arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2512.20845cs.AIcs.MA
收藏

MAR:多智能体反思提升大语言模型的推理能力

MAR:Multi-Agent Reflexion Improves Reasoning Abilities in LLMs

University of California, Berkeley(加州大学伯克利分校)

Onat Ozer, Yuchen Wang, Grace Wu, Daniel Dosti, Honghao Zhang, Vivi De La Rue

AI总结:

提出多智能体反思框架,通过多角色辩论生成多样化反思,解决单模型反思中的思维退化问题,在HotPot QA和HumanEval上分别达到47% EM和82.7%准确率。

摘要:

大语言模型已展现出通过反思自身错误并据此行动来提升推理任务性能的能力。然而,同一LLM对自身的持续反思会表现出思维退化,即即使知道错误,LLM仍会反复重复相同错误。为解决此问题,我们引入多智能体与多角色辩论者作为生成反思的方法。通过大量实验,我们发现这能导致LLM智能体生成的反思具有更好的多样性。我们在HotPot QA(问答)上展示了47%的精确匹配准确率,在HumanEval(编程)上展示了82.7%的准确率,这两项性能均超越了单一LLM的反思。

英文摘要:

LLMs have shown the capacity to improve their performance on reasoning tasks through reflecting on their mistakes, and acting with these reflections in mind. However, continual reflections of the same LLM onto itself exhibit degeneration of thought, where the LLM continues to repeat the same errors again and again even with the knowledge that its wrong. To address this problem, we instead introduce multi-agent with multi-persona debators as the method to generate reflections. Through out extensive experimentation, we've found that the leads to better diversity of in the reflections generated by the llm agent. We demonstrate an accuracy of 47% EM HotPot QA (question answering) and 82.7% on HumanEval (programming), both performances surpassing reflection with a single llm.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