v0.3.1 Add UCI classification datasets, improve binary classification, add LS support for BCEWithLogitsLoss, Remove dependencies

What's Changed

• ✨ Add UCI classification datasets, improve binary classification, add LS support for BCEWithLogitsLoss, remove dependencies by @o-laurent and @alafage in #122

Thanks, @alafage, for the review.

Full Changelog: v0.3.0...v0.3.1

v0.3.0 Add distribution shift evaluation & more

Details

✨ Add evaluation of the metrics on a shifted dataset using the new --eval_shift argument
✨ Add a corrupted dataset
✨ Add new Packed layers
✨ Add the focal loss
👕 Plot segmentation results in the SegmentationRoutine
👕 Rework reliability diagrams for Calibration Error plot
➖ Make some dependencies
🐛 Fix corruption transforms

What's Changed

• 🔧 Maintenance and small improvements by @o-laurent in #114
• 🐛 Fix OOD & Post Processing at the same time & other small changes by @o-laurent in #115
• 👕 Complete metric overhaul, improve PP handling & fix Laplace by @o-laurent in #116
• 🎨 On the road to 0.3.0: Adding shift evaluation & more by @alafage in #117

Full Changelog: v0.2.2...v0.3.0

v0.2.2.post2 Improve metrics plotting, PP handling and fix the Laplace wrapper

What's Changed

• 👕 Complete metric overhaul, improve PP handling & fix Laplace by @alafage in #116

Full Changelog: v0.2.2.post1...v0.2.2.post2

v0.2.2.post1 Small fixes and improvements

What's Changed

• 🐛 Fix OOD & Post Processing at the same time & other small changes by @o-laurent in #115

Full Changelog: v0.2.2.post0...v0.2.2.post1

Maintenance and minor improvements

What's Changed

• 🔧 Maintenance and minor improvements by @o-laurent and @alafage in #114

Full Changelog: v0.2.2...v0.2.2.post0

✨ Add ChannelLayerNorm, Conflictual Loss, AUGRC & improve code quality

What's Changed

• 👕 Small fixes and improvements by @o-laurent in #109
• ✨ Add ChannelLayerNorm, Conflictual Loss, AUGRC & improve code quality by @alafage in #111

Full Changelog: v0.2.1...v0.2.2

:shirt: Minor improvements

What's Changed

• We improve the handling of optional packages.
• We fix AURC metrics in the multi-GPU setting
• We improve the mc-dropout wrapper and its documentation

• 👕 Small fixes and improvements by @o-laurent in #109

Full Changelog: v0.2.1...v0.2.1.post0

v0.2.1 Add Checkpoint Ensembles, EMA, SWA, & SWAG, LaplaceApprox & ABNN

What's Changed

• ✨ Add LPBNN in #90
• ✨ Implement Adaptive ECE metric by @qbouniot in #92
• ✨ Finalize Depth Estimation, add DeeplabV3, & Add Selective Classification in #88
• ✨ Add LPBNN, Adaptive ECE, start supporting Depth estimation & Improve segmentation in #93
• 🐛 Fix documentation in #95
• ✨ Add a Laplace wrapper in #96
• ✨ Add trajectory models, including Snapshot Ensembles in #101
• ✨ Refactor wrappers & PP, Add Checkpoint Ensembles, EMA, SWA, & SWAG, Add LaplaceApprox & ABNN in #98

Thanks to @alafage for the review.

Full Changelog: v0.2.0...v0.2.1

What's Changed

• ✨ Add LPBNN by @o-laurent in #90
• ✨ Implement Adaptive ECE metric by @qbouniot in #92
• ✨ Finalize Depth Estimation, add DeeplabV3, & Add Selective Classification by @o-laurent in #88
• ✨ Add LPBNN, Adaptive ECE, start supporting Depth estimation & Improve segmentation by @o-laurent in #93
• 🐛 Fix documentation by @o-laurent in #95
• ✨ Add a Laplace wrapper by @o-laurent in #96
• ✨ Add trajectory models including Snapshot Ensembles by @o-laurent in #101
• ✨ Refactor wrappers & PP, Add Checkpoint Ensembles, EMA, SWA, & SWAG, Add LaplaceApprox & ABNN by @o-laurent in #98
• ⚡ Bump version by @o-laurent in #108

Full Changelog: v0.2.0...v0.2.1

v0.2.0 Lightning 2.0, RegressionRoutine & SegmentationRoutine

🚀 TorchUncertainty 0.2.0 Released! 🚀

We're thrilled to unveil TorchUncertainty 0.2.0!

This update brings a complete overhaul reconstruction around our uncertainty-aware routines. Highlights include:

• Lightning 2.0: Support and a complete overhaul of the command-line interface.

• RegressionRoutine: Fully functional, now supporting probabilistic regression with PyTorch distributions.

• SegmentationRoutine: Introduces semantic segmentation support for datasets like Cityscapes and MUAD.

Stay tuned for even more (Monocular depth estimation!) in TorchUncertainty 0.2.1!

Breaking Changes

As we are still in pre-release, this version breaks a large part of the routine and CLI components of TorchUncertainty 0.1.6.

CLI

The behavior of the CLI has completely changed and is now based on the configuration files from Lightning 2.0. We provide a new page that explains how to leverage Baselines using the CLI for easy benchmarking.

Routines

Notably, there is no more distinction between ensemble and single routines to reduce code entropy: single routines are ensemble routines with 1 estimator. Furthermore, the routines' loss parameters now take an instantiated loss instead of a type, the optimization_procedure is renamed optim_recipe and is now a dictionary and not a callable. The ood_criterion and the calibration sets are now strings.

Metrics

The NegativeLogLikelihood metric is renamed CategoricalNLL.

Baselines

All baselines have been renamed to explicitly contain "Baselines" in their name.

Tutorials

We have rewritten and updated the tutorials should now be clearer. Send us feedback!

What's Changed

• ➖ Avoid using Argvcontext in tutorials by @o-laurent in #82
• 🚀 Upgrade to Lightning 2.0 by @alafage in #79
• 🚀 Update to Lightning 2.0, Add Segmentation, & Rework Regression by @o-laurent in #85

Full Changelog: v0.1.6...v0.2.0

v0.1.6 Add Grouping Loss, MC-BN, OpenImage-O & MUAD

What's Changed

• ⬆️ Bump tj-actions/changed-files from 34 to 41 in /.github/workflows by @dependabot in #75
• ✨ Add ResNet-20, corruptions and improve docs by @o-laurent in #76
• ✨ Add the grouping loss to single model training by @o-laurent in #77
• ✨ Add Monte-Carlo Batch Normalization by @o-laurent in #78
• ✨ Add grouping loss, Monte-Carlo Batch Normalization, OpenImage-O, MUAD & Improve code quality by @o-laurent in #80

Full Changelog: v0.1.5...v0.1.6