Researchers from UC Berkeley, UIUC, and NYU Developed an Algorithmic Framework that Uses Reinforcement Learning (RL) to Optimize Vision-Language Models (VLMs) – MarkTechPost