Saved in:
Bibliographic Details
Main Authors: Ismithdeen, Mohamed Insaf, Khattak, Muhammad Uzair, Khan, Salman
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.03986
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911138301607936
author Ismithdeen, Mohamed Insaf
Khattak, Muhammad Uzair
Khan, Salman
author_facet Ismithdeen, Mohamed Insaf
Khattak, Muhammad Uzair
Khan, Salman
contents Despite the success of Large Multimodal Models (LMMs) in recent years, prompt design for LMMs in Multiple-Choice Question Answering (MCQA) remains poorly understood. We show that even minor variations in prompt phrasing and structure can lead to accuracy deviations of up to 15% for certain prompts and models. This variability poses a challenge for transparent and fair LMM evaluation, as models often report their best-case performance using carefully selected prompts. To address this, we introduce Promptception, a systematic framework for evaluating prompt sensitivity in LMMs. It consists of 61 prompt types, spanning 15 categories and 6 supercategories, each targeting specific aspects of prompt formulation, and is used to evaluate 10 LMMs ranging from lightweight open-source models to GPT-4o and Gemini 1.5 Pro, across 3 MCQA benchmarks: MMStar, MMMU-Pro, MVBench. Our findings reveal that proprietary models exhibit greater sensitivity to prompt phrasing, reflecting tighter alignment with instruction semantics, while open-source models are steadier but struggle with nuanced and complex phrasing. Based on this analysis, we propose Prompting Principles tailored to proprietary and open-source LMMs, enabling more robust and fair model evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2509_03986
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Promptception: How Sensitive Are Large Multimodal Models to Prompts?
Ismithdeen, Mohamed Insaf
Khattak, Muhammad Uzair
Khan, Salman
Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
Despite the success of Large Multimodal Models (LMMs) in recent years, prompt design for LMMs in Multiple-Choice Question Answering (MCQA) remains poorly understood. We show that even minor variations in prompt phrasing and structure can lead to accuracy deviations of up to 15% for certain prompts and models. This variability poses a challenge for transparent and fair LMM evaluation, as models often report their best-case performance using carefully selected prompts. To address this, we introduce Promptception, a systematic framework for evaluating prompt sensitivity in LMMs. It consists of 61 prompt types, spanning 15 categories and 6 supercategories, each targeting specific aspects of prompt formulation, and is used to evaluate 10 LMMs ranging from lightweight open-source models to GPT-4o and Gemini 1.5 Pro, across 3 MCQA benchmarks: MMStar, MMMU-Pro, MVBench. Our findings reveal that proprietary models exhibit greater sensitivity to prompt phrasing, reflecting tighter alignment with instruction semantics, while open-source models are steadier but struggle with nuanced and complex phrasing. Based on this analysis, we propose Prompting Principles tailored to proprietary and open-source LMMs, enabling more robust and fair model evaluation.
title Promptception: How Sensitive Are Large Multimodal Models to Prompts?
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
url https://arxiv.org/abs/2509.03986