1y#d`0l4sfxsT^mR}*nR{}QoPUYd6C`}c>P zU$~Ha6EY-zk zO^ny{#OsAt9Yg}e?-aD&JRu+t?RtPu&=s4*rH@`0z;Z8JB-o=YCw-EZ-CUh~f8~qH zz`hfk`GaD$j7-eI2zHYDLD_8h;Vc(?b$1WH<2}`W`~qJ0YLN;)*bj8g!LwADFE3GU zsDX3vPgV6pKU>Gq_Omvsq9Sb{J2*MRVh4k78;$|LDqKJ8!9>>O4;f=5nPXX}2%c8! z<*Kuqd?fDV*#)OJG|7kVgKl@pEA6lV9*sJs&q5Uf3W956Qc4PYLuq+=`Fan?-P6kIG61;~6#pk!?xoac56GU= z&}LMGY|b5jNFSZK+N8+5;@yy(PlCahrm58xBS-tgYuaHPe1T?vczMY>=rU9{=$`V; zS?&A~WwqE=Sv$oxyTU!=95AIMt%X8hM?c1`b*SDIUD&8Vf;w=od#$!z&Jw))vkO!wp+=VedYU>~Y{L<06_kBKA+KXR6IrRRMW{3bn@^CYs*yuEv;P+t{@pt7Hq06-%h}{L0D1-hHO@Rbkmc zJXT-hYpGvLFyE`5<6)0A>I36Ory!}0pGWO4+PXx@^u$w>F47O{Yg2kIr(-)I`#Z{E zp@r7E`4DAyM^#C@*Zjnw)(fZ*I*h<{08yX&-IQvr{+EZ|i(dO@2h1N*^GNB zxNeWZoWX;aW53>0lIgtHfDNcNBlj%VzOJrr;g8BFvHeinfdfEeTh|YtJ$sfgi!e57 z-k9@+0V+l|3-UKWowsZo5O>fPes|*&mP<#vZQE&AS67dRPukDuGZdL$aK;vrvzrT% z+0td!IRz=Qsu6d%`IQY(ci0g(MF(IAbQXFp*aW`4@S+Mn-s_EDBc@!ku)`$N=aN$+ zJ{8mdI{u?HH^e_2vo3c zpJi;lqt)q@FPdcQJ%V-TxXU>IP);EXTzS8~2R1a!{IpZ)zED!%>O@BBY9w^-l`J^O1q%=5j3$-TR^@k{If8}#<v^He)4gp)RY;88hmV9nQx^7BJ;Kfd7F8Bv;~=5 z)QR)x6eHe*Rb|2P4z^-ohj5Hgt&-Q!T*@hng^p@Rzf# z9h*~ZSTx=~hVInI#BY+jEo4HN7N#IvSqFVV(vxL;eR--%tUms;Y{+0T%4@Fci zCEkXh+Ec&A>%-3xLQc)CTU~Js&=zBnQsyo>KxA|<1 z`%Y8ZcrR!4^w^lj!OBViqA=bHWs3B-xg}xh`~7_3;@o7C#+=jg=~D!pm=U#n`c4rA zq0PUr-vP|WE`+<(0{LH`;mf{sRrUcSpgB*SRi4#Uz%Lut2`4qm(oT;nkK%5TVBmWe9Ad4Ps^lGb$=O(;hUWn#r0wMW z=e~G!v27mX!^~-t`Q!)Vm8mN+`if(z0!y}0p)AJrCi#EDo?#`HQ{c#`=G^PWNkR>h zJrCVbZm+AwehQiqw6wIeH}iy0ZER4vz4{()`5E-6K)JyJ_w1was3s5(897P-m3mYE z0w~(LyEXV|h#1IdAQU}6I<&W1r*;3`_nn=qrrQ|c_K?l;r(dTv07$B65bp{GuW;1c z{mv)-7+`J`buYdPYJ3DY-|)bd#^T`OLSj%c2Qd9*Zis(%&{HJFV37T&YeCOHza?oo zYYwlfMsjjT&%GmFSl)JC zxB1!VpzZ8>_Jsz!@VNE3#S>6OXcceNYajOCY5Ael`i=8aDsgt!d|q?-pu+e3`N+Uc zw%hL%rZ0xg`L4|p7IptM>f(#Gh9ZFeoQ6R$k5eY`4HL#-{2J=(EFU?mo!DkM zeCtd%7(%F*&#?uukQOtHBu)(0e$aQLw9jI1Hb%k~m*UWfjl4r|APln80ozO~|HHq+ zqHuW0mcgZ@2bou<2ApcY1CJ9qrsB8fOEbSC(sZ~YX6hi`8G0XMUip{oWzh}5R%$QJ zvVkd&o**ZFZB(Q$0oN>kS|FHbMM`Th%uqZueyu?t%QyH5U)Jgfrqf&hfYmI?{DPi1 zBl*Kqr%xJgTf|OJ?&P3h2gikl&URTZ=>8B6q1$C))2?Y}={9WH40V|)G+L#KKc&)< z61;bZY~l}2(yza4{GCZS#P>_*?F$!|PYi|mnR`ekJSr@_&MNOhKQ0)d^5Un42Qyah zh+tGy)CvT7eL}oOps6NnUc^Lv05Yxh`}p8k_6)JFs^9%tC*FGu;EF`*oE$_l02XN+k_ehqLNMgyEl4EPedAR@g-PEOvwyvzum zaslweqPKxFKCXV4f!`z6EEXKR&h~Bal z%1S(JQwAr;@x3NaG+N{QRA05v;bQ;sK*+b8=`u@(&-G(P(!72Y)AZ2p6xshRingYk zd*s)roK4v9`H9g=wLlDO3HPhnsA*BuNUw11lW*g1Xd>`90aYPW^0v%gC)i;6&SO&U z>^o0fwut7&cPARnN{{xHB=Z8sZQWbf1wXx2u-=FOium=hS&|_@?yL z(?}4Gtg{F55<*~$L@OX9q^q7N#|N_7-90DAa`*NLKxI-3KO+dZpi%z)dohUHCoAOX z7v0O^9c(74M(neQPS=k3G*3_K6&B$bO~Q)|Lw!0po+6U9KftLB{`gVqX6!@2od!on z5ElDURjR0Edr3RQ+8OVK>nsKmDer!GF|m)3!^{B7K60h50&khRl)aU-BG zv|4%{I65Dv@;xi{ zY`;TvDoD6AYr@}5CZsMU%um)_ax2w(WGTGx*l@j(;&S&ye?HK5h6@fwzDOJ^=EnPh zv}%ueU&OAQ?qZK_(|7}t_*C)FAY~W=mrA`woZI8qWLJ~2S)$dK18}+8VViv&<+J-7 zmr;o|CGmYJc>88&slixU1O`fe5cNmUbZZjlWhyoBD`m*v-0kA_B|tE+f?w6~?m#Fs zb#WP;&sStdZ}%aRFBr31lY2=hO|02#KCtgu&HVm#C>!>-U%l?=!D^*64gNNDOdEY6 z_3AE3WJSten(^FlNqtL7%dQ%kLcbuY+&5((OS(N>4qq;w4H1(;F8%?m;YC!Um zORe(G&!0G6_Qb@)$FNo9|Ni60N5sehiLn)YYLgp>l1jIwf z<@t|HS@;cWs#L4Y|xN7e9bNw%>VU^6o$k6Vv~b>gK4!9aQvP}U zMP49@@l#mRQx1Q-RgH=y zi@^E$`Nm22zFP_PwA?LwZyZa!^_<(GBR*jagdqsAp}8zzu^DH%H2sRqu^udAy@%*r zT-;l^MQpioQ02)?e0cBezP#~gzIAw5*=GsY1Z<{YYzvb`D=Nx{AFEiWoNJ0XI}KcddhOvhQof~WjS zuj+C*mgHrK2x8c2#qssBL(QLyF!xCW+t0JeAN7W71u60?-j)r_Nuwk2iFhPDR#w4= z(b&|4u6N12jD;p>^|nM<1Xm*#qEv*%`E|R6Rxp&=f_{MP!Gi}#K=u-J7M-vm7b{&s zR-dpfY%83wk_%ynN|=l6J6| zX<-Q$Ii!^hQnbCk@Qi*ex$RibIyPCu%9}Lr-z~Yuj93)>#|=*yb?ihQ$$s@&Zt=U` zR>`WCWyNkjQBAsG`!YrmCo(Xm_||uCsh2U)_@Rv!BAD-EmJ~aA$-SRH7vm1;=qlp^ z$_LKdzMIOTzdoqyhtk3M@QA6?MkIomj4|?(V*+rgaU<^7!LiDBvFm1CaJQQel)+pk z@8=4#CjhXz({n>~Aq+Zd$+{s<3=EV(awtOIeU16(M8kHVb=pD0KHiW@av50d=_~`tbKEA!P6x@ z_z87vKpfqkR1L=s5ElzBul(DVtqHxG%KGLk*PQX=7_#w|b^gf2&xoVQ7z0ku@_FjQ zAisC+rg*xDJT>5vAh0n|kyZ5Zj*bONZ1#Fd zbVGA99_y_hGevg~4{5Z*jR2}9l74hKBNj>gOcU(!_uz%}~tG6j5WBB{Cd&yVvW%oevMofy`eqm_eYT<>y|5D6` zQ!S9c6elUXV89R(1BVO+?ZG>oB*74#r9`WhR0SE?^9Eftj6f6tPJhXK zqqUJ8Aygh+&WcjZQNPmA6%6TLQP@^<yM)oUmduB!&nOD_bk8;(oS6eg~^+p$z zvj}&vT{W1jaM>cZSdx&LHgXx%;2<9F#j|gp3@FED%vN_^xN}9%`AuQcnn`hUVn%>P z5Z#MyYZ^0=wt&*whePzePM~zCet!v^s+}UVBg|c)X~>cSRrd+|&)~ zK^7n2YfMWyNw2LocyHxp4M)g@Vp0n=tdi)lF%Q#e*_xk)z85h2g}zK%kr}UH9WwWM ziTnU3T~-z&Q~$oS@8?{WK74j_oCLq|069KluC^!ldI^$mEbkdAh*qJ+6ZVo_Eo4fy zMv4QBrHBv$Ts8(FRywv%!0Az8GHlW0vB|*}*G*GFhcED{h^(*DD$gy28{~JS;vGaf0!Wtt=3Vi>48t(u9)3{n#bUiyY!4?6 z7;k-Ebcaz4YtjdvG>nzK{;^M>sQ~(kAxi`%G;?z!26!J1af}21cu48Mdy5(yq8#Mn zLItwpjmx;QtPt7`*d>J_f^l+)+z|c&X^@SnyF$jnM-DyPi#CQNN7td3A zc#e554WywgXlQ68rlx*^O}|2`qX{d=Q2bapqo+ID_(g=Y^ zgY#C^hxuNb1ur^ROrK2i?hrNMRo5I$MiaY4S{>M8Y(bT!F;-U!pqXbA7~ z_L~-|T^X$syF>6v8^ zZ(L?s7Q<;fQuvd|mm3^ZoIb^7;1LtUyRU|VMgEp3gOlmIM=uH7nAnu^1N*ZX0Np-< zrh*E0EOv46I`mg7USuQjn1h#$$z^m%$ZthM!C6MtM6P+63fAG^394-)x~R8*axKjV)e zc3Nsfh<8FWfTcZvl4v^7^7p!K`kmJKMF2sgJTrS;#qY0%&kEIoReyv+vSUP8VX2;| zaTDjOl?8PL@?f?7x`U{)&7%eQoKLE`nbheR+bkQ=T~T}R8wV1dP<(15Ke69(N3yX>_3@z9N0IF zM_=7=48+Q}d8)Me+D?gB2pb(AX^kR9w}m-4I7}ZuMq61~NzBOLt}id!OiWB{{{%UI zVHohKc!h+50rK7mCsoxwGxk^c$`L4V@!_)hJgf(61;i=g2^>T9{vrm&a*{>F+sEhc zSVs4!Pk}W3K$yCXP(+)5C^_yopTD^@kOAx&CDVp}lp@&e-xq<>RTZI>nhTidP8~N# z`ui28+e#%I>I?=(1+KyJ=Mxf$U~>Lbr^HnLr9#0Nrinp00Zu&ht&pCPpPyghx}i+| z;AB_N&rbsRxME_4`=@Hen1ufM{LK#zt2?4zYWCt4Sv1sK-U(d|0Trff zvnpw!q4*c^Yf?3%w~CBB%dvx1S(LY8JAHe6{5!uDMGg*&uLi|k!7$Z%^Ai9U#Lft@ zqJ-W9pn#q~e}qZN$Z%;3tSQqW8%*rt>_DD57U@TzJ-l-Dst74ldj>B8HX!r<(^Tgx z@d*j2ajhYkCtlpZ^MK|NxAh=xhqz<|P)g{vChxJXNDr(PkK2Ark7*0 zW2s!7=SFZ^nE{L|9sQSd>JYB)_N1&8#ZWR!B%5%3O5H~%2k)M2-=4CuE^KV?x*@YX zCbuUlV&B!!AOesTCL0I5{0Mg)mY&fHc|3RThF4bd;^E=(2+KpK+6_FdKmbIj1_lN! zUL|tFK(UuR*?a|z;i$nwKLTcaL|l;`#Q-}|MIUHOQBj%{)a%w zQVdME)AjX&gaWn(Z};4e3Cpk#4NaywUL{U@{wnIrYh~JL4PEnXzmM)Ou`(@0avS2X z4B`*&PuwkuKZZMO>qQ{`_svZiWR3(HFw*)V2QD#@MdVpbpLvDF04ZYhNJB&8AzWQv z9<}iWtv!PQ-kjLBlaw}Ojb=-J0BoR7zyiA84AE)V-O}KCWc+)2$}1{ryMQPfk<2Ox zRmF8PGDCbXd;lT5VtVDXn^Cz4zTU{B2UUzATC1>+o~@qw#!AU4u?e<3ap1BfbC2G6 zZd@X8P~g>r0|l_x^r?rtyP2maIb7aH_ukWL7s^kH5KVue6pqSU9gUK5YaVG+IWdTA zP1d$SS-J%5sR4{?! zaA-N42ykBjYydhYHSfxA^j68*LeYB7v9$s39FCo@aHU1L&TBs!SFD?Xf!Dod(R*Ut z?`&6I>T!@FY=0pmG>WE!BOS%%=lAawGKs#o{h>O>6DVdzev8t&$rFu}#F*$AmAxAO z#7R}dKhXaL;dr#a{z|6{fO|4qy4zL}%EXwy9-yp8b)HJ-F{27qT#9^veT9RS%Y07D zY=DcULM^Q=m#+gS=__A}^Y@9WW=aMOC?UKsS;8$u3^zpw)MTJ;Dd61X;N&zzqE&!@ z58W_Qp}%}l`~CYj#x6%_Vcu><9l_KfLoa?#cy+HM|(d2hF+0ul8BWL|D zAbj0$PcuVWYoCV&V<NpIUMdKr*OwKhBiMA8vOQ< zl^6Iq$U#z4(!wjxS7&Xa@rveDFDVWINTgX+i$8%#_StT`emr}F1yAY&DHrV<*pJ(< zhF?mCinN-e$Vy5NXowk$)J&H?6#X>tb(^{W>vns8JeJLC1<4943@@fUR4m7rinn2h zBOnlnY|uTX%#cD4-cu@-gtT)78W21l{mwBJPirAj)_swgR2PQQ`?5yko~@zDfxrGNMCr} z;cbC`kmH)6$!2k3|APFyPsLY#SE>hy6`eOQ!$|npW7QvOKsvo1S7kdU3%`73nC!CG5v$L**@=aPsB)h*_L!0jDBMyXC)E`Ys4%8PR2~6ix%h!(uX8|VkiElw>}Py@Ns7{d*7c?M z(DNt8uSxQ74xlF5=Y<74zNWe)AwVtsahF2v<)8f50Jt&@Hz^cz(W@2j2^F2_XJ%p& z<>E0dkAzpO29cQF;EVnNu+;-OIo`W>ag@e*1Oc;J!-*yyuNqO#?H%ggLx_OD-WX5Y2WVOdWO=)HI49jk^GYP;#_x`#Qb z+_j)(VePkvcz-hAcE2&cs`e5`bW8WO+?PK}^o-d>Y<`cI#7OSnM~kkca(?^894Ab$ zmU!vcnRKP8GJ7vQnYKjh=VX0y625_l@Y{o5Fmlh@2}K~s8D1NhWo~U?J+-t%?nMXycfrzfaH3#NfZj4T!_EFEC{0DvAFI2;y@tcMDGL*uGsH zaU4y6e6m?QP9r^vk-bsCf29UyyxOgE;`<+=KS#P`uxcBTvTZ9d6#@4nvTz(pjIob; zSgv~i*v+qu{cV;FHIuL*WIU!*LBH4hF;?RD3QFD|t1YWl3N1#)J3*)GN#0d!#P?f=0c#YW*VHfgoLxxe~UHPry?20W1*ytG*S` z&%ZqdX%7+frh)iq;495j(WMwL3m4_u9(bt*6 zae@d|Yu(D=2WFG^059XOD@%X_7E|hHW|p;DYuQ|%wa1FZXrr$^#Rsp&NB&KjK>o}h z&!$Oj^^A+I=??&^o(QzDsWKpgSyP3v!QFJUnXSccZdt8|%uFz_!BMo>fu8_EKaukT zmqZhzh2c#?YSdLoFGf-di+e>;G~|+3CdYN}UUzmYKTqb*YzkVd5sG zQ~JI2u)!%uI?(NRy}|n7-@mDDvF%qvCT~X2G4kC7gYQL~M`}hn75_2?uC0SQrN5+I zcqG!2gLYE4&VeRAKAwk{_vPN+-dkn;U^CdD5W!^Xn^GPp!Q$pCF4d!CQ{Pu&!{8Hkp1+;{y7HAD;!Tktra_hQ3Vn{L8VS;3e0J-J(4MhOtV4 zJhO6wR-u$6!TcNtJ>v*KAh+5s{vpeO%8Ck8c-+Mf;Cr^W=lnVZC0z zwQ&a2#E7G6EXzBJHo;4K>LtA8kmJigsnBBtoicJgh>20B6uT5xLaYRO((22(DzVU; ziTE!y|7yN7)z2Wa)g)05Ctw(Ssk-!{eNC$3=S-l`njLo{Q(U(CfR)3nr+Fy!pN)Sq zzHPLoX5W(b(ee~wdmHJ=r&i(eQc>cG@f970F`j)+?=!7mC1es3bvQ{)3S!%?dvkMh zm#M+POn&U(Fns-~zEL+va=sbxP7pV1fZCBy2b~ujhPlvPIO__zKCYJXt@I8iA*e6H~9%`j*;qWuDE6m?X^VX1CrgP77Z`Xx}nvGIPn#lH;|M4YPA<-PFO=Z+^9}H zrDj66B=^)p?_z_9I-{IP?9Rs#oYKy_38bK*UHg4>lns9$;%v0FRYEh&aWCrakR>8a zv74#A_QUPl^{3bI@OBMp0WS$XTJZBj3|TRE>ZpY5+8^FdBkuc)SS2RRdO+BLY}GbB z!U=7~q)B#jImQ3+6d{wV;Y$<0p37wG4j$scLwbEUM6j`ea5#&dnYlT8osO-aKf|5a zUnM9PFy#3U9}Sl>b^NO+TFx>xrI1p(sZYYULQ9;DA9PqeqUC&l?(|U<)lJ+By(1Dr zbuHcOs>goIx3_2DJF^Tqm3+pVkuUw^uwc0NM~d`Z^z6x-Z?zG%2YGM1LD! zA)LY!ZB~IFbw6`ws?Ie(fWJ@nCtcomBali>Fko|=~n5}AzV~R$>;A`49&zyLwJU`Ej~VE z-L#eOF)DS+S2fiVVF*vdDi$)>@3Hk!!Y3e!m`tzyeT~t0M($VM?w>4A4pKU!)i%c< z{#W9QeOKr89cjZUmDaC+IoIjMd+bJ8CG{lDQJ7Ok0h`!;%f>?*3-9`Mge8es5kORL z5L8Y0)bVbI`XDvwenSN|_LN z1GCJ1lFUH`5FY67L^JwMI#JR(4~6};o|6Q7Dl55kz7=+)2xE3W@rm(FPsrO~O@);i z{S=o6?2oQWIZur?8up*aZ|J=1OgemRyRjb{u+*HW{YN;A+A3(!@M^erhF{*UJh?v# zVvf}_D}ucC4&%{J%~uAg=Mw#;^tR(njJunc=`D?c8ysB+ZWD;7BE^{ zSuVG{4Z!ALTh(Qi{E@e8nmD7d3h1nMkt9X!$%$7Jp=Dkr^nOTxwX|dgpw)-*T*

z!5Ep~MDCTzh9-?MTQ$4kvQ?C3TtrT&wOI2ree?fg>%GIV{`>!737Oe5yJYWVCW*}G zls%Ib%HAtvm(f5cqJ+e0WRnpxvbRL`${tx+zvtWMd;gC6{{8ObxQ^@k>*{cx=llJ7 zJ)e(R^Sm)W+g#pI|EYm_2UhzlGF%Rtncf@tcOiu|*w;BqqxO255#?4}eVg4>$AjT*c1jT5+7y(Vb|WYIUttP_UZy?20}DgT9%{93!4xzK${|1G_;32V8A)o4)t;&-=NLV z#omAK7shVk>9~t5T7_5UDQI+UI8yPqCnhAUc+CFa%U&6Yxs*X&Zu&*e0XK%%9jJKt z{u+h+YO}U>$TI#i_o)9=+D}>e7)Y?K$xr!1S8dD1lLSLiEazlvoo-05d=tP~pImD- zpvk{o@1)8Dz$ic|bw>0_y>mpzDTm#2L)AMX+*Idqx6d=CT6XZ-`)Hw_vlbZmojRE7=X#-df zA33HuQB9D^PjU!)6UgpQ6X(KAv*P_;0 z87o9gptxDqt(-lV)I@$@8Lje1COYbsYo@NW%O@dC0fJ?#)=1$fL*sD9?s*FG(!e__ z9W(!$?@UD%Qe<*rgiY08`Tb@Te>c61LLK}KuDE8;ezMzwNPqItEYSyB`o2Ih2KrXM zc-7280OcM6DcrQRRm9XaR&1kLkKoR!sm*Z8!FrDzFfdQU9vMP`O@Of{hQkdS*A(?Y>RE)JYd}x2N2)GszuF%Q&a)d*FK7Obr%_}|@MDA2PHSY&0=NN^jJ~eu zQ_DYq7zhN`gTLvg^~XCfe8rK$$eQ}o)djo9?R)?aAC+F~-}1Y!xGe}%+(onvoy}Uv zT*HKjWaSdAC2o(1M5dFl6QcNagxUi*gEbb+FVMmpL>F(5?P8{ZdUcPT&NY3}F)?et zHC)uH-aj&!b=7&!(-PNH&QT9%`Vl8-$^yZ!N-J-Eke?Pp#xe#Rp)e5qN6aaxEC>Us z0sA5R#URUtOX@{*wE3e)t$K0E5fRzwcTfz!xU|$DJ-&IdqqCDb<-E&@k*dM%@^0*?L;b8KM?x)CSe>$xT4^r1%mMxXmV4QFQcL#VT zE`9^35qU#}h0V(~N+A{L|7T?_{p56&*8;X;QQyW(AN2S~Pc^pa&;wgu%Nd@Ce(g^1}A-N1Bi4tkE0>>Vz9wg+sp`gH*}{)eIa zn&<7zMhgyHL=y>0Di|038?KrrCt05*C|tx2d{obESlbq4yz%KBOGu=X!eYUOAmgMZ_Ha|A{#$$+G?An(hZs_yWrRj;^j> zQytUW(r`T@K`PMq^HSr}s?$8sQjO3rJ+K9Z9zdu(ClaVQt0O|&(7Oa`*TxnI z_#@=3v3@sG4mDZ8cg@rmDik^>b2s7$E6*bQ55|r17w@3dpKolMbT{LzQ7$Q_`HIZcTVzALyk_xr5^oF`y8enjlAIh>xi`p_`Dr*x zx>2C~k1tBhe*cgSx7p61OUSwqc+K^7B_Q9=cuh!xqokxf<2BPN0Kx_9$wg2EFv2;8 z9PS8BwX>u1q*E2X#}1&-x?~R@r@(Qg4-O|p#(D&;LJi=$ivbV(LmwXoPy+#K3bg3r z>S_}JD9;e~9jO&I-O152x3tXY)x4~IHp+IDR8dQ9UurFpjnh9@0;(*0HpEV@z>acC z^U;rWzxo#fLRamLBF7|hzrbq?lE^Wi<-7_gmvI_>QVuBd(8&q>)=1aKPcu&s570b= zypsL*nuq0T_1?TZh+uJ-`8CFiuXqrMoXKvEe}2ZYbh~hfs)mnA29Atgk2YS3t(32*RR6 zAMuCW-1dt$9-f}!kCp*`Nbl|K4MIeoobTMUK8kbbo*i459?LU9We?~ro+;O0c>f_q zGDmf%d#A{d0Frvh`2xRD#iVR$AvP|OAY}?G#6#$F16UJydI{*hTQ+4ziFL`%!&Y%o zcK-o%47cypB0hD2ib-tgkde{R_S`F%Esyhg+@623?)gT69vThL$Nh2*j=Vn~AKJ^t z5zs;WsMAw<@1j{29r?E!rm7WpUdES%%iN}+iuX!7nm-Nv!h8sC3I{Z(5&-GCe)h$2*_}N|L*X?|O^SzQ z@C4!&tsu{v*}uL-KfxA4+G;2tYeo>A%kipgpsK}3^)#on3UYHW`)%Ah0_Q$&#c%!> zl0exG*Q~iqfE>gbKz8Af5mq>WD<=&S&rX=b7CI{9EALjO$CBVy z%N!J1)qO%HR(JYyZ(CHZ5mf%GdFPnwx0iXD+z#$e1zur~O17@@dM2t(6+Nx`*0dmt zN0yE^pxXP}{AyfY-$%km^qf&rqQEEe@Hnaqw?bLYgP7}l&M$-5d2i_q9t=cNSl-b@ zQDGo231IsA|6}ET&T|uy0s*g#XaOJ`X7oM()hJ~r#f+Iua5$5Po;AI8NO2=MOw)q} z0b+<5E2H0*E-dI{j1-~Vkt}{zzb!%%M2{2O-HMg{$CY2RYj^}hS`*B&Cc6RP)I@Sb zkPxS7B-(+3pq5(P5Ohcx2CvDg1U}@Xy!!Nyt0Cn&DQKsEn62E0P4^<#5{?V zd?C@Fa8JTUj1WVc>9jT3Pk3TwL9%^zup-|PqLv#ubO-Xp+k52X-=&FtJh&|DxURok z8tCu0IJr%@C^ag5|M-oujriRHpW&RXAocZ61ll&jOpQ;K-z43a4WSt*EM81OrVCk=^aLh8#9i*90#nUd0laeyACR~?^9 z!?x(ur@}chVPs&iS=vdJ3UY+?#u6+ejr%hS{=teU5Er%%xf@!2md3 z-o2PkjFk_a-iH{4Ty=iU0xe6-U%vxKu~!FLbIU;y+1k~6(VOE^c8c+%!Fp2J&H)9# zT6e^Y;=f(Yhr`3Krc^R5Epgj^}o8opn|1jyT}d__TGN9ByY1Fo&3**ox)t zr$li1$;e-3-$L}aX~!!zn>FkHq<2VkT9HZ}`}|70VPjVkB;eXO;9w5)aFGwxmOW38 z_h$UpeOaLA>qFAoDMXP@pB~qph9eECh`?|2Wvt)#y$cs zw3|QXMF> z5F;jTwfl*&_gs9%MgtMZyOsi00KpTeT4YBoNYG6$221MgpAC=x?(JECOpe1`T}Q_m zu<#k=#~%kTDw0FU5xcs$$hDsYJjY!Wl`_^HvLGot=_qVD9BsD#uCE8i2|BGkLht_> zQJ_-AWS2FF!!HpP3t(}3ZsMSBe}5>^{1Af1v-!%e_kf(ljZn#m%(k$N)&k8|@yqaK zaTrHvJ+uF_0`;dL@2ehZM7AVP7MTJqod*D4*NTaKkA2qbQ)^{(r9_H8#BRaoK~=|4SCnV%vPPYF zTt*qPCFq>z_Nu(!eJ<<+cBJKe!go*HoV3CE5J+f(S`#8}E6nLNY32{)>v$m1=XoC+s*K|)Llcam(^%~L-=#H$ZwAkK?E=G_aBc#w<~gd-;* zZqYkPX;8}0yM=^gA*9?=fU$F~2eeFy>*##2dG^^!c~uF8gLUpkV8Y%(b#mPot`Mw@ zkZbxorFkKKC(X=h?l-tz@N=i@e{!rO{S+~9Enkuua!s_Kv>Xu6-5W?}bRAS(Y!5ya z$+%Nbm|c`f{88Tea$}3%J276~IJ9aeLp6tjf@IV1&ypy_zYy_saNe5KYHLfLJzeVY zSOs~asgLcCRhaWCR?9|n@Ni?HQg{qvZ(xw%KNts+A{w&(IkWHPjY2@7Aq5p~Ao3a^ z2Zn0H+a}q6y|HkK3*!s7YxHzQlICLF68wTgw zvu%uVg|c(=gIVlHjn417HV{v@=yq`S!SBW-DDWXWRC0bx>GIqV8{^Y4{UGsL0WOJy z-i&>zO8c84fin(YkG;>wP96OD;Qhr)65e=}-)09jIf@C99uKRACmOpDAX!pL3CKD$ z39XvYY6oQ09Ix4CO|&V@iIZ@T-}$i$4Cy|klo@YRC#=KSzFAgYlQ+8y3mO%YiR$NQ zBMNdOTuL@f9KEAQy9=JkkMp@!l_)*s&UeFL9iIQ^{yWfI0vr8=ChSw8POf!7tNd(! zZOn)Vsy)Z8;gx|h$opejL0nQN(HmR#(4NNl8@xvl}?QQg}8@~o@P98^k9S?;G; z*l4YW$NJ237YJ$-U}r*Xn)`O~3Ate0(Y{&6$H|`43kwtyNftjAfV2dH=^JBjfK(4_ zKqo6w-ziXW3}J8eh#v)(p`XKLL(2$A3;#D&htshI*#*PS;#c56UET*Gz_jn*5gZR! zyCfGF%|=Ha!;-s_24#$!36Rx8uO50kiojJ?nnIUVG(xVRRmUXPDqW*(?0XGgo_#vb zz`v9vwZ+@xIfVM%6)q#$-;^>RzRTcSvd^Kpa}o({0Tb*AcCE=&;9+I|h8VDn{~~HQU;M~W#1V#ds<_ky+)a~k3*Gt65N>fZ{U+EHX6$bOu-7K z+}l7Z2~_3*ZGcQNOQu7}{tXgZ+EWM^F>yLX_H{%y2*^^+!s_GeV0M7)8!U)8S%Z|K z12Go{HB?7R61l?razjrW^i6|%wk5xqQs`SVu;xTfnGF-dTj^zsuTn%bXiK(Q1PBWF znjxz5t)CPr+4rs$MEVu>jgqTmD&bOgW$dEzZg_P;xHQM*5WFaRO0>DK;@Mh$42*}F1t+(kDfW2Ty)%>yQ z=+mFvGcmDli(dB`{%Y*#c}QQz1ECo4pOI%aGQ>MN3Wy5t_w#M2rCTL@o28m)$kTOkR3-`G3b?19Jr--FUu1L8_eXy$XVZt5Y2GS2<2?8VEN=y-GbE?x%|jt zPMkg6Wx8rXfNIuMX5aIs9fxy(qJElBe6Yv-U<10eR4jITJBvMN{oB@*k|K)7wBk)# zc(1wgN7ipEy0-m?G&&a}DRdroc7q|tb#gfcZuc5wD$#xh;b>zpgK(6m!KDg1^X(FK zMbrj7&$Y9=@YyK<$J!}T7j*c#dVP2ls-fJv+hzV_yzUts)Y+~wq&|=CY^;vbXKhcI zuT&!+zjxOCdX{OTXjo8#`Eu=kqfsND8d}R3H>$364>L6V9L{B6K$)xZpI9R`nLSzGwWBxbU2J3=?sWP zKA-;+BIPw}zX2b?*b&K?Etngc@vJ+1x4asVYQz)Y0A*8P0*B5*lDqvfg5@J1-THZI z4SseQ`vY;@q6cx3bzIF*HNDL1@n?9+bI##yL_{0mnnWa<`qt$!*0&R7)A7vrpUe)MioO2z+_hst z&`pO(z|H+K2k43e@~;Mdm?Ms)7*vy&|0ct78!@l5ZWo03W%#)MasRc_x5Yq5JX zA5t5Vf9W71)<{g9?L8N^9U21t(!V*lYfw=gX6TGH$4?*WUn+j`@hXMei_9R4-pyj?$S0jN z>t;}ke0EnZ*2*mw!fAT#`m|wa4nE2%Tkj$XsM_(6z^vs}i2150n9Db-V}^_hU=1dB zZ=N1>*CrV8nhbA3dO`fy+Mf5hZ;-Y;?{k=TdSrv70oEo?5T{VGL2qH4)>fm3ItNJk zu%r;RS58vFvJa1?DlZ?m2?J%6-}qg8*H`)D{^9YOAhE{Q;TAUD6($dnfSl2b-dmlO zb^PM*c~)M}IAJ)WZnp{lgi5QN+jGa)_m-$q_L*gjjIJ)_b8EGKVJzkS7_&RsRY}L4 zdE64c?K?DM_paq-AI2%03zK$xM-)j|pb{4epxwTZeujM@!*4nR-1A>%%U~m3ft0A) zgxnk+01U%np2=DI*H8$<@GBv%2Y8CdJmti7-K!87(gezl4G3jnMr1NTb3w|h(3w^I z7P3=AU=cAwa3MsDeyLr2!Hzda$btlEUubwO`c3E5!7OJY?Lay0S<4&$-o@g8U_Ir*F!)a%MuM)v1!jLD2S* zG@`XTjC?mhs_!|VEzm0qq+&W7y001p$+5-=*faQhSKh<%!LFW>lh?UDhSa%(P;EzVKpBK~%*?D)f# za?&3~1yt|HRC2doJ}hsE=;`+7iP1-H$N=ccgD9Hz5sLz=#1LRZ=2#XZZMMIL!g`y+Xu~b^J4p-jt7cR@h z&mbt`-%>6Da^NF(1;ER+BUM3t@LQMxg%Y<@nvzxVd<{oU;I=`W{~N6aA>zfbbB#`< z7W22?o<&J9Qq%f5yf#@0f;u#1_PsRyi@8|2o)kJ)wxmZy4tHH9h@SYa+*BNuQ&7Nd z)BUXUZ1PP7eUbIOt5YAvT zU&JH69scIdNqnN3qG*PqseX_d(j^9J85ZOe%>f?`#M^OI*3{MAjGId|>Vmlj+aQ2r zZixK5D@bqiDbJx~1jUi%9h$fv!2o#qB}Nk;wQ)q>`l4K$2bW;VbSf^JP)Cfj)BQ4S z!$5nuP*4L|R(iy%Z7l2!e~k`I33t1$E$h~I(2yQh6t$G^#1(c6J9pgAz$9r!XDb`9 z z-8NPICUk|^B;1_NBed~-*=Gm8NHH;j^Vo7WHZ~u{L)#|D_)0Ia8SRmhqiTK@K~7$j ziw-rQ_weCBfo-x(qD9r6ClRr@;7!1W@X*jJ7V`~>`#CuY00a|&Y$8zH?9Z^AMgX85 z?|+2U(^G0loI~OyOacD&8ILgpy}5v(V0f+URt)&cc>i!=9?;DdKd`pXS)wL|@Mc-g zW$oN&u9L3=tb1Nh^JBKTt$oD?B=gzrwjbLlO3Hplh0N4V<$shtlbGdNIWm~K&NO6F zi|2Dhy<);4^5%EH*)S3iY%5IoMv|F&Vy2A{l_1cv682D|O&Oj0!(blUX?r`jWUD1k zt4QnWpyK@1t6$G_RLG^G9fBNDJR;~9%d4NYdE(|cU#kp{e$3AP8L^W zTZNj7p|VFQ;#xVp8Y2uam{R)q=t%!uhmYx=$2s}5(QK~DP;mX6qe%CQkG(^Q!o3-J zkKdT(@AE5fUpIl<-s?AKnEM-404YoVc>#7(zo%t@Nnu-t#!VG7Smn+9s?J7X*#$6v zUQrCu`qp}v&%jj$)p2>7dT6C%lWO3Ak&vaE^z4P*{Q}22dva&hc&akaYg<9uR=)Ca zR!@nU9~T9*dE!%X1kLFB+Vx|s{4sSe_?y~IQr8#cg$7`q4P!m z&B5FCl;@&z{|S5(GfK+MF%&Yhd>W>+)&A=2xX?A)zPI49&tPv2g$jl8NV*A%7&)c2 zL(?nPdDJv&eY71U=Y}q2+>t5wNgcLtpn0Jb-I3fcnYi|ZAHK-lz>PAu_d6{ zo{tQ7!0w)h%ry*f`u_QT`wDF%m z#`k`H)syb%IF?YXR}1YhWS0q$f4hbZ4laeBQ73z)R*mSs+`uP@z#hxq2g9t??PCw;qR&7U#Sh$ zl6tqOL>9ON#FyHBn;sEv7?Bhc){>Huhvq-f6d-gIY_1V$QWHp_v0MCdf?D@ekl=4W zPHco;&oD@0Y^Ms`y{+puohod;r6bkV?%C5yGa^$ zu&LXywvR)b>GCQ_-{4t%1vKKe-1`Z8`c;aL-LoTEQ6$)ctgivrV~tW)hJ`bUb! z97EN$Px;tbk(ey4+iKf}jc8#X=_~6gbrXn!*nt2JiY61Ocg>KGLwJIWslCaATdHcc zaDGJA?EXS-H{9~O>gqT}kIUykoqWedXU5cbb98j)--ubMMI zhMc_Z)^fgu7nZqkP91seV^=aU##XoSw|VA#3bbkL1Qh5MAf933B!T1oEH;`r`2PD4 ztRsc6%XBd?jN&eQ4GC)CH91awL!x0^_ADY|%=zZsXxh$~uE+M9S~{<@X@)Ry?Alnm zgDSTEqAYDC;i*nCWgF{qWi5#C{Jz4@#uwMFwj2lTHPJ$3Jl97TG%1Vr@SZ)6QKX0) zE%gbw(Vo~qYNq|)t0hdxpp#P%^?j-i$&mdxJ)OR8xkw23_zdz8ZsNMGNo zd0f7rWBbnh5DcQdp0@TeYqU+WAp z)IPcIGS+OI8P=;1P!WAFW*p^If?KX;Q8m!&rrS!4vGP7{CzprVyKKATc+t;d5=P^q zaMVK($o~xFlg~&6D(|XBLQ~_fVVAv?k^5TSlvs$)!mK?{$IL9v{@&*2PpMlqD=s&a zvr~WCODZqtd4l>?%3oEjEcSh>HlZB}Y?&b_fe->Ya(GZ#To(b90@HGg z>IE@b>@L=}!Kb-yd~rUfJu7vT`Fn?*f2 zxKX_GC$wdZ>fK^9V~bc*5Op4(S#%$H2OnjZTi3P9vTFr4zQn(cbGskd)J&<{k@U6d zwW?qFNhX@Vp9BRAc4q&r8`&RYTxyOX{bj3k4EEl_{GHJHCniWgMbQYRV6J(pQnIt7 zqhg%d+TP11RGoEfgY%$&WQ2f#Rh(4yZgW*%UEfiHTtKiPiWU8bD*=RYSCR2ragvm9k!4xUM`&aS=8j=hxvV{JuVLPof*ZAf?)FDris7=wftZzEeX+Pd+H_rf`!bI? zg@|U}mWGL1Ik2mQZ2W6AdyCPn5i)?GnwYLKVLnC8?&8osN~g5mNPcI-h#Wf2Ba$uv zE36=P=Kq2d58`S-By(VxKRN>==SGxXbpzlO2xawkd~Kof$w*OC>FNQlOj}!XzzQUndmqzap+P2%WJO1W3i0ZNQ8&0186{jO zLH}~!vJ3rmMxnmN`0{5IYTq%j4c(+P86S-HU=^K>U!sll2NR zQhm8%wS27|qOO6hbMC%EUayz=MD+v;Y_)l4eBTh&DxXqQL^gRe1~z zJA2g~P_0tGvnyUbMUtBMXBg?*@;Ouktw_(%4?+Oc{+S5J(cjd?@9&9@qO6yH&*eq` zJ2E*WiFD$T?1h&UJ=UBQl_JS0T<=*8(wHqIF_>XrfqqlCUG?{~3={&c@@9AkDZK&g7y0V-+FRAe-HIxCkMmp%(`;y9*M-6z_Sosd#TO zkzjSt&Mrd8xeG3Tx_Qifmis*lT~mG;{eP+w7ZPF3htOPO&Pda}s$p{{C!rE|+M8y0 zA~WrF1{blUh*ZSF+|<_?@Ls@0j}-;uS*UB#*IU|325NV6ZRD#;I4WJ&Ml}=q#;o+`#w|Bc~&G#hRCB$gRDm_lw#ov2ZTa@|RO)7P?z8jHbXqeo^;uo|F! zWhuy9NvQa-M1OT}QUMJa$>^a`GFJ##qjy3@NtnW}d6`>2(^tfW^u=f8AxrhD@cb&Q zK08CGAeA2!bXBvvz%m+pO@%_2{Xd8>pn&_IqudM_ISw6a-byHDVGZFr*aj&>_PiC= zH?X5U(*r7sqkIC9#6fAHooN@qRk8;^G!w>BAu}{m(=UB~?(wgZ&+~`6?3v#$$Znhi zF1vo`kOgGv=M6L^;rx;TR@?QI!X0m^wIG-*i_K*F8TLV>Im}^in!|O{mr*EAwOewUMyUcNNQ{`uJ6%)Sve&P z1nCkKMee6$u~3`ZlbwPYz~};gtop8&(jx8z z+%M=%J}VYGB!`D_kTwE49cx!2-##$xbg6@- z@jJ$alN{ALP5prl(nQ%;t31dV3x-QCGWSzrvm}JxQUW#i@(eU{WHe-c=I@?(!X9Rr>r zm2hW#%O0ZrsWM7a<320WLu&Dmf#wJ(r5i{FuB8Y%ClF>V6_ z*onjOy`^7Q1^+9vF;mSy7Q=cZ#Yx=w;BLndO#{_Edt&qEATn&-DY?!yZ!=%~Vkd)x zr-$fjUCMdO06NM^U9tJB{kf8lLQnZ+DH}hmnQZ?a_-rz8ZK!7T`Lcn(vF0_YC4O-y zULaOvr&fy_eU&oi6C4v6IS=&YtIM9VQ~c^`BX9ZJTQv66*Fj05|A2u{*^DOta_fVv z;W>OP1U0YBL93GRchYwe$5XdLmMY0v)H}ONW4kqMg@2(SlkqG}b?38Yyx%%vo(4XCNTcZSLNno<(Fvy+Cq#CnM2jexOxrMBGlG z`a}X&&ZLrdgFK){ff!08gb%JiBtvBbQmb{1O+gzSaL{h{5P}ePtM8UVk^!z+v6XkV z!K#TH049C%8P_dP<;;7r`U7jrvzd3G}=yu1J4D-AAKO8u!Hmj_>xv+<+ ztg}u%eNK5<{0_8W!92s2>hiCXpxL1sJiQYwZ0C~AqbgaZ>S3d6C%*Ih`n*WaJAjK*lT|(9C3(& z5D2O7M5+gn`U&jOELDhOR#FN#p@04VhL~Wr(1%j!HBR`N$TD1UoRbn&gXcXpwA|c0^5%I(Zx6 z&{VL0B@pJ_o(nhkZ$Y!b(|SmM8LEwellDkK3Cd4o5qPoRx!8ye)sgG%j(eN6GNz=BAaw*Q@T19J0cK+q-t zP8C1?ZX%me&_c{~f`3~fw9&&{GnE04nA@cIDfR(@fg*~Ps`tCiep{LtS(JA%^Y{H9 z{No#6xOBvp$*#M|f2B6;6)zGJvEnEXx)N^G2Xq+L}d0uiwk2;fQ_; z1SLTjxGB1Ev1H(X5F{FTwPs+?Z9vrd5I}&0r@<`)DWSl0H-akzl9$fhKk(y8o@Ky8 zL)NZw{WW~~h+snWnPw`7{dpj^J3rq*j46tG?_8`*NgYKngA!SFJ?C|DM{<89>a4EHuQ;04d=GGX8s@nDd&B zl-T@Ed8z0GQB57Lma}w8En~tk6M{CZZ{URC2TeMNB7W;rW)22xI1`%xJLDfa#i45Q zdy#+WIFsUNu;x#HkT4f$uIzira94kjx_;8QKlhV-R?U6(ma@9m}JO>n;eT7=$4r9FPsNr+Xpkh7nv>ny5v9O9IW-IfBJ=+7gd(b+L+l z@QZbRX&sIRG&vYBcMQqBy`@?(5aOi%sAXGQx7N;@QG@7SCuZ2EX6AlE0R3V%)d86Z zib`Rn#rSCV5Jul7|C@}X#K3b+h91F@2XCC!~%+)-@hu!GJ~xAJ5M*cEpK~+DD8N?1V4D z^=*{UV#Gc?W3WdaZ13Sgtp)Ib4-!Jrnp-lAc!h8H}DqS-E^=H{}XmtsV&-4_v z^SigLFNg;HEYbh8g?VXJs++6!w$Ai0bf!*~U+luUFhje>G{7N?BlkY>*>S*c;(Y(! zeC%|-ZNS}1=Q&pa-qR`q5WxrHZ`dpv>)J{b8!vw8P27}PSflI0yc44_fHt&}WiJlE zQ#?fDOta>9te?7Gt<`A(lF*Dr-7B#Na-@Xiwzyd+KIaOyK9h4$U4ZL+cAF zKX92#v3M+z1lTFBn7b4r*Pn_J+cc+&l{^&B?-+hMRuAM~ybORhGRQCV< zaSHe-AvTa%GgOFaM;ox+$Dzo6e#2HnA(=99ZvARtI;%ys!H3@INW)0KInC_@{mQz@ z)Hl?v#8-Qk>*CApSmx9mTmT zy1PVw@8GAwJ+>hm6kst*Mz~?{7DgbCZMqO}e&7{Cf}R=dw{qfnE$2A}+b3EG8w~23 ztUq#L$2AbAFk}i-E!#W@ze^p&nw3c1bmVPY+oD;U_GaoMwz<5sw%(%#xvbZsGRvG9 zs^`9V5Z}2m)23k=)|@&pPiA|`+19|ciX?sZ@b12cs=`3}^hOuzk3SnTL{HqFKe=T6 z=*MXGygi$);p~H6Re?s18hjh!WMn@5%iIv#_f6HBo_B=fkq-?8MN&dj z>H(87hg6(~yF5ypbo7%ti6?&XBP;AeGuJMsEPvmV6K5~0dBH1sOuwIw31W#Y z;N+z_FPC0R>#1L5JGawXRV{&DD+f332y~G zD15Un#8T^honYiiEfq!NFAe6OxaX6r<+Coq*K(!c=3d-WO|?d-GYx@#b4Kg_L3}YFE6NV|a8LHFt?jO+*?><&4wDVjhlNvOS};)EFy-g|BQYH!P|zQl#J2arwiQm#QAD@k`foMTUy*?$%c&#yyI@ zrOq*2^>unI&FP5ad5*Pe3u)&;j)0ftZTwkY|Dmxy#yrZdHTGN;@jas`k{{8My zq1P(7H8)iL?q2=CgZ*4hYJOfs|HopFTO=fyz56P%xTe)Rk*1)jS2P_`#nk$EGgsU| zO!T3tVZx{)PUGpZQ&BRzo_PmaN64i1rDjK>?ECle4|%KgD9M|Y?kN#~LyC#<35qCI z^ZBVjf!L*7EygDCeYy#LdzB~K7G0f*I%v|@$fHAmUZHTi-CkaY7pI(oJT*F$Lc~nG zXb#I#Vq#m?T$?ztB7I}DlKfq^cPnx{BPeHo59N{}NvXO)A$#^Vz0Mb|68*x%BLY+E z_DBDt``7&cp0aIODU-GrKpv<+pdF;HR9KirmUZLvuZ|m?5$qaYd6Ev{*}PUB`Yh&!pjsZ9BZ&0 zYnAC3Wh(P57F9zry@R>*zB2%;sLeMp%c}5|fx;al;q~O)u+LmPQm@`|OsOiOdPjPn zX<${QHo?1@ln``|HJAT(zg}U5Ot>X`sCE(M4~3lBitzxJ?A`Y4PYhYvb<1k#Ov#fR znyzc>txnb@m>3C&)&z{X)GpzRB3usjNcgHUn_m!VFBmpL3?MvHEMZ09ylJak#bNegX zs=20TE8dSUJ;|EG&|C_>e)6)-IH`v|;#J?bwl)gg5)L|nK_O|aclQRCKB08PBJ;bh zuyPa`;-ObWu9=*F@U3;J^-(*S{4Ji|Ah#k)y9UA@o$NSz6|oLO9U`0bY%#gl;i632 zPo2j}4Z5L?rWn39h1s5*VZ3t47)$RtQ*KeQ120m>7hCQoMZ&EpdyiIhjz>}oEeWqO z-C#l}M;qro%|Ug7m&xBeHN0}8Dw{@L1bzO)(a~w&bIrE=%7wrfH*9!Y-Q2QM3nJE5 zl;0O$KGzuc_Fek}bK&RecjWuU`1qETuXN;~&mVKU_GdhXF(tmx&Ke&>N$=hvk1oK( z&8c~Y)m&wl$gRhzwI{&XS@*NxlaY(5Hzmxv;|cS+GILPl9dT4}e%ff-X_d?$9UQ2* z{e=Mk(J#UTQMJ}zgF+j{XC0!l}i#>>L@_c%F7;JJmbwCjmNa@)&g;{<|dlsA+^umDudg{DKy&SFwPC zXQMLSc&&VCL0WT>xOXBVBEYb}L$HGrEG&o)sPOBB>*pJV3L%pT(XkL?Lo5|~?%ux-fIfeb1r;}xuj=>qcwb(%1T=0PoP z>SGfXxL2{?`Q4`q3Oc!l_pe{yv{y(vxoInk9=p@B`0dEMPRDX{oU(CiYdr<5UqyD0 z9$=h}yDOs7y)xNXUSCz1;u?M>gIS92A{=&@>i9L(t_L*-rlmxxOHWkRo?K#7yY%2(<7-q z0ym6su60+ZjZobK-Oe|al8%jq&6Jw+0DXG~PTdg12+ywC(9U;DzmiHB@BnXqXJ8-} zD=tq88Ht^`?ju{%dt&`3ntl_ekr59+aM5WAL~y4+e>S4nrTQyLTCitkQwCRXPeu6q zT+6EKv-?D~?+rJ&53Cn(GLOsI+_P1&|7c%$MGy<$JZng>TNTkCH(KJ`N>__RFbP3# z=C5RA*?h4vzd)f%_(Or=hl1422~I2s*J(bWHKyv~%TD^^=jv+AX9Pc%(9JIHN_)1B zIEP9N(XNAE5+T4Bl7ebOLBSIJ?p+gTjST$YYkr2heJd2{w@$?6m9j;xS)go^#?=KxBwEP9NC zb{#ARvVmY6Damg^x?J~%%4CQkXc@#H!nnfbmD_1JM*_T|czjvoeGZ_RpyWWR)T8YKvo zUVe6f%Y=Rr@+$d~<|l1pQqT7V;^#P~FF#QvUtCwSDEPws;)UDXu;BKfED5hZ!)jKO zF5}gn<)q`5mG{X`z?E?GViQiisk|tJ>1HbqIb8Yme_3Gy>Rz6K+l;($tNQ! zzsfJXraxZE<{nslzw?9qD4;F_gTQrb9W&)aQ+|3=gYYT51g$o7C)E%QXxZhw_BcE zH9yl^^f;I)%^-zkFt3|e@kW=Lqb&%is6NQ)=x{}8Ip-^@5gkSV!+{HX;v@0-K9Q1ZYam zT4vwYw{j`lKdexd=*sE)$v}C%bB{6OjlO=zLyzmvLphkL9Ad~Dh0e#lW8(3tqVx)o z4<}1Pqh1HTOn%^?FsZ2;X(`Y8fl%&$QT5*8T>kC>c1D>u{O@@xXvAYJSm^7lIbAza(y<#USKf^cNpE6Uajad|5PQ+*J=f1SON|r)k zx=*01lOP#9JLeGc&dt@zY(QUnwW)YD03;dLeTaFIO&8=^<6iWq;hN}*X*~2=F6GAa zy`@HLNh8usBB9P-AUqT(JsXwiAW^@F@3)d?{>o+4NzT_)UO453^lV&YPJ(XxcQXT5 zS1Y}XX8mP8rS|*+@5DH)GDO%|v0lO_Ldm0$^T14Vxo`$REpHIG*^XxNb!9dHyp;zI zfQa%BnDHuDjsAxt4#Hsqz90wuyH+_U{6h~AT3fB-{SB5YJ>7p!x-T?3BC(RIupLpD z_wN857QwAnN)ljt!Tik(d2xKB>(#&jgv`jxA2@jMRQo!@7X=9$%!)==u*L$EP{jtF z%vaAN*R>mxP0U`|`M8nMNlei=R@yr%E6lS^S3neq^z);|%Y@bYO49kmhP-@4 zKNgM0=~*ogE=1YZ1`-mf|{F2>Q8K-8`{zVMzAOVQ@d&8(}FfMn--d%?)eyAr3y0; zezac`^5c-Oxbi7oJgPSt$FIVctv$ANz;Z#P8}s2!-zk5-ygedA&-a zFLOJ;s^f9a8FhlLw7e6D8y%p-3?Q<{XZi}VHXLD=+r z78v!eflVQj)>FK2ngdB=A4<}+dXY-E9I)hZ_w?Hk@=cI&0zRVaU>(la1*guw@cvKl z`@!IqCbg9p@$UR*b3yak(!2A3w}Pd`oPVW{hqK1(BOHbO`z!4!|6HoT<&fkZD1=9& zEuN*{MYfZGvJM?I>B>Au_@cV8d3>W{nh`ZSI)*R8%#}Wt=y9QiT0LObD|Cn&Y_3KP8%k1=Dva0aI3wP9-y zzZHPpRwS>EdGAJ_dH^~qRSr680~AI4^ex}L;#I)X#ZA$RqnD zkObK$g&!Qf$O4)gb=Ozy`7fPFHlKFLcZv=W1A$r;aXJSF0K|XF{>lkW$k&J~?~$LR zw^sr2OhcHRkqzPSn^b)oUEWr@=Hc_y=4lJ7d#Z zTH_{o6jP#@?M|7`14m`VJxpIFTAK3m(G9zi$1@8~6%JktI<#?UAy}~-sps&lZoTcQ z%BlX5XY*lgW-W&o3zO||uz!@g@gmP$N|(FWge;|-@wljo*i4+E2t$?I#rlFDZ4H;y zTCOV|we=~<$%|=7{EVAdw>Nn5uJ@W@-iRESGUdHj*>k}+Wv@=5W(9qhvU4v`s)t;C zHTVNs9TbJ^3g-6CWhL%ebS5mq3dI=_ld&=T$y3qI%{#V^jzaU<&)ynTBPJW#2!w&+-oD20b_?Neg;y7>7NqF`n!d#yq@X-OC97uVbo`a+v z1KE6l80%)&9XN{;Rd!E{Iq$+|HWAqcB8DJe6Yq39y+zYt0&7=96tjl3o)$czu-Rxh zZid*Xw&qXQ&)2UbhJ2|@VD(^)@Fv6(K5ZiJ0+Le8l)_n^52g_$AE3J}4_e~v`Ug*u zA+o%-oCWPD4W3~mcU*l(+y*9y-`bbuS!naP3^m%`O*$e<85$YW&iZDKsLsJQA@S)) zU8@F~OfrThe52#F1P_MP-@bd^ICS$Fp6Mr-GRG|A1!D&DO$RSeHmYV>>*ozF)#L$! zX$LWZj@tK{iAW}|hEI(AN;9Re+AyeQYaRac?K%Eq?Dpk~#RWp?NUq&y09vn)2p6S3 zse2~%i(=O~moeJRoMe78&kfIbU&r)SFHPxeszo}H?L~eJLCow~{u2Lap1sUX&j?{^ zVlmX%j$}lTf-ZgG_3s%{m1ObZiGC9If{coq-O*s~8@6oBoi(nj6MnejJCkLdCC}dQ zreuW$p!?AEdoQL5=Iv&t=@Y%5ag(b5`ex#l(v;*?dVp~oVA~4E)I9h0hu95=P2CmEep6c89r>HYdjz1uLRSaTQc!d|6By ziWn=FYE=w7gNv^{34fFop$-|?&8dWMKGcn0?hm>TD+`uLQ@%RWOXp)^+vL9-=c5fp zc3#&QibPTUsDA1gc+RnLS7S<1&BWs>rWLuikgyt!0~Q_t;oZ_``#)QX}t*^G0EYQ5)Ous+|~DE7C{wb7sI z%PpQv65qy0FOuBS<({0Uj23x_vdyG5#-0_ly!Jpf+uS|BP?J(DY$EY8a@m$OOjgyg z64B_X+%O;P{eF43Aa}6$!)1pCp6A4=>BdEFPV$oa7ejb{6Y()3RA#pqQv#<^BG2CJ zJ!pV6v~JKMaII_FKBFL)_3EzRU;Yx`4iei6+dng(6ND-cYS!rk@q6EZR6+r2!@G>5 zzmaMP5jl+;KM`}M2T<05y#xU$V9L}3Cl091dH!>z1x12(&*F~E|+^BGLTHia2IL(VTMUF0TTUd1|IGxSBuF5gJauRt*zk6`U zU>=Q>tbl6mvtm{VaHiu`6jjd!fir&%fy5!XALEdE* z;l>QN__crS&jr-~*Z%x^M!Q?^E^OeL>Gr+3gL{|C%H5pQBpWs!9!H80v@KxK!8Q4g zTH(!dZ^osv)=4~4T?F?e_`L2Dw^{RC-E2<0ZR*{)7=-@B)`y{m(nPpEI;`@5ql zA*XIgtxG}EW2Rq$?3X&}Y|}{FO@<7@Q4by+6IcCB_Ks(b-!;9jM$OYqyt>F|uK6Y; zB58R2J1vkddxCY^riGf3#d%^YS>j|p=-<+Hb1WCTOji4wKp z?i=_5Sv+0AHM)5WMyPRY>F&qfHf=4rnju>38R}7uv(U^t!%ALp!~-e}%w&mWAM?R{ z1IcoM@VQ^%OSLegMK|)3ysET1v-I$fgWUGlgqBeP1UFxQ61d5{U%FyG^tV=)4@NFO-vx1SD$YJb8E7OH81;1!>o}q zGtYC5#V~dclFf;lI?DDkhqDg*W?M=zmG`VasPH1cn&EO2_vH)0N(ldtNolZT+G5;+ zh-F=m%AfU$I(9n(*XLyvvGq(EkzU@04C}A>yR5YdTwSk>%wJ-DQP7IpmR&@u`b;`S z-CyBSH0{J2qp_x)>6d!!4?m8{7_hlUi=;7-#$VbMkE|K*&1U{1iEevAm`eGZ0rn%H zd@O+zt$WlEqVoe?OF?D@C%)C6-}8UTfIWNK*M6}gZa3ggI5gu!2Vp;N4u6h}jO08f za}@i0etteYr@V#)t;KRG{fGR>#AHJ=F;e&asF^)xWl>t95Z>iB&jqHjkB)sck*+jy z!WB#l<47&X9#8$OxKmFj)*~syuQtR_$ykF@SY_;jNqIR-h1i8gF`=S%$r3VZ} zXeY3AEW{{Wep9UeO077&jw76Rn&V3~ety7qo0-5illwjwmJm*aj3K0ZfSb`ELy6IO zqfMOWN!5A)mt$H8cL`&p^_7|2%<|;77}`vf(`R;Vg6-oAYeVW3RQ^T-yeD@h74Pg% zj?~p`Q#nbb*WoJ5ORq!$k`!HJ&(?sYIr=hLx`e-o)|9szHcafWVPekLz|26!4(esV z;;nsMVZB#HJV8ur>=X0T)8Iz2LR>tgAHx(hf3A06v*Y7-LE+-ZHFKrqf|>hAD_i%v zenF7kj~C0}e^SAkjE9CqMu1VumiyaBcIg$xdco&isi0XIJ^cCMbLOE>&?$&Ig89-u zME%3-gp%LVi)!}4#7o!*L8iCyj$U z1KquDN5Rd0Jl~ag8;_sE@anb=%!q()`zUYYAKHn6=-|&+`v+ z0d1!T37fxo;^W~8MGLKQ2S0?1ID`&jm^lSKAL`{0IN@!W12e;?FYGkRYh&ToNq(p0 z43?VsSKg9t?#P9a&bEJ6w!(e@9n3Sw;uqL zCRaT97f|}zR}-5C!+b9dCC@lP@wtMCRQ7rq`s zEN4VhWISzj7>v5M5Elwb9WQ}pSmHNXgQsGiwK*R-DX-4U>jv8cyAa)o8W!q2u}Zms z0M^$_Y0#m!RQ4Jw3-t-Kj{N+sm=(J8Ww1-H%S^DrCG{Pu0PnbbSBIk+m2EUjU^LE~ zaa$W3>k={U$(DP5P`u@nFwBR=nYE{8`(k;uJ$kOe(%r`9D~FbtF0gkD|v*U36@)O^1|8h`2(DD3q%8NGdJs9P6x`L zNmU6MhlITTmD2_!%#?rWmO%F^?(XhZ-ULUz@;u0!>p%Pn`3%eRK1jH_0e6%JyS~?7 z#{=t8j}o7v5fCMH*L=+LosdFcPy$bI0i#1~vfhJ-5-O;jzx&i@ehCcF?#y@c{y zoHn)a?^Y<$er|Z^p_6W={W`8rE+$W8!*oYH{cW}n$4w;z=6w~sk1Xd(YkGFl&Y&Xm zc`ZtJsVK{!k2js2B>igGb^i6Rcm z>SS?#?nM<6c#jzTJ#T3-@j7{TzW43}tRt!`KWrl0@IS{%e?aJg7+N8O@%aGV4zVR& z5x_9?!O%ztB=;`Q6`X!RLtj`pw6VboW{IQc#Gt77PT_|S*ShnYT;CHBT85xaVOSp| zkK`B%60PL*1K13!o!2WFv6?z_ZY&O(vqt!v>F#`10_|?>tR=eFvbdBy16tLWss*3B zIo%I0&!Ig}5`I@u{WIpxCEx#wUXnOPHR^iisZk$W6fnvuB7Do;ojG` zC5Gq~YRU`}C})PXu8SC^+}5T0c6vlJntV^yaR+KNwoM84-dV@m3g?jYK0^TfWVk7zk}1_g~59^H?`_}8Kz?eGeA=|_F00#iE(QjG2bB{Uaof} zfTCTjWIel-R9Vx8ztntNp!LkK_8t{*_&%G>A(DI~S(@6?Pq$|eF~LV1kpDV}-n9Cj ziW>d3M)DvQ*$!c2N1)k2ytuK2x3cAu(zhYd+XDC%NeZcHhy^$rG-=Zy0)*7gF1ryf zpfy8a6(Tb`vsIX3C3h2S8qrIU7CrJsPELlc9Kai>+7pepotoM!rLD~0(`oZiF@g{EHZzH&V+(TlqG@g{0 zh*)F~4G+hb?G+#YBLg~A7`!5Oa1zeOxvl9APM1h@0sP1iXlwp61ROPxuq#?xdGMBw zB>jcQ!}VlmoYKl=B90!G)VZTJ0oNWrV-tz2pmS0lb5M}KtxXQg>~gnYc3++s_{ML7 zska0{WJYN)p6>KfqjI=X^n+NzsPT9scMwKZPWILdHz${F|Hn2S<0SYg!_SJ>FWNkF zGFYcU)urLz?hkSWb{eo@BP`t-ProUga<62?l`mik+t<@Rr&xyvSH)-OQ`94evYmTX zutiEb-8M`;mr)V5^eO8Oz8J6Bg`)Cp)jLmFR)`emVSBA$G}@a?Zt8(;DjPA7A;mgN zsW+lklKLt7w(r?L2IXXBihp(WycNJm;Hr+K9R4@^6uNtWSGD^D@vm+o_UXXSPr80C z2?!x#sSVO5R`LXVSh_<}g?*uKz~?VtNF08B%xnM;mr{HOD-iu4e%!5RAh_&Bk}W4k zJ9q0x|1xG93eRrItK8@l{0jbl27JdM>8rmXx`df91?I4y83wF zLou{%BbAPqJ9!@@o_-?jiZ_nkZj{MgXu_qR<(i z^8MYZ{bzR*7MbIAEIwb$nP;jarZ4@IXbyXbEr~FnX}s=PgJUZS1#e^z#G=5B0?UA} z;Eibyp8wd;OComV@X7oO0Ys)UqgpG5zR(c&_si_MJD@&u{v0v}yaNopmGSZM*45RS znf0KzXj4l|=JWT~w^5Zfz2FjP>C;}!dHl1vd++3IAIZ1&hHw0Uz)|0T2uuTcGXAn5 zaEkw$k3un|0lVI+T92Z;|0WR!NM<{4K5(P`t~KwE*qVSF`vQpDN2t0l{>u{vgc?#K z!_~MY=8pVjziJYzifFB>^76Z8<&WEi$Vb%l5!>a2eZ=~GPfxXsPLkHDWR8?wDMUdg zMF%`iyoTK!*jFBuLM1^c93X)asH&<$5~e`G1sOq*KjLTE z-2E7$p5Q})4$jWThEcqh3M9S2s&0U%g@kP*^#AXj`#$+J`W5M495Fw>Gtc>W`V;!* zP!a&IE^bfv;C-lf*o^$0Y98>7`3Jryp(Ed^D?4!nP zAUyMbx#U~s3GGO_8oXHGM?D&H*Xt=KJ}a80FUei;8@^g0)m1J?W9@0BD9&T6x={Uo z#+3?T?gX&fiL^R5|IV!cEXC5CxnVRXq4icp`9)BsM1bOFW_oNIOJMwMaI5v(FN@|h z;K!NY^i`DTt5U%Gy2aL!P-(`E_q*hciER3?3hvD0bgwb+g_w&2jD+H6b(Vwl8++-z z&#i53Z{<>P_?ZUC+`)Tf+;qw5AX1RPRIM$`0Il)5zWFSTudk-u-kQdV5Yc%)ZkU7c zBY+fCBBt1YVLP^l|9WTyzMITIFzhVtYKrV_wW;XHP*h$sQdpnaS_5(O%gY5;9^#K`3 zyBXG@rit&|Mn9Zp&nxOuzKn&m*3G<)S5Ai(xFQcT5+YM7Be~d;)lu6!nor8E(mqv_ zaSw057dy%q`Kd%HS)%gsTYq_s@$?}-uj(AUg>C_ue;m$a@CWO8PXn{5L{DpmXwj+< z<;C|S3&FRb?_2yOU93LlLuc)!yfLY0<(smFw zWN9#)@ReB(wTfMEw9YeQ&vV6=SS!zYyZtWhf0zl5QT^z!8xOD}^(5*X9G>T8$Dsa_ zNb`3iUE0LoJ|b~`wd`Qsxt5D$6oG}UBxJ0B@s+dj5HCM71i>bNUy?2C#Sk$EN-m>rs-t3Pn&(3gfwya6B>3qdEbUXJ;-T1`B2cp||F`GP=jJ|im^@aFg zhZ&Hq)Y)PJbw_6Y5bbZu;N#9+N4SD+fIU2tp@$?%AW;JlQiZ=?>3Zq!QVdcS2Z$t= zU;g~a_m8lh{2W{rz-4WhbaQ%I@1N~Bbcv8byS=~vvvU>b&U~1nrE*TbTEp9IVXdva zPFoEoR^m1^p`KZ`V-9-q*3d?Da&gKVZ6B~*dHI8`*H6!IVD|&>I7yZFu~$wV&}4;p z4M^aENmkmS497;HFBSblc*{nOO$5N|Tw%Ifh#HinP?K;pkl*3k!E4|jsJ_58`|J}#CHEZ(R z)p&7hiMkj+Hk=^-=FlB_^Be+Dl(i8seB!5xPRtl12TEbIDB~faD9giV8G)>CD5pn= zOHvt#-fjM#rI_G7OV|t)czj|xKSTRU4L7W-3t-{HwI>n8y<=m9|6+1<_0T;+LJ)nY zC?XC7hfw5zK}V0hB?DmexIoA^0!T9Lsu=>$I`(A$#{^VHpSewvd`9t5Z1-poMhIA% zzXvT4_b)^$=-KtlX%~F6y5@anG$q!+=Ehy^8zcBzgB|i;xfHSFd$jo>buMp&C4~42 zgYNS@{0rFU5h2#Vp512w$R6~tGoSLxKhkZLWU{Lfw|mpbv#-@vRe-0pwzfiuI0PoM zjdgW($vwzKp%fImUIZNH_|ijN{l`fL{_zhBHVw-Q8rAS5E>!c;3#z` zvBfw%4WVRv0AwwczIY~8c*eLH^9(jjzc0Tnr;V^GabsJ~7rC0!w58PMxNh`jt9M?n zUA9RwjniOCUzX}W#^U32aEvjm;~Tjzo70D!McmMNmfCxl4ZALtqEM==X%KE`CO#@x z`s@}ig(gjWWfYZBLQf>oSmulz!&4avV>TaS%=tId^xyEDx&Hz{ei&1SfV7To%3u>Z zyl`~+jejxIIYpx+MJs?gTU9JLcgL75&y;d3^1$4`X5`X!`AGXnQW|>teu(WMmi?gv z33!-D+YgP2I;vz3toad&B9e8EWMaTWg<}x*VuX#^?IM!v3i%JK@Hlr!O)6S6WtSK4 zVQBzK1geKTK^erXef4c zTTfx8eoCwX25&}xJXgc>4zEK*!Rbu~CCQ14>qdSwcK9%W@tE4;`YJM~y><%NW_0 zL$-g?U(6BCt3avw`Lm;Q6@4xYagEdqm0MFEeR)JHM4LG)OO2Irn8OHxud3ki- zjKpRxOPNWi7)J$+1)ZUT4CB478g2U?F&y`7`&63x1$;LVUvEipRmCu! zMBMjN`Lm#TLROvMK>66V=?GsE4EYu>VZiGGN+xyiZ2)7AV;&q|K#cyonzKJJ;grRL zhQKWMz>Tem_ky1#{D^ zjg2bY*mu!v+Kt9V>CQ|eVeO)?%8sr$pE4^-6I}qu_l0P!ftbjj^%_O2>%q!*$7ThS zXD;{BMSNqB^=O@WiZLNX`Yz=XUl;#kzGW2m=7ne^B!NoTVtB)!=vTb=$7{xqE26ou z3=EZpQ$lSR%XO9v`*I3W+=5(}+a9 z0lxb8wsWSzVP|JY0>|M?fL?;6y@4&o&mjx&k$^K|2qAj>Egmq3gz*N!=l$1N2JE?> zKpZHzsF8qR*JI=eBK8m$axeg7N%n`hg>D~@m~)sQT51GKXFfi*PR#Xe!vkL$>}Djn zc?h^mA5CyVeP3b32m>_jUv3%zR02t{6kU#)+Q3ffeC z82_S(2rKuV+Tz?+d5){$6}epFao#cei#i5Wu7=o*MUG-FZB9)>wPs>7ip(OZSMJK9 zZ#2=Od+=|(^VR&qLfE5gaaVn&8F&Ang4TOk&##xk<8UsxvP-6!v7(q0e9JbYI-X~j z#o~Rs*mvv9Eo-1;?Uc5krHW?_yxyhe@*c~Q=s}lcJ+_N~=Dl-=sXA$$QgBf`6eD8<*70juBl!BuwX4d=HIcdC%9pZl1^M|>BEDPuW37|-6E`DC z>BB-v(ECUxq$1yc{Y`X?k)O;)mGdD|(KlS}XT@|Hy=;ak$yu)QQEuEl-AW8KaqR3A z2yPOxAG8_E68d_tv%IF}hQGgm!owU}nuYc|0zdax2x9ueP$+H#@X2p!S-26nu{`BI zz#h&Rdri=1W0*$XNmEnvMiayBt3ax5;xP?DZtXaZf>_fDnm#0o3jjGu`xCQz2esuh zg!bloduA_S9f>JJ)XwD}FDG57?fT7K)A9XPW3%SM1#$;+EO}C;%s9VcdJVSGC{h=9 zcSR2mJ~wywqMbNHH)Unz@~W!-uWNlzv8q~c=u+aMmA<7$H~*u-pUqD4Q1E`pWlCEg z^d5A>3>|~wqB^-3kJ3Wf>TZ^@1~w^WrKqM-UOp0Pl_mGAr}rVGkHmX9-196eb(5J@ zoW8Lpd&L%Bmv~gTFR|jh?V#_t1EGdOi|AdmoPt~8%IrOtMo1OKi()vBz&`g@fTt-6UqEDZ}Fp&uSm7qBP zHhGtMxN()G-dnRbLm~XbSdF^y^KOvh~B2AV|{X=!MW|6NQ5 z9S5>))<{>ikm(TLo zaH9cIdvJX>$3_;qB)9HD{)&C(1EhS%lCk`5+(RnC4cNbZ>Hc#KPxt~Z>o6-WycN76 z3-rPetli5zR)AIR*-S@8MO9v2zI{NTV7-8 zFt4=Dy;k~hAB*g980SPuNT@1tdoyMGkP0Cob0L|M33k8m&fH^mON=oSm9efiTjF%< zj+TD!K8LwKn}6yzsV)axQ8JE8lVQP9nR+a|G;ux`X^6cOzwY8sThjq!40ByDEZsk9ZL|AO1b;Mwj`_a9l$j&Fkr`?rtwSKuT)Gz|f}iKDqD$c9R{he7OGHu`&jZ&(DJ$(hiy#_@S)tmbsX!pFw35m>!Wah_y)OHq^>rhgK`))^H z_9NcctHZhA(}Ihi;+sH<))EUwx%Wp27M>9@2zkYk2MvY9Z=Xzh*63rs&Of)9&fDi> z4&0sXt=JpLvn)wE^r9xx z`gR;Nt#E!nlhn_oYOzrELKjRxJ8dH=EsDUz-L~3}CLz?*$*X96GEBQI;-e+u?Ue3b z8av|dnu9YVYLCMFD7^)BZmLL8$ZuOmBU>2z(W6I55%uu!NEj}8@q((lx_Y6}mEzT_ zS2~7orOgH|4lZ zFXLjt5uBTwi`?Og5C8t%$H2(QZaA4KklGfXsrSz#4Qb)Q@VF~sR#@3*LV?Whzg=>1 zpMU+@^`$83MLXfa_kNt5+ZhOt)wI z+y#G$Ef!tf_>Xcox)S#fj(LsG+i_hslRp(}29-QNk()`;6NB^hWu~0CfN`wp_S>oM zGZ-sZzDKCmW5jHRvMxFImU}QtjHSoQ=%;$n3!@9l0ba($=k25=H^k zMRXmb?Mk#O+WYryaiMEb{Tp@gj8C2}2z(Doyy$Fp)GfA8sx6l*;9bh9#wj465*R3@ zF{5N@$$+Y?)U^iE!{8-4y4MQ}u@L5MfBba~Ifj9Ofi3Y=a16-`p+rjWe}||uTx5x~ zxO4_S6Z(tR{DE-3DDe4jcOK3~DvWp8W&d_~cz4qJ$5lzMy2L)?B}$=$-YyK*m;+|AMU z7;yY7Lseom8Hj0>vAHey*}f~zK5V}@_Q$D1BwRVrzHdC&P~MdJ@f7O8oS zoB>ZJ9(+p8y-RoS9?KBqS0~%vxHCW2IBBIo=2Umfw8el4?J}-Xu;z~$t6bIGj##hu zWxMr|^rM=2@&XE_wguh1>_>Df{6jZ3$WN|wEL>M8{xjn!I>|U9OkYoki{2+_ljGF? z(!Rvq(YinTIhj(M?I1arTIx07u$-*jzu}p6|7A$`QK0v*Mr4M4!=j0Tx0CTi4G+g- zXEH&19YYR-Sv*|y?)`|G6I&ll1+Of0%#amKHN7WXT9Ts17f&*7+2wEPOw#7i(|t@$jOppL_S9YW8;mySyx9_7inPNv@G6G z%I8nlXT@cvOSp3V^7>HH&_w>#rl7 z)NX={yu*uKqpx0l;Md;B%f`?16_t<};+>2I1I{+Z?A_=CR?AMy}~kT%JE*zeyMoqnve)c7a67 z79ZUOmGFTXI|>{1C{>wiW1@7L`;fCOyS|G>`^#`ZN|NO%2iHMw*#zIQdx)N>3`q1UAeyd zyK8vA$}l`O`nQRlVDe<@=Gk`4oOQkrQB=3;g66r(`)<8Bh>&pZQyzxY~ z@g5rb90KBT>>jDpHtM#^Xw>YHBWoXVg2RY8Q*ku?*Jr}PKD^YPX+JMmW^eVc`_w?6YSMt)v*Rye_YFq7V&BG%UeEhZJS~{jYS?OWfNnc<`Y-#cvZ_ zRV?l%yXeiq-d+8w>ptb{!jWh0>@RHXzS{BhkP6{$+fJiO<*^whV`_E;rAV&3t7Ot@ z^G+n(EmYVo;Toi&RL0Y4!Iw4nUOiM8w<)B*F~KA;mN4Bw=N=N9C0#JYc`p?Ei^s*9 zgcJ=m-;-F8Y}TGnoS`$iXcL;pm{v&7io0EUMVjR0`j=2{_bCGo(b?>yEPZCbE=zuD z=r~?A40c{#tBoty+x$iDES*15*KV$&qGFQw>D%4nIRY$VMn#uLDA zqN1XtlCl|NSwC5Qa(QCYXP+bG>y*Gqfp(GKB3Aj~#b+Iz0}rW(s4h(8Nr_Y+3-|EU zc7*fPv@qA`P7P37Cp^3P*y-#}RY?if(;JU^S385?$iuGX{dKi~a1KWZ86 z{`Tm0NvCG7)&H*2qtuT9{*@F8PoVj+xLm>^F&Q3?ft+S2(fp5jVsr<|xq7jRu?^xb z_p_z!RUj1xTLhNR=VYwiZ#B5ra*iv z+E$4uJ|IryDTg%MJ?accS`;e$;~MSi+@0*Yb4Aar?9!dQ#VuAJTgCYaC6iv+C)V*I zY--V?rnuB-GjG;Nj%{erz0lD6F0B0MLz-EaDFbQ)<S`kRoK;{*99xK--x>J1xAYu>!zrn#Z2M!$V3Hd< zSx~B92(*)sCp}ApMHn%x1NtENq&V|B&~J|4PKHFo#KP8DJvV101F5`1qcJ2`5}`e6 zt}QAZ$i3Yamo4HvNEJ7cC4@QpSYYyP)=R`mQ3g`wbVnmeTY$m+4b%~(FJN<^C3E_< z5AO)N{r}eN(ssrf&$}`Ob+`>Dyl+(Li7#+-bL*Iz#)XH6cd(Q3UeMT)mwB7Ik;*Es zn;uE>DW_CE^zGo9AP&>iC$TB$bsJ_vT?3?Sp=0vl=H#A!KeZ98a)}-07Mnk*8=bad zJnfQIuY^f!m^6RHSy_rGKTf{;(%!#Nwvk6Ib1m6X4i^dexKu^OG_T=Kk#dAjK%B}M zPiU(3MNG^BH)T;K<>tLxo`JU>Mm)L7ysKn=Ge$C z>g#72DoMs|GOvAh-nB*wMW?2cr>Cdeyh`VStFQO`>;!IJI{nVd&Q1m-x@L1c7io_Z zXoN6e8VsymRfd9nlg7RUe5kxZ4RwR>=;RFX4MiZ^l9TVh4z`}cuEFV-ro;-R*x(9X zgY)DdQo$Qe5j7h)c%`jQfa~5vWYZQV$GF)50l^Fc9bXlu9IvY&bbX<44Ya?y{fH~4 zXh@RvtEG~^S58&%ws=^F{7D!*+$S`e$vq~DDVB0?vdXA zC@*T|Am_|7k~WU=63?ehmN7c=4Bkn6OAoQO`vkAGEksT1mNr#xZ_QmhgQqvwe(qt(1wm*G`##~nNAxFyQ0Vkfh{g-R6L(V<=U$WOP z-@PP=qik4I&W*KY>s?C9e7a4+>rZbnpYlVlrt`~8g1l#z6kBkC$B$zEsv{%1fplRb zEdd!*u@uP$9u*oxGLFKS2^xc0bVH)HKRABSE0eHG^kBsZOyPC!hxJpspfHJNvAM=a zp1ytQHWRnZ)c%<50%?Z+7m^QS_4UH(85y>Vjjk>(E-#VC4ehdiQ&VdzEpQVSU=s)D zHySD4(DjrlFaRtN?ETj9@f!FW_9F?xFM=UryXD&;-8a|={8)_n10jI@)6Z_&XC~20 zk%Lt7CVQV|JkV5VK$9WfVRd`V8OdNXs6jhh zyqIT*c%%~Ye_;{No{XT8`jJG!9;8GEPlZk3J)4xma})mNpPjqRakwQ8y}yN z4v$vZBzZ1K?RywYZ$hM+QMkyTy$iS*ycZPMv*{CF-?VW0L~cpUKCi(gKk~i6hU|fN zn+2>&6>!T@nQsq7>a@gOYsVf6f3~U;v7RE5-^jh|k-W&3V*UM{#2xuNn6?+Nl2ZnU zS@DqT`lTY2uv8e|E57_1;Y*?5b@xmEwJmc%=1(&ELWCG)MHMq=&y*(dD<;HDx%DUBRauFZDj(nS&g z5f0#^_C1j7Q(|Ldo5o!wgI5P(f=vO>^dK9Bk_6td_q+Q?jt}v3&yl0(~ebPS}4% zFy11;6UKvo;#Ab8;?;ii;19F}|=#C~9nk`-uvMLXynn zS1fr4{LWFV%2^3Q!}Ww}snc5ZCTdl|8N3pjiA_KM;#X(%%o~whl21_N_z%S>sZf>^ zG2)DHYgoozwRw^#tV)iCLnF==SgW);s6hB;#*<+8b53@~o7&m#(o~Z~NiQ5Z!v?Vk zv6d_9_b8O8{Ul(V=(qbaWM_MD92=(am{qWWPb<~Yr6@18g5&$hmpJs}AE`Q#w7;#} zJ~brz|J-8@q1Oo?Q#JXT6;9kbk}2ApB$+0JM;t>3fAVs1u{x$6p|uDyX`n|y3wrg* zu;mAh4Mz;WsfP=D0cAg`RdpR7P&%Vf@w!iM*uE*({d#it4+s`BGdF)aC!#5G3J(N9 zU7=kPbQn%Xm+;PEyrqI6AJRQEIP?RSv;@G1z2XHX?*qY2V0`j-*ALCOH>4TcZch~f z^3(@Bd?e4BhOBG=kz&z`I5|jtH-Z1dYfxJqi9o~~1cfr$pIo{caBCXswjy}c?}Pa)8iE?v5IP!Qz-vCOQ_ITb!%mh5K=BcjL%; zzS+ZMV^><~Up+>j$-8Coq?3iS_s6k@mq<&3ubdGLJ^}Rc%FmNTl)f{Mc|ME9kt8Q5 zeZBV~nQN5HBl=;D`i#=W=eh1q7Z>fe1GHR6S5l8nub$WO&s{!Z6sof=x^_Jzh#=5G z4@S}OI<9~R63>Hzu~O*CX57=x6A!P+GVWve^feN#x!dqC8R_Ua*JcSPuq|UEC{?P1 zWPX$vhM*?qc_?*sMFauy%6q{vG?}h|@_4A&c!u5^yiQD#FO%yTMN+$BMcN-LO~wqr zUh}e9Zt^nye}d6DIC8xFAG~$?9IyF& ze3KokO2$sgXY+Z4NI&yqLYyFrw<6oD&#RNXtquF8cf3)5s!Y)=;zaYBivw zs+f9nDxGm7wF;MlriE=JsLw62mGZ|=r4q$b(!9YSp9DS&k%rfl?UZF)!5jqM8+TN& zY#-pfy)6ivhCeJ#g^3v+WJ*6rYOWU61*HDI^P5Mu)pq*8bd?mB{r9dPw$d_?-X2^P{|EwU&#`*#Xx8bjm^z5D?;r_eK6!AmOu~+tpu&}KSX1e z!WCBUZZZ8ILZk>Gn(|;=i&NY zD=t!>c*fXUeZLifHp$g$yGTT5nGj}H zFGE{JZzm~`s+z~un`YKS!&-@z8GCWPU7_IP6%Kip%FXm7}3<^X+_)s z9Z@?-9kf~j9-{GhG~@LiJGrIVuhYBFMk)A`maS@_(zmqGfWu6dOrS872INox)U7ao zSndxij5;8U-CAj*g6ju5H7c0&*SsW!FH^ufQ_}+ZSYMIL2f5+k9e-6vt93i?-scKr z-tYzp&{?WCtOq$FNZ$&JuEOVczu5tjQP%MYNp42maUrw%G$EIS>}QV_%Q<`qSVD}S z&dJ{y%a^|U0*kDrr6r1*GL+i_UPDWZWUas-qv_J1?>MSjk*;U5OMcK&E{kKLQ&fAP zH{PL{dh~>#^y8E5DF@1nD6e)(IyUFF)cnrJtQ=%ctf`bT)G`r!Wdf?HZ!(^il20@- z`Uvuumb8+R^0p*)-mc)^noRFb^`N;aNvclzHhhsz{>SZYV)k8JOrk;3R+e15I=_On z`tw^EBX{h5`gSF7>F~c-XW{r4Ua6rcr*w`_mK52)&yCeGuAUPiRF0xRea6Cyu|86q zKND<4XObKXMekn4GeyummmGrBbgrji@3H=95D2q;vdW~FGw{4lh4dRk2|@tU6~pBV zBcP(9BIeI8?~=e6fhJ7v=1l@KGc!o&WrxPI2&xpkwK|lkx}-`d!La1_AvUiI6J0;_ z$KCz)@$qxeDOOpul7j;|KVoAJH~ZaAE6IEs1-DAK+sU(Ytu zPVO3-PyRMQ4nIHf*5eNCxVnG|EqO9IaW-&SnXL6c` z^ZB_j3UT#5{})VGu`{%LDGW^U*~a%+C(_?fS$Mhd<4&y2cTOGt?J9^P=)U4eKT*9T zI*X_28~*&q_to$NofBJADu)|(AHxCjF`tmxyl3)!Q3 z(=FWpUhpM6!=joL$o8e8Lf+oqzL%BKo)B;ZMY!Q%`BHB-eU(;nYwh|7RN|STE%!`r%}Y z_lKOjh=+$X;V}K|VA8qHx6DJKt()z$~q?2d--bmM=MBZ@MJ6I0(eWB{9 zW$~1+el0$8kRla&BNsjvB0mT>F{j?mNhXoh-Dc(K9jG!zi!!l(|m4_1t0~8s!8T?2nkjk7w`d?bYCoAcac189rdZfI7ADI*o zC!P)|slyf*L5|=r{@Q#2X5}oo#s+h3Prwb9)By=#8{Ux3uEax6nwK+gPfXOpERCxp zO%uW?)0g)y!iiz&)_RtLHDGDPT(#w|7MXp`C~Z@m{JQYXtaV`mmSKS z0?l%XRdU9cSa>hKn3Po|wCO#v4)1xntehJ9W8V-4?XaZ#L4*8%#P3nF)N3o@a3tyW zGb?x=K@Rz{;06AswvdhFY+<8`={A)#GG}3&5!7+omOdX3_i{WwF8T^CsS-`0V0_14h|QjZZW0gQS7kFEEP=X!tRhixBw@4aU>*_k0D5+#&f zGO|}>W+x+iWF?IwQ8H5Y$cTtUwve5SknZc{obT^<-;c+A9_O!f9_i!re!XAUbv@T* z3)lR_IW&PqRG$7nJ&CovGm+GpKfg?oFKB;!>NpzCEtdLd50V(%3;`S%KpPm&=T*Ax z6l|{a=fgXEN}xA5({zLE@eon$euU!tDK!^MISDPdM=zoqrC=HTxE-Sk2xzoK9W!Ts+2+!RKLt9 zs|g>`(k`Zdj9jveQ-mnJatRWoA?qr@z!;`y90aE~A&zt$9XZBNaP7_P~f?@_$IWQ&#xs(Pue z&H>4~KF;?<+C;1iZUoNndg=XtBmsqRMZ!Bh;Hz@85g<$?3IMyOrk)-VY7{Ljm`o+_ zaT3tQ9n2l~t0{vrr;7Z4mmo`r+mJxEepUH8#bg@Iqkw3kOV09cdXv1VxSA z_#6yscmpqakrbwY8DaUJ+hTIuVXcy+Z+4$?8w)NjVwp0B$S2RFXKRpp>i z0@(wQJG0z$+IN|GyH%YhpnQCDilkxECML3S6X+^UUBFG6ggY>4~T_(|2w0n(h+W5;I!Gj80*5MQAh~XWV|{Wl<&CdFnd5w~EFW zZB&6#^Ktbm`H+msrAskCe%u@$9qoKoT~@{qG&EA<%R`)yq+J`vl1UK}n|LWK|Em^- z)zvhx{`cAz=o4%~Iw(QPvksz=nJ5~gCa}b69!(RT!6?Xlr|1vdKQGK}&_h=Jo-d+) z&O@GZlaSH_>#yaaw%OeXE!s$CP!3Q=}?K9=s*OCgy$VV6DU{gAe&l9Ye0YLgnce*SA2buPoSLQMacH*o*m z9ttGbSY^&l9Ge=WxEK%d;2@4 zdET@P42#ODCnskW`@fT;xm2I@W-^Eb5-VA+hU}ODy->R%+2H|L$P75XI<*U@ax3vbn#=^^(w<4~E#&6R*rXUwee=^o48= z$`%wn&PPS;53bVA*sp@KUm{nDhwB|Yp4{QFl+cU7WMlJnB}1j^Cw3Uz9T< z?XlaAxr&;eP9D>-(J}rw4(pu~Hn-!mSrZi4*w{2BI_JA?*)SE0&-SkK4Q>zKcgcw! zEYY$5wKTZ-JNQ5B=MYz0|Io0f(Z#n3w96gUR^6O24JgVt``In=x2yF1B>zU zy}6E?CgE{I1r1^1XNfhn)6ulg6cW!xv2LUa=^1hN26K03==Q1JT1+7_w6o!g!UEwD zkC=EGqk~0wvEbM;NolNG*>Hu|fcQ@L-J$sV>vY@-S$(;Q|4f9&vTBa1243#694k95 zC5d^dZdoc8)y26CbDq19Z%iw8vw}}6*6F@je?pN(>aWrZW(0z}4HXYsdnC=*FJgsZ z-4~Unj^iWHDy5Y0pEYP!0Gj<((8>(iD~&MCRHi(4vQAcSQV@fkMPq6m5r53(#jihd z=i;P!put`7{g|*gH|eFA&v@pJa>qTEn3@{7rTT(xMz|B9T2uTn#n*!O%o|NqketLQ za;XBeGS)3S=gyt;07`uh!mio|bvW`{F+;5~hHqijJ;H&(lHJIDO^`x)Qw`jUB%mce z1-DN%^C-Afw@`rZKW`?_EuAC6f|jfqy%?|-vFLn+2`q5lpCv(0>+(xGd_btuJia|K zeLS10FV_5Xp$B=nu$t#@saCQ|{PUMzq~%o}Jhgta@<449+TH?{6UGQgh)SDgoVUO- z-B`KNd5H!0OYP^F4_SF7H>XkpHMwi;VI-^nSx#m=Lw%X1<-fvS*JyRY*f@DR1j%RdY{gHm>=bo)3pWvn3S>&am%vH^ zz+1P#)HV$^-R8RtS7Xt$gkf4CR>l{i$^P1|Pt<{d&icX%Y`S>qdCIZ6Uy)=xy`W;dDIl;qm449L!LC^n=>hBpIHC`K(f{agHqQ zTBXrymlS@MSA=7y@@kqqP{w@MY|(*In@F6@b7-E+CSXfaU)rGZio#Nc&GO{i5?S3O zpc9X-nCb0~vTFT}dSs`}JuS!inv1Zd*mvAennKE1w!$a7w_3euZ60pLbEOkKLSLi>{rH>_k$ZbR!-lREY@g$5a*uA^yrb# zaVniH2!R}AFgBLF3IW!ur6we&wY|Dl{XM9t?Ql%yvVc)HkKpUxCaA{C7Ls=$(2@Hj zDo`*l#@KUV%)+aWcXqJ#)=S^}^McfH0dbqH9;-U;?PIuQKT(BzS6*voq+LTOadV}<#7D=K?rXQ?3fO{DyZdNQ_|@T7=KH9;1uM96_p@SmO0kgxqP8hZYC@^}{3 zy`i#{TC{Y~QRUd<@!9Osj)mIR2ay*|D?77bkTC2{mtqI)AH7p=7WVt8smE(?A?D@8 z*93BR8ki^>E6#wXT2Q@ge66Eyu^<7|3nzP1YbQ+n8SOn3^JhY@_i&|ovmd9$0}9*9 zOXcLn>e)tx-lM_6!EofK30SXNtn}X){qnlAZSsd4Exm++VpH6cv3apRUt9JcEF~pMeYwOqOiGfx^buP|CCl~-`>UWI1IxFjjT zdroa!H1O3EkTACocL{uh@L2An#9g!#3|5!Y_}@m36~r`^o|jw$6CA_U09z|bCv|n~ z?;juDN7+1}0eOLAOJPlMDr6(aAw^v_g$v9K@D{mVp18D`M zMmy+JVygQUMT!`QxIq^9LgSOJ{2~Z6Kx5#vd#BhTf>#d=emP~$GaeALeeP$eIspI3 zyBG|QIbb0@U4Ev`v`HH~$C5YrXB~t1(QB$_Q!X8oe^fEewsk@t{mq8#qC2_MY!wa0 z8L4uAaP`RfElnlKdgaWSe5vQZWZJyvw8_FLWBMddjKxJ&*18-uv=H(1=I%1dbNNrs zt7EPSGg`r7?ivKC-lSeFdnF-w!JB#xyJ)xG!R^0^y)rQ0?%c zb&0#8>NT%>5=#%;aW_Bdi`zoavKgRkvVx#!6b6S3F6?_B8Cx%13;FzPUy423rz_Eb z6Khc88MK@{G}Ky73gDgdfHd!-N00DPJ-r&V%r`t@Zr|yS+pt$tRI1z3w#CeU-lJtX zb|Xj6DC-Z$*r)O_!9al>ZFwCBvT$RCpYk-(rv~5VeFW_hH7yTufrjPvu4636s|V-d zN?j#VF+}4x_15>iPC;t>>x8H6rc={KGJw+Fn8>E53Y0yO@-m+F(eu7s+*|N%D>m5X zMe2g>o+`b`Skw{+o||W+f)$rWV7|Mfkp_lqdbIJzo?M~y0I3+MM4Y;>E_x5X_hv_e z`VJ@lpY1b$C`8_~`^hGmjwpZPlJ;s`{~g%yiwV_S^L0_ZuNeybC*MV$ojCXVH!8n=&PN-ibu4?R4l1@f`6BRPpNO>*RV zjDbrTK=d%ou0}D5n^4wWv&blNX3jGbP_X?T#-$lO4WX2Dlc9e@CPMZ$ByDG#Hwk5i zWr>gc^C`p|MqVQc+e~NxKCpN0L5klpZIIJNv-2#?V0P zg9HJ;r0%z$^;VcyhdCzSs^|`LPZxc=M$zme|BX56%*gw0|`j(3%uKB#yzWqvseR<3vt(uS^v3XiCN9ozi7tSgmcWAy<6Dh=M#v>x(Jz^kg zT~B*@Dp(o}m%y`cZFO?kgLTI_`!wYf+j}LyE`CzJkmvj-PN1I`Q_RqYb*u?@Hup*Q zsG9~54tlPVJ&z4@qmkz2kz4+MwraZ!aNLG=!Vo)@Bxzj`^g{^1o_5i{ps%DK?W-0z zhTBfV>~q!v8r9B?2ZFo)CFm@&7T~&)C0CeFeo(WMaP0^OD3t_@;{i>IM++0#q~F-2 zXd2&=aDAe-J(r4SW>%t=H6NWF60FLB+-Jxff$n&}^$8}^)H|R9LAyPp;1!>PrLmB@ zt3*)0tl@^+8uKtkEuzY-KKEFFkqow+V^cK^rf7LVC0{lt->LyD$yQk*|_^R*GlNO~e;86NKEl~}d zvEnS)isRLXrl)FRXyuZwk2o$R&IB`F;$=-v-Uu1(N@mHlpJ~Idce6k65=5_3C|(IN zuuO74n<>FdfF&weWBC2%_fC?Ap6oaTM=Xhl*cA}0Jqc(nYzcDezsD$=9-Xcu(2AKc zJRc>XFA&6FPg+&1W=^YyVT`7t-;DC3RsM@pIf)xL!8xXNbmODHIX4}tbePiw*fo?_-PU{m87Se z^q=X`%DL7?W7A>z@yh6dKQ|2-f=oCuL8&|^PZjzJr|i}l1dz7As_};~HG2?JN%mX| zT7LAZCQ}R``BtQZt3HB2N4OU^GqT1vX{6vX#8`&Gpkfx3I9Wu=KRlFTVfdhK6q#mG0GL*mowi!^oM~mwS5d z_>iM*9+x8Wntx%v(e$7DN?S)C2!b~sn#;Uis#DbB#=X2Ay-V%zwNKJxno3lJU0AQb z!AO=wfuUhSOrEv>YFpMf4ADBDOWB<~vnaJ=YiU!Dt7IIS^y7AyuDR^(8ue%6M%)>i zgzqeN{mUGK{5s9mmh%{HONZ*lM?C)|_9iSu4QSOmd>A}iZ_B@Swx5yKjLclmL$OHk z+#CCn)bDkiqG1nK2U~w`mRiuep4WP>=0Se{1)Xf)c;(|6kyioFxGrN=;E+va-=V(x zyGH~C1fk@7?KnwnwTy-GyshCUrXWxr$O`f3NAfV=l@Z&%9N90mo;<+UmtXWF*Zr(8 zbkOixnQigJ5yD9zH;+s|DFmtyDZo4M-ag}@&nG=WF)<^TT+k>cO7}N`oV6qafhf0k zSDWEo9s272^!e3B@Qk6kV9m|wp(-2LW7+QuI9)>`lG}fd;l3Xn8_OQdDl^EAAfn;7 zL1c}jW2CNn*~dkaOb)#fGHEF)wRr+AJosG3K(hh5!>-?fR1;NI)U7cTF;AcBZ9tNc zmEV|SEyT1T#l%GD-#?dM-=zDhIKK6{3`@>!r>NKp26dxEj+H3Up`ACf!jh3lTW02IsI*F43+97Wx7j!!#pHw;t361 z^1MG*_P+f3WYWXEN;I|sL*tkGqB}iQMrl@2P8DO_Dx#+DSwcO>JD?%cI^edT)iD8PjAxfu>i*|F%lA}UdD1*= zQaqyxz1=A;{OH;xb^jM3BS3X=X8X5YMkwStEf4uUKL3!NrwcC(FctBl68o}1a6Zq8 zZS=f!&WT3;gj+DB4gZ9(utkDY@Z85UI%Ba!n1YScIyiw~E z+=sEE3RPb2EsqZ=CH_pegke!m-(kGUma`hB>ASl>>7erQx^_7|eeCY~+3e4ILRhJy znk0sAr8Q>_gq;gi>c2Lq^g9l4N#9;e=pt1ORl1_Gvx#?w%SI!WiB$|A@^5(&GM@*j zcOjE8L(6Too^wb7hxBso+2(gok$^23!HbYtAt)@Yfy7zD!cwVe{M0n(eVCS0#~Qz8 zBdIca^+11x4-p2}j%FDGl?WobWDT%fv2rvk!~J#f)aS8Ic>U65v{&5!H-xp0Jzj%# z{&=MJs=KroP(0ci19ruNT1v_}C4y+6Tr z_WWanflE!PbB=@|4RA#>ACbgh{MLK6OV+-0z&X-i1)kt&aL9FD7X}QVaH|wqEZ~iW z>=Yy_tAyNH^v3Qlp7|R$;zyx6HQjm8clDxKd6rc7-Ipb-!$z)n)wbPUJr_S4RpkF6 z>&KtHH>y{Kr)|!;C-sDrMGI6ZX-!lQW58^ZVI_UaR?zO7Se>)l#;=ympk^|jQtE{; z3}2-tU*1tYdFX8=+Q&3c$I04fy40+i|Ni>BYcv`C2is|HMa#Y+GOOGIH*mtg_~YC+ zCBK9;#_}(79|Y))bcR26nTXLD+_vnHN4IbHFlfY1YkdeN z)M#s}i<3K#`!$4Yz>tHHq~10oovalBWzZJUK|5kG=Kmi(?m4RkF3b{Wdfh*l?e6-) z@vw)+OYqbwZ8#AiGn#z1j6Zn#`SGQU1wnPEQg2XgKrGw)N}uHmM-YbywGrI)wL2r0 zCKoQGF7lidZSK*5Att~Zn1ca9^~~h(8Bh=Ty51H?0|OyIuERgY-yc<-A2*~Ql0m3; z)Q+_+%5+6DFhZt+vfJVO1P zn4j_LqMrQKI&c5PR%(_R_A@|Yfcl~KJq+{^8T@xia39bjbF&L#E!@ z8y!u6AijXvIDtxNk;;87flq<8kZ7G@w@ urW{d@1>%?mnn?pQD?=d;N|eJMx;jy z(P|Ju{j2`HE1Oe!y8PvJEY6cOpUP2B@g zD=6g-ST+<`xLYjo=_gm|xzI{3{S1|BcjjWX zxMO-tsWieY?R`_Y4LUeuw6gr(z47{yd+{`RJC3768>o$xOY~+AtS=D01 z3I0}C2ck0g$(8fJAvZVZX!2diq1^{B2jMc1=yg@GrVJQ}m0h$+R5zhj2JuD!r1fw8 zUVLZ|di($g&YRu|YF&|)k(oX|+~`EvqbQ&^OUYmC-Ql;5=$=fPWuB;|m%aMDe~Nm0 zJUiK9P@e;S>wx(b9TZYqQs#8V?~dE-N{E}F*aL}WEW$Dauhs?L(OxU=o@RAQ?D^7< zMv7(O4vbm0cv_V?e*QzW9h~~qwAyi0v|kI>E3zXx8C!m}%MK0ic+6UCtm19cM%T1+ zx(Lg>VX1n$ThG?PcrlaaX%CTm4#p?Q2Jc9H)PN#pW+z-o>f#%goN+eVTz=2hCN+ON z-cf_Y!6Dy57p-~b==l%PmMdlFZn!dyk1Xb%3(H#m z!8l1vPyb!$XLuwfPJVtq;u=t?LX3dt=(FmwVEaFJ8!u+Kp#W)tf49A>y*BY1#z?FfAHm zbucCYHO%Ti;dx?}1&(A68qxdI)5Kh0$2?hK4AJHj#MZ=Wc-CYm6 z4;y;|RxirSF>v32haif_>~xLt1<-te@eUV$AmQYlLpNl(-3Nb2WQWt4DlbdFQlt2lFwIRzcrddMnEvW)xq=7PFNqyfYOi0V#cp3qEl;J7ltdkgDanu z>A3jSzcAe_(W;=dS>n3HiFXuDai^7{I+vx4`-w-nk>G`(4I@(0oagFkvKTz;)4r`7 z^!?c_TF+iE5tC%d_zNao-ujB=*=oSt!D+eiOruGH6flU#r3z}Zy#aXmZGK@Sw~4d} zaU;Ba_Hzcm{}VH{gqIa2D*b?#Zg$x8eEZSwZArJUuYzN;WsDxvpzj_U3UI}RBc;~x za9gd+0oX-APzkY38c23}GV2gsr>LXjhLmQxTa(D}J`2r@=`Q`Xai@6eMaVk2M!L0aWzB3!w-Db zoj~g}v5+k<2O<=(d-Ejfzxrs^=HT|YfwTHe)+a70q7RIkke=e+xMMyMyxkW70Pf2h zgWL$XFmca^kJ<9#&W;FR=id-3^SwVm0m4(5FW@wjVuLatekYJE^E>%@@PemaS*aER zt752}VvunHC3-S5GXwTSF%735gxxANYciE9E9Rs^DH4GZG-(30GXtwB!R)bJUZWw- zSz%hlfl0LR>KE*I}fd z06Q~eHiq+jUWmB8G(Z~NU8tCEoNwfjrn%>%v3JO2fXjb+X6LSzw6)j`e0-CNY!ka7 z^-NBlQG88c|=ZBMzQ zuv{3rooQvOPs-+b-3R$RH{>wyPkq%&c1{<4eB{16=zHc$qwmr$*=AaTEIdKx*0#7b z5AfO|A;-X~L`gjbTq<^-O7#F4N)J7+X_}egGU9n?W zy)YH;4sKdO>b#(+XemPw9-9LkVx%aB#n4eJPo5%8IuhfRmO7{vG(ZPFgFfP|&_m0& z0E_O!g4;yj6;(~^lB6FXo5v*WxCqf73T9P^42*aM~Q$5=(z2Gc}*N zn3p`{(`rCQpLfKkWL!wcU3jSJ*o%E;kBf3MQawWW$$$glygOeXVIqqYF=+XoIvZne z8tZLW5KR?{u+d+yQa>1CRHjHZBriyCmA^+r4|mRi<-odI*ri*9Sz)}=wDr~~E&d)j z?&n}BbC-V>Q;cLZQq2NbpC#o;3@@MYNpl(G!Tu)?8D-&Bb^ZE^2w5x8ltsou&%@7} zh~cUP^OUG%4Jk}utzZeCtMG>=#D4N^V3w>4eKa^_I6IjUDjKoG^53CRY_-n&ll3v545(}aN+UlMQD0z^i;AqZCrkohpl~~@CGrY z&h=sWn8tJ)N_5OD^5aZ^r|UPD-B_n-n=EXQZ4A!1MG5^XWGVX~5!B1ro2NiCYVB{`VZgqclup9|oiJ&u z+P`U8_QH{#{iq{`_gOHWT27kL6{pq4j4eslWyA(rP^9UnpOYZQ>nueP-Bv)`g3^eF zv*%}_;`x&%rj+1Bc5^{%bf>n5j!)Y9Cbzm%FDX!0{yYxL^Eb0`)}CJekbzy>1C7Yy~CBo&0k=Ti0K zNk7kot$IX-nu1vzOJVf4uR=Xm{5XF5TEvqTiMLm?>am**F|ARy>E>_leQ6O(c5#wh z`t2<8_oE%|PGE8u?{87P$xA^r*ZFx2_z1-PoHi4E;2$6a25V$l!Q4`NL1~B^hLN>R_pa5Jqthr>qCoj_i4q4x8U zfYx|A5!BJTPQF!G4?+wBij+kiEzH@G9lydkKWPB#cl7YWQ#|!j?0e^+tu56p&irTH zk>eBJpr6-PYUcm$hzhtPkQ^U{7o*Ms$wcW-U5P?QhntOiQ=uFvm=gYWPhjTeK%mtI z7(Ahtlo=*GDTy3KNf;qA9YJ=9n}Z{k_yc}4ZaASM9E%;W~A z7%_J1qE6FX`0}1kI&Y=}QG5^4=QV;fUbR>orq{J@0=%E*xdi$|ABexv$}-Uyz&M!Q zeg2eEqX}Z7I%%$)zw8+kSu8hzU+3hMWL&4Cn6CA{_)LNX4b|9-cACPyPJ8aEdn%Ur ze+ntMYJC*9X#>Ak9=t=e3mRAfc$?V@up6;YoI%@LQP%+x z@uW9+NKbgR{~Q_@cuk(H7EBym;2QLkGAnKn`044|$d<`w|3NlU5*ag*))@7g6labo z6X;L1AZ-ty@$)B7n!(GS0M7O4!>PZjNUQ=m5lpGPI}Hxuqbp@b3~aiAHAu54ga#H$RKEDCW@t@#d}tv+&bZhFC<>`iS#!e0WZq{E zbt(iGHOpWgbO`iNqn!Q*=@IIwj^**xw6xT=z0g~9lC&<~l?bfmRltobb~gSd*7c38 zgekK3asZZG71>>DX$6Y&t{C$Eb0O>v32)xlMiE5vbrjOl;b`9-aJgAn(x_h~!mXvq zb%~=T!b)2DZsp2vd$n3oKJ{OMZLESJ_S~ZeDK?P2VpnwAp^#_x=pQu^wUjHYSH5mn z2gtQ$$L)@VJlon8&?lbA!NJCE9iBOT1YTY2rjH?*eJDwEWaRBket%ed|8G~}JtU*~ zFh2GNUXiy-Q~lkE{kOa1Z`OHo+lH=D=CwZdFYGg=IA2JI0|4J+%XSkTw!26!3V>zM zJFpC+E))#9$odWvWYo|Bca`iWUFqxEo$KqsjcBQMP5a+< z@$Lj0L8Spzo|E?)Jw;F$zl0yQ83gGPJEPY+Q9}T!Rdx_0h6J(T^>IDbONxhsati+E z-UUl`t;0D_&zJw)?_L*-kErurw1Z}QA3BqlAU8jP7F5echw$CRQfJ4hs~5>HNjj#| z%$-#d!cND#P_7c)Dg851^ri8s`r}hwR7=7-V)e_yyh4I44;+}S611}Tao8zHUocpm z)({XEEP$c#X%Bw;;NdE40vck@nxpDZONlINR=+6y_72P%8a`xgyCdeUxGz0t$=`$v zZaL~c;CcSu|8E}WXs^#p_o%3c^Gvt3)aTQ}m!h*pdHZ=?b;ZwA{}M59AQ@b{CMw!N zks)~f`HD(BtY;^2vD2ZRHCRznU%ToR;_|WOW1*nLLEZbsQ___csoyCvjCb%7u?%cA z)S9j{ZZcgJGzc$~2UGV4KCb=a(rzhEJOE0npQR>a8ZvhOrhd2noU&@lt-OEL?P0(w z+2YwQvTCaOWn6n8_DEUGlOO?P8o)^~b7;c}d2`u`5(T{gL>b%BaTeXskO5Uzf7#nx z8hP`P_d7GQ1o52FWRb3Od3ls4Yb1eZeHr}Q3#nPGH%>tHy6?p~rUn>t%eA_pKy1PaEYkPC`)(m%dHTn*7adeOtSQF5sV)#D z8WN&%^X5$u5F|?sivQb;{+mOYDj`9_3h3_T8doNe%Am4^!W#D0t@nc-hX0Xe6ZG)6 zn+*WM&#R;|nVpNKnHQ^9l{|YUSn!RG@{<;A-B=>6d%U2yu`nqV9HuXUk!_UZ+g$j6 z;62_qO>2B(YcUC-o#QECYk`=y3>(pk0I3QWqr9oy$#bbXYf5@4*8@P)Tlce^fOhK^ zwUb{7M}yOMo{7?elAr$U-EP+p7S2Ud4aVsNmtJ%`%~(xbTRHku{!%yq!o)9?+9|6jNfoB;kY0 z{tegc!i;gAv7>KO`jQb-2d4Hci5N5+G(a)&xPv%iYz3Y`o~ zTDgzO^N;B3C$th0;qXo;P?#Oa*9sjB?3$evU-3!h5k+x#h1sLV8T&s@#oKw z+U7Bdlt(p7JW(hi4QaD8L@gqa&hXEgkf2}_N(c3v-Y9;kRTBx<-Rm_Q<0uRxaZe0hn`7m9)9@b)TZ*=hG>XJi4<4H_bzF!FP15 z5hQTW&-$o&H%R~Byh+5tWS`6{G*fu!c#akewW6^i5UjfWI|p}rc;kmlmFkqjjI zuTX*n7N@E<|Gq>4F>V4WDjJ!ex<`%vxA*yCWpr1gvwIJ=UVbnqwB9xqcb*IhZ29Ap z+tog>XBFATTYK@em9UKm?wwY`RvFGFs%h3BPAmeV99B_hv7QJ1IFJcT2YI_5EWc6v2h!J*3-5yn|(LRF#j$lZ`2 z#Q(p&Da4_}s|d_Ho4p74a1bSn9Lh5uQ6iR5Zx)oDP6A_*+hh*nA&~liw&FC9LukAN zL2%vLvv?}4&`2Wek^3Fosg9f|1!OE{QUbippKJi!m5k^G<|xrsY0?j9CGQJoZ^M7| z%)CDPh1$&?B9#IYy63jqe7jap@i%bmp^%6u;Qa{EN}{WGemg23@6Q2lPfu1rk&0 z=n%38EpZ`v^q2Ry#QoMKUtSMoi6?nf=~C9?IQ2=UPAhfQyoz;6^~gpi0@sI&=VSTJ zViYZDc2R;F_DquDh8w}*U}%xh)X*U(P&1b;cV~Fux>Gchy72Z{X&{A0j{||tKV<%O zH_JFKwR~rv%W?k4pdvet>vWygFLU`J)d61?gV@S_g33!uuUL2$ITE#069&Gilgb2n zx2`T*KEbqb_oO}M?qBuH9oY^&z1MFr_4xk)^VoQ#XGq;6!dtKBZ04$V8AS6dO&jO^ zIwlI!yC5MhfEN%YFCB}yq=F$KaT~8Y+a>AdW!>81cv$c%Z*ElO#`bW@&=jedfdrE= z8#Qq5(MW%aW=F%3k35j!hei+Bd;(${rv9zl|I5qYn7ju$QYE1L%A5Q6{YwUB#{{Vt z(QSduPM)3Ypvgo`G~5=pkU^2`m<@Ir)PXqA1qJbe)+-VsL~Os+`Mf;7&z)E26Ch&{^{QZNtcmMhm*{{$fs*oUVYxCq zJMl`784YBDGjW{D&CT5cCU6#*z;>a>c@c0fra@;E_Vg)}{@H-WNsoxK>T2nkAc#E3 zD};NktLt4{FS0@Y4~#$2{((%e1GHwb9rS0esP4@skX=n=FS7f=d-|#H=ZdQ4NJ=5A zD&-L>d5$l24s~kftaS^~&dpeQsYQas^Q85P}T7nk(unhYV z^2n4w6eb;hOjQ$}!=1wY1ms-@ck17>WynkK%jZvShrVsJ{P=S$=&`L0AYA}skSB$$ z>ok$VFw2E!&>A(msb(w88jqVZyTl-tY;ji-z>OGD+#;za2*SLmL2QUa7sjTaGyd{6 z+576Rt5yrl%?hCR1Ic>REgu{MwG5_n@x-2AogdHCjJvl&eEv2Y|KL4j_Rf>1qTx4C zb8!(t_lBt{uJvfi2-;-?MZXfF`O!l|tVek{dH`{)Qegq$KYi)YO%2PgRM&&RW7L+X z;EjkN&?^7sh+c4E|tufUmRz9%<5D5{6lD@2O;nUrXxb z2xWzh+_6{o^8F{9!8a3$V%@-wkOgtaFc1%&D-Kd-hqSI@7|Xy5Pc#ZkIpGpS`R35o zu2I)$1(%(Fn6y0fxCHRD6C(m#Q8F^aw&P;B4ZE+3OE0&~6B)6}H{-|HZ^kWWnN>~j zlV``XvQD4#vk_(Z*Wty(gB)J-$#d8{gtk#(>)Yyk0k#K!!ltn($oQerxV(Mf^04`_<2zs}gM*c4 zI~{_Py2+|JniyMQ#)SktN|XlhU)T(|V54l5bpLNe5y}~Yvw?K8tmHZ$4(T*(Agc0m zl>-r4=vYfPfHe)5dY}N_>l)XU@_l+vz6mm!s+Y}u&r%@ z8lD`vsp&3Rg@dH+g{7UWDwrS8$ARrW8z(_}AWWExmvpo^3nMxCnIjK0Cn59~geQVr zpD36>0)f>1cIelrS#@T;)8;u~$nYbVQT4bja`Bp5+k4d;>eh{Cri#AP>eF2R+Lm&` zT%)*3Iff@HtQ()yG%UABQjTghm)0yCTu9(WDlRWnU*dG)#JZ(VqIlC~#^q-8@{SrK zK@&c048h(&k)WVRWJx9GUiIOl2871;pgl!N zePZwS?GT41L0ZzeM#43{{d*HBO%NsmzO_gcc^xZ>VtPP}O?53~hYnW8-UBLyvY+*0 zDi?+uXT?91^Oy_j5>?RClkx<#+yCFr0y&hm3HeG#JU1=Cb_*kWJTwygYik`!HBiw& zQKI}L3-Axd>Ld6GEl@e7nbSa)ha_;ja%<%Fr18ZkN8110HXG8rpI6?(LuC_E1*1lR z=zj6`C-K$51v-#D(1H1}8MWM8n$b}-j6gL3%Fkl{{Bk&wM{Nc*&TMRe9|$gA%trO| z=cGu116Xr9C%cVqYLxUxWg76~^FxB6Cz53XeyFXlPgCPM`qKUq)+0?mgM#Und0Z>w z&*3T@x~UO+G1eW|9fa;u_cQS?r7}LvV6VH~0h=gjwI4~Te?#fobKx^P5i4f}$7p=$ zlA-79!M7N4P1arax+<%%A^$i` z^h$HQ@HidJAd-GdJoY5AG+O?4-z22}CZ>Fl(;Rx4>eY5%RDW{Z|M^G8%UVK$!NoZ(tFoDWwaXwJH9BlIz2qV1?%19Lct?eSl#sN>_Mb*E^V7kd18;Y_5;?CC7XZC~ z+}9vBGQ<4+`E5{4&o#a3c7KrmwDDFY2e%#;qfOQ>qcPc!hu%t&Y2H*y&JRv3=MG&` zjS*Uayo2)+X7UR!L2Uv`mh0FTACnk-2$iaHp@bmRyjr)rgfJL1Vagc@Q13`&iiC|~C@Ka#JbZMH@nqK?&V%Q2JOQ5}4@LyWji}X%O1>K#ee>gZa$S429&aN#)=b%PBC)^0uL^ zQkWL=fg6xUs+E%m+d@&WcX5W>)H}9^K>u9fd$IE42TGRBS^&AgO`~?#PBwp5U>qZW zGA>^bZ~R4#lKe_v_aSHJ`cKfJ9O{LHLT*=$qSg<-E{#TT{-!%Pn_jzx>xJ&>d?=6RDl zeJr>aNv|8dfSCbT z0Q=NkEql+OQGx0v&#C*z*uuVRsBPZ3$4u$dZS24l|1>N*pNC2!7zTH7()Z$DY@))Z zaiGicR4l5uPIjFY6s;!_Xgzy+a7{P`Bh9W(W*==WoxM<=b{BaIouxdir7bQ*KK7G$ zS87v(u?1_FZi{A@lfgY}EAIWnN*4b6;suFcand4aUIfLBd7X7ATHI~p{x~hAAC`7` zDO-t`zxh%9%mr-7`(sgMEW^f5!de`_#SUnSsK}1Ttx}c{C@RCVKUts#JGkht`&oNa z(eq!~h#6}>H}GZI+7R#BGs3KZUJ2YAdoqRJMlq3uO}uL6Oent&*2FQMo($jtgbrO> z0>s3`61FXPXth}_IOtC@D-aU{;Bm>w_-hvic35N;HJ%N4GPt~%QI`gFVn3|m{>gR3 zC{LOx z7FR(FoVqVL83;}v1Hs+Zd&qDv*~6=MZrszmkW-Jk%qSn_-cAE+(EQw@+W7M2FLVCD zboGPP@$K*LxyUwx+;pjoX&7nv6^>wcQwz`qp5TSFoh{Xz9%cJ-O{DUBE4lLDn>-`O zZ|F|BL6%`3#vI>$2x90np4sVcX6QvMJU!{0xZDabkmvP%ihEgNUn>|sljfp;^MWl>TJ^pgU#SueSK^u#h2rZ9qwp5`6 z^fis2WCS6xpGy1mw%@qj;+NQ7o|e#vzbwI5KK!#hu^3%;l;&fb$kwamTCce$)r_uo#+j=lW5cM#zn5aL$u zJej0IcCBd4l&l}equwkp;L{dBzl&Q?#AcI});T5lu{rk-J;8Hzg6qIGQN-Y!O&2<^ z1;LPaPy`0-4rpI{GnE7*7q$I~B-7UA-(tp(E)pc!uRcOy0TtE_7rvUqNC?nMQPu+3 zZ8`xB$aRl(_U@A%ZL_`u73b00e-MofTMv>fV5ey&Kuo-i`D4}BJAGS@P+sh%uCFvG z(gQgHs*g~;h3b6a>xf$G+?EZ*rJNYeVHX@y885*(XV(IEck=+m2cPnO;fDU-diARPjt-#odW=i7U~?>~Eub&cCG){RTAIOm+l zJc2BbS}l>d0CQJED40+6w}g1_gZnYv)m}@#SSYH=j8kGw3A=MH(2(#1y?u@IA>kq2 z%d8AO(UJBl2CB^#D-lZUB0!GfbZ)dfzZ^!PL_{Q6NHw>4fi^8DOnK@-otQy%h}frp zIl;-4N9A9VN1awzhe;@m&B?=z!d{g^W$#UM3OXLZpJ=^I`^;zwVL@bJz+(-lD zT?6>fyf7!imTn3<^HA{n=faJZ+EsPt70z3zq4Muo2DEHKL1&^lIeFyOZmnhJ-!AK_ zxTr_UwB4ja?t@15vv^t+b|J%-p|4NBy!h2|BBrPrRInk?ypy5|Q!CGI11u@nT%n~i zoI}P4jdkpZyQu-=FiAK% z_S1*=hl{WuxlI6zJ852I!i5^9jQ+7?DQ`C|G4&O7b#}}w@=~7xCWTw))``0qxJ~qr z<-dR53WPzE1fHjXNH?0BbR(LXVVK<+mL$KS!OPrngAGH`=e_Thna6XSGD=*SfItu# zfgTJ|$w6Ob1RGEf;ClfBV9*Qtc<-sBvj}$t{h5L%9v6RsFMMJnuWi6>C;^XeDfje4Y`pZN9$(Ea@$9wTEb?;X~6)M67lPx2_y$4E=; zw{*BdkXnQ+B}Yu}5@=RHsOVIcCC2=@+x(mm*iX&| zr0MJQS#g?zf#my})-MQ3Q|n74j7wg#zvz5Kch05Dl&i}8=@t2r@qt&5IogB;uN+SZ zmsU8f?p03ig`?iVu2+!fxp&S4ALW@x3p@!MQ<^SKo#x9MZG~S|ei{ef+ffj>otCUl zPoMUby1AQ*M(kPkS4N7hu0h39XFp@}PI9;$e$QsYVjRQ|q?Vw?g^vZ8D3GDk94_dY zC@QGhI8`?Ex-TQRb+3hP>AGzP*=o^0s;=}SBLNwqyvSEEVnpC))Bw9nq{B4OTjX@N#{SQDv9?P?+CQp84w zf9=LBlyCU&2-#O@klgJApdSchantEM>7BugOXVl~2Ab&L88Mdn02p04pP(l|X|9V} z+(;h3BoX_wlfauWteg7wE!4SN3?KM0{FxIDO{CDB>3`)vUJ$7Z82bgs7yt^$$mz46 z3sUXq+1s|yMP?R_9H$3%4IuVf^84V%P8;F^W zVe^6F5GQ@MRnx$lp+P{N5>Z)^(G(wo+mi+H5$38g@cK0YWe~;+Jg1m8pf1mzdwUOr z%XQ?p6fJGzfkZ50==GWM-){?QxAo0k@!Q|8=$PP-L-l;REArf_-RTM#^f%E=e|r2W zT!~6cJEY@RVAD+S_{VtJ7--1H#si~MbbqJl8F_tfh?|qU?kIZajyMgjS*`;OjSG|I zS1GUkN=%qfqvc*QEnvr6}soV~*Z%{ZD0tM0Bc2A;mjN~1)WD75cj z`v5S4rPLXSI%`5618D03T@HD<&1hiFS~YPzhw9VOJs|O&%-FCOt4B^6U?9Z=4)vWw z*~l3PUOECP1_V}NJxC}4nEI0AmrLkLBd&%K{LYSbHNXf&EcoAb|BmV^Gi3l|xCFSg z<){1;D@1S(i+QlN>EVFz*MG-#|M1@rGws*F6nCu_HmW}WYIu5J1Lh!BFQxJQVWOBu zlU%1$_sLoWslfm>cdR?A`*%zZY^2j*wK$a21CZ4QEK)GEnos11T|u!|*8qfk2XOD6 z%?>*Yv3$0jUceV@K-@x*ShAe+*}y{nTpp9uO9DPgU07Fhu++BU84 z&w^Q6J#+z2A=CIQ@+SiZEummUB{K}fCw5FQ9E`giso#Czddf$dv}VuYo#8+z-NVOU zI3JcZ&3=s7H9h@0_DXYA6UNp zEx+b6_^nZM>eQ=J>c(CyJGpx_mCDL@L~ZDS;U2qn*^rii&^G>(;>dj9#gMdz)VBQY zI}0i8uP2#_omKRQRb-2jrTR1V1-)bV-?IxHwNGDx(zCX7wXSOL2{{h3C-$4*o9!Ez zpg1@it18l|@P%6QMuL#~y)-K^zXiRtd}VGT<`7g)HC^T7UdIXz1o!M@bxj3Az?=`3 z{nz+Q_}q~Dc>;(5$XO9Fou4SQZjJ3$jn6BK`5b0>T#fHqSW{dxb6d!5HGuG{R#>n- zvm-qi>lh;+N5r{WfaN8aZlkUcGE#mNfkzMuu!G*=>aV0j zYB>UC5Fx@;VbW;Kf7wO!8xFFco(*nmyc_~L1zHKy1A&Lyh+LbO} z-)V!t6#6Fc4DOMhGaU^sIQ^bYmWORNB`zL6gNf93GH}3pqRK0N z%`UWnHs?|2bI&Lq)(cx#C%OgXR|PI8{3+-0=w%J#_iwK{mf|~ocJS8&`~ec9ow;?F zqt0jk@3G_F$vV@_v)6Wivlbe*3B}{Q@rK%|+Wx|Rcx-&-M8>mI^*vH=gx#Bh12%uM z9*}X88PvtNQQf9YZl%H7OCB?pYo&KH>Sp5=?`M*jNbIvD>-9zt3Dj0##xLtMve72!(~~IIZaM=Jb75zQA%6AVG?X|ndn7M z@Jd9l0w=2%oq_JU*n#vlVlpr&53>VYo4g8xXQBxV|5bmCT|Drn=Tl)m#Sz*3q!tptfE(_9 z>J&%<9ANQb0>CGV1yDKI)GA+bszMfsvjCWXL4RIIi9AW1;+E-$b_cD1HgF_K#ZhCZ8$KqN5&6RwBrwh&Gp+)kY)>JUh*mM4W1 zlpxPhF_4y5EcS3G3wKeL)qMXBB$4M_(t0Y1_y4j#n+j&HRQiMYiM|>vOJ6fTE23f* zxZD`|H=4%g{x!>mdrHo2m)SgSe-1WKgOQuCz|7>IYGbG9&Z-|bPJM8e?8>DjeWzT@~AX=MmbEZ1c6uo;_^t6#~IHuOma4sj2IZ6#UHb z9WgkBSB7vCe+IUb{w9?HX1M=#|D){fQaeDsa^VzQlVpuZ4z538G;BhkQS-G z-L^7ikY*(xrh>J@cA*QudQB1Jv@Cx)9E<~%{0o$lggmLh&rDh0bN;2{+h9r@id75P zPnSGuCPMZ_X!@<0gf)R9(6na@-WH+w^6%89?|vGfR~Gn{o8KS5pFtL+FkKj>H2hQs z>l>@@t>9$@=k;^KrR8J*)a6kKHi8QMK&=fWIPZ_|g9JeG{+x!+F|RO=A7-2;q-KTY z_i{FmnLQ$NBe*YGYHuX+%bZgc60qq%IYzrTCEYQ!kVDQ2;Ibmd5Ror~>}=5Fh6D=T z(nmIx`bCj;598R7Sl9y^21nGcL*t8x=HNfP{bqL(EWnVTBZ|U8FgYC4RIhHvlxrVx zuFAp=4ppXc&&BkUmmDqYa6?DIIs1#1`Y*mpA9erR8hC z@v_U2MJF7n&WX~+gD{&}rD#BkJ|uqy*6j3#SA*LF?pO)o;XN40sxpYx_9lrXq!b$& zNu?0f3gnCGn9vRwlm5kAY}~#zSsa`=VIX+X{tlIe+r`OxYl?1M`oNsdEG47zo>ZEv zBznfPy#$TddVK|w!p~&Qiv$izci<&FtKOC5d?&`bGya?gvSg2Xs#cy(Nf$joW3V+n zH~ZA(w+gfbbocK((3t*4LqI@}Ap`z!J1M~_zkx9S$O69(eVoYO7nNgy za&nmVpvb!vhxroAQhYfzQ@v8uguZ_8&6UZ;`G?xKlQ{*P!fd#1@NjW8AZJx{*g#9D zbm$FCM6^0623b*L1V68kY7-)edhJ>>yAIw}deQ?aTEGm20cIeJ#^iK8=<+sD<%QqC zxTQJiIbRD}4I_EOZ#cZdkop16BnQzPTxUSSPm+OvvNelFD6l`SVb>tcc?KB=*eI|L zUNB;?lcWo2ez~+$F!DjZmFcbCW}*n=#!w3x&W;EByALt*u?#7Z>F&YPGnBjob9`OB zNpA@zUp>vKUv_F-b^+=(l}RWJ&0y*vzVdTA|83ahBv$qc#dt6qkHZ<-Y$F{8efDNg zIul*?8nMOcaT3FA9Bj2net<pH+}Eg$E^*K0)IuCL?Y`^j>Zm1H-Ds^p&rcY~hCd-g{b`2)_OLAouv`4>7H&M-n-o02LBpE*3b$gy4`sZi==KZK z_i39pUX|Eht+o42wfn1vVw&ZSP-sLvS^fL9KeTA5fqb9m^zq@(C{BZS0n zj{r#DZYj6mKC4Dt}zjKUkl=lWCYgl`O+1NilzPgtsu9Kk23&nFQVuxyY) zGX_ZVAdHfTm>7j!qm$?N0yDKE>5{z z`al?l7s*s;{4>DrMWVGB*nz8$ZHy)wlJICdG>n>wjoG`W=(*U8&NO#V_^rO>uTQ&D zNhU$B>`3{aB5Lj(We7#@XG2&{bo$GP*3&xCz63$t0O8R5QwQTtlQ-9k^}Q05 zo(PLq*x_rMl1NCGQeQa?tjne>T&eZXsk|$;w&W8wAW~gpenYM(*?@qo0e7zV`a9#L zY$cZL9$sUmGx~z06JgcwWkRfe^B>kP_hsVx5d8km+RiO>lb;?tDQ&DFtGirRNZURb zn=JAU=Yq?W!WyNdC}neWql?Ub*0*s|YP^=WdrX6L2`Nm++FFWi&sbenMcnIyqaB^r zocS)W9d0$CX3)u4>)Ey=`Oju)0*A=wFDqDCVjAtE0Z&EJhZ{bGV))Vx<%P{_E1ZuL&TB-Ir#8{9rsIlsyf zcls>|cYXwO56peN^xu*fRAObD=4Dw}94eI`SueSeC_b+td74(Ef7iWy_%9s=$xC^T zta>pa&ZGZp59cK22*#l^33_%7fFH%j>xBOW|Dmx8`&Mdo$pBy%zsk?&TJxGZ{8r%r zVD&WgMZsggFrTDlNgV#`_SU>IB-iR6-(6z#Fezr46n{7IiSN()``SYyd={5GXD?nL z$dS4JUm!Y|0-cCN!BwKMF@$&%v>$=nBVQ^z<{QCuAO=qSABap#b%s)b8IruCj@D|( z(HZxBOpo<89?b(Io%~2_PH}$6PsC5-46+-a$UJW4eCTmUT%jRkap7L<@^11-b<;DJ znqx~F0b?h;ffqBIhYe|0XjmNS^_xQHg)o2pXZG=L#~t~cqB9Oab(4%VrGr$dB+Nsb zjjK`N%lSvws#VloNyRH{u;a~0BpPl6cD8Vo`RyIlQ<&a>5ILnrWWn6DOmg)a2}`K4 z+(!865LpAK5v+Ib7rXiXycs)=kit!t!@*4!PR(ivcz!of#i^S)Jodbx@2yy|*od|D z3%JQ9xm?#1GUBQFKnP8}`$z@{r==}kuvDFrrx7QReN0*9)uHmm(tkaI9Wj z5NqZp@7{<;Y}G4gL6Rl<*_2axFAfT>e10hu^TqCm3dD1+_|cF1Ur!W`mks0qf0K?6 zaqE1)LMYyBYrh(Ohi}0lG=yb`w{DfE5iWP1;c1$DgOfgypI+CKhWa~Vrc^{}qr!p^ z2l-uc7siE#IV?WHShLXQuWtN6jc}L9RAggg_31uksNN$t-e>0w~+;}MyOH5jSfa_8NGNqv9xo5Qb zqZuaMiv=q#gTih+;Hcqux#jj%MNiD9+xemkKS{KKx6ep`*Lv z{ZI-rl_WkaLG#Xf5CQg%^pUl51l@A;XTtkWOaUxbyLe_2|993Pw7I1FcU<5?iE*(CvjHKO>wJ1>G9>AH-n|31BdTcSa|CCRTMR{MYRYB zAp6U2fIL3chyq9&QCK(EfnJxt7pr}QW`1i%zOdUQ9@O33x2>qaWe#z*(8 z5|gXr*8ETP%^a*0#l-CsdL<-yl^w|J|5wFF5h))>pJqqfde-|gf=*O2r9bbJQU;xQ z`*jHRU{RbI9>_E+qGl4gSh%n0>%$UDYyK&!@&q~m33s70e-Bg~rFf1R$iHeZaGwZ4 zfH0b;AYmiu%`S4r_JNJtZ?JZbeo$&|jMOxM+pv3u(*qmDEYTno!PTkPY#Re{i8p{d zkX=~HvLb?CZ1t9$b{-h)5YM2)34GH8CDz%GqM`2@not@Yqh$_|E*MU=d=xV+-YGibk1@Ncy?v9`QPA>RF6+#D5Q zsTbQf(s;j{+gRj@G(CP+zk6BEQCWm$R>8wlb-b?0JUvaYv8H75FHp1*#-5neSMQyF_V$q*9A>6 z+4W*L%|!y11rn)uvD*Ha&0qrV*-L%@9K~08PUSS)OK1;>PW-81L(Lz_oku~4cE8bx zl!n4h9{xML4|s9X?*Bz`o|HMs>pK7Ae-D2fEE5tiE<|Jsv?qf7`DtWFg>09B*s$RR zInHQ$Iw32A)o26IU=b#aY|j5u@T(-lO{cqh19)cexHiG?0wSqwNAzs7Fh$t^4&G5y zw}|n%ejsjE^mTvsTrMZWhTgQiY%K1EULmaB}ry^BqQJ{}&qrk<@NoQQszAsxrCwNI4CBq2Uvm5SLv! z@W%p$<8&GX7{U+f$_D?XhCq)MF&E_|P(OW5BJPNBWxsgz1!aVyG)cf} zKs3975<6gsxp|)Adg}W#wv*zPABb_gHE61ov)hsjvse`@<*s(9TxndizX^kS#&sSX zm0JQqO}Wzuyocy)e;6~Gu6dF-`hkc501!0tpvQi=yE0oO0-YpUydY-Yo4c>xsQwSW zf7;CKw&&<7TIybU}Qoi2fZyB^4~{(jHMPmEr?_aG%8pSqKW`x z78&fsS%CZyh8Pp@lF{joN&WT-I=52caaUNsc~I#jE+Sp)Mbr1zPJ%`sY3eaIuM*#i zbEK+xDw=^-%CV+&WMVsa&>jKi^m{Qug&&v;UfD@(>hJMkW1T2i$Q$SUrzs?LtkoPX z?-4V^PeIo z)F^&k^)r~Q{j;4jsJF3-D?Q`ehk50Db)7sBEAj#r(Z&w8jBylGZgPy4R&6P?7B;!xtZ(1U{I#4#T>G_8Og zBM|~hYJmj0Pm}*kDL%>j1`gMW=ndkAK_hTUN#V;zl)Vo&N+M~UZ9q}oP}6}!=x)4k zNPSc=c1Ne4&+!wrCTGpvX*J%yD@qM6UJsqfFzqFeW5~zc4(}Sgl^@26@0e~3fW}}3 z*5^q58+S#}jL_xFDfc!SOoV?ccGC*(a8&PguZ<*j#gKb*f$yx!m#9|_<38QLoSia3 zbvKe!T*{n=|A&fG*D5JYwjBbf*N-gQjIK0dtUQz=@wPc<#;A@dlEa5P7fW6FNTZtb}^d?y#kiX znZJ*_sO;nvKGn_NO=+*s`pT&FAjtoE{l-Nj9QejhpEex8&Rh;@q_O;fo$|9SN%Ivi zp@G2d;LPy;evLn?7BWpeetUvw4ICr2M?dMc!aF!DQg2b-Fg6+5LdIZMh@)5}AUY)X z=gHohq;$dI>c4+0?*!wcSg0uap580adHYdkzk0fehW75Y5Q=w&yDR-Qg@5027!s^gXYoiGzw9gzB*S7ED46kP$^@*z6pzT;o-@Kf+a5VQapPQ+1C$3$KU_d>cWIz?0CoP6H`Zf2&yjMYQ zjeJs<87FPAnR3T#g@KpWhudy|U91*co4Xm+4&mJ=4pGKSw-b1eUIO1#&# zs&jLQkGJFJRHnPQY=QCKHPv;U*c=(%!yYZsQumKS!@UH_#&YKL4-eSJwQf5nC@lw&WkkYpkX7l=7$VdF~*Foo&% z0+F=ddWoMVO9-wwz8}|@2K)s|!itB&g7GoSHor@XCXIh_I!aH)X4Ou%?g~jxvC({; z4W6xShXz4xcTjWxr?N{EpT^?H!#g1=B4=A8njL002e&o2hvM{yOBz|Lg-?)fRBfp+x4B-k$+9StOVn)kJ@ zX{CD+>RtK6BAoT>lfg$e5&p;tvEYmA`4n=iQChdRDPpTPdR!;IX$xjEH74uUuy$gO zW%;BR@w8VcINJP5*l8=tx43v0X*+xNydf3v`hZIrbA`~baC-R5n_o^3yDL>X>h~zl z#P9f~nNpFmj1?0m&rJMTWzgq7PH+A2PU`R&oU&e`%T%Bpve6Jj+XFG2)7U<4b`7?a z8!>s(1moYOX9zTQ*>Ws1IG3p+&TsunTZqGLnulB!pXc@V{M9GPKvDy)q)LA5Cr+C#m2z`2oyWfPSJ3ym2>ui1UU|9{D>k`|k9C>68GF+!|+!nf0j(X=rIty+POJ zYH5!XAaol*VJZI1A^U5LfARsf$<*Czvt&>2NRC@;!z0JvmJqFbdCS9ROEFEoF1*yW z=Go%1M{P7FL~N&b6SWFJ?lnpLRVL2k?NFO|zio$w(|}Ykdh63pgu^*f(x&p_i=Om* z8Rx-A$$g%;3rl6;=+Dae zc^IYgcIn~#_^yNha(#eeO4h)Jhp>WDD8$&3+U~uL>xtQ$sG2@Anyb;9~`aM)iFCz*`|Ph1~sSJ>wTnA`pNC_oV-h z&us51;gR)YnO-yOmC+135ymF>bgh12-?)xqm3I!(=Aty=u9fN^*`{8) zbSz|<`RJvs5`CWwe60>A?&QASuR8ewy!t5b|P z6S=o8Ybh@%&ASXcj~ejLA$JNCuv+5lE$sL{TE~~gD@Zm&Dz>rBsPUd`>H1sMsf#=2 zFO66@Wo_R*ZSfG!aO!j6vxNv|`VJsvfxFSl2V!@{srb=y#7Bl{F&yqVWlXDH)PQR` zEWg4a5Q{@C?Gl|20c*9P#L!>w&$YkPf+;4{PA8vy%Ys6xud=fI!7(Kb=68heeSp&& zL^`rUj9t+XGr@#nb&z-&o~kLBU!=Kxx1tYa#m`pY0b1-LjP&zJzl?N1z*qwU*(Aog zM&4ER#gjTZvpQd!7M2%bMR^J5WDGt#u3Ss}nAxIJ$FFp`$ibIos)AnBfe%GFL6v8J zc>MEzjqBgh>*n@imRU4vlF`&rKiK2)UEXGj7r$e|GZ&3ar?V>USGeL*&0dAc-d~_F zCr(b7$I`>3@KQ??MO<}hEGjU+cJAUS(t$EQRpOp)OI`IziGlLcF|?xJdLvN_a&d(+vb}M>1tZnVyJIiFgBeUlg=?FC_ial z{(9XUYWLUsl@K1j;hPB?_yh>CLh$~hKKiJ3TOqb`lb_cPvO%;HTE1GAVl>p3`4|)r{hj_U(AVWm*aWKc&-W#*85VHeDF@kO2jQs$+ z%6({JPEKlj_5c-Brgy|rFra}4h1E1zm*Z7TxG_OlH<@V+@qw{@)KQrWbYxwypaLQp z2@G}SKoA8^Wf6M z>}5vtwwA9QE12T`HBD~XPa1#Kq(aAfCG_IlRR+8poC$SxgqFU)OQ{-|Vow?k%rGDq zx;&sT;QAudTCvLcrD^VU&jPdCPO43OnnQdwCP%SSeLQaiDOlpP6sHG$;oH(Ledf(!Y)vh-4TgsQK`CTFjRqdaYNZF;0Zre+d~t3obR z!%7-VQ!2JL#0h8Bcvpz7+2{r3rD+MP+&0;%J#vCoro;Tp`w-mOTewa8IaLF>G=K2e ze1xKrAc|sap^{onGyr|C>KQ7c%xjvh7}RO82411J5JHDCYjp(13LXz4^d*4IJq5LW zAP3&clR4%xnBX=8IXW`s9QRr&1i4KFfb&~uMl%?mH+pGl2G#^98V|YbNkX0(dWjlR zgxMZz4qen&FZrk|KhPM-v`>q0VZ;}1oN-3J?NrB6ko8+=@RV&5WomTg}M3X_sKGV!6L_J18 z`9Y4C=Q^Ih@xW;q7N22jBBl+9CHN5iW4Yn`TYKPIxq)d7cz9 zg!D;-r20Vvh8|{wY|vIREGE&Zh*pE?CLFQ(r!`nLxz2GiHbu2;q?vLDC)+96C2^RG zlJiidz1@=-i~ICGj8Bii=4|OQn|wXyn}pI(C)r2QzVUZMuWiScQZ7*1AC?qZ=%#nD z(z6D)rSFtihn?->BO$ps&_CYbn9_Kdg8As3eNW~$+b9=&CwVw4!nR&hLBj8gUKph8 zP7e3$%U?7Ng?7bfKHfo-x660_)r+9T>Mv+Y86z=fayze^ViEXWilNO0olAdq@?_@G zn~7MS@Oj9f3Q=w=2uL&H%j1!sO^dIJn3Uy$O*}3NaRa7hZca{4mm}y3^WMFSgh(m~ znW?X5zE3T;BdCn~f$xjq%CjCNg1%pKHvZ^)IAk#x9)$pf14y=N0UOgc7*g8-XVx`% z@CVX}5hcYL8clv!&!S!j`k6WZ1j_+nt33GR;aHT{r+h6oc>v&gk&Ii~O)6{}zXr4m z<|~R;l4khs8RAS3+-R;Gbi7zc{r-Wb#FpR32W`@$r_P3dRpX_{nv=d*tSu<)RLzWe z=Hja5%6s+d`^AL;eJ?4}Xlc~w20l(!o)bs zlyLHPvAP<^Cd?)*RAZ=1w!(F$53eYR1kC1&hq4{xs0k?1kaj@-Z>;jldo0BL@@_Tm z%#Sx97EgmO{)59zqrSQ#PMQYM)cygCfGzOGc^)q3IZk{_r-vqTYpxy1q?jSD5^8m8 z;mPrF`ojLy^^+sTNnt?f!gB)9ikp3X!38~FC}4KLqI3d(^U_%`wD^Fp0wBK!phIF{ zNl}8`ulyi9+EiWqx73qI9nNVsx1Ht>V#&m;E$Pdn(&-wCojf05+B3@7)+<(?9KJTD z=y_?buKZQIW8#(ekcuU3_v%e;!7(Yqv;KKgLLv|7f+=!xPuaXyB-S7m6bbx8m@QL# zq{`bBU=S+ktL3kol3E>Mm3oOLdhxYx%Fz{~TON*ahNle#iw9;@4Aw`9`JD||HNz*e zUp#^!M@g7aU|5=4>m#*ZUh_ z-Iaz8ZupCDfKz1|{XUp_ZNsIhuips}X%*OyGDvy3{3A7wsPDiN@V=CVk&H1F4_=nA zD_4!sAVR00ap#>&6f2z>!F)%Gx3>&>W3i6|&4+HK;{S8f5*BQY2>dn{bIOMOG#@T& zFhQJANRO`l>m&KM3k!p4|2T*EoY^hkdlr8i)NjLy)0|PkCGr_^o9{heE>uCvG**1~ z5|!QQR@EO|$;>qV8N8d^8S$aR(3@ z)&=D0uk4BXjR@vn1g~deUWLc^FsIGj%N_0!aWb-3`$~}NmB+q&+oaBRHzN#VrQ=CS z6iUW!58HbO(M9}7!NN?Ro~C^XOVuJP%j#%5dPoV6D^KDUw+r;804Yqj9oN2TVlb7{mJ)~`w>s47rTOwgeMCwf*T~oM z_P{68WPeOSa%ho7T}(OWHp}p~l+XO)AE|G>1gXYEO^P%j&x)h8==)n+?<(b}VGEa7 zGyZTDwfm5I^}KeL6&%?CuU=eWjDivvyOw2Ri`;WE*TYwg83) zO9(bOtg7Az=-rVkQ|C4%I{Ag{fRM_-9B?cbVLY0JMr;u}`PYDOK_tv&iSBJl#5_Y6 zm=kqCGgJxmMFs)WaAtbyfY|)lQ5Gy=k>|HQvc?E3N+D$~6kZ1)aw8lsDlo#Q&PGcgT7u~nGKTQ<^xR@mC+L&-z0=k7!FN-P zkQEfD-D!mAUf{%GtZ3-ywqKn)4u}0xENG^lLRd+SHo-ES#WRo=&JV0Thw?`Au%Z3n zC9DLqY^<;ADf7MV20vU~g=|LUY{uNiTP}TdZP@2idDd2sMMJ`(T=|A?(YRB~vDQUc zWlH|Quf|wKb)eV1Ae8HSHtz4ob84e9&o4^p3kwe0S6FKYRC=D*hDS8Q;=3sLMyrYB zLuZIUgO`kx!BC0FE#J;>y&tuPR15~j4BpcaJv_)?Z(1%$uJQ{}owy*Qb^WgCdCz-i z!J7d7L%IKGklp&tvtQ>kA~Cgf?g5?73D+@g?71}2vm1bG33_gsBF_wz8dOGPvaq-a z=Aq2UYzKrke~^+V^3NY1u^a3spBk}J$Vn3(uGZhr8;vy3-Qp{krH@tx!3FLMaBkb6 zRVbZ(R6-9Rx&^3ZkiGy;o-purxKm_)M(lZhY?TOa{fqV4FONj~F4fMyUh7cTo`5(V zp13du|9f)`$xqMM5XoD-HB#o;nuznt9C)9!E`N&0bm6botG1u^+Je|}<<@cpl=`-| z*+*YF>0qp>!TFWmes7JWZU1Z5z^(grgJm`%;Rf1W!R9QiPL(*8mLui_VGO|hoj zq6IV=E0oT!bW%cWLrYY>ub%!w6gAjfAh#owOQTnAV+^9`%MU&T1GrlzW;b5x2rgQQ zFiqSVW=HQwxU}4Vb0&p1EZ)%ffb8LstC=52{YwEgLV9pGd|`~-gv+ssR0L&l!yM8C z?tqu@zbVADXSj^~`bkB^c?Ph&jUR+}@_)vCy}uM1%|K`}KaYzsa4n$rTwv3@nOFAF zoVVTE+RayGd${mfL4u1gp($SMx78;f?0eK^R0@ZtuMd1DK9!f3IdFPZcl*>kbl9v+ zjIAWds^kYVxC9)(yF8(|r?b5`mribBy(DXC_5VBW;fGwvj}wiSAM7i3NsXm#5puh# zlixoj_*xFL#FebZid> ztClG@3=Pl1dt~~(%<2=+XF4-wG68Z84h;=mt%Nwjwa@#8=t617s2;;)rTOdiyiw}X z4H%U<)$Kgn+0=xdGR3@-Jof?}CK%)I*X|jf4W*5IPnl0Lyr$jsVz-c3$eXmLAWA>6 z@78^Xu3ZI%Fx?QrRD8jKC+n)(f)e9@UuWvYJTA}RW*?J=hBlEf3)lT|Qpfr48B%Tl zUg8|ko+-rT6=nLfTjt&tEU1b5Q{u4p-Ns90-<@v1``y1Zj4kmz-K2>^R>auGaWd9( z{DaSNCf~|3F=EPhz!Wj1*RDg2@uj8<9NaGg(i#8^4X8^lHCIiYy%ELO4Cf@IIiVpX z;N--u4wYMpryJdAUStbivTG2Hhn@=R7e0Mx!*K1MFbF7nkShd|WWwh(`18~_NFmzL z1z%Vb5D!|Qt+~X-^}INE=}hu<^P2&R^PVt+j_Z(2IS$NoXhvL{C@UtuK+h^c(Xj*- zJeS>*ZlIr&xNMrX#L6p={rTu))4;^fAFRy_PLFi2$eu43#YzfCOJ{XAbQ3Ie74i4jN-ghm^prTiwS9STpR#h} zy4!4jgIDY;u8stC0@m+wZws1hUM(&TC%?~B^`5~Hg?@Oy{5fTE%jk0$*3s+XJktVm zMz4YWxjoZbZ|~Z$_kYr!RP>|$$4ME-H%iTars(U#JWt`um>=;gi`6h}mSb$>O>X&d z{|9{+IG14phv5p1T^uS&h!CV3!jBKjDTpv>hGDMW{$*;NOl`qS z9Pp{6a3J^zt>0nBLkAhmeBcui%d%K~1t?RGh6=PJnM9(&fen#7@yM7j;V*qLBo2{V zzS5l^Pi1%Exvf(4dR~F*A(84Ze-rB}Y!_Z_+tyWUatFOE8f4{d$Ce)!?(NUBAImR} zlNeRNEOm(+2n$})ec5KS89%Nra?*IHRneMuFG$=A#S6ykhq0eXvA4_Wh1HaP@+*e(VDQ|--Q zQtf_#Kx6?zOtAsGN>dJ5FP%}yfQUAbfc>=mK3zCt^gvRCt#sw~xWxOev>~$?t&7Bv6j|>0!+olSo z$a&}NKQ6V#Sl`3&><@WaIx|*Ho%twv_sU8T^cT1Fjv}UQ-U9aegsOO7_%kDp%{||H z<`dt9rFK8-LQNm0J*nx-SZ`7Yjec?*xM3+i$C?}Uq_2*SiBT^|_cX`}tsaTq(SHe# z-CKn#WThISIa6R2g(i8sy5l`){9xY!g%=bK_dTHh}bN^jXJ;_ ze5;fjmuMcrY=zL3|8wgBo+tnJ*2C)YMK?Y8Uy4?8gj;UiyZ4BO)_6g&xbYysQZeH> z;n{U=pR~e2uEI8F<$*F6k84S^0_7CK6s0q2XmD8BL`^x#N()Xop_YJUgncU)zgtP0Q)1c6p+7`brOr7Fd&^F_Z$u?$U zwYeA^qu#F+yAubKB_S;~P5bL0vj<|Sua?qZXxNOEMoTR?;U z7{)Lsw4jX7{1Xg~u}b5OoB_n}zie>xS*zpQ^5?%-iQ|}fHW-hvF5rbHKsXg9uS8WT(K#17P9r6$<2GB2UiI>af?l&(z8&w0IQ`7D+DT}JyZ-M-by2XOVi*!C!3k9V&wVy3a>D(gP40&KYS!PNf5o#= z6_&pru^3+HGugK4iyPnXr#&M9C=k7PbrEap#~`+N1>)~}oNuaa8ZS%K`f!pytpZQ zbpnwkyYhptfqD742Sts^g?cIbb4SS=kF=V39-KRR@ym`jI{C|2L9rhz4^93C9ssRD zNjKpc8CNdEaZJPbhy-Yh$eKdnL2b{0 ze@ZbGi1NSR|2w?;cct#{Gc?$utN>)u&6aVC;$hwMkrtwNve*$XXiP|%!zY`E7BtdB zpy+4DO@6y1qr=-_Mm-Yt#zk+-e2j#&*98_r{9UJ5x<|}9wlanfcrD-Gih&RCie-BD zOWa**jP;&H5hOAj)B86%>Pz-Krp*6ZCP@DC@g=v182*>rO_{a08-V1`^a@6Fi5oO3)+Bg34+h*8ivpFlq?IJ#nJh{df2!4Z7S=>>sk@HU+q%=rQ1IA4Y!0$8$>4VMW2VD0fH}D(jYciI*7)s4` zrKPy=v)ARJb_$Lzwf?q8W4BG8t>9j+&$aTm-k-}NsC2o#HTX|x;++*E#@J{tga()xJC# z>v57WR!Nxl<-E*ptGx6UAU{9@Zvne&El5{uWv3zpo>n^uKNw{EBy)3fk;fAa4b4v6 z#?vLMAbj!#SeTt$pm3Xk$i2m`ipdKM=BlG-F?D#0V?(AQA0oc33nvd6GjDZlJo zaBqGSfA@OO#rc7iE({X)I2*d0QBq!-MMSwJx}0Nqv~3<)u5}#SvMdy=W*?m z=G7CXUNRNcl$ReU6&8`$6|LQ}a#_iDe;j_q-z!_;LPSi2R9RJ^P=>#n51kZG-+iuJ zFgcwDFvE1YKNtK$V50l;dlSs!Af!wPdLQ~qU%+SJwk@&DIh{S3?+oRD{xN1RgF?19 z;U;c@6wV3c2_G3j{qi3I3$kZN5T$Ox@4H$pm%TKM1b7*q3k;LKi!2M>8LFGp4LTVy zHLg6a>z~zk_5R&-ndR^qw%y(rWL25k-;}ya+yB@i8T73N!+oUk+*NyCi~&3f$wjeM zZ~2M|PT_vNf2rr@K|UYm2FGt>$_NGMmK#V-`Y&@+ln5wJ3DJVoT0`PL?Ph{jY*i~LvU1((b_vl=_Bi#DA1t}OR=N;!F#&$!#jifzN+>j zHu<;CNo@G~%)3`kIB3P$M)1Ci@%?%)FE03OD+yEN@fOzjpDCFc=$g#o)={<4{L)aK z@jQzRrv~!W7kM5Y5Y_#ezp59)Z#!oGb1?rCKtg|jU%3F5omYbJNaIZIomA9EG5-Tq z<|OtBMcra+1YNf$q&j+lKJj49y{^L_i}x-4x1k|)Bu0z>ywnZWFldyHzs1Iu#`Rqo zy#M@;S02Kt_b2RtT(bUd*psMm%q!t*RL85t1^;$ATp!D{t{FV+gB>@lTC(pDnB+as7Pv{Lhl zK+svGVf(X<=)iqxQadTyl3kl{zZ>aw*q0Rz?*>O$q_qe>@e4(;BwEHeh>i#KsO)ek zbAbb76>K-fe!h7n3TU(^1hP7~SND1V){B^Rv%r>Jbgw&N0Y>|y0Y&_i8!5MnYxUsD z$(ZyEs0_ep1OFy(L{motRtqc%P;{MMq&FBhdBVkY6qN6#1BckQZSMitCc-YQQ*hE znfuprD2}<8lx3hy7gx+HHC->E;*arKVp~yWK$#GooJ*cvK_J7ME^R&aV)t@uyUG4> zR|EZFTgT`3G;eR83jKFja=nnO>sCl-Ehlxi!tJn+5|&nL9%Q?AyitBEXRoA?-54gO~|q3_md7%Sw|g1Ud?t2tIa|NG5F(C;wuO_Zgz*58hOJa3~>@^ITlLfail*^*q^rbytKb83L z@t79WYp<<_>1s9E5Sd^Kj2TB*nFh)Jqn{Tf%TmH=`am{4afAsL38~^1mmX3&GvzC5 z6?1Lke*Ehu+0+c*sf>>#WK~K$0fSrQHp3Rl1^g$31;b|;4Stj%(8@+bqGCCb2^ov{ z3&cZ2s3zJ4f(C+;CgViSq3M?a?{0&mMvP%{{A8@^S}4BE2`6*<9!IlBlllrUH5G}W^B>s}P55BiPK9t9_>RYz_u+Mk zhb1?P_S)Lo^6J`GJ?UfmYX0}+i|H4Nf~Y1$CUin|1Nce;pdJ1T_DLeHa+iN0U>Piw zrY&TS7{J#L3--J7NQ0)GP2cDGV^b7vBh>Exk&%&wq`p@-ouT^?1g-90z~)L}#U*{P zqd8tJ&bac^>W>dgwy}J~i^wmX#QPh96jH3kvO=nI#$wsuEtQIL3xrZI5ynYfq2*Hr ze9kqP0L=1dlXI_SZ+7VMC49+93n>$Xk2v&ORhMF-+a_84H7-rw>ZWt}Po{>;&e^XE z+g8`h4L1mBW@76PH*b%-gqJSYYCMzQ$aH_$zq5w>%T>LnK*R@+#+z+HG2hvZVwdqw zwVa5s9Bb;uXwg8pN3byN0do;u?Q4VL02P4Jj6Sd)g<^Ol01>gd4 zF0a+;AzT`*JiAg`QAA1M*AwWBRYCo^J+=0XX{d!`d`xL*oGZaDWo@b!hh3f6)_6a_-)}+h?bUIMX|Rv1H#A0?ZonkA+*{^izVab9eGGpyaU>L9h$(~%P_7(&tI+8gb2 ze^}GSd8jj^bj8-#C^JLjH%llVme9_iHm&X7K_OL@fzD7X$)2uBz3m=5s_s=nEZYYe zq`dexhL7HRe0?>UpFLUFd{;>bdndMgnpGZwYsf*DLD*W$?GF|{^L1#t%k{!ms(IJ! zVYvvSa1a7@!FMJNvV8?SHl~f9Jf!ee-a6?m z=U)8Brei&%G#F-poGj#hRS7(;1F&oBfm;sNk$?Lhjf25aGrB;%QIBZVf)->T0Lo%+ zTa|yIf&K^(Y>2eYi=MX8f+RsOOCd!tG-M-d0uEqzyA6sB>Zlu4Ks7>goCg(tQiiIe zyTTurSW|m9W)yCmF$tENI2cq=Z@Y}|*DH4qt8jeuzRO_TcrU*yvB9u}r-g1c6{dtq zTTrl}(|`4u_&6t?QFQ+J5 ziV8+wNE#D^m#~r=tCDmK3 zcy(cuuV5u&vC^&_Vf+gnJO0&Z_|64@XchG3@Q2F@(=6nn=^zXllBezh+Ulqcl}|9S z-arpR1Yy)i6(g{sV&8Pjf<_oi*$rpG(UnrjvQ|=2`Mfq>$AOpDy_59nRRf{t+7S$} z2K(0&Fu|-t1s#zrebD0pY0GW~v~!YPFv6##SL1TyVb%J=&fYbA^_F}wW_aH_<`0^+FZ09L7|}=%rt=I?V0^JG zXwDE9enfs+?_j7Ob+({MSOYq7z)EHe`RA-(s~|KN^_Rd&f4e1~=Q$$MA|fI_oir3b zplUFH=>raQ2wWUqDgE{9*HX}P8@|f^P`4J21{-*wvY`$|e9=Hz0y8s;hYd-#VGdj!VgR0>SwJvH-{u-pLBCqJUmFk)ubHkdsmwmKe*4~ z(SDUCG$M$<-U#3K$|>>;%Sp}1>^RY%u!jAyZ_wG9$&L$oRlgpQOiDvz%P9E`w*;4R z=E?quGQOP?_hYjaK7#618QD#1vdVtWQPI;&s_Nj5lf#hT*VAi#S3Jm(es!) zaAHkesM$S*WPyO^(!e{9uJpNbHf)g-^*bKWWuySw9O$ip2{+-P#$cbWpP%2z1`Ppx zyGfA8jUu^G)Xp@32jif>xl9~NtXJYS;4eSmgH24M@j{LJPPWPRX&8#FGe2_4cXz4c z4%EAzjoweoHIMZ-STd2_sCsKQhxH+TFzU~iOnn-Cq_W`6IFCb8nnvR&V@`hqOxYr%VcG$8A3WX=EJPNMMo?W<1>Qt3v1w{LUYyxa29V4jSY ztH-C%Jc}{s=%^aZc0`+US65|DmgO2p?3Bkcwfl7UT%Yo6U&NYGvX8+c>3Py8kWFVh z;?Tby{lPt5i>4w!KMd>+00qiIjXRnt)YOWH)GbHKqjR-N3^mBENWMUAKp<#uoLxVr zd^cR0-Wi;*i2j3!X)C+$$s#_GQN#~#uMzN~B>Ct6{y2>PewW7ZDi4+smbJ#FlsnxYCO{*Po%g zdZlBhz}p|#!{c#S78)Uvyh=>`A97=+o@Fc!nyZx+X{m8x$*%&Wko7B@jx1M-i|$#d z3a(D~$2wwedB(UGlLs`3ljj_$4Le=03&=6O4lsk*jxVDWCSmAXArK<}yrhq-VP(bX zm>oYD$yys}9r5-;!}^<__f)*3#BP4XyV&kzD~P=&vwvHGGbk+%m)&T*ZRA(uC6YIOPZc!KHMd^FG}UTR!In)T44YzLS1R^ zOeq?25Ksg&6Rc3jOGZeJeyLQU_y{hwqkn05AxeclQU-w*&>s5}Pc)pdYu`wgQbEn7x@6o?u{vovtYF+Dja%ccpO+a~Jem8aE;lPv{}a<`yhD|| z%yX8@#q|B$cwB1uT;GOQw?XdQPp!t~F?NMGbb{V!X<++WqcdyxFN3KIA)mNemuzTe z6qXx>+ONWY%r8TVE8mTUUzeXRMl3pB`O`{AW_|sj-Ef!x>dfy4uBFdcYZx01a&(mE zp+2jIYPrhy(!=K^1ca{+bJxof6O%YG(66%h%@r64D|&^f+CFywfjT>(H}6|NzU1C5CM7??M6epU zI3H-@K7DSB@|~S|;DK_7$V~@27XyTu=T)rmgG>IKwjvh`MNGUVb=3Qh2#bJlcw zwIQVXg{Ye;!J{6bg~(y&hjn6uH8lW7pjN-KgfODi`(cmC|Gp)!j(jm3pb~-7uX;Gc zH9(n#1lI@K%Y(Kc-bMr?@GK~zQWcC)&}#|IOddJqh4 zd0<(k(=b`7)%D!z$G_ez-tX4H-vLcS3s<`|t9Z^KXei^HC~F1^SW-o=o3GUwRFQdw3i0 z0Sf?43p=O^x5u77y_H-J=K+-W87ndr-9P`8{l!wFuZYj-t@8fwJ01{wNeo^pV9)SF zHr6{oaRWj`%UN)e<=J@G@3%9Dgwoa{cS}oSQ*HIyw(JYcf3H%%UeLr1mgLkQz=F@r zhAFzKjrU-G490U)wa$+v3k%tCiHl*<;xY_I^SHiQ2mRdC%>`M5+qc#q;Ge(_HR;Gd zbh5}xoDG-eo?vjwi#Ib2*f`_2(eblQUEm&*Fm=79-(FYx@W zD=RA{CB+N)KhP$(K=~8jueCag+k$+{slIc8pmQT7wE?7%3j#S&rZIXwz^@-DnKoz( zHknB{eSKLkd zvv&zh81gz|ClsblND@l)$;e39Xhcj0l(ORE&T8(5KGPk`q<`Pr6CtQ-%_85;4LJb? z-Xjt2irgQ7C64BD>Pt1x0D%Up57XT}`iU%}j9LD-qOn>1op1pt(Z)n)H>pK>9&S(Z z->+l;*32R*L8`zR?kd=BrOeY^Bz zBW_vfpf+8tj$gRReZBc}!FBhs28V9e*1|8{33^do{fP&O(gA@dwl%qXm+|^4_s=Sp z_3-m>mll`unevkx_)JU?LwbZC^o0l4`e4tx#=Et!bI z1&Sdo02Z1lSSXu9RPUlBIpnPYfE0DHA$=Dlk9n;5{us5R3V?G}A-K&kLV zi}iq6zB^ycU`dI5Y)862jZfkDn3HV;OkyC@9dM*cv7w32aLT^wvRSr$C}cT9u|2t^ z8(7tc?Gr1|h!Iu5CQ(gL?tLP@ZRnrTkvi)}{~ZnpFY|4Xrue=fqGkCmG5#@Tm>vIR z7zLfN|5~&n2vwYAqEz^pgBJP3-@qDer^BT=BK6#@+gn4TL}1Q!GiHV@Z?U+$OWW{i zt;Z>E{twmHW_VNYDv?6e#34P*8;AxbM8F^P{DuZk;fsx6FFXZ+JX94)k+Uyy|#iP(S| z)F`nYvHtTL0&1Gj1G2VTKvtK5mJePK(VH#z0`tH={(m58%{z77Ks)hYz8zFRgN_ei zz(H+QbDMne;5I~iWfl}fz}kV~vmJQ-Py-aVsxVmsZc^ji3i;`vPXM$Kg>#@(N)(hx zX~Yq6{yaVNO(Oy(Iu-2~LiT6MM|bohwY$72-Fu3E$2B&clpeB@vF_4Sw0x9fg6ZrV z-l?)b!xggX_ADtcF$*+=DM~c{oxcqtoW%3BF=x1ynKcEktK-4gec0g=4Pwdt zgH)Kccnar7?x}=x*t83rdrdads!Fq^fDd#CfebhG&-^wxv*RFq2<4E#cJ^(H+f5w6 z3%%bPuWqe#cYs4+o%s z7V6uKphpj}($Ghvs-Y1_{Ysi1d1;Z+9Z{>Gk%2@$RlvAD5STmN3#Ywu&M`U|Z>{%( zXEg8-B%uv1kTtyvZ6XZ-#dv9{ih0Phx_)Eec1O_03!6ptkkqqtSop@Li=RAl8URaN z;n=;;3$N|no}}B5blw}V&5W5Uw{P#o|4<}t{ z8pTQboi^%@Wuf_+Qk5(y76#Fdu zV_zZ1JQibZkdV@I%6Me^xMO9j3KJg4%_WqWUrb)Lb;s7*W7Uk;ikE<2JFZLj=9S{d zR%z)jnyS4mzlq!KVT}lzOmrTO3w?_&y^z*7DAJQ7Wd;EX7%STWg8&p88~~us@Zt6Y zpS(r1+KMaI$?9)Fi=0ZQ`-KV+!Wfr&+pFtEh;moWm#8F(uZV&c3gE{$w zKDjSNOVugOdINnsnWklrzX?`b`RC)LGv9WT(H=WRDiaOAf;|``Ek#XAzu3*K#+Ps_ zp5oH^Wx{%fYrN$*5iZ^k7amyPOE#mWs#K+_w8T`demZvLEtmd}JBa*8fe-+bM=qM3 z4t}VCP(SpDrzt~aY1X@UmuW&v$omHdz5x*X89hI0eJE}{p$EJ1YUEhs(YS{MtuSRH zM%G5Bq?d0PgI1*b+P0_Qd$cTk-rzeo46+$btG`m&n!n`I94(Toy^U0ufK_}766is= zMT`g2J+#on-x~=4(bw4ym~ukRA25ip>wh}70^1%hZ@71Pd1C9b&Y9H8{0rG|CdKKU^39JjDHCXCAy17H}j!jQ7|nTRnAjlha3XxZgOh zqTB=~_>$}J1!u$!(JWEiTlSjpqm@MeMhDjD#6BdcRAxe z?FdDkYj7Z`Pdyz&O-t1#wwm1Q`lpxh_^@cvW!jb*N8L@G_N(JxZGRa*2i(p?KYYb zA!E={c5~)ns+z3L(!7_Z<*cXM!Orhwuy`%JriQg(ch@w-|6v>w-GYA=xn`vq6BWKS zH%~%D$q!k*eGkWJ5O9ihWefuoVW2fP1hgW@AW*csU9YNp zW&rx8;5H>fhVEEM#(@+ZW~A!?<-uc~Z7?eQfcC*3kbpM7l6AH-7j(36xT8qxUlSK< zen6#x5k#h=@M6Ha!Bf&=~2NQTmklVW$NSh@-N+V^0~mdG zvS-QZLCo|WdVA1Q38N|c3s-LMFVaugPAu{gK}v{4#~r#WviYa+0D^O4{^8YexRnSK9Nm#4LJdbZIc}ER;~Uh^PF@ zXo?=Z`_YYjZIS8ZP~Ue>yFTuhHgBKVygk++lc5za&`)=@iELvTA@=SXfw!Ys@3gV-$Lm>k$75`QSm4X}Mp`;=#)MT0}#G0*HT;q|z|W zz>(sI&16XGSyZ|VFVzfUPdZ=te&d+V{Jm_kKVe5AYA=9uKoCz(Pxic6z#5DIz?QMYl9{E@`2~3g z(1UcQklOg=((g5FjPb6!} znw=Fm8NL=+Mt$gF9h@8l^tj>M`*W#QByFS~e!lDiZ#kbbxeE$To8%byw-8~>oX88h zr31E$d$va(&l+oQdPr({HEw72KRM_gy=~}pJc*y*fJJm5_{V;hd2-%t>YB{1ZA9Zq ze%Eqy6LOu%$+HyYc}4-EcR*7KWbJ6$@1Cy8%!e#aO|lik5uOqJkFfDUFh~P}L4Q2t zI|4UIEbyurtJVDw@k}UVz}NgbY5VWO2564;>z7H}+kbx9q31Ln`5ib*zI^%eK%M=> z`rT__(n4T542r)%5NvMuy(2}eY^k6bEGyHH(h_CVld5+BF^%oeI#=-YrhFu_mBJ9m z6Hti2WO1Tzaa3@?q@I0ABNPuO7l+6G&nny?e0WPy=ODroVR0$m9fS-vXRs~bfy=;!y72U%ct+<<_6@1(<}M}qKbp~lL}t<|TK;zsqi z^`tBhCEwHf1V%;a6EZbz*}LJezG=XF@a$Rz^_hW6$8<=$I_3&79A@!e17A3J^yS;;bg_?y21FdBXl;hI3 z;eYZL!vdwuGzmGa98v}^rAF(P9}*qDKFJ}_W0)}f^ACWSOL(k@t{3A!HZQv}yN}rh z7VLT07d`FZ@%V83S%$UY&?#^~SV#E2`n0Klk~wIli%A;@6wThmr7Y(E=Nw^TCb ziDbt6`>tp%fgS+c*JP8hJ!a+RK3pCu{txQqjxR;D3KC}^fES7B!9va3-~k{O+RdJ` zl?PT<3hsq4?^nTs7$41YA5*%FQn#RJ4ILveixSiS<*Ca%;La&usz#AoM+z9IYs2gh z!a7RpCD-`~90sVC(lzZOKRl9^qev{sXXaN^ ze$TR3t@uAgR;-a-hM5XDozXkN%p7r||5oxMKPcqb&;xO;s8R6c6Yt;4#lnvr|G0Jju1vc%Jl~NHG@vzA zgPjswBQn|zv}b?D~+C|>ob%1w8{kUab}LkaJ}%e07| zs}`nf3u`v2z3fZxDQA+diPxKjEXH*WNIFFApR|J}PlC_g3;T)Xohgs7TM&Cw>w ze7(6zedzl2>r0)J;#tXM%@C%7C_r#hJhmH)N=j-61t)T2L(KBY^1SBt>$K1O{ZB(M zT@zW!OL)5gx4b3Ab#zE{d{aIThXmaxla|;{lgpdXfht8{az(toiyMC!18N;RzQy;2 z=*vfrZ?yD4N6?_{#|JMU^czW=Lii@m00mzLGeX6o`5`z6fk!ar99*&cPa;_jixZ05 z&;kPP7e^#-hOkc;sMwa_>lg#6q;Qs+}y}mJ5=yzui$vl zOKKv0|GOn-AWM#&Ys-j!q?ZQ$%HjY#FX)erjnsxiW{UuD@Ihg>U744c4=#O3HETj%UD!AwH4E%?Xf6b^5QU&~{t~wDKqHXplj@7@Py{9G`Fm7v_lo$3 znCfFy?ejj05LIzFB^Qc5Z+++vF@dnBhw%4dwH~W2hu3^;HLNjiQO&e znkHk@9$6NV!9szA z_L?%`oJ{JrW`qT$e;l~18A~-)?}p2j&^uKbtHz%9GkD-DdMY;NO=N%Go|yUk(|e2| zZ5iIN`_?7&V^;{_OnkIO+kC51369O~{n z70Er{r19)#9r=u_!9*$psPf-*1LQC^df*Db?89$P2{W)xI=zNCy_Vq%Q!z7fWo9SW zy+zLF-)AY&o>rOd>G;t%F_Ap4?~5-Pa@_#(D^RTu(D2i)@u??b9-?vun5gBAjEpep zEM=`3SM-`5*aCA=X+|O4R2akKZ5)Mq2`$v7OoJ{BJR4P>N>=p552>T{4iK zF{bKgdNi0pCFa-!1{xe8S8I04R1%SXXXyS;eOOSm_nkSc4DNQSR>{Ecpkag~A3o$- z=s0VFB(!E2>s)~M(>h6SVQ!8JLdxK|1snjH=}M0ggd(B-~1e)n8X90OmP06$2TfOIr z29r&Rj;qdxNdrzpNnY+{7B^RZ(p^jSFMt>MBD19I_T6jt@(l_l^n{{c##>hEr5s%2 znguV-m6J`orMlQE!jW6?);wblhopW!ybYh#=(0_QO7LG#u?R$zLd~DZ|K=&)&ATeDbStC|ZbYr6*|TTQ?xEzu zIt^s;qYL~C|DV*}ys<+uGFUDdI0|0v>h0nW96;BjRGtDjUE#w-^=g_g+yJt3fd`Z? z9;i_0m5w6?aLm++FSb0A(6kQ4NKw#hZ`Ubt!4_D@63Wa)egH2X0UXrzIg5}(J`(v! z;fqEkL+Bs|;id~^s`6{7C5}$uc#ezn>cwSldvaH$98SkfM=hS&2PV2@c)`rkuFk!c z%p|7eiMgrtnwuV>;(~WMDe88#3s@UF*;&sDLmL;0-e~Kj-Dw5+|d3)&aHp?*H zKhG@LY=km!j-3cpXgl8bT1j=};ffvVt@e0zv1>&~(J7P8Bw!~^#o)d6y(Y`H4!n&e zchY~{y>M^5!1&Q3-zo>`c%t7v55IjVYlq#xbsXaHe$+O2?pF<|(!jwBP%K*`gf_f^ z;S$l{2<`GGyp*_T7R;7xMhRLZ6WCKwmB~M)1_1WYK^lz|fG8NCdUxXNk)RJ&A=&_{ zuJJ>b^antsPESt@ZVwnOJ38NjbYs+KipG=_CFu||oK761Us|pCyMa_cZa98*9^oZWAwK(^WBNOL#;aPNJ(2^c+#~F>)CGP z#%mFJ*EW0uZR=OU1INpPEVZ2cKC{bN3J^9EU{4e`tKS&@!kJ_)Gx~o2q3hrpSF=>J zoBT-L&~6nK+WxO3(>ipXHD&#M#j)TbYoi@4PnDcQ^Y-juPm{I$3d#gpCL;|sZ188f4q zpZ#O2A2*KUMtq+C;Il3j7@CGBKzuqI%#x?y^;SmZV?H4l-j!i1(9u28ioeOrdtB-M z{RWtEf)>lKKgfRuYmp87@gUHf%jhc}T&kzvZc{@<7QlQFz|Wv!Pu@$9JCa4f+~I>H z&@V4CHpdWE7{I{&3xv3jZHGbATnmEZs{i_t86JfqD)*QhEZ3Sk)cE-z&Y0Eb6+M~o zzX+`05P>jyg!RKce_Nol*o7>fQWVCt`EJ@`HG{Kf#>vIyOW zZ(E!$oW@n1j_LC_JF-9Tp;hnV#_ zA0sXjGzoo_%IOvKMl)iu-0Lc9SKS=e z*@NkbjCn6hGn$@OrKe|%U7ONsv(iR$X|qU7PtVc5@d7AirEC^~EO5Whfm-QjKA5J# z@YW7gs?=Ze0xUl#`yIej1G6NWC(9r>6xj<=AQ7xjQ^=zO6RXwe4%C1&BDoE}12alO zE1i=M92LJN0fnFkY z2#AiMx-S%m$IZ=c12519jEo-*pc{%>;N6(A4-Ek^JG8~djQH%4AXIZJwoEfTR z!WTLpG1qqaUI1(3OZ#5!TV=&u(+Bh<$`L{2iL=M#MWnimaQx_CVp=rgEv8>J9@co- zb|2FcGAz+67p?oX{Jr_i=zD^=mIs^JamVgvo2dA6i$&)(iX#ErwbA(^WxmZlf_S&ED2dh--iutL6k zlIej0phOZlFViLXx^1aYTrfO-ysL~<1YF=gLMjVi zoO7D?fyAOqM@iH>HGitUSdAU7IPHOPC^3!V=%StbchUBM+*_m6#vnmh0IWoA&b*>O z72-3%UxRF}FnvuRHzgDtjl;+0^gNheKmhnV2!gGygY6iKh39H1o{zqT*wRafb5g(N zq=YAcE4t`3=5Z4-a_i3CyZbNedDZptDW%`OCAawT>9#sgVN&t+)#R&X(1_j@QAv%< z7at>&_18>|rEv#MP!_fenof`m{4o&6?8K*|LfdsDUkCmA%C-JdP< zO1$Ym)qLn89I=^O;u zrE#0se?fr&`XLcD_>ycYqS>)5hi8ChoP1UD>nhT#U9NtN)0v>AuD(%s>39AYD^PAD zjjsf38|Wq+y%X~62a!&f-A2eZcZ zfX*n}>eCv`H#mVJLgB-Hbz&e@48fV`0qqs4ULJ?$?nkm}=U29ud(`fy1=RUU;xPjA zY@k(^xT4!#&azwjs$kLVzI={P%v6TiU9D_K=R{e$(ZEO~D9y?ay^VrvXhEgmPyDQo zm|8vCvMY5cu5V*|iF!GOOpQyKsitb}WzDb}SMk#Ka%Ul{5_1MqCL9w-X^FI zz;~2ju;eL!jX?q>G1xEV*~NJkPF~wPZ?PQuHSq+?EJ4L_;Qu zc$1LLoB{5m_FCp^QApkf?n49%j1{)PgrjClE|3KSE^uqP2pAew1)5+-d>X;n zq~AX?k3t&J$5wr?G6}^J|EqObupqYM8EDd061!CxfBC3iF1+^XIvPBM*CGa4AsAE_*CbzO zi517t@m)ZU0mgYw;k|{B4u0Vmqs&WBX5B1DyDHQ69*Q20@YIYvTp%;~cGr(%2yt7#Nw@?4roIor= znwBC^kKchM0F3W}(kCRfM!@}ju;HK^?D*Yllou~*0O1FN!3^xm)^^{)*MWAjpoU37 z7#dkaEL;*BP;nk|k8uGRh=ig5C8N;#H6?V=DAMnIz0Ee#f;`wZkIk?6C8orD)X zpe03VuY&9K=PWSZop_c^>`CK38z97l>QI2zXQu3%c2R_N54D6r2 z4hq+(7-cw#zsP*;{*VGmbTLyH>aPS69Vga?(nGKhbS{2w!CYS*+DaD4F>$;I3*rS> z`d1lgTL?~NP?`vwXXw6|r=woCloe)#{XvT1byiWx?05dkq>A!5N=jNTxo>!r*Ft?G z&CLjc-zI%pj})gqO1U?{)5#8fWIDtK5u!OJE29S&r}#`%2a0xV??nt5(?Ca7(q;y`>eoN_=0b$>VN8b!g(tgu~5+?@~gw{@p=4S?>InhFc1jmTMlymhIC8 z<J#xwcEv>$fL&T+68k ztFjZb)>_L$^{^U@0zVS&NfKygW%aEyLPdf`h0{BQ^x+|M&5l%_u(dSHbx`mY)bv&@ zHYQ5@t#{QHkLa>pg-bI8bN|)=!9BNxteqU3dE#xH)mPo-u~vhm2i~#4oZl#$h%B5l zpWZNSPvauR?f)CXe3hE#)Ta(PQOPsIxTg&BM51s_%6B~rcLd)&S!uDDx%LVd{$}jF zPoyuiy>8rnzb;%_x^$63_5q=%bLair${)gk2C@~7&-OQ4@V!DC0*R)1Roj}7e*u!x zVBI_ax7Y}R{;{paeuS(d6cDk<@JogYC!%B=n0m}=uGcG|iV+&M zwx=bzll5>n8dfZp5LNX8Xu$WgL#*(ruwy+pZZ7Su*UCHZoG_bBF=G$^{Dv9Y?i+2u zl#L|1p4-^a6@Z13hKne^7Fi|00wVW0Rh=ayr!UbzGQ>&)KqLpX9gfwadU|@tA(}BY8#gWa*;*j6uN0oxO>Zq)>F(}kmkzeo=9ZkapyJ^O z?K*b3>t-^e4C$aKm82uCvpYA~uav@$cW$_?p+L-Q;Pke(>;_EXZ$PQ4)O5D~;4+{O z{`Xq(*l-LSnb>>T1!nei1BBbHoNae#3PB&Lb+5I>3PLODI4!T&eQjWzrasL}M z3_U7j7>ci)$8Ex;oh5&awmcJFdd5<*8 zvqpi#K!rUKlJ7kG|1B9p8GWct0&I!-s!3aLY3d>252~Zy{P6UUWmw(>?4bWc-%~yP zwGNC;Y8;C8+1s2gDP2aiFQl-v{v~uehfmIKNTKUIsR9)!KrMBGa0QUQ)7h@HL$>ho zlLUzuJiWopoCcf8FgeLov~9zPTF@D#&!5>k+l4k?LQwUSX@G`D#} z?#chn5bBM{@H3KaKsxfenG*sYSv>8F8$o_@`U440)jr_50>{$!_R~8yXum-X<>3 zm}P%odb1`c#e#bE(&ggit;3ANhioOEj_6=z=VN{;qSx!hQ`Xj{cX22q9D2Sw63I$B z4XRhyG!}6a!q_RAq8`0*Z~t>#$msZGwFXPqr&{%;C+grhDDeU-$@@@G$(JS64_$oE zioV{>JqOqSDe&L1!4;h5%OMfK`9=c?C?WdbD!(j4r30jzHt=T;^1eoNTBC3%xxy^n_D-owts>aeJy+Wm1FJ8`$$?btJK0zc zkGv6Rd|CUb;oT4o4EdIM8vX?xQW<$dF6IpH(`}Hj$F9O1CnR`<)-@DNXlUqwfgIf- z5c%%;{MpPX;LCs0LBLdtYti@K0REcDunly^QfYZPKe{u3)9je$JMvR%Z<)pGX84fZZu4j4+>D5s zaueUX9V8xQOc(shN+uC^DApt!7~sLOM27rFxs?^}1UgS&z3GBwiBUJ;wFOO6%li!A z^k6{1=2zw6ilvPil|2rO&(&Ubr}Odc4#nSIAP5mOj4QiRoK$TNE6K?_JmuX(+HCL>3Qdn00Ji$R(8NI@o=rBBws=APU2Mhu_+vK7@PeBp`3xb!Q4hlN@ivCjC0Env3@u=`HBvt2cdx;i;k@**0c~=SkM5W- z36dl<#-S<|z_erZGv>hmh#U>hVDR3X>Q=Ok4h=nxkucq52dNKx-T3P^j%tUCTbF)g zli~@im%eUGFB_Q2?Wx(h9rLDr%_GEeol9sW;ybsz#)aXx(hTS5&nas&dZ~C~u9oFp z9B?ETAjPS-=(l5S6u?+DXL9%4BWhwvHU4?OVw%J8{${_)9yYpiu(FSE2^eP7vR*v^H9@n_&J*WI^up-ayCmPu8#ZP3rPX(qXz6 zQR@{S#zuva|?GgA+7t*wfgLDweK=I&_BG z+U}|tYqyQ{$6iQ3A{Oq<N&b^x1qWUrTv1kd>b=f7jTrR1DYa-ea|7y07bBzp>lw{ z*bE}CkI>XfAoR0iX)Ki5^O|2@7**Rlw;q9Em-L13rLi$U`3Y`K!M`aD7}mE=6yZe| zfIRGM6-x!YCr-W6eu4uD`A-QA zcKTimyPFJj3ZSxQcDyS@DGbDN7^_Xt3%^>0UWucKAvlnxfP|u>42>Kp4cg7P`yNZe zSBlJtkj6W2R$rs(wh4RQs@0&GMxQ{b?j(rWDfhC=S@#p-gl4U{qU*x9rpmts)Eb?`&p|>w!!T>X6m-6{=71b zcdW+7;Cl}%*cxe>KIjC99QL$xbK?meJ1BN^Xt#(Piptm~uE@@Qy{AFs9rxd1_yjnN zUW=L8!VHIejp7jgeug3|J1(9fX@?W_>0CbxDJ2e&jLmp)M(QKS{;g9c#D0Bu@L2g4 zCGtik6kU1*czP-v_?W#+p3!ts2pXS3i~&-Hr%AqS>Rb|koEk+(_WYxHFEf&ng31Ee zUGR~@FT#nwZG^jQzfr4ZIUAlDWeD{Vte<0#!C5(L?Z*LV^BLMZN4du0ZD>t|YP9nJ zd>$^LuwNMcpWKlPd^ zp}aDnC#Bg{?dKrRL?0qJ0F(sai*OB;sQI3Crz_(rK@{34#+x!;7xd!#*E3tji(lm| z)@an(aN@2Tja-wsE3pDgU~U6y%R@eXbFP>WSK>_gPX?QoXMIQ?_1wfEbEdR1NK-6_}l z?@Te)1{V!1yW8Ll{|0tB-gZUTu+gPkCK2XNECR2s#DBWu=3QU%>R|MI|H%wbG(;ol8L<1kByHZZq;jPaX1FAfF7JkDvgB-5na_1CSQ>U)kos{ATtBB`DtB^F_owN`?FvrGIX0$B6<`wzQyHTX?sYaJS~q zfpDVx1df8xycX#YuPdF5d!1zVnDg^^o=aix^1kj=Jk{um&&+TQTV)|BF7C-aj9us- zqu=jjd^e;idUrkDL(*=1s?k@JmOJdwlm)jzrsF}4&k(g%AaCD9Ilz{X*QCTGs@$oB zi;HbInDRQ2^d9q;u3KXN__lI4x4lk72fn>dd+loXOx;dtRuCV%`0Ns#SYMB;tfs$=OobwS6BB z%s1P*gCI{bGckUEAhq%9y*oNZ(7H@X30VT#f)2Q1l6aDY7WGA^G4wY-(r!}>lW0lr z=Z4;Q$V}ADaGzFPT+%mMI?MZ|zw-`tIlhlSzVX7<3r9g9;D`tZ3I1JjReS0GTIEn4 z(F|gw(94j+ZUPiBblr}F$oImHnqQoSCarYB!g{b!ova6K#^X0Q=ueXpc& za^rF673lbn=*5*daTAK=6eDF_IuXlFG7>ed^+{X%boNfBmiAesVn^W)@6tjhw)*wZ zKiJZJuQZ)?V1_HhAwOZ2$3iB=-)zuVlo+otAtrJp9M&~(S-6&~+$U_cy5Rj(j(6?c zQW%`t9+Z?ZW(IL&)vJ@G1c|ec6zP6WXU8q8*Ky-t?~Z7HKt1mRijH5JH|5eOmxT`BL5>@E1t2E9Op(zN*uVtFz1`*wSC0AV#=M( zKnD$GO^{lk^lujkrIco%TKG=g^Tknr(E#@)7U=^Cp^mOfGnd7-W5!*wN{i~mpT&hQK7mr1Z_P|wi2A~k0a{FtjJ_hX3rgjjtFo^`(Xgvm|c|$`y@EPKC zfoEp5J^&211uUE8OB^7_Z-6YBP4$us;HA>k9C{dvi~@oB9Z^ve#BjX3w){C9`(q0$ znX0O4_rfD7Y6nsTt%4L=NSFgMa8>RSKmP&|Rlh&Mn7xN`d}c~fd>UlSAR#Xd{`A-j z5}Aikw|E4uqIisPerHoAu?8CMKrsEkPo=Z3T8zmQmuGcYtq$GL_J6uq&sH%uATmU(^U3J zyL`s+^YcV82C)N<^s{R03GF=ZSR0)ly7kWX^jp>woacPfGLD@5i^6-1%=7F6Oevqp zV;AS`S$p}lqnpT1KiZpz`^2_}8@ct8tqyzpvhG#<-vo57x3AP4 zICj>TxA_k|qQCbdMIGlvZf#j2LD--iQ&i1jaIDCgMVvdTl>L zuJ{Mqr(6QL(mP)0>9Qc4oc4|FYN_Jc+-A{!G8gA z%jRjxN)|spDHM&2#tOy4felNibk9+0X%UDxN1&x zw86Rax!`STXn{7I&vXal^Up-slx5xm=EPO}^ekXgRBxC+T3IsjsvTL$s3@aQ| zm^xA>>sd9aoh7lQloYmSbY#88he< zS;rdWFFEjni1&&!q@T0xb1o%GJ z(-Eu)n_%7GFzraDYLPqB*^O8C!$R+X5^1{7Vu^AJpqUTYg^Vwy=x##~1S{DNK7559 zWfc{=@#CE^`vq)54tseJ1)+~MBgWldAQ)V>6-Bt^CP+k z2X@hU?{I40#hcABxL|_*58J=lx$jQNEX9|;S+2n4{s=@IV3GzXwhc)$!N1`XY5&e4 zC(9ulkY(H!EYf(tIk&+sZ3ku3A%p{a@n`y%UOGi>FUu%=BNT3*aErV$e^8y7-?$|4 z^TOS0NQi|(8$KFJJU?rmfI8IwM`gs$PmDNIxD}qr{+;@79>AoJq(~qUvCB!K1{4+< zb5P0%HrF}PDP%wah7C#D(Dn&SeH>tfn{Z2flxWV8PvBig%6pV<4LO~mC}0n~Cuxi2 zR8tq0yBywt#1K`@AfqGN@u2TlLf(qCEruCxe~yh!oPQJ!_OkYHgwFOhUEs_?@Dygb zJ}2A8+`QBM(g_Lin|#jSYme%26NN4gueisu3-G?fDq~Cg4dC=qO5`M^Yj7yI^a2+1q3t&@zfe!Z;l!aOrru<99| z#bsI##hj3pk*jKXPu@8=f+qA|Zpgul(Eu0y@WAj;jl%K2oc9#cA`{|ExDhdK`- zqz!t){#S)jFFV@lj_G-=&bpH6w68KY^ucVGO?XM;=vRLXdjleTEWek2>}}r$Y#1pZ zBtxj;fbgO97VSoTP$=#k@!>$c{WJU)dLHgC@%1}sK-)REM<#yw5M}_@i(I47Yo}{F zp+40%*Lh?VFGK0FATKLq?R)KiIyH%#clxG-d@jm$B0`=xU4$44@tHJ zhL}^#V`bBLhI%M+H4J?C)Blx}jRsXjHzUag{uaLY2tVaHhu8U)LcAQfKP~%&<1% z*U^zOe=1tK^Zvl&!&R4t_|NpU1f3_{97p6pSVUaM&W?8i3al0qx4POaPt1SM9f;;%R6v`HT0+uH($kwcK^-O)Lgp#w-tPQG9sOsQ%5NT2X-5vb# z8lc}cmR*ED=HGejDD&?;2Dk!jB z=zLQFLq{StavGy+1U(3|hbE%k?UEXKKB}0=6RBd@)b>R1i+^&^K^Fup8<(DL;8^QW zzjWB*_N)#8oE&|mGZLC(A~U|`*t2pTdg0%d<=yShE6C%=&!(lFvEaw1u_uDtoCA`F zu0l^_bK)yV)BCeJ;v(laNiXvihATm8ZRz5}J_qPcWiXyiS(ZreVoS&}r^%a3rqkN& zkXXqa;o%$}e8xbN?q|oP!ebX|8t-qVp|9aw9k5vHT-{$_&Jc%N;mKk9CM#Z;!c(x5>GpINvR9wO;rwqF`DoV7aCA6B$>+ziCtNr;CBFw1jAYR% z-gw(Vd8dAExA3V{VXzB)zvke{xO;ytLz2mO6PnWA{DWgFWD_Ij4xoMnlnKbbV%!+P zdm{sz9P*g}iA5=p@GCajlKs7S#WVkXs)msGfs(9|eKOaR*RW%0RQ~D-A zpaMfcaC=wZ8ZdRw{qCfr94DYcAjx^u@H=u>q6g(4qAwGsH&h=w?dvrcW)_bMGtmQF zi2y3eMTwN+leSO{RL}AsIU26~$6|4gV;o#Z%wCX;!Ayl&hW;e*!tI6C`XF^-+5SRK z#d7M&Zp;1PoHegehl1}r+DG=d_mI!PE!WII+E&Nzz0TvglWnlzb*-iE(pC(SN+_gQ zyVZB#zXw>hj$4 zuckH@W`X^NX1seQ#1egXj0$vSj;+6e4Im{bSmkf%(i3e7_*CbL`Yvc~Ln1`ofGdA5S1dA2q<-DUn%G-VAm&mT)oEtU!eY%;Pc{XolB+v zyRvDT-h1xo;6rZ$^MWw6gdOBhLUY;x*wCmZq z&iMGI^D$F#RJ=D;wXb$F}z=nhI4}9gOdnXQUS4 z+ekndgJrP`FKA9aQD-KO#yml2{MJ3zoHdU;eAE5vrMK3c{pio8_Z~0o5f|;W;bQ~Q z-|zRre0EPAo8>x#hSo4p{RU!}Qww-9W);7@MpMM{?el&UpuPZ4J=)BP_xmdGCwDqa zO#`1Jbr-2b#DxT{G!dr58Ev}-E%<$;L&4wcM{oOc^2(2e|8cyz@_Z_Oe4?;)A&Z(A z51I{0Q34I=Y_%m!jxJ+jb*uKJ$G%eMg!wCfdc*+rzAxx>>IePZnoKojB{GV7qBpxg z5lQ9W$t|{dLLS(}+#q{w@*+CMF1V{e9Ji&MRr3Rn%$m*ven$@6;ya%ANwte&GNI(g zzCm}8I04{_9#HK4&J6yrvdru!zuKAmqqKjIs?~B@->sl#qJ;7>SYDmHpB+8H$$&W{>!f8r-)_8|f<&e?AE&g2VIf z#WB^A!VzFzwWVu4x?G{$4W(VWyThuB(#vA`4j?vhE0xAT+DLVy!`_u~Ym|DmuShDd zxbjB^<2LQ~u2!O$<6@n2Nxe~`q+{n>cW2cFA1uV^YT;N$6l&nZsz_HA5 z{#r;wo#uU7&J+G_bd0=J#Hbe}u1au3dN)pSJOfn>{4XL_B8b5MmJbv2_sv-BpXPC} z15J)UOS}ikM$d(#6^7PO=5`h(C6b8`PlUFc}zjPmG9bSFW%ig zrlI$#VLj|A(o7?~6ILO`T&D>t@-nH*`^_{_)H*`Pcy^ozs6^y2yeUC$k7o(eQ7OsawWv%KG z1lYmVV41-)MS?{&pAz0`v-EywLFw*}PcspAw@kZ3)DH43-@9Ssp7{a%KRpOOe@XR{ zg}2ml>;2j3P5mBg-`1|&BPXev;Y8c@VFfDM+ zMvLQ#DklVNt$)kCy=UJnr|9|c=hwDYl!*Z-D1-WeUn%?fT>E2Wn|yU~!}T2ydO6N) zC7?1;x`_cwN`AhSj5sed!O|x1LOj+=SpDCUa}lt$wRKso7FZsgknr~O`6(Wfq&fl8 z%y!u5z@*Kwo=I>1S2eUB8#hEVIdJdLyDhnTZJ%?PGxFN$)jK43VuqXDX)f?{C2reJ zIV`Y+IkTeTZGrSE)$$7+E=`ukV(aGjnn=zDWe?B!s<6LTVD594n0ZzeF<{-LdY9hu z^wvoFiNa-V-YX@Oeya$MmM>uyu;E-?5-oKoWPG2=g>8*e`5E?J{n}g4Yg6VKcdC@# zxVo>3*}SArL&B_Gu0Zv2mie&CPva#<`cbOeLv9?u-CKAA^$gX|0ItnQ6#>tO9(0Ti z?`K^x!;+L~C8tfEK?(ar)gjyldQ7wiFlxwG_yV#gv7tN)*BL8oYvJ-!5`QYL+IWQUfIFtpX}kSKBV zpcl{iE3iDltd7;2Dr-xlMG8}1oZy*DD&d1I zz5*|Gc1h0s_yrEey>c6BbAx7Xghpp>>17v-xVP?~uYo{M8s%ts)OHxE`E?Q{Op}z=Uc~x<9Bg z=aume1!l9#mu&9I8}fPfRRy`kxBvFr_aC|fD^`}3{{%xqdBpJ@owNSrPiyy<|LmczU~DlN8YeT2m>n!OP*_L}A$Q!TR*3 z#U!-k*+`>Te~!w~MttNOH(mxg^DD4TH1XpTO^LsvcJbFcbYb5E73Z&4sByR85J&ub zLi?-Usx@p>Tm8iS+sHdtKJL$%8TZ@vI=ahyJXv4oy36xc^o{2H^C_vbn^-i0=6Ams zI&(gIaBuXi!^UJ#MvAlV@_1PWCV3CGTMB(H*q>1PWli;b{q&)|Ys^wJW5nMHDTx0t zu^IWs;=AUxsrO*)O4qSR6?CuV>MZoxgOp3v?j$AID+D;foDb2u(0FA>ryM*IQt%?k|0OW`^Pcrbo%~HVj%F(Cp zlY;o}^rc3Hc$CD-K2LYvpxbfsyLEUKXyB0UUs<(|w8kD^y4pWKAzcZf>6*WyjfiBk zI=BqKd-Ara0|SXYX|g>Cv5bk-?_O+=_Dq{o(!y~dNSrXSf+bl-#M14K_v%!O7%A-PRXM_3tN6KT;EU#u{W~JOuzW$?Fi&rsluxQEL&!vrxE5ax( zs}r}e^j^N$AXNQKW8;C&(!_8NQQvK2h8Ejcf0=8~(a#(kv!h0pCO&I{pSp&y>~3yh zZ+9N+5q80e18*E7a?JiqQ!4HaL7(WB>j&*j@yO*xT5gfQd5kH^E8*p)Zvm zV~&eEvJ052m6gwXhjmXOfxv6jfF^rHEqlM`#Q0Y zKi+3im%Z9HPns>isx<6W=DIMM+VywR{qJ~>^!F+!_vz$UdA}^^`Af*e9vfxBc|!vc zhgRWHe`n(uKV`G;gq*a@pN-RPapyYMiwY60==8thu4SkxSTdX8ML4(++A?_6*R zZh|{I!QX!^%-5Xa{`<+_6B+O)aB_p1BqTK?I_1Le*$cpRp-A-f$6avu^%KIkCDRDu zjuG!jA4I}7-V*)bbV&8nb18cDXu*eB;_1a_cwYYaT#;IBhO!eDrR7zz#L!Uk!}^K? z3`+Wa#Xajg?f#HrZOcY|a&{QEe+lvkYpLzg?2x#kaoFU+ny2wuu_5E>ABu!p>L<8V z3>R!ugigMfT)5+R^=cMsqrzsM&G%Nl`|#wgN_T+R5ph;fELz>sX`Lmh0>n*pQFS7b z2IK++#;15bkxjUc7>Po-w@anz?+=(YL2A*L3J<$Usx5bQbv2=T6BGtZu}fqvL0r;| zjM#V%fx!wM1i%5M7r-|cT`mB}M4znoV{I6-V2b6FqkK{DNMVxfLJ*Kzr-bxBJp@~d z`aMg>Tg6uPKk}<0xA`^D3TDMB$K{*P*aUgqxLbE1`R&6^_Gj&_`)3V;8YmfeBdvtO zt%Ul{F>lxu*sma>^y*lcvz&i6*Bs+tOC>L+J>GVL>)9aq)nD%{I(9;WeH%Pj00iBi zpU^PNT?F1878A42$tnBku|+x-I@8d9E`Jv9lbm#$ZaXg{w1D69rN95N(>?sy?qHtF z!_r?0Xw_8#&iQE&9^cC*MT2pHzrA)KexS*AfuBQqaF>9X*2i zeoi$##$%njv|;v8mcvR`M@?5F+3mg=BJAdyc#kw5EfPsx`KPMoO~n1u$o#Zk{nzuV zh>yvNqN&7+5KX+&uepAz-h{*F6+7paq5(Pz<;dK^`R_*wV_{nAjp|s_eRcZmzpGl) zn?Khf%UGH%_rm<_AT$F6IeChpM}lr0HlF$z7aSa1gbR{O#YYaXY9Ct2a(6TQG;T#h zyGvS*+cPNgrW9CjI=J?vRP({Uj-|ZRO`(a`tg(Yk-g@W@H+JvshE$o{Wh(|lT`6*nklKzXY&Q+1>Y+53x4Gh;VHX7 zuT<8fwsA6Hf2-o7l%mNGJqlS#eX%z52_vKUJDOUv znelA@1cps)sz@&CPiUYVTy~L$pv!9{i)fy60;n^zb^b#u&006EXuc`H{s2_874Hsq zu9RthUnHYdG*ql-c!t}Si$+6F^M*I4k?4dH5CU(fG6#n<5i+&qVMD7gT+-LiuUu~; zc{AVI7RLLAu-PF*c)9TuJbmpnzkpT7kJ42>ZYBlBz zo6y!osa`hHKd8F=aT2$KEN@&U?GE)*$jy{vh9zSu*(#7e7VWCjnhHkRWk=oL9efJ; zgAdAf?QQu;Oqu`wblq9+E>rvaz38{41oVy%=GKV!!_A2}2M{$l$Q?$s(89%FYvov4LUoQsr^{xwXze_N^<X;|9Cmg;Y_e%pSmfA-f@8g@u{-&H$*o4YlF!)2WS=*b6rjMtu4)JPE zzK2XYy8jhIA2xoJ{;sT7%QfRzVj{`I^QQA$DCJqD(Xp=I%XlcebGKcYPGMYTyo1^N z{=wznn%|2)Y^QhK5jDar)!}a6P02w(_7Qk1vNRKilYQg&(w0@OOfu!>iXHw56592q2WtAMiz!-wy-9>CNpM>4J-Qvv zuz4;=jvV><`B13^IlQnkKsQ!!v$XKf>0a4TQk`J(cRLdU1g1=a#P|~Ah3F1JBPNHN zS|OTx4p{}npdN92yhOK+HjcHHear(5gdsMZx8*OKEZ<)C^$2&indb)585M-=dPsJV z+>TYyTjIrpwWBj{`fItwx=Q4X>;ZHFMpdrVifhbAu%*R2Mod2M78EeI8i}O#txgH2 zB&eV1c%i3kz)7#0@hl)vS~qxMqgPr7g)b`RJQFvsPxgE->?tAo|`*#zi!IOgaTg6=CqCvha z?Z+;+hO(R;^*ym^ud~@lL&rmgBgTCqJk%xKjl{OLzw*W)Nz24sIGW$d_AKr{Ai}4R z_Gg7^|0hbQL4Sh0VZOD1O*j#zFgnuIR2gYE9o&DtcyE^JOYe93mCui%If1jO4(};J z{~>y-@z6h|NI)FGl9Vp7F5F>V%L#wYTJR&u;|ZZnjtguIz>&B_UPv)S3s+el#HOjO z9VrplmI=BP1w$Mw^Ht)=L!#qB_^V_-X!JZQ;M$~aN(z(UR?C&q+a8^kZRDhMc(WL( zo_3IpjL9Cd%T4yGduLHJYQvLeZb3==IX1F~`vfJtvfXHX=olhm9tauj?ssMD4ju1z zO|tlH=Pf)bbvC@*`@~7bZbh4r(!zkWv2VWS)86rGY8{+MGCVfhIFA@mkjCkw5==N) z2qN79^EdkHAP*z0{YQ`L;0hr>A!ezpP(?8O$)buF3dP38PWCL`L~Fc{eoI<05;BWo z{qbMTN4P|kAvoQ{@rmRQ!bGOw^ONamxBrCib|tH-EKD1bkXXb*s{5+4)ZrQI*Vo## zbz@vg)UTl*J;#mER205@7nNr@!foI@5gybUcBI&Ic~Zyb#_6TgF#)yy>B&o-7I3QsTrFz(wDK@L!)!tgl|J~cMl6>rjzHbU^K~Oy5mo zZV#lJ4vIhaSS#(cwt(UC(twPnhZlB=Kb_b^O7a<3SH%(Wz+-x%R|Y-F6U5HJLH7R$ zFW&xka0-#@=g_1O6N~MPsw%zYEczY|4yetq@j9()O_j`fn2KhsVSVn|eSgkrxWxa8 z*TsmZ!UO34hS#yPdLwX91o~VpN5`$C;m1`X0=KZ2%*MHQlNidT!y{yFHD4DKtLYs* zwQ-YI=3!G4QwraKm>8>VG4}?G_8vbMq_DH0C1QY*W=Ywb!ek`F@6!5I+NiJ1BeJD7 ztVF5vr=aLm4l2Xy3no z*SRgXze7U3L^QRn$aND@kpUIALOIV0Cb*qDuVesDoz+QTvk1Oz~fsj6f1% z*t4^(32q1RLY>O|gLmXgAngw^MZ$Dua49xfmB8ghsRF)*zKB)7CxaB<1w&z5t|PP& zl1n=poKj>1wwIVU3S=iU#cGKLtoPjeG`(rh@TWTZcg7KEv+1R=eB;A`!WNz3*1P{u zeKHR%=krokwYDd`-vY~iPse=gFCjmZhKx<)Lue2BzLK?C&%qfW&J_FX*gg}6e@V??z9Z)SOh= z+>kg392-R*pXn2GHeSC85wQChn?F#MH%W0QQ_a7=$F=c zrH8GG!?E)ngu}sJC;$O;2%5>Pzv6I0;oI7fS() zaHPazYp9XF?rRRaE!ryuPeoAH+ftjQkJuGF?D&?y>GP|NxmKP6HqV&Y+`3;}Pms9d zXSm*mBDTM$OwIHQWA}>*u5$@U$vyWbc6n@liBpx<{yH zBq7*$t9JA%V8@}Hf7c!_G4KP$g3)JfhvyH!qAzG4e9aMI z^dpk8-5+CyZ11Wy+k#5RG^W3QQwe{hIyGUI-`$@7K;y1>iOE`1x^EQPg?uxl4dXri z&d%^L!&r6e_K>=`Y};EQWA0dBMN6H1RJC(UrE5w8^3!hI2y#)5>YIK2)mld(EKEA;Yne9Gqq zPJX6Xv-B((9kWU)gdy8V{#wiQ(bL<;#6O{%(%!_gv8YXdTSax~@LvKNdzesz20i$C!&kF(%SCb<@y2-YReNSM) zANm)$E9{H1Yw!#b|2PtM4HU%Iz3ac>%9k0(?jfNVCmyaxrpg-h;WRf@0~j}K+H^HP zKi}gSGcX>qU4v%c2h2mc?5l&eL&chgpO-XB=6(Hcq%i*_khb37`0IiXH(9Gj@0y4m z%2Mx?&=qSMo#+i$er)~aJdbjQblxVq9d~!Pw@o#0FiGj|Yf|Vb=Cq}{O&l$}*;&u+ z8mrtd|3#ElA#C_cJ8RS~>(ie^Z`?7}e{7_A`Os<-^!{_{P9jPq+)>buJW8^)Mb+V7 z3fFmJ{z@G?aOTVoJduoCJFvd#jXk?h5Su|zRS6(S8ZNJIpN%UC_nd$jc)B9u;3+kb?M=pyEHxP?QYx}E{w$%mm+0B;PiaY5~*A26mEQ@WoGFc#{^VP9i zKp27jBlf1%61 z0sjKxVML1l%pp#@*bLtTtk4Cdo@DU&%Rg^O?(4(@h*}dc=!pSKOG^NcAN0LpQBlnq zuFkwy;f47KJ}p>H-Wc&Zr==<@7Gbm^B0-+EX5CP8Cwuoc)~cgd2Ic9ws>HS&Q&)Y* zOmF3Jsh~qbx2!zBR@^>wbT})PLEZgPL{{Hv7Z>>*-_GfDU!TpV44>n%ZdHDhU?^7* zvL!3lOdy9QcEj^k=YxaG)>NM{cqog}2Iu$Q?B`v;{x7X)C6X!T%yglj|pQY`&_TsSBk$ah>x zcR4E*z}ABYVz~N>$BK@kntq1g*4UTuW zg^Bvr_J|aeJUGH)R95ecZm%$4>@Gh|HxcLe zHP?mj|-1Ol4|%dSqOjwy`ls^hx*cPt__h z6Ckk)*DDbTjCB?kDttj6Q8Mmqm>)+_D4P=xU13QMR@0yBHhQo8!&A55b@Ib7Lcz<| zT_ih`XHrF1R9%ku6O4QYlCG}*0Jz&$dIE{*uOR5zjXRoPKKzP@`>F0`1lp8mrV8Jlo7x}U?;G#$(9J({x^3BRpW$w~JL?p9=`0l$LPooc2f?2DlC zL|uAQeTtsHIGNaV??v~5p72a&$Zx<=po^14PoSYzM;dnHsSN9vyY(qM?Kd>PfcvX! zYg6Fk5A*0yh0h{TMB}dh8h4My<&ii4RDMdZ#yiP=*(7npk%N=x%U1pWDyiDWx~an@ z9*z!0!yq^rSH{D`hg1Q8j;t%o3qNK)3=*$CjF<%+^4TYXmU?>3Xy51y1^Y3z)q!Y! z4wj=U=~U0HzcV}+xkH0mB`jK$m~TC^IC5yR^a3CSIjgq@8XwN+Pn&uBbN=R#wRI31i;G}b0eb4zKL`L)dt zrlh1eC7kR{i?YB$APZwcwnuk~yUL|}4YBryLp>c6w_u_wdD6inR)V91MNHP=oJ+7z zN6(mO1TqzxuDO}c)9}quMQ#t_zMhzy=)3;c&ID4grl+Ud+^c6yrj1E``H%i!g9m;p z5EhA*FIjGa(f?S%>HP~ovh~E;&>3D^Rm?4rxL$29cHe^gt+}P8Kw)o_6bXHCauNqf zrc=do&ABVY?0bWxm5SzjJZM4(@8nnY-&@io0s4~$CmT=$pVmi9^^17J zJsctvSFiEd^CW@Co=VG<_{H9VIz2YXs~Pyh?hup^a?GJ4ca&_Y4iD9stpd9EUniXw z3TvnFf5vjqV>vXfT2u8pOiepD|J2ce^qmW~d30H7R-H0n(UY5R(M}Lca8#ihB+^r^ z>{lBM3h@0k-}^iF#LHzQ(i}}%FqqpCz_$o$0}-eMP@A)u$YtS)O!BH>5cCMF9RA6o z%|Ey>gaYT8k@VYj5t5croG#dF-S_X66)OMrKMK{#hJdsc)dc6l4XXzQYp@F??JF@< z0ObkN)dmJ#s^s1V&h|=tEI(&bF$d@l%5iwUjReUg_!fgiSep)o09`o zij1t$u^~IdHe|)p3pX3ow9qnJIc+(j>-)~bFrSf*QiMbED$rpcWyQM36*4+ax%J)? z1tpoPfz%F(uNJ!vkY``35oVbF{bfbVvRG=iQIY?s2F8hyr2Wf_t>4Si@HtUOw>sV$ z`{*SO2(h4@!u5SygH%|oc5y7%_^=e)86FX4$B13lmnNh3#T)59HdSw1b%{A3HIfqj z2I-_vi1{&MK>L-czF4KAM)Tnb2dp;zkMq#D;q;LSl65AlQ;cV#;8iVVLbbV z3gTO~1pirZS0ZLesD7gt?%_G?!P*cF0BDQk`XH2sw{nBBa(r#A4$e|JRGe^(q4|Im za1$bifo)Mg%)m|}T69Sc8@p0nuw&%itfq!AYpFY23raFjw-R%9vQD&p!~QISbT|LL ztIb}y@RCe*_^}5YmtMfEJO%ZZ$9Rf=+iI{aV3>kIW-*`^-Uz0AXCZBsB>zU(tdVVv zqUr2zSVH6XMX@gEV5JG?>uYas9~c-23lDFwUq1-GAF;nrIRC>In=ii+(Wpj=EJ4Gj z6_J$iE9Qb8Q-;UaFtp^lYYVK{QJVbR4$W86n2gRdTnR(+_)Oj%d8u|qz~Smj`Ipv? zx0VE$@9hx#RuJbi%TeXvrC7iwJOIW>`VsrI)Oo@cYO-@rcvi%e_lzC$cxbR;k5gSy zzV)l@UDnatV)AKMS6=|$M|1P?ys5t9dRn5-fBmp0Y--CJ%6rkMXZF8U7!e~%2H1;Z z8~yB)Yv|CkqpcsYxEO@BWIt@*wjKBRx*UOE2jdbyo+S=6#M2WTc{wP)O3oGR8SRGX znAp%m9-NMCQH#!Kv3Qzo1^38?$=Dsm`JPg@>BD`E8@8h+uOyi;#C+t#nG60yTK`=c z-GT@eq96Ii?s2^_yV8a-q6{IT{f~8sh2oL8$?rZ#F^G{%?2Typ9kjAKoivp$B3riH zO-b=botLSEzu=AAw|Rk3czv4;N*Me~0+RuyOtzoe@`!ImT~oc=<&=T*DJQMjG_-Ea zO3mjx-wSP|CiWMbGxK@q0fl^35b^L{X2)KJVa8l$Oqd@8Sob=tf7M(sp|g?o%Lq;9 zz|7UPyPKvDA8z={i~4CU+m+1Jbn3@f)ALd@JND0ena}(hOkvhMo;{aK({+KAVK?>o4Qb@cRvB_$)#AhNBby_b;@6cTd& zi(SX9SP<2);VOsEMe%Pv`i3hev^!|l&EOrG`0JncB;(y5fGpk#rym@N3xTsz{zI&b zqzA&8Ky6_}`VEm~h?(REqHk<_0yB4F%0XV>NWwnH5;^9%&BxiaL=l%OA}*fx-~l5% zj`#*==RqMNqcq+L-19z&u)+T971^Ny&mbP$7$bJ#%1%jgQvCK<@sW>ciZ!FKw4;RA zHs*AOh7FDd=G=$V+A}YAWd+G>w4)RZF1dS+c7x+4ownfo1DiwI`fa#{&u^hO8vkJy z1TN&L_U`cOV;%mZ{Mly=wAQ!@)TG4~U#Xk0mrT;R3rSze#e1&ak1(%^oHQjZt-_nq z{c3zx66cS|$H=cNDr%15k=m2KCAF=S6_a-QElm3^3S-+FV?>&4to-ZM+wk+XPygNthQCB961+S$(N_d=ZmGb9W_>E6k z5<|7qY8w^oddZK2JCKim=7%fqCXRAmX*+hZRl);_lFY0y#1DF7H8AwISeCs(j{{o> zWlGv>6G^7B6Zmnv;6Vz!)D`2IM@ua1ha|@^0zL7}@>Nb@$ ztJ^RT;QZ8iAh+W8m|Yp;sB6i)mH9v2l9C3J9Cae)XzNE4CIuM0ey26CUdTZONQR{?BOt<$u?yvN64{^N3{p(^P zu|LP zr6vxdM>lvSV-fMFCYq}hFauF*=ZIx$XCD`ghRZM!P>_@#vaiEx5MzYd`Zv=`tIb+H zLCUDEs~be2t&pO&%_w%9A-iW1r7-5+{uQ6@6&L^zIbPx6ild^h36w<`co3xBWC^*m zXLQ?tBVca#u_KL$i79M#L4kjEbI_A0u?UbuB$TU75?m^{Ncl$iOyAUZuLZrsbyIDK z&B4crjKI587a(Q?i;IsB!4FlzH6w=SrXStKCjuu+OFw>+ut${XIDZKCBv`qsway!} zoj4o0%lcC+{ndH6rB2 zT}SNjJMD+0KNt97rt{6q))$?)7`Y;-kQqN2PK&Fmt-U@g#b%7Bo^%&cQ7jBsO!zGS z0T^e3TK4@P$n)5<5c2%Zn?f}a!N<&JcclTbhMt@(i;&TLi9`^gutr_K&e+)42-^?9 zUU>+ThyQ>jku~G2HGcYb{KCRFG8J^Y98$YYq^rc?8F4l%h#nPuod9H*!19e5-g~`$ za(XdWf7;vIb3wQO#`$Lmyrq~Rslr6_1Z=!mgpDH&jD}L^wfDCo_o#oBW)<5ey#Zt% z$YMh03|O;jjD~Fdz$a4fROY6prV`Hp<+y%1AcTZ?25|Cn|2d#jt-29S zbfo33Yn6@m9Czw-rb&%c(SE<{Q){!a)0nSctxU9Fs}$xR0moac74}9}^b>Rr+n(Cf zyRu^i9eZBPcbrL$&o-$mW|wtd>bqHi(SS$s5($$6u7>E!0unFY6in(;HqbIxl*w+| zYcD)~-`N`N_LySW0m+;mdo`L^h8tx-{e;-B-Ra!**lhQ+Lo0xyO8G4Pt9QO<_6ZhZ zau8&p-(Ii2h@;{G!3rab>gWL+_0Krr4>15WZ&ab7)CeYTCxHJdq0P#Y`E5_BKM@g?|-OL7w#}CpM zzdR5PrX?!_c&(2wg`3EBk*&{p9MvxUhFq?)irRJE# zIqBdfzLii<@J%mr#ds^11};4N38Shqm=HBfpVbI-Y8D6^=4th&iS7IN&wc-|CxZC= z!&Wk=tu@US>WM^W$e155J`so1>Ms3;a|%f!gyA^*%M*`Y%auvMtbWxkxQ^mtaJq4< z_-%n>fm62@MimUdP=o?4CGQ)z)XK83`r2PVP@`h8m>}fY$L`@997rHp4)MhH%l<`)#Rs~v|Fw8aZT=J|1Oq< z!8=j{H}29!oA(t)NaUM!zSrPqjhgIuZrIUEX-7xn#^tUrYSJtt^GsJt>e)xfj%+^L zdAInXf%>R`jJP@K5|x^5yStU0&DM3gOeht$7FcWX=bb)tQ#-^=GsN`D-BGmi@$lp# ziv^;@y8Yn6gLC?MmB{uaM;BK&!aO^9WnsqA=KVY2$63?GH7&7ZBwIC%tWNG-FZ^i8 zLB$Nfdw#A@dFB^Gjx129q2kcq@!p3cqFc6DP7gJZHNw{CXHnND(G8EbehV&VbG&*K z#vhxRb zC-l{Mj}?B%Zw*zRIgB-%U)A8ICaE09IKpa&fcC813o*6iMR9I^mc8QhMH;kPVvf)E zNo#YRizL>S*V~~oI-MQ#z(t-ImQo}F;t5KXCpeUZWCMcDjSyF}n7;!gnH5JR$l@AJ zESJQIT{l)*?~VB!Tri*(W6@y-)a`-LCQ;FhxKr&VA-wVtR}pl2LZXU@t$`-IMQ~Cu z-~HsL4=8W=;sml{x!Mmu?`bk%zYye^Fp@oS#NzEkvwveGMgks=1|qzz!Zs^%o;eRh zfVpG+J};u_2^(a?t1b)RW!2LrkpfRoO<2eTo~Djyino zYm@or!3T|AFT;t?xr%OQ1I){rFDiy!2BUmifsIf9_zC^fR0p3d;Q`~jD^91g9jJG_ z(~)tfCrjOyp}TqFyBKlcBV(|5(EuxbmAe4AU@&eg$4d>4aS!-*a@z%UV47C+7{zf6 zVC32ZfC%@e3&p+ie_L9^fr8LKfH|T$UjJ2IFL0sZNaAo_ZZ2`DBV}rF|9s5z<67jF zpOg+c2HYvikk_?#`=Y4<*sXZ596NI+J}S?4m62%4XHMF8Ybg6JWGgRyJ5)!?6{5%J z&01Y}SD3W|o*_1r@O7u5#)zEW*6Kb?SkPD~o$Yd2fpnLgRO(V9@fFs)T?M8@ebNXK zZ>rU8V8UI3{hCGTyOx$S#CQhCTiZz@TZL1NfRIq?yLe+zkbmUbyzj%3$C~7qV~EY7 zD~o>9UmgW33N zj+j!sx2#7;t|TW_5#SQ50E;6W1&);kNm`=B1`0S3+|$~t*%Jm~j*?b1VwFic50dyy zL_T1Vg`*i#?s<-Z9cy@8OB*;?q;&~Lng}V850j&Bu9M2%=ge8!t1rPh(5StQEbr(9 z5?zwDlaG6kJ^qQHtkvKef3E+wDqLfcuht=)?!Ss`VFMRtGLJ9ZKf=lOo=DM0Lbq-B zvH8DP15RC4P>qRf-b}Jea$Ha$$$AkR29h-kY@B#G^!F6Nxf@>xl#oL8;Jk=li;x=L zywQ3>^T78t$!8(R35Z-6l0L%B@_oq%i&so*NAEo>msm(iSuLQvk*93>wRf?ycJf~~ z`V+&C>y3HRdfEbPHF&H|*%>;6#M--anVQN^f6qawad_zKNOe$1&g+oqtZCkAxWK`1 z!=R}}XB+l9CtA(9RF)Y>gb(e>Skg11z5W`R_%r3mVO2Lk*Hu9C6cytz1z_C_5=IjK)BXfDg*fAc zXLEVQYGKUF0AVY?<%Jqv1Er0GGa&RZ*q3iV35LYd5%9Fo7gvttLej*TGFe4OMU|V& zBxkDsR?$a2Aqsok1t2C(DT&nx@G`r?a^4lAMdh;C29BDco!KLkxZfqgd4edI{sq=P zPX;2iIaPe>HWfdGHWa=8#+k?FoN}RMko&D!b<5qVbG_qPX-1mT-qkPg(7i&FV$u3; zIs%yLxO#aJo$xH4g8r9)}l_80u>krqmcH@%l3vLWoYrpaEJUnz`0$FHW{sQ==~ z7nF2L=_>FA3SF9QlJ=7%-Gs@N9ZcbO%=2V)m};)mGLB&D*-SkM+!D|9JiYrw_6;bJUBpb{==v z$Oo0@iot%`!{KPFJ>#1DhxYXk*c?(Xqm7IHXwLBuA)VE_KzXt#jXQ-6t{OQ^)2h-@VQtpj1h*-&tlA9xn@ zi|U0>TU1PJ0L^J-!5BwB{Kg<{AWR=P(}G4dFdo1XKlFme^z_VWuMhW!%#ydpqSX8x z*1vFV`Hs7N6__DN1XcN~xQ5@qf#s4Mi6i+o`^832AK?h?pCF1^oIeRv4&%axhB_V6 ziJ}XHO;o>JPUmeIN0U64v1h9rPBtRG7Gv_FzPR&rsiQ|p!|&PeJ`YR_Ls9BNK$flk zCE4IDvTaiTbQ8}w@{el%A#aRIb{1e1KJCFT1VS2=iL#-2tjVX4WiT*cMM4g+$gKWv zZbwi^CXxdQW`@@ZbXqujgr+1`^Eu+OU!SvFiw#-sf9hJn%&UPCOtv>KPEc&@@JFw& zSEdl8neYI;K8xZ==skLVJD@oJmhr{rX)!~|TnWZ@$NFghopbBsjxDPQo2d`5jWr(K z6>THh7{CQmL9vWY2^Q|u^1)RnxQy48bDLfhH+39Mh5b@%gC<~!lb?mtcCNi5FzUZj zz3f3kh`nNXi%z%x3x*TijiXN$aYN7l6cJdFU{)DKdxl4~@czDhYvG=!2LznZjkK*Y zay(=(XxwkRo{|It1BnPH^2}}9^gATjZl&2ykM|~`YPdG<-?!h`7;9X5fXdX0<(GGp zI|1RDAR_G~9r}8uC)z!KiKO1Ue}6H02-gm*S+H7$QwoVUA_Nft)!(0PR9pA;yUFp# zyN%xuKTGaxB5VJh3&Z;UW#(WPx|@mBBtsidLh6xfv19_0VzT?006I{SU6v~rWhfVV zxpzGM`E&4>rxIa%EKd?>wmM1vIHKak*Raq5tIuMm5|t5IvmMGmdinvre1t)hIe2|J zp!g))a)>%UjR)!V7_kyueB(tvXnhmRQBl^RZ7U-M~B!Dgp2h z@3^ILQBmwTQ=B`jG#uoG8(cvH2_spq#O&!O=7Ry;ZG-rUUl;J)6Z;)~Ki*VBYgpgJ zTE;JB`kcP(Qt6Q@TGpu-J_4_OqU^X{HIt?~hkNybpq8Qp?J6h!fdGRyU{riay4x>_ z&G#-yp$ZNGp3hf|;*y{1Fxy}INh<&F#Q_@nLf%8q0 z$p-I?N>39zpVbi#l4?a>=>E_W7{J?DAS%~&P!GUikS)dP)TzkGNL4SdALoBRM+^H$ zVu&T@11eh`8E*k%Jb{(v*ZGA>@6BrKwrt(%_-(Rfk9+djdUw9MsVXYyyECv8GBN2U5LHU#Hk(?!|BgLp1gPydPcQ1L4YyYt8r!OE>NL+85y}Kz$K2){z z`~s)^8LQFBmh;9<`E3!yTB0P)8qv((2^*~6N9Fi~Vabk!cDK~nnr)@Gi5M48+Ns~` zcadg9P)yf-y4jul<;o2gF^p2k8qz?7y~F`Ntmuk<+cq3+=wN<8+!4_W=Vhjgsy)0D z^~&80HxeJ&M&c`-pSh;NIm5&TEUgkT3PIT;dRVS^0kI^gk2X_(YNx=85%+s{ny-v8#!W5W7@{-Emy1IcGYIrByh$<4$E()^{KbHH^!V) z@42y$XOpz9Jx9+^lB>%k+w}sfkl8kh83f62BRco-=An8ird0L-^Y_K##QmB}+U|fu z_2iIBQCchd2L;q86(lRDVd==yew>5p*j_y)sYi~AGP9_9_?fWJ*a&iEyKkhgeR8@9 z8O!&M{&;YKA{~7|S@JyFXMxAZgvL9!z5XAv7S<7)#l$y|ucHh>L&-eCY1HbDCum#E zb%uuJi&7V4%~SEf=mIGZ=GdD2=usFoPPVcFS>t%hYRC)j0R&`0u4(e|(({gAy0XA$ zKnv6dZo)}KbCEI+GFbkc*%wT2sOR9IR*!pblfklq(7Z|ZAo0?9gq{iqO4}hl^bKEl zOYK0^(v*`{kcEWIy7fP(9zXDb!>VE*k2W#BvQ@iu>nn4w(2iDD`VJ=lZWe}Ztmo9= zpp#W0Q!7{czN19V^MSmP`lXO8_kz6-_o=EjNwX$P8>w?^I?8rXxCD3^lkCI!9iH-N zvW3&-2=OSVIy>;lf(-8#$I{!P(>^IEqwk5>Qq&{bXFq_ngbbtz5_0-^6mQqh8R=N5 zBnM+T^#0{#><-`rxZyb4X7~lsi~ni4RI#~VbMuS z0%BsZU50%X!u7~Ap{c8T)A>3}w*%JOi}{S=JHb{EyxtzHe);ASA}t_pjo594I$pkO z{@P}42X{}iyD~?|^mm`ds7nwDh*@*a$c)lMs(~*kzvyvZWJl|@x`PTdfIWdZzKlyQ zdxYSxmcM}5oOpX%o~lKctesYp&BJqA_W3$GtXR*#AkV4#INy=`E&|^8m;S6>6ZJ>VtqKJpnC97Z? z21OAGktEdmMb;chNnpDtJukRkOmRH?>wsZXo`&vUfjyO`%*sz`Yc2P#{_NI!tiMg> z9^uVNxWV;5AVb+eHB`P>Y9$p|49D+Izk<5%Kb8 zFo_!-JgAPQ$h6SuwY%A|3k&Z2xxSuLR9W&Ok+nsjFOby>-n!uBmLx}qtZJc)wCsyX z!`kx+-ddyyzXy8toYvAmFcKqNnqK@Oa9bm-+m8>&@wT;4v8vS298mcfLSzC?a@gqP zv?SaZ>ZUAeBK8#y5-$ddeYW}M=8T+K`i7i_dTaaSMD#CcV+Z zzn&&|Yef-L3aWTbb#)&u-|J1+{fFXEiNdNd=3~ujg!Cx<8i~%TgbPfNtKjox5|Abs_xPCNej@lDjr*)<1Nx;k%K3^pXclCllJC(JPNsH%veB ztD9zf6p+EeRtxj+b(aQP_EJ6}B+vv37s#KHE$Dm{7IDM9FwljvA%<(2Q( zRN#m|y95GB0TNb(?cm7=Me;w|)0AD=(C(;>|C!3M2-<)%p&CrXPX=Sys(p-|t!J*Q9or`SMjwZtf56JeJYXZK`LYso5cPt6+te=pMbW zFf=El8ge;bU8HzAC8+D$0@}l?Qd;It>$c9Y&K)?X<#+9ZeOmUSuZy5%38--^56S;y zfT0arA58!=47yDdM>r01WDkC5W6mD@Vn?MlFl(i7G5vaz$9f_A47QChcSf#NZ$naDG$B|p;|>&w*y)zaA2G7df=HiY(GJ~xNzZ{qfySOA?+=b zq?dPtP?K(x6}JG1+AH>K%2y-SUk9J~PTXEyB zztRz=PQlPj#P8ti6C;0=gmsv+imbIG#+@+y5FSAnEaJk2727q%I{IMoLg*eC^GFC4 z@qmRnIqZ7Ye(b!TA=0a+)xqDmBR9v#@0@<&$e7Th7W>g8rY0#}M%%Kkjg{^z6_~;U z)Sp)Qsk0Yz*dCrX^?P!4PN5^9oomXmfzsX*@hPa8VmtSh>%7uOX_?yr``>IFy^^Ea z2ht?4d7H*dd5TB^_({ib#ZpnuNwBTMho?y$#J%q->8F1(+5S(#EvZMupg&l=SVJW* z*h3L%%Y*`j4TIvMbtc*iaAIs-eSZI1h}~{UM^4;|_2*8=zC955+!o>lcu@Kzp7#JH zj)WjXZTT$nln`r3_Q>M87q|evg9fRoj5M7E4P>sdn7_wZc4F{tJXxj_^C$xme=K;? zE2?eUrU-xoDbJ}<*S-0oc~76-25F;mCS01yBA~1f^9vRsa)jE^l&-%?Qc}`mVV;$B z5oaIY=q*Ra6nvJeNAo=@9@``W7R*K?U>zd_ng&pvN?YWu1gKl9r&PyH!g!uTFh>m{YPi#gsoxjh`hfwYu|8lz@$_ z!8?Q9?vE~o@st~hGhDGfoI*!@SXdrKeBh`!pti?3VRAz0^o@ruSyF++(FV1qwzSnuGt3toH!t zx^4f*+oe0Av}{FENLC0f38Bo&sLbpld#k9Bj7rE}p%ls<$%;s__ewU|r1-rq_w#&z zzwiHl9M63m_whVE`uM!B_j#S?Yn@YQ+g`vz=ymB(y*&T!Gw;a~7qU8N5sG4VpB8B4 zKC?n(x5@1W??bb;*o1_-LRVLo?-KhcFRlBtSmWxeRq!@?LWEFn5Zy#xapR+_^4WAc zrko}kWK0lyA$E+}f%7Bq?#ZQ=8t4*Art_{`Qoj2l`l9l+iNEkQi%QQ~k{3GqvLgl< z!kO6NdAn}k^EO8 zKG10=cQZU1=bv+3{RI5AQD8q&B?==> zDeaJmIQm`Xkor5V(AFD*&nWxr+dc?~KYs8a#b|?*3?6hq@5R#kzD&Ev#&o9bXQ`F{ zvvq97T%FJ9GH+e8t}wsD$NJg!D}9DyhK5%>pQyNs4$16W;5xa6i3j;Udy?YQTsW+w zT0ir3L~lF`u$@~<9U^H+oPTWABUI-F&iH9260k9g-qB-;aQySOr`H3e;rSxc(?DBj0vWlD{ zh^T?83lpg&9E2qQHT4+o`RAKmfh9mE>AmsL{DYc-K0p~yI;84Wo}`)4ms^13=q0^A zLOObZAM}wdr&HwuI`H-=4f`S2lVC|3a9ymQ z8Kv2L&4(>wMkUXE)8#H?taqf=DiHnQ`Adu=Z9}F~q`Ur=g=)5K;sp~Ts(poRnOxbv z@nhHBxPJ)!(kk8rKZ9gz>XMTJAm7F{e^4|74SCoJTcHy}Cuz4HgGOX{WaQYk^-b23i%z(|3Bhs)Tc`Sjzq;g0 zL6D4`fPeQ-)>@0t*&oQ!`wjIu>F=N=GM(4`rbxCM6XhQQoo@g%O)XPkb-1gVtaePN zZJS9+y8Ig?|-@xlQ)Q1^gnQZ$5szs`JHVU?$W1>Fz{%C zIjoitIA@zD{Vg&ZDpDduA8tt24KfNTBZ4^Oz&ECh5MMypfxX+N2sLC_yC1{atPsp* z!-?b!k`;hiF*&uH&el( z0js-s8J*~=+@RzGw_q*jM{5R#@*?GH-VI+qNLxBRX`-RjV`32AcKz-~zT`1ko0;xY z2F?E{5{$pG`+qQg`weIlZX&K|$157f9pV;qNMDNuYJ|xlwIPESG#Fxw$PNLb)!^Nx z`y@g+!?fiS3H*kzkgPzn_CdN82wZ*OwK!i=67yS9pAb<#Mq&-kYcLz=>$AgI*9}xS z9^+p1$O6IDNUkN+`RyM3yTe=sT+mN=UH>S|6iYhf$WH&z_amCn^X3 z8;&^g%i1x#90xJ@#0O2uCoYs))~d?lpj2td8=KQB^@5?I^JX4{Nt<{lDkcc zt?t@Yo-?ZT-1BZ5Gw_k9*{4Es+-Wm;jf*vQmyzr{{0P2-Jruuk%=KDIk%B70?{*zA z)t`}0!w3X-F)yR^SF;8{SrHUbO z2I7KuSGU9`qTdOCW*1q$6_u64zY6H%XaR~*kwOGUFSekdp!5a8!N}8Cd<_CfNfr*- zTV!SVG~(|vL=8~rNrnqx$imQ@94u)O(wq&W8}do>fJ5m|66X~ai|}0%Y7#-jap-^v zY8R*=`Y2Vm+_G8w!CFR>YM;)Olo6y)9w!QfY8N5(oL_j`2`Djcd%*ytK0%?#q9a|( zU|2S~g{i#v95twX%DLq9S61C%lx9&T4OYGqypZqf6bRe2aeq%`{LSoDxm9Wy7Y0P^`;0)L zw)l2>b7@M#+oLp77;i1`SAl}k4Wh)FE^OknGTe6*GP+uOo`MA(BA8)GJK%sMR8pNV zN-}?GUGY4W~Wx5Meu?VBsd!1bBX4WL2Yj6$+f z=yStJn+Fh`EK-|jG85QE^&lGw1B*z2F7ahNp;LYsp8v>uRK1|A$KJQ?h)!N5rCSdY zb2pkUW3+_2So7>eWZf9inZpN4ZBo`C^A1o!Ie~g1X4sSa`e=s1;W%M5m-g_Zg=iHZ za2WJSd(=n(x#3oabvSGUNiz;F9w%}Sv@B25#;rREB`H}ZiP7>L?_SemUwOCjWOc72 zVhFT@xlM8FJetD>>1A&)J{Y3UmA0R~|4tF0g4pJwtobYc?zHf(aU3 zcgdqWsLeE*K4@pe!%{DAE#ODDvaz~o)o7!T-Y7?RuK+>arPVr@$LmoJ$Iv755gDKpKQGYTrOpMEQW)YGR7-DC?cUd5a9S7*f zr#9w1aNdSqM*@-W(Y$?{!Jr}Y+4JXVXd{@Mew_xTuttYFKt7Z|X05_Q(=KNT=BP_s z+(>aEvZ#pVnFPk1F~bGelOps17GX(?SzEO^(c}Nbn(gL^O9oJryJA_;!pd^t%F*3_ zd^$IM#7t1xo_&YhEpsCocZ*nV<^p!wkPG`U_Rvj-{;oaJmg)ki!Qhj=1Sxv3N4Yu0(HO(o2I{6#ORG%df5(TOZ}s z`31TD6+J92R< zqcp2e^Pf6SNhQn@M4CXd3!uKfL`?e(8&4c9dr8ii7IQzP^v71% zF?@ghrE~JC`>X}qE~T73b%C`_Y%h?1`JKfAeVvb0}^VPyV2WQc#=R}LogKNRwSd(=-2RRY!>sZndn0_ zSU2lVq;i zWXBtk?5U=VO=U|zR^l%_dJ983P(`<(YYvYv0vjydiC+6@&?T~Cy~j68=G#`BVS?miAo;{@quLUJN0FJJAl`EtyY6+jdOtbwxak;lnUPrmbD zj!E+g6Nm0edP=Q~S%%gOMvd|Op{>_EwIP$qFbl-Zh-8fgXguUt{Ps><^e`lS3bNAv zWrQ9|usTSY*D)R-Lmm-L0x9;@GPvqr7HsIJ+%#+OCYS_alNE)qq?HYLb6M{Z&uR;$ zqg8u-r3kayx6w2*$>Nn7t(*NjZd+(#(E_n)u*e+)?=R`;=@pQ-CbZFmD5X*>X7SPy*8v;CD##pmHl!9zG!|e}3%?7C< zmH5mhqHh9ARf3z81RcIJv7a`)F$vf0Pux_-yA9g|4VzOza3Nr(B!#~gbh#1D=O1~U^q+q;G&d2u@%sn=feR7@ zv=rElVU{M{3<(pel<{DdSEo3>NFSo-`NDSw8plo2Y4GtzA1)S1@L{#~tOTlY=4`M@+NJD0d zj-1)7lm`g{R>k{8tXYPF&%Sf&Xgi4MoJwSHLqLv*-;li*WPtT*K{Up%*7Zk?_y=B) zdMxTr{9olvTAn=%!fe)8GQYk_xGf$O*>S3wGU}JkX>mjb$T7R{pKj%S_DT6smuE|e zxaYHR7~P6DOxC-QywclzM~vza?qtNf|@er(Pw3z5#!4`o2lI{?&tA@MX@HiV4n&Wk-ky$M6YNyMIOcC5P% zeU|e^a!7~^*=EXM*X=&Gm02p3<0woQWYprqT5V3ql0W+&Ao{QQ#beDK5PoDsEnx?u zM`zmiYxs zL$D{K={?c8|GU?}TFWRaufe~9hq0<|#J#bGJLu}8aXn*)Q$1vr3JDj+DC4WOo6>x) z{@!JBv`7D`nTDR==;|UN1e3Tow4SH`Z*J=W z!2_XC>kK+H6{hI8SFL|?fHUrTP#2RMBbAc!hiAWA8M74jYAFwTN;P(`Zfz#q9aKQ6 z2=aEZ7%|)HTgsX^X2LS`53Scb#{E!Lh;5GJG;9|%*lVCmZ}6vWlcTz~uEq`#zZR8? zj(IU}QnWdm;f(+8yrF7Myiuuh_@hs%JR|h$gRz@R8%(JaL=c2}{sZYle0R zFHHLz#|9ec+h~{==-Y9e2s7IoQdHDI{TZ=IkG0Od(eQZl>4uy3@C@^}2&|O3k1&pb z)ZK?!TM=7^@kTuavCg%0FmT6L3;{VeXiO0Aa*rPzjgiV4XaaRX+PLBj>6^H zx2TtYo77LAf_Msjm@d}Wn0DsvldlWO-y?{M<84qURZ8AU!uq|$#ymE5_OZ*!ItU9X z{N3<&HUACigQ?r>ZES>a!I5VHBWF^>(r&XNQ_&}!%JIK&q{`RDQMc@g06=7x*-?fV zz;v69vN7&e;E%j|8+ax{Ib>3m^^F?TcHlYW5!c*7BtEN<;oloI)Y)Wfm= zkoGh7Z^mdf-~_s?Wv!ZiOtCio1!eI^6#RKe??`$U%tO^wH^wHuM%od?RL2#P-q){5 zDkhP-zShi6Yp_6m8QE`)wH5w>fd`7@YPDe&C93>B5@`qVYgWo9WjfEH1MJq+iNP?W z6n-%;#H1HKpmbeX!E9FOOc9+x_1UJ66Z0hEfj}(3kjV0`rG}_IG;S?=z_=?Ce)cm} zMyux#08#Ffbd%=ADv=*YPCtI^Sf{)$SW!GvkpJPBr>#k~z^`DJb!tix@LAD))&v`d zyYY!L=_|3Jl1R7!8(Em`sZop+2$-e){#tIGLz9H(v!1E4Q}Hcv?p4IfsVtMPanIPa zSit!pP%buf_?QLG++7o0jKw~h*#Xe5!-2b7Bi_aq32j95jx?d3`K5m6Lh5LF=#v@H zp&5C9A&!_kZDyIJ+EPm?ANbrz6(ZJZytEK$crHheJU{M zb0P*VBNH9K>mrVU&h*3aiLakOcN5;17{Cd|VtVac+Q-Pm!S6l9E)5D-2f8lkTt{+J zBThopMH117dEwHF8M1|%F#RLxT;4#$iKrdf{-V~*c?0&E0KQ3np9;oZwK}iPCxg1w zX3IWb6NiLZ=Io@(!|`SiO#RQ``IBDUCyz>A%y48$Ukw0@7T4SEO*F5+#pLUBw5x6l zZ~F*oY@_KwWy-yj5*y!hHNGFX{;3myw5JC=CkO29q_07fvNN*f>c`p$8Z5|#B!JUl0vWxMxhJkBTQ?A(-Z|^ICRTi zUXV@Oo<}YCtV=OxNV(Dr5iK?x;Z2KglT&Lx6)z&X6416;z z=GpxNWRuRSZ12;HZ+A3uP6R4>MKZ=FMFk9dZbj}_xR&?Q?%tg zJ}LXpg^GI|i{aQf6r%9w1f3%XC8bs!uhn|3rZ%x=2CaamX-%ja^u_(qvzz8aTfz_ZB@=7eLMlqku-qH_82^%oUG|ENoK7=M1VtM&4&4+)w`XZ4~lh!c1pESgmNm+tT8`M%1CYZXAj zJ|e~C7Dfv;*W;2PrV(35FhZOzr`_6BM8)6t_;?!jB6ojBRL?8Rz<7cC&shxff_uKaVvLd zc``$CK9w2tH;JIef_k{rf83Gl`%A7~Y_~Sb9bj-41N7_Q*4(!<{|^)fkgnxiHMDY- zJR1!q6ba7k!yxY6ac#!l=yw6~}H_DKH25WPali`;KwH(LgSPL$QX(~EjtK}7%X zMdsg?W0En0U-2$y5SUIGSqw|z@kVyx=arOp6y4y>C@vu#u=m&%zHJ9~4{so$R)a3jXzpTBL3CZW(9`{%kKZy+gg59v$|q! zE%+ZxS!C>c+M|sj`&ihWsfYHp>@^tRlF8Tq>-uiYA#M9!8Ovceg=~eIJBe>P6|`oB zK1Hbu$VZLPRmiL*Q9kBbJ)0E7XY2E?)sU&uJZJK!M2p{07S-tl*6(|zV~wPuzFDu2 z=qbLnv6t!Q=p%WbrRW4fq*`zY1vyljnBJ!>SKu>g4H?Q@*O9?syZ^TMgIVij-u0ry zZG*YckzvAq)t`;t{q>d9bsVGJ*;~c7406T>?_;Cq3~LOzO2d=zsl_wuD#J@Fy9CXf z;Fkc#4KzjyY_I;t>%8eaqctr7ACr<(j5E&Mryh%Xs)IWLZvIqcO(JZmb2y7YV@U0S zCsf+lI4Lh;&l#)F7&5WCQpWpBm(#TTSBa%j;i^~8Av{N!W-(D(qQ;V&P;H)_Ug`&1 zWc*UVdiOPWVsq94PR%L;e+F`R$c;}{jKCKge9vgrmOa#)tNEj%B#W}mY zw%vaH8uUqo#%bwV7e|Yt|6`T9reLj|PseQrJx2Ll&yhGC?qJ&v>ph@9F1(IP=l3Yq zM%IcQpix3?>#l7cMNtPW~q>@p1Htv*UTw3M7$1 zu&rm8_>wi$?Ie5=ukA!TOi08mIX4ZhL*wXf@dH8*&r>n80TzcQUGpIMb6Q#CqakpNPMKs{HymZMxmqu%-XISE44Ru@*r}a5b!o5W%{_ z(z?FJMv-Z6C0nW7fsV?O?{l}ZE52^k!;+4#Z=zso`yuG||BE>5T$w**1-K3lM98bj z9wpp5?ZyjioScscsE4CgfUkNYRUcZ`RwYFH!OHddYuJfUE%T6r2%=^DP2Z^%-Bx1w z?hfDWJexmrorNHNX`i2Ks-&Ndl#ImFYyS=LJ$U`s;63X#5>z*;F_WI`O&) z!TgE=Xh&`rl*s)^^XvuDK&1Xe$yKLtg4;%1J}7oi^l^$!_pqoCMx8;M2aGm*-A}P} z*1^vr{GR{^0UGhMwzIqwEPrsZe{k8On+Kk*y zbOB!i*AP1~*@|)l8LAytZ+kByCgfowl#00biwXXTY$76mv~C9%BM1Hy%~B%QokMdK z#3Krtij3CaXg8-MT#_w*9{^3boUzny&)Ug%ohhe&Zc+Na$!1|dt0^kOm>$sIyW&~H zkWb$PF)>1Nw2dGjcz;KK&7RR~$)HckUHc|zGv8R*yWlYo8;^bf0z>P)?;g>d>JaRE`V%_rSLej@OQbnCg}O|DWX*b3tGPJTuw^)S54n> z2Ix@%W0Pwn&(6Yh;OZ>`LzjeOfJsj7!^zV_)vUH@KQ0JG3$c2Vh&#+QOC@v4R$vE7 z`}_joBuCVodsU%( zl%_|@HYFdG?{8cdtX4YT9&&K*VN|i=^&29Ya=vs4cW>y@$D0qcv2R`lBXC;3qPS$e zwN;!^l;woxj7T#e`T)r135}x8OYOZr^v6k-CP~F?e)Q_h!-9UeQF?&zjlci4^3Tz{ zkAU+#M`Pz+5nUq^A8aeftc2odiJT2uv`eFDhAcH0CDVS$Pc(s2{6aWONZHzltsTPm zI8N;Cu+2kO9XVrQO~1Y7TvBVmQ19X1;CP1AV>MI}I% zeF+MV$BmASVzl)rliAwG*KR2zU6&*u`7ns2X|Yxg>Q3fQI6r~rd;=kD)!H&ZidS8` zmz(?G@Ru)Nj0_DyrBOBa)r?^qQv9^JL&sS5$yJqG$B!^KGS+M;v3ZUW^i@lLE#oH-UmDV?f9;6AzT7$A@*YzV-!)T_DosU2fRM1!BqJE_J&+IJ0^rA-hcr42xYCSUIW>V`2||;arrCawv3^bQ`}F@mwn7i zleaXEef%cAiR$x>sb?p9aD)GVqR#;E#}Z@N{jz!9d@g$JTy5if4JtyTu|y|f|zE_W5JB~2U`Q8N3dLCKIxan^%(34eq0mRId^BDxZRlj zHMWwL>S@zTKBw5YHmc0zsFr*_u&+k(`%z z{YDQ-NtT@uwsy2cnN_kY9s46d_TDB;-Q6N&vtvFX_ZfJH_jF}qn4gJ0cJ5}9mTa8> z)k!hSGv@$_aIpZ!4vFu{;rW*dkw4y15+Z0~hhpqk?>QFWe97F_baLBTZIp;W9(k_$5%}}YD z&TeZOENX~f>qTsFkQFGYs7NB-dUOid)HEJmZq0k$)eSkK3ai0ZWbYJsm>bZ!BXMvr z@*YtBO@Nek5l07qn`*5+3Je+CP+*|*>&d$IT998vf6dTe&+h4j_rKG-%>Ep;dG-9X z$ay$BvYw5o3DjS{q@gPHDr)Of@05E+A$U7diDw+`Tl590M$m;jHSf#{8uo;D`0X<* z-X?NdtUvnwp!-Mo{*F+b^}a=Kpqt?E#6)J4$~lTPT~Ittur8gV<+NtzvHLCHX>M-% zHQ(JR*Y;0(>iCf=H{?vU<>-ZKfDIvA954&EfLcPjQ+54Q1uuWukYJgQ(#EqlZ9OxU zdqzKgcVc*Ib;Wzau`;IUO+r~>Y*8|HE!F?p{O6@jt5LgEojg(Xhq**ZP9!|-1)^a< zg1rD^kj(|6WIr58Eq7NQ2Rfy0H10QH z(OQMujv?$4M`Wevg|6>VeYR}|Y_UWY4i8a_DW{oy9avR@*ASaQ>SXQwqot|vAJN`C zy(CP}x35Ia&Pzzr7o{DTCx8;Bqf0tlWDBN?l~P_1YG4UZq3j;4J6-ykyY{z_}D0R&e4z>RO#G%Mm;XIjBi zZea)P!5(_11z-Xs=Mcd)O5W^B4*aB1ol|{3*#r3KCK-d^6i*GCmtS;}AQ4K)c@f4w z*16_k7D){#^UBFWaSQ`;^77oP-Q^zX1A@-HoJQ+wL*E~D$(-CddWn1egKY=64;-<{ z{x248smXCao^zX)?zc#HO{>~-ot{On}%`AvMg5F}3_a@zE$9M|Z> zPv7MK#4~@HDHUgEto_Dl5n(YulI&jf>dKX-HVuw=&lDqnYQ1^6H}YGb_%R}@;n&kW z|M(aWV%ac>8SPrx-8GKOJ{#G-JcaqNI+JW-(ufTkkKAV5uK7-Y88o7inp8TAy@tEo zWf7IOXZ2+5gcSd&%xy^jB|gIcq!W%#Egf(O8mweQT<@*j@hw_7V*sY(aR4Enm8}b+ zj`oOmuj*g~4MOf90z#6=!MZxMSM~Mv@2p4E2sMjvhq(ob)`96sG#&ir?O7=~|AY?NF!4WRMIYEOB z2P_4ACY1N!w(tRXFn90f(fka}1aUE1xBr`1+vjy8g+*Lk3(L`p8Oo#nbv!nv{N2v1kU5c!2;P+V9ma3$tsU+1TS_-NSyE)nvmE z?1$khES)cy7YEiX3a!{(^W5#x^Oyd+d=MNTHW$BdumJtQ1^SPK+3M-(%{=U~V3l+l z<@C!g#0?5ehXNecc)-ZSR-)L&6FN+7e=_0eClcbW54FU9BBWd*F3Wi>Yo0r|kx&|B ze?od&8rMH<8H4}_fHd+=Vsy?9?;l!A#{~kCyDP&#CBHLHC4J}3x3~sVfEj)j?(=EZ zM|mTgrt`p-ai2*hZ5p1iP3}*c{Oe>uCaC&#fS{&4r%WsCw3-%IT5=~A4pqK!?<;J1 zzKO#9(pkk*u`)Yj4;@mXP2B%Ilyw8Of67rI$)jG^Has7{a4N_?^rrJJ71o#{hC$SsnF4H zshxU-H?$hy4M|<)*RQ1Y+H*8KokNxkLcO=r-YAlP*)FK6>i#|>{N~q}7XPRWo$zLc zQbzZ-EQi_06MXhBG)jcGLqf{pwmvFmi7G%6 zudbx-hupIN$;o8|g86gr_Vy@nUP^ObvH__t9V+zZSM`L)-Fqbq-kM^$5f^A4)T->E z^P{#6{><ZMof1XNT@<8qc$wuRbq9sEavX>oFv$L#?(_oC9d(nJwy zOSaqGBA)l3j;yemb|y~^^1D?V9AIO+k5l~&x{%uJ3mjW_PnQ3P@{?Q_q*Svr*wRz) z@{Tn;FD2O`dXq_RAG7)>seMA8UMFes6YS z8=`=2;HR9L)A=$q^eAX~dAUACt8HvLHu{iM`9bm1#}c~4XJ0G=OCiAogb;@*oBwMJ zhmI%%)k_5qY&_7zw4j-1=MT{$UBwFrxpoAjK72d!rfT-XMC%7{{F?e%-24vK)jFkK zwGhw)E?ghWt}{I0crXwlJxJ^Bhk8)zg0=W*HP>+ls?8`=cJ}tQk*C*h-F5hB%zzv@ zV*ABqKrC-zEjcTH>mDC2o$iQJ!A#$#e*NmkP%+-y?wWn1332x;!oeCJy%p$ z7pM>>nhN-Vicz$L2ppr#ZaO)Mp#de?faIMc*d~&0WWLh2DCIj?6WjoKa(n+4Ej>Lt ze@M;S^8h_G|X=DVK}aO@HQ@$fx*A&`g)fyjbHE(3o;DfVcF`VD*T+ z>q=GoXs)~ehS!OihW0KuIS<*X=lzJ|rMB}ZKoT(-32ruXt?T^#bz!@+ZtUr=C{Sy9~>iN~o!L2dV_re!FeNqq&}K67EPg|YT6V{-oWaCG18fX-k?DJTnk%9ca?X#~=TU)NmSR<{I3g6qJ@){p;DY|#m1W!8a89CrHEd}-thHQ5AGrL`+Lm!?J&h%EI`btgZ&!fmJJtoxZ5F_ z{YQ1$LB#DEWt~FOfsF~nYtHc>UZY-RAjTDbrK(ZvqG2Ah#)YorSbh6OL)R78MwfB9 zLtE{a?<8E#{BT0qB9Vu7dC~b@h^gI`{C6R5>~o~TI-XSq$DT~J*T}N$m%VnhoO<8B z3Varlq>1Rfh2Ou4agJA~WJ^qIAobO_LYL9D%qQ3?53}_N&|rK%%CsH;_P*iuM9AW~ z*J{a5OuNtg@PaPkok3IF8I>mGPASeyF?-5jf^+kBCbG;?7mvfd z4?WfGo=*(Ix7|==^m7lL!N){iQ{3d8c#cn^48GfR^YinEYLRSOuz2v~lMyJ2`x?stLZ|;4Oo4wCK~^SOMkXX zGx;;rAC ziB}VR1QBY8jPliHk`K$Z1sSrKh z*5n3RRFui7sc5_`d2n=DVQ74HKQUAIVpq#*v#~!@`V2OEZDD1Sb?EI!)`WfUBMuyQ zFO6g@`lHygg+)N%DbbmHagVHuz4d6Fo>BQbyyGwH1Rf)MpUf%XmlrVd@-bdc(G-E6 z{4Da24ynBI%aicl{3*wJWU%prMmW2kv0*b~Vyvk0?IZ<(bN33*UNT7ExZQg|Tv2?& zMb0HRLi)_lv0T?=gVxj&%I!%8Q??Cq7fm`5aNzjBHG&P+PyHBJ}D92@Iwt&mK&0u_$=vCW^G@ zggD4W+v^*^%}lDK-JzkPcR(`l^jGdWAfr?$uSk3Na!M88)ir)&dQ~$NFMqcW-4k9o4;?S=$|!zM`r8o}N1?so zn0%$R^j81RN5LWS(MI+W!mlH{wI@Fo_fFb2RM9f3}a4<=|OgO57}pjk>&W9;iM<{_eqeYH4C`~jkwmx zDj=mK70wt&M>XaL_`^IrUk7hQm+Q^wT@0)!!7YcCXx!ZD@3Y?YZo!ovjXs8KfDZ0V zsj<&Jw`X+#jP&t>}Dq?ryu(>sE?zzBz{#HJ^l)0(QJuMfbokn?L4K@ zRrf6MA4S6%C3Br#m~?0Bc4--@6Z4;ov&x8KkWKk`Rs_I>aOkEx8qBAYzI?&wNXco zpx@u#jcd@WXlC7ByTGw20+I?Mw#3!#JjyWCe%^56&9Wh4-fxQ6A^*0 zur}c$vvK_^Z8qKEqW349C)xg%s^bm2g2@6Ov>bp}B~izqX&^riN6eZ&^H!I{CMt%{ zAoDJuP7T&2Y~OWQ#ji=uqAQDjPw2~}K-%+`>Kn!C-|O)T$h}CtD*o%8!%usq7?on$ z?E#~z)%8B!H4bgR@)lauE}<30e=Z3=Q#sFP6SQ7lIW^gaA<;(8hF{2gbGX{u+-ri$ z4;}G1UL{oma3q#!g#Lf*xvjVN<_C2n7(EbB-G_O$*i`7tfGVDeu6&3;JYzuYdrqlffwC1=lX z|BT1>Wo%tM)3!AAHv-1>4>8(TnK>EEEM3pG^gw5mr1qxr#6(^(yhm~nWhEuixgSez z@QOZEQdVX;*?Z2H`Q%q%CyA=byD8}eZa{Tbwm5DTIoKr=k(orxj+vJ3oZ_#!xp-Wj ze{i*vh6yzy@T99trUjlg|FvX1rfM9!)V8sy9L{ljqloLYmbpf-M#KsSJi6e^E+O(< zxD1XgiaAftP8HV;KTD{1K}VMqADzUbqMEPG7vITL^JjkGXtoN+_@1p> z50>7q&?^|4-z=VZK~?Kj>tokFA( zRzgBzX1L*>1&+?wnS&Cp%VPMw`Y{-^giHg7C$cX9PwPG&;_bdI{Oi}R8+bglN7T5^ zk2H3jrD70X^C!KcMvqVXgZd+HZ~o&=Cr4fw?_Q)H)h71K0I}mq_sB3Okwmgs~gNLuyye;Gkij1x`zG zr5+!uGbPFY!7hNs;pDfk5NqgZ0qidxzpKg{^=IJ7?xSF4pJExm>J{x5+Tx)QbPr>@ z<^|t_75;9@){y~o*_CxaJ>J@`brPG>KGv_rzx(m~nmatI@)thk6-^|ZQQr|?p;s3B zVk*xofm$T0sX*?+*Sy6to&zG&6D3u74x5!Py$d#T<+#1)qrv2wQ>7kS9|Ha8YCgp7 z4N%&7J2B;D#HG5=V?V4|EkhBvAZqmx_5_k=L9(}eu|Ysv6T{P0U{o9mFG;xX)5EHV z%gwaVx&A1C@NyS3GcyC#1AnnZ4&m|U)M9Ces?ihP@kUY%`&jnxzf07;c=!os_tUQ+ z9}<%plGv`IL#FvsQ2X*~R7Cn2kFDm@gS#Ze#l_u6%l&5;mM}6>#j4J7U5Z?T?b$q$U|0QW=X|Ml{RwEs*m1jYyTxu!OS=440TGUF8WRg0`Hsg4 z1b}u#DlH*|hvL+BzoBwg?uY%B**2g%MBU%p&5U>CPX79ZTHZY|QG*7V z!mYT6H+*w)LnW5{JXTQp{dbl17(X5W%$N+tUifiCR%F}<#`1z`W^!^e3XkxGTs*y6 zLj98!d>jJBqn_iX=MTHMoo*M8<_oOeQ-Tj=7+W$jGmld}Sf5a5#I||RJq5-bGI$V@ z04q@2f97lb{#z4vq8y*=(W6H_cvBYnY;}>4!cI22(3O-J?w(E#0|Jv6E^owFnle2Th27l3&oV~za8?GkEKbeti za5y0E;P}3fc#EG^`;c5J54$(MvF)I}KJ%tv+Q-Kx{}a&$0-A`UL*LMTekP;e;m_r+ zrxDmNjGNSFgKTaG)eTdo6PTTeFBYLF39c*4nAfNSEA&(X>N&kWf$bo{jc?RHM59re z8Epw9NZs80X-q}12~5t-34M*iGRu9bsi`d?H9CMyKW)UYENN_f7+vAcIaN1kN?wS9@(MmU*^XyyHeeEDrWaDb;ruZQFA+DT9E-C%4zk``MhQ2p@u z;G>sl%n45_aI5nrRO=5wXRp<>Utb{Mim|~*w#s423gZC2SVBTTEMnewtEAr&n}DU% z-y0T%i#xWPn=~ZtQcHdYkuRkiBH)3Op7Ww539RUZj$AO=O`dMn<%*0QsLW@?1ZMP# zaoOcEHr>0x#Ka_E*7_9XmA>NH&`^1bR&E*THsQqaaeL7@uBobuwiV6g4TRK?7p}}+ zRW(lX(T*}Fk{eT0R7BoD+#v7Jl43xrwIMSPK5XyE$OXVJcV?P*onsXd3C3EUeMQ1} z?_ZA*Ocvdt`4f=)B_{#K78l<`?Fq&!VcAb)#L=P;F$@qD=P&4t^y zc|oY8y?;h)tEG0>(8y5uC6!Vbr?84+D-HJ>sqm}E`m|?18K?H!i^h0vqI2cA(fJ~J zaM!OXhtf0cp-&5wC5I;?%}EPI2$S4S{83%e4EPX`>J32j-w zJ(=Sm$+rpda(u!7fXZuUa#2-Li^ag*me#I76O} z5XhwD?5VdB_T3-X5(q+8)~jtSBwi6?DdrwY9Ce?`;Od=pbd_WwpN`IceD1x;6GG@B z4j>mP)^4&_>Khlo19g>oCSGkiPb$Vr9IeM|o=e!q$v(SM;3R_bm-|Lf#@YGg-&^=@lcc7-Fl|(jE|GM0) zBXgrIXQzK2jf^)IdtMmQW|kBm&?~KQ!SuVPq1SRY+Z+9N2@N{~1LjpBoj5(SJL<^a zb&;{!osLRc6(CK{G-&Q~p(@Sqarq;DCy7dlkyBXf>-+TtJ!{b`E$;PKt2o}_XqLwB zjuj3S6A@2H=x1$GX|d!*-xjngxVwoX0qpX&zA6KLQj#G~6)3|k^zzydU_t7czVkIS)}|lSF(o|Z z(F#Fy>@U>T+zyum4qT3?2t=G)QX+}hs%_xh(W6IkM9!nGOwP}5{w}n9R=xKp;0n^S z;(N~4*mcpFfLXC%j?HTn(cnKW>^fAjDxm3g+5ex``p{re3JLau2#(I3P5K|n;A#T` zWsrSpbxRX4!_yp)dZ$}XRfS%OCv3=^#Df~k*PG3Ze`1pR6Ym)ZCzDtZ*s>K3VF~Z&u?4obr z6{HkM^jtjUcfV@s!KO%~S0V8$+{*LcC&E3Y0Ga20@lDF{xv#i!r&=K8#pFG^(>>U3 zuA`~onB;!Z^=9fRb+&vL5>O^}J{1>>-0nUV--p{o>e3}@)H3{*2Q@6#qe{+4kQN)t zXS6XE!VKoAdeX_UiJ6<4a*qvB_^u%tiWo2U^YT)Amwtlty9|h>@a^tZvmU}Ug#jz? z;-9?(+zY542{OQGIIfi}Apuzm>RcLL=mTYMY#)AaB4!afA=8KGTOqy2i6P;s%YUXX z>Pm!GdjgbVpu(Xc5|VuahL_?88V2^PUe?tAg*@#XScZVQt0XFUJbbw2z@fNIN1TuW z1hhtDf;s-iSOikhg2kWo0fB-18=}bykLO3PD}bs=%a?EtCsZrO@@#g!q$F7wmWVqY z@9dYxc0jM8)n&K|2HP@^k`7d3>Pfs&9^zn#+5iaFO%ro`Rr`!Wq^zt2h*%9ck@=qe z-;ZeJ2%J8Bnlvl>&v{Af#DBI1ml#o#~FL8 zt3Ya3Wfi9VG}J7p-&bzK%doA1ug+Em@9rKFL``I-Mt?Lee#DT;xdP!SdAh6cO)zV) zN>IVNcyz+1LVZ@E9j&kN<&2PF7|oF0Vm1jF#w^nBT%v`4q3)%sTYnOZ@Ywe6|74c% zra+gJ?BnWje;hedTvpKd<4J7Avw3`9M%tvmx4*v>IL{%q*Uj8VHJ3w~?C!n)sB+^* zq(*m@snDId(W~e#^%=hSxQGGBg9G_404DHPYtVMx^QVoCxL7Bk%aP8ei*SSSOSB$t zZfi(U#yd~`TBEK4@!j5~W*}7u77#X;2tFN83TYjbL0_JY%~P|QOWv|h08+r%blXTl zMWE{7P6;N71&75R{pqiDdUBMMs)S?$IMGv6xZHwHO;<^7UUDjkdOP37je%1&6AZ@} z!s?FjCMJbEFR`}`8Y+bX4PU!R2JYLSNhYwQ)g-H%r3s7dVZDzHEP7x>^UII>% zOM?-O4g9yYr=HyP>(_~i8o0_LfbGH74@dBN1o&VK+3t@Kz4rB0{xt`L{FZZ{eCg+t zNFd!4(f#2IBG#!tfB*73U6$>dt?25C!2V{!^r~l?o2kSK;7m)o{uL*>E31*l$3SH} zc=zpGmZ39iNsFr;{?ny;oR!-%<Q^;W}<#01z_ zSIg^mG4YQA*pAa35$Vq)IW;r$gS0rO+}D{G#kqzea2K7kWpYZQ5gmR7%v_=Ha~GL) zj)Li=#dL(ZCb#-6hY6UWsrcy;_mql24t`wMXf5PWmq?%_`c27m=d#6J+;9gdW8m9X zPKJ9Q7ez7^kZTi!!W6Nzqi@c%u*nU>EK!$de}`P^SFY@}u&|g4;j(;<;lSvTua~z; zN^(h}qM^Z_YeErXX7b^U+eEl&rVV6X`0>z8Pq4|!XwmeJD@iYj3xvv*ZJe(%9DEea zqat5jS1-2_d~8 zf`>(}dW2e#NGpv0JZ4wWFR{iQ+KQw1@g)69ooh;3iz!hd02xX*?+`su_WIVlrTKw( zGpR<>-V{MxR|5ds;a9)eewjH?qSJZdmo|o)5=;g@i_3q*xSXr<&@&KO)>E#l->-Nt zpZx3MaISjhNp-*M(&#M7Fhz~4MZcC=zB=?wMg~6bU#+8&Lt~c_Fd>Tzhp$6X6YLL} zWdIQxZ}9uxg=2v4P9<6GF`5DT)13LL_Ne=Nv5A(F9?3o{qu**iAP^2mi?TaoIt%Y6 zoGIb!Hg0XV3J3moXSkj-Qx{D(c4Oy!NCbzagWD4b%~_sp6krk}FDt31>TFz%xdP5> zrHMm11lAfHNE}*ZrPDI)8KBlY-@W5S`j%(ixTLkwp%98&cU=CMZk!Zw?E3IQ4_1%> zH(ZC$`?rS|jvR|h%uITec!RH_qf1{^S;q1p^~6K=bMhO`73&@A8R|ccmUIzn_*d5p zm$lR%+xZPP+#nzeZo{kj4j)ok`NRQ?qf@+yYYZ&8yJ8s8SUmLhj=^r`mlDpcZ~kV+ z8$IRLcn{FYkLlo|kuA{d7L4_ogYi1q&dkGD&UP|=dCQe;HP=%_>)iJOWv ztTgOwX<8XYSt+G5lM2}yc1R`3v9d|X%Ew2t9%BIan zZ_6Q#$j{q+QT)lum|@?0A=h#6H#%R+69V#TXlwcDu|}S1;y+jBEnY3@gF&g`S2J~^ zSJNVjZ`)6gDIW%}bD^gcvzymdD4~{n@;SxeX+fBYar7PDsC~P4U#(ecBv*|KGYW96 zK*jHd_eTIx`Mx{d8Mp$9q43TPf=Ud_qcop|h(I^HA_2U6|>i3j-j_83I(c{zl8 zE0f+x04}fZ^YzVagT0W=H<4`)L*GmY|7=P*1;uOp7Z_hoixkXQ>of> zvQjm=Z9RYM!ge+dYrY}b*nJyJnAwcq1(}R^o>-%BfvVX)v8W+!sn0RXR$SjhlAaC0olO0mQ%k4vEP7jeOF?Piz|AiD zI?;h|$py-Zy!?C+^yWJ_wcN%*YSYqHbPxG4ezAquh6IDbfO1Pog@Du#2nktW^u!>x zFulw~+;{bM6U_8P1Q}#I^?{(LF#%K~rhL-OY&-SgBEC=DGHyS?q(bi_9NcwpGIe_E zT^ie?zYOgmr~G1kc|-$8K78m~GiWCSbr8R-Y&e?P)l5O55a8*Kpp^Ok{+@Xyo)KR_ zt82h(`V6#w#%w|2)pX0o%X-&JG1feTqX!*L?CSbdh$AWVU24ms=@?#sx zZH?t++srGrZfP}I`4S-36R^G`BMvm+oI14;7=D?pKHB9@R4f!EV5j?pe_pR?vX!CC zOQ;-LFtJFj72X!@&Ll3s;+RXqq(exywadL9I=Qz7OXde_7r96o#h;FfycjUh;DmS1 zijURQpYO2RNGA0;xk~7QIiQfj7x=#Ma9*ym+Z5g}E{qeTC4pMPm?L`QLJ)oTfoG5u z806ci3sj!=p19hYp36j6c;%nu-n5CyqBm1_R6zixpa402mnu593GN5egn1GqN-MH+ zkUujoeTAVsAVa7T+HiVz9?8VuHTW3yGX*qQp?nKNffFAeG(R;qi0)Q|#m3!= zKhO)$k!8)AQqbXB6DyitP5>h!yoT!V)h?#AaKV^w4;*ebv}wp+=i=wXR*e zC&Lb~o9&wv!4%<9epkK7&P=_j`?Hh zP8WO;Mj}2O*9NyKxe6=%esrS)4LArb&zkeaIAFi?Nhr3%ZVO&6ycO`P{cCf5a;@}) zBQD58spe~;%n|@h0!3+)r4(_M&kt>(Iz;yju!Y6no<0%Jqzevt2Y4tL0eD0!3JX18 zWDrN%u)ehay`Aayy#}PEJZeohKpOqy0)Y4IPrM2J|dF zJwkaQI>)zoNf&P(2PyDiXs+id)ULf?lTwJpB?yF5%xO%(*hTKyDP!Nsh zKJ?PQhN0bx7+S2_PlXl0EJ`Z~Fewn>1hJIv(=~YOs1rl0s#40Hns1(pRUxXRw?iXj zC9SH8P- z9U#O4z=jci;Tg&2YJz|zPx|r8e2-ZD>DBwE2>GF;3jlqbb*_UBQ3fXt>SCxzZO%m= zW*NdxqM-U+UJ(}DXtqb`X~ zjDDCMBY9_u@}guI)swWJ9J)z0KRn#0`fSkkJ9uC#H~o)pgIEag4Us4`oCepY#=5o= zJd1;-j7aXU+0Lm}ntME-pp7HY0=4G_;gcz5Dmk;{k2AdN^I2|=uUr}M{>C@YU6r5W z?w*3mGcL@jBtm)*>19PW@v9WDH*rVa2s&;y_FLjxl#s~O1tCp|%7oVR{WexY5@7>g z0_h0cGO+e{eXK>xK4oRn z4M95fA&#*bZr5GF)YYRaVz!t)Kjb4A<}~bye_bchKn*}Ghm_@dNNdx1y0VT^X#@@= z{v|=dK-sz8?RWR~i~7!qe>4ZwCn2KAcaF})Kxodg21pFyRltQ%4ifT(?5ivRdQBCL z)Hg7u`_uKED_an6+G(e+VVMj*+q^ae_SMhlT%4SCe*gO1aK@o6D^gpO(6pfnY`?7G z#M+7KqpYgB?XvgSI3y44$azZdT;m;iJgNFwO*bH{Qf1!C*=Jw~Yi_;*kt3EFzfq85 z0hy04fjSRdBAWMi&V&jom&OP4O$kKXho`3?ze^ z;&`1zu;@#`XX+Qmimok#s6?~V7>`SrI4%C%X!#83A{n-wZ?SPr8mAtwXME+t>=xz)wr{{rd;&xixXx_b4Z7WJ7kCK}4h%l(me zdJ!jb_45-G>7QzXX0MTV2_YdPqP^NjYUwGnFpK|{Fwfr0eDbc8)n!1q>5S0N#Gc`{ zT-_}GaDFfL2+ka5n^WOm6{`h~F6_)qWx1U%F()kUf}{+!Sqn$_&b;Qq!hV#a=fK&* z)uTVo+H0XeG0PNOjp1YVqZC24hSNpU^8IJkXK}e8XiE+HJWqfB^4oVy&a=_Jiz0Q2ECOc2PndBWS znY(C?aJ)NhqZrkb&PgGYVF`(qpwjwVROlc;#pz?yI%MC1YZFZBb=+@FIb+QY*l3Tz zJd>wg;!YVYAA}3@03=H5yBna^I{}2>AX^440Hx4!N?x9Rtv!Df*J+eYKl0%jaxs!X z4W1r*WzNextH7@Zp3F~aR)*^bzK3L>n0;$*HmucU{x$P+yl!vMaQXYN<$D@|f{mj& zHWxA+Oi4ZUu_e{X}9W(D96d=Jv53{h|zXJ~%M)j%Ka@|7!O zbw*bbaA^GtG_Pphg@N7^u7;O@YABt5Hnv^fn=i3D&oX(^lFI(N>f!zIvrODCRJjx6 zqK~SzhRKET>dFf1Z{!y_ynJ8K78TN9eR{+MIT}hT<}Zl5Pdyd5=s7^+j%Xl+ z)RLhm4Yg8DtdFrDow-pfYB54|c!A=gr5PSJ1^G$uW^f$h`vp51QAk^_L`Brp3 z0+{MJnf(}(5<~@Fn8q&_gh^Nj)NmMfNWrBiNp(xSBumRxY%iE ziG|)7vbEC6f$5VV>%%UTm^$;1dTA!2IBsp$^}~0W#yDn`t`AwZ7)EFOnigTlW9G-M z+HPJ7(D`!pS;@<8pnx#ic!FhXd*?KCp7uHbwdW`tI{FQ_yCE=yWb?W)cbkqbIf(aw zYhx~|cBmbb1$AtM>k;0M8POG;E51Si)Ju3hfErOhFahs^gi1gK#u{XgsYN6Xrao@C z4J~}rdqA$}X+$mRW$E^UU|xLfN)8U?M1$;(y3FnL5UR{>UWzgUMzxHSQw2dFOEwo$ z(dW7_0v+Sh;~FUsWh6eRI@}p(JR{^8UA-u91C!9K|Dk2T<{UBDy0?}`TaE7{r$!b} zkhTP;>ME1#Oc5^fG*<#LXJuiz(dmjg6jb{|M7iM`s&Bo0{WpFrDPmB_X`n+t7u6gl zd2S!Nl2c$sG4uDQBbnjT$7UxWHb!u$V7lV*>#&qMRhuSCI?qbq8ssq%TMC|UIY{l2 zR_m_Qg8b(vK{e(?(Yas`Lc6Uq7?HLrQ;3zT8i znGZD{01NSl-a%pyDbe?}hodKZU!qZYAE8iA%O`-1)!r-5x^RKoI_@6(zsidF0re+; zAi;L`-o2{}@mqp2IS#0Xebd2~pf`LTx&MG}B96~zZHZ3&a*y0MkaDUiD=R;p%LPIf zARs;CL&3#HZ~q8|X~PYS%@z)B6Y?}^!qf^RyI5K2JU~ZKg805DA0zd5Ox*XEcP7x0 zP>>b)!kgI710=l^9(9nngZ|;HMFflNwJ09NR6i|nN7$}mYK^#0(xELlXkp*J9S4CC z^5EPPhY^^;$X|e>psb=&Z0iF{lZeO-4C>#yyO%s-utwl6##_cCD7dil!GqXUEAn{& z_Lp%b1?*UElQ*HlDz?YTPFJwckApS5FGfnUbfHPS7j8IAKFoNwxUbOG!l$!r*)sZd zghDrDqfoKoFhV-`_;lMY^*?wr+IaldK=pYltnkupyYL0A>c&bD0FUzqjS4A@0oTkb zrcfDBsQA5T2(#f6&H(s`2W%LgYb3^L^%ZF}0XXesgF#IauN`h+$aLH70HfOsl!KR> zd*=20lP*CA7`G|g7%e}FzIAe2xV6k(+P*T5&{3r)hc${8FIey`P(vQ5%E;5FDov@= z^*?>@U+L~n;J|c*$v=RPLO7CoxYW$o6yynf%{bN0z1|6 z?g-8aHN6Nx25)34RdsRZ>=A%u7Ki7Ov4xy*2zPJp>+F69iAf&WeBhA~bNwM4Tq3kd z0IWQ;e+N)a$zyltB{#Nu=AvD}lZQxQKLCN40}p1QGKG|xq%L|iCGU?pvNk$Skc1#; zhhc^QA5pDZ$iUD z@7);TC40mh&%O6gOjvJV(z|2mK%Sf`zLB_CUTHQ@`@9>G{Ffq6ya@T(|BGzZ#*i(M z`gwcStBTGR6sg@xO4bmbsulvgON|TdH~94vO%S2Xpy`S3dJ0cXXBwtx7FJe8#EcgO zZ;crD!R|*%0^Kn(J#2bZd77q8+l~V{n7$X^7EJ2GX`;Cg;Dh1wLze0KO0vly;3-rD zq8bQjtV+@`YT$ao7cTGa;4drt?P!|iCEy8nDbj6h(5&9MSMKzeC!wutYiWK#{DqRG zYC}RTEGN*CIaKV<|3SqK@XJigct1`Kjv@>Kz^~eu^u{X$QO}O^0ZvP_`FZD?Pq-*6 zAnezZZ!3*6F;Q}>MaT3A6=Ua-p(p5=maSa57smpCusk4uC`gz;C+G!Z>Pf3R<`!;$ zJuWUB{V2}+_|vhL80swb54#ij!bxL>GfB)fpR9zmwXpB9+MAav_7 zd5nh^&KKmKj({E)!>QouiPJ4`L9P?+zr;b#>K;;n2IdE39$(j(i-lUs+$DW6dU;LU zg|cc&KHlDuP}gD-`KY|6GNJ(=AsnMp1&9jS!*`|u?jp|rI{mJOnJHMNYX#6MGlD)vaWGBkBYZ zQ~0{gFBL#2CvxuJZE=i}>_v@EH=coC#KgB47LkW#vFl{7B}?{!ipCoahGQSZ1a)=~ z&V+$xw%;bIb*u}2F~H9GVNs$`mSX-3>|XQ+IKZnFDTK?YC&bg8R6k*#)&X7WS6J#l677ZSV`DxV|i9^iQcTnad`> zLSg&~aMYyWADem<0&a|2YD(NTGTnf}A5gjb;i@Q)XkU=pbOTdo)jWU>UHDjz>h9fn zOW6LwcmhEcA8rfvv*c(!|VPI1(M=lp&fp2TMJfR3FJGRc#i-h_FI7WNmcQG zd=yh6EDzrzbiQD+!$pcXd5~4@=X8s+9Qrl@q-}%#t)LbLRLs2?TP#V>wz(l};= zDZ%XDXwNP5mi0IH-2iGD6L*^Trp#S+TtWbFETw$b zCrrAab$_C>88@{m_Q7CY6NlmhsDWu|XwC0ygQM^7$>FyeS3G-m3ywTgB~Uk_U83-6 zef)gg^34#u2x16?nAg6yp4t|axa$ySVX&_&*t2&p9Tbc*?-AVd777X9DhI3HgN)li z!U<(ScV0liDzBlSx8ltWd&t>IxkYG*vL7;ZQKmF-RnW<(u!=aCewuH~J(D$2?2!kz z60)PKC4AD}xL)z-0xOGsc6`>GMrGSsH7v^<=T_=R8+Lzz|J2$&(*7Q2;HDB7vm+ytI``j zho2l0MF!nkG+A%Z>zI9voD1fRRR_RFa7Sk^IaG8*~Hua0>3}^)CF@j^O`!Y_s9msH$ zMwvj(=IO)=A4tzgn1(pyH)kaqY+Yg3h6^hRB9iq&%F(@1OBXI|QkWWL#++0#RQDMR zEninuumzXw*|%>YTt{S9A#lM`O3L}y765exVK_VJ2+yUadU(voJ7@ay+Z%G>R(*-8 z;2MB;klfbk4t=+Jt5;~d&`nXxMJ7t{Fyd=a%>J2KiuRN=fIXVy=v{gFc^}9yKCsKG)wiUs zcwN{8AvV{HgWLmenm_65rxse*IWF-C@;-SGzyTrvcEjvN)>K>vWfckZ0zEu`roaVOi=1Yi z=&jGD^*54t%9L$C%sxR!0F5p>^IrEi&j-6L@Zs=Tjlf%wQ))%CNfIZdcA#cM5&XB} zdaL`Higv_408SmAI)g_RbgFGh2&i_9crRdAYK1=!&1&87* zzkhkZ<~|q$3f;u)zN-7Z&E_HmA?${l7a=A$m$j%($zon;$6qF+`yOtnVexFiVhx3f zJY*5lx{0XbEkGzBg;=&`jVehI=^jCzLCzeuUm*O zBh@5rNHaShTMC5%L^O){AyC7I`lVMLRm01>|bDYJnlD?(DdLY&^-haIf2u(`Q8FyT-=cxMXJpcX43&w1#3KCThpfqKu$ zI*Ex8nmI5Dpu>U7Y4)eMW8d$*RpV#FW5Lx8eBC68kP{i(jvGk=n>+ybspHO&1|N!I z%wPa>$74h_D?olYWCvRJo=#iO6^(FB{-FXsLX}PZKHhXCOHGaMo1tUFfFiJQ;GICEJA0v#Vv@r}2poz>=G~`#WUj2}A zMt6LCyv4>HRt#e&{eBbyD%=T95ATIbpLjb5OQNkyc;5|Ti_=d3b#gvmq`<&|u86#G(-_r1tl#k0 zx8vdC3K)1Vx2f88e8dXX`~B49umWC{%lRio14M`FI`PC|^xIq&X9PaEhVWiYCaG5| zsRyTVLO1c=B7ap`ALtaN5$iz&wv0>|Zdb@@P+K0t5kNH*DM@DKZugzK+@m#L!IJd2~B`XXqU^9)7_)cOiS4#nci^ zW1$ed1IFGU);}u_Gv2-nXJ)#bn>@tscr=>}Q{u#?%G;_y3wN8!3JSAdznp(M3f}C& z?iwHz8Aj68O$HC+@aM$&`S~kEy(3vzSa!n92~^^P>+d5@TR&}E!eER-FMv+p6kwZq zL-ct7h7?Loa6W$b)n3~aYkOf$BIpfZ`ZqLJ@NeC~tcZ9Qq(W(m(bzaEr)#+w)=G;b5w>#$Tj`TkCdfg- z*N;2D?5iwp51shR4!cK%Ylenb!M6v}qYR*sI$mV!flcq8w0 z{Ii%q=ggbe*0*rFx7cm>R~x*01*|N&jYI)ZImZ>eUg&GIfN50XK?XX-gfyW^<^7l63;4 za=7R(ot(+BGgzUe4yT5we{SP_uo6Vi70k0!tEg?O0VRb@Fg_GMNx#EeR+Y!9^qO*+7w??bifn7xG<=ml1*{|>B8l0=_F z2p*O4Sye?;@Tkd`S$l_jjo7pEAw)ThpkjRbah zU(~w!fv3k#5b$^b+6YUPl>GIwoj4#|>*;A?V6jm8{ryd$F&OJLV1e-3?5fh)R4T>mHgTZ-4nIenHgPpUq^sEoN6dLbz* zawD}xX!;H&8pt)vI|bi!?#ph&N65>|BVsePT_jaz!8Bv#nfVvJ=6=ELdOjb$uP+F~ z7l@}d+@4y$V*aFx&$dui^URE)8DaPGV&eoUOqvotpq`zTX8iMD1nIA-D8aK}xHq+= zmEtYHQ&O+9qhGc~d#t??g0|x=MK^aId8X}FPo{m8rj44BdFRTxfKS5?2FTRaeUub@ zz&~xYuW8B3B3cIM$5E*Dclx!Hm>-jV@o41LwD}#3J%BP)Fp1*37p-TjK-OIHsgIWd zWdZ+oG?NQsk}GB;x`!TKqt5CJJ>xk^rvKd zZk5i0tx5I~a}z>gIT{u)#XgU3KfxudsN0|Xmb*RwC#n8BOi<{;hxT{_PK7?91Xo~|Vc2e{0R zZiQ!eKFp3Ei*OnW929z!#J256efAF6NJTEEO_|%@s$2gk=P5;s0+&Q^TkDG2p09$+72fm4Js^*Rv<^o|n9KB5H9nl)?P zSk(!@f55zPv9AYDUm;4aJonD7uHPRNOf!JPffZOpxv{FYi{rHRXuXMZxY6)k15HS5 zSZE;xX!T7We&j_L1boYK-=FAiCaRgCYDR*A*OzAZ{R+BjqWt?k2xPV0yXPQjYx^2q z2lv9$+@>y2Gzr)>G2T5FAT7r6 z-LAXYv~J@jWb$sRSd{8inS7^PzREtj95V)a8M|wSF7}Yv5YAJ}nll%$;Rp8QLi5zj z6b}o)HiW>TRHw?NI!KLU)+3GV_9r*Akq1Yy5a7mDp&9k@j>pu87r-5(`!hsB;;7}~ z5}FCTomoCxtrq?gI$4;U7q6AMh3~j!NUYZkUjX7|kiS!;8ozVbkrM6f(`~u8KF7UkPt;5G11C$O zrB3P$i0h!90H3zETIac9%jYsyF${irP+1jX%gcTw0&PUl9Q3Cce<-ztP&m|TixKBJ zBD5Sp?T%e^(AIS`wqMWhw8{?GHu%gYfC&ia8FzawD2=>W5J$cQnOGQ#u0{~Zj(V)nw1-62kbU!0|`Wyw~!Ey)R#>_#!Ya2 zA6C~khL=o=)k4D&M>P<#E&K>F>zvK`Ev3ND|u$m%LAY`>)T$x>^Q|d)zfg_dtmoSsA}&>_&8~C3*~d553ug? zoL!Ck_HM|->7v7OnC}lC%975T^sT+DR{c!xAQBXcAd9H`_v)E2a%lp9-~pZdvaL)N z+pe(pd^S|nZRj@(-rK(ZdXyQWsa+T&DV+!6L#j?_qc{vL2G`&xdc$~i34lfV+FqD! z(5z=AHb$D^?L~J7+#PHNL`YtvGMlsK#L8W2yl-;?l~MR{IkV(%Ix%P z3egfhJN_nISkv=2{06)N0!gqUkUR!=AplV5Co}B2_knLAl!ePPZ;4t3>_MwYxBzC| zy;5XieB9@B$iXtcb~xS>iwmk**HKw zlaA>m3J@~HO2si)v_^UPIYk=7TLY9+yyuR5!71?OD_U=SIq?9~bQZjo!1>C~J*|z& zasf~|-${IW%DtMF`QLo9RGF*j^A(@;l;$H5(fBG6-O2$QNJZ!Fify1U9AUGt_`0FX z?!65-ad}L&B${8}J?oCaa&0bC$vbtIA9db_$&7q0AU{lF!eWnwz;zP^wRgduxpI!= zmo~xAKD=~sSF^A|mNK3zg#34{6tcg!;kM_LbFzKP|HONKIT}&AipAFSLYfG^;g-=R zZtI;4#?p1N%4j_axyL;e>URd}BUDxTmJTI+`7IopYo5BgSO(7NP(OToAkLvgXlPw7 zFH6zKlpTt9(~HT=O}9+@r-)S)PvF$Uirb!M%R{ zMbHf-Z9wr1s48(;j|LM+16soA^T!Zh#J=&kEn8ZH{NB@a?XAIwrDg=7UAz@mq}srI z4czVkkVKVJ%a|VnoqZT5h(5Bs&$J4)4$O-pa^Ku*P_5rV`X2ajH2+E~0@zMMdFVl%|TA#1Gw++L#=IId}~aWXcN37uNuRh#)Sn1 zy`dN4xO;wZj+X1_W;ahJ_bsMjk4XQx};vm`y-gx9Y3R@WqqGk0#ga`dwYUgYD|Cc#dj({XC^t z1$i^sWr3RY+TBB?(9CghoRJ#TiXvkI7ge=-kv_(JsGOVdc$6%^fDd_g{2$|_V{4{?}tOm_>_P?1}8iT4-XG*ZMF^DJ8ts;ASFJ`1B~rA09tL7-IzDid<1hJ?W)zeB z3<$+2c^Z`qPU4=E7IoE@kf>|A(ZLmuG5V&FM(TqIQLttjx|E<+a z(9vd`-*5ie@G5Zj=Z9(q*K9|g2s#<$9vW-Mlzu+ozLPfW6VS44jfQIdeJl5It(Su!WFv6#!8@gu^RcGIjc{aBxuB ztv(H<$|KurH&DGgKWUm;MB@(B_vEaexPa0{qBnZT5NZ9e_DBYGH>8DYN--L?D!Y^} zhByMlKK|>77K<`aFdc|ORDn-4?=>Xk7nARcKAL-o{DfaF99}YKv-jmXb5mSDln?k? zXvPGlsRMX)HBJUSxA3@Uam+j&Tyj3w?qyR?h{y>=SePjnMOiG0C0$vQ9l|J1G(NZ- z|He`kJCuLeqojmvg@^+{)-bNW#|#7MXX=E!y#V~8)(`?wPlpZ}U!S%6(BT9fjJ9g1 zk{Gp@lIqf^G_w>*)`$!V#Cw@!*L~zjvdc0q1&e2Kd6_3Ym1T;rJrd`YYSNT&+`0md zj1A0){2ThV3iC(st6VQMx}vye;in{f4YW$6H-I@To7b@380-x?y{l+ie?lO05I4r+ zsd)%iLS+RAv=mD}wnMaX!z}nx7Wxt#Py+9OM^Hd1RJx(uoJq&uT!yXaDpUZpru!p& zQ~(=!Iki0Xp@s5S^@o8Z(cX~{0Ii!pWmO+@9a>IYmCXA`y{Bh$td1I5@6f+wi( zq1VFSIJclr?_)cr(Tg=sgTu)nWty$i%`^HADG&tX^ZWZ5iBMJAB}gp`WDRaIgDous zYGXa}7{AD$Prg+)eV87=|v4_w6-YnK9W86~YKc4I5cE zysGrR1Nk)F5c<9@mYIjZx7=?D!j$Ffr{xSd+vsfhskb8hqp`hrEgWr_jrp zfQp?ncP^)w@>p1FWLw?skPu}Q2e85Bg|MdGU$|lYQuO9gRI|!+P z_6XR9jD3&Kt5{hW#-ZB_;*um>F1pK_gh`Rz z4&#^Q4{)uHN$ahuj90K_;`PU&Z^n%w;@A27>0ss7q1~!?a~zK}EO?U2H2>$&kTDQ$ z3TsMrajmMdMJO%Cwl>LNgHAE@a=65dVtG^{p#A#FO#yAuse~R|nN4~xw-B@5U6cB% zvf`FdTpT$A2wsJf2#w!-bT`OzK>2ZZHW&X>b<(E(0_%`^FgsBe5Wq4MWtlPC zEON-au1XWyyzW7mRed^m5L-(`pv8cg-RJuZypo_4`Y>u~ zfz94hW+Tv9K0s+%6uu6_^9>yaI6_aN{*Zgu-Y3YVj{YMWcp0Z328h=QYC+TD?j@OZ z@1bhJBxH>5Ih0>^Y-Hpo_`y%97ROKtO|-Ss0}=b`6O@d19qmc=Swg?*lCHd$pWTb> zrW072NY>2H>^tJL=Bm9Zxy7syuW)Y#m#Q~!mazw>93SX9rJ6lc3r|)ByaA2pHJ&r- zh*U2Qv>Rl3>oWG<0wFI^ zZj*}ndP6gM(C3tc(M@7MTUi*6`3Uubq7j?r$82b&5+(0}Gc_L_>7B=5P#fkesSfQc zzPsjM7^KCq^Ho;AF*gjv(D&`eP9U5vXU58w=>>D)k${{F`qzYn2PB_0DvZDW7y=9s0+)heHDl&@zuLy=U)Rmye@O|ZK=JBz1~Y{>tl9Ks@#pghdc>@EA3?uYt%}k5hRXr*|M7t zmZ!ZxD+!B~C%s?LU23dknrU%&o! zg4pAA%=*ZAASnl61yR3FQE9mgI#H^e*sd`D9=vFJPb>cz0cqC&&oZrJuS!}^qB25? zyM2FO5%sc=h0jPWOf*}Cie-uq#V&MYTgh`ZOMhJhKuQp~|3q!I9PH3+b`^4Ud4Zmq zh-lPc6SE6{PL}!Z|Bja76llGU%J^Wt*K8to`Lm&M_6D&>p(@QGadF$9Y*?{;c|PvI zLr{EJWv)|p1s*X95;PVlebAAmkB3q2L>&8#sJ5+v55WAA#zY8&+3?B`ws3f1@^_|C zh;E?R3d8r$d0P)AbqERtWMyj?7_{@rab$sw`y24$D;SM}cRe4|t02`nV11$dnch3# z`=ag|JaC=gc8mJjgn#2%Hdp|p!{krYK8ikyj(CL?rDNmw#Z9S+go!n}2R_E1h|&nt zC^tz0ze6`BfqY{H0U1F8g!K}ub|CEqOg{VZGlalmG6HT zzJ|O7F$56d@pp@JVP7~lK3Hg%sZOmGCilT=hWLbcy+OUxL=A3!)~&g=bMCDJ?c12_s2a-n&jJ3de!Rqp(7NgdK; zaQ^QBA|z99v%}a^8Im+0aOj2Tk?Ag~WmWnMXb3?Wa&_USLl6^%pRQCQEyodi?1>(j z6Ou#`1%!$KxUwzkTpP*t&>GFG4OlfXF+l;`2z;ASWGxnT;ly(=wkR!^6sE=lvDroo zUKuN+%ry2}LH;feK58L1`!cxf#}AW$N1R_}*i0^6AsOvHMt8FtfFCZ7uxG zl~o8cFU2A98p>aUMG^K4@+0Fh;XiicJ`nfGyf7g{Ed+QRQQS`_F`{fg>o{qFmTO|{ zJWAW1|4se)Ge-w>?fIQZKsTqKCW3a0PfF^m)Sq6BMfyL1<+>R# zX%XRhwl3o~yvh*4KG)KpvXJ9gNJ~xRCr1oH2Z@!YT$peoQ9B5_I##he*qhK&9O%3I z%@2d{-}a365X7v*>MEc=r+hxv+-6?2&68r_N(ON0V{$IUdMc?4<7vxE8I0u!jlvt< zi9Yw9!1nLuHNZ}Yss{RH>%0!4Dcnjx&H6E5m}V}FEeeJ78Xx7=;z9u9W!SK2G+qZE zGQMJTCi!5nX}&-{;>hSI?b85jN2WrvFHbme-~!!tardmBPKm)rfgEmhym)U=74Z|= z(2+jG-wB+Uw_^AiMlUgZgP0mjk}sYG{e}xlvfz(71A}gFks<4_BqT9p3q;mD)~Mu} z*+<2Pg>otA{Ucje4~Ws(UAw3>eA=Xw*5iqg6X?F3oE$$qJFyG(Q3*HvfK%uOe@N%p z-~&U&uM1H8YZadNn*477-aDaPC(VqAN0b?bHv=r9fl+n)+L?ql64L-mrm?eg6BG6a z7*d`W$xuwBaga1urI7jiC>Ifmi-4 zU*~tLZ45>zjAxLK%n`{*?FjY-u~-Duo0c+K9OBdRx8V~IXoj;Fz(CuVcq2dbQ?voF zMD!^{&;Z3|KRWkGJt90d9%>&-076&`sQQC|cGU05s(>Ct4l?z>&Tw7!nCaMx zZc>oIgZ3Hm0VU-Te?e~UzaTfWCwMI}Bwb)+hGmldq9z~Kqpa#~L!mc*ck2Xhe;D;j zv1@I^DNtF^gstw*;}&{7M(cV2#V@{H^rFN(6^P9Uo);S7hs%Cemd}2MdllV5V1gq7 zOgM#r6s^we7zA7)g8Z$AiY5l1W>->LA1!#nG-%%a4u)CqNWca85`)JMtU^`|=5dC5 zGbZvjqzC-NV32(SWdYi{(of-FO`6~VY{%gW%u~hT%icHmzx1EiYLw>OEi-wDo)Di| z3sM3|4yfOtU;xBvIJ=Vrg+^73&*a?Ox$`1#{DNG(Ak)4e6ilnCCQE6?fj3deZ8>sr zTt`9mUBYHpg(6<=io04iEEfK)8O^JN|FR{=k26tJ3YrxRf@PQ^oUfu00X;-)335%L z>kByHQ+b-IH3{pGiiXPsC1C?XTwu5|l1YTW0-yFVS%jbnb<8nSTW~WSnr2K*d-e78 z-D@BO-i5%8{^N`$_5J(yHPNsK&eSq~X>95;jJUECkJ$Gw8suYaYu30_1by2M3wkF) z>;9}t;1CtezETFD7=pHbwVI%=_N9oh$5U0>eBRX$nl&pxFnUk-DD1^qqSzctsNKJ=>{DgcT!hqlrQ>$d=d`Hr?`-kUyY zMM8f?#eEl1Iw<`B^CqbX*Pp9Og_c}GVmUHJ@zNF$QegJX3R`#}FCa_f4}+@N$qT@UkD3@w+6=K3%12X8+20L&v+JLx6nz1m<;4R@p*?v}0Db4p$D-!>qV~ ziK)?T>UYF47Wd@!&+E+Fvxn=O@pcA$4{^MY+x>}t`ls%Y>gf$BAhti^7q|eAbWK<$ zz6?t2hxW0h;NUf^6p0f=_l^=F4!hfG{sX^%{|*cYIH|2RnhIb3Cs^W?d^m2J?J?uF z#XJliL=>+S(dk3KJmEJlC4LXW?cd zffp&`VD#E9Y1*Z7A5{(W*t>JPhJD+l;b_23Vt>7)QzaB%Rx4)0BI~9YoN_2TNxKcD z5y}AEisn|^CN#3+qSw9BW}k}(nQU7Xe>N`e^jRa)aDdu-Q==zXLjnD;^zM53>W>qB zy*6VHg8@Nf5!8&huV)dh2Gt65ZOCSvxnS{P%2kPrn6MOD|Tbd<`=uccFv z%Ya&KJSV84d^J1!tyV#yu{8Wq3UW?ucIt{E=L%bK$XIhtj_VE}4-&I#oj%l>r_TO~ zuWYvR<1pF#0wU460HQ%M!VvZGoPeR?k%P8a*lj3z#|&f&j%zcdcNm4$14Q$MAbun1BrawE^T!+>?{XerI2!qN+tc{x zhyvyn>)Y1ncSu{PJ>p^B#GA+Io@%}uHgs+kejLsdinWUT-g8$NzwVk%WIrBcu#xIb zJ_sJ@jioUe>e*MJI#d1#%=oig5idGiQH@ajn*OidifM%uaSkz>i4?qoB+lqRzNg9j zL$fuOEH+|T6!YusdxYx(qO@gOkz+GEyihbow~#d)oi^I1EbGg++1B{EX(~IrNJ3 zorCwo$ov>|9hq@U#_PW~I+U|z9 zDhO;L{v}Naq!#H)fT@u~me!S_;Wc--1TkO`p0o|#H~nLoh_Hf}X>jlnB#%_j(KXfe z%Y)i~1UeBPNYTI}{~XErp=k++wbfsK($o7kS7DP`f%!gu!J%2OAFYve`zJ-PTEwbWHXlH!h?%U-RGM42ubRb*M1Vld7Xv9f9Ihq-S z^Dv8}@;_cx>ydFfT_Jh1RV`Wr<!o{@t^2lzJ{^h%+Sr>Ts8vmC_{<#RSjPs*#70Wq;^ywRi{P!&!jZ$c(qHu$y> zJ@vHc36a_2OMdD??*I;@f$g^5HHQOye0%R0;{nY36qcjXBkALX>_WHg4*FhM*+;rDCVcjW<;5Qy&(t{FB<=yrB~DT zO0QW;+t_eY#cO{-w{Hm+j=aqCH?vJeSv)H1?qzZs|A8)nBtyWaIg_f-^0Gu(RAfV1 z0RF)=FWg!XKLWarW|xZ}GhCoYL>`pX{`UUiVF4^y2`R+#8)(4*I*y|BAlOEUuS8E} zD#XyxkaX?@i-HE>x|nRvOJEE?|2(MfF}{(K2xtK5$idmK!kIG!n!#89XVrAi!atVf zs}2h~x=WK-f5LZ7Xglud+8BK%Gq26)9*LwRG-woy-=cv(oPePV?!vkSrQb8Fq5i-y zQL3|NzgmR@x};FJbeDSM#;|_3l_3fscupo4u{GHdBZ5f@UXa(LHtkl@41~~eZXVroVZ>>~ zkZ~5Rg#OTd0i!4|B1wfeI=uI4X|pF?t~7aonnPw2v?VGoY^iD$+b~TYM!GR-4LPhr z$sGNSrZR+3W1$x6HO1k9J1Vvvr|}5F0th%Gp#ie*=Y~@~RwF{5wm9TkVWz}u8iV?= z?#aFmvHcyHwwK85jt_a6&0y#>h3voG3q!5tCqUJWBHx992!YG2Q->RT6ceO*#-&5B$J4Gpg$r3$$6TLMoUK@!HBYEq z0W>pcVGvl4mSX@H(G8@xE{Hiu}!JV$u^k|*(mw3Y*STK-k%)7sdMO&>Vw<=U##df?AVF$Xa;-Eu70?sTx3HnIWn@j0Go(25Xq@D3f207KlA zVLbO8F@|^_(SOYzuZUXnDS&n8TMycE2ejNTQoi8Yp|~WVNwsl?I*r=Lt4BI57kEyfqN*3E6|&iiPyRhU4aIHi3VH7iYHZpJC{f2vBC;B zH#aKHaB!KANe%v%Kt?-egOe#od|UP-(g?pm^>agyyc~j#(QRn_r6+brp_w5ThsL6d zdd?ApLLoTB`*6eJ-CaBbzic}i7tHqrpAdYVe5tbLRy4Ct7t<{;p>Jzyeg}1Zy7Cf( zBPRYN;CeisXAswsu1_g(=+Yx_9NX&I0(KFV2EZP2i&g)e%a$<#c!cHAc2DPh-1F-; z>|2oBm^#%!Z%gsFE_wqLDucGzSgIbGKsrwjS?Sp<P+#= zS#&8zAU#FiH4L~uao_mh;Jv{#UPgQw^kIOUtu-}Uh~5d2IWI`tSMQzPd;awOX=_2b z4hwT&m8-7Azy*yWZHe1B447FCRvL+zg$VJx^pAc>Hq#=|4eTR8wfBbBbJmy`{n>QB* z{V_{kfE!j+VB8*S7#*wj`%)@xdp}o2yl@_!7DAzIU>NM#hmq4 zp*v$}Z9h>coboOXL0=xt55ul0Jwm`KGWDGZTt*N=1K)zyPfb{1b#?Uo;9+>8{>GZ> zU7INvl7k!oyjH@{ct63kOKQ-H8X^tPBxc*`4PbEwX`HZcxW)=c9ix!n6|10z;d%hir9#9Un)vu&CrRXNb3(>bO zq^AUIFmBf_EeXgAN4L!VnKEjq)8k*hX!wj`)X@|+3xEDMoc`Wy`A?3&4BH4%1G-=V zUfwwx8s&dF(_$egKdCbXI62cJ_vVRkU}2c&31&s1Uy)QDex`RT4Q{BexOFO&XANP0 z2$c$RvCxn@D6S|z!V9~)1kr#T1&jD&gLZxPx3<$0$))7*Tpff6=E4Lcp9=eM1 zv3j7hFN|<@R80V~ zMSS+vl|n=NZh=+=?FE=gSlAEvJ?LgWKHwskk^9tYjh5^L@A~hzX9Iviw*A3(nMx-J6?16TKVK~An104l)NQn5R_eDiQS&^87ltk?1#;G zcyy+~F@7|>iuj*>qR{$K90Wy?;Z+*=lf5=D?y6iH0cGb~S`P~vjcL)Mhjw)W*z#@_ z+MOq}m8R9;Kptf~zSg0GTpM9a`sUfL_=$5vfI4fh_sn>2|A_5EM{H4fe*pYJ;MzSr)=!v~rxww-2QqK_E9!v-#FrCov~l zJ*@G`nJ1@n9?k!Z)JGpmI}|FMB-AGQz8AJ5$%oX@m;QJcJm=qZxih_~g(dXU>?i`< zVPhS~>!);Wr(DOAI67xaT6x;9#@Kq$zrvEc~12^@YNN0J`( zt5R^CyFbS-X$aVvMk?y;AvdV+Guo0=txDV{QuwIy_I9L$V*UEUco7k1&{c7o1Iy(^ zTOfAiNZ$GLMxn=+# zi+N!N$W-PoPHtg$`1JjvD0}L!qFk}|*1b80@gx6oGxS@K=n_9664@9#k6c7HC+Z0@^>u?sUtP#v_FCO73s(n4kPYWuTO{|4 z_P1jf=ZS9-=Mmu#&IdOsNwa1Q=Y?_EVYVn6bhtEYu61~Ayl%)yDuT7vlx@z-72e9) zyL`_ZY_);(>;-NZNsD?_yp}$`Oc@|ZfBuz|K71Qh(i+gBYNdy7ARmJtg`foHgzruX z(>Ol#;+se`>$z8X>3MAWYUw-UHcKZ4I~~IcXx_r}!u)re^_!TQ^fjl->yLWuwQ}|6 za!cf&8NibD#{1T{o*te1?7@*t|0u9`uZ-1taA4!D75gnW9lU>_wyy5(w{`39rHoFz zv2Z`ix{vj8MAg8cSxEK{8`D`k0NB|q^)B}Fh~B}7*FDh}kWWl9L?<`FWLMQ4?OSm$ zejYjD@Fi-(6ioq^*zmDG6UBx`b4u$$Q%$6&xw&~__ay&Shf%O5yHIbjQteJ>1}+NZ zvM8jbr)Rt0%l`ejx3RqBXUxkQ6l@&Einu(8C@>}ee}w&MT#jq|#t)Y{V?ySPnG2ah zNfA+@On zx~c0r&*RvKZQJ&32hNs_4&tr`v-+4i4%aCW^y}h$@nU18=iq3cl=6UtpWv0!o0tAf zomXDYE)$#S!xd#aEQ7m)HgfOn(C2z=>VB-vxRW@c6vr+t9D&HQ+gsgXo{b}(A40bW zOQ@tVzwY)}pY{cW)F(Io0Ppq6wB2lPR~Iw}frdMZroP9U=BH=?GR@`Kd6W5GjXr7sp($p zfo){06HQwIu}7SwXb^VoDXn*RRg-#$H*J|+xYZ+T_p~v#UxvA;=bW5(t?a?4E%kCn zt=w1J({gw01amY!vJZ|(`E5W4U)9c?nj3ebETA}p*Lk$mdQ@K~n1rkYt~x*d*!Yii zjUu~)Cyb_sMUsB_vjvmhvj{v`lRG!kcXbN1pIv9W;g8E3Eq9IeoH&*-6WW;g)i+nz z>$-e;^-@;W(Mb*Z`0=Af9aCq$rK}~YvwbUp%8m0>##F-MFpIDcN7mo*qJ$N(7_i9% zB3@W=)%~2D(^N0lLFX8x%YXcv;Sa#??NmtJa_+n$w`PY2{Ib_=_w2TVX@dZ>04Zt^ zYS|c`N=rN3K|?0%C_+-t^!~FWMKym__e)WK4n}7~g~`l8N{<)xn*=kmPhG2xt?Qt# zr3mR(W{0Y{C!Sw&f7<-_Xdk<|_cTnH+5PyCl}~;2e+8$%3pII8vKB=-N8N41p+sZP ztGN+QebJy~J`B86xC-x*gsm5k==u4@?RTA#Sl~aPwWjfg2CGY*7aG?qt4M!eF5bh; zjmQ)y9ky(rC0g4hcI5fDmjB4&2UFs+3NC-d!t7sWd*U9rcW(ll9gp6MRU9%U>GZMP zj1>(l7ul+|aXtOd7~9AR%lfs=Eis#uRxwUvw$a`jV}0JXd)sK(vuBS)Ov+9CBoiPX zwfEgTVUgkh>xi1`8WGbuB-q8-S@b?KPsPjq_kH=NtiExsVzeMDad<89EH{au=VemhtJgs!ySKq#Ff-@H+r(|0!D*MV>@{#orVD2 z|3oobRE(|0L~_;D{;XQA*GbLTKRvwG(qT($PTJnsv!*FOM8-x+osFJY=`_V)o2vpB zW3IN&%)uL*4H5oWTxeZbFcZTa`rG&{Jh44yL}T2Qv|_pAzkD)xQ9vvir5YTMmUnlD^PWV_9nL$A<8W6%@ z9JbzV+a_VTU}tZC5BV>8Lt%hve9zYITtnXLCw}z{&$n41OMwPyi7F=@9C(2}w0)T1-v}kbMU_ovm2N2#a=3 zhRbEH$MUUyDnoZ!G?JlJsy69;iw6}x64OFDo5=mShg;A3a~{g3#CbHzduNhGwig=< zFgA8LpdVEFmvX6=LDOLCrcYs3d4-q}c_c1%a&lrmHBjqfbc*rBuwWronD7LD2nxL& z!e6-^+R}Cx&nZ4joo7J#G~Xr~=_DzD8JvL543kN;h<7Ofd%|s=apc8| z<0ipPrs39NGZ{W)Pgh}8#Px(|zu{ZPq_wXT_ohuSzpTA4;#y{S8<#__{cm3Dbv*NV zhDWz}jr+xKSIph3XBtpE*l~Mp`})l^$G2PxPaqHp&EB|v7sja#n)(!ElHk(OHaAjG z6}xTyNm}4W)?^~7Go>{ev2nc$3q$ldeVWS^0kIhG=0_gyJyTmuuTXqgj!Mm)Q1mm2 z^GwL7X0?TziTTd}d6C7d+cKbV)pf(~| zQF~U`VN)Men&MoxNCaU?#3{m;wkff1i>P z8;IMX5u2+ z{aq(F-^uB`Fwt{7#kII8ky;4aPyC+C9%>QQ5qoGAeEI6F4#p2zex~ydX9sanJJ@Cr z6SWp+7}`sUr-4BJvyqHn84$ad+DUfvayoXx7M{&~-~Xf;-}uY=J2SG38y%TeXeA@& ztXxHpG#vBC)2Yf*1eEIb>wj(hKRwPSZ`S|S{uO;Y6_}or<+wY$oA;S@1g>sHlkS&- z-<7X#+)g(JiJaPVgJZ*%iU@*%sJu@Lu2CKG{j5^Ge(kJi@<6}Z$t`k-Rh=o-H+x-C zoHK$AK#V#~I*d3EQ@nS52`}Yo$kHZEtYz1_JxqDtr5RcyfcHHfSH?Ylw~jG-*75hN6D9DUWaj->-7j6{9J<`< z*_XJtox_{C9$Rs=Q03sA?Gs$?yfAg^{6yp4^Al~k&ZT{T4kR1h6g* zy`uO^=_z$`W1${r^gb#XsS==E-*5`;vZsU6lTRg%hzOoVWSwH8jUo#fVjm@M7H$2R z%n@JnYI37Xjg(jCtAyn7}|X=VPOEuUPNHyKb;u84I_CjZrqPC);p`AC{zE5YK#cp zd--p-8_hoFl75ygaaR=Y^OePulX2{ta(>*=NY9Bi2F1wxVWPSov*lABK40WAS0fZv zPt5z&L+|uls=mCs^*v(!3q+Yf+`DyiZm!Hyvpt?v$UPxM#{ICM-%-0`&?zp$&_Des zod;lZ68wk2xzINm55EUelH8R^e8Q0|lLVQFS$)%V4Z&3eUx<9k$?Ue%ICgPboc^)4 zjolCKz7fG|ObQCD1j_h2mefxuHHVU1^ z*((R5Ii~%~c3S2THA<_?m6R9YMVt^H{b}fc=6`jX4njjhP*yFBNJRX&wJ^0wVBuUm zBoJE_U!RbC-K8W~pPg=xIObuMn(HhJ5ts{?hNQu;fnLVDbr#)h!d~5aR`BP&?iknV zU|n5nxdcs6QhHXyWVGS5p1&saDSTW+ zZg_HoQvpi7=$fGQ6gSXpn^!3d+-b#8$71v(AG2B0riK_-A<%0=CD$4BJNs_!Qi~to zhY9hC1U$ay*e4ua!p@f6yT@*;ET=iW&LPv4-4(JuW&W*TqdhlIPxQZMGjIw4Bd5vynpymrYsN1IR*pHjWjiVRrKtCQ~3Js2LKunqR>J8Tn|+IC7WlHipy8 zbjNtMHb2+I(z-*#+d~S5Z@I5wdScK&_2#r{yz*_sVF$X^BW5j}Gk z?ekrOmjeDHR1^blvD_-Mn9O0s6b6V^Y);Oc5xaM)YJ<27q9V1&=TkJxR7YKqg~%Bg z0%jY#V2^u?$H*B_PElLRu>B0R92dR$fPAdi7>@!NkC^TZi$kh}pn2sGhJ84hvem23 zc4f$HpetzyZ@P^!!fHqaqGX6rB@+kB6+A83O~SLdt*2=FVnm*uV+JV6HI~JEd=Xvp z*y)aCDkPW?P#ns$wEEfi9;F!iZ%Im(DiA9tT9aj&7RWuWe3GqyUA?3KC@xh?x_EY8 z)}lHfv>or+@i!fWNCVkdS%W4t$(k(~$&ipQ_QVVh3kJ3Eg}4SdnwoTl-u&~;RXJlA zWg)IFMg}Tu8$$-QH9Wk>AMIQl1K~Ubl)l_BIW+j(5Y$_EESMSb{Nt4Mzvc~Q{SD*o z9z=eevdC+uOlNvU9HzueCpIIMfVoY{I}=&DXsL%bo^uQ9r?QQ)wK$nE5`HV z&OZJ2AY@Jc4?o>K*LOCv-Z=6=m-_4Kyel&2Z|;v54i>#CdVz7i$8~S!6#oOQ z$EuCndaceh=b`%W$;8#@Tv|YKFu_E^DOcKSxcT{tmo)T3lb}$u`L*uj>DwQM<0~dJ zJzQ4rZ~(|}yLq)L z*sxP1D|C$~hWIKxhV>tTm`9LMZer-(mD^*tBwNcu4w{}jPd0fCbEpryM1^`VB4Rl7 zSB(3X$>mGY>d?q|DvvH>8}q9~NJpl1OP$#%ZZk{ny6}kh zREF_p#q!l^_zXzi2een$i%TZ(r1c0i_edAY0rMgT!{QY%jN$`6^b@kgLUkdcp0CkW zKMWwRjA-GMGspK{eq(~?#02U_S&9y^@2%eBoeXe`Ut{SvngR<@%?tB=ook5|xmJkSTmj(0!Ec*&I zMfN7o4qJ95e75CvzJ^hDeQD29OuD@Mja`)g5}ZN(m=XVT<23XFR<&hc6|r?=@#~%$ zDxaYOcr;%L{96*5y=0x=pQVK|KOqW))5#ZV4~ZHNw#QrJng=ggG^u4Lh0gqazBUbN zF0WFSYM&4+9ySaQ1V@$x-y^T|m>;DF|6La4MMnk2Ig!n(rKP2LWH@QwIa2rA?ol~x zdaUF{G!UP+N0vjUxZ$*#gm1B%mP!o4TG@>xP!#IPb6}BHNyjJD*d5ovPPd)lxlB+% z(I3iKH0oQ`>;Y5TM3xTrk4?@i^Zqf;Y~+PecB-rPcYV}md}x;`NxR;Cd^v_(Dzh^5 z9cC$0o0qgMKkLh6kKjsFrF#z@I+Q*!ICJXF)${1y#-RU1_4vu9*N$;V0IT9pxBtB4 zDa+$1gEwj>toP(M%cu+n3OZZV*Rf~T5#Lo4b}&6~(`*35aXNVnT#js*uUD^LOmcE@ z_zyzFWh9aCbSq*Hrf5cyE|Q4Z?yrYGAYt`j&%gIQKz~jpK(`!W`xoyS>*i)$-5aDS zVs20=U&FLS++yf?da%uto9$X0d!qFGVE~oa;IlFAI_B%|Xux77TKV$0m4f}I4b@9= zmhmOP;lYWv?0IC9FYTb#EQOWMvW>VR05Zrb)m%}&4fTL$}6smSCue>&~@^afM>%g5O6s`t86=DEIWwLWz| z+^5N!9*4SiNVxK}-oZQf!XA8ncIb}L0d2MIr@i`Q4M_JdDU`mSoHl1$e`7b6l0u2g zkRepTGz1Qmi18}%05W$f^odASGU|t0zBJc+G9T~+USEIy{CU{OyOi12@Zkl$IRM6C z9m4(u86f9Qm`&=}P0#Cd04|#pM@n`hKBEMANSJefJ(sp`EQnNC!Cm)nT?d6MH!9o@ z)H8;dpd7lwSEp4&=k@EC+Z%u&XSbWgCu_XBCR@Z3tsd(vyO?Fz!uTx3k(glsz^!O9 z;Lp1T-5aY@t$o^abDOy)wbzdF77U{VW;jUnalFN>7d z(}&TSfhH7?*-UcBGVnkbyX&8sEsVOhHJ*ISzOXT&7K!Wwzh^!n)EKfzQ#KcYqg0N0 z*7tr+5^m<80|}a5sb2T2!_GZz@uk(cWnb>*{G%T7(nKjPM8|4a^Sb>jZ0=q@J36b& z$ik7fiiM5?1SypX74rQ2gr57H_P_%hZ1xf<9`>dhjN02qK!K_<4#bxu0v$;;=``BS zWY?otI8~hlgC-}|@v_>uPPd)Ba)KdOyYWa$=qyYf81Gf_>OLihVbDqZCd#`-2EvxJn!xtW}JNl`{qUnp$ z-Z{LTFpM`X$3phYwAQh5e(v8bcTv&Oe1a=hiPpr@vKywydbI#>6Dd0(t=1jOzTSI{ z_O$G#e6nu>5aZh?L2Xeg-nAKs032^T?IuTRi0(=zhtfW47d;zybM)?nxVL=i63a==cT#MYl7Bx@R^oM}G*|r+ zuBJpEYdjcOkXm}oE9w2BkStm8EruQ>MmgkQv%Wp*Q~GFP%863KU0e+r;DE#bd^@7g z4(G#^ofO`q5wc44f_k*<26JPrh(rTdJUKug_t5B)b0TtYJ7i}M}Mu!6>FM%*1Ti5J8dOsLuZk4a;~HMH@Up$rn_$sXsG=NBc;Z`?spU;#?z(@<|1+C^7tA<-tm<1CoU&iIZke~B zjP&@i-M5DbRG?}!lPZsiCoMIySvPm}Nf&$6CzBVFvt*-A?B}rbsXU1cQgal_X9pp4TuBcq+(55gM_gsP_U%VmR)@#~*yX(4qprh33rt0JP&ZBzl zGr}1MF=jXPyqxU#oD`@%WGiezKkV)8-6wKiOL1hDI+O3c_t3uBh4wzW%|ATt*ZR@h zgTto$IDdhhBm6CyFY2TA{+kG0c&o@^4SB0{TKATR#V=c`bJB)xi!?|x(@V{!^A1(p zvs`UP$7A@lp`V_yg$GihEf``aqHN!FXONDG&BUsdQ5*lg!S6uN2>AWIrG5(uBXqf} z!u>bJ*mJ%t;04^?ygF;G!-(bF1!-KU-xl_373zA9jzRo1h;-{$s|J4e8seu`?ewZz zQp4rT+&@WaPs>nRGYP9c6V|nDSchwX9*3^5eqK<6_-WqHdr|u^9zTc*zV4dDJUSlu zqrtcfIQ4A7q(FXo+Pig}yKS$SFv z`TH7zpwP)<#YO*E@fz2arZxXzvvci&$m z(D$=QmuAj-*McT)sflZUjmTrM`%?C|7JZcr-bn&xQZL(g>v~J(lovo^2K#x_DR@nw z?&ViAnP?7@OG$E$FJzZ3Iw5)F$n;B*!N}CV%A@8c$9qlhB|sNRU{*zm&JDa-Q4`u= zl02t!%Rc7;0m>tIDX&pXbO#U?oJknmWpgMPe*gYqE5;JUxd}7`Zhmo>xpFfdBjzQ2 z3`pq^V1VtLM{4f*O~KPD-{V8f1WBmQi}iPMHR}zei=x=<_HC{rY7@?vRL*0QW(AJe z)23OodsI>w!HoaZ|4f+e3yn3Ny&sb=hpyd0Tv0aPHo8h63mISIQJhV3Q6e?&p*B;x z@_Bg5s#l$AS4RNL#yw}Vm6iuGT8%6AWxYOm$PZ)&P)9Fin^FuCLXd%wC%;$uRTx13 z!t{uCd{w@s*2@eOsnoeur(An+V558dF?Hjz$eCgyhMkZ3p#$0+l%2DLm0mNiP$+uT zp3^C=w||fN1mHdx#AA{}uxnxaN0}0}(N*kZK7V$-6z$ZYQ`~8@U)t|}Zn@7L5n=`K ztvhS+XkZo&KAlL9#K5k!vDli>9fob}`r7q1!3lT6?gG==3|zkTBw9ooaC?MAz=fw@ z*-tIh7q4!{Tn&j*l>UI-%4 z<1e62RP0BC6oCjM-55Y*Hvi;M-GLf)^{Vw(iC)va5XjJPMwW?ka0ckkXM!ZjuW z7Hvy?%`i>grQj&3v$MBPqt7OJipd2P%73MxQKZqQfUDKXY47K53ZN@Es*lc#quWRfd!Q40G{Z<0;vRakp=WX4_iCO2Regid6{2~^V3u0r|jYj)rr zEC*klS?-qPSZN&T4#pqZ3*9_=`Gp5?a40LA$o2)E(|uT8?9%Hal8{xs95~vU8k+F1 z=wOijmf{gRnH^s!hPP)dudOy@z8ld-EJ+v}nAF_3f!?&52E4)sUGFr!+2-pkjL!*} zhXL`owaUmOYJYjOyOtVm3S&sblasy#R_J3}L5fyNjUG|-9!S?;q$ML1#b6Z+iGZ{E zsE3SY0G62n0oXG9E^t}MhddRiXS+I^z$agljSJ?)UvMM$&ywF8hAJ7 zMuNAjhNcmmr*lb}Pqv#XYb~4Eh5n?U?BA-P+5(D#;1~heVA*?EM(PS<2J(R#OKa-+ z*aa~Q`uX(OSMhDt%^WHyi%I7at@0VNkfpW!ikd1-yrxCl65_mhjp30OH0WA3KD*q^P_% zTmkQ5Y;fj^%x4SwEuh)(Ubhf|aCZK-Q)gpe zt|=w4KL#7=`1`SC-|hC_0_ni}!!?3<8qS(^!gP4y6eZ_ znKAFcq?XExwX+3fpf@1Tt^&)t>Rh%6P0G)^epeNhmlEuAy`lXQOM9}r?^`u4ASsi| z`UA5=xsF6Zk-7_ePc!>)X^1Mysx6*8QZ?`-#Hy&B52mLq=qLUqqA#K~)>J|vY}vH% z+E=hy(cAKwH3tq`JJM z`tEy%h!gPz9!O=e%q{8oC7O~z;05@zv^QNE)~N}KNRAzV0h1`TA~$D1_p#lgR%zFO zjF)Q`gk&-9bN9uIs_@Ufi{3L1V8xUPLWyim-6shN=y$$uC9y@X+5ER5Rq4|Dze=Ol zps#C$V>H+_x53?Mfz4hD-b}e5l$yn|zIoxAvGaEfGc;hiC(rT`-7a$o_mYz*%+Z_F za+KonLXcxdH$5qcWx$3AH;V7_jqyd|^OYMNm24XHQG&pVOHWVNv_HCgcU_Xc-Mggy zUXWW9_w~V6DMgj%s&BS@Qw|7Id6HU-1R>URD3e22hmIer$w?hQ(+GrVR6MPOA)@W1 znXNtJ()8;k^*ZIF#6ult#x5$kRghF1^OoIPeFoiKz~C9Z-XTEV$smD$tG&41z9|dk z&+|Q58L4M9YLs&5x7x;&fx`%i(%BLQWQ2$k!5`w>+0}K$^g+{>zlkTA;6KqFj!a3l zd+$%1F>u#KZestgdkiKn*tgF+<8mr)t!_ISlQhl>Z0=dlko6k0X?Y|9licVKKwnv_ zQGAQS7FZE}6@HexVJWVz${8#sv*V8c*#ga3tDX2)@-lRgI=*(E& zK{2=}^&)LwI-|!#bSLwZW-NoIum}=l9C1h2w5%Q<93#}f%F^-F>QStd@c0u0V9|72KbsD#d6v6G2yVq?|Mn9PikW8ZlS-aPcyLE#eY&zu!6v0k z85hHX;Z(&21>H`(l@8mi>u^V$zr1bA%M;pEUNQ%1Me9Ks-|BMJ1O{U11qsA#O^3yc z76IHFFbLT(UlFxfhYU^ou{oRe#+VR9 zAf|%aC3;OvYPv6*4`#mvK^TF2?CCDVU1{URDNNey|6Re<|HBkwiN!KZJL~qkz#GX{ zI?9I!m|T3_+_Pmbfey*nHZUQuDKQ22-?dE6^{Lb4%AA{rY@@#IqHrYd%+;~Dn6O@b zkn4rp{TB6Wb#+#s?DryhgB0`S8G>$Vb*C2gJ z!o*?nPyFnFes@Clr6ms(Bin9|9Z5T!1_AgU99iwi5= zVB;c}0E0`j4KmJ(m#qMF3?IeW4D9BtJeF&@AE7bkKnDd=#mpCll!k2tJ)vy9qmcy= zb4y*;DR<%je|o+pt;+37T3B4wiv1S;3-3(NTal)qiOneNS6Z%UhTc`>Y>tgOh5y3~4L#^sY`?fvQHaJ6JNpLb!! z90$&p48;-9l|nx)DH`FKv~B3pEI~c@YtwMaDn{9*8=Ihx*YpO&>@=iAy8hywx$>#1 z?Qrqx2wQHGR;jNcJa z05_j#TzAlIci!xzo7hAG0CzL}uP90FRzANPa2;1;d#q}AS1S-GaO|4mBZ9L5glc`_ z{jnQULe!}6SZzGa^AP3rW3(~|t?!EYyUXVbGwL+hX8)`->-a}nH(7idgK0!$NTQxUT^RD zzUrMpMJDpq#Np~jt6=LWY?HX1zb>za1N_&| zkkaAqx7#-P%#r;_)LzT-~P-7Uk8zKC$L1!=CMr}AW7lE;gYHn5^ZNGm# z4=YROMO)cK)!)19&(5nK6|X@~=#)H^%#ShS35d%Ndwk1oaht&_g;@HUx4p^Ltd&rs z3|z6G*FAAZvULHR;ZnHk>$m#SdL4zX1Wgd{T%4ZCEen_EsTdk@Ml1iXQP0gYuGGKRU?2t>tIlEfEJ&~H7^Mir3Ltz;Q%hShX8?_Qs~ z-LtCvH~JkB2IiGLjRLil&2Fqaw8>GsZBYJlxG`x2CEtK3$SdKqk zIBt&&cu)?5d|}+-IniI=Y&~of!Vu4Z5e5_P%pXK`L#49f$CT!^&!vq6Ri}NxcD0+y z5A)T_H*egqA%K&?;Y+u~LaVlhLAzs$_th_V$2i^p#44Rn1}wRQZPbm@e-R;z;1%pu zr0~$@`|tZ6Q?5Jrx^)e4F@VoFf@nza0};P(96q(BHiFna$=3ENx3HT&Khg6P{6^QJ zc8Wg#Y*BBu9_=8PrRpdR)T*QRh7R{bHjDNPd)CRrH*HXuRPp3Lvw}uQwBnx`#x2Xw z&>M>2_K&c%>emoj6T~Y+Wk!SWDx-FmHpm6t0c4v$Rt{LZ{-5!)hp)q56h(%HP`abFn zge?S!7bbr;2e(&6AGgkbKQKDj_7_Q4_h|EAKG>-Ifb{dYv|allrOq8%H#{)~Od0wb{X z$tzfu6BaXi(Dnv1rdD8dTY$qk43F?;#m|(l z0v|Ub*O6d#9so>SXU>~?Oso;U!}%cX3XPmBaz{Y(8NV~d%N4 zMteM7y<7W22~g@K7Yv+xEk_xWXDdB0c%H1Q5X@h{)*Ko%qqh>3hK%xRFY)I!iA;^A=it;&E07T* z6Yo6xdllm9F^4sh$;w1@i_#t=Niyj8Y6DFEPBq5RDVCDqE6!742fUVBn=u8+P&?K$ z(ym+lSgNHSm_Qm6B?zuHcDM@zgMg%x6YTLV$3g9Z=kS}Rr7LLXWU@&-3#Fwy{{FLG z5tdWVN8zw)Q<45@g|8gn?h#WP4YrZi;&5_sH*>E?#OH$K9j`c>|3`(dOMDE+>Y~_u zj`P|XcL4M{aXH$Q3nDL;f(lsf+lVgoFlp(Ahmiy>AZyFWBpmMA_fNDS+SCd+N;?V# zWL9d21X-Andf(=IY22jF-$Abj1=VNO=`@?s{^|Em7EbE$X_l_UL(e&O2(=5bIWX}wc`)jxA^-a zTm46U#PtT`4Be?5cE$xIwJZf6viezGQs3yVaLlAPxVvGVDVh=6c>yF5N0s<#&3DQi zoSIqyNcR_Wiq5`|FJUS;-?($Y_B5>w8egw3naq_i%3kUU%Iz&+|>j!Hz!rz%Ob zw^JK=rw$Z%7id-qVOZn|^0{*5;D^A@xnS}SrgBTP0kaZR~fLWVF3V}$}Ju9_$=ypAM%9VjTgYa z6gO(>>J>(n-OEo&_uU{KaoJ!_q7VtT)B1Jm_@{m5{pKqD*UH58K7YJ<-B6u*R1Nu1 zJcT?wL2Ss6O?ThhiYjQ^kdP%wW9S06bl`Jj5M4hEkt(rhOW|>Sr;q2r@dVQw zzhhq1hLactPy7*b>R2;rXq#)#{d%nSmw*Q52-F^Lv&W%fG|dZI5g!Wr?DFY9-fr;R zshU6ULgss>kXuUoM#NhLf{1!$=)?%Rl{DZ05UG3OBt?Pr+J4F&E1yl?*|!~f6ToFR zc*C}X4}f!ohq9!1Ahto=*)^E5V(Ce{wybOEC}6YB4hL+|2H`49ReX*iKF){-l;*P7 zQaBUJ_~|}tG^UC|fO(FV)QC&I$=ZPXLM5yoJpjXT^`JVv(rrUMVe1xA}1hA zR0?$={rn+=CjGBq4&Hm}8Y4wE&A#$AWFQbg_mqbIdSEtXF$fhoW~r4j@LnFpHx+Ne z%DPW6c_{7re)OV1Lv8x_#Ym4WyfZS1BItY$r1xFAd_I-XJ+hSZ*V40&DO+Y({v3UOU+es70~{jFiuzNiJ>Xg}KL(kqwPGnCKHd6o zRG3`bHr&HnBj?Nl3~(DXwO*N}R8LH%VDKc+5yF{wh<#9BX&3V?;*iMp1ng|oX5a!6 z37iepQ|F^(*8gADzHp%zQi)H@8_4cJ`jfE~aABW9xCoYw9U8Wx3kRN;^@fx;i^*@a zhX64ZLst%LgRqnkx9VrW?+;LPyaHh*XnjnMM+GjfA{l*>ZSQ0^oOr@6HYwV$_4(?| zsB_bUEj@W;GMoULCVWfJdACDnJe>PH7=vQUcv!`x*TeLJ4d~D0sB8bKd5ir#yr%sj z`u!E}&(JY3Br3o@A0d0z{EcsC{83KS>oiSgd^PU}aH5^MBqWRHDf|w0CEIC;*c4fV z3%VDnq-jsv@+x|~wLO)u)Hx5f;05ecN`=)ud_Hy;a`2In;uZydx;0|uFDdpwA{h5=HDctEJnsu+)#o~@t-(>Qa}oFhyYQR zAfsGxZI)$#d6Av0p<|Wped@4z-!Ii!cv1N{UzK^_t(?TX!b1PBu&kAolHwQXrRPF6 zJ8(Qrts`Cn(14Rknzd4!3W^IV7{ZK7@%p#r>qW`I1`lG0AM@ne_PtE}err_UVb%QE z>-L>*^7kblJZBv?Zd^o0g0b9L!@r&RfV83m0&IINld>jdqIE&X(s@rjD@=r4PY$4y_y)A7Pj@Ex76(pH# zZjJua=~vm2LKA@0=e(m=@MA8JRM|t<oLZCrnCzYhy=x((RS2emH7k8BsgqkJn(PDE`_il}( zpyoW{p_}N`h#z%dVOK_1LAb96A&_P{_0mcKdM}@*7nfZTd{*W#HQ_rLRC(=JUIMNr z+7dckWt+vb3X7Uy)b;hp53wvq-q65jNy@aEeI16ihb=xw@@0*Xk!$JHUz_<{AkN)* zv}Gwn;+7KmWSW!wzPMkjNzJhoG$auSyn~>|b!f|K4K*T}&I6YJEta+_RH@!-U zaSCp&#^IK>iUngC?g2KEshqG+ugy!$=QX#>T|KakhzO~hvpetqe(H8Ta=)xH@NRwJ zo%pfh5?o`A*1cltN}>@VZii!viJ!*&cHFS*YW z3i!DZp%|u#v5AOMzzjr21+(RMCfwPU-Uf<9z6_0I2gZiW>MW-3JB36A<4VB>iR11I zAex*TDN=JqoVWS+hP<*?^ESR^xC&V!e_dRy1vsV%kVAt;gED9Ly|WvuOvYuE>=4k3 zXOaF!IsVO!&aQ$d?w#%%%C4uG_-ty^P{7V@BHL2`Nx}QyFnU+PX>Ftu!`bQ$)e7L zLNjkA0&hMG2|AF89_w7Z7#rf2F&;FCJT5#k_k~5|DLkC~RWl*OWYC6$SB4HfW&WAX;^lDJCA;ouY zdNKAbuc4ghIXtO2r;z}!M=W#0?biA5(OE@zMP`U>zhk#6@4TX?3}i3B_?K=v z1V3;j7zLpNS~V2<7sf_31;vI<2#y~2Ii z^uiSkgC)`*+_FX`I|_TWDmpRNI;1Qwwx$ft7FnU>nU=G$ScFILd*#iOd}_Y%6^qS68IjG)t^uN}+-kiCA9DR+-&-+3RmR zFgpK#D)o|jgLJR6-AJHnl*QCYDqAv^XKecfs9Q}I|BKVLWm1nA+~{ z`wa=(EVik0Do-&6{DEx2GAE7sjO72^SEDWxKVEs_koI1ovovEDxE!_LgAK~GM6pJ$r6lutSSSz&WWIDGPv$FUp&3A$urbTZC zS4tek-3uGpVktLg36(sG%dKeG;U*6a8Qk^^L9-#PXM9pp1Lt<<{SUcx&~uKVT2{`z z_2U=EWbeU)_iYB={u0Ef4r7(=#zmFcFA^A#=^rJJ$1eeDCDNL#e>=Iohwam=u}glW zFL{a{R@nG0JG>)ckN8!%ap;SsCR5BdO#-8qMce-+vLZ~ZUshi;=Y)W*ZoJu-Jo$WZ zII;F8{U7aat^o~Mg9zD)K8QkhZDGm;`oY=w`p9ld9tYX%_CHT{CA^M?VNNfsFq}71 zsJeWQ5NO@-FN21Pe+Q#^9wP#>y3;gGB%5L3W4mEYk3n;K`W)L9vFX=s$3_{(JqmaB z;tek8$G9GS3pS7sMHslm~MSWa8W9=C-wrCyT2mK#=6*e-LzwNsYSto5wQdK)WQ+3f0<3k$04qIu}I$ z5Sh;rVmKwNl3r`4i}cn==% zjOpH_tCv@&GMf!&Viz@4)4B@$r^cSCzVr0QVBognEcIKLyp^b8_ku)(?J%=;phiVZ`Kxz%Cm{ zwMg^-we()fG#3U;Q>)PDlX7QXJC zn`58yHcIe1g*y*_z8I*PIBL`=UiXqSwR=bt*C%lHNY5m1ri8Rcy`@hTD`FXXB-6{D z5+GKaAFgTl$MKrl25kLKpXm#XB5FFRF_?2k*%{`@atJCu*?eBVc@HThK~Ce6l1$ah z?)OIaMlZCOF)*>)qLaU$b47-EklS!F;f}O6n-gNI*p!yqYRKw#h+%%LRJ<=2C zu&W;&UAH9|5fO|hgG1QWsOMd=?m|M9?+v72-4GR!gB#qj3_iFj!|Z`c;-SEbf_FtN z_rCLh(#g@r`OZ#!X{?oO^4Ii(yHK@lv^%Hcu3yYovnNsnHe;oQfnU_Y2=9B&o zVphDW5T)Ae?_6|uf+$WXJW78)I$mUyM$@SrTDzme(QX3=4xGKMzbBt4`SsIkw+D0| z*yo?QQM~G`WhF;SXc?vA%E04u2Aob7l$em?rZs19(2nrzc#+UuLs!S(N;Rl5oyh7=J4PD4tb=3xsS$AaySiWQt z2pF>uYv3GT~(5MY-#L3U;Y{T({AZlV1h1kZf7wLLw&RM(nU zdz%wj$C3AQyTk|G3-2;Xyq1divnw;xNadbma&yfHO?OL|IhBDtr?urT4rowNdsZ4 zicU;TJ|$F&UGVqy9I&@c%}`y|{vB~3JiD&LO124vUORp7r7T~e|G;qJFiaaq!|LHb zdm*kU9$F~L;w_5MmKr_W`4SM72BzUlbF@iEM|B)2-5!E{nLX(r8 zs=uomtTc*T{U5Qk!}$Ugc|QOo5i9`n?+*^P^{vsi1kWu1dsx&{Hx|@BGHAsN-rdgN z;F1QFWxNvdv-H2qXPYZNoWOLE2EBh;*09{>g27N=y&~^tmrrf^4P{;useRt24n_IC z_Awt|PXwT3EAq`7$twrb#D-Q0;YZX6T><+#?0aQND$QV6r5KekOM%U2(#&-gU>)nL zdpAlqnRz9{G8NOz-xF~T?S?2+nX=GodrdVJ42wELS_7iXRj!0{j-R==7mtM!sU`2> zSb$Ct#e|ojWZ&>M2wh<0{Bt&&9|!YicKQ4JC*2y~$S9I!I-u4|2qaR`>AkwJ)^gF* z=9&|&lbFAShwR8K-A`}tf};KunC8G3^D1ERICxaz@SG2zx=7n1v6opMdcOEM*XOo3 z05Cf+X7QPr$wVWW;99U?L1YuNi@&A}IC#-8Xn8a`B)Z+#%_MCMw-HG~19h@#n*Syl0@XXJq6Lt4g;=Fr~?Dt|)f|6w({UMbsJ z<1fW*otozoP`@cuGaPb%tvDcWGB~mb0NLnCRZ?;Sr=9G#&?vIOC^$`F;`Tl)`gS`o z^D8(me~Kj1#$&}5ZC?WNVWbXmbN0<^6uK^V6C|=U&vbU8Rjy!Q_u)>fSLJL` zu6d&0!WvW!G)aiZseU*af2*HTwcsr@0xqyb933fn=-X1f4LaCH{J;GNx^F#scjy1h zHb85CdAiJuKwQAC-@A8j)Nc(Uc>{Rq=9E{G=#SM^^glODdwPfbsqP_etFr^mB1+lRP*c^iw2&ww!O^)1F{{Rdixehd=_d4AsRWL|gP zFoUnC*+Q@RWWVsQIVK}-V%YL++0k?~{ox&!THd2_qMnoCov>>+hYe0QSdox%L0=>UJw! zscupZ#Seayr@JKSGg>r$3YduL+L%|e>rIGn3|((b^3(7KJ0YJI1ofKU*jzo4xPK2e z;?D)Y<*yy{?d#Vy5DB%4*^nM?wE`8Gn41D_{2A^d%zP%!4hiG zmFZV+IhS#+Fi+A^ zsBc!(K1&Y!#FOprJU?Z>Ig>yyx%U|rm!huJvEs;klPnTW8-ALAu2<43m+kTOv&^Vw zt&CrEjh$5;Wk>30&(o)Z?hkF2gnkVnZzppvz3dyHZ=b-jd8r#$HBDW8(|y3wi4;}= zeEaL(!Op|=nd&EIh@1tRgK?0&04Ji$1HhQVbRKcH0`s;wQ(`pMi)siz1PQ0>Ovt%H zzU1l)8ViEc)6lU*1t}|4Bsihhgyl}D9UsjA|eiZP~;H3s9;jqDNY6Ju1qN&$?fi6@5+p_R_d;%jCje!kA#1 zcw~sJrAJg?F}1Nw3{X4Rt9)FioqlDtJqExl2RGBuXsoyC!PMluz&ddRY+2(aAiv7( z|ElAu{|Tt-l|2sFKaeMB4GyXOTA?^AW)?z2G18u&s=7YG&Y#98MzP&+_^Ls!b8S&xIscC3@obo2;B*+`$ zW%?h6gLl$`;@|mB z4Xq(jcgCebxtBeQM*JCRwTw7a6eeR@vKE^%Z-S!aKQK18&i_T=cTzRULQQ(A(EEKI zwvN$C`xagG>o6s2yi2gX`U7@Gb9zON$BdBO!;q6^-GKQxE^pYeyrFl<;u_*BCfZ_# zRkS+FP)V3Y;GCP<2y{2}^vi*@kiL<92B@d7t)+3$B-WaF>7C@|YQ8Uv%V8_ZC9&0{ zgXr$kL1uzvxQY5a@#%n0xqT=MDP#JBuFD6RHEUL6v6%P+F^K(2H+Pe5Ww~o{B`tk9 zN}>(54K38P$l~kP$?}ItSMdk{u5aDz^jHYDZGD@x(=C0Jde37#W$>g+=$E(p#7nl9^AE!^bJPIpM!RF=W@{ zw(sq+m!@Qk z%jM7ezfx)DBKR2*aWd6#yU&e){d1Mzi^P5m9HbUQeMCC|G8voz@!m$rqtw?}Ie5+) zqntU7q;pNV0XW@+9 zDpwe621P?nm)V#7mUUc|*cRcjEi(WosJbej{LE?Ixu-(u4xMj5ZZy|V-F7%iB*3|m z0HE;g$6_Ac-1FU2T%81GMr}f@%MWFryn>ZOAYzXZ-#)Sq`RR3k_Emo<;%6C!rmXfpz3JTl<7J~`wp;D3k z&RcYwWXnb)kinxgXrsNy5)S~Zm16$bcIK}t)U)2O{--W^OQhk5O6mLw_>`O+t)*B` zHt#Vh&)|nB4sfn)iGaNkBWR?$1r%AkyL-ZIVN2yTh07YU?`0RU>60z zqiIFK-XSmkjt;tOwQ9CNa1Z3IxyBZqa$A6*W1YufP9id>EYw_oILM`etFa?z!jNU< z-@bw6$-ocJa=B~I=O_F0s86;0|G5oiqGub7PT;q{6z6!L4Zq51VwQX#gQB$-?{hkJ zf04`cd8+va%oZKwV%NN7b*q+aFCqAK)6;8*Y!I4(%Sw}#Coj@91b`3p7f$#fQ=&Er zQ=1zn9S;Yn7X856%|q}Fj6H!?sI6&2<@0QpM*+VtdnFLXm~Hb;+N2C81=&=&S)+qY zvb@AGXd#yaap0`BIkTu0BL4M`_n~@aktWlO)cLVwARln*R?*f!BRh_Qt

%}p#2=hu(8Q73Z=V1~>)>G! zt%XZEA|cTh4|n>gWUii}xPO-lsYaw?%7>yT(|>xzqNNZcG3*ZAVqfX60|Ws{sCpEW zNEIz|6h=wRY5!Q%-i;guOnsfpEkP18D*x9_6hvlH-6L8oyKX*ePv-v#HlW9ZHIS4A z%Bwv;>T=j$&4%bU$hj{|O3VdZ$7M+=YRC?ZFL&(dN8DwOK%7Dfs!8XXHUG1`c;QV; zw2=E9!#qpa@&*^_5ye)iICcau6BRibsLnV=VR)OLvyg0vZYkWJU0Q=uYC0jxI>&^b z#mL1*hh9F5ADX9gwn>+ZC4E^oB1=yERKn(CRv^YyKt^#n-3twtJ~*K6QRqAnT_-{GhA zw0>C>;vfcAq0!h}@_0Z^4_p@vson8+^Z`aE{-}I(y{|B{7Sm~3Xj+%}w&y?Mjsc4O z4v`v4fyd*c`9Va1CgjqCm4lEl0ISJuXQL!zAj2oNMTo67s(s&X6^4VDcMzn87eIQw z6p1ZT*v1-y_cqYpkXAMm@#x0CH z0oz=jbMs@kW|F6V$#>nE34brip31Fy+p2dlNyy zwqB#UlbQGi&V5hEE;v#{V;-wZU2J)uN#xR@t}2l3F972VHzPwaxR$xws&s}a=KiuU6TI5iQ4u+>DE^E|$vfT#)O9pGfM8){*K-(eA zQEdH_29XnC4HHxFs7tBi5{GHQs6!H-;@<+yXt7Jp*)4F2*8Sf)fv*d?9vRJV>6%+i zcJ`rh-J3lVO&6mR4J8SF&L+SK^7a2v_U2(duig81hRwz{Z1X&3$V^2jnJGn7GL(#^ z5<)1Hac2xsRFt7PiV{l69GYm*rU+3gNh*?t=e1(r&++>`e?9lHkMH)q@4f5udB3k~ zt#h5{xz=)F$;|6N@J7?CCLc#S*&u3U(VA5F7T1}}@Qtn&6lcanbps3b!vy}-?#v;q zs7xH-f2O9C3AFzkn-MUl595yl7liv|jo!OX_LvYXr%}Rd9qav|@U!!lIJL&j>x*U! z*e@a5U0xp7f-MzoPCpKVH(5dCAzZJTxNl&@Br)_6Hio5{+ZumoYQB^sLzFLWi2RHs zv<_#5(Rx#(xKr3$VTevP?p7}SZ-;@T^pQR3FxJ`LVZGyj1zYI<3btdb>EF1EHG9=B z`-HvOINQGyBNL2iWsq^7JWSN}1|h!5X<_FO_Mg2DHamKC{$L8@Mhr;*wCU%fc1Ht; z&bx$DmndwJd1mI74&WsUw)&4dtPX0lOGs~sU2V6)fh$iPTmL~=f|4~x;ZOU`1B`;s zK7iteX3k~_0yDj7XrGnhuy>0Qk%)3y3k)j{n^JKFh$w;g#VNdG0>?+}`2dF|Wt|Y7 zObnDc7aEGk>S%GX6vL+go>xT|$({td(hsrG0#ST(6F?Z=UUU1UhS^TQ&pMI=-FzVaIcS;HxZvmeF(x z$9V3jG^~8?MFsNx?#`ph7rf#L4=e(T+BA$Y0ZeUwyrMT8z^z}+Mrg>*T-+!BQX-~= zL6zdKl4&sgW}W^x*YDGU)A0#`MYP_Tj69)cUUG`oEMg^NBP}QH4js9B#d`hZA_Aoh zszLC#zB$W3wf$ZA&*qA2l{VRy455_=ATCUYgY0WKf8NJ9a$g&@CZb!T)1P_2ohPUI zx`^gcov-8fZpjF6kCo9<3=p81k{)gk=+nrn44@(hi0Kfk@O?l^ghI<>!|9R-z`;*L zuh`KHwK@9EssE3Qp~bHevp%G_Z7?%D-oG&lQmQIYHaBejeZmXO_l>Z7*fyD95a-WP zyZM+yq?zzqa1qFx zAEf*3FPG3=_9kKqA)@Svt2S!;2cMga@ymONJFoHa!W;Bx=i%tTol}?z#u76&*_Kss z^TN8$j6Oqk-@)8dG=iei1;s`U5D!DZVDJ?es12px~CX@I% zG?d4!Pj^CA{QqN2IC^NQxFBh7ZN);5hPX0d4maRk;HWn_`bOai!Cl6|+_wEIu#wVY z6s9CAE#Gf}7-()EFS%%ck=H1leI3RprJn9lb`uw=c$l#MyaBx}++c- zH~#AHdSlPg9luAlOc_ZH_7^-x{hn+d0M0A3C=DslrR(jpL-qDD&77UR^ysxvf7~|k zR)|@gPLe$9h27UuSYiU`wx#vpz@b&PpR+pczVl^nk~PMgg%kXU_OD(s!uuhb&c8Tk z!g|b()&K;TJaO)e3AJJ^B?OJU)D2=cpjLiQgGWC|prJ+NdnRCjHpZS9WF&zRJBQ*< zclBG>@g2H~*+$dRL4YLRFz4zZT%gF_><8W>oe2d0nWJ%BgV7NJIlBC9z?KJDkOB@!?9l9 z$E^#Q@;16h-oQ2k83F_aCC$&AaR#~ zzuCb92bLj6p~Ea2<%!tJ6m1!dOMputlH#tp?Ravyqeqk;y-X^ApnB?yRpy4Tfw_|j zWZ>VO@8f&^7#PsZ4DZ=;EJ$>Vu=;1yVoQZwq4#^*G&@*3^_3%y(J)47?K@qX7yNCU zB@2K6nyD`jeos!_xww^Mr_%fTimQK_0d74dWP1nHn*Byn$xuERxu>1pgiIn^-~sv1 z7`OwaSMz$i7n4X|tud#rmli{r%Lr|M|FJ?5_=r37;{UV-ge_&?+Jb`W3C%xl2+{uH zXVgWyE6Ww~O6j(d1`$a+3wpFUPdMWlHj{-9feLSs4j+HhJN?oh&1Eih+ndKX=0)Ji zvi|946VmTo&BctuXNBc#h?g6JmH06<9iz3{1HXj4#HiC!nCuR;Nl*lnLhqCJSs?vs z>vxlh|LMnW*Uck-tuFer&e}!ab^Rj^ny2p_OtU8`P=>?_=m5E4t+p(-Nt$%N=Hk$Q z0)Ju+`&;NV#JQJ~+@ije0YGT&*fVrOoUn{ zlcaW?JZa?et^fU>I$%;v`y2_eunH!X9P6c#>HFS-fQj-2UGkeDke07?8#Zh;*YpR9 z(|pf^;;FRd^mkZ$lQEUDznkzi1xrNsi$V_T7sV~l##GF${y{)vRwWC47;4s$7rY#> zLSxLWq@kMnL>Mt|ij$HRAk79g$K=dUsAAp0FZJz97$i6qk;BDJz z4}*fA3|@FF&uM(nL~rU2u0x&OHk(yM!oJfA(I1)@B~8R;_#T|0R$Q8ziP&$|p12>9jJ=0B;S?!X1Mj$x^N&A8WpxznaHL;L%lR!iKX1?Mt>d z2%Lg@uN%g_iH7$jI4A{k^1@OX=OC)byjPmuQ-@sN3a$}wOQrh19Rs0@HG~dFgkXJG z@%jdevQC$+}A`#C{5 zrP4D;^Yo>|OVgGMv4hjFGppC%R8u=1;QRc|Oobxx^j?`SK_1TW!J6x_>r@++SD~}a z5U<*yL$nuDhF~8gqOm!}>xToc2A>UO)zIwV^jqT-nGTpy7%@So+J^W3kV;`uWoojuXX&J)9@dI*&H`wpT7!U)ZF^s7tOZ(P zQ(r;@BIMvd^jBDy%;*+HVMa$Z62x+jgKR#FI9+&9uR$kss#Dxhk3|jLt5#l1i7Jh( zkL`_ZC(52#Iz7UuC{9`h4`?bgK15ff2yF7YnAal4FI=Z}OWZKN(f6fjmHi+XZyX^y z2aq;2Q@39maXw-xZ}K{Wqb0jF8MnImg;(F1?a7BzDeGJ=#828+JA25)n@~AoA|Y(= z+rNMRe=8*L`4hz-E^>c0e@+!>085<9#*!XC z+vi@ch5O}=e)awyz(r?5{9}wiY{-(+5wVb<|59y1XvT%qd}D`sjvZt!0nS9kQX1Wl ztr#=xSDM|^%I5K>3%?FQ?-%cr&Uo^ArH!2|?e97b{%WDp8fs@Y5mCjQrx+0jB2Kqb z0y)dZy~TmSC}LTaWZeF_0i_tHV&t}pFV{zQY}0+1{X=@m7|+S?{X<>pu>a7`@Vni% z%)L8wjGB3zx?@dr{HCdMq^ThQG2r~J5j|)ig|Oprqu=b4Sur>(o>5>rgVa}V?-1%X z%U_O|c;JfW1-nxCMK(04Jqf6TB1^|APE#9)Y$mgphwb#R4hG%JNKVeFu**~gMJY*C zv9vu?XilP=TeO24Oq|^nAAM^UfS-xGm^@^77eDjJ=#Jqa6buQIZ#F_^IVSU@UoS{< zA3)uS$dtnb580$A1Urog;ncNIB9pO}%ciHwfV%9$cQ0G(e zf4!GtNNCweH#Q&qdiFstmpo?o!a@DHCVpyL7+Bi-rV*@ z3q6W4Jm@C_s=eSIaX2X)>F{Uz;IYhlG3QbZwa!^|pXncAtt$Q*$jCm>?f&pn-MV+5 zMM71AdXvFP6wAF;RN6w!lW1i$EF#x7;_L~QZ~{%DqO(_w6#t|UK?7$`-}l{x!IPw0 z8Gp>Yzp!Nwnlh}WZQg5XI+j7lG_wV-)Y~+)-p_vUhQaAih&>emfdQ;Fm5G@^{GoMj z1vpnJ(P6%R+kB%L_Pw`N?<;A^k(6wb5oKVN1Uz0?M9?2!D(-Yt9oOfqYyLdtl|h-c z7~xJ<_d8Mf$_79x&zjOmxK5ef!m^cb>3o{WsK{Et!gkDys1>6QyPt#pSa$cg zO$s*4tJ`D$EU#v(9G3Z+nIScc#YX{U?VV{3r6VJU-jv*%TTuNH0p@xpePcd27!6nl z%q0{LIxA_J^*VWm=cr5Z6*Iff{hb2#oxE2OH4ev}P`g_;2OQL4N;O5kK4uCq=L6d9 zupCBQ`%*WU&_BTUYPy0>oG9{83{1&rL9d|`1gmX393UT;bwj5?UaL`MsTKCgWk#enFLcy6+jg4lUgRbUlAkc}4Jgt@_v9uJyU;+@xHGH%nwIJV`TzhUT`2uRUB7~zE@9R7}#Aq^^M3vGJ%AWivGr-U$$iV zHu}`*uSSLFh#JAHfJ#{J!E7McGVtbeU?0~|qmOq}*Qs-4C=s=H zhW6pu0PyBfJXCPE#dB-M0ryA|eGAke@^V!89O9FmC5i3Rw|Xx66f2L}upAGBa{_O8)k;d5ywVm$C7aFai$;F+K<4<~DZWZc|f0j@pa8bm08E7T(9G>IkCUjswknLaH z#-8}!Hi+y&KUU9+ZOQlFjo5=?|4+IlGlc&S`$WW&KHMbJ0B<|2-*v`jrh9_yt#4Mo z%Ay~*W!&~JcN)vGIDuEZ2k4-Se>7;%f+&7v>o#rJEC4In|HiO_Q@O+-Db}`}p2*OV zrOaSvYzsBGAKBoPj$eoIo4Fk*S;>HaOJ_pIdM#u|IJ`n1*yJWH&a*IHN$M~TSjw)x zfF|5`*dQ>0R&xeDL2hNNk=yPO)i2mg-q+towZT- zv_xmWR#2T5sUGvc<1u^MlHcl`(Rn@*opz;QXlA||4@sQ?@lU75)3ET{IXVYahD25U#!)>}(eLjS_TOm^s8o%)nFNr%4Sjur9Ffs4bbV#2VXT$~r| zcFZ8=Gr7Ju6wUsZb$fH6)88ztRL7_&IRLo!g{01A%y}nw=x;eaCa!L3wU`qAYXC(% z(!EyN{;_m0LJZBJ;a+B|gFISSEk>u6(J*9{?3a!=Cq*BG(GvW`aBXs+3_cJL2^2Hd zeli}EB8`l=Y5xz0h0=@NJ=#ruizim94~i*>#NQKg;ZudBWl+MlYcv{+vpl+#CnO~d zDI{b3P1~_&zHMLF{GWzmwkK3FQ*q?)u?I)#ms?SI+#$|h-X z3X$~qwq#mj0ezS0KepZ`ufyqCnvPGC7mn1`vOAQ4f}_qFdgz7jp5?#c$Iahaqt(0g zXG%$fmi-n7OPb1UNAsbY5sCnSAd3t^tFizDoJ2!TUg2l6Fue{BAgpA73=&;^uffJa zFN|c&7xt^tlXGwV-ctf zDaPxL?B0V2)AoDKPQ>G#Ta5u#NPH(fV{BmO1yDr9>|3{P4QqQUaF%ZOahnc7zsUqz z9^dIr(B()WDY$bxtE|G-;I{hIq-bBdjAw}&wc4A(;+)oP<=yNLR%5uN!r6lbgTmSE z?>W`05#okGg!nTtuEaUmtHrwp#v?2GyH(yoVa05!M>6tTd5a;HKDwT6{92ibJ}Iy0 zBL5zzt|h~7G*_FceDLK7gXo7${BO~>DSwO&T7DB^ zlW`Dm)wHFqio;T)DqxEyfc*ub|6TUc$5;GbQ$8tr6uIl!$erplmvy-L+MwXwdjh1) z9{%EU=*HsEj4--)QvR56>(i+M8`+o6|H_c@)b=%LJsZ~# zFPZaYvSlkj<%ktA3xo7b&N8t|Je=?cRA#JjIWAL!+L%Uu{EvKX0Ltf6|enDd%|(vaV?05Zf{1cMDq?XjvOidVZX+vOh^Y)Goqg!6}$6s zmJ~zSksWG&m=CcUpsr+8!-$#Vc&AQ&)${uAeHzW4w&`7M#OCa!a}Dw{>} zuLgh4IPq8g)(!Xf-q-Zv`6Y|nG;;Yfs%D*uiSj?`s>>Do?CZI2-;AU=^N$Yhxa7p? ziH%ORFf8s7?SHD@8W*nZnF86?}-y9H2dxui@PbdJov?y!w;4XoPYT&r~1?NsBxWe zNI#`at}aQ)_5Wesq3K?B?0vBFnyasG$Yy8NRZk}%A{r3*uCj^!a1YJbp*uEhaZO;T zsusl1Q!;%Sn0{cZKlT0?(oEK%T4vq2LO|bq--_ZV0}J2U*j-Pbk^1LDX3_|A@11iG zwRIC(j;}atoKfE3kwVfCznIm)%ya+-??#Jv!8oCqtMIM0+n=p6;Gm|#N*MC&U(qlf zRsAi~72RohO=OXsm&{=hsD>W?l6x>Gdo%9WV17i@2e(Bh*hyJ4PC>!NG;uwocR{NE z5yc<9bdpQk?w@jkAvo1JJ!+ql(?Yl(KfN`&I68Z5VC;_pEomu$)(FD-joUSFySDK| zt_MJ-nZG|R|J%xwCL$vJ_&#ZDMK?Mk##^TNJl1mh(JgKcVy_B-R&yWoWU7j1iRh_6 zTf-L^8qC17{?#$okSzA^_V>R^y%c+MM~>Qo2g}HiH`mdft z>t+r2T+w0Ko4XC0JF>@}=Q+Dmr%pE3^%DJ-Vu(wwh#V51Li>~|mPC1SbGO-^rgRU4 zx2Y*PeuW39OM&WA^~5dL*eg$|Yl}gB`%VqLRT+dg#+PW9&)W&5KD+?DF<}UwyX^Yn zO3#NEF3&tUvvIA)@Gi%A)qynX4T}$}+PW{ikA{XxhVh2PznAL7hm}2FwAKRgGp_k^ z?Z%nO%qugXdu=ZN$+GgwqXTQ!LbTLQ-s5&C$6vElduQbA#s)5--eBRBYi?0vA)o4Bo1>U_9zXWYzTF$M^Dm%4BLbmhkI z+!rSzo7VO{Y5cCo=xjThDU(}j_z+y4*qrWjE4AAia+`foLrv%7xSN}nRwf9mmFefFItdROM_Ffc`r#?9(Y4(7XFeDc)T zdX3>>AB}pccDcTD!UP=jsSG(8od5pbWhAMxMe!bP`}-YI(jxpl1^A`KY~5_K{&8O$ zS4%8FAy_){&mSC`0<|9iNcxCVcOJ@gb@KEJa~8Py!WtGrGM!LdnC7)oUdX!|6IF(9 zouiaJlQMsEmW`A2TlaD1AD0`~zgxpbM>Dit5=7MMtoDv%I&4ZWbhv#DzaDMupWx{=pe!XR~g@uJ&L^@Rm#o7G1tp0?4 zW{G8_1}lo~rj2&HymxA@k*rTfKmUDJ zpBK18^Vo~|nXpELK0opFb{nnpsA#^c;jHPoqw?KHt1GW{Sqv-Iznwzjwr*gdx%T$< z##1|%T0yp?Gd?>##AH#*xubbkl8#}X*Z6X~K~$dCR>cR6)w#_`4l2v;@5@O_Y`Aq~ z<829kDzjcu2bGawde~Hj#SblQ5awSHa`SUacFnizeVm8kGYh@e^i9>$%xkhYt+pyp zbK%WRm*~9<9u#up-M;VJYpcK8f4UM^Mtk?eh_NSI)NFqHu*=#s<_!6h%L}i>b=PO2 z#h**--yU)5%?C_7y|f(uSXTN|Zdpr>GQF&y;RhC^j<$NPL(D7BTmHOI?j9jC`QINx zp)6V(W0-QTcx>)#Hc`$`xe|M%YvP&Bua&e8OzCZtly*PXLi6ay%d7h(W6F zMm5>Kczb!ejdjB#ajRS5BQ%Dza^9MFXc6X*7-XA)bI#vaN>o)9~G=pW$;!rX3o4^b>kbpPqbL{+2I4~3 zB}vEj5%6eH%AHoe90uXOC{12^-+O$cEM;c3(|(-?xe49s^**+5aJ-9k=F4ciaQH)i z8Lck&@wI*75Z%*ib6na20sd@QwIgT@J||^KAfsPxJnnO`d9s{-g(UU*?(+99{YQHG z;<@=2yt~_C)vtFN_c*cf^4?PSWl3U?M@}EOW;`i?q-V+GW-33sl;s3@BY>&F!i5Ws z_j=AB?58)tr^Dhcch)*O3{3FLx4!-=H+jz3tM2O5eJfY2xNjIC?Tyl~9m9M@O1E)059QjCNP8tTJ6`9v+$Lw{cpg4@rgKgJ%x)!5-qp>`Z*=xH<SJZz6xRF>C|DjD$ZkOX_wKoy~4U> zm{sT1-QcP5VFt`Eavi*2u1nH>%|(e>5@d)Fa{~hdZQ}Y`pd-~Zurlv`bUt;sAg7dL z>BE1PX7(F z5thE@UUt(cMUMopZz@mw#=6t)DTCxsJC?s>W%Q(TMP5hQ>cR-amOvdgV)7lNZ-60Uj2j z;-!Q5s&d>`T96+-#%%B#8e6qqhB&@0-8-q*t9AF9^Iw`75850brotz`WE?q{UhJ|w z+ur>1`kV`D(SdxQ+P5Kx4*iuLFlRDO=sDB{)z1raZ=PAXpsCZn8nxV;rqs`6|1Lk0 z&sWv1SSVRsA@ABnM}yY#r=jZznOC~oz5hGr>ytrIMb|RrlxV8_i^VQB>WTYbE3_LA zYRTN*`6=hxzb4fyDJjt>;XX9#V{fhgM#uWKKWIU6u4E)o-v`$x0hDi;zWTaxk>Wx( zU%dhEuXi84oNo+cu?ja+B>>y@Uh~lbq+I#NyhV-z|MypRjvZq?@HjR8b@>I1!wH0% z`_U!P;QW|T;S0Rn11(PkEI*8#;;cM!{EiKvrJ(s&&SO_RhneN?hYfc$Jf(UqP<{3D zuzp!0-e3{?dAiARZKv*|f)V_nt^MD>U*UKwcVb|ZHvK+=8a0;Z!F&2qIr=kS$7mw+ zf?{MClh5(m?ngn`uQnuW)O4ISxP2-VJH5_*hs?h$TUqpRqZuG_83xXs3g*(ME=yyU zPIb{*$5d|Ac|mUyT<)DR_$)|xN3$;N{)0~(t9h4#1QGtzWzADmM-@6`4tE_hceZ-T z9>p`yCirzSM=l4cm}J>x*c$S8-ffQ0sMWfL?j&`qclY-LEgklDOfDsE5HT*0${(Kj zxp6i86_ci(0#&LbBSV6T)E_}6u#_?H?YvBza1F+q9cNX2d!opYi>mR&ienZYBr z;LJFuQc8eb?)VHn~TlY1)U^bcJ*8kSDbl0*|garx4r=A1LbUT4<35+Q`m+u zZ8O9_6+DIn%O{#cwfN34GAx?k?tQF$2^q(bIrKN2wkK4UN`~N3^7!8$zcu}O*7}2s zx0OtVmCvsATE4)ISr(Xk?PUd=C3+_pyN{Ok%$fSy+6Ooo_V!UwLw|1`maw_g^ILLe z*Dbv;sh|Ae^dDw@E+YsCS}mth!V(Hwu4G@WowwcX7ALJ;2h~a!&;dvo&LkG&lDaGbz01ldy|73NKqP!V2;Xdw$<)_J>fvU&jb8v;`5d=R8|>7Pr>NE};bFjwwh71j zIC1gp9wsjzrGa-TfXw`D<{Xs$}P>#Xb=C&=P8fdH>Ec* zl~D$sAlR{xaI;fNZZ=eNzW2R_JjaTIOXGeOOXoE=C?~C^oVqawY?BHl=N!-?mCZHm z4S0Lvp$U>1PS*iiH4pvr?Bpy!1imrKRDWfq+~pL6?~!q97P$3L(ACzSOYl~Kgw-gy z-)0~)iG3z7O@aegq7s&;8(UL8OxJt7iDXdeo=DEvqs-(vpNz#@i>~HY8tFlL_HiZ_ zaLxOweCAsp>0BUp4TPtlVEi$`J`!gn>hZ1~_Y}dm6 zagncTg}&LD>9S%}Au9)h7x8f&bD*%V2Hn?en-yW7YiMNR%WE!BOrRn!n&dt;nN9P2He0&8yd+YH9nQ3 zGh3JOi$_3BojGhk15rhdu38;K*#uVz;lqECgtTZ>d**%w~&?-srNgR7CVeHFaG_)8^x+!{+47nhBsx{tv4 z6B`HoJD}TN2DiT4Z8mVxwTqJ4coj3pmTm93q2eoM`&7>SRmWNf#)BJTx-lYLVXrO@ zeT>1c$Vu_IiwhB4N>Ir70O}ZU*bRFo+ZkbhN;FmWyhpx{2w)>1YdLbJa zm}A*@qazM5U;59|YsZvE6HL{*&SN&%Mp!2AGLKPho92T+4FGj=0DByncA-X99=5{K zPQOCKjR|0#8H(q*N|w|%E-fNPobcfnA~jh0tK_bYt@b08&Ty%}a|hYPID@;7m>(fO zf96;R&s#YoJbkdar3ljbw#)3Z+e!(sDb0pOzcF=cvfOJ1HZCa+rFtggo$2vSM_W4u zMp_lOtOh%I{8fvay->|C&S^Yf`6u?0GM;b&)xY|O5jZ%58KB*@Jc0_->$y6>rp6L1 zv4)~L=KJfd2w&c{mydyi8rG*8iz*DcX}}gn!o30d*Br&x&!DwjW%SehD|cI(u6cR& zvE#nNbPJ`C?Tpw*JC z&E=7=+qi9cL;k*_*^U?QPebjJOy3x-ew-EblkYqaQ_3uhED5os-2>yM4{oo2toKwb zCdz-*-B^;t83-aE*ChL|_o7^Z6D&OE)ICKxS-nrb z-pa08__(~QN~_&3y-ywuS)S6MrTW<;?D^f7;bmW-vnJTuVG8ai_x<+{>W&0pX?yn)%_;c&FHF zVG3nt+qk0g9}M{RrTJq3PY|^KcHR6d^AlEe0cM7xi ztaYw?(DjCUo)!v=$5WL9LhpRI0A@lT36tV+mVE+E4VDx`2>B;tUsx^dJr9&;wAjv9 z@yw=EBibJX>r17H!2GSjN%zS<@(@@e)p7IT{_)YV%(ixE@v51=)}w~HO_q1hcQ^NX z#`AOa^6I`ht5jVo4{E71HPywA!F0aIb(N|8_S~0pXs+(%uRZFP371#1wWJbI+oJ9^c?iH+*l z+k0;*wSnd#yQ^PTDz+}sT)XR(Syv!jiQ(z>#EZA|%dn^>zYUj7q$ zo)vHevoLSrLJ!TxS{@ujX14^x$eu~5%-ZHp53(1|u41RTSSACA)pfY&yMpp}HpzYD zk&nqA?stE4UF^R7xX;l#J~(eo_<|^=!^rL6IF^r|o!H2nQ!KQ+e&@~PF?s*B;hmHz zyrQT=$dbKC#lyRt74Z{V?459VCa_YLucISfjvt?GwM{vcP&i=Bm<6z4M=}j&>FTBv zLr=K4tY4U8`UoZPBA+OL$L%oav3SG%$$)|AdcyLdSDu&409Pyu0PDdow)tlS{Uet! z5>zcX$tek&Kqyg>5=UI-Sa?INMQjjv2ze;(lA=I8LzO36c)&- z2kn;TnPpxRMnn$pfk7dd=nhw%{jgun`2yvgFbKkV)P6CfUjSrnQ(l~%?Kl8lnm3)u z3KY2dIe>*(MY(ktWL@h&SJa`rKbFC8J9Js2-)$tND7=k4;Uk&TzN{mFrjmvt=ogMP zIrn1Q5PiQx+=+f#{;xu^wyoi!dC(7+m3z5~&g=E@Lk8;mB`6))UNJ`4Fa)Weo=^OG zKP6UIYo9h>*xNcQxgDV&yIM>c!$iB6jt8(#q|DXuwPALe6_o19mij!}(=FN@`1JJL zZfxBb#ENVfnNa0*&rxGxMM^_u-J{Qj7ixx&x^De*U-2J)A2w5ap`3*R_=1To)0^vi zBx8=yqo7Qa&yEdQ^NW%eM$B}Q%31Q1*T(WEXnO@T*rdab zhPx-evWe>!1-CT&tMjiXW|>s{s+UBF9VALF2vk7Ts3Tr^KhiBYYd-KE6>+m45*Ab; zc(fXiHc!kogHwhI4yLx8IktS%JEtzB1L--c9lJVn9xgtU(7s;>2RBxeR3%HbtIpIlpUr$ zgm1^(wiw_gh%hY#wQGBfw z%5%M~b^ce~W~@M-7Q+)sb5e24cQ0mnpXEr+bP0*f?F?UPkn%X90|a92&+J+i3?o_x znze4vvmok6I96?}@9sBgC3T_X1V^vj8;T~Z2L&FNaXxRXQ>(+P&bli`Eknl5l4mP`krtZbDSR&M)k1%}?F<`>EFD4%aySGr3MF#O`v~!Bq&s z{_4~C7hv-HBc03wSy#3(Bfx#hSoYjUuw&Wv*lJyBw-m(Zvyz+8no`c$Zp$q=ibRuv zkHo-Fr<3F`o{2t|U_<7+!5-^k*_lVe1cvqa_rLlI1Byk(m$x_>~jCCQVUlC4ZjaHdbBn6f~nKY>WUt zzS!(kIzS#_h-z7b6l81kp`<-fEnRAR=}(H zg=169f58emt_^691)zK+aSI*`;bK30)i|}XWSa~-S_PPp+yyX{v33IJ1Re|L$awYv zq!}rBggU_kmQN3cW^o9m`I~J^pMQ|3h$eLvqtFnOC<96aSZoQsLfdfj;%&O3@0#F) zFIjZPQ`^JceR$s6mP<4$KVB%vj*W$z{b2O+s_?V`iy*|eN_r}06}-I!&o+k@suyd| z^eHW$F_+t8An!lOK+EL>S2YpK9PDygc9*{zRXf+CRpF_eBTwLLK(=Ah^~M+EjVR>x z|3zT0t2d&{dH2zpN9LW$-6GegarZ3~@C)6{e)-haq#Yqreix5_%gdz+37>D|w7}Z& zX`gNuta&hG29+-~|4e~bd>swAvT>gmQoH%_S~xWG`r8h1FMf;ukVc2!cGlqmE{t)#l>@UGUa3dmJZz=AYOxhe6O71gtEzjV5^x8dgPoPM@T@yqE!xuCZ1N3~isg$i zU%>@r;J$tP{^;@~ccIS5RGY4~uRXIv*++taHV^z#Ec_nvAIEl9mU-=Uegj;Elq5BIzLTwTnSIf{T-`l6FDgV1C_z+Ts0_y} z^(ms|&HLyV3sJeC zTxQMhmb$w4Fdi8^l0QRAgm$X*yoZMWRXsUdZB(y$`EUHL2VQiKJYdGlB$K4!b=E+8 zU5J8kH$$lZj%Js<>~zok_`csDigpx9e%-gc+1clzYQ3kNnFonETQ)?03X7K@WwIer zwWg9jpcmJQ96OA#+Elv3)x!02R0xIldB>Vk@2DK-af(VI<#n%JVZqJ`btK?~lsjME zGUV<&KBXBh(DYEsih7K@(=G38Ouh6>jr-XH3lP34>Z1UnG1AQ zm>4Ab^xYKR$&uKz%Rj$b^gV0C5iuO>Z@vO3QF;dtzr1sr2yGW=xP~WPbYyGV8YmW{ z%xk=0`Al+d`>T2(yEDaFm4-$QTFy}P_VkEWX{oX@WaHYGCr%DD)CQ`%vD`F}Afv<) zs_81{U*VDV&Wm1ke|nR(WP#h7nDTk;n6?5sPeX&Qdb)z5)%HOC$(PzBb{!g1Hdhrdz~?F_6vS)RR;0axml46D={qUu zT20aSl}+DK1wp}S3~=FD8iTl=(~({Mz|V%kTXzMp8$x-#OJ( zzsHdU`u?14{L%Iw&pAnE0bRw1UUR3dRon!~QUup$*c}Q&^ZGuxlw21#9$C6=KEJ-F zJgyzFKBT5X(oKdKWX0ryF42E=$@Y`UkFi?#h727Be@i$2Y*Qq13H0~3(Oaf|uQ{M1G+w^mQa2m^C+ zjrJV7G&`(WIxww~kj36#%7J)q5<00`p1=)y5#%+~C41ye$}bNr6&M)8w)|tB4W!wa zcV@FCJh_9FO{o>ATLnfVHENi9?AlMSV;1pY=aM6@G6UAv&u<`_TFq8nSNmP3>b~`c zCNb2O{Kp2$hQCAxq#1o4&WtpT?^(y6p7ML}MhH=Ag&R%B`knc%JNlImc~fy?_Z%if z>4BqwZ`r-G$TOTX*pQn(7mVUmieJIF2&&UOWHIU_R=mQfC*GQ|xCvG^M_T8)8%ia* zwc>`~13Rj1>H91ItjvME%K5z0I@5!B0qi(&m=@B5kJlf$!r4CSVxf+0DS5~e;a!4Y zC1lc&1z}TQ88jB9YWOoLY68$iKWjwzwM%n-5Ac-Uy>#~;Tzc6>k&`nMAZ_HOZnyDq z-;)D)5)Q6*`AgF6?*4uj&42M+6S~IPyjOEpPcB`wotfpjF0;1YOWo5J$8`v7K;?4x zrmk3Ap0+bGws~3fRQV@(yw0lOG%7VQn&5H?^Us65y_A)Uy~o-HL5cLqe={w#Q=j$+ zmEhPAVKP7^pBk&OykDgHFqW5*C16}7N#?T36IsRqp-+z%l%nPWWKYN+DDA*Q#g z_Du+05T6GV`Z-U3<@n84%WZRu>wulY>4U$@Lj~t2SKG`!5xDL+C=ac(!u21W4>R(Y z*e2kWRopgIOUx?Fmh3;qIcM#)tzO(XMZ87Z~O={tRSh!aS%L55~%{snR-LXL|{Vt;0=e%|dO6cc6h%#h;nC1zKyBEOvWsbbD z9m@AEHpg+aS;F@d&(Wc4ZWfIKcIhw+?m21TgqK$c)rpu7VUM1RQpgX@_w!sc7rns{ zN=M77`(Ru|JlIO+LxooqB(>PU>s$MrKu+cxVyDvCkeg!w&>rm3i>Iidu;9qAH27xr z%FJtb-(J4;Be#pMzyHj+#_i7V(*sFY=I@Q=_i<4N z@{>~=4mf`NxNV|pMn#trzxw`m)hpbe=OHEv3!&I;^}#ix!WpV#UQMa!uRfGt>%8K* z192urz64I4N#d9i(Pcu3&4~sKN*Ust#oQ zlLPVN1|`^{7xRJ(ytaOpqNOS$;odx~4qz~CEv?5-@`cM#IO_ZRh1RH;HQ0tRYhmPi z-@)-;gWI?dJjz8C!ZG>L4+v^SEgoad!z+bR;>}5;2-Z4~Ic2m?N7}m{cu?8$Z|y#v z`&~R#^gpE0KTPebopPtAZAb`($3jmC)nSp((uqcGSW!k|BiyJ0E^hr}&c|Eh)Kbq7qBhA1oU& z#`yvzq$OWq!S+DM&f%M0g{?g3wdGwCT`%<)u})^&dmMa_##}x!^T>nd&E%_3c;0kH zx$wjID~GmmU@?LVJj}-)GvUi2zR2{GDGxzAy7z@_pWxSIzDMF$&$YJ}XlZHH`2Dzo zjeJUee!flVh;;RCamNayJk)H?qe2B>g$m}H7C^~}Liio3&2f3{e4Y3*f|?Sw(o0+{ z*_~fqvxX?21T=lsDRLMe!*M0+PoD}zf`z}|?Yi|ME(>4V#en8C!E)k%BEPJE|LB~k z%v97ub7>xr+!Zy3*e->|D|n@vvBi^?JijAOUYeoH5mhU{Kf9<;g}`A4`HSi712oeh z|G(th{}1j~q{a|g#IO6>;it8biT8SfYk;8-X%pEVN1DT=_ z2Kwfosy7{-l-r1sr8V~IkJS!|RgoA;w1ms^vditHQJSDEQZ(_$Po&RQY88g#+8_MU zZnv{9;fb^5(26o(YVW^^*19fLIeaiR4bVe6WqNgEvynzHSc^tyt=la>K_;6RZa?8O zy~lO~5zxhSB`hB>0v6n1K=p)4Q|Z zXL8F9?t0>0EKeF&901dD(X=koIofR-R7bj21?mj{N1wUlbzS*w8W*#(vtsU;-h=nc~NVOZ4=2T@WGN}I*;FWLxM5Y zVF7Ztw4HZU z5v)CXwuG9h#uh%4iyj<=gBD96$TJ3z=v;Ae2SSf7 zB!JH)P=|f0tb00Y8BfFN%7%6GP90u{R&M9jHLhgPtbUew&}+u%^h~uiF|p5==6P4e zGhd9Ui=RUYR9U+Et=B^&63?_!t3OvU$5L26jJK=mVE&-q?gO<^O>s+fZNXHZ#US>R z!mv#5ZZ}0BqGL-S4oDi>j$m4ld}sA~&$E`fL8tjxYIcrt&&1iT9>xd952Sb>IuFp7uXmo>#A*JTY%uBbVg1@tj%5wS!F6<_o&QXeaA5I5n{6D-l#f*S-u!invc$UPmD z-V$o~_IX7iW2pr72echq+F^53s$|UezPO{tIJC))$6=w5*Ip)vBY*i!wldeM z8T;Jeu1aAVH5}}X&_~2MacDvM?SA-H-GQ7e+3y0y3W@mELUf$Y% zla8#e9aL1TE)gzXw6yTD-1qd5Elr$14TmR=AuA1zPl3#%_`0aPWvYMD#r?G$m3mI&Z@<|e-rDk5~8?xOYjk6#^XLu*M?2)tqlt;zZPTp3&a?pGzD@TS!-UT#O} zb6Og(xy0UBZl)hi(Ie)2;|z_rW?KoI2uy89Pg86S=nSG*@!5Sxw=_1v^RsxOAZ7e8 zCdu|@Fb@Lt9D<*(|dPxB?SyRyN)bYxc{4bQd_I*K5{CH z&*lthdmyu8i-z^-AI6U3ddOK?Xcxvh#Q0QfI>=>6=L8*5!`~8HvSy&AXsa|r?Su_I zFSM4Z=e!+L!$$^*Cw$m`biLP_+KAgW)Bv8$IwQ*_ynDA%t4~1%AOB+4dg=31T&@#}jzt#FeR9jmZv2ITcjnN@ zDV%~d+`E2g+)`zVPf7BkrjOEN)5cWf9IHNmKdJ*?r?3A4mrnN_Jg4H$!+p)rF};+> z)*YoSh}Z3^7@K~>Vzb4lqC~&)w_W4*TN%2$xSXdfvl8ezV);Kp&^S7qmA-z|wbkXp zGl@(Bs%(;x)z^)G()FWqi_m3WKPI|PN$%Ut*m31}r$A@-RePt0du@g1IN{+@a3s&p z3vkVT&`>^dzm3jQ+y}Y~NvHTDeHcB_5-*5+(y#&sm`kZ57%B$$wxn_fSzH7nR=O(d zKCt`62`HSN-!u4Eu)CdK*WQ3vk`|{4h$Ll4vZ?o6-d}I3**(Kn{1SEg%$?@nB-N?c z63<{F@|$EX?7mnPOvKSQmEMDK#nBf4rUiIx=b|RpcG7>5x@W?$$oH)^kJ^p3Rv{-Q zsp8{tTlx5!wd18Sh<<^~VS^<}V{tM3MD?3>%T9YKcM#QH3EgiML-g7tIUhWF3_0OJU+kRb(C;BE>VMGw@aFv_0ivdg8gnS{-&0ZDaw02i5K0cI> zBWX7`ua(^4BFv)Lx>oOL0Qu>XXZKZo*qY|_(Wu1G`PNz4>`?DxkN)-^G4bJ&;U8lO z1Gvb}adxLuB8~rGR|bKKjrkcbw5s4V$3u>FqZVDaJj_@~pSGN&9`{u;)+2pdF{#}A zle+n&Tva8jS#nEp_dl`z^@l}(v84&5EdDxiqOF$y>#sLO2Zqn5M&_~7MF74A?!Ts9 z*pp{D>Xu1{Yhjdl>|mN}SMdVG2`mzqMEEz+W_OQ(EeNPb`R_w%_WsC;#Yb^0N6n@v zen|Y+MIN5*bIHdLK4QcH(MW~kSk)%|TGE*X90>IL58I#*H!S^6+>%!zHx74qD-H`!*bEohs@y>fhNA zZDH1#p`QtopW&66pVa4^4aYa^z=9rJ2UBcYsnB{CWXC1tG12kPlBX?eDT*SwN4ufC zGS>Aw+lkz!<}B@q8uowb zhaK>?wC<_zC~f?UPE)9f1wIr)G`lHk_IzXKCFq9_v4PXO=XJC-p6ty&Nhn4tE?ei5iuRXPS3v(`&yWb_4-osZX5nS*0>Cx%7Qhh-o@L20}*B zzgH$>;JCr8mf$u?_xALX&_c11*m9XiiOvVz($~tiMfAPv;U>&Fl^fN#d1s+}y^n3E zyW2Z9^!VRr1^i($8R%G3<&fSl|II(u5Yg$NZ=99Ydx}QRP7l0hL~vF+DWloyyv7(D z3fLL#O;a6b$+YiCnhvlL(X}?0BD7qHO=3bigXmN#GGcyyz8M9{Z9s>4?aq#+>vWJy zIN@ap%E=-u_dk7bg9$o$EBZ&C*FHxf{~Nk<~zh8v)3c zJTF7e#@5N|3Tlp*bLwvLgnr^vk^)u>5zTdDfX4@rY#HmH^rZwN-}ikSnoA=4V!qxa z!_@QiOopbN$Alqg*$t;0K?<|nv3Ph5GALAnXh+LUQO6wDTZV&5(GuZ@0HIK!SiCU!p>c5SE``H9j4jc;qHi8|8!zK z#3vDjvcrtW-f&QfEryZGe%`I_I2r%@o9+2h>#q9)mrOpz8MI$yRr76aqFBxy!f!n* zpPpA?n$@WG*+b6Lj6D{46y>}YkaPC8alzXuMx|K`o6q3d`MA=E{lJZlsk);rU|@uF zmY@p7tUb2vij5|tLoV5gp`VngQTBB56Kv6ruX3xIlyMlwUMc^+)g_Q`5CxLUyUgP( zmz{{e(l$%Ibp$=JSXkpi|FldpuYXlUdmv`=qsY2%b35|zFQQJN*87F}=C0EzIt@u6 zIYD!{i!efr-JHy%vpHb&bfzWCT|nKRGq`#SnGIGEkeu@V!(ImZa)K93&Mn^;JUj=XhMWvyaBcV!A^Yy6s+^jR>ny;h>Yq5YxpWRl&+I)uD5@{7&{V8@+R6;bfOuKLP zO31L9H#paWXd5H~IekiN8?32+N_-_%g`;;qy6~98M^HbD+2`Tm zAzF-n&tQ*BZ>1}??qh110+8bXmYp8WApBkqH*M?Qh=M3QDj|1NbGqZ`b<0Y z6rb-!G_HZzM!BJ0d5=rg`KC^)zPp#oG*ygRE0euU1dsJ>b(BOyA=7@?;)`9@+?Xg1 zLR@X(6$9%j+>W%w_imYeD@M#MT3>_p=>D|x@@LOo0QK5}QL*4w_<+SI>UzUFig%Y# z>(9NP3eo4%V*IJ~BPZ(Q>-j!OsP`D;VZ{s|mv_NOGJ=g}V=qw8ou3;m;e*D#OKg02 zWOMq4c5X-^I)Z2kj{{A&`E6hbp{cayf?45|VPpL$+k|1aNzdRR%MKQb)tVEZ>zGLe zwOGAgxF5juD*mKXy;#%gB!dXk#LF1y1|$Iaa~= z*q1!$0AP_`@cmL%s)ay62I5aAf`6TSAoorytxB3a!q(dpq@v#b)8$uh#KuhP>(V+R z|I|I#{!P1?IRreg>B^155c6?whsAmrWu=ADQP<3J=D*i2QG|56{^7=$V(B|TOBtl9 znr?d2eS=mCWqkm#_Tgd2kveC^E{C`%#bl2P86^g15PV^=BmU`GmA{nNH&s)9wR7&s z8SYtn(s=!T=|d(7<)<5&8e-Y4TH?3yvbKR@>-|>d-6_ML?g?zms3cnx>E&Eb%wCae z=S4K}gTK-ie8X+Ox83ha4s2Nj#l*$ae$3qs08f};iiyRW+sBoS@G1K!dUdz5Crx(-8_MPW*hoC-yi({#Lrknq9i<8S)8`>#q@q3iVA zFT2tB5IUkZ7l&NEsv?dg0c_$UNlP=pt}zG&jB>M`WJ)Y!YF*tFrCuVw9_WG!`DPW( zL?fj+I^|BpzQ89P;dNwCLub{2GxJVTHgT(rXKR$7(2UF3|6mz82{p#1tKb?~90T~M zH2Tzp29n0l_iD3@4QIAVP)S#cW%3Jxz~?=0A99HD_xJbGBTAAHH2*>FQ4^b#s0_)c zN=9m`(+gEwN+nsAU0Y)`TeJEql~r-*Pib@{$_Ik%g+c|V2S*LnLaY7p_?7XmNU*r~ zFN^DKEn0D|S;s{@Ukxz=UJc-oiMKwD=7C^RSDnd|q8?fH1XH_kN^=dsQGv`d5P{=e zF!3aBWa-ms{1d{4;@}xRbjqs5v|okx7?}`c2W{yqLBnUcQLxIyY5H30IgfY`Jw?AI z35xp7y&&czd~@U0*7J(&8yUGKK1|pQsIi`FsxH`=7-4_Qh^1{r`p#pIba=&yt$cU=QmGnI868F z*MwfFa(_D?wuE#Byx17Zx~z*_;qKp0_p4$S?!0Oa`ZMrBcAw`<>MU6U@eB(}k|vD! zp*MA4T++lnf=*@u7V8WY1r)t=97>935!QGSc*tvQwWw7u81@ykl!p*9H9Sf)lkAK6Q96!Z$Y4?cZPdW1TQB zS2^-zPcLb=LC*|e#-jAsIA&dR=iZ>3XBdzI(m>_`lHhR=rox{aGC5JJ#Bd}nQMi=F zX%HKfTf$0vQ}JRF`uk|v)HBj|Pkn?C4=HznOk{;KQpi2W*C-yOub_N3BMZ(m^wn-x zTPmUs%xEY^sE3A+2O9>Rkd_~AwAapYbpi}!d)fWE5vUh z=jkjeBBWQ;yOogMzOMz-j?sMLpICH1PR};K)NA{<1sZhBGFW404||n~QlhO>8=PD! zXI>g_L{pS7OxUML7ODXz)77WXq?6R78U-cy=xd;hQ;NxgtjKVm3a3>G)mN<-k#lMJ zHB#xEE)*}=>c^1uz5ot5g@GOaDxmu#74vAj4i8l_jPMRRuFkMsxxkH@{(@MU^MLpK zr$hVSp{=UP$i^zxo}t-49(G)g3hRw=)t<(*UI^D%}? z=bL0R8FHiP9Oz>4(%&=UH2$P? zb+Bt_hr^C?fEOn^8ei)^JLFog{TSytOwl=PP;&?6QdS|R`_aXclUwE zr6dHSUWA-#+%_?qaWR(kH!P|oXm!gd$TSE{NZ2O3OAv0%vqqE_BCJc})fnduclFG2 zEnSgW(wQNqf+>NQG*_IuR8op4eH=U-=X60XpsK99B24^(P*^FMk_vTWS}B+VFMx;( zHPuI3G@1#F#W3M>gd~r$t)6nMEm4# zXT>urqhER)EmKc!xe}qXNX9M@M17BUjyHCcDvT6aq3Jm<2cH(DL+7y@13pLQb&m{z z$DU=mI<<8SCr_ZZNZ3TuRoG5s>I5SK`X-k`CkDY~1OTCH%z5eAyQ)^<2nZC{OS{Rw zyL%-Au$A!5*kshjc7>r`-QxPxnP=oXoDSEiSciro#t7BTpN)9{G~gL6;X=X%kI6rq zENo_D<%@BIm$;;$>&hXea}UDuNv>S z{YOcC?_XVd-rK(JmwJz49V!4(-p)(vCwvrIYMy1Gv*rGK9X)kuAR$!qj>cfO^@*~Ax^|J+~fag z?_9iU%=a+fah!2-W=hknT*l>~L@o=tgy@WMViV1gE#;DHiKM1D6Qk zI7t3B-!r~V>pX3U)^s57j_H+in`;KrpTd!WvSq%3$XV(8q9?z(c0>(orN3J|M>vGo zh!ktHD7ToWBn)wg_XPA&CR4OyegkqzuJb{QbvD}X`susKxipmf%(&x$C6rJUg;Md% zP}IXI7|%+hvQ)~vO0yV=i4=o$ zS)S9>I3<4{?o-j3RR?;q!u(OiI=-oS=I)Kgx{&+fl)9vLwEbK&XbM+rz>>}XJ@3YS zp%0NH?U<@xnmFi%5CXCKn+il)OutfgE9KbUOrnS2FZwdKprJh!2MVi#_f@MLN;9uS zWGYMKP*Du$vYs@;388yVbc$GgVW!AJ>ShIo5+H|EIlz3F&mvYPcQq6C_-ly~`p94O zM+Q+1Vmr0X_3o$UxgAGGd@DR3)OA4KZ%4srs5UM*n~YzASPjXfz+;5kF!l1X3=}yQ zH{139450F#fHaYt%@Evg9sAO}_zsc8pgRJUr#l=BolSNR^bo2z5uU)kZt+1)5*fW0 zAt(}Ju9P)86jgpgUC`P5?LTO#sdz=78{gXXV%oXKgO8sz-RpQ60+J;ii4;oOFQN77 zzW~9^j9OfAT_aKk7%vTKDmmQyS-!CGD{=N|E~}7a0K_2{Vt`*L zeRv|EQ);x{0HC7fqEn56%xsDY0j80-CJ>>}3GkvVq}*}Tu$>di@)k{1r~{dl437nM zgDTu%w<2-3s%~kQSD}c|l;ijD!L&iT4RXe-lIgihOS<6h$*F>l!_tT>Hqi^9+yT3( zh3fQ9-+iq9viNIJk|PHxqIUbFD{o^s;>Q)CQu-BjjB@b9yKCYMxue z_vDEg0r-Cg+;xE;>*jY(QcmjhN(=c*&aT|rzGBSSv0C5*x!l5o%PN@rf}CvYUL89; z{fgg)~8;aF#0{LJLa}&4!vCJp{NbJ zlWG61+k3mGbI^dn6c@gykPZv_DxPZMAx9x5_!&?ukqV1v4=GQiOpEND$AhhyGK!)D zH+fHyD9yW|@!8(Du)2(fLiQ^4( z(7KagGH5xaHp0>Z3VnQl)@Sgfl~)ngcCw>g#I;%x%V7Vq#w{y2Il4<-WkADK22n0t zp{PwqsIpjWFl`pjGb6>ks_iotf(#tQuo-NV9cXL=x+&aND<6I2M70kP5Xm^a7ZagNS{z1j2B z?W2yMbmN5WOZf;>agj`ngKFJJV&KXc#K?Dhk5rM>~Aml=O`_t~BT614QSgOS5|=)3uNinW)R z975NAnez8p?Wg^w;BHo>lXm@CB1b(H5tZN$>=9a6>Q)LAMErm_lDLDO;HG$K(e1<8 zx2e$zS?@&$gMBW@A-tGITPA{E?p`e(@9@JXTd^LHHGImgdBzoqi6r6Fbe#w-6Phv3uY;3Rri zxzm`IIjq(UZdKgxULF*?w3mYn-vAjgr)3RQ+&OVfe2YXlE%Go&(d-Bfa@&{p#&(Wq z#f5%oObgH@@t|pnxYyc&PEeg%TvS`Ba;Fc9nR4*jA{d%-SFWA`#!D>YrpV|v?Zr2+ z)zbP`<>vm-dMl;_<9N7m;pYpFr^7$!+UjGM&zfrnPpW;+qJ3_R=`nI?eDMSIoQ9vi z8~Xm7!rIrkd+R2!p+xpv-{;zXcSu)+_BPB+**8ZZ$2 z=VD^i`I0m-b(ojOT%4n9KC6@(%S45*32X0Kbpp=+&-8^9n_3Q9+c&?Gcv}0{qzwer z4g-sN2T>spd3yLuaR-n>1B{Od1BmPKh3?fQkevgN8Kd@ny@x==s6IFhWp} zJxih^+> z8cd80n#n|48QfmVt~wk}sspSP2ou0Si_ZND$@yu{=z;gVmAeyrs?)EoYx31^me8>- zWAPwhYgzLCM6Z52cak;tzVzloyJ@VcyLV-#5O!kHJ$dsDY#vqWB66Gjp{-N2IJ{57m)iV_v zcl@-&dgjr+mCgD{(~pGkD=~-RbWCZ^%wNLWS|}e@4INZZD*1xLrF1I_YZVk8QE{fP(J}PW^?*+d~i!OVb z8yViw>m1fcZFVTdjBXqKn^-eK(OHGA>;5SZ4oW0n-%iZ3^efBG-W5`y@(NoHLq@4< zW&A@RAR%Wjc$aPfrO0t&P)dO$Vti1Op<*8+T_PXb$yjqgZsUqR>QYQt(;Tve`kL}D zFTFvUJ?d3Kwoz`C{4lbUUBe0S2^5Buz*o6sImKtNdDEMo>My$ana8TF1n*n6@Nqgo zC>PCuf^8m6QUG}2@6li5D_S9LD(pquP<5e{UcVCX9+HXuhh&vVBw3IJG&;s=yM+Y6 z#S)le56C(3p8qt>N_3lgJFDM6E*0!0xoEBWj=zHvf+wFu{|;x|4`jv%@W^~)Y=5sN zmhL>r#xmgMXX~3xxd9LMprd+;U$1Stf64m=Y*lLHlS4`hDuUI|o^s`Ni5g~k-z)53 z7t1fzSQrWFPs!h-TKxGH>c7S|ZTz9eOVGFqtQs#t<0YUtXp9Aom!NSKG;Rc{qu8c( Z9>WNm?*Fo@Vj2Jd diff --git a/test_wandb_scraping.py b/test_wandb_scraping.py deleted file mode 100644 index dff198a..0000000 --- a/test_wandb_scraping.py +++ /dev/null @@ -1,65 +0,0 @@ -#!/usr/bin/env python3 -""" -Test script to check WandB run scraping functionality. -""" - -import wandb -import logging - -# Setup logging -logging.basicConfig(level=logging.INFO) -logger = logging.getLogger(__name__) - -def main(): - """Test WandB scraping.""" - group_name = "noise_check_experiment_multimeasurement_vs_correlation" - project = "sule-shashank/jamun" - - logger.info(f"Testing WandB scraping for group: {group_name}") - - try: - api = wandb.Api() - runs = api.runs(project, filters={'group': group_name}) - runs_list = list(runs) - - logger.info(f"Found {len(runs_list)} total runs in group") - - denoiser_count = 0 - for i, run in enumerate(runs_list): - try: - config = run.config - if 'cfg' in config: - cfg = config['cfg'] - model_target = cfg.get('model', {}).get('_target_') - - logger.info(f"Run {i+1}: {run.name}") - logger.info(f" Model target: {model_target}") - logger.info(f" State: {run.state}") - - if model_target == 'jamun.model.Denoiser': - denoiser_count += 1 - logger.info(f" āœ“ This is a Denoiser run!") - - # Extract sigma value - sigma = cfg.get('model', {}).get('sigma_distribution', {}).get('sigma') - if sigma is not None: - logger.info(f" Sigma: {sigma}") - - logger.info("") - - if i >= 10: # Limit to first 10 runs for testing - logger.info("... (limiting to first 10 runs for testing)") - break - - except Exception as e: - logger.warning(f"Error processing run {run.name}: {e}") - continue - - logger.info(f"Found {denoiser_count} Denoiser runs out of {min(len(runs_list), 10)} examined") - - except Exception as e: - logger.error(f"Error in WandB scraping: {e}") - raise - -if __name__ == "__main__": - main() diff --git a/validation_errors_plot.csv b/validation_errors_plot.csv deleted file mode 100644 index 63f43d9..0000000 --- a/validation_errors_plot.csv +++ /dev/null @@ -1,10 +0,0 @@ -run_name,run_path,validation_rmsd_squared,model_target,data_target,subsample,sigma,total_lag_time,data_datamodule_batch_size,index -noise_check_spatiotemporal_repeated_pos_m2,sule-shashank/jamun/jjx6l8fg,1.287444397287128e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,2,32,0 -noise_check_spatiotemporal_repeated_pos_m4,sule-shashank/jamun/psxh4cl6,1.4849490481435916e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,4,32,1 -noise_check_spatiotemporal_repeated_pos_m5,sule-shashank/jamun/ss5qjtb3,1.3529375963257174e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,5,32,2 -noise_check_spatiotemporal_repeated_pos_m6,sule-shashank/jamun/y8qzwcfy,1.4670496489178373e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,6,32,3 -noise_check_spatiotemporal_repeated_pos_m3,sule-shashank/jamun/bp5p8gbq,1.6283258377672115e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,3,32,4 -noise_check_spatiotemporal_repeated_pos_m7,sule-shashank/jamun/do28iggc,1.7884779541672415e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,7,32,5 -noise_check_spatiotemporal_repeated_pos_m8,sule-shashank/jamun/feo5op22,1.7367485100323023e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,8,32,6 -noise_check_spatiotemporal_repeated_pos_m9,sule-shashank/jamun/4f0aw2ad,1.762919576003462e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,9,32,7 -noise_check_spatiotemporal_repeated_pos_m10,sule-shashank/jamun/lik3brd0,1.891950703725839e-07,jamun.model.denoiser_conditional.Denoiser,jamun.data.parse_repeated_position_datasets_from_directory,1,0.04,10,32,8 diff --git a/validation_errors_plot.png b/validation_errors_plot.png deleted file mode 100644 index afc3849950fbe0713311720d5a62a84b13d211dc..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 169982 zcmeFZcRZGT|39ulC89)pT=#ul_wV!jeeVB$e||k4*VT22^E{8^c)wq-=X$*ZFPxXBrevZdBO{|$I47%) zKibL2ww>Hcf$y~M8jZ#ul8$mZjv6*s9bJs=&B#=Y9c{1MI9|8Ba=_Wl-oeu5h7hj+ zKd%ta0SiY*TL(!#KI?yeg4f30obT$9ydb>FHrsQ$4rFA!hlzhS+)R>jCfh(prXYJt z^JdIgx6{qt+J8hR-TR->%F{95p{L2H+)uf?{q!XZPE|{#OKG`&5z>BJIKvn=?sjq6 z`kS{>!S>1rgZh;%Hq$1~q9r*`gCDOr7`PjZbx34~#tNDH&%`F|I=z*4=YRd8Kghn+ zO?K$Ne#Mto_euWOzdgid&!{gi`(MB0cKDD2{_9urCu7_F|MipYq?e$|IQ(DVI{whz zN~iz&#pKSoF;lev_jgu%{Qm!5^#AM5|KBJ5|6OeVzb)_P_gN+%pR1>^ zWtx0^FsLVOQZunUbII2B&Dct&PS&A{r*g6rfzO{G`20j*(|A{be3YQsWg8nlk^{T) zlSjQ@*lI5xyObZcI&>-jdQ`8-~v-WS0) ze&Em{H`d+N0*sma1OM70$p-}MCMf<>B_2ClT3T+UqWTbf+=Z5&-Y+@%XyU6^@7vlK zoSmJIUVOc~w5;sM_&BwaQr+r}jHb;L6h9ZI`U@R@?Yhxj?AO>HDioT}K4b~TYYZuR%9@_Q{kSSJqy)Kp3>NJf`}GYB=v&1+mmTr&&NSsS zFfhEy&Gn6q-T(dj_r71h0)yze>X~8?c$^{@%Ux_tU5f)*b({o@$C$QlAnz)&bVrstD%KT z9UU*u$4U|}+nH zq=(EB%_B}~m5Er%eCGpPy_1rXetdZR(@9W!e5XO~oqBTy<~HM|oCH-;WPj zhxNK~`}i&lnsydBaByiR-+Cy$y!Gym{p!1QwveTyq!hZ%aYgVN1l``W<=y-Do8r9J zi`2R($T&GU1A>FAi2KrCP0f7bCha|jWABFJymuHG8Hr0syl|Q5`S!+mle&fm_3Ohz zapB?gZgZo4N0cJ^z85(ZdjFB;)i1myv%a{=(9n>qC*#KaV5HHN+Az+uD-ViSC%pfB zWz;b@Hoi?s&-Gj@T`xqtCgjLWnzi3WJv}C`wK?rDUIQjNj`JT{Q?=D zx6tb?|M9{8A7f+pok(qi_V>0I5L>W5NAB3sfp1wR*;&=t=hG#+vZkiwPHEata<6x& zH)}H;)_EKda6grQuJx&yO}*%W^Q=!rtpyb&{$A~5|8&uFH0RHsA8AdEWK-E{9QKMj zwX48})v(O%h(fAXIun!Vb&J{JfaJ@H=g(JT3998>RasqI=q*zJi04mBM;DqP$Rgu) zAD@*mN$T`gkL6#lMtaJQKa=r();}#MC^$L$OXF~^vdp$^+rAY$aW5}S(4cO7jy|%2ox@t5tY;dMph_7CMm_n8a?xgbzm=c~#=x z1g|-nHh=vpVAFF#A-b$gN-IM@@Y^??6E`Qx+A|D7OW9ST!${*?pX1}>8`E@!JXXdF zf_xP66qJ=^&z%d{sdrXZ)=&4;J`&a)y%W|a9_g6}b)2=gC;okFOg*mt2flFWpMAg_ zh<&Wstd*v78+-C}lKS3s78N?S!fX50erTm?9Si=2MHLVfL~N-8zK=XvSXm!Flv!td z`SK;Bz!hJ$t}s+1T-N2;ktT}BERQAo;BnCpcy?Kiztn@{!Z|f4(k~U9&AZmxH#9_X z>C&YSRaFNC1$WV~oOnA}7iqV)?8L^wu`lTu*@a?9OY!5!>u~*FccO@nj9B1jE!Woh zjLJWJ2&9|BYB3q_%vQZhqH$eQArdv@}@{1zm<=856EWOY`oF4E>@* z+x23k+(l7GLsI9;GqFj33=iK`isZFiEhV;jOY+XwnkgGOI5>24b%_sT(^FcBPing~ zZT7vyg=X*Gy{_Nw4;|tfq3qOoh|N#t?d^S^Nu;)Km4;RNLro}0@VCD^KG*1fK%JXq z>udVjaPGUX49m3>l9EjF4_T^8U1x+db_ibm%qVQxyg8g(yI$=$YM=k7PoL5jPz1zo zzH>TzmI6zJQ^CR6xf)wxme*=!ew=9&35C(8*K36uZ``)`Lku3#+RB1SSHAVPeCvHb z`}^l^d@S)=osy^8C#p;mMPDg&U*IKn`1U=#m+@mv+EgVD!#kKnuK8kBCZIS-Tx&0R zt(8tGCMM?V=eL3EcXx?-Zg!5tk9Xv{dKF%4_GLW_tzn!R!W3tx0Ej5pmy6Gp@H|MC8=$<)uPo6EDD>uU=}Ey-$U@~zd?ccC`$ZGT=LSXd}7D=%+e zR*;t`qSw)(W|UW7RYlIr%UfhPu`+_+)+w}o!Q+Xy{8ng}@Utd@g@uJt=ql9?X7O8h zX;^=zW|lV>jvm^%v+NJ$Phw>zA9ZkY@{5a$WBbzRUV~?u{dQr)CJM&yqJOW?TmJxm z(_5&ol^B(nI?EVB#weSXTv_E+ZmCf8BiwY9=<_4I&&v4RH^{!ot!P42m5E6pK2O<0hsLJ)iK* zGk#Bj-l)`E>_qadzKY8C6tL`kHz_AYkj*_aY6eB1OPS)dTP}}VQP#2@I@B!xceR#sn|+kFxIWs-9o+6yY(eQEHv8i5ekL3BdAel7 zzyD8D9FI#~bi=#eURIOp`!#As(OT*IM6WX{tEe~)1>e8_p}WLos;_dB?Af!2czJh- zii+AgJNw_gyM;;Eg3R1}m^m-=Iqm7b>A||2lYJWu%RK_IcTjl~6vMdye@a74 zvS!MzYc0u-ImWkc-~PVHVR&+J$^@0XvZ-n3@87>IzP&kET3R}tNL`eq93_~HHhiPE z{Fnj_6_xD5?`{W8XJQkNjs{Im+LE0-cW&$a@9ukm$(+0sT?^;@RlZt2E3oN3d-JBava<4p(wlZ+$1X9HmzM`sb1vVU z`bqw~r|ceq@8#tW3kdAov112u6TwCQv@aTAP)}}_Li-K!x?P4 zJ`!oWl@)3AM*Ni@KYr+bbkfmbL}UChIvVYyUva!iynw(2+8O$ls9KN1!ZrhxYo_bb zn>XKn{J4dhN%&Li<^EFX9k=fY_0*l<+e}V= z5bcI5oH8RLW27U81K`_fh)=OyA&O{w(`V6O&`j@Ohg_Y@Ly=0pn94BPn)<3R?wq3H zI{?$2jEn(dHa%~fn*~WaDLPqKGIFJGrF(ew-vTkv;_XrJ3JVM0e)&QT5O8{FdN5ee z$jwcXz*QwK6F>190fB+1@ExMLU8xFsfn}JimO!Dbe1D~M2ul+HbPLwRJA4R0vQ6QN zHhz>BqizcdG#m30pKC z(EcqmFf7&ULhoCCHZtDh{K3C!dc6LDzj!GAMTM<;2+U|RSq*7^yo(|->dBL%F)=Z0 zgDIL^QzK0Y*9&@!ol+MU`ya|BDTbR`t9j!u)YA1trdB2@?qP@cbTZP?9^e)`aNt1u zbjrPZ_vlbm%Z_TiRL<}H+SaCZAJyiB*XlQmIioVSi^@+$RfKJN^mX*~M!&aXyI)zH z{Nc97BAlnz8;P&#Y@a)KOuLp>R=96;mGByt9wi!IOl)jkev)k9t}B3vKhfkUTEE82 zZC7wTrkDH3g9Y!uy1MEc7^skMJvlYC6Dx-S0E{B9~L_p1G%XkNq-i<0mq=G@m|t4<5I8Y-rnBl%b=oMHKE;kp|poSdM4$IT9R_I6H?;> z?%&@X%AwkJ^WmYHi`k|M0xrMx(>+`8?p2>2ANGrlHOL>-%rrD0nymTLaP*cv*hIU$ zyu84|6-nznm-5pVAy3Ha=sc;M>#p!NA~0R9Rr`6A<izO^l zZkn^Cc2%8Am$;YL3X6+_u#E#CXpFAbSC{sP4TyiaH{}Gzv^d{oBZsn-`&&#%NT?}M zfiBB=T(7J;AgZMflv1zF9KGYp{8&f4@rPyHc>MDIOy&i{7 za>e!Q*9oX+ulWZW*KUW)MW)G4?T+he^rlJL+sDk`e7A`>qWbgAG5=j+txTiUYvez~KWkp7cF#1bV1NQ7Hgc2adb@(eMV)H5)BJ_J zp;T5n?U>v`2C0tvT+!?s3r9?hXl1 z`vYjBBTG)SVXUN+;L)Q(8eZRLJhcrMMEoBFPXCNK{jXrE`9o+oTSa89?A;9#g{ zKD9Fo3)-4n$n?wH=mi7>C|Wm>;W9_QzM{F>@XXqFud5|+ksb2s6PR1#_rF$r_apxgOd$1f1cy{Ni4e$7J$7c*!5Qvl9(^8+n|byN);}ckPtw> znez^Fa&Fsy-1!5FZGv{W&lZ|^jaTPUv>MdbmJV`pDMkzd^o3r1W@;pi4rFg1uBP6*Uw+SIF-7<^C}1FW#hb-IX{2?%w~zD{H?;f;-{c5Eu)m1 zFhNwp+pLIikdG+00hf`ScedDGt}i0+uGb^gudOe!{z_xVgSs^z#JPibYdAdA@a5L_+cQ)R+gEy``=`Xs^aYpQ8dEK5Re!2Yt2? zy;Iqq_P zq7>rm6A>#0g4Lqh`;0cscDwW%e=oEHhE|W^5Eo~}{S_Hu%Hk&^wSlsiE0xjug|-_2 zdTQNlEG+{e<}xLz;E56t{c3%TWupKczmbkSu?nA+Uajo?=p=^(h zk3+M%O$d;l%e%ER4XY}Bwj2o^^rc<=T{3YY;ViY|@9(;q^Fv8F<_+->iOwk~JWfp2 z99dbq)FOLVcyhlN@KoO4BtW)$*FmOik zNA1;F_-(oX_-+Ed0}NHn&AoT9zkd@4$E&mj)i~+!&8knHK7If3BS+UNJ{!|l^ZJ-T z$hEaZw`_cUkvrGAiva~q7FUGr>R;x*@Dr>l8gmGO&EuzGJd|&RlV!J$RS<`!F>uvO*caXF20oqD#XlQ7(q-bn`rFABp z>tf%}pBuqaw;lwE!;@tpj0JJeWnswZ+0#{3RWH!J3HL%Vk~h|RX<|KMZf(DLc23UM z7587$Qy*FDHtLeSt*xab{Jfc=FKQs$Z0ziZuzS#$9>uakuxiBfBiO%6v~ZKo!&k4& z+jI^?0eQ$G72*M%!!@=y6P&TbDe|q zKhgW}u>6B_)Be7`m&d+keakVYfmG-Nh-w-fm!6(Zlts`e=4a2I5jskO{KGe{GgrTt zdtOqn$Fm^dvCk&e>1A=AqkqBf+3~JD4_RgE6|qH`C0%~s=eqd%ZEGt%p!-%z%6CIw zp3;XeNgHu+s$Hmd^ZQcLZU&V)OgSGaqnsZly<-F}as=rKd=l6}8Q=v9Uy>M7ev{-Mt_36Ja0VsSzR_q;=KowQRGxzJUQ>@Js`{ z?x?Y`vFr>IJhiRz^72sC4j(x}j-m`dRrcgbGQ7#*ngz&N0Q59y%aKyTXF>tJua!Pa zh0@fSZ~auHEh{%yR$rePwumjd`S)U{?ZXXm4a;2&XMt|y<>X!{gt9~8&KS!<2Ox|_ z$WD4~g}lPTyMczPAUE&ax9@45HC{t{b@EJ%w5Q&#>uH*fj>5%`zd}T=3X6%!LojVB zcCrdPsPAV)aItWlgcoXPkrfZrIxT^2(^#MILzvN{+ zk}9%S-1b2E@@Q(B$tw1C;mxT-qwSfQB5})er-6@h%G%oGq4pHJOk_+M9|3?I>CEFM z6hopwSY@LN{{TD)u{MHUh~?G~01-4V_2|!n$96@d1X(rdSO0Q8)mUmG=cfPNct*+y z*Je9eh&JA0j@KL+fE<8#*0q!Q(n}lbTT?z%~ zC%i3M$k|LYkO=~?YPLZ))a?O-KYGmpmMPm(T3TB6`uMM3DNaT}6hQ~i_gN$q!)7yvqAb?5%u0-R#7H z0)7I^#-2KLO0SNJn!5F7%tl(^2zvA>&DB_H5%|hJ59mAzr(R1QYs+y+Nl7Vm?Irp@ zFZAU5ozhAEpxl=CpuO~%#Yj3S!r7>um&YYmp$cbzaY3gG9wx`u+R4nEmSd>$@k7b@ zNOwsA916?)C0yRzdPe71PuZpZiJtG@^*yGl_WVSjKbx)^`*P-eiRWq5?%erp*id`m zn1_e|7#d=K_wL>9ycf8s@#wQxurm$;^r}4r=$&ekXRR3>|6Y_6zj~T+*kzS>Zbl0? zZuj7&3a_&2-ji4eko@?vzZ3wLszD0Sm4;<1x6;&@6(vg8)oIF**zEe=Za@uqY+Vi= zM!>3##~5_-xQ&eKw5)mA=&(ML6lvZl#+RVLa(0!1#O=w?&(C&_6Sq6vg4#;*1V*ib zOC}QW_}MdFujOAH!D}HQcJf}VcgZbF7jAfufy<)^gpb>B96o%e-}TL#f3%F7GAZ?c zi5b-63&E|wl}A0>5F2nBYHCp}?P`aRsA|PJSI1ZJX&P=(AU5{aL&e?pJt@K@}_5Y-UUMK&sc5Nqp%hX@rR4MAG{4`!{c0g~`Vqv2zsU+qI4do~EQvUa|#pr8OUYtxcJ?tXLY&fJ{~ z6TJnwv7^j}g&@2V4`b*4 z{plhT==HJ?EX?K_;$)1>%c3)02{p(J4GmpsOVjOO878*g`sxzXN20`DoGlCP)*cRX zGeZ4vtr><$8|oqLJf@RANa%4`oeIxjQJQ;tO%Et^lxw)L+Tm5QOUXV!!pt+&<}oVQ zhvZu`-I{4+q$cyUcfrH&Q$_jN-d-!iXBz*iHpwGb{)xv8p1XcM?fU5qT-?@e+tleRXxN-6&{!$jP=F{=DfDu;^=xcw2q_7>nd*(as# z02)qr4G9gEL;i)$6-G_C^_w$>pDZt^sFvX5G2-AS{}Lb-hVS zdtzNh_R~dWPW6AZq9&_729Ka|xhqpG!T$-r@f{s(+MDpx)JScPUY|Tr091VM-X-xI zK5;@HGPp{F_lJO;L7k661%j8~Z^HsDbQpdRUrK(kkXKYx3qcv>{0W4K(nksw;Q|G3 za0Yz*8rA;o_DO8n$nbWAhvYswz9f+T#xX|~VTRK)5_@i0^pqAMuF@Li4Y4S@bb`WG zIa*jjZACtq=?FK#sfgsv;DVhF?MHV)rvz>e8r9bRmJw?Inm>4sQGX#K>kdiK2&oNC z&54h!-Z9oj=GU&hK6>t$goN(m;^MhlegT0dCh>4iHa1xXM)j;Ka#g3^oGzv2MOG^; zkqY%EcK9>WNr`)j!(Ri>51-OpSnro(x|=rBf8&8Em?G zZZP;1!$D%LYE_SQ~`AO+3C$#7C!?|ILftOV-?#odlk4+GpyhZdHa5G z+mq8p8FvMt3n;eHl&z5QpYb8xf&AWYL-rHJNPSj{&!CtMewOKy_n$SFSl5}AXEt8* zBFpQBGt>WzuwwtVzJ3P;>F`7%hy-|cQU5F;7+#Xik}z*TYI45xkBlT6o#}1Lx4r}< zc`9Tcx(TDc;SB|7B)~F9EKvf!Wt(mP%Sr|&ftzW&4;tcSFp&#AXdK311tGcM zjfeR88~f%5@{lb$6M^1c@d9KxXxba(9>6{gxt$kT`bR#0`I2SVfA*%U>#y%xN$Qqq z!*6eIqG*kc{Q2`IKYSAAGrnAdu?=v2B1x0iNJ-GH=6a)czw2)k&mwie5 z@ivDMYX}=3hQ4^KqC!S`ebK~YvT_SlTY2EKsiomK6VMA(4FSdD_Ms&uB|OIf6Hx(} z3RKUXJBg-)^4*q}Yx(u%kBN#u2;W&|%fVd?qR6KoB#@Rqdn zbf<_^O{6~ff$WntxqPOFKR_eMd)9uVr!)|%6TyCt8kL^_Vg2jQMhTKbNS*K zBpd|+#7b^BA0`~4MyX>qgnCQt#m{k}0$;$hfCAUj7EW))^84O5J>7}gPP=z6r$Q}gNarGwV3?x<`&Ju~wj*TrVJzP?T| zH-Wdp{+9wicf7HJzWxH_J0ur8o9rYyI=}n;bvdVcMC{nvoBz0dFR+Pw$51!eYNyj! z?f$mQiVoYzXPUu}#O6sKmc_Ey2%Z{0dYN1oe&>v+*XJ=(`O5F(j-)mf^S$sso=7^q zW;=HH@FsA>mZ$~qKP#-T({p;dzJ0rlW=?F)kfBXvpohU|!Ai~0CbJxd&&Q5yv5Ik- znVI>)jDA;L&CZP{VJ=xQ18?ET1Xx7?!!bR^7JnBKuQJtMqgPni$aOqUbQ)>gfU9Uu zjlse&K};hg7hd%tR0zAhQC#>nC|TUaV!{y2i82-v;Tl8Da)L3=b3UqG5a|=hFq}%- zf-iOQuF>D|@yY3V`s4`_oq;6oUtb!9G(?tlmuSrs7Z}5Uky-Hmt^tb^rcg0Bp08vSpEaKn=70^JjSU3okg!5cp~l${PJ(=B0wa3cpgE zjJGtb;G8d!XoB0X7$s2ut@MZO`A-f>LX^Oj(Rk25SEJc)TXB7Tf-KY_SMRyei)?Ic z1J`1gJIg(f8&-H7Q^0CvCK_KM6`BYLa@98#jWopnk~LmEU>nEMT54fhgOs%c5=TU0 zD-tbDrf*zOhLuE%J~y6BM%u)JTJqv>fB)gGKcl@V2hJ%eoq>h}!8#DTr_jzHrHQomO{&F;1BAnm?=+)59Tg^i0dYWrXSHwXKG8TwYk zpw>`1RO#AgFO)#IMO?fI*E_NaGl|U-yG1Ye({s@;Q9>NO8yWDT55WWo2RbPKA(9lZ z5+}c^%De!QvvqQk!w!1`5}S}`xw<%ooI}XZaqf$+ebG9Vn$dc*u2fMbYNrxiU0*@Zc@ZWd>%$D|m-zd&ClL!!20O0hC%p++pZGn>8pdM4$GOG(fd?Jw4JS zPka09QyZ=sc`Y3w1pk{WNGw_^|3Tz(_a7<6u%oxDgZC*l!}rLLvgLq7R!&`^__+9+_+xmOV{AExG}^Ar7q2 z=8nrG-_F3O`3Dzrl@| zXj++%Ky1s+?@1b&ZRN@W*W(#C~ZO zehggSvD> z`$0v8=w-#D5`&OTU^-kz=X6sYU^v|ev4&B?o{bBtEeIuJo^|zemYa_hzK#kY`5SSuz_TUrswKp z8cNkDdMJ7#D4Q*O<^lN`>z-2G>2hp1i|*pQQ)arCE+qg&h6v9|pV{Wwpb9f7BI#m^ zy4=LJU!#Kqrr#g$`DlEKd17$>%vncHs&<8Cu5%_*g3rLE6TrsZ2Af*`B4C+Yn2{EI z4hc3y*noQrtF#BBni@)Y3^NJhenhF*z=;=po|&Z9O{{3JoVrEusYVv{Je8ZW&6IpL zdEg?CUCL7F?F~Z8c6ok0RhvcSV-!Mkrs0A|q!;@()wmmc>>x5D5MglpgY=d!$kY?A z#=!wv0gDvZw#C{+mc(1}PlAsqHLtF%iJYf&ZHI2tAJ2x@OjkqiK`GG5FH+= zQUuR!l#!+`OXLTrf`3h%Pp(V!x&c$hn~$B|u*hC6Kl?JS%k-4J;!QWVcr=@kzF@Qt z#AHrqaulV?n8{S2IDwHi^_CWoF1 zXUc-{UV!5XG4?q3I1nuyzeX!{LK{3EBYqpr@&M905MGJf2Ygm-a+H_%{Is-$#9ov@ z0;crz9^|u>Y+&kq1?!F#5ns6r7cOXZ0+t-~ZTwcj_;h(d?Tpqr*Y=WOmcH>?<7Fa3 zYM9!#;ZfRXcnXv~x9|;QJv}`k3s9mcf{#y+it||XOOa*w)yBB7W`)ITD5963*9au z+oGX3y+<9|fQa1nnWdGLy5?EeJ1cN`1OQ6Yj4D|0B$h?=f-?doT{#aP+yIR>Jh3(X zQhVR@C=wb8eZ^Q$$dpil7KtIGWMM0Dva++G6J1Ddx*Qe|w0_~_G(y7u@IiLh?iYsT zycm3NQ@yt7xgB}($jHd)o%-5<)Aw1VB-=NUZWXK)fE0+^4{W{l=V^-wa+zt?E!yHu zF5%lfp+OLxd89dMJ8&Bj_k&DF8Odw#4v5uuqDKmuSF2+tmGxv0 z^Rt`17LxO%)>Ly_SJa}&&F$L`+5*4>^dOXfh`fCQwS8~+nHm+*{<(!No-$DMqT-M%^oJv8Ks)#M=^LIL~>;X1v`qM@H-@=2(uFO{Cie z!<0uVosjIShV&N*$Zo#UxBUo~B;oSF-9C~UGSOSX3UBT`2w&4HRaSU>EzhmKzkB;= zY7V7-a_GxGg-*1GEcb;QK3y-QZ}5ILq4azxgqR>kTgpgZ!Vr*C`s(s0C37)Z)_K9U zM;rC~6J^Gxo}TyM9M{@53GQGy5kQ3eJbz=`O#0Ba?45D@@b2A4bk(r$LS=vmKzC*| z$7?6t%gM7co05{_{;fG3wfWE3#|FX3F5_AxVg!9AIasfFKu19wg%AdWN|2Xj|$7q`+T5Rw$}rvo=v(!}NqxmwriG@CO$*3zS}Yuzq|_ zwx==SsHrMCZ!T}Be#jc9{pE`m|224i%9@*l7lRiyIASj~*;d>Zd7#$v8nzk!hCR%&VyJz@;j&DtnUM}Wvy8S-mM z_}@SNFC;7l65~ViZeQ_;4WlD>O_N{dCKjI8Z%CLS3oy|O;U4$QU8CE%^Tk}u!t|gL z6?h#iFcC%1;$}lGgq?%bXaMpfL==!>O^Wgml%LUx z(9gmuzx=td@(fbbUr)Ts&CS)9sGl|dni|R3w?^d%F2$t*^A=^z5IZ82`Y@_lx?Q;Dyp~JXJ@Y^QGIbIwH?c~XnJRH+leT59SFBQ zCU@i1B|o`uZrlFR&>|;S)@S7&s8Agw)FF6WXVLYrobTeFx@p9oSbIHYR5A)O@iamz zX3(m0*e_@5=1{;-)Qtmx_cY}Zh`SbazC$B5#vPI#Yf`y z3QZb7isI;3J-IY@&Z~FkSir}iU2;qRa1x(?IEkq+;5|aJN7yW}9K+Lk9ueVWhUNRP}AeNW)9TX}1~D z=-_{w2!Mw2Q6n8EWWFQ+Mt3ynuUy<;A`#g@xP&JdFmuvCpOS69Ct+pxl)hQ+Kcb z&4C9C_MNPI{78v0N5@R0xMC34WV`)z$r!u}GY_|U)_iEB-x8)RDl~dNIZ}6Lm6fT~ z)zwYc-a!GQdhL!NPoc{M3-JMBH<;tL7Gl1nd~N8Ejz71jepw#*`N|Z^NTfAdf0^6E;iq`Hyhk>_%HS!O*Gky|a*T z;O0@Frx5R5TNC_>epu9G!dqoYr1g}IC!D}1f@VQHo}kl3>%P_?q#HMGWJ?u`j*2p# z?5iy5+;IkL-UNyO)RAz{h$2BHFG*+sJn0y*TWSnh4c)Wmt~O?bIZ+>A{A;Vqy}e$B zYf`7roFQx^1-#DQq54>2LINVmR)_+=n7@GfLfDmv=hW76-XGRIS#IP@E1_pBRPlhQ z&bpn*s+ykKCgDjh;%diS{;SCy3$zY2n>8_&+jLL(4qBfN9a z37wFk-@hNA=Jo^GEJ93aL~0gh;XarML}s_JQ|o;7KE!A~#LL~AL?Z2acQHMx_8HVN z;U1Y!VMs~<%o9P8Li@p>N~b;S(bc;#-zGLX!6q&)K2f2=qSoup>M?n{>7mdOn$3sT zL@XM~K-cTs;+{MS06-jc>nm9n|MIQ>q>7ViWJDwcJa?n_|Evav|EvZzV!lcAlmk#< zwnF779UIwY06fV$@5RZ32s5)OkYLY)CPG9 z_~IUm{J7Xx+7hToO(nlqkxM-AGz{aF;R%ARGkEaB$)yY3Sn-(q*`%O{DHAWm%=Oa? zAV@&TJlnQ{?S=%=xoj0;Q1{obsMPUK+fbx~L)S4J6uA)WDvv}N=5vlH*)%6ANawQu zA>`7C2vl^9p7_oR#QJplB@}ym%d;8^I)YW5e}CuP&b|VgOGF6Ew2l*xxNl8@L6}@< z3px3qTiHZOsrZiGf+<6_qVu&W3&v@YNc>?+Q0mE(C!ZcvmYz^eHfnzw9eoA*pha(a znc*O#NvW~5bugk&S0ED-^q5G3j5Ho37CJ$0$PaFOivrUGaD0OFbb4m-=aWO`Eeos* z20sKn7Hw-*z>j_+c%eIiK$Yn*DMr15rKV<)?t+(82vpYI$nux`F@M{Pw*Q%;-+K|^ znnncu6!dIu`LQ{&d+z%CUzzAB3$c!Yl8*%zJIE^OcrHSeXk1-OFu}@Z297}a(S3B~ z{6-VZ2c@>p!{^DSe@V>z!wN4W@s6zMK|~}+L=@5km)A1E3(#wiSVptwH2x6~5vc;v zGyPZzhb9<4+VHjR(a>T9oH$)=#$F{2Q_Q`UCEH3v({MLpH4f4zF-;J1^~Q|{P^ktq z?oHo)`t&IjoHeD(m!I>bA6?+9UK}4zeAwie;q2nlUYNf8$Nm3gf(RiLuw4_fG1X7aNbF>4ImUtS0Eh=`3$ z>0=T!4H(eTk`V*J)a@4tp&M;S{+5%lB`XJ0j2>cHSWsou0Vqva8JVR0y_@I_{&O5k z!q?o5u@x2GZ57(-dVWZ6)eOUnxdUyXK08jzo!!;O%q(d^@fw-wZ4+`JGHOB#P$=hKH6O`pqBKUyvgxi@jXSR=T?#5`vxgHtsb!Ko)if|{)4R@?VT*#=j=H|NA zo)HWcQL!0>k2?Bj#zW}A@87*U-EU5as)){HwAoNSc<>-w*i-ABU_LSqJi6Jk5y)0+ z2U~l_r7RpHGv%iKMfB8=1+H*Kll^jj0$wZCCwe=fpFmG%D!9n{RTw&CSGoMtJJK&} z&fj3&j!}15NTrp7oP!7#xG%m%Mux~Wk(??Tpdom~RlCa#O4cw=Tcr)74!E+?(V1M3 zKx9bxtiw^7lr1?&*x7%=t|~ILQBnil4`%KsHFam5H(f(GqWjJjLh68_5z!ZY;q_UWQn(( zqr%RITrl!H`<3oiI58u8YH2zI+17zIWS!=`m;}fGg^2StAQtAXc_y>i4!nI}_bbh(6R@R^T>4wH!sn(C0=I(Hq|b5b+<>iv`r{IRBdE#7*0BiWJTy zPX4#wR9)LY;?{`FbJMXae_E*?$iegf#GY;4Ti*R#niqr4NbH0qUd+(%=yT{_dCOR_ z2I$L}YyTDv$8KfbT0swO()ddJQpJ_~0x!4G&_t%~NTZo6=gDlyA*AS}y>8FS05SeD zTQNbHO@FW_DTPnWk(FTk)6z$dBwe^KvOT+0a!2ng)W%}eM(jQ8G=(12<1<@dxJWD; zb`5HJ#qsrQK^#0g7x~p;x3cQujNTF#J{YCnVd4US!t)?s@a-6$h>#X6t>4 z9pCe;_8>3f7!k)QGqJ>whg8P@x(9Re^Q-@*iz_RCK(V!3Uv=F~L7^y}3Tq3I`GN z1GKx41y_kE8nVs;8$P8=$d&a}6yOwrG=mbYzseshlEGw8IXQ$@2rWq=nsy_qYD24A z&eN9geX2?g)CBBK8myu`AN?6z2gBl-JY`rC6zMT)-z+89F*6YiVS;4PhAnKyU73#v z6UP_sSz+?%z0+NPK(uG;6MiwL|I&Y^u%@3n;mSY;^O5c8hZH;&b0zE^2<(c@=-;M| zjN<_BVM)Iv8Coi2E-y;sT!|8%alFJ?Ma7^~gH4wRY@-Tn;}~m7wItAnBuVywCxR(s zojQ@XZrwUzRK#1+uxG-Dpk^!?Z*Fu;`rHYaGrmu9Bg3$t%>5?jvgFLs%?C)gRYnoa zxbCNZMyvC$yW)CZv%2I z3GUa2kET`X={4GvaQFI^MiiF0=v#Xj82lmUsZA7%s9cq}Ir-esj5xXlXGAoX{8k18 z3WcCwGb7OikkXQ>tr~(D2{lnMGwm_a?+!e8Hl(e<#;CBQWVkR`u8Kxxl?>;?D0D)5 zAm%~!Q*#hUC#DDn=y&bf$jTb0pJD^_kbN+KsixjvW)Y{_V~6ndCUD1{@}^xOlV5i5S~}C$3&vCOJ2A zT~l*67(XC8ZM0V=Q0G^xBd-YG6IUN1jD11HDy?rNJvTwaDBVURatV_{K@Sne>h*_V zAU%_wOcV4;BsP;zq%4-n`4@ttEk1&IL?{DikXU_Y=Zx<$LeOk)?oUItq0v#U_Kg4iw$Bj%c4~Sda2-Q%(7acJR zD{hS`YCFylL5LhKA_D1?2mlYER?y%kc+7ke%btZYd5DLHhgC1EpyzIRCkCx=-@cta zGXw42AxRq2FJ@H~Ln086MaFXEVm<1z8IAb#^z>=OI*?+`m>*a`k3YKO>h!`wLzAyJ8R6y+tij-&qm<rl_ajz#>l9!J*AdpmWdB zw;3VQ0P|VdO5juz;&=+PQ+k;612VEc|1<^jAe3N_!SLRV|2j|L7!S`I3?w4F-(T{b#>c?u71AYc%!bKcqTO* zai5xx1BnR)C`KZPQ#IU(5<5HEs+4~PEO%4x`EMGmU*YH&BQPA)xd$%~Jij+r@f7>q z3=bJ%@Q}@^uzvY*NPto^RE@hBYv8Q?gDg4ny7Kvw?U>;-OTCcHje!pY9{?M?ykTP($B7FP-oKz`wmq(kJ7j2}Lze zm{M{8EA(@bMPE8QnTQ!^p0$#R^7klF)Vwaxu(DT_CK6Pomo~#73Q3g0NDMIo2B(^u zO9;dJGOLrD0I*H(vkyftBxS+{4vCG!#&<|B$L*?FjzF%-7{_`9O%rp}49~Za8F@^R zAzo&PC*#Dt|Y8g$rhhmpCI^5?*lN_dJN)XCEu>BJ;Izb%KghLCi z&%kQj2{K<7EenG8ys&OEP{`RGP%T=9_! zsJ~4w&TL6nO@D}_{vIQb+t{^NRoKKFjWAxMGc7GC$%HEU9+CkSR|75K+M)Tw-^Qp8 z0;o8q>fLpYIu1;|Ob^sxVl8(};RjGsW8pP&%Z2weGH-!=L-G-Ig()Esr)TN`TZ<5n ziSuNj(Gy9RdB^cFbcf40HY(YmM3nGsS3GdD-ehMBI)`p3jUDqGuHT=gleLi$K8Oqu z5VcjzfR%}f4^o-?aznnY$R?>9V*_T_IAb8>BsN5DMvEgb1~C=HFg7+Yz;-DsA4EnO8f$k*Ts>K~auOEeI) z9-D{OA>yBPbLxujG9fCXOVQ$8JBF}>`LJ+6OP*+Xm`ZIDZ~AD->wMt za9?Xz@Ozw>m&bEi6w@#W8p&P0d^vxKMcN|(Q1A+@2(m|3CdFgPMv)ud!h^=yCzov} zZPf;F+8Qh>915k-OoQWRV251+Vh>`;ro|~xfDX^`#>p36Y(ywBATcrVdVzLD#+F-{ z05(RJr(+1Cf`4hi#92Yz@0$v2zQI!U@brX-c<;~Jgf|lCDe)edWjcksX2!jb20DXP zIjV8Z#(9_VPCw9RD}()R_)82YeqqucoSRA_TCqP~P^lT&2x8yQj&*2Eop}h@Of(R36^q+x;Cew&sYMc>B ze8n%c^-GgLO5E6>zxP5=m$RFOWq=gtZbmdfUmqQ)t&z zm>DHow~6z_iC}K6YsBGfxLB<9T1EFYIH7sh{LonwpaLUo>K_=m4Z$y>q+Uh=2)Pvi zujjdGO$f4V;XS>cIBbVFYk-Z8YGJIyhv;0`t*A4PRGtvSc1_ydQws|piJA0m_Cb&) zz_kbm5glbavK@Y*p>)Iq6W*8ABkzjB39Fdm&#G4tCVLUKm}n_T|u$q7ZD*23?QUkB&K#CB6(CR zjS+}~7CnlXY6CMsX?DyHmBUY4H7a^xT1RAkv7ZJF_8@`-pmr9Wxtx;DV*&WlO*dC& zSBORhF`DTQ3m&LH#F;iXjCc8BLbay;vCP^m9dSAvXz+6^?l(C(?@_Y|+Zc{59l{T_ zoo(=2pqi&vUtDqI#M~y4SC+8B=-G4dnKy6VoWu?{wzA^hzLA)HzkrNR8L93(o;)#S zL8O&K$`K+!uw)D2KLC;v$1Fnk!|5=_xCrQJ(P=!u(-7kI@&z@Wv6gS+Gb=S?zB(9y zBLN2z{A+Y>wt_gjDZ+ZP2g7~DU^V7GPlkB?AMV~f9P52;8`r8;t9jHUnvkJV$(W%@ zMXAgo6iMcp5L!*9q>L3(hRl>XLW7|QB}0h2=uR0k&%EcQwfD34`~HsOJ&xa>Z^!ZM z_3XV$_x=5TKG$%b=XGAq1#>|={ef6=2!Je^kD!ux4I4CQb}sgPUr58A(B8G+`G>H0 zKi{)RFn2LJAG=WHQ}Pca))?0$+&R|k0^}22;&p_qa9j!XiAFpvgevqxmlKRPsi~>a z=n4R9xd-WD*6T*iSiayx(#RHUAR@r&KY9A}4zmHV77DsM2Ax|MI1Y7rQ`<2(G#@i` zm{0DhR?{I%HsW~@;?IC(fk6t)IcfB1+OG3+)}z{zbX?Gv0O2`X+%V6333mlb$wk?A zou05ie1We8mXg>4yY8lK>SAwknxP+5S@9%cdtdhXSJ&o3jlYY^cZg1BqWDu^tl^2I zgb%wE-v$GIZ%QBt*_1cjJU?*&Rv%}f6eKM^=LE9GOMsm3L4*HGvV86wCwUCf#3!Df zdIQ*+^_T!qd~RiErVy)qD%UZ+XpWHyBaa=0z+oWS?;(IvPB^s`CzqwE3)GPX29ak% z*;zhs^;RQte1lJhE(y^3Gyw;L_VpoY4`${Ngsc(|}@=933xv+dY$w*9a7|gg_#3WYyMFml53Cw31U(t!p@a-^sGw zqkP&YOE#6oG@*w$+NLG)%dTl3DQwr;rXKZZf+C$Ro`d#)03^b2Z9a@dl}*lJJKfTE(G9zF-}R)~Sf z1xaJLN+~Lj`pF;V(!@aRFkG!z<#+JQAqxGmaTY!gx9kb@bCVNe`un8s(?l?Xq01T0puPBH^HDpbhyZ zxQ>UYVLaRJLsLxlf0!zhtAxNfsc&2tcri*I!w}B+DwH=Cu+gjg_7$+*Th4KNUM-cA z>rsnbhxiMF|K_re@InGN=n0v6+g6(725KS$vT;Q%E}S{|h1rL1t-IqqNcLv&ZZ6Q< z*9BrG!xI+v+bcS&DGV4_&T+n*RO_L3*P|2J59F2{WYH~^ z3Ma!77B?(3n%}`d0|Etz(6;1khYeTN>eDQHtvcnVTHV&W>;;|r3l2SRRYT^_o%<5m zB_-EQ*Zp13eqew4{8?I1{OEmonFHG!54<62x5dnkes?$-)~yD!4$5F3E*e1Yi*%K_ zbuYKI$|=+PBIdWy`$a%1?AZ_)41GL3RW#`9Fxh#>8TF5CVZ0ox48p|UUS9KOFXq^q zW4{8_CBo|&&?qBS*`HtM)$f1=gN6eDg>7-Vs08Ma33xHJEy8C357FaH0rseU$Q(BG&YP$+>T9BR)>eqzDMWDm8>+lS(PO3)fTL2;Q7 zAnyMJARL`GjWTXPBB4*9hy!)?BOn=X6Jjq;Obwj3XnfJ|!VkfQWcgxzTcXL&5U>#Q*-jK*7)|9V9*c;90(u2Te6}4`(;121u~y>Zj@gUW-5rSR41q9p9SUC( zeZ8ovs^ZaaBTHn2sv@N0rSOdd0v7VOZ5^^QAE`v%8_hsortm>OVg9yKHtMN>=9#Ct z%Cv9Yup8o_(U?;>`#J2vODX&RlZ-FA%V4Og*L&vUBBxCW35njQ z$)PgT)E4M|(}0*9W+I}~u!p#s05X{{**XH{Z&%3!^NH4a4lzIt-!Wfm&tZ$(Q3%S4 z)QB<-G+%&h4EDfFdtkvj(IxP#yo{=TW7#_)V4t6!I|!~6(7uUs=6g&6e4x+3nCp;{ z`q%f*tPMyFhSR-!xM}mR!I$j7arU8I^!}9n`DI?c1)v|Q%)$K4xjK3UJ?t0M0&sc@ zY;x#pe_;Oj)%)6S#sBV1IsT(F{Y_7w!a-gwZWHs8siC3KjG34sBq^Julg$Y3JN1gb zpt_}|<)!WSOw24|;AaVS*EMrjfwjh;;~_08j_|jvCj7Z(oDGNrZr04jPvyQ)m%>Fy z2v1@S*O4~`1+y5=L+sj+Y=GAa2Ob*(BR5PMoz>S5Xz3jQn~fVTOmX%{ZE{$-$44;U z#TX&9ME3|tSGuBMEN>Js`P^79^km{;-?iQg!Ce&^i$l;t+*&7g1>rm#uMc&W(rGeH z#8Gso(2eX0b0^KD^jai%%$kxL03`iiFtJUT0APS(yebkX;3T!6>%I2unL|=~i}Qzk zS62y0vA}S#A(Kbg_C|yDrhbpiYtdlSFir21s3IV(8#*9%!>B&JOv?Wx%CdYxRWO3V z4#>kX?m8FKO!(H>3iUOZ-UEQ%OpVIgb6xoE02;ocwISc4{-kJB^rDu{v#2`*6kym` zL%@(hIr&49IZ1ANWh9a#=Hj=L$0CNORv_TZAYh&`FqpI4@A-szByQ+B5)M=PLrr?2 zu?DjlAQvd3mOqg6LcOqBIT{r|rZul3d+0(rjuSecOcgNY#f+J1gAWfx@>i){M2xYp zu~GI(RB}z4PG_m=1GoQp??>GY$cy+Xwp5=^8H+3Q^TF#HIf&< zVMu7i_TNaV4A-?KBaMIk?p!y8hrl!FC2E`=Na9fQzQptZSYL8c0i8L~lz5l(+6&Z% zAE&-2s89)X7Wf;97GkIwK zQN?~iw~6!nqN%Criz9S-3efY)Vq!`fA%vTTHwsBE1L_Vy@e4#A{I|n+GarDv&2dnZ#9*=wR*~dd?pQ}|R`OS*}We0bPwI?1( z?8Af==FpI1PQpxv`hf*Ec=i*h^Ww#egua8ol2Ms_K@3K|566e|{Ntz1+l~fmwovtB z#PelS69>|&3`C+Xsavc-}RhW3}_}GBKXTxm%Ox2f3)Dd5i9-zXM z0cC-a&Xag-Kr`BMeRcL>bo~l7awyoKKRAr$1pYtRys#MEnR1-+TpSiY3;rI~6;P;7 z+mmu|VCl$T7}R`)$y5A-?c@S*@2^Y2~PLOdyj^2Ae5_X|Isw`fs$_jzXQl^q#e)8{5UyliVe+g=0- z`c*{H^K-7#D&iAx{-Hd?%=2vM*-)asf(b4Xf*YKPTtou?k?{VjK~%x)cKVZawD`&U3=4%z=cRIcyg*aa_|O+lJ+zyId{{7>ay|Is5Hk%l^V82mZG zYd}|N^xAV~JIe-$7GO&78o-+ZTAkg-H4D(SU0WYoRg476xS& zG_9pB{Ivd}`~1?SANG7wTdr^xadg|yXavnbtWx>?t7PfFi$vBzKmYHwD*y2$+}-C%rH-4i{BuNk1wmiH17I#-&b88+85xJr=fFeR z8Qz}cH-*T7*@wM|h-52_GUF>+7IGtFVq{aCj5mfyMwo+Y{~x!$#QxthA--9pPJ$`t zg>j%e=SiOg0Xw#~91YcnLDgA6gczY-18if##%oW5j~*2Nm9=?O_~%MMDh~|Yc}_gZ zpZ*K?@cB3S`R6CJK-7n+qqnanEd_nWm~l{|dU&g7S4GImH6rD|p6c)a zGH?0+yq+_dRoMr%G-nyin!r|*JM~S~6X-O6;HIyeL%SsY_g}0l|6_|pr3rp7AkVa| z&d82{iFY_Uqe!9}UV$ua3Q;pGL9+)X1_s~p!pMLUdeQ28I)D8QD3i=)V1`SmCXy9d z^I~3XB~JO@9&g7cIyP*aMOhSO-Uj4(yniYINKlGC;Wb>R^MmiUFI&Fcvg@-LT|0Qj zpv47<%0_EtfO})F&YIlvcMLPs_mDK5H8Prqzk&?0(E;Mz97_JU1QT>w4t?BRjC>!! zS^#QaqQ*y8gim=rC}lvurFZF5|DE9V~zzR<*V4f^QW~ z^`I5N1N$2sD}@yh##ZblGqRUVQrW zX{*G^p_gMWThF)RZziKBkRxDh78@8Ckb4!NI2z8eN^HXGrfv)w7BApxR1_CFDGCKd zPSn?n*S5`Wx;2OD1;mA~_*uAPmjNx%SB8A(8)S!=a-kuqD{hi zk%u8?TqE#A))7WZ!bB8Wc-g1{58_9n_`ikA!z@`nh@K6?kJ%)OqLQXr>_Da=mh*tu3x=66Z74-pb_H9p3-y2-AQO4M7@ZCu-DbqLhad- z0!dqXTf}dr1U-0gi`WG;KX_FKz=@!Gqz5nu%b4IhM`au;?fH%AiU4hi@A_SKZpLXr z*^GlMgAbqp)*EB9-?AQDu@b0R?c9+=H8xICGgx){K!Ed5Pf{o<9NYGPv2h{IyL(KME?^^v{ZQsbYcprmK)S~E6!38};E&5Db`gc(I zx$mEXMB(o$7#Ysk+e>rtE+R+&+ISFsF=wX{j!X^d?{$G#kmD?ZNS!^)28eCb0$ouf z?u5|+vANzOf;s@TNzy>pTeF=P<8a72&{#uE2q4+F^^dPPKEVH@@4OOT*952HMy5_o zLWfDs9QJ9wd1%)vqBYiu*Fy+6xk)rQfBE`#+G!X9I;+nQP=Ei+7X18J!=D>a-x6a| zfEpX_i}&~08jdxz`(mI1oXtjrA1(lrX&jsHCF3`EBGP%G`SXAr>T3NZup%_DzHPsWqi`<7(U6#bC&FapA+!y=l0b-%G|L^AN}z-mk!bnRXrgKZp^{*d8FqMQQ-_ z(Ryj!Ibhg7z~}d@xq0${6R@DumwMlBp;vn-Rs^~}QY2yC{PZbI8YODN92_6d1OJvT zM*G%1TZ;Xc?1#9{3&K?h?=i=U8*$@rhN%6wny14MPGI=}AMa~?cBHA3-NN9PE=ahU zL(|(Jfrkt~aVC>% z;w1dxPUT&C1}McbV;~p|A4t$2pIz_mp2+QYUL!szG+V299B(ZBmMugYGLR!3B--2- zm(vUIfG3`we$Ha-T`Ro1QxBG*Yc;QdXA_E!Cpcw%HTECaCx>x2tLdt`S<*8OC>DuX{? z)4MP?Pu1XuWn8>#eU5!HbVG^gd#i_?1FG(TwX^zLACRb_X#i4pcW86gu6q}xhB?Zu)RX_3r2T90>Ej)v_mFZv zKtrC}Mwa=~<`wnd&`SccMpO{>^-yKLsd&sBA&UDbeiA*-fV{$3hCG_|A-+Y@*C?P zEeEDpq3~rd22h4OQY+GqUn%XXPBT=>y|Oz^ew>kR98@klIpSj(<{7ZJ`C&u%D+KDoBDCcJMj6NqT zdlGexCILwe2SFw_Dv4V9gfaV{LesWonAX&69r~zxSX*1uaDLUPvjQ8VxGQQIjzU)W#~yVJ4Zge^ zrR`2ni@Nku_o!kXI(gvb%6kH)GfePymV+{J>~?i>a?0)gxOVO7RfwR#>p~;)B2cwq zF^Qsq)}NADNe{;Z=}rO2y={6q=CK-zdl~rbYY(6%!LK_#xeVjyzyJTGtcrjaeWQTZ z%mZch&aH;0@4pPX(PzY{E!>zW*;s1J8VF|wc$35nllm!!f12q)-SJKEFLVcCSk@Yvp&h18Yg!GabBKMRl=4#*MIaNGA2@>fINX&>9CI zo^svUf`vszYeUq6VQN8kgJI!fD_S^e}A=-^`yB-DLz>2A<|MszKn zZDY`$(oB-92HA|k=@6~Ll{s(_1_GH&yhI>`nxIEGE@qK zx43&JwrXY^vwQ#rDF&ubRZ2sVLWDtedp0NJk6?ZeSy()glIPy6=NfVB%$W@k0SzSa zE5UCcts0Re2a_+6trJ>)^n>Rt8n3Hdx+DQr&Q&a9U48n{zGoSu_#26>-AFcE4@e_M z=2xjr(UFm)Wx-hZ)8<_?1`IJO85SfD{$ZZXi;WLHIAE(K@{E5Q(VOS?gim^>My${CMNU^;YCuLWs28LR?6s4`J~KO%)Zxh7gjtoF%&&JfE;^~ zAVM2h{c-Pp=rL?j;*#Gw%79&2H^E=ALlATb15`~8dU|>^%0@!w?g^tKT4G|~S%|5l zRifAwX%?x<%)Qd9S^nnDi`yFyke?K9p4%TRfTSs@4CJ+9Ya{(C(*GYF*`{aYd#ize ze^ysqOsoe)T4>GKro6tVp!3CG`hi`%sUR{yb+I9YVY8bD z5i|&~7=?B>NqGT6h#E5*(syXIi5P)OY6tW=2}2OF&`E};@M84Jus|ox43VJdAGf__ zg9$sRz%w|tp6$V6*4(s42#;h5_EI5lw18aq73P16@db+N zD^KgfOAj<&#M#$3$etc*<)sj_fQ}&Fa2nw;amCy+9>h%C>C4ZaJ=@2S7QuNJZFu+w zx#)F8#bRpb@pN%3aRJ$gfWx&SZn|kqQ0e@a7daDJh+DKBfD~fjH%Q_{#0aj)CRAI# zuBiUNNzOsV5|LGoC`$BlKvYA0Jxb!m)SiQ=LKW|W=AJrhTo|$n;cV!JL<_;=<0l_u zUlgq1RgD*zK?t~y_V7ZuibWc;(E>q64o{(3M5_XeVB75>zaIOz$hi<+_+jxUz1h&};Jg*;G#>(RIKTYDo=v3Qc{KB$Xu9yPzh1xi8d3&ba<`ogA; zxJU!v!BW*#Sj9FM$Uf+e7K9?$67r*t<#aHI6a}A+tcpC_Vdy%SHf>o}I-*c|k2MM) z%h%v19>BXr&3XM_D#W>M5X1h!WCMK7{y^)Nn?HmH_5|YM1ld%31LM~PK`QHF)}<{X(E9Onr6=6z+FFt3}cC2IiUhp zFkWE0AEE>x_eadJ5@CjWEysNr&SSKz3W?0OS#K8tDFWn66gm1~SdhYw=3_5ZfG{zi z^`&!|_>fPi#yM|`Kwk^GawB|{ywR%g#f@nJZos|JQ~i~P(~7i5?vC`HK+zHE1tgk= z(CZHTNtp*oS2#}Q+R`Dg%Mfd&)Q@634I@H}A5KKmb5K^4aO$I#& zkMb*Mx?f?61HVLV+g9M7ZxZ_sh+wB~!Cz$hgXaf7Z)4S8KKK;4_+&JWD%F5p8d>x; zZYtLRwkm!Bll&SSeDC-9txK^fFo7Y5mjw29^T_jQ`_oVqoVSfTOA988eoYpEPR3OC z4S-H)Pya*(0wjw3_5oEF;9pokU09c8MUrkzQ50E*XM^g2a8r5ZZT<7-Z{WnxiWSnM zf{(kLx%eSCr9u>9mdp}X@3EL2BRH0pP_ZTqLC~=mkatR#5(aI)pk=L_eT&uKI3BSEj(pXw2e8GssHj)>R{K#ca=A<3~q zQGKUzOZDRojb3Gj>&KFSS&+--GIvd=Vr@D;EyZy<4E&3>^P%_G!+BU+O1|m;Km7ZD zCcW8-G4d)HT|s(Rw)Q~7|7PHSIl;#PWG>>c8`=BpS%l4?Z`z_SXS$BrR~RD>lm@kX zUFp|^jI0}C!TSkAbVn(@VZR8aHw3hJmjO3c+RBTTejDTr7fv?*PMPOsjbvWHTF~iV z`o#?mnk=`U<6s3b%1SCxG0Ha3e8ptvZ?A7I)b6{#m%YL0pRBBJ8uHPbyks&Ffgt1{ z`C(b1v?wsopWxoQZJWLWTq$CIIH6sX?S;}uIeFr?A>vPpFwdh0#aIpA&VYPw1%A_8 z;cf&58Dyf}frNw=nyNA8un=_7JTCUgFkiommXz;AmNc1~Bp?fZg=8zGS_Hu#Yc&;i zZ_P#es9mxI!*9|TDrS8MV5v-C#fJ|H5LxoaX-alOLziYY1#wMDFE!{m7DpAF(R(c; z*J;d5WKeNb-nm}dWh<@!8vx|tf%?&de03b~`}h#;_>H>};91j!1gP_>kz%1U-;7>J zyieM7@-*$OM#?(g$~Zo;+G$8M84j07`$sYyF!C&eK%w86fP>rI?TEMv(0Jopb5ql8 zcxPOIfz!WB+I8He`QwC1wAcJ`VItw7rKj(nTKmhML z?fXKk2^5`OGHws7u>j#DwJg~%404RsHfFUp+U4sE>OOsf2po2n8*=W<@Sm%r-FxRZEn>iIeEg8p%0 zS2wrQTgyVxz8Mpu$0hQTIOI!9`K}J0`$V26N-%C`1?mvjA$ ztDnw98O{=EmkNO96hL;21A#6S@+ch;8*c{3P&C+n55orMj-1hv5f4;FJaN)IJV(%` zuBS~3x{!{G=Z86J-PqATl++}h>!b}q)Jhbq%OMJ>`>rM}(pq~F7W&c+E$ zHY5G%)1?q4tU?zi-j0N}3_$t{3^CmyoAD;ovW18Sgnp$sW0&tnWLD0m=ME7GU8^=W zb$!FScb=w?oj}B`XkWu#tQ1-85POa0b*3#J$EZi z7iryfU_SDhXMJM#wL54Lk7U@}+OlGat!JC!s`cxWt=R@%$Yb+NqTifqt}R;#P?BAP z7fbS!+39t6QMJbIjIn}D-jn-lEnxU+>sey9$Wc}*D_9LOQIu9YJl-6U~$v*dw zOqlEO5SOsA#^UH=cQ^?eGC7W^3*#y`iUd?D6P`@G73ak59OLdkNrYp7!M3S~Cq= zagM@~J$-jwx^&6r*5U;Vla3xgUgMx$b{r|542>hpQTdksu_#q($(8nb&E|D=b?T?F za#LXf0|tRi3Vh+u|EVu7asO+u>klTr=aT(zD&cO|_eh8?f$Nu+mI$>60J}>n0f}A) zNY7uwfa>^Nh?r8~B!tFl$#yF^k@)w)2Fu3G00U{jUS7k;-Pfi6hXQgqf9Cz5c4+se z*UQ)2!9j-yTVXgmNu1jEf47u3UVx=gfzQzwMt5{4`NO%SreZO>_%$40xynE+B*XjS zPa-Z8V&*IH@y4neK>ukPX<}{geCqOm=^m||yPyN!sHF<`{(AKzTHf(wQ{-O3&=(&6^4R$SGejLqW1ttJ7{fnL|$j9B5)%OvZO5 zIBvlGgfsd|=@l(No389zgIq}aUoACL$&2|W`ySN~m@mCW+f_LYQPEK?G@W(fT=;iF zf_dJ^q2vrSW$ciy`tCk3^fy?2)uhvNQI}oqW&4OwK+~3?4RF)=8lRl!8;PYP8>FN- zz*14o9^PYu8kuHfWi{Z82(ybcTEp{NAb)(5Y&$&CfIiN`z;kNTF(6CDXm>abfndl0 z^KIU>%fZQsR+l4(R6a{YV>&&^Q3^B&l}vw9<__%d;6hGMu+9NTsEd6XzG-Nba2IL8 z7b+EK3lqz&NQaEDEmw&uC|W1t?S)Sao_2HK7p%39l2C)CWC#GBGmsi+zk_oKWrSk) zs0v>|$J)3jcO{*KF|&}Q_F$Z&!mt;cG+H~W0WH!MxwlRr8(wi!z_V42_BhO3PoQx6lQ zslbcBiD{UUq%9b`SqP5WyCHr^mgL^}4cfNnD{Xx^1@P2oqwMk}b30T4OVRIE=14%C zk;sVO(Bvf)=~H$KN@*Nnw`E*%KLWsV{GTfUns?EF2pYbBF9gV~5BJCJg_o@=n4g5y zSFu;q0jLtf(1nEW%e8FXEiV(@U)W5h1M;AcLxR zE^%q=Gl3(I_(y&D(DZFU7=ZZaQRUHa%=T!&vDLdp7hgH)!lZ-%sxLs@8!c2ht`68`OS^+g`n4AvI)+0>%fE z5z3Ae5x2zSUU8NjSR@u%IkWj-B!|J$r*&gO$+?3XB|=1j#oM}!Aka!+hev53yd8BC z54MeF-7>y8Z{ED-yLz`|`JBF9U)7AYRSEsVUAaR|kuy$i+Qh}%*IqIt3%t?`heIy2 z_Pu{Thll%cxqY5eR9|?Bk^7Ktd_%F>yEO(OOPxBN9u8>3wi!ON+<}_Nl=%YBpD11p z3&;ec^SJ-c&%E^+JAs}ZUvuG0dad?RrQIqYmkcqnAuY1sEm#qv9me=~5x-f{H`BM|-B?}Y{ z9ypv2SQfG=Pwt8M)Oep#30upt^{lx*Wnz8r1eFdwGffaQ!yMN0l2o@&d<0(o;QNwE zP{L><)kUU2c&=>JKiQi%*4X`6;)TkpF|83dwf44sLFnWoZ?4n{a0@Bj)B9$#U!cLe zyLb9GSUEd&J!@4Q4Vk|^>r6z%*V=$Mj#lJ|r>zg1xC2bhZWIhPISELOb)-h7)Q@jx zIJPHxbMve0-ZWN$l?rK+9gqTzW5VTv#6XpjEYqnoT@9`l+Yu8(1G9vq9f%GdX8lUcv%cQAh`&%5{f8wS^rIz)Z7U>)4gxffXYZa}%;97bV9P` z!xhy7qZ3XMwj6jhkxaki;<5D;ukd;Ntjbj|JhpI-2v!gP0 zc#^vPe1U0Nqo3L4O_WW=8@Vc};0PTZab|~n5huA8Ct0O>LNNSm zNQHAFZ*+)a)YvxEdkGf8A8ghcG?->;1)Lur%SMOs_}!#RXiI`rQp?)-UJncCEPGze zT&$cui6a`E_Po)q!PM$Ybgy$Igb{ks1_hWUKjL*=7#a`|uvA%);2{2O6Cj z#?28vQl6M5FND(9uf#`k@DqAklYr%!ZATA?ugv^xRk~WpNV(EP!{gwla*4ioq9_oR z&d-S6!ziSKVjTig?%Q+m zz0Qi8?zk(RkQ74IrSjID?G}2L$POQ;?p*tHzYLM+jZI#O$JgiV#CgAor>^o$dqm2h zJux%s!{wdKu@DpmwGAUo%+xkxr;;}Uvcz#QroZU;uh)9mF64pN6YVa1JIseHW zQG_L>LwDl0hl#PUjOTkx9)rDH@9f}N7zk|rWCWk;)8yDGTMClvX@gmg3P1U3(DH0Q zj$VQMi5q5_jdn-nv|~IWsVcOtkPK25wrnn-&uP3l%~8oE?ORFAg(A-~uG)aweAg+y z%$%2S6Y}--Smtr?d0+XSYB~SUcL4(Kne302=8_C?Ku9i8 zI(1i3^@II0d@}zzGwm@G2s0Fw^oFFQAZ4kZ}38~24_<=E=Uot2HfXXM>EStA0 ze`<2=6kEf0j$vs$Cg1Zjj&m;Sxw?LQ107^~a+J(thx*z^ebG?CpvWnpT_~;ZnR~rh zxlw$6QJPEwsv@&pY(hreUl`?fv(=um8BN~C7~t6gvEc);rMiBcETrrVJ%~cCLc+d#qIs8~El-Mbjz@ zEF5%bTX3B86h{Cpx0LWF-$Yp01@XW-Y%aV-@drtFIsA)nqG48x4F#|t?H}M8^YpQi zlWKk(kL$6jL;H1AR16TZ8!JRajL(jAr$83ngrj1Vp1YIPIS8Ls&SA$6_OQPE$$@10 z;=aCrVCzdu5u5qVu!#WOtP*)Z!`uWbC#{l-P9pdY{KmX2l%JGF$jlCMos$rJ>==YX!-$ceaFlhR#nxeA1G-bO@=j^X(8)I4L$nd6h-zD+kd5`7yt$snKk6(^AK0_>tHvxAmJM2K zry5OSA`xOlCh=RM_bd=@zP9}M5&OSbSl*v8Ni7q;gGyxg0hNQKDbp)vw@;|dpxkm1 z=|g06^r%4YiaF~n_S(KWRtb2-s2UE8hLu+Ghr72;-L3^IrUx;A;k}));(TjkGq`Qp zN{_tOQPY%N6RhZx?-f&*r}p9Xb=#d!D&$2PSK#{gxJi#cdQ}VWiY2huzbJuJy|ky= z{VjS%*AU0HSj%RQEvlmYhaCp)3}>Lp5jT(61WS=)4x&-ni*M8%bna+b`g3eMfBn$M z9x+rC4ZQkDrR$rwNRP^b!gkzfwT#SM0dlm7RlQ&LizaZbHWGaEB%noBDx zl0UrLj($_#akZIVzQ@9z)TAy46|ER=sd%Z;+#AD^HEOO5<*J%irnUKVoE~%Qi83G} z%{K+hkF0tBkejb>$q)|9YT#!s<1Oo_Vtx(aLDSU139ZS z8)~Jh8?~Jmyw_Sa^cuk2a(a1}2I~ycGs=p5XMKJ@sGq+anb=#$E_>!>8)KB}>4*20 z+lVw>)I+bCSTYr`-g!InRHew<87v!?W2!3QmD@)I(z`b<@I_}wd`O1-i{UeT3n9;l z@KI*Ng*QgMlWb#7$tPHyw|jq0t$6#^gEB#r%UGueGdHs(fsmgBCKalYR_mp@wZi-4{1>DlXy?hZmkz(hk`~RL4U1 zXQXXPcRv2)smrsCE;aIJ1H(pZmgFCPXeejz+6P#(SgbX~&-WA-O0R8i%!+nByYJ3T_bQ zhC`U$bq`xgjr6B`3YMQJzye)reQ)An9Rzb(BK9&sP5a)BH^xc0?RcHLT_&#w-t@&+ z-qV?;)ZfjJ1}3@$Ft(4DT^$#l8I60f8$V)$Zmgn6Y$+N@m!PyuFJ{aspkP}+Z$1ml z{cDm#pW(ZB_m1*ljiz}N@BM5)sF(Y1Pa=#Nr{!Wt?fHzP8{ypD`A_qQMwLIbR^F8w zquc?kgdL78?)c!sMYq`5fbfiRub-z0ZSc+2J2ySRwZ4enU0D3SHJ~NGKsfm@&T*I| zT4PJB%E5pjD!3Z^+})v0vi^*Sqf#H;u6rnV^o-7I7M6z@_nzIlX<4Ig0Pcgo zIn=(%1AlmPUSDYUK};WRf-0%;k-k45j-msLCH>WUU^D{L@P-UT1wwaRjNY1zq_~P%^1adU2jNE^d7Q_llh$ ztiAK33@anE&?_1z=7N){xnLLEE4bIBL*!}r%hzjPEm*1S-F>GhAtQ+^X8q3&bNHLG zS4i|7Q&h*pX$H@gq^PTPurnmg7PCAqzpkz?*z6;*3?%mSxtY0GVhLlU++H`>G$Zy& zxnPCFa_Qlhi;X_p=|_%zGIDRV_P)*-D2fLtO_`57z#9v>5VrkNx^Nr)tnzeRzURcf zDJ$~^ys_1c02Xv&;2T({Ybe`VtyQQe5-kp;A~F=r-4e0{o;il;4T-U%c{j}XhIEaU zfKHYS=<5Bjrk&bpA@I7Ok00PC17KG%_-r-X*e(Q+ba^ZGH4*q(5Gs4wxZUZQhGgbQ z;jXSPUhjHo?dCsNF5if~v}^Cd0t}ZnUtgttHM?n`)HsjH~p(bq}47 zSw?5?%p~QZuQ^1m#cMkVIoxgaJ0T^@+l~V$|H?3>q}zV!Qc&0m>6vsGK8@$8t;#?j zFZ%V`@@6VjF9ANxFdWdz3q6g@HYKGPwxqC;k1hTZ9S5T48FBlI~!}4 zAyo~EBtP$!D;enJ3OcRG{QDr}P%2wI4A6?2Qd*ww^YN&3L7iJ7zduf}Zw<9CLQntX zLuXY@Pe7Mmskzd?NXOdD-nmw=^0!3a!WF=^%p(h^oL>pHIDNH#!$0CeY|p`;BVqFbD&@K#9Y8mN?_AT_i?b9CtXCaTkpS}5B(Ifr~A2} zA^ohNeiV(YXvz*l1-CnD99bf9>P#Evm?wM#y}Z0AWGB{SK$pD+>EH-o1Db9T^=Q;u zvZYLeXR!(LW#5!oQSF*xGpBNV7qmP|CBj3Kf+`~u+c#04RNd4>7*AkoN(!GzL>|gy z0i0Ua{E1F;CG9bUKJ~Mqo*b|0PHocGOT$yzvJFF3I}$fn)kdLmKmPOo30|NQ~p zxBfs#%`Q!jw5C_u$>f_>CuX7(!C={Af1xNB0i}-Y_9Wh+d%^V&SM1)(Oy0aH&VHMM zva+(k+wQ6dI2aIn$nm0xo``!MI(q*lB=(PbfO!i2=D(9W?leGb6x=iJB3=FOk4 zBsJC+vK;TC03TIL^K3Q$ze7RQBRUg{B!0c%;o)(I8;}GFQjYx5>g48dukD_QV}ev* zXqr9Kr&IY9fSQe)W-IJl`Y2`jmXt_;Li;hEA2$AN3$H#<;L{TLcZTkd&44UO@#2p3pJ)AD+U}?@FgFY4A6eIT{AyYkuuNVLQA}xK#BB{ zF;sG@{7*46K0;BDI{@q?KWGA@6?&%lgf6TYL^#IR@QLS#+d8Wwa&`Fl>wwGJF4J_5 z=P?|CP%n9EZ`(c@1{2}=q=0GAR0y5kM{XQ(K^gqMp`q-6rw@+<^^xb($u!f#=Bk8O z3?Ca~p5kUok&%AK+BYN!oon>lT>;+#%;JcayLRC2J6!TLQqSCu_8-cq@#|N!c7;KU4Um7|KHrh~Oeq z+D;#Rj_MPGLG|_Xj8i-Kg@Vc_&m%2jL#nWRSg;WTsBH~z7T2a+_tVCW#=|?;|ijGJoCN6e2U2O4AVWQI9^K$@qMYQFn zVB6+25M^p1!0%-BQT)fI&)6Twhdj$A3x8hA9-xTTvQ9tqRyCZA>f?~Jt*Q}w37^FE zh$C^T3z%bZqg~M*RFYN+zqLI%2PD}FdJF3ieJFPE0H*ksO_g>84cB-W$d>`3V)Zx} zZt@IvSWjU%7%i{?#l@Ffs6zbJ12^TZeG6)9#Lzuvklp6}KHkN3jjmG@Di6NK0DwT1 zbIZRMb%~AN0P6aHv~gtQA_&srs|L}eor)LSrP41Y+=_R8!hvVVeeA566~caHR7@y?=*Kk>L6-mJ*MSciv9%}V(=p-(!_`;9XkIzzx8=nD;#DI=>1iJ zWQ?>PJ|5VLl=8&>}mo2aT6~?A~pfSmArvObD=Ef z=ip6P%l3<08vZ2;+NU(obs%0RYWzE)pxdASdbKul3J1IhHA}p*0rbN?C=j^yed_Ug8KQVBo7@NAFa(#n%Z0WaJ++msY0jQx!4eJP@gP|+5Jl5 z1lx-llY^P{^I{b{_|rFQO~}KNSgC~1G0R1XZmd7!ZFUU;i~rv`!nTcCbcWiBYw-|2 zK5(x-iO>}9c`KEEf&y^M8yDKt{y9T$%8$FOGkl$w55Ug<<&M0OGV{l$^J}7DH5To! zGT?H?ln+G55zb~nKnZ>smCgbgkHkBkyjGmu49SHuP4^mlUifI#60%`aGZ66u>3Td2 zwQ755{oi$|Y|>Q?>>a8C>3*iYEO z{hNVs7{5-_gDJh6U(H})iN)v6r$H6Jl9d1*=qy6sA}Fs={U9-;o{u@h7i!O96*VoD zsLs!q`pS40@9DsWK=tWllA=8_r_s~PE54tl36DeRd`E07{`UmAtH~HpN_#a&EFYaz zjjpnM#bT6x{wV|?0k~57X^Q<1n_Y%-7+eAR`uZwaoi+p*RrWqUFGK-Ysf1uA?wwvr zkf$}LeLQ}YPczHn?4@oe#!;n%wp+k{yU?NV55qN7Ij$NPNAw2{ahUQTI?Dnyf9lbl zszva7Wv|t7byie|1`x!d#AI(!ek5<4XVloK=gdG&#Y4^vbX6^&^Q;xsp$4Jq|7n?U z182Y#Q@}NLFs6qM>qUgOqw40Xo^FLkd#%gGwlFWXwbNPjpTkR5t&U-JZuiB|M?|2; zkku#Fb{qoI6G{gMqERy<8r@@h#PParBSfEX;KTAQ$8SRaTy1wM*dIOQuJ@OnquMH~ z85nO6it2!{$Udq)C@`=HctE@J`Fj8p3ZZH8D-pJpOhNr5qphv2SVyQOU%ErJikgu7 zjNO&aQnh#4k-mAarcqhu((4x%x#gi_)r!$J(~LT9{lS+Q&=akJCkti-Ev~n=55$g3 z`nt;x9kuOrVvK2H4UKSY2ck-h;(8tR=>Uea#Y=q-XFvdwAaRAssSDIpF+3O!Oh8s& zrBP=RU$Yn>m~!Xp5}m76|8+Lg&~>gBHaS%(9|8I0bQ6fj<4q|V5orY`=Puk_FD(U^ zVRizG>arp0=s~2}Lk0e=U+q;Opl|o*VZHFZ-W_Z#EVqqNAl|~9uvpYM>~4}R&iKT| z$!${58mX&r2~@QEC@)foAFLcZ3P4o%`r-tY^F<-}A9m|=cE%@V$04 zi7X$hfeIC(JAnw7C1rwZ5d9hHr+A243ZYLoNXsr(Sra;6cLa*aDWl#qGI~8_<}QNz zk-eaJrkT^&y7WMpGtgBi*=f`N;c)ajT}36T80nw`Do{uqJs~-uEZ0ahXGHsAHsP%p zW(!l!Ks_$n?7o?%abr{$0KmJ+W|p;)1k|*p{%VaRcVn ziTlt|Q3~dlIQ+;AKP%-`g%iMSUfD+~J+I~q6re(%xPMKx6^!&sVwd@xP)~kbA^$z* z!s8#k&aFihhmK=76FD)i^NBDo<`7Wy2P+RElpR8Z&n^1!DJnj{4t1nriA!HFLz&Mb z0kRR||5Xh)1TDl6+FGhS(&t>rLYY8|3}y9$vpEcb>QRXv+d+6pzj-fClg?LXf13+! zwW_wfcsQ?K-ZabU%*guYdo-@mJQq~a!bwIUY4s)nS znfj~~vr4UvWJ^W=>A!{(GU!#*?4)O?v0*kR$XE0}V!gl8@z@&4Y%Dc0>3v`>-d!ks zj$U`kO$&G&ctS27#p$6|gX*sNu=J<`q&-o>`;YA+f|&{KE4H$hMd10b{o}*M(n$(q z%x=jO_#I8%cj|%=%VXqy+aIo+XkL|nB%9ZcP6yTRTS6@Mf5RHV1dM6d4mU6+SBf;+ zD5=+>k!_~dDClCl(ukYJaWt1$x99A5Wd_UTL*K8=-yZX8y#r+ErJ_K3bU}pETZ@!P z$k+)U8Qsqk;tP}x*~_~*57upAJ{p4Jo+_l~+Wbk|6)RV&o=Qfpnf&D@U4T3rlu}N} z-#`<+ZZ~$jQ;Ky&kv@67sNvgd7W(dhM~VoNSbAqygS@Tjr*a%^J%X^mb$q9;eI2Ts zow!u~rk_!Eq%S+66bOdQzRAI$WMGu|3*vQQ8E+e5OLYl#gwN3f$udc0gGxNAMsM5p z?Y9Vf&A=c5MJ9>iFINNPSO5rl6CX>FA^N$hLR~G?^L(h7Q2+sMq5L|IESYFChIYam zVv!9QxX{DM5bE)B6|1Yk5@z)ju_={6G`(gHyUodC@bgY>xQmoF{_}*ZCq+NtM(V51 z)_9`eBLu{EmtHOUy9m^D)9j<-<|o*aH2wWa1cTKTxr$>i_RV4U@4c-AJ{qBL$8j`8 z{t7SoccK0rMw~?3-|xOk`{LQucD2HxyrEobjFm<&AKm_pb9+uliiT-CcctZ4xW!1& zHS%ZF5cPy5+hN`{YrkACnWVmM$@3L?psRORZ`YT0{lz9ky2@s?)*H;;c1Tts9F0h^X@xq=En zAXTnV`Dl;Xgy{Ycgv^8hm=Q++;yyh zP-W%Ob-l#S>5Mn8l4TONimzps(uIm4NYM^FYY{axAH<2^G%FNz)T_w#fL7b0qE*HKoY87lwKorJ?* z&Kcl5*1$NvhQ+{|Bu=8}z4a&u8YkxpOMx1as~!U|-;Dn|KCD!E zg>TZ)G*Aq0fnv{x-R{#y%1Onz!bXx!`0`>=9f)Uj-xluPxYHALKUuW77H38c^jH{C1KAh zB9LmeolI%=rMx-u1Y9?6;$Cpab^x4^SK1A7PIThCNWKR#3ixUvyA%gV$w7z{r7aLT zVLur8t|m|!d_ckRrlP`!5L)oax_CrAI6TF$ztP3VU^C@s-z>zi`6xf#EH6Bz2zX0P z%T{H4AlcO$0l5&Y_gaJ%PjKEvsG(L9F?U^109DzxpQ9$$NzcCD$5tGh%GXQHa|cwh zCWD`pG(=rRz=QiK>5+Eo9YRRwCgdVzO%evT*I-kgP5OB~zU~~xgqZ|*AcD=I_LVveUj6>IU>iw=PPAz{7@oO-Hu@9lbB18A2Vye;4Fn^OoL zf3>E0u5H}{eMq%0J;!0<6%0YCT}ZuIbnyw)!xF&Vd(cT8LN!T%4VCl<5qG9hPVnwX zyR>_+0R8b2$~lyd_4h=Fxj2^09A{B3;KT;Oyc8qXeI(4T}6p$Wj`s?HN zVccm94)yXd7ubq#!b*`Ib8BRYZ#1SC!%9`G9<(mJsglrn3lnOR)0J&J-LiO_U)l^q zsTDFZGK8Rzew>Rtqj@Su^j!fd0qveS1(*X1={Vu# zvkYU7$$gQLg5{!{2^3?D)@1q)_EG;wubLDoB?oK=wy_tJV|nQ^DPd7G2FQ!$pG+Ec zCfd|`pf4z(NSu;_g=0lPicc}`UBZy>GH@2>D7PsfR?_K*!i?SW)!Z!wbOHg|+yq`+ zjDeIE9N*+$Uq8@A#j*~C*g>p{Cd}+0-FdE;lTeo3q=bfIb3I(@J>YwG5c5q0WswFQ zyM%hctQTFIyIdDSUlqr&sw7moxbk!Oj$L8=V#`L|QV70+FbKLn?!#>_nUGr);xE_3 zpc`-TUcDXY&;1X6qavUOXLvjIW2}W*fj9LihxNOimIKvlwFYs27yG&zhPsg>&&Us%Ke!+$a0b z055T+H49d1n%wdI_MUV}T6+3dYz9Q=%2J6Ss}F&+7s}I}6pN?|z|r9X-?55F2Y|E1 zh*S4%a-9lU)e0ypGdV-4!J?3}SZeqTg^zzpWJpI(2N<7HAlPykl0`4%+efb$t>D^3 z5e75fW=!98Ai4Bh^a&C2pqN2t^Dw)ua^2PrjNAEO5V{a4{AA$&V((qRu};^vakE-8 zt!dhf_JvH5LehqWqM}4m*=Jh`Aw(e|&CIHlqHIG_NhLh8&%P#mq3rj{HXfTn$o@O8 zr>U6#djIeDe&73k-}@cMIF8k7#Pj@~-|xP!`?}8ayw2+} z)<9emYZZQX0dd4#xm-~CrLGz|23xg<_-2Lp>Mf@@g1fp42Jn+`aSsLWoriI;Q`a0A~* zr6crd%jlEhDS<0*x-L+#+=mXADk5Dp^qp+-E>seyN85&qdy>sojSmqI1muwMwGaj902$O)8W?oF_{M61#!qP!8dg6p-@ zDFUJr%cgDufduEysO&vh`PjP;!aKofkeE;wT_Pvr%G4L}C8YyTQ?oEr`Gm;%|NRCj zCs(b?n_vQ@@lLRU4aORClB+P3$H6(4K0YsF3lfztH34x{3j5T*`za{GL(B1N_i4UG zTV>op14;)UIuSw|>1it0AsCf@Ocm06`co9Pb=mvvP#LYk%Uu&la6TRP(p1YVdl;T1^O2!Mp1Dn@(|wo*#g$LY5vx5IQWUx# zg(WXNjpgZ&6rJp@$aZYBMG~VXSnt0=n1mJgfnLN7>XPcz+Y~dq;;xh^Nk(LKnRnk} zbPV;`(78ZO#{4xDkhng?G%ao@Q`Zkv!eP>I{MC%_WkZn5ZbN5kA%>8Ok9RBU>&E4fgQ96f!!qgfV|+DX7D3y=)XEaj zyr=}=d*Ch5j@L&*p+QT?gF0+@;EA`N-mjzTx)N2)6QRLikU&LWJ_ za5!%ZN+UPI7T>F(4{Ibqic#Jn&9^V-o8#{}5*cj5n^YWFj^z4&y{@fwUb$ z&-nxpQZ*@Btq#>PN5-{F-fw_rQffx!?VE})2RAj2AzAc$b6TeK?&*ZKRA<{n>*#yGtQb9(5%Eqj@_uPwFmk_9HGP<+ZlmaG}vh0$fJl(r9)N{n8E?#2Bq zri=w6aaDVQfgN|3idpEy^;|YTMdcoV*`eh@Yf8PGmZ7Jh`t{kK-*(Q3hVd5M+>TXQ z1FJp;?uqb3g-M!bPjE`r7NQVDQar2CI`Is6!@NA7)7I9eLMHRr*fyzAdW3uH2DyyS z_8r>ZmlY_Fq8nd(>gAk`NLg6x84{s6jYnClUw?9>f~KGW&UTFH34_3$c#}+k2~F5- z*I;;~n%qU43QJ&ygL2ET`|qfIWoSM2A};Kvev`WThoqxK%fqjDVi|l3?5LN-IM7BN zkkjh(BPBuQ^6UH>L9JY;`~dCl+(6H2nrA%1VyFE6BeVi5&`IV-hnvYsqU5MpKw+|& zXU{bkAZpTZs!|J0q6}IexO*?@yP$GZjhIQZ)fW07!#wh2$p_pi&bOi_kuq2V?lWsLP{Ef#IO!0 z@vs>qkuUTl`n>4}G=^g^8qJ1Pc_CBFf}Va$k{*Q~I8c!oBa3h>)S6GMYdM-*TUogq zG-oaZT4BzVPH5Una356N5?axSB*YD9{s!vmFwNdspi6=)l*1W>&JwQ~a$Xu5W5}-t zT?nBdhVhn^(>WdJ%?5rnFNm{jkXnI6kvvG72)5(2DLq7bKDoaBQ31KvskO2}?$->{Iia+^so4@!n8xW>P$lS$T|>iqTuIJNvaEt~6NErc;Pfcu>@pzk z5p;^&8aQ-}BMDNF?Z7$~P_060f%qz;0EIFpF3%Aats4TU7;&!@GUH})Jh=ucJR9F! zl0McV?0HB0$zF6kHlfJ44tguKxX1W(A&Nn2d!h=-3X3)RPeUyX$R$&3$;A5un`9on{YR%y> zg;!BiKH5{W6#=aY`L#n;9|+$?1j^0fK}I3432r&Glnel|V33Um{CLfqTUy9kh6;{D zXc5H!Mf9E+8Q=%`K$R@rE{Yo{OYb8_vuZmODN7*DjP!i)oEnu;n=Gf&4aF}>;7_7Y z0=~-W@8Lxm5vMY-GKsYM=W#Br@%9F1Mv!+;`s8^U2Zc_RlZh=pQ3wiIu;dThepNJJ z3-#=wsRnk*Kf!BgDR4b{S6MnZF}u(Eyh|iw;gTiUprlq%V-OnAB_Xq5$B`fupvD3S zp&1P+bQ>bl1!30!95kl>HWJ1_BfyXm8bHo=1^KNULM=zjs{ynnxFyt|qcUmEh}`4uwD{35IH^{#b%-s;Bl9%G#E?;Gju?t)u-Xi zberjo0azcyI=hB~Ro%oVbAOqaTPE(WB+H^RbaUGjO_4JWay;Dmiwph#d>lUmSt0iJv#F2I#ez_=0| zHYBgWAo06OjAg`KA@{isWMEq>QUD^#AbOHRj(Z>Gz$oV3D-mu@y*l;uKnT%iBUlvU z29JV(*n=Q;AN3%;-5#JRUWIKP+5oo9Eq0g8l@xP7d8A80;i$0{bt zgsh+{gh(VTHFxxoWbN9Z2CrlHerygC}^v!Qud278dJ zSOOR)x;LpibIW4-9EjCV!;ARWt-FSn30Y8P$B^5*(odPJs%s^929OMqf-V4Cc4K%j zUd-3YE^Z?ecK9sP{|Rr!&eCm(762PCcx+PZdMJ=w`Y^Nf9nd(}L?B^Ayv9wEHM?!= z%Q3AWSr_rj)Ig>_NMdO`8I2#_fG-j`?~V4z3;PdNHA_H|LSK&+ub-NmO!+SrT3c(r#R!i~vQNN~S? zd64D-ev&nmG35Dx%2)vu{qS-gg>)pCf+nycdqGV=BD)cfzJBG>+lV6<^6=?FLMZ}& z4wR~&1ciQa)T8|Q>%t{jYpzm{9;Y4$;3L}cj`>0fgn-`2aw%?uc-|ccp0`jNp0WIz zK(FaH;HSf%ux;e{E+{J=oumn5<(UMKe#GIrh+^Ld+j6XFDvz);ligNDI1qtV+()1n z+5nUY+Gg7!^8s?g>_r+nhuXS0Kci{(k|Me!^|l1nX;WXQu5G@E5VOF98Nb^Q@v;p! z5Z@=w1JwigKAT@>e2s-GB1c4FP2$^Meqbd;{kFyLP&~TAQnl1xl8)P8=gA41de&|5 z=7HFm!q8ecad#|`A%T7=um6P`?%3~-a;O}UDgl!qtqi7#9-Me9K?4zx0R zNmBL)ChROxhm*bVXa@pBN6MTFeRz|@!cMgqfQbKO%Q-`&cIp_z*E#Jzzn_GSX{zHT zs(t&n(6fBJE!XC_7oD1@i?$@!Vu(7`K--*4qnz!eEeD9o$l&)K^-z3y z$2ziftaHSMJa?XO?|Ip1i0D2Y%; zmrW@0$wXv`q!Jd{GxdVm6loIOKCam4Pz!+qTIUY}`i`o@2~)3zNYA?}YD zR5FqDFS?~qTxyPYvxL7F4(T;QJwQaG0I7t0esvsfRo2;VT!A~-MA%2pPI)c9aY=fu znJ1E_yrGJIUfV|z)vETX(GZ`c11jjR9|(WjqE&RXOhL4Dw5qdU=x_~mQI%OQZdf~4{WTdyDhwD%k#7JBG8c-HYGjOv+y zS|vQITBKT(A?(J2C^n2@>?(o`eAiZnFufMS8K%-W;rQ)R7mzP=CXHnBNmDCyVpPjB~OT z<0B+kC%b+fBF^t!y%6V`nFQf-DRn8_@;97^qAB4!tLCYF*Q5vF)kH#A3w z@Q!*9q2wWWbRjx1FnSRSmg-g z=a-jv9H0Pdj)JR?3nZAdL@8;2@;AVPGm#&!FFKOObZt&L=K}`y-|B}Yu~PB;5^6pG zMkS~X-%k56V7_kHm6C>a#Hc{}%gxeifOUuTeJMLTJq(grpM(3M{?Vzi_4V^{5F0GH zPSN?nm`V`lo_Ya*v2{)th*EnhK?p!GHvU!X`D&msUYWC4fJ z@zJzH0+&UAGIazooxrsb;`|Y>V1TH^!ZYca3uNmpqgRHyfw3-%K^6qmlYlZ(x!wT9 zIW;68{^_=NQ4!@%dZ^HFfd)ILTqQH+vQ`4jsOyjbDQcK;Y?Exb0(NUPSH^%4;4fmR zA4KP`PY(YrBPo&uc{q6PpR@Ti{Pbl%1mHn(iFwIOC z2a$aRT+P~5tE{c9X|K6K;28be5atWBI`njKoOsty0PyuvR|(x|6F`p-fx{PTVkXPOb?%J30!T z;U#d=vxY=Jp7Tf2^wOi_E04K87y&N#q;JbAlH48~9HMTjs;YnkZPft2@+Kzm+@6$m zKO{60LVDeJN3iAP=Quc2{BWbywk=t?tcf{uSB6H3pn4`H9puqe)ral3g!-%_GN!^S zJDTD{4>Px%K%lDXv|dzjd=L@Wcf1beo9s5Lt}9YftzTPnq<*Oh2V?HXPbuTpA*e%1*n9i_kJ%-6;Xc*3fy$k|mE)(jc{tPM5^Qu~6IewE7AN zVGp45fk;S_YZ2B$oNo$Lyf;B|k_;x$wwj*^gh~MAp7k#)oVME;!zfqn&TdP5B z3YE~;?_myv5p(TEE&IK+nkiI#+JRmGt3hax!Q*a8PB{J=I!WTH>+iLKpN(GxwxD~s@Qm8##|RPnfa7q;*V2R17NQR zE6%ug%#E-XmUjACgw2DwsTRr>4>&}I zN1#oyHRgFPT(qc6VcYR|yelO*THoC@d5v0I_j7$L(%$f;*8ZbD)^Hl=Mfb6xvj$&e z@Ph{r!j*O>5&}c=88w2VBCapN6SJn^0RIwrP}6rBt5{Gb;90?1-V%#iet|-LJ6SlnxX?8a z$3;2Ibqh;;L~4Tc7P3Uj#$Tg@g>g6lwz^dCwsoQgo;~hvI%Q>&;Ti$f9x1h_vGmV@7%~!;{>q+U|fM? z%l2ZVv2jhrmnDe5JmvyxW=9nNs#Vz6i(%5&yr+Wt01$E0PbOj!W1?P#Ktdw;4TYkt z!GYsRz~*^V;CUXLQ6L*>f{bz<4w4Fu_fxN4 z#0mlLs)3vM!;{0fr`Ms-T?Z`Evs@c@b*$Ijd0oKn^=OMBjkRfO$Ib$KZ-T_wzIe7{oC)#( zS8T#Pva(b_kQ{-!FN#DO!+6t4KuZoD5?59SzAD%!D_e`YkjN-vEDu*9Ev823>Jx!5 z+;|2ZR}=nAA@6BKL~|15J!;+0iB|zzrNe^2lszWzO5l@`+~{QD-Mh~}w6ZDw^D`92 z+Me>0bEiRJ*q$9>?nU|ef}_cs?785^w=OIuO6LZ#snk*iWtF>2q>(nNiwf%Ix`M}= zK|L=xSKLuUM4Qh+rb;C9(EyZ)g4&U!V>1W0u!R=NXZIx*HwYS!rI>_6Ar)fOLyN2= zPwFMfjB)4%7OSJuF+g{d)yp|z8H6fgr%_V*o{mZkS0x6IWXw3P=n#d_Z7qLcQr`7M zga?*B7DAyt=<|T59XBtc50vloDIP3wY|)fMhXS{g9gIA*L~ly`F`x=ijm4yv73q z0?Zozc4F7R8wI$X9O0Hx)!S)k;BrA!HkbqJ@<8%GJO;ZKzj+ z^yucX1<1dmw?O!!TuolvcwhJnVGF){IpvDsjh$9baBL%di z4%1#R1{u^zK$$1uDkPp~X9xe|Mm@JAklG3zCk4fTx>lmYsZW}=9N8Gqkxj`B{gS=C zJ;~cEAsP=x&CTM)!C3^ZiD&^u<-*Rx3Oc~R#p|JUWL+K53M{0>)X-q**u+7b>qEuf+F%Zz-lI#U6alKAD@uxKTcE639e7@_Ln0U@OG%+bK1k^$qdzbIPiSD$@uaO;4Ojdsr=850wI0vh4Zcj zuzMOgiZ0NVH(BKq&VMa4tUP<xAKt|=qVGQ57I>b|w%-kuuVT6fg*sXnVrm-AMX#lT3> z*}Fx%5G&L#mIk?FwMCxdAT=442Btb>_Cm-1O-eJpdNSW`o1KGRjfpu0G(x24Y}YWMm6_qQ9i@fBz;4=U7O*1(-4^P$5d zs9wzP9Vz$8)7dokIzsC{hM*kaQ>x{m!8C|r769J@1)DreUlO9{Mxd9WWzSGRFJU#B zz`jY2^f46cL90O%%1x9N0ToCqq1NFolZ4u=3GJmDGuKS{Xgijth?yO(9VsqdyIww{ z`o71l#N&$Lf@8?vTkXZ)geu0ofBoce^Ks+dSfry^q)9c+qOKk}ZXU(5>vUC?x~ebA z8o_U+JL;~(u zmlqtX!q3E(~Swp_B0eXnnnN|W*B&ZVizu@RT;_SjQ3^huK%oSTaT zP-HgCH4uF&lr$+ad~w6bh=wGK4D{Szv6*s@*i4=}njfl_80?SENNS(l7oKnFwH)qZYEBg`!ZohZyfT+a_~=AWHUZwh8f`U>7GUyNV_j$UhR0#wR7BanOYGf zGkR_2Scw7^;^+-*0w{0|o?}$HkbB@pRAf{MlL!XKEA?Sf4;Qsoammh^a#cGF@fv53 z)Lzp}P1{<&*gFGTi)TJNmzJ2T1(%k(cqk*K_I76_!yxQd+!^zAy1Gj@FOGD(qumO0 zVn&lu#?a;NgB+GKij}OnkdKeBgutR-h@qH!_lnbTOtM~UVQU6p3e1;UdA-# ziL&?=O+N3-?-?%i4l&#RkS@A2N$cVO(y0eNe# zsh1-%-#}XLN2CMg3(*9=N~^g=cGSn-rrHR9zS$-WvHpUnGm1x*IbYwJ&NW}yb42%$ ze2SQZy{zmyy@9fD8?G7=SKJZ_b1~qt5^*Mr+1XtuD!LtFe6y48dq|is>F}Sg6Akl?my9)Cd~;D$+GCNc3JSroBJ44~j8NbBdyMJqHZEbL zG18-|eBpUf10C1e+c~+68Xzt?*c*!z#r5MR)0Z(H3WI@e;+xJYE&thXUz&oE_Go%@ zyib93N8I?Fu847c>xOHqJ(wqIr}NG`c%B+LS;@*S^!7sX-d{VSIH&q`{iQ*%j2YL# z7*D;MC;onQA4VtN{Mg})7^oK6WqHCvW|X7x=PZRk-S)pFR_Plij5~X1kMrUGl=ZE* zz1IBm9lop9U+dzMlRqr5`h=5}Nosuhz0^5fmM$}w3+7JSoUi|JQv0>Qs#RL8EhwU^ zfActJ?kwXV6yQ{LerPiDUtA^}&IJ4-$|gN($`|>{%A4HRt6tBqn306r$xsYWM0er@>@Z&_Pg^)n~R*746h@#@dNNeRUJ+W6fF z;+czw^$S*s8CO-WUC$rkIe2xSff0RTgcjudx-+fs7 z*n7)Qb$@;>^0qeatlM54n{(XPFE+I?7tXT}ynuvSKsMjx|6<% z9(vz8x=j4-ntNO1x_~GKPNFvqLp}U-CZg((=Gx6=ZzO*iS@WljcduTzI*Y=t&ykAL z`C}*EtXqQ_a~n_h`hPq=@kBG$)XB@sLL!&(K!n{@J;MD!Yvffw*9w%(tF%P&uTbR1 zwIJ8zlHPjc^82#Y>38avmAcpdImQ-Q)KzB}&8^x<`3m59#O1}|I@~{KFx&@ruij|f zUQH)vY)-+I_<|Uo1V#DdhP!db#%OqFTpb9BVm#w*bRM?DsH@M}N_0&0eMQ&Wj}r&4 z*oeRwV(PBH%g(F|F&{IkEVP;o*nF(`TF9FabJxoIiv60=4>v51)Z(V}2hq~YXJK}6 zxJUH4{PJemqbBOItUW`=_+|BiRcl7WqR-b!W-N2CbB`D5EAC6X%lLCdU?2zlIPtp6 z;J0fbbT%BDpoA)*mHu}~Re8DDRfYS{9&hanShQyq`%UjfZ^R!RZfsrq@oqSsia)yD?f%mn@co3?jb-LCfeX3Zubxrqz9a(?Tu$$}Tr&2y$;}0U zebE!?yAH*yy%Oosfnabf!Mr;2n^Z};28KBR2Q!|C9VLH~F8SZUxlm~mW*b$a-gOK& z;+0+!-$P-o1)G;dUMm(AypN|Xj(aT=79VAvJ@)=hsM^Vrdup*c`SNz}4@UlJxi2U~ z{IPw1n_KcsA)tp}NO)RhMh(5FibeL!Z1?`ZrIVU|JhA=fh_C*7(W!gwuY3AmA@KWa zjs3O8zTAPow%A`=?8{<+O837;Q~bXlawgnU9o(a17(JPj^)beLn>}CcxA^;hm+!6R z|MBDM9jR}>%}_aW*!}KcW#i6>Ss!D%qGH$kT-S=rxySICk^e(|$Gw>mOFo6yTAfWz?%6aw z@rD2S%fPHycQ({A^0GTsKR?|6%a`=mW#A6{cUvcaJ(Ry5%761v_SX{qYl;4~L_hBW z^f&%b;C5w6#w|0o?{`ha9wEiWIM&%0jVmE`FE9 z-iPkVHnKB-qxB6^xEw}9aJ*%MUZND3#c-X(zFQS)>wWt%a!H!3DNcji?n&r@Y$kvV zafhGJl)6!^2JcB6L;4-kTXJ!6@#q$=fXFaNknwC443LU;4(LMx zK#qWpXh<(2Q%)kV zm}L6+`}>o69-Vln&Ye3a;Ism&BOS>56CtYEj89{rJ1qr!zA|iC```A`EE5RNZDh?p z#z>FGZ2!fE8%C@#&^S@kVIp+GedHrN)Zpw4HqsZO|EAE;;V$kV$T!+x)*xPPI90uk z+bUVXWng1tL*7l+aDZUle~40o$PF7V48qpvVRYZChz{*nG{s?M+eDx6g8#OZbo5^W60f^%#TS<&!5NGpUMrbl$jD z)YK2gD>y)d&ZGG~{j9LjZ#LPlN9%{u93dE1l9742*LM>}0?o`KXE`#Po42*wee$Dw zKbZ)^8f?WT-|4yO@4GZM7k{hji9U7U7{IQxih!hLMbBSCI(ZUb)r+@cQ#H%_{p>J^k5{ya_icX zlGCpZCQmo?Hf7J-r}+8hvIFQILg~lRPLpQXF0@r8x}|j6L~>ZCKv4DTictvD=ZU95 zy?iNz@AGf5ell48nH$Z(@thB$)QPgH&}!9U3UI>SuSLn zg8ITcRmtLSH}sjBn)cBT7IHLv%=$1mXoNZDJ?AQt`ywYtBK8SQ6>%su zPeYRZC?uriZksdM(tgXyK}+ zG()di2&3j_MQP~lE`)PNAnD*BDOAKT`3@$tq`KPQ$H#}p9(Lsges&wk$fLUP61k1bsDPL1`s&D<(wpFV>#-K|XW z{Haq0|ln&Ge2d8aPPI*H4i!XmU8HU&DDoLmXm-AxXNfP1=D>;RMN02=%i1=b@$qqQafOg#sq~WHF%V6R{BkZkTF4hyRQMJa z7DDdH2f0GN^mvJM%N)zIGN9nAofyMVIk!ScUjT9P;Is0&x!$^aRi3@AD>aVDUOv8n^`yr2H(? z)YHqBE!=bVLX)rEHo1IP(R3aixy*q24-d&75tEMhP4se4ZluOK3;2GJCTx^*GalX8 zfc~V-`N<(?=SpTG4AA@G7(eLJT8V4C@#9aN46Kxs4QZ*?I_rH^w3&-$o!a(0?(cc~ zY~0_Jrf=L1)c3GuM;Z*cHH_y=N@ElBQCN|;E|y=@VkqozxTB+ke4x-zu>hZ>{r&qK z=+@dDG4P_Ht+Bq>4UjB{?}cnDqH8hk8N1*o?tKdAOcF91Hf5 z{pNK^amvPac0+o&`@)lb!kChkw)uD}xu!wdqj>M&UYM;>@P?X)wymzN?&z3B2FxRF zMPfLDxA3d7va%LP2@7Dp*f_c2lwJ>v`v2+xE8sjf4*!y zAT86W!0Apg)JebJzoL0_(C~gdA~CM6=+Dt>E_E;Xw6?aAl*E2GUtg-Z8v5kZ5OP0C zPL_o5t$g9F-jY8mjL$q@Olgz&-b z>AJyO@2xQGJlQj;<~CX5JZfnVZ}^l7;b%6swnh+BRzWeR*QN!5JzD%NRe!O8AC5^q z1axF?R1 zf7SMGt0PB#BEK^7TA5KOsZ)e`1#A~ORB*)n{u|!Kk%y!!QbUPFhAdy zrw2d+)kZ!>8z2Ap$K84R))9)%2A&h%yqk>+U-e8*^k}%z5R@((M-2@PGUrlv!yXrF z!>$~5R%a0 z962eEiqvi!Kgefr1Vy7Oii|6VF{K3(sl+w%(Qpq+c{j&0hs-ll;Ksob?=RJN(|z`E zcEPV?T0D2yCaS!=oQ&2hP4AzZGkf-fgKJgYvfa5Ync5ec{%N=Ed2jybVCmR*9-Emf z%x~tLr$C3jR;|-JBgGNdpUAjH`%2>)XeIzV> zdEamQJusod7()EvU!_0($~tW@R`#7&ev6kv2ozK2r}oJ`5{L%&x-_&zeuIE0Ue<|#BQh{}5)jCJ<5 z1ff+=y)^|&^iAl|8fQ4eQj`@X(emiWOS1`~rb^3BN^V`Z*wvx*88h46vl*h!eCVb( z!vuDmkt{}|=M3-A8d)Z4TKY{kwRn1Or0jUmI2I`zM7HtKSv3#=@@N@63lcl^acuk7 zqp|J8*{BKe`3#<*j~Q!}*ZT@)BR-7w!Ha+7LK6e#mfMgg4A-_v`SS83RJ@my_J;`- zXPnmO}Eg6Y;tGIyBNHAg@<2FlgFYK%*U3c2Q8C$r&?hCx>V{FL7|jVOW=jY)2C@dg-hx zTUWUKky6ZOgjvIsG#3)v4b8PL{?4==MnAzFvVg0rtCRPjF)m1GX*|N^IHf;g#)V(Y zN=wf?mr9=8HTwMd^A!~d<~8aj2mD3a@0Qo4Hgq@5%fvj1H{|VvvZcMNi^fWMlV=Wg zF%rFVg%Zqgy2uX47DLRzWYHxhn}j4n3zW4yt#8*p|vkW`*ELNks}}X$LBMxQ8;wj zAosTAEU!+E6s?`x*1t{j%eC0z<|R)GXj-35pFJ3q9h;EAEieDaJ!0cwL`6l-S7x@i zi$C6&JY;tAB5w{gWL zto>>@VCJ$mZKu|y{O+La7w7NLo1BXMzCtY#hS_t)rEcr5XLfdST5rRF&%@6uTJRG` zGU8%mRnz3&AoE2hM$-j@>gwuC*QAx>;^O?!297o?Z{B4~m*yam@29cwtmBP*sL1VDFOWIsV)1`))L$^HlZxyUrs*I?qn! z^uFsn|5@4e`z~v0v8_hxCi!?H2J+M~KfY5&@J2sVGkoxh$PccK8~vm!$8qV0OeWiU zY_56LnQ1}ah@~s{bUA(0O8IAuyaIGfQwJF5(1#DrbUg2LpL~P@52>M1PGUvcVBAY} zZu)6_J>N_2_90F%{SJwYM7K^5@^#eW){#w@wUIO<}W2S)O z-pV!_)s)h_T39?h>GKZPTU-D8C9wxlQ5)rp-@Lg=^BG1vB{yui5$V)r=G}x-**MRO z|8j~E&d`+Z6o7YI^!f3!&wVdAc!mq1!?5;(imPesc+mwzpYhJt9h7C_E&f^>V9UMlVmVclu z-P?J#lI?_l6;&IK?fdkrnbP5d0|W4;51RCZ|Es0DE|ZGsqY3^9a%3;$=JCy!I4p9|0O-*kQIW70MDEUkC3!Ub zl~pyzC((P+*PdcJNim8{glv30n$f;~9tImh^l*{tnOIsDVVjjSH0UqumvjCghcY_r zXk;R?5ptYQglnKRT0xHEAf+q-4||Jc{cUWl*~9f&yO-#yejnatfld!r5(L8vsnpa| zEPO?&fzShn=;{;4I>AgFkBNzilQr=|s7hWvoviGyU+9`iQxb+V)=1q%^8PR(K@1>> zhBf@#jL~HVJ7>r5zWa`9WAwjmLbAII?S_XUf50cY(6L%a&Z8ObL`C$p=q%^#%Qwck zKzPuQq%Djuh6B=v1IBWL#&X}R|J)Y(_7?xbSyW9=O!Q1j!IPp0Ng~};s=XXOpfUCg z^>!1Ma5+^^Mn-1V#a;Nk9uFcSgs?Y9!@Gnm?f21-I2GE#;H#M``2LH+jw{jncHHAZ zYN}KE&LqrrFvNU@%TDuVZbD<<7_b!GG+;xm8tEfy9#lFEBV$jkm`Jv5ply61c;_>> zs|oP0rGSfQ5!r&;)rbaSj}JsU_d>tq&YlhEA5%mkA$l%dpVsz0f*n8Z8ubv@VxNwdeQ`U*04?MYrYz?rY zJWW{&5_JP~8=ZSDPSaG#$>A{YJQ{{Vs|Y7&Rg{4o5(0Xxc$4FJGpO~IU`^(?>)t8Z zA*-JX3JT~xqlx?q*&y4ax0cFl)XT^qjj<~1tMc`HDAwYVwmo6(i-UEuT6CptzMafP z4@y7K9%J-QO~E^{2%7;e79eZeQoOqR9py?0Zo1-n7$HsvH+#5C({5*HXSBGfir+!E z1uk*V(^YJjT>nKL{H3Vdzf8l)5+R+XwJ5OkWbs^%hyuduf*@{usajVcbM<< z@)ULLElLJw+nVn)d3yVp;!07)KtG9l;w5WD3_CtP6e(RKm!v~4aaZB{FU0bx46G_2 zMvaeykL4?#umq!1PjB<`8*Ny?91$3=9v$s>eY9nbys7Cd`{CS(#IvmrUY(QQhdA$2 z%A%1~*L)?M=V~YF6#k+X(BHJ%Ec&R~ffh$+lMT8~A+u-C<{fb2_x0ShvZ_);39Hw? z!!>lD;^OVet6lHZc`VT8upUlSv7U&v+GR`E_k82ZFVy(+qgB;)L6*`_1%A-zs=dMx z(NvPtz|_aVhY8Lb-cG(>cH4V@24fYs1syJ;q*_Er|JfYgU zpTk#Bf4h*-9GmQkOG|0i#^I+&Ua%Lusja;|g!1a8Dm%UFY~d*%rK)eB#+tePfT3K^ zj_>kYPS(~3#CN4G<6PI%d2sJ?Q&*E^YMxfX`>=vl)H`4Aoa_&2Y_Ac&M)dh8j z)8p2x5pmCZ(0b_bGt<*68lGtVYupd+53SRnd$4IX4R|*i&J=&(690WV_M< zef|R{EAN;r*z+hZ3gfKWa&<=a3v*t1YPr1#>pJ5=UhWe`$>KWwJh%Fv9Y=pwdwFIFT_#)llt=K5kY=3t z4;to$M-p!vOC||OmMd~qoXtrt>)LnkU{0ykb#JcDJ6-Sd@XwbuED9qvY-i`Q(lEB< zud0ow`T&}wwVv_GM_V~qo*Pl%U;Kl8`-bnBw)_ii3$@oRE-ul?%j!y%_uv-R^0ZFe z7H;^SUW znz*Qey5#;eM#q~N4h{CVw(AZ_>pT_>3i)-6~%_v^ao zR5{?x?;c+7&s$ZRc)B9#dxH*@wF})kk^r18$Sz$YUmc`BbhAvmq(ooh36n+ol*xR( zaHn&OV8LI1ZTaB)T}h4ft%V*|iUZllyYQ#{O6-&U2oCty@cLSt{pBpr3peK}?_H9o zDEacFu9KwQHM7S;4>eDpv5Tk)(Nx(hV{X+pQq-fF5Uu;^y_T++`fD*<__eyC%QefH zuGweCb(uEO+}+w`e*~1-^Se%6F>0RERe7oAJVytI77XPP&2G;B zqU$*lu#BGwV~pe>wH}_)Dbm zi;7+>U1YZD34@P+O+fO9TW&S}^suIH*kt;bmM*mJ+xqF29XUH?T_2cZlbJHsT+W;O z&yB$zya5uU%Imr7>X*=<-J>a-sf))NZNF8`L_GRGZlWdH&W5S_x8=ojr~5jL{sC{ ztxMHoLmG-450hCd;iJ&kdplQQpN#HAx~r0+dDM;atfc)x8U|Z;FI=>X2 zcYjzAP;yXzf2e9G4RT1`~l-kfO(J%+*YU&(2Vl9h?9Y zVZ6`}&qqV}e6)9WD>*qiK}J%78V;zgC@C#q7!nBNf1E)G5Pn*i00XMwFs?%sSye+kd9)?ZI}y(BG2ll4{BH~l3=EBkD1eDOoCzKshux)p zftj8seCoMN9bXQoeDn6y*JBNLVngEuU=?vR4V|z@=odf zgxUjnc?Ih`zZWlFtgL`6hCnM=I5rdB^SN}F(-|f01v!WT)|M%p2UY}pDroO~IMHo_ z!HdqBH@k;j(?7a86%G(j>#9F?47^=JcHDIiF(W{E!=#x2Hp1gP?Cb1@{lsZCJ~JAG zqa#SVLRhB+;p}Rze*Sz1kP1(DLpH;;v4|>bODCrG%-g4*-4SEs{QheU!Kd6)N~AXp zHk7BLnaHREcm(jT2!!QIh|YRrSYIZhmA~IR($O$TGT1x6g@hmWn9?QK<{v=%h_>Tj z0h`TuM!j_1=e6b5C~Et?+L5{5aaOAmO&%dU6_)Du8V2c) z@Q{doC7gSv^Y1eq0^10z2jwZ*FkDEKoIH#2gwZoGq{irem-iq~BBBT3Z$3A%8yqqL z*xpS;(Djh1jIwreo5ld9liUINz5#@49aD$)5%J;*PrETb*r6W2tr+KHzn8GcHu|8j z52sBCPuA5z!>^3zCJz33NWi;K%(-98`-m={nbhw;&uhk9;JT(2Gr#*T8uApHH!}0K z9qR*!`rllSPx|GH=*}+oSOpK7LP3m|U^X^Ok6yN&jZF*9oNxIEf;S(q+{DTkNxaK& zF9|zqp#>7Xd50|kF>dt&Qcb`b$&&>!0^_yo64mtWtlin8Xn9Vh9<+G(zz9LjkgxE` z)7U)6_cGko#9^)pJfkc%T2*8OdCn)^1oIBJhL6l^2Jd~C0DfNn#je9lP|Ycrp;X7* z4r;B{?tGhF-<1Q7jdxq6iowae-1#Nx#wWYLJ^+GV96l8dZ3E;Hk5`cQz(&AMGjo^1 z=d*-P%YzqNrT~Lk6N`p{#q_!!}ul`TDC6+-TmZLpGD?qh$s$$rN3Q$K!5m2fpjWOZTnHJVxv=919ESBcrvou%+D>%Rh2!vhtxGlHyS}CC2MsQNaKHh6-X$k6@L4h_jG&C2S2Mhnh zTm%+3ymIJivYVTmf&*-vE~mKa85_49ElZQujz1Y1YuB|(Oy|;< zz8FMV3ZuY__v!1y23mTcRON7@b&Fp1yac2bDUFLFA9IpJXU+hIyx%Q84~g2PQc>c^ zy)gDd-3L7&=!Ke+Uv z24pR5MtF1@gWMDMa>RkmLR&hV?#;KLje$s?v<1pDsmXxiDch{x$*`G1pmcTmjaJw+jNqj5Dt;lCkwk1_Lkf zb6g#!l?Z_?)|G;#k2}gXaS=y4+idNh2vAGA4uTg z*tj4l2To!oH8lZ|jxB@?;KdZd^u!@2;QE&0Ic!iHOA(a z_W9b@_>r#DX;_UvIgchd2MB{8CqaoPZ&z2>=$v9jSP(-_x3$w@>kZ7oZS9%p?&-tC zu8rvN~|rBQC8WGSzVNg~|=n zUd;Y3*m(909yHd2n%f%aIU(I*5C6x7ca5Kglq{iNP)fsBl?_W_ND`tu!7Z8o?V%7= zPdW40*xqV>g)CVAaJPfF%_%=~E&M-3lw3IawW!Cd9K86yPLxcAvG7t)_WmajCBi>q z)N&5&RNYRv;n`45y6oXqwzn6j71VbX2s=;U*7Uv@X4@^Rv3%X5uTTeXXe9q5t23dS zjGEkDe3;WcyXg9pFhcY;rG-jy#1A%yNjnRiet13KJMs>JLk7+=vwI=PXoFkcw)^4; z!n`^&ca%-be;)rWzrr#wIHA;W$8Gw#U6A={ zYNS1N+u^^ED7nc?L!4-y?alQw*z`oEk{Ec3Zg;-WCjA+gy4u?XLp0Ylhb-{l)ReK& zZ>y0iXp`|m@HEF8ZyoZpYP^+lW|zv<%w3=X?2o-1{DadFZsfdY$MT;N!z#Ss_ zH@2lxU(ruob@n|(W}?$1&&?zX1WX9vUW+qMN`9}ssvB=Q1*XJndaCe=7}mZYd2pME z6!0mX@40fYh(!0>eZzL@!}P1Xyg0o2_QkL*#!qz$W4Ls^xtI^#%(gYxbqwMeAMbdU zl{eQG1ePKG`K6#(TZN$HCBr#jnx_Tlw_A~1_6J`y^Wf!37<%|9((72%4d1y2Pc3|R z9c@d=JhpPXQwXteyu-csMZvkfOj^wL{M1+Y2_J21;X2&}A&!IrP+>|n{YMk{4LDS^ zw@K))j1($Ao>!U0D_=8M4@j%(Y22E973Pux%x~Cc&zzb#Txfc=JztDZULt8^(^gF- z8QqNr989+kfBXqZ{ML&X)?4fb=7i3`HrqS(YG(kThY!9WP}tZm@0p(zsFAx#Ap%lTmdWdcMN@r!SkEjQXE><-ys= zZ+bFVdEVTy;#_ihgW5c{=RaKkif1GFu{wT)804NYeLynwqSpV$&pn7YGEVABSeanJ z0n+5L_mz8Dq@RGfjiP z;WqNWm{USyKNyPWsE8aQ=y2yKph1cizkJz@p3$ttmT|m z`!JO@%Vdhk3nLTED2y)4y;>}1Kaz#TI9wLCD)%>L@;Gv}6>lcEQ3C&X7=kVs9^CJH z(IQ?6)e))5DASH!nRpZPj$S1TcPimrG>Kn%WC8B@3g@dhRjr^Q92d3f5oO04GcVK! zZ)#X3xRBp}X#W1Q>Zwr@@>V|##8N#4Y3Ed(aw{qx&Mg6w!2pipvVBRAFC(Z@a{a=g zM_2c^NM>hBv{*SFt63jtU#2;e>yK=E=t7z6Uw=4$tOjOhPSJ&Jb_Qa+i$jfGBR$_7 z^*ZE;AcK`XK2X19emAz&a8`X-Ps8N8MZQ$@QoM0_Ao@}<(%d{$Qk9!iFQL~Edq}Xk zNKL62RS;Ciwwe9W9(mcc@mZjusI392K|c=gT%A8q5}icfE%K!j_1&avTNkE{_9X|4 z{wD(#gG7G+w6=}0@r7USiB=no{qhXO-2jW{;i!uu*3`ysjmrAQraL()CmRM)gQ+_f zSa)I-l|=4sujKnrR`za?KpNFlj(+w$%8zllzV!1EWhDmZ6CFQOUlqrM#EiOkI1++u~9zL-BJ6_AZ@!{g)j+na&b}dHAF;j2sU8szi)*PTqw2+9= zrs~M&hf4m;`om(uU`@=`up3#ZBk8zVnFxi8t;2di`R~ppRW2@_cbr3V2v!~S66@+_ zGUzEz)@n!7?OSRUS=*kcZHoHc3Oj#jb)|3M!P8}fzbT42YAG-8^jZyGVE!FWXUPh6 zRH~Q6C>kGhpMGarcQ4M$41le;goG8pqayF)^HBm1Uy01MELd9qJ0%`)I`-%PR7r)? zW8Zv=2pniL8^RDnS4eC0x$-$9yIIG=nZ;V)gA|Ro9sp>&MDF}A>m3|f6C6^-xterJ zkk1YUrq_avvwHEbT29Hi|4@|TJi%~_nT4}vloUiq8HG>*rOZC{5bJR%&DJ{pTssobpgH>QS<x$aiAS7bI>R3gABU)Sd)qn^=_{ZI)J3@(6qh~*7 z&(6Q_m@*S>U@z_+8z1)!)kSX`Iwt|!sTrv?IB0J~bZ+0IoCXgmZ!&7So5Z-uk zDacjTLztVA3j=6e-sNGN?=PTwcyBD%ADPL`l#@7sXOpb?&LE(ZD;{f&|Ye0Dg zE%(VGqSK`?!=7Q2BZM$wbyoB;8 zm|aqTSUGQF6R7lb=NLkdCBM=T5_=QvqC*G>{UnVlV8E;^ffhyBTrAQSo~ai+!{SV8 zAcUd<)x#W_&W(PE`HM1ms|MTy$`B(nOcASEgTSbCJ7%(_ zIUb%z`kh`nvR*Zr;V3H-snB1qfk+Au#vA$$A~AB=4Xr4c=q(ir=1X)=juf%|qz8hS zyjDF6ZmFzqv6yPL0A&oRc$XLRRz3kNHLF01`^zu&!A7IvB+t&iGg1~PF1T_g8n~<5 zP(zXW`KUZ)G}=(wv1oaR3JM_mNIVhgErD0puGYX2YW6Y=R6J0@E2b1zsM}bFyxsgK z=IG(@Fy0Y0P%4A@Ql5^Dlh2voql?I9}(R0<7POvc8Ff(rcDh8x;)IvyK&?A6LWL94Iaa< zS5DtA&qcF7%I+^LelWum`RH!$qx>=AoC7KV6^wk5D2Pc+G`l37{fCi!KPYL^;H8d- z4S){0^y@cut`onDqy5~ui|(*{+>SmBz&m$Gxgre}9o?F*JI@t`QXL~1G=K9<>zyK# zfW?!sU2II`$HA%ly{;iuXrp;Or?cB|M0vRDUCDM<~fM7aW^%wL&B zz)vEVeAN4 z4E`WeI6ZmS@X#b;qFG=77P$Y8$ugmAcm4(Ajz}6lI5hVGRc{feED5yQVzjWa%-@Q1 z@1Dmir1)#wd9s8sV?@;y905{wM~ z7OP+JRJ`Xnytfy}Xmj4sgYo98yavP6Yddj<&8ehYh4YQ`vx4a0Tg%NSL`;f3`U7YM7+u{If(VisTGrxc8 zKa5Ur!jVZ=gwdJ?CM3o5!S-D(Y&wuC~}>RvJ7!x6rw# z0+e|6{)r}s>SFfrzWSc^K}SO#R0>~xphQLo;HZGIn^)UbM|cc0s=UGjmaf0^GV?$! zEBP{{(iswaT<9d*;Lb@&S7_G|35)vf$VGjNyfX4z#h7{RFumTK9R!%-!X#lQ*XHF|Pme%MF(}J!9#h(d;UM8j}wr9eXKX9~0 zopr0eVJ5ts*sqVV8qmha3N0^N1|p_ioSdBr`9Nv~MPQU$qoN!c6}1K;0pL?>hQlnu zyI5&gx**}$##WS8iv_R)P%jj)V%Ri)mPMWv&QSF3?Q24CKpmHWzBQJ~eDHEvozOk0t%!53eQNgOL@HAw+*YaP7?Vku%?+wp;@B zkERwv#el7;yGEWIH{yaj_vD$3AYKbiHY$VeRKxySO4Kz+YPJ1$yL#XzJW<`DgUl^j^i|zUy+AM@pe|l=nLQL3s zWsblRH4`)gE=zy@JRUMav_jy%9Rc%}P*AuCL0GV?g`so+hi!G%rFll?=4DiT0ui|x z4nrxas-TvRB@qqJk)E_zXE=q5F4W8Qa3au=FZ31_WxLJ))Mp;K`lP;6^p*fB*g3uR zWO+Bk!DgfE-U8l)Yw!4Z@h4lfNF3okkHq7(--yH<+BKUpjM9qry$8p(NfAqw_=I@JK~vP~11c+gyK8jz$J&kG5rGF&@BU za&DME9EXH|N0wQp;pk(@6ekm8H06Cy%-?o5DY-m3Cr>8N!Pwe2GCKNlU_^Z|GG0Zu z1Mgu9UEJ4$TjAUn`QcAf+L>VUpgOE?8z*4p26BMo`V=xl7kys%IM>S9#H2z+_kgx> zN_*`l^}+6d(C2lR3Dt{S)Hf;4KT#uA2ojz7(HdYi)K46lTT}tL+#)>J$z4TBcoxs? zqCZq6>~C&Z7*O%7Do`;jhal{}l{Hr@c4nz8OSLS}8=G%`u5E8smHr>s4lip`iNJmJ zS4GJ9&q;mU&0BTOVd8CpHEb^bGhXm0+2f-#IQN2HC@Zuer$g+7F9V8e@8|3=V>xfz zd91Pa#^vpm-&P36uYJ8YaHcw$nx<8f-uV0AmscrGoCc(1jjNevg^4om!t*XGY<4mb2{CyH^#$ap>$zfP) zpGjOs$;J3uFO@90!{(%?RPKI?$_Xl6D3uME%BdL?WW~K+&f}Po<%+~Xka*Hvptn&~ zp?`i;R606T6pg3jl zdKjSu@W}F+TKv_-a(kgS#}C@&|DSj!@1|4j`wrVx4NXq#I-3V+om%U+VxPx98816G zee&tuW*ZoRZ-;F}OP-=G?~|*H0Oi@oejUIokAv zW^jnG>#OIC!hYMFxP@2Mv+R_6od(|?XJh)qf15Ll*LP+)2dt4lI%Hawzbk0Cy7d4r zOZ>LUL0(qy*2L0fO9fN23){{vRBr9Q|HUr5Poc=`jJU$T(anGqPOI`3w%I&d3^5VpsF-~e$imn{BqWg%Zubk z?x%mHp6#rqd4!SUIR9~E>pIQgiu&yj zloDC-nQ{A^6$S!^A{*Z>VaXNcUFy)9fTGSk7iLvCGWm`(Dj^ z`)S(eKl%DERv*`{PlpDF1?0z!=gIqS-L!f}1EE%W*}N1RnJ^8DPl zbn0N^*Qwsl-p@;4qw|Yq^KyX5fUOC zowjbBUzJpdoPH1f>nFUdqV8&SYxH)J#kDz_|6$zHGDO=9vhCN%uT@&=$5797vo5l7 zh|p0BFgSX<>Fc+Ojwc!ddm8>SH{ph|3epw^)5z`ho%>!pJ(k#>(lB%m4)HqB^Spmi<{A^x=seLa8@q;T=BMbF zK4uhF`fN2aU!S6&=FIJIxb4nux;FF2xx=rzNcP~hZBp{Wi6;Tp306+7a1bqQqBZMMYzIXxUapqo6|yhZzPz8+CMxU+2uL#as_>?+Xsbgjnil#CNh>Wl_-6 zM&Hy+Uz2vh9qWI6SnFK2<2`jjd5)~znG=s?M|*Bo)uyFNDGOXIj1MnW&&=Hcb5vWY zR=g`wCgFm{T@oH8h+RJ zr|tV^;mZFmtS1q}wWO`)!qK)Sw+x0tjd@Y!XD5Yi|F}jrutH72UDsW_PBdnt-_3o| zA{kLpX=`3^Is~Ie!*$(q z?inlAdwsv4x7vT6eOD0{;<_}4(O z|FETe&b`n#Dn4^giNAXkeKqF=;)_FKbJigpv0iS@;>O>1wal@K^n+O5Irf>p@L&DG ztkd7GTm5|#G!B1{#LOuCJrW>}85hHQbjs(4jze#2uB-%EeC z`mdkTA0#(eHvCr>Iu!Sl_NF)g$zx&Ql6mvPe1G`7Q2(vV?jMA+?al5BU%>BheN|Tz zvV9$WeN8V1*o=YHh)Gjdca%ZJTCbfvXSMyq!LaGaS0$}Djw*;4l=G~C>2ABchr1Ax zx}kxPIvRx@O*oQ1S*y170PoK1c(8Kf#e%cLZAc}<89PC-`ClK8<7<*u?51v;jVOV| zAYu3HmZXIYVAH7ijZ}_=1P5Co&A!9@yT1;>A3XTJ!%-+Dd-ETELMivN!%dJ>(nt2_ z^iz&o<&ubwdp7UQAE{7WaT#Yd@ekGka9w-dXzuB_>_YGR@Yx}N*Pk348kz*)zV-fV zACnh9a{NkSuT}wf_F%cff*ul(LOIav;PFhiPH#QbOtu#*CJreJ>9&zw^RyfJ0H)zH2Rk2j~OHvEld+AN+b`d8oot>z2`|cjc zb*zy1eEQK#2$!(rW3@b!TUkm1OkX;Bx4e)-R9+W*qLzv~xU_TMjd( zbH8fgXP!ZZ*m&^2n7_GCng%Lt&artqwGZJlbeDo4cQcT+7qt`(O-;|!-^V9qW@hF# zl$XD1`Zizi6-NHU+5E;OW?sgJukd~zzWjBpFBWt2&H!j)ZUIf0{zH=>fsvN*!qB;$ zrMX^=epU1K_RQcxf4FxsS(sq(KFxAJ&{M7l2*aCtY%+@`8^RRZXlIw{?QxQvq!6-0y*~Ve#nTX zeZwx)U20wb{n| zKg~T8*STW~w-&?(|N6xe?(yXY>e9&ruzB^NkRjc({k^~%`+)qi0>2NL0{ZH2pZNVq z{yh?ZkHp^*uu4S`X9 zz-f+$b2#-L&*EQNHcX5TK%rt>AA-LGM*+Dx5FoaJWlV^(47i`_%NgOn!RZVyHCw*C zW)kY_u8jlXBh84q52y#kRT;Bs%!FC7ju@}s7Ool4@?VOdy?<)@|NPSxckcH>=DkpN zvPyT>tJ8gCK&Xu3kC1kEA>S{@QNL8S?IpJNL+z{Uj|{f21qT3v6mxE z(z|MWOw0c1?gdb#-VC~ULn?nZR6N>9r3CCbM>+?B))yfZmo%O8tUN(jT&Bkl&>878%t0t))e0e#M!x5Lf*K*W{? zWwg}7%?~QWrZr;FGAbp>W+1D025dMFIkm}K!vSL2rBF_?fA@fZc0kSUk$K-mV&>QW z&+fc01$YQJ7b@uZLN0l`6I&>g%j0k|VqcEa=I{IF<4OVYATMe$Et^U5IWeO#*puH$ z5aAsYfPF;sM88L_<`m#%Y-wAPt(QH#-D&m-^#p7|$ga*vA%b?DWS{QksnMqVi- ziH{wSPev)kJX172zmzTfQHp|H)o{=T7B zMqZu_(9KLjske>fO}i4qA!TDlRh2&=`J3qlaswgx)$~C-f@(vfh`@6Okcc4KI|KxT zq2EujE2#kMOm?anLdJt-n{#8rab{dMn@X5z*#mAl(}m~Z%Bd|-zi3FZ21INCoi{`X zGV8PR)%oM$voFr!#yHesu1$|jPZ7p)eoih_c4$)9Xjv;tL*$5LhnNmw8i@g~#z>d9g%GJv<^qi6HH<5d>XGf2$8A1SWw^5KJP z3b(1J8LQ?_4JEgst%Q@0^jJGdRRu{o51;Y=d}f9|@R&rSh9U@WUuBT)1Hg#>c)1(V zqkyyvu+Weivm!nVkh$d*(#sMz7Q3`=izp=8z?ccdn;mKXL(@th7p=zI&(|F|hg4SwiiXvOx%??m2oFFhh*;~`SPMjfVr z_i_q(DG*_bupE7q{ppEJ&kyCZB#@6|tEQo~e z0uG?2N~E*}NiOMi0nY3946$3Opp1VhY<=Z%NF)Dk2>!*{#%!7+7^$e37y+VofY;v-IgZ!sk_W>2 zTnrB#vqyuM)~m_~<&j*QOoRPvBROpn$!8SOZVB-e1K6I1mzgGM8U(o>o1)YWkkZnd zHYrVw*@H4O3M_08aLUpxX|(fd0S`QMrxz*RV$yd9fP8m>6j1W)_F%BP2J}+UJ73e!o(N@kr?hP!K+-G~vxiX#v9yScyvZJ=Wz}mm!>^BkT`*SI(&sgD#4D8IZfb^IZ*xrmAhW-Dp8knlM}-rCxFia9+hd5b-$OOpunA90Xlm3VQr{s0OQWnLc&Bq{$8QB zYP6JKCw1BFsE`Qr19F@+(TL?B4O(z_gZ5NRmuPNB%be|fSfbom>w1pX?PGEi~eZ*SDw3E@FiXh`?v-M!DK9OtfI8xC4so}~22CmLn!V;3>sWeNw zShQz@w;`-w!|xRVrUyP-vbIYmkvxxY$XO^gu7N6^isnzpFc@QlmrbAD;uZ1~?v#@aWsz0=B z&h3g=y~VnuZlAJ;0a>GdhK>aija&6=+}k9Yf=H_pwHRzP;9?bnr{UQ>HZ$1yU({6V zmXErC0CYFEp|LSnyUo1fU4`6RvUOT$Z=mGp zjonEyCM`qTBvvXgkJ%&gVU`=U`*l(8kui&H8@;|?iI!UDT)c_&Hd=%E&d_wTenrjP z$oJfn%RO^XhonG>Nt81VLE$9HI$g>fz<;J=4+=D-=TeBeXEa{AfL2T1Po*-WLsts{8W~=t37#(AWOx_wp4mmG9t(btKhYBeCcXM~R!eapX+JTY%D`V*5H{_CJ~_=E z^SGaUvViw812s`5b&T7P{6Ls)&sQJ1`eG0TA}c=T2t5033+flcboRM%r{@5wX=e4` zu^)QZn(cUpa0pX#UEjU*Bo`?y)V&9J!8TV5;sU zruN|F%{Jk9G3tPBE*ec!RB1xUC-eF({S?hI-(79vNduJs0jclxB_r5*I_}A9wa;}N zeH8h&tKFjhktnywxlWU*=C~Ra4lS71ww`=mT{~Qd+UxT>iL)-fO0kLVcpdeIlTR_O zig&>V;Z%2We0@129x)pneL= z@?rY>Nw|qfif{q6n({j16WBfcP(?cHqX}9lo1;NnkZ)vWrWpAf5f#BCN=VveJ3v&l zpFjCSg_&971Fn^LpR0pY><9l!rZC`qUorv>5<`>e=(Fh>-_hh^O}Aio?9}WF%?Q{l)W~9KHu0C; zPfAEg$jyP_Ij%Vy9><1%b8zP>^m0)njVngsVl}NR)K`ZcV1WBeBv_)E!jJE43q>uj z7S9$j5KIIMViS|#8MG)Sq zw7*cvO;|blKp}`x#F<9>K1o1!@yM%9a$wjP1Q6=qM};hQt5xWj(QT17s1{?tC~S+SH`WxEbk7X$?MIHg3s~=kx z4Z#DEd8i{0T6~WB`S5{*M|lih>yoWZ#~f+{D|;nnqaTAJMqDhyE}eE@k_ihRcA{C1 zjx>{bqZ4g15!iBMhdYvKd zO88-f7INFSzlWqPT?pCO0YuK5iTBOk4yh$6Le3QMw25C10Z%4@CEiDrIwpMRtW2ZR zpWQSmMDLUzVW}iM)Y?18_xNu5qzR&P8#=qfE3d%A*hkoZcN)moz*H5(FKF zw?;>i=W(ZOX8122l+7HfSthy*OwE8*Uf&Zpc^JK6j;oU@Y^_OhPNzwgaoTZp&&S!I zNLwSXQ8m&Br<)f&Y!-%|e|8gD4)mM4-3K*I$72!qKg)k&$pIMnZ!iS?KbScMFPCDE z^k4tEcrRct&!I)KzHDz1i3m>vZw!p-v7|3%ej{lgGG^=u==!%w-GvV~)EHcEds{ib zX#0-EM$0p#oe%Ur0eCj0Akxyv!pwZ|{>(uag%@7x@`dyM0MQ6xv@E82A!Gg&%Yr4}&ThxxvnIT^F~;8!(kUd^~WTuT$H&AUMu z(YV?qRM*eQ0uWSHDkRm)6*?yKeHmto29JG`tC;HP$G;0%knZBS6Yz0AAZa5l&4_`J zJK?uL*l@4?K`GO{;w(+Msd-!3a()vC8{Bki+Q9=u^DN;XznL*srV-eEu-s!}XHD(i zyUv)UOSDIJi4G1MUf-BH0|ze6-<8?o@x76S(ZRNB(KeToqXfOK8U&>oSo8PsJ&B&u zz|HkP>GXCq?TMbn1NXCJ6RexcZ3)3l2xRZc%Zp*kMksXn9T*i51!q0B{PQNKwK8CV z_X^D{ORg_(X|B}qyWnQ!K4xMdI$YKNG(Oz4?g5;NlNur041UlAOg#!@1nc?2cWtdF z>l-v#M--F^3O!<4<~0;j!BQ(OPYX%0JE==3aru6wn#K?w-W(;Jtnokhy8JjZ@H+X( z$j*!d*R2YvOHw*KT{QSN{Z||bK#+8$elY7cHr z!9B4=wnpl5!H%JYY`{29rnmhsR>dCjDkr05YoD?ryj{~aYGkfj=hyqztk$Ie{=Mxo zdC7v{tfbU5TXx_nt`_1sQ|=}QEXCYw+_R73R|88Mm z`PsW=Gvovz;CX5#18FU++crp!V`ZWVkXp5zQklYwWK z+_tOfI2u3}HN5}YS#EJbB4Sd)^{epNwG!X*-_m-b5=%RJ5(xT#AZW|lnugQ^vsmf; z)aBv2UuA}&vGY}Q^s1bOK^J4W_QLP*r6PdWhNOLoM-|iqlt$Yqtp-?C7$qSezLTGM zJJ;*}ACiJ_*qc9Up5E`JME3g@*X)MYWD>H=G~tX=TtV9wlIyaEE2mApI`>J9<{^(>42U-aG?9*_T~O zn^{;6FUhkv$AX(k+KfK7Ugj05Lhe2v>`&6Rjeb@kK7DQ0LHRK&xZ!z zeLXvs&ei?lubJ!FRr{djH_M()rAFq<8)K@?s_IoDbD|uJ$CM=Rn!o(K>5G~D?oA1n zbp!ga$*2EmL3u5^uv8cqHr()L7V^a|2*U~~Tb7WNwVo6diC)VVr@+?P)h48#!U)uL z8@=p4`<`R7Jdc|FRx2v~Rfe~j3IDU9jHHaWzumWZw*LJLXWl+;*#Kn=QimCvLNO>q zyGOs==;a(d5p!h4hqLxo_Z$=%)68PfKMHp=Y6M#vpq=XI06nh)+ifj+~B-=WGQ zK@K3YSXg=#)IrEps&91mDl+dGRh90eE&FP->Z^w`%4Q<~r5HR4wq-xl(IY)z4J;KFpUGZ+lWFxw7Czf!>WMKzWZJN%sCQjB?a4U$dq7Rk`ZUl2VzXyK(Lh2Xteg7+3OW&-20I>pZHdP(%uCd-EV0=gp zC#|8&%-I}~nq9=~JCV!=7x97VND`y4`LmPY6V_Pp59_+9|2IrB3=DgAP$%rMVQ3MN zCImuK$hYMDDDdMx-vgSO_(0dk1F5(xb=ve@Z*U0paqPDKvwT2vvY1=Sw|? zm7KfLd?6Q2uaQ!L=W$bz6oW0ZYc#U$%m9YY9h(CCe4u{8`{`Ex$>z?Idu8Yfyf~=t zZp%6=g=*6S?q+(&s=JaG>K6v1-B=t*OZ6)Hfl0%>Z*G>&@HT-_ein7yQrmZ;Heyn} zDCK}JyYD7-AOGnm0r$JZQ29S`h&PZyx-sd_VQ}E%at|I}L}iD8Ys+Vz*I~Wv;pSh& zD*PA~;!ssc6id?x@;qvB+g*2uN|9Butx1}2k6CHZ@Q)=U>1qd(r+-$OJ#OtLZ>-s>yz{bL5omqt|oRXG3wyl-uD&XaOOe+Mt^?1f>_G z)o-tKrwqzy+sp%Ur3=pR?nX9g2Qq@TI=XzDRDZc@g=mcX^^)+0lNIIlgQ3Yyx+|XS zQToMpUAmS#R_NM$8O==Yu&pX9XLtF}9eSHI7GAmJKJhwehlP$*C~NsMvq?YwL;FVB z+^5>a5Y}dlk5Pw|e&pGawSXD}Pi^5W2XC9eC8fiRF45tZ z@TjhoEJN$b?ED^8ECINZQc9AaB@6ln`4z-~)29)9A=I@}y2I3hFIwrJ?9=6+u6%P} zNS@{;bq`gNlPmDc{gJi2)*C4l>W9&v$#gZ(PS#AQ>bG|vu^;N9ZXT%C4qPsq-CZPW z;V`y#-x`ivV3%swvCmo4+Bi*^C$YN_oIb-xV~v&b!^Kh(;Lw2bHzK4R;dyj8^5}*N z&;ixq9@2enwSabk+!K>}s!-cg3gStp zt4-`5ap@KGWSAIE zwVEwx>0hy=CF8GqJJiL!R7Fh75A6dJBT-2S}T{cGf(H9`1Ei#dVAX`M;~13 z2d^;_o=HdHs)eju7?v|r6yIhYl)Eh_EZ>(5-x;tUj2YARKlLG^8E zpNwt+12d+-4lWC9$S#P=Nkcc9nX%te^q#ioaqhH*S($W&cnb5L&)|?Esq}R9DTYf@ zFQbK_GxOEsS<=4)08Efsrn=R++v8A!%U(6pYjNgAi3@rfJ|Ng2h#9`pRpVq09x{Vr zNnc-V^=Gy5k0rC#d*;aevM5}Nq51QUXHfB*^I2^W<7lOgF6<=NLcNx~`8f{sIBXts zz6(wlvg@gOzyA#P!r}?luu#Jk!n1hu&pMVf=kz!n8+YXb5~8cwd3hwCQ0oGfVM31J zG2M{88GY(0eHGnv60fYDO=-;>x#e?_?Mcp5bfW4BU!Q{N!rfs|#jd;LI8Hu3~Hg83nx72XcYpXeAoO5)gS%d$4~B&PFTcpeSOd+n_SS z6)I8~sEeLHF=s#^0NTEszs9TzAOvchLsDK=C|0zP3s0qJ#?a*?v@@W9hCODu&CzB7 znINJNT{$^35oXW#<{$SEn6t#&ub;J|w$0F|VJ9hfK>5wcxK7wuJpf?>!6r!f#sCR; z#~u^ikYb_ijmY;d4Ci1tcr5ujOgM=ZkGvKMr4f*WbZn6P)dBA#Vs`T z9r%T81Ok<=<}1$NO=fQ@|BK%b^1&TLX0(_#AF4UXj20?UU;0xhNytyV+n!AsMd(tN z;*NtTSxc$%voW*wYj(H|2Y>#pE8P}P)rw5F23wGk+Ei@-sA$Ytb5Ay%nQ8$qII%_D zw{BCuO7Q7-(4bQ% zG;I2T&4s%gAhXjS<31Kc+ak9__no~yvq`m?Rln3p$V-3ys91JRM#) zV^`W^R8Bz?%@H>ej++-T*5bIX&Ubc2_j^eN3ibQ}6xLT$7yAB!m`XDd*Res8Hl%)3 zDJbRbFjnC_)Wa_LzBxkoQNopglCz!k4t*;lj%27}9iv_M6a&58idUd3%I>=#R2}Ff zrpd@rVI@$3)*7vDTB!{bq#L!|U0G*(ff#Tku`?ol5V36eOsCx0t9u;2dBd^+!ju%5 zF_hf)YvYYk*CYdaPNH^rba340(nJK0n*iBB9Q-f~{oUBMgSu=9xc3X7%Z5p_1IGau zkC6R>kQo|cv}~%yn;0O33bhvVG)ttZ{Ls#f;|M!92}P|MV8qCadziHs>&fefkBp+F zqBs9{K;NAn`E5$#WT5sV zN&2b-@Q8%Kc`_N?$?>!9f9^HY%0L;NP#=WwpqLZpfq~%mcfZ**-OLeBPfHlEe@e~_ z!D6WOOO1|mR6>GDb{QHS6jIioo&JwkO_(ohtC_)Zd~AC3UYK&+$xau?=!8aq*m zM}A#|Y@*TysmS^@ubY=eZPH*r#)&i4hN6KvYBx{ogQtF-={ks5)kpsn zkiJ6GT2WC#QHf(7>^i&s!FEkVKHLBYP-mU9f8385Lp_~(_k`M%%MT3@cup;+Q+4Q$ zIqRe$a8G0_$i+Jdp~&N$D@@+*XMvi@7TqZWLfwq{wfe@mS!w$^Z6ARc=1%J5MYW4& zY5+&@GFiwuWayGuXEU}u&+D^}?9B6f2pA!mi>ibRc7+5_EL;-$9nD`LeVMAc7-&c6 zOF_(YV(s(Rm55(=7V^*m{^@!_5Zzw9`_{ zccLR=%81?aF+H7I+#F;rb|;w|Jg$Nf(-Um~8oFVHS2t}g7N-40=>ChJQ9`-cZutIe z?Eio9YYcY(CYuUH*Aa)6~8gbJ3SWUinWDYq^itOH4yjZnJ(5sGvXQezoZJ{|r_j`uDV`?U~A z?yzXQ%;fM9^4qJQ-Zw`h%AuYmIqkAY8&Fw^fhj02jNY9$R&s2W9)}ZQ@=y!wWT9%x zO0ve{SkR+{>d-zza$cs~hkGUK%2Kwh7|CH?b`1Am{A<^_U;gX+fBj4Lvm**|-<^N) z%h%g|^h}JZbU#(u^)I>8e0RoNzs~pi$I;JKc$fZT@#QT_t2{oF-?D1efMnRq$;q!$ z^LW4c;hbN$YgukmTFO{%pIc*F=18GB)5~P2^{;j3QI>qOC2RQU)2D|zIOZ~!%@)2~ zPV>jb+W(Qfb#TR@4lb2EfP*Ma86@dwMYT|D_)*ZBUM6aKjCH ztM^`fg1uw?k}ZXVn4xMob*H)veaxKB`@Id>ZP!lv6cXHu_P)knIRW44N{@XX@l2It=5EJ2;qn9q9a`mu2Ad=F&zZ}Rfr@?$Z9Y*-AdmQz>ee|KbbdQ@O_11 z=BfC8P`%ofp7!Qw;m`FxK$Zyfqnw^wzBTpw*wJ-JmX7Q$b#^IO!sXqj9HUy+%^Vmg7IH2~0i@tyFq51dN5->9^Wv z=l8&cpCK+ZFMt_~OYI1H3O{#KQy)-knc%1j8&}4vxyHy{;umYVs!WYhcmB>{W!ksk z0uX@KS`*mR0#I?Fl53}O2ktZKmgy^KB^Ob-AfOK}VF2(L))yXSxdWA7rmTiJ*htlH&=@X= zBI?BlIGnmW*6V7nz^A=Y8fOc^KIigm`@YZDa)b@fGp}|%fgmQUKJPVOU+kf;=U&G{ zdrzj1B=9e0y%3jgXB2i;CC~-iwELXYbVjmjqMS#1>7}APkVgQ4(YKe}LxF`+`<`P9 z%I7wUwc%QC0=!8OaEzn5@M;N5V2z|;D^Ujkk!w6PL6YKV%SCVz0!^@0-~rFFnt~{~ zrJ5XvrF(8(WGiBh{n4z~lbbIX62fY1Zy1QY8W(o}!PpIG2{wle)V@7?#wYg!v_g(% zQ?$LP(C2IA(RL2Hzjd3wNlO4&1CSZB-cQA4n;fXLE4a-lMFTpXaG# z?dwm(H#tPlPoM`M zrgR{b3uI{R7WGXcCu9huUKXY7S&CLLj9XkvGS!@}< zDOi_w;PYKA&p%vzL3?O--<&(t8FrvMWR}su?U#El}7RW@Upq@UQez{Rh-K|?d307d7@AQ zr7C~OqAt;P%xr=};(?+|2jD*6YuLsyqwehxsHwWIljCGIJW8;~8p-O^8#h)_aU;a9 z(+#GNEp{xd|AgC60?M~mkYY8hkU+z=QurMjxqav_IiH*N{zLWacrQ^$;#_3(^10-i z{$ca=%)`#U4uNqAdL0YeAIy1Ysi}iKtc>!n(l0bkEziibYB>MZ>i+bM9J7Q*8Pp>P zU z3$HMjST$AIO^_xaC%S)sM zl~{M&R?vUHS+4+o1c_|(w_!xMr|C=pcSa#01|PtZJwf(jGQMvD2B`!_xURL9^4N;i z97ZHol<%h22s9#^$;X2gR9^JRjtOV&(FI-H?S9mo)-v15!dR`jn2M08ecsKR<1|!s zefQZqks2iV{YcfjC8Zt(oq-IrIAn*Emt4C~8kw4!`t=ZPiMj~;7dyyXQJhbN&RfMWlv%bq%4TS790waqQl!i-5prr@%lniY%)Y;PRV1SlY z;TlIMHB&DG=s99nMLZxtdcBl`H^}$+gzOK2fC!{tnhl}nVuBI_5G~Xb&ap*a&r!9_ z?4^E})Oj6&CZkIM+i>^O4%-J=@K?y;ShhuP7YRoDM{Gt-63f~`cPNK&!vs1z$IX|u z;RMHHb>GRt`}C!t6NG~G5h^usB-jN6LS-ZQvj$taf{96&=p=f1#J$OgH`mU#E2TiG z8Mws~iN?1+Xw;F{`5YMYA$FTbtN=mMbNle&+g}x~MO{%r4R9BXIGRCP9&|JqYU&-Q zCG@e$OSpOu@2u@3l>6I)m4sPq18|Dea#E0v8|-b6Af#_2Tz|uNZ(jCXhF0~5k6uke zomB&+H1u5$-$pA0>SW=$x!U%}vkQ322nE=&^uw?5r=Iwv@=tV1p%QdDzmuXns2$EM3A9umK)W+9*WJr`|VUESXqACAMHY(!OoQ$T!J zT$5_ZO!`lU^1+?Mi*$v%N1oknot&(4P7~D5h)+kRk9vyx>ca|+V)%}|+|GHSZ&kIE z;IIsk`P_uGF73n+4!OHuj`LZ*iS4w~h7YT_qv0x)y9va+Ov28T@@ptC92K7}M9rR@ z@7SKGd8UO}%7F~bcWGZz@CF)|&CH?5mSEuA$oLKRHObi8vZ}w51lG=OFK7Bk{CvDn zO&9`R^g3@D#}vIOD2zB&%XO%VC6#Q&5mN3lOX##}m@j+ys)F5$z#gxg&pX9L7Y&HI zaab;oI+dZ8tc9OunXM}~GvenM3OpdCv)1!*Og@i;vGK*Wh-(sTIGipn@f(=>94GM5 zR{S`KxWyS}>?%F{J}WwcbFP8~3NPX$0 z=gM$u0U*Z_r(OTtHrA?vI^?wYtF;2#=!hsj_0e5Sy5a?BgFaCqGpVeDVpQMFUpQ{n z=AX9y+;>EPon)ts(Fdm#kDdZxC!32Rx-9$UD7-*w4a187oyZ5}<>g0yYUVhL%B^^r z%p`C*s5IFR#BsV-^alU^^8N9jW-p8o=(^xYg|LC=Coi?2O6HC>EO{nBbEb=cao4g3 z3j|5JQAOXOsLX}GW>I>C5Ru!gnx9~w-yzS|k4JvS5ln#o_FI?I9oaNE;+eHk|M5q; z9EOgo>IPmJ?OtvZuUBS>#r0By5Hc{tkI{6o2ZeSN4Q{V#TI8f@-5Cc?)%We|g0|RO zK6Q)Va=iW6BQ#EHmT21?#?56hM!|;j$6-Sv7i{GzW|%_j-^z1(YC9SObdf?O>sy6e2Yr*sMR(fH&CLgOM_phjnw=Q8`E+C6Y}vYiZhEMCe}|1@ub z`H0rM1>7b0n|f{py25j}aa`^C>v-5LWL%3ODW6uu&CGxa2}nkM=gz|cj%RynscM#) z1g>C97l?g5Z(OiCYrE??CP`&D*|}xHkuHGJSPI(Yu4%RafX+K(Z7 zSf8Ymwi}h(F^4cY4FEC-Dd8xg>@uU9JnqRC@cY>XF)Hf;41nCix)rx0dnU zXXm$oHDkey<+v%EEwHZ&z5rWq17!jhADae~EvpY*q}t9AQXpe@q6;%JQwMd;aY3tvCRAj zm@@8x`Wy!2x|0ZuLBN;1JGUUUXhucU{#0z_?79J4$qJq)B@HVwESlgcZPd=%jx^Q} zl=EY>DE9%m0H+puS(owDs_iU}CqA?(J`0;cOt_OVtqOSLWfU7_ZVY5lWd)fs73m-$ z`O#EH%U$g(n?mqRCqwP(jIt$VIL=cQgRL+9F{oY7sB;t1CW>1!kHo1=rQY?9SPacl zlIDMI^J>IfXWqeA&-J2Qi>)cehh_xjcm4`!`oMQ1;}6Mzj^UHMXR!$0n4EVaji`A zlaOswj^vo`dLa}7aeTUsi$2J_(!W&lws8K$*oSlb;S!0tapVhkxK3u)y6gSHuA9)m zaI)@UE2k@SJzeFtmOZFH?0JAq%Cy7m_12!3RUiMX^51W1u|%ejy1TEI^F@v|=wWvF zu2loTi7W$-tQ?nVf|q~S-iu4037zD4I6ck(H_v-eHLwe2M;WcO&Ye5=0NQpeDJIZs ztP+kLiMHWr^}~x)#W<=WTzs0U5tx36^U67L99-@C*a6!96yvzXI?bxUuDXY8*9ns` z?j!#vez)S)E-Z1?Tx7)TKlS4{_RvD(n_BjuN(;#Ek2*r|^rCYd19tt@^7VW#h_#tv zTz{5F`Wkv`nGQXG6{D^*l--ghq*xJ?cyMOk9I3-y0?Kc#8Zhsa*ZZgMM9qv6uFrdc z{ub2D1M8@v*OB99+IiROB(d~f|D%4`Wy%=$ga09An~j1HWujOjv+(2DgTmyrqvkPK zx}mJPOiK399#M$=-<^t&=Ev*UcZIg3N}w~8N~E;Hv5GdzpKQhSX2pvsL>j_3_ z!=N(uu-L2ftIKc6T^c^mX}<&x!55-Lr5OCb6quZz_+uP*Z*u~>-y027>OS^;g1jQM zGS-06v{4tacJJZ?9M0k{!y~Yaq%`pH%l@TvJ%`uE#bqb*KUi6?IAHBBHwTEPDV z?buVW-pK%>xUMNBhrJCcM7tCFIqpAR+f%ScL`0+owbuP^?;fr^1q@tEo9toJ0WHi# zsaqWL;^`-pl2YNN-}4D zky@8cOifEXKAA^Zaw|*ow6?Z*ho|&#-bHhi*m=5MuvHV6>2&pUcu&c6BLdT!tkx$Qd;Y||Hobtc27-k zZ7ERbkF%vj<~g2~xk!gVFOS{8(BaJsSYs5SHY-WK{j=;d4%4*D^Ke>Amb~BEHR|rh z+9SOU1L&A_CcBlLEQHj;8hMBoP{Argb1%nGiATei$U&zxd0)(1DwM#}WxaZbU{-Ci z8(ISUlXY_7AHmt25qK|%(CeqhHQ#t zFZjr^n$=QGT(dtJMD`S+6Gd>HbOVsP`U?QuCsI zTO_a~7#xAf^;_f!&ALs^hzJ|Huh9`Mli;1bQf&)HWs$s){>i5|_OJvP95>2$q$;3` zh(8o(04!gdB8k*y$VEzm7uCR1D{BeV%`Ln}JqszTfqNRQB@j&5WsZp0!{_D=EK+>{ zBQgOl@ME9^X;BnYYj&nwUcdkwGBLAZ_^3)sNg_gCm&1#ps}V9e6|7Cn7raztdrslq z_`l=aZ&iMam&z29r^+%z9-zMtCHJhtHniL%{f47LWpF&e@tbBNS=fLe62bg;qsh+SG`gRCjUod4V($yV{y?QkeQ8=8i?Z zqE~zI4*Q`%A&7{qHmZrkdU$jy!{+fzIW-4pnmd+SvVkv1H?&djJJ_ot<^=#tiEuUw z80i&6drmP199!r0*R)R4p^>?blBN3ac}G#$zX~owr2F)g;rfj)QbsDsn_j9&Y1TQYyHtA)^$NI?UaLYD*F#FTD)P9W8@A=p|ZlNKGC4NicP8^Vc5^PigE%qOEaNSiGSe;xcfS0^>MlE4L^3fL_C^0f zpA`CkWup1w049+*xysb{jjauXHOE@bd8d5Guz{NvBA;nZ;uJW`eqSxe7Jp(t=%%9g zC3TQvZ(~HeV9z+(ZBmiT4O_;MnD^kmjv$s4laQ6Cg~5bO2vs|>ORd>qJrUa(U0>Ky zV>agkPLpspuFgsuHNK>r;W`GjcN5BLzu6Psh4AzsEC?1NNu{8}Q-(6ukz%rZoJX9&5i5|vf|D75;RU=K zHicerW2@u0n{|jKBY@nVe0ic~2Kj0|d>kIF2X~}#u`MvzCB#SvG#O6+A75;4hTl+1 zH(O;biQ1?bf|vn(eYH&t%eox7ys&G$bMM2$SPf6fU}uzbUB?6ZB4Flu!Pj8p7YMuo zQb9v=7%_G7rpNQxbWtL*pbMLpbXt;9gf!WcyT$z+d;Xy?__%rsc3{8N17|?yoZ9$e zeNtPm$15z@6b(iP`qRO!@WbUa;Nc$xR645Yk%qpIAS40Jvb; zH`lBwgik#{Df(>@qX(RT--wpX$doA@>(^$6OF=iuWi4$+DAsBgCUYFVYcYzB9*I@h zOkALRQc9J938)LGQB97}I#_*b7=U0jX>`zuM;s(W}V?f=`xo#){*N@rs`7 zYRsO(0lJ8`zwFE(C%i~J_J2#}@1r{VA?xMq)84?DMi~PBAMziXycPVS8JyNTE$T!| z76KN;kH~l+gKv>65-nwF&goSfE5NR;FjaI+%K6-~N4=b5ZXmsLER$SQr_%EJ>W_m- zRRohL`qlkgF!}$@GhglE82ebUh5j1v;D;tumJ_sQeL6LY`tb2r!|hS|0ttsJY5trC z8*9>2FOt2hP)&I@ubU@+<+RT6a@uQ_G^swy($3+?|8rF(o)D-}{J+?H52&iobbXYJ z6P-*f5W9P47lTLzGjIv5ZZe5(QAU}A5^G+I6 z+9@qW>uNg9cO)3DAp^%Yl}yYrbOe-zPFW;O0WCBw*D&sLv*(d{CST9`88^Qz&OY{# zMQIo!FdMl}%0eRor&<9mQ6~f)-u`}x4pu!di8sN7plukG_;5sud#)^Pf)}|tRS_>O z;!%4WGvne1_uyIksRGI#3G1+5S<$ajYB_&Q!q?K4z`*R{8X;P+f^Nk@l1;297J~Yo z8E=05GHq?NqJ6tL>lZ_pYn_M%oCf?@1&>(4{YT**p7`yQvF=xiGKqg7Ooe>#5KTGEU;6~*t!ZrKPi)MWf zlV`2fxhR3+xC1u7<6wVR_>GM=KsOtMS>F+~WO^&Odq!f5!j2}uKBVKg1EDLCek^RR z<>_s0ZDh92mS&>zp@aMc<8u58J7f=IU-D48!vfZ3LpV-6nMEV0ccpEw$0HXoVd7c& zo89!gao8DRM`fXZ2AnVea5wCBB2mojPj9big;>$v!a`XtZP&kMd`tWeB90EM)lDpQ z2{X&ar`MWTz=-Q#>5&4friq0ccj?bO3y3G;D;_I_c?s$S*@eW)trs{Qj+}nh%hU@8 zGNi~8{o(45A?Vjd+6N8?X68L)mGNqC9$hd&SfI$b3!R;!eA`tts@ca%tgf^RJ=^eg zCdz9f5S&noKFsSS1z>@Ljnj;WzoUUjhQWp@4!~&Gs=l!&ccePEh-$_p(u5?bmko** z&1*;BthFI<{>K+eC6u(r7D==+`N%AmPFKO}-_AK!I?lNC^})la*1~{OiP8Hw9g&TL zcgd<%t88IC;d8}yJVBlgX%aTBle||cS=Qi;%rg11X8!PptZg(M&l-t(L+YBO`XEQ8 zw7F}gvx));@=3rAb`cxpv+9ExEXb+Rxe?)A>cfRLC|F z7a3SuvWrD&n|oxazk~%a693Yf>EPQ^_P*L@F!%=`t>RBY8R$Pnw&^iOp@fi$*BU7* z)U1Y*Aae6JkImdEgKwLrTb@Jkhq6*LNnIeNagE$l-6THwzrTi_Dwl&woVJ3o(?URD{B_Yq}n-7 zCr?RugiINT^+0+j@=sKGjXUXh+E8#)b$Nhk0gl%U;*1eb_bvNI-NH<6bsQHikykE0pP0)P|D*Eh^) z@PkI;yHP2iR_=kh%?}uX`9!A1K8!g#o~B@*7%1_EfRdlaIPza%1lK-|oA7eZIPkdh zK7WJ21SxtRwsID1#__6xly%*PJXukgu_)4o?49HGp$IC($1)OTjG^0q;Y1BFfiwX~ zUWjHcJI1B-1>h4_HbLNcr2`hh_lHBu-&u=M?xwT_SW*&cbx!7^2qan<3IcLC2QtPS z88fE?RilgF!;jezP8p;m3RuqJ*6lM^ro({;RM<&0AQ8ix+G661y_KX1H2?6@%yYtt z8@v386Gf~8QnR7k16BrWpMr3I`N(b;V2-F8j3Gzr`-j?^@4H``GDxwNo&)8eZiE0l za_PI%-rUTqbgVJ#iF)ZPAGsPy&lM05qWYX2 zQhCP$XZDq)sz#hMl9z0T7VOpYL3Z@c8r;M_1>Lu9U@vqe&%?W7=5jdy2FgxQgG;By z0&-6+5J}hv?97IrGYnnwpA0LcJQcvu+hN}rDLdz$pvWTO5Qay0u=WHaX0p-60ZlFv;&|?n)x#_k{mAs za63x;$WLm#j6cl|4?^+H_<2?Sr(?+fy#Dv!2lIcze*EbK_~kLC`sJYQm)7{w8tfLh z-uz{zObvzqSGPvS-%mr=(e}|i(MFONC0%x;c#+NW9c!m!vrq{Vfz@|&W8DY!{R&m30cEF_7G7RTO2Rv%)r!m6SL!AC_C;Z6uP?j%*~}pB`)SKgh74(=8>R9 zRkFP7CVB!9kF`Rk&kxAYJhYGcupAQp8#alrr+O;R{_5g0zWU%mgm&7u5XTmvkdJc6 z^0G6-1hWGB%!5ct^8EpC$o;=5*s#e;QN#@y0D+DW4X7xrZ^a=F82bQh{>_vu0TJ)z z>XQ$?^7WjtmEO|{(atk=q3r3S1u`Ls|D&R#H&LkrkxLI96$ZvebH_FV#9BZC+7Opi z6`zsBeIR0M!&x1PQzhYq3B5xYO;AzlUUZs2`%K7yQ$kg94z7(Lr(?yHmF0+^p(rZS z&tW(JKLP#Qy&Odj&4QHpscK^T?$-;(IC&}Lx{i>0Al-K&F6Z{k_nC9x6l*jg*qRT@EcaGrjloU+iZ8Cl7>jQ!<-{ z;L6}8>&hOm-@ijq%C&qeogQq7%}!WiLMr-;z998fI6M^M_~8$2$c{TtIXIG&kkM;I zbgRT5^+Q5h@`l`QfH-AR0h3B@)8Q@ej{`s9-u9j(VZbo*2t~1g*h_@&=uUziL;L$p zwo@ww{*-yyoApY75TsPmA{3^8OePAh1Q@*qeXNeeCnS?8SiIdJxGoT^9U3FuF7-Js zf+t@;aLC_fVQy)Ox*c!*g5_Htl%{cn{v)ED=_YAr3nB{@=T5=^rb-00u-Z(u<@%HZc;rgBl5`eHWVTC zM1_Pt)&Aw(?)DEXVG)sAah4-*Dv%Kq9v`MGsei?vQqS6O zO57y=O()oaI={{=vTMa7`SR`+$U4uPU7ugyxdf}s!CF#!=NGPm^dhnH#9)6$c6Pp^7Dox={$S6k^B%9`F~?p&~pTOzAwrPJ@!zDR_JdkL`mY zP6`FL@97i|YJiLp$V&i_AX898U`EiN8uwc{sQ?+HHYQp12t@-MB6?9)fjIfd7zhao z5km@@Ru{JN3sRi6sO_Yy_N27bi%zwm#DR2A&TTjri}a&X6bp8myJr&tg4vq!K(-$5 zIfFby;8ugZI7-$xQF;j=joinlffFI99XGJO@h_mj-<*WRuo1FBm>4GAz_M>6j<_vS zG+ihASn;{iqIIVX@TDqAD@e~$ddZ|XL=wl|9wdpk0zzB!W;osm`H2DGSAHQKhdqaq2kwGIg4Aa*WN#_q!A8|kxs@4` zls%d=wF#KY6xg3giwbuYSCSdA3pOFyYsKXp?(-XooJ|P_ZsM5|1_TS$dSF@!M}VWR z-)6lqqQelL|I;0w2w)%3$5};g@2{i(#em9uf!-ydZY40UX=MZYc(^)uM2q+%Bz2RS zRBuzrPxiaIIbF!ulJ@xA=bJ0b)cgFci`zlbw$&rd4K7t{-LT@p%cFqOb%#A4nMtWb%|0 z6ck88fe!d!kp}xTiA2q88$Gm*Kh{fE{=0RsQ7zo1y!OmrRq z zLd?<@i5IJBu0b*_{muDR1^uY}6%s}M{Q3Q3CXYeF6vPO{if@->n+>GM05O4p79e6B z%!YEnG07~tPjyTTeYYhl|LTUu7eh<6)dH67Ez%wwkkK!bR= zp*{g4hSZkHcY+y`_J@{0Jg{R3uzHJ8Na#vwtsWyO4FX!X4KR^SAyF#Y1F9s)3e(2rih9)JYiO5`n=sVKkD&4DimmIsj;iC$c!3LP3O#L zPk0Hkngvl8hEu@auE9Vj8|`hNGKhqaP|cbWjSfX$b3E68oXm(0K}6L@hDmDx&*aTb za~CUv$O_`E`i;%{2AIl1EQD{=L*o+?wo*J-&`&dKz%JDy!P*le(XwY~m14NG6n)At ze0A?A@uJhV=RmQ~;(%?n#@ETgRypZsAO@Z*fvwuo|8#Xj*HrSyy)SLRV8bi3|B?$< zBuq&{FiMMwZV6VQFri;4U5y90g`kQSDI$@OD^qtT2>#b-=%fDC5%dkm^(3c4t-8d= zB`7_GAvG_Bw>~sbHhy;AMw5}%VPOT{m`>{U780~47$5*^8*^n&yt-#mAcee< z2x|}7d&pIcAEJgZZ?2AsMZ^Q*I(%}Ec2jnE|rf=WxoyQ zC}OWgc?Kd;UVoN|tDE_}l|maSE+s@L<~}VKL`x!X9wK|R)@2W=74=}^q@<5}jeYWB zoKDXj*r3;j3?4-XRek(8<4v| z5ixA9t`@Ud1~A?`ZETnyf($Tv3Yj!6&*T6|*WHUnGWB?^N1ny-XzC{*cY-koWUegb zFCZQv>337zkd{$uXzENN`4Duk$nn)HoNEf0G@gQ0KBYcsKA=)}2uSx$>`GpTxTEv% zpC-sWo}b$lSB1AF2vZ%WiPmC% zY7}Rf0oLOtR;<*XQw`Rua&!5?zW1X9E^#jjo=Ti$IR-hRpsvS0%OO%B9^wFiq5)8v zKk6+;3>F=8tLIf!8N0fyiu;Bo)71X-z@`R(`K^#Xb1KpRda)}r9J_P{qGK`3W9lkT zX`|qn=G;MxyusR`BZMu3UdOlWU7aYB=xmU3`^0e1tEzCGsea`QCiI)#4#)>>@@J8z1Hfaih3jc11ot~0#yUj2G#Wf|3{ro< zI!W)|4R8v#97To%1|k+lVOKAU8GmL3v3u7^&%S%sx2P&%Gc7WFWi$?Ve&s;nRRNO} zD(^UkfzVczQ?HAxJUc2P?`K+rx_jVRcDv|4s{HrPi~vWJ;_TUxV0{bNQKHT*QJh*4 zAH-cJkR9jq?=P%Y&KZUC35nZVSPugPy$x$3E?-BN#{tQ&%l`+BvgRHXFMQ%*+nz6y z&TE7OEY&~?`zdbX3z#|~yX)*EETH&xfVV{u+$%9iJPYNd5>O6oU+(`iAezRUbD)&VuWS=o#`qOJRPLo_qDqjdK*FAAkMh$>(*f!EPEN8fG>WIB%10$zB6i<9 z`pD+@g|bu*-`7=_?PwZHN?PnP69U2(L=iOc5U)KJaXyg(Dhp*@kVC&h%pK_a1Enmyy=j!8P%b6r|(G(E}Nm zO~@Qsg}Dy4QnU_@iGA38x0yb_PStYy8q@}}jG`udKIldDTQ)6OQLTkD*z`SI4R8W{^~#?81mAJMs8!3}%o=kIj>2)g2( zeGXH7dwpaRbZMSPdd+CO4rB#pyNGu-$Zl% zs#2M?(oVYv?aXU>^)8CA#eIN3ln|1|7_&turK+r(y0Q2_iT5HN+0rafsJbZ1sHv(s zdFh^Je!bv6Mc4l)z;FX?-e1*2fuSRTMcV_bqtv9bByb`eQObMYpL?hD+D#v|j9}F$_ zcGG3VbOr`B`>l%!*6c33p{kK&Inru}Hx#_B?*Fl?bbYUV-nP< z6wI$Zj;S#E?XtZ_^-G4jUtE4!`Tk1t9bRykag+?V^{U={VQ!cDrTA%Z>-O*1FYJpK z=3B<{i?b|v_A{T#V(4&edNgRSl-ucnP&Hmm$wg<=IVby~2VpvihW0APn>9j(S(@EZ z>V77_KL~5B&fY2$uwA$%yTScxO?HXEC*$Fi1+MSUQS@G0GmbS39 zX8IA)hRu&N>S|XB+N9OB)<+boEn$hM5zSiG|7uOFqm!eR;vsIqrE$;jGG>;WY|4K( z(@4E##^b^@)ouDFRK>fl0##ueE_b!7L(2E_2d1AiQsy3UcF32W<$Uv3*Syiz>Vtaz zF0=dgjfLqj*3rMKiKZB$lZRW*30AjiO7DLN_Rf7> zIYm+B*+BpsNDVn|4%hET_n7ftmJ~OzUK5sZTc={$WgO-D& zMYfCF)$12}*%ji}De{tObR@S^jzNPoGUX|zU}L3Np+mtmT-1&s6>6eK!IJ`%C+&e2RPDuk7#lwd{$ZTSi9;W0rU; zQ~&BDI*hH+VJIi2MhoxfkHy`$nSZ#WlQ<1sde4BS$GhjoKWOV2Yo%C@T(YiR*`}hG zW~&rieLW}6$8sLxy|Kbe_JT{Nm!o#bQY^OPLSNa9UZd1dq{*ZQbWZpE6p*sdQ-M!*qT25`>n7n0XY7PF+#bQDTojpLxNteKVeyl9>JhvPe zux?dx7M8V9-1jasyKt~qc6j~2dBnao z_j8_AG;D546)xg8Gvjy2ufo8o5Q0;2?#}Wa@@9AA z6_`&lZthzeubHM$p__GFN0}RoPuUSY|BMXf^Zn}orukwqoa(k@SuqUt*{*s(unoi*15Zn@bX-!X6lj3UGQwp*bY z&5ZK>w3^|w!Txt;8>ZEi4vw(<;{E)p2&MBMTm1$zlxO$rx?QemUmx?LrYu9H!raoV z2!HYBKIiPXA1hd9Z-?KxzIRC~)5QdNP&9~LC;l>_Zp)e-dpSFIMy-jl-}d~?-YbXi{`j|le)oOB_h;}v%i-5$mf5PB%vbUYt#8qL(J_> z`&Nx<-L-rz354GI3Cka!Zti!2Ra#}Gn)yFM_utN&$hQjv9!K= za`xi_&(-f$+E;q)9h@WEjSrd^v6dDGhJ|@nBp;SfcdE>9wX~8_44<~EZP(V^Kx^M} zg)3gI#pQ(uialhUC8aoH#hTv5N!E%6=y0V-W*co;rNtL05y77%Uah>zU$;*@Ww7&h zoMYa;^Rni0jJ{NPY9Y@Gc}>11ZI?H48g`1zQ!YZV2FMPYhukD}ps8VuE$eU74k&g?E9##CBoOZZODmKJ#MT(EST*S4z z^@j@k-Z(j^n&*GJAf3-&LF|sXX0!c%h4jM(SNr_5v)c`)>vXREjXWe2<><{!1SGiX zhnq4a@5RaF*`>AIUh$$#bUTAz;=G`wW=UMBW?RbfL4}m&(^_-*Rz!EX-l$EkuxTz1 zNM)*Y`UwR1QU}&{#4ehYKd1h1Pxu~lTfd;q(y9n8S1`gF?1`8oYS@sfBX;~r@@ioN z&AsNmb3H4RIb(0dSa5snRpwpa!)sw?W+7){zCPC1rbkn@Ke;85SzI#kt%=fW9+Y68 z6hi2;xw#K?VFowOs85kQ4g;Y0yX@?21Q=w;Xj=#S<-|e(1xLqzSG_T+mMFP)dy@iQ zW!7n1(A&#uYeVS-Owu`|+-7H=-TfjF4rgzE(94mOBYIvVik!}s2GAJ@hk}P7g!4@4 zU_W=>yvW#Ce&|xezO@+TEFU7I5bUa~vU1=uOD(3H4dn+l$)AXjm8lk57;jEaPQ+1R zOaD?r21kXF(bHFd24K4xxm04(B%$mIdL`F{>m!|Qi0wvFN+CFRcQJdPbPr9ODXSL3$xAmr9!WboJdzY6>4Hav*c9;TMybU#t*ZaD zm(n9IS-shHEEY=vszww8vvu*Pnt@{9yI)MkNG+O-NEKq$s#(3qQ-Xf`?Kh%F)!oDK#yCWS4xhvJv-a3) z5Z`~N{5bi{l65vtZ29B~PZXqhr=ofM%m7SVX4KVj0+$_0K zLtPqnt>HNEN0piy>g#_;VjhI0_IKH)8BzL_M3In;PaDvL-8JQ+;jF`j(W#McV}yrx z=&WH2luJ0YKc(#Ixv&w^xXHoMQ?x%bGn1&Y&ma|s6Tz%mvozrFPRw8zczw6gds1@X za7;NjJq9uLz}$`07GR%gF{1ro$PU4mMvz!dN#2i?$F8@y2_q+{k#puW8tnNQ3nH2) z8px*AUtFzTnprb0CP@}LSW-@W@p07lq$D8CdHJk3NxDd&BzKDpVBuJGuI|DKc zS8mo{51@?{MM$Us?#kYjaUki$ey1*iZa^wD7@ZGiaucxo2d?CPy5gTzOuv-pC?^6pVCM6KEtu4(gmLGl|0Gv z3{=*+NFFI@2&~F}g)o`z+XGPXkSH1<>>@ITP10l5UKspMq;+}#Og-+*S&1Yublq4N zoeRGsEbgHWn#9q`H9X?q?)qb0_()}VlZBOYwL*>?lrBt8o!TyE17-lF%<83&LSY5v zSvF`Il}{cy4EuV+q9EZp!ecSSoDRa+ZNqQYjvkdB{ovKb zOj*!7`ChM|PFP8}tG_3=6NnJJDa;uo&Wx`-!f_xO=FXqrY5NRig-{@C*2@_ehXnG$ zaNQ`Lr5D|nA|Ui+mN3@E%e zDwS1K^j&6B61V^FS-NLNM?q+f=o24Yd3a5br* z?u;FhhXb(>VtdJtbdDpT^_PPNJNZKdO($^*kbd8Mt@5ty0-cIp03j7lJIOKYS% zl3(t$`r@Cem!h-5;Z472CV>7jxdzz>Ds5)Bf=0ZRWv_MzlV8Q($W-FT-1 z5v8V<{e69F-nYkNHVPMY(53`JC})zn*W>;PJ){TMc48jVPyF?yq#ji-pf!p*j;Cal z3<0KLWl2HklO*suO{&& z?~h?!DC$5bt$_zMGLY5>lX4lzpNgdQOYX6}d-dNw!(m>|ZeeVqLDd}{9YglTSG=4* z49q!({2yC%i@ZeHwbQwC=b{SO>qP7?!ZeUaF|iv^%a$a7jx_TXs;N6;vt_e&Wp4K9 z75p|V+0Q(3X0k}Ldc}mr)yV*RJ3Aq+6pXzNrr2ibxQK{`>&#AwvQu4_=+|uLQ`t4A zxk*Ksbj>gc4f|iTvS|jj8TO>qoXZ^8?8zMM>>3qt`zY&X_dMyU@3Jw;zW$a;u+*6E zgaOL@q|VMxW9>AotwiIBcPd6PyI*|lAnc$CcSWlwbszah;k30#Ic^Pa(9b$`iws`V(j5OX;rz_;df)o-)0F!o zQp-4Q3DI(ylBZ9f@~s&o&_6^1H=iCoVs2)Z4|vXtntG6mbi2Z#9#PbR;)H2@`Ub*w z*kDx=?YF{a|4qn5#Np00)i^REFJ$)1%E~fs0<=?)WlOK`74~^5fu_4vXqb@2&YY}f zPfm$4Q=vZAtD;~K(J_6nExc4wDk42|n#ltXm*xC|{b3}-GDWipzujm1b{5$&h@_1{ zIetCO(xVy^$29k7g^5XULBqtv1Pi>QP`}&zb88zRdaHJ)yiGUHqBcXA>;TBVm|HGV zIiFN`&S~}_-ChLIpV47J(w{-uS+6_b+O=!LpYmjiDYz2SBYDI=vCdg*Kkb0Bxi>e( zKlNZgH`1b27&n6IViRFS>M-0yWD1Xvcf3uNS=9PU&JQktr#wRlnp09@S2Be9nV$4e zWyu;pjq#0wWBaxg#feSh;9p5}sC zL5&de-vo3aw(39O7SQI zH6ZB%3P;M13uB!rhT&is`V8ssk$aPx@(-@tHscfH(kY*f5AmOZ7TQVW%cU zLPALbxgAdiir1OuR#xEM$w!p#W1N7KHZJa3S;-bHhr`Kc=WHsblIT7givyd(Qw$9C z{w2FIFe{Cq%s*#j1zVxBhswSKW&Sxn#Aladz~-izv;Y;>8>H=qD#AYC6&ct&Y#_zoOmd@u4RoUwo2W~Q zLnXZ1=YR4I6xrA$<460x#_r+X_iQ-YjQPyr_4 zLxQtJD!2Y3ofkuHRluS2kVF-Rh$Hzr!n4gd&yy2AA%&o4_RMj6KZm%%WIgame3JJl z>!)l`;+DAdmb>g)7a_7HupOfa3`Za7vx~N>UiI`e#t11IAzETeiv3@B{-S^Pe%>ra zoz4#&D*n!r4_-Eo&fj96KI5aNH6?FsZi+}XU79JR$?+t3(kW_ul)FDP3!GTaUYED8 zjBN_c`R#7)wg8jVX%Ti)F{ha80rjQDt+w@sWrJwwe^~34?jsU3r^lY-k zsOUOjIIchl zdbaHMes{cg@8E3NM5DZu=Ve)Pj8q#g&KKrIwjv#OO>X8$gaub<7Kf~Ou`M%+e^pvX zfx27Y?rA6JhOx7zgG6ry@vP&QxUE)ol9QgG90q$QZ}|RIPql4>j*?-^N#WkTT*0%? zC!N;c-s@!0|5YR$Oc|YKP0xaK71;CZY3knf)v!8MjWs%z9A+|fp?eqy@bSz*;VhiV zGd*Te?t_vNzq{_6Sh)1d{xR+ou;tAfCu4KV;mU^Sij^zSxve?BwXdukvCA-v^?epv z>ZcKE5$&T8eKRaDmaE9t%iKbNzv*S02a+u}#}gox_bt>@=Xo^$R2eGA7_)5&8ZP-( zsaPYgDCDBoky9vFo+itUOB0^P92`Lfi=@V%;&C6|Gx%t2>aV54t_?dJJYFJKWb-hj z_y=*Py~sZJ$)|t3%*5RTB@8r|a!PUWJZn{qR4$0Srsy(+Q7-M*P?N(@S@ z)~|jgWQvWhBEG^1 z9knYWWQ{{QhTQ*)#OE=PIURI%He;ti?|ufYVuxaVOH?_!c3zb>-w7I2V(1J z_9eEZcU&-<=v8DL7EO$;&Q7NjDvX;}C%Jo_xH>SN+D)bInisd57A056w_zU5k?jkb zOc`!mGcgkyjx$a97sS^4aaSf9^s7K?#uP(c{4SLcXhd?ix*V-VBBU^`wCbqNB*h>g zWFnrt8e=jzTP0p}n`Vp&W`qv|Be)^4?{0DGO`o}56@r>EQMy)!E@Q9n-<^~sRTP+f z>GCT6OoPcXrJ7Q5+hUE*xT$8FWJ|HS0w>}i zj*Uwv8{v+o6}Q~V%rPX8h2mIU8kQx4&Y#^;E>kKLKM(F7-_#XXJt}Cks{Ve5Ohit0 z%VeH$>Q~xADj3{mvF)QvdQYZE)3vPr{k6o8I+!qSMx-#fSFDWopx0oTR%eg?xa{BD zS980GX>IAy7_mPZ(A;pk3lZ)wOM{8yEp36w z6y;)M7t8jc_n#PbjQ)1*<~2og{-_k?1L@iY)=3k;&cShT_hhfAI-A4%k?GAHC}^O` zFf8=ECbaTGf2%!4dA5Xu>=|3VBmM0SPGzGH7flw2-H$Dt9E1YelAGmdJ#Mnul)2Zu zQe4RC2(w09Cf+l5S`_vBM$@V`eIeE`o2Gh<&*U_1`N!hclYGiTxJgp1LvE>U2 zoys+(R;A8gFX!17gcZss!nrD`QMmTsj(^B|)MePYG7NQ@$;oM|$}z&-c4G%udOQh; zD>rS=v5c*Nedqcf-HQVP&K<*;;aex1%sDu=+e`#C0sm;>5Acor);Lm|w+V~vCSL|} z?~;P&&*~r@B{;GRiTVA}J|SB&1)G#NwFrdx_AD!2<1I|%13&Iu-V?j;H-~tAZ633h z8SZsf5ZC0>*kah2H+?)@5S^^PEiL6^IM3D8NWa=cI+(YWdA*GfANM3;CLmA0IcajV z9JHJ4Sx9#JyzuJCN2=Bk3p{IRPExAl7A30{`&>>sBdEeeJJ8u){Ecl!;z>&r&*1XF zzKVj~UX!8U>aQji21O_AB$Te?DPa}7d)E%IX5>H}v|r1rs>IK}m|Xv4Z*=dTR_&yb z+pvFp)ucbu&yF{fPNw63=h(RVb1*jk$w?wF$=zm^{=X*jhu>p|E#jDe>`#iLFPN)O@y38d7S1>HVXZ}`_@CKyX}LOtsgzajl5GDTwGji$4RH`FmB}EgI-^n{soTt z9``LnQc_dd+Hq8FgpOAcw9j3=&z?PdhUnMtN_8v9uoXIkymDklM&*ToKVD{k;qCNjKc9&W|JH;`Rp^61+&`3u=@9|=0Oz5jIw+Gl*Y$)E z{3Bm)v`McSI$wdWqh~&F=zUNBUXb{t$*i7X2&rA7yKZp!Zu+meZ(SGHupVrz4NPs@ z{kCJ@>V5hbYMi2HXZj3Y<+SD;y3J&0Ygk;G(bl6~0+%m2wt@Y?J) zP)N0&Yuv4<*cXY`OaAtU>+xH2I%~jB5EbgWnz2S6nLywjPoo?${o$57(^=fRkZMJ! z=4_B7df1spuC;;pG^k{c=Un?-45gM&xrI6=d~{pOah0?GlzVic)EfOZ8?QM5K{fSX-4dAbQASOJ zn0Bf~wMZ$~Y9AapA2hy4QMj1^I<{lG)(##d39D~Hra=yoKzYBAd5t@Luq`GDv@h=b zAHV*#8xqOJ$L7pvr|nlI)2^W+QJ+s7;pS1YZUa(G>GkeR9msPGbYT*XH`FNsFq)fP zm6VjgcNcu(eP}nzc{pyT%Q*L)@XGl*U|mhavFX^&|3^iqlQ&1jc!9N1yi2zr%pRjQY*~zKzo0;~;8@&Icyushw$$}}Obo=XEX(%jq+J};`ymNLgy7bb(j_CuQ zV@U4LNB&#am7|lHdL%))!D)uzgu7g%(TzD9g8*HL(v0ow86@sC(-o@&2xEHPE5Ye~ zidx}_*jvzekQ|GtOt&@3>Gq8$_&wsBSFCJA;OIU>1Qu)wKSE6r zStCZ*_HK^Vs(!3=STLXu(h(tsi?1KW84==f6`}=J5~PM$5)6& zFe@w*ybYq>+l~QIYd}!|Ii*@bmqi2_1lP~!vOq^jYvk-YC5ze9 zC*DagH^SvZ;2|4IN!GqMc~RLdLMLD2SfxMO8Xnh8@ELe1Q>`-%_5UH zz1|Fi-l_4uC`xi1QmwcI>iZ^r-^4N+uL>5k%@UhO0WKe5v@DoEzgePw;DGH|nZo8G zczmU+!&Hf+v3MXf4lc?jIfprh5SnJ~EXsoYAsb^%stm;3cgw#gP&x9~Y?+}XnZoM2 zzEuA+W2Bt_{{2ZtBPpKK`Q7DV1~)9(4Qyc2*~r#1hly8ievq*F{NtnQJuu`FA`uk^ z;RGRfUK|WlB>&;7B*Fd&bOC}$*2kN#ZJR`Aga5$hYjHeS{)w~{p~CXssNRjxfbx+H z?#g>dcL5CEFUF?%#OY)*xXHi>ZhapTndTN2MRPD&n@55B3|2Cc_{Zk3h^VL*_mcVs zl3By2-rNKg#ObC_T{0&-+a*3KoimQ31NN+kY=^*cDkO;C1}5i~N!A92))96=(S0kV zmX>9`9m4ULUYG%X6&FyJbu9r>)|!mr{P1BpD5~jx2!duso9f2q)CmxWWKn!SP!4MDUgF>gDF7kKGc@HhUH)H3i`3 z*pe{Ln&WVfe(f7}EL9e;)1FqF&Ar@7SOjmRy8u1g!OS4q0}c+EB?Cj|oi*uVfYg&k zHP{BtkUTj_=_5jN65Lu`T)dAZPI5xJ)dZ>3>kZi(QCiap0PE{YJz(d}gQ^Ky4^I*a zW9FKF*BsA5(7Ka{er8MTfr|j0PGXvzN&v*ZxetA`;L|22oHDoHuGGti{&{e2cKlvv3(1N>}xabh*il^;wPiVTN+c zW$TuZUN`8V+n^oslmzFXJ5KHkbuB9;96Kd%r}QH`bHQ#@K`bD=Wpw4gd2P8Hlp=`+ z^4pw!JB)@o*$&7AU@r7Gk(3!FrN~H$P^zRki1RadX83`AxKoNj{=AT|)7iyt)>(ay zZvBn}5i)}j=faHcW_@Z&$dyr=T-L8|A1{1+??zt>##Q6cPEwQVQY5bD9O`x(8ghxi z5)%_bPB~VMH0#cgv=2GLa1%FxWPNCc7xiF=sjP8@+Sh{9o(b>Y28vkLY_~o)c_$8U z<B4!t~HnKgnl1(vV;Iz5C zM{NF+D<)N#n85w=oLo71UH!!Inu6j#O+xL}Y==l8Z|~l{x*tAa%&x;O?2_M;rwPDl zF$O^MD0;fz2{IogR+chg7CILRA0o*+K!9`7-JqGY|FzKMl^+<6qXYuDM?Wf!384$* z7#KVx_T^e>fcU~PXP~)x6jC?7AepUwAB54g4?AkS0j6tB4N|CetZdL}gdu}x%(z)< z+DQ32T2@#YNaRq4qvXP7d%NE0&?Dw^JNgJmypa}3*^y`q8C0f@~&RJ+DLpg#4w`~168M6 zLeomD*M531r>w3njGJc8wrsG4Lnf%U+T_oW` zj6fIHTh4X4q(fK2~LAmLBtfK=+^e7PmT@uOXC>8fZf2;P^d}@f}~cd63-Oe zvgH?A7KT{!6z;LDXt^n(Bg41c#t+V6Ig^Mt4n>Rc z4Mpyja&#lHBnYKngPtcDq%tG|fBp4W>IB-U@Toru147>aLUIMk5rOIkVD%?`>O5*U zQe`6!q@m*jGLvFck|?W8ib5c7lJx;o&&`EAi-z0Oa|?jEr??i0V5p`D*VfjOk{2-_ z$5QeQ|ZgBtwJ4 zJ^$C2uX@7QU)c1TgbY`hgp}5}HC~go!FrEes^StXpco#RRqNs;^E&isTE?fE2CSJ*32l_upVmAZ+*uuj?6xPih> z2MW8vNmjP!3eTW7HxP}4i~M$3Ng?2h@E+d(bP@oG$Yng^aB}x?p_S=?6}x_Xef|XS zngfHTv0=Ym(*#?!OoXwQ3xLA?Tpr2E$v2Z?85KS~uU}}V9AS$uX%Rq)5Xbm<+!?tjY2!Q`&9ct2axOdLWMs-St+hpF(8e1*8$;YkE&f57!?k0;V$WjrWIRe+J`oY;>PU zcfUVQApR$rVW#Kp&Gr7q$`IE~`tΞP~qJd!3K5d;utH^r2Lzwas)t#0Y8=1;+vS zHbhhyiYtBeajLczV9M?O$5vTJA6Ec z{#$m6++EHbM3yA8p*WBsPPl&W5R9JT_kGmE=iw(W)??zpEhn;IE+#zbA!zj0^gT+P zqS-7O#{ep7(&UySB?ipyK7efZvJYnrbIWF9bD=c3O^_6ILXZ~@_FBrPn- zu3ChIu2qT+7L4Er#pqs~7O(F%$GuvNoBcoAuhq=u2iW<7z0T)x^~9eYrUv+gwXRak zu6v4A;7utY!01WTlQ> zt$8A9IaG4ns%uJqCt&6HyuSHTEOxzInP%k(>&lzsF51lcq`I)){%4cS1dazU$2rs+ zA{QB>U=Cd%X@~`2HK?iRLG0H&@7bWYF)c}g@4M>99hQRZCFC_ zF;@J%q6snq^{nFhiHULCDz*_Acm}tvv7!RwMCDe--iG(m*>T_zI_7m=wCenp73QL? zRHb+7$Ip=?YN_Lml>9%z$V0v4!o=4gH6hWHQLVDc-{vs0GwP|uD{-IWv93odVk+7y z@7ZsZ!^g>o3nEw7z6e_%-zxnB>+@*dVm;CLs<7Uzi0r{p7?A1-y!X%ra7&-q#qKre zw>MAOTUE^)*#EC*j+N#u*yB7}H9wDO#c}%7mgvwleKq1)*+l!Gk8T~7pM8FB&$H#T z9Wc{NAg7{lDio<~^RT_xM-dijnHuU=Z6iP1t7JA+sAd`m1xDVAHrurHqVeT#FWl0 z&6)pf3u|CxXISCFUsePJp6hXuUb|pzXi3W%f9BEm4MMiMEiYnKd=@vup6|y`RXayC z->H6Xg?riK=0vH^Mvs5~P_!^}&rVm~44=h&$1Fp)Nxa&xb$A8eo#VmTS9HWfJ(sAa zY_&LKWSo~VS}VTCpYd?xQ>Dl2xn}Ryus+I{z3%fxZAf;CgX7cT<*zb-5LmJGJC%FY z+3IGCmo>@m+9fW1L~Y57WwDnc&5z&s^@F#9!#t1HBN?|g+}fz8F;hpt-ZdwE|JS97e{#d2Yfgt(mV--eH5 zW@R1zxUu6OQVJ5gb$OSBUitPdW8vkgNhZfuUFA4)XKuUB(lPt)*Pl9m68kV1vZOTg z?sg};&VLlf95<*83c4L_e%0Q7Ug@iYYx$Jt1@@?^HOX2S@UB}YzPNc$e=+BcEQ3Rh zEwSONs^TS!rSI$cELqmvIHhFvJeB08INX&&6W>0(Z}_|AR_o%+r|imxk0c-0`S`Bp zd%BhVJdZ=F%zyOS_a4$bysFi1H}8^C!Q731|NFN$434^@xs2kiC^P3 zt~}-vEwFj>lA+#?e_mM7zQVcV;L_;B#|0ffPsZ!n<1%N`<#6V0&o=OX<(wyw5Eu7- zbL{QQl?pL$&l&la9!U{jx$?!eaR19+aTXqHv5YLubw>`F;pBEMP)#KNy z@_74jc=_`rq1hjQ@W{xxd<1SD&FLsk?>Tz(=={$(eU5ul`Gr%R#2$X8=F=s|@DaD? zA4+{&4d3`>`-Z09$&`A+pr7ga+1-k`yeA!1n zkA(j(x5d}#^B?pPdSWX~ zR8z+4Kw)#WEC4vSTEG4M6ue#=P)GGDC@6T=1GkYYND8OPkx<+3kgZ#}XoKQKbv+kh zB%tp-wuSBds+JJj1H-J`QS*-wGn%J7))LUYUDUU9Xl(7#3ez!T)Q z8&IopO)4gH4k{p9qYUvFFx=IobIAJXb5u=FY5H}6y~R5Z(YlMMu*Ii)mU1syvLpeC zQ&O{zfKtgJ6_rH*ZC}mz<}vBGmj>CI?GQ#ehD_)O%D!K?G-KBcI*@KD0TuNN@1RKq z)j2_v74QREk+;_E%xaXgt)G5>VK;2tL{4iXGb{<`cWcO*88&4PcNdaR#0O;lQ)Ac7 zc)}CHcu}<@N0y^f?Fc)9AK>HQz8sE8cz5X!55L{~JNapND`vJQDv-Q-cXxN{sS^kt zsfhI9=gnunpD)pV@6=F78rQlTFm~DNwT6=nH_Y>9U_f6PqajRR-ahof!|DvzK~bk?+8ksmL7dzT?HKbkA5nfB$8Wv+`(bYkQ_`WMnjd(N`94FcvQJ z#b<3>w{G@=HLD5z?zzatDfHySZvykk?r^?0b=<_yj&8b+nyTs&*oNc>RYL~f)@M@R zg8e(!TYXrsjzhe}4HnWSPpFy5cHqKP5{`Z?XfH0)O|6M~VD{Z~ug^N9?2DwT(?p2c z1y2L2+-=w&w3w3lHSotMef28f^e8m;FOu$}nBw1(eIbyETG`u6+EYLvT;bj{My|m~alrfK>#*2Xq2e69!ad84@7tL);zfslGTRU&w zJPm_aYODNwVpV^Us5s@`SGaP)b61Uquc~-h5<7;BJm#3@-u=MQ;V-=A`bgrF7n;(s zU6<733}O~9S@Ib5rp5l+HgfF{BaF52BD4z`Pg-O=L0y3cGr!M8?ilnG&CTth2V{4i zUGHO&^hR_+1@BW%bIEDf_wMDS>(5HooAwZ9{3JPM|5NN+D;cfG)Kn2NmA)Mt>j8bd zb!76UE&HRx8#11k%t18`2Ul2qY_+Vpe+uZBV_06U$w%Ih^;(lq?>h7t_@z~mk&*jz zMj^uW7U&Bv*xm4;s81VcKm0)a%| zPgqcvsYD&{0MO(ypjj)(IOi?Q+piFs%kk{5p({l#wTHngsWy$Ng_PA3z?&O@M(}{! zqH_Y=tX$vP!w=+}?)Rx{CRi`IVc3cc` zG_5l%t$MWUWsBn6^{T3>AE5LnPVciX>V-y$u#nL2RBmK7!}4b_K~|Q=S1Rj$#zA3X z=oey!-h@1#bwoN%9Wuze-*zYFEn4y_WLe!jfEC|^E0ZC|HGhBq6jnuQMk7PvEuY?i z{P=VUu`81cw1%z34VY2!`(0|h1JjhMg|Rno+)z!;;s4=>XOdx(r)`EIQ|EPX>qYB*_$+a3ev z<08CD_Uh{Do`EOGR#ZYwO-(Oi75a{78^4n&>N}tt1jjxGTVg_4kMuJ0h6j2C-4@_g z7N46v?=p4&Oz-N&-lk9mfpl(c4=eCcX4jlQwpR+0|KWb-MzlL`{DV}q zTZyE=#*L=Fxmu@BpAO6;i$n}%tYGTz6NU&r_5lf*Nr=Ocu^!>0)L-Nq>1Ux z1lDM5Ir&p*4?%eLQc|%vAKw^+kKf)3w!AFoyr`BUA6({Fg-RI@wYiRP<^CZij?vfq|lwG7gM-QM~Hk0P?C)8Y5ViD5F z&;igiP_@z$H$p+GA- z=%t4C9)_N&Njrd>3lnm7IKb}BQneYy@;z94cJH1GYcYbT6uB~XVpk_OPV0hAQ75_w zBs%YNVM?r9_ZFdsCpoOcVR>E}yagbdKl=z5~nFt+3k(falR+ zRn^Bd>y~dlwwoM=6Ckd+M@~-8JC1Ck@8-d+YAo}yZCR^b^9O(gfe*Rc`!{nQRX5l zY5AxpxPh2T#Gp%n9r5kx=u2b}N)ACZjYda?@<0OQq}}!yOuRJWkM1JFPDxkvC9?B% zz`3RJ4)Ktbbn^$<%%>?nexIDCN2B|rwDUv;un`3b*Fl5amPXZK-jpKd?YOvNNdW-? z%QCT5YyMAr-yId@`E@%+6B0WrAOfP;3o2CvDMk<*8f^3;BZ5*4T{=mADuRGPM8E<9 zVxyxJX+~j25Co)3O&A2F3W$Ip+9yncRu63~z})#O%Zxp)qIafs*-?&Q*=lYYa*e-*|bq>8F02U+WJwz$h$4IyIdQcR|US zM~*)hqoSm==;p53u&P+!>2A}}(P`}sT(Kw2{>53g)#=_U5Y&*#~jI>htUj`|IgBHuQ2*PuIhdGBJ_F#2HJ z=WR8+FODG{WAh^K+*yrr(+MyDB9Y+8{l4!u6GXgTz+e`}%!*y#cglW@xKZRD^rYt} zc6|WT7uu_1yu7?b1p+kQqIu}hgI63(>J_BfobtI37c5wihj>A9kT6_TKTVt?QZX;D z1m?+M5gm^RZWN&S5|8E}K?pH10vUOCG^1=W4D+8?jl5$#a~bevFQ&icfkMYe@eqAh zFLXTf!OGvfwrZfOZZmPN(=jpDE{K0vK_v`dY94GJB8AfQBgr~0-9E4W6=urYXiH!TT?x(|#S%nuz8v4oq z6Ge+|SsNnjD+BH#UO@-*{$4OkkGEnXWNMiWO_hpDPnQ!DgJ!b`jpWP18eYvFL^EnL zG~Bn-B-haDmoEwa5XkbzTo#&<-V3K~{od?jSwH6V?IxLB>&xr^c_NI*RqLKPctz7p z?E6hocNKQ1skXd+o&UqCYjZIz_eWJ#MAs!2WS)-|Tda&8&d8XUmGH8ndFJPPa?7J@i(B&VV7nJdwgF2oaK-@>N#$@s=2e!j0g?9>Gm;;aTNY63C$m`xyiC>-}4vgs%u z5sh{=M2oS;;)IKn8|>>uU(0(L;VpE672c_@!VJ_S+j zBaK0M-~tXDpC3SNKAwFuWZO$>e@MYTPlEepkxw)ldLm!~#&~VBj1xM6$oQDNjNa9Y zMgjU2ilvhryJ~^FFC-@?zd;`~VPXmUH>C^Zel|v;!OBHMj}{Y}{H9s3+`ZUL`9QyX z6xJaZM=_CoPottE+JqL$+nWocxhO7da<0fc=g`NfxHAzVr!gDAu5I0L{{rh?Wm##6|7g~s76<+L3J&Z1P^4$=;}eNW-=tjqMNyY~-ps$l z)0l-oJ|x-L;X-(-+{zL?K0m*=-rmmwSZ@rquWqqEVPKH8$xclbNAQ+OZjM3oyI&NU z2_l7Jmv_Hjt}yA;G5{=i=639HnBNxbt-h)_Kn?TZT00gkin^D&6#ICY&&g%i{de9H z(v!vD{}Gl{d*nFIagZq zI_#=?RI71&uMa;=veeM1sF1;))HIaeEiZHIg-%wg#7UuQBs)=NPeYkKlBu~~_u&Ri zSqYVRVB~zw{pZ}(>o!=92&e|1pTCNgcYc1_TZ(0Dps@Y$r+R(>_;LLpv z=mbnF4R`67@}%_SmBELqRGCbdPgCxCf94XrzP<>np$6k{`0c2+LHWXtcA?0acnKwa zO_3Q|alVUhiA0*6ICt6|HGC&cQ+-JjmCje0(rFtd{~`o7gdz3G8r5%JrQH6iLj6Fh z&_TsS<3^^WwWs5bip|kckK&gqX&dr&oUHEISABDn@*}4WK7IabR8Ai?zYu@sVW`~* zKfl`c7Pak6`6GuaSubTCY96ef>w6^5_i&{{lARjc+dIAYR5|D6Q=YsW_srJj6_Zg& zq9DFeOhYeS{$Q-$VFLlX;5a{fc`;>WKKDn#s?~u{cHTOmm$lEsaP1?f?zs)}eEEg# zk3w>j<*O=X^IUz`Mj55;E4vlF$y!KxS&m)HVu?g29ic2qrEW~ly&zSxRjOKI(uVQs z+G_}E*Mo<+G@Fs3&`^_xPEY=h-Fv1pjT2t3?$NG@HfcBuGC`2jdY@lF0(lU<>E&DR zSI?_h_u3!bIkl7B#W7>GJ2=F>_nGvu4D$`I<(FszrPmL>yr!*pP_)uvmW0`>T5o$* z=i8ArxAfSnpEaeLCM{9k==iouqELZPEqYn&dOab$FDW%fsoV9$z8TqH#0R;boO&61 za5Zgg%u@Ge*@L=bc==V5_3ggE<2^^xw!nZouH9JBX*~Ex-#zoaw>b}At!u7)^<1yz zA5IrO$WjPbCV?y4+V^qP+gn>-XMV>?AzELBG~=Z1Wp^t+cC@zLHZic;zCJ$oyf^n4 z*~>RX4o-eb5yL?pUU_t4*BfSU0G%SKCEPSk$zV_9l3a)2C%=wny(xc4$OIb|`+z@A zSKuIjto7xQ$?5n~d)i!d6NNkNbmnP(o8WaSOb$IF7=+u{W82pc$hkL$=kkeW| zUWM6!^Nzw^nOp8vH`vak6C5p8sKEcQAp5xjzgo1|$@WE2@i!$F36)EnQagOLesbXH zrNRC656PIccdvFpfyF1#qV8eX(XBi9J1pJ;ftG~aab>62tJ#OXGM7uPV9`{EjE6r; z>J2DAw_2e#Y~?Z@(5>~U#J1_$>#vC(8PPcUKOWX_S5>W$k%oOa_iL)bn9`#r^}+Gg z|Ih<*HyIc>xcoM8=1mEaj`r#+ z9r9x-_0_qVUl|{j4H&k1KJlO$``Pn7YDYZ`u_1hSs$)M`U2;vhbI)(5I{oAcSVH*w zCR>_3`S;69AM~AhH{otRbkwatKEk1UE^U>H9r6v2GOKTugXVSI{d~fRa!;mF(+17a z5|Mu>=lGOo7-YFbVzy|Fl$h9Bd44rR!?jhFGI=mIN3z@SZ?LPpLDYg>DEwb+&ZO$u zre!Iya((>=%A)K)ZLsavI->5roq4}LA?Kz9alpL2VUop^H}YjaTEI3`2zprf>F;aD zA1Jx8L;}y!&DlKM+xv}9?NT4#zNni^MMPuo_;b(W?%UNyY4}0D+HhHaxM|L%!JhdT zSF6@D84RfO(;7;~8adDG(pu~cR+N4jKGvIzX5FvfX4=b(*l8V7H+3+GKESJ>Wq%>- z&|0PP+^w_U7?p_)PQD`JI@%k7yKgmj{x+fp_TK6>MFAe5-bITYiH-9cymDC0eLF|i zAQv~-Xmo^cT~!s%)JKPo1uM=Bmt~|c$#MKCm*0hN@@HDE!(G}G`pn>5UFPWf>*yi- zHES*aDVRDYUQN9t3HKrEDnQo=+8?tvvarbC;ZvrA&xRFeG%tqx?AbBX8{D)bE^3oM zQ~C<*fh$i|3?MQ%T8Xx~%66t+#xqw!@jUpBWUq@1 zb@{;FoCHH>zR~7aa3UD0u%oHZUsFFMyyMMR&-6E2k;lgu2(+@jZrqZHFgsCEc{q=7 zatW#%;?rl$unU>|RJ^Qyzw$*I{PV|@R&fcF|9t39;jb(bgE3G2D}TW7%A4T3{{1UU zZ2tWbUl!xvYw=|z7>s|f#lP1AEAW5BUaWcMz%bv1M2`*!ql8eU=+Z!Olym{5x=~SB zz|Y?bE{J&k7y1G>{y*maHGP3w4AXtiR@f@qGh~?%)M2hgy^AtRm=b0?|L=Y`dTR4d zqG_HI9Qy2<$gHpF&J3?q8XP_8rP3UY*hId1X~Zq6iX#`CQFI#l6rb55^5Cd+mUepY@t(G zyu?>Qv0$Ymst9i|aaY*(*-}ky?T`+xfL)g^GPrxM%8H|Nc|N_BXjnK{L6C^ z?9HgF_=EajoWg$H*2YGSq#1P#KkDN)fBG-P-|8#(hRj7Z*lz2G8m6Y6Ugj$f#=Xub z4G)qV@CTzpcJKEYmv-MSf2)6vQFDXV9fNx#F;tKz%-)U24V!a^H7GY5cWtIu@Fo-k ziNMHg&O6`Rem{Xlh5gHyFNd~`40H%ky~aR=5!Oje0+JqL`HKiqSqJqsmlSp@t`Srl z!STM0^;?bbnRX!s{58O~l!Q zLTT}laOIN~chbRQ)ZH+b5!Zh6++rVMoh4zm!^-e-D7I*)|9o^^%r_BRUMLRM4?$3> zT&%j!PROfE3o|Z9>vBKo)q9L|m0kPz6Wt=L-+2_FaDofL(Nz;dbEWDbx9L6$4g*S3^fbnE~Kn|q*j7qT*zFQM|=`@TOgn8kF95J zX-RBOHnP%}2hPSqnJRs_^(qsUkY2R2tD~kc`{2~Xhl|lF??-)0k3Vn;3Aox*(z#IF zLG`JV7ZeR|AnSZ8)DWc*COFl7C!