-
Notifications
You must be signed in to change notification settings - Fork 3
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
use onnxruntime-gpu for large-model-test #24
base: mainline
Are you sure you want to change the base?
Conversation
@@ -96,5 +97,7 @@ setenv = | |||
PATH = {env:PATH}{:}{toxinidir}{/}scripts | |||
commands = | |||
pip install --upgrade -r {toxinidir}{/}temp{/}marqo{/}requirements.txt | |||
pip uninstall -y onnxruntime | |||
pip install onnxruntime-gpu==1.13.1 |
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.
i don't think tox envs were meant to use deps like this, why not instead do this:
- don't use
pip install
- install deps with
deps =
{[testenv]deps}
-r{toxinidir}{/}temp{/}marqo{/}requirements.txt
- define
onnxruntime-gpu==1.13.1
inrequirements.txt
But if it's not possible to change this in the requirements file itself, your solution could be a good workaround :)
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.
Yes, the requirements.txt is using onnxruntime in general. We only need onnxruntime-gpu for the large model test. any good idea?
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.
I see. note that the solution you put here will affect both largemodel_unit_test_CI.yml and unit_test_CI.yml. Have you tried running regular unit tests with this?
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.
i missed this point. we should use the onnxruntime in unittest as the instance does not have GPU. do you reckon creating a new test env will solve the problem?
use onnxruntime-gpu for large-model-test