Pytorch With CUDA on Zed Box

I am attempting to use a custom pytorch model deployed on a zed box. It has been very frustrating trying to figure out how to build pytorch with CUDA enabled.
Has anyone had success doing this? Would using a docker container be more effective?

Hi @frigusgulo,

As I’ve mentioned in the other post and for anyone that gets here, I would suggest following the official guide from NVIDIA: Installing PyTorch for Jetson Platform - NVIDIA Docs