search

akanimax / pro_gan_pytorch

Unofficial PyTorch implementation of the paper titled "Progressive growing of GANs for improved Quality, Stability, and Variation"
MIT License
536 stars 100 forks source link

Calculating fan_in for EqualizedConvTranspose2d #52

Closed srihari-humbarwadi closed 3 years ago

srihari-humbarwadi commented 3 years ago

Is there a particular reason for using fan_in = self.in_channels for EqualizedConvTranspose2d in https://github.com/akanimax/pro_gan_pytorch/blob/e3d2341333500b4792d60b42af3719eb524a52ec/pro_gan_pytorch/custom_layers.py#L110

compared to fan_in = np.prod(self.kernel_size) * self.in_channels for EqualizedConv2d in https://github.com/akanimax/pro_gan_pytorch/blob/e3d2341333500b4792d60b42af3719eb524a52ec/pro_gan_pytorch/custom_layers.py#L62

srihari-humbarwadi commented 3 years ago

Answered here https://github.com/akanimax/pro_gan_pytorch/issues/44

akanimax commented 3 years ago

Glad to know that the issue was resolved. Please feel free to reopen if you still have any problem.