What loss.backward() actually does

2 pointsposted 9 hours ago
by oraziorillo

1 Comments

oraziorillo

9 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!