{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:URSHQLWF2FHAUZYGP3ENMJGIYH","short_pith_number":"pith:URSHQLWF","schema_version":"1.0","canonical_sha256":"a464782ec5d14e0a67067ec8d624c8c1f42d441147f043aac9103c3f487d9603","source":{"kind":"arxiv","id":"2606.02628","version":1},"attestation_state":"computed","paper":{"title":"Hallucination Is Linearly Decodable from Mid-Layer Hidden States in Quantized LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Aizierjiang Aiersilan","submitted_at":"2026-05-30T02:04:30Z","abstract_excerpt":"We investigate whether open-source LLMs encode a linearly separable truthfulness signal in their hidden states, and at which network depth this signal is strongest. Across three $7$B--$8$B instruction-tuned models (Llama-3.1-8B, Mistral-7B, Qwen2.5-7B) loaded in $4$-bit NF4 quantization, we extract per-layer hidden states on four hallucination benchmarks (TruthfulQA, HaluEval-QA, FEVER, and a controlled synthetic set) and compare four detection approaches: linear and MLP probes, INSIDE EigenScore, self-consistency, and attention entropy. A linear probe on a single mid-network layer achieves $0"},"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":"2606.02628","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-30T02:04:30Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3ab277cd514831c0449ad6c791531e07baeae5861badbcff7c385d71e7490876","abstract_canon_sha256":"ad068affabba5241929c9a3421ed9e269270be0eab49af18b203d855b17c75a3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T00:05:05.095085Z","signature_b64":"P3vhU6YbusKmuu0U2YUyBdR3rb1l2B9ZQPvuNKIb1zu9UClANe9pRBL1T8NO+OitcYJqmMJ1p5vYURoLIn4SCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a464782ec5d14e0a67067ec8d624c8c1f42d441147f043aac9103c3f487d9603","last_reissued_at":"2026-06-03T00:05:05.094689Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T00:05:05.094689Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hallucination Is Linearly Decodable from Mid-Layer Hidden States in Quantized LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Aizierjiang Aiersilan","submitted_at":"2026-05-30T02:04:30Z","abstract_excerpt":"We investigate whether open-source LLMs encode a linearly separable truthfulness signal in their hidden states, and at which network depth this signal is strongest. Across three $7$B--$8$B instruction-tuned models (Llama-3.1-8B, Mistral-7B, Qwen2.5-7B) loaded in $4$-bit NF4 quantization, we extract per-layer hidden states on four hallucination benchmarks (TruthfulQA, HaluEval-QA, FEVER, and a controlled synthetic set) and compare four detection approaches: linear and MLP probes, INSIDE EigenScore, self-consistency, and attention entropy. A linear probe on a single mid-network layer achieves $0"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.02628","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/2606.02628/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":"2606.02628","created_at":"2026-06-03T00:05:05.094745+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.02628v1","created_at":"2026-06-03T00:05:05.094745+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.02628","created_at":"2026-06-03T00:05:05.094745+00:00"},{"alias_kind":"pith_short_12","alias_value":"URSHQLWF2FHA","created_at":"2026-06-03T00:05:05.094745+00:00"},{"alias_kind":"pith_short_16","alias_value":"URSHQLWF2FHAUZYG","created_at":"2026-06-03T00:05:05.094745+00:00"},{"alias_kind":"pith_short_8","alias_value":"URSHQLWF","created_at":"2026-06-03T00:05:05.094745+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH","json":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH.json","graph_json":"https://pith.science/api/pith-number/URSHQLWF2FHAUZYGP3ENMJGIYH/graph.json","events_json":"https://pith.science/api/pith-number/URSHQLWF2FHAUZYGP3ENMJGIYH/events.json","paper":"https://pith.science/paper/URSHQLWF"},"agent_actions":{"view_html":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH","download_json":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH.json","view_paper":"https://pith.science/paper/URSHQLWF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.02628&json=true","fetch_graph":"https://pith.science/api/pith-number/URSHQLWF2FHAUZYGP3ENMJGIYH/graph.json","fetch_events":"https://pith.science/api/pith-number/URSHQLWF2FHAUZYGP3ENMJGIYH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH/action/storage_attestation","attest_author":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH/action/author_attestation","sign_citation":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH/action/citation_signature","submit_replication":"https://pith.science/pith/URSHQLWF2FHAUZYGP3ENMJGIYH/action/replication_record"}},"created_at":"2026-06-03T00:05:05.094745+00:00","updated_at":"2026-06-03T00:05:05.094745+00:00"}