{"id":"https://openalex.org/W7163183833","doi":"https://doi.org/10.48550/arxiv.2606.01985","title":"MT-EditFlow: Reinforcement Learning for Multi-Turn Image Editing with Flow Matching","display_name":"MT-EditFlow: Reinforcement Learning for Multi-Turn Image Editing with Flow Matching","publication_year":2026,"publication_date":"2026-06-01","ids":{"openalex":"https://openalex.org/W7163183833","doi":"https://doi.org/10.48550/arxiv.2606.01985"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.01985","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01985","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":"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.2606.01985","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137692341","display_name":"Jiahui Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Jiahui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137649655","display_name":"Yasi Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yasi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137644234","display_name":"Tianyu Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Tianyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137703766","display_name":"Shu Wang","orcid":"https://orcid.org/0000-0003-2061-2445"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Shu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137705731","display_name":"Jianwen Xie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Jianwen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137701851","display_name":"Oscar Leong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Leong, Oscar","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137650576","display_name":"Mingyuan Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Mingyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137642552","display_name":"Nanzhu Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Nanzhu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137647143","display_name":"Ying Nian Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Ying Nian","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":0.4235999882221222,"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":0.4235999882221222,"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":0.08919999748468399,"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/T12859","display_name":"Cell Image Analysis Techniques","score":0.06300000101327896,"subfield":{"id":"https://openalex.org/subfields/1304","display_name":"Biophysics"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.8130000233650208},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6334999799728394},{"id":"https://openalex.org/keywords/image-editing","display_name":"Image editing","score":0.4717999994754791},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.3959999978542328},{"id":"https://openalex.org/keywords/signal","display_name":"SIGNAL (programming language)","score":0.3797999918460846},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.36959999799728394},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.3637000024318695}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.8130000233650208},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8048999905586243},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6334999799728394},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6315000057220459},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4738999903202057},{"id":"https://openalex.org/C2776674983","wikidata":"https://www.wikidata.org/wiki/Q545981","display_name":"Image editing","level":3,"score":0.4717999994754791},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.3959999978542328},{"id":"https://openalex.org/C2779843651","wikidata":"https://www.wikidata.org/wiki/Q7390335","display_name":"SIGNAL (programming language)","level":2,"score":0.3797999918460846},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.36959999799728394},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.3637000024318695},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.35899999737739563},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3472999930381775},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.3028999865055084},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2815999984741211},{"id":"https://openalex.org/C2776608160","wikidata":"https://www.wikidata.org/wiki/Q4785462","display_name":"Natural (archaeology)","level":2,"score":0.2815999984741211},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.27950000762939453},{"id":"https://openalex.org/C155542232","wikidata":"https://www.wikidata.org/wiki/Q736111","display_name":"Optical flow","level":3,"score":0.27889999747276306},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.2639999985694885},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.2623000144958496},{"id":"https://openalex.org/C110157686","wikidata":"https://www.wikidata.org/wiki/Q922122","display_name":"Broadcasting (networking)","level":2,"score":0.25029999017715454}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.01985","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01985","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":"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.2606.01985","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01985","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":"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":{"Recent":[0],"breakthroughs":[1],"in":[2,38,207,239],"instruction-based":[3],"image":[4,50,104],"editing":[5,18,28,83,174],"have":[6],"captured":[7],"significant":[8],"attention,":[9],"as":[10,216],"models":[11,29,214],"are":[12],"now":[13],"capable":[14],"of":[15],"handling":[16],"real-world":[17],"demands":[19],"with":[20,111],"the":[21,53,62,71,134,168,172,178],"practicality":[22],"required":[23],"by":[24,137,204],"everyday":[25],"users.":[26],"However,":[27],"trained":[30],"primarily":[31],"for":[32,102,142,232],"single-turn":[33],"edits":[34],"often":[35],"break":[36],"down":[37],"multi-turn":[39,109,185],"editing--the":[40],"natural":[41,236],"interactive":[42],"setting":[43],"where":[44,65,77],"a":[45,66,92,108,112,117,230],"user":[46],"iteratively":[47],"refines":[48],"an":[49],"based":[51],"on":[52],"model's":[54],"own":[55],"previous":[56],"outputs.":[57],"This":[58],"failure":[59],"stems":[60],"from":[61],"all-or-nothing":[63],"requirement,":[64],"single":[67],"failed":[68],"turn":[69],"compromises":[70],"entire":[72,173],"sequence,":[73],"and":[74,124,132,153,155,183,224,235],"error":[75],"propagation,":[76],"exposure":[78,226],"bias":[79,152],"leads":[80],"to":[81,98,115,121,148,159],"compounding":[82],"errors.":[84],"To":[85],"address":[86],"these":[87],"challenges,":[88],"we":[89],"introduce":[90],"MT-EditFlow,":[91],"flow-matching":[93],"reinforcement":[94,126],"learning":[95,127],"framework":[96],"designed":[97],"optimize":[99,133],"reward":[100,135,151,161],"signals":[101],"sequential":[103],"editing.":[105],"MT-EditFlow":[106,192,228],"integrates":[107],"perspective":[110],"multi-reward":[113],"formulation":[114],"provide":[116],"unified":[118],"structure":[119],"applicable":[120],"both":[122],"GRPO":[123],"NFT-based":[125],"methods.":[128],"We":[129],"systematically":[130],"analyze":[131],"signal":[136],"investigating":[138],"effective":[139],"scoring":[140],"strategies":[141],"turn-level":[143],"aggregation,":[144],"VLM":[145],"reasoning":[146],"modes":[147],"trade":[149],"off":[150],"variance,":[154],"advantage":[156,170],"fusion":[157],"levels":[158],"prevent":[160],"hacking.":[162],"Our":[163],"findings":[164],"reveal":[165],"that":[166,191],"broadcasting":[167],"aggregated":[169],"across":[171,196],"trajectory":[175],"effectively":[176],"bridges":[177],"gap":[179],"between":[180],"local":[181],"planning":[182],"global":[184],"task":[186],"success.":[187],"Extensive":[188],"experiments":[189],"demonstrate":[190],"significantly":[193],"improves":[194],"performance":[195],"diverse":[197],"base":[198],"models.":[199],"Notably,":[200],"it":[201],"boosts":[202],"FLUX.1-Kontext-dev":[203],"6.85":[205],"points":[206],"turn-3":[208],"overall":[209],"performance,":[210],"surpassing":[211],"state-of-the-art":[212],"open-source":[213],"such":[215],"Qwen-Image-Edit.":[217],"By":[218],"maintaining":[219],"high":[220],"marginal":[221],"success":[222],"rates":[223],"reducing":[225],"bias,":[227],"provides":[229],"foundation":[231],"more":[233],"reliable":[234],"human-AI":[237],"collaboration":[238],"visual":[240],"content":[241],"creation.":[242]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-03T00:00:00"}