_Wdirw+c7>Pf?3lQz#G|EMrU9A6n{0o)HSFc%f z9(AldBu$wR;tFk3b_o~_AA7&*TZLLh$|GY`;;8zOq%iH;AkPoww0ZuJU=RTqdi1rE~~vVUpe$?eF0gnL@g zWx{}TQ{_f`3uAlzXfRw7D6-z(#Qh_}4QPNuC}b^oFTonm4+Amg$jMLM49a)c4q2Kgd?`fQ3JH>njnU8w>6`n5&DBROO zTK}E#h%Cxy=@7}4R z8=&QfjTt$g>w8_!#$=|b(pE5aB<)BtLLZ#sI6*$!!tsOO4&A0o{e5jG=$KijY6~QJ z2=xQ;tfEIvkI37bXnmMFXFI-%*I41ue?~KgE0RV}y$|GFl*lMy(MR%_BQUMR=;{PN z$p(Ayc+4PA4pM4c4718DXsQw{WCe(dvW4yTTBuHo z;JYBAAbT)&9Y_b{QLc_YUd~Vw!QK}^d9uV|CPQ~8HX9KEL%~JL18K0V1%FswHO12G zkyB63VbPz7`DW%vl6O6!(`sgeH=!rwZrnhiOnhvEBV3OS_-pwzDLvE`Gb({CtNXZ~ z_$LE{?6TlAjp6T%iEal~oi$wgaudWf#6{+KSxLkC5-XGpVRu&l3p)MGH>Ed=J)Ul0 z)CFL6?CLR$nmv`IopzHraWv;THiK$^;v%!Wgx7#c3QmDpaWP~<)K4(XUcAdqGezW8pwtfJGy^^0PCb1CxJBZl$r8!ndaUd8r6T$s!9% zGWoFT)zNm!Iopa+hpRTp;AVB^^6ppuWDm-ZDGNizf1W8Um$<9VW0dDiz}CK<4W&W5@DY8~_>ZwJt<>u)-yoMlRiZnpf*-3@YV-Df|3FPddjRM!W!D%P`&!4Pb9Fo>JXS zxt_vc25i|oO(nh_cc8-okMKE<;Z)1p*j?9U4AZO)E|q*5N)jEvc+mV5(j;Qd7-;)q zktj$~8cJzYgN01uvEpnz4)H)_`^|NPJT|e^Pg+>mU7Y3>{DHV080vgY@b-bbL)Zo3 zPgNPxYG91hZUwGc$jeLQk!AjJV(hz!wPyK=Dl1uRwQYHZvci_#Q&r-rVPFglRkLCP z8lb#;>f>M{5tpH#py)LC3z`N74bgT+VBGn|f7=ymE20uRe*Eb6so%{suNo#MZq@0+ zpfz>ZK>aXxCi&w<0t*(X!f${Q?-fj<>_dfGIrBgz56vJQC2;J*Gk;=87rZHpVgeuW z^1_7%kXx`NCGMegJk}K!qh;rg046O2^Zy!%*jz3K5pA&|WeG;Pa?Re$nz2^{fEU}A zNHF65B5-Cee_e16(Bs*ef`OST47W(oX63-VkmSH8TRoWiHbxr6U^;^`n{lZ4OL*$$ zBURf=OcSDdh1G!*ch$i!Ky(~cZ!0^bty?TQ;~#&1B1o zW+3tNpO1;}Cson|EF$#j36ED-2_ZF-kCx;Rr50k?!FVtaqKU=gkBg@g`2qm}b#Ncw z_cSRhjSrWwp&!IY@D2F5RFx=ui@>Q{O$-oqVH;I$CkKd#c|g!iFx$XP(z=3aR>t~ofuAU0JaD{>VpQJp9XY9ce()fvTKrR;bJf6an2S=H9 z+8TF2;8Nn~lYddUcOKeKwp--Gruzv>5n&`;@Jdk7kPiANRA5&wVE5Bta(48tOT^6g z$DC_93}faHlK*8Y&ko=;VoMN`C+C7$%ZhemM8YQ<1J5m-Gu_->DNJN%7FY|%{PyhL z9p>&A5PnccCqp9?=PP)SKn;zF7If;Wl_6a<;g#P+tbopwkq?({h`58EemsYf_$_%( za?|Ku2X!QOD1Rf)fR}y}URUb*3V!n;08CMMvi5Wzj)|E3E#V>O`$0_#V z5HO+c9>lz0&U$*2Iv8=pYJfL5fBrVd|M`Z@^#@}Soaqsn$RAuXQB6a`8yqc0T*(Uu zm1g%Web8L$J1yL@^apz_;6^eAz(Zzn&)6U+)N*UruC)#N9+nC#tOn*Xc0C(>wcrjp zhTE^;%;gc+-ju~9VSA2`Wz3kS;@gh(5ytY{!0)gkw*5wY9RzXQ{wExIHgUuVB}%bj zB50>Xp~Z+gF@#t07#{>CA+nH#HN^0FdC79*gI`1Rl|mN}_)QzEVm>rC=aM*@WIu8J z!be+H*&xtL^Z1#eTSSK&48vz(nKXYIRxGXz5ef_#*+iJl@S&yZM1CVh&FD|5K1ii$9H^%GfS?4*e&JC`M%YQB*+6fvC>=W&0*(9>K$F~DFIWJ_ zL_7h4SU7z1hKGl(MEJbKDWXCDgS4T`+gsk=#CV^NPLsZs)t1U>p!7QT6|5q<2z5S) zj}@6B$h4u@39kH#%s60dJ{F-SOk#?8%)NW}5V@D{=%@BOQdf&IX<&nyXljGe8Kx|B zuStMG_y&0>U$bu`9NeKrm}!U(r?@*fcs412QeiXtOi3}44(Rw@(q zjp&lO7=w_PZo}cW?clAZ{<&SLb&}F+Mck!O*QH##L-D>{}1_cBzqtJnnt%YS69H$3Av2!Inxw9)aNRnHd5s+K^~_ z(*#+Rw*%s^;DpiTj^$nNj!>i;^nR+T)ZWhb72pQ(B=1%<(A}T_RF|t3uvpPH1o;73 zx1o0nl|9_-QOl8XA`5_^@wRZxH-B;{q(fh$!*Mt8wepdxuod{;>)O3Lcf2WD=+MZ3 zy0C9AhuyY`k8w5m?)455R$LiALc)_4-2m!V=d~FRet|<-gX40&NKRQ1fiJ^5i(9`VMY^6I^G`&m_T(+E_tQTZVAY)=g;y`%jlW}g}dsJDg!tQYxv zyksE~vApqYxIZ5+`7-cVaZb?db4{kZ_F+dV3$`Tn4+Qlb?`8hpuwk2}s*&j-KtN3e zTlRjZVu4#TXr^g5S7h3*Yy}oZO_lg}aDgC7z#uFFvUp*&nqg~A=e|Ny-L#n80#I$Tt9Wth)!YH-l18lp%JSYLubvp>RU3DOi4oFgo^ z2|jQ zh1xmW??h38ps)y7m`zx{28YEC54Fk>hRw{+2pjSr-P^s0~gZk#ohv=A^EoQ^VX9 zp(xjpgHRb9x(-nWOcgPQZ3Sj}$(A8IwHiH~Aa$y$g4M;$f}U*>{%MmFWROO-WB^fm zDKCwFfsy$oI>;earRU^$MA@lqi2q-Frpp;8z#R zk8{`wV6>~*PAPpvTI8T%7F}?~rX%4+yO8g_Vsyb?C3I_H-SX^rJOjW8F75d^8ecc2 z?Ub!q3>JZE-k*Q`mQv_IZU9qCZtiM>mvoHUzh4+h2H#dh8>yBEB)A3xy=Ot89k6m+ z0U5|SwQdq*Slz{DC;W0LYB_R{FwOl##0-Uby!EOYeO6;T+BCwX3wTa2+zQ>@R+Y~h zzdGgdhoepIgT;#y^fQum+B`>xi15i2W7Ix@Qv78ZC4kn*h{t**Q0v^qbn?jfb+YZg!J0Q94z7POJvrlyfmxFCqIV`WIn1^b|lLwXyYsvF3!7sz#o zX84OJu&6gcuI1u(G9jrC@Wp-}cR6i{coN-5h6dk(TU(+z4_1@5KNNcfOTuLC>tvQ> zqd$NfPNizgGIZI||ESxcvZ2726p_|`fc$N`_)v%DsryUJ*L zxjt~-o0b{k#WHW}DI;KkiBhZ5j^VqFXW+nao7j?-mJ|1x@@#UQiBGw^X)_)>yjjr& zC`v9wzt=>dwZ{#byvf(A(Mr zY~)AU^-}07^9JuOCBGj)`05coJt10FW@k-OXaz6^Hfpv$7hw0QTQ~Km z_p(Jx!uppAJ}?6(Yg`)XBkGj9&;zEfU(TAEQu;=u7ktt9P*e}~(76=c-ET$NtBRU* z2^&JJ`fiavl2+V&JjAHmZ~*F>X=V@g))vd(1{<651b&WSnY`SiCkch$(kI+rL23)+ zpgm3wZANrFi)pR^8I{mp?n3_B0O|D2YhNQ&KhdZ+HSxMD_C|`wJ|kqkjD`!fZ1FpB zT(Z5zD4;0|a3C$epSmZhQH;=dtGrWpXSUn}(kB{!K)oDkkiw1SMxE5j68jC3Hxxd> z98A#dam)N}rV$LKDm@>-7Vq_#$4WBv>nMpHMJtvj+?(h(k{7`oGH~?_Uu5a zt_8VzLT=8^UL)XEk!r-QA^nH^!fF{xS*j`H8 zBP1x)T*OGNqW?kHz0vV&`)+g4b8!9?8Ue3OJ6-2dZ- zF9oku0U%r(IP+s&ZivZYzln$)6dX*wB+VqDFzZSQkT^VWY?t+vR3>C~soTM?m3$R) z+ES7nwB+Sc--CuBmun7xedPJ~D&>BQWY%)`B}C18#vAYL-`fc*gGe9^iI&td_zWr; zh1wv;NByzmW1phnwa}Te06vD0CAC3<;caE_x6z~5mz$42H(R7u61_&9lt0GggsoED zSWWm61*cyaHL<+QcI2V)W*+UF<_G2beDMRz55;Mu*?{$vrDS02FkLUwo>5LcjaIyZ zUUBwl3}>a~48+A|JbCg&?oGsR)WU}hhzLP#nj_H>f+aW&HQ_GDDT&uf*qSOT9g4El z!!3+N*_OTEKCTJ_pPwk~N$9SR9mYaa;{+3`0)Bykf!aB4&XmhTkeFU31EZCG8f1My zro=dt?N>p<96|$FRUX=L7t;Q_xJK5eyuq20No}#)GLi8{vzW?YJgbSz)qYya<5%IZa~pH+{uY;kqq8orSwwDBz;b$HMAx}Vc$cb7V5_Ni`Djj&PaviZL&G`aG z7w|;4C;6);+KD{oC)h(tRW|jZU{m`cJFgBgj%L6xKWrs^_SA^;o%%ebIW-xxmk1)+ zXabkS?E*xHHWdi*qezkjL1z**%b>EQT=*QNE$2ZtKG(Se%}XF$~>YQG^<(Fw1Z;m@NYgA#pQQX;q<^=bx8X%fF@P{#lLx+j>9m#_KW%mxw>_ zoAYPW>*!~xziwhmiu>*pG#2ds>1^Dr;Gd=spPR>?`Err`HouwjDPo~=Q5wr*{I}eE zQZsMf%d**F)1P`@EWh|-r?r3R44>Uywz@3~I(q_Q2HV!%c48h)8SD>z6*~ARyd$Tl zWBBUu1gBf&6)z~LlU@;>-=h(S`x@sX;c?i5Cr#K)@~+D#emr*WhtKbLtWfdzm#Mx_ z%-3nfcXs|h@z;NfZ0D+>PW&JK?UlUjKX7yYH{Ut&TR%De>uXZvYK13XYa#|+x4y3U z`726?;Q&2bXod)Nep=O~-a6HkBZyGy>>SY>Xy_6xe-#%OM{UHimNgeK38RfBv3O0H zdO%CdsmM&UkirZktY1@tIo0x2AZ5)76G-7~B3qpMv}aTsN^!Ka1JF?fm>`yv2wV&x)YVi>u8xXg)ecg&TKqfk&BE6s4{WUDO&ORp~#$e zM!eHSB?hI`OO$D0ijHVgAz=B5lP3#rW0;Th=FMp%icp@q2xmj1I@CI*^2}bJRf6o$ z0pzvck3y7Y5B`8Wugx<%gxD$QTPInnL`RzMgxsN0^VYO_+dY(CbHgN%P~^9d_P8LeDv82NXH#x&qAc2pNgN5S_B;cKgC&vlEdkol(!Pp5 zN2FS8YZNHqf;HxAhu1V~c;v7Jt?~i#B@omIqQNXE1S@n7PwXfA#YUe^hnHju{3YEH z%6Ww-8hiYdxM3GMmP1KUiZmO1syuOzi+ej!ZCLPa2bGIx4o`%!K_HzAY7|rykqqgt@ugF?=9&fPSZ_0}5Ja9wO}Agf)`yRHVWU4bJ&78%LPx z%IGp*(_V)`UMu%q(_Zj?1f{l(+7s-+cSd{vKES1}NJrCPsJB&db-W`~XJ9pV8?gnStBK;u}GcwutV_twe*`C8hIO@gDSsUNJB`P`wj$}sK4m-aKcBt(G+aY$`F!6Ufzn1MCV&r^!MuC|-S{)rTC)W!jnp z6jyEmysDEy;*eg~|Mq5uMRiYY)yrX}BVC$u+|PBof*vaJlfrS6q`b@NfKVF-A8!k7(7& z7b2NBJr_bRwb|pyd5hmo$jHcOFcU3CI%#+N{X2J_A-JETtBkBwA7(?`j$%*z{-1L5 zzPNLFeYgr61rjrqfG_j`sR zIliEZMzJ7j<|uvMOYD7LOu(DDI2~#Os6Op(#4|5B4zNL;?QbOXU1WDhMI-8MCHO-A?^zd2dc9z64$>?74qGiabPU zh6^N7o%RT((xrqp@b^DvqpC+WPE2uAuXjSg(FVU| zU2Om|)%uSv4UkmoBfUW259{hJLb8_6C}Z(d$q4){%5VG#3(>*KAQM zyp3pzL?QtnE+bZ#!JaAxYb9_w2x4@-DCHdPGoWcm$xtKFN9#ZG0s8}6P;M$ARSKF- zB#2ZE5i4NeF!?*?)~yxLl7UI@*DPtoV4--V`|)%BDnfg1ty1AzI&Z+c90R7oEo<1> z*#)+M@j#S$h&mkTbDc1@Kbw3pwhWPz$SZIs6@fN+9EHs@12Hi%(lRn46`;^vr~O98 zw9dV{rQlLOe4H{lLf}Ph?WRnpQw41z?Y_>aIX$P^?T?BkdEWInhU$tIkG-bAqdp_+ z46L%eKr5&3GBG~3KuKy(*o2?-njf{t>`Qm(Xpi6)+hYhM?U4U$h3JIRv62(FuLl$D zADn^$WRu1?nK1dmpfdswk%N9))%9$gZH~&6Evi|sp_x()Bqr%OHX7Lmmrf5w|4s>V zSR`ce79^rw{oDbBnEEzj0D0}%(-%!xw9?_VwHV0xrMFv{ha5Aua5Bh`$o(!O<{46w zK=N0xzQR{VpjvvB2rbxqud-d8;48|3#j~KD_Z)SHgNF}S{`I*9`P~b#FJN Date: Fri, 22 Aug 2025 20:20:05 +0000 Subject: [PATCH 26/32] documentation commit --- .../{pretrained_spatiotemporal.md => KALA_JAMUN_documentation.md} | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename docs/{pretrained_spatiotemporal.md => KALA_JAMUN_documentation.md} (100%) diff --git a/docs/pretrained_spatiotemporal.md b/docs/KALA_JAMUN_documentation.md similarity index 100% rename from docs/pretrained_spatiotemporal.md rename to docs/KALA_JAMUN_documentation.md From e5d2f24855f783fdc1055995163e92ab4d9bcb72 Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 22 Aug 2025 23:17:16 +0000 Subject: [PATCH 27/32] documentation commit --- .gitignore | 3 +- ...n_enhanced_spatiotemporal_conditioner.yaml | 4 +- docs/KALA_JAMUN_documentation.md | 65 +- scratch/README_reorganization.md | 122 +++ scratch/analyze_grid_code_distribution.py | 163 ++++ scratch/analyze_trajectory_noise.py | 276 +++++++ scratch/bond_length_issues.py | 69 ++ scratch/check_trajectory_length.py | 78 ++ scratch/diagnose_mdtraj_dataset.py | 225 +++++ scratch/explore_fake_enhanced_dataset.py | 172 ++++ scratch/explore_fake_enhanced_minimal.py | 151 ++++ scratch/inspect_model_minimal.py | 15 + scratch/load_model_state_dict.py | 23 + scratch/my_histogram.png | Bin 0 -> 164601 bytes scratch/organize_data.py | 140 ++++ scratch/reorganize_swarm_data.py | 770 ++++++++++++++++++ scratch/test_conditional_denoiser.py | 191 +++++ scratch/test_conditional_sampling.py | 179 ++++ scratch/test_conditional_simple.py | 150 ++++ scratch/test_conditioners.py | 396 +++++++++ scratch/test_denoised_conditioner.py | 403 +++++++++ scratch/test_gradient_equivalence.py | 270 ++++++ scratch/test_multimeasurement.py | 172 ++++ scratch/test_reorganize_swarm_data.py | 253 ++++++ scratch/test_repeated_position_dataset.py | 125 +++ scratch/transformer/convert_spatiotemporal.py | 194 +++++ scratch/transformer/develop_transformer.py | 354 ++++++++ scratch/transformer/helpers.py | 187 +++++ scratch/transformer/pooling.py | 217 +++++ scratch/transformer/temporal_transformer.py | 304 +++++++ scratch/transformer/test_e3_spatiotemporal.py | 214 +++++ .../test_spatiotemporal_conditioner.py | 433 ++++++++++ scratch/visualize_fake_enhanced_data.py | 247 ++++++ scratch/visualize_noise_denoise.py | 271 ++++++ 34 files changed, 6768 insertions(+), 68 deletions(-) create mode 100644 scratch/README_reorganization.md create mode 100755 scratch/analyze_grid_code_distribution.py create mode 100644 scratch/analyze_trajectory_noise.py create mode 100644 scratch/bond_length_issues.py create mode 100644 scratch/check_trajectory_length.py create mode 100644 scratch/diagnose_mdtraj_dataset.py create mode 100644 scratch/explore_fake_enhanced_dataset.py create mode 100644 scratch/explore_fake_enhanced_minimal.py create mode 100644 scratch/inspect_model_minimal.py create mode 100644 scratch/load_model_state_dict.py create mode 100644 scratch/my_histogram.png create mode 100755 scratch/organize_data.py create mode 100644 scratch/reorganize_swarm_data.py create mode 100644 scratch/test_conditional_denoiser.py create mode 100644 scratch/test_conditional_sampling.py create mode 100644 scratch/test_conditional_simple.py create mode 100644 scratch/test_conditioners.py create mode 100644 scratch/test_denoised_conditioner.py create mode 100644 scratch/test_gradient_equivalence.py create mode 100644 scratch/test_multimeasurement.py create mode 100644 scratch/test_reorganize_swarm_data.py create mode 100644 scratch/test_repeated_position_dataset.py create mode 100644 scratch/transformer/convert_spatiotemporal.py create mode 100644 scratch/transformer/develop_transformer.py create mode 100644 scratch/transformer/helpers.py create mode 100644 scratch/transformer/pooling.py create mode 100644 scratch/transformer/temporal_transformer.py create mode 100644 scratch/transformer/test_e3_spatiotemporal.py create mode 100755 scratch/transformer/test_spatiotemporal_conditioner.py create mode 100644 scratch/visualize_fake_enhanced_data.py create mode 100644 scratch/visualize_noise_denoise.py diff --git a/.gitignore b/.gitignore index 10226f9..314eb10 100644 --- a/.gitignore +++ b/.gitignore @@ -24,5 +24,4 @@ torch_compile_debug *.profile* **/*.log **/*.err -wandb/* -scratch/* \ No newline at end of file +wandb/* \ No newline at end of file diff --git a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml index d8d471e..43c8dce 100644 --- a/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml +++ b/configs/experiment/train_enhanced_spatiotemporal_conditioner.yaml @@ -18,7 +18,7 @@ data: subsample: 1 total_lag_time: 5 lag_subsample_rate: 1 - max_datasets: 500 # Increased for more training data + max_datasets: 1 # Increased for more training data val: _target_: jamun.data.parse_datasets_from_directory @@ -28,7 +28,7 @@ data: subsample: 1 total_lag_time: ${data.datamodule.datasets.train.total_lag_time} lag_subsample_rate: ${data.datamodule.datasets.train.lag_subsample_rate} - max_datasets: 100 + max_datasets: 1 model: sigma_distribution: diff --git a/docs/KALA_JAMUN_documentation.md b/docs/KALA_JAMUN_documentation.md index fad1628..d270791 100644 --- a/docs/KALA_JAMUN_documentation.md +++ b/docs/KALA_JAMUN_documentation.md @@ -815,7 +815,7 @@ Both datasets consist of swarms sampled at **20 fs intervals**, providing high t **Source Location**: `/data/bucket/vanib/ALA_ALA/swarms/swarm_results` -The raw swarm data has been reorganized using the script: `reorganize_swarm_data.py` which sorts the trajectories into training and validation buckets according to different splitting strategies. +The raw swarm data has been reorganized using the script: `scratch/reorganize_swarm_data.py` which sorts the trajectories into training and validation buckets according to different splitting strategies. #### Data Splitting Strategies @@ -916,48 +916,6 @@ Once data selection is completed, models can be trained using the `train_enhance jamun_train --config-dir=configs experiment={experimental_config_name} ``` -#### Conditional Denoiser Configuration - -KALA-JAMUN uses a specialized model configuration for conditional denoisers: - -**Configuration Name**: `denoiser_conditional` - -This configuration differs from the standard denoiser in several key aspects: -- **Conditioning Module**: Includes spatiotemporal conditioning components -- **Architecture Module**: Uses E3ConvConditional variants -- **Training Parameters**: Optimized for temporal conditioning tasks -- **Memory Management**: Handles hidden_state processing during training - -#### Parameter Tuning - -All training parameters are preset in the experimental configurations but can be tuned through the Hydra configuration system: - -**Key Configuration Sections:** -- `model/denoiser_conditional`: Model architecture and conditioning parameters -- `data`: Dataset loading and preprocessing settings -- `training`: Optimization parameters (learning rate, batch size, etc.) -- `callbacks`: Training monitoring and checkpointing - -**Example Parameter Modifications:** -```yaml -# In configs/model/denoiser_conditional/spatiotemporal.yaml -conditioning_module: - total_lag_time: 5 - lag_subsample_rate: 10 - -architecture_module: - hidden_dim: 128 - num_layers: 4 -``` - -#### Training Process - -1. **Data Loading**: Enhanced datasets are loaded with proper temporal subsampling -2. **Model Initialization**: Conditional denoiser with spatiotemporal components -3. **Training Loop**: Includes hidden_state processing and temporal loss computation -4. **Validation**: Tests on held-out swarms/states based on splitting strategy -5. **Checkpointing**: Regular model saves for sampling and analysis - ### 4.3 Sampling #### Sampling Configuration @@ -987,27 +945,6 @@ The `sample_memory` configuration automatically handles: 3. **Sampler Selection**: Uses `SamplerMemory` with `baoab_memory` algorithm 4. **Wrapper Configuration**: Employs `ModelSamplingWrapperMemory` for proper interface -#### Sampling Parameters - -Key parameters in the memory sampling configuration: - -```yaml -sampler: - _target_: jamun.sampling.SamplerMemory - -batch_sampler: - _target_: jamun.sampling.mcmc.splitting.BAOABMemory - history_update_frequency: 10 # Inner loop equilibration steps - steps: 10000 # Outer loop iterations - -model_wrapper: - _target_: jamun.utils.ModelSamplingWrapperMemory - sigma: 0.1 - recenter_on_init: true -``` - -This experimental framework enables comprehensive evaluation of KALA-JAMUN's temporal conditioning capabilities across different data splitting strategies and conformational scenarios. - ### 4.4 Experiments This section describes key experiments designed to evaluate KALA-JAMUN's performance and validate design choices for temporal conditioning. diff --git a/scratch/README_reorganization.md b/scratch/README_reorganization.md new file mode 100644 index 0000000..856b5d0 --- /dev/null +++ b/scratch/README_reorganization.md @@ -0,0 +1,122 @@ +# ALA_ALA Swarm Data Reorganization Scripts + +## Overview +These scripts reorganize molecular dynamics swarm data from `/data/bucket/vanib/ALA_ALA/swarm_results/` into a machine learning-ready format in `/data2/sules/ALA_ALA_enhanced/`. + +## Scripts + +### 1. `reorganize_swarm_data.py` - Main Script +**Input Structure:** +- Source: `/data/bucket/vanib/ALA_ALA/swarm_results/` +- 184 directories: `AA_000/`, `AA_001/`, ..., `AA_183/` +- Each contains: `swarm_1ps_001.xtc`, `swarm_1ps_002.xtc`, ..., `swarm_1ps_005.xtc` +- Single PDB file: `/data/bucket/vanib/ALA_ALA/ALA_ALA.pdb` + +**Output Structure:** +- Target: `/data2/sules/ALA_ALA_enhanced/` +- `train/` - 172 randomly selected grid codes (860 .xtc + 860 .pdb files) +- `val/` - Remaining 12 grid codes (60 .xtc + 60 .pdb files) + +**File Naming Convention:** +- Original: `swarm_1ps_{traj_code}.xtc` ā†’ New: `swarm_1ps_{grid_code}_{traj_code}.xtc` +- PDB files: `swarm_1ps_{grid_code}_{traj_code}.pdb` (copied from single source) + +**Features:** +- āœ… **Progress bars** with tqdm showing copy progress +- āœ… **Reproducible random split** (seed=42) +- āœ… **mdtraj validation** - Tests that .xtc + .pdb pairs load correctly +- āœ… **Comprehensive logging** - Detailed progress and error reporting +- āœ… **Safe operation** - Only copies, never moves/deletes source data +- āœ… **Verification** - File count validation and structure checking + +### 2. `test_reorganize_swarm_data.py` - Test Script +- Creates mock data structure with 5 grid codes +- Tests the reorganization logic with small dataset +- Validates file organization and naming +- Tests mdtraj integration (expected to fail on mock data) +- Verifies progress bar functionality + +## Usage + +```bash +# Activate conda environment +conda activate jamun + +# Navigate to scripts directory +cd scratch + +# Run test first (optional) +python test_reorganize_swarm_data.py + +# Run reorganization with trajectory split (default) +python reorganize_swarm_data.py trajectory_split + +# Or run reorganization with grid split +python reorganize_swarm_data.py grid_split + +# Or run without arguments (defaults to trajectory_split) +python reorganize_swarm_data.py +``` + +## Splitting Strategies + +The script supports two different data splitting strategies: + +### 1. Grid Split (`grid_split`) +- **Random grid codes split**: 172 grids for train, 12 grids for val, all trajectories +- **Output folder**: `/data2/sules/ALA_ALA_enhanced_full_swarm/` +- **Train**: 172 grid codes Ɨ 5 trajectories Ɨ 2 file types = 1,720 files +- **Val**: 12 grid codes Ɨ 5 trajectories Ɨ 2 file types = 120 files + +### 2. Trajectory Split (`trajectory_split`) - **DEFAULT** +- **All grids split by trajectory**: trajectories 001-004 for train, 005 for val +- **Output folder**: `/data2/sules/ALA_ALA_enhanced_full_grid/` +- **Train**: 184 grid codes Ɨ 4 trajectories Ɨ 2 file types = 1,472 files +- **Val**: 184 grid codes Ɨ 1 trajectory Ɨ 2 file types = 368 files + +## Expected Results + +**Grid Split structure:** +``` +/data2/sules/ALA_ALA_enhanced_full_swarm/ +ā”œā”€ā”€ train/ # 1720 files total +ā”‚ ā”œā”€ā”€ swarm_1ps_000_001.xtc # Random 172 grids, all trajectories +ā”‚ ā”œā”€ā”€ swarm_1ps_000_001.pdb +ā”‚ ā””ā”€ā”€ ... (172 grid codes Ɨ 5 trajectories Ɨ 2 file types) +ā””ā”€ā”€ val/ # 120 files total + ā”œā”€ā”€ swarm_1ps_XXX_001.xtc # Remaining 12 grids, all trajectories + ā””ā”€ā”€ ... (12 grid codes Ɨ 5 trajectories Ɨ 2 file types) +``` + +**Trajectory Split structure:** +``` +/data2/sules/ALA_ALA_enhanced_full_grid/ +ā”œā”€ā”€ train/ # 1472 files total +ā”‚ ā”œā”€ā”€ swarm_1ps_000_001.xtc # All 184 grids, trajectories 001-004 +ā”‚ ā”œā”€ā”€ swarm_1ps_000_002.xtc +ā”‚ ā”œā”€ā”€ swarm_1ps_000_003.xtc +ā”‚ ā”œā”€ā”€ swarm_1ps_000_004.xtc +ā”‚ ā””ā”€ā”€ ... (184 grid codes Ɨ 4 trajectories Ɨ 2 file types) +ā””ā”€ā”€ val/ # 368 files total + ā”œā”€ā”€ swarm_1ps_000_005.xtc # All 184 grids, trajectory 005 only + ā”œā”€ā”€ swarm_1ps_001_005.xtc + ā””ā”€ā”€ ... (184 grid codes Ɨ 1 trajectory Ɨ 2 file types) +``` + +## Dependencies +- **mdtraj** - For trajectory validation (available in jamun environment) +- **tqdm** - For progress bars (available in jamun environment) +- **Standard library** - os, shutil, random, logging, pathlib + +## Runtime Estimate +- **Test script**: ~1 second +- **Grid split**: ~52 seconds (1,840 file operations) +- **Trajectory split**: ~37 seconds (1,840 file operations) +- **Disk space needed**: ~Same as source data for each folder (copying, not moving) + +## Safety Features +- Never deletes or moves source data +- Validates all paths before starting +- Reports missing files without stopping +- Tests mdtraj compatibility with random samples +- Detailed error logging and recovery \ No newline at end of file diff --git a/scratch/analyze_grid_code_distribution.py b/scratch/analyze_grid_code_distribution.py new file mode 100755 index 0000000..699a2e2 --- /dev/null +++ b/scratch/analyze_grid_code_distribution.py @@ -0,0 +1,163 @@ +#!/usr/bin/env python3 +""" +Script to analyze the distribution of trajectories by grid codes. +Creates a histogram showing how many trajectory codes exist for each grid code. +""" + +import os +import re +import matplotlib.pyplot as plt +import numpy as np +from collections import defaultdict, Counter +import argparse + +def parse_trajectory_files(data_dir): + """Parse trajectory files and extract grid codes and traj codes.""" + # Pattern to match traj_{grid_code}_{traj_code} + pattern = re.compile(r'^traj_(\d+)_(\d+)') + + grid_traj_mapping = defaultdict(list) + + # Scan directory for trajectory files + if not os.path.exists(data_dir): + raise ValueError(f"Directory {data_dir} does not exist") + + files = os.listdir(data_dir) + trajectory_files = [] + + for filename in files: + match = pattern.match(filename) + if match: + grid_code = int(match.group(1)) + traj_code = int(match.group(2)) + grid_traj_mapping[grid_code].append(traj_code) + trajectory_files.append(filename) + + print(f"Found {len(trajectory_files)} trajectory files") + print(f"Found {len(grid_traj_mapping)} unique grid codes") + + return grid_traj_mapping + +def create_histogram(grid_traj_mapping, output_path=None): + """Create histogram of trajectory counts per grid code.""" + + if not grid_traj_mapping: + print("No trajectory files found!") + return + + # Get the full range of grid codes (min to max) + all_grid_codes = list(grid_traj_mapping.keys()) + min_grid = min(all_grid_codes) + max_grid = max(all_grid_codes) + + print(f"Grid code range: {min_grid} to {max_grid}") + + # Create array for all grid codes in range + full_range = list(range(min_grid, max_grid + 1)) + traj_counts = [] + + for grid_code in full_range: + count = len(grid_traj_mapping.get(grid_code, [])) + traj_counts.append(count) + + # Print some statistics + total_trajs = sum(traj_counts) + non_zero_grids = sum(1 for count in traj_counts if count > 0) + zero_grids = len(full_range) - non_zero_grids + + print(f"Total trajectories: {total_trajs}") + print(f"Grid codes with trajectories: {non_zero_grids}") + print(f"Grid codes with no trajectories: {zero_grids}") + print(f"Average trajectories per grid code: {total_trajs / len(full_range):.2f}") + print(f"Max trajectories for single grid code: {max(traj_counts)}") + print(f"Min trajectories for single grid code: {min(traj_counts)}") + + # Create histogram + plt.figure(figsize=(12, 6)) + plt.bar(full_range, traj_counts, width=0.8, alpha=0.7, edgecolor='black', linewidth=0.5) + plt.xlabel('Grid Code') + plt.ylabel('Number of Trajectories') + plt.title('Distribution of Trajectories by Grid Code') + plt.grid(True, alpha=0.3) + + # Add some formatting + if len(full_range) > 50: + # If too many grid codes, adjust x-axis ticks + step = max(1, len(full_range) // 20) + plt.xticks(full_range[::step], rotation=45) + else: + plt.xticks(full_range) + + plt.tight_layout() + + # Save plot + if output_path: + plt.savefig(output_path, dpi=300, bbox_inches='tight') + print(f"Histogram saved to: {output_path}") + + plt.show() + + return full_range, traj_counts + +def print_detailed_stats(grid_traj_mapping): + """Print detailed statistics about the distribution.""" + + counts = [len(trajs) for trajs in grid_traj_mapping.values()] + + print("\n" + "="*50) + print("DETAILED STATISTICS") + print("="*50) + + print(f"Total unique grid codes found: {len(grid_traj_mapping)}") + print(f"Total trajectories: {sum(counts)}") + + if counts: + print(f"Mean trajectories per grid code: {np.mean(counts):.2f}") + print(f"Median trajectories per grid code: {np.median(counts):.2f}") + print(f"Std dev trajectories per grid code: {np.std(counts):.2f}") + + # Show distribution of counts + count_dist = Counter(counts) + print(f"\nDistribution of trajectory counts:") + for count, frequency in sorted(count_dist.items()): + print(f" {count} trajectories: {frequency} grid codes") + + # Show some examples + print(f"\nExample grid codes and their trajectory counts:") + for i, (grid_code, trajs) in enumerate(sorted(grid_traj_mapping.items())[:10]): + print(f" Grid {grid_code}: {len(trajs)} trajectories") + + if len(grid_traj_mapping) > 10: + print(f" ... and {len(grid_traj_mapping) - 10} more") + +def main(): + parser = argparse.ArgumentParser(description="Analyze trajectory distribution by grid codes") + parser.add_argument("--data-dir", + default="/data2/sules/fake_enhanced_data/ALA_ALA", + help="Directory containing trajectory files") + parser.add_argument("--output", + default="scratch/grid_code_histogram.png", + help="Output path for histogram plot") + + args = parser.parse_args() + + print(f"Analyzing trajectories in: {args.data_dir}") + + # Parse trajectory files + grid_traj_mapping = parse_trajectory_files(args.data_dir) + + if not grid_traj_mapping: + print("No trajectory files found matching pattern traj_{grid_code}_{traj_code}") + return + + # Print detailed statistics + print_detailed_stats(grid_traj_mapping) + + # Create histogram + print(f"\nCreating histogram...") + full_range, traj_counts = create_histogram(grid_traj_mapping, args.output) + + print(f"\nAnalysis complete!") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/analyze_trajectory_noise.py b/scratch/analyze_trajectory_noise.py new file mode 100644 index 0000000..a31e07b --- /dev/null +++ b/scratch/analyze_trajectory_noise.py @@ -0,0 +1,276 @@ +#!/usr/bin/env python3 +""" +Trajectory noise analysis script. +Loads .xtc trajectory files, adds noise, computes norms between successive points, +and creates histograms with flexible time point filtering. +""" + +import os +import glob +import numpy as np +import matplotlib.pyplot as plt +import mdtraj as md +from pathlib import Path +from typing import List, Tuple, Optional +import argparse + + +def load_trajectories(traj_dir: str, topology_file: str, max_files: Optional[int] = None) -> List[md.Trajectory]: + """ + Load trajectory files from directory using MDTraj. + + Args: + traj_dir: Directory containing .xtc files + topology_file: Path to PDB topology file + max_files: Maximum number of files to load (None for all) + + Returns: + List of MDTraj trajectory objects + """ + xtc_files = glob.glob(os.path.join(traj_dir, "*.xtc")) + xtc_files.sort() + + if max_files is not None: + xtc_files = xtc_files[:max_files] + + print(f"Loading {len(xtc_files)} trajectory files...") + + trajectories = [] + for i, xtc_file in enumerate(xtc_files): + try: + traj = md.load(xtc_file, top=topology_file) + trajectories.append(traj) + if (i + 1) % 50 == 0: + print(f"Loaded {i + 1}/{len(xtc_files)} trajectories") + except Exception as e: + print(f"Warning: Failed to load {xtc_file}: {e}") + + print(f"Successfully loaded {len(trajectories)} trajectories") + return trajectories + + +def add_noise_to_trajectory(traj: md.Trajectory, noise_magnitude: float = 0.04) -> md.Trajectory: + """ + Add Gaussian noise to trajectory coordinates. + + Args: + traj: MDTraj trajectory object + noise_magnitude: Standard deviation of Gaussian noise to add (in nm) + + Returns: + New trajectory with added noise + """ + # Copy the trajectory to avoid modifying the original + noisy_traj = traj.slice(range(traj.n_frames)) + + # Add Gaussian noise to xyz coordinates + noise = np.random.normal(0, noise_magnitude, noisy_traj.xyz.shape) + noisy_traj.xyz += noise + + return noisy_traj + + +def compute_successive_norms(traj: md.Trajectory) -> np.ndarray: + """ + Compute norms between successive trajectory points. + + Args: + traj: MDTraj trajectory object + + Returns: + Array of norms between successive points for each atom + """ + if traj.n_frames < 2: + return np.array([]) + + # Calculate differences between successive frames + diff = traj.xyz[1:] - traj.xyz[:-1] # Shape: (n_frames-1, n_atoms, 3) + + # Compute norms for each atom at each time step + norms = np.linalg.norm(diff, axis=2) # Shape: (n_frames-1, n_atoms) + + return norms + + +def compute_norms_for_time_points(traj: md.Trajectory, time_points: List[Tuple[int, int]]) -> np.ndarray: + """ + Compute norms between specific time points. + + Args: + traj: MDTraj trajectory object + time_points: List of (start_frame, end_frame) tuples + + Returns: + Array of norms for specified time point pairs + """ + norms = [] + + for start_frame, end_frame in time_points: + if start_frame < traj.n_frames and end_frame < traj.n_frames: + diff = traj.xyz[end_frame] - traj.xyz[start_frame] # Shape: (n_atoms, 3) + frame_norms = np.linalg.norm(diff, axis=1) # Shape: (n_atoms,) + norms.extend(frame_norms) + + return np.array(norms) + + +def analyze_trajectories(trajectories: List[md.Trajectory], + noise_magnitude: float = 0.04, + time_point_filter: Optional[List[Tuple[int, int]]] = None) -> Tuple[np.ndarray, np.ndarray]: + """ + Analyze trajectories by adding noise and computing norms. + + Args: + trajectories: List of MDTraj trajectory objects + noise_magnitude: Standard deviation of Gaussian noise + time_point_filter: Optional list of (start, end) frame pairs to analyze + + Returns: + Tuple of (original_norms, noisy_norms) + """ + original_norms = [] + noisy_norms = [] + + print(f"Analyzing {len(trajectories)} trajectories...") + + for i, traj in enumerate(trajectories): + if time_point_filter is not None: + # Compute norms for specific time points + orig_norm = compute_norms_for_time_points(traj, time_point_filter) + + # Add noise and compute norms for same time points + noisy_traj = add_noise_to_trajectory(traj, noise_magnitude) + noisy_norm = compute_norms_for_time_points(noisy_traj, time_point_filter) + else: + # Compute successive norms for all time points + orig_norm = compute_successive_norms(traj) + + # Add noise and compute successive norms + noisy_traj = add_noise_to_trajectory(traj, noise_magnitude) + noisy_norm = compute_successive_norms(noisy_traj) + + # Flatten and collect norms + original_norms.extend(orig_norm.flatten()) + noisy_norms.extend(noisy_norm.flatten()) + + if (i + 1) % 10 == 0: + print(f"Analyzed {i + 1}/{len(trajectories)} trajectories") + + return np.array(original_norms), np.array(noisy_norms) + + +def create_histogram(original_norms: np.ndarray, + noisy_norms: np.ndarray, + title: str = "Norm Differences Between Successive Trajectory Points", + bins: int = 50, + save_path: Optional[str] = None): + """ + Create histogram comparing original and noisy trajectory norms. + + Args: + original_norms: Array of norms from original trajectories + noisy_norms: Array of norms from noisy trajectories + title: Plot title + bins: Number of histogram bins + save_path: Optional path to save the plot + """ + plt.figure(figsize=(12, 8)) + + # Create histogram + plt.hist(original_norms, bins=bins, alpha=0.7, label='Original', density=True, color='blue') + plt.hist(noisy_norms, bins=bins, alpha=0.7, label='With Noise (Ļƒ=0.04)', density=True, color='red') + + plt.xlabel('Norm (nm)') + plt.ylabel('Density') + plt.title(title) + plt.legend() + plt.grid(True, alpha=0.3) + + # Add statistics + orig_mean, orig_std = np.mean(original_norms), np.std(original_norms) + noisy_mean, noisy_std = np.mean(noisy_norms), np.std(noisy_norms) + + stats_text = f'Original: Ī¼={orig_mean:.4f}, Ļƒ={orig_std:.4f}\n' + stats_text += f'Noisy: Ī¼={noisy_mean:.4f}, Ļƒ={noisy_std:.4f}' + + plt.text(0.98, 0.98, stats_text, transform=plt.gca().transAxes, + verticalalignment='top', horizontalalignment='right', + bbox=dict(boxstyle='round', facecolor='white', alpha=0.8)) + + plt.tight_layout() + + if save_path: + plt.savefig(save_path, dpi=300, bbox_inches='tight') + print(f"Plot saved to {save_path}") + + plt.show() + + +def main(): + parser = argparse.ArgumentParser(description='Analyze trajectory noise effects') + parser.add_argument('--traj_dir', type=str, + default='/data2/sules/fake_enhanced_data/ALA_ALA_organized/train', + help='Directory containing trajectory files') + parser.add_argument('--topology', type=str, + default='/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb', + help='PDB topology file') + parser.add_argument('--noise_magnitude', type=float, default=0.04, + help='Noise magnitude (standard deviation)') + parser.add_argument('--max_files', type=int, default=50, + help='Maximum number of trajectory files to load') + parser.add_argument('--time_filter', type=str, default=None, + help='Time point filter as "start1,end1;start2,end2" (e.g., "0,1" for initial->next)') + parser.add_argument('--output', type=str, default='trajectory_noise_analysis.png', + help='Output plot filename') + parser.add_argument('--bins', type=int, default=50, + help='Number of histogram bins') + + args = parser.parse_args() + + # Parse time filter if provided + time_point_filter = None + if args.time_filter: + try: + pairs = args.time_filter.split(';') + time_point_filter = [] + for pair in pairs: + start, end = map(int, pair.split(',')) + time_point_filter.append((start, end)) + print(f"Using time point filter: {time_point_filter}") + except: + print("Warning: Invalid time filter format. Using all successive points.") + + # Load trajectories + trajectories = load_trajectories(args.traj_dir, args.topology, args.max_files) + + if not trajectories: + print("No trajectories loaded. Exiting.") + return + + # Analyze trajectories + original_norms, noisy_norms = analyze_trajectories( + trajectories, args.noise_magnitude, time_point_filter + ) + + # Create title based on analysis type + if time_point_filter: + title = f"Norm Differences for Time Points {time_point_filter}" + else: + title = "Norm Differences Between Successive Trajectory Points" + title += f" (Noise Ļƒ={args.noise_magnitude})" + + # Create histogram + create_histogram(original_norms, noisy_norms, title, args.bins, args.output) + + # Print summary statistics + print(f"\nSummary Statistics:") + print(f"Original trajectories: {len(original_norms)} data points") + print(f" Mean norm: {np.mean(original_norms):.6f} nm") + print(f" Std norm: {np.std(original_norms):.6f} nm") + print(f"Noisy trajectories: {len(noisy_norms)} data points") + print(f" Mean norm: {np.mean(noisy_norms):.6f} nm") + print(f" Std norm: {np.std(noisy_norms):.6f} nm") + + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/bond_length_issues.py b/scratch/bond_length_issues.py new file mode 100644 index 0000000..3baee0c --- /dev/null +++ b/scratch/bond_length_issues.py @@ -0,0 +1,69 @@ +import mdtraj as md +import numpy as np +import matplotlib.pyplot as plt +from jamun.metrics._chemical_validity import check_bond_lengths + +# a. Load trajectory and topology +traj_path_conditional = "/data2/sules/jamun-conditional-runs/outputs/sample/dev/runs/2025-08-13_20-22-36/sampler/ALA_ALA/predicted_samples/dcd/joined.dcd" +pdb_path_conditional = "/data2/sules/jamun-conditional-runs/outputs/sample/dev/runs/2025-08-13_20-22-36/sampler/ALA_ALA/topology.pdb" +traj_path_unconditional = "/data2/sules/jamun-conditional-runs//outputs/sample/dev/runs/2025-08-19_18-56-30/sampler/ALA_ALA/predicted_samples/dcd/joined.dcd" +pdb_path_unconditional = "/data2/sules/jamun-conditional-runs//outputs/sample/dev/runs/2025-08-19_18-56-30/sampler/ALA_ALA/topology.pdb" +md_traj_path = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.xtc" +md_pdb_path = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" +# Load trajectory with topology +traj_conditional = md.load(traj_path_conditional, top=pdb_path_conditional) +traj_unconditional = md.load(traj_path_unconditional, top=pdb_path_unconditional) +md_traj = md.load(md_traj_path, top=md_pdb_path) +md_traj = md_traj[::28] +breakpoint() + +# b. Check bond length issues +tolerance = 0.1 # 20% tolerance (commonly used value) +bond_length_issues_conditional = check_bond_lengths(traj_conditional, tolerance=tolerance) +bond_length_issues_unconditional = check_bond_lengths(traj_unconditional, tolerance=tolerance) +bond_length_issues_md = check_bond_lengths(md_traj, tolerance=tolerance) +breakpoint() +print(f"\nBond length analysis (tolerance: {tolerance*100}%):") +print(f"Number of frames analyzed: {len(bond_length_issues_conditional)}") + +# Convert to numpy array for easier analysis +issues_array_conditional = np.array(bond_length_issues_conditional) +total_issues_conditional = np.sum(issues_array_conditional) +cumulants_conditional = np.array([np.sum(issues_array_conditional[:i])/np.sum(issues_array_conditional) for i in range(issues_array_conditional.shape[0])]) + +issues_array_unconditional = np.array(bond_length_issues_unconditional) +total_issues_unconditional = np.sum(issues_array_unconditional) +cumulants_unconditional = np.array([np.sum(issues_array_unconditional[:i])/np.sum(issues_array_unconditional) for i in range(issues_array_unconditional.shape[0])]) + +issues_array_md = np.array(bond_length_issues_md) +total_issues_md = np.sum(issues_array_md) +cumulants_md = np.array([np.sum(issues_array_md[:i])/np.sum(issues_array_md) for i in range(issues_array_md.shape[0])]) + +breakpoint() +# Create histogram +plt.figure(figsize=(12, 8)) + +# Main histogram +plt.subplot(1, 2, 1) +plt.hist(issues_array_conditional, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor='black', color='blue', label=f'KALA-JAMUN') +plt.hist(issues_array_unconditional, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor='black', color='red', label=f'JAMUN') +plt.hist(issues_array_md, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor='black', color='green', label=f'Reference MD Trajectory issues') +plt.legend() +plt.xlabel('Fraction of Bonds with Issues', fontsize=14) +plt.ylabel('Number of Frames', fontsize=14) +plt.ylim(0, 5.0e4) +plt.title(f'Distribution of Bond Length Issues\n(Tolerance: {tolerance*100}%)', fontsize=14) +plt.grid(True, alpha=0.3) + +# Time series plot +plt.subplot(1, 2, 2) +plt.plot(np.linspace(0,1,issues_array_conditional.shape[0]), cumulants_conditional, alpha=0.5, color='blue', label='KALA-JAMUN', linewidth=5) +plt.plot(np.linspace(0,1,issues_array_unconditional.shape[0]), cumulants_unconditional, alpha=0.5, color='red', label='JAMUN', linewidth=5) +plt.plot(np.linspace(0,1,issues_array_md.shape[0]), cumulants_md, alpha=0.5, color='green', label='Reference MD Trajectory', linewidth=5) +plt.legend() +plt.xlabel('Prop. of trajectory length', fontsize=14) +plt.ylabel('Fraction of issues arising', fontsize=14) +plt.title('Bond Issues Over Time', fontsize=14) +plt.grid(True, alpha=0.3) + +plt.savefig("bond_length_issues_conditional_traj.png") \ No newline at end of file diff --git a/scratch/check_trajectory_length.py b/scratch/check_trajectory_length.py new file mode 100644 index 0000000..ffb8e1f --- /dev/null +++ b/scratch/check_trajectory_length.py @@ -0,0 +1,78 @@ +#!/usr/bin/env python3 +""" +Check the raw trajectory length in xtc files. +""" + +import logging +import os +import sys +import mdtraj as md + +# Set up logging +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("traj_length_check") + +def check_trajectory_lengths(): + """Check the length of trajectories in raw xtc files.""" + + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" + pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + + py_logger.info("CHECKING RAW TRAJECTORY LENGTHS") + py_logger.info("=" * 50) + + # Get a few trajectory files to sample + import glob + xtc_files = glob.glob(os.path.join(dataset_root, "*.xtc")) + + py_logger.info(f"Found {len(xtc_files)} total xtc files") + py_logger.info("Checking first 10 files...") + + lengths = [] + + for i, xtc_file in enumerate(xtc_files[:10]): + try: + # Load trajectory with topology + traj = md.load(xtc_file, top=pdb_file) + length = traj.n_frames + lengths.append(length) + + filename = os.path.basename(xtc_file) + py_logger.info(f"{i+1:2d}. {filename}: {length} frames") + + except Exception as e: + py_logger.error(f"Error loading {xtc_file}: {e}") + + if lengths: + py_logger.info("-" * 50) + py_logger.info(f"Statistics from {len(lengths)} files:") + py_logger.info(f" Minimum length: {min(lengths)} frames") + py_logger.info(f" Maximum length: {max(lengths)} frames") + py_logger.info(f" Average length: {sum(lengths)/len(lengths):.1f} frames") + py_logger.info(f" All lengths: {sorted(set(lengths))}") + + # Show how subsampling affects this + py_logger.info("\nEffect of subsampling (with lag requirements):") + original_length = sum(lengths) / len(lengths) + + # The lag requirements mean we need at least total_lag_time * lag_subsample_rate frames + # to get any output, and then we lose some frames at the beginning + test_cases = [ + {"subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1}, + {"subsample": 5, "total_lag_time": 5, "lag_subsample_rate": 1}, + {"subsample": 10, "total_lag_time": 5, "lag_subsample_rate": 1}, + {"subsample": 20, "total_lag_time": 5, "lag_subsample_rate": 1}, + ] + + for params in test_cases: + # Estimate how many frames we'd get after subsampling and lag filtering + # The algorithm starts from frames that have enough history + min_start_frame = (params["total_lag_time"] - 1) * params["lag_subsample_rate"] + available_frames = max(0, original_length - min_start_frame) + subsampled_frames = available_frames // params["subsample"] + + py_logger.info(f" subsample={params['subsample']:2d}: ~{subsampled_frames:.0f} frames " + f"(from {original_length:.0f} original)") + +if __name__ == "__main__": + check_trajectory_lengths() \ No newline at end of file diff --git a/scratch/diagnose_mdtraj_dataset.py b/scratch/diagnose_mdtraj_dataset.py new file mode 100644 index 0000000..f1c8fab --- /dev/null +++ b/scratch/diagnose_mdtraj_dataset.py @@ -0,0 +1,225 @@ +#!/usr/bin/env python3 +""" +Diagnostic script to understand MDtrajDataset subsampling and lag behavior. +""" + +import os +import sys +import numpy as np +import mdtraj as md +from pathlib import Path + +# Add the project root to the path +sys.path.insert(0, '/homefs/home/sules/jamun') + +from jamun.data._mdtraj import MDtrajDataset, get_subsampled_indices + +def test_trajectory_loading(): + """Test basic trajectory loading without any subsampling.""" + print("=" * 60) + print("TESTING BASIC TRAJECTORY LOADING") + print("=" * 60) + + # Use one of your actual trajectory files + traj_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.xtc" + pdb_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb" + + print(f"Loading trajectory: {traj_file}") + print(f"Loading topology: {pdb_file}") + + # Test direct mdtraj loading + direct_traj = md.load(traj_file, top=pdb_file) + print(f"Direct mdtraj load: {direct_traj.n_frames} frames, {direct_traj.n_atoms} atoms") + + # Test MDtrajDataset without any subsampling parameters + print("\n--- Testing MDtrajDataset with default parameters ---") + dataset = MDtrajDataset( + root="/data2/sules/ALA_ALA_enhanced_full_grid/train", + traj_files=["swarm_1ps_000_001.xtc"], + pdb_file="swarm_1ps_000_001.pdb", + label="test_basic", + verbose=True + ) + + print(f"MDtrajDataset length: {len(dataset)}") + print(f"Dataset trajectory frames: {dataset.traj.n_frames}") + print(f"Dataset trajectory atoms: {dataset.traj.n_atoms}") + + # Check if there are any default parameters being set + print(f"Dataset num_frames param: {getattr(dataset, 'num_frames', 'Not set')}") + print(f"Dataset start_frame param: {getattr(dataset, 'start_frame', 'Not set')}") + print(f"Dataset subsample param: {getattr(dataset, 'subsample', 'Not set')}") + +def test_subsampling_behavior(): + """Test different subsampling scenarios.""" + print("\n" + "=" * 60) + print("TESTING SUBSAMPLING BEHAVIOR") + print("=" * 60) + + base_params = { + "root": "/data2/sules/ALA_ALA_enhanced_full_grid/train", + "traj_files": ["swarm_1ps_000_001.xtc"], + "pdb_file": "swarm_1ps_000_001.pdb", + "label": "test_subsample", + "verbose": True + } + + # Test 1: Explicit num_frames + print("\n--- Test 1: Explicit num_frames ---") + dataset1 = MDtrajDataset(**base_params, num_frames=100) + print(f"With num_frames=100: {len(dataset1)} frames") + + # Test 2: Explicit num_frames = -1 (should load all) + print("\n--- Test 2: num_frames=-1 (load all) ---") + dataset2 = MDtrajDataset(**base_params, num_frames=-1) + print(f"With num_frames=-1: {len(dataset2)} frames") + + # Test 3: No num_frames specified + print("\n--- Test 3: No num_frames specified ---") + dataset3 = MDtrajDataset(**base_params) + print(f"With default num_frames: {len(dataset3)} frames") + + # Test 4: Explicit subsample + print("\n--- Test 4: With subsample=2 ---") + dataset4 = MDtrajDataset(**base_params, num_frames=-1, subsample=2) + print(f"With subsample=2: {len(dataset4)} frames") + +def test_lag_subsampling(): + """Test lag-based subsampling behavior.""" + print("\n" + "=" * 60) + print("TESTING LAG SUBSAMPLING") + print("=" * 60) + + # First get the actual trajectory length + traj_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.xtc" + pdb_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb" + direct_traj = md.load(traj_file, top=pdb_file) + print(f"Actual trajectory length: {direct_traj.n_frames} frames") + + # Test the get_subsampled_indices function directly + print("\n--- Testing get_subsampled_indices function ---") + + test_cases = [ + {"N": 50, "subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1}, + {"N": 250, "subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1}, + {"N": 50, "subsample": 1, "total_lag_time": 2, "lag_subsample_rate": 1}, + ] + + for i, params in enumerate(test_cases): + print(f"\nTest case {i+1}: {params}") + try: + indices = get_subsampled_indices(**params) + print(f" Result: {len(indices)} valid starting points") + if len(indices) <= 5: + print(f" Indices: {indices}") + else: + print(f" First 3 indices: {indices[:3]}") + print(f" Last 3 indices: {indices[-3:]}") + except Exception as e: + print(f" Error: {e}") + + # Test actual MDtrajDataset with lag parameters + print("\n--- Testing MDtrajDataset with lag parameters ---") + + base_params = { + "root": "/data2/sules/ALA_ALA_enhanced_full_grid/train", + "traj_files": ["swarm_1ps_000_001.xtc"], + "pdb_file": "swarm_1ps_000_001.pdb", + "label": "test_lag", + "verbose": True + } + + # Test with different configurations + lag_configs = [ + {"total_lag_time": 5, "lag_subsample_rate": 1}, + {"total_lag_time": 5, "lag_subsample_rate": 1, "num_frames": -1}, + {"total_lag_time": 2, "lag_subsample_rate": 1}, + {"total_lag_time": 5, "lag_subsample_rate": 1, "subsample": 1}, + ] + + for i, config in enumerate(lag_configs): + print(f"\nLag config {i+1}: {config}") + try: + dataset = MDtrajDataset(**base_params, **config) + print(f" Dataset length: {len(dataset)}") + print(f" Trajectory frames: {dataset.traj.n_frames}") + if hasattr(dataset, 'hidden_state') and dataset.hidden_state: + print(f" Hidden states: {len(dataset.hidden_state)} sets") + if len(dataset.hidden_state) > 0: + print(f" Hidden state 0 length: {len(dataset.hidden_state[0])}") + print(f" Lagged indices available: {dataset.lagged_indices is not None}") + except Exception as e: + print(f" Error: {e}") + +def test_configuration_parsing(): + """Test how the configuration parameters are being processed.""" + print("\n" + "=" * 60) + print("TESTING CONFIGURATION PARAMETER PROCESSING") + print("=" * 60) + + # Simulate the exact configuration from your experiment + print("Simulating experiment configuration:") + config = { + "root": "/data2/sules/ALA_ALA_enhanced_full_grid/train", + "traj_pattern": "^(.*).xtc", + "pdb_pattern": "^(.*).pdb", + "subsample": 1, + "total_lag_time": 5, + "lag_subsample_rate": 1, + "max_datasets": 1 # This limits to 1 dataset + } + + print(f"Config: {config}") + + # This should be what parse_datasets_from_directory creates + from jamun.data._utils import parse_datasets_from_directory + + print("\nCreating datasets with parse_datasets_from_directory...") + try: + datasets = parse_datasets_from_directory(**config) + print(f"Number of datasets created: {len(datasets)}") + + for i, dataset in enumerate(datasets[:3]): # Show first 3 + print(f"\nDataset {i}: {dataset.label()}") + print(f" Length: {len(dataset)}") + print(f" Trajectory frames: {dataset.traj.n_frames}") + print(f" Has hidden states: {dataset.hidden_state is not None}") + if dataset.hidden_state: + print(f" Hidden states count: {len(dataset.hidden_state)}") + + except Exception as e: + print(f"Error creating datasets: {e}") + import traceback + traceback.print_exc() + +def main(): + """Run all diagnostic tests.""" + print("MDTRAJ DATASET DIAGNOSTIC SCRIPT") + print("=" * 60) + + # Check if files exist + test_files = [ + "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.xtc", + "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb" + ] + + for file_path in test_files: + if os.path.exists(file_path): + print(f"āœ… Found: {file_path}") + else: + print(f"āŒ Missing: {file_path}") + return + + try: + test_trajectory_loading() + test_subsampling_behavior() + test_lag_subsampling() + test_configuration_parsing() + + except Exception as e: + print(f"\nāŒ ERROR: {e}") + import traceback + traceback.print_exc() + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/explore_fake_enhanced_dataset.py b/scratch/explore_fake_enhanced_dataset.py new file mode 100644 index 0000000..b4da660 --- /dev/null +++ b/scratch/explore_fake_enhanced_dataset.py @@ -0,0 +1,172 @@ +#!/usr/bin/env python3 +""" +Script to explore the fake enhanced dataset using parse_datasets_from_directory. + +Specifically explores how many trajectories we get when: +- subsample_rate = 10 (called 'subsample' in the function) +- total_lag_time = 5 +- lag_subsample_rate = 1 +""" + +import logging +import os +import sys + +import dotenv + +# Set up logging +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("fake_enhanced_exploration") + +# Load environment variables +dotenv.load_dotenv(".env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") + +# Add jamun to path if needed +project_root = os.path.abspath(".") +if project_root not in sys.path: + sys.path.insert(0, project_root) + +import jamun +import jamun.data + +def explore_dataset_parameters(): + """Explore the fake enhanced dataset with specified parameters.""" + + # Dataset parameters as requested + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized" + pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + traj_pattern = "^(.*).xtc" + + # Subsampling parameters as specified by user + subsample_rate = 10 # Called 'subsample' in the function + total_lag_time = 5 + lag_subsample_rate = 1 + + py_logger.info("=" * 60) + py_logger.info("EXPLORING FAKE ENHANCED DATASET") + py_logger.info("=" * 60) + py_logger.info(f"Dataset root: {dataset_root}") + py_logger.info(f"PDB file: {pdb_file}") + py_logger.info(f"Trajectory pattern: {traj_pattern}") + py_logger.info(f"Subsample rate: {subsample_rate}") + py_logger.info(f"Total lag time: {total_lag_time}") + py_logger.info(f"Lag subsample rate: {lag_subsample_rate}") + py_logger.info("=" * 60) + + # Parse datasets for each split + for split in ["train", "val", "test"]: + py_logger.info(f"\n--- Exploring {split.upper()} split ---") + + try: + datasets = jamun.data.parse_datasets_from_directory( + root=f"{dataset_root}/{split}", + traj_pattern=traj_pattern, + pdb_file=pdb_file, + as_iterable=False, + subsample=subsample_rate, + total_lag_time=total_lag_time, + lag_subsample_rate=lag_subsample_rate, + max_datasets=None, # Load all datasets to get full count + verbose=True + ) + + py_logger.info(f"Number of datasets found: {len(datasets)}") + + if datasets: + # Analyze first dataset in detail + first_dataset = datasets[0] + py_logger.info(f"First dataset label: {first_dataset.label()}") + py_logger.info(f"Number of frames in first dataset: {len(first_dataset)}") + + # Check hidden state structure + sample_data = first_dataset[0] + if hasattr(sample_data, 'hidden_state') and sample_data.hidden_state: + py_logger.info(f"Hidden state length: {len(sample_data.hidden_state)}") + py_logger.info(f"Shape of first hidden state: {sample_data.hidden_state[0].shape}") + else: + py_logger.info("No hidden state found (expected for regular subsampling)") + + # Calculate total trajectories across all datasets + total_frames = sum(len(dataset) for dataset in datasets) + py_logger.info(f"Total frames across all datasets: {total_frames}") + + # Estimate original frames before subsampling + original_frames_estimate = total_frames * subsample_rate + py_logger.info(f"Estimated original frames (before subsampling): {original_frames_estimate}") + + # Show some dataset labels + py_logger.info(f"First 5 dataset labels: {[ds.label() for ds in datasets[:5]]}") + if len(datasets) > 5: + py_logger.info(f"... and {len(datasets) - 5} more datasets") + + except Exception as e: + py_logger.error(f"Error processing {split} split: {e}") + continue + + py_logger.info("\n" + "=" * 60) + py_logger.info("EXPLORATION COMPLETE") + py_logger.info("=" * 60) + +def compare_with_different_parameters(): + """Compare trajectory counts with different subsampling parameters.""" + + py_logger.info("\n" + "=" * 60) + py_logger.info("PARAMETER COMPARISON") + py_logger.info("=" * 60) + + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" # Just use train split + pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + traj_pattern = "^(.*).xtc" + + # Test different parameter combinations + test_cases = [ + {"subsample": 1, "total_lag_time": None, "lag_subsample_rate": None, "desc": "No subsampling"}, + {"subsample": 10, "total_lag_time": None, "lag_subsample_rate": None, "desc": "Subsample 10, no lag"}, + {"subsample": 10, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "User's requested parameters"}, + {"subsample": 10, "total_lag_time": 3, "lag_subsample_rate": 1, "desc": "Different lag time"}, + {"subsample": 5, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "Different subsample rate"}, + ] + + for i, params in enumerate(test_cases): + py_logger.info(f"\nTest case {i+1}: {params['desc']}") + py_logger.info(f"Parameters: subsample={params['subsample']}, total_lag_time={params['total_lag_time']}, lag_subsample_rate={params['lag_subsample_rate']}") + + try: + # Limit to first few datasets for speed + datasets = jamun.data.parse_datasets_from_directory( + root=dataset_root, + traj_pattern=traj_pattern, + pdb_file=pdb_file, + as_iterable=False, + subsample=params['subsample'], + total_lag_time=params['total_lag_time'], + lag_subsample_rate=params['lag_subsample_rate'], + max_datasets=3, # Limit for speed + verbose=False + ) + + if datasets: + frames_per_dataset = [len(ds) for ds in datasets] + total_frames = sum(frames_per_dataset) + py_logger.info(f" -> {len(datasets)} datasets, {total_frames} total frames") + py_logger.info(f" -> Frames per dataset: {frames_per_dataset}") + + # Check if lagged data exists + sample = datasets[0][0] + if hasattr(sample, 'hidden_state') and sample.hidden_state: + py_logger.info(f" -> Hidden state length: {len(sample.hidden_state)}") + else: + py_logger.info(f" -> No hidden state") + else: + py_logger.warning(f" -> No datasets found") + + except Exception as e: + py_logger.error(f" -> Error: {e}") + +if __name__ == "__main__": + # First explore with user's specific parameters + explore_dataset_parameters() + + # Then compare with different parameters + compare_with_different_parameters() \ No newline at end of file diff --git a/scratch/explore_fake_enhanced_minimal.py b/scratch/explore_fake_enhanced_minimal.py new file mode 100644 index 0000000..872a1b0 --- /dev/null +++ b/scratch/explore_fake_enhanced_minimal.py @@ -0,0 +1,151 @@ +#!/usr/bin/env python3 +""" +Minimal script to explore fake enhanced dataset trajectory counts. + +Answers: How many trajectories do we get when subsample=10, total_lag_time=5, lag_subsample_rate=1? +""" + +import logging +import os +import sys + +import dotenv + +# Set up logging +logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) +py_logger = logging.getLogger("minimal_exploration") + +# Load environment variables +dotenv.load_dotenv(".env", verbose=True) + +# Add jamun to path +project_root = os.path.abspath(".") +if project_root not in sys.path: + sys.path.insert(0, project_root) + +import jamun +import jamun.data + +def quick_exploration(): + """Quick exploration of dataset with limited scope.""" + + # Dataset parameters + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" # Just train for speed + pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + traj_pattern = "^(.*).xtc" + + # User's parameters + subsample_rate = 10 + total_lag_time = 5 + lag_subsample_rate = 1 + + py_logger.info("MINIMAL EXPLORATION OF FAKE ENHANCED DATASET") + py_logger.info("=" * 60) + py_logger.info(f"Parameters: subsample={subsample_rate}, total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + py_logger.info("=" * 60) + + try: + # Limit to first 5 datasets for speed + py_logger.info("Loading first 5 datasets from train split...") + datasets = jamun.data.parse_datasets_from_directory( + root=dataset_root, + traj_pattern=traj_pattern, + pdb_file=pdb_file, + as_iterable=False, + subsample=subsample_rate, + total_lag_time=total_lag_time, + lag_subsample_rate=lag_subsample_rate, + max_datasets=5, # LIMIT for speed + verbose=True + ) + + py_logger.info(f"Successfully loaded {len(datasets)} datasets") + + if datasets: + # Analyze each dataset + total_frames = 0 + for i, dataset in enumerate(datasets): + frames = len(dataset) + total_frames += frames + py_logger.info(f"Dataset {i+1} ('{dataset.label()}'): {frames} frames") + + # Check first dataset in detail + if i == 0: + sample = dataset[0] + py_logger.info(f" Sample position shape: {sample.pos.shape}") + if hasattr(sample, 'hidden_state') and sample.hidden_state: + py_logger.info(f" Hidden state: {len(sample.hidden_state)} lag frames") + py_logger.info(f" First hidden state shape: {sample.hidden_state[0].shape}") + else: + py_logger.info(f" No hidden state found") + + py_logger.info("-" * 40) + py_logger.info(f"TOTAL FRAMES across {len(datasets)} datasets: {total_frames}") + py_logger.info(f"Average frames per dataset: {total_frames / len(datasets):.1f}") + + # Extrapolate to estimate full dataset + py_logger.info("\nESTIMATING FULL DATASET:") + py_logger.info("Assuming all datasets have similar sizes...") + + # We could try to count total datasets but that might be slow + # Instead, let's just report what we found + py_logger.info(f"With subsample={subsample_rate}, each dataset gives ~{total_frames/len(datasets):.0f} trajectories") + py_logger.info(f"Total lag time {total_lag_time} creates hidden states for conditional training") + + else: + py_logger.warning("No datasets found!") + + except Exception as e: + py_logger.error(f"Error: {e}") + import traceback + traceback.print_exc() + +def test_different_subsample_rates(): + """Test how trajectory count changes with different subsample rates.""" + + py_logger.info("\n" + "=" * 60) + py_logger.info("TESTING DIFFERENT SUBSAMPLE RATES") + py_logger.info("=" * 60) + + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" + pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + traj_pattern = "^(.*).xtc" + + # Test different subsample rates (keeping lag parameters constant) + test_params = [ + {"subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "No subsampling"}, + {"subsample": 5, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "Subsample 5"}, + {"subsample": 10, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "User's parameters (subsample 10)"}, + {"subsample": 20, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "Subsample 20"}, + ] + + for params in test_params: + py_logger.info(f"\nTesting: {params['desc']}") + py_logger.info(f" subsample={params['subsample']}, total_lag_time={params['total_lag_time']}, lag_subsample_rate={params['lag_subsample_rate']}") + + try: + # Load just one dataset for comparison + datasets = jamun.data.parse_datasets_from_directory( + root=dataset_root, + traj_pattern=traj_pattern, + pdb_file=pdb_file, + as_iterable=False, + subsample=params['subsample'], + total_lag_time=params['total_lag_time'], + lag_subsample_rate=params['lag_subsample_rate'], + max_datasets=1, # Just one dataset for speed + verbose=False + ) + + if datasets: + frames = len(datasets[0]) + py_logger.info(f" -> {frames} frames in first dataset") + else: + py_logger.info(f" -> No datasets found") + + except Exception as e: + py_logger.info(f" -> Error: {e}") + +if __name__ == "__main__": + quick_exploration() + test_different_subsample_rates() \ No newline at end of file diff --git a/scratch/inspect_model_minimal.py b/scratch/inspect_model_minimal.py new file mode 100644 index 0000000..2d7f0fd --- /dev/null +++ b/scratch/inspect_model_minimal.py @@ -0,0 +1,15 @@ +import sys +sys.path.insert(0, "src") + +import torch +from jamun.model.denoiser_conditional import Denoiser +print(f"Loading model from checkpoint...") +# Load model +model = Denoiser.load_from_checkpoint("/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-08_22-31-01/checkpoints/last.ckpt") + +# Print key info +print(f"Model: {type(model).__name__}") +print(f"Parameters: {sum(p.numel() for p in model.parameters()):,}") +print(f"Conditioner: {model.conditioner}") +print(f"Architecture: {model.g}") +print(f"Hyperparams: {dict(model.hparams)}") \ No newline at end of file diff --git a/scratch/load_model_state_dict.py b/scratch/load_model_state_dict.py new file mode 100644 index 0000000..91fc3a3 --- /dev/null +++ b/scratch/load_model_state_dict.py @@ -0,0 +1,23 @@ +import os +import hydra +from omegaconf import OmegaConf +import torch + +# Load the config +config_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/wandb/run-20250805_042516-yqn9mm7x/files/config.yaml" +cfg = OmegaConf.load(config_path) + +# Find the checkpoint file +checkpoint_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/checkpoints/last.ckpt" +# checkpoint_files = [f for f in os.listdir(checkpoint_dir) if f.endswith('.ckpt')] +# checkpoint_path = os.path.join(checkpoint_dir, checkpoint_files[0]) # or choose specific one +breakpoint() +# Instantiate the model using the config +model = hydra.utils.instantiate(cfg.model) +breakpoint() +# Load the state dict from checkpoint +checkpoint = torch.load(checkpoint_path, map_location='cpu') +model.load_state_dict(checkpoint['state_dict']) + +print(f"Loaded model: {type(model).__name__}") +print(f"From checkpoint: {checkpoint_path}") \ No newline at end of file diff --git a/scratch/my_histogram.png b/scratch/my_histogram.png new file mode 100644 index 0000000000000000000000000000000000000000..716b7834e537b849ac45fd5d1e0a09039da1cd3c GIT binary patch literal 164601 zcmeFZXIRzu)-7s|Cvlrtg4k$wQHmfSh@g;c5d;C1-cb;c9z?2?>}{fSB1J@+p~GT< zbVQ0qiY^2c>4+pCqSOG=+Z}Uh_WS0Z59ibUaG&SRvv)8EYpwsU%sJ*5V^05b>PqXs z=Kgxss#WWiPb+BP%h;+_U%mgwSNKkjSlw}ak-2(G@2aMw|K$17H%mir3>@G5J6IIZuzYL&=d z`u`X1De|tXzF4(NS>byv_t@cX4|lDWCFy0ie|+)Vo^N;kJ7Cwn|Mm4h4vq#aWS5Nn zsMBn~Z0*c!X-$zY?#wJK8clRz^qRP7z5c<%>9MiCc0~P;cDvV3E+m;wTohmXY@fv1 zx}mA#9`UawW@0BRX5t^JijN&T=C81a1E1r+{`s!vh|T-|`cL}P1^c}o8~^q#an*?( zfBVLzJ8w_??Hf-IKfCs~Z~SsGZ02v@*!5)J7@qaN{`u~P@|)HF^`EO&{kSf?Y|VfF z#wsrU|M?*QAFfF2s_SbuhQ!9kP9!|OzUJuF)3y=1rrwg)EB$ZZRxxra4D8F#h?XBo z*-gYXi#wZmiP{bl0RkFf{e5=_%|4?2iv~9~?&$o#YeZ$A`#fv3Pw`>X8?Q_1|N3VX~_N=&o zfB>^9KaTOIdqa<0J-F(eLqqJb@egSUH6K5`GRL2D-N$>MK6~ch$~07pk+M&`cW;;H z%qSm^W0}**n<|IYqQ&=DEKbzjk@NO^o**xOTf~_AkU@c8k$sn9SCM^^RbzZ^hh@y% zl@0WcyM{iW$whh1H1%LLsg7jDp?B}zDdt<&C+g>0zRWg$cu-h42w!U6#W?O*nx6{f zb9*iJ_19nL@rpefv>orQ&aD%9SXH~zH+HJxtBpH&bIPX;S{E{&M+j&>*I9l4?wNg? z#5_N~cOTBJ?XBKs8YG~Z+MI6k{rBJ76}IT)&b+x+%b~=}(f!b^@k(;DjCiIc1yjts1k7)Q4J`=0nWlkdqsZ?F%{f?#}xczn(myK;ddtmMypB=f$&4%*)Hu zE4Y0Ad7SJ~S_AjV`lEdn3!{PZ%j2O=e?DHLv@ktl{`;eUIZd_b*1Gk0x4rnovitYP z*Y_SckhW;`=kJfPJ6!iC#OkrW*!Y59Pnxa2^2g8Hbnce4m3d6I+vV3MnpXNK%gY^Z z9DLuRYpV8OYsLH^V?E#2b0_PfL_=lA$|R(^?DE^K61;m`4IPT@(Lr zL`rsHDL7Vtwa)M1XsM9OnKSv@B|jV#5bz($sgx^q9pP5m;c@Bs#}_9hRbB`B-I4e8 zDZ_J&51FpWNVQ+MzFqv%X)ioS$NKH`?0N&IKLwDd|3*~fXg$!oEv|n%W%6`yysws-u(;TdyQ9b8-KiJ=e{Lg-^KCun-8tB zO`=o%=wGY*A~eWilpH+Og*ys>fTjx$jr z%#TxLWo7M_G4^H)BgLV={r21Cwrt}9hi;jGz(DgiWgcP{tcW{@FaEP^WqE<_R>XYx zfwLi%%TqZ{&&6slUDHhet$gMUPg?916=B1otfn@@B0DbZo-Li-`_)naaIO%LZcD&pBCHn-fp zhd1997rIvBvkCD9ualHQFE>A}Oj-3bdmzDAPW{AGb3|3S4%47|^~U4h?)3We zdtQ&Xx7fGma%X`xulw)+V#ks;Q(wL(H#fS;+g@mtU-61{s^JpwcV!wB2Hf1~BmKr_ zVWb`L!1joQgNca=`?sD;G**>3^_iu`xw3TH>7zfAd2SlXKFWC)D2UL3)jWG*%SdP8 z8IL9TmASJ_{HkpsW#J4 z+njQ#iHV6pZb9~P9{UVC^soN@uQfIJz=ZzWVZ%7;@J9#;DO( z?p94U%I{oSn8|ToICkuq6TV+f-(eaSXFGpFJRBUsJTmlQR{dCLj04v9>&+bj<{8gaML*Y3C}eYkJ_Uw*SyffB{_ z^-_L5OY^5@M!Sb3Z@Vb#>+7rDIlMbnHJY+X$lT~E2S1%Om5w0fX z|HbOB6p|DxR>U}QMt5)?P!FJ*S7cPF$M~!WLd)u5=od zxjU6Dj6P&9shb=4@~)T~i<#i&KcwukdD}K6A0MBB%P;p( z-g@6zXlqy0Vfhk|A2c1aw#RqHo8m3SUW)FUH*c=`ipTC-?1!4&{l{Oe=QAmv|MSVl zJ|rkrfy(Jl+a$NKH(X;mI5XohJ&XKF_FK1XsYW_5^{F?l@baJ+z#Wd78t{MMJWzu) z%-Q2TWBfe+ge=aMd0+6+kbkXSQ^Ox)7uHf?Yl8#jw6Zjl7C|X31c$|X6}9dO|NG3tD;lbOp9;B42;=jd|dqB0sgG-AAt9T2A=+n{74ix#oRcj+MbXeHX>< zO4)PA$+!eyDcibKZxI&o=R8o;oS|8lAl7E+SWa2uq&%aONoST~y5Yw7| zRpGr*^{+KM?v&s>u#r+h{oMl1?nXc_jZWsUbP(^cyv3PaeLTw12kbOSl zjeva!mju5>xtMv4*25l=2XDh7JcX`d>p77@-1ZzjTFkQNJ$&K2-svvKv%;noWu41{ zneDCnqm05_c~C%!*|p!dnK{4b{G5-3ULpJ+^JJyr4me*Mofb>FSF72W}$6m#K z6*&I}Q<~i#M=hV^cJ#Ec zQOV&HCM$dC!*bSf{>(|7*aJALISQ{*)pR@FZ8fyJRq{6M%qD=GoEABQmJ=3#{Pwdy zBfC`aCBmoRLe9bPwkcrNs>+q6u}^E$n$;mV^MuZ+^vdES!%Hs1f2?vv&IurtA8Fno zsjnkhIlSuL_t!W1&OX_>3DG9YrBYJ6+k0WWcIR2U%B5-Tny&QoNmUK9W3lcZUn<$Y z9$0%I_Lrubx6i_d3i$8~s{SLuR~ecy+Q&X5hil9a0(|SttLVDaPIg!&s0}|Dh}V`o ztK(|YGFATgYs5PnGpwoF5u^v!cyrp>geu&XxR1i&8OyJnSORIuRYf0CWB8HZ^4_?q z1?@N?Ls+FuvRmnQ$m9mBY@ZE{IdPg5lY^|~d zjzJ*ha9eK*2^VkjZGAXaag8u^?xkm8+$}K@R_e!>KE84q%t)(=wu+A~oNb!=`08ve zAmLVBV;60FoQ>IzHqj7?2I;*;?Ph@`lMS-wxcCesU&QqBN2@nh5h_Or4odKu z&xqe+Jn4l$ zA`reww}0;vhGd3cxOR1@^^i#E`$WQB`!@M&XUkvw?Wft69G@*Q_C@+-6D*;^(QEX? z0Qa@*8eq0oz895oW>EvOaUYIKpr-n<{W30tT0<2>Hnd=uU)+-(OiwaHv}-X*qU|dg z&Z|qjHZ$skduAydg2z5eCF?d5_lv6W$B&thv^Eo@@SgwZkBn6F{$&znCppq5P%GV_qa|QYO7Ye(_If?0w%zaZmP@ z^8UK0T2#gyI3VupxWxBiRWS{Y2e@tUk}M-t-MCuU}Um>vZ8tQRoPz(V7k zg^G$wZw-%Q5Fj6;!T&Nn`O*VMPrWwy9g~|ok7j^I>&__?tn*4!n%GNF~^&EXLU>0TU z>Oa%tD?d`|{O-9wN|zH2atp%-2fVZXoFB@m1slLK%)5fT!o7y~hwr@RKcrc=EzJ!i zur~l_J*B30CDeD(oPtFQe_K`v)sgtffweoX2{S#<%FSH@)v==`KB3Fup*nWg^*Odx)Ff7`6%dK4_nHM(#Hqlo z)jO^YbA-yxUoM|1r-S$Y-t}HYcGaQi!NN9^Kx5aAl<&jlO+ES}BPNSbzf)1?8ERqC zR~s6NDp!neDWNO)xk{fU zmhTF-A>49{Eg~Z>MpEbYm?PF2N%qnB_v@>#)cTdbCRqJSU6#EzJoW|W-wq5Ao3o9e zH&9n=dVS}7SI_e8Ba=!D9cJ$@CpY-hX=pX|T{6*aoKBrL0F zC#ZHTuztnQaB5XVP6-bgqGa`P9-n3>pRB8q1rLwi{qJk~?n>I~3>Umok8Idp{8EvJ zopH)Nrvm={?|VApT%^7lz`rdhZtv$P#;CO zBoyiA_4Kwo4>SZf$}I@f6U`zjgty{)pEXC>8n97rb{CjB!4cbjC(c{TlD{HZez5Db zu>xzyPSL>+4i&!^2g<8_|EBy}fqj=0p?XuFIg21}$;{5_D_5>0o=uQ@iJbqwg`rC| z6G*M{r}g#0MBwnBjpqeA@xc@EBM7jDB2nJ?aqKlCq>QTQl7*qJj)ul2LX{~J%PwPY zPT*t}zwat`1YlP{ULrzs93Q$Ccb5m%nTTm|LBSDq#Dn)h*D1=Qk?salvc+&^35*aq-SEXUj4`xAEV;5lwy_<%q32C?pgZ<4|fH znXA>0li>GcgRHQC^`#g0z9-UBDN4jRQJbM7h9EMF^|n3Y@XF=!m)oi&+l<0BOwOJA z7L~j1I$r6+s2q||ww%|}+MM7up?n;$4)DTsAJZ50rXm>J-W=bhA;AU0-+ci>*{1~@ zQG5>LcIUR}7@itvh)Y5oAS5Bbvgiy_?*ye(s(N{@*7W%K^(;>pkK?FBi55Hd$Ze1v zzEA60fIlPIaiTVHI*!<>JfK^(vdMS`0M8vUa<8ap4ol8wieb8c|H*n(Njt!rNF!o5 zh2Y`F)Rc?n`$c`DU1cS6C%ip=rV1A~DX+`qwXZxeTw4Zu%WBnPM zG-TzePG6L4PJG9%Tq>Lb%!qGCV$~RYH!yG@8-=HWwY7E16eXeA?lGSzD$`WD!NaC0 zis|yU3GJ*4;gjW84CdCh=%&jrhG-B`(N$m_t_Nx%cs_@Lkj!jy{;<(^_GHR{GRm=t z+qe0bC!#9%A>`lHn8A)RoqD3n_%GqZ7;6+c0p1m6KAe!u1mEKiI8w?g1rh8IMmJ)D zATwRG1B7a-kQ8k;*Mhd-w2E`lVNLne9V*T`zl73xJD!F)^MvkDL}~f88Ff^;-%t|L z)ptagV^>?ZqN1X%_^2WRclr!Gj}9j~LZt^piHZ>tr$4S2GtUQ4PmtB$JoJRrt(cgH zEEn5rE9opS*_^0SE~>%q;2}TwPIH5`^adeT1#M0s! zL)PU)bUd>->Q;FMp)yOnMm62N#{rDyoW~og?f^(kE!F|CDjWUq_{+#QDUcGLW3&w?wqExF=2;DyoG*PVRO=9VzyMpX+RodvQTRkT6ap<+doI##2z> zb#1>$6uWlAX5yo=YVyo(B$6~US~`?1X!XJt2Y+$8xH1?obPBr=Zt@VAHQciCR~sbM z3q)lrTh!l`3|v{BsbsTJSDc-Nm{H(Uedp>-dr>B-dyG(Z6~uE?WJFJgcfp~%w3;|6 zC0-|m$e|*;4)IYiWfTO%Sb+fnz2FDc9*$iN=986jfF_Wfhig{-^wwU=2k8^XpCRA& zERDL20|TcRF$l1G`g^Mt^l}%@y^E=^%Sitz786GN@Y_fb+y3XalZfW2u4rN=Nj z8a~qs*0-9%ofa_-HA*|ZHzU;y3NEl9)ScXT;s_pm8zI3aTPiA9`Hf#geeokLH^UB6 z$9P-ia-m{=KFPXNN{jl=ogs#=FuuI72>89Oc!UZaBb3$JqiMiE)!MqoTc~`~G!>c{#cvg)&@P?@DhtjCcR^0xRbz&Qt@ zWqfh6u@86D?hTqjBMbcg?4& zxsLALQv}fC8GO;gccH6C!!lh2L^08Jo^d9>+IJO^a)4TK`Im`KM1+JZzUh~N?TPA0 zH5vkLFDn85OvDdHw$@>(5TOsndrx;vT)Si+`UkSlPH2Pub)cq0=^#+7)>>H11Ma9M z%t{bOV~)hr2(CI3J}%hr`Q%5qw?oq5wKTA$1`DN6SJ1Sk7}pQ4)Q$$DPAJr|*N*z% z=S@qme%OS&a<)}&HZBD6dfUOr`}XvL%?Jtoi0^zr??Y>Bq$y=A}Z4p>j*@8qjs&6GTeXBT>s)$;$`>9C9!onhQM;~B7>osAH71>aS z(hsQ-6Ufi9^V1`SNmAJ*y#WzFjez=633t$)CIt7*8b)-cJ(t>=L(S5LbJrbPZuP{*&u(8nCY+9ZU03}$0Qvv+4lsvjS(df!v&dwohCxG1?FqNjwdcR9lII4Y+7k;cADQ-vzo za<~xTmU2QOBI?oh&K7MZ*X+TsAI}&LJ0p^z-D~3g5J-?^^{3Gvs4&miz6R$ijqB%| z9HD#rv<4^pyGN4K_;u{^NUK zpY=)VjS%THve$CBobk)v2nc<4!6P+WkpH3-Qz0Wz3{&w(j-rB%c-!U2>d|bu zi3ok9pUVflSEI=)Y4DOQ|M%ku4RCzN>f15-;|mE&uZHj2w1%M z&8wXwq|XyAlbn~C-9x$lzSJ+DtWoD01^GEnv2#U|x7Ol!^W;1^QdfW15{WYq{(-z- zpZNLtK@kxd*>?+}JmtB&4i3F&`9}@@lZHb?Pp44*9;g`Hv@pXiz>rqe1cl(jpoB4U z$40^#Ro?Zm+m$$01zo3WNqCvqJp*o1F>vc4m<&)oHH|QbOTL9QVlnaig`7K=p|U(D z&yCK!=i6IKVYiRMg;8MHt*?*pJ`5RA*@BEQkwZkjDc3>c3o=F6D;#`Sq3t@_%s)*8 z3z5Icd5AY9P0<-T`%y8dqt+`}$O#1bC;+)Ij5T;=lnMsrI_=;%)~2bv=-=SMMciN6sX(vJ1FfBf$+i9&8pZ$z*a+JEeD%)n*s#utUk!*{`` zK=nwTk^tvzQ0?8Me4u4$T?hoxP@aQThOiGC$57}ODc51W1I$B?@#C4IUiu8yVI^b4omEd} z%Homi3W>&12T&Ft~^lCRXR}DnJ|h%WCqi@3o8XlDf4%ie2LGP+)wTxR9&4T z%~}>+E(mPRnM~E;Vwdl+MWlao+2@y(ND~28g0$2NF2>qRO9B$apOLteiiP%Kk_`m~kJVj$AtJ*r-nNu ziU4>G3U&4Wd9AwRTgEF$*b({{^O?Qv9X~u-p9Cab=#bOf9t~bw<;sPsl`z9^zWzFl z)qpSuoN&~p+s&KNI1qYb=CuSEqi8K+u64U~_uTSBqL$3a!u1zes}db?dS70vtCY5? zK|>KM#KXcoiQmRBrzlgVW-5BIBrl-xd-8&TaxK=ePnnFBbn2~&=rA0b++G2$N}x7T zWM(CxZOkuoY=wR4Qvb`lV=K6D&Lfrc6*6A}R60#a!1-mv03D$OZV5^dxTT9l)z(X& zd<)2E>-qlq0GxGerb&XFH>Gl7iRx-XcP3=qZ&lf1wWH&5QgEVqZAhVwE$Hr-2>!`$ zJ2EY$GE`6@+PE}m-PSvTxN^B8&!W@5MWK4c2IWC#eqPoK7z?0wsH)siajRs0gZ#sT zvYE?z2$qr%%4tMs$TZ$D=vYe4_CY?F1#b|>A`!;WzEGRll_hQd8mhUV@fc`Qb%8FY zT0YxuNALOGKv9h_S5#|52F1moE|VkFU>;!p6m+AcSVvW54&}t)5m&)8KgB~=VrAI| zb4v!0XaV*9oQV|t2yC&4IDQghrlE9S?y2x*nm;S*l=IbcjkTRv9A}s=2{y3N>(UZ$;wg}BE$S(m|x36oqIZSM?4{>HJ)(a$^j*D9h_e1NZ?uM< zUi*vP5naH+w%R$k%`ahboQ0!_RVUc6t9nIw-zEWxQ`4}uw2l;_>XC69<;A@h5L{h{`lh-6ZtFw0G482c!Szm#0H}`P zx3`I!@b2`Py{Ncw2fss-OUfcPJA^o^{SWN$4Jr!BS$U|aeGh8Hn{3kUG(wkayMwjUiz(*rf@1Q-d z*YZ z3;I%tr25nAT>~cAkTx;xsDVGOYP$ppoq69-q-t8nBWV*Mz9X|#UB~r}<%GAg%V6KK zY0DH3h0NBgpIzUBFOd_HKxU_uAvQy9gBcHX&+MKCXtp-;2g|(XCbrA_coBU=>4_vo zo2kbGc9#+1PToA1oq6eLXS1<~6i=9Tk>g!zyr*gS%N;>s?l4v6|tx z$r6Nu-YsS54Y+*)<(bOGeo@;Um-q-H3y=<#6IEfGv^(B*sIOyJjG(M(Z7`2+j+nxc z;Yn-n*Gf7myZ=fye}$${vg1YMk--^~b$;srh@|aQ5A_ibPvltxc9J+{*e-SS(~^_- zje_kW+MA4Cwz*7ykUtT*rC|6nZKZ9QtZ6mlX>W_Ew$>$6r-$O@J%U5mI|=Wh;+YZX z-)8i~_rwMu1-Mm#W!Ow|B#5;ApxZb>^qkSJtgvJ}Yiv9#@bHftJQ`tZhvqe-Sz#}> zi5T4(w|{;^)$~@exx0dK@D+{?r#+VuZe%n`>oVV zJ3snjO%-Z`{2e`a&A?}HF}m->)BI&*6_FG> zwdl3!y&hcWi=FfHN*N21Ob~PeGv)K#vf5n0XGX9*l73SK@Va@Y_s8>#@CpR&4UtZ4 zx>NywG|?-s;nSGn9zdaOWT9cJrWIQ#9ybsgkU!=<*B>B{n_}H0D z&GBVZEiJ79D^-dJdb*!Y+UW;Oy{EK^J{0I*!*})EY6Z@hi2d#0Hv16gg5BPD9Kbqm zWiwbo>QtrUD*0OfvRzk+Bsu0Fss>?P?7LbzdU_=+nP8ry>IT{No8H{L;}s3VY*;$^ zcUnrzNjHZ3Bb+68=AQ;U6zi-#z_mj|G8&qeLjjW9UkKAs1dNmi4jr{=K1ddjg0AS} z9#>NTX^=vYKV=vidS+CGEC{;`-fh~|lj4aEZ3O`X7FBnh9CUc)=@Z3epN$4tn#498 zpcoLdX-+2%lt?UUGDyj~2IQrL16*E>8q8qG*$W?o{65KfgYoh@u4KtaS_)@5Jm|lz zEm91KK*pFOdzs~gboXa?!3RP<+g8JMjE4cWWr(L0sHqJK3n7L_SY_IEfv!))^M3LH zAs#+7=_A0!ipp;8H&}+yRs}L$b3R?wVqgR(;QUYp67^(3tK-uJdD@<=4}V>UaPr{y zKZ%A?8%%ZNzxJpUk-!V*J{JPF87s(&g!t8`3g@{HxS4|xT$Kc#x#21rS;Ph8FKfj{ zC#ACeZY{Q#;qgChungqUhXOTA+Yo{9?6H;1t`*zNfCVO=gHnw_p(N-CS~87}&e9%k z7K8Wxi2 zucM>W_DfR}Fqt&OMopF@_A_GOp`8tYjU9GO*}|n`=Le%Y36FKY$S4sV;tb94c|MLv zAF=hc3%{s*^2C34g6%Sg<|ojPb@NKeA3u@|SHUq0sYhhRgE~cPEm74nv&M7i73bOG z=NX6B}>)%wRfq)*h%g#$gMyD(^vOLCQWU&~z)oa1L8B8v|yF{W^`m zadvjr+_X~;XNfGzHe1^6dOD1P;}WnSbhTPR!2Xe$xS=jo4nYgJM#SiPodz#yb*-+`& zqHDwfz*6Vg0~uD?zyvkCPPnt6$r{~{`aF@wuPAsFba|jmCZ{+UzJ+OKOHlBXJ$)!; zIS(ArK0vR__xx=5y9W~O+<^VWDL4 z#c2rFsz3ozi@GaPXCeAUrSFRLyF2C@tVXcwv+x`Zo``R$hrfWA#7DtvkZm!dsp8M94oXRs4K$wl2{TC7TP?3KO01 zA#jo+vRXmAJV1FDM7@_VjtI6nQWl;dhp^>E{g_xFU2En_RORA+ddg2K5X^i9EYB_S zyJW%exg#;!z8*t5>BB>2dEDEgbv(N;dM-zVyy_qyTr@hs*+opRNft#GxpUOfM56;X zqb)!G{Ig4llS{$luXw896uYZ64U1fAoUqrE^+%KUH4~6|iN%zueY%q2&c}8F4K`;; zX0PDKYC!2k{gJLS0$J;{aznJZg-T~zO{er&*`!98CJL5`paTwB=f-8WpI@;31NS91 zti+X=ZprJ0$TqPqt@;A}k&(<0Ac&nLDQH0X)8}=2?L?l?uL3eXBlp*~&OCocmOoOt zm@)(cSN=)%7NQ0J%^;snlydN`&biV1RF z-3WUG9P;bKSn;UHNmzI_2>W+~N_62Uv6N_D5**T!V)T>3T`*sc)4)(w&1yKc%{W5C zmAWwy7W5g@bki>w@BSpu?D~cTF=Fv~!9)q{vEi5VS5)Bw%COb?lBIm*OI;rqBBxlNEgZobv8Xa z?ZsfQipFS_fbdr`&MyJ~{MxkfGHEzWXS;n}I18F=B8wZexz%xBwgCsw6_Uwpqwlh2nrKw|oqs$?<$^tsoN5`v z@z+K(#-sKe2j5RV*-N|g?g|MDXMNmGc4T&6CZev+P*E*1J{2{K$YZAc`L5`3N*U|B zv7Tou2~5HB^RE(?H$&A|)j$nX-wXSzo<;fU4bs%DcuQ*@?xIU}`w zRynobzewGQmXh%#v!Kj)*beHKZTF=PM#U*QjP0N-&B0_V_qXi-60GkA0gJz^bgjy~ z491pxeq*QVEo9afwCdLbB1g1b?*tMj#cP#26ku8mt*$oMqrPRwX<+()k`W-1|L5$( zAUa5D?jz_FaLgE|7m{2bwj~6{x-6ZMmKqp3d8qkNqXTih7g5O%*02oLcp=I}IvVD% zy$S4WmdDa)SV|gou|MKkZfXMoMiC=(hF+1Ke##U~^1ZMb1nxe0BRR|Y{jWDupTAe? zCLqR6zImu{rAbWI=C7+D=PYb6)u@kD9QD+mog%8I+vSHd4ZZYy2ZzaooaA!39 zYwWU(&$WFwaaVL6GnvlFiHa~!hp{NSx1);eyK9@BA@$gj9H4k?I^Jf`ZH@a&ks)&N zLkA-9S!S(ZT+)0Y@1k}1xS3(}qn7!|Ac_;(-~%?m@4 zN=4(P(cQr9CA;R4&%uT#w;9;)XxihxpCt4aFuT(PmdJ z&qqvq?{ zAdEBa$RKRB8#y_(##C+sj#Md&uR0+VRl(+_J9la=HReJP(Ym|J7q>V0`FmYdw86x2 zo#~h+5}V{7LC$-e{tCnZGtlGmzS49^CMB6D&@$VMV%WnvK>T_%V*K8u#P9^R);6O7y=do#^fNNW zVXtVnVKBoiQ_;G7o^SU%Ecf;dOhq2&+ zo+2y1IT?`Q%BV$`5(m&PR2ym(7jt}5#_qXaQ;0g^Fkzc zBC{)%#$;Y2^^Cm8RPxYufWLkxnXz-tYf~fL-Kcp4^jCrwJ8Hog#7C`ipoK2KmY(a$ zCf6oQEO@r(`6aScnWIKfFD+6BFs5dz?Px@!9$}IeF2fy9y@w`v2r7Pr)UxKNS%vC^ z;9?6GC)Ad;@Zi89)weU8l?JTLwlV;HP+w0wG6@-?aMA1I{mnS zJvnjuBXz0D*@3w3PMx-Zj@!twO>aleRF&EpG9-~_GOm}J`}zPd?Nesjz{p9D6ucU2 z_DLnnEipyXs8lp-p5mppEAhE#z3uO>hD>dSxD`m04ZSpkh7fAG#RK*tUxgwcAAbWd za0k>*q_;iv{*E`XLEJEk5N1&JZ$dX;_JenKsmSOo<58;+DY71OaCY`Vg%3uS7QV5LYZHb9_)$|g;OTZmJqPl@D9*LN5Z=c8Mv3ZiOw(>svunN0HWCN~=b z)AXh%g=9`GFCvCCg{&lpt7mHaA`GbzwxXWQ_I!K0cBA~7&3yi9Ozq%Q@@z6yC9W0g?zQu!7EA)zzW z3}aoW-+hN_8%~-gp!u3iyop&^Sv3m`vn6mO$u%xQ;rvP7#^us1DqAz!e;kfxR2p4^Fgo8rqLeQZs`Y0%PO*-1_=@ zHM)5I)N-e|v~BCw){IK(0iX)9py**%mM|6eZJCvH_r%Z>B@

5^RJK_fF%;xU)m zra85ygX~lCv;>!$Vo+ZuBbz*wWXCj53YN+&UIx#qHz2h^=KVc(8#)>&(qGOK6F~Qr zddS-zjJ4wfG%f`=d1x^blHV>QGlo^BAPg_FX7$m}&lz-@+1rb6Q5auo%W_p5m^JimM3}g^KrBjZw6*qI4gG_ahjk473$)3Vx z6??vKYT@GtPQ-YVSPXiAU^22i(g6kCrp+3N>^R(5P5?VCnV~GgSJc^1ix?B<+Koo@ zYRId+NT}w_;%hE!yoij+N6Qq~X^nHki_QaW69(1i(~|cSaT}`O$>0JiSC8w=U?P## zfNhVORt=<_L!(tF#s^i6*O@MhlZ*F|X*We_^1&9rzH){cUn!VR><<(TZlpdBUYF+c zT-Ny&{SeIA=2e8`SDD4sjD$2>VQ}J%Jj@&62cXbdI}!(4-Cre$_kQ5?6sF z&{7(QbG7}r`w!&#b_rQahfPP0=JEy6eySjk=+0?C{(MPEi*d;mdG z+_#;?fl%Z2)Q%qN$0UJzaVp0*7+ZY5W~~t=my1tsRO4uGCAYO5oM_fl*U(c~jpytG z1`eSyCluAGdxx6;M}A_o=#vSPU=PVnbmz4B&wd>1k#~S|l;`>6_Gxn{@BqJ)I?H7SQqdP5H&$G-aR!Mqp|c4xtVv5{jnR1P^5*xzk0j zrk)LQk`?aUF^11Z_X2LR{1RmGp%ZUfPT{_@KN4<4HF@yuuQ*~R1TPY@7>3)FpsPif zvGGGry+N@9A3ZAFx|is$(atsm%&A=H9s?^<;YZ?9{2rr1d}Y7BI|@2B=7)CKmQFr~ zUQjs!1Tg6{&*GSc=~RLlW6utOo=sVl0*1{HIo_yw$DPOHh>c zk#}~q`=Jt0pKV;hVf6pH;~r5TW&qLF=&~qfO`-hZ^F=}5OOappUtTUnWaq6Dt>guI zRL3Baf*MoT7l=Mo`}KlB@F%8f2Qnx8G+9@>sa~b|7UXruZdUq%0dIN#AQ_9#W)!gA zo7h(Yn9SPVFqLTAc!1LFuz9g-`s}Fia8AlK)XPT0Vp@;g^0k1cRedh;25d6-AJ3?1 zlS!87E$Yz0XLvyJcazF#QP`Wo5X*|7wUvA%y|{P*n6Xm$xU%_PHd#>s<*dq3ise#i zF8)Op=0XryuEL?zN&0DEz7?xF3*2qEyyGpU%kewJcn;%5oezp1 z-$nni>S@J?S825{k8;vDE_6r|S580=Sz{Z{4Z9QXWl}A-eBjxWcTgPT#> zcrO}tm8tflfiBHp!8N0v?*JeLDj2hn{b_Va`=xL-8lB+Er?;N^(nz+S1;tH74Rux! zt;uc$L>{sf-aQZFP!N%88FoN7F7bzokVA=RqgA1iT_CP`h^PKM4~0f;PidHDPSPR{F{lw4n%(P{k^-daMv1mca^h=@|wcHW|ObZ99nR`Ef_94>e zs9~?JwgG@95%A?D@*zD6y_7cpU`sQRKlHfI!VlEqG0ZTu0Rq=n+DIz+Xl?*hy9_Cg zwB6od_1OK32r|^le;$p0RNH{uOIiAIo#YgqEJU7gEmzPcH8epZayiceLH=(e-2f@5 z-HJ#(@*Kcj6^x`QY%Vd)9Hmu(TU^HrNR^-tWeD~rZs5HWY1J|RC+$8-L*&~;`%o1K z73vtIN|?sVps^EF%L(!2FZV)32_%RD;)jDq;Sd1;2Q^I$%I&{x*y?e+Gm~I5b!p5( z0x`#1QC&!luT%s{#w(xTq(cQUGZfvtwJ-sgV|A#Wx(xe`9gyuyyfDoAbp}(_B6On9 z@%=Ry@qP$kXbc5J6+t0E!OZdJ9*;`NYJ+nFMUtHw@UBTT&EWM zPpkBlme6IPTt{9zR0WdfjyxKqGep6KKXN^nxI*uz>5N(b{%b0uo|4+#BLdbQ#7QUy zFMh*u+69am4SCS$fJol`-AndMKf3$a_syp_xxjY>t4r)w7z;yMdhipBeDden7kA&* zDnj0_#Q}c(9S31Lv>|`L1ndihu2_K_H%Pw30bWM6v|3?J)&?5g#->j> zIKX8D6Zwq2)@+si{m0eR=9E45C;Qbd1ivcz^{>H_d1HXWSrdf-WF_cO+&B7I#sM_| zqfAFZ0d=YvWns?%Hd(u>)h?L((;%l+tM2`gg@83q9)8e_tuKCE#dv_+Q;XVSGfue8 z`*WY)>%zOwgD~}Ok_=7v4dQYrMR6jS;lRaoy!+f_+@z*oL`imQC%P1skcD|4c@=1^^nl0CUBSS=X{BBJcx}^E);6R56{|*cfk?f9?I8h0}e*W^`7QNrLR4R6*0c0WO!Lil(Hi9)$!&*z|NjBqC@VJ_rQ#cn!Ok?kLn1AFUl z=^}}76~0p5R=qH@lJe`5|HNS(^ROB27sTwKHT@(S>p00WDA?8oNw;rXtNfBz{DFLs>P{nLscN9Jt4He=?Z60Jd z*4g>jYqP)ZlbAF)f!W6-H_}+!unDqRCBfW#3pQK&+++rZXrr|}Pb@b{7-U5vkx!?q z4fO{%xt&xtJV!?YNj=EOi6r%);zW5P;Jbu&S2o^9!OjuP{cPlK2qJxbhyK(s4XIOp z-36Z)dvqmxObj><;y3ze7BdcAHK^$=V0J`I-EeY=MTVfiEVwWS{7;I&8I{-68`DR< zP4tO@CV9yXNrvfhSdLBrC9v8S&?3r12HP7&e*-2Ly7zkI#2Z7dQ?8?FRb`+tsMJW! z>Vm3G0#$EHlt}>L{;0;gl0s}B=(IB!?|t>1#VEF`2yYU{Z9wfM>_N!1P016$vLg63 zoxTzFK%&3a{`ZyBZx~86^+YvJMlyp2zo_1w$`&W160txT!<8#bG_MWS014F*qDQTp z_E9%J)m4!$BtBqZhLEX4clF-}nm<~(Ri+zC2dJP7Zqt?$2QtLo z_C>ia*hZ6?!f&5N7YX;&OOW$4n5iym+aUvX|9-If-?sdB-+f~(Ni?M6P?`k8xY`9B zWN2>!E9}q$FD|HtGk` zz_nYg5ir>UwjcH^iQtZ98ablUR<1)mWnGI|6A2MzGazZxx)o1BjSE0Eo~E7FEBtR* z@Xu>#Xbk9Gut3oX-r{z_y68*)9Y^XQW3DQmnY9L^9pYdSbM&`4TrXu0W=!7D!Z zEq_SZ*A*m|mjQD#DSsZpD>FIiV0Kl;G+RytX88}=#@IBSO>NCkUir|>42)bjg!f0u zJ0V)0HRK}9)1?(PeqSnnr!DskOY+P~&tPQEZC5+29$B}vJ1p&KbZ7u>IHU=h zlMkB3@bjW6ug!!k%6-(;dY|)ue_O+1Y_2fLrOA0%dU>>gg%zk{I@pAHAd%Z!tC zl!KaNw5|Qdk!ytv2Ve>aBK>Xw&QS&qv1l1`Fv`X4H<7+Ce_hKGnMfB$CR@+Vp=cQ)k*k^ul zHsWv1uwId58gARN0M#lO?NJe?4+pRC)zwI~SRBs^9;m(hwERl_%X4ihrQ$VD`TGLt zm8Hys{*N!|j_~d}-ml=1)N35N52;6yC~UYxErR%Dr7$w6f;C01K~f&r&EsHRlUv#X z9J%XV>4>K@1@gDmM0B{U^xi3a9N>MvXW@w$AWu{0tgwb@msYrscws~4AXBX|r=vZ! zw4fNf{fpXq7voQpa5E9Rt*9N1Mv&6<>WMrWikyuR`*+Z8K~*z2yIzWDGL2 z;~3q%9n(WdtkZ62Zo^L`_akT}^O;x6Z0Vv9nU^*buOxXDB*!^(+rG)6&JwlzB2F9F zNcX;sSlXo%cVPr3_MZauL>_hhlJ-DT8p2wsb>AFXh(A!r0h+TQ4e8n(lQ}s!*b_f# ztR_gavmeY|)ZWO_Ku)+bR2dEAEIks;C4**YIm%3AfMv|s>KQj|Y+W}Rekn#wQHq-L5MsWq4mLcc zpkSgOcp2O{=`_}Hg(D_9rd<&k10_Mnha`WWw5`+ASQg%TzIaGaswhLG@1p&2`BsI| zNZXJbr6Mhncrh+hdd(s(^Z)aUw=Ev|`jq{pwj|2apbU*GUIX)x;!YF1G5&>&ZScWt zrgl@1lAy74m&oeuv$?0t7w(`mPD9B1^K85=g5zzkvmq&G#1Ga?9rsPv9T zKtRYKMtV_4M5HK67qHPwhAv0|3qh&Uf`9}80U?A)jS$M+Zve;4eCIjm-skk^<_{}L ze&zk`ckjK{UMuUJl@98r?eZW%q_QtK83R4(yT;8C(2ex)@W6+DCsJKkC$>n0Cx*n$ zZTD4tds~tLePy7fK6E$HR8xa#^vA)MV>9zarj*Nub9F?kwYD5MS(+(CbIPK-XWVq` z_naz*Yu7Fx0nrSC8qAcDnT4%WbJlh5RW#A*9fKGLY;|?byO9!hvnDva{?dh#qnQxS zltPs4Z#ct&e;R6ojT@HcCzB?yI{lW#w#Ik5ukv8e-WfacPp=XDV@pcF4yzEJ%t8Dc zC;`GFfvqv;u!Ug=ObF#Oz=;&MvP1>|K#SEh!9gce$20+SE={Ej&C6kC0U@=9tf9j% zTQ5}+vXisWjzy4wPy~$yz2}VAdU0?d7DcrpE0cmo454`#%SvJp88SoV%Lkx8gfoV9 zpv$ji0%51w;Zg4_q??*1c0jMzHeCQH0P!K@MzO$g$vDsElH)(DNy*p)iG!vv>cpEm zp*v__3=vG23U!Jw5(z0Y2Q3B`4}|FsDcGGz#3ar?!-dMpW{AC16SL6f8n&UImY-S=l>iK)(gduQ`a_+Xt?O0}L)|)*yaj2z3FP z&0{Tky1OE1024%HdqH}o0k?c+9C1X@bXI|nIZ)s+@0=gTY zKZjsO;I$b!5up^@4@C;0+1kKSZYR_FH(0z-NL8fM<`B0C4kC%v$@YyEi}v8rk$DW18nz-D4dUh_ z@$4iN0=cTk3<=#++!38o^gcE{Ww60*JanHe^`>XbPzmt|`(kMp8{7%nrbU zk!UGmVIU|r!yJbwSWwCfp%a3+>LVeYYibK5iO}t%YOn-OndmSYtMO<;0Vv50TEXj+N__O%siM^~U5i28 zMx))vmTm?+ydai=%|Y;5%+bS#RjElC>Fy|Neid9Y*S!_<(-t9m)*6I}!yVQjGoc-! ziIh!D=;&l~(_s5TDmZt?0rWsQDbG(2TlL_}ks*hsja-+8Xqy;lYQmGOap!oP`teM@ z#fo(2ZvAObHKZLEH@09j(hYd8A}qtZn}4K624SK5A%RSlSP+Q%cNoXW0AKPkBZD#Y zRNmso&y)*e4((o=uOuIZUWdpc(P~hB{r0M1=Stu7I+fYq)K3bLb4?hAhFxW^r@72L z7JVnX=3d>{fsG0LT%J>eFWRu4_aB(dXFRhe5*Bz5XLK(2Jr@?f!N}fS#vTqi(JnR} zScg4R|&xYQp2ml%Wwj+wxnF%6Z4g5Vm2Qjw}tT?6_Q+@V7+AAnHi+66m?}DMqw+X^e8Y^eH zFaM;;pFkD-2L^NQ=)w4lCl6nTkGsYO5B+hxrp~eB4K$t{e_jD1S?_@L^S{fW4IxVy ze*=)*fN%mj(f}HPn1%iQK)5Lp2Gg}=550Aug^8MX92{sXLQ@h#f*g)yLqIce{a)kM zhS$r#Cr{%dGCJk6O8G<+yvP^fMx#^;o^8lnjhIfRykV;{k3*ahij)WgL7m4``tWZO z5GAbZ^jjCzh;6ky61QEG3!B*Pa2#%bM!`S1^UDXs&w*Gl@`q##&~Oyu=|Z*^5rLF^ zv6{XQHlA92NZD*Ap)|L(2Y01l-mHJPP)8J36*QQzJ6=BIp0VpLL;+kDW}F2L!@LkR$}u^8Vb3tpcHnd7-b{~sMJHTg z)fw;f{(8YdlhSOgmOsZ(=OI{mw)gp z)|x*1vLlxd#yib;>8~C8qXuz;!xvK>Rz_gypBxS2nV#1{G~)3eRec@i;) z1~Tm%t{Rwl9E2etq(#^GxH-AMp$RQep@nX`j-penp_k}W7>AYNXLdYH$KKka@&O&! zAG?L28;OWnf(vxIQpW5*JGu-AnW=LO)ap0hfx*P&_RiQ^_!h=Y&R;S9BVTBIT&8<^ z`Dn=bW8Y&O9~SicGJAp_9@QWJ`C(XDoIsA!yvc-;TQVUXi&jlVtW9ONhr_Jua6vyV zI3wfyVph}G>YsR7etWl^Rst@@q|bvBj0WBpFS**ZZaCb4`?$&YM!CT^ex)(9qN*2z-_por@W;6#*P-LuIXago1yAhN z7rx-OBL}a(KXs={==GapsYlM)3~(qva$V!|k|X}g=g+V9OFwk|@Ee!-uuQGXmG?WY zU#)k4_Uy_JN!yFJ7Drym5)%_X5gR{pHeOXSZbbHA*Flrl9hxeMgF}6uoYZkMN1_)& zL3HLzPt*VI_kXeOtp5Ek=AJgsy_SxB`PW<0Jxa~aG@Wr{n}pYts+otF8(OU+EX)J$kWqPQ1D?&%APfA z)(GAZ3J<1!gCR8-EAnZEs|jgNG8nqWW8pch!;~~w=aX$BMxA1-vmPF|Ud>f5#hl*EHTxmP#l&+iJJRFVI z&fmJD7^qxHN z)3w1<=#Bnq&nL7o!x2mO3rl``uc72yjDLxIMAb39;!Eg9tZm1Xw)lJQ`ggct|DD#2 zwht_X2PgK%MKuiWjH+oZwJe~1Q=!I9^;z!_A*87mj&APceWQIPMs^KeH#t0J@__%@R}ELL zPP(x!qqIB!#yh(39B>u=`e0Lhe7wy>NlD2=s6z&((k(bKyQ_`qjwkZ|wsxu#$f0|u zw`)xpWt(%aU%#G1N^Z92P!8@TR*M=oUNteZqr`c69omYS^Avyj6r&g+lmi9zXV988 zpXFy+(i9gT=zH#eL-G#k{^)Cc={qWGH)(E$`QW$3M$P={^~kjG~jDn!u>x8p9%|Is9sWTTYWus-hT z&anW2Bh3K@fu<&tcNVqOz zn%}an@XqDS97W{>$H=MA*6y^`t5@Gv+pO2~y;ZnS(u9Q$t<^kbEO<&i!}X&|$SSl{ zl2sahTZQRSwkC;e{lNuBCJ`a`*aYG+pL2Gu|3ifI@Wfsmsj>+k1k~S;1dE<7l|Mfk8W`BOAtBD5PhZ|WM9_Q>T!`xw6W87wd=l#7f+F*t z&^69LkIw6@pRc<6?{nX-Uq5rS8---^54%6y8?~o#O-$$U3oM6kbEURs&bq9a2&vsy z2eUgmGK!CH8Eb#a7_#74t=AJD0EdIouYtpfF6PMverdGH%46cMFLn@5FVbF(8L1g} zRgK4gkG<5?o>*J*yOf!4{j~xy)oTQlpunpgZL7Ntu&ydp7~lHIV0}oPNfZX7{fpkr zfp+b59}xl3uYPdpkE~z+2c-djeY|p$Gg(OQZo_b)lY6&T5A;vWyz0NeZ6I6nU}}0P zqLa3TWG4*ku+oI#hubm!LqBPTf27kTLUPo3tXiAhH!WR@R0-CnOV(|J2XSO%UtC*j z>o&mhmOqTasDe)48S?WaZX*MeBu`hVnVwe=-32~ZtYv+y{m+_b03XP?&|iJjS5?{D zTdv9d508Z&69bm4KL*pn!OESYfBt%fwW2H_TzHZk?g>;rhgF#94WmAuMlTOZ4@;A* zmyWFBx8==!>96S5=8*rCwCk_e`#&f9i-l*vT14xwsowd?{VK8TZb>C>c4idNO7w7)PU0oFUdzTc9Tv*Rt2 z3Dgfft0g(s&?Ptuy0Pg0UkYIDs#i|nnG4Byfv{?nNpoGw2ScAKE!}(nTDh|p=#P?k zqW#nn^nkCt@UP<-wt;13WyVi0Ik7l508t=|RisaV^v96BQl2;@=l;U!%X{ncdrw6D z&e;7Lcn!V0-JjJ(I+3d?A}AS%Jq0xp<2+j(Z_w!vMbwZdwa7c`%8%lKdG(5-5wk1b z&idW9Bp~z0`Ust{dzi=QUanOAkPy3LOAT-ZY40eD0?uy|4 zIxoEwa=2GWaH$vA4Pg`trK!|hoF(~!Bp*wjKE;ApJECvjU0|xtH>*~{;2S40k$Cjr z7qsjA`LfPL=hQvl)m;gkAoB@NBrk;Sdj}K_DxaTi3@xcgHpeiPH^Cd7wF<|%bl zACS-9=a>j8;Ap7wWE&uZI!Zt?OPWTP4ko=$Lv?Xi43g14tN_z^$AwovgUX!U-2o7v z>mPft(PuBx19K(*c`ooLHCedt5w!NaJqrzsGpaT)4?-_J+2#SzV;{O{Y07|XJOK=h z@x$nPJoNCZ0BkmM;QZkl^6DfR!_xZd_1p`OI7Go9MjFw92DfDzlw4czs40yiV*#>4 zF@`}|6$T#0>mv&m`>hssL7C-{$;N~7|3?q8%Bch#VItB?YA)K7g@*5Q)T9zY-=yUI|60v zL5s74L(n!+Lw#Zc5VeXkePB%)5?KQnE1!Uo*h0zBL*gt#nx`BR0aQWrt6?Se2*e_* zfNa|%SdPO!;^6n1=>x+iM*rmfhv(<#$3k%4kIp02?V^S$gm~mm16Sr`vo9dapH3D* zh4|;|5dL?@@Ef3s&AY>uK1L&x$?o9!Q4VheR3LX6m{X^keL;oRoJ`5VYM`4h=d>Jy zmOjNl5OywW96D|jCK{##&ibe;addPv_mYN&h7J>nx56?})790rr@%DlLnt|tX~Aa! znaP1?GM3EEuRcPh-h}7frR|%mL6v7@<>ZdIHXpOP77`M2a^LohVY|vRQn7JyJ5@6d z2nxa#bovqgrn4a4PAG)r5W#2hl;}FX+m(31T2=oK*$zvkVvGUPHviK z%L}|#=X(!sk;H~D53ffSDLzwD-A2JiM87bpDII+Q-Ufyh(btXKEEO`yuX3^L>`yn$ zRRr>EHZB>5%beMR}2U-h+(_od1zH;oy{ zl%-G2kxOdd%!f_wcaUprOKXxc;3+TsFwyw#-PVv|q1d?Chr+QUmMIUWlJV;U#Pz5% z_fwtgB2$eCP16p=RqGe|dF^9zf~R(m zN2{khex58F?R*ioGeqhX4|d2QEX}gY^F@@sgM)}q*m{*jZKq757t!A5h!NS(+_UwP z62+q3?J1&;^X<1gT4uSkIrZ>Lq?Y@BQjL>RyJJ;IYRy&A=Vp!P-?q7?I4l;5g$y3q zN%K>xl8c`;?r5{mFW(tmOO=h#DN-TTG|hIMYL{scx&f&yIN5M&`PhJnDSW_r6W^2{ z>8&v!ol(UoV-Ry;8M;{-P|HAg#Re^}SVnh8hv{|-Ql50AkdBruOcAPJ?n*k_X{4h- z6hO4wTFRR^8aq3Aku*UWDN03jVj!OcR5!C}ZZ4-TkbP6vZT5v8`KOBZ9C+kml8pR2 zS!s)bYqo}I^3$hJW$6{(Qjz?gzp++Dq)*gt&CNPGQ?5)tU8_+VjU{3rCVbzKc_s3#@bgmn%#)SLi~tDfjE`+RCXL1lm$G?vbwOC^`x7{>Myt6KZq z?3DTRlDxd+I-!h(eC<3gD0%vb4K^^4%O&RX{KdX^-;H^jiO1T~1^8wbg8p16NBi28 zL-Me`W@^nN^H^WU%^G|&vBYFiJ}bGt4ATajSk1aFU5%QGzyc%J9R;zksQN?E0b~{L zn4MM=!~5NPu0%I0ZMr!iu&Hj|X&a&Wjc4oncO(ahQ|I#-fqZ~4bnNlQY!6t5y%i=dXDPX(Jh-jEoGY)svJtDa`Fkyq8%Ni zId8AW^9^dvSz%+j`MyQ6b1{QAyUcp?66ep$%F3#}Ec3HU4wGt!!SaaqvZ)cp2{0AT z+gFFSimEu60!=6L@lzb_?bAFKv$7=S$4i^!S6M%)Dk|Tk8GhD6z01R!S;KhqV#B@o z)5s65;-raM++}q+v#&H%Hbe77E^7j)M`&c<(xCyQEs^w@{_y3%K*+v9ggWFo-D;-uUH5hx zjn+se8@869D)V~cFjR4oz3&ir?Zd>(jeQ>K=>z6r`L!^dbZlaPGq$&9kG$Q5c!-=P zlAbU-hM&?J9GmQaK;6AsbJE#M)>dq2p=gn=&Wv_jxur$T75K4h`kbuWWMzF_`Hg|%AF(bE za(WjWhpg_%iHn#mytczu4P1ZdG5UP}IbCu7w@(-slln4jy$z@{SvD~bmA%^PPE3xwoT>pT{5>NRIJ7n*Jmk&I7)iQm^Q4PvUgXl zqj`$N4u}Vvw@knLCSA^;iitlc&Qs;;@2Q-u)d{@-g(ilvo}QivmUu%c`Svi63X#B^ ze$#i%Mj72)}l;KTBARk)v%M{-c+ke;46#b9VySy^rM zWI}_OKp+g`({$tpOPZvwP6TO+|3LJKu}Fcdx}LU)^Z7SFNd&v-lOQApSwk zhI0O?3-L|T9XX*>(bM>Nbt<-uaP+V;?10MJY68co{6^H0Eolj%_DYPW4x~+2w+M0<;?4{BOGRZ?|dFzvHX1j2t^IN9m?V``=T z4!UlXOQZh&KksVNShX?9a|>R3XRDmIrxk%M6<6!1}ur)uVNB9kFdu{m(psV(Wkvk43 zh<{0zFpy<7LNN}cAn!C~FwSrvJlJj0EDzYIDVdUQuct%ZXypr!a3h6c*(|>jX;{B+ zqGa$QEY*YwPSe7~D@CR-wj%1N3b4Ig)*?{SA_~I5RV9|pmf*eL1sU>th4LAlxe+^6 z(WOHzjR*1Ql1v|fcS#_VU_3Po>_bhJHtOu2HAN1NDW%__yi$MVe z^euHqv1KW|^S2!Ve<;{BC?~A1Gs>NB*w% zERr;MWV|o31BqhH^$0BVJSk!ccOcgfzwteOo8dSi<$(>D2ddLJ)icZuTr?Rv_M!fJu7>I{mN> zZYOxQX7v8PdaKr1c^l`0&rg}cYH{HtO8 z^DFZQCj#s1>xlzyc>IxL$BwPpwz5Q;Z!fo>WSRo%x{`)ax71|i7FA5P+`=#YgZ|tZUw0rXWdFB8PBPGM{7XYpYw81?U7ewTc7^^L8n+~uG_Qk+xW=(2`eg-GT`XLwR$}Pd1 zSpQ!G&HmT*2r>8hcWG|E_UnW5mP_mTBl}@ck9I zB0D=9EGS;5A_kVf&(Uz>B4>+C^I~Zp9v;ExU*5Ej)onYZ+zCxu=+f1fpilEp?_y4h zH^?llY0R+y|Ke8c{~ny}KZKU>U$gfA$KDhZ*tM6R8rVI7$?-gJ36aJ$0RRd_4Gq-p zT7H;M8%WNhXBNE<#qZtvLt}e;aoZdVXVhP6rfHk8fbE+RqDJb64<82UkEM-@bnV%s z;0lQWZwnE#Y0_nFLtW+u)~8u}BIxb*&3oM%72n#wTH(W3<8Y=vCzlFV__@BHzI;OJ zSDRJE{*{AbeHI+se_PVbzq&fV+>Z*iC1Q=5+7}vxTDS`fXn?bWr2`u*O;IpO!cTnHeh^RdODH1gvZ4|2lt}tFm+{ zqgSoG_asl;#GlwkE#_Z6lm8qR(!axDkngfmqg&U6GyyulTURqU!ENwoYgk)>^&PcG zfq0~N@L?B?m}>B5hqd~46H>&30ggRDPI&m}5s2UQ2JRny()Tbc%c-_=6ngC_!$!i3sK11C zqL3Ud%CKSJOO|-`%WXh^i0%P|;ELb3@e?N^Ot|tP8P+jh$zmyoFNp1{%qmVa$EZa$@6w z54YT>GOYN(>BJhv0&v;u1ur)I2!;4Gnbr-H+N=Q)91dr1DIRqbu{R^ZE3K=m3(=5} zI=I&2eB#1TVlUuNHxb+iybs9%?4L^D*;_XHEq7n~@-x&Y_bKsY~ z+ix=8RRNv97c{lR3Zf*e_*Lh6Ha-IW*7;&!9ttquR`uRGyx|5a&1DynmE%UAkB4*I z$9hdS{Q0cHcl!sPdsX5elWe??v=+`K7efKXJvoQpa8Htx8)c%N6eZqw_kf~nF?fMg|2 zBMD*9(&`%YM6O#VOsj={E5wg%?O?kew%@=9T>`u4hyvC5A>EX+Fc4QP&a`ma{~K}& z3Q#D0&;mMZZgb4?MX&S%wu7ks_PN)ck3bu~f#d{}reT1MC*|G+t#szwzkzoPNQ;tO z2IEG}B5aq(nzV@$r=~jdTDQ6b`I&sUP^`~z0WD2v%0Rc|oTFog=RAx@MHPUA?SlR= z;*5WJMQZ}=+K8A%xzm7vk^%#dh70+EX(hTEeISyE1VGiGgx+Uax}P#0S@aS_HY$+Zh*A! zH%6u zK}NFEmurIlPj?;XIMzg+bIpY{iX`!(e3EH=u zs!?|S%y@gWCfFv^=%#6y{o|4VT{^W)d2YY70M%fBs8*Fe4oidpz`>!cteoWHpr!`> z`XR6LnSE_RSRzk0z*wRdSI5j#a(olG%K=le_p{P-P7j?*9(E5-`(^RWpUroGDsyIY z2!mH!7o5y^8g8n0EV_p}DzX5jbAB3<5etSw$Cluv1cOkwgR@+B#A@glAvm1cyH%^) zc*i76mqKtw{zu)QW#0!5+Tu()nr+FnBr2AHpI_A;m9ZO5bHGWEVTqhV2+nKs4Yw?8nkWBns>e}a5RE(TkN}niu=IyK+^S-a%EdD#48LYm8r^> zDt45BJbQ3L@;vy2^jyxvf`kZ~-~2wyV>b)R(T4rio^pBS%s%DG_uAQdImY{b{AG{k zX5cmAoBc?hzTjIt(=SVM)ybnTzFJnK`}Sd0Rj%c{CC*7wYA;{)1^;sZ zQ-}QI+{a7Pvh|@eP8>RUjIfz3%^mm;j?=N(tYt>J;A=HP*_=2(Vz*Op1z3%$XTAYs z^-oAsi|Q)bRGhX90No~d;CGb%_Diijf8ddcje!FTg>0+ulW%Ni7Z1Wai&JUE-|gh} z^cD4~x9hMwmmbPUsd;ivFf=VT=O|BszE>FUkXI%ka1l9{Vh+k*0YRxzIhKGIz!D4$ zMC$b{;8&}!#awNlv1H1mS0*@JR`%T77(}6z>61JQ%u{-%l6~SyvmUl{O{GT3^FBJo zX^P~y^ggNNHS;$C0s0CrAFe7(*l0MRB%Z~>$}^MA5O6?zm<3~?Sw|z`uYET5${9Q8 zmjGh#1oDG)r`us$uY}V9yRTf2m9eW2gk4evB5{?~e5+MN9TWB^X8N!AV)N|6U#p&M zz?=|}r@iG1!wL97YbLrl;qBNNrLkIQo57AO9ftFJwtw4FnZX*N+1Q2x@ z-*K8t!sh%2#Cv#<8c_i(AZ{4frdrn2B`815B!ZS|tRr7p{U;~3tw>Ule)pc5larGT zVJbKfISHj(GLZff$OL4|(qWqPCiGEr`-kO98Mni{&k{@-{b=h1%?@G~)!(_D96|9@ifu|8NQ9GrQp|rh|(;@Eh4hJ@H$u=>fjhgQy1DtQ) z)hCh_cWo^AmuCL*(3RjCc)N;QYE$D=TI0FVk2mAv{I=A&a*}#~#a7O@HL>8Y!uUOj zA$)NHK>&sJ-B&Jlx9+1fM}!h$$6AdW=;xZRLYPfl?a8B_NsR4FH84%Qdb?gH3Cm;3 z{A44qG8uR7ykl5sS-@u+J>iD;*v#Kn{gl3JodwrFU-f&lN&s|AvN%@2cyha}DmFIu z6o8oz$=Gz;;CvSNDM{w-Mks6XtICnLdb7&p_jx%936!Pf?Lg(D;(4fpF=NLjEMuk_ za{bA}#<+hGx>qr-n|P9+z!{Rim=o*lvyVodtUf>pCZA#6T{OGBZ;1`!X?TD_`-s&p zd?2A8ciwYaC>CUWOx$r`KI;OK+lIsiu-K`3z5Ois4Gny)C zNse{3u#b7>Q7OcEjZ#*dtp~t`eaRBFylXOX>|Xb?p5eW5@CzGKee5JSYi!DRqD8emt%QA( zUq$FE?p|+E)Wmh>4^mIf%R`%q*M40$IgRp*Nls5j@uCB>?QF=+>LNYg9TCYIie(NY zBdl`yBb|+1^-J`i2=d^4Ry2fdSozwolT+IsslGm9mBSTMd8U%8@0gf||AK>W`=QZj zpSrk%m-Dk%0e)`D^L%m!I+bG!7r8lgUP>9rY}b4gxy0@;;32?XtZ=eQzmChE1)HAE zRG3Dv?1kGm&*T`-@_Ws+8{-tw-m6zI6*{FDGBu&C*i^C{?3BoHK@jXV{4LnMK*5e# zh>3pJ{eGQwNalnkG-f7!rzhdXe)PKj$eo93uEmWz=*adlLJ3 z5HyZJ`D0F-G-UN+j<}Bev}B)k{}SYh&^FDx#p@E?&CPjj(tqNfv_4SK z5Oov))qlAxwsO$_6XmfYaOP?{prDh>QuUqz^DsYxO{mK3$J}WcXXix;YAX{crnJ;f zwwlhm8>Qo5GO``QaEmXpqw@B$+=Y0uCd~U+H*`4000$}Z+P60=&md4W!&3msSC}wC zFFnI^rPZ{Q$yy3R1`Fb0TX9ew9isuq?Q>(8um035%06m3{3bR0WP2~z-E&blVFLLi z!-2`G;2exp`jaj}!e2^r!ex}O%YUGcd<_|~1sRzJWth@I_bSY4VLU475@dO$eSCB- zmv+=ggvTJw2RV3wBwT5O4{|WDp)pDal>wYNW#F@PfXly@YQy%C1t}2nqIW>1EfhO{ zB+Tx0ZGke8z<3GrT`IlfNY)s5szoh`Mgxle6lBSUyH;j)z`qCX` zeSXC&xS}R-JgA4z_~z4i8u|u)znLgMZm1+iENd1&T~`eTFz_V+)26X)6be5od)2!d zP(crLn51b`|KmJr%+7#JgG~)=FP~vuY-dd*p;88U+B+(;Bj%P((4vxQH0z!vBYTbB zM*{Qb>?ua&v#l(s5&T(#jEKVqIj|RgMlMC*B_m?y2B_~?!{S^6(i{sZ2}4>#$T#To z9o3mTFxCK$jD1#(;by3eCMKVComxL*%X*N{AVClig;U($ON{g6v5u!a%tANvQ zj1z^J#LqAH^dgDBKtyzY0v;Bk*)l)~K(+~Iz?jH@)(7I*p-{#iK%}A{J_0_3inw>k zsw;Z}ngbF@cIn*t^O<8XLH^Z?3t^Tr?ZD3Mc><%VERHgKJfZgj;?p4$9VEt>e%0$L z1yI|U@Ohj_y%MT9BlA&oO~9K>iY4j*MXNnq&jxcBC{MwBpzk-9h4T>wi`Ff0>LIA9 zQRv*ogLKZ2+X$!{K0!9aRRv-HMp$=W(oVKCBZk;BrJxi9T!w-zHg4^y&| zKQB!UjLMZ~WQaoPEu9LwThZUE&zd}jfB0wbcmVxxaUB7-Pw#4n;qz=GpCuld+p%=R z4iP8ZVB)gb2>5gG1%IRR*L1h9!f5M4%TxOaX~6pnCIM(e*Es8)cm6-of-FaZmrhs1%w zrO_#UU>#@|0y+U)I-svX=?x9pzMz$h?XNW~U&<&T_=q9Jz`#J{-X^++At<=O^joAO z>&u9a!>iXpE{G^KC?_?)*mz`%kt9a@59nsV&f`@_UmCnwCjt+{Zj11jXs)WiI^y=6 zuP2OG-Tra$`j?UN6F6W#+z$txQ#lxB3=DLwKakRU0flsP@E*He)35b_D5R;i2=kCa znwt}u?*@@?Ozho`Rn~$|Z{JzW0zIl$8;!5}&bMTv0&w}Ieln_u*u5}aGzNnvQ5`rx zW~Zn)GlJl~OYq^=NhYG6qNe(KUip_$6>j2?Al@KA7q@$X)_vT(5hHnnlHbvJ_N$iY z{e)HOb%zq%pkHBg)ndxAJLi*RXhLs?#Qm?G%~l9B)mHNb`**u5}HUl3my zCM={+ZZ8AxRpFq@-n0BaV6+3-pL#gxN2xWRD8TR0YDVhHmHoa86NyzonLFW0rR;$g zDlu}VB}HW$-Q7c;xD`)vr8J9o}Lb>=>lmejORZGZ-44k6{5-F*?JpJo7?v(Af_=Z~BD4i7`%#;|X%Zz0K{a9~xsPRu z$S2j1jgf|Vlu`2u5M2o!P#yMqX%j9a{3AWfIOk#SwB=Zk(cNS#XftjV`9gI;$LU`U z-=hGw6-|?H!s_O>F!R&c)n(yO4nrE5PlyjzRY324j1wa8S1;BD0)5h*y-7vXE|5(_ zSAkYg3y+a`YywDA0rTr|?=(>5PsZ}|tYiPi8z*2nfJL_o4=*o_h#$!@4n`Tc&7uz? zyTgQw8h{-zi3Ba~>eSAspfr4}8MKXG{ob+yd@dcyRhV6~7D5v>i{n-c$L6a1<}}pQ zyhmLVZwhI$T$2EosRWnnt%$iC7ziIp_h|4Eadrgl4;vhX*E@qfchJ6D;ynSwQzvR2PCCNFHxst#=C55$4z(PCv8xQ0*lE75D7zbKgc-8z!f>a#TVF~Di47D z^wsMSv@^-yjntqkAu#_@qYvPbU(R>r3xgm6n5FFpW5h=V1^5;`9bhE)wx<*WhmG;< z^t>CV7(LLPGAbL70A_&6geQnKGyA}Fb@5vi!_Z6H26lOANd$^CHaBlsns?SDUt2mI zNhj?aE4%2S*fl{uW3|a^v0e07ESSC^H0y112M4^zBx>o4D`daQoJ`u`KN<7W#qQZD z64X;m-xqDWLe22!rajJ6jY0G4=%sxa1T0RVVI7twC8|AuvM7s~)ri!prZ)FcL$&&e z;?hrBi#VLGw_8#Vg(Tt4-G9(&6ym{!1^S{2M304^3vu0+)7=mdfp6dI30e2z4o-`U zC)LzyVe$ac@q=pEueK>fQERJFDB#??sFs}2M7;s)@JP6j0fC9u4MGwTGI9n@*9k0+ z32ZSIUOJZLUrQfnzwp|q^yyIPOdT#1q)MNRtoPCk*WcC~2ZKtivAXV=dbiz~hHP)+ z(l6>E!h|Lw>~cfRC{lM2EnOt+4)JCmvgbSp*VDUS)Jpg}hJzFct)j`uV>Hm%Fr92%z|R{ls$Mm1|kneEQJjAExhogsUfs&?wP>c zFCud6>u(oaMn`C70+I~nuAM5g{S=0I0Ldu)shFzi46i zg)Mvx>xBnU08a$ysA_T6YEcx~2BKQl0QS;O%5mJ!;3YM+mte`rDmalr6`B?-96<%-XBfEIf*jo-uUZPS5Mo(;T-YlOw_j&F&Wd&V zZ9PO#_I+7C-&tM>UXmQZFA%;qR(QUBGi(xKVo^exbM(de@?*^$CrziJl@jE~Eb_aF z+}G}4**?p90N70aJ~J;==U$-Fx)*vYqCXC-U!cD)|K)14X823AS$_uLof~cUA2``h zG1!0=QIXX#vcrP}%LpZ%Io8jv91TDF(vJYo49iv^aqh#Ov8w`eKY_K{E&L4IG1N-3 zLR}ki6h8*n0jNSO>C93OS7>1E^z`!TLq7B9I}peF+_~t@F=}idkenvGjl%1IIs`C2 z`&G(@e#)QdeuR(h$NyjShU9E=nj3sZUA^jhdM8g)@wuGjwVGkk*8RbL45=0nlt5^kAH13$uhpMYNmz?Gsy^+zCXY zav7MGR{?Hv1~9}iNXn`Zzy7SncD8dE(Sw{se;h&ikV7X)_y{jlEq)e2nOOhOLf8sbr+1+J{>L|0|zENs;?s#W=Q z5b>Zv|M%U7m%@8jNK^Qwf^023_J+II7G*{1uwtnK*PAUZ?44JUw9ZcypsVD9>JtXT z_2{2a%Wh?Bm2as-${62b4k@z)s4%CN$b3tJSL&_HhyOUEiLPb_4UY7GVEjFonfw)# zNS+E5PvY?_Yj?|Yy4orX*Z+eK1(2@%a}dmhd?j-r4?$QHl&uKug3gyal(GV_%wCnS z#|7IzKsc-u_9t`j-Ef(oYO6x|`fx6Cq{uWa5UB^rVA%Fn?@s%b?M}FBLX$`(LW&iijkEMp4`uNX(eB@__)(TJI{&-j*2yqRem4iH3^L+GJ#fSTvaMo!P%3Iq7WEKeScU{L!!x;? zC}2-1fVRgzun6^%l?|S2ssK44AW!*jN}Q>+Lup;on{LjsPuJX`2;~JRAg=bsSH}s` zUu$>{tTI{)ZQL!8Ekiy4IccTU+!53{MPn+D&ip1J2r6bj9|sRqWVzRivG8+UaS!zh{s7q0~R2Y07(a!p!36*3kohbRCO}6m(OTH)ys|F*2rMs z=HgR51MZ_gZTa^3EzIU`AOF1WoAs%0AMS0E{b|za$ags|72A2Qt9P#IZTWbukmz*v zX!9?`Lq)5vtp4HQZ!x}CeowphyVNhX(p$cfqjw7o?Z^mEe;lBPKfWumVWMSXq`;9@ z(2bu~^dcx8z%54-<@A~lM8IoyX$C!f1?6>A{G|^23MZIcJbnPyrUxqvY^U?!oqE-o4dTJbKNi2ku$)Ra+ND{YTMM1%Tadl1dy)meegGW3T(loV z?490>u^(fAELqV%0XfmffJC;X_gJOSYg9xc07E$Fp%v|}O$_9ld8a528Ce$qur!h* z-v;WW7y@|~0Jj(bE?pg%%;!Q^-40Z%Ye2xloVsM~xUv%Eh4IDXmVn0~xf1w5p&3}u zac)E^vf444eTUjf2v_+K!0UYU`G%~|(h5DH@&;{d2H z+yj2z&Q7Ny3(9Ab!T_|wU#9@iZLX_(-U5BW2+Y470Ar`PizZPdf~ z&V&ykZtCV4+tLDr8-%XiS0>l!BLyghbG z^RsA;Gzv%)i0)vFrQZ}Klu9Ers6?u6wB!TbDdhTzw$ZN54CD`u1moKQBNJspZ~X$8 zV-Ma*oIBgm(Ge~L#X`e)aRVrL%nutecd&1>U9I@#+vGuWw2A{lg)%A$!u2=LkXz-R z5%$%bhXn|gY1tjyHJ}L{~F!s@K0UiyLtT$i32O}?ve>x6l#voQtgZF1NSZJO(L_;hy1Xqgo zqyM=Ynq@iPv(fLPAq52arNP-;MZgdE5|>0(!NSDg23tLD?W$Mw+6$3OgB~8uwLS6# z9_Mwb@lbDFbj(NyXSLJ3XKJkK!5%pe}%ZVh`meIGY0?;!y(HB+doRnM%Il}C0}gMt_e7URGm{Q!3H^=S7) zIlv6+yC4yHUMD&<3226+xaWQFP(6)+E0e>}bvkudM~E_8*}@tIUp8(jcTvYZBPxOcx8nK^ zpUK7pAManyg=*2Z`mQiH*qy^zbu7%J9)#4Q!4=IN4@cR=pweMzWTX~}E=_fe!g_AW zQUdz|-zu1CeFD4_qiSHC54-Uydw{SB3J|`0d81lafD~m9h^Us*Eg=A;?a!ar*6WKk zI}I_uqqCDKTMxfBx3t`qdK4Efq@$zbqjJf$aZ1i%Uv)5dm+bwvGzkJ8hfjA|c;1t?HdG=opPJHmO+1j$ZpcAXEkTkMSFPQsInklcwx#V^?UT5Xfj2#v ztf;8i^ybYU&!{blz7?j*T~+sE0~~@4sN8LxIR%o@2hJ4?5_~);>uonQn#Et-T^TK9 z&!eme3BOKrbR(}fW!lbreN|~{QqnHXrg!h=R2yic^BOJ;L zG<%a*R@R`9(&qlSo}tlEdTu{6Kd0eb`nF?!2`*Ghm1nxltzt{R$s*l(pXuqhI9Fo2 zMG9qB|M5|v1oW#EPUqz;)K8_m#Lw~&iYc%3>g;R^=xrAjP7QkQ#H)~AS$j*h1WN_P zN|8>;kWvH6yi-yfyFv<1(*0fp>+>?b{~&ZbWsnAsD98Ir&Uf4$#kQ6m=bhz$c62uT zsF;%jHdr-{Y!`9Wsfr8xq;))@?DZZxP#sEgiIv;US|r0iKqI0W68I=Is!rU!86~hY zPs^E4QC9B(y_pEWDISkM1)(Yl6!=*vM-{eBhhpZojabH|;A{>BA;mFnHIkSDhoA#j z9k74`nUq|#H|V`AWbfou2YEwx&mXxV?2HQ7e#=G&np|(V|m82+&8fK#AC3C5&Q2LgMBUIo3UTz=Lh|Bu!m9FSgEhgDGvs(JI01YRutD!##_6U6S@s zGkL5YyF72v=521IP86#4Ii?OC(HuwUnPyXadrWJFkzJ9+rP{*2JEZUR@@P+_y=Y^@ zQ-xksE2{yc)jmE7`NYEb?72cyz+R@yuOIt)2BMV^Jk;#R> zFw)A~(gv;>PLzi6>=qSqI*A|MIqlwd~A-1&wPv-RDq}`~j#>7oDb;Vz`U{DES?@7mOC?&oyjr)8Xrr@fiOA;_PL*Rmj8C<3rju z&F~!B9ci2DyWpl{yZ12vN%T|{$`0HD z&Qzs|x$t0nE4lOZtCVaV<#zX@2Zcy{)9(ovk&m0b3|N;3mFRSF?Bw7pC~O1JnT@Pdu_|{EFz>Uw8&C7X!bFT zab1`5(LNkM2xW??ga;0sZRQ;wDWLoiY&6!is>4J9@Li4MeMVnMsv zK4PoM)2^nD$lN@cEKsF4>s;BLCEDyQV0UioRfJ;pG4vAvsevz{UaBvc>LPFjjh4tLy8xq94HN-d^x={-uFTbTee~%bf~&IkRvyj&r?TqWm`FHGtJmlcYT4{U?cPQ4~WqI<|Bqex1AYq#9Pry&+8WxhpOJc*oB7Z;wr zUD}_qo*oh*`ShTcj$~QV=Q|EQh9)7GV3XGqt7=P)jrZW%dgONQzGg_>p(K!mPK{pH zVV+y@Zj@rP;-bf;phSrmRfS=udBuj3sTA#)@Yi?tFf!QNybh(~mbte0QiqnI*HEM1 zL-FyzM6FC7j)1{0qE$XltI2RK^|w~c*?jOY)U zC!0SIf%-Z9q(Uxx;`YqWD=i=aJh&?GTKmv83Rh zf24_bfzNMOH5A=GVN=Pvi3xyc>_Oy)a440Pf!yziGU>@7B~&99jU)JHVJ%w;aY(&Y z(Vqgx_-MTOt#R{?7qJeYeGP+rtxnwl`ZPMuLnE@x%majrd=wk5eD`C?uJ1!3EH`NS zv0aFZMo@`B5m*hCV%~2(&n=B&J9l_P*;)f-tA((Ftn3HWz9URQqxb_Tx#?FMS>O$j zqNGZEN(K_ysfTHqwVT*mjQF_HqO_R>MtLIe5zDNqP$dm6)yIo`P|*XIq@5}S6hi?0 z@0}7ZDl8m;_J|JjQ3zGY6RsFSib#LsBMm5p>xYao*$#N_#s}i*3sAh+Mp;~X4IdDW zD7Gd`n-O8$ zip6xarDIeLPN&-7Ce}niT2Ha)+atyg z0s#~4dZs@E`;$-_j$+>ol#Eee&}9St%XXO4lX2;N30Lqe=oO*T)6@I`4mP?OfcPSy z{C9pd*ba-r6(%41VNOsB$SmO>;Gb(Oy=V~#R%5kBvdv>q6$I)-qnAdcs&?Fw@EJPR zIwbk^C91GZcfP|w6HoW>WE0JB*mXakankb6Xjspnmws< zbd)M90p!e^#Y|Qp3d6-Uxdhd}N|7IchyW1HXdYLdYSFr4X%blsoz@F&7H(7SwFd!U zM>p60+_{ev9$y~?#<&|Q?Kg3v*-r#*Om=G`bz9q()P?IEZ{FP6=>2{*B!V_31K>04 zTkv8eFU)5I%SzB;nM|iGfH9FPNTv=#Pwf@-K;95u5(`j*1rx7+udb_1KG053DsJ;c zIXoJ7><<#nP93WLo$UbPf=bLiaFD%s8D2jK@n!8hGK?adafkda>((X*a^(m4OQ0thdp%j?<4!Ve4BI-g*J(N}wu2_SgPivWNN zp-_RnhW|N>I8%P4S$~P%rpI0#h7@x_dKUD(w-RJAv+g^lLoE;g(SUqs8hH)qvaJC? z4rO^~w6NO-N?A5KeF`1ZI}>M{J`K&H#*>zVmy`Py5vkV+f){qq=P!U{il_u>AgE^# z*Ot0T0`Hi#2v1IyATr+iR*Y|4jmP6)neA zpr>WRQLET}df0u9M&hL=#B)iNmy9OJ&OHsnJ+3XQ1oNx3EQ3!KcRCY9WbYJZkR3SUn-{HEv*FTh>mulMLkX%p`%eaP zqnAj~k6BS3J|vecgwGh5Ad<1V)z6@|Yt87fGDHi4=82Z+@x1d ztRC$R$DvvyLd}G@_uzGTc>9CL^9xap9BV$!fD~(PcwcMDN=8!z4h!O-+-RO>+)N%> z0#X@YddF6A=YoCsX5^vwqxar@PensNW>$wU9)cCbxUmNJtpy#uLikQ6z@tqlAALTx z=8g|Xz`yjo7hcXB1}U};9;7atawhr!irbH$(C|kaBd21jhk2i__IaZv}SW{e;l{eHDZ7;~~+B z(i8VDpLF?#=)^Y+Ve>x)*pXkcg^AFA2+b28qYQ#`H7Lt@OV+EU>uf zJW1m`X?mt#AMn(uZ-g0_ZQwpSPuY%pn)w)8Cy`v%{n%=>yP;tf}vS_m$Gy~6>Fh84Q7`3)QC@rwBR_&i1iApA+p}u}c z#}MPFdb>KeIL6(%b0?tz-V_}}xSq^+gf&r(5RRBFXe7k72I&NS}3JO-9K6QQYjjRnRM0(e>gms9cIQ;4c?JY|hi z2yt(W{?vz~orJb*Y*4E;62}_BJ&+HBGeuJu&3lIrf)s8Co-;Pc%_iqRXMAuDg?XXv z01K`NA?Bj#T|gi^$h2~T@{rLG8f z)f1aFRLAW_krjF-NO@Vyhk>HB5(qPY$Jt1)=-7KM6JlB1cD{tGyxTbpI zYN;D~R_9kl93$Y`0M$sOP_v&wAodeja6UMSh)WGijfZD`C`fpt=78)XpD;|c8E*e9 z%ZONdHb<-TofQ%=lDHLRP!SQk&>T=5!gJCU&dC-+9`+2Zz!1D(BT^iF#goi)=~o%X z?{7i1G{Mf9)S?eXj1_E1yUJ>6)HMv^IHiNk)9o1FQWgGB1mcZ{o$8Fk?KYtNLUX|OoL%I-wU@krO=y`Xt5UnU_0g;a zMTnN?TeKNRHw!UgWjbJ1I$kfNpy4ko++!CbUSU{#-H)0M+h6|c>z`X)n7>j-C7I@W zZdwbCT?r+dajspa$@AY4Zf9QZp}4b{HC!cJGI+1#$YiwYqCJRN#6Bd7M3TQ_v z*XksjsLor(T%pIFBX8;ab+MM&Wh!d-C|5`?V!VUoXRKk?p-?$lDB3JzJ|TR3Erx`G z@$yGuhEHNMQ?YYWM;TI$Nofnz&EuD@5_eE}l%hu{} zA~i#Jz^);uyXV6R;9q*%D$M&Dsafp+?Islxt5?Lq$l&~Pr_;g+oz%k4zkVv~*x~J~ ziJ=pMelFwyvgj(rEEC5tcdkZMRf=)Ynz@xY=ub;@t|1Jmvflc@7`P}{Mv5o|A)rSe zK>mdPPfch5x?3okvxD@biD00S!whBC$o`0okin#GnElKJ5VWy&{GMdtdWn+TP9m`^ z1|Ev(=e8i-D5U!1j*s+u?B>*lr+l|!1tSpbWJxPF(rovBY%ADbgz~X#284x$xh2O} zs%f=GM#UB&^SXkS1SQUNv65Tgm0N#j?%V0K0bqMlrA>^5>FL%xyJ%uJNg3}=h)(y3 zo}0%+gCrxdr9vM9-NvPru)FPX$`QSETPqDaPIU^^_O!fk(o{Z!|8cBAkFloo1ZV?? zr&7U?B&w0do;-9WWAIFfT0Y2vht?Xbg`yJ<4H6wAj9bCoqAdqHUifukR>GiLPOHC` z{RT@ji7X`XDk-4|f8Xt8Mk#(ZR6HmhBhBKp)2y?aeDS%o%;NXln?QPG@c?WX5fE$N ze#XcReJe)0tE^x&{&_l1fR7WSSOlH} zxN7^2jM|;-bdXf3b<$HobQLEvPQiw_A}qS&>j*8`^$Bn@p@1WRuelNE%mEgX&^FR% zIoT1BQwWa|^-=p`9kMo1?VKttRexp|aX(Oe2w_-70ZEIWykM-52WedRgxU$@m@lN@ z26A-^xD8!H<5a((JazXcCo2UOQ5kE0I4{tL6jl!TR1sh+Ydn6)VbKejXar-zFe)a6 zuhEBbTOQO#An{|#$3!3ZHz3%<7GLA8%8Mu{`sL=zw+hCBibrK*Y-V2pk;m2*CM0yi zq`TZ2cHi@KnfI*5tg@@X+4AgTeruv|b8B=!*2LLp=*;Z4iz@`nD}-_Zg8x=4iz*mo zfYO0te{{6eJp?Y=_9O2^6YtLRkzo`vWX$9C!+avsJA-sJ$?|A>PiB7~@{+BT*0wqj zA~rn+>NhEhw*W1=HJACMtVc6HP9CB6G9jsCqosqJ8m%6`^2!<|!r@E>n4vo8mf}js zxG1aUT!Z?DR>|`jcLsS5Zni8jAiNzgwk1|4}xyV<~-lHPa^7b2h=D#m4KdcAKmhBr8ujtjKe&a$GcTa z{9p+VhlCfSJm(pU!($JYd=4!)njB%ej2RaOl<f@iv@!rWB6;)ZRRf&vQ+DH2cJ2``PbQkl63kHYS5Vxh{7wtGiEH~YEU*?P}`sP1=L z+^(B#`Oc`L;Pg`j%+N%DS#7E8K~JI}0npQSFkGn}q4_go^Wjky|8okq_x;JeyAvw^ zEm!l?vFYohjQezy7bWjt-l@44u@d|YGz!a^SNWII(Uli1GkbFjhLE)!k79t}JU2T{ z{4JC;krsQbkofZl3^Uv?`xwfyp&fHA#UbZf_%qAv@hd7WzHYx_VbPZj&=B_D|hA7>ZnTH^NJhD zaZzS8eysMS=SGu-jLX1-%PQ0p48|jJCc-niXuV?8z$>f^_6U) z5k@AplZ%i^VuRK^|L!uq`)g7SzlJ=pfV@vOGSd`~702*HYR065y7c91{)_pmyTXL? zoq^>!8;lcVKTR-ZA>rU{|l3Zl9hAC3=3O%k~E%g*qI|4hOvK zNGUN%%M{kcq)yY!Jy7CgCO-2sXe$m1KpfqEXRmjpxyT>Kd+-l_)u^vSfp?7p#r8)! zu4PRs^q@>yRHG4dhUK#oEKdG$OFR^n71bKGI0U_cz86fb0HWoEKnjpKP)A0L)>UZ}3B@+ZImHNRj#KuC&gwE$8Ro7~0?F!L+c?kv6OPmi01YwM+lH@x4h z=sbC6OMpCO3Z6?T^6*CUraA(1-i4Nz^P>z2HWZ#7)SX6S66HQsCT^+8$ft5iADYiTooI0ag zhutm68ZEDT$;rtHaceJ;|5+`7m_$?xRHkVpoKk2V4$yUTB&NBx>tN+liOAS~=|3(% zy+NmDgW>6on5+Vqq$0FnLj>$=ZdJXU>#5}O>TA?{Ml2x+qgp-Kju0eU|MrhHwg>A0 zZ$9n=5YYnmO~C@Ux912BqM!GXRt*G1Kd1M6V~gK|-+v75-#6X39cGPD06+Kbw(aVs zh8dB+g%ltf?}v87>XCUcuCzwBi{m{P5K`Od6+migV;ZlC>09HB+K6(SI!SQX4WfWQ zYG*twHT6`0wh^gz`{Y3RhE^^%BuUKq; z%}6agLJ0N;MeADvCaqBwg{7rAp0h(07(`=NM#;JF4IZ|~i(WpWhQeqnw2cQwp(ZJL zXw^KszLBa}i&mPh;wo){0;LFD7i|STZ8V`{bEd|YL4iBVhXJqa+0dqAb_nT=+;Cc{ z0puZ6m{a-=Y&JopIx53{ni|KH)ZvGtFr#9duTFO=!&XDdBY+PD76Aw7#EyqP-+jK^$nP)z?tTu0BO73b_rSwY z=p$6E$EiAFc0)JZtmYYV&t#Dx_VO;f<4F29!hgR`JnN8L4!xAsTG>WU~$_mG5J z&c|gZWu#i5l)C^mX~E2{uOjOjoR}%(BfZ*gi^pYYr7Bi3#}C za9X7YPu=Y4f0=n$!5RUG2-irS4q-d94^D$hK-wwu@Z{?`(Un|k=B8;cy?^mVVrHa~ zwq@S@t^T({mJGJmG={KedEZNwQ`MgzZxK9EYXA6dSAvO4d}IB|2Dbr__^UJTn+C^D z*4S$YW(}PkbYf=>#-*A*w{w$tUhS}pZOVS_6k@C|kEMs?aKUxL)tIQhe}iuR7osUJP1$=l7GRZo%!*C^OhwbOYeN8}LpcfDd_DUh{5EWzXZ3<;)NT_xu6>XC zkUJ8MN@F{>gM{=vW&*(MK!&%SuSTjOH=^&A!os=r+3`wlC!us*Tk}WZ!JN#d;lf(DNy9LW{DlJ%iKwBNzYB zHI#6!-nAA4YTdOVwD4=u6fkwN^8QfAh$t%kN5Oo{dG+wa(pSI?KUkiWBI4Zla_v&HC)3#~-3<0Ng1CJDN}H^@6X7fO2)FnZ z2A{mzzL zUW>2)p(j(!gM3Y4_jf9wE!3uQtWS|nD`66Hvia$h8{bUGLfQwt?o>KPwf6yYOgV~P zu%|d$$C05%4L?@c{q1q~S!7RD2OEPxQ_y>SAffHLeOBzO7m!Y&2#U9dOa%%ntb?mybvUFj` zxNurdJC8%cW*<#lDOl6gfN zkq_wRRlhCvoKi7D%udOQuTD*Uw|pG1d`9Qg`A+e5aw?kSBvO;^#>B`97+g*n8rqta z3fYU^uIk)Grxwj&w-aFEfZM%iY&dOCR^(v&I!>cZl&dda#NJmf?#BW-s-`}YS8cLR zohfBQ@{(12eD;=VezVXjZ-<-xw?pc%3dT4PS#c@cRk*1BRT80l)C^*)eK{4?XNw2- zhMCB2h;q$|m1=XECG`~qFUuH$mtX5@n!a~ks=urE#Hw_54OAEHR+VF!_rv=adVqeYWbw8>!z1P4??1stqY=ZSM_yRh`r|IO6L=sC%mOo3j12)ix z#PUR3*GpRvwNz4kLD3|pjYiVx7qfjS!le5_8eT192pURh$BR1~T8ZvGJ_+BX7jfIF zW8|?<5S3Hc{Esk4|2Pt!?+kvmdrGZg&<=wy&MOb1xGEJlc_B z+_!(oUuh?iK?Vgx?aN`G9eW*KEz^xsHy;iD4XtQ&B(i%lgf|X7MP1$)lZl4Vpy+>w!aTHudrN(hRa;Ph`_h;r%Bgx65-5e$;PBR^>^VKTx)eRDv4 z_{+M4JoOXNHWa27BfxnUjF485CYjpRI=;>*(EEN8byUa8*A0QbZ zZ)gRvzz~Cv`DMF_I}7o)?z&R+rODVAYF=vHp{=e`fKsNX+=_9PqvI@&Hy&g)`^an} ze=gJpSB3meWhRIuLib#cS~Lw4aQa~T#^sGckf$>`S(koipKl4VZ9y`-m7BG6Ik9&r zO0-7Brmx3jlV^K<(dx{uikYZ(pRA@nOl33kaxAOh>asy?7zJn%fKD&O$>=POR884d z+Tr5{!ZvIw2Yu>7vQ(V!T%WuJu4FSv2J&7aKx5$zI~9Qxbp_1cXfEzS6vbO}nY!pJ zYFwTTIE4h@V~CQa4an}NK$?z;evsN-3Z&;cRrAO~EUf{ak@-NB-N<7Ll!B-zt?#+u zfoVs-lzO1)l^BekPh9&pxIA^HOw(u~5+{npGoVqkOMIpRJKc#Mx*bzwSVGVmQnf~x z70Ve->Qh1HDf!eD;M13_`7g4~lmE_O66jqKyUiL-zrNL&wL5Db##628gGyeA@~qWF zCMD%15yqW%v<3IL7;Q>xVS&px*XTZDNd(;h?Zo4j4P+bbm?E_c4DSU7&>Ck|k$q1e zI*)6QYg_FBjaA-@i3giq-}|P(IVhca$p#wp(}Ru6!&>1bT0j|IJsJa!SoVD=xeX>p zda1jdK&U3@?GfKU`REjm$fGHO^qzGWB^baK37@}v6nnR16+c1L>d%$Rj(^vXoB5qb z4`@&)F?@u;VSWbqVvEyQ^rCep%a-y4eZlA~Jrd?*uQ`4-K&GR1sWqLDLVrVGz!ODFrb?o3qBUJ^l3xkpb>}Pp!2fnFWHOPV^vZXp zko}J}H-5o-$BvA)6h%aYGDQNq;1H6i)I}(+Ul+_!t?pBskxv|enOl#?5UI0}=|A-OF?<}vxorj@M)&Y&4!p(%3hY{-~0^-0zg zFSb;~)6@fwm<3Y#Ib*})aZ9tP}+~4R(R~P(v{kp(io40A(hf~GB zYA#TW$dE?Ix7K}nlGvj2BeM_<)KW0@@9B5RjL*iye++LxZ32S~PNd_kO0ZRHEi5dS z+cGtG4^|L=-ZxXi?(~=H7tp_s)w%)TnbXjSjTTCnutTWQ8RyA8 zVk#JNTv`|+{ZW6kZ_qal??d8qya40Ms&6tgjTzAl6hIRm9A`dqd%Fq9(6D;(_!H~mX89amc4O{5i%Yujbel4MS_Gh7*ZwR z{8U-J*k4c+fyS&J0rD3$UqGqeY@lHul>wPV3m9&NV4^BN4q@KX?GYTbDQiL1vZfk_ zwlWaj*N@{l87jBar3|G77Tl$Hfj`D_mS$>=z1o752v}>CHSH;q(X#0_!YA zqm~?Zpy6mCO&Lt2sZmYg2fbQzqVH&)|Cm@j_XyG#GNKYv_=Kf|oMfQ%!huGP329nk zBEyZ^M?ShF9_urD^1oRpBh!D|WhNp4g=Kv_>t{HZ|U( z1kHD96Ddzn<>aFp386F@$*`{XO2&fMvyoiM8m8xh#$2YMLb(F@#*jK0D^eUn?HR&V zc9LU`3q(#Os06p)V9K=~1oDGRBuQR)`FQhCXk?OENyowQHHxtt21lPvEU6HknrxcR z`tvH__;I~@Zdm0mC5FVcRLLJL;jZ*v?aA$NQO6q(-Wr9~7vsPp6P);c4!BdKO{ueG zq#_>02O21cQ;US6}PS>+EN(o&nVG>eNeRl0y!}t5+P%W3qa=_$ER3qLmLMv$#DI3 zG?L>;I0s7NpE<;oo+NPFBKFMSY z$!|uJIjMHiFgJ%KM1k6)j&W5yrYQ6i=Np1A5|I`Vok}Z)_S;VTwW}1?u<^`rO-LwoOfba`mTs zmI$6WR2J|3;s{y+ork;GA}DfOuoOO7Eni@c&@TfYhddshsz`;tM(R}GUOQ$7QX7(U zTx{fs?~~~0#>@8LGDv9}W5&VElI$IphT;)Ifg|OkjrS#*L{5!@v?xaCa(GNvfJN#n~ z0b|W!W9(A)=6}1)SeTDOp}1^B3z(j9uCf$HgWEcW-cuCceVok|kOP9~>PS4;z8P-6 zMGmx}y5^8}_UJP{wcfWV5qbzY3Tm2YBRVpv9Af|B!V&Igy!p;7BD0dEI*B7vD>xQ;C(tiK$?7T6j#)20P^K`aL^2Q( z?cfQ68Kke!durs&NpC($k0zdrAVTn{e&jxaxeJ-6+C*N9So8+YzyPCvU>17M-AO@! zobY%`hHxL_7eQ%7qB=*w_Q?OCECt_!%2J46$o=83%Th`Jgy_+egY8-MUzMdY&{XzA z6u%A2o^-<40dzROFmBIHC9VftHf&g0y~IoQYxslyAa^Ad-$(=H3H5+7=Q;B?jzf>B zw+lr%y@>FnQTR+=)14lx9vzHx4<0~5Vk9HDu1gK;Gtp%=Y;m_vxMCKnvK>P4j zK?QjRN@^Qmyd=(7SP%XySO|M6-?M959pfW9>T+VoKl~Pt$K_inS!rfBXy9tBA(GnD z*6j!e^bO$B2Ctr6Gg-R)iP9dy z#=gXR0XhX#=b!2n9({{U6bsv8oXltjQ!?CQ)$ZOu@P26rZ2-PwrdhBBKIVl~TY((5 zCeAC!A7zE+KJGSjj%Z{b7|W`b8PHPG%QrNK8N9lCQT6?BI6WJXoIMoF1VbFfL?u)2 z8EeOI^?e?lX($jBSgsD&w+fjD1P99eTJY%LR0OAx8W^>#?|02|;-nLwE6s{+M zP(DEZMlO`2WSWah3Z_y;x)x(qn$5CIsX1f@Ap)lkzzQ*t4UwK5#^y~nyzQjY{$%}|&6cdC%2VFWWz)mo|0$G@ zB7lgr;h#MJ54P9by)w*d0fE$p!WiTlJ`|wmzQr<`HP*r(r|Y2 zz)6|hslaPTGxIk^Xj-c0`0Md6lmF_=pWa5jJ$XUl>FqaLX8C)IeYa)r-}#sRxb=IF z^l-u9;W;5yNn2&hWo?|hcgTgVj~vr!uqhpGwCT8lisaDBY~Y8V8iTQZ53#e z%(rUQs$PITPXSjflyT_xqPZ|M>qT(#g&-zJ9g?XquFY|GC-HE+>-Z2&7Wrm;sLS)f zy{`bioTW|P3Pi!DG)m@Kd%5v#{2)lm(uaoEgn*^54n%R8O_5kUQnhWkEk`35=i{-Y z7z=ZdQ!V8tW+|OZ!rQO76K&x7ad1PXF?U7@*Y2L!#uF3m^1e=4a<7MI3@JKkB8{*P z&INb1&$u#;W^lG$H+_Yo(5E58LFr*JjWi*WCb?XfM>5~o@L+;#p&VcDP1tjMWP;GxNt#!x?kQ`)P1xt-i}kX?rThj0~nA-`SF+B5hzO214?%^ zW9(I$=<*o_+o`z>C5D;@KEy;RJdwFQ(^EZp;6MzW1y(5Y$qDQ#uGMYyybyLO^Kjw7 zz`&E7QK@O+sY3Tr(Rho7boGQ&-L31E18JoZRJp)*zohG3!5OT{a(0+NjG^J~k||M@ zCcRPshkFqyo&x3Jla-Y{);*5%yy&?peeK?T`}jGSjK!S`PLRmny?bB5X88gxa(Zu% zF1E5)pyhg#rP$#R(T^WckI|DQ!aWUVo4o}B^4t5ZAe_tl{dYa8312idHU0D~%o5Y| z$Z+flKN!&%qiasaQ5&zcWWy*jukyIA00Aoc4PB{C%eli&OQ+tH3dgj=rs2ZDO=)-6 zS@od5=FM;Ma0Q7$bJMNUmR0MzySjjb>V;*ra3K+`Uv);s`M|{{=X(AvMv@?6VuQdA z2d8J!|EiFM2wr*ZqPn^vx{Fx%;zaNs=?loISD&VlJH0e7_0rRmg$X3iLuWrcKGVLd zj@g(yMrnI9ZkKsR2x|Ns<5Be8PRld*1U|g86Y? zd9=rt{ItZjZe5hD857?<3{@dbEV_PX92E2W#-mW1bM9wDQ)^fDiS2ht5o?$v-5Au6 zsA+^DFeb^;QFivXHDhu^WSw8ZI6=b&`MVbC=bs-8WN0QN_+M(Q#$oGTMpte#>9uhPVwP2?tN>lVAqkH+(k~?BV_^E zi_V=pw=s3;(xt|lA!h80gmI3IjonsRFHzOikRj#Ql7*)CdaT7t<-l`S7H;r6pt9a? zY2N?RkE}GoxXiUJx|!b$4Gio~)->-lsx*92{<6_iestnUTBA^u#Ih`ck zy|v`^uf2*={_&-AQoq||{k2(nx@?(TSW=RWMcDf4F8fOta$Q9fSkJ!sT=`9L^66Q_ z8^VlkkMrJ*&~g<4X*DXL-1{DX^KJVfv23%b_90t+W4lAIy;c}XR&nmS7dJlacqBPT za9>pEvEU}P=BU>pP5E-ZO{F7a;m*r~JMCrcoDR7@J-O)f(T@Dp%Fl?IqmeOR*py#C zYt;0%>BC=2_iYjmNxf!i*e7x@W6GlbPTQSP=|V=lpcIeX~SWkb9YnTG(WR zORQK`P{^qCa(%bwIUSO&UKZtf^w-uyLnls0%Guw1Q-8M9d__>FjHyh&m2vSBMO;B2 z(-jSQn!!2tb$XGbUGbC2GDcF_Q%_w4HYLgBer8;(T4F0d#+$gY(Af!c&W4Fr4E4M| zB(MAR-P>96OyNE;2XA=?(RFJ2g9>>{oE^#KvoYPwZjW5w2A_)~#qA-j`|`4X)NTn_zVO`B zx+G3$gh_Ipt*9bPsB>`a>ZH-_=H*M5Dzx74sqUOO{7vC0^IwAWf+l4THsAko+g;go zt3$@d?e&LLW7%Dc7J84Xvvybb5AAl_voX@|`uH~yfgN)GrAA{C|NL9nf3RYVv0|jT zE2iq<=?TeZ-{$*MLM1yUQx}ZZcaCT#|2ooJXe}LLaT$im+5De zHeMXkkFK7yFWB?s#XeEf30E^X1X^`o%zN5#dbMA?&YG?Yi_Bc7ex>WB1xYzUr4qSL z<>I<|`b7)ImNbOfru`(~WAjK)pWRu(Zz__rnlDZNXZJ8zsqjA^r zpX>#LI)n4o%%7VC_4czD1{c@v5%C(m9;kWR)oWJS^{5-K-3G)A*_b~n9wn0);!095 zS;rpa{5>WDi7yeAFKEUrc6D{_1*yrt%zVM&Mh?Dcg)Nyr+h`KX*eecgOl-*lJ1^RF z9B@lUGj+6fOpP?3h*~ud=5l$jl%`sg_zK5|dwgjXl4s&mj49+!pSyE3;tAnABT&11 zLqqJEfo*%@nxzcM2S0f_8uI4vAHUb;B_j8mu1VD$oAV%;O__@kO}X@~nNw zS9LDDqcpkW^K*VRSyX&-z>J z9ocI_YI{Y`C+(B+S+}w4Mo>h7T}JTA+g5_OCXvn#xkE-#V$zy+IV&Z;ZpfS1Dr?-S zCU_8b|B-KO^cydUi9CCcthlWkDf6gakC|H4uPxYxRglAHKai$|KbS-5Xn?lhu1 z|F0gRRR<3Wdu0VTcc@6QBg*xq>9tb<`)(WQ?U}3Q|0K?B(sohQxM7#{=}JkgYWAUnV zSf<-}b;&uesq>B38D_6rlBUi#&hK7WN9nA~_}9woy~DO32n5%8SF2l2DfDXu4mtFv z<*Hjc;TiQB{cid;dq6B*QSc7_D%)J^a4TX={f?REaAeynhijuIkv-8}MS02nCu~xh zq)wb3j*n;zG^?B)ld@DkN8m;6jRy5e&q--*=3_*9kSOQx2Hucqffa08xr3?^Ai0yV z_!OLp7q5+0eLVJOuT@<|v@D4a(u}0c4v3&QhN}_rlA8wiOg8oiBur`_ZscFK%P3$^StO4i)C-tWNGeH*ZqN2u_f`{%_f z&Yd}4@(z_kOIlwPB$XVKG5@84Bc{1&-xke__t(%rTy3`=xOcRo;Es5t{Jds&F^N!Y zkpWm@Gkf{jn9kP&(!5Qx)1E0lP~;5lAHP;sR+1JNsh!GRw~}u-9@Eq`BX&CuJZ+I3 z-cV-y*7eZz3iE~zePh{XO}}6Zkp|u+$%($b23oSI;rs`q^bZ)#73+`J$(S@9%H1<3 zW&6lBP1lgli4;Cv#o>v=Lyi+}xz+LwSVgI{iq`uT;*;F*Z@GOmX0KARv}yIPr_7r* zWZeZ^a!TlDYO9ufU8+?iCGO~d?}WQlw{m}_%d#HPtMNMF`bmL1j!OrZ@@B|g+UOJB zpe~&5o6HwyYxC!Ib6+hAI}J+cz79{D%Ml3P?{@XjGN(Uw_3Y7Y%k`%iyL#aD&Wz`$ zT;J@hP3sR~e=ngblPoS6Q04Ms_h9Lh@9}p1C8uP1zCFA(M>ZwLzWoh8T4<>;-{PZ! z%^pc@>&t6=<#I8h#C>CablV~li_8q)%Ux>p1T!k-cW3GNUN|k(oi%Vz)SJjOy`|8` z_HVMXEP~&~G76{Ht0tjn9{d-3`HjD&qg?grbV9#*zCxoGK<<6-fw`HesTU&LllAtu zFM7$5gIfa<{eo9O!27~11!u%;GYWye&;dMgpYWX)Yd&q%Ek7APU6C|_l2;ye-uzk% zsLd1$*{PO2_e0L5>ZVy6ki_YZ>&J1KXLBiQ807^w=^%F+NwbAHWaIlD?(7G&6W-Pa2ewe0b8H{HR-p4a$?I;q+p{k`U>6t?o7;y}gbqsc!|Fs!GK| zP$GPrp(DHi{5d*CTGFy{$LlG^2RU7OW7;^K^r$*ue&rFwJOV;jSfrt(f|%#} zq$)bEXBC5PpZMD=%%FQE9y)r+%g1N$ zlOOMLr=45C(xWE9;2?REE!`*MO7!??5SeTmWEE~`-1R$H?t$A|NppxnyCl*#L? z&Xp_4H+2NKPm_{n_0RM8A>04PyA=&SDIGMf@?dXNpL8Nk+lSQKs6;E0rf75-VFKpM z-w%+SxkGabja03YHNB_D-KQ-l=!p8&u_tK$;IjR7T6p?g;P88q7WxY|J4p3fL>(nE znB9Dlx(a~pXvVE4@b-0RdN0q)%8Iva%C5gnIyy5CI|t6=Ijv$E5m* z>I}FRRt(qr5JE)2cplm{0z}BYTG+a-@CFDRT%;_=&2<^nB?8r>0huJJbQCA_@67ocjFq&bzOO89)=-+4R_Vx`@>b!%6Z`|E?c1>b%QI<;NjemOG$ z;O@!ZiKB|Qj$A-9T80>b8c~UMX?`H<>G&Zd@sh}=z@QNn+Bp?LgdG%#lw?s zqVTEv;`g8W9Uo@$%(11+b_s-eqgz}~U z{nMw{u3U*Nbnb!$`+6cPFMDd|{=uHGea{?8CnSIW1nB%TUk{#%0@T7-yT=- ztUrXu;;NOm)uNuheS2mgR#PDDnE}X&JUi3z8*}H)+w-p;dG0lQercvbxWCq4VAbW3 zN3*S!8ysf4di)oU|^WgcX{PL@xIwbwSZ!o(i|D&!)YvK%3)esYV*9H`#4qNYc zTPMzl1Sel8&n(j49?lTfJ%(o1ik{AffIDy7Hg#+${8xnb~$uS0f(3fA+b5d;Ih%x9jewHu3*A ze?ue`58#;xp9WPMB~;rL9ObA=E-1dl^C|Y-rF&=doHCx+W^yQAy=d)+{YJeWS9Ehb zq5?pkOBV?2di?nDiwqOHmhX?&U2{tQO389sWA1eNm&jT?S0B`w5Np72>AkwTy16et zvi_kMT0~neU~2tE#-h1)KIILs zv5Q-6Z=ty7Ec|pMGlvLk&)gt?otQL9-+t_Kt{2j|Zc!qXSRRZ2C#4wgGl~b#|8#X9 zO~5X*!X%jsm|XZMc{{WA-BmxK#!cNiei`JKvhYHOZ0^ec<>9zVf!iLR9?Y3KzI(SB zJ7ptl{}!-KxL?=b-Q{bmwd}pN%H_NnX$^e(8P69_^-D9qyu80|CYtM6Xui)hJo6oo zyKViDMPK`yo1^F|Y#hL*7t^eiiO%CWm(Z`-8?Wv~P&9a0d_*lciPn0hG|_dg>Q zrkgNp*sz$3t&rbYN@^Sk2oM&@(5Qzz;4{25E9VtPQFT6Sq-jZ2ak_#^P(a^hlV*cU zi%7TQyc*-gDNP=qp1H2?{#^WZzH=hH;g$!UnS`8&9|kf$BrQTCl$gr#;6vT7=6IWD zL;aj{drveRB|HWg$v`%<5sVvYe&N>9kldgQ$O z$oYrRi!?*=;Y*iyisL>Rr)WS!&1f$YQmBh;{)F2=v;b~J49)&mWG)7K!LO+f+;i*5 zX|rJhC1Noe#dbae>Cs!r%9J`ttPOlnAL^D^f%lq6BZanM4#%2$8t>^%oohLm8A`=W zV`A76!ez7uPXIZgMs;okY zkaVshcQ1S6fpo8u$n;=Pv|6VO6du1<**ha0`uu*S2~J z_GE8gW`M==U#Edoeb0!=pX@X{p6>vQT?x;mc_=`f}t)AsHc(pY)y)!bb zl_WI4=Yl>V2<+`^T$YN`I;@p&oM+FT-8GQ>Kyax3Gr^(jZwwr1=Q&U;Hiec7CRo_o zCbbSxF9;awRzG>zuCZf#z=XHd%!bE6-6n><@sTHy8?YcuHA_H-!75FupMuuKOcMoo zB2hoO#;RgZu#+apAg(KcabYv}_J zGp}98iM~OL*pnu0C~^oPQ7BwCanO|<3-{TPiN9hYIe-(HpK=Yrc$1s zGf|l1g^t!?O;Sj`Yg>1tZN1j-fOqRme{tEf{$J*s-Gc7#Yq|7v?Z6Nvbc-%!2~>xM;Yi7vUB9$Oj44HBOHJJf zvvT3~L9V3-Yo_Mejw_@p+F1yD%692rkDiy?tq;&S_Q-{kma(nLTV7agp7twj4Qo9a zf30(B4GiF}r&s2cIprp=muXOM>$7oAk57^gv(5cdoS$P?mSbqCZL1e_&i77bdqjz4 zRQ$7ZXlV9`cEZQ}_CQg`jMj(C4GL1@dKB-bfeN3dZTCpleCnBORwzwp=Lc(q=88=bgv zo9fpOPyf~|SrwFQzrLk^VRKt7jWoD{}@e@QPGPN6IT!nk29LhqM9<+s3h>(2HY@&z|kqZ_v|~C3b<4 zi%ZPdh)VpSs@Y#BmrsSrF^jO>zm;VKe3Eb-3)+bG|AgI=bp_UZ9OEw#_P-f?37is5iq zESrBit+F>+d~sm+Xz-i<8aGXEMHyM2!*+MjUc2Vgsq|`~Y3jJuMC2WPja$p}mf4+J zWHL+KF_M?B-InPyxY%noUvd8SfjjalRHlgPJD<&(E7% zwATM3mkyIC7gzl{jdM!-RCIv3jQ3zXuauP9(ik*hgE~gM1{a6^XIGWJfmYV>%Ga5V z+Rno!n-rR5AJ}J`mJE%DVNppoI0$Q*i@UkW+)mt{abZ+O_e$7LRe9UKjd9Eh?A|A+ zpF$j{#PonuYT{ozVIW8|QIC)PR*#L8mmGQl5aY?HT|bLwr@dP|CvD_uY~b{Qk1`UE zf&Gm^yWkTaF%sM~8ax8Jb*;JspyfO=Wj&jJP^YX)4!|tNq_|H{`802PSoG^SLqMr# z?RL!ChiH|V0GWMh)O&n4JXE|$PyzjX!;o8>NnhiCKrA4(7E1p&^=aY-IinXj-J9bo*Y~Hka%o&!PoSrO9CzJ#>26m`$QQir z_WAoi*|lK0(91VYRWNkDe}0|bZ}v_uN;?vppP$VWJQhBAmmMZ86*G2L%EvfCk~hN_ znR8{9jQ)6HC9gnCy6(oWy0jmQOT2I`QYya^W-%C_6k-)qa9GMea0h?-WPON*NQ}=q zKd~os^PGYQ;xzb|c(V;Btd^z9SMxS`PlK*gx770xw>NAiACnsvynJ7H`irDK=`==38>lV3R<(ivS0QmwXDewWr8a#s z$A7UVw?jN=m@eq*Rf}PmXmV90wtFD2&;8pH?Jwn_pXKAwiXh*jSNB9+jrWa9KNLq_ zE$U7mbYk@@@~P9>b6W80rA+3)H7k+LVQNu(&{aC1!e2P*k{uC>pmtwnz4G#y-Me?+ zJ8C@lb79SyeN**{Hs}QJ_D$Kng{}MdZuny%G5G`7H&Z~>$!&L$6HL>#D3~gb?D3Wl zH(sV21WxGSHJQ0oNs3H)+%XiN9Cxw z#JhMmGMU5|WN!^3*Itl&#rz6d@2f;AAG8Kf;7+(^urhzq-QX~1S#f9SEC=C1rxyA- z9t`4eal>4dm_Yq(8-vDgP5Z5wo971xfs@*Y-ShS&FQ@qM`$)H3K+z~#>ZwaueB-Oj zRf*P{ygkdd1o#-s`3|- z^p(nbt=LPtvUm%s4$&2Sk~393?W8%r<3mq=6x(f*@hJA+8*0lK2l5PT$?|&Blq0_> z^X+dVs3WS?Y54I|>fM3S-IX-#40gZ7#%b6#_+WC(t*$<>`$KHq8Amy!*rzia$vPyv zmKVX0508(HixcIx?5Ipz+T{#_q7~@BQvTqpAN%L;D!F-!7WTp($!c#e<47WVQC`+| zbetx0SS$EmNmPB~`Sc;AUCr_n=cj);AV8wt@;+~=#u~LWT)1uMm_s2^Xu=oS2Ey`?f=&Wor_1k=9b^=o61CUjD{J2{O`{7@3}F5@Rle6uPEf^)$;G$h zcOb{%RLA)qji^W{B3=1INGAEXB@-&`yL_J<_A;0ePhZKy{(o6&z4bw)+aSDOk}$1$ zf(BvLi)^BOu5({#`S|B~@v!x&Niwq9l_lK}%O5b$j>2=2yfJ zpz9=t53_I~sX8D+7uSJiAQuC)zanlbO*1FzvdR!V?!3t2T_BGkzoNQnqaMxVB4Pr- zfTsP=L2sj>s}G^w6Bg+wvuulpFMq~;2rNGG2P#V^yZ9{kWnU|R3zS;kO zv-jokRIcs6tJ>9WE=>|CR6?XQk)cwi$QTid$dE*#%bjgqJd3R^K$WfMH?N>ie*lq97hSH>P*;S)S9`h*v zroLxE{qS7x+0dOSfzIX5>-POz9llgqYjes6iN)uvWSq(AHG`|}oV-OTaW?(Yy6b!* zTWdBYnViURQNoW8TPM7(IAv@ivRj;+=A&B!*o0~V_PS7P&V!+ z6QF8U`Td$P3()gNO)3HVlSa4w4h4Ar67K^k@Kd@OGV(Y|owsjrD7fH{_(cRngEsK_ z+3fQuCH`1g=cQ&?NhqXJUmm6QQqb#=hsN+vMF2>ixzp6?1H;kha`444y=gE;iIi!h zh1&P5@bf2@4L2kZgDw2I7`r3}SVw%$S>0si&$;(xCae9&CeT+?^SGhF$N|1`j> zJ!)iJEX-y(#kmh#f^TDs<$erud{Axny-suISIX%}(q7CPHh2(pU4LNd^mcBY3S$DHJ!LWKPfeY4cM2%^k;!b5_u&J3d3UpLEkr zBPx7q^Yr5TZyp$NknY_`p{!Ve21atYe?Hc*{6Dla|3#_HY}Jd@zm|`2{YVb=DmKGW zB!y$i8Assx*=)}F#9@QATFUb`HTxTW3Xt`sdnQFHul%zDG<`Qu?`*g?=XAndOvl=B zi)=xcsnF$EIIW*I9zN?lbVX{Z^NkBX08kQ|@3Z%_OG z_uh|B;orN$|GsNCy)w`30dz!J{;AbWh4bhNtI}cp?LM-=ysCV*QI402Jz?4<0&7`m z@~7)|?m|t?c@J{1sG*o>rlB<7-q~5q{iJ7|Q_PEc~Dx*P#50{tk_N=oJzd+3Tq(`WOYI9cC0 zi{j4~TWBjj+0t-3bEkrv@%Jq|%>#shux1G2B3ku-h_=RZzgY#v0zcb;Gq$DK5jdM1o{4SSF^N7wgR!WLV zhHD)nzRSSy01C<3^nBoZA<22lIw7H2 z#bHsUhv}S{xUSPQ-+68fV%QHK6eU@&Oful@{+M7%NtPqWn@w-l?Qy4$ZQnrAFe1SX z+%Pa9a1WkiTSD2fzVOl1_XR5v_--Ga-^VxOcFg_vRl)ya zE&t_(8S{xNRLP>Bb};~QF`=_h-&-gCi1|D8%Gq$GXat9j{Rujq{x=7H z`plq6n#(c8tYNfIfN`|aJLVPT^u4;Df*cn&9N|DT>~yU5Gi8|p@3DW&G{hQnc&opDSL}Bm55$cd+4mOal^{6BG1mCcr#OG&1UiHCuZ;h3 zOWX|&kphmGTviB@dUjy8O~=PbRRTi|KDu>&N-t6O=S5iYAW}j+y$_G@wX6HSAc`_a$21bn(GGx-S z(81Md`!3c48a%V7&&^H}41=ElRuqg1OA3#5gkj;|ro`^L+ zSI_lG#%cA!8&`mq-vA(m6Ek6Y0ARW(*I}%Q47M?`BDeE@j9qx7dEn1yO>vb+!Fr81 zl5ZR&OGL`UHMEbJR;Ko#6Fkqy7#b>+-}_m}IOrch1xpw@((6R{pEB1>N9;e2`20|- zAH%Q=nWoGK;PUAER|bY2OsF@}%mwKUFpJ?EzU#c1??2buc-8#cr(|Q5ihi_AeL9f= z3&1%~Cu=)G0?{Z1Oo)ohPq+Vc!t|%nMIv7u5oE0?)j|+CGK}vsI&<0ttGa#T zG{Qj)?0v#KKT{tNE&r#-vyajZS`z3od*F)%n+Q+d<&JVVZl;=P=sO9x`SWMnza=eo z9y|bqg?ur}!)(2wV5cD&wyTY~4!b;fIhp`~xTBqAZKc}!MVBcC)s0A1WtaIJ9REEB z0zngqv1f~Qiq-No@HMfp%Zd3(CZv6^KG0meh0bMJpf@+HgN(%0e$bAidDKJ)U=yRjYW#o`Tdr;M53<({q+E!1FC-fcxjkJcE|XZ6&T87 zeBe^yTZPXG@4Cxq1LspKUbI1FrdBa(jjt=zMh1Q?&HbV|L%62SI}6l4c)P(%av)>4QbJ4>EF`*Y{1_^%FIc(UW%zGz6mcX&``5 zd@XgH-{p`P`^f9k_OX&|WGp0%xGmWAg7PtppX>rOWu z{VeW&uKse7ZL?>D0iTMS>(X3Ld8=bd`{VtpM}myRgFNV4SmYQ(x%=Ml`>-J=;Ver6 z-(Z5yreIGa!M%Gz6B<_O_b2DEqu3%Jbbxp315bf9uJQ)wfMU>2+HZc(+O4}@q-|8g zJlCVhpM6l^drL3=;auXK6CtF=E~02XI&kLmmH0Ely|1l`kLj-)ixZVvi5ubC!K$SCa?1*hRk;FbK##D;o4uP1mK3?`J!hZ1FV7%Wrj zxGGp?)snb&Y-;!yOwcLQK#$=4{=C&IRz&M03R(9yOf)QOy4rk=g}Z0V z<^q!#v5w)?=5@Mz06vI&E#+7sh#O8surIJwwAj5t|=Zg^R02Whu)im z;6wIzg&E?7u>fk^y?gfxGhgJ|KD;_=e%#hRG&Ix#R9_pI05IwZE<|_Pln4h0GE4{1 zgJn@GQU)9)XA_FW5cCMpA=RG1%lKos>Q;S z4K2nyf~+6V#CCnM+F^~p^V_xMk6y=9j~FD|zTRzd@l#B%`Y~M*#%rNEne&b_H3K z4Ntg8ii7G zk;a-e*MWKnHrLMyaAX8`?7FZ;KZh_n2=PB?7ptzQ*mbUzUM`%svOA(8p+Qo=LtocY zGCNGe^n9a}hEwH^_#9w+;+55$g%61-a5xXg{-`qVnuC$_aY9_Tb*{hW;Q z-2a&{ujdt+%!yLSdv&=v=Vr#}QoV=-x2ZU>Dv0ydwAhNguc^EI_EuaH4jR))8!eIS zZAz(AsJN;dA7pKE&M~X%q4w`1>A#k~y}|bv#00%-E}viVJK1%V)>Pdt;k&mT`sIB0 z#2)uLt2t)h6|k1$=}Bw;bs~G4OBYdgT-vqnT*+QwgoJ9d_)>cvH-S#f`G^T>o~=m< z34@D5t6mE<*c@uzUNZ}q*ul>rE`jFBV@_G`AK3I!LQ+T_#@3hLY(p1{g?Y^mtb}Y~ z_73r&kdqp@T(*=0WOJ0(|XU3G{uLy7mT;L%7}czlh4r z!z%rK8qrshZkyWkN%;&U&=c6qigx9mS3a&c7#S?pTUPb@$c2}CnDgW-IgbX9UR^bh zH^HumQ%NJ2M}d!VS=C9z;z+WMBl7PR@ekrK-3a>sKG}AlqT5@|c2bjRndrEH_Xb^+ z^_8RR>)aYkr=eb3>r0#MngZAa3hnAAPjL?Q1)oa_s?A8gayxl5I9tt{+HIBwM-g&FyY0WAZQBIDM#d$(;M?Q?`z znthv0XerV6-D132?~0BCrN-MW5;(BNxai~ioJ(;iss9~?gPp7jJ>lpli(WOWQMRY+ zdv}*q*(?`v4-d6jUU%i^z47r>=Ps3$ies(x+JMFjyU*2hSn#Pn@>!&=Ibgf%Gil5R zDD9-7Rp+NguV1hk1JTUp8dqMft#pgZfiNa5nGgy6OcbmxY^HiA(szxVSvtc#Gzw6mQ z<1YT`F2n;R@w2;7U{SOyXqvad_JpC}cW0r{HmO2E-+@oF#QmL7Ow9|MM?#}+7-d@Z zKFLz}hZ-$&XRgagB|d=|d;;-5`vm$=pV}TbP9$7RgKe)e`*dENa+se`p&@KHKOrY0 zTY(5fv0U>*Yg_g}Q-`N8WimqqutJ&sf@Z^CAQpe? z;1XiGr!C+Evr?VJ$T~AkG4xoUWBqA}$#W9i%p$|R&~R0Ks3X z>d*P}S2XzG#_u5c7J~2T$0&+5T4@)866^om<*!TY`*Vhq!Z**0vBn?Vaz!A2FVoln zmI-h&N=2%goNNVxLzA%AxwkYlG?32f%ymD3ES~>{I;nF(D}lb7q}aY5CA@^&)uXGf zpjCKFeGw*5Gy_d{{$vqpt|z_XK!@JGolZhrLV0T6b1B%BOd?GnUG{ksb$RZ~?7z&B zZxR3F+Om8c^<&SbpHBATa4+kapb++9(A-Kl%Nbh^A7avEk)bE5>Q9_nL09>%ZR|xT} z6*Fa#&P>6&&=QU2c_V>&W)_X-NMl&TU zp{Ld~-u=+yyJu7>61>P8g)AajyW6+bgZjPB&-ur~o^Dugw{<&5lR=R{DWAuil3UDc zy>z>5`8hZg9N(0?an7P_`Ch61ai9F_XL@U%3bE_`Lh%P5|BrX@Z)^H=y8dt9qfj;) z{Pf{}uNOwpH0m#ab%yN%2w^$|C|nWFJKe)7-wBROEV~L%h5iy zLRB$rn})=Vxxart|F(@Wzj2+n!Pu6-&t07(SN=HdeQe9Qom<=NTT4btl0T_VJ3Zc%DrU=U7)fy-ye-?(w(&e71F0el55UmT(;jRl-v9aB?0lGZLyg zGP4pAOL)!V{G0b}<5>|E73N@ESOOL6Tuv5SI^gsU>ZZdt^F{YU1+Ma$M{yp`-Nxj? zEg*1w?*3SRjcH|Ebp})Ld^=z+RFstN_4oJR|72LI>i1ny@1`ba=uL~<^x^mG7rK!H zQ9_0YfyG_RMe1#GUYV(;56UemlrNf)sWt8$AF)CRU>Y$`s^8ZRRMX5 zi(M$F&gbw<0P@OH_+6eP%qrVT1S9$XDqOT3GkPfGmr02eQLZxw_{ZPQ!hfBQ`28)W z|GYkZCz$v+tL9RiYYR(OEuOboW=5p8@vopQA}kx!JLP)6?o#1mQaW?yF_+rD)Z5y+ zeQnVyh5c`ptNLeK&9!3fpeBoP^76`gG;>WPdG+nbQRxR|^cQSVsq!43MWM1Y2NO}) z7VCtyrpjZpJL~=SBIM@cbsRExBsty|7poWsimIIAshl5PB`URe5G}doJ}h|;wm}35 z!*bWuWSsP=^;x+ZWSoNSCSGM<9^bXpUDRgj!{PLGNrvx~t*x!g#d+5S!--R5zc1vg zc(>w?D-@IB(n92Z z)I0GR>dxeTjHN-O$<=F_zX^L>zV9X$nq4Y8)P_r|oZqAI^6AUFv-d~xBbjV9zZq_9 zrPyF8u5r&mc;<|Bg;>+}j_VRx4e#^T|s`{>=KaEmJJPFi(; z9(8@mF-Wyn*vprPj)a!2^81x?m;1ulhUnyUYS`p+$;H!Yc4>3NEv`pbyGS;9ga%A0 zGwxsZpz@3*B%`FRs7n+jR+;I=X~n}g7k}y*!N3v>Y!jKgnJ6P!PnB~e{`Y<>r&t}{ z!Cd+5%quRLd(x6uuU>t7KIz;Yxw#Z-9F!~}Sk7vbb9I%aEn8@*8EMgcRAxJ*6OYJM zOKuI0Et9*SwkHHmytY{Iai#%X+7qw$E7T0ceN(X!X^W<{oU^9kSDJlqZ(R3-r!fSD zF2UEN8q&We?!BKyjVMK0#K+yWe)9CRa{cBrUB7Uy4ci#uvt}Y$$%FMhZ>2?#hO`w! z|4H>6iiCEvoYl1Rcs{=I4<{mpsI^(&R>5njk>^hleJjH=L&A}w9-I^>XGO}F${y1M zF(GUE=O-#_6QIy(k^D&fQ>VIzpTlEXKUYezE`u1)+&H}T(?jCr58?mmWs1aa|G&L% z5JnWnMvFg7s8YZ&-zLkdd4^9<@8Ir;@7;2*Gu24-o^;ruMmw1?tcI$5pW{a#JYFhS zi5ae8u&r>Sa37Z`Qe@SgEzvC*GryAGhfUz9rO)mQU-l_1l$%HC)AM|#ovijosoZ9J zpTisqPh8@qyCuiKsR@aOs4}(fyjB}G>>n+e{VRpKfk>J(C*A^5-ka+P$BLGgmdK3} zt5;elcW15JdG>Z#UT$E?9=o?n^kvl0uXi1%&bahf&%2D?1XGONFmUYM+#Mn51FJH+ zC6!$aP}@xzHGGHSv;E};h=&STpeq< zr96n2LTjqjB+rgyG|q9p(>%;Xp)J)D4Y7{wx+3W1YB7gmRq(QE#7k03ZBBi!X4BX_ ztcXF|ggxBj7f+T?ELFUq0 zsI@oX(Q|v`PVZq#a=FFM?Pj0&TKDZA$|(2U-Z9p){-}eRCc9#-IXByC^TXw5>>*#7 zA1B{Thzt#fiP2Rkv{Xs=jQgNFPM3ejW2|d9sp(0$L*AePHD0!Jj_lZy@NoY<`IEiog}=?(m?A4&S{^Fp zX{j$tp_(!6X;pq{A<(>McXXJ;{_lZ@qOM?W(9?3RYB}0N83YB$P*IjQcLwj}(a^Y0 z)2m!W9JliG8#e{i;e^Q!mI+*Vc;Pp$7EVe?gGRb=<5-qy*3)gg^A-J;yHYw_#1gD6 zJhvH}@?I!AzP{ykq~b)&*xaVF`R7d!uWAu)p(qCCKROqAEM`6>MgB>XS?$ZpON`AR zki9GGW*!l7*@a$iv4NXHdGlr3N=g(A3hQ4c0fuqtjJ1>qI?OjaZ0l%#ZTQR{PcCFE z$`1vO2HXDDmGu+~7gKW`o>;9Xps30f-vbom>w8vH4><0Xux(1y&Uz;9<=VUUc9A(T zoYEVb%y0tI`Gm`2=4<$$zdX1wU*az>O+Wqr?sbi$M8(3Ay*Pa_t1T%p@!LLy8;x4n z!bU=x9i`ZlQDt;Ff8*ASEB!#aKk4J}MuL>>oW$kq(nAc$%`iFxK?; zn|mH5$JV&pnn?<=_`EqAzCGT|Od3(o&Yh`CI-aVyNiP01?5IH<2fRG1_j@<)zQZme zTh*xRay!0+FYcJn$>Tdy7f_aa>UUH|n>a~j-yFR;w2-=fAG@EUtn2;~zMkHXf_iJa zg%mSYi}oZ0vya)FU&o~ONY$KoIx*s`xp5|eQzt5JTFqy0?wtB?&Of}!2v#3fR-Zw* zo$sRb-mH9m3|n0ocGX~tVMBr^?I>BNJc=`ITWl~-y!bE9zsstg*Ob$*&%Ei++QH4T zw1dI!%WZBZ0hCMX8a|DX-yWU_4_>_Yhw?zYFD)C-?83ZU+Fds72!-}L?V-8EoqSf2 zQF}3&nTwp!D5;+ZR%;$-`hKOFA$KO&`dDb|Yk5O1e6c?@x@~)_zo7n2hD{fN2ektgdEGi`%qCN|@)had^B_Lzw)HO>-^Z>P|vCL{YgPrV26%ZoJFl zW9CY!TvD~a(Ycbubo(utGV_>L=&D8VZ`Kc_Q2O@yD&#~B+2os2&5h!nGp`L;bVP6n zNpYp8q17Sw|H;1JIusqErJi_bW`8j;`4Xz+~Jy_PxvMu>vOnk zL?XbPQrYUZFj<+gn4P_`E)I%oieYkCz9WXzFwiV z55D^IlomTFTi)27SyngH1o@+4`y?~-ZK)><^usRhN8PQ_ z!JIkL4-*T6Oh(70c?zG??nt_+-B0$2iir@0VPzQ`F&xySLe|$;-)!7Z^Q^5kkD#Sf zg$I(}YU##2xLp(_a@5<+K~ZP1(gDZV;!BhYgNJzP>Q>h{3k*w9mw$sGG~B^F8cVOVZ}ipXiyi zt}2)aEa|>&CdFGWKAXZ5pKN^~vIG%!V`0IYZK5|LO%5<8vH5gIv^RT2CpQ<@UVQkl z{bOWUrfREs`u$Oz$Ia=o_rp(&rz)A>Kfz$K>YB&(c`BvI$4tc3?7%(#gbgODeazdI z|7wJ>d~Zui>Mwj6U3E!SxY||C<&uKWcDYQaUJ6r0j77u)F>x<1!Dqr8EA&@yjt%AM zjP&p-yslT8@ZMiA^yu4Jm9h(NFEM@9%Sx9r>rPX#I;9#ra1MaGU##nM zl%u5!#kd7-a&O2x%CW*>f?h+b@BHJJhbJS?81Gc@w}akBxP%dY!sLlo;uu=8p%?ZK=X`M-mhNDYsSu{38m>UG4M3ZNg*nI zEQZI#gk4qAj7P4o(b}LEEApf^Jh56J@_ldP*$YI$I?kuuJsa8XS-&qMmGzbC?hP>t z;z!v1PQCGHj^ZrK+n>g7Wm*$+sCetZa75}vs*;l;L$O9JWwN^Z%9GTIEv6M8*=5@b zrA394`D{Mbo+$E)Oia*;kxW}g@35PHcV9T$oCv)GJlgNya2@8zU2nryrG4Bg|4~+) zyT#7CvM$5&m4De6vRc8>Tk@JE-O9N1*;To2Ui6E+Bx6?)R=n?|nHi^vx1@tqYe9HC7n~3bK`NpfUs)_M?@;CT}ib)u^5{Z6#C6x$;s(hD~bhL`M z=P()ucHW;AB8;qG;aC=l{4~7sT|RFK$%Vxn+~1kG%J&zhZe_FQy>N6@YH>$GgP(ZH z@i(PcO8YhLp>4o_(X}kc{|J;!R4P)wj^Izx*YVYq55h+2osMl1qpWMIS{g59cFm!5 zM8>7W=ae#&_heoLF=#Fz+*iQZxHK^obeVJ;&)i}Yy4Qq@u7Cx>ePgx?06I1`o2y zEy=Z^NStin>O5G~%PJk*cu>2+BlDCsr6WErMae2QTf*RCTaFL|?dmk3pR4X}IxH(+ z?;VC(cqfMDeZY*I=!{Ek)u$KCgsne$vKuGJch|T`k`o3ukf}v!+WAk)pfOVS-XfoD zKVnIAhT#?&%`EHKc}rNrz%i~hF-n&^lJGZ23whuC-p;t|foYtJT>YY7W@n;Yo?`fp zVR~EoC@aRegN0V zN=7|2-sO&PKb~E?cJ+)X1PL2sVzw@LvLuW?GRaj>Hz1zZ(jzCXbuz8|I>Y?sE=kPr!A9ZuC>LZM|F z!vS0ie)E=X=n`Fe>lJ|S5lq@Yqq=9lD-C3F_+#976Tli8nlbyxG}vJ%@`K&2W$YUm z!zkTL8-4ouY^R@vh;6(oup|cqpDjF>iF7?%n8)~Xec6elM{hE&AlZX;Fne-mx{C9c zFVkn^T-!mg=V4b5Ld1o6d-v{D*Ur2QZi_(kTp8(%zkHrN4ap@xw*6!xS5MytyDp_j z2IviXEIii==$z`_`8O*p;cy5y^kkl64r%UY)bnR`DMwnZ(gnR>UcB|+{)E`3s{-%7 zwY3$AUic7l&GOPE-V_gzW)+Awp!ja(`Zudaq01uBaLyma(7xskZQs7B+OiciCV>XQ z#7D*p<=%)!$FW%D-jmr^Eb5!h^CnI_R;1P%%sPbx8xU1WN;kQ7-B`59;Op&&(i+I^ zDuhYji5kyqAqxpCdq!en;;LVN;n7yV{vrg>b&exq%3;#LxkscasCQ+4&cj;^ z7cLBwzW0#oqZOy@C!jwj%OmGvpqr$6ZE@btRb@|iH`+NBU8o?cD5F%MCCgX!z3rjR zj*iB4O83kxIUdXU9r0S2c0TSX+ajOgAYXVrY2r<(p>G?_TZU8&(n4bePm6RYENeLTFq z)7fNP=uC9RR8`K~!Xv9)=*j~k+)nKG>1mB^+#1i0IK`ZN`rG!v)jYMQuIF8kVxL;x zSygYXnkv72(#|(k{HZFJa!b0vP;XOS#+{-0t6f!GnHm_rHuZkg=9D{b1z}J4)zpoH zRW!48>xy+u^>Z-SCWX5?y1F;xt-ih*dkwFt_~vXUejD+s7a9Vy1eL97E2G9!i{tdG zTEvIuA^2|ST3qy zAi|_Fbf+?YG;tXZ)46;x-LgID>32NlW)XKE z9S$oO^2pI@m@b#7eCidx=d@Kbv1q$j23tq^GA1^Czp7j{RjXEw)l(u$R(xgU8T@K# zQ&Am_m#jvwe;io@nH(7zNxZ0d=l9}=>S~|*o=CC{sG9bu?Jv5Z*5O?&;;C~lom$zy z=V_@oA) zJM;~hlXC)zqsU6(4UWTSv@zI9bLYCyj#cXtyw2nH85G=m{Cxlk;STt=9TcX|#G&vQ0cM2K8)6mgAR;Ti$3Aws0*6G=!s7f27sPzpw(wwigLXs^ndc> z|N9f)eFi+N$KSM#D)&c9KfoxDXrwhPN+-~6=A&7(OniJf;|k1M%zv%DmsoXu6zu6k z=97m$3Bup|8Dc*##`H6ZPq5dA(Pd>`-;B9*VJ~|>CmOh7N2<1t{>{fJ2IyN`vV?Wt zj-Ed7>S%psIR5cs{=!?Y8k*6WW(rQIW2$9;^SuxqVzQPn?~vHPe}4n{B}rHVvl|U# zWSFq17`SRzp8N@q`28T%wdDMkun2u1ldTx_@EwaC8Ugg9H@OPfdV*v0GI&ilXWReH zkKXMez6hq-d(Xb+%oxTxRmchCsM7jT=wuPIPe)vv-S!?nLO0Ckk1FHO#|h)~Vw9`i zTFQ2Lq{XjD)CLIgPJGx^?t^z2<~+3nZCD9{-Y(Ia?5mf^0OQ@tKaQc{WkrGj!cEUNwI zvzmUU7=aNjh2xHShVLInqR;Disn}QyV;qdfgN5tO?+-tMsCHmjV-2WdPpK5lP!;@i z9Oq8CHwJPBj($yPu9 z5Yv~0;X$WZw#YgFJvs`iuoUR@lnb9V0h&}t+m%m1Kmbi|_U<^KV>00)HE)>XBKd!? zP5<{hh3=wn=u12FHz!EBt6-1Y{&$a#n=|7({`%g3@GdiVb`yn0=7&W0zV>?s4t+@NQyyoEu^`hqCce^FFs6j%~L`VA7O5udu&4lqKg#?RTYrr`YOT7yg*M&1+g3;8d-LgZiDuEaIHx?nIjlrc`cQv2>MMh1TEwo{i z8ggahj!wi7&_RGlg%2Q%_C_u4v{Dy(&pID&sKwlz7CWnF_L6T6wrDLzuoq zLVLL#>vO?bOdV)dRYT;_vFx5Al5x6D6Dgpm<{|iQugZ{vbI=d0Pd1y%Dd8~hem#f6 zop%@K>q*aRa^R`}e3>fdFltmuRq|q9bXafsz#Qpw`TEj(Xv|D5xH^e!&j&A=#VPka zanFpmvbfU#qNqx@XdWD@GxxLI{D5COvmI3L7demL{ z6icN#I+m$7m$CD^%+cd&z(h4G?D$TQxvM%l=59IuD5jV{i?6ES(1TxZ0YG24KIP)q zL(}nT`~-*tYc}0tZK~R=235(pX{FC2|f~ z5rg5nR-hX6a}i6~gcT$1F8_K_1D$hjJ=p^!On+^cD<8N=7%j(iObp+~$;o+Rv!pQf zSPc8pTalNjzE7TT0PBZ3L6KI`s;kuu&}9tatGc_7p!!dsx_e0czCWI8!MV$1k!Lx;5}E=fj| zZkqPlx&nQ_kz3>+a4-LLhlX1ic<`GCfrB<4Y%hDT=h z>$kbEa!pUlsGzuwT5q1MgeWE#MC9ukwzz!qL9~Ud-(t1A?SXmjt9({j!Idb!@8=g` zo;z}s+}YYxvuJdXpCl92g>8CbN=l9?oCtYauc7NT1bFpe?NsY)X_VXnweF97MAFfd&%W+A6k`$~tRu~u>(m)Alct{}XhClmm1EfVV``_}?L2771 zXf5MYW-=)?73vEgvl@&V2Oa`!OG{#1P4WG1YxO=d*0#L{?FiIBWS;Yn^A!83b+Sn_dP4^~S3+ z;3bfUr^~v)o|-7l^`xUB!W3?lW{Qaqyt<+KOUJb_3qTkR-#IP#5gt8<%HNHIqTnGb zVopI4JZE*9rATjf(#@r8Q7exhP;wX@jC{ew!M40NbPH*YAsaH=!amKsJ=v05M~?uGIoMrMkY zgF_)6zy85a47HWVimy#y)n^!2L>#aA-EQ}?j^^7*8K3~h${}p zzKwNhG!AT173G6WRSeRjTB5dt(pG{>rGR{GG_%KtGBKw$+Pq~6`_AT zny+|sPdMh`+FpKrhW$;p>Epw@BjGED@HYRG8W>^N@-XLS`8#P{O9 zm`Iff8YW_tMX5#KzjyC31|}H81Z85Jhfh$@c9+o`QP6ss7!SGoYX4aX=Ub6JD@+!l zf?U+3E;W|uomh>8LnI@(bJs3l3sHv_avZ;+jhaVf!!MPV&|ZkWD)2lU{_&O4y(X2? zh$3ncryvj1B-;BZ`S#IJYzjYOE7{QIX##kLyC0j?;XaXy(FsTFdfW$y681h1^I}_X zU!r_IC7m;cHjU_&yg>%V5_1K#SPPrJut%toknC*ss1~QGv0$)&vYNZa9uF4FgvKYv zK(^_3v8Y=bkP_CVomjT2``c6oVFRx;C-oc=t{2n_E{#y+4nsZ%e=u9(2!ZEnCL3*X zk>TQI>k~7PMMzJi99!c`+pf*9T8F=0^k0xe-Kpou6=>I21aqejed$nddSQL5r9e8L zj6mLaqWZFHi9``$kuavkf(&`pOgt;&Qk>qruE$uMHIYV2xXP2AZ_uA5($3}&tq;#&<~jqC-MPGCG9L>f5?=Mvels#1^W@VM-LS=>S!q)|+2 z;)1|Rda<*D0wqO8hD6s;BTAr&`^gR69-375*F%Tx=sON_5I6W-X5u}9ryd+)k z-0}b5%Co6eYG=k;g)FkBCRIGW7uG|oDEjehC&R$dV5iEFE&Zx}aX-@UNbiu12I`n? z-I<+nk)MC9G|3;A-eS{G3XzE>`;*=YHCT-l^Ao(H1^$XMoLmIj-*n4Uvu=TIOU-NW zEpPl42Oh;WP=hP`#Jp|@I#%xKIkFaROjv>ITDxwo^S1ZxDi$#c*RqLNS2j)#?bSxc z3(Lgjz-!_=OM-9mt4PWbNk&_`g~IxvERtrxXTPP5;~s&K|w$q#9BEhMbo zS+Ewre*H`*$MzG_Lpe3ZMT%^gbbGx!Z)#G<$rjw#pWsD(&Dj7rwO%Ho_YyEer=>g6 zuP+mmK2oM61$Os>rNZqSFR>Oul;|B5MLM_(=?3-IT`1@zLB{K5&1d~W59*o1R!fHW zRf95Hz;Cec&Q;*M`IE!IfIS+3feR*;D*Nu;XYXngO+}F;3^IgYm}Ye|X4v z?r&x0<>r>A8;1B^8he4`YT@$WiGD(L;mn-*O9w|^La1>XnKr!_z#hMMm!6Pn34FBX zCbFfj#;;~qcU^6F;7!72*~2lO5B|W$sfBmZNxwxh8}h+opQE+zN#`Ts(uJBs8j=ka z(OTCeo{wcguUfP*zcras%v;{exf<4^CFS_JhZZ+Q;iBjSws?%3J9n-Q_q{jg)<<>& znad@|#a)_o97hc5poemX>%b1&`b_y1H z)^ahqd_=`5XMbJKsYl3-l#DWZ%@5yNnu2}$8A`9+y(P0)$mqj!+rjoTx(;-;sJ(OO zivi&&e+dt)!zveHPCCQhcs^^w`~=jK0< znUzg`0+PHk?HKo}ox73{gxNv2$DnrF5N2Vse4g}u2rI&q#5nD&2hID47=u>wAx2dz zaQ*SP_NW{43LB0lEW(fuZ|}DgU7ZhqJuIUDcBOd7-R2GTI`(}>dQ-e;$GDgh;EKPt zw6x@ddYgSsqW;T9MRZp-H*NX-0J@JGOv8h^YO}88mU8Y*IqDG?7Kaw&O9#& z_BHWr{%7wBcZf)s8u|0MI$Tg^00=ms+Ux#!`?X1`N`1o;R^ge>0UbRH2ih`7%GEZm|E69TK93FUH*Yk0vy#{K7rS4Z`vK0O|5{m z;sy*NZ6{r89bE?}LYj+1CcL4ku5I9IT0u8vFvX=eEz|~6*4uaQ=#O-B?K7Sp3wQ!8 zoiH!;7ps#84_*Vy`8)@(ZebQuE!=G(lS=NOi!T>=g&7I)7}s_Cg_cnf;t_)h$M&^d zL)$y#?RY+C*`Q7x#I*H&P&8gJ#s~lz0dbQw%$75OM$pW>90>(VycCk6 z)nNlx$HGAm@yj5nu4?9gq*6PJ8MVVDge!FY{{d3+O4uxIACR71$B)WV2J8sh)0<96IaiP`or~ry7!3 z0iV8ANBI%rS)Du=6b=$nV>A%(@bGx8aysQaO8m;NaNnz-I0cF<&Kk1pw+sU(zmben zRT6Z0yeO6wD#ssF?`7C9!rY{idM_;w;tpc8p7P%ptmJy5wUs8&GMPHP z_=gW!iGxN08n`5?s;XhogmbSS)V|u{L;OZAmyhh5Wlxl_FP>*=eOS?ynA=eZHJ4em zI{dW3_i7IUklH$>jbKo?<-Q6@5Ot3hdKH??g3nrkR{qB~7Oj=_LRc1#&iF$x8utg) zC*WH(=*tMVJ-l=ovmIn=_b}Sv2-I+T9^k06R&if-elCp#J^+(GTnI$Wb@gz>NrL>_ zJ`Z6lS{QMjU)@^fm)xyUQIZUwk*t+1N<~&!jVPz7Nw3r(snR!Z$}l|aGE(!VP?~OL zAxF=Wh7mB$t~PV?mlgUkf!16n{RC-X?Twcj*^d3?hCb35ihb5z#^d;9706ap*Rcr&l_+@$pq=UmYNU7_vgu5UF5PBQ!`Mo@*N%!Kv{!@!hAN zQ)Hav6)ZEhs$H_QBsdQmT;*y+7ud#wSVtQ2V<%v!;ASe1OgQ2X$>{ZF4|sqMF9a^n zF@Q75aO?Q<8smYyI{8?UoBvubC<|2}yl>Al6rG5Ni$oO4!(zdufv4Bittr^RD~5!h zEkYe3^M(VxRyI!p11rVOy=M&~wFm%yQy?me44NRx1t8HNi(~~7t|)DgjT6?r)id2G ztCZ1Ma~v*r4ap)AhnxhU_F};Z9YeAAbV3 z$@U_Sb(DUQ2f{ZOuB$GB30C(~7xQBy6=Qhg6Ig=^y900~cCKWqFT#bBycdc11R`5- z!0@Gy5B+3jkZLvpszxV}BxB8Ol$ND+yB-ZS`S|Sdd*Mz#Hpk(wnecJxD6~ZaN`Ve! zV$n(2izK2+Y>Kd3CBs#YI{?tMXALbDFfuYSYVp1mg12Zmmb4F#4=*uWVzNH%JxQH- zr54zbtA}IH<>PQeoAG1`!WpkHHtUZE0ol23So$aym3Ad$tCYb_SoP${q9Di74(N$Z zXOcHu>b>$1=?x@2NCs1RM18?!f#=JA{RM8{Mx=dtmj8NX>rOC##Q5ph%0Q=7wwHxg z0sRzazR1LnjNQpvIhJ5!#n&x+-K#R$Gp|9O7Xhkqiv!A6{)+ z5l11jbt_Z7G>oJA+^N_}i{2}()k)=Bde*ERvxk$p+wz+<(I{|#%a96cm2VB&WA=cQ zAiuu4rIJS52Kxc13P}aLL+RaX5Y<;B80%V~hM7GCH3(N0tn@W80oQd6Lvk_FkwvtVN5|cs_5i!Gn#~U?)rSS zEr?bYdP`w{%c|Lj^RFVGPf&9*{TJ2=_=r3E-AjmFB=*I54%ZuzpTZ@H@>|2D8-<%yi7ezQ^VQ0mfue)?_h?30gd2SuwbdtN`e|=T zt>lgOR=ZwPfZAZcpJIKDQ2`)6uH)2%O)q_X+&%}|QW&zhTfcB|H{{PPzKS#Q4n0zz z?poS5&~!;1{u-~tje2}{n(N`hg}5!p{6nyhWxUo1=)@pH9ZduV(Ik_mb){qe7yr^bw*S6T;R817 z)AvClmu!-4Ztd}-k|b^* z1Xdy;@D7B)TafD{>CiBB`wtjIx%}q;}q>*AV{ad*|V=LSjnzsL9zzd5fk`j`O;g+g_vJmsiv5Bzj6w$J?b32 zG_x|J%Jp4iZQ@gr@N;Pd5j_XtJs(T+5K4Safb8@0og>=!01kDo^} z6nQ@y=7rr&j@7m5RI-5QQx1z4PMQ@eNPzQMLf6pnDQ*XmlldF2NZ`;JHZ~lU0wml* zyvmJ=GIDe#>n&UM5OkrJARNMs|@ zmhz7doJ6CQE$~Woq;Pq`M?vz3X%{|cw~#C4KSd>|8>EfS*~e~>QH$vYOJv&#_*w(obz9cf*UXb!^Dj#NVxQ5OKkbik=S0XL5LJS6iAM#+0UISifN zOSj+9mJzoN>3LF^lB-QE#gyk5aDIiq)K|aoGFW8mD|znC?++Oux%H2|WQ5&<-!@0$ z7FPXC{S7hM=7xk5UejIk zP&Bu*qOYx2EmTPVCF-sGlUk>#iOpu~H(DRRFGI*eZZHj#U=fXiAaRwQGjH(+#7jqJ z>xzCGav(6SOC-6As%6)b{*0+91kTl{gm9UONcS}4WW5&~JCBUdCE}(JB0?v-QKzw+ z#Ns5qO8mCnDJQ9$>qo1=Z!S|;yH)1FfvbraalS&vX-P6j(GF)uH z|IupVgCVkeu~B;+$nJtl(sc)7KX1HW&K(^I^BEUZdwdkd$GUcjVpzM^+q~I+yhBUv zi8*@@!rkt~VtyUvMpA)9qCm|gh?)iMfxEh^w~u%XBz6@Gjz3g1w`V8F<$MQgstP~5 zzNQ!m?n3756vk}&Ht!q7-p^U|zgW8NK&sdO-`?q#wCPqvQbI)(x=kZQNLCRI*_p>^ zmqbM>A&Q2bz0RRwR7A-hM^PCE2jv_Qey?}m-~FfCz~?-l=XsCUct5MMy={zUdWlDR zrMeAppq~Gfe(E!`Fdw7so}J0NrAR#(@X}t>+a3Ur+4^F@TtKKA)1sJ-yH=`+pBnzv zsR-x@aK-vsUf(8K71@|vCOB1VDyZ+2|mja%nUgZ6)k*S_}qwiK%RPq@`R4xa(Z zn9S4$`0sLiSk9m*qDp0^UkmhjW(y>A=A*7#2smIem{D8oyw+Y%CMqtWg$ob4_hz}QINl%;C^Ce5zZxn(Mehp>t%uFC zwc!o*%IoUa#;}3Vp9pL_j03vmOx?$azGEJ8axflQTk|S4q>T`;S&GEbHoU$(8_HC> ztRz5spCCKgoNNZ{0HZwtuBsTPpF9{~xA^o`=noI06k%l|tu7EiAiYGNjiPP-KNvwIM9hq$feXA`4Gy-% z0|H~%AT6C-$#Lc!0GMcsV#}f{M2Q0OH`07m4v*Q$U*o-BM<9$yn{&*Hxy^`1yG<(3@%_7|}PPh)Y z-G4Y^JZGixN>I@ji0h3T?{X2O4&2}L3jP)wkVO$xLbE+Vuwz@|w9jjoTAYq!%0{MU~CadNlYM%Vw>zLaqPxTneo3AElS+? z1uPSi_mz>CuO&1X@Tc=?|GDIO@7gY}V4Usn2>O6^8?PxL7%;y}I?Gm@wj*=&0K@dt zkZ=fII6*G$=%$lE%`*r%uH??W`X&>)O(nD}Ohu)mFV7V&UNFdplHrTkkDjy&Wgf~c zyDUwFJQN{EO&UFEobu82I>vv%$p*yF!^|H)hBstKY#WNUiN;ywj1N8L&AV0P(jfI} zP=Vqo+hyzNs<#NVTY*r_d0n-pH{S>7BU5m0ub(?U+zq+%7map#AD}1Lme=vvKLHc| z9S5I+{LMtJTTp9s=+JR{X8#o;`5oWQWO)vzqLW@Jf@rPyz$5mG|T z47(k7fhcY3>cR22*YP-dk<(4N+GKNk6b0cs@GkU^RA8_}Yr`{?d23Ve@7J?i{)=?Q z;%gD^Eb%=Y`|rlPPl_CQSm#R@+h%!Li9d<9bLM<53zbU*^<}o5j${~G-_X#Avz2OQ z##D&hJ6xeI>pSL!=XujQ{4T9ralKL4g(KI1?S4nNO~qJh&(S6UrS7A_;@TVgFZ?Go zh-Qs0!DXs-pk;Q&#jS@GR#=xk7J-b+)(J`|5AQbc61rVZzzkf$S~L26AjL4`W4{Y( zDZDbyeB8gH>u%-RlY&2^*P1|EZ^mfjk8D`Kz7*J>tvUEI6K_vE8f*6BoNtKM-i6RbQB$;`krLe&WN>|0xr z0geMqP7e`F&N zfo)hwX8ZZwzrkrSm9_(BVOE`Qtz0lvY?4f5IEz?4!=iO#=%MtWHe8>RID`ppZ-Ki}T z^7cXivptDvD9I2QQVZ=SoEFOxScpuvO&Qhr6DSfHwx5Ca1-)Ip`okz1GqX7@!0HjD zz5Pm&+(*zPwc_)QeBLFntiJ#!kUU9JJ1V*0%CYB}Xr)K&%Z01-pB9PvEzZ6D%>05oNau z`>Y@3J45grkoV&N1lA8WRYQ)}1C`3=f)?&zEK(}!`DPS}*48Ep8|H$UiTbZ+Ay~3k zw%P!~drenta}C^sc32qK8MtEPNh27rUNQ9|tskT4&6$Hj_R%zDGYoOAX-_@2E~ex@ zEzS~(g`{$~#%|4DaD}=%OW`k2+i#vIw#yrm+}(Jzzbq_;*BV5UY(+P0WJa>#kMgvX zlC;qg{>}IT&;;7<^#==nPnmG)F-kiU>r|t|LFPj>zds1FNCA>J27Tj?Kw~7Ct}GV} zaUEg|AcOVX-;kMz<@Y_W#>NA@@c7e7zvStYCenb3Q_WCka+7Cn#@6gvY#IYdV?FXfyq;Y|XI)A(Q77OiLPLoH*nw%lryJ5^SKr=mQqYxO zQHBL@LGeii0yr--7FDx0u9I(XMsnTg=9@bhji9U85p@ykfUcc%J0t+bSqiFXU%#by zyiFpQm z4LH_wv{OMb*t@6UiVVuU`3T&_Gp&ftkhzix!_3?H9uaxyP{m_gUH<7TDysqx^p~|` z7=X0`ZSuGGqgMHp@#x#PZ*>$_Bz;sQ9|#(a(_9wc04c7xxjzfy!XrDP!SK1a0CkPc`d{RRW zBV=YV9=fm0*h9h?EC|S~a#fP61G{iu^3E7@LdTq9c1b=%ib&lXs-a3Y>-#j11 zQyh4^1|{DnIK$L*V@E;&#uo~h;L*xrc;&_v8*H6f}4KQ5HPrfa>|*g%_<4%ctWAG+LlB6lzu(X)w<(6fhR zn&?k;dn)#I>1U~|X75{#ZaPk~;R;C`UY5b*6i>QHlz%jyNt6&0Pq3R}r}C83rR-Q| zTJl+>c_Q)`kQ%63$a$XvD7`Zjpvu}aAaX7pO;sL=TJ>;3^>uW1h791fpH877~tKOoVVQ}3+{Q8V?JJq-wVnfU0f=xdJ2>UKYV1V)qzR4PAMLj zTj@r)xB3C$HF`4XK$SVRJ8#SkCnP3FoBablIth(X^h&H7Q?FC{BqA+ftQvR)aCx!lTu!2|4vK*+hlJKby>bB1;*l|QE6v5_fy+q6I!&*7 z9$_)PRf5|QVPb{Nr@U3kUo`8#kEQ8m!Mc(DA?45CA`WiyRRL>LKp z5|byvMvx;_0!{(HVX!p$-{>lCO=f*gFUzyv^@{@7T0vzDs6I`O7%IJU0;_ z&^GZw&bqAf6<&R>%*sgBiKXv?Xrl(vN2_z2!bl-?(4{^_u$x5a)P<{se`;Mvs%6P_j@1I9~R5fg5tkCC8K?uXNreMp*B+w0_dDXU>PE3i(iW9k%%T@Z*_of;y z6tzn9c?qvyzh0od`O=$Xq@MM}bwRD{$lJG9l{lUSi+K`#K#-(c>*S0^v|NY3S^@l^ zY=nKNj_A*OSOB^0Ae!cQ2p}9UDhh3T^qu$ao~dgn)lC`xxHiH1-)#ZeE|1(JBy3mi zi#|v#2o1*{1nj^eoCIAPg+On-Z5neRdg^1;PWH@{)NGf^#4NjQP4T6VdP(_;fA4}; zr2`dq6*@$M21&2H?x$V7U5hJZ37rs%-WBD(Ua2zrD_BdhQtc{5dz0f02B$8xlL;J zh?{s-A?c$syCMI5elVI$Alw`xoX80XxFa2A8E{gUrsR9tp>M2S_bmJ|x*iWVH3^XM zcv0W^Bh+*g+N)c%HY&RxEJF{g9Z8ymyy0_IEA_VsIM3%q%c?IETyGm*Rz{`_!tSuS z2O2g3SO!|93v~9@p&XeF{IqhFaj(B=ck~F$(?x&l^o)-XCp?1?O)Y4xR(OZ-ytT9X zm1BnbH}Sde3mbYb|BH6e0iZ33_g$_cprI|zg?W_K9=LsAKfyllk&!X{TK*9q*Va`| zp11u=Vx*sg%V+w&SbcqD8EA)4-lanebkV5=mEZg4n%mdf+UOCmN0bIr0Q9 z-BYNX9Q*YiH{4eJ_aDGJ4QkM-`i}ivsa6R(=@D8+l~vzWKNth`CKDx*(*(#9*G?o* z&{$XFd;bhNuyKD6?DM!xIQpUiWC$;cGy6@U_X%=TQ@o;WO)N3e%* z$tMtUYLm>DbppQ-*xsk=jSf-h2)k$U^f^nSh+GM(tr#m9=Z^#DTpJpSu_=PkmpuCK z9iz9f54GJ}$kTdS(N(?s*GU+2(xsXm-pr*>E3o624yn&_zg(ET*{#=vTYmk?+@X+P zUht&=>KX*e4+0TUk1LS*J!4NV2NKE<_z>KxJ>4uO7-}UN;OijUcxzNhp}K#2>m_JmJ2?Jv-@0fehQw^q#;>RS$G8#7v>&@-WPn zitmM6oor3`V`wC%lP*ndo7qk$psOr&4~i)J27y+Rf(IwJ`tOcmGcdLBZ1PqSq)!B) zF5sw*y)k23YE1<1Wy?bk14M^03@E&JLmT_NJtS*plUDX09&PRABDAQ1IScw_==AKD zp(Z7!{c&|6Va@z(!~JuoU4%*^-Pq-jOY5e^Daw*-mdwEq-qibg$&(&E`5JpD=pVYb z&g=pF0~!{m>=sy|UGsLtP$1GRDoI~a(m(tbWvD0DD9g_DwKVU_11R)LIw7VA2E&Vv zk;DlAz71~g7d;m2?#Jh3VOw>OE)89J62ZWPR*|$D`E7HriKT~3)`(|!%oaEF#mesj zep`wU2tXlk&l&@1hwBr=SjL&f~hRWj&E-^<$XK?P^c7UyJ?N zXQjVZ#5u?1DZnN|@Tp@LhTMjvYs!E6c5xrCB!MjHt!9P1;quAq&W@-NjfF)L!njrw zD`6fAvalQb6Z3U>T7+S$M%_7S*+ z!;Okx<=(Ze)Y!_VIdL9Y zr~^G=5Bo)Jde_fpPZp|W>VwQo_|RN3+(stHeVFj?0Z^1-@DUI)j0PBuBtT635A+y~ zKp8iz5C{u3f)JTu-`$2T2oNNc=Tg9ThzeiUEKtRSuB9e^ePCN1)Og20;i`==2^tzn zfHNOoyLnlFu{Em1tzJWY*-oG^%NVSof&KFkFH#X<8z?S@DZ0dwIy@80nr(GG%UlfJ z9RnSmqXd@!4_8I%%*p0o{clws>=>95aoQ5Ks%(iPNr|DO0xS$s{Mpx8 zOSi1gWHq9x=z*vrv1-KJvs26lGGdcvhrhaZ+R~Zukjcp<7i`c*?&v{%@`FDD`8)}Y zLq8Q)Z~5v_hXKiYN@$}tV3JUhsoyBX+e*t(@WMOv8lsp`{e%M}s3%ySK%S}R{hlxG z%0eJ`%pW5KJ~BsKe#(A79f{)@k(3HEsryJJm*0rvVS}6A((&ik-?M~!IfSo4y5#Nx zhaM~*8I*+nvvT6pp`lE~hBRXgkuS!?tDjWyCfbRyw|dq*Fib@}7)^ zLPGTf?gFNEYx3C+7jQQ)o-CBn2B%n5XP}#U?azT+zmsL4o1o(GMU^`U1!V)_XthoK z!6-u;pm>STd$<2Q!AVdA$Nd?rtLsGE{@Jw`x_t|<7~KdWQvEwpEJoU1U7XIaJcDX8 z(dqXsMZfEBNQr-#&MKk)u2l#3)kAu3kDfh?ZjN$A!PY>^pMcuChgzB(uc6#)q7&Ft zyn%Ed_Vw3ST+J>0)m<-V*CZ8cS-fA_&4fTe6e`T&PdI|W$PZfT7LE*7AmpE-#+r-` zVuAipHvNToOG`_Cvh{8cXC8mIpb_~WFVqI)uKYvehN1lR3;YhlvkCV09dzd=O)A)x zg}xK_=eD-XHvilT5(1su8MM?^`H}Zm3LT7aPY?wwk7Y=Jo|?8t`_KBK#pY+MzenzK zGWUhlk~UuaGrZln;c2U_KR76k}^_b)vAb^{Q*SuKs8c95=ka$ zK_1}Vf3t;~g&2xp(ALT}6o#+BPpJectEuPiI(xRZOX;}>^JiV|#?MVhY4b7O*lq=e z8*~;m;&}^vMu{3?%UF9rp7(s~4fmB#)U20V7cWtXD;I9RCM=)W;q?a9aNQU$tH7gV z621m)QP0ybJ4rXMn9V8e zcW>W52Bba(fOqlSTZg_C){)jE(FQa6*gwA?_|c3ea|d)cW|PdRX}Coa9C6-k)qr(u zb_6CZv3Hja6xL)dm9s94*pEg^j2C8=q2K$8hXa;?W|sYh27l=lCGLv5^xrB5;@%mc z*Mrk+EvN5m@@(yevb8(vqChHA3vnviW|yyWwzu0pha%Ps%e0|GH}pp-kw0f9Rc>0NTFDmi5fo3II!(PuJpoL z^=R0DUbz0$x5D7S}Y`F1T>A=Ru8ywgUs@PI}*wp;WJq-e{9|D(S`j!qYDJC+QP8`|3h;2 zyVD@?&6a=TIfEWU{dOFiUE6Bv+!Ni)Rx$$DtEjZV;nsaX64ZJc zt2X=Gk>-QUxk?CZXFv+_iqoOp?*hk(%*NQ-r5V%r8dVbC@h0Y+p-{i35le`{SFrIs7lsnnZekzDx# zjrn0DLVnJiFV)g^uN{fQ64;-bmX?)w9rM#`(DvqG{aEp3cMMiGwTo-s|56kQi3>v>N1@-%_+^#IgK1*`O=sp2=% zN?UKoy|6G6tjMLMP7g*x@Zf;lmESLil9=nFWl$vy{SRNO3z#}2J&`mhkIkv}3m|eH zLGWkSU8SgSd=qycnSqs9f> z_Um`glq>NoI6TMkaP#dlg|a=cp&^_wFi)X%ZGW-WS`_Fn0KrR}HBjM$R#=j+P|13b zt`sWoDqOO{`3B9duvuRkcY-ER%3FW^`MVoFZnZ!6&MiCWWpW0JC15_!JY*ig!-&oy z_VwbNxEa{7kHA9o7L5BeqLR-Ho6sKElQ-U^4CT`s(kMW~q?W^R7;lL} z?-^_$hSmlH+hXbzA*Z=n9AWrH>pz#9jAQXY#n&W#fTJ~(diKkbl+ePiuqMv4Fj zM~wGo7yU-1K`DJ>ZpZ2{S*z>Bz-$3uPn|4p! zgz?j=I|i^QMu@dcwd@qbUQ*EmaxACt2(%^RqJW}A?>7pv-5|@7Yc7db*~O|i$&GMa zTTpqUI6ei06$Z7XFXK<3YdsJuTNGBt8AE_#?{w@IvTG`Ue&ZPN&+;kEgvm8iXD@P} zybV&LOjP7`=p*&Mp3kS#3wJYF5<6ppneDOuE*h&F^Xj}+09o2c*bO|yu7GU)6+j`w zR)z(q4vbJ+keL=-j*=lWV}GJG^>k~35+xnVuL=@(&{NHs2Cahb%kKKLL@?@XeTJna z6~cEsrdC5+0TqiYT42Vz4}obLescl0tsnHJkzl}ROKp@eoU&K!s_aL@tZX$glrDgt zY%wB%ZR1yWk80*uQy05Oeie|SEr2|563jMzq%(XYWU&{69iQIBTS7hsiNRg51e536 za0oP;j=!gjK)iB{uN@*@hJ6d$W?u6_8HhHZ%Pf)8HGyLKO@g?BgV*7r?@NW%EV(O0hA z0)Y4lQ24M@sjJCidzO}Utl>O3UUIB(crZ-KFKZ~9W?P&=H=Kd^6yn}~5AAFn3Ib~c zk`p-`U*gdnucAduE(gx}hGx-_`&7ew9Ph!F>Yavy0)%zHu~|C1uHW=G1?vfs5B7D@ zfko2F^_jWX_@B-D)g0`WJCiXNef)94JP3B+0sU&kSc{76Z^5en9>whDOdW|sk4C+L zt`%im3<>!Cp$q_I*XRM^cqR2^jnWblo8-~}s8!jVXv+LSLeY<&e!q|1{o^iG0|;?} z{9;Q&Mr~4D*1BH!0lVct@*V?ZW(1EEy0qu8>3n#F5d(M|CoG*YB>@@36##A>sXfRK zjWMHzQ~&`0?`x3dHw}TeF#a*@_nVXDpl}A&)SyKQo|6J3(DFA>9dKxlXcBA<9YQRMPey6K-jappr+tz6g`b)bVTIfuA4kiJ`ur8iCGQ1WQ8qN}{u@YI0 zIvb+1Qfti^AIMFH=-K|a=c8zrCrGj?M8zIJEgOG2=Z}vuxEo{8prnI&O0Asi4MA2w z-mFb7DRydmAET1$MmN7mKD(Xw1XPkM@>wHzGK~adngT#s+0iY}6l(9#gC$)-1*1Lo zY#$W}ce+vPtaqtL{;dZI`%O#+%#+yTopAG{>A?WC!8_5&Z$ZKWu>G9>-%qHYCwvGPM^bPUeIlw13 zh>MHcsaA|q>rsaWwi`{?3zAkXws5%D11yH5Hd|a*%@8dIx7ih8a+CG=XsX~4LOuOh zLj1jd;Cr5+Q!h?*Th=_3pSI|6*fv78Z{8dC0*&Fdj@l%K&rt-W3MZ ztBz`!x zvI^1c+-PRXC~8lt$e8(6$PlPe*E4-!rAd6+DwFjghv0bsd$`gaZc}nmi{+2wbZ5fz z*J7NBu*U<1bk9RzVnzkPi4#DK2ZrAx0B!DlyeS11C|lSe}hMd$2fqRW^ z^MBT=p!vr85WIIw#{lDi7@(ZmhO+{Nl^W15$LR+GDyx8xjP)~uNIuj;O1PvNQSK#{ zI{~V6v&KqPtzjsQ`MHhaIU{G`$lOX|zr&~B-&tqnGY!qjAw8UMpHTxr+rv^2y&WVo z3B5!x2;Rfxk6@Xq_+LQo`jMquR7>E%b97jCYSz#6*j1f6qd*Gt>6d1y5B< zlsw1;4n#)CEj4{erJ%(gZj(%ulx2CRS5<23FECoLBij+BU!ky0dwV-$rZ>v%dB|lo z&}48g)lr^-s@}q{Gzq&DL2Q}*@4Cs~!O@0obG8n3iE)o`o45IXS=d;R$JI}DxtpAS zXmVhObF+9Iq1vqwNO_ZRQg9@--8g2+G=qSSvN4K~EY0@v2wTDnA{un$H6xXGBpnC^ z{Kr4i@AElP8tIycLd3%@XJBNMt*QxaC}rEmxy-^O8x)6)bBBOu9KjMKLamVr2B5I{ z!>5Sdn@}J$Li)R)V6JE>tTIe*56m557&>ocuN3lSVk?edGjxLqPub(Ys2)awR>#2Ni95nOYZ$z5O(Wf*_3<)e(+l9Aa^|? z-5|R-0&iLbz}6LuW6smf-UnnMZQw9#)s^z4IN6`Q6NhAhlMM_j^JiVw4;Vb?Ru(2- zA)xSOg7m}d2UN9=PclCf_7iPHBN;9un%r>luhK-;kPPuDg98l- z;|WBjU%a0%w`4X8IW&*Ozj*OcggTqb@7k3g~+_pfad2(}iceiY^ z(^>du85fm%qh?R`2GQoncygdns`K1tUJmrRZvnM#w|@+Y6VKH!&5w2D=}0n=NLM7( z;9t7ai3w%r?;!DP`+8aT`2l--@ZI@rZjC=dx!4Ao#_Yta5a~hIQkm_kzdcrx7Z*o8 z1vQhYJmZk3VD2fi=@UJh4#jVo1Mapxd$$-9DTJ=9=Jk+n$zh~Vdz=1c)gHs z1!Ne@YF94cJflJor1CXRBpo@@gv-C$a2pMDis_~$e_s(j3nZ5Y$vpimOgaI zVEPkDI@KyDXiq}|-p~7q?0W+Aavb0DjLKrhswMm-RV?FD7&I)Q-n7h#Y_ItYnRXuV z16JYEEG?mx&m?@rKVvI|%lIokXFAGc`C^n!UvGj4{*|oNnn=gKp+nrh} zwQ!bxw&|)L$-cTmH%W287G7i_I0Tx*?cjsfwq!b!Py@?^W(h1)@$qWP_rlp{)xPgq zFO39{Y{!Gq@4phS!asu*P~k8~bFiSrA2K93+rNXX`C5IRvNO~wHy7NCKe=)kmb9V> zJqPXtHKsYAMn2cTB4giD%x=z3SNxH==PzRN5{3<(*3=ZvzwBh+)4Aw?@k{DlIjW8^ zB26M$i`){_Nf@EpZaa#9t$?lEzGZPsd&(;|c7!#M(lCtsRqR!_ zxBGH8BEpso4;su2_fyz2%xu5c1Y-d}MT~6kA$YNgk;t49d*Q-`?8!^`6V_l>6Q+}Y z%~li~$p_Zq_l`NZ|8ei8BbClpbAKh8u^KS;NFVjlDD#^x|FkIzVzUuW%_#V^CNoCQ zmLRjPD5{7c>0v6mV&TJ#iY%VjDW0X4bg(dSC5e5J$X}PKbG7X;@|?&hRxUh*twDPr zZSIDgFAZt|v2MHI=N`a!*hfR#6kMUvdSK)wd_8duNZvf+HXJ7x6a{lZ7rXQ})hpw5 z^-OEJUfY6>wjKhyW8?%PzbIm|9AQ#4j)$nEI>-dx2mIDyVcADDc1%+M=EXX8VjR)%8nRW3L@X%@W-SHzdT6~qo!PA zV5>wN%2*b=3zldM-@{ucNFTP5n}UNbI#wk`r1DShaT0$?Mj%BaJk9Q2JS#a z;+C}cS(Z3YyCmCEXk-}2q0NhLam%iIB}0IOwqIB13V(kX#{2E%*rxyez>=H7{3V-? zs`_yqSC%Mimzn$C5ZR$E`%9;4(8JVU$9;=k_W z&aI=lJ@7GIk3T`WpU@DNbaQRf>|{8qUP3AA!X3nZ?m@giAfF`8*7l9H!!RJu)dzNO zuH!_052YqF;mcY-x^1Q=PCvfKA`S?1QKkn~Qm{HUfTU_O?Rd-T+|%8>&oWqz0niM$ zoh(^AsH#g>5iM%DE!Vkqlm$#2@}V_cQ6?ZTIB7=E*_CQGk#oh(YK_}J^8d`Z4;l;T z1mWFg$hDA2w~TLe+)+d$gUpPY)F$FUt|U)D)6M}z9((H$8-I&rgN~Xv_$vDeMTxa3 z7d>c@t1twBY|ClGmBEaw2eV88!Qa3E*GOzjD&xw<#)gf>l#W52tt33$NDH8$MV`Pw!aMB8&>V z7`W+ z4v&7~hF7?J(A*+tn#Rj+v2(U5cZvA!{b8`no89hR&Kb?bM6|0N!K^)yxWF}N2>n@D-2E)R<)Uz+>UP#^b-xDXg3l@e^|bdsuFd zT{9SFBL(9lD%`L0OB~Oj`4eyExsSBGIChc1YeGbU-DZ%cm3ZTN^c1Hse&TI;$u8wD zJb$aTokanDF)^YxNRA&}QP+N37jlv3^VJ1L&?qB5^ch_rriFxZW!J5Hhgpp8qnz6@ zS4GyPJKff+;m_p?kgVZP3U_3QG_n|3@*f{G7SQnH6L~(5KA2~JzskGM=pnXzPO=!r zl$=0HOaoR{$v>hBQPQfDcSE3id;iC0d}e_*U-2iB&M%SYL^^%wKD>j*JJe&qrY&fx znd+Ido5XF$UfX|qyvG32nPm17+zr5IzbjWmtVu(YgKR*NiGg90v3t=zU7j5X;@#llH%H&^iQv$eK<&Iyx>|zT5_!^-QBlk)idFq}h?L3=HP>HzS69*ere2o2HV)~Cq;#Jfrxh1tATVoaX83t75F3Fb^3neGio>&> zIhSuM$$K$p_Utsk0iMEZPf43izDETe6dn#b-ir@s<;*_`4eg9dKE zi?xM!HWU6a#iGYJji4P{3nrWK!8nj9InB9iy@*^Ry0E=5H} zMiv&2g_XS&nrG|}*Pe|r752EnDHCzmbMWH<`R3z31ueR|x)&@~_sqBqk~|%ANgA-+ zwoq(=%g`W4BWsEC%KwrNr#z-~y)YHWKawsW_$uea`#rTvwD( zH%s_zsMXj=IIIkdF*|h1;2UcL>%I3~4<-vq5k&qmM_MjT|H;#*M^2xPBUQ|zMT>r9 zq@0AxnMoH_m2HEAMUQW}02l*?ZY5`NZGjV~iez!#Uf(6q#liWvR# zcPBzcY?aI$xK8vDO5L)_mbmV#nzf>5RrG9hQ$CHzAw505(Co7rCifTm z(9U#o#xTEnavEEQj`x0E+n3|`Ezrpb3naa+I-sEKgURLBPf-5te8<54OQ$vxP;FV6 ztk=sCdVdfdF&5|b25fdaq#^&Vs;AEe{tu!7cSawvE(}N$bm{%nZ}@_m?VS;?FunTi zfbt~E6#JHZvl4T6_v{c!{r8}y+$ujkiEWuB&VS_~uU6$aChx2tMQ-Nh)xa$CFw?vn z2dW{|nL%FaI>)!*}sFuWnzkd^tyK?IEObK=dO~d}pQe^DgW8 zoVxm{M%s&bs;Hl40Blq5-1!1N0SMosrAtG<9z9wAy*M%~>iUIFL{wCF>!tI_t{_!Z zi{)f-resA{M+AuFjT&M@MItBCI8UYmDb{nRSID-(usW;DHv%MQ=zPIVRLQ|AjEq)# zR(O&7+>huVqMUmml8Fp1+B5Nvd4px>jx48nv0uo#wPE$@*EW+OR4<|aG1x8(3h_J$ zZOa@3=N=}(P}F@F?&ayqmMMS^f_v|6_5GT{;M>s;36VBuVzf%NueZ6DewP|?h_WZ0~8o;_(@4TRSVM)me z$@3M7Yl4G=d$1k)9tLQxkFGmA{jA2wvRit8PfMoOotIgp6B=phSktRbpegz*_KbPw zXCkk!j19Gql^09XBVfR1N#oWnTeg@=5$wpm zcBOZh6FDV&n)7NKrE#?&uy!RbBu5GztS0p@&NtKWLh%6~8s5x({&Utwex7sk`ewy0 z(B1+^xKKd6Kn>`@HHYdOHtJ>m{{FL%NUvJ;3Z6|nGGx@+(b41I5|+^$%CUV45tcDj z0NofsSnjZ6!O%aF>2LlyjY&~2X!3~nv!!#R9Asr=WDMXLNS;TC+h#fIh$0y!*X|dn zmg;c?(!#GDwPtrS>{RxlFoJ{3yIyj3&**EKy=oR2rzBx%9Xz2adkBFMWsyhw8#ox- z4>=y3Kp#rO<$iri-SwL{vk;u7HjBggqM@^V?pynwl=PDC52E_&pw+E?>NI^hG*}Ie zE1z#sTC|JIGCHysM%o0PcIS)SFLZk|SN^c3<_*Y46-~RFaz4}W7A~9L-xc`MvlAX~ zf$OnU86%O{lL(wNZ#1Q~4t#+XI{LX?`{ED92GO*yrM&2=#Z&-wW3cJlz8woSNh5*% zHnr#jW>2t2YOG&Xm=+cl)q~3m+E`B6m1tY|#a^P6w*t%et4l8HVZy=aD{d8jb?Y#d zK={h*0L3{|U*Y9+6d5<(sv-Djao+ZvB#uCpR%-Y@GA z-bY_%o_SLMl!}L;t>I@`;hJBZE1EHGCmqjO%81}xczq1n!hq@S zm^{`6CmAZQw2Gf71qB72L0>>NAtNwW7t6gv(ZB4B*N?xu*cWogEpiRoLyxFHw{n(N z^ZStX7p)s5aPjP4_TifJid~F3Ij1nBf2a3+6uK|y(1#`NaM;R(1PT;S+lAh_)k&s} z)8;HW3fJpHQXpdZqRqVRedpwjQ&ye;GNWNw%_5P+R2bk`0| zA+EHg$U)+o2$ve8ohgmTT;xFi=xDlBICA*K7fgF~{qk}m(x+)hG>1L9sqE{a9PO;!{j3>V*j>L_Bh)!cwq!6P~f!#QI=uYs-$l61m^RBpmOc zR8l`~VR`|oG`Tl}pV^{qlYak#%7MUZ{%UmRwUlM;#CZ$ z5xoRWz+Sz2HS<$AT0g{x4W&FE!37JtF_0ir_ldg4yayPKSprQ%#XomNrmCGelK^1o z-;O_9A#aeABVj+k0h7VpE2ms~$-AqxAjfzg$k>g|XAC96Wlx`o=+0=`|HawSRLA`c zND@iIr0yb(jMLSWU#+b(k<-_kF|%=0;mF&YyWHwjj$=Q1E@O14A5dHc|A!`uqF)=Z9$YzOXFU|g70I}Gw%Y~WZDrtVfushj!;E?R9F$bajKs%EaFFrB|DD5UAgI z)>X5A;UFFXd3JHQM9a!SZJ`~tOV<-okv3p(9gE-m>Qa;IVXWp~#o)`#-U^aaG+9o` z{BZJ_&_*2LL2F$%j!en6t;?&akRuBV!Lx8REn6s!*(z%M1f$cZPwO<=(*UaR!@GB1 z3fXAd!1ibtmRl3pr;>B<>Z88L0=s`f`{XIGsH7Hy%^9b*^!j8y|O^26YiV$2dUs=ikamoK56j*dM)ataaUH3vC-sPP}v8V(@G|IIlaS&@59FV3-hK=fl2rHdF=Kx81j!F zkHKHZUb{9!OjJ~2`*ss#_K%Q652DaHZc~8S=j60LggcD^Az?~YC!mHJFnqiI%+&cO z20sHu68XNsj4oWdbe&o*c%*d9$rj%3u2Y$k%LUG$3yd{)t068>dsbC-NUais!)UDN zTU`2625Lx!rjb|to_Bfsy-Z=B$F)t)$Dn~7gabJa7L1^=+OfZ5#lUUQ^v3#)+V1Y| zA7yK&Ufu@}%4S>_rB;$(3%renj;S9wFgNZdI{&9=yuuyGz>XDzsILa-GOdins#QC2 z?*@QMK(Km&z9Qcq3PzO1+~6eK$ad7r_wqOpp{S3Gcyy2mtgPW)gkPizwN(;LG*Z{p z-2OEmY2r?Jcz9>Uxj#L}W!8QYcoVB+R$zdIBrgRB=PjipKtJ>j9lC=>Vb(Scmj+0O zE6X;zJutGH1999gh}{ojYhorOpv5?(LD|NUCoFRZ;-7;qR2(}mX^?Qg$9v!2y>LXm z_n*#hdGX?fpk4djH;&<&5f!(6Bc$?E->|fRUtBEHR8v!9Z)rolDQcA@eC;)rr9pua z{*wwA4VE)P+KTQI;zL#}a*e&g27W}wP@Fv-tb%^B*$x^MN82RJ6xI45+6-ho14*OQ zn3(Zu^Vrx}nkYfmEo}FlGiT1ce(f6kXg?p8&TvthKeZP7HlDacW4H#>ANmZR(oZs} zXP{I0uX4QM7~uL^8W;|V0XEM%NlA4K;R8Tdg;OtkBl%^1ZW*uf(NX#&PqiZAFjCy3 z#THA6FqI5`^AwxuK{LYBPgIC-M^RREL8_>Cd6v+HuDwuhSMeB7Bvd`oSPV`9!Cn$6 zDZb>|n*BM3|9{Zw03nU32l0BFhszZw*}r&K40N|E)o4vE@B!i(HEt9l2PjA!pkJ8hV=-FL5f(nZk=D) zKwox>&ND4x#D>GxPdhqJs#W4~<3UILr_0-D4a__dflfow^?cjU-&&o=oEB;#QH{W2 zL{|@nh7w2s$8sEMRW{GAJ7@gL<;y_8;?bdaLSa^g2t$X^aRtUw0>}C`d}8<5CjjR( zv?3_b<(#OvIEU62aW)CXmwf36ayr&D4hYGZHl`1kTz-8@J^&_~YD-tSwzjqH(%twiqO5pH&!+HM@T{3;1$Q;jq zrd*B;J%u^&U7(*Z?P*P?Z~F_Kb-FRvR$)Y{jgH;Wwn^ zMNo1@Ve>MQ$8j*|?jO+w*R;if=+#2|d|{fkr-SYnJcJo$+%1@u zi@-e!z~{Q3-$AseS>9;!w*ZqsMpEp^-LmP^jxAgCX;uTK6(DrRf)KziW})cChmoFl zRVNU`l ziUXX9V8tAcMIk_UCLKb2!{Cp5(#dF<nn6JV_$3 z|J}G`Y;wiq-*b03O_=b<_xwyR0L_K~C4pciVsEk;=OpBs{BP3e7?cj|i+6gvi zWAYJIbwXdC1G*#Z@=@T|G@OJk$vf`hYyA{xDQp>g5zY-zVt|`PkNOP1`|(m-)#vWk z^U0{`%-XLr<~L$?bO1nY35>#Sx~hbH91HH+09cbA9-~Z?K7iWe%hyI&>I2rV6-4+f zAo87rcALSQ_JaK3DBpJBM3h(dHw%zq<~edHJ764bfSTg(8Q{m<_ohB~;QCZ?K(b^b{chIO#&@mj9f~jF^EWN`(EWRH0bLx`-`_7QC#>M0 zS~IvO{DUbR{5D4Mq`8hJe zix)owBh10>tPS7hOgC%EPBx!-H)2V&@RfvNAkxv}po8>GOibPiPHCEsC@$U0`XSP` zTmD;HYwKoAxA}W%1YV}QPVBkTcS{gxB=w%&po48ubloHQCPtgm=%oEOFaHt8K1Ue{ z`EUeX%yk;)-~qIOWF%I5biflymQq;GPgFH#OJx_}*}(ss3W{r00UCWm5~Id>v#vB< zOMV&igQm-3LD}B#G5YYq&LOQebQEDM$28ujh{GPXJv!U>YcS^M{;-g8_;?>}@?I!` zDlt-t&Lu8H&;n>#iRv*9Cdzme7}syyIE3R#1MA^Zcv}|OSn)ldjp6`F@IVl={@+_pehAgex3GP=8F)?7>IY85;>w0a6u>pSJe1_cN3nn8(*YO+*qA z20&75-Ymbe;=^B9k9a+e9J=rW70PCW0(dVN!|2Uu8cUq-_!xW85`6&fVsKUyw`<8s zp;4%zq4$nr0Mn{rv)O((y8OF;@%65Lb=O*CS#?Z8V$uo0W(ZPaZg2(n?Ayn|Y#u$d z>|#^LSK>?4M7;WTs+DwYpO&}BXSe#g2@@tZP02?IMpNF=P!vFn>D+X1w`}Xy85$vy zPf!CQhkB~?PIgrNSMI0Cg*?Qfknr#X1TxJ3QpMuckTD)0`Uy{8jAX%k#ge0hV9_AZ zJ-c^fD!{t6Yu8R!nK`gt{=77*$(3Gx=aVo&0z<;~VcZrBO*GpCAtShRG)H(71cZk5 zsa6?4*`b1b=ePpx!}qEZ)s4<8qvp<@t>Gr1$xwFoP1bpIAqk@{DIyF*J%>V^3Iyaz zNkzpYh*&K2iFckjZMJPVL^Fa>ldSLbohRd{wP6V7^n5bz)7nSf2|OFy)JiN`vDIYd zkoV}y4;pICO*o}dU)y56xfabL#h86wG~}y5{rmQ9vD8$`k9EM zy@Ns^?w9ceT-u4n zW@8jvNqBS7RaI5DM6GgN^mClP2{FlY*n`3>nT5i;9RISBo= zS`ZY6&vEh&f{32+&heISj+%b&=p(EVrvN%z2rL%`Wp3?}IWAV*6uQS-ITAAv6WJ^V zhpt2u8At+~LidkQ5r$$-^5El3imu%f>zr$^AS>$Q9b~{(hJF8@%a{9CbE4}|-!Zr^ zOhF(-R?gKli8t5)xziCqj*)Z?>63PMQ5q9%BxoHSbGc>Tgp*SaJRUZFoa}Tu`xX{cLiQ>qMqvz>_+aC>7Mfpirl+qQ< z_KZ&7ioK**Lc3B$vS7x8!Zibe1hMRorj{0-e|_*_NSs=_yuG7Kh0gU4Xi}dLK(qi{Fid`5+cXT${L9Zk44Yb zx5<9uy$G(f?Id&Pd1=yR2j+ZIS#-g=>1_hieJ4^t4KBMDLpyls(j}>iq$__;U{LIl z6pPEaOJu!mOTH=OQU>Ux&%(z`y~t@aV>$!sNA8(8`Qa=+MPloOx0;9bc&M7 z%4TQ|^{{@jrVjyXc7fkzJE7$*E7KVM?^ST~5@+9=*RByA9D^L*iOk2}%XU%v4qQA5 zk`J2cyzNxPj~ah^4fs0wNhV^^BH2Nb6I5>R`p{cdnWtrBSX6C_0rWLtb|dk-u=uDt!H;mBnjb@l5Hy?I8o@t|Vi zHE1GlB@_0CQwVnAa@Y3nhCtja94ZMS|7tj^cf)6;*d zxs*y-Q0zC|dRM6UFgHSSPji<DPSkwy*c!s6o7h@e4%)mcx}Lk5O9Iu5`kXyzc~nM{POY5D4nXFz(4mT2)n6NcHupe-^2B!I*eG(IM5;oMGqsQp7 zBJ_$GsOOMZCB($`pdPv-3aDCT>4XVyb>P6x%+yd*laWiq&nQ8Wff)(VuF&;}I*aE0 zHUSKlnF;&hZ8Gvh34r1E_uCW*@uYW0iInZ>Q!svM#1Zpr^0;7TjBw&ew*m*iy+o%~ z*8{TYx@I9_+6Fj4b@ES_S3* z;6bc-B0zrRj|9Qq61q^1Xo8Q*=sxV}{f}aoEwvMT8_*9vjI2gkH!)~ca(CpioOOp} z7IhF~3bGEHkLt*>}B6VR8kpbDJol-$}UTmRMz&Q5))$^ zX+uOQV_!qjjuX}mk z&*$^r<)a-9=Uike8vHuaS&v?J76C`>%#565vw=g_+r9S#y^Pz><>UCs%;*V){bC;4 zo!7d!T+?*6>(a$uKstP;iVoN~=h_1BPm8ksC9-*_w)N;#Lp|qA?zwn=gZw~#*aJwi zpqTG;o&{;UEFWeS_j#Fo7o7SnNbG5;lAQ|l{iyf)y_r3}cIxT%I?0Qu>;M7UO|kRs z@be9NH|Mu2@q;|L;sob*w`KLxG(y`G%v-{9AKLUOvNrh=v`72Z-q&KZx)j$SpFVK&d5=N;CHI_W9hsIPPBU`%kN;5ut?9eM`gDtx@rr`{qZi z(`vs!(<>}Kj@I2BC}&&p1}|SYJp%S7o!{+cIO@;9?b{c!7dwVd>cEzhm%g~i%)R2& zNpNnVJP)4MyRuUps3C4Zy({}RjSclXL%g));Us!z!#=Qef9W!No`DBET~C{%^tDVt zsxk}*I8UYS5bMK-bbO^=uBA1x#a}@d9O6jU?gWu6O0#fln?IVX&vqV?eB?^N9B5!? zKzg!eMku;H8Tp(C&xvr;ahVek6tsXKDc>#9Kg~P&!BfM*gC9f4KNaEf5y6VB!0a$5 zEEzr}eN`%{jpSSO$qR3!H#OGm9c!a6-#wZ?G1QY)&Yx_Hzxdy)Ny+}3(8-;C^nZxHs#ur@rX}41w;8RMC zbi>_)EFoM-D}fN6(0uGJuI=ss=#xdH%bI`~jU?>7=hni@&iSuR~ z;!8x6UejXb(Mbp20yv$A57J<{a%}h|?_o?sb3Ov}iV={-*Cqj1j5}NV+>|+!7$_Im zABod@6N69f$&M=1eUy3&IxmAahlG|rN9SFd`@`VAJR^|~h||voCBN9=oVT1(N~AW$1XL>%lQF^Q__AMpObJtbNUM`^I;tfqHjVk zd1o2l-pISf-Me>T3JP;Kq(fuPv9xSkZUnA$X*$rptNlyiaYqc4;8 zPJC%a;!CPj>T0-*l=Hq-6)7m`LoH~6aZMuyvzB<%EBrQgUVRtfwt^w2CdSgkJni)Q z(VJP|-DfL3-CI|Q0FfH}C(n60z}#Fxw|jfyeQMgX)M15DjtEmmH&Wbdadk$-anWmn*9(wP}ate*mpP)WK zUUN|Y7u}h1$=`FCJS}tFoD>qf71cQ_Ik zcExA&AB&4Fh;J75I=ODyl2WZMH|uGr-0b@G>*7=3Q0`fIm_sqhvwTsVe*q;UJhf{P z2kK#9LC4zMw7psIrPq7=c<3K}e*JpY6L_u$SZu;rliRU z^JCK^?dx3ktjzZA%)VRvbg9oilZ~FG6aCwE@n}2Or)DlC1!KQ!dqzdmOXR41j80PA z=zDty58@*!eHL?#%u);Hf;EgsZ)psyXiSb6W0taOPlGpmTZ6_fBfb*OjZ~izfR9f) z`1|xz+}-{S!rR3a$WXncJN@QN+U{bpD2&Mw>Mnw#cNdnDhG7Ao0fyKRJy4d>clRdG`}d%1IL0$7OA9V5m_FdDT=uyZ38WHH;SE001tf33dcUCYw7@&&PI z$#Q&GwZ9dsJfnA%wdoPz@x~9Dr!XjQOkEnO=kwD|;Q-X}OLxtNbsZcKFwYj+-?Wlj zik0(Y7->AIbI}Oj(Xq8p(FH;>IcrNq$gIm(uAC$d3vXg!?40ML%DB(%2QDcE@rbHv z8JJ?+-uf`rxj;Kfenmh&qmTEU$8=tNuF6NZ_uaJVv|`(Sux@gf0hgYq{v5jUT|_Nw zAI`BVPHm1eAO1lz{RH!QvoA1ubZ55eQPa4S$8ku3FmwiP@%OJRJWU@U)VrUTKgl>= zAL{-xhjz1WA8u%3mNH{rCk?E+_EhMf)_vmSx3!Apmy5D4cqd3B)T(vs<%IA-csU%0 zK%;C;oQjeXgcrE!OuZg_DmdWv((<59xL4=@ME?-)wx+tEPZ2N+5LlyIjN3>HokY8n?xjz zJEo(XfWHH#N;F8MZ5FS7h*;Tu+fMCQfcuep9esy7e}HAAp{*UbvR?!E@~zbwxG7uv z^z^d(n3#8J=YtQAi0CJT`M+7w)|PXfVZDB&Rr2(i_TGi(|3q2oH~C4|GJmcfHAzlQ z$@74cq#_H$*k8QfmIrjymdp~l23HZ;ekarB9G7+_6^oy{6=H1@ernA!YBSr>zHQq; z?vh;xr3rk1NjpC;c7N;D?DK6<-bBj-*`>90Hc#nyrb)em?}10U%kYvDLFIGWbc$bItb|Ii^^N+w&=+?O$q9rPU|GVdx^Il3QdRVDN)Q0EIP_V;k?eF1l^TNhw)Ib{4f z0%P&!zh4s}JOI-z-x(TjnRez^~g~(YQn;Q*!&D%H6k_ahm_3>l%IUU8YzVu6~ z5wxAJ^d2*66kAQU*Y#6ohC_lv9@E<|j>&c&AMa;ec^Etn4YA8mDgwE~Ms8N(zC;Os z=kx`WFFQJFw_9l@j4W9=leMwq@RExX4|xK!IF=_!phoXkfKxt}_4SXfECy&G-PrRO zyL_FeKW&NDn}cGdZJTYq#%*uhLHpg5hRFqd2%R#FZe1~LXl{QklR&>rG7v0xujPi7 ze?CPEFpE`{%Rn1H^4W=)`7OXnjDQ@d$zNTYF0YRXRddfpiVequ5?o78_H?R?{^UFf zAUO(+;{q=!w{C6CH&Pt3X40Cr%?G!%j=nO_eTnH*`(`$W?VVye^Z6?;1-RALwO!eF zSy{hiowL-U5of0`N~UUQ73AI1LD+dYl~2S6?`z-XZx77LiL&!Aes=Mhh$`qZ&j{^p z!h{LwL;4hLdB)$7CvFe@f&^0o(FvF}YMiloHml;YKW5X$Ed_xqKh^8X<;!oGV%#nn zV`Av^hs(j5Zn@$rM(gvwy5qQDjm0RZ^80q)`SDRf&zJ>5d7`ms>w1D5k%bfOx>L$e zt4xO~!}I;hYt4(#~k{cRTkorSqobj7jtakHuOzNR3DE`}_X8y9Ze zP0w1~liXQ^*x-RREk%?tOeq+bko1yjNgW?i0zHC;(;;Jp=)h3u{a9<=aKdx9SzLl* zbb)0#O_!B6_;LXpC!#p}H)V3<^x019MYJl^O43uNi zD*BXl6htxgP_%Ni*#B_)Jd`hjznz6;_2A{pmt{wOOI=(?5`tijcoux$L-}CIdIxTp z+@}QQZHXPb`0Rf)9sV$w>>}o;ryobzWr10~WrW$xb5{)4U#bch?MVvx3w>>-KEE&= zi9$rn?V^yX;gUKhHftP4Q~aaB0U7#5UU~So>RB!twALBglW5LKUeBBOw_SN5o3IPF z%xGI&OLhcX!57jR2!<4x+<*3><&)$HRuY9lGY@uc%fTQTuuQ%CDr{v{t!lr zm5!}8Yc-s1pE8|>RX$mF{>_gMe0*LfM`!*}VlG04^~zSyhgps#{}|+~sDWw=HFo1D zhUd@2CcE_W@!zO!t~9tacgIZYvFS0ziu%SrN`DL_(tx|>8H_NSGwogeMwO-*ZFDaJ z0VM53&JodImlj%=7?uLQ3j;QMeppMjce7*Zw(Bft?Xw@<*3WGw71K)g8(-oOER^1l z7=CNk*O6{_fCLGYOL8i5MaxGqwZDscgEEs<2Zb{w$j#u}XDmlz^+T^^P}8-QNB^kT zH;~~<=t$hC(X&Sn5%bx)7gi`1pSi3Vvw&nci}|hS9fY~%nkJefree#5n=1o&iq!#; zvwiB1Fbg*fTY9seIN8><C+V;?lu{B#cHWEskK`eJ9|BS1 zajTuOzKHuzok|_@$K=mfQcLkDM z%mq+l>Z??3hVt}wZtnpHPnw9S0Iw-Z>UU;(iy05Z%9dQI)mz)|3>0`k-Gr`5+|V*? z){k7h`AoE!Y)77Q?UVuldO@SK{->J^#gisp;T=g3OIv%#%jcqbsJ`|aqw_?Hru^!# z@(K_sy`nYI{<>pVA1Ep(Y0;!71DPM%XtICX<{q8rwhv*IaZDs-!p7&5dqgfAbo9pZ z8z)tU21bV&N;u&Z-Ws}^pjAwC?Ut;#3t@~U$Kbv}^wQwCkvpQB6d}mGUVf&2 zWjNTMFwEKMejPV7cd?C>X5bu42%d=MUv$n74OPmver{!Ln=+&Xq~A^}@!eVjoT}?= z)ToUd1W=&zu>E4veXa-l9TQzx1N9L@@s*`v8p+6H-Yj2%N04VEQ}ZXNOC4lyp~V;{ z+dD_c_;u?3{$N1r@hZh-D;R@4<>R+uSXC#}s+bQve`0K44|f=EPfT#@QRuurowCvv zr5rAqC+I$NR)p@}z58h1u}dANQDfIZv0Zc#e=s|*jwl*m76SFt! zWp7+Wf*xJE?_Bij8<#%F_KrolDgH6=(*eh?t$p%K$@9*XRng$3!>$zTzO)+JU~j_W zEp1nQ>YF^f#AzFVg5|6F)*5~6KAYTr!ntE2b^rL7?2l`nZ1hV@qj_mt9x!wCa42{0 z7o6g^(v>*W_!j;lx)b3(^;!zDPHk; zaI5;!DT;>^H)+S%XP>d&D9_xg%mt=kW-u6{UXCp4VgIFeRFk#?N>&u-G+l8q_e8d` ztKF#Mz{Mx$Bs`QAoC*2Cv))Q;M@ch_PXPY4hf$LQ?XS&8KC(PJ_O7h@jWoYnOX0DV z=5x#I)rt{`079P5s0iH|hIEdZGo1;ou*T9E_t$kPEr-7cw&_k0w1Swa!epW$-$4RAPCz!KoW}hwXE9VDMOlJw|z}ZC7 z`Id6+&a}Zgg;!090xRn8@;veUx|VkE<45~U;^u+q?~uvymNR?o<5_sM@RzM}Nm zGpZ$8>_?z6%jLv1?H_xI7Q01Vo#*QZV^xe~jE5;cgC=_Lf_sKYxya1CR)uj;^n78ZTsqCK_|CoO(6g z>RI=HXVjDX%h7)*>P21<@#$v&xV)eUuTl1Eyb1EfyskTr;kLGvAZ0<7fK^9R6hjjI zB6BSv0aOlsKlR;gr)zCprd6%Q&q8ZnZ ziy%=NR4Pi;{*24%ohpKi3AFN%Nic)sN$KmWaGS9s1z4ZoS?>+kRH?c@ zCNzgMQm%8D*N)DQf3oTqQS6mw3cc|Uc{#ZPOTCD%cBL}52~7UDLt54EATj=`cvmFNNR(^D=}R%NXR0kKB6#?r5v-)Ss2EO|i%uEyr)Yq{?wr#H z-$A@fZ7N7nj;D{_JuX`DR_8T|8=ux|eU1zDAFF^!Gv4GLRH$1=-0G&d-emQbGrqCc z(=4Ah|1?=(_Oa#6TF#?&?DHle43`;zIZo+R*d=edgyZte5(`18jk0NJUq^v*M$cgVP7&KHZQ9K>N?Q$r&y zYh%rG8~3|i)MuAj`--Q%913K7yJgQ0{sQ+RybjG?+HN+2iSp_$yb-_wuiB<5$A?nP zEtg45?TPlau~92)Rv8{=j3-X5?Mq|^cUBZ^Su$sUE?9W$^VYSRkqh3Ae0?F>t)1J9 zKild@?I+noTg1rm&ZMl@)Ew^?0I6G1D(F#w=X@B>s6v(#cUj?8c>1#D@T)UsX3Rf) znB?tor+4)#Lri{b8zdKrv%=EkOYN&Jz&Vd_(;auOr<2lp(^v(C5$A>ECJ?=_RE2=b zG&w*uB&1Wy_1t^i4{z^3XC_!$25Uriz(;{Da|~xaXip>4hj0P6s|te_hK0}`X-Y=> zbwaa2P=~k5vPk~_+1btYZWCcpV(o8zYTepi7q3Db5cUu7kE$uihA`#{gyq1dM=5prT z*>{ruOSgfA>a+Mc@ixy+GuOg6*nH$CeX7>Vj){l=~dwfZbZgjMr2 z>jDEhn&I6{%l7ll)dvllMy$WbEO{S1qkjAr#>ww6wE&+_pktB6kwZ#;z9vLgioj`1 zwF6bjqjT%nbcJWv_Pcm(#!TBobY`9xTU$5W!&)HbezFotxO9jxcrzKG;Zye+H-vVg z*1Pv;R{4f7BT_XL1*nWFK^IL$>mrnFBnK=sFlK-{opR%65PHfVc9l&Kf+Vo{gY(O0 z2}XRR8N=T~_^?707t)=uGnw=i@99@BhD^=7KIB~)py6{k278bg)c}eRu``=$b@@d49RY)RR>%!>%N&LlxvR2 zWVT(%@;`9H!MrJ1QuSVL0bQYIZK>@cCL2enLgR_qZg@C!Ht?*wNrInreoZ(05abk! z<0#4-%_Gg=@ZLpKE=0Aes+IiNFS1(gqxq_~%RJh`NAmVLvwhI1O=s2~=d(sM5UE-~ z4ZAZ}-?`K41IxQr)?O<8ig*28RyORT@s$se+aqN%Ye@vIwOOgOnUe6Pl6LQj8;ySV z^9wG&IOr?whN?{oHX?D?gk}~rkO2u-N0_dR&g9X`m-9BRr?u(QrAxF!%$kO31D$)C zv{3le{klv=iE2BQV}CtI*8fmMNd*NWyl+B9sdy&?NFEUGP=H)}`b zhQF>iW}XNx){Y(}K>SZC11<&kpR+_P!zkY#p_M#IBU&)Jsf^Q;9zQT^bH_*b-n|Rz zkIJl9?R)29EpDe)H5U;01qTJS8ki8DQ1ae(Ow-SCh#Bq`57!($&cHiws_#BWu&@BJ zPb$P!z}|h}is)Bt%;@nZiE3Uvde_=&x(Kgm(EQ5{WUo|uv&Hm(KNUV3yA(3wAbiNs z>YW!Tz$=U$j6pLBr=ZcSIFG-M-bZ0FAyL^Z-KRS!85G`yor799y4Nj5IJ}Fq7BJBX^+zI6J0jrEyabsoaIi7 zWBc*qSdz*c*sH7KPful`*-GzaXfa~oadIj$4aKimsxsS+tV?3hT(ZCr+MX&T2^ z&g;mQ=6T~~cA~KG3S`S=Fg5$k@o!+ba5@Oeu3aiMbPqzIIh?4K*HwH2Ku&(1l?4mj zQz_w~_dI3Cme#47DMR}-ypSX4PxE{s7;rvzzi?;tj`qYx2W>-FgCiZBS-hjt#x+S;hmoom8 z6j*gDd$=DS@9(j?D2ooDQsJ~$|DTrW5wc{pQ!zDdv)|7N;svuWw4|y|K7U5+`6KV! zO+=Ddn&j5bEkmN0sy_0r{7Pp`qdA8bOwyf}sNe<@XZ1wkssv zcF|U&GyNIajkiix4hZDpc&%#@m)Vfub{2}ZM&|J7lz>2sWEDj%y>>$18VV^c>0_Li zm5jv2VALI4<&jX12t!Tcdm~xg3J7v1EBslh((jSP;Vj^W(rXblAMg&&ueKYN(=9pR zKJVvkuhDVW`!IWn5)YPHuSafvC!8*scvfzaAEm3y35t?;4$ z@+5HaejwbwMGYRxnb=}v@A~H`-|iWHJ7-VbKl&#vD^Ve2k;o5yR905DlYt!>s9|v_ z&|;cnZM4()XER@H%hcX*KZTtW_j?_6coKcxJ}$oGe|?9fB?Iq}i5;8?Vwh?f4VLOjZp(lRKo zu0r%r=&IY)G;X3}UW&@iWg!7yKdrL7*y(S*D^z&V^-(K8MzHPIKWN)FAdY>@w%6s1 zMA-p_#0`GOjWkzQwk%>U(2t)RV+Lof(u-RrBw;E=SxZN26U^5`(Wqd;SO{Wqd!4z@ z=ijT0%{U;U0VT2Y2?-$%LYE(6+CJvWtqHVV_la^+$?E%|ZsuDGr+f};^N>Q}9C~>> z2w7IwMxQu{7V3yb>{$`RUcf2bvOrgj;Q!nD?gOsg9Tw*f8k*TH^;y}FWQ(e}+I=9^ z@Oe%q<}*ziM|+0iC_nN1ss#%c1b_4fKoJ#1&;bNE5DI>E|GIcXmGSD_`gJ`0+y7ha9*$d-1Kf$vUQ=2O5IrM2L@Z)Ep1fRBx> zDE?%pqbPNS-0Nn>p?QGAoarQE<6o)}0=eP~&o*b|Z`h=`B6+MPF*Ti?caR=~Q!?ru`_c<;75Xc@_P{mPl%&soYT20z~86RjcQk zn-E8KXH6dhf1EzzikG*TpGR%-%UxI#Jg-mVp;G9l~TDq%gAUSZyKwdfJy6@hN5CD}=|&DCurY zO(%j}_21E{;ScZEht5L;o-)**BMq1Fm!Ny+UQC`e$(6!Ytn71^!Z^PNQot!iLUalX z=Fvi!20A+q9RPt;#Q~?^FK<;bF>~B@T(tniA`=lak5wwvIk1bCU<(!eaXJJ)dEGN( z9rsnfBV*l6T-`i$HCI?w&nJRd6VFIXbSji?Gsu9u>7j*hxAC!iX-?aQ$>Scc!lHbCveHf@L4_NBNsgrg}5x3IA$4jNW8U|oh$xZRlO9@O!`HRA$U#K zLVtz4p&8K}r8rgfsQnH7Y+Iu%P<@n*admq^M*t>UWTjz^Wbbg z=(xlyl3w~;!c$K$KZ|%maoyvB({oJn`)lz@`6!j z22ihPq0*wUVU#Ewq_cCRVkaf-4R0NDU-9;&@y{Tjh^d-H8yVcl7W*JqcH10bOx)*b zct>mCjH!h56(t&wnm(@{uG=(h1Fvj5&u9tH7nh+>@UTNbqHR$`1x1q=Ky9ZGtnNcy z6`znoIU_IaAQZ>5K}NOK!TmT*C;>+#n>o(RLfW)zm;C9&Kyv)>Bgg~gy7SlSoIjdd z0Ht*4lwoi5x>=*EOUjW)F3*n+m-Ha5Dno%mjSPPXSR=5yLckWCJ+t$3PH&~jhLK&4 z>%h1mWK`CIAbCt%oww$yigVDN*V}USCH_^;6fV-QV`tj05`4&N(iO1d&SuD&00w_{PahuoZ%T(FfqCKP<-rck?R>_6nVY&#}< zT#^`)Dzg;8ZVBeQ$^Km<3CmvBr*f%a0|v^1EY4<2>wuBI-R)wpPvt&kh;WC}WG<|g z>r+~-Rd+n(_m`HlAH;00=gNom7Bm}tQW|cm?I|KzFn^Gis8aLpiqbhgD%bk^ZLWH+ z;h1u5Cjknn9%TwSxTbr6L(93TT6G)ir*l|_@Ohd_m#QkdRih|Zg~vv|-G}NFjvt^) zWxWU?XHHU2y&nvj|6>?z$~^16ZM-Z@2Ubd~$OVYp~Sz(BE=v@yL9B=Ne*li8FxD^z-?F`e1D+DkJa$T13063frVwrOuEBd?46FwYzt~P)( z+9t+Ci%#1wH}u`Bb$h1dCrd;T695X|>8*5|cDE)5^i8;4pKC{?6g8@9=LC=Ww!<21 z>~>Yz`S%b3A}D;4;Z#$? zKsr4db{=}-O{mzk=}`N`T_Kq}-!oP*9pYY_4oKs$&?|KBCapYWsyu;e7$u9>QG;|^ z6pGWgG zj_(zZ)uY7KBFID*z*mX-rW_~+XCgNg(KHT{=~UOf`p!v=%dF?4Yc96BL2sh;pSHAa zeDK8iYflX`Fk3jzf(xTFdU%D()ymO0ACh0PpO>ejOKj?gg9{kA-T51h+q9``O!Itz zxfv#olMR9b7BV{a$|?s86V{!o&Ek=ZdB@s$4Pv7D(`0mPU(M&8b7b}@9%&)RP-$E< zcw`r2-d5Ix1lDigBRL>5`@=JGg%Dg!Q-iB+uXX2YgqdnC8$P_pc?{XscQ2{);Ta51n`|dKcX@237nGnTT@;BdV zm?KqHkD}*1Z=|JLJ?)%yztXl$330+dIWc5}{Um5fO^-=Fd1bImjyS31Wm! z(wzF@#n+<3%1^79+U<}aEtQ|Rv8mn;)Zd_DV&B+zXPRQChSegkrt}8wkp^+}sWC&y zA~Qjx*}J79LJzrL{B5S*duDt^td8(REF`{KreB9%jDP(ne1*l-!9?aDReBUtusp}~ zcztgovr@MHU}~YHFPZdQ!YLD3J)g~}nl6WbqvGAJIcpk~`2!5;)J2s-#>iW{ zv`s+e-z4tby}R&S#)WY)yCCLw|XilK(ldJ&M7R@Wt>)r5I zsD6%N@iz*7A_YsIVSR9^>W58$TqTAK6DDj|Wg0MeVYqUte{8)UffY;ujULkF{()tQ zSJeJEo-^? zR2AV3Cy4ilvn$maY|>sh21#o`rv}d#ff-m#d$HOB$ zB0D_AOTbp1+hPhJ=}%NgH&4-~Qs2UO^bO*tVAr>av6T3A&R`>Jx^ySECS?EFMaGz= zR)MOLM>`C5J|Iqdx4lZd4)Z10ZUT3eWxOyOVOK_F#cafmZ zg*2xR!Q1NmuW|;pFMgJmSYswIsb0E-LDcHSlfRr8u{t>o!HQ{64mr| z3?XodupMn&&v>nh$&py?m+q(7Q-{~s*d9j}_wgRJ8YJHaldGVg>kC32lmWw6mS+__ ztgo}B%Ge#*n1$2~Y^@M43XVESz_#Df8*Nu^3$8MrDTvi2m*>f(0Dl!P2tycBr8W6E87w z%-N)Mz+_eT$hg<3q@^N`lSrX?Z_xy|qjzP@a6$zI@mWc>Q1bsn%x2|ULM6}3L=xm35Zjf3^xM&#TUAkA@E^XI70;#DupHnZ<;+&ww zJVqfMwCiq5%Q3!xx*=X&0$Y!$u`NU{2CWKdzdYy8xOxDqOxY(xy23D{jVv9azuxhz zG?US^*bib#GE;yBI?G7fyRp^JCY%>;wM^+c4|X1T@65t`;@Gooq)uN&y_D=v{Q&F( z`uKGk{0ECV*ZtNYPeoc1_W;;MCd^tb=*}o550QPAe);?<$MdT4*)60k)2tVw8ckvm z=MxNMJrMWzkYI$$u}jL%M{q+b|a zl+GT^^inABzp$KxL4?r6|t5ft(M%qv#|R6k%kE(p4G_0*ze5!=;{P@xWr zad4x)=wP@i7A$IeP|zQchD^cMmRKn(tykcKLwRx(t9l-EbC2b<^gMrg6khjk@DSTH zW;K9Z+i6=Sd{x&0gt2$3(cn8pQj{OctR$ zwS4|hKQ6SDei_a(ms+IkPd$rH2A%!*_Lt}J5(;xxefTAbc@f9?W2+avj#3m1Az!+| z>pwz^BPpJ6bC~aEP9Eeyp(cxxT1>VuynNiXjptu=tN@5opQujloZ*)}T~tIF8$-y$n&!q za84Bh#z-N;%`jmOP}bjK?AQ@%bRa+L+5Y2IM73Qx+Ln)sh}E8P=Q~dIB1ZjCyn+y- zC`mf?Duv^BR~(TTw5`28e<5Xe=*UTJojb^U)aA7qA?6oeShP6UzcMes1p}YW@g>5`&GJk~c3&L53OpG*F>4B;$G$$^?;;NKt{C9Rg`#qqvFNM92*Xi- z%_Yn)!jQyxcq&9`q)=E&;pu`51LM9M6 zmj7#ut$xhL78@UK|^FQ;u7Xx*lCuQDbA?We7!dBY1_CnpLRYzuIbFZUTbcf zvfiR613Nm&gM6}dWutMU&0v;y5)ge@$K}^;i<3!RH-z^LW)WYY;Yl3iX5t;5V5A2g4$n}4ip{k)!Qka6^?)@ zN_Fmhe4`3JebXM@f7mBLD(2oZ7$^sdouII-f|%X49Vi0$1L(h$>VsB+mlzig`T zYwKCS(%%!Au5R_@3n={MN%pjL9ck8+j|%45gGXnqD>&xrwExmCN}DKTokdBZX3uX< zNNl-Hzw=*r+x*=CEqNLiHGHT{A9~X4IseFuXs0sWwRz8sO$)GupZ4K|)cqia+Dpibh4>nC(9?c)&Kq0TA*hDm>2vYfn(W#E;VO>Kb@}7p-|TSUb)-}r0@}5 zyWjPxw_fxp|NixS;I1fH|XrSq};Au z?}8aGbN&}-+U=rCT3T6ri+_J1zb{KBZ4U%x5JDO?>R(fXNm%j0O*w(;@)vTYev7Ed~|e8CjoSliTmi{TG@|pJ?Q<# z2LJxX94f49!FC2xCW_oCXynT-j*3;`|Kkabkb6YFdeN)Uw1<x%h1M z^2;47|6=NA|Nr?f%nd5GOI+g+>cUKH{KJ>b|MLq|I2=aC0g5wE1$xBCjNTm7quMsU;RsK4eb(R1A3;BKb(bekz`_KQc{{8>| z=YKu7|Cc_8c7H*If1e4(<4|ap0?P=65Z2!TB1Bs)tJzF(Cn2bP0O*IRNd3-@OsEB- zlFK<8lAr1%d^_2NHLJ@|_Y(}7UUVA#aC0YmZy0kAQ38ug7agKHa|1>&6Zvi#d=~-n zGH>pfEafT1n%7NPsl%n&v7+I(yQzPxZua!|%lj>zinSc!=Nk|gFm z?32ZooLRN=zU_zo0oRbHj2ACRUQZ(+k-d(!^c@APEwOlZ!5iqnYp~z8@x25nt@|}% ze(ufh!{+)wr5~Wmnk{r;h>yBEV*8jv$&w+o5MOxxqhx!RtSi29Q1dghL`k@0R1^4U znL-3YYhUna*=AsU3u@yB2yg}2h_jga&kt2-Gy3~58~TT|$RHLL#>U3+mGpWU3^8Rw zOnMTkFGRrow3Y|P-y3gc_E08x^gzFmm8X&rT?$&a=?~s?4`}Z=*nmydY6zKP^DKM^ z#-maaeIdeQ^43qm?p$cbp6ju>hXQHJp4RxOqaV97wfDD;09hFGnkw( z9cqL=T;s~}7>8%4gvBa~SP^d}?WGNch}KDS#rz5$;4x}qO54flMevXAu{>_S%r8hj z?d=noYSsZg4D0Ct_|l1!dVhNrjye|-`+xhLf`tW)F0DB)Me%=*#Vh>I98;me_AmFU zGWq!B>!&+w%CV#@go-l}2ER*usu9OhL-)gK#duH{VT={Ds+1|H6lNWC7N$D2u2dC% zw)DW#&k2bKM(_249^dj)M2nu^g;*(E2fd-0YmR>4!yWmA4|85^Jl1g3?gOyzCqmEQ zxip)qZLZGW#>mlgc>MaqVrkR48x-g|N#TGiLJHw>Oyg9z{Bh*l+tmN*x7Tj}NhrUr z6}P(~&Uu?+X2ZIbsUmcs^1j8DoZ@soP3g!dYrH71p_3E}_d2TtGMpQB#kOAI;^Qlf z0?te5lK07zEs9TOh{S;)oxtyDv35{jND&}Z#N_lSZd9j z&O#&8?PWe9qz7m+jle z=sNRy1Ot-uOv!#2#ODZvJ8Ur5(|@)q%l-BU+qC=sEL+yOt&TjyCw>Idf85L;Q=Af? zX~jOThty+6P&u>8B~Y$pX1Vr;nkk`9I?;U170eZcn-CM~UYpAp=B|m7!2!~C0dv}d z94#cE;;9O=bp1ffcPla8J4&nF=gm23pedpUuU?iZ1qi|6R_I_cAin9?8HA9J%KU5P zny-gxNNqo<@QporWu>`%E_qqNXg)_w}YA`R*4i^H!} z$h1Lz4k2f{t801R&aMCXq$5th&;PH(^a^~o@ysuf?)Res7}oEWHieludq`8SIrqY_ z`M*QEomCAj`V`=N%4M@ajL+dbg&C)Q%CX8zi}7UzYdF z{CYx|^%S##d~qn}uFGISJ|8Y=XqmQjBph$S6pSLLL+0m;rw~Zb0f5P(X4Yiu5ZW3x>5d^w&kl|$h8|^t?YfIO-F_EI}>i(vJ1ap*ulQclT;=j*~tg;#<`Ol2gWRj7i z;lrfl9(F(F!1Q77^jWvL)Z(QT-MsvhS{C}s8kR~=5}@?L?AewB1@by0*zgXQ$CWMa z4SSV{VXuHR{V!C#jJVU&RLrT6@1JBOXb~jO6Rg~fkP?N*Mc_f}Z&AL@=nLB-zdb31 zOXrYVL;$<0hr7Ot4rPz5x~Yt8WUXm`C-We_dY|Z9YE&d(64`S#r}JdwxRZE0Zs+vN zRu68%&vv&v;WB~q&H#BX-0DV}cIBU3E)zE-raHty){$uA3Jw17cvrDF>f-fQTaW>x zHIWEMXE9GLaPrq)^W}3yB_{Ky&<0KtrYJJ|i`ZIW zbQ9O@Fp)^QhK~tC4R9RF$OuA2rBw|;&(9uZc^aLS{}U+wDQ*xK zV)A$U-j-qgSlPfFZ_t}_Ql#*~ziJXAu2M+7I+T?)5O6n0c~0E>uw^V7BBKPxbb|IJ z1qNQ8eX>un0JPvsZ+N(APMP64HsjLEm^zsciJC{)uSo6bAL)o1Sdfqj9Hrhyy<#XBPFxb!Kp14xm zEWSuUwu=vWWNgEs61XZR`~31*Sne52&N5^W2ilVw=xkcBt?$8X(;QiD3>)pG)Jl|E-6t2-vHR8rhp&9O*;Th+ z7jwo5OdnrpmOf3G4%z%p`I&DiN>TkgB;;4)2zt=L=SwNy8fv1(tO&_DPil*2xM?!y zz_6Vk9qvWWy*R-EJ(DHzR`X4r&*6cTSUfm&er~;26^iT|hC(-8LnKU>R&tC*FqT?= zKFF2%+K-xzdUyVI)=4{BhTQNtPY3!X%YN&lVIEErj!=fmM!0DiSKmt znWxYS2A}5%tXA97S`}UfxSRzWu_~VcmS6@OMCsY4JZU=hu8_U%84@H^q7zB>u0?qd zpLvsejc)RSCe!0Yxt@qdgylFc>pzj;$N!3w7IZ2R`S6k?(G#~99y%|G@PG|LQ#&%C_xw!L_R|uKx1r^{{?F9VmZxo%p|fe~NBYgiVT* znMZiY<1uP}+XC5T7Er2N{OlWK@Z41vCI0g(pdm}rHjtOvmOmk$?9p3)y%$fW$xK7e zYVdjg&6}IpINgf_!Co7T{nd{BDY(xqOLunn+u^<^dQGmc)y+aDorKkNf9rxB5{d_f zJnmZyim5-|=3D2a=KEi#=nb1kf10@(69?gFk2!eVgGy9!<`)7_RCF!yj)`TcD=~Tj zbVRS4fLNYfH$FQ6E+Kct@M(1c2UQzi!>Gq3-5KWg{gc~ft*MMoxn5A(^q+V3+|}s& z_c;ngZhNkNP8mrIYVPM@8WV4WiT{#bM)(gnSiI)*`xJ|IeZr`hjuDBBB84W)p}cX% z`cUm24XU5@7$Y%+-E-6pPp*tO7dwZ++|)64 z{l7t8`g6W-mKR>(dmNQzXE(*N9Ww_~;I@qiZ?qRO3MV38GjevE&YZHx?}x#qBYfZ9 z7gTG`BV(YN9+4@lsw5FT&UsDC_JAMgO;uKqgYF`H)rGH5?WwI+oMZksp}3M1HjU&Q zlMO5&D`=m3((O&6xk;+KJ;;1hGqfRg2OnH`k7BjwA`j?M2>t91#eBmU|NEgnwqW%l zZI$|)b7u;cPgvaNp|>pV+c;IwDgHlW7LS9TUW*9uC8u>wh#1fjSI? zQKVT@yBfLO2oY`p*wcFvo!QNN5m=`839FV8Ud=CE9ba7R&WQUheIC zgZ??K3hUd7N_~#yhPHUcrho+MXg#5}VSMW>yL-Dlc%1@Q#s>?C3kW_w*YoaM$P;wH5`B(7lgG z(Gj6~KsypmtA9c*DeNeFW{*?p(*Zmuoip8;nQq6%%cR=|m9xExJ7@;ntgWp@%PMqc zDiZoN84{%!jHU*3qiAVV{F-z95citx{kzjnn96WF-o5ykXt=&UC4sro7CxP21-nY7 zjf^)#2lUm3$08+h9g0Hp&U0N8?Dy?vTg(_7{P9%Fe;$^?uNonca5gA)-8=seE0Qoh z-XEe=C0oHfz}=qv^Mp0HxMSX5F6N?wm{Y?UlxTNOdB0l5bJ1+!)7Y87$5joXF)4rB z(;>5C?8Brcho3S0O$DJ_wm?Y$;xdLXBTQDgJPI%+psVhw~VubFu92U|am? zj0t`1J<34cRjk0SK4z5^FMFdnUP5Xj&4|yv?(DA+OCNB8cE0cuW*pMSkP0o6@C&DJPur}~_c^fpJKC@}xXEQu`UMwzkA&Pw_!NdZu4ulkPwi2@?sf0X<45?* zlx@2`u21Lj{cB){h|pXhK`(>D+bUK2B4NM;?;+k5vQgP?X;%o4cO^d2euXhU6KT@@4+jm8Fs3 zP(3Ewj6V)tcqW38^XLDQp(d>InDgx&Dntxe1gA(iLR1^9?VaD}mkTaCU;_#TnL#~` zM1L_!3&|Oz1(C&=OPwHwqPe~I3K}Y#*Dz3-?)LVP_RjGuCm^DdHiNvenB9Uy5Gv+> z2>bkT5tLh3Zue4hC(AponF`*-v1IIdaUh2Y*s3_Ti2eW{Uyo|MnaUYKFC0RLA%1hR zCs-gOPF47eAwoo^)TSD+gNvztzs5|~NL7PEor5eQ$%oGNk!j3P2>hZ_qcfSnX2!IS z-_@C|ZAFLp_b=yE-qdm4s9YYA4HF)xu9$KHtu=s@I_R38=N|P3o)$OJA2h^Q#LdKYp5JzQE4oThXPU`yE0o&&c*)M List[str]: + """Get all files from the source directory.""" + source_path = Path(source_dir) + if not source_path.exists(): + raise FileNotFoundError(f"Source directory {source_dir} does not exist") + + files = [f for f in source_path.iterdir() if f.is_file()] + return files + + +def create_target_directories(base_dir: str) -> dict: + """Create train/val/test directories and return their paths.""" + base_path = Path(base_dir) + + directories = { + 'train': base_path / 'train', + 'val': base_path / 'val', + 'test': base_path / 'test' + } + + for dir_path in directories.values(): + dir_path.mkdir(parents=True, exist_ok=True) + print(f"Created directory: {dir_path}") + + return directories + + +def split_files(files: List[Path], train_ratio: float = 0.7, val_ratio: float = 0.1, test_ratio: float = 0.2): + """Split files randomly into train/val/test sets.""" + if abs(train_ratio + val_ratio + test_ratio - 1.0) > 1e-6: + raise ValueError("Split ratios must sum to 1.0") + + # Shuffle files randomly + files_copy = files.copy() + random.shuffle(files_copy) + + total_files = len(files_copy) + train_count = int(total_files * train_ratio) + val_count = int(total_files * val_ratio) + + # Split the files + train_files = files_copy[:train_count] + val_files = files_copy[train_count:train_count + val_count] + test_files = files_copy[train_count + val_count:] + + return { + 'train': train_files, + 'val': val_files, + 'test': test_files + } + + +def copy_files(file_splits: dict, target_dirs: dict, copy_mode: str = 'copy'): + """Copy or move files to their respective directories.""" + for split_name, files in file_splits.items(): + target_dir = target_dirs[split_name] + + print(f"\n{copy_mode.capitalize()}ing {len(files)} files to {split_name} directory...") + + for file_path in files: + target_path = target_dir / file_path.name + + if copy_mode == 'copy': + shutil.copy2(file_path, target_path) + elif copy_mode == 'move': + shutil.move(str(file_path), str(target_path)) + else: + raise ValueError("copy_mode must be either 'copy' or 'move'") + + print(f"Completed {split_name}: {len(files)} files") + + +def main(): + # Configuration + source_directory = "/data2/sules/fake_enhanced_data/ALA_ALA" + target_base_directory = "/data2/sules/fake_enhanced_data/ALA_ALA_organized" + + # Split ratios + train_ratio = 0.8 + val_ratio = 0.1 + test_ratio = 0.1 + + # Set random seed for reproducibility (optional) + random.seed(42) + + print(f"Organizing files from: {source_directory}") + print(f"Target directory: {target_base_directory}") + print(f"Split ratios - Train: {train_ratio}, Val: {val_ratio}, Test: {test_ratio}") + + try: + # Get all files from source directory + print("\nGetting files from source directory...") + files = get_files_from_directory(source_directory) + print(f"Found {len(files)} files") + + if len(files) == 0: + print("No files found in source directory. Exiting.") + return + + # Create target directories + print("\nCreating target directories...") + target_dirs = create_target_directories(target_base_directory) + + # Split files randomly + print("\nSplitting files randomly...") + file_splits = split_files(files, train_ratio, val_ratio, test_ratio) + + # Print split statistics + print(f"\nSplit statistics:") + for split_name, files_in_split in file_splits.items(): + percentage = (len(files_in_split) / len(files)) * 100 + print(f" {split_name}: {len(files_in_split)} files ({percentage:.1f}%)") + + # Copy files to target directories + print("\nCopying files...") + copy_files(file_splits, target_dirs, copy_mode='copy') + + print(f"\nāœ… Successfully organized {len(files)} files!") + print(f"Files copied to: {target_base_directory}") + + except Exception as e: + print(f"āŒ Error: {e}") + return 1 + + +if __name__ == "__main__": + exit(main()) \ No newline at end of file diff --git a/scratch/reorganize_swarm_data.py b/scratch/reorganize_swarm_data.py new file mode 100644 index 0000000..e7808c2 --- /dev/null +++ b/scratch/reorganize_swarm_data.py @@ -0,0 +1,770 @@ +#!/usr/bin/env python3 +""" +Script to reorganize ALA_ALA swarm results data. + +This script takes data from /data/bucket/vanib/ALA_ALA/swarm_results/ and organizes it +into /data2/sules/ALA_ALA_enhanced/ with train/val splits. + +Input structure: +- /data/bucket/vanib/ALA_ALA/swarm_results/AA_{grid_code}/ + - swarm_1ps_{traj_code}.xtc (where traj_code is 001-005) +- /data/bucket/vanib/ALA_ALA/ALA_ALA.pdb (single PDB file to use for all) + +Output structure: +- /data2/sules/ALA_ALA_enhanced/train/ + - swarm_1ps_{grid_code}_{traj_code}.xtc + - swarm_1ps_{grid_code}_{traj_code}.pdb +- /data2/sules/ALA_ALA_enhanced/val/ + - swarm_1ps_{grid_code}_{traj_code}.xtc + - swarm_1ps_{grid_code}_{traj_code}.pdb + +The train folder contains 172 randomly sampled grid codes, val contains the remaining 12. +Each grid code has 5 swarms (001-005), so: +- Train: 172 Ɨ 5 = 860 .xtc files + 860 .pdb files +- Val: 12 Ɨ 5 = 60 .xtc files + 60 .pdb files +""" + +import os +import shutil +import random +import logging +from pathlib import Path +from typing import List, Tuple + +try: + import mdtraj as md + import numpy as np + MDTRAJ_AVAILABLE = True +except ImportError: + MDTRAJ_AVAILABLE = False + logging.warning("mdtraj not available. Trajectory validation will be skipped.") + +try: + from tqdm import tqdm + TQDM_AVAILABLE = True +except ImportError: + TQDM_AVAILABLE = False + logging.warning("tqdm not available. Progress bars will be disabled.") + +# Configure logging +logging.basicConfig( + level=logging.INFO, + format='%(asctime)s - %(levelname)s - %(message)s' +) +logger = logging.getLogger(__name__) + +# Configuration +SOURCE_DIR = "/data/bucket/vanib/ALA_ALA/swarms/swarm_results" +SINGLE_PDB_FILE = "/data/bucket/vanib/ALA_ALA/swarms/ALA_ALA.pdb" +TRAJECTORY_CODES = ['001', '002', '003', '004', '005'] +LONG_TRAJECTORY_CODES = ['001', '003'] # For 2000ps trajectories + +# Splitting strategies +SPLITTING_STRATEGIES = { + 'grid_split': { + 'target_dir': "/data2/sules/ALA_ALA_enhanced_full_swarm", + 'train_size': 172, # Number of grid codes for training + 'description': "Random grid codes split: 172 grids for train, 12 grids for val, all trajectories" + }, + 'trajectory_split': { + 'target_dir': "/data2/sules/ALA_ALA_enhanced_full_grid", + 'train_trajectories': ['001', '002', '003', '004'], # First 4 trajectories for train + 'val_trajectories': ['005'], # Last trajectory for val + 'description': "All grids split by trajectory: trajectories 001-004 for train, 005 for val" + }, + 'long_grid_split': { + 'target_dir': "/data2/sules/ALA_ALA_enhanced_long", + 'trajectory_codes': ['001', '003'], # Only 2000ps trajectories + 'train_size': 172, # Number of grid codes for training + 'description': "Random grid codes split for 2000ps trajectories: 172 grids for train, 12 grids for val" + }, + 'state_split': { + 'target_dir': "/data2/sules/ALA_ALA_enhanced_long_state_split", + 'trajectory_codes': ['001', '003'], # Only 2000ps trajectories + 'phi_range': (0, 100), # First residue phi range for validation set + 'psi_range': (-50, 100), # First residue psi range for validation set + 'description': "Split by conformational state: trajectories with first residue phi,psi in (0,100)x(-50,100) go to val, others to train" + } +} + +def get_all_grid_codes(source_dir: str) -> List[str]: + """ + Get all grid codes from the source directory. + + Args: + source_dir: Path to the swarm results directory + + Returns: + List of grid codes (e.g., ['000', '001', '002', ...]) + """ + grid_codes = [] + for item in os.listdir(source_dir): + if os.path.isdir(os.path.join(source_dir, item)) and item.startswith('AA_'): + grid_code = item[3:] # Remove 'AA_' prefix + grid_codes.append(grid_code) + + grid_codes.sort() + logger.info(f"Found {len(grid_codes)} grid codes") + return grid_codes + +def split_train_val(grid_codes: List[str], train_size: int, random_seed: int = 42) -> Tuple[List[str], List[str]]: + """ + Randomly split grid codes into train and validation sets. + + Args: + grid_codes: List of all grid codes + train_size: Number of grid codes for training + random_seed: Random seed for reproducibility + + Returns: + Tuple of (train_grid_codes, val_grid_codes) + """ + random.seed(random_seed) + shuffled_codes = grid_codes.copy() + random.shuffle(shuffled_codes) + + train_codes = shuffled_codes[:train_size] + val_codes = shuffled_codes[train_size:] + + logger.info(f"Train set: {len(train_codes)} grid codes") + logger.info(f"Val set: {len(val_codes)} grid codes") + + return train_codes, val_codes + +def create_target_directories(target_dir: str): + """Create target directory structure.""" + train_dir = os.path.join(target_dir, 'train') + val_dir = os.path.join(target_dir, 'val') + + os.makedirs(train_dir, exist_ok=True) + os.makedirs(val_dir, exist_ok=True) + + logger.info(f"Created directories: {train_dir}, {val_dir}") + +def copy_files_for_grid_split( + source_dir: str, + target_dir: str, + grid_codes: List[str], + trajectory_codes: List[str], + single_pdb_file: str, + split_name: str, + use_2000ps: bool = False +): + """ + Copy and rename files for a specific split (train or val). + + Args: + source_dir: Source swarm results directory + target_dir: Target directory for this split + grid_codes: List of grid codes for this split + trajectory_codes: List of trajectory codes (001-005) + single_pdb_file: Path to the single PDB file to copy + split_name: Name of the split for logging + use_2000ps: If True, use swarm_2000ps_*.xtc files instead of swarm_1ps_*.xtc + """ + total_operations = len(grid_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb + + # Create progress bar + if TQDM_AVAILABLE: + pbar = tqdm( + total=total_operations, + desc=f"Copying {split_name} files", + unit="files", + bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]" + ) + + copied_files = 0 + missing_files = 0 + + if use_2000ps: + traj_prefix = "swarm_2000ps" + else: + traj_prefix = "swarm_1ps" + for grid_code in grid_codes: + source_grid_dir = os.path.join(source_dir, f"AA_{grid_code}") + + if not os.path.exists(source_grid_dir): + logger.warning(f"Source directory does not exist: {source_grid_dir}") + # Skip all files for this grid code + if TQDM_AVAILABLE: + pbar.update(len(trajectory_codes) * 2) + continue + + for traj_code in trajectory_codes: + # Handle .xtc file + source_xtc = os.path.join(source_grid_dir, f"{traj_prefix}_{traj_code}.xtc") + target_xtc = os.path.join(target_dir, f"{traj_prefix}_{grid_code}_{traj_code}.xtc") + + if os.path.exists(source_xtc): + shutil.copy2(source_xtc, target_xtc) + copied_files += 1 + else: + logger.warning(f"Source file does not exist: {source_xtc}") + missing_files += 1 + + if TQDM_AVAILABLE: + pbar.update(1) + + # Handle .pdb file (copy the single PDB file) + target_pdb = os.path.join(target_dir, f"{traj_prefix}_{grid_code}_{traj_code}.pdb") + if os.path.exists(single_pdb_file): + shutil.copy2(single_pdb_file, target_pdb) + copied_files += 1 + else: + logger.error(f"Single PDB file does not exist: {single_pdb_file}") + missing_files += 1 + + if TQDM_AVAILABLE: + pbar.update(1) + + if TQDM_AVAILABLE: + pbar.close() + + logger.info(f"{split_name}: Completed copying {copied_files} files ({missing_files} missing/failed)") + +def copy_files_for_trajectory_split( + source_dir: str, + target_dir: str, + all_grid_codes: List[str], + trajectory_codes: List[str], + single_pdb_file: str, + split_name: str +): + """ + Copy and rename files for trajectory-based split (all grids, specific trajectories). + + Args: + source_dir: Source swarm results directory + target_dir: Target directory for this split + all_grid_codes: List of all grid codes to include + trajectory_codes: List of trajectory codes for this split + single_pdb_file: Path to the single PDB file to copy + split_name: Name of the split for logging + """ + total_operations = len(all_grid_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb + + # Create progress bar + if TQDM_AVAILABLE: + pbar = tqdm( + total=total_operations, + desc=f"Copying {split_name} files", + unit="files", + bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]" + ) + + copied_files = 0 + missing_files = 0 + + for grid_code in all_grid_codes: + source_grid_dir = os.path.join(source_dir, f"AA_{grid_code}") + + if not os.path.exists(source_grid_dir): + logger.warning(f"Source directory does not exist: {source_grid_dir}") + # Skip all files for this grid code + if TQDM_AVAILABLE: + pbar.update(len(trajectory_codes) * 2) + continue + + for traj_code in trajectory_codes: + # Handle .xtc file + source_xtc = os.path.join(source_grid_dir, f"swarm_1ps_{traj_code}.xtc") + target_xtc = os.path.join(target_dir, f"swarm_1ps_{grid_code}_{traj_code}.xtc") + + if os.path.exists(source_xtc): + shutil.copy2(source_xtc, target_xtc) + copied_files += 1 + else: + logger.warning(f"Source file does not exist: {source_xtc}") + missing_files += 1 + + if TQDM_AVAILABLE: + pbar.update(1) + + # Handle .pdb file (copy the single PDB file) + target_pdb = os.path.join(target_dir, f"swarm_1ps_{grid_code}_{traj_code}.pdb") + if os.path.exists(single_pdb_file): + shutil.copy2(single_pdb_file, target_pdb) + copied_files += 1 + else: + logger.error(f"Single PDB file does not exist: {single_pdb_file}") + missing_files += 1 + + if TQDM_AVAILABLE: + pbar.update(1) + + if TQDM_AVAILABLE: + pbar.close() + + logger.info(f"{split_name}: Completed copying {copied_files} files ({missing_files} missing/failed)") + +def analyze_trajectory_state(xtc_path: str, pdb_path: str, phi_range: tuple, psi_range: tuple) -> bool: + """ + Analyze a trajectory to determine if any point has first residue phi,psi in the specified ranges. + + Args: + xtc_path: Path to trajectory file + pdb_path: Path to topology file + phi_range: Tuple of (min, max) for phi angles in degrees + psi_range: Tuple of (min, max) for psi angles in degrees + + Returns: + True if any point in trajectory has first residue phi,psi in the specified ranges + """ + if not MDTRAJ_AVAILABLE: + logger.error("mdtraj not available, cannot analyze trajectory states") + return False + + try: + # Load trajectory + traj = md.load(xtc_path, top=pdb_path) + traj = traj[:1000] # only use first 1000 frames to avoid memory issues + # Compute phi and psi angles + _, phi_angles = md.compute_phi(traj) + _, psi_angles = md.compute_psi(traj) + + # Convert to degrees + phi_deg = np.degrees(phi_angles) + psi_deg = np.degrees(psi_angles) + + # Check first residue (index 0) for points in specified ranges + first_phi_in_range = (phi_deg[:, 0] > phi_range[0]) & (phi_deg[:, 0] < phi_range[1]) + first_psi_in_range = (psi_deg[:, 0] > psi_range[0]) & (psi_deg[:, 0] < psi_range[1]) + first_residue_in_range = first_phi_in_range & first_psi_in_range + + # Return True if any point is in range + has_points_in_range = np.any(first_residue_in_range) + n_points_in_range = np.sum(first_residue_in_range) + + logger.debug(f"Trajectory {xtc_path}: {n_points_in_range}/{len(phi_deg)} points in target range") + + return has_points_in_range + + except Exception as e: + logger.error(f"Failed to analyze trajectory {xtc_path}: {str(e)}") + return False + +def test_mdtraj_compatibility(target_dir: str, num_samples: int = 3): + """ + Test that mdtraj can successfully load swarm + PDB combinations. + + Args: + target_dir: Target directory containing train/val splits + num_samples: Number of random samples to test from each split + """ + if not MDTRAJ_AVAILABLE: + logger.warning("āš ļø mdtraj not available, skipping trajectory compatibility tests") + return True + + logger.info("=== TESTING MDTRAJ COMPATIBILITY ===") + + for split in ['train', 'val']: + split_dir = os.path.join(target_dir, split) + if not os.path.exists(split_dir): + continue + + # Get all .xtc files + xtc_files = [f for f in os.listdir(split_dir) if f.endswith('.xtc')] + + if not xtc_files: + logger.warning(f"No .xtc files found in {split} directory") + continue + + # Sample a few files to test + test_files = random.sample(xtc_files, min(num_samples, len(xtc_files))) + + success_count = 0 + for xtc_file in test_files: + # Get corresponding PDB file + base_name = xtc_file.replace('.xtc', '') + pdb_file = f"{base_name}.pdb" + + xtc_path = os.path.join(split_dir, xtc_file) + pdb_path = os.path.join(split_dir, pdb_file) + + if not os.path.exists(pdb_path): + logger.error(f"Missing PDB file: {pdb_path}") + continue + + try: + # Try to load trajectory with mdtraj + traj = md.load(xtc_path, top=pdb_path) + logger.info(f"āœ… {split}: Successfully loaded {xtc_file} + {pdb_file} " + f"({traj.n_frames} frames, {traj.n_atoms} atoms)") + success_count += 1 + + # Clean up memory + del traj + + except Exception as e: + logger.error(f"āŒ {split}: Failed to load {xtc_file} + {pdb_file}: {str(e)}") + + logger.info(f"{split}: {success_count}/{len(test_files)} trajectory tests passed") + + logger.info("mdtraj compatibility testing completed") + return True + +def verify_output(target_dir: str, expected_train_files: int, expected_val_files: int): + """ + Verify the output directory structure and file counts. + + Args: + target_dir: Target directory path + expected_train_files: Expected number of files in train directory + expected_val_files: Expected number of files in val directory + """ + train_dir = os.path.join(target_dir, 'train') + val_dir = os.path.join(target_dir, 'val') + + train_files = len([f for f in os.listdir(train_dir) if os.path.isfile(os.path.join(train_dir, f))]) + val_files = len([f for f in os.listdir(val_dir) if os.path.isfile(os.path.join(val_dir, f))]) + + train_xtc = len([f for f in os.listdir(train_dir) if f.endswith('.xtc')]) + train_pdb = len([f for f in os.listdir(train_dir) if f.endswith('.pdb')]) + val_xtc = len([f for f in os.listdir(val_dir) if f.endswith('.xtc')]) + val_pdb = len([f for f in os.listdir(val_dir) if f.endswith('.pdb')]) + + logger.info("=== VERIFICATION RESULTS ===") + logger.info(f"Train directory: {train_files} total files ({train_xtc} .xtc, {train_pdb} .pdb)") + logger.info(f"Val directory: {val_files} total files ({val_xtc} .xtc, {val_pdb} .pdb)") + logger.info(f"Expected train files: {expected_train_files}") + logger.info(f"Expected val files: {expected_val_files}") + + if train_files == expected_train_files and val_files == expected_val_files: + logger.info("āœ… File counts match expectations!") + + # Test mdtraj compatibility + test_mdtraj_compatibility(target_dir) + + else: + logger.warning("āŒ File counts do not match expectations!") + +def reorganize_with_long_grid_split(grid_codes: List[str], strategy_config: dict): + """Reorganize 2000ps data using grid-based splitting strategy.""" + target_dir = strategy_config['target_dir'] + train_size = strategy_config['train_size'] + trajectory_codes = strategy_config['trajectory_codes'] + + logger.info(f"Using long grid split strategy: {strategy_config['description']}") + + if len(grid_codes) < train_size: + logger.error(f"Not enough grid codes found. Expected at least {train_size}, found {len(grid_codes)}") + return + + # Split into train and validation + train_codes, val_codes = split_train_val(grid_codes, train_size) + + # Create target directories + create_target_directories(target_dir) + + # Copy files for train split (using 2000ps trajectories) + logger.info("Copying 2000ps files for train split...") + copy_files_for_grid_split( + SOURCE_DIR, + os.path.join(target_dir, 'train'), + train_codes, + trajectory_codes, + SINGLE_PDB_FILE, + 'TRAIN', + use_2000ps=True + ) + + # Copy files for val split (using 2000ps trajectories) + logger.info("Copying 2000ps files for val split...") + copy_files_for_grid_split( + SOURCE_DIR, + os.path.join(target_dir, 'val'), + val_codes, + trajectory_codes, + SINGLE_PDB_FILE, + 'VAL', + use_2000ps=True + ) + + # Verify output + expected_train_files = len(train_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb + expected_val_files = len(val_codes) * len(trajectory_codes) * 2 + + verify_output(target_dir, expected_train_files, expected_val_files) + +def reorganize_with_grid_split(grid_codes: List[str], strategy_config: dict): + """Reorganize data using grid-based splitting strategy.""" + target_dir = strategy_config['target_dir'] + train_size = strategy_config['train_size'] + + logger.info(f"Using grid split strategy: {strategy_config['description']}") + + if len(grid_codes) < train_size: + logger.error(f"Not enough grid codes found. Expected at least {train_size}, found {len(grid_codes)}") + return + + # Split into train and validation + train_codes, val_codes = split_train_val(grid_codes, train_size) + + # Create target directories + create_target_directories(target_dir) + + # Copy files for train split + logger.info("Copying files for train split...") + copy_files_for_grid_split( + SOURCE_DIR, + os.path.join(target_dir, 'train'), + train_codes, + TRAJECTORY_CODES, + SINGLE_PDB_FILE, + 'TRAIN' + ) + + # Copy files for val split + logger.info("Copying files for val split...") + copy_files_for_grid_split( + SOURCE_DIR, + os.path.join(target_dir, 'val'), + val_codes, + TRAJECTORY_CODES, + SINGLE_PDB_FILE, + 'VAL' + ) + + # Verify output + expected_train_files = len(train_codes) * len(TRAJECTORY_CODES) * 2 # Ɨ2 for .xtc and .pdb + expected_val_files = len(val_codes) * len(TRAJECTORY_CODES) * 2 + + verify_output(target_dir, expected_train_files, expected_val_files) + +def reorganize_with_trajectory_split(grid_codes: List[str], strategy_config: dict): + """Reorganize data using trajectory-based splitting strategy.""" + target_dir = strategy_config['target_dir'] + train_trajectories = strategy_config['train_trajectories'] + val_trajectories = strategy_config['val_trajectories'] + + logger.info(f"Using trajectory split strategy: {strategy_config['description']}") + + # Create target directories + create_target_directories(target_dir) + + # Copy files for train split (all grids, first 4 trajectories) + logger.info("Copying files for train split...") + copy_files_for_trajectory_split( + SOURCE_DIR, + os.path.join(target_dir, 'train'), + grid_codes, + train_trajectories, + SINGLE_PDB_FILE, + 'TRAIN' + ) + + # Copy files for val split (all grids, last trajectory) + logger.info("Copying files for val split...") + copy_files_for_trajectory_split( + SOURCE_DIR, + os.path.join(target_dir, 'val'), + grid_codes, + val_trajectories, + SINGLE_PDB_FILE, + 'VAL' + ) + + # Verify output + expected_train_files = len(grid_codes) * len(train_trajectories) * 2 # Ɨ2 for .xtc and .pdb + expected_val_files = len(grid_codes) * len(val_trajectories) * 2 + + verify_output(target_dir, expected_train_files, expected_val_files) + +def copy_files_for_state_split( + source_dir: str, + target_dir: str, + all_grid_codes: List[str], + trajectory_codes: List[str], + single_pdb_file: str, + phi_range: tuple, + psi_range: tuple +): + """ + Copy and organize files based on conformational state analysis. + + Args: + source_dir: Source swarm results directory + target_dir: Target directory with train/val subdirectories + all_grid_codes: List of all grid codes to process + trajectory_codes: List of trajectory codes to include + single_pdb_file: Path to the single PDB file to copy + phi_range: Tuple of (min, max) for phi angles in degrees + psi_range: Tuple of (min, max) for psi angles in degrees + """ + if not MDTRAJ_AVAILABLE: + logger.error("mdtraj not available, cannot perform state-based splitting") + return + + train_dir = os.path.join(target_dir, 'train') + val_dir = os.path.join(target_dir, 'val') + + total_operations = len(all_grid_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb + + # Create progress bar + if TQDM_AVAILABLE: + pbar = tqdm( + total=total_operations, + desc="Analyzing and copying files", + unit="files", + bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]" + ) + + copied_train = 0 + copied_val = 0 + missing_files = 0 + analysis_errors = 0 + + for grid_code in all_grid_codes: + source_grid_dir = os.path.join(source_dir, f"AA_{grid_code}") + + if not os.path.exists(source_grid_dir): + logger.warning(f"Source directory does not exist: {source_grid_dir}") + # Skip all files for this grid code + if TQDM_AVAILABLE: + pbar.update(len(trajectory_codes) * 2) + continue + + for traj_code in trajectory_codes: + # Handle .xtc file - need to analyze it first + source_xtc = os.path.join(source_grid_dir, f"swarm_2000ps_{traj_code}.xtc") + + if not os.path.exists(source_xtc): + logger.warning(f"Source file does not exist: {source_xtc}") + missing_files += 1 + if TQDM_AVAILABLE: + pbar.update(2) # Skip both .xtc and .pdb + continue + + # Analyze trajectory to determine train/val split + try: + goes_to_val = analyze_trajectory_state(source_xtc, single_pdb_file, phi_range, psi_range) + + if goes_to_val: + target_xtc = os.path.join(val_dir, f"swarm_2000ps_{grid_code}_{traj_code}.xtc") + target_pdb = os.path.join(val_dir, f"swarm_2000ps_{grid_code}_{traj_code}.pdb") + split_name = "VAL" + copied_val += 1 + else: + target_xtc = os.path.join(train_dir, f"swarm_2000ps_{grid_code}_{traj_code}.xtc") + target_pdb = os.path.join(train_dir, f"swarm_2000ps_{grid_code}_{traj_code}.pdb") + split_name = "TRAIN" + copied_train += 1 + + # Copy .xtc file + shutil.copy2(source_xtc, target_xtc) + logger.debug(f"Copied {source_xtc} to {split_name}") + + except Exception as e: + logger.error(f"Failed to analyze trajectory {source_xtc}: {str(e)}") + analysis_errors += 1 + if TQDM_AVAILABLE: + pbar.update(2) # Skip both .xtc and .pdb + continue + + if TQDM_AVAILABLE: + pbar.update(1) + + # Handle .pdb file (copy the single PDB file) + if os.path.exists(single_pdb_file): + shutil.copy2(single_pdb_file, target_pdb) + else: + logger.error(f"Single PDB file does not exist: {single_pdb_file}") + missing_files += 1 + + if TQDM_AVAILABLE: + pbar.update(1) + + if TQDM_AVAILABLE: + pbar.close() + + logger.info(f"State split completed:") + logger.info(f" TRAIN: {copied_train} trajectories") + logger.info(f" VAL: {copied_val} trajectories") + logger.info(f" Missing files: {missing_files}") + logger.info(f" Analysis errors: {analysis_errors}") + +def reorganize_with_state_split(grid_codes: List[str], strategy_config: dict): + """Reorganize data using conformational state-based splitting strategy.""" + target_dir = strategy_config['target_dir'] + trajectory_codes = strategy_config['trajectory_codes'] + phi_range = strategy_config['phi_range'] + psi_range = strategy_config['psi_range'] + + logger.info(f"Using state split strategy: {strategy_config['description']}") + logger.info(f"Target ranges: phi {phi_range}, psi {psi_range}") + logger.info(f"Using trajectory codes: {trajectory_codes}") + + if not MDTRAJ_AVAILABLE: + logger.error("mdtraj not available, cannot perform state-based splitting") + return + + # Create target directories + create_target_directories(target_dir) + + # Copy and split files based on conformational state + logger.info("Analyzing trajectories and copying files...") + copy_files_for_state_split( + SOURCE_DIR, + target_dir, + grid_codes, + trajectory_codes, + SINGLE_PDB_FILE, + phi_range, + psi_range + ) + + # Note: We can't predict exact file counts since they depend on trajectory analysis + logger.info("State-based reorganization completed!") + +def main(strategy: str = 'trajectory_split'): + """ + Main function to reorganize the swarm data. + + Args: + strategy: Either 'grid_split' or 'trajectory_split' + """ + logger.info("Starting swarm data reorganization...") + + # Validate input paths + if not os.path.exists(SOURCE_DIR): + logger.error(f"Source directory does not exist: {SOURCE_DIR}") + return + + if not os.path.exists(SINGLE_PDB_FILE): + logger.error(f"Single PDB file does not exist: {SINGLE_PDB_FILE}") + return + + if strategy not in SPLITTING_STRATEGIES: + logger.error(f"Invalid strategy: {strategy}. Choose from {list(SPLITTING_STRATEGIES.keys())}") + return + + # Get all grid codes + grid_codes = get_all_grid_codes(SOURCE_DIR) + strategy_config = SPLITTING_STRATEGIES[strategy] + + # Execute the appropriate strategy + if strategy == 'grid_split': + reorganize_with_grid_split(grid_codes, strategy_config) + elif strategy == 'trajectory_split': + reorganize_with_trajectory_split(grid_codes, strategy_config) + elif strategy == 'long_grid_split': + reorganize_with_long_grid_split(grid_codes, strategy_config) + elif strategy == 'state_split': + reorganize_with_state_split(grid_codes, strategy_config) + + logger.info("Swarm data reorganization completed!") + +if __name__ == "__main__": + import sys + + # Default to trajectory_split for the new requirement + strategy = 'trajectory_split' + + # Allow command line argument to choose strategy + if len(sys.argv) > 1: + strategy = sys.argv[1] + + print(f"Running reorganization with strategy: {strategy}") + print(f"Description: {SPLITTING_STRATEGIES[strategy]['description']}") + + main(strategy) \ No newline at end of file diff --git a/scratch/test_conditional_denoiser.py b/scratch/test_conditional_denoiser.py new file mode 100644 index 0000000..6a01769 --- /dev/null +++ b/scratch/test_conditional_denoiser.py @@ -0,0 +1,191 @@ +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +from jamun.hydra import instantiate_dict_cfg +import pdb +import jamun +from jamun.utils import compute_average_squared_distance_from_datasets +from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets +from jamun.utils._normalizations import normalization_factors +import math + +def compute_radial_cutoff(max_radius: float, average_squared_distance: float, sigma: float, D: int = 3) -> float: + """ + Compute radial cutoff using the same formula as the denoiser. + + This replicates the computation from denoiser_conditional.py: + radial_cutoff = effective_radial_cutoff(sigma) / c_in + where: + - effective_radial_cutoff = sqrt(max_radiusĀ² + 6ĻƒĀ²) + - c_in = 1.0 / sqrt(average_squared_distance + 2DĻƒĀ²) + + Args: + max_radius: Maximum radius parameter + average_squared_distance: Average squared distance from dataset + sigma: Noise level + D: Dimensionality (default 3 for 3D coordinates) + + Returns: + Computed radial cutoff + """ + # Effective radial cutoff based on noise level + effective_radial_cutoff = math.sqrt(max_radius**2 + 6 * sigma**2) + + # JAMUN normalization factor c_in + A = average_squared_distance + B = 2 * D * sigma**2 + c_in = 1.0 / math.sqrt(A + B) + + # Final radial cutoff + radial_cutoff = effective_radial_cutoff / c_in + + print(f"Radial cutoff computation:") + print(f" max_radius: {max_radius}") + print(f" average_squared_distance: {average_squared_distance}") + print(f" sigma: {sigma}") + print(f" D: {D}") + print(f" effective_radial_cutoff: {effective_radial_cutoff}") + print(f" c_in: {c_in}") + print(f" final radial_cutoff: {radial_cutoff}") + + return radial_cutoff + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + print(f"Added '{project_root}' to sys.path for module discovery.") +else: + print(f"'{project_root}' is already in sys.path.") + +def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: + """Computes the average squared distance for normalization from the data.""" + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + cutoff = cfg.model.max_radius + average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) + return average_squared_distance + +def compute_temporal_average_squared_distance_from_config(cfg: OmegaConf) -> float: + """Computes the temporal average squared distance for normalization from the data.""" + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + + average_squared_distance = compute_temporal_average_squared_distance_from_datasets( + train_datasets, + num_samples=100, # Use reasonable number of samples + verbose=True + ) + return average_squared_distance + + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") +def main(cfg): + # Override configuration to use denoiser_conditional with DenoisingConditioner + # cfg.model._target_ = "jamun.model.denoiser_conditional.Denoiser" + # cfg.model.sigma_distribution._target_ = "jamun.distributions.ConstantSigma" + # cfg.model.sigma_distribution.sigma = 0.04 + breakpoint() + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup('test') + breakpoint() + # Load the test config + average_squared_distance = compute_average_squared_distance_from_config(cfg) + temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) + cfg.model.average_squared_distance = average_squared_distance + + # Compute radial cutoff for spatiotemporal model using the same formula as denoiser + sigma = cfg.model.sigma_distribution.sigma + max_radius = cfg.model.max_radius + spatial_radial_cutoff = compute_radial_cutoff( + max_radius=max_radius, + average_squared_distance=average_squared_distance, # Use temporal for spatiotemporal model + sigma=sigma, + D=3 + ) + temporal_radial_cutoff = compute_radial_cutoff( + max_radius=max_radius, + average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model + sigma=sigma, + D=3 + ) + cfg.model.conditioner.spatiotemporal_model.radial_cutoff = spatial_radial_cutoff + cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff + # Compute c_in using the utility function + sigma = cfg.model.sigma_distribution.sigma + c_in, c_skip, c_out, c_noise = normalization_factors(sigma, average_squared_distance) + c_in_float = float(c_in) + c_noise_float = float(c_noise) + print(f"Computed normalization factors with sigma={sigma}:") + print(f" c_in: {c_in_float}") + print(f" c_skip: {c_skip}") + print(f" c_out: {c_out}") + print(f" c_noise: {c_noise}") + breakpoint() + # Configure DenoisingConditioner with computed c_in + if cfg.model.conditioner._target_ == "jamun.model.conditioners.DenoisedConditioner": + # cfg.model.conditioner.N_structures = 2 # Must match architecture N_structures + cfg.model.conditioner.pretrained_model_path = "sule-shashank/jamun/88i7qkj2" + cfg.model.conditioner.c_in = c_in_float + + if cfg.model.conditioner._target_ == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner": + cfg.model.conditioner.spatiotemporal_model.radial_cutoff = average_squared_distance + max_radius = cfg.model.max_radius + temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) + temporal_radial_cutoff = compute_radial_cutoff( + max_radius=max_radius, + average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model + sigma=sigma, + D=3) + cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff + cfg.model.conditioner.c_noise = c_noise_float + cfg.model.conditioner.c_in = c_in_float + + print("Loading model...") + model = hydra.utils.instantiate(cfg.model) + model.conditioning_module.c_noise = c_noise + print(f"Model loaded: {type(model)}") + print(f"Conditioner: {type(model.conditioning_module)}") + print(f"Sigma: {model.sigma_distribution.sigma}") + # print(f"Conditioner c_in: {model.conditioning_module.c_in}") + breakpoint() + + # Get a single batch + print("Getting a batch of data...") + train_loader = datamodule.train_dataloader() + _, batch = next(enumerate(train_loader)) + + print(f"Batch shape: {batch.pos.shape}") + print(f"Hidden state shape: {[h.shape for h in batch.hidden_state]}") + breakpoint() + + # Test forward pass + print("Testing forward pass...") + with torch.no_grad(): + sigma = model.sigma_distribution.sample() + x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + + print(f"Input shape: {batch.pos.shape}") + print(f"Noisy shape: {y.pos.shape}") + print(f"Output shape: {xhat.pos.shape}") + breakpoint() + + # Test single training step + print("Testing training step...") + loss_output = model.training_step(batch, 0) + print(f"Loss: {loss_output['loss']:.4f}") + breakpoint() + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/test_conditional_sampling.py b/scratch/test_conditional_sampling.py new file mode 100644 index 0000000..b370a93 --- /dev/null +++ b/scratch/test_conditional_sampling.py @@ -0,0 +1,179 @@ +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +from jamun.hydra import instantiate_dict_cfg +import pdb +import jamun +from jamun.data import MDtrajDataModule +from jamun.utils import find_checkpoint +import wandb +from jamun.utils import ModelSamplingWrapperMemory, ModelSamplingWrapper +from tqdm import tqdm + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + print(f"Added '{project_root}' to sys.path for module discovery.") +else: + print(f"'{project_root}' is already in sys.path.") + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample_memory") +def main(cfg): + print("Configuration loaded:") + print(OmegaConf.to_yaml(cfg)) + breakpoint() + + # Load checkpoint using find_checkpoint function + print("Finding checkpoint...") + checkpoint_path = find_checkpoint( + wandb_train_run_path=cfg.get("wandb_train_run_path"), + checkpoint_dir=cfg.get("checkpoint_dir"), + checkpoint_type=cfg.get("checkpoint_type"), + ) + print(f"Checkpoint found at: {checkpoint_path}") + import numpy as np + # cfg.M = 1/6.0 + # cfg.delta = float(cfg.sigma) + # cfg.friction = float(-np.log(np.sqrt(1-4*cfg.M))) + # u = 1/cfg.M + # cfg.inverse_temperature = float(4/(u*(1- np.sqrt(1 - 4/u)))) + print(f"Sampler params: {cfg.M}, {cfg.delta}, {cfg.friction}, {cfg.inverse_temperature}") + breakpoint() + + # Load the model from checkpoint by instantiating it with the checkpoint path + print("Loading model from checkpoint...") + cfg.model.checkpoint_path = checkpoint_path + model = hydra.utils.instantiate(cfg.model) + from e3tools.nn import LayerNorm + model.conditioning_module.spatiotemporal_model.temporal_to_spatial_pooler.layer_norm = LayerNorm(model.conditioning_module.spatiotemporal_model.temporal_module.irreps_out) + model.conditioning_module.spatiotemporal_model.spatial_to_temporal_pooler.layer_norm = LayerNorm(model.conditioning_module.spatiotemporal_model.spatial_module.irreps_out) + print(f"Model loaded: {type(model)}") + breakpoint() + + # Set up initial datasets for sampling + print("Setting up initial datasets...") + init_datasets = hydra.utils.instantiate(cfg.init_datasets) + print(f"Initial datasets loaded: {len(init_datasets)} datasets") + print(f"Dataset types: {[type(ds) for ds in init_datasets]}") + breakpoint() + + # Manually construct the DataModule + print("Creating datamodule for testing...") + datamodule = MDtrajDataModule( + datasets={"train": init_datasets, "val": init_datasets, "test": init_datasets}, + batch_size=1, + num_workers=1 + ) + + datamodule.setup("test") + print("Datamodule setup complete") + breakpoint() + + # Get a sample batch + print("Getting a sample batch...") + test_loader = datamodule.test_dataloader() + batch_idx, batch = next(enumerate(test_loader)) + print(f"Batch shape: {batch.pos.shape}") + print(f"Batch keys: {batch.keys}") + # if hasattr(batch, 'hidden_state') and len(batch.hidden_state) > 0: + # print(f"Hidden state shapes: {[h.shape for h in batch.hidden_state]}") + breakpoint() + + # Set up sampler + print("Setting up sampler...") + sampler = hydra.utils.instantiate(cfg.sampler) + print(f"Sampler created: {type(sampler)}") + breakpoint() + + # set up batch sampler + batch_sampler = hydra.utils.instantiate(cfg.batch_sampler) + print(f"Batch sampler created: {type(batch_sampler)}") + print(f"Batch sampler mcmc: {batch_sampler.mcmc}") + breakpoint() + + # Write test for score + print("Testing score function...") + with torch.no_grad(): + init_graphs = batch + init_graphs = init_graphs.to(sampler.fabric.device) + model_wrapped = ModelSamplingWrapperMemory(model=model, init_graphs=init_graphs, sigma=batch_sampler.sigma, recenter_on_init=True) + y_init = model_wrapped.sample_initial_noisy_positions() + y_hist_init = model_wrapped.sample_initial_noisy_history() + init_score = model_wrapped.score(y_init, y_hist_init, batch_sampler.sigma) + print(f"Initial score: {init_score}") + breakpoint() + + # Test walk + with torch.no_grad(): + y, v, y_hist, y_traj, score_traj, y_hist_traj = batch_sampler.mcmc(y_init, y_hist_init, \ + lambda y, y_hist: model_wrapped.score(y, y_hist, batch_sampler.sigma), \ + v_init="zero", steps=5) + print(f"Score trajectory: {score_traj}") + breakpoint() + + # Test jump + with torch.no_grad(): + xhat_traj = torch.stack( + [ + model_wrapped.xhat(y_traj[i, :], y_hist_traj[i], sigma=batch_sampler.sigma) + for i in tqdm(range(y_traj.size(0)), leave=False, desc="Jump") + ], + dim=0, + ) + print(f"Xhat trajectory: {xhat_traj}") + breakpoint() + + # Test walkjump + with torch.no_grad(): + out = batch_sampler.sample(model_wrapped, y_init=y_init, v_init="zero", y_hist_init=y_hist_init) + print(f"Out: {out}") + breakpoint() + + # Test unbatching + with torch.no_grad(): + samples = model_wrapped.unbatch_samples(out) + print(f"Samples: {samples}") + breakpoint() + + # Test sampling parameters + print("Testing sampling setup...") + print(f"Sigma: {cfg.sigma}") + print(f"M: {cfg.M}") + print(f"Delta: {cfg.delta}") + print(f"Friction: {cfg.friction}") + print(f"Number of sampling steps per batch: {cfg.num_sampling_steps_per_batch}") + print(f"Number of batches: {cfg.num_batches}") + breakpoint() + + # Test a forward pass with the model + print("Testing model forward pass...") + model.eval() + with torch.no_grad(): + # Test if the model can process the batch + if hasattr(model, 'noise_and_denoise'): + sigma_tensor = torch.tensor([cfg.sigma]) + x_target, xhat, y = model.noise_and_denoise(batch, sigma_tensor, align_noisy_input=True) + print(f"Forward pass successful!") + print(f"Input shape: {batch.pos.shape}") + print(f"Noisy shape: {y.pos.shape}") + print(f"Denoised shape: {xhat.pos.shape}") + else: + print("Model doesn't have noise_and_denoise method, testing direct forward pass") + output = model(batch) + print(f"Model output: {type(output)}") + breakpoint() + + print("Script completed successfully!") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/test_conditional_simple.py b/scratch/test_conditional_simple.py new file mode 100644 index 0000000..d2b579e --- /dev/null +++ b/scratch/test_conditional_simple.py @@ -0,0 +1,150 @@ +#!/usr/bin/env python3 +""" +Simple test script for the debugged denoiser_conditional using default hydra config. +Tests with sigma = 0.0 and sigma = 0.1. +""" + +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +from jamun.utils import compute_average_squared_distance_from_datasets + +breakpoint() # Start debugging + +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" +if project_root not in sys.path: + sys.path.insert(0, project_root) + print(f"Added '{project_root}' to sys.path for module discovery.") + +breakpoint() # After setup + +def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: + """Computes the average squared distance for normalization from the data.""" + breakpoint() # Start of function + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + cutoff = cfg.model.max_radius + average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) + breakpoint() # After computation + return average_squared_distance + + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") +def main(cfg): + breakpoint() # Start of main + print("=" * 60) + print("Testing debugged denoiser_conditional") + print("=" * 60) + + # Compute average squared distance + print("Computing average squared distance...") + breakpoint() # Before distance computation + average_squared_distance = compute_average_squared_distance_from_config(cfg) + cfg.model.average_squared_distance = average_squared_distance + print(f"Average squared distance: {average_squared_distance:.6f}") + + # Load datamodule + print("Loading datamodule...") + breakpoint() # Before datamodule + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup('test') + + # Load model + print("Loading model...") + breakpoint() # Before model loading + model = hydra.utils.instantiate(cfg.model) + + # Get a single batch + print("Getting a batch of data...") + breakpoint() # Before getting batch + train_loader = datamodule.train_dataloader() + _, batch = next(enumerate(train_loader)) + + print(f"Batch info:") + print(f" Position shape: {batch.pos.shape}") + print(f" Number of atoms: {batch.pos.shape[0]}") + print(f" Hidden state shapes: {[h.shape for h in batch.hidden_state]}") + print(f" Number of hidden states: {len(batch.hidden_state)}") + + breakpoint() # After batch info + + # Test with sigma = 0.0 + print("\n" + "=" * 40) + print("Testing with sigma = 0.0 (no noise)") + print("=" * 40) + + breakpoint() # Before sigma=0.0 test + + with torch.no_grad(): + breakpoint() # Before noise_and_denoise + sigma = torch.tensor(0.0) + x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + + print(f"Input shape: {batch.pos.shape}") + print(f"Noisy shape: {y.pos.shape}") + print(f"Output shape: {xhat.pos.shape}") + + breakpoint() # After noise_and_denoise + + # Compute loss + loss, aux = model.compute_loss(x_target, xhat, sigma) + print(f"Loss: {loss.mean().item():.6f}") + print(f"Metrics: {aux}") + + # Check if positions are preserved (should be identical with sigma=0) + pos_diff = torch.abs(batch.pos - y.pos).max() + print(f"Max position difference (sigma=0): {pos_diff.item():.8f}") + + breakpoint() # After sigma=0.0 test + + # Test with sigma = 0.1 + print("\n" + "=" * 40) + print("Testing with sigma = 0.1 (with noise)") + print("=" * 40) + + breakpoint() # Before sigma=0.1 test + + with torch.no_grad(): + sigma = torch.tensor(0.1) + breakpoint() # Before noise_and_denoise with sigma=0.1 + x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + + print(f"Input shape: {batch.pos.shape}") + print(f"Noisy shape: {y.pos.shape}") + print(f"Output shape: {xhat.pos.shape}") + + # Compute loss + loss, aux = model.compute_loss(x_target, xhat, sigma) + print(f"Loss: {loss.mean().item():.6f}") + print(f"Metrics: {aux}") + + # Check noise level + pos_diff = torch.abs(batch.pos - y.pos).max() + print(f"Max position difference (sigma=0.1): {pos_diff.item():.6f}") + + # Check denoising quality + denoise_diff = torch.abs(batch.pos - xhat.pos).max() + print(f"Max denoising difference: {denoise_diff.item():.6f}") + + breakpoint() # After sigma=0.1 test + + print("\n" + "=" * 60) + print("Testing complete!") + print("=" * 60) + + breakpoint() # End of main + + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/test_conditioners.py b/scratch/test_conditioners.py new file mode 100644 index 0000000..9a97d38 --- /dev/null +++ b/scratch/test_conditioners.py @@ -0,0 +1,396 @@ +#!/usr/bin/env python3 +""" +Test script for SelfConditioner, PositionConditioner, and MeanConditioner with both +MDtrajDataset and RepeatedPositionDataset. +""" + +import torch +import torch_geometric +import os +from pathlib import Path + +# Add the src directory to the path so we can import jamun modules +import sys +sys.path.insert(0, str(Path(__file__).parent / "src")) + +from jamun.data._mdtraj import MDtrajDataset +from jamun.data.noisy_position_dataset import RepeatedPositionDataset +from jamun.model.conditioners.conditioners import SelfConditioner, PositionConditioner, MeanConditioner + +def print_tensor_summary(tensor, name, max_elements=6): + """Print a summary of a tensor with first few elements.""" + if tensor.numel() <= max_elements: + print(f"{name}: {tensor.flatten().tolist()}") + else: + flat = tensor.flatten() + print(f"{name} (shape {tensor.shape}): [{flat[0]:.6f}, {flat[1]:.6f}, {flat[2]:.6f}, ..., {flat[-3]:.6f}, {flat[-2]:.6f}, {flat[-1]:.6f}]") + +def create_datasets(): + """Create both types of datasets with 3 total structures (2 hidden states).""" + + root = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train" + traj_files = ["ALA_ALA.xtc"] + pdb_file = "ALA_ALA.pdb" + total_lag_time = 3 # This should create 2 hidden states (3 - 1 = 2) + + print(f"Creating datasets with total_lag_time={total_lag_time} (expecting 2 hidden states)...") + + # Create MDtrajDataset (with real lag processing) + mdtraj_dataset = MDtrajDataset( + root=root, + traj_files=traj_files, + pdb_file=pdb_file, + label="ALA_ALA_mdtraj", + total_lag_time=total_lag_time, + lag_subsample_rate=1, + num_frames=10, + verbose=True + ) + + # Create RepeatedPositionDataset (with position copies) + repeated_dataset = RepeatedPositionDataset( + root=root, + traj_files=traj_files, + pdb_file=pdb_file, + label="ALA_ALA_repeated", + total_lag_time=total_lag_time, + lag_subsample_rate=1, + num_frames=10, + verbose=True + ) + + return mdtraj_dataset, repeated_dataset + +def create_batch_from_dataset(dataset, sample_idx=0): + """Create a batched graph from a single dataset sample.""" + graph = dataset[sample_idx] + batch = torch_geometric.data.Batch.from_data_list([graph]) + return batch + +def print_batch_details(batch, batch_name): + """Print detailed information about a batch.""" + print(f"\n--- {batch_name} Details ---") + print(f"Position shape: {batch.pos.shape}") + print_tensor_summary(batch.pos, "Position") + + # Check if position is mean centered + pos_mean = torch.mean(batch.pos, dim=0) # Mean over atoms + pos_mean_magnitude = torch.norm(pos_mean).item() + print(f"Position mean: [{pos_mean[0]:.6f}, {pos_mean[1]:.6f}, {pos_mean[2]:.6f}]") + print(f"Position mean magnitude: {pos_mean_magnitude:.6f}") + if pos_mean_magnitude < 1e-6: + print(f"āœ… Input position is mean centered") + else: + print(f"āŒ Input position is NOT mean centered") + + print(f"Number of hidden states: {len(batch.hidden_state)}") + for i, hidden_state in enumerate(batch.hidden_state): + print(f"Hidden state {i} shape: {hidden_state.shape}") + print_tensor_summary(hidden_state, f"Hidden state {i}") + + # Check if hidden state is mean centered + hidden_mean = torch.mean(hidden_state, dim=0) # Mean over atoms + hidden_mean_magnitude = torch.norm(hidden_mean).item() + print(f"Hidden state {i} mean: [{hidden_mean[0]:.6f}, {hidden_mean[1]:.6f}, {hidden_mean[2]:.6f}]") + print(f"Hidden state {i} mean magnitude: {hidden_mean_magnitude:.6f}") + if hidden_mean_magnitude < 1e-6: + print(f"āœ… Hidden state {i} is mean centered") + else: + print(f"āŒ Hidden state {i} is NOT mean centered") + +def test_conditioner_detailed(conditioner, batch, test_name): + """Test a conditioner with detailed output.""" + print(f"\n{'='*70}") + print(f"{test_name}") + print(f"{'='*70}") + + # Print input details + print_batch_details(batch, "Input Batch") + + # Run the conditioner + try: + print(f"\nRunning {conditioner.__class__.__name__}...") + conditioned_structures = conditioner(batch) + + print(f"\n--- Conditioner Output ---") + print(f"Number of conditioned structures: {len(conditioned_structures)}") + + # Print each conditioned structure + for i, structure in enumerate(conditioned_structures): + print(f"\nConditioned structure {i} shape: {structure.shape}") + print_tensor_summary(structure, f"Conditioned structure {i}") + + # Compare with input position + pos_diff = torch.max(torch.abs(structure - batch.pos)).item() + print(f"Max difference from current position: {pos_diff:.10f}") + + # Compare with hidden states if available + if i < len(batch.hidden_state): + hidden_diff = torch.max(torch.abs(structure - batch.hidden_state[i])).item() + print(f"Max difference from hidden state {i}: {hidden_diff:.10f}") + + # Check if structure is mean centered (for PositionConditioner) + if conditioner.__class__.__name__ == "PositionConditioner": + structure_mean = torch.mean(structure, dim=0) # Mean over atoms + mean_magnitude = torch.norm(structure_mean).item() + print(f"Mean of structure {i}: [{structure_mean[0]:.6f}, {structure_mean[1]:.6f}, {structure_mean[2]:.6f}]") + print(f"Magnitude of mean: {mean_magnitude:.6f}") + + # Check if it's close to zero (mean centered) + if mean_magnitude < 1e-6: + print(f"āœ… Structure {i} is mean centered (mean ā‰ˆ 0)") + else: + print(f"āŒ Structure {i} is NOT mean centered") + + # Check if structure contains means across time steps (for MeanConditioner) + if conditioner.__class__.__name__ == "MeanConditioner": + # For MeanConditioner, each structure should be the mean across time steps + # All structures should be identical, but atoms can have different coordinates + print(f"āœ… Structure {i} contains mean across time steps") + + # Check if this structure is the same as the first structure (all should be the same mean) + if i > 0: + first_structure = conditioned_structures[0] + structures_same = torch.allclose(structure, first_structure, atol=1e-6) + if structures_same: + print(f"āœ… Structure {i} matches structure 0 (all structures are identical means)") + else: + print(f"āŒ Structure {i} doesn't match structure 0 (all should be identical)") + max_diff = torch.max(torch.abs(structure - first_structure)).item() + print(f"Maximum difference: {max_diff:.10f}") + + # Verify the mean computation is correct by manually computing it + if hasattr(batch, "hidden_state") and batch.hidden_state is not None: + all_positions = [batch.pos] + batch.hidden_state + manual_mean = torch.mean(torch.stack(all_positions, dim=0), dim=0) + mean_correct = torch.allclose(structure, manual_mean, atol=1e-6) + if mean_correct: + print(f"āœ… Structure {i} correctly computed as mean across {len(all_positions)} time steps") + else: + print(f"āŒ Structure {i} mean computation incorrect") + max_diff = torch.max(torch.abs(structure - manual_mean)).item() + print(f"Maximum difference from expected mean: {max_diff:.10f}") + else: + # If no hidden states, should just be y.pos repeated + pos_same = torch.allclose(structure, batch.pos, atol=1e-6) + if pos_same: + print(f"āœ… Structure {i} correctly equals y.pos (no hidden states)") + else: + print(f"āŒ Structure {i} should equal y.pos when no hidden states") + + return conditioned_structures + + except Exception as e: + print(f"āŒ ERROR: Exception in {test_name}: {e}") + import traceback + traceback.print_exc() + return False + +def verify_results(conditioned_structures, batch, test_name, expected_behavior): + """Verify the results match expected behavior.""" + print(f"\n--- Verification for {test_name} ---") + print(f"Expected behavior: {expected_behavior}") + + expected_count = 3 # N_structures = 3, so we expect 3 total structures including current position + if len(conditioned_structures) != expected_count: + print(f"āŒ ERROR: Expected {expected_count} structures, got {len(conditioned_structures)}") + return False + + print(f"āœ… Correct count: {len(conditioned_structures)} structures") + + success = True + for i, structure in enumerate(conditioned_structures): + if structure.shape != batch.pos.shape: + print(f"āŒ ERROR: Structure {i} shape mismatch!") + success = False + else: + print(f"āœ… Structure {i} has correct shape") + + # Verify that the first structure is the current position for most conditioners + # Exception: MeanConditioner returns time-averaged means, not current position + first_structure = conditioned_structures[0] + pos_diff = torch.max(torch.abs(first_structure - batch.pos)).item() + + if test_name.startswith("MeanConditioner"): + # For MeanConditioner, first structure should be the time-averaged mean, not y.pos + if hasattr(batch, "hidden_state") and batch.hidden_state is not None: + all_positions = [batch.pos] + batch.hidden_state + expected_mean = torch.mean(torch.stack(all_positions, dim=0), dim=0) + mean_diff = torch.max(torch.abs(first_structure - expected_mean)).item() + if mean_diff < 1e-6: + print(f"āœ… First structure correctly equals time-averaged mean (diff: {mean_diff:.2e})") + else: + print(f"āŒ ERROR: First structure doesn't match expected time-averaged mean (diff: {mean_diff:.2e})") + success = False + else: + # If no hidden states, should equal y.pos + if pos_diff < 1e-10: + print(f"āœ… First structure correctly equals y.pos (no hidden states, diff: {pos_diff:.2e})") + else: + print(f"āŒ ERROR: First structure doesn't match y.pos when no hidden states (diff: {pos_diff:.2e})") + success = False + else: + # For other conditioners, first structure should be y.pos + if pos_diff < 1e-10: + print(f"āœ… First structure matches current position (diff: {pos_diff:.2e})") + else: + print(f"āŒ ERROR: First structure doesn't match current position (diff: {pos_diff:.2e})") + success = False + + return success + +def main(): + """Main test function.""" + print("Testing Conditioners: SelfConditioner, PositionConditioner, MeanConditioner with 3 total structures (2 hidden states)") + print("=" * 70) + + # Create datasets + try: + mdtraj_dataset, repeated_dataset = create_datasets() + print(f"āœ… Created datasets") + print(f" MDtrajDataset length: {len(mdtraj_dataset)}") + print(f" RepeatedPositionDataset length: {len(repeated_dataset)}") + except Exception as e: + print(f"āŒ ERROR: Failed to create datasets: {e}") + return False + + # Create batches + try: + mdtraj_batch = create_batch_from_dataset(mdtraj_dataset, sample_idx=0) + repeated_batch = create_batch_from_dataset(repeated_dataset, sample_idx=0) + print(f"āœ… Created batches") + except Exception as e: + print(f"āŒ ERROR: Failed to create batches: {e}") + return False + + # Create conditioners + try: + self_conditioner = SelfConditioner(N_structures=3) + position_conditioner = PositionConditioner(N_structures=3) + mean_conditioner = MeanConditioner(N_structures=3) + print(f"āœ… Created conditioners") + except Exception as e: + print(f"āŒ ERROR: Failed to create conditioners: {e}") + return False + + # Test 1: SelfConditioner on MDtrajDataset + result1 = test_conditioner_detailed( + self_conditioner, + mdtraj_batch, + "TEST 1: SelfConditioner on MDtrajDataset" + ) + if result1 is False: + return False + success1 = verify_results( + result1, + mdtraj_batch, + "SelfConditioner + MDtrajDataset", + "Should return [y.pos, y.pos, y.pos] - 3 copies of current position" + ) + + # Test 2: SelfConditioner on RepeatedPositionDataset + result2 = test_conditioner_detailed( + self_conditioner, + repeated_batch, + "TEST 2: SelfConditioner on RepeatedPositionDataset" + ) + if result2 is False: + return False + success2 = verify_results( + result2, + repeated_batch, + "SelfConditioner + RepeatedPositionDataset", + "Should return [y.pos, y.pos, y.pos] - 3 copies of current position" + ) + + # Test 3: PositionConditioner on MDtrajDataset + result3 = test_conditioner_detailed( + position_conditioner, + mdtraj_batch, + "TEST 3: PositionConditioner on MDtrajDataset" + ) + if result3 is False: + return False + success3 = verify_results( + result3, + mdtraj_batch, + "PositionConditioner + MDtrajDataset", + "Should return [y.pos, aligned_hidden_state_1, aligned_hidden_state_2] - current position + 2 aligned hidden states" + ) + + # Test 4: PositionConditioner on RepeatedPositionDataset + result4 = test_conditioner_detailed( + position_conditioner, + repeated_batch, + "TEST 4: PositionConditioner on RepeatedPositionDataset" + ) + if result4 is False: + return False + success4 = verify_results( + result4, + repeated_batch, + "PositionConditioner + RepeatedPositionDataset", + "Should return [y.pos, aligned_copy_1, aligned_copy_2] - current position + 2 aligned copies" + ) + + # Test 5: MeanConditioner on MDtrajDataset + result5 = test_conditioner_detailed( + mean_conditioner, + mdtraj_batch, + "TEST 5: MeanConditioner on MDtrajDataset" + ) + if result5 is False: + return False + success5 = verify_results( + result5, + mdtraj_batch, + "MeanConditioner + MDtrajDataset", + "Should return [time_mean, time_mean, time_mean] - 3 copies of mean across time steps (y.pos + hidden states)" + ) + + # Test 6: MeanConditioner on RepeatedPositionDataset + result6 = test_conditioner_detailed( + mean_conditioner, + repeated_batch, + "TEST 6: MeanConditioner on RepeatedPositionDataset" + ) + if result6 is False: + return False + success6 = verify_results( + result6, + repeated_batch, + "MeanConditioner + RepeatedPositionDataset", + "Should return [time_mean, time_mean, time_mean] - 3 copies of mean across time steps (y.pos + hidden states)" + ) + + # Summary + print(f"\n{'='*70}") + print("SUMMARY") + print(f"{'='*70}") + + tests = [ + ("SelfConditioner + MDtrajDataset", success1), + ("SelfConditioner + RepeatedPositionDataset", success2), + ("PositionConditioner + MDtrajDataset", success3), + ("PositionConditioner + RepeatedPositionDataset", success4), + ("MeanConditioner + MDtrajDataset", success5), + ("MeanConditioner + RepeatedPositionDataset", success6) + ] + + all_passed = True + for test_name, success in tests: + status = "āœ… PASS" if success else "āŒ FAIL" + print(f"{test_name}: {status}") + if not success: + all_passed = False + + if all_passed: + print(f"\nšŸŽ‰ All conditioner tests passed!") + return True + else: + print(f"\nšŸ’„ Some conditioner tests failed!") + return False + +if __name__ == "__main__": + success = main() + exit(0 if success else 1) \ No newline at end of file diff --git a/scratch/test_denoised_conditioner.py b/scratch/test_denoised_conditioner.py new file mode 100644 index 0000000..6489c03 --- /dev/null +++ b/scratch/test_denoised_conditioner.py @@ -0,0 +1,403 @@ +#!/usr/bin/env python3 +""" +Comprehensive test for DenoisedConditioner using real ALA_ALA data. +This test emulates the behavior of xhat_normalized to properly test the scaling parameter. +""" +import torch +import torch_geometric +import numpy as np +import os +import e3nn +from jamun.model.conditioners import DenoisedConditioner +from jamun.data import parse_datasets_from_directory +from jamun.utils import unsqueeze_trailing, mean_center +from jamun.utils._normalizations import normalization_factors + +# Fix e3nn optimization for avoiding script issues +e3nn.set_optimization_defaults(jit_script_fx=False) + +def load_ala_ala_data(): + """Load actual ALA_ALA data from the capped diamines dataset.""" + + # Get data path from environment variable + data_path = os.getenv("JAMUN_DATA_PATH") + if not data_path: + raise ValueError("JAMUN_DATA_PATH environment variable not set") + + # Load ALA_ALA dataset + datasets = parse_datasets_from_directory( + root=f"{data_path}/capped_diamines/timewarp_splits/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + filter_codes=['ALA_ALA'], + as_iterable=False, + subsample=1, + total_lag_time=3, # This will give us hidden states + lag_subsample_rate=1, + num_frames=10, + max_datasets=1 + ) + + if not datasets: + raise ValueError("No ALA_ALA datasets found") + + # Get the first dataset + dataset = datasets[0] + print(f"Loaded ALA_ALA dataset with {len(dataset)} frames") + + # Get a few samples to create a batch + samples = [] + for i in range(min(2, len(dataset))): # Get 2 samples for batch + sample = dataset[i] + samples.append(sample) + + # Create batch + batch = torch_geometric.data.Batch.from_data_list(samples) + + print(f"Created batch with {batch.num_graphs} graphs") + print(f"Batch position shape: {batch.pos.shape}") + + if hasattr(batch, 'hidden_state') and batch.hidden_state: + print(f"Hidden states: {len(batch.hidden_state)} states") + for i, hidden_state in enumerate(batch.hidden_state): + print(f" Hidden state {i}: shape {hidden_state.shape}") + else: + print("No hidden states found") + + return batch + +def add_noise_to_batch(x: torch_geometric.data.Batch, sigma: float) -> torch_geometric.data.Batch: + """Add noise to a batch, similar to the denoiser's add_noise method.""" + sigma = unsqueeze_trailing(torch.tensor(sigma), x.pos.ndim) + + y = x.clone() + + # Add noise to positions + noise = torch.randn_like(x.pos) + y.pos = x.pos + sigma * noise + + # Add noise to hidden states if they exist + if hasattr(x, "hidden_state") and x.hidden_state is not None: + y.hidden_state = [] + for hidden_positions in x.hidden_state: + hidden_noise = torch.randn_like(hidden_positions) + y.hidden_state.append(hidden_positions + sigma * hidden_noise) + + return y + +def mean_center_positions(batch: torch_geometric.data.Batch) -> torch_geometric.data.Batch: + """Mean-center positions and hidden states for each graph in the batch.""" + from e3tools import scatter + from jamun.utils import mean_center + + # Mean-center the main positions using the jamun utils function + batch = mean_center(batch) + + # Mean-center each hidden state individually + if hasattr(batch, "hidden_state") and batch.hidden_state is not None: + for i, hidden_positions in enumerate(batch.hidden_state): + # Create a temporary batch with just the hidden state positions to mean-center + temp_batch = batch.clone() + temp_batch.pos = hidden_positions + temp_batch_centered = mean_center(temp_batch) + batch.hidden_state[i] = temp_batch_centered.pos + + return batch + +def emulate_xhat_normalized_scaling(batch, sigma: float, average_squared_distance: float = 0.332): + """ + Emulate the scaling behavior in xhat_normalized method. + This simulates how the denoiser scales data before passing to conditioner. + """ + print(f"\n=== Emulating xhat_normalized scaling ===") + print(f"Input sigma: {sigma}") + print(f"Average squared distance: {average_squared_distance}") + + # Mean-center the batch positions and hidden states (as done in actual xhat_normalized) + batch = mean_center_positions(batch) + + # Compute normalization factors (same as in denoiser) + c_in, c_skip, c_out, c_noise = normalization_factors(sigma, average_squared_distance) + + print(f"Normalization factors:") + print(f" c_in: {c_in}") + print(f" c_skip: {c_skip}") + print(f" c_out: {c_out}") + print(f" c_noise: {c_noise}") + + # Adjust dimensions (same as in denoiser) + c_in = unsqueeze_trailing(c_in, batch.pos.ndim - 1) + c_skip = unsqueeze_trailing(c_skip, batch.pos.ndim - 1) + c_out = unsqueeze_trailing(c_out, batch.pos.ndim - 1) + c_noise = c_noise.unsqueeze(0) + + # Scale the batch (same as in denoiser) + y_scaled = batch.clone() + y_scaled.pos = batch.pos * c_in + + print(f"Original position mean: {batch.pos.mean():.6f}") + print(f"Scaled position mean: {y_scaled.pos.mean():.6f}") + + # Scale hidden states (same as in denoiser) + if hasattr(batch, "hidden_state") and batch.hidden_state is not None: + y_scaled.hidden_state = [] + for i, positions in enumerate(batch.hidden_state): + scaled_positions = positions * c_in + y_scaled.hidden_state.append(scaled_positions) + print(f"Hidden state {i} - Original mean: {positions.mean():.6f}, Scaled mean: {scaled_positions.mean():.6f}") + + return y_scaled, c_in, c_skip, c_out, c_noise + +def test_denoised_conditioner_with_scaling(): + """Test the DenoisedConditioner with proper scaling emulation.""" + + print("=== Testing DenoisedConditioner with xhat_normalized scaling ===") + + # Test parameters + N_structures = 3 # Must match architecture N_structures (updated to match hidden states) + pretrained_model_path = "sule-shashank/jamun/370wpt17" # Update this to your desired checkpoint + test_sigma = 0.04 + + try: + # Load real ALA_ALA data + print("\n1. Loading real ALA_ALA data...") + original_batch = load_ala_ala_data() + print("āœ“ Real ALA_ALA data loaded successfully") + + # Mean-center the original batch (clean reference) - positions and hidden states + print("\n2. Mean-centering the data...") + print(f" Original position mean: {original_batch.pos.mean():.6f}") + if hasattr(original_batch, "hidden_state") and original_batch.hidden_state: + for i, hidden_state in enumerate(original_batch.hidden_state): + print(f" Original hidden state {i} mean: {hidden_state.mean():.6f}") + + x_clean = mean_center_positions(original_batch) + print(f" Mean-centered position mean: {x_clean.pos.mean():.6f}") + + if hasattr(x_clean, "hidden_state") and x_clean.hidden_state: + for i, hidden_state in enumerate(x_clean.hidden_state): + print(f" Mean-centered hidden state {i} mean: {hidden_state.mean():.6f}") + + # Add noise to the mean-centered data + print(f"\n3. Adding noise with sigma={test_sigma}...") + y_noisy = add_noise_to_batch(x_clean, test_sigma) + print(f" Noisy position mean: {y_noisy.pos.mean():.6f}") + print(f" Noisy position std: {y_noisy.pos.std():.6f}") + + if hasattr(y_noisy, "hidden_state") and y_noisy.hidden_state: + for i, hidden_state in enumerate(y_noisy.hidden_state): + print(f" Noisy hidden state {i} mean: {hidden_state.mean():.6f}") + print(f" Noisy hidden state {i} std: {hidden_state.std():.6f}") + + # Initialize conditioner and extract average_squared_distance from checkpoint + print(f"\n4. Initializing DenoisedConditioner and extracting average_squared_distance...") + print(f" N_structures: {N_structures}") + print(f" pretrained_model_path: {pretrained_model_path}") + + # Use a temporary c_in for initialization + temp_c_in, _, _, _ = normalization_factors(test_sigma, 0.332) # temporary default + temp_c_in_float = float(temp_c_in) + + # Initialize conditioner + conditioner = DenoisedConditioner( + N_structures=N_structures, + pretrained_model_path=pretrained_model_path, + c_in=temp_c_in_float + ) + + print("āœ“ DenoisedConditioner initialized successfully") + print(f" Denoiser sigma: {conditioner.denoiser_sigma}") + + # Extract average_squared_distance from the loaded checkpoint + average_squared_distance = None + if hasattr(conditioner.pretrained_denoiser, 'average_squared_distance'): + average_squared_distance = float(conditioner.pretrained_denoiser.average_squared_distance) + print(f" āœ“ Extracted average_squared_distance from checkpoint: {average_squared_distance}") + elif hasattr(conditioner.pretrained_denoiser, 'hparams') and hasattr(conditioner.pretrained_denoiser.hparams, 'average_squared_distance'): + average_squared_distance = float(conditioner.pretrained_denoiser.hparams.average_squared_distance) + print(f" āœ“ Extracted average_squared_distance from hparams: {average_squared_distance}") + else: + # Try to extract from the config if available + if hasattr(conditioner.pretrained_denoiser, 'cfg'): + cfg = conditioner.pretrained_denoiser.cfg + if hasattr(cfg, 'average_squared_distance'): + average_squared_distance = float(cfg.average_squared_distance) + print(f" āœ“ Extracted average_squared_distance from config: {average_squared_distance}") + + if average_squared_distance is None: + print(" āš ļø Could not extract average_squared_distance from checkpoint") + print(" Available attributes on pretrained_denoiser:") + for attr in dir(conditioner.pretrained_denoiser): + if not attr.startswith('_'): + print(f" - {attr}") + # Use default + average_squared_distance = 0.332 + print(f" Using default average_squared_distance: {average_squared_distance}") + + # Recompute c_in with the correct average_squared_distance + c_in, _, _, _ = normalization_factors(test_sigma, average_squared_distance) + c_in_float = float(c_in) + + # Update the conditioner's c_in + if abs(c_in_float - temp_c_in_float) > 1e-6: + print(f" Updating c_in from {temp_c_in_float} to {c_in_float}") + conditioner.c_in = c_in_float + else: + print(f" c_in remains: {c_in_float}") + + # Test sigma consistency + print(f"\n5. Testing sigma consistency...") + if abs(conditioner.denoiser_sigma - test_sigma) < 1e-5: + print(f"āœ“ Test sigma ({test_sigma}) matches denoiser sigma ({conditioner.denoiser_sigma})") + else: + print(f"āš ļø Test sigma ({test_sigma}) differs from denoiser sigma ({conditioner.denoiser_sigma})") + print(" Using denoiser sigma for consistency") + test_sigma = conditioner.denoiser_sigma + # Recompute c_in with corrected sigma + c_in, _, _, _ = normalization_factors(test_sigma, average_squared_distance) + c_in_float = float(c_in) + conditioner.c_in = c_in_float + print(f" Updated c_in to: {c_in_float}") + + # Emulate xhat_normalized scaling on the noisy batch + print(f"\n6. Emulating xhat_normalized scaling...") + scaled_batch, c_in_tensor, c_skip, c_out, c_noise = emulate_xhat_normalized_scaling( + y_noisy, test_sigma, average_squared_distance + ) + + # Verify our c_in calculation matches + assert abs(float(c_in_tensor) - c_in_float) < 1e-6, f"c_in mismatch: {c_in_tensor} vs {c_in_float}" + print("āœ“ c_in calculation verified") + + # Test conditioner with scaled noisy data + print(f"\n7. Testing conditioner with scaled noisy data...") + + # Move scaled_batch to the same device as the conditioner + device = next(conditioner.parameters()).device + scaled_batch = scaled_batch.to(device) + x_clean = x_clean.to(device) + y_noisy = y_noisy.to(device) + print(f" Moved batches to device: {device}") + + conditioned_structures = conditioner.forward(scaled_batch) + + print(f"āœ“ Conditioner forward pass completed") + print(f" Returned {len(conditioned_structures)} structures") + print(f" Expected N_structures: {N_structures}") + + # Verify output structure + assert len(conditioned_structures) == N_structures, f"Expected {N_structures} structures, got {len(conditioned_structures)}" + + for i, structure in enumerate(conditioned_structures): + assert structure.shape == scaled_batch.pos.shape, f"Structure {i} has wrong shape: {structure.shape} vs {scaled_batch.pos.shape}" + print(f" Structure {i}: shape {structure.shape}") + + # Check that first structure is the scaled current position + assert torch.allclose(conditioned_structures[0], scaled_batch.pos), "First structure should be scaled current position" + print("āœ“ First structure matches scaled current position") + + # Comprehensive denoising quality test + print(f"\n8. COMPREHENSIVE DENOISING QUALITY TEST...") + denoising_improvements = [] + + if hasattr(x_clean, "hidden_state") and x_clean.hidden_state and len(conditioned_structures) > 1: + print(f" Testing denoising on {len(x_clean.hidden_state)} hidden states...") + + for i in range(1, len(conditioned_structures)): # Skip first structure (current position) + hidden_idx = i - 1 # Map to hidden state index + if hidden_idx < len(x_clean.hidden_state): + denoised_structure = conditioned_structures[i] + clean_hidden = x_clean.hidden_state[hidden_idx] + noisy_hidden = y_noisy.hidden_state[hidden_idx] + + # Calculate RMSE between denoised and clean + denoised_rmse = torch.sqrt(torch.mean((denoised_structure - clean_hidden)**2)) + + # Calculate RMSE between noisy and clean for comparison + noisy_rmse = torch.sqrt(torch.mean((noisy_hidden - clean_hidden)**2)) + + # Calculate improvement + improvement = noisy_rmse - denoised_rmse + improvement_percent = (improvement / noisy_rmse) * 100 + + print(f" Hidden State {hidden_idx}:") + print(f" Noisy RMSE vs clean: {noisy_rmse.item():.6f}") + print(f" Denoised RMSE vs clean: {denoised_rmse.item():.6f}") + print(f" Improvement: {improvement.item():.6f} ({improvement_percent.item():.2f}%)") + + denoising_improvements.append(improvement.item()) + + if improvement > 0: + print(f" āœ“ DENOISING SUCCESSFUL (RMSE reduced)") + else: + print(f" āŒ DENOISING FAILED (RMSE increased)") + + # Verify denoised is different from both noisy and original + assert not torch.allclose(denoised_structure, noisy_hidden, atol=1e-4), f"Denoised structure {i} should be different from noisy" + assert not torch.allclose(denoised_structure, scaled_batch.pos, atol=1e-4), f"Denoised structure {i} should be different from current position" + + # Overall denoising assessment + print(f"\n9. OVERALL DENOISING ASSESSMENT...") + if denoising_improvements: + avg_improvement = np.mean(denoising_improvements) + successful_denoising = sum(1 for imp in denoising_improvements if imp > 0) + total_tests = len(denoising_improvements) + success_rate = (successful_denoising / total_tests) * 100 + + print(f" Total hidden states tested: {total_tests}") + print(f" Successful denoising: {successful_denoising}/{total_tests} ({success_rate:.1f}%)") + print(f" Average RMSE improvement: {avg_improvement:.6f}") + + if success_rate >= 80: # Require at least 80% success rate + print(f" āœ… DENOISING QUALITY: EXCELLENT (ā‰„80% success)") + elif success_rate >= 60: + print(f" āš ļø DENOISING QUALITY: GOOD (ā‰„60% success)") + elif success_rate >= 40: + print(f" āš ļø DENOISING QUALITY: MODERATE (ā‰„40% success)") + else: + print(f" āŒ DENOISING QUALITY: POOR (<40% success)") + + # Assert that at least some denoising occurred + assert successful_denoising > 0, "At least one hidden state should show denoising improvement" + print(f" āœ“ At least some denoising improvement verified") + + else: + print(" No hidden states available for denoising quality assessment") + + # Additional validation + print(f"\n10. Additional validation...") + for i, structure in enumerate(conditioned_structures): + # Check for NaN values + assert not torch.isnan(structure).any(), f"Structure {i} contains NaN values" + # Check for infinite values + assert not torch.isinf(structure).any(), f"Structure {i} contains infinite values" + print(f"āœ“ Structure {i} contains valid values") + + print("\nšŸŽ‰ All tests passed! DenoisedConditioner works correctly.") + + # Print final summary + print(f"\n=== FINAL SUMMARY ===") + print(f" Checkpoint: {pretrained_model_path}") + print(f" Test sigma: {test_sigma}") + print(f" Denoiser sigma: {conditioner.denoiser_sigma}") + print(f" Average squared distance: {average_squared_distance}") + print(f" Computed c_in: {c_in_float}") + if denoising_improvements: + print(f" Denoising success rate: {success_rate:.1f}%") + print(f" Average RMSE improvement: {avg_improvement:.6f}") + + return True + + except Exception as e: + print(f"āŒ Test failed with error: {e}") + import traceback + traceback.print_exc() + return False + +if __name__ == "__main__": + success = test_denoised_conditioner_with_scaling() + if success: + print("\nāœ… DenoisedConditioner scaling test passed!") + print("The conditioner correctly handles the scaling parameter and emulates xhat_normalized behavior.") + else: + print("\nāŒ DenoisedConditioner scaling test failed!") \ No newline at end of file diff --git a/scratch/test_gradient_equivalence.py b/scratch/test_gradient_equivalence.py new file mode 100644 index 0000000..2192c21 --- /dev/null +++ b/scratch/test_gradient_equivalence.py @@ -0,0 +1,270 @@ +""" +Test gradient equivalence between automatic and manual optimization in DenoiserMultimeasurement. + +To run this test with proper hydra configuration: + +python3 scratch/test_gradient_equivalence.py --config-dir=configs experiment=train_test_single_shape_conditional ++model._target_=jamun.model.denoiser_multimeasurement.DenoiserMultimeasurement ++model.multimeasurement=True ++model.N_measurements_hidden=2 ++model.N_measurements=2 ++model.max_graphs_per_batch=1 + +This will: +- Use the train_test_single_shape_conditional experiment config +- Override model to use DenoiserMultimeasurement +- Enable multimeasurement with 2 hidden measurements and 2 measurements +- Set max_graphs_per_batch=1 for manual optimization testing +""" + +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +import copy +import numpy as np + +dotenv.load_dotenv("../.env", verbose=True) +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") + +project_root = "/homefs/home/sules/jamun" +if project_root not in sys.path: + sys.path.insert(0, project_root) + +import jamun +from jamun.utils import compute_average_squared_distance_from_datasets + +def create_model_and_data(cfg, max_graphs_per_batch=None): + """Create a model and datamodule with specified optimization mode.""" + # Configure DenoiserMultimeasurement + cfg.model._target_ = "jamun.model.denoiser_multimeasurement.DenoiserMultimeasurement" + cfg.model.sigma_distribution._target_ = "jamun.distributions.ConstantSigma" + cfg.model.sigma_distribution.sigma = 0.04 + cfg.model.multimeasurement = True + cfg.model.N_measurements_hidden = 2 + cfg.model.N_measurements = 2 + cfg.model.max_graphs_per_batch = max_graphs_per_batch + + # # Set up data - use correct attribute name filter_codes + # cfg.data.datamodule.datasets.train.filter_codes = ['ALA_ALA'] + # cfg.data.datamodule.datasets.val.filter_codes = ['ALA_ALA'] + # cfg.data.datamodule.datasets.test.filter_codes = ['ALA_ALA'] + # cfg.data.datamodule.batch_size = 8 # Larger batch for meaningful chunking + + # Compute normalization + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + cutoff = cfg.model.max_radius + average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) + cfg.model.average_squared_distance = average_squared_distance + + # Create model and data + model = hydra.utils.instantiate(cfg.model) + datamodule.setup('test') + + return model, datamodule + +def get_model_gradients(model): + """Extract gradients from model parameters.""" + gradients = {} + for name, param in model.named_parameters(): + if param.grad is not None: + gradients[name] = param.grad.clone().detach() + return gradients + +def compute_gradient_norm(gradients): + """Compute the total gradient norm across all parameters.""" + total_norm = 0.0 + for grad in gradients.values(): + total_norm += grad.norm().item() ** 2 + return total_norm ** 0.5 + +def compare_gradients(grad1, grad2, tolerance=1e-3): + """Compare two gradient dictionaries.""" + if set(grad1.keys()) != set(grad2.keys()): + print("ERROR: Different parameter names!") + return False + + max_relative_diff = 0.0 + for name in grad1.keys(): + g1, g2 = grad1[name], grad2[name] + + # Compute relative difference + diff = torch.abs(g1 - g2) + max_val = torch.max(torch.abs(g1), torch.abs(g2)) + relative_diff = torch.where(max_val > 1e-8, diff / (max_val + 1e-8), diff) + max_rel_diff_param = relative_diff.max().item() + max_relative_diff = max(max_relative_diff, max_rel_diff_param) + + print(f"{name:30s}: max_rel_diff = {max_rel_diff_param:.6f}, norm_ratio = {g1.norm().item()/g2.norm().item():.6f}") + + print(f"\nOverall max relative difference: {max_relative_diff:.6f}") + return max_relative_diff < tolerance + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") +def test_gradient_equivalence(cfg): + """Test that automatic and manual optimization produce equivalent gradients.""" + print("="*80) + print("TESTING GRADIENT EQUIVALENCE: AUTOMATIC vs MANUAL OPTIMIZATION") + print("="*80) + + # Set seeds for reproducibility + torch.manual_seed(42) + np.random.seed(42) + + # Create automatic optimization model + print("\n1. Creating AUTOMATIC optimization model...") + model_auto, datamodule_auto = create_model_and_data(cfg.copy(), max_graphs_per_batch=None) + print(f" Automatic optimization: {model_auto.automatic_optimization}") + + # Create manual optimization model with same architecture + print("2. Creating MANUAL optimization model...") + model_manual, datamodule_manual = create_model_and_data(cfg.copy(), max_graphs_per_batch=2) # 2 graphs per chunk + print(f" Automatic optimization: {model_manual.automatic_optimization}") + + # Ensure models are in training mode and parameters require gradients + print("3. Setting up models for gradient computation...") + model_auto.train() + model_manual.train() + + # Ensure all parameters require gradients + for param in model_auto.parameters(): + param.requires_grad_(True) + for param in model_manual.parameters(): + param.requires_grad_(True) + + # Copy weights from auto to manual model to ensure identical starting point + print("4. Synchronizing model weights...") + model_manual.load_state_dict(model_auto.state_dict()) + + # Get the same batch of data + print("5. Getting identical batch...") + torch.manual_seed(42) # Reset seed to get same batch + train_loader_auto = datamodule_auto.train_dataloader() + batch_auto = next(iter(train_loader_auto)) + + torch.manual_seed(42) # Reset seed to get same batch + train_loader_manual = datamodule_manual.train_dataloader() + batch_manual = next(iter(train_loader_manual)) + + print(f" Batch shapes: auto={batch_auto.pos.shape}, manual={batch_manual.pos.shape}") + print(f" Batch equality: {torch.allclose(batch_auto.pos, batch_manual.pos)}") + + # Use the same sigma for both (important!) + print("\n6. Setting identical sigma values...") + sigma_value = 0.04 # Fixed sigma instead of sampling + sigma_auto = torch.tensor(sigma_value) + sigma_manual = torch.tensor(sigma_value) + + print(f" Using fixed sigma: {sigma_value}") + + # Test forward pass equivalence (without multimeasurement first) + print("\n7. Testing forward pass equivalence...") + + # Disable multimeasurement temporarily for cleaner testing + model_auto.multimeasurement = False + model_manual.multimeasurement = False + + with torch.no_grad(): + # Reset random seeds before each forward pass + torch.manual_seed(123) + x_target_auto, xhat_auto, y_auto = model_auto.noise_and_denoise(batch_auto, sigma_auto, align_noisy_input=True) + + torch.manual_seed(123) # Same seed for manual + x_target_manual, xhat_manual, y_manual = model_manual.noise_and_denoise(batch_manual, sigma_manual, align_noisy_input=True) + + forward_equal = torch.allclose(xhat_auto.pos, xhat_manual.pos, atol=1e-5) + print(f" Forward pass output equality: {forward_equal}") + + if not forward_equal: + print(f" Max difference: {(xhat_auto.pos - xhat_manual.pos).abs().max().item():.8f}") + print(" Continuing with test anyway...") + + # Test gradient computation + print("\n8. Computing gradients...") + + # AUTOMATIC OPTIMIZATION + print(" Computing automatic optimization gradients...") + model_auto.zero_grad() + + # Use same random seed for loss computation + torch.manual_seed(456) + x_target_auto, xhat_auto, y_auto = model_auto.noise_and_denoise(batch_auto, sigma_auto, align_noisy_input=True) + loss_auto, aux_auto = model_auto.compute_loss(x_target_auto, xhat_auto, sigma_auto) + loss_auto_mean = loss_auto.mean() + + print(f" Auto loss: {loss_auto_mean.item():.6f}") + print(f" Auto loss requires_grad: {loss_auto_mean.requires_grad}") + + loss_auto_mean.backward() + + gradients_auto = get_model_gradients(model_auto) + grad_norm_auto = compute_gradient_norm(gradients_auto) + + print(f" Auto grad_norm: {grad_norm_auto:.6f}") + + # MANUAL OPTIMIZATION (simulate _manual_step) + print(" Computing manual optimization gradients...") + model_manual.zero_grad() + + # Use same random seed for noise generation + torch.manual_seed(456) + y_manual, x_target_manual_prep = model_manual._prepare_noisy_batch( + batch_manual, sigma_manual, align_noisy_input=True + ) + + # Split into chunks + y_list = y_manual.to_data_list() + x_target_list = x_target_manual_prep.to_data_list() + chunk_size = model_manual.max_graphs_per_batch + num_chunks = (len(y_list) + chunk_size - 1) // chunk_size + + print(f" Manual: {len(y_list)} graphs ā†’ {num_chunks} chunks of size {chunk_size}") + + # Process chunks and accumulate gradients (simulate manual_step) + total_loss_manual = 0.0 + for i in range(num_chunks): + start_idx = i * chunk_size + end_idx = min(start_idx + chunk_size, len(y_list)) + + y_chunk = torch_geometric.data.Batch.from_data_list(y_list[start_idx:end_idx]) + x_target_chunk = torch_geometric.data.Batch.from_data_list(x_target_list[start_idx:end_idx]) + + xhat_chunk = model_manual.xhat(y_chunk, sigma_manual) + loss_chunk, aux_chunk = model_manual.compute_loss(x_target_chunk, xhat_chunk, sigma_manual) + loss_chunk_mean = loss_chunk.mean() + + # Scale loss by number of chunks (the fix we implemented) + scaled_loss = loss_chunk_mean / num_chunks + scaled_loss.backward() + + total_loss_manual += loss_chunk_mean.item() + + gradients_manual = get_model_gradients(model_manual) + grad_norm_manual = compute_gradient_norm(gradients_manual) + + print(f" Manual total loss: {total_loss_manual:.6f}, grad_norm: {grad_norm_manual:.6f}") + + # Compare gradients + print("\n9. COMPARING GRADIENTS:") + print(f" Gradient norm ratio (manual/auto): {grad_norm_manual/grad_norm_auto:.6f}") + print(f" Loss ratio (manual/auto): {total_loss_manual/loss_auto_mean.item():.6f}") + + print("\n Parameter-wise comparison:") + gradients_match = compare_gradients(gradients_auto, gradients_manual, tolerance=1e-2) + + print("\n" + "="*80) + if gradients_match: + print("āœ… SUCCESS: Manual and automatic optimization produce equivalent gradients!") + print(f" Gradient norms: auto={grad_norm_auto:.6f}, manual={grad_norm_manual:.6f}") + print(f" Relative difference: {abs(grad_norm_manual-grad_norm_auto)/grad_norm_auto*100:.3f}%") + else: + print("āŒ FAILURE: Gradients do not match within tolerance!") + print(" This indicates an issue with the manual optimization implementation.") + print("="*80) + + return gradients_match + +if __name__ == "__main__": + test_gradient_equivalence() \ No newline at end of file diff --git a/scratch/test_multimeasurement.py b/scratch/test_multimeasurement.py new file mode 100644 index 0000000..f9c7c9a --- /dev/null +++ b/scratch/test_multimeasurement.py @@ -0,0 +1,172 @@ +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +from jamun.hydra import instantiate_dict_cfg +import pdb +import jamun +from jamun.utils import compute_average_squared_distance_from_datasets +import lightning.pytorch as pl + +# Fix PyTorch Geometric backend issues +try: + import torch_scatter + import torch_sparse + import torch_cluster +except ImportError: + print("Warning: Some PyTorch Geometric extensions not available") + +# Use GPU if available +device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +print(f"Using device: {device}") + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") +JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + +project_root = "/homefs/home/sules/jamun" # Adjust if necessary +if project_root not in sys.path: + sys.path.insert(0, project_root) + print(f"Added '{project_root}' to sys.path for module discovery.") +else: + print(f"'{project_root}' is already in sys.path.") + +def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: + """Computes the average squared distance for normalization from the data.""" + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup("compute_normalization") + train_datasets = datamodule.datasets["train"] + cutoff = cfg.model.max_radius + average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) + return average_squared_distance + +def test_training_mode(cfg, mode_name, max_graphs_per_batch): + """Test training with specified optimization mode.""" + print(f"\n{'='*50}") + print(f"Testing {mode_name} mode (max_graphs_per_batch={max_graphs_per_batch})") + print(f"{'='*50}") + + # # Configure DenoiserMultimeasurement + # cfg.model._target_ = "jamun.model.denoiser_multimeasurement.DenoiserMultimeasurement" + # cfg.model.sigma_distribution._target_ = "jamun.distributions.ConstantSigma" + # cfg.model.sigma_distribution.sigma = 0.04 + + # Set multimeasurement parameters + # cfg.model.multimeasurement = True + cfg.model.N_measurements_hidden = 2 + cfg.model.N_measurements = 2 + cfg.model.max_graphs_per_batch = max_graphs_per_batch + + # Compute normalization + average_squared_distance = compute_average_squared_distance_from_config(cfg) + cfg.model.average_squared_distance = average_squared_distance + breakpoint() + print("Loading datamodule...") + datamodule = hydra.utils.instantiate(cfg.data.datamodule) + datamodule.setup('test') + breakpoint() + print("Loading model...") + model = hydra.utils.instantiate(cfg.model) + print(f"Model loaded: {type(model)}") + print(f"Multimeasurement: {model.multimeasurement}") + print(f"N_measurements_hidden: {model.N_measurements_hidden}") + print(f"N_measurements: {model.N_measurements}") + print(f"Automatic optimization: {model.automatic_optimization}") + print(f"Sigma: {model.sigma_distribution.sigma}") + breakpoint() + # Get a single batch + print("Getting a batch of data...") + train_loader = datamodule.train_dataloader() + _, batch = next(enumerate(train_loader)) + breakpoint() + print(f"Batch shape: {batch.pos.shape}") + print(f"Batch num_graphs: {batch.num_graphs}") + if hasattr(batch, 'hidden_state') and batch.hidden_state is not None: + print(f"Hidden state shapes: {[h.shape for h in batch.hidden_state]}") + else: + print("No hidden states in batch") + breakpoint() + # Test forward pass + print("Testing forward pass...") + with torch.no_grad(): + sigma = model.sigma_distribution.sample() + x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) + breakpoint() + print(f"Input shape: {batch.pos.shape}") + print(f"Noisy shape: {y.pos.shape}") + print(f"Output shape: {xhat.pos.shape}") + print(f"Target shape: {x_target.pos.shape}") + + # Verify multimeasurement expansion + expected_graphs = batch.num_graphs * model.N_measurements_hidden * model.N_measurements + actual_graphs = y.num_graphs + print(f"Expected graphs after multimeasurement: {expected_graphs}") + print(f"Actual graphs: {actual_graphs}") + assert actual_graphs == expected_graphs, f"Graph count mismatch: expected {expected_graphs}, got {actual_graphs}" + + # Test actual training with fast_dev_run + print("Testing training with fast_dev_run...") + + # Configure trainer to use only 1 GPU + if torch.cuda.is_available(): + print(f"CUDA available with {torch.cuda.device_count()} GPUs - using GPU 0 only") + trainer = pl.Trainer( + fast_dev_run=1, # Run 1 train, 1 val batch and stop + enable_checkpointing=False, + logger=False, + enable_progress_bar=True, + enable_model_summary=False, + accelerator='gpu', + devices=[0], # Explicitly use only GPU 0 + strategy='auto', # Single device strategy + ) + else: + print("CUDA not available - using CPU") + trainer = pl.Trainer( + fast_dev_run=1, + enable_checkpointing=False, + logger=False, + enable_progress_bar=True, + enable_model_summary=False, + accelerator='cpu', + devices=1, + ) + + try: + trainer.fit(model, datamodule) + print(f"{mode_name} mode training completed successfully!") + except Exception as e: + print(f"Error during training: {e}") + raise + + print(f"{mode_name} mode test completed successfully!") + return model + +@hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") +def main(cfg): + # # Override data config to use only ALA_ALA + # cfg.data.datamodule.filter_codes = ['ALA_ALA'] + # cfg.data.datamodule.subsample = 10 # Use fewer samples for faster testing + # cfg.data.datamodule.batch_size = 4 # Small batch size for testing + + print("Testing DenoiserMultimeasurement training modes") + print(f"Using ALA_ALA data from: {JAMUN_DATA_PATH}") + + # Test automatic optimization mode + model_auto = test_training_mode(cfg.copy(), "AUTOMATIC", None) + + # Test manual optimization mode + model_manual = test_training_mode(cfg.copy(), "MANUAL", 2) # Process 2 graphs at a time + + print(f"\n{'='*50}") + print("ALL TESTS PASSED!") + print("Both automatic and manual optimization modes work correctly.") + print(f"{'='*50}") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/test_reorganize_swarm_data.py b/scratch/test_reorganize_swarm_data.py new file mode 100644 index 0000000..404cf3c --- /dev/null +++ b/scratch/test_reorganize_swarm_data.py @@ -0,0 +1,253 @@ +#!/usr/bin/env python3 +""" +Test script for the swarm data reorganization. + +This script tests the reorganization functionality on a small subset of data +before running the full reorganization. +""" + +import os +import shutil +import tempfile +import sys +from pathlib import Path + +# Add the scratch directory to the path so we can import the main script +sys.path.insert(0, os.path.dirname(os.path.abspath(__file__))) + +try: + import mdtraj as md + MDTRAJ_AVAILABLE = True +except ImportError: + MDTRAJ_AVAILABLE = False + print("āš ļø mdtraj not available. Trajectory validation will be skipped.") + +try: + from tqdm import tqdm + TQDM_AVAILABLE = True +except ImportError: + TQDM_AVAILABLE = False + print("āš ļø tqdm not available. Progress bars will be disabled.") + +def create_test_data(test_source_dir: str, test_pdb_file: str): + """Create a small test dataset structure.""" + print("Creating test data structure...") + + # Create test source directory + os.makedirs(test_source_dir, exist_ok=True) + + # Create a few test grid directories with mock files + test_grid_codes = ['000', '001', '002', '003', '004'] + trajectory_codes = ['001', '002', '003', '004', '005'] + + for grid_code in test_grid_codes: + grid_dir = os.path.join(test_source_dir, f"AA_{grid_code}") + os.makedirs(grid_dir, exist_ok=True) + + # Create mock .xtc files + for traj_code in trajectory_codes: + xtc_file = os.path.join(grid_dir, f"swarm_1ps_{traj_code}.xtc") + # Create empty files that will be copied, but note they won't be valid XTC format + # This is just for testing the file organization logic + with open(xtc_file, 'w') as f: + f.write(f"Mock XTC data for grid {grid_code}, trajectory {traj_code}\n") + + # Create mock single PDB file with more realistic content + os.makedirs(os.path.dirname(test_pdb_file), exist_ok=True) + with open(test_pdb_file, 'w') as f: + # Write a minimal but valid PDB structure for 2 alanine residues + f.write("TITLE MOCK ALA-ALA DIPEPTIDE\n") + f.write("ATOM 1 N ALA A 1 0.000 0.000 0.000 1.00 0.00 N\n") + f.write("ATOM 2 CA ALA A 1 1.458 0.000 0.000 1.00 0.00 C\n") + f.write("ATOM 3 C ALA A 1 2.009 1.420 0.000 1.00 0.00 C\n") + f.write("ATOM 4 O ALA A 1 1.332 2.445 0.000 1.00 0.00 O\n") + f.write("ATOM 5 CB ALA A 1 1.978 -0.750 1.202 1.00 0.00 C\n") + f.write("ATOM 6 H ALA A 1 -0.481 0.000 0.890 1.00 0.00 H\n") + f.write("ATOM 7 HA ALA A 1 1.804 -0.531 -0.900 1.00 0.00 H\n") + f.write("ATOM 8 HB1 ALA A 1 1.642 -1.785 1.202 1.00 0.00 H\n") + f.write("ATOM 9 HB2 ALA A 1 3.068 -0.750 1.202 1.00 0.00 H\n") + f.write("ATOM 10 HB3 ALA A 1 1.642 -0.281 2.132 1.00 0.00 H\n") + f.write("ATOM 11 N ALA A 2 3.332 1.420 0.000 1.00 0.00 N\n") + f.write("ATOM 12 CA ALA A 2 4.009 2.709 0.000 1.00 0.00 C\n") + f.write("ATOM 13 C ALA A 2 5.509 2.709 0.000 1.00 0.00 C\n") + f.write("ATOM 14 O ALA A 2 6.134 1.649 0.000 1.00 0.00 O\n") + f.write("ATOM 15 CB ALA A 2 3.489 3.459 1.202 1.00 0.00 C\n") + f.write("ATOM 16 H ALA A 2 3.855 0.556 0.000 1.00 0.00 H\n") + f.write("ATOM 17 HA ALA A 2 3.673 3.240 -0.900 1.00 0.00 H\n") + f.write("ATOM 18 HB1 ALA A 2 3.825 4.494 1.202 1.00 0.00 H\n") + f.write("ATOM 19 HB2 ALA A 2 2.399 3.459 1.202 1.00 0.00 H\n") + f.write("ATOM 20 HB3 ALA A 2 3.825 2.990 2.132 1.00 0.00 H\n") + f.write("ATOM 21 OXT ALA A 2 6.032 3.829 0.000 1.00 0.00 O\n") + f.write("TER 22 ALA A 2\n") + f.write("END\n") + + print(f"Created test data with {len(test_grid_codes)} grid codes") + +def test_mdtraj_with_mock_data(train_dir: str, val_dir: str): + """ + Test mdtraj functionality with mock data. + Note: This will likely fail since we're creating mock XTC files that aren't real trajectories. + """ + if not MDTRAJ_AVAILABLE: + print("āš ļø mdtraj not available, skipping trajectory compatibility tests") + return + + print("Testing mdtraj compatibility (expected to fail with mock data)...") + + for split_name, split_dir in [("train", train_dir), ("val", val_dir)]: + xtc_files = [f for f in os.listdir(split_dir) if f.endswith('.xtc')] + if not xtc_files: + continue + + # Test one file from each split + test_file = xtc_files[0] + base_name = test_file.replace('.xtc', '') + pdb_file = f"{base_name}.pdb" + + xtc_path = os.path.join(split_dir, test_file) + pdb_path = os.path.join(split_dir, pdb_file) + + try: + # This will likely fail since we have mock XTC data + traj = md.load(xtc_path, top=pdb_path) + print(f"āœ… {split_name}: Successfully loaded {test_file} + {pdb_file} " + f"({traj.n_frames} frames, {traj.n_atoms} atoms)") + del traj + except Exception as e: + print(f"āŒ {split_name}: Failed to load {test_file} + {pdb_file}: {str(e)}") + print(" (This is expected with mock data - real data should work)") + + print("Note: mdtraj tests with mock data are expected to fail.") + print("The real script will test with actual trajectory files.") + +def test_reorganization(): + """Test the reorganization script with mock data.""" + with tempfile.TemporaryDirectory() as temp_dir: + # Set up test paths + test_source = os.path.join(temp_dir, "test_swarm_results") + test_target = os.path.join(temp_dir, "test_enhanced") + test_pdb = os.path.join(temp_dir, "test_ALA_ALA.pdb") + + # Create test data + create_test_data(test_source, test_pdb) + + # Import and modify the main script for testing + import reorganize_swarm_data + + # Temporarily override the configuration + original_source = reorganize_swarm_data.SOURCE_DIR + original_pdb = reorganize_swarm_data.SINGLE_PDB_FILE + original_strategies = reorganize_swarm_data.SPLITTING_STRATEGIES.copy() + + reorganize_swarm_data.SOURCE_DIR = test_source + reorganize_swarm_data.SINGLE_PDB_FILE = test_pdb + + # Override the splitting strategies for testing + reorganize_swarm_data.SPLITTING_STRATEGIES = { + 'grid_split': { + 'target_dir': test_target + "_grid_split", + 'train_size': 3, # Use 3 for train, 2 for val + 'description': "Test grid split" + }, + 'trajectory_split': { + 'target_dir': test_target + "_trajectory_split", + 'train_trajectories': ['001', '002', '003'], # First 3 for testing + 'val_trajectories': ['004', '005'], # Last 2 for testing + 'description': "Test trajectory split" + } + } + + try: + print("\n" + "="*50) + print("RUNNING TEST REORGANIZATION") + print("="*50) + + # Test both strategies + print("Testing grid split strategy...") + reorganize_swarm_data.main('grid_split') + + print("Testing trajectory split strategy...") + reorganize_swarm_data.main('trajectory_split') + + # Verify results + print("\n" + "="*50) + print("VERIFYING TEST RESULTS") + print("="*50) + + # Check both strategies + for strategy_name in ['grid_split', 'trajectory_split']: + strategy_dir = test_target + f"_{strategy_name}" + train_dir = os.path.join(strategy_dir, 'train') + val_dir = os.path.join(strategy_dir, 'val') + + if os.path.exists(train_dir) and os.path.exists(val_dir): + train_files = os.listdir(train_dir) + val_files = os.listdir(val_dir) + + train_xtc = [f for f in train_files if f.endswith('.xtc')] + train_pdb = [f for f in train_files if f.endswith('.pdb')] + val_xtc = [f for f in val_files if f.endswith('.xtc')] + val_pdb = [f for f in val_files if f.endswith('.pdb')] + + print(f"\n{strategy_name.upper()} STRATEGY:") + print(f"Train directory: {len(train_files)} files ({len(train_xtc)} .xtc, {len(train_pdb)} .pdb)") + print(f"Val directory: {len(val_files)} files ({len(val_xtc)} .xtc, {len(val_pdb)} .pdb)") + + # Calculate expected files based on strategy + if strategy_name == 'grid_split': + # 3 grid codes Ɨ 5 trajectories Ɨ 2 file types = 30 train files + # 2 grid codes Ɨ 5 trajectories Ɨ 2 file types = 20 val files + expected_train = 3 * 5 * 2 + expected_val = 2 * 5 * 2 + else: # trajectory_split + # 5 grid codes Ɨ 3 trajectories Ɨ 2 file types = 30 train files + # 5 grid codes Ɨ 2 trajectories Ɨ 2 file types = 20 val files + expected_train = 5 * 3 * 2 + expected_val = 5 * 2 * 2 + + if len(train_files) == expected_train and len(val_files) == expected_val: + print(f"āœ… {strategy_name} Test PASSED! File counts are correct.") + + # Check a few file names + print("Sample train files:", sorted(train_files)[:3]) + print("Sample val files:", sorted(val_files)[:3]) + else: + print(f"āŒ {strategy_name} Test FAILED! Expected {expected_train} train, {expected_val} val files") + return False + else: + print(f"āŒ {strategy_name} Test FAILED! Output directories were not created") + return False + + # Test mdtraj compatibility on one strategy + strategy_dir = test_target + "_grid_split" + train_dir = os.path.join(strategy_dir, 'train') + val_dir = os.path.join(strategy_dir, 'val') + print("\n=== Testing mdtraj compatibility ===") + test_mdtraj_with_mock_data(train_dir, val_dir) + + print("\nāœ… All tests completed successfully!") + return True + + finally: + # Restore original configuration + reorganize_swarm_data.SOURCE_DIR = original_source + reorganize_swarm_data.SINGLE_PDB_FILE = original_pdb + reorganize_swarm_data.SPLITTING_STRATEGIES = original_strategies + +def main(): + """Run the test.""" + print("Starting test of swarm data reorganization script...") + + success = test_reorganization() + + if success: + print("\nšŸŽ‰ Test passed! The script should work correctly on the real data.") + print("\nTo run the full reorganization, execute:") + print("python scratch/reorganize_swarm_data.py") + else: + print("\nāŒ Test failed! Please check the script before running on real data.") + + return success + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/test_repeated_position_dataset.py b/scratch/test_repeated_position_dataset.py new file mode 100644 index 0000000..87177ce --- /dev/null +++ b/scratch/test_repeated_position_dataset.py @@ -0,0 +1,125 @@ +#!