AdaPrompt: Adaptive Model Training for Prompt-based NLP

TL;DR

AdaPrompt introduces an adaptive approach to enhance prompt-based NLP by retrieving external data and leveraging NLI models, significantly improving performance in few-shot and zero-shot tasks.

Contribution

The paper proposes AdaPrompt, a novel method that adaptively retrieves external data and uses NLI-based verbalizers to improve prompt-based NLP training.

Findings

Outperforms standard PLMs in few-shot settings.

Achieves up to 26.35% relative error reduction in zero-shot tasks.

Enhances prompt-based learning by addressing data and prompt representation gaps.

Abstract

Prompt-based learning, with its capability to tackle zero-shot and few-shot NLP tasks, has gained much attention in community. The main idea is to bridge the gap between NLP downstream tasks and language modeling (LM), by mapping these tasks into natural language prompts, which are then filled by pre-trained language models (PLMs). However, for prompt learning, there are still two salient gaps between NLP tasks and pretraining. First, prompt information is not necessarily sufficiently present during LM pretraining. Second, task-specific data are not necessarily well represented during pretraining. We address these two issues by proposing AdaPrompt, adaptively retrieving external data for continual pretraining of PLMs by making use of both task and prompt characteristics. In addition, we make use of knowledge in Natural Language Inference models for deriving adaptive verbalizers. Experimental results on five NLP benchmarks show that AdaPrompt can improve over standard PLMs in few-shot settings. In addition, in zero-shot settings, our method outperforms standard prompt-based methods by up to 26.35\% relative error reduction.