Boosting the Power of Small Multimodal Reasoning Models to Match Larger Models with Self-Consistency Training

Open in new window