This is a LSTM NMT model of Japanese-English translation using Chianer 1.24. The main idea is baesd on the attention model proposed in the paper: Neural machine translation by jointly learning to align and translate.
It adopted the 'global attention with dot product' introduced in the paper called Effective Approaches to Attention-based Neural Machine Translation
.
It adopted dropout introduced in the paper: RECURRENT NEURAL NETWORK
REGULARIZATION.
For mroe information about Chainer, please refer to the chainer documentation.
The requirement of this code:
miniconda + python 3+
seaborn + pandas + matplotlib
tqdm + chainer=1.24 + ipython