An implementation of Deep Q learning with Pytorch with the combination of game theory for Pyrat - a 1vs1 game created by Telecom Bretagne for learning purpose. Win rate: 82% vs defaut machine greedy algorithm.
minhIMT97/Reinforcement-Learning-for-Game
A project that uses reinforcement learning to train the agent in a game called pyrat, which is developed by Telecom Bretagne.
Python