{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:FHTOVCWBR7GES7QMPXIC2BEHB7","short_pith_number":"pith:FHTOVCWB","canonical_record":{"source":{"id":"2507.23221","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-31T03:26:57Z","cross_cats_sorted":[],"title_canon_sha256":"11b15119a4c22c7935a0ad99c5bccceb3cc50766af1723cec021f3103a562bd1","abstract_canon_sha256":"5f9e6039c14cb06b38fec9575d18518777de0e16c825ef02dcd3eb8a785081a8"},"schema_version":"1.0"},"canonical_sha256":"29e6ea8ac18fcc497e0c7dd02d04870fcedf41b15a8aba07837d4e967a19f131","source":{"kind":"arxiv","id":"2507.23221","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.23221","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"arxiv_version","alias_value":"2507.23221v1","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.23221","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"pith_short_12","alias_value":"FHTOVCWBR7GE","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"pith_short_16","alias_value":"FHTOVCWBR7GES7QM","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"pith_short_8","alias_value":"FHTOVCWB","created_at":"2026-07-05T11:46:07Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:FHTOVCWBR7GES7QMPXIC2BEHB7","target":"record","payload":{"canonical_record":{"source":{"id":"2507.23221","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-31T03:26:57Z","cross_cats_sorted":[],"title_canon_sha256":"11b15119a4c22c7935a0ad99c5bccceb3cc50766af1723cec021f3103a562bd1","abstract_canon_sha256":"5f9e6039c14cb06b38fec9575d18518777de0e16c825ef02dcd3eb8a785081a8"},"schema_version":"1.0"},"canonical_sha256":"29e6ea8ac18fcc497e0c7dd02d04870fcedf41b15a8aba07837d4e967a19f131","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:46:07.799640Z","signature_b64":"wvWj908mXjIppk0hLCK7P4N+sTbYwHY0xFBEkpQZaIoA86feigWdTARi7VePR2s/5nBn5Ncn+4tkkNl0lZnLAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"29e6ea8ac18fcc497e0c7dd02d04870fcedf41b15a8aba07837d4e967a19f131","last_reissued_at":"2026-07-05T11:46:07.799189Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:46:07.799189Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2507.23221","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:46:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nISC+87KcWUupgF5s5ETJmOY4p6gPU43Gbc3+NXw6gbGVwEJVAhQjE1iOWS0XLoHIPiW/IZXOq4hCQKKSUArDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T11:29:50.239293Z"},"content_sha256":"83b2081b86636cbbd6a797013df3d0d5d42b53dc250f5823fabf9681412bce66","schema_version":"1.0","event_id":"sha256:83b2081b86636cbbd6a797013df3d0d5d42b53dc250f5823fabf9681412bce66"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:FHTOVCWBR7GES7QMPXIC2BEHB7","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"A Single Direction of Truth: An Observer Model's Linear Residual Probe Exposes and Steers Contextual Hallucinations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Charles O'Neill, Chi Chi Zhao, Max Kirkby, Mudith Jayasekara, Slava Chalnev","submitted_at":"2025-07-31T03:26:57Z","abstract_excerpt":"Contextual hallucinations -- statements unsupported by given context -- remain a significant challenge in AI. We demonstrate a practical interpretability insight: a generator-agnostic observer model detects hallucinations via a single forward pass and a linear probe on its residual stream. This probe isolates a single, transferable linear direction separating hallucinated from faithful text, outperforming baselines by 5-27 points and showing robust mid-layer performance across Gemma-2 models (2B to 27B). Gradient-times-activation localises this signal to sparse, late-layer MLP activity. Critic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.23221","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/2507.23221/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:46:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tt0lzqc1puPJWO++e0XNhxUNWyAIIblNYQ8g5lQQx4wJ0kgNAymRNQDvsuNYHY89NOvjWTYQlsmYArBZ+l9UAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T11:29:50.239691Z"},"content_sha256":"fb5bd6912ceb88e302a86d69b96839d0a55585a5d6be20d691b21cbf4e4015f0","schema_version":"1.0","event_id":"sha256:fb5bd6912ceb88e302a86d69b96839d0a55585a5d6be20d691b21cbf4e4015f0"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FHTOVCWBR7GES7QMPXIC2BEHB7/bundle.json","state_url":"https://pith.science