{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:B563JTAYOFBILA3HLLH647WUQX","short_pith_number":"pith:B563JTAY","schema_version":"1.0","canonical_sha256":"0f7db4cc1871428583675acfee7ed485dda1fd3b84f5eb81cc13ba54492ecfd9","source":{"kind":"arxiv","id":"2402.07625","version":7},"attestation_state":"computed","paper":{"title":"Autonomous Data Selection with Zero-shot Generative Classifiers for Mathematical Texts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Andrew C Yao, Yang Yuan, Yifan Luo, Yifan Zhang","submitted_at":"2024-02-12T13:09:21Z","abstract_excerpt":"We present Autonomous Data Selection (AutoDS), a method that leverages base language models themselves as zero-shot \"generative classifiers\" to automatically curate high-quality mathematical texts. Unlike prior approaches that require human annotations or training a dedicated data filter, AutoDS relies solely on a model's logits to determine whether a given passage is mathematically informative and educational. By integrating AutoDS into a continual pretraining pipeline, we substantially boost downstream performance on challenging math benchmarks (MATH, GSM8K, and BBH) while using far fewer to"},"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":"2402.07625","kind":"arxiv","version":7},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-12T13:09:21Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"02ae2f6d86ed4ac00ad5e89ba83c9c99b90a882773fa6b2327d3ba04fbe315c3","abstract_canon_sha256":"8dc3a14f071e4e68404da95673e4f7329b5721b842c6a75a33534abf8370bc87"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:40:47.706491Z","signature_b64":"p6yeJ4szfNkXwNft9dBv5r0rhVdM350+wfSkwbRodLMwilIzl+r/G77tneiwrGjZZvcLPu7yB4kJ9BOv++UvDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0f7db4cc1871428583675acfee7ed485dda1fd3b84f5eb81cc13ba54492ecfd9","last_reissued_at":"2026-07-05T11:40:47.705931Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:40:47.705931Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Autonomous Data Selection with Zero-shot Generative Classifiers for Mathematical Texts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Andrew C Yao, Yang Yuan, Yifan Luo, Yifan Zhang","submitted_at":"2024-02-12T13:09:21Z","abstract_excerpt":"We present Autonomous Data Selection (AutoDS), a method that leverages base language models themselves as zero-shot \"generative classifiers\" to automatically curate high-quality mathematical texts. Unlike prior approaches that require human annotations or training a dedicated data filter, AutoDS relies solely on a model's logits to determine whether a given passage is mathematically informative and educational. By integrating AutoDS into a continual pretraining pipeline, we substantially boost downstream performance on challenging math benchmarks (MATH, GSM8K, and BBH) while using far fewer to"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.07625","kind":"arxiv","version":7},"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/2402.07625/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":"2402.07625","created_at":"2026-07-05T11:40:47.705998+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.07625v7","created_at":"2026-07-05T11:40:47.705998+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.07625","created_at":"2026-07-05T11:40:47.705998+00:00"},{"alias_kind":"pith_short_12","alias_value":"B563JTAYOFBI","created_at":"2026-07-05T11:40:47.705998+00:00"},{"alias_kind":"pith_short_16","alias_value":"B563JTAYOFBILA3H","created_at":"2026-07-05T11:40:47.705998+00:00"},{"alias_kind":"pith_short_8","alias_value":"B563JTAY","created_at":"2026-07-05T11:40:47.705998+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.03032","citing_title":"Analyze Feature Flow to Enhance Interpretation and Steering in Language Models","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX","json":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX.json","graph_json":"https://pith.science/api/pith-number/B563JTAYOFBILA3HLLH647WUQX/graph.json","events_json":"https://pith.science/api/pith-number/B563JTAYOFBILA3HLLH647WUQX/events.json","paper":"https://pith.science/paper/B563JTAY"},"agent_actions":{"view_html":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX","download_json":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX.json","view_paper":"https://pith.science/paper/B563JTAY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.07625&json=true","fetch_graph":"https://pith.science/api/pith-number/B563JTAYOFBILA3HLLH647WUQX/graph.json","fetch_events":"https://pith.science/api/pith-number/B563JTAYOFBILA3HLLH647WUQX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX/action/storage_attestation","attest_author":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX/action/author_attestation","sign_citation":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX/action/citation_signature","submit_replication":"https://pith.science/pith/B563JTAYOFBILA3HLLH647WUQX/action/replication_record"}},"created_at":"2026-07-05T11:40:47.705998+00:00","updated_at":"2026-07-05T11:40:47.705998+00:00"}