[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fTO5iGxQQr5EuIA1F-6bzDMatk-HNKjDlOADhH9vSQF8":3},{"item":4,"related":44},{"id":5,"type":6,"title":7,"slug":8,"summary":9,"body":10,"coverUrl":11,"productScreenshots":12,"productLinks":13,"authorName":14,"authorUrl":15,"authorSubject":16,"category":17,"tags":22,"sourceLabel":35,"sourceName":36,"sourceUrl":36,"status":37,"seoTitle":36,"seoDescription":36,"canonicalUrl":36,"isFeatured":38,"sno":39,"sortOrder":40,"publishedAt":41,"updatedAt":42,"createdAt":43},"74dedc2e-6a66-4481-aef2-5cffc3ba338d","article","嵌入模型（Embeddings）：向量数据库能搜「意思」，全靠它","embedding-models-vector-search","向量库怎么懂「意思相近」？靠嵌入模型把文字变成向量。本文讲清它的工作原理、余弦相似度检索，给出 sentence-transformers 最小示例，以及模型选型、维度统一、中英差异等取舍。","你让向量库「找意思相近的句子」，它怎么懂「意思」？靠嵌入模型（Embeddings）：把文字变成一串数字（向量），意思越近，数字越近。它是语义搜索和 RAG 真正的地基——没有它，模型只能靠关键词硬匹配。\n\n## 为什么需要嵌入\n\n传统搜索靠关键词匹配，搜「怎么给猫降温」找不到「猫咪中暑怎么办」。嵌入把文本映射到向量空间，把相近语义聚在一起，才能按「意思」而不是「字面」检索。\n\n## 它是怎么工作的\n\n嵌入模型（如 BGE、OpenAI text-embedding）是个神经网络，把变长文本压成定长向量（常见 768 或 1536 维）。训练目标是「语义相近的文本，向量距离小」。检索时把 query 也编码，算余弦相似度，找最近的那些。\n\n```mermaid\nflowchart LR\n    A[文本] --> B[嵌入模型]\n    B --> C[向量]\n    C --> D[存入向量库]\n    E[查询] --> B\n    D --> F[相似度检索]\n    B --> F\n    F --> G[返回相近文本]\n```\n\n## 取舍与边界\n\n- **模型要选对**：通用嵌入未必适合你的领域（法律、医疗），必要时用领域数据微调。\n- **维度与成本权衡**：维度越高通常越准，但存储、检索都更贵更慢，按场景取舍。\n- **中英文差异**：混用中英文语料要选多语言模型，否则跨语言检索会崩。\n- **维度必须统一**：检索和入库一定要用同一个模型、同一维度，否则向量不可比，检索全乱。\n\n## Tips\n\n- 任何「按意思搜」的需求，第一步就是选好嵌入模型。\n- 中文场景优先试 BGE、m3e 等多语言\u002F中文模型，别直接套英文默认。\n- 入库和检索用同一模型同一维度，这是铁律。\n- 领域强相关的语料，用该领域样本微调嵌入，召回率提升明显。\n- 嵌入质量直接决定 RAG 上限，值得在它上面多花时间。","https:\u002F\u002Foxqtewbrpuiouqqjrvdv.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fpublic-media\u002F2026-07-22\u002Facd40af8-276a-48bc-8d62-dcd52c124590.jpg",[],[],"Foundit AI","https:\u002F\u002Ffoundit.cn","f39339b1-aaa6-4e86-b0c2-a6e6a21113b5",{"id":18,"name":19,"slug":20,"description":21},"6179d3b6-dc34-4483-9ded-3cd9f1b37a47","科普","abbreviation","介绍各领域新兴概念",[23,27,31],{"id":24,"name":25,"slug":26},"4c2bbea6-eab7-40a8-8447-1de478ff7749","分析","analyse",{"id":28,"name":29,"slug":30},"7c76bfc2-f80f-4ee0-a95d-27bd8708b434","技术","slug",{"id":32,"name":33,"slug":34},"a202d639-99a6-488a-a712-4d4c6ffd7e15","开发","dev","资料来源",null,"published",false,68,0,"2026-07-22T00:00:00.000Z","2026-07-23T01:16:32.032Z","2026-07-20T10:23:37.966Z",[45,55,64],{"id":46,"type":6,"title":47,"slug":48,"summary":49,"coverUrl":50,"authorName":51,"sno":52,"publishedAt":53,"createdAt":54},"9ccde95c-d754-4808-91f7-488f392e3eeb","你的品牌在AI眼里到底存不存在？这套系统说了算","automated-geo-monitoring-system","靠手动抽查来验证GEO效果，本质上是在跟概率玩游戏。赢一次，不代表能一直赢。","\u002Fuploads\u002F2026-08-07\u002Fdf111c0d-f14a-4b2a-8347-141e96b71654.jpg","Foundit",1,"2026-08-07T00:00:00.000Z","2026-08-07T04:31:30.843Z",{"id":56,"type":6,"title":57,"slug":58,"summary":59,"coverUrl":60,"authorName":14,"sno":61,"publishedAt":62,"createdAt":63},"544fc658-c911-4de6-93b0-d2520087119a","MCP：AI 的「USB-C」时刻","mcp-ai-usb-c-moment","以前每个 AI 应用都要为 GitHub、数据库、日历各写一套私有连接器，这是 M×N 的集成噩梦，直到 MCP 的出现","https:\u002F\u002Foxqtewbrpuiouqqjrvdv.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fpublic-media\u002F2026-07-19\u002Fc3c06d99-a0ac-40ac-a283-7e77aabb4c4c.jpg",46,"2026-07-20T00:00:00.000Z","2026-07-19T16:13:40.316Z",{"id":65,"type":6,"title":66,"slug":67,"summary":68,"coverUrl":69,"authorName":51,"sno":70,"publishedAt":71,"createdAt":72},"1bb735e6-efec-4ad5-9065-8ccfd38f5b5d","一张 JPEG 为什么会长出方块、蚊子噪点和白色光环？","why-jpeg-images-have-compression-artifacts","JPEG 通过色度抽样、分块变换、量化和熵编码大幅缩小照片，其中量化会永久舍弃部分细节。压缩过强或反复保存后，方块、边缘光环与蚊子噪点便会显现。本文从人眼偏好出发解释这些痕迹如何产生。","\u002Fuploads\u002F2026-08-14\u002F8011c7fa-472c-4599-a65a-c993b095375f.jpg",60,"2026-08-14T00:00:00.000Z","2026-08-14T03:06:16.293Z"]