def preferred_move(self):
"""Preferred move - the mean of selected moves."""
return tl.Fn(lambda dist_inputs, values: rl_layers.PreferredMove(
dist_inputs, self._policy_dist.sample),
n_out=1)
def f(dist_inputs, values): del values return rl_layers.PreferredMove(dist_inputs, self._policy_dist.sample)
def preferred_move(self):
"""Preferred move - the mean of selected moves."""
layer = tl.Fn(lambda dist_inputs, values: rl_layers.PreferredMove(
dist_inputs, self._policy_dist.sample), n_in=2, n_out=1)
return lambda **unused_kwargs: layer