/usr/bin/env python3 +""" +Test script for RepeatedPositionDataset to verify that hidden states +are exact copies of the current position. +""" + +import torch +import os +from pathlib import Path + +# Add the src directory to the path so we can import jamun modules +import sys +sys.path.insert(0, str(Path(__file__).parent / "src")) + +from jamun.data.noisy_position_dataset import RepeatedPositionDataset + +def test_repeated_position_dataset(): + """Test RepeatedPositionDataset with ALA_ALA capped diamines data.""" + + print("Testing RepeatedPositionDataset...") + print("=" * 50) + + # Set up dataset parameters + root = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train" + traj_files = ["ALA_ALA.xtc"] + pdb_file = "ALA_ALA.pdb" + label = "ALA_ALA_test" + total_lag_time = 4 # This should create 3 hidden states (4 - 1 = 3) + + # Check if files exist + xtc_path = os.path.join(root, traj_files[0]) + pdb_path = os.path.join(root, pdb_file) + + if not os.path.exists(xtc_path): + print(f"āŒ ERROR: XTC file not found: {xtc_path}") + return False + if not os.path.exists(pdb_path): + print(f"āŒ ERROR: PDB file not found: {pdb_path}") + return False + + print(f"āœ… Found XTC file: {xtc_path}") + print(f"āœ… Found PDB file: {pdb_path}") + + try: + # Create the dataset + print(f"\nCreating RepeatedPositionDataset with total_lag_time={total_lag_time}...") + dataset = RepeatedPositionDataset( + root=root, + traj_files=traj_files, + pdb_file=pdb_file, + label=label, + total_lag_time=total_lag_time, + num_frames=5, # Only load 5 frames for testing + verbose=True + ) + + print(f"āœ… Dataset created successfully") + print(f" Dataset length: {len(dataset)}") + print(f" Dataset label: {dataset.label()}") + + # Test a few samples + print(f"\nTesting samples...") + + for idx in range(min(3, len(dataset))): + print(f"\n--- Sample {idx} ---") + + # Get sample from dataset + graph = dataset[idx] + + print(f"Graph pos shape: {graph.pos.shape}") + print(f"Number of hidden states: {len(graph.hidden_state)}") + + # Verify we have the expected number of hidden states + expected_hidden_states = total_lag_time - 1 + if len(graph.hidden_state) != expected_hidden_states: + print(f"āŒ ERROR: Expected {expected_hidden_states} hidden states, got {len(graph.hidden_state)}") + return False + + print(f"āœ… Correct number of hidden states: {len(graph.hidden_state)}") + + # Test each hidden state + for i, hidden_pos in enumerate(graph.hidden_state): + print(f"Hidden state {i} shape: {hidden_pos.shape}") + + # Check if shapes match + if hidden_pos.shape != graph.pos.shape: + print(f"āŒ ERROR: Shape mismatch! pos: {graph.pos.shape}, hidden_state[{i}]: {hidden_pos.shape}") + return False + + # Check if values are exactly equal + if not torch.allclose(hidden_pos, graph.pos, atol=1e-10): + print(f"āŒ ERROR: Hidden state {i} is not exactly equal to current position!") + print(f" Max difference: {torch.max(torch.abs(hidden_pos - graph.pos)).item()}") + return False + + # Check if they are the exact same tensor (should be different objects but same values) + if hidden_pos is graph.pos: + print(f"āš ļø WARNING: Hidden state {i} is the same object as pos (should be different objects)") + else: + print(f"āœ… Hidden state {i} is a different object with same values as pos") + + print(f"āœ… Hidden state {i} exactly matches current position") + + print(f"\nšŸŽ‰ All tests passed!") + print(f" āœ… Dataset loads correctly") + print(f" āœ… Correct number of hidden states ({total_lag_time - 1})") + print(f" āœ… Hidden states exactly match current position") + print(f" āœ… Hidden states are separate objects (not references)") + + return True + + except Exception as e: + print(f"āŒ ERROR: Exception occurred: {e}") + import traceback + traceback.print_exc() + return False + +if __name__ == "__main__": + success = test_repeated_position_dataset() + if success: + print(f"\nšŸŽ‰ Test completed successfully!") + exit(0) + else: + print(f"\nšŸ’„ Test failed!") + exit(1) \ No newline at end of file diff --git a/scratch/transformer/convert_spatiotemporal.py b/scratch/transformer/convert_spatiotemporal.py new file mode 100644 index 0000000..ef95b0f --- /dev/null +++ b/scratch/transformer/convert_spatiotemporal.py @@ -0,0 +1,194 @@ +#!/usr/bin/env python3 +""" +Functions for converting between spatial and temporal graph representations. +""" + +import torch +import torch_geometric + + +def calculate_temporal_positions(temporal_length, device=None): + """ + Calculate normalized temporal positions for nodes in a temporal graph. + + Args: + temporal_length: Total number of nodes in the temporal sequence + device: Device to create tensors on + + Returns: + torch.Tensor: Normalized positions [0, 1/T, 2/T, ..., (T-1)/T] + """ + if temporal_length <= 1: + return torch.tensor([0.0], device=device) + + # Create positions [0, 1, 2, ..., T-1] and normalize by T + positions = torch.arange(temporal_length, dtype=torch.float32, device=device) + normalized_positions = positions / temporal_length + + return normalized_positions + + +def spatial_to_temporal_graphs(batch, graph_type="fan"): + """ + Convert a batch of spatial graphs to temporal graphs with configurable connectivity. + + For each spatial node with position + hidden states, create a temporal graph where: + - Node 0: current position + - Nodes 1-T: hidden state positions + - Connectivity depends on graph_type parameter + + Args: + batch: Input spatial graph batch + graph_type: Type of connectivity to use + - "fan": Hub connects to all + sequential connections (0->all, i->(i+1)) + - "hub_n_spoke": Only hub-spoke connections (0->all, no sequential) + - "complete": Complete graph with self-loops (all-to-all including self) + - "complete_no_self": Complete graph without self-loops (all-to-all excluding self) + """ + import torch_geometric + + # Validate graph_type + valid_types = ["fan", "hub_n_spoke", "complete", "complete_no_self"] + if graph_type not in valid_types: + raise ValueError(f"graph_type must be one of {valid_types}, got {graph_type}") + + # Get device from input batch + device = batch.pos.device + + # Get dimensions + num_spatial_nodes = batch.pos.shape[0] + + # Check if we have hidden states + if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + num_hidden_states = len(batch.hidden_state) + temporal_length = 1 + num_hidden_states # current + hidden + else: + # If no hidden states, just use current position + num_hidden_states = 0 + temporal_length = 1 + + # print(f"Creating {graph_type} temporal graphs: {num_spatial_nodes} spatial nodes -> {num_spatial_nodes} temporal graphs of length {temporal_length}") + + # Store reference to spatial graph + spatial_graph = batch.clone() + + # Set connectivity type code for tracking + connectivity_type_map = { + "fan": 0, + "hub_n_spoke": 1, + "complete": 2, + "complete_no_self": 3 + } + + temporal_graphs = [] + + for node_idx in range(num_spatial_nodes): + # Build temporal positions: [current_pos, hidden_1, hidden_2, ...] + temporal_positions = [batch.pos[node_idx]] # Start with current position + + # Add hidden state positions + if num_hidden_states > 0: + for hidden_pos in batch.hidden_state: + temporal_positions.append(hidden_pos[node_idx]) + + temporal_pos = torch.stack(temporal_positions) # Shape: [T, 3] + + # Calculate temporal positions for this sequence + temporal_position = calculate_temporal_positions(temporal_length, device=device) + + # Create edge connectivity based on graph_type + if temporal_length > 1: + if graph_type == "fan": + # Original fan system: hub-spoke + sequential + # Hub connections: 0->1, 0->2, 0->3, ..., 0->T-1 + hub_src = [0] * (temporal_length - 1) + hub_dst = list(range(1, temporal_length)) + + # Sequential connections: 1->2, 2->3, ..., (T-2)->(T-1) + seq_src = list(range(1, temporal_length - 1)) + seq_dst = list(range(2, temporal_length)) + + # Combine all edges + all_src = hub_src + seq_src + all_dst = hub_dst + seq_dst + + edge_index = torch.tensor([all_src, all_dst], dtype=torch.long, device=device) + + elif graph_type == "hub_n_spoke": + # Hub-and-spoke only: 0 connects to all others, no sequential + hub_src = [0] * (temporal_length - 1) + hub_dst = list(range(1, temporal_length)) + + edge_index = torch.tensor([hub_src, hub_dst], dtype=torch.long, device=device) + + elif graph_type == "complete": + # Complete graph without self-loops: all-to-all excluding self + src_nodes = [] + dst_nodes = [] + + for i in range(temporal_length): + for j in range(temporal_length): + if i != j: # Exclude self-loops + src_nodes.append(i) + dst_nodes.append(j) + + edge_index = torch.tensor([src_nodes, dst_nodes], dtype=torch.long, device=device) + + else: + # Single node case + if graph_type == "complete": + # Single node with self-loop + edge_index = torch.tensor([[0], [0]], dtype=torch.long, device=device) + else: + # Single node, no edges for other types + edge_index = torch.tensor([[], []], dtype=torch.long, device=device) + + # Create temporal graph for this spatial node + temporal_graph = torch_geometric.data.Data( + pos=temporal_pos, + edge_index=edge_index, + spatial_node_idx=torch.tensor([node_idx], device=device), # Track which spatial node this came from + temporal_length=torch.tensor([temporal_length], device=device), + temporal_position=temporal_position, # Normalized position in sequence [0, 1/T, 2/T, ...] + connectivity_type=torch.tensor([connectivity_type_map[graph_type]], device=device), + graph_type=graph_type # Store graph type as string for debugging + ) + temporal_graphs.append(temporal_graph) + + # Batch all temporal graphs + temporal_batch = torch_geometric.data.Batch.from_data_list(temporal_graphs) + + # Store spatial graph reference and graph type + temporal_batch.spatial_graph = spatial_graph + temporal_batch.graph_type = graph_type + + return temporal_batch + + +def temporal_to_spatial_graphs(temporal_batch): + """ + Convert temporal graphs back to spatial graphs. + Take the 0th node position from each temporal graph as the updated spatial position. + """ + # Get the spatial graph template + spatial_graph = temporal_batch.spatial_graph.clone() + + # Extract 0th node positions from each temporal graph + num_temporal_graphs = temporal_batch.num_graphs + updated_positions = [] + + # Iterate through each temporal graph in the batch + for graph_idx in range(num_temporal_graphs): + # Get the node range for this temporal graph + start_idx = temporal_batch.ptr[graph_idx] + + # The 0th node of each temporal graph is at the start of its range + updated_positions.append(temporal_batch.pos[start_idx]) + + # Stack to create new position tensor + updated_positions = torch.stack(updated_positions) + + # Update spatial graph with new positions + spatial_graph.pos = updated_positions + + return spatial_graph diff --git a/scratch/transformer/develop_transformer.py b/scratch/transformer/develop_transformer.py new file mode 100644 index 0000000..34a0a15 --- /dev/null +++ b/scratch/transformer/develop_transformer.py @@ -0,0 +1,354 @@ +#!/usr/bin/env python3 +""" +Simple test script for the debugged denoiser_conditional using default hydra config. +Tests with sigma = 0.0 and sigma = 0.1. + +Device Handling: +- This script manually handles CUDA device placement for standalone testing +- PyTorch Lightning DOES handle device placement automatically when using the Trainer +- In Lightning, you typically don't need to call .to(device) manually on models or data +- Lightning moves models to the specified device and handles data loading automatically +- For standalone scripts like this one, manual device handling is required +""" + +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import dotenv +import sys +import os +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +from jamun.utils import compute_average_squared_distance_from_datasets +from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets +from jamun.data import parse_datasets_from_directory +from jamun.utils import unsqueeze_trailing + +# Import spatial-temporal conversion functions +from convert_spatiotemporal import ( + calculate_temporal_positions, + spatial_to_temporal_graphs, + temporal_to_spatial_graphs +) + +# Import node attribute conversion functions +from pooling import ( + spatial_to_temporal_node_attr, + temporal_to_spatial_node_attr, + temporal_to_spatial_node_attr_mean, + SpatialTemporalToTemporalNodeAttr, + TemporalToSpatialNodeAttrMean +) + +# Setup device - use CUDA if available +device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +print(f"Using device: {device}") +if torch.cuda.is_available(): + print(f"CUDA device: {torch.cuda.get_device_name()}") + print(f"CUDA memory: {torch.cuda.get_device_properties(device).total_memory / 1e9:.1f} GB") + +def to_device(obj, device): + """Helper function to move objects to device, handling various types.""" + if hasattr(obj, 'to'): + return obj.to(device) + elif isinstance(obj, (list, tuple)): + return type(obj)(to_device(item, device) for item in obj) + elif isinstance(obj, dict): + return {key: to_device(value, device) for key, value in obj.items()} + else: + return obj + +def move_graph_to_device(graph, device): + """Move a PyTorch Geometric graph and all its tensor attributes to device.""" + # Move the graph using standard .to() method + graph = graph.to(device) + + # Manually move any custom tensor attributes that might not be handled + for attr_name in dir(graph): + if not attr_name.startswith('_'): # Skip private attributes + attr_value = getattr(graph, attr_name, None) + if isinstance(attr_value, torch.Tensor): + setattr(graph, attr_name, attr_value.to(device)) + + return graph + +dataset = parse_datasets_from_directory( + root="/data/bucket/kleinhej/capped_diamines/timewarp_splits/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + filter_codes=['ALA_ALA'], + as_iterable=False, + subsample=80, + total_lag_time=8, + lag_subsample_rate=10, + start_frame=800000, + num_frames=200000 +) +# temporal_distance_cutoff = compute_temporal_average_squared_distance_from_datasets(dataset) +breakpoint() +# convert to dataloader, then pull data + +graph = dataset[0].__getitem__(0) +batch = torch_geometric.data.Batch.from_data_list([graph]) +from helpers import add_edges +batch = add_edges(batch.pos, batch, batch.batch, 0.05) + +# Move batch data to device +batch = move_graph_to_device(batch, device) +print(f"Moved batch to device: {batch.pos.device}") + +# Use E3Conv architecture for spatial feature processing +import e3tools +from e3nn import o3 +from helpers import create_e3conv_network, get_e3conv_output_irreps, apply_e3conv_to_positions + +# Create E3Conv network with yaml configuration parameters +spatial_e3conv = create_e3conv_network() + +# Move E3Conv model to device +spatial_e3conv = spatial_e3conv.to(device) +print(f"Moved E3Conv model to device: {next(spatial_e3conv.parameters()).device}") + +print(f"E3Conv output irreps: {get_e3conv_output_irreps()}") +output_irreps = o3.Irreps(get_e3conv_output_irreps()) +print(f"E3Conv output dimension: {output_irreps.dim}") + +print("\n" + "="*50) +print("TEMPORAL GRAPH CONVERSION") +print("="*50) + +def test_temporal_conversion(batch, graph_type="fan"): + """Test the conversion functions with example output.""" + print("=== Testing Temporal Graph Conversion ===") + + print(f"Original spatial batch:") + print(f" - pos shape: {batch.pos.shape}") + print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") + if hasattr(batch, 'hidden_state') and batch.hidden_state: + print(f" - hidden_state[0] shape: {batch.hidden_state[0].shape}") + + # Convert to temporal + temporal_batch = spatial_to_temporal_graphs(batch) + + print(f"\nTemporal batch:") + print(f" - pos shape: {temporal_batch.pos.shape}") + print(f" - edge_index shape: {temporal_batch.edge_index.shape}") + print(f" - num_graphs: {temporal_batch.num_graphs}") + print(f" - example edge_index for first temporal graph:") + + # Show first temporal graph structure + first_graph_end = temporal_batch.ptr[1] if temporal_batch.num_graphs > 1 else len(temporal_batch.pos) + first_graph_edges = temporal_batch.edge_index[:, temporal_batch.edge_index[0] < first_graph_end] + print(f" {first_graph_edges}") + + print(f"\n - COMPLETE edge_index for entire temporal batch:") + print(f" Shape: {temporal_batch.edge_index.shape}") + print(f" {temporal_batch.edge_index}") + + print(f"\n - Temporal graph boundaries (ptr): {temporal_batch.ptr}") + print(f" - Graph node ranges:") + for i in range(temporal_batch.num_graphs): + start = temporal_batch.ptr[i] + end = temporal_batch.ptr[i+1] if i+1 < len(temporal_batch.ptr) else len(temporal_batch.pos) + print(f" Graph {i}: nodes {start}-{end-1} ({end-start} nodes)") + + print(f"\n - Temporal positions for each graph:") + print(f" Shape: {temporal_batch.temporal_position.shape}") + print(f" First graph temporal_position: {temporal_batch.temporal_position[:5]}") # Show first 5 positions + print(f" All temporal_position values: {temporal_batch.temporal_position}") + + # Convert back to spatial + reconstructed_spatial = temporal_to_spatial_graphs(temporal_batch) + + print(f"\nReconstructed spatial:") + print(f" - pos shape: {reconstructed_spatial.pos.shape}") + print(f" - position difference from original: {torch.norm(reconstructed_spatial.pos - batch.pos)}") + + return temporal_batch, reconstructed_spatial + + +# Test the temporal conversion +temporal_batch, reconstructed_spatial = test_temporal_conversion(batch) + +# Move temporal batch to device (should already be on correct device now) +temporal_batch = move_graph_to_device(temporal_batch, device) +reconstructed_spatial = move_graph_to_device(reconstructed_spatial, device) +print(f"Temporal batch device verification:") +print(f" - pos device: {temporal_batch.pos.device}") +print(f" - edge_index device: {temporal_batch.edge_index.device}") +print(f" - batch device: {temporal_batch.batch.device}") +print(f" - ptr device: {temporal_batch.ptr.device}") +if hasattr(temporal_batch, 'temporal_position'): + print(f" - temporal_position device: {temporal_batch.temporal_position.device}") +if hasattr(temporal_batch, 'spatial_node_idx'): + print(f" - spatial_node_idx device: {temporal_batch.spatial_node_idx.device}") + +print("\n" + "="*50) +print("PROCESSING ALL TEMPORAL POSITIONS WITH E3CONV") +print("="*50) +breakpoint() +# Process all temporal positions with E3Conv +with torch.no_grad(): + # Create topology without positions for E3Conv processing + # add edges to the topology + sigma = torch.tensor(0.0, device=device) + from jamun.utils import unsqueeze_trailing + sigma = unsqueeze_trailing(sigma,1) + topology = batch.clone() + topology = move_graph_to_device(topology, device) + del topology.pos, topology.batch, topology.num_graphs + + # Process current positions: [N, 3] -> [N, 1, num_features] + node_attr_current = spatial_e3conv(batch.pos, topology, batch.batch, \ + num_graphs=batch.num_graphs,\ + c_noise=sigma,\ + effective_radial_cutoff=0.05).unsqueeze(1) + + # Process hidden state positions and collect all temporal features + node_attr_list = [node_attr_current] + breakpoint() + if hasattr(batch, 'hidden_state') and batch.hidden_state: + for hidden_pos in batch.hidden_state: + node_attr_hidden = node_attr_current = spatial_e3conv(hidden_pos, topology, batch.batch, \ + num_graphs=batch.num_graphs,\ + c_noise=sigma,\ + effective_radial_cutoff=0.05).unsqueeze(1) + node_attr_list.append(node_attr_hidden) + + # Stack along temporal dimension: [N, T, num_features] + breakpoint() + node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) + + breakpoint() + # Convert spatial-temporal features to temporal node attributes with proper ordering + spatial_temporal_pooler = SpatialTemporalToTemporalNodeAttr() + spatial_node_attr_all_temporal = spatial_temporal_pooler(node_attr_spatial_temporal, temporal_batch) + +breakpoint() +print(f"Node attributes for all temporal positions: {spatial_node_attr_all_temporal.shape}") +print(f"First spatial node temporal features: {node_attr_spatial_temporal[0].shape}") +print(f"Total norm (should be nonzero): {torch.norm(spatial_node_attr_all_temporal):.6f}") +print(f"Spatial node attributes device: {spatial_node_attr_all_temporal.device}") + +print("\n" + "="*50) +print("E3TRANSFORMER TEST") +print("="*50) + +breakpoint() +def test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, device): + """Test the E3Transformer with temporal graphs.""" + from temporal_transformer import E3Transformer + + print("=== Testing E3Transformer ===") + + # Use the precomputed temporal node attributes (processed by E3Conv) + print(f"Using temporal node attributes (processed by E3Conv): {spatial_node_attr_all_temporal.shape}") + print(f"Sample temporal node attr: {spatial_node_attr_all_temporal[0]}") + + # The node attributes are already arranged to match temporal graph ordering + temporal_node_attr = spatial_node_attr_all_temporal + + print(f"Input shapes:") + print(f" - temporal_node_attr: {temporal_node_attr.shape}") + print(f" - temporal_graph.pos: {temporal_batch.pos.shape}") + print(f" - temporal_graph.edge_index: {temporal_batch.edge_index.shape}") + print(f" - temporal_graph.temporal_position: {temporal_batch.temporal_position.shape}") + print(f" - temporal_graph.batch: {temporal_batch.batch.shape}") + print(f" - temporal_graph.num_graphs: {temporal_batch.num_graphs}") + + # Create E3Transformer model that takes 1x1e node attributes (E3Conv output) + transformer = E3Transformer( + irreps_out="3x1e", # 3D output (like positions) + irreps_hidden="8x0e + 4x1e", # Hidden representations + irreps_sh="1x0e + 1x1e", # Spherical harmonics + irreps_node_attr="1x1e", # Input node attributes match E3Conv output + num_layers=2, + edge_attr_dim=24, # Split into 2 parts: 12+12 (radial+temporal) + num_attention_heads=1, # Single attention head for simpler test + ) + + # Move transformer to device + transformer = transformer.to(device) + print(f"Moved transformer to device: {next(transformer.parameters()).device}") + + print(f"\nTransformer parameters:") + print(f" - irreps_out: {transformer.irreps_out}") + print(f" - irreps_hidden: {transformer.irreps_hidden}") + print(f" - irreps_node_attr: {transformer.irreps_node_attr}") + print(f" - temporal_gate.irreps_out: {transformer.temporal_gate.irreps_out}") + print(f" - radial_edge_attr_dim: {transformer.radial_edge_attr_dim}") + print(f" - temporal_edge_attr_dim: {transformer.temporal_edge_attr_dim}") + + # Forward pass with tensor and graph (like E3Conv) + effective_radial_cutoff = 5.0 # Define the cutoff in forward pass + temporal_cutoff = 1.0 # Default temporal cutoff (no cutoff for temporal contributions) + with torch.no_grad(): + try: + transformer_output = transformer(temporal_node_attr, temporal_batch, effective_radial_cutoff, temporal_cutoff) + print(f"\nāœ… Transformer forward pass successful!") + print(f"Transformer output shape: {transformer_output.shape}") + print(f"Transformer output sample: {transformer_output[0]}") + print(f"Transformer output norm: {torch.norm(transformer_output):.6f}") + print(f"Used effective_radial_cutoff: {effective_radial_cutoff}") + print(f"Used temporal_cutoff: {temporal_cutoff}") + return True + except Exception as e: + print(f"\nāŒ Transformer forward pass failed: {e}") + import traceback + traceback.print_exc() + return False + +# Test the complete workflow: E3Conv -> Transformer +success = test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, device) + +print("\n" + "="*50) +print("TEMPORAL TO SPATIAL MEAN POOLING") +print("="*50) +breakpoint() +# Demonstrate mean pooling from temporal features back to spatial features +print("=== Testing Mean Pooling ===") + +# Use the transformer output or the original temporal features for pooling demonstration +print(f"Input temporal features shape: {spatial_node_attr_all_temporal.shape}") + +# Create mean pooling module and apply it +temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean() +spatial_features_pooled = temporal_to_spatial_pooler(spatial_node_attr_all_temporal, temporal_batch) + +print(f"Output spatial features shape: {spatial_features_pooled.shape}") +print(f"Number of spatial nodes recovered: {spatial_features_pooled.shape[0]}") +print(f"Original spatial nodes: {batch.pos.shape[0]}") +print(f"Feature dimension: {spatial_features_pooled.shape[1]}") +print(f"Sample pooled features (first node): {spatial_features_pooled[0]}") +print(f"Pooled features norm: {torch.norm(spatial_features_pooled):.6f}") + +# Verify that we correctly recovered the spatial dimension +assert spatial_features_pooled.shape[0] == batch.pos.shape[0], f"Spatial node count mismatch: {spatial_features_pooled.shape[0]} vs {batch.pos.shape[0]}" +print("āœ… Mean pooling successfully converted temporal features back to spatial!") + +print("\n" + "="*50) +print("TESTS COMPLETED") +print("="*50) + +print("\n" + "="*50) +print("FINAL SUMMARY") +print("="*50) + +# Device summary +print(f"Device Summary:") +print(f" - Used device: {device}") +if torch.cuda.is_available(): + print(f" - CUDA memory allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") + print(f" - CUDA memory cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") + +print(f"\nTest Results:") +print(f" - Manual workflow tests: {'āœ… PASSED' if success else 'āŒ FAILED'}") + +if success: + print("\nšŸŽ‰ ALL TESTS PASSED!") + print("The manual spatio-temporal workflow is working correctly.") + print("To test the unified E3SpatioTemporal model, run: python3 test_e3_spatiotemporal.py") +else: + print("\nāš ļø Some tests failed. Check the output above for details.") + diff --git a/scratch/transformer/helpers.py b/scratch/transformer/helpers.py new file mode 100644 index 0000000..1c08437 --- /dev/null +++ b/scratch/transformer/helpers.py @@ -0,0 +1,187 @@ +#!/usr/bin/env python3 +""" +Helper functions for creating network architectures used in transformer development. +""" + +import torch +import torch_geometric +import e3tools +import numpy as np +from jamun.model.arch.e3conv import E3Conv +from jamun.utils.average_squared_distance import compute_distance_matrix, compute_average_squared_distance +import functools +from convert_spatiotemporal import spatial_to_temporal_graphs + + +def compute_temporal_average_squared_distance_from_dataset( + dataset, + num_samples: int = 100, + verbose: bool = False +) -> float: + """ + Compute average squared distance between neighboring vertices in temporal graphs. + + Args: + dataset: Dataset containing spatial graphs with hidden states + num_samples: Number of samples to use for estimation + verbose: Whether to print verbose output + + Returns: + float: Average squared distance between temporal neighbors + """ + + avg_sq_dists = [] + num_graphs = 0 + + # Follow pattern from average_squared_distance.py + for item in dataset: + if num_graphs >= num_samples: + break + for graph in item: + if num_graphs >= num_samples: + break + # Convert to temporal graphs here + temporal_batch = spatial_to_temporal_graphs(graph) + temporal_graphs = torch_geometric.data.Batch.to_data_list(temporal_batch) + graph_mean = 0.0 + num_nodes = graph.pos.shape[0] + for temporal_graph in temporal_graphs: + avg_sq_dist = compute_average_squared_distance(temporal_graph.pos, cutoff=None) + graph_mean += avg_sq_dist / num_nodes + avg_sq_dists.append(graph_mean) + num_graphs += 1 + mean_avg_sq_dist = sum(avg_sq_dists) / num_graphs + + + if verbose: + print(f"Total graphs processed: {num_graphs}") + print(f"Total temporal graphs processed: {len(avg_sq_dists)}") + print(f"Mean average squared distance between temporal nodes: {mean_avg_sq_dist:.6f}") + print(f"Standard deviation: {np.std(avg_sq_dists):.6f}") + + return float(mean_avg_sq_dist) + + +def add_edges( + y: torch.Tensor, + topology: torch_geometric.data.Batch, + batch: torch.Tensor, + radial_cutoff: float, +) -> torch_geometric.data.Batch: + """Add edges to the graph based on the effective radial cutoff.""" + if topology.get("edge_index") is not None: + return topology + + topology = topology.clone() + with torch.cuda.nvtx.range("radial_graph"): + radial_edge_index = e3tools.radius_graph(y, radial_cutoff, batch) + + with torch.cuda.nvtx.range("concatenate_edges"): + edge_index = torch.cat((radial_edge_index, topology.bonded_edge_index), dim=-1) + if topology.bonded_edge_index.numel() == 0: + bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.device) + else: + bond_mask = torch.cat( + ( + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.device), + torch.ones(topology.bonded_edge_index.shape[1], dtype=torch.long, device=y.device), + ), + dim=0, + ) + + topology.edge_index = edge_index + topology.bond_mask = bond_mask + return topology + + +def apply_e3conv_to_positions(e3conv_model, pos, topology, batch, effective_radial_cutoff=5.0): + """ + Apply E3Conv model to a set of positions using existing graph topology. + + Args: + e3conv_model: E3Conv model instance + pos (torch.Tensor): Positions [N, 3] + topology (torch_geometric.data.Batch): Existing graph topology from dataloader + batch (torch.Tensor): Batch tensor from the graph + effective_radial_cutoff (float): Radial cutoff for edges + + Returns: + torch.Tensor: Node features [N, feature_dim] + """ + # Clone topology to avoid modifying original + topology_with_edges = topology.clone() + + # Add edges using the local add_edges function + topology_with_edges = add_edges(pos, topology_with_edges, batch, effective_radial_cutoff) + + # Use noise conditioning of 0.0 (no noise) + c_noise = torch.zeros(pos.shape[0], dtype=pos.dtype, device=pos.device) + + # Apply E3Conv + num_graphs = batch.max().item() + 1 # Number of graphs in the batch + node_features = e3conv_model( + pos=pos, + topology=topology_with_edges, + batch=batch, + num_graphs=num_graphs, + c_noise=c_noise, + effective_radial_cutoff=effective_radial_cutoff + ) + + return node_features + + +def create_e3conv_network(): + """ + Create an E3Conv network with parameters matching the yaml configuration. + + Returns: + E3Conv: Configured E3Conv network + """ + + # Hidden layer factory as specified in yaml + hidden_layer_factory = functools.partial( + e3tools.nn.ConvBlock, + conv=functools.partial(e3tools.nn.Conv) + ) + + # Output head factory as specified in yaml + output_head_factory = functools.partial( + e3tools.nn.EquivariantMLP, + irreps_hidden_list=["120x0e + 32x1e"] # Using irreps_hidden from yaml + ) + + # Create E3Conv with exact parameters from yaml + e3conv = E3Conv( + irreps_out="1x1e", # 3D vector output + irreps_hidden="120x0e + 32x1e", # Hidden representations + irreps_sh="1x0e + 1x1e", # Spherical harmonics + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + use_residue_information=True, # Assuming True, matches yaml ${data.use_residue_information} + n_layers=1, # Number of layers + edge_attr_dim=64, # Edge attribute dimension + atom_type_embedding_dim=8, # Atom type embedding + atom_code_embedding_dim=8, # Atom code embedding + residue_code_embedding_dim=32, # Residue code embedding + residue_index_embedding_dim=8, # Residue index embedding + use_residue_sequence_index=False, # As specified in yaml + num_atom_types=20, # Number of atom types + max_sequence_length=10, # Max sequence length + num_atom_codes=10, # Number of atom codes + num_residue_types=25, # Number of residue types + test_equivariance=False, # Disable for production + reduce=None # No reduction + ) + + return e3conv + + +def get_e3conv_output_irreps(): + """ + Get the output irreps of the E3Conv network. + + Returns: + str: Output irreps string + """ + return "1x1e" # 3D vector output as specified in yaml \ No newline at end of file diff --git a/scratch/transformer/pooling.py b/scratch/transformer/pooling.py new file mode 100644 index 0000000..c3a37bc --- /dev/null +++ b/scratch/transformer/pooling.py @@ -0,0 +1,217 @@ +#!/usr/bin/env python3 +""" +Lightning modules for converting node attributes between spatial and temporal representations. +""" + +import torch +import torch_geometric +import pytorch_lightning as pl + + +class SpatialToTemporalNodeAttr(pl.LightningModule): + """ + Lightning module to transfer node attributes from spatial nodes to temporal nodes + by repeating first temporal feature. + """ + + def __init__(self): + super().__init__() + + def forward(self, spatial_node_attr_temporal, temporal_batch): + """ + Transfer node attributes from spatial nodes to temporal nodes by repeating first temporal feature. + Takes the first temporal feature (t=0) and repeats it T times for each spatial node. + + Args: + spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs + + Returns: + torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] + """ + num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape + num_temporal_graphs = temporal_batch.num_graphs + + # Verify consistency + assert num_spatial_nodes == num_temporal_graphs, \ + f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" + + # Verify temporal length consistency + expected_temporal_nodes = temporal_batch.pos.shape[0] + expected_total_nodes = num_spatial_nodes * temporal_length + assert expected_total_nodes == expected_temporal_nodes, \ + f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" + + # Extract first temporal feature (t=0) and repeat it T times for each spatial node + first_temporal_features = spatial_node_attr_temporal[:, 0, :] # [N, attr_dim] + + # Repeat each spatial node's first temporal feature T times + temporal_node_attr = first_temporal_features.repeat_interleave(temporal_length, dim=0) # [N*T, attr_dim] + + # Verify the output shape matches the temporal batch + assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" + + return temporal_node_attr + + +class TemporalToSpatialNodeAttr(pl.LightningModule): + """ + Lightning module to convert temporal node attributes back to spatial node attributes. + Takes the first temporal node attribute from each temporal graph. + """ + + def __init__(self): + super().__init__() + + def forward(self, temporal_node_attr, temporal_batch): + """ + Convert temporal node attributes back to spatial node attributes. + Takes the first temporal node attribute from each temporal graph. + + Args: + temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs + + Returns: + torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] + """ + num_temporal_graphs = temporal_batch.num_graphs + attr_dim = temporal_node_attr.shape[1] + + # Extract the first node attribute from each temporal graph + spatial_node_attr = [] + + for graph_idx in range(num_temporal_graphs): + # Get the node range for this temporal graph + start_idx = temporal_batch.ptr[graph_idx] + + # The 0th node of each temporal graph is at the start of its range + first_node_attr = temporal_node_attr[start_idx] + spatial_node_attr.append(first_node_attr) + + # Stack to create spatial node attribute tensor + spatial_node_attr = torch.stack(spatial_node_attr) + + # Verify output shape + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" + + return spatial_node_attr + + +class TemporalToSpatialNodeAttrMean(pl.LightningModule): + """ + Lightning module to convert temporal node attributes back to spatial node attributes by averaging. + Takes the mean of all temporal node attributes for each temporal graph. + """ + + def __init__(self): + super().__init__() + + def forward(self, temporal_node_attr, temporal_batch): + """ + Convert temporal node attributes back to spatial node attributes by averaging. + Takes the mean of all temporal node attributes for each temporal graph. + + Args: + temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs + + Returns: + torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] + """ + num_temporal_graphs = temporal_batch.num_graphs + attr_dim = temporal_node_attr.shape[1] + + # Extract the mean node attributes from each temporal graph + spatial_node_attr = [] + + for graph_idx in range(num_temporal_graphs): + # Get the node range for this temporal graph + start_idx = temporal_batch.ptr[graph_idx] + end_idx = temporal_batch.ptr[graph_idx + 1] if graph_idx + 1 < len(temporal_batch.ptr) else len(temporal_node_attr) + + # Take the mean of all temporal nodes for this spatial node + temporal_nodes_attr = temporal_node_attr[start_idx:end_idx] # [temporal_length, attr_dim] + mean_node_attr = temporal_nodes_attr.mean(dim=0) # [attr_dim] + spatial_node_attr.append(mean_node_attr) + + # Stack to create spatial node attribute tensor + spatial_node_attr = torch.stack(spatial_node_attr) + + # Verify output shape + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" + + return spatial_node_attr + + +class SpatialTemporalToTemporalNodeAttr(pl.LightningModule): + """ + Lightning module to convert spatial node attributes arranged temporally to temporal node attributes. + Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. + """ + + def __init__(self): + super().__init__() + + def forward(self, spatial_node_attr_temporal, temporal_batch): + """ + Convert spatial node attributes arranged temporally to temporal node attributes. + Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. + + Args: + spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] + temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs for validation + + Returns: + torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] + """ + num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape + num_temporal_graphs = temporal_batch.num_graphs + + # Verify consistency with temporal batch + assert num_spatial_nodes == num_temporal_graphs, \ + f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" + + # Verify temporal length consistency + expected_temporal_nodes = temporal_batch.pos.shape[0] + expected_total_nodes = num_spatial_nodes * temporal_length + assert expected_total_nodes == expected_temporal_nodes, \ + f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" + + # Reshape to match temporal graph ordering: [N, T, features] -> [N*T, features] + # Temporal graph arranges nodes as: [node0_t0, node0_t1, ..., node0_tT-1, node1_t0, ...] + temporal_node_attr = spatial_node_attr_temporal.reshape(num_spatial_nodes * temporal_length, attr_dim) + + # Verify the output shape matches the temporal batch + assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" + + return temporal_node_attr + + +# Legacy function interfaces for backward compatibility +def spatial_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = SpatialToTemporalNodeAttr() + return module(spatial_node_attr_temporal, temporal_batch) + + +def temporal_to_spatial_node_attr(temporal_node_attr, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = TemporalToSpatialNodeAttr() + return module(temporal_node_attr, temporal_batch) + + +def temporal_to_spatial_node_attr_mean(temporal_node_attr, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = TemporalToSpatialNodeAttrMean() + return module(temporal_node_attr, temporal_batch) + + +def spatial_temporal_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch): + """Legacy function interface for backward compatibility.""" + module = SpatialTemporalToTemporalNodeAttr() + return module(spatial_node_attr_temporal, temporal_batch) diff --git a/scratch/transformer/temporal_transformer.py b/scratch/transformer/temporal_transformer.py new file mode 100644 index 0000000..07852c9 --- /dev/null +++ b/scratch/transformer/temporal_transformer.py @@ -0,0 +1,304 @@ +from typing import Dict, Union + +import e3nn +import torch +import torch.nn as nn +from e3nn import o3 +import torch_geometric.data +import e3tools +import e3tools.nn + + +class E3Transformer(nn.Module): + """E(3)-equivariant transformer with temporal graph support.""" + + def __init__( + self, + irreps_out: Union[str, e3nn.o3.Irreps], + irreps_hidden: Union[str, e3nn.o3.Irreps], + irreps_sh: Union[str, e3nn.o3.Irreps], + irreps_node_attr: Union[str, e3nn.o3.Irreps], + num_layers: int, + edge_attr_dim: int, + num_attention_heads: int, + reduce: str | None = None, + ): + super().__init__() + + self.irreps_out = o3.Irreps(irreps_out) + self.irreps_hidden = o3.Irreps(irreps_hidden) + self.irreps_sh = o3.Irreps(irreps_sh) + self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps + self.num_layers = num_layers + self.edge_attr_dim = edge_attr_dim + self.num_attention_heads = num_attention_heads + self.reduce = reduce + self.sh = o3.SphericalHarmonics( + irreps_out=self.irreps_sh, normalize=True, normalization="component" + ) + # Split edge attribute dimensions: radial and temporal (bondedness is optional) + self.radial_edge_attr_dim = self.edge_attr_dim // 2 + self.temporal_edge_attr_dim = self.edge_attr_dim - self.radial_edge_attr_dim + + # Optional bondedness embedding (only used if bond_mask exists in graph) + self.embed_bondedness = nn.Embedding(2, self.edge_attr_dim // 3) + + # Gate for combining node attributes with temporal position + # Input: node_attr (from data) + temporal_position (1x0e scalar) + irreps_with_temporal = self.irreps_node_attr + o3.Irreps("1x0e") + self.temporal_gate = e3tools.nn.GateWrapper(irreps_in=irreps_with_temporal, \ + irreps_out=self.irreps_hidden, \ + irreps_gate=irreps_with_temporal,) + # self.initial_linear = o3.Linear( + # self.temporal_gate.irreps_out, self.irreps_hidden + # ) + + self.layers = nn.ModuleList() + for _ in range(num_layers): + self.layers.append( + e3tools.nn.TransformerBlock( + irreps_in=self.irreps_hidden, + irreps_out=self.irreps_hidden, + irreps_sh=self.irreps_sh, + edge_attr_dim=self.edge_attr_dim, + num_heads=self.num_attention_heads, + ) + ) + self.output_head = e3tools.nn.EquivariantMLP( + irreps_in=self.irreps_hidden, + irreps_out=self.irreps_out, + irreps_hidden_list=[self.irreps_hidden], + ) + + def forward( + self, + node_attr: torch.Tensor, + temporal_graph: torch_geometric.data.Batch, + effective_radial_cutoff: float, + temporal_cutoff: float = 1.0, + ) -> torch.Tensor: + """Forward pass of the E3Transformer model.""" + # Extract graph data + pos = temporal_graph.pos + edge_index = temporal_graph.edge_index + temporal_position = temporal_graph.temporal_position + batch = temporal_graph.batch + num_graphs = temporal_graph.num_graphs + + src, dst = edge_index + edge_vec = pos[src] - pos[dst] + edge_sh = self.sh(edge_vec) + + # Compute edge attributes: radial and temporal + radial_edge_attr = e3nn.math.soft_one_hot_linspace( + edge_vec.norm(dim=1), + 0.0, + effective_radial_cutoff, + self.radial_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + + # Temporal edge attributes from temporal_position differences + temporal_edge_vec = temporal_position[src] - temporal_position[dst] + temporal_edge_attr = e3nn.math.soft_one_hot_linspace( + temporal_edge_vec.abs(), # Use absolute difference + 0.0, + temporal_cutoff, + self.temporal_edge_attr_dim, + basis="gaussian", + cutoff=True, + ) + + # Optional bondedness (if bond_mask exists in the temporal graph) + if hasattr(temporal_graph, 'bond_mask') and temporal_graph.bond_mask is not None: + bonded_edge_attr = self.embed_bondedness(temporal_graph.bond_mask) + edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr, temporal_edge_attr), dim=-1) + else: + edge_attr = torch.cat((radial_edge_attr, temporal_edge_attr), dim=-1) + + # Process node attributes with temporal gating + + # Concatenate node_attr with temporal_position (scalar) + temporal_position_expanded = temporal_position.unsqueeze(-1) # [N, 1] for concatenation + node_attr_with_temporal = torch.cat([node_attr, temporal_position_expanded], dim=-1) + + # Apply temporal gate + node_attr_processed = self.temporal_gate(node_attr_with_temporal) + # node_attr_processed = self.initial_linear(node_attr_gated) + + # Perform message passing with gated node attributes + for layer in self.layers: + node_attr_processed = layer(node_attr_processed, edge_index, edge_attr, edge_sh) + node_attr_processed = self.output_head(node_attr_processed) + + # Pool over nodes. + if self.reduce is not None: + node_attr_processed = e3tools.scatter( + node_attr_processed, + index=batch, + dim=0, + dim_size=num_graphs, + reduce=self.reduce, + ) + + return node_attr_processed + + +class E3SpatioTemporal(nn.Module): + """ + E(3)-equivariant spatio-temporal model that combines spatial and temporal processing. + + This model implements the complete workflow: + 1. Process input spatial graph and hidden states through spatial module + 2. Pool spatial features to temporal graph representation + 3. Process temporal graph through temporal module + 4. Pool temporal features back to spatial representation + 5. Convert temporal graph back to spatial graph + """ + + def __init__( + self, + spatial_module: nn.Module, + temporal_module: nn.Module, + spatial_to_temporal_pooler: nn.Module, + temporal_to_spatial_pooler: nn.Module, + radial_cutoff: float, + temporal_cutoff: float = 1.0, + ): + """ + Initialize the E3SpatioTemporal model. + + Args: + spatial_module: Module for processing spatial positions (e.g., E3Conv) + temporal_module: Module for processing temporal graphs (e.g., E3Transformer) + spatial_to_temporal_pooler: Module to convert spatial-temporal features to temporal node attributes + temporal_to_spatial_pooler: Module to convert temporal features back to spatial features + radial_cutoff: Cutoff for spatial radial edge weights + temporal_cutoff: Cutoff for temporal edge weights + """ + super().__init__() + + self.spatial_module = spatial_module + self.temporal_module = temporal_module + self.spatial_to_temporal_pooler = spatial_to_temporal_pooler + self.temporal_to_spatial_pooler = temporal_to_spatial_pooler + self.radial_cutoff = radial_cutoff + self.temporal_cutoff = temporal_cutoff + + def forward( + self, + batch: torch_geometric.data.Batch, + c_noise: torch.Tensor, + return_temporal_features: bool = False, + return_temporal_graph: bool = False, + ) -> Union[torch.Tensor, Dict[str, torch.Tensor]]: + """ + Forward pass implementing the complete spatio-temporal workflow. + + Args: + batch: Input spatial graph batch with pos, batch, num_graphs, and optionally hidden_state + c_noise: Noise conditioning tensor + return_temporal_features: Whether to return intermediate temporal features + return_temporal_graph: Whether to return the temporal graph + + Returns: + If return_temporal_features or return_temporal_graph is True, returns dict with: + - 'spatial_features': Final spatial features + - 'spatial_graph': Output spatial graph + - 'temporal_features': Temporal features (if requested) + - 'temporal_graph': Temporal graph (if requested) + Otherwise returns just the final spatial features tensor + """ + from convert_spatiotemporal import spatial_to_temporal_graphs, temporal_to_spatial_graphs + + # Store original device + device = batch.pos.device + + # Step 1: Convert spatial graph to temporal graphs + temporal_batch = spatial_to_temporal_graphs(batch) + + # Step 2: Process all positions (current + hidden states) with spatial module + # Create topology for spatial processing (without positions) + topology = batch.clone() + # Remove position-dependent attributes but keep graph structure + if hasattr(topology, 'pos'): + del topology.pos + if hasattr(topology, 'batch'): + del topology.batch + if hasattr(topology, 'num_graphs'): + del topology.num_graphs + + node_attr_list = [] + + # Process current positions + node_attr_current = self.spatial_module( + batch.pos, + topology, + batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff + ).unsqueeze(1) # [N, 1, features] + node_attr_list.append(node_attr_current) + + # Process hidden state positions if they exist + if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + for hidden_pos in batch.hidden_state: + node_attr_hidden = self.spatial_module( + hidden_pos, + topology, + batch.batch, + num_graphs=batch.num_graphs, + c_noise=c_noise, + effective_radial_cutoff=self.radial_cutoff + ).unsqueeze(1) # [N, 1, features] + node_attr_list.append(node_attr_hidden) + + # Step 3: Stack spatial-temporal features + node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) # [N, T, features] + + # Step 4: Convert spatial-temporal features to temporal node attributes + temporal_node_attr = self.spatial_to_temporal_pooler(node_attr_spatial_temporal, temporal_batch) + + # Step 5: Process temporal graph through temporal module + temporal_output = self.temporal_module( + temporal_node_attr, + temporal_batch, + self.radial_cutoff, + self.temporal_cutoff + ) + + # Step 6: Pool temporal features back to spatial features + spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) + + # Step 7: Convert temporal graph back to spatial graph + output_spatial_graph = temporal_to_spatial_graphs(temporal_batch) + + # Prepare return values + if return_temporal_features or return_temporal_graph: + result = { + 'spatial_features': spatial_features, + 'spatial_graph': output_spatial_graph, + } + if return_temporal_features: + result['temporal_features'] = temporal_output + if return_temporal_graph: + result['temporal_graph'] = temporal_batch + return result + else: + return spatial_features + + def get_spatial_output_irreps(self): + """Get the irreps of the spatial module output.""" + if hasattr(self.spatial_module, 'irreps_out'): + return self.spatial_module.irreps_out + else: + raise AttributeError("Spatial module does not have irreps_out attribute") + + def get_temporal_output_irreps(self): + """Get the irreps of the temporal module output.""" + if hasattr(self.temporal_module, 'irreps_out'): + return self.temporal_module.irreps_out + else: + raise AttributeError("Temporal module does not have irreps_out attribute") \ No newline at end of file diff --git a/scratch/transformer/test_e3_spatiotemporal.py b/scratch/transformer/test_e3_spatiotemporal.py new file mode 100644 index 0000000..369238a --- /dev/null +++ b/scratch/transformer/test_e3_spatiotemporal.py @@ -0,0 +1,214 @@ +#!/usr/bin/env python3 +""" +Test script for the E3SpatioTemporal model. + +This script tests the unified E3SpatioTemporal model that encapsulates +the complete spatio-temporal processing workflow. +""" + +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) +import torch +import torch_geometric +from jamun.data import parse_datasets_from_directory +from jamun.utils import unsqueeze_trailing + +# Import modules needed for the test +from helpers import create_e3conv_network, add_edges +from temporal_transformer import E3SpatioTemporal, E3Transformer +from pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean + +def setup_device(): + """Setup CUDA device if available.""" + device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') + print(f"Using device: {device}") + if torch.cuda.is_available(): + print(f"CUDA device: {torch.cuda.get_device_name()}") + print(f"CUDA memory: {torch.cuda.get_device_properties(device).total_memory / 1e9:.1f} GB") + return device + +def move_graph_to_device(graph, device): + """Move a PyTorch Geometric graph and all its tensor attributes to device.""" + # Move the graph using standard .to() method + graph = graph.to(device) + + # Manually move any custom tensor attributes that might not be handled + for attr_name in dir(graph): + if not attr_name.startswith('_'): # Skip private attributes + attr_value = getattr(graph, attr_name, None) + if isinstance(attr_value, torch.Tensor): + setattr(graph, attr_name, attr_value.to(device)) + + return graph + +def load_test_data(device): + """Load and prepare test data.""" + print("Loading test data...") + + dataset = parse_datasets_from_directory( + root="/data2/sules/ALA_ALA_enhanced_full_grid/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + subsample=1, + total_lag_time=5, + lag_subsample_rate=1, + max_datasets=3, + num_frames=10 + ) + + # Get first graph and create batch + graph = dataset[0].__getitem__(0) + batch = torch_geometric.data.Batch.from_data_list([graph]) + batch = add_edges(batch.pos, batch, batch.batch, 0.05) + + # Move to device + batch = move_graph_to_device(batch, device) + print(f"Loaded batch with {batch.pos.shape[0]} nodes on device: {batch.pos.device}") + + return batch + +def create_spatiotemporal_model(device): + """Create and configure the E3SpatioTemporal model.""" + print("Creating E3SpatioTemporal model...") + + # Create component modules + spatial_module = create_e3conv_network().to(device) + + temporal_module = E3Transformer( + irreps_out="3x1e", # 3D output (like positions) + irreps_hidden="8x0e + 4x1e", # Hidden representations + irreps_sh="1x0e + 1x1e", # Spherical harmonics + irreps_node_attr="1x1e", # Input node attributes match E3Conv output + num_layers=2, + edge_attr_dim=24, # Split into 2 parts: 12+12 (radial+temporal) + num_attention_heads=1, # Single attention head for simpler test + ).to(device) + + spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr() + temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean() + + # Create the unified model + spatiotemporal_model = E3SpatioTemporal( + spatial_module=spatial_module, + temporal_module=temporal_module, + spatial_to_temporal_pooler=spatial_to_temporal_pooler, + temporal_to_spatial_pooler=temporal_to_spatial_pooler, + radial_cutoff=0.05, + temporal_cutoff=1.0, + ).to(device) + + print(f"Created E3SpatioTemporal model on device: {next(spatiotemporal_model.parameters()).device}") + return spatiotemporal_model + +def test_spatiotemporal_model(model, batch, device): + """Test the E3SpatioTemporal model with various configurations.""" + print("=" * 50) + print("TESTING E3SPATIOTEMPORAL MODEL") + print("=" * 50) + + # Print model information + print(f"Model components:") + print(f" - Spatial module output irreps: {model.get_spatial_output_irreps()}") + print(f" - Temporal module output irreps: {model.get_temporal_output_irreps()}") + print(f" - Radial cutoff: {model.radial_cutoff}") + print(f" - Temporal cutoff: {model.temporal_cutoff}") + + # Prepare noise conditioning + sigma = torch.tensor(0.0, device=device) + sigma = unsqueeze_trailing(sigma, 1) + + print(f"\nInput batch:") + print(f" - pos shape: {batch.pos.shape}") + print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") + print(f" - batch device: {batch.pos.device}") + + success = True + + try: + with torch.no_grad(): + print("\n1. Testing simple forward pass (spatial features only)...") + spatial_features = model(batch, sigma) + print(f" āœ… Success! Spatial features shape: {spatial_features.shape}") + print(f" Spatial features device: {spatial_features.device}") + print(f" Spatial features norm: {torch.norm(spatial_features):.6f}") + + print("\n2. Testing full forward pass (all outputs)...") + results = model( + batch, + sigma, + return_temporal_features=True, + return_temporal_graph=True + ) + + print(f" āœ… Success! Full output results:") + print(f" - spatial_features shape: {results['spatial_features'].shape}") + print(f" - temporal_features shape: {results['temporal_features'].shape}") + print(f" - temporal_graph num_graphs: {results['temporal_graph'].num_graphs}") + print(f" - spatial_graph pos shape: {results['spatial_graph'].pos.shape}") + + # Verify spatial graph reconstruction + pos_difference = torch.norm(results['spatial_graph'].pos - batch.pos) + print(f" - Spatial position reconstruction error: {pos_difference:.6f}") + + print("\n3. Testing consistency between simple and full forward pass...") + simple_features = spatial_features + full_features = results['spatial_features'] + consistency_error = torch.norm(simple_features - full_features) + print(f" Consistency error: {consistency_error:.6f}") + + if consistency_error < 1e-6: + print(" āœ… Results are consistent!") + else: + print(" āš ļø Results differ between simple and full forward pass") + success = False + + except Exception as e: + print(f"\nāŒ Test failed: {e}") + import traceback + traceback.print_exc() + success = False + + return success + +def main(): + """Main test function.""" + print("E3SpatioTemporal Model Test") + print("=" * 50) + + # Setup + device = setup_device() + batch = load_test_data(device) + model = create_spatiotemporal_model(device) + + # Run tests + success = test_spatiotemporal_model(model, batch, device) + + # Final summary + print("\n" + "=" * 50) + print("FINAL RESULTS") + print("=" * 50) + + # Device summary + print(f"Device Summary:") + print(f" - Used device: {device}") + if torch.cuda.is_available(): + print(f" - CUDA memory allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") + print(f" - CUDA memory cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") + + print(f"\nTest Results:") + if success: + print("šŸŽ‰ ALL TESTS PASSED! The E3SpatioTemporal model works correctly!") + print("\nThe model successfully:") + print(" āœ… Processes spatial graphs with hidden states") + print(" āœ… Converts to temporal representation") + print(" āœ… Applies temporal transformations") + print(" āœ… Pools back to spatial features") + print(" āœ… Reconstructs spatial graphs") + print(" āœ… Maintains consistency across different call patterns") + else: + print("āŒ SOME TESTS FAILED! Check the output above for details.") + + return success + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/transformer/test_spatiotemporal_conditioner.py b/scratch/transformer/test_spatiotemporal_conditioner.py new file mode 100755 index 0000000..055fc8b --- /dev/null +++ b/scratch/transformer/test_spatiotemporal_conditioner.py @@ -0,0 +1,433 @@ +#!/usr/bin/env python3 +""" +Test script for loading and testing a conditional denoiser with spatiotemporal conditioner. +Uses the new approach where SpatioTemporalConditioner outputs [y.pos, spatial_features] +and E3ConvConditionalSpatioTemporal handles concatenated inputs. +""" + +import e3nn +e3nn.set_optimization_defaults(jit_script_fx=False) + +import torch +import torch_geometric +from typing import Dict, Any +import sys +import os + +# Add the src directory to path to import jamun modules +sys.path.insert(0, 'src') + +from jamun.data import parse_datasets_from_directory +from jamun.model.denoiser_conditional import Denoiser # Changed from DenoiserWithInputAttr +from jamun.model.denoiser import add_edges # Import add_edges function +from jamun.model.conditioners.conditioners import SpatioTemporalConditioner +from jamun.model.arch.spatiotemporal import E3SpatioTemporal, E3Transformer +from jamun.model.arch.e3conv import E3Conv +from jamun.model.arch.e3conv_conditional import E3ConvConditionalSpatioTemporal # Changed from E3ConvConditionalWithInputAttr +from jamun.model.pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean +from jamun.distributions._distributions import ConstantSigma +from jamun.utils import unsqueeze_trailing +from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets # Import temporal function + +# Setup device +device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +print(f"Using device: {device}") + +def create_spatial_module() -> E3Conv: + """Create E3Conv spatial module with reasonable parameters.""" + import functools + import e3tools + + # Create factory functions + hidden_layer_factory = functools.partial( + e3tools.nn.ConvBlock, + conv=functools.partial(e3tools.nn.Conv) + ) + + output_head_factory = functools.partial( + e3tools.nn.EquivariantMLP, + irreps_hidden_list=["120x0e + 32x1e"] + ) + + return E3Conv( + irreps_out="3x1e", # Changed to match temporal module input + irreps_hidden="120x0e + 32x1e", + irreps_sh="1x0e + 1x1e", + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + n_layers=1, + edge_attr_dim=64, + use_residue_information=True, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=32, + residue_index_embedding_dim=8, + use_residue_sequence_index=False, + num_atom_types=20, + max_sequence_length=10, + num_atom_codes=10, + num_residue_types=25, + test_equivariance=False, + reduce=None + ) + +def create_temporal_module() -> E3Transformer: + """Create E3Transformer temporal module.""" + return E3Transformer( + irreps_out="3x1e", # Final spatial features output + irreps_hidden="8x0e + 4x1e", + irreps_sh="1x0e + 1x1e", + irreps_node_attr="3x1e", # Match spatial module output + num_layers=2, + edge_attr_dim=24, + num_attention_heads=1, + reduce=None + ) + +def create_spatiotemporal_model() -> E3SpatioTemporal: + """Create the complete E3SpatioTemporal model.""" + spatial_module = create_spatial_module() + temporal_module = create_temporal_module() + + # Create pooling modules + spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr(irreps_out="3x1e") # Match spatial module output + temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean(irreps_out="3x1e") # Match temporal module output + + # Compute radial cutoff using temporal average squared distance + print("Computing radial cutoff from temporal dataset...") + try: + # Load dataset to compute temporal average squared distance + dataset = parse_datasets_from_directory( + root="/data2/sules/ALA_ALA_enhanced_full_grid/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + subsample=1, + total_lag_time=5, + lag_subsample_rate=1, + max_datasets=2, # Keep small for testing + num_frames=5 # Small number of frames + ) + + # Compute temporal average squared distance + temporal_avg_sq_dist = compute_temporal_average_squared_distance_from_datasets( + [dataset], # Pass as list since function expects multiple datasets + num_samples=50, # Use fewer samples for testing + verbose=True + ) + + # Use a multiple of the temporal average squared distance as the radial cutoff + # Typically we might use sqrt(temporal_avg_sq_dist) * some_factor + import math + radial_cutoff = math.sqrt(temporal_avg_sq_dist) * 2.0 # Scale factor of 2.0 + print(f"Computed radial cutoff: {radial_cutoff:.6f} nm") + + except Exception as e: + print(f"Warning: Failed to compute temporal cutoff ({e}), using default value 0.05") + radial_cutoff = 0.05 + + return E3SpatioTemporal( + spatial_module=spatial_module, + temporal_module=temporal_module, + spatial_to_temporal_pooler=spatial_to_temporal_pooler, + temporal_to_spatial_pooler=temporal_to_spatial_pooler, + radial_cutoff=radial_cutoff, + temporal_cutoff=1.0 + ) + +def create_spatiotemporal_conditioner() -> SpatioTemporalConditioner: + """Create SpatioTemporalConditioner with E3SpatioTemporal model.""" + spatiotemporal_model = create_spatiotemporal_model() + + return SpatioTemporalConditioner( + N_structures=1, # Changed to 2 for [y.pos, spatial_features] + spatiotemporal_model=spatiotemporal_model, + c_noise=0.0, + freeze_spatiotemporal_model=False # Keep trainable + ) + +def create_conditional_denoiser_config() -> Dict[str, Any]: + """Create configuration for Denoiser with spatiotemporal conditioner.""" + import functools + import e3tools.nn + + def create_arch(): + """Create the E3ConvConditionalSpatioTemporal architecture module.""" + # Hidden layer factory + hidden_layer_factory = functools.partial( + e3tools.nn.ConvBlock, + conv=functools.partial(e3tools.nn.Conv) + ) + + # Output head factory + output_head_factory = functools.partial( + e3tools.nn.EquivariantMLP, + irreps_hidden_list=["16x0e + 8x1e"] + ) + + return E3ConvConditionalSpatioTemporal( + irreps_out="1x1e", # Output should be 3 components (1x1e) to match position + irreps_hidden="16x0e + 8x1e", + irreps_sh="1x0e + 1x1e", + hidden_layer_factory=hidden_layer_factory, + output_head_factory=output_head_factory, + n_layers=2, + edge_attr_dim=32, + use_residue_information=True, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=16, + residue_index_embedding_dim=8, + use_residue_sequence_index=False, + num_atom_types=20, + max_sequence_length=10, + num_atom_codes=10, + num_residue_types=25, + test_equivariance=False, + reduce=None, + N_structures=1, # Changed to 2 for [y.pos, spatial_features] + input_attr_irreps="3x1e", # spatial_features only (9 components = 3x1e) + ) + + def create_optim(params): + """Create the optimizer.""" + return torch.optim.Adam(params, lr=0.001) + + return { + # Required Denoiser parameters (changed from DenoiserWithInputAttr) + 'arch': create_arch, + 'optim': create_optim, + 'sigma_distribution': ConstantSigma(sigma=0.1), + 'max_radius': 1000.0, + 'average_squared_distance': 10.0, # Dummy value for testing + 'add_fixed_noise': False, + 'add_fixed_ones': False, + 'align_noisy_input_during_training': True, + 'align_noisy_input_during_evaluation': True, + 'mean_center': True, + 'mirror_augmentation_rate': 0.0, + 'bond_loss_coefficient': 1.0, + 'normalization_type': "JAMUN", + 'sigma_data': None, + 'lr_scheduler_config': None, + 'use_torch_compile': False, # Disable for testing + 'torch_compile_kwargs': None, + 'conditioner': create_spatiotemporal_conditioner() + } + +def add_edges_to_batch(batch: torch_geometric.data.Batch, cutoff: float = 0.05) -> torch_geometric.data.Batch: + """Add edges to batch using existing utility from denoiser.""" + # Use e3tools radius_graph directly since we don't need the full denoiser add_edges logic + import e3tools + + if hasattr(batch, 'edge_index') and batch.edge_index is not None: + return batch + + # Add radius-based edges + edge_index = e3tools.radius_graph(batch.pos, cutoff, batch.batch) + batch.edge_index = edge_index + + # Add bonded edges if they exist + if hasattr(batch, 'bonded_edge_index') and batch.bonded_edge_index is not None: + bond_mask = torch.cat([ + torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device), + torch.ones(batch.bonded_edge_index.shape[1], dtype=torch.long, device=batch.pos.device) + ]) + batch.edge_index = torch.cat([edge_index, batch.bonded_edge_index], dim=1) + batch.bond_mask = bond_mask + else: + batch.bond_mask = torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device) + + return batch + +def load_test_data(): + """Load ALA_ALA test dataset.""" + print("Loading ALA_ALA dataset...") + + dataset = parse_datasets_from_directory( + root="/data2/sules/ALA_ALA_enhanced_full_grid/train", + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + subsample=1, + total_lag_time=5, + lag_subsample_rate=1, + max_datasets=2, # Keep small for testing + num_frames=5 # Small number of frames + ) + + print(f"Loaded dataset with {len(dataset)} samples") + + # Get a sample and create batch + graph = dataset[0].__getitem__(0) + batch = torch_geometric.data.Batch.from_data_list([graph]) + + # Add edges + batch = add_edges_to_batch(batch, cutoff=0.05) + + # Move to device + batch = batch.to(device) + + print(f"Batch info:") + print(f" - pos shape: {batch.pos.shape}") + print(f" - edge_index shape: {batch.edge_index.shape}") + print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") + if hasattr(batch, 'hidden_state') and batch.hidden_state: + print(f" - hidden_state[0] shape: {batch.hidden_state[0].shape}") + + return batch + +def test_spatiotemporal_conditioner(conditioner: SpatioTemporalConditioner, batch: torch_geometric.data.Batch): + """Test the spatiotemporal conditioner.""" + print("\n" + "="*50) + print("TESTING SPATIOTEMPORAL CONDITIONER") + print("="*50) + + try: + # Test forward pass + conditioned_structures = conditioner(batch) + + print(f"āœ… Conditioner forward pass successful!") + print(f"Number of conditioned structures: {len(conditioned_structures)} (expected: 2)") + print(f"First structure (y.pos) shape: {conditioned_structures[0].shape}") + print(f"Second structure (spatial_features) shape: {conditioned_structures[1].shape}") + print(f"Original position shape: {batch.pos.shape}") + print(f"Position difference norm: {torch.norm(conditioned_structures[0] - batch.pos):.6f}") + + # Verify we got exactly two structures + assert len(conditioned_structures) == 2, f"Expected 2 structures, got {len(conditioned_structures)}" + + return True, conditioned_structures + + except Exception as e: + print(f"āŒ Conditioner test failed: {e}") + import traceback + traceback.print_exc() + return False, None + +def test_conditional_denoiser_creation(): + """Test creating Denoiser with spatiotemporal conditioner.""" + print("\n" + "="*50) + print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER CREATION") + print("="*50) + + try: + # Create configuration + config = create_conditional_denoiser_config() + + # Create denoiser (this will instantiate all components) + denoiser = Denoiser(**config) + denoiser = denoiser.to(device) + + print(f"āœ… Denoiser created successfully!") + print(f"Denoiser device: {next(denoiser.parameters()).device}") + print(f"Has conditioner: {hasattr(denoiser, 'conditioning_module')}") + print(f"Architecture type: {type(denoiser.g).__name__}") + print(f"Conditioner type: {type(denoiser.conditioning_module).__name__}") + + # Check if spatiotemporal model is properly set up + if hasattr(denoiser.conditioning_module, 'spatiotemporal_model'): + st_model = denoiser.conditioning_module.spatiotemporal_model + print(f"SpatioTemporal model type: {type(st_model).__name__}") + print(f"Spatial module type: {type(st_model.spatial_module).__name__}") + print(f"Temporal module type: {type(st_model.temporal_module).__name__}") + + return True, denoiser + + except Exception as e: + print(f"āŒ Denoiser creation failed: {e}") + import traceback + traceback.print_exc() + return False, None + +def test_denoiser_forward_pass(denoiser: Denoiser, batch: torch_geometric.data.Batch): + """Test the complete denoiser forward pass.""" + print("\n" + "="*50) + print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER FORWARD PASS") + print("="*50) + + try: + # Test with sigma = 0.1 + sigma = 0.1 + + # Debug: check conditioned structures shapes + conditioned_structures = denoiser.conditioning_module(batch) + print(f"DEBUG: Conditioned structures shapes:") + for i, struct in enumerate(conditioned_structures): + print(f" Structure {i}: {struct.shape}") + + concatenated = torch.cat([*conditioned_structures], dim=-1) + print(f"DEBUG: Concatenated shape: {concatenated.shape}") + print(f"DEBUG: Expected irreps: 4x1e = 12 components") + + with torch.no_grad(): + xhat_batch = denoiser.xhat(batch, sigma) + + print(f"āœ… Denoiser forward pass successful!") + print(f"Input shape: {batch.pos.shape}") + print(f"Output shape: {xhat_batch.pos.shape}") + print(f"Output norm: {torch.norm(xhat_batch.pos):.6f}") + print(f"Used sigma: {sigma}") + + # Verify output shapes match input + assert xhat_batch.pos.shape == batch.pos.shape, f"Shape mismatch: {xhat_batch.pos.shape} vs {batch.pos.shape}" + + return True + + except Exception as e: + print(f"āŒ Denoiser forward pass failed: {e}") + import traceback + traceback.print_exc() + return False + +def main(): + """Main test function.""" + print("="*60) + print("CONDITIONAL DENOISER WITH SPATIOTEMPORAL CONDITIONER TEST") + print("="*60) + + # Load test data + batch = load_test_data() + + # Test conditioner creation and forward pass + conditioner = create_spatiotemporal_conditioner() + conditioner = conditioner.to(device) + + conditioner_success, conditioned_structures = test_spatiotemporal_conditioner(conditioner, batch) + + if not conditioner_success: + print("āŒ Conditioner test failed, stopping here.") + return + + # Test complete denoiser creation + denoiser_success, denoiser = test_conditional_denoiser_creation() + + if not denoiser_success: + print("āŒ Denoiser creation failed, stopping here.") + return + + # Test complete forward pass + forward_success = test_denoiser_forward_pass(denoiser, batch) + + # Final summary + print("\n" + "="*60) + print("FINAL SUMMARY") + print("="*60) + + print(f"Test Results:") + print(f" - Conditioner test: {'āœ… PASSED' if conditioner_success else 'āŒ FAILED'}") + print(f" - Denoiser creation: {'āœ… PASSED' if denoiser_success else 'āŒ FAILED'}") + print(f" - Forward pass test: {'āœ… PASSED' if forward_success else 'āŒ FAILED'}") + + if conditioner_success and denoiser_success and forward_success: + print("\nšŸŽ‰ ALL TESTS PASSED!") + print("The conditional denoiser with spatiotemporal conditioner is working correctly!") + else: + print("\nāš ļø Some tests failed. Check the output above for details.") + + # Device memory summary + if torch.cuda.is_available(): + print(f"\nCUDA Memory:") + print(f" - Allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") + print(f" - Cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") + +if __name__ == "__main__": + main() \ No newline at end of file diff --git a/scratch/visualize_fake_enhanced_data.py b/scratch/visualize_fake_enhanced_data.py new file mode 100644 index 0000000..38e0fe1 --- /dev/null +++ b/scratch/visualize_fake_enhanced_data.py @@ -0,0 +1,247 @@ +import glob +import os +import itertools +import re +from collections import defaultdict + +import mdtraj as md +import matplotlib.pyplot as plt +import numpy as np +from tqdm import tqdm +import matplotlib.colors as colors + +def parse_grid_code_from_filename(filename): + """ + Parse grid code from trajectory filename of format traj_{grid_code}_{traj_code}.xtc + """ + basename = os.path.basename(filename) + match = re.match(r'^traj_(\d+)_(\d+)\.xtc$', basename) + if match: + return int(match.group(1)), int(match.group(2)) + return None, None + +def select_trajectories_with_max_per_grid(traj_files, max_traj_per_grid): + """ + Select trajectories ensuring no grid code has more than max_traj_per_grid trajectories. + """ + grid_trajectories = defaultdict(list) + + # Group trajectories by grid code + for traj_file in traj_files: + grid_code, traj_code = parse_grid_code_from_filename(traj_file) + if grid_code is not None: + grid_trajectories[grid_code].append((traj_file, traj_code)) + + print(f"Found {len(grid_trajectories)} unique grid codes") + + # Limit trajectories per grid code + selected_files = [] + grid_stats = {} + + for grid_code, traj_list in grid_trajectories.items(): + # Sort by trajectory code for deterministic selection + traj_list.sort(key=lambda x: x[1]) + + # Select up to max_traj_per_grid trajectories + selected_count = min(len(traj_list), max_traj_per_grid) + selected_for_grid = traj_list[:selected_count] + + grid_stats[grid_code] = { + 'total': len(traj_list), + 'selected': selected_count + } + + for traj_file, _ in selected_for_grid: + selected_files.append(traj_file) + + # Print statistics + print(f"\nGrid code statistics:") + print(f"Total grid codes: {len(grid_stats)}") + total_original = sum(stats['total'] for stats in grid_stats.values()) + total_selected = sum(stats['selected'] for stats in grid_stats.values()) + print(f"Total trajectories: {total_original} -> {total_selected}") + print(f"Max trajectories per grid: {max_traj_per_grid}") + + # Show distribution + selected_counts = [stats['selected'] for stats in grid_stats.values()] + print(f"Distribution of selected trajectories per grid:") + for count in sorted(set(selected_counts)): + num_grids = sum(1 for c in selected_counts if c == count) + print(f" {count} trajectories: {num_grids} grid codes") + + return sorted(selected_files) + +def create_ramachandran_plot(traj_path, topology, output_dir): + """ + Loads a trajectory, computes phi and psi angles, and saves a Ramachandran plot. + """ + # Load trajectory + try: + traj = md.load(traj_path, top=topology) + except Exception as e: + print(f"Could not load trajectory {traj_path}. Error: {e}") + return + + # Compute dihedral angles + phi_indices, phi_angles = md.compute_phi(traj) + psi_indices, psi_angles = md.compute_psi(traj) + + # Convert radians to degrees + phi_degrees = np.rad2deg(phi_angles.flatten()) + psi_degrees = np.rad2deg(psi_angles.flatten()) + + # Create plot + plt.figure(figsize=(8, 8)) + # Use hexbin for a nicer look + plt.hexbin(phi_degrees, psi_degrees, gridsize=180, cmap='viridis', mincnt=1) + plt.colorbar(label='Count in bin') + plt.title(f'Ramachandran Plot for {os.path.basename(traj_path)}') + plt.xlabel('Phi (degrees)') + plt.ylabel('Psi (degrees)') + plt.xlim(-180, 180) + plt.ylim(-180, 180) + plt.grid(True, linestyle='--', alpha=0.6) + plt.axhline(0, color='k', linestyle='--', linewidth=0.5) + plt.axvline(0, color='k', linestyle='--', linewidth=0.5) + + # Save plot + output_filename = f"ramachandran_{os.path.basename(traj_path).replace('.xtc', '.png')}" + output_path = os.path.join(output_dir, output_filename) + plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.close() + + +def create_histogram_plot(dihedrals, name1, name2, output_dir, name_string): + """ + Creates a 2D histogram with density for a pair of dihedrals. + """ + # Flatten all data for the pair of dihedrals + all_x_data = np.concatenate(dihedrals[name1]) + all_y_data = np.concatenate(dihedrals[name2]) + + plt.figure(figsize=(10, 10)) + + # Create 2D histogram with density + plt.hist2d(all_x_data, all_y_data, range=((-np.pi, np.pi), (-np.pi, np.pi)),bins=100, cmap='viridis', alpha=0.8, norm=colors.LogNorm()) + plt.colorbar(label='Density') + + plt.title(f'Histogram (Density): {name1} vs {name2}') + plt.xlabel(f'{name1} (radians)') + plt.ylabel(f'{name2} (radians)') + plt.xlim(-np.pi, np.pi) + plt.ylim(-np.pi, np.pi) + plt.grid(True, linestyle='--', alpha=0.6) + plt.axhline(0, color='k', linestyle='--', linewidth=0.5) + plt.axvline(0, color='k', linestyle='--', linewidth=0.5) + + output_filename = f"histogram_density_{name1}_vs_{name2}_{name_string}.png" + output_path = os.path.join(output_dir, output_filename) + plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.close() + + +def main(max_traj_per_grid=10): + """ + Loads trajectories, computes dihedral angles, and creates pairwise scatter plots and histograms. + Only keeps up to max_traj_per_grid trajectories per grid code. + """ + data_dir = "/data2/sules/fake_enhanced_data/ALA_ALA" + pdb_path = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + output_dir = f"/data2/sules/ramachandran_plots_ala_ala_fake_enhanced_data_max{max_traj_per_grid}" + + if not os.path.exists(output_dir): + os.makedirs(output_dir) + + try: + topology = md.load_pdb(pdb_path) + except Exception as e: + print(f"Could not load topology file {pdb_path}. Error: {e}") + return + + # Get all trajectory files + all_traj_files = sorted(glob.glob(os.path.join(data_dir, "traj_*.xtc"))) + print(f"Found {len(all_traj_files)} total trajectory files") + + if not all_traj_files: + print(f"No trajectory files found in {data_dir}") + return + + # Select trajectories with max per grid code + traj_files = select_trajectories_with_max_per_grid(all_traj_files, max_traj_per_grid) + print(f"Selected {len(traj_files)} trajectory files after filtering") + + all_phi_angles = [] + all_psi_angles = [] + num_phi, num_psi = None, None + + for traj_file in tqdm(traj_files, desc="Loading trajectories"): + try: + traj = md.load(traj_file, top=topology) + _, phi_angles = md.compute_phi(traj) + _, psi_angles = md.compute_psi(traj) + + if num_phi is None: + num_phi = phi_angles.shape[1] + num_psi = psi_angles.shape[1] + + all_phi_angles.append(phi_angles[:100,:]) + all_psi_angles.append(psi_angles[:100,:]) + except Exception as e: + print(f"Could not load or process trajectory {traj_file}. Error: {e}") + continue + + if not all_phi_angles or not all_psi_angles: + print("No valid trajectories were processed.") + return + + # Dynamically create dihedral dictionary + dihedrals = {} + for i in range(num_phi): + dihedrals[f'phi_{i+1}'] = [angles[:, i] for angles in all_phi_angles] + for i in range(num_psi): + dihedrals[f'psi_{i+1}'] = [angles[:, i] for angles in all_psi_angles] + + dihedral_names = list(dihedrals.keys()) + + # Create line plots (existing functionality) + create_line_plots = False + if create_line_plots: + print("Creating line plots...") + for name1, name2 in itertools.combinations(dihedral_names, 2): + plt.figure(figsize=(10, 10)) + + for i in tqdm(range(len(traj_files)), desc=f"Plotting {name1} vs {name2}"): + x_angles = dihedrals[name1][i] + y_angles = dihedrals[name2][i] + plt.plot(x_angles, y_angles, linestyle='-', alpha=0.5) + plt.scatter(x_angles[0], y_angles[0], c='white', marker='o', edgecolor='black', s=50, zorder=5) + + plt.title(f'Ramachandran Plot: {name1} vs {name2}') + plt.xlabel(f'{name1} (degrees)') + plt.ylabel(f'{name2} (degrees)') + plt.xlim(-180, 180) + plt.ylim(-180, 180) + plt.grid(True, linestyle='--', alpha=0.6) + plt.axhline(0, color='k', linestyle='--', linewidth=0.5) + plt.axvline(0, color='k', linestyle='--', linewidth=0.5) + + output_filename = f"ramachandran_{name1}_vs_{name2}.png" + output_path = os.path.join(output_dir, output_filename) + plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.close() + + create_histogram_plots = True + if create_histogram_plots: + # Create histogram plots (new functionality) + print("Creating histogram plots with density...") + for name1, name2 in tqdm(itertools.combinations(dihedral_names, 2), desc="Creating histograms"): + create_histogram_plot(dihedrals, name1, name2, output_dir, f"100_frames_max{max_traj_per_grid}") + + print(f"\nDone. Histogram plots are saved in {output_dir}") + print(f"Used max {max_traj_per_grid} trajectories per grid code") + + +if __name__ == "__main__": + # Set max trajectories per grid code to 10 + max_traj = 10 + main(max_traj_per_grid=max_traj) \ No newline at end of file diff --git a/scratch/visualize_noise_denoise.py b/scratch/visualize_noise_denoise.py new file mode 100644 index 0000000..d4374d7 --- /dev/null +++ b/scratch/visualize_noise_denoise.py @@ -0,0 +1,271 @@ +#!/usr/bin/env python3 +""" +Script to analyze validation trajectories using a trained model. +Generates Ramachandran plots for clean, noisy, and denoised samples. +""" + +import os +import glob +import tempfile +from pathlib import Path +from typing import List, Dict +import numpy as np +from tqdm import tqdm +import hydra +from omegaconf import OmegaConf +import torch +import torch_geometric +import lightning.pytorch as pl +import matplotlib.pyplot as plt +import mdtraj as md +import wandb +import pdb +pdb.set_trace() +from jamun.data import parse_datasets_from_directory, MDtrajDataset +from jamun.metrics._visualize_denoise import VisualizeDenoiseMetrics, plot_ramachandran_grid +from jamun import utils +from jamun.utils.checkpoint import find_checkpoint +from jamun.model.denoiser_conditional import Denoiser +# from jamun.model.denoiser import Denoiser as Denoiser_unconditional + + +def load_model_from_wandb(wandb_run_path: str, checkpoint_type: str = "last", checkpoint_path: str = None): + """Load model from wandb run.""" + print(f"Loading model from {wandb_run_path}...") + + # Use jamun utilities to find the checkpoint + checkpoint_path_wandb = find_checkpoint( + wandb_train_run_path=wandb_run_path, + checkpoint_type=checkpoint_type + ) + + if checkpoint_path is None: + checkpoint_path = checkpoint_path_wandb + + print(f"Loading model from checkpoint: {checkpoint_path}") + + # Load the model + model = Denoiser.load_from_checkpoint(checkpoint_path) + model.eval() + + print(f"āœ“ Model loaded successfully") + return model + + +def create_dataset_from_trajectory(traj_file: str, pdb_file: str, total_lag_time: int = 2): + """Create a dataset from a single trajectory file.""" + # Create temporary directory structure expected by parse_datasets_from_directory + temp_dir = tempfile.mkdtemp() + + # Copy trajectory file to temp directory + traj_name = Path(traj_file).stem + temp_traj_path = os.path.join(temp_dir, f"{traj_name}.xtc") + temp_pdb_path = os.path.join(temp_dir, f"{traj_name}.pdb") + + # Create symlinks + os.symlink(traj_file, temp_traj_path) + os.symlink(pdb_file, temp_pdb_path) + + # Parse dataset + datasets = parse_datasets_from_directory( + root=temp_dir, + traj_pattern="^(.*).xtc", + pdb_pattern="^(.*).pdb", + as_iterable=False, + subsample=1, + total_lag_time=total_lag_time, + lag_subsample_rate=1, + max_datasets=1, + label_override=traj_name + ) + + return datasets[0] if datasets else None + + +def process_trajectory(model, dataset: MDtrajDataset, sigma: float = 0.04): + """Process a single trajectory through the model.""" + # Create dataloader + dataloader = torch.utils.data.DataLoader( + dataset, + batch_size=32, + shuffle=False, + collate_fn=torch_geometric.data.Batch.from_data_list + ) + + # Store all samples + all_clean = [] + all_noisy = [] + all_denoised = [] + + model.eval() + with torch.no_grad(): + for batch in dataloader: + batch = batch.to(model.device) + + # # Ensure all batch attributes are the correct dtype + # if hasattr(batch, 'pos'): + # batch.pos = batch.pos.float() + # if hasattr(batch, 'batch'): + # batch.batch = batch.batch.long() + # if hasattr(batch, 'edge_index'): + # batch.edge_index = batch.edge_index.long() # Keep as long for indexing! + # if hasattr(batch, 'edge_attr') and batch.edge_attr is not None: + # batch.edge_attr = batch.edge_attr.float() + + # Convert sigma to tensor with correct dtype and device + sigma_tensor = torch.tensor(sigma, dtype=torch.float32, device=model.device) + + # Run noise and denoise + _, xhat, y = model.noise_and_denoise( + batch, sigma_tensor, align_noisy_input=model.align_noisy_input_during_evaluation + ) + + # Convert to data lists + clean_samples = torch_geometric.data.Batch.to_data_list(batch) + noisy_samples = torch_geometric.data.Batch.to_data_list(y) + denoised_samples = torch_geometric.data.Batch.to_data_list(xhat) + + all_clean.extend(clean_samples) + all_noisy.extend(noisy_samples) + all_denoised.extend(denoised_samples) + + return all_clean, all_noisy, all_denoised + + +def samples_to_trajectory(samples: List, dataset: MDtrajDataset): + """Convert list of samples to MDTraj trajectory.""" + coordinates = [] + for sample in samples: + coords = sample.pos.cpu().numpy() + coordinates.append(coords) + + coords_array = np.array(coordinates) # Shape: (n_frames, n_atoms, 3) + + # Create trajectory + traj = md.Trajectory(coords_array, dataset.topology) + return traj + + +def create_ramachandran_plot(clean_traj, noisy_traj, denoised_traj, title: str, save_path: str): + """Create and save Ramachandran plot for three trajectories.""" + trajs = { + "x": clean_traj, + "y": noisy_traj, + "xhat": denoised_traj + } + + try: + fig, axes = plot_ramachandran_grid(trajs, title) + fig.savefig(save_path, dpi=300, bbox_inches='tight') + plt.close(fig) + print(f"āœ“ Saved Ramachandran plot: {save_path}") + except Exception as e: + print(f"āœ— Error creating Ramachandran plot for {title}: {e}") + + +def main(): + # Configuration + wandb_run_path = "sule-shashank/jamun/4p0ejn0z" + val_dir = "/data2/sules/ALA_ALA_enhanced_full_grid/val" + pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" + checkpoint_type = 'epoch=49-step=52900-v1.ckpt' + # checkpoint_path = "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/epoch=9-step=10051.ckpt" + total_lag_time = 5 + sigma = 0.04 + output_dir = "val_ramachandrans" + + # Create output directory + os.makedirs(output_dir, exist_ok=True) + + # Load model + checkpoint_path = find_checkpoint(wandb_run_path, checkpoint_type=checkpoint_type) + model = Denoiser.load_from_checkpoint(checkpoint_path) + model.eval() + model.to('cuda:0') + print(f"Model loaded and moved to cuda:0") + # config_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/wandb/run-20250805_042516-yqn9mm7x/files/config.yaml" + # cfg = OmegaConf.load(config_path) + # checkpoint_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/checkpoints/last.ckpt" + # model = hydra.utils.instantiate(cfg.cfg.value.model) + # checkpoint = torch.load(checkpoint_path, weights_only=False) + # model.load_state_dict(checkpoint['state_dict']) + # model.eval() + # model.to('cuda:0') + # print(f"Model loaded and moved to cuda:0") + print(f"Model device: {next(model.parameters()).device}") + # Get all trajectory files + traj_files = glob.glob(os.path.join(val_dir, "*.xtc")) + traj_files.sort() + + print(f"Found {len(traj_files)} trajectory files") + # do one trial run + dataset = create_dataset_from_trajectory(traj_files[0], pdb_file, total_lag_time) + dataloader = torch.utils.data.DataLoader( + dataset, + batch_size=32, + shuffle=False, + collate_fn=torch_geometric.data.Batch.from_data_list + ) + _, batch = next(enumerate(dataloader)) + + batch = batch.to(model.device) + sigma_tensor = torch.tensor(sigma, dtype=torch.float32, device=model.device) + _, xhat, y = model.noise_and_denoise( + batch, sigma_tensor, align_noisy_input=model.align_noisy_input_during_evaluation + ) + # Store all samples for concatenated analysis + all_clean_samples = [] + all_noisy_samples = [] + all_denoised_samples = [] + breakpoint() + # Process each trajectory with progress bar + for traj_file in tqdm(traj_files, desc="Processing trajectories"): + traj_name = Path(traj_file).stem + + try: + # Create dataset + dataset = create_dataset_from_trajectory(traj_file, pdb_file, total_lag_time) + if dataset is None: + tqdm.write(f"Failed to create dataset for {traj_name}") + continue + + # Process trajectory + # breakpoint() + clean_samples, noisy_samples, denoised_samples = process_trajectory(model, dataset, sigma) + + # Store samples for concatenated analysis + all_clean_samples.extend(clean_samples) + all_noisy_samples.extend(noisy_samples) + all_denoised_samples.extend(denoised_samples) + + except Exception as e: + tqdm.write(f"Error processing {traj_name}: {e}") + continue + + # Create concatenated analysis + if all_clean_samples: + print("\nCreating concatenated Ramachandran plot...") + + # Use the last dataset for topology (they should all be the same) + concat_clean_traj = samples_to_trajectory(all_clean_samples, dataset) + concat_noisy_traj = samples_to_trajectory(all_noisy_samples, dataset) + concat_denoised_traj = samples_to_trajectory(all_denoised_samples, dataset) + + # Create concatenated plot + concat_plot_path = os.path.join(output_dir, "concatenated_ramachandran.png") + create_ramachandran_plot( + concat_clean_traj, + concat_noisy_traj, + concat_denoised_traj, + "Concatenated Trajectories", + concat_plot_path + ) + + print(f"\nAnalysis complete! Concatenated Ramachandran plot saved in {output_dir}/") + print(f"Processed {len(all_clean_samples)} total samples from {len(traj_files)} trajectories") + else: + print("No samples were processed successfully.") + + +if __name__ == "__main__": + main() \ No newline at end of file From 148b76a53fff6d8453e655b0f715567879fd701d Mon Sep 17 00:00:00 2001 From: Shashank Sule Date: Fri, 22 Aug 2025 23:36:45 +0000 Subject: [PATCH 28/32] documentation commit --- docs/conditional-gen-branch-documentation.md | 566 ------------------- 1 file changed, 566 deletions(-) delete mode 100644 docs/conditional-gen-branch-documentation.md diff --git a/docs/conditional-gen-branch-documentation.md b/docs/conditional-gen-branch-documentation.md deleted file mode 100644 index 2e45370..0000000 --- a/docs/conditional-gen-branch-documentation.md +++ /dev/null @@ -1,566 +0,0 @@ -# Conditional Generation Branch Documentation - -This document provides a comprehensive walkthrough of the conditional-gen branch, covering the spatiotemporal model architecture, conditional denoising, and memory-based sampling mechanisms. - -## Table of Contents - -1. [Spatiotemporal Model Architecture](#spatiotemporal-model-architecture) -2. [Conditional Denoiser Architecture](#conditional-denoiser-architecture) -3. [Model.g Parameterization and E3Conv Conditional](#modelg-parameterization-and-e3conv-conditional) -4. [Memory-Based Sampling: BAOAB/ABOBA Subroutines](#memory-based-sampling-baoababoba-subroutines) -5. [Sampling Memory Wrapper](#sampling-memory-wrapper) - ---- - -## Spatiotemporal Model Architecture - -### Overview - -The spatiotemporal model (`E3SpatioTemporal`) implements a complete workflow that processes molecular structures with temporal dependencies by converting between spatial and temporal graph representations. - -### Core Components - -#### 1. Spatial to Temporal Graph Conversion - -The model begins by converting spatial graphs to temporal graphs using the `spatial_to_temporal_graphs()` function: - -```python -def spatial_to_temporal_graphs(batch, graph_type="fan"): - """ - Convert a batch of spatial graphs to temporal graphs with configurable connectivity. - - For each spatial node with position + hidden states, create a temporal graph where: - - Node 0: current position - - Nodes 1-T: hidden state positions - - Connectivity depends on graph_type parameter - """ -``` - -**Graph Type Options:** -- **"fan"**: Hub-spoke + sequential connections (0ā†’all, iā†’(i+1)) -- **"hub_n_spoke"**: Only hub-spoke connections (0ā†’all, no sequential) -- **"complete"**: Complete graph without self-loops (all-to-all excluding self) -- **"complete_no_self"**: Complete graph with self-loops (all-to-all including self) - -#### 2. Temporal Position Calculation - -Temporal positions are normalized to create consistent temporal embeddings: - -```python -def calculate_temporal_positions(temporal_length, mode="linear", device=None): - """Calculate normalized temporal positions [0, 1/T, 2/T, ..., (T-1)/T]""" - positions = torch.linspace(0, 1, temporal_length + 1, device=device)[:-1] - return positions -``` - -### Complete Spatiotemporal Workflow - -The `E3SpatioTemporal.forward()` method implements the following pipeline: - -#### Step 1: Spatial Graph ā†’ Temporal Graph Conversion -```python -temporal_batch = spatial_to_temporal_graphs(batch, graph_type=self.graph_type) -``` - -#### Step 2: Spatial Processing of All Time Steps -For each position (current + hidden states), the spatial module processes: -```python -# Current positions -node_attr_current = self.spatial_module( - pos=batch.pos, - topology=topology, - batch=batch.batch, - num_graphs=batch.num_graphs, - c_noise=c_noise, - effective_radial_cutoff=self.radial_cutoff -) - -# Hidden state positions -for hidden_pos in batch.hidden_state: - node_attr_hidden = self.spatial_module( - pos=hidden_pos, - topology=topology, - batch=batch.batch, - num_graphs=batch.num_graphs, - c_noise=c_noise, - effective_radial_cutoff=self.radial_cutoff - ) -``` - -#### Step 3: Spatial-Temporal Feature Assembly -```python -node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) # [N, T, features] -``` - -#### Step 4: Spatialā†’Temporal Pooling -```python -temporal_node_attr = self.spatial_to_temporal_pooler(node_attr_spatial_temporal, temporal_batch) -``` - -#### Step 5: Temporal Processing -The temporal module processes the temporal graph using an E3 Transformer: -```python -temporal_output = self.temporal_module( - temporal_node_attr, - temporal_batch, - self.radial_cutoff, - self.temporal_cutoff -) -``` - -#### Step 6: Temporalā†’Spatial Pooling -```python -spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) -``` - -#### Step 7: Return to Spatial Representation -The final spatial features are returned for use by the conditional architecture. - ---- - -## Conditional Denoiser Architecture - -### Denoiser_Conditional Overview - -The `Denoiser` class in `denoiser_conditional.py` extends the standard JAMUN denoiser with conditional generation capabilities through the integration of conditioner modules. - -### Key Components - -#### 1. Conditioner Integration - -The denoiser accepts a `conditioner` parameter that processes the input batch to generate conditioning structures: - -```python -def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: - if self.conditioning_module is None: - return self.conditioner_default(y) # Returns [y.pos] - elif callable(self.conditioning_module): - return self.conditioning_module(y) - else: - raise ValueError("Conditioner must be a callable or None") -``` - -#### 2. Hidden State Management - -The denoiser properly handles hidden states throughout the pipeline: - -**Adding Noise to Hidden States:** -```python -def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]): - # Add noise to current positions - y.pos = x.pos + sigma * noise - # Add noise to hidden states - for i in range(len(y.hidden_state)): - y.hidden_state[i] = x.hidden_state[i] + sigma * hidden_noise[i] -``` - -**Hidden State Alignment:** -```python -def _align_A_to_B_batched_with_hidden_states(self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch): - # Align positions - A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) - # Align hidden states - if hasattr(A, "hidden_state") and A.hidden_state is not None: - A_aligned.hidden_state = [] - for i in range(len(A.hidden_state)): - A_aligned.hidden_state.append(kabsch_algorithm( - A.hidden_state[i], B.pos, A.batch, A.num_graphs - )) -``` - -### The `xhat_normalized` Method: Core Conditional Processing - -The `xhat_normalized` method is the heart of conditional generation: - -#### Step 1: Normalization Factor Computation -```python -c_in, c_skip, c_out, c_noise = self.normalization_factors(sigma, D) -radial_cutoff = self.effective_radial_cutoff(sigma) / c_in -``` - -#### Step 2: Input Scaling -```python -y_scaled = y.clone() -y_scaled.pos = y.pos * c_in -# Scale hidden states -if hasattr(y, "hidden_state") and y.hidden_state is not None: - y_scaled.hidden_state = [] - for positions in y.hidden_state: - y_scaled.hidden_state.append(positions * c_in) -``` - -#### Step 3: Conditioning Structure Generation -```python -with torch.cuda.nvtx.range("conditioning"): - conditioned_structures = self.conditioner(y_scaled) -``` - -The conditioner returns a list of tensors that will be concatenated for model input. - -#### Step 4: Model Prediction via model.g -```python -with torch.cuda.nvtx.range("g"): - g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), - topology=y_scaled, - c_noise=c_noise, - effective_radial_cutoff=radial_cutoff) -``` - -**Key Point**: `model.g` receives the concatenated conditioned structures as input positions. - -#### Step 5: Output Construction and Hidden State Update -```python -xhat.pos = c_skip * y.pos + c_out * g_pred -if hasattr(y, "hidden_state") and y.hidden_state is not None: - xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] # Hidden state shifts forward -``` - -**Hidden State Evolution**: The hidden state list is updated by: -1. Moving current position (`y.pos`) to the front -2. Keeping all previous hidden states except the oldest one -3. This creates a sliding window of historical positions - ---- - -## Model.g Parameterization and E3Conv Conditional - -### Architecture Variants - -The conditional-gen branch introduces several E3Conv variants designed for different conditioning scenarios: - -#### 1. E3ConvConditional - -**Purpose**: Basic conditional model that handles multiple structure inputs -**Key Parameters**: -- `N_structures`: Number of input structures to concatenate -- `irreps_sh`: Extended to `N_structures * self.irreps_sh` for spherical harmonics applied in parallel to N_structures-many 3D vectors. - -**Forward Pass**: -```python -def forward(self, pos: Tensor, topology, c_noise, effective_radial_cutoff): - # pos should be [batch_size*N, 3T] where T is number of time-steps - positions = torch.split(pos, 3, dim=-1) - edge_sh = [] - for block in positions: - edge_vec = block[src] - block[dst] - edge_sh.append(self.sh(edge_vec)) - edge_sh = torch.cat(edge_sh, dim=-1) # Concatenate spherical harmonics -``` - -#### 2. E3ConvConditionalSpatioTemporal - -**Purpose**: Specialized for spatiotemporal conditioning where input combines physical positions with spatial features - -**Key Design**: -- Expects input: `[y.pos, spatial_features]` concatenated along feature dimension -- `N_structures = 1` (processes combined input as single structure) -- `input_attr_irreps`: Defines irreps for spatial features component - -**Forward Pass Logic**: -```python -def forward(self, pos: Tensor, topology, c_noise, effective_radial_cutoff): - # Split positions: first 3 coords are physical, rest are spatial features - pos_physical = pos[:, :3] # [N, 3] - physical coordinates - pos_features = pos[:, 3:] # [N, spatial_features_dim] - spatial features - - # Compute edge spherical harmonics ONLY for physical positions - edge_vec_physical = pos_physical[src] - pos_physical[dst] - edge_sh = self.sh(edge_vec_physical) - - # Combine node_attr with spatial features as input attributes - # ... -``` - -### Configuration Examples - -**Standard Conditional Model**: -```yaml -arch: - _target_: jamun.model.arch.E3ConvConditional - N_structures: 2 # [current_pos, conditioned_structure] - irreps_sh: "1x0e + 1x1e" -``` - -**Spatiotemporal Conditional Model**: -```yaml -arch: - _target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalSpatioTemporal - N_structures: 1 # Combined [y.pos, spatial_features] - input_attr_irreps: "120x0e + 32x1e" # Match spatiotemporal output -``` - ---- - -## Memory-Based Sampling: BAOAB/ABOBA Subroutines - -### Overview - -The conditional-gen branch introduces memory-enhanced versions of BAOAB and ABOBA splitting schemes that maintain and update a history of molecular states during sampling. - -### Core Memory Concepts - -#### 1. Historical State Management -- `y_hist`: List of previous molecular configurations -- States are maintained in chronological order: `[newest, ..., oldest]` -- History updates occur at configurable intervals via `history_update_frequency` - -#### 2. Conditional Density Equilibration -The key innovation is equilibration to conditional densities `p(y_t | y_hist)` rather than marginal densities. - -### BAOAB_Memory Algorithm - -#### Algorithm Structure -```python -def baoab_memory(y, y_hist, score_fn, steps, history_update_frequency=1, ...): - """BAOAB splitting scheme that updates a state history.""" -``` - -#### Key Parameters -- `y`: Current molecular state -- `y_hist`: History of previous states -- `score_fn`: Score function that accepts both `y` and `y_hist` -- `history_update_frequency`: How often to update the history (inner loop iterations) - -#### Main Algorithm Loop - -**Outer Loop** (Main sampling steps): -```python -for i in steps_iter: - # Inner equilibration loop - for j in range(1, history_update_frequency): - # BAOAB step for conditional density p(y_t | y_hist) - y_current = y.clone().detach() - v = v + u * (delta / 2) * psi # B step - y = y + (delta / 2) * v # A step - R = torch.randn_like(y) - vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R # O step - y = y + (delta / 2) * vhat # A step - psi, orig_score = score_fn_processed(y, y_hist=y_hist) - v = vhat + (delta / 2) * psi # B step - - # Update history - y_hist.pop(-1) # Remove oldest state - y_hist.insert(0, y_current) # Add current state as newest -``` - -### Inner Loop Mechanics: The Role of Equilibration - -#### Purpose of the Inner Loop - -Rather than taking a single MCMC step, the inner loop allows the system to equilibrate to the conditional distribution `p(y_t | y_hist)` given the fixed history. - - -#### Mathematical Justification -``` -Standard MCMC: y_{t+1} ~ p(y | y_t) -Memory MCMC: y_{t+1} ~ p(y | y_hist) after equilibration to p(y | y_hist) -``` - -### ABOBA_Memory Algorithm - -Similar structure to BAOAB_memory but with ABOBA splitting: - -```python -def aboba_memory(y, y_hist, score_fn, steps, history_update_frequency=1, ...): - """ABOBA splitting scheme that updates a state history.""" - for i in steps_iter: - for j in range(1, history_update_frequency): - # ABOBA inner loop for equilibration - y_current = y.clone().detach() - y = y + (delta / 2) * v # A step - psi, orig_score = score_fn_processed(y, y_hist=y_hist) - v = v + u * (delta / 2) * psi # B step - R = torch.randn_like(y) - vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R # O step - v = vhat + (delta / 2) * psi # B step - y = y + (delta / 2) * v # A step -``` - -### Cleanup Option - -Both algorithms support a `cleanup` option for denoising: - -```python -if cleanup is not None and cleanup and sigma is not None: - y_current = y.clone().detach() - _, orig_score = score_fn_processed(y_current, y_hist=y_hist) - y_denoised_and_noised = y_current + (sigma**2)*orig_score + sigma*torch.randn_like(y_current) - y_hist.pop(-1) - y_hist.insert(0, y_denoised_and_noised) - y = y_denoised_and_noised -``` - -This performs a denoising step followed by re-noising before adding to history. - -### Configuration - -```yaml -batch_sampler: - _target_: jamun.sampling.mcmc.BAOAB_memory - delta: ${delta} - friction: ${friction} - steps: ${num_sampling_steps_per_batch} - history_update_frequency: 10 # Inner loop iterations - save_trajectory: true - cpu_offload: true - verbose: true -``` - ---- - -## Sampling Memory Wrapper - -### ModelSamplingWrapperMemory - -The `ModelSamplingWrapperMemory` class extends the standard sampling wrapper to handle models that depend on historical states. - -#### Key Features - -#### 1. History-Aware Initialization -```python -def __init__(self, model, init_graphs, sigma, recenter_on_init=True): - # Apply mean centering to positions - self.init_graphs = mean_center(self.init_graphs) - - # Mean center hidden states if they exist - if hasattr(self.init_graphs, 'hidden_state') and self.init_graphs.hidden_state: - for i in range(len(self.init_graphs.hidden_state)): - mean = scatter(self.init_graphs.hidden_state[i], self.init_graphs.batch, dim=0, reduce="mean") - self.init_graphs.hidden_state[i] = self.init_graphs.hidden_state[i] - mean[self.init_graphs.batch] -``` - -#### 2. History Sampling -```python -def sample_initial_noisy_history(self) -> list: - """Sample initial noisy history from hidden states.""" - noisy_history = [] - for hidden_state in self.init_graphs.hidden_state: - noisy_history.append(hidden_state + torch.randn_like(hidden_state) * self.sigma) - return noisy_history -``` - -#### 3. Memory-Aware Score and Prediction Functions -```python -def score(self, y, y_hist, sigma): - """Score function that includes history.""" - graph = self.positions_to_graph(y, y_hist).to(self.device) - return self._model.score(graph, sigma) - -def xhat(self, y, y_hist, sigma): - """Prediction function that includes history.""" - graph = self.positions_to_graph(y, y_hist).to(self.device) - xhat_graph = self._model.xhat(graph, sigma) - return xhat_graph.pos -``` - -#### 4. Graph Construction with History -```python -def positions_to_graph(self, positions: torch.Tensor, y_hist: list) -> torch_geometric.data.Data: - """Wraps positions to a graph and attaches the historical states.""" - input_graph = self.init_graphs.clone() - input_graph.pos = positions - input_graph.hidden_state = y_hist # Attach history as hidden_state - return input_graph.to(positions.device) -``` - -#### 5. History-Aware Sample Unbatching -The wrapper handles complex unbatching of trajectory data that includes historical information: - -```python -def unbatch_samples(self, samples: dict[str, torch.Tensor]): - """Unbatch samples including history trajectories.""" - for key, value in samples.items(): - if key == "y_hist" or key == "y_hist_traj": - # Special handling for history data - if key == "y_hist": - value = [value] - value = torch.stack([torch.stack(traj, dim=1) for traj in value], dim=1) - # ... complex unbatching logic for history -``` - -### Integration with SamplerMemory - -The `SamplerMemory` class uses `ModelSamplingWrapperMemory`: - -```python -class SamplerMemory(Sampler): - def sample(self, model, batch_sampler, num_batches, init_graphs, continue_chain=False): - model_wrapped = utils.ModelSamplingWrapperMemory( - model=model, - init_graphs=init_graphs, - sigma=batch_sampler.sigma, - ) - - y_init = model_wrapped.sample_initial_noisy_positions() - y_hist_init = model_wrapped.sample_initial_noisy_history() - - # Memory-aware sampling - out = batch_sampler.sample(model=model_wrapped, y_init=y_init, - v_init=v_init, y_hist_init=y_hist_init) -``` - -### Chain Continuation - -The memory wrapper supports continuing chains across batches: - -```python -if continue_chain: - y_init = out["y"] - v_init = out["v"] - y_hist_init = out["y_hist"] # Continue with updated history -else: - y_init = model_wrapped.sample_initial_noisy_positions() - y_hist_init = model_wrapped.sample_initial_noisy_history() - v_init = "gaussian" -``` - ---- - -## Usage Examples - -### Training a Spatiotemporal Conditional Model - -```yaml -model: - _target_: jamun.model.denoiser_conditional.Denoiser - arch: - _target_: jamun.model.arch.e3conv_conditional.E3ConvConditionalSpatioTemporal - N_structures: 1 - input_attr_irreps: "120x0e + 32x1e" - conditioner: - _target_: jamun.model.conditioners.SpatioTemporalConditioner - spatiotemporal_model: - _target_: jamun.model.arch.spatiotemporal.E3SpatioTemporal - spatial_module: # ... spatial E3Conv config - temporal_module: # ... temporal E3Transformer config -``` - -### Memory-Based Sampling Configuration - -```yaml -sampler: - _target_: jamun.sampling.SamplerMemory - devices: 1 - -batch_sampler: - _target_: jamun.sampling.mcmc.BAOAB_memory - delta: 0.04 - friction: 1.0 - steps: 1000 - history_update_frequency: 10 - save_trajectory: true -``` - -### Data Loading for Memory Models - -```yaml -init_datasets: - _target_: jamun.data.parse_repeated_position_datasets_from_directory - root: "/path/to/data" - total_lag_time: 5 # Number of historical states - lag_subsample_rate: 1 # Temporal subsampling - subsample: 1 # Spatial subsampling -``` - -This comprehensive documentation covers the key innovations in the conditional-gen branch, providing both conceptual understanding and practical implementation details for spatiotemporal modeling and memory-based sampling in molecular dynamics. - From e166db1278203679bf4c82385f37d74f867400f7 Mon Sep 17 00:00:00 2001 From: Joseph Kleinhenz Date: Thu, 5 Feb 2026 18:42:58 +0000 Subject: [PATCH 29/32] update deps --- pyproject.toml | 10 +- uv.lock | 3542 ++++++++++++++++++++++-------------------------- 2 files changed, 1630 insertions(+), 1922 deletions(-) diff --git a/pyproject.toml b/pyproject.toml index 2b99b1b..bb4b014 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -8,7 +8,7 @@ requires-python = ">=3.10" dependencies = [ "ase>=3.23.0", "e3nn>=0.5.6", - "e3tools>=0.1.1", + "e3tools>=0.1.2", "einops>=0.8.0", "hydra-core>=1.3.2", "lightning>=2.4.0", @@ -38,12 +38,20 @@ dependencies = [ "universal-pathlib>=0.2.6", "wandb>=0.19.1", "orb_models>=0.5.4", + "optree>=0.17.0", ] [project.scripts] jamun_train = "jamun.cmdline.train:main" jamun_sample = "jamun.cmdline.sample:main" +[project.optional-dependencies] +analysis = [ + "polars>=1.32.0", + "pyarrow>=21.0.0", + "seaborn>=0.13.2", +] + [build-system] requires = ["hatchling", "hatch-vcs"] build-backend = "hatchling.build" diff --git a/uv.lock b/uv.lock index 0cb5bfc..e6c5b66 100644 --- a/uv.lock +++ b/uv.lock @@ -1,5 +1,5 @@ version = 1 -revision = 2 +revision = 3 requires-python = ">=3.10" resolution-markers = [ "python_full_version >= '3.13' and sys_platform == 'linux'", @@ -14,7 +14,7 @@ resolution-markers = [ [[package]] name = "aiobotocore" -version = "2.23.2" +version = "2.21.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "aiohttp" }, @@ -25,9 +25,9 @@ dependencies = [ { name = "python-dateutil" }, { name = "wrapt" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/1d/ae/523d48504902a6f17f6ec94311899f217f1bf64b9ca394c89c690c37434c/aiobotocore-2.23.2.tar.gz", hash = "sha256:9c2cbd6e813bb6c60b7f20fc11897976a583c57b0093a87bebfe80a9b08746b2", size = 115881, upload-time = "2025-07-24T17:48:15.957Z" } +sdist = { url = "https://files.pythonhosted.org/packages/d2/dc/f5f872fb01ce37c09525cedc7ecfad7002ffe2a8a23f77d7d2c234399b51/aiobotocore-2.21.1.tar.gz", hash = "sha256:010357f43004413e92a9d066bb0db1f241aeb29ffed306e9197061ffc94e6577", size = 108900, upload-time = "2025-03-04T18:30:58.945Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c9/2e/55780065672a69ab3d16062368d358ace7eac196b246e6b15b07301f8fbf/aiobotocore-2.23.2-py3-none-any.whl", hash = "sha256:5ca24feb49be73bd6cd92e82e95aefb0647c07bb85ca57000a0361b9554503d8", size = 84301, upload-time = "2025-07-24T17:48:14.494Z" }, + { url = "https://files.pythonhosted.org/packages/95/67/026598918f92145156f2feb7957f57daefda20375cc2ac1a0692a9b8010b/aiobotocore-2.21.1-py3-none-any.whl", hash = "sha256:bd7c49a6d6f8a3d9444b0a94417c8da13813b5c7eec1c4f0ec2db7e8ce8f23e7", size = 78313, upload-time = "2025-03-04T18:30:56.493Z" }, ] [package.optional-dependencies] @@ -46,7 +46,7 @@ wheels = [ [[package]] name = "aiohttp" -version = "3.12.15" +version = "3.11.18" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "aiohappyeyeballs" }, @@ -58,76 +58,72 @@ dependencies = [ { name = "propcache" }, { name = "yarl" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/9b/e7/d92a237d8802ca88483906c388f7c201bbe96cd80a165ffd0ac2f6a8d59f/aiohttp-3.12.15.tar.gz", hash = "sha256:4fc61385e9c98d72fcdf47e6dd81833f47b2f77c114c29cd64a361be57a763a2", size = 7823716, upload-time = "2025-07-29T05:52:32.215Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/47/dc/ef9394bde9080128ad401ac7ede185267ed637df03b51f05d14d1c99ad67/aiohttp-3.12.15-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:b6fc902bff74d9b1879ad55f5404153e2b33a82e72a95c89cec5eb6cc9e92fbc", size = 703921, upload-time = "2025-07-29T05:49:43.584Z" }, - { url = "https://files.pythonhosted.org/packages/8f/42/63fccfc3a7ed97eb6e1a71722396f409c46b60a0552d8a56d7aad74e0df5/aiohttp-3.12.15-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:098e92835b8119b54c693f2f88a1dec690e20798ca5f5fe5f0520245253ee0af", size = 480288, upload-time = "2025-07-29T05:49:47.851Z" }, - { url = "https://files.pythonhosted.org/packages/9c/a2/7b8a020549f66ea2a68129db6960a762d2393248f1994499f8ba9728bbed/aiohttp-3.12.15-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:40b3fee496a47c3b4a39a731954c06f0bd9bd3e8258c059a4beb76ac23f8e421", size = 468063, upload-time = "2025-07-29T05:49:49.789Z" }, - { url = "https://files.pythonhosted.org/packages/8f/f5/d11e088da9176e2ad8220338ae0000ed5429a15f3c9dfd983f39105399cd/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2ce13fcfb0bb2f259fb42106cdc63fa5515fb85b7e87177267d89a771a660b79", size = 1650122, upload-time = "2025-07-29T05:49:51.874Z" }, - { url = "https://files.pythonhosted.org/packages/b0/6b/b60ce2757e2faed3d70ed45dafee48cee7bfb878785a9423f7e883f0639c/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:3beb14f053222b391bf9cf92ae82e0171067cc9c8f52453a0f1ec7c37df12a77", size = 1624176, upload-time = "2025-07-29T05:49:53.805Z" }, - { url = "https://files.pythonhosted.org/packages/dd/de/8c9fde2072a1b72c4fadecf4f7d4be7a85b1d9a4ab333d8245694057b4c6/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4c39e87afe48aa3e814cac5f535bc6199180a53e38d3f51c5e2530f5aa4ec58c", size = 1696583, upload-time = "2025-07-29T05:49:55.338Z" }, - { url = "https://files.pythonhosted.org/packages/0c/ad/07f863ca3d895a1ad958a54006c6dafb4f9310f8c2fdb5f961b8529029d3/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d5f1b4ce5bc528a6ee38dbf5f39bbf11dd127048726323b72b8e85769319ffc4", size = 1738896, upload-time = "2025-07-29T05:49:57.045Z" }, - { url = "https://files.pythonhosted.org/packages/20/43/2bd482ebe2b126533e8755a49b128ec4e58f1a3af56879a3abdb7b42c54f/aiohttp-3.12.15-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1004e67962efabbaf3f03b11b4c43b834081c9e3f9b32b16a7d97d4708a9abe6", size = 1643561, upload-time = "2025-07-29T05:49:58.762Z" }, - { url = "https://files.pythonhosted.org/packages/23/40/2fa9f514c4cf4cbae8d7911927f81a1901838baf5e09a8b2c299de1acfe5/aiohttp-3.12.15-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8faa08fcc2e411f7ab91d1541d9d597d3a90e9004180edb2072238c085eac8c2", size = 1583685, upload-time = "2025-07-29T05:50:00.375Z" }, - { url = "https://files.pythonhosted.org/packages/b8/c3/94dc7357bc421f4fb978ca72a201a6c604ee90148f1181790c129396ceeb/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:fe086edf38b2222328cdf89af0dde2439ee173b8ad7cb659b4e4c6f385b2be3d", size = 1627533, upload-time = "2025-07-29T05:50:02.306Z" }, - { url = "https://files.pythonhosted.org/packages/bf/3f/1f8911fe1844a07001e26593b5c255a685318943864b27b4e0267e840f95/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:79b26fe467219add81d5e47b4a4ba0f2394e8b7c7c3198ed36609f9ba161aecb", size = 1638319, upload-time = "2025-07-29T05:50:04.282Z" }, - { url = "https://files.pythonhosted.org/packages/4e/46/27bf57a99168c4e145ffee6b63d0458b9c66e58bb70687c23ad3d2f0bd17/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:b761bac1192ef24e16706d761aefcb581438b34b13a2f069a6d343ec8fb693a5", size = 1613776, upload-time = "2025-07-29T05:50:05.863Z" }, - { url = "https://files.pythonhosted.org/packages/0f/7e/1d2d9061a574584bb4ad3dbdba0da90a27fdc795bc227def3a46186a8bc1/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:e153e8adacfe2af562861b72f8bc47f8a5c08e010ac94eebbe33dc21d677cd5b", size = 1693359, upload-time = "2025-07-29T05:50:07.563Z" }, - { url = "https://files.pythonhosted.org/packages/08/98/bee429b52233c4a391980a5b3b196b060872a13eadd41c3a34be9b1469ed/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:fc49c4de44977aa8601a00edbf157e9a421f227aa7eb477d9e3df48343311065", size = 1716598, upload-time = "2025-07-29T05:50:09.33Z" }, - { url = "https://files.pythonhosted.org/packages/57/39/b0314c1ea774df3392751b686104a3938c63ece2b7ce0ba1ed7c0b4a934f/aiohttp-3.12.15-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:2776c7ec89c54a47029940177e75c8c07c29c66f73464784971d6a81904ce9d1", size = 1644940, upload-time = "2025-07-29T05:50:11.334Z" }, - { url = "https://files.pythonhosted.org/packages/1b/83/3dacb8d3f8f512c8ca43e3fa8a68b20583bd25636ffa4e56ee841ffd79ae/aiohttp-3.12.15-cp310-cp310-win32.whl", hash = "sha256:2c7d81a277fa78b2203ab626ced1487420e8c11a8e373707ab72d189fcdad20a", size = 429239, upload-time = "2025-07-29T05:50:12.803Z" }, - { url = "https://files.pythonhosted.org/packages/eb/f9/470b5daba04d558c9673ca2034f28d067f3202a40e17804425f0c331c89f/aiohttp-3.12.15-cp310-cp310-win_amd64.whl", hash = "sha256:83603f881e11f0f710f8e2327817c82e79431ec976448839f3cd05d7afe8f830", size = 452297, upload-time = "2025-07-29T05:50:14.266Z" }, - { url = "https://files.pythonhosted.org/packages/20/19/9e86722ec8e835959bd97ce8c1efa78cf361fa4531fca372551abcc9cdd6/aiohttp-3.12.15-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:d3ce17ce0220383a0f9ea07175eeaa6aa13ae5a41f30bc61d84df17f0e9b1117", size = 711246, upload-time = "2025-07-29T05:50:15.937Z" }, - { url = "https://files.pythonhosted.org/packages/71/f9/0a31fcb1a7d4629ac9d8f01f1cb9242e2f9943f47f5d03215af91c3c1a26/aiohttp-3.12.15-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:010cc9bbd06db80fe234d9003f67e97a10fe003bfbedb40da7d71c1008eda0fe", size = 483515, upload-time = "2025-07-29T05:50:17.442Z" }, - { url = "https://files.pythonhosted.org/packages/62/6c/94846f576f1d11df0c2e41d3001000527c0fdf63fce7e69b3927a731325d/aiohttp-3.12.15-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:3f9d7c55b41ed687b9d7165b17672340187f87a773c98236c987f08c858145a9", size = 471776, upload-time = "2025-07-29T05:50:19.568Z" }, - { url = "https://files.pythonhosted.org/packages/f8/6c/f766d0aaafcee0447fad0328da780d344489c042e25cd58fde566bf40aed/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bc4fbc61bb3548d3b482f9ac7ddd0f18c67e4225aaa4e8552b9f1ac7e6bda9e5", size = 1741977, upload-time = "2025-07-29T05:50:21.665Z" }, - { url = "https://files.pythonhosted.org/packages/17/e5/fb779a05ba6ff44d7bc1e9d24c644e876bfff5abe5454f7b854cace1b9cc/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:7fbc8a7c410bb3ad5d595bb7118147dfbb6449d862cc1125cf8867cb337e8728", size = 1690645, upload-time = "2025-07-29T05:50:23.333Z" }, - { url = "https://files.pythonhosted.org/packages/37/4e/a22e799c2035f5d6a4ad2cf8e7c1d1bd0923192871dd6e367dafb158b14c/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:74dad41b3458dbb0511e760fb355bb0b6689e0630de8a22b1b62a98777136e16", size = 1789437, upload-time = "2025-07-29T05:50:25.007Z" }, - { url = "https://files.pythonhosted.org/packages/28/e5/55a33b991f6433569babb56018b2fb8fb9146424f8b3a0c8ecca80556762/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3b6f0af863cf17e6222b1735a756d664159e58855da99cfe965134a3ff63b0b0", size = 1828482, upload-time = "2025-07-29T05:50:26.693Z" }, - { url = "https://files.pythonhosted.org/packages/c6/82/1ddf0ea4f2f3afe79dffed5e8a246737cff6cbe781887a6a170299e33204/aiohttp-3.12.15-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b5b7fe4972d48a4da367043b8e023fb70a04d1490aa7d68800e465d1b97e493b", size = 1730944, upload-time = "2025-07-29T05:50:28.382Z" }, - { url = "https://files.pythonhosted.org/packages/1b/96/784c785674117b4cb3877522a177ba1b5e4db9ce0fd519430b5de76eec90/aiohttp-3.12.15-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6443cca89553b7a5485331bc9bedb2342b08d073fa10b8c7d1c60579c4a7b9bd", size = 1668020, upload-time = "2025-07-29T05:50:30.032Z" }, - { url = "https://files.pythonhosted.org/packages/12/8a/8b75f203ea7e5c21c0920d84dd24a5c0e971fe1e9b9ebbf29ae7e8e39790/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:6c5f40ec615e5264f44b4282ee27628cea221fcad52f27405b80abb346d9f3f8", size = 1716292, upload-time = "2025-07-29T05:50:31.983Z" }, - { url = "https://files.pythonhosted.org/packages/47/0b/a1451543475bb6b86a5cfc27861e52b14085ae232896a2654ff1231c0992/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:2abbb216a1d3a2fe86dbd2edce20cdc5e9ad0be6378455b05ec7f77361b3ab50", size = 1711451, upload-time = "2025-07-29T05:50:33.989Z" }, - { url = "https://files.pythonhosted.org/packages/55/fd/793a23a197cc2f0d29188805cfc93aa613407f07e5f9da5cd1366afd9d7c/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:db71ce547012a5420a39c1b744d485cfb823564d01d5d20805977f5ea1345676", size = 1691634, upload-time = "2025-07-29T05:50:35.846Z" }, - { url = "https://files.pythonhosted.org/packages/ca/bf/23a335a6670b5f5dfc6d268328e55a22651b440fca341a64fccf1eada0c6/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:ced339d7c9b5030abad5854aa5413a77565e5b6e6248ff927d3e174baf3badf7", size = 1785238, upload-time = "2025-07-29T05:50:37.597Z" }, - { url = "https://files.pythonhosted.org/packages/57/4f/ed60a591839a9d85d40694aba5cef86dde9ee51ce6cca0bb30d6eb1581e7/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:7c7dd29c7b5bda137464dc9bfc738d7ceea46ff70309859ffde8c022e9b08ba7", size = 1805701, upload-time = "2025-07-29T05:50:39.591Z" }, - { url = "https://files.pythonhosted.org/packages/85/e0/444747a9455c5de188c0f4a0173ee701e2e325d4b2550e9af84abb20cdba/aiohttp-3.12.15-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:421da6fd326460517873274875c6c5a18ff225b40da2616083c5a34a7570b685", size = 1718758, upload-time = "2025-07-29T05:50:41.292Z" }, - { url = "https://files.pythonhosted.org/packages/36/ab/1006278d1ffd13a698e5dd4bfa01e5878f6bddefc296c8b62649753ff249/aiohttp-3.12.15-cp311-cp311-win32.whl", hash = "sha256:4420cf9d179ec8dfe4be10e7d0fe47d6d606485512ea2265b0d8c5113372771b", size = 428868, upload-time = "2025-07-29T05:50:43.063Z" }, - { url = "https://files.pythonhosted.org/packages/10/97/ad2b18700708452400278039272032170246a1bf8ec5d832772372c71f1a/aiohttp-3.12.15-cp311-cp311-win_amd64.whl", hash = "sha256:edd533a07da85baa4b423ee8839e3e91681c7bfa19b04260a469ee94b778bf6d", size = 453273, upload-time = "2025-07-29T05:50:44.613Z" }, - { url = "https://files.pythonhosted.org/packages/63/97/77cb2450d9b35f517d6cf506256bf4f5bda3f93a66b4ad64ba7fc917899c/aiohttp-3.12.15-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:802d3868f5776e28f7bf69d349c26fc0efadb81676d0afa88ed00d98a26340b7", size = 702333, upload-time = "2025-07-29T05:50:46.507Z" }, - { url = "https://files.pythonhosted.org/packages/83/6d/0544e6b08b748682c30b9f65640d006e51f90763b41d7c546693bc22900d/aiohttp-3.12.15-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:f2800614cd560287be05e33a679638e586a2d7401f4ddf99e304d98878c29444", size = 476948, upload-time = "2025-07-29T05:50:48.067Z" }, - { url = "https://files.pythonhosted.org/packages/3a/1d/c8c40e611e5094330284b1aea8a4b02ca0858f8458614fa35754cab42b9c/aiohttp-3.12.15-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:8466151554b593909d30a0a125d638b4e5f3836e5aecde85b66b80ded1cb5b0d", size = 469787, upload-time = "2025-07-29T05:50:49.669Z" }, - { url = "https://files.pythonhosted.org/packages/38/7d/b76438e70319796bfff717f325d97ce2e9310f752a267bfdf5192ac6082b/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2e5a495cb1be69dae4b08f35a6c4579c539e9b5706f606632102c0f855bcba7c", size = 1716590, upload-time = "2025-07-29T05:50:51.368Z" }, - { url = "https://files.pythonhosted.org/packages/79/b1/60370d70cdf8b269ee1444b390cbd72ce514f0d1cd1a715821c784d272c9/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:6404dfc8cdde35c69aaa489bb3542fb86ef215fc70277c892be8af540e5e21c0", size = 1699241, upload-time = "2025-07-29T05:50:53.628Z" }, - { url = "https://files.pythonhosted.org/packages/a3/2b/4968a7b8792437ebc12186db31523f541943e99bda8f30335c482bea6879/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3ead1c00f8521a5c9070fcb88f02967b1d8a0544e6d85c253f6968b785e1a2ab", size = 1754335, upload-time = "2025-07-29T05:50:55.394Z" }, - { url = "https://files.pythonhosted.org/packages/fb/c1/49524ed553f9a0bec1a11fac09e790f49ff669bcd14164f9fab608831c4d/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6990ef617f14450bc6b34941dba4f12d5613cbf4e33805932f853fbd1cf18bfb", size = 1800491, upload-time = "2025-07-29T05:50:57.202Z" }, - { url = "https://files.pythonhosted.org/packages/de/5e/3bf5acea47a96a28c121b167f5ef659cf71208b19e52a88cdfa5c37f1fcc/aiohttp-3.12.15-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd736ed420f4db2b8148b52b46b88ed038d0354255f9a73196b7bbce3ea97545", size = 1719929, upload-time = "2025-07-29T05:50:59.192Z" }, - { url = "https://files.pythonhosted.org/packages/39/94/8ae30b806835bcd1cba799ba35347dee6961a11bd507db634516210e91d8/aiohttp-3.12.15-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:3c5092ce14361a73086b90c6efb3948ffa5be2f5b6fbcf52e8d8c8b8848bb97c", size = 1635733, upload-time = "2025-07-29T05:51:01.394Z" }, - { url = "https://files.pythonhosted.org/packages/7a/46/06cdef71dd03acd9da7f51ab3a9107318aee12ad38d273f654e4f981583a/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:aaa2234bb60c4dbf82893e934d8ee8dea30446f0647e024074237a56a08c01bd", size = 1696790, upload-time = "2025-07-29T05:51:03.657Z" }, - { url = "https://files.pythonhosted.org/packages/02/90/6b4cfaaf92ed98d0ec4d173e78b99b4b1a7551250be8937d9d67ecb356b4/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:6d86a2fbdd14192e2f234a92d3b494dd4457e683ba07e5905a0b3ee25389ac9f", size = 1718245, upload-time = "2025-07-29T05:51:05.911Z" }, - { url = "https://files.pythonhosted.org/packages/2e/e6/2593751670fa06f080a846f37f112cbe6f873ba510d070136a6ed46117c6/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:a041e7e2612041a6ddf1c6a33b883be6a421247c7afd47e885969ee4cc58bd8d", size = 1658899, upload-time = "2025-07-29T05:51:07.753Z" }, - { url = "https://files.pythonhosted.org/packages/8f/28/c15bacbdb8b8eb5bf39b10680d129ea7410b859e379b03190f02fa104ffd/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5015082477abeafad7203757ae44299a610e89ee82a1503e3d4184e6bafdd519", size = 1738459, upload-time = "2025-07-29T05:51:09.56Z" }, - { url = "https://files.pythonhosted.org/packages/00/de/c269cbc4faa01fb10f143b1670633a8ddd5b2e1ffd0548f7aa49cb5c70e2/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:56822ff5ddfd1b745534e658faba944012346184fbfe732e0d6134b744516eea", size = 1766434, upload-time = "2025-07-29T05:51:11.423Z" }, - { url = "https://files.pythonhosted.org/packages/52/b0/4ff3abd81aa7d929b27d2e1403722a65fc87b763e3a97b3a2a494bfc63bc/aiohttp-3.12.15-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:b2acbbfff69019d9014508c4ba0401822e8bae5a5fdc3b6814285b71231b60f3", size = 1726045, upload-time = "2025-07-29T05:51:13.689Z" }, - { url = "https://files.pythonhosted.org/packages/71/16/949225a6a2dd6efcbd855fbd90cf476052e648fb011aa538e3b15b89a57a/aiohttp-3.12.15-cp312-cp312-win32.whl", hash = "sha256:d849b0901b50f2185874b9a232f38e26b9b3d4810095a7572eacea939132d4e1", size = 423591, upload-time = "2025-07-29T05:51:15.452Z" }, - { url = "https://files.pythonhosted.org/packages/2b/d8/fa65d2a349fe938b76d309db1a56a75c4fb8cc7b17a398b698488a939903/aiohttp-3.12.15-cp312-cp312-win_amd64.whl", hash = "sha256:b390ef5f62bb508a9d67cb3bba9b8356e23b3996da7062f1a57ce1a79d2b3d34", size = 450266, upload-time = "2025-07-29T05:51:17.239Z" }, - { url = "https://files.pythonhosted.org/packages/f2/33/918091abcf102e39d15aba2476ad9e7bd35ddb190dcdd43a854000d3da0d/aiohttp-3.12.15-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:9f922ffd05034d439dde1c77a20461cf4a1b0831e6caa26151fe7aa8aaebc315", size = 696741, upload-time = "2025-07-29T05:51:19.021Z" }, - { url = "https://files.pythonhosted.org/packages/b5/2a/7495a81e39a998e400f3ecdd44a62107254803d1681d9189be5c2e4530cd/aiohttp-3.12.15-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:2ee8a8ac39ce45f3e55663891d4b1d15598c157b4d494a4613e704c8b43112cd", size = 474407, upload-time = "2025-07-29T05:51:21.165Z" }, - { url = "https://files.pythonhosted.org/packages/49/fc/a9576ab4be2dcbd0f73ee8675d16c707cfc12d5ee80ccf4015ba543480c9/aiohttp-3.12.15-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3eae49032c29d356b94eee45a3f39fdf4b0814b397638c2f718e96cfadf4c4e4", size = 466703, upload-time = "2025-07-29T05:51:22.948Z" }, - { url = "https://files.pythonhosted.org/packages/09/2f/d4bcc8448cf536b2b54eed48f19682031ad182faa3a3fee54ebe5b156387/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b97752ff12cc12f46a9b20327104448042fce5c33a624f88c18f66f9368091c7", size = 1705532, upload-time = "2025-07-29T05:51:25.211Z" }, - { url = "https://files.pythonhosted.org/packages/f1/f3/59406396083f8b489261e3c011aa8aee9df360a96ac8fa5c2e7e1b8f0466/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:894261472691d6fe76ebb7fcf2e5870a2ac284c7406ddc95823c8598a1390f0d", size = 1686794, upload-time = "2025-07-29T05:51:27.145Z" }, - { url = "https://files.pythonhosted.org/packages/dc/71/164d194993a8d114ee5656c3b7ae9c12ceee7040d076bf7b32fb98a8c5c6/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5fa5d9eb82ce98959fc1031c28198b431b4d9396894f385cb63f1e2f3f20ca6b", size = 1738865, upload-time = "2025-07-29T05:51:29.366Z" }, - { url = "https://files.pythonhosted.org/packages/1c/00/d198461b699188a93ead39cb458554d9f0f69879b95078dce416d3209b54/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f0fa751efb11a541f57db59c1dd821bec09031e01452b2b6217319b3a1f34f3d", size = 1788238, upload-time = "2025-07-29T05:51:31.285Z" }, - { url = "https://files.pythonhosted.org/packages/85/b8/9e7175e1fa0ac8e56baa83bf3c214823ce250d0028955dfb23f43d5e61fd/aiohttp-3.12.15-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5346b93e62ab51ee2a9d68e8f73c7cf96ffb73568a23e683f931e52450e4148d", size = 1710566, upload-time = "2025-07-29T05:51:33.219Z" }, - { url = "https://files.pythonhosted.org/packages/59/e4/16a8eac9df39b48ae102ec030fa9f726d3570732e46ba0c592aeeb507b93/aiohttp-3.12.15-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:049ec0360f939cd164ecbfd2873eaa432613d5e77d6b04535e3d1fbae5a9e645", size = 1624270, upload-time = "2025-07-29T05:51:35.195Z" }, - { url = "https://files.pythonhosted.org/packages/1f/f8/cd84dee7b6ace0740908fd0af170f9fab50c2a41ccbc3806aabcb1050141/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:b52dcf013b57464b6d1e51b627adfd69a8053e84b7103a7cd49c030f9ca44461", size = 1677294, upload-time = "2025-07-29T05:51:37.215Z" }, - { url = "https://files.pythonhosted.org/packages/ce/42/d0f1f85e50d401eccd12bf85c46ba84f947a84839c8a1c2c5f6e8ab1eb50/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:9b2af240143dd2765e0fb661fd0361a1b469cab235039ea57663cda087250ea9", size = 1708958, upload-time = "2025-07-29T05:51:39.328Z" }, - { url = "https://files.pythonhosted.org/packages/d5/6b/f6fa6c5790fb602538483aa5a1b86fcbad66244997e5230d88f9412ef24c/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:ac77f709a2cde2cc71257ab2d8c74dd157c67a0558a0d2799d5d571b4c63d44d", size = 1651553, upload-time = "2025-07-29T05:51:41.356Z" }, - { url = "https://files.pythonhosted.org/packages/04/36/a6d36ad545fa12e61d11d1932eef273928b0495e6a576eb2af04297fdd3c/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:47f6b962246f0a774fbd3b6b7be25d59b06fdb2f164cf2513097998fc6a29693", size = 1727688, upload-time = "2025-07-29T05:51:43.452Z" }, - { url = "https://files.pythonhosted.org/packages/aa/c8/f195e5e06608a97a4e52c5d41c7927301bf757a8e8bb5bbf8cef6c314961/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:760fb7db442f284996e39cf9915a94492e1896baac44f06ae551974907922b64", size = 1761157, upload-time = "2025-07-29T05:51:45.643Z" }, - { url = "https://files.pythonhosted.org/packages/05/6a/ea199e61b67f25ba688d3ce93f63b49b0a4e3b3d380f03971b4646412fc6/aiohttp-3.12.15-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ad702e57dc385cae679c39d318def49aef754455f237499d5b99bea4ef582e51", size = 1710050, upload-time = "2025-07-29T05:51:48.203Z" }, - { url = "https://files.pythonhosted.org/packages/b4/2e/ffeb7f6256b33635c29dbed29a22a723ff2dd7401fff42ea60cf2060abfb/aiohttp-3.12.15-cp313-cp313-win32.whl", hash = "sha256:f813c3e9032331024de2eb2e32a88d86afb69291fbc37a3a3ae81cc9917fb3d0", size = 422647, upload-time = "2025-07-29T05:51:50.718Z" }, - { url = "https://files.pythonhosted.org/packages/1b/8e/78ee35774201f38d5e1ba079c9958f7629b1fd079459aea9467441dbfbf5/aiohttp-3.12.15-cp313-cp313-win_amd64.whl", hash = "sha256:1a649001580bdb37c6fdb1bebbd7e3bc688e8ec2b5c6f52edbb664662b17dc84", size = 449067, upload-time = "2025-07-29T05:51:52.549Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/63/e7/fa1a8c00e2c54b05dc8cb5d1439f627f7c267874e3f7bb047146116020f9/aiohttp-3.11.18.tar.gz", hash = "sha256:ae856e1138612b7e412db63b7708735cff4d38d0399f6a5435d3dac2669f558a", size = 7678653, upload-time = "2025-04-21T09:43:09.191Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c7/c3/e5f64af7e97a02f547020e6ff861595766bb5ecb37c7492fac9fe3c14f6c/aiohttp-3.11.18-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:96264854fedbea933a9ca4b7e0c745728f01380691687b7365d18d9e977179c4", size = 711703, upload-time = "2025-04-21T09:40:25.487Z" }, + { url = "https://files.pythonhosted.org/packages/5f/2f/53c26e96efa5fd01ebcfe1fefdfb7811f482bb21f4fa103d85eca4dcf888/aiohttp-3.11.18-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:9602044ff047043430452bc3a2089743fa85da829e6fc9ee0025351d66c332b6", size = 471348, upload-time = "2025-04-21T09:40:27.569Z" }, + { url = "https://files.pythonhosted.org/packages/80/47/dcc248464c9b101532ee7d254a46f6ed2c1fd3f4f0f794cf1f2358c0d45b/aiohttp-3.11.18-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:5691dc38750fcb96a33ceef89642f139aa315c8a193bbd42a0c33476fd4a1609", size = 457611, upload-time = "2025-04-21T09:40:28.978Z" }, + { url = "https://files.pythonhosted.org/packages/4c/ca/67d816ef075e8ac834b5f1f6b18e8db7d170f7aebaf76f1be462ea10cab0/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:554c918ec43f8480b47a5ca758e10e793bd7410b83701676a4782672d670da55", size = 1591976, upload-time = "2025-04-21T09:40:30.804Z" }, + { url = "https://files.pythonhosted.org/packages/46/00/0c120287aa51c744438d99e9aae9f8c55ca5b9911c42706966c91c9d68d6/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8a4076a2b3ba5b004b8cffca6afe18a3b2c5c9ef679b4d1e9859cf76295f8d4f", size = 1632819, upload-time = "2025-04-21T09:40:32.731Z" }, + { url = "https://files.pythonhosted.org/packages/54/a3/3923c9040cd4927dfee1aa017513701e35adcfc35d10729909688ecaa465/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:767a97e6900edd11c762be96d82d13a1d7c4fc4b329f054e88b57cdc21fded94", size = 1666567, upload-time = "2025-04-21T09:40:34.901Z" }, + { url = "https://files.pythonhosted.org/packages/e0/ab/40dacb15c0c58f7f17686ea67bc186e9f207341691bdb777d1d5ff4671d5/aiohttp-3.11.18-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f0ddc9337a0fb0e727785ad4f41163cc314376e82b31846d3835673786420ef1", size = 1594959, upload-time = "2025-04-21T09:40:36.714Z" }, + { url = "https://files.pythonhosted.org/packages/0d/98/d40c2b7c4a5483f9a16ef0adffce279ced3cc44522e84b6ba9e906be5168/aiohttp-3.11.18-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f414f37b244f2a97e79b98d48c5ff0789a0b4b4609b17d64fa81771ad780e415", size = 1538516, upload-time = "2025-04-21T09:40:38.263Z" }, + { url = "https://files.pythonhosted.org/packages/cf/10/e0bf3a03524faac45a710daa034e6f1878b24a1fef9c968ac8eb786ae657/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:fdb239f47328581e2ec7744ab5911f97afb10752332a6dd3d98e14e429e1a9e7", size = 1529037, upload-time = "2025-04-21T09:40:40.349Z" }, + { url = "https://files.pythonhosted.org/packages/ad/d6/5ff5282e00e4eb59c857844984cbc5628f933e2320792e19f93aff518f52/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:f2c50bad73ed629cc326cc0f75aed8ecfb013f88c5af116f33df556ed47143eb", size = 1546813, upload-time = "2025-04-21T09:40:42.106Z" }, + { url = "https://files.pythonhosted.org/packages/de/96/f1014f84101f9b9ad2d8acf3cc501426475f7f0cc62308ae5253e2fac9a7/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:0a8d8f20c39d3fa84d1c28cdb97f3111387e48209e224408e75f29c6f8e0861d", size = 1523852, upload-time = "2025-04-21T09:40:44.164Z" }, + { url = "https://files.pythonhosted.org/packages/a5/86/ec772c6838dd6bae3229065af671891496ac1834b252f305cee8152584b2/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:106032eaf9e62fd6bc6578c8b9e6dc4f5ed9a5c1c7fb2231010a1b4304393421", size = 1603766, upload-time = "2025-04-21T09:40:46.203Z" }, + { url = "https://files.pythonhosted.org/packages/84/38/31f85459c9402d409c1499284fc37a96f69afadce3cfac6a1b5ab048cbf1/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:b491e42183e8fcc9901d8dcd8ae644ff785590f1727f76ca86e731c61bfe6643", size = 1620647, upload-time = "2025-04-21T09:40:48.168Z" }, + { url = "https://files.pythonhosted.org/packages/31/2f/54aba0040764dd3d362fb37bd6aae9b3034fcae0b27f51b8a34864e48209/aiohttp-3.11.18-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:ad8c745ff9460a16b710e58e06a9dec11ebc0d8f4dd82091cefb579844d69868", size = 1559260, upload-time = "2025-04-21T09:40:50.219Z" }, + { url = "https://files.pythonhosted.org/packages/ca/d2/a05c7dd9e1b6948c1c5d04f1a8bcfd7e131923fa809bb87477d5c76f1517/aiohttp-3.11.18-cp310-cp310-win32.whl", hash = "sha256:8e57da93e24303a883146510a434f0faf2f1e7e659f3041abc4e3fb3f6702a9f", size = 418051, upload-time = "2025-04-21T09:40:52.272Z" }, + { url = "https://files.pythonhosted.org/packages/39/e2/796a6179e8abe267dfc84614a50291560a989d28acacbc5dab3bcd4cbec4/aiohttp-3.11.18-cp310-cp310-win_amd64.whl", hash = "sha256:cc93a4121d87d9f12739fc8fab0a95f78444e571ed63e40bfc78cd5abe700ac9", size = 442908, upload-time = "2025-04-21T09:40:54.345Z" }, + { url = "https://files.pythonhosted.org/packages/2f/10/fd9ee4f9e042818c3c2390054c08ccd34556a3cb209d83285616434cf93e/aiohttp-3.11.18-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:427fdc56ccb6901ff8088544bde47084845ea81591deb16f957897f0f0ba1be9", size = 712088, upload-time = "2025-04-21T09:40:55.776Z" }, + { url = "https://files.pythonhosted.org/packages/22/eb/6a77f055ca56f7aae2cd2a5607a3c9e7b9554f1497a069dcfcb52bfc9540/aiohttp-3.11.18-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2c828b6d23b984255b85b9b04a5b963a74278b7356a7de84fda5e3b76866597b", size = 471450, upload-time = "2025-04-21T09:40:57.301Z" }, + { url = "https://files.pythonhosted.org/packages/78/dc/5f3c0d27c91abf0bb5d103e9c9b0ff059f60cf6031a5f06f456c90731f42/aiohttp-3.11.18-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5c2eaa145bb36b33af1ff2860820ba0589e165be4ab63a49aebfd0981c173b66", size = 457836, upload-time = "2025-04-21T09:40:59.322Z" }, + { url = "https://files.pythonhosted.org/packages/49/7b/55b65af9ef48b9b811c91ff8b5b9de9650c71147f10523e278d297750bc8/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3d518ce32179f7e2096bf4e3e8438cf445f05fedd597f252de9f54c728574756", size = 1690978, upload-time = "2025-04-21T09:41:00.795Z" }, + { url = "https://files.pythonhosted.org/packages/a2/5a/3f8938c4f68ae400152b42742653477fc625d6bfe02e764f3521321c8442/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0700055a6e05c2f4711011a44364020d7a10fbbcd02fbf3e30e8f7e7fddc8717", size = 1745307, upload-time = "2025-04-21T09:41:02.89Z" }, + { url = "https://files.pythonhosted.org/packages/b4/42/89b694a293333ef6f771c62da022163bcf44fb03d4824372d88e3dc12530/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8bd1cde83e4684324e6ee19adfc25fd649d04078179890be7b29f76b501de8e4", size = 1780692, upload-time = "2025-04-21T09:41:04.461Z" }, + { url = "https://files.pythonhosted.org/packages/e2/ce/1a75384e01dd1bf546898b6062b1b5f7a59b6692ef802e4dd6db64fed264/aiohttp-3.11.18-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:73b8870fe1c9a201b8c0d12c94fe781b918664766728783241a79e0468427e4f", size = 1676934, upload-time = "2025-04-21T09:41:06.728Z" }, + { url = "https://files.pythonhosted.org/packages/a5/31/442483276e6c368ab5169797d9873b5875213cbcf7e74b95ad1c5003098a/aiohttp-3.11.18-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:25557982dd36b9e32c0a3357f30804e80790ec2c4d20ac6bcc598533e04c6361", size = 1621190, upload-time = "2025-04-21T09:41:08.293Z" }, + { url = "https://files.pythonhosted.org/packages/7b/83/90274bf12c079457966008a58831a99675265b6a34b505243e004b408934/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:7e889c9df381a2433802991288a61e5a19ceb4f61bd14f5c9fa165655dcb1fd1", size = 1658947, upload-time = "2025-04-21T09:41:11.054Z" }, + { url = "https://files.pythonhosted.org/packages/91/c1/da9cee47a0350b78fdc93670ebe7ad74103011d7778ab4c382ca4883098d/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:9ea345fda05bae217b6cce2acf3682ce3b13d0d16dd47d0de7080e5e21362421", size = 1654443, upload-time = "2025-04-21T09:41:13.213Z" }, + { url = "https://files.pythonhosted.org/packages/c9/f2/73cbe18dc25d624f79a09448adfc4972f82ed6088759ddcf783cd201956c/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:9f26545b9940c4b46f0a9388fd04ee3ad7064c4017b5a334dd450f616396590e", size = 1644169, upload-time = "2025-04-21T09:41:14.827Z" }, + { url = "https://files.pythonhosted.org/packages/5b/32/970b0a196c4dccb1b0cfa5b4dc3b20f63d76f1c608f41001a84b2fd23c3d/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:3a621d85e85dccabd700294494d7179ed1590b6d07a35709bb9bd608c7f5dd1d", size = 1728532, upload-time = "2025-04-21T09:41:17.168Z" }, + { url = "https://files.pythonhosted.org/packages/0b/50/b1dc810a41918d2ea9574e74125eb053063bc5e14aba2d98966f7d734da0/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:9c23fd8d08eb9c2af3faeedc8c56e134acdaf36e2117ee059d7defa655130e5f", size = 1750310, upload-time = "2025-04-21T09:41:19.353Z" }, + { url = "https://files.pythonhosted.org/packages/95/24/39271f5990b35ff32179cc95537e92499d3791ae82af7dcf562be785cd15/aiohttp-3.11.18-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d9e6b0e519067caa4fd7fb72e3e8002d16a68e84e62e7291092a5433763dc0dd", size = 1691580, upload-time = "2025-04-21T09:41:21.868Z" }, + { url = "https://files.pythonhosted.org/packages/6b/78/75d0353feb77f041460564f12fe58e456436bbc00cbbf5d676dbf0038cc2/aiohttp-3.11.18-cp311-cp311-win32.whl", hash = "sha256:122f3e739f6607e5e4c6a2f8562a6f476192a682a52bda8b4c6d4254e1138f4d", size = 417565, upload-time = "2025-04-21T09:41:24.78Z" }, + { url = "https://files.pythonhosted.org/packages/ed/97/b912dcb654634a813f8518de359364dfc45976f822116e725dc80a688eee/aiohttp-3.11.18-cp311-cp311-win_amd64.whl", hash = "sha256:e6f3c0a3a1e73e88af384b2e8a0b9f4fb73245afd47589df2afcab6b638fa0e6", size = 443652, upload-time = "2025-04-21T09:41:26.48Z" }, + { url = "https://files.pythonhosted.org/packages/b5/d2/5bc436f42bf4745c55f33e1e6a2d69e77075d3e768e3d1a34f96ee5298aa/aiohttp-3.11.18-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:63d71eceb9cad35d47d71f78edac41fcd01ff10cacaa64e473d1aec13fa02df2", size = 706671, upload-time = "2025-04-21T09:41:28.021Z" }, + { url = "https://files.pythonhosted.org/packages/fe/d0/2dbabecc4e078c0474abb40536bbde717fb2e39962f41c5fc7a216b18ea7/aiohttp-3.11.18-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:d1929da615840969929e8878d7951b31afe0bac883d84418f92e5755d7b49508", size = 466169, upload-time = "2025-04-21T09:41:29.783Z" }, + { url = "https://files.pythonhosted.org/packages/70/84/19edcf0b22933932faa6e0be0d933a27bd173da02dc125b7354dff4d8da4/aiohttp-3.11.18-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:7d0aebeb2392f19b184e3fdd9e651b0e39cd0f195cdb93328bd124a1d455cd0e", size = 457554, upload-time = "2025-04-21T09:41:31.327Z" }, + { url = "https://files.pythonhosted.org/packages/32/d0/e8d1f034ae5624a0f21e4fb3feff79342ce631f3a4d26bd3e58b31ef033b/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3849ead845e8444f7331c284132ab314b4dac43bfae1e3cf350906d4fff4620f", size = 1690154, upload-time = "2025-04-21T09:41:33.541Z" }, + { url = "https://files.pythonhosted.org/packages/16/de/2f9dbe2ac6f38f8495562077131888e0d2897e3798a0ff3adda766b04a34/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5e8452ad6b2863709f8b3d615955aa0807bc093c34b8e25b3b52097fe421cb7f", size = 1733402, upload-time = "2025-04-21T09:41:35.634Z" }, + { url = "https://files.pythonhosted.org/packages/e0/04/bd2870e1e9aef990d14b6df2a695f17807baf5c85a4c187a492bda569571/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3b8d2b42073611c860a37f718b3d61ae8b4c2b124b2e776e2c10619d920350ec", size = 1783958, upload-time = "2025-04-21T09:41:37.456Z" }, + { url = "https://files.pythonhosted.org/packages/23/06/4203ffa2beb5bedb07f0da0f79b7d9039d1c33f522e0d1a2d5b6218e6f2e/aiohttp-3.11.18-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:40fbf91f6a0ac317c0a07eb328a1384941872f6761f2e6f7208b63c4cc0a7ff6", size = 1695288, upload-time = "2025-04-21T09:41:39.756Z" }, + { url = "https://files.pythonhosted.org/packages/30/b2/e2285dda065d9f29ab4b23d8bcc81eb881db512afb38a3f5247b191be36c/aiohttp-3.11.18-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:44ff5625413fec55216da5eaa011cf6b0a2ed67a565914a212a51aa3755b0009", size = 1618871, upload-time = "2025-04-21T09:41:41.972Z" }, + { url = "https://files.pythonhosted.org/packages/57/e0/88f2987885d4b646de2036f7296ebea9268fdbf27476da551c1a7c158bc0/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7f33a92a2fde08e8c6b0c61815521324fc1612f397abf96eed86b8e31618fdb4", size = 1646262, upload-time = "2025-04-21T09:41:44.192Z" }, + { url = "https://files.pythonhosted.org/packages/e0/19/4d2da508b4c587e7472a032290b2981f7caeca82b4354e19ab3df2f51d56/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:11d5391946605f445ddafda5eab11caf310f90cdda1fd99865564e3164f5cff9", size = 1677431, upload-time = "2025-04-21T09:41:46.049Z" }, + { url = "https://files.pythonhosted.org/packages/eb/ae/047473ea50150a41440f3265f53db1738870b5a1e5406ece561ca61a3bf4/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:3cc314245deb311364884e44242e00c18b5896e4fe6d5f942e7ad7e4cb640adb", size = 1637430, upload-time = "2025-04-21T09:41:47.973Z" }, + { url = "https://files.pythonhosted.org/packages/11/32/c6d1e3748077ce7ee13745fae33e5cb1dac3e3b8f8787bf738a93c94a7d2/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:0f421843b0f70740772228b9e8093289924359d306530bcd3926f39acbe1adda", size = 1703342, upload-time = "2025-04-21T09:41:50.323Z" }, + { url = "https://files.pythonhosted.org/packages/c5/1d/a3b57bfdbe285f0d45572d6d8f534fd58761da3e9cbc3098372565005606/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:e220e7562467dc8d589e31c1acd13438d82c03d7f385c9cd41a3f6d1d15807c1", size = 1740600, upload-time = "2025-04-21T09:41:52.111Z" }, + { url = "https://files.pythonhosted.org/packages/a5/71/f9cd2fed33fa2b7ce4d412fb7876547abb821d5b5520787d159d0748321d/aiohttp-3.11.18-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ab2ef72f8605046115bc9aa8e9d14fd49086d405855f40b79ed9e5c1f9f4faea", size = 1695131, upload-time = "2025-04-21T09:41:53.94Z" }, + { url = "https://files.pythonhosted.org/packages/97/97/d1248cd6d02b9de6aa514793d0dcb20099f0ec47ae71a933290116c070c5/aiohttp-3.11.18-cp312-cp312-win32.whl", hash = "sha256:12a62691eb5aac58d65200c7ae94d73e8a65c331c3a86a2e9670927e94339ee8", size = 412442, upload-time = "2025-04-21T09:41:55.689Z" }, + { url = "https://files.pythonhosted.org/packages/33/9a/e34e65506e06427b111e19218a99abf627638a9703f4b8bcc3e3021277ed/aiohttp-3.11.18-cp312-cp312-win_amd64.whl", hash = "sha256:364329f319c499128fd5cd2d1c31c44f234c58f9b96cc57f743d16ec4f3238c8", size = 439444, upload-time = "2025-04-21T09:41:57.977Z" }, + { url = "https://files.pythonhosted.org/packages/0a/18/be8b5dd6b9cf1b2172301dbed28e8e5e878ee687c21947a6c81d6ceaa15d/aiohttp-3.11.18-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:474215ec618974054cf5dc465497ae9708543cbfc312c65212325d4212525811", size = 699833, upload-time = "2025-04-21T09:42:00.298Z" }, + { url = "https://files.pythonhosted.org/packages/0d/84/ecdc68e293110e6f6f6d7b57786a77555a85f70edd2b180fb1fafaff361a/aiohttp-3.11.18-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:6ced70adf03920d4e67c373fd692123e34d3ac81dfa1c27e45904a628567d804", size = 462774, upload-time = "2025-04-21T09:42:02.015Z" }, + { url = "https://files.pythonhosted.org/packages/d7/85/f07718cca55884dad83cc2433746384d267ee970e91f0dcc75c6d5544079/aiohttp-3.11.18-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:2d9f6c0152f8d71361905aaf9ed979259537981f47ad099c8b3d81e0319814bd", size = 454429, upload-time = "2025-04-21T09:42:03.728Z" }, + { url = "https://files.pythonhosted.org/packages/82/02/7f669c3d4d39810db8842c4e572ce4fe3b3a9b82945fdd64affea4c6947e/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a35197013ed929c0aed5c9096de1fc5a9d336914d73ab3f9df14741668c0616c", size = 1670283, upload-time = "2025-04-21T09:42:06.053Z" }, + { url = "https://files.pythonhosted.org/packages/ec/79/b82a12f67009b377b6c07a26bdd1b81dab7409fc2902d669dbfa79e5ac02/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:540b8a1f3a424f1af63e0af2d2853a759242a1769f9f1ab053996a392bd70118", size = 1717231, upload-time = "2025-04-21T09:42:07.953Z" }, + { url = "https://files.pythonhosted.org/packages/a6/38/d5a1f28c3904a840642b9a12c286ff41fc66dfa28b87e204b1f242dbd5e6/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f9e6710ebebfce2ba21cee6d91e7452d1125100f41b906fb5af3da8c78b764c1", size = 1769621, upload-time = "2025-04-21T09:42:09.855Z" }, + { url = "https://files.pythonhosted.org/packages/53/2d/deb3749ba293e716b5714dda06e257f123c5b8679072346b1eb28b766a0b/aiohttp-3.11.18-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f8af2ef3b4b652ff109f98087242e2ab974b2b2b496304063585e3d78de0b000", size = 1678667, upload-time = "2025-04-21T09:42:11.741Z" }, + { url = "https://files.pythonhosted.org/packages/b8/a8/04b6e11683a54e104b984bd19a9790eb1ae5f50968b601bb202d0406f0ff/aiohttp-3.11.18-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:28c3f975e5ae3dbcbe95b7e3dcd30e51da561a0a0f2cfbcdea30fc1308d72137", size = 1601592, upload-time = "2025-04-21T09:42:14.137Z" }, + { url = "https://files.pythonhosted.org/packages/5e/9d/c33305ae8370b789423623f0e073d09ac775cd9c831ac0f11338b81c16e0/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c28875e316c7b4c3e745172d882d8a5c835b11018e33432d281211af35794a93", size = 1621679, upload-time = "2025-04-21T09:42:16.056Z" }, + { url = "https://files.pythonhosted.org/packages/56/45/8e9a27fff0538173d47ba60362823358f7a5f1653c6c30c613469f94150e/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:13cd38515568ae230e1ef6919e2e33da5d0f46862943fcda74e7e915096815f3", size = 1656878, upload-time = "2025-04-21T09:42:18.368Z" }, + { url = "https://files.pythonhosted.org/packages/84/5b/8c5378f10d7a5a46b10cb9161a3aac3eeae6dba54ec0f627fc4ddc4f2e72/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:0e2a92101efb9f4c2942252c69c63ddb26d20f46f540c239ccfa5af865197bb8", size = 1620509, upload-time = "2025-04-21T09:42:20.141Z" }, + { url = "https://files.pythonhosted.org/packages/9e/2f/99dee7bd91c62c5ff0aa3c55f4ae7e1bc99c6affef780d7777c60c5b3735/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:e6d3e32b8753c8d45ac550b11a1090dd66d110d4ef805ffe60fa61495360b3b2", size = 1680263, upload-time = "2025-04-21T09:42:21.993Z" }, + { url = "https://files.pythonhosted.org/packages/03/0a/378745e4ff88acb83e2d5c884a4fe993a6e9f04600a4560ce0e9b19936e3/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:ea4cf2488156e0f281f93cc2fd365025efcba3e2d217cbe3df2840f8c73db261", size = 1715014, upload-time = "2025-04-21T09:42:23.87Z" }, + { url = "https://files.pythonhosted.org/packages/f6/0b/b5524b3bb4b01e91bc4323aad0c2fcaebdf2f1b4d2eb22743948ba364958/aiohttp-3.11.18-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9d4df95ad522c53f2b9ebc07f12ccd2cb15550941e11a5bbc5ddca2ca56316d7", size = 1666614, upload-time = "2025-04-21T09:42:25.764Z" }, + { url = "https://files.pythonhosted.org/packages/c7/b7/3d7b036d5a4ed5a4c704e0754afe2eef24a824dfab08e6efbffb0f6dd36a/aiohttp-3.11.18-cp313-cp313-win32.whl", hash = "sha256:cdd1bbaf1e61f0d94aced116d6e95fe25942f7a5f42382195fd9501089db5d78", size = 411358, upload-time = "2025-04-21T09:42:27.558Z" }, + { url = "https://files.pythonhosted.org/packages/1e/3c/143831b32cd23b5263a995b2a1794e10aa42f8a895aae5074c20fda36c07/aiohttp-3.11.18-cp313-cp313-win_amd64.whl", hash = "sha256:bdd619c27e44382cf642223f11cfd4d795161362a5a1fc1fa3940397bc89db01", size = 437658, upload-time = "2025-04-21T09:42:29.209Z" }, ] [[package]] @@ -141,15 +137,14 @@ wheels = [ [[package]] name = "aiosignal" -version = "1.4.0" +version = "1.3.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "frozenlist" }, - { name = "typing-extensions", marker = "python_full_version < '3.13'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/61/62/06741b579156360248d1ec624842ad0edf697050bbaf7c3e46394e106ad1/aiosignal-1.4.0.tar.gz", hash = "sha256:f47eecd9468083c2029cc99945502cb7708b082c232f9aca65da147157b251c7", size = 25007, upload-time = "2025-07-03T22:54:43.528Z" } +sdist = { url = "https://files.pythonhosted.org/packages/ba/b5/6d55e80f6d8a08ce22b982eafa278d823b541c925f11ee774b0b9c43473d/aiosignal-1.3.2.tar.gz", hash = "sha256:a8c255c66fafb1e499c9351d0bf32ff2d8a0321595ebac3b93713656d2436f54", size = 19424, upload-time = "2024-12-13T17:10:40.86Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/fb/76/641ae371508676492379f16e2fa48f4e2c11741bd63c48be4b12a6b09cba/aiosignal-1.4.0-py3-none-any.whl", hash = "sha256:053243f8b92b990551949e63930a839ff0cf0b0ebbe0597b0f3fb19e1a0fe82e", size = 7490, upload-time = "2025-07-03T22:54:42.156Z" }, + { url = "https://files.pythonhosted.org/packages/ec/6a/bc7e17a3e87a2985d3e8f4da4cd0f481060eb78fb08596c42be62c90a4d9/aiosignal-1.3.2-py2.py3-none-any.whl", hash = "sha256:45cde58e409a301715980c2b01d0c28bdde3770d8290b5eb2173759d9acb31a5", size = 7597, upload-time = "2024-12-13T17:10:38.469Z" }, ] [[package]] @@ -178,18 +173,16 @@ wheels = [ [[package]] name = "ase" -version = "3.25.0" +version = "3.24.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "matplotlib" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, + { name = "scipy" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e8/a1/5735ced2f159979f5b27c4083126b7796a5750cee6f027864e59818a5b76/ase-3.25.0.tar.gz", hash = "sha256:374cf8ca9fe588f05d6e856da3c9c17ef262dc968027b231d449334140c962c2", size = 2400055, upload-time = "2025-04-11T17:14:39.975Z" } +sdist = { url = "https://files.pythonhosted.org/packages/5c/c9/9adb9bc641bd7222367886e4e6c753b4c64da4ff2d9565ab39aee1e34734/ase-3.24.0.tar.gz", hash = "sha256:9acc93d6daaf48cd27b844c56f8bf49428b9db0542faa3cc30d9d5b8e1842195", size = 2383264, upload-time = "2024-12-28T22:20:33.711Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/07/f5/007d993fcf3b051acb304d5402e0bd103fd20816b47dee9531bdbfb3aa0c/ase-3.25.0-py3-none-any.whl", hash = "sha256:f9a5295e1154da355af04726d001fa76a311c076616d98e49cd9f34fc3afe188", size = 2951559, upload-time = "2025-04-11T17:14:37.617Z" }, + { url = "https://files.pythonhosted.org/packages/1f/cd/b1253035a1da90e89f31947e052c558cd83df3bcaff34aa199e5e806d773/ase-3.24.0-py3-none-any.whl", hash = "sha256:974922df87ef4ec8cf1140359a55ab4c4dc55c38e26876bdd9c00968da1f463c", size = 2928893, upload-time = "2024-12-28T22:20:29.416Z" }, ] [[package]] @@ -221,30 +214,30 @@ wheels = [ [[package]] name = "boto3" -version = "1.39.8" +version = "1.37.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "botocore" }, { name = "jmespath" }, { name = "s3transfer" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/d1/ef/f8dbe6482bdf9eb0230f2639483cdd40ef5aaa89c2fb651f2edeee9c248a/boto3-1.39.8.tar.gz", hash = "sha256:456ea6baef037eb6205d64e012259d14f0c9300c9b30603890746c1a0882fa01", size = 111829, upload-time = "2025-07-17T19:19:14.828Z" } +sdist = { url = "https://files.pythonhosted.org/packages/21/8c/c2af03daafaacea1db1823d23073facffa75818b61d376c3be77dd297ae8/boto3-1.37.1.tar.gz", hash = "sha256:96d18f7feb0c1fcb95f8837b74b6c8880e1b4e35ce5f8a8f8cb243a090c278ed", size = 111175, upload-time = "2025-02-25T20:33:16.471Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/3c/f0/f3701472b2e6192e62d80e703186ae9c789b3d607ba22943702c500897d2/boto3-1.39.8-py3-none-any.whl", hash = "sha256:dcea5270ccced0b4b962eb5874cb71b6232ccfc6203e05bf834a314442e4a79c", size = 139886, upload-time = "2025-07-17T19:19:12.634Z" }, + { url = "https://files.pythonhosted.org/packages/63/ec/e722c53c9dc41e8df094587c32e19409bace8b43b5eb31fe3536ca57a38b/boto3-1.37.1-py3-none-any.whl", hash = "sha256:4320441f904435a1b85e6ecb81793192e522c737cc9ed6566014e29f0a11cb22", size = 139338, upload-time = "2025-02-25T20:33:11.935Z" }, ] [[package]] name = "botocore" -version = "1.39.8" +version = "1.37.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "jmespath" }, { name = "python-dateutil" }, { name = "urllib3" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/32/57/16d3d21963975b9be180e96695abfb146695ae7db57f9a2d47e92d33ce9d/botocore-1.39.8.tar.gz", hash = "sha256:3848bd9057ea8dbc059e7764eda63bda575727ad1101dbd03636ab4a6f283fa5", size = 14205898, upload-time = "2025-07-17T19:19:03.832Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e5/01/3083bff25fd91193162298920cb093b9095609408416526d52b2826965b7/botocore-1.37.1.tar.gz", hash = "sha256:b194db8fb2a0ffba53568c364ae26166e7eec0445496b2ac86a6e142f3dd982f", size = 13578835, upload-time = "2025-02-25T20:32:56.63Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/11/ac/51462dd35fc60d11cdce93ba82ccf1635a161ceadc646d89f67d666fff31/botocore-1.39.8-py3-none-any.whl", hash = "sha256:ab43f79c6893271934faba7ae1987a313d59576361c544c70a5391ade560891d", size = 13866818, upload-time = "2025-07-17T19:18:58.521Z" }, + { url = "https://files.pythonhosted.org/packages/3d/20/352b2bf99f93ba18986615841786cbd0d38f7856bd49d4e154a540f04afe/botocore-1.37.1-py3-none-any.whl", hash = "sha256:c1db1bfc5d8c6b3b6d1ca6794f605294b4264e82a7e727b88e0fef9c2b9fbb9c", size = 13359164, upload-time = "2025-02-25T20:32:52.347Z" }, ] [[package]] @@ -275,11 +268,11 @@ wheels = [ [[package]] name = "certifi" -version = "2025.8.3" +version = "2025.1.31" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/dc/67/960ebe6bf230a96cda2e0abcf73af550ec4f090005363542f0765df162e0/certifi-2025.8.3.tar.gz", hash = "sha256:e564105f78ded564e3ae7c923924435e1daa7463faeab5bb932bc53ffae63407", size = 162386, upload-time = "2025-08-03T03:07:47.08Z" } +sdist = { url = "https://files.pythonhosted.org/packages/1c/ab/c9f1e32b7b1bf505bf26f0ef697775960db7932abeb7b516de930ba2705f/certifi-2025.1.31.tar.gz", hash = "sha256:3d5da6925056f6f18f119200434a4780a94263f10d1c21d032a6f6b2baa20651", size = 167577, upload-time = "2025-01-31T02:16:47.166Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e5/48/1549795ba7742c948d2ad169c1c8cdbae65bc450d6cd753d124b17c8cd32/certifi-2025.8.3-py3-none-any.whl", hash = "sha256:f6c12493cfb1b06ba2ff328595af9350c65d6644968e5d3a2ffd78699af217a5", size = 161216, upload-time = "2025-08-03T03:07:45.777Z" }, + { url = "https://files.pythonhosted.org/packages/38/fc/bce832fd4fd99766c04d1ee0eead6b0ec6486fb100ae5e74c1d91292b982/certifi-2025.1.31-py3-none-any.whl", hash = "sha256:ca78db4565a652026a4db2bcdf68f2fb589ea80d0be70e03929ed730746b84fe", size = 166393, upload-time = "2025-01-31T02:16:45.015Z" }, ] [[package]] @@ -353,7 +346,7 @@ name = "cftime" version = "1.6.4.post1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "numpy" }, ] sdist = { url = "https://files.pythonhosted.org/packages/ab/c8/1155d1d58003105307c7e5985f422ae5bcb2ca0cbc553cc828f3c5a934a7/cftime-1.6.4.post1.tar.gz", hash = "sha256:50ac76cc9f10ab7bd46e44a71c51a6927051b499b4407df4f29ab13d741b942f", size = 54631, upload-time = "2024-10-22T18:48:34.194Z" } wheels = [ @@ -385,75 +378,75 @@ wheels = [ [[package]] name = "charset-normalizer" -version = "3.4.2" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/e4/33/89c2ced2b67d1c2a61c19c6751aa8902d46ce3dacb23600a283619f5a12d/charset_normalizer-3.4.2.tar.gz", hash = "sha256:5baececa9ecba31eff645232d59845c07aa030f0c81ee70184a90d35099a0e63", size = 126367, upload-time = "2025-05-02T08:34:42.01Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/95/28/9901804da60055b406e1a1c5ba7aac1276fb77f1dde635aabfc7fd84b8ab/charset_normalizer-3.4.2-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:7c48ed483eb946e6c04ccbe02c6b4d1d48e51944b6db70f697e089c193404941", size = 201818, upload-time = "2025-05-02T08:31:46.725Z" }, - { url = "https://files.pythonhosted.org/packages/d9/9b/892a8c8af9110935e5adcbb06d9c6fe741b6bb02608c6513983048ba1a18/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b2d318c11350e10662026ad0eb71bb51c7812fc8590825304ae0bdd4ac283acd", size = 144649, upload-time = "2025-05-02T08:31:48.889Z" }, - { url = "https://files.pythonhosted.org/packages/7b/a5/4179abd063ff6414223575e008593861d62abfc22455b5d1a44995b7c101/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9cbfacf36cb0ec2897ce0ebc5d08ca44213af24265bd56eca54bee7923c48fd6", size = 155045, upload-time = "2025-05-02T08:31:50.757Z" }, - { url = "https://files.pythonhosted.org/packages/3b/95/bc08c7dfeddd26b4be8c8287b9bb055716f31077c8b0ea1cd09553794665/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:18dd2e350387c87dabe711b86f83c9c78af772c748904d372ade190b5c7c9d4d", size = 147356, upload-time = "2025-05-02T08:31:52.634Z" }, - { url = "https://files.pythonhosted.org/packages/a8/2d/7a5b635aa65284bf3eab7653e8b4151ab420ecbae918d3e359d1947b4d61/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8075c35cd58273fee266c58c0c9b670947c19df5fb98e7b66710e04ad4e9ff86", size = 149471, upload-time = "2025-05-02T08:31:56.207Z" }, - { url = "https://files.pythonhosted.org/packages/ae/38/51fc6ac74251fd331a8cfdb7ec57beba8c23fd5493f1050f71c87ef77ed0/charset_normalizer-3.4.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5bf4545e3b962767e5c06fe1738f951f77d27967cb2caa64c28be7c4563e162c", size = 151317, upload-time = "2025-05-02T08:31:57.613Z" }, - { url = "https://files.pythonhosted.org/packages/b7/17/edee1e32215ee6e9e46c3e482645b46575a44a2d72c7dfd49e49f60ce6bf/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:7a6ab32f7210554a96cd9e33abe3ddd86732beeafc7a28e9955cdf22ffadbab0", size = 146368, upload-time = "2025-05-02T08:31:59.468Z" }, - { url = "https://files.pythonhosted.org/packages/26/2c/ea3e66f2b5f21fd00b2825c94cafb8c326ea6240cd80a91eb09e4a285830/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:b33de11b92e9f75a2b545d6e9b6f37e398d86c3e9e9653c4864eb7e89c5773ef", size = 154491, upload-time = "2025-05-02T08:32:01.219Z" }, - { url = "https://files.pythonhosted.org/packages/52/47/7be7fa972422ad062e909fd62460d45c3ef4c141805b7078dbab15904ff7/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:8755483f3c00d6c9a77f490c17e6ab0c8729e39e6390328e42521ef175380ae6", size = 157695, upload-time = "2025-05-02T08:32:03.045Z" }, - { url = "https://files.pythonhosted.org/packages/2f/42/9f02c194da282b2b340f28e5fb60762de1151387a36842a92b533685c61e/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:68a328e5f55ec37c57f19ebb1fdc56a248db2e3e9ad769919a58672958e8f366", size = 154849, upload-time = "2025-05-02T08:32:04.651Z" }, - { url = "https://files.pythonhosted.org/packages/67/44/89cacd6628f31fb0b63201a618049be4be2a7435a31b55b5eb1c3674547a/charset_normalizer-3.4.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:21b2899062867b0e1fde9b724f8aecb1af14f2778d69aacd1a5a1853a597a5db", size = 150091, upload-time = "2025-05-02T08:32:06.719Z" }, - { url = "https://files.pythonhosted.org/packages/1f/79/4b8da9f712bc079c0f16b6d67b099b0b8d808c2292c937f267d816ec5ecc/charset_normalizer-3.4.2-cp310-cp310-win32.whl", hash = "sha256:e8082b26888e2f8b36a042a58307d5b917ef2b1cacab921ad3323ef91901c71a", size = 98445, upload-time = "2025-05-02T08:32:08.66Z" }, - { url = "https://files.pythonhosted.org/packages/7d/d7/96970afb4fb66497a40761cdf7bd4f6fca0fc7bafde3a84f836c1f57a926/charset_normalizer-3.4.2-cp310-cp310-win_amd64.whl", hash = "sha256:f69a27e45c43520f5487f27627059b64aaf160415589230992cec34c5e18a509", size = 105782, upload-time = "2025-05-02T08:32:10.46Z" }, - { url = "https://files.pythonhosted.org/packages/05/85/4c40d00dcc6284a1c1ad5de5e0996b06f39d8232f1031cd23c2f5c07ee86/charset_normalizer-3.4.2-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:be1e352acbe3c78727a16a455126d9ff83ea2dfdcbc83148d2982305a04714c2", size = 198794, upload-time = "2025-05-02T08:32:11.945Z" }, - { url = "https://files.pythonhosted.org/packages/41/d9/7a6c0b9db952598e97e93cbdfcb91bacd89b9b88c7c983250a77c008703c/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa88ca0b1932e93f2d961bf3addbb2db902198dca337d88c89e1559e066e7645", size = 142846, upload-time = "2025-05-02T08:32:13.946Z" }, - { url = "https://files.pythonhosted.org/packages/66/82/a37989cda2ace7e37f36c1a8ed16c58cf48965a79c2142713244bf945c89/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d524ba3f1581b35c03cb42beebab4a13e6cdad7b36246bd22541fa585a56cccd", size = 153350, upload-time = "2025-05-02T08:32:15.873Z" }, - { url = "https://files.pythonhosted.org/packages/df/68/a576b31b694d07b53807269d05ec3f6f1093e9545e8607121995ba7a8313/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:28a1005facc94196e1fb3e82a3d442a9d9110b8434fc1ded7a24a2983c9888d8", size = 145657, upload-time = "2025-05-02T08:32:17.283Z" }, - { url = "https://files.pythonhosted.org/packages/92/9b/ad67f03d74554bed3aefd56fe836e1623a50780f7c998d00ca128924a499/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fdb20a30fe1175ecabed17cbf7812f7b804b8a315a25f24678bcdf120a90077f", size = 147260, upload-time = "2025-05-02T08:32:18.807Z" }, - { url = "https://files.pythonhosted.org/packages/a6/e6/8aebae25e328160b20e31a7e9929b1578bbdc7f42e66f46595a432f8539e/charset_normalizer-3.4.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0f5d9ed7f254402c9e7d35d2f5972c9bbea9040e99cd2861bd77dc68263277c7", size = 149164, upload-time = "2025-05-02T08:32:20.333Z" }, - { url = "https://files.pythonhosted.org/packages/8b/f2/b3c2f07dbcc248805f10e67a0262c93308cfa149a4cd3d1fe01f593e5fd2/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:efd387a49825780ff861998cd959767800d54f8308936b21025326de4b5a42b9", size = 144571, upload-time = "2025-05-02T08:32:21.86Z" }, - { url = "https://files.pythonhosted.org/packages/60/5b/c3f3a94bc345bc211622ea59b4bed9ae63c00920e2e8f11824aa5708e8b7/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:f0aa37f3c979cf2546b73e8222bbfa3dc07a641585340179d768068e3455e544", size = 151952, upload-time = "2025-05-02T08:32:23.434Z" }, - { url = "https://files.pythonhosted.org/packages/e2/4d/ff460c8b474122334c2fa394a3f99a04cf11c646da895f81402ae54f5c42/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:e70e990b2137b29dc5564715de1e12701815dacc1d056308e2b17e9095372a82", size = 155959, upload-time = "2025-05-02T08:32:24.993Z" }, - { url = "https://files.pythonhosted.org/packages/a2/2b/b964c6a2fda88611a1fe3d4c400d39c66a42d6c169c924818c848f922415/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:0c8c57f84ccfc871a48a47321cfa49ae1df56cd1d965a09abe84066f6853b9c0", size = 153030, upload-time = "2025-05-02T08:32:26.435Z" }, - { url = "https://files.pythonhosted.org/packages/59/2e/d3b9811db26a5ebf444bc0fa4f4be5aa6d76fc6e1c0fd537b16c14e849b6/charset_normalizer-3.4.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:6b66f92b17849b85cad91259efc341dce9c1af48e2173bf38a85c6329f1033e5", size = 148015, upload-time = "2025-05-02T08:32:28.376Z" }, - { url = "https://files.pythonhosted.org/packages/90/07/c5fd7c11eafd561bb51220d600a788f1c8d77c5eef37ee49454cc5c35575/charset_normalizer-3.4.2-cp311-cp311-win32.whl", hash = "sha256:daac4765328a919a805fa5e2720f3e94767abd632ae410a9062dff5412bae65a", size = 98106, upload-time = "2025-05-02T08:32:30.281Z" }, - { url = "https://files.pythonhosted.org/packages/a8/05/5e33dbef7e2f773d672b6d79f10ec633d4a71cd96db6673625838a4fd532/charset_normalizer-3.4.2-cp311-cp311-win_amd64.whl", hash = "sha256:e53efc7c7cee4c1e70661e2e112ca46a575f90ed9ae3fef200f2a25e954f4b28", size = 105402, upload-time = "2025-05-02T08:32:32.191Z" }, - { url = "https://files.pythonhosted.org/packages/d7/a4/37f4d6035c89cac7930395a35cc0f1b872e652eaafb76a6075943754f095/charset_normalizer-3.4.2-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:0c29de6a1a95f24b9a1aa7aefd27d2487263f00dfd55a77719b530788f75cff7", size = 199936, upload-time = "2025-05-02T08:32:33.712Z" }, - { url = "https://files.pythonhosted.org/packages/ee/8a/1a5e33b73e0d9287274f899d967907cd0bf9c343e651755d9307e0dbf2b3/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cddf7bd982eaa998934a91f69d182aec997c6c468898efe6679af88283b498d3", size = 143790, upload-time = "2025-05-02T08:32:35.768Z" }, - { url = "https://files.pythonhosted.org/packages/66/52/59521f1d8e6ab1482164fa21409c5ef44da3e9f653c13ba71becdd98dec3/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fcbe676a55d7445b22c10967bceaaf0ee69407fbe0ece4d032b6eb8d4565982a", size = 153924, upload-time = "2025-05-02T08:32:37.284Z" }, - { url = "https://files.pythonhosted.org/packages/86/2d/fb55fdf41964ec782febbf33cb64be480a6b8f16ded2dbe8db27a405c09f/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d41c4d287cfc69060fa91cae9683eacffad989f1a10811995fa309df656ec214", size = 146626, upload-time = "2025-05-02T08:32:38.803Z" }, - { url = "https://files.pythonhosted.org/packages/8c/73/6ede2ec59bce19b3edf4209d70004253ec5f4e319f9a2e3f2f15601ed5f7/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4e594135de17ab3866138f496755f302b72157d115086d100c3f19370839dd3a", size = 148567, upload-time = "2025-05-02T08:32:40.251Z" }, - { url = "https://files.pythonhosted.org/packages/09/14/957d03c6dc343c04904530b6bef4e5efae5ec7d7990a7cbb868e4595ee30/charset_normalizer-3.4.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:cf713fe9a71ef6fd5adf7a79670135081cd4431c2943864757f0fa3a65b1fafd", size = 150957, upload-time = "2025-05-02T08:32:41.705Z" }, - { url = "https://files.pythonhosted.org/packages/0d/c8/8174d0e5c10ccebdcb1b53cc959591c4c722a3ad92461a273e86b9f5a302/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:a370b3e078e418187da8c3674eddb9d983ec09445c99a3a263c2011993522981", size = 145408, upload-time = "2025-05-02T08:32:43.709Z" }, - { url = "https://files.pythonhosted.org/packages/58/aa/8904b84bc8084ac19dc52feb4f5952c6df03ffb460a887b42615ee1382e8/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:a955b438e62efdf7e0b7b52a64dc5c3396e2634baa62471768a64bc2adb73d5c", size = 153399, upload-time = "2025-05-02T08:32:46.197Z" }, - { url = "https://files.pythonhosted.org/packages/c2/26/89ee1f0e264d201cb65cf054aca6038c03b1a0c6b4ae998070392a3ce605/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:7222ffd5e4de8e57e03ce2cef95a4c43c98fcb72ad86909abdfc2c17d227fc1b", size = 156815, upload-time = "2025-05-02T08:32:48.105Z" }, - { url = "https://files.pythonhosted.org/packages/fd/07/68e95b4b345bad3dbbd3a8681737b4338ff2c9df29856a6d6d23ac4c73cb/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:bee093bf902e1d8fc0ac143c88902c3dfc8941f7ea1d6a8dd2bcb786d33db03d", size = 154537, upload-time = "2025-05-02T08:32:49.719Z" }, - { url = "https://files.pythonhosted.org/packages/77/1a/5eefc0ce04affb98af07bc05f3bac9094513c0e23b0562d64af46a06aae4/charset_normalizer-3.4.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:dedb8adb91d11846ee08bec4c8236c8549ac721c245678282dcb06b221aab59f", size = 149565, upload-time = "2025-05-02T08:32:51.404Z" }, - { url = "https://files.pythonhosted.org/packages/37/a0/2410e5e6032a174c95e0806b1a6585eb21e12f445ebe239fac441995226a/charset_normalizer-3.4.2-cp312-cp312-win32.whl", hash = "sha256:db4c7bf0e07fc3b7d89ac2a5880a6a8062056801b83ff56d8464b70f65482b6c", size = 98357, upload-time = "2025-05-02T08:32:53.079Z" }, - { url = "https://files.pythonhosted.org/packages/6c/4f/c02d5c493967af3eda9c771ad4d2bbc8df6f99ddbeb37ceea6e8716a32bc/charset_normalizer-3.4.2-cp312-cp312-win_amd64.whl", hash = "sha256:5a9979887252a82fefd3d3ed2a8e3b937a7a809f65dcb1e068b090e165bbe99e", size = 105776, upload-time = "2025-05-02T08:32:54.573Z" }, - { url = "https://files.pythonhosted.org/packages/ea/12/a93df3366ed32db1d907d7593a94f1fe6293903e3e92967bebd6950ed12c/charset_normalizer-3.4.2-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:926ca93accd5d36ccdabd803392ddc3e03e6d4cd1cf17deff3b989ab8e9dbcf0", size = 199622, upload-time = "2025-05-02T08:32:56.363Z" }, - { url = "https://files.pythonhosted.org/packages/04/93/bf204e6f344c39d9937d3c13c8cd5bbfc266472e51fc8c07cb7f64fcd2de/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:eba9904b0f38a143592d9fc0e19e2df0fa2e41c3c3745554761c5f6447eedabf", size = 143435, upload-time = "2025-05-02T08:32:58.551Z" }, - { url = "https://files.pythonhosted.org/packages/22/2a/ea8a2095b0bafa6c5b5a55ffdc2f924455233ee7b91c69b7edfcc9e02284/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3fddb7e2c84ac87ac3a947cb4e66d143ca5863ef48e4a5ecb83bd48619e4634e", size = 153653, upload-time = "2025-05-02T08:33:00.342Z" }, - { url = "https://files.pythonhosted.org/packages/b6/57/1b090ff183d13cef485dfbe272e2fe57622a76694061353c59da52c9a659/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:98f862da73774290f251b9df8d11161b6cf25b599a66baf087c1ffe340e9bfd1", size = 146231, upload-time = "2025-05-02T08:33:02.081Z" }, - { url = "https://files.pythonhosted.org/packages/e2/28/ffc026b26f441fc67bd21ab7f03b313ab3fe46714a14b516f931abe1a2d8/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c9379d65defcab82d07b2a9dfbfc2e95bc8fe0ebb1b176a3190230a3ef0e07c", size = 148243, upload-time = "2025-05-02T08:33:04.063Z" }, - { url = "https://files.pythonhosted.org/packages/c0/0f/9abe9bd191629c33e69e47c6ef45ef99773320e9ad8e9cb08b8ab4a8d4cb/charset_normalizer-3.4.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e635b87f01ebc977342e2697d05b56632f5f879a4f15955dfe8cef2448b51691", size = 150442, upload-time = "2025-05-02T08:33:06.418Z" }, - { url = "https://files.pythonhosted.org/packages/67/7c/a123bbcedca91d5916c056407f89a7f5e8fdfce12ba825d7d6b9954a1a3c/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:1c95a1e2902a8b722868587c0e1184ad5c55631de5afc0eb96bc4b0d738092c0", size = 145147, upload-time = "2025-05-02T08:33:08.183Z" }, - { url = "https://files.pythonhosted.org/packages/ec/fe/1ac556fa4899d967b83e9893788e86b6af4d83e4726511eaaad035e36595/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:ef8de666d6179b009dce7bcb2ad4c4a779f113f12caf8dc77f0162c29d20490b", size = 153057, upload-time = "2025-05-02T08:33:09.986Z" }, - { url = "https://files.pythonhosted.org/packages/2b/ff/acfc0b0a70b19e3e54febdd5301a98b72fa07635e56f24f60502e954c461/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:32fc0341d72e0f73f80acb0a2c94216bd704f4f0bce10aedea38f30502b271ff", size = 156454, upload-time = "2025-05-02T08:33:11.814Z" }, - { url = "https://files.pythonhosted.org/packages/92/08/95b458ce9c740d0645feb0e96cea1f5ec946ea9c580a94adfe0b617f3573/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:289200a18fa698949d2b39c671c2cc7a24d44096784e76614899a7ccf2574b7b", size = 154174, upload-time = "2025-05-02T08:33:13.707Z" }, - { url = "https://files.pythonhosted.org/packages/78/be/8392efc43487ac051eee6c36d5fbd63032d78f7728cb37aebcc98191f1ff/charset_normalizer-3.4.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4a476b06fbcf359ad25d34a057b7219281286ae2477cc5ff5e3f70a246971148", size = 149166, upload-time = "2025-05-02T08:33:15.458Z" }, - { url = "https://files.pythonhosted.org/packages/44/96/392abd49b094d30b91d9fbda6a69519e95802250b777841cf3bda8fe136c/charset_normalizer-3.4.2-cp313-cp313-win32.whl", hash = "sha256:aaeeb6a479c7667fbe1099af9617c83aaca22182d6cf8c53966491a0f1b7ffb7", size = 98064, upload-time = "2025-05-02T08:33:17.06Z" }, - { url = "https://files.pythonhosted.org/packages/e9/b0/0200da600134e001d91851ddc797809e2fe0ea72de90e09bec5a2fbdaccb/charset_normalizer-3.4.2-cp313-cp313-win_amd64.whl", hash = "sha256:aa6af9e7d59f9c12b33ae4e9450619cf2488e2bbe9b44030905877f0b2324980", size = 105641, upload-time = "2025-05-02T08:33:18.753Z" }, - { url = "https://files.pythonhosted.org/packages/20/94/c5790835a017658cbfabd07f3bfb549140c3ac458cfc196323996b10095a/charset_normalizer-3.4.2-py3-none-any.whl", hash = "sha256:7f56930ab0abd1c45cd15be65cc741c28b1c9a34876ce8c17a2fa107810c0af0", size = 52626, upload-time = "2025-05-02T08:34:40.053Z" }, +version = "3.4.1" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/16/b0/572805e227f01586461c80e0fd25d65a2115599cc9dad142fee4b747c357/charset_normalizer-3.4.1.tar.gz", hash = "sha256:44251f18cd68a75b56585dd00dae26183e102cd5e0f9f1466e6df5da2ed64ea3", size = 123188, upload-time = "2024-12-24T18:12:35.43Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/0d/58/5580c1716040bc89206c77d8f74418caf82ce519aae06450393ca73475d1/charset_normalizer-3.4.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:91b36a978b5ae0ee86c394f5a54d6ef44db1de0815eb43de826d41d21e4af3de", size = 198013, upload-time = "2024-12-24T18:09:43.671Z" }, + { url = "https://files.pythonhosted.org/packages/d0/11/00341177ae71c6f5159a08168bcb98c6e6d196d372c94511f9f6c9afe0c6/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7461baadb4dc00fd9e0acbe254e3d7d2112e7f92ced2adc96e54ef6501c5f176", size = 141285, upload-time = "2024-12-24T18:09:48.113Z" }, + { url = "https://files.pythonhosted.org/packages/01/09/11d684ea5819e5a8f5100fb0b38cf8d02b514746607934134d31233e02c8/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e218488cd232553829be0664c2292d3af2eeeb94b32bea483cf79ac6a694e037", size = 151449, upload-time = "2024-12-24T18:09:50.845Z" }, + { url = "https://files.pythonhosted.org/packages/08/06/9f5a12939db324d905dc1f70591ae7d7898d030d7662f0d426e2286f68c9/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:80ed5e856eb7f30115aaf94e4a08114ccc8813e6ed1b5efa74f9f82e8509858f", size = 143892, upload-time = "2024-12-24T18:09:52.078Z" }, + { url = "https://files.pythonhosted.org/packages/93/62/5e89cdfe04584cb7f4d36003ffa2936681b03ecc0754f8e969c2becb7e24/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b010a7a4fd316c3c484d482922d13044979e78d1861f0e0650423144c616a46a", size = 146123, upload-time = "2024-12-24T18:09:54.575Z" }, + { url = "https://files.pythonhosted.org/packages/a9/ac/ab729a15c516da2ab70a05f8722ecfccc3f04ed7a18e45c75bbbaa347d61/charset_normalizer-3.4.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:4532bff1b8421fd0a320463030c7520f56a79c9024a4e88f01c537316019005a", size = 147943, upload-time = "2024-12-24T18:09:57.324Z" }, + { url = "https://files.pythonhosted.org/packages/03/d2/3f392f23f042615689456e9a274640c1d2e5dd1d52de36ab8f7955f8f050/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:d973f03c0cb71c5ed99037b870f2be986c3c05e63622c017ea9816881d2dd247", size = 142063, upload-time = "2024-12-24T18:09:59.794Z" }, + { url = "https://files.pythonhosted.org/packages/f2/e3/e20aae5e1039a2cd9b08d9205f52142329f887f8cf70da3650326670bddf/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:3a3bd0dcd373514dcec91c411ddb9632c0d7d92aed7093b8c3bbb6d69ca74408", size = 150578, upload-time = "2024-12-24T18:10:02.357Z" }, + { url = "https://files.pythonhosted.org/packages/8d/af/779ad72a4da0aed925e1139d458adc486e61076d7ecdcc09e610ea8678db/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:d9c3cdf5390dcd29aa8056d13e8e99526cda0305acc038b96b30352aff5ff2bb", size = 153629, upload-time = "2024-12-24T18:10:03.678Z" }, + { url = "https://files.pythonhosted.org/packages/c2/b6/7aa450b278e7aa92cf7732140bfd8be21f5f29d5bf334ae987c945276639/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:2bdfe3ac2e1bbe5b59a1a63721eb3b95fc9b6817ae4a46debbb4e11f6232428d", size = 150778, upload-time = "2024-12-24T18:10:06.197Z" }, + { url = "https://files.pythonhosted.org/packages/39/f4/d9f4f712d0951dcbfd42920d3db81b00dd23b6ab520419626f4023334056/charset_normalizer-3.4.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:eab677309cdb30d047996b36d34caeda1dc91149e4fdca0b1a039b3f79d9a807", size = 146453, upload-time = "2024-12-24T18:10:08.848Z" }, + { url = "https://files.pythonhosted.org/packages/49/2b/999d0314e4ee0cff3cb83e6bc9aeddd397eeed693edb4facb901eb8fbb69/charset_normalizer-3.4.1-cp310-cp310-win32.whl", hash = "sha256:c0429126cf75e16c4f0ad00ee0eae4242dc652290f940152ca8c75c3a4b6ee8f", size = 95479, upload-time = "2024-12-24T18:10:10.044Z" }, + { url = "https://files.pythonhosted.org/packages/2d/ce/3cbed41cff67e455a386fb5e5dd8906cdda2ed92fbc6297921f2e4419309/charset_normalizer-3.4.1-cp310-cp310-win_amd64.whl", hash = "sha256:9f0b8b1c6d84c8034a44893aba5e767bf9c7a211e313a9605d9c617d7083829f", size = 102790, upload-time = "2024-12-24T18:10:11.323Z" }, + { url = "https://files.pythonhosted.org/packages/72/80/41ef5d5a7935d2d3a773e3eaebf0a9350542f2cab4eac59a7a4741fbbbbe/charset_normalizer-3.4.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:8bfa33f4f2672964266e940dd22a195989ba31669bd84629f05fab3ef4e2d125", size = 194995, upload-time = "2024-12-24T18:10:12.838Z" }, + { url = "https://files.pythonhosted.org/packages/7a/28/0b9fefa7b8b080ec492110af6d88aa3dea91c464b17d53474b6e9ba5d2c5/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:28bf57629c75e810b6ae989f03c0828d64d6b26a5e205535585f96093e405ed1", size = 139471, upload-time = "2024-12-24T18:10:14.101Z" }, + { url = "https://files.pythonhosted.org/packages/71/64/d24ab1a997efb06402e3fc07317e94da358e2585165930d9d59ad45fcae2/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f08ff5e948271dc7e18a35641d2f11a4cd8dfd5634f55228b691e62b37125eb3", size = 149831, upload-time = "2024-12-24T18:10:15.512Z" }, + { url = "https://files.pythonhosted.org/packages/37/ed/be39e5258e198655240db5e19e0b11379163ad7070962d6b0c87ed2c4d39/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:234ac59ea147c59ee4da87a0c0f098e9c8d169f4dc2a159ef720f1a61bbe27cd", size = 142335, upload-time = "2024-12-24T18:10:18.369Z" }, + { url = "https://files.pythonhosted.org/packages/88/83/489e9504711fa05d8dde1574996408026bdbdbd938f23be67deebb5eca92/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd4ec41f914fa74ad1b8304bbc634b3de73d2a0889bd32076342a573e0779e00", size = 143862, upload-time = "2024-12-24T18:10:19.743Z" }, + { url = "https://files.pythonhosted.org/packages/c6/c7/32da20821cf387b759ad24627a9aca289d2822de929b8a41b6241767b461/charset_normalizer-3.4.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:eea6ee1db730b3483adf394ea72f808b6e18cf3cb6454b4d86e04fa8c4327a12", size = 145673, upload-time = "2024-12-24T18:10:21.139Z" }, + { url = "https://files.pythonhosted.org/packages/68/85/f4288e96039abdd5aeb5c546fa20a37b50da71b5cf01e75e87f16cd43304/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c96836c97b1238e9c9e3fe90844c947d5afbf4f4c92762679acfe19927d81d77", size = 140211, upload-time = "2024-12-24T18:10:22.382Z" }, + { url = "https://files.pythonhosted.org/packages/28/a3/a42e70d03cbdabc18997baf4f0227c73591a08041c149e710045c281f97b/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:4d86f7aff21ee58f26dcf5ae81a9addbd914115cdebcbb2217e4f0ed8982e146", size = 148039, upload-time = "2024-12-24T18:10:24.802Z" }, + { url = "https://files.pythonhosted.org/packages/85/e4/65699e8ab3014ecbe6f5c71d1a55d810fb716bbfd74f6283d5c2aa87febf/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:09b5e6733cbd160dcc09589227187e242a30a49ca5cefa5a7edd3f9d19ed53fd", size = 151939, upload-time = "2024-12-24T18:10:26.124Z" }, + { url = "https://files.pythonhosted.org/packages/b1/82/8e9fe624cc5374193de6860aba3ea8070f584c8565ee77c168ec13274bd2/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:5777ee0881f9499ed0f71cc82cf873d9a0ca8af166dfa0af8ec4e675b7df48e6", size = 149075, upload-time = "2024-12-24T18:10:30.027Z" }, + { url = "https://files.pythonhosted.org/packages/3d/7b/82865ba54c765560c8433f65e8acb9217cb839a9e32b42af4aa8e945870f/charset_normalizer-3.4.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:237bdbe6159cff53b4f24f397d43c6336c6b0b42affbe857970cefbb620911c8", size = 144340, upload-time = "2024-12-24T18:10:32.679Z" }, + { url = "https://files.pythonhosted.org/packages/b5/b6/9674a4b7d4d99a0d2df9b215da766ee682718f88055751e1e5e753c82db0/charset_normalizer-3.4.1-cp311-cp311-win32.whl", hash = "sha256:8417cb1f36cc0bc7eaba8ccb0e04d55f0ee52df06df3ad55259b9a323555fc8b", size = 95205, upload-time = "2024-12-24T18:10:34.724Z" }, + { url = "https://files.pythonhosted.org/packages/1e/ab/45b180e175de4402dcf7547e4fb617283bae54ce35c27930a6f35b6bef15/charset_normalizer-3.4.1-cp311-cp311-win_amd64.whl", hash = "sha256:d7f50a1f8c450f3925cb367d011448c39239bb3eb4117c36a6d354794de4ce76", size = 102441, upload-time = "2024-12-24T18:10:37.574Z" }, + { url = "https://files.pythonhosted.org/packages/0a/9a/dd1e1cdceb841925b7798369a09279bd1cf183cef0f9ddf15a3a6502ee45/charset_normalizer-3.4.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:73d94b58ec7fecbc7366247d3b0b10a21681004153238750bb67bd9012414545", size = 196105, upload-time = "2024-12-24T18:10:38.83Z" }, + { url = "https://files.pythonhosted.org/packages/d3/8c/90bfabf8c4809ecb648f39794cf2a84ff2e7d2a6cf159fe68d9a26160467/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dad3e487649f498dd991eeb901125411559b22e8d7ab25d3aeb1af367df5efd7", size = 140404, upload-time = "2024-12-24T18:10:44.272Z" }, + { url = "https://files.pythonhosted.org/packages/ad/8f/e410d57c721945ea3b4f1a04b74f70ce8fa800d393d72899f0a40526401f/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c30197aa96e8eed02200a83fba2657b4c3acd0f0aa4bdc9f6c1af8e8962e0757", size = 150423, upload-time = "2024-12-24T18:10:45.492Z" }, + { url = "https://files.pythonhosted.org/packages/f0/b8/e6825e25deb691ff98cf5c9072ee0605dc2acfca98af70c2d1b1bc75190d/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2369eea1ee4a7610a860d88f268eb39b95cb588acd7235e02fd5a5601773d4fa", size = 143184, upload-time = "2024-12-24T18:10:47.898Z" }, + { url = "https://files.pythonhosted.org/packages/3e/a2/513f6cbe752421f16d969e32f3583762bfd583848b763913ddab8d9bfd4f/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc2722592d8998c870fa4e290c2eec2c1569b87fe58618e67d38b4665dfa680d", size = 145268, upload-time = "2024-12-24T18:10:50.589Z" }, + { url = "https://files.pythonhosted.org/packages/74/94/8a5277664f27c3c438546f3eb53b33f5b19568eb7424736bdc440a88a31f/charset_normalizer-3.4.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ffc9202a29ab3920fa812879e95a9e78b2465fd10be7fcbd042899695d75e616", size = 147601, upload-time = "2024-12-24T18:10:52.541Z" }, + { url = "https://files.pythonhosted.org/packages/7c/5f/6d352c51ee763623a98e31194823518e09bfa48be2a7e8383cf691bbb3d0/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:804a4d582ba6e5b747c625bf1255e6b1507465494a40a2130978bda7b932c90b", size = 141098, upload-time = "2024-12-24T18:10:53.789Z" }, + { url = "https://files.pythonhosted.org/packages/78/d4/f5704cb629ba5ab16d1d3d741396aec6dc3ca2b67757c45b0599bb010478/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:0f55e69f030f7163dffe9fd0752b32f070566451afe180f99dbeeb81f511ad8d", size = 149520, upload-time = "2024-12-24T18:10:55.048Z" }, + { url = "https://files.pythonhosted.org/packages/c5/96/64120b1d02b81785f222b976c0fb79a35875457fa9bb40827678e54d1bc8/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:c4c3e6da02df6fa1410a7680bd3f63d4f710232d3139089536310d027950696a", size = 152852, upload-time = "2024-12-24T18:10:57.647Z" }, + { url = "https://files.pythonhosted.org/packages/84/c9/98e3732278a99f47d487fd3468bc60b882920cef29d1fa6ca460a1fdf4e6/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:5df196eb874dae23dcfb968c83d4f8fdccb333330fe1fc278ac5ceeb101003a9", size = 150488, upload-time = "2024-12-24T18:10:59.43Z" }, + { url = "https://files.pythonhosted.org/packages/13/0e/9c8d4cb99c98c1007cc11eda969ebfe837bbbd0acdb4736d228ccaabcd22/charset_normalizer-3.4.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e358e64305fe12299a08e08978f51fc21fac060dcfcddd95453eabe5b93ed0e1", size = 146192, upload-time = "2024-12-24T18:11:00.676Z" }, + { url = "https://files.pythonhosted.org/packages/b2/21/2b6b5b860781a0b49427309cb8670785aa543fb2178de875b87b9cc97746/charset_normalizer-3.4.1-cp312-cp312-win32.whl", hash = "sha256:9b23ca7ef998bc739bf6ffc077c2116917eabcc901f88da1b9856b210ef63f35", size = 95550, upload-time = "2024-12-24T18:11:01.952Z" }, + { url = "https://files.pythonhosted.org/packages/21/5b/1b390b03b1d16c7e382b561c5329f83cc06623916aab983e8ab9239c7d5c/charset_normalizer-3.4.1-cp312-cp312-win_amd64.whl", hash = "sha256:6ff8a4a60c227ad87030d76e99cd1698345d4491638dfa6673027c48b3cd395f", size = 102785, upload-time = "2024-12-24T18:11:03.142Z" }, + { url = "https://files.pythonhosted.org/packages/38/94/ce8e6f63d18049672c76d07d119304e1e2d7c6098f0841b51c666e9f44a0/charset_normalizer-3.4.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:aabfa34badd18f1da5ec1bc2715cadc8dca465868a4e73a0173466b688f29dda", size = 195698, upload-time = "2024-12-24T18:11:05.834Z" }, + { url = "https://files.pythonhosted.org/packages/24/2e/dfdd9770664aae179a96561cc6952ff08f9a8cd09a908f259a9dfa063568/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:22e14b5d70560b8dd51ec22863f370d1e595ac3d024cb8ad7d308b4cd95f8313", size = 140162, upload-time = "2024-12-24T18:11:07.064Z" }, + { url = "https://files.pythonhosted.org/packages/24/4e/f646b9093cff8fc86f2d60af2de4dc17c759de9d554f130b140ea4738ca6/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8436c508b408b82d87dc5f62496973a1805cd46727c34440b0d29d8a2f50a6c9", size = 150263, upload-time = "2024-12-24T18:11:08.374Z" }, + { url = "https://files.pythonhosted.org/packages/5e/67/2937f8d548c3ef6e2f9aab0f6e21001056f692d43282b165e7c56023e6dd/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2d074908e1aecee37a7635990b2c6d504cd4766c7bc9fc86d63f9c09af3fa11b", size = 142966, upload-time = "2024-12-24T18:11:09.831Z" }, + { url = "https://files.pythonhosted.org/packages/52/ed/b7f4f07de100bdb95c1756d3a4d17b90c1a3c53715c1a476f8738058e0fa/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:955f8851919303c92343d2f66165294848d57e9bba6cf6e3625485a70a038d11", size = 144992, upload-time = "2024-12-24T18:11:12.03Z" }, + { url = "https://files.pythonhosted.org/packages/96/2c/d49710a6dbcd3776265f4c923bb73ebe83933dfbaa841c5da850fe0fd20b/charset_normalizer-3.4.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:44ecbf16649486d4aebafeaa7ec4c9fed8b88101f4dd612dcaf65d5e815f837f", size = 147162, upload-time = "2024-12-24T18:11:13.372Z" }, + { url = "https://files.pythonhosted.org/packages/b4/41/35ff1f9a6bd380303dea55e44c4933b4cc3c4850988927d4082ada230273/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:0924e81d3d5e70f8126529951dac65c1010cdf117bb75eb02dd12339b57749dd", size = 140972, upload-time = "2024-12-24T18:11:14.628Z" }, + { url = "https://files.pythonhosted.org/packages/fb/43/c6a0b685fe6910d08ba971f62cd9c3e862a85770395ba5d9cad4fede33ab/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:2967f74ad52c3b98de4c3b32e1a44e32975e008a9cd2a8cc8966d6a5218c5cb2", size = 149095, upload-time = "2024-12-24T18:11:17.672Z" }, + { url = "https://files.pythonhosted.org/packages/4c/ff/a9a504662452e2d2878512115638966e75633519ec11f25fca3d2049a94a/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:c75cb2a3e389853835e84a2d8fb2b81a10645b503eca9bcb98df6b5a43eb8886", size = 152668, upload-time = "2024-12-24T18:11:18.989Z" }, + { url = "https://files.pythonhosted.org/packages/6c/71/189996b6d9a4b932564701628af5cee6716733e9165af1d5e1b285c530ed/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:09b26ae6b1abf0d27570633b2b078a2a20419c99d66fb2823173d73f188ce601", size = 150073, upload-time = "2024-12-24T18:11:21.507Z" }, + { url = "https://files.pythonhosted.org/packages/e4/93/946a86ce20790e11312c87c75ba68d5f6ad2208cfb52b2d6a2c32840d922/charset_normalizer-3.4.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:fa88b843d6e211393a37219e6a1c1df99d35e8fd90446f1118f4216e307e48cd", size = 145732, upload-time = "2024-12-24T18:11:22.774Z" }, + { url = "https://files.pythonhosted.org/packages/cd/e5/131d2fb1b0dddafc37be4f3a2fa79aa4c037368be9423061dccadfd90091/charset_normalizer-3.4.1-cp313-cp313-win32.whl", hash = "sha256:eb8178fe3dba6450a3e024e95ac49ed3400e506fd4e9e5c32d30adda88cbd407", size = 95391, upload-time = "2024-12-24T18:11:24.139Z" }, + { url = "https://files.pythonhosted.org/packages/27/f2/4f9a69cc7712b9b5ad8fdb87039fd89abba997ad5cbe690d1835d40405b0/charset_normalizer-3.4.1-cp313-cp313-win_amd64.whl", hash = "sha256:b1ac5992a838106edb89654e0aebfc24f5848ae2547d22c2c3f66454daa11971", size = 102702, upload-time = "2024-12-24T18:11:26.535Z" }, + { url = "https://files.pythonhosted.org/packages/0e/f6/65ecc6878a89bb1c23a086ea335ad4bf21a588990c3f535a227b9eea9108/charset_normalizer-3.4.1-py3-none-any.whl", hash = "sha256:d98b1668f06378c6dbefec3b92299716b931cd4e6061f3c875a71ced1780ab85", size = 49767, upload-time = "2024-12-24T18:12:32.852Z" }, ] [[package]] name = "click" -version = "8.2.1" +version = "8.1.8" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "colorama", marker = "sys_platform == 'win32'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/60/6c/8ca2efa64cf75a977a0d7fac081354553ebe483345c734fb6b6515d96bbc/click-8.2.1.tar.gz", hash = "sha256:27c491cc05d968d271d5a1db13e3b5a184636d9d930f148c50b038f0d0646202", size = 286342, upload-time = "2025-05-20T23:19:49.832Z" } +sdist = { url = "https://files.pythonhosted.org/packages/b9/2e/0090cbf739cee7d23781ad4b89a9894a41538e4fcf4c31dcdd705b78eb8b/click-8.1.8.tar.gz", hash = "sha256:ed53c9d8990d83c2a27deae68e4ee337473f6330c040a31d4225c9574d16096a", size = 226593, upload-time = "2024-12-21T18:38:44.339Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/85/32/10bb5764d90a8eee674e9dc6f4db6a0ab47c8c4d0d83c27f7c39ac415a4d/click-8.2.1-py3-none-any.whl", hash = "sha256:61a3265b914e850b85317d0b3109c7f8cd35a670f963866005d6ef1d5175a12b", size = 102215, upload-time = "2025-05-20T23:19:47.796Z" }, + { url = "https://files.pythonhosted.org/packages/7e/d4/7ebdbd03970677812aac39c869717059dbb71a4cfc033ca6e5221787892c/click-8.1.8-py3-none-any.whl", hash = "sha256:63c132bbbed01578a06712a2d1f497bb62d9c1c0d329b7903a866228027263b2", size = 98188, upload-time = "2024-12-21T18:38:41.666Z" }, ] [[package]] @@ -467,172 +460,78 @@ wheels = [ [[package]] name = "comm" -version = "0.2.3" +version = "0.2.2" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/4c/13/7d740c5849255756bc17888787313b61fd38a0a8304fc4f073dfc46122aa/comm-0.2.3.tar.gz", hash = "sha256:2dc8048c10962d55d7ad693be1e7045d891b7ce8d999c97963a5e3e99c055971", size = 6319, upload-time = "2025-07-25T14:02:04.452Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/60/97/891a0971e1e4a8c5d2b20bbe0e524dc04548d2307fee33cdeba148fd4fc7/comm-0.2.3-py3-none-any.whl", hash = "sha256:c615d91d75f7f04f095b30d1c1711babd43bdc6419c1be9886a85f2f4e489417", size = 7294, upload-time = "2025-07-25T14:02:02.896Z" }, -] - -[[package]] -name = "contourpy" -version = "1.3.2" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version < '3.11' and sys_platform == 'linux'", - "python_full_version < '3.11' and sys_platform != 'linux'", -] dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/66/54/eb9bfc647b19f2009dd5c7f5ec51c4e6ca831725f1aea7a993034f483147/contourpy-1.3.2.tar.gz", hash = "sha256:b6945942715a034c671b7fc54f9588126b0b8bf23db2696e3ca8328f3ff0ab54", size = 13466130, upload-time = "2025-04-15T17:47:53.79Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/12/a3/da4153ec8fe25d263aa48c1a4cbde7f49b59af86f0b6f7862788c60da737/contourpy-1.3.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:ba38e3f9f330af820c4b27ceb4b9c7feee5fe0493ea53a8720f4792667465934", size = 268551, upload-time = "2025-04-15T17:34:46.581Z" }, - { url = "https://files.pythonhosted.org/packages/2f/6c/330de89ae1087eb622bfca0177d32a7ece50c3ef07b28002de4757d9d875/contourpy-1.3.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:dc41ba0714aa2968d1f8674ec97504a8f7e334f48eeacebcaa6256213acb0989", size = 253399, upload-time = "2025-04-15T17:34:51.427Z" }, - { url = "https://files.pythonhosted.org/packages/c1/bd/20c6726b1b7f81a8bee5271bed5c165f0a8e1f572578a9d27e2ccb763cb2/contourpy-1.3.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9be002b31c558d1ddf1b9b415b162c603405414bacd6932d031c5b5a8b757f0d", size = 312061, upload-time = "2025-04-15T17:34:55.961Z" }, - { url = "https://files.pythonhosted.org/packages/22/fc/a9665c88f8a2473f823cf1ec601de9e5375050f1958cbb356cdf06ef1ab6/contourpy-1.3.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8d2e74acbcba3bfdb6d9d8384cdc4f9260cae86ed9beee8bd5f54fee49a430b9", size = 351956, upload-time = "2025-04-15T17:35:00.992Z" }, - { url = "https://files.pythonhosted.org/packages/25/eb/9f0a0238f305ad8fb7ef42481020d6e20cf15e46be99a1fcf939546a177e/contourpy-1.3.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e259bced5549ac64410162adc973c5e2fb77f04df4a439d00b478e57a0e65512", size = 320872, upload-time = "2025-04-15T17:35:06.177Z" }, - { url = "https://files.pythonhosted.org/packages/32/5c/1ee32d1c7956923202f00cf8d2a14a62ed7517bdc0ee1e55301227fc273c/contourpy-1.3.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ad687a04bc802cbe8b9c399c07162a3c35e227e2daccf1668eb1f278cb698631", size = 325027, upload-time = "2025-04-15T17:35:11.244Z" }, - { url = "https://files.pythonhosted.org/packages/83/bf/9baed89785ba743ef329c2b07fd0611d12bfecbedbdd3eeecf929d8d3b52/contourpy-1.3.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:cdd22595308f53ef2f891040ab2b93d79192513ffccbd7fe19be7aa773a5e09f", size = 1306641, upload-time = "2025-04-15T17:35:26.701Z" }, - { url = "https://files.pythonhosted.org/packages/d4/cc/74e5e83d1e35de2d28bd97033426b450bc4fd96e092a1f7a63dc7369b55d/contourpy-1.3.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:b4f54d6a2defe9f257327b0f243612dd051cc43825587520b1bf74a31e2f6ef2", size = 1374075, upload-time = "2025-04-15T17:35:43.204Z" }, - { url = "https://files.pythonhosted.org/packages/0c/42/17f3b798fd5e033b46a16f8d9fcb39f1aba051307f5ebf441bad1ecf78f8/contourpy-1.3.2-cp310-cp310-win32.whl", hash = "sha256:f939a054192ddc596e031e50bb13b657ce318cf13d264f095ce9db7dc6ae81c0", size = 177534, upload-time = "2025-04-15T17:35:46.554Z" }, - { url = "https://files.pythonhosted.org/packages/54/ec/5162b8582f2c994721018d0c9ece9dc6ff769d298a8ac6b6a652c307e7df/contourpy-1.3.2-cp310-cp310-win_amd64.whl", hash = "sha256:c440093bbc8fc21c637c03bafcbef95ccd963bc6e0514ad887932c18ca2a759a", size = 221188, upload-time = "2025-04-15T17:35:50.064Z" }, - { url = "https://files.pythonhosted.org/packages/b3/b9/ede788a0b56fc5b071639d06c33cb893f68b1178938f3425debebe2dab78/contourpy-1.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:6a37a2fb93d4df3fc4c0e363ea4d16f83195fc09c891bc8ce072b9d084853445", size = 269636, upload-time = "2025-04-15T17:35:54.473Z" }, - { url = "https://files.pythonhosted.org/packages/e6/75/3469f011d64b8bbfa04f709bfc23e1dd71be54d05b1b083be9f5b22750d1/contourpy-1.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:b7cd50c38f500bbcc9b6a46643a40e0913673f869315d8e70de0438817cb7773", size = 254636, upload-time = "2025-04-15T17:35:58.283Z" }, - { url = "https://files.pythonhosted.org/packages/8d/2f/95adb8dae08ce0ebca4fd8e7ad653159565d9739128b2d5977806656fcd2/contourpy-1.3.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d6658ccc7251a4433eebd89ed2672c2ed96fba367fd25ca9512aa92a4b46c4f1", size = 313053, upload-time = "2025-04-15T17:36:03.235Z" }, - { url = "https://files.pythonhosted.org/packages/c3/a6/8ccf97a50f31adfa36917707fe39c9a0cbc24b3bbb58185577f119736cc9/contourpy-1.3.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:70771a461aaeb335df14deb6c97439973d253ae70660ca085eec25241137ef43", size = 352985, upload-time = "2025-04-15T17:36:08.275Z" }, - { url = "https://files.pythonhosted.org/packages/1d/b6/7925ab9b77386143f39d9c3243fdd101621b4532eb126743201160ffa7e6/contourpy-1.3.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:65a887a6e8c4cd0897507d814b14c54a8c2e2aa4ac9f7686292f9769fcf9a6ab", size = 323750, upload-time = "2025-04-15T17:36:13.29Z" }, - { url = "https://files.pythonhosted.org/packages/c2/f3/20c5d1ef4f4748e52d60771b8560cf00b69d5c6368b5c2e9311bcfa2a08b/contourpy-1.3.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3859783aefa2b8355697f16642695a5b9792e7a46ab86da1118a4a23a51a33d7", size = 326246, upload-time = "2025-04-15T17:36:18.329Z" }, - { url = "https://files.pythonhosted.org/packages/8c/e5/9dae809e7e0b2d9d70c52b3d24cba134dd3dad979eb3e5e71f5df22ed1f5/contourpy-1.3.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:eab0f6db315fa4d70f1d8ab514e527f0366ec021ff853d7ed6a2d33605cf4b83", size = 1308728, upload-time = "2025-04-15T17:36:33.878Z" }, - { url = "https://files.pythonhosted.org/packages/e2/4a/0058ba34aeea35c0b442ae61a4f4d4ca84d6df8f91309bc2d43bb8dd248f/contourpy-1.3.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d91a3ccc7fea94ca0acab82ceb77f396d50a1f67412efe4c526f5d20264e6ecd", size = 1375762, upload-time = "2025-04-15T17:36:51.295Z" }, - { url = "https://files.pythonhosted.org/packages/09/33/7174bdfc8b7767ef2c08ed81244762d93d5c579336fc0b51ca57b33d1b80/contourpy-1.3.2-cp311-cp311-win32.whl", hash = "sha256:1c48188778d4d2f3d48e4643fb15d8608b1d01e4b4d6b0548d9b336c28fc9b6f", size = 178196, upload-time = "2025-04-15T17:36:55.002Z" }, - { url = "https://files.pythonhosted.org/packages/5e/fe/4029038b4e1c4485cef18e480b0e2cd2d755448bb071eb9977caac80b77b/contourpy-1.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:5ebac872ba09cb8f2131c46b8739a7ff71de28a24c869bcad554477eb089a878", size = 222017, upload-time = "2025-04-15T17:36:58.576Z" }, - { url = "https://files.pythonhosted.org/packages/34/f7/44785876384eff370c251d58fd65f6ad7f39adce4a093c934d4a67a7c6b6/contourpy-1.3.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:4caf2bcd2969402bf77edc4cb6034c7dd7c0803213b3523f111eb7460a51b8d2", size = 271580, upload-time = "2025-04-15T17:37:03.105Z" }, - { url = "https://files.pythonhosted.org/packages/93/3b/0004767622a9826ea3d95f0e9d98cd8729015768075d61f9fea8eeca42a8/contourpy-1.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:82199cb78276249796419fe36b7386bd8d2cc3f28b3bc19fe2454fe2e26c4c15", size = 255530, upload-time = "2025-04-15T17:37:07.026Z" }, - { url = "https://files.pythonhosted.org/packages/e7/bb/7bd49e1f4fa805772d9fd130e0d375554ebc771ed7172f48dfcd4ca61549/contourpy-1.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:106fab697af11456fcba3e352ad50effe493a90f893fca6c2ca5c033820cea92", size = 307688, upload-time = "2025-04-15T17:37:11.481Z" }, - { url = "https://files.pythonhosted.org/packages/fc/97/e1d5dbbfa170725ef78357a9a0edc996b09ae4af170927ba8ce977e60a5f/contourpy-1.3.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d14f12932a8d620e307f715857107b1d1845cc44fdb5da2bc8e850f5ceba9f87", size = 347331, upload-time = "2025-04-15T17:37:18.212Z" }, - { url = "https://files.pythonhosted.org/packages/6f/66/e69e6e904f5ecf6901be3dd16e7e54d41b6ec6ae3405a535286d4418ffb4/contourpy-1.3.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:532fd26e715560721bb0d5fc7610fce279b3699b018600ab999d1be895b09415", size = 318963, upload-time = "2025-04-15T17:37:22.76Z" }, - { url = "https://files.pythonhosted.org/packages/a8/32/b8a1c8965e4f72482ff2d1ac2cd670ce0b542f203c8e1d34e7c3e6925da7/contourpy-1.3.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f26b383144cf2d2c29f01a1e8170f50dacf0eac02d64139dcd709a8ac4eb3cfe", size = 323681, upload-time = "2025-04-15T17:37:33.001Z" }, - { url = "https://files.pythonhosted.org/packages/30/c6/12a7e6811d08757c7162a541ca4c5c6a34c0f4e98ef2b338791093518e40/contourpy-1.3.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:c49f73e61f1f774650a55d221803b101d966ca0c5a2d6d5e4320ec3997489441", size = 1308674, upload-time = "2025-04-15T17:37:48.64Z" }, - { url = "https://files.pythonhosted.org/packages/2a/8a/bebe5a3f68b484d3a2b8ffaf84704b3e343ef1addea528132ef148e22b3b/contourpy-1.3.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:3d80b2c0300583228ac98d0a927a1ba6a2ba6b8a742463c564f1d419ee5b211e", size = 1380480, upload-time = "2025-04-15T17:38:06.7Z" }, - { url = "https://files.pythonhosted.org/packages/34/db/fcd325f19b5978fb509a7d55e06d99f5f856294c1991097534360b307cf1/contourpy-1.3.2-cp312-cp312-win32.whl", hash = "sha256:90df94c89a91b7362e1142cbee7568f86514412ab8a2c0d0fca72d7e91b62912", size = 178489, upload-time = "2025-04-15T17:38:10.338Z" }, - { url = "https://files.pythonhosted.org/packages/01/c8/fadd0b92ffa7b5eb5949bf340a63a4a496a6930a6c37a7ba0f12acb076d6/contourpy-1.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:8c942a01d9163e2e5cfb05cb66110121b8d07ad438a17f9e766317bcb62abf73", size = 223042, upload-time = "2025-04-15T17:38:14.239Z" }, - { url = "https://files.pythonhosted.org/packages/2e/61/5673f7e364b31e4e7ef6f61a4b5121c5f170f941895912f773d95270f3a2/contourpy-1.3.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:de39db2604ae755316cb5967728f4bea92685884b1e767b7c24e983ef5f771cb", size = 271630, upload-time = "2025-04-15T17:38:19.142Z" }, - { url = "https://files.pythonhosted.org/packages/ff/66/a40badddd1223822c95798c55292844b7e871e50f6bfd9f158cb25e0bd39/contourpy-1.3.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3f9e896f447c5c8618f1edb2bafa9a4030f22a575ec418ad70611450720b5b08", size = 255670, upload-time = "2025-04-15T17:38:23.688Z" }, - { url = "https://files.pythonhosted.org/packages/1e/c7/cf9fdee8200805c9bc3b148f49cb9482a4e3ea2719e772602a425c9b09f8/contourpy-1.3.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:71e2bd4a1c4188f5c2b8d274da78faab884b59df20df63c34f74aa1813c4427c", size = 306694, upload-time = "2025-04-15T17:38:28.238Z" }, - { url = "https://files.pythonhosted.org/packages/dd/e7/ccb9bec80e1ba121efbffad7f38021021cda5be87532ec16fd96533bb2e0/contourpy-1.3.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:de425af81b6cea33101ae95ece1f696af39446db9682a0b56daaa48cfc29f38f", size = 345986, upload-time = "2025-04-15T17:38:33.502Z" }, - { url = "https://files.pythonhosted.org/packages/dc/49/ca13bb2da90391fa4219fdb23b078d6065ada886658ac7818e5441448b78/contourpy-1.3.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:977e98a0e0480d3fe292246417239d2d45435904afd6d7332d8455981c408b85", size = 318060, upload-time = "2025-04-15T17:38:38.672Z" }, - { url = "https://files.pythonhosted.org/packages/c8/65/5245ce8c548a8422236c13ffcdcdada6a2a812c361e9e0c70548bb40b661/contourpy-1.3.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:434f0adf84911c924519d2b08fc10491dd282b20bdd3fa8f60fd816ea0b48841", size = 322747, upload-time = "2025-04-15T17:38:43.712Z" }, - { url = "https://files.pythonhosted.org/packages/72/30/669b8eb48e0a01c660ead3752a25b44fdb2e5ebc13a55782f639170772f9/contourpy-1.3.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c66c4906cdbc50e9cba65978823e6e00b45682eb09adbb78c9775b74eb222422", size = 1308895, upload-time = "2025-04-15T17:39:00.224Z" }, - { url = "https://files.pythonhosted.org/packages/05/5a/b569f4250decee6e8d54498be7bdf29021a4c256e77fe8138c8319ef8eb3/contourpy-1.3.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8b7fc0cd78ba2f4695fd0a6ad81a19e7e3ab825c31b577f384aa9d7817dc3bef", size = 1379098, upload-time = "2025-04-15T17:43:29.649Z" }, - { url = "https://files.pythonhosted.org/packages/19/ba/b227c3886d120e60e41b28740ac3617b2f2b971b9f601c835661194579f1/contourpy-1.3.2-cp313-cp313-win32.whl", hash = "sha256:15ce6ab60957ca74cff444fe66d9045c1fd3e92c8936894ebd1f3eef2fff075f", size = 178535, upload-time = "2025-04-15T17:44:44.532Z" }, - { url = "https://files.pythonhosted.org/packages/12/6e/2fed56cd47ca739b43e892707ae9a13790a486a3173be063681ca67d2262/contourpy-1.3.2-cp313-cp313-win_amd64.whl", hash = "sha256:e1578f7eafce927b168752ed7e22646dad6cd9bca673c60bff55889fa236ebf9", size = 223096, upload-time = "2025-04-15T17:44:48.194Z" }, - { url = "https://files.pythonhosted.org/packages/54/4c/e76fe2a03014a7c767d79ea35c86a747e9325537a8b7627e0e5b3ba266b4/contourpy-1.3.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:0475b1f6604896bc7c53bb070e355e9321e1bc0d381735421a2d2068ec56531f", size = 285090, upload-time = "2025-04-15T17:43:34.084Z" }, - { url = "https://files.pythonhosted.org/packages/7b/e2/5aba47debd55d668e00baf9651b721e7733975dc9fc27264a62b0dd26eb8/contourpy-1.3.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:c85bb486e9be652314bb5b9e2e3b0d1b2e643d5eec4992c0fbe8ac71775da739", size = 268643, upload-time = "2025-04-15T17:43:38.626Z" }, - { url = "https://files.pythonhosted.org/packages/a1/37/cd45f1f051fe6230f751cc5cdd2728bb3a203f5619510ef11e732109593c/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:745b57db7758f3ffc05a10254edd3182a2a83402a89c00957a8e8a22f5582823", size = 310443, upload-time = "2025-04-15T17:43:44.522Z" }, - { url = "https://files.pythonhosted.org/packages/8b/a2/36ea6140c306c9ff6dd38e3bcec80b3b018474ef4d17eb68ceecd26675f4/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:970e9173dbd7eba9b4e01aab19215a48ee5dd3f43cef736eebde064a171f89a5", size = 349865, upload-time = "2025-04-15T17:43:49.545Z" }, - { url = "https://files.pythonhosted.org/packages/95/b7/2fc76bc539693180488f7b6cc518da7acbbb9e3b931fd9280504128bf956/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c6c4639a9c22230276b7bffb6a850dfc8258a2521305e1faefe804d006b2e532", size = 321162, upload-time = "2025-04-15T17:43:54.203Z" }, - { url = "https://files.pythonhosted.org/packages/f4/10/76d4f778458b0aa83f96e59d65ece72a060bacb20cfbee46cf6cd5ceba41/contourpy-1.3.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cc829960f34ba36aad4302e78eabf3ef16a3a100863f0d4eeddf30e8a485a03b", size = 327355, upload-time = "2025-04-15T17:44:01.025Z" }, - { url = "https://files.pythonhosted.org/packages/43/a3/10cf483ea683f9f8ab096c24bad3cce20e0d1dd9a4baa0e2093c1c962d9d/contourpy-1.3.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:d32530b534e986374fc19eaa77fcb87e8a99e5431499949b828312bdcd20ac52", size = 1307935, upload-time = "2025-04-15T17:44:17.322Z" }, - { url = "https://files.pythonhosted.org/packages/78/73/69dd9a024444489e22d86108e7b913f3528f56cfc312b5c5727a44188471/contourpy-1.3.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:e298e7e70cf4eb179cc1077be1c725b5fd131ebc81181bf0c03525c8abc297fd", size = 1372168, upload-time = "2025-04-15T17:44:33.43Z" }, - { url = "https://files.pythonhosted.org/packages/0f/1b/96d586ccf1b1a9d2004dd519b25fbf104a11589abfd05484ff12199cca21/contourpy-1.3.2-cp313-cp313t-win32.whl", hash = "sha256:d0e589ae0d55204991450bb5c23f571c64fe43adaa53f93fc902a84c96f52fe1", size = 189550, upload-time = "2025-04-15T17:44:37.092Z" }, - { url = "https://files.pythonhosted.org/packages/b0/e6/6000d0094e8a5e32ad62591c8609e269febb6e4db83a1c75ff8868b42731/contourpy-1.3.2-cp313-cp313t-win_amd64.whl", hash = "sha256:78e9253c3de756b3f6a5174d024c4835acd59eb3f8e2ca13e775dbffe1558f69", size = 238214, upload-time = "2025-04-15T17:44:40.827Z" }, - { url = "https://files.pythonhosted.org/packages/33/05/b26e3c6ecc05f349ee0013f0bb850a761016d89cec528a98193a48c34033/contourpy-1.3.2-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:fd93cc7f3139b6dd7aab2f26a90dde0aa9fc264dbf70f6740d498a70b860b82c", size = 265681, upload-time = "2025-04-15T17:44:59.314Z" }, - { url = "https://files.pythonhosted.org/packages/2b/25/ac07d6ad12affa7d1ffed11b77417d0a6308170f44ff20fa1d5aa6333f03/contourpy-1.3.2-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:107ba8a6a7eec58bb475329e6d3b95deba9440667c4d62b9b6063942b61d7f16", size = 315101, upload-time = "2025-04-15T17:45:04.165Z" }, - { url = "https://files.pythonhosted.org/packages/8f/4d/5bb3192bbe9d3f27e3061a6a8e7733c9120e203cb8515767d30973f71030/contourpy-1.3.2-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:ded1706ed0c1049224531b81128efbd5084598f18d8a2d9efae833edbd2b40ad", size = 220599, upload-time = "2025-04-15T17:45:08.456Z" }, - { url = "https://files.pythonhosted.org/packages/ff/c0/91f1215d0d9f9f343e4773ba6c9b89e8c0cc7a64a6263f21139da639d848/contourpy-1.3.2-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:5f5964cdad279256c084b69c3f412b7801e15356b16efa9d78aa974041903da0", size = 266807, upload-time = "2025-04-15T17:45:15.535Z" }, - { url = "https://files.pythonhosted.org/packages/d4/79/6be7e90c955c0487e7712660d6cead01fa17bff98e0ea275737cc2bc8e71/contourpy-1.3.2-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:49b65a95d642d4efa8f64ba12558fcb83407e58a2dfba9d796d77b63ccfcaff5", size = 318729, upload-time = "2025-04-15T17:45:20.166Z" }, - { url = "https://files.pythonhosted.org/packages/87/68/7f46fb537958e87427d98a4074bcde4b67a70b04900cfc5ce29bc2f556c1/contourpy-1.3.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:8c5acb8dddb0752bf252e01a3035b21443158910ac16a3b0d20e7fed7d534ce5", size = 221791, upload-time = "2025-04-15T17:45:24.794Z" }, + { name = "traitlets" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/e9/a8/fb783cb0abe2b5fded9f55e5703015cdf1c9c85b3669087c538dd15a6a86/comm-0.2.2.tar.gz", hash = "sha256:3fd7a84065306e07bea1773df6eb8282de51ba82f77c72f9c85716ab11fe980e", size = 6210, upload-time = "2024-03-12T16:53:41.133Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/e6/75/49e5bfe642f71f272236b5b2d2691cf915a7283cc0ceda56357b61daa538/comm-0.2.2-py3-none-any.whl", hash = "sha256:e6fb86cb70ff661ee8c9c14e7d36d6de3b4066f1441be4063df9c5009f0a64d3", size = 7180, upload-time = "2024-03-12T16:53:39.226Z" }, ] [[package]] name = "contourpy" -version = "1.3.3" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.13' and sys_platform == 'linux'", - "python_full_version == '3.12.*' and sys_platform == 'linux'", - "python_full_version >= '3.13' and sys_platform != 'linux'", - "python_full_version == '3.12.*' and sys_platform != 'linux'", - "python_full_version == '3.11.*' and sys_platform == 'linux'", - "python_full_version == '3.11.*' and sys_platform != 'linux'", -] -dependencies = [ - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/58/01/1253e6698a07380cd31a736d248a3f2a50a7c88779a1813da27503cadc2a/contourpy-1.3.3.tar.gz", hash = "sha256:083e12155b210502d0bca491432bb04d56dc3432f95a979b429f2848c3dbe880", size = 13466174, upload-time = "2025-07-26T12:03:12.549Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/91/2e/c4390a31919d8a78b90e8ecf87cd4b4c4f05a5b48d05ec17db8e5404c6f4/contourpy-1.3.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:709a48ef9a690e1343202916450bc48b9e51c049b089c7f79a267b46cffcdaa1", size = 288773, upload-time = "2025-07-26T12:01:02.277Z" }, - { url = "https://files.pythonhosted.org/packages/0d/44/c4b0b6095fef4dc9c420e041799591e3b63e9619e3044f7f4f6c21c0ab24/contourpy-1.3.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:23416f38bfd74d5d28ab8429cc4d63fa67d5068bd711a85edb1c3fb0c3e2f381", size = 270149, upload-time = "2025-07-26T12:01:04.072Z" }, - { url = "https://files.pythonhosted.org/packages/30/2e/dd4ced42fefac8470661d7cb7e264808425e6c5d56d175291e93890cce09/contourpy-1.3.3-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:929ddf8c4c7f348e4c0a5a3a714b5c8542ffaa8c22954862a46ca1813b667ee7", size = 329222, upload-time = "2025-07-26T12:01:05.688Z" }, - { url = "https://files.pythonhosted.org/packages/f2/74/cc6ec2548e3d276c71389ea4802a774b7aa3558223b7bade3f25787fafc2/contourpy-1.3.3-cp311-cp311-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:9e999574eddae35f1312c2b4b717b7885d4edd6cb46700e04f7f02db454e67c1", size = 377234, upload-time = "2025-07-26T12:01:07.054Z" }, - { url = "https://files.pythonhosted.org/packages/03/b3/64ef723029f917410f75c09da54254c5f9ea90ef89b143ccadb09df14c15/contourpy-1.3.3-cp311-cp311-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:0bf67e0e3f482cb69779dd3061b534eb35ac9b17f163d851e2a547d56dba0a3a", size = 380555, upload-time = "2025-07-26T12:01:08.801Z" }, - { url = "https://files.pythonhosted.org/packages/5f/4b/6157f24ca425b89fe2eb7e7be642375711ab671135be21e6faa100f7448c/contourpy-1.3.3-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:51e79c1f7470158e838808d4a996fa9bac72c498e93d8ebe5119bc1e6becb0db", size = 355238, upload-time = "2025-07-26T12:01:10.319Z" }, - { url = "https://files.pythonhosted.org/packages/98/56/f914f0dd678480708a04cfd2206e7c382533249bc5001eb9f58aa693e200/contourpy-1.3.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:598c3aaece21c503615fd59c92a3598b428b2f01bfb4b8ca9c4edeecc2438620", size = 1326218, upload-time = "2025-07-26T12:01:12.659Z" }, - { url = "https://files.pythonhosted.org/packages/fb/d7/4a972334a0c971acd5172389671113ae82aa7527073980c38d5868ff1161/contourpy-1.3.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:322ab1c99b008dad206d406bb61d014cf0174df491ae9d9d0fac6a6fda4f977f", size = 1392867, upload-time = "2025-07-26T12:01:15.533Z" }, - { url = "https://files.pythonhosted.org/packages/75/3e/f2cc6cd56dc8cff46b1a56232eabc6feea52720083ea71ab15523daab796/contourpy-1.3.3-cp311-cp311-win32.whl", hash = "sha256:fd907ae12cd483cd83e414b12941c632a969171bf90fc937d0c9f268a31cafff", size = 183677, upload-time = "2025-07-26T12:01:17.088Z" }, - { url = "https://files.pythonhosted.org/packages/98/4b/9bd370b004b5c9d8045c6c33cf65bae018b27aca550a3f657cdc99acdbd8/contourpy-1.3.3-cp311-cp311-win_amd64.whl", hash = "sha256:3519428f6be58431c56581f1694ba8e50626f2dd550af225f82fb5f5814d2a42", size = 225234, upload-time = "2025-07-26T12:01:18.256Z" }, - { url = "https://files.pythonhosted.org/packages/d9/b6/71771e02c2e004450c12b1120a5f488cad2e4d5b590b1af8bad060360fe4/contourpy-1.3.3-cp311-cp311-win_arm64.whl", hash = "sha256:15ff10bfada4bf92ec8b31c62bf7c1834c244019b4a33095a68000d7075df470", size = 193123, upload-time = "2025-07-26T12:01:19.848Z" }, - { url = "https://files.pythonhosted.org/packages/be/45/adfee365d9ea3d853550b2e735f9d66366701c65db7855cd07621732ccfc/contourpy-1.3.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b08a32ea2f8e42cf1d4be3169a98dd4be32bafe4f22b6c4cb4ba810fa9e5d2cb", size = 293419, upload-time = "2025-07-26T12:01:21.16Z" }, - { url = "https://files.pythonhosted.org/packages/53/3e/405b59cfa13021a56bba395a6b3aca8cec012b45bf177b0eaf7a202cde2c/contourpy-1.3.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:556dba8fb6f5d8742f2923fe9457dbdd51e1049c4a43fd3986a0b14a1d815fc6", size = 273979, upload-time = "2025-07-26T12:01:22.448Z" }, - { url = "https://files.pythonhosted.org/packages/d4/1c/a12359b9b2ca3a845e8f7f9ac08bdf776114eb931392fcad91743e2ea17b/contourpy-1.3.3-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:92d9abc807cf7d0e047b95ca5d957cf4792fcd04e920ca70d48add15c1a90ea7", size = 332653, upload-time = "2025-07-26T12:01:24.155Z" }, - { url = "https://files.pythonhosted.org/packages/63/12/897aeebfb475b7748ea67b61e045accdfcf0d971f8a588b67108ed7f5512/contourpy-1.3.3-cp312-cp312-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:b2e8faa0ed68cb29af51edd8e24798bb661eac3bd9f65420c1887b6ca89987c8", size = 379536, upload-time = "2025-07-26T12:01:25.91Z" }, - { url = "https://files.pythonhosted.org/packages/43/8a/a8c584b82deb248930ce069e71576fc09bd7174bbd35183b7943fb1064fd/contourpy-1.3.3-cp312-cp312-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:626d60935cf668e70a5ce6ff184fd713e9683fb458898e4249b63be9e28286ea", size = 384397, upload-time = "2025-07-26T12:01:27.152Z" }, - { url = "https://files.pythonhosted.org/packages/cc/8f/ec6289987824b29529d0dfda0d74a07cec60e54b9c92f3c9da4c0ac732de/contourpy-1.3.3-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4d00e655fcef08aba35ec9610536bfe90267d7ab5ba944f7032549c55a146da1", size = 362601, upload-time = "2025-07-26T12:01:28.808Z" }, - { url = "https://files.pythonhosted.org/packages/05/0a/a3fe3be3ee2dceb3e615ebb4df97ae6f3828aa915d3e10549ce016302bd1/contourpy-1.3.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:451e71b5a7d597379ef572de31eeb909a87246974d960049a9848c3bc6c41bf7", size = 1331288, upload-time = "2025-07-26T12:01:31.198Z" }, - { url = "https://files.pythonhosted.org/packages/33/1d/acad9bd4e97f13f3e2b18a3977fe1b4a37ecf3d38d815333980c6c72e963/contourpy-1.3.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:459c1f020cd59fcfe6650180678a9993932d80d44ccde1fa1868977438f0b411", size = 1403386, upload-time = "2025-07-26T12:01:33.947Z" }, - { url = "https://files.pythonhosted.org/packages/cf/8f/5847f44a7fddf859704217a99a23a4f6417b10e5ab1256a179264561540e/contourpy-1.3.3-cp312-cp312-win32.whl", hash = "sha256:023b44101dfe49d7d53932be418477dba359649246075c996866106da069af69", size = 185018, upload-time = "2025-07-26T12:01:35.64Z" }, - { url = "https://files.pythonhosted.org/packages/19/e8/6026ed58a64563186a9ee3f29f41261fd1828f527dd93d33b60feca63352/contourpy-1.3.3-cp312-cp312-win_amd64.whl", hash = "sha256:8153b8bfc11e1e4d75bcb0bff1db232f9e10b274e0929de9d608027e0d34ff8b", size = 226567, upload-time = "2025-07-26T12:01:36.804Z" }, - { url = "https://files.pythonhosted.org/packages/d1/e2/f05240d2c39a1ed228d8328a78b6f44cd695f7ef47beb3e684cf93604f86/contourpy-1.3.3-cp312-cp312-win_arm64.whl", hash = "sha256:07ce5ed73ecdc4a03ffe3e1b3e3c1166db35ae7584be76f65dbbe28a7791b0cc", size = 193655, upload-time = "2025-07-26T12:01:37.999Z" }, - { url = "https://files.pythonhosted.org/packages/68/35/0167aad910bbdb9599272bd96d01a9ec6852f36b9455cf2ca67bd4cc2d23/contourpy-1.3.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:177fb367556747a686509d6fef71d221a4b198a3905fe824430e5ea0fda54eb5", size = 293257, upload-time = "2025-07-26T12:01:39.367Z" }, - { url = "https://files.pythonhosted.org/packages/96/e4/7adcd9c8362745b2210728f209bfbcf7d91ba868a2c5f40d8b58f54c509b/contourpy-1.3.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:d002b6f00d73d69333dac9d0b8d5e84d9724ff9ef044fd63c5986e62b7c9e1b1", size = 274034, upload-time = "2025-07-26T12:01:40.645Z" }, - { url = "https://files.pythonhosted.org/packages/73/23/90e31ceeed1de63058a02cb04b12f2de4b40e3bef5e082a7c18d9c8ae281/contourpy-1.3.3-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:348ac1f5d4f1d66d3322420f01d42e43122f43616e0f194fc1c9f5d830c5b286", size = 334672, upload-time = "2025-07-26T12:01:41.942Z" }, - { url = "https://files.pythonhosted.org/packages/ed/93/b43d8acbe67392e659e1d984700e79eb67e2acb2bd7f62012b583a7f1b55/contourpy-1.3.3-cp313-cp313-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:655456777ff65c2c548b7c454af9c6f33f16c8884f11083244b5819cc214f1b5", size = 381234, upload-time = "2025-07-26T12:01:43.499Z" }, - { url = "https://files.pythonhosted.org/packages/46/3b/bec82a3ea06f66711520f75a40c8fc0b113b2a75edb36aa633eb11c4f50f/contourpy-1.3.3-cp313-cp313-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:644a6853d15b2512d67881586bd03f462c7ab755db95f16f14d7e238f2852c67", size = 385169, upload-time = "2025-07-26T12:01:45.219Z" }, - { url = "https://files.pythonhosted.org/packages/4b/32/e0f13a1c5b0f8572d0ec6ae2f6c677b7991fafd95da523159c19eff0696a/contourpy-1.3.3-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4debd64f124ca62069f313a9cb86656ff087786016d76927ae2cf37846b006c9", size = 362859, upload-time = "2025-07-26T12:01:46.519Z" }, - { url = "https://files.pythonhosted.org/packages/33/71/e2a7945b7de4e58af42d708a219f3b2f4cff7386e6b6ab0a0fa0033c49a9/contourpy-1.3.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:a15459b0f4615b00bbd1e91f1b9e19b7e63aea7483d03d804186f278c0af2659", size = 1332062, upload-time = "2025-07-26T12:01:48.964Z" }, - { url = "https://files.pythonhosted.org/packages/12/fc/4e87ac754220ccc0e807284f88e943d6d43b43843614f0a8afa469801db0/contourpy-1.3.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ca0fdcd73925568ca027e0b17ab07aad764be4706d0a925b89227e447d9737b7", size = 1403932, upload-time = "2025-07-26T12:01:51.979Z" }, - { url = "https://files.pythonhosted.org/packages/a6/2e/adc197a37443f934594112222ac1aa7dc9a98faf9c3842884df9a9d8751d/contourpy-1.3.3-cp313-cp313-win32.whl", hash = "sha256:b20c7c9a3bf701366556e1b1984ed2d0cedf999903c51311417cf5f591d8c78d", size = 185024, upload-time = "2025-07-26T12:01:53.245Z" }, - { url = "https://files.pythonhosted.org/packages/18/0b/0098c214843213759692cc638fce7de5c289200a830e5035d1791d7a2338/contourpy-1.3.3-cp313-cp313-win_amd64.whl", hash = "sha256:1cadd8b8969f060ba45ed7c1b714fe69185812ab43bd6b86a9123fe8f99c3263", size = 226578, upload-time = "2025-07-26T12:01:54.422Z" }, - { url = "https://files.pythonhosted.org/packages/8a/9a/2f6024a0c5995243cd63afdeb3651c984f0d2bc727fd98066d40e141ad73/contourpy-1.3.3-cp313-cp313-win_arm64.whl", hash = "sha256:fd914713266421b7536de2bfa8181aa8c699432b6763a0ea64195ebe28bff6a9", size = 193524, upload-time = "2025-07-26T12:01:55.73Z" }, - { url = "https://files.pythonhosted.org/packages/c0/b3/f8a1a86bd3298513f500e5b1f5fd92b69896449f6cab6a146a5d52715479/contourpy-1.3.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:88df9880d507169449d434c293467418b9f6cbe82edd19284aa0409e7fdb933d", size = 306730, upload-time = "2025-07-26T12:01:57.051Z" }, - { url = "https://files.pythonhosted.org/packages/3f/11/4780db94ae62fc0c2053909b65dc3246bd7cecfc4f8a20d957ad43aa4ad8/contourpy-1.3.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:d06bb1f751ba5d417047db62bca3c8fde202b8c11fb50742ab3ab962c81e8216", size = 287897, upload-time = "2025-07-26T12:01:58.663Z" }, - { url = "https://files.pythonhosted.org/packages/ae/15/e59f5f3ffdd6f3d4daa3e47114c53daabcb18574a26c21f03dc9e4e42ff0/contourpy-1.3.3-cp313-cp313t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e4e6b05a45525357e382909a4c1600444e2a45b4795163d3b22669285591c1ae", size = 326751, upload-time = "2025-07-26T12:02:00.343Z" }, - { url = "https://files.pythonhosted.org/packages/0f/81/03b45cfad088e4770b1dcf72ea78d3802d04200009fb364d18a493857210/contourpy-1.3.3-cp313-cp313t-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:ab3074b48c4e2cf1a960e6bbeb7f04566bf36b1861d5c9d4d8ac04b82e38ba20", size = 375486, upload-time = "2025-07-26T12:02:02.128Z" }, - { url = "https://files.pythonhosted.org/packages/0c/ba/49923366492ffbdd4486e970d421b289a670ae8cf539c1ea9a09822b371a/contourpy-1.3.3-cp313-cp313t-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:6c3d53c796f8647d6deb1abe867daeb66dcc8a97e8455efa729516b997b8ed99", size = 388106, upload-time = "2025-07-26T12:02:03.615Z" }, - { url = "https://files.pythonhosted.org/packages/9f/52/5b00ea89525f8f143651f9f03a0df371d3cbd2fccd21ca9b768c7a6500c2/contourpy-1.3.3-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:50ed930df7289ff2a8d7afeb9603f8289e5704755c7e5c3bbd929c90c817164b", size = 352548, upload-time = "2025-07-26T12:02:05.165Z" }, - { url = "https://files.pythonhosted.org/packages/32/1d/a209ec1a3a3452d490f6b14dd92e72280c99ae3d1e73da74f8277d4ee08f/contourpy-1.3.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:4feffb6537d64b84877da813a5c30f1422ea5739566abf0bd18065ac040e120a", size = 1322297, upload-time = "2025-07-26T12:02:07.379Z" }, - { url = "https://files.pythonhosted.org/packages/bc/9e/46f0e8ebdd884ca0e8877e46a3f4e633f6c9c8c4f3f6e72be3fe075994aa/contourpy-1.3.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:2b7e9480ffe2b0cd2e787e4df64270e3a0440d9db8dc823312e2c940c167df7e", size = 1391023, upload-time = "2025-07-26T12:02:10.171Z" }, - { url = "https://files.pythonhosted.org/packages/b9/70/f308384a3ae9cd2209e0849f33c913f658d3326900d0ff5d378d6a1422d2/contourpy-1.3.3-cp313-cp313t-win32.whl", hash = "sha256:283edd842a01e3dcd435b1c5116798d661378d83d36d337b8dde1d16a5fc9ba3", size = 196157, upload-time = "2025-07-26T12:02:11.488Z" }, - { url = "https://files.pythonhosted.org/packages/b2/dd/880f890a6663b84d9e34a6f88cded89d78f0091e0045a284427cb6b18521/contourpy-1.3.3-cp313-cp313t-win_amd64.whl", hash = "sha256:87acf5963fc2b34825e5b6b048f40e3635dd547f590b04d2ab317c2619ef7ae8", size = 240570, upload-time = "2025-07-26T12:02:12.754Z" }, - { url = "https://files.pythonhosted.org/packages/80/99/2adc7d8ffead633234817ef8e9a87115c8a11927a94478f6bb3d3f4d4f7d/contourpy-1.3.3-cp313-cp313t-win_arm64.whl", hash = "sha256:3c30273eb2a55024ff31ba7d052dde990d7d8e5450f4bbb6e913558b3d6c2301", size = 199713, upload-time = "2025-07-26T12:02:14.4Z" }, - { url = "https://files.pythonhosted.org/packages/72/8b/4546f3ab60f78c514ffb7d01a0bd743f90de36f0019d1be84d0a708a580a/contourpy-1.3.3-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:fde6c716d51c04b1c25d0b90364d0be954624a0ee9d60e23e850e8d48353d07a", size = 292189, upload-time = "2025-07-26T12:02:16.095Z" }, - { url = "https://files.pythonhosted.org/packages/fd/e1/3542a9cb596cadd76fcef413f19c79216e002623158befe6daa03dbfa88c/contourpy-1.3.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:cbedb772ed74ff5be440fa8eee9bd49f64f6e3fc09436d9c7d8f1c287b121d77", size = 273251, upload-time = "2025-07-26T12:02:17.524Z" }, - { url = "https://files.pythonhosted.org/packages/b1/71/f93e1e9471d189f79d0ce2497007731c1e6bf9ef6d1d61b911430c3db4e5/contourpy-1.3.3-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:22e9b1bd7a9b1d652cd77388465dc358dafcd2e217d35552424aa4f996f524f5", size = 335810, upload-time = "2025-07-26T12:02:18.9Z" }, - { url = "https://files.pythonhosted.org/packages/91/f9/e35f4c1c93f9275d4e38681a80506b5510e9327350c51f8d4a5a724d178c/contourpy-1.3.3-cp314-cp314-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:a22738912262aa3e254e4f3cb079a95a67132fc5a063890e224393596902f5a4", size = 382871, upload-time = "2025-07-26T12:02:20.418Z" }, - { url = "https://files.pythonhosted.org/packages/b5/71/47b512f936f66a0a900d81c396a7e60d73419868fba959c61efed7a8ab46/contourpy-1.3.3-cp314-cp314-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:afe5a512f31ee6bd7d0dda52ec9864c984ca3d66664444f2d72e0dc4eb832e36", size = 386264, upload-time = "2025-07-26T12:02:21.916Z" }, - { url = "https://files.pythonhosted.org/packages/04/5f/9ff93450ba96b09c7c2b3f81c94de31c89f92292f1380261bd7195bea4ea/contourpy-1.3.3-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f64836de09927cba6f79dcd00fdd7d5329f3fccc633468507079c829ca4db4e3", size = 363819, upload-time = "2025-07-26T12:02:23.759Z" }, - { url = "https://files.pythonhosted.org/packages/3e/a6/0b185d4cc480ee494945cde102cb0149ae830b5fa17bf855b95f2e70ad13/contourpy-1.3.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:1fd43c3be4c8e5fd6e4f2baeae35ae18176cf2e5cced681cca908addf1cdd53b", size = 1333650, upload-time = "2025-07-26T12:02:26.181Z" }, - { url = "https://files.pythonhosted.org/packages/43/d7/afdc95580ca56f30fbcd3060250f66cedbde69b4547028863abd8aa3b47e/contourpy-1.3.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:6afc576f7b33cf00996e5c1102dc2a8f7cc89e39c0b55df93a0b78c1bd992b36", size = 1404833, upload-time = "2025-07-26T12:02:28.782Z" }, - { url = "https://files.pythonhosted.org/packages/e2/e2/366af18a6d386f41132a48f033cbd2102e9b0cf6345d35ff0826cd984566/contourpy-1.3.3-cp314-cp314-win32.whl", hash = "sha256:66c8a43a4f7b8df8b71ee1840e4211a3c8d93b214b213f590e18a1beca458f7d", size = 189692, upload-time = "2025-07-26T12:02:30.128Z" }, - { url = "https://files.pythonhosted.org/packages/7d/c2/57f54b03d0f22d4044b8afb9ca0e184f8b1afd57b4f735c2fa70883dc601/contourpy-1.3.3-cp314-cp314-win_amd64.whl", hash = "sha256:cf9022ef053f2694e31d630feaacb21ea24224be1c3ad0520b13d844274614fd", size = 232424, upload-time = "2025-07-26T12:02:31.395Z" }, - { url = "https://files.pythonhosted.org/packages/18/79/a9416650df9b525737ab521aa181ccc42d56016d2123ddcb7b58e926a42c/contourpy-1.3.3-cp314-cp314-win_arm64.whl", hash = "sha256:95b181891b4c71de4bb404c6621e7e2390745f887f2a026b2d99e92c17892339", size = 198300, upload-time = "2025-07-26T12:02:32.956Z" }, - { url = "https://files.pythonhosted.org/packages/1f/42/38c159a7d0f2b7b9c04c64ab317042bb6952b713ba875c1681529a2932fe/contourpy-1.3.3-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:33c82d0138c0a062380332c861387650c82e4cf1747aaa6938b9b6516762e772", size = 306769, upload-time = "2025-07-26T12:02:34.2Z" }, - { url = "https://files.pythonhosted.org/packages/c3/6c/26a8205f24bca10974e77460de68d3d7c63e282e23782f1239f226fcae6f/contourpy-1.3.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:ea37e7b45949df430fe649e5de8351c423430046a2af20b1c1961cae3afcda77", size = 287892, upload-time = "2025-07-26T12:02:35.807Z" }, - { url = "https://files.pythonhosted.org/packages/66/06/8a475c8ab718ebfd7925661747dbb3c3ee9c82ac834ccb3570be49d129f4/contourpy-1.3.3-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d304906ecc71672e9c89e87c4675dc5c2645e1f4269a5063b99b0bb29f232d13", size = 326748, upload-time = "2025-07-26T12:02:37.193Z" }, - { url = "https://files.pythonhosted.org/packages/b4/a3/c5ca9f010a44c223f098fccd8b158bb1cb287378a31ac141f04730dc49be/contourpy-1.3.3-cp314-cp314t-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:ca658cd1a680a5c9ea96dc61cdbae1e85c8f25849843aa799dfd3cb370ad4fbe", size = 375554, upload-time = "2025-07-26T12:02:38.894Z" }, - { url = "https://files.pythonhosted.org/packages/80/5b/68bd33ae63fac658a4145088c1e894405e07584a316738710b636c6d0333/contourpy-1.3.3-cp314-cp314t-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:ab2fd90904c503739a75b7c8c5c01160130ba67944a7b77bbf36ef8054576e7f", size = 388118, upload-time = "2025-07-26T12:02:40.642Z" }, - { url = "https://files.pythonhosted.org/packages/40/52/4c285a6435940ae25d7410a6c36bda5145839bc3f0beb20c707cda18b9d2/contourpy-1.3.3-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b7301b89040075c30e5768810bc96a8e8d78085b47d8be6e4c3f5a0b4ed478a0", size = 352555, upload-time = "2025-07-26T12:02:42.25Z" }, - { url = "https://files.pythonhosted.org/packages/24/ee/3e81e1dd174f5c7fefe50e85d0892de05ca4e26ef1c9a59c2a57e43b865a/contourpy-1.3.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:2a2a8b627d5cc6b7c41a4beff6c5ad5eb848c88255fda4a8745f7e901b32d8e4", size = 1322295, upload-time = "2025-07-26T12:02:44.668Z" }, - { url = "https://files.pythonhosted.org/packages/3c/b2/6d913d4d04e14379de429057cd169e5e00f6c2af3bb13e1710bcbdb5da12/contourpy-1.3.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:fd6ec6be509c787f1caf6b247f0b1ca598bef13f4ddeaa126b7658215529ba0f", size = 1391027, upload-time = "2025-07-26T12:02:47.09Z" }, - { url = "https://files.pythonhosted.org/packages/93/8a/68a4ec5c55a2971213d29a9374913f7e9f18581945a7a31d1a39b5d2dfe5/contourpy-1.3.3-cp314-cp314t-win32.whl", hash = "sha256:e74a9a0f5e3fff48fb5a7f2fd2b9b70a3fe014a67522f79b7cca4c0c7e43c9ae", size = 202428, upload-time = "2025-07-26T12:02:48.691Z" }, - { url = "https://files.pythonhosted.org/packages/fa/96/fd9f641ffedc4fa3ace923af73b9d07e869496c9cc7a459103e6e978992f/contourpy-1.3.3-cp314-cp314t-win_amd64.whl", hash = "sha256:13b68d6a62db8eafaebb8039218921399baf6e47bf85006fd8529f2a08ef33fc", size = 250331, upload-time = "2025-07-26T12:02:50.137Z" }, - { url = "https://files.pythonhosted.org/packages/ae/8c/469afb6465b853afff216f9528ffda78a915ff880ed58813ba4faf4ba0b6/contourpy-1.3.3-cp314-cp314t-win_arm64.whl", hash = "sha256:b7448cb5a725bb1e35ce88771b86fba35ef418952474492cf7c764059933ff8b", size = 203831, upload-time = "2025-07-26T12:02:51.449Z" }, - { url = "https://files.pythonhosted.org/packages/a5/29/8dcfe16f0107943fa92388c23f6e05cff0ba58058c4c95b00280d4c75a14/contourpy-1.3.3-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:cd5dfcaeb10f7b7f9dc8941717c6c2ade08f587be2226222c12b25f0483ed497", size = 278809, upload-time = "2025-07-26T12:02:52.74Z" }, - { url = "https://files.pythonhosted.org/packages/85/a9/8b37ef4f7dafeb335daee3c8254645ef5725be4d9c6aa70b50ec46ef2f7e/contourpy-1.3.3-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:0c1fc238306b35f246d61a1d416a627348b5cf0648648a031e14bb8705fcdfe8", size = 261593, upload-time = "2025-07-26T12:02:54.037Z" }, - { url = "https://files.pythonhosted.org/packages/0a/59/ebfb8c677c75605cc27f7122c90313fd2f375ff3c8d19a1694bda74aaa63/contourpy-1.3.3-pp311-pypy311_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:70f9aad7de812d6541d29d2bbf8feb22ff7e1c299523db288004e3157ff4674e", size = 302202, upload-time = "2025-07-26T12:02:55.947Z" }, - { url = "https://files.pythonhosted.org/packages/3c/37/21972a15834d90bfbfb009b9d004779bd5a07a0ec0234e5ba8f64d5736f4/contourpy-1.3.3-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5ed3657edf08512fc3fe81b510e35c2012fbd3081d2e26160f27ca28affec989", size = 329207, upload-time = "2025-07-26T12:02:57.468Z" }, - { url = "https://files.pythonhosted.org/packages/0c/58/bd257695f39d05594ca4ad60df5bcb7e32247f9951fd09a9b8edb82d1daa/contourpy-1.3.3-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:3d1a3799d62d45c18bafd41c5fa05120b96a28079f2393af559b843d1a966a77", size = 225315, upload-time = "2025-07-26T12:02:58.801Z" }, +version = "1.3.1" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "numpy" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/25/c2/fc7193cc5383637ff390a712e88e4ded0452c9fbcf84abe3de5ea3df1866/contourpy-1.3.1.tar.gz", hash = "sha256:dfd97abd83335045a913e3bcc4a09c0ceadbe66580cf573fe961f4a825efa699", size = 13465753, upload-time = "2024-11-12T11:00:59.118Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/b2/a3/80937fe3efe0edacf67c9a20b955139a1a622730042c1ea991956f2704ad/contourpy-1.3.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:a045f341a77b77e1c5de31e74e966537bba9f3c4099b35bf4c2e3939dd54cdab", size = 268466, upload-time = "2024-11-12T10:52:03.706Z" }, + { url = "https://files.pythonhosted.org/packages/82/1d/e3eaebb4aa2d7311528c048350ca8e99cdacfafd99da87bc0a5f8d81f2c2/contourpy-1.3.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:500360b77259914f7805af7462e41f9cb7ca92ad38e9f94d6c8641b089338124", size = 253314, upload-time = "2024-11-12T10:52:08.721Z" }, + { url = "https://files.pythonhosted.org/packages/de/f3/d796b22d1a2b587acc8100ba8c07fb7b5e17fde265a7bb05ab967f4c935a/contourpy-1.3.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b2f926efda994cdf3c8d3fdb40b9962f86edbc4457e739277b961eced3d0b4c1", size = 312003, upload-time = "2024-11-12T10:52:13.868Z" }, + { url = "https://files.pythonhosted.org/packages/bf/f5/0e67902bc4394daee8daa39c81d4f00b50e063ee1a46cb3938cc65585d36/contourpy-1.3.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:adce39d67c0edf383647a3a007de0a45fd1b08dedaa5318404f1a73059c2512b", size = 351896, upload-time = "2024-11-12T10:52:19.513Z" }, + { url = "https://files.pythonhosted.org/packages/1f/d6/e766395723f6256d45d6e67c13bb638dd1fa9dc10ef912dc7dd3dcfc19de/contourpy-1.3.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:abbb49fb7dac584e5abc6636b7b2a7227111c4f771005853e7d25176daaf8453", size = 320814, upload-time = "2024-11-12T10:52:25.053Z" }, + { url = "https://files.pythonhosted.org/packages/a9/57/86c500d63b3e26e5b73a28b8291a67c5608d4aa87ebd17bd15bb33c178bc/contourpy-1.3.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a0cffcbede75c059f535725c1680dfb17b6ba8753f0c74b14e6a9c68c29d7ea3", size = 324969, upload-time = "2024-11-12T10:52:30.731Z" }, + { url = "https://files.pythonhosted.org/packages/b8/62/bb146d1289d6b3450bccc4642e7f4413b92ebffd9bf2e91b0404323704a7/contourpy-1.3.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:ab29962927945d89d9b293eabd0d59aea28d887d4f3be6c22deaefbb938a7277", size = 1265162, upload-time = "2024-11-12T10:52:46.26Z" }, + { url = "https://files.pythonhosted.org/packages/18/04/9f7d132ce49a212c8e767042cc80ae390f728060d2eea47058f55b9eff1c/contourpy-1.3.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:974d8145f8ca354498005b5b981165b74a195abfae9a8129df3e56771961d595", size = 1324328, upload-time = "2024-11-12T10:53:03.081Z" }, + { url = "https://files.pythonhosted.org/packages/46/23/196813901be3f97c83ababdab1382e13e0edc0bb4e7b49a7bff15fcf754e/contourpy-1.3.1-cp310-cp310-win32.whl", hash = "sha256:ac4578ac281983f63b400f7fe6c101bedc10651650eef012be1ccffcbacf3697", size = 173861, upload-time = "2024-11-12T10:53:06.283Z" }, + { url = "https://files.pythonhosted.org/packages/e0/82/c372be3fc000a3b2005061ca623a0d1ecd2eaafb10d9e883a2fc8566e951/contourpy-1.3.1-cp310-cp310-win_amd64.whl", hash = "sha256:174e758c66bbc1c8576992cec9599ce8b6672b741b5d336b5c74e35ac382b18e", size = 218566, upload-time = "2024-11-12T10:53:09.798Z" }, + { url = "https://files.pythonhosted.org/packages/12/bb/11250d2906ee2e8b466b5f93e6b19d525f3e0254ac8b445b56e618527718/contourpy-1.3.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:3e8b974d8db2c5610fb4e76307e265de0edb655ae8169e8b21f41807ccbeec4b", size = 269555, upload-time = "2024-11-12T10:53:14.707Z" }, + { url = "https://files.pythonhosted.org/packages/67/71/1e6e95aee21a500415f5d2dbf037bf4567529b6a4e986594d7026ec5ae90/contourpy-1.3.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:20914c8c973f41456337652a6eeca26d2148aa96dd7ac323b74516988bea89fc", size = 254549, upload-time = "2024-11-12T10:53:19.42Z" }, + { url = "https://files.pythonhosted.org/packages/31/2c/b88986e8d79ac45efe9d8801ae341525f38e087449b6c2f2e6050468a42c/contourpy-1.3.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:19d40d37c1c3a4961b4619dd9d77b12124a453cc3d02bb31a07d58ef684d3d86", size = 313000, upload-time = "2024-11-12T10:53:23.944Z" }, + { url = "https://files.pythonhosted.org/packages/c4/18/65280989b151fcf33a8352f992eff71e61b968bef7432fbfde3a364f0730/contourpy-1.3.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:113231fe3825ebf6f15eaa8bc1f5b0ddc19d42b733345eae0934cb291beb88b6", size = 352925, upload-time = "2024-11-12T10:53:29.719Z" }, + { url = "https://files.pythonhosted.org/packages/f5/c7/5fd0146c93220dbfe1a2e0f98969293b86ca9bc041d6c90c0e065f4619ad/contourpy-1.3.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4dbbc03a40f916a8420e420d63e96a1258d3d1b58cbdfd8d1f07b49fcbd38e85", size = 323693, upload-time = "2024-11-12T10:53:35.046Z" }, + { url = "https://files.pythonhosted.org/packages/85/fc/7fa5d17daf77306840a4e84668a48ddff09e6bc09ba4e37e85ffc8e4faa3/contourpy-1.3.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3a04ecd68acbd77fa2d39723ceca4c3197cb2969633836ced1bea14e219d077c", size = 326184, upload-time = "2024-11-12T10:53:40.261Z" }, + { url = "https://files.pythonhosted.org/packages/ef/e7/104065c8270c7397c9571620d3ab880558957216f2b5ebb7e040f85eeb22/contourpy-1.3.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c414fc1ed8ee1dbd5da626cf3710c6013d3d27456651d156711fa24f24bd1291", size = 1268031, upload-time = "2024-11-12T10:53:55.876Z" }, + { url = "https://files.pythonhosted.org/packages/e2/4a/c788d0bdbf32c8113c2354493ed291f924d4793c4a2e85b69e737a21a658/contourpy-1.3.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:31c1b55c1f34f80557d3830d3dd93ba722ce7e33a0b472cba0ec3b6535684d8f", size = 1325995, upload-time = "2024-11-12T10:54:11.572Z" }, + { url = "https://files.pythonhosted.org/packages/a6/e6/a2f351a90d955f8b0564caf1ebe4b1451a3f01f83e5e3a414055a5b8bccb/contourpy-1.3.1-cp311-cp311-win32.whl", hash = "sha256:f611e628ef06670df83fce17805c344710ca5cde01edfdc72751311da8585375", size = 174396, upload-time = "2024-11-12T10:54:15.358Z" }, + { url = "https://files.pythonhosted.org/packages/a8/7e/cd93cab453720a5d6cb75588cc17dcdc08fc3484b9de98b885924ff61900/contourpy-1.3.1-cp311-cp311-win_amd64.whl", hash = "sha256:b2bdca22a27e35f16794cf585832e542123296b4687f9fd96822db6bae17bfc9", size = 219787, upload-time = "2024-11-12T10:54:18.836Z" }, + { url = "https://files.pythonhosted.org/packages/37/6b/175f60227d3e7f5f1549fcb374592be311293132207e451c3d7c654c25fb/contourpy-1.3.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:0ffa84be8e0bd33410b17189f7164c3589c229ce5db85798076a3fa136d0e509", size = 271494, upload-time = "2024-11-12T10:54:23.6Z" }, + { url = "https://files.pythonhosted.org/packages/6b/6a/7833cfae2c1e63d1d8875a50fd23371394f540ce809d7383550681a1fa64/contourpy-1.3.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:805617228ba7e2cbbfb6c503858e626ab528ac2a32a04a2fe88ffaf6b02c32bc", size = 255444, upload-time = "2024-11-12T10:54:28.267Z" }, + { url = "https://files.pythonhosted.org/packages/7f/b3/7859efce66eaca5c14ba7619791b084ed02d868d76b928ff56890d2d059d/contourpy-1.3.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ade08d343436a94e633db932e7e8407fe7de8083967962b46bdfc1b0ced39454", size = 307628, upload-time = "2024-11-12T10:54:33.418Z" }, + { url = "https://files.pythonhosted.org/packages/48/b2/011415f5e3f0a50b1e285a0bf78eb5d92a4df000553570f0851b6e309076/contourpy-1.3.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:47734d7073fb4590b4a40122b35917cd77be5722d80683b249dac1de266aac80", size = 347271, upload-time = "2024-11-12T10:54:38.816Z" }, + { url = "https://files.pythonhosted.org/packages/84/7d/ef19b1db0f45b151ac78c65127235239a8cf21a59d1ce8507ce03e89a30b/contourpy-1.3.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2ba94a401342fc0f8b948e57d977557fbf4d515f03c67682dd5c6191cb2d16ec", size = 318906, upload-time = "2024-11-12T10:54:44.132Z" }, + { url = "https://files.pythonhosted.org/packages/ba/99/6794142b90b853a9155316c8f470d2e4821fe6f086b03e372aca848227dd/contourpy-1.3.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:efa874e87e4a647fd2e4f514d5e91c7d493697127beb95e77d2f7561f6905bd9", size = 323622, upload-time = "2024-11-12T10:54:48.788Z" }, + { url = "https://files.pythonhosted.org/packages/3c/0f/37d2c84a900cd8eb54e105f4fa9aebd275e14e266736778bb5dccbf3bbbb/contourpy-1.3.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:1bf98051f1045b15c87868dbaea84f92408337d4f81d0e449ee41920ea121d3b", size = 1266699, upload-time = "2024-11-12T10:55:04.016Z" }, + { url = "https://files.pythonhosted.org/packages/3a/8a/deb5e11dc7d9cc8f0f9c8b29d4f062203f3af230ba83c30a6b161a6effc9/contourpy-1.3.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:61332c87493b00091423e747ea78200659dc09bdf7fd69edd5e98cef5d3e9a8d", size = 1326395, upload-time = "2024-11-12T10:55:20.547Z" }, + { url = "https://files.pythonhosted.org/packages/1a/35/7e267ae7c13aaf12322ccc493531f1e7f2eb8fba2927b9d7a05ff615df7a/contourpy-1.3.1-cp312-cp312-win32.whl", hash = "sha256:e914a8cb05ce5c809dd0fe350cfbb4e881bde5e2a38dc04e3afe1b3e58bd158e", size = 175354, upload-time = "2024-11-12T10:55:24.377Z" }, + { url = "https://files.pythonhosted.org/packages/a1/35/c2de8823211d07e8a79ab018ef03960716c5dff6f4d5bff5af87fd682992/contourpy-1.3.1-cp312-cp312-win_amd64.whl", hash = "sha256:08d9d449a61cf53033612cb368f3a1b26cd7835d9b8cd326647efe43bca7568d", size = 220971, upload-time = "2024-11-12T10:55:27.971Z" }, + { url = "https://files.pythonhosted.org/packages/9a/e7/de62050dce687c5e96f946a93546910bc67e483fe05324439e329ff36105/contourpy-1.3.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:a761d9ccfc5e2ecd1bf05534eda382aa14c3e4f9205ba5b1684ecfe400716ef2", size = 271548, upload-time = "2024-11-12T10:55:32.228Z" }, + { url = "https://files.pythonhosted.org/packages/78/4d/c2a09ae014ae984c6bdd29c11e74d3121b25eaa117eca0bb76340efd7e1c/contourpy-1.3.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:523a8ee12edfa36f6d2a49407f705a6ef4c5098de4f498619787e272de93f2d5", size = 255576, upload-time = "2024-11-12T10:55:36.246Z" }, + { url = "https://files.pythonhosted.org/packages/ab/8a/915380ee96a5638bda80cd061ccb8e666bfdccea38d5741cb69e6dbd61fc/contourpy-1.3.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ece6df05e2c41bd46776fbc712e0996f7c94e0d0543af1656956d150c4ca7c81", size = 306635, upload-time = "2024-11-12T10:55:41.904Z" }, + { url = "https://files.pythonhosted.org/packages/29/5c/c83ce09375428298acd4e6582aeb68b1e0d1447f877fa993d9bf6cd3b0a0/contourpy-1.3.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:573abb30e0e05bf31ed067d2f82500ecfdaec15627a59d63ea2d95714790f5c2", size = 345925, upload-time = "2024-11-12T10:55:47.206Z" }, + { url = "https://files.pythonhosted.org/packages/29/63/5b52f4a15e80c66c8078a641a3bfacd6e07106835682454647aca1afc852/contourpy-1.3.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:a9fa36448e6a3a1a9a2ba23c02012c43ed88905ec80163f2ffe2421c7192a5d7", size = 318000, upload-time = "2024-11-12T10:55:52.264Z" }, + { url = "https://files.pythonhosted.org/packages/9a/e2/30ca086c692691129849198659bf0556d72a757fe2769eb9620a27169296/contourpy-1.3.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3ea9924d28fc5586bf0b42d15f590b10c224117e74409dd7a0be3b62b74a501c", size = 322689, upload-time = "2024-11-12T10:55:57.858Z" }, + { url = "https://files.pythonhosted.org/packages/6b/77/f37812ef700f1f185d348394debf33f22d531e714cf6a35d13d68a7003c7/contourpy-1.3.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:5b75aa69cb4d6f137b36f7eb2ace9280cfb60c55dc5f61c731fdf6f037f958a3", size = 1268413, upload-time = "2024-11-12T10:56:13.328Z" }, + { url = "https://files.pythonhosted.org/packages/3f/6d/ce84e79cdd128542ebeb268f84abb4b093af78e7f8ec504676673d2675bc/contourpy-1.3.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:041b640d4ec01922083645a94bb3b2e777e6b626788f4095cf21abbe266413c1", size = 1326530, upload-time = "2024-11-12T10:56:30.07Z" }, + { url = "https://files.pythonhosted.org/packages/72/22/8282f4eae20c73c89bee7a82a19c4e27af9b57bb602ecaa00713d5bdb54d/contourpy-1.3.1-cp313-cp313-win32.whl", hash = "sha256:36987a15e8ace5f58d4d5da9dca82d498c2bbb28dff6e5d04fbfcc35a9cb3a82", size = 175315, upload-time = "2024-11-12T10:57:42.804Z" }, + { url = "https://files.pythonhosted.org/packages/e3/d5/28bca491f65312b438fbf076589dcde7f6f966b196d900777f5811b9c4e2/contourpy-1.3.1-cp313-cp313-win_amd64.whl", hash = "sha256:a7895f46d47671fa7ceec40f31fae721da51ad34bdca0bee83e38870b1f47ffd", size = 220987, upload-time = "2024-11-12T10:57:46.365Z" }, + { url = "https://files.pythonhosted.org/packages/2f/24/a4b285d6adaaf9746e4700932f579f1a7b6f9681109f694cfa233ae75c4e/contourpy-1.3.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:9ddeb796389dadcd884c7eb07bd14ef12408aaae358f0e2ae24114d797eede30", size = 285001, upload-time = "2024-11-12T10:56:34.483Z" }, + { url = "https://files.pythonhosted.org/packages/48/1d/fb49a401b5ca4f06ccf467cd6c4f1fd65767e63c21322b29b04ec40b40b9/contourpy-1.3.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:19c1555a6801c2f084c7ddc1c6e11f02eb6a6016ca1318dd5452ba3f613a1751", size = 268553, upload-time = "2024-11-12T10:56:39.167Z" }, + { url = "https://files.pythonhosted.org/packages/79/1e/4aef9470d13fd029087388fae750dccb49a50c012a6c8d1d634295caa644/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:841ad858cff65c2c04bf93875e384ccb82b654574a6d7f30453a04f04af71342", size = 310386, upload-time = "2024-11-12T10:56:44.594Z" }, + { url = "https://files.pythonhosted.org/packages/b0/34/910dc706ed70153b60392b5305c708c9810d425bde12499c9184a1100888/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4318af1c925fb9a4fb190559ef3eec206845f63e80fb603d47f2d6d67683901c", size = 349806, upload-time = "2024-11-12T10:56:49.565Z" }, + { url = "https://files.pythonhosted.org/packages/31/3c/faee6a40d66d7f2a87f7102236bf4780c57990dd7f98e5ff29881b1b1344/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:14c102b0eab282427b662cb590f2e9340a9d91a1c297f48729431f2dcd16e14f", size = 321108, upload-time = "2024-11-12T10:56:55.013Z" }, + { url = "https://files.pythonhosted.org/packages/17/69/390dc9b20dd4bb20585651d7316cc3054b7d4a7b4f8b710b2b698e08968d/contourpy-1.3.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:05e806338bfeaa006acbdeba0ad681a10be63b26e1b17317bfac3c5d98f36cda", size = 327291, upload-time = "2024-11-12T10:56:59.897Z" }, + { url = "https://files.pythonhosted.org/packages/ef/74/7030b67c4e941fe1e5424a3d988080e83568030ce0355f7c9fc556455b01/contourpy-1.3.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:4d76d5993a34ef3df5181ba3c92fabb93f1eaa5729504fb03423fcd9f3177242", size = 1263752, upload-time = "2024-11-12T10:57:14.79Z" }, + { url = "https://files.pythonhosted.org/packages/f0/ed/92d86f183a8615f13f6b9cbfc5d4298a509d6ce433432e21da838b4b63f4/contourpy-1.3.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:89785bb2a1980c1bd87f0cb1517a71cde374776a5f150936b82580ae6ead44a1", size = 1318403, upload-time = "2024-11-12T10:57:31.326Z" }, + { url = "https://files.pythonhosted.org/packages/b3/0e/c8e4950c77dcfc897c71d61e56690a0a9df39543d2164040301b5df8e67b/contourpy-1.3.1-cp313-cp313t-win32.whl", hash = "sha256:8eb96e79b9f3dcadbad2a3891672f81cdcab7f95b27f28f1c67d75f045b6b4f1", size = 185117, upload-time = "2024-11-12T10:57:34.735Z" }, + { url = "https://files.pythonhosted.org/packages/c1/31/1ae946f11dfbd229222e6d6ad8e7bd1891d3d48bde5fbf7a0beb9491f8e3/contourpy-1.3.1-cp313-cp313t-win_amd64.whl", hash = "sha256:287ccc248c9e0d0566934e7d606201abd74761b5703d804ff3df8935f523d546", size = 236668, upload-time = "2024-11-12T10:57:39.061Z" }, + { url = "https://files.pythonhosted.org/packages/3e/4f/e56862e64b52b55b5ddcff4090085521fc228ceb09a88390a2b103dccd1b/contourpy-1.3.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:b457d6430833cee8e4b8e9b6f07aa1c161e5e0d52e118dc102c8f9bd7dd060d6", size = 265605, upload-time = "2024-11-12T10:57:51.188Z" }, + { url = "https://files.pythonhosted.org/packages/b0/2e/52bfeeaa4541889f23d8eadc6386b442ee2470bd3cff9baa67deb2dd5c57/contourpy-1.3.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cb76c1a154b83991a3cbbf0dfeb26ec2833ad56f95540b442c73950af2013750", size = 315040, upload-time = "2024-11-12T10:57:56.492Z" }, + { url = "https://files.pythonhosted.org/packages/52/94/86bfae441707205634d80392e873295652fc313dfd93c233c52c4dc07874/contourpy-1.3.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:44a29502ca9c7b5ba389e620d44f2fbe792b1fb5734e8b931ad307071ec58c53", size = 218221, upload-time = "2024-11-12T10:58:00.033Z" }, ] [[package]] @@ -646,27 +545,27 @@ wheels = [ [[package]] name = "debugpy" -version = "1.8.15" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/8c/8b/3a9a28ddb750a76eaec445c7f4d3147ea2c579a97dbd9e25d39001b92b21/debugpy-1.8.15.tar.gz", hash = "sha256:58d7a20b7773ab5ee6bdfb2e6cf622fdf1e40c9d5aef2857d85391526719ac00", size = 1643279, upload-time = "2025-07-15T16:43:29.135Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/69/51/0b4315169f0d945271db037ae6b98c0548a2d48cc036335cd1b2f5516c1b/debugpy-1.8.15-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:e9a8125c85172e3ec30985012e7a81ea5e70bbb836637f8a4104f454f9b06c97", size = 2084890, upload-time = "2025-07-15T16:43:31.239Z" }, - { url = "https://files.pythonhosted.org/packages/36/cc/a5391dedb079280d7b72418022e00ba8227ae0b5bc8b2e3d1ecffc5d6b01/debugpy-1.8.15-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7fd0b6b5eccaa745c214fd240ea82f46049d99ef74b185a3517dad3ea1ec55d9", size = 3561470, upload-time = "2025-07-15T16:43:32.515Z" }, - { url = "https://files.pythonhosted.org/packages/e8/92/acf64b92010c66b33c077dee3862c733798a2c90e7d14b25c01d771e2a0d/debugpy-1.8.15-cp310-cp310-win32.whl", hash = "sha256:8181cce4d344010f6bfe94a531c351a46a96b0f7987750932b2908e7a1e14a55", size = 5229194, upload-time = "2025-07-15T16:43:33.997Z" }, - { url = "https://files.pythonhosted.org/packages/3f/f5/c58c015c9ff78de35901bea3ab4dbf7946d7a4aa867ee73875df06ba6468/debugpy-1.8.15-cp310-cp310-win_amd64.whl", hash = "sha256:af2dcae4e4cd6e8b35f982ccab29fe65f7e8766e10720a717bc80c464584ee21", size = 5260900, upload-time = "2025-07-15T16:43:35.413Z" }, - { url = "https://files.pythonhosted.org/packages/d2/b3/1c44a2ed311199ab11c2299c9474a6c7cd80d19278defd333aeb7c287995/debugpy-1.8.15-cp311-cp311-macosx_14_0_universal2.whl", hash = "sha256:babc4fb1962dd6a37e94d611280e3d0d11a1f5e6c72ac9b3d87a08212c4b6dd3", size = 2183442, upload-time = "2025-07-15T16:43:36.733Z" }, - { url = "https://files.pythonhosted.org/packages/f6/69/e2dcb721491e1c294d348681227c9b44fb95218f379aa88e12a19d85528d/debugpy-1.8.15-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f778e68f2986a58479d0ac4f643e0b8c82fdd97c2e200d4d61e7c2d13838eb53", size = 3134215, upload-time = "2025-07-15T16:43:38.116Z" }, - { url = "https://files.pythonhosted.org/packages/17/76/4ce63b95d8294dcf2fd1820860b300a420d077df4e93afcaa25a984c2ca7/debugpy-1.8.15-cp311-cp311-win32.whl", hash = "sha256:f9d1b5abd75cd965e2deabb1a06b0e93a1546f31f9f621d2705e78104377c702", size = 5154037, upload-time = "2025-07-15T16:43:39.471Z" }, - { url = "https://files.pythonhosted.org/packages/c2/a7/e5a7c784465eb9c976d84408873d597dc7ce74a0fc69ed009548a1a94813/debugpy-1.8.15-cp311-cp311-win_amd64.whl", hash = "sha256:62954fb904bec463e2b5a415777f6d1926c97febb08ef1694da0e5d1463c5c3b", size = 5178133, upload-time = "2025-07-15T16:43:40.969Z" }, - { url = "https://files.pythonhosted.org/packages/ab/4a/4508d256e52897f5cdfee6a6d7580974811e911c6d01321df3264508a5ac/debugpy-1.8.15-cp312-cp312-macosx_14_0_universal2.whl", hash = "sha256:3dcc7225cb317469721ab5136cda9ff9c8b6e6fb43e87c9e15d5b108b99d01ba", size = 2511197, upload-time = "2025-07-15T16:43:42.343Z" }, - { url = "https://files.pythonhosted.org/packages/99/8d/7f6ef1097e7fecf26b4ef72338d08e41644a41b7ee958a19f494ffcffc29/debugpy-1.8.15-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:047a493ca93c85ccede1dbbaf4e66816794bdc214213dde41a9a61e42d27f8fc", size = 4229517, upload-time = "2025-07-15T16:43:44.14Z" }, - { url = "https://files.pythonhosted.org/packages/3f/e8/e8c6a9aa33a9c9c6dacbf31747384f6ed2adde4de2e9693c766bdf323aa3/debugpy-1.8.15-cp312-cp312-win32.whl", hash = "sha256:b08e9b0bc260cf324c890626961dad4ffd973f7568fbf57feb3c3a65ab6b6327", size = 5276132, upload-time = "2025-07-15T16:43:45.529Z" }, - { url = "https://files.pythonhosted.org/packages/e9/ad/231050c6177b3476b85fcea01e565dac83607b5233d003ff067e2ee44d8f/debugpy-1.8.15-cp312-cp312-win_amd64.whl", hash = "sha256:e2a4fe357c92334272eb2845fcfcdbec3ef9f22c16cf613c388ac0887aed15fa", size = 5317645, upload-time = "2025-07-15T16:43:46.968Z" }, - { url = "https://files.pythonhosted.org/packages/28/70/2928aad2310726d5920b18ed9f54b9f06df5aa4c10cf9b45fa18ff0ab7e8/debugpy-1.8.15-cp313-cp313-macosx_14_0_universal2.whl", hash = "sha256:f5e01291ad7d6649aed5773256c5bba7a1a556196300232de1474c3c372592bf", size = 2495538, upload-time = "2025-07-15T16:43:48.927Z" }, - { url = "https://files.pythonhosted.org/packages/9e/c6/9b8ffb4ca91fac8b2877eef63c9cc0e87dd2570b1120054c272815ec4cd0/debugpy-1.8.15-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:94dc0f0d00e528d915e0ce1c78e771475b2335b376c49afcc7382ee0b146bab6", size = 4221874, upload-time = "2025-07-15T16:43:50.282Z" }, - { url = "https://files.pythonhosted.org/packages/55/8a/9b8d59674b4bf489318c7c46a1aab58e606e583651438084b7e029bf3c43/debugpy-1.8.15-cp313-cp313-win32.whl", hash = "sha256:fcf0748d4f6e25f89dc5e013d1129ca6f26ad4da405e0723a4f704583896a709", size = 5275949, upload-time = "2025-07-15T16:43:52.079Z" }, - { url = "https://files.pythonhosted.org/packages/72/83/9e58e6fdfa8710a5e6ec06c2401241b9ad48b71c0a7eb99570a1f1edb1d3/debugpy-1.8.15-cp313-cp313-win_amd64.whl", hash = "sha256:73c943776cb83e36baf95e8f7f8da765896fd94b05991e7bc162456d25500683", size = 5317720, upload-time = "2025-07-15T16:43:53.703Z" }, - { url = "https://files.pythonhosted.org/packages/07/d5/98748d9860e767a1248b5e31ffa7ce8cb7006e97bf8abbf3d891d0a8ba4e/debugpy-1.8.15-py2.py3-none-any.whl", hash = "sha256:bce2e6c5ff4f2e00b98d45e7e01a49c7b489ff6df5f12d881c67d2f1ac635f3d", size = 5282697, upload-time = "2025-07-15T16:44:07.996Z" }, +version = "1.8.13" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/51/d4/f35f539e11c9344652f362c22413ec5078f677ac71229dc9b4f6f85ccaa3/debugpy-1.8.13.tar.gz", hash = "sha256:837e7bef95bdefba426ae38b9a94821ebdc5bea55627879cd48165c90b9e50ce", size = 1641193, upload-time = "2025-03-05T01:02:22.807Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/3f/32/901c7204cceb3262fdf38f4c25c9a46372c11661e8490e9ea702bc4ff448/debugpy-1.8.13-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:06859f68e817966723ffe046b896b1bd75c665996a77313370336ee9e1de3e90", size = 2076250, upload-time = "2025-03-05T01:02:26.028Z" }, + { url = "https://files.pythonhosted.org/packages/95/10/77fe746851c8d84838a807da60c7bd0ac8627a6107d6917dd3293bf8628c/debugpy-1.8.13-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cb56c2db69fb8df3168bc857d7b7d2494fed295dfdbde9a45f27b4b152f37520", size = 3560883, upload-time = "2025-03-05T01:02:28.207Z" }, + { url = "https://files.pythonhosted.org/packages/a1/ef/28f8db2070e453dda0e49b356e339d0b4e1d38058d4c4ea9e88cdc8ee8e7/debugpy-1.8.13-cp310-cp310-win32.whl", hash = "sha256:46abe0b821cad751fc1fb9f860fb2e68d75e2c5d360986d0136cd1db8cad4428", size = 5180149, upload-time = "2025-03-05T01:02:30.64Z" }, + { url = "https://files.pythonhosted.org/packages/89/16/1d53a80caf5862627d3eaffb217d4079d7e4a1df6729a2d5153733661efd/debugpy-1.8.13-cp310-cp310-win_amd64.whl", hash = "sha256:dc7b77f5d32674686a5f06955e4b18c0e41fb5a605f5b33cf225790f114cfeec", size = 5212540, upload-time = "2025-03-05T01:02:32.403Z" }, + { url = "https://files.pythonhosted.org/packages/31/90/dd2fcad8364f0964f476537481985198ce6e879760281ad1cec289f1aa71/debugpy-1.8.13-cp311-cp311-macosx_14_0_universal2.whl", hash = "sha256:eee02b2ed52a563126c97bf04194af48f2fe1f68bb522a312b05935798e922ff", size = 2174802, upload-time = "2025-03-05T01:02:34.607Z" }, + { url = "https://files.pythonhosted.org/packages/5c/c9/06ff65f15eb30dbdafd45d1575770b842ce3869ad5580a77f4e5590f1be7/debugpy-1.8.13-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4caca674206e97c85c034c1efab4483f33971d4e02e73081265ecb612af65377", size = 3133620, upload-time = "2025-03-05T01:02:36.203Z" }, + { url = "https://files.pythonhosted.org/packages/3b/49/798a4092bde16a4650f17ac5f2301d4d37e1972d65462fb25c80a83b4790/debugpy-1.8.13-cp311-cp311-win32.whl", hash = "sha256:7d9a05efc6973b5aaf076d779cf3a6bbb1199e059a17738a2aa9d27a53bcc888", size = 5104764, upload-time = "2025-03-05T01:02:38.64Z" }, + { url = "https://files.pythonhosted.org/packages/cd/d5/3684d7561c8ba2797305cf8259619acccb8d6ebe2117bb33a6897c235eee/debugpy-1.8.13-cp311-cp311-win_amd64.whl", hash = "sha256:62f9b4a861c256f37e163ada8cf5a81f4c8d5148fc17ee31fb46813bd658cdcc", size = 5129670, upload-time = "2025-03-05T01:02:40.371Z" }, + { url = "https://files.pythonhosted.org/packages/79/ad/dff929b6b5403feaab0af0e5bb460fd723f9c62538b718a9af819b8fff20/debugpy-1.8.13-cp312-cp312-macosx_14_0_universal2.whl", hash = "sha256:2b8de94c5c78aa0d0ed79023eb27c7c56a64c68217d881bee2ffbcb13951d0c1", size = 2501004, upload-time = "2025-03-05T01:02:42.602Z" }, + { url = "https://files.pythonhosted.org/packages/d6/4f/b7d42e6679f0bb525888c278b0c0d2b6dff26ed42795230bb46eaae4f9b3/debugpy-1.8.13-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:887d54276cefbe7290a754424b077e41efa405a3e07122d8897de54709dbe522", size = 4222346, upload-time = "2025-03-05T01:02:44.803Z" }, + { url = "https://files.pythonhosted.org/packages/ec/18/d9b3e88e85d41f68f77235112adc31012a784e45a3fcdbb039777d570a0f/debugpy-1.8.13-cp312-cp312-win32.whl", hash = "sha256:3872ce5453b17837ef47fb9f3edc25085ff998ce63543f45ba7af41e7f7d370f", size = 5226639, upload-time = "2025-03-05T01:02:47.144Z" }, + { url = "https://files.pythonhosted.org/packages/c9/f7/0df18a4f530ed3cc06f0060f548efe9e3316102101e311739d906f5650be/debugpy-1.8.13-cp312-cp312-win_amd64.whl", hash = "sha256:63ca7670563c320503fea26ac688988d9d6b9c6a12abc8a8cf2e7dd8e5f6b6ea", size = 5268735, upload-time = "2025-03-05T01:02:48.92Z" }, + { url = "https://files.pythonhosted.org/packages/b1/db/ae7cd645c1826aae557cebccbc448f0cc9a818d364efb88f8d80e7a03f41/debugpy-1.8.13-cp313-cp313-macosx_14_0_universal2.whl", hash = "sha256:31abc9618be4edad0b3e3a85277bc9ab51a2d9f708ead0d99ffb5bb750e18503", size = 2485416, upload-time = "2025-03-05T01:02:50.558Z" }, + { url = "https://files.pythonhosted.org/packages/ec/ed/db4b10ff3b5bb30fe41d9e86444a08bb6448e4d8265e7768450b8408dd36/debugpy-1.8.13-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a0bd87557f97bced5513a74088af0b84982b6ccb2e254b9312e29e8a5c4270eb", size = 4218784, upload-time = "2025-03-05T01:02:53.535Z" }, + { url = "https://files.pythonhosted.org/packages/82/82/ed81852a8d94086f51664d032d83c7f87cd2b087c6ea70dabec7c1ba813d/debugpy-1.8.13-cp313-cp313-win32.whl", hash = "sha256:5268ae7fdca75f526d04465931cb0bd24577477ff50e8bb03dab90983f4ebd02", size = 5226270, upload-time = "2025-03-05T01:02:56.241Z" }, + { url = "https://files.pythonhosted.org/packages/15/63/aa92fb341a78ec40f1c414ec7a7885c2ee17032eee00d12cee0cdc502af4/debugpy-1.8.13-cp313-cp313-win_amd64.whl", hash = "sha256:79ce4ed40966c4c1631d0131606b055a5a2f8e430e3f7bf8fd3744b09943e8e8", size = 5268621, upload-time = "2025-03-05T01:02:57.845Z" }, + { url = "https://files.pythonhosted.org/packages/37/4f/0b65410a08b6452bfd3f7ed6f3610f1a31fb127f46836e82d31797065dcb/debugpy-1.8.13-py2.py3-none-any.whl", hash = "sha256:d4ba115cdd0e3a70942bd562adba9ec8c651fe69ddde2298a1be296fc331906f", size = 5229306, upload-time = "2025-03-05T01:03:16.51Z" }, ] [[package]] @@ -680,11 +579,11 @@ wheels = [ [[package]] name = "distlib" -version = "0.4.0" +version = "0.3.9" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/96/8e/709914eb2b5749865801041647dc7f4e6d00b549cfe88b65ca192995f07c/distlib-0.4.0.tar.gz", hash = "sha256:feec40075be03a04501a973d81f633735b4b69f98b05450592310c0f401a4e0d", size = 614605, upload-time = "2025-07-17T16:52:00.465Z" } +sdist = { url = "https://files.pythonhosted.org/packages/0d/dd/1bec4c5ddb504ca60fc29472f3d27e8d4da1257a854e1d96742f15c1d02d/distlib-0.3.9.tar.gz", hash = "sha256:a60f20dea646b8a33f3e7772f74dc0b2d0772d2837ee1342a00645c81edf9403", size = 613923, upload-time = "2024-10-09T18:35:47.551Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/33/6b/e0547afaf41bf2c42e52430072fa5658766e3d65bd4b03a563d1b6336f57/distlib-0.4.0-py2.py3-none-any.whl", hash = "sha256:9659f7d87e46584a30b5780e43ac7a2143098441670ff0a49d5f9034c54a6c16", size = 469047, upload-time = "2025-07-17T16:51:58.613Z" }, + { url = "https://files.pythonhosted.org/packages/91/a1/cf2472db20f7ce4a6be1253a81cfdf85ad9c7885ffbed7047fb72c24cf87/distlib-0.3.9-py2.py3-none-any.whl", hash = "sha256:47f8c22fd27c27e25a65601af709b38e4f0a45ea4fc2e710f65755fa8caaaf87", size = 468973, upload-time = "2024-10-09T18:35:44.272Z" }, ] [[package]] @@ -718,14 +617,25 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/ad/e3/96f5267fe5a47c882dce7f3d06b26ddd756681fc4fbedd55d51b78b08bca/dm_tree-0.1.8-cp312-cp312-win_amd64.whl", hash = "sha256:96a548a406a6fb15fe58f6a30a57ff2f2aafbf25f05afab00c8f5e5977b6c715", size = 101754, upload-time = "2024-02-06T09:09:20.962Z" }, ] +[[package]] +name = "docker-pycreds" +version = "0.4.