rbm.py 文件源码

def step_gibbs(self, r_h, r_v, h, *params):
        '''Step Gibbs sample.

        Args:
            r_h (theano.randomstream): random variables for hiddens.
            r_v (theano.randomstream): random variables for visibles.
            h (T.tensor): hidden state.
            *params: theano shared variables.

        Returns:
            T.tensor: hidden samples.
            T.tensor: visible samples.
            T.tensor: conditional hidden probability.
            T.tensor: conditional visible probability.

        '''
        v, pv = self.step_sv_h(r_v, h, *params)
        h, ph = self.step_sh_v(r_h, v, *params)
        return h, v, ph, pv