Hi !
I want to change the method with respect to the action_comput of PPO.
The original form of action computing is the sampling by using the mean and variance of output of the Actor-network. But I need to use the parameterization trick. So, Which file should I change?