Building an MLP from Scratch

This repository contains a step-by-step implementation of a Multi-Layer Perceptron (MLP) built from the ground up using basic PyTorch operations, without relying on high-level abstractions like nn.Linear or nn.Sequential.

It is intended as a deep learning educational project to gain a low-level understanding of how neural networks are built, trained, and evaluated—focusing on concepts like weight initialization, forward propagation, loss calculation, and backpropagation.

---

📓 Notebook: Building_MLP_From_Scratch.ipynb

🧩 Features Implemented:

• Manual creation of weights and biases
• Custom forward pass using torch.tanh and matrix operations
• Logits computation and softmax-based prediction
• Manual loss calculation using Cross Entropy
• Manual gradient computation and parameter updates
• Training and validation loops
• Seed control for reproducibility

📊 Dataset

This notebook uses a small character-level dataset to predict next characters (inspired by name-generation tasks). You can swap in your own datasets for experimentation.

---

Acknowledgements

This project was heavily inspired by Andrej Karpathy and his low-level deep learning explorations. Special thanks for his educational content and clean explanations on building neural networks from scratch.