arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.04853cs.AI
收藏

MemMachine:一种保持事实真实性的记忆系统用于个性化AI代理

MemMachine: A Ground-Truth-Preserving Memory System for Personalized AI Agents

MemVerge, Inc.(MemVerge公司)

Shu Wang, Edwin Yu, Oscar Love, Tom Zhang, Tom Wong, Steve Scargall, Charles Fan

AI总结:

MemMachine是一种开源记忆系统,通过整合短期、长期事件和用户档案记忆,实现事实真实性保护,提升多会话交互的准确性和效率。

摘要:

大型语言模型(LLM)代理需要持久记忆以维持个性化、事实连续性和长期推理,但标准上下文窗口和检索增强生成(RAG)管道在多会话交互中会退化。我们提出了MemMachine,一种开源记忆系统,整合短期、长期事件和用户档案记忆,采用保持事实真实性的架构,存储完整对话回合并减少基于LLM的提取损失。MemMachine使用上下文化检索,扩展核对匹配的上下文,提高相关证据跨多个对话回合时的召回率。在多个基准测试中,MemMachine实现了强准确率-效率权衡:在LoCoMo上使用gpt4.1-mini达到0.9169;在LongMemEvalS(ICLR 2025)中,六维消融实验获得93.0%的准确率,检索阶段优化(检索深度调节+4.2%,上下文格式+2.0%,搜索提示设计+1.8%,查询偏差修正+1.4%)优于摄入阶段增益(如句子分块+0.8%)。GPT-5-mini在优化提示下超过GPT-5 2.6%,成为最经济高效的设置。与Mem0相比,MemMachine在匹配条件下使用约80%更少的输入令牌。一个配套的检索代理能够自适应地将查询路由到直接检索、并行分解或迭代查询链策略,分别在随机噪声条件下达到93.2%的HotpotQA-hard和92.6%的WikiMultiHop。这些结果表明,保持事件事实真实性并叠加自适应检索可为个性化LLM代理提供稳健、高效的长期记忆。

英文摘要:

Large Language Model (LLM) agents require persistent memory to maintain personalization, factual continuity, and long-horizon reasoning, yet standard context-window and retrieval-augmented generation (RAG) pipelines degrade over multi-session interactions. We present MemMachine, an open-source memory system that integrates short-term, long-term episodic, and profile memory within a ground-truth-preserving architecture that stores entire conversational episodes and reduces lossy LLM-based extraction. MemMachine uses contextualized retrieval that expands nucleus matches with surrounding context, improving recall when relevant evidence spans multiple dialogue turns. Across benchmarks, MemMachine achieves strong accuracy-efficiency tradeoffs: on LoCoMo it reaches 0.9169 using gpt4.1-mini; on LongMemEvalS (ICLR 2025), a six-dimension ablation yields 93.0 percent accuracy, with retrieval-stage optimizations -- retrieval depth tuning (+4.2 percent), context formatting (+2.0 percent), search prompt design (+1.8 percent), and query bias correction (+1.4 percent) -- outperforming ingestion-stage gains such as sentence chunking (+0.8 percent). GPT-5-mini exceeds GPT-5 by 2.6 percent when paired with optimized prompts, making it the most cost-efficient setup. Compared to Mem0, MemMachine uses roughly 80 percent fewer input tokens under matched conditions. A companion Retrieval Agent adaptively routes queries among direct retrieval, parallel decomposition, or iterative chain-of-query strategies, achieving 93.2 percent on HotpotQA-hard and 92.6 percent on WikiMultiHop under randomized-noise conditions. These results show that preserving episodic ground truth while layering adaptive retrieval yields robust, efficient long-term memory for personalized LLM agents.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