Implementation of stick breaking VAE with pytorch
Different from the origin paper, q(z | x) follows Beta distribution, since the pahtwise gradient for Beta distribution is available in torch.distributions.Beta.
TODO:
- Parameter tuning to reproduce the paper results