shape not matched

[LeeKyungwook]

Thanks for Great code.

I found some errors. After inference network, shape is not matched with fc layer in loss. In loss func, fc layer need 1024 in features, but shape is not 1024. Can you check it?? Thx

[TangDL]

just change it to 512

[swhoosh]

@TangDL I also encountered this problem. Can you please tell me where can I change the feature size? I am very new to pytorch and NN. Thanks

[wlz987]

Thanks for Great code.

I found some errors. After inference network, shape is not matched with fc layer in loss. In loss func, fc layer need 1024 in features, but shape is not 1024. Can you check it?? Thx
loss.py里面，把1024改成512
class PatchLoss(nn.Module):

def __init__(self, alpha1=1.0, alpha2=1.0):
    super().__init__()
    self.alpha1 = alpha1
    self.alpha2 = alpha2
    self.sim_loss = SimilarityLoss()
    self.amsm_loss = AdMSoftmaxLoss(512, 2)