Evaluating Relevance in RAG: Test Results and Findings

Relevance Eval: It takes a question and a context chunk, it then asks if the chunk can answer the question. It evaluates the quality of the retrieval process of RAG. 🧪 Test results of relevance prompt and Claude: 0.34 F1 not usable