1 paper across 1 session
This paper investigates what kind of R1-Zero-like training is suitable for grounding tasks in GUI agents.