Table 5: 
Ablation study for machine translation on WMT16 Ro-En. Results of Stern et al. (2019) are based on greedy decoding with the EOS penalty.
Model Variantsdevtest
Baseline L2R 32.53 31.82 
SAO default 33.60 32.47 
 no bootstrap 32.86 31.88 
 no bootstrap, no noise 32.64 31.72 
 bootstrap from R2L order 33.12 32.02 
 bootstrap from SYN order 33.09 31.93 
Stern et al. (2019) - Uniform 29.99 28.52 
Stern et al. (2019) - Binary 32.27 30.66 
Model Variantsdevtest
Baseline L2R 32.53 31.82 
SAO default 33.60 32.47 
 no bootstrap 32.86 31.88 
 no bootstrap, no noise 32.64 31.72 
 bootstrap from R2L order 33.12 32.02 
 bootstrap from SYN order 33.09 31.93 
Stern et al. (2019) - Uniform 29.99 28.52 
Stern et al. (2019) - Binary 32.27 30.66 
Close Modal

or Create an Account

Close Modal
Close Modal