Preprint
Aug 2026
Naive Prompt Optimization: Rethinking the Need for Complex Prompt Search
In interactive games, NPO remains broadly competitive with GEPA, while GRPO performs better on some tasks less amenable to prompt optimization.
Yuan Chang, Xiao-Qi Chen
· 0 citations