KEMBAR78
[Dynamo] allow dynamic callables on tensor variables by mlazos · Pull Request #137940 · pytorch/pytorch · GitHub
Skip to content

Conversation

@mlazos
Copy link
Contributor

@mlazos mlazos commented Oct 14, 2024

@pytorch-bot
Copy link

pytorch-bot bot commented Oct 14, 2024

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/137940

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

✅ No Failures

As of commit 052e6b8 with merge base bf1b8ad (image):
💚 Looks good so far! There are no failures yet. 💚

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

@anijain2305
Copy link
Contributor

Seems this needs more work.

@mlazos
Copy link
Contributor Author

mlazos commented Oct 18, 2024

Seems this needs more work.

Yeah this broke the world LOL
I'm gonna dig more into this next week.

@xinyu-intel
Copy link
Contributor

@mlazos any updates?

@mlazos
Copy link
Contributor Author

mlazos commented Oct 29, 2024

@mlazos any updates?

Still digging into this - do you absolutely need callable attributes? or are normal attributes okay. Callables are quite a bit more work.

@majian4work
Copy link
Contributor

@mlazos any updates?

Still digging into this - do you absolutely need callable attributes? or are normal attributes okay. Callables are quite a bit more work.

hi @mlazos
the reproducer is extracted from deepspeed, so the callable attribute is needed.

@mlazos
Copy link
Contributor Author

mlazos commented Nov 5, 2024

@mlazos any updates?

Still digging into this - do you absolutely need callable attributes? or are normal attributes okay. Callables are quite a bit more work.

hi @mlazos the reproducer is extracted from deepspeed, so the callable attribute is needed.

Ok, I will get this to work then

@majian4work
Copy link
Contributor

hi @mlazos
I have tried your patch. it's ok for user defined dynamic callable attr.
but for builtin methods like "x.addcmul_", it doesn't work.
it's weird that the check is_aten_op_or_tensor_method(obj) failed when obj is x.addcmul_.

@mlazos mlazos force-pushed the mlazos/dyn-callable branch from 83dace3 to 3ddee83 Compare November 7, 2024 00:23
@mlazos
Copy link
Contributor Author

mlazos commented Nov 7, 2024

is_aten_op_or_tensor_method

Thanks for investigating a bit, you were right! I think I have this working with the latest updates.

@mlazos mlazos force-pushed the mlazos/dyn-callable branch from 1c3d47d to 09c63c5 Compare November 7, 2024 08:20
@mlazos
Copy link
Contributor Author

mlazos commented Nov 8, 2024

@pytorchbot merge

@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

@pytorchmergebot
Copy link
Collaborator

Merge failed

Reason: 1 jobs have failed, first few of them are: trunk / macos-py3-arm64 / test (default, 2, 3, macos-m1-stable)

Details for Dev Infra team Raised by workflow job

@mlazos
Copy link
Contributor Author

mlazos commented Nov 8, 2024

@pytorchbot merge

@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

@pytorchmergebot
Copy link
Collaborator

Merge failed

Reason: 1 mandatory check(s) failed. The first few are:

Dig deeper by viewing the failures on hud

Details for Dev Infra team Raised by workflow job

Failing merge rule: Core Maintainers

@mlazos
Copy link
Contributor Author

mlazos commented Nov 8, 2024

@pytorchbot merge

@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

pobin6 pushed a commit to pobin6/pytorch that referenced this pull request Dec 5, 2024
@github-actions github-actions bot deleted the mlazos/dyn-callable branch December 9, 2024 02:15
aostrowski-hbn pushed a commit to HabanaAI/pytorch-fork that referenced this pull request Jan 7, 2025
cherry-pick: pytorch#137940
Fixes pytorch#134844

Pull Request resolved: pytorch#137940
Approved by: https://github.com/williamwen42

Change-Id: I2383137dc808002a39d285ac20b6033853d2a92d

Co-authored-by: Michael Lazos <mlazos@meta.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Dynamo] Handle tensor attributes

7 participants