Machine Learning Container for Jetson and JetPack

The l4t-ml docker image contains TensorFlow, PyTorch, JupyterLab, and other popular ML and data science frameworks such as scikit-learn, scipy, and Pandas pre-installed in a Python 3 environment. These containers support the following releases of JetPack for Jetson Nano, TX1/TX2, Xavier NX, AGX Xavier, AGX Orin, Orin NX, and Orin Nano:

• JetPack 5.1.1 (L4T R35.3.1)
• JetPack 5.1 (L4T R35.2.1)
• JetPack 5.0.2 (L4T R35.1.0)
• JetPack 5.0.1 Developer Preview (L4T R34.1.1)
• JetPack 5.0.0 Developer Preview (L4T R34.1.0)
• JetPack 4.6.1 (L4T R32.7.1)
• JetPack 4.6 (L4T R32.6.1)
• JetPack 4.5 (L4T R32.5.0)
• JetPack 4.4.1 (L4T R32.4.4)
• JetPack 4.4 (L4T R32.4.3)
• JetPack 4.4 Developer Preview (L4T R32.4.2)

For additional machine learning containers for Jetson, see the l4t-pytorch and l4t-tensorflow images. Note that the TensorFlow and PyTorch pip wheel installers for aarch64 are available to download independently from the Jetson Zoo.

Package Versions

Depending on your version of JetPack-L4T, different tags of the l4t-ml container are available, each with support for Python 3. Be sure to clone a tag that matches the version of JetPack-L4T that you have installed on your Jetson.

• JetPack 5.1.1 (L4T R35.3.1)

  • l4t-ml:r35.3.1-py3
    • TensorFlow 2.11.0
    • PyTorch v2.0.0
    • torchvision v0.15.1
    • torchaudio v2.0.1
    • onnx 1.13.1
    • onnxruntime 1.16.0
    • optimum 1.8.8
    • CuPy 13.0.0
    • numpy 1.23.5
    • numba 0.56.4
    • PyCUDA 2022.2
    • OpenCV 4.5.0 (with CUDA)
    • pandas 2.0.1
    • scipy 1.10.0
    • scikit-learn 1.2.2
    • diffusers 0.17.1
    • transformers 4.30.2
    • xformers 0.0.20
    • JupyterLab 3.6.3

• JetPack 5.1 (L4T R35.2.1)

  • l4t-ml:r35.2.1-py3
    • TensorFlow 2.11.0
    • PyTorch v2.0.0
    • torchvision v0.14.1
    • torchaudio v0.13.1
    • onnx 1.13.0
    • CuPy 11.5.0
    • numpy 1.21.1
    • numba 0.56.4
    • PyCUDA 2022.2
    • OpenCV 4.5.0 (with CUDA)
    • pandas 1.5.3
    • scipy 1.10.0
    • scikit-learn 1.2.1
    • JupyterLab 3.6.1

• JetPack 5.0.2 (L4T R35.1.0)

  • l4t-ml:r35.1.0-py3
    • TensorFlow 1.15.5
    • PyTorch v1.12.0
    • torchvision v0.13.0
    • torchaudio v0.12.0
    • onnx 1.12.0
    • CuPy 10.2.0
    • numpy 1.22.4
    • numba 0.56.0
    • PyCUDA 2022.1
    • OpenCV 4.5.0 (with CUDA)
    • pandas 1.4.3
    • scipy 1.9.0
    • scikit-learn 1.1.1
    • JupyterLab 3.4.4

• JetPack 5.0.1 Developer Preview (L4T R34.1.1)

  • l4t-ml:r34.1.1-py3
    • TensorFlow 1.15.5
    • PyTorch v1.12.0
    • torchvision v0.12.0
    • torchaudio v0.11.0
    • onnx 1.11.0
    • CuPy 10.2.0
    • numpy 1.21.6
    • numba 0.55.1
    • PyCUDA 2021.1
    • OpenCV 4.5.0 (with CUDA)
    • pandas 1.4.2
    • scipy 1.8.1
    • scikit-learn 1.1.1
    • JupyterLab 3.4.2

• JetPack 5.0.0 Developer Preview (L4T R34.1.0)

  • l4t-ml:r34.1.0-py3
    • TensorFlow 1.15.5
    • PyTorch v1.12.0
    • torchvision v0.12.0
    • torchaudio v0.11.0
    • onnx 1.11.0
    • CuPy 10.2.0
    • numpy 1.21.5
    • numba 0.55.1
    • PyCUDA 2021.1
    • OpenCV 4.5.0 (with CUDA)
    • pandas 1.4.1
    • scipy 1.8.0
    • scikit-learn 1.0.2
    • JupyterLab 3.3.3

