1 paper
Jingfeng Chen, Jiawen Qian, Wendi Deng +5
Video understanding in multimodal large language models requires selecting informative frames from long, redundant videos under limited visual-token budgets. Existing methods often…