braindecode 1.8.1.dev182015723__tar.gz → 1.8.1.dev182039595__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (211) hide show
  1. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/NOTICE.txt +4 -0
  2. {braindecode-1.8.1.dev182015723/braindecode.egg-info → braindecode-1.8.1.dev182039595}/PKG-INFO +1 -1
  3. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/augmentation/__init__.py +2 -0
  4. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/augmentation/functional.py +55 -2
  5. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/augmentation/transforms.py +52 -1
  6. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/base.py +236 -12
  7. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/mne.py +5 -0
  8. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/tuh.py +22 -14
  9. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datautil/serialization.py +21 -10
  10. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/eegneuralnet.py +6 -1
  11. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/functional/_real_dft.py +55 -0
  12. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/functional/functions.py +47 -18
  13. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/__init__.py +8 -0
  14. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/attn_sleep.py +9 -11
  15. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/axon.py +54 -13
  16. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/barista.py +25 -13
  17. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/base.py +56 -15
  18. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/biot.py +7 -2
  19. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/brainmodule.py +13 -1
  20. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/brainomni.py +14 -4
  21. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/brant.py +23 -11
  22. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/codebrain.py +132 -19
  23. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/contrawr.py +23 -1
  24. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/csbrain.py +93 -44
  25. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/ctnet.py +9 -1
  26. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/deep4.py +10 -1
  27. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/dgcnn.py +7 -0
  28. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/diver1.py +51 -59
  29. braindecode-1.8.1.dev182039595/braindecode/models/duin.py +454 -0
  30. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eeg_clip.py +71 -9
  31. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegconformer.py +19 -3
  32. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eeginception_mi.py +7 -3
  33. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegnex.py +20 -1
  34. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegpt.py +15 -9
  35. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegsym.py +85 -17
  36. braindecode-1.8.1.dev182039595/braindecode/models/epint.py +315 -0
  37. braindecode-1.8.1.dev182039595/braindecode/models/gcnsnet.py +338 -0
  38. braindecode-1.8.1.dev182039595/braindecode/models/guetschel2026.py +775 -0
  39. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/labram.py +34 -15
  40. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/luna.py +18 -17
  41. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/mapa.py +8 -2
  42. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/medformer.py +68 -33
  43. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/meta_neuromotor.py +57 -32
  44. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/mirepnet.py +15 -4
  45. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/mscformer.py +21 -1
  46. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/msvtnet.py +123 -6
  47. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/mvpformer.py +46 -56
  48. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/neurorvq.py +135 -34
  49. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/neurorvq_tokenizer.py +87 -63
  50. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/patchedtransformer.py +3 -8
  51. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/reve.py +129 -33
  52. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/signal_jepa.py +87 -4
  53. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sinc_shallow.py +2 -1
  54. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sleep_stager_blanco_2020.py +13 -6
  55. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sleepfm.py +62 -12
  56. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/steegformer.py +16 -9
  57. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/summary.csv +4 -0
  58. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/tcn.py +5 -0
  59. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/tfm_tokenizer.py +44 -3
  60. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/tidnet.py +18 -1
  61. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/usleep.py +3 -2
  62. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/util.py +147 -33
  63. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/zuna.py +17 -9
  64. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/__init__.py +2 -1
  65. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/attention.py +124 -15
  66. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/blocks.py +54 -5
  67. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/channels.py +77 -0
  68. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/dance_modules.py +22 -10
  69. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/filter.py +79 -44
  70. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/layers.py +41 -2
  71. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/quantization.py +17 -13
  72. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/preprocessing/preprocess.py +30 -0
  73. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/preprocessing/windowers.py +5 -4
  74. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/samplers/base.py +60 -61
  75. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/training/losses.py +5 -4
  76. braindecode-1.8.1.dev182039595/braindecode/version.py +1 -0
  77. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595/braindecode.egg-info}/PKG-INFO +1 -1
  78. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode.egg-info/SOURCES.txt +4 -0
  79. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/api.rst +14 -0
  80. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/user_guide/channel_strategies.rst +1 -1
  81. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/whats_new.rst +187 -0
  82. braindecode-1.8.1.dev182015723/braindecode/version.py +0 -1
  83. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/LICENSE.txt +0 -0
  84. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/MANIFEST.in +0 -0
  85. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/README.rst +0 -0
  86. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/__init__.py +0 -0
  87. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/augmentation/base.py +0 -0
  88. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/classifier.py +0 -0
  89. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/__init__.py +0 -0
  90. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/_notebook_viewer.py +0 -0
  91. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bbci.py +0 -0
  92. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bcicomp.py +0 -0
  93. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/__init__.py +0 -0
  94. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/datasets.py +0 -0
  95. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/format.py +0 -0
  96. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/hub.py +0 -0
  97. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/hub_format.py +0 -0
  98. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/hub_io.py +0 -0
  99. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/hub_validation.py +0 -0
  100. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/bids/iterable.py +0 -0
  101. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/chb_mit.py +0 -0
  102. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/collate.py +0 -0
  103. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/moabb.py +0 -0
  104. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/nmt.py +0 -0
  105. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/registry.py +0 -0
  106. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/siena.py +0 -0
  107. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/sleep_physio_challe_18.py +0 -0
  108. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/sleep_physionet.py +0 -0
