Distribution Learning with Valid Outputs Beyond the Worst-Case

Open in new window