TypeError: cannot unpack non-iterable NoneType object while importing fairseq

[subnkve479]

Hi Team

I am trying to reproduce indicTrans_python_interface.ipynb. I am not able to import fairseq library .Below is the error I am facing.

I could see same issue is still open in fairseq github repo. I tried installing torch and torchvision packages as mentioned in below link but still I am facing the same issue.

facebookresearch/fairseq#4214

This is issue is blocking me to train my own model as well using IndicTrans_training.ipynb .

I could see the import is successful in indicTrans_python_interface.ipynb. with few warnings.

Below is link to my notebook

https://colab.research.google.com/drive/1e0G_jDe8_0hd-xtj1e4zhwqJQNtE86EC?usp=sharing

Can you please help me here

Regards
Subbu

[subnkve479]

@anoopkunchukuttan, @GokulNC , @sumanthd17 Can you please help me here

Regards
Subbu

[HarryHe11]

@anoopkunchukuttan, @GokulNC , @sumanthd17 Can you please help me here

Regards Subbu

Yeah, I had the same issue. I can't load any translation model. Please check the following error message:

Using cache found in /root/.cache/torch/hub/pytorch_fairseq_main
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
[<ipython-input-24-5863cdfe7acc>](https://localhost:8080/#) in <module>
      2 
      3 # Load an En-Fr Transformer model trained on WMT'14 data :
----> 4 en2fr = torch.hub.load('pytorch/fairseq', 'transformer.wmt14.en-fr', tokenizer='moses', bpe='subword_nmt')
      5 
      6 # Use the GPU (optional):

7 frames
/usr/local/lib/python3.7/dist-packages/torch/hub.py in load(repo_or_dir, model, source, trust_repo, force_reload, verbose, skip_validation, *args, **kwargs)

/usr/local/lib/python3.7/dist-packages/torch/hub.py in _load_local(hubconf_dir, model, *args, **kwargs)

[/usr/local/lib/python3.7/dist-packages/torch/hub.py](https://localhost:8080/#) in _import_module(name, path)
     87     return '[https://github.com/{}/{}/archive/{}.zip](https://github.com/%7B%7D/%7B%7D/archive/%7B%7D.zip)'.format(repo_owner, repo_name, branch)
     88 
---> 89 
     90 def _load_attr_from_module(module, func_name):
     91     # Check if callable is defined in the module

/usr/lib/python3.7/importlib/_bootstrap_external.py in exec_module(self, module)

/usr/lib/python3.7/importlib/_bootstrap.py in _call_with_frames_removed(f, *args, **kwds)

[~/.cache/torch/hub/pytorch_fairseq_main/hubconf.py](https://localhost:8080/#) in <module>
     37 
     38 # only do fairseq imports after checking for dependencies
---> 39 from fairseq.hub_utils import (  # noqa; noqa
     40     BPEHubInterface as bpe,
     41     TokenizerHubInterface as tokenizer,

[~/.cache/torch/hub/pytorch_fairseq_main/fairseq/__init__.py](https://localhost:8080/#) in <module>
     31 hydra_init()
     32 
---> 33 import fairseq.criterions  # noqa
     34 import fairseq.distributed  # noqa
     35 import fairseq.models  # noqa

[~/.cache/torch/hub/pytorch_fairseq_main/fairseq/criterions/__init__.py](https://localhost:8080/#) in <module>
     22     CRITERION_DATACLASS_REGISTRY,
     23 ) = registry.setup_registry(
---> 24     "--criterion", base_class=FairseqCriterion, default="cross_entropy"
     25 )
     26 

TypeError: cannot unpack non-iterable NoneType object

[subnkve479]

Hi Team
Can you please help me on this issue

Regards
Subbu

[HarryHe11]

Hi Team Can you please help me on this issue

Regards Subbu

Hey body, I guess this annoying issue happens because of the fairseq repo instead of the contributors of this repo. I am recently having the same issue. No one has given me any responses at this point.

[subnkve479]

Hi Team Can you please help me on this issue
Regards Subbu

Hey body, I guess this annoying issue happens because of the fairseq repo instead of the contributors of this repo. I am recently having the same issue. No one has given me any responses at this point.

Yes. There are many open issues in fairseq repo regarding this error. But I could see successful run in the notebook shared in this repo.Hence requesting them the workaround they have followed if any