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "six" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/c5/e6/d1f6c00b7221e2d7c4b470132c931325c8b22c51ca62417e300f5ce16009/docker-pycreds-0.4.0.tar.gz", hash = "sha256:6ce3270bcaf404cc4c3e27e4b6c70d3521deae82fb508767870fdbf772d584d4", size = 8754, upload-time = "2018-11-29T03:26:50.996Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/f5/e8/f6bd1eee09314e7e6dee49cbe2c5e22314ccdb38db16c9fc72d2fa80d054/docker_pycreds-0.4.0-py2.py3-none-any.whl", hash = "sha256:7266112468627868005106ec19cd0d722702d2b7d5912a28e19b826c3d37af49", size = 8982, upload-time = "2018-11-29T03:26:49.575Z" }, +] + [[package]] name = "e3nn" version = "0.5.6" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "opt-einsum-fx" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "scipy" }, { name = "sympy" }, { name = "torch" }, ] @@ -761,14 +671,11 @@ wheels = [ [[package]] name = "exceptiongroup" -version = "1.3.0" +version = "1.2.2" source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "typing-extensions", marker = "python_full_version < '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/0b/9f/a65090624ecf468cdca03533906e7c69ed7588582240cfe7cc9e770b50eb/exceptiongroup-1.3.0.tar.gz", hash = "sha256:b241f5885f560bc56a59ee63ca4c6a8bfa46ae4ad651af316d4e81817bb9fd88", size = 29749, upload-time = "2025-05-10T17:42:51.123Z" } +sdist = { url = "https://files.pythonhosted.org/packages/09/35/2495c4ac46b980e4ca1f6ad6db102322ef3ad2410b79fdde159a4b0f3b92/exceptiongroup-1.2.2.tar.gz", hash = "sha256:47c2edf7c6738fafb49fd34290706d1a1a2f4d1c6df275526b62cbb4aa5393cc", size = 28883, upload-time = "2024-07-12T22:26:00.161Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/36/f4/c6e662dade71f56cd2f3735141b265c3c79293c109549c1e6933b0651ffc/exceptiongroup-1.3.0-py3-none-any.whl", hash = "sha256:4d111e6e0c13d0644cad6ddaa7ed0261a0b36971f6d23e7ec9b4b9097da78a10", size = 16674, upload-time = "2025-05-10T17:42:49.33Z" }, + { url = "https://files.pythonhosted.org/packages/02/cc/b7e31358aac6ed1ef2bb790a9746ac2c69bcb3c8588b41616914eb106eaf/exceptiongroup-1.2.2-py3-none-any.whl", hash = "sha256:3111b9d131c238bec2f8f516e123e14ba243563fb135d3fe885990585aa7795b", size = 16453, upload-time = "2024-07-12T22:25:58.476Z" }, ] [[package]] @@ -800,146 +707,121 @@ wheels = [ [[package]] name = "fonttools" -version = "4.59.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/8a/27/ec3c723bfdf86f34c5c82bf6305df3e0f0d8ea798d2d3a7cb0c0a866d286/fonttools-4.59.0.tar.gz", hash = "sha256:be392ec3529e2f57faa28709d60723a763904f71a2b63aabe14fee6648fe3b14", size = 3532521, upload-time = "2025-07-16T12:04:54.613Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/1c/1f/3dcae710b7c4b56e79442b03db64f6c9f10c3348f7af40339dffcefb581e/fonttools-4.59.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:524133c1be38445c5c0575eacea42dbd44374b310b1ffc4b60ff01d881fabb96", size = 2761846, upload-time = "2025-07-16T12:03:33.267Z" }, - { url = "https://files.pythonhosted.org/packages/eb/0e/ae3a1884fa1549acac1191cc9ec039142f6ac0e9cbc139c2e6a3dab967da/fonttools-4.59.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:21e606b2d38fed938dde871c5736822dd6bda7a4631b92e509a1f5cd1b90c5df", size = 2332060, upload-time = "2025-07-16T12:03:36.472Z" }, - { url = "https://files.pythonhosted.org/packages/75/46/58bff92a7216829159ac7bdb1d05a48ad1b8ab8c539555f12d29fdecfdd4/fonttools-4.59.0-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e93df708c69a193fc7987192f94df250f83f3851fda49413f02ba5dded639482", size = 4852354, upload-time = "2025-07-16T12:03:39.102Z" }, - { url = "https://files.pythonhosted.org/packages/05/57/767e31e48861045d89691128bd81fd4c62b62150f9a17a666f731ce4f197/fonttools-4.59.0-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:62224a9bb85b4b66d1b46d45cbe43d71cbf8f527d332b177e3b96191ffbc1e64", size = 4781132, upload-time = "2025-07-16T12:03:41.415Z" }, - { url = "https://files.pythonhosted.org/packages/d7/78/adb5e9b0af5c6ce469e8b0e112f144eaa84b30dd72a486e9c778a9b03b31/fonttools-4.59.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b8974b2a266b54c96709bd5e239979cddfd2dbceed331aa567ea1d7c4a2202db", size = 4832901, upload-time = "2025-07-16T12:03:43.115Z" }, - { url = "https://files.pythonhosted.org/packages/ac/92/bc3881097fbf3d56d112bec308c863c058e5d4c9c65f534e8ae58450ab8a/fonttools-4.59.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:209b75943d158f610b78320eacb5539aa9e920bee2c775445b2846c65d20e19d", size = 4940140, upload-time = "2025-07-16T12:03:44.781Z" }, - { url = "https://files.pythonhosted.org/packages/4a/54/39cdb23f0eeda2e07ae9cb189f2b6f41da89aabc682d3a387b3ff4a4ed29/fonttools-4.59.0-cp310-cp310-win32.whl", hash = "sha256:4c908a7036f0f3677f8afa577bcd973e3e20ddd2f7c42a33208d18bee95cdb6f", size = 2215890, upload-time = "2025-07-16T12:03:46.961Z" }, - { url = "https://files.pythonhosted.org/packages/d8/eb/f8388d9e19f95d8df2449febe9b1a38ddd758cfdb7d6de3a05198d785d61/fonttools-4.59.0-cp310-cp310-win_amd64.whl", hash = "sha256:8b4309a2775e4feee7356e63b163969a215d663399cce1b3d3b65e7ec2d9680e", size = 2260191, upload-time = "2025-07-16T12:03:48.908Z" }, - { url = "https://files.pythonhosted.org/packages/06/96/520733d9602fa1bf6592e5354c6721ac6fc9ea72bc98d112d0c38b967199/fonttools-4.59.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:841b2186adce48903c0fef235421ae21549020eca942c1da773ac380b056ab3c", size = 2782387, upload-time = "2025-07-16T12:03:51.424Z" }, - { url = "https://files.pythonhosted.org/packages/87/6a/170fce30b9bce69077d8eec9bea2cfd9f7995e8911c71be905e2eba6368b/fonttools-4.59.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:9bcc1e77fbd1609198966ded6b2a9897bd6c6bcbd2287a2fc7d75f1a254179c5", size = 2342194, upload-time = "2025-07-16T12:03:53.295Z" }, - { url = "https://files.pythonhosted.org/packages/b0/b6/7c8166c0066856f1408092f7968ac744060cf72ca53aec9036106f57eeca/fonttools-4.59.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:37c377f7cb2ab2eca8a0b319c68146d34a339792f9420fca6cd49cf28d370705", size = 5032333, upload-time = "2025-07-16T12:03:55.177Z" }, - { url = "https://files.pythonhosted.org/packages/eb/0c/707c5a19598eafcafd489b73c4cb1c142102d6197e872f531512d084aa76/fonttools-4.59.0-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fa39475eaccb98f9199eccfda4298abaf35ae0caec676ffc25b3a5e224044464", size = 4974422, upload-time = "2025-07-16T12:03:57.406Z" }, - { url = "https://files.pythonhosted.org/packages/f6/e7/6d33737d9fe632a0f59289b6f9743a86d2a9d0673de2a0c38c0f54729822/fonttools-4.59.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d3972b13148c1d1fbc092b27678a33b3080d1ac0ca305742b0119b75f9e87e38", size = 5010631, upload-time = "2025-07-16T12:03:59.449Z" }, - { url = "https://files.pythonhosted.org/packages/63/e1/a4c3d089ab034a578820c8f2dff21ef60daf9668034a1e4fb38bb1cc3398/fonttools-4.59.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:a408c3c51358c89b29cfa5317cf11518b7ce5de1717abb55c5ae2d2921027de6", size = 5122198, upload-time = "2025-07-16T12:04:01.542Z" }, - { url = "https://files.pythonhosted.org/packages/09/77/ca82b9c12fa4de3c520b7760ee61787640cf3fde55ef1b0bfe1de38c8153/fonttools-4.59.0-cp311-cp311-win32.whl", hash = "sha256:6770d7da00f358183d8fd5c4615436189e4f683bdb6affb02cad3d221d7bb757", size = 2214216, upload-time = "2025-07-16T12:04:03.515Z" }, - { url = "https://files.pythonhosted.org/packages/ab/25/5aa7ca24b560b2f00f260acf32c4cf29d7aaf8656e159a336111c18bc345/fonttools-4.59.0-cp311-cp311-win_amd64.whl", hash = "sha256:84fc186980231a287b28560d3123bd255d3c6b6659828c642b4cf961e2b923d0", size = 2261879, upload-time = "2025-07-16T12:04:05.015Z" }, - { url = "https://files.pythonhosted.org/packages/e2/77/b1c8af22f4265e951cd2e5535dbef8859efcef4fb8dee742d368c967cddb/fonttools-4.59.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:f9b3a78f69dcbd803cf2fb3f972779875b244c1115481dfbdd567b2c22b31f6b", size = 2767562, upload-time = "2025-07-16T12:04:06.895Z" }, - { url = "https://files.pythonhosted.org/packages/ff/5a/aeb975699588176bb357e8b398dfd27e5d3a2230d92b81ab8cbb6187358d/fonttools-4.59.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:57bb7e26928573ee7c6504f54c05860d867fd35e675769f3ce01b52af38d48e2", size = 2335168, upload-time = "2025-07-16T12:04:08.695Z" }, - { url = "https://files.pythonhosted.org/packages/54/97/c6101a7e60ae138c4ef75b22434373a0da50a707dad523dd19a4889315bf/fonttools-4.59.0-cp312-cp312-manylinux1_x86_64.manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:4536f2695fe5c1ffb528d84a35a7d3967e5558d2af58b4775e7ab1449d65767b", size = 4909850, upload-time = "2025-07-16T12:04:10.761Z" }, - { url = "https://files.pythonhosted.org/packages/bd/6c/fa4d18d641054f7bff878cbea14aa9433f292b9057cb1700d8e91a4d5f4f/fonttools-4.59.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:885bde7d26e5b40e15c47bd5def48b38cbd50830a65f98122a8fb90962af7cd1", size = 4955131, upload-time = "2025-07-16T12:04:12.846Z" }, - { url = "https://files.pythonhosted.org/packages/20/5c/331947fc1377deb928a69bde49f9003364f5115e5cbe351eea99e39412a2/fonttools-4.59.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6801aeddb6acb2c42eafa45bc1cb98ba236871ae6f33f31e984670b749a8e58e", size = 4899667, upload-time = "2025-07-16T12:04:14.558Z" }, - { url = "https://files.pythonhosted.org/packages/8a/46/b66469dfa26b8ff0baa7654b2cc7851206c6d57fe3abdabbaab22079a119/fonttools-4.59.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:31003b6a10f70742a63126b80863ab48175fb8272a18ca0846c0482968f0588e", size = 5051349, upload-time = "2025-07-16T12:04:16.388Z" }, - { url = "https://files.pythonhosted.org/packages/2e/05/ebfb6b1f3a4328ab69787d106a7d92ccde77ce66e98659df0f9e3f28d93d/fonttools-4.59.0-cp312-cp312-win32.whl", hash = "sha256:fbce6dae41b692a5973d0f2158f782b9ad05babc2c2019a970a1094a23909b1b", size = 2201315, upload-time = "2025-07-16T12:04:18.557Z" }, - { url = "https://files.pythonhosted.org/packages/09/45/d2bdc9ea20bbadec1016fd0db45696d573d7a26d95ab5174ffcb6d74340b/fonttools-4.59.0-cp312-cp312-win_amd64.whl", hash = "sha256:332bfe685d1ac58ca8d62b8d6c71c2e52a6c64bc218dc8f7825c9ea51385aa01", size = 2249408, upload-time = "2025-07-16T12:04:20.489Z" }, - { url = "https://files.pythonhosted.org/packages/f3/bb/390990e7c457d377b00890d9f96a3ca13ae2517efafb6609c1756e213ba4/fonttools-4.59.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:78813b49d749e1bb4db1c57f2d4d7e6db22c253cb0a86ad819f5dc197710d4b2", size = 2758704, upload-time = "2025-07-16T12:04:22.217Z" }, - { url = "https://files.pythonhosted.org/packages/df/6f/d730d9fcc9b410a11597092bd2eb9ca53e5438c6cb90e4b3047ce1b723e9/fonttools-4.59.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:401b1941ce37e78b8fd119b419b617277c65ae9417742a63282257434fd68ea2", size = 2330764, upload-time = "2025-07-16T12:04:23.985Z" }, - { url = "https://files.pythonhosted.org/packages/75/b4/b96bb66f6f8cc4669de44a158099b249c8159231d254ab6b092909388be5/fonttools-4.59.0-cp313-cp313-manylinux1_x86_64.manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_5_x86_64.whl", hash = "sha256:efd7e6660674e234e29937bc1481dceb7e0336bfae75b856b4fb272b5093c5d4", size = 4890699, upload-time = "2025-07-16T12:04:25.664Z" }, - { url = "https://files.pythonhosted.org/packages/b5/57/7969af50b26408be12baa317c6147588db5b38af2759e6df94554dbc5fdb/fonttools-4.59.0-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:51ab1ff33c19e336c02dee1e9fd1abd974a4ca3d8f7eef2a104d0816a241ce97", size = 4952934, upload-time = "2025-07-16T12:04:27.733Z" }, - { url = "https://files.pythonhosted.org/packages/d6/e2/dd968053b6cf1f46c904f5bd409b22341477c017d8201619a265e50762d3/fonttools-4.59.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:a9bf8adc9e1f3012edc8f09b08336272aec0c55bc677422273e21280db748f7c", size = 4892319, upload-time = "2025-07-16T12:04:30.074Z" }, - { url = "https://files.pythonhosted.org/packages/6b/95/a59810d8eda09129f83467a4e58f84205dc6994ebaeb9815406363e07250/fonttools-4.59.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:37e01c6ec0c98599778c2e688350d624fa4770fbd6144551bd5e032f1199171c", size = 5034753, upload-time = "2025-07-16T12:04:32.292Z" }, - { url = "https://files.pythonhosted.org/packages/a5/84/51a69ee89ff8d1fea0c6997e946657e25a3f08513de8435fe124929f3eef/fonttools-4.59.0-cp313-cp313-win32.whl", hash = "sha256:70d6b3ceaa9cc5a6ac52884f3b3d9544e8e231e95b23f138bdb78e6d4dc0eae3", size = 2199688, upload-time = "2025-07-16T12:04:34.444Z" }, - { url = "https://files.pythonhosted.org/packages/a0/ee/f626cd372932d828508137a79b85167fdcf3adab2e3bed433f295c596c6a/fonttools-4.59.0-cp313-cp313-win_amd64.whl", hash = "sha256:26731739daa23b872643f0e4072d5939960237d540c35c14e6a06d47d71ca8fe", size = 2248560, upload-time = "2025-07-16T12:04:36.034Z" }, - { url = "https://files.pythonhosted.org/packages/d0/9c/df0ef2c51845a13043e5088f7bb988ca6cd5bb82d5d4203d6a158aa58cf2/fonttools-4.59.0-py3-none-any.whl", hash = "sha256:241313683afd3baacb32a6bd124d0bce7404bc5280e12e291bae1b9bba28711d", size = 1128050, upload-time = "2025-07-16T12:04:52.687Z" }, +version = "4.56.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/1c/8c/9ffa2a555af0e5e5d0e2ed7fdd8c9bef474ed676995bb4c57c9cd0014248/fonttools-4.56.0.tar.gz", hash = "sha256:a114d1567e1a1586b7e9e7fc2ff686ca542a82769a296cef131e4c4af51e58f4", size = 3462892, upload-time = "2025-02-07T13:46:29.026Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/1e/5e/6ac30c2cc6a29454260f13c9c6422fc509b7982c13cd4597041260d8f482/fonttools-4.56.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:331954d002dbf5e704c7f3756028e21db07097c19722569983ba4d74df014000", size = 2752190, upload-time = "2025-02-07T13:43:30.593Z" }, + { url = "https://files.pythonhosted.org/packages/92/3a/ac382a8396d1b420ee45eeb0f65b614a9ca7abbb23a1b17524054f0f2200/fonttools-4.56.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:8d1613abd5af2f93c05867b3a3759a56e8bf97eb79b1da76b2bc10892f96ff16", size = 2280624, upload-time = "2025-02-07T13:43:35.349Z" }, + { url = "https://files.pythonhosted.org/packages/8a/ae/00b58bfe20e9ff7fbc3dda38f5d127913942b5e252288ea9583099a31bf5/fonttools-4.56.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:705837eae384fe21cee5e5746fd4f4b2f06f87544fa60f60740007e0aa600311", size = 4562074, upload-time = "2025-02-07T13:43:38.799Z" }, + { url = "https://files.pythonhosted.org/packages/46/d0/0004ca8f6a200252e5bd6982ed99b5fe58c4c59efaf5f516621c4cd8f703/fonttools-4.56.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc871904a53a9d4d908673c6faa15689874af1c7c5ac403a8e12d967ebd0c0dc", size = 4604747, upload-time = "2025-02-07T13:43:41.831Z" }, + { url = "https://files.pythonhosted.org/packages/45/ea/c8862bd3e09d143ef8ed8268ec8a7d477828f960954889e65288ac050b08/fonttools-4.56.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:38b947de71748bab150259ee05a775e8a0635891568e9fdb3cdd7d0e0004e62f", size = 4559025, upload-time = "2025-02-07T13:43:45.525Z" }, + { url = "https://files.pythonhosted.org/packages/8f/75/bb88a9552ec1de31a414066257bfd9f40f4ada00074f7a3799ea39b5741f/fonttools-4.56.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:86b2a1013ef7a64d2e94606632683f07712045ed86d937c11ef4dde97319c086", size = 4728482, upload-time = "2025-02-07T13:43:49.296Z" }, + { url = "https://files.pythonhosted.org/packages/2a/5f/80a2b640df1e1bb7d459d62c8b3f37fe83fd413897e549106d4ebe6371f5/fonttools-4.56.0-cp310-cp310-win32.whl", hash = "sha256:133bedb9a5c6376ad43e6518b7e2cd2f866a05b1998f14842631d5feb36b5786", size = 2155557, upload-time = "2025-02-07T13:43:52.029Z" }, + { url = "https://files.pythonhosted.org/packages/8f/85/0904f9dbe51ac70d878d3242a8583b9453a09105c3ed19c6301247fd0d3a/fonttools-4.56.0-cp310-cp310-win_amd64.whl", hash = "sha256:17f39313b649037f6c800209984a11fc256a6137cbe5487091c6c7187cae4685", size = 2200017, upload-time = "2025-02-07T13:43:54.768Z" }, + { url = "https://files.pythonhosted.org/packages/35/56/a2f3e777d48fcae7ecd29de4d96352d84e5ea9871e5f3fc88241521572cf/fonttools-4.56.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:7ef04bc7827adb7532be3d14462390dd71287644516af3f1e67f1e6ff9c6d6df", size = 2753325, upload-time = "2025-02-07T13:43:57.855Z" }, + { url = "https://files.pythonhosted.org/packages/71/85/d483e9c4e5ed586b183bf037a353e8d766366b54fd15519b30e6178a6a6e/fonttools-4.56.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ffda9b8cd9cb8b301cae2602ec62375b59e2e2108a117746f12215145e3f786c", size = 2281554, upload-time = "2025-02-07T13:44:01.671Z" }, + { url = "https://files.pythonhosted.org/packages/09/67/060473b832b2fade03c127019794df6dc02d9bc66fa4210b8e0d8a99d1e5/fonttools-4.56.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e2e993e8db36306cc3f1734edc8ea67906c55f98683d6fd34c3fc5593fdbba4c", size = 4869260, upload-time = "2025-02-07T13:44:05.746Z" }, + { url = "https://files.pythonhosted.org/packages/28/e9/47c02d5a7027e8ed841ab6a10ca00c93dadd5f16742f1af1fa3f9978adf4/fonttools-4.56.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:003548eadd674175510773f73fb2060bb46adb77c94854af3e0cc5bc70260049", size = 4898508, upload-time = "2025-02-07T13:44:09.965Z" }, + { url = "https://files.pythonhosted.org/packages/bf/8a/221d456d1afb8ca043cfd078f59f187ee5d0a580f4b49351b9ce95121f57/fonttools-4.56.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:bd9825822e7bb243f285013e653f6741954d8147427aaa0324a862cdbf4cbf62", size = 4877700, upload-time = "2025-02-07T13:44:13.598Z" }, + { url = "https://files.pythonhosted.org/packages/a4/8c/e503863adf7a6aeff7b960e2f66fa44dd0c29a7a8b79765b2821950d7b05/fonttools-4.56.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b23d30a2c0b992fb1c4f8ac9bfde44b5586d23457759b6cf9a787f1a35179ee0", size = 5045817, upload-time = "2025-02-07T13:44:17.532Z" }, + { url = "https://files.pythonhosted.org/packages/2b/50/79ba3b7e42f4eaa70b82b9e79155f0f6797858dc8a97862428b6852c6aee/fonttools-4.56.0-cp311-cp311-win32.whl", hash = "sha256:47b5e4680002ae1756d3ae3b6114e20aaee6cc5c69d1e5911f5ffffd3ee46c6b", size = 2154426, upload-time = "2025-02-07T13:44:21.063Z" }, + { url = "https://files.pythonhosted.org/packages/3b/90/4926e653041c4116ecd43e50e3c79f5daae6dcafc58ceb64bc4f71dd4924/fonttools-4.56.0-cp311-cp311-win_amd64.whl", hash = "sha256:14a3e3e6b211660db54ca1ef7006401e4a694e53ffd4553ab9bc87ead01d0f05", size = 2200937, upload-time = "2025-02-07T13:44:24.607Z" }, + { url = "https://files.pythonhosted.org/packages/39/32/71cfd6877999576a11824a7fe7bc0bb57c5c72b1f4536fa56a3e39552643/fonttools-4.56.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:d6f195c14c01bd057bc9b4f70756b510e009c83c5ea67b25ced3e2c38e6ee6e9", size = 2747757, upload-time = "2025-02-07T13:44:28.021Z" }, + { url = "https://files.pythonhosted.org/packages/15/52/d9f716b072c5061a0b915dd4c387f74bef44c68c069e2195c753905bd9b7/fonttools-4.56.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:fa760e5fe8b50cbc2d71884a1eff2ed2b95a005f02dda2fa431560db0ddd927f", size = 2279007, upload-time = "2025-02-07T13:44:31.325Z" }, + { url = "https://files.pythonhosted.org/packages/d1/97/f1b3a8afa9a0d814a092a25cd42f59ccb98a0bb7a295e6e02fc9ba744214/fonttools-4.56.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d54a45d30251f1d729e69e5b675f9a08b7da413391a1227781e2a297fa37f6d2", size = 4783991, upload-time = "2025-02-07T13:44:34.888Z" }, + { url = "https://files.pythonhosted.org/packages/95/70/2a781bedc1c45a0c61d29c56425609b22ed7f971da5d7e5df2679488741b/fonttools-4.56.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:661a8995d11e6e4914a44ca7d52d1286e2d9b154f685a4d1f69add8418961563", size = 4855109, upload-time = "2025-02-07T13:44:40.702Z" }, + { url = "https://files.pythonhosted.org/packages/0c/02/a2597858e61a5e3fb6a14d5f6be9e6eb4eaf090da56ad70cedcbdd201685/fonttools-4.56.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:9d94449ad0a5f2a8bf5d2f8d71d65088aee48adbe45f3c5f8e00e3ad861ed81a", size = 4762496, upload-time = "2025-02-07T13:44:45.929Z" }, + { url = "https://files.pythonhosted.org/packages/f2/00/aaf00100d6078fdc73f7352b44589804af9dc12b182a2540b16002152ba4/fonttools-4.56.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:f59746f7953f69cc3290ce2f971ab01056e55ddd0fb8b792c31a8acd7fee2d28", size = 4990094, upload-time = "2025-02-07T13:44:49.004Z" }, + { url = "https://files.pythonhosted.org/packages/bf/dc/3ff1db522460db60cf3adaf1b64e0c72b43406717d139786d3fa1eb20709/fonttools-4.56.0-cp312-cp312-win32.whl", hash = "sha256:bce60f9a977c9d3d51de475af3f3581d9b36952e1f8fc19a1f2254f1dda7ce9c", size = 2142888, upload-time = "2025-02-07T13:44:54.127Z" }, + { url = "https://files.pythonhosted.org/packages/6f/e3/5a181a85777f7809076e51f7422e0dc77eb04676c40ec8bf6a49d390d1ff/fonttools-4.56.0-cp312-cp312-win_amd64.whl", hash = "sha256:300c310bb725b2bdb4f5fc7e148e190bd69f01925c7ab437b9c0ca3e1c7cd9ba", size = 2189734, upload-time = "2025-02-07T13:44:57.393Z" }, + { url = "https://files.pythonhosted.org/packages/a5/55/f06b48d48e0b4ec3a3489efafe9bd4d81b6e0802ac51026e3ee4634e89ba/fonttools-4.56.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:f20e2c0dfab82983a90f3d00703ac0960412036153e5023eed2b4641d7d5e692", size = 2735127, upload-time = "2025-02-07T13:44:59.966Z" }, + { url = "https://files.pythonhosted.org/packages/59/db/d2c7c9b6dd5cbd46f183e650a47403ffb88fca17484eb7c4b1cd88f9e513/fonttools-4.56.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:f36a0868f47b7566237640c026c65a86d09a3d9ca5df1cd039e30a1da73098a0", size = 2272519, upload-time = "2025-02-07T13:45:03.891Z" }, + { url = "https://files.pythonhosted.org/packages/4d/a2/da62d779c34a0e0c06415f02eab7fa3466de5d46df459c0275a255cefc65/fonttools-4.56.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:62b4c6802fa28e14dba010e75190e0e6228513573f1eeae57b11aa1a39b7e5b1", size = 4762423, upload-time = "2025-02-07T13:45:07.034Z" }, + { url = "https://files.pythonhosted.org/packages/be/6a/fd4018e0448c8a5e12138906411282c5eab51a598493f080a9f0960e658f/fonttools-4.56.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a05d1f07eb0a7d755fbe01fee1fd255c3a4d3730130cf1bfefb682d18fd2fcea", size = 4834442, upload-time = "2025-02-07T13:45:10.6Z" }, + { url = "https://files.pythonhosted.org/packages/6d/63/fa1dec8efb35bc11ef9c39b2d74754b45d48a3ccb2cf78c0109c0af639e8/fonttools-4.56.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:0073b62c3438cf0058488c002ea90489e8801d3a7af5ce5f7c05c105bee815c3", size = 4742800, upload-time = "2025-02-07T13:45:14.096Z" }, + { url = "https://files.pythonhosted.org/packages/dd/f4/963247ae8c73ccc4cf2929e7162f595c81dbe17997d1d0ea77da24a217c9/fonttools-4.56.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e2cad98c94833465bcf28f51c248aaf07ca022efc6a3eba750ad9c1e0256d278", size = 4963746, upload-time = "2025-02-07T13:45:17.479Z" }, + { url = "https://files.pythonhosted.org/packages/ea/e0/46f9600c39c644b54e4420f941f75fa200d9288c9ae171e5d80918b8cbb9/fonttools-4.56.0-cp313-cp313-win32.whl", hash = "sha256:d0cb73ccf7f6d7ca8d0bc7ea8ac0a5b84969a41c56ac3ac3422a24df2680546f", size = 2140927, upload-time = "2025-02-07T13:45:21.084Z" }, + { url = "https://files.pythonhosted.org/packages/27/6d/3edda54f98a550a0473f032d8050315fbc8f1b76a0d9f3879b72ebb2cdd6/fonttools-4.56.0-cp313-cp313-win_amd64.whl", hash = "sha256:62cc1253827d1e500fde9dbe981219fea4eb000fd63402283472d38e7d8aa1c6", size = 2186709, upload-time = "2025-02-07T13:45:23.719Z" }, + { url = "https://files.pythonhosted.org/packages/bf/ff/44934a031ce5a39125415eb405b9efb76fe7f9586b75291d66ae5cbfc4e6/fonttools-4.56.0-py3-none-any.whl", hash = "sha256:1088182f68c303b50ca4dc0c82d42083d176cba37af1937e1a976a31149d4d14", size = 1089800, upload-time = "2025-02-07T13:46:26.415Z" }, ] [[package]] name = "frozenlist" -version = "1.7.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/79/b1/b64018016eeb087db503b038296fd782586432b9c077fc5c7839e9cb6ef6/frozenlist-1.7.0.tar.gz", hash = "sha256:2e310d81923c2437ea8670467121cc3e9b0f76d3043cc1d2331d56c7fb7a3a8f", size = 45078, upload-time = "2025-06-09T23:02:35.538Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/af/36/0da0a49409f6b47cc2d060dc8c9040b897b5902a8a4e37d9bc1deb11f680/frozenlist-1.7.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:cc4df77d638aa2ed703b878dd093725b72a824c3c546c076e8fdf276f78ee84a", size = 81304, upload-time = "2025-06-09T22:59:46.226Z" }, - { url = "https://files.pythonhosted.org/packages/77/f0/77c11d13d39513b298e267b22eb6cb559c103d56f155aa9a49097221f0b6/frozenlist-1.7.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:716a9973a2cc963160394f701964fe25012600f3d311f60c790400b00e568b61", size = 47735, upload-time = "2025-06-09T22:59:48.133Z" }, - { url = "https://files.pythonhosted.org/packages/37/12/9d07fa18971a44150593de56b2f2947c46604819976784bcf6ea0d5db43b/frozenlist-1.7.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:a0fd1bad056a3600047fb9462cff4c5322cebc59ebf5d0a3725e0ee78955001d", size = 46775, upload-time = "2025-06-09T22:59:49.564Z" }, - { url = "https://files.pythonhosted.org/packages/70/34/f73539227e06288fcd1f8a76853e755b2b48bca6747e99e283111c18bcd4/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3789ebc19cb811163e70fe2bd354cea097254ce6e707ae42e56f45e31e96cb8e", size = 224644, upload-time = "2025-06-09T22:59:51.35Z" }, - { url = "https://files.pythonhosted.org/packages/fb/68/c1d9c2f4a6e438e14613bad0f2973567586610cc22dcb1e1241da71de9d3/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:af369aa35ee34f132fcfad5be45fbfcde0e3a5f6a1ec0712857f286b7d20cca9", size = 222125, upload-time = "2025-06-09T22:59:52.884Z" }, - { url = "https://files.pythonhosted.org/packages/b9/d0/98e8f9a515228d708344d7c6986752be3e3192d1795f748c24bcf154ad99/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ac64b6478722eeb7a3313d494f8342ef3478dff539d17002f849101b212ef97c", size = 233455, upload-time = "2025-06-09T22:59:54.74Z" }, - { url = "https://files.pythonhosted.org/packages/79/df/8a11bcec5600557f40338407d3e5bea80376ed1c01a6c0910fcfdc4b8993/frozenlist-1.7.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f89f65d85774f1797239693cef07ad4c97fdd0639544bad9ac4b869782eb1981", size = 227339, upload-time = "2025-06-09T22:59:56.187Z" }, - { url = "https://files.pythonhosted.org/packages/50/82/41cb97d9c9a5ff94438c63cc343eb7980dac4187eb625a51bdfdb7707314/frozenlist-1.7.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:1073557c941395fdfcfac13eb2456cb8aad89f9de27bae29fabca8e563b12615", size = 212969, upload-time = "2025-06-09T22:59:57.604Z" }, - { url = "https://files.pythonhosted.org/packages/13/47/f9179ee5ee4f55629e4f28c660b3fdf2775c8bfde8f9c53f2de2d93f52a9/frozenlist-1.7.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1ed8d2fa095aae4bdc7fdd80351009a48d286635edffee66bf865e37a9125c50", size = 222862, upload-time = "2025-06-09T22:59:59.498Z" }, - { url = "https://files.pythonhosted.org/packages/1a/52/df81e41ec6b953902c8b7e3a83bee48b195cb0e5ec2eabae5d8330c78038/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:24c34bea555fe42d9f928ba0a740c553088500377448febecaa82cc3e88aa1fa", size = 222492, upload-time = "2025-06-09T23:00:01.026Z" }, - { url = "https://files.pythonhosted.org/packages/84/17/30d6ea87fa95a9408245a948604b82c1a4b8b3e153cea596421a2aef2754/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:69cac419ac6a6baad202c85aaf467b65ac860ac2e7f2ac1686dc40dbb52f6577", size = 238250, upload-time = "2025-06-09T23:00:03.401Z" }, - { url = "https://files.pythonhosted.org/packages/8f/00/ecbeb51669e3c3df76cf2ddd66ae3e48345ec213a55e3887d216eb4fbab3/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:960d67d0611f4c87da7e2ae2eacf7ea81a5be967861e0c63cf205215afbfac59", size = 218720, upload-time = "2025-06-09T23:00:05.282Z" }, - { url = "https://files.pythonhosted.org/packages/1a/c0/c224ce0e0eb31cc57f67742071bb470ba8246623c1823a7530be0e76164c/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:41be2964bd4b15bf575e5daee5a5ce7ed3115320fb3c2b71fca05582ffa4dc9e", size = 232585, upload-time = "2025-06-09T23:00:07.962Z" }, - { url = "https://files.pythonhosted.org/packages/55/3c/34cb694abf532f31f365106deebdeac9e45c19304d83cf7d51ebbb4ca4d1/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:46d84d49e00c9429238a7ce02dc0be8f6d7cd0cd405abd1bebdc991bf27c15bd", size = 234248, upload-time = "2025-06-09T23:00:09.428Z" }, - { url = "https://files.pythonhosted.org/packages/98/c0/2052d8b6cecda2e70bd81299e3512fa332abb6dcd2969b9c80dfcdddbf75/frozenlist-1.7.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:15900082e886edb37480335d9d518cec978afc69ccbc30bd18610b7c1b22a718", size = 221621, upload-time = "2025-06-09T23:00:11.32Z" }, - { url = "https://files.pythonhosted.org/packages/c5/bf/7dcebae315436903b1d98ffb791a09d674c88480c158aa171958a3ac07f0/frozenlist-1.7.0-cp310-cp310-win32.whl", hash = "sha256:400ddd24ab4e55014bba442d917203c73b2846391dd42ca5e38ff52bb18c3c5e", size = 39578, upload-time = "2025-06-09T23:00:13.526Z" }, - { url = "https://files.pythonhosted.org/packages/8f/5f/f69818f017fa9a3d24d1ae39763e29b7f60a59e46d5f91b9c6b21622f4cd/frozenlist-1.7.0-cp310-cp310-win_amd64.whl", hash = "sha256:6eb93efb8101ef39d32d50bce242c84bcbddb4f7e9febfa7b524532a239b4464", size = 43830, upload-time = "2025-06-09T23:00:14.98Z" }, - { url = "https://files.pythonhosted.org/packages/34/7e/803dde33760128acd393a27eb002f2020ddb8d99d30a44bfbaab31c5f08a/frozenlist-1.7.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:aa51e147a66b2d74de1e6e2cf5921890de6b0f4820b257465101d7f37b49fb5a", size = 82251, upload-time = "2025-06-09T23:00:16.279Z" }, - { url = "https://files.pythonhosted.org/packages/75/a9/9c2c5760b6ba45eae11334db454c189d43d34a4c0b489feb2175e5e64277/frozenlist-1.7.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:9b35db7ce1cd71d36ba24f80f0c9e7cff73a28d7a74e91fe83e23d27c7828750", size = 48183, upload-time = "2025-06-09T23:00:17.698Z" }, - { url = "https://files.pythonhosted.org/packages/47/be/4038e2d869f8a2da165f35a6befb9158c259819be22eeaf9c9a8f6a87771/frozenlist-1.7.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:34a69a85e34ff37791e94542065c8416c1afbf820b68f720452f636d5fb990cd", size = 47107, upload-time = "2025-06-09T23:00:18.952Z" }, - { url = "https://files.pythonhosted.org/packages/79/26/85314b8a83187c76a37183ceed886381a5f992975786f883472fcb6dc5f2/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4a646531fa8d82c87fe4bb2e596f23173caec9185bfbca5d583b4ccfb95183e2", size = 237333, upload-time = "2025-06-09T23:00:20.275Z" }, - { url = "https://files.pythonhosted.org/packages/1f/fd/e5b64f7d2c92a41639ffb2ad44a6a82f347787abc0c7df5f49057cf11770/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:79b2ffbba483f4ed36a0f236ccb85fbb16e670c9238313709638167670ba235f", size = 231724, upload-time = "2025-06-09T23:00:21.705Z" }, - { url = "https://files.pythonhosted.org/packages/20/fb/03395c0a43a5976af4bf7534759d214405fbbb4c114683f434dfdd3128ef/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a26f205c9ca5829cbf82bb2a84b5c36f7184c4316617d7ef1b271a56720d6b30", size = 245842, upload-time = "2025-06-09T23:00:23.148Z" }, - { url = "https://files.pythonhosted.org/packages/d0/15/c01c8e1dffdac5d9803507d824f27aed2ba76b6ed0026fab4d9866e82f1f/frozenlist-1.7.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:bcacfad3185a623fa11ea0e0634aac7b691aa925d50a440f39b458e41c561d98", size = 239767, upload-time = "2025-06-09T23:00:25.103Z" }, - { url = "https://files.pythonhosted.org/packages/14/99/3f4c6fe882c1f5514b6848aa0a69b20cb5e5d8e8f51a339d48c0e9305ed0/frozenlist-1.7.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:72c1b0fe8fe451b34f12dce46445ddf14bd2a5bcad7e324987194dc8e3a74c86", size = 224130, upload-time = "2025-06-09T23:00:27.061Z" }, - { url = "https://files.pythonhosted.org/packages/4d/83/220a374bd7b2aeba9d0725130665afe11de347d95c3620b9b82cc2fcab97/frozenlist-1.7.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:61d1a5baeaac6c0798ff6edfaeaa00e0e412d49946c53fae8d4b8e8b3566c4ae", size = 235301, upload-time = "2025-06-09T23:00:29.02Z" }, - { url = "https://files.pythonhosted.org/packages/03/3c/3e3390d75334a063181625343e8daab61b77e1b8214802cc4e8a1bb678fc/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:7edf5c043c062462f09b6820de9854bf28cc6cc5b6714b383149745e287181a8", size = 234606, upload-time = "2025-06-09T23:00:30.514Z" }, - { url = "https://files.pythonhosted.org/packages/23/1e/58232c19608b7a549d72d9903005e2d82488f12554a32de2d5fb59b9b1ba/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:d50ac7627b3a1bd2dcef6f9da89a772694ec04d9a61b66cf87f7d9446b4a0c31", size = 248372, upload-time = "2025-06-09T23:00:31.966Z" }, - { url = "https://files.pythonhosted.org/packages/c0/a4/e4a567e01702a88a74ce8a324691e62a629bf47d4f8607f24bf1c7216e7f/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:ce48b2fece5aeb45265bb7a58259f45027db0abff478e3077e12b05b17fb9da7", size = 229860, upload-time = "2025-06-09T23:00:33.375Z" }, - { url = "https://files.pythonhosted.org/packages/73/a6/63b3374f7d22268b41a9db73d68a8233afa30ed164c46107b33c4d18ecdd/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:fe2365ae915a1fafd982c146754e1de6ab3478def8a59c86e1f7242d794f97d5", size = 245893, upload-time = "2025-06-09T23:00:35.002Z" }, - { url = "https://files.pythonhosted.org/packages/6d/eb/d18b3f6e64799a79673c4ba0b45e4cfbe49c240edfd03a68be20002eaeaa/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:45a6f2fdbd10e074e8814eb98b05292f27bad7d1883afbe009d96abdcf3bc898", size = 246323, upload-time = "2025-06-09T23:00:36.468Z" }, - { url = "https://files.pythonhosted.org/packages/5a/f5/720f3812e3d06cd89a1d5db9ff6450088b8f5c449dae8ffb2971a44da506/frozenlist-1.7.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:21884e23cffabb157a9dd7e353779077bf5b8f9a58e9b262c6caad2ef5f80a56", size = 233149, upload-time = "2025-06-09T23:00:37.963Z" }, - { url = "https://files.pythonhosted.org/packages/69/68/03efbf545e217d5db8446acfd4c447c15b7c8cf4dbd4a58403111df9322d/frozenlist-1.7.0-cp311-cp311-win32.whl", hash = "sha256:284d233a8953d7b24f9159b8a3496fc1ddc00f4db99c324bd5fb5f22d8698ea7", size = 39565, upload-time = "2025-06-09T23:00:39.753Z" }, - { url = "https://files.pythonhosted.org/packages/58/17/fe61124c5c333ae87f09bb67186d65038834a47d974fc10a5fadb4cc5ae1/frozenlist-1.7.0-cp311-cp311-win_amd64.whl", hash = "sha256:387cbfdcde2f2353f19c2f66bbb52406d06ed77519ac7ee21be0232147c2592d", size = 44019, upload-time = "2025-06-09T23:00:40.988Z" }, - { url = "https://files.pythonhosted.org/packages/ef/a2/c8131383f1e66adad5f6ecfcce383d584ca94055a34d683bbb24ac5f2f1c/frozenlist-1.7.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:3dbf9952c4bb0e90e98aec1bd992b3318685005702656bc6f67c1a32b76787f2", size = 81424, upload-time = "2025-06-09T23:00:42.24Z" }, - { url = "https://files.pythonhosted.org/packages/4c/9d/02754159955088cb52567337d1113f945b9e444c4960771ea90eb73de8db/frozenlist-1.7.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:1f5906d3359300b8a9bb194239491122e6cf1444c2efb88865426f170c262cdb", size = 47952, upload-time = "2025-06-09T23:00:43.481Z" }, - { url = "https://files.pythonhosted.org/packages/01/7a/0046ef1bd6699b40acd2067ed6d6670b4db2f425c56980fa21c982c2a9db/frozenlist-1.7.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3dabd5a8f84573c8d10d8859a50ea2dec01eea372031929871368c09fa103478", size = 46688, upload-time = "2025-06-09T23:00:44.793Z" }, - { url = "https://files.pythonhosted.org/packages/d6/a2/a910bafe29c86997363fb4c02069df4ff0b5bc39d33c5198b4e9dd42d8f8/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa57daa5917f1738064f302bf2626281a1cb01920c32f711fbc7bc36111058a8", size = 243084, upload-time = "2025-06-09T23:00:46.125Z" }, - { url = "https://files.pythonhosted.org/packages/64/3e/5036af9d5031374c64c387469bfcc3af537fc0f5b1187d83a1cf6fab1639/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:c193dda2b6d49f4c4398962810fa7d7c78f032bf45572b3e04dd5249dff27e08", size = 233524, upload-time = "2025-06-09T23:00:47.73Z" }, - { url = "https://files.pythonhosted.org/packages/06/39/6a17b7c107a2887e781a48ecf20ad20f1c39d94b2a548c83615b5b879f28/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:bfe2b675cf0aaa6d61bf8fbffd3c274b3c9b7b1623beb3809df8a81399a4a9c4", size = 248493, upload-time = "2025-06-09T23:00:49.742Z" }, - { url = "https://files.pythonhosted.org/packages/be/00/711d1337c7327d88c44d91dd0f556a1c47fb99afc060ae0ef66b4d24793d/frozenlist-1.7.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8fc5d5cda37f62b262405cf9652cf0856839c4be8ee41be0afe8858f17f4c94b", size = 244116, upload-time = "2025-06-09T23:00:51.352Z" }, - { url = "https://files.pythonhosted.org/packages/24/fe/74e6ec0639c115df13d5850e75722750adabdc7de24e37e05a40527ca539/frozenlist-1.7.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b0d5ce521d1dd7d620198829b87ea002956e4319002ef0bc8d3e6d045cb4646e", size = 224557, upload-time = "2025-06-09T23:00:52.855Z" }, - { url = "https://files.pythonhosted.org/packages/8d/db/48421f62a6f77c553575201e89048e97198046b793f4a089c79a6e3268bd/frozenlist-1.7.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:488d0a7d6a0008ca0db273c542098a0fa9e7dfaa7e57f70acef43f32b3f69dca", size = 241820, upload-time = "2025-06-09T23:00:54.43Z" }, - { url = "https://files.pythonhosted.org/packages/1d/fa/cb4a76bea23047c8462976ea7b7a2bf53997a0ca171302deae9d6dd12096/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:15a7eaba63983d22c54d255b854e8108e7e5f3e89f647fc854bd77a237e767df", size = 236542, upload-time = "2025-06-09T23:00:56.409Z" }, - { url = "https://files.pythonhosted.org/packages/5d/32/476a4b5cfaa0ec94d3f808f193301debff2ea42288a099afe60757ef6282/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:1eaa7e9c6d15df825bf255649e05bd8a74b04a4d2baa1ae46d9c2d00b2ca2cb5", size = 249350, upload-time = "2025-06-09T23:00:58.468Z" }, - { url = "https://files.pythonhosted.org/packages/8d/ba/9a28042f84a6bf8ea5dbc81cfff8eaef18d78b2a1ad9d51c7bc5b029ad16/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:e4389e06714cfa9d47ab87f784a7c5be91d3934cd6e9a7b85beef808297cc025", size = 225093, upload-time = "2025-06-09T23:01:00.015Z" }, - { url = "https://files.pythonhosted.org/packages/bc/29/3a32959e68f9cf000b04e79ba574527c17e8842e38c91d68214a37455786/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:73bd45e1488c40b63fe5a7df892baf9e2a4d4bb6409a2b3b78ac1c6236178e01", size = 245482, upload-time = "2025-06-09T23:01:01.474Z" }, - { url = "https://files.pythonhosted.org/packages/80/e8/edf2f9e00da553f07f5fa165325cfc302dead715cab6ac8336a5f3d0adc2/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:99886d98e1643269760e5fe0df31e5ae7050788dd288947f7f007209b8c33f08", size = 249590, upload-time = "2025-06-09T23:01:02.961Z" }, - { url = "https://files.pythonhosted.org/packages/1c/80/9a0eb48b944050f94cc51ee1c413eb14a39543cc4f760ed12657a5a3c45a/frozenlist-1.7.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:290a172aae5a4c278c6da8a96222e6337744cd9c77313efe33d5670b9f65fc43", size = 237785, upload-time = "2025-06-09T23:01:05.095Z" }, - { url = "https://files.pythonhosted.org/packages/f3/74/87601e0fb0369b7a2baf404ea921769c53b7ae00dee7dcfe5162c8c6dbf0/frozenlist-1.7.0-cp312-cp312-win32.whl", hash = "sha256:426c7bc70e07cfebc178bc4c2bf2d861d720c4fff172181eeb4a4c41d4ca2ad3", size = 39487, upload-time = "2025-06-09T23:01:06.54Z" }, - { url = "https://files.pythonhosted.org/packages/0b/15/c026e9a9fc17585a9d461f65d8593d281fedf55fbf7eb53f16c6df2392f9/frozenlist-1.7.0-cp312-cp312-win_amd64.whl", hash = "sha256:563b72efe5da92e02eb68c59cb37205457c977aa7a449ed1b37e6939e5c47c6a", size = 43874, upload-time = "2025-06-09T23:01:07.752Z" }, - { url = "https://files.pythonhosted.org/packages/24/90/6b2cebdabdbd50367273c20ff6b57a3dfa89bd0762de02c3a1eb42cb6462/frozenlist-1.7.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:ee80eeda5e2a4e660651370ebffd1286542b67e268aa1ac8d6dbe973120ef7ee", size = 79791, upload-time = "2025-06-09T23:01:09.368Z" }, - { url = "https://files.pythonhosted.org/packages/83/2e/5b70b6a3325363293fe5fc3ae74cdcbc3e996c2a11dde2fd9f1fb0776d19/frozenlist-1.7.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:d1a81c85417b914139e3a9b995d4a1c84559afc839a93cf2cb7f15e6e5f6ed2d", size = 47165, upload-time = "2025-06-09T23:01:10.653Z" }, - { url = "https://files.pythonhosted.org/packages/f4/25/a0895c99270ca6966110f4ad98e87e5662eab416a17e7fd53c364bf8b954/frozenlist-1.7.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:cbb65198a9132ebc334f237d7b0df163e4de83fb4f2bdfe46c1e654bdb0c5d43", size = 45881, upload-time = "2025-06-09T23:01:12.296Z" }, - { url = "https://files.pythonhosted.org/packages/19/7c/71bb0bbe0832793c601fff68cd0cf6143753d0c667f9aec93d3c323f4b55/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dab46c723eeb2c255a64f9dc05b8dd601fde66d6b19cdb82b2e09cc6ff8d8b5d", size = 232409, upload-time = "2025-06-09T23:01:13.641Z" }, - { url = "https://files.pythonhosted.org/packages/c0/45/ed2798718910fe6eb3ba574082aaceff4528e6323f9a8570be0f7028d8e9/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:6aeac207a759d0dedd2e40745575ae32ab30926ff4fa49b1635def65806fddee", size = 225132, upload-time = "2025-06-09T23:01:15.264Z" }, - { url = "https://files.pythonhosted.org/packages/ba/e2/8417ae0f8eacb1d071d4950f32f229aa6bf68ab69aab797b72a07ea68d4f/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:bd8c4e58ad14b4fa7802b8be49d47993182fdd4023393899632c88fd8cd994eb", size = 237638, upload-time = "2025-06-09T23:01:16.752Z" }, - { url = "https://files.pythonhosted.org/packages/f8/b7/2ace5450ce85f2af05a871b8c8719b341294775a0a6c5585d5e6170f2ce7/frozenlist-1.7.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:04fb24d104f425da3540ed83cbfc31388a586a7696142004c577fa61c6298c3f", size = 233539, upload-time = "2025-06-09T23:01:18.202Z" }, - { url = "https://files.pythonhosted.org/packages/46/b9/6989292c5539553dba63f3c83dc4598186ab2888f67c0dc1d917e6887db6/frozenlist-1.7.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:6a5c505156368e4ea6b53b5ac23c92d7edc864537ff911d2fb24c140bb175e60", size = 215646, upload-time = "2025-06-09T23:01:19.649Z" }, - { url = "https://files.pythonhosted.org/packages/72/31/bc8c5c99c7818293458fe745dab4fd5730ff49697ccc82b554eb69f16a24/frozenlist-1.7.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8bd7eb96a675f18aa5c553eb7ddc24a43c8c18f22e1f9925528128c052cdbe00", size = 232233, upload-time = "2025-06-09T23:01:21.175Z" }, - { url = "https://files.pythonhosted.org/packages/59/52/460db4d7ba0811b9ccb85af996019f5d70831f2f5f255f7cc61f86199795/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:05579bf020096fe05a764f1f84cd104a12f78eaab68842d036772dc6d4870b4b", size = 227996, upload-time = "2025-06-09T23:01:23.098Z" }, - { url = "https://files.pythonhosted.org/packages/ba/c9/f4b39e904c03927b7ecf891804fd3b4df3db29b9e487c6418e37988d6e9d/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:376b6222d114e97eeec13d46c486facd41d4f43bab626b7c3f6a8b4e81a5192c", size = 242280, upload-time = "2025-06-09T23:01:24.808Z" }, - { url = "https://files.pythonhosted.org/packages/b8/33/3f8d6ced42f162d743e3517781566b8481322be321b486d9d262adf70bfb/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:0aa7e176ebe115379b5b1c95b4096fb1c17cce0847402e227e712c27bdb5a949", size = 217717, upload-time = "2025-06-09T23:01:26.28Z" }, - { url = "https://files.pythonhosted.org/packages/3e/e8/ad683e75da6ccef50d0ab0c2b2324b32f84fc88ceee778ed79b8e2d2fe2e/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:3fbba20e662b9c2130dc771e332a99eff5da078b2b2648153a40669a6d0e36ca", size = 236644, upload-time = "2025-06-09T23:01:27.887Z" }, - { url = "https://files.pythonhosted.org/packages/b2/14/8d19ccdd3799310722195a72ac94ddc677541fb4bef4091d8e7775752360/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:f3f4410a0a601d349dd406b5713fec59b4cee7e71678d5b17edda7f4655a940b", size = 238879, upload-time = "2025-06-09T23:01:29.524Z" }, - { url = "https://files.pythonhosted.org/packages/ce/13/c12bf657494c2fd1079a48b2db49fa4196325909249a52d8f09bc9123fd7/frozenlist-1.7.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e2cdfaaec6a2f9327bf43c933c0319a7c429058e8537c508964a133dffee412e", size = 232502, upload-time = "2025-06-09T23:01:31.287Z" }, - { url = "https://files.pythonhosted.org/packages/d7/8b/e7f9dfde869825489382bc0d512c15e96d3964180c9499efcec72e85db7e/frozenlist-1.7.0-cp313-cp313-win32.whl", hash = "sha256:5fc4df05a6591c7768459caba1b342d9ec23fa16195e744939ba5914596ae3e1", size = 39169, upload-time = "2025-06-09T23:01:35.503Z" }, - { url = "https://files.pythonhosted.org/packages/35/89/a487a98d94205d85745080a37860ff5744b9820a2c9acbcdd9440bfddf98/frozenlist-1.7.0-cp313-cp313-win_amd64.whl", hash = "sha256:52109052b9791a3e6b5d1b65f4b909703984b770694d3eb64fad124c835d7cba", size = 43219, upload-time = "2025-06-09T23:01:36.784Z" }, - { url = "https://files.pythonhosted.org/packages/56/d5/5c4cf2319a49eddd9dd7145e66c4866bdc6f3dbc67ca3d59685149c11e0d/frozenlist-1.7.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:a6f86e4193bb0e235ef6ce3dde5cbabed887e0b11f516ce8a0f4d3b33078ec2d", size = 84345, upload-time = "2025-06-09T23:01:38.295Z" }, - { url = "https://files.pythonhosted.org/packages/a4/7d/ec2c1e1dc16b85bc9d526009961953df9cec8481b6886debb36ec9107799/frozenlist-1.7.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:82d664628865abeb32d90ae497fb93df398a69bb3434463d172b80fc25b0dd7d", size = 48880, upload-time = "2025-06-09T23:01:39.887Z" }, - { url = "https://files.pythonhosted.org/packages/69/86/f9596807b03de126e11e7d42ac91e3d0b19a6599c714a1989a4e85eeefc4/frozenlist-1.7.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:912a7e8375a1c9a68325a902f3953191b7b292aa3c3fb0d71a216221deca460b", size = 48498, upload-time = "2025-06-09T23:01:41.318Z" }, - { url = "https://files.pythonhosted.org/packages/5e/cb/df6de220f5036001005f2d726b789b2c0b65f2363b104bbc16f5be8084f8/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9537c2777167488d539bc5de2ad262efc44388230e5118868e172dd4a552b146", size = 292296, upload-time = "2025-06-09T23:01:42.685Z" }, - { url = "https://files.pythonhosted.org/packages/83/1f/de84c642f17c8f851a2905cee2dae401e5e0daca9b5ef121e120e19aa825/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:f34560fb1b4c3e30ba35fa9a13894ba39e5acfc5f60f57d8accde65f46cc5e74", size = 273103, upload-time = "2025-06-09T23:01:44.166Z" }, - { url = "https://files.pythonhosted.org/packages/88/3c/c840bfa474ba3fa13c772b93070893c6e9d5c0350885760376cbe3b6c1b3/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:acd03d224b0175f5a850edc104ac19040d35419eddad04e7cf2d5986d98427f1", size = 292869, upload-time = "2025-06-09T23:01:45.681Z" }, - { url = "https://files.pythonhosted.org/packages/a6/1c/3efa6e7d5a39a1d5ef0abeb51c48fb657765794a46cf124e5aca2c7a592c/frozenlist-1.7.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f2038310bc582f3d6a09b3816ab01737d60bf7b1ec70f5356b09e84fb7408ab1", size = 291467, upload-time = "2025-06-09T23:01:47.234Z" }, - { url = "https://files.pythonhosted.org/packages/4f/00/d5c5e09d4922c395e2f2f6b79b9a20dab4b67daaf78ab92e7729341f61f6/frozenlist-1.7.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b8c05e4c8e5f36e5e088caa1bf78a687528f83c043706640a92cb76cd6999384", size = 266028, upload-time = "2025-06-09T23:01:48.819Z" }, - { url = "https://files.pythonhosted.org/packages/4e/27/72765be905619dfde25a7f33813ac0341eb6b076abede17a2e3fbfade0cb/frozenlist-1.7.0-cp313-cp313t-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:765bb588c86e47d0b68f23c1bee323d4b703218037765dcf3f25c838c6fecceb", size = 284294, upload-time = "2025-06-09T23:01:50.394Z" }, - { url = "https://files.pythonhosted.org/packages/88/67/c94103a23001b17808eb7dd1200c156bb69fb68e63fcf0693dde4cd6228c/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:32dc2e08c67d86d0969714dd484fd60ff08ff81d1a1e40a77dd34a387e6ebc0c", size = 281898, upload-time = "2025-06-09T23:01:52.234Z" }, - { url = "https://files.pythonhosted.org/packages/42/34/a3e2c00c00f9e2a9db5653bca3fec306349e71aff14ae45ecc6d0951dd24/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:c0303e597eb5a5321b4de9c68e9845ac8f290d2ab3f3e2c864437d3c5a30cd65", size = 290465, upload-time = "2025-06-09T23:01:53.788Z" }, - { url = "https://files.pythonhosted.org/packages/bb/73/f89b7fbce8b0b0c095d82b008afd0590f71ccb3dee6eee41791cf8cd25fd/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:a47f2abb4e29b3a8d0b530f7c3598badc6b134562b1a5caee867f7c62fee51e3", size = 266385, upload-time = "2025-06-09T23:01:55.769Z" }, - { url = "https://files.pythonhosted.org/packages/cd/45/e365fdb554159462ca12df54bc59bfa7a9a273ecc21e99e72e597564d1ae/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:3d688126c242a6fabbd92e02633414d40f50bb6002fa4cf995a1d18051525657", size = 288771, upload-time = "2025-06-09T23:01:57.4Z" }, - { url = "https://files.pythonhosted.org/packages/00/11/47b6117002a0e904f004d70ec5194fe9144f117c33c851e3d51c765962d0/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:4e7e9652b3d367c7bd449a727dc79d5043f48b88d0cbfd4f9f1060cf2b414104", size = 288206, upload-time = "2025-06-09T23:01:58.936Z" }, - { url = "https://files.pythonhosted.org/packages/40/37/5f9f3c3fd7f7746082ec67bcdc204db72dad081f4f83a503d33220a92973/frozenlist-1.7.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:1a85e345b4c43db8b842cab1feb41be5cc0b10a1830e6295b69d7310f99becaf", size = 282620, upload-time = "2025-06-09T23:02:00.493Z" }, - { url = "https://files.pythonhosted.org/packages/0b/31/8fbc5af2d183bff20f21aa743b4088eac4445d2bb1cdece449ae80e4e2d1/frozenlist-1.7.0-cp313-cp313t-win32.whl", hash = "sha256:3a14027124ddb70dfcee5148979998066897e79f89f64b13328595c4bdf77c81", size = 43059, upload-time = "2025-06-09T23:02:02.072Z" }, - { url = "https://files.pythonhosted.org/packages/bb/ed/41956f52105b8dbc26e457c5705340c67c8cc2b79f394b79bffc09d0e938/frozenlist-1.7.0-cp313-cp313t-win_amd64.whl", hash = "sha256:3bf8010d71d4507775f658e9823210b7427be36625b387221642725b515dcf3e", size = 47516, upload-time = "2025-06-09T23:02:03.779Z" }, - { url = "https://files.pythonhosted.org/packages/ee/45/b82e3c16be2182bff01179db177fe144d58b5dc787a7d4492c6ed8b9317f/frozenlist-1.7.0-py3-none-any.whl", hash = "sha256:9a5af342e34f7e97caf8c995864c7a396418ae2859cc6fdf1b1073020d516a7e", size = 13106, upload-time = "2025-06-09T23:02:34.204Z" }, +version = "1.5.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/8f/ed/0f4cec13a93c02c47ec32d81d11c0c1efbadf4a471e3f3ce7cad366cbbd3/frozenlist-1.5.0.tar.gz", hash = "sha256:81d5af29e61b9c8348e876d442253723928dce6433e0e76cd925cd83f1b4b817", size = 39930, upload-time = "2024-10-23T09:48:29.903Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/54/79/29d44c4af36b2b240725dce566b20f63f9b36ef267aaaa64ee7466f4f2f8/frozenlist-1.5.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:5b6a66c18b5b9dd261ca98dffcb826a525334b2f29e7caa54e182255c5f6a65a", size = 94451, upload-time = "2024-10-23T09:46:20.558Z" }, + { url = "https://files.pythonhosted.org/packages/47/47/0c999aeace6ead8a44441b4f4173e2261b18219e4ad1fe9a479871ca02fc/frozenlist-1.5.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:d1b3eb7b05ea246510b43a7e53ed1653e55c2121019a97e60cad7efb881a97bb", size = 54301, upload-time = "2024-10-23T09:46:21.759Z" }, + { url = "https://files.pythonhosted.org/packages/8d/60/107a38c1e54176d12e06e9d4b5d755b677d71d1219217cee063911b1384f/frozenlist-1.5.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:15538c0cbf0e4fa11d1e3a71f823524b0c46299aed6e10ebb4c2089abd8c3bec", size = 52213, upload-time = "2024-10-23T09:46:22.993Z" }, + { url = "https://files.pythonhosted.org/packages/17/62/594a6829ac5679c25755362a9dc93486a8a45241394564309641425d3ff6/frozenlist-1.5.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e79225373c317ff1e35f210dd5f1344ff31066ba8067c307ab60254cd3a78ad5", size = 240946, upload-time = "2024-10-23T09:46:24.661Z" }, + { url = "https://files.pythonhosted.org/packages/7e/75/6c8419d8f92c80dd0ee3f63bdde2702ce6398b0ac8410ff459f9b6f2f9cb/frozenlist-1.5.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9272fa73ca71266702c4c3e2d4a28553ea03418e591e377a03b8e3659d94fa76", size = 264608, upload-time = "2024-10-23T09:46:26.017Z" }, + { url = "https://files.pythonhosted.org/packages/88/3e/82a6f0b84bc6fb7e0be240e52863c6d4ab6098cd62e4f5b972cd31e002e8/frozenlist-1.5.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:498524025a5b8ba81695761d78c8dd7382ac0b052f34e66939c42df860b8ff17", size = 261361, upload-time = "2024-10-23T09:46:27.787Z" }, + { url = "https://files.pythonhosted.org/packages/fd/85/14e5f9ccac1b64ff2f10c927b3ffdf88772aea875882406f9ba0cec8ad84/frozenlist-1.5.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:92b5278ed9d50fe610185ecd23c55d8b307d75ca18e94c0e7de328089ac5dcba", size = 231649, upload-time = "2024-10-23T09:46:28.992Z" }, + { url = "https://files.pythonhosted.org/packages/ee/59/928322800306f6529d1852323014ee9008551e9bb027cc38d276cbc0b0e7/frozenlist-1.5.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7f3c8c1dacd037df16e85227bac13cca58c30da836c6f936ba1df0c05d046d8d", size = 241853, upload-time = "2024-10-23T09:46:30.211Z" }, + { url = "https://files.pythonhosted.org/packages/7d/bd/e01fa4f146a6f6c18c5d34cab8abdc4013774a26c4ff851128cd1bd3008e/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:f2ac49a9bedb996086057b75bf93538240538c6d9b38e57c82d51f75a73409d2", size = 243652, upload-time = "2024-10-23T09:46:31.758Z" }, + { url = "https://files.pythonhosted.org/packages/a5/bd/e4771fd18a8ec6757033f0fa903e447aecc3fbba54e3630397b61596acf0/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:e66cc454f97053b79c2ab09c17fbe3c825ea6b4de20baf1be28919460dd7877f", size = 241734, upload-time = "2024-10-23T09:46:33.044Z" }, + { url = "https://files.pythonhosted.org/packages/21/13/c83821fa5544af4f60c5d3a65d054af3213c26b14d3f5f48e43e5fb48556/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:5a3ba5f9a0dfed20337d3e966dc359784c9f96503674c2faf015f7fe8e96798c", size = 260959, upload-time = "2024-10-23T09:46:34.916Z" }, + { url = "https://files.pythonhosted.org/packages/71/f3/1f91c9a9bf7ed0e8edcf52698d23f3c211d8d00291a53c9f115ceb977ab1/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:6321899477db90bdeb9299ac3627a6a53c7399c8cd58d25da094007402b039ab", size = 262706, upload-time = "2024-10-23T09:46:36.159Z" }, + { url = "https://files.pythonhosted.org/packages/4c/22/4a256fdf5d9bcb3ae32622c796ee5ff9451b3a13a68cfe3f68e2c95588ce/frozenlist-1.5.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:76e4753701248476e6286f2ef492af900ea67d9706a0155335a40ea21bf3b2f5", size = 250401, upload-time = "2024-10-23T09:46:37.327Z" }, + { url = "https://files.pythonhosted.org/packages/af/89/c48ebe1f7991bd2be6d5f4ed202d94960c01b3017a03d6954dd5fa9ea1e8/frozenlist-1.5.0-cp310-cp310-win32.whl", hash = "sha256:977701c081c0241d0955c9586ffdd9ce44f7a7795df39b9151cd9a6fd0ce4cfb", size = 45498, upload-time = "2024-10-23T09:46:38.552Z" }, + { url = "https://files.pythonhosted.org/packages/28/2f/cc27d5f43e023d21fe5c19538e08894db3d7e081cbf582ad5ed366c24446/frozenlist-1.5.0-cp310-cp310-win_amd64.whl", hash = "sha256:189f03b53e64144f90990d29a27ec4f7997d91ed3d01b51fa39d2dbe77540fd4", size = 51622, upload-time = "2024-10-23T09:46:39.513Z" }, + { url = "https://files.pythonhosted.org/packages/79/43/0bed28bf5eb1c9e4301003b74453b8e7aa85fb293b31dde352aac528dafc/frozenlist-1.5.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:fd74520371c3c4175142d02a976aee0b4cb4a7cc912a60586ffd8d5929979b30", size = 94987, upload-time = "2024-10-23T09:46:40.487Z" }, + { url = "https://files.pythonhosted.org/packages/bb/bf/b74e38f09a246e8abbe1e90eb65787ed745ccab6eaa58b9c9308e052323d/frozenlist-1.5.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2f3f7a0fbc219fb4455264cae4d9f01ad41ae6ee8524500f381de64ffaa077d5", size = 54584, upload-time = "2024-10-23T09:46:41.463Z" }, + { url = "https://files.pythonhosted.org/packages/2c/31/ab01375682f14f7613a1ade30149f684c84f9b8823a4391ed950c8285656/frozenlist-1.5.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:f47c9c9028f55a04ac254346e92977bf0f166c483c74b4232bee19a6697e4778", size = 52499, upload-time = "2024-10-23T09:46:42.451Z" }, + { url = "https://files.pythonhosted.org/packages/98/a8/d0ac0b9276e1404f58fec3ab6e90a4f76b778a49373ccaf6a563f100dfbc/frozenlist-1.5.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0996c66760924da6e88922756d99b47512a71cfd45215f3570bf1e0b694c206a", size = 276357, upload-time = "2024-10-23T09:46:44.166Z" }, + { url = "https://files.pythonhosted.org/packages/ad/c9/c7761084fa822f07dac38ac29f841d4587570dd211e2262544aa0b791d21/frozenlist-1.5.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a2fe128eb4edeabe11896cb6af88fca5346059f6c8d807e3b910069f39157869", size = 287516, upload-time = "2024-10-23T09:46:45.369Z" }, + { url = "https://files.pythonhosted.org/packages/a1/ff/cd7479e703c39df7bdab431798cef89dc75010d8aa0ca2514c5b9321db27/frozenlist-1.5.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1a8ea951bbb6cacd492e3948b8da8c502a3f814f5d20935aae74b5df2b19cf3d", size = 283131, upload-time = "2024-10-23T09:46:46.654Z" }, + { url = "https://files.pythonhosted.org/packages/59/a0/370941beb47d237eca4fbf27e4e91389fd68699e6f4b0ebcc95da463835b/frozenlist-1.5.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:de537c11e4aa01d37db0d403b57bd6f0546e71a82347a97c6a9f0dcc532b3a45", size = 261320, upload-time = "2024-10-23T09:46:47.825Z" }, + { url = "https://files.pythonhosted.org/packages/b8/5f/c10123e8d64867bc9b4f2f510a32042a306ff5fcd7e2e09e5ae5100ee333/frozenlist-1.5.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9c2623347b933fcb9095841f1cc5d4ff0b278addd743e0e966cb3d460278840d", size = 274877, upload-time = "2024-10-23T09:46:48.989Z" }, + { url = "https://files.pythonhosted.org/packages/fa/79/38c505601ae29d4348f21706c5d89755ceded02a745016ba2f58bd5f1ea6/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cee6798eaf8b1416ef6909b06f7dc04b60755206bddc599f52232606e18179d3", size = 269592, upload-time = "2024-10-23T09:46:50.235Z" }, + { url = "https://files.pythonhosted.org/packages/19/e2/39f3a53191b8204ba9f0bb574b926b73dd2efba2a2b9d2d730517e8f7622/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:f5f9da7f5dbc00a604fe74aa02ae7c98bcede8a3b8b9666f9f86fc13993bc71a", size = 265934, upload-time = "2024-10-23T09:46:51.829Z" }, + { url = "https://files.pythonhosted.org/packages/d5/c9/3075eb7f7f3a91f1a6b00284af4de0a65a9ae47084930916f5528144c9dd/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:90646abbc7a5d5c7c19461d2e3eeb76eb0b204919e6ece342feb6032c9325ae9", size = 283859, upload-time = "2024-10-23T09:46:52.947Z" }, + { url = "https://files.pythonhosted.org/packages/05/f5/549f44d314c29408b962fa2b0e69a1a67c59379fb143b92a0a065ffd1f0f/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:bdac3c7d9b705d253b2ce370fde941836a5f8b3c5c2b8fd70940a3ea3af7f4f2", size = 287560, upload-time = "2024-10-23T09:46:54.162Z" }, + { url = "https://files.pythonhosted.org/packages/9d/f8/cb09b3c24a3eac02c4c07a9558e11e9e244fb02bf62c85ac2106d1eb0c0b/frozenlist-1.5.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:03d33c2ddbc1816237a67f66336616416e2bbb6beb306e5f890f2eb22b959cdf", size = 277150, upload-time = "2024-10-23T09:46:55.361Z" }, + { url = "https://files.pythonhosted.org/packages/37/48/38c2db3f54d1501e692d6fe058f45b6ad1b358d82cd19436efab80cfc965/frozenlist-1.5.0-cp311-cp311-win32.whl", hash = "sha256:237f6b23ee0f44066219dae14c70ae38a63f0440ce6750f868ee08775073f942", size = 45244, upload-time = "2024-10-23T09:46:56.578Z" }, + { url = "https://files.pythonhosted.org/packages/ca/8c/2ddffeb8b60a4bce3b196c32fcc30d8830d4615e7b492ec2071da801b8ad/frozenlist-1.5.0-cp311-cp311-win_amd64.whl", hash = "sha256:0cc974cc93d32c42e7b0f6cf242a6bd941c57c61b618e78b6c0a96cb72788c1d", size = 51634, upload-time = "2024-10-23T09:46:57.6Z" }, + { url = "https://files.pythonhosted.org/packages/79/73/fa6d1a96ab7fd6e6d1c3500700963eab46813847f01ef0ccbaa726181dd5/frozenlist-1.5.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:31115ba75889723431aa9a4e77d5f398f5cf976eea3bdf61749731f62d4a4a21", size = 94026, upload-time = "2024-10-23T09:46:58.601Z" }, + { url = "https://files.pythonhosted.org/packages/ab/04/ea8bf62c8868b8eada363f20ff1b647cf2e93377a7b284d36062d21d81d1/frozenlist-1.5.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:7437601c4d89d070eac8323f121fcf25f88674627505334654fd027b091db09d", size = 54150, upload-time = "2024-10-23T09:46:59.608Z" }, + { url = "https://files.pythonhosted.org/packages/d0/9a/8e479b482a6f2070b26bda572c5e6889bb3ba48977e81beea35b5ae13ece/frozenlist-1.5.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:7948140d9f8ece1745be806f2bfdf390127cf1a763b925c4a805c603df5e697e", size = 51927, upload-time = "2024-10-23T09:47:00.625Z" }, + { url = "https://files.pythonhosted.org/packages/e3/12/2aad87deb08a4e7ccfb33600871bbe8f0e08cb6d8224371387f3303654d7/frozenlist-1.5.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:feeb64bc9bcc6b45c6311c9e9b99406660a9c05ca8a5b30d14a78555088b0b3a", size = 282647, upload-time = "2024-10-23T09:47:01.992Z" }, + { url = "https://files.pythonhosted.org/packages/77/f2/07f06b05d8a427ea0060a9cef6e63405ea9e0d761846b95ef3fb3be57111/frozenlist-1.5.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:683173d371daad49cffb8309779e886e59c2f369430ad28fe715f66d08d4ab1a", size = 289052, upload-time = "2024-10-23T09:47:04.039Z" }, + { url = "https://files.pythonhosted.org/packages/bd/9f/8bf45a2f1cd4aa401acd271b077989c9267ae8463e7c8b1eb0d3f561b65e/frozenlist-1.5.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:7d57d8f702221405a9d9b40f9da8ac2e4a1a8b5285aac6100f3393675f0a85ee", size = 291719, upload-time = "2024-10-23T09:47:05.58Z" }, + { url = "https://files.pythonhosted.org/packages/41/d1/1f20fd05a6c42d3868709b7604c9f15538a29e4f734c694c6bcfc3d3b935/frozenlist-1.5.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:30c72000fbcc35b129cb09956836c7d7abf78ab5416595e4857d1cae8d6251a6", size = 267433, upload-time = "2024-10-23T09:47:07.807Z" }, + { url = "https://files.pythonhosted.org/packages/af/f2/64b73a9bb86f5a89fb55450e97cd5c1f84a862d4ff90d9fd1a73ab0f64a5/frozenlist-1.5.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:000a77d6034fbad9b6bb880f7ec073027908f1b40254b5d6f26210d2dab1240e", size = 283591, upload-time = "2024-10-23T09:47:09.645Z" }, + { url = "https://files.pythonhosted.org/packages/29/e2/ffbb1fae55a791fd6c2938dd9ea779509c977435ba3940b9f2e8dc9d5316/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:5d7f5a50342475962eb18b740f3beecc685a15b52c91f7d975257e13e029eca9", size = 273249, upload-time = "2024-10-23T09:47:10.808Z" }, + { url = "https://files.pythonhosted.org/packages/2e/6e/008136a30798bb63618a114b9321b5971172a5abddff44a100c7edc5ad4f/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:87f724d055eb4785d9be84e9ebf0f24e392ddfad00b3fe036e43f489fafc9039", size = 271075, upload-time = "2024-10-23T09:47:11.938Z" }, + { url = "https://files.pythonhosted.org/packages/ae/f0/4e71e54a026b06724cec9b6c54f0b13a4e9e298cc8db0f82ec70e151f5ce/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:6e9080bb2fb195a046e5177f10d9d82b8a204c0736a97a153c2466127de87784", size = 285398, upload-time = "2024-10-23T09:47:14.071Z" }, + { url = "https://files.pythonhosted.org/packages/4d/36/70ec246851478b1c0b59f11ef8ade9c482ff447c1363c2bd5fad45098b12/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:9b93d7aaa36c966fa42efcaf716e6b3900438632a626fb09c049f6a2f09fc631", size = 294445, upload-time = "2024-10-23T09:47:15.318Z" }, + { url = "https://files.pythonhosted.org/packages/37/e0/47f87544055b3349b633a03c4d94b405956cf2437f4ab46d0928b74b7526/frozenlist-1.5.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:52ef692a4bc60a6dd57f507429636c2af8b6046db8b31b18dac02cbc8f507f7f", size = 280569, upload-time = "2024-10-23T09:47:17.149Z" }, + { url = "https://files.pythonhosted.org/packages/f9/7c/490133c160fb6b84ed374c266f42800e33b50c3bbab1652764e6e1fc498a/frozenlist-1.5.0-cp312-cp312-win32.whl", hash = "sha256:29d94c256679247b33a3dc96cce0f93cbc69c23bf75ff715919332fdbb6a32b8", size = 44721, upload-time = "2024-10-23T09:47:19.012Z" }, + { url = "https://files.pythonhosted.org/packages/b1/56/4e45136ffc6bdbfa68c29ca56ef53783ef4c2fd395f7cbf99a2624aa9aaa/frozenlist-1.5.0-cp312-cp312-win_amd64.whl", hash = "sha256:8969190d709e7c48ea386db202d708eb94bdb29207a1f269bab1196ce0dcca1f", size = 51329, upload-time = "2024-10-23T09:47:20.177Z" }, + { url = "https://files.pythonhosted.org/packages/da/3b/915f0bca8a7ea04483622e84a9bd90033bab54bdf485479556c74fd5eaf5/frozenlist-1.5.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:7a1a048f9215c90973402e26c01d1cff8a209e1f1b53f72b95c13db61b00f953", size = 91538, upload-time = "2024-10-23T09:47:21.176Z" }, + { url = "https://files.pythonhosted.org/packages/c7/d1/a7c98aad7e44afe5306a2b068434a5830f1470675f0e715abb86eb15f15b/frozenlist-1.5.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:dd47a5181ce5fcb463b5d9e17ecfdb02b678cca31280639255ce9d0e5aa67af0", size = 52849, upload-time = "2024-10-23T09:47:22.439Z" }, + { url = "https://files.pythonhosted.org/packages/3a/c8/76f23bf9ab15d5f760eb48701909645f686f9c64fbb8982674c241fbef14/frozenlist-1.5.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1431d60b36d15cda188ea222033eec8e0eab488f39a272461f2e6d9e1a8e63c2", size = 50583, upload-time = "2024-10-23T09:47:23.44Z" }, + { url = "https://files.pythonhosted.org/packages/1f/22/462a3dd093d11df623179d7754a3b3269de3b42de2808cddef50ee0f4f48/frozenlist-1.5.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6482a5851f5d72767fbd0e507e80737f9c8646ae7fd303def99bfe813f76cf7f", size = 265636, upload-time = "2024-10-23T09:47:24.82Z" }, + { url = "https://files.pythonhosted.org/packages/80/cf/e075e407fc2ae7328155a1cd7e22f932773c8073c1fc78016607d19cc3e5/frozenlist-1.5.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:44c49271a937625619e862baacbd037a7ef86dd1ee215afc298a417ff3270608", size = 270214, upload-time = "2024-10-23T09:47:26.156Z" }, + { url = "https://files.pythonhosted.org/packages/a1/58/0642d061d5de779f39c50cbb00df49682832923f3d2ebfb0fedf02d05f7f/frozenlist-1.5.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:12f78f98c2f1c2429d42e6a485f433722b0061d5c0b0139efa64f396efb5886b", size = 273905, upload-time = "2024-10-23T09:47:27.741Z" }, + { url = "https://files.pythonhosted.org/packages/ab/66/3fe0f5f8f2add5b4ab7aa4e199f767fd3b55da26e3ca4ce2cc36698e50c4/frozenlist-1.5.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ce3aa154c452d2467487765e3adc730a8c153af77ad84096bc19ce19a2400840", size = 250542, upload-time = "2024-10-23T09:47:28.938Z" }, + { url = "https://files.pythonhosted.org/packages/f6/b8/260791bde9198c87a465224e0e2bb62c4e716f5d198fc3a1dacc4895dbd1/frozenlist-1.5.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9b7dc0c4338e6b8b091e8faf0db3168a37101943e687f373dce00959583f7439", size = 267026, upload-time = "2024-10-23T09:47:30.283Z" }, + { url = "https://files.pythonhosted.org/packages/2e/a4/3d24f88c527f08f8d44ade24eaee83b2627793fa62fa07cbb7ff7a2f7d42/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:45e0896250900b5aa25180f9aec243e84e92ac84bd4a74d9ad4138ef3f5c97de", size = 257690, upload-time = "2024-10-23T09:47:32.388Z" }, + { url = "https://files.pythonhosted.org/packages/de/9a/d311d660420b2beeff3459b6626f2ab4fb236d07afbdac034a4371fe696e/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:561eb1c9579d495fddb6da8959fd2a1fca2c6d060d4113f5844b433fc02f2641", size = 253893, upload-time = "2024-10-23T09:47:34.274Z" }, + { url = "https://files.pythonhosted.org/packages/c6/23/e491aadc25b56eabd0f18c53bb19f3cdc6de30b2129ee0bc39cd387cd560/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:df6e2f325bfee1f49f81aaac97d2aa757c7646534a06f8f577ce184afe2f0a9e", size = 267006, upload-time = "2024-10-23T09:47:35.499Z" }, + { url = "https://files.pythonhosted.org/packages/08/c4/ab918ce636a35fb974d13d666dcbe03969592aeca6c3ab3835acff01f79c/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:140228863501b44b809fb39ec56b5d4071f4d0aa6d216c19cbb08b8c5a7eadb9", size = 276157, upload-time = "2024-10-23T09:47:37.522Z" }, + { url = "https://files.pythonhosted.org/packages/c0/29/3b7a0bbbbe5a34833ba26f686aabfe982924adbdcafdc294a7a129c31688/frozenlist-1.5.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:7707a25d6a77f5d27ea7dc7d1fc608aa0a478193823f88511ef5e6b8a48f9d03", size = 264642, upload-time = "2024-10-23T09:47:38.75Z" }, + { url = "https://files.pythonhosted.org/packages/ab/42/0595b3dbffc2e82d7fe658c12d5a5bafcd7516c6bf2d1d1feb5387caa9c1/frozenlist-1.5.0-cp313-cp313-win32.whl", hash = "sha256:31a9ac2b38ab9b5a8933b693db4939764ad3f299fcaa931a3e605bc3460e693c", size = 44914, upload-time = "2024-10-23T09:47:40.145Z" }, + { url = "https://files.pythonhosted.org/packages/17/c4/b7db1206a3fea44bf3b838ca61deb6f74424a8a5db1dd53ecb21da669be6/frozenlist-1.5.0-cp313-cp313-win_amd64.whl", hash = "sha256:11aabdd62b8b9c4b84081a3c246506d1cddd2dd93ff0ad53ede5defec7886b28", size = 51167, upload-time = "2024-10-23T09:47:41.812Z" }, + { url = "https://files.pythonhosted.org/packages/c6/c8/a5be5b7550c10858fcf9b0ea054baccab474da77d37f1e828ce043a3a5d4/frozenlist-1.5.0-py3-none-any.whl", hash = "sha256:d994863bba198a4a518b467bb971c56e1db3f180a25c6cf7bb1949c267f748c3", size = 11901, upload-time = "2024-10-23T09:48:28.851Z" }, ] [[package]] name = "fsspec" -version = "2025.7.0" +version = "2025.3.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/8b/02/0835e6ab9cfc03916fe3f78c0956cfcdb6ff2669ffa6651065d5ebf7fc98/fsspec-2025.7.0.tar.gz", hash = "sha256:786120687ffa54b8283d942929540d8bc5ccfa820deb555a2b5d0ed2b737bf58", size = 304432, upload-time = "2025-07-15T16:05:21.19Z" } +sdist = { url = "https://files.pythonhosted.org/packages/34/f4/5721faf47b8c499e776bc34c6a8fc17efdf7fdef0b00f398128bc5dcb4ac/fsspec-2025.3.0.tar.gz", hash = "sha256:a935fd1ea872591f2b5148907d103488fc523295e6c64b835cfad8c3eca44972", size = 298491, upload-time = "2025-03-07T21:47:56.461Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/2f/e0/014d5d9d7a4564cf1c40b5039bc882db69fd881111e03ab3657ac0b218e2/fsspec-2025.7.0-py3-none-any.whl", hash = "sha256:8b012e39f63c7d5f10474de957f3ab793b47b45ae7d39f2fb735f8bbe25c0e21", size = 199597, upload-time = "2025-07-15T16:05:19.529Z" }, + { url = "https://files.pythonhosted.org/packages/56/53/eb690efa8513166adef3e0669afd31e95ffde69fb3c52ec2ac7223ed6018/fsspec-2025.3.0-py3-none-any.whl", hash = "sha256:efb87af3efa9103f94ca91a7f8cb7a4df91af9f74fc106c9c7ea0efd7277c1b3", size = 193615, upload-time = "2025-03-07T21:47:54.809Z" }, ] [package.optional-dependencies] @@ -961,14 +843,14 @@ wheels = [ [[package]] name = "gitpython" -version = "3.1.45" +version = "3.1.44" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "gitdb" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/9a/c8/dd58967d119baab745caec2f9d853297cec1989ec1d63f677d3880632b88/gitpython-3.1.45.tar.gz", hash = "sha256:85b0ee964ceddf211c41b9f27a49086010a190fd8132a24e21f362a4b36a791c", size = 215076, upload-time = "2025-07-24T03:45:54.871Z" } +sdist = { url = "https://files.pythonhosted.org/packages/c0/89/37df0b71473153574a5cdef8f242de422a0f5d26d7a9e231e6f169b4ad14/gitpython-3.1.44.tar.gz", hash = "sha256:c87e30b26253bf5418b01b0660f818967f3c503193838337fe5e573331249269", size = 214196, upload-time = "2025-01-02T07:32:43.59Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/01/61/d4b89fec821f72385526e1b9d9a3a0385dda4a72b206d28049e2c7cd39b8/gitpython-3.1.45-py3-none-any.whl", hash = "sha256:8908cb2e02fb3b93b7eb0f2827125cb699869470432cc885f019b8fd0fccff77", size = 208168, upload-time = "2025-07-24T03:45:52.517Z" }, + { url = "https://files.pythonhosted.org/packages/1d/9a/4114a9057db2f1462d5c8f8390ab7383925fe1ac012eaa42402ad65c2963/GitPython-3.1.44-py3-none-any.whl", hash = "sha256:9e0e10cda9bed1ee64bc9a6de50e7e38a9c9943241cd7f585f6df3ed28011110", size = 207599, upload-time = "2025-01-02T07:32:40.731Z" }, ] [[package]] @@ -1107,7 +989,7 @@ wheels = [ [[package]] name = "huggingface-hub" -version = "0.34.3" +version = "0.34.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "filelock" }, @@ -1119,9 +1001,9 @@ dependencies = [ { name = "tqdm" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/91/b4/e6b465eca5386b52cf23cb6df8644ad318a6b0e12b4b96a7e0be09cbfbcc/huggingface_hub-0.34.3.tar.gz", hash = "sha256:d58130fd5aa7408480681475491c0abd7e835442082fbc3ef4d45b6c39f83853", size = 456800, upload-time = "2025-07-29T08:38:53.885Z" } +sdist = { url = "https://files.pythonhosted.org/packages/22/cd/841bc8e0550d69f632a15cdd70004e95ba92cd0fbe13087d6669e2bb5f44/huggingface_hub-0.34.1.tar.gz", hash = "sha256:6978ed89ef981de3c78b75bab100a214843be1cc9d24f8e9c0dc4971808ef1b1", size = 456783, upload-time = "2025-07-25T14:54:54.758Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/59/a8/4677014e771ed1591a87b63a2392ce6923baf807193deef302dcfde17542/huggingface_hub-0.34.3-py3-none-any.whl", hash = "sha256:5444550099e2d86e68b2898b09e85878fbd788fc2957b506c6a79ce060e39492", size = 558847, upload-time = "2025-07-29T08:38:51.904Z" }, + { url = "https://files.pythonhosted.org/packages/8c/cf/dd53c0132f50f258b06dd37a4616817b1f1f6a6b38382c06effd04bb6881/huggingface_hub-0.34.1-py3-none-any.whl", hash = "sha256:60d843dcb7bc335145b20e7d2f1dfe93910f6787b2b38a936fb772ce2a83757c", size = 558788, upload-time = "2025-07-25T14:54:52.957Z" }, ] [[package]] @@ -1140,11 +1022,11 @@ wheels = [ [[package]] name = "identify" -version = "2.6.12" +version = "2.6.9" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/a2/88/d193a27416618628a5eea64e3223acd800b40749a96ffb322a9b55a49ed1/identify-2.6.12.tar.gz", hash = "sha256:d8de45749f1efb108badef65ee8386f0f7bb19a7f26185f74de6367bffbaf0e6", size = 99254, upload-time = "2025-05-23T20:37:53.3Z" } +sdist = { url = "https://files.pythonhosted.org/packages/9b/98/a71ab060daec766acc30fb47dfca219d03de34a70d616a79a38c6066c5bf/identify-2.6.9.tar.gz", hash = "sha256:d40dfe3142a1421d8518e3d3985ef5ac42890683e32306ad614a29490abeb6bf", size = 99249, upload-time = "2025-03-08T15:54:13.632Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/7a/cd/18f8da995b658420625f7ef13f037be53ae04ec5ad33f9b718240dcfd48c/identify-2.6.12-py2.py3-none-any.whl", hash = "sha256:ad9672d5a72e0d2ff7c5c8809b62dfa60458626352fb0eb7b55e69bdc45334a2", size = 99145, upload-time = "2025-05-23T20:37:51.495Z" }, + { url = "https://files.pythonhosted.org/packages/07/ce/0845144ed1f0e25db5e7a79c2354c1da4b5ce392b8966449d5db8dca18f1/identify-2.6.9-py2.py3-none-any.whl", hash = "sha256:c98b4322da415a8e5a70ff6e51fbc2d2932c015532d77e9f8537b4ba7813b150", size = 99101, upload-time = "2025-03-08T15:54:12.026Z" }, ] [[package]] @@ -1167,14 +1049,14 @@ wheels = [ [[package]] name = "ipykernel" -version = "6.30.1" +version = "6.29.5" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "appnope", marker = "sys_platform == 'darwin'" }, { name = "comm" }, { name = "debugpy" }, - { name = "ipython", version = "8.37.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "ipython", version = "9.4.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "ipython", version = "8.34.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "ipython", version = "9.0.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "jupyter-client" }, { name = "jupyter-core" }, { name = "matplotlib-inline" }, @@ -1185,14 +1067,14 @@ dependencies = [ { name = "tornado" }, { name = "traitlets" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/bb/76/11082e338e0daadc89c8ff866185de11daf67d181901038f9e139d109761/ipykernel-6.30.1.tar.gz", hash = "sha256:6abb270161896402e76b91394fcdce5d1be5d45f456671e5080572f8505be39b", size = 166260, upload-time = "2025-08-04T15:47:35.018Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e9/5c/67594cb0c7055dc50814b21731c22a601101ea3b1b50a9a1b090e11f5d0f/ipykernel-6.29.5.tar.gz", hash = "sha256:f093a22c4a40f8828f8e330a9c297cb93dcab13bd9678ded6de8e5cf81c56215", size = 163367, upload-time = "2024-07-01T14:07:22.543Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/fc/c7/b445faca8deb954fe536abebff4ece5b097b923de482b26e78448c89d1dd/ipykernel-6.30.1-py3-none-any.whl", hash = "sha256:aa6b9fb93dca949069d8b85b6c79b2518e32ac583ae9c7d37c51d119e18b3fb4", size = 117484, upload-time = "2025-08-04T15:47:32.622Z" }, + { url = "https://files.pythonhosted.org/packages/94/5c/368ae6c01c7628438358e6d337c19b05425727fbb221d2a3c4303c372f42/ipykernel-6.29.5-py3-none-any.whl", hash = "sha256:afdb66ba5aa354b09b91379bac28ae4afebbb30e8b39510c9690afb7a10421b5", size = 117173, upload-time = "2024-07-01T14:07:19.603Z" }, ] [[package]] name = "ipython" -version = "8.37.0" +version = "8.34.0" source = { registry = "https://pypi.org/simple" } resolution-markers = [ "python_full_version < '3.11' and sys_platform == 'linux'", @@ -1211,14 +1093,14 @@ dependencies = [ { name = "traitlets", marker = "python_full_version < '3.11'" }, { name = "typing-extensions", marker = "python_full_version < '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/85/31/10ac88f3357fc276dc8a64e8880c82e80e7459326ae1d0a211b40abf6665/ipython-8.37.0.tar.gz", hash = "sha256:ca815841e1a41a1e6b73a0b08f3038af9b2252564d01fc405356d34033012216", size = 5606088, upload-time = "2025-05-31T16:39:09.613Z" } +sdist = { url = "https://files.pythonhosted.org/packages/13/18/1a60aa62e9d272fcd7e658a89e1c148da10e1a5d38edcbcd834b52ca7492/ipython-8.34.0.tar.gz", hash = "sha256:c31d658e754673ecc6514583e7dda8069e47136eb62458816b7d1e6625948b5a", size = 5508477, upload-time = "2025-03-08T13:43:17.591Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/91/d0/274fbf7b0b12643cbbc001ce13e6a5b1607ac4929d1b11c72460152c9fc3/ipython-8.37.0-py3-none-any.whl", hash = "sha256:ed87326596b878932dbcb171e3e698845434d8c61b8d8cd474bf663041a9dcf2", size = 831864, upload-time = "2025-05-31T16:39:06.38Z" }, + { url = "https://files.pythonhosted.org/packages/04/78/45615356bb973904856808183ae2a5fba1f360e9d682314d79766f4b88f2/ipython-8.34.0-py3-none-any.whl", hash = "sha256:0419883fa46e0baa182c5d50ebb8d6b49df1889fdb70750ad6d8cfe678eda6e3", size = 826731, upload-time = "2025-03-08T13:43:15.004Z" }, ] [[package]] name = "ipython" -version = "9.4.0" +version = "9.0.2" source = { registry = "https://pypi.org/simple" } resolution-markers = [ "python_full_version >= '3.13' and sys_platform == 'linux'", @@ -1241,9 +1123,9 @@ dependencies = [ { name = "traitlets", marker = "python_full_version >= '3.11'" }, { name = "typing-extensions", marker = "python_full_version == '3.11.*'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/54/80/406f9e3bde1c1fd9bf5a0be9d090f8ae623e401b7670d8f6fdf2ab679891/ipython-9.4.0.tar.gz", hash = "sha256:c033c6d4e7914c3d9768aabe76bbe87ba1dc66a92a05db6bfa1125d81f2ee270", size = 4385338, upload-time = "2025-07-01T11:11:30.606Z" } +sdist = { url = "https://files.pythonhosted.org/packages/7d/ce/012a0f40ca58a966f87a6e894d6828e2817657cbdf522b02a5d3a87d92ce/ipython-9.0.2.tar.gz", hash = "sha256:ec7b479e3e5656bf4f58c652c120494df1820f4f28f522fb7ca09e213c2aab52", size = 4366102, upload-time = "2025-03-08T15:04:52.885Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/63/f8/0031ee2b906a15a33d6bfc12dd09c3dfa966b3cb5b284ecfb7549e6ac3c4/ipython-9.4.0-py3-none-any.whl", hash = "sha256:25850f025a446d9b359e8d296ba175a36aedd32e83ca9b5060430fe16801f066", size = 611021, upload-time = "2025-07-01T11:11:27.85Z" }, + { url = "https://files.pythonhosted.org/packages/20/3a/917cb9e72f4e1a4ea13c862533205ae1319bd664119189ee5cc9e4e95ebf/ipython-9.0.2-py3-none-any.whl", hash = "sha256:143ef3ea6fb1e1bffb4c74b114051de653ffb7737a3f7ab1670e657ca6ae8c44", size = 600524, upload-time = "2025-03-08T15:04:50.667Z" }, ] [[package]] @@ -1270,12 +1152,11 @@ dependencies = [ { name = "lightning" }, { name = "lovelyplots" }, { name = "matplotlib" }, - { name = "mdtraj", version = "1.10.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "mdtraj", version = "1.11.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "mdtraj" }, { name = "ninja" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "omegaconf" }, + { name = "optree" }, { name = "orb-models" }, { name = "pandas" }, { name = "plotly" }, @@ -1286,8 +1167,7 @@ dependencies = [ { name = "rdkit" }, { name = "requests" }, { name = "s3fs", extra = ["boto3"] }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "scipy" }, { name = "statsmodels" }, { name = "tabulate" }, { name = "torch" }, @@ -1299,11 +1179,18 @@ dependencies = [ { name = "wandb" }, ] +[package.optional-dependencies] +analysis = [ + { name = "polars" }, + { name = "pyarrow" }, + { name = "seaborn" }, +] + [package.dev-dependencies] dev = [ { name = "ipykernel" }, - { name = "ipython", version = "8.37.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "ipython", version = "9.4.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "ipython", version = "8.34.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "ipython", version = "9.0.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, { name = "nbstripout" }, { name = "pre-commit" }, { name = "pytest" }, @@ -1314,7 +1201,7 @@ dev = [ requires-dist = [ { name = "ase", specifier = ">=3.23.0" }, { name = "e3nn", specifier = ">=0.5.6" }, - { name = "e3tools", specifier = ">=0.1.1" }, + { name = "e3tools", specifier = ">=0.1.2" }, { name = "einops", specifier = ">=0.8.0" }, { name = "hydra-core", specifier = ">=1.3.2" }, { name = "lightning", specifier = ">=2.4.0" }, @@ -1324,17 +1211,21 @@ requires-dist = [ { name = "ninja", specifier = ">=1.11.1.3" }, { name = "numpy", specifier = ">=2" }, { name = "omegaconf", specifier = ">=2.3.0" }, + { name = "optree", specifier = ">=0.17.0" }, { name = "orb-models", specifier = ">=0.5.4" }, { name = "pandas", specifier = ">=2.1.0" }, { name = "plotly", specifier = ">=5.24.1" }, + { name = "polars", marker = "extra == 'analysis'", specifier = ">=1.32.0" }, { name = "posebusters", specifier = ">=0.3.1" }, { name = "pot", specifier = ">=0.9.5" }, { name = "py3dmol", specifier = ">=2.4.2" }, + { name = "pyarrow", marker = "extra == 'analysis'", specifier = ">=21.0.0" }, { name = "python-dotenv", specifier = ">=1.0.1" }, { name = "rdkit", specifier = ">=2024.3.6" }, { name = "requests", specifier = ">=2.32.3" }, { name = "s3fs", extras = ["boto3"], specifier = ">=2024.10.0" }, { name = "scipy", specifier = ">=1.13.1" }, + { name = "seaborn", marker = "extra == 'analysis'", specifier = ">=0.13.2" }, { name = "statsmodels", specifier = ">=0.14.0" }, { name = "tabulate", specifier = ">=0.9.0" }, { name = "torch", specifier = ">=2.5.1" }, @@ -1345,6 +1236,7 @@ requires-dist = [ { name = "universal-pathlib", specifier = ">=0.2.6" }, { name = "wandb", specifier = ">=0.19.1" }, ] +provides-extras = ["analysis"] [package.metadata.requires-dev] dev = [ @@ -1358,14 +1250,14 @@ dev = [ [[package]] name = "jaxtyping" -version = "0.3.2" +version = "0.3.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "wadler-lindig" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/88/a8/416bd7ea110ec6b68e8868b0f99c985c735adcf7badc491d3c343937260a/jaxtyping-0.3.2.tar.gz", hash = "sha256:f30483fac4b42e915db8ad2414a85c3b63284aa7d3c100b96b59f755cf4a86ad", size = 44989, upload-time = "2025-04-23T09:48:16.907Z" } +sdist = { url = "https://files.pythonhosted.org/packages/65/bc/a3d6e865ed05e93c5f9c54050fb5f6239c024089db93c3e4f8d85465487d/jaxtyping-0.3.0.tar.gz", hash = "sha256:b334b56436295332addd0b6c451548404d3700c9c35c7fa877c6b3b30ea968de", size = 44793, upload-time = "2025-03-18T17:28:27.372Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c9/b9/281e10e2d967ea5e481683eaec99f55ac5a61085ee60551c36942ef32bef/jaxtyping-0.3.2-py3-none-any.whl", hash = "sha256:6a020fd276226ddb5ac4f5725323843dd65e3c7e85c64fd62431e5f738c74e04", size = 55409, upload-time = "2025-04-23T09:48:15.924Z" }, + { url = "https://files.pythonhosted.org/packages/e9/6e/ce242e8f39919e1af817af194189c5ed2cd2a02dae242563932750787e6b/jaxtyping-0.3.0-py3-none-any.whl", hash = "sha256:4b20d4e7c94d6a2850d78d7849cf33e38a87b993f2f78977d8093efb42cdb892", size = 55199, upload-time = "2025-03-18T17:28:26.003Z" }, ] [[package]] @@ -1403,7 +1295,7 @@ wheels = [ [[package]] name = "jsonschema" -version = "4.25.0" +version = "4.23.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "attrs" }, @@ -1411,21 +1303,21 @@ dependencies = [ { name = "referencing" }, { name = "rpds-py" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/d5/00/a297a868e9d0784450faa7365c2172a7d6110c763e30ba861867c32ae6a9/jsonschema-4.25.0.tar.gz", hash = "sha256:e63acf5c11762c0e6672ffb61482bdf57f0876684d8d249c0fe2d730d48bc55f", size = 356830, upload-time = "2025-07-18T15:39:45.11Z" } +sdist = { url = "https://files.pythonhosted.org/packages/38/2e/03362ee4034a4c917f697890ccd4aec0800ccf9ded7f511971c75451deec/jsonschema-4.23.0.tar.gz", hash = "sha256:d71497fef26351a33265337fa77ffeb82423f3ea21283cd9467bb03999266bc4", size = 325778, upload-time = "2024-07-08T18:40:05.546Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/fe/54/c86cd8e011fe98803d7e382fd67c0df5ceab8d2b7ad8c5a81524f791551c/jsonschema-4.25.0-py3-none-any.whl", hash = "sha256:24c2e8da302de79c8b9382fee3e76b355e44d2a4364bb207159ce10b517bd716", size = 89184, upload-time = "2025-07-18T15:39:42.956Z" }, + { url = "https://files.pythonhosted.org/packages/69/4a/4f9dbeb84e8850557c02365a0eee0649abe5eb1d84af92a25731c6c0f922/jsonschema-4.23.0-py3-none-any.whl", hash = "sha256:fbadb6f8b144a8f8cf9f0b89ba94501d143e50411a1278633f56a7acf7fd5566", size = 88462, upload-time = "2024-07-08T18:40:00.165Z" }, ] [[package]] name = "jsonschema-specifications" -version = "2025.4.1" +version = "2024.10.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "referencing" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/bf/ce/46fbd9c8119cfc3581ee5643ea49464d168028cfb5caff5fc0596d0cf914/jsonschema_specifications-2025.4.1.tar.gz", hash = "sha256:630159c9f4dbea161a6a2205c3011cc4f18ff381b189fff48bb39b9bf26ae608", size = 15513, upload-time = "2025-04-23T12:34:07.418Z" } +sdist = { url = "https://files.pythonhosted.org/packages/10/db/58f950c996c793472e336ff3655b13fbcf1e3b359dcf52dcf3ed3b52c352/jsonschema_specifications-2024.10.1.tar.gz", hash = "sha256:0f38b83639958ce1152d02a7f062902c41c8fd20d558b0c34344292d417ae272", size = 15561, upload-time = "2024-10-08T12:29:32.068Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/01/0e/b27cdbaccf30b890c40ed1da9fd4a3593a5cf94dae54fb34f8a4b74fcd3f/jsonschema_specifications-2025.4.1-py3-none-any.whl", hash = "sha256:4653bffbd6584f7de83a67e0d620ef16900b390ddc7939d56684d6c81e33f1af", size = 18437, upload-time = "2025-04-23T12:34:05.422Z" }, + { url = "https://files.pythonhosted.org/packages/d1/0f/8910b19ac0670a0f80ce1008e5e751c4a57e14d2c4c13a482aa6079fa9d6/jsonschema_specifications-2024.10.1-py3-none-any.whl", hash = "sha256:a09a0680616357d9a0ecf05c12ad234479f549239d0f5b55f3deea67475da9bf", size = 18459, upload-time = "2024-10-08T12:29:30.439Z" }, ] [[package]] @@ -1446,16 +1338,16 @@ wheels = [ [[package]] name = "jupyter-core" -version = "5.8.1" +version = "5.7.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "platformdirs" }, { name = "pywin32", marker = "platform_python_implementation != 'PyPy' and sys_platform == 'win32'" }, { name = "traitlets" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/99/1b/72906d554acfeb588332eaaa6f61577705e9ec752ddb486f302dafa292d9/jupyter_core-5.8.1.tar.gz", hash = "sha256:0a5f9706f70e64786b75acba995988915ebd4601c8a52e534a40b51c95f59941", size = 88923, upload-time = "2025-05-27T07:38:16.655Z" } +sdist = { url = "https://files.pythonhosted.org/packages/00/11/b56381fa6c3f4cc5d2cf54a7dbf98ad9aa0b339ef7a601d6053538b079a7/jupyter_core-5.7.2.tar.gz", hash = "sha256:aa5f8d32bbf6b431ac830496da7392035d6f61b4f54872f15c4bd2a9c3f536d9", size = 87629, upload-time = "2024-03-12T12:37:35.652Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/2f/57/6bffd4b20b88da3800c5d691e0337761576ee688eb01299eae865689d2df/jupyter_core-5.8.1-py3-none-any.whl", hash = "sha256:c28d268fc90fb53f1338ded2eb410704c5449a358406e8a948b75706e24863d0", size = 28880, upload-time = "2025-05-27T07:38:15.137Z" }, + { url = "https://files.pythonhosted.org/packages/c9/fb/108ecd1fe961941959ad0ee4e12ee7b8b1477247f30b1fdfd83ceaf017f0/jupyter_core-5.7.2-py3-none-any.whl", hash = "sha256:4f7315d2f6b4bcf2e3e7cb6e46772eba760ae459cd1f59d29eb57b0a01bd7409", size = 28965, upload-time = "2024-03-12T12:37:32.36Z" }, ] [[package]] @@ -1547,7 +1439,7 @@ wheels = [ [[package]] name = "lightning" -version = "2.5.2" +version = "2.5.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "fsspec", extra = ["http"] }, @@ -1560,23 +1452,23 @@ dependencies = [ { name = "tqdm" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/42/3c/6a930ac7c64fb896adbe560a9141570732d9ca890a11e6d158edd5aece76/lightning-2.5.2.tar.gz", hash = "sha256:9550df613cfb22358ebf77b4a8ad45f3767cd7d26ba2d52b7f036bd3cdd701c4", size = 633391, upload-time = "2025-06-20T15:58:22.065Z" } +sdist = { url = "https://files.pythonhosted.org/packages/39/53/21a834e03317d04cc3db7307d4c19af94c0db43b9001e8fabb8d150c5e69/lightning-2.5.1.tar.gz", hash = "sha256:aca88f8abf3fc38d8b40c1f82ce481f4379c2b181a6eeeb9217db0aba8e40736", size = 630918, upload-time = "2025-03-19T20:28:24.517Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/71/a9/5d39280e55dc5df9e98be074029f6b48f86fe3db4929cb9ada6401234b47/lightning-2.5.2-py3-none-any.whl", hash = "sha256:7e7f23245e214c8ec14d5d8119d3856c25cfe96f9856296fd5df4e29c2ff88a7", size = 821145, upload-time = "2025-06-20T15:58:18.609Z" }, + { url = "https://files.pythonhosted.org/packages/80/eb/45f6629b92cb4ed38854d5b76f9f668ff58404a4b9ec1abefa98502afd98/lightning-2.5.1-py3-none-any.whl", hash = "sha256:512cbf9e80859f331b329536b2e2f90776e6f8a399048745bb4dabacc36e2850", size = 818892, upload-time = "2025-03-19T20:28:21.036Z" }, ] [[package]] name = "lightning-utilities" -version = "0.15.1" +version = "0.14.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "packaging" }, { name = "setuptools" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/10/c8/c1cccd09b1c8a63953cd158ed943b9ee054f553cec4d82974561ee3d20c9/lightning_utilities-0.15.1.tar.gz", hash = "sha256:e1052e9ab294a176f5efcfe6a57b2969b9675cfdae33b936612cbf23fffb4d4b", size = 31093, upload-time = "2025-08-04T20:10:16.658Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e2/95/3654879a4213e29f2c17ce622ca289d67e35c22a35f63cc6d79d22456cfa/lightning_utilities-0.14.2.tar.gz", hash = "sha256:0466a4f1bb9dff1c7190d4c7a32d1a8a1109f94fb816931efe8fb8b12bb0ab8d", size = 30290, upload-time = "2025-03-20T12:53:38.95Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c9/b5/2d118bb657a1877cea0fe7ebced4aaccf6eb1339603104e654cb9b782168/lightning_utilities-0.15.1-py3-none-any.whl", hash = "sha256:18437c86f8380fe80beee0cd3f7a7bde8a68d752060e6a2663c1e5784da768d9", size = 29446, upload-time = "2025-08-04T20:10:15.726Z" }, + { url = "https://files.pythonhosted.org/packages/ec/65/0243d079b4124840abd5fa5a472810c84b46aec67d1b05ba3ba96e3d01e8/lightning_utilities-0.14.2-py3-none-any.whl", hash = "sha256:da791fcaa731f651ec76a1a3b12994ed05af4d6841f2e78760233552709ef05d", size = 28891, upload-time = "2025-03-20T12:53:37.873Z" }, ] [[package]] @@ -1663,77 +1555,54 @@ wheels = [ [[package]] name = "matplotlib" -version = "3.10.5" +version = "3.10.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "contourpy", version = "1.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "contourpy", version = "1.3.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "contourpy" }, { name = "cycler" }, { name = "fonttools" }, { name = "kiwisolver" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "packaging" }, { name = "pillow" }, { name = "pyparsing" }, { name = "python-dateutil" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/43/91/f2939bb60b7ebf12478b030e0d7f340247390f402b3b189616aad790c366/matplotlib-3.10.5.tar.gz", hash = "sha256:352ed6ccfb7998a00881692f38b4ca083c691d3e275b4145423704c34c909076", size = 34804044, upload-time = "2025-07-31T18:09:33.805Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/d1/89/5355cdfe43242cb4d1a64a67cb6831398b665ad90e9702c16247cbd8d5ab/matplotlib-3.10.5-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:5d4773a6d1c106ca05cb5a5515d277a6bb96ed09e5c8fab6b7741b8fcaa62c8f", size = 8229094, upload-time = "2025-07-31T18:07:36.507Z" }, - { url = "https://files.pythonhosted.org/packages/34/bc/ba802650e1c69650faed261a9df004af4c6f21759d7a1ec67fe972f093b3/matplotlib-3.10.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:dc88af74e7ba27de6cbe6faee916024ea35d895ed3d61ef6f58c4ce97da7185a", size = 8091464, upload-time = "2025-07-31T18:07:38.864Z" }, - { url = "https://files.pythonhosted.org/packages/ac/64/8d0c8937dee86c286625bddb1902efacc3e22f2b619f5b5a8df29fe5217b/matplotlib-3.10.5-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:64c4535419d5617f7363dad171a5a59963308e0f3f813c4bed6c9e6e2c131512", size = 8653163, upload-time = "2025-07-31T18:07:41.141Z" }, - { url = "https://files.pythonhosted.org/packages/11/dc/8dfc0acfbdc2fc2336c72561b7935cfa73db9ca70b875d8d3e1b3a6f371a/matplotlib-3.10.5-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a277033048ab22d34f88a3c5243938cef776493f6201a8742ed5f8b553201343", size = 9490635, upload-time = "2025-07-31T18:07:42.936Z" }, - { url = "https://files.pythonhosted.org/packages/54/02/e3fdfe0f2e9fb05f3a691d63876639dbf684170fdcf93231e973104153b4/matplotlib-3.10.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:e4a6470a118a2e93022ecc7d3bd16b3114b2004ea2bf014fff875b3bc99b70c6", size = 9539036, upload-time = "2025-07-31T18:07:45.18Z" }, - { url = "https://files.pythonhosted.org/packages/c1/29/82bf486ff7f4dbedfb11ccc207d0575cbe3be6ea26f75be514252bde3d70/matplotlib-3.10.5-cp310-cp310-win_amd64.whl", hash = "sha256:7e44cada61bec8833c106547786814dd4a266c1b2964fd25daa3804f1b8d4467", size = 8093529, upload-time = "2025-07-31T18:07:49.553Z" }, - { url = "https://files.pythonhosted.org/packages/aa/c7/1f2db90a1d43710478bb1e9b57b162852f79234d28e4f48a28cc415aa583/matplotlib-3.10.5-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:dcfc39c452c6a9f9028d3e44d2d721484f665304857188124b505b2c95e1eecf", size = 8239216, upload-time = "2025-07-31T18:07:51.947Z" }, - { url = "https://files.pythonhosted.org/packages/82/6d/ca6844c77a4f89b1c9e4d481c412e1d1dbabf2aae2cbc5aa2da4a1d6683e/matplotlib-3.10.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:903352681b59f3efbf4546985142a9686ea1d616bb054b09a537a06e4b892ccf", size = 8102130, upload-time = "2025-07-31T18:07:53.65Z" }, - { url = "https://files.pythonhosted.org/packages/1d/1e/5e187a30cc673a3e384f3723e5f3c416033c1d8d5da414f82e4e731128ea/matplotlib-3.10.5-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:080c3676a56b8ee1c762bcf8fca3fe709daa1ee23e6ef06ad9f3fc17332f2d2a", size = 8666471, upload-time = "2025-07-31T18:07:55.304Z" }, - { url = "https://files.pythonhosted.org/packages/03/c0/95540d584d7d645324db99a845ac194e915ef75011a0d5e19e1b5cee7e69/matplotlib-3.10.5-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4b4984d5064a35b6f66d2c11d668565f4389b1119cc64db7a4c1725bc11adffc", size = 9500518, upload-time = "2025-07-31T18:07:57.199Z" }, - { url = "https://files.pythonhosted.org/packages/ba/2e/e019352099ea58b4169adb9c6e1a2ad0c568c6377c2b677ee1f06de2adc7/matplotlib-3.10.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:3967424121d3a46705c9fa9bdb0931de3228f13f73d7bb03c999c88343a89d89", size = 9552372, upload-time = "2025-07-31T18:07:59.41Z" }, - { url = "https://files.pythonhosted.org/packages/b7/81/3200b792a5e8b354f31f4101ad7834743ad07b6d620259f2059317b25e4d/matplotlib-3.10.5-cp311-cp311-win_amd64.whl", hash = "sha256:33775bbeb75528555a15ac29396940128ef5613cf9a2d31fb1bfd18b3c0c0903", size = 8100634, upload-time = "2025-07-31T18:08:01.801Z" }, - { url = "https://files.pythonhosted.org/packages/52/46/a944f6f0c1f5476a0adfa501969d229ce5ae60cf9a663be0e70361381f89/matplotlib-3.10.5-cp311-cp311-win_arm64.whl", hash = "sha256:c61333a8e5e6240e73769d5826b9a31d8b22df76c0778f8480baf1b4b01c9420", size = 7978880, upload-time = "2025-07-31T18:08:03.407Z" }, - { url = "https://files.pythonhosted.org/packages/66/1e/c6f6bcd882d589410b475ca1fc22e34e34c82adff519caf18f3e6dd9d682/matplotlib-3.10.5-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:00b6feadc28a08bd3c65b2894f56cf3c94fc8f7adcbc6ab4516ae1e8ed8f62e2", size = 8253056, upload-time = "2025-07-31T18:08:05.385Z" }, - { url = "https://files.pythonhosted.org/packages/53/e6/d6f7d1b59413f233793dda14419776f5f443bcccb2dfc84b09f09fe05dbe/matplotlib-3.10.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ee98a5c5344dc7f48dc261b6ba5d9900c008fc12beb3fa6ebda81273602cc389", size = 8110131, upload-time = "2025-07-31T18:08:07.293Z" }, - { url = "https://files.pythonhosted.org/packages/66/2b/bed8a45e74957549197a2ac2e1259671cd80b55ed9e1fe2b5c94d88a9202/matplotlib-3.10.5-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:a17e57e33de901d221a07af32c08870ed4528db0b6059dce7d7e65c1122d4bea", size = 8669603, upload-time = "2025-07-31T18:08:09.064Z" }, - { url = "https://files.pythonhosted.org/packages/7e/a7/315e9435b10d057f5e52dfc603cd353167ae28bb1a4e033d41540c0067a4/matplotlib-3.10.5-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:97b9d6443419085950ee4a5b1ee08c363e5c43d7176e55513479e53669e88468", size = 9508127, upload-time = "2025-07-31T18:08:10.845Z" }, - { url = "https://files.pythonhosted.org/packages/7f/d9/edcbb1f02ca99165365d2768d517898c22c6040187e2ae2ce7294437c413/matplotlib-3.10.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ceefe5d40807d29a66ae916c6a3915d60ef9f028ce1927b84e727be91d884369", size = 9566926, upload-time = "2025-07-31T18:08:13.186Z" }, - { url = "https://files.pythonhosted.org/packages/3b/d9/6dd924ad5616c97b7308e6320cf392c466237a82a2040381163b7500510a/matplotlib-3.10.5-cp312-cp312-win_amd64.whl", hash = "sha256:c04cba0f93d40e45b3c187c6c52c17f24535b27d545f757a2fffebc06c12b98b", size = 8107599, upload-time = "2025-07-31T18:08:15.116Z" }, - { url = "https://files.pythonhosted.org/packages/0e/f3/522dc319a50f7b0279fbe74f86f7a3506ce414bc23172098e8d2bdf21894/matplotlib-3.10.5-cp312-cp312-win_arm64.whl", hash = "sha256:a41bcb6e2c8e79dc99c5511ae6f7787d2fb52efd3d805fff06d5d4f667db16b2", size = 7978173, upload-time = "2025-07-31T18:08:21.518Z" }, - { url = "https://files.pythonhosted.org/packages/8d/05/4f3c1f396075f108515e45cb8d334aff011a922350e502a7472e24c52d77/matplotlib-3.10.5-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:354204db3f7d5caaa10e5de74549ef6a05a4550fdd1c8f831ab9bca81efd39ed", size = 8253586, upload-time = "2025-07-31T18:08:23.107Z" }, - { url = "https://files.pythonhosted.org/packages/2f/2c/e084415775aac7016c3719fe7006cdb462582c6c99ac142f27303c56e243/matplotlib-3.10.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b072aac0c3ad563a2b3318124756cb6112157017f7431626600ecbe890df57a1", size = 8110715, upload-time = "2025-07-31T18:08:24.675Z" }, - { url = "https://files.pythonhosted.org/packages/52/1b/233e3094b749df16e3e6cd5a44849fd33852e692ad009cf7de00cf58ddf6/matplotlib-3.10.5-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:d52fd5b684d541b5a51fb276b2b97b010c75bee9aa392f96b4a07aeb491e33c7", size = 8669397, upload-time = "2025-07-31T18:08:26.778Z" }, - { url = "https://files.pythonhosted.org/packages/e8/ec/03f9e003a798f907d9f772eed9b7c6a9775d5bd00648b643ebfb88e25414/matplotlib-3.10.5-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ee7a09ae2f4676276f5a65bd9f2bd91b4f9fbaedf49f40267ce3f9b448de501f", size = 9508646, upload-time = "2025-07-31T18:08:28.848Z" }, - { url = "https://files.pythonhosted.org/packages/91/e7/c051a7a386680c28487bca27d23b02d84f63e3d2a9b4d2fc478e6a42e37e/matplotlib-3.10.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:ba6c3c9c067b83481d647af88b4e441d532acdb5ef22178a14935b0b881188f4", size = 9567424, upload-time = "2025-07-31T18:08:30.726Z" }, - { url = "https://files.pythonhosted.org/packages/36/c2/24302e93ff431b8f4173ee1dd88976c8d80483cadbc5d3d777cef47b3a1c/matplotlib-3.10.5-cp313-cp313-win_amd64.whl", hash = "sha256:07442d2692c9bd1cceaa4afb4bbe5b57b98a7599de4dabfcca92d3eea70f9ebe", size = 8107809, upload-time = "2025-07-31T18:08:33.928Z" }, - { url = "https://files.pythonhosted.org/packages/0b/33/423ec6a668d375dad825197557ed8fbdb74d62b432c1ed8235465945475f/matplotlib-3.10.5-cp313-cp313-win_arm64.whl", hash = "sha256:48fe6d47380b68a37ccfcc94f009530e84d41f71f5dae7eda7c4a5a84aa0a674", size = 7978078, upload-time = "2025-07-31T18:08:36.764Z" }, - { url = "https://files.pythonhosted.org/packages/51/17/521fc16ec766455c7bb52cc046550cf7652f6765ca8650ff120aa2d197b6/matplotlib-3.10.5-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3b80eb8621331449fc519541a7461987f10afa4f9cfd91afcd2276ebe19bd56c", size = 8295590, upload-time = "2025-07-31T18:08:38.521Z" }, - { url = "https://files.pythonhosted.org/packages/f8/12/23c28b2c21114c63999bae129fce7fd34515641c517ae48ce7b7dcd33458/matplotlib-3.10.5-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:47a388908e469d6ca2a6015858fa924e0e8a2345a37125948d8e93a91c47933e", size = 8158518, upload-time = "2025-07-31T18:08:40.195Z" }, - { url = "https://files.pythonhosted.org/packages/81/f8/aae4eb25e8e7190759f3cb91cbeaa344128159ac92bb6b409e24f8711f78/matplotlib-3.10.5-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:8b6b49167d208358983ce26e43aa4196073b4702858670f2eb111f9a10652b4b", size = 8691815, upload-time = "2025-07-31T18:08:42.238Z" }, - { url = "https://files.pythonhosted.org/packages/d0/ba/450c39ebdd486bd33a359fc17365ade46c6a96bf637bbb0df7824de2886c/matplotlib-3.10.5-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8a8da0453a7fd8e3da114234ba70c5ba9ef0e98f190309ddfde0f089accd46ea", size = 9522814, upload-time = "2025-07-31T18:08:44.914Z" }, - { url = "https://files.pythonhosted.org/packages/89/11/9c66f6a990e27bb9aa023f7988d2d5809cb98aa39c09cbf20fba75a542ef/matplotlib-3.10.5-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:52c6573dfcb7726a9907b482cd5b92e6b5499b284ffacb04ffbfe06b3e568124", size = 9573917, upload-time = "2025-07-31T18:08:47.038Z" }, - { url = "https://files.pythonhosted.org/packages/b3/69/8b49394de92569419e5e05e82e83df9b749a0ff550d07631ea96ed2eb35a/matplotlib-3.10.5-cp313-cp313t-win_amd64.whl", hash = "sha256:a23193db2e9d64ece69cac0c8231849db7dd77ce59c7b89948cf9d0ce655a3ce", size = 8181034, upload-time = "2025-07-31T18:08:48.943Z" }, - { url = "https://files.pythonhosted.org/packages/47/23/82dc435bb98a2fc5c20dffcac8f0b083935ac28286413ed8835df40d0baa/matplotlib-3.10.5-cp313-cp313t-win_arm64.whl", hash = "sha256:56da3b102cf6da2776fef3e71cd96fcf22103a13594a18ac9a9b31314e0be154", size = 8023337, upload-time = "2025-07-31T18:08:50.791Z" }, - { url = "https://files.pythonhosted.org/packages/ac/e0/26b6cfde31f5383503ee45dcb7e691d45dadf0b3f54639332b59316a97f8/matplotlib-3.10.5-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:96ef8f5a3696f20f55597ffa91c28e2e73088df25c555f8d4754931515512715", size = 8253591, upload-time = "2025-07-31T18:08:53.254Z" }, - { url = "https://files.pythonhosted.org/packages/c1/89/98488c7ef7ea20ea659af7499628c240a608b337af4be2066d644cfd0a0f/matplotlib-3.10.5-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:77fab633e94b9da60512d4fa0213daeb76d5a7b05156840c4fd0399b4b818837", size = 8112566, upload-time = "2025-07-31T18:08:55.116Z" }, - { url = "https://files.pythonhosted.org/packages/52/67/42294dfedc82aea55e1a767daf3263aacfb5a125f44ba189e685bab41b6f/matplotlib-3.10.5-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:27f52634315e96b1debbfdc5c416592edcd9c4221bc2f520fd39c33db5d9f202", size = 9513281, upload-time = "2025-07-31T18:08:56.885Z" }, - { url = "https://files.pythonhosted.org/packages/e7/68/f258239e0cf34c2cbc816781c7ab6fca768452e6bf1119aedd2bd4a882a3/matplotlib-3.10.5-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:525f6e28c485c769d1f07935b660c864de41c37fd716bfa64158ea646f7084bb", size = 9780873, upload-time = "2025-07-31T18:08:59.241Z" }, - { url = "https://files.pythonhosted.org/packages/89/64/f4881554006bd12e4558bd66778bdd15d47b00a1f6c6e8b50f6208eda4b3/matplotlib-3.10.5-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:1f5f3ec4c191253c5f2b7c07096a142c6a1c024d9f738247bfc8e3f9643fc975", size = 9568954, upload-time = "2025-07-31T18:09:01.244Z" }, - { url = "https://files.pythonhosted.org/packages/06/f8/42779d39c3f757e1f012f2dda3319a89fb602bd2ef98ce8faf0281f4febd/matplotlib-3.10.5-cp314-cp314-win_amd64.whl", hash = "sha256:707f9c292c4cd4716f19ab8a1f93f26598222cd931e0cd98fbbb1c5994bf7667", size = 8237465, upload-time = "2025-07-31T18:09:03.206Z" }, - { url = "https://files.pythonhosted.org/packages/cf/f8/153fd06b5160f0cd27c8b9dd797fcc9fb56ac6a0ebf3c1f765b6b68d3c8a/matplotlib-3.10.5-cp314-cp314-win_arm64.whl", hash = "sha256:21a95b9bf408178d372814de7baacd61c712a62cae560b5e6f35d791776f6516", size = 8108898, upload-time = "2025-07-31T18:09:05.231Z" }, - { url = "https://files.pythonhosted.org/packages/9a/ee/c4b082a382a225fe0d2a73f1f57cf6f6f132308805b493a54c8641006238/matplotlib-3.10.5-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:a6b310f95e1102a8c7c817ef17b60ee5d1851b8c71b63d9286b66b177963039e", size = 8295636, upload-time = "2025-07-31T18:09:07.306Z" }, - { url = "https://files.pythonhosted.org/packages/30/73/2195fa2099718b21a20da82dfc753bf2af58d596b51aefe93e359dd5915a/matplotlib-3.10.5-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:94986a242747a0605cb3ff1cb98691c736f28a59f8ffe5175acaeb7397c49a5a", size = 8158575, upload-time = "2025-07-31T18:09:09.083Z" }, - { url = "https://files.pythonhosted.org/packages/f6/e9/a08cdb34618a91fa08f75e6738541da5cacde7c307cea18ff10f0d03fcff/matplotlib-3.10.5-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1ff10ea43288f0c8bab608a305dc6c918cc729d429c31dcbbecde3b9f4d5b569", size = 9522815, upload-time = "2025-07-31T18:09:11.191Z" }, - { url = "https://files.pythonhosted.org/packages/4e/bb/34d8b7e0d1bb6d06ef45db01dfa560d5a67b1c40c0b998ce9ccde934bb09/matplotlib-3.10.5-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f6adb644c9d040ffb0d3434e440490a66cf73dbfa118a6f79cd7568431f7a012", size = 9783514, upload-time = "2025-07-31T18:09:13.307Z" }, - { url = "https://files.pythonhosted.org/packages/12/09/d330d1e55dcca2e11b4d304cc5227f52e2512e46828d6249b88e0694176e/matplotlib-3.10.5-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:4fa40a8f98428f789a9dcacd625f59b7bc4e3ef6c8c7c80187a7a709475cf592", size = 9573932, upload-time = "2025-07-31T18:09:15.335Z" }, - { url = "https://files.pythonhosted.org/packages/eb/3b/f70258ac729aa004aca673800a53a2b0a26d49ca1df2eaa03289a1c40f81/matplotlib-3.10.5-cp314-cp314t-win_amd64.whl", hash = "sha256:95672a5d628b44207aab91ec20bf59c26da99de12b88f7e0b1fb0a84a86ff959", size = 8322003, upload-time = "2025-07-31T18:09:17.416Z" }, - { url = "https://files.pythonhosted.org/packages/5b/60/3601f8ce6d76a7c81c7f25a0e15fde0d6b66226dd187aa6d2838e6374161/matplotlib-3.10.5-cp314-cp314t-win_arm64.whl", hash = "sha256:2efaf97d72629e74252e0b5e3c46813e9eeaa94e011ecf8084a971a31a97f40b", size = 8153849, upload-time = "2025-07-31T18:09:19.673Z" }, - { url = "https://files.pythonhosted.org/packages/e4/eb/7d4c5de49eb78294e1a8e2be8a6ecff8b433e921b731412a56cd1abd3567/matplotlib-3.10.5-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:b5fa2e941f77eb579005fb804026f9d0a1082276118d01cc6051d0d9626eaa7f", size = 8222360, upload-time = "2025-07-31T18:09:21.813Z" }, - { url = "https://files.pythonhosted.org/packages/16/8a/e435db90927b66b16d69f8f009498775f4469f8de4d14b87856965e58eba/matplotlib-3.10.5-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:1fc0d2a3241cdcb9daaca279204a3351ce9df3c0e7e621c7e04ec28aaacaca30", size = 8087462, upload-time = "2025-07-31T18:09:23.504Z" }, - { url = "https://files.pythonhosted.org/packages/0b/dd/06c0e00064362f5647f318e00b435be2ff76a1bdced97c5eaf8347311fbe/matplotlib-3.10.5-pp310-pypy310_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:8dee65cb1424b7dc982fe87895b5613d4e691cc57117e8af840da0148ca6c1d7", size = 8659802, upload-time = "2025-07-31T18:09:25.256Z" }, - { url = "https://files.pythonhosted.org/packages/dc/d6/e921be4e1a5f7aca5194e1f016cb67ec294548e530013251f630713e456d/matplotlib-3.10.5-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:160e125da27a749481eaddc0627962990f6029811dbeae23881833a011a0907f", size = 8233224, upload-time = "2025-07-31T18:09:27.512Z" }, - { url = "https://files.pythonhosted.org/packages/ec/74/a2b9b04824b9c349c8f1b2d21d5af43fa7010039427f2b133a034cb09e59/matplotlib-3.10.5-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:ac3d50760394d78a3c9be6b28318fe22b494c4fcf6407e8fd4794b538251899b", size = 8098539, upload-time = "2025-07-31T18:09:29.629Z" }, - { url = "https://files.pythonhosted.org/packages/fc/66/cd29ebc7f6c0d2a15d216fb572573e8fc38bd5d6dec3bd9d7d904c0949f7/matplotlib-3.10.5-pp311-pypy311_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:6c49465bf689c4d59d174d0c7795fb42a21d4244d11d70e52b8011987367ac61", size = 8672192, upload-time = "2025-07-31T18:09:31.407Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/2f/08/b89867ecea2e305f408fbb417139a8dd941ecf7b23a2e02157c36da546f0/matplotlib-3.10.1.tar.gz", hash = "sha256:e8d2d0e3881b129268585bf4765ad3ee73a4591d77b9a18c214ac7e3a79fb2ba", size = 36743335, upload-time = "2025-02-27T19:19:51.038Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/ee/b1/f70e27cf1cd76ce2a5e1aa5579d05afe3236052c6d9b9a96325bc823a17e/matplotlib-3.10.1-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:ff2ae14910be903f4a24afdbb6d7d3a6c44da210fc7d42790b87aeac92238a16", size = 8163654, upload-time = "2025-02-27T19:18:10.961Z" }, + { url = "https://files.pythonhosted.org/packages/26/af/5ec3d4636106718bb62503a03297125d4514f98fe818461bd9e6b9d116e4/matplotlib-3.10.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:0721a3fd3d5756ed593220a8b86808a36c5031fce489adb5b31ee6dbb47dd5b2", size = 8037943, upload-time = "2025-02-27T19:18:16.742Z" }, + { url = "https://files.pythonhosted.org/packages/a1/3d/07f9003a71b698b848c9925d05979ffa94a75cd25d1a587202f0bb58aa81/matplotlib-3.10.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d0673b4b8f131890eb3a1ad058d6e065fb3c6e71f160089b65f8515373394698", size = 8449510, upload-time = "2025-02-27T19:18:19.56Z" }, + { url = "https://files.pythonhosted.org/packages/12/87/9472d4513ff83b7cd864311821793ab72234fa201ab77310ec1b585d27e2/matplotlib-3.10.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8e875b95ac59a7908978fe307ecdbdd9a26af7fa0f33f474a27fcf8c99f64a19", size = 8586585, upload-time = "2025-02-27T19:18:25.61Z" }, + { url = "https://files.pythonhosted.org/packages/31/9e/fe74d237d2963adae8608faeb21f778cf246dbbf4746cef87cffbc82c4b6/matplotlib-3.10.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:2589659ea30726284c6c91037216f64a506a9822f8e50592d48ac16a2f29e044", size = 9397911, upload-time = "2025-02-27T19:18:28.914Z" }, + { url = "https://files.pythonhosted.org/packages/b6/1b/025d3e59e8a4281ab463162ad7d072575354a1916aba81b6a11507dfc524/matplotlib-3.10.1-cp310-cp310-win_amd64.whl", hash = "sha256:a97ff127f295817bc34517255c9db6e71de8eddaab7f837b7d341dee9f2f587f", size = 8052998, upload-time = "2025-02-27T19:18:31.518Z" }, + { url = "https://files.pythonhosted.org/packages/a5/14/a1b840075be247bb1834b22c1e1d558740b0f618fe3a823740181ca557a1/matplotlib-3.10.1-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:057206ff2d6ab82ff3e94ebd94463d084760ca682ed5f150817b859372ec4401", size = 8174669, upload-time = "2025-02-27T19:18:34.346Z" }, + { url = "https://files.pythonhosted.org/packages/0a/e4/300b08e3e08f9c98b0d5635f42edabf2f7a1d634e64cb0318a71a44ff720/matplotlib-3.10.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a144867dd6bf8ba8cb5fc81a158b645037e11b3e5cf8a50bd5f9917cb863adfe", size = 8047996, upload-time = "2025-02-27T19:18:37.247Z" }, + { url = "https://files.pythonhosted.org/packages/75/f9/8d99ff5a2498a5f1ccf919fb46fb945109623c6108216f10f96428f388bc/matplotlib-3.10.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:56c5d9fcd9879aa8040f196a235e2dcbdf7dd03ab5b07c0696f80bc6cf04bedd", size = 8461612, upload-time = "2025-02-27T19:18:39.642Z" }, + { url = "https://files.pythonhosted.org/packages/40/b8/53fa08a5eaf78d3a7213fd6da1feec4bae14a81d9805e567013811ff0e85/matplotlib-3.10.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0f69dc9713e4ad2fb21a1c30e37bd445d496524257dfda40ff4a8efb3604ab5c", size = 8602258, upload-time = "2025-02-27T19:18:43.217Z" }, + { url = "https://files.pythonhosted.org/packages/40/87/4397d2ce808467af86684a622dd112664553e81752ea8bf61bdd89d24a41/matplotlib-3.10.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:4c59af3e8aca75d7744b68e8e78a669e91ccbcf1ac35d0102a7b1b46883f1dd7", size = 9408896, upload-time = "2025-02-27T19:18:45.852Z" }, + { url = "https://files.pythonhosted.org/packages/d7/68/0d03098b3feb786cbd494df0aac15b571effda7f7cbdec267e8a8d398c16/matplotlib-3.10.1-cp311-cp311-win_amd64.whl", hash = "sha256:11b65088c6f3dae784bc72e8d039a2580186285f87448babb9ddb2ad0082993a", size = 8061281, upload-time = "2025-02-27T19:18:48.919Z" }, + { url = "https://files.pythonhosted.org/packages/7c/1d/5e0dc3b59c034e43de16f94deb68f4ad8a96b3ea00f4b37c160b7474928e/matplotlib-3.10.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:66e907a06e68cb6cfd652c193311d61a12b54f56809cafbed9736ce5ad92f107", size = 8175488, upload-time = "2025-02-27T19:18:51.436Z" }, + { url = "https://files.pythonhosted.org/packages/7a/81/dae7e14042e74da658c3336ab9799128e09a1ee03964f2d89630b5d12106/matplotlib-3.10.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:e9b4bb156abb8fa5e5b2b460196f7db7264fc6d62678c03457979e7d5254b7be", size = 8046264, upload-time = "2025-02-27T19:18:54.344Z" }, + { url = "https://files.pythonhosted.org/packages/21/c4/22516775dcde10fc9c9571d155f90710761b028fc44f660508106c363c97/matplotlib-3.10.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1985ad3d97f51307a2cbfc801a930f120def19ba22864182dacef55277102ba6", size = 8452048, upload-time = "2025-02-27T19:18:56.536Z" }, + { url = "https://files.pythonhosted.org/packages/63/23/c0615001f67ce7c96b3051d856baedc0c818a2ed84570b9bf9bde200f85d/matplotlib-3.10.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c96f2c2f825d1257e437a1482c5a2cf4fee15db4261bd6fc0750f81ba2b4ba3d", size = 8597111, upload-time = "2025-02-27T19:18:59.439Z" }, + { url = "https://files.pythonhosted.org/packages/ca/c0/a07939a82aed77770514348f4568177d7dadab9787ebc618a616fe3d665e/matplotlib-3.10.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:35e87384ee9e488d8dd5a2dd7baf471178d38b90618d8ea147aced4ab59c9bea", size = 9402771, upload-time = "2025-02-27T19:19:01.944Z" }, + { url = "https://files.pythonhosted.org/packages/a6/b6/a9405484fb40746fdc6ae4502b16a9d6e53282ba5baaf9ebe2da579f68c4/matplotlib-3.10.1-cp312-cp312-win_amd64.whl", hash = "sha256:cfd414bce89cc78a7e1d25202e979b3f1af799e416010a20ab2b5ebb3a02425c", size = 8063742, upload-time = "2025-02-27T19:19:04.632Z" }, + { url = "https://files.pythonhosted.org/packages/60/73/6770ff5e5523d00f3bc584acb6031e29ee5c8adc2336b16cd1d003675fe0/matplotlib-3.10.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:c42eee41e1b60fd83ee3292ed83a97a5f2a8239b10c26715d8a6172226988d7b", size = 8176112, upload-time = "2025-02-27T19:19:07.59Z" }, + { url = "https://files.pythonhosted.org/packages/08/97/b0ca5da0ed54a3f6599c3ab568bdda65269bc27c21a2c97868c1625e4554/matplotlib-3.10.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:4f0647b17b667ae745c13721602b540f7aadb2a32c5b96e924cd4fea5dcb90f1", size = 8046931, upload-time = "2025-02-27T19:19:10.515Z" }, + { url = "https://files.pythonhosted.org/packages/df/9a/1acbdc3b165d4ce2dcd2b1a6d4ffb46a7220ceee960c922c3d50d8514067/matplotlib-3.10.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa3854b5f9473564ef40a41bc922be978fab217776e9ae1545c9b3a5cf2092a3", size = 8453422, upload-time = "2025-02-27T19:19:12.738Z" }, + { url = "https://files.pythonhosted.org/packages/51/d0/2bc4368abf766203e548dc7ab57cf7e9c621f1a3c72b516cc7715347b179/matplotlib-3.10.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7e496c01441be4c7d5f96d4e40f7fca06e20dcb40e44c8daa2e740e1757ad9e6", size = 8596819, upload-time = "2025-02-27T19:19:15.306Z" }, + { url = "https://files.pythonhosted.org/packages/ab/1b/8b350f8a1746c37ab69dda7d7528d1fc696efb06db6ade9727b7887be16d/matplotlib-3.10.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:5d45d3f5245be5b469843450617dcad9af75ca50568acf59997bed9311131a0b", size = 9402782, upload-time = "2025-02-27T19:19:17.841Z" }, + { url = "https://files.pythonhosted.org/packages/89/06/f570373d24d93503988ba8d04f213a372fa1ce48381c5eb15da985728498/matplotlib-3.10.1-cp313-cp313-win_amd64.whl", hash = "sha256:8e8e25b1209161d20dfe93037c8a7f7ca796ec9aa326e6e4588d8c4a5dd1e473", size = 8063812, upload-time = "2025-02-27T19:19:20.888Z" }, + { url = "https://files.pythonhosted.org/packages/fc/e0/8c811a925b5a7ad75135f0e5af46408b78af88bbb02a1df775100ef9bfef/matplotlib-3.10.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:19b06241ad89c3ae9469e07d77efa87041eac65d78df4fcf9cac318028009b01", size = 8214021, upload-time = "2025-02-27T19:19:23.412Z" }, + { url = "https://files.pythonhosted.org/packages/4a/34/319ec2139f68ba26da9d00fce2ff9f27679fb799a6c8e7358539801fd629/matplotlib-3.10.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:01e63101ebb3014e6e9f80d9cf9ee361a8599ddca2c3e166c563628b39305dbb", size = 8090782, upload-time = "2025-02-27T19:19:28.33Z" }, + { url = "https://files.pythonhosted.org/packages/77/ea/9812124ab9a99df5b2eec1110e9b2edc0b8f77039abf4c56e0a376e84a29/matplotlib-3.10.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3f06bad951eea6422ac4e8bdebcf3a70c59ea0a03338c5d2b109f57b64eb3972", size = 8478901, upload-time = "2025-02-27T19:19:31.536Z" }, + { url = "https://files.pythonhosted.org/packages/c9/db/b05bf463689134789b06dea85828f8ebe506fa1e37593f723b65b86c9582/matplotlib-3.10.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a3dfb036f34873b46978f55e240cff7a239f6c4409eac62d8145bad3fc6ba5a3", size = 8613864, upload-time = "2025-02-27T19:19:34.233Z" }, + { url = "https://files.pythonhosted.org/packages/c2/04/41ccec4409f3023a7576df3b5c025f1a8c8b81fbfe922ecfd837ac36e081/matplotlib-3.10.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:dc6ab14a7ab3b4d813b88ba957fc05c79493a037f54e246162033591e770de6f", size = 9409487, upload-time = "2025-02-27T19:19:36.924Z" }, + { url = "https://files.pythonhosted.org/packages/ac/c2/0d5aae823bdcc42cc99327ecdd4d28585e15ccd5218c453b7bcd827f3421/matplotlib-3.10.1-cp313-cp313t-win_amd64.whl", hash = "sha256:bc411ebd5889a78dabbc457b3fa153203e22248bfa6eedc6797be5df0164dbf9", size = 8134832, upload-time = "2025-02-27T19:19:39.431Z" }, + { url = "https://files.pythonhosted.org/packages/c8/f6/10adb696d8cbeed2ab4c2e26ecf1c80dd3847bbf3891f4a0c362e0e08a5a/matplotlib-3.10.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:648406f1899f9a818cef8c0231b44dcfc4ff36f167101c3fd1c9151f24220fdc", size = 8158685, upload-time = "2025-02-27T19:19:41.535Z" }, + { url = "https://files.pythonhosted.org/packages/3f/84/0603d917406072763e7f9bb37747d3d74d7ecd4b943a8c947cc3ae1cf7af/matplotlib-3.10.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:02582304e352f40520727984a5a18f37e8187861f954fea9be7ef06569cf85b4", size = 8035491, upload-time = "2025-02-27T19:19:44.186Z" }, + { url = "https://files.pythonhosted.org/packages/fd/7d/6a8b31dd07ed856b3eae001c9129670ef75c4698fa1c2a6ac9f00a4a7054/matplotlib-3.10.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d3809916157ba871bcdd33d3493acd7fe3037db5daa917ca6e77975a94cef779", size = 8590087, upload-time = "2025-02-27T19:19:46.709Z" }, ] [[package]] @@ -1752,16 +1621,12 @@ wheels = [ name = "mdtraj" version = "1.10.3" source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version < '3.11' and sys_platform == 'linux'", - "python_full_version < '3.11' and sys_platform != 'linux'", -] dependencies = [ - { name = "netcdf4", marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "packaging", marker = "python_full_version < '3.11'" }, - { name = "pyparsing", marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "netcdf4" }, + { name = "numpy" }, + { name = "packaging" }, + { name = "pyparsing" }, + { name = "scipy" }, ] sdist = { url = "https://files.pythonhosted.org/packages/6a/ef/44c34823d2cfd935b76be102ea41bce3a2e688c3d84ccff9029df0c1ff96/mdtraj-1.10.3.tar.gz", hash = "sha256:d14a35009263725b784c436a8ac63fb6ceeb2bb366a526715dac6590d21025e5", size = 2548673, upload-time = "2025-02-07T15:52:11.953Z" } wheels = [ @@ -1783,37 +1648,6 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/26/b7/cd45c6bae1566572d96bda6e749c63886c9c6ded079e34615376de5fe26e/mdtraj-1.10.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1c083e080d1ddf3eb25acec343f4efe93671e1508e17f61b656db8c3a50a38d1", size = 7800597, upload-time = "2025-02-07T18:11:16.853Z" }, ] -[[package]] -name = "mdtraj" -version = "1.11.0" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.13' and sys_platform == 'linux'", - "python_full_version == '3.12.*' and sys_platform == 'linux'", - "python_full_version >= '3.13' and sys_platform != 'linux'", - "python_full_version == '3.12.*' and sys_platform != 'linux'", - "python_full_version == '3.11.*' and sys_platform == 'linux'", - "python_full_version == '3.11.*' and sys_platform != 'linux'", -] -dependencies = [ - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "packaging", marker = "python_full_version >= '3.11'" }, - { name = "pyparsing", marker = "python_full_version >= '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/49/87/04f21b321e7bb6f3e09b5339df51699b8caae4846796175845bbc44f298b/mdtraj-1.11.0.tar.gz", hash = "sha256:2cf0ed2ee9a603dc4599743b1806e1387c655e2a1e9d013841b99fb137ad852e", size = 2860049, upload-time = "2025-06-25T02:06:19.716Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/f8/bc/1a6d273467a24dad5048bf527288a7bbe997ecee8342f6e8fdd7465c9132/mdtraj-1.11.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:a6bbf2bc03a2ed23fd1fe696f6f6f970f495e843f3baaa71a0e99da086420de1", size = 2463123, upload-time = "2025-06-25T02:03:06.602Z" }, - { url = "https://files.pythonhosted.org/packages/09/dc/37bd8a22c66d05a293979d8d8cf1de5e6c03d21ebd6950b6a839c8d8a16b/mdtraj-1.11.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c34002c49282d208fdd34cd1b2578859bc7377a22ebd66a194efa1fc17395e5", size = 8023237, upload-time = "2025-06-25T02:03:28.425Z" }, - { url = "https://files.pythonhosted.org/packages/4c/48/6fc195907c056df9b6ff6442bb39c5dd433b971b9deaecf6d18a0761efa8/mdtraj-1.11.0-cp311-cp311-win_amd64.whl", hash = "sha256:4c794aac3db13937ed7e4f8c2b773564a89b2b539c15f43bae9dab6e1ffa5de1", size = 1355146, upload-time = "2025-06-25T02:03:31.524Z" }, - { url = "https://files.pythonhosted.org/packages/21/71/56adb24577052f2cd6f9b678da908e8e2d7963db3a88faff39792be43903/mdtraj-1.11.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:878fb8e62962b881ba75d4200637280b518e891e4aa5f213052ddb1905704558", size = 2448664, upload-time = "2025-06-25T02:03:43.052Z" }, - { url = "https://files.pythonhosted.org/packages/50/af/7d65be36619befb29b51b6118a97121de6685720af7f4cfd6ff1902ca7eb/mdtraj-1.11.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:01eb00d9b62f22056303fb8e4be0f85257a2ce1003f1e6274393bbcc977badaf", size = 8028112, upload-time = "2025-06-25T02:04:12.949Z" }, - { url = "https://files.pythonhosted.org/packages/d2/8f/fd05aa3613fad29b3a779e3384797d2b64aa329bcfd18d25f8fb9c0d725e/mdtraj-1.11.0-cp312-cp312-win_amd64.whl", hash = "sha256:f87117f7e6742b11a3a39e55fe5c9b4fc013d27691ac671bcc6dfb8193beae0f", size = 1341985, upload-time = "2025-06-25T02:04:19.133Z" }, - { url = "https://files.pythonhosted.org/packages/9c/8f/adbeabf14f27a2befd021548c5d8e696ca129924e0473863fb0e9e546f27/mdtraj-1.11.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:125db8d7b8f795ed2a771d5d6f419461e2f9dd21ef74d43311570b9dadbf0d46", size = 2432588, upload-time = "2025-06-25T02:04:25.409Z" }, - { url = "https://files.pythonhosted.org/packages/b5/c4/4a2e8db0f24c1deb85f4719d54a0a52c56403e17db61ddd72a95d39e07d6/mdtraj-1.11.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:46def167a429af5834f4f26a80bca372518952519847c1ceffd0163814b1164e", size = 7983336, upload-time = "2025-06-25T02:04:57.022Z" }, - { url = "https://files.pythonhosted.org/packages/ac/17/239ae1869fa0f6fe11e42388f5d1e094ac774f4d122767722f0451523989/mdtraj-1.11.0-cp313-cp313-win_amd64.whl", hash = "sha256:8ad81b0aabf090b1b0e60ee7974bd676046e48d35a7efdc9a37a8eacaf5b70e4", size = 1340489, upload-time = "2025-06-25T02:05:04.776Z" }, -] - [[package]] name = "mdurl" version = "0.1.2" @@ -1834,113 +1668,98 @@ wheels = [ [[package]] name = "multidict" -version = "6.6.3" +version = "6.2.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "typing-extensions", marker = "python_full_version < '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/3d/2c/5dad12e82fbdf7470f29bff2171484bf07cb3b16ada60a6589af8f376440/multidict-6.6.3.tar.gz", hash = "sha256:798a9eb12dab0a6c2e29c1de6f3468af5cb2da6053a20dfa3344907eed0937cc", size = 101006, upload-time = "2025-06-30T15:53:46.929Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/0b/67/414933982bce2efce7cbcb3169eaaf901e0f25baec69432b4874dfb1f297/multidict-6.6.3-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:a2be5b7b35271f7fff1397204ba6708365e3d773579fe2a30625e16c4b4ce817", size = 77017, upload-time = "2025-06-30T15:50:58.931Z" }, - { url = "https://files.pythonhosted.org/packages/8a/fe/d8a3ee1fad37dc2ef4f75488b0d9d4f25bf204aad8306cbab63d97bff64a/multidict-6.6.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:12f4581d2930840295c461764b9a65732ec01250b46c6b2c510d7ee68872b140", size = 44897, upload-time = "2025-06-30T15:51:00.999Z" }, - { url = "https://files.pythonhosted.org/packages/1f/e0/265d89af8c98240265d82b8cbcf35897f83b76cd59ee3ab3879050fd8c45/multidict-6.6.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:dd7793bab517e706c9ed9d7310b06c8672fd0aeee5781bfad612f56b8e0f7d14", size = 44574, upload-time = "2025-06-30T15:51:02.449Z" }, - { url = "https://files.pythonhosted.org/packages/e6/05/6b759379f7e8e04ccc97cfb2a5dcc5cdbd44a97f072b2272dc51281e6a40/multidict-6.6.3-cp310-cp310-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:72d8815f2cd3cf3df0f83cac3f3ef801d908b2d90409ae28102e0553af85545a", size = 225729, upload-time = "2025-06-30T15:51:03.794Z" }, - { url = "https://files.pythonhosted.org/packages/4e/f5/8d5a15488edd9a91fa4aad97228d785df208ed6298580883aa3d9def1959/multidict-6.6.3-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:531e331a2ee53543ab32b16334e2deb26f4e6b9b28e41f8e0c87e99a6c8e2d69", size = 242515, upload-time = "2025-06-30T15:51:05.002Z" }, - { url = "https://files.pythonhosted.org/packages/6e/b5/a8f317d47d0ac5bb746d6d8325885c8967c2a8ce0bb57be5399e3642cccb/multidict-6.6.3-cp310-cp310-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:42ca5aa9329a63be8dc49040f63817d1ac980e02eeddba763a9ae5b4027b9c9c", size = 222224, upload-time = "2025-06-30T15:51:06.148Z" }, - { url = "https://files.pythonhosted.org/packages/76/88/18b2a0d5e80515fa22716556061189c2853ecf2aa2133081ebbe85ebea38/multidict-6.6.3-cp310-cp310-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:208b9b9757060b9faa6f11ab4bc52846e4f3c2fb8b14d5680c8aac80af3dc751", size = 253124, upload-time = "2025-06-30T15:51:07.375Z" }, - { url = "https://files.pythonhosted.org/packages/62/bf/ebfcfd6b55a1b05ef16d0775ae34c0fe15e8dab570d69ca9941073b969e7/multidict-6.6.3-cp310-cp310-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:acf6b97bd0884891af6a8b43d0f586ab2fcf8e717cbd47ab4bdddc09e20652d8", size = 251529, upload-time = "2025-06-30T15:51:08.691Z" }, - { url = "https://files.pythonhosted.org/packages/44/11/780615a98fd3775fc309d0234d563941af69ade2df0bb82c91dda6ddaea1/multidict-6.6.3-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:68e9e12ed00e2089725669bdc88602b0b6f8d23c0c95e52b95f0bc69f7fe9b55", size = 241627, upload-time = "2025-06-30T15:51:10.605Z" }, - { url = "https://files.pythonhosted.org/packages/28/3d/35f33045e21034b388686213752cabc3a1b9d03e20969e6fa8f1b1d82db1/multidict-6.6.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:05db2f66c9addb10cfa226e1acb363450fab2ff8a6df73c622fefe2f5af6d4e7", size = 239351, upload-time = "2025-06-30T15:51:12.18Z" }, - { url = "https://files.pythonhosted.org/packages/6e/cc/ff84c03b95b430015d2166d9aae775a3985d757b94f6635010d0038d9241/multidict-6.6.3-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:0db58da8eafb514db832a1b44f8fa7906fdd102f7d982025f816a93ba45e3dcb", size = 233429, upload-time = "2025-06-30T15:51:13.533Z" }, - { url = "https://files.pythonhosted.org/packages/2e/f0/8cd49a0b37bdea673a4b793c2093f2f4ba8e7c9d6d7c9bd672fd6d38cd11/multidict-6.6.3-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:14117a41c8fdb3ee19c743b1c027da0736fdb79584d61a766da53d399b71176c", size = 243094, upload-time = "2025-06-30T15:51:14.815Z" }, - { url = "https://files.pythonhosted.org/packages/96/19/5d9a0cfdafe65d82b616a45ae950975820289069f885328e8185e64283c2/multidict-6.6.3-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:877443eaaabcd0b74ff32ebeed6f6176c71850feb7d6a1d2db65945256ea535c", size = 248957, upload-time = "2025-06-30T15:51:16.076Z" }, - { url = "https://files.pythonhosted.org/packages/e6/dc/c90066151da87d1e489f147b9b4327927241e65f1876702fafec6729c014/multidict-6.6.3-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:70b72e749a4f6e7ed8fb334fa8d8496384840319512746a5f42fa0aec79f4d61", size = 243590, upload-time = "2025-06-30T15:51:17.413Z" }, - { url = "https://files.pythonhosted.org/packages/ec/39/458afb0cccbb0ee9164365273be3e039efddcfcb94ef35924b7dbdb05db0/multidict-6.6.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:43571f785b86afd02b3855c5ac8e86ec921b760298d6f82ff2a61daf5a35330b", size = 237487, upload-time = "2025-06-30T15:51:19.039Z" }, - { url = "https://files.pythonhosted.org/packages/35/38/0016adac3990426610a081787011177e661875546b434f50a26319dc8372/multidict-6.6.3-cp310-cp310-win32.whl", hash = "sha256:20c5a0c3c13a15fd5ea86c42311859f970070e4e24de5a550e99d7c271d76318", size = 41390, upload-time = "2025-06-30T15:51:20.362Z" }, - { url = "https://files.pythonhosted.org/packages/f3/d2/17897a8f3f2c5363d969b4c635aa40375fe1f09168dc09a7826780bfb2a4/multidict-6.6.3-cp310-cp310-win_amd64.whl", hash = "sha256:ab0a34a007704c625e25a9116c6770b4d3617a071c8a7c30cd338dfbadfe6485", size = 45954, upload-time = "2025-06-30T15:51:21.383Z" }, - { url = "https://files.pythonhosted.org/packages/2d/5f/d4a717c1e457fe44072e33fa400d2b93eb0f2819c4d669381f925b7cba1f/multidict-6.6.3-cp310-cp310-win_arm64.whl", hash = "sha256:769841d70ca8bdd140a715746199fc6473414bd02efd678d75681d2d6a8986c5", size = 42981, upload-time = "2025-06-30T15:51:22.809Z" }, - { url = "https://files.pythonhosted.org/packages/08/f0/1a39863ced51f639c81a5463fbfa9eb4df59c20d1a8769ab9ef4ca57ae04/multidict-6.6.3-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:18f4eba0cbac3546b8ae31e0bbc55b02c801ae3cbaf80c247fcdd89b456ff58c", size = 76445, upload-time = "2025-06-30T15:51:24.01Z" }, - { url = "https://files.pythonhosted.org/packages/c9/0e/a7cfa451c7b0365cd844e90b41e21fab32edaa1e42fc0c9f68461ce44ed7/multidict-6.6.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ef43b5dd842382329e4797c46f10748d8c2b6e0614f46b4afe4aee9ac33159df", size = 44610, upload-time = "2025-06-30T15:51:25.158Z" }, - { url = "https://files.pythonhosted.org/packages/c6/bb/a14a4efc5ee748cc1904b0748be278c31b9295ce5f4d2ef66526f410b94d/multidict-6.6.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:bf9bd1fd5eec01494e0f2e8e446a74a85d5e49afb63d75a9934e4a5423dba21d", size = 44267, upload-time = "2025-06-30T15:51:26.326Z" }, - { url = "https://files.pythonhosted.org/packages/c2/f8/410677d563c2d55e063ef74fe578f9d53fe6b0a51649597a5861f83ffa15/multidict-6.6.3-cp311-cp311-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:5bd8d6f793a787153956cd35e24f60485bf0651c238e207b9a54f7458b16d539", size = 230004, upload-time = "2025-06-30T15:51:27.491Z" }, - { url = "https://files.pythonhosted.org/packages/fd/df/2b787f80059314a98e1ec6a4cc7576244986df3e56b3c755e6fc7c99e038/multidict-6.6.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1bf99b4daf908c73856bd87ee0a2499c3c9a3d19bb04b9c6025e66af3fd07462", size = 247196, upload-time = "2025-06-30T15:51:28.762Z" }, - { url = "https://files.pythonhosted.org/packages/05/f2/f9117089151b9a8ab39f9019620d10d9718eec2ac89e7ca9d30f3ec78e96/multidict-6.6.3-cp311-cp311-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:0b9e59946b49dafaf990fd9c17ceafa62976e8471a14952163d10a7a630413a9", size = 225337, upload-time = "2025-06-30T15:51:30.025Z" }, - { url = "https://files.pythonhosted.org/packages/93/2d/7115300ec5b699faa152c56799b089a53ed69e399c3c2d528251f0aeda1a/multidict-6.6.3-cp311-cp311-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:e2db616467070d0533832d204c54eea6836a5e628f2cb1e6dfd8cd6ba7277cb7", size = 257079, upload-time = "2025-06-30T15:51:31.716Z" }, - { url = "https://files.pythonhosted.org/packages/15/ea/ff4bab367623e39c20d3b07637225c7688d79e4f3cc1f3b9f89867677f9a/multidict-6.6.3-cp311-cp311-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:7394888236621f61dcdd25189b2768ae5cc280f041029a5bcf1122ac63df79f9", size = 255461, upload-time = "2025-06-30T15:51:33.029Z" }, - { url = "https://files.pythonhosted.org/packages/74/07/2c9246cda322dfe08be85f1b8739646f2c4c5113a1422d7a407763422ec4/multidict-6.6.3-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f114d8478733ca7388e7c7e0ab34b72547476b97009d643644ac33d4d3fe1821", size = 246611, upload-time = "2025-06-30T15:51:34.47Z" }, - { url = "https://files.pythonhosted.org/packages/a8/62/279c13d584207d5697a752a66ffc9bb19355a95f7659140cb1b3cf82180e/multidict-6.6.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cdf22e4db76d323bcdc733514bf732e9fb349707c98d341d40ebcc6e9318ef3d", size = 243102, upload-time = "2025-06-30T15:51:36.525Z" }, - { url = "https://files.pythonhosted.org/packages/69/cc/e06636f48c6d51e724a8bc8d9e1db5f136fe1df066d7cafe37ef4000f86a/multidict-6.6.3-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:e995a34c3d44ab511bfc11aa26869b9d66c2d8c799fa0e74b28a473a692532d6", size = 238693, upload-time = "2025-06-30T15:51:38.278Z" }, - { url = "https://files.pythonhosted.org/packages/89/a4/66c9d8fb9acf3b226cdd468ed009537ac65b520aebdc1703dd6908b19d33/multidict-6.6.3-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:766a4a5996f54361d8d5a9050140aa5362fe48ce51c755a50c0bc3706460c430", size = 246582, upload-time = "2025-06-30T15:51:39.709Z" }, - { url = "https://files.pythonhosted.org/packages/cf/01/c69e0317be556e46257826d5449feb4e6aa0d18573e567a48a2c14156f1f/multidict-6.6.3-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:3893a0d7d28a7fe6ca7a1f760593bc13038d1d35daf52199d431b61d2660602b", size = 253355, upload-time = "2025-06-30T15:51:41.013Z" }, - { url = "https://files.pythonhosted.org/packages/c0/da/9cc1da0299762d20e626fe0042e71b5694f9f72d7d3f9678397cbaa71b2b/multidict-6.6.3-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:934796c81ea996e61914ba58064920d6cad5d99140ac3167901eb932150e2e56", size = 247774, upload-time = "2025-06-30T15:51:42.291Z" }, - { url = "https://files.pythonhosted.org/packages/e6/91/b22756afec99cc31105ddd4a52f95ab32b1a4a58f4d417979c570c4a922e/multidict-6.6.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:9ed948328aec2072bc00f05d961ceadfd3e9bfc2966c1319aeaf7b7c21219183", size = 242275, upload-time = "2025-06-30T15:51:43.642Z" }, - { url = "https://files.pythonhosted.org/packages/be/f1/adcc185b878036a20399d5be5228f3cbe7f823d78985d101d425af35c800/multidict-6.6.3-cp311-cp311-win32.whl", hash = "sha256:9f5b28c074c76afc3e4c610c488e3493976fe0e596dd3db6c8ddfbb0134dcac5", size = 41290, upload-time = "2025-06-30T15:51:45.264Z" }, - { url = "https://files.pythonhosted.org/packages/e0/d4/27652c1c6526ea6b4f5ddd397e93f4232ff5de42bea71d339bc6a6cc497f/multidict-6.6.3-cp311-cp311-win_amd64.whl", hash = "sha256:bc7f6fbc61b1c16050a389c630da0b32fc6d4a3d191394ab78972bf5edc568c2", size = 45942, upload-time = "2025-06-30T15:51:46.377Z" }, - { url = "https://files.pythonhosted.org/packages/16/18/23f4932019804e56d3c2413e237f866444b774b0263bcb81df2fdecaf593/multidict-6.6.3-cp311-cp311-win_arm64.whl", hash = "sha256:d4e47d8faffaae822fb5cba20937c048d4f734f43572e7079298a6c39fb172cb", size = 42880, upload-time = "2025-06-30T15:51:47.561Z" }, - { url = "https://files.pythonhosted.org/packages/0e/a0/6b57988ea102da0623ea814160ed78d45a2645e4bbb499c2896d12833a70/multidict-6.6.3-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:056bebbeda16b2e38642d75e9e5310c484b7c24e3841dc0fb943206a72ec89d6", size = 76514, upload-time = "2025-06-30T15:51:48.728Z" }, - { url = "https://files.pythonhosted.org/packages/07/7a/d1e92665b0850c6c0508f101f9cf0410c1afa24973e1115fe9c6a185ebf7/multidict-6.6.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:e5f481cccb3c5c5e5de5d00b5141dc589c1047e60d07e85bbd7dea3d4580d63f", size = 45394, upload-time = "2025-06-30T15:51:49.986Z" }, - { url = "https://files.pythonhosted.org/packages/52/6f/dd104490e01be6ef8bf9573705d8572f8c2d2c561f06e3826b081d9e6591/multidict-6.6.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:10bea2ee839a759ee368b5a6e47787f399b41e70cf0c20d90dfaf4158dfb4e55", size = 43590, upload-time = "2025-06-30T15:51:51.331Z" }, - { url = "https://files.pythonhosted.org/packages/44/fe/06e0e01b1b0611e6581b7fd5a85b43dacc08b6cea3034f902f383b0873e5/multidict-6.6.3-cp312-cp312-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:2334cfb0fa9549d6ce2c21af2bfbcd3ac4ec3646b1b1581c88e3e2b1779ec92b", size = 237292, upload-time = "2025-06-30T15:51:52.584Z" }, - { url = "https://files.pythonhosted.org/packages/ce/71/4f0e558fb77696b89c233c1ee2d92f3e1d5459070a0e89153c9e9e804186/multidict-6.6.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b8fee016722550a2276ca2cb5bb624480e0ed2bd49125b2b73b7010b9090e888", size = 258385, upload-time = "2025-06-30T15:51:53.913Z" }, - { url = "https://files.pythonhosted.org/packages/e3/25/cca0e68228addad24903801ed1ab42e21307a1b4b6dd2cf63da5d3ae082a/multidict-6.6.3-cp312-cp312-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:e5511cb35f5c50a2db21047c875eb42f308c5583edf96bd8ebf7d770a9d68f6d", size = 242328, upload-time = "2025-06-30T15:51:55.672Z" }, - { url = "https://files.pythonhosted.org/packages/6e/a3/46f2d420d86bbcb8fe660b26a10a219871a0fbf4d43cb846a4031533f3e0/multidict-6.6.3-cp312-cp312-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:712b348f7f449948e0a6c4564a21c7db965af900973a67db432d724619b3c680", size = 268057, upload-time = "2025-06-30T15:51:57.037Z" }, - { url = "https://files.pythonhosted.org/packages/9e/73/1c743542fe00794a2ec7466abd3f312ccb8fad8dff9f36d42e18fb1ec33e/multidict-6.6.3-cp312-cp312-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:e4e15d2138ee2694e038e33b7c3da70e6b0ad8868b9f8094a72e1414aeda9c1a", size = 269341, upload-time = "2025-06-30T15:51:59.111Z" }, - { url = "https://files.pythonhosted.org/packages/a4/11/6ec9dcbe2264b92778eeb85407d1df18812248bf3506a5a1754bc035db0c/multidict-6.6.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8df25594989aebff8a130f7899fa03cbfcc5d2b5f4a461cf2518236fe6f15961", size = 256081, upload-time = "2025-06-30T15:52:00.533Z" }, - { url = "https://files.pythonhosted.org/packages/9b/2b/631b1e2afeb5f1696846d747d36cda075bfdc0bc7245d6ba5c319278d6c4/multidict-6.6.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:159ca68bfd284a8860f8d8112cf0521113bffd9c17568579e4d13d1f1dc76b65", size = 253581, upload-time = "2025-06-30T15:52:02.43Z" }, - { url = "https://files.pythonhosted.org/packages/bf/0e/7e3b93f79efeb6111d3bf9a1a69e555ba1d07ad1c11bceb56b7310d0d7ee/multidict-6.6.3-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:e098c17856a8c9ade81b4810888c5ad1914099657226283cab3062c0540b0643", size = 250750, upload-time = "2025-06-30T15:52:04.26Z" }, - { url = "https://files.pythonhosted.org/packages/ad/9e/086846c1d6601948e7de556ee464a2d4c85e33883e749f46b9547d7b0704/multidict-6.6.3-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:67c92ed673049dec52d7ed39f8cf9ebbadf5032c774058b4406d18c8f8fe7063", size = 251548, upload-time = "2025-06-30T15:52:06.002Z" }, - { url = "https://files.pythonhosted.org/packages/8c/7b/86ec260118e522f1a31550e87b23542294880c97cfbf6fb18cc67b044c66/multidict-6.6.3-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:bd0578596e3a835ef451784053cfd327d607fc39ea1a14812139339a18a0dbc3", size = 262718, upload-time = "2025-06-30T15:52:07.707Z" }, - { url = "https://files.pythonhosted.org/packages/8c/bd/22ce8f47abb0be04692c9fc4638508b8340987b18691aa7775d927b73f72/multidict-6.6.3-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:346055630a2df2115cd23ae271910b4cae40f4e336773550dca4889b12916e75", size = 259603, upload-time = "2025-06-30T15:52:09.58Z" }, - { url = "https://files.pythonhosted.org/packages/07/9c/91b7ac1691be95cd1f4a26e36a74b97cda6aa9820632d31aab4410f46ebd/multidict-6.6.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:555ff55a359302b79de97e0468e9ee80637b0de1fce77721639f7cd9440b3a10", size = 251351, upload-time = "2025-06-30T15:52:10.947Z" }, - { url = "https://files.pythonhosted.org/packages/6f/5c/4d7adc739884f7a9fbe00d1eac8c034023ef8bad71f2ebe12823ca2e3649/multidict-6.6.3-cp312-cp312-win32.whl", hash = "sha256:73ab034fb8d58ff85c2bcbadc470efc3fafeea8affcf8722855fb94557f14cc5", size = 41860, upload-time = "2025-06-30T15:52:12.334Z" }, - { url = "https://files.pythonhosted.org/packages/6a/a3/0fbc7afdf7cb1aa12a086b02959307848eb6bcc8f66fcb66c0cb57e2a2c1/multidict-6.6.3-cp312-cp312-win_amd64.whl", hash = "sha256:04cbcce84f63b9af41bad04a54d4cc4e60e90c35b9e6ccb130be2d75b71f8c17", size = 45982, upload-time = "2025-06-30T15:52:13.6Z" }, - { url = "https://files.pythonhosted.org/packages/b8/95/8c825bd70ff9b02462dc18d1295dd08d3e9e4eb66856d292ffa62cfe1920/multidict-6.6.3-cp312-cp312-win_arm64.whl", hash = "sha256:0f1130b896ecb52d2a1e615260f3ea2af55fa7dc3d7c3003ba0c3121a759b18b", size = 43210, upload-time = "2025-06-30T15:52:14.893Z" }, - { url = "https://files.pythonhosted.org/packages/52/1d/0bebcbbb4f000751fbd09957257903d6e002943fc668d841a4cf2fb7f872/multidict-6.6.3-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:540d3c06d48507357a7d57721e5094b4f7093399a0106c211f33540fdc374d55", size = 75843, upload-time = "2025-06-30T15:52:16.155Z" }, - { url = "https://files.pythonhosted.org/packages/07/8f/cbe241b0434cfe257f65c2b1bcf9e8d5fb52bc708c5061fb29b0fed22bdf/multidict-6.6.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9c19cea2a690f04247d43f366d03e4eb110a0dc4cd1bbeee4d445435428ed35b", size = 45053, upload-time = "2025-06-30T15:52:17.429Z" }, - { url = "https://files.pythonhosted.org/packages/32/d2/0b3b23f9dbad5b270b22a3ac3ea73ed0a50ef2d9a390447061178ed6bdb8/multidict-6.6.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:7af039820cfd00effec86bda5d8debef711a3e86a1d3772e85bea0f243a4bd65", size = 43273, upload-time = "2025-06-30T15:52:19.346Z" }, - { url = "https://files.pythonhosted.org/packages/fd/fe/6eb68927e823999e3683bc49678eb20374ba9615097d085298fd5b386564/multidict-6.6.3-cp313-cp313-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:500b84f51654fdc3944e936f2922114349bf8fdcac77c3092b03449f0e5bc2b3", size = 237124, upload-time = "2025-06-30T15:52:20.773Z" }, - { url = "https://files.pythonhosted.org/packages/e7/ab/320d8507e7726c460cb77117848b3834ea0d59e769f36fdae495f7669929/multidict-6.6.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f3fc723ab8a5c5ed6c50418e9bfcd8e6dceba6c271cee6728a10a4ed8561520c", size = 256892, upload-time = "2025-06-30T15:52:22.242Z" }, - { url = "https://files.pythonhosted.org/packages/76/60/38ee422db515ac69834e60142a1a69111ac96026e76e8e9aa347fd2e4591/multidict-6.6.3-cp313-cp313-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:94c47ea3ade005b5976789baaed66d4de4480d0a0bf31cef6edaa41c1e7b56a6", size = 240547, upload-time = "2025-06-30T15:52:23.736Z" }, - { url = "https://files.pythonhosted.org/packages/27/fb/905224fde2dff042b030c27ad95a7ae744325cf54b890b443d30a789b80e/multidict-6.6.3-cp313-cp313-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:dbc7cf464cc6d67e83e136c9f55726da3a30176f020a36ead246eceed87f1cd8", size = 266223, upload-time = "2025-06-30T15:52:25.185Z" }, - { url = "https://files.pythonhosted.org/packages/76/35/dc38ab361051beae08d1a53965e3e1a418752fc5be4d3fb983c5582d8784/multidict-6.6.3-cp313-cp313-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:900eb9f9da25ada070f8ee4a23f884e0ee66fe4e1a38c3af644256a508ad81ca", size = 267262, upload-time = "2025-06-30T15:52:26.969Z" }, - { url = "https://files.pythonhosted.org/packages/1f/a3/0a485b7f36e422421b17e2bbb5a81c1af10eac1d4476f2ff92927c730479/multidict-6.6.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7c6df517cf177da5d47ab15407143a89cd1a23f8b335f3a28d57e8b0a3dbb884", size = 254345, upload-time = "2025-06-30T15:52:28.467Z" }, - { url = "https://files.pythonhosted.org/packages/b4/59/bcdd52c1dab7c0e0d75ff19cac751fbd5f850d1fc39172ce809a74aa9ea4/multidict-6.6.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:4ef421045f13879e21c994b36e728d8e7d126c91a64b9185810ab51d474f27e7", size = 252248, upload-time = "2025-06-30T15:52:29.938Z" }, - { url = "https://files.pythonhosted.org/packages/bb/a4/2d96aaa6eae8067ce108d4acee6f45ced5728beda55c0f02ae1072c730d1/multidict-6.6.3-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:6c1e61bb4f80895c081790b6b09fa49e13566df8fbff817da3f85b3a8192e36b", size = 250115, upload-time = "2025-06-30T15:52:31.416Z" }, - { url = "https://files.pythonhosted.org/packages/25/d2/ed9f847fa5c7d0677d4f02ea2c163d5e48573de3f57bacf5670e43a5ffaa/multidict-6.6.3-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:e5e8523bb12d7623cd8300dbd91b9e439a46a028cd078ca695eb66ba31adee3c", size = 249649, upload-time = "2025-06-30T15:52:32.996Z" }, - { url = "https://files.pythonhosted.org/packages/1f/af/9155850372563fc550803d3f25373308aa70f59b52cff25854086ecb4a79/multidict-6.6.3-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:ef58340cc896219e4e653dade08fea5c55c6df41bcc68122e3be3e9d873d9a7b", size = 261203, upload-time = "2025-06-30T15:52:34.521Z" }, - { url = "https://files.pythonhosted.org/packages/36/2f/c6a728f699896252cf309769089568a33c6439626648843f78743660709d/multidict-6.6.3-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:fc9dc435ec8699e7b602b94fe0cd4703e69273a01cbc34409af29e7820f777f1", size = 258051, upload-time = "2025-06-30T15:52:35.999Z" }, - { url = "https://files.pythonhosted.org/packages/d0/60/689880776d6b18fa2b70f6cc74ff87dd6c6b9b47bd9cf74c16fecfaa6ad9/multidict-6.6.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9e864486ef4ab07db5e9cb997bad2b681514158d6954dd1958dfb163b83d53e6", size = 249601, upload-time = "2025-06-30T15:52:37.473Z" }, - { url = "https://files.pythonhosted.org/packages/75/5e/325b11f2222a549019cf2ef879c1f81f94a0d40ace3ef55cf529915ba6cc/multidict-6.6.3-cp313-cp313-win32.whl", hash = "sha256:5633a82fba8e841bc5c5c06b16e21529573cd654f67fd833650a215520a6210e", size = 41683, upload-time = "2025-06-30T15:52:38.927Z" }, - { url = "https://files.pythonhosted.org/packages/b1/ad/cf46e73f5d6e3c775cabd2a05976547f3f18b39bee06260369a42501f053/multidict-6.6.3-cp313-cp313-win_amd64.whl", hash = "sha256:e93089c1570a4ad54c3714a12c2cef549dc9d58e97bcded193d928649cab78e9", size = 45811, upload-time = "2025-06-30T15:52:40.207Z" }, - { url = "https://files.pythonhosted.org/packages/c5/c9/2e3fe950db28fb7c62e1a5f46e1e38759b072e2089209bc033c2798bb5ec/multidict-6.6.3-cp313-cp313-win_arm64.whl", hash = "sha256:c60b401f192e79caec61f166da9c924e9f8bc65548d4246842df91651e83d600", size = 43056, upload-time = "2025-06-30T15:52:41.575Z" }, - { url = "https://files.pythonhosted.org/packages/3a/58/aaf8114cf34966e084a8cc9517771288adb53465188843d5a19862cb6dc3/multidict-6.6.3-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:02fd8f32d403a6ff13864b0851f1f523d4c988051eea0471d4f1fd8010f11134", size = 82811, upload-time = "2025-06-30T15:52:43.281Z" }, - { url = "https://files.pythonhosted.org/packages/71/af/5402e7b58a1f5b987a07ad98f2501fdba2a4f4b4c30cf114e3ce8db64c87/multidict-6.6.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:f3aa090106b1543f3f87b2041eef3c156c8da2aed90c63a2fbed62d875c49c37", size = 48304, upload-time = "2025-06-30T15:52:45.026Z" }, - { url = "https://files.pythonhosted.org/packages/39/65/ab3c8cafe21adb45b24a50266fd747147dec7847425bc2a0f6934b3ae9ce/multidict-6.6.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:e924fb978615a5e33ff644cc42e6aa241effcf4f3322c09d4f8cebde95aff5f8", size = 46775, upload-time = "2025-06-30T15:52:46.459Z" }, - { url = "https://files.pythonhosted.org/packages/49/ba/9fcc1b332f67cc0c0c8079e263bfab6660f87fe4e28a35921771ff3eea0d/multidict-6.6.3-cp313-cp313t-manylinux1_i686.manylinux2014_i686.manylinux_2_17_i686.manylinux_2_5_i686.whl", hash = "sha256:b9fe5a0e57c6dbd0e2ce81ca66272282c32cd11d31658ee9553849d91289e1c1", size = 229773, upload-time = "2025-06-30T15:52:47.88Z" }, - { url = "https://files.pythonhosted.org/packages/a4/14/0145a251f555f7c754ce2dcbcd012939bbd1f34f066fa5d28a50e722a054/multidict-6.6.3-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b24576f208793ebae00280c59927c3b7c2a3b1655e443a25f753c4611bc1c373", size = 250083, upload-time = "2025-06-30T15:52:49.366Z" }, - { url = "https://files.pythonhosted.org/packages/9e/d4/d5c0bd2bbb173b586c249a151a26d2fb3ec7d53c96e42091c9fef4e1f10c/multidict-6.6.3-cp313-cp313t-manylinux2014_armv7l.manylinux_2_17_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:135631cb6c58eac37d7ac0df380294fecdc026b28837fa07c02e459c7fb9c54e", size = 228980, upload-time = "2025-06-30T15:52:50.903Z" }, - { url = "https://files.pythonhosted.org/packages/21/32/c9a2d8444a50ec48c4733ccc67254100c10e1c8ae8e40c7a2d2183b59b97/multidict-6.6.3-cp313-cp313t-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:274d416b0df887aef98f19f21578653982cfb8a05b4e187d4a17103322eeaf8f", size = 257776, upload-time = "2025-06-30T15:52:52.764Z" }, - { url = "https://files.pythonhosted.org/packages/68/d0/14fa1699f4ef629eae08ad6201c6b476098f5efb051b296f4c26be7a9fdf/multidict-6.6.3-cp313-cp313t-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:e252017a817fad7ce05cafbe5711ed40faeb580e63b16755a3a24e66fa1d87c0", size = 256882, upload-time = "2025-06-30T15:52:54.596Z" }, - { url = "https://files.pythonhosted.org/packages/da/88/84a27570fbe303c65607d517a5f147cd2fc046c2d1da02b84b17b9bdc2aa/multidict-6.6.3-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2e4cc8d848cd4fe1cdee28c13ea79ab0ed37fc2e89dd77bac86a2e7959a8c3bc", size = 247816, upload-time = "2025-06-30T15:52:56.175Z" }, - { url = "https://files.pythonhosted.org/packages/1c/60/dca352a0c999ce96a5d8b8ee0b2b9f729dcad2e0b0c195f8286269a2074c/multidict-6.6.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:9e236a7094b9c4c1b7585f6b9cca34b9d833cf079f7e4c49e6a4a6ec9bfdc68f", size = 245341, upload-time = "2025-06-30T15:52:57.752Z" }, - { url = "https://files.pythonhosted.org/packages/50/ef/433fa3ed06028f03946f3993223dada70fb700f763f70c00079533c34578/multidict-6.6.3-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:e0cb0ab69915c55627c933f0b555a943d98ba71b4d1c57bc0d0a66e2567c7471", size = 235854, upload-time = "2025-06-30T15:52:59.74Z" }, - { url = "https://files.pythonhosted.org/packages/1b/1f/487612ab56fbe35715320905215a57fede20de7db40a261759690dc80471/multidict-6.6.3-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:81ef2f64593aba09c5212a3d0f8c906a0d38d710a011f2f42759704d4557d3f2", size = 243432, upload-time = "2025-06-30T15:53:01.602Z" }, - { url = "https://files.pythonhosted.org/packages/da/6f/ce8b79de16cd885c6f9052c96a3671373d00c59b3ee635ea93e6e81b8ccf/multidict-6.6.3-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:b9cbc60010de3562545fa198bfc6d3825df430ea96d2cc509c39bd71e2e7d648", size = 252731, upload-time = "2025-06-30T15:53:03.517Z" }, - { url = "https://files.pythonhosted.org/packages/bb/fe/a2514a6aba78e5abefa1624ca85ae18f542d95ac5cde2e3815a9fbf369aa/multidict-6.6.3-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:70d974eaaa37211390cd02ef93b7e938de564bbffa866f0b08d07e5e65da783d", size = 247086, upload-time = "2025-06-30T15:53:05.48Z" }, - { url = "https://files.pythonhosted.org/packages/8c/22/b788718d63bb3cce752d107a57c85fcd1a212c6c778628567c9713f9345a/multidict-6.6.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:3713303e4a6663c6d01d648a68f2848701001f3390a030edaaf3fc949c90bf7c", size = 243338, upload-time = "2025-06-30T15:53:07.522Z" }, - { url = "https://files.pythonhosted.org/packages/22/d6/fdb3d0670819f2228f3f7d9af613d5e652c15d170c83e5f1c94fbc55a25b/multidict-6.6.3-cp313-cp313t-win32.whl", hash = "sha256:639ecc9fe7cd73f2495f62c213e964843826f44505a3e5d82805aa85cac6f89e", size = 47812, upload-time = "2025-06-30T15:53:09.263Z" }, - { url = "https://files.pythonhosted.org/packages/b6/d6/a9d2c808f2c489ad199723197419207ecbfbc1776f6e155e1ecea9c883aa/multidict-6.6.3-cp313-cp313t-win_amd64.whl", hash = "sha256:9f97e181f344a0ef3881b573d31de8542cc0dbc559ec68c8f8b5ce2c2e91646d", size = 53011, upload-time = "2025-06-30T15:53:11.038Z" }, - { url = "https://files.pythonhosted.org/packages/f2/40/b68001cba8188dd267590a111f9661b6256debc327137667e832bf5d66e8/multidict-6.6.3-cp313-cp313t-win_arm64.whl", hash = "sha256:ce8b7693da41a3c4fde5871c738a81490cea5496c671d74374c8ab889e1834fb", size = 45254, upload-time = "2025-06-30T15:53:12.421Z" }, - { url = "https://files.pythonhosted.org/packages/d8/30/9aec301e9772b098c1f5c0ca0279237c9766d94b97802e9888010c64b0ed/multidict-6.6.3-py3-none-any.whl", hash = "sha256:8db10f29c7541fc5da4defd8cd697e1ca429db743fa716325f236079b96f775a", size = 12313, upload-time = "2025-06-30T15:53:45.437Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/82/4a/7874ca44a1c9b23796c767dd94159f6c17e31c0e7d090552a1c623247d82/multidict-6.2.0.tar.gz", hash = "sha256:0085b0afb2446e57050140240a8595846ed64d1cbd26cef936bfab3192c673b8", size = 71066, upload-time = "2025-03-17T16:55:54.689Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/2d/ca/3ae4d9c9ba78e7bcb63e3f12974b8fa16b9a20de44e9785f5d291ccb823c/multidict-6.2.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:b9f6392d98c0bd70676ae41474e2eecf4c7150cb419237a41f8f96043fcb81d1", size = 49238, upload-time = "2025-03-17T16:53:32.192Z" }, + { url = "https://files.pythonhosted.org/packages/25/a4/55e595d2df586e442c85b2610542d1e14def4c6f641761125d35fb38f87c/multidict-6.2.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:3501621d5e86f1a88521ea65d5cad0a0834c77b26f193747615b7c911e5422d2", size = 29748, upload-time = "2025-03-17T16:53:34.057Z" }, + { url = "https://files.pythonhosted.org/packages/35/6f/09bc361a34bbf953e9897f69823f9c4b46aec0aaed6ec94ce63093ede317/multidict-6.2.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:32ed748ff9ac682eae7859790d3044b50e3076c7d80e17a44239683769ff485e", size = 30026, upload-time = "2025-03-17T16:53:35.378Z" }, + { url = "https://files.pythonhosted.org/packages/b6/c7/5b51816f7c38049fc50786f46e63c009e6fecd1953fbbafa8bfe4e2eb39d/multidict-6.2.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cc826b9a8176e686b67aa60fd6c6a7047b0461cae5591ea1dc73d28f72332a8a", size = 132393, upload-time = "2025-03-17T16:53:37.684Z" }, + { url = "https://files.pythonhosted.org/packages/1a/21/c51aca665afa93b397d2c47369f6c267193977611a55a7c9d8683dc095bc/multidict-6.2.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:214207dcc7a6221d9942f23797fe89144128a71c03632bf713d918db99bd36de", size = 139237, upload-time = "2025-03-17T16:53:39.287Z" }, + { url = "https://files.pythonhosted.org/packages/2e/9b/a7b91f8ed63314e7a3c276b4ca90ae5d0267a584ca2e42106baa728622d6/multidict-6.2.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:05fefbc3cddc4e36da209a5e49f1094bbece9a581faa7f3589201fd95df40e5d", size = 134920, upload-time = "2025-03-17T16:53:40.6Z" }, + { url = "https://files.pythonhosted.org/packages/c8/84/4b590a121b1009fe79d1ae5875b4aa9339d37d23e368dd3bcf5e36d27452/multidict-6.2.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e851e6363d0dbe515d8de81fd544a2c956fdec6f8a049739562286727d4a00c3", size = 129764, upload-time = "2025-03-17T16:53:41.881Z" }, + { url = "https://files.pythonhosted.org/packages/b8/de/831be406b5ab0dc0d25430ddf597c6ce1a2e23a4991363f1ca48f16fb817/multidict-6.2.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:32c9b4878f48be3e75808ea7e499d6223b1eea6d54c487a66bc10a1871e3dc6a", size = 122121, upload-time = "2025-03-17T16:53:43.848Z" }, + { url = "https://files.pythonhosted.org/packages/fa/2f/892334f4d3efc7cd11e3a64dc922a85611627380ee2de3d0627ac159a975/multidict-6.2.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:7243c5a6523c5cfeca76e063efa5f6a656d1d74c8b1fc64b2cd1e84e507f7e2a", size = 135640, upload-time = "2025-03-17T16:53:45.698Z" }, + { url = "https://files.pythonhosted.org/packages/6c/53/bf91c5fdede9406247dcbceaa9d7e7fa08e4d0e27fa3c76a0dab126bc6b2/multidict-6.2.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:0e5a644e50ef9fb87878d4d57907f03a12410d2aa3b93b3acdf90a741df52c49", size = 129655, upload-time = "2025-03-17T16:53:47.322Z" }, + { url = "https://files.pythonhosted.org/packages/d4/7a/f98e1c5d14c1bbbb83025a69da9a37344f7556c09fef39979cf62b464d60/multidict-6.2.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:0dc25a3293c50744796e87048de5e68996104d86d940bb24bc3ec31df281b191", size = 140691, upload-time = "2025-03-17T16:53:48.634Z" }, + { url = "https://files.pythonhosted.org/packages/dd/c9/af0ab78b53d5b769bc1fa751e53cc7356cef422bd1cf38ed653985a46ddf/multidict-6.2.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:a49994481b99cd7dedde07f2e7e93b1d86c01c0fca1c32aded18f10695ae17eb", size = 135254, upload-time = "2025-03-17T16:53:49.866Z" }, + { url = "https://files.pythonhosted.org/packages/c9/53/28cc971b17e25487a089bcf720fe284478f264a6fc619427ddf7145fcb2b/multidict-6.2.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:641cf2e3447c9ecff2f7aa6e9eee9eaa286ea65d57b014543a4911ff2799d08a", size = 133620, upload-time = "2025-03-17T16:53:51.713Z" }, + { url = "https://files.pythonhosted.org/packages/b6/9a/d7637fbe1d5928b9f6a33ce36c2ff37e0aab9aa22f5fc9552fd75fe7f364/multidict-6.2.0-cp310-cp310-win32.whl", hash = "sha256:0c383d28857f66f5aebe3e91d6cf498da73af75fbd51cedbe1adfb85e90c0460", size = 27044, upload-time = "2025-03-17T16:53:53.859Z" }, + { url = "https://files.pythonhosted.org/packages/4e/11/04758cc18a51227dbb350a8a25c7db0620d63fb23db5b8d1f87762f05cbe/multidict-6.2.0-cp310-cp310-win_amd64.whl", hash = "sha256:a33273a541f1e1a8219b2a4ed2de355848ecc0254264915b9290c8d2de1c74e1", size = 29149, upload-time = "2025-03-17T16:53:55.076Z" }, + { url = "https://files.pythonhosted.org/packages/97/aa/879cf5581bd56c19f1bd2682ee4ecfd4085a404668d4ee5138b0a08eaf2a/multidict-6.2.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:84e87a7d75fa36839a3a432286d719975362d230c70ebfa0948549cc38bd5b46", size = 49125, upload-time = "2025-03-17T16:53:56.148Z" }, + { url = "https://files.pythonhosted.org/packages/9e/d8/e6d47c166c13c48be8efb9720afe0f5cdc4da4687547192cbc3c03903041/multidict-6.2.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:8de4d42dffd5ced9117af2ce66ba8722402541a3aa98ffdf78dde92badb68932", size = 29689, upload-time = "2025-03-17T16:53:57.381Z" }, + { url = "https://files.pythonhosted.org/packages/a4/20/f3f0a2ca142c81100b6d4cbf79505961b54181d66157615bba3955304442/multidict-6.2.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:e7d91a230c7f8af86c904a5a992b8c064b66330544693fd6759c3d6162382ecf", size = 29975, upload-time = "2025-03-17T16:53:58.549Z" }, + { url = "https://files.pythonhosted.org/packages/ab/2d/1724972c7aeb7aa1916a3276cb32f9c39e186456ee7ed621504e7a758322/multidict-6.2.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9f6cad071960ba1914fa231677d21b1b4a3acdcce463cee41ea30bc82e6040cf", size = 135688, upload-time = "2025-03-17T16:53:59.653Z" }, + { url = "https://files.pythonhosted.org/packages/1a/08/ea54e7e245aaf0bb1c758578e5afba394ffccb8bd80d229a499b9b83f2b1/multidict-6.2.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0f74f2fc51555f4b037ef278efc29a870d327053aba5cb7d86ae572426c7cccc", size = 142703, upload-time = "2025-03-17T16:54:01.552Z" }, + { url = "https://files.pythonhosted.org/packages/97/76/960dee0424f38c71eda54101ee1ca7bb47c5250ed02f7b3e8e50b1ce0603/multidict-6.2.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:14ed9ed1bfedd72a877807c71113deac292bf485159a29025dfdc524c326f3e1", size = 138559, upload-time = "2025-03-17T16:54:02.973Z" }, + { url = "https://files.pythonhosted.org/packages/d0/35/969fd792e2e72801d80307f0a14f5b19c066d4a51d34dded22c71401527d/multidict-6.2.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4ac3fcf9a2d369bd075b2c2965544036a27ccd277fc3c04f708338cc57533081", size = 133312, upload-time = "2025-03-17T16:54:04.265Z" }, + { url = "https://files.pythonhosted.org/packages/a4/b8/f96657a2f744d577cfda5a7edf9da04a731b80d3239eafbfe7ca4d944695/multidict-6.2.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2fc6af8e39f7496047c7876314f4317736eac82bf85b54c7c76cf1a6f8e35d98", size = 125652, upload-time = "2025-03-17T16:54:05.814Z" }, + { url = "https://files.pythonhosted.org/packages/35/9d/97696d052297d8e2e08195a25c7aae873a6186c147b7635f979edbe3acde/multidict-6.2.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5f8cb1329f42fadfb40d6211e5ff568d71ab49be36e759345f91c69d1033d633", size = 139015, upload-time = "2025-03-17T16:54:07.791Z" }, + { url = "https://files.pythonhosted.org/packages/31/a0/5c106e28d42f20288c10049bc6647364287ba049dc00d6ae4f1584eb1bd1/multidict-6.2.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:5389445f0173c197f4a3613713b5fb3f3879df1ded2a1a2e4bc4b5b9c5441b7e", size = 132437, upload-time = "2025-03-17T16:54:09.491Z" }, + { url = "https://files.pythonhosted.org/packages/55/57/d5c60c075fef73422ae3b8f914221485b9ff15000b2db657c03bd190aee0/multidict-6.2.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:94a7bb972178a8bfc4055db80c51efd24baefaced5e51c59b0d598a004e8305d", size = 144037, upload-time = "2025-03-17T16:54:11.189Z" }, + { url = "https://files.pythonhosted.org/packages/eb/56/a23f599c697a455bf65ecb0f69a5b052d6442c567d380ed423f816246824/multidict-6.2.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:da51d8928ad8b4244926fe862ba1795f0b6e68ed8c42cd2f822d435db9c2a8f4", size = 138535, upload-time = "2025-03-17T16:54:12.453Z" }, + { url = "https://files.pythonhosted.org/packages/34/3a/a06ff9b5899090f4bbdbf09e237964c76cecfe75d2aa921e801356314017/multidict-6.2.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:063be88bd684782a0715641de853e1e58a2f25b76388538bd62d974777ce9bc2", size = 136885, upload-time = "2025-03-17T16:54:13.648Z" }, + { url = "https://files.pythonhosted.org/packages/d6/28/489c0eca1df3800cb5d0a66278d5dd2a4deae747a41d1cf553e6a4c0a984/multidict-6.2.0-cp311-cp311-win32.whl", hash = "sha256:52b05e21ff05729fbea9bc20b3a791c3c11da61649ff64cce8257c82a020466d", size = 27044, upload-time = "2025-03-17T16:54:16.495Z" }, + { url = "https://files.pythonhosted.org/packages/d0/b5/c7cd5ba9581add40bc743980f82426b90d9f42db0b56502011f1b3c929df/multidict-6.2.0-cp311-cp311-win_amd64.whl", hash = "sha256:1e2a2193d3aa5cbf5758f6d5680a52aa848e0cf611da324f71e5e48a9695cc86", size = 29145, upload-time = "2025-03-17T16:54:18.009Z" }, + { url = "https://files.pythonhosted.org/packages/a4/e2/0153a8db878aef9b2397be81e62cbc3b32ca9b94e0f700b103027db9d506/multidict-6.2.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:437c33561edb6eb504b5a30203daf81d4a9b727e167e78b0854d9a4e18e8950b", size = 49204, upload-time = "2025-03-17T16:54:19.193Z" }, + { url = "https://files.pythonhosted.org/packages/bb/9d/5ccb3224a976d1286f360bb4e89e67b7cdfb87336257fc99be3c17f565d7/multidict-6.2.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:9f49585f4abadd2283034fc605961f40c638635bc60f5162276fec075f2e37a4", size = 29807, upload-time = "2025-03-17T16:54:20.398Z" }, + { url = "https://files.pythonhosted.org/packages/62/32/ef20037f51b84b074a89bab5af46d4565381c3f825fc7cbfc19c1ee156be/multidict-6.2.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:5dd7106d064d05896ce28c97da3f46caa442fe5a43bc26dfb258e90853b39b44", size = 30000, upload-time = "2025-03-17T16:54:21.845Z" }, + { url = "https://files.pythonhosted.org/packages/97/81/b0a7560bfc3ec72606232cd7e60159e09b9cf29e66014d770c1315868fa2/multidict-6.2.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e25b11a0417475f093d0f0809a149aff3943c2c56da50fdf2c3c88d57fe3dfbd", size = 131820, upload-time = "2025-03-17T16:54:23.404Z" }, + { url = "https://files.pythonhosted.org/packages/49/3b/768bfc0e41179fbccd3a22925329a11755b7fdd53bec66dbf6b8772f0bce/multidict-6.2.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ac380cacdd3b183338ba63a144a34e9044520a6fb30c58aa14077157a033c13e", size = 136272, upload-time = "2025-03-17T16:54:24.636Z" }, + { url = "https://files.pythonhosted.org/packages/71/ac/fd2be3fe98ff54e7739448f771ba730d42036de0870737db9ae34bb8efe9/multidict-6.2.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:61d5541f27533f803a941d3a3f8a3d10ed48c12cf918f557efcbf3cd04ef265c", size = 135233, upload-time = "2025-03-17T16:54:25.884Z" }, + { url = "https://files.pythonhosted.org/packages/93/76/1657047da771315911a927b364a32dafce4135b79b64208ce4ac69525c56/multidict-6.2.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:facaf11f21f3a4c51b62931feb13310e6fe3475f85e20d9c9fdce0d2ea561b87", size = 132861, upload-time = "2025-03-17T16:54:27.154Z" }, + { url = "https://files.pythonhosted.org/packages/19/a5/9f07ffb9bf68b8aaa406c2abee27ad87e8b62a60551587b8e59ee91aea84/multidict-6.2.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:095a2eabe8c43041d3e6c2cb8287a257b5f1801c2d6ebd1dd877424f1e89cf29", size = 122166, upload-time = "2025-03-17T16:54:28.417Z" }, + { url = "https://files.pythonhosted.org/packages/95/23/b5ce3318d9d6c8f105c3679510f9d7202980545aad8eb4426313bd8da3ee/multidict-6.2.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:a0cc398350ef31167e03f3ca7c19313d4e40a662adcb98a88755e4e861170bdd", size = 136052, upload-time = "2025-03-17T16:54:29.689Z" }, + { url = "https://files.pythonhosted.org/packages/ce/5c/02cffec58ffe120873dce520af593415b91cc324be0345f534ad3637da4e/multidict-6.2.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:7c611345bbe7cb44aabb877cb94b63e86f2d0db03e382667dbd037866d44b4f8", size = 130094, upload-time = "2025-03-17T16:54:31.137Z" }, + { url = "https://files.pythonhosted.org/packages/49/f3/3b19a83f4ebf53a3a2a0435f3e447aa227b242ba3fd96a92404b31fb3543/multidict-6.2.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:8cd1a0644ccaf27e9d2f6d9c9474faabee21f0578fe85225cc5af9a61e1653df", size = 140962, upload-time = "2025-03-17T16:54:32.415Z" }, + { url = "https://files.pythonhosted.org/packages/cc/1a/c916b54fb53168c24cb6a3a0795fd99d0a59a0ea93fa9f6edeff5565cb20/multidict-6.2.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:89b3857652183b8206a891168af47bac10b970d275bba1f6ee46565a758c078d", size = 138082, upload-time = "2025-03-17T16:54:33.655Z" }, + { url = "https://files.pythonhosted.org/packages/ef/1a/dcb7fb18f64b3727c61f432c1e1a0d52b3924016124e4bbc8a7d2e4fa57b/multidict-6.2.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:125dd82b40f8c06d08d87b3510beaccb88afac94e9ed4a6f6c71362dc7dbb04b", size = 136019, upload-time = "2025-03-17T16:54:35.086Z" }, + { url = "https://files.pythonhosted.org/packages/fb/02/7695485375106f5c542574f70e1968c391f86fa3efc9f1fd76aac0af7237/multidict-6.2.0-cp312-cp312-win32.whl", hash = "sha256:76b34c12b013d813e6cb325e6bd4f9c984db27758b16085926bbe7ceeaace626", size = 26676, upload-time = "2025-03-17T16:54:36.32Z" }, + { url = "https://files.pythonhosted.org/packages/3c/f5/f147000fe1f4078160157b15b0790fff0513646b0f9b7404bf34007a9b44/multidict-6.2.0-cp312-cp312-win_amd64.whl", hash = "sha256:0b183a959fb88ad1be201de2c4bdf52fa8e46e6c185d76201286a97b6f5ee65c", size = 28899, upload-time = "2025-03-17T16:54:37.583Z" }, + { url = "https://files.pythonhosted.org/packages/a4/6c/5df5590b1f9a821154589df62ceae247537b01ab26b0aa85997c35ca3d9e/multidict-6.2.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:5c5e7d2e300d5cb3b2693b6d60d3e8c8e7dd4ebe27cd17c9cb57020cac0acb80", size = 49151, upload-time = "2025-03-17T16:54:38.756Z" }, + { url = "https://files.pythonhosted.org/packages/d5/ca/c917fbf1be989cd7ea9caa6f87e9c33844ba8d5fbb29cd515d4d2833b84c/multidict-6.2.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:256d431fe4583c5f1e0f2e9c4d9c22f3a04ae96009b8cfa096da3a8723db0a16", size = 29803, upload-time = "2025-03-17T16:54:40.256Z" }, + { url = "https://files.pythonhosted.org/packages/22/19/d97086fc96f73acf36d4dbe65c2c4175911969df49c4e94ef082be59d94e/multidict-6.2.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:a3c0ff89fe40a152e77b191b83282c9664357dce3004032d42e68c514ceff27e", size = 29947, upload-time = "2025-03-17T16:54:41.545Z" }, + { url = "https://files.pythonhosted.org/packages/e3/3b/203476b6e915c3f51616d5f87230c556e2f24b168c14818a3d8dae242b1b/multidict-6.2.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ef7d48207926edbf8b16b336f779c557dd8f5a33035a85db9c4b0febb0706817", size = 130369, upload-time = "2025-03-17T16:54:43.166Z" }, + { url = "https://files.pythonhosted.org/packages/c6/4f/67470007cf03b2bb6df8ae6d716a8eeb0a7d19e0c8dba4e53fa338883bca/multidict-6.2.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1f3c099d3899b14e1ce52262eb82a5f5cb92157bb5106bf627b618c090a0eadc", size = 135231, upload-time = "2025-03-17T16:54:44.572Z" }, + { url = "https://files.pythonhosted.org/packages/6d/f5/7a5ce64dc9a3fecc7d67d0b5cb9c262c67e0b660639e5742c13af63fd80f/multidict-6.2.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e16e7297f29a544f49340012d6fc08cf14de0ab361c9eb7529f6a57a30cbfda1", size = 133634, upload-time = "2025-03-17T16:54:45.998Z" }, + { url = "https://files.pythonhosted.org/packages/05/93/ab2931907e318c0437a4cd156c9cfff317ffb33d99ebbfe2d64200a870f7/multidict-6.2.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:042028348dc5a1f2be6c666437042a98a5d24cee50380f4c0902215e5ec41844", size = 131349, upload-time = "2025-03-17T16:54:47.837Z" }, + { url = "https://files.pythonhosted.org/packages/54/aa/ab8eda83a6a85f5b4bb0b1c28e62b18129b14519ef2e0d4cfd5f360da73c/multidict-6.2.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:08549895e6a799bd551cf276f6e59820aa084f0f90665c0f03dd3a50db5d3c48", size = 120861, upload-time = "2025-03-17T16:54:49.201Z" }, + { url = "https://files.pythonhosted.org/packages/15/2f/7d08ea7c5d9f45786893b4848fad59ec8ea567367d4234691a721e4049a1/multidict-6.2.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:4ccfd74957ef53fa7380aaa1c961f523d582cd5e85a620880ffabd407f8202c0", size = 134611, upload-time = "2025-03-17T16:54:50.811Z" }, + { url = "https://files.pythonhosted.org/packages/8b/07/387047bb1eac563981d397a7f85c75b306df1fff3c20b90da5a6cf6e487e/multidict-6.2.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:83b78c680d4b15d33042d330c2fa31813ca3974197bddb3836a5c635a5fd013f", size = 128955, upload-time = "2025-03-17T16:54:52.48Z" }, + { url = "https://files.pythonhosted.org/packages/8d/6e/7ae18f764a5282c2d682f1c90c6b2a0f6490327730170139a7a63bf3bb20/multidict-6.2.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:b4c153863dd6569f6511845922c53e39c8d61f6e81f228ad5443e690fca403de", size = 139759, upload-time = "2025-03-17T16:54:53.877Z" }, + { url = "https://files.pythonhosted.org/packages/b6/f4/c1b3b087b9379b9e56229bcf6570b9a963975c205a5811ac717284890598/multidict-6.2.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:98aa8325c7f47183b45588af9c434533196e241be0a4e4ae2190b06d17675c02", size = 136426, upload-time = "2025-03-17T16:54:56.506Z" }, + { url = "https://files.pythonhosted.org/packages/a2/0e/ef7b39b161ffd40f9e25dd62e59644b2ccaa814c64e9573f9bc721578419/multidict-6.2.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9e658d1373c424457ddf6d55ec1db93c280b8579276bebd1f72f113072df8a5d", size = 134648, upload-time = "2025-03-17T16:54:57.896Z" }, + { url = "https://files.pythonhosted.org/packages/37/5c/7905acd0ca411c97bcae62ab167d9922f0c5a1d316b6d3af875d4bda3551/multidict-6.2.0-cp313-cp313-win32.whl", hash = "sha256:3157126b028c074951839233647bd0e30df77ef1fedd801b48bdcad242a60f4e", size = 26680, upload-time = "2025-03-17T16:54:59.399Z" }, + { url = "https://files.pythonhosted.org/packages/89/36/96b071d1dad6ac44fe517e4250329e753787bb7a63967ef44bb9b3a659f6/multidict-6.2.0-cp313-cp313-win_amd64.whl", hash = "sha256:2e87f1926e91855ae61769ba3e3f7315120788c099677e0842e697b0bfb659f2", size = 28942, upload-time = "2025-03-17T16:55:00.813Z" }, + { url = "https://files.pythonhosted.org/packages/f5/05/d686cd2a12d648ecd434675ee8daa2901a80f477817e89ab3b160de5b398/multidict-6.2.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:2529ddbdaa424b2c6c2eb668ea684dd6b75b839d0ad4b21aad60c168269478d7", size = 50807, upload-time = "2025-03-17T16:55:02.162Z" }, + { url = "https://files.pythonhosted.org/packages/4c/1f/c7db5aac8fea129fa4c5a119e3d279da48d769138ae9624d1234aa01a06f/multidict-6.2.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:13551d0e2d7201f0959725a6a769b6f7b9019a168ed96006479c9ac33fe4096b", size = 30474, upload-time = "2025-03-17T16:55:04.097Z" }, + { url = "https://files.pythonhosted.org/packages/e5/f1/1fb27514f4d73cea165429dcb7d90cdc4a45445865832caa0c50dd545420/multidict-6.2.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:d1996ee1330e245cd3aeda0887b4409e3930524c27642b046e4fae88ffa66c5e", size = 30841, upload-time = "2025-03-17T16:55:06.098Z" }, + { url = "https://files.pythonhosted.org/packages/d6/6b/9487169e549a23c8958edbb332afaf1ab55d61f0c03cb758ee07ff8f74fb/multidict-6.2.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c537da54ce4ff7c15e78ab1292e5799d0d43a2108e006578a57f531866f64025", size = 148658, upload-time = "2025-03-17T16:55:07.556Z" }, + { url = "https://files.pythonhosted.org/packages/d7/22/79ebb2e4f70857c94999ce195db76886ae287b1b6102da73df24dcad4903/multidict-6.2.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0f249badb360b0b4d694307ad40f811f83df4da8cef7b68e429e4eea939e49dd", size = 151988, upload-time = "2025-03-17T16:55:09.141Z" }, + { url = "https://files.pythonhosted.org/packages/49/5d/63b17f3c1a2861587d26705923a94eb6b2600e5222d6b0d513bce5a78720/multidict-6.2.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:48d39b1824b8d6ea7de878ef6226efbe0773f9c64333e1125e0efcfdd18a24c7", size = 148432, upload-time = "2025-03-17T16:55:11.089Z" }, + { url = "https://files.pythonhosted.org/packages/a3/22/55204eec45c4280fa431c11494ad64d6da0dc89af76282fc6467432360a0/multidict-6.2.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b99aac6bb2c37db336fa03a39b40ed4ef2818bf2dfb9441458165ebe88b793af", size = 143161, upload-time = "2025-03-17T16:55:12.625Z" }, + { url = "https://files.pythonhosted.org/packages/97/e6/202b2cf5af161228767acab8bc49e73a91f4a7de088c9c71f3c02950a030/multidict-6.2.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:07bfa8bc649783e703263f783f73e27fef8cd37baaad4389816cf6a133141331", size = 136820, upload-time = "2025-03-17T16:55:14.073Z" }, + { url = "https://files.pythonhosted.org/packages/7d/16/dbedae0e94c7edc48fddef0c39483f2313205d9bc566fd7f11777b168616/multidict-6.2.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:b2c00ad31fbc2cbac85d7d0fcf90853b2ca2e69d825a2d3f3edb842ef1544a2c", size = 150875, upload-time = "2025-03-17T16:55:15.625Z" }, + { url = "https://files.pythonhosted.org/packages/f3/04/38ccf25d4bf8beef76a22bad7d9833fd088b4594c9765fe6fede39aa6c89/multidict-6.2.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:0d57a01a2a9fa00234aace434d8c131f0ac6e0ac6ef131eda5962d7e79edfb5b", size = 142050, upload-time = "2025-03-17T16:55:17.186Z" }, + { url = "https://files.pythonhosted.org/packages/9e/89/4f6b43386e7b79a4aad560d751981a0a282a1943c312ac72f940d7cf8f9f/multidict-6.2.0-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:abf5b17bc0cf626a8a497d89ac691308dbd825d2ac372aa990b1ca114e470151", size = 154117, upload-time = "2025-03-17T16:55:19.115Z" }, + { url = "https://files.pythonhosted.org/packages/24/e3/3dde5b193f86d30ad6400bd50e116b0df1da3f0c7d419661e3bd79e5ad86/multidict-6.2.0-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:f7716f7e7138252d88607228ce40be22660d6608d20fd365d596e7ca0738e019", size = 149408, upload-time = "2025-03-17T16:55:20.689Z" }, + { url = "https://files.pythonhosted.org/packages/df/b2/ec1e27e8e3da12fcc9053e1eae2f6b50faa8708064d83ea25aa7fb77ffd2/multidict-6.2.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:d5a36953389f35f0a4e88dc796048829a2f467c9197265504593f0e420571547", size = 145767, upload-time = "2025-03-17T16:55:22.271Z" }, + { url = "https://files.pythonhosted.org/packages/3a/8e/c07a648a9d592fa9f3a19d1c7e1c7738ba95aff90db967a5a09cff1e1f37/multidict-6.2.0-cp313-cp313t-win32.whl", hash = "sha256:e653d36b1bf48fa78c7fcebb5fa679342e025121ace8c87ab05c1cefd33b34fc", size = 28950, upload-time = "2025-03-17T16:55:23.807Z" }, + { url = "https://files.pythonhosted.org/packages/dc/a9/bebb5485b94d7c09831638a4df9a1a924c32431a750723f0bf39cd16a787/multidict-6.2.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ca23db5fb195b5ef4fd1f77ce26cadefdf13dba71dab14dadd29b34d457d7c44", size = 32001, upload-time = "2025-03-17T16:55:25.184Z" }, + { url = "https://files.pythonhosted.org/packages/9c/fd/b247aec6add5601956d440488b7f23151d8343747e82c038af37b28d6098/multidict-6.2.0-py3-none-any.whl", hash = "sha256:5d26547423e5e71dcc562c4acdc134b900640a39abd9066d7326a7cc2324c530", size = 10266, upload-time = "2025-03-17T16:55:52.771Z" }, ] [[package]] name = "narwhals" -version = "2.0.1" +version = "1.31.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/e2/8f/51d14e402c4f9d281a2e153a6a805cad5460088027a999faf264b54e7641/narwhals-2.0.1.tar.gz", hash = "sha256:235e61ca807bc21110ca36a4d53888ecc22c42dcdf50a7c886e10dde3fd7f38c", size = 525541, upload-time = "2025-07-29T08:39:04.81Z" } +sdist = { url = "https://files.pythonhosted.org/packages/36/fa/c2b6a4d5dbc4af15aa58c86920d5275a9c65142318179b246685069f57da/narwhals-1.31.0.tar.gz", hash = "sha256:333472e2562343dfdd27407ec9b5114a07c81d0416794e4ac6b703dd925c6a1a", size = 253463, upload-time = "2025-03-17T15:26:26.065Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/7f/26/43caf834e47c63883a5eddc02893b7fdbe6a0a4508ff6dc401907f3cc085/narwhals-2.0.1-py3-none-any.whl", hash = "sha256:837457e36a2ba1710c881fb69e1f79ce44fb81728c92ac378f70892a53af8ddb", size = 385436, upload-time = "2025-07-29T08:39:03.163Z" }, + { url = "https://files.pythonhosted.org/packages/f9/c0/fb39bd876ea2fd9509343d643690cd2f9715e6a77271e7c7b26f1eea70c1/narwhals-1.31.0-py3-none-any.whl", hash = "sha256:2a7b79bb5f511055c4c0142121fc0d4171ea171458e12d44dbd9c8fc6488e997", size = 313124, upload-time = "2025-03-17T15:26:23.87Z" }, ] [[package]] @@ -1984,9 +1803,9 @@ name = "netcdf4" version = "1.7.2" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "certifi", marker = "python_full_version < '3.11'" }, - { name = "cftime", marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, + { name = "certifi" }, + { name = "cftime" }, + { name = "numpy" }, ] sdist = { url = "https://files.pythonhosted.org/packages/71/ed/4d27fcfa40ebfdad3d2088a3de7ee48dbff7f35163e815ec1870d2a7398c/netcdf4-1.7.2.tar.gz", hash = "sha256:a4c6375540b19989896136943abb6d44850ff6f1fa7d3f063253b1ad3f8b7fce", size = 835064, upload-time = "2024-10-22T19:01:25.521Z" } wheels = [ @@ -2016,32 +1835,11 @@ wheels = [ name = "networkx" version = "3.4.2" source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version < '3.11' and sys_platform == 'linux'", - "python_full_version < '3.11' and sys_platform != 'linux'", -] sdist = { url = "https://files.pythonhosted.org/packages/fd/1d/06475e1cd5264c0b870ea2cc6fdb3e37177c1e565c43f56ff17a10e3937f/networkx-3.4.2.tar.gz", hash = "sha256:307c3669428c5362aab27c8a1260aa8f47c4e91d3891f48be0141738d8d053e1", size = 2151368, upload-time = "2024-10-21T12:39:38.695Z" } wheels = [ { url = "https://files.pythonhosted.org/packages/b9/54/dd730b32ea14ea797530a4479b2ed46a6fb250f682a9cfb997e968bf0261/networkx-3.4.2-py3-none-any.whl", hash = "sha256:df5d4365b724cf81b8c6a7312509d0c22386097011ad1abe274afd5e9d3bbc5f", size = 1723263, upload-time = "2024-10-21T12:39:36.247Z" }, ] -[[package]] -name = "networkx" -version = "3.5" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.13' and sys_platform == 'linux'", - "python_full_version == '3.12.*' and sys_platform == 'linux'", - "python_full_version >= '3.13' and sys_platform != 'linux'", - "python_full_version == '3.12.*' and sys_platform != 'linux'", - "python_full_version == '3.11.*' and sys_platform == 'linux'", - "python_full_version == '3.11.*' and sys_platform != 'linux'", -] -sdist = { url = "https://files.pythonhosted.org/packages/6c/4f/ccdb8ad3a38e583f214547fd2f7ff1fc160c43a75af88e6aec213404b96a/networkx-3.5.tar.gz", hash = "sha256:d4c6f9cf81f52d69230866796b82afbccdec3db7ae4fbd1b65ea750feed50037", size = 2471065, upload-time = "2025-05-29T11:35:07.804Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/eb/8d/776adee7bbf76365fdd7f2552710282c79a4ead5d2a46408c9043a2b70ba/networkx-3.5-py3-none-any.whl", hash = "sha256:0030d386a9a06dee3565298b4a734b68589749a544acbb6c412dc9e2489ec6ec", size = 2034406, upload-time = "2025-05-29T11:35:04.961Z" }, -] - [[package]] name = "ninja" version = "1.11.1.4" @@ -2077,236 +1875,139 @@ wheels = [ [[package]] name = "numpy" -version = "2.2.6" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version < '3.11' and sys_platform == 'linux'", - "python_full_version < '3.11' and sys_platform != 'linux'", -] -sdist = { url = "https://files.pythonhosted.org/packages/76/21/7d2a95e4bba9dc13d043ee156a356c0a8f0c6309dff6b21b4d71a073b8a8/numpy-2.2.6.tar.gz", hash = "sha256:e29554e2bef54a90aa5cc07da6ce955accb83f21ab5de01a62c8478897b264fd", size = 20276440, upload-time = "2025-05-17T22:38:04.611Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/9a/3e/ed6db5be21ce87955c0cbd3009f2803f59fa08df21b5df06862e2d8e2bdd/numpy-2.2.6-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:b412caa66f72040e6d268491a59f2c43bf03eb6c96dd8f0307829feb7fa2b6fb", size = 21165245, upload-time = "2025-05-17T21:27:58.555Z" }, - { url = "https://files.pythonhosted.org/packages/22/c2/4b9221495b2a132cc9d2eb862e21d42a009f5a60e45fc44b00118c174bff/numpy-2.2.6-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:8e41fd67c52b86603a91c1a505ebaef50b3314de0213461c7a6e99c9a3beff90", size = 14360048, upload-time = "2025-05-17T21:28:21.406Z" }, - { url = "https://files.pythonhosted.org/packages/fd/77/dc2fcfc66943c6410e2bf598062f5959372735ffda175b39906d54f02349/numpy-2.2.6-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:37e990a01ae6ec7fe7fa1c26c55ecb672dd98b19c3d0e1d1f326fa13cb38d163", size = 5340542, upload-time = "2025-05-17T21:28:30.931Z" }, - { url = "https://files.pythonhosted.org/packages/7a/4f/1cb5fdc353a5f5cc7feb692db9b8ec2c3d6405453f982435efc52561df58/numpy-2.2.6-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:5a6429d4be8ca66d889b7cf70f536a397dc45ba6faeb5f8c5427935d9592e9cf", size = 6878301, upload-time = "2025-05-17T21:28:41.613Z" }, - { url = "https://files.pythonhosted.org/packages/eb/17/96a3acd228cec142fcb8723bd3cc39c2a474f7dcf0a5d16731980bcafa95/numpy-2.2.6-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:efd28d4e9cd7d7a8d39074a4d44c63eda73401580c5c76acda2ce969e0a38e83", size = 14297320, upload-time = "2025-05-17T21:29:02.78Z" }, - { url = "https://files.pythonhosted.org/packages/b4/63/3de6a34ad7ad6646ac7d2f55ebc6ad439dbbf9c4370017c50cf403fb19b5/numpy-2.2.6-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc7b73d02efb0e18c000e9ad8b83480dfcd5dfd11065997ed4c6747470ae8915", size = 16801050, upload-time = "2025-05-17T21:29:27.675Z" }, - { url = "https://files.pythonhosted.org/packages/07/b6/89d837eddef52b3d0cec5c6ba0456c1bf1b9ef6a6672fc2b7873c3ec4e2e/numpy-2.2.6-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:74d4531beb257d2c3f4b261bfb0fc09e0f9ebb8842d82a7b4209415896adc680", size = 15807034, upload-time = "2025-05-17T21:29:51.102Z" }, - { url = "https://files.pythonhosted.org/packages/01/c8/dc6ae86e3c61cfec1f178e5c9f7858584049b6093f843bca541f94120920/numpy-2.2.6-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:8fc377d995680230e83241d8a96def29f204b5782f371c532579b4f20607a289", size = 18614185, upload-time = "2025-05-17T21:30:18.703Z" }, - { url = "https://files.pythonhosted.org/packages/5b/c5/0064b1b7e7c89137b471ccec1fd2282fceaae0ab3a9550f2568782d80357/numpy-2.2.6-cp310-cp310-win32.whl", hash = "sha256:b093dd74e50a8cba3e873868d9e93a85b78e0daf2e98c6797566ad8044e8363d", size = 6527149, upload-time = "2025-05-17T21:30:29.788Z" }, - { url = "https://files.pythonhosted.org/packages/a3/dd/4b822569d6b96c39d1215dbae0582fd99954dcbcf0c1a13c61783feaca3f/numpy-2.2.6-cp310-cp310-win_amd64.whl", hash = "sha256:f0fd6321b839904e15c46e0d257fdd101dd7f530fe03fd6359c1ea63738703f3", size = 12904620, upload-time = "2025-05-17T21:30:48.994Z" }, - { url = "https://files.pythonhosted.org/packages/da/a8/4f83e2aa666a9fbf56d6118faaaf5f1974d456b1823fda0a176eff722839/numpy-2.2.6-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:f9f1adb22318e121c5c69a09142811a201ef17ab257a1e66ca3025065b7f53ae", size = 21176963, upload-time = "2025-05-17T21:31:19.36Z" }, - { url = "https://files.pythonhosted.org/packages/b3/2b/64e1affc7972decb74c9e29e5649fac940514910960ba25cd9af4488b66c/numpy-2.2.6-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c820a93b0255bc360f53eca31a0e676fd1101f673dda8da93454a12e23fc5f7a", size = 14406743, upload-time = "2025-05-17T21:31:41.087Z" }, - { url = "https://files.pythonhosted.org/packages/4a/9f/0121e375000b5e50ffdd8b25bf78d8e1a5aa4cca3f185d41265198c7b834/numpy-2.2.6-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:3d70692235e759f260c3d837193090014aebdf026dfd167834bcba43e30c2a42", size = 5352616, upload-time = "2025-05-17T21:31:50.072Z" }, - { url = "https://files.pythonhosted.org/packages/31/0d/b48c405c91693635fbe2dcd7bc84a33a602add5f63286e024d3b6741411c/numpy-2.2.6-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:481b49095335f8eed42e39e8041327c05b0f6f4780488f61286ed3c01368d491", size = 6889579, upload-time = "2025-05-17T21:32:01.712Z" }, - { url = "https://files.pythonhosted.org/packages/52/b8/7f0554d49b565d0171eab6e99001846882000883998e7b7d9f0d98b1f934/numpy-2.2.6-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b64d8d4d17135e00c8e346e0a738deb17e754230d7e0810ac5012750bbd85a5a", size = 14312005, upload-time = "2025-05-17T21:32:23.332Z" }, - { url = "https://files.pythonhosted.org/packages/b3/dd/2238b898e51bd6d389b7389ffb20d7f4c10066d80351187ec8e303a5a475/numpy-2.2.6-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ba10f8411898fc418a521833e014a77d3ca01c15b0c6cdcce6a0d2897e6dbbdf", size = 16821570, upload-time = "2025-05-17T21:32:47.991Z" }, - { url = "https://files.pythonhosted.org/packages/83/6c/44d0325722cf644f191042bf47eedad61c1e6df2432ed65cbe28509d404e/numpy-2.2.6-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:bd48227a919f1bafbdda0583705e547892342c26fb127219d60a5c36882609d1", size = 15818548, upload-time = "2025-05-17T21:33:11.728Z" }, - { url = "https://files.pythonhosted.org/packages/ae/9d/81e8216030ce66be25279098789b665d49ff19eef08bfa8cb96d4957f422/numpy-2.2.6-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:9551a499bf125c1d4f9e250377c1ee2eddd02e01eac6644c080162c0c51778ab", size = 18620521, upload-time = "2025-05-17T21:33:39.139Z" }, - { url = "https://files.pythonhosted.org/packages/6a/fd/e19617b9530b031db51b0926eed5345ce8ddc669bb3bc0044b23e275ebe8/numpy-2.2.6-cp311-cp311-win32.whl", hash = "sha256:0678000bb9ac1475cd454c6b8c799206af8107e310843532b04d49649c717a47", size = 6525866, upload-time = "2025-05-17T21:33:50.273Z" }, - { url = "https://files.pythonhosted.org/packages/31/0a/f354fb7176b81747d870f7991dc763e157a934c717b67b58456bc63da3df/numpy-2.2.6-cp311-cp311-win_amd64.whl", hash = "sha256:e8213002e427c69c45a52bbd94163084025f533a55a59d6f9c5b820774ef3303", size = 12907455, upload-time = "2025-05-17T21:34:09.135Z" }, - { url = "https://files.pythonhosted.org/packages/82/5d/c00588b6cf18e1da539b45d3598d3557084990dcc4331960c15ee776ee41/numpy-2.2.6-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:41c5a21f4a04fa86436124d388f6ed60a9343a6f767fced1a8a71c3fbca038ff", size = 20875348, upload-time = "2025-05-17T21:34:39.648Z" }, - { url = "https://files.pythonhosted.org/packages/66/ee/560deadcdde6c2f90200450d5938f63a34b37e27ebff162810f716f6a230/numpy-2.2.6-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:de749064336d37e340f640b05f24e9e3dd678c57318c7289d222a8a2f543e90c", size = 14119362, upload-time = "2025-05-17T21:35:01.241Z" }, - { url = "https://files.pythonhosted.org/packages/3c/65/4baa99f1c53b30adf0acd9a5519078871ddde8d2339dc5a7fde80d9d87da/numpy-2.2.6-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:894b3a42502226a1cac872f840030665f33326fc3dac8e57c607905773cdcde3", size = 5084103, upload-time = "2025-05-17T21:35:10.622Z" }, - { url = "https://files.pythonhosted.org/packages/cc/89/e5a34c071a0570cc40c9a54eb472d113eea6d002e9ae12bb3a8407fb912e/numpy-2.2.6-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:71594f7c51a18e728451bb50cc60a3ce4e6538822731b2933209a1f3614e9282", size = 6625382, upload-time = "2025-05-17T21:35:21.414Z" }, - { url = "https://files.pythonhosted.org/packages/f8/35/8c80729f1ff76b3921d5c9487c7ac3de9b2a103b1cd05e905b3090513510/numpy-2.2.6-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f2618db89be1b4e05f7a1a847a9c1c0abd63e63a1607d892dd54668dd92faf87", size = 14018462, upload-time = "2025-05-17T21:35:42.174Z" }, - { url = "https://files.pythonhosted.org/packages/8c/3d/1e1db36cfd41f895d266b103df00ca5b3cbe965184df824dec5c08c6b803/numpy-2.2.6-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fd83c01228a688733f1ded5201c678f0c53ecc1006ffbc404db9f7a899ac6249", size = 16527618, upload-time = "2025-05-17T21:36:06.711Z" }, - { url = "https://files.pythonhosted.org/packages/61/c6/03ed30992602c85aa3cd95b9070a514f8b3c33e31124694438d88809ae36/numpy-2.2.6-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:37c0ca431f82cd5fa716eca9506aefcabc247fb27ba69c5062a6d3ade8cf8f49", size = 15505511, upload-time = "2025-05-17T21:36:29.965Z" }, - { url = "https://files.pythonhosted.org/packages/b7/25/5761d832a81df431e260719ec45de696414266613c9ee268394dd5ad8236/numpy-2.2.6-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:fe27749d33bb772c80dcd84ae7e8df2adc920ae8297400dabec45f0dedb3f6de", size = 18313783, upload-time = "2025-05-17T21:36:56.883Z" }, - { url = "https://files.pythonhosted.org/packages/57/0a/72d5a3527c5ebffcd47bde9162c39fae1f90138c961e5296491ce778e682/numpy-2.2.6-cp312-cp312-win32.whl", hash = "sha256:4eeaae00d789f66c7a25ac5f34b71a7035bb474e679f410e5e1a94deb24cf2d4", size = 6246506, upload-time = "2025-05-17T21:37:07.368Z" }, - { url = "https://files.pythonhosted.org/packages/36/fa/8c9210162ca1b88529ab76b41ba02d433fd54fecaf6feb70ef9f124683f1/numpy-2.2.6-cp312-cp312-win_amd64.whl", hash = "sha256:c1f9540be57940698ed329904db803cf7a402f3fc200bfe599334c9bd84a40b2", size = 12614190, upload-time = "2025-05-17T21:37:26.213Z" }, - { url = "https://files.pythonhosted.org/packages/f9/5c/6657823f4f594f72b5471f1db1ab12e26e890bb2e41897522d134d2a3e81/numpy-2.2.6-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:0811bb762109d9708cca4d0b13c4f67146e3c3b7cf8d34018c722adb2d957c84", size = 20867828, upload-time = "2025-05-17T21:37:56.699Z" }, - { url = "https://files.pythonhosted.org/packages/dc/9e/14520dc3dadf3c803473bd07e9b2bd1b69bc583cb2497b47000fed2fa92f/numpy-2.2.6-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:287cc3162b6f01463ccd86be154f284d0893d2b3ed7292439ea97eafa8170e0b", size = 14143006, upload-time = "2025-05-17T21:38:18.291Z" }, - { url = "https://files.pythonhosted.org/packages/4f/06/7e96c57d90bebdce9918412087fc22ca9851cceaf5567a45c1f404480e9e/numpy-2.2.6-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:f1372f041402e37e5e633e586f62aa53de2eac8d98cbfb822806ce4bbefcb74d", size = 5076765, upload-time = "2025-05-17T21:38:27.319Z" }, - { url = "https://files.pythonhosted.org/packages/73/ed/63d920c23b4289fdac96ddbdd6132e9427790977d5457cd132f18e76eae0/numpy-2.2.6-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:55a4d33fa519660d69614a9fad433be87e5252f4b03850642f88993f7b2ca566", size = 6617736, upload-time = "2025-05-17T21:38:38.141Z" }, - { url = "https://files.pythonhosted.org/packages/85/c5/e19c8f99d83fd377ec8c7e0cf627a8049746da54afc24ef0a0cb73d5dfb5/numpy-2.2.6-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f92729c95468a2f4f15e9bb94c432a9229d0d50de67304399627a943201baa2f", size = 14010719, upload-time = "2025-05-17T21:38:58.433Z" }, - { url = "https://files.pythonhosted.org/packages/19/49/4df9123aafa7b539317bf6d342cb6d227e49f7a35b99c287a6109b13dd93/numpy-2.2.6-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1bc23a79bfabc5d056d106f9befb8d50c31ced2fbc70eedb8155aec74a45798f", size = 16526072, upload-time = "2025-05-17T21:39:22.638Z" }, - { url = "https://files.pythonhosted.org/packages/b2/6c/04b5f47f4f32f7c2b0e7260442a8cbcf8168b0e1a41ff1495da42f42a14f/numpy-2.2.6-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:e3143e4451880bed956e706a3220b4e5cf6172ef05fcc397f6f36a550b1dd868", size = 15503213, upload-time = "2025-05-17T21:39:45.865Z" }, - { url = "https://files.pythonhosted.org/packages/17/0a/5cd92e352c1307640d5b6fec1b2ffb06cd0dabe7d7b8227f97933d378422/numpy-2.2.6-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:b4f13750ce79751586ae2eb824ba7e1e8dba64784086c98cdbbcc6a42112ce0d", size = 18316632, upload-time = "2025-05-17T21:40:13.331Z" }, - { url = "https://files.pythonhosted.org/packages/f0/3b/5cba2b1d88760ef86596ad0f3d484b1cbff7c115ae2429678465057c5155/numpy-2.2.6-cp313-cp313-win32.whl", hash = "sha256:5beb72339d9d4fa36522fc63802f469b13cdbe4fdab4a288f0c441b74272ebfd", size = 6244532, upload-time = "2025-05-17T21:43:46.099Z" }, - { url = "https://files.pythonhosted.org/packages/cb/3b/d58c12eafcb298d4e6d0d40216866ab15f59e55d148a5658bb3132311fcf/numpy-2.2.6-cp313-cp313-win_amd64.whl", hash = "sha256:b0544343a702fa80c95ad5d3d608ea3599dd54d4632df855e4c8d24eb6ecfa1c", size = 12610885, upload-time = "2025-05-17T21:44:05.145Z" }, - { url = "https://files.pythonhosted.org/packages/6b/9e/4bf918b818e516322db999ac25d00c75788ddfd2d2ade4fa66f1f38097e1/numpy-2.2.6-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:0bca768cd85ae743b2affdc762d617eddf3bcf8724435498a1e80132d04879e6", size = 20963467, upload-time = "2025-05-17T21:40:44Z" }, - { url = "https://files.pythonhosted.org/packages/61/66/d2de6b291507517ff2e438e13ff7b1e2cdbdb7cb40b3ed475377aece69f9/numpy-2.2.6-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:fc0c5673685c508a142ca65209b4e79ed6740a4ed6b2267dbba90f34b0b3cfda", size = 14225144, upload-time = "2025-05-17T21:41:05.695Z" }, - { url = "https://files.pythonhosted.org/packages/e4/25/480387655407ead912e28ba3a820bc69af9adf13bcbe40b299d454ec011f/numpy-2.2.6-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:5bd4fc3ac8926b3819797a7c0e2631eb889b4118a9898c84f585a54d475b7e40", size = 5200217, upload-time = "2025-05-17T21:41:15.903Z" }, - { url = "https://files.pythonhosted.org/packages/aa/4a/6e313b5108f53dcbf3aca0c0f3e9c92f4c10ce57a0a721851f9785872895/numpy-2.2.6-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:fee4236c876c4e8369388054d02d0e9bb84821feb1a64dd59e137e6511a551f8", size = 6712014, upload-time = "2025-05-17T21:41:27.321Z" }, - { url = "https://files.pythonhosted.org/packages/b7/30/172c2d5c4be71fdf476e9de553443cf8e25feddbe185e0bd88b096915bcc/numpy-2.2.6-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e1dda9c7e08dc141e0247a5b8f49cf05984955246a327d4c48bda16821947b2f", size = 14077935, upload-time = "2025-05-17T21:41:49.738Z" }, - { url = "https://files.pythonhosted.org/packages/12/fb/9e743f8d4e4d3c710902cf87af3512082ae3d43b945d5d16563f26ec251d/numpy-2.2.6-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f447e6acb680fd307f40d3da4852208af94afdfab89cf850986c3ca00562f4fa", size = 16600122, upload-time = "2025-05-17T21:42:14.046Z" }, - { url = "https://files.pythonhosted.org/packages/12/75/ee20da0e58d3a66f204f38916757e01e33a9737d0b22373b3eb5a27358f9/numpy-2.2.6-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:389d771b1623ec92636b0786bc4ae56abafad4a4c513d36a55dce14bd9ce8571", size = 15586143, upload-time = "2025-05-17T21:42:37.464Z" }, - { url = "https://files.pythonhosted.org/packages/76/95/bef5b37f29fc5e739947e9ce5179ad402875633308504a52d188302319c8/numpy-2.2.6-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:8e9ace4a37db23421249ed236fdcdd457d671e25146786dfc96835cd951aa7c1", size = 18385260, upload-time = "2025-05-17T21:43:05.189Z" }, - { url = "https://files.pythonhosted.org/packages/09/04/f2f83279d287407cf36a7a8053a5abe7be3622a4363337338f2585e4afda/numpy-2.2.6-cp313-cp313t-win32.whl", hash = "sha256:038613e9fb8c72b0a41f025a7e4c3f0b7a1b5d768ece4796b674c8f3fe13efff", size = 6377225, upload-time = "2025-05-17T21:43:16.254Z" }, - { url = "https://files.pythonhosted.org/packages/67/0e/35082d13c09c02c011cf21570543d202ad929d961c02a147493cb0c2bdf5/numpy-2.2.6-cp313-cp313t-win_amd64.whl", hash = "sha256:6031dd6dfecc0cf9f668681a37648373bddd6421fff6c66ec1624eed0180ee06", size = 12771374, upload-time = "2025-05-17T21:43:35.479Z" }, - { url = "https://files.pythonhosted.org/packages/9e/3b/d94a75f4dbf1ef5d321523ecac21ef23a3cd2ac8b78ae2aac40873590229/numpy-2.2.6-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:0b605b275d7bd0c640cad4e5d30fa701a8d59302e127e5f79138ad62762c3e3d", size = 21040391, upload-time = "2025-05-17T21:44:35.948Z" }, - { url = "https://files.pythonhosted.org/packages/17/f4/09b2fa1b58f0fb4f7c7963a1649c64c4d315752240377ed74d9cd878f7b5/numpy-2.2.6-pp310-pypy310_pp73-macosx_14_0_x86_64.whl", hash = "sha256:7befc596a7dc9da8a337f79802ee8adb30a552a94f792b9c9d18c840055907db", size = 6786754, upload-time = "2025-05-17T21:44:47.446Z" }, - { url = "https://files.pythonhosted.org/packages/af/30/feba75f143bdc868a1cc3f44ccfa6c4b9ec522b36458e738cd00f67b573f/numpy-2.2.6-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ce47521a4754c8f4593837384bd3424880629f718d87c5d44f8ed763edd63543", size = 16643476, upload-time = "2025-05-17T21:45:11.871Z" }, - { url = "https://files.pythonhosted.org/packages/37/48/ac2a9584402fb6c0cd5b5d1a91dcf176b15760130dd386bbafdbfe3640bf/numpy-2.2.6-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:d042d24c90c41b54fd506da306759e06e568864df8ec17ccc17e9e884634fd00", size = 12812666, upload-time = "2025-05-17T21:45:31.426Z" }, -] - -[[package]] -name = "numpy" -version = "2.3.2" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.13' and sys_platform == 'linux'", - "python_full_version == '3.12.*' and sys_platform == 'linux'", - "python_full_version >= '3.13' and sys_platform != 'linux'", - "python_full_version == '3.12.*' and sys_platform != 'linux'", - "python_full_version == '3.11.*' and sys_platform == 'linux'", - "python_full_version == '3.11.*' and sys_platform != 'linux'", -] -sdist = { url = "https://files.pythonhosted.org/packages/37/7d/3fec4199c5ffb892bed55cff901e4f39a58c81df9c44c280499e92cad264/numpy-2.3.2.tar.gz", hash = "sha256:e0486a11ec30cdecb53f184d496d1c6a20786c81e55e41640270130056f8ee48", size = 20489306, upload-time = "2025-07-24T21:32:07.553Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/96/26/1320083986108998bd487e2931eed2aeedf914b6e8905431487543ec911d/numpy-2.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:852ae5bed3478b92f093e30f785c98e0cb62fa0a939ed057c31716e18a7a22b9", size = 21259016, upload-time = "2025-07-24T20:24:35.214Z" }, - { url = "https://files.pythonhosted.org/packages/c4/2b/792b341463fa93fc7e55abbdbe87dac316c5b8cb5e94fb7a59fb6fa0cda5/numpy-2.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7a0e27186e781a69959d0230dd9909b5e26024f8da10683bd6344baea1885168", size = 14451158, upload-time = "2025-07-24T20:24:58.397Z" }, - { url = "https://files.pythonhosted.org/packages/b7/13/e792d7209261afb0c9f4759ffef6135b35c77c6349a151f488f531d13595/numpy-2.3.2-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:f0a1a8476ad77a228e41619af2fa9505cf69df928e9aaa165746584ea17fed2b", size = 5379817, upload-time = "2025-07-24T20:25:07.746Z" }, - { url = "https://files.pythonhosted.org/packages/49/ce/055274fcba4107c022b2113a213c7287346563f48d62e8d2a5176ad93217/numpy-2.3.2-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:cbc95b3813920145032412f7e33d12080f11dc776262df1712e1638207dde9e8", size = 6913606, upload-time = "2025-07-24T20:25:18.84Z" }, - { url = "https://files.pythonhosted.org/packages/17/f2/e4d72e6bc5ff01e2ab613dc198d560714971900c03674b41947e38606502/numpy-2.3.2-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f75018be4980a7324edc5930fe39aa391d5734531b1926968605416ff58c332d", size = 14589652, upload-time = "2025-07-24T20:25:40.356Z" }, - { url = "https://files.pythonhosted.org/packages/c8/b0/fbeee3000a51ebf7222016e2939b5c5ecf8000a19555d04a18f1e02521b8/numpy-2.3.2-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:20b8200721840f5621b7bd03f8dcd78de33ec522fc40dc2641aa09537df010c3", size = 16938816, upload-time = "2025-07-24T20:26:05.721Z" }, - { url = "https://files.pythonhosted.org/packages/a9/ec/2f6c45c3484cc159621ea8fc000ac5a86f1575f090cac78ac27193ce82cd/numpy-2.3.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:1f91e5c028504660d606340a084db4b216567ded1056ea2b4be4f9d10b67197f", size = 16370512, upload-time = "2025-07-24T20:26:30.545Z" }, - { url = "https://files.pythonhosted.org/packages/b5/01/dd67cf511850bd7aefd6347aaae0956ed415abea741ae107834aae7d6d4e/numpy-2.3.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:fb1752a3bb9a3ad2d6b090b88a9a0ae1cd6f004ef95f75825e2f382c183b2097", size = 18884947, upload-time = "2025-07-24T20:26:58.24Z" }, - { url = "https://files.pythonhosted.org/packages/a7/17/2cf60fd3e6a61d006778735edf67a222787a8c1a7842aed43ef96d777446/numpy-2.3.2-cp311-cp311-win32.whl", hash = "sha256:4ae6863868aaee2f57503c7a5052b3a2807cf7a3914475e637a0ecd366ced220", size = 6599494, upload-time = "2025-07-24T20:27:09.786Z" }, - { url = "https://files.pythonhosted.org/packages/d5/03/0eade211c504bda872a594f045f98ddcc6caef2b7c63610946845e304d3f/numpy-2.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:240259d6564f1c65424bcd10f435145a7644a65a6811cfc3201c4a429ba79170", size = 13087889, upload-time = "2025-07-24T20:27:29.558Z" }, - { url = "https://files.pythonhosted.org/packages/13/32/2c7979d39dafb2a25087e12310fc7f3b9d3c7d960df4f4bc97955ae0ce1d/numpy-2.3.2-cp311-cp311-win_arm64.whl", hash = "sha256:4209f874d45f921bde2cff1ffcd8a3695f545ad2ffbef6d3d3c6768162efab89", size = 10459560, upload-time = "2025-07-24T20:27:46.803Z" }, - { url = "https://files.pythonhosted.org/packages/00/6d/745dd1c1c5c284d17725e5c802ca4d45cfc6803519d777f087b71c9f4069/numpy-2.3.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:bc3186bea41fae9d8e90c2b4fb5f0a1f5a690682da79b92574d63f56b529080b", size = 20956420, upload-time = "2025-07-24T20:28:18.002Z" }, - { url = "https://files.pythonhosted.org/packages/bc/96/e7b533ea5740641dd62b07a790af5d9d8fec36000b8e2d0472bd7574105f/numpy-2.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:2f4f0215edb189048a3c03bd5b19345bdfa7b45a7a6f72ae5945d2a28272727f", size = 14184660, upload-time = "2025-07-24T20:28:39.522Z" }, - { url = "https://files.pythonhosted.org/packages/2b/53/102c6122db45a62aa20d1b18c9986f67e6b97e0d6fbc1ae13e3e4c84430c/numpy-2.3.2-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:8b1224a734cd509f70816455c3cffe13a4f599b1bf7130f913ba0e2c0b2006c0", size = 5113382, upload-time = "2025-07-24T20:28:48.544Z" }, - { url = "https://files.pythonhosted.org/packages/2b/21/376257efcbf63e624250717e82b4fae93d60178f09eb03ed766dbb48ec9c/numpy-2.3.2-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:3dcf02866b977a38ba3ec10215220609ab9667378a9e2150615673f3ffd6c73b", size = 6647258, upload-time = "2025-07-24T20:28:59.104Z" }, - { url = "https://files.pythonhosted.org/packages/91/ba/f4ebf257f08affa464fe6036e13f2bf9d4642a40228781dc1235da81be9f/numpy-2.3.2-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:572d5512df5470f50ada8d1972c5f1082d9a0b7aa5944db8084077570cf98370", size = 14281409, upload-time = "2025-07-24T20:40:30.298Z" }, - { url = "https://files.pythonhosted.org/packages/59/ef/f96536f1df42c668cbacb727a8c6da7afc9c05ece6d558927fb1722693e1/numpy-2.3.2-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8145dd6d10df13c559d1e4314df29695613575183fa2e2d11fac4c208c8a1f73", size = 16641317, upload-time = "2025-07-24T20:40:56.625Z" }, - { url = "https://files.pythonhosted.org/packages/f6/a7/af813a7b4f9a42f498dde8a4c6fcbff8100eed00182cc91dbaf095645f38/numpy-2.3.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:103ea7063fa624af04a791c39f97070bf93b96d7af7eb23530cd087dc8dbe9dc", size = 16056262, upload-time = "2025-07-24T20:41:20.797Z" }, - { url = "https://files.pythonhosted.org/packages/8b/5d/41c4ef8404caaa7f05ed1cfb06afe16a25895260eacbd29b4d84dff2920b/numpy-2.3.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:fc927d7f289d14f5e037be917539620603294454130b6de200091e23d27dc9be", size = 18579342, upload-time = "2025-07-24T20:41:50.753Z" }, - { url = "https://files.pythonhosted.org/packages/a1/4f/9950e44c5a11636f4a3af6e825ec23003475cc9a466edb7a759ed3ea63bd/numpy-2.3.2-cp312-cp312-win32.whl", hash = "sha256:d95f59afe7f808c103be692175008bab926b59309ade3e6d25009e9a171f7036", size = 6320610, upload-time = "2025-07-24T20:42:01.551Z" }, - { url = "https://files.pythonhosted.org/packages/7c/2f/244643a5ce54a94f0a9a2ab578189c061e4a87c002e037b0829dd77293b6/numpy-2.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:9e196ade2400c0c737d93465327d1ae7c06c7cb8a1756121ebf54b06ca183c7f", size = 12786292, upload-time = "2025-07-24T20:42:20.738Z" }, - { url = "https://files.pythonhosted.org/packages/54/cd/7b5f49d5d78db7badab22d8323c1b6ae458fbf86c4fdfa194ab3cd4eb39b/numpy-2.3.2-cp312-cp312-win_arm64.whl", hash = "sha256:ee807923782faaf60d0d7331f5e86da7d5e3079e28b291973c545476c2b00d07", size = 10194071, upload-time = "2025-07-24T20:42:36.657Z" }, - { url = "https://files.pythonhosted.org/packages/1c/c0/c6bb172c916b00700ed3bf71cb56175fd1f7dbecebf8353545d0b5519f6c/numpy-2.3.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:c8d9727f5316a256425892b043736d63e89ed15bbfe6556c5ff4d9d4448ff3b3", size = 20949074, upload-time = "2025-07-24T20:43:07.813Z" }, - { url = "https://files.pythonhosted.org/packages/20/4e/c116466d22acaf4573e58421c956c6076dc526e24a6be0903219775d862e/numpy-2.3.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:efc81393f25f14d11c9d161e46e6ee348637c0a1e8a54bf9dedc472a3fae993b", size = 14177311, upload-time = "2025-07-24T20:43:29.335Z" }, - { url = "https://files.pythonhosted.org/packages/78/45/d4698c182895af189c463fc91d70805d455a227261d950e4e0f1310c2550/numpy-2.3.2-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:dd937f088a2df683cbb79dda9a772b62a3e5a8a7e76690612c2737f38c6ef1b6", size = 5106022, upload-time = "2025-07-24T20:43:37.999Z" }, - { url = "https://files.pythonhosted.org/packages/9f/76/3e6880fef4420179309dba72a8c11f6166c431cf6dee54c577af8906f914/numpy-2.3.2-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:11e58218c0c46c80509186e460d79fbdc9ca1eb8d8aee39d8f2dc768eb781089", size = 6640135, upload-time = "2025-07-24T20:43:49.28Z" }, - { url = "https://files.pythonhosted.org/packages/34/fa/87ff7f25b3c4ce9085a62554460b7db686fef1e0207e8977795c7b7d7ba1/numpy-2.3.2-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5ad4ebcb683a1f99f4f392cc522ee20a18b2bb12a2c1c42c3d48d5a1adc9d3d2", size = 14278147, upload-time = "2025-07-24T20:44:10.328Z" }, - { url = "https://files.pythonhosted.org/packages/1d/0f/571b2c7a3833ae419fe69ff7b479a78d313581785203cc70a8db90121b9a/numpy-2.3.2-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:938065908d1d869c7d75d8ec45f735a034771c6ea07088867f713d1cd3bbbe4f", size = 16635989, upload-time = "2025-07-24T20:44:34.88Z" }, - { url = "https://files.pythonhosted.org/packages/24/5a/84ae8dca9c9a4c592fe11340b36a86ffa9fd3e40513198daf8a97839345c/numpy-2.3.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:66459dccc65d8ec98cc7df61307b64bf9e08101f9598755d42d8ae65d9a7a6ee", size = 16053052, upload-time = "2025-07-24T20:44:58.872Z" }, - { url = "https://files.pythonhosted.org/packages/57/7c/e5725d99a9133b9813fcf148d3f858df98511686e853169dbaf63aec6097/numpy-2.3.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:a7af9ed2aa9ec5950daf05bb11abc4076a108bd3c7db9aa7251d5f107079b6a6", size = 18577955, upload-time = "2025-07-24T20:45:26.714Z" }, - { url = "https://files.pythonhosted.org/packages/ae/11/7c546fcf42145f29b71e4d6f429e96d8d68e5a7ba1830b2e68d7418f0bbd/numpy-2.3.2-cp313-cp313-win32.whl", hash = "sha256:906a30249315f9c8e17b085cc5f87d3f369b35fedd0051d4a84686967bdbbd0b", size = 6311843, upload-time = "2025-07-24T20:49:24.444Z" }, - { url = "https://files.pythonhosted.org/packages/aa/6f/a428fd1cb7ed39b4280d057720fed5121b0d7754fd2a9768640160f5517b/numpy-2.3.2-cp313-cp313-win_amd64.whl", hash = "sha256:c63d95dc9d67b676e9108fe0d2182987ccb0f11933c1e8959f42fa0da8d4fa56", size = 12782876, upload-time = "2025-07-24T20:49:43.227Z" }, - { url = "https://files.pythonhosted.org/packages/65/85/4ea455c9040a12595fb6c43f2c217257c7b52dd0ba332c6a6c1d28b289fe/numpy-2.3.2-cp313-cp313-win_arm64.whl", hash = "sha256:b05a89f2fb84d21235f93de47129dd4f11c16f64c87c33f5e284e6a3a54e43f2", size = 10192786, upload-time = "2025-07-24T20:49:59.443Z" }, - { url = "https://files.pythonhosted.org/packages/80/23/8278f40282d10c3f258ec3ff1b103d4994bcad78b0cba9208317f6bb73da/numpy-2.3.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:4e6ecfeddfa83b02318f4d84acf15fbdbf9ded18e46989a15a8b6995dfbf85ab", size = 21047395, upload-time = "2025-07-24T20:45:58.821Z" }, - { url = "https://files.pythonhosted.org/packages/1f/2d/624f2ce4a5df52628b4ccd16a4f9437b37c35f4f8a50d00e962aae6efd7a/numpy-2.3.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:508b0eada3eded10a3b55725b40806a4b855961040180028f52580c4729916a2", size = 14300374, upload-time = "2025-07-24T20:46:20.207Z" }, - { url = "https://files.pythonhosted.org/packages/f6/62/ff1e512cdbb829b80a6bd08318a58698867bca0ca2499d101b4af063ee97/numpy-2.3.2-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:754d6755d9a7588bdc6ac47dc4ee97867271b17cee39cb87aef079574366db0a", size = 5228864, upload-time = "2025-07-24T20:46:30.58Z" }, - { url = "https://files.pythonhosted.org/packages/7d/8e/74bc18078fff03192d4032cfa99d5a5ca937807136d6f5790ce07ca53515/numpy-2.3.2-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:a9f66e7d2b2d7712410d3bc5684149040ef5f19856f20277cd17ea83e5006286", size = 6737533, upload-time = "2025-07-24T20:46:46.111Z" }, - { url = "https://files.pythonhosted.org/packages/19/ea/0731efe2c9073ccca5698ef6a8c3667c4cf4eea53fcdcd0b50140aba03bc/numpy-2.3.2-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:de6ea4e5a65d5a90c7d286ddff2b87f3f4ad61faa3db8dabe936b34c2275b6f8", size = 14352007, upload-time = "2025-07-24T20:47:07.1Z" }, - { url = "https://files.pythonhosted.org/packages/cf/90/36be0865f16dfed20f4bc7f75235b963d5939707d4b591f086777412ff7b/numpy-2.3.2-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a3ef07ec8cbc8fc9e369c8dcd52019510c12da4de81367d8b20bc692aa07573a", size = 16701914, upload-time = "2025-07-24T20:47:32.459Z" }, - { url = "https://files.pythonhosted.org/packages/94/30/06cd055e24cb6c38e5989a9e747042b4e723535758e6153f11afea88c01b/numpy-2.3.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:27c9f90e7481275c7800dc9c24b7cc40ace3fdb970ae4d21eaff983a32f70c91", size = 16132708, upload-time = "2025-07-24T20:47:58.129Z" }, - { url = "https://files.pythonhosted.org/packages/9a/14/ecede608ea73e58267fd7cb78f42341b3b37ba576e778a1a06baffbe585c/numpy-2.3.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:07b62978075b67eee4065b166d000d457c82a1efe726cce608b9db9dd66a73a5", size = 18651678, upload-time = "2025-07-24T20:48:25.402Z" }, - { url = "https://files.pythonhosted.org/packages/40/f3/2fe6066b8d07c3685509bc24d56386534c008b462a488b7f503ba82b8923/numpy-2.3.2-cp313-cp313t-win32.whl", hash = "sha256:c771cfac34a4f2c0de8e8c97312d07d64fd8f8ed45bc9f5726a7e947270152b5", size = 6441832, upload-time = "2025-07-24T20:48:37.181Z" }, - { url = "https://files.pythonhosted.org/packages/0b/ba/0937d66d05204d8f28630c9c60bc3eda68824abde4cf756c4d6aad03b0c6/numpy-2.3.2-cp313-cp313t-win_amd64.whl", hash = "sha256:72dbebb2dcc8305c431b2836bcc66af967df91be793d63a24e3d9b741374c450", size = 12927049, upload-time = "2025-07-24T20:48:56.24Z" }, - { url = "https://files.pythonhosted.org/packages/e9/ed/13542dd59c104d5e654dfa2ac282c199ba64846a74c2c4bcdbc3a0f75df1/numpy-2.3.2-cp313-cp313t-win_arm64.whl", hash = "sha256:72c6df2267e926a6d5286b0a6d556ebe49eae261062059317837fda12ddf0c1a", size = 10262935, upload-time = "2025-07-24T20:49:13.136Z" }, - { url = "https://files.pythonhosted.org/packages/c9/7c/7659048aaf498f7611b783e000c7268fcc4dcf0ce21cd10aad7b2e8f9591/numpy-2.3.2-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:448a66d052d0cf14ce9865d159bfc403282c9bc7bb2a31b03cc18b651eca8b1a", size = 20950906, upload-time = "2025-07-24T20:50:30.346Z" }, - { url = "https://files.pythonhosted.org/packages/80/db/984bea9d4ddf7112a04cfdfb22b1050af5757864cfffe8e09e44b7f11a10/numpy-2.3.2-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:546aaf78e81b4081b2eba1d105c3b34064783027a06b3ab20b6eba21fb64132b", size = 14185607, upload-time = "2025-07-24T20:50:51.923Z" }, - { url = "https://files.pythonhosted.org/packages/e4/76/b3d6f414f4eca568f469ac112a3b510938d892bc5a6c190cb883af080b77/numpy-2.3.2-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:87c930d52f45df092f7578889711a0768094debf73cfcde105e2d66954358125", size = 5114110, upload-time = "2025-07-24T20:51:01.041Z" }, - { url = "https://files.pythonhosted.org/packages/9e/d2/6f5e6826abd6bca52392ed88fe44a4b52aacb60567ac3bc86c67834c3a56/numpy-2.3.2-cp314-cp314-macosx_14_0_x86_64.whl", hash = "sha256:8dc082ea901a62edb8f59713c6a7e28a85daddcb67454c839de57656478f5b19", size = 6642050, upload-time = "2025-07-24T20:51:11.64Z" }, - { url = "https://files.pythonhosted.org/packages/c4/43/f12b2ade99199e39c73ad182f103f9d9791f48d885c600c8e05927865baf/numpy-2.3.2-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:af58de8745f7fa9ca1c0c7c943616c6fe28e75d0c81f5c295810e3c83b5be92f", size = 14296292, upload-time = "2025-07-24T20:51:33.488Z" }, - { url = "https://files.pythonhosted.org/packages/5d/f9/77c07d94bf110a916b17210fac38680ed8734c236bfed9982fd8524a7b47/numpy-2.3.2-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fed5527c4cf10f16c6d0b6bee1f89958bccb0ad2522c8cadc2efd318bcd545f5", size = 16638913, upload-time = "2025-07-24T20:51:58.517Z" }, - { url = "https://files.pythonhosted.org/packages/9b/d1/9d9f2c8ea399cc05cfff8a7437453bd4e7d894373a93cdc46361bbb49a7d/numpy-2.3.2-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:095737ed986e00393ec18ec0b21b47c22889ae4b0cd2d5e88342e08b01141f58", size = 16071180, upload-time = "2025-07-24T20:52:22.827Z" }, - { url = "https://files.pythonhosted.org/packages/4c/41/82e2c68aff2a0c9bf315e47d61951099fed65d8cb2c8d9dc388cb87e947e/numpy-2.3.2-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:b5e40e80299607f597e1a8a247ff8d71d79c5b52baa11cc1cce30aa92d2da6e0", size = 18576809, upload-time = "2025-07-24T20:52:51.015Z" }, - { url = "https://files.pythonhosted.org/packages/14/14/4b4fd3efb0837ed252d0f583c5c35a75121038a8c4e065f2c259be06d2d8/numpy-2.3.2-cp314-cp314-win32.whl", hash = "sha256:7d6e390423cc1f76e1b8108c9b6889d20a7a1f59d9a60cac4a050fa734d6c1e2", size = 6366410, upload-time = "2025-07-24T20:56:44.949Z" }, - { url = "https://files.pythonhosted.org/packages/11/9e/b4c24a6b8467b61aced5c8dc7dcfce23621baa2e17f661edb2444a418040/numpy-2.3.2-cp314-cp314-win_amd64.whl", hash = "sha256:b9d0878b21e3918d76d2209c924ebb272340da1fb51abc00f986c258cd5e957b", size = 12918821, upload-time = "2025-07-24T20:57:06.479Z" }, - { url = "https://files.pythonhosted.org/packages/0e/0f/0dc44007c70b1007c1cef86b06986a3812dd7106d8f946c09cfa75782556/numpy-2.3.2-cp314-cp314-win_arm64.whl", hash = "sha256:2738534837c6a1d0c39340a190177d7d66fdf432894f469728da901f8f6dc910", size = 10477303, upload-time = "2025-07-24T20:57:22.879Z" }, - { url = "https://files.pythonhosted.org/packages/8b/3e/075752b79140b78ddfc9c0a1634d234cfdbc6f9bbbfa6b7504e445ad7d19/numpy-2.3.2-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:4d002ecf7c9b53240be3bb69d80f86ddbd34078bae04d87be81c1f58466f264e", size = 21047524, upload-time = "2025-07-24T20:53:22.086Z" }, - { url = "https://files.pythonhosted.org/packages/fe/6d/60e8247564a72426570d0e0ea1151b95ce5bd2f1597bb878a18d32aec855/numpy-2.3.2-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:293b2192c6bcce487dbc6326de5853787f870aeb6c43f8f9c6496db5b1781e45", size = 14300519, upload-time = "2025-07-24T20:53:44.053Z" }, - { url = "https://files.pythonhosted.org/packages/4d/73/d8326c442cd428d47a067070c3ac6cc3b651a6e53613a1668342a12d4479/numpy-2.3.2-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:0a4f2021a6da53a0d580d6ef5db29947025ae8b35b3250141805ea9a32bbe86b", size = 5228972, upload-time = "2025-07-24T20:53:53.81Z" }, - { url = "https://files.pythonhosted.org/packages/34/2e/e71b2d6dad075271e7079db776196829019b90ce3ece5c69639e4f6fdc44/numpy-2.3.2-cp314-cp314t-macosx_14_0_x86_64.whl", hash = "sha256:9c144440db4bf3bb6372d2c3e49834cc0ff7bb4c24975ab33e01199e645416f2", size = 6737439, upload-time = "2025-07-24T20:54:04.742Z" }, - { url = "https://files.pythonhosted.org/packages/15/b0/d004bcd56c2c5e0500ffc65385eb6d569ffd3363cb5e593ae742749b2daa/numpy-2.3.2-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f92d6c2a8535dc4fe4419562294ff957f83a16ebdec66df0805e473ffaad8bd0", size = 14352479, upload-time = "2025-07-24T20:54:25.819Z" }, - { url = "https://files.pythonhosted.org/packages/11/e3/285142fcff8721e0c99b51686426165059874c150ea9ab898e12a492e291/numpy-2.3.2-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:cefc2219baa48e468e3db7e706305fcd0c095534a192a08f31e98d83a7d45fb0", size = 16702805, upload-time = "2025-07-24T20:54:50.814Z" }, - { url = "https://files.pythonhosted.org/packages/33/c3/33b56b0e47e604af2c7cd065edca892d180f5899599b76830652875249a3/numpy-2.3.2-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:76c3e9501ceb50b2ff3824c3589d5d1ab4ac857b0ee3f8f49629d0de55ecf7c2", size = 16133830, upload-time = "2025-07-24T20:55:17.306Z" }, - { url = "https://files.pythonhosted.org/packages/6e/ae/7b1476a1f4d6a48bc669b8deb09939c56dd2a439db1ab03017844374fb67/numpy-2.3.2-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:122bf5ed9a0221b3419672493878ba4967121514b1d7d4656a7580cd11dddcbf", size = 18652665, upload-time = "2025-07-24T20:55:46.665Z" }, - { url = "https://files.pythonhosted.org/packages/14/ba/5b5c9978c4bb161034148ade2de9db44ec316fab89ce8c400db0e0c81f86/numpy-2.3.2-cp314-cp314t-win32.whl", hash = "sha256:6f1ae3dcb840edccc45af496f312528c15b1f79ac318169d094e85e4bb35fdf1", size = 6514777, upload-time = "2025-07-24T20:55:57.66Z" }, - { url = "https://files.pythonhosted.org/packages/eb/46/3dbaf0ae7c17cdc46b9f662c56da2054887b8d9e737c1476f335c83d33db/numpy-2.3.2-cp314-cp314t-win_amd64.whl", hash = "sha256:087ffc25890d89a43536f75c5fe8770922008758e8eeeef61733957041ed2f9b", size = 13111856, upload-time = "2025-07-24T20:56:17.318Z" }, - { url = "https://files.pythonhosted.org/packages/c1/9e/1652778bce745a67b5fe05adde60ed362d38eb17d919a540e813d30f6874/numpy-2.3.2-cp314-cp314t-win_arm64.whl", hash = "sha256:092aeb3449833ea9c0bf0089d70c29ae480685dd2377ec9cdbbb620257f84631", size = 10544226, upload-time = "2025-07-24T20:56:34.509Z" }, - { url = "https://files.pythonhosted.org/packages/cf/ea/50ebc91d28b275b23b7128ef25c3d08152bc4068f42742867e07a870a42a/numpy-2.3.2-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:14a91ebac98813a49bc6aa1a0dfc09513dcec1d97eaf31ca21a87221a1cdcb15", size = 21130338, upload-time = "2025-07-24T20:57:54.37Z" }, - { url = "https://files.pythonhosted.org/packages/9f/57/cdd5eac00dd5f137277355c318a955c0d8fb8aa486020c22afd305f8b88f/numpy-2.3.2-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:71669b5daae692189540cffc4c439468d35a3f84f0c88b078ecd94337f6cb0ec", size = 14375776, upload-time = "2025-07-24T20:58:16.303Z" }, - { url = "https://files.pythonhosted.org/packages/83/85/27280c7f34fcd305c2209c0cdca4d70775e4859a9eaa92f850087f8dea50/numpy-2.3.2-pp311-pypy311_pp73-macosx_14_0_arm64.whl", hash = "sha256:69779198d9caee6e547adb933941ed7520f896fd9656834c300bdf4dd8642712", size = 5304882, upload-time = "2025-07-24T20:58:26.199Z" }, - { url = "https://files.pythonhosted.org/packages/48/b4/6500b24d278e15dd796f43824e69939d00981d37d9779e32499e823aa0aa/numpy-2.3.2-pp311-pypy311_pp73-macosx_14_0_x86_64.whl", hash = "sha256:2c3271cc4097beb5a60f010bcc1cc204b300bb3eafb4399376418a83a1c6373c", size = 6818405, upload-time = "2025-07-24T20:58:37.341Z" }, - { url = "https://files.pythonhosted.org/packages/9b/c9/142c1e03f199d202da8e980c2496213509291b6024fd2735ad28ae7065c7/numpy-2.3.2-pp311-pypy311_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8446acd11fe3dc1830568c941d44449fd5cb83068e5c70bd5a470d323d448296", size = 14419651, upload-time = "2025-07-24T20:58:59.048Z" }, - { url = "https://files.pythonhosted.org/packages/8b/95/8023e87cbea31a750a6c00ff9427d65ebc5fef104a136bfa69f76266d614/numpy-2.3.2-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:aa098a5ab53fa407fded5870865c6275a5cd4101cfdef8d6fafc48286a96e981", size = 16760166, upload-time = "2025-07-24T21:28:56.38Z" }, - { url = "https://files.pythonhosted.org/packages/78/e3/6690b3f85a05506733c7e90b577e4762517404ea78bab2ca3a5cb1aeb78d/numpy-2.3.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:6936aff90dda378c09bea075af0d9c675fe3a977a9d2402f95a87f440f59f619", size = 12977811, upload-time = "2025-07-24T21:29:18.234Z" }, +version = "2.2.4" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/e1/78/31103410a57bc2c2b93a3597340a8119588571f6a4539067546cb9a0bfac/numpy-2.2.4.tar.gz", hash = "sha256:9ba03692a45d3eef66559efe1d1096c4b9b75c0986b5dff5530c378fb8331d4f", size = 20270701, upload-time = "2025-03-16T18:27:00.648Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/04/89/a79e86e5c1433926ed7d60cb267fb64aa578b6101ab645800fd43b4801de/numpy-2.2.4-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:8146f3550d627252269ac42ae660281d673eb6f8b32f113538e0cc2a9aed42b9", size = 21250661, upload-time = "2025-03-16T18:02:13.017Z" }, + { url = "https://files.pythonhosted.org/packages/79/c2/f50921beb8afd60ed9589ad880332cfefdb805422210d327fb48f12b7a81/numpy-2.2.4-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:e642d86b8f956098b564a45e6f6ce68a22c2c97a04f5acd3f221f57b8cb850ae", size = 14389926, upload-time = "2025-03-16T18:02:39.022Z" }, + { url = "https://files.pythonhosted.org/packages/c7/b9/2c4e96130b0b0f97b0ef4a06d6dae3b39d058b21a5e2fa2decd7fd6b1c8f/numpy-2.2.4-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:a84eda42bd12edc36eb5b53bbcc9b406820d3353f1994b6cfe453a33ff101775", size = 5428329, upload-time = "2025-03-16T18:02:50.032Z" }, + { url = "https://files.pythonhosted.org/packages/7f/a5/3d7094aa898f4fc5c84cdfb26beeae780352d43f5d8bdec966c4393d644c/numpy-2.2.4-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:4ba5054787e89c59c593a4169830ab362ac2bee8a969249dc56e5d7d20ff8df9", size = 6963559, upload-time = "2025-03-16T18:03:02.523Z" }, + { url = "https://files.pythonhosted.org/packages/4c/22/fb1be710a14434c09080dd4a0acc08939f612ec02efcb04b9e210474782d/numpy-2.2.4-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7716e4a9b7af82c06a2543c53ca476fa0b57e4d760481273e09da04b74ee6ee2", size = 14368066, upload-time = "2025-03-16T18:03:27.146Z" }, + { url = "https://files.pythonhosted.org/packages/c2/07/2e5cc71193e3ef3a219ffcf6ca4858e46ea2be09c026ddd480d596b32867/numpy-2.2.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:adf8c1d66f432ce577d0197dceaac2ac00c0759f573f28516246351c58a85020", size = 16417040, upload-time = "2025-03-16T18:03:55.999Z" }, + { url = "https://files.pythonhosted.org/packages/1a/97/3b1537776ad9a6d1a41813818343745e8dd928a2916d4c9edcd9a8af1dac/numpy-2.2.4-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:218f061d2faa73621fa23d6359442b0fc658d5b9a70801373625d958259eaca3", size = 15879862, upload-time = "2025-03-16T18:04:23.56Z" }, + { url = "https://files.pythonhosted.org/packages/b0/b7/4472f603dd45ef36ff3d8e84e84fe02d9467c78f92cc121633dce6da307b/numpy-2.2.4-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:df2f57871a96bbc1b69733cd4c51dc33bea66146b8c63cacbfed73eec0883017", size = 18206032, upload-time = "2025-03-16T18:04:53.694Z" }, + { url = "https://files.pythonhosted.org/packages/0d/bd/6a092963fb82e6c5aa0d0440635827bbb2910da229545473bbb58c537ed3/numpy-2.2.4-cp310-cp310-win32.whl", hash = "sha256:a0258ad1f44f138b791327961caedffbf9612bfa504ab9597157806faa95194a", size = 6608517, upload-time = "2025-03-16T18:05:06.647Z" }, + { url = "https://files.pythonhosted.org/packages/01/e3/cb04627bc2a1638948bc13e818df26495aa18e20d5be1ed95ab2b10b6847/numpy-2.2.4-cp310-cp310-win_amd64.whl", hash = "sha256:0d54974f9cf14acf49c60f0f7f4084b6579d24d439453d5fc5805d46a165b542", size = 12943498, upload-time = "2025-03-16T18:05:28.591Z" }, + { url = "https://files.pythonhosted.org/packages/16/fb/09e778ee3a8ea0d4dc8329cca0a9c9e65fed847d08e37eba74cb7ed4b252/numpy-2.2.4-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:e9e0a277bb2eb5d8a7407e14688b85fd8ad628ee4e0c7930415687b6564207a4", size = 21254989, upload-time = "2025-03-16T18:06:04.092Z" }, + { url = "https://files.pythonhosted.org/packages/a2/0a/1212befdbecab5d80eca3cde47d304cad986ad4eec7d85a42e0b6d2cc2ef/numpy-2.2.4-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:9eeea959168ea555e556b8188da5fa7831e21d91ce031e95ce23747b7609f8a4", size = 14425910, upload-time = "2025-03-16T18:06:29.062Z" }, + { url = "https://files.pythonhosted.org/packages/2b/3e/e7247c1d4f15086bb106c8d43c925b0b2ea20270224f5186fa48d4fb5cbd/numpy-2.2.4-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:bd3ad3b0a40e713fc68f99ecfd07124195333f1e689387c180813f0e94309d6f", size = 5426490, upload-time = "2025-03-16T18:06:39.901Z" }, + { url = "https://files.pythonhosted.org/packages/5d/fa/aa7cd6be51419b894c5787a8a93c3302a1ed4f82d35beb0613ec15bdd0e2/numpy-2.2.4-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:cf28633d64294969c019c6df4ff37f5698e8326db68cc2b66576a51fad634880", size = 6967754, upload-time = "2025-03-16T18:06:52.658Z" }, + { url = "https://files.pythonhosted.org/packages/d5/ee/96457c943265de9fadeb3d2ffdbab003f7fba13d971084a9876affcda095/numpy-2.2.4-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2fa8fa7697ad1646b5c93de1719965844e004fcad23c91228aca1cf0800044a1", size = 14373079, upload-time = "2025-03-16T18:07:16.297Z" }, + { url = "https://files.pythonhosted.org/packages/c5/5c/ceefca458559f0ccc7a982319f37ed07b0d7b526964ae6cc61f8ad1b6119/numpy-2.2.4-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f4162988a360a29af158aeb4a2f4f09ffed6a969c9776f8f3bdee9b06a8ab7e5", size = 16428819, upload-time = "2025-03-16T18:07:44.188Z" }, + { url = "https://files.pythonhosted.org/packages/22/31/9b2ac8eee99e001eb6add9fa27514ef5e9faf176169057a12860af52704c/numpy-2.2.4-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:892c10d6a73e0f14935c31229e03325a7b3093fafd6ce0af704be7f894d95687", size = 15881470, upload-time = "2025-03-16T18:08:11.545Z" }, + { url = "https://files.pythonhosted.org/packages/f0/dc/8569b5f25ff30484b555ad8a3f537e0225d091abec386c9420cf5f7a2976/numpy-2.2.4-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:db1f1c22173ac1c58db249ae48aa7ead29f534b9a948bc56828337aa84a32ed6", size = 18218144, upload-time = "2025-03-16T18:08:42.042Z" }, + { url = "https://files.pythonhosted.org/packages/5e/05/463c023a39bdeb9bb43a99e7dee2c664cb68d5bb87d14f92482b9f6011cc/numpy-2.2.4-cp311-cp311-win32.whl", hash = "sha256:ea2bb7e2ae9e37d96835b3576a4fa4b3a97592fbea8ef7c3587078b0068b8f09", size = 6606368, upload-time = "2025-03-16T18:08:55.074Z" }, + { url = "https://files.pythonhosted.org/packages/8b/72/10c1d2d82101c468a28adc35de6c77b308f288cfd0b88e1070f15b98e00c/numpy-2.2.4-cp311-cp311-win_amd64.whl", hash = "sha256:f7de08cbe5551911886d1ab60de58448c6df0f67d9feb7d1fb21e9875ef95e91", size = 12947526, upload-time = "2025-03-16T18:09:16.844Z" }, + { url = "https://files.pythonhosted.org/packages/a2/30/182db21d4f2a95904cec1a6f779479ea1ac07c0647f064dea454ec650c42/numpy-2.2.4-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:a7b9084668aa0f64e64bd00d27ba5146ef1c3a8835f3bd912e7a9e01326804c4", size = 20947156, upload-time = "2025-03-16T18:09:51.975Z" }, + { url = "https://files.pythonhosted.org/packages/24/6d/9483566acfbda6c62c6bc74b6e981c777229d2af93c8eb2469b26ac1b7bc/numpy-2.2.4-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:dbe512c511956b893d2dacd007d955a3f03d555ae05cfa3ff1c1ff6df8851854", size = 14133092, upload-time = "2025-03-16T18:10:16.329Z" }, + { url = "https://files.pythonhosted.org/packages/27/f6/dba8a258acbf9d2bed2525cdcbb9493ef9bae5199d7a9cb92ee7e9b2aea6/numpy-2.2.4-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:bb649f8b207ab07caebba230d851b579a3c8711a851d29efe15008e31bb4de24", size = 5163515, upload-time = "2025-03-16T18:10:26.19Z" }, + { url = "https://files.pythonhosted.org/packages/62/30/82116199d1c249446723c68f2c9da40d7f062551036f50b8c4caa42ae252/numpy-2.2.4-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:f34dc300df798742b3d06515aa2a0aee20941c13579d7a2f2e10af01ae4901ee", size = 6696558, upload-time = "2025-03-16T18:10:38.996Z" }, + { url = "https://files.pythonhosted.org/packages/0e/b2/54122b3c6df5df3e87582b2e9430f1bdb63af4023c739ba300164c9ae503/numpy-2.2.4-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c3f7ac96b16955634e223b579a3e5798df59007ca43e8d451a0e6a50f6bfdfba", size = 14084742, upload-time = "2025-03-16T18:11:02.76Z" }, + { url = "https://files.pythonhosted.org/packages/02/e2/e2cbb8d634151aab9528ef7b8bab52ee4ab10e076509285602c2a3a686e0/numpy-2.2.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4f92084defa704deadd4e0a5ab1dc52d8ac9e8a8ef617f3fbb853e79b0ea3592", size = 16134051, upload-time = "2025-03-16T18:11:32.767Z" }, + { url = "https://files.pythonhosted.org/packages/8e/21/efd47800e4affc993e8be50c1b768de038363dd88865920439ef7b422c60/numpy-2.2.4-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7a4e84a6283b36632e2a5b56e121961f6542ab886bc9e12f8f9818b3c266bfbb", size = 15578972, upload-time = "2025-03-16T18:11:59.877Z" }, + { url = "https://files.pythonhosted.org/packages/04/1e/f8bb88f6157045dd5d9b27ccf433d016981032690969aa5c19e332b138c0/numpy-2.2.4-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:11c43995255eb4127115956495f43e9343736edb7fcdb0d973defd9de14cd84f", size = 17898106, upload-time = "2025-03-16T18:12:31.487Z" }, + { url = "https://files.pythonhosted.org/packages/2b/93/df59a5a3897c1f036ae8ff845e45f4081bb06943039ae28a3c1c7c780f22/numpy-2.2.4-cp312-cp312-win32.whl", hash = "sha256:65ef3468b53269eb5fdb3a5c09508c032b793da03251d5f8722b1194f1790c00", size = 6311190, upload-time = "2025-03-16T18:12:44.46Z" }, + { url = "https://files.pythonhosted.org/packages/46/69/8c4f928741c2a8efa255fdc7e9097527c6dc4e4df147e3cadc5d9357ce85/numpy-2.2.4-cp312-cp312-win_amd64.whl", hash = "sha256:2aad3c17ed2ff455b8eaafe06bcdae0062a1db77cb99f4b9cbb5f4ecb13c5146", size = 12644305, upload-time = "2025-03-16T18:13:06.864Z" }, + { url = "https://files.pythonhosted.org/packages/2a/d0/bd5ad792e78017f5decfb2ecc947422a3669a34f775679a76317af671ffc/numpy-2.2.4-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:1cf4e5c6a278d620dee9ddeb487dc6a860f9b199eadeecc567f777daace1e9e7", size = 20933623, upload-time = "2025-03-16T18:13:43.231Z" }, + { url = "https://files.pythonhosted.org/packages/c3/bc/2b3545766337b95409868f8e62053135bdc7fa2ce630aba983a2aa60b559/numpy-2.2.4-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1974afec0b479e50438fc3648974268f972e2d908ddb6d7fb634598cdb8260a0", size = 14148681, upload-time = "2025-03-16T18:14:08.031Z" }, + { url = "https://files.pythonhosted.org/packages/6a/70/67b24d68a56551d43a6ec9fe8c5f91b526d4c1a46a6387b956bf2d64744e/numpy-2.2.4-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:79bd5f0a02aa16808fcbc79a9a376a147cc1045f7dfe44c6e7d53fa8b8a79392", size = 5148759, upload-time = "2025-03-16T18:14:18.613Z" }, + { url = "https://files.pythonhosted.org/packages/1c/8b/e2fc8a75fcb7be12d90b31477c9356c0cbb44abce7ffb36be39a0017afad/numpy-2.2.4-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:3387dd7232804b341165cedcb90694565a6015433ee076c6754775e85d86f1fc", size = 6683092, upload-time = "2025-03-16T18:14:31.386Z" }, + { url = "https://files.pythonhosted.org/packages/13/73/41b7b27f169ecf368b52533edb72e56a133f9e86256e809e169362553b49/numpy-2.2.4-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6f527d8fdb0286fd2fd97a2a96c6be17ba4232da346931d967a0630050dfd298", size = 14081422, upload-time = "2025-03-16T18:14:54.83Z" }, + { url = "https://files.pythonhosted.org/packages/4b/04/e208ff3ae3ddfbafc05910f89546382f15a3f10186b1f56bd99f159689c2/numpy-2.2.4-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bce43e386c16898b91e162e5baaad90c4b06f9dcbe36282490032cec98dc8ae7", size = 16132202, upload-time = "2025-03-16T18:15:22.035Z" }, + { url = "https://files.pythonhosted.org/packages/fe/bc/2218160574d862d5e55f803d88ddcad88beff94791f9c5f86d67bd8fbf1c/numpy-2.2.4-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:31504f970f563d99f71a3512d0c01a645b692b12a63630d6aafa0939e52361e6", size = 15573131, upload-time = "2025-03-16T18:15:48.546Z" }, + { url = "https://files.pythonhosted.org/packages/a5/78/97c775bc4f05abc8a8426436b7cb1be806a02a2994b195945600855e3a25/numpy-2.2.4-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:81413336ef121a6ba746892fad881a83351ee3e1e4011f52e97fba79233611fd", size = 17894270, upload-time = "2025-03-16T18:16:20.274Z" }, + { url = "https://files.pythonhosted.org/packages/b9/eb/38c06217a5f6de27dcb41524ca95a44e395e6a1decdc0c99fec0832ce6ae/numpy-2.2.4-cp313-cp313-win32.whl", hash = "sha256:f486038e44caa08dbd97275a9a35a283a8f1d2f0ee60ac260a1790e76660833c", size = 6308141, upload-time = "2025-03-16T18:20:15.297Z" }, + { url = "https://files.pythonhosted.org/packages/52/17/d0dd10ab6d125c6d11ffb6dfa3423c3571befab8358d4f85cd4471964fcd/numpy-2.2.4-cp313-cp313-win_amd64.whl", hash = "sha256:207a2b8441cc8b6a2a78c9ddc64d00d20c303d79fba08c577752f080c4007ee3", size = 12636885, upload-time = "2025-03-16T18:20:36.982Z" }, + { url = "https://files.pythonhosted.org/packages/fa/e2/793288ede17a0fdc921172916efb40f3cbc2aa97e76c5c84aba6dc7e8747/numpy-2.2.4-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:8120575cb4882318c791f839a4fd66161a6fa46f3f0a5e613071aae35b5dd8f8", size = 20961829, upload-time = "2025-03-16T18:16:56.191Z" }, + { url = "https://files.pythonhosted.org/packages/3a/75/bb4573f6c462afd1ea5cbedcc362fe3e9bdbcc57aefd37c681be1155fbaa/numpy-2.2.4-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:a761ba0fa886a7bb33c6c8f6f20213735cb19642c580a931c625ee377ee8bd39", size = 14161419, upload-time = "2025-03-16T18:17:22.811Z" }, + { url = "https://files.pythonhosted.org/packages/03/68/07b4cd01090ca46c7a336958b413cdbe75002286295f2addea767b7f16c9/numpy-2.2.4-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:ac0280f1ba4a4bfff363a99a6aceed4f8e123f8a9b234c89140f5e894e452ecd", size = 5196414, upload-time = "2025-03-16T18:17:34.066Z" }, + { url = "https://files.pythonhosted.org/packages/a5/fd/d4a29478d622fedff5c4b4b4cedfc37a00691079623c0575978d2446db9e/numpy-2.2.4-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:879cf3a9a2b53a4672a168c21375166171bc3932b7e21f622201811c43cdd3b0", size = 6709379, upload-time = "2025-03-16T18:17:47.466Z" }, + { url = "https://files.pythonhosted.org/packages/41/78/96dddb75bb9be730b87c72f30ffdd62611aba234e4e460576a068c98eff6/numpy-2.2.4-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f05d4198c1bacc9124018109c5fba2f3201dbe7ab6e92ff100494f236209c960", size = 14051725, upload-time = "2025-03-16T18:18:11.904Z" }, + { url = "https://files.pythonhosted.org/packages/00/06/5306b8199bffac2a29d9119c11f457f6c7d41115a335b78d3f86fad4dbe8/numpy-2.2.4-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e2f085ce2e813a50dfd0e01fbfc0c12bbe5d2063d99f8b29da30e544fb6483b8", size = 16101638, upload-time = "2025-03-16T18:18:40.749Z" }, + { url = "https://files.pythonhosted.org/packages/fa/03/74c5b631ee1ded596945c12027649e6344614144369fd3ec1aaced782882/numpy-2.2.4-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:92bda934a791c01d6d9d8e038363c50918ef7c40601552a58ac84c9613a665bc", size = 15571717, upload-time = "2025-03-16T18:19:04.512Z" }, + { url = "https://files.pythonhosted.org/packages/cb/dc/4fc7c0283abe0981e3b89f9b332a134e237dd476b0c018e1e21083310c31/numpy-2.2.4-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:ee4d528022f4c5ff67332469e10efe06a267e32f4067dc76bb7e2cddf3cd25ff", size = 17879998, upload-time = "2025-03-16T18:19:32.52Z" }, + { url = "https://files.pythonhosted.org/packages/e5/2b/878576190c5cfa29ed896b518cc516aecc7c98a919e20706c12480465f43/numpy-2.2.4-cp313-cp313t-win32.whl", hash = "sha256:05c076d531e9998e7e694c36e8b349969c56eadd2cdcd07242958489d79a7286", size = 6366896, upload-time = "2025-03-16T18:19:43.55Z" }, + { url = "https://files.pythonhosted.org/packages/3e/05/eb7eec66b95cf697f08c754ef26c3549d03ebd682819f794cb039574a0a6/numpy-2.2.4-cp313-cp313t-win_amd64.whl", hash = "sha256:188dcbca89834cc2e14eb2f106c96d6d46f200fe0200310fc29089657379c58d", size = 12739119, upload-time = "2025-03-16T18:20:03.94Z" }, + { url = "https://files.pythonhosted.org/packages/b2/5c/f09c33a511aff41a098e6ef3498465d95f6360621034a3d95f47edbc9119/numpy-2.2.4-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7051ee569db5fbac144335e0f3b9c2337e0c8d5c9fee015f259a5bd70772b7e8", size = 21081956, upload-time = "2025-03-16T18:21:12.955Z" }, + { url = "https://files.pythonhosted.org/packages/ba/30/74c48b3b6494c4b820b7fa1781d441e94d87a08daa5b35d222f06ba41a6f/numpy-2.2.4-pp310-pypy310_pp73-macosx_14_0_x86_64.whl", hash = "sha256:ab2939cd5bec30a7430cbdb2287b63151b77cf9624de0532d629c9a1c59b1d5c", size = 6827143, upload-time = "2025-03-16T18:21:26.748Z" }, + { url = "https://files.pythonhosted.org/packages/54/f5/ab0d2f48b490535c7a80e05da4a98902b632369efc04f0e47bb31ca97d8f/numpy-2.2.4-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d0f35b19894a9e08639fd60a1ec1978cb7f5f7f1eace62f38dd36be8aecdef4d", size = 16233350, upload-time = "2025-03-16T18:21:53.902Z" }, + { url = "https://files.pythonhosted.org/packages/3b/3a/2f6d8c1f8e45d496bca6baaec93208035faeb40d5735c25afac092ec9a12/numpy-2.2.4-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:b4adfbbc64014976d2f91084915ca4e626fbf2057fb81af209c1a6d776d23e3d", size = 12857565, upload-time = "2025-03-16T18:22:17.631Z" }, ] [[package]] name = "nvidia-cublas-cu12" -version = "12.6.4.1" +version = "12.8.4.1" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/af/eb/ff4b8c503fa1f1796679dce648854d58751982426e4e4b37d6fce49d259c/nvidia_cublas_cu12-12.6.4.1-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:08ed2686e9875d01b58e3cb379c6896df8e76c75e0d4a7f7dace3d7b6d9ef8eb", size = 393138322, upload-time = "2024-11-20T17:40:25.65Z" }, + { url = "https://files.pythonhosted.org/packages/dc/61/e24b560ab2e2eaeb3c839129175fb330dfcfc29e5203196e5541a4c44682/nvidia_cublas_cu12-12.8.4.1-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:8ac4e771d5a348c551b2a426eda6193c19aa630236b418086020df5ba9667142", size = 594346921, upload-time = "2025-03-07T01:44:31.254Z" }, ] [[package]] name = "nvidia-cuda-cupti-cu12" -version = "12.6.80" +version = "12.8.90" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/49/60/7b6497946d74bcf1de852a21824d63baad12cd417db4195fc1bfe59db953/nvidia_cuda_cupti_cu12-12.6.80-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:6768bad6cab4f19e8292125e5f1ac8aa7d1718704012a0e3272a6f61c4bce132", size = 8917980, upload-time = "2024-11-20T17:36:04.019Z" }, - { url = "https://files.pythonhosted.org/packages/a5/24/120ee57b218d9952c379d1e026c4479c9ece9997a4fb46303611ee48f038/nvidia_cuda_cupti_cu12-12.6.80-py3-none-manylinux2014_x86_64.whl", hash = "sha256:a3eff6cdfcc6a4c35db968a06fcadb061cbc7d6dde548609a941ff8701b98b73", size = 8917972, upload-time = "2024-10-01T16:58:06.036Z" }, + { url = "https://files.pythonhosted.org/packages/f8/02/2adcaa145158bf1a8295d83591d22e4103dbfd821bcaf6f3f53151ca4ffa/nvidia_cuda_cupti_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:ea0cb07ebda26bb9b29ba82cda34849e73c166c18162d3913575b0c9db9a6182", size = 10248621, upload-time = "2025-03-07T01:40:21.213Z" }, ] [[package]] name = "nvidia-cuda-nvrtc-cu12" -version = "12.6.77" +version = "12.8.93" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/75/2e/46030320b5a80661e88039f59060d1790298b4718944a65a7f2aeda3d9e9/nvidia_cuda_nvrtc_cu12-12.6.77-py3-none-manylinux2014_x86_64.whl", hash = "sha256:35b0cc6ee3a9636d5409133e79273ce1f3fd087abb0532d2d2e8fff1fe9efc53", size = 23650380, upload-time = "2024-10-01T17:00:14.643Z" }, + { url = "https://files.pythonhosted.org/packages/05/6b/32f747947df2da6994e999492ab306a903659555dddc0fbdeb9d71f75e52/nvidia_cuda_nvrtc_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl", hash = "sha256:a7756528852ef889772a84c6cd89d41dfa74667e24cca16bb31f8f061e3e9994", size = 88040029, upload-time = "2025-03-07T01:42:13.562Z" }, ] [[package]] name = "nvidia-cuda-runtime-cu12" -version = "12.6.77" +version = "12.8.90" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e1/23/e717c5ac26d26cf39a27fbc076240fad2e3b817e5889d671b67f4f9f49c5/nvidia_cuda_runtime_cu12-12.6.77-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:ba3b56a4f896141e25e19ab287cd71e52a6a0f4b29d0d31609f60e3b4d5219b7", size = 897690, upload-time = "2024-11-20T17:35:30.697Z" }, - { url = "https://files.pythonhosted.org/packages/f0/62/65c05e161eeddbafeca24dc461f47de550d9fa8a7e04eb213e32b55cfd99/nvidia_cuda_runtime_cu12-12.6.77-py3-none-manylinux2014_x86_64.whl", hash = "sha256:a84d15d5e1da416dd4774cb42edf5e954a3e60cc945698dc1d5be02321c44dc8", size = 897678, upload-time = "2024-10-01T16:57:33.821Z" }, + { url = "https://files.pythonhosted.org/packages/0d/9b/a997b638fcd068ad6e4d53b8551a7d30fe8b404d6f1804abf1df69838932/nvidia_cuda_runtime_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:adade8dcbd0edf427b7204d480d6066d33902cab2a4707dcfc48a2d0fd44ab90", size = 954765, upload-time = "2025-03-07T01:40:01.615Z" }, ] [[package]] name = "nvidia-cudnn-cu12" -version = "9.5.1.17" +version = "9.10.2.21" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "nvidia-cublas-cu12", marker = "sys_platform == 'linux'" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/2a/78/4535c9c7f859a64781e43c969a3a7e84c54634e319a996d43ef32ce46f83/nvidia_cudnn_cu12-9.5.1.17-py3-none-manylinux_2_28_x86_64.whl", hash = "sha256:30ac3869f6db17d170e0e556dd6cc5eee02647abc31ca856634d5a40f82c15b2", size = 570988386, upload-time = "2024-10-25T19:54:26.39Z" }, + { url = "https://files.pythonhosted.org/packages/ba/51/e123d997aa098c61d029f76663dedbfb9bc8dcf8c60cbd6adbe42f76d049/nvidia_cudnn_cu12-9.10.2.21-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:949452be657fa16687d0930933f032835951ef0892b37d2d53824d1a84dc97a8", size = 706758467, upload-time = "2025-06-06T21:54:08.597Z" }, ] [[package]] name = "nvidia-cufft-cu12" -version = "11.3.0.4" +version = "11.3.3.83" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "nvidia-nvjitlink-cu12", marker = "sys_platform == 'linux'" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/8f/16/73727675941ab8e6ffd86ca3a4b7b47065edcca7a997920b831f8147c99d/nvidia_cufft_cu12-11.3.0.4-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:ccba62eb9cef5559abd5e0d54ceed2d9934030f51163df018532142a8ec533e5", size = 200221632, upload-time = "2024-11-20T17:41:32.357Z" }, - { url = "https://files.pythonhosted.org/packages/60/de/99ec247a07ea40c969d904fc14f3a356b3e2a704121675b75c366b694ee1/nvidia_cufft_cu12-11.3.0.4-py3-none-manylinux2014_x86_64.whl", hash = "sha256:768160ac89f6f7b459bee747e8d175dbf53619cfe74b2a5636264163138013ca", size = 200221622, upload-time = "2024-10-01T17:03:58.79Z" }, + { url = "https://files.pythonhosted.org/packages/1f/13/ee4e00f30e676b66ae65b4f08cb5bcbb8392c03f54f2d5413ea99a5d1c80/nvidia_cufft_cu12-11.3.3.83-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:4d2dd21ec0b88cf61b62e6b43564355e5222e4a3fb394cac0db101f2dd0d4f74", size = 193118695, upload-time = "2025-03-07T01:45:27.821Z" }, ] [[package]] name = "nvidia-cufile-cu12" -version = "1.11.1.6" +version = "1.13.1.3" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/b2/66/cc9876340ac68ae71b15c743ddb13f8b30d5244af344ec8322b449e35426/nvidia_cufile_cu12-1.11.1.6-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:cc23469d1c7e52ce6c1d55253273d32c565dd22068647f3aa59b3c6b005bf159", size = 1142103, upload-time = "2024-11-20T17:42:11.83Z" }, + { url = "https://files.pythonhosted.org/packages/bb/fe/1bcba1dfbfb8d01be8d93f07bfc502c93fa23afa6fd5ab3fc7c1df71038a/nvidia_cufile_cu12-1.13.1.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:1d069003be650e131b21c932ec3d8969c1715379251f8d23a1860554b1cb24fc", size = 1197834, upload-time = "2025-03-07T01:45:50.723Z" }, ] [[package]] name = "nvidia-curand-cu12" -version = "10.3.7.77" +version = "10.3.9.90" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/73/1b/44a01c4e70933637c93e6e1a8063d1e998b50213a6b65ac5a9169c47e98e/nvidia_curand_cu12-10.3.7.77-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:a42cd1344297f70b9e39a1e4f467a4e1c10f1da54ff7a85c12197f6c652c8bdf", size = 56279010, upload-time = "2024-11-20T17:42:50.958Z" }, - { url = "https://files.pythonhosted.org/packages/4a/aa/2c7ff0b5ee02eaef890c0ce7d4f74bc30901871c5e45dee1ae6d0083cd80/nvidia_curand_cu12-10.3.7.77-py3-none-manylinux2014_x86_64.whl", hash = "sha256:99f1a32f1ac2bd134897fc7a203f779303261268a65762a623bf30cc9fe79117", size = 56279000, upload-time = "2024-10-01T17:04:45.274Z" }, + { url = "https://files.pythonhosted.org/packages/fb/aa/6584b56dc84ebe9cf93226a5cde4d99080c8e90ab40f0c27bda7a0f29aa1/nvidia_curand_cu12-10.3.9.90-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:b32331d4f4df5d6eefa0554c565b626c7216f87a06a4f56fab27c3b68a830ec9", size = 63619976, upload-time = "2025-03-07T01:46:23.323Z" }, ] [[package]] name = "nvidia-cusolver-cu12" -version = "11.7.1.2" +version = "11.7.3.90" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "nvidia-cublas-cu12", marker = "sys_platform == 'linux'" }, @@ -2314,53 +2015,50 @@ dependencies = [ { name = "nvidia-nvjitlink-cu12", marker = "sys_platform == 'linux'" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/f0/6e/c2cf12c9ff8b872e92b4a5740701e51ff17689c4d726fca91875b07f655d/nvidia_cusolver_cu12-11.7.1.2-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:e9e49843a7707e42022babb9bcfa33c29857a93b88020c4e4434656a655b698c", size = 158229790, upload-time = "2024-11-20T17:43:43.211Z" }, - { url = "https://files.pythonhosted.org/packages/9f/81/baba53585da791d043c10084cf9553e074548408e04ae884cfe9193bd484/nvidia_cusolver_cu12-11.7.1.2-py3-none-manylinux2014_x86_64.whl", hash = "sha256:6cf28f17f64107a0c4d7802be5ff5537b2130bfc112f25d5a30df227058ca0e6", size = 158229780, upload-time = "2024-10-01T17:05:39.875Z" }, + { url = "https://files.pythonhosted.org/packages/85/48/9a13d2975803e8cf2777d5ed57b87a0b6ca2cc795f9a4f59796a910bfb80/nvidia_cusolver_cu12-11.7.3.90-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:4376c11ad263152bd50ea295c05370360776f8c3427b30991df774f9fb26c450", size = 267506905, upload-time = "2025-03-07T01:47:16.273Z" }, ] [[package]] name = "nvidia-cusparse-cu12" -version = "12.5.4.2" +version = "12.5.8.93" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "nvidia-nvjitlink-cu12", marker = "sys_platform == 'linux'" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/06/1e/b8b7c2f4099a37b96af5c9bb158632ea9e5d9d27d7391d7eb8fc45236674/nvidia_cusparse_cu12-12.5.4.2-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:7556d9eca156e18184b94947ade0fba5bb47d69cec46bf8660fd2c71a4b48b73", size = 216561367, upload-time = "2024-11-20T17:44:54.824Z" }, - { url = "https://files.pythonhosted.org/packages/43/ac/64c4316ba163e8217a99680c7605f779accffc6a4bcd0c778c12948d3707/nvidia_cusparse_cu12-12.5.4.2-py3-none-manylinux2014_x86_64.whl", hash = "sha256:23749a6571191a215cb74d1cdbff4a86e7b19f1200c071b3fcf844a5bea23a2f", size = 216561357, upload-time = "2024-10-01T17:06:29.861Z" }, + { url = "https://files.pythonhosted.org/packages/c2/f5/e1854cb2f2bcd4280c44736c93550cc300ff4b8c95ebe370d0aa7d2b473d/nvidia_cusparse_cu12-12.5.8.93-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:1ec05d76bbbd8b61b06a80e1eaf8cf4959c3d4ce8e711b65ebd0443bb0ebb13b", size = 288216466, upload-time = "2025-03-07T01:48:13.779Z" }, ] [[package]] name = "nvidia-cusparselt-cu12" -version = "0.6.3" +version = "0.7.1" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/3b/9a/72ef35b399b0e183bc2e8f6f558036922d453c4d8237dab26c666a04244b/nvidia_cusparselt_cu12-0.6.3-py3-none-manylinux2014_x86_64.whl", hash = "sha256:e5c8a26c36445dd2e6812f1177978a24e2d37cacce7e090f297a688d1ec44f46", size = 156785796, upload-time = "2024-10-15T21:29:17.709Z" }, + { url = "https://files.pythonhosted.org/packages/56/79/12978b96bd44274fe38b5dde5cfb660b1d114f70a65ef962bcbbed99b549/nvidia_cusparselt_cu12-0.7.1-py3-none-manylinux2014_x86_64.whl", hash = "sha256:f1bb701d6b930d5a7cea44c19ceb973311500847f81b634d802b7b539dc55623", size = 287193691, upload-time = "2025-02-26T00:15:44.104Z" }, ] [[package]] name = "nvidia-nccl-cu12" -version = "2.26.2" +version = "2.27.3" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/67/ca/f42388aed0fddd64ade7493dbba36e1f534d4e6fdbdd355c6a90030ae028/nvidia_nccl_cu12-2.26.2-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:694cf3879a206553cc9d7dbda76b13efaf610fdb70a50cba303de1b0d1530ac6", size = 201319755, upload-time = "2025-03-13T00:29:55.296Z" }, + { url = "https://files.pythonhosted.org/packages/5c/5b/4e4fff7bad39adf89f735f2bc87248c81db71205b62bcc0d5ca5b606b3c3/nvidia_nccl_cu12-2.27.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:adf27ccf4238253e0b826bce3ff5fa532d65fc42322c8bfdfaf28024c0fbe039", size = 322364134, upload-time = "2025-06-03T21:58:04.013Z" }, ] [[package]] name = "nvidia-nvjitlink-cu12" -version = "12.6.85" +version = "12.8.93" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/9d/d7/c5383e47c7e9bf1c99d5bd2a8c935af2b6d705ad831a7ec5c97db4d82f4f/nvidia_nvjitlink_cu12-12.6.85-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl", hash = "sha256:eedc36df9e88b682efe4309aa16b5b4e78c2407eac59e8c10a6a47535164369a", size = 19744971, upload-time = "2024-11-20T17:46:53.366Z" }, + { url = "https://files.pythonhosted.org/packages/f6/74/86a07f1d0f42998ca31312f998bd3b9a7eff7f52378f4f270c8679c77fb9/nvidia_nvjitlink_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl", hash = "sha256:81ff63371a7ebd6e6451970684f916be2eab07321b73c9d244dc2b4da7f73b88", size = 39254836, upload-time = "2025-03-07T01:49:55.661Z" }, ] [[package]] name = "nvidia-nvtx-cu12" -version = "12.6.77" +version = "12.8.90" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/56/9a/fff8376f8e3d084cd1530e1ef7b879bb7d6d265620c95c1b322725c694f4/nvidia_nvtx_cu12-12.6.77-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:b90bed3df379fa79afbd21be8e04a0314336b8ae16768b58f2d34cb1d04cd7d2", size = 89276, upload-time = "2024-11-20T17:38:27.621Z" }, - { url = "https://files.pythonhosted.org/packages/9e/4e/0d0c945463719429b7bd21dece907ad0bde437a2ff12b9b12fee94722ab0/nvidia_nvtx_cu12-12.6.77-py3-none-manylinux2014_x86_64.whl", hash = "sha256:6574241a3ec5fdc9334353ab8c479fe75841dbe8f4532a8fc97ce63503330ba1", size = 89265, upload-time = "2024-10-01T17:00:38.172Z" }, + { url = "https://files.pythonhosted.org/packages/a2/eb/86626c1bbc2edb86323022371c39aa48df6fd8b0a1647bc274577f72e90b/nvidia_nvtx_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:5b17e2001cc0d751a5bc2c6ec6d26ad95913324a4adb86788c944f8ce9ba441f", size = 89954, upload-time = "2025-03-07T01:42:44.131Z" }, ] [[package]] @@ -2399,6 +2097,97 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/8d/4c/e0370709aaf9d7ceb68f975cac559751e75954429a77e83202e680606560/opt_einsum_fx-0.1.4-py3-none-any.whl", hash = "sha256:85f489f4c7c31fd88d5faf9669c09e61ec37a30098809fdcfe2a08a9e42f23c9", size = 13213, upload-time = "2021-11-07T20:49:32.395Z" }, ] +[[package]] +name = "optree" +version = "0.17.0" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "typing-extensions" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/56/c7/0853e0c59b135dff770615d2713b547b6b3b5cde7c10995b4a5825244612/optree-0.17.0.tar.gz", hash = "sha256:5335a5ec44479920620d72324c66563bd705ab2a698605dd4b6ee67dbcad7ecd", size = 163111, upload-time = "2025-07-25T11:26:11.586Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/52/a0/d5795ac13390b04822f1c61699f684cde682b57bf0a2d6b406019e1762ae/optree-0.17.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:85ec183b8eec6efc9a5572c2a84c62214c949555efbc69ca2381aca6048d08df", size = 622371, upload-time = "2025-07-25T11:24:23.345Z" }, + { url = "https://files.pythonhosted.org/packages/53/8b/ae8ddb511e680eb9d61edd2f5245be88ce050456658fb165550144f9a509/optree-0.17.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:6e77b6e0b7bb3ecfeb9a92ba605ef21b39bff38829b745af993e2e2b474322e2", size = 337260, upload-time = "2025-07-25T11:24:25.291Z" }, + { url = "https://files.pythonhosted.org/packages/91/f9/6ca076fd4c6f16be031afdc711a2676c1ff15bd1717ee2e699179b1a29bc/optree-0.17.0-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:98990201f352dba253af1a995c1453818db5f08de4cae7355d85aa6023676a52", size = 350398, upload-time = "2025-07-25T11:24:26.672Z" }, + { url = "https://files.pythonhosted.org/packages/95/4c/81344cbdcf8ea8525a21c9d65892d7529010ee2146c53423b2e9a84441ba/optree-0.17.0-cp310-cp310-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:e1a40adf6bb78a6a4b4f480879de2cb6b57d46d680a4d9834aa824f41e69c0d9", size = 404834, upload-time = "2025-07-25T11:24:28.988Z" }, + { url = "https://files.pythonhosted.org/packages/e5/c4/ac1880372a89f5c21514a7965dfa23b1afb2ad683fb9804d366727de9ecf/optree-0.17.0-cp310-cp310-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:78a113436a0a440f900b2799584f3cc2b2eea1b245d81c3583af42ac003e333c", size = 402116, upload-time = "2025-07-25T11:24:30.396Z" }, + { url = "https://files.pythonhosted.org/packages/ff/72/ad6be4d6a03805cf3921b492494cb3371ca28060d5ad19d5a36e10c4d67d/optree-0.17.0-cp310-cp310-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:0e45c16018f4283f028cf839b707b7ac734e8056a31b7198a1577161fcbe146d", size = 398491, upload-time = "2025-07-25T11:24:31.725Z" }, + { url = "https://files.pythonhosted.org/packages/d9/c1/6827fb504351f9a3935699b0eb31c8a6af59d775ee78289a25e0ba54f732/optree-0.17.