将你的LLM输出转化为高效的混合System 1模型。开源ML与LLM推理技术，带有精确延迟控制功能。
Distill your LLM traces into a fast hybrid System 1 models. Open-source ML + LLM inference with calibrated deferral.
