/pokrl

AI for teaching and playing Poker.

Primary LanguagePythonApache License 2.0Apache-2.0

Pokrl

A Reinforcement Learning based approach for playing Poker.

Features:

  • The AI uses a single output regressor (SOR) to take actions in the game. Where 0 represents check/fold, and other number represents the quantity that the AI wants to raise the bet.
  • The current game speed is around 6.3 kgps per thread (to be improved).