Hi thanks for the amazing course about RAG. Really appreciate the effort of making it.
I have a question on the value of Reranker. Say if in the query engine the similary_top_n is 6 and the top n for Reranker is 2, why not we just set similarity_top_n is 2 instead in the query engine? I’m confused why we want 2 models/systems to evaluate and filter the retrieved chunks?
Thank you so much for your help!