CS 236 Entropy Regularized Conditional GANs
Example usage
Running conditional gan (train_cond.py) with regularization term 0.0002
$ python main.py --dataset cifar10 --model presgan --lambda 0.0002
Files: hmc.py -> HMC Sampling, train_cond.py -> Main conditional gan training with nets_cond.py architecture.
Entropy on MNIST
Entropy Saturation on CelebA (0.0002)
Entropy Saturation on Cifar-10 0.0002)
Modulated Lambda on Entropy Saturation on Cifar 10 - 0.0002
