Sampling

A momentum-based, sampling-driven method for scaling textual gradient optimization in LLM prompt engineering, improving performance and efficiency across diverse NLP tasks.