{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "guu-et-al-realm-2020",
    "title": "Guu and colleagues: REALM retrieval-augmented pre-training",
    "summary": "Pre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal.",
    "boundary": "REALM is retrieval-augmented pre-training; distinct from the later RAG generation paper already in the formal catalogue.",
    "area": "Retrieval-augmented language learning",
    "topic": "Artificial intelligence",
    "year": 2020,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "guu-et-al-realm-2020",
    "target": {
      "record_id": "urn:uuid:94be414a-6903-4b01-ae9d-0316deebcc2d",
      "version": 1,
      "content_hash": "sha256:b56ad84aa214ee5fb279c161104dfa21e8cc179b87114168c874ce3f5a215a86"
    },
    "humans": [
      "Kelvin Guu",
      "Kenton Lee",
      "Zora Tung",
      "Panupong Pasupat",
      "Ming-Wei Chang"
    ],
    "authors_complete": true,
    "human_role_note": "Complete named paper byline. Human research, source-reported system execution and experimental validation are distinct; individual effort shares are not measured.",
    "agents": [],
    "related": [],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "arxiv-2002-08909",
        "title": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
        "url": "https://arxiv.org/abs/2002.08909v1"
      },
      {
        "id": "company-guu-et-al-realm-2020",
        "title": "Guu and colleagues: REALM retrieval-augmented pre-training — official company source",
        "url": "https://research.google/pubs/retrieval-augmented-language-model-pre-training/"
      }
    ],
    "points": "32000",
    "published_grade": {
      "id": "guu-et-al-realm-2020",
      "entry_id": "guu-et-al-realm-2020",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:94be414a-6903-4b01-ae9d-0316deebcc2d",
        "version": 1,
        "content_hash": "sha256:b56ad84aa214ee5fb279c161104dfa21e8cc179b87114168c874ce3f5a215a86"
      },
      "what_changed": "Pre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal.",
      "scope": "Pre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal. Scope limitation: REALM is retrieval-augmented pre-training; distinct from the later RAG generation paper already in the formal catalogue.",
      "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:94be414a-6903-4b01-ae9d-0316deebcc2d",
        "version": 1,
        "content_hash": "sha256:b56ad84aa214ee5fb279c161104dfa21e8cc179b87114168c874ce3f5a215a86"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-09-27T00:00:00Z",
    "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": "Kelvin Guu",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
            "url": "https://arxiv.org/abs/2002.08909v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Kenton Lee",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
            "url": "https://arxiv.org/abs/2002.08909v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Zora Tung",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
            "url": "https://arxiv.org/abs/2002.08909v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Panupong Pasupat",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
            "url": "https://arxiv.org/abs/2002.08909v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Ming-Wei Chang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
            "url": "https://arxiv.org/abs/2002.08909v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
            "url": "https://arxiv.org/abs/2002.08909v1"
          }
        },
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-1",
          "kind": "source_reference",
          "source": {
            "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — official company source",
            "url": "https://research.google/pubs/retrieval-augmented-language-model-pre-training/"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "REALM is retrieval-augmented pre-training; distinct from the later RAG generation paper already in the formal catalogue.",
        "domain": "Retrieval-augmented language learning"
      },
      "sources": [
        {
          "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
          "url": "https://arxiv.org/abs/2002.08909v1"
        },
        {
          "citation": "Guu and colleagues: REALM retrieval-augmented pre-training — official company source",
          "url": "https://research.google/pubs/retrieval-augmented-language-model-pre-training/"
        }
      ],
      "statement": "# Guu and colleagues: REALM retrieval-augmented pre-training\n\nPre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal.\n\n## Scientific scope\n\nREALM is retrieval-augmented pre-training; distinct from the later RAG generation paper already in the formal catalogue.\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\nKelvin Guu; Kenton Lee; Zora Tung; Panupong Pasupat; Ming-Wei Chang\n\nComplete named paper byline. Human research, source-reported system execution and experimental validation are distinct; individual effort shares are not measured.\n\n\n\n## Original sources\n\n- [Guu and colleagues: REALM retrieval-augmented pre-training — original paper](https://arxiv.org/abs/2002.08909v1)\n- [Guu and colleagues: REALM retrieval-augmented pre-training — official company source](https://research.google/pubs/retrieval-augmented-language-model-pre-training/)\n",
      "summary": "Pre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal.",
      "title": "Guu and colleagues: REALM retrieval-augmented pre-training"
    },
    "record_id": "urn:uuid:94be414a-6903-4b01-ae9d-0316deebcc2d",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:94be414a-6903-4b01-ae9d-0316deebcc2d",
      "version": 1,
      "content_hash": "sha256:b56ad84aa214ee5fb279c161104dfa21e8cc179b87114168c874ce3f5a215a86"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:b89f03147d3bbf260ed9ba1b01d262db832b442821c3ebe886d4577aabadb8d8"
  },
  "target": {
    "record_id": "urn:uuid:94be414a-6903-4b01-ae9d-0316deebcc2d",
    "version": 1,
    "content_hash": "sha256:b56ad84aa214ee5fb279c161104dfa21e8cc179b87114168c874ce3f5a215a86"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "guu-et-al-realm-2020",
    "entry_id": "guu-et-al-realm-2020",
    "revision": 1,
    "title": "Guu and colleagues: REALM retrieval-augmented pre-training",
    "context": "Retrieval-augmented language learning",
    "track": "Human-designed AI research method or resource",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-09-27",
    "evidence_cutoff": "2026-09-27",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Pre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal.",
    "scope": "Pre-trained a language model with a latent document retriever whose retrieval decisions receive a masked-language-model learning signal. Scope limitation: REALM is retrieval-augmented pre-training; distinct from the later RAG generation paper already in the formal catalogue.",
    "evidence_summary": "The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. REALM is retrieval-augmented pre-training; distinct from the later RAG generation paper already in the formal catalogue.",
    "references": [
      {
        "title": "Guu and colleagues: REALM retrieval-augmented pre-training — original paper",
        "url": "https://arxiv.org/abs/2002.08909v1",
        "locator": "Abstract and complete bibliographic byline for 2002.08909v1; not an independent proof or experimental replication."
      },
      {
        "title": "Guu and colleagues: REALM retrieval-augmented pre-training — official company source",
        "url": "https://research.google/pubs/retrieval-augmented-language-model-pre-training/",
        "locator": "Source-linked contribution and stated scientific limits; not an independent replication."
      }
    ],
    "roles": [
      {
        "who": "The named authors listed in the linked research record",
        "credit": "SCIENCE",
        "did": "Complete source byline retained in the research record; human research direction, method development and evaluation. Individual labor shares are not measured."
      }
    ],
    "target": {
      "record_id": "urn:uuid:94be414a-6903-4b01-ae9d-0316deebcc2d",
      "version": 1,
      "content_hash": "sha256:b56ad84aa214ee5fb279c161104dfa21e8cc179b87114168c874ce3f5a215a86"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "32000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}