{"id":"https://openalex.org/W7160994188","doi":"https://doi.org/10.48550/arxiv.2605.11478","title":"FibQuant: Universal Vector Quantization for Random-Access KV-Cache Compression","display_name":"FibQuant: Universal Vector Quantization for Random-Access KV-Cache Compression","publication_year":2026,"publication_date":"2026-05-12","ids":{"openalex":"https://openalex.org/W7160994188","doi":"https://doi.org/10.48550/arxiv.2605.11478"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.11478","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.11478","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.11478","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5014076559","display_name":"Namyoon Lee","orcid":"https://orcid.org/0000-0003-4321-4108"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Namyoon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5025186204","display_name":"Yongjune Kim","orcid":"https://orcid.org/0000-0003-0120-3750"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Yongjune","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/T11478","display_name":"Caching and Content Delivery","score":0.30140000581741333,"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"}},"topics":[{"id":"https://openalex.org/T11478","display_name":"Caching and Content Delivery","score":0.30140000581741333,"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/T11269","display_name":"Algorithms and Data Compression","score":0.17890000343322754,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.10419999808073044,"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/scalar","display_name":"Scalar (mathematics)","score":0.4778999984264374},{"id":"https://openalex.org/keywords/vector-quantization","display_name":"Vector quantization","score":0.47690001130104065},{"id":"https://openalex.org/keywords/codebook","display_name":"Codebook","score":0.4521999955177307},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.4275999963283539},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.37689998745918274},{"id":"https://openalex.org/keywords/encoder","display_name":"Encoder","score":0.3617999851703644},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.3587999939918518},{"id":"https://openalex.org/keywords/cosine-similarity","display_name":"Cosine similarity","score":0.350600004196167}],"concepts":[{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5292999744415283},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.5051000118255615},{"id":"https://openalex.org/C57691317","wikidata":"https://www.wikidata.org/wiki/Q1289248","display_name":"Scalar (mathematics)","level":2,"score":0.4778999984264374},{"id":"https://openalex.org/C199833920","wikidata":"https://www.wikidata.org/wiki/Q612536","display_name":"Vector quantization","level":2,"score":0.47690001130104065},{"id":"https://openalex.org/C127759330","wikidata":"https://www.wikidata.org/wiki/Q637416","display_name":"Codebook","level":2,"score":0.4521999955177307},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.4275999963283539},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4041000008583069},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.37689998745918274},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.3617999851703644},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.3587999939918518},{"id":"https://openalex.org/C2780762811","wikidata":"https://www.wikidata.org/wiki/Q1784941","display_name":"Cosine similarity","level":3,"score":0.350600004196167},{"id":"https://openalex.org/C171182647","wikidata":"https://www.wikidata.org/wiki/Q126736","display_name":"Scalar multiplication","level":3,"score":0.3181999921798706},{"id":"https://openalex.org/C161765866","wikidata":"https://www.wikidata.org/wiki/Q184748","display_name":"Codec","level":2,"score":0.31700000166893005},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.296999990940094},{"id":"https://openalex.org/C184720557","wikidata":"https://www.wikidata.org/wiki/Q7825049","display_name":"Topology (electrical circuits)","level":2,"score":0.2906000018119812},{"id":"https://openalex.org/C32900221","wikidata":"https://www.wikidata.org/wiki/Q181365","display_name":"Dot product","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.2806999981403351},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.27250000834465027},{"id":"https://openalex.org/C46900642","wikidata":"https://www.wikidata.org/wiki/Q2647","display_name":"Huffman