sinanozel/kubyterlab-img repository overview

This is a custom JupyterLab specifically for working with GPUs, tailored towards image generations and working with images. See here for more info: https://github.com/sinan-ozel/jupyterlab-on-kubernetes/tree/main/kubyterlab-img/kubyterlab-img/⁠

With docker: docker run --gpus all --rm -it sinanozel/kubyterlab-llm:25.09 python -c "import torch; print('CUDA available:', torch.cuda.is_available());"

If you see True, you can use this JupyterLab container with CUDA, meaning that you can use the host computer's GPU memory for AI work.

The host machine needs to have CUDA and nvidia-smi and nvidia-container-toolkit installed.

With Kubernetes, start a cluster with some nodes with GPUs and node images that have CUDA and nvidia-smi and nvidia-container-toolkit installed. Consider using the provision configuration from my IaC template repo, https://github.com/sinan-ozel/iac⁠