
Episode 24: Jack Parker-Holder, DeepMind, on open-endedness, evolving agents and environments, online adaptation, and offline learning
Generally Intelligent
BBTN Genmels
When you're doing this generational training and doing the October search, when you reset to a new generation, what you notice is that on textures, you get bigger over time. And we asked some examples of this in the paper. But I know another thing that we also see is the value function that learns a bigger network architecture than the policy. Oh Yeah. That does kind of make sense. It's really interesting. So yeah, I'm trying to hopefully encourage them to keep working on it because there's still more things to do.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.