SEAM: Semantically Equivalent Across Modalities Benchmark for Vision-Language Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Tang, Zhenwei, Jiao, Difan, Yang, Blair, Anderson, Ashton
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915462186532864
author Tang, Zhenwei
Jiao, Difan
Yang, Blair
Anderson, Ashton
author_facet Tang, Zhenwei
Jiao, Difan
Yang, Blair
Anderson, Ashton
contents Evaluating whether vision-language models (VLMs) reason consistently across representations is challenging because modality comparisons are typically confounded by task differences and asymmetric information. We introduce SEAM, a benchmark that pairs semantically equivalent inputs across four domains that have existing standardized textual and visual notations. By employing distinct notation systems across modalities, in contrast to OCR-based image-text pairing, SEAM provides a rigorous comparative assessment of the textual-symbolic and visual-spatial reasoning capabilities of VLMs. Across 21 contemporary models, we observe systematic modality imbalance: vision frequently lags language in overall performance, despite the problems containing semantically equivalent information, and cross-modal agreement is relatively low. Our error analysis reveals two main drivers: textual perception failures from tokenization in domain notation and visual perception failures that induce hallucinations. We also show that our results are largely robust to visual transformations. SEAM establishes a controlled, semantically equivalent setting for measuring and improving modality-agnostic reasoning.
format Preprint
id arxiv_https___arxiv_org_abs_2508_18179
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SEAM: Semantically Equivalent Across Modalities Benchmark for Vision-Language Models
Tang, Zhenwei
Jiao, Difan
Yang, Blair
Anderson, Ashton
Artificial Intelligence
Computer Vision and Pattern Recognition
Evaluating whether vision-language models (VLMs) reason consistently across representations is challenging because modality comparisons are typically confounded by task differences and asymmetric information. We introduce SEAM, a benchmark that pairs semantically equivalent inputs across four domains that have existing standardized textual and visual notations. By employing distinct notation systems across modalities, in contrast to OCR-based image-text pairing, SEAM provides a rigorous comparative assessment of the textual-symbolic and visual-spatial reasoning capabilities of VLMs. Across 21 contemporary models, we observe systematic modality imbalance: vision frequently lags language in overall performance, despite the problems containing semantically equivalent information, and cross-modal agreement is relatively low. Our error analysis reveals two main drivers: textual perception failures from tokenization in domain notation and visual perception failures that induce hallucinations. We also show that our results are largely robust to visual transformations. SEAM establishes a controlled, semantically equivalent setting for measuring and improving modality-agnostic reasoning.
title SEAM: Semantically Equivalent Across Modalities Benchmark for Vision-Language Models
topic Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.18179