Video models are zero-shot learners and reasoners

Wiedemer, Thaddäus, Li, Yuxuan, Vicol, Paul · arXiv · 2025