Skip to content
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


Literature card game implementation:

Start with python3 -i Built for Python 3.6.0.

See how to train a model to play against with python3 -h. Play against a model that I trained with learning.play_against_model('model_10000.out').

Current limitations:

  • The bots only consider asking for a Card that they know a Player does not possess in the case that there are no other possible Moves. I made this simplification because the initial training took too long otherwise.
  • The Players consider what other Players know about them, but they don't consider any levels further than that, e.g., the Players don't consider what other Players know that other Players know about them.
    • There is definitely information there that is not represented in the current state. For example, player 0 might have all of the minor hearts except for the 5 of hearts. If player 1 asks for a minor hearts, player 0 knows that player 1 does not know that player 0 knows that player 1 has the 5 of hearts. Alternately, if player 0 had previously revealed that they have every other minor heart, then player 0 would know that player 1 knows that player 0 knows what card they have.
    • I chose not to represent this because it vastly increases the dimensionality of the problem, and I don't think that the information is particularly valuable, even though it does have strategy implications.
  • During training, the bots will occasionally get caught in an infinite loop. To mitigate this, I add noise to the scores for each move and kill games after 200 moves.
  • I'm only training the bots for games of four right now. The code can be easily adapted to work for a different number of players.
You can’t perform that action at this time.