{"id":"https://openalex.org/W4411171608","doi":"https://doi.org/10.1109/tcsi.2025.3576232","title":"CLAT: A Clustering-Based Attention Transformer Accelerator for Low-Latency Text Generation in LLMs","display_name":"CLAT: A Clustering-Based Attention Transformer Accelerator for Low-Latency Text Generation in LLMs","publication_year":2025,"publication_date":"2025-06-10","ids":{"openalex":"https://openalex.org/W4411171608","doi":"https://doi.org/10.1109/tcsi.2025.3576232"},"language":"en","primary_location":{"id":"doi:10.1109/tcsi.2025.3576232","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsi.2025.3576232","pdf_url":null,"source":{"id":"https://openalex.org/S116977442","display_name":"IEEE Transactions on Circuits and Systems I Regular Papers","issn_l":"1549-8328","issn":["1549-8328","1558-0806"],"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 Circuits and Systems I: Regular Papers","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/A5100698555","display_name":"Sunwoo Lee","orcid":"https://orcid.org/0000-0001-7760-0168"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Sunwoo Lee","raw_affiliation_strings":["Department of Intelligence and Information, the Research Institute for Convergence Science, The Inter-University Semiconductor Research Center, Seoul National University, Seoul, South Korea"],"raw_orcid":"https://orcid.org/0000-0001-7760-0168","affiliations":[{"raw_affiliation_string":"Department of Intelligence and Information, the Research Institute for Convergence Science, The Inter-University Semiconductor Research Center, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030276609","display_name":"Beomseok Kim","orcid":null},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Beomseok Kim","raw_affiliation_strings":["Department of Intelligence and Information, the Research Institute for Convergence Science, The Inter-University Semiconductor Research Center, Seoul National University, Seoul, South Korea"],"raw_orcid":"https://orcid.org/0009-0007-8824-7997","affiliations":[{"raw_affiliation_string":"Department of Intelligence and Information, the Research Institute for Convergence Science, The Inter-University Semiconductor Research Center, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089174485","display_name":"Jeongwoo Park","orcid":"https://orcid.org/0000-0002-9603-9588"},"institutions":[{"id":"https://openalex.org/I848706","display_name":"Sungkyunkwan University","ror":"https://ror.org/04q78tk20","country_code":"KR","type":"education","lineage":["https://openalex.org/I848706"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jeongwoo Park","raw_affiliation_strings":["Electrical and Computer Engineering Department, Sungkyunkwan University, Suwon, South Korea"],"raw_orcid":"https://orcid.org/0000-0002-9603-9588","affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering Department, Sungkyunkwan University, Suwon, South Korea","institution_ids":["https://openalex.org/I848706"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5053248335","display_name":"Dongsuk Jeon","orcid":"https://orcid.org/0000-0002-0395-8076"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Dongsuk Jeon","raw_affiliation_strings":["Department of Intelligence and Information, the Research Institute for Convergence Science, The Inter-University Semiconductor Research Center, Seoul National University, Seoul, South Korea"],"raw_orcid":"https://orcid.org/0000-0002-0395-8076","affiliations":[{"raw_affiliation_string":"Department of Intelligence and Information, the Research Institute for Convergence Science, The Inter-University Semiconductor Research Center, Seoul National University, Seoul, South Korea","institution_ids":["https://openalex.org/I139264467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5100698555"],"corresponding_institution_ids":["https://openalex.org/I139264467"],"apc_list":null,"apc_paid":null,"fwci":8.693,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.97297923,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":91,"max":99},"biblio":{"volume":"72","issue":"11","first_page":"6553","last_page":"6566"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9972000122070312,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9972000122070312,"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":0.9854000210762024,"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/T13629","display_name":"Text Readability and Simplification","score":0.9398000240325928,"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","score":0.6598271727561951},{"id":"https://openalex.org/keywords/cluster-analysis","display_name":"Cluster analysis","score":0.5453547835350037},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5021781921386719},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.3881785571575165},{"id":"https://openalex.org/keywords/electronic-engineering","display_name":"Electronic engineering","score":0.32819271087646484},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1982046365737915},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.18008625507354736},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.10635790228843689}],"concepts":[{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6598271727561951},{"id":"https://openalex.org/C73555534","wikidata":"https://www.wikidata.org/wiki/Q622825","display_name":"Cluster