class TD3Policy(*, actor: Module, actor_optim: Optimizer, critic: Module, critic_optim: Optimizer, action_space: Space, critic2: Module | None = None, critic2_optim: Optimizer | None = None, tau: float = 0.005, gamma: float = 0.99, exploration_noise: BaseNoise | Literal['default'] | None = 'default', policy_noise: float = 0.2, update_actor_freq: int = 2, noise_clip: float = 0.5, estimation_step: int = 1, observation_space: Space | None = None, action_scaling: bool = True, action_bound_method: Literal['clip'] | None = 'clip', lr_scheduler: LRScheduler | MultipleLRSchedulers | None = None)[source]#

Implementation of TD3, arXiv:1802.09477.

  • actor – the actor network following the rules in BasePolicy. (s -> actions)

  • actor_optim – the optimizer for actor network.

  • critic – the first critic network. (s, a -> Q(s, a))

  • critic_optim – the optimizer for the first critic network.

  • action_space – Env’s action space. Should be gym.spaces.Box.

  • critic2 – the second critic network. (s, a -> Q(s, a)). If None, use the same network as critic (via deepcopy).

  • critic2_optim – the optimizer for the second critic network. If None, clone critic_optim to use for critic2.parameters().

  • tau – param for soft update of the target network.

  • gamma – discount factor, in [0, 1].

  • exploration_noise – add noise to action for exploration. This is useful when solving “hard exploration” problems. “default” is equivalent to GaussianNoise(sigma=0.1).

  • policy_noise – the noise used in updating policy network.

  • update_actor_freq – the update frequency of actor network.

  • noise_clip – the clipping range used in updating policy network.

  • observation_space – Env’s observation space.

  • action_scaling – if True, scale the action from [-1, 1] to the range of action_space. Only used if the action_space is continuous.

  • action_bound_method – method to bound action to range [-1, 1]. Only used if the action_space is continuous.

  • lr_scheduler – a learning rate scheduler that adjusts the learning rate in optimizer in each policy.update()

See also

Please refer to BasePolicy for more detailed explanation.

learn(batch: RolloutBatchProtocol, *args: Any, **kwargs: Any) TTD3TrainingStats[source]#

Update policy with a given batch of data.


A dataclass object, including the data needed to be logged (e.g., loss).


In order to distinguish the collecting state, updating state and testing state, you can check the policy state by and self.updating. Please refer to States for policy for more detailed explanation.


If you use torch.distributions.Normal and torch.distributions.Categorical to calculate the log_prob, please be careful about the shape: Categorical distribution gives “[batch_size]” shape while Normal distribution gives “[batch_size, 1]” shape. The auto-broadcasting of numerical operation with torch tensors will amplify this error.

sync_weight() None[source]#

Soft-update the weight for the target network.

train(mode: bool = True) Self[source]#

Set the module in training mode, except for the target network.

class TD3TrainingStats(*, train_time: float = 0.0, smoothed_loss: dict = <factory>, actor_loss: float, critic1_loss: float, critic2_loss: float)[source]#
actor_loss: float#
critic1_loss: float#
critic2_loss: float#