MuZero and Atari
In our example, we used Connect 4, which is a two-player board game, but we shouldn’t miss the fact that MuZero’s generalization (usage of hidden state) makes it possible to apply it to more classical RL scenarios. In the paper by Schrittwieser et al. [Sch+20], the authors successfully applied the method to 57 Atari games. Of course, the method requires tuning and adaptation to such scenarios, but the core is the same. This has been left as an exercise for you to try by yourself.
