{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XFP3WYZPJL4RFATOICNKXSUQA2","short_pith_number":"pith:XFP3WYZP","schema_version":"1.0","canonical_sha256":"b95fbb632f4af912826e409aabca9006b7bbe828ea9fccc8bb7e60f619f82f99","source":{"kind":"arxiv","id":"2505.18102","version":7},"attestation_state":"computed","paper":{"title":"CapBencher: Give Your LLM Benchmark a Built-in Alarm for Test-Set Overfitting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ME"],"primary_cat":"cs.LG","authors_text":"Ikko Yamane, Takashi Ishida, Thanawat Lodkaew","submitted_at":"2025-05-23T16:57:34Z","abstract_excerpt":"Publishing a large language model (LLM) benchmark (especially its ground-truth answers) on the Internet risks contaminating future LLMs and enabling evaluation gaming: it may be unintentionally (or intentionally) used to train or select a model, or exploited to overfit and hack leaderboards when labels are accessible. A common mitigation is to keep the benchmark private and let participants submit their models or predictions to the organizers, but this still permits test-set overfitting through feedback loops. To overcome this issue, we propose CapBencher, a way to publish benchmarks without f"},"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":"2505.18102","kind":"arxiv","version":7},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-23T16:57:34Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ME"],"title_canon_sha256":"ce38c6be55ca22e34be3d4cb8fe009299248b25531fc5635ffd6e4253ae33e32","abstract_canon_sha256":"437df2639b5a11d442cab29f3579cf4cdc18d8ff0427cf1e0fe0276289e84244"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:31.629562Z","signature_b64":"/4P/6UL4PTARZV6WQ81X+6z+c48FcAMDSiZxtSTCiAi2/SzX3dZr7S2ABWIX9suSlKBy1QzAsmpFQcpmu8fOBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b95fbb632f4af912826e409aabca9006b7bbe828ea9fccc8bb7e60f619f82f99","last_reissued_at":"2026-06-02T01:03:31.629048Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:31.629048Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CapBencher: Give Your LLM Benchmark a Built-in Alarm for Test-Set Overfitting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ME"],"primary_cat":"cs.LG","authors_text":"Ikko Yamane, Takashi Ishida, Thanawat Lodkaew","submitted_at":"2025-05-23T16:57:34Z","abstract_excerpt":"Publishing a large language model (LLM) benchmark (especially its ground-truth answers) on the Internet risks contaminating future LLMs and enabling evaluation gaming: it may be unintentionally (or intentionally) used to train or select a model, or exploited to overfit and hack leaderboards when labels are accessible. A common mitigation is to keep the benchmark private and let participants submit their models or predictions to the organizers, but this still permits test-set overfitting through feedback loops. To overcome this issue, we propose CapBencher, a way to publish benchmarks without f"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.18102","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/2505.18102/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":"2505.18102","created_at":"2026-06-02T01:03:31.629114+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.18102v7","created_at":"2026-06-02T01:03:31.629114+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.18102","created_at":"2026-06-02T01:03:31.629114+00:00"},{"alias_kind":"pith_short_12","alias_value":"XFP3WYZPJL4R","created_at":"2026-06-02T01:03:31.629114+00:00"},{"alias_kind":"pith_short_16","alias_value":"XFP3WYZPJL4RFATO","created_at":"2026-06-02T01:03:31.629114+00:00"},{"alias_kind":"pith_short_8","alias_value":"XFP3WYZP","created_at":"2026-06-02T01:03:31.629114+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.09678","citing_title":"NetAgentBench: A State-Centric Benchmark for Evaluating Agentic Network Configuration","ref_index":15,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2","json":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2.json","graph_json":"https://pith.science/api/pith-number/XFP3WYZPJL4RFATOICNKXSUQA2/graph.json","events_json":"https://pith.science/api/pith-number/XFP3WYZPJL4RFATOICNKXSUQA2/events.json","paper":"https://pith.science/paper/XFP3WYZP"},"agent_actions":{"view_html":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2","download_json":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2.json","view_paper":"https://pith.science/paper/XFP3WYZP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.18102&json=true","fetch_graph":"https://pith.science/api/pith-number/XFP3WYZPJL4RFATOICNKXSUQA2/graph.json","fetch_events":"https://pith.science/api/pith-number/XFP3WYZPJL4RFATOICNKXSUQA2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2/action/storage_attestation","attest_author":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2/action/author_attestation","sign_citation":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2/action/citation_signature","submit_replication":"https://pith.science/pith/XFP3WYZPJL4RFATOICNKXSUQA2/action/replication_record"}},"created_at":"2026-06-02T01:03:31.629114+00:00","updated_at":"2026-06-02T01:03:31.629114+00:00"}