-
Notifications
You must be signed in to change notification settings - Fork 3.6k
Specify packaging version to be more than 17.0 #8030
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
Older version of packaging causes an error when saving model checkpoint. The error is in the atomic_save: `if Version(torch.__version__).release[:3] == (1, 6, 0):` For old version of packaging (e.g., 16.8) the call produce the following errror: `AttributeError: 'Version' object has no attribute 'release'`
Codecov Report
@@ Coverage Diff @@
## master #8030 +/- ##
=======================================
- Coverage 92% 87% -5%
=======================================
Files 210 210
Lines 13576 13576
=======================================
- Hits 12449 11807 -642
- Misses 1127 1769 +642 |
carmocca
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
awaelchli
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
ship it!
|
@awaelchli @Borda @carmocca @justusschock any suggestion of why some tests is failing? |
Older version of the
packagingmodule cause an error when saving model checkpoint.The error is in the atomic_save:
if Version(torch.__version__).release[:3] == (1, 6, 0):For old version of packaging (e.g., 16.8) the call produce the following errror:
AttributeError: 'Version' object has no attribute 'release'What does this PR do?
Specify packaging version to be grater or equal to 17.0 in order to avoid bug in the atomic checkpoints save
I tested packing version from 17.0 to 20.9 (latest at the time of writing) and the bug is fixed
How to reproduce the bug
or reproducing the call at line: https://github.com/PyTorchLightning/pytorch-lightning/blob/master/pytorch_lightning/utilities/cloud_io.py#L61