A liar's dice agent using Monte Carlo Counterfactual Regret Minimization (MCCFR). The implementation is inspired by Pluribus.
To train the model, run python mccfr.py
. The model will be saved in output/
. You can adjust the training time by changing the parameter in mccfr.py
.