Before Conda environments may be used by Jupyter Notebooks they must be installed so that Jupyter knows about them. Older versions of Conda automatically installed a Jupyter kernel for themselves, that installation process now must be performed manually by the user.
To perform the installation the user should load their prefered version of Python, activate their Conda environment and run the following command:
PyTorch is a Python package that provides Tensor computation (like NumPy) with strong GPU acceleration and Deep neural networks built on a tape-based autograd system.
Quote from Pytorch Github documentation.
Availability and Restrictions
The following version of Pytorch is available on OSC clusters: