{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:MBD6DYFKPFIUJENVRUK6GDVT7P","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":"1bd649d66ef533f6a042d345faf1a8110da4ab461c23ec1c1e3c4e73c09d5f9b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-03T16:43:42Z","title_canon_sha256":"406edc43360f8e0fc82421e1b7ca062df2800e46d32d6e3a1a56032961bdc7ce"},"schema_version":"1.0","source":{"id":"2402.02207","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.02207","created_at":"2026-07-05T08:33:00Z"},{"alias_kind":"arxiv_version","alias_value":"2402.02207v2","created_at":"2026-07-05T08:33:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02207","created_at":"2026-07-05T08:33:00Z"},{"alias_kind":"pith_short_12","alias_value":"MBD6DYFKPFIU","created_at":"2026-07-05T08:33:00Z"},{"alias_kind":"pith_short_16","alias_value":"MBD6DYFKPFIUJENV","created_at":"2026-07-05T08:33:00Z"},{"alias_kind":"pith_short_8","alias_value":"MBD6DYFK","created_at":"2026-07-05T08:33:00Z"}],"graph_snapshots":[{"event_id":"sha256:148178dea67474ee9f7ba4d8494947521caa43ce267cb2f88ed3d35248172917","target":"graph","created_at":"2026-07-05T08:33:00Z","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/2402.02207/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Current vision large language models (VLLMs) exhibit remarkable capabilities yet are prone to generate harmful content and are vulnerable to even the simplest jailbreaking attacks. Our initial analysis finds that this is due to the presence of harmful data during vision-language instruction fine-tuning, and that VLLM fine-tuning can cause forgetting of safety alignment previously learned by the underpinning LLM. To address this issue, we first curate a vision-language safe instruction-following dataset VLGuard covering various harmful categories. Our experiments demonstrate that integrating th","authors_text":"Ondrej Bohdal, Timothy Hospedales, Tingyang Yu, Yongshuo Zong, Yongxin Yang","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-03T16:43:42Z","title":"Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02207","kind":"arxiv","version":2},"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":"sha256:4af1f9ed9601985a6ad91ff9bf45cc1868f645d98a133c1d7362f54a5272cabe","target":"record","created_at":"2026-07-05T08:33:00Z","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":"1bd649d66ef533f6a042d345faf1a8110da4ab461c23ec1c1e3c4e73c09d5f9b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-03T16:43:42Z","title_canon_sha256":"406edc43360f8e0fc82421e1b7ca062df2800e46d32d6e3a1a56032961bdc7ce"},"schema_version":"1.0","source":{"id":"2402.02207","kind":"arxiv","version":2}},"canonical_sha256":"6047e1e0aa79514491b58d15e30eb3fbe9832da31e41ddb615a2bd70fd022d7e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6047e1e0aa79514491b58d15e30eb3fbe9832da31e41ddb615a2bd70fd022d7e","first_computed_at":"2026-07-05T08:33:00.235997Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:33:00.235997Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"IrjdzenF7J9baE3TqTB5qNkRt6LE3o9/ZNpLdkaEBZRNY3puPzQiviYawYXGsGEGmKKzWVF1xX+gHbN1JQPVAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:33:00.236455Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.02207","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4af1f9ed9601985a6ad91ff9bf45cc1868f645d98a133c1d7362f54a5272cabe","sha256:148178dea67474ee9f7ba4d8494947521caa43ce267cb2f88ed3d35248172917"],"state_sha256":"48806c3f9627ba3befabcf3329406316d094bc32e97e6e900dcc1d653cb20bad"}