Human preference data for "Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback" - View it on GitHub