β-VAE

Pytorch implementation of β-VAE proposed in this paper

Dependencies

python 3.6.4
pytorch 0.3.1.post2
visdom

Datasets

same with here

Usage

initialize visdom

python -m visdom.server

you can run codes using sh files

e.g.
sh run_celeba_H.sh
sh run_3dchairs_H.sh
sh run_dsprites_B.sh

or you can run your own experiments by setting parameters manually

e.g.
python main.py --beta 4 --lr 1e-4 --z_dim 32 ...

check training process on the visdom server

localhost:8097

Results

3D Chairs latent space traverse. z was sampled from N(0,I)

sh run_celeba_H.sh

CelebA latent space traverse. z was sampled from N(0,I)

sh run_3dchairs_H.sh

Reference

1. β-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework, Higgins et al., ICLR, 2017
2. Understanding disentangling in β-VAE, Burgess et al., arxiv:1804.03599, 2018
3. Github Repo: Tensorflow implementation