  109. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/utils.py +0 -0
  110. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datasets/xy.py +0 -0
  111. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datautil/__init__.py +0 -0
  112. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datautil/channel_utils.py +0 -0
  113. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datautil/hub_formats.py +0 -0
  114. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/datautil/util.py +0 -0
  115. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/functional/__init__.py +0 -0
  116. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/functional/initialization.py +0 -0
  117. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/atcnet.py +0 -0
  118. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/attentionbasenet.py +0 -0
  119. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/bendr.py +0 -0
  120. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/brainbert.py +0 -0
  121. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/cbramod.py +0 -0
  122. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/config.py +0 -0
  123. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/dance.py +0 -0
  124. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/deepsleepnet.py +0 -0
  125. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegdino.py +0 -0
  126. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eeginception_erp.py +0 -0
  127. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegitnet.py +0 -0
  128. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegminer.py +0 -0
  129. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegnet.py +0 -0
  130. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegsimpleconv.py +0 -0
  131. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/eegtcnet.py +0 -0
  132. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/emg2qwerty.py +0 -0
  133. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/fbcnet.py +0 -0
  134. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/fblightconvnet.py +0 -0
  135. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/fbmsnet.py +0 -0
  136. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/hybrid.py +0 -0
  137. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/ifnet.py +0 -0
  138. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/neuropose.py +0 -0
  139. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/popt.py +0 -0
  140. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sccnet.py +0 -0
  141. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/seizure_transformer.py +0 -0
  142. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sensingdynamics.py +0 -0
  143. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/shallow_fbcsp.py +0 -0
  144. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sleep_stager_chambon_2018.py +0 -0
  145. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sparcnet.py +0 -0
  146. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/sstdpn.py +0 -0
  147. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/syncnet.py +0 -0
  148. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/tcformer.py +0 -0
  149. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/tmsanet.py +0 -0
  150. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/tsinception.py +0 -0
  151. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/models/vemg2pose.py +0 -0
  152. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/activation.py +0 -0
  153. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/convolution.py +0 -0
  154. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/linear.py +0 -0
  155. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/linear_attention.py +0 -0
  156. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/parametrization.py +0 -0
  157. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/popt_modules.py +0 -0
  158. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/stats.py +0 -0
  159. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/util.py +0 -0
  160. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/modules/wrapper.py +0 -0
  161. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/preprocessing/__init__.py +0 -0
  162. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/preprocessing/eegprep_preprocess.py +0 -0
  163. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/preprocessing/mne_preprocess.py +0 -0
  164. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/preprocessing/util.py +0 -0
  165. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/regressor.py +0 -0
  166. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/samplers/__init__.py +0 -0
  167. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/samplers/ssl.py +0 -0
  168. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/training/__init__.py +0 -0
  169. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/training/callbacks.py +0 -0
  170. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/training/scoring.py +0 -0
  171. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/util.py +0 -0
  172. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/__init__.py +0 -0
  173. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/activations.py +0 -0
  174. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/attribution.py +0 -0
  175. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/confusion_matrices.py +0 -0
  176. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/frequency.py +0 -0
  177. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/metrics.py +0 -0
  178. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/sanity.py +0 -0
  179. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode/visualization/topology.py +0 -0
  180. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode.egg-info/dependency_links.txt +0 -0
  181. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode.egg-info/requires.txt +0 -0
  182. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/braindecode.egg-info/top_level.txt +0 -0
  183. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/Makefile +0 -0
  184. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/_templates/autosummary/class.rst +0 -0
  185. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/_templates/autosummary/class_in_subdir.rst +0 -0
  186. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/_templates/autosummary/function.rst +0 -0
  187. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/_templates/autosummary/function_in_subdir.rst +0 -0
  188. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/agent_docs.rst +0 -0
  189. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/cite.rst +0 -0
  190. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/conf.py +0 -0
  191. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/help.rst +0 -0
  192. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/index.rst +0 -0
  193. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/install/install.rst +0 -0
  194. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/install/install_pip.rst +0 -0
  195. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/install/install_source.rst +0 -0
  196. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/attention.rst +0 -0
  197. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/channel.rst +0 -0
  198. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/convolution.rst +0 -0
  199. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/filterbank.rst +0 -0
  200. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/gnn.rst +0 -0
  201. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/interpretable.rst +0 -0
  202. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/lbm.rst +0 -0
  203. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/recurrent.rst +0 -0
  204. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/categorization/spd.rst +0 -0
  205. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/models.rst +0 -0
  206. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/models_categorization.rst +0 -0
  207. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/models_table.rst +0 -0
  208. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/models/models_visualization.rst +0 -0
  209. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/docs/sg_execution_times.rst +0 -0
  210. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/pyproject.toml +0 -0
  211. {braindecode-1.8.1.dev182015723 → braindecode-1.8.1.dev182039595}/setup.cfg +0 -0
@@ -40,6 +40,7 @@ The following components are licensed under the MIT License:
40
40
 
