Transformer networks for Neural Machine Translation

This is an implementation of the transformer for the paper

"Attention is all you need"

It also contains the implementation of the Speech Transformer as in

"Very Deep Self-Attention Networks for End-to-End Speech Recognition"

---

Requirement: (mostly recommended to use with Anaconda3 - Python3.7)

PyTorch (1.0.1 or 1.1). The code base is normally kept up-to-date with the latest version of PyTorch.

hdf5

apex (the C++ and CUDA extensions are not required) https://github.com/nvidia/apex

Recipe for the Speech model coming soon.

---