you dont have to install it via anaconda, you could install cuda from their website. after install ends open a new terminal and check your cuda version with:
>>> nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Thu_Nov_18_09:52:33_Pacific_Standard_Time_2021
Cuda compilation tools, release 11.5, V11.5.119
Build cuda_11.5.r11.5/compiler.30672275_0
my is V11.5
then go here and select your os and preferred package manager(pip or anaconda), and the cuda version you installed, and copy the generated install command, I got:
pip3 install torch==1.10.1+cu113 torchvision==0.11.2+cu113 torchaudio===0.10.1+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
notice that for me I had python 3.10 installed but my project run over 3.9 so either use virtual environment or run pip of your wanted base interpreter explicitly (for example C:\Software\Python\Python39\python.exe -m pip install .....)
else you will be stuck with Could not find a version that satisfies the requirement torch errors
then open python console and check for cuda availability
>>> import torch
>>> torch.cuda.is_available()
True