/pith/FHTOVCWBR7GES7QMPXIC2BEHB7/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FHTOVCWBR7GES7QMPXIC2BEHB7/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-07T11:29:50Z","links":{"resolver":"https://pith.science/pith/FHTOVCWBR7GES7QMPXIC2BEHB7","bundle":"https://pith.science/pith/FHTOVCWBR7GES7QMPXIC2BEHB7/bundle.json","state":"https://pith.science/pith/FHTOVCWBR7GES7QMPXIC2BEHB7/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FHTOVCWBR7GES7QMPXIC2BEHB7/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:FHTOVCWBR7GES7QMPXIC2BEHB7","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"5f9e6039c14cb06b38fec9575d18518777de0e16c825ef02dcd3eb8a785081a8","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-31T03:26:57Z","title_canon_sha256":"11b15119a4c22c7935a0ad99c5bccceb3cc50766af1723cec021f3103a562bd1"},"schema_version":"1.0","source":{"id":"2507.23221","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.23221","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"arxiv_version","alias_value":"2507.23221v1","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.23221","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"pith_short_12","alias_value":"FHTOVCWBR7GE","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"pith_short_16","alias_value":"FHTOVCWBR7GES7QM","created_at":"2026-07-05T11:46:07Z"},{"alias_kind":"pith_short_8","alias_value":"FHTOVCWB","created_at":"2026-07-05T11:46:07Z"}],"graph_snapshots":[{"event_id":"sha256:fb5bd6912ceb88e302a86d69b96839d0a55585a5d6be20d691b21cbf4e4015f0","target":"graph","created_at":"2026-07-05T11:46:07Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2507.23221/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Contextual hallucinations -- statements unsupported by given context -- remain a significant challenge in AI. We demonstrate a practical interpretability insight: a generator-agnostic observer model detects hallucinations via a single forward pass and a linear probe on its residual stream. This probe isolates a single, transferable linear direction separating hallucinated from faithful text, outperforming baselines by 5-27 points and showing robust mid-layer performance across Gemma-2 models (2B to 27B). Gradient-times-activation localises this signal to sparse, late-layer MLP activity. Critic","authors_text":"Charles O'Neill, Chi Chi Zhao, Max Kirkby, Mudith Jayasekara, Slava Chalnev","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-31T03:26:57Z","title":"A Single Direction of Truth: An Observer Model's Linear Residual Probe Exposes and Steers Contextual Hallucinations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.23221","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:83b2081b86636cbbd6a797013df3d0d5d42b53dc250f5823fabf9681412bce66","target":"record","created_at":"2026-07-05T11:46:07Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"5f9e6039c14cb06b38fec9575d18518777de0e16c825ef02dcd3eb8a785081a8","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-31T03:26:57Z","title_canon_sha256":"11b15119a4c22c7935a0ad99c5bccceb3cc50766af1723cec021f3103a562bd1"},"schema_version":"1.0","source":{"id":"2507.23221","kind":"arxiv","version":1}},"canonical_sha256":"29e6ea8ac18fcc497e0c7dd02d04870fcedf41b15a8aba07837d4e967a19f131","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"29e6ea8ac18fcc497e0c7dd02d04870fcedf41b15a8aba07837d4e967a19f131","first_computed_at":"2026-07-05T11:46:07.799189Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:46:07.799189Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"wvWj908mXjIppk0hLCK7P4N+sTbYwHY0xFBEkpQZaIoA86feigWdTARi7VePR2s/5nBn5Ncn+4tkkNl0lZnLAA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:46:07.799640Z","signed_message":"canonical_sha256_bytes"},"source_id":"2507.23221","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:83b2081b86636cbbd6a797013df3d0d5d42b53dc250f5823fabf9681412bce66","sha256:fb5bd6912ceb88e302a86d69b96839d0a55585a5d6be20d691b21cbf4e4015f0"],"state_sha256":"293753d5e5d4428548d7705071abcd22fb1016f6c330de90b28324810e068f45"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mjmf4rPMcbGekikWkCTJc94yX4J4jUFV7nQE/1oUQ93crIN7yG4Z7hSfdLp6VmQW2I8A6ZpjfrdXp7gRXflXAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T11:29:50.242488Z","bundle_sha256":"fb98c00cf06b2420dc6ed05c9dadeceaec52bafce1315d270119485971543a02"}}