Training-Free Loosely Speculative Decoding: Accepting Semantically Correct Drafts Beyond Exact Match
Jinze Li · Yixing Xu · Guanchen Li · Shuo Yang · Jinfeng Xu · Xuanwu Yin · Dong Li · Edith Ngai · Emad Barsoum
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。
Jinze Li · Yixing Xu · Guanchen Li · Shuo Yang · Jinfeng Xu · Xuanwu Yin · Dong Li · Edith Ngai · Emad Barsoum
Xiusheng Huang · Zhe Li · Xuanwu Yin · Lu Wang · Yequan Wang · Dong Li · Emad Barsoum · Kang Liu
Post-training quantization has emerged as a widely adopted technique for compressing and accelerating the inference of Large Language Models (LLMs). The primary challenges in LLMs quantization stem from activation outliers, which significantly degrade model performance especially at lower bit pre...
Zekai Li · Ji Liu · Guanchen Li · Yixing Xu · Ziqiong Liu · Xuanwu Yin · Dong Li · Emad Barsoum
Huanxuan Liao · Yixing Xu · Shizhu He · Guanchen Li · Xuanwu Yin · Dong Li · Emad Barsoum · Jun Zhao · Kang Liu