[{"data":1,"prerenderedAt":134},["ShallowReactive",2],{"/cn/open_source/home/overview":3},{"id":4,"title":5,"avatar":6,"banner":7,"body":8,"category":6,"desc":115,"description":107,"extension":116,"links":117,"meta":129,"navigation":6,"path":130,"seo":131,"stem":132,"__hash__":133},"docs/cn/open_source/home/overview.md","MemOS 文档",null,"https://statics.memtensor.com.cn/memos/memos-banner.gif",{"type":9,"value":10,"toc":106},"minimark",[11,16,20,27,34,37,77,80,89,92,100,103],[12,13,15],"h2",{"id":14},"什么是-memos","什么是 MemOS？",[17,18,19],"p",{},"随着大型语言模型（LLMs）的不断演进，其所承担的任务日益复杂，包括多轮对话、规划、决策制定以及个性化代理等。在此背景下，如何高效管理和利用记忆，成为实现长期智能与适应性能力的关键因素。\n然而，主流 LLM 架构往往在记忆结构化、管理和集成方面存在不足，导致知识更新成本高、行为状态不可持续以及难以积累用户偏好。",[17,21,22,26],{},[23,24,25],"strong",{},"MemOS"," 通过将记忆重新定义为具有统一结构、生命周期管理和调度策略的核心一级资源来解决这些挑战。它提供了一个 Python 包，为基于 LLM 的应用程序提供统一的记忆层，实现持久化、结构化和高效的记忆操作。这使 LLMs 具备长期知识保留、强大的上下文管理和记忆增强推理能力，支持更智能和自适应的行为。",[17,28,29],{},[30,31],"img",{"alt":32,"src":33},"MemOS Architecture","https://statics.memtensor.com.cn/memos/memos-architecture.png",[12,35,36],{"id":36},"主要特性",[38,39,40,47,53,59,65,71],"ul",{},[41,42,43,46],"li",{},[23,44,45],{},"模块化记忆架构","：支持明文、激活（KV cache）和参数（适配器/LoRA）记忆。",[41,48,49,52],{},[23,50,51],{},"MemCube","：所有记忆类型的统一容器，易于加载/保存和 API 访问。",[41,54,55,58],{},[23,56,57],{},"MOS","：LLMs 的记忆增强系统，具有即插即用的记忆模块。",[41,60,61,64],{},[23,62,63],{},"基于图的后端","：原生支持 Neo4j 和其他图数据库，用于结构化、可解释的记忆。",[41,66,67,70],{},[23,68,69],{},"易于集成","：与 HuggingFace、Ollama 和自定义 LLMs 兼容。",[41,72,73,76],{},[23,74,75],{},"可扩展","：添加您自己的记忆模块或后端。",[12,78,79],{"id":79},"安装",[17,81,82,83,88],{},"请参阅我们的 ",[84,85,87],"a",{"href":86},"/open_source/getting_started/installation","安装指南"," 获取完整的安装说明，包括基础安装、可选依赖项和外部依赖项。",[12,90,91],{"id":91},"贡献",[17,93,94,95,99],{},"我们欢迎贡献！请参阅 ",[84,96,98],{"href":97},"/open_source/contribution/overview","贡献指南"," 了解设置环境和提交 pull request 的详细信息。",[12,101,102],{"id":102},"许可证",[17,104,105],{},"MemOS 在 Apache 2.0 许可证下发布。",{"title":107,"searchDepth":108,"depth":108,"links":109},"",2,[110,111,112,113,114],{"id":14,"depth":108,"text":15},{"id":36,"depth":108,"text":36},{"id":79,"depth":108,"text":79},{"id":91,"depth":108,"text":91},{"id":102,"depth":108,"text":102},"欢迎来到 MemOS 官方文档 – 一个专为大型语言模型 (LLMs) 提供高级模块化记忆功能的 Python 包。","md",[118,125],{"label":119,"to":120,"target":121,"avatar":122},"PyPI","https://pypi.org/project/MemoryOS/","_blank",{"src":123,"alt":124},"https://statics.memtensor.com.cn/icon/pypi.svg","PyPI logo",{"label":126,"to":127,"target":121,"icon":128},"Open Source","https://github.com/MemTensor/MemOS","i-simple-icons-github",{},"/cn/open_source/home/overview",{"title":5,"description":107},"cn/open_source/home/overview","q7qD3PuG2hhwhJut6OmzULVmNsdLV0WUPX-FmdjwjaQ",1758590736182]