Failing deployment share.streamlit.io

Hi, I dont understand why my deployment fails when everything seems to be working locally…

Am I missing something?

@Borg93
This is sort of a vague question. I suggest posting a screenshot of the error message streamlit deployment gives you, to help others understand better.

Looks like SAVE_STATE_ was removed from the newer version of pytorch.
Try including torch==1.4.0 in the requirements file.

okej thanks! Seems that I’m just getting error with keyBERT libary no instead…

Good to see the pytorch issue was resolved.

This time, apart from the keybert issue, it seems your program has general dependency issues.
Pip is having issues trying to install all the versions of packages that are compatible with each other.

Personally, there are two ways I would try solving this.

  1. Since the app runs locally, make sure the package versions in your requirements.txt are the same as the ones in your local machine. You can do this by moving to your working directory of the project in a bash terminal and typing pip freeze > requirements.txt Open that file and check the versions of the necessary packages.

  2. If the above doesn’t work. You can try using a package manager, (anaconda), which will deal with dependencies. Download anaconda, use it to make a virtual environment in your project’s working directory, use conda to install the needed packages into that virtual environment, then check the package versions it conda decided to go with.

2 Likes