Reinforcement learning algorithms for Blackjack game. Use main.py to train agent, simulate games and print percentage of wins and draws.
usage: main.py [-h] [-a ALGORITHM] [--epsilon EPSILON] [--alpha ALPHA]
[--gamma GAMMA] [-e EPISODES] [-g GAMES]
Blackjack RL
optional arguments:
-h, --help show this help message and exit
-a ALGORITHM, --algorithm ALGORITHM
algorithm used to train agent. One of: "Deterministic,
MCES, MCEpsilonGreedy, Sarsa, QLearning "
--epsilon EPSILON epsilon used for probability in e-greedy algorithms,
default=0.1
--alpha ALPHA alpha used in TD algorithms, default=0.01
--gamma GAMMA gamma used in TD algorithms, default=0.15
-e EPISODES, --episodes EPISODES
number of episodes used for learning agent,
default=100000
-g GAMES, --games GAMES
number of games to test agent, default=100000