-
Notifications
You must be signed in to change notification settings - Fork 3.9k
Fix test_train
to not rely on 'Sanity Checking' stdout in multi-GPU runs
#10478
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
Open
drivanov
wants to merge
8
commits into
pyg-team:master
Choose a base branch
from
drivanov:sanity_check
base: master
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
Open
Changes from all commits
Commits
Show all changes
8 commits
Select commit
Hold shift + click to select a range
7f62ca2
Fix `test_train` to not rely on 'Sanity Checking' stdout in multi-GPU…
drivanov 510e6a0
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] cca4150
Removing some things that were used for debugging
drivanov 28d1ed2
Removing some things that were used for debugging
drivanov 0e0006f
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] 22c18f0
Updating CHANGELOG.md
drivanov e8f77a8
Merge branch 'sanity_check' of https://github.com/drivanov/pytorch_ge…
drivanov 38cfc51
Update CHANGELOG.md
akihironitta File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or 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
Oops, something went wrong.
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.
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.
If I understand this diff correctly, this
assert not trainer.sanity_checking
tests nothing. Previously, it tested that the code ran sanity checking, but with this PR, it doesn't.sanity_checking
is onlyTrue
when the trainer is running a few validation steps at the start of fit.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.
Can we just remove this assert? Testing whether it performed sanity checking doens't make much sense as a test anyway.
Uh oh!
There was an error while loading. Please reload this page.
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.
This line was suggested by ChatGPT. The following explanation is adapted from its reasoning:
could fail (i.e.,
trainer.sanity_checking == True
aftertrainer.fit()
finishes).What would cause it
If training never moved past the sanity check
trainer.fit()
would return early, andtrainer.sanity_checking
could remain True.If sanity checking was the only phase run
max_steps=0
ormax_epochs=0
, training effectively does nothing after the sanity check.trainer.sanity_checking
might stay True.Misconfigured validation loop
Bug in Lightning
sanity_checking
flag isn’t toggled back (it’s set True at the start of sanity check, False afterward).Normal expectation
trainer.fit()
, training always goes past sanity check, sotrainer.sanity_checking
must be False.⚡ So in short:
It would fail only if
trainer.fit(
) exited abnormally (error, misconfig, or 0 training steps) or Lightning had a bug in resetting the flag.As we all know, ChatGPT isn't always accurate. Unfortunately, in this case, I am unable to assess the accuracy of its claims. Please advise.