Reland: Make torch.empty* deterministic by filling with NaN or max int

[kurtamohler]

Relands #101849 after #104302 reverted it.

torchrec PR pytorch/torchrec#1269 fixes the torchrec failure that caused #101849 to be reverted

Part of #82004

cc @mruberry

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/104995

• 📄 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.

✅ No Failures

As of commit b214d41:
💚 Looks good so far! There are no failures yet. 💚

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

[albanD]

Thanks for pushing the fix on the torchrec side!

[kurtamohler]

Hey @albanD, if possible, would you be able to import this PR and run the internal torchrec tests, just to make sure they pass before we merge it?

[albanD]

Sure, I'll do a rebase because there was a revert in trunk responsible for the current CI failure.

[albanD]

@pytorchbot -r

[pytorch-bot]

❌ 🤖 pytorchbot command failed:

@pytorchbot: error: unrecognized arguments: -r

usage: @pytorchbot [-h] {merge,revert,rebase,label,drci} ...

Try @pytorchbot --help for more info.

[albanD]

@pytorchbot rebase

[pytorchmergebot]

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

[pytorchmergebot]

Tried to rebase and push PR #104995, but it was already up to date. Try rebasing against main by issuing:
@pytorchbot rebase -b main

[albanD]

@pytorchbot rebase -b main

[pytorchmergebot]

@pytorchbot started a rebase job onto refs/remotes/origin/main. Check the current status here

[pytorchmergebot]

Successfully rebased empty-deterministic-reland-0 onto refs/remotes/origin/main, please pull locally before adding more changes (for example, via git checkout empty-deterministic-reland-0 && git pull --rebase)

[facebook-github-bot]

@albanD has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[albanD]

There seems to be other issues where code expects empty() to return 0s :(
Working on fixing these call sites!

[kurtamohler]

Darn, that's unfortunate. Thanks for the help @albanD, let me know if there's anything I can do

[albanD]

First change seems to fix some. Waiting on CI to run again.

[albanD]

@pytorchbot merge -r

Ok I landed internal fixes. This should be green internally now!

[pytorchmergebot]

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

[pytorchmergebot]

Successfully rebased empty-deterministic-reland-0 onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout empty-deterministic-reland-0 && git pull --rebase)

[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

[kurtamohler]

Awesome, thank you @albanD!