def __init__(
self,
observation_space,
my_action_space,
other_action_space,
index=None,
mas=None,
exploration="EpsGreedy",
gamma=0.99,
lr=0.1,
target_update_freq=None,
name="MinimaxQAgent",
):
QAgent.__init__(
self,
qmodel=MultiQTable(
gymSpace2dim(observation_space),
[
gymSpace2dim(my_action_space),
gymSpace2dim(other_action_space)
],
),
observation_space=observation_space,
action_space=my_action_space,
experience="ReplayMemory-1",
exploration=exploration,
gamma=gamma,
lr=lr,
batch_size=1,
target_update_freq=target_update_freq,
name=name,
)
MATrainable.__init__(self, mas, index)
def __init__(self, model, observation_space=None, action_space=None):
self.observation_space = observation_space
self.action_space = action_space
self.model = marl.model.make(model,
obs_sp=gymSpace2dim(
self.observation_space),
act_sp=gymSpace2dim(self.action_space))
def __init__(self, model, observation_space=None, action_space=None):
super(StochasticPolicy, self).__init__()
self.observation_space = observation_space
self.action_space = action_space
obs_dim = gymSpace2dim(self.observation_space)
act_dim = gymSpace2dim(self.action_space)
self.model = marl.model.make(model, obs_sp=obs_dim, act_sp=act_dim)
def __init__(self, model, observation_space=None, action_space=None):
super(DeterministicPolicy, self).__init__()
self.observation_space = observation_space
self.action_space = action_space
self.low = (self.action_space.low[0] if isinstance(
self.action_space, gym.spaces.Box) else 0)
self.high = (self.action_space.high[0] if isinstance(
self.action_space, gym.spaces.Box) else 1)
obs_dim = gymSpace2dim(self.observation_space)
act_dim = gymSpace2dim(self.action_space)
self.model = marl.model.make(model, obs_sp=obs_dim, act_sp=act_dim)