← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8704 · 生成模型:GPT-6

向量库管理索引与元数据

理解等级:中级

向量库可提供近似或精确近邻、元数据过滤、更新删除和混合检索等能力,具体支持依产品版本。ANN以召回换延迟,过滤还会影响候选集合;演示能查出一个结果不等于证明实际规模下可靠。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Vector Databases Depending on your application, you may have to deal with millions or billions of vectors, with the need to generate and store new vectors and their associated metadata tags every day. Vector databases facilitate this. Both self-hosted and cloud-based, open source, and proprietary options are available. Weviate, Milvus, Pinecone, Chroma, Qdrant, and LanceDB are some of the popular vector databases. More established players

    参考来源

    - 位置:《Designing Large Language Model Applications》第11章《表示学习与嵌入》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch11.html) - 结论依据:摘录中关于向量库管理索引与元数据的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:Vector Databases Depending on your application, you may have to deal with millions or billions of vectors, with the need to generate and store new vectors and their associated metadata tags every day. Vector databases facilitate this. Both self-hosted and cloud-based, open source, and proprietary options are available. Weviate, Milvus, Pinecone, Chroma, Qdrant, and LanceDB are some of the popular vector databases. More established players