Starting from 2017, we are using two newly developed homeworks for students to get familiar with the knowledge they learned in lectures on neural networks and neural machine translation. This repository contains the guidelines and the start code for these homeworks.
HW4 requires students to implement and train a bi-directional RNN language model to solve a pedagogical problem in language learning. This homework is designed to bring students with no prior deep learning experiences up to speed and be able to use deep learning frameworks to implement and test their ideas.
HW5 requires students to implement and train a neural machine translation model. This homeworks is designed to help students understand most of the bells and whistles of modern attention-based neural machine translation architecture.
As of 2017, all the starter code are written in PyTorch. We welcome contribution of starter code written in other deep learning frameworks.