cost_layers.py 文件源码

def compute_sample(self, state_below, temp=1, use_noise=False):
        """
        Constructs the theano expression that samples from the output layer.

        :type state_below: tensor or layer
        :param state_below: The theano expression (or groundhog layer)
            representing the input of the cost layer

        :type temp: float or tensor scalar
        :param temp: scalar representing the temperature that should be used
            when sampling from the output distribution

        :type use_noise: bool
        :param use_noise: flag. If true, noise is used when computing the
            output of the model
        """
        raise NotImplemented