Transformer is one of deep learning model that exhibits high performance in natural language processing tasks such as translation, document generation, and question answering.

This is a project that involves building an English to Japanese translator using PyTorch, with the neural network architecture based on the Transformer model.

The implementation of the layers for the Transformer model (e.g. ScaledDotProductAttention, PositionalEncoding... etc) is done from scratch using Python/PyTorch.

The dataset used for training is a parallel corpus called kftt.