UCLA Researchers Released OpenVLThinker-7B: A Reinforcement Learning Driven Model for Enhancing Complex Visual Reasoning and Step-by-Step Problem Solving in Multimodal Systems