Skip to content
We move forward to MXNet
Find file
New pull request
Pull request Compare This branch is 3 commits ahead of dmlc:master.
Latest commit 3b670af @antinucleon update
Failed to load latest commit information.
.gitignore fix rand_crop Merge remote-tracking branch 'upstream/master'

We move forward to MXNet !

Dear users,

Thanks for using and supporting cxxnet. Today, we finally make a hard but exciting decision: we decide to deprecate cxxnet and fully move forward to next generation toolkit MXNet.

Please check the feature highlights, speed/memory comparation and examples in MXNet.

cxxnet developers,

28th, Sep, 2015

Note: We provide a very simple converter to MXNet. Check guide to see whether your model is able to be converted.


CXXNET is a fast, concise, distributed deep learning framework.

Note: We changed OFFICIAL repo of CXXNET to Distributed (Deep) Machine Learning Common New OFFICIAL address is: This is NOT the OFFICIAL repo.


Feature Highlights

  • Lightweight: small but sharp knife
    • cxxnet contains concise implementation of state-of-art deep learning models
    • The project maintains a minimum dependency that makes it portable and easy to build
  • Scale beyond single GPU and single machine
    • The library works on multiple GPUs, with nearly linear speedup
    • THe library works distributedly backed by disrtibuted parameter server
  • Easy extensibility with no requirement on GPU programming
    • cxxnet is build on mshadow
    • developer can write numpy-style template expressions to extend the library only once
    • mshadow will generate high performance CUDA and CPU code for users
    • It brings concise and readable code, with performance matching hand crafted kernels
  • Convenient interface for other languages
    • Python interface for training from numpy array, and prediction/extraction to numpy array
    • Matlab interface


  • 24-May, 2015: Pretrained Inception model with 89.9% Top-5 Correctness is ready to use.
  • 09-Apr, 2015: Matlab Interface is ready to use

Backbone Library

CXXNET is built on MShadow: Lightweight CPU/GPU Tensor Template Library

  • MShadow is an efficient, device invariant and simple tensor library
    • MShadow allows user to write expressions for machine learning while still provides
    • This means developer do not need to have knowledge on CUDA kernels to extend cxxnet.
  • MShadow also provides a parameter interface for Multi-GPU and distributed deep learning
    • Improvements to cxxnet can naturally run on Multiple GPUs and being distributed


  • Copy make/ to root foler of the project
  • Modify the config to adjust your enviroment settings
  • Type ./ to build cxxnet
Something went wrong with that request. Please try again.