AI.info

Kewei Zhang

Explore Kewei Zhang on AI.info.

• MHLA: Restoring Expressivity of Linear Attention via Token-Level Multi-Head