I try to run my code with pytorch 10.2, but I get the assertion error described in the title. At first I installed cuda with version 11. Then I uninstalled it and installed version 10.2. Also, I used pytorch before without cuda and installed it using command: "conda install pytorch torchvision cpuonly -c pytorch
"
Today after installing Cuda, I ran "conda install pytorch torchvision cudatoolkit=10.2 -c pytorch"
However, when I run command: "torch.cuda.is_available()"
, the output is False.
After running "nvcc --version" I can see that Cuda 10.2 is installed: "Cuda compilation tools, release 10.2, V10.2.89"
. Any idea how to fix this assertion error?
Other answers describe that I should install pytorch with cuda, but I already did that and I still get the same error. Do I need to enable cuda with pytorch somehow manually?