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b698613d821d80cc216a2444ebc3145c8bf671b55a2223058a6574c1483a65f6", size = 387957, upload-time = "2025-07-25T11:24:32.759Z" }, + { url = "https://files.pythonhosted.org/packages/21/3d/44b3cbe4c9245a13b2677e30db2aafadf00bda976a551d64a31dc92f4977/optree-0.17.0-cp310-cp310-win32.whl", hash = "sha256:d07bfd8ce803dbc005502a89fda5f5e078e237342eaa36fb0c46cfbdf750bc76", size = 280064, upload-time = "2025-07-25T11:24:33.875Z" }, + { url = "https://files.pythonhosted.org/packages/74/fa/83d4cd387043483ee23617b048829a1289bf54afe2f6cb98ec7b27133369/optree-0.17.0-cp310-cp310-win_amd64.whl", hash = "sha256:d009d368ef06b8757891b772cad24d4f84122bd1877f7674fb8227d6e15340b4", size = 304398, upload-time = "2025-07-25T11:24:34.844Z" }, + { url = "https://files.pythonhosted.org/packages/21/4f/752522f318683efa7bba1895667c9841165d0284f6dfadf601769f6398ce/optree-0.17.0-cp310-cp310-win_arm64.whl", hash = "sha256:3571085ed9a5f39ff78ef57def0e9607c6b3f0099b6910524a0b42f5d58e481e", size = 308260, upload-time = "2025-07-25T11:24:36.144Z" }, + { url = "https://files.pythonhosted.org/packages/d8/eb/389a7dae8b113064f53909707aea9d72372fdc2eb918c48783c443cb3438/optree-0.17.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:09fbc0e5e42b20cab11851dffb7abe2fdf289c45d29e5be2b50b4ea93d069a9f", size = 640773, upload-time = "2025-07-25T11:24:37.25Z" }, + { url = "https://files.pythonhosted.org/packages/2b/bb/2d78b524989cabb5720e85ea366addc8589b4bbd0ce3f5ea58e370e5636a/optree-0.17.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:90a5864689268eda75d90abded5d474ae0a7ae2608d510626724fb78a1955948", size = 346402, upload-time = "2025-07-25T11:24:38.25Z" }, + { url = "https://files.pythonhosted.org/packages/73/5c/13a2a864b0c0b39c3c193be534a195a3ab2463c7d0443d4a76e749e3ff83/optree-0.17.0-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3080c564c9760711aa72d1b4d700ce1417f99ad087136f415c4eb8221169e2a3", size = 362797, upload-time = "2025-07-25T11:24:39.509Z" }, + { url = "https://files.pythonhosted.org/packages/da/f5/ff7dcb5a0108ee89c2be09aed2ebd26a7e1333d8122031aa9d9322b24ee6/optree-0.17.0-cp311-cp311-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:834a8fb358b608240b3a38706a09b43974675624485fad64c8ee641dae2eb57d", size = 419450, upload-time = "2025-07-25T11:24:40.555Z" }, + { url = "https://files.pythonhosted.org/packages/1b/e6/48a97aefd18770b55e5ed456d8183891f325cdb6d90592e5f072ed6951f8/optree-0.17.0-cp311-cp311-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:1a2bd263e6b5621d000d0f94de1f245414fd5dbce365a24b7b89b1ed0ef56cf9", size = 417557, upload-time = "2025-07-25T11:24:42.396Z" }, + { url = "https://files.pythonhosted.org/packages/c4/b1/4e280edab8a86be47ec1f9bd9ed4b685d2e15f0950ae62b613b26d12a1da/optree-0.17.0-cp311-cp311-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:9b37daca4ad89339b1f5320cc61ac600dcf976adbb060769d36d5542d6ebfedf", size = 414174, upload-time = "2025-07-25T11:24:43.51Z" }, + { url = "https://files.pythonhosted.org/packages/db/3b/49a9a1986215dd342525974deeb17c260a83fee8fad147276fd710ac8718/optree-0.17.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a146a6917f3e28cfdc268ff1770aa696c346482dd3da681c3ff92153d94450ea", size = 402000, upload-time = "2025-07-25T11:24:44.819Z" }, + { url = "https://files.pythonhosted.org/packages/00/8d/13b79d3394b83f4b1c93daac336f0eca5cb1cd5f58e10618f2c2db779cb7/optree-0.17.0-cp311-cp311-win32.whl", hash = "sha256:6b0446803d08f6aaae84f82f03c51527f36dfa15850873fc0183792247bc0071", size = 285777, upload-time = "2025-07-25T11:24:45.976Z" }, + { url = "https://files.pythonhosted.org/packages/90/32/da5191a347e33a78c2804a0cbfaed8eecb758818efda4b4d70bfd9b9b38d/optree-0.17.0-cp311-cp311-win_amd64.whl", hash = "sha256:e39f4f00b2967116badd9617ad6aa9845d8327fe13b6dbf5bc36d8c7b4a5ea03", size = 313761, upload-time = "2025-07-25T11:24:47.047Z" }, + { url = "https://files.pythonhosted.org/packages/e1/ea/7cae17a37a8ef67a33c354fce6f136d5f253d5afa40f68701252b1b2c2a0/optree-0.17.0-cp311-cp311-win_arm64.whl", hash = "sha256:50d4dbcbca3e379cc6b374f9b5a5626ff7ea41df8373e26c3af41d89d8a4b3d5", size = 318242, upload-time = "2025-07-25T11:24:48.708Z" }, + { url = "https://files.pythonhosted.org/packages/79/ce/471ff57336630f2434238a8cb8401e0d714ee7d54a6117823fd85de5f656/optree-0.17.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:09156e2ea62cde66dcbd9a450a5517ad6bad07d4ffc98fab0982c1e4f538341a", size = 654627, upload-time = "2025-07-25T11:24:49.754Z" }, + { url = "https://files.pythonhosted.org/packages/aa/ef/3143b7840dd2daedf1257643119c0f3addd23cf90cc9d2efc88f8166931e/optree-0.17.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:750f24304d1d437c8b235d4bc9e4afda17d85950706c34a875c16049f707eeb4", size = 351124, upload-time = "2025-07-25T11:24:50.813Z" }, + { url = "https://files.pythonhosted.org/packages/41/90/e12dea2cb5d8a5e17bbe3011ed4e972b89c027272a816db4897589751cad/optree-0.17.0-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e13ae51a63d69db445f269a3a4fd1d6edb064a705188d007ea47c9f034788fc5", size = 365869, upload-time = "2025-07-25T11:24:51.807Z" }, + { url = "https://files.pythonhosted.org/packages/76/ee/21af214663960a479863cd6c03d7a0abc8123ea22a6ea34689c2eed88ccd/optree-0.17.0-cp312-cp312-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:5958f58423cc7870cb011c8c8f92687397380886e8c9d33adac752147e7bbc3f", size = 424465, upload-time = "2025-07-25T11:24:53.124Z" }, + { url = "https://files.pythonhosted.org/packages/54/a3/64b184a79373753f4f46a5cd301ea581f71d6dc1a5c103bd2394f0925d40/optree-0.17.0-cp312-cp312-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:970ae4e47727b4c5526fc583b87d29190e576f6a2b6c19e8671589b73d256250", size = 420686, upload-time = "2025-07-25T11:24:54.212Z" }, + { url = "https://files.pythonhosted.org/packages/6c/6d/b6051b0b1ef9a49df96a66e9e62fc02620d2115d1ba659888c94e67fcfc9/optree-0.17.0-cp312-cp312-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:54177fd3e6e05c08b66329e26d7d44b85f24125f25c6b74c921499a1b31b8f70", size = 421225, upload-time = "2025-07-25T11:24:55.213Z" }, + { url = "https://files.pythonhosted.org/packages/f6/f1/940bc959aaef9eede8bb1b1127833b0929c6ffa9268ec0f6cb19877e2027/optree-0.17.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e1959cfbc38c228c8195354967cda64887b96219924b7b3759e5ee355582c1ec", size = 408819, upload-time = "2025-07-25T11:24:56.315Z" }, + { url = "https://files.pythonhosted.org/packages/56/52/ce527556e27dbf77266c1b1bb313ca446c94bc6edd6d7a882dbded028197/optree-0.17.0-cp312-cp312-win32.whl", hash = "sha256:039ea98c0cd94a64040d6f6d21dbe5cd9731bb380d7893f78d6898672080a232", size = 289107, upload-time = "2025-07-25T11:24:57.357Z" }, + { url = "https://files.pythonhosted.org/packages/eb/f1/aecb0199d269ad8ea41a86182474f98378a72681facbd6a06e94c23a2d02/optree-0.17.0-cp312-cp312-win_amd64.whl", hash = "sha256:c3a21109f635ce353d116ed1d77a7dfd77b898bcdaccef3bf74881ce7d6d54d8", size = 314074, upload-time = "2025-07-25T11:24:58.499Z" }, + { url = "https://files.pythonhosted.org/packages/3a/20/615ad64d24318709a236163dd8620fa7879a7720bfd0c755604d3dceeb76/optree-0.17.0-cp312-cp312-win_arm64.whl", hash = "sha256:1a39f957299426d2d4aa36cbc1acd71edb198ff0f28ddb43029bf58efe34a9a1", size = 316409, upload-time = "2025-07-25T11:24:59.855Z" }, + { url = "https://files.pythonhosted.org/packages/21/04/9706d11b880186e9e9d66d7c21ce249b2ce0212645137cc13fdd18247c26/optree-0.17.0-cp313-cp313-ios_13_0_arm64_iphoneos.whl", hash = "sha256:b5995a3efce4b00a14049268a81ab0379656a41ddf3c3761e3b88937fca44d48", size = 348177, upload-time = "2025-07-25T11:25:00.999Z" }, + { url = "https://files.pythonhosted.org/packages/ae/4b/0415c18816818ac871c9f3d5c7c5f4ceb83baff03ed511c9c94591ace4bc/optree-0.17.0-cp313-cp313-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:d06e8143d16fe6c0708f3cc2807b5b65f815d60ee2b52f3d79e4022c95563482", size = 354389, upload-time = "2025-07-25T11:25:02.337Z" }, + { url = "https://files.pythonhosted.org/packages/88/4d/5ce687b3945a34f0f0e17765745f146473b47177badd93b5979374d6e29c/optree-0.17.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:9537c4f82fe454a689e124462f252c4911cd7c78c6277334e7132f8157fb85e8", size = 661629, upload-time = "2025-07-25T11:25:03.429Z" }, + { url = "https://files.pythonhosted.org/packages/45/17/52ec65b80b6a17a9b7242e4cbf569c3d8035e72c49b6a3baba73aed6aa16/optree-0.17.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:79e8a594002509163d218827476f522d4f9ee6436438d90251d28d413af6740c", size = 354967, upload-time = "2025-07-25T11:25:04.523Z" }, + { url = "https://files.pythonhosted.org/packages/dd/12/24d4a417fd325ec06cfbce52716ac4f816ef696653b868960ac2ccb28436/optree-0.17.0-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:dfeea4aa0fd354d27922aba63ff9d86e4e126c6bf89cfb02849e68515519f1a5", size = 368513, upload-time = "2025-07-25T11:25:05.548Z" }, + { url = "https://files.pythonhosted.org/packages/30/e2/34e392209933e2c582c67594a7a6b4851bca4015c83b51c7508384b616b4/optree-0.17.0-cp313-cp313-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:6b2ff8999a9b84d00f23a032b6b3f13678894432a335d024e0670b9880f238ca", size = 430378, upload-time = "2025-07-25T11:25:06.918Z" }, + { url = "https://files.pythonhosted.org/packages/5f/16/0a0d6139022e9a53ecb1212fb6fbc5b60eff824371071ef5f5fa481d8167/optree-0.17.0-cp313-cp313-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:ea8bef525432b38a84e7448348da1a2dc308375bce79c77675cc50a501305851", size = 423294, upload-time = "2025-07-25T11:25:08.043Z" }, + { url = "https://files.pythonhosted.org/packages/ef/60/2e083dabb6aff6d939d8aab16ba3dbe6eee9429597a13f3fca57b33cdcde/optree-0.17.0-cp313-cp313-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:f95b81aa67538d38316b184a6ff39a3725ee5c8555fba21dcb692f8d7c39302e", size = 424633, upload-time = "2025-07-25T11:25:09.141Z" }, + { url = "https://files.pythonhosted.org/packages/af/fd/0e4229b5fa3fd9d3c779a606c0f358ffbdfee717f49b3477facd04de2cec/optree-0.17.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e808a1125169ae90de623456ef2423eb84a8578a74f03fe48b06b8561c2cc31d", size = 414866, upload-time = "2025-07-25T11:25:10.214Z" }, + { url = "https://files.pythonhosted.org/packages/e7/81/976082e979d42d36f9f81ee300d8fe7e86ca87588b70e372a40cb9203c9b/optree-0.17.0-cp313-cp313-win32.whl", hash = "sha256:4f3e0c5b20a4ef5b5a2688b5a07221cf1d2a8b2a57f82cf0c601f9d16f71450b", size = 289505, upload-time = "2025-07-25T11:25:11.616Z" }, + { url = "https://files.pythonhosted.org/packages/fb/ab/5b2c75c262c106747b5fbf1603a94ca8047896e719c3219ca85cb2d9c300/optree-0.17.0-cp313-cp313-win_amd64.whl", hash = "sha256:057f95213e403ff3a975f287aef6b687299d0c4512d211de24b1b98050cd4fbf", size = 316703, upload-time = "2025-07-25T11:25:12.638Z" }, + { url = "https://files.pythonhosted.org/packages/68/d6/78c0c927867b60d9b010bac84eae4046c761084bf2ed8a8d25521965ab4f/optree-0.17.0-cp313-cp313-win_arm64.whl", hash = "sha256:749dbecfd04edd50493b35bfb1f5be350f31b384533301e2257d4b0d0132544c", size = 318098, upload-time = "2025-07-25T11:25:13.755Z" }, + { url = "https://files.pythonhosted.org/packages/98/fd/6b5fdf3430157eced42d193bb49805668a380c672cc40317efe1dea3d739/optree-0.17.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:98c11fae09c5861f42c400f0fa3851f3d58ceba347267d458332710f094d5f75", size = 750506, upload-time = "2025-07-25T11:25:15.267Z" }, + { url = "https://files.pythonhosted.org/packages/19/0a/d8acb03fbf2edfd240a55363d903fad577e880a30a3117b60545a2a31aa5/optree-0.17.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:0b9f25c47de72044d7e1f42e9ed4c765f0867d321a2e6d194bc5facf69316417", size = 399106, upload-time = "2025-07-25T11:25:16.671Z" }, + { url = "https://files.pythonhosted.org/packages/39/df/b8882f5519c85af146de3a79a08066a56fe634b23052c593fcedc70bfcd7/optree-0.17.0-cp313-cp313t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:8e45a13b35873712e095fe0f7fd6e9c4f98f3bd5af6f5dc33c17b80357bc97fc", size = 386945, upload-time = "2025-07-25T11:25:17.728Z" }, + { url = "https://files.pythonhosted.org/packages/ca/d7/91f4efb509bda601a1591465c4a5bd55320e4bafe06b294bf80754127b0e/optree-0.17.0-cp313-cp313t-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:bfaf04d833dc53e5cfccff3b564e934a49086158472e31d84df31fce6d4f7b1c", size = 444177, upload-time = "2025-07-25T11:25:18.749Z" }, + { url = "https://files.pythonhosted.org/packages/84/17/a4833006e925c6ed5c45ceb02e65c9e9a260e70da6523858fcf628481847/optree-0.17.0-cp313-cp313t-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:b4c1d030ac1c881803f5c8e23d241159ae403fd00cdf57625328f282fc671ebd", size = 439198, upload-time = "2025-07-25T11:25:19.865Z" }, + { url = "https://files.pythonhosted.org/packages/ef/d1/c08fc60f6dfcb1b86ca1fdc0add08a98412a1596cd45830acbdc309f2cdb/optree-0.17.0-cp313-cp313t-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:bd7738709970acab5d963896192b63b2718be93bb6c0bcea91895ea157fa2b13", size = 439391, upload-time = "2025-07-25T11:25:20.942Z" }, + { url = "https://files.pythonhosted.org/packages/05/8f/461e10201003e6ad6bff3c594a29a7e044454aba68c5f795f4c8386ce47c/optree-0.17.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1644bc24b6e93cafccfdeee44157c3d4ae9bb0af3e861300602d716699865b1a", size = 426555, upload-time = "2025-07-25T11:25:21.968Z" }, + { url = "https://files.pythonhosted.org/packages/b5/4a/334d579dcb1ecea722ad37b7a8b7b29bb05ab7fe4464479862932ffd1869/optree-0.17.0-cp313-cp313t-win32.whl", hash = "sha256:f6be1f6f045f326bd419285ee92ebb13f1317149cbea84ca73c5bf06109a61bb", size = 319949, upload-time = "2025-07-25T11:25:23.127Z" }, + { url = "https://files.pythonhosted.org/packages/c8/96/5879944aee653471ad2a1ca5194ece0ca5d59de7c1d1fc5682ea3fb42057/optree-0.17.0-cp313-cp313t-win_amd64.whl", hash = "sha256:9d06b89803b1c72044fa5f07c708e33af7fe38ca2f5001cc9b6463894105b052", size = 352862, upload-time = "2025-07-25T11:25:24.214Z" }, + { url = "https://files.pythonhosted.org/packages/0d/de/cc600c216db4caa5b9ec5372e0c7fa05cd38eacde7e519c969ceab8712b6/optree-0.17.0-cp313-cp313t-win_arm64.whl", hash = "sha256:43f243d04fdba644647b1cabbfe4d7ca5fdb16c02e6d7d56e638d3e0b73566e8", size = 352101, upload-time = "2025-07-25T11:25:25.318Z" }, + { url = "https://files.pythonhosted.org/packages/d5/f7/cc6e920faaf96f78e373bf4ca83f806a40892104c0d437ab03402afeb94d/optree-0.17.0-cp314-cp314-macosx_10_13_universal2.whl", hash = "sha256:8808e0b6bd9d0288b76cac6ed5d589532c9c4f3f2b88157c70591e8a0cc9aa3b", size = 662838, upload-time = "2025-07-25T11:25:26.439Z" }, + { url = "https://files.pythonhosted.org/packages/22/fd/a8859f401de8305bd09f6f0f7491e6153cf8e50a8390eaa2b9d0e1f1fc95/optree-0.17.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:80c9dd735e7990a48f3da981125df6c10c9990d1876be7a034357aece600e07f", size = 355857, upload-time = "2025-07-25T11:25:27.55Z" }, + { url = "https://files.pythonhosted.org/packages/3c/21/6480d23b52b2e23b976fe254b9fbdc4b514e90a349b1ee73565b185c69f1/optree-0.17.0-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:dd21e0a89806cc3b86aaa578a73897d56085038fe432043534a23b2e559d7691", size = 369929, upload-time = "2025-07-25T11:25:28.897Z" }, + { url = "https://files.pythonhosted.org/packages/b3/29/69bb26473ff862a1792f5568c977e7a2580e08afe0fdcd7a7b3e1e4d6933/optree-0.17.0-cp314-cp314-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:9211c61285b8b3e42fd0e803cebd6e2b0987d8b2edffe45b42923debca09a9df", size = 430381, upload-time = "2025-07-25T11:25:29.984Z" }, + { url = "https://files.pythonhosted.org/packages/c8/8b/2c0a38c0d0c2396d698b97216cd6814d6754d11997b6ac66c57d87d71bae/optree-0.17.0-cp314-cp314-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:87938255749a45979c4e331627cb33d81aa08b0a09d024368b3e25ff67f0e9f2", size = 424461, upload-time = "2025-07-25T11:25:31.116Z" }, + { url = "https://files.pythonhosted.org/packages/a7/77/08fda3f97621190d50762225ee8bad87463a8b3a55fba451a999971ff130/optree-0.17.0-cp314-cp314-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:3432858145fd1955a3be12207507466ac40a6911f428bf5d2d6c7f67486530a2", size = 427234, upload-time = "2025-07-25T11:25:32.289Z" }, + { url = "https://files.pythonhosted.org/packages/ea/b5/b4f19952c36d6448c85a6ef6be5f916dd13548de2b684ab123f04b450850/optree-0.17.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5afe3e9e2f6da0a0a5c0892f32f675eb88965036b061aa555b74e6c412a05e17", size = 413863, upload-time = "2025-07-25T11:25:33.379Z" }, + { url = "https://files.pythonhosted.org/packages/b7/8c/1da744bb0cc550aed105f8a252fa8d8270067c5e21db7b95e457f76701da/optree-0.17.0-cp314-cp314-win32.whl", hash = "sha256:db6ce8e0d8585621230446736fa99c2883b34f9e56784957f69c47e2de34bdb4", size = 294314, upload-time = "2025-07-25T11:25:34.49Z" }, + { url = "https://files.pythonhosted.org/packages/84/05/5865e2a33c535c6b47378a43605de17cc286de59b93dc7814eb122861963/optree-0.17.0-cp314-cp314-win_amd64.whl", hash = "sha256:aa963de4146fa1b5cdffb479d324262f245c957df0bb9a9b37f6fd559d027acc", size = 323848, upload-time = "2025-07-25T11:25:35.511Z" }, + { url = "https://files.pythonhosted.org/packages/f1/01/55321c0d7b6bb60d88e5f5927216bcdc03e99f1f42567a0bcc23e786554e/optree-0.17.0-cp314-cp314-win_arm64.whl", hash = "sha256:855bfc78eba74748f931be6d6b739a9b03ac82a5c96511d66f310659903f6812", size = 325642, upload-time = "2025-07-25T11:25:36.649Z" }, + { url = "https://files.pythonhosted.org/packages/ee/be/24ef1e0d4212aedb087ff7b7a324426a093172327ecf9c33d2cf4cb6a69c/optree-0.17.0-cp314-cp314t-macosx_10_13_universal2.whl", hash = "sha256:0ac9626a51148c8497e82e9a9c21746795e179fbdec0b01c1644031e25f0d97e", size = 750484, upload-time = "2025-07-25T11:25:37.897Z" }, + { url = "https://files.pythonhosted.org/packages/4e/80/fc26e7c120849297992b0ecf8e435f213a379cc7923ea6ab1bad7b7d9c3f/optree-0.17.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:769c74ac289cdf108986fad2a36f24f4dd5ac6cf62919f99facdce943cd37359", size = 399067, upload-time = "2025-07-25T11:25:38.953Z" }, + { url = "https://files.pythonhosted.org/packages/88/42/6003f13e66cfbe7f0011bf8509da2479aba93068cdb9d79bf46010255089/optree-0.17.0-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5739c03a3362be42cb7649e82457c90aa818aa3e82af9681d3100c3346f4a90f", size = 386975, upload-time = "2025-07-25T11:25:40.376Z" }, + { url = "https://files.pythonhosted.org/packages/d0/53/621642abd76eda5a941b47adc98be81f0052683160be776499d11b4af83d/optree-0.17.0-cp314-cp314t-manylinux_2_26_i686.manylinux_2_28_i686.whl", hash = "sha256:ee07b59a08bd45aedd5252241a98841f1a5082a7b9b73df2dae6a433aa2a91d8", size = 444173, upload-time = "2025-07-25T11:25:41.474Z" }, + { url = "https://files.pythonhosted.org/packages/5b/d3/8819a2d5105a240d6793d11a61d597db91756ce84da5cee08808c6b8f61f/optree-0.17.0-cp314-cp314t-manylinux_2_26_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:875c017890a4b5d566af5593cab67fe3c4845544942af57e6bb9dea17e060297", size = 439080, upload-time = "2025-07-25T11:25:42.605Z" }, + { url = "https://files.pythonhosted.org/packages/c6/ef/9dbd34dfd1ad89feb239ca9925897a14ac94f190379a3bd991afdfd94186/optree-0.17.0-cp314-cp314t-manylinux_2_26_s390x.manylinux_2_28_s390x.whl", hash = "sha256:ffa5686191139f763e13445a169765c83517164bc28e60dbedb19bed2b2655f1", size = 439422, upload-time = "2025-07-25T11:25:43.672Z" }, + { url = "https://files.pythonhosted.org/packages/86/ca/a7a7549af2951925a692df508902ed2a6a94a51bc846806d2281b1029ef9/optree-0.17.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:575cf48cc2190acb565bd2b26b6f9b15c4e3b60183e86031215badc9d5441345", size = 426579, upload-time = "2025-07-25T11:25:44.765Z" }, + { url = "https://files.pythonhosted.org/packages/e6/0c/eb4d8ef38f1b51116095985b350ac9eede7a71d40c2ffaa283e9646b04e0/optree-0.17.0-cp314-cp314t-win32.whl", hash = "sha256:f1897de02364b7ef4a5bb56ae352b674ebf2cdd33da2b0f3543340282dc1f3e1", size = 329053, upload-time = "2025-07-25T11:25:45.845Z" }, + { url = "https://files.pythonhosted.org/packages/18/c6/f8e8c339e384578e3300215c732c20033f97d5ceb4c3d23a38bdb3527d98/optree-0.17.0-cp314-cp314t-win_amd64.whl", hash = "sha256:08df33cf74518f74b1c1f4ac0b760f544796a0b1cede91191c4daea0df3f314c", size = 367555, upload-time = "2025-07-25T11:25:46.95Z" }, + { url = "https://files.pythonhosted.org/packages/97/6f/1358550954dbbbb93b23fc953800e1ff2283024505255b0f9ba901f25e0e/optree-0.17.0-cp314-cp314t-win_arm64.whl", hash = "sha256:93d08d17b7b1d82b51ee7dd3a5a21ae2391fb30fc65a1369d4855c484923b967", size = 359135, upload-time = "2025-07-25T11:25:48.062Z" }, + { url = "https://files.pythonhosted.org/packages/ca/52/350c58dce327257afd77b92258e43d0bfe00416fc167b0c256ec86dcf9e7/optree-0.17.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:f365328450c1072e7a707dce67eaa6db3f63671907c866e3751e317b27ea187e", size = 342845, upload-time = "2025-07-25T11:26:01.651Z" }, + { url = "https://files.pythonhosted.org/packages/ed/d7/3036d15c028c447b1bd65dcf8f66cfd775bfa4e52daa74b82fb1d3c88faf/optree-0.17.0-pp310-pypy310_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:adde1427e0982cfc5f56939c26b4ebbd833091a176734c79fb95c78bdf833dff", size = 350952, upload-time = "2025-07-25T11:26:02.692Z" }, + { url = "https://files.pythonhosted.org/packages/71/45/e710024ef77324e745de48efd64f6270d8c209f14107a48ffef4049ac57a/optree-0.17.0-pp310-pypy310_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a80b7e5de5dd09b9c8b62d501e29a3850b047565c336c9d004b07ee1c01f4ae1", size = 389568, upload-time = "2025-07-25T11:26:04.094Z" }, + { url = "https://files.pythonhosted.org/packages/a8/63/b5cd1309f76f53e8a3cfbc88642647e58b1d3dd39f7cb0daf60ec516a252/optree-0.17.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:3c2c79652c45d82f23cbe08349456b1067ea513234a086b9a6bf1bcf128962a9", size = 306686, upload-time = "2025-07-25T11:26:05.511Z" }, + { url = "https://files.pythonhosted.org/packages/ca/40/afec131d9dd7a18d129190d407d97c95994f42b70c3d8ab897092d4de1d9/optree-0.17.0-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:bd92011cd0f2de40d28a95842819e778c476ab25c12731bfef1d1a0225554f83", size = 353955, upload-time = "2025-07-25T11:26:06.75Z" }, + { url = "https://files.pythonhosted.org/packages/69/c4/94a187ed3ca71194b9da6a276790e1703c7544c8f695ac915214ae8ce934/optree-0.17.0-pp311-pypy311_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f87f6f39015fc82d7adeee19900d246b89911319726e93cb2dbd4d1a809899bd", size = 363728, upload-time = "2025-07-25T11:26:07.959Z" }, + { url = "https://files.pythonhosted.org/packages/cd/99/23b7a484da8dfb814107b20ef2c93ef27c04f36aeb83bd976964a5b69e06/optree-0.17.0-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:58b0a83a967d2ef0f343db7182f0ad074eb1166bcaea909ae33909462013f151", size = 404649, upload-time = "2025-07-25T11:26:09.463Z" }, + { url = "https://files.pythonhosted.org/packages/bc/1f/7eca6da47eadb9ff2183bc9169eadde3dda0518e9a0187b99d5926fb2994/optree-0.17.0-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:e1ae8cbbcfaa45c57f5e51c544afa554cefbbb9fe9586c108aaf2aebfadf5899", size = 316368, upload-time = "2025-07-25T11:26:10.572Z" }, +] + [[package]] name = "orb-models" version = "0.5.4" @@ -2407,10 +2196,8 @@ dependencies = [ { name = "ase" }, { name = "cached-path" }, { name = "dm-tree" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, + { name = "scipy" }, { name = "torch" }, { name = "tqdm" }, ] @@ -2421,60 +2208,59 @@ wheels = [ [[package]] name = "packaging" -version = "25.0" +version = "24.2" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/a1/d4/1fc4078c65507b51b96ca8f8c3ba19e6a61c8253c72794544580a7b6c24d/packaging-25.0.tar.gz", hash = "sha256:d443872c98d677bf60f6a1f2f8c1cb748e8fe762d2bf9d3148b5599295b0fc4f", size = 165727, upload-time = "2025-04-19T11:48:59.673Z" } +sdist = { url = "https://files.pythonhosted.org/packages/d0/63/68dbb6eb2de9cb10ee4c9c14a0148804425e13c4fb20d61cce69f53106da/packaging-24.2.tar.gz", hash = "sha256:c228a6dc5e932d346bc5739379109d49e8853dd8223571c7c5b55260edc0b97f", size = 163950, upload-time = "2024-11-08T09:47:47.202Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/20/12/38679034af332785aac8774540895e234f4d07f7545804097de4b666afd8/packaging-25.0-py3-none-any.whl", hash = "sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484", size = 66469, upload-time = "2025-04-19T11:48:57.875Z" }, + { url = "https://files.pythonhosted.org/packages/88/ef/eb23f262cca3c0c4eb7ab1933c3b1f03d021f2c48f54763065b6f0e321be/packaging-24.2-py3-none-any.whl", hash = "sha256:09abb1bccd265c01f4a3aa3f7a7db064b36514d2cba19a2f694fe6150451a759", size = 65451, upload-time = "2024-11-08T09:47:44.722Z" }, ] [[package]] name = "pandas" -version = "2.3.1" +version = "2.2.3" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "python-dateutil" }, { name = "pytz" }, { name = "tzdata" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/d1/6f/75aa71f8a14267117adeeed5d21b204770189c0a0025acbdc03c337b28fc/pandas-2.3.1.tar.gz", hash = "sha256:0a95b9ac964fe83ce317827f80304d37388ea77616b1425f0ae41c9d2d0d7bb2", size = 4487493, upload-time = "2025-07-07T19:20:04.079Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/c4/ca/aa97b47287221fa37a49634532e520300088e290b20d690b21ce3e448143/pandas-2.3.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:22c2e866f7209ebc3a8f08d75766566aae02bcc91d196935a1d9e59c7b990ac9", size = 11542731, upload-time = "2025-07-07T19:18:12.619Z" }, - { url = "https://files.pythonhosted.org/packages/80/bf/7938dddc5f01e18e573dcfb0f1b8c9357d9b5fa6ffdee6e605b92efbdff2/pandas-2.3.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3583d348546201aff730c8c47e49bc159833f971c2899d6097bce68b9112a4f1", size = 10790031, upload-time = "2025-07-07T19:18:16.611Z" }, - { url = "https://files.pythonhosted.org/packages/ee/2f/9af748366763b2a494fed477f88051dbf06f56053d5c00eba652697e3f94/pandas-2.3.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0f951fbb702dacd390561e0ea45cdd8ecfa7fb56935eb3dd78e306c19104b9b0", size = 11724083, upload-time = "2025-07-07T19:18:20.512Z" }, - { url = "https://files.pythonhosted.org/packages/2c/95/79ab37aa4c25d1e7df953dde407bb9c3e4ae47d154bc0dd1692f3a6dcf8c/pandas-2.3.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:cd05b72ec02ebfb993569b4931b2e16fbb4d6ad6ce80224a3ee838387d83a191", size = 12342360, upload-time = "2025-07-07T19:18:23.194Z" }, - { url = "https://files.pythonhosted.org/packages/75/a7/d65e5d8665c12c3c6ff5edd9709d5836ec9b6f80071b7f4a718c6106e86e/pandas-2.3.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:1b916a627919a247d865aed068eb65eb91a344b13f5b57ab9f610b7716c92de1", size = 13202098, upload-time = "2025-07-07T19:18:25.558Z" }, - { url = "https://files.pythonhosted.org/packages/65/f3/4c1dbd754dbaa79dbf8b537800cb2fa1a6e534764fef50ab1f7533226c5c/pandas-2.3.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:fe67dc676818c186d5a3d5425250e40f179c2a89145df477dd82945eaea89e97", size = 13837228, upload-time = "2025-07-07T19:18:28.344Z" }, - { url = "https://files.pythonhosted.org/packages/3f/d6/d7f5777162aa9b48ec3910bca5a58c9b5927cfd9cfde3aa64322f5ba4b9f/pandas-2.3.1-cp310-cp310-win_amd64.whl", hash = "sha256:2eb789ae0274672acbd3c575b0598d213345660120a257b47b5dafdc618aec83", size = 11336561, upload-time = "2025-07-07T19:18:31.211Z" }, - { url = "https://files.pythonhosted.org/packages/76/1c/ccf70029e927e473a4476c00e0d5b32e623bff27f0402d0a92b7fc29bb9f/pandas-2.3.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2b0540963d83431f5ce8870ea02a7430adca100cec8a050f0811f8e31035541b", size = 11566608, upload-time = "2025-07-07T19:18:33.86Z" }, - { url = "https://files.pythonhosted.org/packages/ec/d3/3c37cb724d76a841f14b8f5fe57e5e3645207cc67370e4f84717e8bb7657/pandas-2.3.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:fe7317f578c6a153912bd2292f02e40c1d8f253e93c599e82620c7f69755c74f", size = 10823181, upload-time = "2025-07-07T19:18:36.151Z" }, - { url = "https://files.pythonhosted.org/packages/8a/4c/367c98854a1251940edf54a4df0826dcacfb987f9068abf3e3064081a382/pandas-2.3.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e6723a27ad7b244c0c79d8e7007092d7c8f0f11305770e2f4cd778b3ad5f9f85", size = 11793570, upload-time = "2025-07-07T19:18:38.385Z" }, - { url = "https://files.pythonhosted.org/packages/07/5f/63760ff107bcf5146eee41b38b3985f9055e710a72fdd637b791dea3495c/pandas-2.3.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3462c3735fe19f2638f2c3a40bd94ec2dc5ba13abbb032dd2fa1f540a075509d", size = 12378887, upload-time = "2025-07-07T19:18:41.284Z" }, - { url = "https://files.pythonhosted.org/packages/15/53/f31a9b4dfe73fe4711c3a609bd8e60238022f48eacedc257cd13ae9327a7/pandas-2.3.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:98bcc8b5bf7afed22cc753a28bc4d9e26e078e777066bc53fac7904ddef9a678", size = 13230957, upload-time = "2025-07-07T19:18:44.187Z" }, - { url = "https://files.pythonhosted.org/packages/e0/94/6fce6bf85b5056d065e0a7933cba2616dcb48596f7ba3c6341ec4bcc529d/pandas-2.3.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:4d544806b485ddf29e52d75b1f559142514e60ef58a832f74fb38e48d757b299", size = 13883883, upload-time = "2025-07-07T19:18:46.498Z" }, - { url = "https://files.pythonhosted.org/packages/c8/7b/bdcb1ed8fccb63d04bdb7635161d0ec26596d92c9d7a6cce964e7876b6c1/pandas-2.3.1-cp311-cp311-win_amd64.whl", hash = "sha256:b3cd4273d3cb3707b6fffd217204c52ed92859533e31dc03b7c5008aa933aaab", size = 11340212, upload-time = "2025-07-07T19:18:49.293Z" }, - { url = "https://files.pythonhosted.org/packages/46/de/b8445e0f5d217a99fe0eeb2f4988070908979bec3587c0633e5428ab596c/pandas-2.3.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:689968e841136f9e542020698ee1c4fbe9caa2ed2213ae2388dc7b81721510d3", size = 11588172, upload-time = "2025-07-07T19:18:52.054Z" }, - { url = "https://files.pythonhosted.org/packages/1e/e0/801cdb3564e65a5ac041ab99ea6f1d802a6c325bb6e58c79c06a3f1cd010/pandas-2.3.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:025e92411c16cbe5bb2a4abc99732a6b132f439b8aab23a59fa593eb00704232", size = 10717365, upload-time = "2025-07-07T19:18:54.785Z" }, - { url = "https://files.pythonhosted.org/packages/51/a5/c76a8311833c24ae61a376dbf360eb1b1c9247a5d9c1e8b356563b31b80c/pandas-2.3.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9b7ff55f31c4fcb3e316e8f7fa194566b286d6ac430afec0d461163312c5841e", size = 11280411, upload-time = "2025-07-07T19:18:57.045Z" }, - { url = "https://files.pythonhosted.org/packages/da/01/e383018feba0a1ead6cf5fe8728e5d767fee02f06a3d800e82c489e5daaf/pandas-2.3.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7dcb79bf373a47d2a40cf7232928eb7540155abbc460925c2c96d2d30b006eb4", size = 11988013, upload-time = "2025-07-07T19:18:59.771Z" }, - { url = "https://files.pythonhosted.org/packages/5b/14/cec7760d7c9507f11c97d64f29022e12a6cc4fc03ac694535e89f88ad2ec/pandas-2.3.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:56a342b231e8862c96bdb6ab97170e203ce511f4d0429589c8ede1ee8ece48b8", size = 12767210, upload-time = "2025-07-07T19:19:02.944Z" }, - { url = "https://files.pythonhosted.org/packages/50/b9/6e2d2c6728ed29fb3d4d4d302504fb66f1a543e37eb2e43f352a86365cdf/pandas-2.3.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ca7ed14832bce68baef331f4d7f294411bed8efd032f8109d690df45e00c4679", size = 13440571, upload-time = "2025-07-07T19:19:06.82Z" }, - { url = "https://files.pythonhosted.org/packages/80/a5/3a92893e7399a691bad7664d977cb5e7c81cf666c81f89ea76ba2bff483d/pandas-2.3.1-cp312-cp312-win_amd64.whl", hash = "sha256:ac942bfd0aca577bef61f2bc8da8147c4ef6879965ef883d8e8d5d2dc3e744b8", size = 10987601, upload-time = "2025-07-07T19:19:09.589Z" }, - { url = "https://files.pythonhosted.org/packages/32/ed/ff0a67a2c5505e1854e6715586ac6693dd860fbf52ef9f81edee200266e7/pandas-2.3.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9026bd4a80108fac2239294a15ef9003c4ee191a0f64b90f170b40cfb7cf2d22", size = 11531393, upload-time = "2025-07-07T19:19:12.245Z" }, - { url = "https://files.pythonhosted.org/packages/c7/db/d8f24a7cc9fb0972adab0cc80b6817e8bef888cfd0024eeb5a21c0bb5c4a/pandas-2.3.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:6de8547d4fdb12421e2d047a2c446c623ff4c11f47fddb6b9169eb98ffba485a", size = 10668750, upload-time = "2025-07-07T19:19:14.612Z" }, - { url = "https://files.pythonhosted.org/packages/0f/b0/80f6ec783313f1e2356b28b4fd8d2148c378370045da918c73145e6aab50/pandas-2.3.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:782647ddc63c83133b2506912cc6b108140a38a37292102aaa19c81c83db2928", size = 11342004, upload-time = "2025-07-07T19:19:16.857Z" }, - { url = "https://files.pythonhosted.org/packages/e9/e2/20a317688435470872885e7fc8f95109ae9683dec7c50be29b56911515a5/pandas-2.3.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2ba6aff74075311fc88504b1db890187a3cd0f887a5b10f5525f8e2ef55bfdb9", size = 12050869, upload-time = "2025-07-07T19:19:19.265Z" }, - { url = "https://files.pythonhosted.org/packages/55/79/20d746b0a96c67203a5bee5fb4e00ac49c3e8009a39e1f78de264ecc5729/pandas-2.3.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:e5635178b387bd2ba4ac040f82bc2ef6e6b500483975c4ebacd34bec945fda12", size = 12750218, upload-time = "2025-07-07T19:19:21.547Z" }, - { url = "https://files.pythonhosted.org/packages/7c/0f/145c8b41e48dbf03dd18fdd7f24f8ba95b8254a97a3379048378f33e7838/pandas-2.3.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6f3bf5ec947526106399a9e1d26d40ee2b259c66422efdf4de63c848492d91bb", size = 13416763, upload-time = "2025-07-07T19:19:23.939Z" }, - { url = "https://files.pythonhosted.org/packages/b2/c0/54415af59db5cdd86a3d3bf79863e8cc3fa9ed265f0745254061ac09d5f2/pandas-2.3.1-cp313-cp313-win_amd64.whl", hash = "sha256:1c78cf43c8fde236342a1cb2c34bcff89564a7bfed7e474ed2fffa6aed03a956", size = 10987482, upload-time = "2025-07-07T19:19:42.699Z" }, - { url = "https://files.pythonhosted.org/packages/48/64/2fd2e400073a1230e13b8cd604c9bc95d9e3b962e5d44088ead2e8f0cfec/pandas-2.3.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:8dfc17328e8da77be3cf9f47509e5637ba8f137148ed0e9b5241e1baf526e20a", size = 12029159, upload-time = "2025-07-07T19:19:26.362Z" }, - { url = "https://files.pythonhosted.org/packages/d8/0a/d84fd79b0293b7ef88c760d7dca69828d867c89b6d9bc52d6a27e4d87316/pandas-2.3.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:ec6c851509364c59a5344458ab935e6451b31b818be467eb24b0fe89bd05b6b9", size = 11393287, upload-time = "2025-07-07T19:19:29.157Z" }, - { url = "https://files.pythonhosted.org/packages/50/ae/ff885d2b6e88f3c7520bb74ba319268b42f05d7e583b5dded9837da2723f/pandas-2.3.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:911580460fc4884d9b05254b38a6bfadddfcc6aaef856fb5859e7ca202e45275", size = 11309381, upload-time = "2025-07-07T19:19:31.436Z" }, - { url = "https://files.pythonhosted.org/packages/85/86/1fa345fc17caf5d7780d2699985c03dbe186c68fee00b526813939062bb0/pandas-2.3.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2f4d6feeba91744872a600e6edbbd5b033005b431d5ae8379abee5bcfa479fab", size = 11883998, upload-time = "2025-07-07T19:19:34.267Z" }, - { url = "https://files.pythonhosted.org/packages/81/aa/e58541a49b5e6310d89474333e994ee57fea97c8aaa8fc7f00b873059bbf/pandas-2.3.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:fe37e757f462d31a9cd7580236a82f353f5713a80e059a29753cf938c6775d96", size = 12704705, upload-time = "2025-07-07T19:19:36.856Z" }, - { url = "https://files.pythonhosted.org/packages/d5/f9/07086f5b0f2a19872554abeea7658200824f5835c58a106fa8f2ae96a46c/pandas-2.3.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:5db9637dbc24b631ff3707269ae4559bce4b7fd75c1c4d7e13f40edc42df4444", size = 13189044, upload-time = "2025-07-07T19:19:39.999Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/9c/d6/9f8431bacc2e19dca897724cd097b1bb224a6ad5433784a44b587c7c13af/pandas-2.2.3.tar.gz", hash = "sha256:4f18ba62b61d7e192368b84517265a99b4d7ee8912f8708660fb4a366cc82667", size = 4399213, upload-time = "2024-09-20T13:10:04.827Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/aa/70/c853aec59839bceed032d52010ff5f1b8d87dc3114b762e4ba2727661a3b/pandas-2.2.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:1948ddde24197a0f7add2bdc4ca83bf2b1ef84a1bc8ccffd95eda17fd836ecb5", size = 12580827, upload-time = "2024-09-20T13:08:42.347Z" }, + { url = "https://files.pythonhosted.org/packages/99/f2/c4527768739ffa4469b2b4fff05aa3768a478aed89a2f271a79a40eee984/pandas-2.2.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:381175499d3802cde0eabbaf6324cce0c4f5d52ca6f8c377c29ad442f50f6348", size = 11303897, upload-time = "2024-09-20T13:08:45.807Z" }, + { url = "https://files.pythonhosted.org/packages/ed/12/86c1747ea27989d7a4064f806ce2bae2c6d575b950be087837bdfcabacc9/pandas-2.2.3-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:d9c45366def9a3dd85a6454c0e7908f2b3b8e9c138f5dc38fed7ce720d8453ed", size = 66480908, upload-time = "2024-09-20T18:37:13.513Z" }, + { url = "https://files.pythonhosted.org/packages/44/50/7db2cd5e6373ae796f0ddad3675268c8d59fb6076e66f0c339d61cea886b/pandas-2.2.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:86976a1c5b25ae3f8ccae3a5306e443569ee3c3faf444dfd0f41cda24667ad57", size = 13064210, upload-time = "2024-09-20T13:08:48.325Z" }, + { url = "https://files.pythonhosted.org/packages/61/61/a89015a6d5536cb0d6c3ba02cebed51a95538cf83472975275e28ebf7d0c/pandas-2.2.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b8661b0238a69d7aafe156b7fa86c44b881387509653fdf857bebc5e4008ad42", size = 16754292, upload-time = "2024-09-20T19:01:54.443Z" }, + { url = "https://files.pythonhosted.org/packages/ce/0d/4cc7b69ce37fac07645a94e1d4b0880b15999494372c1523508511b09e40/pandas-2.2.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:37e0aced3e8f539eccf2e099f65cdb9c8aa85109b0be6e93e2baff94264bdc6f", size = 14416379, upload-time = "2024-09-20T13:08:50.882Z" }, + { url = "https://files.pythonhosted.org/packages/31/9e/6ebb433de864a6cd45716af52a4d7a8c3c9aaf3a98368e61db9e69e69a9c/pandas-2.2.3-cp310-cp310-win_amd64.whl", hash = "sha256:56534ce0746a58afaf7942ba4863e0ef81c9c50d3f0ae93e9497d6a41a057645", size = 11598471, upload-time = "2024-09-20T13:08:53.332Z" }, + { url = "https://files.pythonhosted.org/packages/a8/44/d9502bf0ed197ba9bf1103c9867d5904ddcaf869e52329787fc54ed70cc8/pandas-2.2.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:66108071e1b935240e74525006034333f98bcdb87ea116de573a6a0dccb6c039", size = 12602222, upload-time = "2024-09-20T13:08:56.254Z" }, + { url = "https://files.pythonhosted.org/packages/52/11/9eac327a38834f162b8250aab32a6781339c69afe7574368fffe46387edf/pandas-2.2.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7c2875855b0ff77b2a64a0365e24455d9990730d6431b9e0ee18ad8acee13dbd", size = 11321274, upload-time = "2024-09-20T13:08:58.645Z" }, + { url = "https://files.pythonhosted.org/packages/45/fb/c4beeb084718598ba19aa9f5abbc8aed8b42f90930da861fcb1acdb54c3a/pandas-2.2.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:cd8d0c3be0515c12fed0bdbae072551c8b54b7192c7b1fda0ba56059a0179698", size = 15579836, upload-time = "2024-09-20T19:01:57.571Z" }, + { url = "https://files.pythonhosted.org/packages/cd/5f/4dba1d39bb9c38d574a9a22548c540177f78ea47b32f99c0ff2ec499fac5/pandas-2.2.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c124333816c3a9b03fbeef3a9f230ba9a737e9e5bb4060aa2107a86cc0a497fc", size = 13058505, upload-time = "2024-09-20T13:09:01.501Z" }, + { url = "https://files.pythonhosted.org/packages/b9/57/708135b90391995361636634df1f1130d03ba456e95bcf576fada459115a/pandas-2.2.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:63cc132e40a2e084cf01adf0775b15ac515ba905d7dcca47e9a251819c575ef3", size = 16744420, upload-time = "2024-09-20T19:02:00.678Z" }, + { url = "https://files.pythonhosted.org/packages/86/4a/03ed6b7ee323cf30404265c284cee9c65c56a212e0a08d9ee06984ba2240/pandas-2.2.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:29401dbfa9ad77319367d36940cd8a0b3a11aba16063e39632d98b0e931ddf32", size = 14440457, upload-time = "2024-09-20T13:09:04.105Z" }, + { url = "https://files.pythonhosted.org/packages/ed/8c/87ddf1fcb55d11f9f847e3c69bb1c6f8e46e2f40ab1a2d2abadb2401b007/pandas-2.2.3-cp311-cp311-win_amd64.whl", hash = "sha256:3fc6873a41186404dad67245896a6e440baacc92f5b716ccd1bc9ed2995ab2c5", size = 11617166, upload-time = "2024-09-20T13:09:06.917Z" }, + { url = "https://files.pythonhosted.org/packages/17/a3/fb2734118db0af37ea7433f57f722c0a56687e14b14690edff0cdb4b7e58/pandas-2.2.3-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:b1d432e8d08679a40e2a6d8b2f9770a5c21793a6f9f47fdd52c5ce1948a5a8a9", size = 12529893, upload-time = "2024-09-20T13:09:09.655Z" }, + { url = "https://files.pythonhosted.org/packages/e1/0c/ad295fd74bfac85358fd579e271cded3ac969de81f62dd0142c426b9da91/pandas-2.2.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:a5a1595fe639f5988ba6a8e5bc9649af3baf26df3998a0abe56c02609392e0a4", size = 11363475, upload-time = "2024-09-20T13:09:14.718Z" }, + { url = "https://files.pythonhosted.org/packages/c6/2a/4bba3f03f7d07207481fed47f5b35f556c7441acddc368ec43d6643c5777/pandas-2.2.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:5de54125a92bb4d1c051c0659e6fcb75256bf799a732a87184e5ea503965bce3", size = 15188645, upload-time = "2024-09-20T19:02:03.88Z" }, + { url = "https://files.pythonhosted.org/packages/38/f8/d8fddee9ed0d0c0f4a2132c1dfcf0e3e53265055da8df952a53e7eaf178c/pandas-2.2.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fffb8ae78d8af97f849404f21411c95062db1496aeb3e56f146f0355c9989319", size = 12739445, upload-time = "2024-09-20T13:09:17.621Z" }, + { url = "https://files.pythonhosted.org/packages/20/e8/45a05d9c39d2cea61ab175dbe6a2de1d05b679e8de2011da4ee190d7e748/pandas-2.2.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6dfcb5ee8d4d50c06a51c2fffa6cff6272098ad6540aed1a76d15fb9318194d8", size = 16359235, upload-time = "2024-09-20T19:02:07.094Z" }, + { url = "https://files.pythonhosted.org/packages/1d/99/617d07a6a5e429ff90c90da64d428516605a1ec7d7bea494235e1c3882de/pandas-2.2.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:062309c1b9ea12a50e8ce661145c6aab431b1e99530d3cd60640e255778bd43a", size = 14056756, upload-time = "2024-09-20T13:09:20.474Z" }, + { url = "https://files.pythonhosted.org/packages/29/d4/1244ab8edf173a10fd601f7e13b9566c1b525c4f365d6bee918e68381889/pandas-2.2.3-cp312-cp312-win_amd64.whl", hash = "sha256:59ef3764d0fe818125a5097d2ae867ca3fa64df032331b7e0917cf5d7bf66b13", size = 11504248, upload-time = "2024-09-20T13:09:23.137Z" }, + { url = "https://files.pythonhosted.org/packages/64/22/3b8f4e0ed70644e85cfdcd57454686b9057c6c38d2f74fe4b8bc2527214a/pandas-2.2.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:f00d1345d84d8c86a63e476bb4955e46458b304b9575dcf71102b5c705320015", size = 12477643, upload-time = "2024-09-20T13:09:25.522Z" }, + { url = "https://files.pythonhosted.org/packages/e4/93/b3f5d1838500e22c8d793625da672f3eec046b1a99257666c94446969282/pandas-2.2.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:3508d914817e153ad359d7e069d752cdd736a247c322d932eb89e6bc84217f28", size = 11281573, upload-time = "2024-09-20T13:09:28.012Z" }, + { url = "https://files.pythonhosted.org/packages/f5/94/6c79b07f0e5aab1dcfa35a75f4817f5c4f677931d4234afcd75f0e6a66ca/pandas-2.2.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:22a9d949bfc9a502d320aa04e5d02feab689d61da4e7764b62c30b991c42c5f0", size = 15196085, upload-time = "2024-09-20T19:02:10.451Z" }, + { url = "https://files.pythonhosted.org/packages/e8/31/aa8da88ca0eadbabd0a639788a6da13bb2ff6edbbb9f29aa786450a30a91/pandas-2.2.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f3a255b2c19987fbbe62a9dfd6cff7ff2aa9ccab3fc75218fd4b7530f01efa24", size = 12711809, upload-time = "2024-09-20T13:09:30.814Z" }, + { url = "https://files.pythonhosted.org/packages/ee/7c/c6dbdb0cb2a4344cacfb8de1c5808ca885b2e4dcfde8008266608f9372af/pandas-2.2.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:800250ecdadb6d9c78eae4990da62743b857b470883fa27f652db8bdde7f6659", size = 16356316, upload-time = "2024-09-20T19:02:13.825Z" }, + { url = "https://files.pythonhosted.org/packages/57/b7/8b757e7d92023b832869fa8881a992696a0bfe2e26f72c9ae9f255988d42/pandas-2.2.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6374c452ff3ec675a8f46fd9ab25c4ad0ba590b71cf0656f8b6daa5202bca3fb", size = 14022055, upload-time = "2024-09-20T13:09:33.462Z" }, + { url = "https://files.pythonhosted.org/packages/3b/bc/4b18e2b8c002572c5a441a64826252ce5da2aa738855747247a971988043/pandas-2.2.3-cp313-cp313-win_amd64.whl", hash = "sha256:61c5ad4043f791b61dd4752191d9f07f0ae412515d59ba8f005832a532f8736d", size = 11481175, upload-time = "2024-09-20T13:09:35.871Z" }, + { url = "https://files.pythonhosted.org/packages/76/a3/a5d88146815e972d40d19247b2c162e88213ef51c7c25993942c39dbf41d/pandas-2.2.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3b71f27954685ee685317063bf13c7709a7ba74fc996b84fc6821c59b0f06468", size = 12615650, upload-time = "2024-09-20T13:09:38.685Z" }, + { url = "https://files.pythonhosted.org/packages/9c/8c/f0fd18f6140ddafc0c24122c8a964e48294acc579d47def376fef12bcb4a/pandas-2.2.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:38cf8125c40dae9d5acc10fa66af8ea6fdf760b2714ee482ca691fc66e6fcb18", size = 11290177, upload-time = "2024-09-20T13:09:41.141Z" }, + { url = "https://files.pythonhosted.org/packages/ed/f9/e995754eab9c0f14c6777401f7eece0943840b7a9fc932221c19d1abee9f/pandas-2.2.3-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:ba96630bc17c875161df3818780af30e43be9b166ce51c9a18c1feae342906c2", size = 14651526, upload-time = "2024-09-20T19:02:16.905Z" }, + { url = "https://files.pythonhosted.org/packages/25/b0/98d6ae2e1abac4f35230aa756005e8654649d305df9a28b16b9ae4353bff/pandas-2.2.3-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1db71525a1538b30142094edb9adc10be3f3e176748cd7acc2240c2f2e5aa3a4", size = 11871013, upload-time = "2024-09-20T13:09:44.39Z" }, + { url = "https://files.pythonhosted.org/packages/cc/57/0f72a10f9db6a4628744c8e8f0df4e6e21de01212c7c981d31e50ffc8328/pandas-2.2.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:15c0e1e02e93116177d29ff83e8b1619c93ddc9c49083f237d4312337a61165d", size = 15711620, upload-time = "2024-09-20T19:02:20.639Z" }, + { url = "https://files.pythonhosted.org/packages/ab/5f/b38085618b950b79d2d9164a711c52b10aefc0ae6833b96f626b7021b2ed/pandas-2.2.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:ad5b65698ab28ed8d7f18790a0dc58005c7629f227be9ecc1072aa74c0c1d43a", size = 13098436, upload-time = "2024-09-20T13:09:48.112Z" }, ] [[package]] @@ -2491,8 +2277,7 @@ name = "patsy" version = "1.0.1" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, ] sdist = { url = "https://files.pythonhosted.org/packages/d1/81/74f6a65b848ffd16c18f920620ce999fe45fe27f01ab3911260ce4ed85e4/patsy-1.0.1.tar.gz", hash = "sha256:e786a9391eec818c054e359b737bbce692f051aee4c661f4141cc88fb459c0c4", size = 396010, upload-time = "2024-11-12T14:10:54.642Z" } wheels = [ @@ -2513,151 +2298,129 @@ wheels = [ [[package]] name = "pillow" -version = "11.3.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/f3/0d/d0d6dea55cd152ce3d6767bb38a8fc10e33796ba4ba210cbab9354b6d238/pillow-11.3.0.tar.gz", hash = "sha256:3828ee7586cd0b2091b6209e5ad53e20d0649bbe87164a459d0676e035e8f523", size = 47113069, upload-time = "2025-07-01T09:16:30.666Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/4c/5d/45a3553a253ac8763f3561371432a90bdbe6000fbdcf1397ffe502aa206c/pillow-11.3.0-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:1b9c17fd4ace828b3003dfd1e30bff24863e0eb59b535e8f80194d9cc7ecf860", size = 5316554, upload-time = "2025-07-01T09:13:39.342Z" }, - { url = "https://files.pythonhosted.org/packages/7c/c8/67c12ab069ef586a25a4a79ced553586748fad100c77c0ce59bb4983ac98/pillow-11.3.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:65dc69160114cdd0ca0f35cb434633c75e8e7fad4cf855177a05bf38678f73ad", size = 4686548, upload-time = "2025-07-01T09:13:41.835Z" }, - { url = "https://files.pythonhosted.org/packages/2f/bd/6741ebd56263390b382ae4c5de02979af7f8bd9807346d068700dd6d5cf9/pillow-11.3.0-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:7107195ddc914f656c7fc8e4a5e1c25f32e9236ea3ea860f257b0436011fddd0", size = 5859742, upload-time = "2025-07-03T13:09:47.439Z" }, - { url = "https://files.pythonhosted.org/packages/ca/0b/c412a9e27e1e6a829e6ab6c2dca52dd563efbedf4c9c6aa453d9a9b77359/pillow-11.3.0-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:cc3e831b563b3114baac7ec2ee86819eb03caa1a2cef0b481a5675b59c4fe23b", size = 7633087, upload-time = "2025-07-03T13:09:51.796Z" }, - { url = "https://files.pythonhosted.org/packages/59/9d/9b7076aaf30f5dd17e5e5589b2d2f5a5d7e30ff67a171eb686e4eecc2adf/pillow-11.3.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f1f182ebd2303acf8c380a54f615ec883322593320a9b00438eb842c1f37ae50", size = 5963350, upload-time = "2025-07-01T09:13:43.865Z" }, - { url = "https://files.pythonhosted.org/packages/f0/16/1a6bf01fb622fb9cf5c91683823f073f053005c849b1f52ed613afcf8dae/pillow-11.3.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4445fa62e15936a028672fd48c4c11a66d641d2c05726c7ec1f8ba6a572036ae", size = 6631840, upload-time = "2025-07-01T09:13:46.161Z" }, - { url = "https://files.pythonhosted.org/packages/7b/e6/6ff7077077eb47fde78739e7d570bdcd7c10495666b6afcd23ab56b19a43/pillow-11.3.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:71f511f6b3b91dd543282477be45a033e4845a40278fa8dcdbfdb07109bf18f9", size = 6074005, upload-time = "2025-07-01T09:13:47.829Z" }, - { url = "https://files.pythonhosted.org/packages/c3/3a/b13f36832ea6d279a697231658199e0a03cd87ef12048016bdcc84131601/pillow-11.3.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:040a5b691b0713e1f6cbe222e0f4f74cd233421e105850ae3b3c0ceda520f42e", size = 6708372, upload-time = "2025-07-01T09:13:52.145Z" }, - { url = "https://files.pythonhosted.org/packages/6c/e4/61b2e1a7528740efbc70b3d581f33937e38e98ef3d50b05007267a55bcb2/pillow-11.3.0-cp310-cp310-win32.whl", hash = "sha256:89bd777bc6624fe4115e9fac3352c79ed60f3bb18651420635f26e643e3dd1f6", size = 6277090, upload-time = "2025-07-01T09:13:53.915Z" }, - { url = "https://files.pythonhosted.org/packages/a9/d3/60c781c83a785d6afbd6a326ed4d759d141de43aa7365725cbcd65ce5e54/pillow-11.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:19d2ff547c75b8e3ff46f4d9ef969a06c30ab2d4263a9e287733aa8b2429ce8f", size = 6985988, upload-time = "2025-07-01T09:13:55.699Z" }, - { url = "https://files.pythonhosted.org/packages/9f/28/4f4a0203165eefb3763939c6789ba31013a2e90adffb456610f30f613850/pillow-11.3.0-cp310-cp310-win_arm64.whl", hash = "sha256:819931d25e57b513242859ce1876c58c59dc31587847bf74cfe06b2e0cb22d2f", size = 2422899, upload-time = "2025-07-01T09:13:57.497Z" }, - { url = "https://files.pythonhosted.org/packages/db/26/77f8ed17ca4ffd60e1dcd220a6ec6d71210ba398cfa33a13a1cd614c5613/pillow-11.3.0-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:1cd110edf822773368b396281a2293aeb91c90a2db00d78ea43e7e861631b722", size = 5316531, upload-time = "2025-07-01T09:13:59.203Z" }, - { url = "https://files.pythonhosted.org/packages/cb/39/ee475903197ce709322a17a866892efb560f57900d9af2e55f86db51b0a5/pillow-11.3.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:9c412fddd1b77a75aa904615ebaa6001f169b26fd467b4be93aded278266b288", size = 4686560, upload-time = "2025-07-01T09:14:01.101Z" }, - { url = "https://files.pythonhosted.org/packages/d5/90/442068a160fd179938ba55ec8c97050a612426fae5ec0a764e345839f76d/pillow-11.3.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:7d1aa4de119a0ecac0a34a9c8bde33f34022e2e8f99104e47a3ca392fd60e37d", size = 5870978, upload-time = "2025-07-03T13:09:55.638Z" }, - { url = "https://files.pythonhosted.org/packages/13/92/dcdd147ab02daf405387f0218dcf792dc6dd5b14d2573d40b4caeef01059/pillow-11.3.0-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:91da1d88226663594e3f6b4b8c3c8d85bd504117d043740a8e0ec449087cc494", size = 7641168, upload-time = "2025-07-03T13:10:00.37Z" }, - { url = "https://files.pythonhosted.org/packages/6e/db/839d6ba7fd38b51af641aa904e2960e7a5644d60ec754c046b7d2aee00e5/pillow-11.3.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:643f189248837533073c405ec2f0bb250ba54598cf80e8c1e043381a60632f58", size = 5973053, upload-time = "2025-07-01T09:14:04.491Z" }, - { url = "https://files.pythonhosted.org/packages/f2/2f/d7675ecae6c43e9f12aa8d58b6012683b20b6edfbdac7abcb4e6af7a3784/pillow-11.3.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:106064daa23a745510dabce1d84f29137a37224831d88eb4ce94bb187b1d7e5f", size = 6640273, upload-time = "2025-07-01T09:14:06.235Z" }, - { url = "https://files.pythonhosted.org/packages/45/ad/931694675ede172e15b2ff03c8144a0ddaea1d87adb72bb07655eaffb654/pillow-11.3.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cd8ff254faf15591e724dc7c4ddb6bf4793efcbe13802a4ae3e863cd300b493e", size = 6082043, upload-time = "2025-07-01T09:14:07.978Z" }, - { url = "https://files.pythonhosted.org/packages/3a/04/ba8f2b11fc80d2dd462d7abec16351b45ec99cbbaea4387648a44190351a/pillow-11.3.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:932c754c2d51ad2b2271fd01c3d121daaa35e27efae2a616f77bf164bc0b3e94", size = 6715516, upload-time = "2025-07-01T09:14:10.233Z" }, - { url = "https://files.pythonhosted.org/packages/48/59/8cd06d7f3944cc7d892e8533c56b0acb68399f640786313275faec1e3b6f/pillow-11.3.0-cp311-cp311-win32.whl", hash = "sha256:b4b8f3efc8d530a1544e5962bd6b403d5f7fe8b9e08227c6b255f98ad82b4ba0", size = 6274768, upload-time = "2025-07-01T09:14:11.921Z" }, - { url = "https://files.pythonhosted.org/packages/f1/cc/29c0f5d64ab8eae20f3232da8f8571660aa0ab4b8f1331da5c2f5f9a938e/pillow-11.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:1a992e86b0dd7aeb1f053cd506508c0999d710a8f07b4c791c63843fc6a807ac", size = 6986055, upload-time = "2025-07-01T09:14:13.623Z" }, - { url = "https://files.pythonhosted.org/packages/c6/df/90bd886fabd544c25addd63e5ca6932c86f2b701d5da6c7839387a076b4a/pillow-11.3.0-cp311-cp311-win_arm64.whl", hash = "sha256:30807c931ff7c095620fe04448e2c2fc673fcbb1ffe2a7da3fb39613489b1ddd", size = 2423079, upload-time = "2025-07-01T09:14:15.268Z" }, - { url = "https://files.pythonhosted.org/packages/40/fe/1bc9b3ee13f68487a99ac9529968035cca2f0a51ec36892060edcc51d06a/pillow-11.3.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:fdae223722da47b024b867c1ea0be64e0df702c5e0a60e27daad39bf960dd1e4", size = 5278800, upload-time = "2025-07-01T09:14:17.648Z" }, - { url = "https://files.pythonhosted.org/packages/2c/32/7e2ac19b5713657384cec55f89065fb306b06af008cfd87e572035b27119/pillow-11.3.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:921bd305b10e82b4d1f5e802b6850677f965d8394203d182f078873851dada69", size = 4686296, upload-time = "2025-07-01T09:14:19.828Z" }, - { url = "https://files.pythonhosted.org/packages/8e/1e/b9e12bbe6e4c2220effebc09ea0923a07a6da1e1f1bfbc8d7d29a01ce32b/pillow-11.3.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:eb76541cba2f958032d79d143b98a3a6b3ea87f0959bbe256c0b5e416599fd5d", size = 5871726, upload-time = "2025-07-03T13:10:04.448Z" }, - { url = "https://files.pythonhosted.org/packages/8d/33/e9200d2bd7ba00dc3ddb78df1198a6e80d7669cce6c2bdbeb2530a74ec58/pillow-11.3.0-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:67172f2944ebba3d4a7b54f2e95c786a3a50c21b88456329314caaa28cda70f6", size = 7644652, upload-time = "2025-07-03T13:10:10.391Z" }, - { url = "https://files.pythonhosted.org/packages/41/f1/6f2427a26fc683e00d985bc391bdd76d8dd4e92fac33d841127eb8fb2313/pillow-11.3.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:97f07ed9f56a3b9b5f49d3661dc9607484e85c67e27f3e8be2c7d28ca032fec7", size = 5977787, upload-time = "2025-07-01T09:14:21.63Z" }, - { url = "https://files.pythonhosted.org/packages/e4/c9/06dd4a38974e24f932ff5f98ea3c546ce3f8c995d3f0985f8e5ba48bba19/pillow-11.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:676b2815362456b5b3216b4fd5bd89d362100dc6f4945154ff172e206a22c024", size = 6645236, upload-time = "2025-07-01T09:14:23.321Z" }, - { url = "https://files.pythonhosted.org/packages/40/e7/848f69fb79843b3d91241bad658e9c14f39a32f71a301bcd1d139416d1be/pillow-11.3.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:3e184b2f26ff146363dd07bde8b711833d7b0202e27d13540bfe2e35a323a809", size = 6086950, upload-time = "2025-07-01T09:14:25.237Z" }, - { url = "https://files.pythonhosted.org/packages/0b/1a/7cff92e695a2a29ac1958c2a0fe4c0b2393b60aac13b04a4fe2735cad52d/pillow-11.3.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:6be31e3fc9a621e071bc17bb7de63b85cbe0bfae91bb0363c893cbe67247780d", size = 6723358, upload-time = "2025-07-01T09:14:27.053Z" }, - { url = "https://files.pythonhosted.org/packages/26/7d/73699ad77895f69edff76b0f332acc3d497f22f5d75e5360f78cbcaff248/pillow-11.3.0-cp312-cp312-win32.whl", hash = "sha256:7b161756381f0918e05e7cb8a371fff367e807770f8fe92ecb20d905d0e1c149", size = 6275079, upload-time = "2025-07-01T09:14:30.104Z" }, - { url = "https://files.pythonhosted.org/packages/8c/ce/e7dfc873bdd9828f3b6e5c2bbb74e47a98ec23cc5c74fc4e54462f0d9204/pillow-11.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:a6444696fce635783440b7f7a9fc24b3ad10a9ea3f0ab66c5905be1c19ccf17d", size = 6986324, upload-time = "2025-07-01T09:14:31.899Z" }, - { url = "https://files.pythonhosted.org/packages/16/8f/b13447d1bf0b1f7467ce7d86f6e6edf66c0ad7cf44cf5c87a37f9bed9936/pillow-11.3.0-cp312-cp312-win_arm64.whl", hash = "sha256:2aceea54f957dd4448264f9bf40875da0415c83eb85f55069d89c0ed436e3542", size = 2423067, upload-time = "2025-07-01T09:14:33.709Z" }, - { url = "https://files.pythonhosted.org/packages/1e/93/0952f2ed8db3a5a4c7a11f91965d6184ebc8cd7cbb7941a260d5f018cd2d/pillow-11.3.0-cp313-cp313-ios_13_0_arm64_iphoneos.whl", hash = "sha256:1c627742b539bba4309df89171356fcb3cc5a9178355b2727d1b74a6cf155fbd", size = 2128328, upload-time = "2025-07-01T09:14:35.276Z" }, - { url = "https://files.pythonhosted.org/packages/4b/e8/100c3d114b1a0bf4042f27e0f87d2f25e857e838034e98ca98fe7b8c0a9c/pillow-11.3.0-cp313-cp313-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:30b7c02f3899d10f13d7a48163c8969e4e653f8b43416d23d13d1bbfdc93b9f8", size = 2170652, upload-time = "2025-07-01T09:14:37.203Z" }, - { url = "https://files.pythonhosted.org/packages/aa/86/3f758a28a6e381758545f7cdb4942e1cb79abd271bea932998fc0db93cb6/pillow-11.3.0-cp313-cp313-ios_13_0_x86_64_iphonesimulator.whl", hash = "sha256:7859a4cc7c9295f5838015d8cc0a9c215b77e43d07a25e460f35cf516df8626f", size = 2227443, upload-time = "2025-07-01T09:14:39.344Z" }, - { url = "https://files.pythonhosted.org/packages/01/f4/91d5b3ffa718df2f53b0dc109877993e511f4fd055d7e9508682e8aba092/pillow-11.3.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:ec1ee50470b0d050984394423d96325b744d55c701a439d2bd66089bff963d3c", size = 5278474, upload-time = "2025-07-01T09:14:41.843Z" }, - { url = "https://files.pythonhosted.org/packages/f9/0e/37d7d3eca6c879fbd9dba21268427dffda1ab00d4eb05b32923d4fbe3b12/pillow-11.3.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:7db51d222548ccfd274e4572fdbf3e810a5e66b00608862f947b163e613b67dd", size = 4686038, upload-time = "2025-07-01T09:14:44.008Z" }, - { url = "https://files.pythonhosted.org/packages/ff/b0/3426e5c7f6565e752d81221af9d3676fdbb4f352317ceafd42899aaf5d8a/pillow-11.3.0-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2d6fcc902a24ac74495df63faad1884282239265c6839a0a6416d33faedfae7e", size = 5864407, upload-time = "2025-07-03T13:10:15.628Z" }, - { url = "https://files.pythonhosted.org/packages/fc/c1/c6c423134229f2a221ee53f838d4be9d82bab86f7e2f8e75e47b6bf6cd77/pillow-11.3.0-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f0f5d8f4a08090c6d6d578351a2b91acf519a54986c055af27e7a93feae6d3f1", size = 7639094, upload-time = "2025-07-03T13:10:21.857Z" }, - { url = "https://files.pythonhosted.org/packages/ba/c9/09e6746630fe6372c67c648ff9deae52a2bc20897d51fa293571977ceb5d/pillow-11.3.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c37d8ba9411d6003bba9e518db0db0c58a680ab9fe5179f040b0463644bc9805", size = 5973503, upload-time = "2025-07-01T09:14:45.698Z" }, - { url = "https://files.pythonhosted.org/packages/d5/1c/a2a29649c0b1983d3ef57ee87a66487fdeb45132df66ab30dd37f7dbe162/pillow-11.3.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:13f87d581e71d9189ab21fe0efb5a23e9f28552d5be6979e84001d3b8505abe8", size = 6642574, upload-time = "2025-07-01T09:14:47.415Z" }, - { url = "https://files.pythonhosted.org/packages/36/de/d5cc31cc4b055b6c6fd990e3e7f0f8aaf36229a2698501bcb0cdf67c7146/pillow-11.3.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:023f6d2d11784a465f09fd09a34b150ea4672e85fb3d05931d89f373ab14abb2", size = 6084060, upload-time = "2025-07-01T09:14:49.636Z" }, - { url = "https://files.pythonhosted.org/packages/d5/ea/502d938cbaeec836ac28a9b730193716f0114c41325db428e6b280513f09/pillow-11.3.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:45dfc51ac5975b938e9809451c51734124e73b04d0f0ac621649821a63852e7b", size = 6721407, upload-time = "2025-07-01T09:14:51.962Z" }, - { url = "https://files.pythonhosted.org/packages/45/9c/9c5e2a73f125f6cbc59cc7087c8f2d649a7ae453f83bd0362ff7c9e2aee2/pillow-11.3.0-cp313-cp313-win32.whl", hash = "sha256:a4d336baed65d50d37b88ca5b60c0fa9d81e3a87d4a7930d3880d1624d5b31f3", size = 6273841, upload-time = "2025-07-01T09:14:54.142Z" }, - { url = "https://files.pythonhosted.org/packages/23/85/397c73524e0cd212067e0c969aa245b01d50183439550d24d9f55781b776/pillow-11.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:0bce5c4fd0921f99d2e858dc4d4d64193407e1b99478bc5cacecba2311abde51", size = 6978450, upload-time = "2025-07-01T09:14:56.436Z" }, - { url = "https://files.pythonhosted.org/packages/17/d2/622f4547f69cd173955194b78e4d19ca4935a1b0f03a302d655c9f6aae65/pillow-11.3.0-cp313-cp313-win_arm64.whl", hash = "sha256:1904e1264881f682f02b7f8167935cce37bc97db457f8e7849dc3a6a52b99580", size = 2423055, upload-time = "2025-07-01T09:14:58.072Z" }, - { url = "https://files.pythonhosted.org/packages/dd/80/a8a2ac21dda2e82480852978416cfacd439a4b490a501a288ecf4fe2532d/pillow-11.3.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:4c834a3921375c48ee6b9624061076bc0a32a60b5532b322cc0ea64e639dd50e", size = 5281110, upload-time = "2025-07-01T09:14:59.79Z" }, - { url = "https://files.pythonhosted.org/packages/44/d6/b79754ca790f315918732e18f82a8146d33bcd7f4494380457ea89eb883d/pillow-11.3.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:5e05688ccef30ea69b9317a9ead994b93975104a677a36a8ed8106be9260aa6d", size = 4689547, upload-time = "2025-07-01T09:15:01.648Z" }, - { url = "https://files.pythonhosted.org/packages/49/20/716b8717d331150cb00f7fdd78169c01e8e0c219732a78b0e59b6bdb2fd6/pillow-11.3.0-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:1019b04af07fc0163e2810167918cb5add8d74674b6267616021ab558dc98ced", size = 5901554, upload-time = "2025-07-03T13:10:27.018Z" }, - { url = "https://files.pythonhosted.org/packages/74/cf/a9f3a2514a65bb071075063a96f0a5cf949c2f2fce683c15ccc83b1c1cab/pillow-11.3.0-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f944255db153ebb2b19c51fe85dd99ef0ce494123f21b9db4877ffdfc5590c7c", size = 7669132, upload-time = "2025-07-03T13:10:33.01Z" }, - { url = "https://files.pythonhosted.org/packages/98/3c/da78805cbdbee9cb43efe8261dd7cc0b4b93f2ac79b676c03159e9db2187/pillow-11.3.0-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1f85acb69adf2aaee8b7da124efebbdb959a104db34d3a2cb0f3793dbae422a8", size = 6005001, upload-time = "2025-07-01T09:15:03.365Z" }, - { url = "https://files.pythonhosted.org/packages/6c/fa/ce044b91faecf30e635321351bba32bab5a7e034c60187fe9698191aef4f/pillow-11.3.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:05f6ecbeff5005399bb48d198f098a9b4b6bdf27b8487c7f38ca16eeb070cd59", size = 6668814, upload-time = "2025-07-01T09:15:05.655Z" }, - { url = "https://files.pythonhosted.org/packages/7b/51/90f9291406d09bf93686434f9183aba27b831c10c87746ff49f127ee80cb/pillow-11.3.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:a7bc6e6fd0395bc052f16b1a8670859964dbd7003bd0af2ff08342eb6e442cfe", size = 6113124, upload-time = "2025-07-01T09:15:07.358Z" }, - { url = "https://files.pythonhosted.org/packages/cd/5a/6fec59b1dfb619234f7636d4157d11fb4e196caeee220232a8d2ec48488d/pillow-11.3.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:83e1b0161c9d148125083a35c1c5a89db5b7054834fd4387499e06552035236c", size = 6747186, upload-time = "2025-07-01T09:15:09.317Z" }, - { url = "https://files.pythonhosted.org/packages/49/6b/00187a044f98255225f172de653941e61da37104a9ea60e4f6887717e2b5/pillow-11.3.0-cp313-cp313t-win32.whl", hash = "sha256:2a3117c06b8fb646639dce83694f2f9eac405472713fcb1ae887469c0d4f6788", size = 6277546, upload-time = "2025-07-01T09:15:11.311Z" }, - { url = "https://files.pythonhosted.org/packages/e8/5c/6caaba7e261c0d75bab23be79f1d06b5ad2a2ae49f028ccec801b0e853d6/pillow-11.3.0-cp313-cp313t-win_amd64.whl", hash = "sha256:857844335c95bea93fb39e0fa2726b4d9d758850b34075a7e3ff4f4fa3aa3b31", size = 6985102, upload-time = "2025-07-01T09:15:13.164Z" }, - { url = "https://files.pythonhosted.org/packages/f3/7e/b623008460c09a0cb38263c93b828c666493caee2eb34ff67f778b87e58c/pillow-11.3.0-cp313-cp313t-win_arm64.whl", hash = "sha256:8797edc41f3e8536ae4b10897ee2f637235c94f27404cac7297f7b607dd0716e", size = 2424803, upload-time = "2025-07-01T09:15:15.695Z" }, - { url = "https://files.pythonhosted.org/packages/73/f4/04905af42837292ed86cb1b1dabe03dce1edc008ef14c473c5c7e1443c5d/pillow-11.3.0-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:d9da3df5f9ea2a89b81bb6087177fb1f4d1c7146d583a3fe5c672c0d94e55e12", size = 5278520, upload-time = "2025-07-01T09:15:17.429Z" }, - { url = "https://files.pythonhosted.org/packages/41/b0/33d79e377a336247df6348a54e6d2a2b85d644ca202555e3faa0cf811ecc/pillow-11.3.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:0b275ff9b04df7b640c59ec5a3cb113eefd3795a8df80bac69646ef699c6981a", size = 4686116, upload-time = "2025-07-01T09:15:19.423Z" }, - { url = "https://files.pythonhosted.org/packages/49/2d/ed8bc0ab219ae8768f529597d9509d184fe8a6c4741a6864fea334d25f3f/pillow-11.3.0-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:0743841cabd3dba6a83f38a92672cccbd69af56e3e91777b0ee7f4dba4385632", size = 5864597, upload-time = "2025-07-03T13:10:38.404Z" }, - { url = "https://files.pythonhosted.org/packages/b5/3d/b932bb4225c80b58dfadaca9d42d08d0b7064d2d1791b6a237f87f661834/pillow-11.3.0-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:2465a69cf967b8b49ee1b96d76718cd98c4e925414ead59fdf75cf0fd07df673", size = 7638246, upload-time = "2025-07-03T13:10:44.987Z" }, - { url = "https://files.pythonhosted.org/packages/09/b5/0487044b7c096f1b48f0d7ad416472c02e0e4bf6919541b111efd3cae690/pillow-11.3.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:41742638139424703b4d01665b807c6468e23e699e8e90cffefe291c5832b027", size = 5973336, upload-time = "2025-07-01T09:15:21.237Z" }, - { url = "https://files.pythonhosted.org/packages/a8/2d/524f9318f6cbfcc79fbc004801ea6b607ec3f843977652fdee4857a7568b/pillow-11.3.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:93efb0b4de7e340d99057415c749175e24c8864302369e05914682ba642e5d77", size = 6642699, upload-time = "2025-07-01T09:15:23.186Z" }, - { url = "https://files.pythonhosted.org/packages/6f/d2/a9a4f280c6aefedce1e8f615baaa5474e0701d86dd6f1dede66726462bbd/pillow-11.3.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:7966e38dcd0fa11ca390aed7c6f20454443581d758242023cf36fcb319b1a874", size = 6083789, upload-time = "2025-07-01T09:15:25.1Z" }, - { url = "https://files.pythonhosted.org/packages/fe/54/86b0cd9dbb683a9d5e960b66c7379e821a19be4ac5810e2e5a715c09a0c0/pillow-11.3.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:98a9afa7b9007c67ed84c57c9e0ad86a6000da96eaa638e4f8abe5b65ff83f0a", size = 6720386, upload-time = "2025-07-01T09:15:27.378Z" }, - { url = "https://files.pythonhosted.org/packages/e7/95/88efcaf384c3588e24259c4203b909cbe3e3c2d887af9e938c2022c9dd48/pillow-11.3.0-cp314-cp314-win32.whl", hash = "sha256:02a723e6bf909e7cea0dac1b0e0310be9d7650cd66222a5f1c571455c0a45214", size = 6370911, upload-time = "2025-07-01T09:15:29.294Z" }, - { url = "https://files.pythonhosted.org/packages/2e/cc/934e5820850ec5eb107e7b1a72dd278140731c669f396110ebc326f2a503/pillow-11.3.0-cp314-cp314-win_amd64.whl", hash = "sha256:a418486160228f64dd9e9efcd132679b7a02a5f22c982c78b6fc7dab3fefb635", size = 7117383, upload-time = "2025-07-01T09:15:31.128Z" }, - { url = "https://files.pythonhosted.org/packages/d6/e9/9c0a616a71da2a5d163aa37405e8aced9a906d574b4a214bede134e731bc/pillow-11.3.0-cp314-cp314-win_arm64.whl", hash = "sha256:155658efb5e044669c08896c0c44231c5e9abcaadbc5cd3648df2f7c0b96b9a6", size = 2511385, upload-time = "2025-07-01T09:15:33.328Z" }, - { url = "https://files.pythonhosted.org/packages/1a/33/c88376898aff369658b225262cd4f2659b13e8178e7534df9e6e1fa289f6/pillow-11.3.0-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:59a03cdf019efbfeeed910bf79c7c93255c3d54bc45898ac2a4140071b02b4ae", size = 5281129, upload-time = "2025-07-01T09:15:35.194Z" }, - { url = "https://files.pythonhosted.org/packages/1f/70/d376247fb36f1844b42910911c83a02d5544ebd2a8bad9efcc0f707ea774/pillow-11.3.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:f8a5827f84d973d8636e9dc5764af4f0cf2318d26744b3d902931701b0d46653", size = 4689580, upload-time = "2025-07-01T09:15:37.114Z" }, - { url = "https://files.pythonhosted.org/packages/eb/1c/537e930496149fbac69efd2fc4329035bbe2e5475b4165439e3be9cb183b/pillow-11.3.0-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:ee92f2fd10f4adc4b43d07ec5e779932b4eb3dbfbc34790ada5a6669bc095aa6", size = 5902860, upload-time = "2025-07-03T13:10:50.248Z" }, - { url = "https://files.pythonhosted.org/packages/bd/57/80f53264954dcefeebcf9dae6e3eb1daea1b488f0be8b8fef12f79a3eb10/pillow-11.3.0-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:c96d333dcf42d01f47b37e0979b6bd73ec91eae18614864622d9b87bbd5bbf36", size = 7670694, upload-time = "2025-07-03T13:10:56.432Z" }, - { url = "https://files.pythonhosted.org/packages/70/ff/4727d3b71a8578b4587d9c276e90efad2d6fe0335fd76742a6da08132e8c/pillow-11.3.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4c96f993ab8c98460cd0c001447bff6194403e8b1d7e149ade5f00594918128b", size = 6005888, upload-time = "2025-07-01T09:15:39.436Z" }, - { url = "https://files.pythonhosted.org/packages/05/ae/716592277934f85d3be51d7256f3636672d7b1abfafdc42cf3f8cbd4b4c8/pillow-11.3.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:41342b64afeba938edb034d122b2dda5db2139b9a4af999729ba8818e0056477", size = 6670330, upload-time = "2025-07-01T09:15:41.269Z" }, - { url = "https://files.pythonhosted.org/packages/e7/bb/7fe6cddcc8827b01b1a9766f5fdeb7418680744f9082035bdbabecf1d57f/pillow-11.3.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:068d9c39a2d1b358eb9f245ce7ab1b5c3246c7c8c7d9ba58cfa5b43146c06e50", size = 6114089, upload-time = "2025-07-01T09:15:43.13Z" }, - { url = "https://files.pythonhosted.org/packages/8b/f5/06bfaa444c8e80f1a8e4bff98da9c83b37b5be3b1deaa43d27a0db37ef84/pillow-11.3.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:a1bc6ba083b145187f648b667e05a2534ecc4b9f2784c2cbe3089e44868f2b9b", size = 6748206, upload-time = "2025-07-01T09:15:44.937Z" }, - { url = "https://files.pythonhosted.org/packages/f0/77/bc6f92a3e8e6e46c0ca78abfffec0037845800ea38c73483760362804c41/pillow-11.3.0-cp314-cp314t-win32.whl", hash = "sha256:118ca10c0d60b06d006be10a501fd6bbdfef559251ed31b794668ed569c87e12", size = 6377370, upload-time = "2025-07-01T09:15:46.673Z" }, - { url = "https://files.pythonhosted.org/packages/4a/82/3a721f7d69dca802befb8af08b7c79ebcab461007ce1c18bd91a5d5896f9/pillow-11.3.0-cp314-cp314t-win_amd64.whl", hash = "sha256:8924748b688aa210d79883357d102cd64690e56b923a186f35a82cbc10f997db", size = 7121500, upload-time = "2025-07-01T09:15:48.512Z" }, - { url = "https://files.pythonhosted.org/packages/89/c7/5572fa4a3f45740eaab6ae86fcdf7195b55beac1371ac8c619d880cfe948/pillow-11.3.0-cp314-cp314t-win_arm64.whl", hash = "sha256:79ea0d14d3ebad43ec77ad5272e6ff9bba5b679ef73375ea760261207fa8e0aa", size = 2512835, upload-time = "2025-07-01T09:15:50.399Z" }, - { url = "https://files.pythonhosted.org/packages/6f/8b/209bd6b62ce8367f47e68a218bffac88888fdf2c9fcf1ecadc6c3ec1ebc7/pillow-11.3.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:3cee80663f29e3843b68199b9d6f4f54bd1d4a6b59bdd91bceefc51238bcb967", size = 5270556, upload-time = "2025-07-01T09:16:09.961Z" }, - { url = "https://files.pythonhosted.org/packages/2e/e6/231a0b76070c2cfd9e260a7a5b504fb72da0a95279410fa7afd99d9751d6/pillow-11.3.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:b5f56c3f344f2ccaf0dd875d3e180f631dc60a51b314295a3e681fe8cf851fbe", size = 4654625, upload-time = "2025-07-01T09:16:11.913Z" }, - { url = "https://files.pythonhosted.org/packages/13/f4/10cf94fda33cb12765f2397fc285fa6d8eb9c29de7f3185165b702fc7386/pillow-11.3.0-pp310-pypy310_pp73-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e67d793d180c9df62f1f40aee3accca4829d3794c95098887edc18af4b8b780c", size = 4874207, upload-time = "2025-07-03T13:11:10.201Z" }, - { url = "https://files.pythonhosted.org/packages/72/c9/583821097dc691880c92892e8e2d41fe0a5a3d6021f4963371d2f6d57250/pillow-11.3.0-pp310-pypy310_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:d000f46e2917c705e9fb93a3606ee4a819d1e3aa7a9b442f6444f07e77cf5e25", size = 6583939, upload-time = "2025-07-03T13:11:15.68Z" }, - { url = "https://files.pythonhosted.org/packages/3b/8e/5c9d410f9217b12320efc7c413e72693f48468979a013ad17fd690397b9a/pillow-11.3.0-pp310-pypy310_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:527b37216b6ac3a12d7838dc3bd75208ec57c1c6d11ef01902266a5a0c14fc27", size = 4957166, upload-time = "2025-07-01T09:16:13.74Z" }, - { url = "https://files.pythonhosted.org/packages/62/bb/78347dbe13219991877ffb3a91bf09da8317fbfcd4b5f9140aeae020ad71/pillow-11.3.0-pp310-pypy310_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:be5463ac478b623b9dd3937afd7fb7ab3d79dd290a28e2b6df292dc75063eb8a", size = 5581482, upload-time = "2025-07-01T09:16:16.107Z" }, - { url = "https://files.pythonhosted.org/packages/d9/28/1000353d5e61498aaeaaf7f1e4b49ddb05f2c6575f9d4f9f914a3538b6e1/pillow-11.3.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:8dc70ca24c110503e16918a658b869019126ecfe03109b754c402daff12b3d9f", size = 6984596, upload-time = "2025-07-01T09:16:18.07Z" }, - { url = "https://files.pythonhosted.org/packages/9e/e3/6fa84033758276fb31da12e5fb66ad747ae83b93c67af17f8c6ff4cc8f34/pillow-11.3.0-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7c8ec7a017ad1bd562f93dbd8505763e688d388cde6e4a010ae1486916e713e6", size = 5270566, upload-time = "2025-07-01T09:16:19.801Z" }, - { url = "https://files.pythonhosted.org/packages/5b/ee/e8d2e1ab4892970b561e1ba96cbd59c0d28cf66737fc44abb2aec3795a4e/pillow-11.3.0-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:9ab6ae226de48019caa8074894544af5b53a117ccb9d3b3dcb2871464c829438", size = 4654618, upload-time = "2025-07-01T09:16:21.818Z" }, - { url = "https://files.pythonhosted.org/packages/f2/6d/17f80f4e1f0761f02160fc433abd4109fa1548dcfdca46cfdadaf9efa565/pillow-11.3.0-pp311-pypy311_pp73-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:fe27fb049cdcca11f11a7bfda64043c37b30e6b91f10cb5bab275806c32f6ab3", size = 4874248, upload-time = "2025-07-03T13:11:20.738Z" }, - { url = "https://files.pythonhosted.org/packages/de/5f/c22340acd61cef960130585bbe2120e2fd8434c214802f07e8c03596b17e/pillow-11.3.0-pp311-pypy311_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:465b9e8844e3c3519a983d58b80be3f668e2a7a5db97f2784e7079fbc9f9822c", size = 6583963, upload-time = "2025-07-03T13:11:26.283Z" }, - { url = "https://files.pythonhosted.org/packages/31/5e/03966aedfbfcbb4d5f8aa042452d3361f325b963ebbadddac05b122e47dd/pillow-11.3.0-pp311-pypy311_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5418b53c0d59b3824d05e029669efa023bbef0f3e92e75ec8428f3799487f361", size = 4957170, upload-time = "2025-07-01T09:16:23.762Z" }, - { url = "https://files.pythonhosted.org/packages/cc/2d/e082982aacc927fc2cab48e1e731bdb1643a1406acace8bed0900a61464e/pillow-11.3.0-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:504b6f59505f08ae014f724b6207ff6222662aab5cc9542577fb084ed0676ac7", size = 5581505, upload-time = "2025-07-01T09:16:25.593Z" }, - { url = "https://files.pythonhosted.org/packages/34/e7/ae39f538fd6844e982063c3a5e4598b8ced43b9633baa3a85ef33af8c05c/pillow-11.3.0-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:c84d689db21a1c397d001aa08241044aa2069e7587b398c8cc63020390b1c1b8", size = 6984598, upload-time = "2025-07-01T09:16:27.732Z" }, +version = "11.1.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/f3/af/c097e544e7bd278333db77933e535098c259609c4eb3b85381109602fb5b/pillow-11.1.0.tar.gz", hash = "sha256:368da70808b36d73b4b390a8ffac11069f8a5c85f29eff1f1b01bcf3ef5b2a20", size = 46742715, upload-time = "2025-01-02T08:13:58.407Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/50/1c/2dcea34ac3d7bc96a1fd1bd0a6e06a57c67167fec2cff8d95d88229a8817/pillow-11.1.0-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:e1abe69aca89514737465752b4bcaf8016de61b3be1397a8fc260ba33321b3a8", size = 3229983, upload-time = "2025-01-02T08:10:16.008Z" }, + { url = "https://files.pythonhosted.org/packages/14/ca/6bec3df25e4c88432681de94a3531cc738bd85dea6c7aa6ab6f81ad8bd11/pillow-11.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c640e5a06869c75994624551f45e5506e4256562ead981cce820d5ab39ae2192", size = 3101831, upload-time = "2025-01-02T08:10:18.774Z" }, + { url = "https://files.pythonhosted.org/packages/d4/2c/668e18e5521e46eb9667b09e501d8e07049eb5bfe39d56be0724a43117e6/pillow-11.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a07dba04c5e22824816b2615ad7a7484432d7f540e6fa86af60d2de57b0fcee2", size = 4314074, upload-time = "2025-01-02T08:10:21.114Z" }, + { url = "https://files.pythonhosted.org/packages/02/80/79f99b714f0fc25f6a8499ecfd1f810df12aec170ea1e32a4f75746051ce/pillow-11.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e267b0ed063341f3e60acd25c05200df4193e15a4a5807075cd71225a2386e26", size = 4394933, upload-time = "2025-01-02T08:10:23.982Z" }, + { url = "https://files.pythonhosted.org/packages/81/aa/8d4ad25dc11fd10a2001d5b8a80fdc0e564ac33b293bdfe04ed387e0fd95/pillow-11.1.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:bd165131fd51697e22421d0e467997ad31621b74bfc0b75956608cb2906dda07", size = 4353349, upload-time = "2025-01-02T08:10:25.887Z" }, + { url = "https://files.pythonhosted.org/packages/84/7a/cd0c3eaf4a28cb2a74bdd19129f7726277a7f30c4f8424cd27a62987d864/pillow-11.1.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:abc56501c3fd148d60659aae0af6ddc149660469082859fa7b066a298bde9482", size = 4476532, upload-time = "2025-01-02T08:10:28.129Z" }, + { url = "https://files.pythonhosted.org/packages/8f/8b/a907fdd3ae8f01c7670dfb1499c53c28e217c338b47a813af8d815e7ce97/pillow-11.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:54ce1c9a16a9561b6d6d8cb30089ab1e5eb66918cb47d457bd996ef34182922e", size = 4279789, upload-time = "2025-01-02T08:10:32.976Z" }, + { url = "https://files.pythonhosted.org/packages/6f/9a/9f139d9e8cccd661c3efbf6898967a9a337eb2e9be2b454ba0a09533100d/pillow-11.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:73ddde795ee9b06257dac5ad42fcb07f3b9b813f8c1f7f870f402f4dc54b5269", size = 4413131, upload-time = "2025-01-02T08:10:36.912Z" }, + { url = "https://files.pythonhosted.org/packages/a8/68/0d8d461f42a3f37432203c8e6df94da10ac8081b6d35af1c203bf3111088/pillow-11.1.0-cp310-cp310-win32.whl", hash = "sha256:3a5fe20a7b66e8135d7fd617b13272626a28278d0e578c98720d9ba4b2439d49", size = 2291213, upload-time = "2025-01-02T08:10:40.186Z" }, + { url = "https://files.pythonhosted.org/packages/14/81/d0dff759a74ba87715509af9f6cb21fa21d93b02b3316ed43bda83664db9/pillow-11.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:b6123aa4a59d75f06e9dd3dac5bf8bc9aa383121bb3dd9a7a612e05eabc9961a", size = 2625725, upload-time = "2025-01-02T08:10:42.404Z" }, + { url = "https://files.pythonhosted.org/packages/ce/1f/8d50c096a1d58ef0584ddc37e6f602828515219e9d2428e14ce50f5ecad1/pillow-11.1.0-cp310-cp310-win_arm64.whl", hash = "sha256:a76da0a31da6fcae4210aa94fd779c65c75786bc9af06289cd1c184451ef7a65", size = 2375213, upload-time = "2025-01-02T08:10:44.173Z" }, + { url = "https://files.pythonhosted.org/packages/dd/d6/2000bfd8d5414fb70cbbe52c8332f2283ff30ed66a9cde42716c8ecbe22c/pillow-11.1.0-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:e06695e0326d05b06833b40b7ef477e475d0b1ba3a6d27da1bb48c23209bf457", size = 3229968, upload-time = "2025-01-02T08:10:48.172Z" }, + { url = "https://files.pythonhosted.org/packages/d9/45/3fe487010dd9ce0a06adf9b8ff4f273cc0a44536e234b0fad3532a42c15b/pillow-11.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:96f82000e12f23e4f29346e42702b6ed9a2f2fea34a740dd5ffffcc8c539eb35", size = 3101806, upload-time = "2025-01-02T08:10:50.981Z" }, + { url = "https://files.pythonhosted.org/packages/e3/72/776b3629c47d9d5f1c160113158a7a7ad177688d3a1159cd3b62ded5a33a/pillow-11.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a3cd561ded2cf2bbae44d4605837221b987c216cff94f49dfeed63488bb228d2", size = 4322283, upload-time = "2025-01-02T08:10:54.724Z" }, + { url = "https://files.pythonhosted.org/packages/e4/c2/e25199e7e4e71d64eeb869f5b72c7ddec70e0a87926398785ab944d92375/pillow-11.1.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f189805c8be5ca5add39e6f899e6ce2ed824e65fb45f3c28cb2841911da19070", size = 4402945, upload-time = "2025-01-02T08:10:57.376Z" }, + { url = "https://files.pythonhosted.org/packages/c1/ed/51d6136c9d5911f78632b1b86c45241c712c5a80ed7fa7f9120a5dff1eba/pillow-11.1.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:dd0052e9db3474df30433f83a71b9b23bd9e4ef1de13d92df21a52c0303b8ab6", size = 4361228, upload-time = "2025-01-02T08:11:02.374Z" }, + { url = "https://files.pythonhosted.org/packages/48/a4/fbfe9d5581d7b111b28f1d8c2762dee92e9821bb209af9fa83c940e507a0/pillow-11.1.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:837060a8599b8f5d402e97197d4924f05a2e0d68756998345c829c33186217b1", size = 4484021, upload-time = "2025-01-02T08:11:04.431Z" }, + { url = "https://files.pythonhosted.org/packages/39/db/0b3c1a5018117f3c1d4df671fb8e47d08937f27519e8614bbe86153b65a5/pillow-11.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:aa8dd43daa836b9a8128dbe7d923423e5ad86f50a7a14dc688194b7be5c0dea2", size = 4287449, upload-time = "2025-01-02T08:11:07.412Z" }, + { url = "https://files.pythonhosted.org/packages/d9/58/bc128da7fea8c89fc85e09f773c4901e95b5936000e6f303222490c052f3/pillow-11.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:0a2f91f8a8b367e7a57c6e91cd25af510168091fb89ec5146003e424e1558a96", size = 4419972, upload-time = "2025-01-02T08:11:09.508Z" }, + { url = "https://files.pythonhosted.org/packages/5f/bb/58f34379bde9fe197f51841c5bbe8830c28bbb6d3801f16a83b8f2ad37df/pillow-11.1.0-cp311-cp311-win32.whl", hash = "sha256:c12fc111ef090845de2bb15009372175d76ac99969bdf31e2ce9b42e4b8cd88f", size = 2291201, upload-time = "2025-01-02T08:11:13.056Z" }, + { url = "https://files.pythonhosted.org/packages/3a/c6/fce9255272bcf0c39e15abd2f8fd8429a954cf344469eaceb9d0d1366913/pillow-11.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:fbd43429d0d7ed6533b25fc993861b8fd512c42d04514a0dd6337fb3ccf22761", size = 2625686, upload-time = "2025-01-02T08:11:16.547Z" }, + { url = "https://files.pythonhosted.org/packages/c8/52/8ba066d569d932365509054859f74f2a9abee273edcef5cd75e4bc3e831e/pillow-11.1.0-cp311-cp311-win_arm64.whl", hash = "sha256:f7955ecf5609dee9442cbface754f2c6e541d9e6eda87fad7f7a989b0bdb9d71", size = 2375194, upload-time = "2025-01-02T08:11:19.897Z" }, + { url = "https://files.pythonhosted.org/packages/95/20/9ce6ed62c91c073fcaa23d216e68289e19d95fb8188b9fb7a63d36771db8/pillow-11.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:2062ffb1d36544d42fcaa277b069c88b01bb7298f4efa06731a7fd6cc290b81a", size = 3226818, upload-time = "2025-01-02T08:11:22.518Z" }, + { url = "https://files.pythonhosted.org/packages/b9/d8/f6004d98579a2596c098d1e30d10b248798cceff82d2b77aa914875bfea1/pillow-11.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:a85b653980faad27e88b141348707ceeef8a1186f75ecc600c395dcac19f385b", size = 3101662, upload-time = "2025-01-02T08:11:25.19Z" }, + { url = "https://files.pythonhosted.org/packages/08/d9/892e705f90051c7a2574d9f24579c9e100c828700d78a63239676f960b74/pillow-11.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9409c080586d1f683df3f184f20e36fb647f2e0bc3988094d4fd8c9f4eb1b3b3", size = 4329317, upload-time = "2025-01-02T08:11:30.371Z" }, + { url = "https://files.pythonhosted.org/packages/8c/aa/7f29711f26680eab0bcd3ecdd6d23ed6bce180d82e3f6380fb7ae35fcf3b/pillow-11.1.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7fdadc077553621911f27ce206ffcbec7d3f8d7b50e0da39f10997e8e2bb7f6a", size = 4412999, upload-time = "2025-01-02T08:11:33.499Z" }, + { url = "https://files.pythonhosted.org/packages/c8/c4/8f0fe3b9e0f7196f6d0bbb151f9fba323d72a41da068610c4c960b16632a/pillow-11.1.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:93a18841d09bcdd774dcdc308e4537e1f867b3dec059c131fde0327899734aa1", size = 4368819, upload-time = "2025-01-02T08:11:37.304Z" }, + { url = "https://files.pythonhosted.org/packages/38/0d/84200ed6a871ce386ddc82904bfadc0c6b28b0c0ec78176871a4679e40b3/pillow-11.1.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:9aa9aeddeed452b2f616ff5507459e7bab436916ccb10961c4a382cd3e03f47f", size = 4496081, upload-time = "2025-01-02T08:11:39.598Z" }, + { url = "https://files.pythonhosted.org/packages/84/9c/9bcd66f714d7e25b64118e3952d52841a4babc6d97b6d28e2261c52045d4/pillow-11.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:3cdcdb0b896e981678eee140d882b70092dac83ac1cdf6b3a60e2216a73f2b91", size = 4296513, upload-time = "2025-01-02T08:11:43.083Z" }, + { url = "https://files.pythonhosted.org/packages/db/61/ada2a226e22da011b45f7104c95ebda1b63dcbb0c378ad0f7c2a710f8fd2/pillow-11.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:36ba10b9cb413e7c7dfa3e189aba252deee0602c86c309799da5a74009ac7a1c", size = 4431298, upload-time = "2025-01-02T08:11:46.626Z" }, + { url = "https://files.pythonhosted.org/packages/e7/c4/fc6e86750523f367923522014b821c11ebc5ad402e659d8c9d09b3c9d70c/pillow-11.1.0-cp312-cp312-win32.whl", hash = "sha256:cfd5cd998c2e36a862d0e27b2df63237e67273f2fc78f47445b14e73a810e7e6", size = 2291630, upload-time = "2025-01-02T08:11:49.401Z" }, + { url = "https://files.pythonhosted.org/packages/08/5c/2104299949b9d504baf3f4d35f73dbd14ef31bbd1ddc2c1b66a5b7dfda44/pillow-11.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:a697cd8ba0383bba3d2d3ada02b34ed268cb548b369943cd349007730c92bddf", size = 2626369, upload-time = "2025-01-02T08:11:52.02Z" }, + { url = "https://files.pythonhosted.org/packages/37/f3/9b18362206b244167c958984b57c7f70a0289bfb59a530dd8af5f699b910/pillow-11.1.0-cp312-cp312-win_arm64.whl", hash = "sha256:4dd43a78897793f60766563969442020e90eb7847463eca901e41ba186a7d4a5", size = 2375240, upload-time = "2025-01-02T08:11:56.193Z" }, + { url = "https://files.pythonhosted.org/packages/b3/31/9ca79cafdce364fd5c980cd3416c20ce1bebd235b470d262f9d24d810184/pillow-11.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:ae98e14432d458fc3de11a77ccb3ae65ddce70f730e7c76140653048c71bfcbc", size = 3226640, upload-time = "2025-01-02T08:11:58.329Z" }, + { url = "https://files.pythonhosted.org/packages/ac/0f/ff07ad45a1f172a497aa393b13a9d81a32e1477ef0e869d030e3c1532521/pillow-11.1.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:cc1331b6d5a6e144aeb5e626f4375f5b7ae9934ba620c0ac6b3e43d5e683a0f0", size = 3101437, upload-time = "2025-01-02T08:12:01.797Z" }, + { url = "https://files.pythonhosted.org/packages/08/2f/9906fca87a68d29ec4530be1f893149e0cb64a86d1f9f70a7cfcdfe8ae44/pillow-11.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:758e9d4ef15d3560214cddbc97b8ef3ef86ce04d62ddac17ad39ba87e89bd3b1", size = 4326605, upload-time = "2025-01-02T08:12:05.224Z" }, + { url = "https://files.pythonhosted.org/packages/b0/0f/f3547ee15b145bc5c8b336401b2d4c9d9da67da9dcb572d7c0d4103d2c69/pillow-11.1.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b523466b1a31d0dcef7c5be1f20b942919b62fd6e9a9be199d035509cbefc0ec", size = 4411173, upload-time = "2025-01-02T08:12:08.281Z" }, + { url = "https://files.pythonhosted.org/packages/b1/df/bf8176aa5db515c5de584c5e00df9bab0713548fd780c82a86cba2c2fedb/pillow-11.1.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:9044b5e4f7083f209c4e35aa5dd54b1dd5b112b108648f5c902ad586d4f945c5", size = 4369145, upload-time = "2025-01-02T08:12:11.411Z" }, + { url = "https://files.pythonhosted.org/packages/de/7c/7433122d1cfadc740f577cb55526fdc39129a648ac65ce64db2eb7209277/pillow-11.1.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:3764d53e09cdedd91bee65c2527815d315c6b90d7b8b79759cc48d7bf5d4f114", size = 4496340, upload-time = "2025-01-02T08:12:15.29Z" }, + { url = "https://files.pythonhosted.org/packages/25/46/dd94b93ca6bd555588835f2504bd90c00d5438fe131cf01cfa0c5131a19d/pillow-11.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:31eba6bbdd27dde97b0174ddf0297d7a9c3a507a8a1480e1e60ef914fe23d352", size = 4296906, upload-time = "2025-01-02T08:12:17.485Z" }, + { url = "https://files.pythonhosted.org/packages/a8/28/2f9d32014dfc7753e586db9add35b8a41b7a3b46540e965cb6d6bc607bd2/pillow-11.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:b5d658fbd9f0d6eea113aea286b21d3cd4d3fd978157cbf2447a6035916506d3", size = 4431759, upload-time = "2025-01-02T08:12:20.382Z" }, + { url = "https://files.pythonhosted.org/packages/33/48/19c2cbe7403870fbe8b7737d19eb013f46299cdfe4501573367f6396c775/pillow-11.1.0-cp313-cp313-win32.whl", hash = "sha256:f86d3a7a9af5d826744fabf4afd15b9dfef44fe69a98541f666f66fbb8d3fef9", size = 2291657, upload-time = "2025-01-02T08:12:23.922Z" }, + { url = "https://files.pythonhosted.org/packages/3b/ad/285c556747d34c399f332ba7c1a595ba245796ef3e22eae190f5364bb62b/pillow-11.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:593c5fd6be85da83656b93ffcccc2312d2d149d251e98588b14fbc288fd8909c", size = 2626304, upload-time = "2025-01-02T08:12:28.069Z" }, + { url = "https://files.pythonhosted.org/packages/e5/7b/ef35a71163bf36db06e9c8729608f78dedf032fc8313d19bd4be5c2588f3/pillow-11.1.0-cp313-cp313-win_arm64.whl", hash = "sha256:11633d58b6ee5733bde153a8dafd25e505ea3d32e261accd388827ee987baf65", size = 2375117, upload-time = "2025-01-02T08:12:30.064Z" }, + { url = "https://files.pythonhosted.org/packages/79/30/77f54228401e84d6791354888549b45824ab0ffde659bafa67956303a09f/pillow-11.1.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:70ca5ef3b3b1c4a0812b5c63c57c23b63e53bc38e758b37a951e5bc466449861", size = 3230060, upload-time = "2025-01-02T08:12:32.362Z" }, + { url = "https://files.pythonhosted.org/packages/ce/b1/56723b74b07dd64c1010fee011951ea9c35a43d8020acd03111f14298225/pillow-11.1.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:8000376f139d4d38d6851eb149b321a52bb8893a88dae8ee7d95840431977081", size = 3106192, upload-time = "2025-01-02T08:12:34.361Z" }, + { url = "https://files.pythonhosted.org/packages/e1/cd/7bf7180e08f80a4dcc6b4c3a0aa9e0b0ae57168562726a05dc8aa8fa66b0/pillow-11.1.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9ee85f0696a17dd28fbcfceb59f9510aa71934b483d1f5601d1030c3c8304f3c", size = 4446805, upload-time = "2025-01-02T08:12:36.99Z" }, + { url = "https://files.pythonhosted.org/packages/97/42/87c856ea30c8ed97e8efbe672b58c8304dee0573f8c7cab62ae9e31db6ae/pillow-11.1.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:dd0e081319328928531df7a0e63621caf67652c8464303fd102141b785ef9547", size = 4530623, upload-time = "2025-01-02T08:12:41.912Z" }, + { url = "https://files.pythonhosted.org/packages/ff/41/026879e90c84a88e33fb00cc6bd915ac2743c67e87a18f80270dfe3c2041/pillow-11.1.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:e63e4e5081de46517099dc30abe418122f54531a6ae2ebc8680bcd7096860eab", size = 4465191, upload-time = "2025-01-02T08:12:45.186Z" }, + { url = "https://files.pythonhosted.org/packages/e5/fb/a7960e838bc5df57a2ce23183bfd2290d97c33028b96bde332a9057834d3/pillow-11.1.0-cp313-cp313t-win32.whl", hash = "sha256:dda60aa465b861324e65a78c9f5cf0f4bc713e4309f83bc387be158b077963d9", size = 2295494, upload-time = "2025-01-02T08:12:47.098Z" }, + { url = "https://files.pythonhosted.org/packages/d7/6c/6ec83ee2f6f0fda8d4cf89045c6be4b0373ebfc363ba8538f8c999f63fcd/pillow-11.1.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ad5db5781c774ab9a9b2c4302bbf0c1014960a0a7be63278d13ae6fdf88126fe", size = 2631595, upload-time = "2025-01-02T08:12:50.47Z" }, + { url = "https://files.pythonhosted.org/packages/cf/6c/41c21c6c8af92b9fea313aa47c75de49e2f9a467964ee33eb0135d47eb64/pillow-11.1.0-cp313-cp313t-win_arm64.whl", hash = "sha256:67cd427c68926108778a9005f2a04adbd5e67c442ed21d95389fe1d595458756", size = 2377651, upload-time = "2025-01-02T08:12:53.356Z" }, + { url = "https://files.pythonhosted.org/packages/fa/c5/389961578fb677b8b3244fcd934f720ed25a148b9a5cc81c91bdf59d8588/pillow-11.1.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:8c730dc3a83e5ac137fbc92dfcfe1511ce3b2b5d7578315b63dbbb76f7f51d90", size = 3198345, upload-time = "2025-01-02T08:13:34.091Z" }, + { url = "https://files.pythonhosted.org/packages/c4/fa/803c0e50ffee74d4b965229e816af55276eac1d5806712de86f9371858fd/pillow-11.1.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:7d33d2fae0e8b170b6a6c57400e077412240f6f5bb2a342cf1ee512a787942bb", size = 3072938, upload-time = "2025-01-02T08:13:37.272Z" }, + { url = "https://files.pythonhosted.org/packages/dc/67/2a3a5f8012b5d8c63fe53958ba906c1b1d0482ebed5618057ef4d22f8076/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a8d65b38173085f24bc07f8b6c505cbb7418009fa1a1fcb111b1f4961814a442", size = 3400049, upload-time = "2025-01-02T08:13:41.565Z" }, + { url = "https://files.pythonhosted.org/packages/e5/a0/514f0d317446c98c478d1872497eb92e7cde67003fed74f696441e647446/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:015c6e863faa4779251436db398ae75051469f7c903b043a48f078e437656f83", size = 3422431, upload-time = "2025-01-02T08:13:43.609Z" }, + { url = "https://files.pythonhosted.org/packages/cd/00/20f40a935514037b7d3f87adfc87d2c538430ea625b63b3af8c3f5578e72/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:d44ff19eea13ae4acdaaab0179fa68c0c6f2f45d66a4d8ec1eda7d6cecbcc15f", size = 3446208, upload-time = "2025-01-02T08:13:46.817Z" }, + { url = "https://files.pythonhosted.org/packages/28/3c/7de681727963043e093c72e6c3348411b0185eab3263100d4490234ba2f6/pillow-11.1.0-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:d3d8da4a631471dfaf94c10c85f5277b1f8e42ac42bade1ac67da4b4a7359b73", size = 3509746, upload-time = "2025-01-02T08:13:50.6Z" }, + { url = "https://files.pythonhosted.org/packages/41/67/936f9814bdd74b2dfd4822f1f7725ab5d8ff4103919a1664eb4874c58b2f/pillow-11.1.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:4637b88343166249fe8aa94e7c4a62a180c4b3898283bb5d3d2fd5fe10d8e4e0", size = 2626353, upload-time = "2025-01-02T08:13:52.725Z" }, ] [[package]] name = "platformdirs" -version = "4.3.8" +version = "4.3.7" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/fe/8b/3c73abc9c759ecd3f1f7ceff6685840859e8070c4d947c93fae71f6a0bf2/platformdirs-4.3.8.tar.gz", hash = "sha256:3d512d96e16bcb959a814c9f348431070822a6496326a4be0911c40b5a74c2bc", size = 21362, upload-time = "2025-05-07T22:47:42.121Z" } +sdist = { url = "https://files.pythonhosted.org/packages/b6/2d/7d512a3913d60623e7eb945c6d1b4f0bddf1d0b7ada5225274c87e5b53d1/platformdirs-4.3.7.tar.gz", hash = "sha256:eb437d586b6a0986388f0d6f74aa0cde27b48d0e3d66843640bfb6bdcdb6e351", size = 21291, upload-time = "2025-03-19T20:36:10.989Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/fe/39/979e8e21520d4e47a0bbe349e2713c0aac6f3d853d0e5b34d76206c439aa/platformdirs-4.3.8-py3-none-any.whl", hash = "sha256:ff7059bb7eb1179e2685604f4aaf157cfd9535242bd23742eadc3c13542139b4", size = 18567, upload-time = "2025-05-07T22:47:40.376Z" }, + { url = "https://files.pythonhosted.org/packages/6d/45/59578566b3275b8fd9157885918fcd0c4d74162928a5310926887b856a51/platformdirs-4.3.7-py3-none-any.whl", hash = "sha256:a03875334331946f13c549dbd8f4bac7a13a50a895a0eb1e8c6a8ace80d40a94", size = 18499, upload-time = "2025-03-19T20:36:09.038Z" }, ] [[package]] name = "plotly" -version = "6.2.0" +version = "6.0.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "narwhals" }, { name = "packaging" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/6e/5c/0efc297df362b88b74957a230af61cd6929f531f72f48063e8408702ffba/plotly-6.2.0.tar.gz", hash = "sha256:9dfa23c328000f16c928beb68927444c1ab9eae837d1fe648dbcda5360c7953d", size = 6801941, upload-time = "2025-06-26T16:20:45.765Z" } +sdist = { url = "https://files.pythonhosted.org/packages/c7/cc/e41b5f697ae403f0b50e47b7af2e36642a193085f553bf7cc1169362873a/plotly-6.0.1.tar.gz", hash = "sha256:dd8400229872b6e3c964b099be699f8d00c489a974f2cfccfad5e8240873366b", size = 8094643, upload-time = "2025-03-17T15:02:23.994Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/ed/20/f2b7ac96a91cc5f70d81320adad24cc41bf52013508d649b1481db225780/plotly-6.2.0-py3-none-any.whl", hash = "sha256:32c444d4c940887219cb80738317040363deefdfee4f354498cc0b6dab8978bd", size = 9635469, upload-time = "2025-06-26T16:20:40.76Z" }, + { url = "https://files.pythonhosted.org/packages/02/65/ad2bc85f7377f5cfba5d4466d5474423a3fb7f6a97fd807c06f92dd3e721/plotly-6.0.1-py3-none-any.whl", hash = "sha256:4714db20fea57a435692c548a4eb4fae454f7daddf15f8d8ba7e1045681d7768", size = 14805757, upload-time = "2025-03-17T15:02:18.73Z" }, ] [[package]] name = "pluggy" -version = "1.6.0" +version = "1.5.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/96/2d/02d4312c973c6050a18b314a5ad0b3210edb65a906f868e31c111dede4a6/pluggy-1.5.0.tar.gz", hash = "sha256:2cffa88e94fdc978c4c574f15f9e59b7f4201d439195c3715ca9e2486f1d0cf1", size = 67955, upload-time = "2024-04-20T21:34:42.531Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/88/5f/e351af9a41f866ac3f1fac4ca0613908d9a41741cfcf2228f4ad853b697d/pluggy-1.5.0-py3-none-any.whl", hash = "sha256:44e1ad92c8ca002de6377e165f3e0f1be63266ab4d554740532335b9d75ea669", size = 20556, upload-time = "2024-04-20T21:34:40.434Z" }, +] + +[[package]] +name = "polars" +version = "1.32.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/f9/e2/3e91f31a7d2b083fe6ef3fa267035b518369d9511ffab804f839851d2779/pluggy-1.6.0.tar.gz", hash = "sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3", size = 69412, upload-time = "2025-05-15T12:30:07.975Z" } +sdist = { url = "https://files.pythonhosted.org/packages/67/23/6a5f151981f3ac409bed6dc48a3eaecd0592a03eb382693d4c7e749eda8b/polars-1.32.0.tar.gz", hash = "sha256:b01045981c0f23eeccfbfc870b782f93e73b74b29212fdfc8aae0be9024bc1fb", size = 4761045, upload-time = "2025-08-01T01:43:22.413Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/54/20/4d324d65cc6d9205fabedc306948156824eb9f0ee1633355a8f7ec5c66bf/pluggy-1.6.0-py3-none-any.whl", hash = "sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746", size = 20538, upload-time = "2025-05-15T12:30:06.134Z" }, + { url = "https://files.pythonhosted.org/packages/e2/40/5b27067d10b5a77ab4094932118e16629ffb20ea9ae5f7d1178e04087891/polars-1.32.0-cp39-abi3-macosx_10_12_x86_64.whl", hash = "sha256:94f7c6a3b30bc99bc6b682ea42bb1ae983e33a302ca21aacbac50ae19e34fcf2", size = 37479518, upload-time = "2025-08-01T01:42:18.603Z" }, + { url = "https://files.pythonhosted.org/packages/08/b7/ca28ac10d340fb91bffb2751efd52aebc9799ae161b867214c6299c8f75b/polars-1.32.0-cp39-abi3-macosx_11_0_arm64.whl", hash = "sha256:8bf14c16164839e62c741a863942a94a9a463db21e797452fca996c8afaf8827", size = 34214196, upload-time = "2025-08-01T01:42:22.667Z" }, + { url = "https://files.pythonhosted.org/packages/61/97/fe3797e8e1d4f9eadab32ffe218a841b8874585b6c9bd0f1a26469fb2992/polars-1.32.0-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f4c15adb97d44766d30c759f5cebbdb64d361e8349ef10b5afc7413f71bf4b72", size = 37985353, upload-time = "2025-08-01T01:42:26.033Z" }, + { url = "https://files.pythonhosted.org/packages/a0/7e/2baa2858556e970cc6a35c0d8ad34b2f9d982f1766c0a1fec20ca529a947/polars-1.32.0-cp39-abi3-manylinux_2_24_aarch64.whl", hash = "sha256:13af55890734f89b76016a395fb2e7460e7d9feecf50ed2f55cf0f05a1c0c991", size = 35183912, upload-time = "2025-08-01T01:42:30.446Z" }, + { url = "https://files.pythonhosted.org/packages/ef/41/0e6821dccc5871186a9b95af3990404aa283318263918d33ac974b35cb37/polars-1.32.0-cp39-abi3-win_amd64.whl", hash = "sha256:0397fc2501a5d5f1bb3fe8d27e0c26c7a5349b4110157c0fb7833cd3f5921c9e", size = 37747905, upload-time = "2025-08-01T01:42:33.975Z" }, + { url = "https://files.pythonhosted.org/packages/c2/93/d06df0817da93f922a67e27e9e0f407856991374daa62687e2a45a18935c/polars-1.32.0-cp39-abi3-win_arm64.whl", hash = "sha256:dd84e24422509e1ec9be46f67f758d0bd9944d1ae4eacecee4f53adaa8ecd822", size = 33978543, upload-time = "2025-08-01T01:42:36.779Z" }, ] [[package]] name = "posebusters" -version = "0.4.5" +version = "0.3.6" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "pandas" }, { name = "pyyaml" }, { name = "rdkit" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/f9/e1/7646585f7620c7689c8542eceb93a09567469b5230cfdf1231d38a948eff/posebusters-0.4.5.tar.gz", hash = "sha256:d90f22d32f7ce3551d28d23e899059070211c91d7104ae4fbc7229b4696b9146", size = 5022845, upload-time = "2025-07-13T15:16:21.846Z" } +sdist = { url = "https://files.pythonhosted.org/packages/a4/40/44c4e544798d92afc30d43af869dc0eca555768793f85bf492e632a8df39/posebusters-0.3.6.tar.gz", hash = "sha256:d60cbc0134431a31dfc11d81dacead8b731bcbfe2be10993878649d34462f1a2", size = 4991179, upload-time = "2025-03-22T00:24:42.385Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/b6/03/6989d7d9d65aa19bcc91ea3f78d791141d4be0957a67f90fe82e6495832a/posebusters-0.4.5-py3-none-any.whl", hash = "sha256:ec8f2b1634b95000efe6c9da0392eefdc681dda07646ef0ed1f4a5efca099ab9", size = 556376, upload-time = "2025-07-13T15:16:20.308Z" }, + { url = "https://files.pythonhosted.org/packages/1f/7b/6f549f8aca269c2aafcd4a0ba453a84df027b81d3f860addf67a32ddb5f1/posebusters-0.3.6-py3-none-any.whl", hash = "sha256:4d33780765abd2ac3c27edcf2963a6322530343034ca18898c64b1a3b45c8e21", size = 554304, upload-time = "2025-03-22T00:24:40.989Z" }, ] [[package]] @@ -2665,10 +2428,8 @@ name = "pot" version = "0.9.5" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, + { name = "scipy" }, ] sdist = { url = "https://files.pythonhosted.org/packages/c1/40/3e0c8dd88328d944f9d82b30cafd2a1c911bddff0b8bccc8dc9dd5e45b7c/pot-0.9.5.tar.gz", hash = "sha256:9644ee7ff51c3cffa3c2632b9dd9dff4f3520266f9fb771450935ffb646d6042", size = 440808, upload-time = "2024-11-07T10:05:05.567Z" } wheels = [ @@ -2713,103 +2474,103 @@ wheels = [ [[package]] name = "prompt-toolkit" -version = "3.0.51" +version = "3.0.50" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "wcwidth" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/bb/6e/9d084c929dfe9e3bfe0c6a47e31f78a25c54627d64a66e884a8bf5474f1c/prompt_toolkit-3.0.51.tar.gz", hash = "sha256:931a162e3b27fc90c86f1b48bb1fb2c528c2761475e57c9c06de13311c7b54ed", size = 428940, upload-time = "2025-04-15T09:18:47.731Z" } +sdist = { url = "https://files.pythonhosted.org/packages/a1/e1/bd15cb8ffdcfeeb2bdc215de3c3cffca11408d829e4b8416dcfe71ba8854/prompt_toolkit-3.0.50.tar.gz", hash = "sha256:544748f3860a2623ca5cd6d2795e7a14f3d0e1c3c9728359013f79877fc89bab", size = 429087, upload-time = "2025-01-20T15:55:35.072Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/ce/4f/5249960887b1fbe561d9ff265496d170b55a735b76724f10ef19f9e40716/prompt_toolkit-3.0.51-py3-none-any.whl", hash = "sha256:52742911fde84e2d423e2f9a4cf1de7d7ac4e51958f648d9540e0fb8db077b07", size = 387810, upload-time = "2025-04-15T09:18:44.753Z" }, + { url = "https://files.pythonhosted.org/packages/e4/ea/d836f008d33151c7a1f62caf3d8dd782e4d15f6a43897f64480c2b8de2ad/prompt_toolkit-3.0.50-py3-none-any.whl", hash = "sha256:9b6427eb19e479d98acff65196a307c555eb567989e6d88ebbb1b509d9779198", size = 387816, upload-time = "2025-01-20T15:55:29.98Z" }, ] [[package]] name = "propcache" -version = "0.3.2" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/a6/16/43264e4a779dd8588c21a70f0709665ee8f611211bdd2c87d952cfa7c776/propcache-0.3.2.tar.gz", hash = "sha256:20d7d62e4e7ef05f221e0db2856b979540686342e7dd9973b815599c7057e168", size = 44139, upload-time = "2025-06-09T22:56:06.081Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/ab/14/510deed325e262afeb8b360043c5d7c960da7d3ecd6d6f9496c9c56dc7f4/propcache-0.3.2-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:22d9962a358aedbb7a2e36187ff273adeaab9743373a272976d2e348d08c7770", size = 73178, upload-time = "2025-06-09T22:53:40.126Z" }, - { url = "https://files.pythonhosted.org/packages/cd/4e/ad52a7925ff01c1325653a730c7ec3175a23f948f08626a534133427dcff/propcache-0.3.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:0d0fda578d1dc3f77b6b5a5dce3b9ad69a8250a891760a548df850a5e8da87f3", size = 43133, upload-time = "2025-06-09T22:53:41.965Z" }, - { url = "https://files.pythonhosted.org/packages/63/7c/e9399ba5da7780871db4eac178e9c2e204c23dd3e7d32df202092a1ed400/propcache-0.3.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3def3da3ac3ce41562d85db655d18ebac740cb3fa4367f11a52b3da9d03a5cc3", size = 43039, upload-time = "2025-06-09T22:53:43.268Z" }, - { url = "https://files.pythonhosted.org/packages/22/e1/58da211eb8fdc6fc854002387d38f415a6ca5f5c67c1315b204a5d3e9d7a/propcache-0.3.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9bec58347a5a6cebf239daba9bda37dffec5b8d2ce004d9fe4edef3d2815137e", size = 201903, upload-time = "2025-06-09T22:53:44.872Z" }, - { url = "https://files.pythonhosted.org/packages/c4/0a/550ea0f52aac455cb90111c8bab995208443e46d925e51e2f6ebdf869525/propcache-0.3.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:55ffda449a507e9fbd4aca1a7d9aa6753b07d6166140e5a18d2ac9bc49eac220", size = 213362, upload-time = "2025-06-09T22:53:46.707Z" }, - { url = "https://files.pythonhosted.org/packages/5a/af/9893b7d878deda9bb69fcf54600b247fba7317761b7db11fede6e0f28bd0/propcache-0.3.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:64a67fb39229a8a8491dd42f864e5e263155e729c2e7ff723d6e25f596b1e8cb", size = 210525, upload-time = "2025-06-09T22:53:48.547Z" }, - { url = "https://files.pythonhosted.org/packages/7c/bb/38fd08b278ca85cde36d848091ad2b45954bc5f15cce494bb300b9285831/propcache-0.3.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9da1cf97b92b51253d5b68cf5a2b9e0dafca095e36b7f2da335e27dc6172a614", size = 198283, upload-time = "2025-06-09T22:53:50.067Z" }, - { url = "https://files.pythonhosted.org/packages/78/8c/9fe55bd01d362bafb413dfe508c48753111a1e269737fa143ba85693592c/propcache-0.3.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:5f559e127134b07425134b4065be45b166183fdcb433cb6c24c8e4149056ad50", size = 191872, upload-time = "2025-06-09T22:53:51.438Z" }, - { url = "https://files.pythonhosted.org/packages/54/14/4701c33852937a22584e08abb531d654c8bcf7948a8f87ad0a4822394147/propcache-0.3.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:aff2e4e06435d61f11a428360a932138d0ec288b0a31dd9bd78d200bd4a2b339", size = 199452, upload-time = "2025-06-09T22:53:53.229Z" }, - { url = "https://files.pythonhosted.org/packages/16/44/447f2253d859602095356007657ee535e0093215ea0b3d1d6a41d16e5201/propcache-0.3.2-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:4927842833830942a5d0a56e6f4839bc484785b8e1ce8d287359794818633ba0", size = 191567, upload-time = "2025-06-09T22:53:54.541Z" }, - { url = "https://files.pythonhosted.org/packages/f2/b3/e4756258749bb2d3b46defcff606a2f47410bab82be5824a67e84015b267/propcache-0.3.2-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:6107ddd08b02654a30fb8ad7a132021759d750a82578b94cd55ee2772b6ebea2", size = 193015, upload-time = "2025-06-09T22:53:56.44Z" }, - { url = "https://files.pythonhosted.org/packages/1e/df/e6d3c7574233164b6330b9fd697beeac402afd367280e6dc377bb99b43d9/propcache-0.3.2-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:70bd8b9cd6b519e12859c99f3fc9a93f375ebd22a50296c3a295028bea73b9e7", size = 204660, upload-time = "2025-06-09T22:53:57.839Z" }, - { url = "https://files.pythonhosted.org/packages/b2/53/e4d31dd5170b4a0e2e6b730f2385a96410633b4833dc25fe5dffd1f73294/propcache-0.3.2-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:2183111651d710d3097338dd1893fcf09c9f54e27ff1a8795495a16a469cc90b", size = 206105, upload-time = "2025-06-09T22:53:59.638Z" }, - { url = "https://files.pythonhosted.org/packages/7f/fe/74d54cf9fbe2a20ff786e5f7afcfde446588f0cf15fb2daacfbc267b866c/propcache-0.3.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:fb075ad271405dcad8e2a7ffc9a750a3bf70e533bd86e89f0603e607b93aa64c", size = 196980, upload-time = "2025-06-09T22:54:01.071Z" }, - { url = "https://files.pythonhosted.org/packages/22/ec/c469c9d59dada8a7679625e0440b544fe72e99311a4679c279562051f6fc/propcache-0.3.2-cp310-cp310-win32.whl", hash = "sha256:404d70768080d3d3bdb41d0771037da19d8340d50b08e104ca0e7f9ce55fce70", size = 37679, upload-time = "2025-06-09T22:54:03.003Z" }, - { url = "https://files.pythonhosted.org/packages/38/35/07a471371ac89d418f8d0b699c75ea6dca2041fbda360823de21f6a9ce0a/propcache-0.3.2-cp310-cp310-win_amd64.whl", hash = "sha256:7435d766f978b4ede777002e6b3b6641dd229cd1da8d3d3106a45770365f9ad9", size = 41459, upload-time = "2025-06-09T22:54:04.134Z" }, - { url = "https://files.pythonhosted.org/packages/80/8d/e8b436717ab9c2cfc23b116d2c297305aa4cd8339172a456d61ebf5669b8/propcache-0.3.2-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:0b8d2f607bd8f80ddc04088bc2a037fdd17884a6fcadc47a96e334d72f3717be", size = 74207, upload-time = "2025-06-09T22:54:05.399Z" }, - { url = "https://files.pythonhosted.org/packages/d6/29/1e34000e9766d112171764b9fa3226fa0153ab565d0c242c70e9945318a7/propcache-0.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:06766d8f34733416e2e34f46fea488ad5d60726bb9481d3cddf89a6fa2d9603f", size = 43648, upload-time = "2025-06-09T22:54:08.023Z" }, - { url = "https://files.pythonhosted.org/packages/46/92/1ad5af0df781e76988897da39b5f086c2bf0f028b7f9bd1f409bb05b6874/propcache-0.3.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a2dc1f4a1df4fecf4e6f68013575ff4af84ef6f478fe5344317a65d38a8e6dc9", size = 43496, upload-time = "2025-06-09T22:54:09.228Z" }, - { url = "https://files.pythonhosted.org/packages/b3/ce/e96392460f9fb68461fabab3e095cb00c8ddf901205be4eae5ce246e5b7e/propcache-0.3.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:be29c4f4810c5789cf10ddf6af80b041c724e629fa51e308a7a0fb19ed1ef7bf", size = 217288, upload-time = "2025-06-09T22:54:10.466Z" }, - { url = "https://files.pythonhosted.org/packages/c5/2a/866726ea345299f7ceefc861a5e782b045545ae6940851930a6adaf1fca6/propcache-0.3.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:59d61f6970ecbd8ff2e9360304d5c8876a6abd4530cb752c06586849ac8a9dc9", size = 227456, upload-time = "2025-06-09T22:54:11.828Z" }, - { url = "https://files.pythonhosted.org/packages/de/03/07d992ccb6d930398689187e1b3c718339a1c06b8b145a8d9650e4726166/propcache-0.3.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:62180e0b8dbb6b004baec00a7983e4cc52f5ada9cd11f48c3528d8cfa7b96a66", size = 225429, upload-time = "2025-06-09T22:54:13.823Z" }, - { url = "https://files.pythonhosted.org/packages/5d/e6/116ba39448753b1330f48ab8ba927dcd6cf0baea8a0ccbc512dfb49ba670/propcache-0.3.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c144ca294a204c470f18cf4c9d78887810d04a3e2fbb30eea903575a779159df", size = 213472, upload-time = "2025-06-09T22:54:15.232Z" }, - { url = "https://files.pythonhosted.org/packages/a6/85/f01f5d97e54e428885a5497ccf7f54404cbb4f906688a1690cd51bf597dc/propcache-0.3.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c5c2a784234c28854878d68978265617aa6dc0780e53d44b4d67f3651a17a9a2", size = 204480, upload-time = "2025-06-09T22:54:17.104Z" }, - { url = "https://files.pythonhosted.org/packages/e3/79/7bf5ab9033b8b8194cc3f7cf1aaa0e9c3256320726f64a3e1f113a812dce/propcache-0.3.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:5745bc7acdafa978ca1642891b82c19238eadc78ba2aaa293c6863b304e552d7", size = 214530, upload-time = "2025-06-09T22:54:18.512Z" }, - { url = "https://files.pythonhosted.org/packages/31/0b/bd3e0c00509b609317df4a18e6b05a450ef2d9a963e1d8bc9c9415d86f30/propcache-0.3.2-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:c0075bf773d66fa8c9d41f66cc132ecc75e5bb9dd7cce3cfd14adc5ca184cb95", size = 205230, upload-time = "2025-06-09T22:54:19.947Z" }, - { url = "https://files.pythonhosted.org/packages/7a/23/fae0ff9b54b0de4e819bbe559508da132d5683c32d84d0dc2ccce3563ed4/propcache-0.3.2-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:5f57aa0847730daceff0497f417c9de353c575d8da3579162cc74ac294c5369e", size = 206754, upload-time = "2025-06-09T22:54:21.716Z" }, - { url = "https://files.pythonhosted.org/packages/b7/7f/ad6a3c22630aaa5f618b4dc3c3598974a72abb4c18e45a50b3cdd091eb2f/propcache-0.3.2-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:eef914c014bf72d18efb55619447e0aecd5fb7c2e3fa7441e2e5d6099bddff7e", size = 218430, upload-time = "2025-06-09T22:54:23.17Z" }, - { url = "https://files.pythonhosted.org/packages/5b/2c/ba4f1c0e8a4b4c75910742f0d333759d441f65a1c7f34683b4a74c0ee015/propcache-0.3.2-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:2a4092e8549031e82facf3decdbc0883755d5bbcc62d3aea9d9e185549936dcf", size = 223884, upload-time = "2025-06-09T22:54:25.539Z" }, - { url = "https://files.pythonhosted.org/packages/88/e4/ebe30fc399e98572019eee82ad0caf512401661985cbd3da5e3140ffa1b0/propcache-0.3.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:85871b050f174bc0bfb437efbdb68aaf860611953ed12418e4361bc9c392749e", size = 211480, upload-time = "2025-06-09T22:54:26.892Z" }, - { url = "https://files.pythonhosted.org/packages/96/0a/7d5260b914e01d1d0906f7f38af101f8d8ed0dc47426219eeaf05e8ea7c2/propcache-0.3.2-cp311-cp311-win32.whl", hash = "sha256:36c8d9b673ec57900c3554264e630d45980fd302458e4ac801802a7fd2ef7897", size = 37757, upload-time = "2025-06-09T22:54:28.241Z" }, - { url = "https://files.pythonhosted.org/packages/e1/2d/89fe4489a884bc0da0c3278c552bd4ffe06a1ace559db5ef02ef24ab446b/propcache-0.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:e53af8cb6a781b02d2ea079b5b853ba9430fcbe18a8e3ce647d5982a3ff69f39", size = 41500, upload-time = "2025-06-09T22:54:29.4Z" }, - { url = "https://files.pythonhosted.org/packages/a8/42/9ca01b0a6f48e81615dca4765a8f1dd2c057e0540f6116a27dc5ee01dfb6/propcache-0.3.2-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:8de106b6c84506b31c27168582cd3cb3000a6412c16df14a8628e5871ff83c10", size = 73674, upload-time = "2025-06-09T22:54:30.551Z" }, - { url = "https://files.pythonhosted.org/packages/af/6e/21293133beb550f9c901bbece755d582bfaf2176bee4774000bd4dd41884/propcache-0.3.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:28710b0d3975117239c76600ea351934ac7b5ff56e60953474342608dbbb6154", size = 43570, upload-time = "2025-06-09T22:54:32.296Z" }, - { url = "https://files.pythonhosted.org/packages/0c/c8/0393a0a3a2b8760eb3bde3c147f62b20044f0ddac81e9d6ed7318ec0d852/propcache-0.3.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ce26862344bdf836650ed2487c3d724b00fbfec4233a1013f597b78c1cb73615", size = 43094, upload-time = "2025-06-09T22:54:33.929Z" }, - { url = "https://files.pythonhosted.org/packages/37/2c/489afe311a690399d04a3e03b069225670c1d489eb7b044a566511c1c498/propcache-0.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bca54bd347a253af2cf4544bbec232ab982f4868de0dd684246b67a51bc6b1db", size = 226958, upload-time = "2025-06-09T22:54:35.186Z" }, - { url = "https://files.pythonhosted.org/packages/9d/ca/63b520d2f3d418c968bf596839ae26cf7f87bead026b6192d4da6a08c467/propcache-0.3.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:55780d5e9a2ddc59711d727226bb1ba83a22dd32f64ee15594b9392b1f544eb1", size = 234894, upload-time = "2025-06-09T22:54:36.708Z" }, - { url = "https://files.pythonhosted.org/packages/11/60/1d0ed6fff455a028d678df30cc28dcee7af77fa2b0e6962ce1df95c9a2a9/propcache-0.3.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:035e631be25d6975ed87ab23153db6a73426a48db688070d925aa27e996fe93c", size = 233672, upload-time = "2025-06-09T22:54:38.062Z" }, - { url = "https://files.pythonhosted.org/packages/37/7c/54fd5301ef38505ab235d98827207176a5c9b2aa61939b10a460ca53e123/propcache-0.3.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ee6f22b6eaa39297c751d0e80c0d3a454f112f5c6481214fcf4c092074cecd67", size = 224395, upload-time = "2025-06-09T22:54:39.634Z" }, - { url = "https://files.pythonhosted.org/packages/ee/1a/89a40e0846f5de05fdc6779883bf46ba980e6df4d2ff8fb02643de126592/propcache-0.3.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:7ca3aee1aa955438c4dba34fc20a9f390e4c79967257d830f137bd5a8a32ed3b", size = 212510, upload-time = "2025-06-09T22:54:41.565Z" }, - { url = "https://files.pythonhosted.org/packages/5e/33/ca98368586c9566a6b8d5ef66e30484f8da84c0aac3f2d9aec6d31a11bd5/propcache-0.3.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7a4f30862869fa2b68380d677cc1c5fcf1e0f2b9ea0cf665812895c75d0ca3b8", size = 222949, upload-time = "2025-06-09T22:54:43.038Z" }, - { url = "https://files.pythonhosted.org/packages/ba/11/ace870d0aafe443b33b2f0b7efdb872b7c3abd505bfb4890716ad7865e9d/propcache-0.3.2-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:b77ec3c257d7816d9f3700013639db7491a434644c906a2578a11daf13176251", size = 217258, upload-time = "2025-06-09T22:54:44.376Z" }, - { url = "https://files.pythonhosted.org/packages/5b/d2/86fd6f7adffcfc74b42c10a6b7db721d1d9ca1055c45d39a1a8f2a740a21/propcache-0.3.2-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:cab90ac9d3f14b2d5050928483d3d3b8fb6b4018893fc75710e6aa361ecb2474", size = 213036, upload-time = "2025-06-09T22:54:46.243Z" }, - { url = "https://files.pythonhosted.org/packages/07/94/2d7d1e328f45ff34a0a284cf5a2847013701e24c2a53117e7c280a4316b3/propcache-0.3.2-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:0b504d29f3c47cf6b9e936c1852246c83d450e8e063d50562115a6be6d3a2535", size = 227684, upload-time = "2025-06-09T22:54:47.63Z" }, - { url = "https://files.pythonhosted.org/packages/b7/05/37ae63a0087677e90b1d14710e532ff104d44bc1efa3b3970fff99b891dc/propcache-0.3.2-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:ce2ac2675a6aa41ddb2a0c9cbff53780a617ac3d43e620f8fd77ba1c84dcfc06", size = 234562, upload-time = "2025-06-09T22:54:48.982Z" }, - { url = "https://files.pythonhosted.org/packages/a4/7c/3f539fcae630408d0bd8bf3208b9a647ccad10976eda62402a80adf8fc34/propcache-0.3.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:62b4239611205294cc433845b914131b2a1f03500ff3c1ed093ed216b82621e1", size = 222142, upload-time = "2025-06-09T22:54:50.424Z" }, - { url = "https://files.pythonhosted.org/packages/7c/d2/34b9eac8c35f79f8a962546b3e97e9d4b990c420ee66ac8255d5d9611648/propcache-0.3.2-cp312-cp312-win32.whl", hash = "sha256:df4a81b9b53449ebc90cc4deefb052c1dd934ba85012aa912c7ea7b7e38b60c1", size = 37711, upload-time = "2025-06-09T22:54:52.072Z" }, - { url = "https://files.pythonhosted.org/packages/19/61/d582be5d226cf79071681d1b46b848d6cb03d7b70af7063e33a2787eaa03/propcache-0.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:7046e79b989d7fe457bb755844019e10f693752d169076138abf17f31380800c", size = 41479, upload-time = "2025-06-09T22:54:53.234Z" }, - { url = "https://files.pythonhosted.org/packages/dc/d1/8c747fafa558c603c4ca19d8e20b288aa0c7cda74e9402f50f31eb65267e/propcache-0.3.2-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:ca592ed634a73ca002967458187109265e980422116c0a107cf93d81f95af945", size = 71286, upload-time = "2025-06-09T22:54:54.369Z" }, - { url = "https://files.pythonhosted.org/packages/61/99/d606cb7986b60d89c36de8a85d58764323b3a5ff07770a99d8e993b3fa73/propcache-0.3.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9ecb0aad4020e275652ba3975740f241bd12a61f1a784df044cf7477a02bc252", size = 42425, upload-time = "2025-06-09T22:54:55.642Z" }, - { url = "https://files.pythonhosted.org/packages/8c/96/ef98f91bbb42b79e9bb82bdd348b255eb9d65f14dbbe3b1594644c4073f7/propcache-0.3.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:7f08f1cc28bd2eade7a8a3d2954ccc673bb02062e3e7da09bc75d843386b342f", size = 41846, upload-time = "2025-06-09T22:54:57.246Z" }, - { url = "https://files.pythonhosted.org/packages/5b/ad/3f0f9a705fb630d175146cd7b1d2bf5555c9beaed54e94132b21aac098a6/propcache-0.3.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d1a342c834734edb4be5ecb1e9fb48cb64b1e2320fccbd8c54bf8da8f2a84c33", size = 208871, upload-time = "2025-06-09T22:54:58.975Z" }, - { url = "https://files.pythonhosted.org/packages/3a/38/2085cda93d2c8b6ec3e92af2c89489a36a5886b712a34ab25de9fbca7992/propcache-0.3.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8a544caaae1ac73f1fecfae70ded3e93728831affebd017d53449e3ac052ac1e", size = 215720, upload-time = "2025-06-09T22:55:00.471Z" }, - { url = "https://files.pythonhosted.org/packages/61/c1/d72ea2dc83ac7f2c8e182786ab0fc2c7bd123a1ff9b7975bee671866fe5f/propcache-0.3.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:310d11aa44635298397db47a3ebce7db99a4cc4b9bbdfcf6c98a60c8d5261cf1", size = 215203, upload-time = "2025-06-09T22:55:01.834Z" }, - { url = "https://files.pythonhosted.org/packages/af/81/b324c44ae60c56ef12007105f1460d5c304b0626ab0cc6b07c8f2a9aa0b8/propcache-0.3.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4c1396592321ac83157ac03a2023aa6cc4a3cc3cfdecb71090054c09e5a7cce3", size = 206365, upload-time = "2025-06-09T22:55:03.199Z" }, - { url = "https://files.pythonhosted.org/packages/09/73/88549128bb89e66d2aff242488f62869014ae092db63ccea53c1cc75a81d/propcache-0.3.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8cabf5b5902272565e78197edb682017d21cf3b550ba0460ee473753f28d23c1", size = 196016, upload-time = "2025-06-09T22:55:04.518Z" }, - { url = "https://files.pythonhosted.org/packages/b9/3f/3bdd14e737d145114a5eb83cb172903afba7242f67c5877f9909a20d948d/propcache-0.3.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:0a2f2235ac46a7aa25bdeb03a9e7060f6ecbd213b1f9101c43b3090ffb971ef6", size = 205596, upload-time = "2025-06-09T22:55:05.942Z" }, - { url = "https://files.pythonhosted.org/packages/0f/ca/2f4aa819c357d3107c3763d7ef42c03980f9ed5c48c82e01e25945d437c1/propcache-0.3.2-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:92b69e12e34869a6970fd2f3da91669899994b47c98f5d430b781c26f1d9f387", size = 200977, upload-time = "2025-06-09T22:55:07.792Z" }, - { url = "https://files.pythonhosted.org/packages/cd/4a/e65276c7477533c59085251ae88505caf6831c0e85ff8b2e31ebcbb949b1/propcache-0.3.2-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:54e02207c79968ebbdffc169591009f4474dde3b4679e16634d34c9363ff56b4", size = 197220, upload-time = "2025-06-09T22:55:09.173Z" }, - { url = "https://files.pythonhosted.org/packages/7c/54/fc7152e517cf5578278b242396ce4d4b36795423988ef39bb8cd5bf274c8/propcache-0.3.2-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:4adfb44cb588001f68c5466579d3f1157ca07f7504fc91ec87862e2b8e556b88", size = 210642, upload-time = "2025-06-09T22:55:10.62Z" }, - { url = "https://files.pythonhosted.org/packages/b9/80/abeb4a896d2767bf5f1ea7b92eb7be6a5330645bd7fb844049c0e4045d9d/propcache-0.3.2-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:fd3e6019dc1261cd0291ee8919dd91fbab7b169bb76aeef6c716833a3f65d206", size = 212789, upload-time = "2025-06-09T22:55:12.029Z" }, - { url = "https://files.pythonhosted.org/packages/b3/db/ea12a49aa7b2b6d68a5da8293dcf50068d48d088100ac016ad92a6a780e6/propcache-0.3.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4c181cad81158d71c41a2bce88edce078458e2dd5ffee7eddd6b05da85079f43", size = 205880, upload-time = "2025-06-09T22:55:13.45Z" }, - { url = "https://files.pythonhosted.org/packages/d1/e5/9076a0bbbfb65d1198007059c65639dfd56266cf8e477a9707e4b1999ff4/propcache-0.3.2-cp313-cp313-win32.whl", hash = "sha256:8a08154613f2249519e549de2330cf8e2071c2887309a7b07fb56098f5170a02", size = 37220, upload-time = "2025-06-09T22:55:15.284Z" }, - { url = "https://files.pythonhosted.org/packages/d3/f5/b369e026b09a26cd77aa88d8fffd69141d2ae00a2abaaf5380d2603f4b7f/propcache-0.3.2-cp313-cp313-win_amd64.whl", hash = "sha256:e41671f1594fc4ab0a6dec1351864713cb3a279910ae8b58f884a88a0a632c05", size = 40678, upload-time = "2025-06-09T22:55:16.445Z" }, - { url = "https://files.pythonhosted.org/packages/a4/3a/6ece377b55544941a08d03581c7bc400a3c8cd3c2865900a68d5de79e21f/propcache-0.3.2-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:9a3cf035bbaf035f109987d9d55dc90e4b0e36e04bbbb95af3055ef17194057b", size = 76560, upload-time = "2025-06-09T22:55:17.598Z" }, - { url = "https://files.pythonhosted.org/packages/0c/da/64a2bb16418740fa634b0e9c3d29edff1db07f56d3546ca2d86ddf0305e1/propcache-0.3.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:156c03d07dc1323d8dacaa221fbe028c5c70d16709cdd63502778e6c3ccca1b0", size = 44676, upload-time = "2025-06-09T22:55:18.922Z" }, - { url = "https://files.pythonhosted.org/packages/36/7b/f025e06ea51cb72c52fb87e9b395cced02786610b60a3ed51da8af017170/propcache-0.3.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:74413c0ba02ba86f55cf60d18daab219f7e531620c15f1e23d95563f505efe7e", size = 44701, upload-time = "2025-06-09T22:55:20.106Z" }, - { url = "https://files.pythonhosted.org/packages/a4/00/faa1b1b7c3b74fc277f8642f32a4c72ba1d7b2de36d7cdfb676db7f4303e/propcache-0.3.2-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f066b437bb3fa39c58ff97ab2ca351db465157d68ed0440abecb21715eb24b28", size = 276934, upload-time = "2025-06-09T22:55:21.5Z" }, - { url = "https://files.pythonhosted.org/packages/74/ab/935beb6f1756e0476a4d5938ff44bf0d13a055fed880caf93859b4f1baf4/propcache-0.3.2-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f1304b085c83067914721e7e9d9917d41ad87696bf70f0bc7dee450e9c71ad0a", size = 278316, upload-time = "2025-06-09T22:55:22.918Z" }, - { url = "https://files.pythonhosted.org/packages/f8/9d/994a5c1ce4389610838d1caec74bdf0e98b306c70314d46dbe4fcf21a3e2/propcache-0.3.2-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ab50cef01b372763a13333b4e54021bdcb291fc9a8e2ccb9c2df98be51bcde6c", size = 282619, upload-time = "2025-06-09T22:55:24.651Z" }, - { url = "https://files.pythonhosted.org/packages/2b/00/a10afce3d1ed0287cef2e09506d3be9822513f2c1e96457ee369adb9a6cd/propcache-0.3.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fad3b2a085ec259ad2c2842666b2a0a49dea8463579c606426128925af1ed725", size = 265896, upload-time = "2025-06-09T22:55:26.049Z" }, - { url = "https://files.pythonhosted.org/packages/2e/a8/2aa6716ffa566ca57c749edb909ad27884680887d68517e4be41b02299f3/propcache-0.3.2-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:261fa020c1c14deafd54c76b014956e2f86991af198c51139faf41c4d5e83892", size = 252111, upload-time = "2025-06-09T22:55:27.381Z" }, - { url = "https://files.pythonhosted.org/packages/36/4f/345ca9183b85ac29c8694b0941f7484bf419c7f0fea2d1e386b4f7893eed/propcache-0.3.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:46d7f8aa79c927e5f987ee3a80205c987717d3659f035c85cf0c3680526bdb44", size = 268334, upload-time = "2025-06-09T22:55:28.747Z" }, - { url = "https://files.pythonhosted.org/packages/3e/ca/fcd54f78b59e3f97b3b9715501e3147f5340167733d27db423aa321e7148/propcache-0.3.2-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:6d8f3f0eebf73e3c0ff0e7853f68be638b4043c65a70517bb575eff54edd8dbe", size = 255026, upload-time = "2025-06-09T22:55:30.184Z" }, - { url = "https://files.pythonhosted.org/packages/8b/95/8e6a6bbbd78ac89c30c225210a5c687790e532ba4088afb8c0445b77ef37/propcache-0.3.2-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:03c89c1b14a5452cf15403e291c0ccd7751d5b9736ecb2c5bab977ad6c5bcd81", size = 250724, upload-time = "2025-06-09T22:55:31.646Z" }, - { url = "https://files.pythonhosted.org/packages/ee/b0/0dd03616142baba28e8b2d14ce5df6631b4673850a3d4f9c0f9dd714a404/propcache-0.3.2-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:0cc17efde71e12bbaad086d679ce575268d70bc123a5a71ea7ad76f70ba30bba", size = 268868, upload-time = "2025-06-09T22:55:33.209Z" }, - { url = "https://files.pythonhosted.org/packages/c5/98/2c12407a7e4fbacd94ddd32f3b1e3d5231e77c30ef7162b12a60e2dd5ce3/propcache-0.3.2-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:acdf05d00696bc0447e278bb53cb04ca72354e562cf88ea6f9107df8e7fd9770", size = 271322, upload-time = "2025-06-09T22:55:35.065Z" }, - { url = "https://files.pythonhosted.org/packages/35/91/9cb56efbb428b006bb85db28591e40b7736847b8331d43fe335acf95f6c8/propcache-0.3.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:4445542398bd0b5d32df908031cb1b30d43ac848e20470a878b770ec2dcc6330", size = 265778, upload-time = "2025-06-09T22:55:36.45Z" }, - { url = "https://files.pythonhosted.org/packages/9a/4c/b0fe775a2bdd01e176b14b574be679d84fc83958335790f7c9a686c1f468/propcache-0.3.2-cp313-cp313t-win32.whl", hash = "sha256:f86e5d7cd03afb3a1db8e9f9f6eff15794e79e791350ac48a8c924e6f439f394", size = 41175, upload-time = "2025-06-09T22:55:38.436Z" }, - { url = "https://files.pythonhosted.org/packages/a4/ff/47f08595e3d9b5e149c150f88d9714574f1a7cbd89fe2817158a952674bf/propcache-0.3.2-cp313-cp313t-win_amd64.whl", hash = "sha256:9704bedf6e7cbe3c65eca4379a9b53ee6a83749f047808cbb5044d40d7d72198", size = 44857, upload-time = "2025-06-09T22:55:39.687Z" }, - { url = "https://files.pythonhosted.org/packages/cc/35/cc0aaecf278bb4575b8555f2b137de5ab821595ddae9da9d3cd1da4072c7/propcache-0.3.2-py3-none-any.whl", hash = "sha256:98f1ec44fb675f5052cccc8e609c46ed23a35a1cfd18545ad4e29002d858a43f", size = 12663, upload-time = "2025-06-09T22:56:04.484Z" }, +version = "0.3.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/92/76/f941e63d55c0293ff7829dd21e7cf1147e90a526756869a9070f287a68c9/propcache-0.3.0.tar.gz", hash = "sha256:a8fd93de4e1d278046345f49e2238cdb298589325849b2645d4a94c53faeffc5", size = 42722, upload-time = "2025-02-20T19:03:29.191Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/8d/f0/dc9ec44d2e63c13f816a16398c039329736712440ff82b682dd9a78d2258/propcache-0.3.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:efa44f64c37cc30c9f05932c740a8b40ce359f51882c70883cc95feac842da4d", size = 79574, upload-time = "2025-02-20T18:59:44.353Z" }, + { url = "https://files.pythonhosted.org/packages/99/3a/33a207dfcb3ee1131ea23a2aeb726c3c4994f89546d7eadf8c50627c8b63/propcache-0.3.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2383a17385d9800b6eb5855c2f05ee550f803878f344f58b6e194de08b96352c", size = 45898, upload-time = "2025-02-20T18:59:46.783Z" }, + { url = "https://files.pythonhosted.org/packages/af/68/0bde765c9f5dc02b4466d2838600af38c81b184c26c6d3cd44643ac668e3/propcache-0.3.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:d3e7420211f5a65a54675fd860ea04173cde60a7cc20ccfbafcccd155225f8bc", size = 45418, upload-time = "2025-02-20T18:59:49.082Z" }, + { url = "https://files.pythonhosted.org/packages/06/a6/c682669bae41199358e16cc7b1c818f91c5f9e925cc863dabd98ce32716a/propcache-0.3.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3302c5287e504d23bb0e64d2a921d1eb4a03fb93a0a0aa3b53de059f5a5d737d", size = 205116, upload-time = "2025-02-20T18:59:50.606Z" }, + { url = "https://files.pythonhosted.org/packages/fb/ae/82cfb50267d9a1baa0340728eb9e32245a68538fef929d7bb786d01c11a8/propcache-0.3.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:7e2e068a83552ddf7a39a99488bcba05ac13454fb205c847674da0352602082f", size = 219405, upload-time = "2025-02-20T18:59:54.016Z" }, + { url = "https://files.pythonhosted.org/packages/ab/16/7b6b2bf8c207cfd0e5ca3d41aea397392de9899867ec024f88c94f9ae2ab/propcache-0.3.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2d913d36bdaf368637b4f88d554fb9cb9d53d6920b9c5563846555938d5450bf", size = 217656, upload-time = "2025-02-20T18:59:55.747Z" }, + { url = "https://files.pythonhosted.org/packages/f4/eb/41447de61eb5454891658d0fb9b1d7d35d49a4a5dd2e0c86f2c332e8b7e1/propcache-0.3.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8ee1983728964d6070ab443399c476de93d5d741f71e8f6e7880a065f878e0b9", size = 205414, upload-time = "2025-02-20T18:59:59.907Z" }, + { url = "https://files.pythonhosted.org/packages/03/b6/9719878f8b5b20d37ee663a40f8dcbf888559e4d3be2ba2fe5c790fc28d2/propcache-0.3.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:36ca5e9a21822cc1746023e88f5c0af6fce3af3b85d4520efb1ce4221bed75cc", size = 195746, upload-time = "2025-02-20T19:00:03.124Z" }, + { url = "https://files.pythonhosted.org/packages/bb/ec/b79c3210ba459800d1a8f1afeb81d7b503893555a7b79c24082ff26d3314/propcache-0.3.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:9ecde3671e62eeb99e977f5221abcf40c208f69b5eb986b061ccec317c82ebd0", size = 198651, upload-time = "2025-02-20T19:00:04.747Z" }, + { url = "https://files.pythonhosted.org/packages/48/f6/2b0140bc47013e43575973068e72ad51ee9f22f2dad42e6d6e362d715125/propcache-0.3.0-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:d383bf5e045d7f9d239b38e6acadd7b7fdf6c0087259a84ae3475d18e9a2ae8b", size = 195858, upload-time = "2025-02-20T19:00:06.723Z" }, + { url = "https://files.pythonhosted.org/packages/97/3d/2fa19303d87aa21f9a42dcd870d6088a2a776ff5518e394d50412c3679a6/propcache-0.3.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:8cb625bcb5add899cb8ba7bf716ec1d3e8f7cdea9b0713fa99eadf73b6d4986f", size = 197181, upload-time = "2025-02-20T19:00:08.31Z" }, + { url = "https://files.pythonhosted.org/packages/09/f3/a2170ffc9fa774c1dfd52294113c0fa6cdc5b71dbfd7129bb9378fdd8b42/propcache-0.3.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:5fa159dcee5dba00c1def3231c249cf261185189205073bde13797e57dd7540a", size = 207411, upload-time = "2025-02-20T19:00:10.546Z" }, + { url = "https://files.pythonhosted.org/packages/d6/1e/cb8a6c82178efffa0b00dc463f36cd086f747345585140aeb95d5cb93666/propcache-0.3.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:a7080b0159ce05f179cfac592cda1a82898ca9cd097dacf8ea20ae33474fbb25", size = 210724, upload-time = "2025-02-20T19:00:12.207Z" }, + { url = "https://files.pythonhosted.org/packages/2b/72/6e273543337a3e22cf462eb836f065a9830b4d41baeb1f58db2695c934f3/propcache-0.3.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:ed7161bccab7696a473fe7ddb619c1d75963732b37da4618ba12e60899fefe4f", size = 203511, upload-time = "2025-02-20T19:00:14.689Z" }, + { url = "https://files.pythonhosted.org/packages/f3/ea/7412c79bcec06597c967d49789f5a1f7fd76a8654908feeaefafb7447c9a/propcache-0.3.0-cp310-cp310-win32.whl", hash = "sha256:bf0d9a171908f32d54f651648c7290397b8792f4303821c42a74e7805bfb813c", size = 40600, upload-time = "2025-02-20T19:00:16.423Z" }, + { url = "https://files.pythonhosted.org/packages/a3/42/488c90190491f3e61bd2c2fb0b3d91c1c78778270dde2f0b6633fc9ff723/propcache-0.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:42924dc0c9d73e49908e35bbdec87adedd651ea24c53c29cac103ede0ea1d340", size = 44714, upload-time = "2025-02-20T19:00:18.709Z" }, + { url = "https://files.pythonhosted.org/packages/45/c9/cf09ff7e6d09f14149094f7cd50d2dec032b24e61af21fc4540da2b17bfb/propcache-0.3.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:9ddd49258610499aab83b4f5b61b32e11fce873586282a0e972e5ab3bcadee51", size = 79568, upload-time = "2025-02-20T19:00:21.457Z" }, + { url = "https://files.pythonhosted.org/packages/c8/32/2424d89da88cd81b7d148e0d2b3131461b570a02aa9d84a2e567509adb0d/propcache-0.3.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:2578541776769b500bada3f8a4eeaf944530516b6e90c089aa368266ed70c49e", size = 45895, upload-time = "2025-02-20T19:00:23.035Z" }, + { url = "https://files.pythonhosted.org/packages/f6/91/ee5b6aa7aa31754fefcf0c5180e09223cac380ef195c4ddc8c266eb641ea/propcache-0.3.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:d8074c5dd61c8a3e915fa8fc04754fa55cfa5978200d2daa1e2d4294c1f136aa", size = 45427, upload-time = "2025-02-20T19:00:25.07Z" }, + { url = "https://files.pythonhosted.org/packages/bf/73/38f0128462b8b616181d8c53bd5d04eac41c50c449b07615c65d56ba0a9b/propcache-0.3.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b58229a844931bca61b3a20efd2be2a2acb4ad1622fc026504309a6883686fbf", size = 232427, upload-time = "2025-02-20T19:00:26.587Z" }, + { url = "https://files.pythonhosted.org/packages/59/82/f3d4e84f4539dcfc9c3d338282b9e915f5b63c921986ecfdf7af2d12f87c/propcache-0.3.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e45377d5d6fefe1677da2a2c07b024a6dac782088e37c0b1efea4cfe2b1be19b", size = 239985, upload-time = "2025-02-20T19:00:28.204Z" }, + { url = "https://files.pythonhosted.org/packages/42/e8/029f58cccbae83c9969a7ee7a06558d5b83a93dfc54e0f4f70234bbaea1b/propcache-0.3.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ec5060592d83454e8063e487696ac3783cc48c9a329498bafae0d972bc7816c9", size = 238827, upload-time = "2025-02-20T19:00:30.147Z" }, + { url = "https://files.pythonhosted.org/packages/8b/a2/c373561777c0cb9b9e7b9b9a10b9b3a7b6bde75a2535b962231cecc8fdb8/propcache-0.3.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:15010f29fbed80e711db272909a074dc79858c6d28e2915704cfc487a8ac89c6", size = 231348, upload-time = "2025-02-20T19:00:32.05Z" }, + { url = "https://files.pythonhosted.org/packages/d7/d2/4673f715beedf6038b485bcd976813149231d9df5bb6196cb69a09c185c9/propcache-0.3.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a254537b9b696ede293bfdbc0a65200e8e4507bc9f37831e2a0318a9b333c85c", size = 220426, upload-time = "2025-02-20T19:00:34.756Z" }, + { url = "https://files.pythonhosted.org/packages/e0/f6/1da65f900927bafd4675a16e890618ec7643f2f922bf0e4d84bb38645618/propcache-0.3.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:2b975528998de037dfbc10144b8aed9b8dd5a99ec547f14d1cb7c5665a43f075", size = 220294, upload-time = "2025-02-20T19:00:38.63Z" }, + { url = "https://files.pythonhosted.org/packages/ff/86/620451bdc02e91b1712cd71890c17077ee97e2a28493836a87e47b8e70ff/propcache-0.3.0-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:19d36bb351ad5554ff20f2ae75f88ce205b0748c38b146c75628577020351e3c", size = 212492, upload-time = "2025-02-20T19:00:41.077Z" }, + { url = "https://files.pythonhosted.org/packages/6e/1b/e8f86921ed4016da80faf3b8f515f7829decabdbff106736bfff353bceba/propcache-0.3.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:6032231d4a5abd67c7f71168fd64a47b6b451fbcb91c8397c2f7610e67683810", size = 215113, upload-time = "2025-02-20T19:00:43.577Z" }, + { url = "https://files.pythonhosted.org/packages/1a/95/a61d86cc49aa0945f6c06f3a4614fc543e311a50558c92861f5e9691a37c/propcache-0.3.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:6985a593417cdbc94c7f9c3403747335e450c1599da1647a5af76539672464d3", size = 228330, upload-time = "2025-02-20T19:00:45.163Z" }, + { url = "https://files.pythonhosted.org/packages/8f/7d/10dbae48ff2bb189e92c2b3487a48f3229146a25941ad0d485934d1104d4/propcache-0.3.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:6a1948df1bb1d56b5e7b0553c0fa04fd0e320997ae99689488201f19fa90d2e7", size = 231942, upload-time = "2025-02-20T19:00:46.771Z" }, + { url = "https://files.pythonhosted.org/packages/39/ce/82d16aec96c5513ae7db13ab901a65a1e54c915292fb5b2390e33275b61d/propcache-0.3.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:8319293e85feadbbfe2150a5659dbc2ebc4afdeaf7d98936fb9a2f2ba0d4c35c", size = 223077, upload-time = "2025-02-20T19:00:53.044Z" }, + { url = "https://files.pythonhosted.org/packages/c8/e0/cb077e8e7a583c733df7f53327fcbdb92e42be59b976ce60bf1d904a0efe/propcache-0.3.0-cp311-cp311-win32.whl", hash = "sha256:63f26258a163c34542c24808f03d734b338da66ba91f410a703e505c8485791d", size = 40455, upload-time = "2025-02-20T19:00:55.338Z" }, + { url = "https://files.pythonhosted.org/packages/d8/35/57abeb6146fe3c19081eeaf3d9d4cfea256f87f1e5101acf80d3332c1820/propcache-0.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:cacea77ef7a2195f04f9279297684955e3d1ae4241092ff0cfcef532bb7a1c32", size = 44705, upload-time = "2025-02-20T19:00:56.947Z" }, + { url = "https://files.pythonhosted.org/packages/8d/2c/921f15dc365796ec23975b322b0078eae72995c7b4d49eba554c6a308d70/propcache-0.3.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:e53d19c2bf7d0d1e6998a7e693c7e87300dd971808e6618964621ccd0e01fe4e", size = 79867, upload-time = "2025-02-20T19:00:59.948Z" }, + { url = "https://files.pythonhosted.org/packages/11/a5/4a6cc1a559d1f2fb57ea22edc4245158cdffae92f7f92afcee2913f84417/propcache-0.3.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:a61a68d630e812b67b5bf097ab84e2cd79b48c792857dc10ba8a223f5b06a2af", size = 46109, upload-time = "2025-02-20T19:01:04.447Z" }, + { url = "https://files.pythonhosted.org/packages/e1/6d/28bfd3af3a567ad7d667348e7f46a520bda958229c4d545ba138a044232f/propcache-0.3.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:fb91d20fa2d3b13deea98a690534697742029f4fb83673a3501ae6e3746508b5", size = 45635, upload-time = "2025-02-20T19:01:07.024Z" }, + { url = "https://files.pythonhosted.org/packages/73/20/d75b42eaffe5075eac2f4e168f6393d21c664c91225288811d85451b2578/propcache-0.3.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:67054e47c01b7b349b94ed0840ccae075449503cf1fdd0a1fdd98ab5ddc2667b", size = 242159, upload-time = "2025-02-20T19:01:10.047Z" }, + { url = "https://files.pythonhosted.org/packages/a5/fb/4b537dd92f9fd4be68042ec51c9d23885ca5fafe51ec24c58d9401034e5f/propcache-0.3.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:997e7b8f173a391987df40f3b52c423e5850be6f6df0dcfb5376365440b56667", size = 248163, upload-time = "2025-02-20T19:01:12.883Z" }, + { url = "https://files.pythonhosted.org/packages/e7/af/8a9db04ac596d531ca0ef7dde518feaadfcdabef7b17d6a5ec59ee3effc2/propcache-0.3.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:8d663fd71491dde7dfdfc899d13a067a94198e90695b4321084c6e450743b8c7", size = 248794, upload-time = "2025-02-20T19:01:15.291Z" }, + { url = "https://files.pythonhosted.org/packages/9d/c4/ecfc988879c0fd9db03228725b662d76cf484b6b46f7e92fee94e4b52490/propcache-0.3.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8884ba1a0fe7210b775106b25850f5e5a9dc3c840d1ae9924ee6ea2eb3acbfe7", size = 243912, upload-time = "2025-02-20T19:01:16.95Z" }, + { url = "https://files.pythonhosted.org/packages/04/a2/298dd27184faa8b7d91cc43488b578db218b3cc85b54d912ed27b8c5597a/propcache-0.3.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:aa806bbc13eac1ab6291ed21ecd2dd426063ca5417dd507e6be58de20e58dfcf", size = 229402, upload-time = "2025-02-20T19:01:20.913Z" }, + { url = "https://files.pythonhosted.org/packages/be/0d/efe7fec316ca92dbf4bc4a9ba49ca889c43ca6d48ab1d6fa99fc94e5bb98/propcache-0.3.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6f4d7a7c0aff92e8354cceca6fe223973ddf08401047920df0fcb24be2bd5138", size = 226896, upload-time = "2025-02-20T19:01:23.57Z" }, + { url = "https://files.pythonhosted.org/packages/60/63/72404380ae1d9c96d96e165aa02c66c2aae6072d067fc4713da5cde96762/propcache-0.3.0-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:9be90eebc9842a93ef8335291f57b3b7488ac24f70df96a6034a13cb58e6ff86", size = 221447, upload-time = "2025-02-20T19:01:26.142Z" }, + { url = "https://files.pythonhosted.org/packages/9d/18/b8392cab6e0964b67a30a8f4dadeaff64dc7022b5a34bb1d004ea99646f4/propcache-0.3.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:bf15fc0b45914d9d1b706f7c9c4f66f2b7b053e9517e40123e137e8ca8958b3d", size = 222440, upload-time = "2025-02-20T19:01:28.438Z" }, + { url = "https://files.pythonhosted.org/packages/6f/be/105d9ceda0f97eff8c06bac1673448b2db2a497444de3646464d3f5dc881/propcache-0.3.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:5a16167118677d94bb48bfcd91e420088854eb0737b76ec374b91498fb77a70e", size = 234104, upload-time = "2025-02-20T19:01:31.256Z" }, + { url = "https://files.pythonhosted.org/packages/cb/c9/f09a4ec394cfcce4053d8b2a04d622b5f22d21ba9bb70edd0cad061fa77b/propcache-0.3.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:41de3da5458edd5678b0f6ff66691507f9885f5fe6a0fb99a5d10d10c0fd2d64", size = 239086, upload-time = "2025-02-20T19:01:33.753Z" }, + { url = "https://files.pythonhosted.org/packages/ea/aa/96f7f9ed6def82db67c972bdb7bd9f28b95d7d98f7e2abaf144c284bf609/propcache-0.3.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:728af36011bb5d344c4fe4af79cfe186729efb649d2f8b395d1572fb088a996c", size = 230991, upload-time = "2025-02-20T19:01:35.433Z" }, + { url = "https://files.pythonhosted.org/packages/5a/11/bee5439de1307d06fad176f7143fec906e499c33d7aff863ea8428b8e98b/propcache-0.3.0-cp312-cp312-win32.whl", hash = "sha256:6b5b7fd6ee7b54e01759f2044f936dcf7dea6e7585f35490f7ca0420fe723c0d", size = 40337, upload-time = "2025-02-20T19:01:37.655Z" }, + { url = "https://files.pythonhosted.org/packages/e4/17/e5789a54a0455a61cb9efc4ca6071829d992220c2998a27c59aeba749f6f/propcache-0.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:2d15bc27163cd4df433e75f546b9ac31c1ba7b0b128bfb1b90df19082466ff57", size = 44404, upload-time = "2025-02-20T19:01:38.946Z" }, + { url = "https://files.pythonhosted.org/packages/3a/0f/a79dd23a0efd6ee01ab0dc9750d8479b343bfd0c73560d59d271eb6a99d4/propcache-0.3.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:a2b9bf8c79b660d0ca1ad95e587818c30ccdb11f787657458d6f26a1ea18c568", size = 77287, upload-time = "2025-02-20T19:01:40.897Z" }, + { url = "https://files.pythonhosted.org/packages/b8/51/76675703c90de38ac75adb8deceb3f3ad99b67ff02a0fa5d067757971ab8/propcache-0.3.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b0c1a133d42c6fc1f5fbcf5c91331657a1ff822e87989bf4a6e2e39b818d0ee9", size = 44923, upload-time = "2025-02-20T19:01:42.397Z" }, + { url = "https://files.pythonhosted.org/packages/01/9b/fd5ddbee66cf7686e73c516227c2fd9bf471dbfed0f48329d095ea1228d3/propcache-0.3.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:bb2f144c6d98bb5cbc94adeb0447cfd4c0f991341baa68eee3f3b0c9c0e83767", size = 44325, upload-time = "2025-02-20T19:01:43.976Z" }, + { url = "https://files.pythonhosted.org/packages/13/1c/6961f11eb215a683b34b903b82bde486c606516c1466bf1fa67f26906d51/propcache-0.3.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d1323cd04d6e92150bcc79d0174ce347ed4b349d748b9358fd2e497b121e03c8", size = 225116, upload-time = "2025-02-20T19:01:45.488Z" }, + { url = "https://files.pythonhosted.org/packages/ef/ea/f8410c40abcb2e40dffe9adeed017898c930974650a63e5c79b886aa9f73/propcache-0.3.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:3b812b3cb6caacd072276ac0492d249f210006c57726b6484a1e1805b3cfeea0", size = 229905, upload-time = "2025-02-20T19:01:49.454Z" }, + { url = "https://files.pythonhosted.org/packages/ef/5a/a9bf90894001468bf8e6ea293bb00626cc9ef10f8eb7996e9ec29345c7ed/propcache-0.3.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:742840d1d0438eb7ea4280f3347598f507a199a35a08294afdcc560c3739989d", size = 233221, upload-time = "2025-02-20T19:01:51.142Z" }, + { url = "https://files.pythonhosted.org/packages/dd/ce/fffdddd9725b690b01d345c1156b4c2cc6dca09ab5c23a6d07b8f37d6e2f/propcache-0.3.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7c6e7e4f9167fddc438cd653d826f2222222564daed4116a02a184b464d3ef05", size = 227627, upload-time = "2025-02-20T19:01:53.695Z" }, + { url = "https://files.pythonhosted.org/packages/58/ae/45c89a5994a334735a3032b48e8e4a98c05d9536ddee0719913dc27da548/propcache-0.3.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a94ffc66738da99232ddffcf7910e0f69e2bbe3a0802e54426dbf0714e1c2ffe", size = 214217, upload-time = "2025-02-20T19:01:55.309Z" }, + { url = "https://files.pythonhosted.org/packages/01/84/bc60188c3290ff8f5f4a92b9ca2d93a62e449c8daf6fd11ad517ad136926/propcache-0.3.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:3c6ec957025bf32b15cbc6b67afe233c65b30005e4c55fe5768e4bb518d712f1", size = 212921, upload-time = "2025-02-20T19:01:57.893Z" }, + { url = "https://files.pythonhosted.org/packages/14/b3/39d60224048feef7a96edabb8217dc3f75415457e5ebbef6814f8b2a27b5/propcache-0.3.0-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:549722908de62aa0b47a78b90531c022fa6e139f9166be634f667ff45632cc92", size = 208200, upload-time = "2025-02-20T19:02:00.026Z" }, + { url = "https://files.pythonhosted.org/packages/9d/b3/0a6720b86791251273fff8a01bc8e628bc70903513bd456f86cde1e1ef84/propcache-0.3.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:5d62c4f6706bff5d8a52fd51fec6069bef69e7202ed481486c0bc3874912c787", size = 208400, upload-time = "2025-02-20T19:02:03.997Z" }, + { url = "https://files.pythonhosted.org/packages/e9/4f/bb470f3e687790547e2e78105fb411f54e0cdde0d74106ccadd2521c6572/propcache-0.3.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:24c04f8fbf60094c531667b8207acbae54146661657a1b1be6d3ca7773b7a545", size = 218116, upload-time = "2025-02-20T19:02:06.042Z" }, + { url = "https://files.pythonhosted.org/packages/34/71/277f7f9add469698ac9724c199bfe06f85b199542121a71f65a80423d62a/propcache-0.3.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:7c5f5290799a3f6539cc5e6f474c3e5c5fbeba74a5e1e5be75587746a940d51e", size = 222911, upload-time = "2025-02-20T19:02:08.748Z" }, + { url = "https://files.pythonhosted.org/packages/92/e3/a7b9782aef5a2fc765b1d97da9ec7aed2f25a4e985703608e73232205e3f/propcache-0.3.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4fa0e7c9c3cf7c276d4f6ab9af8adddc127d04e0fcabede315904d2ff76db626", size = 216563, upload-time = "2025-02-20T19:02:11.322Z" }, + { url = "https://files.pythonhosted.org/packages/ab/76/0583ca2c551aa08ffcff87b2c6849c8f01c1f6fb815a5226f0c5c202173e/propcache-0.3.0-cp313-cp313-win32.whl", hash = "sha256:ee0bd3a7b2e184e88d25c9baa6a9dc609ba25b76daae942edfb14499ac7ec374", size = 39763, upload-time = "2025-02-20T19:02:12.977Z" }, + { url = "https://files.pythonhosted.org/packages/80/ec/c6a84f9a36f608379b95f0e786c111d5465926f8c62f12be8cdadb02b15c/propcache-0.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:1c8f7d896a16da9455f882870a507567d4f58c53504dc2d4b1e1d386dfe4588a", size = 43650, upload-time = "2025-02-20T19:02:15.041Z" }, + { url = "https://files.pythonhosted.org/packages/ee/95/7d32e3560f5bf83fc2f2a4c1b0c181d327d53d5f85ebd045ab89d4d97763/propcache-0.3.0-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:e560fd75aaf3e5693b91bcaddd8b314f4d57e99aef8a6c6dc692f935cc1e6bbf", size = 82140, upload-time = "2025-02-20T19:02:16.562Z" }, + { url = "https://files.pythonhosted.org/packages/86/89/752388f12e6027a5e63f5d075f15291ded48e2d8311314fff039da5a9b11/propcache-0.3.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:65a37714b8ad9aba5780325228598a5b16c47ba0f8aeb3dc0514701e4413d7c0", size = 47296, upload-time = "2025-02-20T19:02:17.974Z" }, + { url = "https://files.pythonhosted.org/packages/1b/4c/b55c98d586c69180d3048984a57a5ea238bdeeccf82dbfcd598e935e10bb/propcache-0.3.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:07700939b2cbd67bfb3b76a12e1412405d71019df00ca5697ce75e5ef789d829", size = 46724, upload-time = "2025-02-20T19:02:19.588Z" }, + { url = "https://files.pythonhosted.org/packages/0f/b6/67451a437aed90c4e951e320b5b3d7eb584ade1d5592f6e5e8f678030989/propcache-0.3.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7c0fdbdf6983526e269e5a8d53b7ae3622dd6998468821d660d0daf72779aefa", size = 291499, upload-time = "2025-02-20T19:02:21.1Z" }, + { url = "https://files.pythonhosted.org/packages/ee/ff/e4179facd21515b24737e1e26e02615dfb5ed29416eed4cf5bc6ac5ce5fb/propcache-0.3.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:794c3dd744fad478b6232289c866c25406ecdfc47e294618bdf1697e69bd64a6", size = 293911, upload-time = "2025-02-20T19:02:24.248Z" }, + { url = "https://files.pythonhosted.org/packages/76/8d/94a8585992a064a23bd54f56c5e58c3b8bf0c0a06ae10e56f2353ae16c3d/propcache-0.3.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4544699674faf66fb6b4473a1518ae4999c1b614f0b8297b1cef96bac25381db", size = 293301, upload-time = "2025-02-20T19:02:26.034Z" }, + { url = "https://files.pythonhosted.org/packages/b0/b8/2c860c92b4134f68c7716c6f30a0d723973f881c32a6d7a24c4ddca05fdf/propcache-0.3.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fddb8870bdb83456a489ab67c6b3040a8d5a55069aa6f72f9d872235fbc52f54", size = 281947, upload-time = "2025-02-20T19:02:27.838Z" }, + { url = "https://files.pythonhosted.org/packages/cd/72/b564be7411b525d11757b713c757c21cd4dc13b6569c3b2b8f6d3c96fd5e/propcache-0.3.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f857034dc68d5ceb30fb60afb6ff2103087aea10a01b613985610e007053a121", size = 268072, upload-time = "2025-02-20T19:02:29.594Z" }, + { url = "https://files.pythonhosted.org/packages/37/68/d94649e399e8d7fc051e5a4f2334efc567993525af083db145a70690a121/propcache-0.3.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:02df07041e0820cacc8f739510078f2aadcfd3fc57eaeeb16d5ded85c872c89e", size = 275190, upload-time = "2025-02-20T19:02:32.255Z" }, + { url = "https://files.pythonhosted.org/packages/d8/3c/446e125f5bbbc1922964dd67cb541c01cdb678d811297b79a4ff6accc843/propcache-0.3.0-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:f47d52fd9b2ac418c4890aad2f6d21a6b96183c98021f0a48497a904199f006e", size = 254145, upload-time = "2025-02-20T19:02:33.932Z" }, + { url = "https://files.pythonhosted.org/packages/f4/80/fd3f741483dc8e59f7ba7e05eaa0f4e11677d7db2077522b92ff80117a2a/propcache-0.3.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:9ff4e9ecb6e4b363430edf2c6e50173a63e0820e549918adef70515f87ced19a", size = 257163, upload-time = "2025-02-20T19:02:35.675Z" }, + { url = "https://files.pythonhosted.org/packages/dc/cf/6292b5ce6ed0017e6a89024a827292122cc41b6259b30ada0c6732288513/propcache-0.3.0-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:ecc2920630283e0783c22e2ac94427f8cca29a04cfdf331467d4f661f4072dac", size = 280249, upload-time = "2025-02-20T19:02:38.406Z" }, + { url = "https://files.pythonhosted.org/packages/e8/f0/fd9b8247b449fe02a4f96538b979997e229af516d7462b006392badc59a1/propcache-0.3.0-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:c441c841e82c5ba7a85ad25986014be8d7849c3cfbdb6004541873505929a74e", size = 288741, upload-time = "2025-02-20T19:02:40.149Z" }, + { url = "https://files.pythonhosted.org/packages/64/71/cf831fdc2617f86cfd7f414cfc487d018e722dac8acc098366ce9bba0941/propcache-0.3.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:6c929916cbdb540d3407c66f19f73387f43e7c12fa318a66f64ac99da601bcdf", size = 277061, upload-time = "2025-02-20T19:02:42.309Z" }, + { url = "https://files.pythonhosted.org/packages/42/78/9432542a35d944abeca9e02927a0de38cd7a298466d8ffa171536e2381c3/propcache-0.3.0-cp313-cp313t-win32.whl", hash = "sha256:0c3e893c4464ebd751b44ae76c12c5f5c1e4f6cbd6fbf67e3783cd93ad221863", size = 42252, upload-time = "2025-02-20T19:02:44.447Z" }, + { url = "https://files.pythonhosted.org/packages/6f/45/960365f4f8978f48ebb56b1127adf33a49f2e69ecd46ac1f46d6cf78a79d/propcache-0.3.0-cp313-cp313t-win_amd64.whl", hash = "sha256:75e872573220d1ee2305b35c9813626e620768248425f58798413e9c39741f46", size = 46425, upload-time = "2025-02-20T19:02:48.071Z" }, + { url = "https://files.pythonhosted.org/packages/b5/35/6c4c6fc8774a9e3629cd750dc24a7a4fb090a25ccd5c3246d127b70f9e22/propcache-0.3.0-py3-none-any.whl", hash = "sha256:67dda3c7325691c2081510e92c561f465ba61b975f481735aefdfc845d2cd043", size = 12101, upload-time = "2025-02-20T19:03:27.202Z" }, ] [[package]] @@ -2826,16 +2587,16 @@ wheels = [ [[package]] name = "protobuf" -version = "6.31.1" +version = "5.29.4" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/52/f3/b9655a711b32c19720253f6f06326faf90580834e2e83f840472d752bc8b/protobuf-6.31.1.tar.gz", hash = "sha256:d8cac4c982f0b957a4dc73a80e2ea24fab08e679c0de9deb835f4a12d69aca9a", size = 441797, upload-time = "2025-05-28T19:25:54.947Z" } +sdist = { url = "https://files.pythonhosted.org/packages/17/7d/b9dca7365f0e2c4fa7c193ff795427cfa6290147e5185ab11ece280a18e7/protobuf-5.29.4.tar.gz", hash = "sha256:4f1dfcd7997b31ef8f53ec82781ff434a28bf71d9102ddde14d076adcfc78c99", size = 424902, upload-time = "2025-03-19T21:23:24.25Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/f3/6f/6ab8e4bf962fd5570d3deaa2d5c38f0a363f57b4501047b5ebeb83ab1125/protobuf-6.31.1-cp310-abi3-win32.whl", hash = "sha256:7fa17d5a29c2e04b7d90e5e32388b8bfd0e7107cd8e616feef7ed3fa6bdab5c9", size = 423603, upload-time = "2025-05-28T19:25:41.198Z" }, - { url = "https://files.pythonhosted.org/packages/44/3a/b15c4347dd4bf3a1b0ee882f384623e2063bb5cf9fa9d57990a4f7df2fb6/protobuf-6.31.1-cp310-abi3-win_amd64.whl", hash = "sha256:426f59d2964864a1a366254fa703b8632dcec0790d8862d30034d8245e1cd447", size = 435283, upload-time = "2025-05-28T19:25:44.275Z" }, - { url = "https://files.pythonhosted.org/packages/6a/c9/b9689a2a250264a84e66c46d8862ba788ee7a641cdca39bccf64f59284b7/protobuf-6.31.1-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:6f1227473dc43d44ed644425268eb7c2e488ae245d51c6866d19fe158e207402", size = 425604, upload-time = "2025-05-28T19:25:45.702Z" }, - { url = "https://files.pythonhosted.org/packages/76/a1/7a5a94032c83375e4fe7e7f56e3976ea6ac90c5e85fac8576409e25c39c3/protobuf-6.31.1-cp39-abi3-manylinux2014_aarch64.whl", hash = "sha256:a40fc12b84c154884d7d4c4ebd675d5b3b5283e155f324049ae396b95ddebc39", size = 322115, upload-time = "2025-05-28T19:25:47.128Z" }, - { url = "https://files.pythonhosted.org/packages/fa/b1/b59d405d64d31999244643d88c45c8241c58f17cc887e73bcb90602327f8/protobuf-6.31.1-cp39-abi3-manylinux2014_x86_64.whl", hash = "sha256:4ee898bf66f7a8b0bd21bce523814e6fbd8c6add948045ce958b73af7e8878c6", size = 321070, upload-time = "2025-05-28T19:25:50.036Z" }, - { url = "https://files.pythonhosted.org/packages/f7/af/ab3c51ab7507a7325e98ffe691d9495ee3d3aa5f589afad65ec920d39821/protobuf-6.31.1-py3-none-any.whl", hash = "sha256:720a6c7e6b77288b85063569baae8536671b39f15cc22037ec7045658d80489e", size = 168724, upload-time = "2025-05-28T19:25:53.926Z" }, + { url = "https://files.pythonhosted.org/packages/9a/b2/043a1a1a20edd134563699b0e91862726a0dc9146c090743b6c44d798e75/protobuf-5.29.4-cp310-abi3-win32.whl", hash = "sha256:13eb236f8eb9ec34e63fc8b1d6efd2777d062fa6aaa68268fb67cf77f6839ad7", size = 422709, upload-time = "2025-03-19T21:23:08.293Z" }, + { url = "https://files.pythonhosted.org/packages/79/fc/2474b59570daa818de6124c0a15741ee3e5d6302e9d6ce0bdfd12e98119f/protobuf-5.29.4-cp310-abi3-win_amd64.whl", hash = "sha256:bcefcdf3976233f8a502d265eb65ea740c989bacc6c30a58290ed0e519eb4b8d", size = 434506, upload-time = "2025-03-19T21:23:11.253Z" }, + { url = "https://files.pythonhosted.org/packages/46/de/7c126bbb06aa0f8a7b38aaf8bd746c514d70e6a2a3f6dd460b3b7aad7aae/protobuf-5.29.4-cp38-abi3-macosx_10_9_universal2.whl", hash = "sha256:307ecba1d852ec237e9ba668e087326a67564ef83e45a0189a772ede9e854dd0", size = 417826, upload-time = "2025-03-19T21:23:13.132Z" }, + { url = "https://files.pythonhosted.org/packages/a2/b5/bade14ae31ba871a139aa45e7a8183d869efe87c34a4850c87b936963261/protobuf-5.29.4-cp38-abi3-manylinux2014_aarch64.whl", hash = "sha256:aec4962f9ea93c431d5714ed1be1c93f13e1a8618e70035ba2b0564d9e633f2e", size = 319574, upload-time = "2025-03-19T21:23:14.531Z" }, + { url = "https://files.pythonhosted.org/packages/46/88/b01ed2291aae68b708f7d334288ad5fb3e7aa769a9c309c91a0d55cb91b0/protobuf-5.29.4-cp38-abi3-manylinux2014_x86_64.whl", hash = "sha256:d7d3f7d1d5a66ed4942d4fefb12ac4b14a29028b209d4bfb25c68ae172059922", size = 319672, upload-time = "2025-03-19T21:23:15.839Z" }, + { url = "https://files.pythonhosted.org/packages/12/fb/a586e0c973c95502e054ac5f81f88394f24ccc7982dac19c515acd9e2c93/protobuf-5.29.4-py3-none-any.whl", hash = "sha256:3fde11b505e1597f71b875ef2fc52062b6a9740e5f7c8997ce878b6009145862", size = 172551, upload-time = "2025-03-19T21:23:22.682Z" }, ] [[package]] @@ -2873,11 +2634,54 @@ wheels = [ [[package]] name = "py3dmol" -version = "2.5.2" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/33/e9/fa4fce24843554d4e025fcdb2fcebfd0ec14db88a4a57545eb38b15825be/py3dmol-2.5.2.tar.gz", hash = "sha256:9ef1c72c786b22e33541e05b27e97bb99ea1d3c962819ed55c2203d10a515198", size = 7857, upload-time = "2025-07-31T19:35:41.833Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/a3/17/c476487ba903c7d793db633b4e8ca4a420ae272890302189d9402ba8ff85/py3dmol-2.5.2-py2.py3-none-any.whl", hash = "sha256:b921940ff046cf7ca008a249cbd5debec561dcf337f1e5f3df7ac5d4a1954e8e", size = 7154, upload-time = "2025-07-31T19:35:41.022Z" }, +version = "2.4.2" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/8d/ba/78bc5b451f314c06e6c2a3ca0f0ba18ee751f10e99fed94fc09175b16031/py3Dmol-2.4.2.tar.gz", hash = "sha256:990ed67b2dda5493d21192fef53a52be6128b8afbdc13da8a40a009060120749", size = 7724, upload-time = "2024-11-08T22:19:23.15Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/04/20/923885064f4e4d4392eb2be798532d91b315f9e60ef44f49f4800ba3c57a/py3Dmol-2.4.2-py2.py3-none-any.whl", hash = "sha256:bec23d9a015d692279a5f7d4db92803e4e82ba3bdcc1434a5b6a2be98a347856", size = 7046, upload-time = "2024-11-08T22:19:21.631Z" }, +] + +[[package]] +name = "pyarrow" +version = "21.0.0" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/ef/c2/ea068b8f00905c06329a3dfcd40d0fcc2b7d0f2e355bdb25b65e0a0e4cd4/pyarrow-21.0.0.tar.gz", hash = "sha256:5051f2dccf0e283ff56335760cbc8622cf52264d67e359d5569541ac11b6d5bc", size = 1133487, upload-time = "2025-07-18T00:57:31.761Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/17/d9/110de31880016e2afc52d8580b397dbe47615defbf09ca8cf55f56c62165/pyarrow-21.0.0-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:e563271e2c5ff4d4a4cbeb2c83d5cf0d4938b891518e676025f7268c6fe5fe26", size = 31196837, upload-time = "2025-07-18T00:54:34.755Z" }, + { url = "https://files.pythonhosted.org/packages/df/5f/c1c1997613abf24fceb087e79432d24c19bc6f7259cab57c2c8e5e545fab/pyarrow-21.0.0-cp310-cp310-macosx_12_0_x86_64.whl", hash = "sha256:fee33b0ca46f4c85443d6c450357101e47d53e6c3f008d658c27a2d020d44c79", size = 32659470, upload-time = "2025-07-18T00:54:38.329Z" }, + { url = "https://files.pythonhosted.org/packages/3e/ed/b1589a777816ee33ba123ba1e4f8f02243a844fed0deec97bde9fb21a5cf/pyarrow-21.0.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:7be45519b830f7c24b21d630a31d48bcebfd5d4d7f9d3bdb49da9cdf6d764edb", size = 41055619, upload-time = "2025-07-18T00:54:42.172Z" }, + { url = "https://files.pythonhosted.org/packages/44/28/b6672962639e85dc0ac36f71ab3a8f5f38e01b51343d7aa372a6b56fa3f3/pyarrow-21.0.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:26bfd95f6bff443ceae63c65dc7e048670b7e98bc892210acba7e4995d3d4b51", size = 42733488, upload-time = "2025-07-18T00:54:47.132Z" }, + { url = "https://files.pythonhosted.org/packages/f8/cc/de02c3614874b9089c94eac093f90ca5dfa6d5afe45de3ba847fd950fdf1/pyarrow-21.0.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:bd04ec08f7f8bd113c55868bd3fc442a9db67c27af098c5f814a3091e71cc61a", size = 43329159, upload-time = "2025-07-18T00:54:51.686Z" }, + { url = "https://files.pythonhosted.org/packages/a6/3e/99473332ac40278f196e105ce30b79ab8affab12f6194802f2593d6b0be2/pyarrow-21.0.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:9b0b14b49ac10654332a805aedfc0147fb3469cbf8ea951b3d040dab12372594", size = 45050567, upload-time = "2025-07-18T00:54:56.679Z" }, + { url = "https://files.pythonhosted.org/packages/7b/f5/c372ef60593d713e8bfbb7e0c743501605f0ad00719146dc075faf11172b/pyarrow-21.0.0-cp310-cp310-win_amd64.whl", hash = "sha256:9d9f8bcb4c3be7738add259738abdeddc363de1b80e3310e04067aa1ca596634", size = 26217959, upload-time = "2025-07-18T00:55:00.482Z" }, + { url = "https://files.pythonhosted.org/packages/94/dc/80564a3071a57c20b7c32575e4a0120e8a330ef487c319b122942d665960/pyarrow-21.0.0-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:c077f48aab61738c237802836fc3844f85409a46015635198761b0d6a688f87b", size = 31243234, upload-time = "2025-07-18T00:55:03.812Z" }, + { url = "https://files.pythonhosted.org/packages/ea/cc/3b51cb2db26fe535d14f74cab4c79b191ed9a8cd4cbba45e2379b5ca2746/pyarrow-21.0.0-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:689f448066781856237eca8d1975b98cace19b8dd2ab6145bf49475478bcaa10", size = 32714370, upload-time = "2025-07-18T00:55:07.495Z" }, + { url = "https://files.pythonhosted.org/packages/24/11/a4431f36d5ad7d83b87146f515c063e4d07ef0b7240876ddb885e6b44f2e/pyarrow-21.0.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:479ee41399fcddc46159a551705b89c05f11e8b8cb8e968f7fec64f62d91985e", size = 41135424, upload-time = "2025-07-18T00:55:11.461Z" }, + { url = "https://files.pythonhosted.org/packages/74/dc/035d54638fc5d2971cbf1e987ccd45f1091c83bcf747281cf6cc25e72c88/pyarrow-21.0.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:40ebfcb54a4f11bcde86bc586cbd0272bac0d516cfa539c799c2453768477569", size = 42823810, upload-time = "2025-07-18T00:55:16.301Z" }, + { url = "https://files.pythonhosted.org/packages/2e/3b/89fced102448a9e3e0d4dded1f37fa3ce4700f02cdb8665457fcc8015f5b/pyarrow-21.0.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:8d58d8497814274d3d20214fbb24abcad2f7e351474357d552a8d53bce70c70e", size = 43391538, upload-time = "2025-07-18T00:55:23.82Z" }, + { url = "https://files.pythonhosted.org/packages/fb/bb/ea7f1bd08978d39debd3b23611c293f64a642557e8141c80635d501e6d53/pyarrow-21.0.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:585e7224f21124dd57836b1530ac8f2df2afc43c861d7bf3d58a4870c42ae36c", size = 45120056, upload-time = "2025-07-18T00:55:28.231Z" }, + { url = "https://files.pythonhosted.org/packages/6e/0b/77ea0600009842b30ceebc3337639a7380cd946061b620ac1a2f3cb541e2/pyarrow-21.0.0-cp311-cp311-win_amd64.whl", hash = "sha256:555ca6935b2cbca2c0e932bedd853e9bc523098c39636de9ad4693b5b1df86d6", size = 26220568, upload-time = "2025-07-18T00:55:32.122Z" }, + { url = "https://files.pythonhosted.org/packages/ca/d4/d4f817b21aacc30195cf6a46ba041dd1be827efa4a623cc8bf39a1c2a0c0/pyarrow-21.0.0-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:3a302f0e0963db37e0a24a70c56cf91a4faa0bca51c23812279ca2e23481fccd", size = 31160305, upload-time = "2025-07-18T00:55:35.373Z" }, + { url = "https://files.pythonhosted.org/packages/a2/9c/dcd38ce6e4b4d9a19e1d36914cb8e2b1da4e6003dd075474c4cfcdfe0601/pyarrow-21.0.0-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:b6b27cf01e243871390474a211a7922bfbe3bda21e39bc9160daf0da3fe48876", size = 32684264, upload-time = "2025-07-18T00:55:39.303Z" }, + { url = "https://files.pythonhosted.org/packages/4f/74/2a2d9f8d7a59b639523454bec12dba35ae3d0a07d8ab529dc0809f74b23c/pyarrow-21.0.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:e72a8ec6b868e258a2cd2672d91f2860ad532d590ce94cdf7d5e7ec674ccf03d", size = 41108099, upload-time = "2025-07-18T00:55:42.889Z" }, + { url = "https://files.pythonhosted.org/packages/ad/90/2660332eeb31303c13b653ea566a9918484b6e4d6b9d2d46879a33ab0622/pyarrow-21.0.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:b7ae0bbdc8c6674259b25bef5d2a1d6af5d39d7200c819cf99e07f7dfef1c51e", size = 42829529, upload-time = "2025-07-18T00:55:47.069Z" }, + { url = "https://files.pythonhosted.org/packages/33/27/1a93a25c92717f6aa0fca06eb4700860577d016cd3ae51aad0e0488ac899/pyarrow-21.0.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:58c30a1729f82d201627c173d91bd431db88ea74dcaa3885855bc6203e433b82", size = 43367883, upload-time = "2025-07-18T00:55:53.069Z" }, + { url = "https://files.pythonhosted.org/packages/05/d9/4d09d919f35d599bc05c6950095e358c3e15148ead26292dfca1fb659b0c/pyarrow-21.0.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:072116f65604b822a7f22945a7a6e581cfa28e3454fdcc6939d4ff6090126623", size = 45133802, upload-time = "2025-07-18T00:55:57.714Z" }, + { url = "https://files.pythonhosted.org/packages/71/30/f3795b6e192c3ab881325ffe172e526499eb3780e306a15103a2764916a2/pyarrow-21.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:cf56ec8b0a5c8c9d7021d6fd754e688104f9ebebf1bf4449613c9531f5346a18", size = 26203175, upload-time = "2025-07-18T00:56:01.364Z" }, + { url = "https://files.pythonhosted.org/packages/16/ca/c7eaa8e62db8fb37ce942b1ea0c6d7abfe3786ca193957afa25e71b81b66/pyarrow-21.0.0-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:e99310a4ebd4479bcd1964dff9e14af33746300cb014aa4a3781738ac63baf4a", size = 31154306, upload-time = "2025-07-18T00:56:04.42Z" }, + { url = "https://files.pythonhosted.org/packages/ce/e8/e87d9e3b2489302b3a1aea709aaca4b781c5252fcb812a17ab6275a9a484/pyarrow-21.0.0-cp313-cp313-macosx_12_0_x86_64.whl", hash = "sha256:d2fe8e7f3ce329a71b7ddd7498b3cfac0eeb200c2789bd840234f0dc271a8efe", size = 32680622, upload-time = "2025-07-18T00:56:07.505Z" }, + { url = "https://files.pythonhosted.org/packages/84/52/79095d73a742aa0aba370c7942b1b655f598069489ab387fe47261a849e1/pyarrow-21.0.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:f522e5709379d72fb3da7785aa489ff0bb87448a9dc5a75f45763a795a089ebd", size = 41104094, upload-time = "2025-07-18T00:56:10.994Z" }, + { url = "https://files.pythonhosted.org/packages/89/4b/7782438b551dbb0468892a276b8c789b8bbdb25ea5c5eb27faadd753e037/pyarrow-21.0.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:69cbbdf0631396e9925e048cfa5bce4e8c3d3b41562bbd70c685a8eb53a91e61", size = 42825576, upload-time = "2025-07-18T00:56:15.569Z" }, + { url = "https://files.pythonhosted.org/packages/b3/62/0f29de6e0a1e33518dec92c65be0351d32d7ca351e51ec5f4f837a9aab91/pyarrow-21.0.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:731c7022587006b755d0bdb27626a1a3bb004bb56b11fb30d98b6c1b4718579d", size = 43368342, upload-time = "2025-07-18T00:56:19.531Z" }, + { url = "https://files.pythonhosted.org/packages/90/c7/0fa1f3f29cf75f339768cc698c8ad4ddd2481c1742e9741459911c9ac477/pyarrow-21.0.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:dc56bc708f2d8ac71bd1dcb927e458c93cec10b98eb4120206a4091db7b67b99", size = 45131218, upload-time = "2025-07-18T00:56:23.347Z" }, + { url = "https://files.pythonhosted.org/packages/01/63/581f2076465e67b23bc5a37d4a2abff8362d389d29d8105832e82c9c811c/pyarrow-21.0.0-cp313-cp313-win_amd64.whl", hash = "sha256:186aa00bca62139f75b7de8420f745f2af12941595bbbfa7ed3870ff63e25636", size = 26087551, upload-time = "2025-07-18T00:56:26.758Z" }, + { url = "https://files.pythonhosted.org/packages/c9/ab/357d0d9648bb8241ee7348e564f2479d206ebe6e1c47ac5027c2e31ecd39/pyarrow-21.0.0-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:a7a102574faa3f421141a64c10216e078df467ab9576684d5cd696952546e2da", size = 31290064, upload-time = "2025-07-18T00:56:30.214Z" }, + { url = "https://files.pythonhosted.org/packages/3f/8a/5685d62a990e4cac2043fc76b4661bf38d06efed55cf45a334b455bd2759/pyarrow-21.0.0-cp313-cp313t-macosx_12_0_x86_64.whl", hash = "sha256:1e005378c4a2c6db3ada3ad4c217b381f6c886f0a80d6a316fe586b90f77efd7", size = 32727837, upload-time = "2025-07-18T00:56:33.935Z" }, + { url = "https://files.pythonhosted.org/packages/fc/de/c0828ee09525c2bafefd3e736a248ebe764d07d0fd762d4f0929dbc516c9/pyarrow-21.0.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:65f8e85f79031449ec8706b74504a316805217b35b6099155dd7e227eef0d4b6", size = 41014158, upload-time = "2025-07-18T00:56:37.528Z" }, + { url = "https://files.pythonhosted.org/packages/6e/26/a2865c420c50b7a3748320b614f3484bfcde8347b2639b2b903b21ce6a72/pyarrow-21.0.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:3a81486adc665c7eb1a2bde0224cfca6ceaba344a82a971ef059678417880eb8", size = 42667885, upload-time = "2025-07-18T00:56:41.483Z" }, + { url = "https://files.pythonhosted.org/packages/0a/f9/4ee798dc902533159250fb4321267730bc0a107d8c6889e07c3add4fe3a5/pyarrow-21.0.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:fc0d2f88b81dcf3ccf9a6ae17f89183762c8a94a5bdcfa09e05cfe413acf0503", size = 43276625, upload-time = "2025-07-18T00:56:48.002Z" }, + { url = "https://files.pythonhosted.org/packages/5a/da/e02544d6997037a4b0d22d8e5f66bc9315c3671371a8b18c79ade1cefe14/pyarrow-21.0.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:6299449adf89df38537837487a4f8d3bd91ec94354fdd2a7d30bc11c48ef6e79", size = 44951890, upload-time = "2025-07-18T00:56:52.568Z" }, + { url = "https://files.pythonhosted.org/packages/e5/4e/519c1bc1876625fe6b71e9a28287c43ec2f20f73c658b9ae1d485c0c206e/pyarrow-21.0.0-cp313-cp313t-win_amd64.whl", hash = "sha256:222c39e2c70113543982c6b34f3077962b44fca38c0bd9e68bb6781534425c10", size = 26371006, upload-time = "2025-07-18T00:56:56.379Z" }, ] [[package]] @@ -2912,127 +2716,114 @@ wheels = [ [[package]] name = "pydantic" -version = "2.11.7" +version = "2.10.6" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "annotated-types" }, { name = "pydantic-core" }, { name = "typing-extensions" }, - { name = "typing-inspection" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/00/dd/4325abf92c39ba8623b5af936ddb36ffcfe0beae70405d456ab1fb2f5b8c/pydantic-2.11.7.tar.gz", hash = "sha256:d989c3c6cb79469287b1569f7447a17848c998458d49ebe294e975b9baf0f0db", size = 788350, upload-time = "2025-06-14T08:33:17.137Z" } +sdist = { url = "https://files.pythonhosted.org/packages/b7/ae/d5220c5c52b158b1de7ca89fc5edb72f304a70a4c540c84c8844bf4008de/pydantic-2.10.6.tar.gz", hash = "sha256:ca5daa827cce33de7a42be142548b0096bf05a7e7b365aebfa5f8eeec7128236", size = 761681, upload-time = "2025-01-24T01:42:12.693Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/6a/c0/ec2b1c8712ca690e5d61979dee872603e92b8a32f94cc1b72d53beab008a/pydantic-2.11.7-py3-none-any.whl", hash = "sha256:dde5df002701f6de26248661f6835bbe296a47bf73990135c7d07ce741b9623b", size = 444782, upload-time = "2025-06-14T08:33:14.905Z" }, + { url = "https://files.pythonhosted.org/packages/f4/3c/8cc1cc84deffa6e25d2d0c688ebb80635dfdbf1dbea3e30c541c8cf4d860/pydantic-2.10.6-py3-none-any.whl", hash = "sha256:427d664bf0b8a2b34ff5dd0f5a18df00591adcee7198fbd71981054cef37b584", size = 431696, upload-time = "2025-01-24T01:42:10.371Z" }, ] [[package]] name = "pydantic-core" -version = "2.33.2" +version = "2.27.2" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/ad/88/5f2260bdfae97aabf98f1778d43f69574390ad787afb646292a638c923d4/pydantic_core-2.33.2.tar.gz", hash = "sha256:7cb8bc3605c29176e1b105350d2e6474142d7c1bd1d9327c4a9bdb46bf827acc", size = 435195, upload-time = "2025-04-23T18:33:52.104Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/e5/92/b31726561b5dae176c2d2c2dc43a9c5bfba5d32f96f8b4c0a600dd492447/pydantic_core-2.33.2-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:2b3d326aaef0c0399d9afffeb6367d5e26ddc24d351dbc9c636840ac355dc5d8", size = 2028817, upload-time = "2025-04-23T18:30:43.919Z" }, - { url = "https://files.pythonhosted.org/packages/a3/44/3f0b95fafdaca04a483c4e685fe437c6891001bf3ce8b2fded82b9ea3aa1/pydantic_core-2.33.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:0e5b2671f05ba48b94cb90ce55d8bdcaaedb8ba00cc5359f6810fc918713983d", size = 1861357, upload-time = "2025-04-23T18:30:46.372Z" }, - { url = "https://files.pythonhosted.org/packages/30/97/e8f13b55766234caae05372826e8e4b3b96e7b248be3157f53237682e43c/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0069c9acc3f3981b9ff4cdfaf088e98d83440a4c7ea1bc07460af3d4dc22e72d", size = 1898011, upload-time = "2025-04-23T18:30:47.591Z" }, - { url = "https://files.pythonhosted.org/packages/9b/a3/99c48cf7bafc991cc3ee66fd544c0aae8dc907b752f1dad2d79b1b5a471f/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:d53b22f2032c42eaaf025f7c40c2e3b94568ae077a606f006d206a463bc69572", size = 1982730, upload-time = "2025-04-23T18:30:49.328Z" }, - { url = "https://files.pythonhosted.org/packages/de/8e/a5b882ec4307010a840fb8b58bd9bf65d1840c92eae7534c7441709bf54b/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0405262705a123b7ce9f0b92f123334d67b70fd1f20a9372b907ce1080c7ba02", size = 2136178, upload-time = "2025-04-23T18:30:50.907Z" }, - { url = "https://files.pythonhosted.org/packages/e4/bb/71e35fc3ed05af6834e890edb75968e2802fe98778971ab5cba20a162315/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:4b25d91e288e2c4e0662b8038a28c6a07eaac3e196cfc4ff69de4ea3db992a1b", size = 2736462, upload-time = "2025-04-23T18:30:52.083Z" }, - { url = "https://files.pythonhosted.org/packages/31/0d/c8f7593e6bc7066289bbc366f2235701dcbebcd1ff0ef8e64f6f239fb47d/pydantic_core-2.33.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6bdfe4b3789761f3bcb4b1ddf33355a71079858958e3a552f16d5af19768fef2", size = 2005652, upload-time = "2025-04-23T18:30:53.389Z" }, - { url = "https://files.pythonhosted.org/packages/d2/7a/996d8bd75f3eda405e3dd219ff5ff0a283cd8e34add39d8ef9157e722867/pydantic_core-2.33.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:efec8db3266b76ef9607c2c4c419bdb06bf335ae433b80816089ea7585816f6a", size = 2113306, upload-time = "2025-04-23T18:30:54.661Z" }, - { url = "https://files.pythonhosted.org/packages/ff/84/daf2a6fb2db40ffda6578a7e8c5a6e9c8affb251a05c233ae37098118788/pydantic_core-2.33.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:031c57d67ca86902726e0fae2214ce6770bbe2f710dc33063187a68744a5ecac", size = 2073720, upload-time = "2025-04-23T18:30:56.11Z" }, - { url = "https://files.pythonhosted.org/packages/77/fb/2258da019f4825128445ae79456a5499c032b55849dbd5bed78c95ccf163/pydantic_core-2.33.2-cp310-cp310-musllinux_1_1_armv7l.whl", hash = "sha256:f8de619080e944347f5f20de29a975c2d815d9ddd8be9b9b7268e2e3ef68605a", size = 2244915, upload-time = "2025-04-23T18:30:57.501Z" }, - { url = "https://files.pythonhosted.org/packages/d8/7a/925ff73756031289468326e355b6fa8316960d0d65f8b5d6b3a3e7866de7/pydantic_core-2.33.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:73662edf539e72a9440129f231ed3757faab89630d291b784ca99237fb94db2b", size = 2241884, upload-time = "2025-04-23T18:30:58.867Z" }, - { url = "https://files.pythonhosted.org/packages/0b/b0/249ee6d2646f1cdadcb813805fe76265745c4010cf20a8eba7b0e639d9b2/pydantic_core-2.33.2-cp310-cp310-win32.whl", hash = "sha256:0a39979dcbb70998b0e505fb1556a1d550a0781463ce84ebf915ba293ccb7e22", size = 1910496, upload-time = "2025-04-23T18:31:00.078Z" }, - { url = "https://files.pythonhosted.org/packages/66/ff/172ba8f12a42d4b552917aa65d1f2328990d3ccfc01d5b7c943ec084299f/pydantic_core-2.33.2-cp310-cp310-win_amd64.whl", hash = "sha256:b0379a2b24882fef529ec3b4987cb5d003b9cda32256024e6fe1586ac45fc640", size = 1955019, upload-time = "2025-04-23T18:31:01.335Z" }, - { url = "https://files.pythonhosted.org/packages/3f/8d/71db63483d518cbbf290261a1fc2839d17ff89fce7089e08cad07ccfce67/pydantic_core-2.33.2-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:4c5b0a576fb381edd6d27f0a85915c6daf2f8138dc5c267a57c08a62900758c7", size = 2028584, upload-time = "2025-04-23T18:31:03.106Z" }, - { url = "https://files.pythonhosted.org/packages/24/2f/3cfa7244ae292dd850989f328722d2aef313f74ffc471184dc509e1e4e5a/pydantic_core-2.33.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:e799c050df38a639db758c617ec771fd8fb7a5f8eaaa4b27b101f266b216a246", size = 1855071, upload-time = "2025-04-23T18:31:04.621Z" }, - { url = "https://files.pythonhosted.org/packages/b3/d3/4ae42d33f5e3f50dd467761304be2fa0a9417fbf09735bc2cce003480f2a/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:dc46a01bf8d62f227d5ecee74178ffc448ff4e5197c756331f71efcc66dc980f", size = 1897823, upload-time = "2025-04-23T18:31:06.377Z" }, - { url = "https://files.pythonhosted.org/packages/f4/f3/aa5976e8352b7695ff808599794b1fba2a9ae2ee954a3426855935799488/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:a144d4f717285c6d9234a66778059f33a89096dfb9b39117663fd8413d582dcc", size = 1983792, upload-time = "2025-04-23T18:31:07.93Z" }, - { url = "https://files.pythonhosted.org/packages/d5/7a/cda9b5a23c552037717f2b2a5257e9b2bfe45e687386df9591eff7b46d28/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:73cf6373c21bc80b2e0dc88444f41ae60b2f070ed02095754eb5a01df12256de", size = 2136338, upload-time = "2025-04-23T18:31:09.283Z" }, - { url = "https://files.pythonhosted.org/packages/2b/9f/b8f9ec8dd1417eb9da784e91e1667d58a2a4a7b7b34cf4af765ef663a7e5/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3dc625f4aa79713512d1976fe9f0bc99f706a9dee21dfd1810b4bbbf228d0e8a", size = 2730998, upload-time = "2025-04-23T18:31:11.7Z" }, - { url = "https://files.pythonhosted.org/packages/47/bc/cd720e078576bdb8255d5032c5d63ee5c0bf4b7173dd955185a1d658c456/pydantic_core-2.33.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:881b21b5549499972441da4758d662aeea93f1923f953e9cbaff14b8b9565aef", size = 2003200, upload-time = "2025-04-23T18:31:13.536Z" }, - { url = "https://files.pythonhosted.org/packages/ca/22/3602b895ee2cd29d11a2b349372446ae9727c32e78a94b3d588a40fdf187/pydantic_core-2.33.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:bdc25f3681f7b78572699569514036afe3c243bc3059d3942624e936ec93450e", size = 2113890, upload-time = "2025-04-23T18:31:15.011Z" }, - { url = "https://files.pythonhosted.org/packages/ff/e6/e3c5908c03cf00d629eb38393a98fccc38ee0ce8ecce32f69fc7d7b558a7/pydantic_core-2.33.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:fe5b32187cbc0c862ee201ad66c30cf218e5ed468ec8dc1cf49dec66e160cc4d", size = 2073359, upload-time = "2025-04-23T18:31:16.393Z" }, - { url = "https://files.pythonhosted.org/packages/12/e7/6a36a07c59ebefc8777d1ffdaf5ae71b06b21952582e4b07eba88a421c79/pydantic_core-2.33.2-cp311-cp311-musllinux_1_1_armv7l.whl", hash = "sha256:bc7aee6f634a6f4a95676fcb5d6559a2c2a390330098dba5e5a5f28a2e4ada30", size = 2245883, upload-time = "2025-04-23T18:31:17.892Z" }, - { url = "https://files.pythonhosted.org/packages/16/3f/59b3187aaa6cc0c1e6616e8045b284de2b6a87b027cce2ffcea073adf1d2/pydantic_core-2.33.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:235f45e5dbcccf6bd99f9f472858849f73d11120d76ea8707115415f8e5ebebf", size = 2241074, upload-time = "2025-04-23T18:31:19.205Z" }, - { url = "https://files.pythonhosted.org/packages/e0/ed/55532bb88f674d5d8f67ab121a2a13c385df382de2a1677f30ad385f7438/pydantic_core-2.33.2-cp311-cp311-win32.whl", hash = "sha256:6368900c2d3ef09b69cb0b913f9f8263b03786e5b2a387706c5afb66800efd51", size = 1910538, upload-time = "2025-04-23T18:31:20.541Z" }, - { url = "https://files.pythonhosted.org/packages/fe/1b/25b7cccd4519c0b23c2dd636ad39d381abf113085ce4f7bec2b0dc755eb1/pydantic_core-2.33.2-cp311-cp311-win_amd64.whl", hash = "sha256:1e063337ef9e9820c77acc768546325ebe04ee38b08703244c1309cccc4f1bab", size = 1952909, upload-time = "2025-04-23T18:31:22.371Z" }, - { url = "https://files.pythonhosted.org/packages/49/a9/d809358e49126438055884c4366a1f6227f0f84f635a9014e2deb9b9de54/pydantic_core-2.33.2-cp311-cp311-win_arm64.whl", hash = "sha256:6b99022f1d19bc32a4c2a0d544fc9a76e3be90f0b3f4af413f87d38749300e65", size = 1897786, upload-time = "2025-04-23T18:31:24.161Z" }, - { url = "https://files.pythonhosted.org/packages/18/8a/2b41c97f554ec8c71f2a8a5f85cb56a8b0956addfe8b0efb5b3d77e8bdc3/pydantic_core-2.33.2-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:a7ec89dc587667f22b6a0b6579c249fca9026ce7c333fc142ba42411fa243cdc", size = 2009000, upload-time = "2025-04-23T18:31:25.863Z" }, - { url = "https://files.pythonhosted.org/packages/a1/02/6224312aacb3c8ecbaa959897af57181fb6cf3a3d7917fd44d0f2917e6f2/pydantic_core-2.33.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3c6db6e52c6d70aa0d00d45cdb9b40f0433b96380071ea80b09277dba021ddf7", size = 1847996, upload-time = "2025-04-23T18:31:27.341Z" }, - { url = "https://files.pythonhosted.org/packages/d6/46/6dcdf084a523dbe0a0be59d054734b86a981726f221f4562aed313dbcb49/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4e61206137cbc65e6d5256e1166f88331d3b6238e082d9f74613b9b765fb9025", size = 1880957, upload-time = "2025-04-23T18:31:28.956Z" }, - { url = "https://files.pythonhosted.org/packages/ec/6b/1ec2c03837ac00886ba8160ce041ce4e325b41d06a034adbef11339ae422/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:eb8c529b2819c37140eb51b914153063d27ed88e3bdc31b71198a198e921e011", size = 1964199, upload-time = "2025-04-23T18:31:31.025Z" }, - { url = "https://files.pythonhosted.org/packages/2d/1d/6bf34d6adb9debd9136bd197ca72642203ce9aaaa85cfcbfcf20f9696e83/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c52b02ad8b4e2cf14ca7b3d918f3eb0ee91e63b3167c32591e57c4317e134f8f", size = 2120296, upload-time = "2025-04-23T18:31:32.514Z" }, - { url = "https://files.pythonhosted.org/packages/e0/94/2bd0aaf5a591e974b32a9f7123f16637776c304471a0ab33cf263cf5591a/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:96081f1605125ba0855dfda83f6f3df5ec90c61195421ba72223de35ccfb2f88", size = 2676109, upload-time = "2025-04-23T18:31:33.958Z" }, - { url = "https://files.pythonhosted.org/packages/f9/41/4b043778cf9c4285d59742281a769eac371b9e47e35f98ad321349cc5d61/pydantic_core-2.33.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:8f57a69461af2a5fa6e6bbd7a5f60d3b7e6cebb687f55106933188e79ad155c1", size = 2002028, upload-time = "2025-04-23T18:31:39.095Z" }, - { url = "https://files.pythonhosted.org/packages/cb/d5/7bb781bf2748ce3d03af04d5c969fa1308880e1dca35a9bd94e1a96a922e/pydantic_core-2.33.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:572c7e6c8bb4774d2ac88929e3d1f12bc45714ae5ee6d9a788a9fb35e60bb04b", size = 2100044, upload-time = "2025-04-23T18:31:41.034Z" }, - { url = "https://files.pythonhosted.org/packages/fe/36/def5e53e1eb0ad896785702a5bbfd25eed546cdcf4087ad285021a90ed53/pydantic_core-2.33.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:db4b41f9bd95fbe5acd76d89920336ba96f03e149097365afe1cb092fceb89a1", size = 2058881, upload-time = "2025-04-23T18:31:42.757Z" }, - { url = "https://files.pythonhosted.org/packages/01/6c/57f8d70b2ee57fc3dc8b9610315949837fa8c11d86927b9bb044f8705419/pydantic_core-2.33.2-cp312-cp312-musllinux_1_1_armv7l.whl", hash = "sha256:fa854f5cf7e33842a892e5c73f45327760bc7bc516339fda888c75ae60edaeb6", size = 2227034, upload-time = "2025-04-23T18:31:44.304Z" }, - { url = "https://files.pythonhosted.org/packages/27/b9/9c17f0396a82b3d5cbea4c24d742083422639e7bb1d5bf600e12cb176a13/pydantic_core-2.33.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:5f483cfb75ff703095c59e365360cb73e00185e01aaea067cd19acffd2ab20ea", size = 2234187, upload-time = "2025-04-23T18:31:45.891Z" }, - { url = "https://files.pythonhosted.org/packages/b0/6a/adf5734ffd52bf86d865093ad70b2ce543415e0e356f6cacabbc0d9ad910/pydantic_core-2.33.2-cp312-cp312-win32.whl", hash = "sha256:9cb1da0f5a471435a7bc7e439b8a728e8b61e59784b2af70d7c169f8dd8ae290", size = 1892628, upload-time = "2025-04-23T18:31:47.819Z" }, - { url = "https://files.pythonhosted.org/packages/43/e4/5479fecb3606c1368d496a825d8411e126133c41224c1e7238be58b87d7e/pydantic_core-2.33.2-cp312-cp312-win_amd64.whl", hash = "sha256:f941635f2a3d96b2973e867144fde513665c87f13fe0e193c158ac51bfaaa7b2", size = 1955866, upload-time = "2025-04-23T18:31:49.635Z" }, - { url = "https://files.pythonhosted.org/packages/0d/24/8b11e8b3e2be9dd82df4b11408a67c61bb4dc4f8e11b5b0fc888b38118b5/pydantic_core-2.33.2-cp312-cp312-win_arm64.whl", hash = "sha256:cca3868ddfaccfbc4bfb1d608e2ccaaebe0ae628e1416aeb9c4d88c001bb45ab", size = 1888894, upload-time = "2025-04-23T18:31:51.609Z" }, - { url = "https://files.pythonhosted.org/packages/46/8c/99040727b41f56616573a28771b1bfa08a3d3fe74d3d513f01251f79f172/pydantic_core-2.33.2-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:1082dd3e2d7109ad8b7da48e1d4710c8d06c253cbc4a27c1cff4fbcaa97a9e3f", size = 2015688, upload-time = "2025-04-23T18:31:53.175Z" }, - { url = "https://files.pythonhosted.org/packages/3a/cc/5999d1eb705a6cefc31f0b4a90e9f7fc400539b1a1030529700cc1b51838/pydantic_core-2.33.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:f517ca031dfc037a9c07e748cefd8d96235088b83b4f4ba8939105d20fa1dcd6", size = 1844808, upload-time = "2025-04-23T18:31:54.79Z" }, - { url = "https://files.pythonhosted.org/packages/6f/5e/a0a7b8885c98889a18b6e376f344da1ef323d270b44edf8174d6bce4d622/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0a9f2c9dd19656823cb8250b0724ee9c60a82f3cdf68a080979d13092a3b0fef", size = 1885580, upload-time = "2025-04-23T18:31:57.393Z" }, - { url = "https://files.pythonhosted.org/packages/3b/2a/953581f343c7d11a304581156618c3f592435523dd9d79865903272c256a/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:2b0a451c263b01acebe51895bfb0e1cc842a5c666efe06cdf13846c7418caa9a", size = 1973859, upload-time = "2025-04-23T18:31:59.065Z" }, - { url = "https://files.pythonhosted.org/packages/e6/55/f1a813904771c03a3f97f676c62cca0c0a4138654107c1b61f19c644868b/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1ea40a64d23faa25e62a70ad163571c0b342b8bf66d5fa612ac0dec4f069d916", size = 2120810, upload-time = "2025-04-23T18:32:00.78Z" }, - { url = "https://files.pythonhosted.org/packages/aa/c3/053389835a996e18853ba107a63caae0b9deb4a276c6b472931ea9ae6e48/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0fb2d542b4d66f9470e8065c5469ec676978d625a8b7a363f07d9a501a9cb36a", size = 2676498, upload-time = "2025-04-23T18:32:02.418Z" }, - { url = "https://files.pythonhosted.org/packages/eb/3c/f4abd740877a35abade05e437245b192f9d0ffb48bbbbd708df33d3cda37/pydantic_core-2.33.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9fdac5d6ffa1b5a83bca06ffe7583f5576555e6c8b3a91fbd25ea7780f825f7d", size = 2000611, upload-time = "2025-04-23T18:32:04.152Z" }, - { url = "https://files.pythonhosted.org/packages/59/a7/63ef2fed1837d1121a894d0ce88439fe3e3b3e48c7543b2a4479eb99c2bd/pydantic_core-2.33.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:04a1a413977ab517154eebb2d326da71638271477d6ad87a769102f7c2488c56", size = 2107924, upload-time = "2025-04-23T18:32:06.129Z" }, - { url = "https://files.pythonhosted.org/packages/04/8f/2551964ef045669801675f1cfc3b0d74147f4901c3ffa42be2ddb1f0efc4/pydantic_core-2.33.2-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:c8e7af2f4e0194c22b5b37205bfb293d166a7344a5b0d0eaccebc376546d77d5", size = 2063196, upload-time = "2025-04-23T18:32:08.178Z" }, - { url = "https://files.pythonhosted.org/packages/26/bd/d9602777e77fc6dbb0c7db9ad356e9a985825547dce5ad1d30ee04903918/pydantic_core-2.33.2-cp313-cp313-musllinux_1_1_armv7l.whl", hash = "sha256:5c92edd15cd58b3c2d34873597a1e20f13094f59cf88068adb18947df5455b4e", size = 2236389, upload-time = "2025-04-23T18:32:10.242Z" }, - { url = "https://files.pythonhosted.org/packages/42/db/0e950daa7e2230423ab342ae918a794964b053bec24ba8af013fc7c94846/pydantic_core-2.33.2-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:65132b7b4a1c0beded5e057324b7e16e10910c106d43675d9bd87d4f38dde162", size = 2239223, upload-time = "2025-04-23T18:32:12.382Z" }, - { url = "https://files.pythonhosted.org/packages/58/4d/4f937099c545a8a17eb52cb67fe0447fd9a373b348ccfa9a87f141eeb00f/pydantic_core-2.33.2-cp313-cp313-win32.whl", hash = "sha256:52fb90784e0a242bb96ec53f42196a17278855b0f31ac7c3cc6f5c1ec4811849", size = 1900473, upload-time = "2025-04-23T18:32:14.034Z" }, - { url = "https://files.pythonhosted.org/packages/a0/75/4a0a9bac998d78d889def5e4ef2b065acba8cae8c93696906c3a91f310ca/pydantic_core-2.33.2-cp313-cp313-win_amd64.whl", hash = "sha256:c083a3bdd5a93dfe480f1125926afcdbf2917ae714bdb80b36d34318b2bec5d9", size = 1955269, upload-time = "2025-04-23T18:32:15.783Z" }, - { url = "https://files.pythonhosted.org/packages/f9/86/1beda0576969592f1497b4ce8e7bc8cbdf614c352426271b1b10d5f0aa64/pydantic_core-2.33.2-cp313-cp313-win_arm64.whl", hash = "sha256:e80b087132752f6b3d714f041ccf74403799d3b23a72722ea2e6ba2e892555b9", size = 1893921, upload-time = "2025-04-23T18:32:18.473Z" }, - { url = "https://files.pythonhosted.org/packages/a4/7d/e09391c2eebeab681df2b74bfe6c43422fffede8dc74187b2b0bf6fd7571/pydantic_core-2.33.2-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:61c18fba8e5e9db3ab908620af374db0ac1baa69f0f32df4f61ae23f15e586ac", size = 1806162, upload-time = "2025-04-23T18:32:20.188Z" }, - { url = "https://files.pythonhosted.org/packages/f1/3d/847b6b1fed9f8ed3bb95a9ad04fbd0b212e832d4f0f50ff4d9ee5a9f15cf/pydantic_core-2.33.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:95237e53bb015f67b63c91af7518a62a8660376a6a0db19b89acc77a4d6199f5", size = 1981560, upload-time = "2025-04-23T18:32:22.354Z" }, - { url = "https://files.pythonhosted.org/packages/6f/9a/e73262f6c6656262b5fdd723ad90f518f579b7bc8622e43a942eec53c938/pydantic_core-2.33.2-cp313-cp313t-win_amd64.whl", hash = "sha256:c2fc0a768ef76c15ab9238afa6da7f69895bb5d1ee83aeea2e3509af4472d0b9", size = 1935777, upload-time = "2025-04-23T18:32:25.088Z" }, - { url = "https://files.pythonhosted.org/packages/30/68/373d55e58b7e83ce371691f6eaa7175e3a24b956c44628eb25d7da007917/pydantic_core-2.33.2-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:5c4aa4e82353f65e548c476b37e64189783aa5384903bfea4f41580f255fddfa", size = 2023982, upload-time = "2025-04-23T18:32:53.14Z" }, - { url = "https://files.pythonhosted.org/packages/a4/16/145f54ac08c96a63d8ed6442f9dec17b2773d19920b627b18d4f10a061ea/pydantic_core-2.33.2-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:d946c8bf0d5c24bf4fe333af284c59a19358aa3ec18cb3dc4370080da1e8ad29", size = 1858412, upload-time = "2025-04-23T18:32:55.52Z" }, - { url = "https://files.pythonhosted.org/packages/41/b1/c6dc6c3e2de4516c0bb2c46f6a373b91b5660312342a0cf5826e38ad82fa/pydantic_core-2.33.2-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:87b31b6846e361ef83fedb187bb5b4372d0da3f7e28d85415efa92d6125d6e6d", size = 1892749, upload-time = "2025-04-23T18:32:57.546Z" }, - { url = "https://files.pythonhosted.org/packages/12/73/8cd57e20afba760b21b742106f9dbdfa6697f1570b189c7457a1af4cd8a0/pydantic_core-2.33.2-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:aa9d91b338f2df0508606f7009fde642391425189bba6d8c653afd80fd6bb64e", size = 2067527, upload-time = "2025-04-23T18:32:59.771Z" }, - { url = "https://files.pythonhosted.org/packages/e3/d5/0bb5d988cc019b3cba4a78f2d4b3854427fc47ee8ec8e9eaabf787da239c/pydantic_core-2.33.2-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:2058a32994f1fde4ca0480ab9d1e75a0e8c87c22b53a3ae66554f9af78f2fe8c", size = 2108225, upload-time = "2025-04-23T18:33:04.51Z" }, - { url = "https://files.pythonhosted.org/packages/f1/c5/00c02d1571913d496aabf146106ad8239dc132485ee22efe08085084ff7c/pydantic_core-2.33.2-pp310-pypy310_pp73-musllinux_1_1_aarch64.whl", hash = "sha256:0e03262ab796d986f978f79c943fc5f620381be7287148b8010b4097f79a39ec", size = 2069490, upload-time = "2025-04-23T18:33:06.391Z" }, - { url = "https://files.pythonhosted.org/packages/22/a8/dccc38768274d3ed3a59b5d06f59ccb845778687652daa71df0cab4040d7/pydantic_core-2.33.2-pp310-pypy310_pp73-musllinux_1_1_armv7l.whl", hash = "sha256:1a8695a8d00c73e50bff9dfda4d540b7dee29ff9b8053e38380426a85ef10052", size = 2237525, upload-time = "2025-04-23T18:33:08.44Z" }, - { url = "https://files.pythonhosted.org/packages/d4/e7/4f98c0b125dda7cf7ccd14ba936218397b44f50a56dd8c16a3091df116c3/pydantic_core-2.33.2-pp310-pypy310_pp73-musllinux_1_1_x86_64.whl", hash = "sha256:fa754d1850735a0b0e03bcffd9d4b4343eb417e47196e4485d9cca326073a42c", size = 2238446, upload-time = "2025-04-23T18:33:10.313Z" }, - { url = "https://files.pythonhosted.org/packages/ce/91/2ec36480fdb0b783cd9ef6795753c1dea13882f2e68e73bce76ae8c21e6a/pydantic_core-2.33.2-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:a11c8d26a50bfab49002947d3d237abe4d9e4b5bdc8846a63537b6488e197808", size = 2066678, upload-time = "2025-04-23T18:33:12.224Z" }, - { url = "https://files.pythonhosted.org/packages/7b/27/d4ae6487d73948d6f20dddcd94be4ea43e74349b56eba82e9bdee2d7494c/pydantic_core-2.33.2-pp311-pypy311_pp73-macosx_10_12_x86_64.whl", hash = "sha256:dd14041875d09cc0f9308e37a6f8b65f5585cf2598a53aa0123df8b129d481f8", size = 2025200, upload-time = "2025-04-23T18:33:14.199Z" }, - { url = "https://files.pythonhosted.org/packages/f1/b8/b3cb95375f05d33801024079b9392a5ab45267a63400bf1866e7ce0f0de4/pydantic_core-2.33.2-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:d87c561733f66531dced0da6e864f44ebf89a8fba55f31407b00c2f7f9449593", size = 1859123, upload-time = "2025-04-23T18:33:16.555Z" }, - { url = "https://files.pythonhosted.org/packages/05/bc/0d0b5adeda59a261cd30a1235a445bf55c7e46ae44aea28f7bd6ed46e091/pydantic_core-2.33.2-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2f82865531efd18d6e07a04a17331af02cb7a651583c418df8266f17a63c6612", size = 1892852, upload-time = "2025-04-23T18:33:18.513Z" }, - { url = "https://files.pythonhosted.org/packages/3e/11/d37bdebbda2e449cb3f519f6ce950927b56d62f0b84fd9cb9e372a26a3d5/pydantic_core-2.33.2-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2bfb5112df54209d820d7bf9317c7a6c9025ea52e49f46b6a2060104bba37de7", size = 2067484, upload-time = "2025-04-23T18:33:20.475Z" }, - { url = "https://files.pythonhosted.org/packages/8c/55/1f95f0a05ce72ecb02a8a8a1c3be0579bbc29b1d5ab68f1378b7bebc5057/pydantic_core-2.33.2-pp311-pypy311_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:64632ff9d614e5eecfb495796ad51b0ed98c453e447a76bcbeeb69615079fc7e", size = 2108896, upload-time = "2025-04-23T18:33:22.501Z" }, - { url = "https://files.pythonhosted.org/packages/53/89/2b2de6c81fa131f423246a9109d7b2a375e83968ad0800d6e57d0574629b/pydantic_core-2.33.2-pp311-pypy311_pp73-musllinux_1_1_aarch64.whl", hash = "sha256:f889f7a40498cc077332c7ab6b4608d296d852182211787d4f3ee377aaae66e8", size = 2069475, upload-time = "2025-04-23T18:33:24.528Z" }, - { url = "https://files.pythonhosted.org/packages/b8/e9/1f7efbe20d0b2b10f6718944b5d8ece9152390904f29a78e68d4e7961159/pydantic_core-2.33.2-pp311-pypy311_pp73-musllinux_1_1_armv7l.whl", hash = "sha256:de4b83bb311557e439b9e186f733f6c645b9417c84e2eb8203f3f820a4b988bf", size = 2239013, upload-time = "2025-04-23T18:33:26.621Z" }, - { url = "https://files.pythonhosted.org/packages/3c/b2/5309c905a93811524a49b4e031e9851a6b00ff0fb668794472ea7746b448/pydantic_core-2.33.2-pp311-pypy311_pp73-musllinux_1_1_x86_64.whl", hash = "sha256:82f68293f055f51b51ea42fafc74b6aad03e70e191799430b90c13d643059ebb", size = 2238715, upload-time = "2025-04-23T18:33:28.656Z" }, - { url = "https://files.pythonhosted.org/packages/32/56/8a7ca5d2cd2cda1d245d34b1c9a942920a718082ae8e54e5f3e5a58b7add/pydantic_core-2.33.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:329467cecfb529c925cf2bbd4d60d2c509bc2fb52a20c1045bf09bb70971a9c1", size = 2066757, upload-time = "2025-04-23T18:33:30.645Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/fc/01/f3e5ac5e7c25833db5eb555f7b7ab24cd6f8c322d3a3ad2d67a952dc0abc/pydantic_core-2.27.2.tar.gz", hash = "sha256:eb026e5a4c1fee05726072337ff51d1efb6f59090b7da90d30ea58625b1ffb39", size = 413443, upload-time = "2024-12-18T11:31:54.917Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/3a/bc/fed5f74b5d802cf9a03e83f60f18864e90e3aed7223adaca5ffb7a8d8d64/pydantic_core-2.27.2-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:2d367ca20b2f14095a8f4fa1210f5a7b78b8a20009ecced6b12818f455b1e9fa", size = 1895938, upload-time = "2024-12-18T11:27:14.406Z" }, + { url = "https://files.pythonhosted.org/packages/71/2a/185aff24ce844e39abb8dd680f4e959f0006944f4a8a0ea372d9f9ae2e53/pydantic_core-2.27.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:491a2b73db93fab69731eaee494f320faa4e093dbed776be1a829c2eb222c34c", size = 1815684, upload-time = "2024-12-18T11:27:16.489Z" }, + { url = "https://files.pythonhosted.org/packages/c3/43/fafabd3d94d159d4f1ed62e383e264f146a17dd4d48453319fd782e7979e/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7969e133a6f183be60e9f6f56bfae753585680f3b7307a8e555a948d443cc05a", size = 1829169, upload-time = "2024-12-18T11:27:22.16Z" }, + { url = "https://files.pythonhosted.org/packages/a2/d1/f2dfe1a2a637ce6800b799aa086d079998959f6f1215eb4497966efd2274/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:3de9961f2a346257caf0aa508a4da705467f53778e9ef6fe744c038119737ef5", size = 1867227, upload-time = "2024-12-18T11:27:25.097Z" }, + { url = "https://files.pythonhosted.org/packages/7d/39/e06fcbcc1c785daa3160ccf6c1c38fea31f5754b756e34b65f74e99780b5/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e2bb4d3e5873c37bb3dd58714d4cd0b0e6238cebc4177ac8fe878f8b3aa8e74c", size = 2037695, upload-time = "2024-12-18T11:27:28.656Z" }, + { url = "https://files.pythonhosted.org/packages/7a/67/61291ee98e07f0650eb756d44998214231f50751ba7e13f4f325d95249ab/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:280d219beebb0752699480fe8f1dc61ab6615c2046d76b7ab7ee38858de0a4e7", size = 2741662, upload-time = "2024-12-18T11:27:30.798Z" }, + { url = "https://files.pythonhosted.org/packages/32/90/3b15e31b88ca39e9e626630b4c4a1f5a0dfd09076366f4219429e6786076/pydantic_core-2.27.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:47956ae78b6422cbd46f772f1746799cbb862de838fd8d1fbd34a82e05b0983a", size = 1993370, upload-time = "2024-12-18T11:27:33.692Z" }, + { url = "https://files.pythonhosted.org/packages/ff/83/c06d333ee3a67e2e13e07794995c1535565132940715931c1c43bfc85b11/pydantic_core-2.27.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:14d4a5c49d2f009d62a2a7140d3064f686d17a5d1a268bc641954ba181880236", size = 1996813, upload-time = "2024-12-18T11:27:37.111Z" }, + { url = "https://files.pythonhosted.org/packages/7c/f7/89be1c8deb6e22618a74f0ca0d933fdcb8baa254753b26b25ad3acff8f74/pydantic_core-2.27.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:337b443af21d488716f8d0b6164de833e788aa6bd7e3a39c005febc1284f4962", size = 2005287, upload-time = "2024-12-18T11:27:40.566Z" }, + { url = "https://files.pythonhosted.org/packages/b7/7d/8eb3e23206c00ef7feee17b83a4ffa0a623eb1a9d382e56e4aa46fd15ff2/pydantic_core-2.27.2-cp310-cp310-musllinux_1_1_armv7l.whl", hash = "sha256:03d0f86ea3184a12f41a2d23f7ccb79cdb5a18e06993f8a45baa8dfec746f0e9", size = 2128414, upload-time = "2024-12-18T11:27:43.757Z" }, + { url = "https://files.pythonhosted.org/packages/4e/99/fe80f3ff8dd71a3ea15763878d464476e6cb0a2db95ff1c5c554133b6b83/pydantic_core-2.27.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:7041c36f5680c6e0f08d922aed302e98b3745d97fe1589db0a3eebf6624523af", size = 2155301, upload-time = "2024-12-18T11:27:47.36Z" }, + { url = "https://files.pythonhosted.org/packages/2b/a3/e50460b9a5789ca1451b70d4f52546fa9e2b420ba3bfa6100105c0559238/pydantic_core-2.27.2-cp310-cp310-win32.whl", hash = "sha256:50a68f3e3819077be2c98110c1f9dcb3817e93f267ba80a2c05bb4f8799e2ff4", size = 1816685, upload-time = "2024-12-18T11:27:50.508Z" }, + { url = "https://files.pythonhosted.org/packages/57/4c/a8838731cb0f2c2a39d3535376466de6049034d7b239c0202a64aaa05533/pydantic_core-2.27.2-cp310-cp310-win_amd64.whl", hash = "sha256:e0fd26b16394ead34a424eecf8a31a1f5137094cabe84a1bcb10fa6ba39d3d31", size = 1982876, upload-time = "2024-12-18T11:27:53.54Z" }, + { url = "https://files.pythonhosted.org/packages/c2/89/f3450af9d09d44eea1f2c369f49e8f181d742f28220f88cc4dfaae91ea6e/pydantic_core-2.27.2-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:8e10c99ef58cfdf2a66fc15d66b16c4a04f62bca39db589ae8cba08bc55331bc", size = 1893421, upload-time = "2024-12-18T11:27:55.409Z" }, + { url = "https://files.pythonhosted.org/packages/9e/e3/71fe85af2021f3f386da42d291412e5baf6ce7716bd7101ea49c810eda90/pydantic_core-2.27.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:26f32e0adf166a84d0cb63be85c562ca8a6fa8de28e5f0d92250c6b7e9e2aff7", size = 1814998, upload-time = "2024-12-18T11:27:57.252Z" }, + { url = "https://files.pythonhosted.org/packages/a6/3c/724039e0d848fd69dbf5806894e26479577316c6f0f112bacaf67aa889ac/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8c19d1ea0673cd13cc2f872f6c9ab42acc4e4f492a7ca9d3795ce2b112dd7e15", size = 1826167, upload-time = "2024-12-18T11:27:59.146Z" }, + { url = "https://files.pythonhosted.org/packages/2b/5b/1b29e8c1fb5f3199a9a57c1452004ff39f494bbe9bdbe9a81e18172e40d3/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:5e68c4446fe0810e959cdff46ab0a41ce2f2c86d227d96dc3847af0ba7def306", size = 1865071, upload-time = "2024-12-18T11:28:02.625Z" }, + { url = "https://files.pythonhosted.org/packages/89/6c/3985203863d76bb7d7266e36970d7e3b6385148c18a68cc8915fd8c84d57/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d9640b0059ff4f14d1f37321b94061c6db164fbe49b334b31643e0528d100d99", size = 2036244, upload-time = "2024-12-18T11:28:04.442Z" }, + { url = "https://files.pythonhosted.org/packages/0e/41/f15316858a246b5d723f7d7f599f79e37493b2e84bfc789e58d88c209f8a/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:40d02e7d45c9f8af700f3452f329ead92da4c5f4317ca9b896de7ce7199ea459", size = 2737470, upload-time = "2024-12-18T11:28:07.679Z" }, + { url = "https://files.pythonhosted.org/packages/a8/7c/b860618c25678bbd6d1d99dbdfdf0510ccb50790099b963ff78a124b754f/pydantic_core-2.27.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1c1fd185014191700554795c99b347d64f2bb637966c4cfc16998a0ca700d048", size = 1992291, upload-time = "2024-12-18T11:28:10.297Z" }, + { url = "https://files.pythonhosted.org/packages/bf/73/42c3742a391eccbeab39f15213ecda3104ae8682ba3c0c28069fbcb8c10d/pydantic_core-2.27.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:d81d2068e1c1228a565af076598f9e7451712700b673de8f502f0334f281387d", size = 1994613, upload-time = "2024-12-18T11:28:13.362Z" }, + { url = "https://files.pythonhosted.org/packages/94/7a/941e89096d1175d56f59340f3a8ebaf20762fef222c298ea96d36a6328c5/pydantic_core-2.27.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:1a4207639fb02ec2dbb76227d7c751a20b1a6b4bc52850568e52260cae64ca3b", size = 2002355, upload-time = "2024-12-18T11:28:16.587Z" }, + { url = "https://files.pythonhosted.org/packages/6e/95/2359937a73d49e336a5a19848713555605d4d8d6940c3ec6c6c0ca4dcf25/pydantic_core-2.27.2-cp311-cp311-musllinux_1_1_armv7l.whl", hash = "sha256:3de3ce3c9ddc8bbd88f6e0e304dea0e66d843ec9de1b0042b0911c1663ffd474", size = 2126661, upload-time = "2024-12-18T11:28:18.407Z" }, + { url = "https://files.pythonhosted.org/packages/2b/4c/ca02b7bdb6012a1adef21a50625b14f43ed4d11f1fc237f9d7490aa5078c/pydantic_core-2.27.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:30c5f68ded0c36466acede341551106821043e9afaad516adfb6e8fa80a4e6a6", size = 2153261, upload-time = "2024-12-18T11:28:21.471Z" }, + { url = "https://files.pythonhosted.org/packages/72/9d/a241db83f973049a1092a079272ffe2e3e82e98561ef6214ab53fe53b1c7/pydantic_core-2.27.2-cp311-cp311-win32.whl", hash = "sha256:c70c26d2c99f78b125a3459f8afe1aed4d9687c24fd677c6a4436bc042e50d6c", size = 1812361, upload-time = "2024-12-18T11:28:23.53Z" }, + { url = "https://files.pythonhosted.org/packages/e8/ef/013f07248041b74abd48a385e2110aa3a9bbfef0fbd97d4e6d07d2f5b89a/pydantic_core-2.27.2-cp311-cp311-win_amd64.whl", hash = "sha256:08e125dbdc505fa69ca7d9c499639ab6407cfa909214d500897d02afb816e7cc", size = 1982484, upload-time = "2024-12-18T11:28:25.391Z" }, + { url = "https://files.pythonhosted.org/packages/10/1c/16b3a3e3398fd29dca77cea0a1d998d6bde3902fa2706985191e2313cc76/pydantic_core-2.27.2-cp311-cp311-win_arm64.whl", hash = "sha256:26f0d68d4b235a2bae0c3fc585c585b4ecc51382db0e3ba402a22cbc440915e4", size = 1867102, upload-time = "2024-12-18T11:28:28.593Z" }, + { url = "https://files.pythonhosted.org/packages/d6/74/51c8a5482ca447871c93e142d9d4a92ead74de6c8dc5e66733e22c9bba89/pydantic_core-2.27.2-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:9e0c8cfefa0ef83b4da9588448b6d8d2a2bf1a53c3f1ae5fca39eb3061e2f0b0", size = 1893127, upload-time = "2024-12-18T11:28:30.346Z" }, + { url = "https://files.pythonhosted.org/packages/d3/f3/c97e80721735868313c58b89d2de85fa80fe8dfeeed84dc51598b92a135e/pydantic_core-2.27.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:83097677b8e3bd7eaa6775720ec8e0405f1575015a463285a92bfdfe254529ef", size = 1811340, upload-time = "2024-12-18T11:28:32.521Z" }, + { url = "https://files.pythonhosted.org/packages/9e/91/840ec1375e686dbae1bd80a9e46c26a1e0083e1186abc610efa3d9a36180/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:172fce187655fece0c90d90a678424b013f8fbb0ca8b036ac266749c09438cb7", size = 1822900, upload-time = "2024-12-18T11:28:34.507Z" }, + { url = "https://files.pythonhosted.org/packages/f6/31/4240bc96025035500c18adc149aa6ffdf1a0062a4b525c932065ceb4d868/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:519f29f5213271eeeeb3093f662ba2fd512b91c5f188f3bb7b27bc5973816934", size = 1869177, upload-time = "2024-12-18T11:28:36.488Z" }, + { url = "https://files.pythonhosted.org/packages/fa/20/02fbaadb7808be578317015c462655c317a77a7c8f0ef274bc016a784c54/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:05e3a55d124407fffba0dd6b0c0cd056d10e983ceb4e5dbd10dda135c31071d6", size = 2038046, upload-time = "2024-12-18T11:28:39.409Z" }, + { url = "https://files.pythonhosted.org/packages/06/86/7f306b904e6c9eccf0668248b3f272090e49c275bc488a7b88b0823444a4/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9c3ed807c7b91de05e63930188f19e921d1fe90de6b4f5cd43ee7fcc3525cb8c", size = 2685386, upload-time = "2024-12-18T11:28:41.221Z" }, + { url = "https://files.pythonhosted.org/packages/8d/f0/49129b27c43396581a635d8710dae54a791b17dfc50c70164866bbf865e3/pydantic_core-2.27.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6fb4aadc0b9a0c063206846d603b92030eb6f03069151a625667f982887153e2", size = 1997060, upload-time = "2024-12-18T11:28:44.709Z" }, + { url = "https://files.pythonhosted.org/packages/0d/0f/943b4af7cd416c477fd40b187036c4f89b416a33d3cc0ab7b82708a667aa/pydantic_core-2.27.2-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:28ccb213807e037460326424ceb8b5245acb88f32f3d2777427476e1b32c48c4", size = 2004870, upload-time = "2024-12-18T11:28:46.839Z" }, + { url = "https://files.pythonhosted.org/packages/35/40/aea70b5b1a63911c53a4c8117c0a828d6790483f858041f47bab0b779f44/pydantic_core-2.27.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:de3cd1899e2c279b140adde9357c4495ed9d47131b4a4eaff9052f23398076b3", size = 1999822, upload-time = "2024-12-18T11:28:48.896Z" }, + { url = "https://files.pythonhosted.org/packages/f2/b3/807b94fd337d58effc5498fd1a7a4d9d59af4133e83e32ae39a96fddec9d/pydantic_core-2.27.2-cp312-cp312-musllinux_1_1_armv7l.whl", hash = "sha256:220f892729375e2d736b97d0e51466252ad84c51857d4d15f5e9692f9ef12be4", size = 2130364, upload-time = "2024-12-18T11:28:50.755Z" }, + { url = "https://files.pythonhosted.org/packages/fc/df/791c827cd4ee6efd59248dca9369fb35e80a9484462c33c6649a8d02b565/pydantic_core-2.27.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:a0fcd29cd6b4e74fe8ddd2c90330fd8edf2e30cb52acda47f06dd615ae72da57", size = 2158303, upload-time = "2024-12-18T11:28:54.122Z" }, + { url = "https://files.pythonhosted.org/packages/9b/67/4e197c300976af185b7cef4c02203e175fb127e414125916bf1128b639a9/pydantic_core-2.27.2-cp312-cp312-win32.whl", hash = "sha256:1e2cb691ed9834cd6a8be61228471d0a503731abfb42f82458ff27be7b2186fc", size = 1834064, upload-time = "2024-12-18T11:28:56.074Z" }, + { url = "https://files.pythonhosted.org/packages/1f/ea/cd7209a889163b8dcca139fe32b9687dd05249161a3edda62860430457a5/pydantic_core-2.27.2-cp312-cp312-win_amd64.whl", hash = "sha256:cc3f1a99a4f4f9dd1de4fe0312c114e740b5ddead65bb4102884b384c15d8bc9", size = 1989046, upload-time = "2024-12-18T11:28:58.107Z" }, + { url = "https://files.pythonhosted.org/packages/bc/49/c54baab2f4658c26ac633d798dab66b4c3a9bbf47cff5284e9c182f4137a/pydantic_core-2.27.2-cp312-cp312-win_arm64.whl", hash = "sha256:3911ac9284cd8a1792d3cb26a2da18f3ca26c6908cc434a18f730dc0db7bfa3b", size = 1885092, upload-time = "2024-12-18T11:29:01.335Z" }, + { url = "https://files.pythonhosted.org/packages/41/b1/9bc383f48f8002f99104e3acff6cba1231b29ef76cfa45d1506a5cad1f84/pydantic_core-2.27.2-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:7d14bd329640e63852364c306f4d23eb744e0f8193148d4044dd3dacdaacbd8b", size = 1892709, upload-time = "2024-12-18T11:29:03.193Z" }, + { url = "https://files.pythonhosted.org/packages/10/6c/e62b8657b834f3eb2961b49ec8e301eb99946245e70bf42c8817350cbefc/pydantic_core-2.27.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:82f91663004eb8ed30ff478d77c4d1179b3563df6cdb15c0817cd1cdaf34d154", size = 1811273, upload-time = "2024-12-18T11:29:05.306Z" }, + { url = "https://files.pythonhosted.org/packages/ba/15/52cfe49c8c986e081b863b102d6b859d9defc63446b642ccbbb3742bf371/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:71b24c7d61131bb83df10cc7e687433609963a944ccf45190cfc21e0887b08c9", size = 1823027, upload-time = "2024-12-18T11:29:07.294Z" }, + { url = "https://files.pythonhosted.org/packages/b1/1c/b6f402cfc18ec0024120602bdbcebc7bdd5b856528c013bd4d13865ca473/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:fa8e459d4954f608fa26116118bb67f56b93b209c39b008277ace29937453dc9", size = 1868888, upload-time = "2024-12-18T11:29:09.249Z" }, + { url = "https://files.pythonhosted.org/packages/bd/7b/8cb75b66ac37bc2975a3b7de99f3c6f355fcc4d89820b61dffa8f1e81677/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:ce8918cbebc8da707ba805b7fd0b382816858728ae7fe19a942080c24e5b7cd1", size = 2037738, upload-time = "2024-12-18T11:29:11.23Z" }, + { url = "https://files.pythonhosted.org/packages/c8/f1/786d8fe78970a06f61df22cba58e365ce304bf9b9f46cc71c8c424e0c334/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:eda3f5c2a021bbc5d976107bb302e0131351c2ba54343f8a496dc8783d3d3a6a", size = 2685138, upload-time = "2024-12-18T11:29:16.396Z" }, + { url = "https://files.pythonhosted.org/packages/a6/74/d12b2cd841d8724dc8ffb13fc5cef86566a53ed358103150209ecd5d1999/pydantic_core-2.27.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bd8086fa684c4775c27f03f062cbb9eaa6e17f064307e86b21b9e0abc9c0f02e", size = 1997025, upload-time = "2024-12-18T11:29:20.25Z" }, + { url = "https://files.pythonhosted.org/packages/a0/6e/940bcd631bc4d9a06c9539b51f070b66e8f370ed0933f392db6ff350d873/pydantic_core-2.27.2-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:8d9b3388db186ba0c099a6d20f0604a44eabdeef1777ddd94786cdae158729e4", size = 2004633, upload-time = "2024-12-18T11:29:23.877Z" }, + { url = "https://files.pythonhosted.org/packages/50/cc/a46b34f1708d82498c227d5d80ce615b2dd502ddcfd8376fc14a36655af1/pydantic_core-2.27.2-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:7a66efda2387de898c8f38c0cf7f14fca0b51a8ef0b24bfea5849f1b3c95af27", size = 1999404, upload-time = "2024-12-18T11:29:25.872Z" }, + { url = "https://files.pythonhosted.org/packages/ca/2d/c365cfa930ed23bc58c41463bae347d1005537dc8db79e998af8ba28d35e/pydantic_core-2.27.2-cp313-cp313-musllinux_1_1_armv7l.whl", hash = "sha256:18a101c168e4e092ab40dbc2503bdc0f62010e95d292b27827871dc85450d7ee", size = 2130130, upload-time = "2024-12-18T11:29:29.252Z" }, + { url = "https://files.pythonhosted.org/packages/f4/d7/eb64d015c350b7cdb371145b54d96c919d4db516817f31cd1c650cae3b21/pydantic_core-2.27.2-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:ba5dd002f88b78a4215ed2f8ddbdf85e8513382820ba15ad5ad8955ce0ca19a1", size = 2157946, upload-time = "2024-12-18T11:29:31.338Z" }, + { url = "https://files.pythonhosted.org/packages/a4/99/bddde3ddde76c03b65dfd5a66ab436c4e58ffc42927d4ff1198ffbf96f5f/pydantic_core-2.27.2-cp313-cp313-win32.whl", hash = "sha256:1ebaf1d0481914d004a573394f4be3a7616334be70261007e47c2a6fe7e50130", size = 1834387, upload-time = "2024-12-18T11:29:33.481Z" }, + { url = "https://files.pythonhosted.org/packages/71/47/82b5e846e01b26ac6f1893d3c5f9f3a2eb6ba79be26eef0b759b4fe72946/pydantic_core-2.27.2-cp313-cp313-win_amd64.whl", hash = "sha256:953101387ecf2f5652883208769a79e48db18c6df442568a0b5ccd8c2723abee", size = 1990453, upload-time = "2024-12-18T11:29:35.533Z" }, + { url = "https://files.pythonhosted.org/packages/51/b2/b2b50d5ecf21acf870190ae5d093602d95f66c9c31f9d5de6062eb329ad1/pydantic_core-2.27.2-cp313-cp313-win_arm64.whl", hash = "sha256:ac4dbfd1691affb8f48c2c13241a2e3b60ff23247cbcf981759c768b6633cf8b", size = 1885186, upload-time = "2024-12-18T11:29:37.649Z" }, + { url = "https://files.pythonhosted.org/packages/46/72/af70981a341500419e67d5cb45abe552a7c74b66326ac8877588488da1ac/pydantic_core-2.27.2-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:2bf14caea37e91198329b828eae1618c068dfb8ef17bb33287a7ad4b61ac314e", size = 1891159, upload-time = "2024-12-18T11:30:54.382Z" }, + { url = "https://files.pythonhosted.org/packages/ad/3d/c5913cccdef93e0a6a95c2d057d2c2cba347815c845cda79ddd3c0f5e17d/pydantic_core-2.27.2-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:b0cb791f5b45307caae8810c2023a184c74605ec3bcbb67d13846c28ff731ff8", size = 1768331, upload-time = "2024-12-18T11:30:58.178Z" }, + { url = "https://files.pythonhosted.org/packages/f6/f0/a3ae8fbee269e4934f14e2e0e00928f9346c5943174f2811193113e58252/pydantic_core-2.27.2-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:688d3fd9fcb71f41c4c015c023d12a79d1c4c0732ec9eb35d96e3388a120dcf3", size = 1822467, upload-time = "2024-12-18T11:31:00.6Z" }, + { url = "https://files.pythonhosted.org/packages/d7/7a/7bbf241a04e9f9ea24cd5874354a83526d639b02674648af3f350554276c/pydantic_core-2.27.2-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3d591580c34f4d731592f0e9fe40f9cc1b430d297eecc70b962e93c5c668f15f", size = 1979797, upload-time = "2024-12-18T11:31:07.243Z" }, + { url = "https://files.pythonhosted.org/packages/4f/5f/4784c6107731f89e0005a92ecb8a2efeafdb55eb992b8e9d0a2be5199335/pydantic_core-2.27.2-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:82f986faf4e644ffc189a7f1aafc86e46ef70372bb153e7001e8afccc6e54133", size = 1987839, upload-time = "2024-12-18T11:31:09.775Z" }, + { url = "https://files.pythonhosted.org/packages/6d/a7/61246562b651dff00de86a5f01b6e4befb518df314c54dec187a78d81c84/pydantic_core-2.27.2-pp310-pypy310_pp73-musllinux_1_1_aarch64.whl", hash = "sha256:bec317a27290e2537f922639cafd54990551725fc844249e64c523301d0822fc", size = 1998861, upload-time = "2024-12-18T11:31:13.469Z" }, + { url = "https://files.pythonhosted.org/packages/86/aa/837821ecf0c022bbb74ca132e117c358321e72e7f9702d1b6a03758545e2/pydantic_core-2.27.2-pp310-pypy310_pp73-musllinux_1_1_armv7l.whl", hash = "sha256:0296abcb83a797db256b773f45773da397da75a08f5fcaef41f2044adec05f50", size = 2116582, upload-time = "2024-12-18T11:31:17.423Z" }, + { url = "https://files.pythonhosted.org/packages/81/b0/5e74656e95623cbaa0a6278d16cf15e10a51f6002e3ec126541e95c29ea3/pydantic_core-2.27.2-pp310-pypy310_pp73-musllinux_1_1_x86_64.whl", hash = "sha256:0d75070718e369e452075a6017fbf187f788e17ed67a3abd47fa934d001863d9", size = 2151985, upload-time = "2024-12-18T11:31:19.901Z" }, + { url = "https://files.pythonhosted.org/packages/63/37/3e32eeb2a451fddaa3898e2163746b0cffbbdbb4740d38372db0490d67f3/pydantic_core-2.27.2-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:7e17b560be3c98a8e3aa66ce828bdebb9e9ac6ad5466fba92eb74c4c95cb1151", size = 2004715, upload-time = "2024-12-18T11:31:22.821Z" }, ] [[package]] name = "pygments" -version = "2.19.2" +version = "2.19.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/b0/77/a5b8c569bf593b0140bde72ea885a803b82086995367bf2037de0159d924/pygments-2.19.2.tar.gz", hash = "sha256:636cb2477cec7f8952536970bc533bc43743542f70392ae026374600add5b887", size = 4968631, upload-time = "2025-06-21T13:39:12.283Z" } +sdist = { url = "https://files.pythonhosted.org/packages/7c/2d/c3338d48ea6cc0feb8446d8e6937e1408088a72a39937982cc6111d17f84/pygments-2.19.1.tar.gz", hash = "sha256:61c16d2a8576dc0649d9f39e089b5f02bcd27fba10d8fb4dcc28173f7a45151f", size = 4968581, upload-time = "2025-01-06T17:26:30.443Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/c7/21/705964c7812476f378728bdf590ca4b771ec72385c533964653c68e86bdc/pygments-2.19.2-py3-none-any.whl", hash = "sha256:86540386c03d588bb81d44bc3928634ff26449851e99741617ecb9037ee5ec0b", size = 1225217, upload-time = "2025-06-21T13:39:07.939Z" }, + { url = "https://files.pythonhosted.org/packages/8a/0b/9fcc47d19c48b59121088dd6da2488a49d5f72dacf8262e2790a1d2c7d15/pygments-2.19.1-py3-none-any.whl", hash = "sha256:9ea1544ad55cecf4b8242fab6dd35a93bbce657034b0611ee383099054ab6d8c", size = 1225293, upload-time = "2025-01-06T17:26:25.553Z" }, ] [[package]] name = "pyparsing" -version = "3.2.3" +version = "3.2.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/bb/22/f1129e69d94ffff626bdb5c835506b3a5b4f3d070f17ea295e12c2c6f60f/pyparsing-3.2.3.tar.gz", hash = "sha256:b9c13f1ab8b3b542f72e28f634bad4de758ab3ce4546e4301970ad6fa77c38be", size = 1088608, upload-time = "2025-03-25T05:01:28.114Z" } +sdist = { url = "https://files.pythonhosted.org/packages/8b/1a/3544f4f299a47911c2ab3710f534e52fea62a633c96806995da5d25be4b2/pyparsing-3.2.1.tar.gz", hash = "sha256:61980854fd66de3a90028d679a954d5f2623e83144b5afe5ee86f43d762e5f0a", size = 1067694, upload-time = "2024-12-31T20:59:46.157Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/05/e7/df2285f3d08fee213f2d041540fa4fc9ca6c2d44cf36d3a035bf2a8d2bcc/pyparsing-3.2.3-py3-none-any.whl", hash = "sha256:a749938e02d6fd0b59b356ca504a24982314bb090c383e3cf201c95ef7e2bfcf", size = 111120, upload-time = "2025-03-25T05:01:24.908Z" }, + { url = "https://files.pythonhosted.org/packages/1c/a7/c8a2d361bf89c0d9577c934ebb7421b25dc84bf3a8e3ac0a40aed9acc547/pyparsing-3.2.1-py3-none-any.whl", hash = "sha256:506ff4f4386c4cec0590ec19e6302d3aedb992fdc02c761e90416f158dacf8e1", size = 107716, upload-time = "2024-12-31T20:59:42.738Z" }, ] [[package]] name = "pytest" -version = "8.4.1" +version = "8.3.5" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "colorama", marker = "sys_platform == 'win32'" }, @@ -3040,12 +2831,11 @@ dependencies = [ { name = "iniconfig" }, { name = "packaging" }, { name = "pluggy" }, - { name = "pygments" }, { name = "tomli", marker = "python_full_version < '3.11'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/08/ba/45911d754e8eba3d5a841a5ce61a65a685ff1798421ac054f85aa8747dfb/pytest-8.4.1.tar.gz", hash = "sha256:7c67fd69174877359ed9371ec3af8a3d2b04741818c51e5e99cc1742251fa93c", size = 1517714, upload-time = "2025-06-18T05:48:06.109Z" } +sdist = { url = "https://files.pythonhosted.org/packages/ae/3c/c9d525a414d506893f0cd8a8d0de7706446213181570cdbd766691164e40/pytest-8.3.5.tar.gz", hash = "sha256:f4efe70cc14e511565ac476b57c279e12a855b11f48f212af1080ef2263d3845", size = 1450891, upload-time = "2025-03-02T12:54:54.503Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/29/16/c8a903f4c4dffe7a12843191437d7cd8e32751d5de349d45d3fe69544e87/pytest-8.4.1-py3-none-any.whl", hash = "sha256:539c70ba6fcead8e78eebbf1115e8b589e7565830d7d006a8723f19ac8a0afb7", size = 365474, upload-time = "2025-06-18T05:48:03.955Z" }, + { url = "https://files.pythonhosted.org/packages/30/3d/64ad57c803f1fa1e963a7946b6e0fea4a70df53c1a7fed304586539c2bac/pytest-8.3.5-py3-none-any.whl", hash = "sha256:c69214aa47deac29fad6c2a4f590b9c4a9fdb16a403176fe154b79c0b4d4d820", size = 343634, upload-time = "2025-03-02T12:54:52.069Z" }, ] [[package]] @@ -3062,16 +2852,16 @@ wheels = [ [[package]] name = "python-dotenv" -version = "1.1.1" +version = "1.0.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/f6/b0/4bc07ccd3572a2f9df7e6782f52b0c6c90dcbb803ac4a167702d7d0dfe1e/python_dotenv-1.1.1.tar.gz", hash = "sha256:a8a6399716257f45be6a007360200409fce5cda2661e3dec71d23dc15f6189ab", size = 41978, upload-time = "2025-06-24T04:21:07.341Z" } +sdist = { url = "https://files.pythonhosted.org/packages/bc/57/e84d88dfe0aec03b7a2d4327012c1627ab5f03652216c63d49846d7a6c58/python-dotenv-1.0.1.tar.gz", hash = "sha256:e324ee90a023d808f1959c46bcbc04446a10ced277783dc6ee09987c37ec10ca", size = 39115, upload-time = "2024-01-23T06:33:00.505Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/5f/ed/539768cf28c661b5b068d66d96a2f155c4971a5d55684a514c1a0e0dec2f/python_dotenv-1.1.1-py3-none-any.whl", hash = "sha256:31f23644fe2602f88ff55e1f5c79ba497e01224ee7737937930c448e4d0e24dc", size = 20556, upload-time = "2025-06-24T04:21:06.073Z" }, + { url = "https://files.pythonhosted.org/packages/6a/3e/b68c118422ec867fa7ab88444e1274aa40681c606d59ac27de5a5588f082/python_dotenv-1.0.1-py3-none-any.whl", hash = "sha256:f7b63ef50f1b690dddf550d03497b66d609393b40b564ed0d674909a68ebf16a", size = 19863, upload-time = "2024-01-23T06:32:58.246Z" }, ] [[package]] name = "pytorch-lightning" -version = "2.5.2" +version = "2.5.1" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "fsspec", extra = ["http"] }, @@ -3083,40 +2873,37 @@ dependencies = [ { name = "tqdm" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/01/3e/728fbdc671d07727ad447f9401d98a43570573965beb3cb2060f9a330b4f/pytorch_lightning-2.5.2.tar.gz", hash = "sha256:f817087d611be8d43b777dd4e543d72703e235510936677a13e6c29f7fd790e3", size = 636859, upload-time = "2025-06-20T15:58:27.062Z" } +sdist = { url = "https://files.pythonhosted.org/packages/d8/75/b8ac01fd3974328ae9239c39550d48899b2fa1b0e064c01615b72a574082/pytorch_lightning-2.5.1.tar.gz", hash = "sha256:27a8adb799c13b8202afad518352248d61303fb230ec1f9fa60e0f81d431d6b1", size = 634309, upload-time = "2025-03-19T20:28:22.444Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/e2/42/47c186c8f9e956e559c89e6c764d5d5d0d0af517c04ca0ad39bd0a357d3a/pytorch_lightning-2.5.2-py3-none-any.whl", hash = "sha256:17cfdf89bd98074e389101f097cdf34c486a1f5c6d3fdcefbaf4dea7f97ff0bf", size = 825366, upload-time = "2025-06-20T15:58:25.534Z" }, + { url = "https://files.pythonhosted.org/packages/82/ff/5701f79317a1a03e5ee8a1bf48e7273a8445162a2774e51fc06411a67c89/pytorch_lightning-2.5.1-py3-none-any.whl", hash = "sha256:0bfbbd3ad80281d3062f5d8029a759093bd969ff8162e7c1fe2918552b269f9e", size = 822982, upload-time = "2025-03-19T20:28:20.1Z" }, ] [[package]] name = "pytz" -version = "2025.2" +version = "2025.1" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/f8/bf/abbd3cdfb8fbc7fb3d4d38d320f2441b1e7cbe29be4f23797b4a2b5d8aac/pytz-2025.2.tar.gz", hash = "sha256:360b9e3dbb49a209c21ad61809c7fb453643e048b38924c765813546746e81c3", size = 320884, upload-time = "2025-03-25T02:25:00.538Z" } +sdist = { url = "https://files.pythonhosted.org/packages/5f/57/df1c9157c8d5a05117e455d66fd7cf6dbc46974f832b1058ed4856785d8a/pytz-2025.1.tar.gz", hash = "sha256:c2db42be2a2518b28e65f9207c4d05e6ff547d1efa4086469ef855e4ab70178e", size = 319617, upload-time = "2025-01-31T01:54:48.615Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/81/c4/34e93fe5f5429d7570ec1fa436f1986fb1f00c3e0f43a589fe2bbcd22c3f/pytz-2025.2-py2.py3-none-any.whl", hash = "sha256:5ddf76296dd8c44c26eb8f4b6f35488f3ccbf6fbbd7adee0b7262d43f0ec2f00", size = 509225, upload-time = "2025-03-25T02:24:58.468Z" }, + { url = "https://files.pythonhosted.org/packages/eb/38/ac33370d784287baa1c3d538978b5e2ea064d4c1b93ffbd12826c190dd10/pytz-2025.1-py2.py3-none-any.whl", hash = "sha256:89dd22dca55b46eac6eda23b2d72721bf1bdfef212645d81513ef5d03038de57", size = 507930, upload-time = "2025-01-31T01:54:45.634Z" }, ] [[package]] name = "pywin32" -version = "311" +version = "310" source = { registry = "https://pypi.org/simple" } wheels = [ - { url = "https://files.pythonhosted.org/packages/7b/40/44efbb0dfbd33aca6a6483191dae0716070ed99e2ecb0c53683f400a0b4f/pywin32-311-cp310-cp310-win32.whl", hash = "sha256:d03ff496d2a0cd4a5893504789d4a15399133fe82517455e78bad62efbb7f0a3", size = 8760432, upload-time = "2025-07-14T20:13:05.9Z" }, - { url = "https://files.pythonhosted.org/packages/5e/bf/360243b1e953bd254a82f12653974be395ba880e7ec23e3731d9f73921cc/pywin32-311-cp310-cp310-win_amd64.whl", hash = "sha256:797c2772017851984b97180b0bebe4b620bb86328e8a884bb626156295a63b3b", size = 9590103, upload-time = "2025-07-14T20:13:07.698Z" }, - { url = "https://files.pythonhosted.org/packages/57/38/d290720e6f138086fb3d5ffe0b6caa019a791dd57866940c82e4eeaf2012/pywin32-311-cp310-cp310-win_arm64.whl", hash = "sha256:0502d1facf1fed4839a9a51ccbcc63d952cf318f78ffc00a7e78528ac27d7a2b", size = 8778557, upload-time = "2025-07-14T20:13:11.11Z" }, - { url = "https://files.pythonhosted.org/packages/7c/af/449a6a91e5d6db51420875c54f6aff7c97a86a3b13a0b4f1a5c13b988de3/pywin32-311-cp311-cp311-win32.whl", hash = "sha256:184eb5e436dea364dcd3d2316d577d625c0351bf237c4e9a5fabbcfa5a58b151", size = 8697031, upload-time = "2025-07-14T20:13:13.266Z" }, - { url = "https://files.pythonhosted.org/packages/51/8f/9bb81dd5bb77d22243d33c8397f09377056d5c687aa6d4042bea7fbf8364/pywin32-311-cp311-cp311-win_amd64.whl", hash = "sha256:3ce80b34b22b17ccbd937a6e78e7225d80c52f5ab9940fe0506a1a16f3dab503", size = 9508308, upload-time = "2025-07-14T20:13:15.147Z" }, - { url = "https://files.pythonhosted.org/packages/44/7b/9c2ab54f74a138c491aba1b1cd0795ba61f144c711daea84a88b63dc0f6c/pywin32-311-cp311-cp311-win_arm64.whl", hash = "sha256:a733f1388e1a842abb67ffa8e7aad0e70ac519e09b0f6a784e65a136ec7cefd2", size = 8703930, upload-time = "2025-07-14T20:13:16.945Z" }, - { url = "https://files.pythonhosted.org/packages/e7/ab/01ea1943d4eba0f850c3c61e78e8dd59757ff815ff3ccd0a84de5f541f42/pywin32-311-cp312-cp312-win32.whl", hash = "sha256:750ec6e621af2b948540032557b10a2d43b0cee2ae9758c54154d711cc852d31", size = 8706543, upload-time = "2025-07-14T20:13:20.765Z" }, - { url = "https://files.pythonhosted.org/packages/d1/a8/a0e8d07d4d051ec7502cd58b291ec98dcc0c3fff027caad0470b72cfcc2f/pywin32-311-cp312-cp312-win_amd64.whl", hash = "sha256:b8c095edad5c211ff31c05223658e71bf7116daa0ecf3ad85f3201ea3190d067", size = 9495040, upload-time = "2025-07-14T20:13:22.543Z" }, - { url = "https://files.pythonhosted.org/packages/ba/3a/2ae996277b4b50f17d61f0603efd8253cb2d79cc7ae159468007b586396d/pywin32-311-cp312-cp312-win_arm64.whl", hash = "sha256:e286f46a9a39c4a18b319c28f59b61de793654af2f395c102b4f819e584b5852", size = 8710102, upload-time = "2025-07-14T20:13:24.682Z" }, - { url = "https://files.pythonhosted.org/packages/a5/be/3fd5de0979fcb3994bfee0d65ed8ca9506a8a1260651b86174f6a86f52b3/pywin32-311-cp313-cp313-win32.whl", hash = "sha256:f95ba5a847cba10dd8c4d8fefa9f2a6cf283b8b88ed6178fa8a6c1ab16054d0d", size = 8705700, upload-time = "2025-07-14T20:13:26.471Z" }, - { url = "https://files.pythonhosted.org/packages/e3/28/e0a1909523c6890208295a29e05c2adb2126364e289826c0a8bc7297bd5c/pywin32-311-cp313-cp313-win_amd64.whl", hash = "sha256:718a38f7e5b058e76aee1c56ddd06908116d35147e133427e59a3983f703a20d", size = 9494700, upload-time = "2025-07-14T20:13:28.243Z" }, - { url = "https://files.pythonhosted.org/packages/04/bf/90339ac0f55726dce7d794e6d79a18a91265bdf3aa70b6b9ca52f35e022a/pywin32-311-cp313-cp313-win_arm64.whl", hash = "sha256:7b4075d959648406202d92a2310cb990fea19b535c7f4a78d3f5e10b926eeb8a", size = 8709318, upload-time = "2025-07-14T20:13:30.348Z" }, - { url = "https://files.pythonhosted.org/packages/c9/31/097f2e132c4f16d99a22bfb777e0fd88bd8e1c634304e102f313af69ace5/pywin32-311-cp314-cp314-win32.whl", hash = "sha256:b7a2c10b93f8986666d0c803ee19b5990885872a7de910fc460f9b0c2fbf92ee", size = 8840714, upload-time = "2025-07-14T20:13:32.449Z" }, - { url = "https://files.pythonhosted.org/packages/90/4b/07c77d8ba0e01349358082713400435347df8426208171ce297da32c313d/pywin32-311-cp314-cp314-win_amd64.whl", hash = "sha256:3aca44c046bd2ed8c90de9cb8427f581c479e594e99b5c0bb19b29c10fd6cb87", size = 9656800, upload-time = "2025-07-14T20:13:34.312Z" }, - { url = "https://files.pythonhosted.org/packages/c0/d2/21af5c535501a7233e734b8af901574572da66fcc254cb35d0609c9080dd/pywin32-311-cp314-cp314-win_arm64.whl", hash = "sha256:a508e2d9025764a8270f93111a970e1d0fbfc33f4153b388bb649b7eec4f9b42", size = 8932540, upload-time = "2025-07-14T20:13:36.379Z" }, + { url = "https://files.pythonhosted.org/packages/95/da/a5f38fffbba2fb99aa4aa905480ac4b8e83ca486659ac8c95bce47fb5276/pywin32-310-cp310-cp310-win32.whl", hash = "sha256:6dd97011efc8bf51d6793a82292419eba2c71cf8e7250cfac03bba284454abc1", size = 8848240, upload-time = "2025-03-17T00:55:46.783Z" }, + { url = "https://files.pythonhosted.org/packages/aa/fe/d873a773324fa565619ba555a82c9dabd677301720f3660a731a5d07e49a/pywin32-310-cp310-cp310-win_amd64.whl", hash = "sha256:c3e78706e4229b915a0821941a84e7ef420bf2b77e08c9dae3c76fd03fd2ae3d", size = 9601854, upload-time = "2025-03-17T00:55:48.783Z" }, + { url = "https://files.pythonhosted.org/packages/3c/84/1a8e3d7a15490d28a5d816efa229ecb4999cdc51a7c30dd8914f669093b8/pywin32-310-cp310-cp310-win_arm64.whl", hash = "sha256:33babed0cf0c92a6f94cc6cc13546ab24ee13e3e800e61ed87609ab91e4c8213", size = 8522963, upload-time = "2025-03-17T00:55:50.969Z" }, + { url = "https://files.pythonhosted.org/packages/f7/b1/68aa2986129fb1011dabbe95f0136f44509afaf072b12b8f815905a39f33/pywin32-310-cp311-cp311-win32.whl", hash = "sha256:1e765f9564e83011a63321bb9d27ec456a0ed90d3732c4b2e312b855365ed8bd", size = 8784284, upload-time = "2025-03-17T00:55:53.124Z" }, + { url = "https://files.pythonhosted.org/packages/b3/bd/d1592635992dd8db5bb8ace0551bc3a769de1ac8850200cfa517e72739fb/pywin32-310-cp311-cp311-win_amd64.whl", hash = "sha256:126298077a9d7c95c53823934f000599f66ec9296b09167810eb24875f32689c", size = 9520748, upload-time = "2025-03-17T00:55:55.203Z" }, + { url = "https://files.pythonhosted.org/packages/90/b1/ac8b1ffce6603849eb45a91cf126c0fa5431f186c2e768bf56889c46f51c/pywin32-310-cp311-cp311-win_arm64.whl", hash = "sha256:19ec5fc9b1d51c4350be7bb00760ffce46e6c95eaf2f0b2f1150657b1a43c582", size = 8455941, upload-time = "2025-03-17T00:55:57.048Z" }, + { url = "https://files.pythonhosted.org/packages/6b/ec/4fdbe47932f671d6e348474ea35ed94227fb5df56a7c30cbbb42cd396ed0/pywin32-310-cp312-cp312-win32.whl", hash = "sha256:8a75a5cc3893e83a108c05d82198880704c44bbaee4d06e442e471d3c9ea4f3d", size = 8796239, upload-time = "2025-03-17T00:55:58.807Z" }, + { url = "https://files.pythonhosted.org/packages/e3/e5/b0627f8bb84e06991bea89ad8153a9e50ace40b2e1195d68e9dff6b03d0f/pywin32-310-cp312-cp312-win_amd64.whl", hash = "sha256:bf5c397c9a9a19a6f62f3fb821fbf36cac08f03770056711f765ec1503972060", size = 9503839, upload-time = "2025-03-17T00:56:00.8Z" }, + { url = "https://files.pythonhosted.org/packages/1f/32/9ccf53748df72301a89713936645a664ec001abd35ecc8578beda593d37d/pywin32-310-cp312-cp312-win_arm64.whl", hash = "sha256:2349cc906eae872d0663d4d6290d13b90621eaf78964bb1578632ff20e152966", size = 8459470, upload-time = "2025-03-17T00:56:02.601Z" }, + { url = "https://files.pythonhosted.org/packages/1c/09/9c1b978ffc4ae53999e89c19c77ba882d9fce476729f23ef55211ea1c034/pywin32-310-cp313-cp313-win32.whl", hash = "sha256:5d241a659c496ada3253cd01cfaa779b048e90ce4b2b38cd44168ad555ce74ab", size = 8794384, upload-time = "2025-03-17T00:56:04.383Z" }, + { url = "https://files.pythonhosted.org/packages/45/3c/b4640f740ffebadd5d34df35fecba0e1cfef8fde9f3e594df91c28ad9b50/pywin32-310-cp313-cp313-win_amd64.whl", hash = "sha256:667827eb3a90208ddbdcc9e860c81bde63a135710e21e4cb3348968e4bd5249e", size = 9503039, upload-time = "2025-03-17T00:56:06.207Z" }, + { url = "https://files.pythonhosted.org/packages/b4/f4/f785020090fb050e7fb6d34b780f2231f302609dc964672f72bfaeb59a28/pywin32-310-cp313-cp313-win_arm64.whl", hash = "sha256:e308f831de771482b7cf692a1f308f8fca701b2d8f9dde6cc440c7da17e47b33", size = 8458152, upload-time = "2025-03-17T00:56:07.819Z" }, ] [[package]] @@ -3165,112 +2952,106 @@ wheels = [ [[package]] name = "pyzmq" -version = "27.0.1" +version = "26.3.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "cffi", marker = "implementation_name == 'pypy'" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/30/5f/557d2032a2f471edbcc227da724c24a1c05887b5cda1e3ae53af98b9e0a5/pyzmq-27.0.1.tar.gz", hash = "sha256:45c549204bc20e7484ffd2555f6cf02e572440ecf2f3bdd60d4404b20fddf64b", size = 281158, upload-time = "2025-08-03T05:05:40.352Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/72/0b/ccf4d0b152a6a11f0fc01e73978202fe0e8fe0e91e20941598e83a170bee/pyzmq-27.0.1-cp310-cp310-macosx_10_15_universal2.whl", hash = "sha256:90a4da42aa322de8a3522461e3b5fe999935763b27f69a02fced40f4e3cf9682", size = 1329293, upload-time = "2025-08-03T05:02:56.001Z" }, - { url = "https://files.pythonhosted.org/packages/bc/76/48706d291951b1300d3cf985e503806901164bf1581f27c4b6b22dbab2fa/pyzmq-27.0.1-cp310-cp310-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:e648dca28178fc879c814cf285048dd22fd1f03e1104101106505ec0eea50a4d", size = 905953, upload-time = "2025-08-03T05:02:59.061Z" }, - { url = "https://files.pythonhosted.org/packages/aa/8a/df3135b96712068d184c53120c7dbf3023e5e362a113059a4f85cd36c6a0/pyzmq-27.0.1-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4bca8abc31799a6f3652d13f47e0b0e1cab76f9125f2283d085a3754f669b607", size = 666165, upload-time = "2025-08-03T05:03:00.789Z" }, - { url = "https://files.pythonhosted.org/packages/ee/ed/341a7148e08d2830f480f53ab3d136d88fc5011bb367b516d95d0ebb46dd/pyzmq-27.0.1-cp310-cp310-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:092f4011b26d6b0201002f439bd74b38f23f3aefcb358621bdc3b230afc9b2d5", size = 853756, upload-time = "2025-08-03T05:03:03.347Z" }, - { url = "https://files.pythonhosted.org/packages/c2/bc/d26fe010477c3e901f0f5a3e70446950dde9aa217f1d1a13534eb0fccfe5/pyzmq-27.0.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:6f02f30a4a6b3efe665ab13a3dd47109d80326c8fd286311d1ba9f397dc5f247", size = 1654870, upload-time = "2025-08-03T05:03:05.331Z" }, - { url = "https://files.pythonhosted.org/packages/32/21/9b488086bf3f55b2eb26db09007a3962f62f3b81c5c6295a6ff6aaebd69c/pyzmq-27.0.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:f293a1419266e3bf3557d1f8778f9e1ffe7e6b2c8df5c9dca191caf60831eb74", size = 2033444, upload-time = "2025-08-03T05:03:07.318Z" }, - { url = "https://files.pythonhosted.org/packages/3d/53/85b64a792223cd43393d25e03c8609df41aac817ea5ce6a27eceeed433ee/pyzmq-27.0.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:ce181dd1a7c6c012d0efa8ab603c34b5ee9d86e570c03415bbb1b8772eeb381c", size = 1891289, upload-time = "2025-08-03T05:03:08.96Z" }, - { url = "https://files.pythonhosted.org/packages/23/5b/078aae8fe1c4cdba1a77a598870c548fd52b4d4a11e86b8116bbef47d9f3/pyzmq-27.0.1-cp310-cp310-win32.whl", hash = "sha256:f65741cc06630652e82aa68ddef4986a3ab9073dd46d59f94ce5f005fa72037c", size = 566693, upload-time = "2025-08-03T05:03:10.711Z" }, - { url = "https://files.pythonhosted.org/packages/24/e1/4471fff36416ebf1ffe43577b9c7dcf2ff4798f2171f0d169640a48d2305/pyzmq-27.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:44909aa3ed2234d69fe81e1dade7be336bcfeab106e16bdaa3318dcde4262b93", size = 631649, upload-time = "2025-08-03T05:03:12.232Z" }, - { url = "https://files.pythonhosted.org/packages/e8/4c/8edac8dd56f223124aa40403d2c097bbad9b0e2868a67cad9a2a029863aa/pyzmq-27.0.1-cp310-cp310-win_arm64.whl", hash = "sha256:4401649bfa0a38f0f8777f8faba7cd7eb7b5b8ae2abc7542b830dd09ad4aed0d", size = 559274, upload-time = "2025-08-03T05:03:13.728Z" }, - { url = "https://files.pythonhosted.org/packages/ae/18/a8e0da6ababbe9326116fb1c890bf1920eea880e8da621afb6bc0f39a262/pyzmq-27.0.1-cp311-cp311-macosx_10_15_universal2.whl", hash = "sha256:9729190bd770314f5fbba42476abf6abe79a746eeda11d1d68fd56dd70e5c296", size = 1332721, upload-time = "2025-08-03T05:03:15.237Z" }, - { url = "https://files.pythonhosted.org/packages/75/a4/9431ba598651d60ebd50dc25755402b770322cf8432adcc07d2906e53a54/pyzmq-27.0.1-cp311-cp311-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:696900ef6bc20bef6a242973943574f96c3f97d2183c1bd3da5eea4f559631b1", size = 908249, upload-time = "2025-08-03T05:03:16.933Z" }, - { url = "https://files.pythonhosted.org/packages/f0/7a/e624e1793689e4e685d2ee21c40277dd4024d9d730af20446d88f69be838/pyzmq-27.0.1-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f96a63aecec22d3f7fdea3c6c98df9e42973f5856bb6812c3d8d78c262fee808", size = 668649, upload-time = "2025-08-03T05:03:18.49Z" }, - { url = "https://files.pythonhosted.org/packages/6c/29/0652a39d4e876e0d61379047ecf7752685414ad2e253434348246f7a2a39/pyzmq-27.0.1-cp311-cp311-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c512824360ea7490390566ce00bee880e19b526b312b25cc0bc30a0fe95cb67f", size = 856601, upload-time = "2025-08-03T05:03:20.194Z" }, - { url = "https://files.pythonhosted.org/packages/36/2d/8d5355d7fc55bb6e9c581dd74f58b64fa78c994079e3a0ea09b1b5627cde/pyzmq-27.0.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:dfb2bb5e0f7198eaacfb6796fb0330afd28f36d985a770745fba554a5903595a", size = 1657750, upload-time = "2025-08-03T05:03:22.055Z" }, - { url = "https://files.pythonhosted.org/packages/ab/f4/cd032352d5d252dc6f5ee272a34b59718ba3af1639a8a4ef4654f9535cf5/pyzmq-27.0.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:4f6886c59ba93ffde09b957d3e857e7950c8fe818bd5494d9b4287bc6d5bc7f1", size = 2034312, upload-time = "2025-08-03T05:03:23.578Z" }, - { url = "https://files.pythonhosted.org/packages/e4/1a/c050d8b6597200e97a4bd29b93c769d002fa0b03083858227e0376ad59bc/pyzmq-27.0.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:b99ea9d330e86ce1ff7f2456b33f1bf81c43862a5590faf4ef4ed3a63504bdab", size = 1893632, upload-time = "2025-08-03T05:03:25.167Z" }, - { url = "https://files.pythonhosted.org/packages/6a/29/173ce21d5097e7fcf284a090e8beb64fc683c6582b1f00fa52b1b7e867ce/pyzmq-27.0.1-cp311-cp311-win32.whl", hash = "sha256:571f762aed89025ba8cdcbe355fea56889715ec06d0264fd8b6a3f3fa38154ed", size = 566587, upload-time = "2025-08-03T05:03:26.769Z" }, - { url = "https://files.pythonhosted.org/packages/53/ab/22bd33e7086f0a2cc03a5adabff4bde414288bb62a21a7820951ef86ec20/pyzmq-27.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:ee16906c8025fa464bea1e48128c048d02359fb40bebe5333103228528506530", size = 632873, upload-time = "2025-08-03T05:03:28.685Z" }, - { url = "https://files.pythonhosted.org/packages/90/14/3e59b4a28194285ceeff725eba9aa5ba8568d1cb78aed381dec1537c705a/pyzmq-27.0.1-cp311-cp311-win_arm64.whl", hash = "sha256:ba068f28028849da725ff9185c24f832ccf9207a40f9b28ac46ab7c04994bd41", size = 558918, upload-time = "2025-08-03T05:03:30.085Z" }, - { url = "https://files.pythonhosted.org/packages/0e/9b/c0957041067c7724b310f22c398be46399297c12ed834c3bc42200a2756f/pyzmq-27.0.1-cp312-abi3-macosx_10_15_universal2.whl", hash = "sha256:af7ebce2a1e7caf30c0bb64a845f63a69e76a2fadbc1cac47178f7bb6e657bdd", size = 1305432, upload-time = "2025-08-03T05:03:32.177Z" }, - { url = "https://files.pythonhosted.org/packages/8e/55/bd3a312790858f16b7def3897a0c3eb1804e974711bf7b9dcb5f47e7f82c/pyzmq-27.0.1-cp312-abi3-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:8f617f60a8b609a13099b313e7e525e67f84ef4524b6acad396d9ff153f6e4cd", size = 895095, upload-time = "2025-08-03T05:03:33.918Z" }, - { url = "https://files.pythonhosted.org/packages/20/50/fc384631d8282809fb1029a4460d2fe90fa0370a0e866a8318ed75c8d3bb/pyzmq-27.0.1-cp312-abi3-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1d59dad4173dc2a111f03e59315c7bd6e73da1a9d20a84a25cf08325b0582b1a", size = 651826, upload-time = "2025-08-03T05:03:35.818Z" }, - { url = "https://files.pythonhosted.org/packages/7e/0a/2356305c423a975000867de56888b79e44ec2192c690ff93c3109fd78081/pyzmq-27.0.1-cp312-abi3-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f5b6133c8d313bde8bd0d123c169d22525300ff164c2189f849de495e1344577", size = 839751, upload-time = "2025-08-03T05:03:37.265Z" }, - { url = "https://files.pythonhosted.org/packages/d7/1b/81e95ad256ca7e7ccd47f5294c1c6da6e2b64fbace65b84fe8a41470342e/pyzmq-27.0.1-cp312-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:58cca552567423f04d06a075f4b473e78ab5bdb906febe56bf4797633f54aa4e", size = 1641359, upload-time = "2025-08-03T05:03:38.799Z" }, - { url = "https://files.pythonhosted.org/packages/50/63/9f50ec965285f4e92c265c8f18344e46b12803666d8b73b65d254d441435/pyzmq-27.0.1-cp312-abi3-musllinux_1_2_i686.whl", hash = "sha256:4b9d8e26fb600d0d69cc9933e20af08552e97cc868a183d38a5c0d661e40dfbb", size = 2020281, upload-time = "2025-08-03T05:03:40.338Z" }, - { url = "https://files.pythonhosted.org/packages/02/4a/19e3398d0dc66ad2b463e4afa1fc541d697d7bc090305f9dfb948d3dfa29/pyzmq-27.0.1-cp312-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:2329f0c87f0466dce45bba32b63f47018dda5ca40a0085cc5c8558fea7d9fc55", size = 1877112, upload-time = "2025-08-03T05:03:42.012Z" }, - { url = "https://files.pythonhosted.org/packages/bf/42/c562e9151aa90ed1d70aac381ea22a929d6b3a2ce4e1d6e2e135d34fd9c6/pyzmq-27.0.1-cp312-abi3-win32.whl", hash = "sha256:57bb92abdb48467b89c2d21da1ab01a07d0745e536d62afd2e30d5acbd0092eb", size = 558177, upload-time = "2025-08-03T05:03:43.979Z" }, - { url = "https://files.pythonhosted.org/packages/40/96/5c50a7d2d2b05b19994bf7336b97db254299353dd9b49b565bb71b485f03/pyzmq-27.0.1-cp312-abi3-win_amd64.whl", hash = "sha256:ff3f8757570e45da7a5bedaa140489846510014f7a9d5ee9301c61f3f1b8a686", size = 618923, upload-time = "2025-08-03T05:03:45.438Z" }, - { url = "https://files.pythonhosted.org/packages/13/33/1ec89c8f21c89d21a2eaff7def3676e21d8248d2675705e72554fb5a6f3f/pyzmq-27.0.1-cp312-abi3-win_arm64.whl", hash = "sha256:df2c55c958d3766bdb3e9d858b911288acec09a9aab15883f384fc7180df5bed", size = 552358, upload-time = "2025-08-03T05:03:46.887Z" }, - { url = "https://files.pythonhosted.org/packages/6c/a0/f26e276211ec8090a4d11e4ec70eb8a8b15781e591c1d44ce62f372963a0/pyzmq-27.0.1-cp313-cp313-android_24_arm64_v8a.whl", hash = "sha256:497bd8af534ae55dc4ef67eebd1c149ff2a0b0f1e146db73c8b5a53d83c1a5f5", size = 1122287, upload-time = "2025-08-03T05:03:48.838Z" }, - { url = "https://files.pythonhosted.org/packages/9c/d8/af4b507e4f7eeea478cc8ee873995a6fd55582bfb99140593ed460e1db3c/pyzmq-27.0.1-cp313-cp313-android_24_x86_64.whl", hash = "sha256:a066ea6ad6218b4c233906adf0ae67830f451ed238419c0db609310dd781fbe7", size = 1155756, upload-time = "2025-08-03T05:03:50.907Z" }, - { url = "https://files.pythonhosted.org/packages/ac/55/37fae0013e11f88681da42698e550b08a316d608242551f65095cc99232a/pyzmq-27.0.1-cp313-cp313t-macosx_10_15_universal2.whl", hash = "sha256:72d235d6365ca73d8ce92f7425065d70f5c1e19baa458eb3f0d570e425b73a96", size = 1340826, upload-time = "2025-08-03T05:03:52.568Z" }, - { url = "https://files.pythonhosted.org/packages/f2/e4/3a87854c64b26fcf63a9d1b6f4382bd727d4797c772ceb334a97b7489be9/pyzmq-27.0.1-cp313-cp313t-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:313a7b374e3dc64848644ca348a51004b41726f768b02e17e689f1322366a4d9", size = 897283, upload-time = "2025-08-03T05:03:54.167Z" }, - { url = "https://files.pythonhosted.org/packages/17/3e/4296c6b0ad2d07be11ae1395dccf9cae48a0a655cf9be1c3733ad2b591d1/pyzmq-27.0.1-cp313-cp313t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:119ce8590409702394f959c159d048002cbed2f3c0645ec9d6a88087fc70f0f1", size = 660565, upload-time = "2025-08-03T05:03:56.152Z" }, - { url = "https://files.pythonhosted.org/packages/72/41/a33ba3aa48b45b23c4cd4ac49aafde46f3e0f81939f2bfb3b6171a437122/pyzmq-27.0.1-cp313-cp313t-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:45c3e00ce16896ace2cd770ab9057a7cf97d4613ea5f2a13f815141d8b6894b9", size = 847680, upload-time = "2025-08-03T05:03:57.696Z" }, - { url = "https://files.pythonhosted.org/packages/3f/8c/bf2350bb25b3b58d2e5b5d2290ffab0e923f0cc6d02288d3fbf4baa6e4d1/pyzmq-27.0.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:678e50ec112bdc6df5a83ac259a55a4ba97a8b314c325ab26b3b5b071151bc61", size = 1650151, upload-time = "2025-08-03T05:03:59.387Z" }, - { url = "https://files.pythonhosted.org/packages/f7/1a/a5a07c54890891344a8ddc3d5ab320dd3c4e39febb6e4472546e456d5157/pyzmq-27.0.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:d0b96c30be9f9387b18b18b6133c75a7b1b0065da64e150fe1feb5ebf31ece1c", size = 2023766, upload-time = "2025-08-03T05:04:01.883Z" }, - { url = "https://files.pythonhosted.org/packages/62/5e/514dcff08f02c6c8a45a6e23621901139cf853be7ac5ccd0b9407c3aa3de/pyzmq-27.0.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:88dc92d9eb5ea4968123e74db146d770b0c8d48f0e2bfb1dbc6c50a8edb12d64", size = 1885195, upload-time = "2025-08-03T05:04:03.923Z" }, - { url = "https://files.pythonhosted.org/packages/c8/91/87f74f98a487fbef0b115f6025e4a295129fd56b2b633a03ba7d5816ecc2/pyzmq-27.0.1-cp313-cp313t-win32.whl", hash = "sha256:6dcbcb34f5c9b0cefdfc71ff745459241b7d3cda5b27c7ad69d45afc0821d1e1", size = 574213, upload-time = "2025-08-03T05:04:05.905Z" }, - { url = "https://files.pythonhosted.org/packages/e6/d7/07f7d0d7f4c81e08be7b60e52ff2591c557377c017f96204d33d5fca1b07/pyzmq-27.0.1-cp313-cp313t-win_amd64.whl", hash = "sha256:b9fd0fda730461f510cfd9a40fafa5355d65f5e3dbdd8d6dfa342b5b3f5d1949", size = 640202, upload-time = "2025-08-03T05:04:07.439Z" }, - { url = "https://files.pythonhosted.org/packages/ab/83/21d66bcef6fb803647a223cbde95111b099e2176277c0cbc8b099c485510/pyzmq-27.0.1-cp313-cp313t-win_arm64.whl", hash = "sha256:56a3b1853f3954ec1f0e91085f1350cc57d18f11205e4ab6e83e4b7c414120e0", size = 561514, upload-time = "2025-08-03T05:04:09.071Z" }, - { url = "https://files.pythonhosted.org/packages/5a/0b/d5ea75cf46b52cdce85a85200c963cb498932953df443892238be49b1a01/pyzmq-27.0.1-cp314-cp314t-macosx_10_15_universal2.whl", hash = "sha256:f98f6b7787bd2beb1f0dde03f23a0621a0c978edf673b7d8f5e7bc039cbe1b60", size = 1340836, upload-time = "2025-08-03T05:04:10.774Z" }, - { url = "https://files.pythonhosted.org/packages/be/4c/0dbce882550e17db6846b29e9dc242aea7590e7594e1ca5043e8e58fff2d/pyzmq-27.0.1-cp314-cp314t-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:351bf5d8ca0788ca85327fda45843b6927593ff4c807faee368cc5aaf9f809c2", size = 897236, upload-time = "2025-08-03T05:04:13.221Z" }, - { url = "https://files.pythonhosted.org/packages/1b/22/461e131cf16b8814f3c356fa1ea0912697dbc4c64cddf01f7756ec704c1e/pyzmq-27.0.1-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5268a5a9177afff53dc6d70dffe63114ba2a6e7b20d9411cc3adeba09eeda403", size = 660374, upload-time = "2025-08-03T05:04:15.032Z" }, - { url = "https://files.pythonhosted.org/packages/3f/0c/bbd65a814395bf4fc3e57c6c13af27601c07e4009bdfb75ebcf500537bbd/pyzmq-27.0.1-cp314-cp314t-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a4aca06ba295aa78bec9b33ec028d1ca08744c36294338c41432b7171060c808", size = 847497, upload-time = "2025-08-03T05:04:16.967Z" }, - { url = "https://files.pythonhosted.org/packages/1e/df/3d1f4a03b561d824cbd491394f67591957e2f1acf6dc85d96f970312a76a/pyzmq-27.0.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:1c363c6dc66352331d5ad64bb838765c6692766334a6a02fdb05e76bd408ae18", size = 1650028, upload-time = "2025-08-03T05:04:19.398Z" }, - { url = "https://files.pythonhosted.org/packages/41/c9/a3987540f59a412bdaae3f362f78e00e6769557a598c63b7e32956aade5a/pyzmq-27.0.1-cp314-cp314t-musllinux_1_2_i686.whl", hash = "sha256:87aebf4acd7249bdff8d3df03aed4f09e67078e6762cfe0aecf8d0748ff94cde", size = 2023808, upload-time = "2025-08-03T05:04:21.145Z" }, - { url = "https://files.pythonhosted.org/packages/b0/a5/c388f4cd80498a8eaef7535f2a8eaca0a35b82b87a0b47fa1856fc135004/pyzmq-27.0.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:e4f22d67756518d71901edf73b38dc0eb4765cce22c8fe122cc81748d425262b", size = 1884970, upload-time = "2025-08-03T05:04:22.908Z" }, - { url = "https://files.pythonhosted.org/packages/9a/ac/b2a89a1ed90526a1b9a260cdc5cd42f055fd44ee8d2a59902b5ac35ddeb1/pyzmq-27.0.1-cp314-cp314t-win32.whl", hash = "sha256:8c62297bc7aea2147b472ca5ca2b4389377ad82898c87cabab2a94aedd75e337", size = 586905, upload-time = "2025-08-03T05:04:24.492Z" }, - { url = "https://files.pythonhosted.org/packages/68/62/7aa5ea04e836f7a788b2a67405f83011cef59ca76d7bac91d1fc9a0476da/pyzmq-27.0.1-cp314-cp314t-win_amd64.whl", hash = "sha256:bee5248d5ec9223545f8cc4f368c2d571477ae828c99409125c3911511d98245", size = 660503, upload-time = "2025-08-03T05:04:26.382Z" }, - { url = "https://files.pythonhosted.org/packages/89/32/3836ed85947b06f1d67c07ce16c00b0cf8c053ab0b249d234f9f81ff95ff/pyzmq-27.0.1-cp314-cp314t-win_arm64.whl", hash = "sha256:0fc24bf45e4a454e55ef99d7f5c8b8712539200ce98533af25a5bfa954b6b390", size = 575098, upload-time = "2025-08-03T05:04:27.974Z" }, - { url = "https://files.pythonhosted.org/packages/6f/87/fc96f224dd99070fe55d0afc37ac08d7d4635d434e3f9425b232867e01b9/pyzmq-27.0.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:544b995a6a1976fad5d7ff01409b4588f7608ccc41be72147700af91fd44875d", size = 835950, upload-time = "2025-08-03T05:05:04.193Z" }, - { url = "https://files.pythonhosted.org/packages/d1/b6/802d96017f176c3a7285603d9ed2982550095c136c6230d3e0b53f52c7e5/pyzmq-27.0.1-pp310-pypy310_pp73-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:0f772eea55cccce7f45d6ecdd1d5049c12a77ec22404f6b892fae687faa87bee", size = 799876, upload-time = "2025-08-03T05:05:06.263Z" }, - { url = "https://files.pythonhosted.org/packages/4e/52/49045c6528007cce385f218f3a674dc84fc8b3265330d09e57c0a59b41f4/pyzmq-27.0.1-pp310-pypy310_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c9d63d66059114a6756d09169c9209ffceabacb65b9cb0f66e6fc344b20b73e6", size = 567402, upload-time = "2025-08-03T05:05:08.028Z" }, - { url = "https://files.pythonhosted.org/packages/bc/fe/c29ac0d5a817543ecf0cb18f17195805bad0da567a1c64644aacf11b2779/pyzmq-27.0.1-pp310-pypy310_pp73-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1da8e645c655d86f0305fb4c65a0d848f461cd90ee07d21f254667287b5dbe50", size = 747030, upload-time = "2025-08-03T05:05:10.116Z" }, - { url = "https://files.pythonhosted.org/packages/17/d1/cc1fbfb65b4042016e4e035b2548cdfe0945c817345df83aa2d98490e7fc/pyzmq-27.0.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:1843fd0daebcf843fe6d4da53b8bdd3fc906ad3e97d25f51c3fed44436d82a49", size = 544567, upload-time = "2025-08-03T05:05:11.856Z" }, - { url = "https://files.pythonhosted.org/packages/b4/1a/49f66fe0bc2b2568dd4280f1f520ac8fafd73f8d762140e278d48aeaf7b9/pyzmq-27.0.1-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:7fb0ee35845bef1e8c4a152d766242164e138c239e3182f558ae15cb4a891f94", size = 835949, upload-time = "2025-08-03T05:05:13.798Z" }, - { url = "https://files.pythonhosted.org/packages/49/94/443c1984b397eab59b14dd7ae8bc2ac7e8f32dbc646474453afcaa6508c4/pyzmq-27.0.1-pp311-pypy311_pp73-manylinux2014_i686.manylinux_2_17_i686.whl", hash = "sha256:f379f11e138dfd56c3f24a04164f871a08281194dd9ddf656a278d7d080c8ad0", size = 799875, upload-time = "2025-08-03T05:05:15.632Z" }, - { url = "https://files.pythonhosted.org/packages/30/f1/fd96138a0f152786a2ba517e9c6a8b1b3516719e412a90bb5d8eea6b660c/pyzmq-27.0.1-pp311-pypy311_pp73-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b978c0678cffbe8860ec9edc91200e895c29ae1ac8a7085f947f8e8864c489fb", size = 567403, upload-time = "2025-08-03T05:05:17.326Z" }, - { url = "https://files.pythonhosted.org/packages/16/57/34e53ef2b55b1428dac5aabe3a974a16c8bda3bf20549ba500e3ff6cb426/pyzmq-27.0.1-pp311-pypy311_pp73-manylinux_2_26_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7ebccf0d760bc92a4a7c751aeb2fef6626144aace76ee8f5a63abeb100cae87f", size = 747032, upload-time = "2025-08-03T05:05:19.074Z" }, - { url = "https://files.pythonhosted.org/packages/81/b7/769598c5ae336fdb657946950465569cf18803140fe89ce466d7f0a57c11/pyzmq-27.0.1-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:77fed80e30fa65708546c4119840a46691290efc231f6bfb2ac2a39b52e15811", size = 544566, upload-time = "2025-08-03T05:05:20.798Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/3a/ed/c3876f3b3e8beba336214ce44e1efa1792dd537027cef24192ac2b077d7c/pyzmq-26.3.0.tar.gz", hash = "sha256:f1cd68b8236faab78138a8fc703f7ca0ad431b17a3fcac696358600d4e6243b3", size = 276733, upload-time = "2025-03-12T08:04:30.804Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/41/a8/cc21dcd6f0f96dbd636fcaab345f9664cd54e6577a21a74694202479d3fa/pyzmq-26.3.0-cp310-cp310-macosx_10_15_universal2.whl", hash = "sha256:1586944f4736515af5c6d3a5b150c7e8ca2a2d6e46b23057320584d6f2438f4a", size = 1345312, upload-time = "2025-03-12T08:01:58.084Z" }, + { url = "https://files.pythonhosted.org/packages/0b/6d/7e0e52798697536d572a105849c4ab621ca00511674b6ce694cb05e437fc/pyzmq-26.3.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa7efc695d1fc9f72d91bf9b6c6fe2d7e1b4193836ec530a98faf7d7a7577a58", size = 678336, upload-time = "2025-03-12T08:01:59.912Z" }, + { url = "https://files.pythonhosted.org/packages/91/86/8914875e2341a40da460feaa9cace727e50a6b640a20ac36186686bde7d9/pyzmq-26.3.0-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bd84441e4021cec6e4dd040550386cd9c9ea1d9418ea1a8002dbb7b576026b2b", size = 916965, upload-time = "2025-03-12T08:02:01.24Z" }, + { url = "https://files.pythonhosted.org/packages/9a/59/72b390b31ed0cc825881435f21baaae9d57e263aba526fa833863b90d667/pyzmq-26.3.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9176856f36c34a8aa5c0b35ddf52a5d5cd8abeece57c2cd904cfddae3fd9acd3", size = 874003, upload-time = "2025-03-12T08:02:02.994Z" }, + { url = "https://files.pythonhosted.org/packages/97/d4/4dd152dbbaac35d4e1fe8e8fd26d73640fcd84ec9c3915b545692df1ffb7/pyzmq-26.3.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:49334faa749d55b77f084389a80654bf2e68ab5191c0235066f0140c1b670d64", size = 867989, upload-time = "2025-03-12T08:02:04.321Z" }, + { url = "https://files.pythonhosted.org/packages/a4/22/1c5dc761dff13981d27d8225aedb19e70ce9149d16cf0c97c7547570e986/pyzmq-26.3.0-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:fd30fc80fe96efb06bea21667c5793bbd65c0dc793187feb39b8f96990680b00", size = 1207989, upload-time = "2025-03-12T08:02:06.048Z" }, + { url = "https://files.pythonhosted.org/packages/03/89/227ffb9e30b3fbe8196e7c97704345feb750b468e852ab64b0d19fa89e1a/pyzmq-26.3.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:b2eddfbbfb473a62c3a251bb737a6d58d91907f6e1d95791431ebe556f47d916", size = 1520523, upload-time = "2025-03-12T08:02:07.764Z" }, + { url = "https://files.pythonhosted.org/packages/29/d3/e9b99b8404b6a470762cb947bc342e462a853a22ce0b0f2982c65a9b698f/pyzmq-26.3.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:70b3acb9ad729a53d4e751dace35404a024f188aad406013454216aba5485b4e", size = 1419912, upload-time = "2025-03-12T08:02:09.563Z" }, + { url = "https://files.pythonhosted.org/packages/bb/69/074e2cde8135cae9452778e644ea5c91493bc536367d956005fe83072f63/pyzmq-26.3.0-cp310-cp310-win32.whl", hash = "sha256:c1bd75d692cd7c6d862a98013bfdf06702783b75cffbf5dae06d718fecefe8f2", size = 583733, upload-time = "2025-03-12T08:02:11.647Z" }, + { url = "https://files.pythonhosted.org/packages/00/f0/55e57d40f6e21877e96507c0c2dd7e32afffc37b0dde7b834df1170cd749/pyzmq-26.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:d7165bcda0dbf203e5ad04d79955d223d84b2263df4db92f525ba370b03a12ab", size = 647229, upload-time = "2025-03-12T08:02:12.992Z" }, + { url = "https://files.pythonhosted.org/packages/38/1d/6e935b5f06d674c931540b29932a0dd5e1b9d29d047c2764a9c8c6f3ce08/pyzmq-26.3.0-cp310-cp310-win_arm64.whl", hash = "sha256:e34a63f71d2ecffb3c643909ad2d488251afeb5ef3635602b3448e609611a7ed", size = 561038, upload-time = "2025-03-12T08:02:14.305Z" }, + { url = "https://files.pythonhosted.org/packages/22/75/774e9a4a4291864dd37a03a7bfaf46a82d61cd36c16edd33a5739ad49be3/pyzmq-26.3.0-cp311-cp311-macosx_10_15_universal2.whl", hash = "sha256:2833602d9d42c94b9d0d2a44d2b382d3d3a4485be018ba19dddc401a464c617a", size = 1345893, upload-time = "2025-03-12T08:02:15.725Z" }, + { url = "https://files.pythonhosted.org/packages/ca/51/d3eedd2bd46ef851bea528d8a2688a5091183b27fc238801fcac70e80dbb/pyzmq-26.3.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d8270d104ec7caa0bdac246d31d48d94472033ceab5ba142881704350b28159c", size = 678261, upload-time = "2025-03-12T08:02:17.444Z" }, + { url = "https://files.pythonhosted.org/packages/de/5e/521d7c6613769dcc3ed5e44e7082938b6dab27fffe02755784e54e98e17b/pyzmq-26.3.0-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c208a977843d18d3bd185f323e4eaa912eb4869cb230947dc6edd8a27a4e558a", size = 915311, upload-time = "2025-03-12T08:02:18.912Z" }, + { url = "https://files.pythonhosted.org/packages/78/db/3be86dd82adc638a2eb07c3028c1747ead49a71d7d334980b007f593fd9f/pyzmq-26.3.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:eddc2be28a379c218e0d92e4a432805dcb0ca5870156a90b54c03cd9799f9f8a", size = 873193, upload-time = "2025-03-12T08:02:20.311Z" }, + { url = "https://files.pythonhosted.org/packages/63/1a/81a31920d5113113ccd50271649dd2d0cfcfe46925d8f8a196fe560ed0e6/pyzmq-26.3.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:c0b519fa2159c42272f8a244354a0e110d65175647e5185b04008ec00df9f079", size = 867648, upload-time = "2025-03-12T08:02:22.148Z" }, + { url = "https://files.pythonhosted.org/packages/55/79/bbf57979ff2d89b5465d7205db08de7222d2560edc11272eb054c5a68cb5/pyzmq-26.3.0-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:1595533de3a80bf8363372c20bafa963ec4bf9f2b8f539b1d9a5017f430b84c9", size = 1208475, upload-time = "2025-03-12T08:02:23.952Z" }, + { url = "https://files.pythonhosted.org/packages/50/fc/1246dfc4b165e7ff97ac3c4117bdd3747e03ebb62269f71f65e216bfac8b/pyzmq-26.3.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:bbef99eb8d18ba9a40f00e8836b8040cdcf0f2fa649684cf7a66339599919d21", size = 1519428, upload-time = "2025-03-12T08:02:25.706Z" }, + { url = "https://files.pythonhosted.org/packages/5f/9a/143aacb6b372b0e2d812aec73a06fc5df3e169a361d4302226f8563954c6/pyzmq-26.3.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:979486d444ca3c469cd1c7f6a619ce48ff08b3b595d451937db543754bfacb65", size = 1419530, upload-time = "2025-03-12T08:02:27.119Z" }, + { url = "https://files.pythonhosted.org/packages/47/f7/b437e77d496089e17e77866eb126dd97ea47041b58e53892f57e82869198/pyzmq-26.3.0-cp311-cp311-win32.whl", hash = "sha256:4b127cfe10b4c56e4285b69fd4b38ea1d368099ea4273d8fb349163fce3cd598", size = 582538, upload-time = "2025-03-12T08:02:28.576Z" }, + { url = "https://files.pythonhosted.org/packages/a1/2c/99a01a2d7865aaf44e47c2182cbdbc15da1f2e4cfee92dc8e1fb5114f993/pyzmq-26.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:cf736cc1298ef15280d9fcf7a25c09b05af016656856dc6fe5626fd8912658dd", size = 647989, upload-time = "2025-03-12T08:02:29.897Z" }, + { url = "https://files.pythonhosted.org/packages/60/b3/36ac1cb8fafeadff09935f4bdc1232e511af8f8893d6cebc7ceb93c6753a/pyzmq-26.3.0-cp311-cp311-win_arm64.whl", hash = "sha256:2dc46ec09f5d36f606ac8393303149e69d17121beee13c8dac25e2a2078e31c4", size = 561533, upload-time = "2025-03-12T08:02:31.265Z" }, + { url = "https://files.pythonhosted.org/packages/7b/03/7170c3814bb9106c1bca67700c731aaf1cd990fd2f0097c754acb600330e/pyzmq-26.3.0-cp312-cp312-macosx_10_15_universal2.whl", hash = "sha256:c80653332c6136da7f4d4e143975e74ac0fa14f851f716d90583bc19e8945cea", size = 1348354, upload-time = "2025-03-12T08:02:32.699Z" }, + { url = "https://files.pythonhosted.org/packages/74/f3/908b17f9111cdc764aef1de3d36026a2984c46ed90c3c2c85f28b66142f0/pyzmq-26.3.