{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JCLUTLQWPVHJB6SL7PZTPRAQ7Q","short_pith_number":"pith:JCLUTLQW","schema_version":"1.0","canonical_sha256":"489749ae167d4e90fa4bfbf337c410fc357e300b7234efa95b3113d7646a380b","source":{"kind":"arxiv","id":"2510.00866","version":3},"attestation_state":"computed","paper":{"title":"Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"David Grangier, Louis Bethune, Marco Cuturi, Michal Klein, Pierre Ablin, Thiziri Nait Saada","submitted_at":"2025-10-01T13:15:15Z","abstract_excerpt":"Large-scale models are pretrained on massive web-crawled datasets containing documents of mixed quality, making data filtering essential. A popular method is Classifier-based Quality Filtering (CQF), which trains a binary classifier to distinguish between pretraining data and a small, high-quality set. It assigns each pretraining document a quality score defined as the classifier's score and retains only the top-scoring ones. We provide an in-depth analysis of CQF. We show that while CQF improves downstream task performance, it does not necessarily enhance language modeling on the high-quality"},"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":"2510.00866","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-10-01T13:15:15Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"f4c6c85256023cdecf1912810654e9c036d88949f08e59736d81a25fa37b37e2","abstract_canon_sha256":"8dcff0c2750c4df1ff8e68b918db9700adea64b7ca6ece0789d004be113f4862"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-24T00:14:21.161735Z","signature_b64":"dN3OCdmZY/MAIPceQfQs/CMC/2XWztKCCJ5sd4jgUd1na5AgBILEUhG5CSjsrPOjLp8rMCvfELLir9kjFgazBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"489749ae167d4e90fa4bfbf337c410fc357e300b7234efa95b3113d7646a380b","last_reissued_at":"2026-06-24T00:14:21.161280Z","signature_status":"signed_v1","first_computed_at":"2026-06-24T00:14:21.161280Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"David Grangier, Louis Bethune, Marco Cuturi, Michal Klein, Pierre Ablin, Thiziri Nait Saada","submitted_at":"2025-10-01T13:15:15Z","abstract_excerpt":"Large-scale models are pretrained on massive web-crawled datasets containing documents of mixed quality, making data filtering essential. A popular method is Classifier-based Quality Filtering (CQF), which trains a binary classifier to distinguish between pretraining data and a small, high-quality set. It assigns each pretraining document a quality score defined as the classifier's score and retains only the top-scoring ones. We provide an in-depth analysis of CQF. We show that while CQF improves downstream task performance, it does not necessarily enhance language modeling on the high-quality"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.00866","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/2510.00866/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":"2510.00866","created_at":"2026-06-24T00:14:21.161342+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.00866v3","created_at":"2026-06-24T00:14:21.161342+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.00866","created_at":"2026-06-24T00:14:21.161342+00:00"},{"alias_kind":"pith_short_12","alias_value":"JCLUTLQWPVHJ","created_at":"2026-06-24T00:14:21.161342+00:00"},{"alias_kind":"pith_short_16","alias_value":"JCLUTLQWPVHJB6SL","created_at":"2026-06-24T00:14:21.161342+00:00"},{"alias_kind":"pith_short_8","alias_value":"JCLUTLQW","created_at":"2026-06-24T00:14:21.161342+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.23721","citing_title":"Is a Document Educational or Just Wikipedia-Style? -- Pitfalls of Classifier-Based Quality Filtering","ref_index":11,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q","json":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q.json","graph_json":"https://pith.science/api/pith-number/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/graph.json","events_json":"https://pith.science/api/pith-number/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/events.json","paper":"https://pith.science/paper/JCLUTLQW"},"agent_actions":{"view_html":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q","download_json":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q.json","view_paper":"https://pith.science/paper/JCLUTLQW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.00866&json=true","fetch_graph":"https://pith.science/api/pith-number/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/graph.json","fetch_events":"https://pith.science/api/pith-number/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/action/storage_attestation","attest_author":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/action/author_attestation","sign_citation":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/action/citation_signature","submit_replication":"https://pith.science/pith/JCLUTLQWPVHJB6SL7PZTPRAQ7Q/action/replication_record"}},"created_at":"2026-06-24T00:14:21.161342+00:00","updated_at":"2026-06-24T00:14:21.161342+00:00"}