I found the optimal top-k for my RAG system. Then I ran it again.

I built a small retrieval-augmented generation system over a corpus of research papers and wanted to...