RAG (retrieval-augmented generation)

GPTKB entity

Statements (44)
Predicate Object
gptkbp:instanceOf machine learning technique
natural language processing method
gptkbp:application open-domain question answering
knowledge-intensive tasks
gptkbp:architecture retriever-generator pipeline
gptkbp:arXivID 2005.11401
gptkbp:author gptkb:Mike_Lewis
gptkb:Sebastian_Riedel
gptkb:Patrick_Lewis
gptkb:Douwe_Kiela
gptkb:Fabio_Petroni
gptkb:Tim_Rocktäschel
gptkb:Aleksandra_Piktus
gptkb:Ethan_Perez
gptkb:Heinrich_Küttler
gptkb:Naman_Goyal
gptkb:Vladimir_Karpukhin
gptkb:Wen-tau_Yih
gptkbp:citation gptkb:NeurIPS_2020
gptkb:Retrieval-Augmented_Generation_for_Knowledge-Intensive_NLP_Tasks
2020
gptkbp:combines generator
retrieval
gptkbp:developedBy gptkb:Facebook_AI_Research
gptkbp:generatorModelExample gptkb:BART
gptkb:T5
gptkbp:hasConcept retrieves relevant documents and generates answers conditioned on them
https://www.w3.org/2000/01/rdf-schema#label RAG (retrieval-augmented generation)
gptkbp:improves factual accuracy of language models
gptkbp:input query
gptkbp:introducedIn 2020
gptkbp:openSource gptkb:Hugging_Face_Transformers
gptkbp:output generated answer
gptkbp:reduces hallucination in language models
gptkbp:relatedTo generative models
hybrid models
retrieval-based models
gptkbp:retrievalSource document corpus
external knowledge base
gptkbp:retrieverModelExample gptkb:DPR_(Dense_Passage_Retriever)
gptkbp:uses generator model
retriever model
gptkbp:bfsParent gptkb:Llama_3
gptkbp:bfsLayer 6