Skip to Main Content
Table 1: 
Hyperparameter search space for the NMT systems.
datasetbpe (1k)#layers#embed#hidden#att_headsinit_lr (10−4)
zh, ru, ja, en 10, 30, 50 2, 4 256, 512, 1024 1024, 2048 8, 16 3, 6, 10 
sw 1, 2, 4, 8, 16, 32 1, 2, 4, 6 256, 512, 1024 1024, 2048 8, 16 3, 6, 10 
so 1, 2, 4, 8, 16, 32 1, 2, 4 256, 512, 1024 1024, 2048 8, 16 3, 6, 10 
datasetbpe (1k)#layers#embed#hidden#att_headsinit_lr (10−4)
zh, ru, ja, en 10, 30, 50 2, 4 256, 512, 1024 1024, 2048 8, 16 3, 6, 10 
sw 1, 2, 4, 8, 16, 32 1, 2, 4, 6 256, 512, 1024 1024, 2048 8, 16 3, 6, 10 
so 1, 2, 4, 8, 16, 32 1, 2, 4 256, 512, 1024 1024, 2048 8, 16 3, 6, 10 
Close Modal

or Create an Account

Close Modal
Close Modal