The dataset contains a total of 1,508 questions, out of which 740 are multimodal (i.e., containing images).