LLM推理论文精读2 -- PowerNPU[ASPLOS'25]
PowerNPU:Fast On-device LLM Inference with NPUs
I am Qingwei Ji (纪 清玮 in Chinese), pursing the CS Ph.D. degree at Southeast University since 2024 fall. I am currently advised by Prof. Fang Dong.
Previously, I received my M.Sc. degree from University of Electronic Science and Technology of China in 2024 and my B.Eng. degree from Shandong University Of Science And Technology in 2021.
My research mainly focuses on LLM inference on edge AI system & smartphones, distributed training system and cluster resource scheduling.
Email: qingweiji@seu.edu.cn
PowerNPU:Fast On-device LLM Inference with NPUs
Transformer模型入门
ASPLOS'23弹性调度论文ElasticFlow精读