ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yan, Yibo, Wang, Shen, Huo, Jiahao, Li, Hang, Li, Boyan, Su, Jiamin, Gao, Xiong, Zhang, Yi-Fan, Xu, Tianlong, Chu, Zhendong, Zhong, Aoxiao, Wang, Kun, Xiong, Hui, Yu, Philip S., Hu, Xuming, Wen, Qingsong
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911604800487424
author Yan, Yibo
Wang, Shen
Huo, Jiahao
Li, Hang
Li, Boyan
Su, Jiamin
Gao, Xiong
Zhang, Yi-Fan
Xu, Tianlong
Chu, Zhendong
Zhong, Aoxiao
Wang, Kun
Xiong, Hui
Yu, Philip S.
Hu, Xuming
Wen, Qingsong
author_facet Yan, Yibo
Wang, Shen
Huo, Jiahao
Li, Hang
Li, Boyan
Su, Jiamin
Gao, Xiong
Zhang, Yi-Fan
Xu, Tianlong
Chu, Zhendong
Zhong, Aoxiao
Wang, Kun
Xiong, Hui
Yu, Philip S.
Hu, Xuming
Wen, Qingsong
contents As the field of Multimodal Large Language Models (MLLMs) continues to evolve, their potential to revolutionize artificial intelligence is particularly promising, especially in addressing mathematical reasoning tasks. Current mathematical benchmarks predominantly focus on evaluating MLLMs' problem-solving ability, yet there is a crucial gap in addressing more complex scenarios such as error detection, for enhancing reasoning capability in complicated settings. To fill this gap, we formally formulate the new task: multimodal error detection, and introduce ErrorRadar, the first benchmark designed to assess MLLMs' capabilities in such a task. ErrorRadar evaluates two sub-tasks: error step identification and error categorization, providing a comprehensive framework for evaluating MLLMs' complex mathematical reasoning ability. It consists of 2,500 high-quality multimodal K-12 mathematical problems, collected from real-world student interactions in an educational organization, with rigorous annotation and rich metadata such as problem type and error category. Through extensive experiments, we evaluated both open-source and closed-source representative MLLMs, benchmarking their performance against educational expert evaluators. Results indicate significant challenges still remain, as GPT-4o with best performance is still around 10% behind human evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2410_04509
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection
Yan, Yibo
Wang, Shen
Huo, Jiahao
Li, Hang
Li, Boyan
Su, Jiamin
Gao, Xiong
Zhang, Yi-Fan
Xu, Tianlong
Chu, Zhendong
Zhong, Aoxiao
Wang, Kun
Xiong, Hui
Yu, Philip S.
Hu, Xuming
Wen, Qingsong
Computation and Language
As the field of Multimodal Large Language Models (MLLMs) continues to evolve, their potential to revolutionize artificial intelligence is particularly promising, especially in addressing mathematical reasoning tasks. Current mathematical benchmarks predominantly focus on evaluating MLLMs' problem-solving ability, yet there is a crucial gap in addressing more complex scenarios such as error detection, for enhancing reasoning capability in complicated settings. To fill this gap, we formally formulate the new task: multimodal error detection, and introduce ErrorRadar, the first benchmark designed to assess MLLMs' capabilities in such a task. ErrorRadar evaluates two sub-tasks: error step identification and error categorization, providing a comprehensive framework for evaluating MLLMs' complex mathematical reasoning ability. It consists of 2,500 high-quality multimodal K-12 mathematical problems, collected from real-world student interactions in an educational organization, with rigorous annotation and rich metadata such as problem type and error category. Through extensive experiments, we evaluated both open-source and closed-source representative MLLMs, benchmarking their performance against educational expert evaluators. Results indicate significant challenges still remain, as GPT-4o with best performance is still around 10% behind human evaluation.
title ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection
topic Computation and Language
url https://arxiv.org/abs/2410.04509