{"@context":"https://schema.org","@type":"CreativeWork","@id":"https://forgecascade.org/public/capsules/fd61123d-3625-42a8-a575-82cbf4c01d55","identifier":"fd61123d-3625-42a8-a575-82cbf4c01d55","url":"https://forgecascade.org/public/capsules/fd61123d-3625-42a8-a575-82cbf4c01d55","name":"Retrieval-Augmented Generation for Knowledge-Intensive NLP","text":"Lewis and coauthors introduce retrieval-augmented generation models that combine a pretrained seq2seq generator with a dense vector index of Wikipedia accessed through a neural retriever. The paper compares sequence-level and token-level retrieval conditioning and reports gains on knowledge-intensive NLP tasks. Use this as a source-backed reference for the original RAG formulation and its motivation around provenance and updateable external memory.\n\nSources:\n- https://arxiv.org/abs/2005.11401","keywords":["rag","retrieval","non-parametric-memory","provenance","source-backed","public-reference","free-public-reference"],"about":[],"citation":[],"isPartOf":{"@type":"Dataset","name":"Forge Cascade Knowledge Graph","url":"https://forgecascade.org"},"publisher":{"@type":"Organization","name":"Forge Cascade","url":"https://forgecascade.org"},"dateCreated":"2026-04-11T21:13:02.685048Z","dateModified":"2026-06-19T10:29:06.586000Z","isBasedOn":"https://arxiv.org/abs/2005.11401","additionalProperty":[{"@type":"PropertyValue","name":"trust_level","value":100},{"@type":"PropertyValue","name":"verification_status","value":"sources_verified"},{"@type":"PropertyValue","name":"provenance_status","value":"valid"},{"@type":"PropertyValue","name":"evidence_level","value":"primary_source"},{"@type":"PropertyValue","name":"content_hash","value":"f04784888a87e71e264d32992ee43265b689b8df7bea07c58eb37663078c9482"}]}