{"id":"https://openalex.org/W7204886679","doi":"https://doi.org/10.48550/arxiv.2608.29335","title":"GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling","display_name":"GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling","publication_year":2026,"publication_date":"2026-08-29","ids":{"openalex":"https://openalex.org/W7204886679","doi":"https://doi.org/10.48550/arxiv.2608.29335"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.29335","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29335","pdf_url":"https://arxiv.org/pdf/2608.29335","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.29335","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5153508384","display_name":"Guangting Zheng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Guangting","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143782814","display_name":"Yiyuan Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yiyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123819522","display_name":"Tao Yang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Tao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127781696","display_name":"Yunpeng Chen","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Yunpeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125990352","display_name":"Rui Zhu","orcid":"https://orcid.org/0009-0007-7226-6923","observed_orcids":["https://orcid.org/0009-0007-7226-6923"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124303994","display_name":"Jiajun Deng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Jiajun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5124895167","display_name":"Yanyong Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yanyong","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":0.9990000128746033,"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":0.9990000128746033,"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":0.00019999999494757503,"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/T10036","display_name":"Advanced Neural Network Applications","score":9.999999747378752e-05,"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/end-to-end-learning","display_name":"end-to-end learning","score":0.8999999761581421},{"id":"https://openalex.org/keywords/variational-autoencoder","display_name":"variational autoencoder","score":0.796999990940094},{"id":"https://openalex.org/keywords/image-to-image-translation","display_name":"image-to-image translation","score":0.5320000052452087},{"id":"https://openalex.org/keywords/text-to-image-generation","display_name":"text-to-image generation","score":0.8050000071525574},{"id":"https://openalex.org/keywords/representation-learning","display_name":"representation learning","score":0.7369999885559082}],"concepts":[{"id":"https://openalex.org/C177769412","wikidata":"https://www.wikidata.org/wiki/Q278090","display_name":"Prior probability","level":3,"score":0.8406999707221985},{"id":"https://openalex.org/C101738243","wikidata":"https://www.wikidata.org/wiki/Q786435","display_name":"Autoencoder","level":3,"score":0.6589999794960022},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5939000248908997},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.5572999715805054},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5153999924659729},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.5077999830245972},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.4327999949455261},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.4246000051498413},{"id":"https://openalex.org/C207390915","wikidata":"https://www.wikidata.org/wiki/Q1230525","display_name":"Divergence (linguistics)","level":2,"score":0.4178999960422516},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.40230000019073486},{"id":"https://openalex.org/C51167844","wikidata":"https://www.wikidata.org/wiki/Q4422623","display_name":"Latent variable","level":2,"score":0.39879998564720154},{"id":"https://openalex.org/C2780767217","wikidata":"https://www.wikidata.org/wiki/Q5532421","display_name":"Generality","level":2,"score":0.37439998984336853},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36980000138282776},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3449999988079071},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.3427000045776367},{"id":"https://openalex.org/C65965080","wikidata":"https://www.wikidata.org/wiki/Q1806885","display_name":"Latent variable model","level":3,"score":0.320499986410141},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.31310001015663147},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.3109999895095825},{"id":"https://openalex.org/C141379421","wikidata":"https://www.wikidata.org/wiki/Q6094427","display_name":"Iterative