{ "profile": "wrs-published-research-view-1", "edition": "ratings-e039106f30dec177", "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS", "entry": { "id": "li-et-al-datacomp-language-models-2024", "title": "DataComp-LM: controlled language-model data-curation testbed", "summary": "Fixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models.", "boundary": "The candidate pool is not a claim that every baseline trained on240T tokens. Comparisons are for specified412M–7B scales, compute budgets and evaluation protocols; no universally optimal data filter or corpus-quality theorem. Dataset design and OpenLM/benchmark inheritance remain distinct.", "area": "Language-model dataset evaluation", "topic": "Artificial intelligence", "year": 2024, "lane": "HUMAN", "result": "ALGORITHM", "package_id": "li-et-al-datacomp-language-models-2024", "target": { "record_id": "urn:uuid:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "content_hash": "sha256:ce38a63cac0c2143977f30eb21cf3112e1bb687a41c343eb025d42bb42e6799f" }, "humans": [ "Jeffrey Li", "Alex Fang", "Georgios Smyrnis", "Maor Ivgi", "Matt Jordan", "Samir Gadre", "Hritik Bansal", "Etash Guha", "Sedrick Keh", "Kushal Arora", "Saurabh Garg", "Rui Xin", "Niklas Muenninghoff", "Reinhard Heckel", "Jean Mercat", "Mayee Chen", "Suchin Gururangan", "Mitchell Wortsman", "Alon Albalak", "Yonatan Bitton", "Marianna Nezhurina", "Amro Abbas", "Cheng-Yu Hsieh", "Dhruba Ghosh", "Josh Gardner", "Maciej Kilian", "Hanlin Zhang", "Rulin Shao", "Sarah Pratt", "Sunny Sanyal", "Gabriel Ilharco", "Giannis Daras", "Kalyani Marathe", "Aaron Gokaslan", "Jieyu Zhang", "Khyathi Chandu", "Thao Nguyen", "Igor Vasiljevic", "Sham Kakade", "Shuran Song", "Sujay Sanghavi", "Fartash Faghri", "Sewoong Oh", "Luke Zettlemoyer", "Kyle Lo", "Alaaeldin El-Nouby", "Hadi Pouransari", "Alexander Toshev", "Stephanie Wang", "Dirk Groeneveld", "Luca Soldani", "Pang Wei Koh", "Jenia Jitsev", "Thomas Kollar", "Alexandros G. Dimakis", "Yair Carmon", "Achal Dave", "Ludwig Schmidt", "Vaishaal Shankar" ], "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-2406-11794v1", "title": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } ], "points": "120000", "published_grade": { "id": "li-et-al-datacomp-language-models-2024", "entry_id": "li-et-al-datacomp-language-models-2024", "band": "A", "modifier": null, "target": { "record_id": "urn:uuid:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "content_hash": "sha256:ce38a63cac0c2143977f30eb21cf3112e1bb687a41c343eb025d42bb42e6799f" }, "what_changed": "Fixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models.", "scope": "Fixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models. The candidate pool is not a claim that every baseline trained on240T tokens. Comparisons are for specified412M–7B scales, compute budgets and evaluation protocols; no universally optimal data filter or corpus-quality theorem. Dataset design and OpenLM/benchmark inheritance remain distinct.", "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:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "content_hash": "sha256:ce38a63cac0c2143977f30eb21cf3112e1bb687a41c343eb025d42bb42e6799f" }, "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": "Jeffrey Li", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Alex Fang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Georgios Smyrnis", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Maor Ivgi", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Matt Jordan", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Samir Gadre", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Hritik Bansal", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Etash Guha", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Sedrick Keh", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Kushal Arora", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Saurabh Garg", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Rui Xin", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Niklas Muenninghoff", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Reinhard Heckel", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Jean Mercat", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Mayee Chen", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Suchin Gururangan", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Mitchell Wortsman", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Alon Albalak", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Yonatan Bitton", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Marianna Nezhurina", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Amro Abbas", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Cheng-Yu Hsieh", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Dhruba Ghosh", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Josh Gardner", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Maciej Kilian", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Hanlin Zhang", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Rulin Shao", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Sarah