RAG Evaluation

A specialized evaluation discipline for Retrieval-Augmented Generation applications. Standard metrics include retrieval recall (does the retriever surface the right documents), context relevance (are the retrieved chunks actually about the query), answer groundedness (does the answer stay within the retrieved context), and answer faithfulness (does the answer accurately reflect what the context says).

Definición editorial de DecipherU. Para definiciones técnicas autorizadas, consulta NIST AI RMF, ISO/IEC 42001, o las publicaciones oficiales de los laboratorios de IA.