The Prompt Report: A Systematic Survey of Prompting Techniques

  • 2024-12-23 18:38:36
  • Sander Schulhoff, Michael Ilie, Nishant Balepur, Konstantine Kahadze, Amanda Liu, Chenglei Si, Yinheng Li, Aayush Gupta, HyoJung Han, Sevien Schulhoff, Pranav Sandeep Dulepet, Saurav Vidyadhara, Dayeon Ki, Sweta Agrawal, Chau Pham, Gerson Kroiz, Feileen Li, Hudson Tao, Ashay Srivastava, Hevander Da Costa, Saloni Gupta, Megan L. Rogers, Inna Goncearenco, Giuseppe Sarli, Igor Galynker, Denis Peskoff, Marine Carpuat, Jules White, Shyamal Anadkat, Alexander Hoyle, Philip Resnik
  • 0

Abstract

Generative Artificial Intelligence (GenAI) systems are increasingly beingdeployed across diverse industries and research domains. Developers andend-users interact with these systems through the use of prompting and promptengineering. Although prompt engineering is a widely adopted and extensivelyresearched area, it suffers from conflicting terminology and a fragmentedontological understanding of what constitutes an effective prompt due to itsrelatively recent emergence. We establish a structured understanding of promptengineering by assembling a taxonomy of prompting techniques and analyzingtheir applications. We present a detailed vocabulary of 33 vocabulary terms, ataxonomy of 58 LLM prompting techniques, and 40 techniques for othermodalities. Additionally, we provide best practices and guidelines for promptengineering, including advice for prompting state-of-the-art (SOTA) LLMs suchas ChatGPT. We further present a meta-analysis of the entire literature onnatural language prefix-prompting. As a culmination of these efforts, thispaper presents the most comprehensive survey on prompt engineering to date.

 

Quick Read (beta)

loading the full paper ...