EEGDash Corpus: A Training-Ready Public Corpus for EEG Foundation Models
Organizations: Swartz Center for Computational Neuroscience, University of California San Diego, USA · Yneuro and Université Paris-Saclay, France
Abstract
Electroencephalography (EEG) foundation models are pretrained on tens of thousands of hours. The largest clinical archives restrict redistribution, while public datasets use different formats and licences. Here, we rebuild the 77 redistributable datasets of the REVE pretraining corpus that reach its 200 Hz input (14,075 hours across 7,272 subjects) and release them as eegdash-corpus-v0, with raw-file access by name through EEGDash and a checksummed Zarr store on Hugging Face, converted with REVE's preprocessing recipe. Every dataset keeps its licence, and any subset of datasets can be selected by name. To test whether this public store pretrains REVE-Base, we train the model twice from scratch with REVE's trainer and score it on seven public OpenEEGBench tasks. First, under linear probing, the frozen encoder reaches 0.331 in chance-adjusted balanced accuracy, against 0.333 for the released model (untrained: 0.200), a difference smaller than the spread between our two pretraining seeds. Second, after fine-tuning, it reaches 0.273 against 0.325 published (0.301 in our rerun), and part of this gap comes from the scoring pipeline. A public, training-ready EEG corpus can therefore pretrain the frozen representations of a foundation model and provide a shared base that other groups can audit.
Figures & tables
Appendix figures & tables3 assets
Supplementary material from the paper’s appendix.
Appendix
| corpus | datasets | hours | mean ch. | tokens |
|---|---|---|---|---|
| REVE paper [ 3 ] | 92 | 61,415 | 32 | 7.9 B |
| REVE’s release [ 8 ] | 70 | 34,400 | 49 | 6.59 B |
| eegdash-corpus-v0 | 77 | 14,075 | 112 | 6.21 B |
| id | short name | paradigm | subj. | hours | licence | id | short name | paradigm | subj. | hours | licence |
|---|---|---|---|---|---|---|---|---|---|---|---|
| nm000338 | Lee2019-MI | BCI | 54 | 183 | CC0 | ds002680 | Go/no-go | perception | 14 | 21 | CC0 |
| nm000250 | Dreyer2023 | BCI | 87 | 127 | CC BY | ds004324 | Toon faces | perception | 26 | 19 | CC0 |
| nm000323 | Lee2019ERP | BCI | 54 | 116 | CC0 | ds004043 | Feature attention | perception | 20 | 18 | CC0 |
| nm000267 | Shin2017A | BCI | 29 | 29 | GPL-3.0 | ds005594 | Alphabetic decision | perception | 16 | 12 | CC0 |
| nm000172 | Schirrmeister2017 | BCI | 14 | 29 | CC BY | ds004817 | RSVP attention 2 | perception | 20 | 12 | CC0 |
| nm000173 | Ofner2017 | BCI | 15 | 27 | CC BY | ds004357 | Features-EEG | perception | 16 | 11 | CC0 |
| dataset | reason |
|---|---|
| TUH EEG [ 15 ] ; Harmful Brain Activity; TDBRAIN; I-CARE [ 23 ] | registration or a data use agreement [ 22 ] ; about 50,000 of REVE’s 61,415 hours |
| BNCI2014004/008/009, BNCI2015001/003/004, Nakanishi2015, EPFL P300 (MOABB) | no-derivatives terms (six datasets) or no explicit redistribution licence (two); excluded under this release’s policy |
| Zhou2016 (MOABB) [ 24 ] | the card’s deposit ( nm000226 ) is a 100 Hz copy, below the 200 Hz floor; the 250 Hz original ( nm000115 ) can enter a later version |
| PhysioNet MI [ 34 ] , ds004362 ; ds005597 ; “other PhysioBank items” | on the card only: a test task; fMRI without EEG; no dataset named |
Electroencephalography
Electroencephalography (EEG) studies brain-generated electrical activity to understand cognition and support diagnosis, monitoring, and brain-computer interfaces.
See all 200 papers