This is the repository of the course project implement of a summer course called " Deep Learning" taught by Hehe Fan in ZJU.
We implement four methods in ./train , original GAN, high dimension GAN, diffusion GAN in Diffusion-GAN: Training GANs with Diffusion, and our own DifGAN.
./models includes the models of discriminator and generator (with and without tilmestep as input).
./result should record the generated images and the model parameters(.pth) after training.
There is no dataset because we use Mnist in official library.