Public Repository

Last pushed: 6 months ago
Short Description
nvidia-docker image based on nvidia/cuda:9.0-cudnn7-runtime with Jupyter Stacks as framework.
Full Description

Note: Apologies to anyone who pulled this early on. The password configuration was not functioning. All is fixed now.
Make sure to restart the container if you'd like to use a password on your notebook

Make sure to run this container with nvidia-docker! For more information please see this link for info.
https://github.com/NVIDIA/nvidia-docker

If you need help setting up an Nvidia powered ubuntu server (nvidia-docker requires Ubuntu) please see my guide, especially parts 2 and 3.
http://dimakarpa.com/coding/ml-server-2/

cudaconda is big. It contains Anaconda3 (python 3.6), CUDA and cuDNN but relies heavily on Jupyter Notebook Stacks so it should be very familiar to anyone who uses Jupyter Notebook Scipy or other Jupyter images.

At the time of this writing (May 2018), you must compile TensorFlow in order to use CUDA toolkit 9.1 or 9.2. Hence, this image uses nvidia CUDA toolkit 9.0 with cuDNN 7 as a base. This can be altered by changing the first line of the Dockerfile to nvidia/cuda:9.1-cudnn7-runtime or nvidia/cuda:9.2-cudnn7-runtime.

Many of the Jupyter Notebook Stacks environmental variables apply to this image as well.

Big thank you to Nvidia and Jupyter!

Docker Pull Command
Owner
dimak415