Pratt", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Sunny Sanyal", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Gabriel Ilharco", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } } ], "evidence": [ { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-0", "kind": "source_reference", "source": { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } } ], "origin": "HUMAN_LEGACY", "record_profile": "wrs-local-research-notes-0.1", "record_type": "research_note", "rrp_version": "0.1", "scope": { "description": "The candidate pool is not a claim that every baseline trained on240T tokens. Comparisons are for specified412M–7B scales, compute budgets and evaluation protocols; no universally optimal data filter or corpus-quality theorem. Dataset design and OpenLM/benchmark inheritance remain distinct.", "domain": "Language-model dataset evaluation" }, "sources": [ { "citation": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1" } ], "statement": "# DataComp-LM: controlled language-model data-curation testbed\n\nFixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models.\n\n## Scientific scope\n\nThe candidate pool is not a claim that every baseline trained on240T tokens. Comparisons are for specified412M–7B scales, compute budgets and evaluation protocols; no universally optimal data filter or corpus-quality theorem. Dataset design and OpenLM/benchmark inheritance remain distinct.\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\nJeffrey Li; Alex Fang; Georgios Smyrnis; Maor Ivgi; Matt Jordan; Samir Gadre; Hritik Bansal; Etash Guha; Sedrick Keh; Kushal Arora; Saurabh Garg; Rui Xin; Niklas Muenninghoff; Reinhard Heckel; Jean Mercat; Mayee Chen; Suchin Gururangan; Mitchell Wortsman; Alon Albalak; Yonatan Bitton; Marianna Nezhurina; Amro Abbas; Cheng-Yu Hsieh; Dhruba Ghosh; Josh Gardner; Maciej Kilian; Hanlin Zhang; Rulin Shao; Sarah Pratt; Sunny Sanyal; Gabriel Ilharco; Giannis Daras; Kalyani Marathe; Aaron Gokaslan; Jieyu Zhang; Khyathi Chandu; Thao Nguyen; Igor Vasiljevic; Sham Kakade; Shuran Song; Sujay Sanghavi; Fartash Faghri; Sewoong Oh; Luke Zettlemoyer; Kyle Lo; Alaaeldin El-Nouby; Hadi Pouransari; Alexander Toshev; Stephanie Wang; Dirk Groeneveld; Luca Soldani; Pang Wei Koh; Jenia Jitsev; Thomas Kollar; Alexandros G. Dimakis; Yair Carmon; Achal Dave; Ludwig Schmidt; Vaishaal Shankar\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- [DataComp-LM: In search of the next generation of training sets for language models](https://arxiv.org/abs/2406.11794v1)\n", "summary": "Fixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models.", "title": "DataComp-LM: controlled language-model data-curation testbed" }, "record_id": "urn:uuid:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "source_ref": { "record_id": "urn:uuid:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "content_hash": "sha256:ce38a63cac0c2143977f30eb21cf3112e1bb687a41c343eb025d42bb42e6799f" }, "export_view": "wrs-public-editorial-view-1", "public_view_hash": "sha256:c57acfcbe2c12016d275fc733b98b280aa772190b6ab2b4035b75447cd6a8346" }, "target": { "record_id": "urn:uuid:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "content_hash": "sha256:ce38a63cac0c2143977f30eb21cf3112e1bb687a41c343eb025d42bb42e6799f" }, "published_grade": { "profile": "wrs-published-grade-1", "edition": "2026-10-28.1", "id": "li-et-al-datacomp-language-models-2024", "entry_id": "li-et-al-datacomp-language-models-2024", "revision": 1, "title": "DataComp-LM: controlled language-model data-curation testbed", "context": "Language-model dataset evaluation", "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": "Fixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models.", "scope": "Fixed training recipes and evaluation tasks around a240T-token candidate pool, enabling controlled experiments in filtering, deduplication and mixing, with DCLM-Baseline and released language models. The candidate pool is not a claim that every baseline trained on240T tokens. Comparisons are for specified412M–7B scales, compute budgets and evaluation protocols; no universally optimal data filter or corpus-quality theorem. Dataset design and OpenLM/benchmark inheritance remain distinct.", "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 candidate pool is not a claim that every baseline trained on240T tokens. Comparisons are for specified412M–7B scales, compute budgets and evaluation protocols; no universally optimal data filter or corpus-quality theorem. Dataset design and OpenLM/benchmark inheritance remain distinct.", "references": [ { "title": "DataComp-LM: In search of the next generation of training sets for language models", "url": "https://arxiv.org/abs/2406.11794v1", "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:4e5907fe-fb45-4ef5-af73-45869e1adcd8", "version": 1, "content_hash": "sha256:ce38a63cac0c2143977f30eb21cf3112e1bb687a41c343eb025d42bb42e6799f" }, "scientific_correctness_certified": false, "independent_expert_calibration": false, "method": "PROPRIETARY", "points": "120000" }, "canonical_agent_submission": false, "scientific_verification_awarded": false, "reader_guide": null, "reading_status": "SOURCE_LINKED_NOTE" }