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'): ('text', {}), NamedSplit('validation'): ('text', {}), NamedSplit('test'): ('hdf5', {})}
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.

Spliceformer — GTEx per-tissue test sets

Per-tissue, per-donor splice test sets used to evaluate Spliceformer across tissues.

python scripts/download_assets.py --gtex-tissues lung
python scripts/evaluate_splice.py --context 10k --ensemble --tissue lung

Tissues

brain_cortex · lung · testis · whole_blood · haec10

Layout

{tissue}/
├── ml_data_var/
│   ├── individual/test_GTEX-*.h5   per-donor test sets (~650 MB each)
│   └── combined_test.h5            all donors combined (~6.5 GB)
├── splits_config.yaml              split configuration
├── splits_summary.txt
├── test_samples.txt / test_chroms.txt
└── dataset_options.txt

HDF5 layout matches the main splice datasets: paired X{i} / Y{i} chunks, with X as one-hot DNA and Y as one-hot [neither, acceptor, donor].

How this was published

The source is a 36 GB archive that does not fit on a workstation alongside its own contents. scripts/upload_to_hf.py --gtex-zip extracts one member at a time, uploads it, and deletes it before moving on, so peak extra disk is the largest single member (~6.5 GB) rather than the size of the archive. The upload is resumable — re-running skips members already on the Hub.

Related

License

MIT

Downloads last month
150