{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:WALU5PW2SN3KKH5HLNHDSNJKTZ","short_pith_number":"pith:WALU5PW2","schema_version":"1.0","canonical_sha256":"b0174ebeda9376a51fa75b4e39352a9e704090ef5ab033bc4a3cf3ccc40348b8","source":{"kind":"arxiv","id":"2605.28365","version":1},"attestation_state":"computed","paper":{"title":"Risk-Controlled Lean-as-Judge for Natural-Language Mathematical Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LO"],"primary_cat":"cs.AI","authors_text":"Haitham Bou Ammar, Matthieu Zimmer, Pauline Bourigault, Rasul Tutunov, Xiaotong Ji","submitted_at":"2026-05-27T11:59:28Z","abstract_excerpt":"Lean is increasingly used to judge natural-language mathematical answers, but its signal is partial: many answers never formalize, and a failed proof may reflect an ill-typed statement or a missing library fact, not a wrong answer. On MATH-500 we show this signal is (i) sharply coverage-dependent, that is the proof-winning answer is correct 96% of the time at high proved coverage but 20% at low, and (ii) sparse and often unfaithful: a 7B autoformalizer proves a class for only 28% of problems, and a manual audit finds only approximately 43% of those proofs faithful. We propose COVCAL, a selecto"},"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":"2605.28365","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-27T11:59:28Z","cross_cats_sorted":["cs.CL","cs.LO"],"title_canon_sha256":"39085c431845fde385cb9036c79112693b2b9d5891d971021b950a81dd4db57d","abstract_canon_sha256":"7022bf24598c08051a4293ca102d3879f5dcb74d4cb4007eef58fc6eb0cb869e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T01:05:16.059088Z","signature_b64":"2K9jZBQP1PEqGH3YJn4SqDYpc+uPstezC6/zvWKkgE9UJEtab6cP2vpIml8wDuxRdsNzsXe6Gq9l83+TPUNQDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b0174ebeda9376a51fa75b4e39352a9e704090ef5ab033bc4a3cf3ccc40348b8","last_reissued_at":"2026-05-28T01:05:16.058666Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T01:05:16.058666Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Risk-Controlled Lean-as-Judge for Natural-Language Mathematical Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LO"],"primary_cat":"cs.AI","authors_text":"Haitham Bou Ammar, Matthieu Zimmer, Pauline Bourigault, Rasul Tutunov, Xiaotong Ji","submitted_at":"2026-05-27T11:59:28Z","abstract_excerpt":"Lean is increasingly used to judge natural-language mathematical answers, but its signal is partial: many answers never formalize, and a failed proof may reflect an ill-typed statement or a missing library fact, not a wrong answer. On MATH-500 we show this signal is (i) sharply coverage-dependent, that is the proof-winning answer is correct 96% of the time at high proved coverage but 20% at low, and (ii) sparse and often unfaithful: a 7B autoformalizer proves a class for only 28% of problems, and a manual audit finds only approximately 43% of those proofs faithful. We propose COVCAL, a selecto"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28365","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/2605.28365/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":"2605.28365","created_at":"2026-05-28T01:05:16.058724+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.28365v1","created_at":"2026-05-28T01:05:16.058724+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28365","created_at":"2026-05-28T01:05:16.058724+00:00"},{"alias_kind":"pith_short_12","alias_value":"WALU5PW2SN3K","created_at":"2026-05-28T01:05:16.058724+00:00"},{"alias_kind":"pith_short_16","alias_value":"WALU5PW2SN3KKH5H","created_at":"2026-05-28T01:05:16.058724+00:00"},{"alias_kind":"pith_short_8","alias_value":"WALU5PW2","created_at":"2026-05-28T01:05:16.058724+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ","json":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ.json","graph_json":"https://pith.science/api/pith-number/WALU5PW2SN3KKH5HLNHDSNJKTZ/graph.json","events_json":"https://pith.science/api/pith-number/WALU5PW2SN3KKH5HLNHDSNJKTZ/events.json","paper":"https://pith.science/paper/WALU5PW2"},"agent_actions":{"view_html":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ","download_json":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ.json","view_paper":"https://pith.science/paper/WALU5PW2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.28365&json=true","fetch_graph":"https://pith.science/api/pith-number/WALU5PW2SN3KKH5HLNHDSNJKTZ/graph.json","fetch_events":"https://pith.science/api/pith-number/WALU5PW2SN3KKH5HLNHDSNJKTZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ/action/storage_attestation","attest_author":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ/action/author_attestation","sign_citation":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ/action/citation_signature","submit_replication":"https://pith.science/pith/WALU5PW2SN3KKH5HLNHDSNJKTZ/action/replication_record"}},"created_at":"2026-05-28T01:05:16.058724+00:00","updated_at":"2026-05-28T01:05:16.058724+00:00"}