{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:UOHLXNIKQVC4APWN4GNZJIO5OC","short_pith_number":"pith:UOHLXNIK","schema_version":"1.0","canonical_sha256":"a38ebbb50a8545c03ecde19b94a1dd70b4aba82374293610603ed2ad6387bb05","source":{"kind":"arxiv","id":"2406.10670","version":3},"attestation_state":"computed","paper":{"title":"CoLoR-Filter: Conditional Loss Reduction Filtering for Targeted Language Model Pre-training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Andreas Kirsch, David Brandfonbrener, Hanlin Zhang, Jonathan Richard Schwarz, Sham Kakade","submitted_at":"2024-06-15T15:28:02Z","abstract_excerpt":"Selecting high-quality data for pre-training is crucial in shaping the downstream task performance of language models. A major challenge lies in identifying this optimal subset, a problem generally considered intractable, thus necessitating scalable and effective heuristics. In this work, we propose a data selection method, CoLoR-Filter (Conditional Loss Reduction Filtering), which leverages an empirical Bayes-inspired approach to derive a simple and computationally efficient selection criterion based on the relative loss values of two auxiliary models.\n  In addition to the modeling rationale,"},"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":"2406.10670","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-15T15:28:02Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"2140454f0bdb845e989e4db18cefb4b0e0f2bbaa50ea45e8c3a1104a050c2b7a","abstract_canon_sha256":"5c704b3693ee2fe3ed5ffe50ce349316594aa100040924f33feb874029df5f67"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:28:12.631611Z","signature_b64":"AmOmN3kQ2BNIqbDw9Q3H1hBbnoVQf78clYagY3xTujTcn6dwpzSCRLDy6q+oEVPdKjRQ3C00a4UdaBzVYgCBAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a38ebbb50a8545c03ecde19b94a1dd70b4aba82374293610603ed2ad6387bb05","last_reissued_at":"2026-07-05T09:28:12.631134Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:28:12.631134Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CoLoR-Filter: Conditional Loss Reduction Filtering for Targeted Language Model Pre-training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Andreas Kirsch, David Brandfonbrener, Hanlin Zhang, Jonathan Richard Schwarz, Sham Kakade","submitted_at":"2024-06-15T15:28:02Z","abstract_excerpt":"Selecting high-quality data for pre-training is crucial in shaping the downstream task performance of language models. A major challenge lies in identifying this optimal subset, a problem generally considered intractable, thus necessitating scalable and effective heuristics. In this work, we propose a data selection method, CoLoR-Filter (Conditional Loss Reduction Filtering), which leverages an empirical Bayes-inspired approach to derive a simple and computationally efficient selection criterion based on the relative loss values of two auxiliary models.\n  In addition to the modeling rationale,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.10670","kind":"arxiv","version":3},"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/2406.10670/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":"2406.10670","created_at":"2026-07-05T09:28:12.631193+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.10670v3","created_at":"2026-07-05T09:28:12.631193+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.10670","created_at":"2026-07-05T09:28:12.631193+00:00"},{"alias_kind":"pith_short_12","alias_value":"UOHLXNIKQVC4","created_at":"2026-07-05T09:28:12.631193+00:00"},{"alias_kind":"pith_short_16","alias_value":"UOHLXNIKQVC4APWN","created_at":"2026-07-05T09:28:12.631193+00:00"},{"alias_kind":"pith_short_8","alias_value":"UOHLXNIK","created_at":"2026-07-05T09:28:12.631193+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/UOHLXNIKQVC4APWN4GNZJIO5OC","json":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC.json","graph_json":"https://pith.science/api/pith-number/UOHLXNIKQVC4APWN4GNZJIO5OC/graph.json","events_json":"https://pith.science/api/pith-number/UOHLXNIKQVC4APWN4GNZJIO5OC/events.json","paper":"https://pith.science/paper/UOHLXNIK"},"agent_actions":{"view_html":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC","download_json":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC.json","view_paper":"https://pith.science/paper/UOHLXNIK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.10670&json=true","fetch_graph":"https://pith.science/api/pith-number/UOHLXNIKQVC4APWN4GNZJIO5OC/graph.json","fetch_events":"https://pith.science/api/pith-number/UOHLXNIKQVC4APWN4GNZJIO5OC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC/action/storage_attestation","attest_author":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC/action/author_attestation","sign_citation":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC/action/citation_signature","submit_replication":"https://pith.science/pith/UOHLXNIKQVC4APWN4GNZJIO5OC/action/replication_record"}},"created_at":"2026-07-05T09:28:12.631193+00:00","updated_at":"2026-07-05T09:28:12.631193+00:00"}