1

scripts/train_chengdu_mahjong_model.py

@@ -1,19 +1,24 @@
 import gym
 from stable_baselines3 import PPO
 from src.environment.chengdu_majiang_env import MahjongEnv
+import torch
 
 def train_model():
     # 创建 MahjongEnv 环境实例
     env = MahjongEnv()
 
+    # 检查是否有可用的GPU
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    print(f"使用设备: {device}")
+
     # 使用 PPO 算法训练模型
-    model = PPO("MlpPolicy", env, verbose=1, tensorboard_log="../logs/ppo_mahjong_tensorboard/")
+    model = PPO("MlpPolicy", env, verbose=1, tensorboard_log="../logs/ppo_mahjong_tensorboard/", device=device)
 
     # 训练模型，训练总步数为100000
-    model.learn(total_timesteps=100000)
+    model.learn(total_timesteps=100)
 
     # 保存训练后的模型
-    model.save("ppo_mahjong_model")
+    model.save("../models/ppo_mahjong_model")
 
     # 测试模型
     obs = env.reset()