Micrograd Implementation

This repository contains an implementation of a simple autograd engine and neural network, inspired by Andrej Karpathy's "micrograd" project.

Overview

The project is a lightweight autograd engine that supports backpropagation and neural network implementation.

Features

• Autograd Engine: Implements custom backpropagation for scalar computations.
• Basic Neural Network: Includes simple feedforward layers with tanh activation.
• Gradient Descent: Updates parameters using backpropagation.
• Visualization: Uses graphviz for computational graph visualization.

File Structure

• micrograd1.ipynb: Main implementation of autograd and neural network.
• README.md: Documentation for the project.

Acknowledgments

Inspired by Andrej Karpathy’s micrograd project.

License

This project is open-source under the MIT License.