{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QW7INYAXE2TBZ4E34UJLWNVCYQ","short_pith_number":"pith:QW7INYAX","schema_version":"1.0","canonical_sha256":"85be86e01726a61cf09be512bb36a2c415fefea38c1c13f3f3d63667464aed7d","source":{"kind":"arxiv","id":"2608.02957","version":1},"attestation_state":"computed","paper":{"title":"Inverted Detection and Control in Steering Vectors","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aria Masoomi, Jennifer Dy, Max Torop","submitted_at":"2026-08-03T23:42:24Z","abstract_excerpt":"Steering vectors (SVs) are widely used to influence the expression of concepts (e.g., truthfulness) in large language model outputs. A key assumption underpinning SVs is that they are linearly discriminative with respect to the concept: representations of texts that exhibit the concept are more aligned with the SV than those that do not, motivating shifts along the positive or negative SV direction to respectively promote or suppress the concept. In this work, we identify an inverted detection-control phenomenon in which some highly discriminative SVs that are aligned with positive representat"},"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":"2608.02957","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-08-03T23:42:24Z","cross_cats_sorted":[],"title_canon_sha256":"40e45415bb8f65e8a4a96083858f9bdb331001c46d3c7dc9a173b475f6439f31","abstract_canon_sha256":"7b89467ec85de90925c9dbdb5b7a3641ab9c057ff100c98aa7ae2865d9d893d3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-05T00:44:14.557376Z","signature_b64":"X9YNF9A6rAMoASNM39vogb5VB9h3H899akWtqrbV0Sz+C6JwILiHNdHsMBMZ0aKXhGHHvL02Dxr6/okBg3YOAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"85be86e01726a61cf09be512bb36a2c415fefea38c1c13f3f3d63667464aed7d","last_reissued_at":"2026-08-05T00:44:14.555082Z","signature_status":"signed_v1","first_computed_at":"2026-08-05T00:44:14.555082Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Inverted Detection and Control in Steering Vectors","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aria Masoomi, Jennifer Dy, Max Torop","submitted_at":"2026-08-03T23:42:24Z","abstract_excerpt":"Steering vectors (SVs) are widely used to influence the expression of concepts (e.g., truthfulness) in large language model outputs. A key assumption underpinning SVs is that they are linearly discriminative with respect to the concept: representations of texts that exhibit the concept are more aligned with the SV than those that do not, motivating shifts along the positive or negative SV direction to respectively promote or suppress the concept. In this work, we identify an inverted detection-control phenomenon in which some highly discriminative SVs that are aligned with positive representat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.02957","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/2608.02957/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":"2608.02957","created_at":"2026-08-05T00:44:14.555867+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.02957v1","created_at":"2026-08-05T00:44:14.555867+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.02957","created_at":"2026-08-05T00:44:14.555867+00:00"},{"alias_kind":"pith_short_12","alias_value":"QW7INYAXE2TB","created_at":"2026-08-05T00:44:14.555867+00:00"},{"alias_kind":"pith_short_16","alias_value":"QW7INYAXE2TBZ4E3","created_at":"2026-08-05T00:44:14.555867+00:00"},{"alias_kind":"pith_short_8","alias_value":"QW7INYAX","created_at":"2026-08-05T00:44:14.555867+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/QW7INYAXE2TBZ4E34UJLWNVCYQ","json":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ.json","graph_json":"https://pith.science/api/pith-number/QW7INYAXE2TBZ4E34UJLWNVCYQ/graph.json","events_json":"https://pith.science/api/pith-number/QW7INYAXE2TBZ4E34UJLWNVCYQ/events.json","paper":"https://pith.science/paper/QW7INYAX"},"agent_actions":{"view_html":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ","download_json":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ.json","view_paper":"https://pith.science/paper/QW7INYAX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.02957&json=true","fetch_graph":"https://pith.science/api/pith-number/QW7INYAXE2TBZ4E34UJLWNVCYQ/graph.json","fetch_events":"https://pith.science/api/pith-number/QW7INYAXE2TBZ4E34UJLWNVCYQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ/action/storage_attestation","attest_author":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ/action/author_attestation","sign_citation":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ/action/citation_signature","submit_replication":"https://pith.science/pith/QW7INYAXE2TBZ4E34UJLWNVCYQ/action/replication_record"}},"created_at":"2026-08-05T00:44:14.555867+00:00","updated_at":"2026-08-05T00:44:14.555867+00:00"}