feat(projects): add LitePT

projects/PTv3/Dockerfile

@@ -4,15 +4,9 @@ FROM ${AWML_BASE_IMAGE}
 ENV FLASH_ATTN_CUDA_ARCHS="120"
 ENV PYTHONPATH=/workspace/projects:/workspace/projects/PTv3
 
-RUN python3 -m pip --no-cache-dir install \
-  addict \
-  open3d \
-  flash-attn --no-build-isolation \
-  regex \
-  sharedarray \
-  spconv-cu120 \
-  tensorboardx \
-  termcolor \
-  torch-scatter -f https://data.pyg.org/whl/torch-2.8.0+cu129.html \
-  tqdm \
-  yapf==0.40.1
+RUN curl -LsSf https://astral.sh/uv/install.sh | sh \
+ && . /root/.local/bin/env \
+ && uv pip install --system --python /opt/conda/bin/python \
+    --upgrade pip setuptools wheel \
+ && uv pip install --system --python /opt/conda/bin/python \
+    --no-build-isolation flash-attn /workspace/projects/PTv3

projects/PTv3/configs/semseg-litept-small-v1m1-nuscenes.py

@@ -0,0 +1,245 @@
+_base_ = ["./_base_/default_runtime.py"]
+
+save_path = "work_dirs/litept"
+
+# misc custom setting
+batch_size = 12  # bs: total bs in all gpus
+num_worker = 24
+mix_prob = 0.8
+empty_cache = False
+enable_amp = True
+
+# dataset settings
+dataset_type = "NuScenesDataset"
+data_root = "data/nuscenes"
+ignore_index = -1
+info_paths_train = ["info/nuscenes_infos_10sweeps_train.pkl"]
+info_paths_val = ["info/nuscenes_infos_10sweeps_val.pkl"]
+info_paths_test = ["info/nuscenes_infos_10sweeps_test.pkl"]
+class_names = [
+    "barrier",
+    "bicycle",
+    "bus",
+    "car",
+    "construction_vehicle",
+    "motorcycle",
+    "pedestrian",
+    "traffic_cone",
+    "trailer",
+    "truck",
+    "driveable_surface",
+    "other_flat",
+    "sidewalk",
+    "terrain",
+    "manmade",
+    "vegetation",
+]
+class_mapping = {
+    0: 0,
+    1: 1,
+    2: 2,
+    3: 3,
+    4: 4,
+    5: 5,
+    6: 6,
+    7: 7,
+    8: 8,
+    9: 9,
+    10: 10,
+    11: 11,
+    12: 12,
+    13: 13,
+    14: 14,
+    15: 15,
+}
+num_classes = 16
+
+# model settings
+model = dict(
+    type="DefaultSegmentorV2",
+    num_classes=num_classes,
+    backbone_out_channels=72,
+    backbone=dict(
+        type="LitePT",
+        in_channels=4,
+        order=["z", "z-trans", "hilbert", "hilbert-trans"],
+        stride=(2, 2, 2, 2),
+        enc_depths=(2, 2, 2, 6, 2),
+        enc_channels=(36, 72, 144, 252, 504),
+        enc_num_head=(2, 4, 8, 14, 28),
+        enc_patch_size=(1024, 1024, 1024, 1024, 1024),
+        enc_conv=(True, True, True, False, False),
+        enc_attn=(False, False, False, True, True),
+        enc_rope_freq=(100.0, 100.0, 100.0, 100.0, 100.0),
+        dec_depths=(0, 0, 0, 0),
+        dec_channels=(72, 72, 144, 252),
+        dec_num_head=(4, 4, 8, 14),
+        dec_patch_size=(1024, 1024, 1024, 1024),
+        dec_conv=(False, False, False, False),
+        dec_attn=(False, False, False, False),
+        dec_rope_freq=(100.0, 100.0, 100.0, 100.0),
+        mlp_ratio=4,
+        qkv_bias=True,
+        qk_scale=None,
+        attn_drop=0.0,
+        proj_drop=0.0,
+        drop_path=0.3,
+        shuffle_orders=True,
+        pre_norm=True,
+        enc_mode=False,
+    ),
+    criteria=[
+        dict(type="CrossEntropyLoss", loss_weight=1.0, ignore_index=ignore_index),
+        dict(type="LovaszLoss", mode="multiclass", loss_weight=1.0, ignore_index=ignore_index),
+    ],
+)
+
+# scheduler settings
+epoch = 50
+eval_epoch = 50
+optimizer = dict(type="AdamW", lr=0.002, weight_decay=0.005)
+scheduler = dict(
+    type="OneCycleLR",
+    max_lr=[0.002, 0.0002],
+    pct_start=0.04,
+    anneal_strategy="cos",
+    div_factor=10.0,
+    final_div_factor=100.0,
+)
+param_dicts = [dict(keyword="block", lr=0.0002)]
+
+# dataset settings
+data = dict(
+    num_classes=num_classes,
+    ignore_index=ignore_index,
+    train=dict(
+        type=dataset_type,
+        split="train",
+        data_root=data_root,
+        info_paths=info_paths_train,
+        transform=[
+            # dict(type="RandomDropout", dropout_ratio=0.2, dropout_application_ratio=0.2),
+            # dict(type="RandomRotateTargetAngle", angle=(1/2, 1, 3/2), center=[0, 0, 0], axis="z", p=0.75),
+            dict(type="RandomRotate", angle=[-1, 1], axis="z", center=[0, 0, 0], p=0.5),
+            # dict(type="RandomRotate", angle=[-1/6, 1/6], axis="x", p=0.5),
+            # dict(type="RandomRotate", angle=[-1/6, 1/6], axis="y", p=0.5),
+            dict(type="RandomScale", scale=[0.9, 1.1]),
