Cache results of is_torch_tpu_available() #18777
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The documentation is not available anymore as the PR was closed or merged. |
comaniac
commented
Aug 26, 2022
comaniac
commented
Aug 26, 2022
|
Not sure the reason of CI failure. It seems not relevant to this PR. |
LysandreJik
approved these changes
Aug 30, 2022
There was a problem hiding this comment.
This sounds fair to me, but I'll let @sgugger, mastermind of the Trainer, review when he's back from leave at the end of the week!
|
@LysandreJik thanks for the review and sure we could wait for @sgugger. |
sgugger
approved these changes
Aug 31, 2022
|
Thanks for bearing with us. The test failures are spurious and unrelated, so we can merge this. |
oneraghavan
pushed a commit
to oneraghavan/transformers
that referenced
this pull request
Sep 26, 2022
* Cache results of is_torch_tpu_available() * Update src/transformers/utils/import_utils.py * Update src/transformers/utils/import_utils.py
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
xm.xla_device()(called byis_torch_tpu_available()) hangs when calling multiple times but no XLA devices are available, and this results in Trainer hanging. Since currentlytorch_xlawill be used as long as it is installed in the current active Python environment, I encountered this issue even when I only want to run the Trainer with PyTorch on GPU.The detail reason behind
torch_xlais still under investigation (see pytorch/xla#3939).To workaround this issue, this PR adds
lru_cachetois_torch_tpu_available(), so thatxm.xla_device()is guaranteed to be called only once when no XLA device is available.Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@muellerzr @sgugger