{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:GUD5X3KTZYOHJ3YCGCV4XF3LGQ","short_pith_number":"pith:GUD5X3KT","canonical_record":{"source":{"id":"2605.12765","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T21:26:25Z","cross_cats_sorted":[],"title_canon_sha256":"f7ac0148a6ac44443446671da58bd82b96960f87e1e7ee0de219e17cb0d846d7","abstract_canon_sha256":"98b122ffa4d5a003357308d804bbb6194227a3a3366125f7e6df692a29b1174a"},"schema_version":"1.0"},"canonical_sha256":"3507dbed53ce1c74ef0230abcb976b343c0a43b6b181873b53bc79e4fbdab568","source":{"kind":"arxiv","id":"2605.12765","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12765","created_at":"2026-05-18T03:09:48Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12765v1","created_at":"2026-05-18T03:09:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12765","created_at":"2026-05-18T03:09:48Z"},{"alias_kind":"pith_short_12","alias_value":"GUD5X3KTZYOH","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"GUD5X3KTZYOHJ3YC","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"GUD5X3KT","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:GUD5X3KTZYOHJ3YCGCV4XF3LGQ","target":"record","payload":{"canonical_record":{"source":{"id":"2605.12765","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T21:26:25Z","cross_cats_sorted":[],"title_canon_sha256":"f7ac0148a6ac44443446671da58bd82b96960f87e1e7ee0de219e17cb0d846d7","abstract_canon_sha256":"98b122ffa4d5a003357308d804bbb6194227a3a3366125f7e6df692a29b1174a"},"schema_version":"1.0"},"canonical_sha256":"3507dbed53ce1c74ef0230abcb976b343c0a43b6b181873b53bc79e4fbdab568","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:48.304324Z","signature_b64":"I9KJTG/g9JMSLIHIGBYPafqocAzvbXnlHcsQhb16/SwTMkriIMHjtljDQjdOpnoPrXk80L+7HvN0xoRcAA/ECQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3507dbed53ce1c74ef0230abcb976b343c0a43b6b181873b53bc79e4fbdab568","last_reissued_at":"2026-05-18T03:09:48.303388Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:48.303388Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.12765","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":"2026-05-18T03:09:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"iSNaQzOURm7/kzqPAttuKq8kVAteUdxL8MrE3o6OKTJUXlCDBV8BuXoLkC+3vRvG+u0894ocVHzODtIMq4l0BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T08:38:27.244880Z"},"content_sha256":"e284859eb8bb5e507151bac02c8b449a7496290f29bd41dd3a8ed392d5a52239","schema_version":"1.0","event_id":"sha256:e284859eb8bb5e507151bac02c8b449a7496290f29bd41dd3a8ed392d5a52239"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:GUD5X3KTZYOHJ3YCGCV4XF3LGQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Inference-Time Machine Unlearning via Gated Activation Redirection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A gated input-dependent rotation in the residual stream enables inference-time unlearning of specific data in LLMs without altering weights.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Christian Mattjie, Flavio du Pin Calmon, Joana Pasquali, Jo\\~ao Vitor Boer Abitante, Kristen K. Arguello, Lucas S. Kupssinsk\\\"u, Ot\\'avio Parraga, Ramiro N. Barros, Rodrigo C. Barros, Vin\\'icius Conte Turani","submitted_at":"2026-05-12T21:26:25Z","abstract_excerpt":"Large Language Models memorize vast amounts of training data, raising concerns regarding privacy, copyright infringement, and safety. Machine unlearning seeks to remove the influence of a targeted forget set while preserving model performance, ideally approximating a model retrained from scratch without the forget set. Existing approaches aim to achieve this by updating model parameters via gradient-based methods. However, these updates are computationally expensive, lead to irreversible weight changes, and degrade when the model is quantized for deployment. A recent alternative to changing mo"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"GUARD-IT matches or exceeds 12 gradient-based baselines across three model scales, while being the only method to simultaneously preserve utility, suppress memorization, and avoid catastrophic collapse across all settings.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That an input-dependent norm-preserving rotation in the residual stream can selectively remove the influence of a forget set without introducing unintended changes to model behavior on unrelated inputs.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"GUARD-IT performs machine unlearning in LLMs via inference-time gated activation redirection, matching or exceeding gradient-based baselines on TOFU and MUSE while preserving utility and working under quantization.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A gated input-dependent rotation in the residual stream enables inference-time unlearning of specific data in LLMs without altering weights.