A framework that combines retrieval and generation to produce responses grounded in external knowledge, leveraging techniques like dense retrieval and knowledge retrieval.
More about Retrieval Augmented Generation (RAG)
Retrieval Augmented Generation (RAG) is an AI framework that combines a retriever model with a generator model to produce responses grounded in external knowledge. The retriever fetches relevant documents or pieces of information using methods like dense retrieval or semantic search, while the generator uses that information to create contextually accurate and fact-based outputs.
RAG frameworks are particularly effective for applications such as open-domain question answering, knowledge-grounded dialogue systems, and other knowledge-intensive tasks, ensuring outputs are accurate and well-grounded.
SiteSpeak uses RAG to answer from your own website and documents. See AI training.