41
41
  - `braindecode/models/ctnet.py`
42
42
  - `braindecode/models/dance.py`
43
+ - `braindecode/models/duin.py`
43
44
  - `braindecode/models/medformer.py`
44
45
  - `braindecode/models/tcformer.py`
45
46
  - `braindecode/models/ifnet.py`
@@ -52,7 +53,10 @@ The following components are licensed under the MIT License:
52
53
  - `braindecode/modules/linear_attention.py`
53
54
  - `braindecode/models/seizure_transformer.py`
54
55
  - `docs/_static/model/seizure_transformer_arch.png`
56
+ - `braindecode/models/guetschel2026.py`
55
57
  - `braindecode/models/tmsanet.py`
58
+ - `braindecode/models/epint.py`
59
+ - `braindecode/models/gcnsnet.py`
56
60
 
57
61
  The `brainomni.py` file is a derivative of `OpenTSLab/BrainOmni` (MIT), whose
58
62
  SEANet codec derives from Meta's `facebookresearch/encodec` (MIT). The
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: braindecode
3
- Version: 1.8.1.dev182015723
3
+ Version: 1.8.1.dev182039595
4
4
  Summary: Deep learning software to decode EEG, ECG or MEG signals
5
5
  Author-email: Robin Tibor Schirrmeister <robintibor@gmail.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Alexandre Gramfort <agramfort@meta.com>
6
6
  Maintainer-email: Alexandre Gramfort <agramfort@meta.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Robin Tibor Schirrmeister <robintibor@gmail.com>
@@ -10,6 +10,7 @@ from .transforms import (
10
10
  ChannelsReref,
11
11
  ChannelsShuffle,
12
12
  ChannelsSymmetry,
13
+ CutMix,
13
14
  FrequencyShift,
14
15
  FTSurrogate,
15
16
  GaussianNoise,
@@ -36,6 +37,7 @@ __all__ = [
36
37
  "SignFlip",
37
38
  "FTSurrogate",
38
39
  "ChannelsShuffle",
40
+ "CutMix",
39
41
  "ChannelsDropout",
40
42
  "GaussianNoise",
41
43
  "ChannelsSymmetry",
@@ -3,6 +3,7 @@
3
3
  # Gustavo Rodrigues <gustavenrique01@gmail.com>
4
4
  # Bruna Lopes <brunajaflopes@gmail.com>
5
5
  # Sarthak Tayal <sarthaktayal2@gmail.com>
6
+ # Anton Soloviev <anton@praviel.com>
6
7
  #
7
8
  # License: BSD (3-clause)
8
9
 
@@ -579,6 +580,7 @@ def _frequency_shift(X: torch.Tensor, fs: float, f_shift: float) -> torch.Tensor
579
580
  _f_shift = torch.as_tensor(f_shift).float()
580
581
  else:
581
582
  raise ValueError(f"Invalid f_shift type: {type(f_shift)}")
583
+ _f_shift = _f_shift.to(X.device)
582
584
  f_shift_stack = _f_shift.repeat(N_padded, n_channels, 1)
583
585
  reshaped_f_shift = f_shift_stack.permute(
584
586
  *torch.arange(f_shift_stack.ndim - 1, -1, -1)
@@ -931,7 +933,7 @@ def _rotate_signals(
931
933
  transformed_X = X.clone()
932
934
  sensors_positions = list(sensors_positions_matrix)
933
935
  for s, rot_sensors_matrix in enumerate(rot_sensors_matrices):
934
- rot_sensors_positions = list(rot_sensors_matrix.T)
936
+ rot_sensors_positions = list(rot_sensors_matrix)
935
937
  for time in range(X.shape[-1]):
936
938
  interpolator_t = Rbf(*sensors_positions, X[s, :, time])
937
939
  transformed_X[s, :, time] = torch.from_numpy(
@@ -1074,6 +1076,54 @@ def mixup(
1074
1076
  return X_mix, (y_a, y_b, lam)
1075
1077
 
