{"id":"https://openalex.org/W7138202757","doi":"https://doi.org/10.48550/arxiv.2603.15271","title":"Flash-Unified: A Training-Free and Task-Aware Acceleration Framework for Native Unified Models","display_name":"Flash-Unified: A Training-Free and Task-Aware Acceleration Framework for Native Unified Models","publication_year":2026,"publication_date":"2026-03-16","ids":{"openalex":"https://openalex.org/W7138202757","doi":"https://doi.org/10.48550/arxiv.2603.15271"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.15271","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.15271","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.15271","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129670163","display_name":"Junlong Ke","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ke, Junlong","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102635197","display_name":"Zichen Wen","orcid":"https://orcid.org/0009-0002-6157-5898"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wen, Zichen","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124108288","display_name":"Boxue Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Boxue","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121516269","display_name":"Yantai Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Yantai","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121190807","display_name":"Xuyang Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Xuyang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103034424","display_name":"Chenfei Liao","orcid":"https://orcid.org/0009-0005-6952-1300"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liao, Chenfei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129654630","display_name":"Zhaorun Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Zhaorun","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129746191","display_name":"Shaobo Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Shaobo","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129642313","display_name":"Linfeng Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Linfeng","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5129670163"],"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7114999890327454,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7114999890327454,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.19220000505447388,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.0215000007301569,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6403999924659729},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.5339999794960022},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.5198000073432922},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.4950000047683716},{"id":"https://openalex.org/keywords/unified-model","display_name":"Unified Model","score":0.4657999873161316},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.4616999924182892},{"id":"https://openalex.org/keywords/graphical-model","display_name":"Graphical model","score":0.4108999967575073},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.39820000529289246},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.37779998779296875}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7731999754905701},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6403999924659729},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.5339999794960022},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.5198000073432922},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.4950000047683716},{"id":"https://openalex.org/C45493050","wikidata":"https://www.wikidata.org/wiki/Q7884934","display_name":"Unified Model","level":2,"score":0.4657999873161316},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.4616999924182892},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.44670000672340393},{"id":"https://openalex.org/C155846161","wikidata":"https://www.wikidata.org/wiki/Q1143367","display_name":"Graphical model","level":2,"score":0.4108999967575073},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.39820000529289246},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.37779998779296875},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.37610000371932983},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.34389999508857727},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.34389999508857727},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.32010000944137573},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3151000142097473},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.3151000142097473},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.3091000020503998},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.30649998784065247},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.2921999990940094},{"id":"https://openalex.org/C207390915","wikidata":"https://www.wikidata.org/wiki/Q1230525","display_name":"Divergence (linguistics)","level":2,"score":0.27959999442100525},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.27410000562667847},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.27090001106262207},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.27079999446868896},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.2651999890804291},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.2630000114440918},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.25459998846054077},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.25}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.15271","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.15271","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.15271","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.15271","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":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Native":[0],"unified":[1,59,81,211],"multimodal":[2,162],"models,":[3,60],"which":[4],"integrate":[5],"both":[6,118,200],"generative":[7],"and":[8,43,91,105,125,133,149,202,213],"understanding":[9,45,92,201],"capabilities,":[10],"face":[11],"substantial":[12],"computational":[13,34],"overhead":[14],"that":[15,110,191],"hinders":[16],"their":[17],"real-world":[18],"deployment.":[19],"Existing":[20],"acceleration":[21,107,198,217],"techniques":[22],"typically":[23],"employ":[24],"a":[25,94,103,141,150,175],"static,":[26],"monolithic":[27],"strategy,":[28],"ignoring":[29],"the":[30,54,78,146,154,166,180],"fundamental":[31],"divergence":[32],"in":[33],"profiles":[35],"between":[36],"iterative":[37],"generation":[38,90,203],"tasks":[39,46,204],"(e.g.,":[40,47],"image":[41],"generation)":[42],"single-pass":[44],"VQA).":[48],"In":[49],"this":[50],"work,":[51],"we":[52,101,120,139,169],"present":[53],"first":[55],"systematic":[56],"analysis":[57],"of":[58,183],"revealing":[61],"pronounced":[62],"parameter":[63,79],"specialization,":[64],"where":[65],"distinct":[66],"neuron":[67],"sets":[68],"are":[69],"critical":[70],"for":[71,89,145,153],"each":[72,114],"task.":[73],"This":[74],"implies":[75],"that,":[76],"at":[77,224],"level,":[80],"models":[82,212],"have":[83],"implicitly":[84],"internalized":[85],"separate":[86],"inference":[87,197],"pathways":[88],"within":[93],"single":[95],"architecture.":[96],"Based":[97],"on":[98,188],"these":[99],"insights,":[100],"introduce":[102,121,170],"training-free":[104],"task-aware":[106,216],"framework,":[108],"FlashU,":[109],"tailors":[111],"optimization":[112],"to":[113,130,178,195],"task's":[115],"demands.":[116],"Across":[117],"tasks,":[119],"Task-Specific":[122],"Network":[123],"Pruning":[124,173],"Dynamic":[126,171],"Layer":[127],"Skipping,":[128],"aiming":[129],"eliminate":[131],"inter-layer":[132],"task-specific":[134],"redundancy.":[135],"For":[136,161],"visual":[137,184],"generation,":[138],"implement":[140],"time-varying":[142],"control":[143],"signal":[144],"guidance":[147],"scale":[148],"temporal":[151],"approximation":[152],"diffusion":[155],"head":[156],"via":[157,174],"Diffusion":[158],"Head":[159],"Cache.":[160],"understanding,":[163],"building":[164],"upon":[165],"pruned":[167],"model,":[168],"Token":[172],"V-Norm":[176],"Proxy":[177],"exploit":[179],"spatial":[181],"redundancy":[182],"inputs.":[185],"Extensive":[186],"experiments":[187],"Show-o2":[189],"demonstrate":[190],"FlashU":[192],"achieves":[193],"1.78$\\times$":[194],"2.01$\\times$":[196],"across":[199],"while":[205],"maintaining":[206],"SOTA":[207],"performance,":[208],"outperforming":[209],"competing":[210],"validating":[214],"our":[215],"paradigm.":[218],"Our":[219],"code":[220],"is":[221],"publicly":[222],"available":[223],"https://github.com/Rirayh/FlashU.":[225]},"counts_by_year":[],"updated_date":"2026-03-18T06:31:55.123368","created_date":"2026-03-18T00:00:00"}