v0.6.0a2

What's Changed

This release adds functionality to run inference on datasets that do not fully fit into memory (batched inference, #168 ) and adds support to train unified encoders across multiple sessions in a dataset (#251 ).

🚧 A preprint is pending "Unified CEBRA Encoders for Integrating Neural Recordings via Behavioral Alignment" by Célia Benquet (@CeliaBenquet ), Hossein Mirzaei (@hsirm) , Steffen Schneider (@stes), Mackenzie W. Mathis (@MMathisLab ).

💻 A DEMO Notebook is available at: https://cebra.ai/docs/demos.html

• Fix quantized_embedding_norm undefined when normalize=False by @CeliaBenquet in #249
• Batched inference CEBRA & padding at the Solver level by @CeliaBenquet in #168
• Add unified CEBRA encoder: pytorch implementation by @CeliaBenquet in #251
• Release 0.6.0a2 by @stes in #256

Full Changelog: v0.6.0a1...v0.6.0a2

v0.6.0a1

What's Changed

• Add xCEBRA implementation (AISTATS 2025) by @stes @gonlairo et al. & @MMathisLab in #225

• • this code accompanies https://openreview.net/forum?id=aGrCXoTB4P

• Upgrade docs build by @stes in #241
• Allow indexing of the cebra docs by @stes in #242
• Fix broken docs coverage workflows by @stes in #246
• Update conf.py by @MMathisLab in #237
• Update docs.yml to build from main by @MMathisLab in #248
• Update installation.rst --> add link to docker hub by @MMathisLab in #247

Full Changelog: v0.5.0...v0.6.0a1

v0.5.0

What's Changed

• Fix pandas intersphinx mapping; fix test setup by @stes in #136
• Implementation for discrete multisession by @introspective-swallow in #135
• Update imports in pivae from keras to tensorflow.keras by @MMathisLab in #164
• Upgrade dependencies and fix testsuite by @stes in #172
• Add device as arg to TensorDataset by @stes in #170
• Update installation.rst by @MMathisLab in #163
• Add OS indepdent paths for monkey, hippocampus and synthetic data by @stes in #169
• Change pytorch version in docs to stable by @CeliaBenquet in #176
• Fix tests for upgraded dependencies by @stes in #184
• Add ruff for linting the codebase by @stes in #167
• Extend type checking to all float datatypes by @CeliaBenquet in #166
• Fix linting issues by @stes in #185
• Set default offset to an Offset object by @icarosadero in #180
• Add additional tests for TensorDataset by @stes in #187
• Fix linting errors in tests by @stes in #188
• Fix scikit-learn reference in conda environment files by @stes in #195
• Add support for new sklearn_tags by @stes in #205
• Update workflows to actions/setup-python@v5, actions/cache@v4 by @stes in #212
• Fix deprecation warning force_all_finite -> ensure_all_finite for sklearn>=1.6 by @icarosadero in #206
• Add tests to check legacy model loading by @stes in #214
• Add improved goodness of fit implementation by @stes in #190
• Support numpy 2, upgrade tests to support torch 2.6 by @stes in #221
• Release 0.5.0rc1 by @stes in #189
• Fix pypi action by @stes in #222
• Update base.py by @icarosadero in #224
• Change max consistency value to 100 instead of 99 by @CeliaBenquet in #227
• Update assets.py --> force check for parent dir by @MMathisLab in #230
• User docs minor edit by @MMathisLab in #229
• General Doc refresher by @MMathisLab in #232
• render plotly in our docs, show code/doc version by @MMathisLab in #231
• Update layout.html by @MMathisLab in #233
• Update conf.py by @MMathisLab in #234
• Refactoring setup.cfg by @MMathisLab in #228
• Home page landing update by @MMathisLab in #235
• v0.5.0 by @MMathisLab in #238

New Contributors

• @introspective-swallow made their first contribution in #135
• @icarosadero made their first contribution in #180

Full Changelog: v0.4.0...v0.5.0

v0.5.0rc1

What's Changed

• Fix pandas intersphinx mapping; fix test setup by @stes in #136
• Implementation for discrete multisession by @introspective-swallow in #135
• Update imports in pivae from keras to tensorflow.keras by @MMathisLab in #164
• Upgrade dependencies and fix testsuite by @stes in #172
• Add device as arg to TensorDataset by @stes in #170
• Update installation.rst by @MMathisLab in #163
• Add OS indepdent paths for monkey, hippocampus and synthetic data by @stes in #169
• Change pytorch version in docs to stable by @CeliaBenquet in #176
• Fix tests for upgraded dependencies by @stes in #184
• Add ruff for linting the codebase by @stes in #167
• Extend type checking to all float datatypes by @CeliaBenquet in #166
• Fix linting issues by @stes in #185
• Set default offset to an Offset object by @icarosadero in #180
• Add additional tests for TensorDataset by @stes in #187
• Fix linting errors in tests by @stes in #188
• Fix scikit-learn reference in conda environment files by @stes in #195
• Add support for new sklearn_tags by @stes in #205
• Update workflows to actions/setup-python@v5, actions/cache@v4 by @stes in #212
• Fix deprecation warning force_all_finite -> ensure_all_finite for sklearn>=1.6 by @icarosadero in #206
• Add tests to check legacy model loading by @stes in #214
• Add improved goodness of fit implementation by @stes in #190
• Support numpy 2, upgrade tests to support torch 2.6 by @stes in #221
• Release 0.5.0rc1 by @stes in #189

