新闻
Humming in SGLang: Flexible Low-Bit Kernels for Efficient MoE Serving
LMSYS · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- Humming 0.1.18 adds flexible low-bit quantized kernels for MoE inference in SGLang, supporting MXFP4 weights and FP8 activations on Hopper and Blackwell GPUs.
- 为何重要
- Engineers deploying large mixture-of-experts models need efficient serving with reduced memory and compute requirements while maintaining model quality.
- 注意
- Scale-range adjustment affected 11.27% of draft groups due to padding issues; a fix exists upstream but requires a follow-up Humming dependency upgrade.
- kernel
- serving
这条新闻背后的模式
每个模式都讲清楚技术如何运作、何时值得投入,以及在哪里会失效。
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。