python3 UnicodeDecodeError: 'ascii' codec can't decode byte 0xe9 in position 0

源码

    if os.path.isfile(model_path):
            print('==> loading checkpoint {}'.format(args.resume))
            checkpoint = torch.load(model_path)
            start_epoch = checkpoint['epoch']
            net.load_state_dict(checkpoint['net'])
    else:
            print('==> no checkpoint found at {}'.format(args.resume))

问题

Traceback (most recent call last):
  File "train_img_model_xent.py", line 251, in <module>
    main()
  File "train_img_model_xent.py", line 136, in main
    checkpoint = torch.load(model_path)
  File "C:\Users\Liang\Anaconda3\lib\site-packages\torch\serialization.py", line 268, in load
    return _load(f, map_location, pickle_module)
  File "C:\Users\Liang\Anaconda3\lib\site-packages\torch\serialization.py", line 421, in _load
    result = unpickler.load()
    
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe9 in position 0: ordinal not in range(128)

原因

python版本不兼容，源码python版本低，现有环境python版本高
python3 取消了reload，在编码解码方法做了很大的调整

解决办法

按照要求打开 .\torch\serialization.py 找出错误代码段

_sys_info = pickle_module.load(f)
    unpickler = pickle_module.Unpickler(f)
    unpickler.persistent_load = persistent_load
    result = unpickler.load()

torch库函数 serialization.py 改成

_sys_info = pickle_module.load(f, encoding='iso-8859-1')  # 改动处
    unpickler = pickle_module.Unpickler(f, encoding='iso-8859-1')  # 改动处
    unpickler.persistent_load = persistent_load
    result = unpickler.load()

源码 改成

 if os.path.isfile(model_path):
            print('==> loading checkpoint {}'.format(args.resume))
            checkpoint = torch.load(model_path)
            start_epoch = checkpoint['epoch']
            net.load_state_dict(checkpoint['net'],strict=False)  # 改动处
 else:
            print('==> no checkpoint found at {}'.format(args.resume))

注意事项

如果不改源码 *.load_state_dict（）会出错，如下

Traceback (most recent call last):
  File "E:/wj-lab/expStad/test.py", line 79, in <module>
    net.load_state_dict(checkpoint['net'])
  File "D:\Anaconda3\envs\python35\lib\site-packages\torch\nn\modules\module.py", line 721, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for embed_net:
	Missing key(s) in state_dict: * * *
	Unexpected key(s) in state_dict: "visible_net.visible.bn1.num_batches_tracked", "visible_net.visible.layer1.0.bn1.n