{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4NXQMYKOSAPBQUEQDEW6M763NC","short_pith_number":"pith:4NXQMYKO","schema_version":"1.0","canonical_sha256":"e36f06614e901e185090192de67fdb68a6329854da4aabb16e764d40894b3454","source":{"kind":"arxiv","id":"2507.22940","version":2},"attestation_state":"computed","paper":{"title":"Trustworthy Reasoning: Evaluating and Enhancing Factual Accuracy in LLM Intermediate Thought Processes","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jinku Li, Rui Jiao, Yue Zhang","submitted_at":"2025-07-25T10:34:51Z","abstract_excerpt":"We present a novel framework addressing a critical vulnerability in Large Language Models (LLMs): the prevalence of factual inaccuracies within intermediate reasoning steps despite correct final answers. This phenomenon poses substantial risks in high-stakes domains including healthcare, legal analysis, and scientific research, where erroneous yet confidently presented reasoning can mislead users into dangerous decisions. Our framework integrates three core components: (1) a specialized fact-checking classifier trained on counterfactually augmented data to detect subtle factual inconsistencies"},"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":"2507.22940","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-25T10:34:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"6b3a368b163192d9f0c0c912361259ca435e9ae3c383349afb51319ea5a2a46a","abstract_canon_sha256":"a613e95aa4f4bd71c67ac7d41f37b7d252915e9e915412fa49db6447382cceff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:47:27.503188Z","signature_b64":"QwhjROk0yd0U8Ml3BmtfgfZpWTzvGZOXwlfzqvbcoWGZ4jqJEKNpQL6yMr4RKuwr27ySUGO3gywdvLaPunstAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e36f06614e901e185090192de67fdb68a6329854da4aabb16e764d40894b3454","last_reissued_at":"2026-07-05T11:47:27.502661Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:47:27.502661Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Trustworthy Reasoning: Evaluating and Enhancing Factual Accuracy in LLM Intermediate Thought Processes","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jinku Li, Rui Jiao, Yue Zhang","submitted_at":"2025-07-25T10:34:51Z","abstract_excerpt":"We present a novel framework addressing a critical vulnerability in Large Language Models (LLMs): the prevalence of factual inaccuracies within intermediate reasoning steps despite correct final answers. This phenomenon poses substantial risks in high-stakes domains including healthcare, legal analysis, and scientific research, where erroneous yet confidently presented reasoning can mislead users into dangerous decisions. Our framework integrates three core components: (1) a specialized fact-checking classifier trained on counterfactually augmented data to detect subtle factual inconsistencies"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.22940","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/2507.22940/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":"2507.22940","created_at":"2026-07-05T11:47:27.502731+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.22940v2","created_at":"2026-07-05T11:47:27.502731+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.22940","created_at":"2026-07-05T11:47:27.502731+00:00"},{"alias_kind":"pith_short_12","alias_value":"4NXQMYKOSAPB","created_at":"2026-07-05T11:47:27.502731+00:00"},{"alias_kind":"pith_short_16","alias_value":"4NXQMYKOSAPBQUEQ","created_at":"2026-07-05T11:47:27.502731+00:00"},{"alias_kind":"pith_short_8","alias_value":"4NXQMYKO","created_at":"2026-07-05T11:47:27.502731+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02258","citing_title":"Matter to Mechanism: A Benchmark for AI Co-Scientists in Materials and Battery Research","ref_index":48,"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":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC","json":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC.json","graph_json":"https://pith.science/api/pith-number/4NXQMYKOSAPBQUEQDEW6M763NC/graph.json","events_json":"https://pith.science/api/pith-number/4NXQMYKOSAPBQUEQDEW6M763NC/events.json","paper":"https://pith.science/paper/4NXQMYKO"},"agent_actions":{"view_html":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC","download_json":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC.json","view_paper":"https://pith.science/paper/4NXQMYKO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.22940&json=true","fetch_graph":"https://pith.science/api/pith-number/4NXQMYKOSAPBQUEQDEW6M763NC/graph.json","fetch_events":"https://pith.science/api/pith-number/4NXQMYKOSAPBQUEQDEW6M763NC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC/action/storage_attestation","attest_author":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC/action/author_attestation","sign_citation":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC/action/citation_signature","submit_replication":"https://pith.science/pith/4NXQMYKOSAPBQUEQDEW6M763NC/action/replication_record"}},"created_at":"2026-07-05T11:47:27.502731+00:00","updated_at":"2026-07-05T11:47:27.502731+00:00"}