Reproduce “Quality-Diversity through AI Feedback” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Herbie Bradley, Andrew Dai, Hannah Teufel, Jenny Zhang, Koen Oostermeijer, Marco Bellagente, Jeff Clune, Kenneth Stanley and 2 more
QDAIF uses quality-diversity search to generate and retain strong candidates across user-specified diversity bins while language models assess quality and bin membership. It applies MAP-Elites with LMX mutation and AI feedback. In Opinions and Stories, it outperformed baselines on AI-computed QD score. Human evaluations found competitive performance but lower ratings than Fixed-Few-Shot; results are limited to specified axes and AI-assessed quality.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.