1 paper match
Haoqing Wang, Xingrun Xing, Wei Xia +2
FaithEyes proposes a multi‑agent framework where a vision‑language model judges its own tool calls to ensure they are useful, improving both accuracy and tool faithfulness on visua…