When training AlexNet, several key differences emerge compared to training the simpler LeNet-5 architecture. Because AlexNet is a much deeper and wider network, and because it processes higher-resolution images with more computationally costly convolutions, its training process is significantly slower. Consequently, training AlexNet typically requires the use of a smaller learning rate to ensure stable convergence, whereas LeNet-5 can often be trained with a larger learning rate and converges much faster due to its smaller size and lower-resolution inputs.