Deep Learning from Scratch

Write the forward and backward passes yourself, train a network end to end, then build the layers, normalisation and training tricks that make deep networks work, finishing with your own autograd engine.

16 problems · 425 XP · 2 tracks