v1.3.0

Current 1.3 (stable)

Enhancements

• Adding Hugging Face Hub integration documentation with comprehensive examples for
  loading pre-trained models (:class:braindecode.models.BIOT,
  :class:braindecode.models.Labram, :class:braindecode.models.REVE,
  :class:braindecode.models.LUNA, :class:braindecode.models.BENDR,
  :class:braindecode.models.SignalJEPA) and pushing models to the Hub (:gh:879 by
  Bruno Aristimunha_)
• Adding :class:braindecode.models.REVE with comprehensive documentation covering
  architecture details, 4D positional encoding, and pretraining scale (:gh:866 by
  Jonathan Lys_ )
• Adding augmentation :class:braindecode.augmentation.AmplitudeScale (:gh:848 by
  Bruna Lopes_)
• Adding augmentation :class:braindecode.augmentation.ChannelsReref (:gh:848 by
  Bruna Lopes_)
• Added Patched Brain Transformer from Klein T et al 2025 at
  :class:braindecode.models.PBT (:gh:787 by José Mauricio_)
• Including typing in the augmentation module (:gh:709 by Aphel_)
• Adding :class:braindecode.models.SSTDPN (:gh:790 by Can Han_ and Bruno Aristimunha_ )
• Adding :class:braindecode.models.BrainModule (:gh:743 by Bruno Aristimunha_ )
• Added :class:braindecode.models.BENDR (:gh:802 by Bruno Aristimunha_ )
• Adding a tutorial for fine-tuning a pre-trained foundation model (:gh:794 by Pierre Guetschel_)
• Added 25 new MNE preprocessing functions, expanding from 6 to 31 total preprocessing
  classes. New functions include: :class:braindecode.preprocessing.NotchFilter,
  :class:braindecode.preprocessing.SavgolFilter,
  :class:braindecode.preprocessing.ApplyHilbert,
  :class:braindecode.preprocessing.InterpolateBads,
  :class:braindecode.preprocessing.InterpolateTo,
  :class:braindecode.preprocessing.RenameChannels,
  :class:braindecode.preprocessing.ReorderChannels,
  :class:braindecode.preprocessing.EqualizeChannels,
  :class:braindecode.preprocessing.AddChannels,
  :class:braindecode.preprocessing.SetMontage,
  :class:braindecode.preprocessing.AddReferenceChannels,
  :class:braindecode.preprocessing.AddProj,
  :class:braindecode.preprocessing.ApplyProj,
  :class:braindecode.preprocessing.DelProj,
  :class:braindecode.preprocessing.ComputeCurrentSourceDensity,
  :class:braindecode.preprocessing.FixStimArtifact,
  :class:braindecode.preprocessing.CropByAnnotations,
  :class:braindecode.preprocessing.Anonymize,
  :class:braindecode.preprocessing.SetAnnotations,
  :class:braindecode.preprocessing.SetChannelTypes,
  :class:braindecode.preprocessing.SetMeasDate,
  :class:braindecode.preprocessing.FixMagCoilTypes,
  :class:braindecode.preprocessing.ApplyGradientCompensation,
  :class:braindecode.preprocessing.AddEvents, and
  :class:braindecode.preprocessing.Rescale (:gh:801 by Bruno Aristimunha_)
• Adding artifact removal preprocessing steps, incl.
  :class:braindecode.preprocessing.EEGPrep (:gh:796 by Christian Kothe_)
• Including huggingface hub mixin to all models (:gh:795 by Kuntal Kokate_ )
• Better inheritance between the different dataset classes (:gh:806 by Pierre Guetschel_ )
• Fix minor documentation issues in Labram (:gh:808 by Matthew Chen_)
• Including missed tag in all the models and creating some test for it (:gh:811 by
  Bruno Aristimunha_ )
• Added :class:braindecode.models.LUNA (:gh:802 by Bruno Aristimunha, Thorir Mar Ingolfsson)
• Improve :class:braindecode.preprocessing.Preprocessor serialisation (:gh:813 by
  Pierre Guetschel_ )
• Modify verbosity of create_window_from_events (:gh:814 by Matthew Chen_)
• Including SIENA dataset and CHB_MIT datasets (:gh:818 by Jonathan Dan_ and Bruno Aristimunha_ )
• Modify the interface of FBMSNet for consistency with FBCNet + Unit Tests (:gh:819 by
  Matthew Chen_)
• Adding :class:braindecode.models.MEDFormer (:gh:821 by Bruno Aristimunha_ )
• Adding :class:braindecode.models.EEGSym model (:gh:697 by Pierre Guetschel_ and
  Bruno Aristimunha_ )
• Adding tutorial for
  :func:braindecode.preprocessing.windowers.create_fixed_length_windows (:gh:746 by
  Aniela Bulicz and Arina Schippers and Aron van Zuijlen)
• Including interface with Hugging Face Hub for dataset sharing and loading (:gh:820
  by Kuntal Kokate_ and Bruno Aristimunha_ )
• Adding a CITATION.cff file, allowing researchers to directly cite the Braindecode
  library via a Zenodo DOI (:gh:847 and :gh:872 by Pierre Guetschel_ )
• Add automatically-generated pydantic models configs (:gh:850 by Pierre Guetschel_
  )
• Refactor: standardized parameter names in transformer and LabRam modules (:gh:850 by
  José Mauricio_)
• Added a tutorial for temporal generalization (:gh:857 by Matthew Chen_)
• Including colab buttons to all tutorials (:gh:861 by Bruno Aristimunha_ )
• Adding example Pydantic and Exca (:gh:858 by Pierre Guetschel_)
• Renaming the model category "Foundation Models" from "Large Brain Models (LBM)"
  (:gh:867 by Bruno Aristimunha_ )
• Make the dataset at huggingface hub more bids compatible (:gh:871 by Bruno Aristimunha_ )
• Reorganized BIDS-related modules into :py:mod:braindecode.datasets.bids subpackage
  for better code organization. The subpackage includes: dataset loading
  (datasets.py), Hub format utilities (hub_format.py), Hub integration
  (hub.py), and validation utilities (hub_validation.py) (:gh:871 by Bruno Aristimunha_)
• Changed metadata storage in Zarr format from JSON attributes to TSV files
  (metadata.tsv) for better scalability with large datasets (:gh:871 by Bruno Aristimunha_)
• Improved JSON serialization in zarr.json by storing description and info as
  proper JSON objects instead of stringified JSON, with NaN/Inf sanitization for valid
  JSON output (:gh:871 by Bruno Aristimunha_)

