No HNSW index in pgvector vector store

[holasoftware]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

Not applicable

Error Message and Stack Trace (if applicable)

No response

Description

There is no HNSW index in the pgvector vector store:

https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/vectorstores/pgvector.py

Unlike the pgembedding vectore store:

https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/vectorstores/pgembedding.py#L192

System Info

Not applicable

[wulifu2hao]

Hi @holasoftware , I wonder why you need the hnsw index functionality to be available in pgvector.py if it is already in pgembedding.py ?

[wulifu2hao]

PR: langchain-ai/langchain-postgres#85

[jackbravo]

Hi @holasoftware , I wonder why you need the hnsw index functionality to be available in pgvector.py if it is already in pgembedding.py ?

pgembedding is not for pgvector, but for pg_embedding. They are different postgres extensions. And actually there is also pgvecto_rs (docs), meant for the third pg extension pgvecto.rs :-p.

So this PR is for pgvector, which is I think more widely used, since that extension is supported by AWS and GCP, for example.

[holasoftware]

pg_embedding and pgvector are 2 different postgresSQL extensions. Without the HNSW index in the embedding field, the pgvector vector store is not making an efficient search. I suppose that it's making an exhaustive search.