Vincent Italiano

Hi, I'm trying to modify the loss by adding a confidence weight for each pixel. I don't understand why do you multiply translated mask: ``` mask_x = torch.mul(mask[:, :, 1:],...