{"id":"https://openalex.org/W7161001380","doi":"https://doi.org/10.48550/arxiv.2605.12495","title":"AlphaGRPO: Unlocking Self-Reflective Multimodal Generation in UMMs via Decompositional Verifiable Reward","display_name":"AlphaGRPO: Unlocking Self-Reflective Multimodal Generation in UMMs via Decompositional Verifiable Reward","publication_year":2026,"publication_date":"2026-05-12","ids":{"openalex":"https://openalex.org/W7161001380","doi":"https://doi.org/10.48550/arxiv.2605.12495"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.12495","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.12495","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.2605.12495","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5001888307","display_name":"Runhui Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Runhui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136055381","display_name":"Jie Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Jie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136058501","display_name":"Rui Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136017073","display_name":"Zhe Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Zhe","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136068877","display_name":"Hengshuang Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Hengshuang","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":0.49149999022483826,"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":0.49149999022483826,"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/T10028","display_name":"Topic Modeling","score":0.2281000018119812,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.02930000051856041,"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/verifiable-secret-sharing","display_name":"Verifiable secret sharing","score":0.823199987411499},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.40700000524520874},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.38690000772476196},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.32350000739097595},{"id":"https://openalex.org/keywords/medical-diagnosis","display_name":"Medical diagnosis","score":0.30059999227523804},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.2750999927520752}],"concepts":[{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.823199987411499},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8169000148773193},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5385000109672546},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4607999920845032},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.4413999915122986},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.40700000524520874},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.38690000772476196},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.32350000739097595},{"id":"https://openalex.org/C534262118","wikidata":"https://www.wikidata.org/wiki/Q177719","display_name":"Medical diagnosis","level":2,"score":0.30059999227523804},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.2750999927520752},{"id":"https://openalex.org/C2779038628","wikidata":"https://www.wikidata.org/wiki/Q7248497","display_name":"Programming by demonstration","level":3,"score":0.26980000734329224},{"id":"https://openalex.org/C115901376","wikidata":"https://www.wikidata.org/wiki/Q184199","display_name":"Automation","level":2,"score":0.26330000162124634},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.25870001316070557},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.2574000060558319},{"id":"https://openalex.org/C157657479","wikidata":"https://www.wikidata.org/wiki/Q2367247","display_name":"Closed captioning","level":3,"score":0.25060001015663147}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.12495","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.12495","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.2605.12495","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.12495","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0],"this":[1],"paper,":[2],"we":[3,80],"propose":[4],"AlphaGRPO,":[5],"a":[6,112],"novel":[7],"framework":[8],"that":[9,124,157],"applies":[10],"Group":[11],"Relative":[12],"Policy":[13],"Optimization":[14],"(GRPO)":[15],"to":[16,22,39,95,115,166],"AR-Diffusion":[17],"Unified":[18],"Multimodal":[19],"Models":[20],"(UMMs)":[21],"enhance":[23],"multimodal":[24,78,130],"generation":[25,131],"capabilities":[26],"without":[27,149],"an":[28,93],"additional":[29],"cold-start":[30],"stage.":[31],"Our":[32],"approach":[33,161],"unlocks":[34],"the":[35,48,70,82],"model's":[36],"intrinsic":[37],"potential":[38],"perform":[40],"advanced":[41],"reasoning":[42],"tasks:":[43],"Reasoning":[44],"Text-to-Image":[45],"Generation,":[46],"where":[47,58],"model":[49],"actively":[50],"infers":[51],"implicit":[52],"user":[53,98],"intents,":[54],"and":[55,62,104,118,137],"Self-Reflective":[56],"Refinement,":[57],"it":[59],"autonomously":[60],"diagnoses":[61],"corrects":[63],"misalignments":[64],"in":[65,144],"generated":[66],"outputs.":[67],"To":[68],"address":[69],"challenge":[71],"of":[72],"providing":[73],"stable":[74],"supervision":[75],"for":[76],"real-world":[77],"generation,":[79],"introduce":[81],"Decompositional":[83],"Verifiable":[84],"Reward":[85],"(DVReward).":[86],"Unlike":[87],"holistic":[88],"scalar":[89],"rewards,":[90],"DVReward":[91],"utilizes":[92],"LLM":[94],"decompose":[96],"complex":[97],"requests":[99],"into":[100],"atomic,":[101],"verifiable":[102],"semantic":[103],"quality":[105],"questions,":[106],"which":[107],"are":[108],"then":[109],"evaluated":[110],"by":[111],"general":[113],"MLLM":[114],"provide":[116],"reliable":[117],"interpretable":[119],"feedback.":[120],"Extensive":[121],"experiments":[122],"demonstrate":[123],"AlphaGRPO":[125],"yields":[126],"robust":[127],"improvements":[128],"across":[129],"benchmarks,":[132],"including":[133],"GenEval,":[134],"TIIF-Bench,":[135],"DPG-Bench":[136],"WISE,":[138],"while":[139],"also":[140],"achieving":[141],"significant":[142],"gains":[143],"editing":[145,152],"tasks":[146],"on":[147,151],"GEdit":[148],"training":[150],"tasks.":[153],"These":[154],"results":[155],"validate":[156],"our":[158],"self-reflective":[159],"reinforcement":[160],"effectively":[162],"leverages":[163],"inherent":[164],"understanding":[165],"guide":[167],"high-fidelity":[168],"generation.":[169],"Project":[170],"page:":[171],"https://huangrh99.github.io/AlphaGRPO/":[172]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-14T00:00:00"}