让AI分析类比下，向大模型训练一样去反观人类的学习。 互相印证启发，很有趣。
#### 向阳乔木
@vista8 · 21h ago
书籍、文章、播客都只是大脑的训练语料。 大模型不因为读完了整个互联网而变得聪明，而是因为在海量token中涌现了注意力的远程依赖。
人也一样：一本书翻到第三页就放下，没关系。
但如果那三页，让你大脑里两个从未对话的区域突然通了电。
这就是一次有效的参数更新，也就是我们的学习收获。
6
11
76
16.4K
4 Replies
9 Retweets
43 Likes
10.6K Views 
One Sentence Summary
The author guides readers to use the AI training perspective to reflect on the learning process, drawing cross-domain insights.
Summary
This is a supplementary tweet where the author references their previous tweet about the LLM and learning analogy, guiding readers to engage in cross-domain thinking. The tweet includes an image (content unknown) that further illustrates the value of this analogy. This tweet serves as a reference and extension to the previous one, aiming to encourage readers to compare AI training mechanisms with human learning.
AI Score
80
Influence Score 11
Published At Yesterday
Language
Chinese
Tags
AI Analogy
Learning Reflection
Interdisciplinary Thinking