{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GPY3QQDI4ADGSHALB2IQJ2SPU2","short_pith_number":"pith:GPY3QQDI","schema_version":"1.0","canonical_sha256":"33f1b84068e006691c0b0e9104ea4fa6bb4c4c6cb7523d57a0122a8c7775bbfc","source":{"kind":"arxiv","id":"2404.17140","version":2},"attestation_state":"computed","paper":{"title":"Small Language Models Need Strong Verifiers to Self-Correct Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Honglak Lee, Jaekyeom Kim, Lajanugen Logeswaran, Lu Wang, Moontae Lee, Muhammad Khalifa, Yunxiang Zhang","submitted_at":"2024-04-26T03:41:28Z","abstract_excerpt":"Self-correction has emerged as a promising solution to boost the reasoning performance of large language models (LLMs), where LLMs refine their solutions using self-generated critiques that pinpoint the errors. This work explores whether small (<= 13B) language models (LMs) have the ability of self-correction on reasoning tasks with minimal inputs from stronger LMs. We propose a novel pipeline that prompts smaller LMs to collect self-correction data that supports the training of self-refinement abilities. First, we leverage correct solutions to guide the model in critiquing their incorrect res"},"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":"2404.17140","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-04-26T03:41:28Z","cross_cats_sorted":[],"title_canon_sha256":"4ff76ff0c6fbfa55ea55455555c09dc0e507af24b7156fa0eb50dc645a2e3645","abstract_canon_sha256":"13a6e7db5c7be5f8e821318141ccc2824fbf93cef53e8e123112eb1b8d615a71"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:28:09.526447Z","signature_b64":"yKFU+c9BxyZ/NRCqCbjrmB+eGYJGhFy9esRXIy+c1+mhA1JlflFqPUa3fxEaPfAxUCNzzoiHmlfGCcC7HL2LDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"33f1b84068e006691c0b0e9104ea4fa6bb4c4c6cb7523d57a0122a8c7775bbfc","last_reissued_at":"2026-07-05T08:28:09.526013Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:28:09.526013Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Small Language Models Need Strong Verifiers to Self-Correct Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Honglak Lee, Jaekyeom Kim, Lajanugen Logeswaran, Lu Wang, Moontae Lee, Muhammad Khalifa, Yunxiang Zhang","submitted_at":"2024-04-26T03:41:28Z","abstract_excerpt":"Self-correction has emerged as a promising solution to boost the reasoning performance of large language models (LLMs), where LLMs refine their solutions using self-generated critiques that pinpoint the errors. This work explores whether small (<= 13B) language models (LMs) have the ability of self-correction on reasoning tasks with minimal inputs from stronger LMs. We propose a novel pipeline that prompts smaller LMs to collect self-correction data that supports the training of self-refinement abilities. First, we leverage correct solutions to guide the model in critiquing their incorrect res"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.17140","kind":"arxiv","version":2},"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/2404.17140/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":"2404.17140","created_at":"2026-07-05T08:28:09.526074+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.17140v2","created_at":"2026-07-05T08:28:09.526074+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.17140","created_at":"2026-07-05T08:28:09.526074+00:00"},{"alias_kind":"pith_short_12","alias_value":"GPY3QQDI4ADG","created_at":"2026-07-05T08:28:09.526074+00:00"},{"alias_kind":"pith_short_16","alias_value":"GPY3QQDI4ADGSHAL","created_at":"2026-07-05T08:28:09.526074+00:00"},{"alias_kind":"pith_short_8","alias_value":"GPY3QQDI","created_at":"2026-07-05T08:28:09.526074+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21724","citing_title":"Denoising Iterative Self-Correction: Structured Verification Loops for Reliable LLM Reasoning","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2511.09907","citing_title":"Learning to Pose Problems: Reasoning-Driven and Solver-Adaptive Data Synthesis","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2409.12917","citing_title":"Training Language Models to Self-Correct via Reinforcement Learning","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2503.16419","citing_title":"Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models","ref_index":241,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06165","citing_title":"Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost","ref_index":98,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21268","citing_title":"Measure Twice, Click Once: Co-evolving Proposer and Visual Critic via Reinforcement Learning for GUI Grounding","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2","json":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2.json","graph_json":"https://pith.science/api/pith-number/GPY3QQDI4ADGSHALB2IQJ2SPU2/graph.json","events_json":"https://pith.science/api/pith-number/GPY3QQDI4ADGSHALB2IQJ2SPU2/events.json","paper":"https://pith.science/paper/GPY3QQDI"},"agent_actions":{"view_html":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2","download_json":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2.json","view_paper":"https://pith.science/paper/GPY3QQDI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.17140&json=true","fetch_graph":"https://pith.science/api/pith-number/GPY3QQDI4ADGSHALB2IQJ2SPU2/graph.json","fetch_events":"https://pith.science/api/pith-number/GPY3QQDI4ADGSHALB2IQJ2SPU2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2/action/storage_attestation","attest_author":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2/action/author_attestation","sign_citation":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2/action/citation_signature","submit_replication":"https://pith.science/pith/GPY3QQDI4ADGSHALB2IQJ2SPU2/action/replication_record"}},"created_at":"2026-07-05T08:28:09.526074+00:00","updated_at":"2026-07-05T08:28:09.526074+00:00"}