RLHF

RLHF: How Human Preferences Shape AI

Inside Reinforcement Learning from Human Feedback — reward modeling, PPO optimization, and why DPO is changing the game.

Mixture of Experts

Mixture of Experts: Scaling LLMs Efficiently

How Mixture of Experts architectures achieve massive parameter counts while keeping compute costs manageable — routing, load balancing, and the sparsity trade-off.