I want to train variants using puzzle training.

We will implement that in the future, first goal for now is to add a computer opponent and analysis