Vcae

Contents of this repo are as follows: . ├── README (this file) ├── data.py (downloading and data transformations) ├── layers.py (convolutional layers for AE) ├── main.py (starting point for the experiments) ├── metric.py (functions for calcucating multiple scores, based on https://arxiv.org/pdf/1806.07755.pdf) ├── model.py (stores all generative models ae_vine, vae, cvae, dec_vine and their dcgan versions, as dcgan itself) ├── train.py (most of the logics/execution for the experiment is done here) ├── utils.py (loading datasets, saving models)

A simple demo for VCAE is available by just running the main.py script. Note that this code requires Python 3.6, pytorch 4.0 and R 3.6 installed on your machine.

Run a demo for VCAE: python main.py --dataset 'mnist' --model 'ae_vine' --results-dir 'results_vcae/'