torchrl.trainers.algorithms.configs.hooks.LogScalarConfig

class torchrl.trainers.algorithms.configs.hooks.LogScalarConfig(key: Any = ('next', 'reward'), logname: str | None = None, log_pbar: bool = False, include_std: bool = True, reduction: str = 'mean', _target_: str = 'torchrl.trainers.trainers.LogScalar')

Configuration for the LogScalar hook.

Examples

>>> from torchrl.trainers.algorithms.configs.hooks import LogScalarConfig
>>> from hydra.utils import instantiate
>>> hook = instantiate(
...     LogScalarConfig(key=["next", "reward"], logname="train_reward")
... )