Order-AgnosticCrossEntropyforNon-AutoregressiveMachineTranslationCunxiaoDu1ZhaopengTu2JingJiang1AbstractAnumberofrecenteffortshaveexploredwaystoimprovetheNATmodels’abilitytohandlemultimodality.One...
Cross-modelBack-translatedDistillationforUnsupervisedMachineTranslationXuan-PhiNguyen12ShafiqJoty13Thanh-TungNguyen12WuKui2AiTiAw2Abstractmonolingualdatahasbeenactive.WhileRavi&Knight(2011)andKleme...
Non-autoregressiveMachineTranslationwithDisentangledContextTransformerJungoKasai1JamesCross2MarjanGhazvininejad2JiataoGu2Abstractconditionalindependenceandpreventsthemodelfromprop-erlycapturingtheh...
AlignedCrossEntropyforNon-AutoregressiveMachineTranslationMarjanGhazvininejad1VladimirKarpukhin1LukeZettlemoyer1OmerLevy1AbstractTargetYittastesprettygoodthoughNon-autoregressivemachineTranslationm...
MixtureModelsforDiverseMachineTranslation:TricksoftheTradeTianxiaoShen1MyleOtt2MichaelAuli2Marc’AurelioRanzato2Abstractstylessuchasformal/informal,literal/notliteral,etc.Thisraisesthequestionofhow...
TheoreticalAnalysisofImage-to-ImageTranslationwithAdversarialLearningXudongPan1MiZhang1DaizongDing1Abstractagenerator(i.e.anadaptivemodelthatmapsagaussiannoisetoafakesample)andadiscriminator(i.e.an...
AnalyzingUncertaintyinNeuralMachineTranslationMyleOtt1MichaelAuli1DavidGrangier1Marc’AurelioRanzato1Abstractpublicbenchmarks.However,severalopenquestionsre-main(Koehn&Knowles,2017).Inthiswork,wean...