Why the reward function must be called before updating the observation in PointWiseEnv.py?
2017040264 opened this issue · comments
Shouldn't the reward be an evaluation of the observed new state S(t+1)?
For example，openai_gym_cartpole
2017040264 opened this issue · comments
Shouldn't the reward be an evaluation of the observed new state S(t+1)?
For example，openai_gym_cartpole