{"id":"https://openalex.org/W4409261373","doi":"https://doi.org/10.1109/jetcas.2025.3558300","title":"LightRot: A Light-Weighted Rotation Scheme and Architecture for Accurate Low-Bit Large Language Model Inference","display_name":"LightRot: A Light-Weighted Rotation Scheme and Architecture for Accurate Low-Bit Large Language Model Inference","publication_year":2025,"publication_date":"2025-04-08","ids":{"openalex":"https://openalex.org/W4409261373","doi":"https://doi.org/10.1109/jetcas.2025.3558300"},"language":"en","primary_location":{"id":"doi:10.1109/jetcas.2025.3558300","is_oa":false,"landing_page_url":"https://doi.org/10.1109/jetcas.2025.3558300","pdf_url":null,"source":{"id":"https://openalex.org/S142323794","display_name":"IEEE Journal on Emerging and Selected Topics in Circuits and Systems","issn_l":"2156-3357","issn":["2156-3357","2156-3365"],"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 Journal on Emerging and Selected Topics in 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/A5100748581","display_name":"Sangjin Kim","orcid":"https://orcid.org/0000-0001-6665-9973"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]},{"id":"https://openalex.org/I4210089054","display_name":"Seoul Semiconductor (South Korea)","ror":"https://ror.org/007vmjm88","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210089054"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Sangjin Kim","raw_affiliation_strings":["PIM Semiconductor Design Research Center (AI-PIM), Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0000-0001-6665-9973","affiliations":[{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea","institution_ids":["https://openalex.org/I4210089054"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033586441","display_name":"Yuseon Choi","orcid":"https://orcid.org/0009-0008-0698-8116"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]},{"id":"https://openalex.org/I4210089054","display_name":"Seoul Semiconductor (South Korea)","ror":"https://ror.org/007vmjm88","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210089054"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Yuseon Choi","raw_affiliation_strings":["Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0009-0008-0698-8116","affiliations":[{"raw_affiliation_string":"Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea","institution_ids":["https://openalex.org/I4210089054"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Jungjun Oh","orcid":"https://orcid.org/0009-0006-3998-7385"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]},{"id":"https://openalex.org/I4210089054","display_name":"Seoul Semiconductor (South Korea)","ror":"https://ror.org/007vmjm88","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210089054"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jungjun Oh","raw_affiliation_strings":["Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0009-0006-3998-7385","affiliations":[{"raw_affiliation_string":"Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea","institution_ids":["https://openalex.org/I4210089054"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Byeongcheol Kim","orcid":"https://orcid.org/0009-0003-8354-2170"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]},{"id":"https://openalex.org/I4210089054","display_name":"Seoul Semiconductor (South Korea)","ror":"https://ror.org/007vmjm88","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210089054"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Byeongcheol Kim","raw_affiliation_strings":["Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0009-0003-8354-2170","affiliations":[{"raw_affiliation_string":"Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea","institution_ids":["https://openalex.org/I4210089054"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077896259","display_name":"Hoi\u2010Jun Yoo","orcid":"https://orcid.org/0000-0002-6661-4879"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]},{"id":"https://openalex.org/I4210089054","display_name":"Seoul Semiconductor (South Korea)","ror":"https://ror.org/007vmjm88","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210089054"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hoi-Jun Yoo","raw_affiliation_strings":["PIM Semiconductor Design Research Center (AI-PIM), Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0000-0002-6661-4879","affiliations":[{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"PIM Semiconductor Design Research Center (AI-PIM), Daejeon, South Korea","institution_ids":["https://openalex.org/I4210089054"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5100748581"],"corresponding_institution_ids":["https://openalex.org/I157485424","https://openalex.org/I4210089054"],"apc_list":null,"apc_paid":null,"fwci":6.5198,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":{"value":0.95899577,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":99},"biblio":{"volume":"15","issue":"2","first_page":"231","last_page":"243"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9980999827384949,"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":0.9980999827384949,"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":0.996399998664856,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9797000288963318,"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/inference","display_name":"Inference","score":0.6797610521316528},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5959027409553528},{"id":"https://openalex.org/keywords/scheme","display_name":"Scheme (mathematics)","score":0.5894419550895691},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.5638846158981323},{"id":"https://openalex.org/keywords/rotation","display_name":"Rotation (mathematics)","score":0.5591917634010315},{"id":"https://openalex.org/keywords/bit","display_name":"Bit (key)","score":0.5579911470413208},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.46604812145233154},{"id":"https://openalex.org/keywords/arithmetic","display_name":"Arithmetic","score":0.37362515926361084},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.35350799560546875},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3017999529838562},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.26166391372680664},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.11027243733406067},{"id":"https://openalex.org/keywords/mathematical-analysis","display_name":"Mathematical analysis","score":0.10388603806495667}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6797610521316528},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5959027409553528},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.5894419550895691},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.5638846158981323},{"id":"https://openalex.org/C74050887","wikidata":"https://www.wikidata.org/wiki/Q848368","display_name":"Rotation (mathematics)","level":2,"score":0.5591917634010315},{"id":"https://openalex.org/C117011727","wikidata":"https://www.wikidata.org/wiki/Q1278488","display_name":"Bit (key)","level":2,"score":0.5579911470413208},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.46604812145233154},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.37362515926361084},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.35350799560546875},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3017999529838562},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.26166391372680664},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.11027243733406067},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.10388603806495667},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/jetcas.2025.3558300","is_oa":false,"landing_page_url":"https://doi.org/10.1109/jetcas.2025.3558300","pdf_url":null,"source":{"id":"https://openalex.org/S142323794","display_name":"IEEE Journal on Emerging and Selected Topics in Circuits and Systems","issn_l":"2156-3357","issn":["2156-3357","2156-3365"],"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 Journal on Emerging and Selected Topics in Circuits and Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.4300000071525574}],"awards":[{"id":"https://openalex.org/G6815536168","display_name":null,"funder_award_id":"IO201207-07799-01","funder_id":"https://openalex.org/F4320332195","funder_display_name":"Samsung"}],"funders":[{"id":"https://openalex.org/F4320332195","display_name":"Samsung","ror":"https://ror.org/04w3jy968"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":25,"referenced_works":["https://openalex.org/W4220865834","https://openalex.org/W4319996342","https://openalex.org/W4360605545","https://openalex.org/W4387250389","https://openalex.org/W4387725429","https://openalex.org/W4392739437","https://openalex.org/W4393147284","https://openalex.org/W4396909834","https://openalex.org/W4406650295","https://openalex.org/W4408183360","https://openalex.org/W6727099177","https://openalex.org/W6773820404","https://openalex.org/W6778883912","https://openalex.org/W6847478871","https://openalex.org/W6852874933","https://openalex.org/W6853804809","https://openalex.org/W6854866820","https://openalex.org/W6857799723","https://openalex.org/W6860710830","https://openalex.org/W6861393904","https://openalex.org/W6862909604","https://openalex.org/W6866033557","https://openalex.org/W6868720932","https://openalex.org/W6873648542","https://openalex.org/W6874991662"],"related_works":["https://openalex.org/W4327546585","https://openalex.org/W2411923897","https://openalex.org/W4394546135","https://openalex.org/W4285347720","https://openalex.org/W4200259850","https://openalex.org/W2333831899","https://openalex.org/W2484894494","https://openalex.org/W2367385042","https://openalex.org/W4381186982","https://openalex.org/W2040781570"],"abstract_inverted_index":{"As":[0],"large":[1],"language":[2,114],"models":[3,124],"(LLMs)":[4],"continue":[5],"to":[6,64,140],"demonstrate":[7],"exceptional":[8],"capabilities":[9],"across":[10],"various":[11],"domains,":[12],"the":[13,72,170],"challenge":[14],"of":[15,75,92],"achieving":[16],"energy-efficient":[17],"and":[18,32,49,128,144,155],"accurate":[19],"inference":[20,109],"becomes":[21],"increasingly":[22],"critical.":[23],"This":[24],"work":[25,159],"presents":[26],"LightRot,":[27],"a":[28,56,83,88,161],"lightweight":[29],"rotation":[30,62,76],"scheme":[31],"dedicated":[33],"hardware":[34,156],"accelerator":[35],"designed":[36],"for":[37,95,122,147,164,172],"low-bit":[38,69,166],"LLM":[39,167],"inference.":[40],"The":[41,78],"proposed":[42,79],"architecture":[43],"integrates":[44],"Grouped":[45],"Local":[46],"Rotation":[47],"(GLR)":[48],"Outlier":[50],"Direction":[51],"Aligning":[52],"(ODA)":[53],"algorithms":[54],"with":[55],"hierarchical":[57],"Fast":[58],"Hadamard":[59],"Transform":[60],"(FHT)-based":[61],"unit":[63],"address":[65],"key":[66],"challenges":[67],"in":[68,82],"quantization,":[70],"including":[71],"energy":[73,90],"overhead":[74],"operations.":[77],"accelerator,":[80],"implemented":[81],"28nm":[84],"CMOS":[85],"process,":[86],"achieves":[87],"peak":[89],"efficiency":[91],"27.4":[93],"TOPS/W":[94],"4-bit":[96],"inference,":[97,168],"surpassing":[98],"prior":[99],"state-of-the-art":[100],"designs.":[101],"Unlike":[102],"conventional":[103],"approaches":[104],"that":[105],"rely":[106],"on":[107,112,135],"higher-precision":[108],"or":[110],"evaluate":[111],"basic":[113],"modeling":[115],"tasks":[116],"like":[117],"GPT-2,":[118],"LightRot":[119],"is":[120,132],"optimized":[121],"advanced":[123],"such":[125],"as":[126],"LLaMA2-13B":[127],"LLaMA3-8B.":[129],"Its":[130],"performance":[131],"further":[133],"validated":[134],"MT-Bench,":[136],"demonstrating":[137],"robust":[138],"applicability":[139],"real-world":[141],"conversational":[142],"scenarios":[143],"redefining":[145],"benchmarks":[146],"chat-based":[148],"AI":[149,174],"systems.":[150],"By":[151],"synergizing":[152],"algorithmic":[153],"innovations":[154],"efficiency,":[157],"this":[158],"sets":[160],"new":[162],"paradigm":[163],"scalable,":[165],"paving":[169],"way":[171],"sustainable":[173],"advancements.":[175]},"counts_by_year":[{"year":2026,"cited_by_count":3}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}