{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:7JXLSP7TIFWZU3RDWHFO545S65","short_pith_number":"pith:7JXLSP7T","schema_version":"1.0","canonical_sha256":"fa6eb93ff3416d9a6e23b1caeef3b2f77712642a2edd5cff2c40092acf34de5a","source":{"kind":"arxiv","id":"2311.08516","version":3},"attestation_state":"computed","paper":{"title":"LLMs cannot find reasoning errors, but can correct them given the error location","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Gladys Tyen, Hassan Mansoor, Peter Chen, Tony Mak, Victor C\\u{a}rbune","submitted_at":"2023-11-14T20:12:38Z","abstract_excerpt":"While self-correction has shown promise in improving LLM outputs in terms of style and quality (e.g. Chen et al., 2023b; Madaan et al., 2023), recent attempts to self-correct logical or reasoning errors often cause correct answers to become incorrect, resulting in worse performances overall (Huang et al., 2023). In this paper, we show that poor self-correction performance stems from LLMs' inability to find logical mistakes, rather than their ability to correct a known mistake. Firstly, we benchmark several state-of-the-art LLMs on their mistake-finding ability and demonstrate that they general"},"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":"2311.08516","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2023-11-14T20:12:38Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"19fd9dbc17d041370882cfef716d784c7e99358d21ec76a8fcc127750e147783","abstract_canon_sha256":"d311493af79df1487cd825cd225b15fc7e5d2e0c16f80c371e63461fd704c622"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:27:00.263535Z","signature_b64":"g2N81jwxX8mp+cDNuPxlAVIA52FiEvaq6KnkOWCTK/6qjn+kqfmG9y6K1uv7CYMrV1fPM+1cedqCnSBlCyATCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fa6eb93ff3416d9a6e23b1caeef3b2f77712642a2edd5cff2c40092acf34de5a","last_reissued_at":"2026-07-05T08:27:00.263012Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:27:00.263012Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLMs cannot find reasoning errors, but can correct them given the error location","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Gladys Tyen, Hassan Mansoor, Peter Chen, Tony Mak, Victor C\\u{a}rbune","submitted_at":"2023-11-14T20:12:38Z","abstract_excerpt":"While self-correction has shown promise in improving LLM outputs in terms of style and quality (e.g. Chen et al., 2023b; Madaan et al., 2023), recent attempts to self-correct logical or reasoning errors often cause correct answers to become incorrect, resulting in worse performances overall (Huang et al., 2023). In this paper, we show that poor self-correction performance stems from LLMs' inability to find logical mistakes, rather than their ability to correct a known mistake. Firstly, we benchmark several state-of-the-art LLMs on their mistake-finding ability and demonstrate that they general"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.08516","kind":"arxiv","version":3},"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/2311.08516/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":"2311.08516","created_at":"2026-07-05T08:27:00.263077+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.08516v3","created_at":"2026-07-05T08:27:00.263077+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.08516","created_at":"2026-07-05T08:27:00.263077+00:00"},{"alias_kind":"pith_short_12","alias_value":"7JXLSP7TIFWZ","created_at":"2026-07-05T08:27:00.263077+00:00"},{"alias_kind":"pith_short_16","alias_value":"7JXLSP7TIFWZU3RD","created_at":"2026-07-05T08:27:00.263077+00:00"},{"alias_kind":"pith_short_8","alias_value":"7JXLSP7T","created_at":"2026-07-05T08:27:00.263077+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09060","citing_title":"ATTAIN: Automated Exploit Failure Analysis through Trace-Driven Diff Analysis","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08068","citing_title":"DICE: Entropy-Regularized Equilibrium Selection for Stable Multi-Agent LLM Coordination","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27378","citing_title":"Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs","ref_index":76,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05976","citing_title":"The Self-Correction Illusion: LLMs Correct Others but Not Themselves","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2503.19786","citing_title":"Gemma 3 Technical Report","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19228","citing_title":"Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05579","citing_title":"LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods","ref_index":231,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06066","citing_title":"From Hallucination to Structure Snowballing: The Alignment Tax of Constrained Decoding in LLM Reflection","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2403.07974","citing_title":"LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code","ref_index":101,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65","json":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65.json","graph_json":"https://pith.science/api/pith-number/7JXLSP7TIFWZU3RDWHFO545S65/graph.json","events_json":"https://pith.science/api/pith-number/7JXLSP7TIFWZU3RDWHFO545S65/events.json","paper":"https://pith.science/paper/7JXLSP7T"},"agent_actions":{"view_html":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65","download_json":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65.json","view_paper":"https://pith.science/paper/7JXLSP7T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.08516&json=true","fetch_graph":"https://pith.science/api/pith-number/7JXLSP7TIFWZU3RDWHFO545S65/graph.json","fetch_events":"https://pith.science/api/pith-number/7JXLSP7TIFWZU3RDWHFO545S65/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65/action/storage_attestation","attest_author":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65/action/author_attestation","sign_citation":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65/action/citation_signature","submit_replication":"https://pith.science/pith/7JXLSP7TIFWZU3RDWHFO545S65/action/replication_record"}},"created_at":"2026-07-05T08:27:00.263077+00:00","updated_at":"2026-07-05T08:27:00.263077+00:00"}