{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6P44DDHVM5RTARRLKTHWFJ64PJ","short_pith_number":"pith:6P44DDHV","schema_version":"1.0","canonical_sha256":"f3f9c18cf5676330462b54cf62a7dc7a4fa1da807fa7448932d3ff8bb08aae78","source":{"kind":"arxiv","id":"2505.12886","version":1},"attestation_state":"computed","paper":{"title":"Detection and Mitigation of Hallucination in Large Reasoning Models: A Mechanistic Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CY"],"primary_cat":"cs.AI","authors_text":"Haoyu Wang, Jun Xu, Qipeng Wang, Xiao Zhang, Zhongxiang Sun","submitted_at":"2025-05-19T09:16:40Z","abstract_excerpt":"Large Reasoning Models (LRMs) have shown impressive capabilities in multi-step reasoning tasks. However, alongside these successes, a more deceptive form of model error has emerged--Reasoning Hallucination--where logically coherent but factually incorrect reasoning traces lead to persuasive yet faulty conclusions. Unlike traditional hallucinations, these errors are embedded within structured reasoning, making them more difficult to detect and potentially more harmful. In this work, we investigate reasoning hallucinations from a mechanistic perspective. We propose the Reasoning Score, which qua"},"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":"2505.12886","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-05-19T09:16:40Z","cross_cats_sorted":["cs.CL","cs.CY"],"title_canon_sha256":"b4a7e87cb0a70cad4beb668bf56685a7a5d2f8b5379b17d409f4f9b73df311a6","abstract_canon_sha256":"646efe2324264bca0735e6772508d2ac394c4196d32d0362e1faadaa4588af35"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:05:11.233522Z","signature_b64":"6zjL9pvdJ5WuAwiPA9IsBdoR7AivA9/XzxRHFVoxEN2y5kIl6sbM5qS2liKTkxiSRUppbGSqZwTy/G+tADY1Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3f9c18cf5676330462b54cf62a7dc7a4fa1da807fa7448932d3ff8bb08aae78","last_reissued_at":"2026-07-05T11:05:11.233016Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:05:11.233016Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Detection and Mitigation of Hallucination in Large Reasoning Models: A Mechanistic Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CY"],"primary_cat":"cs.AI","authors_text":"Haoyu Wang, Jun Xu, Qipeng Wang, Xiao Zhang, Zhongxiang Sun","submitted_at":"2025-05-19T09:16:40Z","abstract_excerpt":"Large Reasoning Models (LRMs) have shown impressive capabilities in multi-step reasoning tasks. However, alongside these successes, a more deceptive form of model error has emerged--Reasoning Hallucination--where logically coherent but factually incorrect reasoning traces lead to persuasive yet faulty conclusions. Unlike traditional hallucinations, these errors are embedded within structured reasoning, making them more difficult to detect and potentially more harmful. In this work, we investigate reasoning hallucinations from a mechanistic perspective. We propose the Reasoning Score, which qua"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.12886","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/2505.12886/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":"2505.12886","created_at":"2026-07-05T11:05:11.233078+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.12886v1","created_at":"2026-07-05T11:05:11.233078+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.12886","created_at":"2026-07-05T11:05:11.233078+00:00"},{"alias_kind":"pith_short_12","alias_value":"6P44DDHVM5RT","created_at":"2026-07-05T11:05:11.233078+00:00"},{"alias_kind":"pith_short_16","alias_value":"6P44DDHVM5RTARRL","created_at":"2026-07-05T11:05:11.233078+00:00"},{"alias_kind":"pith_short_8","alias_value":"6P44DDHV","created_at":"2026-07-05T11:05:11.233078+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05861","citing_title":"Mitigating Factual Hallucination in Large Reasoning Models via Mixed-Mode Advantage Regularization","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17028","citing_title":"PARALLAX: Separating Genuine Hallucination Detection from Benchmark Construction Artifacts","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2601.17467","citing_title":"Harnessing Reasoning Trajectories for Hallucination Detection via Answer-agreement Representation Shaping","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2603.25412","citing_title":"Beyond Content Safety: Real-Time Monitoring for Reasoning Vulnerabilities in Large Language Models","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ","json":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ.json","graph_json":"https://pith.science/api/pith-number/6P44DDHVM5RTARRLKTHWFJ64PJ/graph.json","events_json":"https://pith.science/api/pith-number/6P44DDHVM5RTARRLKTHWFJ64PJ/events.json","paper":"https://pith.science/paper/6P44DDHV"},"agent_actions":{"view_html":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ","download_json":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ.json","view_paper":"https://pith.science/paper/6P44DDHV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.12886&json=true","fetch_graph":"https://pith.science/api/pith-number/6P44DDHVM5RTARRLKTHWFJ64PJ/graph.json","fetch_events":"https://pith.science/api/pith-number/6P44DDHVM5RTARRLKTHWFJ64PJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ/action/storage_attestation","attest_author":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ/action/author_attestation","sign_citation":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ/action/citation_signature","submit_replication":"https://pith.science/pith/6P44DDHVM5RTARRLKTHWFJ64PJ/action/replication_record"}},"created_at":"2026-07-05T11:05:11.233078+00:00","updated_at":"2026-07-05T11:05:11.233078+00:00"}