class AGENT:
def __init__(self, num_states, num_actions, args):
self.args = args
self.brain = BRAIN(num_states, num_actions, self.args)
def get_action(self, state, weight):
action = self.brain.decide_action(state, weight)
return action
def get_Q_value(self, state, action):
Q_vector = self.brain.compute_Q_value(state, action)
return Q_vector
def get_next_value(self, next_state, weight):
V_next_vector = self.brain.compute_next_value(next_state, weight)
return V_next_vector
class AGENT:
def __init__(self, num_states, num_actions, args):
self.args = args
self.brain = BRAIN(num_states, num_actions, self.args)
def get_action(self, state, exploration_noise):
action = self.brain.decide_action(state)
if exploration_noise is not None:
action += torch.Tensor(exploration_noise.noise())
return action
class AGENT:
def __init__(self, num_states, num_actions, args):
self.args = args
self.brain = BRAIN(num_states, num_actions, self.args)
def update_DNNs(self, batch):
self.brain.update_network(batch)
def get_action(self, state, exploration_noise):
action = self.brain.decide_action(state)
#add noise############################################
if exploration_noise is not None:
action += torch.Tensor(exploration_noise.noise())
if action[0, 0] > 1.0:
action[0, 0] = 1.0
elif action[0, 0] < -1.0:
action[0, 0] = -1.0
else:
action[0, 0] = action[0, 0]
######################################################
return action
def update_target_DNNs(self):
self.brain.update_target_network()
def __init__(self, num_states, num_actions, args):
self.args = args
self.brain = BRAIN(num_states, num_actions, self.args)