{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2EN5SFX37SOYDGCALPOK2SJT7N","short_pith_number":"pith:2EN5SFX3","schema_version":"1.0","canonical_sha256":"d11bd916fbfc9d8198405bdcad4933fb74d0ecd839540308a3c2ed41ae9386bf","source":{"kind":"arxiv","id":"2503.01052","version":2},"attestation_state":"computed","paper":{"title":"ALinFiK: Learning to Approximate Linearized Future Influence Kernel for Scalable Third-Party LLM Data Valuation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Denghui Zhang, Huawei Lin, Jiamin Chen, Weijie Zhao, Xiaodong Yu, Yanzhou Pan, Yide Ran, Zhaozhuo Xu","submitted_at":"2025-03-02T22:51:12Z","abstract_excerpt":"Large Language Models (LLMs) heavily rely on high-quality training data, making data valuation crucial for optimizing model performance, especially when working within a limited budget. In this work, we aim to offer a third-party data valuation approach that benefits both data providers and model developers. We introduce a linearized future influence kernel (LinFiK), which assesses the value of individual data samples in improving LLM performance during training. We further propose ALinFiK, a learning strategy to approximate LinFiK, enabling scalable data valuation. Our comprehensive evaluatio"},"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":"2503.01052","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-02T22:51:12Z","cross_cats_sorted":[],"title_canon_sha256":"55b1277c091e6a549f87e7fc1fec39f497835aa99f5d2afe33829163e8f4ba57","abstract_canon_sha256":"783298585e6f890ae235f900a06960b87ede3901e61ba72139a6335e8622dd56"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:01:56.885470Z","signature_b64":"Rg2szGZEIggQojrIuEiHl1J5RwO6fUAEOWTo1gxODVK6ZC11mItN33TcAM0KNVgKA7cCUIJNRjK/iQwMSvZNBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d11bd916fbfc9d8198405bdcad4933fb74d0ecd839540308a3c2ed41ae9386bf","last_reissued_at":"2026-07-05T11:01:56.884956Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:01:56.884956Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ALinFiK: Learning to Approximate Linearized Future Influence Kernel for Scalable Third-Party LLM Data Valuation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Denghui Zhang, Huawei Lin, Jiamin Chen, Weijie Zhao, Xiaodong Yu, Yanzhou Pan, Yide Ran, Zhaozhuo Xu","submitted_at":"2025-03-02T22:51:12Z","abstract_excerpt":"Large Language Models (LLMs) heavily rely on high-quality training data, making data valuation crucial for optimizing model performance, especially when working within a limited budget. In this work, we aim to offer a third-party data valuation approach that benefits both data providers and model developers. We introduce a linearized future influence kernel (LinFiK), which assesses the value of individual data samples in improving LLM performance during training. We further propose ALinFiK, a learning strategy to approximate LinFiK, enabling scalable data valuation. Our comprehensive evaluatio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.01052","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/2503.01052/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":"2503.01052","created_at":"2026-07-05T11:01:56.885013+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.01052v2","created_at":"2026-07-05T11:01:56.885013+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.01052","created_at":"2026-07-05T11:01:56.885013+00:00"},{"alias_kind":"pith_short_12","alias_value":"2EN5SFX37SOY","created_at":"2026-07-05T11:01:56.885013+00:00"},{"alias_kind":"pith_short_16","alias_value":"2EN5SFX37SOYDGCA","created_at":"2026-07-05T11:01:56.885013+00:00"},{"alias_kind":"pith_short_8","alias_value":"2EN5SFX3","created_at":"2026-07-05T11:01:56.885013+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.16197","citing_title":"Sketching the Readout of Large Language Models for Scalable Data Attribution and Valuation","ref_index":38,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N","json":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N.json","graph_json":"https://pith.science/api/pith-number/2EN5SFX37SOYDGCALPOK2SJT7N/graph.json","events_json":"https://pith.science/api/pith-number/2EN5SFX37SOYDGCALPOK2SJT7N/events.json","paper":"https://pith.science/paper/2EN5SFX3"},"agent_actions":{"view_html":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N","download_json":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N.json","view_paper":"https://pith.science/paper/2EN5SFX3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.01052&json=true","fetch_graph":"https://pith.science/api/pith-number/2EN5SFX37SOYDGCALPOK2SJT7N/graph.json","fetch_events":"https://pith.science/api/pith-number/2EN5SFX37SOYDGCALPOK2SJT7N/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N/action/storage_attestation","attest_author":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N/action/author_attestation","sign_citation":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N/action/citation_signature","submit_replication":"https://pith.science/pith/2EN5SFX37SOYDGCALPOK2SJT7N/action/replication_record"}},"created_at":"2026-07-05T11:01:56.885013+00:00","updated_at":"2026-07-05T11:01:56.885013+00:00"}