{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:447TLTGR6BRGPM6FRGEOSWNLZO","short_pith_number":"pith:447TLTGR","schema_version":"1.0","canonical_sha256":"e73f35ccd1f06267b3c58988e959abcb8d2cec5677bc11cd3f72ef3d1dcc0503","source":{"kind":"arxiv","id":"2501.06208","version":1},"attestation_state":"computed","paper":{"title":"Enhancing AI Safety Through the Fusion of Low Rank Adapters","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Harpreet Singh, Ponnurangam Kumaraguru, Satya Swaroop Gudipudi, Shashwat Goel, Sreeram Vipparla","submitted_at":"2024-12-30T13:12:27Z","abstract_excerpt":"Instruction fine-tuning of large language models (LLMs) is a powerful method for improving task-specific performance, but it can inadvertently lead to a phenomenon where models generate harmful responses when faced with malicious prompts. In this paper, we explore Low-Rank Adapter Fusion (LoRA) as a means to mitigate these risks while preserving the model's ability to handle diverse instructions effectively. Through an extensive comparative analysis against established baselines using recognized benchmark datasets, we demonstrate a 42\\% reduction in the harmfulness rate by leveraging LoRA fusi"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2501.06208","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-30T13:12:27Z","cross_cats_sorted":[],"title_canon_sha256":"b8d13febd0e42e75f974ad790f978d96d13ccff782c394e07c7dfa0b0f37d90e","abstract_canon_sha256":"eec239f8306ce6e1d8a1f50a000f211b5aba92ed7d07e9af33e057199d334dfe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:52:16.941835Z","signature_b64":"CeCcm6GePNCvxk8c2qLAf1DAkFdopRBzhFf5QtvaA6wJ4IfDVuleJvTYZFhVENiLAVM/0SMThw8XlE+lyothAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e73f35ccd1f06267b3c58988e959abcb8d2cec5677bc11cd3f72ef3d1dcc0503","last_reissued_at":"2026-07-05T11:52:16.941411Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:52:16.941411Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Enhancing AI Safety Through the Fusion of Low Rank Adapters","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Harpreet Singh, Ponnurangam Kumaraguru, Satya Swaroop Gudipudi, Shashwat Goel, Sreeram Vipparla","submitted_at":"2024-12-30T13:12:27Z","abstract_excerpt":"Instruction fine-tuning of large language models (LLMs) is a powerful method for improving task-specific performance, but it can inadvertently lead to a phenomenon where models generate harmful responses when faced with malicious prompts. In this paper, we explore Low-Rank Adapter Fusion (LoRA) as a means to mitigate these risks while preserving the model's ability to handle diverse instructions effectively. Through an extensive comparative analysis against established baselines using recognized benchmark datasets, we demonstrate a 42\\% reduction in the harmfulness rate by leveraging LoRA fusi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.06208","kind":"arxiv","version":1},"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/2501.06208/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2501.06208","created_at":"2026-07-05T11:52:16.941466+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.06208v1","created_at":"2026-07-05T11:52:16.941466+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.06208","created_at":"2026-07-05T11:52:16.941466+00:00"},{"alias_kind":"pith_short_12","alias_value":"447TLTGR6BRG","created_at":"2026-07-05T11:52:16.941466+00:00"},{"alias_kind":"pith_short_16","alias_value":"447TLTGR6BRGPM6F","created_at":"2026-07-05T11:52:16.941466+00:00"},{"alias_kind":"pith_short_8","alias_value":"447TLTGR","created_at":"2026-07-05T11:52:16.941466+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO","json":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO.json","graph_json":"https://pith.science/api/pith-number/447TLTGR6BRGPM6FRGEOSWNLZO/graph.json","events_json":"https://pith.science/api/pith-number/447TLTGR6BRGPM6FRGEOSWNLZO/events.json","paper":"https://pith.science/paper/447TLTGR"},"agent_actions":{"view_html":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO","download_json":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO.json","view_paper":"https://pith.science/paper/447TLTGR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.06208&json=true","fetch_graph":"https://pith.science/api/pith-number/447TLTGR6BRGPM6FRGEOSWNLZO/graph.json","fetch_events":"https://pith.science/api/pith-number/447TLTGR6BRGPM6FRGEOSWNLZO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO/action/storage_attestation","attest_author":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO/action/author_attestation","sign_citation":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO/action/citation_signature","submit_replication":"https://pith.science/pith/447TLTGR6BRGPM6FRGEOSWNLZO/action/replication_record"}},"created_at":"2026-07-05T11:52:16.941466+00:00","updated_at":"2026-07-05T11:52:16.941466+00:00"}