coopihczoo.teaching.scripts_to_sort.behavioral_cloning_original

Functions

check_for_correct_spaces	Checks that the environment has same spaces as provided ones.
dataclass_quick_asdict	Extract dataclass to items using dataclasses.fields + dict comprehension.
flatten_trajectories	Flatten a series of trajectory dictionaries into arrays.
generate_trajectories	Generate trajectory dictionaries from a policy and an environment.
make_min_episodes	Terminate after collecting n episodes of data.
make_min_timesteps	Terminate at the first episode after collecting n timesteps of data.
make_sample_until	Returns a termination condition sampling for a number of timesteps and episodes.
rollout	Generate policy rollouts.
transitions_collate_fn	Custom torch.utils.data.DataLoader collate_fn for TransitionsMinimal.
unwrap_traj	Uses RolloutInfoWrapper-captured obs and rews to replace fields.

Classes

BC	Behavioral cloning (BC).
ConstantLRSchedule	A callable that returns a constant learning rate.
EpochOrBatchIteratorWithProgress	Wraps DataLoader so that all BC batches can be processed in one for-loop.
FeedForward32Policy	A feed forward policy network with two hidden layers of 32 units.
RolloutInfoWrapper	Add the entire episode's rewards and observations to info at episode end.
Trajectory	A trajectory, e.g.
TrajectoryAccumulator	Accumulates trajectories step-by-step.
TrajectoryWithRew	A Trajectory that additionally includes reward information.
Transitions	A batch of obs-act-obs-done transitions.
TransitionsMinimal	A Torch-compatible Dataset of obs-act transitions.