{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "raffel-et-al-t5-2019",
    "title": "Raffel and colleagues: unified text-to-text transfer learning (T5)",
    "summary": "Cast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family.",
    "boundary": "The paper's benchmark results depend on its data, tasks and training settings; it does not originate all transformer transfer learning or establish a universal best objective.",
    "area": "Language-model transfer learning",
    "topic": "Artificial intelligence",
    "year": 2019,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "raffel-et-al-t5-2019",
    "target": {
      "record_id": "urn:uuid:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
      "version": 1,
      "content_hash": "sha256:36cad3880ded5a7cb29ef97a5dc1a4b3f28c67f51b9e12ca7320bf4eaa7b44a5"
    },
    "humans": [
      "Colin Raffel",
      "Noam Shazeer",
      "Adam Roberts",
      "Katherine Lee",
      "Sharan Narang",
      "Michael Matena",
      "Yanqi Zhou",
      "Wei Li",
      "Peter J. Liu"
    ],
    "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-t5-1910-10683",
        "title": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
        "url": "https://arxiv.org/abs/1910.10683v4"
      },
      {
        "id": "google-t5-2020",
        "title": "Raffel and colleagues: unified text-to-text transfer learning (T5) — company research page",
        "url": "https://research.google/blog/exploring-transfer-learning-with-t5-the-text-to-text-transfer-transformer/"
      }
    ],
    "points": "65000",
    "published_grade": {
      "id": "raffel-et-al-t5-2019",
      "entry_id": "raffel-et-al-t5-2019",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
        "version": 1,
        "content_hash": "sha256:36cad3880ded5a7cb29ef97a5dc1a4b3f28c67f51b9e12ca7320bf4eaa7b44a5"
      },
      "what_changed": "Cast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family.",
      "scope": "Cast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family. Scope limitation: The paper's benchmark results depend on its data, tasks and training settings; it does not originate all transformer transfer learning or establish a universal best objective.",
      "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:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
        "version": 1,
        "content_hash": "sha256:36cad3880ded5a7cb29ef97a5dc1a4b3f28c67f51b9e12ca7320bf4eaa7b44a5"
      },
      "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": "Colin Raffel",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Noam Shazeer",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Adam Roberts",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Katherine Lee",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Sharan Narang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Michael Matena",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yanqi Zhou",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Wei Li",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Peter J. Liu",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
            "url": "https://arxiv.org/abs/1910.10683v4"
          }
        },
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-1",
          "kind": "source_reference",
          "source": {
            "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — company research page",
            "url": "https://research.google/blog/exploring-transfer-learning-with-t5-the-text-to-text-transfer-transformer/"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "The paper's benchmark results depend on its data, tasks and training settings; it does not originate all transformer transfer learning or establish a universal best objective.",
        "domain": "Language-model transfer learning"
      },
      "sources": [
        {
          "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
          "url": "https://arxiv.org/abs/1910.10683v4"
        },
        {
          "citation": "Raffel and colleagues: unified text-to-text transfer learning (T5) — company research page",
          "url": "https://research.google/blog/exploring-transfer-learning-with-t5-the-text-to-text-transfer-transformer/"
        }
      ],
      "statement": "# Raffel and colleagues: unified text-to-text transfer learning (T5)\n\nCast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family.\n\n## Scientific scope\n\nThe paper's benchmark results depend on its data, tasks and training settings; it does not originate all transformer transfer learning or establish a universal best objective.\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\nColin Raffel; Noam Shazeer; Adam Roberts; Katherine Lee; Sharan Narang; Michael Matena; Yanqi Zhou; Wei Li; Peter J. Liu\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- [Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper](https://arxiv.org/abs/1910.10683v4)\n- [Raffel and colleagues: unified text-to-text transfer learning (T5) — company research page](https://research.google/blog/exploring-transfer-learning-with-t5-the-text-to-text-transfer-transformer/)\n",
      "summary": "Cast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family.",
      "title": "Raffel and colleagues: unified text-to-text transfer learning (T5)"
    },
    "record_id": "urn:uuid:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
      "version": 1,
      "content_hash": "sha256:36cad3880ded5a7cb29ef97a5dc1a4b3f28c67f51b9e12ca7320bf4eaa7b44a5"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:75e08931440435b21701d9922f16b7d82cdbbccae6def9da344b2cb0c7eadb32"
  },
  "target": {
    "record_id": "urn:uuid:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
    "version": 1,
    "content_hash": "sha256:36cad3880ded5a7cb29ef97a5dc1a4b3f28c67f51b9e12ca7320bf4eaa7b44a5"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "raffel-et-al-t5-2019",
    "entry_id": "raffel-et-al-t5-2019",
    "revision": 1,
    "title": "Raffel and colleagues: unified text-to-text transfer learning (T5)",
    "context": "Language-model transfer 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": "Cast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family.",
    "scope": "Cast diverse natural-language tasks into a unified text-to-text format, systematically compared transfer-learning choices, and introduced the C4 corpus and T5 model family. Scope limitation: The paper's benchmark results depend on its data, tasks and training settings; it does not originate all transformer transfer learning or establish a universal best objective.",
    "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. The paper's benchmark results depend on its data, tasks and training settings; it does not originate all transformer transfer learning or establish a universal best objective.",
    "references": [
      {
        "title": "Raffel and colleagues: unified text-to-text transfer learning (T5) — original paper",
        "url": "https://arxiv.org/abs/1910.10683v4",
        "locator": "Abstract and complete bibliographic byline for 1910.10683v4; not an independent proof or experimental replication."
      },
      {
        "title": "Raffel and colleagues: unified text-to-text transfer learning (T5) — company research page",
        "url": "https://research.google/blog/exploring-transfer-learning-with-t5-the-text-to-text-transfer-transformer/",
        "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:ea79f9ab-cfed-4a0c-ac32-0626b5c01f67",
      "version": 1,
      "content_hash": "sha256:36cad3880ded5a7cb29ef97a5dc1a4b3f28c67f51b9e12ca7320bf4eaa7b44a5"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "65000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}