{"id":"https://openalex.org/W4375928954","doi":"https://doi.org/10.1109/tcad.2023.3273992","title":"AccelTran: A Sparsity-Aware Accelerator for Dynamic Inference With Transformers","display_name":"AccelTran: A Sparsity-Aware Accelerator for Dynamic Inference With Transformers","publication_year":2023,"publication_date":"2023-05-08","ids":{"openalex":"https://openalex.org/W4375928954","doi":"https://doi.org/10.1109/tcad.2023.3273992"},"language":"en","primary_location":{"id":"doi:10.1109/tcad.2023.3273992","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2023.3273992","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-article"},"type":"article","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/A5011008635","display_name":"Shikhar Tuli","orcid":"https://orcid.org/0000-0002-9230-5877"},"institutions":[{"id":"https://openalex.org/I20089843","display_name":"Princeton University","ror":"https://ror.org/00hx57361","country_code":"US","type":"education","lineage":["https://openalex.org/I20089843"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Shikhar Tuli","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Princeton University, Princeton, NJ, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Princeton University, Princeton, NJ, USA","institution_ids":["https://openalex.org/I20089843"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5086131079","display_name":"Niraj K. Jha","orcid":"https://orcid.org/0000-0002-1539-0369"},"institutions":[{"id":"https://openalex.org/I20089843","display_name":"Princeton University","ror":"https://ror.org/00hx57361","country_code":"US","type":"education","lineage":["https://openalex.org/I20089843"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Niraj K. Jha","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Princeton University, Princeton, NJ, USA"],"affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Princeton University, Princeton, NJ, USA","institution_ids":["https://openalex.org/I20089843"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5011008635"],"corresponding_institution_ids":["https://openalex.org/I20089843"],"apc_list":null,"apc_paid":null,"fwci":6.9066,"has_fulltext":false,"cited_by_count":40,"citation_normalized_percentile":{"value":0.97611536,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":97,"max":100},"biblio":{"volume":"42","issue":"11","first_page":"4038","last_page":"4051"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","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/T10028","display_name":"Topic Modeling","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":0.9972000122070312,"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":0.9954000115394592,"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/inference","display_name":"Inference","score":0.7516499161720276},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6579918265342712},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5800614356994629},{"id":"https://openalex.org/keywords/notation","display_name":"Notation","score":0.5614162683486938},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.4600306749343872},{"id":"https://openalex.org/keywords/quadratic-equation","display_name":"Quadratic equation","score":0.432952880859375},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3688458800315857},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.32953792810440063},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.3241633176803589},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2199820578098297},{"id":"https://openalex.org/keywords/arithmetic","display_name":"Arithmetic","score":0.20114931464195251},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.18722844123840332},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.09490576386451721},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.0879659354686737}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7516499161720276},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6579918265342712},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5800614356994629},{"id":"https://openalex.org/C45357846","wikidata":"https://www.wikidata.org/wiki/Q2001982","display_name":"Notation","level":2,"score":0.5614162683486938},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4600306749343872},{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.432952880859375},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3688458800315857},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.32953792810440063},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3241633176803589},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2199820578098297},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.20114931464195251},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.18722844123840332},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.09490576386451721},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0879659354686737},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","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:10.1109/tcad.2023.3273992","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2023.3273992","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Affordable and clean