OakEmu-RL v0.1.0

Appearance

TrajectoryTrainingConfiguration ​

RL-side configuration for replayed and midpoint-started trajectories.

Source

Methods ​

resolve_aliases ​

python
resolve_aliases() -> TrajectoryTrainingConfiguration

Source

validate_archive_group_partitions_utilities ​

python
validate_archive_group_partitions_utilities() -> TrajectoryTrainingConfiguration

Source

validate_cell_utility_bucketing ​

python
validate_cell_utility_bucketing() -> TrajectoryTrainingConfiguration

Source

validate_episode_types ​

python
validate_episode_types(value: Any) -> EpisodeTypeSchedule

Source

validate_midpoint_partition ​

python
validate_midpoint_partition() -> TrajectoryTrainingConfiguration

Source

validate_utility_references ​

python
validate_utility_references() -> TrajectoryTrainingConfiguration

Source

Values ​

aliases ​

python
aliases: dict[str, str] = Field(default_factory=dict)

Source

episode_types ​

python
episode_types: EpisodeTypeSchedule = Field(default_factory=EpisodeTypeSchedule.default)

Source

midpoint ​

python
midpoint: TrajectoryMidpointConfiguration = Field(default_factory=TrajectoryMidpointConfiguration)

Source

random_seed ​

python
random_seed: int | None = None

Source

replay ​

python
replay: TrajectoryReplayConfiguration = Field(default_factory=TrajectoryReplayConfiguration)

Source

store ​

python
store: TrajectoryStoreConfiguration = Field(default_factory=TrajectoryStoreConfiguration)

The single segment store, keyed by destination cell; replay and midpoint selection share it.

Source

utilities ​

python
utilities: TrajectoryUtilitiesConfiguration | None = None

Source