{"id":"https://openalex.org/W4402349265","doi":"https://doi.org/10.1145/3665314.3670806","title":"Hardware-friendly Hessian-driven Row-wise Quantization and FPGA Acceleration for Transformer-based Models","display_name":"Hardware-friendly Hessian-driven Row-wise Quantization and FPGA Acceleration for Transformer-based Models","publication_year":2024,"publication_date":"2024-08-05","ids":{"openalex":"https://openalex.org/W4402349265","doi":"https://doi.org/10.1145/3665314.3670806"},"language":"en","primary_location":{"id":"doi:10.1145/3665314.3670806","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3665314.3670806","pdf_url":null,"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 29th ACM/IEEE International Symposium on Low Power Electronics and Design","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3665314.3670806","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5015661744","display_name":"Woo-Hong Byun","orcid":null},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Woohong Byun","raw_affiliation_strings":["School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States"],"raw_orcid":"https://orcid.org/0009-0003-5087-4582","affiliations":[{"raw_affiliation_string":"School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080060539","display_name":"Jongseok Woo","orcid":"https://orcid.org/0000-0001-8030-2356"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jongseok Woo","raw_affiliation_strings":["School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States"],"raw_orcid":"https://orcid.org/0000-0001-8030-2356","affiliations":[{"raw_affiliation_string":"School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5009591041","display_name":"Saibal Mukhopadhyay","orcid":"https://orcid.org/0000-0002-8894-3390"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Saibal Mukhopadhyay","raw_affiliation_strings":["School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States"],"raw_orcid":"https://orcid.org/0000-0002-8894-3390","affiliations":[{"raw_affiliation_string":"School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States","institution_ids":["https://openalex.org/I130701444"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I130701444"],"apc_list":null,"apc_paid":null,"fwci":1.0668,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":{"value":0.78662135,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":96},"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/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.9685999751091003,"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/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.9685999751091003,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9366000294685364,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10715","display_name":"Distributed and Parallel Computing Systems","score":0.9021000266075134,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.7572392225265503},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6243796348571777},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.5984957814216614},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5609754920005798},{"id":"https://openalex.org/keywords/hessian-matrix","display_name":"Hessian matrix","score":0.5483246445655823},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.5331375598907471},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.4831666350364685},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.37663522362709045},{"id":"https://openalex.org/keywords/computational-science","display_name":"Computational science","score":0.34166446328163147},{"id":"https://openalex.org/keywords/electronic-engineering","display_name":"Electronic engineering","score":0.3228321671485901},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.2586749494075775},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.24511975049972534},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.22165626287460327},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.2033563256263733},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.19182425737380981},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.18255099654197693},{"id":"https://openalex.org/keywords/applied-mathematics","display_name":"Applied mathematics","score":0.09659519791603088}],"concepts":[{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.7572392225265503},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6243796348571777},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.5984957814216614},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5609754920005798},{"id":"https://openalex.org/C203616005","wikidata":"https://www.wikidata.org/wiki/Q620495","display_name":"Hessian