garage.torch.algos.maml_vpg

Model-Agnostic Meta-Learning (MAML) algorithm applied to VPG.

class MAMLVPG(env, policy, value_function, inner_lr=_Default(0.1), outer_lr=0.001, max_episode_length=100, discount=0.99, gae_lambda=1, center_adv=True, positive_adv=False, policy_ent_coeff=0.0, use_softplus_entropy=False, stop_entropy_gradient=False, entropy_method='no_entropy', meta_batch_size=20, num_grad_updates=1, meta_evaluator=None, evaluate_every_n_epochs=1)

Bases: garage.torch.algos.maml.MAML

Inheritance diagram of garage.torch.algos.maml_vpg.MAMLVPG

Model-Agnostic Meta-Learning (MAML) applied to VPG.

Parameters:
  • env (Environment) – A multi-task environment.
  • policy (garage.torch.policies.Policy) – Policy.
  • value_function (garage.np.baselines.Baseline) – The value function.
  • inner_lr (float) – Adaptation learning rate.
  • outer_lr (float) – Meta policy learning rate.
  • max_episode_length (int) – Maximum length of a single episode.
  • discount (float) – Discount.
  • gae_lambda (float) – Lambda used for generalized advantage estimation.
  • center_adv (bool) – Whether to rescale the advantages so that they have mean 0 and standard deviation 1.
  • positive_adv (bool) – Whether to shift the advantages so that they are always positive. When used in conjunction with center_adv the advantages will be standardized before shifting.
  • policy_ent_coeff (float) – The coefficient of the policy entropy. Setting it to zero would mean no entropy regularization.
  • use_softplus_entropy (bool) – Whether to estimate the softmax distribution of the entropy to prevent the entropy from being negative.
  • stop_entropy_gradient (bool) – Whether to stop the entropy gradient.
  • entropy_method (str) – A string from: ‘max’, ‘regularized’, ‘no_entropy’. The type of entropy method to use. ‘max’ adds the dense entropy to the reward for each time step. ‘regularized’ adds the mean entropy to the surrogate objective. See https://arxiv.org/abs/1805.00909 for more details.
  • meta_batch_size (int) – Number of tasks sampled per batch.
  • num_grad_updates (int) – Number of adaptation gradient steps.
  • meta_evaluator (garage.experiment.MetaEvaluator) – A meta evaluator for meta-testing. If None, don’t do meta-testing.
  • evaluate_every_n_epochs (int) – Do meta-testing every this epochs.
policy

Current policy of the inner algorithm.

Returns:
Current policy of the inner
algorithm.
Return type:garage.torch.policies.Policy
train(self, runner)

Obtain samples and start training for each epoch.

Parameters:runner (LocalRunner) – Gives the algorithm access to :method:`~LocalRunner.step_epochs()`, which provides services such as snapshotting and sampler control.
Returns:The average return in last epoch cycle.
Return type:float
train_once(self, runner, all_samples, all_params)

Train the algorithm once.

Parameters:
  • runner (LocalRunner) – The experiment runner.
  • all_samples (list[list[MAMLEpisodeBatch]]) – A two dimensional list of MAMLEpisodeBatch of size [meta_batch_size * (num_grad_updates + 1)]
  • all_params (list[dict]) – A list of named parameter dictionaries. Each dictionary contains key value pair of names (str) and parameters (torch.Tensor).
Returns:

Average return.
Return type:

float
log_performance(self, itr, all_samples, loss_before, loss_after, kl_before, kl, policy_entropy)

Evaluate performance of this batch.

Parameters:
  • itr (int) – Iteration number.
  • all_samples (list[list[MAMLEpisodeBatch]]) – Two dimensional list of MAMLEpisodeBatch of size [meta_batch_size * (num_grad_updates + 1)]
  • loss_before (float) – Loss before optimization step.
  • loss_after (float) – Loss after optimization step.
  • kl_before (float) – KL divergence before optimization step.
  • kl (float) – KL divergence after optimization step.
  • policy_entropy (float) – Policy entropy.
Returns:

The average return in last epoch cycle.
Return type:

float
get_exploration_policy(self)

Return a policy used before adaptation to a specific task.

Each time it is retrieved, this policy should only be evaluated in one task.

Returns:
The policy used to obtain samples that are later used for
meta-RL adaptation.
Return type:Policy
adapt_policy(self, exploration_policy, exploration_episodes)

Adapt the policy by one gradient steps for a task.

Parameters:
  • exploration_policy (Policy) – A policy which was returned from get_exploration_policy(), and which generated exploration_episodes by interacting with an environment. The caller may not use this object after passing it into this method.
  • exploration_episodes (EpisodeBatch) – Episodes with which to adapt, generated by exploration_policy exploring the environment.
Returns:

A policy adapted to the task represented by the

exploration_episodes.
Return type:

Policy