Comment on Every Model Learned by Gradient Descent Is Approximately a Kernel MachineparentComments−cs7025yYou're absolutely right. I substituted SGD for GD without giving it any thought because everyone uses SGD!
Comments
You're absolutely right. I substituted SGD for GD without giving it any thought because everyone uses SGD!