Fix projection for class conditioned networks

[namangup]

Currently, the code produces an error for networks trained with class conditioning. Fixed it by allowing addition of an explicit label.
File "projector.py", line 197, in run_projection verbose=True File "projector.py", line 60, in project w_samples = G.mapping(torch.from_numpy(z_samples).to(device), label) # [N, L, C] File "/home/ubuntu/anaconda3/envs/pytorch_latest_p37/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "<string>", line 222, in forward File "/home/ubuntu/misc/stylegan2-ada-pytorch/torch_utils/misc.py", line 81, in assert_shape if tensor.ndim != len(ref_shape): AttributeError: 'NoneType' object has no attribute 'ndim'

[yixin1024]

Hi,I change the projector.py . But it also has this question. Do you solve it? thank you!

[namangup]

Hi,I change the projector.py . But it also has this question. Do you solve it? thank you!

Yes, with the changes it should work. Make sure to use the new options --label and --label-dim