[论文解读] ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs

Jingming Zhuo, Songyang Zhang+4 more|arXiv (Cornell University)|Oct 16, 2024

Biomedical Text Mining and Ontologies被引用 8

一句话总结

ProSA 引入 PromptSensiScore (PSS) 来衡量大模型在实例层面的提示敏感性，分析客观与主观评估的鲁棒性，并将敏感性与解码置信度相关联。

ABSTRACT

Large language models (LLMs) have demonstrated impressive capabilities across various tasks, but their performance is highly sensitive to the prompts utilized. This variability poses challenges for accurate assessment and user satisfaction. Current research frequently overlooks instance-level prompt variations and their implications on subjective evaluations. To address these shortcomings, we introduce ProSA, a framework designed to evaluate and comprehend prompt sensitivity in LLMs. ProSA incorporates a novel sensitivity metric, PromptSensiScore, and leverages decoding confidence to elucidate underlying mechanisms. Our extensive study, spanning multiple tasks, uncovers that prompt sensitivity fluctuates across datasets and models, with larger models exhibiting enhanced robustness. We observe that few-shot examples can alleviate this sensitivity issue, and subjective evaluations are also susceptible to prompt sensitivities, particularly in complex, reasoning-oriented tasks. Furthermore, our findings indicate that higher model confidence correlates with increased prompt robustness. We believe this work will serve as a helpful tool in studying prompt sensitivity of LLMs. The project is released at: https://github.com/open-compass/ProSA .

研究动机与目标

• 从实例层面而非数据集层面，激励并量化提示如何影响 LLM 回应。
• 开发用于客观与主观评估的实例级敏感性指标（PSS）。
• 研究提示敏感性如何随模型、数据集和提示风格而变化。
• 探究解码置信度与提示鲁棒性之间的关系。

提出的方法

• 将 PSS 定义为同一实例在所有提示变体中的响应平均成对差异。
• 在客观任务中对多种开源 LLM 与数据集评估 PSS。
• 评估少量示例提示对提示鲁棒性的影响。
• 使用 LC AlpacaEval 2.0 与 Arena Hard Auto 进行带提示重写的主观评估分析。
• 利用解码置信度分析提示敏感性的潜在原因。

Figure 1: A Showcase of the Four Prompt Templates on MATH. These four prompt templates represent four different styles of constructing prompts, serving as an example of the diversity in human prompt expression.

实验结果

研究问题

• RQ1 实例级提示敏感性如何在不同数据集和模型规模之间变化？
• RQ2 少量示例提示是否会降低提示敏感性并提高鲁棒性？
• RQ3 在实践中，LLM 的解码置信度与其提示鲁棒性之间有什么关系？
• RQ4 提示敏感性在主观评估与客观评估中的表现有何差异？
• RQ5 哪些任务类别显示出更高或更低的提示敏感性？

主要发现

• 提示敏感性因数据集和模型而异，较大模型通常表现出更高的鲁棒性。
• 少量示例提示可降低提示敏感性，尤指从 0-shot 到 1-shot 设置；并且大模型在更多示例下获得更高的鲁棒性。
• 主观评估在复杂任务上揭示更高的敏感性，尽管在简单任务上表现出鲁棒性。
• 解码置信度与较低的提示敏感性相关，表明鲁棒性反映了潜在的解码动态。
• 提示类别会影响敏感性；知识密集型任务往往比编码或创造性任务更鲁棒。

Figure 2: A Comparision of Evaluating LLMs’ Prompt Sensitivity. ✓ and ✗ indicate the accuracy of the LLM’s responses. In this example, LLMs appear robust at the dataset level evaluation (calculated from the variance of different templates), but this overlooks the sensitivity of LLMs to different tem