BehaviourCloningSettings
Full path:
schola.scripts.rllib.launch_imitation.BehaviourCloningSettings
schola.scripts.rllib.launch_imitation.BehaviourCloningSettings
BehaviourCloningSettings
BehaviourCloningSettings( cloning_steps=1000000, expert_path=None, converted_expert_path=None)Bases: object
Dataclass for behavior cloning settings used in the RLlib training process. This class defines the parameters for behavior cloning. This includes the path to the trajectory data that is collected from the environment and processed into rllib compatible format through conversion script.
Methods
| Item | Description |
|---|---|
| init | — |
Attributes
| Item | Description |
|---|---|
| cloning_steps | The number of timesteps to clone. |
| converted_expert_path | Path to the expert data converted to rllib format. |
| expert_path | Path to the original expert data for behavior cloning. |
| name | — |
Parameters
cloning_steps (Annotated[int, Parameter(validator=(Number(lt=None, lte=None, gt=0, gte=None, modulo=None),))])
expert_path (Annotated[Path | None, Parameter(validator=(Path(exists=True, file_okay=True, dir_okay=False, ext=()),))])
converted_expert_path (Annotated[Path | None, Parameter(validator=(Path(exists=False, file_okay=False, dir_okay=True, ext=()),))])
init
__init__(cloning_steps=1000000, expert_path=None, converted_expert_path=None)Parameters
cloning_steps (Annotated[int, Parameter(validator=(Number(lt=None, lte=None, gt=0, gte=None, modulo=None),))])
expert_path (Annotated[Path | None, Parameter(validator=(Path(exists=True, file_okay=True, dir_okay=False, ext=()),))])
converted_expert_path (Annotated[Path | None, Parameter(validator=(Path(exists=False, file_okay=False, dir_okay=True, ext=()),))])
Returns
None
cloning_steps
= 1000000 cloning_steps: Annotated[int, Parameter(validator=Number(lt=None, lte=None, gt=0, gte=None, modulo=None))]The number of timesteps to clone. This specifies the total number of timesteps to use for behavior cloning. This can be adjusted based on the amount of data available and the desired amount of training data for the behavior cloning process.
converted_expert_path
= None converted_expert_path: Annotated[Path | None, Parameter(validator=Path(exists=False, file_okay=False, dir_okay=True, ext=()))]Path to the expert data converted to rllib format. This specifies the location of the trajectory data that has been processed into a format compatible with RLlib. This data will be used for training the behavior cloning model.
expert_path
= None expert_path: Annotated[Path | None, Parameter(validator=Path(exists=True, file_okay=True, dir_okay=False, ext=()))]Path to the original expert data for behavior cloning. This specifies the location of the trajectory data that is collected from the environment. This data needs to be converted into a format compatible with RLlib before it can be used for training the behavior cloning model.
name
name: str