jupyterlab-gpu — bare JupyterLab on GPU
A clean JupyterLab environment on top of CUDA 12 — no opinionated ML libs preinstalled, just Jupyter and Python. Bring your own pip install.
Image tag
docker.io/manvarharsh/jupyterlab-gpu:cuda12
What’s in this image
- Base:
nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04 - Python 3.10 (conda)
- JupyterLab + JupyterHub
- Podstack authenticator
- OpenSSH server
- Build essentials and common CUDA dev headers
Default ports
| Port | Service |
|---|---|
| 22 | SSH |
| 8000 | JupyterHub |
Use cases
- A blank-slate notebook environment when none of the preset images fits
- Teaching / coursework where students install their own stack
- Quick sandbox for trying a new library on GPU
Environment variables
| Variable | Description |
|---|---|
ENABLE_SSH | Enable SSH server |
ENABLE_JUPYTERHUB | Enable JupyterHub on port 8000 |
PODSTACK_API_URL | Backend URL for JupyterHub token validation |
SSH_PUBLIC_KEY | Public key for SSH |
Persistence
Mount at /data. Notebooks and any installed packages (pip install --target=/data/site-packages) live there.
See also
- Conventions and shared environment variables
- pytorch and tensorflow for opinionated alternatives
- Accessing output files in the browser