anything-llm/server/utils/vectorDbProviders/pinecone/PINECONE_SETUP.md
Sean Hatfield 2f3db0e63a
[FEAT] support pinecone serverless (#639)
* migrate pinecone package to latest version and migrate pinecone vectordb provider class

* remove pinecone environment name env variable and update docs to reflect removal & serverless support complete

* migrate query for pinecone db

* typo in log

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-01-22 16:41:20 -08:00

750 B

How to setup Pinecone Vector Database for AnythingLLM

Official Pinecone Docs for reference.

How to get started

Requirements

  • Pinecone account with index that allows namespaces.

Note: Namespaces are not supported in gcp-starter environments and are required to work with AnythingLLM.

Instructions

  • Create an index on your Pinecone account. Name can be anything eg: my-primary-index
  • Metric cosine
  • Dimensions 1536 since we use OpenAI for embeddings
  • 1 pod, all other default settings are fine.
VECTOR_DB="pinecone"
PINECONE_API_KEY=sklive-123xyz
PINECONE_INDEX=my-primary-index # the value from the first instruction!