iot-salzburg/gpu-jupyter

What is it?

Key features

• 01Preconfigured JupyterLab with PyTorch and TensorFlow on NVIDIA CUDA.
• 02Version-tagged images for reproducible experiments.
• 03Local folder mounting so work persists outside the container.
• 04Python-only, slim and full variants with optional Julia and R.
• 05Docker Compose support and a static Jupyter token option.
• 06Generator script to build a customized Dockerfile and image.

Use cases

• Running GPU-accelerated deep learning experiments in notebooks on a local machine or server.
• Sharing a reproducible research environment with collaborators or in publications.
• Setting up a standard data science workspace on a GPU host without manual driver and library setup.
• Building a team-specific image by layering custom packages on top of the base image.

Quick start

1. 1

   Install the requirements: NVIDIA GPU drivers, CUDA, Docker and the NVIDIA Container Toolkit

2. 2

   Verify GPU access in Docker

   docker run --rm --gpus all nvidia/cuda:13.0.3-cudnn-runtime-ubuntu24.04 nvidia-smi
3. 3

   Pull and run the GPU-Jupyter image, then open http://localhost:8848 with the printed token

   docker run --gpus all -it -p 8848:8888 -v "$(pwd)/data:/home/jovyan/work" -e GRANT_SUDO=yes -e JUPYTER_ENABLE_LAB=yes --user root cschranz/gpu-jupyter:v1.11_cuda-13.0_ubuntu-24.04_python-only

Things to know

• !Requires an NVIDIA GPU with drivers, CUDA, Docker and the NVIDIA Container Toolkit installed.
• !The image is large and the first download can take a long time.

Topics

FAQ

What is gpu-jupyter?

A Docker image that gives you a GPU-enabled JupyterLab with PyTorch, TensorFlow and a data science toolstack for reproducible deep learning work. GPU-Jupyter is a Docker-based JupyterLab environment built on NVIDIA's official CUDA image and the Jupyter Docker Stacks. It bundles PyTorch, TensorFlow and a set of data science packages, so you can run GPU-accelerated notebooks without installing CUDA libraries and frameworks by hand.

Is gpu-jupyter free to use? What license does it have?

gpu-jupyter is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.

Is gpu-jupyter still maintained?

Updated within 6 months. The last commit to gpu-jupyter was on Aug 11, 2026.

What can I use gpu-jupyter for?

Running GPU-accelerated deep learning experiments in notebooks on a local machine or server. Sharing a reproducible research environment with collaborators or in publications. Setting up a standard data science workspace on a GPU host without manual driver and library setup. Building a team-specific image by layering custom packages on top of the base image.

How do I install gpu-jupyter?

Install the requirements: NVIDIA GPU drivers, CUDA, Docker and the NVIDIA Container Toolkit Verify GPU access in Docker: docker run --rm --gpus all nvidia/cuda:13.0.3-cudnn-runtime-ubuntu24.04 nvidia-smi Pull and run the GPU-Jupyter image, then open http://localhost:8848 with the printed token: docker run --gpus all -it -p 8848:8888 -v "$(pwd)/data:/home/jovyan/work" -e GRANT_SUDO=yes -e JUPYTER_ENABLE_LAB=yes --user root cschranz/gpu-jupyter:v1.11_cuda-13.0_ubuntu-24.04_python-only

Where can I get gpu-jupyter?

The source code and installation instructions are on GitHub: https://github.com/iot-salzburg/gpu-jupyter