AI Flow

Introduction

AI Flow, which offers various reusable operators & processing units in AI modeling, helps AI engineer to write less, reuse more, integrate easily.

Install

pip install aiflow

Concepts

Operators VS. Units

Ideally, we agree:

• An Operator would contain lot of units, which will be integrated into airflow for building non-realtime processing workflow;
• A Unit is a small calculation unit, which could be a function, or just a simple modeling logic, and it could be picked as bricks to build an operator. Besides, it could be reused anywhere for realtime calculation.

Classes

Operators

MongoToCSVOperator

Elastic2CSVOperator

RegExLabellingOperator

Units

Doc2VecUnit

Doc2MatUnit

Tests & Examples

Example: Use Units to Build Your Castle

Example: Working with Airflow

In tests/docker/ folder, we provide examples on how to use aiflow with airflow. It is a docker image, you could simply copy and start to use it!

In project root directory, run commands first:

docker-compose up --build aiflow

Then open localhost:8080 in your browser, you can see all the examples aiflow provided! Note: both the default username & password are admin

Enjoy!

Contribution