{"id":"https://openalex.org/W3006552058","doi":"https://doi.org/10.48550/arxiv.2002.06261","title":"Stress Test Evaluation of Transformer-based Models in Natural Language Understanding Tasks","display_name":"Stress Test Evaluation of Transformer-based Models in Natural Language Understanding Tasks","publication_year":2020,"publication_date":"2020-02-14","ids":{"openalex":"https://openalex.org/W3006552058","doi":"https://doi.org/10.48550/arxiv.2002.06261","mag":"3006552058"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2002.06261","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2002.06261","pdf_url":"https://arxiv.org/pdf/2002.06261","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2002.06261","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5043133456","display_name":"Carlos Aspillaga","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I162148367","display_name":"Pontificia Universidad Cat\u00f3lica de Chile","ror":"https://ror.org/04teye511","country_code":"CL","type":"education","lineage":["https://openalex.org/I162148367"]}],"countries":["CL"],"is_corresponding":false,"raw_author_name":"Aspillaga, Carlos","raw_affiliation_strings":["Pontificia Universidad Cat\u00f3lica de Chile, Santiago, Chile"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Pontificia Universidad Cat\u00f3lica de Chile, Santiago, Chile","institution_ids":["https://openalex.org/I162148367"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087152717","display_name":"Andr\u00e9s Carvallo","orcid":"https://orcid.org/0000-0002-5042-8772","observed_orcids":["https://orcid.org/0000-0002-5042-8772"]},"institutions":[{"id":"https://openalex.org/I162148367","display_name":"Pontificia Universidad Cat\u00f3lica de Chile","ror":"https://ror.org/04teye511","country_code":"CL","type":"education","lineage":["https://openalex.org/I162148367"]}],"countries":["CL"],"is_corresponding":false,"raw_author_name":"Carvallo, Andr\u00e9s","raw_affiliation_strings":["IMFD , Santiago , Chile","Pontificia Universidad Cat\u00f3lica de Chile, Santiago, Chile"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IMFD , Santiago , Chile","institution_ids":[]},{"raw_affiliation_string":"Pontificia Universidad Cat\u00f3lica de Chile, Santiago, Chile","institution_ids":["https://openalex.org/I162148367"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5058447624","display_name":"Vladimir Araujo","orcid":"https://orcid.org/0000-0001-5760-8410","observed_orcids":["https://orcid.org/0000-0001-5760-8410"]},"institutions":[{"id":"https://openalex.org/I162148367","display_name":"Pontificia Universidad Cat\u00f3lica de Chile","ror":"https://ror.org/04teye511","country_code":"CL","type":"education","lineage":["https://openalex.org/I162148367"]}],"countries":["CL"],"is_corresponding":false,"raw_author_name":"Araujo, Vladimir","raw_affiliation_strings":["IMFD , Santiago , Chile","Pontificia Universidad Cat\u00f3lica de Chile , Santiago , Chile","Pontifical Catholic University of Chile#TAB#"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IMFD , Santiago , Chile","institution_ids":[]},{"raw_affiliation_string":"Pontificia Universidad Cat\u00f3lica de Chile , Santiago , Chile","institution_ids":["https://openalex.org/I162148367"]},{"raw_affiliation_string":"Pontifical Catholic University of Chile#TAB#","institution_ids":["https://openalex.org/I162148367"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I162148367"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":15,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1882","last_page":"1894"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.8842999935150146,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.8842999935150146,"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/T10028","display_name":"Topic Modeling","score":0.10779999941587448,"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":0.004900000058114529,"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/adversarial-examples","display_name":"adversarial examples","score":0.7689999938011169},{"id":"https://openalex.org/keywords/stress-testing","display_name":"stress testing","score":0.6169999837875366},{"id":"https://openalex.org/keywords/transformer","display_name":"transformer models","score":0.7580000162124634},{"id":"https://openalex.org/keywords/model-robustness","display_name":"model robustness","score":0.7960000038146973},{"id":"https://openalex.org/keywords/natural-language-inference","display_name":"natural language inference","score":0.8849999904632568},{"id":"https://openalex.org/keywords/question-answering","display_name":"question answering","score":0.9980000257492065},{"id":"https://openalex.org/keywords/roberta","display_name":"RoBERTa","score":0.9879999756813049},{"id":"https://openalex.org/keywords/xlnet","display_name":"XLNet","score":0.972000002861023},{"id":"https://openalex.org/keywords/bert","display_name":"BERT","score":0.9990000128746033},{"id":"https://openalex.org/keywords/recurrent-neural-networks","display_name":"recurrent neural networks","score":0.6140000224113464},{"id":"https://openalex.org/keywords/data-leakage","display_name":"data leakage","score":0.3269999921321869}],"concepts":[{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.7897517681121826},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7769469022750854},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6489468812942505},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5614284873008728},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5168778896331787},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.48594462871551514},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4816388189792633},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.42686140537261963},{"id":"https://openalex.org/C2779439875","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Natural language