Improving Disentangled Representatoin Learning with the Beta Bernoulli Process [link].
The most of the contents in this code base are copied from this repository, so credit goes to them!
Use the environment.yml file provided in this repo to set the conda environment to run the code.
conda env create -f environment.yml
Finally, to run the code (e.g., IBP-VAE for MNIST with beta = 5):
python IBP-VAE-MNIST.py --beta 5
The most important part in this code is the compilation in order to run the code with GPU support. For this you need to navigate to lgamma and run ./make.sh. You might need to change architecture setting inside make.sh according to your GPU card's compute capability and the CUDA_PATH might also need to be customized. Please refer to the original code base for IBP-VAE [link] if you have any troble compiling or contact me at [email protected].