GAN Experiments These are experiments of trying to train GANS from scratch. Each notebook has all the code needed to train GAN. There are 4 notebooks that differ in techniques used for training. Below are results. Vanilla Minibatch Discrimination Augmenting images Multiscale gradient + pixelwise norm + minibatch std + equalized learning + relativistic hinge loss DCGAN DCGAN + Spectral Norm