{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:R4ONF3OHSP5U7S4MYZG6HCW3KC","short_pith_number":"pith:R4ONF3OH","schema_version":"1.0","canonical_sha256":"8f1cd2edc793fb4fcb8cc64de38adb50a0e5e3f0f4e92e769f540ab146b864fb","source":{"kind":"arxiv","id":"2409.03444","version":1},"attestation_state":"computed","paper":{"title":"Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cond-mat.mtrl-sci","cs.AI"],"primary_cat":"cs.CL","authors_text":"Markus J. Buehler, Rachel K. Luu, Wei Lu","submitted_at":"2024-09-05T11:49:53Z","abstract_excerpt":"The advancement of Large Language Models (LLMs) for domain applications in fields such as materials science and engineering depends on the development of fine-tuning strategies that adapt models for specialized, technical capabilities. In this work, we explore the effects of Continued Pretraining (CPT), Supervised Fine-Tuning (SFT), and various preference-based optimization approaches, including Direct Preference Optimization (DPO) and Odds Ratio Preference Optimization (ORPO), on fine-tuned LLM performance. Our analysis shows how these strategies influence model outcomes and reveals that the "},"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":"2409.03444","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-05T11:49:53Z","cross_cats_sorted":["cond-mat.mtrl-sci","cs.AI"],"title_canon_sha256":"6bb3ee3557965ddb1791e5d7ab8a927373568578d81d40d2bfcdeb66ad901636","abstract_canon_sha256":"9a45ef64857911713d3ce08cbc7d29677a870154eedfd7b8211118f27813f6f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:03:31.458200Z","signature_b64":"S3abmR95LZ0Zwx0Eje8KEkjQRH8DA2c/1GjTFUCNZ8j03bJ/80RwdvtSCzrZTSWMqF/yttnsGoM4msjSllRYDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8f1cd2edc793fb4fcb8cc64de38adb50a0e5e3f0f4e92e769f540ab146b864fb","last_reissued_at":"2026-07-05T09:03:31.457606Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:03:31.457606Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cond-mat.mtrl-sci","cs.AI"],"primary_cat":"cs.CL","authors_text":"Markus J. Buehler, Rachel K. Luu, Wei Lu","submitted_at":"2024-09-05T11:49:53Z","abstract_excerpt":"The advancement of Large Language Models (LLMs) for domain applications in fields such as materials science and engineering depends on the development of fine-tuning strategies that adapt models for specialized, technical capabilities. In this work, we explore the effects of Continued Pretraining (CPT), Supervised Fine-Tuning (SFT), and various preference-based optimization approaches, including Direct Preference Optimization (DPO) and Odds Ratio Preference Optimization (ORPO), on fine-tuned LLM performance. Our analysis shows how these strategies influence model outcomes and reveals that the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.03444","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/2409.03444/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":"2409.03444","created_at":"2026-07-05T09:03:31.457667+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.03444v1","created_at":"2026-07-05T09:03:31.457667+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.03444","created_at":"2026-07-05T09:03:31.457667+00:00"},{"alias_kind":"pith_short_12","alias_value":"R4ONF3OHSP5U","created_at":"2026-07-05T09:03:31.457667+00:00"},{"alias_kind":"pith_short_16","alias_value":"R4ONF3OHSP5U7S4M","created_at":"2026-07-05T09:03:31.457667+00:00"},{"alias_kind":"pith_short_8","alias_value":"R4ONF3OH","created_at":"2026-07-05T09:03:31.457667+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2503.04901","citing_title":"Multiscale Analysis of Woven Composites Using Hierarchical Physically Recurrent Neural Networks","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27351","citing_title":"Heterogeneous Scientific Foundation Model Collaboration","ref_index":145,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC","json":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC.json","graph_json":"https://pith.science/api/pith-number/R4ONF3OHSP5U7S4MYZG6HCW3KC/graph.json","events_json":"https://pith.science/api/pith-number/R4ONF3OHSP5U7S4MYZG6HCW3KC/events.json","paper":"https://pith.science/paper/R4ONF3OH"},"agent_actions":{"view_html":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC","download_json":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC.json","view_paper":"https://pith.science/paper/R4ONF3OH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.03444&json=true","fetch_graph":"https://pith.science/api/pith-number/R4ONF3OHSP5U7S4MYZG6HCW3KC/graph.json","fetch_events":"https://pith.science/api/pith-number/R4ONF3OHSP5U7S4MYZG6HCW3KC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC/action/storage_attestation","attest_author":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC/action/author_attestation","sign_citation":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC/action/citation_signature","submit_replication":"https://pith.science/pith/R4ONF3OHSP5U7S4MYZG6HCW3KC/action/replication_record"}},"created_at":"2026-07-05T09:03:31.457667+00:00","updated_at":"2026-07-05T09:03:31.457667+00:00"}