Two time Scale update is not used in this train script

[mostafaelaraby]

The two time scale update rule as mentioned in https://arxiv.org/pdf/1706.08500.pdf is about giving higher learning rate to the discriminator and instead of training the discriminator for 5 steps then updating the generator they are both trained at the same number of iters but with different learning rate ,but in your code you added a comment stating that you used this technique however both optimizers are having the same learning rate and the generator is updated once every 5 steps of discriminator

WaveGAN-pytorch/train.py

Line 50 in 3c99f1d

# "Two time-scale update rule"(TTUR) to update netD 4x faster than netG.