def step_async()

in understanding_rl_vision/rl_clarity/training.py [0:0]

• 9 lines of code
• 3 McCabe index (conditional complexity)

    def step_async(self, actions):
        mask = np.random.uniform(size=self.num_envs) < self.epsilon
        new_actions = np.array(
            [
                self.action_space.sample() if mask[i] else actions[i]
                for i in range(self.num_envs)
            ]
        )
        self.venv.step_async(new_actions)