ExpertWeaver: Unlocking the Inherent MoE in Dense LLMs with GLU Activation Patterns
Ziyu Zhao · Tong Zhu · Xin Yu · Zhi Zhang · Tiantian Fan · Jinluan Yang · Kun Kuang · Zhongyu Wei · Fei Wu · Yu Cheng
首次获取该候选人的完整档案及 StartLux 评估共 20 积分;每位用户单独解锁,之后查看免费。主动刷新 20 积分。
实时查找电子邮箱,费用不超过20积分;未找到邮箱不扣费。
Ziyu Zhao · Tong Zhu · Xin Yu · Zhi Zhang · Tiantian Fan · Jinluan Yang · Kun Kuang · Zhongyu Wei · Fei Wu · Yu Cheng
Kairong Han · Nuanqiao Shan · Ziyu Zhao · Zijing Hu · Xinpeng Dong · Ye Jun Jian · Lujia Pan · Fei Wu · Kun Kuang
Autoregressive (AR) language models and Diffusion Language Models (DLMs) constitute the two principal paradigms of large language models. However, both paradigms suffer from insufficient reasoning capabilities. Human reasoning inherently relies on causal knowledge and thought, which are reflected...
Ziyu Zhao · Yixiao Zhou · Xin Yu · Zhi Zhang · Didi Zhu · Tao Shen · Zexi Li · Jinluan Yang · Xuwu Wang · Jing Su · Kun Kuang · zhongyu wei · Fei Wu · Yu Cheng