Alex Tamkin on Self-Supervised Learning and Large Language Models
The Gradient: Perspectives on AI
00:00
The Principles of Self-Supervised Learning
In online web data, you have images, speech, text, sensors, and captioned images. It would be interesting if you could actually come up with a single thing that works for all of them. What are the principles that make self-supervised learning work? And how do we get your network to learn that? Those are the sort of two things that kind of motivate DABs.
Play episode from 36:44
Transcript
