MMRefine: Unveiling the Obstacles to Robust Refinement in Multimodal Large Language Models
This addresses the problem of enhancing reasoning in MLLMs for researchers, but it is incremental as it focuses on benchmarking rather than a new method.
The paper introduces MMRefine, a benchmark to evaluate error refinement in Multimodal Large Language Models across six scenarios and error types, revealing bottlenecks that hinder performance improvement.
This paper introduces MMRefine, a MultiModal Refinement benchmark designed to evaluate the error refinement capabilities of Multimodal Large Language Models (MLLMs). As the emphasis shifts toward enhancing reasoning during inference, MMRefine provides a framework that evaluates MLLMs' abilities to detect and correct errors across six distinct scenarios beyond just comparing final accuracy before and after refinement. Furthermore, the benchmark analyzes the refinement performance by categorizing errors into six error types. Experiments with various open and closed MLLMs reveal bottlenecks and factors impeding refinement performance, highlighting areas for improvement in effective reasoning enhancement. Our code and dataset are publicly available at https://github.com/naver-ai/MMRefine.