model.py 文件源码

def build_summary(self, name):
        # Distribution of generator activations
        tf.summary.histogram('generator/{}/f_deconv2_outputs'.format(name), self.net['f_deconv2_outputs'])
        tf.summary.histogram('generator/{}/f_deconv3_outputs'.format(name), self.net['f_deconv3_outputs'])
        tf.summary.histogram('generator/{}/f_deconv4_outputs'.format(name), self.net['f_deconv4_outputs'])
        tf.summary.histogram('generator/{}/f_deconv5i_outputs'.format(name), self.net['f_deconv5i_outputs'])
        tf.summary.histogram('generator/{}/f_deconv5m_outputs'.format(name), self.net['f_deconv5m_outputs'])
        tf.summary.histogram('generator/{}/b_deconv2_outputs'.format(name), self.net['b_deconv2_outputs'])
        tf.summary.histogram('generator/{}/b_deconv3_outputs'.format(name), self.net['b_deconv3_outputs'])
        tf.summary.histogram('generator/{}/b_deconv4_outputs'.format(name), self.net['b_deconv4_outputs'])
        tf.summary.histogram('generator/{}/b_deconv5_outputs'.format(name), self.net['b_deconv5_outputs'])

        # Generator weights, biases
        tf.summary.scalar('generator/{}/w2_f'.format(name), tf.norm(self.net['w2_f']))
        tf.summary.scalar('generator/{}/w3_f'.format(name), tf.norm(self.net['w3_f']))
        tf.summary.scalar('generator/{}/w4_f'.format(name), tf.norm(self.net['w4_f']))
        tf.summary.scalar('generator/{}/w5_fi'.format(name), tf.norm(self.net['w5_fi']))
        tf.summary.scalar('generator/{}/w5_fm'.format(name), tf.norm(self.net['w5_fm']))
        tf.summary.scalar('generator/{}/w2_b'.format(name), tf.norm(self.net['w2_b']))
        tf.summary.scalar('generator/{}/w3_b'.format(name), tf.norm(self.net['w3_b']))
        tf.summary.scalar('generator/{}/w4_b'.format(name), tf.norm(self.net['w4_b']))
        tf.summary.scalar('generator/{}/w5_b'.format(name), tf.norm(self.net['w5_b']))

        tf.summary.scalar('generator/{}/b2_f'.format(name), tf.norm(self.net['b2_f']))
        tf.summary.scalar('generator/{}/b3_f'.format(name), tf.norm(self.net['b3_f']))
        tf.summary.scalar('generator/{}/b4_f'.format(name), tf.norm(self.net['b4_f']))
        tf.summary.scalar('generator/{}/b5_fi'.format(name), tf.norm(self.net['b5_fi']))
        tf.summary.scalar('generator/{}/b5_fm'.format(name), tf.norm(self.net['b5_fm']))
        tf.summary.scalar('generator/{}/b2_b'.format(name), tf.norm(self.net['b2_b']))
        tf.summary.scalar('generator/{}/b3_b'.format(name), tf.norm(self.net['b3_b']))
        tf.summary.scalar('generator/{}/b4_b'.format(name), tf.norm(self.net['b4_b']))
        tf.summary.scalar('generator/{}/b5_b'.format(name), tf.norm(self.net['b5_b']))