{ "profile": "wrs-published-research-view-1", "edition": "ratings-e039106f30dec177", "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS", "entry": { "id": "borgeaud-et-al-retro-2021", "title": "Borgeaud and colleagues: RETRO chunked retrieval-enhanced language modeling", "summary": "Conditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores.", "boundary": "Retrieval-augmented generation is inherited; the contribution is the chunked architecture and scale study. Database storage, retrieval costs and data overlap remain relevant.", "area": "Retrieval-enhanced language models", "topic": "Artificial intelligence", "year": 2021, "lane": "HUMAN", "result": "ALGORITHM", "package_id": "borgeaud-et-al-retro-2021", "target": { "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "content_hash": "sha256:707a39ba4c7d2a04c6777d8b4a9e447b9a91c7d1266963f793bc43f614876931" }, "humans": [ "Sebastian Borgeaud", "Arthur Mensch", "Jordan Hoffmann", "Trevor Cai", "Eliza Rutherford", "Katie Millican", "George van den Driessche", "Jean-Baptiste Lespiau", "Bogdan Damoc", "Aidan Clark", "Diego de Las Casas", "Aurelia Guy", "Jacob Menick", "Roman Ring", "Tom Hennigan", "Saffron Huang", "Loren Maggiore", "Chris Jones", "Albin Cassirer", "Andy Brock", "Michela Paganini", "Geoffrey Irving", "Oriol Vinyals", "Simon Osindero", "Karen Simonyan", "Jack W. Rae", "Erich Elsen", "Laurent Sifre" ], "authors_complete": true, "human_role_note": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.", "agents": [], "related": [], "subresults": [], "scientific_assessment": { "source_review": "SOURCE_CURATED", "scientific_review": "NOT_REVIEWED_BY_WRS", "formal_replay": "NOT_RUN", "provider_attestation": "NONE" }, "sources": [ { "id": "cur20261015-arxiv-2112-04426v3", "title": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } ], "points": "95000", "published_grade": { "id": "borgeaud-et-al-retro-2021", "entry_id": "borgeaud-et-al-retro-2021", "band": "A", "modifier": "-", "target": { "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "content_hash": "sha256:707a39ba4c7d2a04c6777d8b4a9e447b9a91c7d1266963f793bc43f614876931" }, "what_changed": "Conditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores.", "scope": "Conditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores. Retrieval-augmented generation is inherited; the contribution is the chunked architecture and scale study. Database storage, retrieval costs and data overlap remain relevant.", "status": "PROVISIONAL_EDITORIAL", "scientific_correctness_certified": false }, "assessment_disposition": { "status": "ASSESSED", "label": "Provisionally assessed", "reason": "A current grade and separate reference score identify this exact contribution.", "target": { "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "content_hash": "sha256:707a39ba4c7d2a04c6777d8b4a9e447b9a91c7d1266963f793bc43f614876931" }, "edition": "ratings-e039106f30dec177" } }, "version": { "kind": "PUBLIC_RECORD_VIEW", "previous": null, "published_at": "2026-10-16T00:00:00Z", "publisher": { "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33", "role": "WRS_EDITORIAL_PUBLICATION" }, "record": { "contributors": [ { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Sebastian Borgeaud", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Arthur Mensch", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Jordan Hoffmann", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Trevor Cai", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Eliza Rutherford", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Katie Millican", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "George van den Driessche", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Jean-Baptiste Lespiau", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Bogdan Damoc", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Aidan Clark", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Diego de Las Casas", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Aurelia Guy", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Jacob Menick", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Roman Ring", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Tom Hennigan", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Saffron Huang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Loren Maggiore", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Chris Jones", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Albin Cassirer", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Andy Brock", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Michela Paganini", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Geoffrey Irving", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Oriol Vinyals", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Simon