No, you aren't wrong. For ML people, it is quite simple and hopefully the final code reflects that

The aim is really to give a good base for follow up research / modifications, which I think there will be many for this paper