Softmax function - Wikipedia
5
公开标注数
5
参与人数
2026-07-19 09:39:34
首次 Whisper
本页的公开 Whisper
划选高亮2026-07-19 13:00:34
原文高亮摘录
“The Softmax function is a smooth approximation to the arg max function”
Whisper 随想笔记
Wait, so the name is technically wrong but everyone still uses it? Classic ML.
划选高亮2026-07-19 12:51:34
原文高亮摘录
“The Softmax function is a smooth approximation to the arg max function”
Whisper 随想笔记
Huh, so 'softmax' is actually a misnomer? I always assumed it was a smooth max.
划选高亮2026-07-19 09:57:34
原文高亮摘录
“The softmax function takes as input a tuple z of K real numbers, and normalizes it into a probability distribution”
Whisper 随想笔记
I always forget that the inputs don't have to be probabilities before softmax.
划选高亮2026-07-19 09:48:34
原文高亮摘录
“The softmax function takes as input a tuple z of K real numbers, and normalizes it into a probability distribution”
Whisper 随想笔记
So it's like a fancy way to make numbers into percentages that add to 100.
划选高亮2026-07-19 09:39:34
原文高亮摘录
“The softmax function takes as input a tuple z of K real numbers, and normalizes it into a probability distribution”
Whisper 随想笔记
This is basically how my brain decides what to eat for dinner, just with exponentials.
分享本页 Whisper
短链接
https://domwhisper.com/s/c5218183f8db嵌入代码
<iframe src="https://domwhisper.com/embed/c5218183f8db" width="100%" height="480" style="border:0;border-radius:16px" loading="lazy"></iframe>