This repository contains my ongoing experiments with machine learning. The goal is to understand the fundamental concepts of machine learning (neural networks, LLMs, research papers, etc) and implement them from scratch.
- gridworld – A reinforcement learning implementation demonstrating Q-learning to solve a GridWorld environment, balancing exploration and exploitation.
- MNIST – A feedforward neural network for handwritten digit recognition, trained on the MNIST dataset with support for multiple hidden layers.
- trigram – A character-level trigram model implemented from scratch to generate names, demonstrating basic statistical language modeling.
- iris – A neural network trained on the Iris dataset using backpropagation, capable of classifying flowers into three species.
- xor – A simple MLP solving the XOR problem using backpropagation with configurable neurons.