What is RLHF (Reinforcement Learning from Human Feedback), and how does it align LLMs? What was the original Deep RL from Human Preferences paper?
Interview insights are lockedSee how often this is asked and which companies asked it with Buncho Pro.Describe the three stages of RLHF, the reward-model loss, the KL-regularised RL objective, and the history from the 2017 Deep RL from Human Preferences paper to InstructGPT.