Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Patrick Lewis and coauthors submitted 'Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks' on May 22, 2020. The paper introduces RAG models that combine a pre-trained sequence-to-sequence model as parametric memory with a dense vector index of Wikipedia accessed through a neural retriever as non-parametric memory.
Historical classification
This verified record belongs to LINEAiGE's artificial-intelligence history layer beginning in 1956.
Why it matters
A documented research-publication milestone for retrieval-augmented generation (RAG) models that jointly use parametric and retrieved non-parametric memory for language generation.
People, organizations & technologies
PEOPLE
Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Kuttler, Mike Lewis, Wen-tau Yih, Tim Rocktaschel, Sebastian Riedel, Douwe Kiela
TECHNOLOGIES
Retrieval-Augmented Generation, RAG, Natural language processing, Neural retrieval
Evidence & verification
STATE
verified
CONFIDENCE
high
LAST VERIFIED
2026-09-12T03:04:00+08:00
The original arXiv record identifies the twelve authors and the May 22 2020 submission date and directly describes retrieval-augmented generation models combining parametric and non-parametric memory. The official NeurIPS 2020 proceedings independently corroborate the paper title, author list, venue, and the same technical description. The day-precision date is scoped only to the arXiv submission milestone, not asserted as a universal invention or system-birth date. The NeurIPS paper lists Facebook AI Research, University College London, and New York University affiliations, but this candidate leaves organizations empty because author-affiliation mapping is not represented here without per-author semantics. No geography, ancestry, influence, causality, or relationship edges are inferred.
Sources
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
arXiv
PUBLISHED
2020-05-22
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
NeurIPS 2020
Trace the chronology
Record identity
LINEAiGE IDretrieval-augmented-generation-2020