garage.torch.value_functions.gaussian_mlp_value_function

A value function based on a GaussianMLP model.

class GaussianMLPValueFunction(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')

Bases: garage.torch.value_functions.value_function.ValueFunction

Inheritance diagram of garage.torch.value_functions.gaussian_mlp_value_function.GaussianMLPValueFunction

Gaussian MLP Value Function with Model.

It fits the input data to a gaussian distribution estimated by a MLP.

Parameters
  • env_spec (EnvSpec) – Environment specification.

  • hidden_sizes (list[int]) – Output dimension of dense layer(s) for the MLP for mean. For example, (32, 32) means the MLP consists of two hidden layers, each with 32 hidden units.

  • hidden_nonlinearity (callable) – Activation function for intermediate dense layer(s). It should return a torch.Tensor. Set it to None to maintain a linear activation.

  • hidden_w_init (callable) – Initializer function for the weight of intermediate dense layer(s). The function should return a torch.Tensor.

  • hidden_b_init (callable) – Initializer function for the bias of intermediate dense layer(s). The function should return a torch.Tensor.

  • output_nonlinearity (callable) – Activation function for output dense layer. It should return a torch.Tensor. Set it to None to maintain a linear activation.

  • output_w_init (callable) – Initializer function for the weight of output dense layer(s). The function should return a torch.Tensor.

  • output_b_init (callable) – Initializer function for the bias of output dense layer(s). The function should return a torch.Tensor.

  • learn_std (bool) – Is std trainable.

  • init_std (float) – Initial value for std. (plain value - not log or exponentiated).

  • layer_normalization (bool) – Bool for using layer normalization or not.

  • name (str) – The name of the value function.

compute_loss(self, obs, returns)

Compute mean value of loss.

Parameters
  • obs (torch.Tensor) – Observation from the environment with shape \((N \dot [T], O*)\).

  • returns (torch.Tensor) – Acquired returns with shape \((N, )\).

Returns

Calculated negative mean scalar value of

objective (float).

Return type

torch.Tensor

forward(self, obs)

Predict value based on paths.

Parameters

obs (torch.Tensor) – Observation from the environment with shape \((P, O*)\).

Returns

Calculated baselines given observations with

shape \((P, O*)\).

Return type

torch.Tensor