Episode 27: Noam Brown, FAIR, on achieving human-level performance in poker and Diplomacy, and the power of spending compute at inference time
Generally Intelligent
00:00
The Problem of Not Controlling the Dialogue Model
The data set is very large. We have about like 50,000 human games, 13 million messages. It's really just there to give you a grounding in how humans play the game. You're not going to get a super human or an expert level bot by just doing supervised learning on this data set.
Transcript
Play full episode
