New method reveals multimodal benchmark item weaknesses
Are Multimodal Benchmarks Really Useful? Item-Level Multimodal Benchmark Diagnosis via Structure-Response Co-Calibration
Artificial Intelligence
Summary
Multimodal benchmarks test if computer models can understand information from different sources like images and text together. But sometimes, these tests have questions that can be answered using only one source, making the test less useful. The authors created a new method called SRCoD that looks at both the test content and model responses to check how much each question really depends on multiple sources. This method helps spot weak questions and better understand how well models truly combine different types of information.
What this means in practice
- •For machine learning engineers: Improve evaluation datasets by identifying questions that don't require multimodal understanding.
- •For qa system builders: Refine multimodal question-answering benchmarks to better assess model understanding of combined modalities.
Authors
Shiqi Zhang, Weixin Zeng, Ziheng Zhang, Jiuyang Tang, Xiang Zhao
Abstract
Multimodal benchmarks are central to evaluating whether models can reason over information from multiple modalities. However, existing literature has revealed that some benchmarks may contain shortcut items that can be answered from a single modality or from answer-option regularities alone, thus failing to reflect models' true cross-modal inference ability. Existing methods for assessing benchmarks typically rely either on intrinsic item structure or extrinsic model response behavior, which are not adequate for a comprehensive diagnosis. To fill in this gap, we propose SRCoD, a structure-response co-calibrated diagnosis framework for item-level diagnosis of modality dependence in multimodal benchmarks. SRCoD estimates sample-wise multimodal information structure from benchmark content and target answers, and uses this signal as a soft structural anchor for modality-conditioned item response modeling. The learned model produces a calibrated item-level diagnostic profile that reflects how strongly an item depends on different forms of modality evidence, enabling fine-grained benchmark analysis beyond aggregate accuracy. SRCoD also provides interpretable attribution of the diagnosis results. Extensive experiments under two complementary protocols, i.e., benchmark-refinement oriented validation and human-labeled item validation, demonstrate that SRCoD outperforms state-of-the-art benchmark diagnosis methods.