1
Post-Training Recipe, More Than Model Family, Shapes Multi-Agent LLM Conversational Behavior
研究表明，多LLM系统的对话行为更多取决于后训练配方，而非模型家族差异，挑战了传统选型策略。
arXiv:2606.20632v1 Announce Type: cross Abstract: Multi-LLM systems use multiple language models to deliberate, judge each other's outputs, or coordin…