RAG Retriever

Noun · AI & Machine Learning

Definitions

  1. The retriever component in a retrieval-augmented generation system that selects candidate documents or chunks for the model to use as context. Retriever quality is often the main bottleneck in RAG performance.

    In plain English: The component that retrieves relevant source material in a RAG system.

    Example: "They replaced the original RAG retriever after evaluation showed it overfavored semantically similar but outdated help pages."

Related Terms