{"id":"https://openalex.org/W2949012179","doi":"https://doi.org/10.18653/v1/d19-5608","title":"Making Asynchronous Stochastic Gradient Descent Work for Transformers","display_name":"Making Asynchronous Stochastic Gradient Descent Work for Transformers","publication_year":2019,"publication_date":"2019-01-01","ids":{"openalex":"https://openalex.org/W2949012179","doi":"https://doi.org/10.18653/v1/d19-5608","mag":"2949012179"},"language":"en","primary_location":{"id":"doi:10.18653/v1/d19-5608","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/d19-5608","pdf_url":"https://www.aclweb.org/anthology/D19-5608.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":"Proceedings of the 3rd Workshop on Neural Generation and Translation","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://www.aclweb.org/anthology/D19-5608.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112924039","display_name":"Alham Fikri Aji","orcid":null},"institutions":[{"id":"https://openalex.org/I98677209","display_name":"University of Edinburgh","ror":"https://ror.org/01nrxwf90","country_code":"GB","type":"education","lineage":["https://openalex.org/I98677209"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Alham Fikri Aji","raw_affiliation_strings":["University of Edinburgh,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Edinburgh,","institution_ids":["https://openalex.org/I98677209"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5070589644","display_name":"Kenneth Heafield","orcid":"https://orcid.org/0000-0002-6344-9927"},"institutions":[{"id":"https://openalex.org/I1320481043","display_name":"European Union","ror":"https://ror.org/019w4f821","country_code":"BE","type":"government","lineage":["https://openalex.org/I1320481043"]},{"id":"https://openalex.org/I98677209","display_name":"University of Edinburgh","ror":"https://ror.org/01nrxwf90","country_code":"GB","type":"education","lineage":["https://openalex.org/I98677209"]}],"countries":["BE","GB"],"is_corresponding":false,"raw_author_name":"Kenneth Heafield","raw_affiliation_strings":["School of Informatics, University of Edinburgh 10 Crichton Street Edinburgh EH8 9AB Scotland, European Union","University of Edinburgh,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Informatics, University of Edinburgh 10 Crichton Street Edinburgh EH8 9AB Scotland, European Union","institution_ids":["https://openalex.org/I1320481043","https://openalex.org/I98677209"]},{"raw_affiliation_string":"University of Edinburgh,","institution_ids":["https://openalex.org/I98677209"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"80","last_page":"89"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9995999932289124,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9995999932289124,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9994000196456909,"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"}},{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9986000061035156,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/asynchronous-communication","display_name":"Asynchronous communication","score":0.9373397827148438},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7753703594207764},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6854580044746399},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.6750370264053345},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.602821946144104},{"id":"https://openalex.org/keywords/convergence","display_name":"Convergence (economics)","score":0.4416271448135376},{"id":"https://openalex.org/keywords/synchronization","display_name":"Synchronization (alternating current)","score":0.42502784729003906},{"id":"https://openalex.org/keywords/asynchronous-learning","display_name":"Asynchronous learning","score":0.4123150110244751},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.34120532870292664},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.30895042419433594},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.28851351141929626},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.15394511818885803},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.1389232873916626},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.11085626482963562},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.07606875896453857}],"concepts":[{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.9373397827148438},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7753703594207764},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6854580044746399},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.6750370264053345},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.602821946144104},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.4416271448135376},{"id":"https://openalex.org/C2778562939","wikidata":"https://www.wikidata.org/wiki/Q1298791","display_name":"Synchronization (alternating current)","level":3,"score":0.42502784729003906},{"id":"https://openalex.org/C2777072894","wikidata":"https://www.wikidata.org/wiki/Q4812204","display_name":"Asynchronous learning","level":5,"score":0.4123150110244751},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.34120532870292664},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.30895042419433594},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.28851351141929626},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.15394511818885803},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.1389232873916626},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.11085626482963562},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.07606875896453857},{"id":"https://openalex.org/C50522688","wikidata":"https://www.wikidata.org/wiki/Q189833","display_name":"Economic growth","level":1,"score":0.0},{"id":"https://openalex.org/C88610354","wikidata":"https://www.wikidata.org/wiki/Q1813494","display_name":"Teaching method","level":2,"score":0.0},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.0},{"id":"https://openalex.org/C19122763","wikidata":"https://www.wikidata.org/wiki/Q7662215","display_name":"Synchronous learning","level":4,"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/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0},{"id":"https://openalex.org/C51672120","wikidata":"https://www.wikidata.org/wiki/Q303446","display_name":"Cooperative learning","level":3,"score":0.0},{"id":"https://openalex.org/C145420912","wikidata":"https://www.wikidata.org/wiki/Q853077","display_name":"Mathematics education","level":1,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.18653/v1/d19-5608","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/d19-5608","pdf_url":"https://www.aclweb.org/anthology/D19-5608.