📄 中文摘要
在使用大型语言模型（LLM）时，用户的接受标准对模型的输出质量有显著影响。明确的标准可以帮助模型更好地理解用户的需求，从而生成更符合预期的结果。通过设定具体的评价标准，用户能够引导模型的生成过程，提升交互的有效性和满意度。这种方法不仅适用于文本生成，还可以扩展到其他应用场景，如代码生成和数据分析等。研究表明，用户在与 LLM 交互时，清晰的目标和标准能够显著改善模型的表现。
用户首先定义接受标准时，LLM 的表现最佳
出处: LLMs work best when the user defines their acceptance criteria first
发布: 2026年3月7日
📄 English Summary
LLMs work best when the user defines their acceptance criteria first
The performance of large language models (LLMs) significantly improves when users define their acceptance criteria upfront. Clear criteria help models better understand user needs, leading to outputs that align more closely with expectations. By establishing specific evaluation standards, users can guide the generation process of the model, enhancing the effectiveness and satisfaction of the interaction. This approach is applicable not only to text generation but also to other scenarios such as code generation and data analysis. Research indicates that having clear goals and standards during interactions with LLMs can markedly enhance their performance.