Inspired by AlphaGo, this project seeks to train an AI to play the dice game, Farkle. Much like AlphaGo Zero, AlphaFarkle is designed to learn to effectively play the game of Farkle by exclusively playing against itself.
Because Farkle is a folk game, the rules are not well standardized. AlphaFarkle is trained using the rules defined by Elverson Puzzle.