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"056e6c7803e400d74a30cce0f9bf9c96f56486ac3b14985c55b5492b53e5da86"},"source":{"id":"2605.12765","kind":"arxiv","version":1},"verdict":{"id":"9d004799-7886-4cb1-a9fd-78e8401a902b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:54:49.153797Z","strongest_claim":"GUARD-IT matches or exceeds 12 gradient-based baselines across three model scales, while being the only method to simultaneously preserve utility, suppress memorization, and avoid catastrophic collapse across all settings.","one_line_summary":"GUARD-IT performs machine unlearning in LLMs via inference-time gated activation redirection, matching or exceeding gradient-based baselines on TOFU and MUSE while preserving utility and working under quantization.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That an input-dependent norm-preserving rotation in the residual stream can selectively remove the influence of a forget set without introducing unintended changes to model behavior on unrelated inputs.","pith_extraction_headline":"A gated input-dependent rotation in the residual stream enables inference-time unlearning of specific data in LLMs without altering weights."},"references":{"count":59,"sample":[{"doi":"","year":null,"title":"TOFU: A Task of Fictitious Unlearning for LLMs","work_id":"b05148d7-e817-49f0-be96-3e297fddec9b","ref_index":1,"cited_arxiv_id":"2401.06121","is_internal_anchor":true},{"doi":"","year":null,"title":"Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks , url =","work_id":"35a3edc1-58a2-4a50-b179-e2a6ccd771d2","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019","work_id":"e3609b2c-d523-4f5c-9a53-c3fd668e03ca","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , year =","work_id":"3296f847-5cec-41c6-a43b-fe9b45396798","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Smith and Chiyuan Zhang , booktitle=","work_id":"69af5d1c-1676-41d0-87e8-3d2cca8af021","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":59,"snapshot_sha256":"41719a12ce8e1ff2064c05920c0a3b51fb65c535a3c9bd192bfaaab45e3370f5","internal_anchors":4},"formal_canon":{"evidence_count":2,"snapshot_sha256":"98b4bb989dfa4a6cf31c8e24baa4ad047ace9d455f8adfb46d21374b4c86be51"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"9d004799-7886-4cb1-a9fd-78e8401a902b"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZGprZPTPXgDwbOGKJyMuU1AL9qVo4uoEfEq0lzeLpbXScnWv2aJL3NPQLcxaa/2bHLnmA37XhPRdSRrTkQtUCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T08:38:27.245854Z"},"content_sha256":"966be1cff3af3fc3aa5156f495efe0633645622697a40198d6daf08b85540071","schema_version":"1.0","event_id":"sha256:966be1cff3af3fc3aa5156f495efe0633645622697a40198d6daf08b85540071"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ/bundle.json","state_url":"https://pith.science/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ/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":"2026-05-27T08:38:27Z","links":{"resolver":"https://pith.science/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ","bundle":"https://pith.science/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ/bundle.json","state":"https://pith.science/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GUD5X3KTZYOHJ3YCGCV4XF3LGQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:GUD5X3KTZYOHJ3YCGCV4XF3LGQ","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":"98b122ffa4d5a003357308d804bbb6194227a3a3366125f7e6df692a29b1174a","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T21:26:25Z","title_canon_sha256":"f7ac0148a6ac44443446671da58bd82b96960f87e1e7ee0de219e17cb0d846d7"},"schema_version":"1.0","source":{"id":"2605.12765","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12765","created_at":"2026-05-18T03:09:48Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12765v1","created_at":"2026-05-18T03:09:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12765","created_at":"2026-05-18T03:09:48Z"},{"alias_kind":"pith_short_12","alias_value":"GUD5X3KTZYOH","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"GUD5X3KTZYOHJ3YC","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"GUD5X3KT","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:966be1cff3af3fc3aa5156f495efe0633645622697a40198d6daf08b85540071","target":"graph","created_at":"2026-05-18T03:09:48Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"GUARD-IT matches or exceeds 12 gradient-based baselines across three model scales, while being the only method to simultaneously preserve utility, suppress memorization, and avoid catastrophic collapse across all settings."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That an input-dependent norm-preserving rotation in the residual stream can selectively remove the influence of a forget set without introducing unintended changes to model behavior on unrelated inputs."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"GUARD-IT performs machine unlearning in LLMs via inference-time gated activation redirection, matching or exceeding gradient-based baselines on TOFU and MUSE while preserving utility and working under quantization."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"A gated input-dependent rotation in the residual stream enables inference-time unlearning of specific data in LLMs without altering weights."