API changes

• BIDS and Hub modules moved to :py:mod:braindecode.datasets.bids subpackage:
  braindecode.datasets.bids.hub, braindecode.datasets.bids.hub_format,
  braindecode.datasets.bids.datasets, braindecode.datasets.bids.hub_validation
  (:gh:871 by Bruno Aristimunha_)

• Deprecating the old naming of MOABB Dataset name (:gh:826 by Bruno Aristimunha_)

• Exposing the :func:braindecode.datautil.infer_signal_properties utility function
  (:gh:856 by Pierre Guetschel_)

• Deprecating the old naming of MOABB Dataset name :gh:826 by Bruno Aristimunha_

• Drop support for Python 3.10 and increase support to Python 3.13 and python 3.14
  (:gh:840 by Bruno Aristimunha_)

  • Model config helpers now soft-import pydantic/numpydantic; if the optional
    dependencies are missing the module skips config generation and warns to install
    pip install braindecode[pydantic].

Bugs

• Fixing the issues with :class:braindecode.models.ATCNet (:gh:782 by Hamdi Altaheri_)
• Fixing expired path when loading a preprocessed dataset that was moved (:gh:786 by
  Pierre Guetschel_)
• Fix OverflowError sometimes when saving description to a json file (:gh:789 by
  Pierre Guetschel_)
• Small adjustment in :class:braindecode.models.Labram model to allow pre-loading of
  weights (:gh:792 by Bruno Aristimunha_)
• Fixing many warnings for generating the documentation (:gh:805 by Bruno Aristimunha_)
• Fix test compatibility with scikit-learn 1.8+ by adding __sklearn_tags__ support
  to test mocks (:gh:868 by Bruno Aristimunha_)

Release 1.2.0

Check the whats new file

v1.1.1

What's Changed

• Fix whats_new by @PierreGtch in #749
• [FIX] Fix Deep4Net initialization issue by @bruAristimunha in #752
• [FIX] Initialization issue with atcnet. by @bruAristimunha in #754
• [FIX] Update and fix various neural network models by @bruAristimunha in #756
• [FIX] Labram with small input by @bruAristimunha in #757
• Consolidate pre-commit repositories and update configurations by @bruAristimunha in #759
• [FIX] Embedding size parameters should be index based on the conv and depth parameters by @bruAristimunha in #760
• Optimize installation time by updating dependencies by @bruAristimunha in #761
• Release 1.1.0 by @bruAristimunha in #762
• Release 1.1.1 by @bruAristimunha in #764

Full Changelog: v1.0...v1.1.1

v1.1

What's Changed

• Fix whats_new by @PierreGtch in #749
• [FIX] Fix Deep4Net initialization issue by @bruAristimunha in #752
• [FIX] Initialization issue with atcnet. by @bruAristimunha in #754
• [FIX] Update and fix various neural network models by @bruAristimunha in #756
• [FIX] Labram with small input by @bruAristimunha in #757
• Consolidate pre-commit repositories and update configurations by @bruAristimunha in #759
• [FIX] Embedding size parameters should be index based on the conv and depth parameters by @bruAristimunha in #760
• Optimize installation time by updating dependencies by @bruAristimunha in #761

