video_id stringclasses 4
values | youtube_url stringclasses 4
values | subtitle_url stringclasses 4
values | availability stringclasses 1
value | in_prepared_selection bool 1
class | video_url stringclasses 4
values | size_bytes int64 13.9M 408M | last_attempt_at timestamp[s]date 2026-11-17 14:21:32 2026-11-17 14:21:38 |
|---|---|---|---|---|---|---|---|
h80t-2JRLl8 | https://www.youtube.com/watch?v=h80t-2JRLl8 | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/h80t-2JRLl8.vtt | published | true | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/h80t-2JRLl8.mp4 | 262,373,326 | 2026-11-17T14:21:38 |
OzPdil_lRl4 | https://www.youtube.com/watch?v=OzPdil_lRl4 | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/OzPdil_lRl4.vtt | published | true | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/OzPdil_lRl4.mp4 | 407,998,444 | 2026-11-17T14:21:36 |
zCPeCgPfYQs | https://www.youtube.com/watch?v=zCPeCgPfYQs | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/zCPeCgPfYQs.vtt | published | true | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/zCPeCgPfYQs.mp4 | 216,105,311 | 2026-11-17T14:21:35 |
Y9IKQ50Qx8Y | https://www.youtube.com/watch?v=Y9IKQ50Qx8Y | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/Y9IKQ50Qx8Y.vtt | published | true | https://huggingface.co/buckets/TornadoLabs/staging/resolve/videos/test-youtube/Y9IKQ50Qx8Y.mp4 | 13,892,445 | 2026-11-17T14:21:32 |
Test_Youtube β YouTube source videos
4 source IDs with their source links and availability. 4 videos (0.9 GB) with 4 subtitle files, downloaded at source quality and hosted for direct download.
Coverage (2026-11-17): 4 complete YouTube source videos are hosted here. They come from a prepared selection of 4 YouTube IDs read from
Decre99/Test_Youtube(revision4ef6c1994c2a); the selection stopped because all matching upstream metadata files were read. Of the prepared IDs: 4 hosted, 0 unavailable at download time, 0 failed downloads, 0 not added because of the 6 GB cap, 0 in progress and 0 not attempted. Thevideosconfiguration lists every prepared ID, hosted or not:video_urlis filled only for the 4 hosted videos, andavailabilitygives the state of each row when this card was generated. The annotations cover all listed IDs.
Public release: capped at 6 GB (6,000,000,000 bytes), including media and indexes. Unit: complete YouTube source videos. Upstream clip intervals, where present, remain in the annotations; the videos have not been cut to those intervals. The index lists every prepared ID; only part of them is hosted.
What's inside
metadata.jsonlβ 4 rows, one per prepared source ID:video_id,youtube_url,availability,in_prepared_selection,video_url(direct HTTPS download, CDN-backed; empty when the video is not hosted),subtitle_url,size_bytes,last_attempt_at- 4 of these videos (0.9 GB) are hosted on Hugging Face storage; the other rows keep their source link only
captions.parquetβ 4 academic annotations (source:annotations.csv)
Quick start
from datasets import load_dataset
# videos (default config): one row per source ID; video_url is set for hosted videos only
ds = load_dataset("TornadoLabs/test-youtube", "videos", split="train")
print(ds[0])
# keep the hosted videos, then download one
import requests
hosted = ds.filter(lambda r: r["video_url"] is not None)
r = requests.get(hosted[0]["video_url"])
open(f"{hosted[0]['video_id']}.mp4", "wb").write(r.content)
# annotations for all listed source videos
ann = load_dataset("TornadoLabs/test-youtube", "annotations", split="train")
How this dataset was built
Each file is the complete YouTube source video (best available quality), fetched through the Tornado API into a temporary location, measured, then copied into the public storage only while the release stayed under 6 GB. Clip intervals from the upstream annotations are kept as annotations and are not applied to the files. Videos that were unavailable, failed or did not fit are counted in the coverage note above.
Provenance & citation
- Original dataset: https://huggingface.co/datasets/Decre99/Test_Youtube
- Paper: See the upstream dataset card
- The upstream repository declares MIT for its annotations; rights to the third-party videos are separate.
The video files remain the property of their respective YouTube uploaders. The annotations belong to the original dataset authors β cite their paper if you use them. This re-hosting is provided for reproducible research; if you are a rights holder and want content removed, open a discussion on this repo.
Request a larger or custom delivery
This release is capped at 6 GB. Additional videos from the prepared selection or from the complete upstream ID list can be collected on request: request a delivery quote from Tornado API. What can be delivered depends on each video's current availability on YouTube and on the applicable rights: recovery of every listed source is not guaranteed, and no commercial rights to the source videos are granted. The upstream terms continue to apply.

Collected and delivered with Tornado API
- Downloads last month
- 16