Skip to content
ai

RLHF

Reinforcement Learning from Human Feedback

Definition

RLHF is a training methodology that uses human preference judgments to align language model behavior with human values. Human annotators compare model outputs and rank them, training a reward model on these preferences.

The language model is then fine-tuned with PPO to maximize the reward model's score. RLHF was key to creating ChatGPT-style helpful, harmless, and honest assistants.

Can this ship? Are we exposed? Can you prove it?

Every CISO is asking these three questions. We answer them with facts, not guesses, from repository to runtime.