Surprise Calibration for Better In-Context Learning

TL;DR

This paper introduces Surprise Calibration (SC), a novel method that uses the concept of surprise to adaptively calibrate biases in large language models during in-context learning, improving task performance.

Contribution

The paper proposes a new bias calibration technique based on surprise, leveraging implicit Bayesian inference to adapt to dynamic class prior shifts in ICL.

Findings

SC outperforms existing bias calibration methods on NLP benchmarks.

Surprise effectively captures class prior shifts in in-context learning.

SC is computationally efficient and adaptable to different tasks.

Abstract

In-context learning (ICL) has emerged as a powerful paradigm for task adaptation in large language models (LLMs), where models infer underlying task structures from a few demonstrations. However, ICL remains susceptible to biases that arise from prior knowledge and contextual demonstrations, which can degrade the performance of LLMs. Existing bias calibration methods typically apply fixed class priors across all inputs, limiting their efficacy in dynamic ICL settings where the context for each query differs. To address these limitations, we adopt implicit sequential Bayesian inference as a framework for interpreting ICL, identify "surprise" as an informative signal for class prior shift, and introduce a novel method--Surprise Calibration (SC). SC leverages the notion of surprise to capture the temporal dynamics of class priors, providing a more adaptive and computationally efficient solution for in-context learning. We empirically demonstrate the superiority of SC over existing bias calibration techniques across a range of benchmark natural language processing tasks.