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6e317ee1d4528a03506cb1c282cd9db73660a35b3564096de37de7350e7d87a7", size = 671056, upload-time = "2025-03-12T08:02:34.086Z" }, + { url = "https://files.pythonhosted.org/packages/02/ad/afcb8484b65ceacd1609f709c2caeed31bd6c49261a7507cd5c175cc105f/pyzmq-26.3.0-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:943a22ebb3daacb45f76a9bcca9a7b74e7d94608c0c0505da30af900b998ca8d", size = 908597, upload-time = "2025-03-12T08:02:35.536Z" }, + { url = "https://files.pythonhosted.org/packages/a1/b5/4eeeae0aaaa6ef0c74cfa8b2273b53382bd858df6d99485f2fc8211e7002/pyzmq-26.3.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3fc9e71490d989144981ea21ef4fdfaa7b6aa84aff9632d91c736441ce2f6b00", size = 865260, upload-time = "2025-03-12T08:02:37.562Z" }, + { url = "https://files.pythonhosted.org/packages/74/6a/63db856e93e3a3c3dc98a1de28a902cf1b21c7b0d3856cd5931d7cfd30af/pyzmq-26.3.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:e281a8071a06888575a4eb523c4deeefdcd2f5fe4a2d47e02ac8bf3a5b49f695", size = 859916, upload-time = "2025-03-12T08:02:38.954Z" }, + { url = "https://files.pythonhosted.org/packages/e1/ce/d522c9b46ee3746d4b98c81969c568c2c6296e931a65f2c87104b645654c/pyzmq-26.3.0-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:be77efd735bb1064605be8dec6e721141c1421ef0b115ef54e493a64e50e9a52", size = 1201368, upload-time = "2025-03-12T08:02:40.774Z" }, + { url = "https://files.pythonhosted.org/packages/5a/56/29dcd3647a39e933eb489fda261a1e2700a59d4a9432889a85166e15651c/pyzmq-26.3.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:7a4ac2ffa34f1212dd586af90f4ba894e424f0cabb3a49cdcff944925640f6ac", size = 1512663, upload-time = "2025-03-12T08:02:42.2Z" }, + { url = "https://files.pythonhosted.org/packages/6b/36/7c570698127a43398ed1b1832dada59496e633115016addbce5eda9938a6/pyzmq-26.3.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:ba698c7c252af83b6bba9775035263f0df5f807f0404019916d4b71af8161f66", size = 1411693, upload-time = "2025-03-12T08:02:43.583Z" }, + { url = "https://files.pythonhosted.org/packages/de/54/51d39bef85a7cdbca36227f7defdbfcdc5011b8361a3bfc0e8df431f5a5d/pyzmq-26.3.0-cp312-cp312-win32.whl", hash = "sha256:214038aaa88e801e54c2ef0cfdb2e6df27eb05f67b477380a452b595c5ecfa37", size = 581244, upload-time = "2025-03-12T08:02:45.072Z" }, + { url = "https://files.pythonhosted.org/packages/f2/6a/9512b11a1d0c5648534f03d5ab0c3222f55dc9c192029c1cb00a0ca044e2/pyzmq-26.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:bad7fe0372e505442482ca3ccbc0d6f38dae81b1650f57a0aa6bbee18e7df495", size = 643559, upload-time = "2025-03-12T08:02:46.485Z" }, + { url = "https://files.pythonhosted.org/packages/27/9f/faf5c9cf91b61eeb82a5e919d024d3ac28a795c92cce817be264ccd757d3/pyzmq-26.3.0-cp312-cp312-win_arm64.whl", hash = "sha256:b7b578d604e79e99aa39495becea013fd043fa9f36e4b490efa951f3d847a24d", size = 557664, upload-time = "2025-03-12T08:02:47.896Z" }, + { url = "https://files.pythonhosted.org/packages/37/16/97b8c5107bfccb39120e611671a452c9ff6e8626fb3f8d4c15afd652b6ae/pyzmq-26.3.0-cp313-cp313-macosx_10_15_universal2.whl", hash = "sha256:fa85953df84beb7b8b73cb3ec3f5d92b62687a09a8e71525c6734e020edf56fd", size = 1345691, upload-time = "2025-03-12T08:02:49.508Z" }, + { url = "https://files.pythonhosted.org/packages/a5/61/d5572d95040c0bb5b31eed5b23f3f0f992d94e4e0de0cea62e3c7f3a85c1/pyzmq-26.3.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:209d09f0ab6ddbcebe64630d1e6ca940687e736f443c265ae15bc4bfad833597", size = 670622, upload-time = "2025-03-12T08:02:51.112Z" }, + { url = "https://files.pythonhosted.org/packages/1c/0c/f0235d27388aacf4ed8bcc1d574f6f2f629da0a20610faa0a8e9d363c2b0/pyzmq-26.3.0-cp313-cp313-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d35cc1086f1d4f907df85c6cceb2245cb39a04f69c3f375993363216134d76d4", size = 908683, upload-time = "2025-03-12T08:02:52.659Z" }, + { url = "https://files.pythonhosted.org/packages/cb/52/664828f9586c396b857eec088d208230463e3dc991a24df6adbad98fbaa3/pyzmq-26.3.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b380e9087078ba91e45fb18cdd0c25275ffaa045cf63c947be0ddae6186bc9d9", size = 865212, upload-time = "2025-03-12T08:02:54.187Z" }, + { url = "https://files.pythonhosted.org/packages/2b/14/213b2967030b7d7aecc32dd453830f98799b3cbf2b10a40232e9f22a6520/pyzmq-26.3.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:6d64e74143587efe7c9522bb74d1448128fdf9897cc9b6d8b9927490922fd558", size = 860068, upload-time = "2025-03-12T08:02:55.609Z" }, + { url = "https://files.pythonhosted.org/packages/aa/e5/ff50c8fade69d1c0469652832c626d1910668697642c10cb0e1b6183ef9a/pyzmq-26.3.0-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:efba4f53ac7752eea6d8ca38a4ddac579e6e742fba78d1e99c12c95cd2acfc64", size = 1201303, upload-time = "2025-03-12T08:02:57.073Z" }, + { url = "https://files.pythonhosted.org/packages/9a/e2/fff5e483be95ccc11a05781323e001e63ec15daec1d0f6f08de72ca534db/pyzmq-26.3.0-cp313-cp313-musllinux_1_1_i686.whl", hash = "sha256:9b0137a1c40da3b7989839f9b78a44de642cdd1ce20dcef341de174c8d04aa53", size = 1512892, upload-time = "2025-03-12T08:02:58.68Z" }, + { url = "https://files.pythonhosted.org/packages/21/75/cc44d276e43136e5692e487c3c019f816e11ed445261e434217c28cc98c4/pyzmq-26.3.0-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:a995404bd3982c089e57b428c74edd5bfc3b0616b3dbcd6a8e270f1ee2110f36", size = 1411736, upload-time = "2025-03-12T08:03:00.202Z" }, + { url = "https://files.pythonhosted.org/packages/ee/1c/d070cbc9a7961fe772641c51bb3798d88cb1f8e20ca718407363462624cf/pyzmq-26.3.0-cp313-cp313-win32.whl", hash = "sha256:240b1634b9e530ef6a277d95cbca1a6922f44dfddc5f0a3cd6c722a8de867f14", size = 581214, upload-time = "2025-03-12T08:03:02.412Z" }, + { url = "https://files.pythonhosted.org/packages/38/d3/91082f1151ff5b54e0bed40eb1a26f418530ab07ecaec4dbb83e3d9fa9a9/pyzmq-26.3.0-cp313-cp313-win_amd64.whl", hash = "sha256:fe67291775ea4c2883764ba467eb389c29c308c56b86c1e19e49c9e1ed0cbeca", size = 643412, upload-time = "2025-03-12T08:03:04.007Z" }, + { url = "https://files.pythonhosted.org/packages/e0/cf/dabe68dfdf3e67bea6152eeec4b251cf899ee5b853cfb5c97e4719f9e6e9/pyzmq-26.3.0-cp313-cp313-win_arm64.whl", hash = "sha256:73ca9ae9a9011b714cf7650450cd9c8b61a135180b708904f1f0a05004543dce", size = 557444, upload-time = "2025-03-12T08:03:05.53Z" }, + { url = "https://files.pythonhosted.org/packages/c0/56/e7576ac71c1566da4f4ec586351462a2bb202143fb074bf56df8fe85dcc3/pyzmq-26.3.0-cp313-cp313t-macosx_10_15_universal2.whl", hash = "sha256:fea7efbd7e49af9d7e5ed6c506dfc7de3d1a628790bd3a35fd0e3c904dc7d464", size = 1340288, upload-time = "2025-03-12T08:03:07.638Z" }, + { url = "https://files.pythonhosted.org/packages/f1/ab/0bca97e94d420b5908968bc479e51c3686a9f80d8893450eefcd673b1b1d/pyzmq-26.3.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c4430c7cba23bb0e2ee203eee7851c1654167d956fc6d4b3a87909ccaf3c5825", size = 662462, upload-time = "2025-03-12T08:03:10.039Z" }, + { url = "https://files.pythonhosted.org/packages/ee/be/99e89b55863808da322ac3ab52d8e135dcf2241094aaa468bfe2923d5194/pyzmq-26.3.0-cp313-cp313t-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:016d89bee8c7d566fad75516b4e53ec7c81018c062d4c51cd061badf9539be52", size = 896464, upload-time = "2025-03-12T08:03:11.51Z" }, + { url = "https://files.pythonhosted.org/packages/38/d4/a4be06a313c8d6a5fe1d92975db30aca85f502e867fca392532e06a28c3c/pyzmq-26.3.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:04bfe59852d76d56736bfd10ac1d49d421ab8ed11030b4a0332900691507f557", size = 853432, upload-time = "2025-03-12T08:03:12.948Z" }, + { url = "https://files.pythonhosted.org/packages/12/e6/e608b4c34106bbf5b3b382662ea90a43b2e23df0aa9c1f0fd4e21168d523/pyzmq-26.3.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:1fe05bd0d633a0f672bb28cb8b4743358d196792e1caf04973b7898a0d70b046", size = 845884, upload-time = "2025-03-12T08:03:14.429Z" }, + { url = "https://files.pythonhosted.org/packages/c3/a9/d5e6355308ba529d9cd3576ee8bb3b2e2b726571748f515fbb8559401f5b/pyzmq-26.3.0-cp313-cp313t-musllinux_1_1_aarch64.whl", hash = "sha256:2aa1a9f236d5b835fb8642f27de95f9edcfd276c4bc1b6ffc84f27c6fb2e2981", size = 1191454, upload-time = "2025-03-12T08:03:16.348Z" }, + { url = "https://files.pythonhosted.org/packages/6a/9a/a21dc6c73ac242e425709c1e0049368d8f5db5de7c1102a45f93f5c492b3/pyzmq-26.3.0-cp313-cp313t-musllinux_1_1_i686.whl", hash = "sha256:21399b31753bf321043ea60c360ed5052cc7be20739785b1dff1820f819e35b3", size = 1500397, upload-time = "2025-03-12T08:03:17.918Z" }, + { url = "https://files.pythonhosted.org/packages/87/88/0236056156da0278c9ca2e2562463643597808b5bbd6c34009ba217e7e92/pyzmq-26.3.0-cp313-cp313t-musllinux_1_1_x86_64.whl", hash = "sha256:d015efcd96aca8882057e7e6f06224f79eecd22cad193d3e6a0a91ec67590d1f", size = 1398401, upload-time = "2025-03-12T08:03:19.493Z" }, + { url = "https://files.pythonhosted.org/packages/7e/ec/2e02dde6b1a436b02a6c0e3cb64c779bf6e76cc41c12131f29d9b10a088f/pyzmq-26.3.0-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:ad03f4252d9041b0635c37528dfa3f44b39f46024ae28c8567f7423676ee409b", size = 835672, upload-time = "2025-03-12T08:03:52.729Z" }, + { url = "https://files.pythonhosted.org/packages/22/ee/30c2c3f162912cff31af2b9d87295533d16f867e7621bd6f9ed62d9cc807/pyzmq-26.3.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0f3dfb68cf7bf4cfdf34283a75848e077c5defa4907506327282afe92780084d", size = 570837, upload-time = "2025-03-12T08:03:54.3Z" }, + { url = "https://files.pythonhosted.org/packages/51/a5/5aead624f5f1033dab9bdaf3e2bc692a8042fcb59355c919a2c042061780/pyzmq-26.3.0-pp310-pypy310_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:356ec0e39c5a9cda872b65aca1fd8a5d296ffdadf8e2442b70ff32e73ef597b1", size = 799508, upload-time = "2025-03-12T08:03:55.841Z" }, + { url = "https://files.pythonhosted.org/packages/ca/8a/dcc0a24cfed80cc004abcba710077147ec9178a12865914e73a60a70cb62/pyzmq-26.3.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:749d671b0eec8e738bbf0b361168369d8c682b94fcd458c20741dc4d69ef5278", size = 758001, upload-time = "2025-03-12T08:03:57.87Z" }, + { url = "https://files.pythonhosted.org/packages/1a/74/f18e63540340f5c740396eb6408d154a84e9f0e9e1ae931b192bf2aa7425/pyzmq-26.3.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:f950f17ae608e0786298340163cac25a4c5543ef25362dd5ddb6dcb10b547be9", size = 556425, upload-time = "2025-03-12T08:03:59.479Z" }, + { url = "https://files.pythonhosted.org/packages/f4/c6/e36b2a2ff6534cb1d1f6b3fb37901ac54675caf7b2e1239613aa40d1d217/pyzmq-26.3.0-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:b4fc9903a73c25be9d5fe45c87faababcf3879445efa16140146b08fccfac017", size = 835670, upload-time = "2025-03-12T08:04:01.037Z" }, + { url = "https://files.pythonhosted.org/packages/1d/b9/8059c5af94b245068e7f7379c08c7e409ec854139d6021aecf2c111d8547/pyzmq-26.3.0-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c15b69af22030960ac63567e98ad8221cddf5d720d9cf03d85021dfd452324ef", size = 570838, upload-time = "2025-03-12T08:04:03.125Z" }, + { url = "https://files.pythonhosted.org/packages/80/a4/f0a4266ff2d94a87f7c32895b1716f9ac0edc0471d518462beeb0a9a94b5/pyzmq-26.3.0-pp311-pypy311_pp73-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2cf9ab0dff4dbaa2e893eb608373c97eb908e53b7d9793ad00ccbd082c0ee12f", size = 799507, upload-time = "2025-03-12T08:04:04.704Z" }, + { url = "https://files.pythonhosted.org/packages/78/14/3d7d459f496fab8e487b23423ccba57abf7153a4fde0c3e000500fa02ff8/pyzmq-26.3.0-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3ec332675f6a138db57aad93ae6387953763f85419bdbd18e914cb279ee1c451", size = 758002, upload-time = "2025-03-12T08:04:06.678Z" }, + { url = "https://files.pythonhosted.org/packages/22/65/cc1f0e1db1290770285430e36d51767e620487523e6a04094be637e55698/pyzmq-26.3.0-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:eb96568a22fe070590942cd4780950e2172e00fb033a8b76e47692583b1bd97c", size = 556425, upload-time = "2025-03-12T08:04:08.37Z" }, ] [[package]] name = "rdkit" -version = "2025.3.5" +version = "2024.9.6" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "pillow" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/f7/dd/325588bac4d71a7cc4acfe9c5565b896de4cf9d46cce9da325858a1075c4/rdkit-2025.3.5-cp310-cp310-macosx_10_15_x86_64.whl", hash = "sha256:323fd21ce62f8b6ca34cebb8d0d7cbd9c4d757f5bff76143e2d2c04c94c1304a", size = 31161247, upload-time = "2025-08-01T08:37:56.364Z" }, - { url = "https://files.pythonhosted.org/packages/e3/5f/7341e6b03e1cf53ee2e82ebd7602e25c0596667c47a3e0bcd6c76ca039ed/rdkit-2025.3.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:1b8aee3c6c9fb0a8a3596a50595e4d88d1e02a70e9cb2d4626ab6134d4a5eceb", size = 28703723, upload-time = "2025-08-01T08:38:01.039Z" }, - { url = "https://files.pythonhosted.org/packages/96/d0/50ce769d29262ca6fa6801dfcc30e9ce6b031a5590aec617f884ac79e209/rdkit-2025.3.5-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:1c70bb2ebcb42125ba6500d692a8ba4738f518f4c334ed1dd0885499617bff53", size = 35372548, upload-time = "2025-08-01T08:38:04.997Z" }, - { url = "https://files.pythonhosted.org/packages/6f/f9/dae8b9785e2cf0361a211166e28f673f283ed2d4829c30b8614537eb66f2/rdkit-2025.3.5-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:07a6d212ce88a91bf8bc187cbd4c98d8a5af3e7c01a7a64b4215f46825456b80", size = 36272060, upload-time = "2025-08-01T08:38:09.233Z" }, - { url = "https://files.pythonhosted.org/packages/e2/bd/af721a3b97d4a59b7d209c71561e9e1c17c08dc5cd2f6f26b69deea94ab1/rdkit-2025.3.5-cp310-cp310-win_amd64.whl", hash = "sha256:1bbc026a8bc873d890b9f79f5df9eec3f322f580dfb1aec9b50aaef8409be6f4", size = 23517085, upload-time = "2025-08-01T08:38:12.942Z" }, - { url = "https://files.pythonhosted.org/packages/9d/04/7724512f61e30cefb6e4af54cd3a1ccf622196acb199bc351017f207747e/rdkit-2025.3.5-cp311-cp311-macosx_10_15_x86_64.whl", hash = "sha256:010c18f84f474f8ff97b5fec6571296ab17475c4378beb5870aa8ea412cc61f2", size = 31161777, upload-time = "2025-08-01T08:38:16.552Z" }, - { url = "https://files.pythonhosted.org/packages/6d/f3/c6c8352008b202351b57191cd9805090a2077d7ebcc301a7e65b38e62ce4/rdkit-2025.3.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:fb10ebcd3f2901a48922bfecb00de64c4589fd121af79b985768952a65436413", size = 28704087, upload-time = "2025-08-01T08:38:20.704Z" }, - { url = "https://files.pythonhosted.org/packages/21/e1/5bec35b13ed5f88d11841bd73dbb32bdae02a317e094ab8737ee1e020920/rdkit-2025.3.5-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:9303de802d03f07936ffab6cd7fe4b9f488314e33fd0eaf3d29edd78cfd9cd88", size = 35367566, upload-time = "2025-08-01T08:38:24.535Z" }, - { url = "https://files.pythonhosted.org/packages/40/17/d3df2ef69504418432364cd42add91da85012985320a571116d2a5492079/rdkit-2025.3.5-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:eb92045567bfda71721f607d52a1d1a7a5dd101b853d62c13c01995f4052b6fb", size = 36271424, upload-time = "2025-08-01T08:38:28.8Z" }, - { url = "https://files.pythonhosted.org/packages/54/7a/98ed651841c61ef761f401e569addf213c28867ab84862fbe4b7d8539f70/rdkit-2025.3.5-cp311-cp311-win_amd64.whl", hash = "sha256:58e7d4498c53205dfb1fa35850daadd7bb18da2791317f251216dc53616e8754", size = 23517947, upload-time = "2025-08-01T08:38:32.187Z" }, - { url = "https://files.pythonhosted.org/packages/42/3c/a377c7117ae5b983c121594e60e8c511a97ebe2c8836548754ca013f5b6c/rdkit-2025.3.5-cp312-cp312-macosx_10_15_x86_64.whl", hash = "sha256:0419a2a19135cfbdd46fc4794dd45a69138e80632883489a499cfc0b750d6c95", size = 31237587, upload-time = "2025-08-01T08:38:35.805Z" }, - { url = "https://files.pythonhosted.org/packages/e7/1d/e293366515c9ec157312db7aa0e4bd5b9f8f00fea526afca033fa8e5df7b/rdkit-2025.3.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:f8a1459a595c2d20928bb477aff155fa3480b09df8a866be9a6c7d334dd2ca2e", size = 28746581, upload-time = "2025-08-01T08:38:40.223Z" }, - { url = "https://files.pythonhosted.org/packages/ba/c7/972600433ab966db58ceb3e1cda4f688bb15ffe4078219323682e6ad9ccd/rdkit-2025.3.5-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:5c5f69325802014aa5a535b8c680636f1c1fb9201fbabf250220bbae429721b7", size = 35243285, upload-time = "2025-08-01T08:38:49.396Z" }, - { url = "https://files.pythonhosted.org/packages/c4/c7/79030fec5fbc72ee46fc35724b707025c0c45c7b0864050d28bff7977baf/rdkit-2025.3.5-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:36b43611c0e455cb5b5f08207fd27d83db405d4f88a5bb8df093aa38c55527a1", size = 36192701, upload-time = "2025-08-01T08:38:53.442Z" }, - { url = "https://files.pythonhosted.org/packages/39/38/0e6ce9905c798abb776a7de1b9582c41359468d9efc82c939146d32264eb/rdkit-2025.3.5-cp312-cp312-win_amd64.whl", hash = "sha256:7f02fb64f0cf5e798357bc49f762eb9854c25c363579b660348557fd1f8f5c00", size = 23537346, upload-time = "2025-08-01T08:38:56.913Z" }, - { url = "https://files.pythonhosted.org/packages/47/2e/c7cc6ea1ce2eca4e03c25d7b7b61e23646d96fff138f472bf7e0015bdd42/rdkit-2025.3.5-cp313-cp313-macosx_10_15_x86_64.whl", hash = "sha256:b34bac98122e0e95c958c96a923cfb8f34b554bff652a571bca1ed85d567d61a", size = 31236417, upload-time = "2025-08-01T08:39:00.325Z" }, - { url = "https://files.pythonhosted.org/packages/c6/1d/4cc7d00c388dde4cb4ce6ff7c4314f0f2c0e342d5b7a7b779fc12c32cf5a/rdkit-2025.3.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b564067066c97136c448c58ac796c66010d041a5fd5dccf8997461a1846e63b7", size = 28745429, upload-time = "2025-08-01T08:39:04.627Z" }, - { url = "https://files.pythonhosted.org/packages/28/ea/c8e6ea43a9f352b9a71e4417ddfc4a66f8bda7813e782a412c206d6bdfbd/rdkit-2025.3.5-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:252288a0d93c5c31344349a85d769d5fa6a7e5ee132efe1fffbc4a0d6d3833c6", size = 35241653, upload-time = "2025-08-01T08:39:08.345Z" }, - { url = "https://files.pythonhosted.org/packages/8b/6a/c6ec8e124bc621ce0706894f877c6f57d7c9c53cbc73510ac046d40b4dc7/rdkit-2025.3.5-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:65c636e4ea0e4bbf80261155e19f96de8e178271513f8823a4364e2939ee5791", size = 36192312, upload-time = "2025-08-01T08:39:12.253Z" }, - { url = "https://files.pythonhosted.org/packages/19/04/6d5c40e8a4b3edd0cef5fa9cda96e36739a7f3e6573d80722b719fd22e13/rdkit-2025.3.5-cp313-cp313-win_amd64.whl", hash = "sha256:8b2c51020f446adb0f56d0fcb38b45a01a665ef4523eea368ea30dc0026cbebd", size = 23535970, upload-time = "2025-08-01T08:39:16.131Z" }, - { url = "https://files.pythonhosted.org/packages/88/c3/548f36267ff114c5a76cea9603eecf51b24bb6a09599a41f19627b6f3662/rdkit-2025.3.5-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:5c84758f67ef8ba09cd8a6310686ffff237337ad65a38c64308ca119132c7914", size = 31238791, upload-time = "2025-08-01T08:39:19.429Z" }, - { url = "https://files.pythonhosted.org/packages/ff/c6/09bc6e443b5e3a526f295f060c743a48c1652fd050d5fead3efb52689102/rdkit-2025.3.5-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:07e2c35d3781bf013c16957b4d8daaa90e4914830ea6490291b276698d2a366d", size = 28763880, upload-time = "2025-08-01T08:39:23.058Z" }, - { url = "https://files.pythonhosted.org/packages/08/80/e90d6fff82972e82e1ff086562d0be7c90dc6367f9f351c8298fbbe88fc8/rdkit-2025.3.5-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:83930691cc76051ca01850d28cba7e21c0f8fcc43605a6355753f9df2fbf855a", size = 35276980, upload-time = "2025-08-01T08:39:27.331Z" }, - { url = "https://files.pythonhosted.org/packages/2b/c2/56cc024b8c9b34c460d0fa17d3180606a96e0e4b1614b9d7e07160ea4f56/rdkit-2025.3.5-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:193a4147598116655b978fe28d9995c864c445a751c9755ba39753a053bbb4a5", size = 36204963, upload-time = "2025-08-01T08:39:31.247Z" }, - { url = "https://files.pythonhosted.org/packages/2c/11/8dfd2d397d45c951571f575e539338f8f16cb85f0997b865b23211ebbc6c/rdkit-2025.3.5-cp314-cp314-win_amd64.whl", hash = "sha256:b715c506f99d3b29ea5eb903fa1fd6787fef567d8a249dfdc14f5f57f372f377", size = 24014579, upload-time = "2025-08-01T08:39:34.935Z" }, + { url = "https://files.pythonhosted.org/packages/a0/44/8908dc5e26248a1e27d5a0f149e17af2b5e85585a227e57d2d4f53f68d07/rdkit-2024.9.6-cp310-cp310-macosx_10_15_x86_64.whl", hash = "sha256:015c7200fffefdae53b59a4658e0323e2d16ac16b2b00174188f769a0cabe52a", size = 29960044, upload-time = "2025-03-12T10:34:47.284Z" }, + { url = "https://files.pythonhosted.org/packages/8f/2e/36a2cee64073df724c7fc635c969de604440d754c65ac1dd2627e5aacc16/rdkit-2024.9.6-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:ffe05be4354178b48bfd60ea40c847e6364e75e71893bd2c659da8819c8afab6", size = 27718856, upload-time = "2025-03-12T10:34:51.256Z" }, + { url = "https://files.pythonhosted.org/packages/13/28/c9abf09ade43a8ed6165bc1e7660cbefdf449f89cb400eb20377e84ceb2c/rdkit-2024.9.6-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:6e6bc9b6edc3e5efa88f796d8a86192b72b5d54ce9488f8ccb043e6b74308b79", size = 33528869, upload-time = "2025-03-12T10:34:54.957Z" }, + { url = "https://files.pythonhosted.org/packages/a3/f6/3a2278acc1d3831fd1d5b81ae355297be1c2ea8d8d0e200291aa2edfbdeb/rdkit-2024.9.6-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:2b5573055c8defbad7ce25db10786a56e0699faf3282daea21100c59f7af7298", size = 34344527, upload-time = "2025-03-12T10:34:59.482Z" }, + { url = "https://files.pythonhosted.org/packages/cb/61/18b1fec82a2c4ab60234eab6673174c7dbcfd82490fc834a4d2b10d42d8b/rdkit-2024.9.6-cp310-cp310-win_amd64.whl", hash = "sha256:9d64c865de57c15bbe7726f20a967016f2ac04f738889c83210091be944b512b", size = 22472377, upload-time = "2025-03-12T10:35:03.297Z" }, + { url = "https://files.pythonhosted.org/packages/cc/c1/b21d793ab572545917af8d00c7519ea1e0fc10d3e74ac936de4a06fc3649/rdkit-2024.9.6-cp311-cp311-macosx_10_15_x86_64.whl", hash = "sha256:bbb3fe860e5ccb8debad592ded991b2aca611c997640fc6e68b0f41beb648639", size = 29960558, upload-time = "2025-03-12T10:35:07.236Z" }, + { url = "https://files.pythonhosted.org/packages/72/64/f96f4a72052be7580ab825279cc71653a476c023c53828765be520db47e6/rdkit-2024.9.6-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:a28971024ba0300f7c566874212b8c2c8f7db984e85418eaceec5cb739bcab1d", size = 27719061, upload-time = "2025-03-12T10:35:11.167Z" }, + { url = "https://files.pythonhosted.org/packages/1d/2f/3befa6b3a061665351d5a58aba66e318175777c40cbfbfc511bd57d2aa7b/rdkit-2024.9.6-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:6e236da74308b19f695ed259b12b75e100e56593333f2b47cff2420e9746e571", size = 33524980, upload-time = "2025-03-12T10:35:15.419Z" }, + { url = "https://files.pythonhosted.org/packages/cc/3f/472c33312ca8a55242fc2cf6179809f4a967185e9dc6e76ea28ddd37a097/rdkit-2024.9.6-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:7b473a715711e7f1f747173fc18a16b02d5952eadbeedb09d05d1f0a6879c282", size = 34343341, upload-time = "2025-03-12T10:35:19.528Z" }, + { url = "https://files.pythonhosted.org/packages/94/85/d8d7da7ba8d00ad8977193f30d0062324aa9279e8af67e413674d3f60483/rdkit-2024.9.6-cp311-cp311-win_amd64.whl", hash = "sha256:f302e8a347debe24ffeeb303b8b031ef2e61265872a6450f46aa819b25f5c684", size = 22472536, upload-time = "2025-03-12T10:35:23.174Z" }, + { url = "https://files.pythonhosted.org/packages/b0/b4/54a543114b58bfbb0f659d30c7902d10e7fa908544691455b36e0783fa77/rdkit-2024.9.6-cp312-cp312-macosx_10_15_x86_64.whl", hash = "sha256:f57a2ad6ceff54917586f47a01ecbe43bbdf4a6f00defd746b500228752ccf84", size = 30035515, upload-time = "2025-03-12T10:35:27.512Z" }, + { url = "https://files.pythonhosted.org/packages/d1/5a/83deae41ff455e417f67e91393d3fae9f604e15fc633544412fa16d9bcc6/rdkit-2024.9.6-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:4f3931ff192aeee29ce41cf7c30650883fc10f1504af2a9dcc67d8554a0779d3", size = 27760806, upload-time = "2025-03-12T10:35:31.794Z" }, + { url = "https://files.pythonhosted.org/packages/95/41/c52930491f52d7bdb50e297d5064fdc959a82adea9f735a9ab7c412ac874/rdkit-2024.9.6-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:97661c84a9dc10797d197db1024757676e0ca29a306d5eec289bbfdf4c53c80c", size = 33414832, upload-time = "2025-03-12T10:35:36.122Z" }, + { url = "https://files.pythonhosted.org/packages/e9/f1/2f978f26b2c58716e70ca1e442e51c260b43e55c2eb64bd2ad76584b60da/rdkit-2024.9.6-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:82e91020e700979f6fc94a4722b47d770912d54cd9accb1c85961c57550cb945", size = 34276560, upload-time = "2025-03-12T10:35:40.561Z" }, + { url = "https://files.pythonhosted.org/packages/36/a7/738b3eb302b95ece0efea4d9b9c90d8dd7bb9ef529a56ee3bbbcfd239cde/rdkit-2024.9.6-cp312-cp312-win_amd64.whl", hash = "sha256:995df4e0b09a866fc628b6c9f1c10dca32726964ff1c7506cbf7975af0a9ba60", size = 22487552, upload-time = "2025-03-12T10:35:44.277Z" }, + { url = "https://files.pythonhosted.org/packages/07/be/e0fb0bf1301ce3add94a248a24b29c7784b7d504de4776554fafe3372c54/rdkit-2024.9.6-cp313-cp313-macosx_10_15_x86_64.whl", hash = "sha256:5001786e1f06749fa4cf7d6a9d23b36afc6d9b028d500777a3861c8a4d91238b", size = 30034429, upload-time = "2025-03-12T10:35:47.789Z" }, + { url = "https://files.pythonhosted.org/packages/d4/0e/918410c8975cb365ad51d3e487965c3c5ccfc3dea5715c81f0e89339cec0/rdkit-2024.9.6-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b2578458414600beb3a5691fc900e0640da3c571b30ed3dc76381cfe5caadc9e", size = 27759699, upload-time = "2025-03-12T10:35:52.004Z" }, + { url = "https://files.pythonhosted.org/packages/41/ee/9758cbcde26bacb820579ecbfbd3069ab8bd10b55e08f6731c208ce2eb3f/rdkit-2024.9.6-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:4fbf4d680a1070a15f6d9c7cf93218bed87c7110e4e7a467d1b449e430e20ff5", size = 33413200, upload-time = "2025-03-12T10:35:56.306Z" }, + { url = "https://files.pythonhosted.org/packages/d9/f9/6e5b8f6445450a3b01e28915f749f9d644a8b855f6e7b26c09f193b1978d/rdkit-2024.9.6-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:4b0efe61d9eee664dc19203478e3312e596be0d200c3bbd4ef1b584aef7146d3", size = 34275798, upload-time = "2025-03-12T10:36:00.826Z" }, + { url = "https://files.pythonhosted.org/packages/3f/9b/3c32fd6dbb1210a8e9dc83feee82778c8bf2ad483c9e8394138c25789110/rdkit-2024.9.6-cp313-cp313-win_amd64.whl", hash = "sha256:6c740c543b55f99d8d3bdea92c68617ef5669bcf2cc1652752a65f156f26dbc0", size = 22486705, upload-time = "2025-03-12T10:36:05.44Z" }, ] [[package]] @@ -3289,7 +3070,7 @@ wheels = [ [[package]] name = "requests" -version = "2.32.4" +version = "2.32.3" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "certifi" }, @@ -3297,9 +3078,9 @@ dependencies = [ { name = "idna" }, { name = "urllib3" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/e1/0a/929373653770d8a0d7ea76c37de6e41f11eb07559b103b1c02cafb3f7cf8/requests-2.32.4.tar.gz", hash = "sha256:27d0316682c8a29834d3264820024b62a36942083d52caf2f14c0591336d3422", size = 135258, upload-time = "2025-06-09T16:43:07.34Z" } +sdist = { url = "https://files.pythonhosted.org/packages/63/70/2bf7780ad2d390a8d301ad0b550f1581eadbd9a20f896afe06353c2a2913/requests-2.32.3.tar.gz", hash = "sha256:55365417734eb18255590a9ff9eb97e9e1da868d4ccd6402399eaf68af20a760", size = 131218, upload-time = "2024-05-29T15:37:49.536Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/7c/e4/56027c4a6b4ae70ca9de302488c5ca95ad4a39e190093d6c1a8ace08341b/requests-2.32.4-py3-none-any.whl", hash = "sha256:27babd3cda2a6d50b30443204ee89830707d396671944c998b5975b031ac2b2c", size = 64847, upload-time = "2025-06-09T16:43:05.728Z" }, + { url = "https://files.pythonhosted.org/packages/f9/9b/335f9764261e915ed497fcdeb11df5dfd6f7bf257d4a6a2a686d80da4d54/requests-2.32.3-py3-none-any.whl", hash = "sha256:70761cfe03c773ceb22aa2f671b4757976145175cdfca038c02654d061d6dcc6", size = 64928, upload-time = "2024-05-29T15:37:47.027Z" }, ] [[package]] @@ -3318,128 +3099,87 @@ wheels = [ [[package]] name = "rpds-py" -version = "0.26.0" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/a5/aa/4456d84bbb54adc6a916fb10c9b374f78ac840337644e4a5eda229c81275/rpds_py-0.26.0.tar.gz", hash = "sha256:20dae58a859b0906f0685642e591056f1e787f3a8b39c8e8749a45dc7d26bdb0", size = 27385, upload-time = "2025-07-01T15:57:13.958Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/b9/31/1459645f036c3dfeacef89e8e5825e430c77dde8489f3b99eaafcd4a60f5/rpds_py-0.26.0-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:4c70c70f9169692b36307a95f3d8c0a9fcd79f7b4a383aad5eaa0e9718b79b37", size = 372466, upload-time = "2025-07-01T15:53:40.55Z" }, - { url = "https://files.pythonhosted.org/packages/dd/ff/3d0727f35836cc8773d3eeb9a46c40cc405854e36a8d2e951f3a8391c976/rpds_py-0.26.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:777c62479d12395bfb932944e61e915741e364c843afc3196b694db3d669fcd0", size = 357825, upload-time = "2025-07-01T15:53:42.247Z" }, - { url = "https://files.pythonhosted.org/packages/bf/ce/badc5e06120a54099ae287fa96d82cbb650a5f85cf247ffe19c7b157fd1f/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec671691e72dff75817386aa02d81e708b5a7ec0dec6669ec05213ff6b77e1bd", size = 381530, upload-time = "2025-07-01T15:53:43.585Z" }, - { url = "https://files.pythonhosted.org/packages/1e/a5/fa5d96a66c95d06c62d7a30707b6a4cfec696ab8ae280ee7be14e961e118/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:6a1cb5d6ce81379401bbb7f6dbe3d56de537fb8235979843f0d53bc2e9815a79", size = 396933, upload-time = "2025-07-01T15:53:45.78Z" }, - { url = "https://files.pythonhosted.org/packages/00/a7/7049d66750f18605c591a9db47d4a059e112a0c9ff8de8daf8fa0f446bba/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4f789e32fa1fb6a7bf890e0124e7b42d1e60d28ebff57fe806719abb75f0e9a3", size = 513973, upload-time = "2025-07-01T15:53:47.085Z" }, - { url = "https://files.pythonhosted.org/packages/0e/f1/528d02c7d6b29d29fac8fd784b354d3571cc2153f33f842599ef0cf20dd2/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9c55b0a669976cf258afd718de3d9ad1b7d1fe0a91cd1ab36f38b03d4d4aeaaf", size = 402293, upload-time = "2025-07-01T15:53:48.117Z" }, - { url = "https://files.pythonhosted.org/packages/15/93/fde36cd6e4685df2cd08508f6c45a841e82f5bb98c8d5ecf05649522acb5/rpds_py-0.26.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c70d9ec912802ecfd6cd390dadb34a9578b04f9bcb8e863d0a7598ba5e9e7ccc", size = 383787, upload-time = "2025-07-01T15:53:50.874Z" }, - { url = "https://files.pythonhosted.org/packages/69/f2/5007553aaba1dcae5d663143683c3dfd03d9395289f495f0aebc93e90f24/rpds_py-0.26.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:3021933c2cb7def39d927b9862292e0f4c75a13d7de70eb0ab06efed4c508c19", size = 416312, upload-time = "2025-07-01T15:53:52.046Z" }, - { url = "https://files.pythonhosted.org/packages/8f/a7/ce52c75c1e624a79e48a69e611f1c08844564e44c85db2b6f711d76d10ce/rpds_py-0.26.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:8a7898b6ca3b7d6659e55cdac825a2e58c638cbf335cde41f4619e290dd0ad11", size = 558403, upload-time = "2025-07-01T15:53:53.192Z" }, - { url = "https://files.pythonhosted.org/packages/79/d5/e119db99341cc75b538bf4cb80504129fa22ce216672fb2c28e4a101f4d9/rpds_py-0.26.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:12bff2ad9447188377f1b2794772f91fe68bb4bbfa5a39d7941fbebdbf8c500f", size = 588323, upload-time = "2025-07-01T15:53:54.336Z" }, - { url = "https://files.pythonhosted.org/packages/93/94/d28272a0b02f5fe24c78c20e13bbcb95f03dc1451b68e7830ca040c60bd6/rpds_py-0.26.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:191aa858f7d4902e975d4cf2f2d9243816c91e9605070aeb09c0a800d187e323", size = 554541, upload-time = "2025-07-01T15:53:55.469Z" }, - { url = "https://files.pythonhosted.org/packages/93/e0/8c41166602f1b791da892d976057eba30685486d2e2c061ce234679c922b/rpds_py-0.26.0-cp310-cp310-win32.whl", hash = "sha256:b37a04d9f52cb76b6b78f35109b513f6519efb481d8ca4c321f6a3b9580b3f45", size = 220442, upload-time = "2025-07-01T15:53:56.524Z" }, - { url = "https://files.pythonhosted.org/packages/87/f0/509736bb752a7ab50fb0270c2a4134d671a7b3038030837e5536c3de0e0b/rpds_py-0.26.0-cp310-cp310-win_amd64.whl", hash = "sha256:38721d4c9edd3eb6670437d8d5e2070063f305bfa2d5aa4278c51cedcd508a84", size = 231314, upload-time = "2025-07-01T15:53:57.842Z" }, - { url = "https://files.pythonhosted.org/packages/09/4c/4ee8f7e512030ff79fda1df3243c88d70fc874634e2dbe5df13ba4210078/rpds_py-0.26.0-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:9e8cb77286025bdb21be2941d64ac6ca016130bfdcd228739e8ab137eb4406ed", size = 372610, upload-time = "2025-07-01T15:53:58.844Z" }, - { url = "https://files.pythonhosted.org/packages/fa/9d/3dc16be00f14fc1f03c71b1d67c8df98263ab2710a2fbd65a6193214a527/rpds_py-0.26.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5e09330b21d98adc8ccb2dbb9fc6cb434e8908d4c119aeaa772cb1caab5440a0", size = 358032, upload-time = "2025-07-01T15:53:59.985Z" }, - { url = "https://files.pythonhosted.org/packages/e7/5a/7f1bf8f045da2866324a08ae80af63e64e7bfaf83bd31f865a7b91a58601/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2c9c1b92b774b2e68d11193dc39620d62fd8ab33f0a3c77ecdabe19c179cdbc1", size = 381525, upload-time = "2025-07-01T15:54:01.162Z" }, - { url = "https://files.pythonhosted.org/packages/45/8a/04479398c755a066ace10e3d158866beb600867cacae194c50ffa783abd0/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:824e6d3503ab990d7090768e4dfd9e840837bae057f212ff9f4f05ec6d1975e7", size = 397089, upload-time = "2025-07-01T15:54:02.319Z" }, - { url = "https://files.pythonhosted.org/packages/72/88/9203f47268db488a1b6d469d69c12201ede776bb728b9d9f29dbfd7df406/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8ad7fd2258228bf288f2331f0a6148ad0186b2e3643055ed0db30990e59817a6", size = 514255, upload-time = "2025-07-01T15:54:03.38Z" }, - { url = "https://files.pythonhosted.org/packages/f5/b4/01ce5d1e853ddf81fbbd4311ab1eff0b3cf162d559288d10fd127e2588b5/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0dc23bbb3e06ec1ea72d515fb572c1fea59695aefbffb106501138762e1e915e", size = 402283, upload-time = "2025-07-01T15:54:04.923Z" }, - { url = "https://files.pythonhosted.org/packages/34/a2/004c99936997bfc644d590a9defd9e9c93f8286568f9c16cdaf3e14429a7/rpds_py-0.26.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d80bf832ac7b1920ee29a426cdca335f96a2b5caa839811803e999b41ba9030d", size = 383881, upload-time = "2025-07-01T15:54:06.482Z" }, - { url = "https://files.pythonhosted.org/packages/05/1b/ef5fba4a8f81ce04c427bfd96223f92f05e6cd72291ce9d7523db3b03a6c/rpds_py-0.26.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:0919f38f5542c0a87e7b4afcafab6fd2c15386632d249e9a087498571250abe3", size = 415822, upload-time = "2025-07-01T15:54:07.605Z" }, - { url = "https://files.pythonhosted.org/packages/16/80/5c54195aec456b292f7bd8aa61741c8232964063fd8a75fdde9c1e982328/rpds_py-0.26.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d422b945683e409000c888e384546dbab9009bb92f7c0b456e217988cf316107", size = 558347, upload-time = "2025-07-01T15:54:08.591Z" }, - { url = "https://files.pythonhosted.org/packages/f2/1c/1845c1b1fd6d827187c43afe1841d91678d7241cbdb5420a4c6de180a538/rpds_py-0.26.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:77a7711fa562ba2da1aa757e11024ad6d93bad6ad7ede5afb9af144623e5f76a", size = 587956, upload-time = "2025-07-01T15:54:09.963Z" }, - { url = "https://files.pythonhosted.org/packages/2e/ff/9e979329dd131aa73a438c077252ddabd7df6d1a7ad7b9aacf6261f10faa/rpds_py-0.26.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:238e8c8610cb7c29460e37184f6799547f7e09e6a9bdbdab4e8edb90986a2318", size = 554363, upload-time = "2025-07-01T15:54:11.073Z" }, - { url = "https://files.pythonhosted.org/packages/00/8b/d78cfe034b71ffbe72873a136e71acc7a831a03e37771cfe59f33f6de8a2/rpds_py-0.26.0-cp311-cp311-win32.whl", hash = "sha256:893b022bfbdf26d7bedb083efeea624e8550ca6eb98bf7fea30211ce95b9201a", size = 220123, upload-time = "2025-07-01T15:54:12.382Z" }, - { url = "https://files.pythonhosted.org/packages/94/c1/3c8c94c7dd3905dbfde768381ce98778500a80db9924731d87ddcdb117e9/rpds_py-0.26.0-cp311-cp311-win_amd64.whl", hash = "sha256:87a5531de9f71aceb8af041d72fc4cab4943648d91875ed56d2e629bef6d4c03", size = 231732, upload-time = "2025-07-01T15:54:13.434Z" }, - { url = "https://files.pythonhosted.org/packages/67/93/e936fbed1b734eabf36ccb5d93c6a2e9246fbb13c1da011624b7286fae3e/rpds_py-0.26.0-cp311-cp311-win_arm64.whl", hash = "sha256:de2713f48c1ad57f89ac25b3cb7daed2156d8e822cf0eca9b96a6f990718cc41", size = 221917, upload-time = "2025-07-01T15:54:14.559Z" }, - { url = "https://files.pythonhosted.org/packages/ea/86/90eb87c6f87085868bd077c7a9938006eb1ce19ed4d06944a90d3560fce2/rpds_py-0.26.0-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:894514d47e012e794f1350f076c427d2347ebf82f9b958d554d12819849a369d", size = 363933, upload-time = "2025-07-01T15:54:15.734Z" }, - { url = "https://files.pythonhosted.org/packages/63/78/4469f24d34636242c924626082b9586f064ada0b5dbb1e9d096ee7a8e0c6/rpds_py-0.26.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:fc921b96fa95a097add244da36a1d9e4f3039160d1d30f1b35837bf108c21136", size = 350447, upload-time = "2025-07-01T15:54:16.922Z" }, - { url = "https://files.pythonhosted.org/packages/ad/91/c448ed45efdfdade82348d5e7995e15612754826ea640afc20915119734f/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3e1157659470aa42a75448b6e943c895be8c70531c43cb78b9ba990778955582", size = 384711, upload-time = "2025-07-01T15:54:18.101Z" }, - { url = "https://files.pythonhosted.org/packages/ec/43/e5c86fef4be7f49828bdd4ecc8931f0287b1152c0bb0163049b3218740e7/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:521ccf56f45bb3a791182dc6b88ae5f8fa079dd705ee42138c76deb1238e554e", size = 400865, upload-time = "2025-07-01T15:54:19.295Z" }, - { url = "https://files.pythonhosted.org/packages/55/34/e00f726a4d44f22d5c5fe2e5ddd3ac3d7fd3f74a175607781fbdd06fe375/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9def736773fd56b305c0eef698be5192c77bfa30d55a0e5885f80126c4831a15", size = 517763, upload-time = "2025-07-01T15:54:20.858Z" }, - { url = "https://files.pythonhosted.org/packages/52/1c/52dc20c31b147af724b16104500fba13e60123ea0334beba7b40e33354b4/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:cdad4ea3b4513b475e027be79e5a0ceac8ee1c113a1a11e5edc3c30c29f964d8", size = 406651, upload-time = "2025-07-01T15:54:22.508Z" }, - { url = "https://files.pythonhosted.org/packages/2e/77/87d7bfabfc4e821caa35481a2ff6ae0b73e6a391bb6b343db2c91c2b9844/rpds_py-0.26.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:82b165b07f416bdccf5c84546a484cc8f15137ca38325403864bfdf2b5b72f6a", size = 386079, upload-time = "2025-07-01T15:54:23.987Z" }, - { url = "https://files.pythonhosted.org/packages/e3/d4/7f2200c2d3ee145b65b3cddc4310d51f7da6a26634f3ac87125fd789152a/rpds_py-0.26.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:d04cab0a54b9dba4d278fe955a1390da3cf71f57feb78ddc7cb67cbe0bd30323", size = 421379, upload-time = "2025-07-01T15:54:25.073Z" }, - { url = "https://files.pythonhosted.org/packages/ae/13/9fdd428b9c820869924ab62236b8688b122baa22d23efdd1c566938a39ba/rpds_py-0.26.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:79061ba1a11b6a12743a2b0f72a46aa2758613d454aa6ba4f5a265cc48850158", size = 562033, upload-time = "2025-07-01T15:54:26.225Z" }, - { url = "https://files.pythonhosted.org/packages/f3/e1/b69686c3bcbe775abac3a4c1c30a164a2076d28df7926041f6c0eb5e8d28/rpds_py-0.26.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:f405c93675d8d4c5ac87364bb38d06c988e11028a64b52a47158a355079661f3", size = 591639, upload-time = "2025-07-01T15:54:27.424Z" }, - { url = "https://files.pythonhosted.org/packages/5c/c9/1e3d8c8863c84a90197ac577bbc3d796a92502124c27092413426f670990/rpds_py-0.26.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:dafd4c44b74aa4bed4b250f1aed165b8ef5de743bcca3b88fc9619b6087093d2", size = 557105, upload-time = "2025-07-01T15:54:29.93Z" }, - { url = "https://files.pythonhosted.org/packages/9f/c5/90c569649057622959f6dcc40f7b516539608a414dfd54b8d77e3b201ac0/rpds_py-0.26.0-cp312-cp312-win32.whl", hash = "sha256:3da5852aad63fa0c6f836f3359647870e21ea96cf433eb393ffa45263a170d44", size = 223272, upload-time = "2025-07-01T15:54:31.128Z" }, - { url = "https://files.pythonhosted.org/packages/7d/16/19f5d9f2a556cfed454eebe4d354c38d51c20f3db69e7b4ce6cff904905d/rpds_py-0.26.0-cp312-cp312-win_amd64.whl", hash = "sha256:cf47cfdabc2194a669dcf7a8dbba62e37a04c5041d2125fae0233b720da6f05c", size = 234995, upload-time = "2025-07-01T15:54:32.195Z" }, - { url = "https://files.pythonhosted.org/packages/83/f0/7935e40b529c0e752dfaa7880224771b51175fce08b41ab4a92eb2fbdc7f/rpds_py-0.26.0-cp312-cp312-win_arm64.whl", hash = "sha256:20ab1ae4fa534f73647aad289003f1104092890849e0266271351922ed5574f8", size = 223198, upload-time = "2025-07-01T15:54:33.271Z" }, - { url = "https://files.pythonhosted.org/packages/6a/67/bb62d0109493b12b1c6ab00de7a5566aa84c0e44217c2d94bee1bd370da9/rpds_py-0.26.0-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:696764a5be111b036256c0b18cd29783fab22154690fc698062fc1b0084b511d", size = 363917, upload-time = "2025-07-01T15:54:34.755Z" }, - { url = "https://files.pythonhosted.org/packages/4b/f3/34e6ae1925a5706c0f002a8d2d7f172373b855768149796af87bd65dcdb9/rpds_py-0.26.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1e6c15d2080a63aaed876e228efe4f814bc7889c63b1e112ad46fdc8b368b9e1", size = 350073, upload-time = "2025-07-01T15:54:36.292Z" }, - { url = "https://files.pythonhosted.org/packages/75/83/1953a9d4f4e4de7fd0533733e041c28135f3c21485faaef56a8aadbd96b5/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:390e3170babf42462739a93321e657444f0862c6d722a291accc46f9d21ed04e", size = 384214, upload-time = "2025-07-01T15:54:37.469Z" }, - { url = "https://files.pythonhosted.org/packages/48/0e/983ed1b792b3322ea1d065e67f4b230f3b96025f5ce3878cc40af09b7533/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:7da84c2c74c0f5bc97d853d9e17bb83e2dcafcff0dc48286916001cc114379a1", size = 400113, upload-time = "2025-07-01T15:54:38.954Z" }, - { url = "https://files.pythonhosted.org/packages/69/7f/36c0925fff6f660a80be259c5b4f5e53a16851f946eb080351d057698528/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4c5fe114a6dd480a510b6d3661d09d67d1622c4bf20660a474507aaee7eeeee9", size = 515189, upload-time = "2025-07-01T15:54:40.57Z" }, - { url = "https://files.pythonhosted.org/packages/13/45/cbf07fc03ba7a9b54662c9badb58294ecfb24f828b9732970bd1a431ed5c/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3100b3090269f3a7ea727b06a6080d4eb7439dca4c0e91a07c5d133bb1727ea7", size = 406998, upload-time = "2025-07-01T15:54:43.025Z" }, - { url = "https://files.pythonhosted.org/packages/6c/b0/8fa5e36e58657997873fd6a1cf621285ca822ca75b4b3434ead047daa307/rpds_py-0.26.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2c03c9b0c64afd0320ae57de4c982801271c0c211aa2d37f3003ff5feb75bb04", size = 385903, upload-time = "2025-07-01T15:54:44.752Z" }, - { url = "https://files.pythonhosted.org/packages/4b/f7/b25437772f9f57d7a9fbd73ed86d0dcd76b4c7c6998348c070d90f23e315/rpds_py-0.26.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:5963b72ccd199ade6ee493723d18a3f21ba7d5b957017607f815788cef50eaf1", size = 419785, upload-time = "2025-07-01T15:54:46.043Z" }, - { url = "https://files.pythonhosted.org/packages/a7/6b/63ffa55743dfcb4baf2e9e77a0b11f7f97ed96a54558fcb5717a4b2cd732/rpds_py-0.26.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:9da4e873860ad5bab3291438525cae80169daecbfafe5657f7f5fb4d6b3f96b9", size = 561329, upload-time = "2025-07-01T15:54:47.64Z" }, - { url = "https://files.pythonhosted.org/packages/2f/07/1f4f5e2886c480a2346b1e6759c00278b8a69e697ae952d82ae2e6ee5db0/rpds_py-0.26.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:5afaddaa8e8c7f1f7b4c5c725c0070b6eed0228f705b90a1732a48e84350f4e9", size = 590875, upload-time = "2025-07-01T15:54:48.9Z" }, - { url = "https://files.pythonhosted.org/packages/cc/bc/e6639f1b91c3a55f8c41b47d73e6307051b6e246254a827ede730624c0f8/rpds_py-0.26.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:4916dc96489616a6f9667e7526af8fa693c0fdb4f3acb0e5d9f4400eb06a47ba", size = 556636, upload-time = "2025-07-01T15:54:50.619Z" }, - { url = "https://files.pythonhosted.org/packages/05/4c/b3917c45566f9f9a209d38d9b54a1833f2bb1032a3e04c66f75726f28876/rpds_py-0.26.0-cp313-cp313-win32.whl", hash = "sha256:2a343f91b17097c546b93f7999976fd6c9d5900617aa848c81d794e062ab302b", size = 222663, upload-time = "2025-07-01T15:54:52.023Z" }, - { url = "https://files.pythonhosted.org/packages/e0/0b/0851bdd6025775aaa2365bb8de0697ee2558184c800bfef8d7aef5ccde58/rpds_py-0.26.0-cp313-cp313-win_amd64.whl", hash = "sha256:0a0b60701f2300c81b2ac88a5fb893ccfa408e1c4a555a77f908a2596eb875a5", size = 234428, upload-time = "2025-07-01T15:54:53.692Z" }, - { url = "https://files.pythonhosted.org/packages/ed/e8/a47c64ed53149c75fb581e14a237b7b7cd18217e969c30d474d335105622/rpds_py-0.26.0-cp313-cp313-win_arm64.whl", hash = "sha256:257d011919f133a4746958257f2c75238e3ff54255acd5e3e11f3ff41fd14256", size = 222571, upload-time = "2025-07-01T15:54:54.822Z" }, - { url = "https://files.pythonhosted.org/packages/89/bf/3d970ba2e2bcd17d2912cb42874107390f72873e38e79267224110de5e61/rpds_py-0.26.0-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:529c8156d7506fba5740e05da8795688f87119cce330c244519cf706a4a3d618", size = 360475, upload-time = "2025-07-01T15:54:56.228Z" }, - { url = "https://files.pythonhosted.org/packages/82/9f/283e7e2979fc4ec2d8ecee506d5a3675fce5ed9b4b7cb387ea5d37c2f18d/rpds_py-0.26.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:f53ec51f9d24e9638a40cabb95078ade8c99251945dad8d57bf4aabe86ecee35", size = 346692, upload-time = "2025-07-01T15:54:58.561Z" }, - { url = "https://files.pythonhosted.org/packages/e3/03/7e50423c04d78daf391da3cc4330bdb97042fc192a58b186f2d5deb7befd/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7ab504c4d654e4a29558eaa5bb8cea5fdc1703ea60a8099ffd9c758472cf913f", size = 379415, upload-time = "2025-07-01T15:54:59.751Z" }, - { url = "https://files.pythonhosted.org/packages/57/00/d11ee60d4d3b16808432417951c63df803afb0e0fc672b5e8d07e9edaaae/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:fd0641abca296bc1a00183fe44f7fced8807ed49d501f188faa642d0e4975b83", size = 391783, upload-time = "2025-07-01T15:55:00.898Z" }, - { url = "https://files.pythonhosted.org/packages/08/b3/1069c394d9c0d6d23c5b522e1f6546b65793a22950f6e0210adcc6f97c3e/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:69b312fecc1d017b5327afa81d4da1480f51c68810963a7336d92203dbb3d4f1", size = 512844, upload-time = "2025-07-01T15:55:02.201Z" }, - { url = "https://files.pythonhosted.org/packages/08/3b/c4fbf0926800ed70b2c245ceca99c49f066456755f5d6eb8863c2c51e6d0/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c741107203954f6fc34d3066d213d0a0c40f7bb5aafd698fb39888af277c70d8", size = 402105, upload-time = "2025-07-01T15:55:03.698Z" }, - { url = "https://files.pythonhosted.org/packages/1c/b0/db69b52ca07413e568dae9dc674627a22297abb144c4d6022c6d78f1e5cc/rpds_py-0.26.0-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc3e55a7db08dc9a6ed5fb7103019d2c1a38a349ac41901f9f66d7f95750942f", size = 383440, upload-time = "2025-07-01T15:55:05.398Z" }, - { url = "https://files.pythonhosted.org/packages/4c/e1/c65255ad5b63903e56b3bb3ff9dcc3f4f5c3badde5d08c741ee03903e951/rpds_py-0.26.0-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:9e851920caab2dbcae311fd28f4313c6953993893eb5c1bb367ec69d9a39e7ed", size = 412759, upload-time = "2025-07-01T15:55:08.316Z" }, - { url = "https://files.pythonhosted.org/packages/e4/22/bb731077872377a93c6e93b8a9487d0406c70208985831034ccdeed39c8e/rpds_py-0.26.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:dfbf280da5f876d0b00c81f26bedce274e72a678c28845453885a9b3c22ae632", size = 556032, upload-time = "2025-07-01T15:55:09.52Z" }, - { url = "https://files.pythonhosted.org/packages/e0/8b/393322ce7bac5c4530fb96fc79cc9ea2f83e968ff5f6e873f905c493e1c4/rpds_py-0.26.0-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:1cc81d14ddfa53d7f3906694d35d54d9d3f850ef8e4e99ee68bc0d1e5fed9a9c", size = 585416, upload-time = "2025-07-01T15:55:11.216Z" }, - { url = "https://files.pythonhosted.org/packages/49/ae/769dc372211835bf759319a7aae70525c6eb523e3371842c65b7ef41c9c6/rpds_py-0.26.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:dca83c498b4650a91efcf7b88d669b170256bf8017a5db6f3e06c2bf031f57e0", size = 554049, upload-time = "2025-07-01T15:55:13.004Z" }, - { url = "https://files.pythonhosted.org/packages/6b/f9/4c43f9cc203d6ba44ce3146246cdc38619d92c7bd7bad4946a3491bd5b70/rpds_py-0.26.0-cp313-cp313t-win32.whl", hash = "sha256:4d11382bcaf12f80b51d790dee295c56a159633a8e81e6323b16e55d81ae37e9", size = 218428, upload-time = "2025-07-01T15:55:14.486Z" }, - { url = "https://files.pythonhosted.org/packages/7e/8b/9286b7e822036a4a977f2f1e851c7345c20528dbd56b687bb67ed68a8ede/rpds_py-0.26.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ff110acded3c22c033e637dd8896e411c7d3a11289b2edf041f86663dbc791e9", size = 231524, upload-time = "2025-07-01T15:55:15.745Z" }, - { url = "https://files.pythonhosted.org/packages/55/07/029b7c45db910c74e182de626dfdae0ad489a949d84a468465cd0ca36355/rpds_py-0.26.0-cp314-cp314-macosx_10_12_x86_64.whl", hash = "sha256:da619979df60a940cd434084355c514c25cf8eb4cf9a508510682f6c851a4f7a", size = 364292, upload-time = "2025-07-01T15:55:17.001Z" }, - { url = "https://files.pythonhosted.org/packages/13/d1/9b3d3f986216b4d1f584878dca15ce4797aaf5d372d738974ba737bf68d6/rpds_py-0.26.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:ea89a2458a1a75f87caabefe789c87539ea4e43b40f18cff526052e35bbb4fdf", size = 350334, upload-time = "2025-07-01T15:55:18.922Z" }, - { url = "https://files.pythonhosted.org/packages/18/98/16d5e7bc9ec715fa9668731d0cf97f6b032724e61696e2db3d47aeb89214/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:feac1045b3327a45944e7dcbeb57530339f6b17baff154df51ef8b0da34c8c12", size = 384875, upload-time = "2025-07-01T15:55:20.399Z" }, - { url = "https://files.pythonhosted.org/packages/f9/13/aa5e2b1ec5ab0e86a5c464d53514c0467bec6ba2507027d35fc81818358e/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:b818a592bd69bfe437ee8368603d4a2d928c34cffcdf77c2e761a759ffd17d20", size = 399993, upload-time = "2025-07-01T15:55:21.729Z" }, - { url = "https://files.pythonhosted.org/packages/17/03/8021810b0e97923abdbab6474c8b77c69bcb4b2c58330777df9ff69dc559/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1a8b0dd8648709b62d9372fc00a57466f5fdeefed666afe3fea5a6c9539a0331", size = 516683, upload-time = "2025-07-01T15:55:22.918Z" }, - { url = "https://files.pythonhosted.org/packages/dc/b1/da8e61c87c2f3d836954239fdbbfb477bb7b54d74974d8f6fcb34342d166/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:6d3498ad0df07d81112aa6ec6c95a7e7b1ae00929fb73e7ebee0f3faaeabad2f", size = 408825, upload-time = "2025-07-01T15:55:24.207Z" }, - { url = "https://files.pythonhosted.org/packages/38/bc/1fc173edaaa0e52c94b02a655db20697cb5fa954ad5a8e15a2c784c5cbdd/rpds_py-0.26.0-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:24a4146ccb15be237fdef10f331c568e1b0e505f8c8c9ed5d67759dac58ac246", size = 387292, upload-time = "2025-07-01T15:55:25.554Z" }, - { url = "https://files.pythonhosted.org/packages/7c/eb/3a9bb4bd90867d21916f253caf4f0d0be7098671b6715ad1cead9fe7bab9/rpds_py-0.26.0-cp314-cp314-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:a9a63785467b2d73635957d32a4f6e73d5e4df497a16a6392fa066b753e87387", size = 420435, upload-time = "2025-07-01T15:55:27.798Z" }, - { url = "https://files.pythonhosted.org/packages/cd/16/e066dcdb56f5632713445271a3f8d3d0b426d51ae9c0cca387799df58b02/rpds_py-0.26.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:de4ed93a8c91debfd5a047be327b7cc8b0cc6afe32a716bbbc4aedca9e2a83af", size = 562410, upload-time = "2025-07-01T15:55:29.057Z" }, - { url = "https://files.pythonhosted.org/packages/60/22/ddbdec7eb82a0dc2e455be44c97c71c232983e21349836ce9f272e8a3c29/rpds_py-0.26.0-cp314-cp314-musllinux_1_2_i686.whl", hash = "sha256:caf51943715b12af827696ec395bfa68f090a4c1a1d2509eb4e2cb69abbbdb33", size = 590724, upload-time = "2025-07-01T15:55:30.719Z" }, - { url = "https://files.pythonhosted.org/packages/2c/b4/95744085e65b7187d83f2fcb0bef70716a1ea0a9e5d8f7f39a86e5d83424/rpds_py-0.26.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:4a59e5bc386de021f56337f757301b337d7ab58baa40174fb150accd480bc953", size = 558285, upload-time = "2025-07-01T15:55:31.981Z" }, - { url = "https://files.pythonhosted.org/packages/37/37/6309a75e464d1da2559446f9c811aa4d16343cebe3dbb73701e63f760caa/rpds_py-0.26.0-cp314-cp314-win32.whl", hash = "sha256:92c8db839367ef16a662478f0a2fe13e15f2227da3c1430a782ad0f6ee009ec9", size = 223459, upload-time = "2025-07-01T15:55:33.312Z" }, - { url = "https://files.pythonhosted.org/packages/d9/6f/8e9c11214c46098b1d1391b7e02b70bb689ab963db3b19540cba17315291/rpds_py-0.26.0-cp314-cp314-win_amd64.whl", hash = "sha256:b0afb8cdd034150d4d9f53926226ed27ad15b7f465e93d7468caaf5eafae0d37", size = 236083, upload-time = "2025-07-01T15:55:34.933Z" }, - { url = "https://files.pythonhosted.org/packages/47/af/9c4638994dd623d51c39892edd9d08e8be8220a4b7e874fa02c2d6e91955/rpds_py-0.26.0-cp314-cp314-win_arm64.whl", hash = "sha256:ca3f059f4ba485d90c8dc75cb5ca897e15325e4e609812ce57f896607c1c0867", size = 223291, upload-time = "2025-07-01T15:55:36.202Z" }, - { url = "https://files.pythonhosted.org/packages/4d/db/669a241144460474aab03e254326b32c42def83eb23458a10d163cb9b5ce/rpds_py-0.26.0-cp314-cp314t-macosx_10_12_x86_64.whl", hash = "sha256:5afea17ab3a126006dc2f293b14ffc7ef3c85336cf451564a0515ed7648033da", size = 361445, upload-time = "2025-07-01T15:55:37.483Z" }, - { url = "https://files.pythonhosted.org/packages/3b/2d/133f61cc5807c6c2fd086a46df0eb8f63a23f5df8306ff9f6d0fd168fecc/rpds_py-0.26.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:69f0c0a3df7fd3a7eec50a00396104bb9a843ea6d45fcc31c2d5243446ffd7a7", size = 347206, upload-time = "2025-07-01T15:55:38.828Z" }, - { url = "https://files.pythonhosted.org/packages/05/bf/0e8fb4c05f70273469eecf82f6ccf37248558526a45321644826555db31b/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:801a71f70f9813e82d2513c9a96532551fce1e278ec0c64610992c49c04c2dad", size = 380330, upload-time = "2025-07-01T15:55:40.175Z" }, - { url = "https://files.pythonhosted.org/packages/d4/a8/060d24185d8b24d3923322f8d0ede16df4ade226a74e747b8c7c978e3dd3/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:df52098cde6d5e02fa75c1f6244f07971773adb4a26625edd5c18fee906fa84d", size = 392254, upload-time = "2025-07-01T15:55:42.015Z" }, - { url = "https://files.pythonhosted.org/packages/b9/7b/7c2e8a9ee3e6bc0bae26bf29f5219955ca2fbb761dca996a83f5d2f773fe/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9bc596b30f86dc6f0929499c9e574601679d0341a0108c25b9b358a042f51bca", size = 516094, upload-time = "2025-07-01T15:55:43.603Z" }, - { url = "https://files.pythonhosted.org/packages/75/d6/f61cafbed8ba1499b9af9f1777a2a199cd888f74a96133d8833ce5eaa9c5/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9dfbe56b299cf5875b68eb6f0ebaadc9cac520a1989cac0db0765abfb3709c19", size = 402889, upload-time = "2025-07-01T15:55:45.275Z" }, - { url = "https://files.pythonhosted.org/packages/92/19/c8ac0a8a8df2dd30cdec27f69298a5c13e9029500d6d76718130f5e5be10/rpds_py-0.26.0-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ac64f4b2bdb4ea622175c9ab7cf09444e412e22c0e02e906978b3b488af5fde8", size = 384301, upload-time = "2025-07-01T15:55:47.098Z" }, - { url = "https://files.pythonhosted.org/packages/41/e1/6b1859898bc292a9ce5776016c7312b672da00e25cec74d7beced1027286/rpds_py-0.26.0-cp314-cp314t-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:181ef9b6bbf9845a264f9aa45c31836e9f3c1f13be565d0d010e964c661d1e2b", size = 412891, upload-time = "2025-07-01T15:55:48.412Z" }, - { url = "https://files.pythonhosted.org/packages/ef/b9/ceb39af29913c07966a61367b3c08b4f71fad841e32c6b59a129d5974698/rpds_py-0.26.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:49028aa684c144ea502a8e847d23aed5e4c2ef7cadfa7d5eaafcb40864844b7a", size = 557044, upload-time = "2025-07-01T15:55:49.816Z" }, - { url = "https://files.pythonhosted.org/packages/2f/27/35637b98380731a521f8ec4f3fd94e477964f04f6b2f8f7af8a2d889a4af/rpds_py-0.26.0-cp314-cp314t-musllinux_1_2_i686.whl", hash = "sha256:e5d524d68a474a9688336045bbf76cb0def88549c1b2ad9dbfec1fb7cfbe9170", size = 585774, upload-time = "2025-07-01T15:55:51.192Z" }, - { url = "https://files.pythonhosted.org/packages/52/d9/3f0f105420fecd18551b678c9a6ce60bd23986098b252a56d35781b3e7e9/rpds_py-0.26.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:c1851f429b822831bd2edcbe0cfd12ee9ea77868f8d3daf267b189371671c80e", size = 554886, upload-time = "2025-07-01T15:55:52.541Z" }, - { url = "https://files.pythonhosted.org/packages/6b/c5/347c056a90dc8dd9bc240a08c527315008e1b5042e7a4cf4ac027be9d38a/rpds_py-0.26.0-cp314-cp314t-win32.whl", hash = "sha256:7bdb17009696214c3b66bb3590c6d62e14ac5935e53e929bcdbc5a495987a84f", size = 219027, upload-time = "2025-07-01T15:55:53.874Z" }, - { url = "https://files.pythonhosted.org/packages/75/04/5302cea1aa26d886d34cadbf2dc77d90d7737e576c0065f357b96dc7a1a6/rpds_py-0.26.0-cp314-cp314t-win_amd64.whl", hash = "sha256:f14440b9573a6f76b4ee4770c13f0b5921f71dde3b6fcb8dabbefd13b7fe05d7", size = 232821, upload-time = "2025-07-01T15:55:55.167Z" }, - { url = "https://files.pythonhosted.org/packages/ef/9a/1f033b0b31253d03d785b0cd905bc127e555ab496ea6b4c7c2e1f951f2fd/rpds_py-0.26.0-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:3c0909c5234543ada2515c05dc08595b08d621ba919629e94427e8e03539c958", size = 373226, upload-time = "2025-07-01T15:56:16.578Z" }, - { url = "https://files.pythonhosted.org/packages/58/29/5f88023fd6aaaa8ca3c4a6357ebb23f6f07da6079093ccf27c99efce87db/rpds_py-0.26.0-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:c1fb0cda2abcc0ac62f64e2ea4b4e64c57dfd6b885e693095460c61bde7bb18e", size = 359230, upload-time = "2025-07-01T15:56:17.978Z" }, - { url = "https://files.pythonhosted.org/packages/6c/6c/13eaebd28b439da6964dde22712b52e53fe2824af0223b8e403249d10405/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:84d142d2d6cf9b31c12aa4878d82ed3b2324226270b89b676ac62ccd7df52d08", size = 382363, upload-time = "2025-07-01T15:56:19.977Z" }, - { url = "https://files.pythonhosted.org/packages/55/fc/3bb9c486b06da19448646f96147796de23c5811ef77cbfc26f17307b6a9d/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:a547e21c5610b7e9093d870be50682a6a6cf180d6da0f42c47c306073bfdbbf6", size = 397146, upload-time = "2025-07-01T15:56:21.39Z" }, - { url = "https://files.pythonhosted.org/packages/15/18/9d1b79eb4d18e64ba8bba9e7dec6f9d6920b639f22f07ee9368ca35d4673/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:35e9a70a0f335371275cdcd08bc5b8051ac494dd58bff3bbfb421038220dc871", size = 514804, upload-time = "2025-07-01T15:56:22.78Z" }, - { url = "https://files.pythonhosted.org/packages/4f/5a/175ad7191bdbcd28785204621b225ad70e85cdfd1e09cc414cb554633b21/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0dfa6115c6def37905344d56fb54c03afc49104e2ca473d5dedec0f6606913b4", size = 402820, upload-time = "2025-07-01T15:56:24.584Z" }, - { url = "https://files.pythonhosted.org/packages/11/45/6a67ecf6d61c4d4aff4bc056e864eec4b2447787e11d1c2c9a0242c6e92a/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:313cfcd6af1a55a286a3c9a25f64af6d0e46cf60bc5798f1db152d97a216ff6f", size = 384567, upload-time = "2025-07-01T15:56:26.064Z" }, - { url = "https://files.pythonhosted.org/packages/a1/ba/16589da828732b46454c61858950a78fe4c931ea4bf95f17432ffe64b241/rpds_py-0.26.0-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:f7bf2496fa563c046d05e4d232d7b7fd61346e2402052064b773e5c378bf6f73", size = 416520, upload-time = "2025-07-01T15:56:27.608Z" }, - { url = "https://files.pythonhosted.org/packages/81/4b/00092999fc7c0c266045e984d56b7314734cc400a6c6dc4d61a35f135a9d/rpds_py-0.26.0-pp310-pypy310_pp73-musllinux_1_2_aarch64.whl", hash = "sha256:aa81873e2c8c5aa616ab8e017a481a96742fdf9313c40f14338ca7dbf50cb55f", size = 559362, upload-time = "2025-07-01T15:56:29.078Z" }, - { url = "https://files.pythonhosted.org/packages/96/0c/43737053cde1f93ac4945157f7be1428724ab943e2132a0d235a7e161d4e/rpds_py-0.26.0-pp310-pypy310_pp73-musllinux_1_2_i686.whl", hash = "sha256:68ffcf982715f5b5b7686bdd349ff75d422e8f22551000c24b30eaa1b7f7ae84", size = 588113, upload-time = "2025-07-01T15:56:30.485Z" }, - { url = "https://files.pythonhosted.org/packages/46/46/8e38f6161466e60a997ed7e9951ae5de131dedc3cf778ad35994b4af823d/rpds_py-0.26.0-pp310-pypy310_pp73-musllinux_1_2_x86_64.whl", hash = "sha256:6188de70e190847bb6db3dc3981cbadff87d27d6fe9b4f0e18726d55795cee9b", size = 555429, upload-time = "2025-07-01T15:56:31.956Z" }, - { url = "https://files.pythonhosted.org/packages/2c/ac/65da605e9f1dd643ebe615d5bbd11b6efa1d69644fc4bf623ea5ae385a82/rpds_py-0.26.0-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:1c962145c7473723df9722ba4c058de12eb5ebedcb4e27e7d902920aa3831ee8", size = 231950, upload-time = "2025-07-01T15:56:33.337Z" }, - { url = "https://files.pythonhosted.org/packages/51/f2/b5c85b758a00c513bb0389f8fc8e61eb5423050c91c958cdd21843faa3e6/rpds_py-0.26.0-pp311-pypy311_pp73-macosx_10_12_x86_64.whl", hash = "sha256:f61a9326f80ca59214d1cceb0a09bb2ece5b2563d4e0cd37bfd5515c28510674", size = 373505, upload-time = "2025-07-01T15:56:34.716Z" }, - { url = "https://files.pythonhosted.org/packages/23/e0/25db45e391251118e915e541995bb5f5ac5691a3b98fb233020ba53afc9b/rpds_py-0.26.0-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:183f857a53bcf4b1b42ef0f57ca553ab56bdd170e49d8091e96c51c3d69ca696", size = 359468, upload-time = "2025-07-01T15:56:36.219Z" }, - { url = "https://files.pythonhosted.org/packages/0b/73/dd5ee6075bb6491be3a646b301dfd814f9486d924137a5098e61f0487e16/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:941c1cfdf4799d623cf3aa1d326a6b4fdb7a5799ee2687f3516738216d2262fb", size = 382680, upload-time = "2025-07-01T15:56:37.644Z" }, - { url = "https://files.pythonhosted.org/packages/2f/10/84b522ff58763a5c443f5bcedc1820240e454ce4e620e88520f04589e2ea/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:72a8d9564a717ee291f554eeb4bfeafe2309d5ec0aa6c475170bdab0f9ee8e88", size = 397035, upload-time = "2025-07-01T15:56:39.241Z" }, - { url = "https://files.pythonhosted.org/packages/06/ea/8667604229a10a520fcbf78b30ccc278977dcc0627beb7ea2c96b3becef0/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:511d15193cbe013619dd05414c35a7dedf2088fcee93c6bbb7c77859765bd4e8", size = 514922, upload-time = "2025-07-01T15:56:40.645Z" }, - { url = "https://files.pythonhosted.org/packages/24/e6/9ed5b625c0661c4882fc8cdf302bf8e96c73c40de99c31e0b95ed37d508c/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:aea1f9741b603a8d8fedb0ed5502c2bc0accbc51f43e2ad1337fe7259c2b77a5", size = 402822, upload-time = "2025-07-01T15:56:42.137Z" }, - { url = "https://files.pythonhosted.org/packages/8a/58/212c7b6fd51946047fb45d3733da27e2fa8f7384a13457c874186af691b1/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4019a9d473c708cf2f16415688ef0b4639e07abaa569d72f74745bbeffafa2c7", size = 384336, upload-time = "2025-07-01T15:56:44.239Z" }, - { url = "https://files.pythonhosted.org/packages/aa/f5/a40ba78748ae8ebf4934d4b88e77b98497378bc2c24ba55ebe87a4e87057/rpds_py-0.26.0-pp311-pypy311_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:093d63b4b0f52d98ebae33b8c50900d3d67e0666094b1be7a12fffd7f65de74b", size = 416871, upload-time = "2025-07-01T15:56:46.284Z" }, - { url = "https://files.pythonhosted.org/packages/d5/a6/33b1fc0c9f7dcfcfc4a4353daa6308b3ece22496ceece348b3e7a7559a09/rpds_py-0.26.0-pp311-pypy311_pp73-musllinux_1_2_aarch64.whl", hash = "sha256:2abe21d8ba64cded53a2a677e149ceb76dcf44284202d737178afe7ba540c1eb", size = 559439, upload-time = "2025-07-01T15:56:48.549Z" }, - { url = "https://files.pythonhosted.org/packages/71/2d/ceb3f9c12f8cfa56d34995097f6cd99da1325642c60d1b6680dd9df03ed8/rpds_py-0.26.0-pp311-pypy311_pp73-musllinux_1_2_i686.whl", hash = "sha256:4feb7511c29f8442cbbc28149a92093d32e815a28aa2c50d333826ad2a20fdf0", size = 588380, upload-time = "2025-07-01T15:56:50.086Z" }, - { url = "https://files.pythonhosted.org/packages/c8/ed/9de62c2150ca8e2e5858acf3f4f4d0d180a38feef9fdab4078bea63d8dba/rpds_py-0.26.0-pp311-pypy311_pp73-musllinux_1_2_x86_64.whl", hash = "sha256:e99685fc95d386da368013e7fb4269dd39c30d99f812a8372d62f244f662709c", size = 555334, upload-time = "2025-07-01T15:56:51.703Z" }, +version = "0.23.1" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/0a/79/2ce611b18c4fd83d9e3aecb5cba93e1917c050f556db39842889fa69b79f/rpds_py-0.23.1.tar.gz", hash = "sha256:7f3240dcfa14d198dba24b8b9cb3b108c06b68d45b7babd9eefc1038fdf7e707", size = 26806, upload-time = "2025-02-21T15:04:23.169Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/34/fe/e5326459863bd525122f4e9c80ac8d7c6cfa171b7518d04cc27c12c209b0/rpds_py-0.23.1-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:2a54027554ce9b129fc3d633c92fa33b30de9f08bc61b32c053dc9b537266fed", size = 372123, upload-time = "2025-02-21T15:01:14.816Z" }, + { url = "https://files.pythonhosted.org/packages/f9/db/f10a3795f7a89fb27594934012d21c61019bbeb516c5bdcfbbe9e9e617a7/rpds_py-0.23.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:b5ef909a37e9738d146519657a1aab4584018746a18f71c692f2f22168ece40c", size = 356778, upload-time = "2025-02-21T15:01:16.788Z" }, + { url = "https://files.pythonhosted.org/packages/21/27/0d3678ad7f432fa86f8fac5f5fc6496a4d2da85682a710d605219be20063/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3ee9d6f0b38efb22ad94c3b68ffebe4c47865cdf4b17f6806d6c674e1feb4246", size = 385775, upload-time = "2025-02-21T15:01:18.192Z" }, + { url = "https://files.pythonhosted.org/packages/99/a0/1786defa125b2ad228027f22dff26312ce7d1fee3c7c3c2682f403db2062/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:f7356a6da0562190558c4fcc14f0281db191cdf4cb96e7604c06acfcee96df15", size = 391181, upload-time = "2025-02-21T15:01:20.214Z" }, + { url = "https://files.pythonhosted.org/packages/f1/5c/1240934050a7ffd020a915486d0cc4c7f6e7a2442a77aedf13664db55d36/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:9441af1d25aed96901f97ad83d5c3e35e6cd21a25ca5e4916c82d7dd0490a4fa", size = 444607, upload-time = "2025-02-21T15:01:22.221Z" }, + { url = "https://files.pythonhosted.org/packages/b7/1b/cee6905b47817fd0a377716dbe4df35295de46df46ee2ff704538cc371b0/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:3d8abf7896a91fb97e7977d1aadfcc2c80415d6dc2f1d0fca5b8d0df247248f3", size = 445550, upload-time = "2025-02-21T15:01:24.742Z" }, + { url = "https://files.pythonhosted.org/packages/54/f7/f0821ca34032892d7a67fcd5042f50074ff2de64e771e10df01085c88d47/rpds_py-0.23.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1b08027489ba8fedde72ddd233a5ea411b85a6ed78175f40285bd401bde7466d", size = 386148, upload-time = "2025-02-21T15:01:26.23Z" }, + { url = "https://files.pythonhosted.org/packages/eb/ef/2afe53bc857c4bcba336acfd2629883a5746e7291023e017ac7fc98d85aa/rpds_py-0.23.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:fee513135b5a58f3bb6d89e48326cd5aa308e4bcdf2f7d59f67c861ada482bf8", size = 416780, upload-time = "2025-02-21T15:01:27.778Z" }, + { url = "https://files.pythonhosted.org/packages/ae/9a/38d2236cf669789b8a3e1a014c9b6a8d7b8925b952c92e7839ae2749f9ac/rpds_py-0.23.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:35d5631ce0af26318dba0ae0ac941c534453e42f569011585cb323b7774502a5", size = 558265, upload-time = "2025-02-21T15:01:28.979Z" }, + { url = "https://files.pythonhosted.org/packages/e6/0a/f2705530c42578f20ed0b5b90135eecb30eef6e2ba73e7ba69087fad2dba/rpds_py-0.23.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:a20cb698c4a59c534c6701b1c24a968ff2768b18ea2991f886bd8985ce17a89f", size = 585270, upload-time = "2025-02-21T15:01:30.879Z" }, + { url = "https://files.pythonhosted.org/packages/29/4e/3b597dc84ed82c3d757ac9aa620de224a94e06d2e102069795ae7e81c015/rpds_py-0.23.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:5e9c206a1abc27e0588cf8b7c8246e51f1a16a103734f7750830a1ccb63f557a", size = 553850, upload-time = "2025-02-21T15:01:32.269Z" }, + { url = "https://files.pythonhosted.org/packages/00/cc/6498b6f79e4375e6737247661e52a2d18f6accf4910e0c8da978674b4241/rpds_py-0.23.1-cp310-cp310-win32.whl", hash = "sha256:d9f75a06ecc68f159d5d7603b734e1ff6daa9497a929150f794013aa9f6e3f12", size = 220660, upload-time = "2025-02-21T15:01:33.643Z" }, + { url = "https://files.pythonhosted.org/packages/17/2b/08db023d23e8c7032c99d8d2a70d32e450a868ab73d16e3ff5290308a665/rpds_py-0.23.1-cp310-cp310-win_amd64.whl", hash = "sha256:f35eff113ad430b5272bbfc18ba111c66ff525828f24898b4e146eb479a2cdda", size = 232551, upload-time = "2025-02-21T15:01:35.529Z" }, + { url = "https://files.pythonhosted.org/packages/1c/67/6e5d4234bb9dee062ffca2a5f3c7cd38716317d6760ec235b175eed4de2c/rpds_py-0.23.1-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:b79f5ced71efd70414a9a80bbbfaa7160da307723166f09b69773153bf17c590", size = 372264, upload-time = "2025-02-21T15:01:37.918Z" }, + { url = "https://files.pythonhosted.org/packages/a7/0a/3dedb2daee8e783622427f5064e2d112751d8276ee73aa5409f000a132f4/rpds_py-0.23.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c9e799dac1ffbe7b10c1fd42fe4cd51371a549c6e108249bde9cd1200e8f59b4", size = 356883, upload-time = "2025-02-21T15:01:39.131Z" }, + { url = "https://files.pythonhosted.org/packages/ed/fc/e1acef44f9c24b05fe5434b235f165a63a52959ac655e3f7a55726cee1a4/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:721f9c4011b443b6e84505fc00cc7aadc9d1743f1c988e4c89353e19c4a968ee", size = 385624, upload-time = "2025-02-21T15:01:40.589Z" }, + { url = "https://files.pythonhosted.org/packages/97/0a/a05951f6465d01622720c03ef6ef31adfbe865653e05ed7c45837492f25e/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:f88626e3f5e57432e6191cd0c5d6d6b319b635e70b40be2ffba713053e5147dd", size = 391500, upload-time = "2025-02-21T15:01:42.584Z" }, + { url = "https://files.pythonhosted.org/packages/ea/2e/cca0583ec0690ea441dceae23c0673b99755710ea22f40bccf1e78f41481/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:285019078537949cecd0190f3690a0b0125ff743d6a53dfeb7a4e6787af154f5", size = 444869, upload-time = "2025-02-21T15:01:44.004Z" }, + { url = "https://files.pythonhosted.org/packages/cc/e6/95cda68b33a6d814d1e96b0e406d231ed16629101460d1740e92f03365e6/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b92f5654157de1379c509b15acec9d12ecf6e3bc1996571b6cb82a4302060447", size = 444930, upload-time = "2025-02-21T15:01:46.191Z" }, + { url = "https://files.pythonhosted.org/packages/5f/a7/e94cdb73411ae9c11414d3c7c9a6ad75d22ad4a8d094fb45a345ba9e3018/rpds_py-0.23.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e768267cbe051dd8d1c5305ba690bb153204a09bf2e3de3ae530de955f5b5580", size = 386254, upload-time = "2025-02-21T15:01:48.038Z" }, + { url = "https://files.pythonhosted.org/packages/dd/c5/a4a943d90a39e85efd1e04b1ad5129936786f9a9aa27bb7be8fc5d9d50c9/rpds_py-0.23.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:c5334a71f7dc1160382d45997e29f2637c02f8a26af41073189d79b95d3321f1", size = 417090, upload-time = "2025-02-21T15:01:50.252Z" }, + { url = "https://files.pythonhosted.org/packages/0c/a0/80d0013b12428d1fce0ab4e71829400b0a32caec12733c79e6109f843342/rpds_py-0.23.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d6adb81564af0cd428910f83fa7da46ce9ad47c56c0b22b50872bc4515d91966", size = 557639, upload-time = "2025-02-21T15:01:51.488Z" }, + { url = "https://files.pythonhosted.org/packages/a6/92/ec2e6980afb964a2cd7a99cbdef1f6c01116abe94b42cbe336ac93dd11c2/rpds_py-0.23.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:cafa48f2133d4daa028473ede7d81cd1b9f9e6925e9e4003ebdf77010ee02f35", size = 584572, upload-time = "2025-02-21T15:01:53.13Z" }, + { url = "https://files.pythonhosted.org/packages/3d/ce/75b6054db34a390789a82523790717b27c1bd735e453abb429a87c4f0f26/rpds_py-0.23.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:0fced9fd4a07a1ded1bac7e961ddd9753dd5d8b755ba8e05acba54a21f5f1522", size = 553028, upload-time = "2025-02-21T15:01:54.84Z" }, + { url = "https://files.pythonhosted.org/packages/cc/24/f45abe0418c06a5cba0f846e967aa27bac765acd927aabd857c21319b8cc/rpds_py-0.23.1-cp311-cp311-win32.whl", hash = "sha256:243241c95174b5fb7204c04595852fe3943cc41f47aa14c3828bc18cd9d3b2d6", size = 220862, upload-time = "2025-02-21T15:01:56.966Z" }, + { url = "https://files.pythonhosted.org/packages/2d/a6/3c0880e8bbfc36451ef30dc416266f6d2934705e468db5d21c8ba0ab6400/rpds_py-0.23.1-cp311-cp311-win_amd64.whl", hash = "sha256:11dd60b2ffddba85715d8a66bb39b95ddbe389ad2cfcf42c833f1bcde0878eaf", size = 232953, upload-time = "2025-02-21T15:02:00.297Z" }, + { url = "https://files.pythonhosted.org/packages/f3/8c/d17efccb9f5b9137ddea706664aebae694384ae1d5997c0202093e37185a/rpds_py-0.23.1-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:3902df19540e9af4cc0c3ae75974c65d2c156b9257e91f5101a51f99136d834c", size = 364369, upload-time = "2025-02-21T15:02:02.396Z" }, + { url = "https://files.pythonhosted.org/packages/6e/c0/ab030f696b5c573107115a88d8d73d80f03309e60952b64c584c70c659af/rpds_py-0.23.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:66f8d2a17e5838dd6fb9be6baaba8e75ae2f5fa6b6b755d597184bfcd3cb0eba", size = 349965, upload-time = "2025-02-21T15:02:04.527Z" }, + { url = "https://files.pythonhosted.org/packages/b3/55/b40170f5a079c4fb0b6a82b299689e66e744edca3c3375a8b160fb797660/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:112b8774b0b4ee22368fec42749b94366bd9b536f8f74c3d4175d4395f5cbd31", size = 389064, upload-time = "2025-02-21T15:02:06.633Z" }, + { url = "https://files.pythonhosted.org/packages/ab/1c/b03a912c59ec7c1e16b26e587b9dfa8ddff3b07851e781e8c46e908a365a/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:e0df046f2266e8586cf09d00588302a32923eb6386ced0ca5c9deade6af9a149", size = 397741, upload-time = "2025-02-21T15:02:08.195Z" }, + { url = "https://files.pythonhosted.org/packages/52/6f/151b90792b62fb6f87099bcc9044c626881fdd54e31bf98541f830b15cea/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0f3288930b947cbebe767f84cf618d2cbe0b13be476e749da0e6a009f986248c", size = 448784, upload-time = "2025-02-21T15:02:09.473Z" }, + { url = "https://files.pythonhosted.org/packages/71/2a/6de67c0c97ec7857e0e9e5cd7c52405af931b303eb1e5b9eff6c50fd9a2e/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ce473a2351c018b06dd8d30d5da8ab5a0831056cc53b2006e2a8028172c37ce5", size = 440203, upload-time = "2025-02-21T15:02:11.745Z" }, + { url = "https://files.pythonhosted.org/packages/db/5e/e759cd1c276d98a4b1f464b17a9bf66c65d29f8f85754e27e1467feaa7c3/rpds_py-0.23.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d550d7e9e7d8676b183b37d65b5cd8de13676a738973d330b59dc8312df9c5dc", size = 391611, upload-time = "2025-02-21T15:02:13.76Z" }, + { url = "https://files.pythonhosted.org/packages/1c/1e/2900358efcc0d9408c7289769cba4c0974d9db314aa884028ed7f7364f61/rpds_py-0.23.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:e14f86b871ea74c3fddc9a40e947d6a5d09def5adc2076ee61fb910a9014fb35", size = 423306, upload-time = "2025-02-21T15:02:15.096Z" }, + { url = "https://files.pythonhosted.org/packages/23/07/6c177e6d059f5d39689352d6c69a926ee4805ffdb6f06203570234d3d8f7/rpds_py-0.23.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:1bf5be5ba34e19be579ae873da515a2836a2166d8d7ee43be6ff909eda42b72b", size = 562323, upload-time = "2025-02-21T15:02:17.379Z" }, + { url = "https://files.pythonhosted.org/packages/70/e4/f9097fd1c02b516fff9850792161eb9fc20a2fd54762f3c69eae0bdb67cb/rpds_py-0.23.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:d7031d493c4465dbc8d40bd6cafefef4bd472b17db0ab94c53e7909ee781b9ef", size = 588351, upload-time = "2025-02-21T15:02:19.688Z" }, + { url = "https://files.pythonhosted.org/packages/87/39/5db3c6f326bfbe4576ae2af6435bd7555867d20ae690c786ff33659f293b/rpds_py-0.23.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:55ff4151cfd4bc635e51cfb1c59ac9f7196b256b12e3a57deb9e5742e65941ad", size = 557252, upload-time = "2025-02-21T15:02:21.875Z" }, + { url = "https://files.pythonhosted.org/packages/fd/14/2d5ad292f144fa79bafb78d2eb5b8a3a91c358b6065443cb9c49b5d1fedf/rpds_py-0.23.1-cp312-cp312-win32.whl", hash = "sha256:a9d3b728f5a5873d84cba997b9d617c6090ca5721caaa691f3b1a78c60adc057", size = 222181, upload-time = "2025-02-21T15:02:23.353Z" }, + { url = "https://files.pythonhosted.org/packages/a3/4f/0fce63e0f5cdd658e71e21abd17ac1bc9312741ebb8b3f74eeed2ebdf771/rpds_py-0.23.1-cp312-cp312-win_amd64.whl", hash = "sha256:b03a8d50b137ee758e4c73638b10747b7c39988eb8e6cd11abb7084266455165", size = 237426, upload-time = "2025-02-21T15:02:25.326Z" }, + { url = "https://files.pythonhosted.org/packages/13/9d/b8b2c0edffb0bed15be17b6d5ab06216f2f47f9ee49259c7e96a3ad4ca42/rpds_py-0.23.1-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:4caafd1a22e5eaa3732acb7672a497123354bef79a9d7ceed43387d25025e935", size = 363672, upload-time = "2025-02-21T15:02:26.528Z" }, + { url = "https://files.pythonhosted.org/packages/bd/c2/5056fa29e6894144d7ba4c938b9b0445f75836b87d2dd00ed4999dc45a8c/rpds_py-0.23.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:178f8a60fc24511c0eb756af741c476b87b610dba83270fce1e5a430204566a4", size = 349602, upload-time = "2025-02-21T15:02:27.82Z" }, + { url = "https://files.pythonhosted.org/packages/b0/bc/33779a1bb0ee32d8d706b173825aab75c628521d23ce72a7c1e6a6852f86/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c632419c3870507ca20a37c8f8f5352317aca097639e524ad129f58c125c61c6", size = 388746, upload-time = "2025-02-21T15:02:29.914Z" }, + { url = "https://files.pythonhosted.org/packages/62/0b/71db3e36b7780a619698ec82a9c87ab44ad7ca7f5480913e8a59ff76f050/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:698a79d295626ee292d1730bc2ef6e70a3ab135b1d79ada8fde3ed0047b65a10", size = 397076, upload-time = "2025-02-21T15:02:31.255Z" }, + { url = "https://files.pythonhosted.org/packages/bb/2e/494398f613edf77ba10a916b1ddea2acce42ab0e3b62e2c70ffc0757ce00/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:271fa2184cf28bdded86bb6217c8e08d3a169fe0bbe9be5e8d96e8476b707122", size = 448399, upload-time = "2025-02-21T15:02:32.637Z" }, + { url = "https://files.pythonhosted.org/packages/dd/53/4bd7f5779b1f463243ee5fdc83da04dd58a08f86e639dbffa7a35f969a84/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b91cceb5add79ee563bd1f70b30896bd63bc5f78a11c1f00a1e931729ca4f1f4", size = 439764, upload-time = "2025-02-21T15:02:34.372Z" }, + { url = "https://files.pythonhosted.org/packages/f6/55/b3c18c04a460d951bf8e91f2abf46ce5b6426fb69784166a6a25827cb90a/rpds_py-0.23.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f3a6cb95074777f1ecda2ca4fa7717caa9ee6e534f42b7575a8f0d4cb0c24013", size = 390662, upload-time = "2025-02-21T15:02:36.616Z" }, + { url = "https://files.pythonhosted.org/packages/2a/65/cc463044a3cbd616029b2aa87a651cdee8288d2fdd7780b2244845e934c1/rpds_py-0.23.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:50fb62f8d8364978478b12d5f03bf028c6bc2af04082479299139dc26edf4c64", size = 422680, upload-time = "2025-02-21T15:02:38.02Z" }, + { url = "https://files.pythonhosted.org/packages/fa/8e/1fa52990c7836d72e8d70cd7753f2362c72fbb0a49c1462e8c60e7176d0b/rpds_py-0.23.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:c8f7e90b948dc9dcfff8003f1ea3af08b29c062f681c05fd798e36daa3f7e3e8", size = 561792, upload-time = "2025-02-21T15:02:41.184Z" }, + { url = "https://files.pythonhosted.org/packages/57/b8/fe3b612979b1a29d0c77f8585903d8b3a292604b26d4b300e228b8ac6360/rpds_py-0.23.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:5b98b6c953e5c2bda51ab4d5b4f172617d462eebc7f4bfdc7c7e6b423f6da957", size = 588127, upload-time = "2025-02-21T15:02:42.641Z" }, + { url = "https://files.pythonhosted.org/packages/44/2d/fde474de516bbc4b9b230f43c98e7f8acc5da7fc50ceed8e7af27553d346/rpds_py-0.23.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:2893d778d4671ee627bac4037a075168b2673c57186fb1a57e993465dbd79a93", size = 556981, upload-time = "2025-02-21T15:02:43.969Z" }, + { url = "https://files.pythonhosted.org/packages/18/57/767deeb27b81370bbab8f74ef6e68d26c4ea99018f3c71a570e506fede85/rpds_py-0.23.1-cp313-cp313-win32.whl", hash = "sha256:2cfa07c346a7ad07019c33fb9a63cf3acb1f5363c33bc73014e20d9fe8b01cdd", size = 221936, upload-time = "2025-02-21T15:02:45.339Z" }, + { url = "https://files.pythonhosted.org/packages/7d/6c/3474cfdd3cafe243f97ab8474ea8949236eb2a1a341ca55e75ce00cd03da/rpds_py-0.23.1-cp313-cp313-win_amd64.whl", hash = "sha256:3aaf141d39f45322e44fc2c742e4b8b4098ead5317e5f884770c8df0c332da70", size = 237145, upload-time = "2025-02-21T15:02:47.461Z" }, + { url = "https://files.pythonhosted.org/packages/ec/77/e985064c624230f61efa0423759bb066da56ebe40c654f8b5ba225bd5d63/rpds_py-0.23.1-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:759462b2d0aa5a04be5b3e37fb8183615f47014ae6b116e17036b131985cb731", size = 359623, upload-time = "2025-02-21T15:02:49.02Z" }, + { url = "https://files.pythonhosted.org/packages/62/d9/a33dcbf62b29e40559e012d525bae7d516757cf042cc9234bd34ca4b6aeb/rpds_py-0.23.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:3e9212f52074fc9d72cf242a84063787ab8e21e0950d4d6709886fb62bcb91d5", size = 345900, upload-time = "2025-02-21T15:02:51.268Z" }, + { url = "https://files.pythonhosted.org/packages/92/eb/f81a4be6397861adb2cb868bb6a28a33292c2dcac567d1dc575226055e55/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9e9f3a3ac919406bc0414bbbd76c6af99253c507150191ea79fab42fdb35982a", size = 386426, upload-time = "2025-02-21T15:02:53.489Z" }, + { url = "https://files.pythonhosted.org/packages/09/47/1f810c9b5e83be005341201b5389f1d240dfa440346ea7189f9b3fd6961d/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:c04ca91dda8a61584165825907f5c967ca09e9c65fe8966ee753a3f2b019fe1e", size = 392314, upload-time = "2025-02-21T15:02:55.721Z" }, + { url = "https://files.pythonhosted.org/packages/83/bd/bc95831432fd6c46ed8001f01af26de0763a059d6d7e6d69e3c5bf02917a/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4ab923167cfd945abb9b51a407407cf19f5bee35001221f2911dc85ffd35ff4f", size = 447706, upload-time = "2025-02-21T15:02:59.224Z" }, + { url = "https://files.pythonhosted.org/packages/19/3e/567c04c226b1802dc6dc82cad3d53e1fa0a773258571c74ac5d8fbde97ed/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:ed6f011bedca8585787e5082cce081bac3d30f54520097b2411351b3574e1219", size = 437060, upload-time = "2025-02-21T15:03:00.953Z" }, + { url = "https://files.pythonhosted.org/packages/fe/77/a77d2c6afe27ae7d0d55fc32f6841502648070dc8d549fcc1e6d47ff8975/rpds_py-0.23.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6959bb9928c5c999aba4a3f5a6799d571ddc2c59ff49917ecf55be2bbb4e3722", size = 389347, upload-time = "2025-02-21T15:03:02.287Z" }, + { url = "https://files.pythonhosted.org/packages/3f/47/6b256ff20a74cfebeac790ab05586e0ac91f88e331125d4740a6c86fc26f/rpds_py-0.23.1-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:1ed7de3c86721b4e83ac440751329ec6a1102229aa18163f84c75b06b525ad7e", size = 415554, upload-time = "2025-02-21T15:03:03.816Z" }, + { url = "https://files.pythonhosted.org/packages/fc/29/d4572469a245bc9fc81e35166dca19fc5298d5c43e1a6dd64bf145045193/rpds_py-0.23.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:5fb89edee2fa237584e532fbf78f0ddd1e49a47c7c8cfa153ab4849dc72a35e6", size = 557418, upload-time = "2025-02-21T15:03:05.489Z" }, + { url = "https://files.pythonhosted.org/packages/9c/0a/68cf7228895b1a3f6f39f51b15830e62456795e61193d2c8b87fd48c60db/rpds_py-0.23.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:7e5413d2e2d86025e73f05510ad23dad5950ab8417b7fc6beaad99be8077138b", size = 583033, upload-time = "2025-02-21T15:03:07.493Z" }, + { url = "https://files.pythonhosted.org/packages/14/18/017ab41dcd6649ad5db7d00155b4c212b31ab05bd857d5ba73a1617984eb/rpds_py-0.23.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:d31ed4987d72aabdf521eddfb6a72988703c091cfc0064330b9e5f8d6a042ff5", size = 554880, upload-time = "2025-02-21T15:03:08.967Z" }, + { url = "https://files.pythonhosted.org/packages/2e/dd/17de89431268da8819d8d51ce67beac28d9b22fccf437bc5d6d2bcd1acdb/rpds_py-0.23.1-cp313-cp313t-win32.whl", hash = "sha256:f3429fb8e15b20961efca8c8b21432623d85db2228cc73fe22756c6637aa39e7", size = 219743, upload-time = "2025-02-21T15:03:10.429Z" }, + { url = "https://files.pythonhosted.org/packages/68/15/6d22d07e063ce5e9bfbd96db9ec2fbb4693591b4503e3a76996639474d02/rpds_py-0.23.1-cp313-cp313t-win_amd64.whl", hash = "sha256:d6f6512a90bd5cd9030a6237f5346f046c6f0e40af98657568fa45695d4de59d", size = 235415, upload-time = "2025-02-21T15:03:12.664Z" }, + { url = "https://files.pythonhosted.org/packages/95/a9/6fafd35fc6bac05f59bcbc800b57cef877911ff1c015397c519fec888642/rpds_py-0.23.1-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:c1f8afa346ccd59e4e5630d5abb67aba6a9812fddf764fd7eb11f382a345f8cc", size = 373463, upload-time = "2025-02-21T15:03:37.242Z" }, + { url = "https://files.pythonhosted.org/packages/5b/ac/44f00029b8fbe0903a19e9a87a9b86063bf8700df2cc58868373d378418c/rpds_py-0.23.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:fad784a31869747df4ac968a351e070c06ca377549e4ace94775aaa3ab33ee06", size = 358400, upload-time = "2025-02-21T15:03:38.752Z" }, + { url = "https://files.pythonhosted.org/packages/5e/9c/3da199346c68d785f10dccab123b74c8c5f73be3f742c9e33d1116e07931/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b5a96fcac2f18e5a0a23a75cd27ce2656c66c11c127b0318e508aab436b77428", size = 386815, upload-time = "2025-02-21T15:03:40.216Z" }, + { url = "https://files.pythonhosted.org/packages/d3/45/8f6533c33c0d33da8c2c8b2fb8f2ee90b23c05c679b86b0ac6aee4653749/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:3e77febf227a1dc3220159355dba68faa13f8dca9335d97504abf428469fb18b", size = 392974, upload-time = "2025-02-21T15:03:42.286Z" }, + { url = "https://files.pythonhosted.org/packages/ca/56/6a9ac1bf0455ba07385d8fe98c571c519b4f2000cff6581487bf9fab9272/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:26bb3e8de93443d55e2e748e9fd87deb5f8075ca7bc0502cfc8be8687d69a2ec", size = 446019, upload-time = "2025-02-21T15:03:43.811Z" }, + { url = "https://files.pythonhosted.org/packages/f4/83/5d9a3f9731cdccf49088bcc4ce821a5cf50bd1737cdad83e9959a7b9054d/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:db7707dde9143a67b8812c7e66aeb2d843fe33cc8e374170f4d2c50bd8f2472d", size = 445811, upload-time = "2025-02-21T15:03:45.218Z" }, + { url = "https://files.pythonhosted.org/packages/44/50/f2e0a98c62fc1fe68b176caca587714dc5c8bb2c3d1dd1eeb2bd4cc787ac/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1eedaaccc9bb66581d4ae7c50e15856e335e57ef2734dbc5fd8ba3e2a4ab3cb6", size = 388070, upload-time = "2025-02-21T15:03:46.905Z" }, + { url = "https://files.pythonhosted.org/packages/f2/d0/4981878f8f157e6dbea01d95e0119bf3d6b4c2c884fe64a9e6987f941104/rpds_py-0.23.1-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:28358c54fffadf0ae893f6c1050e8f8853e45df22483b7fff2f6ab6152f5d8bf", size = 419173, upload-time = "2025-02-21T15:03:48.552Z" }, + { url = "https://files.pythonhosted.org/packages/ce/13/fc971c470da96b270d2f64fedee987351bd935dc3016932a5cdcb1a88a2a/rpds_py-0.23.1-pp310-pypy310_pp73-musllinux_1_2_aarch64.whl", hash = "sha256:633462ef7e61d839171bf206551d5ab42b30b71cac8f10a64a662536e057fdef", size = 559048, upload-time = "2025-02-21T15:03:50.226Z" }, + { url = "https://files.pythonhosted.org/packages/42/02/be91e1de139ec8b4f9fec4192fd779ba48af281cfc762c0ca4c15b945484/rpds_py-0.23.1-pp310-pypy310_pp73-musllinux_1_2_i686.whl", hash = "sha256:a98f510d86f689fcb486dc59e6e363af04151e5260ad1bdddb5625c10f1e95f8", size = 584773, upload-time = "2025-02-21T15:03:52.678Z" }, + { url = "https://files.pythonhosted.org/packages/27/28/3af8a1956df3edc41d884267d766dc096496dafc83f02f764a475eca0b4a/rpds_py-0.23.1-pp310-pypy310_pp73-musllinux_1_2_x86_64.whl", hash = "sha256:e0397dd0b3955c61ef9b22838144aa4bef6f0796ba5cc8edfc64d468b93798b4", size = 555153, upload-time = "2025-02-21T15:03:55.21Z" }, + { url = "https://files.pythonhosted.org/packages/5e/bb/e45f51c4e1327dea3c72b846c6de129eebacb7a6cb309af7af35d0578c80/rpds_py-0.23.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:75307599f0d25bf6937248e5ac4e3bde5ea72ae6618623b86146ccc7845ed00b", size = 233827, upload-time = "2025-02-21T15:03:56.853Z" }, ] [[package]] @@ -3456,41 +3196,41 @@ wheels = [ [[package]] name = "ruff" -version = "0.12.7" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/a1/81/0bd3594fa0f690466e41bd033bdcdf86cba8288345ac77ad4afbe5ec743a/ruff-0.12.7.tar.gz", hash = "sha256:1fc3193f238bc2d7968772c82831a4ff69252f673be371fb49663f0068b7ec71", size = 5197814, upload-time = "2025-07-29T22:32:35.877Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/e1/d2/6cb35e9c85e7a91e8d22ab32ae07ac39cc34a71f1009a6f9e4a2a019e602/ruff-0.12.7-py3-none-linux_armv6l.whl", hash = "sha256:76e4f31529899b8c434c3c1dede98c4483b89590e15fb49f2d46183801565303", size = 11852189, upload-time = "2025-07-29T22:31:41.281Z" }, - { url = "https://files.pythonhosted.org/packages/63/5b/a4136b9921aa84638f1a6be7fb086f8cad0fde538ba76bda3682f2599a2f/ruff-0.12.7-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:789b7a03e72507c54fb3ba6209e4bb36517b90f1a3569ea17084e3fd295500fb", size = 12519389, upload-time = "2025-07-29T22:31:54.265Z" }, - { url = "https://files.pythonhosted.org/packages/a8/c9/3e24a8472484269b6b1821794141f879c54645a111ded4b6f58f9ab0705f/ruff-0.12.7-py3-none-macosx_11_0_arm64.whl", hash = "sha256:2e1c2a3b8626339bb6369116e7030a4cf194ea48f49b64bb505732a7fce4f4e3", size = 11743384, upload-time = "2025-07-29T22:31:59.575Z" }, - { url = "https://files.pythonhosted.org/packages/26/7c/458dd25deeb3452c43eaee853c0b17a1e84169f8021a26d500ead77964fd/ruff-0.12.7-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:32dec41817623d388e645612ec70d5757a6d9c035f3744a52c7b195a57e03860", size = 11943759, upload-time = "2025-07-29T22:32:01.95Z" }, - { url = "https://files.pythonhosted.org/packages/7f/8b/658798472ef260ca050e400ab96ef7e85c366c39cf3dfbef4d0a46a528b6/ruff-0.12.7-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:47ef751f722053a5df5fa48d412dbb54d41ab9b17875c6840a58ec63ff0c247c", size = 11654028, upload-time = "2025-07-29T22:32:04.367Z" }, - { url = "https://files.pythonhosted.org/packages/a8/86/9c2336f13b2a3326d06d39178fd3448dcc7025f82514d1b15816fe42bfe8/ruff-0.12.7-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a828a5fc25a3efd3e1ff7b241fd392686c9386f20e5ac90aa9234a5faa12c423", size = 13225209, upload-time = "2025-07-29T22:32:06.952Z" }, - { url = "https://files.pythonhosted.org/packages/76/69/df73f65f53d6c463b19b6b312fd2391dc36425d926ec237a7ed028a90fc1/ruff-0.12.7-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:5726f59b171111fa6a69d82aef48f00b56598b03a22f0f4170664ff4d8298efb", size = 14182353, upload-time = "2025-07-29T22:32:10.053Z" }, - { url = "https://files.pythonhosted.org/packages/58/1e/de6cda406d99fea84b66811c189b5ea139814b98125b052424b55d28a41c/ruff-0.12.7-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:74e6f5c04c4dd4aba223f4fe6e7104f79e0eebf7d307e4f9b18c18362124bccd", size = 13631555, upload-time = "2025-07-29T22:32:12.644Z" }, - { url = "https://files.pythonhosted.org/packages/6f/ae/625d46d5164a6cc9261945a5e89df24457dc8262539ace3ac36c40f0b51e/ruff-0.12.7-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:5d0bfe4e77fba61bf2ccadf8cf005d6133e3ce08793bbe870dd1c734f2699a3e", size = 12667556, upload-time = "2025-07-29T22:32:15.312Z" }, - { url = "https://files.pythonhosted.org/packages/55/bf/9cb1ea5e3066779e42ade8d0cd3d3b0582a5720a814ae1586f85014656b6/ruff-0.12.7-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:06bfb01e1623bf7f59ea749a841da56f8f653d641bfd046edee32ede7ff6c606", size = 12939784, upload-time = "2025-07-29T22:32:17.69Z" }, - { url = "https://files.pythonhosted.org/packages/55/7f/7ead2663be5627c04be83754c4f3096603bf5e99ed856c7cd29618c691bd/ruff-0.12.7-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:e41df94a957d50083fd09b916d6e89e497246698c3f3d5c681c8b3e7b9bb4ac8", size = 11771356, upload-time = "2025-07-29T22:32:20.134Z" }, - { url = "https://files.pythonhosted.org/packages/17/40/a95352ea16edf78cd3a938085dccc55df692a4d8ba1b3af7accbe2c806b0/ruff-0.12.7-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:4000623300563c709458d0ce170c3d0d788c23a058912f28bbadc6f905d67afa", size = 11612124, upload-time = "2025-07-29T22:32:22.645Z" }, - { url = "https://files.pythonhosted.org/packages/4d/74/633b04871c669e23b8917877e812376827c06df866e1677f15abfadc95cb/ruff-0.12.7-py3-none-musllinux_1_2_i686.whl", hash = "sha256:69ffe0e5f9b2cf2b8e289a3f8945b402a1b19eff24ec389f45f23c42a3dd6fb5", size = 12479945, upload-time = "2025-07-29T22:32:24.765Z" }, - { url = "https://files.pythonhosted.org/packages/be/34/c3ef2d7799c9778b835a76189c6f53c179d3bdebc8c65288c29032e03613/ruff-0.12.7-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:a07a5c8ffa2611a52732bdc67bf88e243abd84fe2d7f6daef3826b59abbfeda4", size = 12998677, upload-time = "2025-07-29T22:32:27.022Z" }, - { url = "https://files.pythonhosted.org/packages/77/ab/aca2e756ad7b09b3d662a41773f3edcbd262872a4fc81f920dc1ffa44541/ruff-0.12.7-py3-none-win32.whl", hash = "sha256:c928f1b2ec59fb77dfdf70e0419408898b63998789cc98197e15f560b9e77f77", size = 11756687, upload-time = "2025-07-29T22:32:29.381Z" }, - { url = "https://files.pythonhosted.org/packages/b4/71/26d45a5042bc71db22ddd8252ca9d01e9ca454f230e2996bb04f16d72799/ruff-0.12.7-py3-none-win_amd64.whl", hash = "sha256:9c18f3d707ee9edf89da76131956aba1270c6348bfee8f6c647de841eac7194f", size = 12912365, upload-time = "2025-07-29T22:32:31.517Z" }, - { url = "https://files.pythonhosted.org/packages/4c/9b/0b8aa09817b63e78d94b4977f18b1fcaead3165a5ee49251c5d5c245bb2d/ruff-0.12.7-py3-none-win_arm64.whl", hash = "sha256:dfce05101dbd11833a0776716d5d1578641b7fddb537fe7fa956ab85d1769b69", size = 11982083, upload-time = "2025-07-29T22:32:33.881Z" }, +version = "0.11.2" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/90/61/fb87430f040e4e577e784e325351186976516faef17d6fcd921fe28edfd7/ruff-0.11.2.tar.gz", hash = "sha256:ec47591497d5a1050175bdf4e1a4e6272cddff7da88a2ad595e1e326041d8d94", size = 3857511, upload-time = "2025-03-21T13:31:17.419Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/62/99/102578506f0f5fa29fd7e0df0a273864f79af044757aef73d1cae0afe6ad/ruff-0.11.2-py3-none-linux_armv6l.whl", hash = "sha256:c69e20ea49e973f3afec2c06376eb56045709f0212615c1adb0eda35e8a4e477", size = 10113146, upload-time = "2025-03-21T13:30:26.68Z" }, + { url = "https://files.pythonhosted.org/packages/74/ad/5cd4ba58ab602a579997a8494b96f10f316e874d7c435bcc1a92e6da1b12/ruff-0.11.2-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:2c5424cc1c4eb1d8ecabe6d4f1b70470b4f24a0c0171356290b1953ad8f0e272", size = 10867092, upload-time = "2025-03-21T13:30:37.949Z" }, + { url = "https://files.pythonhosted.org/packages/fc/3e/d3f13619e1d152c7b600a38c1a035e833e794c6625c9a6cea6f63dbf3af4/ruff-0.11.2-py3-none-macosx_11_0_arm64.whl", hash = "sha256:ecf20854cc73f42171eedb66f006a43d0a21bfb98a2523a809931cda569552d9", size = 10224082, upload-time = "2025-03-21T13:30:39.962Z" }, + { url = "https://files.pythonhosted.org/packages/90/06/f77b3d790d24a93f38e3806216f263974909888fd1e826717c3ec956bbcd/ruff-0.11.2-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0c543bf65d5d27240321604cee0633a70c6c25c9a2f2492efa9f6d4b8e4199bb", size = 10394818, upload-time = "2025-03-21T13:30:42.551Z" }, + { url = "https://files.pythonhosted.org/packages/99/7f/78aa431d3ddebfc2418cd95b786642557ba8b3cb578c075239da9ce97ff9/ruff-0.11.2-py3-none-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:20967168cc21195db5830b9224be0e964cc9c8ecf3b5a9e3ce19876e8d3a96e3", size = 9952251, upload-time = "2025-03-21T13:30:45.196Z" }, + { url = "https://files.pythonhosted.org/packages/30/3e/f11186d1ddfaca438c3bbff73c6a2fdb5b60e6450cc466129c694b0ab7a2/ruff-0.11.2-py3-none-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:955a9ce63483999d9f0b8f0b4a3ad669e53484232853054cc8b9d51ab4c5de74", size = 11563566, upload-time = "2025-03-21T13:30:47.516Z" }, + { url = "https://files.pythonhosted.org/packages/22/6c/6ca91befbc0a6539ee133d9a9ce60b1a354db12c3c5d11cfdbf77140f851/ruff-0.11.2-py3-none-manylinux_2_17_ppc64.manylinux2014_ppc64.whl", hash = "sha256:86b3a27c38b8fce73bcd262b0de32e9a6801b76d52cdb3ae4c914515f0cef608", size = 12208721, upload-time = "2025-03-21T13:30:49.56Z" }, + { url = "https://files.pythonhosted.org/packages/19/b0/24516a3b850d55b17c03fc399b681c6a549d06ce665915721dc5d6458a5c/ruff-0.11.2-py3-none-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a3b66a03b248c9fcd9d64d445bafdf1589326bee6fc5c8e92d7562e58883e30f", size = 11662274, upload-time = "2025-03-21T13:30:52.055Z" }, + { url = "https://files.pythonhosted.org/packages/d7/65/76be06d28ecb7c6070280cef2bcb20c98fbf99ff60b1c57d2fb9b8771348/ruff-0.11.2-py3-none-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:0397c2672db015be5aa3d4dac54c69aa012429097ff219392c018e21f5085147", size = 13792284, upload-time = "2025-03-21T13:30:54.24Z" }, + { url = "https://files.pythonhosted.org/packages/ce/d2/4ceed7147e05852876f3b5f3fdc23f878ce2b7e0b90dd6e698bda3d20787/ruff-0.11.2-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:869bcf3f9abf6457fbe39b5a37333aa4eecc52a3b99c98827ccc371a8e5b6f1b", size = 11327861, upload-time = "2025-03-21T13:30:56.757Z" }, + { url = "https://files.pythonhosted.org/packages/c4/78/4935ecba13706fd60ebe0e3dc50371f2bdc3d9bc80e68adc32ff93914534/ruff-0.11.2-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:2a2b50ca35457ba785cd8c93ebbe529467594087b527a08d487cf0ee7b3087e9", size = 10276560, upload-time = "2025-03-21T13:30:58.881Z" }, + { url = "https://files.pythonhosted.org/packages/81/7f/1b2435c3f5245d410bb5dc80f13ec796454c21fbda12b77d7588d5cf4e29/ruff-0.11.2-py3-none-musllinux_1_2_armv7l.whl", hash = "sha256:7c69c74bf53ddcfbc22e6eb2f31211df7f65054bfc1f72288fc71e5f82db3eab", size = 9945091, upload-time = "2025-03-21T13:31:01.45Z" }, + { url = "https://files.pythonhosted.org/packages/39/c4/692284c07e6bf2b31d82bb8c32f8840f9d0627d92983edaac991a2b66c0a/ruff-0.11.2-py3-none-musllinux_1_2_i686.whl", hash = "sha256:6e8fb75e14560f7cf53b15bbc55baf5ecbe373dd5f3aab96ff7aa7777edd7630", size = 10977133, upload-time = "2025-03-21T13:31:04.013Z" }, + { url = "https://files.pythonhosted.org/packages/94/cf/8ab81cb7dd7a3b0a3960c2769825038f3adcd75faf46dd6376086df8b128/ruff-0.11.2-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:842a472d7b4d6f5924e9297aa38149e5dcb1e628773b70e6387ae2c97a63c58f", size = 11378514, upload-time = "2025-03-21T13:31:06.166Z" }, + { url = "https://files.pythonhosted.org/packages/d9/3a/a647fa4f316482dacf2fd68e8a386327a33d6eabd8eb2f9a0c3d291ec549/ruff-0.11.2-py3-none-win32.whl", hash = "sha256:aca01ccd0eb5eb7156b324cfaa088586f06a86d9e5314b0eb330cb48415097cc", size = 10319835, upload-time = "2025-03-21T13:31:10.7Z" }, + { url = "https://files.pythonhosted.org/packages/86/54/3c12d3af58012a5e2cd7ebdbe9983f4834af3f8cbea0e8a8c74fa1e23b2b/ruff-0.11.2-py3-none-win_amd64.whl", hash = "sha256:3170150172a8f994136c0c66f494edf199a0bbea7a409f649e4bc8f4d7084080", size = 11373713, upload-time = "2025-03-21T13:31:13.148Z" }, + { url = "https://files.pythonhosted.org/packages/d6/d4/dd813703af8a1e2ac33bf3feb27e8a5ad514c9f219df80c64d69807e7f71/ruff-0.11.2-py3-none-win_arm64.whl", hash = "sha256:52933095158ff328f4c77af3d74f0379e34fd52f175144cefc1b192e7ccd32b4", size = 10441990, upload-time = "2025-03-21T13:31:15.206Z" }, ] [[package]] name = "s3fs" -version = "2025.7.0" +version = "2025.3.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "aiobotocore" }, { name = "aiohttp" }, { name = "fsspec" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/bf/13/37438c4672ba1d23ec46df0e4b57e98469e5c5f4f98313cf6842b631652b/s3fs-2025.7.0.tar.gz", hash = "sha256:5e7f9ec0cad7745155e3eb86fae15b1481fa29946bf5b3a4ce3a60701ce6022d", size = 77795, upload-time = "2025-07-15T16:35:22.177Z" } +sdist = { url = "https://files.pythonhosted.org/packages/e4/cd/5dde2fed1699ff48120336249d9857a574e39feb8afaff694568ab1499b3/s3fs-2025.3.0.tar.gz", hash = "sha256:446dd539eb0d0678209723cb7ad1bedbb172185b0d34675b09be1ad81843a644", size = 77153, upload-time = "2025-03-07T21:58:32.114Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/ff/c7/30d13b7fd4f866ca3f30e9a6e7ae038f0c45226f6e26b3cc98d6d197f93b/s3fs-2025.7.0-py3-none-any.whl", hash = "sha256:b6b2d3f84b6aa1c2ba5e62e39dd9410cf54f10a2cce1ea6db1ba0d1a6bcce685", size = 30315, upload-time = "2025-07-15T16:35:20.734Z" }, + { url = "https://files.pythonhosted.org/packages/3a/3f/35f4041a82a68df89fe4af97c8bb44aa492dad924799cbb02078e9e303e6/s3fs-2025.3.0-py3-none-any.whl", hash = "sha256:88d803615baa04945156ca0e1498009b7acd3132c07198bd81b3e874846e0aa2", size = 30454, upload-time = "2025-03-07T21:58:30.998Z" }, ] [package.optional-dependencies] @@ -3500,160 +3240,157 @@ boto3 = [ [[package]] name = "s3transfer" -version = "0.13.1" +version = "0.11.3" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "botocore" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/6d/05/d52bf1e65044b4e5e27d4e63e8d1579dbdec54fce685908ae09bc3720030/s3transfer-0.13.1.tar.gz", hash = "sha256:c3fdba22ba1bd367922f27ec8032d6a1cf5f10c934fb5d68cf60fd5a23d936cf", size = 150589, upload-time = "2025-07-18T19:22:42.31Z" } +sdist = { url = "https://files.pythonhosted.org/packages/39/24/1390172471d569e281fcfd29b92f2f73774e95972c965d14b6c802ff2352/s3transfer-0.11.3.tar.gz", hash = "sha256:edae4977e3a122445660c7c114bba949f9d191bae3b34a096f18a1c8c354527a", size = 148042, upload-time = "2025-02-26T20:44:57.459Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/6d/4f/d073e09df851cfa251ef7840007d04db3293a0482ce607d2b993926089be/s3transfer-0.13.1-py3-none-any.whl", hash = "sha256:a981aa7429be23fe6dfc13e80e4020057cbab622b08c0315288758d67cabc724", size = 85308, upload-time = "2025-07-18T19:22:40.947Z" }, + { url = "https://files.pythonhosted.org/packages/e4/81/48c41b554a54d75d4407740abb60e3a102ae416284df04d1dbdcbe3dbf24/s3transfer-0.11.3-py3-none-any.whl", hash = "sha256:ca855bdeb885174b5ffa95b9913622459d4ad8e331fc98eb01e6d5eb6a30655d", size = 84246, upload-time = "2025-02-26T20:44:55.509Z" }, ] [[package]] name = "scipy" -version = "1.15.3" +version = "1.15.2" +source = { registry = "https://pypi.org/simple" } +dependencies = [ + { name = "numpy" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/b7/b9/31ba9cd990e626574baf93fbc1ac61cf9ed54faafd04c479117517661637/scipy-1.15.2.tar.gz", hash = "sha256:cd58a314d92838f7e6f755c8a2167ead4f27e1fd5c1251fd54289569ef3495ec", size = 59417316, upload-time = "2025-02-17T00:42:24.791Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/95/df/ef233fff6838fe6f7840d69b5ef9f20d2b5c912a8727b21ebf876cb15d54/scipy-1.15.2-cp310-cp310-macosx_10_13_x86_64.whl", hash = "sha256:a2ec871edaa863e8213ea5df811cd600734f6400b4af272e1c011e69401218e9", size = 38692502, upload-time = "2025-02-17T00:28:56.118Z" }, + { url = "https://files.pythonhosted.org/packages/5c/20/acdd4efb8a68b842968f7bc5611b1aeb819794508771ad104de418701422/scipy-1.15.2-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:6f223753c6ea76983af380787611ae1291e3ceb23917393079dcc746ba60cfb5", size = 30085508, upload-time = "2025-02-17T00:29:06.048Z" }, + { url = "https://files.pythonhosted.org/packages/42/55/39cf96ca7126f1e78ee72a6344ebdc6702fc47d037319ad93221063e6cf4/scipy-1.15.2-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:ecf797d2d798cf7c838c6d98321061eb3e72a74710e6c40540f0e8087e3b499e", size = 22359166, upload-time = "2025-02-17T00:29:13.553Z" }, + { url = "https://files.pythonhosted.org/packages/51/48/708d26a4ab8a1441536bf2dfcad1df0ca14a69f010fba3ccbdfc02df7185/scipy-1.15.2-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:9b18aa747da280664642997e65aab1dd19d0c3d17068a04b3fe34e2559196cb9", size = 25112047, upload-time = "2025-02-17T00:29:23.204Z" }, + { url = "https://files.pythonhosted.org/packages/dd/65/f9c5755b995ad892020381b8ae11f16d18616208e388621dfacc11df6de6/scipy-1.15.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:87994da02e73549dfecaed9e09a4f9d58a045a053865679aeb8d6d43747d4df3", size = 35536214, upload-time = "2025-02-17T00:29:33.215Z" }, + { url = "https://files.pythonhosted.org/packages/de/3c/c96d904b9892beec978562f64d8cc43f9cca0842e65bd3cd1b7f7389b0ba/scipy-1.15.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:69ea6e56d00977f355c0f84eba69877b6df084516c602d93a33812aa04d90a3d", size = 37646981, upload-time = "2025-02-17T00:29:46.188Z" }, + { url = "https://files.pythonhosted.org/packages/3d/74/c2d8a24d18acdeae69ed02e132b9bc1bb67b7bee90feee1afe05a68f9d67/scipy-1.15.2-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:888307125ea0c4466287191e5606a2c910963405ce9671448ff9c81c53f85f58", size = 37230048, upload-time = "2025-02-17T00:29:56.646Z" }, + { url = "https://files.pythonhosted.org/packages/42/19/0aa4ce80eca82d487987eff0bc754f014dec10d20de2f66754fa4ea70204/scipy-1.15.2-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:9412f5e408b397ff5641080ed1e798623dbe1ec0d78e72c9eca8992976fa65aa", size = 40010322, upload-time = "2025-02-17T00:30:07.422Z" }, + { url = "https://files.pythonhosted.org/packages/d0/d2/f0683b7e992be44d1475cc144d1f1eeae63c73a14f862974b4db64af635e/scipy-1.15.2-cp310-cp310-win_amd64.whl", hash = "sha256:b5e025e903b4f166ea03b109bb241355b9c42c279ea694d8864d033727205e65", size = 41233385, upload-time = "2025-02-17T00:30:20.268Z" }, + { url = "https://files.pythonhosted.org/packages/40/1f/bf0a5f338bda7c35c08b4ed0df797e7bafe8a78a97275e9f439aceb46193/scipy-1.15.2-cp311-cp311-macosx_10_13_x86_64.whl", hash = "sha256:92233b2df6938147be6fa8824b8136f29a18f016ecde986666be5f4d686a91a4", size = 38703651, upload-time = "2025-02-17T00:30:31.09Z" }, + { url = "https://files.pythonhosted.org/packages/de/54/db126aad3874601048c2c20ae3d8a433dbfd7ba8381551e6f62606d9bd8e/scipy-1.15.2-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:62ca1ff3eb513e09ed17a5736929429189adf16d2d740f44e53270cc800ecff1", size = 30102038, upload-time = "2025-02-17T00:30:40.219Z" }, + { url = "https://files.pythonhosted.org/packages/61/d8/84da3fffefb6c7d5a16968fe5b9f24c98606b165bb801bb0b8bc3985200f/scipy-1.15.2-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:4c6676490ad76d1c2894d77f976144b41bd1a4052107902238047fb6a473e971", size = 22375518, upload-time = "2025-02-17T00:30:47.547Z" }, + { url = "https://files.pythonhosted.org/packages/44/78/25535a6e63d3b9c4c90147371aedb5d04c72f3aee3a34451f2dc27c0c07f/scipy-1.15.2-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:a8bf5cb4a25046ac61d38f8d3c3426ec11ebc350246a4642f2f315fe95bda655", size = 25142523, upload-time = "2025-02-17T00:30:56.002Z" }, + { url = "https://files.pythonhosted.org/packages/e0/22/4b4a26fe1cd9ed0bc2b2cb87b17d57e32ab72c346949eaf9288001f8aa8e/scipy-1.15.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:6a8e34cf4c188b6dd004654f88586d78f95639e48a25dfae9c5e34a6dc34547e", size = 35491547, upload-time = "2025-02-17T00:31:07.599Z" }, + { url = "https://files.pythonhosted.org/packages/32/ea/564bacc26b676c06a00266a3f25fdfe91a9d9a2532ccea7ce6dd394541bc/scipy-1.15.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:28a0d2c2075946346e4408b211240764759e0fabaeb08d871639b5f3b1aca8a0", size = 37634077, upload-time = "2025-02-17T00:31:15.191Z" }, + { url = "https://files.pythonhosted.org/packages/43/c2/bfd4e60668897a303b0ffb7191e965a5da4056f0d98acfb6ba529678f0fb/scipy-1.15.2-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:42dabaaa798e987c425ed76062794e93a243be8f0f20fff6e7a89f4d61cb3d40", size = 37231657, upload-time = "2025-02-17T00:31:22.041Z" }, + { url = "https://files.pythonhosted.org/packages/4a/75/5f13050bf4f84c931bcab4f4e83c212a36876c3c2244475db34e4b5fe1a6/scipy-1.15.2-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:6f5e296ec63c5da6ba6fa0343ea73fd51b8b3e1a300b0a8cae3ed4b1122c7462", size = 40035857, upload-time = "2025-02-17T00:31:29.836Z" }, + { url = "https://files.pythonhosted.org/packages/b9/8b/7ec1832b09dbc88f3db411f8cdd47db04505c4b72c99b11c920a8f0479c3/scipy-1.15.2-cp311-cp311-win_amd64.whl", hash = "sha256:597a0c7008b21c035831c39927406c6181bcf8f60a73f36219b69d010aa04737", size = 41217654, upload-time = "2025-02-17T00:31:43.65Z" }, + { url = "https://files.pythonhosted.org/packages/4b/5d/3c78815cbab499610f26b5bae6aed33e227225a9fa5290008a733a64f6fc/scipy-1.15.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:c4697a10da8f8765bb7c83e24a470da5797e37041edfd77fd95ba3811a47c4fd", size = 38756184, upload-time = "2025-02-17T00:31:50.623Z" }, + { url = "https://files.pythonhosted.org/packages/37/20/3d04eb066b471b6e171827548b9ddb3c21c6bbea72a4d84fc5989933910b/scipy-1.15.2-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:869269b767d5ee7ea6991ed7e22b3ca1f22de73ab9a49c44bad338b725603301", size = 30163558, upload-time = "2025-02-17T00:31:56.721Z" }, + { url = "https://files.pythonhosted.org/packages/a4/98/e5c964526c929ef1f795d4c343b2ff98634ad2051bd2bbadfef9e772e413/scipy-1.15.2-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:bad78d580270a4d32470563ea86c6590b465cb98f83d760ff5b0990cb5518a93", size = 22437211, upload-time = "2025-02-17T00:32:03.042Z" }, + { url = "https://files.pythonhosted.org/packages/1d/cd/1dc7371e29195ecbf5222f9afeedb210e0a75057d8afbd942aa6cf8c8eca/scipy-1.15.2-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:b09ae80010f52efddb15551025f9016c910296cf70adbf03ce2a8704f3a5ad20", size = 25232260, upload-time = "2025-02-17T00:32:07.847Z" }, + { url = "https://files.pythonhosted.org/packages/f0/24/1a181a9e5050090e0b5138c5f496fee33293c342b788d02586bc410c6477/scipy-1.15.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5a6fd6eac1ce74a9f77a7fc724080d507c5812d61e72bd5e4c489b042455865e", size = 35198095, upload-time = "2025-02-17T00:32:14.565Z" }, + { url = "https://files.pythonhosted.org/packages/c0/53/eaada1a414c026673eb983f8b4a55fe5eb172725d33d62c1b21f63ff6ca4/scipy-1.15.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2b871df1fe1a3ba85d90e22742b93584f8d2b8e6124f8372ab15c71b73e428b8", size = 37297371, upload-time = "2025-02-17T00:32:21.411Z" }, + { url = "https://files.pythonhosted.org/packages/e9/06/0449b744892ed22b7e7b9a1994a866e64895363572677a316a9042af1fe5/scipy-1.15.2-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:03205d57a28e18dfd39f0377d5002725bf1f19a46f444108c29bdb246b6c8a11", size = 36872390, upload-time = "2025-02-17T00:32:29.421Z" }, + { url = "https://files.pythonhosted.org/packages/6a/6f/a8ac3cfd9505ec695c1bc35edc034d13afbd2fc1882a7c6b473e280397bb/scipy-1.15.2-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:601881dfb761311045b03114c5fe718a12634e5608c3b403737ae463c9885d53", size = 39700276, upload-time = "2025-02-17T00:32:37.431Z" }, + { url = "https://files.pythonhosted.org/packages/f5/6f/e6e5aff77ea2a48dd96808bb51d7450875af154ee7cbe72188afb0b37929/scipy-1.15.2-cp312-cp312-win_amd64.whl", hash = "sha256:e7c68b6a43259ba0aab737237876e5c2c549a031ddb7abc28c7b47f22e202ded", size = 40942317, upload-time = "2025-02-17T00:32:45.47Z" }, + { url = "https://files.pythonhosted.org/packages/53/40/09319f6e0f276ea2754196185f95cd191cb852288440ce035d5c3a931ea2/scipy-1.15.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:01edfac9f0798ad6b46d9c4c9ca0e0ad23dbf0b1eb70e96adb9fa7f525eff0bf", size = 38717587, upload-time = "2025-02-17T00:32:53.196Z" }, + { url = "https://files.pythonhosted.org/packages/fe/c3/2854f40ecd19585d65afaef601e5e1f8dbf6758b2f95b5ea93d38655a2c6/scipy-1.15.2-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:08b57a9336b8e79b305a143c3655cc5bdbe6d5ece3378578888d2afbb51c4e37", size = 30100266, upload-time = "2025-02-17T00:32:59.318Z" }, + { url = "https://files.pythonhosted.org/packages/dd/b1/f9fe6e3c828cb5930b5fe74cb479de5f3d66d682fa8adb77249acaf545b8/scipy-1.15.2-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:54c462098484e7466362a9f1672d20888f724911a74c22ae35b61f9c5919183d", size = 22373768, upload-time = "2025-02-17T00:33:04.091Z" }, + { url = "https://files.pythonhosted.org/packages/15/9d/a60db8c795700414c3f681908a2b911e031e024d93214f2d23c6dae174ab/scipy-1.15.2-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:cf72ff559a53a6a6d77bd8eefd12a17995ffa44ad86c77a5df96f533d4e6c6bb", size = 25154719, upload-time = "2025-02-17T00:33:08.909Z" }, + { url = "https://files.pythonhosted.org/packages/37/3b/9bda92a85cd93f19f9ed90ade84aa1e51657e29988317fabdd44544f1dd4/scipy-1.15.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9de9d1416b3d9e7df9923ab23cd2fe714244af10b763975bea9e4f2e81cebd27", size = 35163195, upload-time = "2025-02-17T00:33:15.352Z" }, + { url = "https://files.pythonhosted.org/packages/03/5a/fc34bf1aa14dc7c0e701691fa8685f3faec80e57d816615e3625f28feb43/scipy-1.15.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fb530e4794fc8ea76a4a21ccb67dea33e5e0e60f07fc38a49e821e1eae3b71a0", size = 37255404, upload-time = "2025-02-17T00:33:22.21Z" }, + { url = "https://files.pythonhosted.org/packages/4a/71/472eac45440cee134c8a180dbe4c01b3ec247e0338b7c759e6cd71f199a7/scipy-1.15.2-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:5ea7ed46d437fc52350b028b1d44e002646e28f3e8ddc714011aaf87330f2f32", size = 36860011, upload-time = "2025-02-17T00:33:29.446Z" }, + { url = "https://files.pythonhosted.org/packages/01/b3/21f890f4f42daf20e4d3aaa18182dddb9192771cd47445aaae2e318f6738/scipy-1.15.2-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:11e7ad32cf184b74380f43d3c0a706f49358b904fa7d5345f16ddf993609184d", size = 39657406, upload-time = "2025-02-17T00:33:39.019Z" }, + { url = "https://files.pythonhosted.org/packages/0d/76/77cf2ac1f2a9cc00c073d49e1e16244e389dd88e2490c91d84e1e3e4d126/scipy-1.15.2-cp313-cp313-win_amd64.whl", hash = "sha256:a5080a79dfb9b78b768cebf3c9dcbc7b665c5875793569f48bf0e2b1d7f68f6f", size = 40961243, upload-time = "2025-02-17T00:34:51.024Z" }, + { url = "https://files.pythonhosted.org/packages/4c/4b/a57f8ddcf48e129e6054fa9899a2a86d1fc6b07a0e15c7eebff7ca94533f/scipy-1.15.2-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:447ce30cee6a9d5d1379087c9e474628dab3db4a67484be1b7dc3196bfb2fac9", size = 38870286, upload-time = "2025-02-17T00:33:47.62Z" }, + { url = "https://files.pythonhosted.org/packages/0c/43/c304d69a56c91ad5f188c0714f6a97b9c1fed93128c691148621274a3a68/scipy-1.15.2-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:c90ebe8aaa4397eaefa8455a8182b164a6cc1d59ad53f79943f266d99f68687f", size = 30141634, upload-time = "2025-02-17T00:33:54.131Z" }, + { url = "https://files.pythonhosted.org/packages/44/1a/6c21b45d2548eb73be9b9bff421aaaa7e85e22c1f9b3bc44b23485dfce0a/scipy-1.15.2-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:def751dd08243934c884a3221156d63e15234a3155cf25978b0a668409d45eb6", size = 22415179, upload-time = "2025-02-17T00:33:59.948Z" }, + { url = "https://files.pythonhosted.org/packages/74/4b/aefac4bba80ef815b64f55da06f62f92be5d03b467f2ce3668071799429a/scipy-1.15.2-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:302093e7dfb120e55515936cb55618ee0b895f8bcaf18ff81eca086c17bd80af", size = 25126412, upload-time = "2025-02-17T00:34:06.328Z" }, + { url = "https://files.pythonhosted.org/packages/b1/53/1cbb148e6e8f1660aacd9f0a9dfa2b05e9ff1cb54b4386fe868477972ac2/scipy-1.15.2-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7cd5b77413e1855351cdde594eca99c1f4a588c2d63711388b6a1f1c01f62274", size = 34952867, upload-time = "2025-02-17T00:34:12.928Z" }, + { url = "https://files.pythonhosted.org/packages/2c/23/e0eb7f31a9c13cf2dca083828b97992dd22f8184c6ce4fec5deec0c81fcf/scipy-1.15.2-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6d0194c37037707b2afa7a2f2a924cf7bac3dc292d51b6a925e5fcb89bc5c776", size = 36890009, upload-time = "2025-02-17T00:34:19.55Z" }, + { url = "https://files.pythonhosted.org/packages/03/f3/e699e19cabe96bbac5189c04aaa970718f0105cff03d458dc5e2b6bd1e8c/scipy-1.15.2-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:bae43364d600fdc3ac327db99659dcb79e6e7ecd279a75fe1266669d9a652828", size = 36545159, upload-time = "2025-02-17T00:34:26.724Z" }, + { url = "https://files.pythonhosted.org/packages/af/f5/ab3838e56fe5cc22383d6fcf2336e48c8fe33e944b9037fbf6cbdf5a11f8/scipy-1.15.2-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:f031846580d9acccd0044efd1a90e6f4df3a6e12b4b6bd694a7bc03a89892b28", size = 39136566, upload-time = "2025-02-17T00:34:34.512Z" }, + { url = "https://files.pythonhosted.org/packages/0a/c8/b3f566db71461cabd4b2d5b39bcc24a7e1c119535c8361f81426be39bb47/scipy-1.15.2-cp313-cp313t-win_amd64.whl", hash = "sha256:fe8a9eb875d430d81755472c5ba75e84acc980e4a8f6204d402849234d3017db", size = 40477705, upload-time = "2025-02-17T00:34:43.619Z" }, +] + +[[package]] +name = "seaborn" +version = "0.13.2" source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version < '3.11' and sys_platform == 'linux'", - "python_full_version < '3.11' and sys_platform != 'linux'", -] dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/0f/37/6964b830433e654ec7485e45a00fc9a27cf868d622838f6b6d9c5ec0d532/scipy-1.15.3.tar.gz", hash = "sha256:eae3cf522bc7df64b42cad3925c876e1b0b6c35c1337c93e12c0f366f55b0eaf", size = 59419214, upload-time = "2025-05-08T16:13:05.955Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/78/2f/4966032c5f8cc7e6a60f1b2e0ad686293b9474b65246b0c642e3ef3badd0/scipy-1.15.3-cp310-cp310-macosx_10_13_x86_64.whl", hash = "sha256:a345928c86d535060c9c2b25e71e87c39ab2f22fc96e9636bd74d1dbf9de448c", size = 38702770, upload-time = "2025-05-08T16:04:20.849Z" }, - { url = "https://files.pythonhosted.org/packages/a0/6e/0c3bf90fae0e910c274db43304ebe25a6b391327f3f10b5dcc638c090795/scipy-1.15.3-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:ad3432cb0f9ed87477a8d97f03b763fd1d57709f1bbde3c9369b1dff5503b253", size = 30094511, upload-time = "2025-05-08T16:04:27.103Z" }, - { url = "https://files.pythonhosted.org/packages/ea/b1/4deb37252311c1acff7f101f6453f0440794f51b6eacb1aad4459a134081/scipy-1.15.3-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:aef683a9ae6eb00728a542b796f52a5477b78252edede72b8327a886ab63293f", size = 22368151, upload-time = "2025-05-08T16:04:31.731Z" }, - { url = "https://files.pythonhosted.org/packages/38/7d/f457626e3cd3c29b3a49ca115a304cebb8cc6f31b04678f03b216899d3c6/scipy-1.15.3-cp310-cp310-macosx_14_0_x86_64.whl", hash = "sha256:1c832e1bd78dea67d5c16f786681b28dd695a8cb1fb90af2e27580d3d0967e92", size = 25121732, upload-time = "2025-05-08T16:04:36.596Z" }, - { url = "https://files.pythonhosted.org/packages/db/0a/92b1de4a7adc7a15dcf5bddc6e191f6f29ee663b30511ce20467ef9b82e4/scipy-1.15.3-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:263961f658ce2165bbd7b99fa5135195c3a12d9bef045345016b8b50c315cb82", size = 35547617, upload-time = "2025-05-08T16:04:43.546Z" }, - { url = "https://files.pythonhosted.org/packages/8e/6d/41991e503e51fc1134502694c5fa7a1671501a17ffa12716a4a9151af3df/scipy-1.15.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9e2abc762b0811e09a0d3258abee2d98e0c703eee49464ce0069590846f31d40", size = 37662964, upload-time = "2025-05-08T16:04:49.431Z" }, - { url = "https://files.pythonhosted.org/packages/25/e1/3df8f83cb15f3500478c889be8fb18700813b95e9e087328230b98d547ff/scipy-1.15.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:ed7284b21a7a0c8f1b6e5977ac05396c0d008b89e05498c8b7e8f4a1423bba0e", size = 37238749, upload-time = "2025-05-08T16:04:55.215Z" }, - { url = "https://files.pythonhosted.org/packages/93/3e/b3257cf446f2a3533ed7809757039016b74cd6f38271de91682aa844cfc5/scipy-1.15.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:5380741e53df2c566f4d234b100a484b420af85deb39ea35a1cc1be84ff53a5c", size = 40022383, upload-time = "2025-05-08T16:05:01.914Z" }, - { url = "https://files.pythonhosted.org/packages/d1/84/55bc4881973d3f79b479a5a2e2df61c8c9a04fcb986a213ac9c02cfb659b/scipy-1.15.3-cp310-cp310-win_amd64.whl", hash = "sha256:9d61e97b186a57350f6d6fd72640f9e99d5a4a2b8fbf4b9ee9a841eab327dc13", size = 41259201, upload-time = "2025-05-08T16:05:08.166Z" }, - { url = "https://files.pythonhosted.org/packages/96/ab/5cc9f80f28f6a7dff646c5756e559823614a42b1939d86dd0ed550470210/scipy-1.15.3-cp311-cp311-macosx_10_13_x86_64.whl", hash = "sha256:993439ce220d25e3696d1b23b233dd010169b62f6456488567e830654ee37a6b", size = 38714255, upload-time = "2025-05-08T16:05:14.596Z" }, - { url = "https://files.pythonhosted.org/packages/4a/4a/66ba30abe5ad1a3ad15bfb0b59d22174012e8056ff448cb1644deccbfed2/scipy-1.15.3-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:34716e281f181a02341ddeaad584205bd2fd3c242063bd3423d61ac259ca7eba", size = 30111035, upload-time = "2025-05-08T16:05:20.152Z" }, - { url = "https://files.pythonhosted.org/packages/4b/fa/a7e5b95afd80d24313307f03624acc65801846fa75599034f8ceb9e2cbf6/scipy-1.15.3-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:3b0334816afb8b91dab859281b1b9786934392aa3d527cd847e41bb6f45bee65", size = 22384499, upload-time = "2025-05-08T16:05:24.494Z" }, - { url = "https://files.pythonhosted.org/packages/17/99/f3aaddccf3588bb4aea70ba35328c204cadd89517a1612ecfda5b2dd9d7a/scipy-1.15.3-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:6db907c7368e3092e24919b5e31c76998b0ce1684d51a90943cb0ed1b4ffd6c1", size = 25152602, upload-time = "2025-05-08T16:05:29.313Z" }, - { url = "https://files.pythonhosted.org/packages/56/c5/1032cdb565f146109212153339f9cb8b993701e9fe56b1c97699eee12586/scipy-1.15.3-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:721d6b4ef5dc82ca8968c25b111e307083d7ca9091bc38163fb89243e85e3889", size = 35503415, upload-time = "2025-05-08T16:05:34.699Z" }, - { url = "https://files.pythonhosted.org/packages/bd/37/89f19c8c05505d0601ed5650156e50eb881ae3918786c8fd7262b4ee66d3/scipy-1.15.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:39cb9c62e471b1bb3750066ecc3a3f3052b37751c7c3dfd0fd7e48900ed52982", size = 37652622, upload-time = "2025-05-08T16:05:40.762Z" }, - { url = "https://files.pythonhosted.org/packages/7e/31/be59513aa9695519b18e1851bb9e487de66f2d31f835201f1b42f5d4d475/scipy-1.15.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:795c46999bae845966368a3c013e0e00947932d68e235702b5c3f6ea799aa8c9", size = 37244796, upload-time = "2025-05-08T16:05:48.119Z" }, - { url = "https://files.pythonhosted.org/packages/10/c0/4f5f3eeccc235632aab79b27a74a9130c6c35df358129f7ac8b29f562ac7/scipy-1.15.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:18aaacb735ab38b38db42cb01f6b92a2d0d4b6aabefeb07f02849e47f8fb3594", size = 40047684, upload-time = "2025-05-08T16:05:54.22Z" }, - { url = "https://files.pythonhosted.org/packages/ab/a7/0ddaf514ce8a8714f6ed243a2b391b41dbb65251affe21ee3077ec45ea9a/scipy-1.15.3-cp311-cp311-win_amd64.whl", hash = "sha256:ae48a786a28412d744c62fd7816a4118ef97e5be0bee968ce8f0a2fba7acf3bb", size = 41246504, upload-time = "2025-05-08T16:06:00.437Z" }, - { url = "https://files.pythonhosted.org/packages/37/4b/683aa044c4162e10ed7a7ea30527f2cbd92e6999c10a8ed8edb253836e9c/scipy-1.15.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:6ac6310fdbfb7aa6612408bd2f07295bcbd3fda00d2d702178434751fe48e019", size = 38766735, upload-time = "2025-05-08T16:06:06.471Z" }, - { url = "https://files.pythonhosted.org/packages/7b/7e/f30be3d03de07f25dc0ec926d1681fed5c732d759ac8f51079708c79e680/scipy-1.15.3-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:185cd3d6d05ca4b44a8f1595af87f9c372bb6acf9c808e99aa3e9aa03bd98cf6", size = 30173284, upload-time = "2025-05-08T16:06:11.686Z" }, - { url = "https://files.pythonhosted.org/packages/07/9c/0ddb0d0abdabe0d181c1793db51f02cd59e4901da6f9f7848e1f96759f0d/scipy-1.15.3-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:05dc6abcd105e1a29f95eada46d4a3f251743cfd7d3ae8ddb4088047f24ea477", size = 22446958, upload-time = "2025-05-08T16:06:15.97Z" }, - { url = "https://files.pythonhosted.org/packages/af/43/0bce905a965f36c58ff80d8bea33f1f9351b05fad4beaad4eae34699b7a1/scipy-1.15.3-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:06efcba926324df1696931a57a176c80848ccd67ce6ad020c810736bfd58eb1c", size = 25242454, upload-time = "2025-05-08T16:06:20.394Z" }, - { url = "https://files.pythonhosted.org/packages/56/30/a6f08f84ee5b7b28b4c597aca4cbe545535c39fe911845a96414700b64ba/scipy-1.15.3-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c05045d8b9bfd807ee1b9f38761993297b10b245f012b11b13b91ba8945f7e45", size = 35210199, upload-time = "2025-05-08T16:06:26.159Z" }, - { url = "https://files.pythonhosted.org/packages/0b/1f/03f52c282437a168ee2c7c14a1a0d0781a9a4a8962d84ac05c06b4c5b555/scipy-1.15.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:271e3713e645149ea5ea3e97b57fdab61ce61333f97cfae392c28ba786f9bb49", size = 37309455, upload-time = "2025-05-08T16:06:32.778Z" }, - { url = "https://files.pythonhosted.org/packages/89/b1/fbb53137f42c4bf630b1ffdfc2151a62d1d1b903b249f030d2b1c0280af8/scipy-1.15.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:6cfd56fc1a8e53f6e89ba3a7a7251f7396412d655bca2aa5611c8ec9a6784a1e", size = 36885140, upload-time = "2025-05-08T16:06:39.249Z" }, - { url = "https://files.pythonhosted.org/packages/2e/2e/025e39e339f5090df1ff266d021892694dbb7e63568edcfe43f892fa381d/scipy-1.15.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:0ff17c0bb1cb32952c09217d8d1eed9b53d1463e5f1dd6052c7857f83127d539", size = 39710549, upload-time = "2025-05-08T16:06:45.729Z" }, - { url = "https://files.pythonhosted.org/packages/e6/eb/3bf6ea8ab7f1503dca3a10df2e4b9c3f6b3316df07f6c0ded94b281c7101/scipy-1.15.3-cp312-cp312-win_amd64.whl", hash = "sha256:52092bc0472cfd17df49ff17e70624345efece4e1a12b23783a1ac59a1b728ed", size = 40966184, upload-time = "2025-05-08T16:06:52.623Z" }, - { url = "https://files.pythonhosted.org/packages/73/18/ec27848c9baae6e0d6573eda6e01a602e5649ee72c27c3a8aad673ebecfd/scipy-1.15.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:2c620736bcc334782e24d173c0fdbb7590a0a436d2fdf39310a8902505008759", size = 38728256, upload-time = "2025-05-08T16:06:58.696Z" }, - { url = "https://files.pythonhosted.org/packages/74/cd/1aef2184948728b4b6e21267d53b3339762c285a46a274ebb7863c9e4742/scipy-1.15.3-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:7e11270a000969409d37ed399585ee530b9ef6aa99d50c019de4cb01e8e54e62", size = 30109540, upload-time = "2025-05-08T16:07:04.209Z" }, - { url = "https://files.pythonhosted.org/packages/5b/d8/59e452c0a255ec352bd0a833537a3bc1bfb679944c4938ab375b0a6b3a3e/scipy-1.15.3-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:8c9ed3ba2c8a2ce098163a9bdb26f891746d02136995df25227a20e71c396ebb", size = 22383115, upload-time = "2025-05-08T16:07:08.998Z" }, - { url = "https://files.pythonhosted.org/packages/08/f5/456f56bbbfccf696263b47095291040655e3cbaf05d063bdc7c7517f32ac/scipy-1.15.3-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:0bdd905264c0c9cfa74a4772cdb2070171790381a5c4d312c973382fc6eaf730", size = 25163884, upload-time = "2025-05-08T16:07:14.091Z" }, - { url = "https://files.pythonhosted.org/packages/a2/66/a9618b6a435a0f0c0b8a6d0a2efb32d4ec5a85f023c2b79d39512040355b/scipy-1.15.3-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:79167bba085c31f38603e11a267d862957cbb3ce018d8b38f79ac043bc92d825", size = 35174018, upload-time = "2025-05-08T16:07:19.427Z" }, - { url = "https://files.pythonhosted.org/packages/b5/09/c5b6734a50ad4882432b6bb7c02baf757f5b2f256041da5df242e2d7e6b6/scipy-1.15.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c9deabd6d547aee2c9a81dee6cc96c6d7e9a9b1953f74850c179f91fdc729cb7", size = 37269716, upload-time = "2025-05-08T16:07:25.712Z" }, - { url = "https://files.pythonhosted.org/packages/77/0a/eac00ff741f23bcabd352731ed9b8995a0a60ef57f5fd788d611d43d69a1/scipy-1.15.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:dde4fc32993071ac0c7dd2d82569e544f0bdaff66269cb475e0f369adad13f11", size = 36872342, upload-time = "2025-05-08T16:07:31.468Z" }, - { url = "https://files.pythonhosted.org/packages/fe/54/4379be86dd74b6ad81551689107360d9a3e18f24d20767a2d5b9253a3f0a/scipy-1.15.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:f77f853d584e72e874d87357ad70f44b437331507d1c311457bed8ed2b956126", size = 39670869, upload-time = "2025-05-08T16:07:38.002Z" }, - { url = "https://files.pythonhosted.org/packages/87/2e/892ad2862ba54f084ffe8cc4a22667eaf9c2bcec6d2bff1d15713c6c0703/scipy-1.15.3-cp313-cp313-win_amd64.whl", hash = "sha256:b90ab29d0c37ec9bf55424c064312930ca5f4bde15ee8619ee44e69319aab163", size = 40988851, upload-time = "2025-05-08T16:08:33.671Z" }, - { url = "https://files.pythonhosted.org/packages/1b/e9/7a879c137f7e55b30d75d90ce3eb468197646bc7b443ac036ae3fe109055/scipy-1.15.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:3ac07623267feb3ae308487c260ac684b32ea35fd81e12845039952f558047b8", size = 38863011, upload-time = "2025-05-08T16:07:44.039Z" }, - { url = "https://files.pythonhosted.org/packages/51/d1/226a806bbd69f62ce5ef5f3ffadc35286e9fbc802f606a07eb83bf2359de/scipy-1.15.3-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:6487aa99c2a3d509a5227d9a5e889ff05830a06b2ce08ec30df6d79db5fcd5c5", size = 30266407, upload-time = "2025-05-08T16:07:49.891Z" }, - { url = "https://files.pythonhosted.org/packages/e5/9b/f32d1d6093ab9eeabbd839b0f7619c62e46cc4b7b6dbf05b6e615bbd4400/scipy-1.15.3-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:50f9e62461c95d933d5c5ef4a1f2ebf9a2b4e83b0db374cb3f1de104d935922e", size = 22540030, upload-time = "2025-05-08T16:07:54.121Z" }, - { url = "https://files.pythonhosted.org/packages/e7/29/c278f699b095c1a884f29fda126340fcc201461ee8bfea5c8bdb1c7c958b/scipy-1.15.3-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:14ed70039d182f411ffc74789a16df3835e05dc469b898233a245cdfd7f162cb", size = 25218709, upload-time = "2025-05-08T16:07:58.506Z" }, - { url = "https://files.pythonhosted.org/packages/24/18/9e5374b617aba742a990581373cd6b68a2945d65cc588482749ef2e64467/scipy-1.15.3-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0a769105537aa07a69468a0eefcd121be52006db61cdd8cac8a0e68980bbb723", size = 34809045, upload-time = "2025-05-08T16:08:03.929Z" }, - { url = "https://files.pythonhosted.org/packages/e1/fe/9c4361e7ba2927074360856db6135ef4904d505e9b3afbbcb073c4008328/scipy-1.15.3-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9db984639887e3dffb3928d118145ffe40eff2fa40cb241a306ec57c219ebbbb", size = 36703062, upload-time = "2025-05-08T16:08:09.558Z" }, - { url = "https://files.pythonhosted.org/packages/b7/8e/038ccfe29d272b30086b25a4960f757f97122cb2ec42e62b460d02fe98e9/scipy-1.15.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:40e54d5c7e7ebf1aa596c374c49fa3135f04648a0caabcb66c52884b943f02b4", size = 36393132, upload-time = "2025-05-08T16:08:15.34Z" }, - { url = "https://files.pythonhosted.org/packages/10/7e/5c12285452970be5bdbe8352c619250b97ebf7917d7a9a9e96b8a8140f17/scipy-1.15.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:5e721fed53187e71d0ccf382b6bf977644c533e506c4d33c3fb24de89f5c3ed5", size = 38979503, upload-time = "2025-05-08T16:08:21.513Z" }, - { url = "https://files.pythonhosted.org/packages/81/06/0a5e5349474e1cbc5757975b21bd4fad0e72ebf138c5592f191646154e06/scipy-1.15.3-cp313-cp313t-win_amd64.whl", hash = "sha256:76ad1fb5f8752eabf0fa02e4cc0336b4e8f021e2d5f061ed37d6d264db35e3ca", size = 40308097, upload-time = "2025-05-08T16:08:27.627Z" }, -] - -[[package]] -name = "scipy" -version = "1.16.1" -source = { registry = "https://pypi.org/simple" } -resolution-markers = [ - "python_full_version >= '3.13' and sys_platform == 'linux'", - "python_full_version == '3.12.*' and sys_platform == 'linux'", - "python_full_version >= '3.13' and sys_platform != 'linux'", - "python_full_version == '3.12.*' and sys_platform != 'linux'", - "python_full_version == '3.11.*' and sys_platform == 'linux'", - "python_full_version == '3.11.*' and sys_platform != 'linux'", + { name = "matplotlib" }, + { name = "numpy" }, + { name = "pandas" }, ] -dependencies = [ - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/f5/4a/b927028464795439faec8eaf0b03b011005c487bb2d07409f28bf30879c4/scipy-1.16.1.tar.gz", hash = "sha256:44c76f9e8b6e8e488a586190ab38016e4ed2f8a038af7cd3defa903c0a2238b3", size = 30580861, upload-time = "2025-07-27T16:33:30.834Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/da/91/812adc6f74409b461e3a5fa97f4f74c769016919203138a3bf6fc24ba4c5/scipy-1.16.1-cp311-cp311-macosx_10_14_x86_64.whl", hash = "sha256:c033fa32bab91dc98ca59d0cf23bb876454e2bb02cbe592d5023138778f70030", size = 36552519, upload-time = "2025-07-27T16:26:29.658Z" }, - { url = "https://files.pythonhosted.org/packages/47/18/8e355edcf3b71418d9e9f9acd2708cc3a6c27e8f98fde0ac34b8a0b45407/scipy-1.16.1-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:6e5c2f74e5df33479b5cd4e97a9104c511518fbd979aa9b8f6aec18b2e9ecae7", size = 28638010, upload-time = "2025-07-27T16:26:38.196Z" }, - { url = "https://files.pythonhosted.org/packages/d9/eb/e931853058607bdfbc11b86df19ae7a08686121c203483f62f1ecae5989c/scipy-1.16.1-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:0a55ffe0ba0f59666e90951971a884d1ff6f4ec3275a48f472cfb64175570f77", size = 20909790, upload-time = "2025-07-27T16:26:43.93Z" }, - { url = "https://files.pythonhosted.org/packages/45/0c/be83a271d6e96750cd0be2e000f35ff18880a46f05ce8b5d3465dc0f7a2a/scipy-1.16.1-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:f8a5d6cd147acecc2603fbd382fed6c46f474cccfcf69ea32582e033fb54dcfe", size = 23513352, upload-time = "2025-07-27T16:26:50.017Z" }, - { url = "https://files.pythonhosted.org/packages/7c/bf/fe6eb47e74f762f933cca962db7f2c7183acfdc4483bd1c3813cfe83e538/scipy-1.16.1-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:cb18899127278058bcc09e7b9966d41a5a43740b5bb8dcba401bd983f82e885b", size = 33534643, upload-time = "2025-07-27T16:26:57.503Z" }, - { url = "https://files.pythonhosted.org/packages/bb/ba/63f402e74875486b87ec6506a4f93f6d8a0d94d10467280f3d9d7837ce3a/scipy-1.16.1-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:adccd93a2fa937a27aae826d33e3bfa5edf9aa672376a4852d23a7cd67a2e5b7", size = 35376776, upload-time = "2025-07-27T16:27:06.639Z" }, - { url = "https://files.pythonhosted.org/packages/c3/b4/04eb9d39ec26a1b939689102da23d505ea16cdae3dbb18ffc53d1f831044/scipy-1.16.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:18aca1646a29ee9a0625a1be5637fa798d4d81fdf426481f06d69af828f16958", size = 35698906, upload-time = "2025-07-27T16:27:14.943Z" }, - { url = "https://files.pythonhosted.org/packages/04/d6/bb5468da53321baeb001f6e4e0d9049eadd175a4a497709939128556e3ec/scipy-1.16.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d85495cef541729a70cdddbbf3e6b903421bc1af3e8e3a9a72a06751f33b7c39", size = 38129275, upload-time = "2025-07-27T16:27:23.873Z" }, - { url = "https://files.pythonhosted.org/packages/c4/94/994369978509f227cba7dfb9e623254d0d5559506fe994aef4bea3ed469c/scipy-1.16.1-cp311-cp311-win_amd64.whl", hash = "sha256:226652fca853008119c03a8ce71ffe1b3f6d2844cc1686e8f9806edafae68596", size = 38644572, upload-time = "2025-07-27T16:27:32.637Z" }, - { url = "https://files.pythonhosted.org/packages/f8/d9/ec4864f5896232133f51382b54a08de91a9d1af7a76dfa372894026dfee2/scipy-1.16.1-cp312-cp312-macosx_10_14_x86_64.whl", hash = "sha256:81b433bbeaf35728dad619afc002db9b189e45eebe2cd676effe1fb93fef2b9c", size = 36575194, upload-time = "2025-07-27T16:27:41.321Z" }, - { url = "https://files.pythonhosted.org/packages/5c/6d/40e81ecfb688e9d25d34a847dca361982a6addf8e31f0957b1a54fbfa994/scipy-1.16.1-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:886cc81fdb4c6903a3bb0464047c25a6d1016fef77bb97949817d0c0d79f9e04", size = 28594590, upload-time = "2025-07-27T16:27:49.204Z" }, - { url = "https://files.pythonhosted.org/packages/0e/37/9f65178edfcc629377ce9a64fc09baebea18c80a9e57ae09a52edf84880b/scipy-1.16.1-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:15240c3aac087a522b4eaedb09f0ad061753c5eebf1ea430859e5bf8640d5919", size = 20866458, upload-time = "2025-07-27T16:27:54.98Z" }, - { url = "https://files.pythonhosted.org/packages/2c/7b/749a66766871ea4cb1d1ea10f27004db63023074c22abed51f22f09770e0/scipy-1.16.1-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:65f81a25805f3659b48126b5053d9e823d3215e4a63730b5e1671852a1705921", size = 23539318, upload-time = "2025-07-27T16:28:01.604Z" }, - { url = "https://files.pythonhosted.org/packages/c4/db/8d4afec60eb833a666434d4541a3151eedbf2494ea6d4d468cbe877f00cd/scipy-1.16.1-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:6c62eea7f607f122069b9bad3f99489ddca1a5173bef8a0c75555d7488b6f725", size = 33292899, upload-time = "2025-07-27T16:28:09.147Z" }, - { url = "https://files.pythonhosted.org/packages/51/1e/79023ca3bbb13a015d7d2757ecca3b81293c663694c35d6541b4dca53e98/scipy-1.16.1-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f965bbf3235b01c776115ab18f092a95aa74c271a52577bcb0563e85738fd618", size = 35162637, upload-time = "2025-07-27T16:28:17.535Z" }, - { url = "https://files.pythonhosted.org/packages/b6/49/0648665f9c29fdaca4c679182eb972935b3b4f5ace41d323c32352f29816/scipy-1.16.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:f006e323874ffd0b0b816d8c6a8e7f9a73d55ab3b8c3f72b752b226d0e3ac83d", size = 35490507, upload-time = "2025-07-27T16:28:25.705Z" }, - { url = "https://files.pythonhosted.org/packages/62/8f/66cbb9d6bbb18d8c658f774904f42a92078707a7c71e5347e8bf2f52bb89/scipy-1.16.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e8fd15fc5085ab4cca74cb91fe0a4263b1f32e4420761ddae531ad60934c2119", size = 37923998, upload-time = "2025-07-27T16:28:34.339Z" }, - { url = "https://files.pythonhosted.org/packages/14/c3/61f273ae550fbf1667675701112e380881905e28448c080b23b5a181df7c/scipy-1.16.1-cp312-cp312-win_amd64.whl", hash = "sha256:f7b8013c6c066609577d910d1a2a077021727af07b6fab0ee22c2f901f22352a", size = 38508060, upload-time = "2025-07-27T16:28:43.242Z" }, - { url = "https://files.pythonhosted.org/packages/93/0b/b5c99382b839854a71ca9482c684e3472badc62620287cbbdab499b75ce6/scipy-1.16.1-cp313-cp313-macosx_10_14_x86_64.whl", hash = "sha256:5451606823a5e73dfa621a89948096c6528e2896e40b39248295d3a0138d594f", size = 36533717, upload-time = "2025-07-27T16:28:51.706Z" }, - { url = "https://files.pythonhosted.org/packages/eb/e5/69ab2771062c91e23e07c12e7d5033a6b9b80b0903ee709c3c36b3eb520c/scipy-1.16.1-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:89728678c5ca5abd610aee148c199ac1afb16e19844401ca97d43dc548a354eb", size = 28570009, upload-time = "2025-07-27T16:28:57.017Z" }, - { url = "https://files.pythonhosted.org/packages/f4/69/bd75dbfdd3cf524f4d753484d723594aed62cfaac510123e91a6686d520b/scipy-1.16.1-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:e756d688cb03fd07de0fffad475649b03cb89bee696c98ce508b17c11a03f95c", size = 20841942, upload-time = "2025-07-27T16:29:01.152Z" }, - { url = "https://files.pythonhosted.org/packages/ea/74/add181c87663f178ba7d6144b370243a87af8476664d5435e57d599e6874/scipy-1.16.1-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:5aa2687b9935da3ed89c5dbed5234576589dd28d0bf7cd237501ccfbdf1ad608", size = 23498507, upload-time = "2025-07-27T16:29:05.202Z" }, - { url = "https://files.pythonhosted.org/packages/1d/74/ece2e582a0d9550cee33e2e416cc96737dce423a994d12bbe59716f47ff1/scipy-1.16.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:0851f6a1e537fe9399f35986897e395a1aa61c574b178c0d456be5b1a0f5ca1f", size = 33286040, upload-time = "2025-07-27T16:29:10.201Z" }, - { url = "https://files.pythonhosted.org/packages/e4/82/08e4076df538fb56caa1d489588d880ec7c52d8273a606bb54d660528f7c/scipy-1.16.1-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fedc2cbd1baed37474b1924c331b97bdff611d762c196fac1a9b71e67b813b1b", size = 35176096, upload-time = "2025-07-27T16:29:17.091Z" }, - { url = "https://files.pythonhosted.org/packages/fa/79/cd710aab8c921375711a8321c6be696e705a120e3011a643efbbcdeeabcc/scipy-1.16.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:2ef500e72f9623a6735769e4b93e9dcb158d40752cdbb077f305487e3e2d1f45", size = 35490328, upload-time = "2025-07-27T16:29:22.928Z" }, - { url = "https://files.pythonhosted.org/packages/71/73/e9cc3d35ee4526d784520d4494a3e1ca969b071fb5ae5910c036a375ceec/scipy-1.16.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:978d8311674b05a8f7ff2ea6c6bce5d8b45a0cb09d4c5793e0318f448613ea65", size = 37939921, upload-time = "2025-07-27T16:29:29.108Z" }, - { url = "https://files.pythonhosted.org/packages/21/12/c0efd2941f01940119b5305c375ae5c0fcb7ec193f806bd8f158b73a1782/scipy-1.16.1-cp313-cp313-win_amd64.whl", hash = "sha256:81929ed0fa7a5713fcdd8b2e6f73697d3b4c4816d090dd34ff937c20fa90e8ab", size = 38479462, upload-time = "2025-07-27T16:30:24.078Z" }, - { url = "https://files.pythonhosted.org/packages/7a/19/c3d08b675260046a991040e1ea5d65f91f40c7df1045fffff412dcfc6765/scipy-1.16.1-cp313-cp313t-macosx_10_14_x86_64.whl", hash = "sha256:bcc12db731858abda693cecdb3bdc9e6d4bd200213f49d224fe22df82687bdd6", size = 36938832, upload-time = "2025-07-27T16:29:35.057Z" }, - { url = "https://files.pythonhosted.org/packages/81/f2/ce53db652c033a414a5b34598dba6b95f3d38153a2417c5a3883da429029/scipy-1.16.1-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:744d977daa4becb9fc59135e75c069f8d301a87d64f88f1e602a9ecf51e77b27", size = 29093084, upload-time = "2025-07-27T16:29:40.201Z" }, - { url = "https://files.pythonhosted.org/packages/a9/ae/7a10ff04a7dc15f9057d05b33737ade244e4bd195caa3f7cc04d77b9e214/scipy-1.16.1-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:dc54f76ac18073bcecffb98d93f03ed6b81a92ef91b5d3b135dcc81d55a724c7", size = 21365098, upload-time = "2025-07-27T16:29:44.295Z" }, - { url = "https://files.pythonhosted.org/packages/36/ac/029ff710959932ad3c2a98721b20b405f05f752f07344622fd61a47c5197/scipy-1.16.1-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:367d567ee9fc1e9e2047d31f39d9d6a7a04e0710c86e701e053f237d14a9b4f6", size = 23896858, upload-time = "2025-07-27T16:29:48.784Z" }, - { url = "https://files.pythonhosted.org/packages/71/13/d1ef77b6bd7898720e1f0b6b3743cb945f6c3cafa7718eaac8841035ab60/scipy-1.16.1-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:4cf5785e44e19dcd32a0e4807555e1e9a9b8d475c6afff3d21c3c543a6aa84f4", size = 33438311, upload-time = "2025-07-27T16:29:54.164Z" }, - { url = "https://files.pythonhosted.org/packages/2d/e0/e64a6821ffbb00b4c5b05169f1c1fddb4800e9307efe3db3788995a82a2c/scipy-1.16.1-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:3d0b80fb26d3e13a794c71d4b837e2a589d839fd574a6bbb4ee1288c213ad4a3", size = 35279542, upload-time = "2025-07-27T16:30:00.249Z" }, - { url = "https://files.pythonhosted.org/packages/57/59/0dc3c8b43e118f1e4ee2b798dcc96ac21bb20014e5f1f7a8e85cc0653bdb/scipy-1.16.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:8503517c44c18d1030d666cb70aaac1cc8913608816e06742498833b128488b7", size = 35667665, upload-time = "2025-07-27T16:30:05.916Z" }, - { url = "https://files.pythonhosted.org/packages/45/5f/844ee26e34e2f3f9f8febb9343748e72daeaec64fe0c70e9bf1ff84ec955/scipy-1.16.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:30cc4bb81c41831ecfd6dc450baf48ffd80ef5aed0f5cf3ea775740e80f16ecc", size = 38045210, upload-time = "2025-07-27T16:30:11.655Z" }, - { url = "https://files.pythonhosted.org/packages/8d/d7/210f2b45290f444f1de64bc7353aa598ece9f0e90c384b4a156f9b1a5063/scipy-1.16.1-cp313-cp313t-win_amd64.whl", hash = "sha256:c24fa02f7ed23ae514460a22c57eca8f530dbfa50b1cfdbf4f37c05b5309cc39", size = 38593661, upload-time = "2025-07-27T16:30:17.825Z" }, - { url = "https://files.pythonhosted.org/packages/81/ea/84d481a5237ed223bd3d32d6e82d7a6a96e34756492666c260cef16011d1/scipy-1.16.1-cp314-cp314-macosx_10_14_x86_64.whl", hash = "sha256:796a5a9ad36fa3a782375db8f4241ab02a091308eb079746bc0f874c9b998318", size = 36525921, upload-time = "2025-07-27T16:30:30.081Z" }, - { url = "https://files.pythonhosted.org/packages/4e/9f/d9edbdeff9f3a664807ae3aea383e10afaa247e8e6255e6d2aa4515e8863/scipy-1.16.1-cp314-cp314-macosx_12_0_arm64.whl", hash = "sha256:3ea0733a2ff73fd6fdc5fecca54ee9b459f4d74f00b99aced7d9a3adb43fb1cc", size = 28564152, upload-time = "2025-07-27T16:30:35.336Z" }, - { url = "https://files.pythonhosted.org/packages/3b/95/8125bcb1fe04bc267d103e76516243e8d5e11229e6b306bda1024a5423d1/scipy-1.16.1-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:85764fb15a2ad994e708258bb4ed8290d1305c62a4e1ef07c414356a24fcfbf8", size = 20836028, upload-time = "2025-07-27T16:30:39.421Z" }, - { url = "https://files.pythonhosted.org/packages/77/9c/bf92e215701fc70bbcd3d14d86337cf56a9b912a804b9c776a269524a9e9/scipy-1.16.1-cp314-cp314-macosx_14_0_x86_64.whl", hash = "sha256:ca66d980469cb623b1759bdd6e9fd97d4e33a9fad5b33771ced24d0cb24df67e", size = 23489666, upload-time = "2025-07-27T16:30:43.663Z" }, - { url = "https://files.pythonhosted.org/packages/5e/00/5e941d397d9adac41b02839011594620d54d99488d1be5be755c00cde9ee/scipy-1.16.1-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e7cc1ffcc230f568549fc56670bcf3df1884c30bd652c5da8138199c8c76dae0", size = 33358318, upload-time = "2025-07-27T16:30:48.982Z" }, - { url = "https://files.pythonhosted.org/packages/0e/87/8db3aa10dde6e3e8e7eb0133f24baa011377d543f5b19c71469cf2648026/scipy-1.16.1-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:3ddfb1e8d0b540cb4ee9c53fc3dea3186f97711248fb94b4142a1b27178d8b4b", size = 35185724, upload-time = "2025-07-27T16:30:54.26Z" }, - { url = "https://files.pythonhosted.org/packages/89/b4/6ab9ae443216807622bcff02690262d8184078ea467efee2f8c93288a3b1/scipy-1.16.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:4dc0e7be79e95d8ba3435d193e0d8ce372f47f774cffd882f88ea4e1e1ddc731", size = 35554335, upload-time = "2025-07-27T16:30:59.765Z" }, - { url = "https://files.pythonhosted.org/packages/9c/9a/d0e9dc03c5269a1afb60661118296a32ed5d2c24298af61b676c11e05e56/scipy-1.16.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:f23634f9e5adb51b2a77766dac217063e764337fbc816aa8ad9aaebcd4397fd3", size = 37960310, upload-time = "2025-07-27T16:31:06.151Z" }, - { url = "https://files.pythonhosted.org/packages/5e/00/c8f3130a50521a7977874817ca89e0599b1b4ee8e938bad8ae798a0e1f0d/scipy-1.16.1-cp314-cp314-win_amd64.whl", hash = "sha256:57d75524cb1c5a374958a2eae3d84e1929bb971204cc9d52213fb8589183fc19", size = 39319239, upload-time = "2025-07-27T16:31:59.942Z" }, - { url = "https://files.pythonhosted.org/packages/f2/f2/1ca3eda54c3a7e4c92f6acef7db7b3a057deb135540d23aa6343ef8ad333/scipy-1.16.1-cp314-cp314t-macosx_10_14_x86_64.whl", hash = "sha256:d8da7c3dd67bcd93f15618938f43ed0995982eb38973023d46d4646c4283ad65", size = 36939460, upload-time = "2025-07-27T16:31:11.865Z" }, - { url = "https://files.pythonhosted.org/packages/80/30/98c2840b293a132400c0940bb9e140171dcb8189588619048f42b2ce7b4f/scipy-1.16.1-cp314-cp314t-macosx_12_0_arm64.whl", hash = "sha256:cc1d2f2fd48ba1e0620554fe5bc44d3e8f5d4185c8c109c7fbdf5af2792cfad2", size = 29093322, upload-time = "2025-07-27T16:31:17.045Z" }, - { url = "https://files.pythonhosted.org/packages/c1/e6/1e6e006e850622cf2a039b62d1a6ddc4497d4851e58b68008526f04a9a00/scipy-1.16.1-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:21a611ced9275cb861bacadbada0b8c0623bc00b05b09eb97f23b370fc2ae56d", size = 21365329, upload-time = "2025-07-27T16:31:21.188Z" }, - { url = "https://files.pythonhosted.org/packages/8e/02/72a5aa5b820589dda9a25e329ca752842bfbbaf635e36bc7065a9b42216e/scipy-1.16.1-cp314-cp314t-macosx_14_0_x86_64.whl", hash = "sha256:8dfbb25dffc4c3dd9371d8ab456ca81beeaf6f9e1c2119f179392f0dc1ab7695", size = 23897544, upload-time = "2025-07-27T16:31:25.408Z" }, - { url = "https://files.pythonhosted.org/packages/2b/dc/7122d806a6f9eb8a33532982234bed91f90272e990f414f2830cfe656e0b/scipy-1.16.1-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:f0ebb7204f063fad87fc0a0e4ff4a2ff40b2a226e4ba1b7e34bf4b79bf97cd86", size = 33442112, upload-time = "2025-07-27T16:31:30.62Z" }, - { url = "https://files.pythonhosted.org/packages/24/39/e383af23564daa1021a5b3afbe0d8d6a68ec639b943661841f44ac92de85/scipy-1.16.1-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f1b9e5962656f2734c2b285a8745358ecb4e4efbadd00208c80a389227ec61ff", size = 35286594, upload-time = "2025-07-27T16:31:36.112Z" }, - { url = "https://files.pythonhosted.org/packages/95/47/1a0b0aff40c3056d955f38b0df5d178350c3d74734ec54f9c68d23910be5/scipy-1.16.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:5e1a106f8c023d57a2a903e771228bf5c5b27b5d692088f457acacd3b54511e4", size = 35665080, upload-time = "2025-07-27T16:31:42.025Z" }, - { url = "https://files.pythonhosted.org/packages/64/df/ce88803e9ed6e27fe9b9abefa157cf2c80e4fa527cf17ee14be41f790ad4/scipy-1.16.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:709559a1db68a9abc3b2c8672c4badf1614f3b440b3ab326d86a5c0491eafae3", size = 38050306, upload-time = "2025-07-27T16:31:48.109Z" }, - { url = "https://files.pythonhosted.org/packages/6e/6c/a76329897a7cae4937d403e623aa6aaea616a0bb5b36588f0b9d1c9a3739/scipy-1.16.1-cp314-cp314t-win_amd64.whl", hash = "sha256:c0c804d60492a0aad7f5b2bb1862f4548b990049e27e828391ff2bf6f7199998", size = 39427705, upload-time = "2025-07-27T16:31:53.96Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/86/59/a451d7420a77ab0b98f7affa3a1d78a313d2f7281a57afb1a34bae8ab412/seaborn-0.13.2.tar.gz", hash = "sha256:93e60a40988f4d65e9f4885df477e2fdaff6b73a9ded434c1ab356dd57eefff7", size = 1457696, upload-time = "2024-01-25T13:21:52.551Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/83/11/00d3c3dfc25ad54e731d91449895a79e4bf2384dc3ac01809010ba88f6d5/seaborn-0.13.2-py3-none-any.whl", hash = "sha256:636f8336facf092165e27924f223d3c62ca560b1f2bb5dff7ab7fad265361987", size = 294914, upload-time = "2024-01-25T13:21:49.598Z" }, ] [[package]] name = "sentry-sdk" -version = "2.34.1" +version = "2.24.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "certifi" }, { name = "urllib3" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/3a/38/10d6bfe23df1bfc65ac2262ed10b45823f47f810b0057d3feeea1ca5c7ed/sentry_sdk-2.34.1.tar.gz", hash = "sha256:69274eb8c5c38562a544c3e9f68b5be0a43be4b697f5fd385bf98e4fbe672687", size = 336969, upload-time = "2025-07-30T11:13:37.93Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/2d/3e/bb34de65a5787f76848a533afbb6610e01fbcdd59e76d8679c254e02255c/sentry_sdk-2.34.1-py2.py3-none-any.whl", hash = "sha256:b7a072e1cdc5abc48101d5146e1ae680fa81fe886d8d95aaa25a0b450c818d32", size = 357743, upload-time = "2025-07-30T11:13:36.145Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/fe/c6/61d3d5eecca9f0237251a89043d1bfa3fbeb4e35d48c1cbba2f6dde72664/sentry_sdk-2.24.0.tar.gz", hash = "sha256:4a4a8de31573c8ab14c9b866fd44cf783df062ca7b4a56ed0a108453abbc2a24", size = 317058, upload-time = "2025-03-21T12:33:59.537Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/c4/5d/dbfbda95daa8fdb22276513f930269149b338b67efb9ebf24ebf6d62ca9a/sentry_sdk-2.24.0-py2.py3-none-any.whl", hash = "sha256:7150cfe61dfd37d30b33d8d6b153d25e14c69bbcf6f4a98ffc97e92de88be215", size = 336890, upload-time = "2025-03-21T12:33:57.754Z" }, +] + +[[package]] +name = "setproctitle" +version = "1.3.5" +source = { registry = "https://pypi.org/simple" } +sdist = { url = "https://files.pythonhosted.org/packages/c4/4d/6a840c8d2baa07b57329490e7094f90aac177a1d5226bc919046f1106860/setproctitle-1.3.5.tar.gz", hash = "sha256:1e6eaeaf8a734d428a95d8c104643b39af7d247d604f40a7bebcf3960a853c5e", size = 26737, upload-time = "2025-02-22T21:52:43.276Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/9d/e1/9ccff2682c38061baa07e128b60712bc18e3398aa7d5471c51a704f9d24c/setproctitle-1.3.5-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:02870e0cb0de7f68a7a8a5b23c2bc0ce63821cab3d9b126f9be80bb6cd674c80", size = 17256, upload-time = "2025-02-22T21:50:22.744Z" }, + { url = "https://files.pythonhosted.org/packages/ed/64/936c1f92d60052f11a8de9f90a4b7ec4996b8ebd6d67ba425ed214c80771/setproctitle-1.3.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:55b278135be742b8901067479626d909f6613bd2d2c4fd0de6bb46f80e07a919", size = 11893, upload-time = "2025-02-22T21:50:25.255Z" }, + { url = "https://files.pythonhosted.org/packages/01/2d/abc817b3778d9b1f7675020030379a0c39e0bf74b36af211b26191a63da3/setproctitle-1.3.5-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:53fc971f7bf7a674f571a23cdec70f2f0ac88152c59c06aa0808d0be6d834046", size = 31295, upload-time = "2025-02-22T21:50:27.389Z" }, + { url = "https://files.pythonhosted.org/packages/03/4d/e2055dfb1b492fd3a3b27deeaa642d81c580d48a16bc9b07afc3504af677/setproctitle-1.3.5-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fb0500e1bc6f00b8ba696c3743ddff14c8679e3c2ca9d292c008ac51488d17cf", size = 32637, upload-time = "2025-02-22T21:50:29.47Z" }, + { url = "https://files.pythonhosted.org/packages/89/28/a1f23d7d127dff59fe75ad671d1d5c83ab8cba10d0e343820b96d5d8a2f7/setproctitle-1.3.5-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:995b3ac1b5fe510f4e1d1c19ebf19f4bceb448f2d6e8d99ea23f33cb6f1a277e", size = 29772, upload-time = "2025-02-22T21:50:30.597Z" }, + { url = "https://files.pythonhosted.org/packages/df/46/2ea4d436c7d664d41df7e60fbd3103f1139a931638e998f478e870e72255/setproctitle-1.3.5-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a5a05e2c3fdfbda32b9c9da72d0506398d1efb5bd2c5981b9e12d3622eb3d4f9", size = 30811, upload-time = "2025-02-22T21:50:32.802Z" }, + { url = "https://files.pythonhosted.org/packages/45/60/4c17211c2d80e6fe9fa486fa3214d565d0cd9a6eff0b67e6219ddb2ba49c/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:310c7f4ca4c8476a9840b2cd4b22ee602a49a3c902fdcd2dd8284685abd10a9a", size = 30442, upload-time = "2025-02-22T21:50:35.057Z" }, + { url = "https://files.pythonhosted.org/packages/7e/bf/65a8f8f2d03cd9a9429cfa0d6b22282ff7a609a4d08602bcb8351a271bec/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:867af4a5c3d85484fbcc50ea88bcd375acf709cff88a3259575361849c0da351", size = 29492, upload-time = "2025-02-22T21:50:37.23Z" }, + { url = "https://files.pythonhosted.org/packages/c6/96/56f45f0b81fcc776f925c34e2699040df39cfc6b3cc7520d9b378314435b/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:8ec0a7fe9f1ba90900144489bc93ce7dd4dec3f3df1e7f188c9e58364fe4a4c5", size = 31947, upload-time = "2025-02-22T21:50:38.65Z" }, + { url = "https://files.pythonhosted.org/packages/ec/9d/6b697c1562b21368e579d820bca2a607e565638fd332247841eb65dec4b2/setproctitle-1.3.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:aaee7acba2733a14a886488b7495bfec4a8d6407124c04a0946dbde1684230a3", size = 29863, upload-time = "2025-02-22T21:50:40.775Z" }, + { url = "https://files.pythonhosted.org/packages/ba/0f/4551cbb120d003fa1284ee35d559366e09b513a87dfee02f804da1936054/setproctitle-1.3.5-cp310-cp310-win32.whl", hash = "sha256:bd2cccd972e4282af4ce2c13cd9ebdf07be157eabafd8ce648fffdc8ae6fbe28", size = 11471, upload-time = "2025-02-22T21:50:42.749Z" }, + { url = "https://files.pythonhosted.org/packages/a6/f4/2dd926687b7a3bdaa83533e2898f929e1ff3bdeb6aa271bdb1d4d5923c7e/setproctitle-1.3.5-cp310-cp310-win_amd64.whl", hash = "sha256:81f2328ac34c9584e1e5f87eea916c0bc48476a06606a07debae07acdd7ab5ea", size = 12196, upload-time = "2025-02-22T21:50:43.852Z" }, + { url = "https://files.pythonhosted.org/packages/ec/4a/9e0243c5df221102fb834a947f5753d9da06ad5f84e36b0e2e93f7865edb/setproctitle-1.3.5-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:1c8dcc250872385f2780a5ea58050b58cbc8b6a7e8444952a5a65c359886c593", size = 17256, upload-time = "2025-02-22T21:50:45.928Z" }, + { url = "https://files.pythonhosted.org/packages/c7/a1/76ad2ba6f5bd00609238e3d64eeded4598e742a5f25b5cc1a0efdae5f674/setproctitle-1.3.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:ca82fae9eb4800231dd20229f06e8919787135a5581da245b8b05e864f34cc8b", size = 11893, upload-time = "2025-02-22T21:50:47.167Z" }, + { url = "https://files.pythonhosted.org/packages/47/3a/75d11fedff5b21ba9a4c5fe3dfa5e596f831d094ef1896713a72e9e38833/setproctitle-1.3.5-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0424e1d33232322541cb36fb279ea5242203cd6f20de7b4fb2a11973d8e8c2ce", size = 31631, upload-time = "2025-02-22T21:50:50.863Z" }, + { url = "https://files.pythonhosted.org/packages/5a/12/58220de5600e0ed2e5562297173187d863db49babb03491ffe9c101299bc/setproctitle-1.3.5-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:fec8340ab543144d04a9d805d80a0aad73fdeb54bea6ff94e70d39a676ea4ec0", size = 32975, upload-time = "2025-02-22T21:50:52.188Z" }, + { url = "https://files.pythonhosted.org/packages/fa/c4/fbb308680d83c1c7aa626950308318c6e6381a8273779163a31741f3c752/setproctitle-1.3.5-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:eab441c89f181271ab749077dcc94045a423e51f2fb0b120a1463ef9820a08d0", size = 30126, upload-time = "2025-02-22T21:50:53.496Z" }, + { url = "https://files.pythonhosted.org/packages/31/6e/baaf70bd9a881dd8c12cbccdd7ca0ff291024a37044a8245e942e12e7135/setproctitle-1.3.5-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d2c371550a2288901a0dcd84192691ebd3197a43c95f3e0b396ed6d1cedf5c6c", size = 31135, upload-time = "2025-02-22T21:50:54.931Z" }, + { url = "https://files.pythonhosted.org/packages/a6/dc/d8ab6b1c3d844dc14f596e3cce76604570848f8a67ba6a3812775ed2c015/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:78288ff5f9c415c56595b2257ad218936dd9fa726b36341b373b31ca958590fe", size = 30874, upload-time = "2025-02-22T21:50:57.042Z" }, + { url = "https://files.pythonhosted.org/packages/d4/84/62a359b3aa51228bd88f78b44ebb0256a5b96dd2487881c1e984a59b617d/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:f1f13a25fc46731acab518602bb1149bfd8b5fabedf8290a7c0926d61414769d", size = 29893, upload-time = "2025-02-22T21:50:59.644Z" }, + { url = "https://files.pythonhosted.org/packages/e2/d6/b3c52c03ee41e7f006e1a737e0db1c58d1dc28e258b83548e653d0c34f1c/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:1534d6cd3854d035e40bf4c091984cbdd4d555d7579676d406c53c8f187c006f", size = 32293, upload-time = "2025-02-22T21:51:01.777Z" }, + { url = "https://files.pythonhosted.org/packages/55/09/c0ba311879d9c05860503a7e2708ace85913b9a816786402a92c664fe930/setproctitle-1.3.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:62a01c76708daac78b9688ffb95268c57cb57fa90b543043cda01358912fe2db", size = 30247, upload-time = "2025-02-22T21:51:03.779Z" }, + { url = "https://files.pythonhosted.org/packages/9e/43/cc7155461f0b5a48aebdb87d78239ff3a51ebda0905de478d9fa6ab92d9c/setproctitle-1.3.5-cp311-cp311-win32.whl", hash = "sha256:ea07f29735d839eaed985990a0ec42c8aecefe8050da89fec35533d146a7826d", size = 11476, upload-time = "2025-02-22T21:51:05.746Z" }, + { url = "https://files.pythonhosted.org/packages/e7/57/6e937ac7aa52db69225f02db2cfdcb66ba1db6fdc65a4ddbdf78e214f72a/setproctitle-1.3.5-cp311-cp311-win_amd64.whl", hash = "sha256:ab3ae11e10d13d514d4a5a15b4f619341142ba3e18da48c40e8614c5a1b5e3c3", size = 12189, upload-time = "2025-02-22T21:51:07.837Z" }, + { url = "https://files.pythonhosted.org/packages/2b/19/04755958495de57e4891de50f03e77b3fe9ca6716a86de00faa00ad0ee5a/setproctitle-1.3.5-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:523424b9be4dea97d95b8a584b183f35c7bab2d0a3d995b01febf5b8a8de90e4", size = 17250, upload-time = "2025-02-22T21:51:09.785Z" }, + { url = "https://files.pythonhosted.org/packages/b9/3d/2ca9df5aa49b975296411dcbbe272cdb1c5e514c43b8be7d61751bb71a46/setproctitle-1.3.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:b6ec1d86c1b4d7b5f2bdceadf213310cf24696b82480a2a702194b8a0bfbcb47", size = 11878, upload-time = "2025-02-22T21:51:11.679Z" }, + { url = "https://files.pythonhosted.org/packages/36/d6/e90e23b4627e016a4f862d4f892be92c9765dd6bf1e27a48e52cd166d4a3/setproctitle-1.3.5-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ea6c505264275a43e9b2acd2acfc11ac33caf52bc3167c9fced4418a810f6b1c", size = 31940, upload-time = "2025-02-22T21:51:12.977Z" }, + { url = "https://files.pythonhosted.org/packages/15/13/167cdd55e00a8e10b36aad79646c3bf3c23fba0c08a9b8db9b74622c1b13/setproctitle-1.3.5-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0b91e68e6685998e6353f296100ecabc313a6cb3e413d66a03d74b988b61f5ff", size = 33370, upload-time = "2025-02-22T21:51:15.115Z" }, + { url = "https://files.pythonhosted.org/packages/9b/22/574a110527df133409a75053b7d6ff740993ccf30b8713d042f26840d351/setproctitle-1.3.5-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bc1fda208ae3a2285ad27aeab44c41daf2328abe58fa3270157a739866779199", size = 30628, upload-time = "2025-02-22T21:51:16.324Z" }, + { url = "https://files.pythonhosted.org/packages/52/79/78b05c7d792c9167b917acdab1773b1ff73b016560f45d8155be2baa1a82/setproctitle-1.3.5-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:828727d220e46f048b82289018300a64547b46aaed96bf8810c05fe105426b41", size = 31672, upload-time = "2025-02-22T21:51:17.791Z" }, + { url = "https://files.pythonhosted.org/packages/b0/62/4509735be062129694751ac55d5e1fbb6d86fa46a8689b7d5e2c23dae5b0/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:83b016221cf80028b2947be20630faa14e3e72a403e35f0ba29550b4e856767b", size = 31378, upload-time = "2025-02-22T21:51:19.404Z" }, + { url = "https://files.pythonhosted.org/packages/72/e7/b394c55934b89f00c2ef7d5e6f18cca5d8dfa26ef628700c4de0c85e3f3d/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:6d8a411e752e794d052434139ca4234ffeceeb8d8d8ddc390a9051d7942b2726", size = 30370, upload-time = "2025-02-22T21:51:21.218Z" }, + { url = "https://files.pythonhosted.org/packages/13/ee/e1f27bf52d2bec7060bb6311ab0ccede8de98ed5394e3a59e7a14a453fb5/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:50cfbf86b9c63a2c2903f1231f0a58edeb775e651ae1af84eec8430b0571f29b", size = 32875, upload-time = "2025-02-22T21:51:22.505Z" }, + { url = "https://files.pythonhosted.org/packages/6e/08/13b561085d2de53b9becfa5578545d99114e9ff2aa3dc151bcaadf80b17e/setproctitle-1.3.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:f3b5e2eacd572444770026c9dd3ddc7543ce427cdf452d40a408d1e95beefb30", size = 30903, upload-time = "2025-02-22T21:51:23.732Z" }, + { url = "https://files.pythonhosted.org/packages/65/f0/6cd06fffff2553be7b0571447d0c0ef8b727ef44cc2d6a33452677a311c8/setproctitle-1.3.5-cp312-cp312-win32.whl", hash = "sha256:cf4e3ded98027de2596c6cc5bbd3302adfb3ca315c848f56516bb0b7e88de1e9", size = 11468, upload-time = "2025-02-22T21:51:25.45Z" }, + { url = "https://files.pythonhosted.org/packages/c1/8c/e8a7cb568c4552618838941b332203bfc77ab0f2d67c1cb8f24dee0370ec/setproctitle-1.3.5-cp312-cp312-win_amd64.whl", hash = "sha256:f7a8c01ffd013dda2bed6e7d5cb59fbb609e72f805abf3ee98360f38f7758d9b", size = 12190, upload-time = "2025-02-22T21:51:26.78Z" }, + { url = "https://files.pythonhosted.org/packages/ab/78/d6b5aa3af2dd64f6c32e78fb85797b9725a3cdcbdf17dffc5838019918c3/setproctitle-1.3.5-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:162fd76781f57f42ddf27c475e5fef6a8df4fdd69b28dd554e53e2eb2bfe0f95", size = 17238, upload-time = "2025-02-22T21:51:28.451Z" }, + { url = "https://files.pythonhosted.org/packages/3d/00/14781f0ac28c7a37fe2ba321c276188ddd5ca73d69dab8a0f739d57b776b/setproctitle-1.3.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:4969d996bdfbe23bbd023cd0bae6c73a27371615c4ec5296a60cecce268659ef", size = 11867, upload-time = "2025-02-22T21:51:29.649Z" }, + { url = "https://files.pythonhosted.org/packages/f0/22/8430c879a8e3201508924a6cf45dba92b9a7b105fac8eebd0ef62e60fba9/setproctitle-1.3.5-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bd70c95a94473216e7c7a7a1f7d8ecbaca5b16d4ba93ddbfd32050fc485a8451", size = 32001, upload-time = "2025-02-22T21:51:32.21Z" }, + { url = "https://files.pythonhosted.org/packages/01/f2/b00fe72c20897695f85932d193a5c57ecf94cbf825c0fd4082e3fa3e00bd/setproctitle-1.3.5-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:7a887582bfdb6dcbc482db0ef9e630ad23ca95875806ef2b444bf6fbd7b7d7ca", size = 33415, upload-time = "2025-02-22T21:51:33.427Z" }, + { url = "https://files.pythonhosted.org/packages/11/5b/e497bf702ea5d553a331ca879e73a18bbd8f7d66d18d275cb2324e4144c4/setproctitle-1.3.5-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:755671c39a9e70834eeec6dc6b61e344399c49881d2e7ea3534a1c69669dd9cc", size = 30606, upload-time = "2025-02-22T21:51:34.729Z" }, + { url = "https://files.pythonhosted.org/packages/16/99/1bcb837134c71f332bfeaf923e68279566362b7d1504aa106af8046696e8/setproctitle-1.3.5-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9ab52b4c2ce056a1b60d439991a81ca90f019488d4b4f64b2779e6badd3677e6", size = 31679, upload-time = "2025-02-22T21:51:37.018Z" }, + { url = "https://files.pythonhosted.org/packages/77/55/72af3dbb0b1304bad54ea3b7cf1b524a8a2868da0b4c38bc18290f0097f7/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:36178b944019ec7fc52bb967ffeee296a11d373734a7be276755bedb3db5c141", size = 31388, upload-time = "2025-02-22T21:51:38.377Z" }, + { url = "https://files.pythonhosted.org/packages/f3/08/fa13f2da6bd10ca756a45f8fed2888f439e9ce7d6402258e87ceef2d4c71/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:269d41cd4f085b69821d1ee6599124f02dbbc79962b256e260b6c9021d037994", size = 30370, upload-time = "2025-02-22T21:51:39.879Z" }, + { url = "https://files.pythonhosted.org/packages/25/4b/83575bb403967f1069b68a8799979fe7979b5a7c17703d2984965d8f4e92/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d880630fd81d1b3bde121c352ca7ea2f2ff507ef40c3c011d0928ed491f912c9", size = 32897, upload-time = "2025-02-22T21:51:42.376Z" }, + { url = "https://files.pythonhosted.org/packages/1a/71/0c1e151ef6899260da4009e7170f56261486d3149e9bad40990b52bdd620/setproctitle-1.3.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8a7fed67ab49f60bd51f3b4cffff3f8d754d1bb0a40e42869911301ec6519b65", size = 30944, upload-time = "2025-02-22T21:51:43.698Z" }, + { url = "https://files.pythonhosted.org/packages/38/34/a3bdaeaee03e11aef82b45014738f1210f90e37359c41eda3e49b4ce891c/setproctitle-1.3.5-cp313-cp313-win32.whl", hash = "sha256:e9c0d0cfcf715631b10d5950d04a9978f63bc46535724ef7c2eaf1dca9988642", size = 11463, upload-time = "2025-02-22T21:51:44.869Z" }, + { url = "https://files.pythonhosted.org/packages/ef/f1/a19cde9f3f4054aed7c6077e7fc3420a5151ec6173cf3235fe000722ccb8/setproctitle-1.3.5-cp313-cp313-win_amd64.whl", hash = "sha256:e1d28eb98c91fbebd3e443a45c7da5d84974959851ef304c330eabd654a386f1", size = 12182, upload-time = "2025-02-22T21:51:46.033Z" }, + { url = "https://files.pythonhosted.org/packages/4a/ba/2524329ce958599069f0d0e4cfd3d6fbb7c58a4408b9e5609698e47353ec/setproctitle-1.3.5-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:dc66b84beb0d5eb03abf0c3140c6d2cbe3d67ae9f0824a09dfa8c6ff164319a6", size = 11418, upload-time = "2025-02-22T21:52:24.881Z" }, + { url = "https://files.pythonhosted.org/packages/a6/5f/a049640b05c609585ad0f471e667be0fd9ab533219127b455826d31587d5/setproctitle-1.3.5-pp310-pypy310_pp73-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:31dc9b330e7cac7685bdef790747c07914081c11ee1066eb0c597303dfb52010", size = 13425, upload-time = "2025-02-22T21:52:26.833Z" }, + { url = "https://files.pythonhosted.org/packages/a9/15/caa47039e267ea67316b285e2e308ae529872ad6a143edf03a7d8edf6175/setproctitle-1.3.5-pp310-pypy310_pp73-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4028639b511f5e641d116b3b54ad70c637ebd1b4baac0948283daf11b104119f", size = 13026, upload-time = "2025-02-22T21:52:28.783Z" }, + { url = "https://files.pythonhosted.org/packages/c1/a2/1fb0647a251f4c788b94f751cf23171b2a905758fd13ef8d126222d41428/setproctitle-1.3.5-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:6bddef4e27d0ed74e44b58bf050bc3108591bf17d20d461fc59cd141282f849c", size = 12222, upload-time = "2025-02-22T21:52:31.088Z" }, ] [[package]] @@ -3699,43 +3436,41 @@ wheels = [ [[package]] name = "statsmodels" -version = "0.14.5" +version = "0.14.4" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "packaging" }, { name = "pandas" }, { name = "patsy" }, - { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "scipy", version = "1.16.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/64/cc/8c1bf59bf8203dea1bf2ea811cfe667d7bcc6909c83d8afb02b08e30f50b/statsmodels-0.14.5.tar.gz", hash = "sha256:de260e58cccfd2ceddf835b55a357233d6ca853a1aa4f90f7553a52cc71c6ddf", size = 20525016, upload-time = "2025-07-07T12:14:23.195Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/3a/2c/55b2a5d10c1a211ecab3f792021d2581bbe1c5ca0a1059f6715dddc6899d/statsmodels-0.14.5-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:9fc2b5cdc0c95cba894849651fec1fa1511d365e3eb72b0cc75caac44077cd48", size = 10058241, upload-time = "2025-07-07T12:13:16.286Z" }, - { url = "https://files.pythonhosted.org/packages/66/d9/6967475805de06691e951072d05e40e3f1c71b6221bb92401193ee19bd2a/statsmodels-0.14.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:b8d96b0bbaeabd3a557c35cc7249baa9cfbc6dd305c32a9f2cbdd7f46c037e7f", size = 9734017, upload-time = "2025-07-07T12:05:08.498Z" }, - { url = "https://files.pythonhosted.org/packages/df/a8/803c280419a7312e2472969fe72cf461c1210a27770a662cbe3b5cd7c6fe/statsmodels-0.14.5-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:145bc39b2cb201efb6c83cc3f2163c269e63b0d4809801853dec6f440bd3bc37", size = 10459677, upload-time = "2025-07-07T14:21:51.809Z" }, - { url = "https://files.pythonhosted.org/packages/a1/25/edf20acbd670934b02cd9344e29c9a03ce040122324b3491bb075ae76b2d/statsmodels-0.14.5-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d7c14fb2617bb819fb2532e1424e1da2b98a3419a80e95f33365a72d437d474e", size = 10678631, upload-time = "2025-07-07T14:22:05.496Z" }, - { url = "https://files.pythonhosted.org/packages/64/22/8b1e38310272e766abd6093607000a81827420a3348f09eff08a9e54cbaf/statsmodels-0.14.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:1e9742d8a5ac38a3bfc4b7f4b0681903920f20cbbf466d72b1fd642033846108", size = 10699273, upload-time = "2025-07-07T14:22:19.487Z" }, - { url = "https://files.pythonhosted.org/packages/d1/6f/6de51f1077b7cef34611f1d6721392ea170153251b4d977efcf6d100f779/statsmodels-0.14.5-cp310-cp310-win_amd64.whl", hash = "sha256:1cab9e6fce97caf4239cdb2df375806937da5d0b7ba2699b13af33a07f438464", size = 9644785, upload-time = "2025-07-07T12:05:20.927Z" }, - { url = "https://files.pythonhosted.org/packages/14/30/fd49902b30416b828de763e161c0d6e2cc04d119ae4fbdd3f3b43dc8f1be/statsmodels-0.14.5-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:4b7091a8442076c708c926de3603653a160955e80a2b6d931475b7bb8ddc02e5", size = 10053330, upload-time = "2025-07-07T12:07:39.689Z" }, - { url = "https://files.pythonhosted.org/packages/ca/c1/2654541ff6f5790d01d1e5ba36405fde873f4a854f473e90b4fe56b37333/statsmodels-0.14.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:128872be8f3208f4446d91ea9e4261823902fc7997fee7e1a983eb62fd3b7c6e", size = 9735555, upload-time = "2025-07-07T12:13:28.935Z" }, - { url = "https://files.pythonhosted.org/packages/ce/da/6ebb64d0db4e86c0d2d9cde89e03247702da0ab191789f7813d4f9a348da/statsmodels-0.14.5-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f2ad5aee04ae7196c429df2174df232c057e478c5fa63193d01c8ec9aae04d31", size = 10307522, upload-time = "2025-07-07T14:22:32.853Z" }, - { url = "https://files.pythonhosted.org/packages/67/49/ac803ca093ec3845184a752a91cd84511245e1f97103b15cfe32794a3bb0/statsmodels-0.14.5-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f402fc793458dd6d96e099acb44cd1de1428565bf7ef3030878a8daff091f08a", size = 10474665, upload-time = "2025-07-07T14:22:46.011Z" }, - { url = "https://files.pythonhosted.org/packages/f0/c8/ae82feb00582f4814fac5d2cb3ec32f93866b413cf5878b2fe93688ec63c/statsmodels-0.14.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:26c028832730aebfbfd4e7501694e1f9ad31ec8536e776716673f4e7afd4059a", size = 10713120, upload-time = "2025-07-07T14:23:00.067Z" }, - { url = "https://files.pythonhosted.org/packages/05/ac/4276459ea71aa46e2967ea283fc88ee5631c11f29a06787e16cf4aece1b8/statsmodels-0.14.5-cp311-cp311-win_amd64.whl", hash = "sha256:ec56f771d9529cdc17ed2fb2a950d100b6e83a7c5372aae8ac5bb065c474b856", size = 9640980, upload-time = "2025-07-07T12:05:33.085Z" }, - { url = "https://files.pythonhosted.org/packages/5f/a5/fcc4f5f16355660ce7a1742e28a43e3a9391b492fc4ff29fdd6893e81c05/statsmodels-0.14.5-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:37e7364a39f9aa3b51d15a208c2868b90aadb8412f868530f5cba9197cb00eaa", size = 10042891, upload-time = "2025-07-07T12:13:41.671Z" }, - { url = "https://files.pythonhosted.org/packages/1c/6f/db0cf5efa48277ac6218d9b981c8fd5e63c4c43e0d9d65015fdc38eed0ef/statsmodels-0.14.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:4263d7f4d0f1d5ac6eb4db22e1ee34264a14d634b9332c975c9d9109b6b46e12", size = 9698912, upload-time = "2025-07-07T12:07:54.674Z" }, - { url = "https://files.pythonhosted.org/packages/4a/93/4ddc3bc4a59c51e6a57c49df1b889882c40d9e141e855b3517f6a8de3232/statsmodels-0.14.5-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:86224f6e36f38486e471e75759d241fe2912d8bc25ab157d54ee074c6aedbf45", size = 10237801, upload-time = "2025-07-07T14:23:12.593Z" }, - { url = "https://files.pythonhosted.org/packages/66/de/dc6bf2f6e8c8eb4c5815560ebdbdf2d69a767bc0f65fde34bc086cf5b36d/statsmodels-0.14.5-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c3dd760a6fa80cd5e0371685c697bb9c2c0e6e1f394d975e596a1e6d0bbb9372", size = 10424154, upload-time = "2025-07-07T14:23:25.365Z" }, - { url = "https://files.pythonhosted.org/packages/16/4f/2d5a8d14bebdf2b03b3ea89b8c6a2c837bb406ba5b7a41add8bd303bce29/statsmodels-0.14.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:6264fb00e02f858b86bd01ef2dc05055a71d4a0cc7551b9976b07b0f0e6cf24f", size = 10652915, upload-time = "2025-07-07T14:23:39.337Z" }, - { url = "https://files.pythonhosted.org/packages/df/4c/2feda3a9f0e17444a84ba5398ada6a4d2e1b8f832760048f04e2b8ea0c41/statsmodels-0.14.5-cp312-cp312-win_amd64.whl", hash = "sha256:b2ed065bfbaf8bb214c7201656df840457c2c8c65e1689e3eb09dc7440f9c61c", size = 9611236, upload-time = "2025-07-07T12:08:06.794Z" }, - { url = "https://files.pythonhosted.org/packages/84/fd/4c374108cf108b3130240a5b45847a61f70ddf973429044a81a05189b046/statsmodels-0.14.5-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:906263134dd1a640e55ecb01fda4a9be7b9e08558dba9e4c4943a486fdb0c9c8", size = 10013958, upload-time = "2025-07-07T14:35:01.04Z" }, - { url = "https://files.pythonhosted.org/packages/5a/36/bf3d7f0e36acd3ba9ec0babd79ace25506b6872780cbd710fb7cd31f0fa2/statsmodels-0.14.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:9118f76344f77cffbb3a9cbcff8682b325be5eed54a4b3253e09da77a74263d3", size = 9674243, upload-time = "2025-07-07T12:08:22.571Z" }, - { url = "https://files.pythonhosted.org/packages/90/ce/a55a6f37b5277683ceccd965a5828b24672bbc427db6b3969ae0b0fc29fb/statsmodels-0.14.5-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:9dc4ee159070557c9a6c000625d85f653de437772fe7086857cff68f501afe45", size = 10219521, upload-time = "2025-07-07T14:23:52.646Z" }, - { url = "https://files.pythonhosted.org/packages/1e/48/973da1ee8bc0743519759e74c3615b39acdc3faf00e0a0710f8c856d8c9d/statsmodels-0.14.5-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5a085d47c8ef5387279a991633883d0e700de2b0acc812d7032d165888627bef", size = 10453538, upload-time = "2025-07-07T14:24:06.959Z" }, - { url = "https://files.pythonhosted.org/packages/c7/d6/18903fb707afd31cf1edaec5201964dbdacb2bfae9a22558274647a7c88f/statsmodels-0.14.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9f866b2ebb2904b47c342d00def83c526ef2eb1df6a9a3c94ba5fe63d0005aec", size = 10681584, upload-time = "2025-07-07T14:24:21.038Z" }, - { url = "https://files.pythonhosted.org/packages/44/d6/80df1bbbfcdc50bff4152f43274420fa9856d56e234d160d6206eb1f5827/statsmodels-0.14.5-cp313-cp313-win_amd64.whl", hash = "sha256:2a06bca03b7a492f88c8106103ab75f1a5ced25de90103a89f3a287518017939", size = 9604641, upload-time = "2025-07-07T12:08:36.23Z" }, + { name = "scipy" }, +] +sdist = { url = "https://files.pythonhosted.org/packages/1f/3b/963a015dd8ea17e10c7b0e2f14d7c4daec903baf60a017e756b57953a4bf/statsmodels-0.14.4.tar.gz", hash = "sha256:5d69e0f39060dc72c067f9bb6e8033b6dccdb0bae101d76a7ef0bcc94e898b67", size = 20354802, upload-time = "2024-10-03T16:15:36.273Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/af/2c/23bf5ad9e8a77c0c8d9750512bff89e32154dea91998114118e0e147ae67/statsmodels-0.14.4-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:7a62f1fc9086e4b7ee789a6f66b3c0fc82dd8de1edda1522d30901a0aa45e42b", size = 10216574, upload-time = "2024-10-03T16:13:31.472Z" }, + { url = "https://files.pythonhosted.org/packages/ba/a5/2f09ab918296e534ea5d132e90efac51ae12ff15992d77539bbfca1158fa/statsmodels-0.14.4-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:46ac7ddefac0c9b7b607eed1d47d11e26fe92a1bc1f4d9af48aeed4e21e87981", size = 9912430, upload-time = "2024-10-03T16:13:44.683Z" }, + { url = "https://files.pythonhosted.org/packages/93/6a/b86f8c9b799dc93e5b4a3267eb809843e6328e34248a53496b96f50d732e/statsmodels-0.14.4-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2a337b731aa365d09bb0eab6da81446c04fde6c31976b1d8e3d3a911f0f1e07b", size = 10444673, upload-time = "2024-10-03T17:09:04.647Z" }, + { url = "https://files.pythonhosted.org/packages/78/44/d72c634211797ed07dd8c63ced4ae11debd7a40b24ee80e79346a526194f/statsmodels-0.14.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:631bb52159117c5da42ba94bd94859276b68cab25dc4cac86475bc24671143bc", size = 10811248, upload-time = "2024-10-03T17:09:20.337Z" }, + { url = "https://files.pythonhosted.org/packages/35/64/df81426924fcc48a0402534efa96cde13275629ae52f123189d16c4b75ff/statsmodels-0.14.4-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:3bb2e580d382545a65f298589809af29daeb15f9da2eb252af8f79693e618abc", size = 10946447, upload-time = "2024-10-03T17:09:35.135Z" }, + { url = "https://files.pythonhosted.org/packages/5c/f9/205130cceeda0eebd5a1a58c04e060c2f87a1d63cbbe37a9caa0fcb50c68/statsmodels-0.14.4-cp310-cp310-win_amd64.whl", hash = "sha256:9729642884147ee9db67b5a06a355890663d21f76ed608a56ac2ad98b94d201a", size = 9845796, upload-time = "2024-10-03T16:13:58.307Z" }, + { url = "https://files.pythonhosted.org/packages/48/88/326f5f689e69d9c47a68a22ffdd20a6ea6410b53918f9a8e63380dfc181c/statsmodels-0.14.4-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:5ed7e118e6e3e02d6723a079b8c97eaadeed943fa1f7f619f7148dfc7862670f", size = 10221032, upload-time = "2024-10-03T16:22:48.191Z" }, + { url = "https://files.pythonhosted.org/packages/07/0b/9a0818be42f6689ebdc7a2277ea984d6299f0809d0e0277128df4f7dc606/statsmodels-0.14.4-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:f5f537f7d000de4a1708c63400755152b862cd4926bb81a86568e347c19c364b", size = 9912219, upload-time = "2024-10-03T17:17:03.799Z" }, + { url = "https://files.pythonhosted.org/packages/b1/f2/91c70a3b4a3e416f76ead61b04c87bc60080d634d7fa2ab893976bdd86fa/statsmodels-0.14.4-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:aa74aaa26eaa5012b0a01deeaa8a777595d0835d3d6c7175f2ac65435a7324d2", size = 10424053, upload-time = "2024-10-03T17:09:49.325Z" }, + { url = "https://files.pythonhosted.org/packages/9d/4f/a96e682f82b675e4a6f3de8ad990587d8b1fde500a630a2aabcaabee11d8/statsmodels-0.14.4-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e332c2d9b806083d1797231280602340c5c913f90d4caa0213a6a54679ce9331", size = 10752529, upload-time = "2024-10-03T17:10:03.489Z" }, + { url = "https://files.pythonhosted.org/packages/4b/c6/47549345d32da1530a819a3699f6f34f9f70733a245eeb29f5e05e53f362/statsmodels-0.14.4-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:d9c8fa28dfd75753d9cf62769ba1fecd7e73a0be187f35cc6f54076f98aa3f3f", size = 10959003, upload-time = "2024-10-03T17:10:17.477Z" }, + { url = "https://files.pythonhosted.org/packages/4b/e4/f9e96896278308e17dfd4f60a84826c48117674c980234ee38f59ab28a12/statsmodels-0.14.4-cp311-cp311-win_amd64.whl", hash = "sha256:a6087ecb0714f7c59eb24c22781491e6f1cfffb660b4740e167625ca4f052056", size = 9853281, upload-time = "2024-10-03T16:14:11.019Z" }, + { url = "https://files.pythonhosted.org/packages/f5/99/654fd41a9024643ee70b239e5ebc987bf98ce9fc2693bd550bee58136564/statsmodels-0.14.4-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:5221dba7424cf4f2561b22e9081de85f5bb871228581124a0d1b572708545199", size = 10220508, upload-time = "2024-10-03T17:10:31.183Z" }, + { url = "https://files.pythonhosted.org/packages/67/d8/ac30cf4cf97adaa48548be57e7cf02e894f31b45fd55bf9213358d9781c9/statsmodels-0.14.4-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:17672b30c6b98afe2b095591e32d1d66d4372f2651428e433f16a3667f19eabb", size = 9912317, upload-time = "2024-10-03T16:22:29.504Z" }, + { url = "https://files.pythonhosted.org/packages/e0/77/2440d551eaf27f9c1d3650e13b3821a35ad5b21d3a19f62fb302af9203e8/statsmodels-0.14.4-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ab5e6312213b8cfb9dca93dd46a0f4dccb856541f91d3306227c3d92f7659245", size = 10301662, upload-time = "2024-10-03T17:13:04.537Z" }, + { url = "https://files.pythonhosted.org/packages/fa/e1/60a652f18996a40a7410aeb7eb476c18da8a39792c7effe67f06883e9852/statsmodels-0.14.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:4bbb150620b53133d6cd1c5d14c28a4f85701e6c781d9b689b53681effaa655f", size = 10741763, upload-time = "2024-10-03T17:13:17.594Z" }, + { url = "https://files.pythonhosted.org/packages/81/0c/2453eec3ac25e300847d9ed97f41156de145e507391ecb5ac989e111e525/statsmodels-0.14.4-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:bb695c2025d122a101c2aca66d2b78813c321b60d3a7c86bb8ec4467bb53b0f9", size = 10879534, upload-time = "2024-10-03T17:13:31.19Z" }, + { url = "https://files.pythonhosted.org/packages/59/9a/e466a1b887a1441141e52dbcc98152f013d85076576da6eed2357f2016ae/statsmodels-0.14.4-cp312-cp312-win_amd64.whl", hash = "sha256:7f7917a51766b4e074da283c507a25048ad29a18e527207883d73535e0dc6184", size = 9823866, upload-time = "2024-10-03T16:14:23.828Z" }, + { url = "https://files.pythonhosted.org/packages/31/f8/2662e6a101315ad336f75168fa9bac71f913ebcb92a6be84031d84a0f21f/statsmodels-0.14.4-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b5a24f5d2c22852d807d2b42daf3a61740820b28d8381daaf59dcb7055bf1a79", size = 10186886, upload-time = "2024-10-03T17:10:44.074Z" }, + { url = "https://files.pythonhosted.org/packages/fa/c0/ee6e8ed35fc1ca9c7538c592f4974547bf72274bc98db1ae4a6e87481a83/statsmodels-0.14.4-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:df4f7864606fa843d7e7c0e6af288f034a2160dba14e6ccc09020a3cf67cb092", size = 9880066, upload-time = "2024-10-03T17:10:56.972Z" }, + { url = "https://files.pythonhosted.org/packages/d1/97/3380ca6d8fd66cfb3d12941e472642f26e781a311c355a4e97aab2ed0216/statsmodels-0.14.4-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:91341cbde9e8bea5fb419a76e09114e221567d03f34ca26e6d67ae2c27d8fe3c", size = 10283521, upload-time = "2024-10-03T17:14:06.216Z" }, + { url = "https://files.pythonhosted.org/packages/fe/2a/55c5b5c5e5124a202ea3fe0bcdbdeceaf91b4ec6164b8434acb9dd97409c/statsmodels-0.14.4-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1322286a7bfdde2790bf72d29698a1b76c20b8423a55bdcd0d457969d0041f72", size = 10723228, upload-time = "2024-10-03T17:14:19.587Z" }, + { url = "https://files.pythonhosted.org/packages/4f/76/67747e49dc758daae06f33aad8247b718cd7d224f091d2cd552681215bb2/statsmodels-0.14.4-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:e31b95ac603415887c9f0d344cb523889cf779bc52d68e27e2d23c358958fec7", size = 10859503, upload-time = "2024-10-03T17:14:32.798Z" }, + { url = "https://files.pythonhosted.org/packages/1d/eb/cb8b01f5edf8f135eb3d0553d159db113a35b2948d0e51eeb735e7ae09ea/statsmodels-0.14.4-cp313-cp313-win_amd64.whl", hash = "sha256:81030108d27aecc7995cac05aa280cf8c6025f6a6119894eef648997936c2dd0", size = 9817574, upload-time = "2024-10-03T16:14:37.461Z" }, ] [[package]] @@ -3800,14 +3535,13 @@ wheels = [ [[package]] name = "torch" -version = "2.7.1" +version = "2.8.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "filelock" }, { name = "fsspec" }, { name = "jinja2" }, - { name = "networkx", version = "3.4.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "networkx", version = "3.5", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "networkx" }, { name = "nvidia-cublas-cu12", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" }, { name = "nvidia-cuda-cupti-cu12", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" }, { name = "nvidia-cuda-nvrtc-cu12", marker = "platform_machine == 'x86_64' and sys_platform == 'linux'" }, @@ -3828,26 +3562,26 @@ dependencies = [ { name = "typing-extensions" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/6a/27/2e06cb52adf89fe6e020963529d17ed51532fc73c1e6d1b18420ef03338c/torch-2.7.1-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:a103b5d782af5bd119b81dbcc7ffc6fa09904c423ff8db397a1e6ea8fd71508f", size = 99089441, upload-time = "2025-06-04T17:38:48.268Z" }, - { url = "https://files.pythonhosted.org/packages/0a/7c/0a5b3aee977596459ec45be2220370fde8e017f651fecc40522fd478cb1e/torch-2.7.1-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:fe955951bdf32d182ee8ead6c3186ad54781492bf03d547d31771a01b3d6fb7d", size = 821154516, upload-time = "2025-06-04T17:36:28.556Z" }, - { url = "https://files.pythonhosted.org/packages/f9/91/3d709cfc5e15995fb3fe7a6b564ce42280d3a55676dad672205e94f34ac9/torch-2.7.1-cp310-cp310-win_amd64.whl", hash = "sha256:885453d6fba67d9991132143bf7fa06b79b24352f4506fd4d10b309f53454162", size = 216093147, upload-time = "2025-06-04T17:39:38.132Z" }, - { url = "https://files.pythonhosted.org/packages/92/f6/5da3918414e07da9866ecb9330fe6ffdebe15cb9a4c5ada7d4b6e0a6654d/torch-2.7.1-cp310-none-macosx_11_0_arm64.whl", hash = "sha256:d72acfdb86cee2a32c0ce0101606f3758f0d8bb5f8f31e7920dc2809e963aa7c", size = 68630914, upload-time = "2025-06-04T17:39:31.162Z" }, - { url = "https://files.pythonhosted.org/packages/11/56/2eae3494e3d375533034a8e8cf0ba163363e996d85f0629441fa9d9843fe/torch-2.7.1-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:236f501f2e383f1cb861337bdf057712182f910f10aeaf509065d54d339e49b2", size = 99093039, upload-time = "2025-06-04T17:39:06.963Z" }, - { url = "https://files.pythonhosted.org/packages/e5/94/34b80bd172d0072c9979708ccd279c2da2f55c3ef318eceec276ab9544a4/torch-2.7.1-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:06eea61f859436622e78dd0cdd51dbc8f8c6d76917a9cf0555a333f9eac31ec1", size = 821174704, upload-time = "2025-06-04T17:37:03.799Z" }, - { url = "https://files.pythonhosted.org/packages/50/9e/acf04ff375b0b49a45511c55d188bcea5c942da2aaf293096676110086d1/torch-2.7.1-cp311-cp311-win_amd64.whl", hash = "sha256:8273145a2e0a3c6f9fd2ac36762d6ee89c26d430e612b95a99885df083b04e52", size = 216095937, upload-time = "2025-06-04T17:39:24.83Z" }, - { url = "https://files.pythonhosted.org/packages/5b/2b/d36d57c66ff031f93b4fa432e86802f84991477e522adcdffd314454326b/torch-2.7.1-cp311-none-macosx_11_0_arm64.whl", hash = "sha256:aea4fc1bf433d12843eb2c6b2204861f43d8364597697074c8d38ae2507f8730", size = 68640034, upload-time = "2025-06-04T17:39:17.989Z" }, - { url = "https://files.pythonhosted.org/packages/87/93/fb505a5022a2e908d81fe9a5e0aa84c86c0d5f408173be71c6018836f34e/torch-2.7.1-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:27ea1e518df4c9de73af7e8a720770f3628e7f667280bce2be7a16292697e3fa", size = 98948276, upload-time = "2025-06-04T17:39:12.852Z" }, - { url = "https://files.pythonhosted.org/packages/56/7e/67c3fe2b8c33f40af06326a3d6ae7776b3e3a01daa8f71d125d78594d874/torch-2.7.1-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:c33360cfc2edd976c2633b3b66c769bdcbbf0e0b6550606d188431c81e7dd1fc", size = 821025792, upload-time = "2025-06-04T17:34:58.747Z" }, - { url = "https://files.pythonhosted.org/packages/a1/37/a37495502bc7a23bf34f89584fa5a78e25bae7b8da513bc1b8f97afb7009/torch-2.7.1-cp312-cp312-win_amd64.whl", hash = "sha256:d8bf6e1856ddd1807e79dc57e54d3335f2b62e6f316ed13ed3ecfe1fc1df3d8b", size = 216050349, upload-time = "2025-06-04T17:38:59.709Z" }, - { url = "https://files.pythonhosted.org/packages/3a/60/04b77281c730bb13460628e518c52721257814ac6c298acd25757f6a175c/torch-2.7.1-cp312-none-macosx_11_0_arm64.whl", hash = "sha256:787687087412c4bd68d315e39bc1223f08aae1d16a9e9771d95eabbb04ae98fb", size = 68645146, upload-time = "2025-06-04T17:38:52.97Z" }, - { url = "https://files.pythonhosted.org/packages/66/81/e48c9edb655ee8eb8c2a6026abdb6f8d2146abd1f150979ede807bb75dcb/torch-2.7.1-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:03563603d931e70722dce0e11999d53aa80a375a3d78e6b39b9f6805ea0a8d28", size = 98946649, upload-time = "2025-06-04T17:38:43.031Z" }, - { url = "https://files.pythonhosted.org/packages/3a/24/efe2f520d75274fc06b695c616415a1e8a1021d87a13c68ff9dce733d088/torch-2.7.1-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:d632f5417b6980f61404a125b999ca6ebd0b8b4bbdbb5fbbba44374ab619a412", size = 821033192, upload-time = "2025-06-04T17:38:09.146Z" }, - { url = "https://files.pythonhosted.org/packages/dd/d9/9c24d230333ff4e9b6807274f6f8d52a864210b52ec794c5def7925f4495/torch-2.7.1-cp313-cp313-win_amd64.whl", hash = "sha256:23660443e13995ee93e3d844786701ea4ca69f337027b05182f5ba053ce43b38", size = 216055668, upload-time = "2025-06-04T17:38:36.253Z" }, - { url = "https://files.pythonhosted.org/packages/95/bf/e086ee36ddcef9299f6e708d3b6c8487c1651787bb9ee2939eb2a7f74911/torch-2.7.1-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:0da4f4dba9f65d0d203794e619fe7ca3247a55ffdcbd17ae8fb83c8b2dc9b585", size = 68925988, upload-time = "2025-06-04T17:38:29.273Z" }, - { url = "https://files.pythonhosted.org/packages/69/6a/67090dcfe1cf9048448b31555af6efb149f7afa0a310a366adbdada32105/torch-2.7.1-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:e08d7e6f21a617fe38eeb46dd2213ded43f27c072e9165dc27300c9ef9570934", size = 99028857, upload-time = "2025-06-04T17:37:50.956Z" }, - { url = "https://files.pythonhosted.org/packages/90/1c/48b988870823d1cc381f15ec4e70ed3d65e043f43f919329b0045ae83529/torch-2.7.1-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:30207f672328a42df4f2174b8f426f354b2baa0b7cca3a0adb3d6ab5daf00dc8", size = 821098066, upload-time = "2025-06-04T17:37:33.939Z" }, - { url = "https://files.pythonhosted.org/packages/7b/eb/10050d61c9d5140c5dc04a89ed3257ef1a6b93e49dd91b95363d757071e0/torch-2.7.1-cp313-cp313t-win_amd64.whl", hash = "sha256:79042feca1c634aaf6603fe6feea8c6b30dfa140a6bbc0b973e2260c7e79a22e", size = 216336310, upload-time = "2025-06-04T17:36:09.862Z" }, - { url = "https://files.pythonhosted.org/packages/b1/29/beb45cdf5c4fc3ebe282bf5eafc8dfd925ead7299b3c97491900fe5ed844/torch-2.7.1-cp313-none-macosx_11_0_arm64.whl", hash = "sha256:988b0cbc4333618a1056d2ebad9eb10089637b659eb645434d0809d8d937b946", size = 68645708, upload-time = "2025-06-04T17:34:39.852Z" }, + { url = "https://files.pythonhosted.org/packages/63/28/110f7274254f1b8476c561dada127173f994afa2b1ffc044efb773c15650/torch-2.8.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:0be92c08b44009d4131d1ff7a8060d10bafdb7ddcb7359ef8d8c5169007ea905", size = 102052793, upload-time = "2025-08-06T14:53:15.852Z" }, + { url = "https://files.pythonhosted.org/packages/70/1c/58da560016f81c339ae14ab16c98153d51c941544ae568da3cb5b1ceb572/torch-2.8.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:89aa9ee820bb39d4d72b794345cccef106b574508dd17dbec457949678c76011", size = 888025420, upload-time = "2025-08-06T14:54:18.014Z" }, + { url = "https://files.pythonhosted.org/packages/70/87/f69752d0dd4ba8218c390f0438130c166fa264a33b7025adb5014b92192c/torch-2.8.0-cp310-cp310-win_amd64.whl", hash = "sha256:e8e5bf982e87e2b59d932769938b698858c64cc53753894be25629bdf5cf2f46", size = 241363614, upload-time = "2025-08-06T14:53:31.496Z" }, + { url = "https://files.pythonhosted.org/packages/ef/d6/e6d4c57e61c2b2175d3aafbfb779926a2cfd7c32eeda7c543925dceec923/torch-2.8.0-cp310-none-macosx_11_0_arm64.whl", hash = "sha256:a3f16a58a9a800f589b26d47ee15aca3acf065546137fc2af039876135f4c760", size = 73611154, upload-time = "2025-08-06T14:53:10.919Z" }, + { url = "https://files.pythonhosted.org/packages/8f/c4/3e7a3887eba14e815e614db70b3b529112d1513d9dae6f4d43e373360b7f/torch-2.8.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:220a06fd7af8b653c35d359dfe1aaf32f65aa85befa342629f716acb134b9710", size = 102073391, upload-time = "2025-08-06T14:53:20.937Z" }, + { url = "https://files.pythonhosted.org/packages/5a/63/4fdc45a0304536e75a5e1b1bbfb1b56dd0e2743c48ee83ca729f7ce44162/torch-2.8.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:c12fa219f51a933d5f80eeb3a7a5d0cbe9168c0a14bbb4055f1979431660879b", size = 888063640, upload-time = "2025-08-06T14:55:05.325Z" }, + { url = "https://files.pythonhosted.org/packages/84/57/2f64161769610cf6b1c5ed782bd8a780e18a3c9d48931319f2887fa9d0b1/torch-2.8.0-cp311-cp311-win_amd64.whl", hash = "sha256:8c7ef765e27551b2fbfc0f41bcf270e1292d9bf79f8e0724848b1682be6e80aa", size = 241366752, upload-time = "2025-08-06T14:53:38.692Z" }, + { url = "https://files.pythonhosted.org/packages/a4/5e/05a5c46085d9b97e928f3f037081d3d2b87fb4b4195030fc099aaec5effc/torch-2.8.0-cp311-none-macosx_11_0_arm64.whl", hash = "sha256:5ae0524688fb6707c57a530c2325e13bb0090b745ba7b4a2cd6a3ce262572916", size = 73621174, upload-time = "2025-08-06T14:53:25.44Z" }, + { url = "https://files.pythonhosted.org/packages/49/0c/2fd4df0d83a495bb5e54dca4474c4ec5f9c62db185421563deeb5dabf609/torch-2.8.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:e2fab4153768d433f8ed9279c8133a114a034a61e77a3a104dcdf54388838705", size = 101906089, upload-time = "2025-08-06T14:53:52.631Z" }, + { url = "https://files.pythonhosted.org/packages/99/a8/6acf48d48838fb8fe480597d98a0668c2beb02ee4755cc136de92a0a956f/torch-2.8.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:b2aca0939fb7e4d842561febbd4ffda67a8e958ff725c1c27e244e85e982173c", size = 887913624, upload-time = "2025-08-06T14:56:44.33Z" }, + { url = "https://files.pythonhosted.org/packages/af/8a/5c87f08e3abd825c7dfecef5a0f1d9aa5df5dd0e3fd1fa2f490a8e512402/torch-2.8.0-cp312-cp312-win_amd64.whl", hash = "sha256:2f4ac52f0130275d7517b03a33d2493bab3693c83dcfadf4f81688ea82147d2e", size = 241326087, upload-time = "2025-08-06T14:53:46.503Z" }, + { url = "https://files.pythonhosted.org/packages/be/66/5c9a321b325aaecb92d4d1855421e3a055abd77903b7dab6575ca07796db/torch-2.8.0-cp312-none-macosx_11_0_arm64.whl", hash = "sha256:619c2869db3ada2c0105487ba21b5008defcc472d23f8b80ed91ac4a380283b0", size = 73630478, upload-time = "2025-08-06T14:53:57.144Z" }, + { url = "https://files.pythonhosted.org/packages/10/4e/469ced5a0603245d6a19a556e9053300033f9c5baccf43a3d25ba73e189e/torch-2.8.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:2b2f96814e0345f5a5aed9bf9734efa913678ed19caf6dc2cddb7930672d6128", size = 101936856, upload-time = "2025-08-06T14:54:01.526Z" }, + { url = "https://files.pythonhosted.org/packages/16/82/3948e54c01b2109238357c6f86242e6ecbf0c63a1af46906772902f82057/torch-2.8.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:65616ca8ec6f43245e1f5f296603e33923f4c30f93d65e103d9e50c25b35150b", size = 887922844, upload-time = "2025-08-06T14:55:50.78Z" }, + { url = "https://files.pythonhosted.org/packages/e3/54/941ea0a860f2717d86a811adf0c2cd01b3983bdd460d0803053c4e0b8649/torch-2.8.0-cp313-cp313-win_amd64.whl", hash = "sha256:659df54119ae03e83a800addc125856effda88b016dfc54d9f65215c3975be16", size = 241330968, upload-time = "2025-08-06T14:54:45.293Z" }, + { url = "https://files.pythonhosted.org/packages/de/69/8b7b13bba430f5e21d77708b616f767683629fc4f8037564a177d20f90ed/torch-2.8.0-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:1a62a1ec4b0498930e2543535cf70b1bef8c777713de7ceb84cd79115f553767", size = 73915128, upload-time = "2025-08-06T14:54:34.769Z" }, + { url = "https://files.pythonhosted.org/packages/15/0e/8a800e093b7f7430dbaefa80075aee9158ec22e4c4fc3c1a66e4fb96cb4f/torch-2.8.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:83c13411a26fac3d101fe8035a6b0476ae606deb8688e904e796a3534c197def", size = 102020139, upload-time = "2025-08-06T14:54:39.047Z" }, + { url = "https://files.pythonhosted.org/packages/4a/15/5e488ca0bc6162c86a33b58642bc577c84ded17c7b72d97e49b5833e2d73/torch-2.8.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:8f0a9d617a66509ded240add3754e462430a6c1fc5589f86c17b433dd808f97a", size = 887990692, upload-time = "2025-08-06T14:56:18.286Z" }, + { url = "https://files.pythonhosted.org/packages/b4/a8/6a04e4b54472fc5dba7ca2341ab219e529f3c07b6941059fbf18dccac31f/torch-2.8.0-cp313-cp313t-win_amd64.whl", hash = "sha256:a7242b86f42be98ac674b88a4988643b9bc6145437ec8f048fea23f72feb5eca", size = 241603453, upload-time = "2025-08-06T14:55:22.945Z" }, + { url = "https://files.pythonhosted.org/packages/04/6e/650bb7f28f771af0cb791b02348db8b7f5f64f40f6829ee82aa6ce99aabe/torch-2.8.0-cp313-none-macosx_11_0_arm64.whl", hash = "sha256:7b677e17f5a3e69fdef7eb3b9da72622f8d322692930297e4ccb52fefc6c8211", size = 73632395, upload-time = "2025-08-06T14:55:28.645Z" }, ] [[package]] @@ -3858,8 +3592,7 @@ dependencies = [ { name = "aiohttp" }, { name = "fsspec" }, { name = "jinja2" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "psutil" }, { name = "pyparsing" }, { name = "requests" }, @@ -3872,70 +3605,67 @@ wheels = [ [[package]] name = "torchmetrics" -version = "1.8.0" +version = "1.7.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "lightning-utilities" }, - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "packaging" }, { name = "torch" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/b0/1d/01fdf2595ac344dcfb2d837e7596c71cd8659e99f0b1fd72a93c76ea2c05/torchmetrics-1.8.0.tar.gz", hash = "sha256:8b4d011963a602109fb8255018c2386391e8c4c7f48a09669fbf7bb7889fda8c", size = 578941, upload-time = "2025-07-23T17:39:11.719Z" } +sdist = { url = "https://files.pythonhosted.org/packages/47/99/3a28bc2be4f562b9aae5b3557ec4c6cbcfcdd0a7d2a6fcb1ab44e7c55334/torchmetrics-1.7.0.tar.gz", hash = "sha256:7a26d5cb73a6ae51ab5cb514aa4dc0543af7287a507719986a06e15df12ea68b", size = 564173, upload-time = "2025-03-20T19:12:01.289Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/0b/cc/f41157b446d555bf1446915960f92928dc5e8393fc09b6de6189aef2d9dd/torchmetrics-1.8.0-py3-none-any.whl", hash = "sha256:009832f0df5be9aca72f51a1e6c6a555a131ba53c1ce46a38348a202250c22df", size = 981886, upload-time = "2025-07-23T17:39:09.246Z" }, + { url = "https://files.pythonhosted.org/packages/71/1d/dbbb54743865bad58a9adc7ac2c9bedd28bc76253cb8ec4b69765ccc8190/torchmetrics-1.7.0-py3-none-any.whl", hash = "sha256:39a72cf40c8452e041f5315b997ef811c2baaae01478131cf6ed9813f553a668", size = 960916, upload-time = "2025-03-20T19:11:58.841Z" }, ] [[package]] name = "torchvision" -version = "0.22.1" +version = "0.23.0" source = { registry = "https://pypi.org/simple" } dependencies = [ - { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" }, - { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" }, + { name = "numpy" }, { name = "pillow" }, { name = "torch" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/15/2c/7b67117b14c6cc84ae3126ca6981abfa3af2ac54eb5252b80d9475fb40df/torchvision-0.22.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3b47d8369ee568c067795c0da0b4078f39a9dfea6f3bc1f3ac87530dfda1dd56", size = 1947825, upload-time = "2025-06-04T17:43:15.523Z" }, - { url = "https://files.pythonhosted.org/packages/6c/9f/c4dcf1d232b75e28bc37e21209ab2458d6d60235e16163544ed693de54cb/torchvision-0.22.1-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:990de4d657a41ed71680cd8be2e98ebcab55371f30993dc9bd2e676441f7180e", size = 2512611, upload-time = "2025-06-04T17:43:03.951Z" }, - { url = "https://files.pythonhosted.org/packages/e2/99/db71d62d12628111d59147095527a0ab492bdfecfba718d174c04ae6c505/torchvision-0.22.1-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:3347f690c2eed6d02aa0edfb9b01d321e7f7cf1051992d96d8d196c39b881d49", size = 7485668, upload-time = "2025-06-04T17:43:09.453Z" }, - { url = "https://files.pythonhosted.org/packages/32/ff/4a93a4623c3e5f97e8552af0f9f81d289dcf7f2ac71f1493f1c93a6b973d/torchvision-0.22.1-cp310-cp310-win_amd64.whl", hash = "sha256:86ad938f5a6ca645f0d5fb19484b1762492c2188c0ffb05c602e9e9945b7b371", size = 1707961, upload-time = "2025-06-04T17:43:13.038Z" }, - { url = "https://files.pythonhosted.org/packages/f6/00/bdab236ef19da050290abc2b5203ff9945c84a1f2c7aab73e8e9c8c85669/torchvision-0.22.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:4addf626e2b57fc22fd6d329cf1346d474497672e6af8383b7b5b636fba94a53", size = 1947827, upload-time = "2025-06-04T17:43:10.84Z" }, - { url = "https://files.pythonhosted.org/packages/ac/d0/18f951b2be3cfe48c0027b349dcc6fde950e3dc95dd83e037e86f284f6fd/torchvision-0.22.1-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:8b4a53a6067d63adba0c52f2b8dd2290db649d642021674ee43c0c922f0c6a69", size = 2514021, upload-time = "2025-06-04T17:43:07.608Z" }, - { url = "https://files.pythonhosted.org/packages/c3/1a/63eb241598b36d37a0221e10af357da34bd33402ccf5c0765e389642218a/torchvision-0.22.1-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:b7866a3b326413e67724ac46f1ee594996735e10521ba9e6cdbe0fa3cd98c2f2", size = 7487300, upload-time = "2025-06-04T17:42:58.349Z" }, - { url = "https://files.pythonhosted.org/packages/e5/73/1b009b42fe4a7774ba19c23c26bb0f020d68525c417a348b166f1c56044f/torchvision-0.22.1-cp311-cp311-win_amd64.whl", hash = "sha256:bb3f6df6f8fd415ce38ec4fd338376ad40c62e86052d7fc706a0dd51efac1718", size = 1707989, upload-time = "2025-06-04T17:43:14.332Z" }, - { url = "https://files.pythonhosted.org/packages/02/90/f4e99a5112dc221cf68a485e853cc3d9f3f1787cb950b895f3ea26d1ea98/torchvision-0.22.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:153f1790e505bd6da123e21eee6e83e2e155df05c0fe7d56347303067d8543c5", size = 1947827, upload-time = "2025-06-04T17:43:11.945Z" }, - { url = "https://files.pythonhosted.org/packages/25/f6/53e65384cdbbe732cc2106bb04f7fb908487e4fb02ae4a1613ce6904a122/torchvision-0.22.1-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:964414eef19459d55a10e886e2fca50677550e243586d1678f65e3f6f6bac47a", size = 2514576, upload-time = "2025-06-04T17:43:02.707Z" }, - { url = "https://files.pythonhosted.org/packages/17/8b/155f99042f9319bd7759536779b2a5b67cbd4f89c380854670850f89a2f4/torchvision-0.22.1-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:699c2d70d33951187f6ed910ea05720b9b4aaac1dcc1135f53162ce7d42481d3", size = 7485962, upload-time = "2025-06-04T17:42:43.606Z" }, - { url = "https://files.pythonhosted.org/packages/05/17/e45d5cd3627efdb47587a0634179a3533593436219de3f20c743672d2a79/torchvision-0.22.1-cp312-cp312-win_amd64.whl", hash = "sha256:75e0897da7a8e43d78632f66f2bdc4f6e26da8d3f021a7c0fa83746073c2597b", size = 1707992, upload-time = "2025-06-04T17:42:53.207Z" }, - { url = "https://files.pythonhosted.org/packages/7a/30/fecdd09fb973e963da68207fe9f3d03ec6f39a935516dc2a98397bf495c6/torchvision-0.22.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:9c3ae3319624c43cc8127020f46c14aa878406781f0899bb6283ae474afeafbf", size = 1947818, upload-time = "2025-06-04T17:42:51.954Z" }, - { url = "https://files.pythonhosted.org/packages/55/f4/b45f6cd92fa0acfac5e31b8e9258232f25bcdb0709a604e8b8a39d76e411/torchvision-0.22.1-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:4a614a6a408d2ed74208d0ea6c28a2fbb68290e9a7df206c5fef3f0b6865d307", size = 2471597, upload-time = "2025-06-04T17:42:48.838Z" }, - { url = "https://files.pythonhosted.org/packages/8d/b0/3cffd6a285b5ffee3fe4a31caff49e350c98c5963854474d1c4f7a51dea5/torchvision-0.22.1-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:7ee682be589bb1a002b7704f06b8ec0b89e4b9068f48e79307d2c6e937a9fdf4", size = 7485894, upload-time = "2025-06-04T17:43:01.371Z" }, - { url = "https://files.pythonhosted.org/packages/fd/1d/0ede596fedc2080d18108149921278b59f220fbb398f29619495337b0f86/torchvision-0.22.1-cp313-cp313-win_amd64.whl", hash = "sha256:2566cafcfa47ecfdbeed04bab8cef1307c8d4ef75046f7624b9e55f384880dfe", size = 1708020, upload-time = "2025-06-04T17:43:06.085Z" }, - { url = "https://files.pythonhosted.org/packages/0f/ca/e9a06bd61ee8e04fb4962a3fb524fe6ee4051662db07840b702a9f339b24/torchvision-0.22.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:043d9e35ed69c2e586aff6eb9e2887382e7863707115668ac9d140da58f42cba", size = 2137623, upload-time = "2025-06-04T17:43:05.028Z" }, - { url = "https://files.pythonhosted.org/packages/ab/c8/2ebe90f18e7ffa2120f5c3eab62aa86923185f78d2d051a455ea91461608/torchvision-0.22.1-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:27142bcc8a984227a6dcf560985e83f52b82a7d3f5fe9051af586a2ccc46ef26", size = 2476561, upload-time = "2025-06-04T17:42:59.691Z" }, - { url = "https://files.pythonhosted.org/packages/94/8b/04c6b15f8c29b39f0679589753091cec8b192ab296d4fdaf9055544c4ec9/torchvision-0.22.1-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:ef46e065502f7300ad6abc98554131c35dc4c837b978d91306658f1a65c00baa", size = 7658543, upload-time = "2025-06-04T17:42:46.064Z" }, - { url = "https://files.pythonhosted.org/packages/ab/c0/131628e6d42682b0502c63fd7f647b8b5ca4bd94088f6c85ca7225db8ac4/torchvision-0.22.1-cp313-cp313t-win_amd64.whl", hash = "sha256:7414eeacfb941fa21acddcd725f1617da5630ec822e498660a4b864d7d998075", size = 1629892, upload-time = "2025-06-04T17:42:57.156Z" }, + { url = "https://files.pythonhosted.org/packages/4d/49/5ad5c3ff4920be0adee9eb4339b4fb3b023a0fc55b9ed8dbc73df92946b8/torchvision-0.23.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:7266871daca00ad46d1c073e55d972179d12a58fa5c9adec9a3db9bbed71284a", size = 1856885, upload-time = "2025-08-06T14:57:55.024Z" }, + { url = "https://files.pythonhosted.org/packages/25/44/ddd56d1637bac42a8c5da2c8c440d8a28c431f996dd9790f32dd9a96ca6e/torchvision-0.23.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:31c583ba27426a3a04eca8c05450524105c1564db41be6632f7536ef405a6de2", size = 2394251, upload-time = "2025-08-06T14:58:01.725Z" }, + { url = "https://files.pythonhosted.org/packages/93/f3/3cdf55bbf0f737304d997561c34ab0176222e0496b6743b0feab5995182c/torchvision-0.23.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:3932bf67256f2d095ce90a9f826f6033694c818856f4bb26794cf2ce64253e53", size = 8627497, upload-time = "2025-08-06T14:58:09.317Z" }, + { url = "https://files.pythonhosted.org/packages/97/90/02afe57c3ef4284c5cf89d3b7ae203829b3a981f72b93a7dd2a3fd2c83c1/torchvision-0.23.0-cp310-cp310-win_amd64.whl", hash = "sha256:83ee5bf827d61a8af14620c0a61d8608558638ac9c3bac8adb7b27138e2147d1", size = 1600760, upload-time = "2025-08-06T14:57:56.783Z" }, + { url = "https://files.pythonhosted.org/packages/f0/d7/15d3d7bd8d0239211b21673d1bac7bc345a4ad904a8e25bb3fd8a9cf1fbc/torchvision-0.23.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:49aa20e21f0c2bd458c71d7b449776cbd5f16693dd5807195a820612b8a229b7", size = 1856884, upload-time = "2025-08-06T14:58:00.237Z" }, + { url = "https://files.pythonhosted.org/packages/dd/14/7b44fe766b7d11e064c539d92a172fa9689a53b69029e24f2f1f51e7dc56/torchvision-0.23.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:01dc33ee24c79148aee7cdbcf34ae8a3c9da1674a591e781577b716d233b1fa6", size = 2395543, upload-time = "2025-08-06T14:58:04.373Z" }, + { url = "https://files.pythonhosted.org/packages/79/9c/fcb09aff941c8147d9e6aa6c8f67412a05622b0c750bcf796be4c85a58d4/torchvision-0.23.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:35c27941831b653f5101edfe62c03d196c13f32139310519e8228f35eae0e96a", size = 8628388, upload-time = "2025-08-06T14:58:07.802Z" }, + { url = "https://files.pythonhosted.org/packages/93/40/3415d890eb357b25a8e0a215d32365a88ecc75a283f75c4e919024b22d97/torchvision-0.23.0-cp311-cp311-win_amd64.whl", hash = "sha256:09bfde260e7963a15b80c9e442faa9f021c7e7f877ac0a36ca6561b367185013", size = 1600741, upload-time = "2025-08-06T14:57:59.158Z" }, + { url = "https://files.pythonhosted.org/packages/df/1d/0ea0b34bde92a86d42620f29baa6dcbb5c2fc85990316df5cb8f7abb8ea2/torchvision-0.23.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:e0e2c04a91403e8dd3af9756c6a024a1d9c0ed9c0d592a8314ded8f4fe30d440", size = 1856885, upload-time = "2025-08-06T14:58:06.503Z" }, + { url = "https://files.pythonhosted.org/packages/e2/00/2f6454decc0cd67158c7890364e446aad4b91797087a57a78e72e1a8f8bc/torchvision-0.23.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:6dd7c4d329a0e03157803031bc856220c6155ef08c26d4f5bbac938acecf0948", size = 2396614, upload-time = "2025-08-06T14:58:03.116Z" }, + { url = "https://files.pythonhosted.org/packages/e4/b5/3e580dcbc16f39a324f3dd71b90edbf02a42548ad44d2b4893cc92b1194b/torchvision-0.23.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:4e7d31c43bc7cbecbb1a5652ac0106b436aa66e26437585fc2c4b2cf04d6014c", size = 8627108, upload-time = "2025-08-06T14:58:12.956Z" }, + { url = "https://files.pythonhosted.org/packages/82/c1/c2fe6d61e110a8d0de2f94276899a2324a8f1e6aee559eb6b4629ab27466/torchvision-0.23.0-cp312-cp312-win_amd64.whl", hash = "sha256:a2e45272abe7b8bf0d06c405e78521b5757be1bd0ed7e5cd78120f7fdd4cbf35", size = 1600723, upload-time = "2025-08-06T14:57:57.986Z" }, + { url = "https://files.pythonhosted.org/packages/91/37/45a5b9407a7900f71d61b2b2f62db4b7c632debca397f205fdcacb502780/torchvision-0.23.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:1c37e325e09a184b730c3ef51424f383ec5745378dc0eca244520aca29722600", size = 1856886, upload-time = "2025-08-06T14:58:05.491Z" }, + { url = "https://files.pythonhosted.org/packages/ac/da/a06c60fc84fc849377cf035d3b3e9a1c896d52dbad493b963c0f1cdd74d0/torchvision-0.23.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:2f7fd6c15f3697e80627b77934f77705f3bc0e98278b989b2655de01f6903e1d", size = 2353112, upload-time = "2025-08-06T14:58:26.265Z" }, + { url = "https://files.pythonhosted.org/packages/a0/27/5ce65ba5c9d3b7d2ccdd79892ab86a2f87ac2ca6638f04bb0280321f1a9c/torchvision-0.23.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:a76fafe113b2977be3a21bf78f115438c1f88631d7a87203acb3dd6ae55889e6", size = 8627658, upload-time = "2025-08-06T14:58:15.999Z" }, + { url = "https://files.pythonhosted.org/packages/1f/e4/028a27b60aa578a2fa99d9d7334ff1871bb17008693ea055a2fdee96da0d/torchvision-0.23.0-cp313-cp313-win_amd64.whl", hash = "sha256:07d069cb29691ff566e3b7f11f20d91044f079e1dbdc9d72e0655899a9b06938", size = 1600749, upload-time = "2025-08-06T14:58:10.719Z" }, + { url = "https://files.pythonhosted.org/packages/05/35/72f91ad9ac7c19a849dedf083d347dc1123f0adeb401f53974f84f1d04c8/torchvision-0.23.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:2df618e1143805a7673aaf82cb5720dd9112d4e771983156aaf2ffff692eebf9", size = 2047192, upload-time = "2025-08-06T14:58:11.813Z" }, + { url = "https://files.pythonhosted.org/packages/1d/9d/406cea60a9eb9882145bcd62a184ee61e823e8e1d550cdc3c3ea866a9445/torchvision-0.23.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:2a3299d2b1d5a7aed2d3b6ffb69c672ca8830671967eb1cee1497bacd82fe47b", size = 2359295, upload-time = "2025-08-06T14:58:17.469Z" }, + { url = "https://files.pythonhosted.org/packages/2b/f4/34662f71a70fa1e59de99772142f22257ca750de05ccb400b8d2e3809c1d/torchvision-0.23.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:76bc4c0b63d5114aa81281390f8472a12a6a35ce9906e67ea6044e5af4cab60c", size = 8800474, upload-time = "2025-08-06T14:58:22.53Z" }, + { url = "https://files.pythonhosted.org/packages/6e/f5/b5a2d841a8d228b5dbda6d524704408e19e7ca6b7bb0f24490e081da1fa1/torchvision-0.23.0-cp313-cp313t-win_amd64.whl", hash = "sha256:b9e2dabf0da9c8aa9ea241afb63a8f3e98489e706b22ac3f30416a1be377153b", size = 1527667, upload-time = "2025-08-06T14:58:14.446Z" }, ] [[package]] name = "tornado" -version = "6.5.1" +version = "6.4.2" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/51/89/c72771c81d25d53fe33e3dca61c233b665b2780f21820ba6fd2c6793c12b/tornado-6.5.1.tar.gz", hash = "sha256:84ceece391e8eb9b2b95578db65e920d2a61070260594819589609ba9bc6308c", size = 509934, upload-time = "2025-05-22T18:15:38.788Z" } +sdist = { url = "https://files.pythonhosted.org/packages/59/45/a0daf161f7d6f36c3ea5fc0c2de619746cc3dd4c76402e9db545bd920f63/tornado-6.4.2.tar.gz", hash = "sha256:92bad5b4746e9879fd7bf1eb21dce4e3fc5128d71601f80005afa39237ad620b", size = 501135, upload-time = "2024-11-22T03:06:38.036Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/77/89/f4532dee6843c9e0ebc4e28d4be04c67f54f60813e4bf73d595fe7567452/tornado-6.5.1-cp39-abi3-macosx_10_9_universal2.whl", hash = "sha256:d50065ba7fd11d3bd41bcad0825227cc9a95154bad83239357094c36708001f7", size = 441948, upload-time = "2025-05-22T18:15:20.862Z" }, - { url = "https://files.pythonhosted.org/packages/15/9a/557406b62cffa395d18772e0cdcf03bed2fff03b374677348eef9f6a3792/tornado-6.5.1-cp39-abi3-macosx_10_9_x86_64.whl", hash = "sha256:9e9ca370f717997cb85606d074b0e5b247282cf5e2e1611568b8821afe0342d6", size = 440112, upload-time = "2025-05-22T18:15:22.591Z" }, - { url = "https://files.pythonhosted.org/packages/55/82/7721b7319013a3cf881f4dffa4f60ceff07b31b394e459984e7a36dc99ec/tornado-6.5.1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b77e9dfa7ed69754a54c89d82ef746398be82f749df69c4d3abe75c4d1ff4888", size = 443672, upload-time = "2025-05-22T18:15:24.027Z" }, - { url = "https://files.pythonhosted.org/packages/7d/42/d11c4376e7d101171b94e03cef0cbce43e823ed6567ceda571f54cf6e3ce/tornado-6.5.1-cp39-abi3-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:253b76040ee3bab8bcf7ba9feb136436a3787208717a1fb9f2c16b744fba7331", size = 443019, upload-time = "2025-05-22T18:15:25.735Z" }, - { url = "https://files.pythonhosted.org/packages/7d/f7/0c48ba992d875521ac761e6e04b0a1750f8150ae42ea26df1852d6a98942/tornado-6.5.1-cp39-abi3-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:308473f4cc5a76227157cdf904de33ac268af770b2c5f05ca6c1161d82fdd95e", size = 443252, upload-time = "2025-05-22T18:15:27.499Z" }, - { url = "https://files.pythonhosted.org/packages/89/46/d8d7413d11987e316df4ad42e16023cd62666a3c0dfa1518ffa30b8df06c/tornado-6.5.1-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:caec6314ce8a81cf69bd89909f4b633b9f523834dc1a352021775d45e51d9401", size = 443930, upload-time = "2025-05-22T18:15:29.299Z" }, - { url = "https://files.pythonhosted.org/packages/78/b2/f8049221c96a06df89bed68260e8ca94beca5ea532ffc63b1175ad31f9cc/tornado-6.5.1-cp39-abi3-musllinux_1_2_i686.whl", hash = "sha256:13ce6e3396c24e2808774741331638ee6c2f50b114b97a55c5b442df65fd9692", size = 443351, upload-time = "2025-05-22T18:15:31.038Z" }, - { url = "https://files.pythonhosted.org/packages/76/ff/6a0079e65b326cc222a54720a748e04a4db246870c4da54ece4577bfa702/tornado-6.5.1-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:5cae6145f4cdf5ab24744526cc0f55a17d76f02c98f4cff9daa08ae9a217448a", size = 443328, upload-time = "2025-05-22T18:15:32.426Z" }, - { url = "https://files.pythonhosted.org/packages/49/18/e3f902a1d21f14035b5bc6246a8c0f51e0eef562ace3a2cea403c1fb7021/tornado-6.5.1-cp39-abi3-win32.whl", hash = "sha256:e0a36e1bc684dca10b1aa75a31df8bdfed656831489bc1e6a6ebed05dc1ec365", size = 444396, upload-time = "2025-05-22T18:15:34.205Z" }, - { url = "https://files.pythonhosted.org/packages/7b/09/6526e32bf1049ee7de3bebba81572673b19a2a8541f795d887e92af1a8bc/tornado-6.5.1-cp39-abi3-win_amd64.whl", hash = "sha256:908e7d64567cecd4c2b458075589a775063453aeb1d2a1853eedb806922f568b", size = 444840, upload-time = "2025-05-22T18:15:36.1Z" }, - { url = "https://files.pythonhosted.org/packages/55/a7/535c44c7bea4578e48281d83c615219f3ab19e6abc67625ef637c73987be/tornado-6.5.1-cp39-abi3-win_arm64.whl", hash = "sha256:02420a0eb7bf617257b9935e2b754d1b63897525d8a289c9d65690d580b4dcf7", size = 443596, upload-time = "2025-05-22T18:15:37.433Z" }, + { url = "https://files.pythonhosted.org/packages/26/7e/71f604d8cea1b58f82ba3590290b66da1e72d840aeb37e0d5f7291bd30db/tornado-6.4.2-cp38-abi3-macosx_10_9_universal2.whl", hash = "sha256:e828cce1123e9e44ae2a50a9de3055497ab1d0aeb440c5ac23064d9e44880da1", size = 436299, upload-time = "2024-11-22T03:06:20.162Z" }, + { url = "https://files.pythonhosted.org/packages/96/44/87543a3b99016d0bf54fdaab30d24bf0af2e848f1d13d34a3a5380aabe16/tornado-6.4.2-cp38-abi3-macosx_10_9_x86_64.whl", hash = "sha256:072ce12ada169c5b00b7d92a99ba089447ccc993ea2143c9ede887e0937aa803", size = 434253, upload-time = "2024-11-22T03:06:22.39Z" }, + { url = "https://files.pythonhosted.org/packages/cb/fb/fdf679b4ce51bcb7210801ef4f11fdac96e9885daa402861751353beea6e/tornado-6.4.2-cp38-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1a017d239bd1bb0919f72af256a970624241f070496635784d9bf0db640d3fec", size = 437602, upload-time = "2024-11-22T03:06:24.214Z" }, + { url = "https://files.pythonhosted.org/packages/4f/3b/e31aeffffc22b475a64dbeb273026a21b5b566f74dee48742817626c47dc/tornado-6.4.2-cp38-abi3-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c36e62ce8f63409301537222faffcef7dfc5284f27eec227389f2ad11b09d946", size = 436972, upload-time = "2024-11-22T03:06:25.559Z" }, + { url = "https://files.pythonhosted.org/packages/22/55/b78a464de78051a30599ceb6983b01d8f732e6f69bf37b4ed07f642ac0fc/tornado-6.4.2-cp38-abi3-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bca9eb02196e789c9cb5c3c7c0f04fb447dc2adffd95265b2c7223a8a615ccbf", size = 437173, upload-time = "2024-11-22T03:06:27.584Z" }, + { url = "https://files.pythonhosted.org/packages/79/5e/be4fb0d1684eb822c9a62fb18a3e44a06188f78aa466b2ad991d2ee31104/tornado-6.4.2-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:304463bd0772442ff4d0f5149c6f1c2135a1fae045adf070821c6cdc76980634", size = 437892, upload-time = "2024-11-22T03:06:28.933Z" }, + { url = "https://files.pythonhosted.org/packages/f5/33/4f91fdd94ea36e1d796147003b490fe60a0215ac5737b6f9c65e160d4fe0/tornado-6.4.2-cp38-abi3-musllinux_1_2_i686.whl", hash = "sha256:c82c46813ba483a385ab2a99caeaedf92585a1f90defb5693351fa7e4ea0bf73", size = 437334, upload-time = "2024-11-22T03:06:30.428Z" }, + { url = "https://files.pythonhosted.org/packages/2b/ae/c1b22d4524b0e10da2f29a176fb2890386f7bd1f63aacf186444873a88a0/tornado-6.4.2-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:932d195ca9015956fa502c6b56af9eb06106140d844a335590c1ec7f5277d10c", size = 437261, upload-time = "2024-11-22T03:06:32.458Z" }, + { url = "https://files.pythonhosted.org/packages/b5/25/36dbd49ab6d179bcfc4c6c093a51795a4f3bed380543a8242ac3517a1751/tornado-6.4.2-cp38-abi3-win32.whl", hash = "sha256:2876cef82e6c5978fde1e0d5b1f919d756968d5b4282418f3146b79b58556482", size = 438463, upload-time = "2024-11-22T03:06:34.71Z" }, + { url = "https://files.pythonhosted.org/packages/61/cc/58b1adeb1bb46228442081e746fcdbc4540905c87e8add7c277540934edb/tornado-6.4.2-cp38-abi3-win_amd64.whl", hash = "sha256:908b71bf3ff37d81073356a5fadcc660eb10c1476ee6e2725588626ce7e5ca38", size = 438907, upload-time = "2024-11-22T03:06:36.71Z" }, ] [[package]] @@ -3961,38 +3691,26 @@ wheels = [ [[package]] name = "triton" -version = "3.3.1" +version = "3.4.0" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "setuptools", marker = "sys_platform == 'linux'" }, ] wheels = [ - { url = "https://files.pythonhosted.org/packages/8d/a9/549e51e9b1b2c9b854fd761a1d23df0ba2fbc60bd0c13b489ffa518cfcb7/triton-3.3.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b74db445b1c562844d3cfad6e9679c72e93fdfb1a90a24052b03bb5c49d1242e", size = 155600257, upload-time = "2025-05-29T23:39:36.085Z" }, - { url = "https://files.pythonhosted.org/packages/21/2f/3e56ea7b58f80ff68899b1dbe810ff257c9d177d288c6b0f55bf2fe4eb50/triton-3.3.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b31e3aa26f8cb3cc5bf4e187bf737cbacf17311e1112b781d4a059353dfd731b", size = 155689937, upload-time = "2025-05-29T23:39:44.182Z" }, - { url = "https://files.pythonhosted.org/packages/24/5f/950fb373bf9c01ad4eb5a8cd5eaf32cdf9e238c02f9293557a2129b9c4ac/triton-3.3.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9999e83aba21e1a78c1f36f21bce621b77bcaa530277a50484a7cb4a822f6e43", size = 155669138, upload-time = "2025-05-29T23:39:51.771Z" }, - { url = "https://files.pythonhosted.org/packages/74/1f/dfb531f90a2d367d914adfee771babbd3f1a5b26c3f5fbc458dee21daa78/triton-3.3.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b89d846b5a4198317fec27a5d3a609ea96b6d557ff44b56c23176546023c4240", size = 155673035, upload-time = "2025-05-29T23:40:02.468Z" }, - { url = "https://files.pythonhosted.org/packages/28/71/bd20ffcb7a64c753dc2463489a61bf69d531f308e390ad06390268c4ea04/triton-3.3.1-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a3198adb9d78b77818a5388bff89fa72ff36f9da0bc689db2f0a651a67ce6a42", size = 155735832, upload-time = "2025-05-29T23:40:10.522Z" }, + { url = "https://files.pythonhosted.org/packages/62/ee/0ee5f64a87eeda19bbad9bc54ae5ca5b98186ed00055281fd40fb4beb10e/triton-3.4.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7ff2785de9bc02f500e085420273bb5cc9c9bb767584a4aa28d6e360cec70128", size = 155430069, upload-time = "2025-07-30T19:58:21.715Z" }, + { url = "https://files.pythonhosted.org/packages/7d/39/43325b3b651d50187e591eefa22e236b2981afcebaefd4f2fc0ea99df191/triton-3.4.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7b70f5e6a41e52e48cfc087436c8a28c17ff98db369447bcaff3b887a3ab4467", size = 155531138, upload-time = "2025-07-30T19:58:29.908Z" }, + { url = "https://files.pythonhosted.org/packages/d0/66/b1eb52839f563623d185f0927eb3530ee4d5ffe9d377cdaf5346b306689e/triton-3.4.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:31c1d84a5c0ec2c0f8e8a072d7fd150cab84a9c239eaddc6706c081bfae4eb04", size = 155560068, upload-time = "2025-07-30T19:58:37.081Z" }, + { url = "https://files.pythonhosted.org/packages/30/7b/0a685684ed5322d2af0bddefed7906674f67974aa88b0fae6e82e3b766f6/triton-3.4.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:00be2964616f4c619193cb0d1b29a99bd4b001d7dc333816073f92cf2a8ccdeb", size = 155569223, upload-time = "2025-07-30T19:58:44.017Z" }, + { url = "https://files.pythonhosted.org/packages/20/63/8cb444ad5cdb25d999b7d647abac25af0ee37d292afc009940c05b82dda0/triton-3.4.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7936b18a3499ed62059414d7df563e6c163c5e16c3773678a3ee3d417865035d", size = 155659780, upload-time = "2025-07-30T19:58:51.171Z" }, ] [[package]] name = "typing-extensions" -version = "4.14.1" -source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/98/5a/da40306b885cc8c09109dc2e1abd358d5684b1425678151cdaed4731c822/typing_extensions-4.14.1.tar.gz", hash = "sha256:38b39f4aeeab64884ce9f74c94263ef78f3c22467c8724005483154c26648d36", size = 107673, upload-time = "2025-07-04T13:28:34.16Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/b5/00/d631e67a838026495268c2f6884f3711a15a9a2a96cd244fdaea53b823fb/typing_extensions-4.14.1-py3-none-any.whl", hash = "sha256:d1e1e3b58374dc93031d6eda2420a48ea44a36c2b4766a4fdeb3710755731d76", size = 43906, upload-time = "2025-07-04T13:28:32.743Z" }, -] - -[[package]] -name = "typing-inspection" -version = "0.4.1" +version = "4.12.2" source = { registry = "https://pypi.org/simple" } -dependencies = [ - { name = "typing-extensions" }, -] -sdist = { url = "https://files.pythonhosted.org/packages/f8/b1/0c11f5058406b3af7609f121aaa6b609744687f1d158b3c3a5bf4cc94238/typing_inspection-0.4.1.tar.gz", hash = "sha256:6ae134cc0203c33377d43188d4064e9b357dba58cff3185f22924610e70a9d28", size = 75726, upload-time = "2025-05-21T18:55:23.885Z" } +sdist = { url = "https://files.pythonhosted.org/packages/df/db/f35a00659bc03fec321ba8bce9420de607a1d37f8342eee1863174c69557/typing_extensions-4.12.2.tar.gz", hash = "sha256:1a7ead55c7e559dd4dee8856e3a88b41225abfe1ce8df57b7c13915fe121ffb8", size = 85321, upload-time = "2024-06-07T18:52:15.995Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/17/69/cd203477f944c353c31bade965f880aa1061fd6bf05ded0726ca845b6ff7/typing_inspection-0.4.1-py3-none-any.whl", hash = "sha256:389055682238f53b04f7badcb49b989835495a96700ced5dab2d8feae4b26f51", size = 14552, upload-time = "2025-05-21T18:55:22.152Z" }, + { url = "https://files.pythonhosted.org/packages/26/9f/ad63fc0248c5379346306f8668cda6e2e2e9c95e01216d2b8ffd9ff037d0/typing_extensions-4.12.2-py3-none-any.whl", hash = "sha256:04e5ca0351e0f3f85c6853954072df659d0d13fac324d0072316b67d7794700d", size = 37438, upload-time = "2024-06-07T18:52:13.582Z" }, ] [[package]] @@ -4018,64 +3736,67 @@ wheels = [ [[package]] name = "urllib3" -version = "2.5.0" +version = "2.3.0" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/15/22/9ee70a2574a4f4599c47dd506532914ce044817c7752a79b6a51286319bc/urllib3-2.5.0.tar.gz", hash = "sha256:3fc47733c7e419d4bc3f6b3dc2b4f890bb743906a30d56ba4a5bfa4bbff92760", size = 393185, upload-time = "2025-06-18T14:07:41.644Z" } +sdist = { url = "https://files.pythonhosted.org/packages/aa/63/e53da845320b757bf29ef6a9062f5c669fe997973f966045cb019c3f4b66/urllib3-2.3.0.tar.gz", hash = "sha256:f8c5449b3cf0861679ce7e0503c7b44b5ec981bec0d1d3795a07f1ba96f0204d", size = 307268, upload-time = "2024-12-22T07:47:30.032Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/a7/c2/fe1e52489ae3122415c51f387e221dd0773709bad6c6cdaa599e8a2c5185/urllib3-2.5.0-py3-none-any.whl", hash = "sha256:e6b01673c0fa6a13e374b50871808eb3bf7046c4b125b216f6bf1cc604cff0dc", size = 129795, upload-time = "2025-06-18T14:07:40.39Z" }, + { url = "https://files.pythonhosted.org/packages/c8/19/4ec628951a74043532ca2cf5d97b7b14863931476d117c471e8e2b1eb39f/urllib3-2.3.0-py3-none-any.whl", hash = "sha256:1cee9ad369867bfdbbb48b7dd50374c0967a0bb7710050facf0dd6911440e3df", size = 128369, upload-time = "2024-12-22T07:47:28.074Z" }, ] [[package]] name = "virtualenv" -version = "20.33.0" +version = "20.29.3" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "distlib" }, { name = "filelock" }, { name = "platformdirs" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/db/2e/8a70dcbe8bf15213a08f9b0325ede04faca5d362922ae0d62ef0fa4b069d/virtualenv-20.33.0.tar.gz", hash = "sha256:47e0c0d2ef1801fce721708ccdf2a28b9403fa2307c3268aebd03225976f61d2", size = 6082069, upload-time = "2025-08-03T08:09:19.014Z" } +sdist = { url = "https://files.pythonhosted.org/packages/c7/9c/57d19fa093bcf5ac61a48087dd44d00655f85421d1aa9722f8befbf3f40a/virtualenv-20.29.3.tar.gz", hash = "sha256:95e39403fcf3940ac45bc717597dba16110b74506131845d9b687d5e73d947ac", size = 4320280, upload-time = "2025-03-06T19:54:19.055Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/43/87/b22cf40cdf7e2b2bf83f38a94d2c90c5ad6c304896e5a12d0c08a602eb59/virtualenv-20.33.0-py3-none-any.whl", hash = "sha256:106b6baa8ab1b526d5a9b71165c85c456fbd49b16976c88e2bc9352ee3bc5d3f", size = 6060205, upload-time = "2025-08-03T08:09:16.674Z" }, + { url = "https://files.pythonhosted.org/packages/c2/eb/c6db6e3001d58c6a9e67c74bb7b4206767caa3ccc28c6b9eaf4c23fb4e34/virtualenv-20.29.3-py3-none-any.whl", hash = "sha256:3e3d00f5807e83b234dfb6122bf37cfadf4be216c53a49ac059d02414f819170", size = 4301458, upload-time = "2025-03-06T19:54:16.923Z" }, ] [[package]] name = "wadler-lindig" -version = "0.1.7" +version = "0.1.4" source = { registry = "https://pypi.org/simple" } -sdist = { url = "https://files.pythonhosted.org/packages/1e/67/cbae4bf7683a64755c2c1778c418fea96d00e34395bb91743f08bd951571/wadler_lindig-0.1.7.tar.gz", hash = "sha256:81d14d3fe77d441acf3ebd7f4aefac20c74128bf460e84b512806dccf7b2cd55", size = 15842, upload-time = "2025-06-18T07:00:42.843Z" } +sdist = { url = "https://files.pythonhosted.org/packages/39/7a/fea25d7985211556bbe2511d42e07453b484bf8e0d5d6109aabb08f52784/wadler_lindig-0.1.4.tar.gz", hash = "sha256:75aa3ddd384573c41d5c910fd990e655c2a641e5093cf5081650d0229daf87ad", size = 15356, upload-time = "2025-03-15T21:53:42.05Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/8d/96/04e7b441807b26b794da5b11e59ed7f83b2cf8af202bd7eba8ad2fa6046e/wadler_lindig-0.1.7-py3-none-any.whl", hash = "sha256:e3ec83835570fd0a9509f969162aeb9c65618f998b1f42918cfc8d45122fe953", size = 20516, upload-time = "2025-06-18T07:00:41.684Z" }, + { url = "https://files.pythonhosted.org/packages/7b/69/cfb1af44622044d4db0cad65721d283a921a4795f0ad121616b9eaa6ccd7/wadler_lindig-0.1.4-py3-none-any.whl", hash = "sha256:5c463aeb1f4ddc4acc12c3708d22ae21bcfc3e19e7c4d7aeef6642ea57b1a8b8", size = 20126, upload-time = "2025-03-15T21:53:40.835Z" }, ] [[package]] name = "wandb" -version = "0.21.0" +version = "0.19.11" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "click" }, + { name = "docker-pycreds" }, { name = "gitpython" }, - { name = "packaging" }, { name = "platformdirs" }, { name = "protobuf" }, + { name = "psutil" }, { name = "pydantic" }, { name = "pyyaml" }, { name = "requests" }, { name = "sentry-sdk" }, + { name = "setproctitle" }, + { name = "setuptools" }, { name = "typing-extensions" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/73/09/c84264a219e20efd615e4d5d150cc7d359d57d51328d3fa94ee02d70ed9c/wandb-0.21.0.tar.gz", hash = "sha256:473e01ef200b59d780416062991effa7349a34e51425d4be5ff482af2dc39e02", size = 40085784, upload-time = "2025-07-02T00:24:15.516Z" } +sdist = { url = "https://files.pythonhosted.org/packages/39/98/0ff2925a21b998d4b84731429f4554ca3d9b5cad42c09c075e7306c3aca0/wandb-0.19.11.tar.gz", hash = "sha256:3f50a27dfadbb25946a513ffe856c0e8e538b5626ef207aa50b00c3b0356bff8", size = 39511477, upload-time = "2025-05-07T20:50:01.341Z" } wheels = [ - { url = "https://files.pythonhosted.org/packages/38/dd/65eac086e1bc337bb5f0eed65ba1fe4a6dbc62c97f094e8e9df1ef83ffed/wandb-0.21.0-py3-none-any.whl", hash = "sha256:316e8cd4329738f7562f7369e6eabeeb28ef9d473203f7ead0d03e5dba01c90d", size = 6504284, upload-time = "2025-07-02T00:23:46.671Z" }, - { url = "https://files.pythonhosted.org/packages/17/a7/80556ce9097f59e10807aa68f4a9b29d736a90dca60852a9e2af1641baf8/wandb-0.21.0-py3-none-macosx_10_14_x86_64.whl", hash = "sha256:701d9cbdfcc8550a330c1b54a26f1585519180e0f19247867446593d34ace46b", size = 21717388, upload-time = "2025-07-02T00:23:49.348Z" }, - { url = "https://files.pythonhosted.org/packages/23/ae/660bc75aa37bd23409822ea5ed616177d94873172d34271693c80405c820/wandb-0.21.0-py3-none-macosx_11_0_arm64.whl", hash = "sha256:01689faa6b691df23ba2367e0a1ecf6e4d0be44474905840098eedd1fbcb8bdf", size = 21141465, upload-time = "2025-07-02T00:23:52.602Z" }, - { url = "https://files.pythonhosted.org/packages/23/ab/9861929530be56557c74002868c85d0d8ac57050cc21863afe909ae3d46f/wandb-0.21.0-py3-none-macosx_11_0_x86_64.whl", hash = "sha256:55d3f42ddb7971d1699752dff2b85bcb5906ad098d18ab62846c82e9ce5a238d", size = 21793511, upload-time = "2025-07-02T00:23:55.447Z" }, - { url = "https://files.pythonhosted.org/packages/de/52/e5cad2eff6fbed1ac06f4a5b718457fa2fd437f84f5c8f0d31995a2ef046/wandb-0.21.0-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:893508f0c7da48917448daa5cd622c27ce7ce15119adaa861185034c2bd7b14c", size = 20704643, upload-time = "2025-07-02T00:23:58.255Z" }, - { url = "https://files.pythonhosted.org/packages/83/8f/6bed9358cc33767c877b221d4f565e1ddf00caf4bbbe54d2e3bbc932c6a7/wandb-0.21.0-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a4e8245a8912247ddf7654f7b5330f583a6c56ab88fee65589158490d583c57d", size = 22243012, upload-time = "2025-07-02T00:24:01.423Z" }, - { url = "https://files.pythonhosted.org/packages/be/61/9048015412ea5ca916844af55add4fed7c21fe1ad70bb137951e70b550c5/wandb-0.21.0-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:2e4c4f951e0d02755e315679bfdcb5bc38c1b02e2e5abc5432b91a91bb0cf246", size = 20716440, upload-time = "2025-07-02T00:24:04.198Z" }, - { url = "https://files.pythonhosted.org/packages/02/d9/fcd2273d8ec3f79323e40a031aba5d32d6fa9065702010eb428b5ffbab62/wandb-0.21.0-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:873749966eeac0069e0e742e6210641b6227d454fb1dae2cf5c437c6ed42d3ca", size = 22320652, upload-time = "2025-07-02T00:24:07.175Z" }, - { url = "https://files.pythonhosted.org/packages/80/68/b8308db6b9c3c96dcd03be17c019aee105e1d7dc1e74d70756cdfb9241c6/wandb-0.21.0-py3-none-win32.whl", hash = "sha256:9d3cccfba658fa011d6cab9045fa4f070a444885e8902ae863802549106a5dab", size = 21484296, upload-time = "2025-07-02T00:24:10.147Z" }, - { url = "https://files.pythonhosted.org/packages/cf/96/71cc033e8abd00e54465e68764709ed945e2da2d66d764f72f4660262b22/wandb-0.21.0-py3-none-win_amd64.whl", hash = "sha256:28a0b2dad09d7c7344ac62b0276be18a2492a5578e4d7c84937a3e1991edaac7", size = 21484301, upload-time = "2025-07-02T00:24:12.658Z" }, + { url = "https://files.pythonhosted.org/packages/4f/2c/f8bab58c73fdde4442f1baffd9ea5d1bb3113906a97a27e8d9ab72db7a69/wandb-0.19.11-py3-none-any.whl", hash = "sha256:ff3bf050ba25ebae7aedc9a775ffab90c28068832edfe5458423f488c2558f82", size = 6481327, upload-time = "2025-05-07T20:49:33.461Z" }, + { url = "https://files.pythonhosted.org/packages/45/4a/34b364280f690f4c6d7660f528fba9f13bdecabc4c869d266a4632cf836e/wandb-0.19.11-py3-none-macosx_10_14_x86_64.whl", hash = "sha256:0823fd9aa6343f40c04e01959997ca8c6d6adf1bd81c8d45261fa4915f1c6b67", size = 20555751, upload-time = "2025-05-07T20:49:36.392Z" }, + { url = "https://files.pythonhosted.org/packages/d8/e6/a27868fdb83a60df37b9d15e52c3353dd88d74442f27ae48cf765c6b9554/wandb-0.19.11-py3-none-macosx_11_0_arm64.whl", hash = "sha256:c758ef5439599d9023db5b3cf1698477055d82f9fae48af2779f63f1d289167c", size = 20377587, upload-time = "2025-05-07T20:49:39.126Z" }, + { url = "https://files.pythonhosted.org/packages/21/f7/d5cf5b58c2b3015364c7b2b6af6a440cbeda4103b67332e1e64b30f6252d/wandb-0.19.11-py3-none-macosx_11_0_x86_64.whl", hash = "sha256:de2dfd4911e7691735e271654c735e7b90cdee9d29a3796fbf06e9e92d48f3d7", size = 20985041, upload-time = "2025-05-07T20:49:41.571Z" }, + { url = "https://files.pythonhosted.org/packages/68/06/8b827f16a0b8f18002d2fffa7c5a7fd447946e0d0c68aeec0dd7eb18cdd3/wandb-0.19.11-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cfff738850770d26b13f8f3fe400a6456f1e39e87f3f29d5aa241b249476df95", size = 20017696, upload-time = "2025-05-07T20:49:44.04Z" }, + { url = "https://files.pythonhosted.org/packages/f9/31/eeb2878b26566c04c3e9b8b20b3ec3c54a2be50535088d36a37c008e07a3/wandb-0.19.11-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e8ff673007448df11cc69379ae0df28ead866800dc1ec7bc151b402db0bbcf40", size = 21425857, upload-time = "2025-05-07T20:49:46.347Z" }, + { url = "https://files.pythonhosted.org/packages/10/30/08988360678ae78334bb16625c28260fcaba49f500b89f8766807cb74d71/wandb-0.19.11-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:858bc5023fa1b3285d89d15f62be78afdb28301064daa49ea3f4ebde5dcedad2", size = 20023145, upload-time = "2025-05-07T20:49:48.965Z" }, + { url = "https://files.pythonhosted.org/packages/c8/e9/a639c42c8ca517c4d25e8970d64d0c5a9bd35b784faed5f47d9cca3dcd12/wandb-0.19.11-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:90e4b57649896acb16c3dd41b3093df1a169c2f1d94ff15d76af86b8a60dcdac", size = 21504842, upload-time = "2025-05-07T20:49:51.628Z" }, + { url = "https://files.pythonhosted.org/packages/44/74/dbe9277dd935b77dd16939cdf15357766fec0813a6e336cf5f1d07eb016e/wandb-0.19.11-py3-none-win32.whl", hash = "sha256:38dea43c7926d8800405a73b80b9adfe81eb315fc6f2ac6885c77eb966634421", size = 20767584, upload-time = "2025-05-07T20:49:56.629Z" }, + { url = "https://files.pythonhosted.org/packages/36/d5/215cac3edec5c5ac6e7231beb9d22466d5d4e4a132fa3a1d044f7d682c15/wandb-0.19.11-py3-none-win_amd64.whl", hash = "sha256:73402003c56ddc2198878492ab2bff55bb49bce5587eae5960e737d27c0c48f7", size = 20767588, upload-time = "2025-05-07T20:49:58.85Z" }, ] [[package]] @@ -4153,99 +3874,78 @@ wheels = [ [[package]] name = "yarl" -version = "1.20.1" +version = "1.18.3" source = { registry = "https://pypi.org/simple" } dependencies = [ { name = "idna" }, { name = "multidict" }, { name = "propcache" }, ] -sdist = { url = "https://files.pythonhosted.org/packages/3c/fb/efaa23fa4e45537b827620f04cf8f3cd658b76642205162e072703a5b963/yarl-1.20.1.tar.gz", hash = "sha256:d017a4997ee50c91fd5466cef416231bb82177b93b029906cefc542ce14c35ac", size = 186428, upload-time = "2025-06-10T00:46:09.923Z" } -wheels = [ - { url = "https://files.pythonhosted.org/packages/cb/65/7fed0d774abf47487c64be14e9223749468922817b5e8792b8a64792a1bb/yarl-1.20.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:6032e6da6abd41e4acda34d75a816012717000fa6839f37124a47fcefc49bec4", size = 132910, upload-time = "2025-06-10T00:42:31.108Z" }, - { url = "https://files.pythonhosted.org/packages/8a/7b/988f55a52da99df9e56dc733b8e4e5a6ae2090081dc2754fc8fd34e60aa0/yarl-1.20.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:2c7b34d804b8cf9b214f05015c4fee2ebe7ed05cf581e7192c06555c71f4446a", size = 90644, upload-time = "2025-06-10T00:42:33.851Z" }, - { url = "https://files.pythonhosted.org/packages/f7/de/30d98f03e95d30c7e3cc093759982d038c8833ec2451001d45ef4854edc1/yarl-1.20.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:0c869f2651cc77465f6cd01d938d91a11d9ea5d798738c1dc077f3de0b5e5fed", size = 89322, upload-time = "2025-06-10T00:42:35.688Z" }, - { url = "https://files.pythonhosted.org/packages/e0/7a/f2f314f5ebfe9200724b0b748de2186b927acb334cf964fd312eb86fc286/yarl-1.20.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:62915e6688eb4d180d93840cda4110995ad50c459bf931b8b3775b37c264af1e", size = 323786, upload-time = "2025-06-10T00:42:37.817Z" }, - { url = "https://files.pythonhosted.org/packages/15/3f/718d26f189db96d993d14b984ce91de52e76309d0fd1d4296f34039856aa/yarl-1.20.1-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:41ebd28167bc6af8abb97fec1a399f412eec5fd61a3ccbe2305a18b84fb4ca73", size = 319627, upload-time = "2025-06-10T00:42:39.937Z" }, - { url = "https://files.pythonhosted.org/packages/a5/76/8fcfbf5fa2369157b9898962a4a7d96764b287b085b5b3d9ffae69cdefd1/yarl-1.20.1-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:21242b4288a6d56f04ea193adde174b7e347ac46ce6bc84989ff7c1b1ecea84e", size = 339149, upload-time = "2025-06-10T00:42:42.627Z" }, - { url = "https://files.pythonhosted.org/packages/3c/95/d7fc301cc4661785967acc04f54a4a42d5124905e27db27bb578aac49b5c/yarl-1.20.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:bea21cdae6c7eb02ba02a475f37463abfe0a01f5d7200121b03e605d6a0439f8", size = 333327, upload-time = "2025-06-10T00:42:44.842Z" }, - { url = "https://files.pythonhosted.org/packages/65/94/e21269718349582eee81efc5c1c08ee71c816bfc1585b77d0ec3f58089eb/yarl-1.20.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1f8a891e4a22a89f5dde7862994485e19db246b70bb288d3ce73a34422e55b23", size = 326054, upload-time = "2025-06-10T00:42:47.149Z" }, - { url = "https://files.pythonhosted.org/packages/32/ae/8616d1f07853704523519f6131d21f092e567c5af93de7e3e94b38d7f065/yarl-1.20.1-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:dd803820d44c8853a109a34e3660e5a61beae12970da479cf44aa2954019bf70", size = 315035, upload-time = "2025-06-10T00:42:48.852Z" }, - { url = "https://files.pythonhosted.org/packages/48/aa/0ace06280861ef055855333707db5e49c6e3a08840a7ce62682259d0a6c0/yarl-1.20.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:b982fa7f74c80d5c0c7b5b38f908971e513380a10fecea528091405f519b9ebb", size = 338962, upload-time = "2025-06-10T00:42:51.024Z" }, - { url = "https://files.pythonhosted.org/packages/20/52/1e9d0e6916f45a8fb50e6844f01cb34692455f1acd548606cbda8134cd1e/yarl-1.20.1-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:33f29ecfe0330c570d997bcf1afd304377f2e48f61447f37e846a6058a4d33b2", size = 335399, upload-time = "2025-06-10T00:42:53.007Z" }, - { url = "https://files.pythonhosted.org/packages/f2/65/60452df742952c630e82f394cd409de10610481d9043aa14c61bf846b7b1/yarl-1.20.1-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:835ab2cfc74d5eb4a6a528c57f05688099da41cf4957cf08cad38647e4a83b30", size = 338649, upload-time = "2025-06-10T00:42:54.964Z" }, - { url = "https://files.pythonhosted.org/packages/7b/f5/6cd4ff38dcde57a70f23719a838665ee17079640c77087404c3d34da6727/yarl-1.20.1-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:46b5e0ccf1943a9a6e766b2c2b8c732c55b34e28be57d8daa2b3c1d1d4009309", size = 358563, upload-time = "2025-06-10T00:42:57.28Z" }, - { url = "https://files.pythonhosted.org/packages/d1/90/c42eefd79d0d8222cb3227bdd51b640c0c1d0aa33fe4cc86c36eccba77d3/yarl-1.20.1-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:df47c55f7d74127d1b11251fe6397d84afdde0d53b90bedb46a23c0e534f9d24", size = 357609, upload-time = "2025-06-10T00:42:59.055Z" }, - { url = "https://files.pythonhosted.org/packages/03/c8/cea6b232cb4617514232e0f8a718153a95b5d82b5290711b201545825532/yarl-1.20.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:76d12524d05841276b0e22573f28d5fbcb67589836772ae9244d90dd7d66aa13", size = 350224, upload-time = "2025-06-10T00:43:01.248Z" }, - { url = "https://files.pythonhosted.org/packages/ce/a3/eaa0ab9712f1f3d01faf43cf6f1f7210ce4ea4a7e9b28b489a2261ca8db9/yarl-1.20.1-cp310-cp310-win32.whl", hash = "sha256:6c4fbf6b02d70e512d7ade4b1f998f237137f1417ab07ec06358ea04f69134f8", size = 81753, upload-time = "2025-06-10T00:43:03.486Z" }, - { url = "https://files.pythonhosted.org/packages/8f/34/e4abde70a9256465fe31c88ed02c3f8502b7b5dead693a4f350a06413f28/yarl-1.20.1-cp310-cp310-win_amd64.whl", hash = "sha256:aef6c4d69554d44b7f9d923245f8ad9a707d971e6209d51279196d8e8fe1ae16", size = 86817, upload-time = "2025-06-10T00:43:05.231Z" }, - { url = "https://files.pythonhosted.org/packages/b1/18/893b50efc2350e47a874c5c2d67e55a0ea5df91186b2a6f5ac52eff887cd/yarl-1.20.1-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:47ee6188fea634bdfaeb2cc420f5b3b17332e6225ce88149a17c413c77ff269e", size = 133833, upload-time = "2025-06-10T00:43:07.393Z" }, - { url = "https://files.pythonhosted.org/packages/89/ed/b8773448030e6fc47fa797f099ab9eab151a43a25717f9ac043844ad5ea3/yarl-1.20.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:d0f6500f69e8402d513e5eedb77a4e1818691e8f45e6b687147963514d84b44b", size = 91070, upload-time = "2025-06-10T00:43:09.538Z" }, - { url = "https://files.pythonhosted.org/packages/e3/e3/409bd17b1e42619bf69f60e4f031ce1ccb29bd7380117a55529e76933464/yarl-1.20.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:7a8900a42fcdaad568de58887c7b2f602962356908eedb7628eaf6021a6e435b", size = 89818, upload-time = "2025-06-10T00:43:11.575Z" }, - { url = "https://files.pythonhosted.org/packages/f8/77/64d8431a4d77c856eb2d82aa3de2ad6741365245a29b3a9543cd598ed8c5/yarl-1.20.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:bad6d131fda8ef508b36be3ece16d0902e80b88ea7200f030a0f6c11d9e508d4", size = 347003, upload-time = "2025-06-10T00:43:14.088Z" }, - { url = "https://files.pythonhosted.org/packages/8d/d2/0c7e4def093dcef0bd9fa22d4d24b023788b0a33b8d0088b51aa51e21e99/yarl-1.20.1-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:df018d92fe22aaebb679a7f89fe0c0f368ec497e3dda6cb81a567610f04501f1", size = 336537, upload-time = "2025-06-10T00:43:16.431Z" }, - { url = "https://files.pythonhosted.org/packages/f0/f3/fc514f4b2cf02cb59d10cbfe228691d25929ce8f72a38db07d3febc3f706/yarl-1.20.1-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8f969afbb0a9b63c18d0feecf0db09d164b7a44a053e78a7d05f5df163e43833", size = 362358, upload-time = "2025-06-10T00:43:18.704Z" }, - { url = "https://files.pythonhosted.org/packages/ea/6d/a313ac8d8391381ff9006ac05f1d4331cee3b1efaa833a53d12253733255/yarl-1.20.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:812303eb4aa98e302886ccda58d6b099e3576b1b9276161469c25803a8db277d", size = 357362, upload-time = "2025-06-10T00:43:20.888Z" }, - { url = "https://files.pythonhosted.org/packages/00/70/8f78a95d6935a70263d46caa3dd18e1f223cf2f2ff2037baa01a22bc5b22/yarl-1.20.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:98c4a7d166635147924aa0bf9bfe8d8abad6fffa6102de9c99ea04a1376f91e8", size = 348979, upload-time = "2025-06-10T00:43:23.169Z" }, - { url = "https://files.pythonhosted.org/packages/cb/05/42773027968968f4f15143553970ee36ead27038d627f457cc44bbbeecf3/yarl-1.20.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:12e768f966538e81e6e7550f9086a6236b16e26cd964cf4df35349970f3551cf", size = 337274, upload-time = "2025-06-10T00:43:27.111Z" }, - { url = "https://files.pythonhosted.org/packages/05/be/665634aa196954156741ea591d2f946f1b78ceee8bb8f28488bf28c0dd62/yarl-1.20.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:fe41919b9d899661c5c28a8b4b0acf704510b88f27f0934ac7a7bebdd8938d5e", size = 363294, upload-time = "2025-06-10T00:43:28.96Z" }, - { url = "https://files.pythonhosted.org/packages/eb/90/73448401d36fa4e210ece5579895731f190d5119c4b66b43b52182e88cd5/yarl-1.20.1-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:8601bc010d1d7780592f3fc1bdc6c72e2b6466ea34569778422943e1a1f3c389", size = 358169, upload-time = "2025-06-10T00:43:30.701Z" }, - { url = "https://files.pythonhosted.org/packages/c3/b0/fce922d46dc1eb43c811f1889f7daa6001b27a4005587e94878570300881/yarl-1.20.1-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:daadbdc1f2a9033a2399c42646fbd46da7992e868a5fe9513860122d7fe7a73f", size = 362776, upload-time = "2025-06-10T00:43:32.51Z" }, - { url = "https://files.pythonhosted.org/packages/f1/0d/b172628fce039dae8977fd22caeff3eeebffd52e86060413f5673767c427/yarl-1.20.1-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:03aa1e041727cb438ca762628109ef1333498b122e4c76dd858d186a37cec845", size = 381341, upload-time = "2025-06-10T00:43:34.543Z" }, - { url = "https://files.pythonhosted.org/packages/6b/9b/5b886d7671f4580209e855974fe1cecec409aa4a89ea58b8f0560dc529b1/yarl-1.20.1-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:642980ef5e0fa1de5fa96d905c7e00cb2c47cb468bfcac5a18c58e27dbf8d8d1", size = 379988, upload-time = "2025-06-10T00:43:36.489Z" }, - { url = "https://files.pythonhosted.org/packages/73/be/75ef5fd0fcd8f083a5d13f78fd3f009528132a1f2a1d7c925c39fa20aa79/yarl-1.20.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:86971e2795584fe8c002356d3b97ef6c61862720eeff03db2a7c86b678d85b3e", size = 371113, upload-time = "2025-06-10T00:43:38.592Z" }, - { url = "https://files.pythonhosted.org/packages/50/4f/62faab3b479dfdcb741fe9e3f0323e2a7d5cd1ab2edc73221d57ad4834b2/yarl-1.20.1-cp311-cp311-win32.whl", hash = "sha256:597f40615b8d25812f14562699e287f0dcc035d25eb74da72cae043bb884d773", size = 81485, upload-time = "2025-06-10T00:43:41.038Z" }, - { url = "https://files.pythonhosted.org/packages/f0/09/d9c7942f8f05c32ec72cd5c8e041c8b29b5807328b68b4801ff2511d4d5e/yarl-1.20.1-cp311-cp311-win_amd64.whl", hash = "sha256:26ef53a9e726e61e9cd1cda6b478f17e350fb5800b4bd1cd9fe81c4d91cfeb2e", size = 86686, upload-time = "2025-06-10T00:43:42.692Z" }, - { url = "https://files.pythonhosted.org/packages/5f/9a/cb7fad7d73c69f296eda6815e4a2c7ed53fc70c2f136479a91c8e5fbdb6d/yarl-1.20.1-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:bdcc4cd244e58593a4379fe60fdee5ac0331f8eb70320a24d591a3be197b94a9", size = 133667, upload-time = "2025-06-10T00:43:44.369Z" }, - { url = "https://files.pythonhosted.org/packages/67/38/688577a1cb1e656e3971fb66a3492501c5a5df56d99722e57c98249e5b8a/yarl-1.20.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b29a2c385a5f5b9c7d9347e5812b6f7ab267193c62d282a540b4fc528c8a9d2a", size = 91025, upload-time = "2025-06-10T00:43:46.295Z" }, - { url = "https://files.pythonhosted.org/packages/50/ec/72991ae51febeb11a42813fc259f0d4c8e0507f2b74b5514618d8b640365/yarl-1.20.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:1112ae8154186dfe2de4732197f59c05a83dc814849a5ced892b708033f40dc2", size = 89709, upload-time = "2025-06-10T00:43:48.22Z" }, - { url = "https://files.pythonhosted.org/packages/99/da/4d798025490e89426e9f976702e5f9482005c548c579bdae792a4c37769e/yarl-1.20.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:90bbd29c4fe234233f7fa2b9b121fb63c321830e5d05b45153a2ca68f7d310ee", size = 352287, upload-time = "2025-06-10T00:43:49.924Z" }, - { url = "https://files.pythonhosted.org/packages/1a/26/54a15c6a567aac1c61b18aa0f4b8aa2e285a52d547d1be8bf48abe2b3991/yarl-1.20.1-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:680e19c7ce3710ac4cd964e90dad99bf9b5029372ba0c7cbfcd55e54d90ea819", size = 345429, upload-time = "2025-06-10T00:43:51.7Z" }, - { url = "https://files.pythonhosted.org/packages/d6/95/9dcf2386cb875b234353b93ec43e40219e14900e046bf6ac118f94b1e353/yarl-1.20.1-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4a979218c1fdb4246a05efc2cc23859d47c89af463a90b99b7c56094daf25a16", size = 365429, upload-time = "2025-06-10T00:43:53.494Z" }, - { url = "https://files.pythonhosted.org/packages/91/b2/33a8750f6a4bc224242a635f5f2cff6d6ad5ba651f6edcccf721992c21a0/yarl-1.20.1-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:255b468adf57b4a7b65d8aad5b5138dce6a0752c139965711bdcb81bc370e1b6", size = 363862, upload-time = "2025-06-10T00:43:55.766Z" }, - { url = "https://files.pythonhosted.org/packages/98/28/3ab7acc5b51f4434b181b0cee8f1f4b77a65919700a355fb3617f9488874/yarl-1.20.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a97d67108e79cfe22e2b430d80d7571ae57d19f17cda8bb967057ca8a7bf5bfd", size = 355616, upload-time = "2025-06-10T00:43:58.056Z" }, - { url = "https://files.pythonhosted.org/packages/36/a3/f666894aa947a371724ec7cd2e5daa78ee8a777b21509b4252dd7bd15e29/yarl-1.20.1-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8570d998db4ddbfb9a590b185a0a33dbf8aafb831d07a5257b4ec9948df9cb0a", size = 339954, upload-time = "2025-06-10T00:43:59.773Z" }, - { url = "https://files.pythonhosted.org/packages/f1/81/5f466427e09773c04219d3450d7a1256138a010b6c9f0af2d48565e9ad13/yarl-1.20.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:97c75596019baae7c71ccf1d8cc4738bc08134060d0adfcbe5642f778d1dca38", size = 365575, upload-time = "2025-06-10T00:44:02.051Z" }, - { url = "https://files.pythonhosted.org/packages/2e/e3/e4b0ad8403e97e6c9972dd587388940a032f030ebec196ab81a3b8e94d31/yarl-1.20.1-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:1c48912653e63aef91ff988c5432832692ac5a1d8f0fb8a33091520b5bbe19ef", size = 365061, upload-time = "2025-06-10T00:44:04.196Z" }, - { url = "https://files.pythonhosted.org/packages/ac/99/b8a142e79eb86c926f9f06452eb13ecb1bb5713bd01dc0038faf5452e544/yarl-1.20.1-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:4c3ae28f3ae1563c50f3d37f064ddb1511ecc1d5584e88c6b7c63cf7702a6d5f", size = 364142, upload-time = "2025-06-10T00:44:06.527Z" }, - { url = "https://files.pythonhosted.org/packages/34/f2/08ed34a4a506d82a1a3e5bab99ccd930a040f9b6449e9fd050320e45845c/yarl-1.20.1-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:c5e9642f27036283550f5f57dc6156c51084b458570b9d0d96100c8bebb186a8", size = 381894, upload-time = "2025-06-10T00:44:08.379Z" }, - { url = "https://files.pythonhosted.org/packages/92/f8/9a3fbf0968eac704f681726eff595dce9b49c8a25cd92bf83df209668285/yarl-1.20.1-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:2c26b0c49220d5799f7b22c6838409ee9bc58ee5c95361a4d7831f03cc225b5a", size = 383378, upload-time = "2025-06-10T00:44:10.51Z" }, - { url = "https://files.pythonhosted.org/packages/af/85/9363f77bdfa1e4d690957cd39d192c4cacd1c58965df0470a4905253b54f/yarl-1.20.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:564ab3d517e3d01c408c67f2e5247aad4019dcf1969982aba3974b4093279004", size = 374069, upload-time = "2025-06-10T00:44:12.834Z" }, - { url = "https://files.pythonhosted.org/packages/35/99/9918c8739ba271dcd935400cff8b32e3cd319eaf02fcd023d5dcd487a7c8/yarl-1.20.1-cp312-cp312-win32.whl", hash = "sha256:daea0d313868da1cf2fac6b2d3a25c6e3a9e879483244be38c8e6a41f1d876a5", size = 81249, upload-time = "2025-06-10T00:44:14.731Z" }, - { url = "https://files.pythonhosted.org/packages/eb/83/5d9092950565481b413b31a23e75dd3418ff0a277d6e0abf3729d4d1ce25/yarl-1.20.1-cp312-cp312-win_amd64.whl", hash = "sha256:48ea7d7f9be0487339828a4de0360d7ce0efc06524a48e1810f945c45b813698", size = 86710, upload-time = "2025-06-10T00:44:16.716Z" }, - { url = "https://files.pythonhosted.org/packages/8a/e1/2411b6d7f769a07687acee88a062af5833cf1966b7266f3d8dfb3d3dc7d3/yarl-1.20.1-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:0b5ff0fbb7c9f1b1b5ab53330acbfc5247893069e7716840c8e7d5bb7355038a", size = 131811, upload-time = "2025-06-10T00:44:18.933Z" }, - { url = "https://files.pythonhosted.org/packages/b2/27/584394e1cb76fb771371770eccad35de400e7b434ce3142c2dd27392c968/yarl-1.20.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:14f326acd845c2b2e2eb38fb1346c94f7f3b01a4f5c788f8144f9b630bfff9a3", size = 90078, upload-time = "2025-06-10T00:44:20.635Z" }, - { url = "https://files.pythonhosted.org/packages/bf/9a/3246ae92d4049099f52d9b0fe3486e3b500e29b7ea872d0f152966fc209d/yarl-1.20.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:f60e4ad5db23f0b96e49c018596707c3ae89f5d0bd97f0ad3684bcbad899f1e7", size = 88748, upload-time = "2025-06-10T00:44:22.34Z" }, - { url = "https://files.pythonhosted.org/packages/a3/25/35afe384e31115a1a801fbcf84012d7a066d89035befae7c5d4284df1e03/yarl-1.20.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:49bdd1b8e00ce57e68ba51916e4bb04461746e794e7c4d4bbc42ba2f18297691", size = 349595, upload-time = "2025-06-10T00:44:24.314Z" }, - { url = "https://files.pythonhosted.org/packages/28/2d/8aca6cb2cabc8f12efcb82749b9cefecbccfc7b0384e56cd71058ccee433/yarl-1.20.1-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:66252d780b45189975abfed839616e8fd2dbacbdc262105ad7742c6ae58f3e31", size = 342616, upload-time = "2025-06-10T00:44:26.167Z" }, - { url = "https://files.pythonhosted.org/packages/0b/e9/1312633d16b31acf0098d30440ca855e3492d66623dafb8e25b03d00c3da/yarl-1.20.1-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:59174e7332f5d153d8f7452a102b103e2e74035ad085f404df2e40e663a22b28", size = 361324, upload-time = "2025-06-10T00:44:27.915Z" }, - { url = "https://files.pythonhosted.org/packages/bc/a0/688cc99463f12f7669eec7c8acc71ef56a1521b99eab7cd3abb75af887b0/yarl-1.20.1-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:e3968ec7d92a0c0f9ac34d5ecfd03869ec0cab0697c91a45db3fbbd95fe1b653", size = 359676, upload-time = "2025-06-10T00:44:30.041Z" }, - { url = "https://files.pythonhosted.org/packages/af/44/46407d7f7a56e9a85a4c207724c9f2c545c060380718eea9088f222ba697/yarl-1.20.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d1a4fbb50e14396ba3d375f68bfe02215d8e7bc3ec49da8341fe3157f59d2ff5", size = 352614, upload-time = "2025-06-10T00:44:32.171Z" }, - { url = "https://files.pythonhosted.org/packages/b1/91/31163295e82b8d5485d31d9cf7754d973d41915cadce070491778d9c9825/yarl-1.20.1-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:11a62c839c3a8eac2410e951301309426f368388ff2f33799052787035793b02", size = 336766, upload-time = "2025-06-10T00:44:34.494Z" }, - { url = "https://files.pythonhosted.org/packages/b4/8e/c41a5bc482121f51c083c4c2bcd16b9e01e1cf8729e380273a952513a21f/yarl-1.20.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:041eaa14f73ff5a8986b4388ac6bb43a77f2ea09bf1913df7a35d4646db69e53", size = 364615, upload-time = "2025-06-10T00:44:36.856Z" }, - { url = "https://files.pythonhosted.org/packages/e3/5b/61a3b054238d33d70ea06ebba7e58597891b71c699e247df35cc984ab393/yarl-1.20.1-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:377fae2fef158e8fd9d60b4c8751387b8d1fb121d3d0b8e9b0be07d1b41e83dc", size = 360982, upload-time = "2025-06-10T00:44:39.141Z" }, - { url = "https://files.pythonhosted.org/packages/df/a3/6a72fb83f8d478cb201d14927bc8040af901811a88e0ff2da7842dd0ed19/yarl-1.20.1-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:1c92f4390e407513f619d49319023664643d3339bd5e5a56a3bebe01bc67ec04", size = 369792, upload-time = "2025-06-10T00:44:40.934Z" }, - { url = "https://files.pythonhosted.org/packages/7c/af/4cc3c36dfc7c077f8dedb561eb21f69e1e9f2456b91b593882b0b18c19dc/yarl-1.20.1-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d25ddcf954df1754ab0f86bb696af765c5bfaba39b74095f27eececa049ef9a4", size = 382049, upload-time = "2025-06-10T00:44:42.854Z" }, - { url = "https://files.pythonhosted.org/packages/19/3a/e54e2c4752160115183a66dc9ee75a153f81f3ab2ba4bf79c3c53b33de34/yarl-1.20.1-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:909313577e9619dcff8c31a0ea2aa0a2a828341d92673015456b3ae492e7317b", size = 384774, upload-time = "2025-06-10T00:44:45.275Z" }, - { url = "https://files.pythonhosted.org/packages/9c/20/200ae86dabfca89060ec6447649f219b4cbd94531e425e50d57e5f5ac330/yarl-1.20.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:793fd0580cb9664548c6b83c63b43c477212c0260891ddf86809e1c06c8b08f1", size = 374252, upload-time = "2025-06-10T00:44:47.31Z" }, - { url = "https://files.pythonhosted.org/packages/83/75/11ee332f2f516b3d094e89448da73d557687f7d137d5a0f48c40ff211487/yarl-1.20.1-cp313-cp313-win32.whl", hash = "sha256:468f6e40285de5a5b3c44981ca3a319a4b208ccc07d526b20b12aeedcfa654b7", size = 81198, upload-time = "2025-06-10T00:44:49.164Z" }, - { url = "https://files.pythonhosted.org/packages/ba/ba/39b1ecbf51620b40ab402b0fc817f0ff750f6d92712b44689c2c215be89d/yarl-1.20.1-cp313-cp313-win_amd64.whl", hash = "sha256:495b4ef2fea40596bfc0affe3837411d6aa3371abcf31aac0ccc4bdd64d4ef5c", size = 86346, upload-time = "2025-06-10T00:44:51.182Z" }, - { url = "https://files.pythonhosted.org/packages/43/c7/669c52519dca4c95153c8ad96dd123c79f354a376346b198f438e56ffeb4/yarl-1.20.1-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:f60233b98423aab21d249a30eb27c389c14929f47be8430efa7dbd91493a729d", size = 138826, upload-time = "2025-06-10T00:44:52.883Z" }, - { url = "https://files.pythonhosted.org/packages/6a/42/fc0053719b44f6ad04a75d7f05e0e9674d45ef62f2d9ad2c1163e5c05827/yarl-1.20.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:6f3eff4cc3f03d650d8755c6eefc844edde99d641d0dcf4da3ab27141a5f8ddf", size = 93217, upload-time = "2025-06-10T00:44:54.658Z" }, - { url = "https://files.pythonhosted.org/packages/4f/7f/fa59c4c27e2a076bba0d959386e26eba77eb52ea4a0aac48e3515c186b4c/yarl-1.20.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:69ff8439d8ba832d6bed88af2c2b3445977eba9a4588b787b32945871c2444e3", size = 92700, upload-time = "2025-06-10T00:44:56.784Z" }, - { url = "https://files.pythonhosted.org/packages/2f/d4/062b2f48e7c93481e88eff97a6312dca15ea200e959f23e96d8ab898c5b8/yarl-1.20.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3cf34efa60eb81dd2645a2e13e00bb98b76c35ab5061a3989c7a70f78c85006d", size = 347644, upload-time = "2025-06-10T00:44:59.071Z" }, - { url = "https://files.pythonhosted.org/packages/89/47/78b7f40d13c8f62b499cc702fdf69e090455518ae544c00a3bf4afc9fc77/yarl-1.20.1-cp313-cp313t-manylinux_2_17_armv7l.manylinux2014_armv7l.manylinux_2_31_armv7l.whl", hash = "sha256:8e0fe9364ad0fddab2688ce72cb7a8e61ea42eff3c7caeeb83874a5d479c896c", size = 323452, upload-time = "2025-06-10T00:45:01.605Z" }, - { url = "https://files.pythonhosted.org/packages/eb/2b/490d3b2dc66f52987d4ee0d3090a147ea67732ce6b4d61e362c1846d0d32/yarl-1.20.1-cp313-cp313t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8f64fbf81878ba914562c672024089e3401974a39767747691c65080a67b18c1", size = 346378, upload-time = "2025-06-10T00:45:03.946Z" }, - { url = "https://files.pythonhosted.org/packages/66/ad/775da9c8a94ce925d1537f939a4f17d782efef1f973039d821cbe4bcc211/yarl-1.20.1-cp313-cp313t-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f6342d643bf9a1de97e512e45e4b9560a043347e779a173250824f8b254bd5ce", size = 353261, upload-time = "2025-06-10T00:45:05.992Z" }, - { url = "https://files.pythonhosted.org/packages/4b/23/0ed0922b47a4f5c6eb9065d5ff1e459747226ddce5c6a4c111e728c9f701/yarl-1.20.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:56dac5f452ed25eef0f6e3c6a066c6ab68971d96a9fb441791cad0efba6140d3", size = 335987, upload-time = "2025-06-10T00:45:08.227Z" }, - { url = "https://files.pythonhosted.org/packages/3e/49/bc728a7fe7d0e9336e2b78f0958a2d6b288ba89f25a1762407a222bf53c3/yarl-1.20.1-cp313-cp313t-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c7d7f497126d65e2cad8dc5f97d34c27b19199b6414a40cb36b52f41b79014be", size = 329361, upload-time = "2025-06-10T00:45:10.11Z" }, - { url = "https://files.pythonhosted.org/packages/93/8f/b811b9d1f617c83c907e7082a76e2b92b655400e61730cd61a1f67178393/yarl-1.20.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:67e708dfb8e78d8a19169818eeb5c7a80717562de9051bf2413aca8e3696bf16", size = 346460, upload-time = "2025-06-10T00:45:12.055Z" }, - { url = "https://files.pythonhosted.org/packages/70/fd/af94f04f275f95da2c3b8b5e1d49e3e79f1ed8b6ceb0f1664cbd902773ff/yarl-1.20.1-cp313-cp313t-musllinux_1_2_armv7l.whl", hash = "sha256:595c07bc79af2494365cc96ddeb772f76272364ef7c80fb892ef9d0649586513", size = 334486, upload-time = "2025-06-10T00:45:13.995Z" }, - { url = "https://files.pythonhosted.org/packages/84/65/04c62e82704e7dd0a9b3f61dbaa8447f8507655fd16c51da0637b39b2910/yarl-1.20.1-cp313-cp313t-musllinux_1_2_i686.whl", hash = "sha256:7bdd2f80f4a7df852ab9ab49484a4dee8030023aa536df41f2d922fd57bf023f", size = 342219, upload-time = "2025-06-10T00:45:16.479Z" }, - { url = "https://files.pythonhosted.org/packages/91/95/459ca62eb958381b342d94ab9a4b6aec1ddec1f7057c487e926f03c06d30/yarl-1.20.1-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:c03bfebc4ae8d862f853a9757199677ab74ec25424d0ebd68a0027e9c639a390", size = 350693, upload-time = "2025-06-10T00:45:18.399Z" }, - { url = "https://files.pythonhosted.org/packages/a6/00/d393e82dd955ad20617abc546a8f1aee40534d599ff555ea053d0ec9bf03/yarl-1.20.1-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:344d1103e9c1523f32a5ed704d576172d2cabed3122ea90b1d4e11fe17c66458", size = 355803, upload-time = "2025-06-10T00:45:20.677Z" }, - { url = "https://files.pythonhosted.org/packages/9e/ed/c5fb04869b99b717985e244fd93029c7a8e8febdfcffa06093e32d7d44e7/yarl-1.20.