{"id":"https://openalex.org/W7166845075","doi":"https://doi.org/10.18653/v1/2026.acl-long.594","title":"HybridKV: Hybrid KV Cache Compression for Efficient Multimodal Large Language Model Inference","display_name":"HybridKV: Hybrid KV Cache Compression for Efficient Multimodal Large Language Model Inference","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166845075","doi":"https://doi.org/10.18653/v1/2026.acl-long.594"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.594","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.594","pdf_url":"https://aclanthology.org/2026.acl-long.594.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.594.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139747104","display_name":"Bowen Zeng","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156488","display_name":"Institute of Zoology of the Slovak Academy of Sciences","ror":"https://ror.org/05tm6j853","country_code":"SK","type":"facility","lineage":["https://openalex.org/I207624831","https://openalex.org/I4210156488"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","SK"],"is_corresponding":false,"raw_author_name":"Bowen Zeng","raw_affiliation_strings":["Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","The State Key Laboratory of Blockchain and Data Security , Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","institution_ids":["https://openalex.org/I4210156488"]},{"raw_affiliation_string":"The State Key Laboratory of Blockchain and Data Security , Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001812354","display_name":"Feiyang Ren","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156488","display_name":"Institute of Zoology of the Slovak Academy of Sciences","ror":"https://ror.org/05tm6j853","country_code":"SK","type":"facility","lineage":["https://openalex.org/I207624831","https://openalex.org/I4210156488"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","SK"],"is_corresponding":false,"raw_author_name":"Feiyang Ren","raw_affiliation_strings":["Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","The State Key Laboratory of Blockchain and Data Security , Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","institution_ids":["https://openalex.org/I4210156488"]},{"raw_affiliation_string":"The State Key Laboratory of Blockchain and Data Security , Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139834593","display_name":"Jun Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156488","display_name":"Institute of Zoology of the Slovak Academy of Sciences","ror":"https://ror.org/05tm6j853","country_code":"SK","type":"facility","lineage":["https://openalex.org/I207624831","https://openalex.org/I4210156488"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","SK"],"is_corresponding":false,"raw_author_name":"Jun Zhang","raw_affiliation_strings":["Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","The State Key Laboratory of Blockchain and Data Security , Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","institution_ids":["https://openalex.org/I4210156488"]},{"raw_affiliation_string":"The State Key Laboratory of Blockchain and Data Security , Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111925362","display_name":"Xiaoling Gu","orcid":null},"institutions":[{"id":"https://openalex.org/I50760025","display_name":"Hangzhou Dianzi University","ror":"https://ror.org/0576gt767","country_code":"CN","type":"education","lineage":["https://openalex.org/I50760025"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xiaoling Gu","raw_affiliation_strings":["Hangzhou Dianzi University , Hangzhou , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou Dianzi University , Hangzhou , China","institution_ids":["https://openalex.org/I50760025"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139721963","display_name":"Ke Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156488","display_name":"Institute of Zoology of the Slovak Academy of Sciences","ror":"https://ror.org/05tm6j853","country_code":"SK","type":"facility","lineage":["https://openalex.org/I207624831","https://openalex.org/I4210156488"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","SK"],"is_corresponding":false,"raw_author_name":"Ke Chen","raw_affiliation_strings":["Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","The State Key Laboratory of Blockchain and Data Security , Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","institution_ids":["https://openalex.org/I4210156488"]},{"raw_affiliation_string":"The State Key Laboratory of Blockchain and Data Security , Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139730956","display_name":"Lidan Shou","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156488","display_name":"Institute of Zoology of the Slovak Academy of Sciences","ror":"https://ror.org/05tm6j853","country_code":"SK","type":"facility","lineage":["https://openalex.org/I207624831","https://openalex.org/I4210156488"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","SK"],"is_corresponding":false,"raw_author_name":"Lidan Shou","raw_affiliation_strings":["Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","The State Key Laboratory of Blockchain and Data Security , Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","institution_ids":["https://openalex.org/I4210156488"]},{"raw_affiliation_string":"The State Key Laboratory of Blockchain and Data Security , Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5139756701","display_name":"Huan Li","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156488","display_name":"Institute of Zoology of the Slovak Academy of Sciences","ror":"https://ror.org/05tm6j853","country_code":"SK","type":"facility","lineage":["https://openalex.org/I207624831","https://openalex.org/I4210156488"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","SK"],"is_corresponding":false,"raw_author_name":"Huan