1076
1078
 
1079
+ def cutmix(
1080
+ X: torch.Tensor,
1081
+ y: torch.Tensor,
1082
+ idx_perm: torch.Tensor,
1083
+ cut_t_start: torch.Tensor,
1084
+ cut_t_len: torch.Tensor,
1085
+ ) -> tuple[torch.Tensor, torch.Tensor]:
1086
+ """Paste a time region of shuffled examples into the original examples.
1087
+
1088
+ See [1]_ for details.
1089
+
1090
+ Parameters
1091
+ ----------
1092
+ X : torch.Tensor
1093
+ EEG data in form ``batch_size, n_channels, n_times``
1094
+ y : torch.Tensor
1095
+ Target of length ``batch_size``
1096
+ idx_perm : torch.Tensor
1097
+ Permuted indices of example that are mixed into original examples.
1098
+ cut_t_start : torch.Tensor
1099
+ Start index of the pasted time region for each example.
1100
+ cut_t_len : torch.Tensor
1101
+ Length of the pasted time region for each example.
1102
+
1103
+ Returns
1104
+ -------
1105
+ tuple
1106
+ ``X``, ``y``. Where ``X`` is augmented and ``y`` is a tuple of length
1107
+ 3 containing the labels of the two mixed examples and the fraction of
1108
+ each example kept from the original, ``1 - cut_t_len / n_times``.
1109
+
1110
+ References
1111
+ ----------
1112
+ .. [1] Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun,
1113
+ Junsuk Choe, Youngjoon Yoo (2019). CutMix: Regularization Strategy
1114
+ to Train Strong Classifiers with Localizable Features. In 2019
1115
+ IEEE/CVF International Conference on Computer Vision (ICCV)
1116
+ Online: https://arxiv.org/abs/1905.04899
1117
+ """
1118
+ n_times = X.shape[-1]
1119
+ t = torch.arange(n_times, device=X.device)
1120
+ start = cut_t_start[:, None]
1121
+ in_cut = (t >= start) & (t < start + cut_t_len[:, None])
1122
+ X_mix = torch.where(in_cut[:, None], X[idx_perm], X)
1123
+ lam = (n_times - cut_t_len).to(X.dtype) / n_times
1124
+ return X_mix, (y.clone(), y[idx_perm].clone(), lam)
1125
+
1126
+
1077
1127
  def segmentation_reconstruction(
1078
1128
  X: torch.Tensor,
1079
1129
  y: torch.Tensor,
@@ -1093,7 +1143,8 @@ def segmentation_reconstruction(
1093
1143
  y : torch.Tensor
1094
1144
  EEG labels for the example or batch.
1095
1145
  n_segments : int
1096
- Number of segments to use in the batch.
1146
+ Number of segments to use in the batch. Any remaining samples are
1147
+ included in the last segment.
1097
1148
  data_classes : list[tuple[int, torch.Tensor]]
1098
1149
  List of tuples. Each tuple contains the class index and the corresponding EEG data.
1099
1150
  rand_indices : array-like
@@ -1133,6 +1184,8 @@ def segmentation_reconstruction(
1133
1184
  for idx_segment in range(n_segments):
1134
1185
  start = idx_segment * segment_size
1135
1186
  end = (idx_segment + 1) * segment_size
1187
+ if idx_segment == n_segments - 1:
1188
+ end = window_size
1136
1189
 
1137
1190
  # Perform the data augmentation
1138
1191
  X_aug[np.arange(n_trials), :, start:end] = X_class[
@@ -3,6 +3,7 @@
3
3
  # Gustavo Rodrigues <gustavenrique01@gmail.com>
4
4
  # Bruna Lopes <brunajaflopes@gmail.com>
5
5
  # Sarthak Tayal <sarthaktayal2@gmail.com>
6
+ # Anton Soloviev <anton@praviel.com>
6
7
  #
7
8
  # License: BSD (3-clause)
8
9
 
@@ -24,6 +25,7 @@ from .functional import (
24
25
  channels_permute,
25
26
  channels_rereference,
26
27
  channels_shuffle,
28
+ cutmix,
27
29
  frequency_shift,
28
30
  ft_surrogate,
29
31
  gaussian_noise,
@@ -1095,6 +1097,55 @@ class Mixup(Transform):
1095
1097
  }
1096
1098
 
1097
1099
 
1100
+ class CutMix(Mixup):
1101
+ """Implements CutMix for EEG data.
1102
+
1103
+ A random time region of each example is replaced by the same region of
1104
+ a shuffled example of the batch, see [1]_. As in the reference code [2]_,
1105
+ a coefficient ``lam`` is drawn from a Beta distribution, the region is
1106
+ ``(1 - lam) * n_times`` long (the 1-D analogue of the cut area ratio
1107
+ ``1 - lam``) around a uniformly drawn center, it is clipped at the trial
1108
+ boundaries, and ``lam`` is then recomputed from the pasted length.
1109
+
1110
+ Like :class:`Mixup`, it applies to whole batches and returns targets
1111
+ ``(y_a, y_b, lam)`` for :func:`braindecode.training.mixup_criterion`.
1112
+
1113
+ Parameters
1114
+ ----------
1115
+ alpha : float
1116
+ Parameter of the Beta distribution of ``lam``. ``alpha <= 0`` cuts
1117
+ nothing.
1118
+ beta_per_sample : bool (default=False)
1119
+ By default, one ``lam`` per batch is drawn. If True, one per sample.
1120
+ random_state : int | numpy.random.Generator, optional
1121
+ Seed to be used to instantiate numpy random number generator instance.
1122
+ Defaults to None.
1123
+
1124
+ References
1125
+ ----------
1126
+ .. [1] Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun,
1127
+ Junsuk Choe, Youngjoon Yoo (2019). CutMix: Regularization Strategy
1128
+ to Train Strong Classifiers with Localizable Features. In 2019
1129
+ IEEE/CVF International Conference on Computer Vision (ICCV)
1130
+ Online: https://arxiv.org/abs/1905.04899
1131
+ .. [2] https://github.com/clovaai/CutMix-PyTorch/blob/master/train.py
1132
+ """
1133
+
1134
+ operation = staticmethod(cutmix) # type: ignore[assignment]
1135
+
1136
+ def get_augmentation_params(self, *batch):
1137
+ """Return idx_perm and the cut_t_start, cut_t_len of each region."""
1138
+ params = super().get_augmentation_params(*batch)
1139
+ lam = params.pop("lam")
1140
+ n_times = batch[0].shape[-1]
1141
+ half = ((1 - lam) * n_times).long() // 2
1142
+ center = torch.as_tensor(self.rng.randint(n_times, size=len(lam)))
1143
+ center = center.to(lam.device)
1144
+ start = (center - half).clamp(min=0)
1145
+ end = (center + half).clamp(max=n_times)
1146
+ return dict(params, cut_t_start=start, cut_t_len=end - start)
1147
+
1148
+
1098
1149
  class SegmentationReconstruction(Transform):
1099
1150
  """Segmentation Reconstruction from Lotte (2015) [Lotte2015]_.
1100
1151
 
