{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "transformer-architecture",
    "title": "The Transformer: attention-based sequence modelling",
    "summary": "Introduced an attention-based encoder–decoder architecture for sequence transduction, with multi-head attention and positional information, and reported strong machine-translation and parsing results.",
    "boundary": "This record credits the human-designed method. The eight authors explicitly state equal contribution and random listing order; order is not a merit score. The paper does not establish general intelligence or correctness of later language-model outputs. Its 2017 version and evaluations are distinct from subsequent Transformer systems.",
    "area": "Machine learning",
    "topic": "Computing & AI",
    "year": 2017,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "transformer-architecture",
    "target": {
      "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
      "version": 2,
      "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
    },
    "humans": [
      "Ashish Vaswani",
      "Noam Shazeer",
      "Niki Parmar",
      "Jakob Uszkoreit",
      "Llion Jones",
      "Aidan N. Gomez",
      "Łukasz Kaiser",
      "Illia Polosukhin"
    ],
    "authors_complete": true,
    "human_role_note": "Listed source contributors; individual labor shares are not measured",
    "agents": [],
    "related": [
      "lstm-architecture",
      "backprop-representations"
    ],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "transformer-2017",
        "title": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
        "url": "https://arxiv.org/abs/1706.03762v5"
      }
    ],
    "geography": null,
    "points": "250000",
    "published_grade": {
      "id": "transformer-architecture",
      "entry_id": "transformer-architecture",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
        "version": 2,
        "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
      },
      "what_changed": "Sequence transduction gained an attention-based architecture with greater parallelism than recurrent designs.",
      "scope": "The human-designed attention-based encoder–decoder and the experiments in version 5 of the 2017 paper.",
      "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:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
        "version": 2,
        "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": {
      "content_hash": "sha256:f1c6327fdc1bc8dffe69a77f1577b969649c2a9d13139b6fb9445e190487fbd2",
      "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
      "version": 1
    },
    "published_at": "2026-09-12T00:00:00Z",
    "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": "Ashish Vaswani",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "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": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Niki Parmar",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jakob Uszkoreit",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Llion Jones",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Aidan N. Gomez",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Łukasz Kaiser",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Illia Polosukhin",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
            "url": "https://arxiv.org/abs/1706.03762v5"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "This record credits the human-designed method. The eight authors explicitly state equal contribution and random listing order; order is not a merit score. The paper does not establish general intelligence or correctness of later language-model outputs. Its 2017 version and evaluations are distinct from subsequent Transformer systems.",
        "domain": "Machine learning"
      },
      "sources": [
        {
          "citation": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
          "url": "https://arxiv.org/abs/1706.03762v5"
        }
      ],
      "statement": "# The Transformer: attention-based sequence modelling\n\n## The question\n\nHow can a model use relationships between parts of a sequence without processing everything through a recurrent chain? The 2017 Transformer paper proposes an attention-based encoder–decoder architecture and evaluates it on specified translation tasks. This record concerns an architectural and experimental contribution by its human authors. It is not an AI-origin discovery just because the architecture later became central to AI systems.\n\n## The idea\n\nAttention lets a representation combine information from other positions according to learned relationships. Multiple attention heads can provide different representation paths, while positional information and other components remain important. The original architecture removes recurrence and convolutions from the central sequence-transduction design. Its paper reports translation quality and training conditions; those measurements describe the tested setup rather than every future language model or every possible attention implementation.\n\n## Why it matters\n\nThe contribution offers a reusable way to organize sequence modeling and make training more parallelizable. WRS's A grade refers to the recorded architecture contribution. Later large language models add their own data, objectives, scales and engineering. Those developments do not turn the original paper into a report of modern general intelligence, and a familiar family name is not sufficient provenance for a specific model's behavior.\n\n## A teaching example\n\nConsider the sentence: the trophy did not fit in the suitcase because it was too large. To interpret it, a reader must connect words that may be separated in the sequence. This is a useful intuition for relationships between positions. A trained attention weight is not automatically a trustworthy explanation of an answer, however. The example is illustrative and does not reproduce a translation benchmark. To test a model, choose defined inputs, reference outputs and a metric before inspecting which patterns appear convincing.\n\n## Scope and a useful next step\n\nThe selected arXiv version is v5, dated December 2017; newer archive versions are not silently substituted. Begin with the architecture section and the experimental setup. A bounded contribution could map one component to its exact equation and show a small shape-checked implementation. Keep that exercise separate from retraining the reported system. Record dimensions, masks and test cases, since a visually plausible attention diagram can still hide a mathematical or implementation mismatch.\n\n## Scientific scope\n\nThis record credits the human-designed method. The eight authors explicitly state equal contribution and random listing order; order is not a merit score. The paper does not establish general intelligence or correctness of later language-model outputs. Its 2017 version and evaluations are distinct from subsequent Transformer systems.\n\n## Sources for this explanation\n\n- [Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)](https://arxiv.org/abs/1706.03762v5): Identified original contribution source. This explanatory guide is not an independent proof or experimental replication.\n",
