GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning

4 points | by handfuloflight 9 hours ago