Conference proceeding
Instances Need More Care: Rewriting Prompts for Instances with LLMs in the Loop Yields Better Zero-Shot Performance
FINDINGS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS: ACL 2024, pp.6211-6232
01/01/2024
DOI: 10.18653/v1/2024.findings-acl.371
Abstract
Large language models (LLMs) have revolutionized zero-shot task performance, mitigating the need for task-specific annotations while enhancing task generalizability. Despite its advancements, current methods using trigger phrases such as "Let's think step by step" remain limited. This study introduces PROMPTED, an approach that optimizes the zero-shot prompts for individual task instances following an innovative manner of "LLMs in the loop". Our comprehensive evaluation across 13 datasets and 10 task types based on GPT-4 reveals that PROMPTED significantly outperforms both the naive zero-shot approaches and a strong baseline (i.e., "Output Refinement") which refines the task output instead of the input prompt. Our experimental results also confirmed the generalization of this advantage to the relatively weaker GPT-3.5. Even more intriguingly, we found that leveraging GPT-3.5 to rewrite prompts for the stronger GPT-4 not only matches but occasionally exceeds the efficacy of using GPT-4 as the prompt rewriter. Our research thus presents a huge value in not only enhancing zero-shot LLM performance but also potentially enabling supervising LLMs with their weaker counterparts, a capability attracting much interest recently. Finally, our additional experiments confirm the generalization of the advantages to open-source LLMs such as Mistral 7B and Mixtral 8x7B.(1)
Details
- Title: Subtitle
- Instances Need More Care: Rewriting Prompts for Instances with LLMs in the Loop Yields Better Zero-Shot Performance
- Creators
- Saurabh Srivastava - George Mason UniversityChengyue Huang - Univ Iowa, Iowa City, IA 52242 USAWeiguo Fan - Univ Iowa, Iowa City, IA 52242 USAZiyu Yao - George Mason University
- Contributors
- Andre Martins (Editor)Vivek Srikumar (Editor)Lun-Wei Ku (Editor)
- Resource Type
- Conference proceeding
- Publication Details
- FINDINGS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS: ACL 2024, pp.6211-6232
- DOI
- 10.18653/v1/2024.findings-acl.371
- Publisher
- Assoc Computational Linguistics-Acl
- Number of pages
- 22
- Grant note
- GMU College of Computing and Engineering 1625039; 2018631 / National Science Foundation; National Science Foundation (NSF) SHF 2311468 / NSF; National Science Foundation (NSF) Office of Research Computing at GMU GMU Department of Computer Science
- Language
- English
- Date published
- 01/01/2024
- Academic Unit
- Business Analytics
- Record Identifier
- 9984797930102771
Metrics
59 Record Views