def __init__(self, env_spec, **kwargs):
self._obs_dim = env_spec.observation_space.flat_dim
self._action_dim = env_spec.action_space.flat_dim
Policy.__init__(self, env_spec)
GaussianMLPModule.__init__(self,
input_dim=self._obs_dim,
output_dim=self._action_dim,
**kwargs)
def __init__(self,
env_spec,
hidden_sizes=(32, 32),
hidden_nonlinearity=torch.tanh,
hidden_w_init=nn.init.xavier_uniform_,
hidden_b_init=nn.init.zeros_,
output_nonlinearity=None,
output_w_init=nn.init.xavier_uniform_,
output_b_init=nn.init.zeros_,
learn_std=True,
init_std=1.0,
layer_normalization=False,
name='GaussianMLPValueFunction'):
super(GaussianMLPValueFunction, self).__init__(env_spec, name)
input_dim = env_spec.observation_space.flat_dim
output_dim = 1
self.module = GaussianMLPModule(
input_dim=input_dim,
output_dim=output_dim,
hidden_sizes=hidden_sizes,
hidden_nonlinearity=hidden_nonlinearity,
hidden_w_init=hidden_w_init,
hidden_b_init=hidden_b_init,
output_nonlinearity=output_nonlinearity,
output_w_init=output_w_init,
output_b_init=output_b_init,
learn_std=learn_std,
init_std=init_std,
min_std=None,
max_std=None,
std_parameterization='exp',
layer_normalization=layer_normalization)
def __init__(self,
env_spec,
hidden_sizes=(32, 32),
hidden_nonlinearity=torch.tanh,
hidden_w_init=nn.init.xavier_uniform_,
hidden_b_init=nn.init.zeros_,
output_nonlinearity=None,
output_w_init=nn.init.xavier_uniform_,
output_b_init=nn.init.zeros_,
learn_std=True,
init_std=1.0,
min_std=1e-6,
max_std=None,
std_parameterization='exp',
layer_normalization=False,
name='GaussianMLPPolicy'):
self._obs_dim = env_spec.observation_space.flat_dim
self._action_dim = env_spec.action_space.flat_dim
Policy.__init__(self, env_spec, name)
GaussianMLPModule.__init__(self,
input_dim=self._obs_dim,
output_dim=self._action_dim,
hidden_sizes=hidden_sizes,
hidden_nonlinearity=hidden_nonlinearity,
hidden_w_init=hidden_w_init,
hidden_b_init=hidden_b_init,
output_nonlinearity=output_nonlinearity,
output_w_init=output_w_init,
output_b_init=output_b_init,
learn_std=learn_std,
init_std=init_std,
min_std=min_std,
max_std=max_std,
std_parameterization=std_parameterization,
layer_normalization=layer_normalization)