I am trying to implement a DDPG agent to control the Gym’s Pendulum. Since I am new to gym, I was wondering if the state data collected via env.step(action) is already normalized or I should do that manually. Also, should action be normalized or ...