Exploring Explanations Improves the Robustness of In-Context Learning
This work addresses robustness issues in ICL for natural language processing, representing an incremental improvement over existing explanation-based methods.
The paper tackles the problem of in-context learning (ICL) struggling to generalize beyond demonstration distributions by introducing X^2-ICL, a framework that systematically explores explanations for all possible labels, resulting in significantly improved robustness to out-of-distribution data on multiple natural language understanding datasets.
In-context learning (ICL) has emerged as a successful paradigm for leveraging large language models (LLMs). However, it often struggles to generalize beyond the distribution of the provided demonstrations. A recent advancement in enhancing robustness is ICL with explanations (X-ICL), which improves prediction reliability by guiding LLMs to understand and articulate the reasoning behind correct labels. Building on this approach, we introduce an advanced framework that extends X-ICL by systematically exploring explanations for all possible labels (X$^2$-ICL), thereby enabling more comprehensive and robust decision-making. Experimental results on multiple natural language understanding datasets validate the effectiveness of X$^2$-ICL, demonstrating significantly improved robustness to out-of-distribution data compared to the existing ICL approaches.