Hi everyone,
First of all, thank you for this great open-source project and all the effort behind it!
I am the author of GPU-Jupyter, a GPU-enabled Jupyter environment built on top of Jupyter’s Docker Stacks, NVIDIA’s base images, and deep learning frameworks like PyTorch and TensorFlow. The goal of GPU-Jupyter is to provide an easy-to-set-up, reproducible deep learning environment for research and development.
Due to the project’s adoption by the open-source community (~700 GitHub stars and 400k+ image pulls), I am planning to migrate it to a more suitable namespace. Currently, the source-code repository is under the iot-salzburg organization on GitHub, and images are hosted in the personal cschranz namespace, which is not optimal.
Given its close integration with Jupyter’s Docker Stacks, I am considering whether the official Jupyter organization would be a suitable home for this repository. With the increasing demand for GPU support in data science environments, an official GPU-enabled Jupyter stack could be a valuable addition to the Jupyter ecosystem. This could create a synergy by combining Jupyter’s data science suite with NVIDIA’s GPU capabilities, making it easier to integrate GPU frameworks into Jupyter-based workflows.
Regarding integration, GPU-Jupyter is not just a single notebook or an extension of the Docker Stacks but provides also a framework for generating configured GPU-supported Dockerfiles and thus images. Therefore, it would make more sense as a separate repository within the Jupyter organization rather than being merged into the existing docker-stacks repository.
I would love to hear your thoughts on this and discuss whether such an integration would be beneficial for the Jupyter project.
Best,
Chris