Train an agent with discrete action

Hello,

I want to train an agent with discrete action space, but I don’t know how to implement.
It seems that all samples in IsaacGym have continuous action space, so I am confused.
Is there any good example codes?

Thank you very much.