Full Changelog: v1.0...v1.1

v1.0

What's Changed

• V0.8 by @bruAristimunha in #557
• Fix annotations by @dcwil in #563
• [MNT] changing the whats new checker and fixing tests with deprecated things by @bruAristimunha in #569
• [MRG] Adding BIOT to braindecode by @bruAristimunha in #573
• Allow inferring signal parameters from torch tensors by @PierreGtch in #574
• [MRG] Creating tests to ensure that all the models work with default parameters, mapping the mandatory parameters for the models by @bruAristimunha in #571
• Adding Labram by @bruAristimunha in #578
• [ENH] Add new model: EEG-SimpleConv by @elouayas in #581
• adding NMT dataset by @MohammadJavadD in #443
• [MNT] Black codebase by @bruAristimunha in #584
• Adding extra options for MOABB dataset by @bruAristimunha in #587
• ENH Add type hints to datasets by @PierreGtch in #590
• MNT trying to solve the issue with test by @bruAristimunha in #591
• EHN Increase test coverage by @bruAristimunha in #592
• MNT temporary solution for MNE version by @bruAristimunha in #594
• [WIP] Adding Attention Base Net by @bruAristimunha in #572
• [MRG] TUH: standardize channel name and define their positions by @PierreGtch in #593
• Add offset arg to preproces function by @PierreGtch in #599
• Add type hints to preprocessing functions by @PierreGtch in #600
• Update nmt.py by @MohammadJavadD in #601
• Add mypy to CI by @PierreGtch in #606
• Preprocessing EEGWindowsDataset by @PierreGtch in #607
• Fix dead link & Sphinx URL formatting by @OverLordGoldDragon in #602
• [MNT] Fix matplotlib colormaps deprecation by @bruAristimunha in #617
• Adding SPARCNet and ContraWR by @bruAristimunha in #611
• MNT Ensure mypy to work for every commit by @bruAristimunha in #619
• Add SegmentationReconstruction augmentation by @gustavohenriquesr in #608
• Code clarity changes in windowers.py by @OverLordGoldDragon in #615
• MNT Optimizing the CI by @bruAristimunha in #620
• Lazy index for fixed length windower by @PierreGtch in #597
• Deprecate moabb 1.0.0 by @PierreGtch in #627
• Log Softmax as False by @bruAristimunha in #624
• MNT Removing MNE version pin by @bruAristimunha in #629
• Add MaskEncoding augmentation by @gustavohenriquesr in #631
• MNT Increasing the moabb version by @bruAristimunha in #632
• EHN: Including EEGNex by @bruAristimunha in #635
• EHN: Ensure consistency in the last layer using tests by @bruAristimunha in #642
• MNT: removing deprecated EEGInception and deprecating EEGNetV1 by @bruAristimunha in #639
• Exposing activation function by @bruAristimunha in #637
• EHN: including EEGTCNet by @bruAristimunha in #640
• Adding SleepPhysionetChallenge2018 dataset by @bruAristimunha in #621
• MNT: solving windows CI. by @bruAristimunha in #644
• EHN: Including Syncnet by @bruAristimunha in #643
• DOC: Solving issues with Sphinx by @bruAristimunha in #646
• [EHN] Normalizing the parameters for dropout by @bruAristimunha in #654
• [EHN]: Add TSInceptionV1 by @bruAristimunha in #641
• [WIP] No more relative imports and moving some functions by @bruAristimunha in #653
• FIX: small merge issue from #653 by @bruAristimunha in #655
• [MNT] Updating the only tutorial that doesn't run on CI (Lazy benchmark) by @bruAristimunha in #652
• EHN: Including pytorch audio by @bruAristimunha in #661
• [EHN] Filterbank as module by @bruAristimunha in #656
• FIX: Suppress TUHAbnormal warnings when running in parallel by @itsaphel in #670
• [MNT] Bye old parameters by @bruAristimunha in #672
• [ENH] Including CTNet into braindecode by @bruAristimunha in #666
• [MNT] or [EHN] (?) Exposing EEGITNet parameters by @bruAristimunha in #673
• [EHN] EEGMiner by @bruAristimunha in #667
• [MNT] EEGNetv4 FIX by @bruAristimunha in #676
• [EHN] Including SCC by @bruAristimunha in #679
• [EHN] Including MSVTNet by @bruAristimunha in #659
• [EHN] Inclusion of SincShallowNet by @bruAristimunha in #678
• [MNT] Including figures in the docstring of all models that I could find by @bruAristimunha in #682
• [MNT] Including figures by @bruAristimunha in #683
• [MNT] TCN and Hybrid are not longer models by @bruAristimunha in #674
• [MNT] Adjusting doc from EEGMiner by @bruAristimunha in #685
• MTN unpin sckit-learn after skorch release by @tomMoral in #688
• Bump actions/upload-artifact to v4 by @itsaphel in #691
• fix joblib usage for NMT dataset by @itsaphel in #690
• MTN fix mne caching in doc build by @tomMoral in #693
• [MNT] updating the pre-commit and applying by @bruAristimunha in #699
• [MNT] Eegnetv4 fix v2 by @bruAristimunha in #680
• [FIX] Update preprocess.py for Windows compatibility by @arnodelorme in #700
• [FIX] Making the MOABB dataset work with BIDS cache by @bruAristimunha in #694
• Add BIDSDataset by @PierreGtch in #701
• Add BIDSDataset to doc by @PierreGtch in #702
• Distributed samplers for recording and relative positioning sampling by @dungscout96 in #695
• Include Signal-JEPA models by @PierreGtch in #703
• [MNT] updating the CI by @bruAristimunha in #705
• Small fixes on BIDS datasets by @PierreGtch in #704
• Fixes on sJEPA and BIDS docstrings by @PierreGtch in #706
• [EHN] Allowing Slicedataset inference by @bruAristimunha in #716
• [EHN] Including model list into documentation by @lucas-heck in #715
• Include summary.csv in package data and update MANIFEST.in by @Samuel-Boehm in #719
• Enhance model summary page with structured layout, tags, and dropdown filters by @Ganasekhar-gif in #718
• [EHN] FBCNet model by @bruAristimunha in #722
• [EHN] FBLightConvNet by @bruAristimunha in #723
• Add FBMSNet model implementation by @bruAristimunha in #724
• Add IFNet model to the library by @bruAristimunha in #726
• DOC: Fixing broken links and warnings by @lucas-heck in #727
• Refactor and update modules for improved organization and functionality by @bruArist...

