Docker for Fione

What is Fione?

Fione is very powerful and easily deployable Enterprise AI Platform. You can install and run Fione quickly on any platforms, which have Docker runtime environment. Fione is provided under Apache license.

Getting Started

(For more details, see Quick Start)

Prerequisite

• Docker

Start Fione

First of all, start Fione with the following command:

$ git clone https://github.com/codelibs/docker-fione.git -b v14.1.0
$ cd docker-fione/compose
$ docker compose up

Create Project

1. Access http://localhost:8080/admin/easyml/ and login as admin user (password is admin).
2. Type Project Name and Training Data. (for example, you can download train.csv for Titanic)
3. Click Create button.

Run Machine Learning

1. Select Predicted Column and Input Columns in Data Analysis page. (for Titanic data, the predicted column is Survived and the type is Enum)
2. Select Max Execution Time.
3. Click Run button.

You can check the result in Prediction Summary page. For Machine Learning experts, you can also use H2O Flow in http://localhost:54321/flow/.

Start Serving API with Docker

Fione provides Serving API container with Docker. To download Dockerfile for Serving API, you can find it at Serving page for each Model page.

Stop Fione and Clean Up

$ docker compose down

Some data remains in docker volumes. To delete them, run the following command:

$ docker volume rm compose_es-data compose_es-dict compose_fess-data compose_minio-data

Docker Images

• snapshot (snapshot/Dockerfile)
• 15.3.2, 15.3, 15, latest (15.3/Dockerfile)
• 14.6.0, 14.6, 14 (14.6/Dockerfile)
• 14.1.0, 14.1 (14.1/Dockerfile)
• 13.10.0, 13.10, 13 (13.10/Dockerfile)
• 13.9.0, 13.9 (13.9/Dockerfile)
• 13.8.0, 13.8 (13.8/Dockerfile)
• 13.7.0, 13.7 (13.7/Dockerfile)
• 13.6.3, 13.6 (13.6/Dockerfile)

Development

Build Fione

To build docker images, run as below:

$ docker build --rm -t ghcr.io/codelibs/fione:<tag name> ./<version_dir>/

License

Apache License 2.0