1 paper
Yuhang Fu, Ruishan Fang, Jiaqi Shao +4
Does adding more agents help an LLM workflow once compared systems share the same benchmark loader, tool access, answer contract, usage accounting, and trajectory logging? We intro…