{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3PXB3R2ZY5KBLYI5MRDYREXU4H","short_pith_number":"pith:3PXB3R2Z","schema_version":"1.0","canonical_sha256":"dbee1dc759c75415e11d64478892f4e1f37ef2e83265c6ffafd302780323ef0e","source":{"kind":"arxiv","id":"2505.17454","version":1},"attestation_state":"computed","paper":{"title":"Self-Training Large Language Models with Confident Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Hyosoon Jang, Jungseul Ok, Sungjae Lee, Sungsoo Ahn, Yunhui Jang","submitted_at":"2025-05-23T04:25:10Z","abstract_excerpt":"Large language models (LLMs) have shown impressive performance by generating reasoning paths before final answers, but learning such a reasoning path requires costly human supervision. To address this issue, recent studies have explored self-training methods that improve reasoning capabilities using pseudo-labels generated by the LLMs themselves. Among these, confidence-based self-training fine-tunes LLMs to prefer reasoning paths with high-confidence answers, where confidence is estimated via majority voting. However, such methods exclusively focus on the quality of the final answer and may i"},"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":"2505.17454","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-23T04:25:10Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"e207338cf67a0b619718b73a4e53226c36ec313b5cb811aa6a833f9f0ab2836a","abstract_canon_sha256":"9846930a147ee1c6490c04bf169747c159a574faeaa7066fe83efe8659922c16"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:26.447547Z","signature_b64":"Knrg6TeemApnoU5pd64lv3LTWCgruVBLdjQ734u7hySdnY7zFO3uVtrHjSt7VYnrC0HXZtmnItr73X04k4QgDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dbee1dc759c75415e11d64478892f4e1f37ef2e83265c6ffafd302780323ef0e","last_reissued_at":"2026-07-05T11:08:26.447149Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:26.447149Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Training Large Language Models with Confident Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Hyosoon Jang, Jungseul Ok, Sungjae Lee, Sungsoo Ahn, Yunhui Jang","submitted_at":"2025-05-23T04:25:10Z","abstract_excerpt":"Large language models (LLMs) have shown impressive performance by generating reasoning paths before final answers, but learning such a reasoning path requires costly human supervision. To address this issue, recent studies have explored self-training methods that improve reasoning capabilities using pseudo-labels generated by the LLMs themselves. Among these, confidence-based self-training fine-tunes LLMs to prefer reasoning paths with high-confidence answers, where confidence is estimated via majority voting. However, such methods exclusively focus on the quality of the final answer and may i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17454","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/2505.17454/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":"2505.17454","created_at":"2026-07-05T11:08:26.447205+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17454v1","created_at":"2026-07-05T11:08:26.447205+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17454","created_at":"2026-07-05T11:08:26.447205+00:00"},{"alias_kind":"pith_short_12","alias_value":"3PXB3R2ZY5KB","created_at":"2026-07-05T11:08:26.447205+00:00"},{"alias_kind":"pith_short_16","alias_value":"3PXB3R2ZY5KBLYI5","created_at":"2026-07-05T11:08:26.447205+00:00"},{"alias_kind":"pith_short_8","alias_value":"3PXB3R2Z","created_at":"2026-07-05T11:08:26.447205+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.23183","citing_title":"ZeroSiam: An Efficient Asymmetry for Test-Time Entropy Optimization without Collapse","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H","json":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H.json","graph_json":"https://pith.science/api/pith-number/3PXB3R2ZY5KBLYI5MRDYREXU4H/graph.json","events_json":"https://pith.science/api/pith-number/3PXB3R2ZY5KBLYI5MRDYREXU4H/events.json","paper":"https://pith.science/paper/3PXB3R2Z"},"agent_actions":{"view_html":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H","download_json":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H.json","view_paper":"https://pith.science/paper/3PXB3R2Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17454&json=true","fetch_graph":"https://pith.science/api/pith-number/3PXB3R2ZY5KBLYI5MRDYREXU4H/graph.json","fetch_events":"https://pith.science/api/pith-number/3PXB3R2ZY5KBLYI5MRDYREXU4H/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H/action/storage_attestation","attest_author":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H/action/author_attestation","sign_citation":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H/action/citation_signature","submit_replication":"https://pith.science/pith/3PXB3R2ZY5KBLYI5MRDYREXU4H/action/replication_record"}},"created_at":"2026-07-05T11:08:26.447205+00:00","updated_at":"2026-07-05T11:08:26.447205+00:00"}