---

一、现象

报错显示

    s = torch.from_numpy(padding_seq([s])).cuda().long()  
  File "D:\Anaconda3\lib\site-packages\torch\cuda\__init__.py", line 221, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

二、解决方案

将

s = torch.from_numpy(padding_seq([s])).cuda().long()

改成

s = torch.from_numpy(padding_seq([s])).long()

报错即可解决

---

解决方案：网上各种资料尝试一番