@@ -1283,7 +1334,7 @@ class MaskEncoding(Transform):
1283
1334
  )
1284
1335
 
1285
1336
  time_start = self.rng.randint(
1286
- 0, n_times - segment_length, (batch_size, self.n_segments)
1337
+ 0, n_times - segment_length + 1, (batch_size, self.n_segments)
1287
1338
  )
1288
1339
  time_start = torch.from_numpy(time_start)
1289
1340
 
@@ -14,13 +14,16 @@ from __future__ import annotations
14
14
  import bisect
15
15
  import html as _html
16
16
  import json
17
+ import mmap
17
18
  import os
18
19
  import shutil
19
20
  import warnings
20
21
  from abc import abstractmethod
21
22
  from collections import Counter
22
- from collections.abc import Callable, Hashable
23
+ from collections.abc import Callable, Hashable, Sequence
23
24
  from glob import glob
25
+ from itertools import chain
26
+ from operator import attrgetter
24
27
  from pathlib import Path
25
28
  from typing import Any, Generic, Iterable, no_type_check
26
29
 
@@ -28,6 +31,7 @@ import mne
28
31
  import mne.io
29
32
  import numpy as np
30
33
  import pandas as pd
34
+ from mne.io.constants import FIFF
31
35
  from mne.utils import _soft_import
32
36
  from mne.utils.docs import deprecated
33
37
  from torch.utils.data import ConcatDataset, Dataset, IterableDataset
@@ -220,6 +224,125 @@ def _compute_ch_pos(info, targets_from):
220
224
  return pos
221
225
 
