Deep Machine Learning Package [dmlpack] > header-only cpp file > no dependencies except -std=c++11 > inbuilt matrix lib [ Currently in process of adding SIMD instruction and CUDA support ] > clean interface > tons of testing TO DO : > Impelement better structure for organizing the code into seperate classes. Make an abstract class make everything inherit from it ? > Add CUDA and SSE support to the matrix lib > Implement covnets testing : testing will be done by using catch. The user should not have to worry about testing. So keep it seperate. dmlpack.h will not have any dependence on testing.