迈向理解上下文学习：对比演示与显著性图
Towards Understanding In-Context Learning with Contrastive Demonstrations and Saliency Maps
AI总结：
本研究通过可解释NLP方法和对比演示的显著性图，探究真实标签、输入分布和补充解释对大型语言模型上下文学习性能的影响，发现翻转标签影响最大且补充解释的效果因任务而异。
摘要：
我们研究了各种演示组件在大型语言模型（LLMs）的上下文学习（ICL）性能中的作用。具体而言，我们探讨了真实标签、输入分布和补充解释的影响，特别是当这些元素被改变或扰动时。我们建立在先前的工作之上，这些工作对于这些元素如何影响ICL提供了混合的发现。为了探究这些问题，我们采用可解释NLP（XNLP）方法，并利用对比演示的显著性图进行定性和定量分析。我们的发现表明，翻转真实标签会显著影响显著性，尽管这在较大的LLMs中更为明显。我们对细粒度级别的输入分布的分析表明，在情感分析任务中将指示情感的术语更改为中性术语，其影响不如更改真实标签那样显著。最后，我们发现补充解释在提升ICL性能方面的有效性取决于任务，与符号推理任务相比，在情感分析任务中看到的益处有限。这些见解对于理解LLMs的功能和指导有效演示的开发至关重要，随着LLMs在ChatGPT等应用中的日益广泛使用，这一点变得越来越相关。我们的研究代码公开在https://github.com/paihengxu/XICL。
英文摘要：
We investigate the role of various demonstration components in the in-context learning (ICL) performance of large language models (LLMs). Specifically, we explore the impacts of ground-truth labels, input distribution, and complementary explanations, particularly when these are altered or perturbed. We build on previous work, which offers mixed findings on how these elements influence ICL. To probe these questions, we employ explainable NLP (XNLP) methods and utilize saliency maps of contrastive demonstrations for both qualitative and quantitative analysis. Our findings reveal that flipping ground-truth labels significantly affects the saliency, though it's more noticeable in larger LLMs. Our analysis of the input distribution at a granular level reveals that changing sentiment-indicative terms in a sentiment analysis task to neutral ones does not have as substantial an impact as altering ground-truth labels. Finally, we find that the effectiveness of complementary explanations in boosting ICL performance is task-dependent, with limited benefits seen in sentiment analysis tasks compared to symbolic reasoning tasks. These insights are critical for understanding the functionality of LLMs and guiding the development of effective demonstrations, which is increasingly relevant in light of the growing use of LLMs in applications such as ChatGPT. Our research code is publicly available at https://github.com/paihengxu/XICL.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
