Video-LevelGauge: Investigating Contextual Positional Bias in Large Video Language Models
Video-LevelGauge is under the CC-BY-NC-SA-4.0 license.
It is derived from several previously published datasets (VideoMME, MLVU, VisDrone, UCF-Crime, and Ego4D). Please note that the original datasets may have their own licenses. Users must comply with the licenses of the original datasets when using this derived… See the full description on the dataset page:
https://huggingface.co/datasets/Cola-any/Video-LevelGauge.
