Skip to content

Fix device of masks in tests#27887

Merged
fxmarty merged 1 commit into
huggingface:mainfrom
fxmarty:fix-device-mask-tests
Dec 7, 2023
Merged

Fix device of masks in tests#27887
fxmarty merged 1 commit into
huggingface:mainfrom
fxmarty:fix-device-mask-tests

Conversation

@fxmarty

@fxmarty fxmarty commented Dec 7, 2023

Copy link
Copy Markdown
Contributor

As per title, fix a bug introduced in #24587 in the GPU tests.

@fxmarty
fxmarty requested review from ArthurZucker and LysandreJik and removed request for ArthurZucker December 7, 2023 11:47
@fxmarty fxmarty assigned amyeroberts and unassigned amyeroberts Dec 7, 2023
@fxmarty
fxmarty requested a review from amyeroberts December 7, 2023 11:48

@ArthurZucker ArthurZucker left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks

@ArthurZucker
ArthurZucker removed the request for review from amyeroberts December 7, 2023 11:53
@ydshieh

ydshieh commented Dec 7, 2023

Copy link
Copy Markdown
Collaborator

Strange that we don't see the issue on our CI. Do you know why @fxmarty?

@fxmarty

fxmarty commented Dec 7, 2023

Copy link
Copy Markdown
Contributor Author

@ydshieh Can you try RUN_SLOW=1 CUDA_VISIBLE_DEVICES=0 pytest tests/models/llama -s -vvvvv -k "test_resize_tokens_embeddings" on main?

@fxmarty

fxmarty commented Dec 7, 2023

Copy link
Copy Markdown
Contributor Author

I am not sure it's weird. The push tests should run on GPU so those should have failed...

machine_type: [single-gpu]

@fxmarty

fxmarty commented Dec 7, 2023

Copy link
Copy Markdown
Contributor Author

As discussed offline it did actually fail (https://github.com/huggingface/transformers/actions/runs/7124821720) so all good!

@fxmarty
fxmarty merged commit c99f254 into huggingface:main Dec 7, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants