Evaluation Methodology

This page compares a baseline LLM answer against the Agentic RAG system. The baseline answer is generated without document retrieval. The Agentic RAG answer uses the planner, query rewriter, retriever, evidence evaluator, and answer generator. An evaluator agent then scores both answers from 1 to 5.

Run Evaluation