Algorithms

Built-in policy implementations.

RandomPolicy

Uniform random policy over discrete actions.

from tinyrl import GridWorld, Runner, RandomPolicy

env = GridWorld()
policy = RandomPolicy(n_actions=env.n_actions)
runner = Runner(env)

result = runner.run_episode(policy)

Constructor

RandomPolicy(n_actions: int)

Args:

• n_actions — number of discrete actions to sample from uniformly