{"id":"https://openalex.org/W7133557855","doi":"https://doi.org/10.1109/hpca68181.2026.11408453","title":"GyRot: Leveraging Hidden Synergy Between Rotation and Fine-Grained Group Quantization for Low-Bit LLM Inference","display_name":"GyRot: Leveraging Hidden Synergy Between Rotation and Fine-Grained Group Quantization for Low-Bit LLM Inference","publication_year":2026,"publication_date":"2026-01-31","ids":{"openalex":"https://openalex.org/W7133557855","doi":"https://doi.org/10.1109/hpca68181.2026.11408453"},"language":null,"primary_location":{"id":"doi:10.1109/hpca68181.2026.11408453","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408453","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","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/A5000767990","display_name":"S. -P. Kim","orcid":null},"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"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sangjin Kim","raw_affiliation_strings":["School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128111706","display_name":"Yuseon Chou","orcid":null},"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"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Yuseon Choi","raw_affiliation_strings":["School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5091616745","display_name":"Byeongcheol Kim","orcid":"https://orcid.org/0000-0003-1079-7220"},"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"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Byeongcheol Kim","raw_affiliation_strings":["School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046949698","display_name":"Ji Hye Oh","orcid":"https://orcid.org/0000-0003-3510-980X"},"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"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jungjun Oh","raw_affiliation_strings":["School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5126396423","display_name":"Hoi-Jun Yoo","orcid":null},"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"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hoi-Jun Yoo","raw_affiliation_strings":["School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electrical Engineering, Korea Advanced Institue of Science and Technology (KAIST)","institution_ids":["https://openalex.org/I157485424"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157485424"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.19948095,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"15"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.10970000177621841,"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":0.10970000177621841,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.04190000146627426,"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/T10048","display_name":"Particle physics theoretical and experimental studies","score":0.037300001829862595,"subfield":{"id":"https://openalex.org/subfields/3106","display_name":"Nuclear and High Energy Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.4896000027656555},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.4830999970436096},{"id":"https://openalex.org/keywords/rotation","display_name":"Rotation (mathematics)","score":0.4171999990940094},{"id":"https://openalex.org/keywords/group","display_name":"Group (periodic table)","score":0.3725000023841858},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.32679998874664307}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.564300000667572},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5357000231742859},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4896000027656555},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.4830999970436096},{"id":"https://openalex.org/C74050887","wikidata":"https://www.wikidata.org/wiki/Q848368","display_name":"Rotation (mathematics)","level":2,"score":0.4171999990940094},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3921999931335449},{"id":"https://openalex.org/C2781311116","wikidata":"https://www.wikidata.org/wiki/Q83306","display_name":"Group (periodic table)","level":2,"score":0.3725000023841858},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.32679998874664307},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3059999942779541},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C104267543","wikidata":"https://www.wikidata.org/wiki/Q208163","display_name":"Signal processing","level":3,"score":0.25279998779296875},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.2524999976158142}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/hpca68181.2026.11408453","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408453","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","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":21,"referenced_works":["https://openalex.org/W2883920103","https://openalex.org/W2946609015","https://openalex.org/W2965373594","https://openalex.org/W2998617917","https://openalex.org/W3100985894","https://openalex.org/W3194676777","https://openalex.org/W4308083739","https://openalex.org/W4366341968","https://openalex.org/W4380874652","https://openalex.org/W4388979610","https://openalex.org/W4401211807","https://openalex.org/W4402671081","https://openalex.org/W4408183360","https://openalex.org/W4409248530","https://openalex.org/W4409248600","https://openalex.org/W4409248709","https://openalex.org/W4409261373","https://openalex.org/W4411486471","https://openalex.org/W4415796210","https://openalex.org/W4415796640","https://openalex.org/W7133239970"],"related_works":[],"abstract_inverted_index":{"Low-bit":[0],"quantization":[1,14,51,98],"is":[2],"essential":[3],"for":[4,145],"efficient":[5],"LLM":[6,137,149],"inference,":[7],"and":[8,11,40,53,70,80,99,131,147],"both":[9],"rotation":[10,39,79],"fine-grained":[12],"group":[13,45,81],"have":[15],"shown":[16],"individual":[17],"promise.":[18],"However,":[19],"their":[20],"combination":[21],"often":[22],"leads":[23],"to":[24,31,74,128],"accuracy":[25,121],"degradation":[26],"or":[27],"hardware":[28,54,93],"overhead":[29],"due":[30],"a":[32,50,101],"mismatch":[33],"between":[34],"the":[35,41],"global":[36],"nature":[37],"of":[38,44,78],"localized":[42],"behavior":[43],"scaling.":[46],"We":[47],"propose":[48],"GyRot,":[49],"framework":[52],"accelerator":[55],"that":[56,105],"bridges":[57],"this":[58],"gap":[59],"through":[60],"algorithm-hardware":[61],"co-design.":[62],"GyRot":[63,117],"introduces":[64],"Coarse":[65],"Rotation,":[66],"Fine":[67],"Grouping":[68],"(CoRFiG)":[69],"Harmonic-Aligned":[71],"Permutation":[72],"(HAP)":[73],"enable":[75],"cooperative":[76],"integration":[77],"quantization,":[82],"enhancing":[83],"quantizability":[84],"while":[85,125],"relaxing":[86],"scaling":[87],"factor":[88],"precision.":[89],"To":[90],"further":[91],"reduce":[92],"cost,":[94],"we":[95],"reformulate":[96],"asymmetric":[97],"introduce":[100],"zero-point":[102],"rounding":[103],"strategy":[104],"enables":[106],"fully":[107],"integer":[108],"dequantization.":[109],"Implemented":[110],"on":[111],"an":[112],"INT4-based":[113],"tensor":[114],"PE":[115],"architecture,":[116],"achieves":[118],"state-of-the-art":[119],"4-bit":[120],"across":[122],"LLaMA-family":[123],"models,":[124],"delivering":[126],"up":[127],"3.4\u00d7":[129],"speedup":[130],"3.6\u00d7":[132],"energy":[133],"efficiency":[134],"over":[135],"baseline":[136],"accelerators.":[138],"These":[139],"results":[140],"validate":[141],"GyRot's":[142],"practical":[143],"effectiveness":[144],"scalable":[146],"energy-efficient":[148],"deployment.":[150]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-03-05T00:00:00"}