Zeteng/meva-reversible-events-vjepa2.1-vitb-features
MEVA reversible events — frozen V-JEPA 2.1 ViT-B features Pre-computed V-JEPA 2.1 ViT-B/16 (384) features for 10 pairs of reversible activities from the MEVA dataset, for quick probing of temporal-direction understanding without re-encoding video. No video frames or images are included. Contents file content instances_train.jsonl (2,060), instances_dev.jsonl (834) id, split, clip name, camera, event class, event pair, label, activity frame range, crop… See the full description on the dataset page: https://huggingface.co/datasets/Zeteng/meva-reversible-events-vjepa2.1-vitb-features.
| Author / organization | Zeteng |
|---|---|
| License (as declared) | cc-by-4.0 |
| Downloads | 0 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 21 October 2026 |
| Last updated at source | 21 October 2026 |
Popularity, last 2 days
0 now. Counts as reported by the source, snapshotted daily by Silytics.
Recent changes
- 21 Oct 2026 License: none → cc-by-4.0
- 21 Oct 2026 Description: none → MEVA reversible events — frozen V-JEPA 2.1 ViT-B features Pre-computed V-JEPA 2.1 ViT-B/16 (384) features for 10 pairs of reversible activities from the MEVA dataset, for quick probing of temporal-direction understanding without re-encoding
- 21 Oct 2026 Topics added task_categories:video-classification, video, v-jepa, meva, features, temporal-events
Changes detected by Silytics between refreshes of the source.
More from Zeteng
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 22 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.