      "summary": "Introduced an attention-based encoder–decoder architecture for sequence transduction, with multi-head attention and positional information, and reported strong machine-translation and parsing results.",
      "title": "The Transformer: attention-based sequence modelling"
    },
    "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
    "version": 2,
    "source_ref": {
      "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
      "version": 2,
      "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:859e69d25ebc0d4d6aca17bfb2611bdcbe598d0b085199a9280a13271979cbc6"
  },
  "target": {
    "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
    "version": 2,
    "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "transformer-architecture",
    "entry_id": "transformer-architecture",
    "revision": 1,
    "title": "The Transformer: attention-based sequence modelling",
    "context": "Neural sequence-model architecture",
    "track": "Human-designed AI methods",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-09-13",
    "evidence_cutoff": "2026-09-13",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Sequence transduction gained an attention-based architecture with greater parallelism than recurrent designs.",
    "scope": "The human-designed attention-based encoder–decoder and the experiments in version 5 of the 2017 paper.",
    "evidence_summary": "The cited historical record or published author account supports this scoped contribution. The source-reading boundary is stated at each link; a WRS grade is not an independent scientific validation.",
    "references": [
      {
        "title": "Attention Is All You Need, v5",
        "url": "https://arxiv.org/abs/1706.03762v5",
        "locator": "Version-specific abstract, byline and task evaluations; not a new training replay"
      }
    ],
    "roles": [
      {
        "who": "The eight named paper authors",
        "credit": "SCIENCE",
        "did": "Architecture and experiments; equal contribution stated by the paper, full byline retained in the record."
      }
    ],
    "target": {
      "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
      "version": 2,
      "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "250000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": {
    "id": "transformer-architecture",
    "sections": [
      {
        "title": "The question",
        "text": "How can a model use relationships between parts of a sequence without processing everything through a recurrent chain? The 2017 Transformer paper proposes an attention-based encoder–decoder architecture and evaluates it on specified translation tasks. This record concerns an architectural and experimental contribution by its human authors. It is not an AI-origin discovery just because the architecture later became central to AI systems."
      },
      {
        "title": "The idea",
        "text": "Attention lets a representation combine information from other positions according to learned relationships. Multiple attention heads can provide different representation paths, while positional information and other components remain important. The original architecture removes recurrence and convolutions from the central sequence-transduction design. Its paper reports translation quality and training conditions; those measurements describe the tested setup rather than every future language model or every possible attention implementation."
      },
      {
        "title": "Why it matters",
        "text": "The contribution offers a reusable way to organize sequence modeling and make training more parallelizable. WRS's A grade refers to the recorded architecture contribution. Later large language models add their own data, objectives, scales and engineering. Those developments do not turn the original paper into a report of modern general intelligence, and a familiar family name is not sufficient provenance for a specific model's behavior."
      },
      {
        "title": "A teaching example",
        "text": "Consider the sentence: the trophy did not fit in the suitcase because it was too large. To interpret it, a reader must connect words that may be separated in the sequence. This is a useful intuition for relationships between positions. A trained attention weight is not automatically a trustworthy explanation of an answer, however. The example is illustrative and does not reproduce a translation benchmark. To test a model, choose defined inputs, reference outputs and a metric before inspecting which patterns appear convincing."
      },
      {
        "title": "Scope and a useful next step",
        "text": "The selected arXiv version is v5, dated December 2017; newer archive versions are not silently substituted. Begin with the architecture section and the experimental setup. A bounded contribution could map one component to its exact equation and show a small shape-checked implementation. Keep that exercise separate from retraining the reported system. Record dimensions, masks and test cases, since a visually plausible attention diagram can still hide a mathematical or implementation mismatch."
      }
    ],
    "target": {
      "record_id": "urn:uuid:6efcc53a-1ab6-4be3-a83a-f0774ea63bcc",
      "version": 2,
      "content_hash": "sha256:e154639b3bf4fbc96c9a05412dfaee3177583b3424a487ea845b128c3689d412"
    },
    "references": [
      {
        "title": "Vaswani et al., Attention Is All You Need, NeurIPS 2017; arXiv:1706.03762v5 (6 December 2017)",
        "url": "https://arxiv.org/abs/1706.03762v5",
        "locator": "Identified original contribution source. This explanatory guide is not an independent proof or experimental replication."
      }
    ]
  },
  "reading_status": "EXPLANATORY_GUIDE"
}