Sam McCandlish

Scaling Laws for Neural Language Models
Constitutional AI: Harmlessness from AI Feedback
Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback