1 paper
Chenggang Zhao, Chengqi Deng, Chong Ruan +12
The rapid scaling of large language models (LLMs) has unveiled critical limitations in current hardware architectures, including constraints in memory capacity, computational effic…