{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:SITYFITXWHMUC7APCBMV4BSVOZ","short_pith_number":"pith:SITYFITX","canonical_record":{"source":{"id":"2407.12404","kind":"arxiv","version":8},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-17T08:32:03Z","cross_cats_sorted":[],"title_canon_sha256":"30336cd36e12a9132df86bd3e3a02c8b27d2e825de089a95b7f03c3f548b1c58","abstract_canon_sha256":"541d4838bd3544a065e4175e8ea52a4573fad47f93174ce431ec6f03bad9a029"},"schema_version":"1.0"},"canonical_sha256":"922782a277b1d9417c0f10595e06557675fe940efdb7f49f65e29c268286c9d0","source":{"kind":"arxiv","id":"2407.12404","version":8},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.12404","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"arxiv_version","alias_value":"2407.12404v8","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.12404","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"pith_short_12","alias_value":"SITYFITXWHMU","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"pith_short_16","alias_value":"SITYFITXWHMUC7AP","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"pith_short_8","alias_value":"SITYFITX","created_at":"2026-07-05T10:58:17Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:SITYFITXWHMUC7APCBMV4BSVOZ","target":"record","payload":{"canonical_record":{"source":{"id":"2407.12404","kind":"arxiv","version":8},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-17T08:32:03Z","cross_cats_sorted":[],"title_canon_sha256":"30336cd36e12a9132df86bd3e3a02c8b27d2e825de089a95b7f03c3f548b1c58","abstract_canon_sha256":"541d4838bd3544a065e4175e8ea52a4573fad47f93174ce431ec6f03bad9a029"},"schema_version":"1.0"},"canonical_sha256":"922782a277b1d9417c0f10595e06557675fe940efdb7f49f65e29c268286c9d0","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:58:17.201926Z","signature_b64":"IghTryp+5Mj+DueX2KXU+cWdZ3TB1N/pkLFZOrzFsFQHSLmXk4J3C9/z3eCS8kP1zYZPO8xqaeYRny/XPazuAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"922782a277b1d9417c0f10595e06557675fe940efdb7f49f65e29c268286c9d0","last_reissued_at":"2026-07-05T10:58:17.201375Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:58:17.201375Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2407.12404","source_version":8,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:58:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"kKViX8iKj82uNbNb3bC3E9CRicGofja7cPc3F/JIABPuuspyCJlbD93QbLPG3hpFvEI3u48mUtHTdNfUGJJeAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-06T11:22:27.198570Z"},"content_sha256":"eb9c51dda6037e17bfa05bfc634d9cc829542ac6a002b5ec2cbad4aab7a9e695","schema_version":"1.0","event_id":"sha256:eb9c51dda6037e17bfa05bfc634d9cc829542ac6a002b5ec2cbad4aab7a9e695"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:SITYFITXWHMUC7APCBMV4BSVOZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Analyzing the Generalization and Reliability of Steering Vectors","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Adria Garriga-Alonso, Aengus Lynch, Brooks Paige, Daniel Tan, David Chanin, Dimitrios Kanoulas, Robert Kirk","submitted_at":"2024-07-17T08:32:03Z","abstract_excerpt":"Steering vectors (SVs) have been proposed as an effective approach to adjust language model behaviour at inference time by intervening on intermediate model activations. They have shown promise in terms of improving both capabilities and model alignment. However, the reliability and generalisation properties of this approach are unknown. In this work, we rigorously investigate these properties, and show that steering vectors have substantial limitations both in- and out-of-distribution. In-distribution, steerability is highly variable across different inputs. Depending on the concept, spurious"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.12404","kind":"arxiv","version":8},"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/2407.12404/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":"2026-07-05T10:58:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xL+JGCk9/nZouGGagvuZo2CPXmjGgICy94b4WvIFa2Td3s01Im/oATsvwLVroPShvA9Z4C2Da1E1YzMPGY1yBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-06T11:22:27.198943Z"},"content_sha256":"b6f23807894be113529883c4eab7401db0133282ab0ca81469dd2ff66f878c30","schema_version":"1.0","event_id":"sha256:b6f23807894be113529883c4eab7401db0133282ab0ca81469dd2ff66f878c30"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SITYFITXWHMUC7APCBMV4BSVOZ/bundle.json","state_url":"https://