energy","score":0.9100000262260437,"id":"https://metadata.un.org/sdg/7"}],"awards":[{"id":"https://openalex.org/G6011117703","display_name":null,"funder_award_id":"CCF-2203399","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":65,"referenced_works":["https://openalex.org/W1964678809","https://openalex.org/W1989931041","https://openalex.org/W2010202670","https://openalex.org/W2033493616","https://openalex.org/W2041811089","https://openalex.org/W2048266589","https://openalex.org/W2064675550","https://openalex.org/W2204664557","https://openalex.org/W2289252105","https://openalex.org/W2588191434","https://openalex.org/W2626211758","https://openalex.org/W2766736793","https://openalex.org/W2805634098","https://openalex.org/W2880869856","https://openalex.org/W2896457183","https://openalex.org/W2904902077","https://openalex.org/W2923014074","https://openalex.org/W2963748441","https://openalex.org/W2969601108","https://openalex.org/W2971033911","https://openalex.org/W2971381503","https://openalex.org/W2978017171","https://openalex.org/W3017024317","https://openalex.org/W3033529678","https://openalex.org/W3035251378","https://openalex.org/W3036120435","https://openalex.org/W3038012435","https://openalex.org/W3047848469","https://openalex.org/W3085139254","https://openalex.org/W3094502228","https://openalex.org/W3099513244","https://openalex.org/W3099576124","https://openalex.org/W3102446692","https://openalex.org/W3104894901","https://openalex.org/W3134756900","https://openalex.org/W3159727696","https://openalex.org/W3162542754","https://openalex.org/W3192478068","https://openalex.org/W3196923642","https://openalex.org/W3205192296","https://openalex.org/W3210705689","https://openalex.org/W3213229701","https://openalex.org/W4206634569","https://openalex.org/W4220912491","https://openalex.org/W4221167110","https://openalex.org/W4223424212","https://openalex.org/W4240168186","https://openalex.org/W4251575795","https://openalex.org/W4300171661","https://openalex.org/W4310997605","https://openalex.org/W4372306494","https://openalex.org/W4385245566","https://openalex.org/W6638783484","https://openalex.org/W6718683173","https://openalex.org/W6758657797","https://openalex.org/W6766795243","https://openalex.org/W6768851824","https://openalex.org/W6771713106","https://openalex.org/W6777017071","https://openalex.org/W6778883912","https://openalex.org/W6779163297","https://openalex.org/W6780125659","https://openalex.org/W6780226713","https://openalex.org/W6810296985","https://openalex.org/W6810362000"],"related_works":["https://openalex.org/W2504004674","https://openalex.org/W1987679298","https://openalex.org/W2963177394","https://openalex.org/W4313359513","https://openalex.org/W2498744856","https://openalex.org/W4390482104","https://openalex.org/W4206178588","https://openalex.org/W3094491777","https://openalex.org/W3214715529","https://openalex.org/W4287635093"],"abstract_inverted_index":{"Self-attention-based":[0],"transformer":[1,19,33,101,110,244],"models":[2,143],"have":[3],"achieved":[4],"tremendous":[5],"success":[6],"in":[7,58,109],"the":[8,18,59,91,98,107,153,215,242],"domain":[9],"of":[10,93,100,179],"natural":[11],"language":[12],"processing.":[13],"Despite":[14],"their":[15],"efficacy,":[16],"accelerating":[17],"is":[20,251],"challenging":[21],"due":[22],"to":[23,36,40,68,116,168,209,241],"its":[24,38],"quadratic":[25],"computational":[26],"complexity":[27],"and":[28,144,229],"large":[29,55],"activation":[30],"sizes.":[31],"Existing":[32],"accelerators":[34],"attempt":[35],"prune":[37],"tokens":[39],"reduce":[41],"memory":[42],"access,":[43],"albeit":[44],"with":[45,86,113,141,195],"high":[46],"compute":[47],"overheads.":[48],"Moreover,":[49],"previous":[50],"works":[51],"directly":[52],"operate":[53],"on":[54],"matrices":[56,108],"involved":[57],"attention":[60],"operation,":[61],"which":[62,81],"limits":[63],"hardware":[64],"utilization.":[65],"In":[66],"order":[67],"address":[69],"these":[70,128],"challenges,":[71],"this":[72],"work":[73],"proposes":[74],"a":[75,133,210],"novel":[76,134],"dynamic":[77],"inference":[78],"scheme,":[79],"DynaTran,":[80],"prunes":[82],"activations":[83],"at":[84,253],"runtime":[85],"low":[87],"overhead,":[88],"substantially":[89],"reducing":[90],"number":[92],"ineffectual":[94],"operations.":[95],"This":[96],"improves":[97],"throughput":[99,194,228],"inference.":[102],"We":[103],"further":[104],"propose":[105,131],"tiling":[106],"operations":[111],"along":[112],"diverse":[114],"dataflows":[115],"improve":[117],"data":[118],"reuse,":[119],"thus,":[120],"enabling":[121],"higher":[122,150,176,193,227],"energy":[123,205,238],"efficiency.":[124],"To":[125],"effectively":[126],"implement":[127],"methods,":[129],"we":[130],"AccelTran,":[132],"accelerator":[135],"architecture":[136],"for":[137],"transformers.":[138],"Extensive":[139],"experiments":[140],"different":[142],"benchmarks":[145],"demonstrate":[146],"that":[147],"DynaTran":[148],"achieves":[149,184,219],"accuracy":[151],"than":[152],"state-of-the-art":[154,243],"top-":[155],"<inline-formula":[156,169,186,197,220,230],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[157,170,187,198,221,231,255],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[158,171,188,199,222,232],"<tex-math":[159,172,189,200,223,233],"notation=\"LaTeX\">$k$":[160],"</tex-math></inline-formula>":[161,175,192,203,226,236],"hardware-aware":[162],"pruning":[163],"strategy":[164],"while":[165],"attaining":[166],"up":[167],"notation=\"LaTeX\">$1.2\\times":[173],"$":[174,191,202,225,235],"sparsity.":[177],"One":[178],"our":[180],"proposed":[181],"accelerators,":[182],"AccelTran-Edge,":[183],"330K":[185],"notation=\"LaTeX\">$\\times":[190,201],"93K":[196],"lower":[204,237],"requirement":[206],"when":[207],"compared":[208,240],"Raspberry":[211],"Pi":[212],"device.":[213],"On":[214],"other":[216],"hand,":[217],"AccelTran-Server":[218],"notation=\"LaTeX\">$5.73\\times":[224],"notation=\"LaTeX\">$3.69\\times":[234],"consumption":[239],"co-processor,":[245],"Energon.":[246],"The":[247],"simulation":[248],"source":[249],"code":[250],"available":[252],"<uri":[254],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">https://github.com/jha-lab/acceltran</uri>":[256],".":[257]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":20},{"year":2024,"cited_by_count":13},{"year":2023,"cited_by_count":4}],"updated_date":"2026-04-11T08:14:18.477133","created_date":"2025-10-10T00:00:00"}