
Episode 16: Yilun Du, MIT, on energy-based models, implicit functions, and modularity
Generally Intelligent
00:00
Dolty, What's the Most Interesting Thing About This Approach?
This is a different way to do generative muslin, you could say, clearly, it's not nearly as good as other approaches right now. But i think there's some interesting ideas that you can take from these odor techniques and use them in new ways. The only thing it workd was three otherm but you guys would lie. No, we're just gont use three later, use the whole time. That's itlo, a little, it's roi for that to come through. I guess i could also imagine the way it works basically as if there's this hiber network that predicts the weights of this other network. Rit so when you're predicting the weights
Transcript
Play full episode