Dragon - Python API

Dragon is a computation graph based distributed deep learning framework.

For using it, import as follows:

import dragon

Style Orientation

However, it will not help you much because Dragon is designed without systemic API.

We have extended it with FOUR Programming Styles:

TinyDragon

TinyDragon takes a very light weight programming style.

Our goal is to reduce extra unnecessary structures or interfaces. Therefore, in addition to feed or fetch, the last thing is designing a objective function through available operators.

It mainly uses the following components:

Caffe

Caffe is implemented basing on the backend of Dragon thorough native Python language.

Our work is very different from the official Python wrappers, a.k.a, the PyCaffe, which comes from the exports of BoostPython based on C++ language.

It mainly uses the following components:

Theano

Theano is an inception of the modern deep learning frameworks.

We implement it based on the backend of Dragon thorough native Python language. All operators are compiled offline, which is more similar to TensorFlow but not the original Theano.

It mainly uses the following components:

TensorFlow

COMING SOON……

Table of Contents

Quick Shortcut

dragon.config The global configures.
dragon.ops The exhibition of available operators.
dragon.updaters The exhibition of available updaters.
dragon.memonger The extreme memory optimizer.

Packages

dragon.core The core package.
dragon.utils The utils package.
dragon.operators The operators package.
dragon.tools The tools package.
dragon.vm The vm package.