{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UEPS22F7R4GJI5AB5CI7AVUDEX","short_pith_number":"pith:UEPS22F7","schema_version":"1.0","canonical_sha256":"a11f2d68bf8f0c947401e891f0568325d218e526f771be9a616601e9b0c7bb05","source":{"kind":"arxiv","id":"2502.00678","version":2},"attestation_state":"computed","paper":{"title":"How Contaminated Is Your Benchmark? Quantifying Dataset Leakage in Large Language Models with Kernel Divergence","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Hongxin Wei, Hyeong Kyu Choi, Maxim Khanov, Yixuan Li","submitted_at":"2025-02-02T05:50:39Z","abstract_excerpt":"Dataset contamination, where evaluation datasets overlap with pre-training corpora, inflates performance metrics and undermines the reliability of model evaluations. Measuring dataset contamination thus becomes essential to ensure that performance evaluations genuinely reflect a model's ability to generalize to unseen data, rather than relying on memorized examples. To address this problem, we propose Kernel Divergence Score (KDS), a novel method that evaluates dataset contamination by computing the divergence between the kernel similarity matrix of sample embeddings, before and after fine-tun"},"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":"2502.00678","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-02T05:50:39Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"4fad277095ff039e7c76509344cd176df39c50ab053d0723188811949f1ea7b2","abstract_canon_sha256":"a12cb1fa1e2ec97c2465d66c638d9fc068cd2ef8f6ba7aed34fd8a25d256739e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:27.519855Z","signature_b64":"nDh1C57pxIwMV/xebtRZsDfANJyzNvgUwBa7RlLYzDEz+PS2d1+wowoEHf7A3sesmbZ68uzvm/ZOK5KFIj+DBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a11f2d68bf8f0c947401e891f0568325d218e526f771be9a616601e9b0c7bb05","last_reissued_at":"2026-07-05T11:06:27.519286Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:27.519286Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How Contaminated Is Your Benchmark? Quantifying Dataset Leakage in Large Language Models with Kernel Divergence","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Hongxin Wei, Hyeong Kyu Choi, Maxim Khanov, Yixuan Li","submitted_at":"2025-02-02T05:50:39Z","abstract_excerpt":"Dataset contamination, where evaluation datasets overlap with pre-training corpora, inflates performance metrics and undermines the reliability of model evaluations. Measuring dataset contamination thus becomes essential to ensure that performance evaluations genuinely reflect a model's ability to generalize to unseen data, rather than relying on memorized examples. To address this problem, we propose Kernel Divergence Score (KDS), a novel method that evaluates dataset contamination by computing the divergence between the kernel similarity matrix of sample embeddings, before and after fine-tun"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.00678","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/2502.00678/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":"2502.00678","created_at":"2026-07-05T11:06:27.519358+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.00678v2","created_at":"2026-07-05T11:06:27.519358+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.00678","created_at":"2026-07-05T11:06:27.519358+00:00"},{"alias_kind":"pith_short_12","alias_value":"UEPS22F7R4GJ","created_at":"2026-07-05T11:06:27.519358+00:00"},{"alias_kind":"pith_short_16","alias_value":"UEPS22F7R4GJI5AB","created_at":"2026-07-05T11:06:27.519358+00:00"},{"alias_kind":"pith_short_8","alias_value":"UEPS22F7","created_at":"2026-07-05T11:06:27.519358+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07996","citing_title":"MC-PDD: Masked Corpus-Level Pretraining Data Detection for Black-Box Large Language Models","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07805","citing_title":"Beyond Goodhart's Law: A Dynamic Benchmark for Evaluating Compliance in Multi-Agent Systems","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21255","citing_title":"When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX","json":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX.json","graph_json":"https://pith.science/api/pith-number/UEPS22F7R4GJI5AB5CI7AVUDEX/graph.json","events_json":"https://pith.science/api/pith-number/UEPS22F7R4GJI5AB5CI7AVUDEX/events.json","paper":"https://pith.science/paper/UEPS22F7"},"agent_actions":{"view_html":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX","download_json":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX.json","view_paper":"https://pith.science/paper/UEPS22F7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.00678&json=true","fetch_graph":"https://pith.science/api/pith-number/UEPS22F7R4GJI5AB5CI7AVUDEX/graph.json","fetch_events":"https://pith.science/api/pith-number/UEPS22F7R4GJI5AB5CI7AVUDEX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX/action/storage_attestation","attest_author":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX/action/author_attestation","sign_citation":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX/action/citation_signature","submit_replication":"https://pith.science/pith/UEPS22F7R4GJI5AB5CI7AVUDEX/action/replication_record"}},"created_at":"2026-07-05T11:06:27.519358+00:00","updated_at":"2026-07-05T11:06:27.519358+00:00"}