Mmr Total

The MMR total task in the Multi-Modal Reading (MMR) Benchmark aims to evaluate the model's overall performance in handling 11 different sub-tasks, including text, font, visual elements, bounding boxes, spatial relationships, and basic positioning. This task comprehensively measures the model's multimodal reading comprehension ability by summing up the scores of each sub-task, providing a crucial reference for research and applications in the field of computer vision.