Zero-Shot 3D Visual Grounding from Vision-Language Models | Read Paper on Bytez