On the Foundations of Deep Learning: SGD, Overparametrization, and Generalization 

Excellent work. Thank you. At 6:12 into the video, I think I hear you say "gram matrix" and in the transcript, it does say "gram matrix" but what is that? What is a "gram matrix" ? I just searched for that online and it seems to be a basic concept. Okay, I am always lacking some basics.