让大模型自主决定何时查记忆、何时反思，提升问答准确性。
MemR$^3$: Memory Retrieval via Reflective Reasoning for LLM Agents
- 用路由机制自动选择查、思、答三种动作，实现闭环控制。
- 在LoCoMo基准上，对RAG和Zep的性能分别提升7.29%和1.94%。
- 适合需要精准记忆检索的智能体系统，可直接接入现有存储结构。
记忆系统旨在让大语言模型智能体利用过往经验。然而，多数已部署的记忆系统主要优化压缩与存储，对记忆检索的显式闭环控制关注较少。基于此，我们构建了名为MemR$^3$的自主、准确且兼容的记忆检索系统，包含两个核心机制：1）路由模块，可在检索、反思、回答三类动作中动态选择，以优化答案质量；2）全局证据缺口追踪器，显式呈现推理过程并追踪证据收集状态。该设计突破传统‘检索后回答’流程，引入闭环控制，实现自主决策。在LoCoMo基准上的实验证明，MemR$^3$在LLM-as-a-Judge评分上超越强基线，尤其在四类检索场景中表现优异，使用GPT-4.1-mini后端时，整体提升RAG（+7.29%）与Zep（+1.94%），可作为即插即用的控制器集成至现有记忆存储系统。
原文摘要 · Abstract (English)
Memory systems have been designed to leverage past experiences in Large Language Model (LLM) agents. However, many deployed memory systems primarily optimize compression and storage, with comparatively less emphasis on explicit, closed-loop control of memory retrieval. From this observation, we build memory retrieval as an autonomous, accurate, and compatible agent system, named MemR$^3$, which has two core mechanisms: 1) a router that selects among retrieve, reflect, and answer actions to optimize answer quality; 2) a global evidence-gap tracker that explicitly renders the answering process transparent and tracks the evidence collection process. This design departs from the standard retrieve-then-answer pipeline by introducing a closed-loop control mechanism that enables autonomous decision-making. Empirical results on the LoCoMo benchmark demonstrate that MemR$^3$ surpasses strong baselines on LLM-as-a-Judge score, and particularly, it improves existing retrievers across four categories with an overall improvement on RAG (+7.29%) and Zep (+1.94%) using GPT-4.1-mini backend, offering a plug-and-play controller for existing memory stores.
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。