coopihczoo.teaching.scripts_to_sort.behavioral_cloning_original.TrajectoryWithRew¶

class coopihczoo.teaching.scripts_to_sort.behavioral_cloning_original.TrajectoryWithRew(obs: numpy.ndarray, acts: numpy.ndarray, infos: Optional[numpy.ndarray], terminal: bool, rews: numpy.ndarray)[source]¶

Bases: coopihczoo.teaching.scripts_to_sort.behavioral_cloning_original.Trajectory

A Trajectory that additionally includes reward information.

Methods

Attributes

rews

Reward, shape (trajectory_len, ).

acts: numpy.ndarray¶: Actions, shape (trajectory_len, ) + action_shape.

infos: Optional[numpy.ndarray]¶: An array of info dicts, length trajectory_len.

obs: numpy.ndarray¶: Observations, shape (trajectory_len + 1, ) + observation_shape.

rews: numpy.ndarray¶: Reward, shape (trajectory_len, ). dtype float.

terminal: bool¶

Does this trajectory (fragment) end in a terminal state?

Episodes are always terminal. Trajectory fragments are also terminal when they contain the final state of an episode (even if missing the start of the episode).