tensorlite

A lightweight C++ tensor library capable to perform extremely fast tensor computations.

Features

• Made with ❤️ using modern C++
• Fully templated library
• Fully extensible
• Uses smart pointers for safe memory management
• Cross platform

Installation

Install tensorlite as follows

  git clone 
  cd tensorlite
  make

Usage/Examples

#include "Tensor.h"
#include "Tensor2D.h"
#include "Layer.h"
#include <iostream>
using namespace std;


Tensor<double> t1  = {{1,2,3,4}};
Tensor<double> t2 = {{5,6,7,8}};
Tensor<double> t3 = t1+t2;
cout << t3 << endl;

output:{ 6,8,10,12 }

Tensor2D<double> t5({t1,t1,t1,t2,t2,t1,t2,t1,t2,t1,t2});
cout << t5 << endl;

output:
Tensor2D {
{ 1,2,3,4 }
{ 1,2,3,4 }
{ 1,2,3,4 }
{ 5,6,7,8 }
{ 5,6,7,8 }
{ 1,2,3,4 }
{ 5,6,7,8 }
{ 1,2,3,4 }
{ 5,6,7,8 }
{ 1,2,3,4 }
{ 5,6,7,8 }
}

auto t6 = t5.transpose();
cout <<  t6 << endl;

output:
Tensor2D {
{ 1,1,1,5,5,1,5,1,5,1,5 }
{ 2,2,2,6,6,2,6,2,6,2,6 }
{ 3,3,3,7,7,3,7,3,7,3,7 }
{ 4,4,4,8,8,4,8,4,8,4,8 }
 }

auto t7 = t5*t6;
cout << t7 << endl;

output:
Tensor2D {
{ 30,30,30,70,70,30,70,30,70,30,70 }
{ 30,30,30,70,70,30,70,30,70,30,70 }
{ 30,30,30,70,70,30,70,30,70,30,70 }
{ 70,70,70,174,174,70,174,70,174,70,174 }
{ 70,70,70,174,174,70,174,70,174,70,174 }
{ 30,30,30,70,70,30,70,30,70,30,70 }
{ 70,70,70,174,174,70,174,70,174,70,174 }
{ 30,30,30,70,70,30,70,30,70,30,70 }
{ 70,70,70,174,174,70,174,70,174,70,174 }
{ 30,30,30,70,70,30,70,30,70,30,70 }
{ 70,70,70,174,174,70,174,70,174,70,174 }
}

Layer<float> l(5,5);
cout << l1 << endl;
output:
Layer {
{ -1.15307,-0.228105,0.167813,0.86608,-0.62505 }
{ -0.294773,0.292243,-1.50031,0.93002,-0.470464 }
{ 0.838648,0.320883,-0.149071,-0.373821,-0.30948 }
{ 0.45579,-0.565829,-0.139341,-1.63598,0.288001 }
{ 0.674396,0.281527,-0.044332,-0.312636,1.20646 }
}

l1.multiply(100);
output:
Layer {
{ -115.307,-22.8105,16.7813,86.608,-62.505 }
{ -29.4773,29.2243,-150.031,93.002,-47.0464 }
{ 83.8648,32.0883,-14.9071,-37.3821,-30.948 }
{ 45.579,-56.5829,-13.9341,-163.598,28.8001 }
{ 67.4396,28.1527,-4.4332,-31.2636,120.646 }
}

Layer<double> l2 (5,1);
Layer<double> ll = l1.multiply(l2);
cout << ll << endl;
output:
Layer {
{ -44.5445 }
{ 200.929 }
{ 128.982 }
{ -51.4861 }
{ -38.3569 }
}

Running Tests

To run tests, run the following command

  ./test_tensor
  ./test_tensor2D

Contributing

Contributions are always welcome!

See contributing.md for ways to get started.

Please adhere to this project's code of conduct.

The following are some pressing features that need to be completed:

• Make Layer class
• Activation Functions
• Example for Forward Propagation
• Function to calculate gradients in back propagation
• Support Multi-Threading and GPU computations

Authors

• @gautamsharma

Appendix

Any additional information goes here