Torch.cuda.is_available() returns false, mismatch


A very strange problem.

I have NVIDIA RTX 3050 Ti Laptop.

I am trying to setup a repo that requires:
" This repo is tested with PyTorch 1.2, cuda 10.0 and Python 3.6."

The graphics card does not support CUDA 10.0. If I use CUDA 11.6 or similar, PyTorch 1.2 cannot run on that cuda.

Any solution??

Try this? I had the same problem and it worked for me:

conda install pytorch torchvision cudatoolkit=11.1 -c pytorch -c conda-forge 

full fix:

PS: Your RTX 3050 is well equipped for pretty much any version of CUDA, see:

CUDA Compatibility of NVIDIA Display / GPU Drivers | Amikelive | Technology Blog.