analysis","level":2,"score":0.5453547835350037},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5021781921386719},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.3881785571575165},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.32819271087646484},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1982046365737915},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.18008625507354736},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.10635790228843689}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcsi.2025.3576232","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsi.2025.3576232","pdf_url":null,"source":{"id":"https://openalex.org/S116977442","display_name":"IEEE Transactions on Circuits and Systems I Regular Papers","issn_l":"1549-8328","issn":["1549-8328","1558-0806"],"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 Circuits and Systems I: Regular Papers","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","score":0.46000000834465027,"display_name":"Affordable and clean energy"}],"awards":[{"id":"https://openalex.org/G2411785074","display_name":null,"funder_award_id":"RS-2025-02218733","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G3369221025","display_name":null,"funder_award_id":"RS-2023-00228970","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G3598079552","display_name":null,"funder_award_id":"RS-2021-II211343","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G4294718623","display_name":null,"funder_award_id":"RS-2024-00347394","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G5791007231","display_name":null,"funder_award_id":"RS-2024-00355678","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G7934916028","display_name":null,"funder_award_id":"IITP-2023-RS-2023-00256081","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"}],"funders":[{"id":"https://openalex.org/F4320335489","display_name":"Institute for Information and Communications Technology Promotion","ror":"https://ror.org/01g0hqq23"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":26,"referenced_works":["https://openalex.org/W1790954942","https://openalex.org/W2798405286","https://openalex.org/W2965373594","https://openalex.org/W2979826702","https://openalex.org/W2984100107","https://openalex.org/W3017024317","https://openalex.org/W3096609285","https://openalex.org/W3130554079","https://openalex.org/W3131922516","https://openalex.org/W3138516171","https://openalex.org/W3159727696","https://openalex.org/W3189877953","https://openalex.org/W3205192296","https://openalex.org/W3206453033","https://openalex.org/W4206289265","https://openalex.org/W4214686755","https://openalex.org/W4308083513","https://openalex.org/W4312358791","https://openalex.org/W4313021454","https://openalex.org/W4380881077","https://openalex.org/W4385245566","https://openalex.org/W4393578753","https://openalex.org/W4394625766","https://openalex.org/W4400448267","https://openalex.org/W4402670692","https://openalex.org/W4404954358"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W2864363823"],"abstract_inverted_index":{"Transformer-based":[0],"large":[1,17],"language":[2],"models":[3],"(LLMs)":[4],"excel":[5],"in":[6],"text":[7,139,149],"generation":[8,140,150],"but":[9,38],"face":[10],"challenges":[11],"like":[12],"memory":[13,81],"bandwidth":[14,82],"bottlenecks":[15],"and":[16,35,88,105,125,137],"key-value":[18],"(KV)":[19],"cache":[20],"sizes":[21],"as":[22],"context":[23,51],"lengths":[24],"grow,":[25],"impacting":[26],"low-latency":[27],"performance.":[28],"Existing":[29],"accelerators":[30],"adopt":[31],"parallelism,":[32],"model":[33,100],"compression,":[34],"sparsity":[36],"exploitation":[37],"often":[39],"fail":[40],"to":[41,92,102,108,135,144,160],"fully":[42],"utilize":[43],"token-specific":[44],"sparsity,":[45],"limiting":[46],"their":[47],"effectiveness":[48],"for":[49,68,85],"long":[50],"lengths.":[52],"CLAT":[53,98,128],"addresses":[54],"these":[55],"issues":[56],"with":[57,76],"a":[58,113],"low-overhead":[59],"clustering":[60],"algorithm":[61],"that":[62,117],"identifies":[63],"relevant":[64,74],"key":[65],"vector":[66],"clusters":[67],"each":[69],"token\u2019s":[70],"query,":[71],"omitting":[72],"less":[73],"vectors":[75],"minimal":[77],"impact.":[78],"It":[79,146],"optimizes":[80],"using":[83],"routing":[84],"single-batch":[86,148],"inference":[87],"introduces":[89],"scheduling":[90],"techniques":[91],"reduce":[93],"attention":[94,130],"layer":[95,131],"latency.":[96],"Additionally,":[97],"compresses":[99],"parameters":[101],"4-bit":[103],"precision":[104],"KV":[106],"caches":[107],"8-bit":[109],"precision,":[110],"supported":[111],"by":[112,133,142,152],"multi-precision":[114],"MAC":[115],"structure":[116],"avoids":[118],"extra":[119],"overhead.":[120],"Validated":[121],"on":[122],"Llama2-7B,":[123],"OPT-6.7B,":[124],"Llama3-8B":[126],"models,":[127],"reduces":[129],"latency":[132,141],"up":[134,143],"88.6%":[136],"overall":[138],"34.9%.":[145],"improves":[147],"throughput":[151],"<inline-formula":[153,161],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[154,162],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[155,163],"<tex-math":[156,164],"notation=\"LaTeX\">$1.66\\times":[157],"$</tex-math>":[158,166],"</inline-formula>":[159,167],"notation=\"LaTeX\">$2.42\\times":[165],"over":[168],"an":[169],"A100":[170],"GPU,":[171],"demonstrating":[172],"significant":[173],"performance":[174],"gains.":[175]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}