Tweeted By @ilyasut

i like how it suggests that backprop may really be the best algorithm for training neural networkshttps://t.co/pCQJj8wFJ5

— Ilya Sutskever (@ilyasut) March 17, 2020