{"id":"https://openalex.org/W4385572001","doi":"https://doi.org/10.18653/v1/2023.findings-acl.806","title":"GLUE-X: Evaluating Natural Language Understanding Models from an Out-of-Distribution Generalization Perspective","display_name":"GLUE-X: Evaluating Natural Language Understanding Models from an Out-of-Distribution Generalization Perspective","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4385572001","doi":"https://doi.org/10.18653/v1/2023.findings-acl.806"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2023.findings-acl.806","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.findings-acl.806","pdf_url":"https://aclanthology.org/2023.findings-acl.806.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 2023","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2023.findings-acl.806.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5035082722","display_name":"Linyi Yang","orcid":"https://orcid.org/0000-0003-0667-7349"},"institutions":[{"id":"https://openalex.org/I4210137766","display_name":"Institute for Advanced Study","ror":"https://ror.org/03xg85719","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210137766"]},{"id":"https://openalex.org/I3133055985","display_name":"Westlake University","ror":"https://ror.org/05hfa4n20","country_code":"CN","type":"education","lineage":["https://openalex.org/I3133055985"]}],"countries":["CN","DE"],"is_corresponding":true,"raw_author_name":"Linyi Yang","raw_affiliation_strings":["Institute of Advanced Technology, Westlake Institute for Advanced Study","School of Engineering, Westlake University"],"affiliations":[{"raw_affiliation_string":"Institute of Advanced Technology, Westlake Institute for Advanced Study","institution_ids":["https://openalex.org/I4210137766"]},{"raw_affiliation_string":"School of Engineering, Westlake University","institution_ids":["https://openalex.org/I3133055985"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021587333","display_name":"Shuibai Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]},{"id":"https://openalex.org/I3133055985","display_name":"Westlake University","ror":"https://ror.org/05hfa4n20","country_code":"CN","type":"education","lineage":["https://openalex.org/I3133055985"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shuibai Zhang","raw_affiliation_strings":["School of Engineering, Westlake University","University of Electronic Science and Technology of China"],"affiliations":[{"raw_affiliation_string":"School of Engineering, Westlake University","institution_ids":["https://openalex.org/I3133055985"]},{"raw_affiliation_string":"University of Electronic Science and Technology of China","institution_ids":["https://openalex.org/I150229711"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029082837","display_name":"Libo Qin","orcid":"https://orcid.org/0000-0002-3619-675X"},"institutions":[{"id":"https://openalex.org/I139660479","display_name":"Central South University","ror":"https://ror.org/00f1zfq44","country_code":"CN","type":"education","lineage":["https://openalex.org/I139660479"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Libo Qin","raw_affiliation_strings":["School of Computer Science and Engineering, Central South University"],"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Central South University","institution_ids":["https://openalex.org/I139660479"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032649444","display_name":"Yafu Li","orcid":"https://orcid.org/0000-0002-7895-9997"},"institutions":[{"id":"https://openalex.org/I3133055985","display_name":"Westlake University","ror":"https://ror.org/05hfa4n20","country_code":"CN","type":"education","lineage":["https://openalex.org/I3133055985"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yafu Li","raw_affiliation_strings":["School of Engineering, Westlake University"],"affiliations":[{"raw_affiliation_string":"School of Engineering, Westlake University","institution_ids":["https://openalex.org/I3133055985"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100685716","display_name":"Yidong Wang","orcid":"https://orcid.org/0009-0007-9969-8259"},"institutions":[{"id":"https://openalex.org/I3133055985","display_name":"Westlake University","ror":"https://ror.org/05hfa4n20","country_code":"CN","type":"education","lineage":["https://openalex.org/I3133055985"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yidong Wang","raw_affiliation_strings":["School of Engineering, Westlake University"],"affiliations":[{"raw_affiliation_string":"School of Engineering, Westlake University","institution_ids":["https://openalex.org/I3133055985"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081383763","display_name":"Hanmeng Liu","orcid":"https://orcid.org/0009-0007-1060-6064"},"institutions":[{"id":"https://openalex.org/I3133055985","display_name":"Westlake University","ror":"https://ror.org/05hfa4n20","country_code":"CN","type":"education","lineage":["https://openalex.org/I3133055985"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hanmeng Liu","raw_affiliation_strings":["School of Engineering, Westlake University"],"affiliations":[{"raw_affiliation_string":"School of Engineering, Westlake University","institution_ids":["https://openalex.org/I3133055985"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100700956","display_name":"Jindong