tf_agents.environments.suite_gym.wrap_env

Wraps given gym environment with TF Agent's GymWrapper.

tf_agents.environments.suite_gym.wrap_env(
    gym_env: gym.Env,
    discount: tf_agents.typing.types.Float = 1.0,
    max_episode_steps: Optional[types.Int] = None,
    gym_env_wrappers: Sequence[tf_agents.typing.types.GymEnvWrapper] = (),
    time_limit_wrapper: tf_agents.environments.suite_gym.TimeLimitWrapperType = tf_agents.environments.TimeLimit,
    env_wrappers: Sequence[tf_agents.typing.types.PyEnvWrapper] = (),
    spec_dtype_map: Optional[Dict[gym.Space, np.dtype]] = None,
    auto_reset: bool = True,
    render_kwargs: Optional[Dict[str, Any]] = None
) -> tf_agents.environments.PyEnvironment

Note that by default a TimeLimit wrapper is used to limit episode lengths to the default benchmarks defined by the registered environments.

gym_env An instance of OpenAI gym environment.
discount Discount to use for the environment.
max_episode_steps Used to create a TimeLimitWrapper. No limit is applied if set to None or 0. Usually set to gym_spec.max_episode_steps in load.
gym_env_wrappers Iterable with references to wrapper classes to use directly on the gym environment.
time_limit_wrapper Wrapper that accepts (env, max_episode_steps) params to enforce a TimeLimit. Usuaully this should be left as the default, wrappers.TimeLimit.
env_wrappers Iterable with references to wrapper classes to use on the gym_wrapped environment.
spec_dtype_map A dict that maps gym specs to tf dtypes to use as the default dtype for the tensors. An easy way how to configure a custom mapping through Gin is to define a gin-configurable function that returns desired mapping and call it in your Gin config file, for example: suite_gym.load.spec_dtype_map = @get_custom_mapping().
auto_reset If True (default), reset the environment automatically after a terminal state is reached.
render_kwargs Optional dict of keywoard arguments for rendering.

A PyEnvironment instance.