{"id":"https://openalex.org/W7051896053","doi":"https://doi.org/10.48550/arxiv.2310.00426","title":"PixArt-$\u03b1$: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis","display_name":"PixArt-$\u03b1$: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis","publication_year":2023,"publication_date":"2023-09-30","ids":{"openalex":"https://openalex.org/W7051896053","doi":"https://doi.org/10.48550/arxiv.2310.00426"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2310.00426","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.00426","pdf_url":"https://arxiv.org/pdf/2310.00426","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-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","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/2310.00426","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Chen, Junsong","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Junsong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Yu, Jincheng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Jincheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030551813","display_name":"Chongjian Ge","orcid":"https://orcid.org/0000-0003-1142-9171","observed_orcids":["https://orcid.org/0000-0003-1142-9171","https://orcid.org/0000-0003-2746-2216"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ge, Chongjian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027453000","display_name":"Lewei Yao","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yao, Lewei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Xie, Enze","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Enze","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wu, Yue","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Yue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wang, Zhongdao","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zhongdao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070273088","display_name":"James Tin-Yau Kwok","orcid":"https://orcid.org/0000-0002-4828-8248","observed_orcids":["https://orcid.org/0000-0002-4828-8248"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kwok, James","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Luo, Ping","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Ping","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lu, Huchuan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Huchuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Li, Zhenguo","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zhenguo","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":30,"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":true,"primary_topic":{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":1.0,"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":1.0,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":9.999999747378752e-05,"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":0.0,"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/text-to-image-generation","display_name":"text-to-image generation","score":0.9120000004768372},{"id":"https://openalex.org/keywords/diffusion-transformer","display_name":"diffusion transformer","score":0.925000011920929},{"id":"https://openalex.org/keywords/efficient-training","display_name":"efficient training","score":0.6589999794960022},{"id":"https://openalex.org/keywords/training-cost-reduction","display_name":"training cost reduction","score":0.7369999885559082},{"id":"https://openalex.org/keywords/co2-emission-reduction","display_name":"CO2 emission reduction","score":0.8029999732971191},{"id":"https://openalex.org/keywords/cross-attention","display_name":"cross-attention","score":0.7139999866485596},{"id":"https://openalex.org/keywords/image-text-alignment","display_name":"image-text alignment","score":0.9430000185966492},{"id":"https://openalex.org/keywords/high-resolution-image-synthesis","display_name":"high-resolution image generation","score":0.6470000147819519},{"id":"https://openalex.org/keywords/semantic-control","display_name":"semantic control","score":0.5},{"id":"https://openalex.org/keywords/generative-ai","display_name":"generative AI","score":0.6200000047683716}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6805999875068665},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.6280999779701233},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5986999869346619},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.48489999771118164},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.4494999945163727},{"id":"https://openalex.org/C160633673","wikidata":"https://www.wikidata.org/wiki/Q355198","display_name":"Pixel","level":2,"score":0.4156000018119812},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4097000062465668},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.39629998803138733},{"id":"https://openalex.org/C55020928","wikidata":"https://www.wikidata.org/wiki/Q3813865","display_name":"Image quality","level":3,"score":0.38370001316070557},{"id":"https://openalex.org/C2989087649","wikidata":"https://www.wikidata.org/wiki/Q176953","display_name":"Image synthesis","level":3,"score":0.36899998784065247},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3668999969959259},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.34700000286102295},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.3456000089645386},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.31949999928474426},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.31310001015663147},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.30309998989105225},{"id":"https://openalex.org/C2776674983","wikidata":"https://www.wikidata.org/wiki/Q545981","display_name":"Image editing","level":3,"score":0.29269999265670776},{"id":"https://openalex.org/C9417928","wikidata":"https://www.wikidata.org/wiki/Q1070689","display_name":"Image