back

Learning 'n-step' Actions for Control Tasks

Learning a policy that outputs both an action and the number of time-steps for which the action should be repeated.

Open in Google Slides