向量存储#
向量存储包含已摄取文档块(有时也包含文档块本身)的嵌入向量。
简单向量存储#
默认情况下,LlamaIndex 使用一个简单的内存向量存储,非常适合快速实验。它们可以通过调用 vector_store.persist()(以及 SimpleVectorStore.from_persist_path(...) 分别)持久化到(并从)磁盘。
向量存储选项和功能支持#
LlamaIndex 支持 20 多种不同的向量存储选项。我们正在积极添加更多集成并改进每个选项的功能覆盖范围。
| Vector Store | 类型 | 元数据过滤 | 混合搜索 | 删除 | 存储文档 | 异步 |
|---|---|---|---|---|---|---|
| Alibaba Cloud OpenSearch | 云 | ✓ | ✓ | ✓ | ✓ | |
| Apache Cassandra® | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| Astra DB | 云 | ✓ | ✓ | ✓ | ||
| Azure AI Search | 云 | ✓ | ✓ | ✓ | ✓ | |
| Azure CosmosDB MongoDB | 云 | ✓ | ✓ | |||
| BaiduVectorDB | 云 | ✓ | ✓ | ✓ | ||
| ChatGPT Retrieval Plugin | 聚合器 | ✓ | ✓ | |||
| Chroma | 自托管 | ✓ | ✓ | ✓ | ||
| Couchbase | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| DashVector | 云 | ✓ | ✓ | ✓ | ✓ | |
| Databricks | 云 | ✓ | ✓ | ✓ | ||
| Deeplake | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| DocArray | 聚合器 | ✓ | ✓ | ✓ | ||
| DuckDB | 内存 / 自托管 | ✓ | ✓ | ✓ | ||
| DynamoDB | 云 | ✓ | ||||
| Elasticsearch | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | ✓ |
| FAISS | 内存 | |||||
| Google AlloyDB | 云 | ✓ | ✓ | ✓ | ✓ | |
| Google Cloud SQL Postgres | 云 | ✓ | ✓ | ✓ | ✓ | |
| Hnswlib | 内存 | |||||
| txtai | 内存 | |||||
| Jaguar | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| LanceDB | 云 | ✓ | ✓ | ✓ | ||
| Lantern | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | ✓ |
| Metal | 云 | ✓ | ✓ | ✓ | ||
| MongoDB Atlas | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| MyScale | 云 | ✓ | ✓ | ✓ | ✓ | |
| Milvus / Zilliz | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| Neo4jVector | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| OpenSearch | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | ✓ |
| Pinecone | 云 | ✓ | ✓ | ✓ | ✓ | |
| Postgres | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | ✓ |
| pgvecto.rs | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| Qdrant | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | ✓ |
| Redis | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| Simple | 内存 | ✓ | ✓ | |||
| SingleStore | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| Supabase | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| Tablestore | 云 | ✓ | ✓ | ✓ | ✓ | |
| Tair | 云 | ✓ | ✓ | ✓ | ||
| TiDB | 云 | ✓ | ✓ | ✓ | ||
| TencentVectorDB | 云 | ✓ | ✓ | ✓ | ✓ | |
| Timescale | ✓ | ✓ | ✓ | ✓ | ||
| Typesense | 自托管 / 云 | ✓ | ✓ | ✓ | ||
| Upstash | 云 | ✓ | ||||
| Vearch | 自托管 | ✓ | ✓ | ✓ | ||
| Vespa | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| Vertex AI Vector Search | 云 | ✓ | ✓ | ✓ | ||
| Weaviate | 自托管 / 云 | ✓ | ✓ | ✓ | ✓ | |
| WordLift | 云 | ✓ | ✓ | ✓ | ✓ | ✓ |
有关更多详细信息,请参阅向量存储集成。
示例 Notebooks#
- Alibaba Cloud OpenSearch
- Astra DB
- 异步索引创建
- Azure AI Search
- Azure Cosmos DB
- Baidu
- Cassandra
- Chromadb
- Couchbase
- Dash
- Databricks
- Deeplake
- DocArray HNSW
- DocArray 内存版
- DuckDB
- Espilla
- Google AlloyDB for PostgreSQL
- Google Cloud SQL for PostgreSQL
- Jaguar
- LanceDB
- Lantern
- Metal
- Milvus
- Milvus Async API
- Milvus Full-Text Search
- Milvus Hybrid Search
- MyScale
- ElasticSearch
- FAISS
- Hnswlib
- MongoDB Atlas
- Neo4j
- OpenSearch
- Pinecone
- Pinecone Hybrid Search
- PGvectoRS
- Postgres
- Redis
- Qdrant
- Qdrant 混合搜索
- Rockset
- Simple
- Supabase
- Tablestore
- Tair
- TiDB
- Tencent
- Timescale
- Upstash
- Vearch
- Vespa
- Vertex AI Vector Search
- Weaviate
- Weaviate Hybrid Search
- WordLift
- Zep