Skip to content
forked from dmlc/minpy

Pure NumPy practice with third-party operator integration

License

Notifications You must be signed in to change notification settings

sneakerkg/minpy

 
 

Repository files navigation

MinPy

Build Status PyPI version Docs

This repository aims at prototyping a pure NumPy interface above MXNet backend. The key features include:

  • Autograd support. Automatic gradient generation.
  • Seamless MXNet symbol integration.
  • Graceful fallback for missing operations to NumPy.
  • Transparent device and partition specification.

MXNet version

Currently both MXNet and MinPy are going through rapid development. MinPy is not guaranteed to work with all MXNet versions.

This version of MinPy is tested to work with MXNet at 3fb29a3.

How to get started

The project is still a work-in-progress. You could look at this tutorial to understand its concept. Documents are hosted here.

Easy installation

pip install minpy

What we could achieve

In one word, if you have NumPy code, you could replace the import by:

import minpy.numpy as np

Other numpy code remain the same. And you could have:

  • Auto differentiation support.
  • Speed up with some operations executed on GPUs.
  • Missing operations will fall back to NumPy version automatically.
  • Hybrid programming with efficient MXNet's symbol and flexible MinPy arrays.

About

Pure NumPy practice with third-party operator integration

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 99.5%
  • Shell 0.5%