{"id":"https://openalex.org/W4415329327","doi":"https://doi.org/10.52202/085713-4984","title":"dKV-Cache: The Cache for Diffusion Language Models","display_name":"dKV-Cache: The Cache for Diffusion Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4415329327","doi":"https://doi.org/10.52202/085713-4984"},"language":"en","primary_location":{"id":"doi:10.52202/085713-4984","doi":"https://doi.org/10.52202/085713-4984","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-4984","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2505.15781","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5109748042","display_name":"Xinyin Ma","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Xinyin Ma","raw_affiliation_strings":["National University of Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National University of Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109242003","display_name":"Ruotong Yu","orcid":"https://orcid.org/0000-0003-1983-2694","observed_orcids":["https://orcid.org/0000-0003-1983-2694"]},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Runpeng Yu","raw_affiliation_strings":["National University of Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National University of Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102483331","display_name":"Gongfan Fang","orcid":"https://orcid.org/0009-0009-6935-0432","observed_orcids":["https://orcid.org/0009-0009-6935-0432"]},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Gongfan Fang","raw_affiliation_strings":["National University of Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National University of Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5015574447","display_name":"Xinchao Wang","orcid":"https://orcid.org/0000-0003-0057-1404","observed_orcids":["https://orcid.org/0000-0003-0057-1404","https://orcid.org/0009-0005-8055-9094"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xinchao Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":48.0942,"has_fulltext":true,"cited_by_count":136,"citation_normalized_percentile":{"value":0.99876424,"is_in_top_1_percent":true,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":99,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"165470","last_page":"165494"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.7167999744415283,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.7167999744415283,"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":0.24379999935626984,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.006500000134110451,"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/diffusion-language-models","display_name":"diffusion language models","score":0.9980000257492065},{"id":"https://openalex.org/keywords/key-value-cache","display_name":"key-value cache","score":0.9020000100135803},{"id":"https://openalex.org/keywords/inference-acceleration","display_name":"inference acceleration","score":0.8220000267028809},{"id":"https://openalex.org/keywords/noise-reduction","display_name":"noise reduction","score":0.609000027179718},{"id":"https://openalex.org/keywords/non-autoregressive-decoding","display_name":"non-autoregressive decoding","score":0.7829999923706055},{"id":"https://openalex.org/keywords/bidirectional-attention","display_name":"bidirectional attention","score":0.8569999933242798},{"id":"https://openalex.org/keywords/greedy-decoding","display_name":"greedy decoding","score":0.574999988079071},{"id":"https://openalex.org/keywords/long-sequence-generation","display_name":"long sequence generation","score":0.6880000233650208},{"id":"https://openalex.org/keywords/code-generation","display_name":"code generation","score":0.6909999847412109}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8123999834060669},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.7706000208854675},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.6154999732971191},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.49889999628067017},{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.4970000088214874},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.45339998602867126},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.44589999318122864},{"id":"https://openalex.org/C38556500","wikidata":"https://www.wikidata.org/wiki/Q13404475","display_name":"Cache algorithms","level":4,"score":0.4440999925136566},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.41609999537467957},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.41269999742507935},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.40070000290870667},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3853999972343445},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.382099986076355},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.37549999356269836},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3677000105381012},{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.36570000648498535},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.328900009393692},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.3276999890804291},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.3206999897956848},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.313400000333786},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.301800012588501},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.26499998569488525},{"id":"https://openalex.org/C167713795","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"Smart