• JetPack 4.6.1 (L4T R32.7.1)

  • l4t-ml:r32.7.1-py3
    • TensorFlow 1.15.5
    • PyTorch v1.10.0
    • torchvision v0.11.0
    • torchaudio v0.10.0
    • onnx 1.11.0
    • CuPy 9.2.0
    • numpy 1.19.5
    • numba 0.53.1
    • OpenCV 4.5.0 (with CUDA)
    • pandas 1.1.5
    • scipy 1.5.4
    • scikit-learn 0.24.2
    • JupyterLab 2.2.9

• JetPack 4.6 (L4T R32.6.1)

  • l4t-ml:r32.6.1-py3
    • TensorFlow 1.15.5
    • PyTorch v1.9.0
    • torchvision v0.10.0
    • torchaudio v0.9.0
    • onnx 1.8.0
    • CuPy 8.0.0
    • numpy 1.19.5
    • numba 0.53.1
    • OpenCV 4.4.0 (with CUDA)
    • pandas 1.1.5
    • scipy 1.5.4
    • scikit-learn 0.23.2
    • JupyterLab 2.2.9

• JetPack 4.5 (L4T R32.5.0)

  • l4t-ml:r32.5.0-py3
    • TensorFlow 1.15
    • PyTorch v1.7.0
    • torchvision v0.8.0
    • torchaudio v0.7.0
    • onnx 1.8.0
    • CuPy 8.0.0
    • numpy 1.19.4
    • numba 0.52.0
    • OpenCV 4.4.1
    • pandas 1.1.5
    • scipy 1.5.4
    • scikit-learn 0.23.2
    • JupyterLab 2.2.9

• JetPack 4.4.1 (L4T R32.4.4)

  • l4t-ml:r32.4.4-py3
    • TensorFlow 1.15
    • PyTorch v1.6.0
    • torchvision v0.7.0
    • torchaudio v0.6.0
    • onnx 1.7.0
    • CuPy 8.0.0
    • numpy 1.19.2
    • numba 0.51.2
    • pandas 1.1.3
    • scipy 1.5.3
    • scikit-learn 0.23.2
    • JupyterLab 2.2.8

• JetPack 4.4 (L4T R32.4.3)

  • l4t-ml:r32.4.3-py3
    • TensorFlow 1.15
    • PyTorch v1.6.0
    • torchvision v0.7.0
    • torchaudio v0.6.0
    • onnx 1.7.0
    • CuPy 8.0.0
    • numpy 1.19.0
    • numba 0.50.0
    • pandas 1.0.5
    • scipy 1.5.0
    • scikit-learn 0.23.1
    • JupyterLab 2.1.5

• JetPack 4.4 Developer Preview (L4T R32.4.2)

  • l4t-ml:r32.4.2-py3
    • TensorFlow 1.15
    • PyTorch v1.5.0
    • torchvision v0.6.0
    • onnx 1.6.0
    • numpy 1.18.2
    • pandas 1.0.3
    • scipy 1.4.1
    • scikit-learn 0.22.2
    • JupyterLab 2.0.1

note: the l4t-ml containers require JetPack 4.4 or newer

Running the Container

First pull one of the l4t-ml container tags from above, corresponding to the version of JetPack-L4T that you have installed on your Jetson. For example, if you are running the latest JetPack 5.1 (L4T R35.2.1) release:

sudo docker pull nvcr.io/nvidia/l4t-ml:r35.2.1-py3

Then to start an interactive session in the container, run the following command:

sudo docker run -it --rm --runtime nvidia --network host nvcr.io/nvidia/l4t-ml:r35.2.1-py3

You should then be able to start a Python3 interpreter and import the packages above.

Connecting to JupyterLab Server

Unless a user-provided run command overrides the default, a JupyterLab server instance is automatically started along with the container.

You can connect to it by navigating your browser to http://localhost:8888 (or substitute the IP address of your Jetson device if you wish to connect from a remote host, i.e. with the Jetson in headless mode). Note that the default password used to login to JupyterLab is nvidia.

Mounting Directories from the Host Device

To mount scripts, data, ect. from your Jetson's filesystem to run inside the container, use Docker's -v flag when starting your Docker instance:

sudo docker run -it --rm --runtime nvidia --network host -v /home/user/project:/location/in/container nvcr.io/nvidia/l4t-ml:r35.2.1-py3

Dockerfiles

To access or modify the Dockerfiles and scripts used to build this container, see this GitHub repo.

License

The l4t-ml container includes various software packages with their respective licenses included within the container.

Getting Help & Support

If you have any questions or need help, please visit the Jetson Developer Forums.