Experimental Design for Active Transductive Inference in Large Language Models
This work addresses the challenge of optimizing in-context learning efficiency for LLMs, representing an incremental improvement in prompt design techniques.
The paper tackles the problem of adaptively selecting few-shot examples for LLM prompts using active learning, resulting in improved performance over other methods across various tasks and model sizes.
One emergent ability of large language models (LLMs) is that query-specific examples can be included in the prompt at inference time. In this work, we use active learning for adaptive prompt design and call it Active In-context Prompt Design (AIPD). We design the LLM prompt by adaptively choosing few-shot examples from a training set to optimize performance on a test set. The training examples are initially unlabeled and we obtain the label of the most informative ones, which maximally reduces uncertainty in the LLM prediction. We propose two algorithms, GO and SAL, which differ in how the few-shot examples are chosen. We analyze these algorithms in linear models: first GO and then use its equivalence with SAL. We experiment with many different tasks in small, medium-sized, and large language models; and show that GO and SAL outperform other methods for choosing few-shot examples in the LLM prompt at inference time.