{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:4LAF4NMRUTBVZDCVZQTZCBGNYW","short_pith_number":"pith:4LAF4NMR","canonical_record":{"source":{"id":"2205.05124","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-10T19:04:37Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"7f5039c7c9e11a95acf22817bd7f95df51e102012a87559a01c01f167fe21ae4","abstract_canon_sha256":"bd334c8cc7216733f7b740247a5acb688dabf70e4520a5f8358946f6e9689298"},"schema_version":"1.0"},"canonical_sha256":"e2c05e3591a4c35c8c55cc279104cdc59d1a5c8d8a8e26a749a566a526f9cf2f","source":{"kind":"arxiv","id":"2205.05124","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2205.05124","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"arxiv_version","alias_value":"2205.05124v1","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.05124","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"pith_short_12","alias_value":"4LAF4NMRUTBV","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"pith_short_16","alias_value":"4LAF4NMRUTBVZDCV","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"pith_short_8","alias_value":"4LAF4NMR","created_at":"2026-07-05T04:22:22Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:4LAF4NMRUTBVZDCVZQTZCBGNYW","target":"record","payload":{"canonical_record":{"source":{"id":"2205.05124","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-10T19:04:37Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"7f5039c7c9e11a95acf22817bd7f95df51e102012a87559a01c01f167fe21ae4","abstract_canon_sha256":"bd334c8cc7216733f7b740247a5acb688dabf70e4520a5f8358946f6e9689298"},"schema_version":"1.0"},"canonical_sha256":"e2c05e3591a4c35c8c55cc279104cdc59d1a5c8d8a8e26a749a566a526f9cf2f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:22:22.561144Z","signature_b64":"byEU7jir6tIl1jRT+H5T2/44MDB7A/pg26v4fwM++G8e1XHH8Q2oFZuXoHX9Xx/LJir5YmAwIEM7opkWfzMHAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e2c05e3591a4c35c8c55cc279104cdc59d1a5c8d8a8e26a749a566a526f9cf2f","last_reissued_at":"2026-07-05T04:22:22.560733Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:22:22.560733Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2205.05124","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":"2026-07-05T04:22:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"13+X9AKQsG6ef8I14w2syk25wBPzO0eEZzOoxlbwdcTST1BBEJFU+H5Eq6/D03a1PHmdCEABFoeSkkXwIKj7BA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T21:49:16.664289Z"},"content_sha256":"70bff8a6f4ba46e9b190f54b650618b3eef95f4a7b6eca90b14a50ad646c4004","schema_version":"1.0","event_id":"sha256:70bff8a6f4ba46e9b190f54b650618b3eef95f4a7b6eca90b14a50ad646c4004"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:4LAF4NMRUTBVZDCVZQTZCBGNYW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Extracting Latent Steering Vectors from Pretrained Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Matthew E. Peters, Nishant Subramani, Nivedita Suresh","submitted_at":"2022-05-10T19:04:37Z","abstract_excerpt":"Prior work on controllable text generation has focused on learning how to control language models through trainable decoding, smart-prompt design, or fine-tuning based on a desired objective. We hypothesize that the information needed to steer the model to generate a target sentence is already encoded within the model. Accordingly, we explore a different approach altogether: extracting latent vectors directly from pretrained language model decoders without fine-tuning. Experiments show that there exist steering vectors, which, when added to the hidden states of the language model, generate a t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.05124","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/2205.05124/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":"2026-07-05T04:22:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FaOa0mzyV+cBjLZlJ6kX31SVEZoP0h929nsNhCM9KOUWteEZYvbDMBP1s4MdbXJZ4r6F345n1auOIm3l5W7+CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T21:49:16.664791Z"},"content_sha256":"979fe993f7108a04955066a0eec84f20bc812e91f5ef0805cb0f9407191ba53b","schema_version":"1.0","event_id":"sha256:979fe993f7108a04955066a0eec84f20bc812e91f5ef0805cb0f9407191ba53b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW/bundle.json","state_url":"https://pith.science