The step method of the SelfPlayEnv calss

selfplay.py

class SelfPlayEnv(env):
  # ...

    def step(self, action):
        self.render()
        observation, reward, done, _ = super(SelfPlayEnv, self).step(action)
        logger.debug(f'Action played by agent: {action}')
        logger.debug(f'Rewards: {reward}')
        logger.debug(f'Done: {done}')

        if not done:
            observation, reward, done, _ = self.continue_game()

        agent_reward = reward[self.agent_player_num]
        logger.debug(f'\nReward To Agent: {agent_reward}')

        if done:
            self.render()

        return observation, agent_reward, done, {}