q_derived.q_network_derived_epsilon_greedy_policy

Classes

QNetworkDerivedEpsilonGreedyPolicy

Helper class that provides a standard way to create an ABC using

Module Contents

class q_derived.q_network_derived_epsilon_greedy_policy.QNetworkDerivedEpsilonGreedyPolicy(q_network: torch.nn.Module, num_actions: int, action_space=None, epsilon=0.1, allow_decay=True, epsilon_min=0.001, decay_rate=0.01)

Bases: gridmind.policies.soft.q_derived.base_q_derived_soft_policy.BaseQDerivedSoftPolicy

Helper class that provides a standard way to create an ABC using inheritance.

action_space = None

allow_decay = True

epsilon_min = 0.001

decay_rate = 0.01

device

get_network()

set_network(network)

update(state, action)

update_q(state, action, value: float)

_get_greedy_action(state)

set_epsilon(value: float)

decay_epsilon()