Find answers from the community

Updated 2 months ago

Hi! In your experience, what are the

Hi! In your experience, what are the best approaches to improve retrieval?

Now all I do is: chunking using chunk size 500 tokens, embedding to qdrant, then search “my question vs embeddings”.

The questions are predefined btw

2 comments

LLogan M

hybrid search, reranking, query fusion retriever, sub question query engine

LLogan M

qdrant has sparse embedding support, which works pretty well with reranking 🙂 The other two would be additional improvements

Add a reply