研究人员开发了GRASP，一个旨在提高用作评估论点的模型（LLM-as-a-Judge）的一致性和透明度的新框架。当前LLM-as-a-Judge方法由于过度简化复杂的辩论结构，常常产生不稳定的全局判决。GRASP通过攻击-防御传播算子聚合稳定的局部交互判断来解决这一问题，从而产生更具可复现性的全局排名，侧重于结构充分性而非主观说服力。 AI
影响 引入了一种更透明、可审计的LLM论点评估方法，有望提高AI评委的可靠性。
排序理由 学术论文，介绍LLM评估新框架。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
Large language models are increasingly deployed as automated judges to evaluate the strength of arguments. As this role expands, their legitimacy depends on consistency, transparency, and the ability to separate argumentative structure from rhetorical appeal. However, we show tha…
Large language models are increasingly deployed as automated judges to evaluate the strength of arguments. As this role expands, their legitimacy depends on consistency, transparency, and the ability to separate argumentative structure from rhetorical appeal. However, we show tha…