给出导入四个.pth文件的具体代码

假设您有四个.pth文件分别为file1.pth、file2.pth、file3.pth和file4.pth，以下是导入这四个.pth文件的代码：

import torch

# 加载 file1.pth 文件
checkpoint1 = torch.load('file1.pth')
model_state_dict1 = checkpoint1['model_state_dict']
optimizer_state_dict1 = checkpoint1['optimizer_state_dict']
epoch1 = checkpoint1['epoch']
step1 = checkpoint1['step']

# 加载 file2.pth 文件
checkpoint2 = torch.load('file2.pth')
model_state_dict2 = checkpoint2['model_state_dict']
optimizer_state_dict2 = checkpoint2['optimizer_state_dict']
epoch2 = checkpoint2['epoch']
step2 = checkpoint2['step']

# 加载 file3.pth 文件
checkpoint3 = torch.load('file3.pth')
model_state_dict3 = checkpoint3['model_state_dict']
optimizer_state_dict3 = checkpoint3['optimizer_state_dict']
epoch3 = checkpoint3['epoch']
step3 = checkpoint3['step']

# 加载 file4.pth 文件
checkpoint4 = torch.load('file4.pth')
model_state_dict4 = checkpoint4['model_state_dict']
optimizer_state_dict4 = checkpoint4['optimizer_state_dict']
epoch4 = checkpoint4['epoch']
step4 = checkpoint4['step']

以上代码中，我们使用torch.load()方法加载.pth文件，然后从检查点中提取模型和优化器的状态，以及训练的epoch和step。您可以根据自己的需要更改变量名称。