{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:AAY3BCOV7UMIHK3DF4XV7RVDKG","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":"e8231812a2f74367422d707d85f583543a8aaa0e532213eddc97ab5bba0fafae","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-06T15:34:04Z","title_canon_sha256":"3853a9d73ce09ec9c126298b887316f715de998699cc5b4803c058ed017d4fd6"},"schema_version":"1.0","source":{"id":"2410.04524","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2410.04524","created_at":"2026-07-05T10:15:19Z"},{"alias_kind":"arxiv_version","alias_value":"2410.04524v2","created_at":"2026-07-05T10:15:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.04524","created_at":"2026-07-05T10:15:19Z"},{"alias_kind":"pith_short_12","alias_value":"AAY3BCOV7UMI","created_at":"2026-07-05T10:15:19Z"},{"alias_kind":"pith_short_16","alias_value":"AAY3BCOV7UMIHK3D","created_at":"2026-07-05T10:15:19Z"},{"alias_kind":"pith_short_8","alias_value":"AAY3BCOV","created_at":"2026-07-05T10:15:19Z"}],"graph_snapshots":[{"event_id":"sha256:4f47915f9680b4c6b410cebc0f83500296695784773399d165324521e2bfeee3","target":"graph","created_at":"2026-07-05T10:15:19Z","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/2410.04524/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Instruction fine-tuning has emerged as a critical technique for customizing Large Language Models (LLMs) to specific applications. However, recent studies have highlighted significant security vulnerabilities in fine-tuned LLMs. Existing defense efforts focus more on pre-training and post-training methods, yet there remains underexplored in in-training methods. To fill this gap, we introduce a novel secure-tuning strategy called SWAT. By analyzing how module-level parameters (e.g. Q/K/V/O) affect the security feature space drift, we identify a robust subset of modules, termed Mods_Rob. Our SWA","authors_text":"Bing Qin, Danyang Zhao, Fenglei Fan, Jiawei Cao, Ming Ma, Sendong Zhao, Shuren Qi, Ting Liu, Yanrui Du","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-06T15:34:04Z","title":"Toward Secure Tuning: Mitigating Security Risks from Instruction Fine-Tuning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.04524","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":"sha256:4515e12a5ca14e385fbdd7ca0a9e5753782f33cd8a4395d1cc558a39e3573038","target":"record","created_at":"2026-07-05T10:15:19Z","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":"e8231812a2f74367422d707d85f583543a8aaa0e532213eddc97ab5bba0fafae","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-06T15:34:04Z","title_canon_sha256":"3853a9d73ce09ec9c126298b887316f715de998699cc5b4803c058ed017d4fd6"},"schema_version":"1.0","source":{"id":"2410.04524","kind":"arxiv","version":2}},"canonical_sha256":"0031b089d5fd1883ab632f2f5fc6a35194bb9b3810cd6975afa8a5b328fa9372","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0031b089d5fd1883ab632f2f5fc6a35194bb9b3810cd6975afa8a5b328fa9372","first_computed_at":"2026-07-05T10:15:19.717950Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:15:19.717950Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Yh056Lec20PIzltA9Fu6GitoIoLMwnRnG1S8u6S1ZLpsTQ+liDNgoLoV2gaGZRisgjFe2YaN6GyHDxW7dfguBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:15:19.718448Z","signed_message":"canonical_sha256_bytes"},"source_id":"2410.04524","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4515e12a5ca14e385fbdd7ca0a9e5753782f33cd8a4395d1cc558a39e3573038","sha256:4f47915f9680b4c6b410cebc0f83500296695784773399d165324521e2bfeee3"],"state_sha256":"b350b199c163a6a98414d6accbb0a76b517e09c9bb447414a19ef5b3712d660e"}