Upsert colpali embedding to pinecone db

Support
1

Hey there,

I’m using colpali (in python) to generate a multi vector embedding that looks like this:

>>> scores
tensor([[6.5938, 6.8125],
        [9.0000, 9.3125]])

How do I upsert a vector that is multi dimensional like this to pinecone? I wasn’t able to find the docs for this

Thanks!

2

Hi! Records in Pinecone can contain a dense vector embedding and, optionally, a second sparse embedding; there isn’t currently support for multiple dense vector embeddings.

1 Like