Flair: Error using flair embeddings: ValueError: invalid literal for int() with base 10: '0+cu101'

[ChessMateK]

Describe the bug
When I run this:

embedding_types: List[TokenEmbeddings] = [               
    FlairEmbeddings('news-forward'),
    FlairEmbeddings('news-backward')
]

I obtain the following errors:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-4-177704bfc5f6> in <module>()
---> 51     FlairEmbeddings('news-forward'),
     52     FlairEmbeddings('news-backward')

5 frames
/usr/local/lib/python3.6/dist-packages/flair/embeddings.py in __init__(self, model, fine_tune, chars_per_chunk)
   1986             self.name = f"Task-LSTM-{self.lm.hidden_size}-{self.lm.nlayers}-{self.lm.is_forward_lm}"
   1987         else:
-> 1988             self.lm: LanguageModel = LanguageModel.load_language_model(model)
   1989             self.name = str(model)
   1990 

/usr/local/lib/python3.6/dist-packages/flair/models/language_model.py in load_language_model(cls, model_file)
    195             state["embedding_size"],
    196             state["nout"],
--> 197             state["dropout"],
    198         )
    199         model.load_state_dict(state["state_dict"])

/usr/local/lib/python3.6/dist-packages/flair/models/language_model.py in __init__(self, dictionary, is_forward_lm, hidden_size, nlayers, embedding_size, nout, dropout)
     59 
     60         # auto-spawn on GPU if available
---> 61         self.to(flair.device)
     62 
     63     def init_weights(self):

/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py in to(self, *args, **kwargs)
    441             return t.to(device, dtype if t.is_floating_point() else None, non_blocking)
    442 
--> 443         return self._apply(convert)
    444 
    445     def register_backward_hook(self, hook):

/usr/local/lib/python3.6/dist-packages/flair/models/language_model.py in _apply(self, fn)
    392     def _apply(self, fn):
    393         major, minor, build, *_ = (int(info)
--> 394                                 for info in torch.__version__.split('.'))
    395 
    396         # fixed RNN change format for torch 1.4.0

/usr/local/lib/python3.6/dist-packages/flair/models/language_model.py in <genexpr>(.0)
    392     def _apply(self, fn):
    393         major, minor, build, *_ = (int(info)
--> 394                                 for info in torch.__version__.split('.'))
    395 
    396         # fixed RNN change format for torch 1.4.0

ValueError: invalid literal for int() with base 10: '0+cu101'

To Reproduce
Follow this flair guide for NER and use flair embeddings

Expected behavior
Embeddings are stacked without errors.

Screenshots
None

Environment (please complete the following information):

• Google Colab

Additional context
None

[iam-kevin]

This issue was solved for me when I used:

!pip install --upgrade git+https://github.com/flairNLP/flair.git

[alanakbik]

Yes, this issue is fixed in master branch. Fix is part of next release.

[alanakbik]

Closing but feel free to reopen if you have more questions.

[abeermohamed1]

this issue generated again whenever i use PooledFlairEmbeddings. Any help please as the above fix doesn't work with me.