Reproducing Neural Discrete Representation Learning

Course Project for IFT 6135 - Representation Learning

Project Report link: final_project.pdf

Download the dataset

cd data/miniimagenet
gdown --id 1pQK7CDStL4Pvzf4AlMNWcYcwS0D-3pJa
unzip mini.zip
rm mini.zip
cd ../..

Download the miniimagenet vqvae model k=64

gdown --id 1UGlBPd7U5nBloHDbYRMtbH2x5Zf2FjVa

Check the results of this trained vqvae

See loadvqvaek64.ipynb

Run it yourself

https://colab.research.google.com/drive/1BH2RK088d5-w-H4oSrs4t5zJwLxctRXV?usp=sharing

Instructions

1. To train the VQVAE with default arguments as discussed in the report, execute:

python vqvae.py --data-folder /tmp/miniimagenet --output-folder models/vqvae

2. To train the PixelCNN prior on the latents, execute:

python pixelcnn_prior.py --data-folder /tmp/miniimagenet --model models/vqvae --output-folder models/pixelcnn_prior