Łukasz Kaiser - Research Scientist at Google Brain - talks about attentional neural network models and the quick developments that have been made in this recent field. In his talk, he explains how such models "look at the past and generate the next word of the output" and how to train them.
