MemRefine: LLM-Guided Compression for Long-Term Agent Memory
- ID: 97936646
- 原文链接: https://arxiv.org/abs/2606.13177
- PDF: https://arxiv.org/pdf/2606.13177
- 作者: Minjae Kim, Jinheon Baek, Soyeong Jeong, Sung Ju Hwang
- 日期: 2026-06-11
- 更新: 2026-06-11
- 分类: infra
- 来源类型: paper
- 标签: agent-memory, compression, memory-budget, storage-management, retrieval, arxiv
- 质量评分: 4/5
- 抓取时间: 2026-09-06 12:21 UTC
中文导读
MemRefine：面向资源受限平台的预算化长期记忆管理长程交互中记忆库无限增长，冗余条目推高存储成本并通过挤占最有用证据而劣化检索；论文提出存储预算下的记忆管理任务在固定预算内维持已构建的记忆库，同时保留对未来交互有用的信息由于表面相似度难以反映事实价值，MemRefine 仅用相似度提出候选，由 LLM 引导压缩决策
为什么值得关注
记忆库也会超载：把预算化记忆管理立成任务，相似度只提名LLM 决定压缩谁，保未来有用的信息
关键信息
- 论文标题：MemRefine: LLM-Guided Compression for Long-Term Agent Memory
- 作者：Minjae Kim, Jinheon Baek, Soyeong Jeong, Sung Ju Hwang
- arXiv：https://arxiv.org/abs/2606.13177
- 发布时间：2026-06-11
- arXiv 分类：cs.CL, cs.AI, cs.LG
- 关联标签：agent-memory, compression, memory-budget, storage-management, retrieval, arxiv
English Abstract
Large language model (LLM) agents are increasingly expected to operate over long-term interactions, where information from past dialogues must be preserved and recalled to support future tasks. However, as interactions accumulate, the memory store grows without bound and fills with redundant entries that inflate storage cost and degrade retrieval by crowding out the most useful evidence. Furthermore, this is especially limiting on resource-constrained platforms with hard memory budgets, motivating us to formulate storage-budgeted memory management, the task of keeping an already constructed memory store within a fixed budget while preserving information useful for future interactions. To this end, we then propose MemRefine, an LLM-guided framework that, since surface similarity poorly reflects factual value, uses similarity only to propose candidate pairs and defers delete, merge, and preserve decisions to an LLM judge based on factual content, iterating until the budget is met. Across multiple memory frameworks and long-term conversation benchmarks, MemRefine consistently meets target budgets while preserving downstream performance and outperforming rule-based baselines under tight budgets.
English Summary
Large language model (LLM) agents are increasingly expected to operate over long-term interactions, where information from past dialogues must be preserved and recalled to support future tasks. However, as interactions accumulate, the memory store grows without bound and fills with redundant entries that inflate storage cost and degrade retrieval by crowding out the most useful evidence. Furthermore, this is especially limiting on resource-constrained platforms with hard memory budgets, motivating us to formulate storage-budgeted memory management, the task of keeping an already constructed memory store within a fixed budget while preserving information useful for future interactions. To this end, we then propose MemRefine, an LLM-guided framework that, since surface similarity poorly reflects factual value, uses similarity only to propose candidate pairs and defers delete, merge, and preserve decisions to an LLM judge based on factual content, iterating until the budget is met. Across multiple memory frameworks and long-term conversation benchmarks, MemRefine consistently meets target budgets while preserving downstream performance and outperforming rule-based baselines under tight budgets.
Obsidian Notes
- 内容由
opencli arxiv paper拉取 arXiv 元数据与摘要生成。 - 中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。