{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:G4FO4KAWZGNZGA4L6O2OQWDU7R","short_pith_number":"pith:G4FO4KAW","schema_version":"1.0","canonical_sha256":"370aee2816c99b93038bf3b4e85874fc6f2ae5233def67253e05e80969af6693","source":{"kind":"arxiv","id":"2402.03744","version":2},"attestation_state":"computed","paper":{"title":"INSIDE: LLMs' Internal States Retain the Power of Hallucination Detection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chao Chen, Jieping Ye, Kai Liu, Mingyuan Tao, Yi Gu, Yue Wu, Ze Chen, Zhihang Fu","submitted_at":"2024-02-06T06:23:12Z","abstract_excerpt":"Knowledge hallucination have raised widespread concerns for the security and reliability of deployed LLMs. Previous efforts in detecting hallucinations have been employed at logit-level uncertainty estimation or language-level self-consistency evaluation, where the semantic information is inevitably lost during the token-decoding procedure. Thus, we propose to explore the dense semantic information retained within LLMs' \\textbf{IN}ternal \\textbf{S}tates for halluc\\textbf{I}nation \\textbf{DE}tection (\\textbf{INSIDE}). In particular, a simple yet effective \\textbf{EigenScore} metric is proposed "},"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":"2402.03744","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-06T06:23:12Z","cross_cats_sorted":[],"title_canon_sha256":"32a87b16bbcbb9c09ab6fd4ea09593e19f1ba2ef1b6b11ab2bbca261aa1d3586","abstract_canon_sha256":"2c4db05f6f55a8a586bd7dc5e522f48a3b194ee549535db0c181ab7beb4ed6ff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:23:02.832067Z","signature_b64":"uPn14oPRst3wFRZQavMIAJPIzAwby0Uf74MnE5JDSgem0qHIiaqJYOW9k+h/mU++TQ5fbR6ay6cyS14xrmMXCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"370aee2816c99b93038bf3b4e85874fc6f2ae5233def67253e05e80969af6693","last_reissued_at":"2026-07-05T09:23:02.831592Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:23:02.831592Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"INSIDE: LLMs' Internal States Retain the Power of Hallucination Detection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chao Chen, Jieping Ye, Kai Liu, Mingyuan Tao, Yi Gu, Yue Wu, Ze Chen, Zhihang Fu","submitted_at":"2024-02-06T06:23:12Z","abstract_excerpt":"Knowledge hallucination have raised widespread concerns for the security and reliability of deployed LLMs. Previous efforts in detecting hallucinations have been employed at logit-level uncertainty estimation or language-level self-consistency evaluation, where the semantic information is inevitably lost during the token-decoding procedure. Thus, we propose to explore the dense semantic information retained within LLMs' \\textbf{IN}ternal \\textbf{S}tates for halluc\\textbf{I}nation \\textbf{DE}tection (\\textbf{INSIDE}). In particular, a simple yet effective \\textbf{EigenScore} metric is proposed "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.03744","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/2402.03744/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":"2402.03744","created_at":"2026-07-05T09:23:02.831650+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.03744v2","created_at":"2026-07-05T09:23:02.831650+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.03744","created_at":"2026-07-05T09:23:02.831650+00:00"},{"alias_kind":"pith_short_12","alias_value":"G4FO4KAWZGNZ","created_at":"2026-07-05T09:23:02.831650+00:00"},{"alias_kind":"pith_short_16","alias_value":"G4FO4KAWZGNZGA4L","created_at":"2026-07-05T09:23:02.831650+00:00"},{"alias_kind":"pith_short_8","alias_value":"G4FO4KAW","created_at":"2026-07-05T09:23:02.831650+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":21,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08277","citing_title":"Remember with Confidence: Uncertainty Quantification for Spatio-temporal Memory with Probabilistic Guarantees","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03535","citing_title":"Can LLM Rerankers Predict Their Own Ranking Performance?","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24919","citing_title":"MultiHaluDet: Multilingual Hallucination Detection via LLM Hidden State Probing","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29545","citing_title":"AURORA: Asymmetry and Update-Induced Rotation for Robust Hallucination Detection in Large Language Models","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29605","citing_title":"VLAConf: Calibrated Task-Success Confidence for Vision-Language-Action Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28664","citing_title":"Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28264","citing_title":"Entropy Distribution as a Fingerprint for Hallucinations in Generative Models","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02628","citing_title":"Hallucination Is Linearly Decodable from Mid-Layer Hidden States in Quantized LLMs","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02443","citing_title":"HalluScan: A Systematic Benchmark for Detecting and Mitigating Hallucinations in Instruction-Following LLMs","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2505.11737","citing_title":"TokUR: Token-Level Uncertainty Estimation for Large Language Model Reasoning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2509.05489","citing_title":"Self-Aligned Reward: Towards Effective and Efficient Reasoners","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2509.17314","citing_title":"Clotho: Measuring Task-Specific Pre-Generation Test Adequacy for LLM Inputs","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2602.13224","citing_title":"A Geometric Taxonomy of Hallucinations in LLMs","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18901","citing_title":"Harmful Intent as a Geometrically Recoverable Feature of LLM Residual Streams","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05134","citing_title":"Low-Cost Black-Box Detection of LLM Hallucinations via Dynamical System Prediction","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19162","citing_title":"Mind the Unseen Mass: Unmasking LLM Hallucinations via Soft-Hybrid Alphabet Estimation","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04743","citing_title":"Hallucination Basins: A Dynamic Framework for Understanding and Controlling LLM Hallucinations","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05348","citing_title":"From Retinal Evidence to Safe Decisions: RETINA-SAFE and ECRT for Hallucination Risk Triage in Medical LLMs","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18901","citing_title":"Harmful Intent as a Geometrically Recoverable Feature of LLM Residual Streams","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15741","citing_title":"Learning Uncertainty from Sequential Internal Dispersion in Large Language Models","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15945","citing_title":"RAGognizer: Hallucination-Aware Fine-Tuning via Detection Head Integration","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R","json":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R.json","graph_json":"https://pith.science/api/pith-number/G4FO4KAWZGNZGA4L6O2OQWDU7R/graph.json","events_json":"https://pith.science/api/pith-number/G4FO4KAWZGNZGA4L6O2OQWDU7R/events.json","paper":"https://pith.science/paper/G4FO4KAW"},"agent_actions":{"view_html":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R","download_json":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R.json","view_paper":"https://pith.science/paper/G4FO4KAW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.03744&json=true","fetch_graph":"https://pith.science/api/pith-number/G4FO4KAWZGNZGA4L6O2OQWDU7R/graph.json","fetch_events":"https://pith.science/api/pith-number/G4FO4KAWZGNZGA4L6O2OQWDU7R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R/action/storage_attestation","attest_author":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R/action/author_attestation","sign_citation":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R/action/citation_signature","submit_replication":"https://pith.science/pith/G4FO4KAWZGNZGA4L6O2OQWDU7R/action/replication_record"}},"created_at":"2026-07-05T09:23:02.831650+00:00","updated_at":"2026-07-05T09:23:02.831650+00:00"}