{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "openai-berner-et-al-openai-five-2019",
    "title": "OpenAI Five: scaled self-play for team Dota2",
    "summary": "Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting.",
    "boundary": "Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.",
    "area": "Large-scale reinforcement-learning systems",
    "topic": "Artificial intelligence",
    "year": 2019,
    "lane": "HUMAN",
    "result": "EMPIRICAL",
    "package_id": "openai-berner-et-al-openai-five-2019",
    "target": {
      "record_id": "urn:uuid:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
      "version": 1,
      "content_hash": "sha256:6a08c993156e0e2855626319c08e2f357dd2a0f0931cb96f9b15f35e1986ad3e"
    },
    "humans": [
      "Christopher Berner",
      "Greg Brockman",
      "Brooke Chan",
      "Vicki Cheung",
      "Przemysław Dębiak",
      "Christy Dennison",
      "David Farhi",
      "Quirin Fischer",
      "Shariq Hashme",
      "Chris Hesse",
      "Rafal Józefowicz",
      "Scott Gray",
      "Catherine Olsson",
      "Jakub Pachocki",
      "Michael Petrov",
      "Henrique Pondé de Oliveira Pinto",
      "Jonathan Raiman",
      "Tim Salimans",
      "Jeremy Schlatter",
      "Jonas Schneider",
      "Szymon Sidor",
      "Ilya Sutskever",
      "Jie Tang",
      "Filip Wolski",
      "Susan Zhang"
    ],
    "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-openai-dota2-paper2019",
        "title": "Dota2 with Large Scale Deep Reinforcement Learning",
        "url": "https://cdn.openai.com/dota-2.pdf"
      }
    ],
    "points": "185000",
    "published_grade": {
      "id": "openai-berner-et-al-openai-five-2019",
      "entry_id": "openai-berner-et-al-openai-five-2019",
      "band": "A",
      "modifier": "+",
      "target": {
        "record_id": "urn:uuid:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
        "version": 1,
        "content_hash": "sha256:6a08c993156e0e2855626319c08e2f357dd2a0f0931cb96f9b15f35e1986ad3e"
      },
      "what_changed": "Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting.",
      "scope": "Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting. Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.",
      "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:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
        "version": 1,
        "content_hash": "sha256:6a08c993156e0e2855626319c08e2f357dd2a0f0931cb96f9b15f35e1986ad3e"
      },
      "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": "Christopher Berner",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Greg Brockman",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Brooke Chan",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Vicki Cheung",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Przemysław Dębiak",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Christy Dennison",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "David Farhi",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Quirin Fischer",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Shariq Hashme",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Chris Hesse",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Rafal Józefowicz",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Scott Gray",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Catherine Olsson",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jakub Pachocki",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Michael Petrov",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Henrique Pondé de Oliveira Pinto",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jonathan Raiman",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Tim Salimans",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jeremy Schlatter",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jonas Schneider",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Szymon Sidor",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Ilya Sutskever",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jie Tang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Filip Wolski",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Susan Zhang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "organization",
          "name": "OpenAI",
          "note": "Cited historical collaboration, not a platform organization account; individual attribution remains incomplete.",
          "roles": [
            "discoverer"
          ],
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
            "url": "https://cdn.openai.com/dota-2.pdf"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.",
        "domain": "Large-scale reinforcement-learning systems"
      },
      "sources": [
        {
          "citation": "Dota2 with Large Scale Deep Reinforcement Learning",
          "url": "https://cdn.openai.com/dota-2.pdf"
        }
      ],
      "statement": "# OpenAI Five: scaled self-play for team Dota2\n\nScaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting.\n\n## Scientific scope\n\nUses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.\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\nChristopher Berner; Greg Brockman; Brooke Chan; Vicki Cheung; Przemysław Dębiak; Christy Dennison; David Farhi; Quirin Fischer; Shariq Hashme; Chris Hesse; Rafal Józefowicz; Scott Gray; Catherine Olsson; Jakub Pachocki; Michael Petrov; Henrique Pondé de Oliveira Pinto; Jonathan Raiman; Tim Salimans; Jeremy Schlatter; Jonas Schneider; Szymon Sidor; Ilya Sutskever; Jie Tang; Filip Wolski; Susan Zhang\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- [Dota2 with Large Scale Deep Reinforcement Learning](https://cdn.openai.com/dota-2.pdf)\n",
      "summary": "Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting.",
      "title": "OpenAI Five: scaled self-play for team Dota2"
    },
    "record_id": "urn:uuid:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
      "version": 1,
      "content_hash": "sha256:6a08c993156e0e2855626319c08e2f357dd2a0f0931cb96f9b15f35e1986ad3e"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:e9ee914a02b85bdab5aaf7438928dd5614cdff3a717b6ebb1d23e4b08f9efb11"
  },
  "target": {
    "record_id": "urn:uuid:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
    "version": 1,
    "content_hash": "sha256:6a08c993156e0e2855626319c08e2f357dd2a0f0931cb96f9b15f35e1986ad3e"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "openai-berner-et-al-openai-five-2019",
    "entry_id": "openai-berner-et-al-openai-five-2019",
    "revision": 1,
    "title": "OpenAI Five: scaled self-play for team Dota2",
    "context": "Large-scale reinforcement-learning systems",
    "track": "Human-designed research method or resource",
    "band": "A",
    "modifier": "+",
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-16",
    "evidence_cutoff": "2026-10-16",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting.",
    "scope": "Scaled recurrent PPO self-play with distributed rollout/inference/optimization and continual-training infrastructure, producing a team agent that defeated Team OG in the documented2019 Dota2 setting. Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.",
    "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. Uses inherited PPO/GAE/LSTM/Adam and shaped rewards. Evaluation supports17heroes, excludes specified multi-unit items, uses semantic engine observations rather than pixels, and retains scripted item/ability/courier logic. Defeating the team in that setting is not solving unrestricted Dota2 or certifying unexploitable equilibrium.",
    "references": [
      {
        "title": "Dota2 with Large Scale Deep Reinforcement Learning",
        "url": "https://cdn.openai.com/dota-2.pdf",
        "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:59426c63-056f-4d0e-a99d-2ea8c55e1a2c",
      "version": 1,
      "content_hash": "sha256:6a08c993156e0e2855626319c08e2f357dd2a0f0931cb96f9b15f35e1986ad3e"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "185000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}