New Contributors

• @introspective-swallow made their first contribution in #135
• @icarosadero made their first contribution in #180

Full Changelog: v0.4.0...v0.5.0rc1

v0.4.0

What's Changed

Starting with v0.4.0, CEBRA is released under an Apache 2.0 license.

• Add missing error raising and test for adapt on Multissession by @CeliaBenquet in #114
• Change license to Apache 2.0 by @MMathisLab and @stes in #117
• Apply updated license headers by @stes in #118
• Update setup.cfg license classifier by @stes in #119
• Update README.md by @MMathisLab in #120
• Pass showlegend/template kwargs to plotly by @nastya236 in #122

Full Changelog: v0.3.1...v0.4.0

v0.3.1

What's Changed

• Auto-import plotly integration by @MMathisLab in #96
• Update usage.rst by @MMathisLab in #98
• Simplify plotly docstring example by @stes in #102
• Restructure API docs for integrations by @stes in #103
• Release v0.3.1rc1 by @stes in #104
• Fix windows tests by @gonlairo in #90
• Add torch API usage example by @timonmerk in #99
• Update icon link for dandi demo data by @MMathisLab in #113
• Update cebra_paper_m1.yml and cebra.yml by @MMathisLab in #112
• Release 0.3.1 by @MMathisLab and @stes in #110

New Contributors

• @timonmerk made their first contribution in #99

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

v0.3.0

What's Changed

• Update usage.rst docs by @MMathisLab in #21
• Remove subsample default in OrthogonalProcrustesAlignment, improve tests by @stes in #28
• Remove grid_search_models/ after running tests by @stes in #26
• Fix action workflow definitions for external contributors by @stes in #31
• For docs build, limit scikit-learn version by @stes in #37
• Temporarily disable PR comment to report doc coverage by @stes in #38
• Update repo link on landing page by @stes in #36
• Fix typo in pi_vae.py by @eltociear in #20
• Automatically download the rat-hippocampus dataset by @gonlairo in #35
• Fix dtypes handled in the labels of the decoders by @CeliaBenquet in #30
• Temporarily disable test runs on dev repo by @stes in #40
• Add ... pattern to copybutton config by @stes in #41
• Expose n_bins argument from align_embeddings by @stes in #25
• Add pre-commit config to standardize files by @stes in #42
• Fix Device in Multisession Training by @gonlairo in #44
• Support More Data Types in the sklearn Dataset and Decoder by @gonlairo in #43
• Improve device support and add support for Apple Silicon chipset (mps) by @gonlairo in #34
• Add NLB tools pypi package back as dataset dependency by @stes in #50
• Add missing mark in test_datasets.py by @gonlairo in #52
• Add Matplotlib ImportError when using compare_models() by @gonlairo in #53
• Fix test failures due to http rate limit by @stes in #62
• Autodownload monkey reaching data by @gonlairo in #59
• add demo link to homepage by @MMathisLab in #64
• Fix type bug in to() method by @sofiagilardini in #55
• Release 0.3.0rc1 by @stes in #66
• Autodownload synthetic data by @gonlairo in #68
• Add thumbnail for openscope notebook by @stes in #71
• Improve saving and loading of models by @gonlairo in #69
• Update installation.rst by @MMathisLab in #80
• Improve documentation for local docs build by @stes in #70
• Extend Delta distribution to multiple dimensions by @gonlairo in #76
• Fix broadcasting in InfoNCE loss by @stes in #86
• Enhance ModuleNotFoundError messages by @stes in #85
• Add OS independent paths for Allen dataset by @gonlairo in #88
• Fix consistency labels ordering and simplify by @CeliaBenquet in #87
• Inspired by @obarnstedt work at Lindoscope workshop we added interactive plots using plotly by @nastya236 in #82
• Release v0.3.0rc2 by @stes in #67
• Pin tables <= 3.8 by @stes in #91
• Release v0.3.0 by @stes in #92

Contributors

• @nastya236 made their first contribution in #82
• @eltociear made their first contribution in #20
• @gonlairo made their first contribution in #35
• @CeliaBenquet made their first contribution in #30
• @sofiagilardini made their first contribution in #55

Full Changelog: v0.2.0...v0.3.0