{"id":"https://openalex.org/W7165655920","doi":"https://doi.org/10.48550/arxiv.2606.23087","title":"FlowTrain: Flow-Based Decoupled Training for Industrial-Grade Vision-Language Models","display_name":"FlowTrain: Flow-Based Decoupled Training for Industrial-Grade Vision-Language Models","publication_year":2026,"publication_date":"2026-06-22","ids":{"openalex":"https://openalex.org/W7165655920","doi":"https://doi.org/10.48550/arxiv.2606.23087"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.23087","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.23087","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.2606.23087","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5052014893","display_name":"Zhida Jiang","orcid":"https://orcid.org/0000-0002-7338-0724"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Zhida","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121622277","display_name":"Zhaolong Xing","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xing, Zhaolong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139201843","display_name":"Yang Pei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pei, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139193215","display_name":"Xiaolong Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Xiaolong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133363812","display_name":"Y H Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Yuanhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139189334","display_name":"Chengzhi Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Chengzhi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139171939","display_name":"Xiyu Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Xiyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139205929","display_name":"Haopeng Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Haopeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070874925","display_name":"Qingyuan Sang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sang, Qingyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048566538","display_name":"Lingfeng Zhou","orcid":"https://orcid.org/0000-0002-8160-4273"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Lingfeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139213458","display_name":"Jiaxing Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Jiaxing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139210758","display_name":"Zicheng Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zicheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101401728","display_name":"Wenzhe Wang","orcid":"https://orcid.org/0000-0003-2378-0540"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Wenzhe","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139189835","display_name":"Xinyu Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Xinyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139135851","display_name":"Yan Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139194680","display_name":"Zhen Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Zhen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139146148","display_name":"Ke Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Ke","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9079999923706055,"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":0.9079999923706055,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.010599999688565731,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.008299999870359898,"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/dataflow","display_name":"Dataflow","score":0.739300012588501},{"id":"https://openalex.org/keywords/allocator","display_name":"Allocator","score":0.5999000072479248},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.5594000220298767},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5306000113487244},{"id":"https://openalex.org/keywords/decoupling","display_name":"Decoupling (probability)","score":0.5055000185966492},{"id":"https://openalex.org/keywords/encoder","display_name":"Encoder","score":0.5029000043869019},{"id":"https://openalex.org/keywords/construct","display_name":"Construct (python library)","score":0.4788999855518341}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8154000043869019},{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.739300012588501},{"id":"https://openalex.org/C162262903","wikidata":"https://www.wikidata.org/wiki/Q343527","display_name":"Allocator","level":2,"score":0.5999000072479248},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.5594000220298767},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5306000113487244},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.5299000144004822},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5110999941825867},{"id":"https://openalex.org/C205606062","wikidata":"https://www.wikidata.org/wiki/Q5249645","display_name":"Decoupling (probability)","level":2,"score":0.5055000185966492},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.5029000043869019},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.4788999855518341},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.45159998536109924},{"id":"https://openalex.org/C2781172179","wikidata":"https://www.wikidata.org/wiki/Q853109","display_name":"Parallelism (grammar)","level":2,"score":0.4424999952316284},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.3978999853134155},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3903999924659729},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.3813000023365021},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C160403385","wikidata":"https://www.wikidata.org/wiki/Q220543","display_name":"Queue","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.25440001487731934}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.23087","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.23087","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.2606.23087","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.23087","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":[{"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9","score":0.5903576016426086}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Industrial-grade":[0],"distributed":[1],"training":[2,60,74,79],"of":[3,13,112],"vision-language":[4],"models":[5],"(VLMs)":[6],"remains":[7],"far":[8],"less":[9],"efficient":[10],"than":[11],"that":[12,23,35,50,76,123,159],"unimodal":[14],"LLMs.":[15],"Existing":[16],"solutions":[17,53],"either":[18],"follow":[19],"a":[20,32,42,71,81,86,98,119,130],"monolithic":[21],"design":[22],"assigns":[24,124],"uniform":[25],"parallelism":[26,126],"to":[27,140,147,167,175],"heterogeneous":[28,120],"modules":[29,37],"or":[30],"adopt":[31],"disaggregated":[33],"deployment":[34],"separates":[36],"while":[38],"executing":[39],"them":[40],"as":[41,80],"batch-synchronized":[43],"pipeline.":[44],"In":[45],"this":[46,66,104],"paper,":[47],"we":[48,68],"highlight":[49],"the":[51,109,148,172],"above":[52],"are":[54],"still":[55],"not":[56],"sufficient,":[57],"and":[58,92,114,165],"VLM":[59,78],"can":[61,94],"be":[62],"further":[63,117],"decoupled.":[64],"To":[65],"end,":[67],"present":[69],"FlowTrain,":[70],"flow-based":[72],"decoupled":[73],"framework":[75],"reformulates":[77],"producer-consumer":[82],"dataflow":[83],"coordinated":[84],"through":[85],"unified":[87],"memory":[88],"pool.":[89],"The":[90,134],"encoder":[91],"backbone":[93],"progress":[95],"independently":[96],"over":[97,162],"global":[99],"virtual":[100],"address":[101],"space.":[102],"Since":[103],"execution":[105],"decoupling":[106],"fundamentally":[107],"changes":[108],"optimization":[110],"objective":[111],"allocation":[113],"scheduling,":[115],"FlowTrain":[116,160],"introduces":[118],"parallel":[121],"allocator":[122],"module-specific":[125],"strategies":[127],"by":[128],"solving":[129],"throughput":[131,169],"matching":[132],"problem.":[133],"dynamic":[135],"packing":[136],"scheduler":[137],"is":[138],"used":[139],"construct":[141],"balanced":[142],"microbatches":[143],"at":[144],"runtime":[145],"according":[146],"actual":[149],"LLM-side":[150],"computation":[151],"cost.":[152],"Extensive":[153],"experiments":[154],"on":[155],"real-world":[156],"workloads":[157],"show":[158],"achieves":[161],"50%":[163],"MFU":[164],"up":[166],"1.7x":[168],"improvement,":[170],"narrowing":[171],"efficiency":[173],"gap":[174],"LLM-only":[176],"training.":[177]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-24T00:00:00"}