FastVLM: Efficient Vision Encoding for Vision Language Modelsmachinelearning.apple.com93 points2bita year ago