}],"snapshot_sha256":"056e6c7803e400d74a30cce0f9bf9c96f56486ac3b14985c55b5492b53e5da86"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"98b4bb989dfa4a6cf31c8e24baa4ad047ace9d455f8adfb46d21374b4c86be51"},"paper":{"abstract_excerpt":"Large Language Models memorize vast amounts of training data, raising concerns regarding privacy, copyright infringement, and safety. Machine unlearning seeks to remove the influence of a targeted forget set while preserving model performance, ideally approximating a model retrained from scratch without the forget set. Existing approaches aim to achieve this by updating model parameters via gradient-based methods. However, these updates are computationally expensive, lead to irreversible weight changes, and degrade when the model is quantized for deployment. A recent alternative to changing mo","authors_text":"Christian Mattjie, Flavio du Pin Calmon, Joana Pasquali, Jo\\~ao Vitor Boer Abitante, Kristen K. Arguello, Lucas S. Kupssinsk\\\"u, Ot\\'avio Parraga, Ramiro N. Barros, Rodrigo C. Barros, Vin\\'icius Conte Turani","cross_cats":[],"headline":"A gated input-dependent rotation in the residual stream enables inference-time unlearning of specific data in LLMs without altering weights.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T21:26:25Z","title":"Inference-Time Machine Unlearning via Gated Activation Redirection"},"references":{"count":59,"internal_anchors":4,"resolved_work":59,"sample":[{"cited_arxiv_id":"2401.06121","doi":"","is_internal_anchor":true,"ref_index":1,"title":"TOFU: A Task of Fictitious Unlearning for LLMs","work_id":"b05148d7-e817-49f0-be96-3e297fddec9b","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks , url =","work_id":"35a3edc1-58a2-4a50-b179-e2a6ccd771d2","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019","work_id":"e3609b2c-d523-4f5c-9a53-c3fd668e03ca","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , year =","work_id":"3296f847-5cec-41c6-a43b-fe9b45396798","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Smith and Chiyuan Zhang , booktitle=","work_id":"69af5d1c-1676-41d0-87e8-3d2cca8af021","year":2025}],"snapshot_sha256":"41719a12ce8e1ff2064c05920c0a3b51fb65c535a3c9bd192bfaaab45e3370f5"},"source":{"id":"2605.12765","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:54:49.153797Z","id":"9d004799-7886-4cb1-a9fd-78e8401a902b","model_set":{"reader":"grok-4.3"},"one_line_summary":"GUARD-IT performs machine unlearning in LLMs via inference-time gated activation redirection, matching or exceeding gradient-based baselines on TOFU and MUSE while preserving utility and working under quantization.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"A gated input-dependent rotation in the residual stream enables inference-time unlearning of specific data in LLMs without altering weights.","strongest_claim":"GUARD-IT matches or exceeds 12 gradient-based baselines across three model scales, while being the only method to simultaneously preserve utility, suppress memorization, and avoid catastrophic collapse across all settings.","weakest_assumption":"That an input-dependent norm-preserving rotation in the residual stream can selectively remove the influence of a forget set without introducing unintended changes to model behavior on unrelated inputs."}},"verdict_id":"9d004799-7886-4cb1-a9fd-78e8401a902b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e284859eb8bb5e507151bac02c8b449a7496290f29bd41dd3a8ed392d5a52239","target":"record","created_at":"2026-05-18T03:09:48Z","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":"98b122ffa4d5a003357308d804bbb6194227a3a3366125f7e6df692a29b1174a","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T21:26:25Z","title_canon_sha256":"f7ac0148a6ac44443446671da58bd82b96960f87e1e7ee0de219e17cb0d846d7"},"schema_version":"1.0","source":{"id":"2605.12765","kind":"arxiv","version":1}},"canonical_sha256":"3507dbed53ce1c74ef0230abcb976b343c0a43b6b181873b53bc79e4fbdab568","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3507dbed53ce1c74ef0230abcb976b343c0a43b6b181873b53bc79e4fbdab568","first_computed_at":"2026-05-18T03:09:48.303388Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:48.303388Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"I9KJTG/g9JMSLIHIGBYPafqocAzvbXnlHcsQhb16/SwTMkriIMHjtljDQjdOpnoPrXk80L+7HvN0xoRcAA/ECQ==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:48.304324Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12765","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e284859eb8bb5e507151bac02c8b449a7496290f29bd41dd3a8ed392d5a52239","sha256:966be1cff3af3fc3aa5156f495efe0633645622697a40198d6daf08b85540071"],"state_sha256":"a13bf507d867ebb562f2577c74d550ecf559a65a6b1cc0a9b6b6015aec68f4f0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oOs3OZRPLY3Z8xqCLHeB+MsiiDjOVpXhWjwDsQIPLjO7h6iWSp5OxlzX3VifporyL9qSqCfFDAE62x5E8ip4CA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T08:38:27.250002Z","bundle_sha256":"b791e077aad2dda82296ae619f3f9e6c61a85d535bd5f91fc5d70a7b96ad0470"}}