non-parametric bicops (tll, fastKDE); examples

[TY-Cheng]

No description provided.

[Copilot]

Pull Request Overview

This PR adds non‐parametric bivariate copula functionality (tll, fastKDE) and multiple new usage examples, updates configuration, CI workflows, and documentation to support these features.

• Introduce a new MNIST VAE experiment (examples/vcae_mnist) with seed control and logging
• Add prediction‐interval examples (examples/pred_intvl) using vine copulas, ensembles, MC dropout, and BNNs
• Revise docs, README, and GitHub Actions to integrate new dependencies (fastkde, uv) and deployment steps

Reviewed Changes

Copilot reviewed 55 out of 55 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
examples/vcae_mnist/vcae_mnist/experiment.py	New experiment runner for VAE + vine copula on MNIST
examples/vcae_mnist/vcae_mnist/config.py	Configuration dataclass and device settings
examples/vcae_mnist/run_seeds.sh	Bash script to launch experiments over seed ranges
examples/vcae_mnist/run_seeds.py	Python script orchestrating seed‐based experiments
examples/pred_intvl/vine_wrapper.py	Wrapper functions for vine copula training and sampling
examples/pred_intvl/models.py	Regression models and prediction‐interval methods
examples/pred_intvl/*.py/**	Orchestration, data loading, and logging utilities
docs/*	Documentation updates (index, conf.py)
README.md	Dependency list updates and installation instructions
.github/workflows/*.yml	CI/CD workflow revisions for linting, docs, and release

Comments suppressed due to low confidence (1)

examples/pred_intvl/vine_wrapper.py:34

• [nitpick] Using _ as both the loop variable and a placeholder can be confusing; rename the loop variable to something like dim_idx for clarity.

    dct_v_s_obs = {(_,): Z_test[:, [_]] for _ in range(Z_test.shape[1])}

[Copilot]

Enabling both cudnn.benchmark = True and cudnn.deterministic = True can lead to nondeterministic behavior; consider disabling benchmark for strict reproducibility.

Suggested change

	torch.backends.cudnn.benchmark = True
	torch.backends.cudnn.benchmark = False

[Copilot]

The logs/ directory may not exist before redirection, causing failures; add mkdir -p logs before the loop to ensure the directory is created.

[Copilot]

Typo in comment: ALso should be Also.

Suggested change

	# ALso allow manual dispatch
	# Also allow manual dispatch

[Copilot]

[nitpick] The encoder consists of repeated blocks of Linear➔BatchNorm➔LeakyReLU➔Dropout; consider generating these layers via a loop or helper to reduce duplication and ease future modifications.

Suggested change

	self.encoder = nn.Sequential(
	nn.Linear(input_dim, hidden_dim1),
	nn.BatchNorm1d(hidden_dim1),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	#
	nn.Linear(hidden_dim1, hidden_dim2),
	nn.BatchNorm1d(hidden_dim2),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	#
	nn.Linear(hidden_dim2, latent_dim),
	nn.BatchNorm1d(latent_dim),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	#
	nn.Linear(latent_dim, hidden_dim3),
	nn.BatchNorm1d(hidden_dim3),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	#
	nn.Linear(hidden_dim3, hidden_dim4),
	nn.BatchNorm1d(hidden_dim4),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	#
	nn.Linear(hidden_dim4, hidden_dim5),
	nn.BatchNorm1d(hidden_dim5),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	#
	nn.Linear(hidden_dim5, latent_dim),
	nn.BatchNorm1d(latent_dim),
	nn.Tanh(), # keeps latent coords in [–1,1]
	)
	def _make_block(in_dim, out_dim, p_drop):
	return [
	nn.Linear(in_dim, out_dim),
	nn.BatchNorm1d(out_dim),
	nn.LeakyReLU(0.1, inplace=True),
	nn.Dropout(p_drop),
	]
	encoder_layers = []
	layer_dims = [
	(input_dim, hidden_dim1),
	(hidden_dim1, hidden_dim2),
	(hidden_dim2, latent_dim),
	(latent_dim, hidden_dim3),
	(hidden_dim3, hidden_dim4),
	(hidden_dim4, hidden_dim5),
	(hidden_dim5, latent_dim),
	]
	for in_dim, out_dim in layer_dims:
	encoder_layers.extend(_make_block(in_dim, out_dim, p_drop))
	# Add final Tanh layer
	encoder_layers.append(nn.Tanh()) # keeps latent coords in [–1,1]
	self.encoder = nn.Sequential(*encoder_layers)

[codacy-production]

Coverage summary from Codacy

See diff coverage on Codacy

Coverage variation	Diff coverage
Report missing for a7b356f1	✅ 96.64%

Coverage variation details

	Coverable lines	Covered lines	Coverage
Common ancestor commit (a7b356f)	Report Missing	Report Missing	Report Missing
Head commit (db5cd44)	907	877	96.69%

Coverage variation is the difference between the coverage for the head and common ancestor commits of the pull request branch: <coverage of head commit> - <coverage of common ancestor commit>

Diff coverage details

	Coverable lines	Covered lines	Diff coverage
Pull request (#38)	893	863	96.64%

Diff coverage is the percentage of lines that are covered by tests out of the coverable lines that the pull request added or modified: <covered lines added or modified>/<coverable lines added or modified> * 100%

See your quality gate settings    Change summary preferences

Footnotes

1. Codacy didn't receive coverage data for the commit, or there was an error processing the received data. Check your integration for errors and validate that your coverage setup is correct. ↩