Welcome to my projects showcase! This is where I highlight some of my best work.
PROJECT 01: Neural Network From Scratch
Project Type: Personal
Project Name: Rawtron
In this project I built a two layer deep multilayer perceptron from scratch without using any ML libraries such as TensorFlow PyTorch. I start by explaining how derivatives can be used to quantify the sensitivity of a function on its input. Then I implement the forward pass and back propagation on a single neuron. Finally I train a ~50 parameter network using gradient descent using an L2 loss function. Visit my GitHub to check out the details:
PROJECT 02: CHARACTER LEVEL LANGUAGE MODEL
Project Type: Personal
Project Name: Baby LLM
In this project I created a character level model that can predict new names given a data set of names. The first model I used is a simple statistical model based on the counts of bigrams. To predict a new character we simple query from the probability distribution. For the second model I used a simple one layer linear neural network with a cross-entropy loss function to train the network. Visit my GitHub to check out the details:
