Sai Krishna Gottipati
TalkRL: The Reinforcement Learning Podcast
00:00
How to Find the Reactant in a Continuous Space?
For the reactan, there is this databis called inemine data set. It contains about a hundred and 50 thousand molecules. And we have a librari like ardicete that does a, that computes most of these things. So when you get a new molecule, you pass it through this reward model to obtain scalar value. For example, let's say you want to invent a drug to cure hechivy then what you do as they doelop some cusar models so you know what a heChivy ar target is. Whatis the reward a design hereye? We did it againnst three hecvebest targets.
Play episode from 45:42
Transcript
