The code is clear while the README file could be more implemented.
I have seen that you make your reinforcement learning player plays always as first in each match, this aspect could change the final results in favour of your RL player.
Anyway the RL player is well trained especially thanks to the "clever_game" function, I still prefer the random way in order to train the player so it starts with zero knowledge about the game.
One more advice, you could improve the way you show the results with some more informations and also a graph.
Hi Ludovico, I took a look to your code.
The code is clear while the README file could be more implemented.
I have seen that you make your reinforcement learning player plays always as first in each match, this aspect could change the final results in favour of your RL player. Anyway the RL player is well trained especially thanks to the "clever_game" function, I still prefer the random way in order to train the player so it starts with zero knowledge about the game. One more advice, you could improve the way you show the results with some more informations and also a graph.
Good job! Keep pushing!