def __init__(
self,
mdp_rep_for_rl: MDPRepForRLTabular,
first_visit: bool,
softmax: bool,
epsilon: float,
epsilon_half_life: float,
num_episodes: int,
max_steps: int
) -> None:
super().__init__(
mdp_rep_for_rl=mdp_rep_for_rl,
softmax=softmax,
epsilon=epsilon,
epsilon_half_life=epsilon_half_life,
num_episodes=num_episodes,
max_steps=max_steps
)
self.first_visit: bool = first_visit
self.nt_return_eval_steps = get_nt_return_eval_steps(
max_steps,
mdp_rep_for_rl.gamma,
1e-4
)
def __init__(self, mdp_rep_for_rl: MDPRepForRLFA, first_visit: bool,
softmax: bool, epsilon: float, num_episodes: int,
max_steps: int, fa_spec: FuncApproxSpec) -> None:
super().__init__(mdp_rep_for_rl=mdp_rep_for_rl,
softmax=softmax,
epsilon=epsilon,
num_episodes=num_episodes,
max_steps=max_steps,
fa_spec=fa_spec)
self.first_visit: bool = first_visit
self.nt_return_eval_steps = get_nt_return_eval_steps(
max_steps, mdp_rep_for_rl.gamma, 1e-4)
def __init__(self, mdp_rep_for_rl: MDPRepForRLFA, exploring_start: bool,
softmax: bool, epsilon: float, epsilon_half_life: float,
num_episodes: int, max_steps: int,
fa_spec: FuncApproxSpec) -> None:
super().__init__(mdp_rep_for_rl=mdp_rep_for_rl,
exploring_start=exploring_start,
softmax=softmax,
epsilon=epsilon,
epsilon_half_life=epsilon_half_life,
num_episodes=num_episodes,
max_steps=max_steps,
fa_spec=fa_spec)
self.nt_return_eval_steps = get_nt_return_eval_steps(
max_steps, mdp_rep_for_rl.gamma, 1e-4)
