{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:GGIKJM6L7PZ67OTDAIKAREOUWX","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":"3afed9d3d891c92c620f82f4c828fccb1132f69903699cbd8be88f57bd9e7bfa","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2025-03-24T20:41:57Z","title_canon_sha256":"8dc2cc9bba9912be6a21491d6f668073cad27bf203b017a2380604817ab3dbb2"},"schema_version":"1.0","source":{"id":"2503.20807","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.20807","created_at":"2026-07-05T10:39:48Z"},{"alias_kind":"arxiv_version","alias_value":"2503.20807v1","created_at":"2026-07-05T10:39:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.20807","created_at":"2026-07-05T10:39:48Z"},{"alias_kind":"pith_short_12","alias_value":"GGIKJM6L7PZ6","created_at":"2026-07-05T10:39:48Z"},{"alias_kind":"pith_short_16","alias_value":"GGIKJM6L7PZ67OTD","created_at":"2026-07-05T10:39:48Z"},{"alias_kind":"pith_short_8","alias_value":"GGIKJM6L","created_at":"2026-07-05T10:39:48Z"}],"graph_snapshots":[{"event_id":"sha256:18ac34f1d4e41a3017357c476b4c3f051599e6cb18a4b435cd1cf4bdbb544af5","target":"graph","created_at":"2026-07-05T10:39:48Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2503.20807/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Fine-tuning Large Language Models (LLMs) on some task-specific datasets has been a primary use of LLMs. However, it has been empirically observed that this approach to enhancing capability inevitably compromises safety, a phenomenon also known as the safety-capability trade-off in LLM fine-tuning. This paper presents a theoretical framework for understanding the interplay between safety and capability in two primary safety-aware LLM fine-tuning strategies, providing new insights into the effects of data similarity, context overlap, and alignment loss landscape. Our theoretical results characte","authors_text":"Han Shen, Payel Das, Pin-Yu Chen, Tianyi Chen","cross_cats":["cs.AI","cs.CL","cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2025-03-24T20:41:57Z","title":"Fundamental Safety-Capability Trade-offs in Fine-tuning Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.20807","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:ace5573defb2c99a3feb43868e04a4a1630cd6be165940f853646b0c959e43bf","target":"record","created_at":"2026-07-05T10:39:48Z","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":"3afed9d3d891c92c620f82f4c828fccb1132f69903699cbd8be88f57bd9e7bfa","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2025-03-24T20:41:57Z","title_canon_sha256":"8dc2cc9bba9912be6a21491d6f668073cad27bf203b017a2380604817ab3dbb2"},"schema_version":"1.0","source":{"id":"2503.20807","kind":"arxiv","version":1}},"canonical_sha256":"3190a4b3cbfbf3efba6302140891d4b5fac56ac8e0412d6db1ad15557c1c6ba7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3190a4b3cbfbf3efba6302140891d4b5fac56ac8e0412d6db1ad15557c1c6ba7","first_computed_at":"2026-07-05T10:39:48.552695Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:39:48.552695Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"3oa1cPQsrYyqfgKRyPzqX76Jw0pwNsGy/KSYrKskqLjXsXoF5a6ftZeR9Up6mLWosPEpr1Hu3x9IbLLC6a7eCg==","signature_status":"signed_v1","signed_at":"2026-07-05T10:39:48.553186Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.20807","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ace5573defb2c99a3feb43868e04a4a1630cd6be165940f853646b0c959e43bf","sha256:18ac34f1d4e41a3017357c476b4c3f051599e6cb18a4b435cd1cf4bdbb544af5"],"state_sha256":"82b28001b0f308163b35d9b05759c0869f68c63d339b4b6dc82c586f35eccf83"}