{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:7VGUQQSPMLO5ZNJH5W5OROIX3X","short_pith_number":"pith:7VGUQQSP","canonical_record":{"source":{"id":"2405.18641","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-28T22:53:43Z","cross_cats_sorted":[],"title_canon_sha256":"374f6a70764857f22a5600a4010adcf627aa833794c0a8e3f7b22fbc5cd21025","abstract_canon_sha256":"6740aac01dd169aaa490212311628df0da47ddbe1c68d9af2c8eb0bcb9efae1a"},"schema_version":"1.0"},"canonical_sha256":"fd4d48424f62dddcb527edbae8b917ddfe45c5c36a5abf59ef2e388e56234953","source":{"kind":"arxiv","id":"2405.18641","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.18641","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"arxiv_version","alias_value":"2405.18641v5","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18641","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"pith_short_12","alias_value":"7VGUQQSPMLO5","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"pith_short_16","alias_value":"7VGUQQSPMLO5ZNJH","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"pith_short_8","alias_value":"7VGUQQSP","created_at":"2026-07-05T09:27:22Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:7VGUQQSPMLO5ZNJH5W5OROIX3X","target":"record","payload":{"canonical_record":{"source":{"id":"2405.18641","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-28T22:53:43Z","cross_cats_sorted":[],"title_canon_sha256":"374f6a70764857f22a5600a4010adcf627aa833794c0a8e3f7b22fbc5cd21025","abstract_canon_sha256":"6740aac01dd169aaa490212311628df0da47ddbe1c68d9af2c8eb0bcb9efae1a"},"schema_version":"1.0"},"canonical_sha256":"fd4d48424f62dddcb527edbae8b917ddfe45c5c36a5abf59ef2e388e56234953","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:27:22.373973Z","signature_b64":"TKij3P+oYL/CLTc67Zlx4kUf/RvmpGoyhRZl5XEof5DamzFZ1nVtdQNlowzI2Jo4PFBjO6JRjhxCC78r0+A9CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fd4d48424f62dddcb527edbae8b917ddfe45c5c36a5abf59ef2e388e56234953","last_reissued_at":"2026-07-05T09:27:22.373444Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:27:22.373444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.18641","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:27:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xHEZJtK/tSm/mTmzBtxoP/dlgUQFhhJ9VHFUogObKimcJTyJCPswkF+vHh1W6HGMEppYmaj9W93AzkLkHgxMCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T10:16:14.914626Z"},"content_sha256":"b02e1845eb227abdf4c9e741d47ce3d1337fe3978b3ee73baa211b754ae972a2","schema_version":"1.0","event_id":"sha256:b02e1845eb227abdf4c9e741d47ce3d1337fe3978b3ee73baa211b754ae972a2"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:7VGUQQSPMLO5ZNJH5W5OROIX3X","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Fatih Ilhan, Ling Liu, Selim Furkan Tekin, Sihao Hu, Tiansheng Huang","submitted_at":"2024-05-28T22:53:43Z","abstract_excerpt":"Recent studies show that Large Language Models (LLMs) with safety alignment can be jail-broken by fine-tuning on a dataset mixed with harmful data. First time in the literature, we show that the jail-broken effect can be mitigated by separating states in the finetuning stage to optimize the alignment and user datasets. Unfortunately, our subsequent study shows that this simple Bi-State Optimization (BSO) solution experiences convergence instability when steps invested in its alignment state is too small, leading to downgraded alignment performance. By statistical analysis, we show that the \\te"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18641","kind":"arxiv","version":5},"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/2405.18641/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":"2026-07-05T09:27:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"bkjUWXLQM8q0Eco8lVjsvsFBx6Dvm4IQ+XlJimm8Ogx7FKvvOVacAyjijKCvWpsbyY/vQQKzZaqzKaMDDB3/AA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T10:16:14.915518Z"},"content_sha256":"96287f17013057c9e06caca1ce201860c5a4999ead4e1dadfda527522159e4e9","schema_version":"1.0","event_id":"sha256:96287f17013057c9e06caca1ce201860c5a4999ead4e1dadfda527522159e4e9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X/bundle.json","state_url":"https://pith.science/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