ALTA 2021 - Tutorial Code

^1,2Jacob Parnell, ^1,2Inigo Jauregi Unanue and ¹Prof. Massimo Piccardi

¹University of Technology Sydney (UTS)

²RoZetta Technology

Introduction

This is the code used in the tutorial at the 19th Annual Workshop of the Australasian Language Technology Association (ALTA 2021). You can find the full recording of the tutorial here.

In this code repository we show how to train neural machine translation and abstractive summarisation models using PyTorch Lightning, a framework meant to remove the "boilerplate" from the development of research models.

Getting Started

We provide two notebooks (one for translation and one for summarisation) as part of the tutorial. You can run this on Jupyter notebooks or Google-colab.

Translation

Check the translation folder to start with this section of the tutorial.

Summarisation

Check the summarisation folder to start with this section of the tutorial.

Shared Materials

The NMT and summarisation slide sets and accompanying videos are available here.