Prompt Optimization Via Diffusion Language Models
• Diffusion-based framework refines system prompts via masked denoising in an iterative manner. • Conditions on interaction traces: user queries, model responses, and optional feed
Prompt-Optimization