Osindero", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Karen Simonyan", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Jack W. Rae", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Erich Elsen", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Laurent Sifre", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } } ], "evidence": [ { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-0", "kind": "source_reference", "source": { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } } ], "origin": "HUMAN_LEGACY", "record_profile": "wrs-local-research-notes-0.1", "record_type": "research_note", "rrp_version": "0.1", "scope": { "description": "Retrieval-augmented generation is inherited; the contribution is the chunked architecture and scale study. Database storage, retrieval costs and data overlap remain relevant.", "domain": "Retrieval-enhanced language models" }, "sources": [ { "citation": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3" } ], "statement": "# Borgeaud and colleagues: RETRO chunked retrieval-enhanced language modeling\n\nConditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores.\n\n## Scientific scope\n\nRetrieval-augmented generation is inherited; the contribution is the chunked architecture and scale study. Database storage, retrieval costs and data overlap remain relevant.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nSebastian Borgeaud; Arthur Mensch; Jordan Hoffmann; Trevor Cai; Eliza Rutherford; Katie Millican; George van den Driessche; Jean-Baptiste Lespiau; Bogdan Damoc; Aidan Clark; Diego de Las Casas; Aurelia Guy; Jacob Menick; Roman Ring; Tom Hennigan; Saffron Huang; Loren Maggiore; Chris Jones; Albin Cassirer; Andy Brock; Michela Paganini; Geoffrey Irving; Oriol Vinyals; Simon Osindero; Karen Simonyan; Jack W. Rae; Erich Elsen; Laurent Sifre\n\nSource-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.\n\n\n\n## Original sources\n\n- [Improving language models by retrieving from trillions of tokens](https://arxiv.org/abs/2112.04426v3)\n", "summary": "Conditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores.", "title": "Borgeaud and colleagues: RETRO chunked retrieval-enhanced language modeling" }, "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "source_ref": { "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "content_hash": "sha256:707a39ba4c7d2a04c6777d8b4a9e447b9a91c7d1266963f793bc43f614876931" }, "export_view": "wrs-public-editorial-view-1", "public_view_hash": "sha256:27d94387939d133e490500550b47dfdb836f1652bb712cf7b4e23c58202f0227" }, "target": { "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "content_hash": "sha256:707a39ba4c7d2a04c6777d8b4a9e447b9a91c7d1266963f793bc43f614876931" }, "published_grade": { "profile": "wrs-published-grade-1", "edition": "2026-10-28.1", "id": "borgeaud-et-al-retro-2021", "entry_id": "borgeaud-et-al-retro-2021", "revision": 1, "title": "Borgeaud and colleagues: RETRO chunked retrieval-enhanced language modeling", "context": "Retrieval-enhanced language models", "track": "Human-designed research method or resource", "band": "A", "modifier": "-", "status": "PROVISIONAL_EDITORIAL", "assessed_on": "2026-10-16", "evidence_cutoff": "2026-10-16", "achievement_state": "SUPPORTED_SCOPED_RESULT", "what_changed": "Conditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores.", "scope": "Conditioned generation on retrieved text chunks using a frozen retriever and trainable encoding and cross-attention, testing large external text stores. Retrieval-augmented generation is inherited; the contribution is the chunked architecture and scale study. Database storage, retrieval costs and data overlap remain relevant.", "evidence_summary": "The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Retrieval-augmented generation is inherited; the contribution is the chunked architecture and scale study. Database storage, retrieval costs and data overlap remain relevant.", "references": [ { "title": "Improving language models by retrieving from trillions of tokens", "url": "https://arxiv.org/abs/2112.04426v3", "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification." } ], "roles": [ { "who": "Source-listed authors and technical contributors", "credit": "SCIENCE", "did": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities." } ], "target": { "record_id": "urn:uuid:bca5746a-9a2a-4206-a577-d4a1d5af5b10", "version": 1, "content_hash": "sha256:707a39ba4c7d2a04c6777d8b4a9e447b9a91c7d1266963f793bc43f614876931" }, "scientific_correctness_certified": false, "independent_expert_calibration": false, "method": "PROPRIETARY", "points": "95000" }, "canonical_agent_submission": false, "scientific_verification_awarded": false, "reader_guide": null, "reading_status": "SOURCE_LINKED_NOTE" }