using OpenNMT-tf (Klein et al., 2017). Using the model, we then calculate the average per-form neg- ative log likelihood ( ) of held out ...
確定! 回上一頁