/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW/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":"2026-08-08T21:49:16Z","links":{"resolver":"https://pith.science/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW","bundle":"https://pith.science/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW/bundle.json","state":"https://pith.science/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/4LAF4NMRUTBVZDCVZQTZCBGNYW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:4LAF4NMRUTBVZDCVZQTZCBGNYW","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":"bd334c8cc7216733f7b740247a5acb688dabf70e4520a5f8358946f6e9689298","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-10T19:04:37Z","title_canon_sha256":"7f5039c7c9e11a95acf22817bd7f95df51e102012a87559a01c01f167fe21ae4"},"schema_version":"1.0","source":{"id":"2205.05124","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2205.05124","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"arxiv_version","alias_value":"2205.05124v1","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.05124","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"pith_short_12","alias_value":"4LAF4NMRUTBV","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"pith_short_16","alias_value":"4LAF4NMRUTBVZDCV","created_at":"2026-07-05T04:22:22Z"},{"alias_kind":"pith_short_8","alias_value":"4LAF4NMR","created_at":"2026-07-05T04:22:22Z"}],"graph_snapshots":[{"event_id":"sha256:979fe993f7108a04955066a0eec84f20bc812e91f5ef0805cb0f9407191ba53b","target":"graph","created_at":"2026-07-05T04:22:22Z","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/2205.05124/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Prior work on controllable text generation has focused on learning how to control language models through trainable decoding, smart-prompt design, or fine-tuning based on a desired objective. We hypothesize that the information needed to steer the model to generate a target sentence is already encoded within the model. Accordingly, we explore a different approach altogether: extracting latent vectors directly from pretrained language model decoders without fine-tuning. Experiments show that there exist steering vectors, which, when added to the hidden states of the language model, generate a t","authors_text":"Matthew E. Peters, Nishant Subramani, Nivedita Suresh","cross_cats":["cs.AI","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-10T19:04:37Z","title":"Extracting Latent Steering Vectors from Pretrained Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.05124","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":"sha256:70bff8a6f4ba46e9b190f54b650618b3eef95f4a7b6eca90b14a50ad646c4004","target":"record","created_at":"2026-07-05T04:22:22Z","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":"bd334c8cc7216733f7b740247a5acb688dabf70e4520a5f8358946f6e9689298","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-10T19:04:37Z","title_canon_sha256":"7f5039c7c9e11a95acf22817bd7f95df51e102012a87559a01c01f167fe21ae4"},"schema_version":"1.0","source":{"id":"2205.05124","kind":"arxiv","version":1}},"canonical_sha256":"e2c05e3591a4c35c8c55cc279104cdc59d1a5c8d8a8e26a749a566a526f9cf2f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e2c05e3591a4c35c8c55cc279104cdc59d1a5c8d8a8e26a749a566a526f9cf2f","first_computed_at":"2026-07-05T04:22:22.560733Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:22:22.560733Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"byEU7jir6tIl1jRT+H5T2/44MDB7A/pg26v4fwM++G8e1XHH8Q2oFZuXoHX9Xx/LJir5YmAwIEM7opkWfzMHAA==","signature_status":"signed_v1","signed_at":"2026-07-05T04:22:22.561144Z","signed_message":"canonical_sha256_bytes"},"source_id":"2205.05124","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:70bff8a6f4ba46e9b190f54b650618b3eef95f4a7b6eca90b14a50ad646c4004","sha256:979fe993f7108a04955066a0eec84f20bc812e91f5ef0805cb0f9407191ba53b"],"state_sha256":"142c9528caf4a209df438ec8089d6a15fc5b9697a1a1ab60fa52bf84988b4478"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HrmAGWSfSOGWTC29AGflzEGcZVQH6JdLlN/RT8ywTWPNXXO0xybEHAebcC+xzbNWGLWBk1MI81U25N3gsR5XAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T21:49:16.668221Z","bundle_sha256":"96d896a7b7871864ddf9c8f597d24e48842511afdd2ac98a83ba9f8d55191869"}}