Experience

Experience tuple at time step \(t\). It contains the observation, action, next observation, reward, terminated. Each item shape is (num_envs, *item_shape). *item_shape depends on the item.

Module: aine_drl.exp

@dataclass(frozen=True)
class Experience

Fields

obs

The observation tuple. See Observation docs.

obs: Observation

action

See Action docs.

action: Action

next_obs

The next observation tuple. See Observation docs.

next_obs: Observation

reward

The reward tensor. *item_shape = 1

reward: Tensor

terminated

The terminated tensor. *item_shape = 1

terminated: Tensor

Methods

transform()

Transform the experience tuple with the callable function.

def transform(self, func: Callable[[Tensor], Tensor]) -> Experience

Parameters:

Returns: