{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4LEWBFGBV6E7IQ7R5Z5DTPJZUE","short_pith_number":"pith:4LEWBFGB","schema_version":"1.0","canonical_sha256":"e2c96094c1af89f443f1ee7a39bd39a13b71456e4a2af5cde7dc28a2e6eb3d7f","source":{"kind":"arxiv","id":"2404.10306","version":6},"attestation_state":"computed","paper":{"title":"Balancing Speciality and Versatility: A Coarse to Fine Framework for Mitigating Catastrophic Forgetting in Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dawei Li, Fei Tan, Hengyuan Zhang, Rui Zhao, Sak Yang, Yanru Wu, Yong Jiang","submitted_at":"2024-04-16T06:27:39Z","abstract_excerpt":"Aligned Large Language Models (LLMs) showcase remarkable versatility, capable of handling diverse real-world tasks. Meanwhile, aligned LLMs are also expected to exhibit speciality, excelling in specific applications. However, fine-tuning with extra data, a common practice to gain speciality, often leads to catastrophic forgetting (CF) of previously acquired versatility, hindering the model's performance across diverse tasks. In response to this challenge, we propose CoFiTune, a coarse to fine framework in an attempt to strike the balance between speciality and versatility. At the coarse-graine"},"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":"2404.10306","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-16T06:27:39Z","cross_cats_sorted":[],"title_canon_sha256":"0e91ed6e6daf8e976748b6f6868ee8fe6dd2ee7de233213d52b75ffb5518b9f3","abstract_canon_sha256":"fa8939bee9586b12c7c15930197f05194eaecba117dd6ca557dc8b958c4a16f8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:18:36.282144Z","signature_b64":"X8R2N3examFCH8nxHWIYKDeEMFVf0DMC3TRGppj8YzFtSRLbVnXs8tsinhwCxsYOF8UvL3q5acClSGYCgWBhAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e2c96094c1af89f443f1ee7a39bd39a13b71456e4a2af5cde7dc28a2e6eb3d7f","last_reissued_at":"2026-07-05T10:18:36.281635Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:18:36.281635Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Balancing Speciality and Versatility: A Coarse to Fine Framework for Mitigating Catastrophic Forgetting in Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dawei Li, Fei Tan, Hengyuan Zhang, Rui Zhao, Sak Yang, Yanru Wu, Yong Jiang","submitted_at":"2024-04-16T06:27:39Z","abstract_excerpt":"Aligned Large Language Models (LLMs) showcase remarkable versatility, capable of handling diverse real-world tasks. Meanwhile, aligned LLMs are also expected to exhibit speciality, excelling in specific applications. However, fine-tuning with extra data, a common practice to gain speciality, often leads to catastrophic forgetting (CF) of previously acquired versatility, hindering the model's performance across diverse tasks. In response to this challenge, we propose CoFiTune, a coarse to fine framework in an attempt to strike the balance between speciality and versatility. At the coarse-graine"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.10306","kind":"arxiv","version":6},"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/2404.10306/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":"2404.10306","created_at":"2026-07-05T10:18:36.281695+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.10306v6","created_at":"2026-07-05T10:18:36.281695+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.10306","created_at":"2026-07-05T10:18:36.281695+00:00"},{"alias_kind":"pith_short_12","alias_value":"4LEWBFGBV6E7","created_at":"2026-07-05T10:18:36.281695+00:00"},{"alias_kind":"pith_short_16","alias_value":"4LEWBFGBV6E7IQ7R","created_at":"2026-07-05T10:18:36.281695+00:00"},{"alias_kind":"pith_short_8","alias_value":"4LEWBFGB","created_at":"2026-07-05T10:18:36.281695+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2411.15594","citing_title":"A Survey on LLM-as-a-Judge","ref_index":205,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20296","citing_title":"Spectral Unforgetting: Post-Hoc Recovery of Damaged Capabilities Without Retraining","ref_index":28,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE","json":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE.json","graph_json":"https://pith.science/api/pith-number/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/graph.json","events_json":"https://pith.science/api/pith-number/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/events.json","paper":"https://pith.science/paper/4LEWBFGB"},"agent_actions":{"view_html":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE","download_json":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE.json","view_paper":"https://pith.science/paper/4LEWBFGB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.10306&json=true","fetch_graph":"https://pith.science/api/pith-number/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/graph.json","fetch_events":"https://pith.science/api/pith-number/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/action/storage_attestation","attest_author":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/action/author_attestation","sign_citation":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/action/citation_signature","submit_replication":"https://pith.science/pith/4LEWBFGBV6E7IQ7R5Z5DTPJZUE/action/replication_record"}},"created_at":"2026-07-05T10:18:36.281695+00:00","updated_at":"2026-07-05T10:18:36.281695+00:00"}