Skip to content

Latest commit

 

History

History
62 lines (45 loc) · 1.94 KB

Docker.md

File metadata and controls

62 lines (45 loc) · 1.94 KB

OpenML-Tensorflow container

The docker container has the latest version of OpenML-Tensorflow downloaded and pre-installed. It can be used to run TensorFlow Deep Learning analysis on OpenML datasets. This document contains information about:

Usage: how to use the image

Using Locally Stored Datasets: mounting datasets from the local cache

Environment Variables: setting the cache directory path

Usage

These are the steps to use the image:

  1. Pull the docker image
docker pull openml/openml-tensorflow:latest
  1. If you want to run a local script, it needs to be mounted first. Mount it into the 'app' folder:
docker run -it -v PATH/TO/CODE_FOLDER:/app openml/openml-tensorflow /bin/bash

You can also mount multiple directories into the container (such as your code file directory and dataset directory ) using:

docker run -t -i -v PATH/TO/CODE_FOLDER:/app -v PATH/TO/DATASET_FOLDER:/app/dataset openml/openml-tensorflow /bin/bash
  1. Please make sure to give the correct path to the dataset. For example,
openml_tensorflow.config.dir = 'dataset/Images'
  1. Run your code scripts using, for example:
 python docs/Examples/tf_image_classification.py

Using Locally Stored Datasets

If you don't want to download the dataset each time you run your script, you can mount your dataset saved in your local cache directory to the container.

Example Usage

  1. Mount the dataset to the 'app/dataset' folder
docker run -t -i -v PATH/TO/CODE_FOLDER:/app -v PATH/TO/DATASET_FOLDER:/app/dataset openml/openml-tensorflow /bin/bash
  1. Set correct path to the dataset.
openml_tensorflow.config.dir = '/app/dataset/Images'

Environment Variable

You can configure the cache directory to control where 'OpenML' datasets are downloaded and cached.

cache_dir = "/app/.openml"
openml.config.set_root_cache_directory(cache_dir)