[Tests] clean up and refactor gradient checkpointing tests #9494
Conversation
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
sayakpaul
commented
Sep 23, 2024
Comment on lines
+341
to
+343
| @unittest.skip( | ||
| "Gradient checkpointing is supported but this test doesn't apply to this class because it's forward is a bit different from the rest." | ||
| ) |
There was a problem hiding this comment.
Other forwards don't apply scaling and unscaling like it does. So, that makes this a little different.
|
@yiyixuxu WDYT? |
|
@DN6 could you give this a look? |
|
@DN6 a gentle ping here. |
DN6
approved these changes
Oct 31, 2024
|
Failing test is unrelated. |
a-r-r-o-w
pushed a commit
that referenced
this pull request
Nov 1, 2024
* check. * fixes * fixes * updates * fixes * fixes
sayakpaul
added a commit
that referenced
this pull request
Dec 23, 2024
* check. * fixes * fixes * updates * fixes * fixes
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Outdated suggestions cannot be applied.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?
Gradient checkpointing is an essential component of model training. We need to ensure it's implemented properly.
If we had them implemented properly we could have avoided the fix from #9489 beforehand. Another related issue: #9496.
Additionally, I took the liberty of properly skipping the related tests with "unittest.skip". This way, we know that tests are being actually skipped.