You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
I am using Colab to run some pytorch examples. Installing using pip worked with the original version, but from time to time it leads to problems, since tensorflow is preinstalled with a different numpy version. In our CI we runs the demo in a clean virtual environment for testing that everything work. I would then like to be able to assure that the demos also run in Colab. But unfortunately, every now and then I realize the the demos are broken and need to do some workaround to fix some dependency problems.
Looking for a solution I found many posts / issues where people are asking for additional or newer versions of some preinstalled libraries.
Describe the solution you'd like
I would suggest to add an option to use Colab without any preinstalled libraries. In case of problems with the dependencies I could at least start with a clean environment.
Describe alternatives you've considered
I played around with a virtual environment inside of Colab, but this was always a bit unstable and complicates the demo code.
A more advanced feature would be to be able to provide the required dependencies via requirements.txt or something similar, but this would probably be much more complicated to implement.
Additional context
Add any other context, screenshots or notebooks about the feature request here.
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
I am using Colab to run some
pytorch
examples. Installing usingpip
worked with the original version, but from time to time it leads to problems, sincetensorflow
is preinstalled with a differentnumpy
version. In our CI we runs the demo in a clean virtual environment for testing that everything work. I would then like to be able to assure that the demos also run in Colab. But unfortunately, every now and then I realize the the demos are broken and need to do some workaround to fix some dependency problems.Looking for a solution I found many posts / issues where people are asking for additional or newer versions of some preinstalled libraries.
Describe the solution you'd like
I would suggest to add an option to use Colab without any preinstalled libraries. In case of problems with the dependencies I could at least start with a clean environment.
Describe alternatives you've considered
I played around with a virtual environment inside of Colab, but this was always a bit unstable and complicates the demo code.
A more advanced feature would be to be able to provide the required dependencies via
requirements.txt
or something similar, but this would probably be much more complicated to implement.Additional context
Add any other context, screenshots or notebooks about the feature request here.
The text was updated successfully, but these errors were encountered: