I’m trying to use Autotrain-Advanced with an Nvidiea Geforce 4070 and i got the error:“No GPU available”. This occurs because of the pytorch-command: “torch.cuda.is_available()”. So my questions is, has the Nvidea Geforce 4070 CUDA-ability.
Additionally, i can install the NVIDEA-CUDA-Toolkit and have the Nvidea Compiler Driver on my PC and a Cuda-Version also, but the command returns “false”.
Yes. You do not say which version of Cuda you have installed. I know nothing about Pytorch, but the home page for it states you need either 11.8 or 12.1.
I have installed the 12.3 Version of Cuda.
Now I installed 12.1 Version and it don’t work eiter. But i don’t know if i uninstalled Cuda 12.3 the right way. I uninstalled everything in the App-Manager that has something to do with Cuda 12.3 is that right? Because the command “nvidia-smi” still shows Cuda Version 12.3
And another Question
Is it possible to use Cuda with python on Windows? Because the instructions in this guidance: 1. Introduction — Installation Guide Windows 12.3 documentation
only works on Linux, because of the module
Did you get CUDA 12.3 working?
I guess I did, but don’t know for sure.
I switched to ubuntu and it works there. So my problem is done.