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
Fixes the checkpoint directory structure for pytorch and pytorch lightning #3362
Conversation
Signed-off-by: Kamal Sharma <kamalbhardwaj020@gmail.com>
24fb500
to
05149d2
Compare
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.
minor formatting, can't say much about the logic
Unit Test Results (with flaky tests) 901 files + 11 901 suites +11 9h 46m 46s ⏱️ + 30m 59s For more details on these failures and errors, see this check. Results for commit 06f1c4a. ± Comparison against base commit b96ecae. ♻️ This comment has been updated with latest results. |
Signed-off-by: Kamal Sharma <kamalbhardwaj020@gmail.com>
9a7c45e
to
5a5642d
Compare
Signed-off-by: Kamal Sharma <kamalbhardwaj020@gmail.com>
52055b2
to
460fee0
Compare
@EnricoMi resolved the comments, please take a look |
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.
left a question.
Signed-off-by: Kamal Sharma <kamalbhardwaj020@gmail.com>
9c109b6
to
06f1c4a
Compare
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.
LGTM
fsspec version requirement introduced from this commit |
Checklist before submitting
Description
Fixes #3268. Creating a
copy()
method to replaceput()
while syncing from local temporary directory to run path.put()
is inconsistent with cp and currently has no mechanism to paste the contents from source directory to target without pasting the source directory itself if the target directory is already present, this case is unwanted for horovod and replacing this creates consistency for checkpoints path for both pytorch and pytorch lightning.