Streamlit deployment error during processing dependencies

[barky]

Hello,

This is my first time deploying the app through streamlit.
I have put the packages that I want to import to requirements.txt as the following.

However, everytime I run the app, I am getting the following error and not sure how to fix it. Could anyone give me help? Thank you in advance!

[dataprofessor]

Hi @barky

Lines 1-8 should work perfectly as is. However the problem lies in lines 9-12. For lines 9 and 10, you’ll need to delete the empty spaces so that it becomes torch==1.9.0 and torchvision==0.10.0.

As for installing from GitHub, you can refer to the following:

PIP Installing from Github

Hi @sophiegeoghan and @C-Roensholt I had no luck via the requirements.txt route either What worked was installing the package in the main app script itself via a subprocess. Say your script is streamlit_app.py. Include the following at the very top of your app: import streamlit as st import subprocess import sys import time try: # replace "yourpackage" with the package you want to import import yourpackage # This block executes only on the first run when your pac…

Hope this helps!

[barky]

Thank you for your help!
I stopped getting such error message after fixing the requirements.txt like this.

However, now I am facing with another solution which is telling me that there is no module named torch. I’ve tried to look for a solution in the community to solve this problem, but is still running to error. Would there be any reasons why this is happening?