def start(self, observation): self.episode_counter += 1 if self.episode_counter < episodes_only_dqn: action = BaseDynaAgent.start(self, observation) elif self.episode_counter < episodes_only_dqn + episodes_only_mcts: action = MCTSAgent.start(self, observation) else: if self.episode_counter % 2 == 0: action = BaseDynaAgent.start(self, observation) else: action = MCTSAgent.start(self, observation) return action
def start(self, observation): self.episode_counter += 1 if self.episode_counter % 2 == 0: action = BaseDynaAgent.start(self, observation) else: action = MCTSAgent.start(self, observation) return action
def start(self, observation): self.episode_counter += 1 if self.episode_counter % 2 == 0: action = BaseDynaAgent.start(self, observation) else: action = MCTSAgent.start(self, observation) self.mcts_prev_state = self.getStateRepresentation(observation) self.mcts_prev_action = action return action
def start(self, observation): self.episode_counter += 1 if self._sr['network'] is None: self.init_s_representation_network(observation) action = MCTSAgent.start(self, observation) return action