222
226
 
227
+ _TAG_ROW = attrgetter("kind", "type", "size", "next", "pos")
228
+
229
+
230
+ def _tag_rows(tags):
231
+ """(kind, type, size, next, pos) of FIF tags as an int64 array (n, 5)."""
232
+ flat = chain.from_iterable(map(_TAG_ROW, tags))
233
+ return np.fromiter(flat, np.int64, 5 * len(tags)).reshape(-1, 5)
234
+
235
+
236
+ class _FifTagTable(Sequence):
237
+ """The data-buffer tags of a lazy FIF raw as one int64 array.
238
+
239
+ mne's FIF reader keeps one ``Tag`` object per data buffer in
240
+ ``raw._raw_extras[i]["ent"]`` (one per second of signal with mne's default
241
+ buffers), each with its attribute dict and integers once unpickled. A lazy
242
+ dataset of many recordings holds millions of them in RAM, and every forked
243
+ DataLoader worker copies the memory pages of those it reads (reference
244
+ counts). This holds the same fields in 40 bytes per buffer and returns an
245
+ equal ``Tag`` on access, so mne's own reads are unchanged.
246
+ """
247
+
248
+ def __init__(self, tags):
249
+ self.tag_class = type(tags[0])
250
+ self.table = _tag_rows(tags)
251
+
252
+ def __len__(self):
253
+ return len(self.table)
254
+
255
+ def __getitem__(self, i):
256
+ if isinstance(i, slice):
257
+ return [self.tag_class(*row) for row in self.table[i].tolist()]
258
+ return self.tag_class(*self.table[i].tolist())
259
+
260
+
261
+ _TAG_FIELDS = ("kind", "type", "size", "next", "pos", "data")
262
+
263
+
264
+ def _compact_fif_tags(inst):
265
+ """Swap the ``Tag`` lists of a FIF raw for `_FifTagTable` (in place).
266
+
267
+ Only gap-free lists of directory tags (no ``None``, no ``data``) with mne's
268
+ ``Tag`` fields are swapped; anything else is left as mne built it.
269
+ """
270
+ for extra in getattr(inst, "_raw_extras", ()):
271
+ tags = extra.get("ent")
272
+ if not isinstance(tags, list) or not tags:
273
+ continue
274
+ cls = type(tags[0])
275
+ if tuple(getattr(cls, "__dataclass_fields__", ())) == _TAG_FIELDS and all(
276
+ type(t) is cls and t.data is None for t in tags
277
+ ):
278
+ extra["ent"] = _FifTagTable(tags)
279
+ return inst
280
+
281
+
282
+ def _read_fif_window(raw, start, stop, willneed=False):
283
+ """Samples ``start:stop`` of a lazy FIF raw, copied from the file in one go.
284
+
285
+ mne reads, casts and calibrates the FIF data buffers of a range one at a
286
+ time (60 Python steps for a 60-s window saved with 1-s buffers, each through
287
+ a buffered file whose buffer is 4 MB on network file systems). Here one
288
+ slice of the memory-mapped file holds the bytes of every buffer in the
289
+ range and one multiply calibrates them, with mne's arithmetic (float64 data
290
+ times ``raw._cals``), so the values equal ``raw._getitem``'s. Returns None
291
+ (the caller uses mne) unless ``raw`` is a lazy, single-file, uncompressed
292
+ FIF raw of float32 buffers without projector, compensation or gaps.
293
+
294
+ With ``willneed=True`` nothing is read and None is returned: the OS is
295
+ only asked to start reading the window's bytes into the page cache in the
296
+ background (``posix_fadvise(POSIX_FADV_WILLNEED)``).
297
+ """
298
+ if (
299
+ not isinstance(raw, mne.io.Raw)
300
+ or raw.preload
301
+ or len(raw._raw_extras) != 1
302
+ or raw._projector is not None
303
+ or raw._comp is not None
304
+ ):
305
+ return None
306
+ extra = raw._raw_extras[0]
307
+ fname, bounds = extra["filename"], extra["bounds"]
308
+ if not isinstance(fname, Path) or fname.suffix == ".gz":
309
+ return None
310
+ start, stop = start + raw._first_samps[0], stop + raw._first_samps[0]
311
+ first = np.searchsorted(bounds, start, side="right") - 1
312
+ tags = extra["ent"]
313
+ if isinstance(tags, _FifTagTable):
314
+ rows = tags.table[first : np.searchsorted(bounds, stop)]
315
+ else:
316
+ tags = tags[first : np.searchsorted(bounds, stop)]
317
+ if any(t is None for t in tags):
318
+ return None
319
+ rows = _tag_rows(tags)
320
+ if not len(rows) or (rows[:, 1] != FIFF.FIFFT_FLOAT).any():
321
+ return None
322
+ size, pos = rows[:, 2].tolist(), rows[:, 4].tolist()
323
+ if willneed: # only a hint: any error is left to the read itself
324
+ try:
325
+ with open(fname, "rb") as f:
326
+ os.posix_fadvise(
327
+ f.fileno(),
328
+ pos[0],
329
+ pos[-1] + 16 + size[-1] - pos[0],
330
+ os.POSIX_FADV_WILLNEED,
331
+ )
332
+ except OSError:
333
+ pass
334
+ return None
335
+ # mmap: no copy through a 4 MB read buffer, and a cold file is read around
336
+ # the window by the kernel (as mne's 4 MB reads did), not window by window
337
+ with open(fname, "rb") as f, mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ) as m:
338
+ buf = m[pos[0] : pos[-1] + 16 + size[-1]]
339
+ one = np.concatenate(
340
+ [np.frombuffer(buf, ">f4", s // 4, p + 16 - pos[0]) for s, p in zip(size, pos)]
341
+ ).reshape(-1, extra["orig_nchan"])[start - bounds[first] : stop - bounds[first]]
342
+ picked = one[:, raw._read_picks[0]].T
343
+ return np.multiply(picked, raw._cals[:, np.newaxis], order="C")
344
+
345
+
223
346
  def _window_info(crop_inds, sfreq):
224
347
  """Extract window size from crop indices.
225
348
 
@@ -511,7 +634,9 @@ class RecordDataset(Dataset[tuple[np.ndarray, int | str, tuple[int, int, int]]])
511
634
  f"'{key}' already in description. Please "
512
635
  f"rename or set overwrite to True."
513
636
  )
514
- self._description.pop(key)
637
+ # drop on a copy: windows datasets share the description
638
+ # Series of the dataset they were created from
639
+ self._description = self._description.drop(key)
515
640
  self._description = pd.concat([self.description, description])
516
641
 
517
642
  @property
@@ -823,9 +948,11 @@ class EEGWindowsDataset(_ZarrMixin, RecordDataset):
823
948
  if self._zarr_data is not None:
824
949
  X = self._zarr_data[:, i_start:i_stop]
825
950
  else:
826
- X = self.raw._getitem(
827
- (slice(None), slice(i_start, i_stop)), return_times=False
828
- )
951
+ X = _read_fif_window(self.raw, i_start, i_stop)
952
+ if X is None:
953
+ X = self.raw._getitem(
954
+ (slice(None), slice(i_start, i_stop)), return_times=False
955
+ )
829
956
  X = np.array(X, dtype="float32")
830
957
  if self.transform is not None:
831
958
  X = self.transform(X)
@@ -1156,6 +1283,49 @@ class WindowsDataset(_ZarrMixin, RecordDataset):
1156
1283
  )
1157
1284
 
