You need to agree to share your Hugging Face information to access this dataset
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
These are the original media files of public research datasets, re-hosted so that JevArena-MM can be run. Each file stays under its source's license (table below); several sources allow non-commercial research use only. By requesting access you agree to use the files only to evaluate models or for other non-commercial research, to follow each source's license and terms, not to redistribute them, and to delete any file whose owner asks for its removal.
Log in or Sign Up to review the conditions and access this dataset content.
JevArena-MM (video): media
The media files of the video tasks of JevArena-MM, a benchmark of JevArena decision requests whose state carries video and audio. The items, code and result tables are in the JevArena-MM repository; this dataset holds only the files the items point to: 10,078 files, 438 GB.
- A file an item references as
<folder>/<path>is at<folder>/<path>here: the source's original file, unchanged. Items select a window of it and record its sha256. - Media whose terms forbid redistribution, or require each user to accept them first, are not hosted; JevArena-MM's fetch scripts get them from their owners or distributors (table at the end).
Download
Accept the terms on this page, log in with hf auth login (or set HF_TOKEN), then in a clone of JevArena-MM:
pip install -e '.[video]'
export JEVARENA_MM_MEDIA=/path/to/media
jevarena-mm fetch --leaf vid.
jevarena-mm fetch --leaf lvid.
jevarena-mm verify
One folder without the repository:
hf download JevArena-mm/jevarena-MM_video --repo-type dataset --include 'nexar/*' --local-dir /path/to/media
Hosted folders
| Folder | Source | License | Files | Size |
|---|---|---|---|---|
ami |
AMI Meeting Corpus, decision annotations (Carletta et al., 2005) | CC BY 4.0 | 266 | 17.3 GB |
captaincook4d |
CaptainCook4D, step error annotations (Peddi et al., 2024) | Apache-2.0 (data); MIT (annotation repository) | 96 | 10.9 GB |
changeit |
ChangeIt, test split (Soucek et al., 2022) | MIT (annotations); video rights not established | 81 | 5.7 GB |
countix |
Countix, test split (Dwibedi et al., 2020) | No license stated; Kinetics YouTube clips, video rights not established | 250 | 0.3 GB |
epic-kitchens-100 |
EPIC-KITCHENS-100, action anticipation, validation split (Damen et al., 2022) | CC BY-NC 4.0 | 135 | 169.4 GB |
jaad |
JAAD, pedestrian crossing behaviour (Rasouli et al., 2017) | CC BY 4.0 (videos); MIT (annotations) | 130 | 1.3 GB |
jevarena-marks |
JAAD, pedestrian crossing behaviour (Rasouli et al., 2017) | CC BY 4.0 (videos); MIT (annotations) | 500 | 0.0 GB |
jevarena-marks |
PIE, pedestrian crossing intention (Rasouli et al., 2019) | MIT (videos and annotations) | 500 | 0.0 GB |
kinetics700 |
Kinetics-700-2020, validation split (Smaira et al., 2020) | CC BY 4.0 (annotations); YouTube videos keep their own terms | 500 | 0.7 GB |
konvid-1k |
KoNViD-1k (Hosu et al., 2017) | Free for research, no license stated; Flickr (YFCC100m) videos under their own Creative Commons terms | 500 | 1.0 GB |
meetingbank |
TofuEval, MeetingBank topic summaries (Tang et al., 2024; Hu et al., 2023) | MeetingBank: CC BY-NC-ND 4.0 (the strictest of its stated licenses); TofuEval: MIT-0 | 50 | 91.7 GB |
msrvtt |
MSR-VTT multiple-choice test (Xu et al., 2016; Yu et al., 2018) | No license stated; video rights not established | 500 | 0.3 GB |
mvbench |
MVBench (Li et al., 2024) | MIT (dataset card); source videos keep their own licenses | 346 | 2.4 GB |
needle |
Generated needle videos (JevArena-MM video team) | CC BY 4.0 (generated by the JevArena-MM video team; no third-party assets) | 250 | 2.6 GB |
nexar |
Nexar Collision Prediction, test set (Moura and Zvitia, 2025) | Nexar Open Data License (attribution; no resale) | 500 | 2.2 GB |
nextqa |
NExT-QA, multiple choice, validation split (Xiao et al., 2021) | MIT (annotations); VidOR videos from YFCC100m under their own Creative Commons terms | 500 | 2.2 GB |
perception-test |
MVBench (Li et al., 2024) | MIT (dataset card); source videos keep their own licenses | 511 | 5.4 GB |
perception-test |
Perception Test, multiple-choice video QA, valid split (Patraucean et al., 2023) | CC BY 4.0 | 511 | 5.4 GB |
physgame |
PhysGame (Cao et al., 2024) | Apache-2.0 (dataset card); gameplay clips from Reddit and YouTube, rights not established | 500 | 4.7 GB |
physion |
Physion object contact prediction, test set (Bear et al., 2021) | MIT (code and data) | 500 | 0.1 GB |
pie |
PIE, pedestrian crossing intention (Rasouli et al., 2019) | MIT (videos and annotations) | 49 | 69.1 GB |
qvhighlights |
QVHighlights saliency scores, validation split (Lei et al., 2021) | CC BY-NC-SA 4.0 (annotations); YouTube videos, no license stated | 520 | 6.0 GB |
