Understanding Generalization
Preetum Nakkiran discusses the confusion around understanding generalization and the importance of expressing claims within a causal framework. They emphasize the need to define the causal graph and variables when making claims about causal explanations for generalization. Additionally, they highlight the value of studying objects of interest in deep learning systems, even if they are not design choices.In this clip
From this podcast
The Gradient
Preetum Nakkiran: An Empirical Theory of Deep Learning
Related Questions