SEPM：基于任务图的自演化程序记忆用于多智能体协作
原标题：SEPM: Self-Evolving Procedural Memory Grounded in Task Graphs for Multi-Agent Collaboration
AI 摘要
该论文提出 SEPM（自演化程序记忆）框架，以任务依赖图作为多智能体在线协调与程序记忆演化的统一表示。执行中通过共享黑板记录结构化证据与任务依赖，检测目标、计划与世界状态偏差并基于证据重对齐；执行后将验证过的动作与依赖蒸馏为可复用程序，失败则转为警告与约束，程序池经创建、精炼、特化、退役演化。在 ALFWorld、WebArena、OfficeBench 上基于 AutoGen 与 DyLAN 的六个基准设置中均取得最高任务分数，AutoGen 下较无额外记忆基线分别提升 29.85、25.86、13.00 个百分点，代码已开源。
正文节选
SEPM: Self-Evolving Procedural Memory Grounded in Task Graphs for Multi-Agent Collaboration Thanks: This work was conducted during an internship at China Telecom and was supported in part by the China Telecom Key Research Project (No. 65996). Abstract Long-horizon multi-agent collaboration requires memory that preserves both successful actions and the coordination structure underlying them. Existing memory systems support information sharing or procedural reuse, but often leave the evolution of