1158
1285
 
1286
+ def _lazy_source_files(ds) -> list:
1287
+ """Return the files ``ds`` still reads its signal from (none once preloaded)."""
1288
+ signal = ds.raw if hasattr(ds, "raw") else getattr(ds, "windows", None)
1289
+ if signal is None or getattr(signal, "preload", True):
1290
+ return []
1291
+ files = list(
1292
+ getattr(signal, "filenames", None) or [getattr(signal, "filename", None)]
1293
+ )
1294
+ raw = getattr(signal, "_raw", None)
1295
+ if isinstance(raw, mne.io.BaseRaw) and not raw.preload:
1296
+ files += raw.filenames # mne.Epochs of a lazily read Raw read through it
1297
+ return [f for f in files if f is not None]
1298
+
1299
+
1300
+ def _dir_key(directory):
1301
+ """Identify a directory by device and inode, whatever its spelling."""
1302
+ stat = os.stat(directory)
1303
+ return stat.st_dev, stat.st_ino
1304
+
1305
+
1306
+ def _overwrite_conflicts(datasets, path, offset):
1307
+ """Yield ``(i_ds, i_target, target_dir)`` for each dataset ``i_ds`` that
1308
+ lazily reads from inside ``target_dir = path/<i_target + offset>``, which
1309
+ ``save(overwrite=True)`` deletes before writing dataset ``i_target``.
1310
+ """
1311
+ targets = {}
1312
+ for i in range(len(datasets)):
1313
+ target_dir = os.path.join(path, str(i + offset))
1314
+ if os.path.isdir(target_dir):
1315
+ targets[_dir_key(target_dir)] = (i, target_dir)
1316
+ if not targets:
1317
+ return
1318
+ for i_ds, ds in enumerate(datasets):
1319
+ for file in _lazy_source_files(ds):
1320
+ # ancestors of the name and of the file it links to (DataLad)
1321
+ parents = {
1322
+ *Path(os.path.abspath(file)).parents,
1323
+ *Path(os.path.realpath(file)).parents,
1324
+ }
1325
+ for key in {_dir_key(p) for p in parents if p.is_dir()} & targets.keys():
1326
+ yield (i_ds, *targets[key])
1327
+
1328
+
1159
1329
  @register_dataset
1160
1330
  class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1161
1331
  """A base class for concatenated datasets.
@@ -1253,7 +1423,8 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1253
1423
  def __len__(self):
1254
1424
  return self._ensure_cumulative_sizes()[-1]
1255
1425
 
1256
- def _get_single_item(self, idx: int):
1426
+ def _locate(self, idx: int):
1427
+ """The dataset that holds item ``idx`` and the item's index in it."""
1257
1428
  if idx < 0:
1258
1429
  if -idx > len(self):
1259
1430
  raise ValueError(
@@ -1267,7 +1438,33 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1267
1438
  sample_idx = idx
1268
1439
  else:
1269
1440
  sample_idx = idx - cumulative_sizes[dataset_idx - 1]
1270
- return self.datasets[dataset_idx][sample_idx]
1441
+ return self.datasets[dataset_idx], sample_idx
1442
+
1443
+ def _get_single_item(self, idx: int):
1444
+ ds, i = self._locate(idx)
1445
+ return ds[i]
1446
+
1447
+ def __getitems__(self, indices):
1448
+ """Items ``indices``, as ``[self[i] for i in indices]``.
1449
+
1450
+ A :class:`~torch.utils.data.DataLoader` fetches each batch through
1451
+ this. The lazy FIF windows of the batch are first requested from the
1452
+ OS all at once (``posix_fadvise(POSIX_FADV_WILLNEED)``, where it
1453
+ exists), so their reads from disk overlap instead of running one after
1454
+ another: on a network file system a batch read from cold files then
1455
+ costs about one round trip, not one per window.
1456
+ """
1457
+ # ponytail: advises the whole batch at once; chunk it if a batch's
1458
+ # windows ever outgrow the page cache
1459
+ if hasattr(os, "posix_fadvise"):
1460
+ for idx in indices:
1461
+ if not isinstance(idx, (int, np.integer)):
1462
+ continue # e.g. a SequenceSampler tuple: read, not advised
1463
+ ds, i = self._locate(idx)
1464
+ if isinstance(ds, EEGWindowsDataset) and ds._zarr_data is None:
1465
+ _, start, stop = ds.crop_inds[i].tolist()
1466
+ _read_fif_window(ds.raw, start, stop, willneed=True)
1467
+ return [self[idx] for idx in indices]
1271
1468
 
