{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:P2HXRADDF7BSAP4UI5JPZVCW7K","short_pith_number":"pith:P2HXRADD","schema_version":"1.0","canonical_sha256":"7e8f7880632fc3203f944752fcd456fa84323f5cd31c5a1c54b3d4f95c85f443","source":{"kind":"arxiv","id":"2503.06643","version":1},"attestation_state":"computed","paper":{"title":"Is Your Benchmark (Still) Useful? Dynamic Benchmarking for Code Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.SE","authors_text":"Batu Guan, Shaohua Li, Xiao Wu, Yuanyuan Yuan","submitted_at":"2025-03-09T14:41:18Z","abstract_excerpt":"In this paper, we tackle a critical challenge in model evaluation: how to keep code benchmarks useful when models might have already seen them during training. We introduce a novel solution, dynamic benchmarking framework, to address this challenge. Given a code understanding or reasoning benchmark, our framework dynamically transforms each input, i.e., programs, with various semantic-preserving mutations to build a syntactically new while semantically identical benchmark. We evaluated ten popular language models on our dynamic benchmarks. Our evaluation reveals several interesting or surprisi"},"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":"2503.06643","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2025-03-09T14:41:18Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"fb23adece69f559224aa87bfe918b4e019c2eb7ef63d66dbc781a29eac22c5e8","abstract_canon_sha256":"175d92e794c4a9c0b1208b299d56bfe35ed7ecbd4ff712befc7c87fce809e333"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:27:39.030284Z","signature_b64":"bSc13ZAWW0jQtJs2S2Q3+8YyaIUbfnlE/RicuXfNgXQ34WKdaFRjH5rAM6ThRD+Xe2dbK+bUdG6Yl2RK6P0wBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7e8f7880632fc3203f944752fcd456fa84323f5cd31c5a1c54b3d4f95c85f443","last_reissued_at":"2026-07-05T10:27:39.029654Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:27:39.029654Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Is Your Benchmark (Still) Useful? Dynamic Benchmarking for Code Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.SE","authors_text":"Batu Guan, Shaohua Li, Xiao Wu, Yuanyuan Yuan","submitted_at":"2025-03-09T14:41:18Z","abstract_excerpt":"In this paper, we tackle a critical challenge in model evaluation: how to keep code benchmarks useful when models might have already seen them during training. We introduce a novel solution, dynamic benchmarking framework, to address this challenge. Given a code understanding or reasoning benchmark, our framework dynamically transforms each input, i.e., programs, with various semantic-preserving mutations to build a syntactically new while semantically identical benchmark. We evaluated ten popular language models on our dynamic benchmarks. Our evaluation reveals several interesting or surprisi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.06643","kind":"arxiv","version":1},"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/2503.06643/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":"2503.06643","created_at":"2026-07-05T10:27:39.029725+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.06643v1","created_at":"2026-07-05T10:27:39.029725+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.06643","created_at":"2026-07-05T10:27:39.029725+00:00"},{"alias_kind":"pith_short_12","alias_value":"P2HXRADDF7BS","created_at":"2026-07-05T10:27:39.029725+00:00"},{"alias_kind":"pith_short_16","alias_value":"P2HXRADDF7BSAP4U","created_at":"2026-07-05T10:27:39.029725+00:00"},{"alias_kind":"pith_short_8","alias_value":"P2HXRADD","created_at":"2026-07-05T10:27:39.029725+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.29490","citing_title":"CODEFUSE-DEBENCH: An Empirical Study on Readability, Recompilability, and Functionality","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10448","citing_title":"Can Agent Benchmarks Support Their Scores? Evidence-Supported Bounds for Interactive-Agent Evaluation","ref_index":6,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K","json":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K.json","graph_json":"https://pith.science/api/pith-number/P2HXRADDF7BSAP4UI5JPZVCW7K/graph.json","events_json":"https://pith.science/api/pith-number/P2HXRADDF7BSAP4UI5JPZVCW7K/events.json","paper":"https://pith.science/paper/P2HXRADD"},"agent_actions":{"view_html":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K","download_json":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K.json","view_paper":"https://pith.science/paper/P2HXRADD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.06643&json=true","fetch_graph":"https://pith.science/api/pith-number/P2HXRADDF7BSAP4UI5JPZVCW7K/graph.json","fetch_events":"https://pith.science/api/pith-number/P2HXRADDF7BSAP4UI5JPZVCW7K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K/action/storage_attestation","attest_author":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K/action/author_attestation","sign_citation":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K/action/citation_signature","submit_replication":"https://pith.science/pith/P2HXRADDF7BSAP4UI5JPZVCW7K/action/replication_record"}},"created_at":"2026-07-05T10:27:39.029725+00:00","updated_at":"2026-07-05T10:27:39.029725+00:00"}