让大模型自己判断哪些记忆该保留，提升长期智能体效率
What Deserves Memory: Adaptive Memory Distillation for LLM Agents
- 用可预测性评估记忆价值，替代人工设定的规则
- 实验显示记忆量减少50%以上，任务表现仍更优
- 适合长期决策类AI系统，如对话代理、自主机器人
大模型智能体的记忆系统难以自主判断何信息值得留存。现有方法依赖预设启发式规则，如重要性评分、情感标签或事实模板，反映的是设计者直觉而非数据本身。受认知科学启发，我们提出NEMORI——一种自适应记忆蒸馏框架，将经验未来效用的评估转化为可预测性的判断。该框架包含两个级联模块：情景记忆融合将原始交互整合为连贯叙事；语义知识蒸馏通过预测误差提取深层洞见。以蒸馏为核心，框架对下游管理方式保持无偏。大量实验表明，NEMORI在性能、效率和存储压缩方面均表现优异。研究证明，观察交互序列的内在特性，可为记忆设计提供一种数据驱动的可行替代方案。代码：https://github.com/nemori-ai/nemori。
原文摘要 · Abstract (English)
Memory systems for LLM agents struggle to determine what information deserves retention. Existing approaches rely on predefined heuristics such as importance scores, emotional tags, or factual templates, encoding designer intuition rather than learning from the data itself. Inspired by cognitive ideas, we propose NEMORI, an adaptive memory distillation framework that casts the assessment of experience's future utility as a matter of predictability. Specifically, NEMORI comprises two cascading modules: Episodic Memory Integration transforms raw interactions into coherent narratives, and Semantic Knowledge Distillation extracts insights via prediction error. Centering on distillation, the framework remains agnostic to downstream management. Extensive experiments confirm that NEMORI achieves strong performance, efficiency, and storage reduction. Our work suggests that observing the intrinsic properties of interaction sequences offers a viable, data-driven alternative to heuristic-based memory design. Code: https://github.com/nemori-ai/nemori.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。