def act()

in ddppo_agents.py [0:0]

• 14 lines of code
• 1 McCabe index (conditional complexity)

    def act(self, observations: Observations) -> Dict[str, int]:
        batch = batch_obs([observations], device=self.device)
        batch = apply_obs_transforms_batch(batch, self.obs_transforms)
        with torch.no_grad():
            (_, actions, _, self.test_recurrent_hidden_states) = self.actor_critic.act(
                batch,
                self.test_recurrent_hidden_states,
                self.prev_actions,
                self.not_done_masks,
                deterministic=False,
            )
            #  Make masks not done till reset (end of episode) will be called
            self.not_done_masks.fill_(True)
            self.prev_actions.copy_(actions)  # type: ignore

        return {"action": actions[0][0].item()}