two-timescale update rule (TTUR)

heykeetae / Self-Attention-GAN

Pytorch implementation of Self-Attention Generative Adversarial Networks (SAGAN)

2.53k stars 475 forks source link

two-timescale update rule (TTUR) #4

Closed taki0112 closed 6 years ago

taki0112 commented 6 years ago

Could you tell me where in the code is TTUR ?

Thank you

heykeetae commented 6 years ago

To my knowledge, TTUR refers to setting different learning rate for the generator and the discriminator, g_lr=0.0001 and d_lr=0.0004 in the code respectively, as guided by the paper.

zhangqianhui commented 6 years ago

very simple