+            # dict(type="RandomShift", shift=[0.2, 0.2, 0.2]),
+            dict(type="RandomFlip", p=0.5),
+            dict(type="RandomJitter", sigma=0.005, clip=0.02),
+            # dict(type="ElasticDistortion", distortion_params=[[0.2, 0.4], [0.8, 1.6]]),
+            dict(
+                type="GridSample",
+                grid_size=0.05,
+                hash_type="fnv",
+                mode="train",
+                keys=("coord", "strength", "segment"),
+                return_grid_coord=True,
+            ),
+            # dict(type="SphereCrop", point_max=1000000, mode="random"),
+            # dict(type="CenterShift", apply_z=False),
+            dict(type="ToTensor"),
+            dict(type="Update", keys_dict={"grid_size": 0.05}),
+            dict(
+                type="Collect",
+                keys=("coord", "grid_coord", "segment", "grid_size"),
+                feat_keys=("coord", "strength"),
+            ),
+        ],
+        test_mode=False,
+        ignore_index=ignore_index,
+        class_mapping=class_mapping,
+    ),
+    val=dict(
+        type=dataset_type,
+        split="val",
+        data_root=data_root,
+        info_paths=info_paths_val,
+        transform=[
+            dict(type="Copy", keys_dict={"segment": "origin_segment"}),
+            # dict(type="PointClip", point_cloud_range=(-51.2, -51.2, -4, 51.2, 51.2, 2.4)),
+            dict(
+                type="GridSample",
+                grid_size=0.05,
+                hash_type="fnv",
+                mode="train",
+                keys=("coord", "strength", "segment"),
+                return_grid_coord=True,
+                return_inverse=True,
+            ),
+            # dict(type="SphereCrop", point_max=1000000, mode='center'),
+            dict(type="ToTensor"),
+            dict(
+                type="Collect",
+                keys=("coord", "grid_coord", "segment", "origin_segment", "inverse"),
+                feat_keys=("coord", "strength"),
+            ),
+        ],
+        test_mode=False,
+        ignore_index=ignore_index,
+        class_mapping=class_mapping,
+    ),
+    test=dict(
+        type=dataset_type,
+        split="val",
+        data_root=data_root,
+        info_paths=info_paths_test,
+        transform=[
+            dict(type="Copy", keys_dict={"segment": "origin_segment"}),
+            dict(
+                type="GridSample",
+                grid_size=0.025,
+                hash_type="fnv",
+                mode="train",
+                keys=("coord", "strength", "segment"),
+                return_inverse=True,
+            ),
+        ],
+        test_mode=True,
+        test_cfg=dict(
+            voxelize=dict(
+                type="GridSample",
+                grid_size=0.05,
+                hash_type="fnv",
+                mode="test",
+                keys=("coord", "strength", "segment"),
+                return_grid_coord=True,
+            ),
+            crop=None,
+            post_transform=[
+                dict(type="ToTensor"),
+                dict(
+                    type="Collect",
+                    keys=("coord", "grid_coord", "index"),
+                    feat_keys=("coord", "strength"),
+                ),
+            ],
+            aug_transform=[
+                [dict(type="RandomScale", scale=[0.9, 0.9])],
+                [dict(type="RandomScale", scale=[0.95, 0.95])],
+                [dict(type="RandomScale", scale=[1, 1])],
+                [dict(type="RandomScale", scale=[1.05, 1.05])],
+                [dict(type="RandomScale", scale=[1.1, 1.1])],
+                [
+                    dict(type="RandomScale", scale=[0.9, 0.9]),
+                    dict(type="RandomFlip", p=1),
+                ],
+                [
+                    dict(type="RandomScale", scale=[0.95, 0.95]),
+                    dict(type="RandomFlip", p=1),
+                ],
+                [dict(type="RandomScale", scale=[1, 1]), dict(type="RandomFlip", p=1)],
+                [
+                    dict(type="RandomScale", scale=[1.05, 1.05]),
+                    dict(type="RandomFlip", p=1),
+                ],
+                [
+                    dict(type="RandomScale", scale=[1.1, 1.1]),
+                    dict(type="RandomFlip", p=1),
+                ],
+            ],
+        ),
+        ignore_index=ignore_index,
+        class_mapping=class_mapping,
+    ),
+)