foamliu / Tacotron2-Mandarin

PyTorch reimplementation of Tacotron2 in Mandarin
MIT License
80 stars 28 forks source link

BEST_checkpoint.tar无法解压 #8

Closed ZengHorace closed 4 years ago

ZengHorace commented 4 years ago

前辈,BEST_checkpoint.tar无法解压呀,可以在上传一下嘛

phecda-xu commented 4 years ago

见代码: https://github.com/foamliu/Tacotron2-Mandarin/blob/4fc5825626c804497c2b8ef364784cca8921804a/utils.py#L28

很明显这个模型不是 压缩文件,是torch.save()保存的模型,一般是用'.pth'做后缀的,也可以用其他的后缀,本质上是个checkpoint模型,应该是直接用torch.load()加载的。

ZengHorace commented 4 years ago

试了,报错 File "infer.py", line 33, in waveglow = torch.load(waveglow_path)['model'] File "/root/anaconda3/envs/horace/lib/python3.6/site-packages/torch/serialization.py", line 387, in load return _load(f, map_location, pickle_module, **pickle_load_args) File "/root/anaconda3/envs/horace/lib/python3.6/site-packages/torch/serialization.py", line 574, in _load result = unpickler.load() ModuleNotFoundError: No module named 'models'

foamliu commented 4 years ago

Thanks @phecda-xu , you're right.