Meta-Video / README.md
soumyasj's picture
Card: add paper, project page and code links, task categories, citation
8811c0c verified
|
Raw History Blame Contribute Delete
3.93 kB
---
license: cc-by-4.0
task_categories:
- video-classification
- visual-question-answering
- video-text-to-text
- multiple-choice
language:
- en
pretty_name: Meta-Video
size_categories:
- 1K<n<10K
configs:
- config_name: meta_v
data_files:
- split: test
path: data/meta_v-*.parquet
---
# Meta-Video (meta_v)
[Project page](https://soumyasj.github.io/meta-video/) · [GitHub](https://github.com/soumyasj/lmms_eval_video) · [Paper](https://arxiv.org/abs/2610.18535)
Meta-Video is a 1,000-question evaluation set for video understanding, sampled from five existing benchmarks
(Video-MME, LongVideoBench, LVBench, VideoMMMU, VSI-Bench) over 791 videos (all <= 60 min).
It keeps only questions that need the video: every question that at least two of three models
(Qwen2.5-VL-7B-Instruct, Qwen3-VL-8B-Instruct, InternVL3.5-8B) answer correctly from black frames was removed
before sampling. Items were then selected to preserve the model ranking of the full benchmarks at ~5x lower cost.
Companion dataset with the video-dependent remainder of each source benchmark:
[soumyasj/video-benchmarks-nonblind](https://huggingface.co/datasets/soumyasj/video-benchmarks-nonblind).
## Files
* `data/meta_v-00000-of-00001.parquet` - the 1,000 items (config `meta_v`, split `test`).
* `videos/<source>/...` - the 791 videos, referenced by the `video` column.
## Columns
| column | description |
|---|---|
| `id` | `<source_task>::<source_id>` |
| `source` | Video-MME, LongVideoBench, LVBench, VideoMMMU or VSI-Bench |
| `source_task` | lmms-eval task name of the source (`video_mmmu_perception` etc.) |
| `source_id` | native question id in the source benchmark |
| `video` | repo-relative path of the video |
| `duration_s`, `duration_bucket` | video length in seconds; short < 5 min, medium 5-20 min, long 20-60 min |
| `question`, `options`, `answer` | `options` is empty for the 63 numerical VSI-Bench items; `answer` is the option letter or the number |
| `answer_type` | `multiple_choice` (937) or `numerical` (63) |
| `categories` | JSON string with the source-specific tags (task type, domain, duration group, discipline, ...; LVBench `question_type` is a list) |
| `blind_correct_models` | how many of the three filter models solved the item from black frames (0 or 1) |
## Composition
| source | items |
|---|---|
| Video-MME | 312 |
| LongVideoBench (val, video) | 240 |
| VideoMMMU (perception / comprehension / adaptation) | 60 / 60 / 60 |
| LVBench | 146 |
| VSI-Bench | 122 |
## Running it
The task `metav` of [lmms_eval_video](https://github.com/soumyasj/lmms_eval_video) evaluates this dataset directly
(one subtask per source, each scored with the source benchmark's own parser and metric); the first run downloads
this repo to `$HF_HOME/meta_video`:
```bash
python -m lmms_eval --model qwen2_5_vl --model_args pretrained=Qwen/Qwen2.5-VL-7B-Instruct,max_num_frames=32 \
--tasks metav --batch_size 1 --log_samples --output_path logs/metav
python eval_scripts/score_metav.py --write logs/metav/*/*_results.json
```
## Scoring
Accuracy per source (exact option match; VSI-Bench numerical items use mean relative accuracy),
then the macro average over the five sources (VideoMMMU = mean of its three tracks).
## Licenses
The question lists are released under CC BY 4.0. The videos remain under the licenses and terms of
their original benchmarks and sources (Video-MME, LongVideoBench, LVBench, VideoMMMU, VSI-Bench /
ScanNet, ScanNet++, ARKitScenes); please follow those terms.
## Citation
```bibtex
@article{jahagirdar2026metavideo,
title = {Meta-Video: Efficient Standardized Evaluation for Video Understanding},
author = {Jahagirdar, Soumya Shamarao and Kukleva, Anna and Udandarao, Vishaal and Thomas, Samuel and Kingsbury, Brian and Feris, Rogerio and Glass, James R. and Prabhu, Ameya and Kuehne, Hilde},
journal = {arXiv preprint arXiv:2610.18535},
year = {2026}
}
```