Wang","orcid":"https://orcid.org/0000-0002-4833-0880"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jindong Wang","raw_affiliation_strings":["Microsoft Research Asia"],"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100723568","display_name":"Xing Xie","orcid":"https://orcid.org/0000-0002-1636-4816"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xing Xie","raw_affiliation_strings":["Microsoft Research Asia"],"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100333738","display_name":"Yue Zhang","orcid":"https://orcid.org/0000-0002-6327-5023"},"institutions":[{"id":"https://openalex.org/I4210137766","display_name":"Institute for Advanced Study","ror":"https://ror.org/03xg85719","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210137766"]},{"id":"https://openalex.org/I3133055985","display_name":"Westlake University","ror":"https://ror.org/05hfa4n20","country_code":"CN","type":"education","lineage":["https://openalex.org/I3133055985"]}],"countries":["CN","DE"],"is_corresponding":false,"raw_author_name":"Yue Zhang","raw_affiliation_strings":["School of Engineering, Westlake University","Institute of Advanced Technology, Westlake Institute for Advanced Study"],"affiliations":[{"raw_affiliation_string":"School of Engineering, Westlake University","institution_ids":["https://openalex.org/I3133055985"]},{"raw_affiliation_string":"Institute of Advanced Technology, Westlake Institute for Advanced Study","institution_ids":["https://openalex.org/I4210137766"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5035082722"],"corresponding_institution_ids":["https://openalex.org/I3133055985","https://openalex.org/I4210137766"],"apc_list":null,"apc_paid":null,"fwci":4.3498,"has_fulltext":true,"cited_by_count":25,"citation_normalized_percentile":{"value":0.95474258,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":97,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"12731","last_page":"12750"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9998999834060669,"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":0.9998999834060669,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9986000061035156,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9873999953269958,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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":0.844874918460846},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8034839034080505},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.682217001914978},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.595399022102356},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5756406784057617},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.5335204005241394},{"id":"https://openalex.org/keywords/natural-language-understanding","display_name":"Natural language understanding","score":0.5037881731987},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.49783825874328613},{"id":"https://openalex.org/keywords/limiting","display_name":"Limiting","score":0.48525166511535645},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.48417240381240845},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.4639069437980652},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.08372944593429565},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.06696689128875732}],"concepts":[{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.844874918460846},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8034839034080505},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.682217001914978},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.595399022102356},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5756406784057617},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5335204005241394},{"id":"https://openalex.org/C2779439875","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Natural language understanding","level":3,"score":0.5037881731987},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.49783825874328613},{"id":"https://openalex.org/C188198153","wikidata":"https://www.wikidata.org/wiki/Q1613840","display_name":"Limiting","level":2,"score":0.48525166511535645},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.48417240381240845},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.4639069437980652},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.08372944593429565},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.06696689128875732},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0},{"id":"https://openalex.org/C78519656","wikidata":"https://www.wikidata.org/wiki/Q101333","display_name":"Mechanical engineering","level":1,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2023.findings-acl.806","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.findings-acl.806","pdf_url":"https://aclanthology.org/2023.findings-acl.806.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 2023","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2023.findings-acl.806","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.findings-acl.806","pdf_url":"https://aclanthology.org/2023.findings-acl.806.