1 paper
Pengyiang Liu, Zhongyue Shi, Hongye Hao +7
Video understanding requires models to continuously track and update world state during playback. Although existing benchmarks have advanced video understanding evaluation across m…