This repository contains the code and various resources used to complete our
final project for LING 380. Please note that the repositiory contains some large
files (containing training data), and as such, git lfs must be installed
(see https://git-lfs.github.com/).
The paper directory contains the various tex files and images used to
generate the final writeup.
The src directory contains the code used tot train and analyse the model. To
train a new model, run python train.py, and for more information on how to
run, apply the -h command line option. The file main.py contains various
utility functions that were used in anlysing data; it does not do anything when
run in isolation. Dependencies include: pytorch, pickle, pandas,
matplotlib, seaborn, scipy, and pingouin.
The src/data directory contains the datasets used for training the model.
The experiment_data directory contains the test items used for experiments
outlined in the paper.