Restrict PyTorch version to <2.3.0 to resolve import error #577
Conversation
Addressing the import error encountered with PyTorch 2.3.0 as detailed in issue bigscience-workshop#576. fixes bigscience-workshop#576
|
Tested this PR locally, seems to have fixed the issue for me! 😁 |
|
Yep! Even I have tested in my local. It's working fine. |
setup.cfg
Outdated
| @@ -32,7 +32,7 @@ package_dir = | |||
| packages = find: | |||
| python_requires = >=3.8 | |||
| install_requires = | |||
| torch>=1.12 | |||
| torch==2.2.2 | |||
There was a problem hiding this comment.
Thanks for the contribution! The requirement seems to be overly restrictive though (it will force users with lower versions of PyTorch to update), can you change it to inequality constraints?
There was a problem hiding this comment.
Thank you, that's right! Updated PyTorch version constraint in the cfg to torch>=1.12,<2.3.0 to have compatibility with earlier versions while avoiding the issue introduced in 2.3.0. Please review the updated pull request.
Modified the version constraint for PyTorch in setup.cfg to torch>=1.12,<2.3.0 to avoid the import errors introduced in version 2.3.0 while still supporting earlier compatible versions. This change follows feedback from @mryab to allow flexibility for users on different versions.
There was a problem hiding this comment.
Modified the version constraint for PyTorch in setup.cfg to torch>=1.12,<2.3.0 to avoid the import errors introduced in version 2.3.0 while still supporting earlier compatible versions. This change follows feedback from @mryab to allow flexibility for users on different versions.
mryab
changed the title
Fixes #576
This addresses the import error encountered with PyTorch 2.3.0 as detailed in issue #576. The error 'cannot import name '_refresh_per_optimizer_state' from 'torch.cuda.amp.grad_scaler' is resolved by pinning the PyTorch version to 2.2.2 in setup.cfg.
Small change:
setup.cfgto specifytorch==2.2.2underinstall_requires.