1 paper · 1 filter
Kazuki Takamura, Satoshi Yamane
In order to obtain a model which can process sequential data related to machine translation and speech recognition faster and more accurately, we propose adopting Chrono Initialize…