In Retrieval-Augmented Generation (RAG), what are the two main components and their roles?

• A. The retriever stores the model's parameters; the generator updates them.
• B. The retriever generates responses; the generator selects documents.
• C. The retriever filters outputs; the generator rewrites them.
• D. The retriever fetches relevant documents and the generator uses those documents to produce answers.

Answer: (D)

In Retrieval-Augmented Generation, the system combines a retriever and a generator to work together. The retriever searches a large document collection and fetches passages that are relevant to the user's query. The generator then uses those retrieved documents as context to produce a final answer. This setup allows the model to ground its response in external information rather than relying solely on internal memorized knowledge, which often leads to more accurate and verifiable outputs.

This aligns with the statement that the retriever fetches relevant documents and the generator uses those documents to produce answers. The other descriptions mix up the roles—for example, treating the retriever as generating responses or as storing model parameters, or having the generator merely select or rewrite documents—so they don’t capture how the two parts actually collaborate in RAG.