Code to show the simulation results in ARM:Augment-REINFORCE-merge gradient for discrete latent variable models

The MNIST data is self-contained and the Omniglot data is in the repository.

Below are the paper to cite if you find the algorithms in this repository useful in your own research:

@article{
  yin2018arm,
  title={ARM: Augment-REINFORCE-Merge Gradient for Discrete Latent Variable Models},
  author={Yin, Mingzhang and Zhou, Mingyuan},
  journal={arXiv preprint arXiv:1807.11143},
  year={2018}
}

This code is offered under the MIT License.