qvhighlights |
QVHighlights, validation split (Lei et al., 2021) | CC BY-NC-SA 4.0 (annotations); YouTube videos, no license stated | 520 | 6.0 GB |
repcount-a |
RepCount-A, test and validation splits (Hu et al., 2022) | No license stated; YouTube videos, rights not established | 250 | 2.4 GB |
robofac |
RoboFAC robotic failure analysis dataset (2025) | MIT | 500 | 0.2 GB |
tempcompass |
TempCompass, multiple-choice questions (Liu et al., 2024) | CC BY-NC 4.0 | 410 | 0.4 GB |
ucf-crime |
UCF-Crime, test split (Sultani et al., 2018) | No license stated; video rights not established | 270 | 7.6 GB |
ucf-crime |
UCF-Crime, untrimmed test videos (Sultani et al., 2018) | No license stated; video rights not established | 270 | 7.6 GB |
ucf101 |
UCF101, test split 1 (Soomro et al., 2012) | No license stated (cite Soomro et al., 2012); YouTube videos | 500 | 0.3 GB |
videofeedback |
VideoFeedback, text-to-video alignment, test split (He et al., 2024) | Apache-2.0 (dataset card) | 500 | 0.1 GB |
videohallucer |
VideoHallucer, object-relation and temporal subsets (Wang et al., 2024) | MIT (dataset card) | 250 | 0.9 GB |
videommmu |
VideoMMMU (Hu et al., 2025) | CC BY-NC-SA 4.0 (dataset); videos remain their creators' | 255 | 25.5 GB |
vsi-bench |
VSI-Bench (Yang et al., 2025) | Apache-2.0 (annotations); videos under the ARKitScenes, ScanNet and ScanNet++ terms | 121 | 1.5 GB |
xd-violence |
XD-Violence, test split (Wu et al., 2020) | No license stated; video rights not established | 338 | 5.7 GB |
Not hosted
jevarena-mm fetch downloads these from their owners or distributors as well; you accept each source's terms by
doing so. Only Ego4D's own files need something from you: an Ego4D license and its keys.
| Files | Source | Terms | fetch gets them from |
|---|---|---|---|
charades/* |
Charades-STA (Gao et al., 2017) | Charades license, local evaluation only (https://prior.allenai.org/projects/charades) | AI2's public Charades archive |
charades/* |
STAR, prediction and feasibility questions, validation split (Wu et al., 2021) | Charades license, local evaluation only (https://prior.allenai.org/projects/charades) | AI2's public Charades archive |
ego4d/* |
Ego4D natural language queries, validation split (Grauman et al., 2022) | Ego4D license agreement (https://ego4d-data.org/) | Ego4D's S3 buckets, with your Ego4D license keys |
ego4d/* |
Ego4D object state change classification (OSCC), validation split (Grauman et al., 2022) | Ego4D license agreement (https://ego4d-data.org/) | Ego4D's S3 buckets, with your Ego4D license keys |
egoplan-bench2/* |
EgoPlan-Bench2 (Qiu et al., 2024) | cut from Ego4D; Ego4D license agreement (https://ego4d-data.org/) | EgoPlan-Bench2's reference clips (TencentARC/SEED-Bench-R1 on Hugging Face) |
egoschema/* |
EgoSchema, public answer subset (Mangalam et al., 2023) | cut from Ego4D; Ego4D license agreement (https://ego4d-data.org/) | EgoSchema's own release (Google Drive, or the lmms-lab/egoschema mirror) |
mvbench/original-publisher/charades-official-pinned/* |
MVBench (Li et al., 2024) | a Charades video; Charades license | AI2's public Charades archive |
mvbench/ssv2_video/* |
MVBench (Li et al., 2024) | Something-Something v2 clips; Something-Something v2 license | MVBench's release (OpenGVLab/MVBench) |
mvbench/sta/* |
MVBench (Li et al., 2024) | Charades-STA clips cut from Charades; Charades license | MVBench's release (OpenGVLab/MVBench) |
mvbench/star/* |
MVBench (Li et al., 2024) | STAR clips cut from Charades; Charades license | MVBench's release (OpenGVLab/MVBench) |
soccernet-mvfoul/* |
SoccerNet-MVFoul, test split (Held et al., 2023) | broadcast footage whose rights are not established; from SoccerNet's own release (https://huggingface.co/datasets/SoccerNet/SN-MVFouls-2024) | SoccerNet's release (SoccerNet/SN-MVFouls-2024) |
ssv2/* |
Something-Something v2, validation split (Goyal et al., 2017) | Something-Something v2 license (https://www.qualcomm.com/developer/software/something-something-v-2-dataset) | Qualcomm's official package (streamed; no login) |
videomme/* |
Video-MME, long videos (Fu et al., 2025) | Video-MME terms: no distribution without the authors' approval (https://github.com/MME-Benchmarks/Video-MME) | Video-MME's release (lmms-lab/Video-MME) |
vsi-bench/*scannet_* |
VSI-Bench (Yang et al., 2025) | ScanNet terms of use: shared only with people who signed them (http://www.scan-net.org/) | VSI-Bench's release (nyu-visionx/VSI-Bench) |
vsi-bench/*scannetpp_* |
VSI-Bench (Yang et al., 2025) | ScanNet++ terms of use: shared only after signing them (https://kaldir.vc.in.tum.de/scannetpp/) | VSI-Bench's release (nyu-visionx/VSI-Bench) |
Removal
To have a file removed, open a discussion on this page naming the file; it is taken down.
- Downloads last month
- 155