This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(128, 74%, 22%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ES%3C/text%3E%3C/svg%3E)
Hello!
I have a python web app on Azure.
How can I install torch with requirements.txt?
With the Web SSH, it always restarts.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(201.6, 12%, 29%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ESS%3C/text%3E%3C/svg%3E)
Try adding this to the requirements file. It worked for me.
-f https://download.pytorch.org/whl/torch_stable.html
torch==1.11.0+cpu
torchvision==0.12.0+cpu
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(25.6, 68%, 12%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ECR%3C/text%3E%3C/svg%3E)
Solved my issue! Thank you so much
Hi @SuperMX ,
When you deploy your Python app, Azure App Service will create a virtual environment to and run pip install -r requirements.txt. If you don't see this happening, make sure SCM_DO_BUILD_DURING_DEPLOYMENT is set to 1. See https://learn.microsoft.com/en-us/azure/app-service/configure-language-python for more details. If you're seeing errors during app startup, please comment down below.
Hello @Ryan Hill , thank you for your answer!
SCM_DO_BUILD_DURING_DEPLOYMENT was already set to 1.
I've tried a lot and it seems the build only fails if i add torch to requirements.txt. But without, of course it cannot find the module.
This is the end of the deployment log:
*[21:11:00+0000] Collecting numpy
[21:11:00+0000] Downloading numpy-1.21.1-cp38-cp38-manylinux_2_12_x86_64.manylinux2010_x86_64.whl (15.8 MB)
[21:11:02+0000] Collecting torch
[21:11:02+0000] Downloading torch-1.9.0-cp38-cp38-manylinux1_x86_64.whl (831.4 MB)
An unknown error has occurred. Check the diagnostic log for details.
Error: Failed to deploy web package to App Service.
Error: Deployment Failed with Error: Package deployment using ZIP Deploy failed. Refer logs for more details.*
It is linux based but I was able to install it with putting the pytorch path in the requirements.txt! (Install from path).
Do you still want the logs?
Glad to hear that @SuperMX . However, I'm not familiar with install from path. Did you do something similar to https://github.com/pytorch/pytorch/issues/29745?
yes!
--find-links https://download.pytorch.org/whl/torch_stable.html
torch==1.9.0+cpu
Cool. Glad you were able resolve the issue! Happy coding!