Mitigating Visual Context Degradation in Large Multimodal Models: A Training-Free Decoupled Agentic Framework

Hongrui Jia, Chaoya Jiang, Shikun Zhang, Wei Ye

Keywords: Vision, Language, and Reasoning

Chat is not available.