A benchmark to evaluate language models on questions I've previously asked them to solve. - View it on GitHub