arXiv / Factual Memory / Token-level

WorldMM: Dynamic Multimodal Memory Agent for Long Video Reasoning

作者：Woongyeong Yeo, Kangsan Kim, Jaehong Yoon, Sung Ju Hwang · 来源：arXiv · 发布日期：2025/12

事实记忆词元级记忆
新标签打开 AlphaXiv
WorldMM: Dynamic Multimodal Memory Agent for Long Video Reasoning figure
AlphaXiv 中文概览（可滚动查看）