X/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":"2026-08-08T10:16:14Z","links":{"resolver":"https://pith.science/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X","bundle":"https://pith.science/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X/bundle.json","state":"https://pith.science/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X/state.json","well_known_bundle":"https://pith.science/.well-known/pith/7VGUQQSPMLO5ZNJH5W5OROIX3X/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:7VGUQQSPMLO5ZNJH5W5OROIX3X","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":"6740aac01dd169aaa490212311628df0da47ddbe1c68d9af2c8eb0bcb9efae1a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-28T22:53:43Z","title_canon_sha256":"374f6a70764857f22a5600a4010adcf627aa833794c0a8e3f7b22fbc5cd21025"},"schema_version":"1.0","source":{"id":"2405.18641","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.18641","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"arxiv_version","alias_value":"2405.18641v5","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18641","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"pith_short_12","alias_value":"7VGUQQSPMLO5","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"pith_short_16","alias_value":"7VGUQQSPMLO5ZNJH","created_at":"2026-07-05T09:27:22Z"},{"alias_kind":"pith_short_8","alias_value":"7VGUQQSP","created_at":"2026-07-05T09:27:22Z"}],"graph_snapshots":[{"event_id":"sha256:96287f17013057c9e06caca1ce201860c5a4999ead4e1dadfda527522159e4e9","target":"graph","created_at":"2026-07-05T09:27:22Z","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/2405.18641/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent studies show that Large Language Models (LLMs) with safety alignment can be jail-broken by fine-tuning on a dataset mixed with harmful data. First time in the literature, we show that the jail-broken effect can be mitigated by separating states in the finetuning stage to optimize the alignment and user datasets. Unfortunately, our subsequent study shows that this simple Bi-State Optimization (BSO) solution experiences convergence instability when steps invested in its alignment state is too small, leading to downgraded alignment performance. By statistical analysis, we show that the \\te","authors_text":"Fatih Ilhan, Ling Liu, Selim Furkan Tekin, Sihao Hu, Tiansheng Huang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-28T22:53:43Z","title":"Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18641","kind":"arxiv","version":5},"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:b02e1845eb227abdf4c9e741d47ce3d1337fe3978b3ee73baa211b754ae972a2","target":"record","created_at":"2026-07-05T09:27:22Z","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":"6740aac01dd169aaa490212311628df0da47ddbe1c68d9af2c8eb0bcb9efae1a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-28T22:53:43Z","title_canon_sha256":"374f6a70764857f22a5600a4010adcf627aa833794c0a8e3f7b22fbc5cd21025"},"schema_version":"1.0","source":{"id":"2405.18641","kind":"arxiv","version":5}},"canonical_sha256":"fd4d48424f62dddcb527edbae8b917ddfe45c5c36a5abf59ef2e388e56234953","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"fd4d48424f62dddcb527edbae8b917ddfe45c5c36a5abf59ef2e388e56234953","first_computed_at":"2026-07-05T09:27:22.373444Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:27:22.373444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"TKij3P+oYL/CLTc67Zlx4kUf/RvmpGoyhRZl5XEof5DamzFZ1nVtdQNlowzI2Jo4PFBjO6JRjhxCC78r0+A9CQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:27:22.373973Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.18641","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b02e1845eb227abdf4c9e741d47ce3d1337fe3978b3ee73baa211b754ae972a2","sha256:96287f17013057c9e06caca1ce201860c5a4999ead4e1dadfda527522159e4e9"],"state_sha256":"b2c26ead52b631586587af4fd06e4cb24fc353f2dd0242a4c176f14b9df3650d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"VI6PVqyqV7MmvUcQQ9iWEznt/zb/D6wBvQOr/9cyzGI+6ouXx43Dj3uP76TN/0s6h+4YUVb6dK1LA6h9kMBEAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T10:16:14.920992Z","bundle_sha256":"bcffceaf8adfc89810e8a016989d08144548b380bc62f76f5809f568440d8963"}}