Swagan support for sample generation, latent space projection, closed form factorization

[JadAssaf]

If you've trained your model using swagan, you might face issues with model weights incompatibilities:

RuntimeError: Error(s) in loading state_dict for Generator:
size mismatch for to_rgb1.bias: copying a param with shape torch.Size([1, 12, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 1, 1]).
size mismatch for to_rgb1.conv.weight: copying a param with shape torch.Size([1, 12, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 512, 1, 1]).
size mismatch for to_rgbs.0.bias: copying a param with shape torch.Size([1, 12, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 1, 1]).
size mismatch for to_rgbs.0.conv.weight: copying a param with shape torch.Size([1, 12, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 512, 1, 1]).
size mismatch for to_rgbs.1.bias: copying a param with shape torch.Size([1, 12, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 1, 1]).
size mismatch for to_rgbs.1.conv.weight: copying a param with shape torch.Size([1, 12, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 512, 1, 1]).
size mismatch for to_rgbs.2.bias: copying a param with shape torch.Size([1, 12, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 1, 1]).
size mismatch for to_rgbs.2.conv.weight: copying a param with shape torch.Size([1, 12, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 512, 1, 1]).
size mismatch for to_rgbs.3.bias: copying a param with shape torch.Size([1, 12, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 1, 1]).
size mismatch for to_rgbs.3.conv.weight: copying a param with shape torch.Size([1, 12, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 512, 1, 1]).
size mismatch for to_rgbs.4.bias: copying a param with shape torch.Size([1, 12, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 1, 1]).
size mismatch for to_rgbs.4.conv.weight: copying a param with shape torch.Size([1, 12, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1, 3, 256, 1, 1]).

To fix this, in generator.py, change "from model import Generator" to "from swagan import Generator".
Same applies for other scripts.