This website is not finished :)
A gallery of many transformer variants, specifically long sequence transformers to try to benchmark and open source research. (Longformer, Recurrent Memory Transformer etc...)
A DQN based AI agent that was able to play all 57 atari games at a superhuman level. It was a big milestone in reinforcement learning.
A deep learning stock trading bot. It uses many transformer variants integrated with distributed reinforcement learning algorithms to try to make sense of the news feed from google news.
A reimplementation of the ICM paper that endows DQN agents with curiosity, incentivizing them to explore unfamiliar things in their environment.
A generative AI that synthesizes images from noise. Reimplemented from the landmark paper that became the predecessor to current image AI generators.
An algorithm that learned to play atari games entirely from self play using Q learning