PyTorch学习笔记：模型保存和加载方法

# 方法一：保存整个模型
torch.save(mlp1, "model_saved/mlp1.pkl")
mlp1load = torch.load("model_saved/mlp1.pkl")  # 导入保存的模型
print(mlp1load)
print(mlp1load.hidden2.weight)

# 方法二：只保存模型的参数
torch.save(mlp1.state_dict(), "model_saved/mlp1_param.pkl")
mlp1load_param = torch.load("model_saved/mlp1_param.pkl")  # 导入保存的模型的参数
print(mlp1load_param)

代码实际运行中遇到报错及解决方法可见以下笔记：

_pickle.PicklingError: Can't pickle <class '__main__.MLPmodel'>: attribute lookup MLPmodel on __main__ failed