AI Cambrian Explosion: Conversation With Three AI Engineers — #37
Manifold
00:00
The Importance of Quantization in Machine Learning
When you're doing specific sub domains, there also comes into play like a size question. So it could be that smaller model isn't good across the board, but your ability to make it good at a fixed cost or effort on like just say reading legal contracts or financial documents is where maybe those more smaller models shot. I think yeah, that specialization aspect is going to play a huge part kind of as we come and get into the conversation. Like a GI. There was a model called the gorilla model that was specifically trained on a data set of interacting with APIs. It actually performed better at kind of reasoning about API's than Gbeat 4,. And so you can almost imagine if you were
Transcript
Play full episode
