Table of contents

pytorch — PyTorch + Jupyter

A GPU-ready PyTorch environment with JupyterHub, SSH, and common computer-vision libraries. Use this when you want a clean PyTorch sandbox without committing to a specific framework on top.

Image tag

docker.io/manvarharsh/pytorch:cuda12

What’s in this image

  • Base: nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04
  • Python 3.10 (conda)
  • PyTorch (CUDA 12), torchvision, torchaudio
  • ultralytics (YOLO), opencv-python-headless
  • NumPy, Pandas, Matplotlib, scikit-learn
  • JupyterHub with the Podstack authenticator
  • OpenSSH server

Default ports

PortService
22SSH
8000JupyterHub

Use cases

  • General-purpose PyTorch development
  • Training computer-vision models (detection, segmentation, classification)
  • Notebook-driven ML experimentation
  • A solid base for installing any PyTorch ecosystem package

Environment variables

VariableDescription
ENABLE_SSHEnable SSH server
ENABLE_JUPYTERHUBEnable JupyterHub on port 8000
PODSTACK_API_URLBackend URL for JupyterHub token validation
SSH_PUBLIC_KEYPublic key for SSH

Persistence

Mount at /data. Notebooks under /data/notebooks/, models under /data/models/, datasets under /data/datasets/.

See also