Zhiyong Wang

Transformer context for in-context RL can be compressed to the Krylov dimension, revealing a $Θ(γ^m)$ decay in approximation error below this threshold.

Bridging the gap between real and synthetic video representations can dramatically enhance zero-shot video captioning accuracy.

Redundant neuron encoding can thwart white-box attacks, ensuring safety without sacrificing model performance.

HiLaR reveals that optimizing user preference modeling through hierarchical latent reasoning can significantly outperform traditional recommendation approaches.