{"id":"https://openalex.org/W4415181426","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1731","title":"Multi-lingual Functional Evaluation for Large Language Models","display_name":"Multi-lingual Functional Evaluation for Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4415181426","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1731"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1731","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1731","pdf_url":null,"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":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.18653/v1/2026.findings-acl.1731","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5119998728","display_name":"Victor Ojewale","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Victor Ojewale","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011015504","display_name":"Inioluwa Deborah Raji","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Inioluwa Deborah Raji","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5061790878","display_name":"Suresh Venkatasubramanian","orcid":"https://orcid.org/0000-0001-7679-7130"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Suresh Venkatasubramanian","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":0.01284711,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"34672","last_page":"34691"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.7882000207901001,"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/T10028","display_name":"Topic Modeling","score":0.7882000207901001,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.5310999751091003},{"id":"https://openalex.org/keywords/arabic","display_name":"Arabic","score":0.31690001487731934},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.2921000123023987},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.29010000824928284}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7628999948501587},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5310999751091003},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4830999970436096},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4726000130176544},{"id":"https://openalex.org/C96455323","wikidata":"https://www.wikidata.org/wiki/Q13955","display_name":"Arabic","level":2,"score":0.31690001487731934},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2921000123023987},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.29010000824928284},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.28349998593330383},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2531000077724457},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.2409999966621399}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1731","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1731","pdf_url":null,"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"},{"id":"pmh:oai:arXiv.org:2506.20793","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.20793","pdf_url":"https://arxiv.org/pdf/2506.20793","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.20793","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.20793","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1731","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1731","pdf_url":null,"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":[],"funders":[{"id":"https://openalex.org/F4320306142","display_name":"John D. and Catherine T. MacArthur Foundation","ror":"https://ror.org/00dxczh48"},{"id":"https://openalex.org/F4320327705","display_name":"Mozilla Foundation","ror":"https://ror.org/01y8r3379"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multi-lingual":[0],"competence":[1],"in":[2,114,121],"large":[3],"language":[4],"models":[5,36],"is":[6,107],"often":[7,22],"evaluated":[8],"via":[9],"static":[10,92],"data":[11],"benchmarks":[12,46,94],"such":[13],"as":[14],"Belebele,":[15],"M-MMLU":[16,150],"and":[17,33,55,85,111,118,124,139,141,151],"M-GSM.":[18],"However,":[19],"these":[20],"evaluations":[21],"fail":[23],"to":[24,68,145],"provide":[25],"an":[26],"adequate":[27],"understanding":[28],"of":[29,35,76],"the":[30,74,170],"practical":[31],"performance":[32,97,115,133,147],"robustness":[34,158],"across":[37,104,135,159,175],"multi-lingual":[38,44,93],"settings.":[39],"In":[40],"response,":[41],"we":[42,154],"create":[43],"functional":[45,63,96],"--":[47],"Cross-Lingual":[48,56],"Grade":[49],"School":[50],"Math":[51],"Symbolic":[52,120],"(CL-GSM":[53],"Symbolic)":[54],"Instruction-Following":[57],"Eval":[58],"(CL-IFEval)--":[59],"by":[60],"translating":[61],"existing":[62],"benchmark":[64],"templates":[65],"from":[66],"English":[67],"five":[69],"additional":[70],"languages":[71,136,160,165],"that":[72,90,156],"span":[73],"range":[75],"resources":[77],"available":[78],"for":[79],"NLP:":[80],"French,":[81],"Spanish,":[82],"Hindi,":[83],"Arabic":[84],"Yoruba.":[86],"Our":[87],"results":[88],"reveal":[89],"some":[91],"capture":[95],"much":[98],"more":[99],"closely":[100],"than":[101],"others":[102],"(i.e.":[103],"models,":[105],"there":[106],"a":[108,129,143],"24%,":[109],"17%":[110],"18%":[112],"decrease":[113],"between":[116,137,149],"M-GSM":[117],"CL-GSM":[119],"English,":[122],"French":[123],"Spanish":[125],"respectively;":[126],"similarly":[127],"there's":[128],"15":[130],"-":[131],"24%":[132],"drop":[134,148],"Belebele":[138],"CL-IFEval,":[140],"only":[142],"0.5%":[144],"3%":[146],"CL-IFEval).":[152],"Similarly,":[153],"find":[155],"model":[157],"varies":[161],"significantly,":[162],"with":[163],"certain":[164],"(eg.":[166],"Arabic,":[167],"English)":[168],"being":[169],"most":[171],"consistently":[172],"well":[173],"performing":[174],"evaluation":[176],"iterations.":[177]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-15T00:00:00"}