danny-avila / rag_api

ID-based RAG FastAPI: Integration with Langchain and PostgreSQL/pgvector
https://librechat.ai/
143 stars 60 forks source link

adds pinecone vector store #34

Open paulmikulskis opened 1 month ago

paulmikulskis commented 1 month ago

Allows the rag-api to function with Pinecone as a vector store. Configuration follows suit with the PGVector and MongoDB solutions by having details added in the .env file which are then templated into a connectionString.

The COLLECTION_NAME is used for the Pinecone Index name.

Settings needed to use Pinecone within the rag-api:

VECTOR_DB_TYPE=pinecone
COLLECTION_NAME=
PINECONE_API_KEY=
PINECONE_NAMESPACE=
danny-avila commented 1 month ago

Hi @paulmikulskis

There are some new conflicts.

I've added enums to help manage the addition of vector stores, please use them. Also, I would like to see all routes working with screenshots and configuration settings for your setup, because the last addition of a vector store (not yours) does not seem fully functional.

Thanks