CUDA and CUDNN issue in running openpose model

[Mobina-Zibandepour]

I have a problem and need help!
I encountered several errors in running openpose, which ran perfectly until last week BEFORE Google Colab updates with CUDA12.2 and CUDANN 8.9.6. It seems that we must manually downgrade the CUDA and CuDNN in Google Colab, but it isn't easy work. I faced many errors, so we thank Google Colab developers for explaining what to do in this situation?

[metrizable]

@Mobina-Zibandepour Thanks for filing the issue and thanks for using Colab.

Colab runtimes were upgraded to CUDA 12.5, announced in #5061. During large upgrades of the runtime, we provide the ability to select the fallback version from the Command Palette via the "Use fallback runtime version" option. In your case, it would allow you to select the prior runtime which provides CUDA 12.2 while you are updating your notebook to be compatible with the newer version.