Unleashing the potential of prompt engineering for large language models
Раскрытие потенциала инженерии промптов для больших языковых моделей
2025-05-08
SCID: 54.1/7xb96ccu
Discuss with AI
chain-of-thought promptinglarge language models (LLMs)prompt engineeringself-consistencyvision-language models (VLMs)
Figures from the paper
Abstract (AI)
This review explores the role of prompt engineering in unleashing the capabilities of large language models (LLMs). Prompt engineering is the process of structuring inputs, and it has emerged as a crucial technique for maximizing the utility and accuracy of these models. Both foundational and advanced prompt engineering methodologies-including techniques such as self-consistency, chain of thought, and generated knowledge, which can significantly enhance the performance of models-are explored in this paper. Additionally, the prompt methods for vision language models (VLMs) are examined in detail. Prompt methods are evaluated with subjective and objective metrics, ensuring a robust analysis of their efficacy. Critical to this discussion is the role of prompt engineering in artificial intelligence (AI) security, particularly in terms of defending against adversarial attacks that exploit vulnerabilities in LLMs. Strategies for minimizing these risks and improving the robustness of models are thoroughly reviewed. Finally, we provide a perspective for future research and applications.
Key Findings
1
Advanced methods including self-consistency, chain-of-thought, and generated knowledge can significantly enhance LLM performance.
2
Prompt engineering contributes to AI security by addressing adversarial attacks and supporting strategies to improve model robustness.
3
Prompt engineering structures inputs to improve the utility and accuracy of large language models.
4
Prompt methods are assessed using both subjective and objective metrics to analyze their efficacy robustly.
5
The review examines prompt-engineering methods for vision-language models in addition to text-based LLMs.
Research Object
large language models (LLMs), including vision language models (VLMs)
Research Subject
the effectiveness, performance enhancement, and security robustness of prompt engineering methodologies
Publication Details
Publication Date
2025-05-08
Journal
Publisher
ISSN
Cited by
1098
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai20
ImageNet: A large-scale hierarchical image database2009
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer2019
Proceedings of the 24th international conference on Machine learning2007
Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank2013
A Survey on Evaluation of Large Language Models2024
Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing2022
A survey on large language model based autonomous agents2024
ConceptNet 5.5: An Open Multilingual Graph of General Knowledge2017
PaLM: Scaling Language Modeling with Pathways2022
GPT-4 Technical Report2023
Explainability for artificial intelligence in healthcare: a multidisciplinary perspective2020
Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering2021
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing2021
Sparks of Artificial General Intelligence: Early experiments with GPT-42023
A survey on large language model (LLM) security and privacy: The Good, The Bad, and The Ugly2024
Active Retrieval Augmented Generation2023
Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models2022
Prompt Injection attack against LLM-integrated Applications2023
Sentiment Analysis in the Era of Large Language Models: A Reality Check2024
A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT2023