def test()

in train.py [0:0]
78 lines of code
8 McCabe index (conditional complexity)

def test(target_vars, saver, sess, logger, dataloader):
    X_NOISE = target_vars['X_NOISE']
    X = target_vars['X']
    Y = target_vars['Y']
    LABEL = target_vars['LABEL']
    energy_start = target_vars['energy_start']
    x_mod = target_vars['x_mod']
    x_mod = target_vars['test_x_mod']
    energy_neg = target_vars['energy_neg']

    np.random.seed(1)
    random.seed(1)

    output = [x_mod, energy_start, energy_neg]

    dataloader_iterator = iter(dataloader)
    data_corrupt, data, label = next(dataloader_iterator)
    data_corrupt, data, label = data_corrupt.numpy(), data.numpy(), label.numpy()

    orig_im = try_im = data_corrupt

    if FLAGS.cclass:
        try_im, energy_orig, energy = sess.run(
            output, {X_NOISE: orig_im, Y: label[0:1], LABEL: label})
    else:
        try_im, energy_orig, energy = sess.run(
            output, {X_NOISE: orig_im, Y: label[0:1]})

    orig_im = rescale_im(orig_im)
    try_im = rescale_im(try_im)
    actual_im = rescale_im(data)

    for i, (im, energy_i, t_im, energy, label_i, actual_im_i) in enumerate(
            zip(orig_im, energy_orig, try_im, energy, label, actual_im)):
        label_i = np.array(label_i)

        shape = im.shape[1:]
        new_im = np.zeros((shape[0], shape[1] * 3, *shape[2:]))
        size = shape[1]
        new_im[:, :size] = im
        new_im[:, size:2 * size] = t_im

        if FLAGS.cclass:
            label_i = np.where(label_i == 1)[0][0]
            if FLAGS.dataset == 'cifar10':
                log_image(new_im, logger, '{}_{:.4f}_now_{:.4f}_{}'.format(
                    i, energy_i[0], energy[0], cifar10_map[label_i]), step=i)
            else:
                log_image(
                    new_im,
                    logger,
                    '{}_{:.4f}_now_{:.4f}_{}'.format(
                        i,
                        energy_i[0],
                        energy[0],
                        label_i),
                    step=i)
        else:
            log_image(
                new_im,
                logger,
                '{}_{:.4f}_now_{:.4f}'.format(
                    i,
                    energy_i[0],
                    energy[0]),
                step=i)

    test_ims = list(try_im)
    real_ims = list(actual_im)

    for i in tqdm(range(50000 // FLAGS.batch_size + 1)):
        try:
            data_corrupt, data, label = dataloader_iterator.next()
        except BaseException:
            dataloader_iterator = iter(dataloader)
            data_corrupt, data, label = dataloader_iterator.next()

        data_corrupt, data, label = data_corrupt.numpy(), data.numpy(), label.numpy()

        if FLAGS.cclass:
            try_im, energy_orig, energy = sess.run(
                output, {X_NOISE: data_corrupt, Y: label[0:1], LABEL: label})
        else:
            try_im, energy_orig, energy = sess.run(
                output, {X_NOISE: data_corrupt, Y: label[0:1]})

        try_im = rescale_im(try_im)
        real_im = rescale_im(data)

        test_ims.extend(list(try_im))
        real_ims.extend(list(real_im))

    score, std = get_inception_score(test_ims)
    print("Inception score of {} with std of {}".format(score, std))