pytorch-lightning 1.7.6 conda package is broken #14765
Comments
deGravity
added
bug
|
Hi, which tensorboard version got installed for you? I can't reproduce the same error, but something is broken. I got (tested py 3.9 and 3.10): >>> import pytorch_lightning
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/site-packages/pytorch_lightning/__init__.py", line 34, in <module>
from pytorch_lightning.callbacks import Callback # noqa: E402
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/site-packages/pytorch_lightning/callbacks/__init__.py", line 14, in <module>
from pytorch_lightning.callbacks.callback import Callback
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/site-packages/pytorch_lightning/callbacks/callback.py", line 25, in <module>
from pytorch_lightning.utilities.types import STEP_OUTPUT
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/site-packages/pytorch_lightning/utilities/types.py", line 148, in <module>
class DistributedDataParallel(Protocol):
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/site-packages/pytorch_lightning/utilities/types.py", line 156, in DistributedDataParallel
process_group: Optional[ProcessGroup] = None,
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/typing.py", line 277, in inner
return func(*args, **kwds)
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/typing.py", line 354, in __getitem__
return self._getitem(self, parameters)
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/typing.py", line 478, in Optional
arg = _type_check(parameters, f"{self} requires a single type.")
File "/Users/adrian/miniconda3/envs/conda-pl39/lib/python3.9/typing.py", line 166, in _type_check
raise TypeError(f"{msg} Got {arg!r:.100}.")
TypeError: typing.Optional requires a single type. Got Ellipsis. |
awaelchli
added
priority: 0
|
I get tensorboard 2.10.0 and setuptools 65.3.0 (I think those are the relevant packages in my error) Here's the full package list from |
|
I realized that I was using mamba instead of conda to create the environment. When I use conda, the error does not occur. The biggest difference in the packages installed with conda instead of mamba is that pytorch is 1.12.1 instead of 1.10.2: Package list when installed using conda |
|
Also what python and OS are you using? |
|
Ubuntu under WSL2 on Windows. Python 3.10 |
|
Hi, @deGravity, I have to say, I am unable to reproduce this - I tried installing a fresh environment using The Mamba and Conda were installed using mamba-forge as of today, micromamba was also installed today, below is Conda base envProduced environmentsConda envMamba envMicromamba envHowever!The real issue is something else - installing There are numer of weird reasons how did this happen:
If you specifically request this broken environment ( My suggestion is to do either of these:
Hope this helps! |
|
I'm closing this issue as it's not really an issue with Lightning. However, something weird was going on with @awaelchli's installation. If you could please provide me with your |
|
@otaj here they are: conda listconda list -n base |
12 tasks
|
I found out the issue happening to @awaelchli on 1.7.6, this will be affecting all users with |
Red54
added a commit
to Red54/stable-diffusion-webui
that referenced
this issue
Feb 24, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
First check
Bug description
Pytorch lightning cannot be successfully imported after a fresh conda install due to a mismatch between tensorboard and distutils versions.
I tried going back through versions, and the latest it works on is 1.7.2.
How to reproduce the bug
Then try to
import pytorch_lightningError messages and logs
Important info
More info
No response
The text was updated successfully, but these errors were encountered: