{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:TLBZPLOPVN2J237W6YVXQ6LCCA","short_pith_number":"pith:TLBZPLOP","schema_version":"1.0","canonical_sha256":"9ac397adcfab749d6ff6f62b787962101d1a7e3919c4cfceb043c77a35e00f8b","source":{"kind":"arxiv","id":"2309.10677","version":2},"attestation_state":"computed","paper":{"title":"Estimating Contamination via Perplexity: Quantifying Memorisation in Language Model Evaluation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Yucheng Li","submitted_at":"2023-09-19T15:02:58Z","abstract_excerpt":"Data contamination in model evaluation is getting increasingly prevalent as the massive training corpora of large language models often unintentionally include benchmark samples. Therefore, contamination analysis has became an inevitable part of reliable model evaluation. However, existing method of contamination analysis requires the access of the entire training data which is often confidential for recent models. This prevent the community to rigorously audit these models and conduct accurate assessment of their capability. In this paper, we propose a novel method to quantify contamination w"},"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":"2309.10677","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-09-19T15:02:58Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"17853f95a306c42e6ba6de67d6e28efb1d20841815e39dbebff3d0acd2ed14ff","abstract_canon_sha256":"a3dd82e8e3593474ffef8bde8b230eff5f73bbc59e36efc3528416f2a0e5c5fc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:54:40.625305Z","signature_b64":"+H+YQrXeMCWiq22mPZn9Z175l5tfcofpJKjp2Xymzs8K7EOT2RDDJj2bai2sXdrD09diRUJfcwYf6xZJME6CDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9ac397adcfab749d6ff6f62b787962101d1a7e3919c4cfceb043c77a35e00f8b","last_reissued_at":"2026-07-05T06:54:40.624824Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:54:40.624824Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Estimating Contamination via Perplexity: Quantifying Memorisation in Language Model Evaluation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Yucheng Li","submitted_at":"2023-09-19T15:02:58Z","abstract_excerpt":"Data contamination in model evaluation is getting increasingly prevalent as the massive training corpora of large language models often unintentionally include benchmark samples. Therefore, contamination analysis has became an inevitable part of reliable model evaluation. However, existing method of contamination analysis requires the access of the entire training data which is often confidential for recent models. This prevent the community to rigorously audit these models and conduct accurate assessment of their capability. In this paper, we propose a novel method to quantify contamination w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2309.10677","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/2309.10677/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":"2309.10677","created_at":"2026-07-05T06:54:40.624887+00:00"},{"alias_kind":"arxiv_version","alias_value":"2309.10677v2","created_at":"2026-07-05T06:54:40.624887+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2309.10677","created_at":"2026-07-05T06:54:40.624887+00:00"},{"alias_kind":"pith_short_12","alias_value":"TLBZPLOPVN2J","created_at":"2026-07-05T06:54:40.624887+00:00"},{"alias_kind":"pith_short_16","alias_value":"TLBZPLOPVN2J237W","created_at":"2026-07-05T06:54:40.624887+00:00"},{"alias_kind":"pith_short_8","alias_value":"TLBZPLOP","created_at":"2026-07-05T06:54:40.624887+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.26133","citing_title":"Pretraining Data Exposure in Large Language Models: A Survey of Membership Inference, Data Contamination, and Security Implications","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2406.04244","citing_title":"Benchmark Data Contamination of Large Language Models: A Survey","ref_index":94,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21543","citing_title":"Provable Joint Decontamination for Benchmarking Multiple Large Language Models","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07046","citing_title":"An Interpretable and Scalable Framework for Evaluating Large Language Models","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06327","citing_title":"Measuring Evaluation-Context Divergence in Open-Weight LLMs: A Paired-Prompt Protocol with Pilot Evidence of Alignment-Pipeline-Specific Heterogeneity","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA","json":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA.json","graph_json":"https://pith.science/api/pith-number/TLBZPLOPVN2J237W6YVXQ6LCCA/graph.json","events_json":"https://pith.science/api/pith-number/TLBZPLOPVN2J237W6YVXQ6LCCA/events.json","paper":"https://pith.science/paper/TLBZPLOP"},"agent_actions":{"view_html":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA","download_json":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA.json","view_paper":"https://pith.science/paper/TLBZPLOP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2309.10677&json=true","fetch_graph":"https://pith.science/api/pith-number/TLBZPLOPVN2J237W6YVXQ6LCCA/graph.json","fetch_events":"https://pith.science/api/pith-number/TLBZPLOPVN2J237W6YVXQ6LCCA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA/action/storage_attestation","attest_author":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA/action/author_attestation","sign_citation":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA/action/citation_signature","submit_replication":"https://pith.science/pith/TLBZPLOPVN2J237W6YVXQ6LCCA/action/replication_record"}},"created_at":"2026-07-05T06:54:40.624887+00:00","updated_at":"2026-07-05T06:54:40.624887+00:00"}