def _init_pars(self):
spec = varprop_rnn.parameters(
self.n_inpt, self.n_hiddens, self.n_output, self.skip_to_out,
self.hidden_transfers, self.out_transfer)
self.parameters = ParameterSet(**spec)
self.parameters.data[:] = np.random.standard_normal(
self.parameters.data.shape).astype(theano.config.floatX)
def _init_pars(self):
spec = varprop_rnn.parameters(self.n_inpt, self.n_hiddens,
self.n_output, self.skip_to_out,
self.hidden_transfers, self.out_transfer)
self.parameters = ParameterSet(**spec)
self.parameters.data[:] = np.random.standard_normal(
self.parameters.data.shape).astype(theano.config.floatX)
def _recog_par_spec(self):
"""Return the parameter specification of the recognition model."""
spec = vprnn.parameters(self.n_inpt, self.n_hiddens_recog, self.n_latent)
spec['p_dropout'] = {
'inpt': 1,
'hiddens': [1 for _ in self.n_hiddens_recog],
'hidden_to_out': 1,
}
return spec
def _gen_par_spec(self):
"""Return the parameter specification of the generating model."""
n_output = self.assumptions.visible_layer_size(self.n_inpt)
spec = vprnn.parameters(
self.n_latent + self.n_hiddens_recog[-1], self.n_hiddens_gen,
n_output,
hidden_transfers=self.gen_transfers,
)
return spec
def _gen_par_spec(self):
"""Return the parameter specification of the generating model."""
n_output = self.assumptions.visible_layer_size(self.n_inpt)
spec = vprnn.parameters(
self.n_latent + self.n_hiddens_recog[-1],
self.n_hiddens_gen,
n_output,
hidden_transfers=self.gen_transfers,
)
return spec
def _recog_par_spec(self):
"""Return the parameter specification of the recognition model."""
spec = vprnn.parameters(
self.n_inpt,
self.n_hiddens_recog,
self.n_latent,
hidden_transfers=self.recog_transfers,
)
spec['p_dropout'] = {
'inpt': 1,
'hiddens': [1 for _ in self.n_hiddens_recog],
'hidden_to_out': 1,
}
return spec
