{"id":"https://openalex.org/W7162579502","doi":"https://doi.org/10.48550/arxiv.2605.26636","title":"JetViT: Efficient High-Resolution Vision Transformer with Post-Training Attention Search","display_name":"JetViT: Efficient High-Resolution Vision Transformer with Post-Training Attention Search","publication_year":2026,"publication_date":"2026-05-26","ids":{"openalex":"https://openalex.org/W7162579502","doi":"https://doi.org/10.48550/arxiv.2605.26636"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.26636","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.26636","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":null,"license_id":null,"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.2605.26636","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137134266","display_name":"Dongyun Zou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zou, Dongyun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5009067861","display_name":"Zhuoyang Zhang","orcid":"https://orcid.org/0000-0002-3312-6246"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zhuoyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137128775","display_name":"Junyu Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Junyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137138400","display_name":"Wenkun He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Wenkun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137133977","display_name":"Qinhe Peng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Qinhe","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058387246","display_name":"Hanrong Ye","orcid":"https://orcid.org/0000-0002-7986-6143"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ye, Hanrong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137106375","display_name":"Yao Lu","orcid":"https://orcid.org/0000-0002-6025-1878"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Yao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137136763","display_name":"Hongxu Yin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yin, Hongxu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137138298","display_name":"Yu Wang","orcid":"https://orcid.org/0009-0000-5937-8778"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137143225","display_name":"Song Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Song","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100940805","display_name":"Han Cai","orcid":"https://orcid.org/0009-0007-2242-6984"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cai, Han","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/T10036","display_name":"Advanced Neural Network Applications","score":0.3889999985694885,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.3889999985694885,"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/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.36239999532699585,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.06459999829530716,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"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/transformer","display_name":"Transformer","score":0.5339000225067139},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5159000158309937},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.3610999882221222},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.3154999911785126},{"id":"https://openalex.org/keywords/block","display_name":"Block (permutation group theory)","score":0.3057999908924103},{"id":"https://openalex.org/keywords/machine-vision","display_name":"Machine vision","score":0.30059999227523804}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6819999814033508},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5339000225067139},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5159000158309937},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.47099998593330383},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3612000048160553},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3610999882221222},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3361999988555908},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.3154999911785126},{"id":"https://openalex.org/C2777210771","wikidata":"https://www.wikidata.org/wiki/Q4927124","display_name":"Block (permutation group theory)","level":2,"score":0.3057999908924103},{"id":"https://openalex.org/C5339829","wikidata":"https://www.wikidata.org/wiki/Q1425977","display_name":"Machine vision","level":2,"score":0.30059999227523804},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.29829999804496765},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2897999882698059},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.2870999872684479},{"id":"https://openalex.org/C75165309","wikidata":"https://www.wikidata.org/wiki/Q2258979","display_name":"Search engine indexing","level":2,"score":0.28049999475479126},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.2759999930858612},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.2531999945640564}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.26636","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.26636","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.26636","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.26636","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":null,"license_id":null,"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":{"We":[0,114,148],"introduce":[1],"JetViT,":[2],"a":[3,41],"novel":[4],"family":[5],"of":[6,16,34,101],"hybrid-architecture":[7],"Vision":[8],"Transformer":[9],"(ViT)":[10],"models":[11,21,156],"that":[12,45],"match":[13],"the":[14,32,68,74,82,92,98,128],"accuracy":[15],"state-of-the-art":[17],"full-attention":[18,48,59,112],"vision":[19,121],"foundation":[20,122],"while":[22],"achieving":[23],"substantially":[24],"higher":[25,137],"inference":[26],"efficiency":[27],"on":[28,117],"high-resolution":[29,120],"images.":[30],"At":[31],"core":[33],"our":[35,151],"approach":[36],"is":[37],"Post-Training":[38,77],"Attention":[39,78],"Search,":[40],"post-training":[42],"acceleration":[43],"framework":[44],"converts":[46],"pre-trained":[47],"ViTs":[49],"into":[50],"efficient":[51],"hybrid-attention":[52],"variants":[53],"by":[54],"identifying":[55,108],"and":[56,70,103,106,109,125,139,153],"replacing":[57],"redundant":[58],"blocks":[60],"with":[61],"linear":[62],"or":[63],"window-attention":[64,104],"blocks.":[65,113],"By":[66],"inheriting":[67],"MLP":[69],"attention":[71],"weights":[72],"from":[73],"base":[75],"model,":[76],"Search":[79],"efficiently":[80],"explores":[81],"architectural":[83],"design":[84],"space":[85],"through":[86],"three":[87],"key":[88],"steps:":[89],"(1)":[90],"optimizing":[91],"linear-attention":[93,102],"block":[94],"design;":[95],"(2)":[96],"finding":[97],"best":[99],"combination":[100],"blocks;":[105],"(3)":[107],"preserving":[110],"critical":[111],"evaluate":[115],"JetViT":[116,132],"two":[118],"representative":[119],"models,":[123],"DINOv3":[124],"DepthAnythingV2.":[126],"On":[127],"NVIDIA":[129],"H100":[130],"GPU,":[131],"achieves":[133],"up":[134,140],"to":[135,141],"1.79x":[136],"throughput":[138],"44.81%":[142],"lower":[143],"latency":[144],"without":[145],"sacrificing":[146],"accuracy.":[147],"will":[149],"release":[150],"code":[152],"accelerated":[154],"ViT":[155],"soon.":[157]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-28T00:00:00"}