Li","raw_affiliation_strings":["Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","The State Key Laboratory of Blockchain and Data Security , Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hangzhou High-Tech Zone (Binjiang ) Institute of Blockchain and Data Security","institution_ids":["https://openalex.org/I4210156488"]},{"raw_affiliation_string":"The State Key Laboratory of Blockchain and Data Security , Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.80700807,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"13018","last_page":"13034"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.3564999997615814,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.3564999997615814,"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":0.34290000796318054,"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":0.0697999969124794,"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/compression","display_name":"Compression (physics)","score":0.4788999855518341},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.44909998774528503},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.42590001225471497},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.4034999907016754},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.37450000643730164}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7720000147819519},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.4788999855518341},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.44909998774528503},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.42590001225471497},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.4034999907016754},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3865000009536743},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.37450000643730164},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.35019999742507935},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2824000120162964},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2694999873638153},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2685000002384186},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.25850000977516174},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2574999928474426},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.25609999895095825}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.594","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.594","pdf_url":"https://aclanthology.org/2026.acl-long.594.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.594","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.594","pdf_url":"https://aclanthology.org/2026.acl-long.594.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6320710235","display_name":null,"funder_award_id":"62471168","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320318547","display_name":"Baidu","ror":"https://ror.org/03vs3wt56"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320338464","display_name":"Natural Science Foundation of Zhejiang Province","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166845075.pdf","grobid_xml":"https://content.openalex.org/works/W7166845075.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multimodal":[0],"Large":[1],"Language":[2],"Models":[3],"(MLLMs)":[4],"have":[5],"advanced":[6],"unified":[7],"reasoning":[8],"over":[9],"text,":[10],"images,":[11],"and":[12,42,56,87,98,156,178],"videos,":[13],"but":[14],"their":[15],"inference":[16],"is":[17,64],"hindered":[18],"by":[19,153,159,174],"the":[20,100,193],"rapid":[21],"growth":[22],"of":[23,32,103],"key-value":[24],"(KV)":[25],"caches.Each":[26],"visual":[27],"input":[28],"expands":[29],"into":[30,130],"thousands":[31],"tokens,":[33,81],"causing":[34],"caches":[35,67],"to":[36,52,65,176,192],"scale":[37],"linearly":[38],"with":[39,166,183],"context":[40],"length":[41],"remain":[43],"resident":[44],"in":[45,123],"GPU":[46],"memory":[47,54,173],"throughout":[48],"decoding,":[49,182],"which":[50],"leads":[51],"prohibitive":[53],"overhead":[55],"latency":[57],"even":[58,189],"on":[59,162],"high-end":[60],"GPUs.A":[61],"common":[62],"solution":[63],"compress":[66],"under":[68],"a":[69,113,139],"fixed":[70],"allocated":[71],"budget":[72,141],"at":[73,96],"different":[74],"granularities:":[75],"tokenlevel":[76],"uniformly":[77],"discards":[78],"less":[79],"important":[80],"layer-level":[82],"varies":[83],"retention":[84],"across":[85,91],"layers,":[86],"head-level":[88],"redistributes":[89],"budgets":[90],"heads.Yet":[92],"these":[93],"approaches":[94],"stop":[95],"allocation":[97,142],"overlook":[99],"heterogeneous":[101],"behaviors":[102],"attention":[104],"heads":[105,126,150,158],"that":[106,119,168],"require":[107],"distinct":[108],"compression":[109,117],"strategies.We":[110],"propose":[111],"HYBRIDKV,":[112],"hybrid":[114],"KV":[115,146,171],"cache":[116,172],"framework":[118],"integrates":[120],"complementary":[121],"strategies":[122],"three":[124],"stages:":[125],"are":[127,151],"first":[128],"classified":[129],"static":[131,149],"or":[132,188],"dynamic":[133,157],"types":[134],"using":[135],"text-centric":[136],"attention;":[137],"then":[138],"top-down":[140],"scheme":[143],"hierarchically":[144],"assigns":[145],"budgets;":[147],"finally,":[148],"compressed":[152],"text-prior":[154],"pruning":[155],"chunk-wise":[160],"retrieval.Experiments":[161],"11":[163],"multimodal":[164],"benchmarks":[165],"Qwen2.5-VL-7Bshow":[167],"HYBRIDKV":[169],"reduces":[170],"up":[175],"7.9\u00d7":[177],"achieves":[179],"1.52\u00d7":[180],"faster":[181],"almost":[184],"no":[185],"performance":[186],"drop":[187],"higher":[190],"relative":[191],"full-cache":[194],"MLLM.":[195]},"counts_by_year":[],"updated_date":"2026-08-24T07:32:12.397491","created_date":"2026-07-02T00:00:00"}
