“How is SGLang built?” 👇 Together with Datawhale, we’re launching zero-to-sglang to help you go from reading inference code to building a mini-SGLang yourself, then finding your way around the real…

Glad to share zero-to-sglang course. We cooperated with Datawhale to make this official AI infra intro class for the community, in which we will introduce optimization techniques used in SGLang, and…

从零手搓SGLang，读懂推理，动手写代码。#sglang #datawhale See link below 👇

原来读懂推理引擎，不必一上来就硬啃源码。 记一条认知更新：SGLang 的性能不是某个神奇算子换来的，而是 KV Cache、Continuous Batching、Paged KV Cache、RadixAttention 等机制共同作用。教程先讲 prefill/decode 和软硬件瓶颈，再手搓 mini-sglang，最后过渡到真实源码、profiling 与…