4fe443104940890fabe2ff7b4876584093464337,implementations/dualgan/dualgan.py,,,#,126

Before Change

            n_samples = 10
            // Concatenate samples by column
            real_A = torch.cat(imgs_A.data[:n_samples], -1)
            real_B = torch.cat(imgs_B.data[:n_samples], -1)
            fake_A = torch.cat(fake_A.data[:n_samples], -1)
            fake_B = torch.cat(fake_B.data[:n_samples], -1)
            recov_A = torch.cat(recov_A.data[:n_samples], -1)
            recov_B = torch.cat(recov_B.data[:n_samples], -1)

After Change

        BAB = torch.cat((imgs_B.data, fake_A.data, recov_B.data), -2)
        sample = torch.cat((ABA, BAB), -2)
        image_samples.append(sample)
        if len(image_samples) > 5:
            image_samples.pop(0)

        // Check sample interval => save sample if there

Instances