Dataset Viewer

The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.

Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): (None, {}), NamedSplit('test'): ('json', {})}

Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Portable synthetic-data / EEG-foundation-model study

This directory is the complete transfer payload: archive, upload, or copy the whole directory as-is. No transfer-time allowlist or exclusion rule is needed.

It is self-contained except for the licensed TUH EEG datasets. Put them at:

test/AB: TUAB
test/EV: TUEV
test/SZ: TUSZ

Selected foundation models are NeuroLM-B, NeuroLM-L, NeuroLM-XL, ST-EEGFormer-Large and LUNA-Huge. Selected generators are conditional diffusion, conditional EEG WGAN-GP and JET.

The study evaluates the influence of synthetic data on both native self-supervised foundation-model training and fixed real-only downstream linear probes. All jobs use frozen, shared evaluation indices and report confidence intervals.

Experiment tracking is entirely local; no result is sent to W&B. Start with docs/RUNBOOK.md and docs/WANDB_AND_COMBINATION_RUNBOOK.md (the legacy filename now documents console logging): rebuild the isolated environment, validate all assets with env/runtime/bin/python src/validate_bundle.py --deep-hash --require-trained-generators, then use one of the 75 GPU-selectable combination scripts indexed by configs/combination_scripts.tsv.

Downloads last month: 89

Total file size:

38.2 GB