In-Context Learning Demonstration Selection via Influence Analysis
This addresses a key bottleneck in improving few-shot learning efficiency for LLM users, though it is an incremental advancement in demonstration selection methods.
The paper tackles the problem of selecting effective demonstrations for in-context learning in large language models by proposing InfICL, a method using influence functions to identify influential training samples, which enhances generalization performance compared to state-of-the-art baselines on various real-world datasets.
Large Language Models (LLMs) have showcased their In-Context Learning (ICL) capabilities, enabling few-shot learning without the need for gradient updates. Despite its advantages, the effectiveness of ICL heavily depends on the choice of demonstrations. Selecting the most effective demonstrations for ICL remains a significant research challenge. To tackle this issue, we propose a demonstration selection method named InfICL, which utilizes influence functions to analyze impacts of training samples. By identifying the most influential training samples as demonstrations, InfICL aims to enhance the ICL generalization performance. To keep InfICL cost-effective, we only use the LLM to generate sample input embeddings, avoiding expensive fine-tuning. Through empirical studies on various real-world datasets, we demonstrate advantages of InfICL compared to state-of-the-art baselines.