OakEmu-RL v0.1.0

Appearance

ObservationSample ​

python
ObservationSample(observations: np.ndarray, episode_ids: np.ndarray, episodes: int, wins: int, over_budget: int) -> None

Observations a policy actually visited, with where each came from.

Source

Values ​

episode_ids ​

python
episode_ids: np.ndarray

(N,) index of the episode each observation belongs to.

Source

episodes ​

python
episodes: int

Episodes played, including an unfinished last one.

Source

observations ​

python
observations: np.ndarray

(N, *observation_shape) in the environment's observation dtype. Every row is one the policy acted on.

Source

over_budget ​

python
over_budget: int

Episodes abandoned at the action budget.

Source

wins ​

python
wins: int

Episodes that ended in a win.

Source