Yes, the CUDA runtime used by your PyTorch binaries will be shown via print(torch.version.cuda) and would be 10.2 in your use case, which is why the error is raised as your Ampere GPU needs CUDA>=11.
Thanks again for your answer!! I am a new starter and your answers towards many questions have helped me a great deal! 
I am trying to replicate work in a github repo that requires PyTorch 1.7.0 to run. My conda environment was working yesterday and all of a sudden i am now getting the error in this thread. I have tried creating a new conda environment and using
conda install pytorch==1.7.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=11.0 -c pytorch
but am still getting the same error.
Your PyTorch version might be too old so you would need to update it to the latest stable or nightly release.