Policy

tinyrl.Policy is the abstract base class for policies. Subclass it to create reusable, structured policies that work with Runner.run_episode().

from tinyrl import Policy, PolicyOutput

class MyPolicy(Policy):
    def __call__(self, obs):
        action = ...
        return PolicyOutput(action=action, logprob=-0.5, entropy=1.2)

Using the Policy ABC is optional — run_episode also accepts any callable. The ABC is useful when building real policies because it provides a clear interface and catches missing implementations at instantiation time.

Abstract methods

__call__(obs: np.ndarray) -> int | np.ndarray | PolicyOutput

Choose an action given an observation.

Args:

• obs — current observation from the environment

Returns: one of:

• int — discrete action index
• np.ndarray — continuous action vector
• PolicyOutput — action with optional logprob and entropy