{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:P3TOMQJPZ5OM5H6MMTSHBNBYYP","short_pith_number":"pith:P3TOMQJP","schema_version":"1.0","canonical_sha256":"7ee6e6412fcf5cce9fcc64e470b438c3ce7c630e6f831bdc5c35fc40efda53fd","source":{"kind":"arxiv","id":"2505.17656","version":3},"attestation_state":"computed","paper":{"title":"Too Consistent to Detect: A Study of Self-Consistent Errors in LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Du Su, Fei Sun, Hexiang Tan, Huawei Shen, Jingang Wang, Qi Cao, Sha Liu, Xin Chen, Xueqi Cheng, Xunliang Cai, Yuanzhuo Wang","submitted_at":"2025-05-23T09:18:56Z","abstract_excerpt":"As large language models (LLMs) often generate plausible but incorrect content, error detection has become increasingly critical to ensure truthfulness. However, existing detection methods often overlook a critical problem we term as self-consistent error, where LLMs repeatedly generate the same incorrect response across multiple stochastic samples. This work formally defines self-consistent errors and evaluates mainstream detection methods on them. Our investigation reveals two key findings: (1) Unlike inconsistent errors, whose frequency diminishes significantly as the LLM scale increases, t"},"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":"2505.17656","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-23T09:18:56Z","cross_cats_sorted":[],"title_canon_sha256":"a52a797fc8897da23a068b22cef1c6340e0e6d307454f8139694f345b1f50ff8","abstract_canon_sha256":"dc6afcaa6cf39c988aab47979ab201817f89c4e775b9505bac4a1ef6597394a2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:06:12.951559Z","signature_b64":"BJ8YfmY91PZwf3p21IrCYP5mD7P6Ma+mKc48hGA/F709qM2bJDHphRP8dZA8BZ4eS+IUIj5aJI9KiNHvPel1CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7ee6e6412fcf5cce9fcc64e470b438c3ce7c630e6f831bdc5c35fc40efda53fd","last_reissued_at":"2026-07-05T12:06:12.951023Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:06:12.951023Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Too Consistent to Detect: A Study of Self-Consistent Errors in LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Du Su, Fei Sun, Hexiang Tan, Huawei Shen, Jingang Wang, Qi Cao, Sha Liu, Xin Chen, Xueqi Cheng, Xunliang Cai, Yuanzhuo Wang","submitted_at":"2025-05-23T09:18:56Z","abstract_excerpt":"As large language models (LLMs) often generate plausible but incorrect content, error detection has become increasingly critical to ensure truthfulness. However, existing detection methods often overlook a critical problem we term as self-consistent error, where LLMs repeatedly generate the same incorrect response across multiple stochastic samples. This work formally defines self-consistent errors and evaluates mainstream detection methods on them. Our investigation reveals two key findings: (1) Unlike inconsistent errors, whose frequency diminishes significantly as the LLM scale increases, t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17656","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/2505.17656/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":"2505.17656","created_at":"2026-07-05T12:06:12.951084+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17656v3","created_at":"2026-07-05T12:06:12.951084+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17656","created_at":"2026-07-05T12:06:12.951084+00:00"},{"alias_kind":"pith_short_12","alias_value":"P3TOMQJPZ5OM","created_at":"2026-07-05T12:06:12.951084+00:00"},{"alias_kind":"pith_short_16","alias_value":"P3TOMQJPZ5OM5H6M","created_at":"2026-07-05T12:06:12.951084+00:00"},{"alias_kind":"pith_short_8","alias_value":"P3TOMQJP","created_at":"2026-07-05T12:06:12.951084+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.03535","citing_title":"Can LLM Rerankers Predict Their Own Ranking Performance?","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2509.09438","citing_title":"GrACE: A Generative Approach to Better Confidence Elicitation and Efficient Test-Time Scaling in Large Language Models","ref_index":59,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP","json":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP.json","graph_json":"https://pith.science/api/pith-number/P3TOMQJPZ5OM5H6MMTSHBNBYYP/graph.json","events_json":"https://pith.science/api/pith-number/P3TOMQJPZ5OM5H6MMTSHBNBYYP/events.json","paper":"https://pith.science/paper/P3TOMQJP"},"agent_actions":{"view_html":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP","download_json":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP.json","view_paper":"https://pith.science/paper/P3TOMQJP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17656&json=true","fetch_graph":"https://pith.science/api/pith-number/P3TOMQJPZ5OM5H6MMTSHBNBYYP/graph.json","fetch_events":"https://pith.science/api/pith-number/P3TOMQJPZ5OM5H6MMTSHBNBYYP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP/action/storage_attestation","attest_author":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP/action/author_attestation","sign_citation":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP/action/citation_signature","submit_replication":"https://pith.science/pith/P3TOMQJPZ5OM5H6MMTSHBNBYYP/action/replication_record"}},"created_at":"2026-07-05T12:06:12.951084+00:00","updated_at":"2026-07-05T12:06:12.951084+00:00"}