The predictability of generalization error with scale had already been investigated before scaling laws became a mainstream concept.
这一观点指出，缩放定律成为主流概念之前，研究者就已经开始研究泛化误差的可预测性。这提醒我们，在AI领域，许多看似新颖的发现往往建立在早期研究基础上。初学者应关注历史文献，避免重复造轮子。
2 Matching Annotations
- Last 7 days
-
lilianweng.github.io lilianweng.github.io
-
- Jan 2023
-
www.complexityexplorer.org www.complexityexplorer.org
-
https://www.youtube.com/watch?v=5pSGniUOyLc
Digital humanities aka Humanities Analytics
5:54 Simon DeDeo mentioned Alastair McKinnon the philosopher in the 60s did a stylopheric study of Kierkegaard pseudonyms - Kierkegaard's Pseudonyms: A New Hierarchy by Alastair McKinnon https://www.jstor.org/stable/20009297
Tools for supplementing research and scholarship
core audience is Ph.D. students
-