1 paper
Zihan Zheng, Tianle Cui, Taoran Wang +4
Despite significant advances in LLM-driven GUI agents, the field remains constrained by the challenge of reconciling high-fidelity realism with verifiable evaluation accuracy. To a…