{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:STOKPVAD7EPF5LCG4CEXFEECQP","short_pith_number":"pith:STOKPVAD","canonical_record":{"source":{"id":"2505.22637","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T17:53:31Z","cross_cats_sorted":[],"title_canon_sha256":"f9c4670352a5b4ade7de80ff884fa804eb08622df06f11933389fb1ccca42837","abstract_canon_sha256":"93c11547f668287b36be561896abd19feaf06a51e2d25a4f138edb9b7077f7da"},"schema_version":"1.0"},"canonical_sha256":"94dca7d403f91e5eac46e08972908283dc73010aff5e7c5182cc4c6351baf8df","source":{"kind":"arxiv","id":"2505.22637","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.22637","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"arxiv_version","alias_value":"2505.22637v1","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.22637","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"pith_short_12","alias_value":"STOKPVAD7EPF","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"pith_short_16","alias_value":"STOKPVAD7EPF5LCG","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"pith_short_8","alias_value":"STOKPVAD","created_at":"2026-07-05T11:11:28Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:STOKPVAD7EPF5LCG4CEXFEECQP","target":"record","payload":{"canonical_record":{"source":{"id":"2505.22637","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T17:53:31Z","cross_cats_sorted":[],"title_canon_sha256":"f9c4670352a5b4ade7de80ff884fa804eb08622df06f11933389fb1ccca42837","abstract_canon_sha256":"93c11547f668287b36be561896abd19feaf06a51e2d25a4f138edb9b7077f7da"},"schema_version":"1.0"},"canonical_sha256":"94dca7d403f91e5eac46e08972908283dc73010aff5e7c5182cc4c6351baf8df","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:11:28.204342Z","signature_b64":"ehqN0z0mOZrPG1JaYMT9wLSSED0dbCHEddNgx1vS5OqbWygwl5oqYmH0u6aQZ4ce0aSjOu8/WbsrLGoCWpQ/BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"94dca7d403f91e5eac46e08972908283dc73010aff5e7c5182cc4c6351baf8df","last_reissued_at":"2026-07-05T11:11:28.203783Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:11:28.203783Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2505.22637","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":"2026-07-05T11:11:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ngUUIsAJk7lNEz3cNjClpoAjvSfuNzacHaL7/XOfg+EWsdFmQxFfp/m7Fsi84bPoRuBXfq/m50t58z/AkD95Cw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T08:54:56.645559Z"},"content_sha256":"de15c743ac33ca2c35fb95816f193f054688dec3fb7dab4fbd03251d0c8221ec","schema_version":"1.0","event_id":"sha256:de15c743ac33ca2c35fb95816f193f054688dec3fb7dab4fbd03251d0c8221ec"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:STOKPVAD7EPF5LCG4CEXFEECQP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Understanding (Un)Reliability of Steering Vectors in Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Carsten Eickhoff, David Krueger, Dmitrii Krasheninnikov, Joschka Braun, Seyed Ali Bahrainian","submitted_at":"2025-05-28T17:53:31Z","abstract_excerpt":"Steering vectors are a lightweight method to control language model behavior by adding a learned bias to the activations at inference time. Although steering demonstrates promising performance, recent work shows that it can be unreliable or even counterproductive in some cases. This paper studies the influence of prompt types and the geometry of activation differences on steering reliability. First, we find that all seven prompt types used in our experiments produce a net positive steering effect, but exhibit high variance across samples, and often give an effect opposite of the desired one. N"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.22637","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/2505.22637/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":"2026-07-05T11:11:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Qgf/McHA+MEC1EyDfz08slWE5PVVoPVNqduo1FDh2fsNumuL0ijlL+OSgEuphnXRAxdfpGIKmhyIm48UIT3SAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T08:54:56.646065Z"},"content_sha256":"b4dc72d7e88fb440973f4732c7c5ab456900c4b5a79deb2948c8a6cd08716bd0","schema_version":"1.0","event_id":"sha256:b4dc72d7e88fb440973f4732c7c5ab456900c4b5a79deb2948c8a6cd08716bd0"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/STOKPVAD7EPF5LCG4CEXFEECQP/bundle.json","state_url":"https://