def _build_replay_memory(self, use_staging): """Creates the replay memory used by the agent. Args: use_staging: bool, if True, uses a staging area for replaying. Returns: A replay memory object. """ return replay_memory.WrappedReplayMemory( num_actions=self.num_actions, observation_size=self.observation_size, batch_size=32, stack_size=1, use_staging=use_staging, update_horizon=self.update_horizon, gamma=self.gamma)
def _build_replay_memory(self, use_staging): """Creates the replay memory used by the agent. Rainbow uses prioritized replay. Args: use_staging: bool, whether to use a staging area in the replay memory. Returns: A replay memory object. """ return replay_memory.WrappedReplayMemory( num_actions=self.num_actions, observation_size=self.observation_size, stack_size=1, use_staging=use_staging, update_horizon=self.update_horizon, gamma=self.gamma)