Skip to content

Latest commit

 

History

History
14 lines (12 loc) · 900 Bytes

README.md

File metadata and controls

14 lines (12 loc) · 900 Bytes

mini-dnn

a vs2017 C++ demo of deep neural networks. It is implemented purely in C++, only depends on Eigen lib.

How to build

Win10 (Visual Studio2017)

A Visual Studio solution mini-dnn.sln has been generated. Open the solution file and build all projects, the static lib and examples have been built. We can ref the blog which introduces the configuration of Eigen under VS2017

Download and unzip mnist dataset by the baiduyun code 45x2 in mini-dnn/mini-dnn Run demo.cpp

Result

simple neural network with 3 FC layers can obtain 0.97+ accuracy on MNIST testset. you can select layers by yourself.

Acknowledge

Thanks for the repo