{"id":"https://openalex.org/W4283701884","doi":"https://doi.org/10.1109/isqed54688.2022.9806238","title":"Low-Precision Quantization Techniques for Hardware-Implementation-Friendly BERT Models","display_name":"Low-Precision Quantization Techniques for Hardware-Implementation-Friendly BERT Models","publication_year":2022,"publication_date":"2022-04-06","ids":{"openalex":"https://openalex.org/W4283701884","doi":"https://doi.org/10.1109/isqed54688.2022.9806238"},"language":"en","primary_location":{"id":"doi:10.1109/isqed54688.2022.9806238","is_oa":false,"landing_page_url":"https://doi.org/10.1109/isqed54688.2022.9806238","pdf_url":null,"source":{"id":"https://openalex.org/S4363607671","display_name":"2022 23rd International Symposium on Quality Electronic Design (ISQED)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2022 23rd International Symposium on Quality Electronic Design (ISQED)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5074083393","display_name":"Xinpei Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I74801974","display_name":"The University of Tokyo","ror":"https://ror.org/057zh3y96","country_code":"JP","type":"education","lineage":["https://openalex.org/I74801974"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Xinpei Zhang","raw_affiliation_strings":["The University of Tokyo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Tokyo","institution_ids":["https://openalex.org/I74801974"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000662681","display_name":"Yi Ding","orcid":"https://orcid.org/0000-0002-8850-9447"},"institutions":[{"id":"https://openalex.org/I74801974","display_name":"The University of Tokyo","ror":"https://ror.org/057zh3y96","country_code":"JP","type":"education","lineage":["https://openalex.org/I74801974"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Yi Ding","raw_affiliation_strings":["The University of Tokyo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Tokyo","institution_ids":["https://openalex.org/I74801974"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112911034","display_name":"Mingfei Yu","orcid":null},"institutions":[{"id":"https://openalex.org/I74801974","display_name":"The University of Tokyo","ror":"https://ror.org/057zh3y96","country_code":"JP","type":"education","lineage":["https://openalex.org/I74801974"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Mingfei Yu","raw_affiliation_strings":["The University of Tokyo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Tokyo","institution_ids":["https://openalex.org/I74801974"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030090929","display_name":"S. O\u2019uchi","orcid":"https://orcid.org/0000-0002-9386-3571"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shin-Ichi O'Uchi","raw_affiliation_strings":["AIST-UTokyo AI Chip Design Open Innovation"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AIST-UTokyo AI Chip Design Open Innovation","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5027837299","display_name":"Masahiro Fujita","orcid":"https://orcid.org/0000-0002-6516-4175"},"institutions":[{"id":"https://openalex.org/I74801974","display_name":"The University of Tokyo","ror":"https://ror.org/057zh3y96","country_code":"JP","type":"education","lineage":["https://openalex.org/I74801974"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Masahiro Fujita","raw_affiliation_strings":["The University of Tokyo","AIST-UTokyo AI Chip Design Open Innovation"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Tokyo","institution_ids":["https://openalex.org/I74801974"]},{"raw_affiliation_string":"AIST-UTokyo AI Chip Design Open Innovation","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9991000294685364,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9991000294685364,"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/T10028","display_name":"Topic Modeling","score":0.9975000023841858,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9955000281333923,"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/quantization","display_name":"Quantization (signal processing)","score":0.8275864124298096},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7754768133163452},{"id":"https://openalex.org/keywords/encoder","display_name":"Encoder","score":0.6247307062149048},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.5650634169578552},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5396682024002075},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5343747735023499},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.4835083782672882},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.4551337957382202},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.44005751609802246},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4235977232456207},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3812544345855713},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.3460445702075958},{"id":"https://openalex.org/keywords/arithmetic","display_name":"Arithmetic","score":0.32831281423568726},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.1214425265789032},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.07313555479049683}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