This repository holds the dataset Vinoground, a temporal counterfactual dataset composed of 1000 short and natural video-caption pairs.This benchmark has also been integrated into \[lmms-eval\]. One can begin using it by cloning their repository. Evaluation is now made easier. You can also use the evaluation codes we provided in this repository.
4
13 commits
3 linked in READMEs
updated Aug 7, 2025
This repository holds the dataset Vinoground, a temporal counterfactual dataset composed of 1000 short and natural video-caption pairs.This benchmark has also been integrated into [lmms-eval]. One can begin using it by cloning their repository. Evaluation is now made easier. You can also use the evaluation codes we provided in this repository.
Project page: https://vinoground.github.io
Paper: https://arxiv.org/abs/2410.02763
Code: https://github.com/Vinoground/Vinoground
Leaderboard: https://vinoground.github.io/#leaderboard
9 commits
4 commits
This repository holds the dataset Vinoground, a temporal counterfactual dataset composed of 1000 short and natural video-caption pairs.This benchmark has also been integrated into \[lmms-eval\]. One can begin using it by cloning their repository. Evaluation is now made easier. You can also use the evaluation codes we provided in this repository.
4
13 commits
3 linked in READMEs
updated Aug 7, 2025
This repository holds the dataset Vinoground, a temporal counterfactual dataset composed of 1000 short and natural video-caption pairs.This benchmark has also been integrated into [lmms-eval]. One can begin using it by cloning their repository. Evaluation is now made easier. You can also use the evaluation codes we provided in this repository.
Project page: https://vinoground.github.io
Paper: https://arxiv.org/abs/2410.02763
Code: https://github.com/Vinoground/Vinoground
Leaderboard: https://vinoground.github.io/#leaderboard
9 commits
4 commits