{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MPAELKDIAUAURJ2AU6VXPYECFQ","short_pith_number":"pith:MPAELKDI","schema_version":"1.0","canonical_sha256":"63c045a868050148a740a7ab77e0822c38d4feafb6e38b761458dac2b4020480","source":{"kind":"arxiv","id":"2411.06387","version":4},"attestation_state":"computed","paper":{"title":"Self-Training Meets Consistency: Improving LLMs' Reasoning with Consistency-Driven Rationale Evaluation","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Jaehyeok Lee, JinYeong Bak, Keisuke Sakaguchi","submitted_at":"2024-11-10T08:11:05Z","abstract_excerpt":"Self-training approach for large language models (LLMs) improves reasoning abilities by training the models on their self-generated rationales. Previous approaches have labeled rationales that produce correct answers for a given question as appropriate for training. However, a single measure risks misjudging rationale quality, leading the models to learn flawed reasoning patterns. To address this issue, we propose CREST (Consistency-driven Rationale Evaluation for Self-Training), a self-training framework that further evaluates each rationale through follow-up questions and leverages this eval"},"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":"2411.06387","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-11-10T08:11:05Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"88d0a2f75aed19c6186437e01eb6a4629fe56dc414f4865a8e1a0d7f48e7298e","abstract_canon_sha256":"e5237d915eecc3cd88507c178763ea81bc3b241b86bf78a722a2ee67466f7d80"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:10:09.707778Z","signature_b64":"ftXK/0HWc7Q03mV0h2omel18lOljzehFPXruRiZKrTwbJ/5O2Psq7Ufkyv0yb1mAI7AcT+rM6sRAkROxFf/YDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"63c045a868050148a740a7ab77e0822c38d4feafb6e38b761458dac2b4020480","last_reissued_at":"2026-07-05T10:10:09.707346Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:10:09.707346Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Training Meets Consistency: Improving LLMs' Reasoning with Consistency-Driven Rationale Evaluation","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Jaehyeok Lee, JinYeong Bak, Keisuke Sakaguchi","submitted_at":"2024-11-10T08:11:05Z","abstract_excerpt":"Self-training approach for large language models (LLMs) improves reasoning abilities by training the models on their self-generated rationales. Previous approaches have labeled rationales that produce correct answers for a given question as appropriate for training. However, a single measure risks misjudging rationale quality, leading the models to learn flawed reasoning patterns. To address this issue, we propose CREST (Consistency-driven Rationale Evaluation for Self-Training), a self-training framework that further evaluates each rationale through follow-up questions and leverages this eval"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.06387","kind":"arxiv","version":4},"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/2411.06387/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":"2411.06387","created_at":"2026-07-05T10:10:09.707417+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.06387v4","created_at":"2026-07-05T10:10:09.707417+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.06387","created_at":"2026-07-05T10:10:09.707417+00:00"},{"alias_kind":"pith_short_12","alias_value":"MPAELKDIAUAU","created_at":"2026-07-05T10:10:09.707417+00:00"},{"alias_kind":"pith_short_16","alias_value":"MPAELKDIAUAURJ2A","created_at":"2026-07-05T10:10:09.707417+00:00"},{"alias_kind":"pith_short_8","alias_value":"MPAELKDI","created_at":"2026-07-05T10:10:09.707417+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.18354","citing_title":"PRISMA: Preference-Reinforced Self-Training Approach for Interpretable Emotionally Intelligent Negotiation Dialogues","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ","json":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ.json","graph_json":"https://pith.science/api/pith-number/MPAELKDIAUAURJ2AU6VXPYECFQ/graph.json","events_json":"https://pith.science/api/pith-number/MPAELKDIAUAURJ2AU6VXPYECFQ/events.json","paper":"https://pith.science/paper/MPAELKDI"},"agent_actions":{"view_html":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ","download_json":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ.json","view_paper":"https://pith.science/paper/MPAELKDI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.06387&json=true","fetch_graph":"https://pith.science/api/pith-number/MPAELKDIAUAURJ2AU6VXPYECFQ/graph.json","fetch_events":"https://pith.science/api/pith-number/MPAELKDIAUAURJ2AU6VXPYECFQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ/action/storage_attestation","attest_author":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ/action/author_attestation","sign_citation":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ/action/citation_signature","submit_replication":"https://pith.science/pith/MPAELKDIAUAURJ2AU6VXPYECFQ/action/replication_record"}},"created_at":"2026-07-05T10:10:09.707417+00:00","updated_at":"2026-07-05T10:10:09.707417+00:00"}