The bsuite-challenge-starter-kit notebook has been updated which allows the Runner class to give the Agent's learn() access to the done flag.
- The major change is in the
Agentclass where thedoneflag is passed as the final parameter. - The minor changes are to (a) the
RandomAgentto mirror the same change in thelearn()and (b) the example train loop now passes the done flag to thelearn()
Please make sure to use the updated notebook.