gym_yatzy

Yatzy Gym is an OpenAI Gym environment designed for training reinforcement learning agents to play a version of poker dice, also known as Yatzy, Yahtzee, or Kniffel, among other names. In this environment, the agent interacts with a game setup where it rolls dice, chooses which dice to keep, and fills in scoring fields based on the outcome. The observation space includes information such as the current dice values, the number of rolls, and the agent’s progress on various scoring fields. The action space allows the agent to select which dice to keep, whether to roll again, and which field to fill based on the game’s rules. Additional customization options include penalty settings for illegal moves and an optional discrete action space. This environment is a versatile tool for developing agents that can solve complex decision-making problems in turn-based games with partial information.