The implementation details in the code.

google-research / xmcgan_image_generation

97 stars 15 forks source link

The implementation details in the code. #2

Closed SUNJIMENG closed 3 years ago

SUNJIMENG commented 3 years ago

Excuse me, line 216 in xmc_net.py, "x = dense_fn(self.gf_dim 16 4 * 4)(z)", which input the noise "z" into the linear layer. However, Table 7 in supp shows that the input is the concatenation of noise and reshaped condition. I wonder which one is right.

kohjingyu commented 3 years ago

Thanks for catching this. In this case, the code is correct - the input to the linear layer is just z. However, I believe there won't be a significant effect on results either way.