.8275864124298096},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7754768133163452},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.6247307062149048},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.5650634169578552},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5396682024002075},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5343747735023499},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4835083782672882},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.4551337957382202},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.44005751609802246},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4235977232456207},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3812544345855713},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3460445702075958},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.32831281423568726},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.1214425265789032},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.07313555479049683},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/isqed54688.2022.9806238","is_oa":false,"landing_page_url":"https://doi.org/10.1109/isqed54688.2022.9806238","pdf_url":null,"source":{"id":"https://openalex.org/S4363607671","display_name":"2022 23rd International Symposium on Quality Electronic Design (ISQED)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2022 23rd International Symposium on Quality Electronic Design (ISQED)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":22,"referenced_works":["https://openalex.org/W2521740029","https://openalex.org/W2896457183","https://openalex.org/W2896718327","https://openalex.org/W2915478146","https://openalex.org/W2916954108","https://openalex.org/W2923014074","https://openalex.org/W2962735857","https://openalex.org/W2963122961","https://openalex.org/W2998183051","https://openalex.org/W3043504674","https://openalex.org/W3098576111","https://openalex.org/W3100083812","https://openalex.org/W3122752795","https://openalex.org/W3170233084","https://openalex.org/W3177265267","https://openalex.org/W6754503602","https://openalex.org/W6755207826","https://openalex.org/W6759144272","https://openalex.org/W6760069825","https://openalex.org/W6785530804","https://openalex.org/W6788001715","https://openalex.org/W6788635982"],"related_works":["https://openalex.org/W3049463507","https://openalex.org/W2936497627","https://openalex.org/W4288365749","https://openalex.org/W3013624417","https://openalex.org/W4287826556","https://openalex.org/W4287598411","https://openalex.org/W3098382480","https://openalex.org/W3198458223","https://openalex.org/W4288267738","https://openalex.org/W3126642501"],"abstract_inverted_index":{"Transformer-based":[0],"deep":[1],"learning":[2],"models":[3,91],"have":[4,83,110],"been":[5,84],"widely":[6],"recognized":[7],"as":[8],"highly":[9],"effective":[10],"for":[11,149,187,199],"NLP(natural":[12],"language":[13],"processing)":[14],"tasks,":[15],"among":[16],"which":[17,82,191],"BERT(Bidirectional":[18],"Encoder":[19],"Representations":[20],"from":[21,194],"Transformers)":[22],"[1]":[23],"is":[24,33,128,179],"achieving":[25],"outstanding":[26],"performance":[27,178],"on":[28,50,154],"popular":[29],"benchmarks.":[30],"However,":[31],"it":[32,127],"also":[34],"noticed":[35],"that":[36,193],"the":[37,100,114,132,135,167,175,215],"large":[38],"amount":[39],"of":[40,52,105,116,134,177,201],"parameters,":[41],"along":[42],"with":[43,55,113,141,161,208,211],"computational":[44],"burden,":[45],"are":[46,169],"constraining":[47],"its":[48],"implementation":[49,66,115],"top":[51],"hardware":[53,65,151],"platforms":[54],"limited":[56],"computing":[57],"and":[58,71,78,118,146,189,203],"memory":[59],"resources.":[60],"In":[61],"this":[62,95],"work,":[63],"having":[64],"in":[67,86,121,144,214],"mind,":[68],"we":[69],"introduce":[70],"evaluate":[72],"2":[73],"quantization":[74,120,140,185,198],"techniques":[75],"-":[76,92],"clipping,":[77],"two":[79],"piece-wise":[80,119],"quantization,":[81],"implemented":[85],"convolutional":[87],"neural":[88],"network":[89],"(CNN)":[90],"to":[93,130,171,196],"quantize":[94],"originally":[96],"heavy":[97],"model":[98,137,147,216],"into":[99],"one":[101],"requiring":[102],"smaller":[103,150],"number":[104],"bits.":[106],"Our":[107],"experimental":[108],"results":[109,186],"revealed":[111],"that,":[112,160],"clipping":[117],"an":[122],"independent":[123],"or":[124],"joint":[125],"manner,":[126],"possible":[129],"maintain":[131],"accuracy":[133,210],"BERT":[136],"after":[138],"operating":[139],"lower":[142],"bit-width":[143],"activation":[145],"quantized":[148,170],"implementation.":[152],"Evaluations":[153],"four":[155],"typical":[156],"NLP":[157],"tasks":[158],"prove":[159],"8-bit":[162,197],"integer":[163],"activations,":[164,190],"even":[165],"if":[166],"weights":[168,188,202,213],"only":[172],"4-bit":[173,195],"integer,":[174],"loss":[176],"relatively":[180],"small.":[181],"We":[182],"show":[183],"various":[184],"indicate":[192],"both":[200],"activations":[204],"can":[205],"be":[206],"used":[207],"good":[209],"all":[212],"quantized.":[217]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":2}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}