Hi,
We are trying to train SERL for a different set of tasks. We see that the compliance params and action_scale very widely between different tasks in SERL. What do the compliance and precision params control? Do you have tips and best practices on how to go about setting them for each task?
Hi,
We are trying to train SERL for a different set of tasks. We see that the compliance params and action_scale very widely between different tasks in SERL. What do the compliance and precision params control? Do you have tips and best practices on how to go about setting them for each task?