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 2610.19581cs.MAcs.AI
收藏

XferAgent:面向自改进大语言模型智能体的跨智能体经验迁移

XferAgent: Cross-Agent Experience Transfer for Self-Improving LLM Agents

McGill University(麦吉尔大学);McCormick School of Engineering, Northwestern University(西北大学麦考密克工程学院);Mila – Quebec AI Institute(米拉-魁北克人工智能研究所);Ohio University(俄亥俄大学)

Weihao Li, Jun Bai, Shijie Chen, Qincheng Lu, Xinyu Wang, Ziyang Song

AI总结:

本文提出XferAgent框架,解决现有跨智能体经验共享复用难问题,通过经验图组织经验,实验表明其性能优于同类方法且可提升累计任务成功率。

摘要:

大语言模型(LLM)智能体可复用先前任务执行的经验以提升后续任务性能。现有方法虽支持跨智能体经验共享,但智能体执行历史与任务需求的差异导致共享经验难以有效复用。本文提出XferAgent,一种跨智能体经验迁移框架,其在本地与共享经验图中组织可复用经验,协调异构智能体的兼容贡献同时避免不可靠合并。针对每个新任务,它选择共享经验并将其与目标智能体的本地经验整合以支撑任务执行。跨多样任务域的实验显示,该方法优于代表性基于记忆的方法,且能在异构智能体框架间实现有效经验迁移;随着智能体更新并复用经验,累计任务成功率在连续任务执行中也有所提升。

英文摘要:

Large language model (LLM) agents can reuse experience from prior task executions to improve performance on subsequent tasks. Although existing methods enable cross-agent experience sharing, differences in agents' execution histories and task requirements make shared experiences difficult to reuse effectively. We propose XferAgent, a framework for cross-agent experience transfer that organizes reusable experience in local and shared experience graphs. XferAgent reconciles compatible contributions from heterogeneous agents while avoiding unreliable merges. For each new task, it selects shared experiences and integrates them with the target agent's local experience to support task execution. Experiments across diverse task domains show improvements over representative memory-based methods and effective experience transfer between heterogeneous agent frameworks. Cumulative task success also increases over successive task passes as agents update and reuse their experiences.

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。

↑