research
∙
03/20/2022
Small Batch Sizes Improve Training of Low-Resource Neural MT
We study the role of an essential hyper-parameter that governs the train...
research
∙
06/13/2019