← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 8683 · 生成模型:GPT-6

推理优化先辨别瓶颈阶段

理解等级:中级

自回归逐token生成存在顺序依赖,长输入的稠密注意力还带来较高预填充成本。预填充、逐步解码及跨请求调度的瓶颈不同;不能将全部请求耗时一概说成随输入长度平方增长,也不能仅用参数量解释所有延迟。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      LLM Inference Challenges What are the bottlenecks affecting LLM inference? As we all know, their gargantuan sizes necessitate vast computing and memory resources. Apart from that, two additional factors exacerbate the situation: As seen in Chapter 4, contemporary LLMs are based largely on decoder-only models that operate autoregressively. This means that each token is generated one after the other, thus imposing a sequential limitation. Later

    参考来源

    - 位置:《Designing Large Language Model Applications》第09章《推理优化》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch09.html) - 结论依据:摘录中关于推理优化先辨别瓶颈阶段的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:LLM Inference Challenges What are the bottlenecks affecting LLM inference? As we all know, their gargantuan sizes necessitate vast computing and memory resources. Apart from that, two additional factors exacerbate the situation: As seen in Chapter 4, contemporary LLMs are based largely on decoder-only models that operate autoregressively. This means that each token is generated one after the other, thus imposing a sequential limitation. Later