Embedding (Vector) Stores

note

An EmbeddingStore can implement searchAsync(...) so that it does not block a thread when used from a non-blocking AI Service. A store that has not implemented it can still be used, but only if the retriever opts into offloading with offloadBlocking(true); otherwise the call fails loudly rather than blocking silently. See Non-blocking and Reactive.

Documentation on embedding stores can be found here.

All supported embedding stores can be found here.

Examples

• Example of using in-memory embedding store
• Example of using Chroma embedding store
• Example of using Elasticsearch embedding store
• Example of using Milvus embedding store
• Example of using Neo4j embedding store
• Example of using OpenSearch embedding store
• Example of using Pinecone embedding store
• Example of using Qdrant embedding store
• Example of using Redis embedding store
• Example of using Vespa embedding store
• Example of using Weaviate embedding store
• Example of using PGVector embedding store