Probing the Decision Boundaries of In-context Learning in Large Language Models

TL;DR

This paper investigates how large language models form decision boundaries during in-context learning, revealing irregularities and proposing methods to improve their robustness and generalization in binary classification tasks.

Contribution

It introduces a novel decision boundary perspective to analyze in-context learning and evaluates techniques to smooth boundaries and enhance model robustness.

Findings

Decision boundaries in LLMs are often irregular and non-smooth.

Model architecture and prompting methods influence boundary quality.

Active prompting can help smooth decision boundaries.

Abstract

In-context learning is a key paradigm in large language models (LLMs) that enables them to generalize to new tasks and domains by simply prompting these models with a few exemplars without explicit parameter updates. Many attempts have been made to understand in-context learning in LLMs as a function of model scale, pretraining data, and other factors. In this work, we propose a new mechanism to probe and understand in-context learning from the lens of decision boundaries for in-context binary classification. Decision boundaries are straightforward to visualize and provide important information about the qualitative behavior of the inductive biases of standard classifiers. To our surprise, we find that the decision boundaries learned by current LLMs in simple binary classification tasks are often irregular and non-smooth, regardless of linear separability in the underlying task. This paper investigates the factors influencing these decision boundaries and explores methods to enhance their generalizability. We assess various approaches, including training-free and fine-tuning methods for LLMs, the impact of model architecture, and the effectiveness of active prompting techniques for smoothing decision boundaries in a data-efficient manner. Our findings provide a deeper understanding of in-context learning dynamics and offer practical improvements for enhancing robustness and generalizability of in-context learning.