Agent Memory Distillation：通过层次化教师记忆增强小型LLM智能体
原标题：Agent Memory Distillation: Empowering Small LLM Agents with Hierarchical Teacher Memory
AI 摘要
本文提出了一种名为Agent Memory Distillation (AMD)的无训练框架，通过层次化记忆将大型教师智能体的结构化知识迁移到小型学生智能体。AMD构建了工作流记忆、子任务记忆和函数记忆三种互补记忆类型，并在三个工具使用基准上使用四个学生模型进行评估，平均准确率分别提升27.2%、11.2%和3.4%。分析表明子任务记忆贡献最大，教师有效性取决于教师能力和学生兼容性，4B参数的学生模型受益最多。
正文节选
Agent Memory Distillation: Empowering Small LLM Agents with Hierarchical Teacher Memory Abstract Memory systems have shown promise for improving agent performance, but their potential remains largely unexplored for small language models, which struggle to generate sufficient successful trajectories on their own. We propose Agent Memory Distillation (AMD), a training-free framework that transfers structured knowledge from a large teacher agent to a small student agent through hierarchical memory.