Customize NGC Container for Jetson AGX Xavier

Hey there, I have a Jetson AGX Xavier which I want to use with an NGC Docker Container from nvcr.io. I have Jetpack 4.6.2 installed and would like to use the L4T-ML container. I took the l4t-ml:r32.7.1-py3 assuming it works for Jetpack 4.6.2 even though it is only stated to work with Jetpack 4.6.1.

However, I do not really need tensorflow or Jupyter on my container so I’d like to cutomize according the Dockerfile I found here: jetson-containers/Dockerfile.ml at master · dusty-nv/jetson-containers · GitHub by deleting the lines concerning with tensorflow and jupyter.

I am however unsure how to set “ARG PYTORCH_IMAGE”. I’ve tried “ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:22.04-py3” but that doesn’t seem to work. I’d be very thankful if someone could tell me how to do this correctly!

Thanks in advance!

Hi @user152354, these get set in the scripts/docker_build_ml.sh script here:

If you don’t want TensorFlow and JupyterLab though, it may be easier for you just to use l4t-pytorch container?

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.