FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow

Benchmark Model Rank Results
machine-translation-on-iwslt2015-germanFlowSeq-base#12BLEU score: 24.75
machine-translation-on-wmt2014-english-germanFlowSeq-large (NPD n = 30)#64BLEU score: 23.64
machine-translation-on-wmt2014-english-germanFlowSeq-large (NPD n = 15)#65BLEU score: 23.14
machine-translation-on-wmt2014-english-germanFlowSeq-large (IWD n = 15)#66BLEU score: 22.94
machine-translation-on-wmt2014-english-germanFlowSeq-large#69BLEU score: 20.85
machine-translation-on-wmt2014-english-germanFlowSeq-base#73BLEU score: 18.55
machine-translation-on-wmt2014-german-englishFlowSeq-large (NPD n = 30)#9BLEU score: 28.29
machine-translation-on-wmt2014-german-englishFlowSeq-large (NPD n = 15)#10BLEU score: 27.71
machine-translation-on-wmt2014-german-englishFlowSeq-large (IWD n=15)#11BLEU score: 27.16
machine-translation-on-wmt2014-german-englishFlowSeq-large#13BLEU score: 25.4
machine-translation-on-wmt2014-german-englishFlowSeq-base#14BLEU score: 23.36
machine-translation-on-wmt2016-english-1FlowSeq-large (NPD n = 30)#3BLEU score: 32.35
machine-translation-on-wmt2016-english-1FlowSeq-large (NPD n=15)#4BLEU score: 31.97
machine-translation-on-wmt2016-english-1FlowSeq-large (IWD n = 15)#5BLEU score: 31.08
machine-translation-on-wmt2016-english-1FlowSeq-large#8BLEU score: 29.86
machine-translation-on-wmt2016-english-1FlowSeq-base#11BLEU score: 29.26
machine-translation-on-wmt2016-romanianFlowSeq-large (NPD n = 30)#10BLEU score: 32.91
machine-translation-on-wmt2016-romanianFlowSeq-large (NPD n = 15)#12BLEU score: 32.46
machine-translation-on-wmt2016-romanianFlowSeq-large (IWD n = 15)#13BLEU score: 32.03
machine-translation-on-wmt2016-romanianFlowSeq-large#16BLEU score: 30.69
machine-translation-on-wmt2016-romanianFlowSeq-base#18BLEU score: 30.16