Evaluate a Retrieval-Augmented Generation System

You are a data scientist working on a retrieval-augmented generation (RAG) assistant. The assistant accepts a user question, retrieves supporting documents from a large corpus, and produces an answer with citations back to those documents. Your task is to design a rigorous evaluation framework for this system. The framework must measure the system along these axes: how much relevant evidence the retriever finds, whether the final answer is correct, whether every claim in the…