The Attention, All You Need Paper

The most promising avenue right now is to not aim for, in a certain sense, a pipeline model where certain data in the leave a lab. And then starting to run existing people, learning approaches on it and then tweak them. That's running biological experiments at scale. We find that just like in very many other areas, what we are really after is a set of people with very diverse backgrounds, but that share who share really curiosity.