KEMBAR78
[aten decomp] Update sdpa decom by kimishpatel · Pull Request #108371 · pytorch/pytorch · GitHub
Skip to content

[aten decomp] Update sdpa decom #108371

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.

Sign up for GitHub

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

Jump to bottom
Closed
wants to merge 5 commits into from
Closed

[aten decomp] Update sdpa decom #108371

wants to merge 5 commits into from

Conversation

@kimishpatel
Copy link
Contributor

@kimishpatel kimishpatel commented Aug 31, 2023 •
edited
Loading

Stack from ghstack (oldest at bottom):

Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: D48917461

@kimishpatel
[aten decomp] Update sdpa decom
a55a4f5 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

[ghstack-poisoned]
@pytorch-bot
Copy link

pytorch-bot bot commented Aug 31, 2023 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/108371

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 49068e0 with merge base b9fc6d7 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

kimishpatel added a commit that referenced this pull request Aug 31, 2023
@kimishpatel
[aten decomp] Update sdpa decom
522bde0 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

ghstack-source-id: 3262538
Pull Request resolved: #108371
@kimishpatel
Copy link
Contributor Author

talked to @larryliu0820 and realized that this is wrong. Need a different solution

@kimishpatel
Update on "[aten decomp] Update sdpa decom"
190849d 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

[ghstack-poisoned]
@kimishpatel
Update on "[aten decomp] Update sdpa decom"
25f41fd 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

[ghstack-poisoned]
larryliu0820
larryliu0820 approved these changes Sep 1, 2023
View reviewed changes
torch/_decomp/decompositions.py
)
return (
output,
output.transpose(1, 2),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

oh wow this is so weird

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

it is indeed. flash attnetion, https://fburl.com/38pwyabk, does that while the other one does not. I am gonna check if i trigger non flash variant what happens. But surprising calls to contiguous, https://github.com/pytorch/pytorch/blob/main/torch/nn/functional.py#L5441, does not appear in trace. which is also very weird.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it a bug? It would be great to add this explanation in the comment. What's the relationship to decomposition statement in the summary?

@kimishpatel
Copy link
Contributor Author

@kimishpatel has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

@kimishpatel
Update on "[aten decomp] Update sdpa decom"
543ac0e 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: [D48917461](https://our.internmc.facebook.com/intern/diff/D48917461)

[ghstack-poisoned]
kimishpatel added a commit that referenced this pull request Sep 1, 2023
@kimishpatel
[aten decomp] Update sdpa decom
4c26adb 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

ghstack-source-id: 730e479
Pull Request resolved: #108371
@kimishpatel
Copy link
Contributor Author

@kimishpatel has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

@kimishpatel
Update on "[aten decomp] Update sdpa decom"
49068e0 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: [D48917461](https://our.internmc.facebook.com/intern/diff/D48917461)

[ghstack-poisoned]
kimishpatel added a commit that referenced this pull request Sep 2, 2023
@kimishpatel
[aten decomp] Update sdpa decom
b9d4790 
Summary:
Earlier decomp was routing _flash* variant to _match variant and this
was result in failure during torch.export, for some reason that I
couldnt trace.

However, it seems that we should really have a decomp for
scaled_dot_product_attention, instead of
scaled_dot_product_flash_attention. Right?

This diff adds that. Plus it adds a test to check if the model exported
via two stage export, has decomposed the op. This test needs improvement
to figur eout what the core aten opset is and check for anything that is
not inside.

Test Plan:
test_model_exports_to_core_aten

Reviewers:

Subscribers:

Tasks:

Tags:

ghstack-source-id: 44a94e7
Pull Request resolved: #108371
@kimishpatel
Copy link
Contributor Author

@kimishpatel has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Contributor

@pytorchbot merge

(Initiating merge automatically since Phabricator Diff has merged)

@pytorch-bot pytorch-bot bot added the ciflow/trunk Trigger trunk jobs on your pull request label Sep 3, 2023
@pytorchmergebot
Copy link
Collaborator

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging
Check the merge workflow status
here

@facebook-github-bot facebook-github-bot deleted the gh/kimishpatel/178/head branch September 7, 2023 14:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@iseeyuan iseeyuan iseeyuan left review comments

@larryliu0820 larryliu0820 larryliu0820 approved these changes

Assignees

No one assigned

Labels

ciflow/inductor ciflow/trunk Trigger trunk jobs on your pull request Merged

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@kimishpatel @facebook-github-bot @pytorchmergebot @larryliu0820 @iseeyuan