MMRefine: Unveiling the Obstacles to Robust Refinement in Multimodal Large Language Models

This paper introduces MMRefine, a Multi-Modal 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 . * Most work was done during the internship at NAVER Cloud AI. † Corresponding author Refinement Failure (RF) Error Detection Success (ED) Error Correction Success (EC) Refinement Success (RS) False Error Detection (FD) Verification Success (VS) RefScore mRecall