#preference learning

Research

Reinforcement Learning From Human Feedback: The Secret Sauce Behind ChatGPT

ChatGPT didn’t become useful because of a bigger model. It became useful because OpenAI figured out how to teach it…