[Pallas] Introduce GMM(torch.autograd.Function) #7152
Merged
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
JackCaoG
reviewed
May 30, 2024
| # Make sure gmm doesn't fallback. | ||
| self.assertNotIn("aten::", met.short_metrics_report()) | ||
|
|
||
| @unittest.skipIf(xr.device_type() != 'TPU', "This test only works on TPU.") |
Collaborator
There was a problem hiding this comment.
do you need TPU version check here?
Collaborator
Author
There was a problem hiding this comment.
lol, good question.
Collaborator
Author
There was a problem hiding this comment.
v2 is pretty happy on the tree.
Collaborator
There was a problem hiding this comment.
interesting.. I thought pallas is not supported on v2.
Collaborator
Author
There was a problem hiding this comment.
gmm is just mm... The kernel is simple... Other fuses softmax, etc...
JackCaoG
reviewed
May 30, 2024
JackCaoG
reviewed
May 30, 2024
JackCaoG
approved these changes
May 30, 2024
Collaborator
Author
|
Thanks Jack for approving. |
Collaborator
Author
|
Skip GPU tests to move fast. |
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Outdated suggestions cannot be applied.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.
Summary:
This pull request make GMM as a torch.autograd.Function such that we can use torch.autograd.backward instead of manual backpropagation.
Test Plan:
python test/test_gmm.py