[AC] Return None from apply_activation_checkpointing()
#87871
Conversation
[ghstack-poisoned]
awgu
requested review from
H-Huang,
kwen2501,
mrshenli,
pritamdamania87,
rohan-varma and
zhaojuanmao
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/87871
Note: Links to docs will display an error until the docs builds have been completed. ✅ No Failures, 6 PendingAs of commit 5bdd481: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
awgu
added
module: activation checkpointing
|
@pytorchbot merge |
This PR needs a labelIf your changes are user facing and intended to be a part of release notes, please use a label starting with If not, please add the For more information, see https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work. |
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 |
Merge failedReason: The following mandatory check(s) failed (Rule Dig deeper by viewing the failures on hud Details for Dev Infra teamRaised by workflow job |
|
@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 |
…7871) `_recursive_wrap()` returns `Tuple[nn.Module, int]`, where the `nn.Module` is the in-place modified module and the `int` is the numel wrapped. In that sense, the return value is not meant to be publicly used. The `apply_activation_checkpointing()` docs already suggest that the function returns `None`, so this PR simply follows that. **Test Plan** CI Pull Request resolved: pytorch#87871 Approved by: https://github.com/zhaojuanmao
…7871) `_recursive_wrap()` returns `Tuple[nn.Module, int]`, where the `nn.Module` is the in-place modified module and the `int` is the numel wrapped. In that sense, the return value is not meant to be publicly used. The `apply_activation_checkpointing()` docs already suggest that the function returns `None`, so this PR simply follows that. **Test Plan** CI Pull Request resolved: pytorch#87871 Approved by: https://github.com/zhaojuanmao
Stack from ghstack:
Nonefromapply_activation_checkpointing()#87871 [AC] ReturnNonefromapply_activation_checkpointing()_recursive_wrap()returnsTuple[nn.Module, int], where thenn.Moduleis the in-place modified module and theintis the numel wrapped. In that sense, the return value is not meant to be publicly used. Theapply_activation_checkpointing()docs already suggest that the function returnsNone, so this PR simply follows that.Test Plan
CI