dusty-nv/jetson-containers

CUDA Containers for Edge AI & Robotics

Modular container build system that provides the latest AI/ML packages for NVIDIA Jetson :rocket::robot:

See the packages directory for the full list, including pre-built container images for JetPack/L4T.

Using the included tools, you can easily combine packages together for building your own containers. Want to run ROS2 with PyTorch and Transformers? No problem - just do the system setup, and build it on your Jetson:

$ jetson-containers build --name=my_container pytorch transformers ros:humble-desktop

There are shortcuts for running containers too - this will pull or build a l4t-pytorch image that's compatible:

$ jetson-containers run $(autotag l4t-pytorch)

^{jetson-containers run launches docker run with some added defaults (like --runtime nvidia, mounted /data cache and devices)
autotag finds a container image that's compatible with your version of JetPack/L4T - either locally, pulled from a registry, or by building it.}

If you look at any package's readme (like l4t-pytorch), it will have detailed instructions for running it.

Changing CUDA Versions

You can rebuild the container stack for different versions of CUDA by setting the CUDA_VERSION variable:

CUDA_VERSION=12.6 jetson-containers build transformers

It will then go off and either pull or build all the dependencies needed, including PyTorch and other packages that would be time-consuming to compile. There is a Pip server that caches the wheels to accelerate builds. You can also request specific versions of cuDNN, TensorRT, Python, and PyTorch with similar environment variables like here.

Documentation

• Package List
• Package Definitions
• System Setup
• Building Containers
• Running Containers

Check out the tutorials at the Jetson Generative AI Lab!

Getting Started

Refer to the System Setup page for tips about setting up your Docker daemon and memory/storage tuning.

# install the container tools
git clone https://github.com/dusty-nv/jetson-containers
bash jetson-containers/install.sh

# automatically pull & run any container
jetson-containers run $(autotag l4t-pytorch)

To also spin up a local devpi PyPI + APT caching server during install, pass --pypi:

sudo bash jetson-containers/install.sh --pypi

This launches a containerized devpi instance with per-architecture indexes (jp6, jp7, sbsa, amd64) and an nginx-based APT cache, so subsequent builds resolve wheels and packages locally instead of hitting the network.

Or you can manually run a container image of your choice without using the helper scripts above:

sudo docker run --runtime nvidia -it --rm --network=host dustynv/l4t-pytorch:r36.2.0

Looking for the old jetson-containers? See the legacy branch.

Only Tested and supported Jetpack 6.2 (Cuda 12.6) and JetPack 7 (CUDA 13.x).

[!NOTE] Ubuntu 24.04 containers for JetPack 6 and JetPack 7 are now available (with CUDA support)

     LSB_RELEASE=24.04 jetson-containers build pytorch:2.8      jetson-containers run dustynv/pytorch:2.8-r36.4-cu128-24.04

ARM SBSA (Server Base System Architecture) is supported for GH200 / GB200. To install CUDA 13.0 SBSA wheels for Python 3.12 / 24.04:

     uv pip install torch torchvision torchaudio \             --index-url https://pypi.jetson-ai-lab.io/sbsa/cu129

See the Ubuntu 24.04 section of the docs for details and a list of available containers 🤗 Thanks to all our contributors from Discord and AI community for their support 🤗

[!IMPORTANT] Jetson Orin now receives ARM SBSA CUDA in CUDA 13.2 — Starting with CUDA 13.2, the same Arm SBSA CUDA Toolkit is supported on Jetson Orin devices, unifying the toolkit across server-class (GH200/GB200) and embedded (Orin) platforms. This requires JetPack 7.2 with CUDA driver R595 minimum.

Code Style

The project uses automated code formatting tools to maintain consistent code style. See Code Style Guide for details on:

• Setting up formatting tools
• Adding your package to formatting checks
• Troubleshooting common issues

Troubleshooting

InvalidDefaultArgInFrom: Default value for ARG ${BASE_IMAGE} results in empty or invalid base image name

This may be caused by corrupted caches. Docker caches both image results and build downloads. Try these simple steps first:

• Remove non-essential docker images using: docker image rm --force <image id>
• Prune resulting docker image cache: docker image prune -f
• Prune docker build image cache: docker buildx prune -f

Gallery

Multimodal Voice Chat with LLaVA-1.5 13B on NVIDIA Jetson AGX Orin (container: NanoLLM)

Interactive Voice Chat with Llama-2-70B on NVIDIA Jetson AGX Orin (container: NanoLLM)

Realtime Multimodal VectorDB on NVIDIA Jetson (container: nanodb)

NanoOWL - Open Vocabulary Object Detection ViT (container: nanoowl)

Live Llava on Jetson AGX Orin (container: NanoLLM)

Live Llava 2.0 - VILA + Multimodal NanoDB on Jetson Orin (container: NanoLLM)

Small Language Models (SLM) on Jetson Orin Nano (container: NanoLLM)

Realtime Video Vision/Language Model with VILA1.5-3b (container: NanoLLM)

Citation

Please see CITATION.cff for citation information.