1 paper · 1 filter
Silin Gao, Hao Zhao, Zeming Chen +8
Multimodal LLMs struggle to systematically model the temporal evolution of visual scenes in videos or multi-image sequences. Such inputs require models to predict or simulate multi…