1 paper · 1 filter
Mingkuan Feng, Jinyang Wu, Hao Gu +5
Agentic multimodal models perform diverse operations on an image via code and reason over the returned view, an effective paradigm for fine-grained visual question answering. Howev…