{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:JALUH5OTSAZV5EVF4EFYBEMQJL","short_pith_number":"pith:JALUH5OT","schema_version":"1.0","canonical_sha256":"481743f5d390335e92a5e10b8091904af832b500596a92023a72aa488873b0c8","source":{"kind":"arxiv","id":"2312.10793","version":3},"attestation_state":"computed","paper":{"title":"Demystifying Instruction Mixing for Fine-tuning Large Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chiyu Zhang, Haonan Li, Minghao Wu, Renxi Wang, Timothy Baldwin, Xudong Han, Yuxia Wang","submitted_at":"2023-12-17T18:44:26Z","abstract_excerpt":"Instruction tuning significantly enhances the performance of large language models (LLMs) across various tasks. However, the procedure to optimizing the mixing of instruction datasets for LLM fine-tuning is still poorly understood. This study categorizes instructions into three primary types: NLP downstream tasks, coding, and general chat. We explore the effects of instruction tuning on different combinations of datasets on LLM performance, and find that certain instruction types are more advantageous for specific applications but can negatively impact other areas. This work provides insights "},"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":"2312.10793","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-12-17T18:44:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"f182218126f52c0ab7ea871bc919f4414e574ab66c3be1a7df7ddce24ac235c3","abstract_canon_sha256":"996b76e5b6342b9218ec4553c8b740847da5329856502d3caeedb15f94383ff7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:46:28.913887Z","signature_b64":"0lq6m2oAi2A8G0nurHLgXw0EZt7U+FraSvg2OwQcRILhv0eoEX8TfnbEIxgZ+D4PFb2umZRkw07RZ6XmZMGbBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"481743f5d390335e92a5e10b8091904af832b500596a92023a72aa488873b0c8","last_reissued_at":"2026-07-05T07:46:28.913389Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:46:28.913389Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Demystifying Instruction Mixing for Fine-tuning Large Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chiyu Zhang, Haonan Li, Minghao Wu, Renxi Wang, Timothy Baldwin, Xudong Han, Yuxia Wang","submitted_at":"2023-12-17T18:44:26Z","abstract_excerpt":"Instruction tuning significantly enhances the performance of large language models (LLMs) across various tasks. However, the procedure to optimizing the mixing of instruction datasets for LLM fine-tuning is still poorly understood. This study categorizes instructions into three primary types: NLP downstream tasks, coding, and general chat. We explore the effects of instruction tuning on different combinations of datasets on LLM performance, and find that certain instruction types are more advantageous for specific applications but can negatively impact other areas. This work provides insights "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.10793","kind":"arxiv","version":3},"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/2312.10793/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":"2312.10793","created_at":"2026-07-05T07:46:28.913448+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.10793v3","created_at":"2026-07-05T07:46:28.913448+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.10793","created_at":"2026-07-05T07:46:28.913448+00:00"},{"alias_kind":"pith_short_12","alias_value":"JALUH5OTSAZV","created_at":"2026-07-05T07:46:28.913448+00:00"},{"alias_kind":"pith_short_16","alias_value":"JALUH5OTSAZV5EVF","created_at":"2026-07-05T07:46:28.913448+00:00"},{"alias_kind":"pith_short_8","alias_value":"JALUH5OT","created_at":"2026-07-05T07:46:28.913448+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.07037","citing_title":"Beyond Monolingual Assumptions: A Survey of Code-Switched NLP in the Era of Large Language Models across Modalities","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL","json":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL.json","graph_json":"https://pith.science/api/pith-number/JALUH5OTSAZV5EVF4EFYBEMQJL/graph.json","events_json":"https://pith.science/api/pith-number/JALUH5OTSAZV5EVF4EFYBEMQJL/events.json","paper":"https://pith.science/paper/JALUH5OT"},"agent_actions":{"view_html":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL","download_json":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL.json","view_paper":"https://pith.science/paper/JALUH5OT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.10793&json=true","fetch_graph":"https://pith.science/api/pith-number/JALUH5OTSAZV5EVF4EFYBEMQJL/graph.json","fetch_events":"https://pith.science/api/pith-number/JALUH5OTSAZV5EVF4EFYBEMQJL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL/action/storage_attestation","attest_author":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL/action/author_attestation","sign_citation":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL/action/citation_signature","submit_replication":"https://pith.science/pith/JALUH5OTSAZV5EVF4EFYBEMQJL/action/replication_record"}},"created_at":"2026-07-05T07:46:28.913448+00:00","updated_at":"2026-07-05T07:46:28.913448+00:00"}