understanding","level":3,"score":0.42366015911102295},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.4199143946170807},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.40629732608795166},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.40007439255714417},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.15708637237548828},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.10797694325447083},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"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/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","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}],"mesh":[],"locations_count":4,"locations":[{"id":"pmh:oai:arXiv.org:2002.06261","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2002.06261","pdf_url":"https://arxiv.org/pdf/2002.06261","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"mag:3006552058","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/pdf/2002.06261.pdf","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"arXiv (Cornell University)","raw_type":null},{"id":"doi:10.48550/arxiv.2002.06261","doi":"https://doi.org/10.48550/arxiv.2002.06261","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2002.06261","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,"listed_in":[],"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":"Preprint"},{"id":"mag:3029637777","doi":null,"is_oa":false,"landing_page_url":"https://www.aclweb.org/anthology/2020.lrec-1.232.pdf","pdf_url":null,"source":{"id":"https://openalex.org/S4306424877","display_name":"Language Resources and Evaluation","issn_l":"1574-020X","issn":["1574-020X","1574-0218"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["anvur-class-a","anvur-scientific","cwts-core","dongbi-c","erih-plus","jufo-3","norway-2","poland-70","russia-white-list-1","vabb-shw"],"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":"Language Resources and Evaluation","raw_type":null}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2002.06261","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2002.06261","pdf_url":"https://arxiv.org/pdf/2002.06261","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[{"id":"https://metadata.un.org/sdg/4","score":0.8700000047683716,"display_name":"Quality education"}],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":34,"referenced_works":["https://openalex.org/W48507986","https://openalex.org/W1600009974","https://openalex.org/W1603015195","https://openalex.org/W2020278455","https://openalex.org/W2140679639","https://openalex.org/W2146241755","https://openalex.org/W2250539671","https://openalex.org/W2296076036","https://openalex.org/W2463267937","https://openalex.org/W2516930406","https://openalex.org/W2551396370","https://openalex.org/W2607892599","https://openalex.org/W2617685061","https://openalex.org/W2626778328","https://openalex.org/W2737504179","https://openalex.org/W2790415926","https://openalex.org/W2798139452","https://openalex.org/W2799007037","https://openalex.org/W2805083708","https://openalex.org/W2811010710","https://openalex.org/W2949202705","https://openalex.org/W2950220847","https://openalex.org/W2950813464","https://openalex.org/W2952984539","https://openalex.org/W2962685628","https://openalex.org/W2963341956","https://openalex.org/W2963418779","https://openalex.org/W2963661177","https://openalex.org/W2963748441","https://openalex.org/W2963969878","https://openalex.org/W2965373594","https://openalex.org/W2965570258","https://openalex.org/W2980282514","https://openalex.org/W2996851481"],"related_works":["https://openalex.org/W4389519854","https://openalex.org/W7203669052","https://openalex.org/W4399642142","https://openalex.org/W2971524212","https://openalex.org/W2993439425","https://openalex.org/W2996601440","https://openalex.org/W4388787832","https://openalex.org/W3199638386","https://openalex.org/W4412189824","https://openalex.org/W4404752309"],"abstract_inverted_index":{"There":[0],"has":[1],"been":[2,93],"significant":[3],"progress":[4],"in":[5,8,98,121,197],"recent":[6,89],"years":[7],"the":[9,17,20,79,142],"field":[10],"of":[11,19,30,68,72],"Natural":[12,122],"Language":[13,123],"Processing":[14],"thanks":[15],"to":[16,53,100,131,167],"introduction":[18],"Transformer":[21],"architecture.":[22],"Current":[23],"state-of-the-art":[24],"models,":[25],"via":[26],"a":[27,149],"large":[28],"number":[29],"parameters":[31],"and":[32,74,119,126,157,173,182],"pre-training":[33],"on":[34,42,78],"massive":[35],"text":[36],"corpus,":[37],"have":[38,48,91,141],"shown":[39],"impressive":[40],"results":[41],"several":[43],"downstream":[44],"tasks.":[45,175],"Many":[46],"researchers":[47],"studied":[49],"previous":[50],"(non-Transformer)":[51],"models":[52,64,90,116,166],"understand":[54],"their":[55,85,102,146],"actual":[56],"behavior":[57],"under":[58,104],"different":[59],"scenarios,":[60],"showing":[61],"that":[62,75,109,154,189],"these":[63],"are":[65,135,159,178],"taking":[66],"advantage":[67],"clues":[69],"or":[70,138],"failures":[71],"datasets":[73],"slight":[76],"perturbations":[77],"input":[80],"data":[81],"can":[82],"severely":[83],"reduce":[84],"performance.":[86],"In":[87],"contrast,":[88],"not":[92],"systematically":[94],"tested":[95],"with":[96],"adversarial-examples":[97],"order":[99],"show":[101],"robustness":[103],"severe":[105],"stress":[106,168],"conditions.":[107],"For":[108],"reason,":[110],"this":[111,198],"work":[112],"evaluates":[113],"three":[114],"Transformer-based":[115],"(RoBERTa,":[117],"XLNet,":[118],"BERT)":[120],"Inference":[124],"(NLI)":[125],"Question":[127],"Answering":[128],"(QA)":[129],"tasks":[130],"know":[132],"if":[133,139],"they":[134,140,177],"more":[136,160],"robust":[137,161],"same":[143],"flaws":[144],"as":[145],"predecessors.":[147],"As":[148],"result,":[150],"our":[151],"experiments":[152],"reveal":[153],"RoBERTa,":[155],"XLNet":[156],"BERT":[158],"than":[162],"recurrent":[163],"neural":[164],"network":[165],"tests":[169],"for":[170,194],"both":[171],"NLI":[172],"QA":[174],"Nevertheless,":[176],"still":[179,192],"very":[180],"fragile":[181],"demonstrate":[183],"various":[184],"unexpected":[185],"behaviors,":[186],"thus":[187],"revealing":[188],"there":[190],"is":[191],"room":[193],"future":[195],"improvement":[196],"field.":[199]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":4},{"year":2021,"cited_by_count":5},{"year":2020,"cited_by_count":2}],"updated_date":"2026-10-30T12:39:25.321706","created_date":"2020-02-24T00:00:00"}