Episode 28: Sergey Levine, UC Berkeley, on the bottlenecks to generalization in reinforcement learning, why simulation is doomed to succeed, and how to pick good research problems
Generally Intelligent
00:00
Is There a Principle to It?
In offline RL, do you feel like there's a lot left to be done in offline RL? Or are we sort of at the point where we have decent techniques. We're learning a lot from these datasets that we have, and we need something else to move forward," he says.
Transcript
Play full episode
