{"id":"https://openalex.org/W7204959638","doi":"https://doi.org/10.48550/arxiv.2608.30963","title":"A Universal Context-Reuse Layer for Cross-Model KV Sharing","display_name":"A Universal Context-Reuse Layer for Cross-Model KV Sharing","publication_year":2026,"publication_date":"2026-08-31","ids":{"openalex":"https://openalex.org/W7204959638","doi":"https://doi.org/10.48550/arxiv.2608.30963"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.30963","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30963","pdf_url":"https://arxiv.org/pdf/2608.30963","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.30963","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5122275445","display_name":"Yi Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121536196","display_name":"Dongming Jiang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Dongming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138780759","display_name":"Yi Zhao","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Yi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5144342972","display_name":"Bingzhe Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Bingzhe","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":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.5576000213623047,"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":0.5576000213623047,"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":0.361299991607666,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.024700000882148743,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/kv-cache-reuse","display_name":"KV cache reuse","score":0.7979999780654907},{"id":"https://openalex.org/keywords/large-language-model-serving","display_name":"large language model serving","score":0.8090000152587891},{"id":"https://openalex.org/keywords/inference-latency-reduction","display_name":"inference latency reduction","score":0.6129999756813049}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7705000042915344},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.7121999859809875},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.5803999900817871},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.57669997215271},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5590999722480774},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5081999897956848},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.5016000270843506},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.49970000982284546},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.4620000123977661},{"id":"https://openalex.org/C100279451","wikidata":"https://www.wikidata.org/wiki/Q372193","display_name":"Perplexity","level":3,"score":0.4244999885559082},{"id":"https://openalex.org/C111852164","wikidata":"https://www.wikidata.org/wiki/Q1414679","display_name":"Handover","level":2,"score":0.40639999508857727},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.3953999876976013},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.3919999897480011},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.36480000615119934},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3630000054836273},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.3610999882221222},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.3483000099658966},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.337799996137619},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.30250000953674316},{"id":"https://openalex.org/C112313634","wikidata":"https://www.wikidata.org/wiki/Q7886648","display_name":"Complement (music)","level":5,"score":0.26910001039505005},{"id":"https://openalex.org/C186967261","wikidata":"https://www.wikidata.org/wiki/Q5082128","display_name":"Mobile device","level":2,"score":0.262800008058548},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.26100000739097595},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.25609999895095825},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.250900000333786}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.30963","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30963","pdf_url":"https://arxiv.org/pdf/2608.30963","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.30963","doi":"https://doi.org/10.48550/arxiv.2608.30963","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.30963","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.30963","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30963","pdf_url":"https://arxiv.org/pdf/2608.30963","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7172487991","https://openalex.org/W7221171401","https://openalex.org/W7214719603","https://openalex.org/W4417092586","https://openalex.org/W7161727518","https://openalex.org/W7221804650","https://openalex.org/W4414836780","https://openalex.org/W7206178193","https://openalex.org/W4411119414","https://openalex.org/W7214804908"],"abstract_inverted_index":{"Modern":[0],"large":[1],"language":[2],"model":[3,16,26,72,96],"(LLM)":[4],"serving":[5],"systems":[6,224],"increasingly":[7],"operate":[8],"over":[9,128],"repeated":[10],"or":[11],"shared":[12],"context,":[13],"yet":[14],"each":[15],"typically":[17],"performs":[18],"its":[19],"own":[20],"prefill":[21,153,229],"computation":[22,40],"even":[23],"when":[24],"another":[25],"has":[27],"already":[28],"processed":[29],"the":[30,49,65,100,129,143],"same":[31],"input.":[32],"Existing":[33],"KV-cache":[34],"reuse":[35],"mechanisms":[36],"substantially":[37],"reduce":[38],"redundant":[39,228],"within":[41],"a":[42,54,70,74,81,122,172,223],"single":[43],"model,":[44,84],"but":[45],"generally":[46],"assume":[47],"that":[48,76,87,204],"producer":[50],"and":[51,95,105,218,233],"consumer":[52],"of":[53,124],"cache":[55],"are":[56],"identical.":[57],"We":[58,98],"study":[59],"\\emph{cross-model":[60],"KV":[61,66,113,149,205],"sharing},":[62],"which":[63],"translates":[64],"state":[67],"produced":[68],"by":[69,80,155],"source":[71],"into":[73],"representation":[75],"can":[77,207],"be":[78],"consumed":[79],"different":[82],"target":[83,140],"including":[85],"models":[86],"differ":[88],"in":[89,102],"scale,":[90],"architecture,":[91],"attention":[92],"configuration,":[93],"tokenizer,":[94],"family.":[97],"evaluate":[99],"approach":[101],"both":[103],"within-family":[104],"cross-family":[106,144,179],"settings.":[107],"For":[108,142],"Qwen2.5-7B":[109,177,189],"$\\rightarrow$":[110,146,176],"Qwen2.5-1.5B,":[111],"translated":[112],"states":[114,206],"improve":[115],"LongBench2":[116],"accuracy":[117,183],"from":[118,195],"27.59\\%":[119],"to":[120,138,157,168,197],"34.48\\%,":[121],"gain":[123],"6.89":[125],"percentage":[126],"points":[127],"native":[130,139,188],"1.5B":[131],"baseline,":[132],"while":[133,163,191],"reducing":[134,192,227],"handoff":[135,150,180],"cost":[136,154],"relative":[137],"prefill.":[141],"Qwen2.5-1.5B":[145],"Gemma-2-2B":[147],"setting,":[148,178],"reduces":[151],"target-side":[152],"up":[156],"67.05\\%":[158],"at":[159],"4K":[160],"context":[161],"length":[162],"maintaining":[164],"decoding":[165],"perplexity":[166],"close":[167],"native-model":[169],"baselines.":[170],"In":[171],"more":[173],"heterogeneous":[174,231],"Llama3.1-70B":[175],"achieves":[181],"44.0\\%":[182],"compared":[184],"with":[185],"45.7\\%":[186],"for":[187,226],"inference,":[190],"measured":[193],"latency":[194],"899ms":[196],"138ms.":[198],"These":[199],"results":[200],"provide":[201],"initial":[202],"evidence":[203],"serve":[208],"as":[209,222],"transferable":[210],"computational":[211],"representations":[212],"rather":[213],"than":[214],"strictly":[215],"model-local":[216],"caches,":[217],"motivate":[219],"\\emph{context":[220],"mobility}":[221],"abstraction":[225],"across":[230],"LLM":[232],"multi-agent":[234],"inference":[235],"workflows.":[236]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-02T00:00:00"}