What Are We Measuring When We Evaluate Large Vision-Language Models? An Analysis of Latent Factors and Biases

Anthony Tiong, Junqi Zhao, Boyang Li, Junnan Li, Steven Hoi, Caiming Xiong


Abstract
Vision-language (VL) models, pretrained on colossal image-text datasets, have attained broad VL competence that is difficult to evaluate. A common belief is that a small number of VL skills underlie the variety of VL tests. In this paper, we perform a large-scale transfer learning experiment aimed at discovering latent VL skills from data. We reveal interesting characteristics that have important implications for test suite design. First, generation tasks suffer from a length bias, suggesting benchmarks should balance tasks with varying output lengths. Second, we demonstrate that factor analysis successfully identifies reasonable yet surprising VL skill factors, suggesting benchmarks could leverage similar analyses for task selection.Finally, we present a new dataset, OLIVE1, which simulates user instructions in the wild and presents challenges dissimilar to all datasets we tested. Our findings contribute to the design of balanced and broad-coverage vision-language evaluation methods. 1https://github.com/jq-zh/olive-dataset
Anthology ID:
2024.naacl-long.188
Volume:
Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
Month:
June
Year:
2024
Address:
Mexico City, Mexico
Editors:
Kevin Duh, Helena Gomez, Steven Bethard
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3427–3454
Language:
URL:
https://aclanthology.org/2024.naacl-long.188
DOI:
10.18653/v1/2024.naacl-long.188
Bibkey:
Cite (ACL):
Anthony Tiong, Junqi Zhao, Boyang Li, Junnan Li, Steven Hoi, and Caiming Xiong. 2024. What Are We Measuring When We Evaluate Large Vision-Language Models? An Analysis of Latent Factors and Biases. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 3427–3454, Mexico City, Mexico. Association for Computational Linguistics.
Cite (Informal):
What Are We Measuring When We Evaluate Large Vision-Language Models? An Analysis of Latent Factors and Biases (Tiong et al., NAACL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.naacl-long.188.pdf