Reputation: 15
torch is not installing and getting failed and I need that module so is there any option to get it done with requirement.txt
Upvotes: 0
Views: 658
Reputation: 8008
Torch is not installing and getting failed, and I need that module so is there any option to get it done with requirement.txt:
This seems to be an issue with oryx build releases and that is why deployment patch is failing with an error oryx deployment files doesn't exists and unable to fetch the logs also.
After I've done a workaround,
To avoid this, I added ENABLE_ORYX_BUILD = false in function app->Configuration-> Application Settings
(App that will deploy my function) Portal as shown:
After doing it, I've installed torch & torchvision
modules with the compatible versions by using:
pip install torch===1.8.0 torchvision===0.9.0 -f https://download.pytorch.org/whl/torch_stable.html
Note: If the provided torch & torch_vision
versions are not compatible then it throws an error asking us to check the dependencies to avoid conflicts.
And
pip install -r requirements.txt
requirements.txt:
azure-functions==1.12.0 //Give latest version
requests==2.28.1
torch==1.8.0
torchvision==0.9.0
typing_extensions==4.4.0
Sometimes requirements.txt
file will not get properly configured.
To avoid that, Freeze the requirements by using terminal:
pip freeze > requirements.txt
I created an Azure function with python runtime stack, it worked for me and deployment got succeeded:
Triggered successfully after deployment and function app is running:
Verification in Azure Portal:
Version_check:
Python_Version: Python 3.9.3
(taken)
Upgrade pip
to the latest version if prompted.
PyTorch: pytorch.org
Restart the Visual Studio code (ctrl + shift + p
) by viewing command palette -> Reload the window and clear cache
to avoid module conflicts if needed.
Upvotes: 1