Generative modeling tries to recreate images from the training set either recreating the original images or creating totally new images. To run the generative_modeling.py, move all your images to Google drive in a directory. Mount the google drive to the Google colab environment. You can upload the dataset as a zip file and unzip in the drive. Install all the libraries imported and then run each block of code in every cell. 

In vae_classification, we compare generative classification ability of different representation learning models and compare there results. There are two methods to run vae_classification.py. You can use the dataset available on kaggle directly onto the google colab environment. To do this, create a kaggle account, get API and credential to access dataset, find the chest x-ray dataset and install the dataset onto your colab environment using kaggle API. Or, you can directly use the dataset on the drive. You need to separately keep disease and non-disease images on two folders. Then, import all the libraries of the code, install if needed and then run each block of code into the google colab cell. 

Dataset: https://nihcc.app.box.com/v/ChestXray-NIHCC