Setting the embedding model for a specific retriever

Question

Hiya, I have a simple question but couldn't quite figure it out from the docs. Is it possible to set the embedding model to be used for a specific retriever - for example when I set a retriever using VectorStoreIndex.from_vector_store(my_store).as_retriever() ? - or when I'm setting the vector store itself for example the WeaviateVectorStore? - So far I see I can do this as a part of the Settings.embed_model but want to make it so that I can define a separate model per retriever if possible.

Logan M · Answer

By default it pulls the embed model attached to the index, i.e. from VectorStoreIndex.from_vector_store(..., embed_model=embed_model)

You can pass it in again though too into as_retriever(embed_model=embed_model, ...)

Tuana · Answer

Thank you! 🙏

Tuana · Answer

Thanks again! That helped and I presented my first llama index based demo yesterday at an Agents meetup in Amsterdam 😄

Logan M · Answer

Amazing! Glad that went well! :LlamaIndex:

Find answers from the community

Setting the embedding model for a specific retriever