{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FTWCIISVQFO4V67U5IVD6HXBBT","short_pith_number":"pith:FTWCIISV","schema_version":"1.0","canonical_sha256":"2cec242255815dcafbf4ea2a3f1ee10ccd2096d4ba5dc8421ded7545db78bde9","source":{"kind":"arxiv","id":"2602.09805","version":2},"attestation_state":"computed","paper":{"title":"Beyond Accuracy: Decomposing the Reasoning Efficiency of LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ali Ramezani-Kebrya, Arnoldo Frigessi, Benjamin Ricaud, Daniel Kaiser","submitted_at":"2026-02-10T14:09:18Z","abstract_excerpt":"As reasoning LLMs increasingly trade tokens for accuracy through deliberation, search, and self-correction, a single accuracy score can no longer tell whether those tokens buy useful reasoning, recovery from hard instances, or unnecessary verbosity. We introduce a trace-optional evaluation protocol that exactly decomposes token efficiency using three observables available even for closed models: completion rate, conditional correctness given completion, and generated length. When instance-level workload metadata is available, we further normalize generated length by declared task-implied work "},"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":"2602.09805","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-02-10T14:09:18Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"cc0daca22f1f586ae53326ef8a843d884695b4fb59d472a843fd10f4ad82e61e","abstract_canon_sha256":"02d2eb5c50362a01fbde10ae09994f29570c6d0f9c0eec65dfdb21ad01a61e6d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:05:41.928882Z","signature_b64":"6HzVdxTsGfpD41iCMBpw7ASJZM4QEv58SiKP88SBddEOhFoTnnHCzrTV0Wwz8K+D2lZYARrhp3CmLsRdrhsCAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2cec242255815dcafbf4ea2a3f1ee10ccd2096d4ba5dc8421ded7545db78bde9","last_reissued_at":"2026-05-20T00:05:41.928426Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:05:41.928426Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Accuracy: Decomposing the Reasoning Efficiency of LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ali Ramezani-Kebrya, Arnoldo Frigessi, Benjamin Ricaud, Daniel Kaiser","submitted_at":"2026-02-10T14:09:18Z","abstract_excerpt":"As reasoning LLMs increasingly trade tokens for accuracy through deliberation, search, and self-correction, a single accuracy score can no longer tell whether those tokens buy useful reasoning, recovery from hard instances, or unnecessary verbosity. We introduce a trace-optional evaluation protocol that exactly decomposes token efficiency using three observables available even for closed models: completion rate, conditional correctness given completion, and generated length. When instance-level workload metadata is available, we further normalize generated length by declared task-implied work "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.09805","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/2602.09805/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":"2602.09805","created_at":"2026-05-20T00:05:41.928492+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.09805v2","created_at":"2026-05-20T00:05:41.928492+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.09805","created_at":"2026-05-20T00:05:41.928492+00:00"},{"alias_kind":"pith_short_12","alias_value":"FTWCIISVQFO4","created_at":"2026-05-20T00:05:41.928492+00:00"},{"alias_kind":"pith_short_16","alias_value":"FTWCIISVQFO4V67U","created_at":"2026-05-20T00:05:41.928492+00:00"},{"alias_kind":"pith_short_8","alias_value":"FTWCIISV","created_at":"2026-05-20T00:05:41.928492+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/FTWCIISVQFO4V67U5IVD6HXBBT","json":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT.json","graph_json":"https://pith.science/api/pith-number/FTWCIISVQFO4V67U5IVD6HXBBT/graph.json","events_json":"https://pith.science/api/pith-number/FTWCIISVQFO4V67U5IVD6HXBBT/events.json","paper":"https://pith.science/paper/FTWCIISV"},"agent_actions":{"view_html":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT","download_json":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT.json","view_paper":"https://pith.science/paper/FTWCIISV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.09805&json=true","fetch_graph":"https://pith.science/api/pith-number/FTWCIISVQFO4V67U5IVD6HXBBT/graph.json","fetch_events":"https://pith.science/api/pith-number/FTWCIISVQFO4V67U5IVD6HXBBT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT/action/storage_attestation","attest_author":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT/action/author_attestation","sign_citation":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT/action/citation_signature","submit_replication":"https://pith.science/pith/FTWCIISVQFO4V67U5IVD6HXBBT/action/replication_record"}},"created_at":"2026-05-20T00:05:41.928492+00:00","updated_at":"2026-05-20T00:05:41.928492+00:00"}