Change default torch_function behavior to be disabled when torch_dispatch is defined (take 2)

[albanD]

This does not introduce a new test but is tested by checking that all the classes we already have still behave as before now that they don't explicitly disable torch_function.

cc @mrshenli @pritamdamania87 @zhaojuanmao @satgera @rohan-varma @gqchen @aazzolini @osalpekar @jiayisuse @H-Huang @kwen2501 @awgu @penguinwu @fegin @XilunWu @wanchaol @fduwjj @wz337 @tianyu-l @wconstab @yf225 @chauhang @ezyang @gchanan @voznesenskym @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @chenyang78 @kadeng @aakhundov

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/120632

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

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

✅ You can merge normally! (2 Unrelated Failures)

As of commit 337e6b8 with merge base 7b4f70e ():

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / linux-focal-cuda11.8-py3.10-gcc9 / test (distributed, 3, 3, linux.8xlarge.nvidia.gpu) (gh)
  distributed/checkpoint/test_state_dict.py::TestStateDict::test_compiled_fsdp
• pull / linux-focal-py3_8-clang9-xla / test (xla, 1, 1, linux.12xlarge) (gh)
  npm ERR! code EEXIST

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

[ezyang]

yoloooo

[albanD]

@JackCaoG is there an easy way for me to run this failing xla test on a cpu linux machine?
If not, would it be ok for you to add the following to XLAShardedTensor such that this PR will be a no-op for your Tensor and we can look into why this is happening later?

    #EDIT: added the classmethod here to avoid this weird Dynamo behavior https://github.com/pytorch/pytorch/issues/120799
    @classmethod
    def __torch_function__(cls, func, types, args=(), kwargs=None):
        return super().__torch_function__(func, types, args, kwargs)

[JackCaoG]

Yea I can add that to XLAShardedTensor and create a pr

[albanD]

@JackCaoG where I can know if the PR you did is already included in PT trunk? Or anything I should do to make your PR in xla included for this PR?

[JackCaoG]

You can check https://github.com/pytorch/xla/tree/master/.circleci#coodinating-merges-for-breaking-pytorch-prs for detail but pretty much you can update https://github.com/pytorch/pytorch/blob/main/.github/ci_commit_pins/xla.txt to the head of the pt/xla. I checked that file and it seems like pin was updated today so if you rebase this should just work.

[albanD]

@pytorchbot merge

[pytorchmergebot]

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