{"id":"https://openalex.org/W7164530017","doi":"https://doi.org/10.48550/arxiv.2606.13679","title":"InterleaveThinker: Reinforcing Agentic Interleaved Generation","display_name":"InterleaveThinker: Reinforcing Agentic Interleaved Generation","publication_year":2026,"publication_date":"2026-06-11","ids":{"openalex":"https://openalex.org/W7164530017","doi":"https://doi.org/10.48550/arxiv.2606.13679"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.13679","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.13679","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.2606.13679","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138530884","display_name":"Dian Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Dian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138547284","display_name":"Harry Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Harry","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138489309","display_name":"Manyuan Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Manyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138531864","display_name":"Kaituo Feng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Feng, Kaituo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049150264","display_name":"Zoey Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Zoey","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138551009","display_name":"Ray Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Ray","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138510433","display_name":"Hongsheng Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Hongsheng","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":0.5411999821662903,"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":0.5411999821662903,"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":0.22750000655651093,"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/T11019","display_name":"Image Enhancement Techniques","score":0.017500000074505806,"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/generator","display_name":"Generator (circuit theory)","score":0.7222999930381775},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.6725999712944031},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.6299999952316284},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.46149998903274536},{"id":"https://openalex.org/keywords/construct","display_name":"Construct (python library)","score":0.412200003862381},{"id":"https://openalex.org/keywords/planner","display_name":"Planner","score":0.3513999879360199}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8082000017166138},{"id":"https://openalex.org/C2780992000","wikidata":"https://www.wikidata.org/wiki/Q17016113","display_name":"Generator (circuit theory)","level":3,"score":0.7222999930381775},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.6725999712944031},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6299999952316284},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4805000126361847},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.46149998903274536},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.412200003862381},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3596000075340271},{"id":"https://openalex.org/C2776999362","wikidata":"https://www.wikidata.org/wiki/Q2349274","display_name":"Planner","level":2,"score":0.3513999879360199},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.3160000145435333},{"id":"https://openalex.org/C173246807","wikidata":"https://www.wikidata.org/wiki/Q7833062","display_name":"Trajectory optimization","level":3,"score":0.3075000047683716},{"id":"https://openalex.org/C151346624","wikidata":"https://www.wikidata.org/wiki/Q5276129","display_name":"Digital pattern generator","level":3,"score":0.295199990272522},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.29499998688697815},{"id":"https://openalex.org/C3018907156","wikidata":"https://www.wikidata.org/wiki/Q7785058","display_name":"Third generation","level":2,"score":0.27880001068115234},{"id":"https://openalex.org/C133731056","wikidata":"https://www.wikidata.org/wiki/Q4917288","display_name":"Control engineering","level":1,"score":0.27619999647140503},{"id":"https://openalex.org/C133162039","wikidata":"https://www.wikidata.org/wiki/Q1061077","display_name":"Code generation","level":3,"score":0.2754000127315521},{"id":"https://openalex.org/C90509273","wikidata":"https://www.wikidata.org/wiki/Q11012","display_name":"Robot","level":2,"score":0.2563999891281128}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.13679","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.13679","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.2606.13679","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.13679","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"image":[1,67,87,199],"generators":[2],"have":[3],"demonstrated":[4],"impressive":[5],"photorealism":[6],"and":[7,13,35,115,129,177,212,236],"instruction-following":[8],"capabilities":[9],"in":[10,31,49],"single-image":[11],"generation":[12,24,71,149,157,188,203],"editing.":[14],"However,":[15],"constrained":[16],"by":[17],"their":[18],"architectures,":[19],"they":[20],"cannot":[21],"achieve":[22],"interleaved":[23,70,156,202],"(text-image":[25],"sequence),":[26],"which":[27],"has":[28],"crucial":[29],"applications":[30],"visual":[32],"narratives,":[33],"guidance,":[34],"embodied":[36],"manipulation.":[37],"Even":[38],"the":[39,58,81,86,90,104,112,117,127,142,166,186,219],"latest":[40],"open-source":[41],"Unified":[42],"Multimodal":[43],"Models":[44],"(UMMs)":[45],"exhibit":[46],"limited":[47],"performance":[48,196,207],"this":[50,53,123],"regard.":[51],"In":[52],"paper,":[54],"we":[55,74,97,125,137,173,230],"introduce":[56,98],"InterleaveThinker,":[57],"first":[59],"multi-agent":[60],"pipeline":[61],"designed":[62],"to":[63,79,102,131,140,183,209],"endow":[64],"any":[65],"existing":[66],"generator":[68,88,163],"with":[69],"capabilities.":[72],"Specifically,":[73],"employ":[75],"a":[76,99,133,148,154],"planner":[77],"agent":[78,101],"organize":[80],"image-text":[82],"input":[83],"sequence,":[84],"instructing":[85],"on":[89,222,227,234],"required":[91],"execution":[92],"at":[93],"each":[94],"step.":[95],"Subsequently,":[96],"critic":[100],"evaluate":[103],"generator's":[105],"outputs,":[106],"identify":[107],"samples":[108],"that":[109,193],"deviate":[110],"from":[111],"planned":[113],"instructions,":[114],"refine":[116],"instructions":[118],"for":[119,225],"regeneration.":[120],"To":[121],"implement":[122],"pipeline,":[124],"construct":[126],"Interleave-Planner-SFT-80k":[128],"Interleave-Critic-SFT-112k":[130],"perform":[132],"format":[134],"cold-start.":[135],"Then":[136],"develop":[138],"Interleave-Critic-RL-13k":[139],"reinforce":[141],"step-wise":[143,178],"instruction":[144],"correction":[145],"capability":[146],"within":[147],"trajectory":[150,158,168],"using":[151],"GRPO.":[152],"Since":[153],"single":[155],"may":[159],"involve":[160],"over":[161],"25":[162],"calls,":[164],"optimizing":[165],"entire":[167,187],"is":[169],"computationally":[170],"impractical.":[171],"Therefore,":[172],"propose":[174],"accuracy":[175],"reward":[176],"reward,":[179],"allowing":[180],"single-step":[181],"RL":[182],"effectively":[184],"guide":[185],"trajectory.":[189],"The":[190],"results":[191],"show":[192],"InterleaveThinker":[194],"improves":[195],"across":[197],"various":[198],"generators.":[200],"On":[201],"benchmarks,":[204],"it":[205,215],"achieves":[206],"comparable":[208],"Nano":[210],"Banana":[211],"GPT-5.":[213],"Surprisingly,":[214],"also":[216],"significantly":[217],"enhances":[218],"base":[220],"model":[221],"reasoning-based":[223],"benchmarks;":[224],"example,":[226],"4-step":[228],"FLUX.2-klein,":[229],"observe":[231],"substantial":[232],"gains":[233],"WISE":[235],"RISE.":[237]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-13T00:00:00"}