✨ Everything as a second level import

[o-laurent]

To make imports shorter...

I didn't include the baselines since they are bound to be deleted in the short/medium term, if I remember correctly.

[Copilot]

Pull Request Overview

This PR refactors the package to expose utilities, models, layers, datasets, and datamodules at a second-level import, and consolidates duplicate factory functions into single, parameterized versions.

• Moved plotting logic out of misc.py into a dedicated plotting module and updated exports.
• Unified deep_lab_v3_resnet50/101 and bts_resnet50/101 into deep_lab_v3_resnet and bts_resnet with an arch parameter.
• Replaced wildcard imports across __init__.py files with explicit second-level exports for shorter import paths.

Reviewed Changes

Copilot reviewed 21 out of 21 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
torch_uncertainty/utils/misc.py	Removed plotting imports and plot_hist (moved to plotting.py).
torch_uncertainty/utils/init.py	Added plot_hist and show from new plotting module.
torch_uncertainty/transforms/init.py	Re-exported corruption transforms for direct imports.
torch_uncertainty/models/segmentation/deeplab.py	Merged two deep_lab_v3_resnet* functions into one with arch.
torch_uncertainty/models/segmentation/init.py	Exposed deep_lab_v3_resnet at package level.
torch_uncertainty/models/depth/bts.py	Merged bts_resnet50/101 into bts_resnet with arch.
torch_uncertainty/models/depth/init.py	Exposed bts_resnet at package level.
torch_uncertainty/models/classification/wideresnet/init.py	Replaced wildcards with explicit exports.
torch_uncertainty/models/classification/vgg/init.py	Replaced wildcards with explicit exports.
torch_uncertainty/models/classification/resnet/init.py	Replaced wildcards with explicit exports.
torch_uncertainty/models/classification/init.py	Consolidated all model exports explicitly.
torch_uncertainty/models/init.py	Added unified model factories to top-level exports.
torch_uncertainty/layers/init.py	Re-exported new distribution layers for second-level import.
torch_uncertainty/datasets/init.py	Re-exported classification, regression, segmentation datasets.
torch_uncertainty/datamodules/classification/init.py	Added UCIClassificationDataModule.
torch_uncertainty/datamodules/init.py	Added MUADDataModule alongside existing modules.
torch_uncertainty/baselines/segmentation/deeplab.py	Updated DeepLabBaseline to call the new factory.
tests/models/test_deeplab.py	Updated tests to exercise deep_lab_v3_resnet.
tests/baselines/test_standard.py	Removed obsolete error test for legacy version handling.

Comments suppressed due to low confidence (3)

torch_uncertainty/models/classification/resnet/init.py:2

• Missing re-exports for mimo_resnet, packed_resnet, and resnet, which are referenced by the top-level classification package. Consider adding:

from .mimo import mimo_resnet
from .packed import packed_resnet
from .std import resnet

from .batched import batched_resnet

torch_uncertainty/baselines/segmentation/deeplab.py:18

• The version parameter is no longer used after consolidating to a single factory call. Consider removing this argument to simplify the constructor signature.

version: Literal["std"],

torch_uncertainty/models/depth/bts.py:579

• The new bts_resnet function is not covered by existing tests. Add unit tests to verify correct behavior for supported arch values (50, 101) and expected errors for unsupported inputs.

def bts_resnet(

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.09%. Comparing base (eff9313) to head (99574e1).
Report is 3 commits behind head on dev.

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #173      +/-   ##
==========================================
- Coverage   98.13%   96.09%   -2.04%     
==========================================
  Files         153      156       +3     
  Lines        7863     8073     +210     
  Branches     1034     1049      +15     
==========================================
+ Hits         7716     7758      +42     
- Misses         91      259     +168     
  Partials       56       56              

Flag	Coverage Δ
pytest	96.09% <100.00%> (-2.04%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[alafage]

Thanks @o-laurent! Indeed, the baselines will be removed for better clarity and to simplify things.