863a9dcee691e850e79d97a16abb977e24324fa1,tests/convolution_test.py,ConvTest,test_conv_3d_bias_reg_shape,#ConvTest#,42

Before Change

            self.assertAllClose((2, 16, 16, 16, 10), out_3d.shape)

    def test_conv_3d_bias_reg_shape(self):
        x_3d = self.get_3d_input()
        conv_reg = ConvLayer(10, 3, 2,
                             w_regularizer=regularizers.l2_regularizer(0.5),
                             with_bias=True,
                             b_regularizer=regularizers.l2_regularizer(0.5))
        conv_reg_out = conv_reg(x_3d)
        print(conv_reg)
        with self.test_session() as sess:
            sess.run(tf.global_variables_initializer())
            out_3d = sess.run(conv_reg_out)
            self.assertAllClose((2, 8, 8, 8, 10), out_3d.shape)

    def test_3d_convlayer_default_shape(self):
        x_3d = self.get_3d_input()
        conv_3d = ConvolutionalLayer(10, 3, 1)
        conv_3d_out = conv_3d(x_3d, is_training=True)

After Change

                                     output_shape=(2, 16, 16, 8, 10))

    def test_conv_3d_bias_reg_shape(self):
        input_param = {"n_output_chns": 10,
                       "kernel_size": [3, 1, 3],
                       "stride": [2, 2, 2],
                       "with_bias": True,
                       "w_regularizer": regularizers.l2_regularizer(0.5),
                       "b_regularizer": regularizers.l2_regularizer(0.5)}
        self._test_conv_output_shape(rank=3,
                                     param_dict=input_param,
                                     output_shape=(2, 8, 8, 8, 10))

    def test_3d_convlayer_default_shape(self):
        input_param = {"n_output_chns": 10,
                       "kernel_size": 3,

Instances