LogValidationReward

class torchrl.trainers.LogValidationReward(*, record_interval: int, record_frames: int, frame_skip: int = 1, policy_exploration: TensorDictModule, environment: EnvBase = None, exploration_type: ExplorationType = InteractionType.RANDOM, log_keys: list[str] | tuple[str]] | None = None, out_keys: dict[str | tuple[str], str] | None = None, suffix: str | None = None, log_pbar: bool = False, recorder: EnvBase = None)

Recorder hook for Trainer.

参数:

• record_interval (int) – total number of optimization steps between two calls to the recorder for testing.

• record_frames (int) – number of frames to be recorded during testing.

• frame_skip (int) – frame_skip used in the environment. It is important to let the trainer know the number of frames skipped at each iteration, otherwise the frame count can be underestimated. For logging, this parameter is important to normalize the reward. Finally, to compare different runs with different frame_skip, one must normalize the frame count and rewards. Defaults to 1.

• policy_exploration (ProbabilisticTDModule) –

  a policy instance used for

  1. updating the exploration noise schedule;

  2. testing the policy on the recorder.

  Given that this instance is supposed to both explore and render the performance of the policy, it should be possible to turn off the explorative behavior by calling the set_exploration_type(ExplorationType.DETERMINISTIC) context manager.

• environment (EnvBase) – An environment instance to be used for testing.

• exploration_type (ExplorationType, optional) – exploration mode to use for the policy. By default, no exploration is used and the value used is ExplorationType.DETERMINISTIC. Set to ExplorationType.RANDOM to enable exploration

• log_keys (sequence of str or tuples or str, optional) – keys to read in the tensordict for logging. Defaults to [("next", "reward")].

• out_keys (Dict[str, str], optional) – a dictionary mapping the log_keys to their name in the logs. Defaults to {("next", "reward"): "r_evaluation"}.

• suffix (str, optional) – suffix of the video to be recorded.

• log_pbar (bool, optional) – if True, the reward value will be logged on the progression bar. Default is False.

register(trainer: Trainer, name: str = 'recorder')

Registers the hook in the trainer at a default location.

参数:

• trainer (Trainer) – the trainer where the hook must be registered.

• name (str) – the name of the hook.

注意

To register the hook at another location than the default, use register_op().