{"id":"https://openalex.org/W2995273672","doi":"https://doi.org/10.48550/arxiv.1912.10077","title":"Are Transformers universal approximators of sequence-to-sequence functions?","display_name":"Are Transformers universal approximators of sequence-to-sequence functions?","publication_year":2019,"publication_date":"2019-12-20","ids":{"openalex":"https://openalex.org/W2995273672","doi":"https://doi.org/10.48550/arxiv.1912.10077","mag":"2995273672"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:1912.10077","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/1912.10077","pdf_url":"https://arxiv.org/pdf/1912.10077","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/1912.10077","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5045562332","display_name":"Chulhee Yun","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yun, Chulhee","raw_affiliation_strings":["Massachusetts Institute of Technology, Cambridge, United States"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Massachusetts Institute of Technology, Cambridge, United States","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057451920","display_name":"Srinadh Bhojanapalli","orcid":"https://orcid.org/0000-0002-4147-2106","observed_orcids":["https://orcid.org/0000-0002-4147-2106"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Bhojanapalli, Srinadh","raw_affiliation_strings":["Google (United States), Mountain View, United States"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google (United States), Mountain View, United States","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079621556","display_name":"Ankit Singh Rawat","orcid":"https://orcid.org/0000-0001-9790-6500","observed_orcids":["https://orcid.org/0000-0001-9790-6500"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Rawat, Ankit Singh","raw_affiliation_strings":["Google (United States), Mountain View, United States"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google (United States), Mountain View, United States","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5082314161","display_name":"Sashank Jakkam Reddi","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Reddi, Sashank J.","raw_affiliation_strings":["Google (United States), Mountain View, United States"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google (United States), Mountain View, United States","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5018301052","display_name":"Sanjiv Kumar","orcid":"https://orcid.org/0000-0002-4080-1414","observed_orcids":["https://orcid.org/0000-0002-4080-1414"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kumar, Sanjiv","raw_affiliation_strings":["Google (United States), Mountain View, United States"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google (United States), Mountain View, United States","institution_ids":["https://openalex.org/I1291425158"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":120,"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":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.6509000062942505,"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/T10320","display_name":"Neural Networks and Applications","score":0.6509000062942505,"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":0.23989999294281006,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.015300000086426735,"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/transformer","display_name":"transformer models","score":0.7120000123977661},{"id":"https://openalex.org/keywords/universal-approximation","display_name":"universal approximation","score":0.9819999933242798},{"id":"https://openalex.org/keywords/permutation-equivariance","display_name":"permutation equivariance","score":0.8700000047683716},{"id":"https://openalex.org/keywords/positional-encoding","display_name":"positional encoding","score":0.7360000014305115},{"id":"https://openalex.org/keywords/self-attention","display_name":"self-attention","score":0.9449999928474426},{"id":"https://openalex.org/keywords/expressiveness","display_name":"expressiveness","score":0.7480000257492065},{"id":"https://openalex.org/keywords/compact-support","display_name":"compact support","score":0.5529999732971191}],"concepts":[{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.7501744031906128},{"id":"https://openalex.org/C21308566","wikidata":"https://www.wikidata.org/wiki/Q7169365","display_name":"Permutation (music)","level":2,"score":0.5919854044914246},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5719190835952759},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.5654259920120239},{"id":"https://openalex.org/C195818886","wikidata":"https://www.wikidata.org/wiki/Q5421724","display_name":"Expressive power","level":2,"score":0.4847692847251892},{"id":"https://openalex.org/C171036898","wikidata":"https://www.wikidata.org/wiki/Q256355","display_name":"Equivariant map","level":2,"score":0.4647011160850525},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.44050899147987366},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.42722564935684204},{"id":"https://openalex.org/C184720557","wikidata":"https://www.wikidata.org/wiki/Q7825049","display_name":"Topology (electrical circuits)","level":2,"score":0.34857648611068726},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.30317121744155884},{"id":"https://openalex.org/C202444582","wikidata":"https://www.wikidata.org/wiki/Q837863","display_name":"Pure mathematics","level":1,"score":0.17895281314849854},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.08920422196388245},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.0763913094997406},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.07569873332977295},{"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/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C24890656","wikidata":"https://www.wikidata.org/wiki/Q82811","display_name":"Acoustics","level":1,"score":0.0},{"id":"https://openalex.org/C54355233","wikidata":"https://www.wikidata.org/wiki/Q7162","display_name":"Genetics","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"pmh:oai:arXiv.org:1912.10077","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/1912.10077","pdf_url":"https://arxiv.org/pdf/1912.10077","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":"doi:10.48550/arxiv.1912.10077","doi":"https://doi.org/10.48550/arxiv.1912.10077","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.1912.10077","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:2995273672","doi":null,"is_oa":false,"landing_page_url":"https://openreview.net/pdf?id=ByxRM0Ntvr","pdf_url":null,"source":{"id":"https://openalex.org/S4306419637","display_name":"International Conference on Learning Representations","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":"International Conference on Learning Representations","raw_type":null}],"best_oa_location":{"id":"pmh:oai:arXiv.org:1912.10077","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/1912.10077","pdf_url":"https://arxiv.org/pdf/1912.10077","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":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":23,"referenced_works":["https://openalex.org/W1566289585","https://openalex.org/W1902237438","https://openalex.org/W1988115241","https://openalex.org/W2093813380","https://openalex.org/W2103496339","https://openalex.org/W2153579005","https://openalex.org/W2162064258","https://openalex.org/W2531409750","https://openalex.org/W2626778328","https://openalex.org/W2766099245","https://openalex.org/W2810898455","https://openalex.org/W2908336025","https://openalex.org/W2908802752","https://openalex.org/W2920448302","https://openalex.org/W2949603537","https://openalex.org/W2950813464","https://openalex.org/W2953333557","https://openalex.org/W2953337107","https://openalex.org/W2963097630","https://openalex.org/W2963341956","https://openalex.org/W2963748441","https://openalex.org/W2965373594","https://openalex.org/W2970727289"],"related_works":["https://openalex.org/W4385327870","https://openalex.org/W4414634062","https://openalex.org/W7108838949","https://openalex.org/W4383473342","https://openalex.org/W7133213348","https://openalex.org/W4415798446","https://openalex.org/W4392121088","https://openalex.org/W4414942720","https://openalex.org/W7133568806","https://openalex.org/W4415797069"],"abstract_inverted_index":{"Despite":[0],"the":[1,10,44,58,85,89,92,111,119],"widespread":[2],"adoption":[3],"of":[4,13,30,46,60,88,110,123],"Transformer":[5,25,66],"models":[6,15,26,67],"for":[7],"NLP":[8],"tasks,":[9],"expressive":[11],"power":[12],"these":[14,50],"is":[16,40],"not":[17],"well-understood.":[18],"In":[19,97],"this":[20,127],"paper,":[21],"we":[22,56,99,132],"establish":[23],"that":[24,65,101],"are":[27],"universal":[28,120],"approximators":[29],"continuous":[31,72],"permutation":[32,61],"equivariant":[33],"sequence-to-sequence":[34,73],"functions":[35,74],"with":[36],"compact":[37,77],"support,":[38],"which":[39],"quite":[41],"surprising":[42],"given":[43],"amount":[45],"shared":[47],"parameters":[48],"in":[49,95,118],"models.":[51],"Furthermore,":[52],"using":[53],"positional":[54],"encodings,":[55],"circumvent":[57],"restriction":[59],"equivariance,":[62],"and":[63,91,140],"show":[64],"can":[68,106],"universally":[69],"approximate":[70],"arbitrary":[71],"on":[75,126],"a":[76,115],"domain.":[78],"Interestingly,":[79],"our":[80,130],"proof":[81],"techniques":[82],"clearly":[83],"highlight":[84],"different":[86],"roles":[87],"self-attention":[90,104,138],"feed-forward":[93],"layers":[94,105,139],"Transformers.":[96,124],"particular,":[98],"prove":[100],"fixed":[102],"width":[103],"compute":[107],"contextual":[108],"mappings":[109],"input":[112],"sequences,":[113],"playing":[114],"key":[116],"role":[117],"approximation":[121],"property":[122],"Based":[125],"insight":[128],"from":[129],"analysis,":[131],"consider":[133],"other":[134],"simpler":[135],"alternatives":[136],"to":[137],"empirically":[141],"evaluate":[142],"them.":[143]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":14},{"year":2024,"cited_by_count":17},{"year":2023,"cited_by_count":20},{"year":2022,"cited_by_count":15},{"year":2021,"cited_by_count":38},{"year":2020,"cited_by_count":13}],"updated_date":"2026-10-16T09:59:22.235173","created_date":"2019-12-26T00:00:00"}