A Neural Machine Translation Model for Arabic Dialects That Utilizes Multitask Learning (MTL)

Table 1

Multitask neural translation vs single model.

Model	Pairs	Embedding Size	Hidden Size	Epochs	Blue
Single NMT	LA-MSA	150	170	120	0.17
Multitask	LA-MSA	150	170	170	0.41
Single NMT	MA-MSA	180	180	120	0.16
Multitask	MA-MSA	180	180	230	0.30
Single NMT	MSA-EN	160	160	120	0.10
Multitask	MSA-EN	150	170	170	0.27