COMET not working with python 3.8 about comet

[joncamgui]

Thanks a lot @ricardorei . It works perfectly!

[i-am-nika]

It still doesn't work for Python 3.8...

[mjpost]

Yes, it works (with a few command-line parameter changes). Thank you!

[FayeXXX]

@AnanyaCoder
the same error occours in my python3.7 env, have you figured it out?

[ricardorei]

Is there any plan to make it work for python 3.8?

Yes, I already tested it before in python 3.8 and 3.9. It should support python 3.6, 3.7, 3.8, and 3.9. I'll take a look at it tomorrow. It is probably just a reqs/dependency problem...

[ricardorei]

Btw does the command: comet-score -s src.de -t hyp.en -r ref.en work?

[ShivaliGoel]

Hi, I had the same issue. I switched to Python 3.7 and pandas==1.1.5 to make it work for me. (Installed from pip & running locally- no GPU)

[joncamgui]

Btw does the command: comet-score -s src.de -t hyp.en -r ref.en work?

It does not work either. I get:

File "/usr/local/Cellar/[email protected]/3.8.12/Frameworks/Python.framework/Versions/3.8/lib/python3.8/multiprocessing/reduction.py", line 60, in dump
    ForkingPickler(file, protocol).dump(obj)
AttributeError: Can't pickle local object 'CometModel.predict.<locals>.<lambda>'

[ShivaliGoel]

@joncamgui I got the same error but using 3.7 and pandas==1.1.5 resolved it.

[ricardorei]

I found the error.

I believe during Inference Pytorch Lightning will create model copies and it will complain about the fact that there is a lambda function defines locally: https://github.com/Unbabel/COMET/blob/master/comet/models/base.py#L445

After fixing that error it will complain about the local definition of PredictProgressBar.

I am just wondering why it works for some python versions and not for others.

[mjpost]

I also have an issue with Python 3.8:

$ python3 -V
3.8.10
$ pip list | grep comet
unbabel-comet          0.1.0
$ comet score --model wmt-large-da-estimator-1719 [...]
Traceback (most recent call last):
  File "/usr/local/bin/comet", line 8, in <module>
    sys.exit(comet())
  File "/usr/local/lib/python3.8/dist-packages/click/core.py", line 829, in __call__
    return self.main(*args, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/click/core.py", line 782, in main
    rv = self.invoke(ctx)
  File "/usr/local/lib/python3.8/dist-packages/click/core.py", line 1259, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/usr/local/lib/python3.8/dist-packages/click/core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/usr/local/lib/python3.8/dist-packages/click/core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/comet/cli.py", line 133, in score
    model = load_checkpoint(model) if os.path.exists(model) else download_model(model)
  File "/usr/local/lib/python3.8/dist-packages/comet/models/__init__.py", line 98, in download_model
    return load_checkpoint(checkpoint_path)
  File "/usr/local/lib/python3.8/dist-packages/comet/models/__init__.py", line 131, in load_checkpoint
    model = str2model[hparams["model"]].load_from_checkpoint(
  File "/usr/local/lib/python3.8/dist-packages/pytorch_lightning/core/saving.py", line 157, in load_from_checkpoint
    model = cls._load_model_state(checkpoint, strict=strict, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/pytorch_lightning/core/saving.py", line 199, in _load_model_state
    model = cls(**_cls_kwargs)
  File "/usr/local/lib/python3.8/dist-packages/comet/models/estimators/comet_estimator.py", line 35, in __init__
    super().__init__(hparams)
  File "/usr/local/lib/python3.8/dist-packages/comet/models/estimators/estimator_base.py", line 70, in __init__
    super().__init__(hparams)
  File "/usr/local/lib/python3.8/dist-packages/comet/models/model_base.py", line 134, in __init__
    self.hparams = Namespace(**hparams)
  File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 819, in __setattr__
    object.__setattr__(self, name, value)
AttributeError: can't set attribute

[ricardorei]

Hi @mjpost can you test it using the pre-release version?

pip install unbabel-comet==1.0.0rc8

I fixed some of the problems reported on this thread already.

[ricardorei]

@joncamgui please test it now: pip install unbabel-comet==1.0.0rc8
I fixed the error you reported already

[WilliamTambellini]

Hi @ricardorei
Any plan to make the official 1.0 available in standard 3.8 pip:

pip3.8 install unbabel-comet==1.0.0
ERROR: Could not find a version that satisfies the requirement unbabel-comet==1.0.0 (from versions: 0.0.1, 0.0.2, 0.0.3, 0.0.4, 0.0.6, 0.0.6.post1, 0.0.6.post2, 0.0.7, 0.1.0, 1.0.0rc1, 1.0.0rc2, 1.0.0rc3, 1.0.0rc4, 1.0.0rc5, 1.0.0rc6, 1.0.0rc7, 1.0.0rc8, 1.0.0rc9)
ERROR: No matching distribution found for unbabel-comet==1.0.0

Best
W.

[ricardorei]

@WilliamTambellini I release version 1.0 a few hours ago and I tested it on python 3.8. As you can see your pip should return you 1.0.0 and 1.0.1 versions

The install test for python 3.8 can be seen here.

[ricardorei]

have you tried to upgrade pip?

[ricardorei]

I am reopening this until its fixed.

[AnanyaCoder]

---

ImportError Traceback (most recent call last)
/tmp/ipykernel_33/376669550.py in
----> 1 from comet import download_model, load_from_checkpoint
2 import os
3 import sys
4 #from comet.models import download_model
5 #Load COMET Models before hand

ImportError: cannot import name 'download_model' from 'comet' (/opt/conda/lib/python3.7/site-packages/comet/init.py)