[ONNX] Support ONNXFakeContext with op_level_debug #105874
Closed
Conversation
This file contains 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]
titaiwangms
requested review from
BowenBao,
abock and
thiagocrepaldi
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/105874
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f76d61c: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
release notes: onnx
titaiwangms
added a commit
that referenced
this pull request
Jul 24, 2023
ghstack-source-id: 0c0e27f87b8a70ae171d5642d755b1637b6c7ba9 Pull Request resolved: #105874
titaiwangms
added
module: onnx
titaiwangms
commented
Jul 24, 2023
thiagocrepaldi
approved these changes
Jul 24, 2023
BowenBao
reviewed
Jul 24, 2023
BowenBao
approved these changes
Jul 24, 2023
justinchuby
reviewed
Jul 24, 2023
| list_int_shape.append(symbolic_shapes.hint_int(dim)) | ||
| else: | ||
| raise ValueError( | ||
| f"Unexpected SymInt found in shape. SymInt: {dim}; " |
There was a problem hiding this comment.
Is this unexpected, or is it a case we can expect but cannot handle? Just to make sure for clarity.
There was a problem hiding this comment.
nice catch, overlooked this one. Please reveal keyword "unbacked", will easily lead to many helpful resources such as this https://pytorch-dev-podcast.simplecast.com/episodes/unbacked-symints-oyqu0_P6
Previous to the PR, op_level_debug doesn't support OnnxFakeConext because it relies on real tensor in args to do shape type inference propagation in fx graph to get static shapes helping simulating the op args/kwargs. However, OnnxFakeContext will fakify the args/kwargs at the very begining, so the op_level_debug can't have the static_shapes to utilize. This PR uses SymInt API: `has_hint` and `hint_int` to fully replace the functionality of shape type inference propagation. The static shapes are obtained through SymInt. Therefore, the pass `ShapeInferenceWithFakeTensor` is eliminated. Also moved the args/kwargs processing into op_validation to live under the rule `op_level_debug`. [ghstack-poisoned]
titaiwangms
added a commit
that referenced
this pull request
Jul 25, 2023
ghstack-source-id: 19a82b4636e3400e93aa498108549b3be7a6c864 Pull Request resolved: #105874
|
@pytorchbot merge |
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 |
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):
Previous to the PR, op_level_debug doesn't support OnnxFakeConext because it relies on real tensor in args to do shape type inference propagation in fx graph to get static shapes helping simulating the op args/kwargs. However, OnnxFakeContext will fakify the args/kwargs at the very begining, so the op_level_debug can't have the static_shapes to utilize.
This PR uses SymInt API:
has_hintandhint_intto fully replace the functionality of shape type inference propagation. The static shapes are obtained through SymInt. Therefore, the passShapeInferenceWithFakeTensoris eliminated.Also moved the args/kwargs processing into op_validation to live under the rule
op_level_debug.