{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "gehring-et-al-convs2s-2017",
    "title": "Convolutional sequence-to-sequence learning with multi-step attention",
    "summary": "Used residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions.",
    "boundary": "Training parallelism does not remove autoregressive dependencies during generation. GLUs, residual learning, convolutions and attention are inherited components. Benchmarked speed/quality improvements are not a universal advantage over recurrent or Transformer models.",
    "area": "Neural machine translation",
    "topic": "Artificial intelligence",
    "year": 2017,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "gehring-et-al-convs2s-2017",
    "target": {
      "record_id": "urn:uuid:eee71d82-5fc4-49d3-a620-bab22f08f00b",
      "version": 1,
      "content_hash": "sha256:829f645c937ecf7afa5d513613478b153d6bb0d277cf7efdcd71481a92e673fa"
    },
    "humans": [
      "Jonas Gehring",
      "Michael Auli",
      "David Grangier",
      "Denis Yarats",
      "Yann N. Dauphin"
    ],
    "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-1705-03122v1",
        "title": "Convolutional Sequence to Sequence Learning",
        "url": "https://arxiv.org/abs/1705.03122v1"
      }
    ],
    "points": "130000",
    "published_grade": {
      "id": "gehring-et-al-convs2s-2017",
      "entry_id": "gehring-et-al-convs2s-2017",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:eee71d82-5fc4-49d3-a620-bab22f08f00b",
        "version": 1,
        "content_hash": "sha256:829f645c937ecf7afa5d513613478b153d6bb0d277cf7efdcd71481a92e673fa"
      },
      "what_changed": "Used residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions.",
      "scope": "Used residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions. Training parallelism does not remove autoregressive dependencies during generation. GLUs, residual learning, convolutions and attention are inherited components. Benchmarked speed/quality improvements are not a universal advantage over recurrent or Transformer models.",
      "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:eee71d82-5fc4-49d3-a620-bab22f08f00b",
        "version": 1,
        "content_hash": "sha256:829f645c937ecf7afa5d513613478b153d6bb0d277cf7efdcd71481a92e673fa"
      },
      "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": "Jonas Gehring",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Convolutional Sequence to Sequence Learning",
            "url": "https://arxiv.org/abs/1705.03122v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Michael Auli",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Convolutional Sequence to Sequence Learning",
            "url": "https://arxiv.org/abs/1705.03122v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "David Grangier",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Convolutional Sequence to Sequence Learning",
            "url": "https://arxiv.org/abs/1705.03122v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Denis Yarats",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Convolutional Sequence to Sequence Learning",
            "url": "https://arxiv.org/abs/1705.03122v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yann N. Dauphin",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Convolutional Sequence to Sequence Learning",
            "url": "https://arxiv.org/abs/1705.03122v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Convolutional Sequence to Sequence Learning",
            "url": "https://arxiv.org/abs/1705.03122v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "Training parallelism does not remove autoregressive dependencies during generation. GLUs, residual learning, convolutions and attention are inherited components. Benchmarked speed/quality improvements are not a universal advantage over recurrent or Transformer models.",
        "domain": "Neural machine translation"
      },
      "sources": [
        {
          "citation": "Convolutional Sequence to Sequence Learning",
          "url": "https://arxiv.org/abs/1705.03122v1"
        }
      ],
      "statement": "# Convolutional sequence-to-sequence learning with multi-step attention\n\nUsed residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions.\n\n## Scientific scope\n\nTraining parallelism does not remove autoregressive dependencies during generation. GLUs, residual learning, convolutions and attention are inherited components. Benchmarked speed/quality improvements are not a universal advantage over recurrent or Transformer models.\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\nJonas Gehring; Michael Auli; David Grangier; Denis Yarats; Yann N. Dauphin\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- [Convolutional Sequence to Sequence Learning](https://arxiv.org/abs/1705.03122v1)\n",
      "summary": "Used residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions.",
      "title": "Convolutional sequence-to-sequence learning with multi-step attention"
    },
    "record_id": "urn:uuid:eee71d82-5fc4-49d3-a620-bab22f08f00b",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:eee71d82-5fc4-49d3-a620-bab22f08f00b",
      "version": 1,
      "content_hash": "sha256:829f645c937ecf7afa5d513613478b153d6bb0d277cf7efdcd71481a92e673fa"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:77ed17d2fa33c3525a7f80304f4c2ed643abef3012da0c279a8a506533a5f22d"
  },
  "target": {
    "record_id": "urn:uuid:eee71d82-5fc4-49d3-a620-bab22f08f00b",
    "version": 1,
    "content_hash": "sha256:829f645c937ecf7afa5d513613478b153d6bb0d277cf7efdcd71481a92e673fa"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "gehring-et-al-convs2s-2017",
    "entry_id": "gehring-et-al-convs2s-2017",
    "revision": 1,
    "title": "Convolutional sequence-to-sequence learning with multi-step attention",
    "context": "Neural machine translation",
    "track": "Human-designed research method or resource",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-16",
    "evidence_cutoff": "2026-10-16",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Used residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions.",
    "scope": "Used residual gated convolutions for source and target sequence representations, with separate attention at decoder layers and parallel computation across training positions. Training parallelism does not remove autoregressive dependencies during generation. GLUs, residual learning, convolutions and attention are inherited components. Benchmarked speed/quality improvements are not a universal advantage over recurrent or Transformer models.",
    "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. Training parallelism does not remove autoregressive dependencies during generation. GLUs, residual learning, convolutions and attention are inherited components. Benchmarked speed/quality improvements are not a universal advantage over recurrent or Transformer models.",
    "references": [
      {
        "title": "Convolutional Sequence to Sequence Learning",
        "url": "https://arxiv.org/abs/1705.03122v1",
        "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:eee71d82-5fc4-49d3-a620-bab22f08f00b",
      "version": 1,
      "content_hash": "sha256:829f645c937ecf7afa5d513613478b153d6bb0d277cf7efdcd71481a92e673fa"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "130000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}