AlphaGomoku is a Gomoku AI based on Alpha Zero's algorithm.
Chinese: AlphaGomoku 是個基於Alpha Zero 演算法演算法的五子棋AI。
After cloning the project, training can be started by running model_training/agc_training.py. Testing of trained models can be done in playground.py.
Please note that the training may take while, because it requires at least a 4000+ games to achieve reasonable results. It is also recommended to train on a GPU.
Chinese: 將專案clone下來後,執行model_training/agc_training.py 即可開始訓練。若訓練好想要測試結果,可在執player都設定好後,執行playground.py。
訓練過程可能會花上較長一段時間,因為AI需訓練大約4000+場後才會逐漸有成果。建議使用GPU來跑訓練。
https://drive.google.com/file/d/15E1Rx6rJS6t5jMG6L_YsadBM39MD6T5X/view?usp=sharing