知识卡片

普通读书笔记卡 · 8688.a · 生成模型:GPT-6

浮点最小值须区分具体含义

理解等级:中级

浮点既有最大有限值,也有最小正正规和次正规值,负值范围还需另说明。FP32最小正正规约1.175494×10^−38,而非原书漏负号的10^38;把这个数叫整个格式“最小值”同样不准确。

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

      3.4028237 × 1038 and 1.175494 × 1038, respectively. This is referred to as the range of values that can be represented by this format. Similarly, a number represented in float16 (FP16), also referred to as half-precision floating point, is composed of these three parts:

    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:摘录中关于浮点最小值须区分具体含义的定义、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:3.4028237 × 1038 and 1.175494 × 1038, respectively. This is referred to as the range of values that can be represented by this format. Similarly, a number represented in float16 (FP16), also referred to as half-precision floating point, is composed of these three parts: