This repository contains a self-project in which I designed and implemented a neural network library from scratch. The project includes an automatic differentiation engine for performing backpropagation through dynamically constructed computational graphs, demonstrating foundational machine learning concepts. The library allows the creation, training, and evaluation of neural network models for various tasks, such as classification.
- Automatic Differentiation Engine: Implements backpropagation through dynamic computational graphs, which allows gradients to be computed automatically for neural network training.
- Neural Network Library: Built atop the autograd engine, this library enables the creation and training of different neural network architectures.
- Support for Classification: A variety of classification tasks can be tackled, reinforcing key deep learning principles through practical implementation.
- Modular Design: The project is organized into separate components, making it easy to extend and experiment with different parts of the codebase.