Reordering Examples Helps during Priming-based Few-Shot Learning

TL;DR

This paper introduces PERO, a method that optimizes the order of examples in prompts for few-shot learning, significantly improving generalization with minimal data across NLP tasks.

Contribution

The paper proposes PERO, a novel approach that searches for optimal example orderings in prompts, enhancing few-shot learning performance without model fine-tuning.

Findings

PERO outperforms existing methods with as few as 10 examples.

Learning a new separator token can improve prompt effectiveness.

Order of examples in prompts critically affects model generalization.

Abstract

The ability to learn from limited data, or few-shot learning, is a desirable and often critical requirement for NLP systems. While many existing methods do poorly at learning from a handful of examples, large pretrained language models have recently been shown to be efficient few-shot learners. One approach to few-shot learning, which does not require finetuning of model parameters, is to augment the language model's input with priming text which is typically constructed using task specific descriptions and examples. In this work, we further explore priming-based few-shot learning, with focus on using examples as prompts. We show that presenting examples in the right order is key for generalization. We introduce PERO (Prompting with Examples in the Right Order), where we formulate few-shot learning as search over the set of permutations of the training examples. We show that PERO can learn to generalize efficiently using as few as 10 examples, in contrast to existing approaches. While the newline token is a natural choice for separating the examples in the prompt, we show that learning a new separator token can potentially provide further gains in performance. We demonstrate the effectiveness of the proposed method on the tasks of sentiment classification, natural language inference and fact retrieval. Finally, we analyze the learned prompts to reveal novel insights, including the idea that two training examples in the right order alone can provide competitive performance for sentiment classification and natural language inference.