coding","level":3,"score":0.2678999900817871},{"id":"https://openalex.org/C136886441","wikidata":"https://www.wikidata.org/wiki/Q926129","display_name":"Normalization (sociology)","level":2,"score":0.26190000772476196},{"id":"https://openalex.org/C93372532","wikidata":"https://www.wikidata.org/wiki/Q6552455","display_name":"Linde\u2013Buzo\u2013Gray algorithm","level":3,"score":0.26100000739097595},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.25859999656677246},{"id":"https://openalex.org/C37312099","wikidata":"https://www.wikidata.org/wiki/Q203484","display_name":"Vector Laplacian","level":4,"score":0.25589999556541443},{"id":"https://openalex.org/C39927690","wikidata":"https://www.wikidata.org/wiki/Q11197","display_name":"Logarithm","level":2,"score":0.25270000100135803},{"id":"https://openalex.org/C66746571","wikidata":"https://www.wikidata.org/wiki/Q1134833","display_name":"ENCODE","level":3,"score":0.25049999356269836}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.11478","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.11478","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.11478","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.11478","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Long-context":[0],"inference":[1],"is":[2,9,26,80,86,210],"increasingly":[3],"a":[4,41,44,53,71,74,82,87,97,113,152,158,162,169,191],"memory-traffic":[5],"problem.":[6],"The":[7,122,165],"culprit":[8],"the":[10,63,67,91,104,137],"key--value":[11],"(KV)":[12],"cache:":[13],"it":[14,25,85,209],"grows":[15],"with":[16,151],"context":[17],"length,":[18],"batch":[19],"size,":[20],"layers,":[21],"and":[22,24,48,58,130,161,175,219],"heads,":[23],"read":[27],"at":[28,52,148,197,204,216,227],"every":[29],"decoding":[30],"step.":[31,69],"Rotation-based":[32],"scalar":[33,110,145,225,234],"codecs":[34],"meet":[35],"this":[36,119],"systems":[37],"constraint":[38],"by":[39,66,112],"storing":[40],"norm,":[42],"applying":[43],"shared":[45,114],"random":[46],"rotation,":[47,73],"quantizing":[49],"one":[50],"coordinate":[51],"time.":[54],"They":[55],"are":[56],"universal":[57,98],"random-access,":[59],"but":[60],"they":[61],"discard":[62],"geometry":[64],"created":[65],"normalization":[68],"After":[70],"Haar":[72],"block":[75],"of":[76,214],"$k$":[77],"consecutive":[78],"coordinates":[79],"not":[81],"product":[83,146],"source;":[84],"spherical-Beta":[88],"source":[89],"on":[90,143,207],"unit":[92],"ball.":[93],"We":[94,134],"introduce":[95],"\\textsc{FibQuant},":[96],"fixed-rate":[99],"vector":[100,139],"quantizer":[101],"that":[102,136,155],"keeps":[103],"same":[105,166],"normalize--rotate--store":[106],"interface":[107],"while":[108],"replacing":[109],"tables":[111],"radial--angular":[115],"codebook":[116,123],"matched":[117,149],"to":[118,202,238],"canonical":[120],"source.":[121],"combines":[124],"Beta-quantile":[125],"radii,":[126],"Fibonacci\\,/\\,Roberts--Kronecker":[127],"quasi-uniform":[128],"directions,":[129],"multi-restart":[131],"Lloyd--Max":[132],"refinement.":[133],"prove":[135],"resulting":[138],"code":[140],"strictly":[141],"improves":[142],"its":[144],"specialization":[147],"rate,":[150],"high-rate":[153],"gain":[154],"separates":[156],"into":[157],"cell-shaping":[159],"factor":[160],"density-matching":[163],"factor.":[164],"construction":[167],"gives":[168],"dense":[170],"rate":[171],"axis,":[172],"including":[173],"fractional-bit":[174],"sub-one-bit":[176],"operating":[177],"points,":[178],"without":[179],"calibration":[180],"or":[181],"variable-length":[182],"addresses.":[183],"On":[184],"GPT-2":[185],"small":[186],"KV":[187],"caches,":[188],"\\textsc{FibQuant}":[189],"traces":[190],"memory--fidelity":[192],"frontier":[193],"from":[194],"$5\\times$":[195],"compression":[196,218],"$0.99$":[198],"attention":[199],"cosine":[200],"similarity":[201],"$34\\times$":[203],"$0.95$.":[205],"End-to-end":[206],"TinyLlama-1.1B,":[208],"within":[211],"$0.10$":[212],"perplexity":[213,223],"fp16":[215],"$4\\times$":[217],"has":[220],"$3.6\\times$":[221],"lower":[222],"than":[224],"\\textsc{TurboQuant}":[226],"$b":[228],"=":[229],"2$":[230],"($8\\times$":[231],"compression),":[232],"where":[233],"random-access":[235],"quantization":[236],"begins":[237],"fail.":[239]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-14T00:00:00"}