logo

What loss.backward() actually does

Posted by oraziorillo |2 hours ago |1 comments

oraziorillo 2 hours ago

Following up on my post on microcrad — many of you liked and supported the project, thank you for that — I published a long-form piece on automatic differentiation, where I explain backpropagation in simple terms so that everyone will (hopefully) be able to understand it. Questions/suggestions are welcome!