matrix","level":2,"score":0.5483246445655823},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.5331375598907471},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.4831666350364685},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.37663522362709045},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.34166446328163147},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.3228321671485901},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.2586749494075775},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.24511975049972534},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.22165626287460327},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2033563256263733},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.19182425737380981},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.18255099654197693},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.09659519791603088},{"id":"https://openalex.org/C74650414","wikidata":"https://www.wikidata.org/wiki/Q11397","display_name":"Classical mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3665314.3670806","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3665314.3670806","pdf_url":null,"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 29th ACM/IEEE International Symposium on Low Power Electronics and Design","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3665314.3670806","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3665314.3670806","pdf_url":null,"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 29th ACM/IEEE International Symposium on Low Power Electronics and Design","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7","score":0.75}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":28,"referenced_works":["https://openalex.org/W2799054028","https://openalex.org/W2896457183","https://openalex.org/W2923014074","https://openalex.org/W2963367920","https://openalex.org/W2965373594","https://openalex.org/W2973061659","https://openalex.org/W2975059944","https://openalex.org/W2981852735","https://openalex.org/W2996428491","https://openalex.org/W3017746288","https://openalex.org/W3022810465","https://openalex.org/W3034999214","https://openalex.org/W3047848469","https://openalex.org/W3089271032","https://openalex.org/W3134292414","https://openalex.org/W3155487259","https://openalex.org/W3158831985","https://openalex.org/W3161584439","https://openalex.org/W4211085526","https://openalex.org/W4213068294","https://openalex.org/W4288089799","https://openalex.org/W4309591680","https://openalex.org/W4364382916","https://openalex.org/W4388214800","https://openalex.org/W4391410261","https://openalex.org/W6768807518","https://openalex.org/W6786271619","https://openalex.org/W6788001715"],"related_works":["https://openalex.org/W2611031068","https://openalex.org/W1704347466","https://openalex.org/W4283017538","https://openalex.org/W1996936972","https://openalex.org/W1545275724","https://openalex.org/W2802707792","https://openalex.org/W2569979269","https://openalex.org/W2075777916","https://openalex.org/W2518118925","https://openalex.org/W3159273459"],"abstract_inverted_index":{"Recent":[0],"advancements":[1],"in":[2,12],"using":[3],"FPGAs":[4],"as":[5],"co-processors":[6],"for":[7,128],"language":[8,33],"model":[9,121],"acceleration,":[10],"particularly":[11],"terms":[13],"of":[14,31,100,114,118],"energy":[15,148],"efficiency":[16,149],"and":[17,56,75,104],"flexibility,":[18],"face":[19],"challenges":[20],"due":[21],"to":[22,81,154,157],"limited":[23],"memory":[24,130],"capacity.":[25],"This":[26],"issue":[27],"hinders":[28],"the":[29,83,94,98,112,119,123,126],"deployment":[30],"transformer-based":[32],"models.":[34],"To":[35],"address":[36],"these":[37],"issues,":[38],"we":[39],"propose":[40],"a":[41,49],"novel":[42],"software-hardware":[43],"co-optimization":[44],"approach.":[45],"Our":[46],"approach":[47],"incorporates":[48],"hardware-friendly":[50],"hessian-based":[51],"row-wise":[52,101],"mixed-precision":[53,59,91,102],"quantization":[54,103],"algorithm":[55],"an":[57],"intra-layer":[58],"compute":[60],"fabric.":[61],"The":[62,134],"software":[63],"algorithm,":[64],"based":[65],"on":[66,93,122],"Hessian":[67],"analysis,":[68],"quantizes":[69],"important":[70],"rows":[71,77],"with":[72,78],"high":[73],"precision":[74,80],"unimportant":[76],"low":[79],"compress":[82],"parameters":[84,117],"effectively":[85],"while":[86],"maintaining":[87],"accuracy,":[88],"enabling":[89],"fine-grained":[90],"computation":[92],"FPGA":[95,140,145,159],"accelerator.":[96],"Moreover,":[97],"integration":[99],"our":[105,139],"energy-efficient":[106],"data":[107],"flow":[108],"optimization":[109],"scheme":[110],"enables":[111],"accommodation":[113],"all":[115],"necessary":[116],"BERT-base":[120],"FPGA,":[124],"eliminating":[125],"need":[127],"off-chip":[129],"access":[131],"during":[132],"runtime.":[133],"experimental":[135],"results":[136],"demonstrate":[137],"that":[138],"accelerator":[141],"generally":[142],"outperforms":[143],"existing":[144,158],"accelerators,":[146],"exhibiting":[147],"improvements":[150],"ranging":[151],"from":[152],"4.12X":[153],"14.57X":[155],"compared":[156],"accelerators.":[160]},"counts_by_year":[{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}