Super cool; training game players is super appealing to me as a math-software-engineer-person. My babble: if DeepPink can gage its uncertainty on a move, it'd

Super cool; training game players is super appealing to me as a math-software-engineer-person. My babble: if DeepPink can gage its uncertainty on a move, it'd be cool to see a hybrid system in-action. Plus, "DeepFish" has a cool name. Either way - nice! And thanks for putting the source on GitHub; I will have a goof with it!