{ "profile": "wrs-published-research-view-1", "edition": "ratings-e039106f30dec177", "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS", "entry": { "id": "xin-et-al-deepseek-prover-synthesis-rmaxts-family-2024", "title": "DeepSeek-Prover: synthetic proof data and feedback-driven RMaxTS", "summary": "V1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion.", "boundary": "One two-source method family distinct from already formal ProverV2 recursive decomposition. Inherits DeepSeekMath, Lean and expert iteration. Source-specific initials retained; autoformalization fidelity is not implied by valid proofs. Sampling and search budgets are part of results.", "area": "Formal-proof training and search", "topic": "Artificial intelligence", "year": 2024, "lane": "HUMAN", "result": "ALGORITHM", "package_id": "xin-et-al-deepseek-prover-synthesis-rmaxts-family-2024", "target": { "record_id": "urn:uuid:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "content_hash": "sha256:dea80560d0e6a5b68f23e302234fc900a231cd11dadd291e8b4258e483dbf076" }, "humans": [ "Huajian Xin", "Daya Guo", "Zhihong Shao", "Z.Z. Ren", "Qihao Zhu", "Bo Liu", "Chong Ruan", "Wenda Li", "Xiaodan Liang", "Junxiao Song", "Wanjia Zhao", "Haocheng Wang", "Liyue Zhang", "Xuan Lu", "Qiushi Du", "Wenjun Gao", "Dejian Yang", "Zhibin Gou", "Z.F. Wu", "Fuli Luo" ], "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-2405-14333v1", "title": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" }, { "id": "cur20261015-arxiv-2408-08152v1", "title": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } ], "points": "140000", "published_grade": { "id": "xin-et-al-deepseek-prover-synthesis-rmaxts-family-2024", "entry_id": "xin-et-al-deepseek-prover-synthesis-rmaxts-family-2024", "band": "A", "modifier": null, "target": { "record_id": "urn:uuid:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "content_hash": "sha256:dea80560d0e6a5b68f23e302234fc900a231cd11dadd291e8b4258e483dbf076" }, "what_changed": "V1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion.", "scope": "V1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion. One two-source method family distinct from already formal ProverV2 recursive decomposition. Inherits DeepSeekMath, Lean and expert iteration. Source-specific initials retained; autoformalization fidelity is not implied by valid proofs. Sampling and search budgets are part of results.", "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:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "content_hash": "sha256:dea80560d0e6a5b68f23e302234fc900a231cd11dadd291e8b4258e483dbf076" }, "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": "Huajian Xin", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Daya Guo", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Zhihong Shao", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Z.Z. Ren", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Qihao Zhu", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Bo Liu", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Chong Ruan", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Wenda Li", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Xiaodan Liang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Junxiao Song", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Wanjia Zhao", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Haocheng Wang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Liyue Zhang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Xuan Lu", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Qiushi Du", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Wenjun Gao", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Dejian Yang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Zhibin Gou", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Z.F. Wu", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Fuli Luo", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } } ], "evidence": [ { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-0", "kind": "source_reference", "source": { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" } }, { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-1", "kind": "source_reference", "source": { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } } ], "origin": "HUMAN_LEGACY", "record_profile": "wrs-local-research-notes-0.1", "record_type": "research_note", "rrp_version": "0.1", "scope": { "description": "One two-source method family distinct from already formal ProverV2 recursive decomposition. Inherits DeepSeekMath, Lean and expert iteration. Source-specific initials retained; autoformalization fidelity is not implied by valid proofs. Sampling and search budgets are part of results.", "domain": "Formal-proof training and search" }, "sources": [ { "citation": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1" }, { "citation": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1" } ], "statement": "# DeepSeek-Prover: synthetic proof data and feedback-driven RMaxTS\n\nV1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion.\n\n## Scientific scope\n\nOne two-source method family distinct from already formal ProverV2 recursive decomposition. Inherits DeepSeekMath, Lean and expert iteration. Source-specific initials retained; autoformalization fidelity is not implied by valid proofs. Sampling and search budgets are part of results.\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\nHuajian Xin; Daya Guo; Zhihong Shao; Z.Z. Ren; Qihao Zhu; Bo Liu; Chong Ruan; Wenda Li; Xiaodan Liang; Junxiao Song; Wanjia Zhao; Haocheng Wang; Liyue Zhang; Xuan Lu; Qiushi Du; Wenjun Gao; Dejian Yang; Zhibin Gou; Z.F. Wu; Fuli Luo\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- [DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data](https://arxiv.org/abs/2405.14333v1)\n- [DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search](https://arxiv.org/abs/2408.08152v1)\n", "summary": "V1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion.", "title": "DeepSeek-Prover: synthetic proof data and feedback-driven RMaxTS" }, "record_id": "urn:uuid:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "source_ref": { "record_id": "urn:uuid:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "content_hash": "sha256:dea80560d0e6a5b68f23e302234fc900a231cd11dadd291e8b4258e483dbf076" }, "export_view": "wrs-public-editorial-view-1", "public_view_hash": "sha256:5f1ce056776f7f0fa503aff9e96435a20c6e2e9923f31640c185d4e25fa2f49b" }, "target": { "record_id": "urn:uuid:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "content_hash": "sha256:dea80560d0e6a5b68f23e302234fc900a231cd11dadd291e8b4258e483dbf076" }, "published_grade": { "profile": "wrs-published-grade-1", "edition": "2026-10-28.1", "id": "xin-et-al-deepseek-prover-synthesis-rmaxts-family-2024", "entry_id": "xin-et-al-deepseek-prover-synthesis-rmaxts-family-2024", "revision": 1, "title": "DeepSeek-Prover: synthetic proof data and feedback-driven RMaxTS", "context": "Formal-proof training and search", "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": "V1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion.", "scope": "V1 iteratively synthesized formal statements/proofs with filtering and parallel negation checks; V1.5 added proof-assistant-feedback RL and intrinsic-reward MonteCarlo search with truncate-and-resume proof completion. One two-source method family distinct from already formal ProverV2 recursive decomposition. Inherits DeepSeekMath, Lean and expert iteration. Source-specific initials retained; autoformalization fidelity is not implied by valid proofs. Sampling and search budgets are part of results.", "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. One two-source method family distinct from already formal ProverV2 recursive decomposition. Inherits DeepSeekMath, Lean and expert iteration. Source-specific initials retained; autoformalization fidelity is not implied by valid proofs. Sampling and search budgets are part of results.", "references": [ { "title": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "url": "https://arxiv.org/abs/2405.14333v1", "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification." }, { "title": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search", "url": "https://arxiv.org/abs/2408.08152v1", "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:b0b67022-eed4-45a7-a21b-47c900b956d3", "version": 1, "content_hash": "sha256:dea80560d0e6a5b68f23e302234fc900a231cd11dadd291e8b4258e483dbf076" }, "scientific_correctness_certified": false, "independent_expert_calibration": false, "method": "PROPRIETARY", "points": "140000" }, "canonical_agent_submission": false, "scientific_verification_awarded": false, "reader_guide": null, "reading_status": "SOURCE_LINKED_NOTE" }