GenPRM: Scaling Test-Time Compute of Process Reward Models via Generative Reasoning Paper • 2504.00891 • Published 4 days ago • 7