{"id":"https://openalex.org/W7143348494","doi":"https://doi.org/10.48550/arxiv.2603.26357","title":"MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Model","display_name":"MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Model","publication_year":2026,"publication_date":"2026-03-27","ids":{"openalex":"https://openalex.org/W7143348494","doi":"https://doi.org/10.48550/arxiv.2603.26357"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.26357","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.26357","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,"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.2603.26357","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112879567","display_name":"Quan Dao","orcid":"https://orcid.org/0009-0006-0996-0472"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dao, Quan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5130969078","display_name":"Dimitris Metaxas","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Metaxas, Dimitris","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.8119999766349792,"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"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.8119999766349792,"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"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.016499999910593033,"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"}},{"id":"https://openalex.org/T11751","display_name":"Lattice Boltzmann Simulation Studies","score":0.014700000174343586,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/flops","display_name":"FLOPS","score":0.7861999869346619},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.730400025844574},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6948999762535095},{"id":"https://openalex.org/keywords/generative-design","display_name":"Generative Design","score":0.4690999984741211},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.4253000020980835},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.3508000075817108},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.3303000032901764}],"concepts":[{"id":"https://openalex.org/C3826847","wikidata":"https://www.wikidata.org/wiki/Q188768","display_name":"FLOPS","level":2,"score":0.7861999869346619},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.730400025844574},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6948999762535095},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.664900004863739},{"id":"https://openalex.org/C184408114","wikidata":"https://www.wikidata.org/wiki/Q1502022","display_name":"Generative Design","level":3,"score":0.4690999984741211},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.4253000020980835},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.36550000309944153},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.3508000075817108},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3312000036239624},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3303000032901764},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.32179999351501465},{"id":"https://openalex.org/C2985918086","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel architecture","level":3,"score":0.295199990272522},{"id":"https://openalex.org/C37135326","wikidata":"https://www.wikidata.org/wiki/Q931942","display_name":"Design flow","level":2,"score":0.28769999742507935},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.2847000062465668},{"id":"https://openalex.org/C187691185","wikidata":"https://www.wikidata.org/wiki/Q2020720","display_name":"Grid","level":2,"score":0.2840000092983246},{"id":"https://openalex.org/C133162039","wikidata":"https://www.wikidata.org/wiki/Q1061077","display_name":"Code generation","level":3,"score":0.28110000491142273},{"id":"https://openalex.org/C2984241579","wikidata":"https://www.wikidata.org/wiki/Q323611","display_name":"Architectural design","level":3,"score":0.26589998602867126},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.2612999975681305},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.2574000060558319},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.25380000472068787}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.26357","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.26357","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,"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.2603.26357","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.26357","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,"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":{"Transformer":[0],"architectures,":[1],"particularly":[2],"Diffusion":[3],"Transformers":[4],"(DiTs),":[5],"have":[6],"become":[7],"widely":[8],"used":[9],"in":[10,37,57,91],"diffusion":[11],"and":[12,107],"flow-matching":[13],"models":[14],"due":[15],"to":[16,21,41,65,76,89],"their":[17],"strong":[18],"performance":[19],"compared":[20],"convolutional":[22],"UNets.":[23],"However,":[24],"the":[25,31,117,121],"isotropic":[26],"design":[27,56,82],"of":[28,34,123],"DiTs":[29],"processes":[30],"same":[32],"number":[33],"patchified":[35],"tokens":[36],"every":[38],"block,":[39],"leading":[40],"relatively":[42],"heavy":[43],"computation":[44],"during":[45],"training":[46,112],"process.":[47],"In":[48,98],"this":[49],"work,":[50],"we":[51,100],"introduce":[52],"a":[53],"multi-patch":[54],"transformer":[55],"which":[58],"early":[59],"blocks":[60,72],"operate":[61],"on":[62,116],"larger":[63],"patches":[64,75],"capture":[66],"coarse":[67],"global":[68],"context,":[69],"while":[70,93],"later":[71],"use":[73],"smaller":[74],"refine":[77],"local":[78],"details.":[79],"This":[80],"hierarchical":[81],"could":[83],"reduces":[84],"computational":[85],"cost":[86],"by":[87],"up":[88],"50%":[90],"GFLOPs":[92],"achieving":[94],"good":[95],"generative":[96],"performance.":[97],"addition,":[99],"also":[101],"propose":[102],"improved":[103],"designs":[104],"for":[105],"time":[106],"class":[108],"embeddings":[109],"that":[110],"accelerate":[111],"convergence.":[113],"Extensive":[114],"experiments":[115],"ImageNet":[118],"dataset":[119],"demonstrate":[120],"effectiveness":[122],"our":[124],"architectural":[125],"choices.":[126],"Code":[127],"is":[128],"released":[129],"at:":[130],"https://github.com/quandao10/MPDiT":[131]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-03-31T00:00:00"}