{"id":"https://openalex.org/W6947993922","doi":"https://doi.org/10.48550/arxiv.2503.01275","title":"Enhancing Non-English Capabilities of English-Centric Large Language Models through Deep Supervision Fine-Tuning","display_name":"Enhancing Non-English Capabilities of English-Centric Large Language Models through Deep Supervision Fine-Tuning","publication_year":2025,"publication_date":"2025-03-03","ids":{"openalex":"https://openalex.org/W6947993922","doi":"https://doi.org/10.48550/arxiv.2503.01275"},"language":"en","primary_location":{"id":"doi:10.48550/arxiv.2503.01275","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2503.01275","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2503.01275","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Huo, Wenshuai","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Huo, Wenshuai","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Feng, Xiaocheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Feng, Xiaocheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Huang, Yichong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Yichong","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Fu, Chengpeng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fu, Chengpeng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Li, Baohang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Baohang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Ye, Yangfan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ye, Yangfan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zhang, Zhirui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zhirui","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Tu, Dandan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tu, Dandan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Tang, Duyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Duyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lu, Yunfei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Yunfei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wang, Hui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Hui","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Qin, Bing","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qin, Bing","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":12,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":true,"primary_topic":{"id":"https://openalex.org/T11926","display_name":"Subterranean biodiversity and taxonomy","score":0.05959999933838844,"subfield":{"id":"https://openalex.org/subfields/1911","display_name":"Paleontology"},"field":{"id":"https://openalex.org/fields/19","display_name":"Earth and Planetary Sciences"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11926","display_name":"Subterranean biodiversity and taxonomy","score":0.05959999933838844,"subfield":{"id":"https://openalex.org/subfields/1911","display_name":"Paleontology"},"field":{"id":"https://openalex.org/fields/19","display_name":"Earth and Planetary Sciences"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T14427","display_name":"Environmental Monitoring and Data Management","score":0.04659999907016754,"subfield":{"id":"https://openalex.org/subfields/1907","display_name":"Geology"},"field":{"id":"https://openalex.org/fields/19","display_name":"Earth and Planetary Sciences"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12561","display_name":"bioluminescence and chemiluminescence research","score":0.034299999475479126,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/constraint","display_name":"Constraint (computer-aided design)","score":0.5946000218391418},{"id":"https://openalex.org/keywords/internal-model","display_name":"Internal model","score":0.42480000853538513},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4034000039100647},{"id":"https://openalex.org/keywords/mechanism","display_name":"Mechanism (biology)","score":0.40290001034736633},{"id":"https://openalex.org/keywords/language-understanding","display_name":"Language understanding","score":0.3580000102519989},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.35409998893737793},{"id":"https://openalex.org/keywords/second-language","display_name":"Second language","score":0.34389999508857727},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.32499998807907104}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7472000122070312},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.5946000218391418},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5062999725341797},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5019999742507935},{"id":"https://openalex.org/C28427503","wikidata":"https://www.wikidata.org/wiki/Q13580300","display_name":"Internal model","level":3,"score":0.42480000853538513},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4034000039100647},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.40290001034736633},{"id":"https://openalex.org/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.3580000102519989},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.35409998893737793},{"id":"https://openalex.org/C2987567764","wikidata":"https://www.wikidata.org/wiki/Q125421","display_name":"Second language","level":2,"score":0.34389999508857727},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.32499998807907104},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.31470000743865967},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.29010000824928284},{"id":"https://openalex.org/C2778883600","wikidata":"https://www.wikidata.org/wiki/Q2390977","display_name":"Language proficiency","level":2,"score":0.28850001096725464},{"id":"https://openalex.org/C2780035574","wikidata":"https://www.wikidata.org/wiki/Q30081","display_name":"Multilingualism","level":2,"score":0.28519999980926514},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.28279998898506165},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2721000015735626},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.26980000734329224},{"id":"https://openalex.org/C2775907273","wikidata":"https://www.wikidata.org/wiki/Q7805281","display_name":"Time constraint","level":2,"score":0.2671000063419342},{"id":"https://openalex.org/C129792486","wikidata":"https://www.wikidata.org/wiki/Q1050419","display_name":"Language identification","level":3,"score":0.265500009059906}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2503.01275","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2503.01275","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2503.01275","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2503.01275","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"score":0.8110284209251404,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1,10,135],"models":[2],"(LLMs)":[3],"have":[4],"demonstrated":[5],"significant":[6],"progress":[7],"in":[8,19,24,71,101,147],"multilingual":[9,34,218],"understanding":[11],"and":[12,51,138,163,170,211,213],"generation.":[13],"However,":[14,60],"due":[15,61],"to":[16,62,108,128,144,181],"the":[17,32,48,57,63,72,77,102,106,125,130,133,141,152,179,185,196,214],"imbalance":[18],"training":[20,116],"data,":[21],"their":[22],"capabilities":[23],"non-English":[25,42,191],"languages":[26],"are":[27],"limited.":[28],"Recent":[29],"studies":[30],"revealed":[31],"English-pivot":[33],"mechanism":[35],"of":[36,65,75,105,121,132,159,198],"LLMs,":[37,76],"where":[38],"LLMs":[39],"implicitly":[40],"convert":[41],"queries":[43],"into":[44,136],"English":[45,53],"ones":[46],"at":[47,56,124,140],"bottom":[49,126],"layers":[50,74,104,120,127,143],"adopt":[52],"for":[54,68],"thinking":[55],"middle":[58,142],"layers.":[59],"absence":[64],"explicit":[66],"supervision":[67,93,100,160],"cross-lingual":[69],"alignment":[70],"intermediate":[73],"internal":[78,103,199],"representations":[79],"during":[80],"these":[81],"stages":[82],"may":[83],"become":[84],"inaccurate.":[85],"In":[86],"this":[87],"work,":[88],"we":[89,113,155],"introduce":[90,114],"a":[91,167,171],"deep":[92],"fine-tuning":[94,227],"method":[95,177,223],"(DFT)":[96],"that":[97,221],"incorporates":[98],"additional":[99],"model":[107,180],"guide":[109],"its":[110],"workflow.":[111],"Specifically,":[112],"two":[115,157],"objectives":[117],"on":[118,205,216],"different":[119],"LLMs:":[122],"one":[123],"constrain":[129,145],"conversion":[131],"target":[134],"English,":[137],"another":[139],"reasoning":[146],"English.":[148],"To":[149],"effectively":[150],"achieve":[151],"guiding":[153],"purpose,":[154],"designed":[156],"types":[158],"signals:":[161],"logits":[162],"feature,":[164],"which":[165],"represent":[166],"stricter":[168],"constraint":[169],"relatively":[172],"more":[173],"relaxed":[174],"guidance.":[175],"Our":[176],"guides":[178],"not":[182],"only":[183],"consider":[184],"final":[186],"generated":[187],"result":[188],"when":[189],"processing":[190],"inputs":[192],"but":[193],"also":[194],"ensure":[195],"accuracy":[197],"representations.":[200],"We":[201],"conducted":[202],"extensive":[203],"experiments":[204],"typical":[206],"English-centric":[207],"large":[208],"models,":[209],"LLaMA-2":[210],"Gemma-2,":[212],"results":[215],"multiple":[217],"datasets":[219],"show":[220],"our":[222],"significantly":[224],"outperforms":[225],"traditional":[226],"methods.":[228]},"counts_by_year":[],"updated_date":"2025-11-06T06:51:31.235846","created_date":"2025-10-10T00:00:00"}