提出新型记忆架构，让智能体检索更准更快。
Storage Is Not Memory: A Retrieval-Centered Architecture for Agent Recall
- 记忆按事件原样保存，检索时动态处理，避免信息丢失
- 在多个测试集上准确率超现有方法，最高达93.0%
- 仅用普通CPU和SQLite，无需GPU或外部存储
传统智能体记忆在录入时就提取内容，导致查询前无法恢复丢弃的信息。我们提出True Memory，一种六层架构，将系统中心从存储模式转向多阶段检索流水线，对原始事件进行无损保留。整个系统以单个SQLite文件运行于通用CPU，无需外部数据库、向量索引、图存储或GPU。在LoCoMo数据集（1,540个问题，10次多会话对话）上，True Memory Pro达到93.0%准确率（3次运行均值），优于Mem0（61.4%）、Supermemory（65.4%）、Zep（约71%）和EverMemOS（94.5%，使用相同gpt-4.1-mini模型）。在LongMemEval（500个问题）上达87.8%；在BEAM-1M（700个问题，百万级令牌规模）上达76.6%，超过此前最优结果Hindsight的73.9%。56种配置的消融实验显示，性能最优组内差异仅1.3个百分点。
原文摘要 · Abstract (English)
Extraction at ingestion is the wrong primitive for agent memory: content discarded before the query is known cannot be recovered at retrieval time. We propose True Memory, a six-layer architecture that shifts the center of the system from a storage schema to a multi-stage retrieval pipeline operating over events preserved verbatim. The full system runs as a single SQLite file on commodity CPU with no external database, vector index, graph store, or GPU. On LoCoMo (1,540 questions across 10 multi-session conversations), True Memory Pro reaches 93.0% accuracy (3-run mean) against 61.4% for Mem0, 65.4% for Supermemory, approximately 71% for Zep, and 94.5% for EverMemOS under a matched gpt-4.1-mini answer model. On LongMemEval (500 questions), True Memory Pro reaches 87.8% (3-run mean). On BEAM-1M (700 questions at the 1-million-token scale), True Memory Pro reaches 76.6% (3-run mean), above the prior published result of 73.9% for Hindsight. A 56-configuration ablation shows a 1.3-percentage-point spread within the top-performing configuration family.
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。