/docker-tensorflow-haskell-ihaskell

Image for convenient development with Tensorflow + Haskell/IHaskell

docker-tensorflow-haskell-ihaskell

Images for convenient development for Tensorflow + Haskell/IHaskell.

Has open source versions:gpu, cpu. And advanced closed source versions: adv-gpu, adv-cpu.

Includes Jupyter notebooks, IHaskell environment, Haskell Tensorflow bindings and Tensorflow libraries.

We encourage to use GPU images wherever possible, so latest images defaults to GPU version.

Quickstart - running GPU image

To use Tensorflow with GPU you need to:

  1. Install nvidia-docker (https://github.com/NVIDIA/nvidia-docker)
  2. Use GPU versions of images: nvidia-docker run -e PASSWORD=<password> -p 8888:8888 <image_name>:<tag>
  3. Open http://<host>:8888 URL
  4. Enter password
  5. Open IHaskell -> notebooks -> IHaskell.ipynb

Quickstart - running CPU image

  1. Use CPU versions of images: docker run -e PASSWORD=<password> -p 8888:8888 <image_name>:<tag>
  2. Open http://<host>:8888 URL
  3. Enter password
  4. Open IHaskell -> notebooks -> IHaskell.ipynb

Environment

Environment variables:

required

  • PASSWORD - set password to connect to Jupyter Notebook. If password is not set, image going to configure URL token that is going to be posted to containers STDOUT at the end of starting process.

optional

  • PORT - set port for Jupyter inside container, but beware that Dockerfiles open 8888 by default configuration.