{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HBWF4XMTXRSQTWGD2FXIQGUBBK","short_pith_number":"pith:HBWF4XMT","schema_version":"1.0","canonical_sha256":"386c5e5d93bc6509d8c3d16e881a810a8a86c177df73f674df25011cd99a6b7e","source":{"kind":"arxiv","id":"2505.23540","version":1},"attestation_state":"computed","paper":{"title":"Probability-Consistent Preference Optimization for Enhanced LLM Reasoning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aojun Zhou, Hongsheng Li, Houxing Ren, Junting Pan, Ke Wang, Mingjie Zhan, Weikang Shi, Yunqiao Yang, Zimu Lu","submitted_at":"2025-05-29T15:20:44Z","abstract_excerpt":"Recent advances in preference optimization have demonstrated significant potential for improving mathematical reasoning capabilities in large language models (LLMs). While current approaches leverage high-quality pairwise preference data through outcome-based criteria like answer correctness or consistency, they fundamentally neglect the internal logical coherence of responses. To overcome this, we propose Probability-Consistent Preference Optimization (PCPO), a novel framework that establishes dual quantitative metrics for preference selection: (1) surface-level answer correctness and (2) int"},"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":"2505.23540","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-29T15:20:44Z","cross_cats_sorted":[],"title_canon_sha256":"c43e1c1f8527398265ce444c7497a4a7d68f4e2bfd9ea9456931b44d4b859b7a","abstract_canon_sha256":"9a0eba3433021fd67b5580adfb44f0e52ab7d03b684c2c3cef49f227f2b3ee59"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:12:03.652372Z","signature_b64":"KL1Qu0oh7mZxcohTpL19HjlQQCB8z2dCUiA90kdwc26E16Rt5IkQ01mzlVZo8v/l+SJBvXITdiRKB7NJVQxaBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"386c5e5d93bc6509d8c3d16e881a810a8a86c177df73f674df25011cd99a6b7e","last_reissued_at":"2026-07-05T11:12:03.651866Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:12:03.651866Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Probability-Consistent Preference Optimization for Enhanced LLM Reasoning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aojun Zhou, Hongsheng Li, Houxing Ren, Junting Pan, Ke Wang, Mingjie Zhan, Weikang Shi, Yunqiao Yang, Zimu Lu","submitted_at":"2025-05-29T15:20:44Z","abstract_excerpt":"Recent advances in preference optimization have demonstrated significant potential for improving mathematical reasoning capabilities in large language models (LLMs). While current approaches leverage high-quality pairwise preference data through outcome-based criteria like answer correctness or consistency, they fundamentally neglect the internal logical coherence of responses. To overcome this, we propose Probability-Consistent Preference Optimization (PCPO), a novel framework that establishes dual quantitative metrics for preference selection: (1) surface-level answer correctness and (2) int"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.23540","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/2505.23540/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":"2505.23540","created_at":"2026-07-05T11:12:03.651933+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.23540v1","created_at":"2026-07-05T11:12:03.651933+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.23540","created_at":"2026-07-05T11:12:03.651933+00:00"},{"alias_kind":"pith_short_12","alias_value":"HBWF4XMTXRSQ","created_at":"2026-07-05T11:12:03.651933+00:00"},{"alias_kind":"pith_short_16","alias_value":"HBWF4XMTXRSQTWGD","created_at":"2026-07-05T11:12:03.651933+00:00"},{"alias_kind":"pith_short_8","alias_value":"HBWF4XMT","created_at":"2026-07-05T11:12:03.651933+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.07353","citing_title":"Confidence-Aware Alignment Makes Reasoning LLMs More Reliable","ref_index":51,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK","json":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK.json","graph_json":"https://pith.science/api/pith-number/HBWF4XMTXRSQTWGD2FXIQGUBBK/graph.json","events_json":"https://pith.science/api/pith-number/HBWF4XMTXRSQTWGD2FXIQGUBBK/events.json","paper":"https://pith.science/paper/HBWF4XMT"},"agent_actions":{"view_html":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK","download_json":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK.json","view_paper":"https://pith.science/paper/HBWF4XMT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.23540&json=true","fetch_graph":"https://pith.science/api/pith-number/HBWF4XMTXRSQTWGD2FXIQGUBBK/graph.json","fetch_events":"https://pith.science/api/pith-number/HBWF4XMTXRSQTWGD2FXIQGUBBK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK/action/storage_attestation","attest_author":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK/action/author_attestation","sign_citation":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK/action/citation_signature","submit_replication":"https://pith.science/pith/HBWF4XMTXRSQTWGD2FXIQGUBBK/action/replication_record"}},"created_at":"2026-07-05T11:12:03.651933+00:00","updated_at":"2026-07-05T11:12:03.651933+00:00"}