Hi,
I am very new to the topic. I was following a tutorial about simple AI and such. Nothing worked and the reason seems to be, that Python cannot Communicate with CUDA (Or my Graphicscard in general.).
So I learned that every Pytorch comes with its own CUDA Version. OFC I installed the newest, which probatly was the problem itself. My Graphic Card has CUDA 7.5 and 240 Tensor Cores (Gen 2).
According to this list, I need PyTorch 0.4.1 which was a real hustle to unistall the new one and install the old version.
Now my Tutorial is not working anymore, because it was not meant to process with a historic version of Pytorch.
Does that mean I have no chance to work with AI (of this century) at all? Or do you guys know a trick to surepass the limitation?
GPU: INNO3D GeForce RTX 2060 Compact X1
CPU: Intel Core I7 4790k
RAM: 16 GB DDR3
OS: Windows 10
Python: 3.10.4
All drivers up to date, CUDA Toolkit installed, Path Variables checked and seemed ok.