
Episode 25: Nicklas Hansen, UCSD, on long-horizon planning and why algorithms don't drive research progress
Generally Intelligent
00:00
How Do You Think You Can Predict the Future?
I think you also compare like a like dreamer and things that work for pixels right in that paper. I was really curious if you had done any experiments to see if you do to my hyperparameters, how much closer do they get to because dreamer is definitely tuned. Yeah, we basically just look at what other people were using an image based around. And we just try to stay close to those hyperparameters and not really tweak it too much. But I think if you tweak it, you're going to always squeeze out a lot more performance.
Transcript
Play full episode