Skip to Main Content
Table 3: 
Adding auxiliary OpenSubtitles MT data to the training. The two-stage models benefit much more strongly than the direct model, with our proposed model yielding the strongest overall results.
ModelFisherFisher+OpenSub
Cascade 32.45 34.58 (+6.2% rel.) 
Direct model 35.30 36.45 (+3.2% rel.) 
Basic two-stage 34.36 36.91 (+6.9% rel.) 
Best APM 36.70 38.81 (+5.4% rel.) 
ModelFisherFisher+OpenSub
Cascade 32.45 34.58 (+6.2% rel.) 
Direct model 35.30 36.45 (+3.2% rel.) 
Basic two-stage 34.36 36.91 (+6.9% rel.) 
Best APM 36.70 38.81 (+5.4% rel.) 
Close Modal

or Create an Account

Close Modal
Close Modal