Machine Translation On Wmt2016 English 1

Metrics

BLEU score

Results

Performance results of various models on this benchmark

Model Name	BLEU score	Paper Title	Repository
PBSMT + NMT	25.13	Phrase-Based & Neural Unsupervised Machine Translation	-
Denoising autoencoders (non-autoregressive)	29.66	Deterministic Non-Autoregressive Neural Sequence Modeling by Iterative Refinement	-
ConvS2S BPE40k	29.9	Convolutional Sequence to Sequence Learning	-
CMLM+LAT+4 iterations	32.87	Incorporating a Local Translation Mechanism into Non-autoregressive Translation	-
NAT +FT + NPD	29.79	Non-Autoregressive Neural Machine Translation	-
DeLighT	34.7	DeLighT: Deep and Light-weight Transformer	-
GRU BPE90k	28.9	-	-
Unsupervised PBSMT	21.33	Phrase-Based & Neural Unsupervised Machine Translation	-
CMLM+LAT+1 iterations	30.74	Incorporating a Local Translation Mechanism into Non-autoregressive Translation	-
Deep Convolutional Encoder; single-layer decoder	27.8	A Convolutional Encoder Model for Neural Machine Translation	-
FlowSeq-large (NPD n=15)	31.97	FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow	-
FLAN 137B (few-shot, k=9)	20.5	Finetuned Language Models Are Zero-Shot Learners	-
BiLSTM	27.5	A Convolutional Encoder Model for Neural Machine Translation	-
BART (TextBox 2.0)	-	TextBox 2.0: A Text Generation Library with Pre-trained Language Models	-
BiGRU	28.1	Edinburgh Neural Machine Translation Systems for WMT 16	-
FlowSeq-base	29.26	FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow	-
FlowSeq-large	29.86	FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow	-
FLAN 137B (zero-shot)	18.9	Finetuned Language Models Are Zero-Shot Learners	-
Unsupervised NMT + Transformer	21.18	Phrase-Based & Neural Unsupervised Machine Translation	-
FlowSeq-large (NPD n = 30)	32.35	FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow	-

0 of 21 row(s) selected.