POET: אופטימיזציה של העדפות ליצירת תמונות מטקסט

תקציר מקורי באנגליתarXiv:2510.12041v3 Announce Type: replace Abstract: Recent advances in text-to-image (T2I) generation have achieved impressive results, yet existing models often struggle with simple or underspecified user prompts due to a distributional gap with their descriptive training captions. This frequently leads to suboptimal image-text alignment, aesthetics, and overall visual quality. To bridge this gap, we propose POET (\textbf{P}reference \textbf{O}ptimization for \textbf{E}nhanced \textbf{T}ext-to-Image generation), an automated prompt rewriting framework that leverages large language models (LLMs) to refine user inputs before feeding them into frozen T2I backbones. POET introduces a carefully designed composite reward system and an iterative Direct Preference Optimization (DPO) training pipe