{"id":"https://openalex.org/W4415536856","doi":"https://doi.org/10.1145/3746027.3755479","title":"Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching","display_name":"Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching","publication_year":2025,"publication_date":"2025-10-25","ids":{"openalex":"https://openalex.org/W4415536856","doi":"https://doi.org/10.1145/3746027.3755479"},"language":null,"primary_location":{"id":"doi:10.1145/3746027.3755479","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3746027.3755479","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Zhixin Zheng","orcid":"https://orcid.org/0009-0003-0228-4000"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhixin Zheng","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China and University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":"https://orcid.org/0009-0003-0228-4000","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China and University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079329431","display_name":"Xinyu Wang","orcid":"https://orcid.org/0009-0003-5492-8015"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xinyu Wang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China and Shandong University, Qingdao, China"],"raw_orcid":"https://orcid.org/0009-0003-5492-8015","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China and Shandong University, Qingdao, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113844313","display_name":"Chang Zou","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chang Zou","raw_affiliation_strings":["University of Electronic Science and Technology of China, Chengdu, China and Shanghai Jiao Tong University, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0001-2663-2204","affiliations":[{"raw_affiliation_string":"University of Electronic Science and Technology of China, Chengdu, China and Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107930049","display_name":"Shaobo Wang","orcid":"https://orcid.org/0000-0001-8156-7081"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shaobo Wang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0001-8156-7081","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100689114","display_name":"Linfeng Zhang","orcid":"https://orcid.org/0000-0002-3341-183X"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Linfeng Zhang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0002-3341-183X","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I183067930"],"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":"10181","last_page":"10189"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11269","display_name":"Algorithms and Data Compression","score":0.9983999729156494,"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/T11269","display_name":"Algorithms and Data Compression","score":0.9983999729156494,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.9975000023841858,"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/T12029","display_name":"DNA and Biological Computing","score":0.9972000122070312,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.6787999868392944},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6035000085830688},{"id":"https://openalex.org/keywords/cluster-analysis","display_name":"Cluster analysis","score":0.5893999934196472},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.5537999868392944},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5372999906539917},{"id":"https://openalex.org/keywords/granularity","display_name":"Granularity","score":0.41839998960494995},{"id":"https://openalex.org/keywords/feature-model","display_name":"Feature model","score":0.3978999853134155},{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.39309999346733093}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.723800003528595},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6787999868392944},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6035000085830688},{"id":"https://openalex.org/C73555534","wikidata":"https://www.wikidata.org/wiki/Q622825","display_name":"Cluster analysis","level":2,"score":0.5893999934196472},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.5537999868392944},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5372999906539917},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.42250001430511475},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.41839998960494995},{"id":"https://openalex.org/C101814296","wikidata":"https://www.wikidata.org/wiki/Q5439685","display_name":"Feature model","level":3,"score":0.3978999853134155},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.39309999346733093},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3894999921321869},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3529999852180481},{"id":"https://openalex.org/C83665646","wikidata":"https://www.wikidata.org/wiki/Q42139305","display_name":"Feature vector","level":2,"score":0.33329999446868896},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.3199999928474426},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.3052999973297119},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.302700012922287},{"id":"https://openalex.org/C2781142347","wikidata":"https://www.wikidata.org/wiki/Q1366592","display_name":"Hilbert curve","level":2,"score":0.2897999882698059},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.2687999904155731},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.2676999866962433},{"id":"https://openalex.org/C159694833","wikidata":"https://www.wikidata.org/wiki/Q2321565","display_name":"Iterative method","level":2,"score":0.26570001244544983},{"id":"https://openalex.org/C99138194","wikidata":"https://www.wikidata.org/wiki/Q183427","display_name":"Hash function","level":2,"score":0.260699987411499},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.25529998540878296}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3746027.3755479","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3746027.3755479","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":8,"referenced_works":["https://openalex.org/W2117539524","https://openalex.org/W4386065704","https://openalex.org/W4390872297","https://openalex.org/W4390874074","https://openalex.org/W4402660110","https://openalex.org/W4402778510","https://openalex.org/W4402816687","https://openalex.org/W4413157724"],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"transformers":[1,41],"have":[2],"gained":[3],"significant":[4],"attention":[5],"in":[6,47,53,69,99,106,140],"recent":[7],"years":[8],"for":[9,88,158],"their":[10,28,111],"ability":[11],"to":[12,27,38,113,121,152],"generate":[13],"high-quality":[14],"images":[15],"and":[16,50,85,109,135,143],"videos,":[17],"yet":[18],"still":[19],"suffer":[20],"from":[21],"a":[22],"huge":[23],"computational":[24],"cost":[25],"due":[26],"iterative":[29],"denoising":[30],"process.":[31],"Recently,":[32],"feature":[33,45,90],"caching":[34,43],"has":[35],"been":[36],"introduced":[37],"accelerate":[39],"diffusion":[40,63,154],"by":[42,127,177],"the":[44,54,59,67,70,115,123,174],"computation":[46],"previous":[48,89],"timesteps":[49],"reusing":[51],"it":[52,147],"following":[55],"timesteps,":[56],"which":[57,118],"leverage":[58],"temporal":[60],"similarity":[61,68],"of":[62,125,171],"models":[64],"while":[65],"ignoring":[66],"spatial":[71,95],"dimension.":[72],"In":[73],"this":[74],"paper,":[75],"we":[76],"introduce":[77],"Cluster-Driven":[78],"Feature":[79],"Caching":[80],"(ClusCa)":[81],"as":[82],"an":[83,169],"orthogonal":[84],"complementary":[86],"perspective":[87],"caching.":[91],"Specifically,":[92],"ClusCa":[93,162],"performs":[94],"clustering":[96],"on":[97,132,166],"tokens":[98,126],"each":[100,107],"timestep,":[101],"computes":[102],"only":[103],"one":[104],"token":[105],"cluster":[108],"propagates":[110],"information":[112],"all":[114],"other":[116],"tokens,":[117],"is":[119,181],"able":[120],"reduce":[122],"number":[124],"over":[128],"90%.":[129],"Extensive":[130],"experiments":[131],"DiT,":[133],"FLUX":[134,167],"HunyuanVideo":[136],"demonstrate":[137],"its":[138],"effectiveness":[139],"both":[141],"text-to-image":[142],"text-to-video":[144],"generation.":[145],"Besides,":[146],"can":[148],"be":[149],"directly":[150],"applied":[151],"any":[153],"transformer":[155],"without":[156],"requirements":[157],"training.":[159],"For":[160],"instance,":[161],"achieves":[163],"4.96x":[164],"acceleration":[165],"with":[168],"ImageReward":[170],"99.49%,":[172],"surpassing":[173],"original":[175],"model":[176],"0.51%.":[178],"The":[179],"code":[180],"available":[182],"at":[183],"https://github.com/zhixin-zheng/ClusCa.":[184]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-25T00:00:00"}