dataset-grid / collections
Collections
Themed bundles of datasets under a single index. A collection is a pinned list of revisions; it is pulled with one command and stays reproducible.
| Collection | Modality | Format | Datasets | Size | Revision |
|---|---|---|---|---|---|
| vault-labs/ru-pretrain-suite | Text | parquet · zstd | 7 | 4.62 TB | v3 · 8c1e0a2 |
| volga-ai/asr-ru-bundle | Audio | webdataset · flac | 4 | 1.73 TB | v1.2 · 44b9f71 |
| northgrid/vision-defects-pack | Images | webdataset · jpeg | 3 | 918 GB | v2 · d07a3be |
| finlab/fin-instruct-ru | Text | parquet · gzip | 5 | 38.4 GB | v5 · 19e6c4d |
| medsight/medimg-segmentation | Images | arrow · nifti | 2 | 3.41 TB | v1 · a5f20c8 |
| northgrid/timeseries-grid | Tabular | parquet · snappy | 6 | 1.59 TB | v2.1 · 6b3d9e0 |
partial collection pull
$ dgrid pull vault-labs/ru-pretrain-suite --include 'shard-000*.parquet'Need to assemble a private collection over 10 TB?
Contact sales