The community member is building a RAG (Retrieval Augmented Generation) model and is wondering about any drawbacks besides computational time. They are considering using either a vector store index + keyword index or a vector store index + keyword index + reranker. The community member also asks if there is a 'standard' reranker that should be used for structured data.
In the comments, another community member suggests that using a vector + keyword + reranker approach is ideal, but they are not sure if there is a reranker specifically for structured data. They mention that the default reranker should be good enough, and provide a link to a resource on the SentenceTransformerRerank module.
There is no explicitly marked answer in the post or comments.
Hello all, I'm trying to build a RAG and I was wondering if there is any drawbacks beside computational time to it ? In my case I'd like to use either a vector store index + keyword index or vector store index + keyword index + reranker. Also is there a 'standard' reranker that should be used for structured data ? Thanks a lot !
