{"id":"https://openalex.org/W4408345918","doi":"https://doi.org/10.1109/icassp49660.2025.10889000","title":"KVPruner: Structural Pruning for Faster and Memory-Efficient Large Language Models","display_name":"KVPruner: Structural Pruning for Faster and Memory-Efficient Large Language Models","publication_year":2025,"publication_date":"2025-03-12","ids":{"openalex":"https://openalex.org/W4408345918","doi":"https://doi.org/10.1109/icassp49660.2025.10889000"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49660.2025.10889000","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10889000","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-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/A5101655322","display_name":"Bo Lv","orcid":"https://orcid.org/0000-0002-3448-8986"},"institutions":[{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bo Lv","raw_affiliation_strings":["Huazhong University of Science and Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Huazhong University of Science and Technology","institution_ids":["https://openalex.org/I47720641"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5051170836","display_name":"Quan Zhou","orcid":"https://orcid.org/0000-0002-9351-5614"},"institutions":[{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Quan Zhou","raw_affiliation_strings":["Huazhong University of Science and Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Huazhong University of Science and Technology","institution_ids":["https://openalex.org/I47720641"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044430721","display_name":"Xuanang Ding","orcid":"https://orcid.org/0000-0001-9512-7044"},"institutions":[{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xuanang Ding","raw_affiliation_strings":["Huazhong University of Science and Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Huazhong University of Science and Technology","institution_ids":["https://openalex.org/I47720641"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5118092688","display_name":"Yan Wang","orcid":"https://orcid.org/0009-0009-0503-4879"},"institutions":[{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yan Wang","raw_affiliation_strings":["Huazhong University of Science and Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Huazhong University of Science and Technology","institution_ids":["https://openalex.org/I47720641"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5047143528","display_name":"Zeyu Ma","orcid":"https://orcid.org/0000-0002-1846-8889"},"institutions":[{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zeming Ma","raw_affiliation_strings":["Huazhong University of Science and Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Huazhong University of Science and Technology","institution_ids":["https://openalex.org/I47720641"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I47720641"],"apc_list":null,"apc_paid":null,"fwci":1.6508,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.85010229,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9782999753952026,"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":0.9782999753952026,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.972000002861023,"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/computer-science","display_name":"Computer science","score":0.7837077975273132},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6356821656227112},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5309749841690063},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.42693257331848145},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.34575456380844116},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3291111886501312}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7837077975273132},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6356821656227112},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5309749841690063},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.42693257331848145},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.34575456380844116},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3291111886501312},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49660.2025.10889000","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10889000","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":33,"referenced_works":["https://openalex.org/W3092879656","https://openalex.org/W3138516171","https://openalex.org/W3159481202","https://openalex.org/W3206996142","https://openalex.org/W4249013746","https://openalex.org/W4372269772","https://openalex.org/W4389520322","https://openalex.org/W4392903033","https://openalex.org/W4392931430","https://openalex.org/W4393160747","https://openalex.org/W4401857375","https://openalex.org/W4402671307","https://openalex.org/W4411403450","https://openalex.org/W6739901393","https://openalex.org/W6755207826","https://openalex.org/W6763832098","https://openalex.org/W6777615688","https://openalex.org/W6784333009","https://openalex.org/W6796581206","https://openalex.org/W6838322825","https://openalex.org/W6846164622","https://openalex.org/W6848451824","https://openalex.org/W6850625674","https://openalex.org/W6852874933","https://openalex.org/W6852962002","https://openalex.org/W6854094408","https://openalex.org/W6854308872","https://openalex.org/W6856574389","https://openalex.org/W6859846904","https://openalex.org/W6860155063","https://openalex.org/W6861391505","https://openalex.org/W6869820326","https://openalex.org/W6877022392"],"related_works":["https://openalex.org/W2373300491","https://openalex.org/W2395294869","https://openalex.org/W2378744544","https://openalex.org/W2594301978","https://openalex.org/W2379704676","https://openalex.org/W1998810860","https://openalex.org/W4206442282","https://openalex.org/W2384505857","https://openalex.org/W2355171581","https://openalex.org/W3204019825"],"abstract_inverted_index":{"The":[0],"bottleneck":[1],"associated":[2],"with":[3],"the":[4,12,37,45,74,93,127],"key-value(KV)":[5],"cache":[6],"presents":[7],"a":[8],"significant":[9],"challenge":[10],"during":[11],"inference":[13],"processes":[14],"of":[15,44,116,126],"large":[16],"language":[17],"models.":[18],"While":[19],"depth":[20],"pruning":[21,41],"accelerates":[22],"inference,":[23],"it":[24],"requires":[25,112],"extensive":[26],"recovery":[27],"training,":[28],"which":[29],"can":[30],"take":[31],"up":[32],"to":[33,59,72,84,92,123],"two":[34,114],"weeks.":[35],"On":[36],"other":[38],"hand,":[39],"width":[40],"retains":[42],"much":[43],"performance":[46],"but":[47],"offers":[48],"slight":[49],"speed":[50],"gains.":[51],"To":[52],"tackle":[53],"these":[54],"challenges,":[55],"we":[56],"propose":[57],"KVPruner":[58,96],"improve":[60],"model":[61],"efficiency":[62],"while":[63],"maintaining":[64],"performance.":[65,128],"Our":[66],"method":[67,111],"uses":[68],"global":[69],"perplexity-based":[70],"analysis":[71],"determine":[73],"importance":[75],"ratio":[76],"for":[77],"each":[78],"block":[79],"and":[80,103],"provides":[81],"multiple":[82],"strategies":[83],"prune":[85],"non-essential":[86],"KV":[87],"channels":[88],"within":[89],"blocks.":[90],"Compared":[91],"original":[94],"model,":[95],"reduces":[97],"runtime":[98],"memory":[99],"usage":[100],"by":[101,106],"50%":[102],"boosts":[104],"throughput":[105],"over":[107],"35%.":[108],"Additionally,":[109],"our":[110],"only":[113],"hours":[115],"LoRA":[117],"fine-tuning":[118],"techniques":[119],"on":[120],"small":[121],"datasets":[122],"recover":[124],"most":[125]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2025-10-10T00:00:00"}