When2Think: Learning Difficulty-Aware Length Control for Efficient Hybrid Reasoning Models Paper • 2609.19671 • Published 7 days ago • 49
Harnessing CLIP and DINO: An Uncertainty-Aware Cascaded Fusion Network for Generalizable Deepfake Image Detection Paper • 2609.07670 • Published 17 days ago • 18
WorldReward: Reward Modeling for Camera-Conditioned World Models Paper • 2609.03952 • Published 21 days ago • 27
Listening Forward: Next Patch Embedding Prediction Enables Scalable Audio Learners Paper • 2608.19863 • Published Aug 20 • 6
SKILLER: Language-Level Reinforcement Learning for Reusable Skill Extraction in Small Language Models Paper • 2608.10538 • Published Aug 11 • 17
CAPEval: A Decoupled Caption Evaluation across Understanding and Generation Paper • 2608.02589 • Published Aug 3 • 26