Architecture
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Meta·May 22, 2020
Patrick Lewis, Ethan Perez, Aleksandra Piktus
View on arXivTL;DR
Combines a retriever with a generator so answers are grounded in fetched documents.
Why it matters
Coined RAG and defined the dominant pattern for question answering over private or fresh data.