Spotlight on Token Perception for Multimodal Reinforcement Learning Paper • 2510.09285 • Published 10 days ago • 35
FrameThinker: Learning to Think with Long Videos via Multi-Turn Frame Spotlighting Paper • 2509.24304 • Published 22 days ago • 4
Sampling-Efficient Test-Time Scaling: Self-Estimating the Best-of-N Sampling in Early Decoding Paper • 2503.01422 • Published Mar 3