Which embedding model does Vanna uses to convert text to embedding? #378
-
|
Is it OpenAI text embedding or some kind of Bert transformer, etc? Thanks! |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
|
This depends on the setup you're using. Generally speaking for each vector store, it'll be whatever the "default" is. For example, for ChromaDB, it used the default embedding function as defined here: https://github.com/vanna-ai/vanna/blob/main/src/vanna/chromadb/chromadb_vector.py#L12 Which can then be overridden as necessary. If you're setting up with vector storage that doesn't include it's own embedding functionality, then you can reference the |
Beta Was this translation helpful? Give feedback.
This depends on the setup you're using. Generally speaking for each vector store, it'll be whatever the "default" is. For example, for ChromaDB, it used the default embedding function as defined here:
https://github.com/vanna-ai/vanna/blob/main/src/vanna/chromadb/chromadb_vector.py#L12
Which can then be overridden as necessary.
If you're setting up with vector storage that doesn't include it's own embedding functionality, then you can reference the
generate_embeddingfunction, which would have an implementation like this (example for using OpenAI):https://github.com/vanna-ai/vanna/blob/main/src/vanna/openai/openai_embeddings.py#L34-L46