VLind-Bench: Measuring Language Priors in Large Vision-Language Models
This addresses a critical issue for researchers and developers of LVLMs by providing a tool to measure language priors, which is incremental as it builds on existing benchmarks but specifically disentangles this factor.
The paper tackles the problem of language priors in Large Vision-Language Models (LVLMs), where models rely on textual patterns over image information, leading to biases or hallucinations, and proposes VLind-Bench, a new benchmark that measures this issue, revealing that almost all models show significant reliance on language priors.
Large Vision-Language Models (LVLMs) have demonstrated outstanding performance across various multimodal tasks. However, they suffer from a problem known as language prior, where responses are generated based solely on textual patterns while disregarding image information. Addressing the issue of language prior is crucial, as it can lead to undesirable biases or hallucinations when dealing with images that are out of training distribution. Despite its importance, current methods for accurately measuring language priors in LVLMs are poorly studied. Although existing benchmarks based on counterfactual or out-of-distribution images can partially be used to measure language priors, they fail to disentangle language priors from other confounding factors. To this end, we propose a new benchmark called VLind-Bench, which is the first benchmark specifically designed to measure the language priors, or blindness, of LVLMs. It not only includes tests on counterfactual images to assess language priors but also involves a series of tests to evaluate more basic capabilities such as commonsense knowledge, visual perception, and commonsense biases. For each instance in our benchmark, we ensure that all these basic tests are passed before evaluating the language priors, thereby minimizing the influence of other factors on the assessment. The evaluation and analysis of recent LVLMs in our benchmark reveal that almost all models exhibit a significant reliance on language priors, presenting a strong challenge in the field.