> pip uninstall torch torchvision torchaudio
> git clone https://github.com/pytorch/pytorch.git
> cd pytorch
> git submodule sync && git submodule update --init --recursive
> sudo docker run -d --gpus=1 --runtime=nvidia --rm -p8000:8000 -p8001:8001 -p8002:8002 -v/models:/models nvcr.io/nvidia/tritonserver:24.01-py3-igpu tritonserver --model-repository=/models --strict-model-config=false
While building applications from source on a Jetson Orin Nano, you may run into performance issues .. notably running out of resources which causes lock-ups and reboots as it only has 8 GB of RAM. This gist explains how to give yourself more SWAP space to better the chances of getting a successful build.
sudo apt-get install -y dphys-swapfile