processing","level":3,"score":0.2815999984741211},{"id":"https://openalex.org/C206658404","wikidata":"https://www.wikidata.org/wiki/Q12725","display_name":"Electricity","level":2,"score":0.26919999718666077},{"id":"https://openalex.org/C121684516","wikidata":"https://www.wikidata.org/wiki/Q7600677","display_name":"Computer graphics (images)","level":1,"score":0.26820001006126404},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.26330000162124634},{"id":"https://openalex.org/C199639397","wikidata":"https://www.wikidata.org/wiki/Q1788588","display_name":"Engineering drawing","level":1,"score":0.2565000057220459},{"id":"https://openalex.org/C166704113","wikidata":"https://www.wikidata.org/wiki/Q861092","display_name":"Image registration","level":3,"score":0.2556999921798706},{"id":"https://openalex.org/C205372480","wikidata":"https://www.wikidata.org/wiki/Q210521","display_name":"Image resolution","level":2,"score":0.2533000111579895}],"mesh":[],"locations_count":3,"locations":[{"id":"pmh:oai:arXiv.org:2310.00426","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.00426","pdf_url":"https://arxiv.org/pdf/2310.00426","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-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-136838","doi":null,"is_oa":false,"landing_page_url":"http://www.scopus.com/record/display.url?eid=2-s2.0-85200569896&origin=inward","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"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":"Conference paper"},{"id":"doi:10.48550/arxiv.2310.00426","doi":"https://doi.org/10.48550/arxiv.2310.00426","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2310.00426","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2310.00426","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.00426","pdf_url":"https://arxiv.org/pdf/2310.00426","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-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4390832578","https://openalex.org/W4379258791","https://openalex.org/W7103754349","https://openalex.org/W4404625217","https://openalex.org/W4413147203","https://openalex.org/W7130724513","https://openalex.org/W4308163867","https://openalex.org/W4393027713","https://openalex.org/W4281485151","https://openalex.org/W4403444038"],"abstract_inverted_index":{"The":[0],"most":[1],"advanced":[2],"text-to-image":[3],"(T2I)":[4],"models":[5,247],"require":[6],"significant":[7],"training":[8,69,95,162,179,207],"costs":[9],"(e.g.,":[10,47],"millions":[11],"of":[12,137,175],"GPU":[13,185],"hours),":[14],"seriously":[15],"hindering":[16],"the":[17,21,126,135,233],"fundamental":[18],"innovation":[19],"for":[20],"AIGC":[22,234],"community":[23,235],"while":[24],"increasing":[25],"CO2":[26,196],"emissions.":[27,197],"This":[28],"paper":[29],"introduces":[30],"PIXART-$\u03b1$,":[31],"a":[32,145,159,201],"Transformer-based":[33],"T2I":[34,110,168],"diffusion":[35],"model":[36,148],"whose":[37],"image":[38,45,61,105,219],"generation":[39],"quality":[40],"is":[41,209],"competitive":[42],"with":[43,67,200],"state-of-the-art":[44],"generators":[46],"Imagen,":[48],"SDXL,":[49],"and":[50,76,104,124,143,193,222,236],"even":[51],"Midjourney),":[52],"reaching":[53],"near-commercial":[54],"application":[55],"standards.":[56],"Additionally,":[57],"it":[58],"supports":[59],"high-resolution":[60],"synthesis":[62],"up":[63],"to":[64,120,149,153,232,238],"1024px":[65],"resolution":[66],"low":[68],"cost,":[70],"as":[71],"shown":[72],"in":[73,140,218],"Figure":[74],"1":[75],"2.":[77],"To":[78],"achieve":[79],"this":[80],"goal,":[81],"three":[82,93],"core":[83],"designs":[84],"are":[85],"proposed:":[86],"(1)":[87],"Training":[88],"strategy":[89],"decomposition:":[90],"We":[91,112,133,225],"devise":[92],"distinct":[94],"steps":[96],"that":[97,215],"separately":[98],"optimize":[99],"pixel":[100],"dependency,":[101],"text-image":[102,141,155],"alignment,":[103],"aesthetic":[106],"quality;":[107],"(2)":[108],"Efficient":[109],"Transformer:":[111],"incorporate":[113],"cross-attention":[114],"modules":[115],"into":[116],"Diffusion":[117,177],"Transformer":[118],"(DiT)":[119],"inject":[121],"text":[122],"conditions":[123],"streamline":[125],"computation-intensive":[127],"class-condition":[128],"branch;":[129],"(3)":[130],"High-informative":[131],"data:":[132],"emphasize":[134],"significance":[136],"concept":[138],"density":[139],"pairs":[142],"leverage":[144],"large":[146],"Vision-Language":[147],"auto-label":[150],"dense":[151],"pseudo-captions":[152],"assist":[154],"alignment":[156],"learning.":[157],"As":[158],"result,":[160],"PIXART-$\u03b1$'s":[161],"speed":[163],"markedly":[164],"surpasses":[165],"existing":[166],"large-scale":[167],"models,":[169],"e.g.,":[170],"PIXART-$\u03b1$":[171,216,227],"only":[172],"takes":[173],"10.8%":[174],"Stable":[176],"v1.5's":[178],"time":[180],"(675":[181],"vs.":[182,191],"6,250":[183],"A100":[184],"days),":[186],"saving":[187],"nearly":[188],"\\$300,000":[189],"(\\$26,000":[190],"\\$320,000)":[192],"reducing":[194],"90%":[195],"Moreover,":[198],"compared":[199],"larger":[202],"SOTA":[203],"model,":[204],"RAPHAEL,":[205],"our":[206],"cost":[208],"merely":[210],"1%.":[211],"Extensive":[212],"experiments":[213],"demonstrate":[214],"excels":[217],"quality,":[220],"artistry,":[221],"semantic":[223],"control.":[224],"hope":[226],"will":[228],"provide":[229],"new":[230],"insights":[231],"startups":[237],"accelerate":[239],"building":[240],"their":[241],"own":[242],"high-quality":[243],"yet":[244],"low-cost":[245],"generative":[246],"from":[248],"scratch.":[249]},"counts_by_year":[{"year":2025,"cited_by_count":18},{"year":2024,"cited_by_count":12}],"updated_date":"2026-10-16T09:59:22.235173","created_date":"2025-10-10T00:00:00"}