Large language model - Wikipedia
6
公开标注数
6
参与人数
2026-06-15 21:09:18
首次 Whisper
本页的公开 Whisper
划选高亮2026-07-15 13:25:22
原文高亮摘录
“reinforcement learning from human feedback (RLHF)”
Whisper 随想笔记
Kinda wild that we're literally training software with our own opinions now.
划选高亮2026-07-15 13:16:22
原文高亮摘录
“reinforcement learning from human feedback (RLHF)”
Whisper 随想笔记
RLHF is basically the secret sauce that makes them sound less like robots.
划选高亮2026-07-15 10:22:22
原文高亮摘录
“Attention Is All You Need”
Whisper 随想笔记
Bahdanau's attention was the real hero here, transformer just polished it.
划选高亮2026-07-15 10:13:22
原文高亮摘录
“Attention Is All You Need”
Whisper 随想笔记
I still miss the old BERT era, decoder-only models feel like cheating sometimes.
划选高亮2026-07-15 10:04:22
原文高亮摘录
“Attention Is All You Need”
Whisper 随想笔记
Transformer paper was such a game changer, everything since is just tweaking.
划选高亮2026-06-15 21:09:18
原文高亮摘录
“initially deemed it too powerful to release publicly”
Whisper 随想笔记
OpenAI really played us with the whole 'too dangerous to release' stunt, classic marketing.
分享本页 Whisper
短链接
https://domwhisper.com/s/dcc1964c3e14嵌入代码
<iframe src="https://domwhisper.com/embed/dcc1964c3e14" width="100%" height="480" style="border:0;border-radius:16px" loading="lazy"></iframe>