eywalker/tensorflow repository overview

This is a collection of images offering a complete TensorFlow + Python 3 development environment in Ubuntu OS with all essential scientific Python packages such as numpy, matplotlib, and scipy. All images are built on Python 3.5 + Ubuntu 16.04 LTS environment.

⁠Images

⁠Python 3.5 + Ubuntu 16.04 LTS

⁠CPU images

Slim images designed to run on standard machines:

• latest, v0.12.1 - Latest stable TensorFlow v0.12.1 compiled from the source.
• v1.0.0rc0 - Latest release candidate TensorFlow v1.0.0-rc0 compiled from the source.

⁠GPU images

Images designed to be used with nvidia-docker:

• cuda, v0.12.1-cuda8.0-cudnn5 - Latest stable release TensorFlow v0.12.1 compiled against CUDA 8.0 + cuDNN 5
• v1.0.0rc0-cuda8.0-cudnn5 - Latest release candidate TensorFlow v1.0.0-rc0 compiled against CUDA 8.0 + cuDNN 5
• v0.11.0rc0-cuda8.0-cudnn5 - Release candidate TensorFlow v0.11.0rc0 compiled against CUDA 8.0 + cuDNN 5
• v0.10.0-cuda8.0-cudnn5 - Stable release TensorFlow v0.10.0 compiled against CUDA 8.0 + cuDNN 5
• v0.10.0rc0-cuda8.0-cudnn5 - Previous release candidate TensorFlow v0.10.0rc0 compiled against CUDA 8.0 + cuDNN 5

Dockerfile and Makefile to build images can be found at https://github.com/eywalker/tensorflow-jupyter-docker⁠. Please file issue reports and requests on the issue page of GitHub.