{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QZ5WOFU4SMYCII5KPN6WUM3PRK","short_pith_number":"pith:QZ5WOFU4","schema_version":"1.0","canonical_sha256":"867b67169c93302423aa7b7d6a336f8a85c29b9cbfd55ba2129c965040d86231","source":{"kind":"arxiv","id":"2605.19436","version":1},"attestation_state":"computed","paper":{"title":"CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Abdelrahman M. Shaker, Ahmed Heakl, Fahad Shahbaz Khan, Omar Fetouh, Rania Elbadry, Salman Khan, Youssef Mohamed","submitted_at":"2026-05-19T06:46:19Z","abstract_excerpt":"When a model produces a correct solution under reinforcement learning with verifiable rewards (RLVR), every token receives the same reward signal regardless of whether it was a decisive reasoning step or a grammatical filler. A natural fix is to condition the model on the correct answer as a teacher, identifying tokens it would have generated differently had it known the answer. Prior work shows this either corrupts training by leaking the answer into the gradient, or produces a weak signal that cannot distinguish decisive steps from filler, since both look equally surprising relative to the m"},"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":"2605.19436","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-19T06:46:19Z","cross_cats_sorted":["cs.CL","cs.CV"],"title_canon_sha256":"ab233004695d1df74b4b3fb4f5044a1e8cbd66df1a1d998abb8a9dc92db56e4c","abstract_canon_sha256":"daa6d1692c4ea02489b34ce56a4fd82ffa071161d61c7c2a44ec8e507a20f2e7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:05:45.791885Z","signature_b64":"j77kXe25cCpsGbNu/aki34dTvvxpQ1ea6s9/mfEGFxCtN8IUvbqtRY9Z4WldcKlGwP8uCOCNkpM89EHZsGqWCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"867b67169c93302423aa7b7d6a336f8a85c29b9cbfd55ba2129c965040d86231","last_reissued_at":"2026-05-20T01:05:45.791306Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:05:45.791306Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Abdelrahman M. Shaker, Ahmed Heakl, Fahad Shahbaz Khan, Omar Fetouh, Rania Elbadry, Salman Khan, Youssef Mohamed","submitted_at":"2026-05-19T06:46:19Z","abstract_excerpt":"When a model produces a correct solution under reinforcement learning with verifiable rewards (RLVR), every token receives the same reward signal regardless of whether it was a decisive reasoning step or a grammatical filler. A natural fix is to condition the model on the correct answer as a teacher, identifying tokens it would have generated differently had it known the answer. Prior work shows this either corrupts training by leaking the answer into the gradient, or produces a weak signal that cannot distinguish decisive steps from filler, since both look equally surprising relative to the m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19436","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/2605.19436/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":"2605.19436","created_at":"2026-05-20T01:05:45.791404+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.19436v1","created_at":"2026-05-20T01:05:45.791404+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19436","created_at":"2026-05-20T01:05:45.791404+00:00"},{"alias_kind":"pith_short_12","alias_value":"QZ5WOFU4SMYC","created_at":"2026-05-20T01:05:45.791404+00:00"},{"alias_kind":"pith_short_16","alias_value":"QZ5WOFU4SMYCII5K","created_at":"2026-05-20T01:05:45.791404+00:00"},{"alias_kind":"pith_short_8","alias_value":"QZ5WOFU4","created_at":"2026-05-20T01:05:45.791404+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/QZ5WOFU4SMYCII5KPN6WUM3PRK","json":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK.json","graph_json":"https://pith.science/api/pith-number/QZ5WOFU4SMYCII5KPN6WUM3PRK/graph.json","events_json":"https://pith.science/api/pith-number/QZ5WOFU4SMYCII5KPN6WUM3PRK/events.json","paper":"https://pith.science/paper/QZ5WOFU4"},"agent_actions":{"view_html":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK","download_json":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK.json","view_paper":"https://pith.science/paper/QZ5WOFU4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.19436&json=true","fetch_graph":"https://pith.science/api/pith-number/QZ5WOFU4SMYCII5KPN6WUM3PRK/graph.json","fetch_events":"https://pith.science/api/pith-number/QZ5WOFU4SMYCII5KPN6WUM3PRK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK/action/storage_attestation","attest_author":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK/action/author_attestation","sign_citation":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK/action/citation_signature","submit_replication":"https://pith.science/pith/QZ5WOFU4SMYCII5KPN6WUM3PRK/action/replication_record"}},"created_at":"2026-05-20T01:05:45.791404+00:00","updated_at":"2026-05-20T01:05:45.791404+00:00"}