clipping values, and replacing transposes with permute

Without clipping, any values larger than 255 will be replaced with
int(v mod 256) by byte(), which results in high freq noise in image.

torchvision/utils.py

@@ -90,6 +90,6 @@ def save_image(tensor, filename, nrow=8, padding=2,
     tensor = tensor.cpu()
     grid = make_grid(tensor, nrow=nrow, padding=padding,
                      normalize=normalize, range=range, scale_each=scale_each)
-    ndarr = grid.mul(255).byte().transpose(0, 2).transpose(0, 1).numpy()
+    ndarr = grid.mul(255).clip(0, 255).byte().permute(1, 2, 0).numpy()
     im = Image.fromarray(ndarr)
     im.save(filename)