Advancing Multimodal Large Language Models: Optimizing Prompt Engineering Strategies for Enhanced Performance

This study investigates prompt engineering (PE) strategies to mitigate hallucination, a key limitation of multimodal large language models (MLLMs). To address this issue, we explore five prominent multimodal PE techniques: in-context learning (ICL), chain of thought (CoT), step-by-step reasoning (SS...

Celý popis

Podrobná bibliografie

Hlavní autoři:	Minjun Son, Sungjin Lee
Médium:	Artigo
Jazyk:	Inglês
Vydáno:	MDPI AG 2025-04-01
Edice:	Applied Sciences
Témata:	multimodal large language model prompt engineering in-context learning chain of thought retrieval-augmented generation step-by-step reasoning
On-line přístup:	https://www.mdpi.com/2076-3417/15/7/3992
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!