Make test_torchbind.py training IR compatible #138658
Closed
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
[ghstack-poisoned]
tugsbayasgalan
requested review from
angelayi,
avikchaudhuri,
ydwu4 and
zhxchen17
as code owners
This was referenced Oct 23, 2024
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/138658
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (1 Unrelated Failure)As of commit 8b114b1 with merge base 924e726 ( FLAKY - The following job failed but was likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
ydwu4
approved these changes
Oct 23, 2024
test/export/test_torchbind.py
Outdated
| return (getitem, sin)""", # noqa: B950 | ||
| ) | ||
|
|
||
| @unittest.expectedFailure # T205481814 |
Contributor
There was a problem hiding this comment.
Not sure what's going on here. Other parts looks good.
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. [ghstack-poisoned]
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. [ghstack-poisoned]
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
1 similar comment
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
1 similar comment
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to the serializer. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to the serializer. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
pianpwk
reviewed
Nov 1, 2024
torch/export/_trace.py
Outdated
|
|
||
| from torch._export.verifier import TrainingIRVerifier | ||
|
|
||
| print("GRAPH", gm.graph) |
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to the serializer. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to the serializer. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to the serializer. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to the serializer. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) [ghstack-poisoned]
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
1 similar comment
Contributor
Author
|
@tugsbayasgalan has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
tugsbayasgalan
commented
Nov 3, 2024
| # as input, runtime assertion should fail. This is because we would create | ||
| # guard on y.shape[0] > x.shape[0] but somehow in old export, we dce this | ||
| # assertion. | ||
| if is_training_ir_test(self._testMethodName) and is_non_strict_test( |
zhxchen17
approved these changes
Nov 4, 2024
Contributor
|
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
Collaborator
Merge startedYour 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 |
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 4, 2024
Differential Revision: [D65158939](https://our.internmc.facebook.com/intern/diff/D65158939) Pull Request resolved: #139209 Approved by: https://github.com/ydwu4 ghstack dependencies: #138658
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 4, 2024
Differential Revision: [D65282662](https://our.internmc.facebook.com/intern/diff/D65282662) Pull Request resolved: #139233 Approved by: https://github.com/kwen2501 ghstack dependencies: #138658, #139209
Ryo-not-rio
pushed a commit
to Ryo-not-rio/pytorch
that referenced
this pull request
Dec 2, 2024
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs: 1. ep.module() doesn't register all aliased constants in the module. 2. When we retrace, we need to fakify the original Torchbind object. 3. We don't run any DCE on training IR so we need to add some more torch ops to verifier. Differential Revision: [D64853530](https://our.internmc.facebook.com/intern/diff/D64853530) Pull Request resolved: pytorch#138658 Approved by: https://github.com/ydwu4, https://github.com/zhxchen17
Ryo-not-rio
pushed a commit
to Ryo-not-rio/pytorch
that referenced
this pull request
Dec 2, 2024
Differential Revision: [D65158939](https://our.internmc.facebook.com/intern/diff/D65158939) Pull Request resolved: pytorch#139209 Approved by: https://github.com/ydwu4 ghstack dependencies: pytorch#138658
Ryo-not-rio
pushed a commit
to Ryo-not-rio/pytorch
that referenced
this pull request
Dec 2, 2024
Differential Revision: [D65282662](https://our.internmc.facebook.com/intern/diff/D65282662) Pull Request resolved: pytorch#139233 Approved by: https://github.com/kwen2501 ghstack dependencies: pytorch#138658, pytorch#139209
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.
Stack from ghstack (oldest at bottom):
In this diff, i make test_torchbind.py tests to handle training IR. Today in the training IR, we don't see the effect token and HOP because this happens at the FunctionalTensorMode. Maybe in the future, we should move this logic up to the training IR so that writing passes etc on training Ir is safer. But for the migration purposes, i think it is ok for now. I also fixed two bugs:
Differential Revision: D64853530