pretrained model can't be read

[kpis-msa]

I used these source code on google colab by porting. But these codes can't read the pretrained model. The error message is as follows:

`

AttributeError Traceback (most recent call last) in () 208 "dont_smooth": 'smooth'}) 209 # Load model --> 210 checkpoint = torch.load(args.model, map_location=str(device)) 211 decoder = checkpoint['decoder'] 212 decoder = decoder.to(device)

1 frames /usr/local/lib/python3.6/dist-packages/torch/serialization.py in _legacy_load(f, map_location, pickle_module, **pickle_load_args) 772 unpickler = pickle_module.Unpickler(f, **pickle_load_args) 773 unpickler.persistent_load = persistent_load --> 774 result = unpickler.load() 775 776 deserialized_storage_keys = pickle_module.load(f, **pickle_load_args)

AttributeError: Can't get attribute 'Encoder' on <module 'models' (namespace)> `

Please provide a correct pretrain model.

[adib0073]

@sgrvinod @kpis-msa - Getting a similar issue. Is there a solution to this? Migrating to an old pytorch version will work here?

    772     unpickler = pickle_module.Unpickler(f, **pickle_load_args)
    773     unpickler.persistent_load = persistent_load
--> 774     result = unpickler.load()
    775 
    776     deserialized_storage_keys = pickle_module.load(f, **pickle_load_args)

ModuleNotFoundError: No module named 'models'

[adib0073]

It seems that models.py is required to be in the same folder as the trained model while loading the pytorch model in the old way. This link was helpful for me: https://github.com/pytorch/pytorch/issues/3678