Fusing Pre-Trained Language Models with Multimodal Prompts through Reinforcement Learning | IEEE Conference Publication | IEEE Xplore