{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "ofa-progressive-shrinking-2019",
    "title": "Once-for-All progressive shrinking for deployable subnetworks",
    "summary": "Trains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch.",
    "boundary": "The O(1) training claim is relative to deployment-scenario count after one expensive supported-space training; per-device search/evaluation and hardware measurements still cost resources. Accuracy equivalence to independent training is empirical, not a theorem over all 10^19 subnetworks.",
    "area": "Hardware-specialized model deployment",
    "topic": "Artificial intelligence",
    "year": 2019,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "ofa-progressive-shrinking-2019",
    "target": {
      "record_id": "urn:uuid:83d61579-947c-48a5-a250-a0a94eac8d43",
      "version": 1,
      "content_hash": "sha256:94e22f994891ffc6b5c540b1d7030eb0d25c48b8e4f090a30dbec596c4ef0675"
    },
    "humans": [
      "Han Cai",
      "Chuang Gan",
      "Song Han"
    ],
    "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": "ai-oct15-arxiv-1908-09791v1",
        "title": "Once for All: Train One Network and Specialize it for Efficient Deployment",
        "url": "https://arxiv.org/pdf/1908.09791v1"
      }
    ],
    "points": "145000",
    "published_grade": {
      "id": "ofa-progressive-shrinking-2019",
      "entry_id": "ofa-progressive-shrinking-2019",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:83d61579-947c-48a5-a250-a0a94eac8d43",
        "version": 1,
        "content_hash": "sha256:94e22f994891ffc6b5c540b1d7030eb0d25c48b8e4f090a30dbec596c4ef0675"
      },
      "what_changed": "Trains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch.",
      "scope": "Trains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch. The O(1) training claim is relative to deployment-scenario count after one expensive supported-space training; per-device search/evaluation and hardware measurements still cost resources. Accuracy equivalence to independent training is empirical, not a theorem over all 10^19 subnetworks.",
      "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:83d61579-947c-48a5-a250-a0a94eac8d43",
        "version": 1,
        "content_hash": "sha256:94e22f994891ffc6b5c540b1d7030eb0d25c48b8e4f090a30dbec596c4ef0675"
      },
      "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": "Han Cai",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Once for All: Train One Network and Specialize it for Efficient Deployment",
            "url": "https://arxiv.org/pdf/1908.09791v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Chuang Gan",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Once for All: Train One Network and Specialize it for Efficient Deployment",
            "url": "https://arxiv.org/pdf/1908.09791v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Song Han",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "Once for All: Train One Network and Specialize it for Efficient Deployment",
            "url": "https://arxiv.org/pdf/1908.09791v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Once for All: Train One Network and Specialize it for Efficient Deployment",
            "url": "https://arxiv.org/pdf/1908.09791v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "The O(1) training claim is relative to deployment-scenario count after one expensive supported-space training; per-device search/evaluation and hardware measurements still cost resources. Accuracy equivalence to independent training is empirical, not a theorem over all 10^19 subnetworks.",
        "domain": "Hardware-specialized model deployment"
      },
      "sources": [
        {
          "citation": "Once for All: Train One Network and Specialize it for Efficient Deployment",
          "url": "https://arxiv.org/pdf/1908.09791v1"
        }
      ],
      "statement": "# Once-for-All progressive shrinking for deployable subnetworks\n\nTrains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch.\n\n## Scientific scope\n\nThe O(1) training claim is relative to deployment-scenario count after one expensive supported-space training; per-device search/evaluation and hardware measurements still cost resources. Accuracy equivalence to independent training is empirical, not a theorem over all 10^19 subnetworks.\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\nHan Cai; Chuang Gan; Song Han\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- [Once for All: Train One Network and Specialize it for Efficient Deployment](https://arxiv.org/pdf/1908.09791v1)\n",
      "summary": "Trains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch.",
      "title": "Once-for-All progressive shrinking for deployable subnetworks"
    },
    "record_id": "urn:uuid:83d61579-947c-48a5-a250-a0a94eac8d43",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:83d61579-947c-48a5-a250-a0a94eac8d43",
      "version": 1,
      "content_hash": "sha256:94e22f994891ffc6b5c540b1d7030eb0d25c48b8e4f090a30dbec596c4ef0675"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:23e4fb36216b38cf231640662f7e6a95ec710bc3623e70ed305134827652f55b"
  },
  "target": {
    "record_id": "urn:uuid:83d61579-947c-48a5-a250-a0a94eac8d43",
    "version": 1,
    "content_hash": "sha256:94e22f994891ffc6b5c540b1d7030eb0d25c48b8e4f090a30dbec596c4ef0675"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "ofa-progressive-shrinking-2019",
    "entry_id": "ofa-progressive-shrinking-2019",
    "revision": 1,
    "title": "Once-for-All progressive shrinking for deployable subnetworks",
    "context": "Hardware-specialized model deployment",
    "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": "Trains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch.",
    "scope": "Trains a shared-weight network progressively over kernel size, depth, width and resolution, then selects subnetworks for multiple deployment constraints without separately training each from scratch. The O(1) training claim is relative to deployment-scenario count after one expensive supported-space training; per-device search/evaluation and hardware measurements still cost resources. Accuracy equivalence to independent training is empirical, not a theorem over all 10^19 subnetworks.",
    "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. The O(1) training claim is relative to deployment-scenario count after one expensive supported-space training; per-device search/evaluation and hardware measurements still cost resources. Accuracy equivalence to independent training is empirical, not a theorem over all 10^19 subnetworks.",
    "references": [
      {
        "title": "Once for All: Train One Network and Specialize it for Efficient Deployment",
        "url": "https://arxiv.org/pdf/1908.09791v1",
        "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:83d61579-947c-48a5-a250-a0a94eac8d43",
      "version": 1,
      "content_hash": "sha256:94e22f994891ffc6b5c540b1d7030eb0d25c48b8e4f090a30dbec596c4ef0675"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "145000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}