GUI-G1: Understanding R1-Zero-Like Training for Visual Grounding in GUI Agents | Read Paper on Bytez