You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have a question regarding a difference between your implementation and a recommendation from the repository "ganhacks" that you referred to. In their repository they state to use an embedding layer for labels, and add as additional channels to images when implementing the conditional gan. I have often seen this done by concatenating the embedding and the noise vectors. However, in your code you use a multiplication, as below;
label_embedding = self.embedding(label)
x = torch.mul(noise,label_embedding)
x = x.view(-1,100,1,1)
Is this the same as the concatenation and if not, what is the difference and would you prefer one over the other?
Thank you in advance!
The text was updated successfully, but these errors were encountered:
Hi, first of all, thanks for your amazing work!
I have a question regarding a difference between your implementation and a recommendation from the repository "ganhacks" that you referred to. In their repository they state to use an embedding layer for labels, and add as additional channels to images when implementing the conditional gan. I have often seen this done by concatenating the embedding and the noise vectors. However, in your code you use a multiplication, as below;
label_embedding = self.embedding(label)
x = torch.mul(noise,label_embedding)
x = x.view(-1,100,1,1)
Is this the same as the concatenation and if not, what is the difference and would you prefer one over the other?
Thank you in advance!
The text was updated successfully, but these errors were encountered: