1 paper
Kaiyu Li, Jiayu Wang, Zhi Wang +4
LLM-driven agents, particularly those using general frameworks like ReAct or human-inspired role-playing, often struggle in specialized domains that necessitate rigorously structur…