Uncertainty Unveiled: Can Exposure to More In-context Examples Mitigate Uncertainty for Large Language Models?

TL;DR

This paper investigates how increasing in-context examples affects the predictive uncertainty of large language models, revealing that more examples generally reduce uncertainty and improve trustworthiness, especially after addressing input noise.

Contribution

It introduces a systematic analysis of uncertainty in in-context learning, emphasizing the role of epistemic uncertainty and internal confidence evolution in large language models.

Findings

Additional examples reduce total uncertainty and epistemic uncertainty.

Performance improvements are linked to injecting task-specific knowledge.

Addressing input noise is crucial for complex tasks.

Abstract

Recent advances in handling long sequences have facilitated the exploration of long-context in-context learning (ICL). While much of the existing research emphasizes performance improvements driven by additional in-context examples, the influence on the trustworthiness of generated responses remains underexplored. This paper addresses this gap by investigating how increased examples influence predictive uncertainty, an essential aspect in trustworthiness. We begin by systematically quantifying the uncertainty of ICL with varying shot counts, analyzing the impact of example quantity. Through uncertainty decomposition, we introduce a novel perspective on performance enhancement, with a focus on epistemic uncertainty (EU). Our results reveal that additional examples reduce total uncertainty in both simple and complex tasks by injecting task-specific knowledge, thereby diminishing EU and enhancing performance. For complex tasks, these advantages emerge only after addressing the increased noise and uncertainty associated with longer inputs. Finally, we explore the evolution of internal confidence across layers, unveiling the mechanisms driving the reduction in uncertainty.