Attention Is All You Need - Wikipedia
6
公开标注数
6
参与人数
2026-07-20 09:17:01
首次 Whisper
本页的公开 Whisper
划选高亮2026-07-20 15:41:01
原文高亮摘录
“We chose the sinusoidal version because it may allow the model to extrapolate to sequence lengths longer than the ones encountered during training.”
Whisper 随想笔记
Sinusoidal positions really do help with longer sequences, nice choice.
划选高亮2026-07-20 12:38:01
原文高亮摘录
“the paper has been cited more than 250,000 times, placing it among the top ten most-cited papers of the 21st century”
Whisper 随想笔记
Guess the authors knew it was peak, all jumped ship right after lol.
划选高亮2026-07-20 12:29:01
原文高亮摘录
“the paper has been cited more than 250,000 times, placing it among the top ten most-cited papers of the 21st century”
Whisper 随想笔记
250k citations is insane, but honestly the paper deserves it, it changed everything.
划选高亮2026-07-20 09:35:01
原文高亮摘录
“The paper introduced a new deep learning architecture known as the transformer”
Whisper 随想笔记
I remember reading this back then; felt like a neat tweak, not a revolution.
划选高亮2026-07-20 09:26:01
原文高亮摘录
“The paper introduced a new deep learning architecture known as the transformer”
Whisper 随想笔记
Honestly, the 'foreseeing' part is impressive, but they didn't fully predict the scale.
划选高亮2026-07-20 09:17:01
原文高亮摘录
“The paper introduced a new deep learning architecture known as the transformer”
Whisper 随想笔记
So this is where the whole GPT craze started—crazy to think it's just 2017.
分享本页 Whisper
短链接
https://domwhisper.com/s/34c4c8ee90c4嵌入代码
<iframe src="https://domwhisper.com/embed/34c4c8ee90c4" width="100%" height="480" style="border:0;border-radius:16px" loading="lazy"></iframe>