Principia Labs (source code)

= Principia Labs
{c}

https://www.principialabs.org
> We combine large-scale pretraining with reinforcement learning to create models that can rederive and learn from the entire corpus of human mathematics. Our goal is automated mathematical discovery: AI that does the creative, generative work that was previously only possible for the world's best researchers—and can be deployed on the hardest problems in science and engineering.