Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing

TL;DR

This paper provides a comprehensive survey of prompt-based learning in NLP, highlighting its methodology, advantages, and organizing existing research into a unified framework for better understanding and accessibility.

Contribution

It introduces a unified mathematical notation, organizes existing prompt-based methods along key dimensions, and offers resources to facilitate further research in the field.

Findings

Prompt-based learning enables few-shot and zero-shot NLP tasks.

A structured typology of prompt methods is proposed.

Resources and tools are provided for the research community.

Abstract

This paper surveys and organizes research works in a new paradigm in natural language processing, which we dub "prompt-based learning". Unlike traditional supervised learning, which trains a model to take in an input x and predict an output y as P(y|x), prompt-based learning is based on language models that model the probability of text directly. To use these models to perform prediction tasks, the original input x is modified using a template into a textual string prompt x' that has some unfilled slots, and then the language model is used to probabilistically fill the unfilled information to obtain a final string x, from which the final output y can be derived. This framework is powerful and attractive for a number of reasons: it allows the language model to be pre-trained on massive amounts of raw text, and by defining a new prompting function the model is able to perform few-shot or even zero-shot learning, adapting to new scenarios with few or no labeled data. In this paper we introduce the basics of this promising paradigm, describe a unified set of mathematical notations that can cover a wide variety of existing work, and organize existing work along several dimensions, e.g.the choice of pre-trained models, prompts, and tuning strategies. To make the field more accessible to interested beginners, we not only make a systematic review of existing works and a highly structured typology of prompt-based concepts, but also release other resources, e.g., a website http://pretrain.nlpedia.ai/ including constantly-updated survey, and paperlist.