XNews

Datawhale and SGLang launch open-source course: build a mini inference engine

Datawhale and SGLang launch open-source course: build a mini inference engine

Datawhale and RadixArk released an open-source course teaching LLM inference from scratch by building a mini-sglang. The four-part curriculum covers concepts, implementation, advanced optimizations, and contribution workflow. SGLang core team members authored advanced sections including attention backends, quantization, and PD separation.

Source posts