c8a9b2b2f14226567f2e1b42e7cf007b38c0fba3,tests/classifiers/test_pytorch.py,TestPyTorchClassifier,test_class_gradient,#TestPyTorchClassifier#,143

Before Change

        grads = ptc.class_gradient(x_test, label=label)

        self.assertTrue(np.array(grads.shape == (NB_TEST, 1, 1, 28, 28)).all())
        self.assertNotEqual(np.sum(grads), 0)

    def test_class_gradient_target(self):
        // Get MNIST
        (_, _), (x_test, _) = self.mnist

After Change

        self.assertEqual(gradients.shape, (NB_TEST, 10, 1, 28, 28))

        expected_gradients_1 = np.asarray([-1.5104107e-04, -1.4420391e-04, -8.5643369e-05, 4.2904957e-04,
                                           4.6917787e-04, -3.3433505e-05, 4.3190207e-04, 5.4948201e-04,
                                           9.0309686e-04, -1.3793766e-04, 6.4290554e-04, -2.8910500e-04,
                                           -8.5047574e-04, -4.8866379e-04, 1.3007881e-03, 1.5724849e-04,
                                           6.1079778e-04, -5.2589108e-05, -6.7920942e-04, -2.2616469e-04,
                                           4.6133125e-04, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00,
                                           0.0000000e+00, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00])
        np.testing.assert_array_almost_equal(gradients[0, 5, 0, :, 14], expected_gradients_1, decimal=4)

        expected_gradients_2 = np.asarray([-4.7446613e-04, -9.8460718e-05, -1.1919734e-04, -2.2912446e-04,
                                           2.8542569e-04, -8.9842360e-06, 1.7257492e-04, 0.0000000e+00,
                                           0.0000000e+00, -7.2422711e-04, -9.8898512e-05, 7.9051330e-04,
                                           -3.9680302e-04, 7.3898572e-04, 1.3007881e-03, 7.1805675e-04,
                                           7.7588821e-04, 9.9411258e-04, 4.4845918e-04, -1.9343558e-04,
                                           -1.8357937e-04, 5.9276586e-04, -6.4391940e-04, 0.0000000e+00,
                                           0.0000000e+00, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00])
        np.testing.assert_array_almost_equal(gradients[0, 5, 0, 14, :], expected_gradients_2, decimal=4)

        // Test 1 gradient label = 5
        gradients = classifier.class_gradient(self.x_test, label=5)

        self.assertEqual(gradients.shape, (NB_TEST, 1, 1, 28, 28))

        expected_gradients_1 = np.asarray([-1.5104107e-04, -1.4420391e-04, -8.5643369e-05, 4.2904957e-04,
                                           4.6917787e-04, -3.3433505e-05, 4.3190207e-04, 5.4948201e-04,
                                           9.0309686e-04, -1.3793766e-04, 6.4290554e-04, -2.8910500e-04,
                                           -8.5047574e-04, -4.8866379e-04, 1.3007881e-03, 1.5724849e-04,
                                           6.1079778e-04, -5.2589108e-05, -6.7920942e-04, -2.2616469e-04,
                                           4.6133125e-04, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00,
                                           0.0000000e+00, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00])
        np.testing.assert_array_almost_equal(gradients[0, 0, 0, :, 14], expected_gradients_1, decimal=4)

        expected_gradients_2 = np.asarray([-4.7446613e-04, -9.8460718e-05, -1.1919734e-04, -2.2912446e-04,
                                           2.8542569e-04, -8.9842360e-06, 1.7257492e-04, 0.0000000e+00,
                                           0.0000000e+00, -7.2422711e-04, -9.8898512e-05, 7.9051330e-04,
                                           -3.9680302e-04, 7.3898572e-04, 1.3007881e-03, 7.1805675e-04,
                                           7.7588821e-04, 9.9411258e-04, 4.4845918e-04, -1.9343558e-04,
                                           -1.8357937e-04, 5.9276586e-04, -6.4391940e-04, 0.0000000e+00,
                                           0.0000000e+00, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00])
        np.testing.assert_array_almost_equal(gradients[0, 0, 0, 14, :], expected_gradients_2, decimal=4)

        // Test a set of gradients label = array
        label = np.random.randint(5, size=NB_TEST)
        gradients = classifier.class_gradient(self.x_test, label=label)

        self.assertEqual(gradients.shape, (NB_TEST, 1, 1, 28, 28))

        expected_gradients_1 = np.asarray([-2.39315428e-04, -2.28482357e-04, -2.39842790e-04, -6.74667899e-05,
                                           5.54567552e-04, 5.59428358e-04, 3.71058501e-04, 4.47539205e-04,
                                           5.91437332e-04, -1.86359655e-04, 2.78284366e-04, 1.86634657e-04,
                                           -3.62139835e-05, 1.46655992e-04, 7.07449333e-04, 1.51086148e-04,
                                           1.42195524e-04, 5.60022090e-06, -3.43588385e-04, -1.13900256e-04,
                                           2.32334001e-04, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
                                           0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00])
        np.testing.assert_array_almost_equal(gradients[0, 0, 0, :, 14], expected_gradients_1, decimal=4)

        expected_gradients_2 = np.asarray([-3.3879644e-04, 1.1766610e-04, 4.3186071e-04, 3.2926403e-04,
                                           8.9179150e-05, 4.6644508e-04, 5.1162497e-04, 0.0000000e+00,
                                           0.0000000e+00, -3.7474511e-04, -4.9881939e-05, 4.0110818e-04,
                                           -1.9942035e-04, 3.7868350e-04, 7.0744933e-04, 5.2536512e-04,
                                           1.0788649e-04, -3.9121151e-06, 2.3214625e-04, 1.5785350e-05,
                                           1.0586554e-05, -2.1646731e-04, 8.7655055e-05, 0.0000000e+00,
                                           0.0000000e+00, 0.0000000e+00, 0.0000000e+00, 0.0000000e+00])
        np.testing.assert_array_almost_equal(gradients[0, 0, 0, 14, :], expected_gradients_2, decimal=4)

    def test_class_gradient_target(self):
        classifier = get_classifier_pt()

Instances