a0f77cc8b87512e691a7c22b6fb9ba60f603b632,network/convolutional_variational_autoencoder.py,VAE_convolutional,layer_op,#VAE_convolutional#Any#Any#,138

Before Change

        tf.add_to_collection("NiftyNetCollectionConsole", KL)

        LL = tf.summary.scalar("log_likelihood", log_likelihood)
        tf.add_to_collection("NiftyNetCollectionConsole", LL)

        return [posterior_means, posterior_logvariances, flow_means, data_logvariances,
                images, data_variances, posterior_variances]

After Change

                name="decoder_trans_conv_means_{}_{}".format(self.trans_conv_kernels_sizes[i], self.trans_conv_features[i])))
            print(decoders_means_cnn[-1])

        for i in range(0, len(self.trans_conv_features)):
            if self.upsampling_mode == "DECONV":
                decoders_logvariances_upsamplers.append(DeconvolutionalLayer(
                    n_output_chns=self.trans_conv_features[i],
                    kernel_size=2,
                    stride=2,
                    padding="SAME",
                    with_bias=True,
                    with_bn=True,
                    w_initializer=self.initializers["w"],
                    w_regularizer=None,
                    acti_func="identity",
                    name="decoder_upsampler_variances_{}_{}".format(2, 2)))
                print(decoders_logvariances_upsamplers[-1])

            decoders_logvariances_cnn.append(DeconvolutionalLayer(
                n_output_chns=self.trans_conv_features[i],
                kernel_size=self.trans_conv_kernels_sizes[i],
                stride=1,
                padding="SAME",
                with_bias=True,
                with_bn=not (i == len(self.trans_conv_features) - 1), // No BN on output
                w_initializer=self.initializers["w"],
                w_regularizer=None,
                acti_func=self.acti_func_trans_conv_logvariances[i],
                name="decoder_trans_conv_variances_{}_{}".format(self.trans_conv_kernels_sizes[i],
                                                                  self.trans_conv_features[i])))
            print(decoders_logvariances_cnn[-1])

        // Convolutional encoder layers
        flow = images
        for i in range(0, len(self.conv_features)):
            flow = encoders_cnn[i](flow, is_training)
            flow = encoders_downsamplers[i](flow, is_training)

Instances