-
Notifications
You must be signed in to change notification settings - Fork 1.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[bnb-nightly
] Address final comments
#1287
Conversation
@@ -25,7 +25,7 @@ jobs: | |||
TEST_TYPE: "single_gpu_${{ matrix.docker-image-name }}" | |||
container: | |||
image: ${{ matrix.docker-image-name }} | |||
options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/ -e NVIDIA_DISABLE_REQUIRE=true | |||
options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the env variable NVIDIA_DISABLE_REQUIRE
is set on the very top so these are not needed
@@ -39,16 +39,19 @@ jobs: | |||
mkdir transformers-clone && git clone https://github.com/huggingface/transformers.git transformers-clone # rename to transformers clone to avoid modules conflict | |||
|
|||
- name: Run examples on single GPU | |||
if: always() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@ydshieh I thought the fail-fast: false
already takes care of running the tests even if they fail, do you know why here we need to put always()
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The fail-fast:
(under jobs.<job_id>.strategy
) applies to the jobs in the defined matrix. But here each job in the matrix has 3 testing steps in each job, but those steps won't be controlled by jobs.<job_id>.strategy
.
That is my understanding
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
well, let me double check maybe!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
See this example
https://github.com/huggingface/transformers/actions/runs/7286009365/job/19853943708
all 3 jobs (in the matrix) are run. But the 3rd step in each job are not.
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
bnb-nightly
] Address final comments
run: | | ||
source activate peft | ||
make tests_examples_single_gpu_bnb | ||
|
||
- name: Run core tests on single GPU | ||
if: always() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is the quick and simple way. Just want to point out:
if: always()
will make the step uncancellable. A better approach would beif: !cancelled()
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
But I believe we only want to run the 3 tests only if the step Pip install
succeed. The condition gets a bit more complex in this case. OK for me if you are happy to keep the current version.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Perfect thanks! I will stick to the current plan then :D
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, just give you some info for you to make the final decision.
Thanks a lot for your review @ydshieh ! |
What does this PR do?
Address final comments from @ydshieh on. #1282 (review)
I left few questions !
cc @ydshieh