In this article we explore new DeepMind model which is able to achieve SOTA results on different types of input, without changes in the architecture.

Read more of this post