Installation

GTNet is available on the Python Package Index.

pip install gtnet

GPU acceleration

GTNet uses PyTorch, so it is capable of GPU acceleration with CUDA. As long as CUDA is available on your system, GTNet will detect if CUDA is available and make GPU acceleration available.

If your system is equipped with NVIDIA GPUs, but are unsure if CUDA is installed, we recommend installing PyTorch and the CUDA Toolkit using Conda.

For example, if you would like to run PyTorch with CUDA Toolkit 11.8, you can run the following commands:

conda create -n gtnet-env
conda activate gtnet-env
conda install pytorch pytorch-cuda=11.8 -c pytorch -c nvidia
pip install gtnet