On the page https://pytorch.org/get-started/previous-versions/ the following instructions are provided:

Linux and Windows
# CUDA 10.2
conda install pytorch==1.9.1 torchvision==0.10.1 torchaudio==0.9.1 cudatoolkit=10.2 -c pytorch

# CUDA 11.3
conda install pytorch==1.9.1 torchvision==0.10.1 torchaudio==0.9.1 cudatoolkit=11.3 -c pytorch -c conda-forge

# CPU Only
conda install pytorch==1.9.0 torchvision==0.10.1 torchaudio==0.9.1 cpuonly -c pytorch

However, when checking https://anaconda.org/pytorch/pytorch/files?version=1.9.1, I noticed that there are no PyTorch packages that fulfill the requirement for CUDA 11.3. It appears that CUDA 11.1 should be used instead of 11.3.

Users who follow the current instructions will end up installing the CPU-only packages, which may not be their intention.

Please consider updating the page with the correct CUDA version to prevent this confusion.