ML Runtime

This is a Ubuntu Docker image with various machine learning runtime.

It is based on the Base Runtime.

Major Packages

All the variants can be retrieved at Docker Hub.

• Variants:

  • CPU variant: based on haowenxu/base-runtime:cpu
  • GPU variant: based on haowenxu/base-runtime:gpu

Installation

Generate the Dockerfile

We use configure.py to generate the Dockerfile according to configurations.

You should first install the dependencies of configure.py:

pip install -r requirements.txt

Then for example, you can use the following statement to generate the CPU variant Dockerfile:

python configure.py \
    -c config/cpu.yml \
    -c config/tensorflow1.12.yml

Build the Docker Image

After generate the Dockerfile, you can build the docker image by:

docker build .

Usage

The basic usage of this docker image is shown as below. Note that you may specify the TZ environmental variable, such that the container will have the correct timezone:

docker run \
    -ti --rm \
    -e TZ=Asia/Shanghai \
    haowenxu/ml-runtime:cpu \
    /bin/bash