[lxe]

I've noticed that the learning curve stary fairly flat when it comes to understanding weights, and layers, and neural networks, heck, even what gradient descent is for... but then when it comes to actually understanding why optimization algorithms are needed, and how they work, things just spiral into very hard math territory.

I do think that maybe it feels inaccessible because we transition from discrete concepts easily digestible by CS grads into some complicated math with very terse mathematic notation, yet the math might not be as hard if presented in a way that doesn't scare away programmers.