Avoid data-dependent errors in NJT tests via capture_scalar_outputs=True #144588
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]
This was referenced Jan 10, 2025
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/144588
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (3 Unrelated Failures)As of commit abb5e73 with merge base 803017f ( FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This was referenced Jan 10, 2025
soulitzer
reviewed
Jan 11, 2025
| @ops( | ||
| [op for op in njt_op_db if op.supports_njt], | ||
| allowed_dtypes=(torch.float32,), | ||
| ) | ||
| @torch._dynamo.config.patch(capture_dynamic_output_shape_ops=True) | ||
| # needed to avoid "data dependent operator: aten._local_scalar_dense.default" | ||
| @torch._dynamo.config.patch(capture_scalar_outputs=True) |
There was a problem hiding this comment.
Yep saw these changes as part of the narrow PR as well haha
Though one reason for not doing it by default for all ops is if we want to get signal on which ops need which flags to work with compile, but probably not worth the hassle.
There was a problem hiding this comment.
I think what I want to do is have a section of the upcoming NJT docs dedicated to torch.compile-based stuff, and mention the trick to add this flag as needed for data-dependent operators.
Yep saw these changes as part of the narrow PR as well haha
ah yeah, sadly that's part of a different stack. I shouldn't have done that :x
…r_outputs=True" Part of my BE project addressing NJT bugs surfaced via OpInfo tests. There are several xfails related to data-dependent errors in torch.compile. This PR sets `torch._dynamo.config.capture_scalar_outputs=True` to avoid these, which tends to exercise unbacked SymInt logic and will require `torch._check()`-related fixes. [ghstack-poisoned]
…r_outputs=True" Part of my BE project addressing NJT bugs surfaced via OpInfo tests. There are several xfails related to data-dependent errors in torch.compile. This PR sets `torch._dynamo.config.capture_scalar_outputs=True` to avoid these, which tends to exercise unbacked SymInt logic and will require `torch._check()`-related fixes. [ghstack-poisoned]
…r_outputs=True" Part of my BE project addressing NJT bugs surfaced via OpInfo tests. There are several xfails related to data-dependent errors in torch.compile. This PR sets `torch._dynamo.config.capture_scalar_outputs=True` to avoid these, which tends to exercise unbacked SymInt logic and will require `torch._check()`-related fixes. [ghstack-poisoned]
…r_outputs=True" Part of my BE project addressing NJT bugs surfaced via OpInfo tests. There are several xfails related to data-dependent errors in torch.compile. This PR sets `torch._dynamo.config.capture_scalar_outputs=True` to avoid these, which tends to exercise unbacked SymInt logic and will require `torch._check()`-related fixes. [ghstack-poisoned]
…r_outputs=True" Part of my BE project addressing NJT bugs surfaced via OpInfo tests. There are several xfails related to data-dependent errors in torch.compile. This PR sets `torch._dynamo.config.capture_scalar_outputs=True` to avoid these, which tends to exercise unbacked SymInt logic and will require `torch._check()`-related fixes. [ghstack-poisoned]
|
@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 |
nWEIdia
pushed a commit
to nWEIdia/pytorch
that referenced
this pull request
Jan 27, 2025
…rue (pytorch#144588) Part of my BE project addressing NJT bugs surfaced via OpInfo tests. There are several xfails related to data-dependent errors in torch.compile. This PR sets `torch._dynamo.config.capture_scalar_outputs=True` to avoid these, which tends to exercise unbacked SymInt logic and will require `torch._check()`-related fixes. Pull Request resolved: pytorch#144588 Approved by: https://github.com/soulitzer ghstack dependencies: pytorch#144586, pytorch#144587
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):
Part of my BE project addressing NJT bugs surfaced via OpInfo tests.
There are several xfails related to data-dependent errors in torch.compile. This PR sets
torch._dynamo.config.capture_scalar_outputs=Trueto avoid these, which tends to exercise unbacked SymInt logic and will requiretorch._check()-related fixes.