I have a custom gymnasium Mujoco environment that has an action space that has specific ranges for actions. For example, for some action vector [x1, x2, x3, …, x20], x1 to x3 have ranges from -1 to 1, whereas x20 has range 0 to 1. This corresponds to actuation in the Mujoco model.
I am wondering: is there a way to incorporate this into something like TD3 policy in RLLib?