{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:SIEAEP2PAYI5ZK2PNRLE3XSHUB","short_pith_number":"pith:SIEAEP2P","schema_version":"1.0","canonical_sha256":"9208023f4f0611dcab4f6c564dde47a07bc0aa03ae30b4ac2e5df1ae2723155d","source":{"kind":"arxiv","id":"2406.06046","version":2},"attestation_state":"computed","paper":{"title":"MATES: Model-Aware Data Selection for Efficient Pretraining with Data Influence Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Chenyan Xiong, Spandan Das, Zichun Yu","submitted_at":"2024-06-10T06:27:42Z","abstract_excerpt":"Pretraining data selection has the potential to improve language model pretraining efficiency by utilizing higher-quality data from massive web data corpora. Current data selection methods, which rely on either hand-crafted rules or larger reference models, are conducted statically and do not capture the evolving data preferences during pretraining. In this paper, we introduce model-aware data selection with data influence models (MATES), where a data influence model continuously adapts to the evolving data preferences of the pretraining model and then selects the data most effective for the c"},"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":"2406.06046","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-10T06:27:42Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ea8eecb552bf1ff7813d24c697c398d163506f6c2d2c7c61ccbe69f0fbbcfb2d","abstract_canon_sha256":"561067d09f867032e95c738f4ead26f2707c1cc8cc5a2b406562b9c9b542a7b9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:36:12.235857Z","signature_b64":"5ldiMP9dgzhkFwaPV648nfcX7LvZuF192JrYJsP0F8utJ7wXmme7vwBcj38youJsSEk5w+HAcnMqA51cKS0NCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9208023f4f0611dcab4f6c564dde47a07bc0aa03ae30b4ac2e5df1ae2723155d","last_reissued_at":"2026-07-05T09:36:12.235348Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:36:12.235348Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MATES: Model-Aware Data Selection for Efficient Pretraining with Data Influence Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Chenyan Xiong, Spandan Das, Zichun Yu","submitted_at":"2024-06-10T06:27:42Z","abstract_excerpt":"Pretraining data selection has the potential to improve language model pretraining efficiency by utilizing higher-quality data from massive web data corpora. Current data selection methods, which rely on either hand-crafted rules or larger reference models, are conducted statically and do not capture the evolving data preferences during pretraining. In this paper, we introduce model-aware data selection with data influence models (MATES), where a data influence model continuously adapts to the evolving data preferences of the pretraining model and then selects the data most effective for the c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.06046","kind":"arxiv","version":2},"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/2406.06046/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":"2406.06046","created_at":"2026-07-05T09:36:12.235406+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.06046v2","created_at":"2026-07-05T09:36:12.235406+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.06046","created_at":"2026-07-05T09:36:12.235406+00:00"},{"alias_kind":"pith_short_12","alias_value":"SIEAEP2PAYI5","created_at":"2026-07-05T09:36:12.235406+00:00"},{"alias_kind":"pith_short_16","alias_value":"SIEAEP2PAYI5ZK2P","created_at":"2026-07-05T09:36:12.235406+00:00"},{"alias_kind":"pith_short_8","alias_value":"SIEAEP2P","created_at":"2026-07-05T09:36:12.235406+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.02266","citing_title":"HERMES: A Multi-Granularity Labeling Substrate for Pre-training Data Mixtures","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11810","citing_title":"GRACE: A Dynamic Coreset Selection Framework for Large Language Model Optimization","ref_index":80,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07769","citing_title":"An Empirical Study on Influence-Based Pretraining Data Selection for Code Large Language Models","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07739","citing_title":"Efficient Dataset Selection for Continual Adaptation of Generative Recommenders","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16197","citing_title":"Sketching the Readout of Large Language Models for Scalable Data Attribution and Valuation","ref_index":57,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB","json":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB.json","graph_json":"https://pith.science/api/pith-number/SIEAEP2PAYI5ZK2PNRLE3XSHUB/graph.json","events_json":"https://pith.science/api/pith-number/SIEAEP2PAYI5ZK2PNRLE3XSHUB/events.json","paper":"https://pith.science/paper/SIEAEP2P"},"agent_actions":{"view_html":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB","download_json":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB.json","view_paper":"https://pith.science/paper/SIEAEP2P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.06046&json=true","fetch_graph":"https://pith.science/api/pith-number/SIEAEP2PAYI5ZK2PNRLE3XSHUB/graph.json","fetch_events":"https://pith.science/api/pith-number/SIEAEP2PAYI5ZK2PNRLE3XSHUB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB/action/storage_attestation","attest_author":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB/action/author_attestation","sign_citation":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB/action/citation_signature","submit_replication":"https://pith.science/pith/SIEAEP2PAYI5ZK2PNRLE3XSHUB/action/replication_record"}},"created_at":"2026-07-05T09:36:12.235406+00:00","updated_at":"2026-07-05T09:36:12.235406+00:00"}