[easy] Run autograd if any mutations on inputs that require grad

[jamesjwu]

Stack from ghstack (oldest at bottom):

• Run TestAOTAutograd test suite with cache #128222
• Add functionality to make ViewAndMutationData (slightly more) cache safe #127618
• Save backward graphs lazily to cache #126999
• First version of AOTAutogradCache #126791
• -> [easy] Run autograd if any mutations on inputs that require grad #128229

If any inputs are mutated that require grad, even if all the outputs don't require grad, we should still run autograd with a backwards graph. This fixes two tests: test_input_mutation_alias_everything and test_view_detach.

Fixes #128035

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/128229

• 📄 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.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Rebase your PRs: Unstable CUDA signal in CI caused by cudnn 9 update

✅ You can merge normally! (2 Unrelated Failures)

As of commit 3e8b9bb with merge base failed to retrieve merge base, please contact dev infra:

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• inductor / cuda12.1-py3.10-gcc9-sm86 / test (dynamic_inductor_timm, 1, 2, linux.g5.4xlarge.nvidia.gpu) (gh) (similar failure)
  gluon_inception_v3

UNSTABLE - The following job failed but was likely due to flakiness present on trunk and has been marked as unstable:

• inductor / linux-jammy-cpu-py3.8-gcc11-inductor / test (inductor_torchbench_cpu_smoketest_perf, 1, 1, linux.24xl.spr-metal, unstable) (gh) (#126993)
  Process completed with exit code 1.

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

[jamesjwu]

My previous PR didn't actually test xfail properly here, it was just skipping them anyway. It turns out in python 3 you can't xfail tests from methods, only at the class level. So I implemented expected failure with xfail_inherited_tests instead.

[jamesjwu]

@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

[bdhirsh]

cc @IvanKobzarev / @tugsbayasgalan - another subtle AOTAutograd correctness issue that would be good to understand (also around properly picking inference vs. training)

[bdhirsh]

thanks for the fix!