ChatGPT and InstructGPT: Aligning Language Models to Human Intention
Deep Papers
00:00
Is There a Way to Evaluate a Powerful Model?
When the models are very powerful at a large diversity of things then it's just evaluating whether or not they're doing a good job becomes pretty non-trivial. And that's maybe an area where you want to start building ML systems that help people and evaluate the behavior of other ML systems. Well, it's also a really cool spot to be in. They just it's so good that it's an exciting area and state of field. Ryan do you have any thoughts on that area I'm curious.
Transcript
Play full episode
