WorldMM: Dynamic Multimodal Memory Agent for Long Video Reasoning

Yeo, Woongyeong, Kim, Kangsan, Yoon, Jaehong · arXiv · 2025