AI
逐次プロンプト最適化のためのコントラスティブ・リフレクション
Contrastive Reflection for Iterative Prompt Optimization
arxiv2026年7月1日
日本語要約
本論文は、大規模言語モデルで使用されるプロンプトを逐次的に最適化するための新しい手法「コントラスティブ・リフレクション」を提案します。この手法は、LLMとの対話や制御の方法を大幅に改善する可能性があります。
English Summary
This paper presents 'Contrastive Reflection,' a new technique for iteratively optimizing prompts used with large language models. This method could significantly improve the efficiency and effectiveness of how we interact with and control LLMs.