JAMMEval: A Refined Collection of Japanese Benchmarks for Reliable VLM Evaluation

TL;DR

JAMMEval is a systematically refined collection of Japanese vision-language benchmarks designed to improve evaluation reliability and better reflect model capabilities.

Contribution

It introduces a human-annotated refinement process for Japanese VQA benchmarks, enhancing data quality and evaluation consistency.

Findings

Refined benchmarks yield more accurate model evaluation scores.

Evaluation scores show lower variance and better distinguish model capabilities.

The dataset and code are publicly released for community use.

Abstract

Reliable evaluation is essential for the development of vision-language models (VLMs). However, Japanese VQA benchmarks have undergone far less iterative refinement than their English counterparts. As a result, many existing benchmarks contain issues such as ambiguous questions, incorrect answers, and instances that can be solved without visual grounding, undermining evaluation reliability and leading to misleading conclusions in model comparisons. To address these limitations, we introduce JAMMEval, a refined collection of Japanese benchmarks for reliable VLM evaluation. It is constructed by systematically refining seven existing Japanese benchmark datasets through two rounds of human annotation, improving both data quality and evaluation reliability. In our experiments, we evaluate open-weight and proprietary VLMs on JAMMEval and analyze the capabilities of recent models on Japanese VQA. We further demonstrate the effectiveness of our refinement by showing that the resulting benchmarks yield evaluation scores that better reflect model capability, exhibit lower run-to-run variance, and improve the ability to distinguish between models of different capability levels. We release our dataset and code to advance reliable evaluation of VLMs.