arXiv cs.CL· Jingwei Sun, Jianing Zhu, Jiangchao Yao, Tongliang Liu, Bo Han·· 12 小时前AI 评分40
TriMem：超越原子事实的终身 LLM 智能体记忆新方案
Rethinking How to Remember: Beyond Atomic Facts in Lifelong LLM Agent Memory
AI 导读
TriMem 提出三种共存的记忆表示粒度——带来源标识的原始对话片段、抽取的原子事实、以及聚合分散事实的合成画像，兼顾存储保真、检索效率与深度推理。该方法采用 TextGrad 提示词优化，通过响应质量反馈迭代改进抽取与画像提示词，无需更新参数即可实现终身进化。
正文 · 原文
Abstract:To enable reliable long-term interaction, LLM agents require a memory system that can faithfully store, efficiently retrieve, and deeply reason over accumulated dialogue history. Most existing methods adopt an extracted fact based paradigm: handcrafted static prompts compress raw dialogues into atomic facts, which are then stored, matched, and injected into downstream reasoning. Nevertheless, such fact-centric designs inevitably discard fine-grained details in original dialogues and fail to support deep reasoning over scattered isolated facts. Moreover, static prompts cannot maintain consistent extraction granularity across diverse dialogue styles. To address these limitations, we propose TriMem, which maintains three coexisting representation granularities, including raw dialogue segments anchored by source identifiers for storage fidelity, extracted atomic facts for efficient memory retrieval, synthesized profiles that aggregate dispersed facts into holistic semantic understanding for deep reasoning. We further adopt TextGrad-based prompt optimization, which iteratively refines extraction and profiling prompts via response quality feedback, achieving lifelong evolution without any parameter updating. Extensive experiments on LoCoMo and PerLTQA across multiple LLM backbones demonstrate that TriMem consistently outperforms strong memory baselines. The code is available at this https URL .
| Comments: | Accepted by NeurIPS 2026 |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2605.19952 [cs.CL] |
| (or arXiv:2605.19952v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.19952 arXiv-issued DOI via DataCite |
Submission history
From: Jingwei Sun [view email]
[v1]
Tue, 19 May 2026 15:05:06 UTC (3,421 KB)
[v2]
Tue, 20 Oct 2026 11:37:16 UTC (2,977 KB)
来源：arXiv cs.CL · arxiv.org