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":"Proceedings of the 3rd Workshop on Neural Generation and Translation","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:1906.03496","is_oa":true,"landing_page_url":"http://arxiv.org/abs/1906.03496","pdf_url":"https://arxiv.org/pdf/1906.03496","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"mag:2949012179","is_oa":true,"landing_page_url":"https://arxiv.org/pdf/1906.03496.pdf","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"arXiv (Cornell University)","raw_type":null},{"id":"doi:10.48550/arxiv.1906.03496","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.1906.03496","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-journal"}],"best_oa_location":{"id":"doi:10.18653/v1/d19-5608","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/d19-5608","pdf_url":"https://www.aclweb.org/anthology/D19-5608.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":"Proceedings of the 3rd Workshop on Neural Generation and Translation","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/8","display_name":"Decent work and economic growth","score":0.4399999976158142}],"awards":[{"id":"https://openalex.org/G2465893628","display_name":"Peta-5: A National Facility for Petascale Data Intensive Computation and Analytics","funder_award_id":"EP/P020259/1","funder_id":"https://openalex.org/F4320334627","funder_display_name":"Engineering and Physical Sciences Research Council"},{"id":"https://openalex.org/G5258663329","display_name":null,"funder_award_id":"EP/P020259/","funder_id":"https://openalex.org/F4320334627","funder_display_name":"Engineering and Physical Sciences Research Council"},{"id":"https://openalex.org/G6661763476","display_name":null,"funder_award_id":"EP/P020259/1","funder_id":"https://openalex.org/F4320320273","funder_display_name":"University of Cambridge"}],"funders":[{"id":"https://openalex.org/F4320312927","display_name":"Dell EMC","ror":"https://ror.org/05rejmm18"},{"id":"https://openalex.org/F4320320273","display_name":"University of Cambridge","ror":"https://ror.org/013meh722"},{"id":"https://openalex.org/F4320334627","display_name":"Engineering and Physical Sciences Research Council","ror":"https://ror.org/0439y7842"},{"id":"https://openalex.org/F4320334632","display_name":"Science and Technology Facilities Council","ror":"https://ror.org/057g20z61"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W2949012179.pdf","grobid_xml":"https://content.openalex.org/works/W2949012179.grobid-xml"},"referenced_works_count":31,"referenced_works":["https://openalex.org/W1522301498","https://openalex.org/W2133941677","https://openalex.org/W2138243089","https://openalex.org/W2145832734","https://openalex.org/W2168231600","https://openalex.org/W2336650964","https://openalex.org/W2407022425","https://openalex.org/W2541884796","https://openalex.org/W2622263826","https://openalex.org/W2740433069","https://openalex.org/W2766164908","https://openalex.org/W2769644379","https://openalex.org/W2792630208","https://openalex.org/W2796108585","https://openalex.org/W2888555799","https://openalex.org/W2896060389","https://openalex.org/W2903193068","https://openalex.org/W2911002035","https://openalex.org/W2949526915","https://openalex.org/W2951781666","https://openalex.org/W2962784628","https://openalex.org/W2962950660","https://openalex.org/W2963216553","https://openalex.org/W2963403868","https://openalex.org/W2963532001","https://openalex.org/W2963631431","https://openalex.org/W2963803379","https://openalex.org/W2963807318","https://openalex.org/W2963903325","https://openalex.org/W2963964896","https://openalex.org/W3101036738"],"related_works":["https://openalex.org/W2985517040","https://openalex.org/W2774769449","https://openalex.org/W2205410708","https://openalex.org/W2888555799","https://openalex.org/W2963179579","https://openalex.org/W3032687999","https://openalex.org/W2886404071","https://openalex.org/W3113081714","https://openalex.org/W3089807438","https://openalex.org/W3041891006","https://openalex.org/W3205436659","https://openalex.org/W2769644379","https://openalex.org/W3106387899","https://openalex.org/W2750919002","https://openalex.org/W2951862792","https://openalex.org/W3204244704","https://openalex.org/W2770274930","https://openalex.org/W2954365017","https://openalex.org/W2802686027","https://openalex.org/W2982408624"],"abstract_inverted_index":{"Asynchronous":[0],"stochastic":[1],"gradient":[2],"descent":[3],"(SGD)":[4],"converges":[5],"poorly":[6],"for":[7,17,36,45,48],"Transformer":[8,18,40,92],"models,":[9],"so":[10,54],"synchronous":[11,71],"SGD":[12,25,62],"has":[13],"become":[14],"the":[15,39,43,66,91,94],"norm":[16],"training.":[19],"This":[20],"is":[21,26,42],"unfortunate":[22],"because":[23],"asynchronous":[24,61,69,78],"faster":[27],"at":[28],"raw":[29],"training":[30,55],"speed":[31,56],"since":[32],"it":[33],"avoids":[34],"waiting":[35],"synchronization.":[37],"Moreover,":[38],"model":[41],"basis":[44],"state-of-the-art":[46],"models":[47],"several":[49,77],"tasks,":[50],"including":[51],"machine":[52],"translation,":[53],"matters.":[57],"To":[58],"understand":[59],"why":[60],"under-performs,":[63],"we":[64],"blur":[65],"lines":[67],"between":[68],"and":[70],"methods.":[72],"We":[73],"find":[74],"that":[75],"summing":[76],"updates,":[79],"rather":[80],"than":[81],"applying":[82],"them":[83],"immediately,":[84],"restores":[85],"convergence":[86],"behavior.":[87],"With":[88],"this":[89],"method,":[90],"attains":[93],"same":[95],"BLEU":[96],"score":[97],"1.36":[98],"times":[99],"as":[100],"fast.":[101]},"counts_by_year":[{"year":2019,"cited_by_count":2}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}