Tools Used: C++, Graphviz
Topics: Deep Learning, Machine Learning
MicroGradCpp is a C++ implementation of MicroGrad, a minimalistic deep learning library. It includes an API to calculate gradients, and to train neural networks in single neuron level. It is a very simple library, and it is very easy to use. It is a very good tool to learn how neural networks work, and how they are trained.
The project is a replicate of Andrej Karpathy’s MicroGrad with a few extra features and in C++.
- Custom value and gradient system.
- Graphviz support for visualizing the graph.
- Fully working neural network implementation.
Here is an implementation of a single neuron:
Which gives the following graph: