bayesian optimization

This work introduces a novel language-in-the-loop framework that leverages large language models to transform free-form textual feedback into scalar utilities, enabling principled bayesian optimization with Gaussian Processes.