microgradpp

A pure C++ machine learning library inspired by Andrej Karpathy's micrograd

Features • Getting Started • Examples • Usage • Contributing

---

Are you tired of using outdated hard-to-use C++ machine learning libraries???

Welcome to microgradpp! This repository is inspired by Andrej Karpathy's micrograd. microgradpp is a pure C++ machine learning library designed to make machine learning accessible to everyone.

---

What is microgradpp?

microgradpp (micrograd++) is a C++ implementation of an automatic differentiation (autograd) engine with a neural network library built on top. It implements backpropagation (reverse-mode autodiff) over a dynamically built computational graph and provides tools to build and train neural networks.

Why microgradpp?

Feature	microgradpp	Python micrograd
Language	Modern C++17	Python
Performance	Fast (compiled)	Interpreted
Computer Vision	✅ OpenCV support	❌
Header-Only Option	✅	✅

---

Features

• ✅ Pure C++: Entirely implemented in C++ for high performance
• ✅ Inspired by micrograd: Brings the simplicity and educational value of micrograd to the C++ ecosystem
• ✅ Accessible Machine Learning: Designed to be easy to use, even for those new to machine learning or C++
• ✅ Computer Vision Support: OpenCV integration for image-based neural networks
• ✅ Header-Only Option: Can be used as a header-only library
• ✅ Tensor Class: Simplified data loading and manipulation
• ✅ Multiple Activation Functions: ReLU, Tanh, Sigmoid, and more

---

Getting Started

Prerequisites

• CMake: Version 3.15 or higher
• C++ Compiler: Supports C++17 standard
• OpenCV: For visualization (optional)

Building the Library

1. Clone the repository:

   git clone https://github.com/gautam-sharma1/microgradpp.git
   cd microgradpp

2. Create a build directory:

   mkdir build
   cd build

3. Configure the project with CMake:

   Build microgradpp only:

   cmake ..

   Build with examples and tests:

   cmake .. -DBUILD_EXAMPLES=ON -DBUILD_TESTS=ON

   Release build:

   cmake .. -DCMAKE_BUILD_TYPE=Release

4. Build the project:

   make

Using as Header-Only Library

microgradpp is also available as a header-only library. Simply include the headers from include/ in your project:

#include "microgradpp/Value.hpp"
#include "microgradpp/MLP.hpp"

Running the Example

After building, run the provided example:

./m++

---

Usage

Here's a brief example of how to use microgradpp:

#include <iostream>
#include "Value.h"

int main() {
    // Create values
    auto a = microgradpp::Value::create(2.0);
    auto b = microgradpp::Value::create(3.0);
    
    // Build computation graph
    auto c = a * b;
    
    // Backpropagation
    c->backProp();
    
    std::cout << "a->grad: " << a->grad << std::endl;  // 3.0
    std::cout << "b->grad: " << b->grad << std::endl;  // 2.0

    return 0;
}

---

Examples

Head over to the examples/ directory to explore various examples. To build examples:

cd build
cmake -DBUILD_EXAMPLES=ON .. && make
cd examples
./example_mlp  # Run MLP example

Multi-Layer Perceptron (MLP)

A simple multi-layer perceptron defined in mlp.cpp. Running it for 50 iterations:

Computer Vision Example 🐕

The example in images.cpp trains a neural network to learn a German Shepherd puppy face:

---

Project Structure

microgradpp/
├── include/          # Header files
├── examples/         # Example programs
├── tests/            # Unit tests
├── public/           # Images and assets
├── scripts/          # Utility scripts
├── CMakeLists.txt    # CMake configuration
├── main.cpp          # Main example
└── README.md

---

API Overview

Value Class

The core building block for automatic differentiation:

auto x = microgradpp::Value::create(2.0);
auto y = microgradpp::Value::create(3.0);
auto z = x * y + x->pow(2);
z->backProp();
// x->grad and y->grad now contain gradients

Neural Network Classes

// Create a Multi-Layer Perceptron
// Input: 3 features, Hidden: 4 neurons, Output: 1 neuron
MLP model(3, {4, 4, 1});

// Forward pass
auto output = model(input_data);

// Training
model.zeroGrad();
loss->backProp();
// Update weights...

---

Contributing

We welcome contributions to microgradpp! Here are areas where you can help:

TODO

• Modify CMakeLists to Add a Flag to Build Tests: ✅ Done
• Make a Tensor Class: ✅ Done
• Add an Activation Function Enum or Class: ✅ Done
• Make an Abstract Base Class for Layer and Value: ✅ Done
• CI/CD Pipeline: Develop a pipeline using GitHub Actions
• More Examples: Add tutorials and use cases
• Python Interface: Create Python bindings

How to Contribute

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

---

Acknowledgments

• Andrej Karpathy: For the original micrograd library and inspiration
• The C++ community for feedback and contributions

---

Citation

If you use microgradpp in your research or projects:

@software{microgradpp,
  author = {Gautam Sharma},
  title = {microgradpp: A C++ Machine Learning Library},
  year = {2024},
  url = {https://github.com/ggsharma/microgradpp}
}

---

License

This project is licensed under the GNU GPL-3.0 License - see the LICENSE file for details.

---

Author

Gautam Sharma - gsharma.dev

---

If you find microgradpp useful, please consider giving it a ⭐!