OakEmu-RL v0.1.0

Appearance

Experiment ​

python
Experiment(config: TConfig)

Source

Properties ​

config ​

python
config: TConfig

Source

dump_directory ​

python
dump_directory: Path

Source

experiment_name ​

python
experiment_name: str

Source

framework ​

python
framework: str

Source

game_configuration ​

python
game_configuration: GameConfiguration

Source

game_name ​

python
game_name: str

Source

model_configuration ​

python
model_configuration: ModelConfiguration

Source

output_directory ​

python
output_directory: Path

Source

training_configuration ​

python
training_configuration: TrainingConfiguration

Source

Methods ​

diagnose ​

python
diagnose(*, checkpoint_number: int | None = None, episodes: int = 16, max_observations: int = 8192) -> Path

Measure a checkpoint's network on the states its own policy visits and write a report.

Parameters:

  • checkpoint_number (int | None) – The checkpoint to measure, or None for the latest.
  • episodes (int) – Episodes to play to gather observations.
  • max_observations (int) – Stop gathering once this many observations are held.

Returns:

  • Path – The path of the written report.

Source

export ​

python
export(output_path: str | PathLike | None = None) -> Path

Source

train ​

python
train() -> None

Source