{"id":"https://openalex.org/W7166810043","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1541","title":"ReFreeKV: Towards Threshold-Free KV Cache Compression","display_name":"ReFreeKV: Towards Threshold-Free KV Cache Compression","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166810043","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1541"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1541","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1541","pdf_url":"https://aclanthology.org/2026.findings-acl.1541.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":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.1541.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5035601720","display_name":"Xuanfan Ni","orcid":"https://orcid.org/0009-0006-9785-2956"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xuanfan Ni","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066949435","display_name":"Liyan Xu","orcid":"https://orcid.org/0000-0002-6933-3394"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liyan Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139748030","display_name":"Chenyang Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenyang Lyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100364552","display_name":"Lu Wang","orcid":"https://orcid.org/0000-0003-4979-4895"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Longyue Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139728673","display_name":"Mo Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mo Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139758728","display_name":"Lemao Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lemao Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139717797","display_name":"Fandong Meng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fandong Meng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139746769","display_name":"Jie Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jie Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139779590","display_name":"Piji Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Piji Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"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":0.80812641,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"30827","last_page":"30841"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.4133000075817108,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.4133000075817108,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T11181","display_name":"Advanced Data Storage Technologies","score":0.11789999902248383,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11478","display_name":"Caching and Content Delivery","score":0.10939999669790268,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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":0.3693999946117401},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.3089999854564667},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.27160000801086426},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.25529998540878296}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5200999975204468},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.3693999946117401},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.30559998750686646},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.28850001096725464},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.28850001096725464},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2847999930381775},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.27160000801086426},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.25529998540878296},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.23849999904632568}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1541","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1541","pdf_url":"https://aclanthology.org/2026.findings-acl.1541.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":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1541","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1541","pdf_url":"https://aclanthology.org/2026.findings-acl.1541.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":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166810043.pdf","grobid_xml":"https://content.openalex.org/works/W7166810043.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"To":[0],"reduce":[1],"memory":[2,29],"consumption":[3],"during":[4],"LLM":[5],"inference,":[6],"prior":[7],"works":[8],"have":[9],"proposed":[10],"numerous":[11],"methods":[12,119],"that":[13,93,107,120],"focus":[14],"on":[15,20,31,37,97],"KV":[16,114],"cache":[17],"pruning":[18],"based":[19],"various":[21],"designed":[22],"criteria.While":[23],"these":[24],"techniques":[25],"often":[26,35],"accomplish":[27],"lossless":[28],"reduction":[30],"many":[32],"datasets,":[33],"they":[34],"hinge":[36],"an":[38,41],"under-emphasized":[39],"condition:":[40],"input/domain-specific":[42],"budget":[43,123],"threshold":[44,78,87,110],"needs":[45],"to":[46,49],"be":[47,58,89],"pre-determined":[48],"achieve":[50],"the":[51,82,109,137],"optimal":[52],"performance.However,":[53],"such":[54,85],"input-sensitive":[55,86],"design":[56],"may":[57],"considerably":[59],"limited":[60],"in":[61],"real-world":[62],"scenarios,":[63],"as":[64,136],"open-domain":[65],"inputs":[66],"span":[67],"diverse":[68,148],"domains,":[69],"lengths":[70],"and":[71,153,159],"difficulty":[72],"levels,":[73],"without":[74],"clear":[75],"boundaries":[76],"for":[77,112,117],"selection.As":[79],"a":[80,90,104,131],"result,":[81],"dependence":[83],"of":[84,140],"can":[88],"fundamental":[91],"limitation":[92],"causes":[94],"large":[95],"degradation":[96],"arbitrary":[98],"inputs.In":[99],"this":[100,141],"work,":[101],"we":[102],"propose":[103,130],"new":[105],"objective":[106],"lifts":[108],"constraints":[111],"robust":[113],"compression,":[115],"advocating":[116],"\"threshold-free\"":[118],"adaptively":[121],"adjust":[122],"allocation":[124],"while":[125],"preserving":[126],"full-cache":[127],"performance.We":[128],"then":[129],"novel":[132],"method,":[133],"ReFreeKV,":[134],"serving":[135],"first":[138],"instantiation":[139],"objective.Extensive":[142],"experiments":[143],"across":[144],"13":[145],"datasets":[146],"with":[147],"context":[149],"lengths,":[150],"task":[151],"types,":[152],"model":[154],"sizes":[155],"demonstrate":[156],"its":[157],"efficacy":[158],"efficiency.":[160]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}