Empowering Hybrid Attention Models on NPUs
Hybrid attention models have emerged as a crucial architecture for Large Language Models (LLMs) (e.g., the Qwen3.5 and Kimi series). Their memory and computational efficiency make them highly attractive for on-device inference, forming a promising synergy with edge Neural Processing Units (NPUs). However, naive executi...
Yin-Yuan Zhang, Da-Liang Xu, Xiao-Long Huang et al.
· 0 citations