UC San Diego 提出通过心理模型奖励训练 7B 智能体
UC San Diego 用心理模型奖励训练 7B 智能体
UC San Diego 研究人员在论文《Mental-Models for Multi-Agent Systems》中提出，将心理理论（ToM）融入训练奖励机制，使 7B 智能体能够推断伙伴的信念与目标。该方法通过训练模型在 RL 阶段为候选回复打分，从而让智能体在部署时无需额外模型即可独立运行，有效提升了多智能体协作中的意图识别准确性。
Rohan Paul· @rohanpaul_ai · X·2026-10-12 15:14· 50 分钟前AI 评分39
AI 导读
UC San Diego 新论文提出把心理理论（ToM）写进训练奖励，让 7B 智能体学会推断伙伴的信念与目标，部署时无需额外模型即可独立运行。该方法训练一个推断伙伴信念、意图和想法的模型，在 RL 阶段为候选回复打分，最终智能体单独运行。论文题为《Mental-Models for Multi-Agent Systems》，编号 arXiv:2610.12453。
正文
A new UC San Diego paper builds theory of mind into a 7B agent through its training reward, so the agent learns to read a partner’s beliefs and goals and then runs on its own with no extra model at deployment.
LLM agents that negotiate, cooperate or role-play usually guess the other side's goals from prompt text or memory summaries. They keep no explicit estimate of the partner's knowledge or plans, so they can misread intent.
They train a model that infers the partner's beliefs, intents and thoughts, including what the partner thinks of the agent. That model scores candidate replies during RL, and the final agent runs alone without it.
For agents that negotiate or collaborate, train with a reward that judges replies against an inferred model of the partner's goals and beliefs.
– arxiv. org/abs/2610.12453
Title: "Mental-Models for Multi-Agent Systems"
来源：Rohan Paul · x.com
来源：AIHot · aihot.news