Cache","level":5,"score":0.26109999418258667},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.2517000138759613}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.52202/085713-4984","doi":"https://doi.org/10.52202/085713-4984","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-4984","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2505.15781","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.15781","pdf_url":"https://arxiv.org/pdf/2505.15781","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2505.15781","doi":"https://doi.org/10.48550/arxiv.2505.15781","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.15781","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:2505.15781","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.15781","pdf_url":"https://arxiv.org/pdf/2505.15781","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":"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":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4415329327.pdf","grobid_xml":"https://content.openalex.org/works/W4415329327.grobid-xml"},"referenced_works_count":22,"referenced_works":["https://openalex.org/W2767206889","https://openalex.org/W3195140297","https://openalex.org/W3210904070","https://openalex.org/W4321471763","https://openalex.org/W4379089709","https://openalex.org/W4385569744","https://openalex.org/W4385571393","https://openalex.org/W4387964173","https://openalex.org/W4389518760","https://openalex.org/W4393572898","https://openalex.org/W4393969060","https://openalex.org/W4399795196","https://openalex.org/W4402778510","https://openalex.org/W4402816687","https://openalex.org/W4404307915","https://openalex.org/W4404311645","https://openalex.org/W4405096483","https://openalex.org/W4405664066","https://openalex.org/W4407359553","https://openalex.org/W4407679583","https://openalex.org/W4415798933","https://openalex.org/W6910504028"],"related_works":["https://openalex.org/W4415332076","https://openalex.org/W4415037641","https://openalex.org/W7092289163","https://openalex.org/W6904055037","https://openalex.org/W7164158727","https://openalex.org/W7214068343","https://openalex.org/W7139935094","https://openalex.org/W7204596471","https://openalex.org/W4416048624","https://openalex.org/W7166199774"],"abstract_inverted_index":{"Diffusion":[0],"Language":[1],"Models":[2],"(DLMs)":[3],"have":[4,19,71],"been":[5,21],"seen":[6],"as":[7],"a":[8,50,82,197],"promising":[9],"competitor":[10],"for":[11,55,88],"autoregressive":[12],"language":[13,17,179],"models.":[14],"However,":[15],"diffusion":[16,77],"models":[18],"long":[20,116],"constrained":[22],"by":[23,48,65],"slow":[24],"inference.":[25,127],"A":[26],"core":[27],"challenge":[28],"is":[29,63],"that":[30,41,68,119,187],"their":[31],"non-autoregressive":[32],"architecture":[33],"and":[34,84,90,101,111,166,182],"bidirectional":[35],"attention":[36],"preclude":[37],"the":[38,56,66,76,145,162],"key-value":[39],"cache":[40,99,188],"accelerates":[42],"decoding.":[43],"We":[44,93,168],"address":[45],"this":[46],"bottleneck":[47],"proposing":[49],"KV-cache-like":[51],"mechanism,":[52],"delayed":[53,83],"KV-Cache,":[54],"denoising":[57],"process":[58],"of":[59,147],"DLMs.":[60,167,202],"Our":[61],"approach":[62],"motivated":[64],"observation":[67],"different":[69],"tokens":[70],"distinct":[72],"representation":[73],"dynamics":[74],"throughout":[75],"process.":[78],"Accordingly,":[79],"we":[80],"propose":[81],"conditioned":[85],"caching":[86,133],"strategy":[87],"key":[89,100],"value":[91,102],"states.":[92],"design":[94],"two":[95],"complementary":[96],"variants":[97],"to":[98],"step-by-step:":[103],"(1)":[104],"dKV-Cache-Decode,":[105],"which":[106,130],"provides":[107],"almost":[108],"lossless":[109],"acceleration,":[110],"even":[112,195],"improves":[113],"performance":[114,149],"on":[115,172],"sequences,":[117],"suggesting":[118],"existing":[120],"DLMs":[121],"may":[122],"under-utilise":[123],"contextual":[124],"information":[125],"during":[126],"(2)":[128],"dKV-Cache-Greedy,":[129],"has":[131],"aggressive":[132],"with":[134,140],"reduced":[135],"lifespan,":[136],"achieving":[137],"higher":[138],"speed-ups":[139],"quadratic":[141],"time":[142],"complexity":[143],"at":[144],"cost":[146],"some":[148],"degradation.":[150],"dKV-Cache,":[151],"in":[152,158,193,196],"final,":[153],"achieves":[154],"from":[155,200],"2-10x":[156],"speedup":[157],"inference,":[159],"largely":[160],"narrowing":[161],"gap":[163],"between":[164],"ARs":[165],"evaluate":[169],"our":[170],"dKV-Cache":[171],"several":[173],"benchmarks,":[174],"delivering":[175],"acceleration":[176],"across":[177],"general":[178],"understanding,":[180],"mathematical,":[181],"code-generation":[183],"benchmarks.":[184],"Experiments":[185],"demonstrate":[186],"can":[189],"also":[190],"be":[191],"used":[192],"DLMs,":[194],"training-free":[198],"manner":[199],"current":[201]},"counts_by_year":[{"year":2026,"cited_by_count":86},{"year":2025,"cited_by_count":50}],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2025-10-19T00:00:00"}