Fix last checkpoint finding in filtered files with correct extension #17072
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
yassersouri
requested review from
williamFalcon,
awaelchli,
carmocca and
justusschock
as code owners
|
@carmocca Where are the tests for the |
awaelchli
changed the title
|
They are in |
Borda
changed the title
Borda
approved these changes
Nov 18, 2023
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #17072 +/- ##
==========================================
- Coverage 84% 49% -35%
==========================================
Files 443 435 -8
Lines 36154 36007 -147
==========================================
- Hits 30252 17610 -12642
- Misses 5902 18397 +12495 |
for more information, see https://pre-commit.ci
awaelchli
approved these changes
Nov 19, 2023
|
Hi. At the moment I don't have the capacity.
…On Mon, Nov 20, 2023 at 7:22 AM Carlos Mocholí ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In src/lightning/pytorch/callbacks/model_checkpoint.py
<#17072 (comment)>
:
> if self.CHECKPOINT_NAME_LAST in os.path.split(p)[1]
+ and os.path.split(p)[1].endswith(self.FILE_EXTENSION)
@yassersouri <https://github.com/yassersouri> are you still interested in
finishing this PR?
—
Reply to this email directly, view it on GitHub
<#17072 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACGSB55HY4ZSSN2K4WKQDDYFNYRZAVCNFSM6AAAAAAVZ4ZX6SVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMYTONBQGA4DAMRYHE>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Borda
approved these changes
Nov 21, 2023
carmocca
approved these changes
Nov 21, 2023
Borda
pushed a commit
that referenced
this pull request
Dec 19, 2023
…17072) Co-authored-by: awaelchli <[email protected]> (cherry picked from commit 67d3844)
lantiga
pushed a commit
that referenced
this pull request
Dec 20, 2023
…17072) Co-authored-by: awaelchli <[email protected]> (cherry picked from commit 67d3844)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Simple change. The last checkpoint candidate finding code, would look at all the files with "last" in their name.
But we could use a better filtering mechanism to include files that have "last" in their name and have the correct extension.
(since the fix is very simple, I didn't create an issue for it)
I don't think this breaks anything.
What was my issue?
After the training finished, I was saving some other files in the directory with "last" and "best" in their names. These files were tsv, text or other kind of files.
After the save, the testing using
ckpt_path="last"would stop working and I noticed in the logs that one of the files that I had created was being loaded instead of the correct "last.ckpt" file.I have tested locally and with this simple change, this issue for me is resolved.