Alaya-EVOKE: From Linear-Scaling Supervision to Endless World Paper • 2608.13546 • Published 17 days ago • 165
Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMs Paper • 2608.20492 • Published 10 days ago • 105