Scaling GUI reinforcement learning requires environments that preserve real-app behavior. Yet coupling behavior inference with coding can turn partial traces…
机构:阿里
来源:arXiv 2610.30792 | AI4Papers 论文推荐平台