pith.science/pith/STOKPVAD7EPF5LCG4CEXFEECQP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/STOKPVAD7EPF5LCG4CEXFEECQP/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":"2026-08-03T08:54:56Z","links":{"resolver":"https://pith.science/pith/STOKPVAD7EPF5LCG4CEXFEECQP","bundle":"https://pith.science/pith/STOKPVAD7EPF5LCG4CEXFEECQP/bundle.json","state":"https://pith.science/pith/STOKPVAD7EPF5LCG4CEXFEECQP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/STOKPVAD7EPF5LCG4CEXFEECQP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:STOKPVAD7EPF5LCG4CEXFEECQP","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":"93c11547f668287b36be561896abd19feaf06a51e2d25a4f138edb9b7077f7da","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T17:53:31Z","title_canon_sha256":"f9c4670352a5b4ade7de80ff884fa804eb08622df06f11933389fb1ccca42837"},"schema_version":"1.0","source":{"id":"2505.22637","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.22637","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"arxiv_version","alias_value":"2505.22637v1","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.22637","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"pith_short_12","alias_value":"STOKPVAD7EPF","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"pith_short_16","alias_value":"STOKPVAD7EPF5LCG","created_at":"2026-07-05T11:11:28Z"},{"alias_kind":"pith_short_8","alias_value":"STOKPVAD","created_at":"2026-07-05T11:11:28Z"}],"graph_snapshots":[{"event_id":"sha256:b4dc72d7e88fb440973f4732c7c5ab456900c4b5a79deb2948c8a6cd08716bd0","target":"graph","created_at":"2026-07-05T11:11:28Z","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/2505.22637/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Steering vectors are a lightweight method to control language model behavior by adding a learned bias to the activations at inference time. Although steering demonstrates promising performance, recent work shows that it can be unreliable or even counterproductive in some cases. This paper studies the influence of prompt types and the geometry of activation differences on steering reliability. First, we find that all seven prompt types used in our experiments produce a net positive steering effect, but exhibit high variance across samples, and often give an effect opposite of the desired one. N","authors_text":"Carsten Eickhoff, David Krueger, Dmitrii Krasheninnikov, Joschka Braun, Seyed Ali Bahrainian","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T17:53:31Z","title":"Understanding (Un)Reliability of Steering Vectors in Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.22637","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:de15c743ac33ca2c35fb95816f193f054688dec3fb7dab4fbd03251d0c8221ec","target":"record","created_at":"2026-07-05T11:11:28Z","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":"93c11547f668287b36be561896abd19feaf06a51e2d25a4f138edb9b7077f7da","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-28T17:53:31Z","title_canon_sha256":"f9c4670352a5b4ade7de80ff884fa804eb08622df06f11933389fb1ccca42837"},"schema_version":"1.0","source":{"id":"2505.22637","kind":"arxiv","version":1}},"canonical_sha256":"94dca7d403f91e5eac46e08972908283dc73010aff5e7c5182cc4c6351baf8df","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"94dca7d403f91e5eac46e08972908283dc73010aff5e7c5182cc4c6351baf8df","first_computed_at":"2026-07-05T11:11:28.203783Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:11:28.203783Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ehqN0z0mOZrPG1JaYMT9wLSSED0dbCHEddNgx1vS5OqbWygwl5oqYmH0u6aQZ4ce0aSjOu8/WbsrLGoCWpQ/BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:11:28.204342Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.22637","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:de15c743ac33ca2c35fb95816f193f054688dec3fb7dab4fbd03251d0c8221ec","sha256:b4dc72d7e88fb440973f4732c7c5ab456900c4b5a79deb2948c8a6cd08716bd0"],"state_sha256":"90307ae3fdf35ec91eab24c55a5343c50984205cb0a8ef4505219acbf59fcb5c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"o/vmWecFgUFxD7t9cs5pYL+S/HkGun1SJYr8MjMzmKIiOdcBJf+DuWpttLUBOLxgeuBnZ30b5ECwyblcWFTpAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-03T08:54:56.650190Z","bundle_sha256":"24c51475cd6446cde19224345b50afc1a4bb2ec7590aea6bbd2dfdffb3681ce5"}}