{"id":"https://openalex.org/W7166840366","doi":"https://doi.org/10.18653/v1/2026.findings-acl.5","title":"Accelerating LLM Fine-Tuning via Embedding Knowledge Transfer","display_name":"Accelerating LLM Fine-Tuning via Embedding Knowledge Transfer","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166840366","doi":"https://doi.org/10.18653/v1/2026.findings-acl.5"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.5","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.5","pdf_url":"https://aclanthology.org/2026.findings-acl.5.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.5.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139724409","display_name":"Meishu Peng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Meishu Peng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139832697","display_name":"Ziyue Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ziyue Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139756979","display_name":"Yi Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yi Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036270316","display_name":"Pengyang Wang","orcid":"https://orcid.org/0000-0003-3961-5523"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pengyang Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049462386","display_name":"Zixuan Yuan","orcid":"https://orcid.org/0000-0003-1197-0347"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zixuan Yuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139755229","display_name":"Denghui Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Denghui Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.83000851,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"90","last_page":"107"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.08299999684095383,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.08299999684095383,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.06889999657869339,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11206","display_name":"Model Reduction and Neural Networks","score":0.03689999878406525,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.47609999775886536},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.35260000824928284},{"id":"https://openalex.org/keywords/knowledge-transfer","display_name":"Knowledge transfer","score":0.3449999988079071},{"id":"https://openalex.org/keywords/knowledge-representation-and-reasoning","display_name":"Knowledge representation and reasoning","score":0.29019999504089355},{"id":"https://openalex.org/keywords/knowledge-based-systems","display_name":"Knowledge-based systems","score":0.2718999981880188}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5601000189781189},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.47609999775886536},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3571999967098236},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.35260000824928284},{"id":"https://openalex.org/C2776960227","wikidata":"https://www.wikidata.org/wiki/Q2586354","display_name":"Knowledge transfer","level":2,"score":0.3449999988079071},{"id":"https://openalex.org/C161301231","wikidata":"https://www.wikidata.org/wiki/Q3478658","display_name":"Knowledge representation and reasoning","level":2,"score":0.29019999504089355},{"id":"https://openalex.org/C115925183","wikidata":"https://www.wikidata.org/wiki/Q1412694","display_name":"Knowledge-based systems","level":2,"score":0.2718999981880188},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.2712000012397766},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.2565999925136566},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2558000087738037},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2554999887943268},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2533000111579895},{"id":"https://openalex.org/C199639397","wikidata":"https://www.wikidata.org/wiki/Q1788588","display_name":"Engineering drawing","level":1,"score":0.2531999945640564}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.5","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.5","pdf_url":"https://aclanthology.org/2026.findings-acl.5.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.5","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.5","pdf_url":"https://aclanthology.org/2026.findings-acl.5.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2460922350","display_name":null,"funder_award_id":"2024A03J0630","funder_id":"https://openalex.org/F4320323537","funder_display_name":"Hong Kong University of Science and Technology"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320323537","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166840366.pdf","grobid_xml":"https://content.openalex.org/works/W7166840366.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Incorporating":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"for":[5,38,102],"downstream":[6],"tasks":[7],"has":[8],"recently":[9],"garnered":[10],"considerable":[11],"attention,":[12],"where":[13],"fine-tuning":[14,100,125,158],"plays":[15],"a":[16,51,59,98,110,115,137,156],"key":[17],"role":[18],"in":[19,179,183],"LLMs'":[20],"adaptation.These":[21],"LLMs,":[22],"often":[23],"consisting":[24],"of":[25,27,32,58,127],"billions":[26],"parameters,":[28],"require":[29],"vast":[30],"amounts":[31],"computational":[33],"resources":[34],"when":[35],"customizing":[36],"them":[37],"new":[39],"tasks.To":[40],"mitigate":[41],"this,":[42],"researchers":[43],"have":[44],"proposed":[45],"the":[46,124],"Parameter-Efficient":[47],"Fine-Tuning":[48],"(PEFT)":[49],"as":[50],"practical":[52],"solution":[53],"by":[54],"adjusting":[55],"fewer":[56],"parameters":[57],"pre-trained":[60],"LLM.However,":[61],"these":[62],"methods":[63],"heavily":[64],"rely":[65],"on":[66],"their":[67],"own":[68],"structural":[69],"modifications":[70],"that":[71,166],"fail":[72],"to":[73,79,90,114],"establish":[74],"an":[75],"efficient":[76],"knowledge-sharing":[77],"mechanism":[78],"distill":[80],"rich":[81],"knowledge":[82,108,142],"from":[83,109],"other":[84],"expert":[85,112,134],"models,":[86],"which":[87,105],"may":[88],"lead":[89],"inefficient":[91],"fine-tuning.In":[92],"this":[93],"paper,":[94],"we":[95,130],"propose":[96],"Pen2Sword,":[97],"lightweight":[99],"framework":[101,169],"domain":[103,153],"adaptation":[104,154],"efficiently":[106],"transfers":[107],"small":[111],"model":[113,118,175,190],"large":[116,128],"target":[117],"via":[119,136],"embedding":[120,148],"layers,":[121],"significantly":[122],"enhancing":[123],"efficiency":[126],"models.Specifically,":[129],"first":[131],"selects":[132],"optimal":[133],"models":[135],"preserving":[138],"function,":[139],"then":[140],"facilitates":[141],"transfer":[143],"through":[144],"vocabulary":[145],"alignment":[146],"and":[147,150,181,185,192,196],"expansion,":[149],"finally":[151],"accelerates":[152,171],"with":[155],"fast":[157],"paradigm.Extensive":[159],"empirical":[160],"evaluations":[161],"across":[162,188],"multiple":[163],"domains":[164],"demonstrate":[165],"our":[167],"Pen2Sword":[168],"consistently":[170],"domain-specific":[172],"fine-tuning,":[173],"improves":[174],"performance":[176],"(e.g.,":[177],"+13.6%":[178],"code":[180],"+20.1%":[182],"math),":[184],"remains":[186],"robust":[187],"diverse":[189],"families":[191],"PEFT":[193],"methods.The":[194],"codes":[195],"data":[197],"are":[198],"available":[199],"at":[200],"https://github.com/pengmeishu/Pen2Sword.":[201]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}