{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6FKI3MVNGBUAVHFC2SQN23HO5W","short_pith_number":"pith:6FKI3MVN","schema_version":"1.0","canonical_sha256":"f1548db2ad30680a9ca2d4a0dd6ceeed806f05a517d84bda579f26e826c16bde","source":{"kind":"arxiv","id":"2508.15096","version":1},"attestation_state":"computed","paper":{"title":"Nemotron-CC-Math: A 133 Billion-Token-Scale High Quality Math Pretraining Dataset","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bryan Catanzaro, Mohammad Shoeybi, Mostofa Patwary, Rabeeh Karimi Mahabadi, Sanjeev Satheesh, Shrimai Prabhumoye","submitted_at":"2025-08-20T22:16:57Z","abstract_excerpt":"Pretraining large language models (LLMs) on high-quality, structured data such as mathematics and code substantially enhances reasoning capabilities. However, existing math-focused datasets built from Common Crawl suffer from degraded quality due to brittle extraction heuristics, lossy HTML-to-text conversion, and the failure to reliably preserve mathematical structure. In this work, we introduce Nemotron-CC-Math, a large-scale, high-quality mathematical corpus constructed from Common Crawl using a novel, domain-agnostic pipeline specifically designed for robust scientific text extraction.\n  U"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.15096","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-20T22:16:57Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"6aa295f797c75336a3d0980416668363292f54630f3ae7143d76b900b7b98ee5","abstract_canon_sha256":"7cfa4424d1bb74c99a833bc03487d01df5e5f0c4bd82ec2f595598ede235c86c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:57:06.406120Z","signature_b64":"8RzMmZwZcdWx9xb+2dTpREmIOiRJOCS3k768FIAW7GBY1TXtgXZBwASpvuqLDNcT0Y/jnF3kon+Oi2VpCaS4DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f1548db2ad30680a9ca2d4a0dd6ceeed806f05a517d84bda579f26e826c16bde","last_reissued_at":"2026-07-05T11:57:06.405622Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:57:06.405622Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Nemotron-CC-Math: A 133 Billion-Token-Scale High Quality Math Pretraining Dataset","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bryan Catanzaro, Mohammad Shoeybi, Mostofa Patwary, Rabeeh Karimi Mahabadi, Sanjeev Satheesh, Shrimai Prabhumoye","submitted_at":"2025-08-20T22:16:57Z","abstract_excerpt":"Pretraining large language models (LLMs) on high-quality, structured data such as mathematics and code substantially enhances reasoning capabilities. However, existing math-focused datasets built from Common Crawl suffer from degraded quality due to brittle extraction heuristics, lossy HTML-to-text conversion, and the failure to reliably preserve mathematical structure. In this work, we introduce Nemotron-CC-Math, a large-scale, high-quality mathematical corpus constructed from Common Crawl using a novel, domain-agnostic pipeline specifically designed for robust scientific text extraction.\n  U"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.15096","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.15096/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.15096","created_at":"2026-07-05T11:57:06.405681+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.15096v1","created_at":"2026-07-05T11:57:06.405681+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.15096","created_at":"2026-07-05T11:57:06.405681+00:00"},{"alias_kind":"pith_short_12","alias_value":"6FKI3MVNGBUA","created_at":"2026-07-05T11:57:06.405681+00:00"},{"alias_kind":"pith_short_16","alias_value":"6FKI3MVNGBUAVHFC","created_at":"2026-07-05T11:57:06.405681+00:00"},{"alias_kind":"pith_short_8","alias_value":"6FKI3MVN","created_at":"2026-07-05T11:57:06.405681+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.15079","citing_title":"Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale","ref_index":160,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26097","citing_title":"Forgetting in Language Models: Capacity, Optimization, and Self-Generated Replay","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17283","citing_title":"OProver: A Unified Framework for Agentic Formal Theorem Proving","ref_index":167,"is_internal_anchor":false},{"citing_arxiv_id":"2512.20856","citing_title":"NVIDIA Nemotron 3: Efficient and Open Intelligence","ref_index":167,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W","json":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W.json","graph_json":"https://pith.science/api/pith-number/6FKI3MVNGBUAVHFC2SQN23HO5W/graph.json","events_json":"https://pith.science/api/pith-number/6FKI3MVNGBUAVHFC2SQN23HO5W/events.json","paper":"https://pith.science/paper/6FKI3MVN"},"agent_actions":{"view_html":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W","download_json":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W.json","view_paper":"https://pith.science/paper/6FKI3MVN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.15096&json=true","fetch_graph":"https://pith.science/api/pith-number/6FKI3MVNGBUAVHFC2SQN23HO5W/graph.json","fetch_events":"https://pith.science/api/pith-number/6FKI3MVNGBUAVHFC2SQN23HO5W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W/action/storage_attestation","attest_author":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W/action/author_attestation","sign_citation":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W/action/citation_signature","submit_replication":"https://pith.science/pith/6FKI3MVNGBUAVHFC2SQN23HO5W/action/replication_record"}},"created_at":"2026-07-05T11:57:06.405681+00:00","updated_at":"2026-07-05T11:57:06.405681+00:00"}