reconstruction","level":2,"score":0.29670000076293945},{"id":"https://openalex.org/C9679016","wikidata":"https://www.wikidata.org/wiki/Q1417473","display_name":"Principle of maximum entropy","level":2,"score":0.27959999442100525},{"id":"https://openalex.org/C160234255","wikidata":"https://www.wikidata.org/wiki/Q812535","display_name":"Bayesian inference","level":3,"score":0.2551000118255615},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.2551000118255615},{"id":"https://openalex.org/C167981619","wikidata":"https://www.wikidata.org/wiki/Q1685498","display_name":"Cross entropy","level":3,"score":0.2506999969482422}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.29335","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29335","pdf_url":"https://arxiv.org/pdf/2608.29335","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.29335","doi":"https://doi.org/10.48550/arxiv.2608.29335","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.29335","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.29335","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29335","pdf_url":"https://arxiv.org/pdf/2608.29335","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7166829976","https://openalex.org/W7160095423","https://openalex.org/W7106136681","https://openalex.org/W7214936584","https://openalex.org/W4417461713","https://openalex.org/W7140202574","https://openalex.org/W6948626004","https://openalex.org/W7106789420","https://openalex.org/W7172008835","https://openalex.org/W7214771596"],"abstract_inverted_index":{"Latent":[0],"generative":[1,17,148,249],"models":[2,34],"typically":[3],"follow":[4],"a":[5,9,16,54,143,193,208],"two-stage":[6],"pipeline,":[7],"training":[8,32,42,136],"variational":[10],"autoencoder":[11],"for":[12,86,227],"reconstruction":[13,89,105,112,157,161],"and":[14,52,70,90,106,115,122,140,179,189,199,229,232,251],"then":[15],"model":[18],"on":[19,127,203,213],"the":[20,67,76,80,96,132,152,221,241],"frozen":[21],"latent":[22,50,68,102,138,153,246],"space.":[23],"Since":[24],"reconstruction-optimized":[25],"latents":[26,226],"are":[27],"not":[28],"necessarily":[29],"generation-friendly,":[30],"jointly":[31],"both":[33],"is":[35,47,84,113,120,162],"an":[36],"appealing":[37],"alternative.":[38],"However,":[39],"direct":[40,134],"end-to-end":[41,135,187,245],"remains":[43],"challenging,":[44],"as":[45],"it":[46],"prone":[48],"to":[49,94,124,165,223,233],"collapse":[51,139],"faces":[53],"generation-reconstruction":[55],"conflict.":[56],"We":[57,169],"revisit":[58],"this":[59],"problem":[60],"by":[61],"analyzing":[62],"how":[63],"different":[64],"objectives":[65],"shape":[66],"space":[69,154],"identify":[71],"two":[72],"key":[73],"insights.":[74],"First,":[75],"entropy":[77,99],"term":[78],"in":[79],"Kullback-Leibler":[81],"divergence":[82],"objective":[83,149,188],"essential":[85],"preventing":[87],"collapse:":[88],"prior":[91],"fitting":[92],"tend":[93],"shrink":[95],"posterior,":[97],"while":[98,205],"preserves":[100],"non-degenerate":[101],"uncertainty.":[103],"Second,":[104],"generation":[107,119,228],"exhibit":[108],"asymmetric":[109],"learning":[110,247],"dynamics:":[111],"fast":[114],"strongly":[116],"supervised,":[117],"whereas":[118],"slower":[121],"harder":[123],"optimize.":[125],"Based":[126],"these":[128],"insights,":[129],"we":[130,219],"achieve":[131],"first":[133,150],"without":[137,201],"propose":[141],"GenFirst,":[142,190],"simple":[144],"generation-before-reconstruction":[145],"strategy.":[146],"The":[147],"shapes":[151],"under":[155],"weak":[156],"pressure,":[158],"after":[159],"which":[160],"progressively":[163],"strengthened":[164],"recover":[166],"visual":[167,225],"details.":[168],"validate":[170],"GenFirst":[171],"with":[172,176,182,197],"continuous":[173,234],"autoregressive":[174],"priors":[175,181,250],"exact":[177],"likelihoods":[178],"SiT":[180,191],"implicit":[183],"likelihoods.":[184],"With":[185],"our":[186],"achieves":[192],"gFID":[194],"of":[195,211,243],"0.97":[196],"CFG":[198,202],"1.45":[200],"ImageNet-256,":[204],"MMDiT":[206],"reaches":[207],"GenEval":[209],"score":[210],"0.90":[212],"text-to-image":[214],"generation.":[215,237],"Beyond":[216],"image":[217],"generation,":[218],"extend":[220],"framework":[222],"shared":[224],"representation":[230],"learning,":[231],"unified":[235],"text-image":[236],"These":[238],"results":[239],"demonstrate":[240],"generality":[242],"stable":[244],"across":[248],"modalities.":[252]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-02T00:00:00"}