You’ve probably heard about GPT-3, the best and worst, revolutionary and mediocre, genius and stupid language model built by OpenAI. Maybe you saw a really cool application. Maybe you saw a Twitter thread of all the ways it’s going to fail. This, however, is a summary of the original paper…

Earlier this week, Project Slippi announced a huge new update — they’re bringing rollback netcode to Super Smash Bros. Melee. Melee is a competitive fighting game, and one of the most popular games at that. For 19 years (it was released in 2001) the game has drawn in players from…

The secrets of how we keep information secure — Today’s world runs on data. We’ve generated more data in the past two years than in our entire history combined. This data revolution has also allowed us to communicate more than ever before. Humanity revolves around communication, especially private communication — but how do we keep communication data private? Gone…

How pregnancy works and how we’re trying to fix it. — Pregnancy is hard. It’s a process that brings about many issues, some small and some big. I’ve also been told that pregnancy is one of the most rewarding experiences, but I think that’s just wishful thinking. The War of Pregnancy Pregnancy in the human body is much like wartime operation for nations. Every part…

Music seems to be one of the fields that, at a surface level at least, AI just can’t seem to penetrate. AI is rapidly taking over so many fields, and there’s huge progress in music too! …

The superpower of the trained ear and an automated system to help develop it — It’d be incredibly naive to think a majority of people reading this know what ear training is. It’s a unique skill that only musicians have and train. It’s esoteric in essence, and its incredible specificity is marvelous and kind of terrifying. What if I told you that you have this…

Using Deep Learning combined with Q-learning to make the computer play Pong at superhuman levels — If you’re coming from my previous article, welcome! The basics you learned there are going to prove crucial to get to the juicy rewards today: Deep Q Learning. …

An in-depth look at classic Q-Learning and how it works. — How do human beings learn? Let’s picture a child (this example was taken from an excellent article by Thomas Simonini). The child approaches a fireplace. They feel warm and cozy. Reward += 1 The child wants more of that feeling — they touch the fireplace. Ouch! reward-= 1 . Now…

Using Convolutional Neural Networks to classify pictures of apples and oranges. — I have a lid but I’m not a can of paint I have a lens but I’m not a camera I’m in a socket but I’m not a plug I have an iris but I’m not a bunch of flowers I have lashes but I don’t have a whip. What…