Episode 20: Hattie Zhou, Mila, on supermasks, iterative learning, and fortuitous forgetting
Generally Intelligent
Tendexure, I Love That!
In a lot of cases, like almost monotonically towards. When it's very nicely converging, also monotonically toward something. In some way, we don't have a good number for that at all. For easy data sets, that actually might be true. I think there was a paper that showed if you take the initial ways and the final ways, just linearly interpolate, the law just goes straight down. No idea why. Not the case for obviously a more complicated data set.
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.
