An overview of the work done in this repository can be seen in the following presentation : presentation
The most important parts to understand this work and code are :
- Metropolis-Hastings algorithm - The metropolis-hastings implementation is in metropolis_hastings_hybrid_kernels.py
- Inference on VAE - The inference done using an already trained VAE model is in mcmc_on_vae.py
- VAE - The VAE model is defined in network.py using the PyTorch library.
- Trainning - The trainning was done using the script train.py