pith.science/pith/SITYFITXWHMUC7APCBMV4BSVOZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SITYFITXWHMUC7APCBMV4BSVOZ/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":"2026-07-06T11:22:27Z","links":{"resolver":"https://pith.science/pith/SITYFITXWHMUC7APCBMV4BSVOZ","bundle":"https://pith.science/pith/SITYFITXWHMUC7APCBMV4BSVOZ/bundle.json","state":"https://pith.science/pith/SITYFITXWHMUC7APCBMV4BSVOZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SITYFITXWHMUC7APCBMV4BSVOZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:SITYFITXWHMUC7APCBMV4BSVOZ","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":"541d4838bd3544a065e4175e8ea52a4573fad47f93174ce431ec6f03bad9a029","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-17T08:32:03Z","title_canon_sha256":"30336cd36e12a9132df86bd3e3a02c8b27d2e825de089a95b7f03c3f548b1c58"},"schema_version":"1.0","source":{"id":"2407.12404","kind":"arxiv","version":8}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.12404","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"arxiv_version","alias_value":"2407.12404v8","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.12404","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"pith_short_12","alias_value":"SITYFITXWHMU","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"pith_short_16","alias_value":"SITYFITXWHMUC7AP","created_at":"2026-07-05T10:58:17Z"},{"alias_kind":"pith_short_8","alias_value":"SITYFITX","created_at":"2026-07-05T10:58:17Z"}],"graph_snapshots":[{"event_id":"sha256:b6f23807894be113529883c4eab7401db0133282ab0ca81469dd2ff66f878c30","target":"graph","created_at":"2026-07-05T10:58:17Z","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/2407.12404/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Steering vectors (SVs) have been proposed as an effective approach to adjust language model behaviour at inference time by intervening on intermediate model activations. They have shown promise in terms of improving both capabilities and model alignment. However, the reliability and generalisation properties of this approach are unknown. In this work, we rigorously investigate these properties, and show that steering vectors have substantial limitations both in- and out-of-distribution. In-distribution, steerability is highly variable across different inputs. Depending on the concept, spurious","authors_text":"Adria Garriga-Alonso, Aengus Lynch, Brooks Paige, Daniel Tan, David Chanin, Dimitrios Kanoulas, Robert Kirk","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-17T08:32:03Z","title":"Analyzing the Generalization and Reliability of Steering Vectors"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.12404","kind":"arxiv","version":8},"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:eb9c51dda6037e17bfa05bfc634d9cc829542ac6a002b5ec2cbad4aab7a9e695","target":"record","created_at":"2026-07-05T10:58:17Z","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":"541d4838bd3544a065e4175e8ea52a4573fad47f93174ce431ec6f03bad9a029","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-17T08:32:03Z","title_canon_sha256":"30336cd36e12a9132df86bd3e3a02c8b27d2e825de089a95b7f03c3f548b1c58"},"schema_version":"1.0","source":{"id":"2407.12404","kind":"arxiv","version":8}},"canonical_sha256":"922782a277b1d9417c0f10595e06557675fe940efdb7f49f65e29c268286c9d0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"922782a277b1d9417c0f10595e06557675fe940efdb7f49f65e29c268286c9d0","first_computed_at":"2026-07-05T10:58:17.201375Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:58:17.201375Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"IghTryp+5Mj+DueX2KXU+cWdZ3TB1N/pkLFZOrzFsFQHSLmXk4J3C9/z3eCS8kP1zYZPO8xqaeYRny/XPazuAw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:58:17.201926Z","signed_message":"canonical_sha256_bytes"},"source_id":"2407.12404","source_kind":"arxiv","source_version":8}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:eb9c51dda6037e17bfa05bfc634d9cc829542ac6a002b5ec2cbad4aab7a9e695","sha256:b6f23807894be113529883c4eab7401db0133282ab0ca81469dd2ff66f878c30"],"state_sha256":"3ff7498d1e340e561d133a5479b291d99242fa608fbf384b81ec803e1005de78"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UlmhK9Easg9pql67Z+314uOAJAffSfUvPyxcaBa2DRCuZjxW/UTKFmEi44ySNyPoMwU69XWqZeLXBm/OdzglBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-06T11:22:27.200940Z","bundle_sha256":"3dc7e558fec98c19fb81368df3d42356f9840aa0ffddd646afc89ba01b1e4700"}}