{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QQNJULOQ6SRGGHG3OC4CHMWNV6","short_pith_number":"pith:QQNJULOQ","schema_version":"1.0","canonical_sha256":"841a9a2dd0f4a2631cdb70b823b2cdaf90af562eaf55da5c8346634e259006d0","source":{"kind":"arxiv","id":"2606.06888","version":1},"attestation_state":"computed","paper":{"title":"Data-Constrained Language Model Pretraining: Improved Regularization and Scaling Laws","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hanseul Cho, Shihao Wu, Wei Hu, Yixin Wang, Zhiwei Xu","submitted_at":"2026-06-05T04:10:09Z","abstract_excerpt":"Classical scaling laws for language model pretraining balance model size against training dataset size under a fixed compute budget, assuming abundant data and a single pass over the corpus. As training compute grows faster than the supply of natural language data, pretraining is likely to enter a data-constrained, compute-rich regime where models train for multiple epochs over a finite dataset. We study data-constrained pretraining along two axes, regularization and scaling. For regularization, we study masked-input regularization (MIR), an auxiliary next-token prediction loss on randomly mas"},"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":"2606.06888","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T04:10:09Z","cross_cats_sorted":[],"title_canon_sha256":"84ef3f5230565c2879248e56407818db6e66a3eeb50e933c6fb707d833ee8b9b","abstract_canon_sha256":"33293a16bcd9c12a236b15758895bf83096da825e9e46a9512d902e643eec350"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:04:34.094486Z","signature_b64":"B4lnD7+0CuIo6Ln2L60d3N3iCSIK7iykW7m6yrmBuuG6Fr0vwUczW8x+xxF16AdclFEyMrpKtAVDC2U9tShlCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"841a9a2dd0f4a2631cdb70b823b2cdaf90af562eaf55da5c8346634e259006d0","last_reissued_at":"2026-06-08T01:04:34.093532Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:04:34.093532Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Data-Constrained Language Model Pretraining: Improved Regularization and Scaling Laws","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hanseul Cho, Shihao Wu, Wei Hu, Yixin Wang, Zhiwei Xu","submitted_at":"2026-06-05T04:10:09Z","abstract_excerpt":"Classical scaling laws for language model pretraining balance model size against training dataset size under a fixed compute budget, assuming abundant data and a single pass over the corpus. As training compute grows faster than the supply of natural language data, pretraining is likely to enter a data-constrained, compute-rich regime where models train for multiple epochs over a finite dataset. We study data-constrained pretraining along two axes, regularization and scaling. For regularization, we study masked-input regularization (MIR), an auxiliary next-token prediction loss on randomly mas"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.06888","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/2606.06888/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":"2606.06888","created_at":"2026-06-08T01:04:34.093721+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.06888v1","created_at":"2026-06-08T01:04:34.093721+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.06888","created_at":"2026-06-08T01:04:34.093721+00:00"},{"alias_kind":"pith_short_12","alias_value":"QQNJULOQ6SRG","created_at":"2026-06-08T01:04:34.093721+00:00"},{"alias_kind":"pith_short_16","alias_value":"QQNJULOQ6SRGGHG3","created_at":"2026-06-08T01:04:34.093721+00:00"},{"alias_kind":"pith_short_8","alias_value":"QQNJULOQ","created_at":"2026-06-08T01:04:34.093721+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6","json":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6.json","graph_json":"https://pith.science/api/pith-number/QQNJULOQ6SRGGHG3OC4CHMWNV6/graph.json","events_json":"https://pith.science/api/pith-number/QQNJULOQ6SRGGHG3OC4CHMWNV6/events.json","paper":"https://pith.science/paper/QQNJULOQ"},"agent_actions":{"view_html":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6","download_json":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6.json","view_paper":"https://pith.science/paper/QQNJULOQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.06888&json=true","fetch_graph":"https://pith.science/api/pith-number/QQNJULOQ6SRGGHG3OC4CHMWNV6/graph.json","fetch_events":"https://pith.science/api/pith-number/QQNJULOQ6SRGGHG3OC4CHMWNV6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6/action/storage_attestation","attest_author":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6/action/author_attestation","sign_citation":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6/action/citation_signature","submit_replication":"https://pith.science/pith/QQNJULOQ6SRGGHG3OC4CHMWNV6/action/replication_record"}},"created_at":"2026-06-08T01:04:34.093721+00:00","updated_at":"2026-06-08T01:04:34.093721+00:00"}