1272
1469
  def __getitem__(self, idx: int | list):
1273
1470
  """
@@ -1366,7 +1563,9 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1366
1563
  metadata : pd.DataFrame
1367
1564
  DataFrame containing as many rows as there are windows in the
1368
1565
  BaseConcatDataset, with the metadata and description information
1369
- for each window.
1566
+ for each window, and an ``i_dataset`` column giving the position
1567
+ in ``datasets`` of the dataset (recording) each window comes from.
1568
+ Samplers use it to keep recordings apart.
1370
1569
  """
1371
1570
  if not all(
1372
1571
  [
@@ -1392,9 +1591,15 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1392
1591
  "Dataset descriptions cannot contain the reserved window "
1393
1592
  "metadata key 'i_trial_in_dataset'."
1394
1593
  )
1594
+ if "i_dataset" in df.columns or "i_dataset" in ds.description:
1595
+ raise ValueError(
1596
+ "Window metadata and dataset descriptions cannot contain the "
1597
+ "key 'i_dataset', which get_metadata() reserves for the "
1598
+ "position of the dataset each window comes from. Rename it."
1599
+ )
1395
1600
 
1396
1601
  all_dfs = list()
1397
- for ds in self.datasets:
1602
+ for i_ds, ds in enumerate(self.datasets):
1398
1603
  if hasattr(ds, "_windows") and ds._windows is not None:
1399
1604
  df = ds._windows.metadata
1400
1605
  else:
@@ -1404,6 +1609,7 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1404
1609
  df = df.copy()
1405
1610
  for k, v in ds.description.items():
1406
1611
  df[k] = v
1612
+ df["i_dataset"] = i_ds
1407
1613
  all_dfs.append(df)
1408
1614
 
1409
1615
  return pd.concat(all_dfs)
@@ -1672,7 +1878,8 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1672
1878
  kwargs_path = os.path.join(path, ".".join([kwarg_name, "json"]))
1673
1879
  kwargs = getattr(self, kwarg_name)
1674
1880
  if kwargs is not None:
1675
- json.dump(kwargs, open(kwargs_path, "w"))
1881
+ with open(kwargs_path, "w") as f:
1882
+ json.dump(kwargs, f)
1676
1883
 
1677
1884
  @property
1678
1885
  def description(self) -> pd.DataFrame:
@@ -1695,6 +1902,11 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1695
1902
  dataset description.
1696
1903
  """
1697
1904
  description = pd.DataFrame(description)
1905
+ if len(description.columns) and len(description) != len(self.datasets):
1906
+ raise ValueError(
1907
+ f"Length of values ({len(description)}) does not match "
1908
+ f"number of datasets ({len(self.datasets)})."
1909
+ )
1698
1910
  for key, value in description.items():
1699
1911
  for ds, value_ in zip(self.datasets, value):
1700
1912
  ds.set_description({key: value_}, overwrite=overwrite)
@@ -1725,7 +1937,9 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1725
1937
  -raw.fif | -epo.fif and .json files to.
1726
1938
  overwrite : bool
1727
1939
  Whether to delete old subdirectories that will be saved to in this
1728
- call.
1940
+ call. If a dataset still reads its data lazily from one of them, a
1941
+ ``ValueError`` is raised before anything is deleted: load the data
1942
+ into memory first (``preload=True``) or save to another directory.
1729
1943
  offset : int
1730
1944
  If provided, the integer is added to the id of the dataset in the
1731
1945
  concat. This is useful in the setting of very large datasets, where
@@ -1738,11 +1952,21 @@ class BaseConcatDataset(ConcatDataset, HubDatasetMixin, Generic[T]):
1738
1952
  hasattr(self.datasets[0], "raw") or hasattr(self.datasets[0], "windows")
1739
1953
  ):
1740
1954
  raise ValueError("dataset should have either raw or windows attribute")
1955
+ if overwrite:
1956
+ for _, _, target_dir in _overwrite_conflicts(self.datasets, path, offset):
1957
+ raise ValueError(
1958
+ f"Saving with overwrite=True would delete {target_dir} while "
1959
+ "a dataset still reads its data from it. Load the data into "
1960
+ "memory first (preload=True) or save to another directory."
1961
+ )
1741
1962
 
1742
1963
  # Create path if it doesn't exist
1743
1964
  os.makedirs(path, exist_ok=True)
1744
1965
 
1745
- path_contents = os.listdir(path)
1966
+ # With an offset the caller fills one directory piece by piece (preprocess
1967
+ # saves each recording at its index), so the other entries are expected:
1968
+ # skip the scan that would stat them all on every call and list them in a warning.
1969
+ path_contents = os.listdir(path) if offset == 0 else []
1746
1970
  n_sub_dirs = len(
1747
1971
  [e for e in path_contents if os.path.isdir(os.path.join(path, e))]
1748
1972
  )
@@ -159,6 +159,11 @@ def create_from_mne_epochs(
159
159
 
160
160
  list_of_windows_ds = []
161
161
  for i, epochs in enumerate(list_of_epochs):
162
+ # drop rejected epochs first (on a copy) so that iterating over the
163
+ # epochs stays aligned with ``epochs.events``; preloaded epochs have
164
+ # already dropped them, so skip copying their data
165
+ if not epochs.preload:
166
+ epochs = epochs.copy().drop_bad()
162
167
  event_descriptions = epochs.events[:, 2]
163
168
  original_trial_starts = epochs.events[:, 0]
164
169
  stop = len(epochs.times) - window_size_samples