{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:75O3ODZLFB3DF7ZYYBMOXBJOAT","short_pith_number":"pith:75O3ODZL","schema_version":"1.0","canonical_sha256":"ff5db70f2b287632ff38c058eb852e04f18d6da20043ddb5098e580e757cdf47","source":{"kind":"arxiv","id":"2607.13753","version":1},"attestation_state":"computed","paper":{"title":"Post-Training Shifts Confidence: A Three-Stage Analysis of How SFT, RL, and OPD Shape Pre-, Intra-, and Post-CoT Calibration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anhao Zhao, Guodong Du, Shuhao Li, Tianyu Yuan, Wanyu Lin, Xiaoyu Shen","submitted_at":"2026-07-15T12:12:37Z","abstract_excerpt":"Large language models have made strong reasoning gains through supervised fine-tuning, reinforcement learning, and on-policy distillation, yet these post-training methods are usually evaluated only by final-answer accuracy. We study how they reshape confidence during reasoning. We introduce a three-stage calibration framework that evaluates confidence before, during, and after chain-of-thought generation, corresponding to difficulty estimation, early termination, and answer aggregation. Through a controlled comparison on mathematical reasoning benchmarks, we find that OPD provides the most use"},"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":"2607.13753","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-15T12:12:37Z","cross_cats_sorted":[],"title_canon_sha256":"79e566e8993b52e6e594a800df820605b565b14300fdc30b4ad3341ee01b4571","abstract_canon_sha256":"0a5424fc7e33c507184ec7b9bc936624f568c9ef5ad2425d5596f29b2083a869"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-16T01:23:04.297351Z","signature_b64":"HzBnC8RBNQY40uhykMnbXzNwPuRINEN5vC3FPCnwATbyw73N4en5h6/Sn9/1w/afrEOUPmRgQr9Xor+HyYNkAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff5db70f2b287632ff38c058eb852e04f18d6da20043ddb5098e580e757cdf47","last_reissued_at":"2026-07-16T01:23:04.296005Z","signature_status":"signed_v1","first_computed_at":"2026-07-16T01:23:04.296005Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Post-Training Shifts Confidence: A Three-Stage Analysis of How SFT, RL, and OPD Shape Pre-, Intra-, and Post-CoT Calibration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anhao Zhao, Guodong Du, Shuhao Li, Tianyu Yuan, Wanyu Lin, Xiaoyu Shen","submitted_at":"2026-07-15T12:12:37Z","abstract_excerpt":"Large language models have made strong reasoning gains through supervised fine-tuning, reinforcement learning, and on-policy distillation, yet these post-training methods are usually evaluated only by final-answer accuracy. We study how they reshape confidence during reasoning. We introduce a three-stage calibration framework that evaluates confidence before, during, and after chain-of-thought generation, corresponding to difficulty estimation, early termination, and answer aggregation. Through a controlled comparison on mathematical reasoning benchmarks, we find that OPD provides the most use"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.13753","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/2607.13753/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":"2607.13753","created_at":"2026-07-16T01:23:04.296569+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.13753v1","created_at":"2026-07-16T01:23:04.296569+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.13753","created_at":"2026-07-16T01:23:04.296569+00:00"},{"alias_kind":"pith_short_12","alias_value":"75O3ODZLFB3D","created_at":"2026-07-16T01:23:04.296569+00:00"},{"alias_kind":"pith_short_16","alias_value":"75O3ODZLFB3DF7ZY","created_at":"2026-07-16T01:23:04.296569+00:00"},{"alias_kind":"pith_short_8","alias_value":"75O3ODZL","created_at":"2026-07-16T01:23:04.296569+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/75O3ODZLFB3DF7ZYYBMOXBJOAT","json":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT.json","graph_json":"https://pith.science/api/pith-number/75O3ODZLFB3DF7ZYYBMOXBJOAT/graph.json","events_json":"https://pith.science/api/pith-number/75O3ODZLFB3DF7ZYYBMOXBJOAT/events.json","paper":"https://pith.science/paper/75O3ODZL"},"agent_actions":{"view_html":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT","download_json":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT.json","view_paper":"https://pith.science/paper/75O3ODZL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.13753&json=true","fetch_graph":"https://pith.science/api/pith-number/75O3ODZLFB3DF7ZYYBMOXBJOAT/graph.json","fetch_events":"https://pith.science/api/pith-number/75O3ODZLFB3DF7ZYYBMOXBJOAT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT/action/storage_attestation","attest_author":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT/action/author_attestation","sign_citation":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT/action/citation_signature","submit_replication":"https://pith.science/pith/75O3ODZLFB3DF7ZYYBMOXBJOAT/action/replication_record"}},"created_at":"2026-07-16T01:23:04.296569+00:00","updated_at":"2026-07-16T01:23:04.296569+00:00"}