You are reviewing the retrieval layer for an LLM feature that answers user questions from an internal document index. The team wants a clear way to tell whether retrieval is good enough before tuning prompts or generation.
Walk me through how you would evaluate retrieval quality with metrics like recall@k and MRR.
You are reviewing the retrieval layer for an LLM feature that answers user questions from an internal document index. The team wants a clear way to tell whether retrieval is good enough before tuning prompts or generation.
Walk me through how you would evaluate retrieval quality with metrics like recall@k and MRR.
