
Episode 25: Nicklas Hansen, UCSD, on long-horizon planning and why algorithms don't drive research progress
Generally Intelligent
How Much More All Time Does It Take Kind of First Step Than a Model Free Algorithm?
It seems like this model worked quite well on a dog, which I was very surprised by. Like nothing works on dog. Is that something you expected? Or kind of did it take a lot of tweaking to get to a place where it worked in such a, I think dogs like 38-dimensional or something, continuous action space? Did it take aAinslie: We used more or less the same how our hyper-remers that were used for humanoid just let it run a lot longer. It's ultimately a trade-off. TDMPC actually also learned to policy in an extra credit fashion. And we use the policy in our planning. But you could just, if you
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.