v0.8.1

See changes at https://braindecode.org/stable/whats_new.html

New Contributors

• @Sara04
• @ZamboniMarco99
• @PierreGtch
• @tgy
• @matthieutrs
• @remidbs
• @brunaafl
• @eeyhsong
• @ostormer

Version 0.7.0

See changes at https://braindecode.org/stable/whats_new.html

What's Changed

• Update base.py by @MohammadJavadD in #371
• Adding source code button for documentation. by @bruAristimunha in #380
• [MRG] Fix predict_trials by @gemeinl in #381
• [WIP] Move stored preds in callbacks to CPU by @robintibor in #384
• Changing the name of function for sklearn compatibility. by @bruAristimunha in #388
• [WIP] Adding docstrings for models by @dcwil in #281
• [WIP] Data augmentation search by @bruAristimunha in #389
• [WIP] Scikit-learn hyperparameter tuning example by @gemeinl in #349
• [WIP] Nicer more structured docs by @robintibor and @bruAristimunha in #272
• More natural phrasing in README.rst by @cedricrommel in #396
• WIP: added unified validation scheme by @martinwimpff and @bruAristimunha in #397
• added verbosity args to windowers by @gemeinl in #391
• Fix torch.linalg issue in rotation augmentations by @cedricrommel in #398
• Resolving some Sphinx Doc Warnings. Related to Issue #266 by @bruAristimunha in #403
• adding EEG-ITNET model by @GhBlg, @bruAristimunha and @sliwy in #400
• Updating What's New, related with PR #391. by @bruAristimunha in #407
• Adding a new parameter in the class AugmentedDataloader to send transformation to the device. by @bruAristimunha in #406
• [WIP] EEG-Inception by @bruAristimunha in #390
• bugfix ft_surrogate by @martinwimpff in #409
• Removing subject_ids argument in TUHAbnormal example. by @bruAristimunha in #402
• Removing functions by @bruAristimunha in #415
• [WIP] Drop last in train by @robintibor in #411
• Updated depreciation by @bruAristimunha in #413
• Updating the versions.json. by @bruAristimunha in #416
• V0.7 release by @bruAristimunha in #421
• Updating Path to CI by @bruAristimunha in #423

New Contributors

• @MohammadJavadD made their first contribution in #371
• @bruAristimunha made their first contribution in #380
• @martinwimpff made their first contribution in #397
• @GhBlg made their first contribution in #400

Full Changelog: v0.6.0...v0.7.0

Version 0.6.0

See changes at https://braindecode.org/whats_new.html

Version 0.5.1

See Changes at https://braindecode.org/whats_new.html

Version 0.5.0

First release on PyPI.