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 2023","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.6700000166893005}],"awards":[],"funders":[{"id":"https://openalex.org/F4320322927","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884"}],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4385572001.pdf"},"referenced_works_count":85,"referenced_works":["https://openalex.org/W131533222","https://openalex.org/W1840435438","https://openalex.org/W2016944307","https://openalex.org/W2113459411","https://openalex.org/W2155858138","https://openalex.org/W2170240176","https://openalex.org/W2557764419","https://openalex.org/W2593833795","https://openalex.org/W2889965839","https://openalex.org/W2891012317","https://openalex.org/W2896457183","https://openalex.org/W2923014074","https://openalex.org/W2943552823","https://openalex.org/W2951286828","https://openalex.org/W2951936329","https://openalex.org/W2952984539","https://openalex.org/W2953039212","https://openalex.org/W2953163841","https://openalex.org/W2953556549","https://openalex.org/W2962727366","https://openalex.org/W2962736243","https://openalex.org/W2963846996","https://openalex.org/W2965373594","https://openalex.org/W2970119729","https://openalex.org/W2970597249","https://openalex.org/W2977235550","https://openalex.org/W2978017171","https://openalex.org/W2994972936","https://openalex.org/W2996428491","https://openalex.org/W2996507500","https://openalex.org/W3011574394","https://openalex.org/W3021037761","https://openalex.org/W3021336872","https://openalex.org/W3034408878","https://openalex.org/W3034850762","https://openalex.org/W3034917890","https://openalex.org/W3034999214","https://openalex.org/W3035503910","https://openalex.org/W3035507081","https://openalex.org/W3035990676","https://openalex.org/W3038585207","https://openalex.org/W3043411673","https://openalex.org/W3095992020","https://openalex.org/W3105928338","https://openalex.org/W3112486745","https://openalex.org/W3114651185","https://openalex.org/W3114925695","https://openalex.org/W3169283738","https://openalex.org/W3171654528","https://openalex.org/W3172642864","https://openalex.org/W3174057701","https://openalex.org/W3176197839","https://openalex.org/W3177397812","https://openalex.org/W3182414670","https://openalex.org/W3195577433","https://openalex.org/W3201734921","https://openalex.org/W3206010442","https://openalex.org/W3208933101","https://openalex.org/W4205725534","https://openalex.org/W4206236515","https://openalex.org/W4221140922","https://openalex.org/W4221149036","https://openalex.org/W4221167889","https://openalex.org/W4226040778","https://openalex.org/W4226278401","https://openalex.org/W4281690148","https://openalex.org/W4283204295","https://openalex.org/W4285223438","https://openalex.org/W4285298351","https://openalex.org/W4286224803","https://openalex.org/W4287076014","https://openalex.org/W4287646194","https://openalex.org/W4287824654","https://openalex.org/W4288089799","https://openalex.org/W4292779060","https://openalex.org/W4293167850","https://openalex.org/W4294435801","https://openalex.org/W4297725990","https://openalex.org/W4300756893","https://openalex.org/W4303648884","https://openalex.org/W4307415565","https://openalex.org/W4308352263","https://openalex.org/W4321854923","https://openalex.org/W4385567015","https://openalex.org/W4385573449"],"related_works":["https://openalex.org/W2367925007","https://openalex.org/W4288263119","https://openalex.org/W3015724364","https://openalex.org/W2967994095","https://openalex.org/W4285240985","https://openalex.org/W2900126711","https://openalex.org/W4286930972","https://openalex.org/W3202115945","https://openalex.org/W2542958340","https://openalex.org/W4389520438"],"abstract_inverted_index":{"Pre-trained":[0],"language":[1,13],"models":[2,15],"(PLMs)":[3],"are":[4,101],"known":[5],"to":[6,77,86,135],"improve":[7,87],"the":[8,23,27,40,49,67,79,116],"generalization":[9,30],"performance":[10,127],"of":[11,20,43,69,81],"natural":[12],"understanding":[14],"by":[16],"leveraging":[17],"large":[18],"amounts":[19],"data":[21],"during":[22],"pre-training":[24],"phase.":[25],"However,":[26],"out-of-distribution":[28],"(OOD)":[29],"problem":[31],"remains":[32],"a":[33,54,82],"challenge":[34],"in":[35,63,122,131],"many":[36],"NLP":[37,64,106,123],"tasks,":[38,124],"limiting":[39],"real-world":[41],"deployment":[42],"these":[44],"methods.":[45],"This":[46],"paper":[47],"presents":[48],"first":[50],"attempt":[51],"at":[52],"creating":[53],"unified":[55],"benchmark":[56,90],"named":[57],"GLUE-X":[58],"for":[59,96,118],"evaluating":[60],"OOD":[61,70,97,120],"robustness":[62,71,80],"models,":[65],"highlighting":[66],"importance":[68],"and":[72,84,99],"providing":[73],"insights":[74],"on":[75,103],"how":[76,85],"measure":[78],"model":[83],"it.":[88],"The":[89],"includes":[91],"13":[92],"publicly":[93],"available":[94],"datasets":[95],"testing,":[98],"evaluations":[100],"conducted":[102],"8":[104],"classic":[105],"tasks":[107],"over":[108],"21":[109],"popularly":[110],"used":[111],"PLMs.":[112],"Our":[113],"findings":[114],"confirm":[115],"need":[117],"improved":[119],"accuracy":[121],"as":[125],"significant":[126],"degradation":[128],"was":[129],"observed":[130],"all":[132],"settings":[133],"compared":[134],"in-distribution":[136],"(ID)":[137],"accuracy.":[138]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":13},{"year":2024,"cited_by_count":5},{"year":2023,"cited_by_count":4}],"updated_date":"2026-04-02T15:55:50.835912","created_date":"2025-10-10T00:00:00"}