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:88cab98aa4e13e1ade8c141daeedd300a4603b7132819c484841bb7af3edce9e", size = 341709, upload-time = "2025-06-10T00:45:23.221Z" }, - { url = "https://files.pythonhosted.org/packages/24/fd/725b8e73ac2a50e78a4534ac43c6addf5c1c2d65380dd48a9169cc6739a9/yarl-1.20.1-cp313-cp313t-win32.whl", hash = "sha256:b121ff6a7cbd4abc28985b6028235491941b9fe8fe226e6fdc539c977ea1739d", size = 86591, upload-time = "2025-06-10T00:45:25.793Z" }, - { url = "https://files.pythonhosted.org/packages/94/c3/b2e9f38bc3e11191981d57ea08cab2166e74ea770024a646617c9cddd9f6/yarl-1.20.1-cp313-cp313t-win_amd64.whl", hash = "sha256:541d050a355bbbc27e55d906bc91cb6fe42f96c01413dd0f4ed5a5240513874f", size = 93003, upload-time = "2025-06-10T00:45:27.752Z" }, - { url = "https://files.pythonhosted.org/packages/b4/2d/2345fce04cfd4bee161bf1e7d9cdc702e3e16109021035dbb24db654a622/yarl-1.20.1-py3-none-any.whl", hash = "sha256:83b8eb083fe4683c6115795d9fc1cfaf2cbbefb19b3a1cb68f6527460f483a77", size = 46542, upload-time = "2025-06-10T00:46:07.521Z" }, +sdist = { url = "https://files.pythonhosted.org/packages/b7/9d/4b94a8e6d2b51b599516a5cb88e5bc99b4d8d4583e468057eaa29d5f0918/yarl-1.18.3.tar.gz", hash = "sha256:ac1801c45cbf77b6c99242eeff4fffb5e4e73a800b5c4ad4fc0be5def634d2e1", size = 181062, upload-time = "2024-12-01T20:35:23.292Z" } +wheels = [ + { url = "https://files.pythonhosted.org/packages/d2/98/e005bc608765a8a5569f58e650961314873c8469c333616eb40bff19ae97/yarl-1.18.3-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:7df647e8edd71f000a5208fe6ff8c382a1de8edfbccdbbfe649d263de07d8c34", size = 141458, upload-time = "2024-12-01T20:32:32.604Z" }, + { url = "https://files.pythonhosted.org/packages/df/5d/f8106b263b8ae8a866b46d9be869ac01f9b3fb7f2325f3ecb3df8003f796/yarl-1.18.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:c69697d3adff5aa4f874b19c0e4ed65180ceed6318ec856ebc423aa5850d84f7", size = 94365, upload-time = "2024-12-01T20:32:35.736Z" }, + { url = "https://files.pythonhosted.org/packages/56/3e/d8637ddb9ba69bf851f765a3ee288676f7cf64fb3be13760c18cbc9d10bd/yarl-1.18.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:602d98f2c2d929f8e697ed274fbadc09902c4025c5a9963bf4e9edfc3ab6f7ed", size = 92181, upload-time = "2024-12-01T20:32:37.944Z" }, + { url = "https://files.pythonhosted.org/packages/76/f9/d616a5c2daae281171de10fba41e1c0e2d8207166fc3547252f7d469b4e1/yarl-1.18.3-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c654d5207c78e0bd6d749f6dae1dcbbfde3403ad3a4b11f3c5544d9906969dde", size = 315349, upload-time = "2024-12-01T20:32:40.126Z" }, + { url = "https://files.pythonhosted.org/packages/bb/b4/3ea5e7b6f08f698b3769a06054783e434f6d59857181b5c4e145de83f59b/yarl-1.18.3-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:5094d9206c64181d0f6e76ebd8fb2f8fe274950a63890ee9e0ebfd58bf9d787b", size = 330494, upload-time = "2024-12-01T20:32:41.833Z" }, + { url = "https://files.pythonhosted.org/packages/55/f1/e0fc810554877b1b67420568afff51b967baed5b53bcc983ab164eebf9c9/yarl-1.18.3-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:35098b24e0327fc4ebdc8ffe336cee0a87a700c24ffed13161af80124b7dc8e5", size = 326927, upload-time = "2024-12-01T20:32:43.73Z" }, + { url = "https://files.pythonhosted.org/packages/a9/42/b1753949b327b36f210899f2dd0a0947c0c74e42a32de3f8eb5c7d93edca/yarl-1.18.3-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3236da9272872443f81fedc389bace88408f64f89f75d1bdb2256069a8730ccc", size = 319703, upload-time = "2024-12-01T20:32:46.131Z" }, + { url = "https://files.pythonhosted.org/packages/f0/6d/e87c62dc9635daefb064b56f5c97df55a2e9cc947a2b3afd4fd2f3b841c7/yarl-1.18.3-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e2c08cc9b16f4f4bc522771d96734c7901e7ebef70c6c5c35dd0f10845270bcd", size = 310246, upload-time = "2024-12-01T20:32:48.577Z" }, + { url = "https://files.pythonhosted.org/packages/e3/ef/e2e8d1785cdcbd986f7622d7f0098205f3644546da7919c24b95790ec65a/yarl-1.18.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:80316a8bd5109320d38eef8833ccf5f89608c9107d02d2a7f985f98ed6876990", size = 319730, upload-time = "2024-12-01T20:32:50.209Z" }, + { url = "https://files.pythonhosted.org/packages/fc/15/8723e22345bc160dfde68c4b3ae8b236e868f9963c74015f1bc8a614101c/yarl-1.18.3-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:c1e1cc06da1491e6734f0ea1e6294ce00792193c463350626571c287c9a704db", size = 321681, upload-time = "2024-12-01T20:32:52.498Z" }, + { url = "https://files.pythonhosted.org/packages/86/09/bf764e974f1516efa0ae2801494a5951e959f1610dd41edbfc07e5e0f978/yarl-1.18.3-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:fea09ca13323376a2fdfb353a5fa2e59f90cd18d7ca4eaa1fd31f0a8b4f91e62", size = 324812, upload-time = "2024-12-01T20:32:54.947Z" }, + { url = "https://files.pythonhosted.org/packages/f6/4c/20a0187e3b903c97d857cf0272d687c1b08b03438968ae8ffc50fe78b0d6/yarl-1.18.3-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:e3b9fd71836999aad54084906f8663dffcd2a7fb5cdafd6c37713b2e72be1760", size = 337011, upload-time = "2024-12-01T20:32:57.692Z" }, + { url = "https://files.pythonhosted.org/packages/c9/71/6244599a6e1cc4c9f73254a627234e0dad3883ece40cc33dce6265977461/yarl-1.18.3-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:757e81cae69244257d125ff31663249b3013b5dc0a8520d73694aed497fb195b", size = 338132, upload-time = "2024-12-01T20:33:00.247Z" }, + { url = "https://files.pythonhosted.org/packages/af/f5/e0c3efaf74566c4b4a41cb76d27097df424052a064216beccae8d303c90f/yarl-1.18.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:b1771de9944d875f1b98a745bc547e684b863abf8f8287da8466cf470ef52690", size = 331849, upload-time = "2024-12-01T20:33:02.492Z" }, + { url = "https://files.pythonhosted.org/packages/8a/b8/3d16209c2014c2f98a8f658850a57b716efb97930aebf1ca0d9325933731/yarl-1.18.3-cp310-cp310-win32.whl", hash = "sha256:8874027a53e3aea659a6d62751800cf6e63314c160fd607489ba5c2edd753cf6", size = 84309, upload-time = "2024-12-01T20:33:04.832Z" }, + { url = "https://files.pythonhosted.org/packages/fd/b7/2e9a5b18eb0fe24c3a0e8bae994e812ed9852ab4fd067c0107fadde0d5f0/yarl-1.18.3-cp310-cp310-win_amd64.whl", hash = "sha256:93b2e109287f93db79210f86deb6b9bbb81ac32fc97236b16f7433db7fc437d8", size = 90484, upload-time = "2024-12-01T20:33:06.615Z" }, + { url = "https://files.pythonhosted.org/packages/40/93/282b5f4898d8e8efaf0790ba6d10e2245d2c9f30e199d1a85cae9356098c/yarl-1.18.3-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:8503ad47387b8ebd39cbbbdf0bf113e17330ffd339ba1144074da24c545f0069", size = 141555, upload-time = "2024-12-01T20:33:08.819Z" }, + { url = "https://files.pythonhosted.org/packages/6d/9c/0a49af78df099c283ca3444560f10718fadb8a18dc8b3edf8c7bd9fd7d89/yarl-1.18.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:02ddb6756f8f4517a2d5e99d8b2f272488e18dd0bfbc802f31c16c6c20f22193", size = 94351, upload-time = "2024-12-01T20:33:10.609Z" }, + { url = "https://files.pythonhosted.org/packages/5a/a1/205ab51e148fdcedad189ca8dd587794c6f119882437d04c33c01a75dece/yarl-1.18.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:67a283dd2882ac98cc6318384f565bffc751ab564605959df4752d42483ad889", size = 92286, upload-time = "2024-12-01T20:33:12.322Z" }, + { url = "https://files.pythonhosted.org/packages/ed/fe/88b690b30f3f59275fb674f5f93ddd4a3ae796c2b62e5bb9ece8a4914b83/yarl-1.18.3-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d980e0325b6eddc81331d3f4551e2a333999fb176fd153e075c6d1c2530aa8a8", size = 340649, upload-time = "2024-12-01T20:33:13.842Z" }, + { url = "https://files.pythonhosted.org/packages/07/eb/3b65499b568e01f36e847cebdc8d7ccb51fff716dbda1ae83c3cbb8ca1c9/yarl-1.18.3-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:b643562c12680b01e17239be267bc306bbc6aac1f34f6444d1bded0c5ce438ca", size = 356623, upload-time = "2024-12-01T20:33:15.535Z" }, + { url = "https://files.pythonhosted.org/packages/33/46/f559dc184280b745fc76ec6b1954de2c55595f0ec0a7614238b9ebf69618/yarl-1.18.3-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c017a3b6df3a1bd45b9fa49a0f54005e53fbcad16633870104b66fa1a30a29d8", size = 354007, upload-time = "2024-12-01T20:33:17.518Z" }, + { url = "https://files.pythonhosted.org/packages/af/ba/1865d85212351ad160f19fb99808acf23aab9a0f8ff31c8c9f1b4d671fc9/yarl-1.18.3-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:75674776d96d7b851b6498f17824ba17849d790a44d282929c42dbb77d4f17ae", size = 344145, upload-time = "2024-12-01T20:33:20.071Z" }, + { url = "https://files.pythonhosted.org/packages/94/cb/5c3e975d77755d7b3d5193e92056b19d83752ea2da7ab394e22260a7b824/yarl-1.18.3-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:ccaa3a4b521b780a7e771cc336a2dba389a0861592bbce09a476190bb0c8b4b3", size = 336133, upload-time = "2024-12-01T20:33:22.515Z" }, + { url = "https://files.pythonhosted.org/packages/19/89/b77d3fd249ab52a5c40859815765d35c91425b6bb82e7427ab2f78f5ff55/yarl-1.18.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:2d06d3005e668744e11ed80812e61efd77d70bb7f03e33c1598c301eea20efbb", size = 347967, upload-time = "2024-12-01T20:33:24.139Z" }, + { url = "https://files.pythonhosted.org/packages/35/bd/f6b7630ba2cc06c319c3235634c582a6ab014d52311e7d7c22f9518189b5/yarl-1.18.3-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:9d41beda9dc97ca9ab0b9888cb71f7539124bc05df02c0cff6e5acc5a19dcc6e", size = 346397, upload-time = "2024-12-01T20:33:26.205Z" }, + { url = "https://files.pythonhosted.org/packages/18/1a/0b4e367d5a72d1f095318344848e93ea70da728118221f84f1bf6c1e39e7/yarl-1.18.3-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:ba23302c0c61a9999784e73809427c9dbedd79f66a13d84ad1b1943802eaaf59", size = 350206, upload-time = "2024-12-01T20:33:27.83Z" }, + { url = "https://files.pythonhosted.org/packages/b5/cf/320fff4367341fb77809a2d8d7fe75b5d323a8e1b35710aafe41fdbf327b/yarl-1.18.3-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:6748dbf9bfa5ba1afcc7556b71cda0d7ce5f24768043a02a58846e4a443d808d", size = 362089, upload-time = "2024-12-01T20:33:29.565Z" }, + { url = "https://files.pythonhosted.org/packages/57/cf/aadba261d8b920253204085268bad5e8cdd86b50162fcb1b10c10834885a/yarl-1.18.3-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:0b0cad37311123211dc91eadcb322ef4d4a66008d3e1bdc404808992260e1a0e", size = 366267, upload-time = "2024-12-01T20:33:31.449Z" }, + { url = "https://files.pythonhosted.org/packages/54/58/fb4cadd81acdee6dafe14abeb258f876e4dd410518099ae9a35c88d8097c/yarl-1.18.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:0fb2171a4486bb075316ee754c6d8382ea6eb8b399d4ec62fde2b591f879778a", size = 359141, upload-time = "2024-12-01T20:33:33.79Z" }, + { url = "https://files.pythonhosted.org/packages/9a/7a/4c571597589da4cd5c14ed2a0b17ac56ec9ee7ee615013f74653169e702d/yarl-1.18.3-cp311-cp311-win32.whl", hash = "sha256:61b1a825a13bef4a5f10b1885245377d3cd0bf87cba068e1d9a88c2ae36880e1", size = 84402, upload-time = "2024-12-01T20:33:35.689Z" }, + { url = "https://files.pythonhosted.org/packages/ae/7b/8600250b3d89b625f1121d897062f629883c2f45339623b69b1747ec65fa/yarl-1.18.3-cp311-cp311-win_amd64.whl", hash = "sha256:b9d60031cf568c627d028239693fd718025719c02c9f55df0a53e587aab951b5", size = 91030, upload-time = "2024-12-01T20:33:37.511Z" }, + { url = "https://files.pythonhosted.org/packages/33/85/bd2e2729752ff4c77338e0102914897512e92496375e079ce0150a6dc306/yarl-1.18.3-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:1dd4bdd05407ced96fed3d7f25dbbf88d2ffb045a0db60dbc247f5b3c5c25d50", size = 142644, upload-time = "2024-12-01T20:33:39.204Z" }, + { url = "https://files.pythonhosted.org/packages/ff/74/1178322cc0f10288d7eefa6e4a85d8d2e28187ccab13d5b844e8b5d7c88d/yarl-1.18.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:7c33dd1931a95e5d9a772d0ac5e44cac8957eaf58e3c8da8c1414de7dd27c576", size = 94962, upload-time = "2024-12-01T20:33:40.808Z" }, + { url = "https://files.pythonhosted.org/packages/be/75/79c6acc0261e2c2ae8a1c41cf12265e91628c8c58ae91f5ff59e29c0787f/yarl-1.18.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:25b411eddcfd56a2f0cd6a384e9f4f7aa3efee14b188de13048c25b5e91f1640", size = 92795, upload-time = "2024-12-01T20:33:42.322Z" }, + { url = "https://files.pythonhosted.org/packages/6b/32/927b2d67a412c31199e83fefdce6e645247b4fb164aa1ecb35a0f9eb2058/yarl-1.18.3-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:436c4fc0a4d66b2badc6c5fc5ef4e47bb10e4fd9bf0c79524ac719a01f3607c2", size = 332368, upload-time = "2024-12-01T20:33:43.956Z" }, + { url = "https://files.pythonhosted.org/packages/19/e5/859fca07169d6eceeaa4fde1997c91d8abde4e9a7c018e371640c2da2b71/yarl-1.18.3-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e35ef8683211db69ffe129a25d5634319a677570ab6b2eba4afa860f54eeaf75", size = 342314, upload-time = "2024-12-01T20:33:46.046Z" }, + { url = "https://files.pythonhosted.org/packages/08/75/76b63ccd91c9e03ab213ef27ae6add2e3400e77e5cdddf8ed2dbc36e3f21/yarl-1.18.3-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:84b2deecba4a3f1a398df819151eb72d29bfeb3b69abb145a00ddc8d30094512", size = 341987, upload-time = "2024-12-01T20:33:48.352Z" }, + { url = "https://files.pythonhosted.org/packages/1a/e1/a097d5755d3ea8479a42856f51d97eeff7a3a7160593332d98f2709b3580/yarl-1.18.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:00e5a1fea0fd4f5bfa7440a47eff01d9822a65b4488f7cff83155a0f31a2ecba", size = 336914, upload-time = "2024-12-01T20:33:50.875Z" }, + { url = "https://files.pythonhosted.org/packages/0b/42/e1b4d0e396b7987feceebe565286c27bc085bf07d61a59508cdaf2d45e63/yarl-1.18.3-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d0e883008013c0e4aef84dcfe2a0b172c4d23c2669412cf5b3371003941f72bb", size = 325765, upload-time = "2024-12-01T20:33:52.641Z" }, + { url = "https://files.pythonhosted.org/packages/7e/18/03a5834ccc9177f97ca1bbb245b93c13e58e8225276f01eedc4cc98ab820/yarl-1.18.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:5a3f356548e34a70b0172d8890006c37be92995f62d95a07b4a42e90fba54272", size = 344444, upload-time = "2024-12-01T20:33:54.395Z" }, + { url = "https://files.pythonhosted.org/packages/c8/03/a713633bdde0640b0472aa197b5b86e90fbc4c5bc05b727b714cd8a40e6d/yarl-1.18.3-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:ccd17349166b1bee6e529b4add61727d3f55edb7babbe4069b5764c9587a8cc6", size = 340760, upload-time = "2024-12-01T20:33:56.286Z" }, + { url = "https://files.pythonhosted.org/packages/eb/99/f6567e3f3bbad8fd101886ea0276c68ecb86a2b58be0f64077396cd4b95e/yarl-1.18.3-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:b958ddd075ddba5b09bb0be8a6d9906d2ce933aee81100db289badbeb966f54e", size = 346484, upload-time = "2024-12-01T20:33:58.375Z" }, + { url = "https://files.pythonhosted.org/packages/8e/a9/84717c896b2fc6cb15bd4eecd64e34a2f0a9fd6669e69170c73a8b46795a/yarl-1.18.3-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:c7d79f7d9aabd6011004e33b22bc13056a3e3fb54794d138af57f5ee9d9032cb", size = 359864, upload-time = "2024-12-01T20:34:00.22Z" }, + { url = "https://files.pythonhosted.org/packages/1e/2e/d0f5f1bef7ee93ed17e739ec8dbcb47794af891f7d165fa6014517b48169/yarl-1.18.3-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:4891ed92157e5430874dad17b15eb1fda57627710756c27422200c52d8a4e393", size = 364537, upload-time = "2024-12-01T20:34:03.54Z" }, + { url = "https://files.pythonhosted.org/packages/97/8a/568d07c5d4964da5b02621a517532adb8ec5ba181ad1687191fffeda0ab6/yarl-1.18.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ce1af883b94304f493698b00d0f006d56aea98aeb49d75ec7d98cd4a777e9285", size = 357861, upload-time = "2024-12-01T20:34:05.73Z" }, + { url = "https://files.pythonhosted.org/packages/7d/e3/924c3f64b6b3077889df9a1ece1ed8947e7b61b0a933f2ec93041990a677/yarl-1.18.3-cp312-cp312-win32.whl", hash = "sha256:f91c4803173928a25e1a55b943c81f55b8872f0018be83e3ad4938adffb77dd2", size = 84097, upload-time = "2024-12-01T20:34:07.664Z" }, + { url = "https://files.pythonhosted.org/packages/34/45/0e055320daaabfc169b21ff6174567b2c910c45617b0d79c68d7ab349b02/yarl-1.18.3-cp312-cp312-win_amd64.whl", hash = "sha256:7e2ee16578af3b52ac2f334c3b1f92262f47e02cc6193c598502bd46f5cd1477", size = 90399, upload-time = "2024-12-01T20:34:09.61Z" }, + { url = "https://files.pythonhosted.org/packages/30/c7/c790513d5328a8390be8f47be5d52e141f78b66c6c48f48d241ca6bd5265/yarl-1.18.3-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:90adb47ad432332d4f0bc28f83a5963f426ce9a1a8809f5e584e704b82685dcb", size = 140789, upload-time = "2024-12-01T20:34:11.414Z" }, + { url = "https://files.pythonhosted.org/packages/30/aa/a2f84e93554a578463e2edaaf2300faa61c8701f0898725842c704ba5444/yarl-1.18.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:913829534200eb0f789d45349e55203a091f45c37a2674678744ae52fae23efa", size = 94144, upload-time = "2024-12-01T20:34:13.485Z" }, + { url = "https://files.pythonhosted.org/packages/c6/fc/d68d8f83714b221a85ce7866832cba36d7c04a68fa6a960b908c2c84f325/yarl-1.18.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:ef9f7768395923c3039055c14334ba4d926f3baf7b776c923c93d80195624782", size = 91974, upload-time = "2024-12-01T20:34:15.234Z" }, + { url = "https://files.pythonhosted.org/packages/56/4e/d2563d8323a7e9a414b5b25341b3942af5902a2263d36d20fb17c40411e2/yarl-1.18.3-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:88a19f62ff30117e706ebc9090b8ecc79aeb77d0b1f5ec10d2d27a12bc9f66d0", size = 333587, upload-time = "2024-12-01T20:34:17.358Z" }, + { url = "https://files.pythonhosted.org/packages/25/c9/cfec0bc0cac8d054be223e9f2c7909d3e8442a856af9dbce7e3442a8ec8d/yarl-1.18.3-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:e17c9361d46a4d5addf777c6dd5eab0715a7684c2f11b88c67ac37edfba6c482", size = 344386, upload-time = "2024-12-01T20:34:19.842Z" }, + { url = "https://files.pythonhosted.org/packages/ab/5d/4c532190113b25f1364d25f4c319322e86232d69175b91f27e3ebc2caf9a/yarl-1.18.3-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:1a74a13a4c857a84a845505fd2d68e54826a2cd01935a96efb1e9d86c728e186", size = 345421, upload-time = "2024-12-01T20:34:21.975Z" }, + { url = "https://files.pythonhosted.org/packages/23/d1/6cdd1632da013aa6ba18cee4d750d953104a5e7aac44e249d9410a972bf5/yarl-1.18.3-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:41f7ce59d6ee7741af71d82020346af364949314ed3d87553763a2df1829cc58", size = 339384, upload-time = "2024-12-01T20:34:24.717Z" }, + { url = "https://files.pythonhosted.org/packages/9a/c4/6b3c39bec352e441bd30f432cda6ba51681ab19bb8abe023f0d19777aad1/yarl-1.18.3-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f52a265001d830bc425f82ca9eabda94a64a4d753b07d623a9f2863fde532b53", size = 326689, upload-time = "2024-12-01T20:34:26.886Z" }, + { url = "https://files.pythonhosted.org/packages/23/30/07fb088f2eefdc0aa4fc1af4e3ca4eb1a3aadd1ce7d866d74c0f124e6a85/yarl-1.18.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:82123d0c954dc58db301f5021a01854a85bf1f3bb7d12ae0c01afc414a882ca2", size = 345453, upload-time = "2024-12-01T20:34:29.605Z" }, + { url = "https://files.pythonhosted.org/packages/63/09/d54befb48f9cd8eec43797f624ec37783a0266855f4930a91e3d5c7717f8/yarl-1.18.3-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:2ec9bbba33b2d00999af4631a3397d1fd78290c48e2a3e52d8dd72db3a067ac8", size = 341872, upload-time = "2024-12-01T20:34:31.454Z" }, + { url = "https://files.pythonhosted.org/packages/91/26/fd0ef9bf29dd906a84b59f0cd1281e65b0c3e08c6aa94b57f7d11f593518/yarl-1.18.3-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:fbd6748e8ab9b41171bb95c6142faf068f5ef1511935a0aa07025438dd9a9bc1", size = 347497, upload-time = "2024-12-01T20:34:34.004Z" }, + { url = "https://files.pythonhosted.org/packages/d9/b5/14ac7a256d0511b2ac168d50d4b7d744aea1c1aa20c79f620d1059aab8b2/yarl-1.18.3-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:877d209b6aebeb5b16c42cbb377f5f94d9e556626b1bfff66d7b0d115be88d0a", size = 359981, upload-time = "2024-12-01T20:34:36.624Z" }, + { url = "https://files.pythonhosted.org/packages/ca/b3/d493221ad5cbd18bc07e642894030437e405e1413c4236dd5db6e46bcec9/yarl-1.18.3-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:b464c4ab4bfcb41e3bfd3f1c26600d038376c2de3297760dfe064d2cb7ea8e10", size = 366229, upload-time = "2024-12-01T20:34:38.657Z" }, + { url = "https://files.pythonhosted.org/packages/04/56/6a3e2a5d9152c56c346df9b8fb8edd2c8888b1e03f96324d457e5cf06d34/yarl-1.18.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8d39d351e7faf01483cc7ff7c0213c412e38e5a340238826be7e0e4da450fdc8", size = 360383, upload-time = "2024-12-01T20:34:40.501Z" }, + { url = "https://files.pythonhosted.org/packages/fd/b7/4b3c7c7913a278d445cc6284e59b2e62fa25e72758f888b7a7a39eb8423f/yarl-1.18.3-cp313-cp313-win32.whl", hash = "sha256:61ee62ead9b68b9123ec24bc866cbef297dd266175d53296e2db5e7f797f902d", size = 310152, upload-time = "2024-12-01T20:34:42.814Z" }, + { url = "https://files.pythonhosted.org/packages/f5/d5/688db678e987c3e0fb17867970700b92603cadf36c56e5fb08f23e822a0c/yarl-1.18.3-cp313-cp313-win_amd64.whl", hash = "sha256:578e281c393af575879990861823ef19d66e2b1d0098414855dd367e234f5b3c", size = 315723, upload-time = "2024-12-01T20:34:44.699Z" }, + { url = "https://files.pythonhosted.org/packages/f5/4b/a06e0ec3d155924f77835ed2d167ebd3b211a7b0853da1cf8d8414d784ef/yarl-1.18.3-py3-none-any.whl", hash = "sha256:b57f4f58099328dfb26c6a771d09fb20dbbae81d20cfb66141251ea063bd101b", size = 45109, upload-time = "2024-12-01T20:35:20.834Z" }, ] From cc4b42749da2d4771babe021c8c0e8ebe4dc7181 Mon Sep 17 00:00:00 2001 From: Joseph Kleinhenz Date: Thu, 5 Feb 2026 18:45:47 +0000 Subject: [PATCH 30/32] format + lint --- scratch/analyze_grid_code_distribution.py | 112 ++-- scratch/analyze_trajectory_noise.py | 201 ++++---- scratch/bond_length_issues.py | 90 +++- scratch/check_denoiser.py | 115 +++-- scratch/check_hidden_state.py | 34 +- scratch/check_model_arch.py | 92 ++-- scratch/check_trajectory_length.py | 45 +- scratch/diagnose_mdtraj_dataset.py | 93 ++-- scratch/explore_fake_enhanced_dataset.py | 75 +-- scratch/explore_fake_enhanced_minimal.py | 74 +-- scratch/hydra_trials.py | 33 +- scratch/inspect_model_minimal.py | 11 +- scratch/load_model_state_dict.py | 13 +- scratch/load_wandb_checkpoint.py | 5 +- scratch/organize_data.py | 82 ++- scratch/reorganize_swarm_data.py | 478 +++++++++--------- scratch/test_conditional_denoiser.py | 81 +-- scratch/test_conditional_sampling.py | 95 ++-- scratch/test_conditional_simple.py | 65 +-- scratch/test_conditioners.py | 213 ++++---- scratch/test_denoised_conditioner.py | 250 ++++----- scratch/test_gradient_equivalence.py | 143 +++--- scratch/test_multimeasurement.py | 71 +-- scratch/test_reorganize_swarm_data.py | 157 +++--- scratch/test_repeated_position_dataset.py | 73 +-- scratch/training_prototype.py | 133 ++--- scratch/transformer/convert_spatiotemporal.py | 90 ++-- scratch/transformer/develop_transformer.py | 202 ++++---- scratch/transformer/helpers.py | 100 ++-- scratch/transformer/pooling.py | 113 +++-- scratch/transformer/temporal_transformer.py | 129 +++-- scratch/transformer/test_e3_spatiotemporal.py | 111 ++-- .../test_spatiotemporal_conditioner.py | 301 +++++------ scratch/visualize_fake_enhanced_data.py | 134 ++--- scratch/visualize_noise_denoise.py | 132 ++--- scratch/visualize_traj_data.py | 31 +- src/jamun/cmdline/sample.py | 22 +- src/jamun/cmdline/train.py | 55 +- src/jamun/data/__init__.py | 2 +- src/jamun/data/_mdtraj.py | 58 ++- src/jamun/data/_subsample.py | 43 +- src/jamun/data/_utils.py | 25 +- src/jamun/data/noisy_position_dataset.py | 21 +- src/jamun/data/tests/test_subsample.py | 178 ++++--- src/jamun/model/__init__.py | 4 +- src/jamun/model/arch/__init__.py | 4 +- src/jamun/model/arch/e3conv_conditional.py | 92 ++-- src/jamun/model/arch/spatiotemporal.py | 228 ++++----- src/jamun/model/conditioner_usage_example.py | 171 +++---- src/jamun/model/conditioners/__init__.py | 10 +- src/jamun/model/conditioners/conditioners.py | 145 +++--- src/jamun/model/denoiser_conditional.py | 104 ++-- src/jamun/model/denoiser_multimeasurement.py | 152 ++---- src/jamun/model/denoiser_spiked.py | 100 ++-- src/jamun/model/noise_test.py | 32 +- src/jamun/model/pooling.py | 124 +++-- src/jamun/sampling/_sampler.py | 6 +- src/jamun/sampling/mcmc/__init__.py | 2 +- src/jamun/sampling/mcmc/_splitting.py | 6 +- .../sampling/mcmc/functional/__init__.py | 2 +- .../sampling/mcmc/functional/_splitting.py | 68 ++- .../sampling/walkjump/_single_measurement.py | 60 ++- src/jamun/utils/__init__.py | 7 +- src/jamun/utils/_normalizations.py | 17 +- src/jamun/utils/average_squared_distance.py | 15 +- src/jamun/utils/checkpoint.py | 2 +- src/jamun/utils/data_with_residue_info.py | 9 +- src/jamun/utils/inspect_pretrained.py | 133 +++-- src/jamun/utils/pretrained.py | 155 +++--- src/jamun/utils/pretrained_wrapper.py | 73 ++- src/jamun/utils/sampling_wrapper.py | 27 +- 71 files changed, 3265 insertions(+), 3064 deletions(-) diff --git a/scratch/analyze_grid_code_distribution.py b/scratch/analyze_grid_code_distribution.py index 699a2e2..610d14b 100755 --- a/scratch/analyze_grid_code_distribution.py +++ b/scratch/analyze_grid_code_distribution.py @@ -4,27 +4,29 @@ Creates a histogram showing how many trajectory codes exist for each grid code. """ +import argparse import os import re +from collections import Counter, defaultdict + import matplotlib.pyplot as plt import numpy as np -from collections import defaultdict, Counter -import argparse + def parse_trajectory_files(data_dir): """Parse trajectory files and extract grid codes and traj codes.""" # Pattern to match traj_{grid_code}_{traj_code} - pattern = re.compile(r'^traj_(\d+)_(\d+)') - + pattern = re.compile(r"^traj_(\d+)_(\d+)") + grid_traj_mapping = defaultdict(list) - + # Scan directory for trajectory files if not os.path.exists(data_dir): raise ValueError(f"Directory {data_dir} does not exist") - + files = os.listdir(data_dir) trajectory_files = [] - + for filename in files: match = pattern.match(filename) if match: @@ -32,54 +34,55 @@ def parse_trajectory_files(data_dir): traj_code = int(match.group(2)) grid_traj_mapping[grid_code].append(traj_code) trajectory_files.append(filename) - + print(f"Found {len(trajectory_files)} trajectory files") print(f"Found {len(grid_traj_mapping)} unique grid codes") - + return grid_traj_mapping + def create_histogram(grid_traj_mapping, output_path=None): """Create histogram of trajectory counts per grid code.""" - + if not grid_traj_mapping: print("No trajectory files found!") return - + # Get the full range of grid codes (min to max) all_grid_codes = list(grid_traj_mapping.keys()) min_grid = min(all_grid_codes) max_grid = max(all_grid_codes) - + print(f"Grid code range: {min_grid} to {max_grid}") - + # Create array for all grid codes in range full_range = list(range(min_grid, max_grid + 1)) traj_counts = [] - + for grid_code in full_range: count = len(grid_traj_mapping.get(grid_code, [])) traj_counts.append(count) - + # Print some statistics total_trajs = sum(traj_counts) non_zero_grids = sum(1 for count in traj_counts if count > 0) zero_grids = len(full_range) - non_zero_grids - + print(f"Total trajectories: {total_trajs}") print(f"Grid codes with trajectories: {non_zero_grids}") print(f"Grid codes with no trajectories: {zero_grids}") print(f"Average trajectories per grid code: {total_trajs / len(full_range):.2f}") print(f"Max trajectories for single grid code: {max(traj_counts)}") print(f"Min trajectories for single grid code: {min(traj_counts)}") - + # Create histogram plt.figure(figsize=(12, 6)) - plt.bar(full_range, traj_counts, width=0.8, alpha=0.7, edgecolor='black', linewidth=0.5) - plt.xlabel('Grid Code') - plt.ylabel('Number of Trajectories') - plt.title('Distribution of Trajectories by Grid Code') + plt.bar(full_range, traj_counts, width=0.8, alpha=0.7, edgecolor="black", linewidth=0.5) + plt.xlabel("Grid Code") + plt.ylabel("Number of Trajectories") + plt.title("Distribution of Trajectories by Grid Code") plt.grid(True, alpha=0.3) - + # Add some formatting if len(full_range) > 50: # If too many grid codes, adjust x-axis ticks @@ -87,77 +90,78 @@ def create_histogram(grid_traj_mapping, output_path=None): plt.xticks(full_range[::step], rotation=45) else: plt.xticks(full_range) - + plt.tight_layout() - + # Save plot if output_path: - plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.savefig(output_path, dpi=300, bbox_inches="tight") print(f"Histogram saved to: {output_path}") - + plt.show() - + return full_range, traj_counts + def print_detailed_stats(grid_traj_mapping): """Print detailed statistics about the distribution.""" - + counts = [len(trajs) for trajs in grid_traj_mapping.values()] - - print("\n" + "="*50) + + print("\n" + "=" * 50) print("DETAILED STATISTICS") - print("="*50) - + print("=" * 50) + print(f"Total unique grid codes found: {len(grid_traj_mapping)}") print(f"Total trajectories: {sum(counts)}") - + if counts: print(f"Mean trajectories per grid code: {np.mean(counts):.2f}") print(f"Median trajectories per grid code: {np.median(counts):.2f}") print(f"Std dev trajectories per grid code: {np.std(counts):.2f}") - + # Show distribution of counts count_dist = Counter(counts) - print(f"\nDistribution of trajectory counts:") + print("\nDistribution of trajectory counts:") for count, frequency in sorted(count_dist.items()): print(f" {count} trajectories: {frequency} grid codes") - + # Show some examples - print(f"\nExample grid codes and their trajectory counts:") + print("\nExample grid codes and their trajectory counts:") for i, (grid_code, trajs) in enumerate(sorted(grid_traj_mapping.items())[:10]): print(f" Grid {grid_code}: {len(trajs)} trajectories") - + if len(grid_traj_mapping) > 10: print(f" ... and {len(grid_traj_mapping) - 10} more") + def main(): parser = argparse.ArgumentParser(description="Analyze trajectory distribution by grid codes") - parser.add_argument("--data-dir", - default="/data2/sules/fake_enhanced_data/ALA_ALA", - help="Directory containing trajectory files") - parser.add_argument("--output", - default="scratch/grid_code_histogram.png", - help="Output path for histogram plot") - + parser.add_argument( + "--data-dir", default="/data2/sules/fake_enhanced_data/ALA_ALA", help="Directory containing trajectory files" + ) + parser.add_argument("--output", default="scratch/grid_code_histogram.png", help="Output path for histogram plot") + args = parser.parse_args() - + print(f"Analyzing trajectories in: {args.data_dir}") - + # Parse trajectory files grid_traj_mapping = parse_trajectory_files(args.data_dir) - + if not grid_traj_mapping: print("No trajectory files found matching pattern traj_{grid_code}_{traj_code}") return - + # Print detailed statistics print_detailed_stats(grid_traj_mapping) - + # Create histogram - print(f"\nCreating histogram...") + print("\nCreating histogram...") full_range, traj_counts = create_histogram(grid_traj_mapping, args.output) - - print(f"\nAnalysis complete!") + + print("\nAnalysis complete!") + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/analyze_trajectory_noise.py b/scratch/analyze_trajectory_noise.py index a31e07b..8db0ba1 100644 --- a/scratch/analyze_trajectory_noise.py +++ b/scratch/analyze_trajectory_noise.py @@ -5,36 +5,35 @@ and creates histograms with flexible time point filtering. """ -import os +import argparse import glob -import numpy as np +import os + import matplotlib.pyplot as plt import mdtraj as md -from pathlib import Path -from typing import List, Tuple, Optional -import argparse +import numpy as np -def load_trajectories(traj_dir: str, topology_file: str, max_files: Optional[int] = None) -> List[md.Trajectory]: +def load_trajectories(traj_dir: str, topology_file: str, max_files: int | None = None) -> list[md.Trajectory]: """ Load trajectory files from directory using MDTraj. - + Args: traj_dir: Directory containing .xtc files topology_file: Path to PDB topology file max_files: Maximum number of files to load (None for all) - + Returns: List of MDTraj trajectory objects """ xtc_files = glob.glob(os.path.join(traj_dir, "*.xtc")) xtc_files.sort() - + if max_files is not None: xtc_files = xtc_files[:max_files] - + print(f"Loading {len(xtc_files)} trajectory files...") - + trajectories = [] for i, xtc_file in enumerate(xtc_files): try: @@ -44,7 +43,7 @@ def load_trajectories(traj_dir: str, topology_file: str, max_files: Optional[int print(f"Loaded {i + 1}/{len(xtc_files)} trajectories") except Exception as e: print(f"Warning: Failed to load {xtc_file}: {e}") - + print(f"Successfully loaded {len(trajectories)} trajectories") return trajectories @@ -52,121 +51,125 @@ def load_trajectories(traj_dir: str, topology_file: str, max_files: Optional[int def add_noise_to_trajectory(traj: md.Trajectory, noise_magnitude: float = 0.04) -> md.Trajectory: """ Add Gaussian noise to trajectory coordinates. - + Args: traj: MDTraj trajectory object noise_magnitude: Standard deviation of Gaussian noise to add (in nm) - + Returns: New trajectory with added noise """ # Copy the trajectory to avoid modifying the original noisy_traj = traj.slice(range(traj.n_frames)) - + # Add Gaussian noise to xyz coordinates noise = np.random.normal(0, noise_magnitude, noisy_traj.xyz.shape) noisy_traj.xyz += noise - + return noisy_traj def compute_successive_norms(traj: md.Trajectory) -> np.ndarray: """ Compute norms between successive trajectory points. - + Args: traj: MDTraj trajectory object - + Returns: Array of norms between successive points for each atom """ if traj.n_frames < 2: return np.array([]) - + # Calculate differences between successive frames diff = traj.xyz[1:] - traj.xyz[:-1] # Shape: (n_frames-1, n_atoms, 3) - + # Compute norms for each atom at each time step norms = np.linalg.norm(diff, axis=2) # Shape: (n_frames-1, n_atoms) - + return norms -def compute_norms_for_time_points(traj: md.Trajectory, time_points: List[Tuple[int, int]]) -> np.ndarray: +def compute_norms_for_time_points(traj: md.Trajectory, time_points: list[tuple[int, int]]) -> np.ndarray: """ Compute norms between specific time points. - + Args: traj: MDTraj trajectory object time_points: List of (start_frame, end_frame) tuples - + Returns: Array of norms for specified time point pairs """ norms = [] - + for start_frame, end_frame in time_points: if start_frame < traj.n_frames and end_frame < traj.n_frames: diff = traj.xyz[end_frame] - traj.xyz[start_frame] # Shape: (n_atoms, 3) frame_norms = np.linalg.norm(diff, axis=1) # Shape: (n_atoms,) norms.extend(frame_norms) - + return np.array(norms) -def analyze_trajectories(trajectories: List[md.Trajectory], - noise_magnitude: float = 0.04, - time_point_filter: Optional[List[Tuple[int, int]]] = None) -> Tuple[np.ndarray, np.ndarray]: +def analyze_trajectories( + trajectories: list[md.Trajectory], + noise_magnitude: float = 0.04, + time_point_filter: list[tuple[int, int]] | None = None, +) -> tuple[np.ndarray, np.ndarray]: """ Analyze trajectories by adding noise and computing norms. - + Args: trajectories: List of MDTraj trajectory objects noise_magnitude: Standard deviation of Gaussian noise time_point_filter: Optional list of (start, end) frame pairs to analyze - + Returns: Tuple of (original_norms, noisy_norms) """ original_norms = [] noisy_norms = [] - + print(f"Analyzing {len(trajectories)} trajectories...") - + for i, traj in enumerate(trajectories): if time_point_filter is not None: # Compute norms for specific time points orig_norm = compute_norms_for_time_points(traj, time_point_filter) - + # Add noise and compute norms for same time points noisy_traj = add_noise_to_trajectory(traj, noise_magnitude) noisy_norm = compute_norms_for_time_points(noisy_traj, time_point_filter) else: # Compute successive norms for all time points orig_norm = compute_successive_norms(traj) - + # Add noise and compute successive norms noisy_traj = add_noise_to_trajectory(traj, noise_magnitude) noisy_norm = compute_successive_norms(noisy_traj) - + # Flatten and collect norms original_norms.extend(orig_norm.flatten()) noisy_norms.extend(noisy_norm.flatten()) - + if (i + 1) % 10 == 0: print(f"Analyzed {i + 1}/{len(trajectories)} trajectories") - + return np.array(original_norms), np.array(noisy_norms) -def create_histogram(original_norms: np.ndarray, - noisy_norms: np.ndarray, - title: str = "Norm Differences Between Successive Trajectory Points", - bins: int = 50, - save_path: Optional[str] = None): +def create_histogram( + original_norms: np.ndarray, + noisy_norms: np.ndarray, + title: str = "Norm Differences Between Successive Trajectory Points", + bins: int = 50, + save_path: str | None = None, +): """ Create histogram comparing original and noisy trajectory norms. - + Args: original_norms: Array of norms from original trajectories noisy_norms: Array of norms from noisy trajectories @@ -175,95 +178,105 @@ def create_histogram(original_norms: np.ndarray, save_path: Optional path to save the plot """ plt.figure(figsize=(12, 8)) - + # Create histogram - plt.hist(original_norms, bins=bins, alpha=0.7, label='Original', density=True, color='blue') - plt.hist(noisy_norms, bins=bins, alpha=0.7, label='With Noise (Ļƒ=0.04)', density=True, color='red') - - plt.xlabel('Norm (nm)') - plt.ylabel('Density') + plt.hist(original_norms, bins=bins, alpha=0.7, label="Original", density=True, color="blue") + plt.hist(noisy_norms, bins=bins, alpha=0.7, label="With Noise (Ļƒ=0.04)", density=True, color="red") + + plt.xlabel("Norm (nm)") + plt.ylabel("Density") plt.title(title) plt.legend() plt.grid(True, alpha=0.3) - + # Add statistics orig_mean, orig_std = np.mean(original_norms), np.std(original_norms) noisy_mean, noisy_std = np.mean(noisy_norms), np.std(noisy_norms) - - stats_text = f'Original: Ī¼={orig_mean:.4f}, Ļƒ={orig_std:.4f}\n' - stats_text += f'Noisy: Ī¼={noisy_mean:.4f}, Ļƒ={noisy_std:.4f}' - - plt.text(0.98, 0.98, stats_text, transform=plt.gca().transAxes, - verticalalignment='top', horizontalalignment='right', - bbox=dict(boxstyle='round', facecolor='white', alpha=0.8)) - + + stats_text = f"Original: Ī¼={orig_mean:.4f}, Ļƒ={orig_std:.4f}\n" + stats_text += f"Noisy: Ī¼={noisy_mean:.4f}, Ļƒ={noisy_std:.4f}" + + plt.text( + 0.98, + 0.98, + stats_text, + transform=plt.gca().transAxes, + verticalalignment="top", + horizontalalignment="right", + bbox=dict(boxstyle="round", facecolor="white", alpha=0.8), + ) + plt.tight_layout() - + if save_path: - plt.savefig(save_path, dpi=300, bbox_inches='tight') + plt.savefig(save_path, dpi=300, bbox_inches="tight") print(f"Plot saved to {save_path}") - + plt.show() def main(): - parser = argparse.ArgumentParser(description='Analyze trajectory noise effects') - parser.add_argument('--traj_dir', type=str, - default='/data2/sules/fake_enhanced_data/ALA_ALA_organized/train', - help='Directory containing trajectory files') - parser.add_argument('--topology', type=str, - default='/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb', - help='PDB topology file') - parser.add_argument('--noise_magnitude', type=float, default=0.04, - help='Noise magnitude (standard deviation)') - parser.add_argument('--max_files', type=int, default=50, - help='Maximum number of trajectory files to load') - parser.add_argument('--time_filter', type=str, default=None, - help='Time point filter as "start1,end1;start2,end2" (e.g., "0,1" for initial->next)') - parser.add_argument('--output', type=str, default='trajectory_noise_analysis.png', - help='Output plot filename') - parser.add_argument('--bins', type=int, default=50, - help='Number of histogram bins') - + parser = argparse.ArgumentParser(description="Analyze trajectory noise effects") + parser.add_argument( + "--traj_dir", + type=str, + default="/data2/sules/fake_enhanced_data/ALA_ALA_organized/train", + help="Directory containing trajectory files", + ) + parser.add_argument( + "--topology", + type=str, + default="/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb", + help="PDB topology file", + ) + parser.add_argument("--noise_magnitude", type=float, default=0.04, help="Noise magnitude (standard deviation)") + parser.add_argument("--max_files", type=int, default=50, help="Maximum number of trajectory files to load") + parser.add_argument( + "--time_filter", + type=str, + default=None, + help='Time point filter as "start1,end1;start2,end2" (e.g., "0,1" for initial->next)', + ) + parser.add_argument("--output", type=str, default="trajectory_noise_analysis.png", help="Output plot filename") + parser.add_argument("--bins", type=int, default=50, help="Number of histogram bins") + args = parser.parse_args() - + # Parse time filter if provided time_point_filter = None if args.time_filter: try: - pairs = args.time_filter.split(';') + pairs = args.time_filter.split(";") time_point_filter = [] for pair in pairs: - start, end = map(int, pair.split(',')) + start, end = map(int, pair.split(",")) time_point_filter.append((start, end)) print(f"Using time point filter: {time_point_filter}") except: print("Warning: Invalid time filter format. Using all successive points.") - + # Load trajectories trajectories = load_trajectories(args.traj_dir, args.topology, args.max_files) - + if not trajectories: print("No trajectories loaded. Exiting.") return - + # Analyze trajectories - original_norms, noisy_norms = analyze_trajectories( - trajectories, args.noise_magnitude, time_point_filter - ) - + original_norms, noisy_norms = analyze_trajectories(trajectories, args.noise_magnitude, time_point_filter) + # Create title based on analysis type if time_point_filter: title = f"Norm Differences for Time Points {time_point_filter}" else: title = "Norm Differences Between Successive Trajectory Points" title += f" (Noise Ļƒ={args.noise_magnitude})" - + # Create histogram create_histogram(original_norms, noisy_norms, title, args.bins, args.output) - + # Print summary statistics - print(f"\nSummary Statistics:") + print("\nSummary Statistics:") print(f"Original trajectories: {len(original_norms)} data points") print(f" Mean norm: {np.mean(original_norms):.6f} nm") print(f" Std norm: {np.std(original_norms):.6f} nm") @@ -273,4 +286,4 @@ def main(): if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/bond_length_issues.py b/scratch/bond_length_issues.py index 3baee0c..9f1b2f3 100644 --- a/scratch/bond_length_issues.py +++ b/scratch/bond_length_issues.py @@ -1,13 +1,18 @@ +import matplotlib.pyplot as plt import mdtraj as md import numpy as np -import matplotlib.pyplot as plt + from jamun.metrics._chemical_validity import check_bond_lengths # a. Load trajectory and topology traj_path_conditional = "/data2/sules/jamun-conditional-runs/outputs/sample/dev/runs/2025-08-13_20-22-36/sampler/ALA_ALA/predicted_samples/dcd/joined.dcd" -pdb_path_conditional = "/data2/sules/jamun-conditional-runs/outputs/sample/dev/runs/2025-08-13_20-22-36/sampler/ALA_ALA/topology.pdb" +pdb_path_conditional = ( + "/data2/sules/jamun-conditional-runs/outputs/sample/dev/runs/2025-08-13_20-22-36/sampler/ALA_ALA/topology.pdb" +) traj_path_unconditional = "/data2/sules/jamun-conditional-runs//outputs/sample/dev/runs/2025-08-19_18-56-30/sampler/ALA_ALA/predicted_samples/dcd/joined.dcd" -pdb_path_unconditional = "/data2/sules/jamun-conditional-runs//outputs/sample/dev/runs/2025-08-19_18-56-30/sampler/ALA_ALA/topology.pdb" +pdb_path_unconditional = ( + "/data2/sules/jamun-conditional-runs//outputs/sample/dev/runs/2025-08-19_18-56-30/sampler/ALA_ALA/topology.pdb" +) md_traj_path = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.xtc" md_pdb_path = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" # Load trajectory with topology @@ -23,21 +28,33 @@ bond_length_issues_unconditional = check_bond_lengths(traj_unconditional, tolerance=tolerance) bond_length_issues_md = check_bond_lengths(md_traj, tolerance=tolerance) breakpoint() -print(f"\nBond length analysis (tolerance: {tolerance*100}%):") +print(f"\nBond length analysis (tolerance: {tolerance * 100}%):") print(f"Number of frames analyzed: {len(bond_length_issues_conditional)}") # Convert to numpy array for easier analysis issues_array_conditional = np.array(bond_length_issues_conditional) total_issues_conditional = np.sum(issues_array_conditional) -cumulants_conditional = np.array([np.sum(issues_array_conditional[:i])/np.sum(issues_array_conditional) for i in range(issues_array_conditional.shape[0])]) +cumulants_conditional = np.array( + [ + np.sum(issues_array_conditional[:i]) / np.sum(issues_array_conditional) + for i in range(issues_array_conditional.shape[0]) + ] +) issues_array_unconditional = np.array(bond_length_issues_unconditional) total_issues_unconditional = np.sum(issues_array_unconditional) -cumulants_unconditional = np.array([np.sum(issues_array_unconditional[:i])/np.sum(issues_array_unconditional) for i in range(issues_array_unconditional.shape[0])]) +cumulants_unconditional = np.array( + [ + np.sum(issues_array_unconditional[:i]) / np.sum(issues_array_unconditional) + for i in range(issues_array_unconditional.shape[0]) + ] +) issues_array_md = np.array(bond_length_issues_md) total_issues_md = np.sum(issues_array_md) -cumulants_md = np.array([np.sum(issues_array_md[:i])/np.sum(issues_array_md) for i in range(issues_array_md.shape[0])]) +cumulants_md = np.array( + [np.sum(issues_array_md[:i]) / np.sum(issues_array_md) for i in range(issues_array_md.shape[0])] +) breakpoint() # Create histogram @@ -45,25 +62,58 @@ # Main histogram plt.subplot(1, 2, 1) -plt.hist(issues_array_conditional, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor='black', color='blue', label=f'KALA-JAMUN') -plt.hist(issues_array_unconditional, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor='black', color='red', label=f'JAMUN') -plt.hist(issues_array_md, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor='black', color='green', label=f'Reference MD Trajectory issues') +plt.hist( + issues_array_conditional, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor="black", color="blue", label="KALA-JAMUN" +) +plt.hist( + issues_array_unconditional, bins=10, alpha=0.7, range=(0.0, 1.0), edgecolor="black", color="red", label="JAMUN" +) +plt.hist( + issues_array_md, + bins=10, + alpha=0.7, + range=(0.0, 1.0), + edgecolor="black", + color="green", + label="Reference MD Trajectory issues", +) plt.legend() -plt.xlabel('Fraction of Bonds with Issues', fontsize=14) -plt.ylabel('Number of Frames', fontsize=14) +plt.xlabel("Fraction of Bonds with Issues", fontsize=14) +plt.ylabel("Number of Frames", fontsize=14) plt.ylim(0, 5.0e4) -plt.title(f'Distribution of Bond Length Issues\n(Tolerance: {tolerance*100}%)', fontsize=14) +plt.title(f"Distribution of Bond Length Issues\n(Tolerance: {tolerance * 100}%)", fontsize=14) plt.grid(True, alpha=0.3) # Time series plot plt.subplot(1, 2, 2) -plt.plot(np.linspace(0,1,issues_array_conditional.shape[0]), cumulants_conditional, alpha=0.5, color='blue', label='KALA-JAMUN', linewidth=5) -plt.plot(np.linspace(0,1,issues_array_unconditional.shape[0]), cumulants_unconditional, alpha=0.5, color='red', label='JAMUN', linewidth=5) -plt.plot(np.linspace(0,1,issues_array_md.shape[0]), cumulants_md, alpha=0.5, color='green', label='Reference MD Trajectory', linewidth=5) +plt.plot( + np.linspace(0, 1, issues_array_conditional.shape[0]), + cumulants_conditional, + alpha=0.5, + color="blue", + label="KALA-JAMUN", + linewidth=5, +) +plt.plot( + np.linspace(0, 1, issues_array_unconditional.shape[0]), + cumulants_unconditional, + alpha=0.5, + color="red", + label="JAMUN", + linewidth=5, +) +plt.plot( + np.linspace(0, 1, issues_array_md.shape[0]), + cumulants_md, + alpha=0.5, + color="green", + label="Reference MD Trajectory", + linewidth=5, +) plt.legend() -plt.xlabel('Prop. of trajectory length', fontsize=14) -plt.ylabel('Fraction of issues arising', fontsize=14) -plt.title('Bond Issues Over Time', fontsize=14) +plt.xlabel("Prop. of trajectory length", fontsize=14) +plt.ylabel("Fraction of issues arising", fontsize=14) +plt.title("Bond Issues Over Time", fontsize=14) plt.grid(True, alpha=0.3) -plt.savefig("bond_length_issues_conditional_traj.png") \ No newline at end of file +plt.savefig("bond_length_issues_conditional_traj.png") diff --git a/scratch/check_denoiser.py b/scratch/check_denoiser.py index f546c1b..b8e527a 100644 --- a/scratch/check_denoiser.py +++ b/scratch/check_denoiser.py @@ -1,20 +1,18 @@ # %% -import functools import logging import os import dotenv -import tqdm -from typing import Union + logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) py_logger = logging.getLogger("jamun") import torch + torch.cuda.is_available() torch.set_float32_matmul_precision("high") import e3nn -import e3tools.nn e3nn.set_optimization_defaults(jit_script_fx=False) @@ -24,7 +22,7 @@ import jamun.model import jamun.model.arch -# %% +# %% # dataset dotenv.load_dotenv("../.env", verbose=True) JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") @@ -36,7 +34,7 @@ root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", traj_pattern="^(.*)-traj-arrays.npz", pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], + filter_codes=["AA"], as_iterable=False, subsample=100, max_datasets=1, @@ -48,12 +46,13 @@ batch_size=3, num_workers=2, ) -datamodule.setup('test') +datamodule.setup("test") _, data_batch = next(enumerate(datamodule.test_dataloader())) # %% check paths import sys -project_root = "/homefs/home/sules/jamun" # Or use os.path.abspath("..") if your notebook is in a subdir of jamun + +project_root = "/homefs/home/sules/jamun" # Or use os.path.abspath("..") if your notebook is in a subdir of jamun if project_root not in sys.path: sys.path.insert(0, project_root) @@ -61,10 +60,11 @@ else: py_logger.info(f"'{project_root}' is already in sys.path.") -# %% get configs -from hydra import compose, initialize +# %% get configs import hydra +from hydra import compose, initialize from omegaconf import OmegaConf + # Load the configuration file with initialize(config_path="", job_name="jamun_test"): cfg = compose( @@ -73,7 +73,7 @@ "model.arch._target_=scratch.e3conv_test.E3Conv", # Override to use E3Conv from e3conv_test.py "model._target_=scratch.denoiser_test.Denoiser", # Override to use Denoiser from denoiser_test.py "+model.arch.N_structures=2", # Ensure N_structures is set, defaulting to 2 - ] + ], ) # Log the configuration py_logger.info("Loaded configuration:") @@ -85,15 +85,20 @@ py_logger.info("Attempting to re-instantiate model with updated arch...") # Ensure average_squared_distance is still set correctly if not hasattr(cfg.model, "average_squared_distance") or cfg.model.average_squared_distance is None: - from jamun.utils import compute_average_squared_distance_from_datasets # Ensure import - average_squared_distance = compute_average_squared_distance_from_datasets(datasets['test'], cfg.model.max_radius) + from jamun.utils import compute_average_squared_distance_from_datasets # Ensure import + + average_squared_distance = compute_average_squared_distance_from_datasets( + datasets["test"], cfg.model.max_radius + ) cfg.model.average_squared_distance = average_squared_distance py_logger.info(f"Set cfg.model.average_squared_distance to {cfg.model.average_squared_distance}") # Provide conditioner if needed if not hasattr(cfg.model, "conditioner"): OmegaConf.set_struct(cfg.model, False) # Allow modification cfg.model.conditioner = OmegaConf.create({}) - cfg.model.conditioner._target_ = 'scratch.conditioners.PositionConditioner' # Use the PositionConditioner from scratch.conditioners + cfg.model.conditioner._target_ = ( + "scratch.conditioners.PositionConditioner" # Use the PositionConditioner from scratch.conditioners + ) OmegaConf.set_struct(cfg.model, True) # Lock structure again py_logger.info("Set cfg.model.conditioner to instantiate 'scratch.conditioners.PositionConditioner'") model = hydra.utils.instantiate(cfg.model) @@ -105,6 +110,7 @@ except Exception as e: py_logger.error(f"Error during model re-instantiation: {e}") import traceback + traceback.print_exc() # %% Tests for Denoiser.noise_and_denoise @@ -112,12 +118,12 @@ # Ensure 'model' (your Denoiser instance) and 'data_batch' are available from previous cells. # If 'model' is not the correct Denoiser instance, re-instantiate it as needed. # For example, if you were using the custom_denoiser_model: -# denoiser_to_test = custom_denoiser_model +# denoiser_to_test = custom_denoiser_model # Or if you are using the one from the cfg re-instantiation: -denoiser_to_test = model +denoiser_to_test = model # Make sure data_batch is on the same device as the model -if hasattr(denoiser_to_test, 'device'): +if hasattr(denoiser_to_test, "device"): data_batch = data_batch.to(denoiser_to_test.device) elif next(denoiser_to_test.parameters()).is_cuda: data_batch = data_batch.to(next(denoiser_to_test.parameters()).device) @@ -128,15 +134,15 @@ # %% Tests for Denoiser object (denoiser_to_test) -import torch_geometric.data # For isinstance checks py_logger.info("Starting tests for Denoiser object...") _, data_batch = next(enumerate(datamodule.test_dataloader())) # Ensure data_batch has hidden_state for the tests, matching model's N_structures -if not hasattr(data_batch, 'hidden_state') or \ - not isinstance(data_batch.hidden_state, list) or \ - len(data_batch.hidden_state) != denoiser_to_test.g._orig_mod.N_structures - 1: # Use _orig_mod to access N_structures if g is compiled - +if ( + not hasattr(data_batch, "hidden_state") + or not isinstance(data_batch.hidden_state, list) + or len(data_batch.hidden_state) != denoiser_to_test.g._orig_mod.N_structures - 1 +): # Use _orig_mod to access N_structures if g is compiled n_structures = denoiser_to_test.g._orig_mod.N_structures py_logger.info(f"data_batch.hidden_state is missing or incorrect. Re-creating with {n_structures} structures.") data_batch.hidden_state = [torch.randn_like(data_batch.pos) for _ in range(n_structures)] @@ -151,18 +157,19 @@ py_logger.info("Test 1: Denoiser.noise_and_denoise (align_noisy_input=False)") original_x = data_batch.clone() # sigma_test1 = torch.tensor(0.5, device=denoiser_to_test.device) - sigma = denoiser_to_test.sigma_distribution.sample()*1e-5 - xhat1, y_processed1 = denoiser_to_test.noise_and_denoise(original_x.clone(), sigma, \ - align_noisy_input=True) + sigma = denoiser_to_test.sigma_distribution.sample() * 1e-5 + xhat1, y_processed1 = denoiser_to_test.noise_and_denoise(original_x.clone(), sigma, align_noisy_input=True) # assert isinstance(xhat1, torch_geometric.data.Batch), "xhat1 is not a PyG Batch object" # assert isinstance(y_processed1, torch_geometric.data.Batch), "y_processed1 is not a PyG Batch object" - + assert xhat1.pos.shape == original_x.pos.shape, "xhat1.pos shape mismatch" assert y_processed1.pos.shape == original_x.pos.shape, "y_processed1.pos shape mismatch" - - assert not torch.allclose(y_processed1.pos, original_x.pos), "y_processed1.pos should be different from original x.pos" - + + assert not torch.allclose(y_processed1.pos, original_x.pos), ( + "y_processed1.pos should be different from original x.pos" + ) + assert xhat1.num_graphs == original_x.num_graphs, "xhat1 num_graphs mismatch" assert y_processed1.num_graphs == original_x.num_graphs, "y_processed1 num_graphs mismatch" assert xhat1.num_nodes == original_x.num_nodes, "xhat1 num_nodes mismatch" @@ -171,61 +178,69 @@ assert torch.allclose(y_processed1.batch, original_x.batch), "y_processed1.batch mismatch" # Check hidden_state in y_processed1 (noisy input) - if hasattr(original_x, 'hidden_state') and original_x.hidden_state: - assert hasattr(y_processed1, 'hidden_state') and len(y_processed1.hidden_state) == len(original_x.hidden_state), "y_processed1.hidden_state length mismatch" + if hasattr(original_x, "hidden_state") and original_x.hidden_state: + assert hasattr(y_processed1, "hidden_state") and len(y_processed1.hidden_state) == len( + original_x.hidden_state + ), "y_processed1.hidden_state length mismatch" for i in range(len(original_x.hidden_state)): - assert not torch.allclose(y_processed1.hidden_state[i], original_x.hidden_state[i]), f"y_processed1.hidden_state[{i}] should be different" - + assert not torch.allclose(y_processed1.hidden_state[i], original_x.hidden_state[i]), ( + f"y_processed1.hidden_state[{i}] should be different" + ) + # xhat inherits attributes from the input to xhat_normalized, which is the noisy graph (y_processed1) # So, xhat1 should also have hidden_state if y_processed1 does. - if hasattr(y_processed1, 'hidden_state') and y_processed1.hidden_state: - assert hasattr(xhat1, 'hidden_state') and len(xhat1.hidden_state) == len(y_processed1.hidden_state), "xhat1.hidden_state length mismatch with y_processed1" + if hasattr(y_processed1, "hidden_state") and y_processed1.hidden_state: + assert hasattr(xhat1, "hidden_state") and len(xhat1.hidden_state) == len(y_processed1.hidden_state), ( + "xhat1.hidden_state length mismatch with y_processed1" + ) py_logger.info("Test 1 PASSED.") except Exception as e: py_logger.error(f"Test 1 FAILED: {e}") import traceback + traceback.print_exc() # %% Test 3: Denoiser.training_step try: py_logger.info("Test 3: Denoiser.training_step") # Get a fresh batch for training_step to avoid issues with modified data_batch from other tests - _, train_batch = next(enumerate(datamodule.test_dataloader())) # Using test_dataloader for convenience - + _, train_batch = next(enumerate(datamodule.test_dataloader())) # Using test_dataloader for convenience + # Ensure train_batch has hidden_state - if not hasattr(train_batch, 'hidden_state') or \ - not isinstance(train_batch.hidden_state, list) or \ - len(train_batch.hidden_state) != denoiser_to_test.g._orig_mod.N_structures-1: + if ( + not hasattr(train_batch, "hidden_state") + or not isinstance(train_batch.hidden_state, list) + or len(train_batch.hidden_state) != denoiser_to_test.g._orig_mod.N_structures - 1 + ): n_structures = denoiser_to_test.g._orig_mod.N_structures train_batch.hidden_state = [torch.randn_like(train_batch.pos) for _ in range(n_structures)] - else: + else: py_logger.info(f"train_batch.hidden_state already exists with {len(train_batch.hidden_state)} structures.") train_batch.hidden_state = [hs.to(denoiser_to_test.device) for hs in train_batch.hidden_state] train_batch = train_batch.to(denoiser_to_test.device) - # Manually set align_noisy_input_during_training if not set (it's a param of Denoiser) - if not hasattr(denoiser_to_test, 'align_noisy_input_during_training'): + if not hasattr(denoiser_to_test, "align_noisy_input_during_training"): py_logger.warning("Denoiser missing 'align_noisy_input_during_training', defaulting to False for this test.") - denoiser_to_test.align_noisy_input_during_training = False # Or True, as needed + denoiser_to_test.align_noisy_input_during_training = False # Or True, as needed logs_dict = denoiser_to_test.training_step(train_batch, 0) - + assert isinstance(logs_dict, dict), "Logs is not a dictionary" expected_keys = ["mse", "rmsd", "scaled_rmsd", "loss"] for key in expected_keys: assert key in logs_dict, f"Key '{key}' missing in logs" assert isinstance(logs_dict[key], torch.Tensor), f"Log value for '{key}' is not a tensor" - if key == 'loss': - assert logs_dict[key].ndim == 0, f"logs_dict['loss'] is not scalar, shape: {logs_dict[key].shape}" - else: # mse, rmsd, scaled_rmsd are averaged in training_step's aux_mean - assert logs_dict[key].ndim == 0, f"logs_dict['{key}'] is not scalar, shape: {logs_dict[key].shape}" - + if key == "loss": + assert logs_dict[key].ndim == 0, f"logs_dict['loss'] is not scalar, shape: {logs_dict[key].shape}" + else: # mse, rmsd, scaled_rmsd are averaged in training_step's aux_mean + assert logs_dict[key].ndim == 0, f"logs_dict['{key}'] is not scalar, shape: {logs_dict[key].shape}" py_logger.info(f"Test 3 PASSED. Loss: {logs_dict['mse'].item()}") except Exception as e: py_logger.error(f"Test 3 FAILED: {e}") import traceback + traceback.print_exc() # %% diff --git a/scratch/check_hidden_state.py b/scratch/check_hidden_state.py index 9abfb69..e0c7a6c 100644 --- a/scratch/check_hidden_state.py +++ b/scratch/check_hidden_state.py @@ -1,19 +1,17 @@ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) import logging import os import sys -import torch -import torch_geometric.data -import hydra -from hydra import compose, initialize -from omegaconf import OmegaConf -import wandb -from jamun.utils import find_checkpoint -import jamun.data + import dotenv +import torch from denoiser_test import Denoiser +from hydra import compose, initialize +import jamun.data +from jamun.utils import find_checkpoint # Setup logging logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) @@ -34,17 +32,17 @@ "+model.arch.N_structures=2", # We need at least 2 structures to test hidden state "model.use_torch_compile=false", # Disable torch.compile to avoid ScriptModule issues "+model.conditioner._target_=scratch.conditioners.SelfConditioner", - ] + ], ) # Load checkpoint -checkpoint_path = find_checkpoint(wandb_train_run_path="sule-shashank/jamun/y4rm5488", checkpoint_type='last') -checkpoint = torch.load(checkpoint_path, map_location='cpu', weights_only=False) +checkpoint_path = find_checkpoint(wandb_train_run_path="sule-shashank/jamun/y4rm5488", checkpoint_type="last") +checkpoint = torch.load(checkpoint_path, map_location="cpu", weights_only=False) # Modify hyperparameters to disable torch.compile -if 'hyper_parameters' in checkpoint: - checkpoint['hyper_parameters']['use_torch_compile'] = False - checkpoint['hyper_parameters']['torch_compile_kwargs'] = None +if "hyper_parameters" in checkpoint: + checkpoint["hyper_parameters"]["use_torch_compile"] = False + checkpoint["hyper_parameters"]["torch_compile_kwargs"] = None # Load model with modified hyperparameters breakpoint() @@ -61,7 +59,7 @@ root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", traj_pattern="^(.*)-traj-arrays.npz", pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], + filter_codes=["AA"], as_iterable=False, subsample=100, max_datasets=1, @@ -73,12 +71,12 @@ batch_size=3, num_workers=2, ) -datamodule.setup('test') +datamodule.setup("test") _, test_data = next(enumerate(datamodule.test_dataloader())) test_data = test_data.to(model.device) # Ensure test_data has hidden_state -if not hasattr(test_data, 'hidden_state') or not test_data.hidden_state: +if not hasattr(test_data, "hidden_state") or not test_data.hidden_state: py_logger.info("Adding hidden state to test data") test_data.hidden_state = [torch.randn_like(test_data.pos) for _ in range(model.g.N_structures - 1)] @@ -105,4 +103,4 @@ # Check if noisy positions are different from original noisy_pos_diff = torch.abs(y.pos - test_data.pos).mean() -print(f"Mean absolute difference between original and noisy positions: {noisy_pos_diff.item()}") \ No newline at end of file +print(f"Mean absolute difference between original and noisy positions: {noisy_pos_diff.item()}") diff --git a/scratch/check_model_arch.py b/scratch/check_model_arch.py index 1830cec..849a0df 100644 --- a/scratch/check_model_arch.py +++ b/scratch/check_model_arch.py @@ -4,12 +4,12 @@ import os import dotenv -import tqdm -from typing import Union + logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) py_logger = logging.getLogger("jamun") import torch + torch.cuda.is_available() torch.set_float32_matmul_precision("high") @@ -24,7 +24,7 @@ import jamun.model import jamun.model.arch -# %% +# %% # dataset dotenv.load_dotenv("../.env", verbose=True) JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") @@ -36,7 +36,7 @@ root=f"{JAMUN_DATA_PATH}/capped_diamines/timewarp_splits/train/", traj_pattern="^(.*).xtc", pdb_file="ALA_ALA.pdb", - filter_codes=['ALA_ALA'], + filter_codes=["ALA_ALA"], as_iterable=False, subsample=100, total_lag_time=10, @@ -51,59 +51,61 @@ batch_size=5, num_workers=2, ) -datamodule.setup('test') +datamodule.setup("test") _, data_batch = next(enumerate(datamodule.test_dataloader())) -print(f'Number of hidden states: {len(data_batch.hidden_state)}') -print(f'Size of one hidden state: {data_batch.hidden_state[0].shape}') -# %% test the new e3conv_test class -from e3conv_test import E3Conv +print(f"Number of hidden states: {len(data_batch.hidden_state)}") +print(f"Size of one hidden state: {data_batch.hidden_state[0].shape}") +# %% test the new e3conv_test class import torch_geometric +from e3conv_test import E3Conv from e3tools import radius_graph trial_model = E3Conv( - irreps_out="1x1e", - irreps_hidden="120x0e + 32x1e", - irreps_sh="1x0e + 1x1e", - n_layers=1, - edge_attr_dim=8, - atom_type_embedding_dim=8, - atom_code_embedding_dim=8, - residue_code_embedding_dim=32, - residue_index_embedding_dim=8, - use_residue_information=False, - use_residue_sequence_index=False, - hidden_layer_factory=functools.partial( - e3tools.nn.ConvBlock, - conv=e3tools.nn.Conv, - ), - output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), - N_structures=2 - ) + irreps_out="1x1e", + irreps_hidden="120x0e + 32x1e", + irreps_sh="1x0e + 1x1e", + n_layers=1, + edge_attr_dim=8, + atom_type_embedding_dim=8, + atom_code_embedding_dim=8, + residue_code_embedding_dim=32, + residue_index_embedding_dim=8, + use_residue_information=False, + use_residue_sequence_index=False, + hidden_layer_factory=functools.partial( + e3tools.nn.ConvBlock, + conv=e3tools.nn.Conv, + ), + output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]), + N_structures=2, +) + -# %% postprocess data for plugging into model +# %% postprocess data for plugging into model def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_geometric.data.Batch: radial_edge_index = radius_graph(y.pos, cutoff, batch=y["batch"]) bonded_edge_index = y.edge_index edge_index = torch.cat((radial_edge_index, bonded_edge_index), dim=-1) bond_mask = torch.cat( - ( - torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), - torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), - ), - dim=0, - ) + ( + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), + torch.ones(bonded_edge_index.shape[1], dtype=torch.long, device=y.edge_index.device), + ), + dim=0, + ) y.edge_index = edge_index y.bond_mask = bond_mask - return y + return y + -# add bond mask--do this only once! -bond_mask_exists = hasattr(data_batch, 'bond_mask') and data_batch.bond_mask is not None +# add bond mask--do this only once! +bond_mask_exists = hasattr(data_batch, "bond_mask") and data_batch.bond_mask is not None if not bond_mask_exists: py_logger.info("Adding bond mask to data_batch...") # Ensure data_batch is a torch_geometric.data.Batch if not isinstance(data_batch, torch_geometric.data.Batch): raise TypeError(f"Expected data_batch to be a torch_geometric.data.Batch, got {type(data_batch)}") - + # Add bond mask data_batch = add_bond_mask(data_batch) else: @@ -114,15 +116,15 @@ def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_g raise TypeError(f"Expected data_batch.bond_mask to be a torch.Tensor, got {type(data_batch.bond_mask)}") if data_batch.bond_mask.dtype != torch.long: raise ValueError(f"Expected data_batch.bond_mask to be of dtype torch.long, got {data_batch.bond_mask.dtype}") - + # Ensure edge_index is set correctly - if not hasattr(data_batch, 'edge_index') or data_batch.edge_index is None: + if not hasattr(data_batch, "edge_index") or data_batch.edge_index is None: raise ValueError("data_batch.edge_index is not set. Please ensure it is initialized before adding bond mask.") - + # If everything is fine, we can proceed with the existing bond mask py_logger.info(f"data_batch has {data_batch.num_graphs} graphs and {data_batch.num_nodes} nodes.") # Ensure data_batch is on the same device as the model -if hasattr(trial_model, 'device'): +if hasattr(trial_model, "device"): data_batch = data_batch.to(trial_model.device) elif next(trial_model.parameters()).is_cuda: data_batch = data_batch.to(next(trial_model.parameters()).device) @@ -131,9 +133,9 @@ def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_g py_logger.info("Testing E3Conv model with data_batch...") # Ensure data_batch is on the same device as the model y = data_batch -if hasattr(trial_model, 'device'): +if hasattr(trial_model, "device"): y = y.to(trial_model.device) -elif next(trial_model.parameters()).is_cuda: +elif next(trial_model.parameters()).is_cuda: y = y.to(next(trial_model.parameters()).device) # Run a forward pass through the model try: @@ -143,5 +145,5 @@ def add_bond_mask(y: torch_geometric.data.Batch, cutoff: float = 1.0) -> torch_g except Exception as e: py_logger.error(f"Error during forward pass: {e}") import traceback - traceback.print_exc() + traceback.print_exc() diff --git a/scratch/check_trajectory_length.py b/scratch/check_trajectory_length.py index ffb8e1f..7197083 100644 --- a/scratch/check_trajectory_length.py +++ b/scratch/check_trajectory_length.py @@ -5,74 +5,79 @@ import logging import os -import sys + import mdtraj as md # Set up logging logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) py_logger = logging.getLogger("traj_length_check") + def check_trajectory_lengths(): """Check the length of trajectories in raw xtc files.""" - + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" - + py_logger.info("CHECKING RAW TRAJECTORY LENGTHS") py_logger.info("=" * 50) - + # Get a few trajectory files to sample import glob + xtc_files = glob.glob(os.path.join(dataset_root, "*.xtc")) - + py_logger.info(f"Found {len(xtc_files)} total xtc files") py_logger.info("Checking first 10 files...") - + lengths = [] - + for i, xtc_file in enumerate(xtc_files[:10]): try: # Load trajectory with topology traj = md.load(xtc_file, top=pdb_file) length = traj.n_frames lengths.append(length) - + filename = os.path.basename(xtc_file) - py_logger.info(f"{i+1:2d}. {filename}: {length} frames") - + py_logger.info(f"{i + 1:2d}. {filename}: {length} frames") + except Exception as e: py_logger.error(f"Error loading {xtc_file}: {e}") - + if lengths: py_logger.info("-" * 50) py_logger.info(f"Statistics from {len(lengths)} files:") py_logger.info(f" Minimum length: {min(lengths)} frames") py_logger.info(f" Maximum length: {max(lengths)} frames") - py_logger.info(f" Average length: {sum(lengths)/len(lengths):.1f} frames") + py_logger.info(f" Average length: {sum(lengths) / len(lengths):.1f} frames") py_logger.info(f" All lengths: {sorted(set(lengths))}") - + # Show how subsampling affects this py_logger.info("\nEffect of subsampling (with lag requirements):") original_length = sum(lengths) / len(lengths) - + # The lag requirements mean we need at least total_lag_time * lag_subsample_rate frames # to get any output, and then we lose some frames at the beginning test_cases = [ {"subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1}, - {"subsample": 5, "total_lag_time": 5, "lag_subsample_rate": 1}, + {"subsample": 5, "total_lag_time": 5, "lag_subsample_rate": 1}, {"subsample": 10, "total_lag_time": 5, "lag_subsample_rate": 1}, {"subsample": 20, "total_lag_time": 5, "lag_subsample_rate": 1}, ] - + for params in test_cases: # Estimate how many frames we'd get after subsampling and lag filtering # The algorithm starts from frames that have enough history min_start_frame = (params["total_lag_time"] - 1) * params["lag_subsample_rate"] available_frames = max(0, original_length - min_start_frame) subsampled_frames = available_frames // params["subsample"] - - py_logger.info(f" subsample={params['subsample']:2d}: ~{subsampled_frames:.0f} frames " - f"(from {original_length:.0f} original)") + + py_logger.info( + f" subsample={params['subsample']:2d}: ~{subsampled_frames:.0f} frames " + f"(from {original_length:.0f} original)" + ) + if __name__ == "__main__": - check_trajectory_lengths() \ No newline at end of file + check_trajectory_lengths() diff --git a/scratch/diagnose_mdtraj_dataset.py b/scratch/diagnose_mdtraj_dataset.py index f1c8fab..97916a0 100644 --- a/scratch/diagnose_mdtraj_dataset.py +++ b/scratch/diagnose_mdtraj_dataset.py @@ -5,32 +5,32 @@ import os import sys -import numpy as np + import mdtraj as md -from pathlib import Path # Add the project root to the path -sys.path.insert(0, '/homefs/home/sules/jamun') +sys.path.insert(0, "/homefs/home/sules/jamun") from jamun.data._mdtraj import MDtrajDataset, get_subsampled_indices + def test_trajectory_loading(): """Test basic trajectory loading without any subsampling.""" print("=" * 60) print("TESTING BASIC TRAJECTORY LOADING") print("=" * 60) - + # Use one of your actual trajectory files traj_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.xtc" pdb_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb" - + print(f"Loading trajectory: {traj_file}") print(f"Loading topology: {pdb_file}") - + # Test direct mdtraj loading direct_traj = md.load(traj_file, top=pdb_file) print(f"Direct mdtraj load: {direct_traj.n_frames} frames, {direct_traj.n_atoms} atoms") - + # Test MDtrajDataset without any subsampling parameters print("\n--- Testing MDtrajDataset with default parameters ---") dataset = MDtrajDataset( @@ -38,75 +38,77 @@ def test_trajectory_loading(): traj_files=["swarm_1ps_000_001.xtc"], pdb_file="swarm_1ps_000_001.pdb", label="test_basic", - verbose=True + verbose=True, ) - + print(f"MDtrajDataset length: {len(dataset)}") print(f"Dataset trajectory frames: {dataset.traj.n_frames}") print(f"Dataset trajectory atoms: {dataset.traj.n_atoms}") - + # Check if there are any default parameters being set print(f"Dataset num_frames param: {getattr(dataset, 'num_frames', 'Not set')}") print(f"Dataset start_frame param: {getattr(dataset, 'start_frame', 'Not set')}") print(f"Dataset subsample param: {getattr(dataset, 'subsample', 'Not set')}") + def test_subsampling_behavior(): """Test different subsampling scenarios.""" print("\n" + "=" * 60) print("TESTING SUBSAMPLING BEHAVIOR") print("=" * 60) - + base_params = { "root": "/data2/sules/ALA_ALA_enhanced_full_grid/train", "traj_files": ["swarm_1ps_000_001.xtc"], "pdb_file": "swarm_1ps_000_001.pdb", "label": "test_subsample", - "verbose": True + "verbose": True, } - + # Test 1: Explicit num_frames print("\n--- Test 1: Explicit num_frames ---") dataset1 = MDtrajDataset(**base_params, num_frames=100) print(f"With num_frames=100: {len(dataset1)} frames") - + # Test 2: Explicit num_frames = -1 (should load all) print("\n--- Test 2: num_frames=-1 (load all) ---") dataset2 = MDtrajDataset(**base_params, num_frames=-1) print(f"With num_frames=-1: {len(dataset2)} frames") - + # Test 3: No num_frames specified print("\n--- Test 3: No num_frames specified ---") dataset3 = MDtrajDataset(**base_params) print(f"With default num_frames: {len(dataset3)} frames") - + # Test 4: Explicit subsample print("\n--- Test 4: With subsample=2 ---") dataset4 = MDtrajDataset(**base_params, num_frames=-1, subsample=2) print(f"With subsample=2: {len(dataset4)} frames") + def test_lag_subsampling(): """Test lag-based subsampling behavior.""" print("\n" + "=" * 60) print("TESTING LAG SUBSAMPLING") print("=" * 60) - + # First get the actual trajectory length traj_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.xtc" pdb_file = "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb" direct_traj = md.load(traj_file, top=pdb_file) print(f"Actual trajectory length: {direct_traj.n_frames} frames") - + # Test the get_subsampled_indices function directly print("\n--- Testing get_subsampled_indices function ---") - + test_cases = [ {"N": 50, "subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1}, {"N": 250, "subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1}, {"N": 50, "subsample": 1, "total_lag_time": 2, "lag_subsample_rate": 1}, ] - + for i, params in enumerate(test_cases): - print(f"\nTest case {i+1}: {params}") + print(f"\nTest case {i + 1}: {params}") try: indices = get_subsampled_indices(**params) print(f" Result: {len(indices)} valid starting points") @@ -117,18 +119,18 @@ def test_lag_subsampling(): print(f" Last 3 indices: {indices[-3:]}") except Exception as e: print(f" Error: {e}") - + # Test actual MDtrajDataset with lag parameters print("\n--- Testing MDtrajDataset with lag parameters ---") - + base_params = { "root": "/data2/sules/ALA_ALA_enhanced_full_grid/train", "traj_files": ["swarm_1ps_000_001.xtc"], "pdb_file": "swarm_1ps_000_001.pdb", "label": "test_lag", - "verbose": True + "verbose": True, } - + # Test with different configurations lag_configs = [ {"total_lag_time": 5, "lag_subsample_rate": 1}, @@ -136,14 +138,14 @@ def test_lag_subsampling(): {"total_lag_time": 2, "lag_subsample_rate": 1}, {"total_lag_time": 5, "lag_subsample_rate": 1, "subsample": 1}, ] - + for i, config in enumerate(lag_configs): - print(f"\nLag config {i+1}: {config}") + print(f"\nLag config {i + 1}: {config}") try: dataset = MDtrajDataset(**base_params, **config) print(f" Dataset length: {len(dataset)}") print(f" Trajectory frames: {dataset.traj.n_frames}") - if hasattr(dataset, 'hidden_state') and dataset.hidden_state: + if hasattr(dataset, "hidden_state") and dataset.hidden_state: print(f" Hidden states: {len(dataset.hidden_state)} sets") if len(dataset.hidden_state) > 0: print(f" Hidden state 0 length: {len(dataset.hidden_state[0])}") @@ -151,34 +153,35 @@ def test_lag_subsampling(): except Exception as e: print(f" Error: {e}") + def test_configuration_parsing(): """Test how the configuration parameters are being processed.""" print("\n" + "=" * 60) print("TESTING CONFIGURATION PARAMETER PROCESSING") print("=" * 60) - + # Simulate the exact configuration from your experiment print("Simulating experiment configuration:") config = { "root": "/data2/sules/ALA_ALA_enhanced_full_grid/train", "traj_pattern": "^(.*).xtc", - "pdb_pattern": "^(.*).pdb", + "pdb_pattern": "^(.*).pdb", "subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1, - "max_datasets": 1 # This limits to 1 dataset + "max_datasets": 1, # This limits to 1 dataset } - + print(f"Config: {config}") - + # This should be what parse_datasets_from_directory creates from jamun.data._utils import parse_datasets_from_directory - + print("\nCreating datasets with parse_datasets_from_directory...") try: datasets = parse_datasets_from_directory(**config) print(f"Number of datasets created: {len(datasets)}") - + for i, dataset in enumerate(datasets[:3]): # Show first 3 print(f"\nDataset {i}: {dataset.label()}") print(f" Length: {len(dataset)}") @@ -186,40 +189,44 @@ def test_configuration_parsing(): print(f" Has hidden states: {dataset.hidden_state is not None}") if dataset.hidden_state: print(f" Hidden states count: {len(dataset.hidden_state)}") - + except Exception as e: print(f"Error creating datasets: {e}") import traceback + traceback.print_exc() + def main(): """Run all diagnostic tests.""" print("MDTRAJ DATASET DIAGNOSTIC SCRIPT") print("=" * 60) - + # Check if files exist test_files = [ "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.xtc", - "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb" + "/data2/sules/ALA_ALA_enhanced_full_grid/train/swarm_1ps_000_001.pdb", ] - + for file_path in test_files: if os.path.exists(file_path): print(f"āœ… Found: {file_path}") else: print(f"āŒ Missing: {file_path}") return - + try: test_trajectory_loading() - test_subsampling_behavior() + test_subsampling_behavior() test_lag_subsampling() test_configuration_parsing() - + except Exception as e: print(f"\nāŒ ERROR: {e}") import traceback + traceback.print_exc() + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/explore_fake_enhanced_dataset.py b/scratch/explore_fake_enhanced_dataset.py index b4da660..a882b6a 100644 --- a/scratch/explore_fake_enhanced_dataset.py +++ b/scratch/explore_fake_enhanced_dataset.py @@ -30,19 +30,20 @@ import jamun import jamun.data + def explore_dataset_parameters(): """Explore the fake enhanced dataset with specified parameters.""" - + # Dataset parameters as requested dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized" pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" traj_pattern = "^(.*).xtc" - + # Subsampling parameters as specified by user subsample_rate = 10 # Called 'subsample' in the function total_lag_time = 5 lag_subsample_rate = 1 - + py_logger.info("=" * 60) py_logger.info("EXPLORING FAKE ENHANCED DATASET") py_logger.info("=" * 60) @@ -53,11 +54,11 @@ def explore_dataset_parameters(): py_logger.info(f"Total lag time: {total_lag_time}") py_logger.info(f"Lag subsample rate: {lag_subsample_rate}") py_logger.info("=" * 60) - + # Parse datasets for each split for split in ["train", "val", "test"]: py_logger.info(f"\n--- Exploring {split.upper()} split ---") - + try: datasets = jamun.data.parse_datasets_from_directory( root=f"{dataset_root}/{split}", @@ -68,57 +69,58 @@ def explore_dataset_parameters(): total_lag_time=total_lag_time, lag_subsample_rate=lag_subsample_rate, max_datasets=None, # Load all datasets to get full count - verbose=True + verbose=True, ) - + py_logger.info(f"Number of datasets found: {len(datasets)}") - + if datasets: # Analyze first dataset in detail first_dataset = datasets[0] py_logger.info(f"First dataset label: {first_dataset.label()}") py_logger.info(f"Number of frames in first dataset: {len(first_dataset)}") - + # Check hidden state structure sample_data = first_dataset[0] - if hasattr(sample_data, 'hidden_state') and sample_data.hidden_state: + if hasattr(sample_data, "hidden_state") and sample_data.hidden_state: py_logger.info(f"Hidden state length: {len(sample_data.hidden_state)}") py_logger.info(f"Shape of first hidden state: {sample_data.hidden_state[0].shape}") else: py_logger.info("No hidden state found (expected for regular subsampling)") - + # Calculate total trajectories across all datasets total_frames = sum(len(dataset) for dataset in datasets) py_logger.info(f"Total frames across all datasets: {total_frames}") - + # Estimate original frames before subsampling original_frames_estimate = total_frames * subsample_rate py_logger.info(f"Estimated original frames (before subsampling): {original_frames_estimate}") - + # Show some dataset labels py_logger.info(f"First 5 dataset labels: {[ds.label() for ds in datasets[:5]]}") if len(datasets) > 5: py_logger.info(f"... and {len(datasets) - 5} more datasets") - + except Exception as e: py_logger.error(f"Error processing {split} split: {e}") continue - + py_logger.info("\n" + "=" * 60) py_logger.info("EXPLORATION COMPLETE") py_logger.info("=" * 60) + def compare_with_different_parameters(): """Compare trajectory counts with different subsampling parameters.""" - + py_logger.info("\n" + "=" * 60) py_logger.info("PARAMETER COMPARISON") py_logger.info("=" * 60) - + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" # Just use train split pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" traj_pattern = "^(.*).xtc" - + # Test different parameter combinations test_cases = [ {"subsample": 1, "total_lag_time": None, "lag_subsample_rate": None, "desc": "No subsampling"}, @@ -127,11 +129,13 @@ def compare_with_different_parameters(): {"subsample": 10, "total_lag_time": 3, "lag_subsample_rate": 1, "desc": "Different lag time"}, {"subsample": 5, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "Different subsample rate"}, ] - + for i, params in enumerate(test_cases): - py_logger.info(f"\nTest case {i+1}: {params['desc']}") - py_logger.info(f"Parameters: subsample={params['subsample']}, total_lag_time={params['total_lag_time']}, lag_subsample_rate={params['lag_subsample_rate']}") - + py_logger.info(f"\nTest case {i + 1}: {params['desc']}") + py_logger.info( + f"Parameters: subsample={params['subsample']}, total_lag_time={params['total_lag_time']}, lag_subsample_rate={params['lag_subsample_rate']}" + ) + try: # Limit to first few datasets for speed datasets = jamun.data.parse_datasets_from_directory( @@ -139,34 +143,35 @@ def compare_with_different_parameters(): traj_pattern=traj_pattern, pdb_file=pdb_file, as_iterable=False, - subsample=params['subsample'], - total_lag_time=params['total_lag_time'], - lag_subsample_rate=params['lag_subsample_rate'], + subsample=params["subsample"], + total_lag_time=params["total_lag_time"], + lag_subsample_rate=params["lag_subsample_rate"], max_datasets=3, # Limit for speed - verbose=False + verbose=False, ) - + if datasets: frames_per_dataset = [len(ds) for ds in datasets] total_frames = sum(frames_per_dataset) py_logger.info(f" -> {len(datasets)} datasets, {total_frames} total frames") py_logger.info(f" -> Frames per dataset: {frames_per_dataset}") - + # Check if lagged data exists sample = datasets[0][0] - if hasattr(sample, 'hidden_state') and sample.hidden_state: + if hasattr(sample, "hidden_state") and sample.hidden_state: py_logger.info(f" -> Hidden state length: {len(sample.hidden_state)}") else: - py_logger.info(f" -> No hidden state") + py_logger.info(" -> No hidden state") else: - py_logger.warning(f" -> No datasets found") - + py_logger.warning(" -> No datasets found") + except Exception as e: py_logger.error(f" -> Error: {e}") + if __name__ == "__main__": # First explore with user's specific parameters explore_dataset_parameters() - - # Then compare with different parameters - compare_with_different_parameters() \ No newline at end of file + + # Then compare with different parameters + compare_with_different_parameters() diff --git a/scratch/explore_fake_enhanced_minimal.py b/scratch/explore_fake_enhanced_minimal.py index 872a1b0..b666e37 100644 --- a/scratch/explore_fake_enhanced_minimal.py +++ b/scratch/explore_fake_enhanced_minimal.py @@ -26,24 +26,27 @@ import jamun import jamun.data + def quick_exploration(): """Quick exploration of dataset with limited scope.""" - + # Dataset parameters dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" # Just train for speed pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" traj_pattern = "^(.*).xtc" - + # User's parameters subsample_rate = 10 total_lag_time = 5 lag_subsample_rate = 1 - + py_logger.info("MINIMAL EXPLORATION OF FAKE ENHANCED DATASET") py_logger.info("=" * 60) - py_logger.info(f"Parameters: subsample={subsample_rate}, total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") + py_logger.info( + f"Parameters: subsample={subsample_rate}, total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}" + ) py_logger.info("=" * 60) - + try: # Limit to first 5 datasets for speed py_logger.info("Loading first 5 datasets from train split...") @@ -56,61 +59,65 @@ def quick_exploration(): total_lag_time=total_lag_time, lag_subsample_rate=lag_subsample_rate, max_datasets=5, # LIMIT for speed - verbose=True + verbose=True, ) - + py_logger.info(f"Successfully loaded {len(datasets)} datasets") - + if datasets: # Analyze each dataset total_frames = 0 for i, dataset in enumerate(datasets): frames = len(dataset) total_frames += frames - py_logger.info(f"Dataset {i+1} ('{dataset.label()}'): {frames} frames") - + py_logger.info(f"Dataset {i + 1} ('{dataset.label()}'): {frames} frames") + # Check first dataset in detail if i == 0: sample = dataset[0] py_logger.info(f" Sample position shape: {sample.pos.shape}") - if hasattr(sample, 'hidden_state') and sample.hidden_state: + if hasattr(sample, "hidden_state") and sample.hidden_state: py_logger.info(f" Hidden state: {len(sample.hidden_state)} lag frames") py_logger.info(f" First hidden state shape: {sample.hidden_state[0].shape}") else: - py_logger.info(f" No hidden state found") - + py_logger.info(" No hidden state found") + py_logger.info("-" * 40) py_logger.info(f"TOTAL FRAMES across {len(datasets)} datasets: {total_frames}") py_logger.info(f"Average frames per dataset: {total_frames / len(datasets):.1f}") - + # Extrapolate to estimate full dataset py_logger.info("\nESTIMATING FULL DATASET:") py_logger.info("Assuming all datasets have similar sizes...") - + # We could try to count total datasets but that might be slow # Instead, let's just report what we found - py_logger.info(f"With subsample={subsample_rate}, each dataset gives ~{total_frames/len(datasets):.0f} trajectories") + py_logger.info( + f"With subsample={subsample_rate}, each dataset gives ~{total_frames / len(datasets):.0f} trajectories" + ) py_logger.info(f"Total lag time {total_lag_time} creates hidden states for conditional training") - + else: py_logger.warning("No datasets found!") - + except Exception as e: py_logger.error(f"Error: {e}") import traceback + traceback.print_exc() + def test_different_subsample_rates(): """Test how trajectory count changes with different subsample rates.""" - + py_logger.info("\n" + "=" * 60) py_logger.info("TESTING DIFFERENT SUBSAMPLE RATES") py_logger.info("=" * 60) - + dataset_root = "/data2/sules/fake_enhanced_data/ALA_ALA_organized/train" pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" traj_pattern = "^(.*).xtc" - + # Test different subsample rates (keeping lag parameters constant) test_params = [ {"subsample": 1, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "No subsampling"}, @@ -118,11 +125,13 @@ def test_different_subsample_rates(): {"subsample": 10, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "User's parameters (subsample 10)"}, {"subsample": 20, "total_lag_time": 5, "lag_subsample_rate": 1, "desc": "Subsample 20"}, ] - + for params in test_params: py_logger.info(f"\nTesting: {params['desc']}") - py_logger.info(f" subsample={params['subsample']}, total_lag_time={params['total_lag_time']}, lag_subsample_rate={params['lag_subsample_rate']}") - + py_logger.info( + f" subsample={params['subsample']}, total_lag_time={params['total_lag_time']}, lag_subsample_rate={params['lag_subsample_rate']}" + ) + try: # Load just one dataset for comparison datasets = jamun.data.parse_datasets_from_directory( @@ -130,22 +139,23 @@ def test_different_subsample_rates(): traj_pattern=traj_pattern, pdb_file=pdb_file, as_iterable=False, - subsample=params['subsample'], - total_lag_time=params['total_lag_time'], - lag_subsample_rate=params['lag_subsample_rate'], + subsample=params["subsample"], + total_lag_time=params["total_lag_time"], + lag_subsample_rate=params["lag_subsample_rate"], max_datasets=1, # Just one dataset for speed - verbose=False + verbose=False, ) - + if datasets: frames = len(datasets[0]) py_logger.info(f" -> {frames} frames in first dataset") else: - py_logger.info(f" -> No datasets found") - + py_logger.info(" -> No datasets found") + except Exception as e: py_logger.info(f" -> Error: {e}") + if __name__ == "__main__": quick_exploration() - test_different_subsample_rates() \ No newline at end of file + test_different_subsample_rates() diff --git a/scratch/hydra_trials.py b/scratch/hydra_trials.py index 832011d..65008fa 100644 --- a/scratch/hydra_trials.py +++ b/scratch/hydra_trials.py @@ -1,12 +1,13 @@ -# %% imports -import hydra -from omegaconf import OmegaConf +# %% imports +import logging import os import sys -import dotenv -import logging import traceback +import dotenv +import hydra +from omegaconf import OmegaConf + # --- Basic Setup --- logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) py_logger = logging.getLogger("jamun_sampling_script") @@ -23,38 +24,41 @@ JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") + def print_config_sections(cfg): """Print different sections of the configuration.""" print("\nFull Configuration:") print(OmegaConf.to_yaml(cfg)) - - if hasattr(cfg, 'model'): + + if hasattr(cfg, "model"): print("\nModel Configuration:") print(OmegaConf.to_yaml(cfg.model)) - - if hasattr(cfg, 'init_datasets'): + + if hasattr(cfg, "init_datasets"): print("\nDataset Configuration:") print(OmegaConf.to_yaml(cfg.init_datasets)) - - if hasattr(cfg, 'sampler'): + + if hasattr(cfg, "sampler"): print("\nSampler Configuration:") print(OmegaConf.to_yaml(cfg.sampler)) + def run(cfg): """Main function to run the config loading and printing.""" try: # Print the loaded configuration print_config_sections(cfg) - + # Print specific config values - if hasattr(cfg, 'model'): + if hasattr(cfg, "model"): print("\nModel target:", cfg.model._target_) - if hasattr(cfg, 'sampler'): + if hasattr(cfg, "sampler"): print("Sampler target:", cfg.sampler._target_) except Exception: traceback.print_exc(file=sys.stderr) raise + @hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample") def main(cfg): try: @@ -63,5 +67,6 @@ def main(cfg): traceback.print_exc(file=sys.stderr) raise + if __name__ == "__main__": main() diff --git a/scratch/inspect_model_minimal.py b/scratch/inspect_model_minimal.py index 2d7f0fd..9a94a58 100644 --- a/scratch/inspect_model_minimal.py +++ b/scratch/inspect_model_minimal.py @@ -1,15 +1,18 @@ import sys + sys.path.insert(0, "src") -import torch from jamun.model.denoiser_conditional import Denoiser -print(f"Loading model from checkpoint...") + +print("Loading model from checkpoint...") # Load model -model = Denoiser.load_from_checkpoint("/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-08_22-31-01/checkpoints/last.ckpt") +model = Denoiser.load_from_checkpoint( + "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-08_22-31-01/checkpoints/last.ckpt" +) # Print key info print(f"Model: {type(model).__name__}") print(f"Parameters: {sum(p.numel() for p in model.parameters()):,}") print(f"Conditioner: {model.conditioner}") print(f"Architecture: {model.g}") -print(f"Hyperparams: {dict(model.hparams)}") \ No newline at end of file +print(f"Hyperparams: {dict(model.hparams)}") diff --git a/scratch/load_model_state_dict.py b/scratch/load_model_state_dict.py index 91fc3a3..2470b83 100644 --- a/scratch/load_model_state_dict.py +++ b/scratch/load_model_state_dict.py @@ -1,14 +1,15 @@ -import os import hydra -from omegaconf import OmegaConf import torch +from omegaconf import OmegaConf # Load the config config_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/wandb/run-20250805_042516-yqn9mm7x/files/config.yaml" cfg = OmegaConf.load(config_path) # Find the checkpoint file -checkpoint_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/checkpoints/last.ckpt" +checkpoint_path = ( + "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/checkpoints/last.ckpt" +) # checkpoint_files = [f for f in os.listdir(checkpoint_dir) if f.endswith('.ckpt')] # checkpoint_path = os.path.join(checkpoint_dir, checkpoint_files[0]) # or choose specific one breakpoint() @@ -16,8 +17,8 @@ model = hydra.utils.instantiate(cfg.model) breakpoint() # Load the state dict from checkpoint -checkpoint = torch.load(checkpoint_path, map_location='cpu') -model.load_state_dict(checkpoint['state_dict']) +checkpoint = torch.load(checkpoint_path, map_location="cpu") +model.load_state_dict(checkpoint["state_dict"]) print(f"Loaded model: {type(model).__name__}") -print(f"From checkpoint: {checkpoint_path}") \ No newline at end of file +print(f"From checkpoint: {checkpoint_path}") diff --git a/scratch/load_wandb_checkpoint.py b/scratch/load_wandb_checkpoint.py index 78d8d37..6a52906 100644 --- a/scratch/load_wandb_checkpoint.py +++ b/scratch/load_wandb_checkpoint.py @@ -1,6 +1,8 @@ import os + from jamun.model.denoiser_conditional import Denoiser + def load_model_from_local_checkpoint(checkpoint_dir: str): """ Loads a model from a local checkpoint directory. @@ -32,6 +34,7 @@ def load_model_from_local_checkpoint(checkpoint_dir: str): print(f"An error occurred: {e}") return None + if __name__ == "__main__": checkpoint_dir = "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-06-30_19-07-58/checkpoints" - load_model_from_local_checkpoint(checkpoint_dir) \ No newline at end of file + load_model_from_local_checkpoint(checkpoint_dir) diff --git a/scratch/organize_data.py b/scratch/organize_data.py index cc3e45b..eca00f1 100755 --- a/scratch/organize_data.py +++ b/scratch/organize_data.py @@ -4,19 +4,17 @@ with random 70/10/20 split. """ -import os -import shutil import random +import shutil from pathlib import Path -from typing import List -def get_files_from_directory(source_dir: str) -> List[str]: +def get_files_from_directory(source_dir: str) -> list[str]: """Get all files from the source directory.""" source_path = Path(source_dir) if not source_path.exists(): raise FileNotFoundError(f"Source directory {source_dir} does not exist") - + files = [f for f in source_path.iterdir() if f.is_file()] return files @@ -24,62 +22,54 @@ def get_files_from_directory(source_dir: str) -> List[str]: def create_target_directories(base_dir: str) -> dict: """Create train/val/test directories and return their paths.""" base_path = Path(base_dir) - - directories = { - 'train': base_path / 'train', - 'val': base_path / 'val', - 'test': base_path / 'test' - } - + + directories = {"train": base_path / "train", "val": base_path / "val", "test": base_path / "test"} + for dir_path in directories.values(): dir_path.mkdir(parents=True, exist_ok=True) print(f"Created directory: {dir_path}") - + return directories -def split_files(files: List[Path], train_ratio: float = 0.7, val_ratio: float = 0.1, test_ratio: float = 0.2): +def split_files(files: list[Path], train_ratio: float = 0.7, val_ratio: float = 0.1, test_ratio: float = 0.2): """Split files randomly into train/val/test sets.""" if abs(train_ratio + val_ratio + test_ratio - 1.0) > 1e-6: raise ValueError("Split ratios must sum to 1.0") - + # Shuffle files randomly files_copy = files.copy() random.shuffle(files_copy) - + total_files = len(files_copy) train_count = int(total_files * train_ratio) val_count = int(total_files * val_ratio) - + # Split the files train_files = files_copy[:train_count] - val_files = files_copy[train_count:train_count + val_count] - test_files = files_copy[train_count + val_count:] - - return { - 'train': train_files, - 'val': val_files, - 'test': test_files - } + val_files = files_copy[train_count : train_count + val_count] + test_files = files_copy[train_count + val_count :] + + return {"train": train_files, "val": val_files, "test": test_files} -def copy_files(file_splits: dict, target_dirs: dict, copy_mode: str = 'copy'): +def copy_files(file_splits: dict, target_dirs: dict, copy_mode: str = "copy"): """Copy or move files to their respective directories.""" for split_name, files in file_splits.items(): target_dir = target_dirs[split_name] - + print(f"\n{copy_mode.capitalize()}ing {len(files)} files to {split_name} directory...") - + for file_path in files: target_path = target_dir / file_path.name - - if copy_mode == 'copy': + + if copy_mode == "copy": shutil.copy2(file_path, target_path) - elif copy_mode == 'move': + elif copy_mode == "move": shutil.move(str(file_path), str(target_path)) else: raise ValueError("copy_mode must be either 'copy' or 'move'") - + print(f"Completed {split_name}: {len(files)} files") @@ -87,54 +77,54 @@ def main(): # Configuration source_directory = "/data2/sules/fake_enhanced_data/ALA_ALA" target_base_directory = "/data2/sules/fake_enhanced_data/ALA_ALA_organized" - + # Split ratios train_ratio = 0.8 val_ratio = 0.1 test_ratio = 0.1 - + # Set random seed for reproducibility (optional) random.seed(42) - + print(f"Organizing files from: {source_directory}") print(f"Target directory: {target_base_directory}") print(f"Split ratios - Train: {train_ratio}, Val: {val_ratio}, Test: {test_ratio}") - + try: # Get all files from source directory print("\nGetting files from source directory...") files = get_files_from_directory(source_directory) print(f"Found {len(files)} files") - + if len(files) == 0: print("No files found in source directory. Exiting.") return - + # Create target directories print("\nCreating target directories...") target_dirs = create_target_directories(target_base_directory) - + # Split files randomly print("\nSplitting files randomly...") file_splits = split_files(files, train_ratio, val_ratio, test_ratio) - + # Print split statistics - print(f"\nSplit statistics:") + print("\nSplit statistics:") for split_name, files_in_split in file_splits.items(): percentage = (len(files_in_split) / len(files)) * 100 print(f" {split_name}: {len(files_in_split)} files ({percentage:.1f}%)") - + # Copy files to target directories print("\nCopying files...") - copy_files(file_splits, target_dirs, copy_mode='copy') - + copy_files(file_splits, target_dirs, copy_mode="copy") + print(f"\nāœ… Successfully organized {len(files)} files!") print(f"Files copied to: {target_base_directory}") - + except Exception as e: print(f"āŒ Error: {e}") return 1 if __name__ == "__main__": - exit(main()) \ No newline at end of file + exit(main()) diff --git a/scratch/reorganize_swarm_data.py b/scratch/reorganize_swarm_data.py index e7808c2..d05350c 100644 --- a/scratch/reorganize_swarm_data.py +++ b/scratch/reorganize_swarm_data.py @@ -2,7 +2,7 @@ """ Script to reorganize ALA_ALA swarm results data. -This script takes data from /data/bucket/vanib/ALA_ALA/swarm_results/ and organizes it +This script takes data from /data/bucket/vanib/ALA_ALA/swarm_results/ and organizes it into /data2/sules/ALA_ALA_enhanced/ with train/val splits. Input structure: @@ -15,7 +15,7 @@ - swarm_1ps_{grid_code}_{traj_code}.xtc - swarm_1ps_{grid_code}_{traj_code}.pdb - /data2/sules/ALA_ALA_enhanced/val/ - - swarm_1ps_{grid_code}_{traj_code}.xtc + - swarm_1ps_{grid_code}_{traj_code}.xtc - swarm_1ps_{grid_code}_{traj_code}.pdb The train folder contains 172 randomly sampled grid codes, val contains the remaining 12. @@ -24,16 +24,15 @@ - Val: 12 Ɨ 5 = 60 .xtc files + 60 .pdb files """ +import logging import os -import shutil import random -import logging -from pathlib import Path -from typing import List, Tuple +import shutil try: import mdtraj as md import numpy as np + MDTRAJ_AVAILABLE = True except ImportError: MDTRAJ_AVAILABLE = False @@ -41,118 +40,120 @@ try: from tqdm import tqdm + TQDM_AVAILABLE = True except ImportError: TQDM_AVAILABLE = False logging.warning("tqdm not available. Progress bars will be disabled.") # Configure logging -logging.basicConfig( - level=logging.INFO, - format='%(asctime)s - %(levelname)s - %(message)s' -) +logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s") logger = logging.getLogger(__name__) # Configuration SOURCE_DIR = "/data/bucket/vanib/ALA_ALA/swarms/swarm_results" SINGLE_PDB_FILE = "/data/bucket/vanib/ALA_ALA/swarms/ALA_ALA.pdb" -TRAJECTORY_CODES = ['001', '002', '003', '004', '005'] -LONG_TRAJECTORY_CODES = ['001', '003'] # For 2000ps trajectories +TRAJECTORY_CODES = ["001", "002", "003", "004", "005"] +LONG_TRAJECTORY_CODES = ["001", "003"] # For 2000ps trajectories # Splitting strategies SPLITTING_STRATEGIES = { - 'grid_split': { - 'target_dir': "/data2/sules/ALA_ALA_enhanced_full_swarm", - 'train_size': 172, # Number of grid codes for training - 'description': "Random grid codes split: 172 grids for train, 12 grids for val, all trajectories" + "grid_split": { + "target_dir": "/data2/sules/ALA_ALA_enhanced_full_swarm", + "train_size": 172, # Number of grid codes for training + "description": "Random grid codes split: 172 grids for train, 12 grids for val, all trajectories", + }, + "trajectory_split": { + "target_dir": "/data2/sules/ALA_ALA_enhanced_full_grid", + "train_trajectories": ["001", "002", "003", "004"], # First 4 trajectories for train + "val_trajectories": ["005"], # Last trajectory for val + "description": "All grids split by trajectory: trajectories 001-004 for train, 005 for val", }, - 'trajectory_split': { - 'target_dir': "/data2/sules/ALA_ALA_enhanced_full_grid", - 'train_trajectories': ['001', '002', '003', '004'], # First 4 trajectories for train - 'val_trajectories': ['005'], # Last trajectory for val - 'description': "All grids split by trajectory: trajectories 001-004 for train, 005 for val" + "long_grid_split": { + "target_dir": "/data2/sules/ALA_ALA_enhanced_long", + "trajectory_codes": ["001", "003"], # Only 2000ps trajectories + "train_size": 172, # Number of grid codes for training + "description": "Random grid codes split for 2000ps trajectories: 172 grids for train, 12 grids for val", }, - 'long_grid_split': { - 'target_dir': "/data2/sules/ALA_ALA_enhanced_long", - 'trajectory_codes': ['001', '003'], # Only 2000ps trajectories - 'train_size': 172, # Number of grid codes for training - 'description': "Random grid codes split for 2000ps trajectories: 172 grids for train, 12 grids for val" + "state_split": { + "target_dir": "/data2/sules/ALA_ALA_enhanced_long_state_split", + "trajectory_codes": ["001", "003"], # Only 2000ps trajectories + "phi_range": (0, 100), # First residue phi range for validation set + "psi_range": (-50, 100), # First residue psi range for validation set + "description": "Split by conformational state: trajectories with first residue phi,psi in (0,100)x(-50,100) go to val, others to train", }, - 'state_split': { - 'target_dir': "/data2/sules/ALA_ALA_enhanced_long_state_split", - 'trajectory_codes': ['001', '003'], # Only 2000ps trajectories - 'phi_range': (0, 100), # First residue phi range for validation set - 'psi_range': (-50, 100), # First residue psi range for validation set - 'description': "Split by conformational state: trajectories with first residue phi,psi in (0,100)x(-50,100) go to val, others to train" - } } -def get_all_grid_codes(source_dir: str) -> List[str]: + +def get_all_grid_codes(source_dir: str) -> list[str]: """ Get all grid codes from the source directory. - + Args: source_dir: Path to the swarm results directory - + Returns: List of grid codes (e.g., ['000', '001', '002', ...]) """ grid_codes = [] for item in os.listdir(source_dir): - if os.path.isdir(os.path.join(source_dir, item)) and item.startswith('AA_'): + if os.path.isdir(os.path.join(source_dir, item)) and item.startswith("AA_"): grid_code = item[3:] # Remove 'AA_' prefix grid_codes.append(grid_code) - + grid_codes.sort() logger.info(f"Found {len(grid_codes)} grid codes") return grid_codes -def split_train_val(grid_codes: List[str], train_size: int, random_seed: int = 42) -> Tuple[List[str], List[str]]: + +def split_train_val(grid_codes: list[str], train_size: int, random_seed: int = 42) -> tuple[list[str], list[str]]: """ Randomly split grid codes into train and validation sets. - + Args: grid_codes: List of all grid codes train_size: Number of grid codes for training random_seed: Random seed for reproducibility - + Returns: Tuple of (train_grid_codes, val_grid_codes) """ random.seed(random_seed) shuffled_codes = grid_codes.copy() random.shuffle(shuffled_codes) - + train_codes = shuffled_codes[:train_size] val_codes = shuffled_codes[train_size:] - + logger.info(f"Train set: {len(train_codes)} grid codes") logger.info(f"Val set: {len(val_codes)} grid codes") - + return train_codes, val_codes + def create_target_directories(target_dir: str): """Create target directory structure.""" - train_dir = os.path.join(target_dir, 'train') - val_dir = os.path.join(target_dir, 'val') - + train_dir = os.path.join(target_dir, "train") + val_dir = os.path.join(target_dir, "val") + os.makedirs(train_dir, exist_ok=True) os.makedirs(val_dir, exist_ok=True) - + logger.info(f"Created directories: {train_dir}, {val_dir}") + def copy_files_for_grid_split( source_dir: str, target_dir: str, - grid_codes: List[str], - trajectory_codes: List[str], + grid_codes: list[str], + trajectory_codes: list[str], single_pdb_file: str, split_name: str, - use_2000ps: bool = False + use_2000ps: bool = False, ): """ Copy and rename files for a specific split (train or val). - + Args: source_dir: Source swarm results directory target_dir: Target directory for this split @@ -163,48 +164,48 @@ def copy_files_for_grid_split( use_2000ps: If True, use swarm_2000ps_*.xtc files instead of swarm_1ps_*.xtc """ total_operations = len(grid_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb - + # Create progress bar if TQDM_AVAILABLE: pbar = tqdm( total=total_operations, desc=f"Copying {split_name} files", unit="files", - bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]" + bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]", ) - + copied_files = 0 missing_files = 0 - + if use_2000ps: traj_prefix = "swarm_2000ps" else: traj_prefix = "swarm_1ps" for grid_code in grid_codes: source_grid_dir = os.path.join(source_dir, f"AA_{grid_code}") - + if not os.path.exists(source_grid_dir): logger.warning(f"Source directory does not exist: {source_grid_dir}") # Skip all files for this grid code if TQDM_AVAILABLE: pbar.update(len(trajectory_codes) * 2) continue - + for traj_code in trajectory_codes: # Handle .xtc file source_xtc = os.path.join(source_grid_dir, f"{traj_prefix}_{traj_code}.xtc") target_xtc = os.path.join(target_dir, f"{traj_prefix}_{grid_code}_{traj_code}.xtc") - + if os.path.exists(source_xtc): shutil.copy2(source_xtc, target_xtc) copied_files += 1 else: logger.warning(f"Source file does not exist: {source_xtc}") missing_files += 1 - + if TQDM_AVAILABLE: pbar.update(1) - + # Handle .pdb file (copy the single PDB file) target_pdb = os.path.join(target_dir, f"{traj_prefix}_{grid_code}_{traj_code}.pdb") if os.path.exists(single_pdb_file): @@ -213,26 +214,27 @@ def copy_files_for_grid_split( else: logger.error(f"Single PDB file does not exist: {single_pdb_file}") missing_files += 1 - + if TQDM_AVAILABLE: pbar.update(1) - + if TQDM_AVAILABLE: pbar.close() - + logger.info(f"{split_name}: Completed copying {copied_files} files ({missing_files} missing/failed)") + def copy_files_for_trajectory_split( source_dir: str, target_dir: str, - all_grid_codes: List[str], - trajectory_codes: List[str], + all_grid_codes: list[str], + trajectory_codes: list[str], single_pdb_file: str, - split_name: str + split_name: str, ): """ Copy and rename files for trajectory-based split (all grids, specific trajectories). - + Args: source_dir: Source swarm results directory target_dir: Target directory for this split @@ -242,44 +244,44 @@ def copy_files_for_trajectory_split( split_name: Name of the split for logging """ total_operations = len(all_grid_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb - + # Create progress bar if TQDM_AVAILABLE: pbar = tqdm( total=total_operations, desc=f"Copying {split_name} files", unit="files", - bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]" + bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]", ) - + copied_files = 0 missing_files = 0 - + for grid_code in all_grid_codes: source_grid_dir = os.path.join(source_dir, f"AA_{grid_code}") - + if not os.path.exists(source_grid_dir): logger.warning(f"Source directory does not exist: {source_grid_dir}") # Skip all files for this grid code if TQDM_AVAILABLE: pbar.update(len(trajectory_codes) * 2) continue - + for traj_code in trajectory_codes: # Handle .xtc file source_xtc = os.path.join(source_grid_dir, f"swarm_1ps_{traj_code}.xtc") target_xtc = os.path.join(target_dir, f"swarm_1ps_{grid_code}_{traj_code}.xtc") - + if os.path.exists(source_xtc): shutil.copy2(source_xtc, target_xtc) copied_files += 1 else: logger.warning(f"Source file does not exist: {source_xtc}") missing_files += 1 - + if TQDM_AVAILABLE: pbar.update(1) - + # Handle .pdb file (copy the single PDB file) target_pdb = os.path.join(target_dir, f"swarm_1ps_{grid_code}_{traj_code}.pdb") if os.path.exists(single_pdb_file): @@ -288,65 +290,67 @@ def copy_files_for_trajectory_split( else: logger.error(f"Single PDB file does not exist: {single_pdb_file}") missing_files += 1 - + if TQDM_AVAILABLE: pbar.update(1) - + if TQDM_AVAILABLE: pbar.close() - + logger.info(f"{split_name}: Completed copying {copied_files} files ({missing_files} missing/failed)") + def analyze_trajectory_state(xtc_path: str, pdb_path: str, phi_range: tuple, psi_range: tuple) -> bool: """ Analyze a trajectory to determine if any point has first residue phi,psi in the specified ranges. - + Args: xtc_path: Path to trajectory file pdb_path: Path to topology file phi_range: Tuple of (min, max) for phi angles in degrees psi_range: Tuple of (min, max) for psi angles in degrees - + Returns: True if any point in trajectory has first residue phi,psi in the specified ranges """ if not MDTRAJ_AVAILABLE: logger.error("mdtraj not available, cannot analyze trajectory states") return False - + try: # Load trajectory traj = md.load(xtc_path, top=pdb_path) - traj = traj[:1000] # only use first 1000 frames to avoid memory issues + traj = traj[:1000] # only use first 1000 frames to avoid memory issues # Compute phi and psi angles _, phi_angles = md.compute_phi(traj) _, psi_angles = md.compute_psi(traj) - + # Convert to degrees phi_deg = np.degrees(phi_angles) psi_deg = np.degrees(psi_angles) - + # Check first residue (index 0) for points in specified ranges first_phi_in_range = (phi_deg[:, 0] > phi_range[0]) & (phi_deg[:, 0] < phi_range[1]) first_psi_in_range = (psi_deg[:, 0] > psi_range[0]) & (psi_deg[:, 0] < psi_range[1]) first_residue_in_range = first_phi_in_range & first_psi_in_range - + # Return True if any point is in range has_points_in_range = np.any(first_residue_in_range) n_points_in_range = np.sum(first_residue_in_range) - + logger.debug(f"Trajectory {xtc_path}: {n_points_in_range}/{len(phi_deg)} points in target range") - + return has_points_in_range - + except Exception as e: logger.error(f"Failed to analyze trajectory {xtc_path}: {str(e)}") return False + def test_mdtraj_compatibility(target_dir: str, num_samples: int = 3): """ Test that mdtraj can successfully load swarm + PDB combinations. - + Args: target_dir: Target directory containing train/val splits num_samples: Number of random samples to test from each split @@ -354,234 +358,221 @@ def test_mdtraj_compatibility(target_dir: str, num_samples: int = 3): if not MDTRAJ_AVAILABLE: logger.warning("āš ļø mdtraj not available, skipping trajectory compatibility tests") return True - + logger.info("=== TESTING MDTRAJ COMPATIBILITY ===") - - for split in ['train', 'val']: + + for split in ["train", "val"]: split_dir = os.path.join(target_dir, split) if not os.path.exists(split_dir): continue - + # Get all .xtc files - xtc_files = [f for f in os.listdir(split_dir) if f.endswith('.xtc')] - + xtc_files = [f for f in os.listdir(split_dir) if f.endswith(".xtc")] + if not xtc_files: logger.warning(f"No .xtc files found in {split} directory") continue - + # Sample a few files to test test_files = random.sample(xtc_files, min(num_samples, len(xtc_files))) - + success_count = 0 for xtc_file in test_files: # Get corresponding PDB file - base_name = xtc_file.replace('.xtc', '') + base_name = xtc_file.replace(".xtc", "") pdb_file = f"{base_name}.pdb" - + xtc_path = os.path.join(split_dir, xtc_file) pdb_path = os.path.join(split_dir, pdb_file) - + if not os.path.exists(pdb_path): logger.error(f"Missing PDB file: {pdb_path}") continue - + try: # Try to load trajectory with mdtraj traj = md.load(xtc_path, top=pdb_path) - logger.info(f"āœ… {split}: Successfully loaded {xtc_file} + {pdb_file} " - f"({traj.n_frames} frames, {traj.n_atoms} atoms)") + logger.info( + f"āœ… {split}: Successfully loaded {xtc_file} + {pdb_file} " + f"({traj.n_frames} frames, {traj.n_atoms} atoms)" + ) success_count += 1 - + # Clean up memory del traj - + except Exception as e: logger.error(f"āŒ {split}: Failed to load {xtc_file} + {pdb_file}: {str(e)}") - + logger.info(f"{split}: {success_count}/{len(test_files)} trajectory tests passed") - + logger.info("mdtraj compatibility testing completed") return True + def verify_output(target_dir: str, expected_train_files: int, expected_val_files: int): """ Verify the output directory structure and file counts. - + Args: target_dir: Target directory path expected_train_files: Expected number of files in train directory expected_val_files: Expected number of files in val directory """ - train_dir = os.path.join(target_dir, 'train') - val_dir = os.path.join(target_dir, 'val') - + train_dir = os.path.join(target_dir, "train") + val_dir = os.path.join(target_dir, "val") + train_files = len([f for f in os.listdir(train_dir) if os.path.isfile(os.path.join(train_dir, f))]) val_files = len([f for f in os.listdir(val_dir) if os.path.isfile(os.path.join(val_dir, f))]) - - train_xtc = len([f for f in os.listdir(train_dir) if f.endswith('.xtc')]) - train_pdb = len([f for f in os.listdir(train_dir) if f.endswith('.pdb')]) - val_xtc = len([f for f in os.listdir(val_dir) if f.endswith('.xtc')]) - val_pdb = len([f for f in os.listdir(val_dir) if f.endswith('.pdb')]) - + + train_xtc = len([f for f in os.listdir(train_dir) if f.endswith(".xtc")]) + train_pdb = len([f for f in os.listdir(train_dir) if f.endswith(".pdb")]) + val_xtc = len([f for f in os.listdir(val_dir) if f.endswith(".xtc")]) + val_pdb = len([f for f in os.listdir(val_dir) if f.endswith(".pdb")]) + logger.info("=== VERIFICATION RESULTS ===") logger.info(f"Train directory: {train_files} total files ({train_xtc} .xtc, {train_pdb} .pdb)") logger.info(f"Val directory: {val_files} total files ({val_xtc} .xtc, {val_pdb} .pdb)") logger.info(f"Expected train files: {expected_train_files}") logger.info(f"Expected val files: {expected_val_files}") - + if train_files == expected_train_files and val_files == expected_val_files: logger.info("āœ… File counts match expectations!") - + # Test mdtraj compatibility test_mdtraj_compatibility(target_dir) - + else: logger.warning("āŒ File counts do not match expectations!") -def reorganize_with_long_grid_split(grid_codes: List[str], strategy_config: dict): + +def reorganize_with_long_grid_split(grid_codes: list[str], strategy_config: dict): """Reorganize 2000ps data using grid-based splitting strategy.""" - target_dir = strategy_config['target_dir'] - train_size = strategy_config['train_size'] - trajectory_codes = strategy_config['trajectory_codes'] - + target_dir = strategy_config["target_dir"] + train_size = strategy_config["train_size"] + trajectory_codes = strategy_config["trajectory_codes"] + logger.info(f"Using long grid split strategy: {strategy_config['description']}") - + if len(grid_codes) < train_size: logger.error(f"Not enough grid codes found. Expected at least {train_size}, found {len(grid_codes)}") return - + # Split into train and validation train_codes, val_codes = split_train_val(grid_codes, train_size) - + # Create target directories create_target_directories(target_dir) - + # Copy files for train split (using 2000ps trajectories) logger.info("Copying 2000ps files for train split...") copy_files_for_grid_split( SOURCE_DIR, - os.path.join(target_dir, 'train'), + os.path.join(target_dir, "train"), train_codes, trajectory_codes, SINGLE_PDB_FILE, - 'TRAIN', - use_2000ps=True + "TRAIN", + use_2000ps=True, ) - + # Copy files for val split (using 2000ps trajectories) logger.info("Copying 2000ps files for val split...") copy_files_for_grid_split( SOURCE_DIR, - os.path.join(target_dir, 'val'), + os.path.join(target_dir, "val"), val_codes, trajectory_codes, SINGLE_PDB_FILE, - 'VAL', - use_2000ps=True + "VAL", + use_2000ps=True, ) - + # Verify output expected_train_files = len(train_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb expected_val_files = len(val_codes) * len(trajectory_codes) * 2 - + verify_output(target_dir, expected_train_files, expected_val_files) -def reorganize_with_grid_split(grid_codes: List[str], strategy_config: dict): + +def reorganize_with_grid_split(grid_codes: list[str], strategy_config: dict): """Reorganize data using grid-based splitting strategy.""" - target_dir = strategy_config['target_dir'] - train_size = strategy_config['train_size'] - + target_dir = strategy_config["target_dir"] + train_size = strategy_config["train_size"] + logger.info(f"Using grid split strategy: {strategy_config['description']}") - + if len(grid_codes) < train_size: logger.error(f"Not enough grid codes found. Expected at least {train_size}, found {len(grid_codes)}") return - + # Split into train and validation train_codes, val_codes = split_train_val(grid_codes, train_size) - + # Create target directories create_target_directories(target_dir) - + # Copy files for train split logger.info("Copying files for train split...") copy_files_for_grid_split( - SOURCE_DIR, - os.path.join(target_dir, 'train'), - train_codes, - TRAJECTORY_CODES, - SINGLE_PDB_FILE, - 'TRAIN' + SOURCE_DIR, os.path.join(target_dir, "train"), train_codes, TRAJECTORY_CODES, SINGLE_PDB_FILE, "TRAIN" ) - + # Copy files for val split logger.info("Copying files for val split...") copy_files_for_grid_split( - SOURCE_DIR, - os.path.join(target_dir, 'val'), - val_codes, - TRAJECTORY_CODES, - SINGLE_PDB_FILE, - 'VAL' + SOURCE_DIR, os.path.join(target_dir, "val"), val_codes, TRAJECTORY_CODES, SINGLE_PDB_FILE, "VAL" ) - + # Verify output expected_train_files = len(train_codes) * len(TRAJECTORY_CODES) * 2 # Ɨ2 for .xtc and .pdb expected_val_files = len(val_codes) * len(TRAJECTORY_CODES) * 2 - + verify_output(target_dir, expected_train_files, expected_val_files) -def reorganize_with_trajectory_split(grid_codes: List[str], strategy_config: dict): + +def reorganize_with_trajectory_split(grid_codes: list[str], strategy_config: dict): """Reorganize data using trajectory-based splitting strategy.""" - target_dir = strategy_config['target_dir'] - train_trajectories = strategy_config['train_trajectories'] - val_trajectories = strategy_config['val_trajectories'] - + target_dir = strategy_config["target_dir"] + train_trajectories = strategy_config["train_trajectories"] + val_trajectories = strategy_config["val_trajectories"] + logger.info(f"Using trajectory split strategy: {strategy_config['description']}") - + # Create target directories create_target_directories(target_dir) - + # Copy files for train split (all grids, first 4 trajectories) logger.info("Copying files for train split...") copy_files_for_trajectory_split( - SOURCE_DIR, - os.path.join(target_dir, 'train'), - grid_codes, - train_trajectories, - SINGLE_PDB_FILE, - 'TRAIN' + SOURCE_DIR, os.path.join(target_dir, "train"), grid_codes, train_trajectories, SINGLE_PDB_FILE, "TRAIN" ) - + # Copy files for val split (all grids, last trajectory) logger.info("Copying files for val split...") copy_files_for_trajectory_split( - SOURCE_DIR, - os.path.join(target_dir, 'val'), - grid_codes, - val_trajectories, - SINGLE_PDB_FILE, - 'VAL' + SOURCE_DIR, os.path.join(target_dir, "val"), grid_codes, val_trajectories, SINGLE_PDB_FILE, "VAL" ) - + # Verify output expected_train_files = len(grid_codes) * len(train_trajectories) * 2 # Ɨ2 for .xtc and .pdb expected_val_files = len(grid_codes) * len(val_trajectories) * 2 - + verify_output(target_dir, expected_train_files, expected_val_files) + def copy_files_for_state_split( source_dir: str, target_dir: str, - all_grid_codes: List[str], - trajectory_codes: List[str], + all_grid_codes: list[str], + trajectory_codes: list[str], single_pdb_file: str, phi_range: tuple, - psi_range: tuple + psi_range: tuple, ): """ Copy and organize files based on conformational state analysis. - + Args: source_dir: Source swarm results directory target_dir: Target directory with train/val subdirectories @@ -594,51 +585,51 @@ def copy_files_for_state_split( if not MDTRAJ_AVAILABLE: logger.error("mdtraj not available, cannot perform state-based splitting") return - - train_dir = os.path.join(target_dir, 'train') - val_dir = os.path.join(target_dir, 'val') - + + train_dir = os.path.join(target_dir, "train") + val_dir = os.path.join(target_dir, "val") + total_operations = len(all_grid_codes) * len(trajectory_codes) * 2 # Ɨ2 for .xtc and .pdb - + # Create progress bar if TQDM_AVAILABLE: pbar = tqdm( total=total_operations, desc="Analyzing and copying files", unit="files", - bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]" + bar_format="{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, {rate_fmt}]", ) - + copied_train = 0 copied_val = 0 missing_files = 0 analysis_errors = 0 - + for grid_code in all_grid_codes: source_grid_dir = os.path.join(source_dir, f"AA_{grid_code}") - + if not os.path.exists(source_grid_dir): logger.warning(f"Source directory does not exist: {source_grid_dir}") # Skip all files for this grid code if TQDM_AVAILABLE: pbar.update(len(trajectory_codes) * 2) continue - + for traj_code in trajectory_codes: # Handle .xtc file - need to analyze it first source_xtc = os.path.join(source_grid_dir, f"swarm_2000ps_{traj_code}.xtc") - + if not os.path.exists(source_xtc): logger.warning(f"Source file does not exist: {source_xtc}") missing_files += 1 if TQDM_AVAILABLE: pbar.update(2) # Skip both .xtc and .pdb continue - + # Analyze trajectory to determine train/val split try: goes_to_val = analyze_trajectory_state(source_xtc, single_pdb_file, phi_range, psi_range) - + if goes_to_val: target_xtc = os.path.join(val_dir, f"swarm_2000ps_{grid_code}_{traj_code}.xtc") target_pdb = os.path.join(val_dir, f"swarm_2000ps_{grid_code}_{traj_code}.pdb") @@ -649,122 +640,119 @@ def copy_files_for_state_split( target_pdb = os.path.join(train_dir, f"swarm_2000ps_{grid_code}_{traj_code}.pdb") split_name = "TRAIN" copied_train += 1 - + # Copy .xtc file shutil.copy2(source_xtc, target_xtc) logger.debug(f"Copied {source_xtc} to {split_name}") - + except Exception as e: logger.error(f"Failed to analyze trajectory {source_xtc}: {str(e)}") analysis_errors += 1 if TQDM_AVAILABLE: pbar.update(2) # Skip both .xtc and .pdb continue - + if TQDM_AVAILABLE: pbar.update(1) - + # Handle .pdb file (copy the single PDB file) if os.path.exists(single_pdb_file): shutil.copy2(single_pdb_file, target_pdb) else: logger.error(f"Single PDB file does not exist: {single_pdb_file}") missing_files += 1 - + if TQDM_AVAILABLE: pbar.update(1) - + if TQDM_AVAILABLE: pbar.close() - - logger.info(f"State split completed:") + + logger.info("State split completed:") logger.info(f" TRAIN: {copied_train} trajectories") logger.info(f" VAL: {copied_val} trajectories") logger.info(f" Missing files: {missing_files}") logger.info(f" Analysis errors: {analysis_errors}") -def reorganize_with_state_split(grid_codes: List[str], strategy_config: dict): + +def reorganize_with_state_split(grid_codes: list[str], strategy_config: dict): """Reorganize data using conformational state-based splitting strategy.""" - target_dir = strategy_config['target_dir'] - trajectory_codes = strategy_config['trajectory_codes'] - phi_range = strategy_config['phi_range'] - psi_range = strategy_config['psi_range'] - + target_dir = strategy_config["target_dir"] + trajectory_codes = strategy_config["trajectory_codes"] + phi_range = strategy_config["phi_range"] + psi_range = strategy_config["psi_range"] + logger.info(f"Using state split strategy: {strategy_config['description']}") logger.info(f"Target ranges: phi {phi_range}, psi {psi_range}") logger.info(f"Using trajectory codes: {trajectory_codes}") - + if not MDTRAJ_AVAILABLE: logger.error("mdtraj not available, cannot perform state-based splitting") return - + # Create target directories create_target_directories(target_dir) - + # Copy and split files based on conformational state logger.info("Analyzing trajectories and copying files...") copy_files_for_state_split( - SOURCE_DIR, - target_dir, - grid_codes, - trajectory_codes, - SINGLE_PDB_FILE, - phi_range, - psi_range + SOURCE_DIR, target_dir, grid_codes, trajectory_codes, SINGLE_PDB_FILE, phi_range, psi_range ) - + # Note: We can't predict exact file counts since they depend on trajectory analysis logger.info("State-based reorganization completed!") -def main(strategy: str = 'trajectory_split'): + +def main(strategy: str = "trajectory_split"): """ Main function to reorganize the swarm data. - + Args: strategy: Either 'grid_split' or 'trajectory_split' """ logger.info("Starting swarm data reorganization...") - + # Validate input paths if not os.path.exists(SOURCE_DIR): logger.error(f"Source directory does not exist: {SOURCE_DIR}") return - + if not os.path.exists(SINGLE_PDB_FILE): logger.error(f"Single PDB file does not exist: {SINGLE_PDB_FILE}") return - + if strategy not in SPLITTING_STRATEGIES: logger.error(f"Invalid strategy: {strategy}. Choose from {list(SPLITTING_STRATEGIES.keys())}") return - + # Get all grid codes grid_codes = get_all_grid_codes(SOURCE_DIR) strategy_config = SPLITTING_STRATEGIES[strategy] - + # Execute the appropriate strategy - if strategy == 'grid_split': + if strategy == "grid_split": reorganize_with_grid_split(grid_codes, strategy_config) - elif strategy == 'trajectory_split': + elif strategy == "trajectory_split": reorganize_with_trajectory_split(grid_codes, strategy_config) - elif strategy == 'long_grid_split': + elif strategy == "long_grid_split": reorganize_with_long_grid_split(grid_codes, strategy_config) - elif strategy == 'state_split': + elif strategy == "state_split": reorganize_with_state_split(grid_codes, strategy_config) - + logger.info("Swarm data reorganization completed!") + if __name__ == "__main__": import sys - + # Default to trajectory_split for the new requirement - strategy = 'trajectory_split' - + strategy = "trajectory_split" + # Allow command line argument to choose strategy if len(sys.argv) > 1: strategy = sys.argv[1] - + print(f"Running reorganization with strategy: {strategy}") print(f"Description: {SPLITTING_STRATEGIES[strategy]['description']}") - - main(strategy) \ No newline at end of file + + main(strategy) diff --git a/scratch/test_conditional_denoiser.py b/scratch/test_conditional_denoiser.py index 6a01769..999ce1e 100644 --- a/scratch/test_conditional_denoiser.py +++ b/scratch/test_conditional_denoiser.py @@ -1,51 +1,51 @@ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys +import math import os +import sys + +import dotenv import hydra -from omegaconf import OmegaConf import torch -import torch_geometric -from jamun.hydra import instantiate_dict_cfg -import pdb -import jamun +from omegaconf import OmegaConf + from jamun.utils import compute_average_squared_distance_from_datasets -from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets from jamun.utils._normalizations import normalization_factors -import math +from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets + def compute_radial_cutoff(max_radius: float, average_squared_distance: float, sigma: float, D: int = 3) -> float: """ Compute radial cutoff using the same formula as the denoiser. - + This replicates the computation from denoiser_conditional.py: radial_cutoff = effective_radial_cutoff(sigma) / c_in where: - effective_radial_cutoff = sqrt(max_radiusĀ² + 6ĻƒĀ²) - c_in = 1.0 / sqrt(average_squared_distance + 2DĻƒĀ²) - + Args: max_radius: Maximum radius parameter average_squared_distance: Average squared distance from dataset sigma: Noise level D: Dimensionality (default 3 for 3D coordinates) - + Returns: Computed radial cutoff """ # Effective radial cutoff based on noise level effective_radial_cutoff = math.sqrt(max_radius**2 + 6 * sigma**2) - + # JAMUN normalization factor c_in A = average_squared_distance B = 2 * D * sigma**2 c_in = 1.0 / math.sqrt(A + B) - + # Final radial cutoff radial_cutoff = effective_radial_cutoff / c_in - - print(f"Radial cutoff computation:") + + print("Radial cutoff computation:") print(f" max_radius: {max_radius}") print(f" average_squared_distance: {average_squared_distance}") print(f" sigma: {sigma}") @@ -53,20 +53,22 @@ def compute_radial_cutoff(max_radius: float, average_squared_distance: float, si print(f" effective_radial_cutoff: {effective_radial_cutoff}") print(f" c_in: {c_in}") print(f" final radial_cutoff: {radial_cutoff}") - + return radial_cutoff -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") -project_root = "/homefs/home/sules/jamun" # Adjust if necessary +project_root = "/homefs/home/sules/jamun" # Adjust if necessary if project_root not in sys.path: sys.path.insert(0, project_root) print(f"Added '{project_root}' to sys.path for module discovery.") else: print(f"'{project_root}' is already in sys.path.") + def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the average squared distance for normalization from the data.""" datamodule = hydra.utils.instantiate(cfg.data.datamodule) @@ -76,16 +78,17 @@ def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) return average_squared_distance + def compute_temporal_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the temporal average squared distance for normalization from the data.""" datamodule = hydra.utils.instantiate(cfg.data.datamodule) datamodule.setup("compute_normalization") train_datasets = datamodule.datasets["train"] - + average_squared_distance = compute_temporal_average_squared_distance_from_datasets( - train_datasets, + train_datasets, num_samples=100, # Use reasonable number of samples - verbose=True + verbose=True, ) return average_squared_distance @@ -98,13 +101,13 @@ def main(cfg): # cfg.model.sigma_distribution.sigma = 0.04 breakpoint() datamodule = hydra.utils.instantiate(cfg.data.datamodule) - datamodule.setup('test') + datamodule.setup("test") breakpoint() # Load the test config average_squared_distance = compute_average_squared_distance_from_config(cfg) temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) cfg.model.average_squared_distance = average_squared_distance - + # Compute radial cutoff for spatiotemporal model using the same formula as denoiser sigma = cfg.model.sigma_distribution.sigma max_radius = cfg.model.max_radius @@ -112,13 +115,13 @@ def main(cfg): max_radius=max_radius, average_squared_distance=average_squared_distance, # Use temporal for spatiotemporal model sigma=sigma, - D=3 + D=3, ) temporal_radial_cutoff = compute_radial_cutoff( max_radius=max_radius, average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model sigma=sigma, - D=3 + D=3, ) cfg.model.conditioner.spatiotemporal_model.radial_cutoff = spatial_radial_cutoff cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff @@ -138,20 +141,21 @@ def main(cfg): # cfg.model.conditioner.N_structures = 2 # Must match architecture N_structures cfg.model.conditioner.pretrained_model_path = "sule-shashank/jamun/88i7qkj2" cfg.model.conditioner.c_in = c_in_float - + if cfg.model.conditioner._target_ == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner": cfg.model.conditioner.spatiotemporal_model.radial_cutoff = average_squared_distance max_radius = cfg.model.max_radius temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) temporal_radial_cutoff = compute_radial_cutoff( - max_radius=max_radius, - average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model - sigma=sigma, - D=3) + max_radius=max_radius, + average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model + sigma=sigma, + D=3, + ) cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff cfg.model.conditioner.c_noise = c_noise_float cfg.model.conditioner.c_in = c_in_float - + print("Loading model...") model = hydra.utils.instantiate(cfg.model) model.conditioning_module.c_noise = c_noise @@ -160,32 +164,33 @@ def main(cfg): print(f"Sigma: {model.sigma_distribution.sigma}") # print(f"Conditioner c_in: {model.conditioning_module.c_in}") breakpoint() - + # Get a single batch print("Getting a batch of data...") train_loader = datamodule.train_dataloader() _, batch = next(enumerate(train_loader)) - + print(f"Batch shape: {batch.pos.shape}") print(f"Hidden state shape: {[h.shape for h in batch.hidden_state]}") breakpoint() - + # Test forward pass print("Testing forward pass...") with torch.no_grad(): sigma = model.sigma_distribution.sample() x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) - + print(f"Input shape: {batch.pos.shape}") print(f"Noisy shape: {y.pos.shape}") print(f"Output shape: {xhat.pos.shape}") breakpoint() - + # Test single training step print("Testing training step...") loss_output = model.training_step(batch, 0) print(f"Loss: {loss_output['loss']:.4f}") breakpoint() + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/test_conditional_sampling.py b/scratch/test_conditional_sampling.py index b370a93..d00616c 100644 --- a/scratch/test_conditional_sampling.py +++ b/scratch/test_conditional_sampling.py @@ -1,38 +1,36 @@ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys import os +import sys + +import dotenv import hydra -from omegaconf import OmegaConf import torch -import torch_geometric -from jamun.hydra import instantiate_dict_cfg -import pdb -import jamun -from jamun.data import MDtrajDataModule -from jamun.utils import find_checkpoint -import wandb -from jamun.utils import ModelSamplingWrapperMemory, ModelSamplingWrapper +from omegaconf import OmegaConf from tqdm import tqdm -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +from jamun.data import MDtrajDataModule +from jamun.utils import ModelSamplingWrapperMemory, find_checkpoint + +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") -project_root = "/homefs/home/sules/jamun" # Adjust if necessary +project_root = "/homefs/home/sules/jamun" # Adjust if necessary if project_root not in sys.path: sys.path.insert(0, project_root) print(f"Added '{project_root}' to sys.path for module discovery.") else: print(f"'{project_root}' is already in sys.path.") + @hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="sample_memory") def main(cfg): print("Configuration loaded:") print(OmegaConf.to_yaml(cfg)) breakpoint() - + # Load checkpoint using find_checkpoint function print("Finding checkpoint...") checkpoint_path = find_checkpoint( @@ -41,44 +39,46 @@ def main(cfg): checkpoint_type=cfg.get("checkpoint_type"), ) print(f"Checkpoint found at: {checkpoint_path}") - import numpy as np - # cfg.M = 1/6.0 + # cfg.M = 1/6.0 # cfg.delta = float(cfg.sigma) # cfg.friction = float(-np.log(np.sqrt(1-4*cfg.M))) # u = 1/cfg.M # cfg.inverse_temperature = float(4/(u*(1- np.sqrt(1 - 4/u)))) print(f"Sampler params: {cfg.M}, {cfg.delta}, {cfg.friction}, {cfg.inverse_temperature}") breakpoint() - + # Load the model from checkpoint by instantiating it with the checkpoint path print("Loading model from checkpoint...") cfg.model.checkpoint_path = checkpoint_path model = hydra.utils.instantiate(cfg.model) from e3tools.nn import LayerNorm - model.conditioning_module.spatiotemporal_model.temporal_to_spatial_pooler.layer_norm = LayerNorm(model.conditioning_module.spatiotemporal_model.temporal_module.irreps_out) - model.conditioning_module.spatiotemporal_model.spatial_to_temporal_pooler.layer_norm = LayerNorm(model.conditioning_module.spatiotemporal_model.spatial_module.irreps_out) + + model.conditioning_module.spatiotemporal_model.temporal_to_spatial_pooler.layer_norm = LayerNorm( + model.conditioning_module.spatiotemporal_model.temporal_module.irreps_out + ) + model.conditioning_module.spatiotemporal_model.spatial_to_temporal_pooler.layer_norm = LayerNorm( + model.conditioning_module.spatiotemporal_model.spatial_module.irreps_out + ) print(f"Model loaded: {type(model)}") breakpoint() - + # Set up initial datasets for sampling print("Setting up initial datasets...") init_datasets = hydra.utils.instantiate(cfg.init_datasets) print(f"Initial datasets loaded: {len(init_datasets)} datasets") print(f"Dataset types: {[type(ds) for ds in init_datasets]}") breakpoint() - + # Manually construct the DataModule print("Creating datamodule for testing...") datamodule = MDtrajDataModule( - datasets={"train": init_datasets, "val": init_datasets, "test": init_datasets}, - batch_size=1, - num_workers=1 + datasets={"train": init_datasets, "val": init_datasets, "test": init_datasets}, batch_size=1, num_workers=1 ) - + datamodule.setup("test") print("Datamodule setup complete") breakpoint() - + # Get a sample batch print("Getting a sample batch...") test_loader = datamodule.test_dataloader() @@ -88,7 +88,7 @@ def main(cfg): # if hasattr(batch, 'hidden_state') and len(batch.hidden_state) > 0: # print(f"Hidden state shapes: {[h.shape for h in batch.hidden_state]}") breakpoint() - + # Set up sampler print("Setting up sampler...") sampler = hydra.utils.instantiate(cfg.sampler) @@ -101,12 +101,14 @@ def main(cfg): print(f"Batch sampler mcmc: {batch_sampler.mcmc}") breakpoint() - # Write test for score + # Write test for score print("Testing score function...") with torch.no_grad(): - init_graphs = batch + init_graphs = batch init_graphs = init_graphs.to(sampler.fabric.device) - model_wrapped = ModelSamplingWrapperMemory(model=model, init_graphs=init_graphs, sigma=batch_sampler.sigma, recenter_on_init=True) + model_wrapped = ModelSamplingWrapperMemory( + model=model, init_graphs=init_graphs, sigma=batch_sampler.sigma, recenter_on_init=True + ) y_init = model_wrapped.sample_initial_noisy_positions() y_hist_init = model_wrapped.sample_initial_noisy_history() init_score = model_wrapped.score(y_init, y_hist_init, batch_sampler.sigma) @@ -115,21 +117,25 @@ def main(cfg): # Test walk with torch.no_grad(): - y, v, y_hist, y_traj, score_traj, y_hist_traj = batch_sampler.mcmc(y_init, y_hist_init, \ - lambda y, y_hist: model_wrapped.score(y, y_hist, batch_sampler.sigma), \ - v_init="zero", steps=5) + y, v, y_hist, y_traj, score_traj, y_hist_traj = batch_sampler.mcmc( + y_init, + y_hist_init, + lambda y, y_hist: model_wrapped.score(y, y_hist, batch_sampler.sigma), + v_init="zero", + steps=5, + ) print(f"Score trajectory: {score_traj}") breakpoint() # Test jump with torch.no_grad(): xhat_traj = torch.stack( - [ - model_wrapped.xhat(y_traj[i, :], y_hist_traj[i], sigma=batch_sampler.sigma) - for i in tqdm(range(y_traj.size(0)), leave=False, desc="Jump") - ], - dim=0, - ) + [ + model_wrapped.xhat(y_traj[i, :], y_hist_traj[i], sigma=batch_sampler.sigma) + for i in tqdm(range(y_traj.size(0)), leave=False, desc="Jump") + ], + dim=0, + ) print(f"Xhat trajectory: {xhat_traj}") breakpoint() @@ -154,16 +160,16 @@ def main(cfg): print(f"Number of sampling steps per batch: {cfg.num_sampling_steps_per_batch}") print(f"Number of batches: {cfg.num_batches}") breakpoint() - + # Test a forward pass with the model print("Testing model forward pass...") model.eval() with torch.no_grad(): # Test if the model can process the batch - if hasattr(model, 'noise_and_denoise'): + if hasattr(model, "noise_and_denoise"): sigma_tensor = torch.tensor([cfg.sigma]) x_target, xhat, y = model.noise_and_denoise(batch, sigma_tensor, align_noisy_input=True) - print(f"Forward pass successful!") + print("Forward pass successful!") print(f"Input shape: {batch.pos.shape}") print(f"Noisy shape: {y.pos.shape}") print(f"Denoised shape: {xhat.pos.shape}") @@ -172,8 +178,9 @@ def main(cfg): output = model(batch) print(f"Model output: {type(output)}") breakpoint() - + print("Script completed successfully!") + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/test_conditional_simple.py b/scratch/test_conditional_simple.py index d2b579e..bae7188 100644 --- a/scratch/test_conditional_simple.py +++ b/scratch/test_conditional_simple.py @@ -5,14 +5,16 @@ """ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys import os +import sys + +import dotenv import hydra -from omegaconf import OmegaConf import torch -import torch_geometric +from omegaconf import OmegaConf + from jamun.utils import compute_average_squared_distance_from_datasets breakpoint() # Start debugging @@ -28,6 +30,7 @@ breakpoint() # After setup + def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the average squared distance for normalization from the data.""" breakpoint() # Start of function @@ -46,105 +49,105 @@ def main(cfg): print("=" * 60) print("Testing debugged denoiser_conditional") print("=" * 60) - + # Compute average squared distance print("Computing average squared distance...") breakpoint() # Before distance computation average_squared_distance = compute_average_squared_distance_from_config(cfg) cfg.model.average_squared_distance = average_squared_distance print(f"Average squared distance: {average_squared_distance:.6f}") - + # Load datamodule print("Loading datamodule...") breakpoint() # Before datamodule datamodule = hydra.utils.instantiate(cfg.data.datamodule) - datamodule.setup('test') - + datamodule.setup("test") + # Load model print("Loading model...") breakpoint() # Before model loading model = hydra.utils.instantiate(cfg.model) - + # Get a single batch print("Getting a batch of data...") breakpoint() # Before getting batch train_loader = datamodule.train_dataloader() _, batch = next(enumerate(train_loader)) - - print(f"Batch info:") + + print("Batch info:") print(f" Position shape: {batch.pos.shape}") print(f" Number of atoms: {batch.pos.shape[0]}") print(f" Hidden state shapes: {[h.shape for h in batch.hidden_state]}") print(f" Number of hidden states: {len(batch.hidden_state)}") - + breakpoint() # After batch info - + # Test with sigma = 0.0 print("\n" + "=" * 40) print("Testing with sigma = 0.0 (no noise)") print("=" * 40) - + breakpoint() # Before sigma=0.0 test - + with torch.no_grad(): breakpoint() # Before noise_and_denoise sigma = torch.tensor(0.0) x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) - + print(f"Input shape: {batch.pos.shape}") print(f"Noisy shape: {y.pos.shape}") print(f"Output shape: {xhat.pos.shape}") - + breakpoint() # After noise_and_denoise - + # Compute loss loss, aux = model.compute_loss(x_target, xhat, sigma) print(f"Loss: {loss.mean().item():.6f}") print(f"Metrics: {aux}") - + # Check if positions are preserved (should be identical with sigma=0) pos_diff = torch.abs(batch.pos - y.pos).max() print(f"Max position difference (sigma=0): {pos_diff.item():.8f}") - + breakpoint() # After sigma=0.0 test - + # Test with sigma = 0.1 print("\n" + "=" * 40) print("Testing with sigma = 0.1 (with noise)") print("=" * 40) - + breakpoint() # Before sigma=0.1 test - + with torch.no_grad(): sigma = torch.tensor(0.1) breakpoint() # Before noise_and_denoise with sigma=0.1 x_target, xhat, y = model.noise_and_denoise(batch, sigma, align_noisy_input=True) - + print(f"Input shape: {batch.pos.shape}") print(f"Noisy shape: {y.pos.shape}") print(f"Output shape: {xhat.pos.shape}") - + # Compute loss loss, aux = model.compute_loss(x_target, xhat, sigma) print(f"Loss: {loss.mean().item():.6f}") print(f"Metrics: {aux}") - + # Check noise level pos_diff = torch.abs(batch.pos - y.pos).max() print(f"Max position difference (sigma=0.1): {pos_diff.item():.6f}") - + # Check denoising quality denoise_diff = torch.abs(batch.pos - xhat.pos).max() print(f"Max denoising difference: {denoise_diff.item():.6f}") - + breakpoint() # After sigma=0.1 test - + print("\n" + "=" * 60) print("Testing complete!") print("=" * 60) - + breakpoint() # End of main if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/test_conditioners.py b/scratch/test_conditioners.py index 9a97d38..bc66643 100644 --- a/scratch/test_conditioners.py +++ b/scratch/test_conditioners.py @@ -1,21 +1,22 @@ #!/usr/bin/env python3 """ -Test script for SelfConditioner, PositionConditioner, and MeanConditioner with both +Test script for SelfConditioner, PositionConditioner, and MeanConditioner with both MDtrajDataset and RepeatedPositionDataset. """ +# Add the src directory to the path so we can import jamun modules +import sys +from pathlib import Path + import torch import torch_geometric -import os -from pathlib import Path -# Add the src directory to the path so we can import jamun modules -import sys sys.path.insert(0, str(Path(__file__).parent / "src")) from jamun.data._mdtraj import MDtrajDataset from jamun.data.noisy_position_dataset import RepeatedPositionDataset -from jamun.model.conditioners.conditioners import SelfConditioner, PositionConditioner, MeanConditioner +from jamun.model.conditioners.conditioners import MeanConditioner, PositionConditioner, SelfConditioner + def print_tensor_summary(tensor, name, max_elements=6): """Print a summary of a tensor with first few elements.""" @@ -23,18 +24,21 @@ def print_tensor_summary(tensor, name, max_elements=6): print(f"{name}: {tensor.flatten().tolist()}") else: flat = tensor.flatten() - print(f"{name} (shape {tensor.shape}): [{flat[0]:.6f}, {flat[1]:.6f}, {flat[2]:.6f}, ..., {flat[-3]:.6f}, {flat[-2]:.6f}, {flat[-1]:.6f}]") + print( + f"{name} (shape {tensor.shape}): [{flat[0]:.6f}, {flat[1]:.6f}, {flat[2]:.6f}, ..., {flat[-3]:.6f}, {flat[-2]:.6f}, {flat[-1]:.6f}]" + ) + def create_datasets(): """Create both types of datasets with 3 total structures (2 hidden states).""" - + root = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train" traj_files = ["ALA_ALA.xtc"] pdb_file = "ALA_ALA.pdb" total_lag_time = 3 # This should create 2 hidden states (3 - 1 = 2) - + print(f"Creating datasets with total_lag_time={total_lag_time} (expecting 2 hidden states)...") - + # Create MDtrajDataset (with real lag processing) mdtraj_dataset = MDtrajDataset( root=root, @@ -44,9 +48,9 @@ def create_datasets(): total_lag_time=total_lag_time, lag_subsample_rate=1, num_frames=10, - verbose=True + verbose=True, ) - + # Create RepeatedPositionDataset (with position copies) repeated_dataset = RepeatedPositionDataset( root=root, @@ -56,38 +60,40 @@ def create_datasets(): total_lag_time=total_lag_time, lag_subsample_rate=1, num_frames=10, - verbose=True + verbose=True, ) - + return mdtraj_dataset, repeated_dataset + def create_batch_from_dataset(dataset, sample_idx=0): """Create a batched graph from a single dataset sample.""" graph = dataset[sample_idx] batch = torch_geometric.data.Batch.from_data_list([graph]) return batch + def print_batch_details(batch, batch_name): """Print detailed information about a batch.""" print(f"\n--- {batch_name} Details ---") print(f"Position shape: {batch.pos.shape}") print_tensor_summary(batch.pos, "Position") - + # Check if position is mean centered pos_mean = torch.mean(batch.pos, dim=0) # Mean over atoms pos_mean_magnitude = torch.norm(pos_mean).item() print(f"Position mean: [{pos_mean[0]:.6f}, {pos_mean[1]:.6f}, {pos_mean[2]:.6f}]") print(f"Position mean magnitude: {pos_mean_magnitude:.6f}") if pos_mean_magnitude < 1e-6: - print(f"āœ… Input position is mean centered") + print("āœ… Input position is mean centered") else: - print(f"āŒ Input position is NOT mean centered") - + print("āŒ Input position is NOT mean centered") + print(f"Number of hidden states: {len(batch.hidden_state)}") for i, hidden_state in enumerate(batch.hidden_state): print(f"Hidden state {i} shape: {hidden_state.shape}") print_tensor_summary(hidden_state, f"Hidden state {i}") - + # Check if hidden state is mean centered hidden_mean = torch.mean(hidden_state, dim=0) # Mean over atoms hidden_mean_magnitude = torch.norm(hidden_mean).item() @@ -98,56 +104,59 @@ def print_batch_details(batch, batch_name): else: print(f"āŒ Hidden state {i} is NOT mean centered") + def test_conditioner_detailed(conditioner, batch, test_name): """Test a conditioner with detailed output.""" - print(f"\n{'='*70}") + print(f"\n{'=' * 70}") print(f"{test_name}") - print(f"{'='*70}") - + print(f"{'=' * 70}") + # Print input details print_batch_details(batch, "Input Batch") - + # Run the conditioner try: print(f"\nRunning {conditioner.__class__.__name__}...") conditioned_structures = conditioner(batch) - - print(f"\n--- Conditioner Output ---") + + print("\n--- Conditioner Output ---") print(f"Number of conditioned structures: {len(conditioned_structures)}") - + # Print each conditioned structure for i, structure in enumerate(conditioned_structures): print(f"\nConditioned structure {i} shape: {structure.shape}") print_tensor_summary(structure, f"Conditioned structure {i}") - + # Compare with input position pos_diff = torch.max(torch.abs(structure - batch.pos)).item() print(f"Max difference from current position: {pos_diff:.10f}") - + # Compare with hidden states if available if i < len(batch.hidden_state): hidden_diff = torch.max(torch.abs(structure - batch.hidden_state[i])).item() print(f"Max difference from hidden state {i}: {hidden_diff:.10f}") - + # Check if structure is mean centered (for PositionConditioner) if conditioner.__class__.__name__ == "PositionConditioner": structure_mean = torch.mean(structure, dim=0) # Mean over atoms mean_magnitude = torch.norm(structure_mean).item() - print(f"Mean of structure {i}: [{structure_mean[0]:.6f}, {structure_mean[1]:.6f}, {structure_mean[2]:.6f}]") + print( + f"Mean of structure {i}: [{structure_mean[0]:.6f}, {structure_mean[1]:.6f}, {structure_mean[2]:.6f}]" + ) print(f"Magnitude of mean: {mean_magnitude:.6f}") - + # Check if it's close to zero (mean centered) if mean_magnitude < 1e-6: print(f"āœ… Structure {i} is mean centered (mean ā‰ˆ 0)") else: print(f"āŒ Structure {i} is NOT mean centered") - + # Check if structure contains means across time steps (for MeanConditioner) if conditioner.__class__.__name__ == "MeanConditioner": # For MeanConditioner, each structure should be the mean across time steps # All structures should be identical, but atoms can have different coordinates print(f"āœ… Structure {i} contains mean across time steps") - + # Check if this structure is the same as the first structure (all should be the same mean) if i > 0: first_structure = conditioned_structures[0] @@ -158,7 +167,7 @@ def test_conditioner_detailed(conditioner, batch, test_name): print(f"āŒ Structure {i} doesn't match structure 0 (all should be identical)") max_diff = torch.max(torch.abs(structure - first_structure)).item() print(f"Maximum difference: {max_diff:.10f}") - + # Verify the mean computation is correct by manually computing it if hasattr(batch, "hidden_state") and batch.hidden_state is not None: all_positions = [batch.pos] + batch.hidden_state @@ -177,27 +186,29 @@ def test_conditioner_detailed(conditioner, batch, test_name): print(f"āœ… Structure {i} correctly equals y.pos (no hidden states)") else: print(f"āŒ Structure {i} should equal y.pos when no hidden states") - + return conditioned_structures - + except Exception as e: print(f"āŒ ERROR: Exception in {test_name}: {e}") import traceback + traceback.print_exc() return False + def verify_results(conditioned_structures, batch, test_name, expected_behavior): """Verify the results match expected behavior.""" print(f"\n--- Verification for {test_name} ---") print(f"Expected behavior: {expected_behavior}") - + expected_count = 3 # N_structures = 3, so we expect 3 total structures including current position if len(conditioned_structures) != expected_count: print(f"āŒ ERROR: Expected {expected_count} structures, got {len(conditioned_structures)}") return False - + print(f"āœ… Correct count: {len(conditioned_structures)} structures") - + success = True for i, structure in enumerate(conditioned_structures): if structure.shape != batch.pos.shape: @@ -205,12 +216,12 @@ def verify_results(conditioned_structures, batch, test_name, expected_behavior): success = False else: print(f"āœ… Structure {i} has correct shape") - + # Verify that the first structure is the current position for most conditioners # Exception: MeanConditioner returns time-averaged means, not current position first_structure = conditioned_structures[0] pos_diff = torch.max(torch.abs(first_structure - batch.pos)).item() - + if test_name.startswith("MeanConditioner"): # For MeanConditioner, first structure should be the time-averaged mean, not y.pos if hasattr(batch, "hidden_state") and batch.hidden_state is not None: @@ -236,161 +247,149 @@ def verify_results(conditioned_structures, batch, test_name, expected_behavior): else: print(f"āŒ ERROR: First structure doesn't match current position (diff: {pos_diff:.2e})") success = False - + return success + def main(): """Main test function.""" - print("Testing Conditioners: SelfConditioner, PositionConditioner, MeanConditioner with 3 total structures (2 hidden states)") + print( + "Testing Conditioners: SelfConditioner, PositionConditioner, MeanConditioner with 3 total structures (2 hidden states)" + ) print("=" * 70) - + # Create datasets try: mdtraj_dataset, repeated_dataset = create_datasets() - print(f"āœ… Created datasets") + print("āœ… Created datasets") print(f" MDtrajDataset length: {len(mdtraj_dataset)}") print(f" RepeatedPositionDataset length: {len(repeated_dataset)}") except Exception as e: print(f"āŒ ERROR: Failed to create datasets: {e}") return False - + # Create batches try: mdtraj_batch = create_batch_from_dataset(mdtraj_dataset, sample_idx=0) repeated_batch = create_batch_from_dataset(repeated_dataset, sample_idx=0) - print(f"āœ… Created batches") + print("āœ… Created batches") except Exception as e: print(f"āŒ ERROR: Failed to create batches: {e}") return False - + # Create conditioners try: self_conditioner = SelfConditioner(N_structures=3) position_conditioner = PositionConditioner(N_structures=3) mean_conditioner = MeanConditioner(N_structures=3) - print(f"āœ… Created conditioners") + print("āœ… Created conditioners") except Exception as e: print(f"āŒ ERROR: Failed to create conditioners: {e}") return False - + # Test 1: SelfConditioner on MDtrajDataset - result1 = test_conditioner_detailed( - self_conditioner, - mdtraj_batch, - "TEST 1: SelfConditioner on MDtrajDataset" - ) + result1 = test_conditioner_detailed(self_conditioner, mdtraj_batch, "TEST 1: SelfConditioner on MDtrajDataset") if result1 is False: return False success1 = verify_results( - result1, - mdtraj_batch, + result1, + mdtraj_batch, "SelfConditioner + MDtrajDataset", - "Should return [y.pos, y.pos, y.pos] - 3 copies of current position" + "Should return [y.pos, y.pos, y.pos] - 3 copies of current position", ) - - # Test 2: SelfConditioner on RepeatedPositionDataset + + # Test 2: SelfConditioner on RepeatedPositionDataset result2 = test_conditioner_detailed( - self_conditioner, - repeated_batch, - "TEST 2: SelfConditioner on RepeatedPositionDataset" + self_conditioner, repeated_batch, "TEST 2: SelfConditioner on RepeatedPositionDataset" ) if result2 is False: return False success2 = verify_results( - result2, - repeated_batch, + result2, + repeated_batch, "SelfConditioner + RepeatedPositionDataset", - "Should return [y.pos, y.pos, y.pos] - 3 copies of current position" + "Should return [y.pos, y.pos, y.pos] - 3 copies of current position", ) - + # Test 3: PositionConditioner on MDtrajDataset result3 = test_conditioner_detailed( - position_conditioner, - mdtraj_batch, - "TEST 3: PositionConditioner on MDtrajDataset" + position_conditioner, mdtraj_batch, "TEST 3: PositionConditioner on MDtrajDataset" ) if result3 is False: return False success3 = verify_results( - result3, - mdtraj_batch, + result3, + mdtraj_batch, "PositionConditioner + MDtrajDataset", - "Should return [y.pos, aligned_hidden_state_1, aligned_hidden_state_2] - current position + 2 aligned hidden states" + "Should return [y.pos, aligned_hidden_state_1, aligned_hidden_state_2] - current position + 2 aligned hidden states", ) - + # Test 4: PositionConditioner on RepeatedPositionDataset result4 = test_conditioner_detailed( - position_conditioner, - repeated_batch, - "TEST 4: PositionConditioner on RepeatedPositionDataset" + position_conditioner, repeated_batch, "TEST 4: PositionConditioner on RepeatedPositionDataset" ) if result4 is False: return False success4 = verify_results( - result4, - repeated_batch, + result4, + repeated_batch, "PositionConditioner + RepeatedPositionDataset", - "Should return [y.pos, aligned_copy_1, aligned_copy_2] - current position + 2 aligned copies" + "Should return [y.pos, aligned_copy_1, aligned_copy_2] - current position + 2 aligned copies", ) - + # Test 5: MeanConditioner on MDtrajDataset - result5 = test_conditioner_detailed( - mean_conditioner, - mdtraj_batch, - "TEST 5: MeanConditioner on MDtrajDataset" - ) + result5 = test_conditioner_detailed(mean_conditioner, mdtraj_batch, "TEST 5: MeanConditioner on MDtrajDataset") if result5 is False: return False success5 = verify_results( - result5, - mdtraj_batch, + result5, + mdtraj_batch, "MeanConditioner + MDtrajDataset", - "Should return [time_mean, time_mean, time_mean] - 3 copies of mean across time steps (y.pos + hidden states)" + "Should return [time_mean, time_mean, time_mean] - 3 copies of mean across time steps (y.pos + hidden states)", ) - + # Test 6: MeanConditioner on RepeatedPositionDataset result6 = test_conditioner_detailed( - mean_conditioner, - repeated_batch, - "TEST 6: MeanConditioner on RepeatedPositionDataset" + mean_conditioner, repeated_batch, "TEST 6: MeanConditioner on RepeatedPositionDataset" ) if result6 is False: return False success6 = verify_results( - result6, - repeated_batch, + result6, + repeated_batch, "MeanConditioner + RepeatedPositionDataset", - "Should return [time_mean, time_mean, time_mean] - 3 copies of mean across time steps (y.pos + hidden states)" + "Should return [time_mean, time_mean, time_mean] - 3 copies of mean across time steps (y.pos + hidden states)", ) - + # Summary - print(f"\n{'='*70}") + print(f"\n{'=' * 70}") print("SUMMARY") - print(f"{'='*70}") - + print(f"{'=' * 70}") + tests = [ ("SelfConditioner + MDtrajDataset", success1), - ("SelfConditioner + RepeatedPositionDataset", success2), + ("SelfConditioner + RepeatedPositionDataset", success2), ("PositionConditioner + MDtrajDataset", success3), ("PositionConditioner + RepeatedPositionDataset", success4), ("MeanConditioner + MDtrajDataset", success5), - ("MeanConditioner + RepeatedPositionDataset", success6) + ("MeanConditioner + RepeatedPositionDataset", success6), ] - + all_passed = True for test_name, success in tests: status = "āœ… PASS" if success else "āŒ FAIL" print(f"{test_name}: {status}") if not success: all_passed = False - + if all_passed: - print(f"\nšŸŽ‰ All conditioner tests passed!") + print("\nšŸŽ‰ All conditioner tests passed!") return True else: - print(f"\nšŸ’„ Some conditioner tests failed!") + print("\nšŸ’„ Some conditioner tests failed!") return False + if __name__ == "__main__": success = main() - exit(0 if success else 1) \ No newline at end of file + exit(0 if success else 1) diff --git a/scratch/test_denoised_conditioner.py b/scratch/test_denoised_conditioner.py index 6489c03..fef2df2 100644 --- a/scratch/test_denoised_conditioner.py +++ b/scratch/test_denoised_conditioner.py @@ -3,96 +3,100 @@ Comprehensive test for DenoisedConditioner using real ALA_ALA data. This test emulates the behavior of xhat_normalized to properly test the scaling parameter. """ -import torch -import torch_geometric -import numpy as np + import os + import e3nn -from jamun.model.conditioners import DenoisedConditioner +import numpy as np +import torch +import torch_geometric + from jamun.data import parse_datasets_from_directory -from jamun.utils import unsqueeze_trailing, mean_center +from jamun.model.conditioners import DenoisedConditioner +from jamun.utils import mean_center, unsqueeze_trailing from jamun.utils._normalizations import normalization_factors # Fix e3nn optimization for avoiding script issues e3nn.set_optimization_defaults(jit_script_fx=False) + def load_ala_ala_data(): """Load actual ALA_ALA data from the capped diamines dataset.""" - + # Get data path from environment variable data_path = os.getenv("JAMUN_DATA_PATH") if not data_path: raise ValueError("JAMUN_DATA_PATH environment variable not set") - + # Load ALA_ALA dataset datasets = parse_datasets_from_directory( root=f"{data_path}/capped_diamines/timewarp_splits/train", traj_pattern="^(.*).xtc", pdb_pattern="^(.*).pdb", - filter_codes=['ALA_ALA'], + filter_codes=["ALA_ALA"], as_iterable=False, subsample=1, total_lag_time=3, # This will give us hidden states lag_subsample_rate=1, num_frames=10, - max_datasets=1 + max_datasets=1, ) - + if not datasets: raise ValueError("No ALA_ALA datasets found") - + # Get the first dataset dataset = datasets[0] print(f"Loaded ALA_ALA dataset with {len(dataset)} frames") - + # Get a few samples to create a batch samples = [] for i in range(min(2, len(dataset))): # Get 2 samples for batch sample = dataset[i] samples.append(sample) - + # Create batch batch = torch_geometric.data.Batch.from_data_list(samples) - + print(f"Created batch with {batch.num_graphs} graphs") print(f"Batch position shape: {batch.pos.shape}") - - if hasattr(batch, 'hidden_state') and batch.hidden_state: + + if hasattr(batch, "hidden_state") and batch.hidden_state: print(f"Hidden states: {len(batch.hidden_state)} states") for i, hidden_state in enumerate(batch.hidden_state): print(f" Hidden state {i}: shape {hidden_state.shape}") else: print("No hidden states found") - + return batch + def add_noise_to_batch(x: torch_geometric.data.Batch, sigma: float) -> torch_geometric.data.Batch: """Add noise to a batch, similar to the denoiser's add_noise method.""" sigma = unsqueeze_trailing(torch.tensor(sigma), x.pos.ndim) - + y = x.clone() - + # Add noise to positions noise = torch.randn_like(x.pos) y.pos = x.pos + sigma * noise - + # Add noise to hidden states if they exist if hasattr(x, "hidden_state") and x.hidden_state is not None: y.hidden_state = [] for hidden_positions in x.hidden_state: hidden_noise = torch.randn_like(hidden_positions) y.hidden_state.append(hidden_positions + sigma * hidden_noise) - + return y + def mean_center_positions(batch: torch_geometric.data.Batch) -> torch_geometric.data.Batch: """Mean-center positions and hidden states for each graph in the batch.""" - from e3tools import scatter - from jamun.utils import mean_center - + # Mean-center the main positions using the jamun utils function batch = mean_center(batch) - + # Mean-center each hidden state individually if hasattr(batch, "hidden_state") and batch.hidden_state is not None: for i, hidden_positions in enumerate(batch.hidden_state): @@ -101,152 +105,156 @@ def mean_center_positions(batch: torch_geometric.data.Batch) -> torch_geometric. temp_batch.pos = hidden_positions temp_batch_centered = mean_center(temp_batch) batch.hidden_state[i] = temp_batch_centered.pos - + return batch + def emulate_xhat_normalized_scaling(batch, sigma: float, average_squared_distance: float = 0.332): """ Emulate the scaling behavior in xhat_normalized method. This simulates how the denoiser scales data before passing to conditioner. """ - print(f"\n=== Emulating xhat_normalized scaling ===") + print("\n=== Emulating xhat_normalized scaling ===") print(f"Input sigma: {sigma}") print(f"Average squared distance: {average_squared_distance}") - + # Mean-center the batch positions and hidden states (as done in actual xhat_normalized) batch = mean_center_positions(batch) - + # Compute normalization factors (same as in denoiser) c_in, c_skip, c_out, c_noise = normalization_factors(sigma, average_squared_distance) - - print(f"Normalization factors:") + + print("Normalization factors:") print(f" c_in: {c_in}") print(f" c_skip: {c_skip}") print(f" c_out: {c_out}") print(f" c_noise: {c_noise}") - + # Adjust dimensions (same as in denoiser) c_in = unsqueeze_trailing(c_in, batch.pos.ndim - 1) c_skip = unsqueeze_trailing(c_skip, batch.pos.ndim - 1) c_out = unsqueeze_trailing(c_out, batch.pos.ndim - 1) c_noise = c_noise.unsqueeze(0) - + # Scale the batch (same as in denoiser) y_scaled = batch.clone() y_scaled.pos = batch.pos * c_in - + print(f"Original position mean: {batch.pos.mean():.6f}") print(f"Scaled position mean: {y_scaled.pos.mean():.6f}") - + # Scale hidden states (same as in denoiser) if hasattr(batch, "hidden_state") and batch.hidden_state is not None: y_scaled.hidden_state = [] for i, positions in enumerate(batch.hidden_state): scaled_positions = positions * c_in y_scaled.hidden_state.append(scaled_positions) - print(f"Hidden state {i} - Original mean: {positions.mean():.6f}, Scaled mean: {scaled_positions.mean():.6f}") - + print( + f"Hidden state {i} - Original mean: {positions.mean():.6f}, Scaled mean: {scaled_positions.mean():.6f}" + ) + return y_scaled, c_in, c_skip, c_out, c_noise + def test_denoised_conditioner_with_scaling(): """Test the DenoisedConditioner with proper scaling emulation.""" - + print("=== Testing DenoisedConditioner with xhat_normalized scaling ===") - + # Test parameters N_structures = 3 # Must match architecture N_structures (updated to match hidden states) pretrained_model_path = "sule-shashank/jamun/370wpt17" # Update this to your desired checkpoint test_sigma = 0.04 - + try: # Load real ALA_ALA data print("\n1. Loading real ALA_ALA data...") original_batch = load_ala_ala_data() print("āœ“ Real ALA_ALA data loaded successfully") - + # Mean-center the original batch (clean reference) - positions and hidden states print("\n2. Mean-centering the data...") print(f" Original position mean: {original_batch.pos.mean():.6f}") if hasattr(original_batch, "hidden_state") and original_batch.hidden_state: for i, hidden_state in enumerate(original_batch.hidden_state): print(f" Original hidden state {i} mean: {hidden_state.mean():.6f}") - + x_clean = mean_center_positions(original_batch) print(f" Mean-centered position mean: {x_clean.pos.mean():.6f}") - + if hasattr(x_clean, "hidden_state") and x_clean.hidden_state: for i, hidden_state in enumerate(x_clean.hidden_state): print(f" Mean-centered hidden state {i} mean: {hidden_state.mean():.6f}") - + # Add noise to the mean-centered data print(f"\n3. Adding noise with sigma={test_sigma}...") y_noisy = add_noise_to_batch(x_clean, test_sigma) print(f" Noisy position mean: {y_noisy.pos.mean():.6f}") print(f" Noisy position std: {y_noisy.pos.std():.6f}") - + if hasattr(y_noisy, "hidden_state") and y_noisy.hidden_state: for i, hidden_state in enumerate(y_noisy.hidden_state): print(f" Noisy hidden state {i} mean: {hidden_state.mean():.6f}") print(f" Noisy hidden state {i} std: {hidden_state.std():.6f}") - + # Initialize conditioner and extract average_squared_distance from checkpoint - print(f"\n4. Initializing DenoisedConditioner and extracting average_squared_distance...") + print("\n4. Initializing DenoisedConditioner and extracting average_squared_distance...") print(f" N_structures: {N_structures}") print(f" pretrained_model_path: {pretrained_model_path}") - + # Use a temporary c_in for initialization temp_c_in, _, _, _ = normalization_factors(test_sigma, 0.332) # temporary default temp_c_in_float = float(temp_c_in) - + # Initialize conditioner conditioner = DenoisedConditioner( - N_structures=N_structures, - pretrained_model_path=pretrained_model_path, - c_in=temp_c_in_float + N_structures=N_structures, pretrained_model_path=pretrained_model_path, c_in=temp_c_in_float ) - + print("āœ“ DenoisedConditioner initialized successfully") print(f" Denoiser sigma: {conditioner.denoiser_sigma}") - + # Extract average_squared_distance from the loaded checkpoint average_squared_distance = None - if hasattr(conditioner.pretrained_denoiser, 'average_squared_distance'): + if hasattr(conditioner.pretrained_denoiser, "average_squared_distance"): average_squared_distance = float(conditioner.pretrained_denoiser.average_squared_distance) print(f" āœ“ Extracted average_squared_distance from checkpoint: {average_squared_distance}") - elif hasattr(conditioner.pretrained_denoiser, 'hparams') and hasattr(conditioner.pretrained_denoiser.hparams, 'average_squared_distance'): + elif hasattr(conditioner.pretrained_denoiser, "hparams") and hasattr( + conditioner.pretrained_denoiser.hparams, "average_squared_distance" + ): average_squared_distance = float(conditioner.pretrained_denoiser.hparams.average_squared_distance) print(f" āœ“ Extracted average_squared_distance from hparams: {average_squared_distance}") else: # Try to extract from the config if available - if hasattr(conditioner.pretrained_denoiser, 'cfg'): + if hasattr(conditioner.pretrained_denoiser, "cfg"): cfg = conditioner.pretrained_denoiser.cfg - if hasattr(cfg, 'average_squared_distance'): + if hasattr(cfg, "average_squared_distance"): average_squared_distance = float(cfg.average_squared_distance) print(f" āœ“ Extracted average_squared_distance from config: {average_squared_distance}") - + if average_squared_distance is None: print(" āš ļø Could not extract average_squared_distance from checkpoint") print(" Available attributes on pretrained_denoiser:") for attr in dir(conditioner.pretrained_denoiser): - if not attr.startswith('_'): + if not attr.startswith("_"): print(f" - {attr}") # Use default average_squared_distance = 0.332 print(f" Using default average_squared_distance: {average_squared_distance}") - + # Recompute c_in with the correct average_squared_distance c_in, _, _, _ = normalization_factors(test_sigma, average_squared_distance) c_in_float = float(c_in) - + # Update the conditioner's c_in if abs(c_in_float - temp_c_in_float) > 1e-6: print(f" Updating c_in from {temp_c_in_float} to {c_in_float}") conditioner.c_in = c_in_float else: print(f" c_in remains: {c_in_float}") - + # Test sigma consistency - print(f"\n5. Testing sigma consistency...") + print("\n5. Testing sigma consistency...") if abs(conditioner.denoiser_sigma - test_sigma) < 1e-5: print(f"āœ“ Test sigma ({test_sigma}) matches denoiser sigma ({conditioner.denoiser_sigma})") else: @@ -258,125 +266,135 @@ def test_denoised_conditioner_with_scaling(): c_in_float = float(c_in) conditioner.c_in = c_in_float print(f" Updated c_in to: {c_in_float}") - + # Emulate xhat_normalized scaling on the noisy batch - print(f"\n6. Emulating xhat_normalized scaling...") + print("\n6. Emulating xhat_normalized scaling...") scaled_batch, c_in_tensor, c_skip, c_out, c_noise = emulate_xhat_normalized_scaling( y_noisy, test_sigma, average_squared_distance ) - + # Verify our c_in calculation matches assert abs(float(c_in_tensor) - c_in_float) < 1e-6, f"c_in mismatch: {c_in_tensor} vs {c_in_float}" print("āœ“ c_in calculation verified") - + # Test conditioner with scaled noisy data - print(f"\n7. Testing conditioner with scaled noisy data...") - + print("\n7. Testing conditioner with scaled noisy data...") + # Move scaled_batch to the same device as the conditioner device = next(conditioner.parameters()).device scaled_batch = scaled_batch.to(device) x_clean = x_clean.to(device) y_noisy = y_noisy.to(device) print(f" Moved batches to device: {device}") - + conditioned_structures = conditioner.forward(scaled_batch) - - print(f"āœ“ Conditioner forward pass completed") + + print("āœ“ Conditioner forward pass completed") print(f" Returned {len(conditioned_structures)} structures") print(f" Expected N_structures: {N_structures}") - + # Verify output structure - assert len(conditioned_structures) == N_structures, f"Expected {N_structures} structures, got {len(conditioned_structures)}" - + assert len(conditioned_structures) == N_structures, ( + f"Expected {N_structures} structures, got {len(conditioned_structures)}" + ) + for i, structure in enumerate(conditioned_structures): - assert structure.shape == scaled_batch.pos.shape, f"Structure {i} has wrong shape: {structure.shape} vs {scaled_batch.pos.shape}" + assert structure.shape == scaled_batch.pos.shape, ( + f"Structure {i} has wrong shape: {structure.shape} vs {scaled_batch.pos.shape}" + ) print(f" Structure {i}: shape {structure.shape}") - + # Check that first structure is the scaled current position - assert torch.allclose(conditioned_structures[0], scaled_batch.pos), "First structure should be scaled current position" + assert torch.allclose(conditioned_structures[0], scaled_batch.pos), ( + "First structure should be scaled current position" + ) print("āœ“ First structure matches scaled current position") - + # Comprehensive denoising quality test - print(f"\n8. COMPREHENSIVE DENOISING QUALITY TEST...") + print("\n8. COMPREHENSIVE DENOISING QUALITY TEST...") denoising_improvements = [] - + if hasattr(x_clean, "hidden_state") and x_clean.hidden_state and len(conditioned_structures) > 1: print(f" Testing denoising on {len(x_clean.hidden_state)} hidden states...") - + for i in range(1, len(conditioned_structures)): # Skip first structure (current position) hidden_idx = i - 1 # Map to hidden state index if hidden_idx < len(x_clean.hidden_state): denoised_structure = conditioned_structures[i] clean_hidden = x_clean.hidden_state[hidden_idx] noisy_hidden = y_noisy.hidden_state[hidden_idx] - + # Calculate RMSE between denoised and clean - denoised_rmse = torch.sqrt(torch.mean((denoised_structure - clean_hidden)**2)) - + denoised_rmse = torch.sqrt(torch.mean((denoised_structure - clean_hidden) ** 2)) + # Calculate RMSE between noisy and clean for comparison - noisy_rmse = torch.sqrt(torch.mean((noisy_hidden - clean_hidden)**2)) - + noisy_rmse = torch.sqrt(torch.mean((noisy_hidden - clean_hidden) ** 2)) + # Calculate improvement improvement = noisy_rmse - denoised_rmse improvement_percent = (improvement / noisy_rmse) * 100 - + print(f" Hidden State {hidden_idx}:") print(f" Noisy RMSE vs clean: {noisy_rmse.item():.6f}") print(f" Denoised RMSE vs clean: {denoised_rmse.item():.6f}") print(f" Improvement: {improvement.item():.6f} ({improvement_percent.item():.2f}%)") - + denoising_improvements.append(improvement.item()) - + if improvement > 0: - print(f" āœ“ DENOISING SUCCESSFUL (RMSE reduced)") + print(" āœ“ DENOISING SUCCESSFUL (RMSE reduced)") else: - print(f" āŒ DENOISING FAILED (RMSE increased)") - + print(" āŒ DENOISING FAILED (RMSE increased)") + # Verify denoised is different from both noisy and original - assert not torch.allclose(denoised_structure, noisy_hidden, atol=1e-4), f"Denoised structure {i} should be different from noisy" - assert not torch.allclose(denoised_structure, scaled_batch.pos, atol=1e-4), f"Denoised structure {i} should be different from current position" - + assert not torch.allclose(denoised_structure, noisy_hidden, atol=1e-4), ( + f"Denoised structure {i} should be different from noisy" + ) + assert not torch.allclose(denoised_structure, scaled_batch.pos, atol=1e-4), ( + f"Denoised structure {i} should be different from current position" + ) + # Overall denoising assessment - print(f"\n9. OVERALL DENOISING ASSESSMENT...") + print("\n9. OVERALL DENOISING ASSESSMENT...") if denoising_improvements: avg_improvement = np.mean(denoising_improvements) successful_denoising = sum(1 for imp in denoising_improvements if imp > 0) total_tests = len(denoising_improvements) success_rate = (successful_denoising / total_tests) * 100 - + print(f" Total hidden states tested: {total_tests}") print(f" Successful denoising: {successful_denoising}/{total_tests} ({success_rate:.1f}%)") print(f" Average RMSE improvement: {avg_improvement:.6f}") - + if success_rate >= 80: # Require at least 80% success rate - print(f" āœ… DENOISING QUALITY: EXCELLENT (ā‰„80% success)") + print(" āœ… DENOISING QUALITY: EXCELLENT (ā‰„80% success)") elif success_rate >= 60: - print(f" āš ļø DENOISING QUALITY: GOOD (ā‰„60% success)") + print(" āš ļø DENOISING QUALITY: GOOD (ā‰„60% success)") elif success_rate >= 40: - print(f" āš ļø DENOISING QUALITY: MODERATE (ā‰„40% success)") + print(" āš ļø DENOISING QUALITY: MODERATE (ā‰„40% success)") else: - print(f" āŒ DENOISING QUALITY: POOR (<40% success)") - + print(" āŒ DENOISING QUALITY: POOR (<40% success)") + # Assert that at least some denoising occurred assert successful_denoising > 0, "At least one hidden state should show denoising improvement" - print(f" āœ“ At least some denoising improvement verified") - + print(" āœ“ At least some denoising improvement verified") + else: print(" No hidden states available for denoising quality assessment") - + # Additional validation - print(f"\n10. Additional validation...") + print("\n10. Additional validation...") for i, structure in enumerate(conditioned_structures): # Check for NaN values assert not torch.isnan(structure).any(), f"Structure {i} contains NaN values" # Check for infinite values assert not torch.isinf(structure).any(), f"Structure {i} contains infinite values" print(f"āœ“ Structure {i} contains valid values") - + print("\nšŸŽ‰ All tests passed! DenoisedConditioner works correctly.") - + # Print final summary - print(f"\n=== FINAL SUMMARY ===") + print("\n=== FINAL SUMMARY ===") print(f" Checkpoint: {pretrained_model_path}") print(f" Test sigma: {test_sigma}") print(f" Denoiser sigma: {conditioner.denoiser_sigma}") @@ -385,19 +403,21 @@ def test_denoised_conditioner_with_scaling(): if denoising_improvements: print(f" Denoising success rate: {success_rate:.1f}%") print(f" Average RMSE improvement: {avg_improvement:.6f}") - + return True - + except Exception as e: print(f"āŒ Test failed with error: {e}") import traceback + traceback.print_exc() return False + if __name__ == "__main__": success = test_denoised_conditioner_with_scaling() if success: print("\nāœ… DenoisedConditioner scaling test passed!") print("The conditioner correctly handles the scaling parameter and emulates xhat_normalized behavior.") else: - print("\nāŒ DenoisedConditioner scaling test failed!") \ No newline at end of file + print("\nāŒ DenoisedConditioner scaling test failed!") diff --git a/scratch/test_gradient_equivalence.py b/scratch/test_gradient_equivalence.py index 2192c21..e338c63 100644 --- a/scratch/test_gradient_equivalence.py +++ b/scratch/test_gradient_equivalence.py @@ -7,22 +7,22 @@ This will: - Use the train_test_single_shape_conditional experiment config -- Override model to use DenoiserMultimeasurement +- Override model to use DenoiserMultimeasurement - Enable multimeasurement with 2 hidden measurements and 2 measurements - Set max_graphs_per_batch=1 for manual optimization testing """ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys import os +import sys + +import dotenv import hydra -from omegaconf import OmegaConf +import numpy as np import torch import torch_geometric -import copy -import numpy as np dotenv.load_dotenv("../.env", verbose=True) JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") @@ -31,9 +31,9 @@ if project_root not in sys.path: sys.path.insert(0, project_root) -import jamun from jamun.utils import compute_average_squared_distance_from_datasets + def create_model_and_data(cfg, max_graphs_per_batch=None): """Create a model and datamodule with specified optimization mode.""" # Configure DenoiserMultimeasurement @@ -44,13 +44,13 @@ def create_model_and_data(cfg, max_graphs_per_batch=None): cfg.model.N_measurements_hidden = 2 cfg.model.N_measurements = 2 cfg.model.max_graphs_per_batch = max_graphs_per_batch - + # # Set up data - use correct attribute name filter_codes # cfg.data.datamodule.datasets.train.filter_codes = ['ALA_ALA'] - # cfg.data.datamodule.datasets.val.filter_codes = ['ALA_ALA'] + # cfg.data.datamodule.datasets.val.filter_codes = ['ALA_ALA'] # cfg.data.datamodule.datasets.test.filter_codes = ['ALA_ALA'] # cfg.data.datamodule.batch_size = 8 # Larger batch for meaningful chunking - + # Compute normalization datamodule = hydra.utils.instantiate(cfg.data.datamodule) datamodule.setup("compute_normalization") @@ -58,13 +58,14 @@ def create_model_and_data(cfg, max_graphs_per_batch=None): cutoff = cfg.model.max_radius average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) cfg.model.average_squared_distance = average_squared_distance - + # Create model and data model = hydra.utils.instantiate(cfg.model) - datamodule.setup('test') - + datamodule.setup("test") + return model, datamodule + def get_model_gradients(model): """Extract gradients from model parameters.""" gradients = {} @@ -73,198 +74,206 @@ def get_model_gradients(model): gradients[name] = param.grad.clone().detach() return gradients + def compute_gradient_norm(gradients): """Compute the total gradient norm across all parameters.""" total_norm = 0.0 for grad in gradients.values(): total_norm += grad.norm().item() ** 2 - return total_norm ** 0.5 + return total_norm**0.5 + def compare_gradients(grad1, grad2, tolerance=1e-3): """Compare two gradient dictionaries.""" if set(grad1.keys()) != set(grad2.keys()): print("ERROR: Different parameter names!") return False - + max_relative_diff = 0.0 for name in grad1.keys(): g1, g2 = grad1[name], grad2[name] - + # Compute relative difference diff = torch.abs(g1 - g2) max_val = torch.max(torch.abs(g1), torch.abs(g2)) relative_diff = torch.where(max_val > 1e-8, diff / (max_val + 1e-8), diff) max_rel_diff_param = relative_diff.max().item() max_relative_diff = max(max_relative_diff, max_rel_diff_param) - - print(f"{name:30s}: max_rel_diff = {max_rel_diff_param:.6f}, norm_ratio = {g1.norm().item()/g2.norm().item():.6f}") - + + print( + f"{name:30s}: max_rel_diff = {max_rel_diff_param:.6f}, norm_ratio = {g1.norm().item() / g2.norm().item():.6f}" + ) + print(f"\nOverall max relative difference: {max_relative_diff:.6f}") return max_relative_diff < tolerance + @hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") def test_gradient_equivalence(cfg): """Test that automatic and manual optimization produce equivalent gradients.""" - print("="*80) + print("=" * 80) print("TESTING GRADIENT EQUIVALENCE: AUTOMATIC vs MANUAL OPTIMIZATION") - print("="*80) - + print("=" * 80) + # Set seeds for reproducibility torch.manual_seed(42) np.random.seed(42) - + # Create automatic optimization model print("\n1. Creating AUTOMATIC optimization model...") model_auto, datamodule_auto = create_model_and_data(cfg.copy(), max_graphs_per_batch=None) print(f" Automatic optimization: {model_auto.automatic_optimization}") - + # Create manual optimization model with same architecture print("2. Creating MANUAL optimization model...") model_manual, datamodule_manual = create_model_and_data(cfg.copy(), max_graphs_per_batch=2) # 2 graphs per chunk print(f" Automatic optimization: {model_manual.automatic_optimization}") - + # Ensure models are in training mode and parameters require gradients print("3. Setting up models for gradient computation...") model_auto.train() model_manual.train() - + # Ensure all parameters require gradients for param in model_auto.parameters(): param.requires_grad_(True) for param in model_manual.parameters(): param.requires_grad_(True) - + # Copy weights from auto to manual model to ensure identical starting point print("4. Synchronizing model weights...") model_manual.load_state_dict(model_auto.state_dict()) - + # Get the same batch of data print("5. Getting identical batch...") torch.manual_seed(42) # Reset seed to get same batch train_loader_auto = datamodule_auto.train_dataloader() batch_auto = next(iter(train_loader_auto)) - + torch.manual_seed(42) # Reset seed to get same batch train_loader_manual = datamodule_manual.train_dataloader() batch_manual = next(iter(train_loader_manual)) - + print(f" Batch shapes: auto={batch_auto.pos.shape}, manual={batch_manual.pos.shape}") print(f" Batch equality: {torch.allclose(batch_auto.pos, batch_manual.pos)}") - + # Use the same sigma for both (important!) print("\n6. Setting identical sigma values...") sigma_value = 0.04 # Fixed sigma instead of sampling sigma_auto = torch.tensor(sigma_value) sigma_manual = torch.tensor(sigma_value) - + print(f" Using fixed sigma: {sigma_value}") - + # Test forward pass equivalence (without multimeasurement first) print("\n7. Testing forward pass equivalence...") - + # Disable multimeasurement temporarily for cleaner testing model_auto.multimeasurement = False model_manual.multimeasurement = False - + with torch.no_grad(): # Reset random seeds before each forward pass torch.manual_seed(123) x_target_auto, xhat_auto, y_auto = model_auto.noise_and_denoise(batch_auto, sigma_auto, align_noisy_input=True) - + torch.manual_seed(123) # Same seed for manual - x_target_manual, xhat_manual, y_manual = model_manual.noise_and_denoise(batch_manual, sigma_manual, align_noisy_input=True) - + x_target_manual, xhat_manual, y_manual = model_manual.noise_and_denoise( + batch_manual, sigma_manual, align_noisy_input=True + ) + forward_equal = torch.allclose(xhat_auto.pos, xhat_manual.pos, atol=1e-5) print(f" Forward pass output equality: {forward_equal}") - + if not forward_equal: print(f" Max difference: {(xhat_auto.pos - xhat_manual.pos).abs().max().item():.8f}") print(" Continuing with test anyway...") - + # Test gradient computation print("\n8. Computing gradients...") - + # AUTOMATIC OPTIMIZATION print(" Computing automatic optimization gradients...") model_auto.zero_grad() - + # Use same random seed for loss computation torch.manual_seed(456) x_target_auto, xhat_auto, y_auto = model_auto.noise_and_denoise(batch_auto, sigma_auto, align_noisy_input=True) loss_auto, aux_auto = model_auto.compute_loss(x_target_auto, xhat_auto, sigma_auto) loss_auto_mean = loss_auto.mean() - + print(f" Auto loss: {loss_auto_mean.item():.6f}") print(f" Auto loss requires_grad: {loss_auto_mean.requires_grad}") - + loss_auto_mean.backward() - + gradients_auto = get_model_gradients(model_auto) grad_norm_auto = compute_gradient_norm(gradients_auto) - + print(f" Auto grad_norm: {grad_norm_auto:.6f}") - + # MANUAL OPTIMIZATION (simulate _manual_step) print(" Computing manual optimization gradients...") model_manual.zero_grad() - + # Use same random seed for noise generation torch.manual_seed(456) y_manual, x_target_manual_prep = model_manual._prepare_noisy_batch( batch_manual, sigma_manual, align_noisy_input=True ) - + # Split into chunks y_list = y_manual.to_data_list() x_target_list = x_target_manual_prep.to_data_list() chunk_size = model_manual.max_graphs_per_batch num_chunks = (len(y_list) + chunk_size - 1) // chunk_size - + print(f" Manual: {len(y_list)} graphs ā†’ {num_chunks} chunks of size {chunk_size}") - + # Process chunks and accumulate gradients (simulate manual_step) total_loss_manual = 0.0 for i in range(num_chunks): start_idx = i * chunk_size end_idx = min(start_idx + chunk_size, len(y_list)) - + y_chunk = torch_geometric.data.Batch.from_data_list(y_list[start_idx:end_idx]) x_target_chunk = torch_geometric.data.Batch.from_data_list(x_target_list[start_idx:end_idx]) - + xhat_chunk = model_manual.xhat(y_chunk, sigma_manual) loss_chunk, aux_chunk = model_manual.compute_loss(x_target_chunk, xhat_chunk, sigma_manual) loss_chunk_mean = loss_chunk.mean() - + # Scale loss by number of chunks (the fix we implemented) scaled_loss = loss_chunk_mean / num_chunks scaled_loss.backward() - + total_loss_manual += loss_chunk_mean.item() - + gradients_manual = get_model_gradients(model_manual) grad_norm_manual = compute_gradient_norm(gradients_manual) - + print(f" Manual total loss: {total_loss_manual:.6f}, grad_norm: {grad_norm_manual:.6f}") - + # Compare gradients print("\n9. COMPARING GRADIENTS:") - print(f" Gradient norm ratio (manual/auto): {grad_norm_manual/grad_norm_auto:.6f}") - print(f" Loss ratio (manual/auto): {total_loss_manual/loss_auto_mean.item():.6f}") - + print(f" Gradient norm ratio (manual/auto): {grad_norm_manual / grad_norm_auto:.6f}") + print(f" Loss ratio (manual/auto): {total_loss_manual / loss_auto_mean.item():.6f}") + print("\n Parameter-wise comparison:") gradients_match = compare_gradients(gradients_auto, gradients_manual, tolerance=1e-2) - - print("\n" + "="*80) + + print("\n" + "=" * 80) if gradients_match: print("āœ… SUCCESS: Manual and automatic optimization produce equivalent gradients!") print(f" Gradient norms: auto={grad_norm_auto:.6f}, manual={grad_norm_manual:.6f}") - print(f" Relative difference: {abs(grad_norm_manual-grad_norm_auto)/grad_norm_auto*100:.3f}%") + print(f" Relative difference: {abs(grad_norm_manual - grad_norm_auto) / grad_norm_auto * 100:.3f}%") else: print("āŒ FAILURE: Gradients do not match within tolerance!") print(" This indicates an issue with the manual optimization implementation.") - print("="*80) - + print("=" * 80) + return gradients_match + if __name__ == "__main__": - test_gradient_equivalence() \ No newline at end of file + test_gradient_equivalence() diff --git a/scratch/test_multimeasurement.py b/scratch/test_multimeasurement.py index f9c7c9a..4b04537 100644 --- a/scratch/test_multimeasurement.py +++ b/scratch/test_multimeasurement.py @@ -1,41 +1,41 @@ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys import os +import sys + +import dotenv import hydra -from omegaconf import OmegaConf +import lightning.pytorch as pl import torch -import torch_geometric -from jamun.hydra import instantiate_dict_cfg -import pdb -import jamun +from omegaconf import OmegaConf + from jamun.utils import compute_average_squared_distance_from_datasets -import lightning.pytorch as pl # Fix PyTorch Geometric backend issues try: + import torch_cluster import torch_scatter import torch_sparse - import torch_cluster except ImportError: print("Warning: Some PyTorch Geometric extensions not available") # Use GPU if available -device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +device = torch.device("cuda" if torch.cuda.is_available() else "cpu") print(f"Using device: {device}") -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") -project_root = "/homefs/home/sules/jamun" # Adjust if necessary +project_root = "/homefs/home/sules/jamun" # Adjust if necessary if project_root not in sys.path: sys.path.insert(0, project_root) print(f"Added '{project_root}' to sys.path for module discovery.") else: print(f"'{project_root}' is already in sys.path.") + def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the average squared distance for normalization from the data.""" datamodule = hydra.utils.instantiate(cfg.data.datamodule) @@ -45,30 +45,31 @@ def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) return average_squared_distance + def test_training_mode(cfg, mode_name, max_graphs_per_batch): """Test training with specified optimization mode.""" - print(f"\n{'='*50}") + print(f"\n{'=' * 50}") print(f"Testing {mode_name} mode (max_graphs_per_batch={max_graphs_per_batch})") - print(f"{'='*50}") - + print(f"{'=' * 50}") + # # Configure DenoiserMultimeasurement # cfg.model._target_ = "jamun.model.denoiser_multimeasurement.DenoiserMultimeasurement" # cfg.model.sigma_distribution._target_ = "jamun.distributions.ConstantSigma" # cfg.model.sigma_distribution.sigma = 0.04 - + # Set multimeasurement parameters # cfg.model.multimeasurement = True cfg.model.N_measurements_hidden = 2 cfg.model.N_measurements = 2 cfg.model.max_graphs_per_batch = max_graphs_per_batch - + # Compute normalization average_squared_distance = compute_average_squared_distance_from_config(cfg) cfg.model.average_squared_distance = average_squared_distance breakpoint() print("Loading datamodule...") datamodule = hydra.utils.instantiate(cfg.data.datamodule) - datamodule.setup('test') + datamodule.setup("test") breakpoint() print("Loading model...") model = hydra.utils.instantiate(cfg.model) @@ -86,7 +87,7 @@ def test_training_mode(cfg, mode_name, max_graphs_per_batch): breakpoint() print(f"Batch shape: {batch.pos.shape}") print(f"Batch num_graphs: {batch.num_graphs}") - if hasattr(batch, 'hidden_state') and batch.hidden_state is not None: + if hasattr(batch, "hidden_state") and batch.hidden_state is not None: print(f"Hidden state shapes: {[h.shape for h in batch.hidden_state]}") else: print("No hidden states in batch") @@ -101,17 +102,17 @@ def test_training_mode(cfg, mode_name, max_graphs_per_batch): print(f"Noisy shape: {y.pos.shape}") print(f"Output shape: {xhat.pos.shape}") print(f"Target shape: {x_target.pos.shape}") - + # Verify multimeasurement expansion expected_graphs = batch.num_graphs * model.N_measurements_hidden * model.N_measurements actual_graphs = y.num_graphs print(f"Expected graphs after multimeasurement: {expected_graphs}") print(f"Actual graphs: {actual_graphs}") assert actual_graphs == expected_graphs, f"Graph count mismatch: expected {expected_graphs}, got {actual_graphs}" - + # Test actual training with fast_dev_run print("Testing training with fast_dev_run...") - + # Configure trainer to use only 1 GPU if torch.cuda.is_available(): print(f"CUDA available with {torch.cuda.device_count()} GPUs - using GPU 0 only") @@ -121,9 +122,9 @@ def test_training_mode(cfg, mode_name, max_graphs_per_batch): logger=False, enable_progress_bar=True, enable_model_summary=False, - accelerator='gpu', + accelerator="gpu", devices=[0], # Explicitly use only GPU 0 - strategy='auto', # Single device strategy + strategy="auto", # Single device strategy ) else: print("CUDA not available - using CPU") @@ -133,40 +134,42 @@ def test_training_mode(cfg, mode_name, max_graphs_per_batch): logger=False, enable_progress_bar=True, enable_model_summary=False, - accelerator='cpu', + accelerator="cpu", devices=1, ) - + try: trainer.fit(model, datamodule) print(f"{mode_name} mode training completed successfully!") except Exception as e: print(f"Error during training: {e}") raise - + print(f"{mode_name} mode test completed successfully!") return model + @hydra.main(version_base=None, config_path="../src/jamun/hydra_config", config_name="train") def main(cfg): # # Override data config to use only ALA_ALA # cfg.data.datamodule.filter_codes = ['ALA_ALA'] # cfg.data.datamodule.subsample = 10 # Use fewer samples for faster testing # cfg.data.datamodule.batch_size = 4 # Small batch size for testing - + print("Testing DenoiserMultimeasurement training modes") print(f"Using ALA_ALA data from: {JAMUN_DATA_PATH}") - + # Test automatic optimization mode model_auto = test_training_mode(cfg.copy(), "AUTOMATIC", None) - + # Test manual optimization mode model_manual = test_training_mode(cfg.copy(), "MANUAL", 2) # Process 2 graphs at a time - - print(f"\n{'='*50}") + + print(f"\n{'=' * 50}") print("ALL TESTS PASSED!") print("Both automatic and manual optimization modes work correctly.") - print(f"{'='*50}") + print(f"{'=' * 50}") + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/test_reorganize_swarm_data.py b/scratch/test_reorganize_swarm_data.py index 404cf3c..456d2b9 100644 --- a/scratch/test_reorganize_swarm_data.py +++ b/scratch/test_reorganize_swarm_data.py @@ -7,16 +7,15 @@ """ import os -import shutil -import tempfile import sys -from pathlib import Path +import tempfile # Add the scratch directory to the path so we can import the main script sys.path.insert(0, os.path.dirname(os.path.abspath(__file__))) try: import mdtraj as md + MDTRAJ_AVAILABLE = True except ImportError: MDTRAJ_AVAILABLE = False @@ -24,37 +23,39 @@ try: from tqdm import tqdm + TQDM_AVAILABLE = True except ImportError: TQDM_AVAILABLE = False print("āš ļø tqdm not available. Progress bars will be disabled.") + def create_test_data(test_source_dir: str, test_pdb_file: str): """Create a small test dataset structure.""" print("Creating test data structure...") - + # Create test source directory os.makedirs(test_source_dir, exist_ok=True) - + # Create a few test grid directories with mock files - test_grid_codes = ['000', '001', '002', '003', '004'] - trajectory_codes = ['001', '002', '003', '004', '005'] - + test_grid_codes = ["000", "001", "002", "003", "004"] + trajectory_codes = ["001", "002", "003", "004", "005"] + for grid_code in test_grid_codes: grid_dir = os.path.join(test_source_dir, f"AA_{grid_code}") os.makedirs(grid_dir, exist_ok=True) - + # Create mock .xtc files for traj_code in trajectory_codes: xtc_file = os.path.join(grid_dir, f"swarm_1ps_{traj_code}.xtc") # Create empty files that will be copied, but note they won't be valid XTC format # This is just for testing the file organization logic - with open(xtc_file, 'w') as f: + with open(xtc_file, "w") as f: f.write(f"Mock XTC data for grid {grid_code}, trajectory {traj_code}\n") - + # Create mock single PDB file with more realistic content os.makedirs(os.path.dirname(test_pdb_file), exist_ok=True) - with open(test_pdb_file, 'w') as f: + with open(test_pdb_file, "w") as f: # Write a minimal but valid PDB structure for 2 alanine residues f.write("TITLE MOCK ALA-ALA DIPEPTIDE\n") f.write("ATOM 1 N ALA A 1 0.000 0.000 0.000 1.00 0.00 N\n") @@ -80,9 +81,10 @@ def create_test_data(test_source_dir: str, test_pdb_file: str): f.write("ATOM 21 OXT ALA A 2 6.032 3.829 0.000 1.00 0.00 O\n") f.write("TER 22 ALA A 2\n") f.write("END\n") - + print(f"Created test data with {len(test_grid_codes)} grid codes") + def test_mdtraj_with_mock_data(train_dir: str, val_dir: str): """ Test mdtraj functionality with mock data. @@ -91,35 +93,38 @@ def test_mdtraj_with_mock_data(train_dir: str, val_dir: str): if not MDTRAJ_AVAILABLE: print("āš ļø mdtraj not available, skipping trajectory compatibility tests") return - + print("Testing mdtraj compatibility (expected to fail with mock data)...") - + for split_name, split_dir in [("train", train_dir), ("val", val_dir)]: - xtc_files = [f for f in os.listdir(split_dir) if f.endswith('.xtc')] + xtc_files = [f for f in os.listdir(split_dir) if f.endswith(".xtc")] if not xtc_files: continue - + # Test one file from each split test_file = xtc_files[0] - base_name = test_file.replace('.xtc', '') + base_name = test_file.replace(".xtc", "") pdb_file = f"{base_name}.pdb" - + xtc_path = os.path.join(split_dir, test_file) pdb_path = os.path.join(split_dir, pdb_file) - + try: # This will likely fail since we have mock XTC data traj = md.load(xtc_path, top=pdb_path) - print(f"āœ… {split_name}: Successfully loaded {test_file} + {pdb_file} " - f"({traj.n_frames} frames, {traj.n_atoms} atoms)") + print( + f"āœ… {split_name}: Successfully loaded {test_file} + {pdb_file} " + f"({traj.n_frames} frames, {traj.n_atoms} atoms)" + ) del traj except Exception as e: print(f"āŒ {split_name}: Failed to load {test_file} + {pdb_file}: {str(e)}") print(" (This is expected with mock data - real data should work)") - + print("Note: mdtraj tests with mock data are expected to fail.") print("The real script will test with actual trajectory files.") + def test_reorganization(): """Test the reorganization script with mock data.""" with tempfile.TemporaryDirectory() as temp_dir: @@ -127,127 +132,131 @@ def test_reorganization(): test_source = os.path.join(temp_dir, "test_swarm_results") test_target = os.path.join(temp_dir, "test_enhanced") test_pdb = os.path.join(temp_dir, "test_ALA_ALA.pdb") - + # Create test data create_test_data(test_source, test_pdb) - + # Import and modify the main script for testing import reorganize_swarm_data - + # Temporarily override the configuration original_source = reorganize_swarm_data.SOURCE_DIR original_pdb = reorganize_swarm_data.SINGLE_PDB_FILE original_strategies = reorganize_swarm_data.SPLITTING_STRATEGIES.copy() - + reorganize_swarm_data.SOURCE_DIR = test_source reorganize_swarm_data.SINGLE_PDB_FILE = test_pdb - + # Override the splitting strategies for testing reorganize_swarm_data.SPLITTING_STRATEGIES = { - 'grid_split': { - 'target_dir': test_target + "_grid_split", - 'train_size': 3, # Use 3 for train, 2 for val - 'description': "Test grid split" + "grid_split": { + "target_dir": test_target + "_grid_split", + "train_size": 3, # Use 3 for train, 2 for val + "description": "Test grid split", + }, + "trajectory_split": { + "target_dir": test_target + "_trajectory_split", + "train_trajectories": ["001", "002", "003"], # First 3 for testing + "val_trajectories": ["004", "005"], # Last 2 for testing + "description": "Test trajectory split", }, - 'trajectory_split': { - 'target_dir': test_target + "_trajectory_split", - 'train_trajectories': ['001', '002', '003'], # First 3 for testing - 'val_trajectories': ['004', '005'], # Last 2 for testing - 'description': "Test trajectory split" - } } - + try: - print("\n" + "="*50) + print("\n" + "=" * 50) print("RUNNING TEST REORGANIZATION") - print("="*50) - + print("=" * 50) + # Test both strategies print("Testing grid split strategy...") - reorganize_swarm_data.main('grid_split') - + reorganize_swarm_data.main("grid_split") + print("Testing trajectory split strategy...") - reorganize_swarm_data.main('trajectory_split') - + reorganize_swarm_data.main("trajectory_split") + # Verify results - print("\n" + "="*50) + print("\n" + "=" * 50) print("VERIFYING TEST RESULTS") - print("="*50) - + print("=" * 50) + # Check both strategies - for strategy_name in ['grid_split', 'trajectory_split']: + for strategy_name in ["grid_split", "trajectory_split"]: strategy_dir = test_target + f"_{strategy_name}" - train_dir = os.path.join(strategy_dir, 'train') - val_dir = os.path.join(strategy_dir, 'val') - + train_dir = os.path.join(strategy_dir, "train") + val_dir = os.path.join(strategy_dir, "val") + if os.path.exists(train_dir) and os.path.exists(val_dir): train_files = os.listdir(train_dir) val_files = os.listdir(val_dir) - - train_xtc = [f for f in train_files if f.endswith('.xtc')] - train_pdb = [f for f in train_files if f.endswith('.pdb')] - val_xtc = [f for f in val_files if f.endswith('.xtc')] - val_pdb = [f for f in val_files if f.endswith('.pdb')] - + + train_xtc = [f for f in train_files if f.endswith(".xtc")] + train_pdb = [f for f in train_files if f.endswith(".pdb")] + val_xtc = [f for f in val_files if f.endswith(".xtc")] + val_pdb = [f for f in val_files if f.endswith(".pdb")] + print(f"\n{strategy_name.upper()} STRATEGY:") print(f"Train directory: {len(train_files)} files ({len(train_xtc)} .xtc, {len(train_pdb)} .pdb)") print(f"Val directory: {len(val_files)} files ({len(val_xtc)} .xtc, {len(val_pdb)} .pdb)") - + # Calculate expected files based on strategy - if strategy_name == 'grid_split': + if strategy_name == "grid_split": # 3 grid codes Ɨ 5 trajectories Ɨ 2 file types = 30 train files # 2 grid codes Ɨ 5 trajectories Ɨ 2 file types = 20 val files expected_train = 3 * 5 * 2 expected_val = 2 * 5 * 2 else: # trajectory_split - # 5 grid codes Ɨ 3 trajectories Ɨ 2 file types = 30 train files + # 5 grid codes Ɨ 3 trajectories Ɨ 2 file types = 30 train files # 5 grid codes Ɨ 2 trajectories Ɨ 2 file types = 20 val files expected_train = 5 * 3 * 2 expected_val = 5 * 2 * 2 - + if len(train_files) == expected_train and len(val_files) == expected_val: print(f"āœ… {strategy_name} Test PASSED! File counts are correct.") - + # Check a few file names print("Sample train files:", sorted(train_files)[:3]) print("Sample val files:", sorted(val_files)[:3]) else: - print(f"āŒ {strategy_name} Test FAILED! Expected {expected_train} train, {expected_val} val files") + print( + f"āŒ {strategy_name} Test FAILED! Expected {expected_train} train, {expected_val} val files" + ) return False else: print(f"āŒ {strategy_name} Test FAILED! Output directories were not created") return False - + # Test mdtraj compatibility on one strategy strategy_dir = test_target + "_grid_split" - train_dir = os.path.join(strategy_dir, 'train') - val_dir = os.path.join(strategy_dir, 'val') + train_dir = os.path.join(strategy_dir, "train") + val_dir = os.path.join(strategy_dir, "val") print("\n=== Testing mdtraj compatibility ===") test_mdtraj_with_mock_data(train_dir, val_dir) - + print("\nāœ… All tests completed successfully!") return True - + finally: # Restore original configuration reorganize_swarm_data.SOURCE_DIR = original_source reorganize_swarm_data.SINGLE_PDB_FILE = original_pdb reorganize_swarm_data.SPLITTING_STRATEGIES = original_strategies + def main(): """Run the test.""" print("Starting test of swarm data reorganization script...") - + success = test_reorganization() - + if success: print("\nšŸŽ‰ Test passed! The script should work correctly on the real data.") print("\nTo run the full reorganization, execute:") print("python scratch/reorganize_swarm_data.py") else: print("\nāŒ Test failed! Please check the script before running on real data.") - + return success + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/test_repeated_position_dataset.py b/scratch/test_repeated_position_dataset.py index 87177ce..2474f54 100644 --- a/scratch/test_repeated_position_dataset.py +++ b/scratch/test_repeated_position_dataset.py @@ -1,46 +1,49 @@ #!/usr/bin/env python3 """ -Test script for RepeatedPositionDataset to verify that hidden states +Test script for RepeatedPositionDataset to verify that hidden states are exact copies of the current position. """ -import torch import os -from pathlib import Path # Add the src directory to the path so we can import jamun modules import sys +from pathlib import Path + +import torch + sys.path.insert(0, str(Path(__file__).parent / "src")) from jamun.data.noisy_position_dataset import RepeatedPositionDataset + def test_repeated_position_dataset(): """Test RepeatedPositionDataset with ALA_ALA capped diamines data.""" - + print("Testing RepeatedPositionDataset...") print("=" * 50) - + # Set up dataset parameters root = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train" traj_files = ["ALA_ALA.xtc"] pdb_file = "ALA_ALA.pdb" label = "ALA_ALA_test" total_lag_time = 4 # This should create 3 hidden states (4 - 1 = 3) - + # Check if files exist xtc_path = os.path.join(root, traj_files[0]) pdb_path = os.path.join(root, pdb_file) - + if not os.path.exists(xtc_path): print(f"āŒ ERROR: XTC file not found: {xtc_path}") return False if not os.path.exists(pdb_path): print(f"āŒ ERROR: PDB file not found: {pdb_path}") return False - + print(f"āœ… Found XTC file: {xtc_path}") print(f"āœ… Found PDB file: {pdb_path}") - + try: # Create the dataset print(f"\nCreating RepeatedPositionDataset with total_lag_time={total_lag_time}...") @@ -51,75 +54,77 @@ def test_repeated_position_dataset(): label=label, total_lag_time=total_lag_time, num_frames=5, # Only load 5 frames for testing - verbose=True + verbose=True, ) - - print(f"āœ… Dataset created successfully") + + print("āœ… Dataset created successfully") print(f" Dataset length: {len(dataset)}") print(f" Dataset label: {dataset.label()}") - + # Test a few samples - print(f"\nTesting samples...") - + print("\nTesting samples...") + for idx in range(min(3, len(dataset))): print(f"\n--- Sample {idx} ---") - + # Get sample from dataset graph = dataset[idx] - + print(f"Graph pos shape: {graph.pos.shape}") print(f"Number of hidden states: {len(graph.hidden_state)}") - + # Verify we have the expected number of hidden states expected_hidden_states = total_lag_time - 1 if len(graph.hidden_state) != expected_hidden_states: print(f"āŒ ERROR: Expected {expected_hidden_states} hidden states, got {len(graph.hidden_state)}") return False - + print(f"āœ… Correct number of hidden states: {len(graph.hidden_state)}") - + # Test each hidden state for i, hidden_pos in enumerate(graph.hidden_state): print(f"Hidden state {i} shape: {hidden_pos.shape}") - + # Check if shapes match if hidden_pos.shape != graph.pos.shape: print(f"āŒ ERROR: Shape mismatch! pos: {graph.pos.shape}, hidden_state[{i}]: {hidden_pos.shape}") return False - + # Check if values are exactly equal if not torch.allclose(hidden_pos, graph.pos, atol=1e-10): print(f"āŒ ERROR: Hidden state {i} is not exactly equal to current position!") print(f" Max difference: {torch.max(torch.abs(hidden_pos - graph.pos)).item()}") return False - + # Check if they are the exact same tensor (should be different objects but same values) if hidden_pos is graph.pos: print(f"āš ļø WARNING: Hidden state {i} is the same object as pos (should be different objects)") else: print(f"āœ… Hidden state {i} is a different object with same values as pos") - + print(f"āœ… Hidden state {i} exactly matches current position") - - print(f"\nšŸŽ‰ All tests passed!") - print(f" āœ… Dataset loads correctly") + + print("\nšŸŽ‰ All tests passed!") + print(" āœ… Dataset loads correctly") print(f" āœ… Correct number of hidden states ({total_lag_time - 1})") - print(f" āœ… Hidden states exactly match current position") - print(f" āœ… Hidden states are separate objects (not references)") - + print(" āœ… Hidden states exactly match current position") + print(" āœ… Hidden states are separate objects (not references)") + return True - + except Exception as e: print(f"āŒ ERROR: Exception occurred: {e}") import traceback + traceback.print_exc() return False + if __name__ == "__main__": success = test_repeated_position_dataset() if success: - print(f"\nšŸŽ‰ Test completed successfully!") + print("\nšŸŽ‰ Test completed successfully!") exit(0) else: - print(f"\nšŸ’„ Test failed!") - exit(1) \ No newline at end of file + print("\nšŸ’„ Test failed!") + exit(1) diff --git a/scratch/training_prototype.py b/scratch/training_prototype.py index 5d99f02..8452e53 100644 --- a/scratch/training_prototype.py +++ b/scratch/training_prototype.py @@ -1,44 +1,41 @@ # %% Imports and Basic Setup -import functools import logging import os import sys -from typing import Union import dotenv -import tqdm # Often used in notebooks, can be optional in scripts -import torch import e3nn -import e3tools.nn import hydra -from hydra import compose, initialize -from omegaconf import OmegaConf import lightning.pytorch as pl +import torch import torch_geometric.data +from hydra import compose, initialize +from omegaconf import OmegaConf import jamun import jamun.data import jamun.distributions import jamun.model import jamun.model.arch +from jamun.hydra import instantiate_dict_cfg + # Assuming these are in jamun.utils and jamun.hydra respectively from jamun.utils import compute_average_squared_distance_from_datasets, find_checkpoint -from jamun.hydra import instantiate_dict_cfg # --- Basic Setup --- logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) py_logger = logging.getLogger("jamun_script") -torch.cuda.is_available() # Good to check, but PL trainer will also handle device +torch.cuda.is_available() # Good to check, but PL trainer will also handle device torch.set_float32_matmul_precision("high") e3nn.set_optimization_defaults(jit_script_fx=False) # %% Environment and Paths -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") -project_root = "/homefs/home/sules/jamun" # Adjust if necessary +project_root = "/homefs/home/sules/jamun" # Adjust if necessary if project_root not in sys.path: sys.path.insert(0, project_root) py_logger.info(f"Added '{project_root}' to sys.path for module discovery.") @@ -50,16 +47,16 @@ # Adjust config_path relative to the script's location if it's not in scratch/ # If jamun_training_script.py is in scratch/, and configs are in jamun/configs/ # then config_path should be "../configs" -with initialize(config_path=".", job_name="conditioning_initial_run"): # Corrected job_name from previous context +with initialize(config_path=".", job_name="conditioning_initial_run"): # Corrected job_name from previous context cfg = compose( - config_name="config", # Main config file + config_name="config", # Main config file overrides=[ - "model.arch._target_=scratch.e3conv_test.E3Conv", # Relative to project_root - "model._target_=scratch.denoiser_test.Denoiser", # Relative to project_root + "model.arch._target_=scratch.e3conv_test.E3Conv", # Relative to project_root + "model._target_=scratch.denoiser_test.Denoiser", # Relative to project_root "+model.arch.N_structures=2", - "trainer.max_epochs=100", # Example: train for 10 epochs + "trainer.max_epochs=100", # Example: train for 10 epochs # Add other overrides, e.g. "logger=null" if you don't want default loggers for a quick test - ] + ], ) py_logger.info("Loaded configuration:") py_logger.info(OmegaConf.to_yaml(cfg)) @@ -68,13 +65,13 @@ # %% Initial Dataset Setup (for model properties like average_squared_distance) py_logger.info("Setting up initial dataset for model properties...") initial_datasets_for_props = { - "props_dataset": jamun.data.parse_datasets_from_directory( # Renamed key for clarity + "props_dataset": jamun.data.parse_datasets_from_directory( # Renamed key for clarity root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", traj_pattern="^(.*)-traj-arrays.npz", pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], + filter_codes=["AA"], as_iterable=False, - subsample=100, # Keep this small for this purpose + subsample=100, # Keep this small for this purpose max_datasets=1, ) } @@ -85,27 +82,30 @@ if not hasattr(cfg.model, "average_squared_distance") or cfg.model.average_squared_distance is None: py_logger.info("Computing average_squared_distance for the model...") average_squared_distance = compute_average_squared_distance_from_datasets( - initial_datasets_for_props['props_dataset'], # Use the small dataset for this - cfg.model.max_radius + initial_datasets_for_props["props_dataset"], # Use the small dataset for this + cfg.model.max_radius, ) cfg.model.average_squared_distance = average_squared_distance py_logger.info(f"Set cfg.model.average_squared_distance to {cfg.model.average_squared_distance}") - + # Provide conditioner if needed if not hasattr(cfg.model, "conditioner"): OmegaConf.set_struct(cfg.model, False) # Allow modification cfg.model.conditioner = OmegaConf.create({}) - cfg.model.conditioner._target_ = 'scratch.conditioners.SelfConditioner' # Use the SelfConditioner from scratch.conditioners + cfg.model.conditioner._target_ = ( + "scratch.conditioners.SelfConditioner" # Use the SelfConditioner from scratch.conditioners + ) OmegaConf.set_struct(cfg.model, True) # Lock structure again py_logger.info(f"Set cfg.model.conditioner to instantiate {cfg.model.conditioner._target_}.") model = hydra.utils.instantiate(cfg.model) - + py_logger.info("Successfully instantiated model.") py_logger.info(f"Instantiated model architecture type: {type(model.g)}") except Exception as e: py_logger.error(f"Error during model instantiation: {e}") import traceback + traceback.print_exc() sys.exit(1) @@ -125,7 +125,7 @@ # After .to(device), the model's internal device attribute should update. # We can also check a parameter's device as a fallback verification. final_model_device = None -if hasattr(model, 'device') and model.device is not None: +if hasattr(model, "device") and model.device is not None: final_model_device = model.device elif next(model.parameters(), None) is not None: final_model_device = next(model.parameters()).device @@ -142,19 +142,19 @@ # You might want to parse a larger dataset here for actual training. py_logger.info("Parsing dataset for training...") training_dataset_source = jamun.data.parse_datasets_from_directory( - root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", # Consider using full dataset + root=f"{JAMUN_DATA_PATH}/timewarp/2AA-1-large/train/", # Consider using full dataset traj_pattern="^(.*)-traj-arrays.npz", pdb_file="AA-traj-state0.pdb", - filter_codes=['AA'], - as_iterable=False, # Set to True for very large datasets if memory is an issue - subsample=cfg.data.datamodule.datasets.train[0].subsample, # Use subsample from config or None for full - max_datasets=1, # Use from config or None for all + filter_codes=["AA"], + as_iterable=False, # Set to True for very large datasets if memory is an issue + subsample=cfg.data.datamodule.datasets.train[0].subsample, # Use subsample from config or None for full + max_datasets=1, # Use from config or None for all ) datasets_for_training = { "train": training_dataset_source, - "val": training_dataset_source, # Replace with actual validation set - "test": training_dataset_source, # Replace with actual test set + "val": training_dataset_source, # Replace with actual validation set + "test": training_dataset_source, # Replace with actual test set } py_logger.info(f"Prepared datasets for training: { {k: type(v).__name__ for k, v in datasets_for_training.items()} }") if isinstance(training_dataset_source, torch_geometric.data.Dataset): @@ -178,12 +178,12 @@ try: # This requires JAMUN_ROOT_PATH, task_name, run_group, and run_key to be correctly # defined and resolvable in your configuration. - wandb_save_dir = str(cfg.paths.run_path) # Resolve the path from OmegaConf + wandb_save_dir = str(cfg.paths.run_path) # Resolve the path from OmegaConf # Update the logger config before instantiation OmegaConf.update(cfg, "logger.wandb.save_dir", wandb_save_dir, merge=False) py_logger.info(f"Explicitly setting WandbLogger save_dir to: {wandb_save_dir}") - + # Ensure the target directory for wandb files exists. # WandbLogger will create its 'wandb/' subdirectory and run-specific folders inside this save_dir. os.makedirs(wandb_save_dir, exist_ok=True) @@ -196,21 +196,24 @@ # 1. Instantiate Loggers and Callbacks from Hydra config loggers_list = [] if cfg.get("logger"): - if hasattr(cfg.logger, '_target_'): - loggers_list.append(hydra.utils.instantiate(cfg.logger)) - else: + if hasattr(cfg.logger, "_target_"): + loggers_list.append(hydra.utils.instantiate(cfg.logger)) + else: # Assuming instantiate_dict_cfg iterates and calls hydra.utils.instantiate for each logger config - loggers_list = instantiate_dict_cfg(cfg.logger) + loggers_list = instantiate_dict_cfg(cfg.logger) py_logger.info(f"Instantiated loggers: {[type(l).__name__ for l in loggers_list]}") # %% 2. Determine and set the ModelCheckpoint directory path final_checkpoint_dir = None # Check if the first logger is a WandbLogger and provides a directory -if (loggers_list and - isinstance(loggers_list[0], pl.loggers.WandbLogger) and - hasattr(loggers_list[0], 'experiment') and loggers_list[0].experiment and - hasattr(loggers_list[0].experiment, 'dir') and loggers_list[0].experiment.dir): - +if ( + loggers_list + and isinstance(loggers_list[0], pl.loggers.WandbLogger) + and hasattr(loggers_list[0], "experiment") + and loggers_list[0].experiment + and hasattr(loggers_list[0].experiment, "dir") + and loggers_list[0].experiment.dir +): wandb_run_root_dir = loggers_list[0].experiment.dir # Adjust if wandb_run_root_dir points to a 'files' subdirectory if os.path.basename(wandb_run_root_dir) == "files": @@ -223,7 +226,9 @@ if not loggers_list: py_logger.info(f"No loggers configured. Defaulting checkpoint directory: {final_checkpoint_dir}") else: - py_logger.info(f"First logger is not a suitable WandB logger. Defaulting checkpoint directory: {final_checkpoint_dir}") + py_logger.info( + f"First logger is not a suitable WandB logger. Defaulting checkpoint directory: {final_checkpoint_dir}" + ) # Update the config if model_checkpoint callback is defined if cfg.get("callbacks") and cfg.callbacks.get("model_checkpoint"): @@ -240,10 +245,10 @@ # %% Instantiate Callbacks callbacks_list = [] if cfg.get("callbacks"): - if hasattr(cfg.callbacks, '_target_'): # Single callback config + if hasattr(cfg.callbacks, "_target_"): # Single callback config callbacks_list.append(hydra.utils.instantiate(cfg.callbacks)) - else: # Dictionary of callback configs - callbacks_list = instantiate_dict_cfg(cfg.callbacks) # This will now use the modified dirpath + else: # Dictionary of callback configs + callbacks_list = instantiate_dict_cfg(cfg.callbacks) # This will now use the modified dirpath py_logger.info(f"Instantiated callbacks: {[type(c).__name__ for c in callbacks_list]}") # 2. Instantiate PyTorch Lightning Trainer @@ -284,9 +289,9 @@ py_logger.info("Starting training...") try: trainer.fit( - model=model, # Use the main model instance + model=model, # Use the main model instance datamodule=datamodule_for_training, - ckpt_path=checkpoint_path if checkpoint_path else None + ckpt_path=checkpoint_path if checkpoint_path else None, ) py_logger.info("Training finished.") @@ -304,17 +309,19 @@ # %% Log the final configuration and save it locally wandb_logger_instance = None # loggers_list should still be in scope from when it was passed to the Trainer -for logger_from_list in loggers_list: # Use a different variable name to avoid conflict if logger is defined elsewhere +for logger_from_list in loggers_list: # Use a different variable name to avoid conflict if logger is defined elsewhere if isinstance(logger_from_list, pl.loggers.WandbLogger): wandb_logger_instance = logger_from_list break -if wandb_logger_instance and hasattr(wandb_logger_instance, 'experiment') and wandb_logger_instance.experiment: - py_logger.info(f"WandbLogger experiment active (run_id: {wandb_logger_instance.experiment.id}). Logging final script config to wandb.") +if wandb_logger_instance and hasattr(wandb_logger_instance, "experiment") and wandb_logger_instance.experiment: + py_logger.info( + f"WandbLogger experiment active (run_id: {wandb_logger_instance.experiment.id}). Logging final script config to wandb." + ) # Convert the current OmegaConf object 'cfg' to a plain dictionary # This 'cfg' includes all modifications made throughout the script final_script_cfg_dict = OmegaConf.to_container(cfg, resolve=True, throw_on_missing=True) - + try: wandb_logger_instance.experiment.config.update( {"cfg": final_script_cfg_dict, "jamun_version_at_end": jamun.__version__, "script_cwd_at_end": os.getcwd()} @@ -324,32 +331,38 @@ py_logger.error(f"Failed to update wandb.config with final script config: {e}") else: if cfg.get("logger") and cfg.logger.get("wandb"): - py_logger.warning("WandbLogger was configured but not found or experiment not active at script end. Final script config not logged to wandb.config.") + py_logger.warning( + "WandbLogger was configured but not found or experiment not active at script end. Final script config not logged to wandb.config." + ) # 2. Explicitly save the final state of the OmegaConf object 'cfg' to a local file final_config_output_dir = None if cfg.get("logger") and cfg.logger.get("wandb") and cfg.logger.wandb.get("save_dir"): final_config_output_dir = cfg.logger.wandb.save_dir -elif 'wandb_save_dir' in locals() and wandb_save_dir: # If it was set in a previous cell - final_config_output_dir = wandb_save_dir +elif "wandb_save_dir" in locals() and wandb_save_dir: # If it was set in a previous cell + final_config_output_dir = wandb_save_dir else: # Fallback if a specific run directory isn't easily available # This might not be ideal as it won't be co-located with W&B run files if save_dir wasn't set - final_config_output_dir = os.path.join(os.getcwd(), "outputs", cfg.get("task_name", "unknown_task"), cfg.get("run_key", "unknown_run")) + final_config_output_dir = os.path.join( + os.getcwd(), "outputs", cfg.get("task_name", "unknown_task"), cfg.get("run_key", "unknown_run") + ) os.makedirs(final_config_output_dir, exist_ok=True) if final_config_output_dir: final_config_path = os.path.join(final_config_output_dir, "final_resolved_script_config.yaml") try: - with open(final_config_path, 'w') as f: + with open(final_config_path, "w") as f: OmegaConf.save(config=cfg, f=f) py_logger.info(f"Final script configuration saved locally to: {final_config_path}") except Exception as e: py_logger.error(f"Failed to save final script configuration locally: {e}") else: - py_logger.warning("Could not determine a definitive output directory for final_resolved_script_config.yaml. Not saving locally.") + py_logger.warning( + "Could not determine a definitive output directory for final_resolved_script_config.yaml. Not saving locally." + ) loggers_list[0].experiment.finish() if loggers_list and isinstance(loggers_list[0], pl.loggers.WandbLogger) else None py_logger.info("Script finished.") diff --git a/scratch/transformer/convert_spatiotemporal.py b/scratch/transformer/convert_spatiotemporal.py index ef95b0f..2cb5053 100644 --- a/scratch/transformer/convert_spatiotemporal.py +++ b/scratch/transformer/convert_spatiotemporal.py @@ -10,33 +10,33 @@ def calculate_temporal_positions(temporal_length, device=None): """ Calculate normalized temporal positions for nodes in a temporal graph. - + Args: temporal_length: Total number of nodes in the temporal sequence device: Device to create tensors on - + Returns: torch.Tensor: Normalized positions [0, 1/T, 2/T, ..., (T-1)/T] """ if temporal_length <= 1: return torch.tensor([0.0], device=device) - + # Create positions [0, 1, 2, ..., T-1] and normalize by T positions = torch.arange(temporal_length, dtype=torch.float32, device=device) normalized_positions = positions / temporal_length - + return normalized_positions def spatial_to_temporal_graphs(batch, graph_type="fan"): """ Convert a batch of spatial graphs to temporal graphs with configurable connectivity. - + For each spatial node with position + hidden states, create a temporal graph where: - Node 0: current position - Nodes 1-T: hidden state positions - Connectivity depends on graph_type parameter - + Args: batch: Input spatial graph batch graph_type: Type of connectivity to use @@ -45,57 +45,51 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): - "complete": Complete graph with self-loops (all-to-all including self) - "complete_no_self": Complete graph without self-loops (all-to-all excluding self) """ - import torch_geometric - + # Validate graph_type valid_types = ["fan", "hub_n_spoke", "complete", "complete_no_self"] if graph_type not in valid_types: raise ValueError(f"graph_type must be one of {valid_types}, got {graph_type}") - + # Get device from input batch device = batch.pos.device - + # Get dimensions num_spatial_nodes = batch.pos.shape[0] - + # Check if we have hidden states - if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + if hasattr(batch, "hidden_state") and batch.hidden_state is not None and len(batch.hidden_state) > 0: num_hidden_states = len(batch.hidden_state) temporal_length = 1 + num_hidden_states # current + hidden else: # If no hidden states, just use current position num_hidden_states = 0 temporal_length = 1 - + # print(f"Creating {graph_type} temporal graphs: {num_spatial_nodes} spatial nodes -> {num_spatial_nodes} temporal graphs of length {temporal_length}") - + # Store reference to spatial graph spatial_graph = batch.clone() - + # Set connectivity type code for tracking - connectivity_type_map = { - "fan": 0, - "hub_n_spoke": 1, - "complete": 2, - "complete_no_self": 3 - } - + connectivity_type_map = {"fan": 0, "hub_n_spoke": 1, "complete": 2, "complete_no_self": 3} + temporal_graphs = [] - + for node_idx in range(num_spatial_nodes): # Build temporal positions: [current_pos, hidden_1, hidden_2, ...] temporal_positions = [batch.pos[node_idx]] # Start with current position - + # Add hidden state positions if num_hidden_states > 0: for hidden_pos in batch.hidden_state: temporal_positions.append(hidden_pos[node_idx]) - + temporal_pos = torch.stack(temporal_positions) # Shape: [T, 3] - + # Calculate temporal positions for this sequence temporal_position = calculate_temporal_positions(temporal_length, device=device) - + # Create edge connectivity based on graph_type if temporal_length > 1: if graph_type == "fan": @@ -103,37 +97,37 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): # Hub connections: 0->1, 0->2, 0->3, ..., 0->T-1 hub_src = [0] * (temporal_length - 1) hub_dst = list(range(1, temporal_length)) - + # Sequential connections: 1->2, 2->3, ..., (T-2)->(T-1) seq_src = list(range(1, temporal_length - 1)) seq_dst = list(range(2, temporal_length)) - + # Combine all edges all_src = hub_src + seq_src all_dst = hub_dst + seq_dst - + edge_index = torch.tensor([all_src, all_dst], dtype=torch.long, device=device) - + elif graph_type == "hub_n_spoke": # Hub-and-spoke only: 0 connects to all others, no sequential hub_src = [0] * (temporal_length - 1) hub_dst = list(range(1, temporal_length)) - + edge_index = torch.tensor([hub_src, hub_dst], dtype=torch.long, device=device) - + elif graph_type == "complete": # Complete graph without self-loops: all-to-all excluding self src_nodes = [] dst_nodes = [] - + for i in range(temporal_length): for j in range(temporal_length): if i != j: # Exclude self-loops src_nodes.append(i) dst_nodes.append(j) - + edge_index = torch.tensor([src_nodes, dst_nodes], dtype=torch.long, device=device) - + else: # Single node case if graph_type == "complete": @@ -142,7 +136,7 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): else: # Single node, no edges for other types edge_index = torch.tensor([[], []], dtype=torch.long, device=device) - + # Create temporal graph for this spatial node temporal_graph = torch_geometric.data.Data( pos=temporal_pos, @@ -151,17 +145,17 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): temporal_length=torch.tensor([temporal_length], device=device), temporal_position=temporal_position, # Normalized position in sequence [0, 1/T, 2/T, ...] connectivity_type=torch.tensor([connectivity_type_map[graph_type]], device=device), - graph_type=graph_type # Store graph type as string for debugging + graph_type=graph_type, # Store graph type as string for debugging ) temporal_graphs.append(temporal_graph) - + # Batch all temporal graphs temporal_batch = torch_geometric.data.Batch.from_data_list(temporal_graphs) - + # Store spatial graph reference and graph type temporal_batch.spatial_graph = spatial_graph temporal_batch.graph_type = graph_type - + return temporal_batch @@ -170,25 +164,25 @@ def temporal_to_spatial_graphs(temporal_batch): Convert temporal graphs back to spatial graphs. Take the 0th node position from each temporal graph as the updated spatial position. """ - # Get the spatial graph template + # Get the spatial graph template spatial_graph = temporal_batch.spatial_graph.clone() - + # Extract 0th node positions from each temporal graph num_temporal_graphs = temporal_batch.num_graphs updated_positions = [] - + # Iterate through each temporal graph in the batch for graph_idx in range(num_temporal_graphs): # Get the node range for this temporal graph start_idx = temporal_batch.ptr[graph_idx] - + # The 0th node of each temporal graph is at the start of its range updated_positions.append(temporal_batch.pos[start_idx]) - + # Stack to create new position tensor updated_positions = torch.stack(updated_positions) - + # Update spatial graph with new positions spatial_graph.pos = updated_positions - + return spatial_graph diff --git a/scratch/transformer/develop_transformer.py b/scratch/transformer/develop_transformer.py index 34a0a15..482afd8 100644 --- a/scratch/transformer/develop_transformer.py +++ b/scratch/transformer/develop_transformer.py @@ -12,45 +12,31 @@ """ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) -import dotenv -import sys -import os -import hydra -from omegaconf import OmegaConf import torch import torch_geometric -from jamun.utils import compute_average_squared_distance_from_datasets -from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets -from jamun.data import parse_datasets_from_directory -from jamun.utils import unsqueeze_trailing # Import spatial-temporal conversion functions -from convert_spatiotemporal import ( - calculate_temporal_positions, - spatial_to_temporal_graphs, - temporal_to_spatial_graphs -) +from convert_spatiotemporal import spatial_to_temporal_graphs, temporal_to_spatial_graphs -# Import node attribute conversion functions -from pooling import ( - spatial_to_temporal_node_attr, - temporal_to_spatial_node_attr, - temporal_to_spatial_node_attr_mean, - SpatialTemporalToTemporalNodeAttr, - TemporalToSpatialNodeAttrMean -) +# Import node attribute conversion functions +from pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean + +from jamun.data import parse_datasets_from_directory +from jamun.utils import unsqueeze_trailing # Setup device - use CUDA if available -device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +device = torch.device("cuda" if torch.cuda.is_available() else "cpu") print(f"Using device: {device}") if torch.cuda.is_available(): print(f"CUDA device: {torch.cuda.get_device_name()}") print(f"CUDA memory: {torch.cuda.get_device_properties(device).total_memory / 1e9:.1f} GB") + def to_device(obj, device): """Helper function to move objects to device, handling various types.""" - if hasattr(obj, 'to'): + if hasattr(obj, "to"): return obj.to(device) elif isinstance(obj, (list, tuple)): return type(obj)(to_device(item, device) for item in obj) @@ -59,31 +45,33 @@ def to_device(obj, device): else: return obj + def move_graph_to_device(graph, device): """Move a PyTorch Geometric graph and all its tensor attributes to device.""" # Move the graph using standard .to() method graph = graph.to(device) - + # Manually move any custom tensor attributes that might not be handled for attr_name in dir(graph): - if not attr_name.startswith('_'): # Skip private attributes + if not attr_name.startswith("_"): # Skip private attributes attr_value = getattr(graph, attr_name, None) if isinstance(attr_value, torch.Tensor): setattr(graph, attr_name, attr_value.to(device)) - + return graph + dataset = parse_datasets_from_directory( root="/data/bucket/kleinhej/capped_diamines/timewarp_splits/train", traj_pattern="^(.*).xtc", pdb_pattern="^(.*).pdb", - filter_codes=['ALA_ALA'], + filter_codes=["ALA_ALA"], as_iterable=False, subsample=80, total_lag_time=8, lag_subsample_rate=10, - start_frame=800000, - num_frames=200000 + start_frame=800000, + num_frames=200000, ) # temporal_distance_cutoff = compute_temporal_average_squared_distance_from_datasets(dataset) breakpoint() @@ -92,6 +80,7 @@ def move_graph_to_device(graph, device): graph = dataset[0].__getitem__(0) batch = torch_geometric.data.Batch.from_data_list([graph]) from helpers import add_edges + batch = add_edges(batch.pos, batch, batch.batch, 0.05) # Move batch data to device @@ -99,9 +88,8 @@ def move_graph_to_device(graph, device): print(f"Moved batch to device: {batch.pos.device}") # Use E3Conv architecture for spatial feature processing -import e3tools from e3nn import o3 -from helpers import create_e3conv_network, get_e3conv_output_irreps, apply_e3conv_to_positions +from helpers import create_e3conv_network, get_e3conv_output_irreps # Create E3Conv network with yaml configuration parameters spatial_e3conv = create_e3conv_network() @@ -111,60 +99,63 @@ def move_graph_to_device(graph, device): print(f"Moved E3Conv model to device: {next(spatial_e3conv.parameters()).device}") print(f"E3Conv output irreps: {get_e3conv_output_irreps()}") -output_irreps = o3.Irreps(get_e3conv_output_irreps()) +output_irreps = o3.Irreps(get_e3conv_output_irreps()) print(f"E3Conv output dimension: {output_irreps.dim}") -print("\n" + "="*50) +print("\n" + "=" * 50) print("TEMPORAL GRAPH CONVERSION") -print("="*50) +print("=" * 50) + def test_temporal_conversion(batch, graph_type="fan"): """Test the conversion functions with example output.""" print("=== Testing Temporal Graph Conversion ===") - - print(f"Original spatial batch:") + + print("Original spatial batch:") print(f" - pos shape: {batch.pos.shape}") - print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") - if hasattr(batch, 'hidden_state') and batch.hidden_state: + print( + f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}" + ) + if hasattr(batch, "hidden_state") and batch.hidden_state: print(f" - hidden_state[0] shape: {batch.hidden_state[0].shape}") - + # Convert to temporal temporal_batch = spatial_to_temporal_graphs(batch) - - print(f"\nTemporal batch:") + + print("\nTemporal batch:") print(f" - pos shape: {temporal_batch.pos.shape}") print(f" - edge_index shape: {temporal_batch.edge_index.shape}") print(f" - num_graphs: {temporal_batch.num_graphs}") - print(f" - example edge_index for first temporal graph:") - + print(" - example edge_index for first temporal graph:") + # Show first temporal graph structure first_graph_end = temporal_batch.ptr[1] if temporal_batch.num_graphs > 1 else len(temporal_batch.pos) first_graph_edges = temporal_batch.edge_index[:, temporal_batch.edge_index[0] < first_graph_end] print(f" {first_graph_edges}") - - print(f"\n - COMPLETE edge_index for entire temporal batch:") + + print("\n - COMPLETE edge_index for entire temporal batch:") print(f" Shape: {temporal_batch.edge_index.shape}") print(f" {temporal_batch.edge_index}") - + print(f"\n - Temporal graph boundaries (ptr): {temporal_batch.ptr}") - print(f" - Graph node ranges:") + print(" - Graph node ranges:") for i in range(temporal_batch.num_graphs): start = temporal_batch.ptr[i] - end = temporal_batch.ptr[i+1] if i+1 < len(temporal_batch.ptr) else len(temporal_batch.pos) - print(f" Graph {i}: nodes {start}-{end-1} ({end-start} nodes)") - - print(f"\n - Temporal positions for each graph:") + end = temporal_batch.ptr[i + 1] if i + 1 < len(temporal_batch.ptr) else len(temporal_batch.pos) + print(f" Graph {i}: nodes {start}-{end - 1} ({end - start} nodes)") + + print("\n - Temporal positions for each graph:") print(f" Shape: {temporal_batch.temporal_position.shape}") print(f" First graph temporal_position: {temporal_batch.temporal_position[:5]}") # Show first 5 positions print(f" All temporal_position values: {temporal_batch.temporal_position}") - + # Convert back to spatial reconstructed_spatial = temporal_to_spatial_graphs(temporal_batch) - - print(f"\nReconstructed spatial:") + + print("\nReconstructed spatial:") print(f" - pos shape: {reconstructed_spatial.pos.shape}") print(f" - position difference from original: {torch.norm(reconstructed_spatial.pos - batch.pos)}") - + return temporal_batch, reconstructed_spatial @@ -174,19 +165,19 @@ def test_temporal_conversion(batch, graph_type="fan"): # Move temporal batch to device (should already be on correct device now) temporal_batch = move_graph_to_device(temporal_batch, device) reconstructed_spatial = move_graph_to_device(reconstructed_spatial, device) -print(f"Temporal batch device verification:") +print("Temporal batch device verification:") print(f" - pos device: {temporal_batch.pos.device}") print(f" - edge_index device: {temporal_batch.edge_index.device}") print(f" - batch device: {temporal_batch.batch.device}") print(f" - ptr device: {temporal_batch.ptr.device}") -if hasattr(temporal_batch, 'temporal_position'): +if hasattr(temporal_batch, "temporal_position"): print(f" - temporal_position device: {temporal_batch.temporal_position.device}") -if hasattr(temporal_batch, 'spatial_node_idx'): +if hasattr(temporal_batch, "spatial_node_idx"): print(f" - spatial_node_idx device: {temporal_batch.spatial_node_idx.device}") -print("\n" + "="*50) +print("\n" + "=" * 50) print("PROCESSING ALL TEMPORAL POSITIONS WITH E3CONV") -print("="*50) +print("=" * 50) breakpoint() # Process all temporal positions with E3Conv with torch.no_grad(): @@ -194,32 +185,36 @@ def test_temporal_conversion(batch, graph_type="fan"): # add edges to the topology sigma = torch.tensor(0.0, device=device) from jamun.utils import unsqueeze_trailing - sigma = unsqueeze_trailing(sigma,1) + + sigma = unsqueeze_trailing(sigma, 1) topology = batch.clone() topology = move_graph_to_device(topology, device) del topology.pos, topology.batch, topology.num_graphs - + # Process current positions: [N, 3] -> [N, 1, num_features] - node_attr_current = spatial_e3conv(batch.pos, topology, batch.batch, \ - num_graphs=batch.num_graphs,\ - c_noise=sigma,\ - effective_radial_cutoff=0.05).unsqueeze(1) - + node_attr_current = spatial_e3conv( + batch.pos, topology, batch.batch, num_graphs=batch.num_graphs, c_noise=sigma, effective_radial_cutoff=0.05 + ).unsqueeze(1) + # Process hidden state positions and collect all temporal features node_attr_list = [node_attr_current] breakpoint() - if hasattr(batch, 'hidden_state') and batch.hidden_state: + if hasattr(batch, "hidden_state") and batch.hidden_state: for hidden_pos in batch.hidden_state: - node_attr_hidden = node_attr_current = spatial_e3conv(hidden_pos, topology, batch.batch, \ - num_graphs=batch.num_graphs,\ - c_noise=sigma,\ - effective_radial_cutoff=0.05).unsqueeze(1) + node_attr_hidden = node_attr_current = spatial_e3conv( + hidden_pos, + topology, + batch.batch, + num_graphs=batch.num_graphs, + c_noise=sigma, + effective_radial_cutoff=0.05, + ).unsqueeze(1) node_attr_list.append(node_attr_hidden) - + # Stack along temporal dimension: [N, T, num_features] breakpoint() node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) - + breakpoint() # Convert spatial-temporal features to temporal node attributes with proper ordering spatial_temporal_pooler = SpatialTemporalToTemporalNodeAttr() @@ -231,32 +226,34 @@ def test_temporal_conversion(batch, graph_type="fan"): print(f"Total norm (should be nonzero): {torch.norm(spatial_node_attr_all_temporal):.6f}") print(f"Spatial node attributes device: {spatial_node_attr_all_temporal.device}") -print("\n" + "="*50) +print("\n" + "=" * 50) print("E3TRANSFORMER TEST") -print("="*50) +print("=" * 50) breakpoint() + + def test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, device): """Test the E3Transformer with temporal graphs.""" from temporal_transformer import E3Transformer - + print("=== Testing E3Transformer ===") - + # Use the precomputed temporal node attributes (processed by E3Conv) print(f"Using temporal node attributes (processed by E3Conv): {spatial_node_attr_all_temporal.shape}") print(f"Sample temporal node attr: {spatial_node_attr_all_temporal[0]}") - + # The node attributes are already arranged to match temporal graph ordering temporal_node_attr = spatial_node_attr_all_temporal - - print(f"Input shapes:") + + print("Input shapes:") print(f" - temporal_node_attr: {temporal_node_attr.shape}") print(f" - temporal_graph.pos: {temporal_batch.pos.shape}") print(f" - temporal_graph.edge_index: {temporal_batch.edge_index.shape}") print(f" - temporal_graph.temporal_position: {temporal_batch.temporal_position.shape}") print(f" - temporal_graph.batch: {temporal_batch.batch.shape}") print(f" - temporal_graph.num_graphs: {temporal_batch.num_graphs}") - + # Create E3Transformer model that takes 1x1e node attributes (E3Conv output) transformer = E3Transformer( irreps_out="3x1e", # 3D output (like positions) @@ -267,26 +264,28 @@ def test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, d edge_attr_dim=24, # Split into 2 parts: 12+12 (radial+temporal) num_attention_heads=1, # Single attention head for simpler test ) - + # Move transformer to device transformer = transformer.to(device) print(f"Moved transformer to device: {next(transformer.parameters()).device}") - - print(f"\nTransformer parameters:") + + print("\nTransformer parameters:") print(f" - irreps_out: {transformer.irreps_out}") print(f" - irreps_hidden: {transformer.irreps_hidden}") print(f" - irreps_node_attr: {transformer.irreps_node_attr}") print(f" - temporal_gate.irreps_out: {transformer.temporal_gate.irreps_out}") print(f" - radial_edge_attr_dim: {transformer.radial_edge_attr_dim}") print(f" - temporal_edge_attr_dim: {transformer.temporal_edge_attr_dim}") - + # Forward pass with tensor and graph (like E3Conv) effective_radial_cutoff = 5.0 # Define the cutoff in forward pass temporal_cutoff = 1.0 # Default temporal cutoff (no cutoff for temporal contributions) with torch.no_grad(): try: - transformer_output = transformer(temporal_node_attr, temporal_batch, effective_radial_cutoff, temporal_cutoff) - print(f"\nāœ… Transformer forward pass successful!") + transformer_output = transformer( + temporal_node_attr, temporal_batch, effective_radial_cutoff, temporal_cutoff + ) + print("\nāœ… Transformer forward pass successful!") print(f"Transformer output shape: {transformer_output.shape}") print(f"Transformer output sample: {transformer_output[0]}") print(f"Transformer output norm: {torch.norm(transformer_output):.6f}") @@ -296,15 +295,17 @@ def test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, d except Exception as e: print(f"\nāŒ Transformer forward pass failed: {e}") import traceback + traceback.print_exc() return False -# Test the complete workflow: E3Conv -> Transformer + +# Test the complete workflow: E3Conv -> Transformer success = test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, device) -print("\n" + "="*50) +print("\n" + "=" * 50) print("TEMPORAL TO SPATIAL MEAN POOLING") -print("="*50) +print("=" * 50) breakpoint() # Demonstrate mean pooling from temporal features back to spatial features print("=== Testing Mean Pooling ===") @@ -324,25 +325,27 @@ def test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, d print(f"Pooled features norm: {torch.norm(spatial_features_pooled):.6f}") # Verify that we correctly recovered the spatial dimension -assert spatial_features_pooled.shape[0] == batch.pos.shape[0], f"Spatial node count mismatch: {spatial_features_pooled.shape[0]} vs {batch.pos.shape[0]}" +assert spatial_features_pooled.shape[0] == batch.pos.shape[0], ( + f"Spatial node count mismatch: {spatial_features_pooled.shape[0]} vs {batch.pos.shape[0]}" +) print("āœ… Mean pooling successfully converted temporal features back to spatial!") -print("\n" + "="*50) +print("\n" + "=" * 50) print("TESTS COMPLETED") -print("="*50) +print("=" * 50) -print("\n" + "="*50) +print("\n" + "=" * 50) print("FINAL SUMMARY") -print("="*50) +print("=" * 50) # Device summary -print(f"Device Summary:") +print("Device Summary:") print(f" - Used device: {device}") if torch.cuda.is_available(): print(f" - CUDA memory allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") print(f" - CUDA memory cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") -print(f"\nTest Results:") +print("\nTest Results:") print(f" - Manual workflow tests: {'āœ… PASSED' if success else 'āŒ FAILED'}") if success: @@ -351,4 +354,3 @@ def test_e3_transformer(batch, temporal_batch, spatial_node_attr_all_temporal, d print("To test the unified E3SpatioTemporal model, run: python3 test_e3_spatiotemporal.py") else: print("\nāš ļø Some tests failed. Check the output above for details.") - diff --git a/scratch/transformer/helpers.py b/scratch/transformer/helpers.py index 1c08437..d06d705 100644 --- a/scratch/transformer/helpers.py +++ b/scratch/transformer/helpers.py @@ -3,36 +3,36 @@ Helper functions for creating network architectures used in transformer development. """ -import torch -import torch_geometric +import functools + import e3tools import numpy as np -from jamun.model.arch.e3conv import E3Conv -from jamun.utils.average_squared_distance import compute_distance_matrix, compute_average_squared_distance -import functools +import torch +import torch_geometric from convert_spatiotemporal import spatial_to_temporal_graphs +from jamun.model.arch.e3conv import E3Conv +from jamun.utils.average_squared_distance import compute_average_squared_distance + def compute_temporal_average_squared_distance_from_dataset( - dataset, - num_samples: int = 100, - verbose: bool = False + dataset, num_samples: int = 100, verbose: bool = False ) -> float: """ Compute average squared distance between neighboring vertices in temporal graphs. - + Args: dataset: Dataset containing spatial graphs with hidden states num_samples: Number of samples to use for estimation verbose: Whether to print verbose output - + Returns: float: Average squared distance between temporal neighbors """ - + avg_sq_dists = [] num_graphs = 0 - + # Follow pattern from average_squared_distance.py for item in dataset: if num_graphs >= num_samples: @@ -52,13 +52,12 @@ def compute_temporal_average_squared_distance_from_dataset( num_graphs += 1 mean_avg_sq_dist = sum(avg_sq_dists) / num_graphs - if verbose: print(f"Total graphs processed: {num_graphs}") print(f"Total temporal graphs processed: {len(avg_sq_dists)}") print(f"Mean average squared distance between temporal nodes: {mean_avg_sq_dist:.6f}") print(f"Standard deviation: {np.std(avg_sq_dists):.6f}") - + return float(mean_avg_sq_dist) @@ -97,26 +96,26 @@ def add_edges( def apply_e3conv_to_positions(e3conv_model, pos, topology, batch, effective_radial_cutoff=5.0): """ Apply E3Conv model to a set of positions using existing graph topology. - + Args: e3conv_model: E3Conv model instance - pos (torch.Tensor): Positions [N, 3] + pos (torch.Tensor): Positions [N, 3] topology (torch_geometric.data.Batch): Existing graph topology from dataloader batch (torch.Tensor): Batch tensor from the graph effective_radial_cutoff (float): Radial cutoff for edges - + Returns: torch.Tensor: Node features [N, feature_dim] """ # Clone topology to avoid modifying original topology_with_edges = topology.clone() - + # Add edges using the local add_edges function topology_with_edges = add_edges(pos, topology_with_edges, batch, effective_radial_cutoff) - + # Use noise conditioning of 0.0 (no noise) c_noise = torch.zeros(pos.shape[0], dtype=pos.dtype, device=pos.device) - + # Apply E3Conv num_graphs = batch.max().item() + 1 # Number of graphs in the batch node_features = e3conv_model( @@ -125,63 +124,60 @@ def apply_e3conv_to_positions(e3conv_model, pos, topology, batch, effective_radi batch=batch, num_graphs=num_graphs, c_noise=c_noise, - effective_radial_cutoff=effective_radial_cutoff + effective_radial_cutoff=effective_radial_cutoff, ) - + return node_features def create_e3conv_network(): """ Create an E3Conv network with parameters matching the yaml configuration. - + Returns: E3Conv: Configured E3Conv network """ - + # Hidden layer factory as specified in yaml - hidden_layer_factory = functools.partial( - e3tools.nn.ConvBlock, - conv=functools.partial(e3tools.nn.Conv) - ) - - # Output head factory as specified in yaml + hidden_layer_factory = functools.partial(e3tools.nn.ConvBlock, conv=functools.partial(e3tools.nn.Conv)) + + # Output head factory as specified in yaml output_head_factory = functools.partial( e3tools.nn.EquivariantMLP, - irreps_hidden_list=["120x0e + 32x1e"] # Using irreps_hidden from yaml + irreps_hidden_list=["120x0e + 32x1e"], # Using irreps_hidden from yaml ) - + # Create E3Conv with exact parameters from yaml e3conv = E3Conv( - irreps_out="1x1e", # 3D vector output - irreps_hidden="120x0e + 32x1e", # Hidden representations - irreps_sh="1x0e + 1x1e", # Spherical harmonics + irreps_out="1x1e", # 3D vector output + irreps_hidden="120x0e + 32x1e", # Hidden representations + irreps_sh="1x0e + 1x1e", # Spherical harmonics hidden_layer_factory=hidden_layer_factory, output_head_factory=output_head_factory, - use_residue_information=True, # Assuming True, matches yaml ${data.use_residue_information} - n_layers=1, # Number of layers - edge_attr_dim=64, # Edge attribute dimension - atom_type_embedding_dim=8, # Atom type embedding - atom_code_embedding_dim=8, # Atom code embedding - residue_code_embedding_dim=32, # Residue code embedding - residue_index_embedding_dim=8, # Residue index embedding - use_residue_sequence_index=False, # As specified in yaml - num_atom_types=20, # Number of atom types - max_sequence_length=10, # Max sequence length - num_atom_codes=10, # Number of atom codes - num_residue_types=25, # Number of residue types - test_equivariance=False, # Disable for production - reduce=None # No reduction + use_residue_information=True, # Assuming True, matches yaml ${data.use_residue_information} + n_layers=1, # Number of layers + edge_attr_dim=64, # Edge attribute dimension + atom_type_embedding_dim=8, # Atom type embedding + atom_code_embedding_dim=8, # Atom code embedding + residue_code_embedding_dim=32, # Residue code embedding + residue_index_embedding_dim=8, # Residue index embedding + use_residue_sequence_index=False, # As specified in yaml + num_atom_types=20, # Number of atom types + max_sequence_length=10, # Max sequence length + num_atom_codes=10, # Number of atom codes + num_residue_types=25, # Number of residue types + test_equivariance=False, # Disable for production + reduce=None, # No reduction ) - + return e3conv def get_e3conv_output_irreps(): """ Get the output irreps of the E3Conv network. - + Returns: str: Output irreps string """ - return "1x1e" # 3D vector output as specified in yaml \ No newline at end of file + return "1x1e" # 3D vector output as specified in yaml diff --git a/scratch/transformer/pooling.py b/scratch/transformer/pooling.py index c3a37bc..418aed8 100644 --- a/scratch/transformer/pooling.py +++ b/scratch/transformer/pooling.py @@ -3,55 +3,57 @@ Lightning modules for converting node attributes between spatial and temporal representations. """ -import torch -import torch_geometric import pytorch_lightning as pl +import torch class SpatialToTemporalNodeAttr(pl.LightningModule): """ - Lightning module to transfer node attributes from spatial nodes to temporal nodes + Lightning module to transfer node attributes from spatial nodes to temporal nodes by repeating first temporal feature. """ - + def __init__(self): super().__init__() - + def forward(self, spatial_node_attr_temporal, temporal_batch): """ Transfer node attributes from spatial nodes to temporal nodes by repeating first temporal feature. Takes the first temporal feature (t=0) and repeats it T times for each spatial node. - + Args: spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs - + Returns: torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] """ num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape num_temporal_graphs = temporal_batch.num_graphs - + # Verify consistency - assert num_spatial_nodes == num_temporal_graphs, \ + assert num_spatial_nodes == num_temporal_graphs, ( f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" - + ) + # Verify temporal length consistency expected_temporal_nodes = temporal_batch.pos.shape[0] expected_total_nodes = num_spatial_nodes * temporal_length - assert expected_total_nodes == expected_temporal_nodes, \ + assert expected_total_nodes == expected_temporal_nodes, ( f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" - + ) + # Extract first temporal feature (t=0) and repeat it T times for each spatial node first_temporal_features = spatial_node_attr_temporal[:, 0, :] # [N, attr_dim] - + # Repeat each spatial node's first temporal feature T times temporal_node_attr = first_temporal_features.repeat_interleave(temporal_length, dim=0) # [N*T, attr_dim] - + # Verify the output shape matches the temporal batch - assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + assert temporal_node_attr.shape[0] == expected_temporal_nodes, ( f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" - + ) + return temporal_node_attr @@ -60,43 +62,44 @@ class TemporalToSpatialNodeAttr(pl.LightningModule): Lightning module to convert temporal node attributes back to spatial node attributes. Takes the first temporal node attribute from each temporal graph. """ - + def __init__(self): super().__init__() - + def forward(self, temporal_node_attr, temporal_batch): """ Convert temporal node attributes back to spatial node attributes. Takes the first temporal node attribute from each temporal graph. - + Args: temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs - + Returns: torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] """ num_temporal_graphs = temporal_batch.num_graphs attr_dim = temporal_node_attr.shape[1] - + # Extract the first node attribute from each temporal graph spatial_node_attr = [] - + for graph_idx in range(num_temporal_graphs): # Get the node range for this temporal graph start_idx = temporal_batch.ptr[graph_idx] - + # The 0th node of each temporal graph is at the start of its range first_node_attr = temporal_node_attr[start_idx] spatial_node_attr.append(first_node_attr) - + # Stack to create spatial node attribute tensor spatial_node_attr = torch.stack(spatial_node_attr) - + # Verify output shape - assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), ( f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" - + ) + return spatial_node_attr @@ -105,45 +108,50 @@ class TemporalToSpatialNodeAttrMean(pl.LightningModule): Lightning module to convert temporal node attributes back to spatial node attributes by averaging. Takes the mean of all temporal node attributes for each temporal graph. """ - + def __init__(self): super().__init__() - + def forward(self, temporal_node_attr, temporal_batch): """ Convert temporal node attributes back to spatial node attributes by averaging. Takes the mean of all temporal node attributes for each temporal graph. - + Args: temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs - + Returns: torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] """ num_temporal_graphs = temporal_batch.num_graphs attr_dim = temporal_node_attr.shape[1] - + # Extract the mean node attributes from each temporal graph spatial_node_attr = [] - + for graph_idx in range(num_temporal_graphs): # Get the node range for this temporal graph start_idx = temporal_batch.ptr[graph_idx] - end_idx = temporal_batch.ptr[graph_idx + 1] if graph_idx + 1 < len(temporal_batch.ptr) else len(temporal_node_attr) - + end_idx = ( + temporal_batch.ptr[graph_idx + 1] + if graph_idx + 1 < len(temporal_batch.ptr) + else len(temporal_node_attr) + ) + # Take the mean of all temporal nodes for this spatial node temporal_nodes_attr = temporal_node_attr[start_idx:end_idx] # [temporal_length, attr_dim] mean_node_attr = temporal_nodes_attr.mean(dim=0) # [attr_dim] spatial_node_attr.append(mean_node_attr) - + # Stack to create spatial node attribute tensor spatial_node_attr = torch.stack(spatial_node_attr) - + # Verify output shape - assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), ( f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" - + ) + return spatial_node_attr @@ -152,43 +160,46 @@ class SpatialTemporalToTemporalNodeAttr(pl.LightningModule): Lightning module to convert spatial node attributes arranged temporally to temporal node attributes. Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. """ - + def __init__(self): super().__init__() - + def forward(self, spatial_node_attr_temporal, temporal_batch): """ Convert spatial node attributes arranged temporally to temporal node attributes. Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. - + Args: spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs for validation - + Returns: torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] """ num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape num_temporal_graphs = temporal_batch.num_graphs - + # Verify consistency with temporal batch - assert num_spatial_nodes == num_temporal_graphs, \ + assert num_spatial_nodes == num_temporal_graphs, ( f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" - + ) + # Verify temporal length consistency expected_temporal_nodes = temporal_batch.pos.shape[0] expected_total_nodes = num_spatial_nodes * temporal_length - assert expected_total_nodes == expected_temporal_nodes, \ + assert expected_total_nodes == expected_temporal_nodes, ( f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" - + ) + # Reshape to match temporal graph ordering: [N, T, features] -> [N*T, features] # Temporal graph arranges nodes as: [node0_t0, node0_t1, ..., node0_tT-1, node1_t0, ...] temporal_node_attr = spatial_node_attr_temporal.reshape(num_spatial_nodes * temporal_length, attr_dim) - + # Verify the output shape matches the temporal batch - assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + assert temporal_node_attr.shape[0] == expected_temporal_nodes, ( f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" - + ) + return temporal_node_attr diff --git a/scratch/transformer/temporal_transformer.py b/scratch/transformer/temporal_transformer.py index 07852c9..d2b82f8 100644 --- a/scratch/transformer/temporal_transformer.py +++ b/scratch/transformer/temporal_transformer.py @@ -1,12 +1,10 @@ -from typing import Dict, Union - import e3nn +import e3tools +import e3tools.nn import torch import torch.nn as nn -from e3nn import o3 import torch_geometric.data -import e3tools -import e3tools.nn +from e3nn import o3 class E3Transformer(nn.Module): @@ -14,10 +12,10 @@ class E3Transformer(nn.Module): def __init__( self, - irreps_out: Union[str, e3nn.o3.Irreps], - irreps_hidden: Union[str, e3nn.o3.Irreps], - irreps_sh: Union[str, e3nn.o3.Irreps], - irreps_node_attr: Union[str, e3nn.o3.Irreps], + irreps_out: str | e3nn.o3.Irreps, + irreps_hidden: str | e3nn.o3.Irreps, + irreps_sh: str | e3nn.o3.Irreps, + irreps_node_attr: str | e3nn.o3.Irreps, num_layers: int, edge_attr_dim: int, num_attention_heads: int, @@ -28,27 +26,27 @@ def __init__( self.irreps_out = o3.Irreps(irreps_out) self.irreps_hidden = o3.Irreps(irreps_hidden) self.irreps_sh = o3.Irreps(irreps_sh) - self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps + self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps self.num_layers = num_layers self.edge_attr_dim = edge_attr_dim self.num_attention_heads = num_attention_heads self.reduce = reduce - self.sh = o3.SphericalHarmonics( - irreps_out=self.irreps_sh, normalize=True, normalization="component" - ) + self.sh = o3.SphericalHarmonics(irreps_out=self.irreps_sh, normalize=True, normalization="component") # Split edge attribute dimensions: radial and temporal (bondedness is optional) self.radial_edge_attr_dim = self.edge_attr_dim // 2 self.temporal_edge_attr_dim = self.edge_attr_dim - self.radial_edge_attr_dim - + # Optional bondedness embedding (only used if bond_mask exists in graph) self.embed_bondedness = nn.Embedding(2, self.edge_attr_dim // 3) - + # Gate for combining node attributes with temporal position # Input: node_attr (from data) + temporal_position (1x0e scalar) irreps_with_temporal = self.irreps_node_attr + o3.Irreps("1x0e") - self.temporal_gate = e3tools.nn.GateWrapper(irreps_in=irreps_with_temporal, \ - irreps_out=self.irreps_hidden, \ - irreps_gate=irreps_with_temporal,) + self.temporal_gate = e3tools.nn.GateWrapper( + irreps_in=irreps_with_temporal, + irreps_out=self.irreps_hidden, + irreps_gate=irreps_with_temporal, + ) # self.initial_linear = o3.Linear( # self.temporal_gate.irreps_out, self.irreps_hidden # ) @@ -98,7 +96,7 @@ def forward( basis="gaussian", cutoff=True, ) - + # Temporal edge attributes from temporal_position differences temporal_edge_vec = temporal_position[src] - temporal_position[dst] temporal_edge_attr = e3nn.math.soft_one_hot_linspace( @@ -109,9 +107,9 @@ def forward( basis="gaussian", cutoff=True, ) - + # Optional bondedness (if bond_mask exists in the temporal graph) - if hasattr(temporal_graph, 'bond_mask') and temporal_graph.bond_mask is not None: + if hasattr(temporal_graph, "bond_mask") and temporal_graph.bond_mask is not None: bonded_edge_attr = self.embed_bondedness(temporal_graph.bond_mask) edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr, temporal_edge_attr), dim=-1) else: @@ -122,7 +120,7 @@ def forward( # Concatenate node_attr with temporal_position (scalar) temporal_position_expanded = temporal_position.unsqueeze(-1) # [N, 1] for concatenation node_attr_with_temporal = torch.cat([node_attr, temporal_position_expanded], dim=-1) - + # Apply temporal gate node_attr_processed = self.temporal_gate(node_attr_with_temporal) # node_attr_processed = self.initial_linear(node_attr_gated) @@ -141,22 +139,22 @@ def forward( dim_size=num_graphs, reduce=self.reduce, ) - + return node_attr_processed class E3SpatioTemporal(nn.Module): """ E(3)-equivariant spatio-temporal model that combines spatial and temporal processing. - + This model implements the complete workflow: 1. Process input spatial graph and hidden states through spatial module 2. Pool spatial features to temporal graph representation - 3. Process temporal graph through temporal module + 3. Process temporal graph through temporal module 4. Pool temporal features back to spatial representation 5. Convert temporal graph back to spatial graph """ - + def __init__( self, spatial_module: nn.Module, @@ -168,7 +166,7 @@ def __init__( ): """ Initialize the E3SpatioTemporal model. - + Args: spatial_module: Module for processing spatial positions (e.g., E3Conv) temporal_module: Module for processing temporal graphs (e.g., E3Transformer) @@ -178,30 +176,30 @@ def __init__( temporal_cutoff: Cutoff for temporal edge weights """ super().__init__() - + self.spatial_module = spatial_module self.temporal_module = temporal_module self.spatial_to_temporal_pooler = spatial_to_temporal_pooler self.temporal_to_spatial_pooler = temporal_to_spatial_pooler self.radial_cutoff = radial_cutoff self.temporal_cutoff = temporal_cutoff - + def forward( self, batch: torch_geometric.data.Batch, c_noise: torch.Tensor, return_temporal_features: bool = False, return_temporal_graph: bool = False, - ) -> Union[torch.Tensor, Dict[str, torch.Tensor]]: + ) -> torch.Tensor | dict[str, torch.Tensor]: """ Forward pass implementing the complete spatio-temporal workflow. - + Args: batch: Input spatial graph batch with pos, batch, num_graphs, and optionally hidden_state c_noise: Noise conditioning tensor return_temporal_features: Whether to return intermediate temporal features return_temporal_graph: Whether to return the temporal graph - + Returns: If return_temporal_features or return_temporal_graph is True, returns dict with: - 'spatial_features': Final spatial features @@ -211,39 +209,39 @@ def forward( Otherwise returns just the final spatial features tensor """ from convert_spatiotemporal import spatial_to_temporal_graphs, temporal_to_spatial_graphs - + # Store original device device = batch.pos.device - + # Step 1: Convert spatial graph to temporal graphs temporal_batch = spatial_to_temporal_graphs(batch) - + # Step 2: Process all positions (current + hidden states) with spatial module # Create topology for spatial processing (without positions) topology = batch.clone() # Remove position-dependent attributes but keep graph structure - if hasattr(topology, 'pos'): + if hasattr(topology, "pos"): del topology.pos - if hasattr(topology, 'batch'): - del topology.batch - if hasattr(topology, 'num_graphs'): + if hasattr(topology, "batch"): + del topology.batch + if hasattr(topology, "num_graphs"): del topology.num_graphs - + node_attr_list = [] - + # Process current positions node_attr_current = self.spatial_module( - batch.pos, - topology, + batch.pos, + topology, batch.batch, num_graphs=batch.num_graphs, c_noise=c_noise, - effective_radial_cutoff=self.radial_cutoff + effective_radial_cutoff=self.radial_cutoff, ).unsqueeze(1) # [N, 1, features] node_attr_list.append(node_attr_current) - + # Process hidden state positions if they exist - if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + if hasattr(batch, "hidden_state") and batch.hidden_state is not None and len(batch.hidden_state) > 0: for hidden_pos in batch.hidden_state: node_attr_hidden = self.spatial_module( hidden_pos, @@ -251,54 +249,51 @@ def forward( batch.batch, num_graphs=batch.num_graphs, c_noise=c_noise, - effective_radial_cutoff=self.radial_cutoff + effective_radial_cutoff=self.radial_cutoff, ).unsqueeze(1) # [N, 1, features] node_attr_list.append(node_attr_hidden) - + # Step 3: Stack spatial-temporal features node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) # [N, T, features] - + # Step 4: Convert spatial-temporal features to temporal node attributes temporal_node_attr = self.spatial_to_temporal_pooler(node_attr_spatial_temporal, temporal_batch) - + # Step 5: Process temporal graph through temporal module temporal_output = self.temporal_module( - temporal_node_attr, - temporal_batch, - self.radial_cutoff, - self.temporal_cutoff + temporal_node_attr, temporal_batch, self.radial_cutoff, self.temporal_cutoff ) - + # Step 6: Pool temporal features back to spatial features spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) - + # Step 7: Convert temporal graph back to spatial graph output_spatial_graph = temporal_to_spatial_graphs(temporal_batch) - + # Prepare return values if return_temporal_features or return_temporal_graph: result = { - 'spatial_features': spatial_features, - 'spatial_graph': output_spatial_graph, + "spatial_features": spatial_features, + "spatial_graph": output_spatial_graph, } if return_temporal_features: - result['temporal_features'] = temporal_output + result["temporal_features"] = temporal_output if return_temporal_graph: - result['temporal_graph'] = temporal_batch + result["temporal_graph"] = temporal_batch return result else: return spatial_features - + def get_spatial_output_irreps(self): """Get the irreps of the spatial module output.""" - if hasattr(self.spatial_module, 'irreps_out'): + if hasattr(self.spatial_module, "irreps_out"): return self.spatial_module.irreps_out else: raise AttributeError("Spatial module does not have irreps_out attribute") - + def get_temporal_output_irreps(self): - """Get the irreps of the temporal module output.""" - if hasattr(self.temporal_module, 'irreps_out'): + """Get the irreps of the temporal module output.""" + if hasattr(self.temporal_module, "irreps_out"): return self.temporal_module.irreps_out else: - raise AttributeError("Temporal module does not have irreps_out attribute") \ No newline at end of file + raise AttributeError("Temporal module does not have irreps_out attribute") diff --git a/scratch/transformer/test_e3_spatiotemporal.py b/scratch/transformer/test_e3_spatiotemporal.py index 369238a..c6a65a7 100644 --- a/scratch/transformer/test_e3_spatiotemporal.py +++ b/scratch/transformer/test_e3_spatiotemporal.py @@ -2,49 +2,54 @@ """ Test script for the E3SpatioTemporal model. -This script tests the unified E3SpatioTemporal model that encapsulates +This script tests the unified E3SpatioTemporal model that encapsulates the complete spatio-temporal processing workflow. """ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) import torch import torch_geometric -from jamun.data import parse_datasets_from_directory -from jamun.utils import unsqueeze_trailing # Import modules needed for the test -from helpers import create_e3conv_network, add_edges -from temporal_transformer import E3SpatioTemporal, E3Transformer +from helpers import add_edges, create_e3conv_network from pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean +from temporal_transformer import E3SpatioTemporal, E3Transformer + +from jamun.data import parse_datasets_from_directory +from jamun.utils import unsqueeze_trailing + def setup_device(): """Setup CUDA device if available.""" - device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') + device = torch.device("cuda" if torch.cuda.is_available() else "cpu") print(f"Using device: {device}") if torch.cuda.is_available(): print(f"CUDA device: {torch.cuda.get_device_name()}") print(f"CUDA memory: {torch.cuda.get_device_properties(device).total_memory / 1e9:.1f} GB") return device + def move_graph_to_device(graph, device): """Move a PyTorch Geometric graph and all its tensor attributes to device.""" # Move the graph using standard .to() method graph = graph.to(device) - + # Manually move any custom tensor attributes that might not be handled for attr_name in dir(graph): - if not attr_name.startswith('_'): # Skip private attributes + if not attr_name.startswith("_"): # Skip private attributes attr_value = getattr(graph, attr_name, None) if isinstance(attr_value, torch.Tensor): setattr(graph, attr_name, attr_value.to(device)) - + return graph + def load_test_data(device): """Load and prepare test data.""" print("Loading test data...") - + dataset = parse_datasets_from_directory( root="/data2/sules/ALA_ALA_enhanced_full_grid/train", traj_pattern="^(.*).xtc", @@ -53,27 +58,28 @@ def load_test_data(device): total_lag_time=5, lag_subsample_rate=1, max_datasets=3, - num_frames=10 + num_frames=10, ) - + # Get first graph and create batch graph = dataset[0].__getitem__(0) batch = torch_geometric.data.Batch.from_data_list([graph]) batch = add_edges(batch.pos, batch, batch.batch, 0.05) - + # Move to device batch = move_graph_to_device(batch, device) print(f"Loaded batch with {batch.pos.shape[0]} nodes on device: {batch.pos.device}") - + return batch + def create_spatiotemporal_model(device): """Create and configure the E3SpatioTemporal model.""" print("Creating E3SpatioTemporal model...") - + # Create component modules spatial_module = create_e3conv_network().to(device) - + temporal_module = E3Transformer( irreps_out="3x1e", # 3D output (like positions) irreps_hidden="8x0e + 4x1e", # Hidden representations @@ -83,10 +89,10 @@ def create_spatiotemporal_model(device): edge_attr_dim=24, # Split into 2 parts: 12+12 (radial+temporal) num_attention_heads=1, # Single attention head for simpler test ).to(device) - + spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr() temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean() - + # Create the unified model spatiotemporal_model = E3SpatioTemporal( spatial_module=spatial_module, @@ -96,34 +102,37 @@ def create_spatiotemporal_model(device): radial_cutoff=0.05, temporal_cutoff=1.0, ).to(device) - + print(f"Created E3SpatioTemporal model on device: {next(spatiotemporal_model.parameters()).device}") return spatiotemporal_model + def test_spatiotemporal_model(model, batch, device): """Test the E3SpatioTemporal model with various configurations.""" print("=" * 50) print("TESTING E3SPATIOTEMPORAL MODEL") print("=" * 50) - + # Print model information - print(f"Model components:") + print("Model components:") print(f" - Spatial module output irreps: {model.get_spatial_output_irreps()}") print(f" - Temporal module output irreps: {model.get_temporal_output_irreps()}") print(f" - Radial cutoff: {model.radial_cutoff}") print(f" - Temporal cutoff: {model.temporal_cutoff}") - + # Prepare noise conditioning sigma = torch.tensor(0.0, device=device) sigma = unsqueeze_trailing(sigma, 1) - - print(f"\nInput batch:") + + print("\nInput batch:") print(f" - pos shape: {batch.pos.shape}") - print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") + print( + f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}" + ) print(f" - batch device: {batch.pos.device}") - + success = True - + try: with torch.no_grad(): print("\n1. Testing simple forward pass (spatial features only)...") @@ -131,71 +140,68 @@ def test_spatiotemporal_model(model, batch, device): print(f" āœ… Success! Spatial features shape: {spatial_features.shape}") print(f" Spatial features device: {spatial_features.device}") print(f" Spatial features norm: {torch.norm(spatial_features):.6f}") - + print("\n2. Testing full forward pass (all outputs)...") - results = model( - batch, - sigma, - return_temporal_features=True, - return_temporal_graph=True - ) - - print(f" āœ… Success! Full output results:") + results = model(batch, sigma, return_temporal_features=True, return_temporal_graph=True) + + print(" āœ… Success! Full output results:") print(f" - spatial_features shape: {results['spatial_features'].shape}") print(f" - temporal_features shape: {results['temporal_features'].shape}") print(f" - temporal_graph num_graphs: {results['temporal_graph'].num_graphs}") print(f" - spatial_graph pos shape: {results['spatial_graph'].pos.shape}") - + # Verify spatial graph reconstruction - pos_difference = torch.norm(results['spatial_graph'].pos - batch.pos) + pos_difference = torch.norm(results["spatial_graph"].pos - batch.pos) print(f" - Spatial position reconstruction error: {pos_difference:.6f}") - + print("\n3. Testing consistency between simple and full forward pass...") simple_features = spatial_features - full_features = results['spatial_features'] + full_features = results["spatial_features"] consistency_error = torch.norm(simple_features - full_features) print(f" Consistency error: {consistency_error:.6f}") - + if consistency_error < 1e-6: print(" āœ… Results are consistent!") else: print(" āš ļø Results differ between simple and full forward pass") success = False - + except Exception as e: print(f"\nāŒ Test failed: {e}") import traceback + traceback.print_exc() success = False - + return success + def main(): """Main test function.""" print("E3SpatioTemporal Model Test") print("=" * 50) - + # Setup device = setup_device() batch = load_test_data(device) model = create_spatiotemporal_model(device) - + # Run tests success = test_spatiotemporal_model(model, batch, device) - + # Final summary print("\n" + "=" * 50) print("FINAL RESULTS") print("=" * 50) - + # Device summary - print(f"Device Summary:") + print("Device Summary:") print(f" - Used device: {device}") if torch.cuda.is_available(): print(f" - CUDA memory allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") print(f" - CUDA memory cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") - - print(f"\nTest Results:") + + print("\nTest Results:") if success: print("šŸŽ‰ ALL TESTS PASSED! The E3SpatioTemporal model works correctly!") print("\nThe model successfully:") @@ -207,8 +213,9 @@ def main(): print(" āœ… Maintains consistency across different call patterns") else: print("āŒ SOME TESTS FAILED! Check the output above for details.") - + return success + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/transformer/test_spatiotemporal_conditioner.py b/scratch/transformer/test_spatiotemporal_conditioner.py index 055fc8b..b525237 100755 --- a/scratch/transformer/test_spatiotemporal_conditioner.py +++ b/scratch/transformer/test_spatiotemporal_conditioner.py @@ -6,52 +6,51 @@ """ import e3nn + e3nn.set_optimization_defaults(jit_script_fx=False) +import sys +from typing import Any + import torch import torch_geometric -from typing import Dict, Any -import sys -import os # Add the src directory to path to import jamun modules -sys.path.insert(0, 'src') +sys.path.insert(0, "src") from jamun.data import parse_datasets_from_directory -from jamun.model.denoiser_conditional import Denoiser # Changed from DenoiserWithInputAttr -from jamun.model.denoiser import add_edges # Import add_edges function -from jamun.model.conditioners.conditioners import SpatioTemporalConditioner -from jamun.model.arch.spatiotemporal import E3SpatioTemporal, E3Transformer +from jamun.distributions._distributions import ConstantSigma from jamun.model.arch.e3conv import E3Conv -from jamun.model.arch.e3conv_conditional import E3ConvConditionalSpatioTemporal # Changed from E3ConvConditionalWithInputAttr +from jamun.model.arch.e3conv_conditional import ( + E3ConvConditionalSpatioTemporal, # Changed from E3ConvConditionalWithInputAttr +) +from jamun.model.arch.spatiotemporal import E3SpatioTemporal, E3Transformer +from jamun.model.conditioners.conditioners import SpatioTemporalConditioner +from jamun.model.denoiser_conditional import Denoiser # Changed from DenoiserWithInputAttr from jamun.model.pooling import SpatialTemporalToTemporalNodeAttr, TemporalToSpatialNodeAttrMean -from jamun.distributions._distributions import ConstantSigma -from jamun.utils import unsqueeze_trailing -from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets # Import temporal function +from jamun.utils.average_squared_distance import ( + compute_temporal_average_squared_distance_from_datasets, # Import temporal function +) # Setup device -device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') +device = torch.device("cuda" if torch.cuda.is_available() else "cpu") print(f"Using device: {device}") + def create_spatial_module() -> E3Conv: """Create E3Conv spatial module with reasonable parameters.""" import functools + import e3tools - + # Create factory functions - hidden_layer_factory = functools.partial( - e3tools.nn.ConvBlock, - conv=functools.partial(e3tools.nn.Conv) - ) - - output_head_factory = functools.partial( - e3tools.nn.EquivariantMLP, - irreps_hidden_list=["120x0e + 32x1e"] - ) - + hidden_layer_factory = functools.partial(e3tools.nn.ConvBlock, conv=functools.partial(e3tools.nn.Conv)) + + output_head_factory = functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["120x0e + 32x1e"]) + return E3Conv( irreps_out="3x1e", # Changed to match temporal module input - irreps_hidden="120x0e + 32x1e", + irreps_hidden="120x0e + 32x1e", irreps_sh="1x0e + 1x1e", hidden_layer_factory=hidden_layer_factory, output_head_factory=output_head_factory, @@ -68,31 +67,33 @@ def create_spatial_module() -> E3Conv: num_atom_codes=10, num_residue_types=25, test_equivariance=False, - reduce=None + reduce=None, ) + def create_temporal_module() -> E3Transformer: """Create E3Transformer temporal module.""" return E3Transformer( irreps_out="3x1e", # Final spatial features output irreps_hidden="8x0e + 4x1e", - irreps_sh="1x0e + 1x1e", + irreps_sh="1x0e + 1x1e", irreps_node_attr="3x1e", # Match spatial module output num_layers=2, edge_attr_dim=24, num_attention_heads=1, - reduce=None + reduce=None, ) + def create_spatiotemporal_model() -> E3SpatioTemporal: """Create the complete E3SpatioTemporal model.""" spatial_module = create_spatial_module() temporal_module = create_temporal_module() - + # Create pooling modules spatial_to_temporal_pooler = SpatialTemporalToTemporalNodeAttr(irreps_out="3x1e") # Match spatial module output temporal_to_spatial_pooler = TemporalToSpatialNodeAttrMean(irreps_out="3x1e") # Match temporal module output - + # Compute radial cutoff using temporal average squared distance print("Computing radial cutoff from temporal dataset...") try: @@ -100,73 +101,71 @@ def create_spatiotemporal_model() -> E3SpatioTemporal: dataset = parse_datasets_from_directory( root="/data2/sules/ALA_ALA_enhanced_full_grid/train", traj_pattern="^(.*).xtc", - pdb_pattern="^(.*).pdb", + pdb_pattern="^(.*).pdb", subsample=1, total_lag_time=5, lag_subsample_rate=1, max_datasets=2, # Keep small for testing - num_frames=5 # Small number of frames + num_frames=5, # Small number of frames ) - + # Compute temporal average squared distance temporal_avg_sq_dist = compute_temporal_average_squared_distance_from_datasets( [dataset], # Pass as list since function expects multiple datasets num_samples=50, # Use fewer samples for testing - verbose=True + verbose=True, ) - + # Use a multiple of the temporal average squared distance as the radial cutoff # Typically we might use sqrt(temporal_avg_sq_dist) * some_factor import math + radial_cutoff = math.sqrt(temporal_avg_sq_dist) * 2.0 # Scale factor of 2.0 print(f"Computed radial cutoff: {radial_cutoff:.6f} nm") - + except Exception as e: print(f"Warning: Failed to compute temporal cutoff ({e}), using default value 0.05") radial_cutoff = 0.05 - + return E3SpatioTemporal( spatial_module=spatial_module, temporal_module=temporal_module, spatial_to_temporal_pooler=spatial_to_temporal_pooler, temporal_to_spatial_pooler=temporal_to_spatial_pooler, radial_cutoff=radial_cutoff, - temporal_cutoff=1.0 + temporal_cutoff=1.0, ) + def create_spatiotemporal_conditioner() -> SpatioTemporalConditioner: """Create SpatioTemporalConditioner with E3SpatioTemporal model.""" spatiotemporal_model = create_spatiotemporal_model() - + return SpatioTemporalConditioner( N_structures=1, # Changed to 2 for [y.pos, spatial_features] spatiotemporal_model=spatiotemporal_model, c_noise=0.0, - freeze_spatiotemporal_model=False # Keep trainable + freeze_spatiotemporal_model=False, # Keep trainable ) -def create_conditional_denoiser_config() -> Dict[str, Any]: + +def create_conditional_denoiser_config() -> dict[str, Any]: """Create configuration for Denoiser with spatiotemporal conditioner.""" import functools + import e3tools.nn - + def create_arch(): """Create the E3ConvConditionalSpatioTemporal architecture module.""" # Hidden layer factory - hidden_layer_factory = functools.partial( - e3tools.nn.ConvBlock, - conv=functools.partial(e3tools.nn.Conv) - ) - + hidden_layer_factory = functools.partial(e3tools.nn.ConvBlock, conv=functools.partial(e3tools.nn.Conv)) + # Output head factory - output_head_factory = functools.partial( - e3tools.nn.EquivariantMLP, - irreps_hidden_list=["16x0e + 8x1e"] - ) - + output_head_factory = functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["16x0e + 8x1e"]) + return E3ConvConditionalSpatioTemporal( irreps_out="1x1e", # Output should be 3 components (1x1e) to match position - irreps_hidden="16x0e + 8x1e", + irreps_hidden="16x0e + 8x1e", irreps_sh="1x0e + 1x1e", hidden_layer_factory=hidden_layer_factory, output_head_factory=output_head_factory, @@ -187,247 +186,261 @@ def create_arch(): N_structures=1, # Changed to 2 for [y.pos, spatial_features] input_attr_irreps="3x1e", # spatial_features only (9 components = 3x1e) ) - + def create_optim(params): """Create the optimizer.""" return torch.optim.Adam(params, lr=0.001) - + return { # Required Denoiser parameters (changed from DenoiserWithInputAttr) - 'arch': create_arch, - 'optim': create_optim, - 'sigma_distribution': ConstantSigma(sigma=0.1), - 'max_radius': 1000.0, - 'average_squared_distance': 10.0, # Dummy value for testing - 'add_fixed_noise': False, - 'add_fixed_ones': False, - 'align_noisy_input_during_training': True, - 'align_noisy_input_during_evaluation': True, - 'mean_center': True, - 'mirror_augmentation_rate': 0.0, - 'bond_loss_coefficient': 1.0, - 'normalization_type': "JAMUN", - 'sigma_data': None, - 'lr_scheduler_config': None, - 'use_torch_compile': False, # Disable for testing - 'torch_compile_kwargs': None, - 'conditioner': create_spatiotemporal_conditioner() + "arch": create_arch, + "optim": create_optim, + "sigma_distribution": ConstantSigma(sigma=0.1), + "max_radius": 1000.0, + "average_squared_distance": 10.0, # Dummy value for testing + "add_fixed_noise": False, + "add_fixed_ones": False, + "align_noisy_input_during_training": True, + "align_noisy_input_during_evaluation": True, + "mean_center": True, + "mirror_augmentation_rate": 0.0, + "bond_loss_coefficient": 1.0, + "normalization_type": "JAMUN", + "sigma_data": None, + "lr_scheduler_config": None, + "use_torch_compile": False, # Disable for testing + "torch_compile_kwargs": None, + "conditioner": create_spatiotemporal_conditioner(), } + def add_edges_to_batch(batch: torch_geometric.data.Batch, cutoff: float = 0.05) -> torch_geometric.data.Batch: """Add edges to batch using existing utility from denoiser.""" # Use e3tools radius_graph directly since we don't need the full denoiser add_edges logic import e3tools - - if hasattr(batch, 'edge_index') and batch.edge_index is not None: + + if hasattr(batch, "edge_index") and batch.edge_index is not None: return batch - + # Add radius-based edges edge_index = e3tools.radius_graph(batch.pos, cutoff, batch.batch) batch.edge_index = edge_index - + # Add bonded edges if they exist - if hasattr(batch, 'bonded_edge_index') and batch.bonded_edge_index is not None: - bond_mask = torch.cat([ - torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device), - torch.ones(batch.bonded_edge_index.shape[1], dtype=torch.long, device=batch.pos.device) - ]) + if hasattr(batch, "bonded_edge_index") and batch.bonded_edge_index is not None: + bond_mask = torch.cat( + [ + torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device), + torch.ones(batch.bonded_edge_index.shape[1], dtype=torch.long, device=batch.pos.device), + ] + ) batch.edge_index = torch.cat([edge_index, batch.bonded_edge_index], dim=1) batch.bond_mask = bond_mask else: batch.bond_mask = torch.zeros(edge_index.shape[1], dtype=torch.long, device=batch.pos.device) - + return batch + def load_test_data(): """Load ALA_ALA test dataset.""" print("Loading ALA_ALA dataset...") - + dataset = parse_datasets_from_directory( root="/data2/sules/ALA_ALA_enhanced_full_grid/train", traj_pattern="^(.*).xtc", - pdb_pattern="^(.*).pdb", + pdb_pattern="^(.*).pdb", subsample=1, total_lag_time=5, lag_subsample_rate=1, max_datasets=2, # Keep small for testing - num_frames=5 # Small number of frames + num_frames=5, # Small number of frames ) - + print(f"Loaded dataset with {len(dataset)} samples") - + # Get a sample and create batch graph = dataset[0].__getitem__(0) batch = torch_geometric.data.Batch.from_data_list([graph]) - + # Add edges batch = add_edges_to_batch(batch, cutoff=0.05) - + # Move to device batch = batch.to(device) - - print(f"Batch info:") + + print("Batch info:") print(f" - pos shape: {batch.pos.shape}") print(f" - edge_index shape: {batch.edge_index.shape}") - print(f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}") - if hasattr(batch, 'hidden_state') and batch.hidden_state: + print( + f" - hidden_state length: {len(batch.hidden_state) if hasattr(batch, 'hidden_state') and batch.hidden_state else 0}" + ) + if hasattr(batch, "hidden_state") and batch.hidden_state: print(f" - hidden_state[0] shape: {batch.hidden_state[0].shape}") - + return batch + def test_spatiotemporal_conditioner(conditioner: SpatioTemporalConditioner, batch: torch_geometric.data.Batch): """Test the spatiotemporal conditioner.""" - print("\n" + "="*50) + print("\n" + "=" * 50) print("TESTING SPATIOTEMPORAL CONDITIONER") - print("="*50) - + print("=" * 50) + try: # Test forward pass conditioned_structures = conditioner(batch) - - print(f"āœ… Conditioner forward pass successful!") + + print("āœ… Conditioner forward pass successful!") print(f"Number of conditioned structures: {len(conditioned_structures)} (expected: 2)") print(f"First structure (y.pos) shape: {conditioned_structures[0].shape}") print(f"Second structure (spatial_features) shape: {conditioned_structures[1].shape}") print(f"Original position shape: {batch.pos.shape}") print(f"Position difference norm: {torch.norm(conditioned_structures[0] - batch.pos):.6f}") - + # Verify we got exactly two structures assert len(conditioned_structures) == 2, f"Expected 2 structures, got {len(conditioned_structures)}" - + return True, conditioned_structures - + except Exception as e: print(f"āŒ Conditioner test failed: {e}") import traceback + traceback.print_exc() return False, None + def test_conditional_denoiser_creation(): """Test creating Denoiser with spatiotemporal conditioner.""" - print("\n" + "="*50) + print("\n" + "=" * 50) print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER CREATION") - print("="*50) - + print("=" * 50) + try: # Create configuration config = create_conditional_denoiser_config() - + # Create denoiser (this will instantiate all components) denoiser = Denoiser(**config) denoiser = denoiser.to(device) - - print(f"āœ… Denoiser created successfully!") + + print("āœ… Denoiser created successfully!") print(f"Denoiser device: {next(denoiser.parameters()).device}") print(f"Has conditioner: {hasattr(denoiser, 'conditioning_module')}") print(f"Architecture type: {type(denoiser.g).__name__}") print(f"Conditioner type: {type(denoiser.conditioning_module).__name__}") - + # Check if spatiotemporal model is properly set up - if hasattr(denoiser.conditioning_module, 'spatiotemporal_model'): + if hasattr(denoiser.conditioning_module, "spatiotemporal_model"): st_model = denoiser.conditioning_module.spatiotemporal_model print(f"SpatioTemporal model type: {type(st_model).__name__}") print(f"Spatial module type: {type(st_model.spatial_module).__name__}") print(f"Temporal module type: {type(st_model.temporal_module).__name__}") - + return True, denoiser - + except Exception as e: print(f"āŒ Denoiser creation failed: {e}") import traceback + traceback.print_exc() return False, None + def test_denoiser_forward_pass(denoiser: Denoiser, batch: torch_geometric.data.Batch): """Test the complete denoiser forward pass.""" - print("\n" + "="*50) + print("\n" + "=" * 50) print("TESTING DENOISER WITH SPATIOTEMPORAL CONDITIONER FORWARD PASS") - print("="*50) - + print("=" * 50) + try: # Test with sigma = 0.1 sigma = 0.1 - + # Debug: check conditioned structures shapes conditioned_structures = denoiser.conditioning_module(batch) - print(f"DEBUG: Conditioned structures shapes:") + print("DEBUG: Conditioned structures shapes:") for i, struct in enumerate(conditioned_structures): print(f" Structure {i}: {struct.shape}") - + concatenated = torch.cat([*conditioned_structures], dim=-1) print(f"DEBUG: Concatenated shape: {concatenated.shape}") - print(f"DEBUG: Expected irreps: 4x1e = 12 components") - + print("DEBUG: Expected irreps: 4x1e = 12 components") + with torch.no_grad(): xhat_batch = denoiser.xhat(batch, sigma) - - print(f"āœ… Denoiser forward pass successful!") + + print("āœ… Denoiser forward pass successful!") print(f"Input shape: {batch.pos.shape}") print(f"Output shape: {xhat_batch.pos.shape}") print(f"Output norm: {torch.norm(xhat_batch.pos):.6f}") print(f"Used sigma: {sigma}") - + # Verify output shapes match input assert xhat_batch.pos.shape == batch.pos.shape, f"Shape mismatch: {xhat_batch.pos.shape} vs {batch.pos.shape}" - + return True - + except Exception as e: print(f"āŒ Denoiser forward pass failed: {e}") import traceback + traceback.print_exc() return False + def main(): """Main test function.""" - print("="*60) + print("=" * 60) print("CONDITIONAL DENOISER WITH SPATIOTEMPORAL CONDITIONER TEST") - print("="*60) - + print("=" * 60) + # Load test data batch = load_test_data() - + # Test conditioner creation and forward pass conditioner = create_spatiotemporal_conditioner() conditioner = conditioner.to(device) - + conditioner_success, conditioned_structures = test_spatiotemporal_conditioner(conditioner, batch) - + if not conditioner_success: print("āŒ Conditioner test failed, stopping here.") return - + # Test complete denoiser creation denoiser_success, denoiser = test_conditional_denoiser_creation() - + if not denoiser_success: print("āŒ Denoiser creation failed, stopping here.") return - + # Test complete forward pass forward_success = test_denoiser_forward_pass(denoiser, batch) - + # Final summary - print("\n" + "="*60) + print("\n" + "=" * 60) print("FINAL SUMMARY") - print("="*60) - - print(f"Test Results:") + print("=" * 60) + + print("Test Results:") print(f" - Conditioner test: {'āœ… PASSED' if conditioner_success else 'āŒ FAILED'}") print(f" - Denoiser creation: {'āœ… PASSED' if denoiser_success else 'āŒ FAILED'}") print(f" - Forward pass test: {'āœ… PASSED' if forward_success else 'āŒ FAILED'}") - + if conditioner_success and denoiser_success and forward_success: print("\nšŸŽ‰ ALL TESTS PASSED!") print("The conditional denoiser with spatiotemporal conditioner is working correctly!") else: print("\nāš ļø Some tests failed. Check the output above for details.") - + # Device memory summary if torch.cuda.is_available(): - print(f"\nCUDA Memory:") + print("\nCUDA Memory:") print(f" - Allocated: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") print(f" - Cached: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") + if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/visualize_fake_enhanced_data.py b/scratch/visualize_fake_enhanced_data.py index 38e0fe1..9189cf3 100644 --- a/scratch/visualize_fake_enhanced_data.py +++ b/scratch/visualize_fake_enhanced_data.py @@ -1,76 +1,76 @@ import glob -import os import itertools +import os import re from collections import defaultdict -import mdtraj as md +import matplotlib.colors as colors import matplotlib.pyplot as plt +import mdtraj as md import numpy as np from tqdm import tqdm -import matplotlib.colors as colors + def parse_grid_code_from_filename(filename): """ Parse grid code from trajectory filename of format traj_{grid_code}_{traj_code}.xtc """ basename = os.path.basename(filename) - match = re.match(r'^traj_(\d+)_(\d+)\.xtc$', basename) + match = re.match(r"^traj_(\d+)_(\d+)\.xtc$", basename) if match: return int(match.group(1)), int(match.group(2)) return None, None + def select_trajectories_with_max_per_grid(traj_files, max_traj_per_grid): """ Select trajectories ensuring no grid code has more than max_traj_per_grid trajectories. """ grid_trajectories = defaultdict(list) - + # Group trajectories by grid code for traj_file in traj_files: grid_code, traj_code = parse_grid_code_from_filename(traj_file) if grid_code is not None: grid_trajectories[grid_code].append((traj_file, traj_code)) - + print(f"Found {len(grid_trajectories)} unique grid codes") - + # Limit trajectories per grid code selected_files = [] grid_stats = {} - + for grid_code, traj_list in grid_trajectories.items(): # Sort by trajectory code for deterministic selection traj_list.sort(key=lambda x: x[1]) - + # Select up to max_traj_per_grid trajectories selected_count = min(len(traj_list), max_traj_per_grid) selected_for_grid = traj_list[:selected_count] - - grid_stats[grid_code] = { - 'total': len(traj_list), - 'selected': selected_count - } - + + grid_stats[grid_code] = {"total": len(traj_list), "selected": selected_count} + for traj_file, _ in selected_for_grid: selected_files.append(traj_file) - + # Print statistics - print(f"\nGrid code statistics:") + print("\nGrid code statistics:") print(f"Total grid codes: {len(grid_stats)}") - total_original = sum(stats['total'] for stats in grid_stats.values()) - total_selected = sum(stats['selected'] for stats in grid_stats.values()) + total_original = sum(stats["total"] for stats in grid_stats.values()) + total_selected = sum(stats["selected"] for stats in grid_stats.values()) print(f"Total trajectories: {total_original} -> {total_selected}") print(f"Max trajectories per grid: {max_traj_per_grid}") - + # Show distribution - selected_counts = [stats['selected'] for stats in grid_stats.values()] - print(f"Distribution of selected trajectories per grid:") + selected_counts = [stats["selected"] for stats in grid_stats.values()] + print("Distribution of selected trajectories per grid:") for count in sorted(set(selected_counts)): num_grids = sum(1 for c in selected_counts if c == count) print(f" {count} trajectories: {num_grids} grid codes") - + return sorted(selected_files) + def create_ramachandran_plot(traj_path, topology, output_dir): """ Loads a trajectory, computes phi and psi angles, and saves a Ramachandran plot. @@ -93,21 +93,21 @@ def create_ramachandran_plot(traj_path, topology, output_dir): # Create plot plt.figure(figsize=(8, 8)) # Use hexbin for a nicer look - plt.hexbin(phi_degrees, psi_degrees, gridsize=180, cmap='viridis', mincnt=1) - plt.colorbar(label='Count in bin') - plt.title(f'Ramachandran Plot for {os.path.basename(traj_path)}') - plt.xlabel('Phi (degrees)') - plt.ylabel('Psi (degrees)') + plt.hexbin(phi_degrees, psi_degrees, gridsize=180, cmap="viridis", mincnt=1) + plt.colorbar(label="Count in bin") + plt.title(f"Ramachandran Plot for {os.path.basename(traj_path)}") + plt.xlabel("Phi (degrees)") + plt.ylabel("Psi (degrees)") plt.xlim(-180, 180) plt.ylim(-180, 180) - plt.grid(True, linestyle='--', alpha=0.6) - plt.axhline(0, color='k', linestyle='--', linewidth=0.5) - plt.axvline(0, color='k', linestyle='--', linewidth=0.5) + plt.grid(True, linestyle="--", alpha=0.6) + plt.axhline(0, color="k", linestyle="--", linewidth=0.5) + plt.axvline(0, color="k", linestyle="--", linewidth=0.5) # Save plot output_filename = f"ramachandran_{os.path.basename(traj_path).replace('.xtc', '.png')}" output_path = os.path.join(output_dir, output_filename) - plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.savefig(output_path, dpi=300, bbox_inches="tight") plt.close() @@ -118,25 +118,33 @@ def create_histogram_plot(dihedrals, name1, name2, output_dir, name_string): # Flatten all data for the pair of dihedrals all_x_data = np.concatenate(dihedrals[name1]) all_y_data = np.concatenate(dihedrals[name2]) - + plt.figure(figsize=(10, 10)) - + # Create 2D histogram with density - plt.hist2d(all_x_data, all_y_data, range=((-np.pi, np.pi), (-np.pi, np.pi)),bins=100, cmap='viridis', alpha=0.8, norm=colors.LogNorm()) - plt.colorbar(label='Density') - - plt.title(f'Histogram (Density): {name1} vs {name2}') - plt.xlabel(f'{name1} (radians)') - plt.ylabel(f'{name2} (radians)') + plt.hist2d( + all_x_data, + all_y_data, + range=((-np.pi, np.pi), (-np.pi, np.pi)), + bins=100, + cmap="viridis", + alpha=0.8, + norm=colors.LogNorm(), + ) + plt.colorbar(label="Density") + + plt.title(f"Histogram (Density): {name1} vs {name2}") + plt.xlabel(f"{name1} (radians)") + plt.ylabel(f"{name2} (radians)") plt.xlim(-np.pi, np.pi) plt.ylim(-np.pi, np.pi) - plt.grid(True, linestyle='--', alpha=0.6) - plt.axhline(0, color='k', linestyle='--', linewidth=0.5) - plt.axvline(0, color='k', linestyle='--', linewidth=0.5) - + plt.grid(True, linestyle="--", alpha=0.6) + plt.axhline(0, color="k", linestyle="--", linewidth=0.5) + plt.axvline(0, color="k", linestyle="--", linewidth=0.5) + output_filename = f"histogram_density_{name1}_vs_{name2}_{name_string}.png" output_path = os.path.join(output_dir, output_filename) - plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.savefig(output_path, dpi=300, bbox_inches="tight") plt.close() @@ -179,13 +187,13 @@ def main(max_traj_per_grid=10): traj = md.load(traj_file, top=topology) _, phi_angles = md.compute_phi(traj) _, psi_angles = md.compute_psi(traj) - + if num_phi is None: num_phi = phi_angles.shape[1] num_psi = psi_angles.shape[1] - all_phi_angles.append(phi_angles[:100,:]) - all_psi_angles.append(psi_angles[:100,:]) + all_phi_angles.append(phi_angles[:100, :]) + all_psi_angles.append(psi_angles[:100, :]) except Exception as e: print(f"Could not load or process trajectory {traj_file}. Error: {e}") continue @@ -197,10 +205,10 @@ def main(max_traj_per_grid=10): # Dynamically create dihedral dictionary dihedrals = {} for i in range(num_phi): - dihedrals[f'phi_{i+1}'] = [angles[:, i] for angles in all_phi_angles] + dihedrals[f"phi_{i + 1}"] = [angles[:, i] for angles in all_phi_angles] for i in range(num_psi): - dihedrals[f'psi_{i+1}'] = [angles[:, i] for angles in all_psi_angles] - + dihedrals[f"psi_{i + 1}"] = [angles[:, i] for angles in all_psi_angles] + dihedral_names = list(dihedrals.keys()) # Create line plots (existing functionality) @@ -209,25 +217,25 @@ def main(max_traj_per_grid=10): print("Creating line plots...") for name1, name2 in itertools.combinations(dihedral_names, 2): plt.figure(figsize=(10, 10)) - + for i in tqdm(range(len(traj_files)), desc=f"Plotting {name1} vs {name2}"): x_angles = dihedrals[name1][i] y_angles = dihedrals[name2][i] - plt.plot(x_angles, y_angles, linestyle='-', alpha=0.5) - plt.scatter(x_angles[0], y_angles[0], c='white', marker='o', edgecolor='black', s=50, zorder=5) + plt.plot(x_angles, y_angles, linestyle="-", alpha=0.5) + plt.scatter(x_angles[0], y_angles[0], c="white", marker="o", edgecolor="black", s=50, zorder=5) - plt.title(f'Ramachandran Plot: {name1} vs {name2}') - plt.xlabel(f'{name1} (degrees)') - plt.ylabel(f'{name2} (degrees)') + plt.title(f"Ramachandran Plot: {name1} vs {name2}") + plt.xlabel(f"{name1} (degrees)") + plt.ylabel(f"{name2} (degrees)") plt.xlim(-180, 180) plt.ylim(-180, 180) - plt.grid(True, linestyle='--', alpha=0.6) - plt.axhline(0, color='k', linestyle='--', linewidth=0.5) - plt.axvline(0, color='k', linestyle='--', linewidth=0.5) - + plt.grid(True, linestyle="--", alpha=0.6) + plt.axhline(0, color="k", linestyle="--", linewidth=0.5) + plt.axvline(0, color="k", linestyle="--", linewidth=0.5) + output_filename = f"ramachandran_{name1}_vs_{name2}.png" output_path = os.path.join(output_dir, output_filename) - plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.savefig(output_path, dpi=300, bbox_inches="tight") plt.close() create_histogram_plots = True @@ -244,4 +252,4 @@ def main(max_traj_per_grid=10): if __name__ == "__main__": # Set max trajectories per grid code to 10 max_traj = 10 - main(max_traj_per_grid=max_traj) \ No newline at end of file + main(max_traj_per_grid=max_traj) diff --git a/scratch/visualize_noise_denoise.py b/scratch/visualize_noise_denoise.py index d4374d7..c329688 100644 --- a/scratch/visualize_noise_denoise.py +++ b/scratch/visualize_noise_denoise.py @@ -4,51 +4,45 @@ Generates Ramachandran plots for clean, noisy, and denoised samples. """ -import os import glob +import os +import pdb import tempfile from pathlib import Path -from typing import List, Dict + +import matplotlib.pyplot as plt +import mdtraj as md import numpy as np -from tqdm import tqdm -import hydra -from omegaconf import OmegaConf import torch import torch_geometric -import lightning.pytorch as pl -import matplotlib.pyplot as plt -import mdtraj as md -import wandb -import pdb +from tqdm import tqdm + pdb.set_trace() -from jamun.data import parse_datasets_from_directory, MDtrajDataset -from jamun.metrics._visualize_denoise import VisualizeDenoiseMetrics, plot_ramachandran_grid -from jamun import utils -from jamun.utils.checkpoint import find_checkpoint +from jamun.data import MDtrajDataset, parse_datasets_from_directory +from jamun.metrics._visualize_denoise import plot_ramachandran_grid from jamun.model.denoiser_conditional import Denoiser +from jamun.utils.checkpoint import find_checkpoint + # from jamun.model.denoiser import Denoiser as Denoiser_unconditional def load_model_from_wandb(wandb_run_path: str, checkpoint_type: str = "last", checkpoint_path: str = None): """Load model from wandb run.""" print(f"Loading model from {wandb_run_path}...") - + # Use jamun utilities to find the checkpoint - checkpoint_path_wandb = find_checkpoint( - wandb_train_run_path=wandb_run_path, - checkpoint_type=checkpoint_type - ) + checkpoint_path_wandb = find_checkpoint(wandb_train_run_path=wandb_run_path, checkpoint_type=checkpoint_type) if checkpoint_path is None: checkpoint_path = checkpoint_path_wandb - + print(f"Loading model from checkpoint: {checkpoint_path}") - + # Load the model model = Denoiser.load_from_checkpoint(checkpoint_path) model.eval() - - print(f"āœ“ Model loaded successfully") + + print("āœ“ Model loaded successfully") return model @@ -56,16 +50,16 @@ def create_dataset_from_trajectory(traj_file: str, pdb_file: str, total_lag_time """Create a dataset from a single trajectory file.""" # Create temporary directory structure expected by parse_datasets_from_directory temp_dir = tempfile.mkdtemp() - + # Copy trajectory file to temp directory traj_name = Path(traj_file).stem temp_traj_path = os.path.join(temp_dir, f"{traj_name}.xtc") temp_pdb_path = os.path.join(temp_dir, f"{traj_name}.pdb") - + # Create symlinks os.symlink(traj_file, temp_traj_path) os.symlink(pdb_file, temp_pdb_path) - + # Parse dataset datasets = parse_datasets_from_directory( root=temp_dir, @@ -76,9 +70,9 @@ def create_dataset_from_trajectory(traj_file: str, pdb_file: str, total_lag_time total_lag_time=total_lag_time, lag_subsample_rate=1, max_datasets=1, - label_override=traj_name + label_override=traj_name, ) - + return datasets[0] if datasets else None @@ -86,22 +80,19 @@ def process_trajectory(model, dataset: MDtrajDataset, sigma: float = 0.04): """Process a single trajectory through the model.""" # Create dataloader dataloader = torch.utils.data.DataLoader( - dataset, - batch_size=32, - shuffle=False, - collate_fn=torch_geometric.data.Batch.from_data_list + dataset, batch_size=32, shuffle=False, collate_fn=torch_geometric.data.Batch.from_data_list ) - + # Store all samples all_clean = [] all_noisy = [] all_denoised = [] - + model.eval() with torch.no_grad(): for batch in dataloader: batch = batch.to(model.device) - + # # Ensure all batch attributes are the correct dtype # if hasattr(batch, 'pos'): # batch.pos = batch.pos.float() @@ -111,36 +102,36 @@ def process_trajectory(model, dataset: MDtrajDataset, sigma: float = 0.04): # batch.edge_index = batch.edge_index.long() # Keep as long for indexing! # if hasattr(batch, 'edge_attr') and batch.edge_attr is not None: # batch.edge_attr = batch.edge_attr.float() - + # Convert sigma to tensor with correct dtype and device sigma_tensor = torch.tensor(sigma, dtype=torch.float32, device=model.device) - + # Run noise and denoise _, xhat, y = model.noise_and_denoise( batch, sigma_tensor, align_noisy_input=model.align_noisy_input_during_evaluation ) - + # Convert to data lists clean_samples = torch_geometric.data.Batch.to_data_list(batch) noisy_samples = torch_geometric.data.Batch.to_data_list(y) denoised_samples = torch_geometric.data.Batch.to_data_list(xhat) - + all_clean.extend(clean_samples) all_noisy.extend(noisy_samples) all_denoised.extend(denoised_samples) - + return all_clean, all_noisy, all_denoised -def samples_to_trajectory(samples: List, dataset: MDtrajDataset): +def samples_to_trajectory(samples: list, dataset: MDtrajDataset): """Convert list of samples to MDTraj trajectory.""" coordinates = [] for sample in samples: coords = sample.pos.cpu().numpy() coordinates.append(coords) - + coords_array = np.array(coordinates) # Shape: (n_frames, n_atoms, 3) - + # Create trajectory traj = md.Trajectory(coords_array, dataset.topology) return traj @@ -148,15 +139,11 @@ def samples_to_trajectory(samples: List, dataset: MDtrajDataset): def create_ramachandran_plot(clean_traj, noisy_traj, denoised_traj, title: str, save_path: str): """Create and save Ramachandran plot for three trajectories.""" - trajs = { - "x": clean_traj, - "y": noisy_traj, - "xhat": denoised_traj - } - + trajs = {"x": clean_traj, "y": noisy_traj, "xhat": denoised_traj} + try: fig, axes = plot_ramachandran_grid(trajs, title) - fig.savefig(save_path, dpi=300, bbox_inches='tight') + fig.savefig(save_path, dpi=300, bbox_inches="tight") plt.close(fig) print(f"āœ“ Saved Ramachandran plot: {save_path}") except Exception as e: @@ -168,22 +155,22 @@ def main(): wandb_run_path = "sule-shashank/jamun/4p0ejn0z" val_dir = "/data2/sules/ALA_ALA_enhanced_full_grid/val" pdb_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.pdb" - checkpoint_type = 'epoch=49-step=52900-v1.ckpt' + checkpoint_type = "epoch=49-step=52900-v1.ckpt" # checkpoint_path = "/data2/sules/jamun-conditional-runs/outputs/train/dev/runs/2025-07-31_00-43-14/checkpoints/epoch=9-step=10051.ckpt" total_lag_time = 5 sigma = 0.04 output_dir = "val_ramachandrans" - + # Create output directory os.makedirs(output_dir, exist_ok=True) - + # Load model checkpoint_path = find_checkpoint(wandb_run_path, checkpoint_type=checkpoint_type) model = Denoiser.load_from_checkpoint(checkpoint_path) model.eval() - model.to('cuda:0') - print(f"Model loaded and moved to cuda:0") - # config_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/wandb/run-20250805_042516-yqn9mm7x/files/config.yaml" + model.to("cuda:0") + print("Model loaded and moved to cuda:0") + # config_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/wandb/run-20250805_042516-yqn9mm7x/files/config.yaml" # cfg = OmegaConf.load(config_path) # checkpoint_path = "/data2/sules/jamun-conditional-runs//outputs/train/dev/runs/2025-08-05_04-24-31/checkpoints/last.ckpt" # model = hydra.utils.instantiate(cfg.cfg.value.model) @@ -196,15 +183,12 @@ def main(): # Get all trajectory files traj_files = glob.glob(os.path.join(val_dir, "*.xtc")) traj_files.sort() - + print(f"Found {len(traj_files)} trajectory files") - # do one trial run + # do one trial run dataset = create_dataset_from_trajectory(traj_files[0], pdb_file, total_lag_time) dataloader = torch.utils.data.DataLoader( - dataset, - batch_size=32, - shuffle=False, - collate_fn=torch_geometric.data.Batch.from_data_list + dataset, batch_size=32, shuffle=False, collate_fn=torch_geometric.data.Batch.from_data_list ) _, batch = next(enumerate(dataloader)) @@ -221,46 +205,42 @@ def main(): # Process each trajectory with progress bar for traj_file in tqdm(traj_files, desc="Processing trajectories"): traj_name = Path(traj_file).stem - + try: # Create dataset dataset = create_dataset_from_trajectory(traj_file, pdb_file, total_lag_time) if dataset is None: tqdm.write(f"Failed to create dataset for {traj_name}") continue - + # Process trajectory # breakpoint() clean_samples, noisy_samples, denoised_samples = process_trajectory(model, dataset, sigma) - + # Store samples for concatenated analysis all_clean_samples.extend(clean_samples) all_noisy_samples.extend(noisy_samples) all_denoised_samples.extend(denoised_samples) - + except Exception as e: tqdm.write(f"Error processing {traj_name}: {e}") continue - + # Create concatenated analysis if all_clean_samples: print("\nCreating concatenated Ramachandran plot...") - + # Use the last dataset for topology (they should all be the same) concat_clean_traj = samples_to_trajectory(all_clean_samples, dataset) concat_noisy_traj = samples_to_trajectory(all_noisy_samples, dataset) concat_denoised_traj = samples_to_trajectory(all_denoised_samples, dataset) - + # Create concatenated plot concat_plot_path = os.path.join(output_dir, "concatenated_ramachandran.png") create_ramachandran_plot( - concat_clean_traj, - concat_noisy_traj, - concat_denoised_traj, - "Concatenated Trajectories", - concat_plot_path + concat_clean_traj, concat_noisy_traj, concat_denoised_traj, "Concatenated Trajectories", concat_plot_path ) - + print(f"\nAnalysis complete! Concatenated Ramachandran plot saved in {output_dir}/") print(f"Processed {len(all_clean_samples)} total samples from {len(traj_files)} trajectories") else: @@ -268,4 +248,4 @@ def main(): if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/scratch/visualize_traj_data.py b/scratch/visualize_traj_data.py index 4da7d24..5cb9780 100644 --- a/scratch/visualize_traj_data.py +++ b/scratch/visualize_traj_data.py @@ -1,9 +1,10 @@ -import mdtraj as md +import itertools import os + +import matplotlib.colors as colors import matplotlib.pyplot as plt +import mdtraj as md import numpy as np -import itertools -import matplotlib.colors as colors # Set file paths for ALA_ALA in capped diamines xtc_file = "/data/bucket/kleinhej/capped_diamines/timewarp_splits/train/ALA_ALA.xtc" @@ -30,9 +31,9 @@ # Each entry is (n_frames,) dihedrals = {} for i in range(num_phi): - dihedrals[f'phi_{i+1}'] = phi_angles[:, i] + dihedrals[f"phi_{i + 1}"] = phi_angles[:, i] for i in range(num_psi): - dihedrals[f'psi_{i+1}'] = psi_angles[:, i] + dihedrals[f"psi_{i + 1}"] = psi_angles[:, i] dihedral_names = list(dihedrals.keys()) @@ -41,19 +42,19 @@ x = dihedrals[name1] y = dihedrals[name2] plt.figure(figsize=(8, 8)) - plt.hist2d(x, y, bins=100, range=((-np.pi, np.pi), (-np.pi, np.pi)), cmap='viridis', norm=colors.LogNorm()) - plt.colorbar(label='Density') - plt.title(f'2D Histogram: {name1} vs {name2}') - plt.xlabel(f'{name1} (radians)') - plt.ylabel(f'{name2} (radians)') + plt.hist2d(x, y, bins=100, range=((-np.pi, np.pi), (-np.pi, np.pi)), cmap="viridis", norm=colors.LogNorm()) + plt.colorbar(label="Density") + plt.title(f"2D Histogram: {name1} vs {name2}") + plt.xlabel(f"{name1} (radians)") + plt.ylabel(f"{name2} (radians)") plt.xlim(-np.pi, np.pi) plt.ylim(-np.pi, np.pi) - plt.grid(True, linestyle='--', alpha=0.6) - plt.axhline(0, color='k', linestyle='--', linewidth=0.5) - plt.axvline(0, color='k', linestyle='--', linewidth=0.5) + plt.grid(True, linestyle="--", alpha=0.6) + plt.axhline(0, color="k", linestyle="--", linewidth=0.5) + plt.axvline(0, color="k", linestyle="--", linewidth=0.5) output_filename = f"hist2d_{name1}_vs_{name2}_true_distribution.png" output_path = os.path.join(output_dir, output_filename) - plt.savefig(output_path, dpi=300, bbox_inches='tight') + plt.savefig(output_path, dpi=300, bbox_inches="tight") plt.close() -print(f"All 2D histograms saved in {output_dir}") \ No newline at end of file +print(f"All 2D histograms saved in {output_dir}") diff --git a/src/jamun/cmdline/sample.py b/src/jamun/cmdline/sample.py index ac4bdf6..f23d22a 100644 --- a/src/jamun/cmdline/sample.py +++ b/src/jamun/cmdline/sample.py @@ -1,8 +1,7 @@ import os import sys import traceback -from typing import Sequence -import pdb +from collections.abc import Sequence import dotenv import e3nn @@ -29,17 +28,18 @@ logging.basicConfig(format="[%(asctime)s][%(name)s][%(levelname)s] - %(message)s", level=logging.INFO) logger = logging.getLogger("load_wandb_checkpoint") -dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ +dotenv.load_dotenv("../.env", verbose=True) # Adjust path if script is not in scratch/ JAMUN_DATA_PATH = os.getenv("JAMUN_DATA_PATH") JAMUN_ROOT_PATH = os.getenv("JAMUN_ROOT_PATH") -project_root = "/homefs/home/sules/jamun" # Adjust if necessary +project_root = "/homefs/home/sules/jamun" # Adjust if necessary if project_root not in sys.path: sys.path.insert(0, project_root) logger.info(f"Added '{project_root}' to sys.path for module discovery.") else: logger.info(f"'{project_root}' is already in sys.path.") + def get_initial_graphs( datasets: Sequence[MDtrajDataset], num_init_samples_per_dataset: int, repeat: int = 1 ) -> torch_geometric.data.Batch: @@ -86,14 +86,16 @@ def run(cfg): # Overwrite the checkpoint path in the config. cfg.model.checkpoint_path = checkpoint_path model = hydra.utils.instantiate(cfg.model) - + # Set default graph_type to "fan" if spatiotemporal model exists but doesn't have graph_type - if (hasattr(model, 'conditioner') and hasattr(model.conditioner, 'spatiotemporal_model') and - not hasattr(model.conditioner.spatiotemporal_model, 'graph_type')): + if ( + hasattr(model, "conditioner") + and hasattr(model.conditioner, "spatiotemporal_model") + and not hasattr(model.conditioner.spatiotemporal_model, "graph_type") + ): model.conditioner.spatiotemporal_model.graph_type = "fan" - - - print(f'Checkpoint path at: {checkpoint_path}') + + print(f"Checkpoint path at: {checkpoint_path}") init_datasets = hydra.utils.instantiate(cfg.init_datasets) # breakpoint() init_graphs = get_initial_graphs( diff --git a/src/jamun/cmdline/train.py b/src/jamun/cmdline/train.py index 5a461ac..f43cdaa 100644 --- a/src/jamun/cmdline/train.py +++ b/src/jamun/cmdline/train.py @@ -14,49 +14,50 @@ e3nn.set_optimization_defaults(jit_script_fx=False) +import math + import jamun # noqa: E402 from jamun.hydra import instantiate_dict_cfg # noqa: E402 from jamun.hydra.utils import format_resolver # noqa: E402 from jamun.utils import compute_average_squared_distance_from_datasets, dist_log, find_checkpoint # noqa: E402 from jamun.utils._normalizations import normalization_factors # noqa: E402 from jamun.utils.average_squared_distance import compute_temporal_average_squared_distance_from_datasets # noqa: E402 -import math -from jamun.utils._normalizations import normalization_factors dotenv.load_dotenv(".env", verbose=True) OmegaConf.register_new_resolver("format", format_resolver) + def compute_radial_cutoff(max_radius: float, average_squared_distance: float, sigma: float, D: int = 3) -> float: """ Compute radial cutoff using the same formula as the denoiser. - + This replicates the computation from denoiser_conditional.py: radial_cutoff = effective_radial_cutoff(sigma) / c_in where: - effective_radial_cutoff = sqrt(max_radiusĀ² + 6ĻƒĀ²) - c_in = 1.0 / sqrt(average_squared_distance + 2DĻƒĀ²) - + Args: max_radius: Maximum radius parameter average_squared_distance: Average squared distance from dataset sigma: Noise level D: Dimensionality (default 3 for 3D coordinates) - + Returns: Computed radial cutoff """ # Effective radial cutoff based on noise level effective_radial_cutoff = math.sqrt(max_radius**2 + 6 * sigma**2) - + # JAMUN normalization factor c_in A = average_squared_distance B = 2 * D * sigma**2 c_in = 1.0 / math.sqrt(A + B) - + # Final radial cutoff radial_cutoff = effective_radial_cutoff / c_in - - print(f"Radial cutoff computation:") + + print("Radial cutoff computation:") print(f" max_radius: {max_radius}") print(f" average_squared_distance: {average_squared_distance}") print(f" sigma: {sigma}") @@ -64,9 +65,10 @@ def compute_radial_cutoff(max_radius: float, average_squared_distance: float, si print(f" effective_radial_cutoff: {effective_radial_cutoff}") print(f" c_in: {c_in}") print(f" final radial_cutoff: {radial_cutoff}") - + return radial_cutoff + def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the average squared distance for normalization from the data.""" datamodule = hydra.utils.instantiate(cfg.data.datamodule) @@ -76,19 +78,21 @@ def compute_average_squared_distance_from_config(cfg: OmegaConf) -> float: average_squared_distance = compute_average_squared_distance_from_datasets(train_datasets, cutoff) return average_squared_distance + def compute_temporal_average_squared_distance_from_config(cfg: OmegaConf) -> float: """Computes the temporal average squared distance for normalization from the data.""" datamodule = hydra.utils.instantiate(cfg.data.datamodule) datamodule.setup("compute_normalization") train_datasets = datamodule.datasets["train"] - + average_squared_distance = compute_temporal_average_squared_distance_from_datasets( - train_datasets, + train_datasets, num_samples=100, # Use reasonable number of samples - verbose=True + verbose=True, ) return average_squared_distance + def run(cfg): log_cfg = OmegaConf.to_container(cfg, throw_on_missing=True, resolve=True) @@ -97,7 +101,7 @@ def run(cfg): dist_log(f"{torch.__config__.parallel_info()}") dist_log(f"CUDA_VISIBLE_DEVICES: {os.environ.get('CUDA_VISIBLE_DEVICES')}") dist_log(f"{os.sched_getaffinity(0)=}") - + # Set the start method to spawn to avoid issues with the default fork method. torch.multiprocessing.set_start_method("spawn", force=True) @@ -121,7 +125,10 @@ def run(cfg): c_noise_float = float(c_noise) # Compute normalization factors for conditioner c_in parameter - if cfg.model.get("conditioner") and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.DenoisedConditioner": + if ( + cfg.model.get("conditioner") + and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.DenoisedConditioner" + ): if hasattr(cfg.model.sigma_distribution, "sigma"): dist_log(f"Computing normalization factors for DenoisedConditioner with sigma={sigma}") dist_log(f" average_squared_distance: {average_squared_distance}") @@ -129,19 +136,23 @@ def run(cfg): dist_log(f" c_skip: {c_skip}") dist_log(f" c_out: {c_out}") dist_log(f" c_noise: {c_noise}") - + cfg.model.conditioner.c_in = c_in_float dist_log(f"Set cfg.model.conditioner.c_in to {c_in_float}") # breakpoint() - if cfg.model.get("conditioner") and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner": + if ( + cfg.model.get("conditioner") + and cfg.model.conditioner.get("_target_") == "jamun.model.conditioners.conditioners.SpatioTemporalConditioner" + ): cfg.model.conditioner.spatiotemporal_model.radial_cutoff = average_squared_distance max_radius = cfg.model.max_radius temporal_average_squared_distance = compute_temporal_average_squared_distance_from_config(cfg) temporal_radial_cutoff = compute_radial_cutoff( - max_radius=max_radius, - average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model - sigma=sigma, - D=3) + max_radius=max_radius, + average_squared_distance=temporal_average_squared_distance, # Use temporal for spatiotemporal model + sigma=sigma, + D=3, + ) cfg.model.conditioner.spatiotemporal_model.temporal_cutoff = temporal_radial_cutoff cfg.model.conditioner.c_noise = c_noise_float cfg.model.conditioner.c_in = c_in_float @@ -202,7 +213,7 @@ def run(cfg): ) else: checkpoint_path = None - print(f'Saving checkpoints @ {checkpoint_path}') + print(f"Saving checkpoints @ {checkpoint_path}") trainer.fit(model, datamodule=datamodule, ckpt_path=checkpoint_path) # breakpoint() diff --git a/src/jamun/data/__init__.py b/src/jamun/data/__init__.py index ab98bb5..a49bd3c 100644 --- a/src/jamun/data/__init__.py +++ b/src/jamun/data/__init__.py @@ -1,7 +1,6 @@ from ._dloader import MDtrajDataModule, RandomChainDataset, StreamingRandomChainDataset from ._mdtraj import MDtrajDataset, MDtrajIterableDataset from ._sdf import MDtrajSDFDataset -from .noisy_position_dataset import RepeatedPositionDataset from ._utils import ( concatenate_datasets, create_dataset_from_pdbs, @@ -11,3 +10,4 @@ parse_repeated_position_datasets_from_directory, parse_sdf_datasets_from_directory, ) +from .noisy_position_dataset import RepeatedPositionDataset diff --git a/src/jamun/data/_mdtraj.py b/src/jamun/data/_mdtraj.py index 4e21b8d..3e83386 100644 --- a/src/jamun/data/_mdtraj.py +++ b/src/jamun/data/_mdtraj.py @@ -1,6 +1,6 @@ import functools import os -from typing import Callable, Optional, Sequence, Tuple, List +from collections.abc import Callable, Sequence import mdtraj as md import numpy as np @@ -16,39 +16,39 @@ def get_subsampled_indices( subsample_rate: int, total_lag_time: int, lag_subsample_rate: int, -) -> List[np.ndarray]: +) -> list[np.ndarray]: """ Generate subsampled indices and their corresponding lagged indices. - + Args: N: Total number of frames subsample_rate: Rate at which to subsample the frames total_lag_time: Number of lagged frames to generate for each subsampled frame lag_subsample_rate: Rate at which to subsample the lagged frames - + Returns: List of arrays, where each array contains the lagged indices for a subsampled frame - + Raises: ValueError: If the input parameters don't satisfy the required constraints """ # Check guardrails if N / subsample_rate < 1: - raise ValueError(f"Number of samples (N/subsample_rate = {N/subsample_rate}) must be >= 1") - + raise ValueError(f"Number of samples (N/subsample_rate = {N / subsample_rate}) must be >= 1") + # Generate subsampled indices subsampled_indices = np.arange(0, N, subsample_rate) - + # Generate lagged indices for each subsampled index lagged_indices = [] for idx in subsampled_indices: # Calculate lagged indices lagged = [int(idx - j * lag_subsample_rate) for j in range(total_lag_time)] - + # Check if we have enough lagged indices if len(lagged) == total_lag_time and all(x >= 0 for x in lagged): - lagged_indices.append(lagged) - + lagged_indices.append(lagged) + return lagged_indices @@ -77,7 +77,7 @@ def make_graph_from_topology( num_residues=residue_sequence_index.max().item() + 1, bonded_edge_index=bonds, pos=None, - hidden_state=None + hidden_state=None, ) graph.residues = [x.residue.name for x in topology.atoms] graph.atom_names = [x.name for x in topology.atoms] @@ -203,12 +203,12 @@ def __init__( traj_files: Sequence[str], pdb_file: str, label: str, - num_frames: Optional[int] = None, - start_frame: Optional[int] = None, - transform: Optional[Callable] = None, - subsample: Optional[int] = None, - total_lag_time: Optional[int] = None, - lag_subsample_rate: Optional[int] = None, + num_frames: int | None = None, + start_frame: int | None = None, + transform: Callable | None = None, + subsample: int | None = None, + total_lag_time: int | None = None, + lag_subsample_rate: int | None = None, loss_weight: float = 1.0, verbose: bool = False, keep_hydrogens: bool = False, @@ -244,21 +244,19 @@ def __init__( if subsample is None or subsample == 0: subsample = 1 - + # Get lagged indices if lag parameters are provided if total_lag_time is not None and lag_subsample_rate is not None: # print(f"total_lag_time: {total_lag_time}, lag_subsample_rate: {lag_subsample_rate}") - self.traj = self.traj[start_frame : start_frame + num_frames] # accommodate for start_frame and num_frames - lagged_indices = get_subsampled_indices( - self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate - ) + self.traj = self.traj[start_frame : start_frame + num_frames] # accommodate for start_frame and num_frames + lagged_indices = get_subsampled_indices(self.traj.n_frames, subsample, total_lag_time, lag_subsample_rate) # Extract subsampled indices (first element of each list) subsampled_indices = [indices[0] for indices in lagged_indices] # Extract lagged indices (all except first element) self.lagged_indices = [indices[1:] for indices in lagged_indices] # Subsample the trajectory using the subsampled indices self.hidden_state = [self.traj[indices] for indices in self.lagged_indices] - self.traj = self.traj[subsampled_indices] # self.traj is permanently modified. + self.traj = self.traj[subsampled_indices] # self.traj is permanently modified. else: # Regular subsampling without lag # print(f"subsample: {subsample}, regular subsampling") @@ -271,7 +269,9 @@ def __init__( self.graph = make_graph_from_topology(self.top) self.traj = self.traj.atom_slice(atom_selection) if self.hidden_state is not None: - self.hidden_state = [traj.atom_slice(atom_selection) for traj in self.hidden_state] # select protein atoms for hidden state(s) + self.hidden_state = [ + traj.atom_slice(atom_selection) for traj in self.hidden_state + ] # select protein atoms for hidden state(s) self.graph.pos = torch.tensor(self.traj.xyz[0], dtype=torch.float32) self.graph.loss_weight = torch.tensor([loss_weight], dtype=torch.float32) @@ -299,12 +299,14 @@ def save_topology_pdb(self, filename: str | None = None): def __getitem__(self, idx): graph = self.graph.clone() graph.pos = torch.tensor(self.traj.xyz[idx]) - + if self.hidden_state is not None: - graph.hidden_state = [torch.tensor(self.hidden_state[idx].xyz[i]) for i in range(self.hidden_state[idx].n_frames)] + graph.hidden_state = [ + torch.tensor(self.hidden_state[idx].xyz[i]) for i in range(self.hidden_state[idx].n_frames) + ] else: graph.hidden_state = [] - + if self.transform: graph = self.transform(graph) return graph diff --git a/src/jamun/data/_subsample.py b/src/jamun/data/_subsample.py index 635d04c..fcc3973 100644 --- a/src/jamun/data/_subsample.py +++ b/src/jamun/data/_subsample.py @@ -1,5 +1,4 @@ import numpy as np -from typing import Tuple, List def get_subsampled_indices( @@ -7,45 +6,45 @@ def get_subsampled_indices( subsample_rate: int, total_lag_time: int, lag_subsample_rate: int, -) -> List[np.ndarray]: +) -> list[np.ndarray]: """ Generate subsampled indices and their corresponding lagged indices. - + Args: N: Total number of frames subsample_rate: Rate at which to subsample the frames total_lag_time: Number of lagged frames to generate for each subsampled frame lag_subsample_rate: Rate at which to subsample the lagged frames - + Returns: List of arrays, where each array contains the lagged indices for a subsampled frame - + Raises: ValueError: If the input parameters don't satisfy the required constraints """ # Check guardrails if N / subsample_rate < 1: - raise ValueError(f"Number of samples (N/subsample_rate = {N/subsample_rate}) must be >= 1") - + raise ValueError(f"Number of samples (N/subsample_rate = {N / subsample_rate}) must be >= 1") + # if total_lag_time * lag_subsample_rate > subsample_rate: # raise ValueError( # f"total_lag_time * lag_subsample_rate ({total_lag_time * lag_subsample_rate}) " # f"must be <= subsample_rate ({subsample_rate})" # ) - + # Generate subsampled indices subsampled_indices = np.arange(0, N, subsample_rate) - + # Generate lagged indices for each subsampled index lagged_indices = [] for idx in subsampled_indices: # Calculate lagged indices lagged = [int(idx - j * lag_subsample_rate) for j in range(total_lag_time)] - + # Check if we have enough lagged indices if len(lagged) == total_lag_time and all(x >= 0 for x in lagged): - lagged_indices.append(lagged) - + lagged_indices.append(lagged) + return lagged_indices @@ -54,34 +53,34 @@ def get_subsampled_trajectory( subsample_rate: int, total_lag_time: int, lag_subsample_rate: int, -) -> Tuple[np.ndarray, List[np.ndarray]]: +) -> tuple[np.ndarray, list[np.ndarray]]: """ Subsample a trajectory and generate lagged states for each subsampled frame. - + Args: positions: Array of shape (N, ...) containing trajectory positions subsample_rate: Rate at which to subsample the frames total_lag_time: Number of lagged frames to generate for each subsampled frame lag_subsample_rate: Rate at which to subsample the lagged frames - + Returns: Tuple containing: - subsampled_positions: Array of subsampled positions - lagged_positions: List of arrays, where each array contains the lagged positions for the corresponding subsampled frame - + Raises: ValueError: If the input parameters don't satisfy the required constraints """ N = len(positions) - + # Get the lagged indices lagged_indices = get_subsampled_indices(N, subsample_rate, total_lag_time, lag_subsample_rate) - + # Extract subsampled positions (first element of each lagged indices list) subsampled_positions = np.array([positions[indices[0]] for indices in lagged_indices]) - + # Generate lagged positions for each subsampled frame - lagged_positions = [ [positions[idx] for idx in indices[1:]] for indices in lagged_indices] - - return subsampled_positions, lagged_positions \ No newline at end of file + lagged_positions = [[positions[idx] for idx in indices[1:]] for indices in lagged_indices] + + return subsampled_positions, lagged_positions diff --git a/src/jamun/data/_utils.py b/src/jamun/data/_utils.py index 75bef0f..20fc614 100644 --- a/src/jamun/data/_utils.py +++ b/src/jamun/data/_utils.py @@ -11,7 +11,6 @@ from jamun.data._mdtraj import MDtrajDataset, MDtrajIterableDataset from jamun.data._sdf import MDtrajSDFDataset -from typing import Optional, List, Sequence, Dict, Any def dloader_map_reduce(f, dloader, reduce_fn=torch.cat, verbose: bool = False): @@ -44,7 +43,7 @@ def parse_datasets_from_directory( max_datasets_offset: int | None = None, filter_codes: Sequence[str] | None = None, as_iterable: bool = False, - label_override: Optional[str] = None, + label_override: str | None = None, **dataset_kwargs, ) -> list[MDtrajDataset]: """Helper function to create MDtrajDataset objects from a directory of trajectory files.""" @@ -362,24 +361,24 @@ def create_dataset_from_pdbs(pdbfiles: str, label_prefix: str | None = None) -> def parse_repeated_position_datasets_from_directory( root: str, traj_pattern: str, - pdb_pattern: Optional[str] = None, - pdb_file: Optional[Sequence[str]] = None, - max_datasets: Optional[int] = None, - max_datasets_offset: Optional[int] = None, - filter_codes: Optional[Sequence[str]] = None, + pdb_pattern: str | None = None, + pdb_file: Sequence[str] | None = None, + max_datasets: int | None = None, + max_datasets_offset: int | None = None, + filter_codes: Sequence[str] | None = None, as_iterable: bool = False, - label_override: Optional[str] = None, + label_override: str | None = None, **dataset_kwargs, -) -> List: +) -> list: """Helper function to create RepeatedPositionDataset objects from a directory of trajectory files.""" # Import here to avoid circular imports from jamun.data.noisy_position_dataset import RepeatedPositionDataset - + # Print the dataset_kwargs for debugging - print(f"=== parse_repeated_position_datasets_from_directory dataset_kwargs ===") + print("=== parse_repeated_position_datasets_from_directory dataset_kwargs ===") print(f"dataset_kwargs: {dataset_kwargs}") - print(f"=== End dataset_kwargs ===") - + print("=== End dataset_kwargs ===") + if pdb_file is not None and pdb_pattern is not None: raise ValueError("Exactly one of pdb_file and pdb_pattern should be provided.") diff --git a/src/jamun/data/noisy_position_dataset.py b/src/jamun/data/noisy_position_dataset.py index 62687f6..c8d070c 100644 --- a/src/jamun/data/noisy_position_dataset.py +++ b/src/jamun/data/noisy_position_dataset.py @@ -1,4 +1,3 @@ -import torch from jamun.data._mdtraj import MDtrajDataset @@ -8,30 +7,30 @@ class RepeatedPositionDataset(MDtrajDataset): This is used for Model 3 experiment where the structures passed to the denoiser are copies of the same structure given by y.pos. The denoiser will add noise during training. """ - + def __init__(self, *args, **kwargs): """Initialize but store total_lag_time before modifying parent behavior.""" # Store the total_lag_time for our own use - self._target_total_lag_time = kwargs.get('total_lag_time', 2) - + self._target_total_lag_time = kwargs.get("total_lag_time", 2) + # Prevent parent from doing lag processing by removing lag parameters kwargs_no_lag = kwargs.copy() - kwargs_no_lag['total_lag_time'] = None - kwargs_no_lag['lag_subsample_rate'] = None - + kwargs_no_lag["total_lag_time"] = None + kwargs_no_lag["lag_subsample_rate"] = None + super().__init__(*args, **kwargs_no_lag) def __getitem__(self, idx): """Override to create position copies instead of using real hidden states.""" # Get the normal item from parent class (without lag processing) graph = super().__getitem__(idx) - + # Create the number of hidden states we want based on our target total_lag_time num_hidden_states = self._target_total_lag_time - 1 - + graph.hidden_state = [] for _ in range(num_hidden_states): # Create a copy of the current position (no noise added here) graph.hidden_state.append(graph.pos.clone()) - - return graph \ No newline at end of file + + return graph diff --git a/src/jamun/data/tests/test_subsample.py b/src/jamun/data/tests/test_subsample.py index 23b3248..3e94d01 100644 --- a/src/jamun/data/tests/test_subsample.py +++ b/src/jamun/data/tests/test_subsample.py @@ -11,37 +11,39 @@ def test_basic_functionality(): total_lag_time = 3 lag_subsample_rate = 10 - print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " - f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") - - breakpoint() # Debug point 1: Check input parameters before function call - - lagged_indices = get_subsampled_indices( - N, subsample_rate, total_lag_time, lag_subsample_rate + print( + f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}" ) + breakpoint() # Debug point 1: Check input parameters before function call + + lagged_indices = get_subsampled_indices(N, subsample_rate, total_lag_time, lag_subsample_rate) + breakpoint() # Debug point 2: Check function output - + # Extract subsampled indices (first element of each list) subsampled_indices = np.array([indices[0] for indices in lagged_indices]) print(f"Subsampled indices: {subsampled_indices}") print(f"Number of lagged indices lists: {len(lagged_indices)}") - + # Check subsampled indices expected_subsampled = np.array([20, 30, 40, 50, 60, 70, 80, 90]) - assert np.array_equal(subsampled_indices, expected_subsampled), \ + assert np.array_equal(subsampled_indices, expected_subsampled), ( f"Expected {expected_subsampled}, got {subsampled_indices}" + ) # Check lagged indices for i, lagged in enumerate(lagged_indices): - expected_lagged = np.array([ - subsampled_indices[i], - subsampled_indices[i] - lag_subsample_rate, - subsampled_indices[i] - 2 * lag_subsample_rate - ]) - assert np.array_equal(lagged, expected_lagged), \ - f"For index {i}, expected {expected_lagged}, got {lagged}" - + expected_lagged = np.array( + [ + subsampled_indices[i], + subsampled_indices[i] - lag_subsample_rate, + subsampled_indices[i] - 2 * lag_subsample_rate, + ] + ) + assert np.array_equal(lagged, expected_lagged), f"For index {i}, expected {expected_lagged}, got {lagged}" + print("Basic functionality test passed!") @@ -53,29 +55,27 @@ def test_edge_cases(): total_lag_time = 1 lag_subsample_rate = 1 - print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " - f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") - - breakpoint() # Debug point 5: Check edge case parameters before function call - - lagged_indices = get_subsampled_indices( - N, subsample_rate, total_lag_time, lag_subsample_rate + print( + f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}" ) + breakpoint() # Debug point 5: Check edge case parameters before function call + + lagged_indices = get_subsampled_indices(N, subsample_rate, total_lag_time, lag_subsample_rate) + breakpoint() # Debug point 6: Check edge case results - + # Extract subsampled indices (first element of each list) subsampled_indices = np.array([indices[0] for indices in lagged_indices]) print(f"Subsampled indices: {subsampled_indices}") print(f"Lagged indices: {lagged_indices}") - + assert len(subsampled_indices) == 1, f"Expected 1 subsampled index, got {len(subsampled_indices)}" assert len(lagged_indices) == 1, f"Expected 1 lagged indices list, got {len(lagged_indices)}" - assert np.array_equal(subsampled_indices, np.array([0])), \ - f"Expected [0], got {subsampled_indices}" - assert np.array_equal(lagged_indices[0], np.array([0])), \ - f"Expected [0], got {lagged_indices[0]}" - + assert np.array_equal(subsampled_indices, np.array([0])), f"Expected [0], got {subsampled_indices}" + assert np.array_equal(lagged_indices[0], np.array([0])), f"Expected [0], got {lagged_indices[0]}" + print("Edge cases test passed!") @@ -87,33 +87,36 @@ def test_lagged_indices_filtering(): total_lag_time = 3 lag_subsample_rate = 3 - print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " - f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") - - breakpoint() # Debug point 7: Check filtering parameters before function call - - lagged_indices = get_subsampled_indices( - N, subsample_rate, total_lag_time, lag_subsample_rate + print( + f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}" ) + breakpoint() # Debug point 7: Check filtering parameters before function call + + lagged_indices = get_subsampled_indices(N, subsample_rate, total_lag_time, lag_subsample_rate) + breakpoint() # Debug point 8: Check filtering results - + # Extract subsampled indices (first element of each list) subsampled_indices = np.array([indices[0] for indices in lagged_indices]) print(f"Subsampled indices: {subsampled_indices}") print(f"Number of lagged indices lists: {len(lagged_indices)}") - + expected_subsampled = np.array([5, 10, 15]) - assert np.array_equal(subsampled_indices, expected_subsampled), \ + assert np.array_equal(subsampled_indices, expected_subsampled), ( f"Expected {expected_subsampled}, got {subsampled_indices}" + ) - assert len(lagged_indices) == len(subsampled_indices), \ + assert len(lagged_indices) == len(subsampled_indices), ( f"Expected {len(subsampled_indices)} lagged indices lists, got {len(lagged_indices)}" + ) expected_first_lagged = np.array([5, 2, -1]) - assert not any(np.array_equal(lagged, expected_first_lagged) for lagged in lagged_indices), \ + assert not any(np.array_equal(lagged, expected_first_lagged) for lagged in lagged_indices), ( "Found unexpected lagged indices that should have been filtered out" - + ) + print("Lagged indices filtering test passed!") @@ -125,97 +128,104 @@ def test_large_numbers(): total_lag_time = 5 lag_subsample_rate = 10 - print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " - f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") - - breakpoint() # Debug point 9: Check large number parameters before function call - - lagged_indices = get_subsampled_indices( - N, subsample_rate, total_lag_time, lag_subsample_rate + print( + f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}" ) + breakpoint() # Debug point 9: Check large number parameters before function call + + lagged_indices = get_subsampled_indices(N, subsample_rate, total_lag_time, lag_subsample_rate) + breakpoint() # Debug point 10: Check large number results - + # Extract subsampled indices (first element of each list) subsampled_indices = np.array([indices[0] for indices in lagged_indices]) print(f"Number of subsampled indices: {len(subsampled_indices)}") print(f"Number of lagged indices lists: {len(lagged_indices)}") - - assert len(subsampled_indices) == N // subsample_rate, \ + + assert len(subsampled_indices) == N // subsample_rate, ( f"Expected {N // subsample_rate} subsampled indices, got {len(subsampled_indices)}" + ) for i, lagged in enumerate(lagged_indices): print(f"\nChecking lagged indices list {i}:") print(f"Lagged indices: {lagged}") print(f"Type of lagged indices: {type(lagged)}") - print(f"Individual values and their types:") + print("Individual values and their types:") for j, val in enumerate(lagged): print(f" Index {j}: value={val}, type={type(val)}") - - assert len(lagged) == total_lag_time, \ - f"Expected lagged indices length {total_lag_time}, got {len(lagged)}" - + + assert len(lagged) == total_lag_time, f"Expected lagged indices length {total_lag_time}, got {len(lagged)}" + # Check each value individually for j, val in enumerate(lagged): - assert isinstance(val, (int, np.integer)), \ + assert isinstance(val, (int, np.integer)), ( f"Value at index {j} is not an integer: {val} (type: {type(val)})" - assert val >= 0, \ - f"Found negative value at index {j}: {val}" - + ) + assert val >= 0, f"Found negative value at index {j}: {val}" + print("Large numbers test passed!") def test_trajectory_subsampling(): """Test subsampling of trajectory positions.""" print("\nTesting trajectory subsampling...") - + # Create a random trajectory with 100 frames, 10 particles, and 3 coordinates N = 100 np.random.seed(42) # For reproducibility positions = np.random.randn(N, 10, 3) - + subsample_rate = 10 total_lag_time = 3 lag_subsample_rate = 10 - print(f"Input parameters: N={N}, subsample_rate={subsample_rate}, " - f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}") - + print( + f"Input parameters: N={N}, subsample_rate={subsample_rate}, " + f"total_lag_time={total_lag_time}, lag_subsample_rate={lag_subsample_rate}" + ) + breakpoint() # Debug point 11: Check trajectory parameters - + subsampled_positions, lagged_positions = get_subsampled_trajectory( positions, subsample_rate, total_lag_time, lag_subsample_rate ) breakpoint() # Debug point 12: Check trajectory results - + print(f"Original positions shape: {positions.shape}") print(f"Subsampled positions shape: {subsampled_positions.shape}") print(f"Number of lagged position lists: {len(lagged_positions)}") - + # Check shapes expected_num_subsampled = (N - 20) // subsample_rate # Starting from index 20 - assert subsampled_positions.shape[0] == expected_num_subsampled, \ + assert subsampled_positions.shape[0] == expected_num_subsampled, ( f"Expected {expected_num_subsampled} subsampled positions, got {subsampled_positions.shape[0]}" - assert subsampled_positions.shape[1:] == (10, 3), \ + ) + assert subsampled_positions.shape[1:] == (10, 3), ( f"Expected subsampled positions to have shape (N, 10, 3), got {subsampled_positions.shape}" - + ) + # Check values for i in range(expected_num_subsampled): # Check subsampled position - expected_sub_pos = positions[20 + i*subsample_rate] - assert np.array_equal(subsampled_positions[i], expected_sub_pos), \ + expected_sub_pos = positions[20 + i * subsample_rate] + assert np.array_equal(subsampled_positions[i], expected_sub_pos), ( f"For index {i}, expected subsampled position {expected_sub_pos}, got {subsampled_positions[i]}" - + ) + # Check lagged positions - assert len(lagged_positions[i]) == total_lag_time, \ + assert len(lagged_positions[i]) == total_lag_time, ( f"For index {i}, expected {total_lag_time} lagged positions, got {len(lagged_positions[i])}" - + ) + for j, lag_pos in enumerate(lagged_positions[i]): - expected_lag_pos = positions[20 + i*subsample_rate - j*lag_subsample_rate] - assert np.array_equal(lag_pos, expected_lag_pos), \ + expected_lag_pos = positions[20 + i * subsample_rate - j * lag_subsample_rate] + assert np.array_equal(lag_pos, expected_lag_pos), ( f"For index {i}, lag {j}, expected position {expected_lag_pos}, got {lag_pos}" - + ) + print("Trajectory subsampling test passed!") @@ -226,4 +236,4 @@ def test_trajectory_subsampling(): # test_lagged_indices_filtering() # test_large_numbers() test_trajectory_subsampling() - print("\nAll tests completed!") \ No newline at end of file + print("\nAll tests completed!") diff --git a/src/jamun/model/__init__.py b/src/jamun/model/__init__.py index b237bc2..3369ea0 100644 --- a/src/jamun/model/__init__.py +++ b/src/jamun/model/__init__.py @@ -1,5 +1,5 @@ +from .conditioners import ConditionerSpiked from .denoiser import Denoiser -from .energy import EnergyModel from .denoiser_multimeasurement import DenoiserMultimeasurement from .denoiser_spiked import DenoiserSpiked -from .conditioners import ConditionerSpiked +from .energy import EnergyModel diff --git a/src/jamun/model/arch/__init__.py b/src/jamun/model/arch/__init__.py index 313896a..3a35d9d 100644 --- a/src/jamun/model/arch/__init__.py +++ b/src/jamun/model/arch/__init__.py @@ -1,4 +1,4 @@ from .e3conv import E3Conv -from .ophiuchus import Ophiuchus from .e3conv_conditional import E3ConvConditional -from .spatiotemporal import E3Transformer, E3SpatioTemporal +from .ophiuchus import Ophiuchus +from .spatiotemporal import E3SpatioTemporal, E3Transformer diff --git a/src/jamun/model/arch/e3conv_conditional.py b/src/jamun/model/arch/e3conv_conditional.py index 757df8a..a199712 100644 --- a/src/jamun/model/arch/e3conv_conditional.py +++ b/src/jamun/model/arch/e3conv_conditional.py @@ -1,15 +1,16 @@ -from typing import Callable +from collections.abc import Callable import e3nn +import e3tools.nn import torch import torch_geometric from e3nn import o3 from e3nn.o3 import Irreps from e3tools import scatter from torch import Tensor + from jamun.model.atom_embedding import AtomEmbeddingWithResidueInformation, SimpleAtomEmbedding from jamun.model.noise_conditioning import NoiseConditionalScaling, NoiseConditionalSkipConnection -import e3tools.nn class E3ConvConditional(torch.nn.Module): @@ -36,7 +37,7 @@ def __init__( num_residue_types: int = 25, test_equivariance: bool = False, reduce: str | None = None, - N_structures: int = 1 + N_structures: int = 1, ): super().__init__() @@ -75,7 +76,7 @@ def __init__( self.initial_projector = hidden_layer_factory( irreps_in=self.initial_noise_scaling.irreps_out, irreps_out=self.irreps_hidden, - irreps_sh=N_structures*self.irreps_sh, + irreps_sh=N_structures * self.irreps_sh, edge_attr_dim=edge_attr_dim, ) @@ -87,7 +88,7 @@ def __init__( hidden_layer_factory( irreps_in=self.irreps_hidden, irreps_out=self.irreps_hidden, - irreps_sh=N_structures*self.irreps_sh, + irreps_sh=N_structures * self.irreps_sh, edge_attr_dim=self.edge_attr_dim, ) ) @@ -100,7 +101,7 @@ def __init__( def forward( self, - pos: Tensor, # should be [batch_size*N, 3T], T is the number of previous time-steps + pos: Tensor, # should be [batch_size*N, 3T], T is the number of previous time-steps topology: torch_geometric.data.Batch, c_noise: Tensor, effective_radial_cutoff: float, @@ -109,13 +110,13 @@ def forward( edge_index = topology["edge_index"] bond_mask = topology["bond_mask"] - src, dst = edge_index # compute edge spherical harmonics over concat structures + src, dst = edge_index # compute edge spherical harmonics over concat structures positions = torch.split(pos, 3, dim=-1) edge_sh = [] - for block in positions: + for block in positions: edge_vec = block[src] - block[dst] edge_sh.append(self.sh(edge_vec)) - edge_sh = torch.cat(edge_sh, dim=-1) + edge_sh = torch.cat(edge_sh, dim=-1) # print(f"Edge spherical harmonics: {type(edge_sh)}") bonded_edge_attr = self.embed_bondedness(bond_mask) @@ -149,7 +150,7 @@ class E3ConvConditionalWithInputAttr(E3ConvConditional): Extension of E3ConvConditional that can accept additional input attributes and combine them with the computed node attributes. """ - + def __init__( self, irreps_out: str | Irreps, @@ -176,7 +177,7 @@ def __init__( ): """ Initialize E3ConvConditionalWithInputAttr. - + Args: input_attr_irreps: Irreps of the input attributes that will be combined with node_attr. If None, the model behaves like the parent class. @@ -204,14 +205,14 @@ def __init__( reduce=reduce, N_structures=N_structures, ) - + self.input_attr_irreps = o3.Irreps(input_attr_irreps) if input_attr_irreps is not None else None - + # Create input irrep aggregator if input attributes are provided if self.input_attr_irreps is not None: # Combined irreps: node_attr irreps + input_attr irreps combined_irreps = self.irreps_hidden + self.input_attr_irreps - + # Create aggregator that takes combined input and outputs node_attr irreps self.input_irrep_aggregator = e3tools.nn.EquivariantMLP( irreps_in=combined_irreps, @@ -220,7 +221,7 @@ def __init__( ) else: self.input_irrep_aggregator = None - + def forward( self, pos: Tensor, @@ -231,7 +232,7 @@ def forward( ) -> Tensor: """ Forward pass with optional input attributes. - + Args: pos: Node positions topology: Graph topology @@ -239,7 +240,7 @@ def forward( effective_radial_cutoff: Radial cutoff for edges input_attr: Optional input attributes to combine with node_attr. Should have shape [N, input_attr_irreps.dim] where N is number of nodes. - + Returns: Node attributes after processing """ @@ -247,13 +248,13 @@ def forward( edge_index = topology["edge_index"] bond_mask = topology["bond_mask"] - src, dst = edge_index # compute edge spherical harmonics over concat structures + src, dst = edge_index # compute edge spherical harmonics over concat structures positions = torch.split(pos, 3, dim=-1) edge_sh = [] - for block in positions: + for block in positions: edge_vec = block[src] - block[dst] edge_sh.append(self.sh(edge_vec)) - edge_sh = torch.cat(edge_sh, dim=-1) + edge_sh = torch.cat(edge_sh, dim=-1) # print(f"Edge spherical harmonics: {type(edge_sh)}") bonded_edge_attr = self.embed_bondedness(bond_mask) @@ -271,32 +272,28 @@ def forward( node_attr = self.atom_embedder(topology) node_attr = self.initial_noise_scaling(node_attr, c_noise) node_attr = self.initial_projector(node_attr, edge_index, edge_attr, edge_sh) - + # Combine with input attributes if provided if input_attr is not None and self.input_irrep_aggregator is not None: # Validate input_attr shape expected_dim = self.input_attr_irreps.dim if input_attr.shape[-1] != expected_dim: raise ValueError( - f"Expected input_attr to have dimension {expected_dim}, " - f"but got {input_attr.shape[-1]}" + f"Expected input_attr to have dimension {expected_dim}, but got {input_attr.shape[-1]}" ) if input_attr.shape[0] != node_attr.shape[0]: raise ValueError( - f"Expected input_attr to have {node_attr.shape[0]} nodes, " - f"but got {input_attr.shape[0]}" + f"Expected input_attr to have {node_attr.shape[0]} nodes, but got {input_attr.shape[0]}" ) - + # Concatenate node_attr with input_attr combined_attr = torch.cat([node_attr, input_attr], dim=-1) - + # Aggregate to get back to node_attr irreps node_attr = self.input_irrep_aggregator(combined_attr) elif input_attr is not None and self.input_irrep_aggregator is None: - raise ValueError( - "input_attr provided but input_attr_irreps was not specified during initialization" - ) - + raise ValueError("input_attr provided but input_attr_irreps was not specified during initialization") + # Continue with normal processing for scaling, skip, layer in zip(self.noise_scalings, self.skip_connections, self.layers): node_attr = skip(node_attr, layer(scaling(node_attr, c_noise), edge_index, edge_attr, edge_sh), c_noise) @@ -312,17 +309,17 @@ def forward( class E3ConvConditionalSpatioTemporal(E3ConvConditional): """ E3ConvConditional specifically designed for spatiotemporal conditioning. - + This class expects input positions to be concatenated as [y.pos, spatial_features] where y.pos are the physical 3D coordinates and spatial_features are additional attributes from the spatiotemporal model. - + Key differences from E3ConvConditional: - Edge spherical harmonics are only computed for the first 3 coordinates (y.pos) - Remaining coordinates are treated as per-node input attributes - Input attributes are combined with computed node attributes """ - + def __init__( self, irreps_out: str | Irreps, @@ -349,7 +346,7 @@ def __init__( ): """ Initialize E3ConvConditionalSpatioTemporal. - + Args: input_attr_irreps: Irreps of the spatial features from spatiotemporal model. Should match the irreps_out of the spatiotemporal model. @@ -378,21 +375,21 @@ def __init__( reduce=reduce, N_structures=N_structures, ) - + # Set up input attribute handling self.input_attr_irreps = o3.Irreps(input_attr_irreps) self.input_attr_irreps_dim = self.input_attr_irreps.dim # Create input irrep aggregator to combine node_attr with input_attr # Combined irreps: node_attr irreps + input_attr irreps combined_irreps = self.irreps_hidden + self.input_attr_irreps - + # Create aggregator that takes combined input and outputs node_attr irreps self.input_irrep_aggregator = e3tools.nn.EquivariantMLP( irreps_in=combined_irreps, irreps_out=self.irreps_hidden, irreps_hidden_list=[self.irreps_hidden], # Single hidden layer ) - + def forward( self, pos: Tensor, # should be [N, 3 + spatial_features_dim] from [y.pos, spatial_features] @@ -402,13 +399,13 @@ def forward( ) -> Tensor: """ Forward pass with spatiotemporal conditioning. - + Args: pos: Concatenated positions [y.pos, spatial_features] with shape [N, 3 + spatial_features_dim] topology: Graph topology c_noise: Noise conditioning effective_radial_cutoff: Radial cutoff for edges - + Returns: Node attributes after processing """ @@ -417,11 +414,11 @@ def forward( bond_mask = topology["bond_mask"] src, dst = edge_index - + # Split positions: first 3 coords are physical positions, rest are spatial features pos_physical = pos[:, :3] # [N, 3] - physical coordinates pos_features = pos[:, 3:] # [N, spatial_features_dim] - spatial features - + # Compute edge spherical harmonics ONLY for physical positions edge_vec_physical = pos_physical[src] - pos_physical[dst] edge_sh = self.sh(edge_vec_physical) @@ -442,22 +439,21 @@ def forward( node_attr = self.atom_embedder(topology) node_attr = self.initial_noise_scaling(node_attr, c_noise) node_attr = self.initial_projector(node_attr, edge_index, edge_attr, edge_sh) - + # Combine node_attr with spatial features (input_attr) # Validate spatial features shape expected_dim = self.input_attr_irreps_dim if pos_features.shape[-1] != expected_dim: raise ValueError( - f"Expected spatial features to have dimension {expected_dim}, " - f"but got {pos_features.shape[-1]}" + f"Expected spatial features to have dimension {expected_dim}, but got {pos_features.shape[-1]}" ) - + # Concatenate node_attr with spatial features combined_attr = torch.cat([node_attr, pos_features], dim=-1) - + # Aggregate to get back to node_attr irreps node_attr = self.input_irrep_aggregator(combined_attr) - + # Continue with normal processing using only physical positions for edge computations for scaling, skip, layer in zip(self.noise_scalings, self.skip_connections, self.layers): node_attr = skip(node_attr, layer(scaling(node_attr, c_noise), edge_index, edge_attr, edge_sh), c_noise) diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index 1904fdf..991f2d7 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -7,34 +7,30 @@ - Spatial-temporal graph conversion utilities """ -from typing import Dict, Union, Optional - import e3nn +import e3tools +import e3tools.nn import torch import torch.nn as nn -from e3nn import o3 import torch_geometric import torch_geometric.data -import e3tools -import e3tools.nn -import logging -from jamun.model.arch.e3conv import E3Conv +from e3nn import o3 def calculate_temporal_positions(temporal_length, mode="linear", device=None): """ Calculate normalized temporal positions for nodes in a temporal graph. - + Args: temporal_length: Total number of nodes in the temporal sequence device: Device to create tensors on - + Returns: torch.Tensor: Normalized positions [0, 1/T, 2/T, ..., (T-1)/T] """ if temporal_length <= 1: return torch.tensor([0.0], device=device) - + if mode == "linear": # Create positions [0, 1, 2, ..., T-1] and normalize by T positions = torch.arange(temporal_length, dtype=torch.float32, device=device) @@ -43,19 +39,19 @@ def calculate_temporal_positions(temporal_length, mode="linear", device=None): # Create positions [0, 1, 2, ..., T-1] and normalize by T positions = torch.arange(temporal_length, dtype=torch.float32, device=device) positions = torch.zeros_like(positions) - + return normalized_positions def spatial_to_temporal_graphs(batch, graph_type="fan"): """ Convert a batch of spatial graphs to temporal graphs with configurable connectivity. - + For each spatial node with position + hidden states, create a temporal graph where: - Node 0: current position - Nodes 1-T: hidden state positions - Connectivity depends on graph_type parameter - + Args: batch: Input spatial graph batch graph_type: Type of connectivity to use @@ -65,56 +61,51 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): - "complete_no_self": Complete graph without self-loops (all-to-all excluding self) """ import torch_geometric - + # Validate graph_type valid_types = ["fan", "hub_n_spoke", "complete", "complete_no_self"] if graph_type not in valid_types: raise ValueError(f"graph_type must be one of {valid_types}, got {graph_type}") - + # Get device from input batch device = batch.pos.device - + # Get dimensions num_spatial_nodes = batch.pos.shape[0] - + # Check if we have hidden states - if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + if hasattr(batch, "hidden_state") and batch.hidden_state is not None and len(batch.hidden_state) > 0: num_hidden_states = len(batch.hidden_state) temporal_length = 1 + num_hidden_states # current + hidden else: # If no hidden states, just use current position num_hidden_states = 0 temporal_length = 1 - + # print(f"Creating {graph_type} temporal graphs: {num_spatial_nodes} spatial nodes -> {num_spatial_nodes} temporal graphs of length {temporal_length}") - + # Store reference to spatial graph spatial_graph = batch.clone() - + # Set connectivity type code for tracking - connectivity_type_map = { - "fan": 0, - "hub_n_spoke": 1, - "complete": 2, - "complete_no_self": 3 - } - + connectivity_type_map = {"fan": 0, "hub_n_spoke": 1, "complete": 2, "complete_no_self": 3} + temporal_graphs = [] - + for node_idx in range(num_spatial_nodes): # Build temporal positions: [current_pos, hidden_1, hidden_2, ...] temporal_positions = [batch.pos[node_idx]] # Start with current position - + # Add hidden state positions if num_hidden_states > 0: for hidden_pos in batch.hidden_state: temporal_positions.append(hidden_pos[node_idx]) - + temporal_pos = torch.stack(temporal_positions) # Shape: [T, 3] - + # Calculate temporal positions for this sequence temporal_position = calculate_temporal_positions(temporal_length, device=device) - + # Create edge connectivity based on graph_type if temporal_length > 1: if graph_type == "fan": @@ -122,37 +113,37 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): # Hub connections: 0->1, 0->2, 0->3, ..., 0->T-1 hub_src = [0] * (temporal_length - 1) hub_dst = list(range(1, temporal_length)) - + # Sequential connections: 1->2, 2->3, ..., (T-2)->(T-1) seq_src = list(range(1, temporal_length - 1)) seq_dst = list(range(2, temporal_length)) - + # Combine all edges all_src = hub_src + seq_src all_dst = hub_dst + seq_dst - + edge_index = torch.tensor([all_src, all_dst], dtype=torch.long, device=device) - + elif graph_type == "hub_n_spoke": # Hub-and-spoke only: 0 connects to all others, no sequential hub_src = [0] * (temporal_length - 1) hub_dst = list(range(1, temporal_length)) - + edge_index = torch.tensor([hub_src, hub_dst], dtype=torch.long, device=device) - + elif graph_type == "complete": # Complete graph without self-loops: all-to-all excluding self src_nodes = [] dst_nodes = [] - + for i in range(temporal_length): for j in range(temporal_length): if i != j: # Exclude self-loops src_nodes.append(i) dst_nodes.append(j) - + edge_index = torch.tensor([src_nodes, dst_nodes], dtype=torch.long, device=device) - + else: # Single node case if graph_type == "complete": @@ -161,7 +152,7 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): else: # Single node, no edges for other types edge_index = torch.tensor([[], []], dtype=torch.long, device=device) - + # Create temporal graph for this spatial node temporal_graph = torch_geometric.data.Data( pos=temporal_pos, @@ -173,15 +164,15 @@ def spatial_to_temporal_graphs(batch, graph_type="fan"): # Note: Removed graph_type string to avoid batching issues with PyTorch Geometric ) temporal_graphs.append(temporal_graph) - + # Batch all temporal graphs temporal_batch = torch_geometric.data.Batch.from_data_list(temporal_graphs) - + # Store spatial graph reference temporal_batch.spatial_graph = spatial_graph # Note: Removed graph_type string to avoid batching issues with PyTorch Geometric # Graph type can be inferred from connectivity_type tensor attribute - + return temporal_batch @@ -190,27 +181,27 @@ def temporal_to_spatial_graphs(temporal_batch): Convert temporal graphs back to spatial graphs. Take the 0th node position from each temporal graph as the updated spatial position. """ - # Get the spatial graph template + # Get the spatial graph template spatial_graph = temporal_batch.spatial_graph.clone() - + # Extract 0th node positions from each temporal graph num_temporal_graphs = temporal_batch.num_graphs updated_positions = [] - + # Iterate through each temporal graph in the batch for graph_idx in range(num_temporal_graphs): # Get the node range for this temporal graph start_idx = temporal_batch.ptr[graph_idx] - + # The 0th node of each temporal graph is at the start of its range updated_positions.append(temporal_batch.pos[start_idx]) - + # Stack to create new position tensor updated_positions = torch.stack(updated_positions) - + # Update spatial graph with new positions spatial_graph.pos = updated_positions - + return spatial_graph @@ -219,16 +210,16 @@ class E3Transformer(nn.Module): def __init__( self, - irreps_out: Union[str, e3nn.o3.Irreps], - irreps_hidden: Union[str, e3nn.o3.Irreps], - irreps_sh: Union[str, e3nn.o3.Irreps], - irreps_node_attr: Union[str, e3nn.o3.Irreps], + irreps_out: str | e3nn.o3.Irreps, + irreps_hidden: str | e3nn.o3.Irreps, + irreps_sh: str | e3nn.o3.Irreps, + irreps_node_attr: str | e3nn.o3.Irreps, num_layers: int, edge_attr_dim: int, num_attention_heads: int, reduce: str | None = None, - conv = e3tools.nn.Conv, - irreps_node_attr_temporal: Union[str, e3nn.o3.Irreps] = "1x1e", + conv=e3tools.nn.Conv, + irreps_node_attr_temporal: str | e3nn.o3.Irreps = "1x1e", radial_edge_attr_encoding_function: str = "gaussian", node_attr_temporal_encoding_function: str = "gaussian", edge_attr_temporal_encoding_function: str = "gaussian", @@ -238,15 +229,13 @@ def __init__( self.irreps_out = o3.Irreps(irreps_out) self.irreps_hidden = o3.Irreps(irreps_hidden) self.irreps_sh = o3.Irreps(irreps_sh) - self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps + self.irreps_node_attr = o3.Irreps(irreps_node_attr) # input irreps self.irreps_node_attr_temporal = o3.Irreps(irreps_node_attr_temporal) self.num_layers = num_layers self.edge_attr_dim = edge_attr_dim self.num_attention_heads = num_attention_heads self.reduce = reduce - self.sh = o3.SphericalHarmonics( - irreps_out=self.irreps_sh, normalize=True, normalization="component" - ) + self.sh = o3.SphericalHarmonics(irreps_out=self.irreps_sh, normalize=True, normalization="component") # Split edge attribute dimensions: radial and temporal (bondedness is optional) self.radial_edge_attr_dim = self.edge_attr_dim // 2 self.temporal_edge_attr_dim = self.edge_attr_dim - self.radial_edge_attr_dim @@ -260,9 +249,11 @@ def __init__( # Input: node_attr (from data) + temporal_position (1x0e scalar) # irreps_with_temporal = self.irreps_node_attr + o3.Irreps("1x0e") irreps_with_temporal = self.irreps_node_attr + self.irreps_node_attr_temporal - self.temporal_gate = e3tools.nn.GateWrapper(irreps_in=irreps_with_temporal, \ - irreps_out=self.irreps_hidden, \ - irreps_gate=irreps_with_temporal,) + self.temporal_gate = e3tools.nn.GateWrapper( + irreps_in=irreps_with_temporal, + irreps_out=self.irreps_hidden, + irreps_gate=irreps_with_temporal, + ) self.layers = nn.ModuleList() self.conv = conv @@ -322,7 +313,7 @@ def forward( cutoff=True, ) radial_edge_attr = torch.ones_like(radial_edge_attr) - + # Temporal edge attributes from temporal_position differences temporal_edge_vec = temporal_position[src] - temporal_position[dst] if self.edge_attr_temporal_encoding_function != "ones": @@ -345,10 +336,10 @@ def forward( ) temporal_edge_attr = torch.ones_like(temporal_edge_attr) - # temporal_edge_attr = torch.ones_like(temporal_edge_attr) # TODO: remove this, this is hacking. + # temporal_edge_attr = torch.ones_like(temporal_edge_attr) # TODO: remove this, this is hacking. # Optional bondedness (if bond_mask exists in the temporal graph) - if hasattr(temporal_graph, 'bond_mask') and temporal_graph.bond_mask is not None: + if hasattr(temporal_graph, "bond_mask") and temporal_graph.bond_mask is not None: bonded_edge_attr = self.embed_bondedness(temporal_graph.bond_mask) edge_attr = torch.cat((bonded_edge_attr, radial_edge_attr, temporal_edge_attr), dim=-1) else: @@ -360,8 +351,8 @@ def forward( if self.node_attr_temporal_encoding_function != "ones": temporal_position = e3nn.math.soft_one_hot_linspace( temporal_position, # Use absolute difference - 0.0, # time always starts at 0 - 1.0, # time always ends at 1 + 0.0, # time always starts at 0 + 1.0, # time always ends at 1 self.temporal_node_attr_dim, basis=self.node_attr_temporal_encoding_function, cutoff=True, @@ -369,8 +360,8 @@ def forward( else: temporal_position = e3nn.math.soft_one_hot_linspace( temporal_position, # Use absolute difference - 0.0, # time always starts at 0 - 1.0, # time always ends at 1 + 0.0, # time always starts at 0 + 1.0, # time always ends at 1 self.temporal_node_attr_dim, basis="gaussian", cutoff=True, @@ -378,7 +369,7 @@ def forward( temporal_position = torch.ones_like(temporal_position) temporal_position_expanded = temporal_position # [N, 1] for concatenation node_attr_with_temporal = torch.cat([node_attr, temporal_position_expanded], dim=-1) - + # Apply temporal gate node_attr_processed = self.temporal_gate(node_attr_with_temporal) @@ -396,22 +387,22 @@ def forward( dim_size=num_graphs, reduce=self.reduce, ) - + return node_attr_processed class E3SpatioTemporal(nn.Module): """ E(3)-equivariant spatio-temporal model that combines spatial and temporal processing. - + This model implements the complete workflow: 1. Process input spatial graph and hidden states through spatial module 2. Pool spatial features to temporal graph representation - 3. Process temporal graph through temporal module + 3. Process temporal graph through temporal module 4. Pool temporal features back to spatial representation 5. Convert temporal graph back to spatial graph """ - + def __init__( self, spatial_module: nn.Module, @@ -424,7 +415,7 @@ def __init__( ): """ Initialize the E3SpatioTemporal model. - + Args: spatial_module: Module for processing spatial positions (e.g., E3Conv) temporal_module: Module for processing temporal graphs (e.g., E3Transformer) @@ -434,7 +425,7 @@ def __init__( temporal_cutoff: Cutoff for temporal edge weights """ super().__init__() - + self.spatial_module = spatial_module self.temporal_module = temporal_module self.spatial_to_temporal_pooler = spatial_to_temporal_pooler @@ -442,23 +433,23 @@ def __init__( self.radial_cutoff = radial_cutoff self.temporal_cutoff = temporal_cutoff self.graph_type = graph_type - + def forward( self, batch: torch_geometric.data.Batch, c_noise: torch.Tensor, return_temporal_features: bool = False, return_temporal_graph: bool = False, - ) -> Union[torch.Tensor, Dict[str, torch.Tensor]]: + ) -> torch.Tensor | dict[str, torch.Tensor]: """ Forward pass implementing the complete spatio-temporal workflow. - + Args: batch: Input spatial graph batch with pos, batch, num_graphs, and optionally hidden_state c_noise: Noise conditioning tensor return_temporal_features: Whether to return intermediate temporal features return_temporal_graph: Whether to return the temporal graph - + Returns: If return_temporal_features or return_temporal_graph is True, returns dict with: - 'spatial_features': Final spatial features @@ -470,39 +461,39 @@ def forward( """ # Store original device device = batch.pos.device - + # Step 1: Convert spatial graph to temporal graphs - if hasattr(self, 'graph_type') and self.graph_type is not None: + if hasattr(self, "graph_type") and self.graph_type is not None: temporal_batch = spatial_to_temporal_graphs(batch, graph_type=self.graph_type) else: - temporal_batch = spatial_to_temporal_graphs(batch) # default to fan graph type - + temporal_batch = spatial_to_temporal_graphs(batch) # default to fan graph type + # Step 2: Process all positions (current + hidden states) with spatial module # Create topology for spatial processing (without positions) topology = batch.clone() # Remove position-dependent attributes but keep graph structure - if hasattr(topology, 'pos'): + if hasattr(topology, "pos"): del topology.pos - if hasattr(topology, 'batch'): - del topology.batch - if hasattr(topology, 'num_graphs'): + if hasattr(topology, "batch"): + del topology.batch + if hasattr(topology, "num_graphs"): del topology.num_graphs - + node_attr_list = [] - + # Process current positions node_attr_current = self.spatial_module( - pos=batch.pos, - topology=topology, + pos=batch.pos, + topology=topology, batch=batch.batch, num_graphs=batch.num_graphs, c_noise=c_noise, - effective_radial_cutoff=self.radial_cutoff + effective_radial_cutoff=self.radial_cutoff, ).unsqueeze(1) # [N, 1, features] node_attr_list.append(node_attr_current) - + # Process hidden state positions if they exist - if hasattr(batch, 'hidden_state') and batch.hidden_state is not None and len(batch.hidden_state) > 0: + if hasattr(batch, "hidden_state") and batch.hidden_state is not None and len(batch.hidden_state) > 0: for hidden_pos in batch.hidden_state: node_attr_hidden = self.spatial_module( pos=hidden_pos, @@ -510,55 +501,52 @@ def forward( batch=batch.batch, num_graphs=batch.num_graphs, c_noise=c_noise, - effective_radial_cutoff=self.radial_cutoff + effective_radial_cutoff=self.radial_cutoff, ).unsqueeze(1) # [N, 1, features] node_attr_list.append(node_attr_hidden) - + # Step 3: Stack spatial-temporal features node_attr_spatial_temporal = torch.cat(node_attr_list, dim=1) # [N, T, features] - + # Step 4: Convert spatial-temporal features to temporal node attributes temporal_node_attr = self.spatial_to_temporal_pooler(node_attr_spatial_temporal, temporal_batch) - + # Step 5: Process temporal graph through temporal module temporal_output = self.temporal_module( - temporal_node_attr, - temporal_batch, - self.radial_cutoff, - self.temporal_cutoff + temporal_node_attr, temporal_batch, self.radial_cutoff, self.temporal_cutoff ) - + # Step 6: Pool temporal features back to spatial features spatial_features = self.temporal_to_spatial_pooler(temporal_output, temporal_batch) - + # Step 7: Convert temporal graph back to spatial graph # output_spatial_graph = temporal_to_spatial_graphs(temporal_batch) - output_spatial_graph = batch + output_spatial_graph = batch # Prepare return values if return_temporal_features or return_temporal_graph: result = { - 'spatial_features': spatial_features, - 'spatial_graph': output_spatial_graph, + "spatial_features": spatial_features, + "spatial_graph": output_spatial_graph, } if return_temporal_features: - result['temporal_features'] = temporal_output + result["temporal_features"] = temporal_output if return_temporal_graph: - result['temporal_graph'] = temporal_batch + result["temporal_graph"] = temporal_batch return result else: return spatial_features - + def get_spatial_output_irreps(self): """Get the irreps of the spatial module output.""" - if hasattr(self.spatial_module, 'irreps_out'): + if hasattr(self.spatial_module, "irreps_out"): return self.spatial_module.irreps_out else: raise AttributeError("Spatial module does not have irreps_out attribute") - + def get_temporal_output_irreps(self): - """Get the irreps of the temporal module output.""" - if hasattr(self.temporal_module, 'irreps_out'): + """Get the irreps of the temporal module output.""" + if hasattr(self.temporal_module, "irreps_out"): return self.temporal_module.irreps_out else: - raise AttributeError("Temporal module does not have irreps_out attribute") \ No newline at end of file + raise AttributeError("Temporal module does not have irreps_out attribute") diff --git a/src/jamun/model/conditioner_usage_example.py b/src/jamun/model/conditioner_usage_example.py index d00387b..ab97a09 100644 --- a/src/jamun/model/conditioner_usage_example.py +++ b/src/jamun/model/conditioner_usage_example.py @@ -6,26 +6,26 @@ import functools import os -import torch -import numpy as np from pathlib import Path +import torch + import jamun +import jamun.data +import jamun.distributions from jamun.model import DenoiserSpiked -from jamun.model.conditioners import ConditionerSpiked from jamun.model.arch import E3ConvConditional -import jamun.distributions -import jamun.data +from jamun.model.conditioners import ConditionerSpiked def get_ala_ala_data(num_frames=20, total_lag_time=5): """ Load ALA_ALA data with specified parameters. - + Args: num_frames: Number of frames to load per dataset total_lag_time: Number of hidden states (total time lag) - + Returns: List of datasets """ @@ -33,54 +33,52 @@ def get_ala_ala_data(num_frames=20, total_lag_time=5): data_path = os.getenv("JAMUN_DATA_PATH") if data_path is None: # Try common locations - possible_paths = [ - "/data/bucket/kleinhej/", - "/data2/sules/", - "/path/to/data/" - ] + possible_paths = ["/data/bucket/kleinhej/", "/data2/sules/", "/path/to/data/"] for path in possible_paths: ala_path = Path(path) / "capped_diamines/timewarp_splits/train" if ala_path.exists(): data_path = path break - + if data_path is None: - raise ValueError("JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable.") - + raise ValueError( + "JAMUN_DATA_PATH not set and cannot find data. Please set JAMUN_DATA_PATH environment variable." + ) + print(f"Using data path: {data_path}") root_path = f"{data_path}/capped_diamines/timewarp_splits/train" - + datasets = jamun.data.parse_datasets_from_directory( root=root_path, traj_pattern="^(.*).xtc", pdb_pattern="^(.*).pdb", - filter_codes=['ALA_ALA'], + filter_codes=["ALA_ALA"], as_iterable=False, subsample=1, total_lag_time=total_lag_time, lag_subsample_rate=1, num_frames=num_frames, - max_datasets=1 # Just use one dataset for testing + max_datasets=1, # Just use one dataset for testing ) - + return datasets def create_test_denoiser_spiked(total_lag_time=5): """ Create a simple DenoiserSpiked model for testing. - + Args: total_lag_time: Number of structures for conditioning - + Returns: DenoiserSpiked model """ import e3tools.nn - + # Note: The actual data has 4 hidden states, so we'll have 4 + 1 clean = 5 structures actual_n_structures = 5 # 4 hidden states + 1 clean structure - + arch = functools.partial( E3ConvConditional, irreps_out="1x1e", @@ -99,14 +97,11 @@ def create_test_denoiser_spiked(total_lag_time=5): e3tools.nn.ConvBlock, conv=e3tools.nn.Conv, ), - output_head_factory=functools.partial( - e3tools.nn.EquivariantMLP, - irreps_hidden_list=["32x0e + 8x1e"] - ), + output_head_factory=functools.partial(e3tools.nn.EquivariantMLP, irreps_hidden_list=["32x0e + 8x1e"]), ) - + conditioner = ConditionerSpiked(N_structures=actual_n_structures) - + denoiser = DenoiserSpiked( arch=arch, optim=functools.partial(torch.optim.Adam, lr=1e-3), @@ -121,7 +116,7 @@ def create_test_denoiser_spiked(total_lag_time=5): mirror_augmentation_rate=0.0, conditioner=conditioner, ) - + return denoiser @@ -132,114 +127,118 @@ def test_noise_and_denoise(): print("=" * 60) print("Testing DenoiserSpiked with ConditionerSpiked on ALA_ALA data") print("=" * 60) - + # Load data try: total_lag_time = 5 datasets = get_ala_ala_data(num_frames=10, total_lag_time=total_lag_time) print(f"āœ… Successfully loaded {len(datasets)} datasets") - + dataset = datasets[0] print(f" Dataset label: {dataset.label()}") print(f" Dataset length: {len(dataset)}") - + # Get a sample sample = dataset[0] print(f" Sample positions shape: {sample.pos.shape}") - print(f" Sample hidden states: {len(sample.hidden_state) if hasattr(sample, 'hidden_state') and sample.hidden_state else 0}") - if hasattr(sample, 'hidden_state') and sample.hidden_state: + print( + f" Sample hidden states: {len(sample.hidden_state) if hasattr(sample, 'hidden_state') and sample.hidden_state else 0}" + ) + if hasattr(sample, "hidden_state") and sample.hidden_state: for i, h in enumerate(sample.hidden_state): print(f" Hidden state {i} shape: {h.shape}") - + except Exception as e: print(f"āŒ Failed to load data: {e}") return False - + # Create model try: denoiser = create_test_denoiser_spiked(total_lag_time=total_lag_time) - print(f"āœ… Successfully created DenoiserSpiked model") + print("āœ… Successfully created DenoiserSpiked model") print(f" Conditioner: {type(denoiser.conditioning_module).__name__}") - + except Exception as e: print(f"āŒ Failed to create model: {e}") return False - + # Test noise_and_denoise try: print("\n" + "-" * 40) print("Testing noise_and_denoise method...") print("-" * 40) - + # Convert to batch for testing import torch_geometric.data + batch = torch_geometric.data.Batch.from_data_list([sample]) print(f" Batch positions shape: {batch.pos.shape}") print(f" Batch num_graphs: {batch.num_graphs}") - + # Set model to eval mode denoiser.eval() - + # Test with different sigma values sigma_values = [0.01, 0.04, 0.1] - + for sigma in sigma_values: print(f"\n Testing with sigma = {sigma}") - + # Run noise_and_denoise with torch.no_grad(): - x_target, xhat, y_noisy = denoiser.noise_and_denoise( - batch, - sigma=sigma, - align_noisy_input=True - ) - - print(f" āœ… noise_and_denoise completed successfully") + x_target, xhat, y_noisy = denoiser.noise_and_denoise(batch, sigma=sigma, align_noisy_input=True) + + print(" āœ… noise_and_denoise completed successfully") print(f" Target shape: {x_target.pos.shape}") print(f" Prediction shape: {xhat.pos.shape}") print(f" Noisy input shape: {y_noisy.pos.shape}") - + # Check that shapes match assert x_target.pos.shape == xhat.pos.shape == y_noisy.pos.shape - + # Test conditioning - print(f" Testing conditioner...") - print(f" y_noisy has hidden_state: {hasattr(y_noisy, 'hidden_state') and y_noisy.hidden_state is not None}") - if hasattr(y_noisy, 'hidden_state') and y_noisy.hidden_state is not None: + print(" Testing conditioner...") + print( + f" y_noisy has hidden_state: {hasattr(y_noisy, 'hidden_state') and y_noisy.hidden_state is not None}" + ) + if hasattr(y_noisy, "hidden_state") and y_noisy.hidden_state is not None: print(f" y_noisy hidden_state count: {len(y_noisy.hidden_state)}") print(f" x_target is not None: {x_target is not None}") if x_target is not None: print(f" x_target.pos shape: {x_target.pos.shape}") - print(f" x_target has hidden_state: {hasattr(x_target, 'hidden_state') and x_target.hidden_state is not None}") - + print( + f" x_target has hidden_state: {hasattr(x_target, 'hidden_state') and x_target.hidden_state is not None}" + ) + conditioned_structures = denoiser.conditioner(y_noisy, x_target) print(f" Conditioned structures count: {len(conditioned_structures)}") - + for i, struct in enumerate(conditioned_structures): print(f" Structure {i} shape: {struct.shape}") - + # Verify that the last structure is the clean structure if len(conditioned_structures) > 0: last_structure = conditioned_structures[-1] clean_structure = x_target.pos if torch.allclose(last_structure, clean_structure, atol=1e-6): - print(f" āœ… Last conditioned structure matches x_clean.pos") + print(" āœ… Last conditioned structure matches x_clean.pos") else: - print(f" āŒ Last conditioned structure does NOT match x_clean.pos") + print(" āŒ Last conditioned structure does NOT match x_clean.pos") print(f" Max difference: {torch.max(torch.abs(last_structure - clean_structure)).item():.8f}") - + # Calculate some basic metrics noise_level = torch.mean(torch.norm(y_noisy.pos - x_target.pos, dim=-1)) prediction_error = torch.mean(torch.norm(xhat.pos - x_target.pos, dim=-1)) print(f" Average noise level: {noise_level:.4f}") print(f" Average prediction error: {prediction_error:.4f}") - - print(f"\nāœ… All noise_and_denoise tests passed!") + + print("\nāœ… All noise_and_denoise tests passed!") return True - + except Exception as e: print(f"āŒ Failed during noise_and_denoise testing: {e}") import traceback + traceback.print_exc() return False @@ -251,52 +250,53 @@ def test_conditioning_shapes(): print("\n" + "=" * 60) print("Testing ConditionerSpiked shape outputs") print("=" * 60) - + # Create dummy data N_atoms = 22 # ALA_ALA has 22 atoms N_structures = 5 - + # Create fake batch pos = torch.randn(N_atoms, 3) hidden_states = [torch.randn(N_atoms, 3) for _ in range(N_structures - 2)] # -2 for current pos and clean pos - + # Create fake torch_geometric batch import torch_geometric.data + y = torch_geometric.data.Data(pos=pos, hidden_state=hidden_states) x_clean = torch_geometric.data.Data(pos=torch.randn(N_atoms, 3)) - + # Test conditioner conditioner = ConditionerSpiked(N_structures=N_structures) conditioned_structures = conditioner.forward(y, x_clean) - - print(f"Input shapes:") + + print("Input shapes:") print(f" y.pos: {y.pos.shape}") print(f" y.hidden_state: {[h.shape for h in y.hidden_state]}") print(f" x_clean.pos: {x_clean.pos.shape}") - - print(f"\nConditioned structures:") + + print("\nConditioned structures:") for i, struct in enumerate(conditioned_structures): print(f" Structure {i}: {struct.shape}") - + # Test concatenation (like in the model) concatenated = torch.cat(conditioned_structures, dim=-1) print(f"\nConcatenated shape: {concatenated.shape}") expected_dim = len(conditioned_structures) * 3 # Each structure has 3D coordinates print(f"Expected last dimension: {expected_dim}") - + # Verify that the last structure is the clean structure if len(conditioned_structures) > 0: last_structure = conditioned_structures[-1] clean_structure = x_clean.pos if torch.allclose(last_structure, clean_structure, atol=1e-6): - print(f"āœ… Last conditioned structure matches x_clean.pos") + print("āœ… Last conditioned structure matches x_clean.pos") else: - print(f"āŒ Last conditioned structure does NOT match x_clean.pos") + print("āŒ Last conditioned structure does NOT match x_clean.pos") print(f" Max difference: {torch.max(torch.abs(last_structure - clean_structure)).item():.8f}") assert False, "Last conditioned structure should match x_clean.pos" - + assert concatenated.shape == (N_atoms, expected_dim) - print(f"āœ… Shape test passed!") + print("āœ… Shape test passed!") if __name__ == "__main__": @@ -304,10 +304,10 @@ def test_conditioning_shapes(): try: # Test data loading and noise_and_denoise success = test_noise_and_denoise() - + # Test conditioning shapes test_conditioning_shapes() - + if success: print("\n" + "=" * 60) print("šŸŽ‰ ALL TESTS PASSED! šŸŽ‰") @@ -317,10 +317,11 @@ def test_conditioning_shapes(): print("\n" + "=" * 60) print("āŒ SOME TESTS FAILED") print("=" * 60) - + except KeyboardInterrupt: print("\nāš ļø Tests interrupted by user") except Exception as e: print(f"\nāŒ Unexpected error: {e}") import traceback - traceback.print_exc() \ No newline at end of file + + traceback.print_exc() diff --git a/src/jamun/model/conditioners/__init__.py b/src/jamun/model/conditioners/__init__.py index c43d2ab..1d5ecb2 100644 --- a/src/jamun/model/conditioners/__init__.py +++ b/src/jamun/model/conditioners/__init__.py @@ -1,2 +1,8 @@ -from .conditioners import Conditioner, PositionConditioner, SelfConditioner, MeanConditioner, DenoisedConditioner, ConditionerSpiked - +from .conditioners import ( + Conditioner, + ConditionerSpiked, + DenoisedConditioner, + MeanConditioner, + PositionConditioner, + SelfConditioner, +) diff --git a/src/jamun/model/conditioners/conditioners.py b/src/jamun/model/conditioners/conditioners.py index 8879c12..e6ebc82 100644 --- a/src/jamun/model/conditioners/conditioners.py +++ b/src/jamun/model/conditioners/conditioners.py @@ -1,39 +1,43 @@ import logging -from typing import Callable, Dict, Tuple, Union +import e3nn import lightning.pytorch as pl -import numpy as np import torch import torch_geometric -import e3nn -from e3tools import radius_graph + from jamun.model.denoiser_conditional import Denoiser + # Fix e3nn optimization for avoiding script issues e3nn.set_optimization_defaults(jit_script_fx=False) -from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils import mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm from jamun.utils.checkpoint import find_checkpoint + class Conditioner(pl.LightningModule): """ Base class for conditioners. """ + def __init__(self, N_structures: int, **kwargs): super().__init__() self.N_structures = N_structures + class PositionConditioner(pl.LightningModule): """ Condition the hidden state on the position of the structure. """ + def __init__(self, N_structures: int, align_hidden_states: bool = True, **kwargs): super().__init__() self.N_structures = N_structures self.align_hidden_states = align_hidden_states + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures = [y.pos] # Start with current position - for positions in y.hidden_state: + for positions in y.hidden_state: if self.align_hidden_states: aligned_positions = kabsch_algorithm(positions, y.pos, y.batch, y.num_graphs) conditioned_structures.append(aligned_positions) @@ -41,34 +45,39 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures.append(positions) return conditioned_structures + class SelfConditioner(pl.LightningModule): """ No conditioning, but add the position of the structure to itself to make it compatible with the denoiser. """ + def __init__(self, N_structures: int, **kwargs): super().__init__() self.N_structures = N_structures + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: conditioned_structures = [y.pos for _ in range(self.N_structures)] # Include current position return conditioned_structures + class MeanConditioner(pl.LightningModule): """ Condition on the mean across time steps of positions and hidden states. For each atom and coordinate, averages across all T+1 structures (current + hidden states). """ + def __init__(self, N_structures: int, **kwargs): super().__init__() self.N_structures = N_structures - + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: # Start with current position all_positions = [y.pos] - + # Add all hidden states if they exist if hasattr(y, "hidden_state") and y.hidden_state is not None: all_positions.extend(y.hidden_state) - + # Stack all positions along a new dimension and compute mean across time steps # Shape: (T+1, N, 3) -> (N, 3) where T is number of hidden states stacked_positions = torch.stack(all_positions, dim=0) # (T+1, N, 3) @@ -83,16 +92,18 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: # Return the mean repeated N_structures times conditioned_structures = [aligned_mean_positions for _ in range(self.N_structures)] - + return conditioned_structures + class DenoisedConditioner(pl.LightningModule): """ Conditioner that uses a pretrained denoiser to denoise hidden states. - + Takes hidden states, unscales them using c_in, denoises each structure, then recenters and aligns them to the current noisy positions. """ + def __init__(self, N_structures: int, pretrained_model_path: str, c_in: float, **kwargs): super().__init__() self.N_structures = N_structures @@ -102,162 +113,159 @@ def __init__(self, N_structures: int, pretrained_model_path: str, c_in: float, * # Load the pretrained denoiser py_logger = logging.getLogger("jamun") py_logger.info(f"Loading pretrained denoiser from wandb run: {pretrained_model_path}") - + # Find the checkpoint for the wandb run - checkpoint_path = find_checkpoint( - wandb_train_run_path=pretrained_model_path, - checkpoint_type="best_so_far" - ) - + checkpoint_path = find_checkpoint(wandb_train_run_path=pretrained_model_path, checkpoint_type="best_so_far") + # Load the denoiser from checkpoint self.pretrained_denoiser = Denoiser.load_from_checkpoint(checkpoint_path, strict=False) self.pretrained_denoiser.eval() # Set to evaluation mode - + # Freeze the pretrained model parameters for param in self.pretrained_denoiser.parameters(): param.requires_grad = False - + # Extract sigma from the pretrained denoiser self.denoiser_sigma = self._extract_sigma_from_denoiser() py_logger.info(f"Extracted sigma from pretrained denoiser: {self.denoiser_sigma}") py_logger.info(f"Successfully loaded pretrained denoiser with c_in={c_in}") - + def _extract_sigma_from_denoiser(self) -> float: """Extract sigma value from the pretrained denoiser's sigma distribution.""" sigma_distribution = self.pretrained_denoiser.sigma_distribution - + # Handle different types of sigma distributions - if hasattr(sigma_distribution, 'sigma'): + if hasattr(sigma_distribution, "sigma"): # For ConstantSigma distribution return float(sigma_distribution.sigma) - elif hasattr(sigma_distribution, 'mean'): + elif hasattr(sigma_distribution, "mean"): # For other distributions that might have a mean return float(sigma_distribution.mean) else: # Fallback - sample from the distribution sample = sigma_distribution.sample() return float(sample) - + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: """ Forward pass that denoises hidden states and returns conditioned structures. - + Args: y: Batch containing current positions and hidden states - + Returns: List of tensors: [y.pos, *denoised_hidden_states] """ # Use the sigma from the pretrained denoiser sigma_to_use = self.denoiser_sigma - + conditioned_structures = [y.pos] # Start with current position - + # Check if we have hidden states to process if not hasattr(y, "hidden_state") or y.hidden_state is None: # If no hidden states, just repeat current position conditioned_structures.extend([y.pos for _ in range(self.N_structures - 1)]) return conditioned_structures - + # # Move pretrained denoiser to same device as input # device = y.pos.device # self.pretrained_denoiser = self.pretrained_denoiser.to(device) - + # Process each hidden state for i, hidden_positions in enumerate(y.hidden_state): # Unscale the hidden state positions unscaled_positions = hidden_positions / self.c_in - + # Create a batch for denoising denoising_batch = y.clone() denoising_batch.pos = unscaled_positions - + # Remove hidden states from the denoising batch to avoid recursion if hasattr(denoising_batch, "hidden_state"): delattr(denoising_batch, "hidden_state") - + # Denoise the unscaled positions using the denoiser's sigma with torch.no_grad(): denoised_batch = self.pretrained_denoiser.xhat(denoising_batch, sigma_to_use) denoised_positions = denoised_batch.pos - + # Align the denoised positions to the current noisy positions aligned_positions = kabsch_algorithm(denoised_positions, y.pos, y.batch, y.num_graphs) - + conditioned_structures.append(aligned_positions) - + # Break if we've processed enough structures if len(conditioned_structures) >= self.N_structures: break - + # If we don't have enough hidden states, pad with the last denoised structure while len(conditioned_structures) < self.N_structures: conditioned_structures.append(conditioned_structures[-1]) - + return conditioned_structures class ConditionerSpiked(Conditioner): """ A conditioner that concatenates hidden states with the clean structure. - + The conditioning order is: 1. Hidden states (y.hidden_state) - if present 2. Clean structure positions (x_clean.pos) - if provided at the end """ - + def __init__(self, N_structures: int, **kwargs): super().__init__(N_structures, **kwargs) - + def forward(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: """ Create conditioning structures by concatenating hidden states with clean structure. - + Args: y: The noisy sample batch containing positions and hidden states x_clean: The clean sample batch containing ground truth positions - + Returns: List of tensors to be concatenated for conditioning """ conditioned_structures = [y.pos] - + # Add hidden states if they exist if hasattr(y, "hidden_state") and y.hidden_state is not None: for hidden_pos in y.hidden_state: conditioned_structures.append(hidden_pos) - + # Add clean structure positions at the end if provided if x_clean is not None: conditioned_structures.pop(-1) conditioned_structures.append(x_clean.pos) - + return conditioned_structures class SpatioTemporalConditioner(pl.LightningModule): """ Conditioner that uses a spatio-temporal model to process hidden states. - + This conditioner takes the current positions and hidden states, processes them through a spatio-temporal model, and returns [y.pos, spatial_features]. Always returns exactly 2 structures: the original positions and computed spatial features. - + By default, the spatiotemporal model is trainable. Set freeze_spatiotemporal_model=True to freeze the parameters (e.g., when using a pretrained model). """ - + def __init__( - self, + self, N_structures: int, spatiotemporal_model: torch.nn.Module, c_noise: float = 0.0, freeze_spatiotemporal_model: bool = False, - **kwargs + **kwargs, ): """ Initialize the SpatioTemporalConditioner. - + Args: N_structures: Number of structures parameter (ignored - this conditioner always returns 1 structure) spatiotemporal_model: The E3SpatioTemporal model to use for processing @@ -270,68 +278,70 @@ def __init__( self.spatiotemporal_model = spatiotemporal_model self.c_noise = c_noise self.freeze_spatiotemporal_model = freeze_spatiotemporal_model - + # Only freeze parameters if explicitly requested if self.freeze_spatiotemporal_model: self.freeze_spatiotemporal_parameters() # Set to evaluation mode when frozen self.spatiotemporal_model.eval() - + def freeze_spatiotemporal_parameters(self): """Freeze the spatiotemporal model parameters.""" for param in self.spatiotemporal_model.parameters(): param.requires_grad = False - + def unfreeze_spatiotemporal_parameters(self): """Unfreeze the spatiotemporal model parameters.""" for param in self.spatiotemporal_model.parameters(): param.requires_grad = True - + def configure_for_inference(self): """Configure the conditioner for inference (freeze parameters and set eval mode).""" self.freeze_spatiotemporal_model = True self.freeze_spatiotemporal_parameters() self.spatiotemporal_model.eval() - + def configure_for_training(self): """Configure the conditioner for training (unfreeze parameters and set train mode).""" self.freeze_spatiotemporal_model = False self.unfreeze_spatiotemporal_parameters() self.spatiotemporal_model.train() - + def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: """ Forward pass that processes the batch through the spatio-temporal model. - + Args: y: Batch containing current positions and hidden states - + Returns: List containing [y.pos, spatial_features] for concatenation by the denoiser """ # Align hidden positions with current position before processing - if hasattr(y, 'hidden_state') and y.hidden_state is not None and len(y.hidden_state) > 0: + if hasattr(y, "hidden_state") and y.hidden_state is not None and len(y.hidden_state) > 0: # Create a copy of the batch to avoid modifying the original y_aligned = y.clone() - + # Align each hidden state to the current position aligned_hidden_states = [] for hidden_pos in y.hidden_state: # Align hidden_pos to y.pos using Kabsch algorithm (same as PositionConditioner) aligned_hidden_pos = kabsch_algorithm(hidden_pos, y.pos, y.batch, y.num_graphs) aligned_hidden_states.append(aligned_hidden_pos) - + # Update the hidden states in the aligned batch y_aligned.hidden_state = aligned_hidden_states else: # If no hidden states, use original batch y_aligned = y - + # Prepare noise conditioning device = y_aligned.pos.device sigma = torch.tensor(self.c_noise, device=device) - sigma = unsqueeze_trailing(sigma, 1) # actually this is correct, but this is bad variable naming. the positional e3conv will take a c_noise, so this is right, but it is not right to call it sigma. - + sigma = unsqueeze_trailing( + sigma, 1 + ) # actually this is correct, but this is bad variable naming. the positional e3conv will take a c_noise, so this is right, but it is not right to call it sigma. + # Process through spatio-temporal model with aligned hidden states # Only disable gradients if the model is frozen if self.freeze_spatiotemporal_model: @@ -340,8 +350,7 @@ def forward(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: else: # Allow gradients to flow when training spatial_features = self.spatiotemporal_model(y_aligned, sigma) - + # Return list containing [y.pos, spatial_features] for concatenation # The denoiser will concatenate these along the feature dimension return [y.pos, spatial_features] - diff --git a/src/jamun/model/denoiser_conditional.py b/src/jamun/model/denoiser_conditional.py index 60761de..c8008ee 100644 --- a/src/jamun/model/denoiser_conditional.py +++ b/src/jamun/model/denoiser_conditional.py @@ -1,17 +1,16 @@ import logging -from typing import Callable, Dict, Optional, Tuple, Union +import os +from collections.abc import Callable +import e3tools import lightning.pytorch as pl import numpy as np import torch import torch_geometric -import e3tools from e3tools import scatter -from tqdm import tqdm -from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils import mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm -import os class Denoiser(pl.LightningModule): @@ -31,11 +30,11 @@ def __init__( mean_center: bool, mirror_augmentation_rate: float, bond_loss_coefficient: float = 1.0, - normalization_type: Optional[str] = "JAMUN", - sigma_data: Optional[float] = None, # Only used if normalization_type is "EDM" - lr_scheduler_config: Optional[Dict] = None, + normalization_type: str | None = "JAMUN", + sigma_data: float | None = None, # Only used if normalization_type is "EDM" + lr_scheduler_config: dict | None = None, use_torch_compile: bool = True, - torch_compile_kwargs: Optional[Dict] = None, + torch_compile_kwargs: dict | None = None, conditioner: Callable[..., list[torch.Tensor]] = None, rotational_augmentation: bool = False, alignment_correction_order: int = 0, @@ -128,7 +127,7 @@ def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: return self.conditioning_module(y) else: raise ValueError("Conditioner must be a callable or None") - + def _align_A_to_B_batched_with_hidden_states( self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch ) -> torch_geometric.data.Batch: @@ -138,13 +137,11 @@ def _align_A_to_B_batched_with_hidden_states( # Align positions A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) - # Align hidden states + # Align hidden states if hasattr(A, "hidden_state") and A.hidden_state is not None: A_aligned.hidden_state = [] for i in range(len(A.hidden_state)): - A_aligned.hidden_state.append(kabsch_algorithm( - A.hidden_state[i], B.pos, A.batch, A.num_graphs - )) + A_aligned.hidden_state.append(kabsch_algorithm(A.hidden_state[i], B.pos, A.batch, A.num_graphs)) return A_aligned def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): @@ -153,8 +150,8 @@ def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): mean = scatter(data.hidden_state[i], data.batch, dim=0, reduce="mean") data.hidden_state[i] = data.hidden_state[i] - mean[data.batch] return data - - def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: + + def add_noise(self, x: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.pos.ndim) @@ -167,12 +164,17 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten num_batches = x.batch.max().item() + 1 if len(x.pos.shape) == 2: num_nodes_per_batch = x.pos.shape[0] // num_batches - noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) - hidden_noise = [torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat(num_batches, 1) for i in range(len(x.hidden_state))] + noise = torch.randn_like(x.pos[:num_nodes_per_batch]).repeat(num_batches, 1) + hidden_noise = [ + torch.randn_like(x.hidden_state[i][:num_nodes_per_batch]).repeat(num_batches, 1) + for i in range(len(x.hidden_state)) + ] if len(x.pos.shape) == 3: num_nodes_per_batch = x.pos.shape[1] - noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) - hidden_noise = [torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state))] + noise = torch.randn_like(x.pos[0]).repeat(num_batches, 1, 1) + hidden_noise = [ + torch.randn_like(x.hidden_state[i][0]).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state)) + ] else: noise = torch.randn_like(x.pos) hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] @@ -183,12 +185,12 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten y.pos = -y.pos return y - def score(self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: + def score(self, y: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: """Compute the score function.""" sigma = torch.as_tensor(sigma).to(y.pos) return (self.xhat(y, sigma).pos - y.pos) / (unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2) - def normalization_factors(self, sigma: float, D: int = 3) -> Tuple[float, float, float, float]: + def normalization_factors(self, sigma: float, D: int = 3) -> tuple[float, float, float, float]: """Normalization factors for the input and output.""" sigma = torch.as_tensor(sigma) @@ -219,7 +221,7 @@ def loss_weight(self, sigma: float, D: int = 3) -> float: _, _, c_out, _ = self.normalization_factors(sigma, D) return 1 / (c_out**2) - def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Tensor: + def effective_radial_cutoff(self, sigma: float | torch.Tensor) -> torch.Tensor: """Compute the effective radial cutoff for the noise level.""" return torch.sqrt((self.max_radius**2) + 6 * (sigma**2)) @@ -254,9 +256,7 @@ def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torc y.bond_mask = bond_mask return y - def xhat_normalized( - self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] - ) -> torch_geometric.data.Batch: + def xhat_normalized(self, y: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: """Compute the denoised prediction using the normalization factors from JAMUN.""" sigma = torch.as_tensor(sigma).to(y.pos) D = y.pos.shape[-1] @@ -291,19 +291,23 @@ def xhat_normalized( if hasattr(y, "hidden_state") and y.hidden_state is not None: xhat.hidden_state = [h.clone() for h in y.hidden_state] - with torch.cuda.nvtx.range("conditioning"): + with torch.cuda.nvtx.range("conditioning"): conditioned_structures = self.conditioner(y_scaled) # print(f"Conditioner is working, number of conditioned structures: {len(conditioned_structures)}") - with torch.cuda.nvtx.range("g"): - g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), topology=y_scaled, \ - c_noise=c_noise, effective_radial_cutoff=radial_cutoff) + with torch.cuda.nvtx.range("g"): + g_pred = self.g( + torch.cat([*conditioned_structures], dim=-1), + topology=y_scaled, + c_noise=c_noise, + effective_radial_cutoff=radial_cutoff, + ) xhat.pos = c_skip * y.pos + c_out * g_pred if hasattr(y, "hidden_state") and y.hidden_state is not None: xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] return xhat - def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): + def xhat(self, y: torch.Tensor, sigma: float | torch.Tensor): """Compute the denoised prediction.""" if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): @@ -323,9 +327,9 @@ def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): def noise_and_denoise( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, - ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: + ) -> tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: """ Add noise to the input and denoise it. Returns the target for the loss, the prediction, and the noisy input. @@ -366,8 +370,8 @@ def compute_loss( self, x: torch_geometric.data.Batch, xhat: torch.Tensor, - sigma: Union[float, torch.Tensor], - ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + sigma: float | torch.Tensor, + ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Compute the loss.""" if self.mean_center: with torch.cuda.nvtx.range("mean_center_x"): @@ -402,23 +406,25 @@ def compute_loss( def noise_and_compute_loss( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, - ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Add noise to the input and compute the loss.""" x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) return self.compute_loss(x_target, xhat, sigma) def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): """The standard step for automatic optimization.""" - align_noisy_input = self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation + align_noisy_input = ( + self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation + ) sigma = self.sigma_distribution.sample().to(self.device) - + loss, aux = self.noise_and_compute_loss( batch, sigma, align_noisy_input=align_noisy_input, - ) # check if the loss is nan. if nan then save the model, and the batch and see what went on. + ) # check if the loss is nan. if nan then save the model, and the batch and see what went on. if torch.isnan(loss.sum()): print(f"Loss is nan at step {self.global_step}") print(f"Batch: {batch}") @@ -429,17 +435,17 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): # Create debug directory if it doesn't exist debug_dir = f"/homefs/home/sules/jamun/debug_nan_loss_step_{self.global_step}" os.makedirs(debug_dir, exist_ok=True) - + # Save model checkpoint checkpoint_path = os.path.join(debug_dir, "model_nan_loss.ckpt") self.trainer.save_checkpoint(checkpoint_path) print(f"Model saved to {checkpoint_path}") - + torch.save(batch, debug_dir + "/batch_nan_loss.pt") - + # Optionally raise an exception to stop training raise RuntimeError(f"NaN loss detected at step {self.global_step}. Debug files saved to {debug_dir}") - + # Average the loss and other metrics over all graphs. with torch.cuda.nvtx.range("mean_over_graphs"): aux["loss"] = loss @@ -449,18 +455,20 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): self.log(f"train/{key}", aux[key], prog_bar=False, batch_size=batch.num_graphs, sync_dist=False) elif stage == "val": self.log( - f"val/{key}", aux[key], prog_bar=(key == "scaled_rmsd"), batch_size=batch.num_graphs, sync_dist=True - ) + f"val/{key}", + aux[key], + prog_bar=(key == "scaled_rmsd"), + batch_size=batch.num_graphs, + sync_dist=True, + ) else: continue - return { "sigma": sigma, **aux, } - def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): """Called during training.""" return self._automatic_step(batch, "train") diff --git a/src/jamun/model/denoiser_multimeasurement.py b/src/jamun/model/denoiser_multimeasurement.py index 681d035..d003945 100644 --- a/src/jamun/model/denoiser_multimeasurement.py +++ b/src/jamun/model/denoiser_multimeasurement.py @@ -1,5 +1,5 @@ import logging -from typing import Callable, Dict, Optional, Tuple, Union +from collections.abc import Callable import e3tools import lightning.pytorch as pl @@ -7,11 +7,9 @@ import torch import torch_geometric from e3tools import scatter -from tqdm import tqdm -from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils import mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm -import os class DenoiserMultimeasurement(pl.LightningModule): @@ -31,11 +29,11 @@ def __init__( mean_center: bool, mirror_augmentation_rate: float, bond_loss_coefficient: float = 1.0, - normalization_type: Optional[str] = "JAMUN", - sigma_data: Optional[float] = None, # Only used if normalization_type is "EDM" - lr_scheduler_config: Optional[Dict] = None, + normalization_type: str | None = "JAMUN", + sigma_data: float | None = None, # Only used if normalization_type is "EDM" + lr_scheduler_config: dict | None = None, use_torch_compile: bool = True, - torch_compile_kwargs: Optional[Dict] = None, + torch_compile_kwargs: dict | None = None, conditioner: Callable[..., list[torch.Tensor]] = None, multimeasurement: bool = False, N_measurements_hidden: int = 1, @@ -122,9 +120,7 @@ def __init__( self.N_measurements = N_measurements self.max_graphs_per_batch = max_graphs_per_batch if not self.automatic_optimization: - py_logger.info( - f"Manual optimization enabled with micro-batch size of {self.max_graphs_per_batch} graphs." - ) + py_logger.info(f"Manual optimization enabled with micro-batch size of {self.max_graphs_per_batch} graphs.") def on_before_optimizer_step(self, optimizer): # Log gradients and parameters. @@ -142,13 +138,11 @@ def _align_A_to_B_batched_with_hidden_states( # Align positions A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) - # Align hidden states + # Align hidden states if hasattr(A, "hidden_state") and A.hidden_state is not None: A_aligned.hidden_state = [] for i in range(len(A.hidden_state)): - A_aligned.hidden_state.append(kabsch_algorithm( - A.hidden_state[i], B.pos, A.batch, A.num_graphs - )) + A_aligned.hidden_state.append(kabsch_algorithm(A.hidden_state[i], B.pos, A.batch, A.num_graphs)) return A_aligned def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): @@ -161,7 +155,7 @@ def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): def _prepare_noisy_batch( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, ): """Prepare a batch of noisy graphs and their targets.""" @@ -174,19 +168,13 @@ def _prepare_noisy_batch( sigma_tensor = torch.as_tensor(sigma).to(x_processed.pos.device) - y = self.add_noise_hiddens( - x_processed, self.N_measurements_hidden, self.N_measurements, sigma_tensor - ) + y = self.add_noise_hiddens(x_processed, self.N_measurements_hidden, self.N_measurements, sigma_tensor) x_list = x_processed.to_data_list() repeated_x_list = [ - graph.clone() - for graph in x_list - for _ in range(self.N_measurements_hidden * self.N_measurements) + graph.clone() for graph in x_list for _ in range(self.N_measurements_hidden * self.N_measurements) ] - x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to( - x_processed.pos.device - ) + x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to(x_processed.pos.device) if self.mean_center: y = mean_center(y) @@ -209,9 +197,7 @@ def conditioner(self, y: torch_geometric.data.Batch) -> list[torch.Tensor]: else: raise ValueError("Conditioner must be a callable or None") - def add_noise( - self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] - ) -> torch_geometric.data.Batch: + def add_noise(self, x: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.pos.ndim) @@ -219,9 +205,7 @@ def add_noise( if self.add_fixed_ones: noise = torch.ones_like(x.pos) if hasattr(x, "hidden_state") and x.hidden_state is not None: - hidden_noise = [ - torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) - ] + hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] else: hidden_noise = [] elif self.add_fixed_noise: @@ -229,22 +213,20 @@ def add_noise( num_batches = x.batch.max().item() + 1 if len(x.pos.shape) == 2: num_nodes_per_batch = x.pos.shape[0] // num_batches - noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) + noise = torch.randn_like(x.pos[:num_nodes_per_batch]).repeat(num_batches, 1) if hasattr(x, "hidden_state") and x.hidden_state is not None: hidden_noise = [ - torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat( - num_batches, 1 - ) + torch.randn_like(x.hidden_state[i][:num_nodes_per_batch]).repeat(num_batches, 1) for i in range(len(x.hidden_state)) ] else: hidden_noise = [] if len(x.pos.shape) == 3: num_nodes_per_batch = x.pos.shape[1] - noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) + noise = torch.randn_like(x.pos[0]).repeat(num_batches, 1, 1) if hasattr(x, "hidden_state") and x.hidden_state is not None: hidden_noise = [ - torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) + torch.randn_like(x.hidden_state[i][0]).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state)) ] else: @@ -252,15 +234,13 @@ def add_noise( else: noise = torch.randn_like(x.pos) if hasattr(x, "hidden_state") and x.hidden_state is not None: - hidden_noise = [ - torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state)) - ] + hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] else: hidden_noise = [] y.pos = x.pos + sigma * noise if hasattr(y, "hidden_state") and y.hidden_state is not None and hidden_noise: for i in range(len(y.hidden_state)): - y.hidden_state[i] = x.hidden_state[i] + sigma*hidden_noise[i] + y.hidden_state[i] = x.hidden_state[i] + sigma * hidden_noise[i] if torch.rand(()) < self.mirror_augmentation_rate: y.pos = -y.pos return y @@ -270,7 +250,7 @@ def add_noise_hiddens( x: torch_geometric.data.Batch, N_measurements_hidden: int, N_measurements: int, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, ) -> torch_geometric.data.Batch: """ Makes N_measurements_hidden number of noisy copies of the hidden states of x @@ -313,18 +293,12 @@ def add_noise_hiddens( return torch_geometric.data.Batch.from_data_list(noisy_y_list) - def score( - self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] - ) -> torch_geometric.data.Batch: + def score(self, y: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: """Compute the score function.""" sigma = torch.as_tensor(sigma).to(y.pos) - return (self.xhat(y, sigma).pos - y.pos) / ( - unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2 - ) + return (self.xhat(y, sigma).pos - y.pos) / (unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2) - def normalization_factors( - self, sigma: float, D: int = 3 - ) -> Tuple[float, float, float, float]: + def normalization_factors(self, sigma: float, D: int = 3) -> tuple[float, float, float, float]: """Normalization factors for the input and output.""" sigma = torch.as_tensor(sigma) @@ -355,13 +329,11 @@ def loss_weight(self, sigma: float, D: int = 3) -> float: _, _, c_out, _ = self.normalization_factors(sigma, D) return 1 / (c_out**2) - def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Tensor: + def effective_radial_cutoff(self, sigma: float | torch.Tensor) -> torch.Tensor: """Compute the effective radial cutoff for the noise level.""" return torch.sqrt((self.max_radius**2) + 6 * (sigma**2)) - def add_edges( - self, y: torch_geometric.data.Batch, radial_cutoff: float - ) -> torch_geometric.data.Batch: + def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torch_geometric.data.Batch: """Add edges to the graph based on the effective radial cutoff.""" if y.get("edge_index") is not None: return y @@ -378,18 +350,12 @@ def add_edges( with torch.cuda.nvtx.range("concatenate_edges"): edge_index = torch.cat((radial_edge_index, y.bonded_edge_index), dim=-1) if y.bonded_edge_index.numel() == 0: - bond_mask = torch.zeros( - radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device - ) + bond_mask = torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device) else: bond_mask = torch.cat( ( - torch.zeros( - radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device - ), - torch.ones( - y.bonded_edge_index.shape[1], dtype=torch.long, device=y.pos.device - ), + torch.zeros(radial_edge_index.shape[1], dtype=torch.long, device=y.pos.device), + torch.ones(y.bonded_edge_index.shape[1], dtype=torch.long, device=y.pos.device), ), dim=0, ) @@ -398,9 +364,7 @@ def add_edges( y.bond_mask = bond_mask return y - def xhat_normalized( - self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor] - ) -> torch_geometric.data.Batch: + def xhat_normalized(self, y: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: """Compute the denoised prediction using the normalization factors from JAMUN.""" sigma = torch.as_tensor(sigma).to(y.pos) D = y.pos.shape[-1] @@ -445,10 +409,10 @@ def xhat_normalized( xhat.pos = c_skip * y.pos + c_out * g_pred if hasattr(y, "hidden_state") and y.hidden_state is not None: - xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] # the hidden state updates! + xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] # the hidden state updates! return xhat - def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): + def xhat(self, y: torch.Tensor, sigma: float | torch.Tensor): """Compute the denoised prediction.""" if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): @@ -469,9 +433,9 @@ def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor]): def noise_and_denoise( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, - ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: + ) -> tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: """ Add noise to the input and denoise it. Returns the target for the loss, the prediction, and the noisy input. @@ -488,20 +452,14 @@ def noise_and_denoise( if self.multimeasurement: with torch.cuda.nvtx.range("add_noise_hiddens"): - y = self.add_noise_hiddens( - x_processed, self.N_measurements_hidden, self.N_measurements, sigma - ) + y = self.add_noise_hiddens(x_processed, self.N_measurements_hidden, self.N_measurements, sigma) # Repeat x_processed to match y's batch size for alignment and loss calculation. x_list = x_processed.to_data_list() repeated_x_list = [ - graph.clone() - for graph in x_list - for _ in range(self.N_measurements_hidden * self.N_measurements) + graph.clone() for graph in x_list for _ in range(self.N_measurements_hidden * self.N_measurements) ] - x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to( - x_processed.pos.device - ) + x_target = torch_geometric.data.Batch.from_data_list(repeated_x_list).to(x_processed.pos.device) else: with torch.cuda.nvtx.range("add_noise"): @@ -527,8 +485,8 @@ def compute_loss( self, x: torch_geometric.data.Batch, xhat: torch.Tensor, - sigma: Union[float, torch.Tensor], - ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + sigma: float | torch.Tensor, + ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Compute the loss.""" if self.mean_center: with torch.cuda.nvtx.range("mean_center_x"): @@ -564,9 +522,9 @@ def compute_loss( def noise_and_compute_loss( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, - ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Add noise to the input and compute the loss.""" x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) return self.compute_loss(x_target, xhat, sigma) @@ -574,9 +532,7 @@ def noise_and_compute_loss( def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): """The standard step for automatic optimization.""" align_noisy_input = ( - self.align_noisy_input_during_training - if stage == "train" - else self.align_noisy_input_during_evaluation + self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation ) sigma = self.sigma_distribution.sample().to(self.device) @@ -584,7 +540,7 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): batch, sigma, align_noisy_input=align_noisy_input, - ) # check if the loss is nan. if nan then save the model, and the batch and see what went on. + ) # check if the loss is nan. if nan then save the model, and the batch and see what went on. # if torch.isnan(loss.sum()): # print(f"Loss is nan at step {self.global_step}") # print(f"Batch: {batch}") @@ -595,17 +551,17 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): # # Create debug directory if it doesn't exist # debug_dir = f"/homefs/home/sules/jamun/debug_nan_loss_step_{self.global_step}" # os.makedirs(debug_dir, exist_ok=True) - # + # # # Save model checkpoint # checkpoint_path = os.path.join(debug_dir, "model_nan_loss.ckpt") # self.trainer.save_checkpoint(checkpoint_path) # print(f"Model saved to {checkpoint_path}") - # + # # torch.save(batch, debug_dir + "/batch_nan_loss.pt") - # + # # # Optionally raise an exception to stop training # raise RuntimeError(f"NaN loss detected at step {self.global_step}. Debug files saved to {debug_dir}") - + # Average the loss and other metrics over all graphs. with torch.cuda.nvtx.range("mean_over_graphs"): aux["loss"] = loss @@ -639,9 +595,7 @@ def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): """A shared step for training and validation with manual optimization.""" sigma = self.sigma_distribution.sample().to(self.device) align_noisy_input = ( - self.align_noisy_input_during_training - if stage == "train" - else self.align_noisy_input_during_evaluation + self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation ) y, x_target = self._prepare_noisy_batch(batch, sigma, align_noisy_input) @@ -667,9 +621,7 @@ def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): continue y_micro_batch = torch_geometric.data.Batch.from_data_list(y_micro_batch_list) - x_target_micro_batch = torch_geometric.data.Batch.from_data_list( - x_target_micro_batch_list - ) + x_target_micro_batch = torch_geometric.data.Batch.from_data_list(x_target_micro_batch_list) xhat_micro_batch = self.xhat(y_micro_batch, sigma) @@ -697,7 +649,7 @@ def _manual_step(self, batch: torch_geometric.data.Batch, stage: str): "batch_size": len(y_list), "sync_dist": (stage == "val"), # Only sync for validation } - + # Ensure training metrics are always logged if stage == "train": log_opts["on_step"] = True @@ -735,4 +687,4 @@ def configure_optimizers(self): **self.lr_scheduler_config, } - return out \ No newline at end of file + return out diff --git a/src/jamun/model/denoiser_spiked.py b/src/jamun/model/denoiser_spiked.py index 4dd6bd9..401b20f 100644 --- a/src/jamun/model/denoiser_spiked.py +++ b/src/jamun/model/denoiser_spiked.py @@ -1,5 +1,5 @@ import logging -from typing import Callable, Dict, Optional, Tuple, Union +from collections.abc import Callable import e3tools import lightning.pytorch as pl @@ -7,9 +7,8 @@ import torch import torch_geometric from e3tools import scatter -from tqdm import tqdm -from jamun.utils import align_A_to_B_batched, mean_center, unsqueeze_trailing +from jamun.utils import mean_center, unsqueeze_trailing from jamun.utils.align import kabsch_algorithm @@ -30,11 +29,11 @@ def __init__( mean_center: bool, mirror_augmentation_rate: float, bond_loss_coefficient: float = 1.0, - normalization_type: Optional[str] = "JAMUN", - sigma_data: Optional[float] = None, # Only used if normalization_type is "EDM" - lr_scheduler_config: Optional[Dict] = None, + normalization_type: str | None = "JAMUN", + sigma_data: float | None = None, # Only used if normalization_type is "EDM" + lr_scheduler_config: dict | None = None, use_torch_compile: bool = True, - torch_compile_kwargs: Optional[Dict] = None, + torch_compile_kwargs: dict | None = None, conditioner: Callable[..., list[torch.Tensor]] = None, ): super().__init__() @@ -113,20 +112,24 @@ def on_before_optimizer_step(self, optimizer): if param.grad is not None: self.log(f"gradient_norms/{name}", param.grad.norm(), sync_dist=True) - def conditioner_default(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: + def conditioner_default( + self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None + ) -> list[torch.Tensor]: conditioned_structures = [y.pos] # Return complete list starting with current position if x_clean is not None: conditioned_structures.append(x_clean.pos) # Add clean sample positions return conditioned_structures - def conditioner(self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None) -> list[torch.Tensor]: + def conditioner( + self, y: torch_geometric.data.Batch, x_clean: torch_geometric.data.Batch = None + ) -> list[torch.Tensor]: if self.conditioning_module is None: return self.conditioner_default(y, x_clean) elif callable(self.conditioning_module): return self.conditioning_module(y, x_clean) else: raise ValueError("Conditioner must be a callable or None") - + def _align_A_to_B_batched_with_hidden_states( self, A: torch_geometric.data.Batch, B: torch_geometric.data.Batch ) -> torch_geometric.data.Batch: @@ -136,13 +139,11 @@ def _align_A_to_B_batched_with_hidden_states( # Align positions A_aligned.pos = kabsch_algorithm(A.pos, B.pos, A.batch, A.num_graphs) - # Align hidden states + # Align hidden states if hasattr(A, "hidden_state") and A.hidden_state is not None: A_aligned.hidden_state = [] for i in range(len(A.hidden_state)): - A_aligned.hidden_state.append(kabsch_algorithm( - A.hidden_state[i], B.pos, A.batch, A.num_graphs - )) + A_aligned.hidden_state.append(kabsch_algorithm(A.hidden_state[i], B.pos, A.batch, A.num_graphs)) return A_aligned def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): @@ -151,8 +152,8 @@ def _mean_center_hidden_states(self, data: torch_geometric.data.Batch): mean = scatter(data.hidden_state[i], data.batch, dim=0, reduce="mean") data.hidden_state[i] = data.hidden_state[i] - mean[data.batch] return data - - def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor]) -> torch_geometric.data.Batch: + + def add_noise(self, x: torch_geometric.data.Batch, sigma: float | torch.Tensor) -> torch_geometric.data.Batch: # pos [B, ...] sigma = unsqueeze_trailing(sigma, x.pos.ndim) @@ -165,12 +166,17 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten num_batches = x.batch.max().item() + 1 if len(x.pos.shape) == 2: num_nodes_per_batch = x.pos.shape[0] // num_batches - noise = torch.randn_like((x.pos[:num_nodes_per_batch])).repeat(num_batches, 1) - hidden_noise = [torch.randn_like((x.hidden_state[i][:num_nodes_per_batch])).repeat(num_batches, 1) for i in range(len(x.hidden_state))] + noise = torch.randn_like(x.pos[:num_nodes_per_batch]).repeat(num_batches, 1) + hidden_noise = [ + torch.randn_like(x.hidden_state[i][:num_nodes_per_batch]).repeat(num_batches, 1) + for i in range(len(x.hidden_state)) + ] if len(x.pos.shape) == 3: num_nodes_per_batch = x.pos.shape[1] - noise = torch.randn_like((x.pos[0])).repeat(num_batches, 1, 1) - hidden_noise = [torch.randn_like((x.hidden_state[i][0])).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state))] + noise = torch.randn_like(x.pos[0]).repeat(num_batches, 1, 1) + hidden_noise = [ + torch.randn_like(x.hidden_state[i][0]).repeat(num_batches, 1, 1) for i in range(len(x.hidden_state)) + ] else: noise = torch.randn_like(x.pos) hidden_noise = [torch.randn_like(x.hidden_state[i]) for i in range(len(x.hidden_state))] @@ -181,12 +187,14 @@ def add_noise(self, x: torch_geometric.data.Batch, sigma: Union[float, torch.Ten y.pos = -y.pos return y - def score(self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor], x_clean: torch_geometric.data.Batch) -> torch_geometric.data.Batch: + def score( + self, y: torch_geometric.data.Batch, sigma: float | torch.Tensor, x_clean: torch_geometric.data.Batch + ) -> torch_geometric.data.Batch: """Compute the score function.""" sigma = torch.as_tensor(sigma).to(y.pos) return (self.xhat(y, sigma, x_clean).pos - y.pos) / (unsqueeze_trailing(sigma, y.pos.ndim - 1) ** 2) - def normalization_factors(self, sigma: float, D: int = 3) -> Tuple[float, float, float, float]: + def normalization_factors(self, sigma: float, D: int = 3) -> tuple[float, float, float, float]: """Normalization factors for the input and output.""" sigma = torch.as_tensor(sigma) @@ -217,7 +225,7 @@ def loss_weight(self, sigma: float, D: int = 3) -> float: _, _, c_out, _ = self.normalization_factors(sigma, D) return 1 / (c_out**2) - def effective_radial_cutoff(self, sigma: Union[float, torch.Tensor]) -> torch.Tensor: + def effective_radial_cutoff(self, sigma: float | torch.Tensor) -> torch.Tensor: """Compute the effective radial cutoff for the noise level.""" return torch.sqrt((self.max_radius**2) + 6 * (sigma**2)) @@ -253,7 +261,7 @@ def add_edges(self, y: torch_geometric.data.Batch, radial_cutoff: float) -> torc return y def xhat_normalized( - self, y: torch_geometric.data.Batch, sigma: Union[float, torch.Tensor], x_clean: torch_geometric.data.Batch + self, y: torch_geometric.data.Batch, sigma: float | torch.Tensor, x_clean: torch_geometric.data.Batch ) -> torch_geometric.data.Batch: """Compute the denoised prediction using the normalization factors from JAMUN.""" sigma = torch.as_tensor(sigma).to(y.pos) @@ -290,19 +298,23 @@ def xhat_normalized( if hasattr(y, "hidden_state") and y.hidden_state is not None: xhat.hidden_state = [h.clone() for h in y.hidden_state] - with torch.cuda.nvtx.range("conditioning"): + with torch.cuda.nvtx.range("conditioning"): conditioned_structures = self.conditioner(y_scaled, x_clean) # print(f"Conditioner is working, number of conditioned structures: {len(conditioned_structures)}") - with torch.cuda.nvtx.range("g"): - g_pred = self.g(torch.cat([*conditioned_structures], dim=-1), topology=y_scaled, \ - c_noise=c_noise, effective_radial_cutoff=radial_cutoff) + with torch.cuda.nvtx.range("g"): + g_pred = self.g( + torch.cat([*conditioned_structures], dim=-1), + topology=y_scaled, + c_noise=c_noise, + effective_radial_cutoff=radial_cutoff, + ) xhat.pos = c_skip * y.pos + c_out * g_pred if hasattr(y, "hidden_state") and y.hidden_state is not None: xhat.hidden_state = [y.pos, *y.hidden_state[:-1]] return xhat - def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor], x_clean: torch_geometric.data.Batch): + def xhat(self, y: torch.Tensor, sigma: float | torch.Tensor, x_clean: torch_geometric.data.Batch): """Compute the denoised prediction.""" if self.mean_center: with torch.cuda.nvtx.range("mean_center_y"): @@ -325,9 +337,9 @@ def xhat(self, y: torch.Tensor, sigma: Union[float, torch.Tensor], x_clean: torc def noise_and_denoise( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, - ) -> Tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: + ) -> tuple[torch_geometric.data.Batch, torch_geometric.data.Batch, torch_geometric.data.Batch]: """ Add noise to the input and denoise it. Returns the target for the loss, the prediction, and the noisy input. @@ -369,8 +381,8 @@ def compute_loss( self, x: torch_geometric.data.Batch, xhat: torch.Tensor, - sigma: Union[float, torch.Tensor], - ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + sigma: float | torch.Tensor, + ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Compute the loss.""" if self.mean_center: with torch.cuda.nvtx.range("mean_center_x"): @@ -405,18 +417,20 @@ def compute_loss( def noise_and_compute_loss( self, x: torch_geometric.data.Batch, - sigma: Union[float, torch.Tensor], + sigma: float | torch.Tensor, align_noisy_input: bool, - ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: + ) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]: """Add noise to the input and compute the loss.""" x_target, xhat, _ = self.noise_and_denoise(x, sigma, align_noisy_input=align_noisy_input) return self.compute_loss(x_target, xhat, sigma) def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): """The standard step for automatic optimization.""" - align_noisy_input = self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation + align_noisy_input = ( + self.align_noisy_input_during_training if stage == "train" else self.align_noisy_input_during_evaluation + ) sigma = self.sigma_distribution.sample().to(self.device) - + loss, aux = self.noise_and_compute_loss( batch, sigma, @@ -432,18 +446,20 @@ def _automatic_step(self, batch: torch_geometric.data.Batch, stage: str): self.log(f"train/{key}", aux[key], prog_bar=False, batch_size=batch.num_graphs, sync_dist=False) elif stage == "val": self.log( - f"val/{key}", aux[key], prog_bar=(key == "scaled_rmsd"), batch_size=batch.num_graphs, sync_dist=True - ) + f"val/{key}", + aux[key], + prog_bar=(key == "scaled_rmsd"), + batch_size=batch.num_graphs, + sync_dist=True, + ) else: continue - return { "sigma": sigma, **aux, } - def training_step(self, batch: torch_geometric.data.Batch, batch_idx: int): """Called during training.""" return self._automatic_step(batch, "train") @@ -464,4 +480,4 @@ def configure_optimizers(self): **self.lr_scheduler_config, } - return out \ No newline at end of file + return out diff --git a/src/jamun/model/noise_test.py b/src/jamun/model/noise_test.py index 5c316b4..1eae2c4 100644 --- a/src/jamun/model/noise_test.py +++ b/src/jamun/model/noise_test.py @@ -1,7 +1,6 @@ import torch -import torch_geometric -from torch_geometric.data import Data, Batch -from typing import List +from torch_geometric.data import Batch, Data + def add_noise_hiddens(x: Batch, N_measurements_hidden: int, N_measurements: int, sigma: float) -> Batch: """ @@ -25,7 +24,7 @@ def add_noise_hiddens(x: Batch, N_measurements_hidden: int, N_measurements: int, for _ in range(N_measurements_hidden): # Create a noisy version of the hidden state noisy_hidden_state = [] - if hasattr(graph, 'hidden_state') and graph.hidden_state is not None: + if hasattr(graph, "hidden_state") and graph.hidden_state is not None: for hs_tensor in graph.hidden_state: noise = torch.randn_like(hs_tensor) * sigma noisy_hidden_state.append(hs_tensor + noise) @@ -38,9 +37,9 @@ def add_noise_hiddens(x: Batch, N_measurements_hidden: int, N_measurements: int, noisy_graph.pos = graph.pos + pos_noise # Assign the noisy hidden state - if hasattr(graph, 'hidden_state') and graph.hidden_state is not None: + if hasattr(graph, "hidden_state") and graph.hidden_state is not None: noisy_graph.hidden_state = [hs.clone() for hs in noisy_hidden_state] - + noisy_y_list.append(noisy_graph) return Batch.from_data_list(noisy_y_list) @@ -53,14 +52,10 @@ def run_test(): # 1. Create dummy data num_nodes = 5 # single data object - data1 = Data( - pos=torch.randn(num_nodes, 3), - hidden_state=[torch.randn(num_nodes, 4), torch.randn(num_nodes, 8)] - ) + data1 = Data(pos=torch.randn(num_nodes, 3), hidden_state=[torch.randn(num_nodes, 4), torch.randn(num_nodes, 8)]) # another data object data2 = Data( - pos=torch.randn(num_nodes + 2, 3), - hidden_state=[torch.randn(num_nodes + 2, 4), torch.randn(num_nodes + 2, 8)] + pos=torch.randn(num_nodes + 2, 3), hidden_state=[torch.randn(num_nodes + 2, 4), torch.randn(num_nodes + 2, 8)] ) original_batch = Batch.from_data_list([data1, data2]) @@ -76,35 +71,36 @@ def run_test(): # 4. Assertions # Check total number of graphs expected_num_graphs = original_batch.num_graphs * N_measurements_hidden * N_measurements - assert noisy_batch.num_graphs == expected_num_graphs, \ + assert noisy_batch.num_graphs == expected_num_graphs, ( f"Expected {expected_num_graphs} graphs, but got {noisy_batch.num_graphs}" + ) print(f"Correct number of graphs in output batch: {noisy_batch.num_graphs}") noisy_graphs = noisy_batch.to_data_list() - + # Check that noise was added assert not torch.allclose(noisy_graphs[0].pos, data1.pos) assert not torch.allclose(noisy_graphs[0].hidden_state[0], data1.hidden_state[0]) print("Noise was added to pos and hidden_state.") - + # Check hidden state logic # The first N_measurements graphs (for the first original graph) should have the same hidden state first_hidden_state_set = noisy_graphs[0].hidden_state for i in range(1, N_measurements): assert torch.allclose(noisy_graphs[i].hidden_state[0], first_hidden_state_set[0]) assert torch.allclose(noisy_graphs[i].hidden_state[1], first_hidden_state_set[1]) - + # But their positions should be different assert not torch.allclose(noisy_graphs[0].pos, noisy_graphs[1].pos) # The (N_measurements+1)-th graph should have a different hidden state (from the second hidden measurement) next_hidden_state_set = noisy_graphs[N_measurements].hidden_state assert not torch.allclose(next_hidden_state_set[0], first_hidden_state_set[0]) - + print("Hidden state noise logic seems correct.") print("Test passed!") if __name__ == "__main__": - run_test() \ No newline at end of file + run_test() diff --git a/src/jamun/model/pooling.py b/src/jamun/model/pooling.py index 0e90cf9..831659c 100644 --- a/src/jamun/model/pooling.py +++ b/src/jamun/model/pooling.py @@ -3,60 +3,63 @@ Lightning modules for converting node attributes between spatial and temporal representations. """ -import torch -import torch_geometric import pytorch_lightning as pl +import torch from e3tools.nn import LayerNorm + class SpatialToTemporalNodeAttr(pl.LightningModule): """ - Lightning module to transfer node attributes from spatial nodes to temporal nodes + Lightning module to transfer node attributes from spatial nodes to temporal nodes by repeating first temporal feature. """ - + def __init__(self, irreps_out): super().__init__() self.irreps_out = irreps_out self.layer_norm = LayerNorm(irreps_out) - + def forward(self, spatial_node_attr_temporal, temporal_batch): """ Transfer node attributes from spatial nodes to temporal nodes by repeating first temporal feature. Takes the first temporal feature (t=0) and repeats it T times for each spatial node. - + Args: spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs - + Returns: torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] """ num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape num_temporal_graphs = temporal_batch.num_graphs - + # Verify consistency - assert num_spatial_nodes == num_temporal_graphs, \ + assert num_spatial_nodes == num_temporal_graphs, ( f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" - + ) + # Verify temporal length consistency expected_temporal_nodes = temporal_batch.pos.shape[0] expected_total_nodes = num_spatial_nodes * temporal_length - assert expected_total_nodes == expected_temporal_nodes, \ + assert expected_total_nodes == expected_temporal_nodes, ( f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" - + ) + # Extract first temporal feature (t=0) and repeat it T times for each spatial node first_temporal_features = spatial_node_attr_temporal[:, 0, :] # [N, attr_dim] - + # Repeat each spatial node's first temporal feature T times temporal_node_attr = first_temporal_features.repeat_interleave(temporal_length, dim=0) # [N*T, attr_dim] - + # Verify the output shape matches the temporal batch - assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + assert temporal_node_attr.shape[0] == expected_temporal_nodes, ( f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" - + ) + # Apply layer normalization before returning temporal_node_attr = self.layer_norm(temporal_node_attr) - + return temporal_node_attr @@ -65,48 +68,49 @@ class TemporalToSpatialNodeAttr(pl.LightningModule): Lightning module to convert temporal node attributes back to spatial node attributes. Takes the first temporal node attribute from each temporal graph. """ - + def __init__(self, irreps_out): super().__init__() self.irreps_out = irreps_out self.layer_norm = LayerNorm(irreps_out) - + def forward(self, temporal_node_attr, temporal_batch): """ Convert temporal node attributes back to spatial node attributes. Takes the first temporal node attribute from each temporal graph. - + Args: temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs - + Returns: torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] """ num_temporal_graphs = temporal_batch.num_graphs attr_dim = temporal_node_attr.shape[1] - + # Extract the first node attribute from each temporal graph spatial_node_attr = [] - + for graph_idx in range(num_temporal_graphs): # Get the node range for this temporal graph start_idx = temporal_batch.ptr[graph_idx] - + # The 0th node of each temporal graph is at the start of its range first_node_attr = temporal_node_attr[start_idx] spatial_node_attr.append(first_node_attr) - + # Stack to create spatial node attribute tensor spatial_node_attr = torch.stack(spatial_node_attr) - + # Verify output shape - assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), ( f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" - + ) + # Apply layer normalization before returning spatial_node_attr = self.layer_norm(spatial_node_attr) - + return spatial_node_attr @@ -115,50 +119,55 @@ class TemporalToSpatialNodeAttrMean(pl.LightningModule): Lightning module to convert temporal node attributes back to spatial node attributes by averaging. Takes the mean of all temporal node attributes for each temporal graph. """ - + def __init__(self, irreps_out): super().__init__() self.irreps_out = irreps_out self.layer_norm = LayerNorm(irreps_out) - + def forward(self, temporal_node_attr, temporal_batch): """ Convert temporal node attributes back to spatial node attributes by averaging. Takes the mean of all temporal node attributes for each temporal graph. - + Args: temporal_node_attr (torch.Tensor): Node attributes for temporal nodes [N_temporal, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs - + Returns: torch.Tensor: Node attributes for spatial nodes [N_spatial, attr_dim] """ num_temporal_graphs = temporal_batch.num_graphs attr_dim = temporal_node_attr.shape[1] - + # Extract the mean node attributes from each temporal graph spatial_node_attr = [] - + for graph_idx in range(num_temporal_graphs): # Get the node range for this temporal graph start_idx = temporal_batch.ptr[graph_idx] - end_idx = temporal_batch.ptr[graph_idx + 1] if graph_idx + 1 < len(temporal_batch.ptr) else len(temporal_node_attr) - + end_idx = ( + temporal_batch.ptr[graph_idx + 1] + if graph_idx + 1 < len(temporal_batch.ptr) + else len(temporal_node_attr) + ) + # Take the mean of all temporal nodes for this spatial node temporal_nodes_attr = temporal_node_attr[start_idx:end_idx] # [temporal_length, attr_dim] mean_node_attr = temporal_nodes_attr.mean(dim=0) # [attr_dim] spatial_node_attr.append(mean_node_attr) - + # Stack to create spatial node attribute tensor spatial_node_attr = torch.stack(spatial_node_attr) - + # Verify output shape - assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), \ + assert spatial_node_attr.shape == (num_temporal_graphs, attr_dim), ( f"Output shape mismatch: {spatial_node_attr.shape} vs expected ({num_temporal_graphs}, {attr_dim})" - + ) + # Apply layer normalization before returning spatial_node_attr = self.layer_norm(spatial_node_attr) - + return spatial_node_attr @@ -167,48 +176,51 @@ class SpatialTemporalToTemporalNodeAttr(pl.LightningModule): Lightning module to convert spatial node attributes arranged temporally to temporal node attributes. Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. """ - + def __init__(self, irreps_out): super().__init__() self.irreps_out = irreps_out self.layer_norm = LayerNorm(irreps_out) - + def forward(self, spatial_node_attr_temporal, temporal_batch): """ Convert spatial node attributes arranged temporally to temporal node attributes. Converts from [N, T, features] to [NT, features] with correct temporal graph ordering. - + Args: spatial_node_attr_temporal (torch.Tensor): Node attributes [N_spatial, T, attr_dim] temporal_batch (torch_geometric.data.Batch): Batch of temporal graphs for validation - + Returns: torch.Tensor: Node attributes for temporal nodes [N_temporal, attr_dim] """ num_spatial_nodes, temporal_length, attr_dim = spatial_node_attr_temporal.shape num_temporal_graphs = temporal_batch.num_graphs - + # Verify consistency with temporal batch - assert num_spatial_nodes == num_temporal_graphs, \ + assert num_spatial_nodes == num_temporal_graphs, ( f"Mismatch: {num_spatial_nodes} spatial nodes vs {num_temporal_graphs} temporal graphs" - + ) + # Verify temporal length consistency expected_temporal_nodes = temporal_batch.pos.shape[0] expected_total_nodes = num_spatial_nodes * temporal_length - assert expected_total_nodes == expected_temporal_nodes, \ + assert expected_total_nodes == expected_temporal_nodes, ( f"Temporal length mismatch: {expected_total_nodes} vs {expected_temporal_nodes}" - + ) + # Reshape to match temporal graph ordering: [N, T, features] -> [N*T, features] # Temporal graph arranges nodes as: [node0_t0, node0_t1, ..., node0_tT-1, node1_t0, ...] temporal_node_attr = spatial_node_attr_temporal.reshape(num_spatial_nodes * temporal_length, attr_dim) - + # Verify the output shape matches the temporal batch - assert temporal_node_attr.shape[0] == expected_temporal_nodes, \ + assert temporal_node_attr.shape[0] == expected_temporal_nodes, ( f"Output shape mismatch: {temporal_node_attr.shape[0]} vs expected {expected_temporal_nodes}" - + ) + # Apply layer normalization before returning temporal_node_attr = self.layer_norm(temporal_node_attr) - + return temporal_node_attr @@ -234,4 +246,4 @@ def temporal_to_spatial_node_attr_mean(temporal_node_attr, temporal_batch, irrep def spatial_temporal_to_temporal_node_attr(spatial_node_attr_temporal, temporal_batch, irreps_out): """Legacy function interface for backward compatibility.""" module = SpatialTemporalToTemporalNodeAttr(irreps_out) - return module(spatial_node_attr_temporal, temporal_batch) \ No newline at end of file + return module(spatial_node_attr_temporal, temporal_batch) diff --git a/src/jamun/sampling/_sampler.py b/src/jamun/sampling/_sampler.py index 16500bb..a803def 100644 --- a/src/jamun/sampling/_sampler.py +++ b/src/jamun/sampling/_sampler.py @@ -98,18 +98,18 @@ def sample( self.fabric.call("on_sample_end", sampler=self) + class SamplerMemory(Sampler): """A sampler for molecular dynamics simulations that uses memory.""" def sample( - self, + self, model, batch_sampler, num_batches: int, init_graphs: torch_geometric.data.Data, continue_chain: bool = False, ): - self.fabric.launch() self.fabric.setup(model) model.eval() @@ -150,4 +150,4 @@ def sample( self.fabric.call("on_after_sample_batch", sample=samples, sampler=self) self.fabric.log("sampler/global_step", batch_idx) - self.fabric.call("on_sample_end", sampler=self) \ No newline at end of file + self.fabric.call("on_sample_end", sampler=self) diff --git a/src/jamun/sampling/mcmc/__init__.py b/src/jamun/sampling/mcmc/__init__.py index 1558500..70a59c5 100644 --- a/src/jamun/sampling/mcmc/__init__.py +++ b/src/jamun/sampling/mcmc/__init__.py @@ -1 +1 @@ -from ._splitting import ABOBA, BAOAB, BAOAB_memory, ABOBA_memory +from ._splitting import ABOBA, BAOAB, ABOBA_memory, BAOAB_memory diff --git a/src/jamun/sampling/mcmc/_splitting.py b/src/jamun/sampling/mcmc/_splitting.py index b08e627..d53c252 100644 --- a/src/jamun/sampling/mcmc/_splitting.py +++ b/src/jamun/sampling/mcmc/_splitting.py @@ -5,7 +5,7 @@ import torch from torch import Tensor -from jamun.sampling.mcmc.functional import aboba, baoab, aboba_memory, baoab_memory +from jamun.sampling.mcmc.functional import aboba, aboba_memory, baoab, baoab_memory @dataclass @@ -61,7 +61,7 @@ def __call__(self, y: torch.Tensor, score_fn: Callable, **kwargs): @dataclass class ABOBA_memory(ABOBA): history_update_frequency: int = 1 - + def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): kwargs = dataclasses.asdict(self) | kwargs return aboba_memory(y=y, y_hist=y_hist, score_fn=score_fn, **kwargs) @@ -70,7 +70,7 @@ def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): @dataclass class BAOAB_memory(BAOAB): history_update_frequency: int = 1 - + def __call__(self, y: torch.Tensor, y_hist: list, score_fn: Callable, **kwargs): kwargs = dataclasses.asdict(self) | kwargs return baoab_memory(y=y, y_hist=y_hist, score_fn=score_fn, **kwargs) diff --git a/src/jamun/sampling/mcmc/functional/__init__.py b/src/jamun/sampling/mcmc/functional/__init__.py index a91fb53..0490fe6 100644 --- a/src/jamun/sampling/mcmc/functional/__init__.py +++ b/src/jamun/sampling/mcmc/functional/__init__.py @@ -1 +1 @@ -from ._splitting import aboba, baoab, aboba_memory, baoab_memory +from ._splitting import aboba, aboba_memory, baoab, baoab_memory diff --git a/src/jamun/sampling/mcmc/functional/_splitting.py b/src/jamun/sampling/mcmc/functional/_splitting.py index 4b9707f..e5d126e 100644 --- a/src/jamun/sampling/mcmc/functional/_splitting.py +++ b/src/jamun/sampling/mcmc/functional/_splitting.py @@ -1,7 +1,7 @@ import logging import math -from typing import Callable, Optional, Tuple, Union -from copy import deepcopy +from collections.abc import Callable + import torch from tqdm.auto import tqdm @@ -26,7 +26,7 @@ def initialize_velocity(v_init: str | torch.Tensor, y: torch.Tensor, u: float) - def create_score_fn(score_fn: Callable, inverse_temperature: float, score_fn_clip: float | None) -> Callable: """Create a score function that is clipped and scaled by the inverse temperature.""" - def score_fn_processed(y: torch.Tensor, *args, **kwargs) -> Tuple[torch.Tensor, torch.Tensor]: + def score_fn_processed(y: torch.Tensor, *args, **kwargs) -> tuple[torch.Tensor, torch.Tensor]: """Score function clipped and scaled by the inverse temperature.""" orig_score = score_fn(y, *args, **kwargs).to(dtype=y.dtype) # Clip the score by norm. @@ -183,7 +183,7 @@ def aboba_memory( y_hist: list, score_fn: Callable, steps: int, - v_init: Union[str, torch.Tensor] = "zero", + v_init: str | torch.Tensor = "zero", save_trajectory=False, save_every_n_steps=1, burn_in_steps=0, @@ -194,9 +194,9 @@ def aboba_memory( friction: float = 1.0, M: float = 1.0, inverse_temperature: float = 1.0, - score_fn_clip: Optional[float] = None, - cleanup: Optional[bool] = None, - sigma: Optional[float] = None, + score_fn_clip: float | None = None, + cleanup: bool | None = None, + sigma: float | None = None, **_, ): """ABOBA splitting scheme that updates a state history.""" @@ -204,13 +204,13 @@ def aboba_memory( y_traj = [] if save_trajectory else None score_traj = [] y_hist_traj = [] - + # Initialize trajectory with initial state if y_traj is not None and i >= burn_in_steps: y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) score_traj.append(torch.zeros_like(y).detach().cpu() if cpu_offload else torch.zeros_like(y).detach()) y_hist_traj.append(list(y_hist)) - + u = pow(M, -1) zeta2 = math.sqrt(1 - math.exp(-2 * friction)) v = initialize_velocity(v_init=v_init, y=y, u=u) @@ -221,7 +221,7 @@ def aboba_memory( steps_iter = tqdm(steps_iter, leave=False, desc="ABOBA Memory") for i in steps_iter: - for j in range(1,history_update_frequency): + for j in range(1, history_update_frequency): # inner aboba loop for equilibration to conditional density p(y_t | y_hist) y_current = y.clone().detach() y = y + (delta / 2) * v @@ -240,7 +240,7 @@ def aboba_memory( if cleanup is not None and cleanup and sigma is not None: y_current = y.clone().detach() _, orig_score = score_fn_processed(y_current, y_hist=y_hist) - y_denoised_and_noised = y_current + (sigma**2)*orig_score + y_denoised_and_noised = y_current + (sigma**2) * orig_score y_hist.pop(-1) y_hist.insert(0, y_denoised_and_noised) y = y_denoised_and_noised @@ -249,7 +249,14 @@ def aboba_memory( y_hist.pop(-1) y_hist.insert(0, y_current) - return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj + return ( + y, + v, + y_hist, + torch.stack(y_traj) if y_traj else None, + torch.stack(score_traj) if score_traj else None, + y_hist_traj, + ) def baoab_memory( @@ -257,7 +264,7 @@ def baoab_memory( y_hist: list, score_fn: Callable, steps: int, - v_init: Union[str, torch.Tensor] = "zero", + v_init: str | torch.Tensor = "zero", save_trajectory=False, save_every_n_steps=1, burn_in_steps=0, @@ -268,9 +275,9 @@ def baoab_memory( friction: float = 1.0, M: float = 1.0, inverse_temperature: float = 1.0, - score_fn_clip: Optional[float] = None, - cleanup: Optional[bool] = None, - sigma: Optional[float] = None, + score_fn_clip: float | None = None, + cleanup: bool | None = None, + sigma: float | None = None, **_, ): """BAOAB splitting scheme that updates a state history.""" @@ -278,7 +285,7 @@ def baoab_memory( y_traj = [] if save_trajectory else None score_traj = [] y_hist_traj = [] - + u = pow(M, -1) zeta2 = math.sqrt(1 - math.exp(-2 * friction)) v = initialize_velocity(v_init=v_init, y=y, u=u) @@ -289,7 +296,7 @@ def baoab_memory( steps_iter = tqdm(steps_iter, leave=False, desc="BAOAB Memory") psi, orig_score = score_fn_processed(y, y_hist=y_hist) - + # Initialize trajectory with initial state if y_traj is not None and i >= burn_in_steps: y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) @@ -298,30 +305,39 @@ def baoab_memory( for i in steps_iter: # print(f"Equilibrating to conditional density p(y_t | y_hist) for {history_update_frequency} steps...") - for j in range(1,history_update_frequency): + for j in range(1, history_update_frequency): # inner baoab loop for equilibration to conditional density p(y_t | y_hist) y_current = y.clone().detach() - v = v + u * (delta / 2) * psi # update with previous psi - y = y + (delta / 2) * v # update with previous v + v = v + u * (delta / 2) * psi # update with previous psi + y = y + (delta / 2) * v # update with previous v R = torch.randn_like(y) vhat = math.exp(-friction) * v + zeta2 * math.sqrt(u) * R y = y + (delta / 2) * vhat psi, orig_score = score_fn_processed(y, y_hist=y_hist) v = vhat + (delta / 2) * psi - + if cleanup is not None and cleanup and sigma is not None: y_current = y.clone().detach() _, orig_score = score_fn_processed(y_current, y_hist=y_hist) - y_denoised_and_noised = y_current + (sigma**2)*orig_score + sigma*torch.randn_like(y_current) # clean and add noise + y_denoised_and_noised = ( + y_current + (sigma**2) * orig_score + sigma * torch.randn_like(y_current) + ) # clean and add noise y_hist.pop(-1) y_hist.insert(0, y_denoised_and_noised) y = y_denoised_and_noised else: - y_hist.pop(-1) # remove the last element of the history - y_hist.insert(0, y_current) # present point is the first element of the history + y_hist.pop(-1) # remove the last element of the history + y_hist.insert(0, y_current) # present point is the first element of the history if save_trajectory and ((i % save_every_n_steps) == 0) and (i >= burn_in_steps): y_traj.append(y.detach().cpu() if cpu_offload else y.detach()) score_traj.append(orig_score.detach().cpu() if cpu_offload else orig_score.detach()) y_hist_traj.append(list(y_hist)) - return y, v, y_hist, torch.stack(y_traj) if y_traj else None, torch.stack(score_traj) if score_traj else None, y_hist_traj + return ( + y, + v, + y_hist, + torch.stack(y_traj) if y_traj else None, + torch.stack(score_traj) if score_traj else None, + y_hist_traj, + ) diff --git a/src/jamun/sampling/walkjump/_single_measurement.py b/src/jamun/sampling/walkjump/_single_measurement.py index d846cab..711b39a 100644 --- a/src/jamun/sampling/walkjump/_single_measurement.py +++ b/src/jamun/sampling/walkjump/_single_measurement.py @@ -1,6 +1,5 @@ import torch from torch import Tensor -from typing import Optional from tqdm.auto import tqdm @@ -91,12 +90,7 @@ def sample( class SingleMeasurementSamplerMemory: """Single Measurement Walk-Jump Sampler.""" - def __init__( - self, - mcmc, - sigma: float, - y_init_distribution: Optional[torch.distributions.Distribution] = None - ): + def __init__(self, mcmc, sigma: float, y_init_distribution: torch.distributions.Distribution | None = None): self.mcmc = mcmc self.sigma = float(sigma) self.y_init_distribution = y_init_distribution @@ -104,10 +98,10 @@ def __init__( def walk( self, model, - batch_size: Optional[int] = None, - y_init: Optional[torch.Tensor] = None, + batch_size: int | None = None, + y_init: torch.Tensor | None = None, v_init: str | Tensor = "gaussian", - y_hist_init: Optional[list] = None, + y_hist_init: list | None = None, ): if y_init is None: if self.y_init_distribution is None: @@ -115,23 +109,37 @@ def walk( y_init = self.y_init_distribution.sample(sample_shape=(batch_size,)).to(model.device) if y_hist_init is None: raise RuntimeError("y_hist_init must be supplied") - y, v, y_hist,y_traj, score_traj, y_hist_traj = self.mcmc(y_init, y_hist_init, lambda y, y_hist: model.score(y, y_hist, self.sigma), \ - v_init=v_init, cleanup=True, sigma=self.sigma) + y, v, y_hist, y_traj, score_traj, y_hist_traj = self.mcmc( + y_init, + y_hist_init, + lambda y, y_hist: model.score(y, y_hist, self.sigma), + v_init=v_init, + cleanup=True, + sigma=self.sigma, + ) if y_traj is not None: t_traj = torch.ones(y_traj.size(0), device=y_traj.device, dtype=int) else: t_traj = None - return {"y": y, "v": v, "y_hist": y_hist, "y_traj": y_traj, "t_traj": t_traj, "score_traj": score_traj, "y_hist_traj": y_hist_traj} + return { + "y": y, + "v": v, + "y_hist": y_hist, + "y_traj": y_traj, + "t_traj": t_traj, + "score_traj": score_traj, + "y_hist_traj": y_hist_traj, + } def walk_jump( self, model, - batch_size: Optional[int] = None, - y_init: Optional[torch.Tensor] = None, + batch_size: int | None = None, + y_init: torch.Tensor | None = None, v_init: str | Tensor = "gaussian", - y_hist_init: Optional[list] = None, + y_hist_init: list | None = None, ): out = self.walk( model, @@ -140,14 +148,22 @@ def walk_jump( v_init=v_init, y_hist_init=y_hist_init, ) - y, v, y_hist, y_traj, t_traj, score_traj, y_hist_traj = out["y"], out["v"], out["y_hist"], out["y_traj"], out["t_traj"], out["score_traj"], out["y_hist_traj"] + y, v, y_hist, y_traj, t_traj, score_traj, y_hist_traj = ( + out["y"], + out["v"], + out["y_hist"], + out["y_traj"], + out["t_traj"], + out["score_traj"], + out["y_hist_traj"], + ) xhat = model.xhat(y, y_hist, sigma=self.sigma) if y_traj is not None: xhat_traj = torch.stack( [ - model.xhat(y_traj[i, :].to(model.device), y_hist_traj[i], sigma=self.sigma) + model.xhat(y_traj[i, :].to(model.device), y_hist_traj[i], sigma=self.sigma) for i in tqdm(range(y_traj.size(0)), leave=False, desc="Jump") ], dim=0, @@ -170,11 +186,11 @@ def walk_jump( def sample( self, model, - batch_size: Optional[int] = None, - y_init: Optional[torch.Tensor] = None, + batch_size: int | None = None, + y_init: torch.Tensor | None = None, v_init: str | Tensor = "gaussian", - y_hist_init: Optional[list] = None, + y_hist_init: list | None = None, ): out = self.walk_jump(model, batch_size=batch_size, y_init=y_init, v_init=v_init, y_hist_init=y_hist_init) out["sample"] = out["xhat"] - return out \ No newline at end of file + return out diff --git a/src/jamun/utils/__init__.py b/src/jamun/utils/__init__.py index dcb87ad..0790d7f 100644 --- a/src/jamun/utils/__init__.py +++ b/src/jamun/utils/__init__.py @@ -1,6 +1,10 @@ from .align import align_A_to_B, align_A_to_B_batched, align_A_to_B_batched_f, find_rigid_alignment from .atom_graphs import to_atom_graphs -from .average_squared_distance import compute_average_squared_distance, compute_average_squared_distance_from_datasets, compute_temporal_average_squared_distance_from_datasets +from .average_squared_distance import ( + compute_average_squared_distance, + compute_average_squared_distance_from_datasets, + compute_temporal_average_squared_distance_from_datasets, +) from .checkpoint import find_checkpoint, find_checkpoint_directory, get_run_path_for_wandb_run, get_wandb_run_config from .data_with_residue_info import DataWithResidueInformation from .dist_log import dist_log, wandb_dist_log @@ -12,7 +16,6 @@ get_side_chain_torsion_idxs, one_k_encoding, ) - from .mdtraj import coordinates_to_trajectories, save_pdb from .mean_center import mean_center, mean_center_f from .plot import animate_trajectory_with_py3Dmol, plot_molecules_with_py3Dmol diff --git a/src/jamun/utils/_normalizations.py b/src/jamun/utils/_normalizations.py index 43fc66a..7e00bcf 100644 --- a/src/jamun/utils/_normalizations.py +++ b/src/jamun/utils/_normalizations.py @@ -2,27 +2,26 @@ Normalization utilities for jamun models. """ -from typing import Tuple import torch def normalization_factors( - sigma: float, - average_squared_distance: float, - normalization_type: str = "JAMUN", + sigma: float, + average_squared_distance: float, + normalization_type: str = "JAMUN", sigma_data: float = None, - D: int = 3 -) -> Tuple[float, float, float, float]: + D: int = 3, +) -> tuple[float, float, float, float]: """ Compute normalization factors for the input and output. - + Args: sigma: Noise level average_squared_distance: Average squared distance from the dataset normalization_type: Type of normalization ("JAMUN", "EDM", or None) sigma_data: Sigma data parameter (only used for EDM normalization) D: Dimensionality (default: 3) - + Returns: Tuple of (c_in, c_skip, c_out, c_noise) normalization factors """ @@ -50,4 +49,4 @@ def normalization_factors( c_noise = torch.log(sigma) / 4 return c_in, c_skip, c_out, c_noise - raise ValueError(f"Unknown normalization type: {normalization_type}") \ No newline at end of file + raise ValueError(f"Unknown normalization type: {normalization_type}") diff --git a/src/jamun/utils/average_squared_distance.py b/src/jamun/utils/average_squared_distance.py index 5f9c616..b11d287 100644 --- a/src/jamun/utils/average_squared_distance.py +++ b/src/jamun/utils/average_squared_distance.py @@ -75,26 +75,24 @@ def compute_average_squared_distance_from_datasets( def compute_temporal_average_squared_distance_from_datasets( - datasets, - num_samples: int = 100, - verbose: bool = False + datasets, num_samples: int = 100, verbose: bool = False ) -> float: """ Compute average squared distance between neighboring vertices in temporal graphs. - + Args: datasets: Collection of datasets containing spatial graphs with hidden states num_samples: Number of samples to use for estimation verbose: Whether to print verbose output - + Returns: float: Average squared distance between temporal neighbors """ from jamun.model.arch.spatiotemporal import spatial_to_temporal_graphs - + avg_sq_dists = [] num_graphs = 0 - + # Follow pattern from existing functions in this module for item in datasets: if num_graphs >= num_samples: @@ -114,11 +112,10 @@ def compute_temporal_average_squared_distance_from_datasets( num_graphs += 1 mean_avg_sq_dist = sum(avg_sq_dists) / num_graphs - if verbose: print(f"Total graphs processed: {num_graphs}") print(f"Total temporal graphs processed: {len(avg_sq_dists)}") print(f"Mean average squared distance between temporal nodes: {mean_avg_sq_dist:.6f}") print(f"Standard deviation: {np.std(avg_sq_dists):.6f}") - + return float(mean_avg_sq_dist) diff --git a/src/jamun/utils/checkpoint.py b/src/jamun/utils/checkpoint.py index 73911dd..8a76084 100644 --- a/src/jamun/utils/checkpoint.py +++ b/src/jamun/utils/checkpoint.py @@ -16,7 +16,7 @@ def get_wandb_run_config(wandb_run_path: str) -> dict[str, Any]: run = wandb.Api().run(wandb_run_path) py_logger = logging.getLogger("jamun") py_logger.info(f"Loading checkpoint corresponding to wandb run {run.name} at {run.url}") - key = next(iter(run.config)) # the key might be named differently in the future + key = next(iter(run.config)) # the key might be named differently in the future return run.config[key] diff --git a/src/jamun/utils/data_with_residue_info.py b/src/jamun/utils/data_with_residue_info.py index 1c136e4..d85708a 100644 --- a/src/jamun/utils/data_with_residue_info.py +++ b/src/jamun/utils/data_with_residue_info.py @@ -1,6 +1,8 @@ +from typing import Any + import torch import torch_geometric -from typing import Any + class DataWithResidueInformation(torch_geometric.data.Data): """Graph with residue-level information.""" @@ -13,7 +15,8 @@ class DataWithResidueInformation(torch_geometric.data.Data): residue_index: torch.Tensor # batched version of residue_sequence_index num_residues: int loss_weight: float - hidden_state: Any + hidden_state: Any + def __inc__(self, key, value, *args, **kwargs): del value, args, kwargs if key in [ @@ -24,7 +27,7 @@ def __inc__(self, key, value, *args, **kwargs): "residue_sequence_index", "num_residues", "loss_weight", - "hidden_state" + "hidden_state", ]: return 0 if key in ["edge_index", "bonded_edge_index"]: diff --git a/src/jamun/utils/inspect_pretrained.py b/src/jamun/utils/inspect_pretrained.py index f95a811..ec056c4 100644 --- a/src/jamun/utils/inspect_pretrained.py +++ b/src/jamun/utils/inspect_pretrained.py @@ -6,27 +6,26 @@ import argparse import sys from pathlib import Path -from typing import Optional, Dict, Any -import torch import hydra +import torch from omegaconf import DictConfig +from jamun.utils.checkpoint import find_checkpoint from jamun.utils.pretrained import ( - load_checkpoint_state_dict, - load_pretrained_model_from_checkpoint, + check_model_compatibility, extract_module_from_model, inspect_model_structure, - check_model_compatibility + load_checkpoint_state_dict, + load_pretrained_model_from_checkpoint, ) -from jamun.utils.checkpoint import find_checkpoint def print_checkpoint_structure(checkpoint_path: str, max_depth: int = 2, use_model_loading: bool = True): """Print the structure of a checkpoint file.""" print(f"\nšŸ“ Checkpoint structure: {checkpoint_path}") print("=" * 60) - + if use_model_loading: # Try to load as a complete model first try: @@ -38,24 +37,24 @@ def print_checkpoint_structure(checkpoint_path: str, max_depth: int = 2, use_mod print("āš ļø Could not load as complete model, falling back to state_dict inspection") except Exception as e: print(f"āš ļø Model loading failed ({e}), falling back to state_dict inspection") - + # Fallback to state_dict inspection try: state_dict = load_checkpoint_state_dict(checkpoint_path) - + # Group keys by their prefixes - key_groups: Dict[str, list] = {} + key_groups: dict[str, list] = {} for key in state_dict.keys(): - parts = key.split('.') + parts = key.split(".") if len(parts) >= max_depth: - prefix = '.'.join(parts[:max_depth]) + prefix = ".".join(parts[:max_depth]) else: prefix = key - + if prefix not in key_groups: key_groups[prefix] = [] key_groups[prefix].append(key) - + # Print grouped structure for prefix in sorted(key_groups.keys()): keys = key_groups[prefix] @@ -67,50 +66,46 @@ def print_checkpoint_structure(checkpoint_path: str, max_depth: int = 2, use_mod # Group of parameters total_params = sum(state_dict[key].numel() for key in keys) print(f" {prefix}.* : {len(keys)} parameters ({total_params:,} total elements)") - + # Show a few example keys if len(keys) <= 5: for key in sorted(keys)[:5]: tensor = state_dict[key] - sub_key = key[len(prefix)+1:] if key.startswith(prefix + '.') else key + sub_key = key[len(prefix) + 1 :] if key.startswith(prefix + ".") else key print(f" ā””ā”€ {sub_key}: {list(tensor.shape)}") else: for key in sorted(keys)[:3]: tensor = state_dict[key] - sub_key = key[len(prefix)+1:] if key.startswith(prefix + '.') else key + sub_key = key[len(prefix) + 1 :] if key.startswith(prefix + ".") else key print(f" ā””ā”€ {sub_key}: {list(tensor.shape)}") - print(f" ā””ā”€ ... and {len(keys)-3} more") - + print(f" ā””ā”€ ... and {len(keys) - 3} more") + total_params = sum(tensor.numel() for tensor in state_dict.values()) print(f"\nšŸ“Š Total parameters: {total_params:,}") - + except Exception as e: print(f"āŒ Error loading checkpoint: {e}") -def check_compatibility_with_config( - checkpoint_path: str, - config_path: str, - module_path: Optional[str] = None -): +def check_compatibility_with_config(checkpoint_path: str, config_path: str, module_path: str | None = None): """Check if a checkpoint is compatible and can be loaded.""" - print(f"\nšŸ” Checking compatibility...") + print("\nšŸ” Checking compatibility...") print(f"Checkpoint: {checkpoint_path}") print(f"Config: {config_path}") if module_path: print(f"Module path: {module_path}") print("=" * 60) - + try: # Check if checkpoint can be loaded as a model compatibility = check_model_compatibility(checkpoint_path=checkpoint_path) - - if compatibility['loadable']: + + if compatibility["loadable"]: print("āœ… Checkpoint can be loaded as a complete model!") print(f"Model class: {compatibility['model_class'].__name__}") print(f"Total parameters: {compatibility['total_params']:,}") print(f"Trainable parameters: {compatibility['trainable_params']:,}") - + # If a module path is specified, try to extract it if module_path: try: @@ -124,7 +119,7 @@ def check_compatibility_with_config( print(f"āŒ Module at path '{module_path}' not found in model") except Exception as e: print(f"āŒ Error extracting module: {e}") - + # Try loading with the config if provided if config_path and Path(config_path).exists(): try: @@ -132,63 +127,61 @@ def check_compatibility_with_config( cfg = hydra.compose(config_name=Path(config_path).stem) if isinstance(cfg, DictConfig): target_model = hydra.utils.instantiate(cfg) - if isinstance(target_model, compatibility['model_class']): + if isinstance(target_model, compatibility["model_class"]): print("āœ… Config model class matches checkpoint model class!") else: - print(f"āš ļø Config model class ({type(target_model).__name__}) differs from checkpoint ({compatibility['model_class'].__name__})") + print( + f"āš ļø Config model class ({type(target_model).__name__}) differs from checkpoint ({compatibility['model_class'].__name__})" + ) except Exception as e: print(f"āš ļø Could not instantiate model from config: {e}") else: print("āŒ Checkpoint cannot be loaded as a model") - if 'error' in compatibility: + if "error" in compatibility: print(f"Error: {compatibility['error']}") - + except Exception as e: print(f"āŒ Error checking compatibility: {e}") -def extract_and_save_module( - checkpoint_path: str, - module_path: str, - output_path: str -): +def extract_and_save_module(checkpoint_path: str, module_path: str, output_path: str): """Extract a specific module from a checkpoint and save it separately.""" print(f"\nšŸ“¤ Extracting module: {module_path}") print(f"From: {checkpoint_path}") print(f"To: {output_path}") print("=" * 60) - + try: # Load the full model model = load_pretrained_model_from_checkpoint(checkpoint_path=checkpoint_path) if model is None: print("āŒ Could not load model from checkpoint") return - + # Extract the specific module extracted_module = extract_module_from_model(model, module_path) if extracted_module is None: print(f"āŒ Module at path '{module_path}' not found in model") return - + # Save the extracted module # We'll save it as a state dict that can be loaded later module_state_dict = extracted_module.state_dict() - + save_data = { - 'state_dict': module_state_dict, - 'module_class': type(extracted_module).__name__, - 'module_path': module_path, - 'source_checkpoint': checkpoint_path + "state_dict": module_state_dict, + "module_class": type(extracted_module).__name__, + "module_path": module_path, + "source_checkpoint": checkpoint_path, } - + torch.save(save_data, output_path) - + param_count = sum(tensor.numel() for tensor in module_state_dict.values()) print(f"āœ… Extracted {len(module_state_dict)} parameters ({param_count:,} elements)") print(f"Module class: {type(extracted_module).__name__}") print(f"Saved to: {output_path}") - + except Exception as e: print(f"āŒ Error extracting module: {e}") @@ -196,49 +189,47 @@ def extract_and_save_module( def main(): parser = argparse.ArgumentParser(description="Inspect pretrained checkpoints for spatiotemporal models") parser.add_argument("command", choices=["inspect", "check", "extract"], help="Command to run") - + # Common arguments parser.add_argument("--checkpoint", type=str, help="Path to checkpoint file") parser.add_argument("--wandb_run", type=str, help="WandB run path (e.g., user/project/run_id)") - parser.add_argument("--checkpoint_type", type=str, default="best_so_far", - help="Type of checkpoint to load") - + parser.add_argument("--checkpoint_type", type=str, default="best_so_far", help="Type of checkpoint to load") + # Inspect command arguments - parser.add_argument("--max_depth", type=int, default=2, - help="Maximum depth for structure inspection") - + parser.add_argument("--max_depth", type=int, default=2, help="Maximum depth for structure inspection") + # Check command arguments parser.add_argument("--config", type=str, help="Path to model config file") - parser.add_argument("--module_path", type=str, - help="Module path to extract (e.g., 'conditioner.spatiotemporal_model.spatial_module')") - + parser.add_argument( + "--module_path", + type=str, + help="Module path to extract (e.g., 'conditioner.spatiotemporal_model.spatial_module')", + ) + # Extract command arguments parser.add_argument("--output", type=str, help="Output path for extracted module") - + args = parser.parse_args() - + # Get checkpoint path if args.checkpoint: checkpoint_path = args.checkpoint elif args.wandb_run: - checkpoint_path = find_checkpoint( - wandb_train_run_path=args.wandb_run, - checkpoint_type=args.checkpoint_type - ) + checkpoint_path = find_checkpoint(wandb_train_run_path=args.wandb_run, checkpoint_type=args.checkpoint_type) else: print("āŒ Must specify either --checkpoint or --wandb_run") sys.exit(1) - + # Execute command if args.command == "inspect": print_checkpoint_structure(checkpoint_path, args.max_depth) - + elif args.command == "check": if not args.config: print("āŒ --config is required for check command") sys.exit(1) check_compatibility_with_config(checkpoint_path, args.config, args.module_path) - + elif args.command == "extract": if not args.module_path or not args.output: print("āŒ --module_path and --output are required for extract command") @@ -247,4 +238,4 @@ def main(): if __name__ == "__main__": - main() \ No newline at end of file + main() diff --git a/src/jamun/utils/pretrained.py b/src/jamun/utils/pretrained.py index 57788cf..e40b862 100644 --- a/src/jamun/utils/pretrained.py +++ b/src/jamun/utils/pretrained.py @@ -2,139 +2,137 @@ import logging import os +from typing import Any + +import lightning.pytorch as pl import torch import torch.nn as nn -import lightning.pytorch as pl -from typing import Dict, Optional, Union, Any -from pathlib import Path from jamun.utils.checkpoint import find_checkpoint py_logger = logging.getLogger("jamun") -def load_checkpoint_state_dict(checkpoint_path: str) -> Dict[str, torch.Tensor]: +def load_checkpoint_state_dict(checkpoint_path: str) -> dict[str, torch.Tensor]: """Load state dict from a checkpoint file.""" if not os.path.exists(checkpoint_path): raise FileNotFoundError(f"Checkpoint file not found: {checkpoint_path}") - - checkpoint = torch.load(checkpoint_path, map_location='cpu') - + + checkpoint = torch.load(checkpoint_path, map_location="cpu") + # Handle different checkpoint formats - if 'state_dict' in checkpoint: - return checkpoint['state_dict'] - elif isinstance(checkpoint, dict) and any(k.startswith('model.') for k in checkpoint.keys()): + if "state_dict" in checkpoint: + return checkpoint["state_dict"] + elif isinstance(checkpoint, dict) and any(k.startswith("model.") for k in checkpoint.keys()): return checkpoint else: raise ValueError(f"Unrecognized checkpoint format in {checkpoint_path}") def load_pretrained_model_from_checkpoint( - checkpoint_path: Optional[str] = None, - wandb_run_path: Optional[str] = None, + checkpoint_path: str | None = None, + wandb_run_path: str | None = None, checkpoint_type: str = "best_so_far", - model_class: Optional[type] = None -) -> Optional[pl.LightningModule]: + model_class: type | None = None, +) -> pl.LightningModule | None: """ Load an entire pretrained model from checkpoint. - + Args: checkpoint_path: Direct path to checkpoint file (mutually exclusive with wandb_run_path) wandb_run_path: WandB run path to find checkpoint (mutually exclusive with checkpoint_path) checkpoint_type: Type of checkpoint to load ("best_so_far", "last", etc.) model_class: Optional model class to use for loading (if checkpoint doesn't contain class info) - + Returns: Loaded model or None if loading failed """ if not checkpoint_path and not wandb_run_path: py_logger.warning("No checkpoint path or wandb run path provided, skipping pretrained loading") return None - + if checkpoint_path and wandb_run_path: raise ValueError("Cannot specify both checkpoint_path and wandb_run_path") - + try: # Find the checkpoint file if wandb_run_path: - checkpoint_path = find_checkpoint( - wandb_train_run_path=wandb_run_path, - checkpoint_type=checkpoint_type - ) - + checkpoint_path = find_checkpoint(wandb_train_run_path=wandb_run_path, checkpoint_type=checkpoint_type) + py_logger.info(f"Loading pretrained model from: {checkpoint_path}") - + # Load the entire model from checkpoint if model_class: model = model_class.load_from_checkpoint(checkpoint_path, strict=False) else: # Try to auto-detect model class from checkpoint - checkpoint = torch.load(checkpoint_path, map_location='cpu') - if 'hyper_parameters' in checkpoint and '_target_' in checkpoint['hyper_parameters']: + checkpoint = torch.load(checkpoint_path, map_location="cpu") + if "hyper_parameters" in checkpoint and "_target_" in checkpoint["hyper_parameters"]: # Try to import and use the model class from checkpoint import importlib - target = checkpoint['hyper_parameters']['_target_'] - module_path, class_name = target.rsplit('.', 1) + + target = checkpoint["hyper_parameters"]["_target_"] + module_path, class_name = target.rsplit(".", 1) module = importlib.import_module(module_path) model_class = getattr(module, class_name) model = model_class.load_from_checkpoint(checkpoint_path, strict=False) else: py_logger.error("Cannot determine model class from checkpoint and no model_class provided") return None - + py_logger.info(f"Successfully loaded pretrained model of type {type(model).__name__}") return model - + except Exception as e: py_logger.error(f"Error loading pretrained model: {e}") return None -def extract_module_from_model(model: pl.LightningModule, module_path: str) -> Optional[nn.Module]: +def extract_module_from_model(model: pl.LightningModule, module_path: str) -> nn.Module | None: """ Extract a specific module from a loaded model using dot notation. - + Args: model: Loaded PyTorch Lightning model module_path: Dot-separated path to module (e.g., "conditioner.spatiotemporal_model.spatial_module") - + Returns: Extracted module or None if not found """ try: current = model - for attr in module_path.split('.'): + for attr in module_path.split("."): if hasattr(current, attr): current = getattr(current, attr) else: py_logger.warning(f"Module path '{module_path}' not found in model") return None - + py_logger.info(f"Successfully extracted module at path: {module_path}") return current - + except Exception as e: py_logger.error(f"Error extracting module '{module_path}': {e}") return None def load_pretrained_module_from_checkpoint( - checkpoint_path: Optional[str] = None, - wandb_run_path: Optional[str] = None, + checkpoint_path: str | None = None, + wandb_run_path: str | None = None, checkpoint_type: str = "best_so_far", - module_path: Optional[str] = None, - model_class: Optional[type] = None -) -> Optional[nn.Module]: + module_path: str | None = None, + model_class: type | None = None, +) -> nn.Module | None: """ Load a specific module from a pretrained model checkpoint. - + Args: checkpoint_path: Direct path to checkpoint file - wandb_run_path: WandB run path to find checkpoint + wandb_run_path: WandB run path to find checkpoint checkpoint_type: Type of checkpoint to load module_path: Dot notation path to extract specific module (e.g., "conditioner.spatiotemporal_model.spatial_module") model_class: Optional model class for loading - + Returns: Extracted module or None if loading failed """ @@ -143,12 +141,12 @@ def load_pretrained_module_from_checkpoint( checkpoint_path=checkpoint_path, wandb_run_path=wandb_run_path, checkpoint_type=checkpoint_type, - model_class=model_class + model_class=model_class, ) - + if model is None: return None - + # Extract the specific module if path provided if module_path: return extract_module_from_model(model, module_path) @@ -161,22 +159,22 @@ def inspect_model_structure(model: pl.LightningModule, max_depth: int = 3) -> No """Print the structure of a loaded model.""" print(f"\nšŸ“ Model structure: {type(model).__name__}") print("=" * 60) - + def print_module_tree(module, prefix="", depth=0): if depth >= max_depth: return - + for name, child in module.named_children(): full_name = f"{prefix}.{name}" if prefix else name param_count = sum(p.numel() for p in child.parameters()) - + print(f"{' ' * depth}ā”œā”€ {name}: {type(child).__name__} ({param_count:,} params)") - + if depth < max_depth - 1: print_module_tree(child, full_name, depth + 1) - + print_module_tree(model) - + total_params = sum(p.numel() for p in model.parameters()) trainable_params = sum(p.numel() for p in model.parameters() if p.requires_grad) print(f"\nšŸ“Š Total parameters: {total_params:,}") @@ -184,52 +182,41 @@ def print_module_tree(module, prefix="", depth=0): def check_model_compatibility( - checkpoint_path: Optional[str] = None, - wandb_run_path: Optional[str] = None, + checkpoint_path: str | None = None, + wandb_run_path: str | None = None, checkpoint_type: str = "best_so_far", - expected_model_class: Optional[type] = None -) -> Dict[str, Any]: + expected_model_class: type | None = None, +) -> dict[str, Any]: """ Check if a checkpoint can be loaded and optionally verify model class. - + Returns: Dict with 'loadable', 'model_class', 'error' info """ try: if wandb_run_path: - checkpoint_path = find_checkpoint( - wandb_train_run_path=wandb_run_path, - checkpoint_type=checkpoint_type - ) - + checkpoint_path = find_checkpoint(wandb_train_run_path=wandb_run_path, checkpoint_type=checkpoint_type) + # Try loading the model model = load_pretrained_model_from_checkpoint(checkpoint_path=checkpoint_path) - + if model is None: - return { - 'loadable': False, - 'model_class': None, - 'error': 'Failed to load model from checkpoint' - } - + return {"loadable": False, "model_class": None, "error": "Failed to load model from checkpoint"} + model_class = type(model) class_compatible = True - + if expected_model_class: class_compatible = isinstance(model, expected_model_class) - + return { - 'loadable': True, - 'model_class': model_class, - 'class_compatible': class_compatible, - 'checkpoint_path': checkpoint_path, - 'total_params': sum(p.numel() for p in model.parameters()), - 'trainable_params': sum(p.numel() for p in model.parameters() if p.requires_grad) + "loadable": True, + "model_class": model_class, + "class_compatible": class_compatible, + "checkpoint_path": checkpoint_path, + "total_params": sum(p.numel() for p in model.parameters()), + "trainable_params": sum(p.numel() for p in model.parameters() if p.requires_grad), } - + except Exception as e: - return { - 'loadable': False, - 'model_class': None, - 'error': str(e) - } \ No newline at end of file + return {"loadable": False, "model_class": None, "error": str(e)} diff --git a/src/jamun/utils/pretrained_wrapper.py b/src/jamun/utils/pretrained_wrapper.py index 75570a7..a434406 100644 --- a/src/jamun/utils/pretrained_wrapper.py +++ b/src/jamun/utils/pretrained_wrapper.py @@ -2,15 +2,13 @@ Pretrained model wrapper utilities for seamless integration with Hydra configs. """ +import logging + import torch import torch.nn as nn -from typing import Optional, Union -import logging -from jamun.utils.pretrained import load_pretrained_model_from_checkpoint -from jamun.utils import mean_center_f, unsqueeze_trailing from jamun.model import Denoiser -from jamun.utils import find_checkpoint +from jamun.utils import find_checkpoint, mean_center_f, unsqueeze_trailing def compute_normalization_factors( @@ -52,20 +50,19 @@ def compute_normalization_factors( raise ValueError(f"Unknown normalization type: {normalization_type}") - class DenoiserWrapper(nn.Module): """ Wrapper around a denoiser model that matches the spatial module interface. - + This allows pretrained denoiser models to be used as spatial/temporal modules in the spatiotemporal architecture by replicating the full denoiser logic including normalization factors computed from the denoiser's own parameters. """ - + def __init__(self, denoiser_model: nn.Module, c_in: float = 1.0, trainable: bool = True): """ Initialize the wrapper. - + Args: denoiser_model: The pretrained denoiser model c_in: Rescaling factor to convert positions from overlaying model scale @@ -74,16 +71,16 @@ def __init__(self, denoiser_model: nn.Module, c_in: float = 1.0, trainable: bool super().__init__() self.denoiser = denoiser_model self.c_in = c_in - + # Set trainability if not trainable: for param in self.denoiser.parameters(): param.requires_grad = False - + def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cutoff): """ Forward pass that replicates the denoiser's xhat and xhat_normalized methods. - + Args: pos: Node positions (input to spatial module) topology: Graph topology information (already contains bonded edges) @@ -91,20 +88,20 @@ def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cu num_graphs: Number of graphs in batch c_noise: Noise conditioning parameter (already computed) effective_radial_cutoff: Radial cutoff - + Returns: Denoised positions from the pretrained model """ # Sample sigma from the denoiser's own sigma distribution sigma = self.denoiser.sigma_distribution.sample().to(pos.device) - + # Rescale positions from overlaying model scale y = pos / self.c_in - + # Replicate xhat logic if self.denoiser.mean_center: y = mean_center_f(y, batch, num_graphs) - + # Replicate xhat_normalized logic # Compute the normalization factors for the rescaled positions c_in, c_skip, c_out, _ = compute_normalization_factors( @@ -121,7 +118,7 @@ def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cu c_skip = unsqueeze_trailing(c_skip, y.ndim - 1) c_out = unsqueeze_trailing(c_out, y.ndim - 1) c_noise = c_noise.unsqueeze(0) if c_noise.dim() == 0 else c_noise - + # Ensure c_noise is float type (fix for dtype mismatch) c_noise = c_noise.float() @@ -163,29 +160,29 @@ def forward(self, pos, topology, batch, num_graphs, c_noise, effective_radial_cu def return_wrapped_denoiser( - wandb_run_path: Optional[str] = None, - checkpoint_dir: Optional[str] = None, + wandb_run_path: str | None = None, + checkpoint_dir: str | None = None, checkpoint_type: str = "best_so_far", c_in: float = 1.0, - trainable: bool = True + trainable: bool = True, ) -> DenoiserWrapper: """ Load a pretrained denoiser model and return it wrapped for use in spatiotemporal architecture. - + This function is designed to be used directly as a _target_ in Hydra configs. The wrapper replicates the full denoiser logic including normalization factors computed from the denoiser's own training parameters. - + Args: wandb_run_path: Path to wandb run (e.g., "entity/project/run_id") checkpoint_path: Direct path to checkpoint file checkpoint_type: Type of checkpoint to load ("best_so_far", "latest", etc.) c_in: Rescaling factor to convert positions from overlaying model scale trainable: Whether to keep the loaded model trainable - + Returns: DenoiserWrapper containing the pretrained model - + Example usage in config: spatial_module: _target_: jamun.utils.pretrained_wrapper.return_wrapped_denoiser @@ -194,40 +191,42 @@ def return_wrapped_denoiser( trainable: false """ py_logger = logging.getLogger("jamun") - + if not wandb_run_path and not checkpoint_dir: raise ValueError("Either wandb_run_path or checkpoint_path must be provided") - + # Load the pretrained model py_logger.info(f"Loading pretrained denoiser from: {wandb_run_path or checkpoint_dir}") - + # pretrained_model = load_pretrained_model_from_checkpoint( # checkpoint_path=checkpoint_path, # wandb_run_path=wandb_run_path, # checkpoint_type=checkpoint_type # ) - checkpoint_path = find_checkpoint(wandb_train_run_path=wandb_run_path, checkpoint_dir=checkpoint_dir, checkpoint_type=checkpoint_type) + checkpoint_path = find_checkpoint( + wandb_train_run_path=wandb_run_path, checkpoint_dir=checkpoint_dir, checkpoint_type=checkpoint_type + ) pretrained_model = Denoiser.load_from_checkpoint(checkpoint_path) - + if pretrained_model is None: raise RuntimeError(f"Failed to load pretrained model from {wandb_run_path or checkpoint_path}") - + py_logger.info("āœ“ Successfully loaded pretrained denoiser") - + # Wrap the model wrapped_model = DenoiserWrapper(pretrained_model, c_in=c_in, trainable=trainable) - + py_logger.info(f"āœ“ Using c_in rescaling factor: {c_in}") - py_logger.info(f"āœ“ Using denoiser's own normalization parameters:") + py_logger.info("āœ“ Using denoiser's own normalization parameters:") py_logger.info(f" - normalization_type: {pretrained_model.normalization_type}") py_logger.info(f" - average_squared_distance: {pretrained_model.average_squared_distance}") - if hasattr(pretrained_model, 'sigma_data') and pretrained_model.sigma_data is not None: + if hasattr(pretrained_model, "sigma_data") and pretrained_model.sigma_data is not None: py_logger.info(f" - sigma_data: {pretrained_model.sigma_data}") py_logger.info(f" - mean_center: {pretrained_model.mean_center}") - + if not trainable: py_logger.info("āœ“ Frozen pretrained denoiser (not trainable)") else: py_logger.info("āœ“ Pretrained denoiser is trainable") - - return wrapped_model \ No newline at end of file + + return wrapped_model diff --git a/src/jamun/utils/sampling_wrapper.py b/src/jamun/utils/sampling_wrapper.py index 3be05e8..663e00c 100644 --- a/src/jamun/utils/sampling_wrapper.py +++ b/src/jamun/utils/sampling_wrapper.py @@ -2,19 +2,21 @@ import torch import torch.nn as nn import torch_geometric +from e3tools import scatter from jamun.utils import mean_center -from typing import Dict, List, Optional -from e3tools import scatter + class ModelSamplingWrapper: """Wrapper to sample positions from a model.""" - def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True): + def __init__( + self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True + ): self._model = model self.init_graphs = init_graphs self.sigma = sigma - + # Apply mean centering if requested if recenter_on_init: self.init_graphs = mean_center(self.init_graphs) @@ -95,18 +97,20 @@ def unbatch_samples(self, samples: dict[str, torch.Tensor]) -> list[torch_geomet class ModelSamplingWrapperMemory: """Wrapper for models that depend on a memory of states.""" - def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True): + def __init__( + self, model: nn.Module, init_graphs: torch_geometric.data.Data, sigma: float, recenter_on_init: bool = True + ): self._model = model self.init_graphs = init_graphs self.sigma = sigma - + # Apply mean centering if requested if recenter_on_init: # Mean center positions self.init_graphs = mean_center(self.init_graphs) - + # Mean center hidden states if they exist and aren't empty - if hasattr(self.init_graphs, 'hidden_state') and self.init_graphs.hidden_state: + if hasattr(self.init_graphs, "hidden_state") and self.init_graphs.hidden_state: for i in range(len(self.init_graphs.hidden_state)): # Mean center each hidden state in-place mean = scatter(self.init_graphs.hidden_state[i], self.init_graphs.batch, dim=0, reduce="mean") @@ -115,17 +119,18 @@ def __init__(self, model: nn.Module, init_graphs: torch_geometric.data.Data, sig @property def device(self) -> torch.device: return next(self._model.parameters()).device + def sample_initial_noisy_positions(self) -> torch.Tensor: pos = self.init_graphs.pos pos = pos + torch.randn_like(pos) * self.sigma return pos - + def sample_initial_noisy_history(self) -> list: noisy_history = [] for hidden_state in self.init_graphs.hidden_state: noisy_history.append(hidden_state + torch.randn_like(hidden_state) * self.sigma) return noisy_history - + def __getattr__(self, name): return getattr(self._model, name) @@ -188,4 +193,4 @@ def unbatch_samples(self, samples: dict[str, torch.Tensor]) -> list[torch_geomet unbatched_value = [t.squeeze(-2) for t in torch.split(unbatched_value, 1, dim=-2)] output_graph[key] = unbatched_value - return output_graphs \ No newline at end of file + return output_graphs From 6d2a9921ab15211b897e134b08d8010b0da10e2d Mon Sep 17 00:00:00 2001 From: Joseph Kleinhenz Date: Thu, 5 Feb 2026 18:47:25 +0000 Subject: [PATCH 31/32] lint unsafe fixes --- scratch/test_multimeasurement.py | 4 ++-- scratch/transformer/develop_transformer.py | 2 +- scratch/transformer/temporal_transformer.py | 1 - src/jamun/data/tests/test_subsample.py | 2 +- src/jamun/model/arch/spatiotemporal.py | 1 - src/jamun/model/conditioner_usage_example.py | 2 +- 6 files changed, 5 insertions(+), 7 deletions(-) diff --git a/scratch/test_multimeasurement.py b/scratch/test_multimeasurement.py index 4b04537..d6fe553 100644 --- a/scratch/test_multimeasurement.py +++ b/scratch/test_multimeasurement.py @@ -160,10 +160,10 @@ def main(cfg): print(f"Using ALA_ALA data from: {JAMUN_DATA_PATH}") # Test automatic optimization mode - model_auto = test_training_mode(cfg.copy(), "AUTOMATIC", None) + test_training_mode(cfg.copy(), "AUTOMATIC", None) # Test manual optimization mode - model_manual = test_training_mode(cfg.copy(), "MANUAL", 2) # Process 2 graphs at a time + test_training_mode(cfg.copy(), "MANUAL", 2) # Process 2 graphs at a time print(f"\n{'=' * 50}") print("ALL TESTS PASSED!") diff --git a/scratch/transformer/develop_transformer.py b/scratch/transformer/develop_transformer.py index 482afd8..69a7aec 100644 --- a/scratch/transformer/develop_transformer.py +++ b/scratch/transformer/develop_transformer.py @@ -38,7 +38,7 @@ def to_device(obj, device): """Helper function to move objects to device, handling various types.""" if hasattr(obj, "to"): return obj.to(device) - elif isinstance(obj, (list, tuple)): + elif isinstance(obj, list | tuple): return type(obj)(to_device(item, device) for item in obj) elif isinstance(obj, dict): return {key: to_device(value, device) for key, value in obj.items()} diff --git a/scratch/transformer/temporal_transformer.py b/scratch/transformer/temporal_transformer.py index d2b82f8..2824943 100644 --- a/scratch/transformer/temporal_transformer.py +++ b/scratch/transformer/temporal_transformer.py @@ -211,7 +211,6 @@ def forward( from convert_spatiotemporal import spatial_to_temporal_graphs, temporal_to_spatial_graphs # Store original device - device = batch.pos.device # Step 1: Convert spatial graph to temporal graphs temporal_batch = spatial_to_temporal_graphs(batch) diff --git a/src/jamun/data/tests/test_subsample.py b/src/jamun/data/tests/test_subsample.py index 3e94d01..99392e2 100644 --- a/src/jamun/data/tests/test_subsample.py +++ b/src/jamun/data/tests/test_subsample.py @@ -160,7 +160,7 @@ def test_large_numbers(): # Check each value individually for j, val in enumerate(lagged): - assert isinstance(val, (int, np.integer)), ( + assert isinstance(val, int | np.integer), ( f"Value at index {j} is not an integer: {val} (type: {type(val)})" ) assert val >= 0, f"Found negative value at index {j}: {val}" diff --git a/src/jamun/model/arch/spatiotemporal.py b/src/jamun/model/arch/spatiotemporal.py index 991f2d7..34845ea 100644 --- a/src/jamun/model/arch/spatiotemporal.py +++ b/src/jamun/model/arch/spatiotemporal.py @@ -460,7 +460,6 @@ def forward( Otherwise returns just the final spatial features tensor """ # Store original device - device = batch.pos.device # Step 1: Convert spatial graph to temporal graphs if hasattr(self, "graph_type") and self.graph_type is not None: diff --git a/src/jamun/model/conditioner_usage_example.py b/src/jamun/model/conditioner_usage_example.py index ab97a09..a684ff6 100644 --- a/src/jamun/model/conditioner_usage_example.py +++ b/src/jamun/model/conditioner_usage_example.py @@ -293,7 +293,7 @@ def test_conditioning_shapes(): else: print("āŒ Last conditioned structure does NOT match x_clean.pos") print(f" Max difference: {torch.max(torch.abs(last_structure - clean_structure)).item():.8f}") - assert False, "Last conditioned structure should match x_clean.pos" + raise AssertionError("Last conditioned structure should match x_clean.pos") assert concatenated.shape == (N_atoms, expected_dim) print("āœ… Shape test passed!") From 133c6b29761d36a156df86224b43de05d0c84f4c Mon Sep 17 00:00:00 2001 From: Joseph Kleinhenz Date: Thu, 5 Feb 2026 18:56:59 +0000 Subject: [PATCH 32/32] update .gitignore --- .gitignore | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/.gitignore b/.gitignore index 314eb10..2b478af 100644 --- a/.gitignore +++ b/.gitignore @@ -24,4 +24,5 @@ torch_compile_debug *.profile* **/*.log **/*.err -wandb/* \ No newline at end of file +scripts/study +wandb/*

SGE!6d(OzqvvZ=Sp|% z0kerMFOvLm%j7YD)agWBMOMejylf=6zYdJc5;f`%M?4=H*K$AMYke$!u+f2?R*<_z zxPv!>i*p}(=$CyZIf#zD&lPIu-?$jxucx(o&Kuz@CBKlUQSb_JlpH-h-}29bO>o7$ z&#$eE%`JZO%qV)EI*CwsbhN9$CI@E{rEb?8!oYor)RGa0o@xA#4wFIc|C+dep7f9h z%y(6#1&I3LF*ot>PAblv1@%X~CkIKI|Drd;hO5MDs)7|EBdKdd*At0v$y^kn{!3G( ztvs(I;y%Y73LCYt*x*93Uu<*d*^7}xWZy88Jd~ZocZR*Tyzp#T`un@F zqMMV9QXA^BLF{9(>w*sV5~35r#UC+>f^_V^pW1swlBrN7d=dk&YrNHJo8FDIr@Hf+ zQ79=}H9A|SOINu(j-&7KWy&C9Y{PG5zhM=k9(s%M%MGjkI7=4|z1>%PX@RPGZdg{~ zqfDtHH+l&;qK237^CV;OO-zHx;$u%4)&G->six>sfM(v4++&Y<>!^BrKW;xotui^J zVo@r4m@MaMfv1urGJ=}2AR~p#PS3qvuzK~1VakPTpmC9bV=mpUWDXhnd<|9*x|$&j z?sJLN%+D$%fIr0u^L%vax*bNEJHCYa;^tJ<=oOadzJ8Mp9vlb_4vDGo|9jp4p}&92 zVV%tM)nwEXvsiOIrF-$q)^8+z4;UH`p+SBr0Bj5*@k5B)OA%Su87Q#zDn9ZO*s+=( zuJkpLV>Yi5=ZMYMYoyK2(SbO#x|<8uP0HW%)B{aeHi{Efui0behIKyGEbcP4cw`dM zd2Q^2M1Zx>_*VFYIgbnR!fUvZ-e-zDUu%g4Lq?hI-xgC7eS!@=YrMPO=|OK9lnQS5 zM<$p@2Vf1Vz|4ncK-NEX^>*P+VwRhEa2Da$5)1|1QtyM4&u!cl($}8(lwRQu)ld-n^Tb^Fv@3=2Pd#8iaD7iwQvBA00d3h)0_$h7W=PFCh;G{8~L#|vZ ztp6=yr1<`Mwu247*9h12!VEMEF;p}zo+Y#x> z8Dzg z7Z_Ty_{~I}=&|*fsv?$DKh`!X=ms)ta z^jl_@e!}m6cI@)|SLQDNzKB$RWkVDfDDZ~6r(Ug^V?QxC(xfYuFQ+`8qrRX8iXq7n zuq&Dv6Z5N=kw#|B(1#7#bN}P2!NThEQg6(H;1p1feAg6==8<(H+(FWqmA)^4`D!9t z!*cy!*_hk>a{*zzgk3IRaRn4d3!RzS!pr-)jV&zaE-90E_XyXOrs{|utyj;uPP}O2 zjC=f7uVr5kG%(J6D?p+K1hiRIO#`KQ)<}L_nf%bw3YuinGpwO7Vmp%o*Ftr+?*lI8 zk>k$dpRP0>4pwJzU~orWC3RE4uQtSt8kZsdMTa}?z1LEDBeey+qWg+jGyJ%r~*VO!du z>Qe8fKkjW7Rjf-=v&RJY`dK4bpS=5p&P5b%h=EC%qk>!IttOo6!#ccLt~b4-L|CI> zSUjoawQ|ejLY7Gbhmqym%QVaB;qOb%c`;@qxIX;lz<(6OcqIY&Gh9cX{$=|03$0MV z0 z`NBM2n=PgOQZ-NKO8n(;2iFE`xu4p+0bt9}DU!#PXF|f8q)OmLxbZPq05|j?7L}{5 z@A1C+@)0US| zxZuPsiPb?pN|-vLyOZy0UpyUiUtXGVs*WNDwe0yVr9xjx&FOJ*<6Slr9XB>xWl)KE9lBZbF#G$ z+J67WT4F=JugABn3xGB|3MZhL>eGgloV)fQGx_x-3n@qO<0;j0N-taDkFl-H8f`e= z6KyiU{@c^}?~nOSo2KrUZy8Ihdg30vL6ci<`j+;e6e<1a$Xstx zPpm#szy4lZnh6~akTN}o{%;tNtziwa!5mabyoC|^7_;&bJsw>R;g+CbpoPPfWx3#M zv-ge8c{(N0Yi!+~AAb-{rWY2=yp0feARnyw^rjgO@rI9se342U^1x5Dx^z)?ZPd>8 zBSpZWcDiQyJpXKVsjHbhe~(IwsuKa_`(UTc2L#>UQzxUpOUT4PJ(T^JUV;W+TTO>6 z?(tE@nK`%4_V)Ytv_{EW_lo%tNn9_kz*O0?Tj0Y0gjcl)z_Opgy7tzo} z-u8g2k>ti_m+Zp)sLr<(HylZV_0!GIXx4IdGJNyYv4$BRQopc`{lq8HOLgx9^K2qx zO{th&Ec{35eba_m=Q|I*{G4Lf?Z=#5XgooRp?-8yp(gG+f|auSLkqz?!*Huh7O-ZJ zr8qD1r|_kzV!B1Ml2K6bqVtB}i%OB2#-@TN_1Dy$oQzJ&N!Hi1HO+XXs_!n}+`SgE z)I(Jm{#4WN6Bqg_pla-)}5hDWZ^&5OAzD-2Msfo*byR zX=(XKM}Pm8ai9((xIz#TmOg+Dc4AgK#E5LJ4(^lRh6CN*k2|`^k9Qq^4CpddXG`Q& zu{!xX(jWpVZ|z*$L_22l^!cyLmN3T{+Nl-@#%=Zwj3q49I28VvYJU*l}5~=S?ZoJ$7B-VMq*>cL+(py8&Egq??*)VQhC9>bvjF9Pm zL-B9S;$rFChbF5bB`4xFR)!COppFWIY zQy3cYU-DaQJSgfcm8|+$lRv>MimH7jLEApxMmv;mgMG|UO6hg8X)jqKmCGm6aAfwk z5)ow)X@BH*C+VO0F=k;wT0%T3#9e}{ob4jRL5eqDNi|eFF!mDUF-v|7#^ihh35Ae? zP2QF^dF+s5jQur@p`Sv5e$bi%+nNIH8X3KO5b4$Y*a>x(F7pQTM^nXiz0Vz1#&QCP zTzVfcZzsmw4VI{zru;UUu60sB)K}xkAn8HXlc<|jJ6ZVj8J-!%*241!eg;I>*e9fU4e3X2NBH7OT&j|-9s0lY@UDTsv;fZy5CPAZ0BNV_>X3IsuzxxJ0YuDxn`YFfq@_T*_SMq@2T9_H^0KnB z`E3Z%tlG5DI?<6_@SBm(=L21>59ENg-PFu~Q91g{eyDcphDnK=Y^srm;G`pJXGBA| zqeiXB{a2;;%OuMsk`tqUug&P+f|2g;>$<@b(H%#vrkFe_Fn}I>myP*-&o&Cbx@%Q! zXjA1#MJN1fHjehw=%KCzZmfArX!n6+zpKbIaGQ1*btSF@$LgTK&zq^g2R4DIB%a;H z#}!Xtz-9dAeEl^~?HGADfBH4IuW4p+pS943TU$E;+J+8PwdoCqBCnL!6m$Fy8K5}`N)o`%rtcYb^nt6p~^ z{fDkjJ^<)?OvEyB4y?`%qC`3wF#-Xb70Z$1&`bV&o^JW#F#SF>nq!xT1&( zbo2CqirBHaV+&m8F`{RG+ub=lnk;yCR@m53!V3QW;-udRbl1`P*rJv-?sH<)K79F{ zxzr!pn%^tLDMqu01Z7R=Bh?hN%sU~jr%|O#Mb@VdZ&qEo59(Fl$KECKf0_G(H+V`_ z2eUGTYDB9j$R=vo3alA| zmvxnpwvF>v$OLyrT`*4&qrJ2pX^tHJy?%S9ptQr6sn`)>&|xOnwmW<}KVWn;Rt}Z< z`?2M>WEpGIJrxD9lrE-nDKY2S(9v3LrR(h|5A$~lXzqhtl5G(;9HtIDo zIY=XHhiHvHEcsly@Q=WFC(81Hu#*v@?#A7IVHUCGnUnMlP|VUGTx1r%O(b(^%X_Th zPQ?uJ({^ZS0I@ z9j5)@|3 z#H`G4AR*VX!A;hf4XRcBmSURK`p|_ozUy|>yP`W?pH-DGu@p#kszN5-h0X0OaI;S) zJ;!Ecd~0mu67I{dfo(Ejh{v+@%E~6ADR4yhxVrD32Q@f~ zJSS8;xLl4s)ElHav(+)&;!A+(X@}G`*lc(?m}kj-q9>9`Idrmy;=21Cp?gGd4?M_1 z?p8jVA3Xu1cZ&Jo@%3-;#}00T9qtK`^?O!!1*0W_Oam`B0YoKy3wKB8RTS_+^qwqK zEs#&XenQ3k;sw082!U-~*Rcswe+K?S`d9V*3L@6pL4qZ54K87Vh2J|1)9TqE_4!jJ zRIfSr*-@9c;qDQTvKz9)b7z6sW#!^;1e=gVH_6mDO35r&a?2cU+-|;BmNLmUcJCKe zkHUQiJ-eyBB3(mot}y)(Epo+$n&R8UNpT#yG8jf(-|N4a&f_ceFzt+YRN!0QY=y?; z)14ehrO5t|Nx>_LCTbM=zupgTKN|1cZggIz7tM>|e0oyPiL(;IQ#}YDkWAJ+tISe9LIXd@`@AuV|%i-U=b3pKxDr6t*^JxD+r= zzjF1;6*;L#>eUl?xrRq9EWug5g$TD#>~YG=x7xO`4TN| zSp?m`Yh5gtUI*KdKoKEJK{Sf@V}J`sieWS`(_1k`h$XkDktILoR;gjlNaNk5VEpoq zkTx5ZjFazOEnjbhyE39+X(B(TRZEa3Lr5=`c#BS|JW=0!RisZeHs5mfRHOAt=(q5d zMt805y=}e^sQ&hdjl7Ndp{N;W$3g~&KIzG!iL3BD-isS6{Aaynt|^6C{vHw8tALxR zRL)cGwBC@w?h*fQ?QyzxTje|7QXHifi`@;_METuduOA*`uumK+im-&hAK9gK4`>q)Mo<$tADh>-2J zAbA=BW0y(xDCvKi>e1-ZW0kBC_j9Dkr0ta2dI`pO==;#3dlK$;;k>>gUAo2*%yYe- zH)Grg$Etugf>Aky0Y_-ZTrO3pNL7%hAOx#cL9o(+^xNznHsh8?o4#o|YtR}#ja2G= zlMpNzTyv869y`uysxMo8R|*~RB{$f7>FbF__6;pf8x^6US+2{N==j~BoeVB0c{{%>wgTCC7bU~lGaoI$Ve!cRAlx*8?52kOJ6>`LmxV5t4W5)Jb!WBj& zepzHCT4*}biCDp&fg*tq_Os(V_29)refksnM=>zb0|c_7pk|U|_PE~zA*->3NO6FK z2!ZAKs|`9BQuNzS0XF4piD@~XY+1T_`S)gFI_YP~#XAO%@h!fXOC-olmeO5)db0jj z?L6IMno7dG{e3UBuxPc>Wu0dkd|McKL$5S~I5qN5S|Q@i)lajV)ksS;us6WqV8iae zX*XO0b-T@~O6yc3)#>?`Y^=E|-RlIxHN#ueN0JN>z7_*XLgg zhRT6SKxYa)XD4YQe%{YcRuo+)DVI<04J-Cl8)Bu#+x@-V?S%JPOI@6Q9Z$qaYDSzbUFr&JHRe;NrBnP?uBPFWNbp9y z7oJtD-}WKMeiX?C!33bsKK7jv#t>ajHPdah=({It42cKQbhqC9%=T(}vf#~}dFGUh zy{yS%p~>^8!q8e7Hb^tuco&N*7v_Dth@J)!=6Imiy-p`*`vHtg@0#67t}U{5X-ZRI zuM0v>*pBKFlz3fmViikEQ@*=>jbM;;F9lV6JcTg)`Ns5z_XxKEymQ&^ASpe*K!{S? z2<~lHU{!K2p$E2Sjqm4I5C_$Q&5bGWNs2ji;eX>`?XLB-B{1;n*1}6O%xmU5&FiVA zf_drqmgUS}Mz!XiUmp{r=SIzHR`F|VI+zXF>ELaQC>T7WEOy`bnysH97YM6JGH@oy zX0RpmF6b;JWU(&R(aoSw5|I!~2!L0vj8Cffj@9>Y#7JC26VT4DeH1RD|3@-2m-CtH z^=rc!lr-`P=3$HQgA33`^#jy03Sc?PrRvzD;XHUsr-WryGuA9`rxfo++>66Ra2r9V zBTSbDF(vmtHTRJ)K2R2>E+D0TnlFZKo-x-P?_!{rd3)ZejvNtS%I;@_iTZB zcJK83`)jeUBU(mH%gFg`BTd8!Zs{-rPN9?NKZFqJ-8z!SU+BI-!gfNcDF|_Z@>yC| zRaG^=jpAs@0A8TU@aR&Ve)PXALEz5ViFqKFw_sr~Ikr*!LB#k?kNKTuv8P$Bt*!Id z0ZZNR^?IbPK(1yxP_?}qPMIc8h^nr+UK?(-E_+dUCYt+piOuRaZfA_)Z;3=5JT{i^ zdelbP8g2%Z0;M%N7LU>riHoxvFHzf4JT}F09LjO>RJ18!ueg9gLqPYi>_@b3Vi_On z>Gh|X4ri)0MOP#vdCX%{+R`d>oh^~qGj{nQQ-;vPRMHo=#=Gbq_^1j}n%;-N%4ohl z5ShYkg_jH;?jK1>Jk|UkLMx3ewao+3&@)u=)!wh>3U>u)7aqTt^_nH-+wodG@_77f zA->UzN@Kt8)(GT>Ujkv~B1lg9)cJ6|)wrxfj}95a9e1E%5cl27pn zw6dTXE|BOsSjdWBu3VmF5yjXQ9~)h+`)>omClbjBrzMUkKAu1{d5vpIPV-j4T}&U4 z&Uq?!G8m&-%~s8P(_9|4b3g+B47qLz6LKT>4YWdkA9Bi{KD3HiFUh{KE6ZucDd_p@ zu;?a_&HQ-YQEi-m`~IC)$|Am)QVjI5saB2!L*B{c=VRRL=kKZ+Gt(K91Qpp6uH}FYP;0E{qemo-HOov14b0}jD4%*obdURSq#s5@ zfqcO#1_3eQ1yAlI_Ijoyx7c-VqJ6F6BF2oh=kIjtC}kYIe;eyELvslrx-4*^ehmN; zTL*Ejfz@L+ER+I{d>=s2xVFB&P`_THSDvBG$_U3T@h+iL9G z9wTt7wpO>!Hhb=oyHUYvH~eH#XeFaPNz1T~ksk9eB9}~15laIPgft|=!OYqk79*_? z{oe6Uu|%sphA{F$mD^VDe|;ppYoT{wK~xf_t}1N*ng|<87taH{+z0Xbu5ph% zpg!&HyLRMNt`J%rcOx!J9XLI)@>Gl|(XR%ze_q2Pp!xFd4xMETmo=`Lg5cxO5KQf> zA!)`x-@bdd`?U|pMO5M2SC>f15C&CP*v4{}v*TLHkkY7PTQTbJZsWG6nV8N9Sv4rE z-@XA&TC}1L<}iMCO6VPn08C;GMYH~5se9I4vkl`_@qSGx4%@CR1Kx=2{uf?0q{ro3 zKRPevdJEdpHlMF9mD%PjPf0D1sulU(!F){d5Oa{;@ifPAI0`bHH0qbr?GSaWEOXLM zG>jhnQ-5~X|M^;gA5X2RkieAlf7_l@C^mNjAv8i{Yfhmee*R1hiK>-eQXNZBq@$^FNAE6Pv) zNEzD7HutziB(z33nb@D|#|}(Sc>nhng^puk?<9-sl<_CJR2QuUx7cu;Eb#hZ@Hcit z&7;L{{`~x0-}$OW&4=7R%)joooPS5j5oAM1Qxp<9krPyW3w(tK&9Aa_V+xa%d#HYg z%E;>?ii(#@Y)fNOr7w$AqK17-w(NejcjXHW94i?z8eQ6qG=S=^LQUnOv;ExZY-u*j zS@A#(+oV5^0ccox0iF2SugmV3#=cLQECzp$bUz@ML!b`asF;{tNM)|coqVz$_L$A-8j^nvE+;n;dwpA5X|RP! zF?db}uv?wtWBY7)R}f?JM)0ayMm!IcP>{TexGk-CfDG=%Onp?6R(H^R;6p(Q$KU*+ zzqOOm?_j$s*U1mFMgE-NEmoDzp|)vG8aeDXJj7#fk_+MdX~CsFk(fXF;ChqO9|uG9^- zyF9FE^wCV7bBdEcVXh0JMU-FY9n0Ugsj{|3eWnFNnw?WewGk1~0I^_w-B!Pnh&s6he)1+iE&2 zIXXeUM1(r=?4u-|UwH(uD_xAk6Av57A;x5 zhv5G)_SR8Z?d!j&s3=k*-S`?vccX|%mvjn9N_R-BgdnNFOACTDFCCI9-AGHTGzcgN zNP3>hTEBhnzW1Cl?iljda_qI&WX|vRc|N)36!3b>?^+eb+}IbvlN~F!09TCP9hQP8 z9$Q{RXNmi49X2)@De0Eg($CW^-B*~5le2zRMBpVZF-&4~llq@AJ7DL15(dn^i! zp;nCVhL$;cNvf+hX%taXH&EXs%vca;DUX ze>}ee7svHL4UZr!)*xOAr4+!kEutOsF#GAD2U;t$14%brT6)>SS{hgq2wT5;GZ_~x=m#&BN%r6`gO0VYWI*}4$K%fGms8MO!o z%b)bg$;op?bsOJ1ANZ9OuUYdY{vwA%>^(Gv?^ab;dZ||hAr0WjxL~tmVQUQ^K}=_b zr9z9=cvS=zF7_pr3W;M>k6XNAIm5VHC%;@ z&1=+qDdJ892y8g*^0vm(Gy+L_%fETY;NO41NOucJ0YxteXuD8Sy8KvNDnu_69`ITD z{0f`_rRQ3gO-Bd9o%n=Pe?b5e`9Pa-o`x{*qqD8iSMQ&%yPcBA1BFhVaT7^P9b69C zVc!%~X~r_|g& z#UG5naO>67EpuId^#ul&V(*A&V3j>C5(GEcj8a;L)zoS+WKH4>@STu_*>fW3c4;Ue zM*iOhWEc!c+Mwff0fGiDAP_4})kY4)3!Ekkm2Myc2!xG+Piq}9#$G!A^2KdnJw3w( zx$Ayy$La~^O%~ZIc#;e#W1ST<5G|_CC^g3-Bw&*HlFO8886s^?9LBt|%lc7|JOnM2 zZCNyEIl0U~U|$^1*lF>dtu5_pIbW3f)|afm62Hk_e)Fhk+i;aOBvy$<>lVFY4xFt=FStEWq?+x! zSTR{vPjIWd%ON*-OZX!um>0{uv!Lb@SQGzMwN;93y^1QlWJ-vS{}a*9e8aBl6Io*e zGkh`T$&Y6j=`EKE>_sDgN%&%A&xhqcIY$$So!kK;9+ zrs{h~+BcjeRsNCqQkl73?@})aS@A`{OU-*$JeU-(Cz@}2zQb$1;}cOv)IEOryU09~ zRyiA5+rlkny3gA*rT*AxRlF9fjgWFDf7Oxk=_wJu;QDcIJT#m;Jv}EUXC(Pr{&1wR zEi3|)?MTKS^tqROr+mQNikQs>vL;tz-u!;YABEs(1V+WP2*74Zx%aN_Xrq1WL(sBD z!7Z+L=AMTx`e@O}z)>GfJDj1^N0>RgbWJy!hWhxeTVLvEI|?`JnD0^%H+dL#Z&B~2 zM$pEvvY2&!q3EqRp-Er~vvDtfDA4)6iNp-O=$P^_h-eW<)w^(Qf(Q?FG+?>|_(r&) z$y@mNn$?)$hPoi6zmUdfgg*~IfygIW2By2yNBF{60pMo5rS zWFd%`OmPP8>k#nD7CNDNHvCvgXe>8Bn|eE>F55P0b(}W}&J$3!;>OTyR~FXPI$Hel z4>qy)zq_Oz{QWeH-R{=#mbF?TZXnqV%pPrZgG^S5L4WC?Qf7fj+ z98Hvw^^c3rrZu!WMLf)YXl$(d`|PPqH%2*9qoXj5St)lAr~=WF!ohc3!9Fultg)>O zAw@2#_A>Y9IyBY4-PKgMT29sOyYTHC)xo4Z8?z-N1cwfj98d@hq19S$yHfS zUlzq-YH$(|YpA(_71OOj>#k^_>G-G8cWzxEl#|Phxc?gaJB(AUj*BXLLA-FjMWbwJ z!%bdct1&7^vU!$0ZA!m1cwZ-i2o-GZHr6j*k-SJhH>`H0xH*ShGFdP3a8wsHl(X=0 z;u-};b+)$P(qnYLAle`EP(MFvvg{$UJSNFJTWRz45cGQ>QMef?y75!Yy}+~)@^F~k z-`PI5n#i)N_V@;`kVHsr+zTjYyYLl=8B2MR->VAuP&+b=);?-L^2DsyF#ZLT19)nJ^GOM*cg}t7KmrzN4O3iibukC@a zJl$s-5ABgIQOF{P6k4-#-L3tjBoa%eDr+w!P79>UCWs#pF}eha1X*fBJiug?ll+Pp z7}!qG(CBVySD?7k=UPSDq+v)@{!LGZR*89{$lMe3jpObk6Ou8p25fnH_eIhAGj2M2 z*W!rS$FRpPK#s%fU$jcaJlW>QrWcgR&leOfdMEpzJ<_LefH4 zqA-t8w)k^T@`Rga62D29wIY60fY9uuK<#r|8JlgpL}rQTx;F`3zj(u|3ihI`+JIZa zb7T?oy*~T=#S6{O72F#J{bIJCNBQR}8Nc-`a#NDp;3HBAy#<#Y%W@BbKc`~mD=7!R=o&N(6e0%kWV>?JeVfY0qk+M~i7YIWRd(hplkgjkG zBEZCu9SF(4AFT)?&oE%hLQE5J@>uYNk^oi}{tCU2Y&A)==#~E=Se-)f+~OZqFy^Ot zH3YH8Wxc$6F|mRq;_J;91ys2kVPylRg9>K21KCMOjGCDeV)x;ISF{e65K|{_`jPd3 zfT&bDblmheldvF(1Ydj^=`sGG13{VkCO|fLUP=E^Kbb$HlE{96bq#o4wTP(K`~kGz zL$sLs&hvezY%AQ?-m!6HrL6kC`OITuAy&{xUcXf z_k~!HLca>Pfp-ftZ*jE?(%8Y`8#fW-VWKLCz_0m>!Q6;(6j(vdU&Vh9ib2Ht`Ov47 z|K~(q?I@mgZ9m{82v!6C=r4T2Pp}cKHR6gl{Xhl3>!G1Jx^UT`p53L5cv=|_8_Z%b zmmuB{>lufPa^~bw_d*k4`7N>YJjY$Hbq%wW(hAO3l@RCUN(A^nu&#-5!%4d%M>m{-C5S7&-hKYgZ%8Ky(YfC8gWF5fzd%qE2X#g;g(`a%w_Kf} zoj4Snp1NCRPT|dGd3$)3I(~D>24b1$(7AjsJ8^gSCX%TraHsW1^*y>dkt|w!geN}o zftJ(t3%aiD3pGf3XO53bgI=QFm)4ylCdoMUHP5>fB zx8sU^f;K@j24NsZAg$xD9Q0sp8HM+2W#!iiJE(IhaD&b`Rw#ZA5H+sxq%pZ8VRb54>df(9S?I4*CEw6VzgQ`vVAQ3j)-Pz;ld)%~s(IM9V5LE(8KZR_0jcHr&d+V z^uY1a(%mJ>AD@|YgEkP-B`)+4{ z7Blj{MWWFwhOE)z*_P)E4xZ(Y9>4gDAA$LPy@k5#&}h;*g*J3Nzzl7{Y!+XlqH`YX zYeWr_>sL%hg?zPKCc0rNE@x5zj{TkqrJW8R9$Ny{Yx8eCp1iHMW6r0lEhw7I_-Gye zD(*@m(y#O}n@)M9ZE(Bmvv755i1OSU{ZWQ-{h|du&SvLy6*krnuV}lL#Hy+L&c{|p z%oq`X$qGTCGejV9?KXt;_P=aLl5!C^7#<OhRHh;m@ow)KbmxOoeXH*uQ^PQzLwO zJ2dV!M3wi6ggxTIDJ*Q7*Xs6ZrkVK&LA_JMo4O%8Oj%lrCZ|VT72BBD`IQO1ytCP( z)a?V74StmBc_waBlVvYQBG@mP+PB&18Z!Oiij;r*#%q$DB1|x0LQmeI0;&@kMYOGM zEVB&$z)f;49WLf6m(Vaj?kem)>>Z8lY(Ct0X*BFU-G z=5B_Akk+e9EpQ#>eo=XT_~VdppIEBu8k3ji-Fse+9pxWMhiGjF$l+25OSx>Ty)G&M zWp|c^gE-q(COz@!g>0RhH8c6C(|zj`7L!$D4R&{0h1CPxX)D}6OM4tEy=hJ72KE`2 z>=yFaD5FcJNJc~!vTo0`+jCbU&9iHfld?z&J(74kT2Fyszz%K5+851e)X)4;PX7P8>namZI_gy0nePYBA*7|fI#`LKrpFA7+9MlvD zAbTn(XlI&{WJM(22sj*d+@R;xF`DwPz>SUkBT%l~j1*o>SUfosm@BS__8=FG*8M45 zF-orJS^sl;s}rjp0=#Qi+i4uGbcwH*|R9 zT*v)48p}om?y|3aArTySHPuzvOX)F!X{IbTd~{tv+SbMJtM-U4zo+ND9rhOsS5w3H zG3uJoS9Q`K%OYF5oi$DVoVV>`{&!;7yj7<4hU@?Fb0!1%WJ3>@Kb0j;C`9;VMcn?uPs<<<)Lk=Q?aj}U9b zo@cVumpb!dnk5kdWfPN=ZN7Lp)GUw}H6E;z363*m{pp=OQUaMgb=8dYKOq1cp-Yg` zE(c?EgVut5@)puVfOr}fz|+J1@g+EoJ`dWxQ1KfCVwk0<--&hxej!L8B~gTQ!`H`e zRq7x80cBwl60p)*b!os>K9B64DKd0V&Bimt5$LzRj|os_4y?H+r3k!(0?yEbER$F< z%??+*%cIq7-D+8N%Ay{&VY`~@H^;KqZRB_EnJ$lN$QP_`hWuEs+gm9>Q_OwX3fBW@^6oHgp3L~G=_~~Cmudd7dL8#Sl zLN*28k+iG4P4HYI=lSSDFv=f!-vA}|iW;v{f;18&t(he4&w02F3#(0C^+Tem(sckmM7IPOh*xPV6OVC(pjYSdPp{7rbQUS3{YOeusm zy7u_X01njw1o%cj<>5oJCep?-wjebXra!}F5BJqkBsZdJrul!2ZEAKDG8R@=Yr@w{7TU~bj1NvP*vK#YRdM6uRT!NT;>rJFQ%RKq zlk%`?a}wVdsS;~iz>(rZ-+eZ_fSV{*!REngc0f!_Rs?jNtDn%&zu3a?GUv&6&gIR` zy_0pV&U(;IdUh%pHZ?UpcUbd+u7)S002mJ0E(M{syFI;>eU3cse1DD1vd za|FXYC!+9kxV7em0ImSAhKx=W1nbIW)oo5eEhUML;4nhC#VrBo{jp253Bq{DApkcp zLb4m~y7LkLKH872(x7WcDVBp_cOv+(by4|w?H;X=7%B1gnzlKll!)iqGg9Z|;}JH7 zud{IUxPNIS8dwp4;urf-d($-j4#}UCaAg^@Ir5=K6lToiplz`aMbAFfeq1vtoZz1z z_4uBlkwi$u4hx~v}Xqatz+5_-ZXWUiEr&fH)N z%q^tiXR^6l`;&8-vBi4heZ?xl-Ge88emp*OQ5{cm%&v5L@m3lCF}0~m$u&jP#DiPm zH({KZ?1!=sqVRS1x5Kf zq(vm>^%}4wLUtE6z99M^tS4`~CEk8SY^ccbdp0?~zc|~xrrU`Yt|-dWJaHr+(ryq7 zDRP0_xqc)|d_5${vI<^WL7M)IovCA$I;{@AEO((~rADSRn)rrYMaBCszsW2pbUSM| zY}=u|?*=MPg!Q@ffkZ|1Yl)Rb_w>+V(pv?0-uv6N#V_IL?}?S98rsR{nS9kJ+Y6exy)Lrk zv9SHVE4jd~JuLwX``kH5`yhg+E8&`s_I|mJv%bg7#2jUsj7cuH+p__@2&YxXc#AC@&2LRh!s^JIckcS6L)N+IS~Q z%!qrEt%$wQw}mU1zpC7(+*Cc5940)#(Kcd!+e?@_IY%cMVy;xzng4Q$qn&$!?E2AC z1N)y*`l%P<7FGp?Xyps`{bR`)q(y8&yyIqs)dtOp6U&1_YGlO&rx7-`jBo-XJXZMQ z+X0YuM)p>biT)zEPNOdSF5gG;lFvRGD>5L9-$9&;aHNJKt+f>u^S*a98sDxUTrm;> zd%Y2p{m)ewb@xOFxO#CX8676lFVH}Nh_=!3stQ+!rdN=*7$kan%Kz{2Oz4jvKiFg< zX|dkLey}9S8h(ijVcl9ypX`c?i!TRA9O{ghkP|l*8s**dP|(R5mGaZjGZ53wa}cib z=HuLak#S{;PKDjH@tdjhexOmvVjk+;Eqvc#&%p&l$%!OZ^L_jii?RR(S0AHm&vupEBM_aO-6dY9~GWbDjYIE zspO8*I>kC7`GTED*~u+=yU_-A ziNbFL#nu;mSI(!zQ7CBtt+OvnBgK4;sEgRGnJ&Fc2>Oyjr(nvQbeEPLHU$w?=zfx2 z5sK-X`(3~V@}h)0(&E@49g+wkj2lttS+fEsJDU}F=A=n7;!F4y1C2rUE`^!ZcG$@# zw$kCKogBTqnmn#-V0(9FxonlzsO^JnL)|;~2G2*LEFYfHcP;sxE^H$ev$@=n)f*>% zE6#{aA!H)exr=$q?{NKSWAlUqict2Ft}d+lYm10Yh(p^@cI){`U!IR52S`(WkFsr>8ZoG|~STc3Z*UZeUmuK$R`#oBe-rIU} zPmXw-Z|D#V;C0yNYN=A~l$-`#6;O(mWdxC5o|B_HT!DWO)6m=TTX5s(DtjlNZh6&Q z^Mix6o`3WAK5=lEXA{t%tH%%Drxk}kB{ON6=)@82T)_uW`)14y2Xsd=n_R<3WCCN+?pDJEUgjPAnkGZr!qDe5>B2&$jBgT#qdhi z9Ktq0tt6I??xk|zaq>l1SyF#M7)W&!yVOJa<=B^j86#ci;M1)N(!W;KH2ru*8fGDZ zz!l~9owN$RNj}6s9J;r@&+Y_w+9sR0DMD6-^iL56ZQjPCFQs-_Z!DNo*-ef3Ml9ya zQrk>Ts}O3>noN|^>^HF-JhXDz8MDr0S~=c!)b156?u=bZ=oYC9wIdHATF0bsvn80^ z?711!!i~*GNJxnF96z5@m0#lHYGml%MCH(7D3M=*n@^q5*q1rn#O(1R`MH9PIV}%i zzX~=^_WZ<5wHZHaEdu~J`8Cyl$Qvcy+7%6d{M$KKkSJQ{DFV2xB2tMLV!u&_d$nUu zQ1P-oQtUU<-POdouJP+yLws%N3JdE~PjjcZ8TF)K=|sPILyzAzrZv<`8_kT(E$EjY$W<3$ zPT}u4k>{^F(-p9f3;|W`{X?_hijT#B#?SK&npIF^WJ}Y)z~H>fnWLy_3LdxR22#X(q1+&$4ZfK zR$k|wd^UFcuzRnK-OCDGPX~oudNzZMauWTo_!Z_mU%T-1tjS6ICMpdM zg~&$zE5g#m8DUfyFcFad=94RbYPtoXn7B-~U^KqM|2?P{UcR>r?c6lhKB8`=sWoxu zeW9_w*{Mi$jIF&V6$2EgMeAncn;LUdEpovtsrL={;9U}Fg&t@J|HDZ{Ij+nRKL-)t zi^wDbej1Q?$0+A8lj8o~-6bY`fm{|)GIaKFny&!#@M30;f@4nO_ZMaF0d@a^S0kE37EcNAA-NX5SAvAlrTLr}EiH*{VviU#6 zjJhjBH^<^b+Pf@f=o4)ZfHW9Apu~y#V;!ZYW{@Jog*uaYW90xhyHV zyD3}6n8o+vH1K~5d8rm5?jyx_cO3b_%*0lqPBpEPW;n>#WIi(qa zG6`u)LPQDB#r)=TwBRw^3~Gz0(|f>7cJLV|5UkEV+;{+Ci0o3 z$1``Ojv+jGx+`IgtiKSIvLa@LOa|FJ`G0aIUFY_01X>E6>fcFEQ*XfQ;IMW3l`PlY zyZv9jd_kkXT^Q@SbyEP(j3rCR=MUFkByoDt1yL!v^a;;w2U-TfL9}UV+Jlzp%5d}Y z_U4v0=CZT{{na6%U|@~z`zoDr70-K$rM~ay+J2$_E-Y}bJydp5cH4GhtyLd#znA^; z-ln}q!+q!SGdT~6QK3BDG*qG@6HkemJ-%K^@@=Ouq zmn|gi!_R-yOTiXs8Cs!Snw|sMj?VAk(!CgvEwgh)GZUZKrkjYdVv^n&)9Zjono*kM z(xSeuH}8g`HZ_kfd$WfrTln*fh37&EOP}$}jf_~9NUF-IvZ*x3Y#6{ud5T$)RpORh zBo0b|G^|w!dRhK}?|bD=RY}Rm>ruDM7Q`1+%un2P?8CXht1?xN)Ry zFBsgf$hp(&t*Fc>RS=8F6%!1P6D%2EO_^!mKl?lIB?HB+5`nbe(L~v29E)!qs-lL9 zKPTDMj9fJTuFUsT8}l}wljV4NnS13yoGvQAJnq-@1b3cZBX%+#GGJW_PDuZEz`8g3 zjyZvrnQkehJk5^`WgS6W-j0(CJC%Nj9VaxxL&6^w`g4|i0<}^IJxF~*sGG_}jB9Il zyt_sno)N|UnYH&kb{o!Cw~xM(YUZh^=cbYQkh|=-!_+Ixsy~@V(v(U5TbG(pjM8F} zAns#p1AH_2E2oP<$Y^vX7wGIW2AJRgAABjkcyx3d1}#CzvFdxNwYZi?s^)SoG0u#K z3URF>VnSk^^MJ?dpo6%Br=e&7n*XjuiylhaXyU?|Gcr4CbRheUluzUld}yYX9M&e2 ztv-7Et!Ky7Yxf#9EV`@M-|`KJXN8^7Y*QNpx++@!>9T9uve%_>Mma*wnX=MTzyCbDy|8>Xqhj+lQFiXfswvT)H6!#EvkVmugdy^&D4na66Mb+qRB_;}{J^gE z&FQWx_L*?Kz(PZ5OeR*xGU=tqkri^U1$CVn6{dPjNN#}cc-&O7@SDh_JT8ibdHJ>z z?|ldw)~+g9N~ z#^Xzw(?)tO5ly>^t3^)u#agKRi4x||9(0sY(a^+LpT-ptDcz7U1XbP6?yfTc9Z&vN zvEj8ly`PH|eTtdRTYzKd0nzsU?K2P$2*q>ujr$ge$Z-qttQpD<5ty=MakV+_jz4w* z!1Ku@N|F^WEsOGTa=Xyn8>z~?ckcB5{>}I639ZhI5gBelw(LlkJC{V^W8rAszYDiM zY`9jf587?P2tkl;jbnyhx_p4?bj1(974KX-od+$$euoT5t55s`B!;2acl}nJJZP28 zcxo26(dNdA+|haR?EWt?MtUWs8P}N$gdX~In4-z(3muHBuuD$pewid&4i|qwUyI#Q zrY~SG>dzUACK^%Gp!b;Ra7adt3(%-hoZ5n(l1bb2zLtv~b}>^*x~=Q19PD{V*EZEQ zf~dT0rMk%(r4Yi-H4)7P($9b4A;-_8DmozOfN#>!CVeD%g3AtKG0RcYrI#sEI|%L^ z9ItY%`8+%()}SN95BidjUw{+%*L!1-SpUhxNrBB6ll#`GfD^Zp-?e`sg5gKnRh5*W zH61Q?5b-cFGHN|+CV>w$kmv-6MJjt&N|TN zb?do2Gn8W^YvjwNnmDfR0+X{p#>-#$ALRId2l8k)dyP0l7xZ~vZIsa2&_HCt5;yU4 zzqRvLP@c+7pDZE-r5+uuY+rtE`*gD0SDUc}~M@4gn_?F2isBLI|F zV0vO@zWa^!vVoQweM^O1GUCTZ_|5WuF&d1Uwh<9gs`-vVCX|8@n&LxQ9ynkip1Q+~ z2U_5nrP#<#q3;T`(-1;gTM{_O=M*3{N47+s+*IXjuiH2x|C9zHg_$%(Hu9~X#5G){ z5$3S5%;5+$Vu=z7y~r11t}982GKoredlqO&_!wf&ncpS26Cz>_z3cenB!L>$m3rhR zHsJ#AbOFS@GFCp{`JM9Oi?ffGRgYVSiOxxy^=0J=+zo!^K|a8;F1(AjF~j|KYi>si z<6^u2#6G{W4eyCwFU_{I`?TFlgN)!8FFcVJ(+~wQieAGKl ziRaEa2k5iPs#3$l^Ya3z5dv*~w$2{)&Q^GlaHR=umw8N1>)Q^p3fD$p6+NL>BO%E~9o>l7Cx^>qzUAiA%%RH#iG zO?q;BL$hg$>1ee|AI?@&SJ&amzvA5U7`(5fW?rQ+mChnv91RZp17%Y-)eo0UNm#(N1+~wl?FY5 z2~?;-R1r~B`%LLP8yY#X#)wBS%M-=g%d;=`=)zn&^VcqnV;7m*e3J z&aeSHE)t)y4V)MCWL1NFVj6h*kdR-2!q-%|K_zxH-DO%2KMS^{R~r7(;iF!NI$4f- zg7|TPlUTkpL*PK1QCDa`je^Y5tV#<%KPcH{wy?11SlV6_8!KazmX@vu@zr6b_*HQ+ zJZz&uR})b+_q^OJkaVvZoj-$Br(vNkVU>ZU!Tz@0g04L8`j@8&)$pWP<8*Rg_sD#Jo4;TnLC zVgShC{NRvW6q`aD)Z1jz40uI|l;q_oNnUTx*vg;mJHoqmsDp}G*cC&kScW8#-Qw!T zBR}(dZ_foHhMeH%Ri{~fS0d!)+Mr9`zwapaKeq7$GwRUnY(MOmqV7Mg5{Gv1i^7YW z(d_DM9It$5BYIpA)q*}oIuM(in}MSRehVi^0=5~kv{|lbDrC&Pm?M$)6@GgwIca_W zQ&orv{))H0N@IG1Ecu}){%*TdLUOXHwe?HT=YXrSw!hLYd0;>lskx`8Z*%WHZ~qa@ zRG@O3z`Hs-J9AqbCS($e05Zbt!uwy$5Wf-CiWz$bo( zv#AH!SF50=1a{eZXW=tEiHc#>xx8e!*zjgTq?(1hJ2_^I^A08cg$AO<&ni2Z9dP1M z=0+1xq`nJsAV9Me&Bo&i4j%{O!UEa<`BK^s;gu5J{MX+~_902sET&z7rbHZJ`ji zg4BKTS;qugCeF+lav}w&)qSa-6XD_6+uJu^dZEJ90@IeMsi|bUVZ^nKhbghImsgVI2UZu#f#JgwkL1cYGs>?yK~jFKM71ZcyVS3ObHBB3J@+= zb_S{C(bwgjDt6Cc_^rt0L|jhC$J3EV3UY>R5&-zXVrBWJI6z)8YKY@9!1gu4tvf|zIy4+VZ z0yDW1i(X(hh5!dr_=(k4`PT93Isik(D}nLd3qnvGiV0#;krcWQPPlOtvq? z8_ldLeW{vXmVuWl_nF#DczZTy2g`ggt|IF}Tywwv1oy84F)on+vm-&xXM+?UHCT#< zdkB;coNgt;9(2y!D92puQJs~~vmR=aMAoL)5fk!)SJ~mY#0b8ZiiZvk$Ov3FhWd^q z&)0po{{sw(NJgy!%ny6~ismtNQw(H<3$P`ghtd$ZHy#igKtRdo%!ANey5CpO$rd*A zK@NJ9Wsv}N_{8CA>`4c^ly+q$me#Wg3;P+Uk63Y2T_%=QISpXDqL$r^z-I9m2P8ty9@H%AHUcRedB`q8c|-DM$M5-s#^5y%Gsjej7Y z3PZClVevF-CTLp0$wMHY7REuZSEml1=<5NdFAng$04 z^Lu%E`udOxV&A>#cK914BqWN8ihRdj&k*iI^12*|lfSWI_}eILCwI@r6W5bY%Q9D* z@R*8LvI({U1`YlQNW|Eyk!Ln31)Hty0$uT^M??sAJZxKIUURqGWtM);;LRQfD<-)b zBYnIWH`2DA7=O`^wOYbfkj~X`Xp8Fs1+ETCx-d1oBqpWrwW9?{D?`~oVWp>e8b^GBYm`bQoZfM<;r7pu=io1Jri_P2>aD*!63Kt4PDY2(-+Z^ywkd}*;K);cD=;^i z`+_Mx+T1h71WKSUGWbLz3^|nX@Z7nhbnVquB)$zrihBhx$JN|SAs}|4uj~&&#(-tm z2ND1RYb78W+2#Q{f*Z^!5Kb4t^1Rkg2Kn~r=yvpis`a>Z3bp1zzoUA6QqNwusB(8* zLIpem)6`A?pHl+>hB$Exx0 z1h%*|=POqWU7oOcr7FrkJ{Q{WR{%ZCy#svEC#`isL&n$716cZEDAIBxu8YO!9YzJNCM8xT4Hk;&KlLAse?>x>f^J16pWf}cUqZj0q`aW_3UL(r0a;BKFD^XPMlN?ceUHVL zMGB`nlEk(M<{pqD(SHRp176M7ij19RJfyhTn;A7~xIpB_Fk$!+v;(nNASG48P(1|0 zmhD0{V@t4>cE^u4^n#wv;?ffA>|$v~do^%h@$h_y#(`(MqAiZM(x63R>X%+zYw46N z_1ERW4{PaAkFrV_>$&|3ZoOISH$OO7|5^`3Qxr@k_y>%Hm6R0cC{QC<`pQ;g9tC5zdAc^^ktphZJT`Ss6S z)DryMN>v7b+W!7Nf4>rA%#9ei|Ig6xy0fn#=BXv^`=U4V9&I1|#jxr(wM`6aoH4m{{r^J5F+UH4t^ZVSw9%OGp6U3J zv*#5;4#6Go-4d|Xm5s9@5E6XuqM~(+i#u6aH|c$Ut0F(Zc|h{@eIPKBk@IgN9=RPf zU78c}G+u&S`3(}BiDcc+EZI#;GcQ!~yueh)xE9#W0Fv|&qSg=_@f_02@dVr%U!pgq zLcC(D+4{|R*r#3}z*}tb_;CYK_UKwp^Yc&X6iu`c;M=X0x`ssAXTCG9$xVzG)>2`z zJhHqWbk+c9x#Gv3$CKA3YxHhaCabCC6yoL(^_H2~7W*++#)pZom(fioo9iOxjL3{i zivhNh+@;QcMO)*x~dj9J`jFZH0D zhA0NT4{n&*vjf)aZnsg~ssply*rT=g4ce-wbH>aIS27KE|4z50->cLP>iO@99*UcI zd3kM$G=UmGWNFYy2LDBdLghn6AnAlH^bom5RLPKo(&^7nV=<5nZV9 z&w3n%R@CnMvbwD|yK^aeZ*reey@QM^iY@fgNW$D}=kI1p!MBBZf=GE+FL`O)Jr!k1 zgD5L>z0b0lSdRQR>7_d|Z`dN89GToH0f9P|WO@A9kFjK{j<8&)CKqnHf*DBpYRX=D zyugp5F5bXRoUv~>(T*0VtsoWtOXS96kZCi9o49uALcouS$}sTvt1^)$Bqshl^3N4o z`^NnrW`2>37~zZ$aE{gmvXv_E(WtP_#ZjV{LU8q1$iknOmD8jX31H& za^gh$xLMx2XvA1)Z#GEclTFykhR26wxTKsR{}{OD2D>X6ab4${Qi(cxU=M!q072YA z78-o(5SCx$OBWMUzA;FLA)OY7-0sb@{O4PIbTMi=@^883w|5*qI$tesG}k5ndNuBB z(BJycrZiFPRC^hDU6C&;tme)pZq-_9Lt2w5q87SGA-k%pCZY^@Xfm7C{}6yZ2yaTF zv2HYAAu0#yG+r;kwI)+b+T_Y*iD=^ZyRN2slOz+F)VUOF@f$N}wU)-bqE5_shTpnI zu><|qq!q>Pa(;H|m1|_vEz=&mdmbCThG1o(6Ll{!Z6~B{H+FYN!`y=)mP^~LZ<70$ zgk6`3UJ%kjy)I{pu!xBK)vHd(%PPT_M*9p0RnM9Y(i-)Z(56mm`=DvnWP)GqADGQ$4wZyn2;S*zp=f3qB@1bg>#QjtrQ z9WNvp!tM>X@8Tj-$^!7;Lyqe2HQ<+-@3;aJ=kxz2M7-0~-6r{|kR@LeBA?PdoEkf;|e=@WErXA{G&w@h-AMB@{%gap1$H#yy6xNXRQr?Hu zl)op(h)&dJzuRaLiLZH{3Uis=XDp=5sKjH_@Z?ATNg;c)gUO+aJSL$>$9Hg6UO_1Js^Ae}4Iy;$J=VVMThSf^?JpqyeT~ zcO!w#yJPdMhP+I7tL}h?)?u3W6Md!;Q&IVkxZEb|W`U{muEBx4&98O;k$r(~qPZ^< zKOZV!AeQSTMkEv4D5vZHOk7??^$;#-VcB}De+bIZ=lApDn$D8hPOoO7}eCq zEx&N}b<@h;{0?SGZE&zolXS2x1+&S>$RHCDQgntu2g=H?JI%CiYuog~1rC=UQx;$b zkTJ1SWbW3>#Yh-_*P!o zTT6=GqG$_UTx?QYShOkup~bpxZ2C=bld_yvB#)cw51V* z9H_2Yp5YOZYD&CrN6yWNXv86(@Oj_}5j_2pmvNc|7Bc2nY+UNGQA_iZ~@tD3`qDE_Ukh*bY zQ+E8AM=$6DBL#tD02C?T68L|RY#kbLaS{f1Q*T; zMN)*j{*9E*JPnsan%*Jsf2*<2PS8+EZJM%q^QdpramI7$H1%g7%5NCYpmq(ui@F*h z!-Nihl#6m;bJi;C%N?wWa?-#jbnVqoi3BL`Xdi`I0CGyvUS@(?S6bqCOr5^_QVqd>S6i4tSDFYn2z4 zD=$DpI2oBsB3K=Cui^X_NxT zXZKs44LGWDnP(F#>9abVA}kBPKl43qR}l+dnck-pE+TQ`W@@cs=}3>gMjb&MAuYV} zE=K3!!`~B0+N(DNfX;zc4zZUZ>-*!XYd=s+PB7MqAf2iEbsr8zPaqR;y6)GfP5Fdr zNn{rxsImLx59vUTh|(U@#ooQul%W043EO}7vHyjyzRJv+>I?FF?=q{a<1hRFX@>m} zsjf9 zc?Ka`D!GbH5OBCg)=2j1*aD7%1l>EcGCAgC@hr`TIHd%D(cd*bz<^v0t@!-io9iH* zIhCvB$w??*zG4%Yz_0uPHuZl5z+d_;{NJ3as#sdOk)#l%NAg9+D zAQSMW(>kcvI7f;ZXnZlcUT2=M%^p=G0w8@w%J;G<$DI1hArqFF4Nk*Waqo^YVz1E> zl7JKqy3F!4AnT!R?iYW5RQntR^V54Rqkz!*2QJ_e5@H%-fP{IRucu-0z~r!0U3ZxG zHotXI3>?ORU!v7s0z(F>n@F;#I?+bK`fmh=lsb-YCAW;k#kW zcBJL!quKq_dOq6qT(7f}`=Pb@75Q|N&>pmd`u@Q}qX%a3r;o$(quW6ZC}aE^caA&ZE5fycN+MNHgM)_N^N ztTx8-d|FeKm?6_>TXBA>$t2H(z4LB$7kIS)|15O>@MFN@@Ey4^7kk4ExenGlCn!t% z72)91Eb8!Z^EiKSr%%3rdb!q8uz^Be?LJOv9s)f;wK56Q2JQ6jH)L1woSdAwF#NgA znM5A4(V$Z;0P*%hQSQNn2!fo z6*LThH+OV|IJ{l9*EsTJuDHP?MK@Jl`~31;N3rtUh5*%dQ~f^*aSW51yp6HeKCN1t z?;b{;liH5ejxgZe(o!O=V6gqsNb@bDt&wD$y-?~X2+Jm6z@#D6^(3$n8-!D7$VHq$K6}t_W?v3?qI+=z6z^D`m)}eU3LOt)uH6@}HF0_+=7P5rF z5BU}3qc(5L+gn2#XkEW;|K62SUkvT5h~&rln4gXHhLUqnC_>ig_f8^)zL$nt+kH_` zW)n%`pn>Bi0G8ZcL@f>8azwrgfEWO;jJIwjCMNz8*{y-u_;7FYF^p+2wQ<3)*wPaC zF!M@A%73aztGp9`VU1q+?2LM0LR2w)9U&fV;+wdX*!2(Imh$aMRAEP@x`#_DY! zLD?wc*ydkbYst_+h$34Ss9Ki20Sn|zeY9MYI z*?aH33E3(mWW`NZlo=V3ohY*RmYHmKwq$#+@ALb=$ML?;q2qXtMc;GisPv`(;q3xgUF2Uk)nQ_Kx(E z=GDsRjYnVEa&FG;FKPeTwka9O^qp);qs9|`BrpJQg~inf7SW5oV-2aXt_4OdYrutK>=N5_?D^E@#*|7((xn!57>Ky!fJd|)24TT91iWQ#^UyF)D* z6ge9m!C%NOg+Izp_gP9&nndJ+e)L}^Rb6DH7}vCn8Ob6o+&|<0PY_24g({$-j+8&v z*4M3|wTBDE^g28w7Bf6M8+$>x{cB{*@YIz==Y6+erYeJ_s)ph(Qh9!XLsfGO0|jVx ztDhzIZ8{ch*%Y=({VBF}ott9rXtoKhegeKuHB)KreEAhUP<|s*1>Euv+RQ3rVN+n&zl;-G)qrQ)h7;hYYPf*!zyxKDl-8=?3(N004;r zGXobMRt=H&^(^m#{SX=QXS}Lse4G*?x;#8QhA=Yg69LF#(gVbFSiJ!KH2sft=+W_U z?f!&=*!jt>gZ>6E%gE>&@&_>lIs8r0*bYNO%J2r%6y8@ZWkOm-Cg*Fxq!wV=z(50~ z|4D$fU-lZ7={m}gHjqp$#>U2$dJ}4E-@2Yf&WXd1d-?L{{Y`LuVyYR$#RCAdHi+2X==LC`0I?FP#zkdh|L+KXVN-0bj4{o}ajb43fCk z8p$T6*=nH5^`sq*#1UG5`NE*aT!{|1sF*0a6DlF)paxl=Aax zdX2lYubDaK8W$OI>E~qUH9{4Gq_WJZ6C|yhYZ$vg3!;2)T@V4_U9Dla-nQX?f<0rd zQGiK6s`uZz^E$2YlA)59XsPhq$J?<%-AKswT678AO$)eUku(XB3IfV`%VXmmQp82W z=Y2Q*`9WL#x|7TNJpFmiCdTo<*>7QTC{D{PaDnICR{o`4a2+wWVttXHm-lS3BfOzO z9K5|yl5A`!Owl#V8CdTH@=1v~s7~o%&-Hj@-!u}xkRj)Pn{q5o|C0^mc1C9nBsDJr zpsgk%gXvwJ&hkI2;4TfK!m5(l#{Iqf%XzU+uY%-X$d^chA_CiklF>9g+r7vTT_>)4 zo1Nego0WmkpB?d@gWaplmsVL}LD%Qj6^*&Dmjcnk zEIl+Jx`1^sl5~)V5vgP$(jmblt#q}t2iH$og0j&aOqSSf=jtZfCr zQ>OV6Lj0!&+QkdxV;vQj_|v2~TDG=$K|}!~mmHw~M`%7c@qlu0WjI8cSzCvJ{uYK# z!~w7Z8}PxaszCC6Ih-R3s#z+7JOMTtqn|wYVZZ3<-?9Jpi$YD=eZ>lKbXW*F-#qUb z%Gj?6EB>PPpBIm&D9sWWaQp_SVa5Uy!u`E|>HDA!cRDx{NFfxN_8|L2(@uCf7%_6k zhQFb2^EYDWBbcOmX8KJ!Rd1i42AyYuH*dn9CCvJI3uUNq>z4%CaC<%NvVbSwUlOFK za=+Gl+U2elr3_0W2k^R%JNFenQv^$1#@7tbdvfJl)w;M29L#mtIMEGq47~r;mO1dS z`4>7~yQz1tNbmK_4%4w(M-MstJ+2DOzRE#n10&%5pdNg$R7T1sS-N)1aXly2HK9pL zN-*P3W$1^ndh?L;lft;w<)7c(L%Ww~H!Cndqd!D*^vrqK`}Ha3{QrAR1m=P;0z#~s zr9+?%!BQN%>%gU_b4_U6)+5>xqGNHz18 zYRA&Go;Z&&=Lqyhgv4M;x%F0J-43G{6aWW6rS)EwUE^z;g=7;!w2H-SoAA7`Ttxwo ztj}9dGNUzjrnbJ~_y7NH)B#q?w~gJGcr90C_w)`_>J@_^y02T}tJc-8BnZ9uUe&g% z?;P6u0EF4vs9mPxq2B{Y!hl#Qp=+^bH81S|=GZ#H*}M&^TBzS) zObL)KL)Y&($jT3wgSSlPm?VQ{uJjj-l(vnp2+d#Vk3XJ#+cqq&*{&(BrT2MVT_1Lc zO&SK1nih_;^sL-eZx@Q{*r|I%9Vnu{s9PcUS8j6i$)q0!m0$LMn;SIx#Yh+D6cd