Training AI Without Writing A Reward Function, with Reward Modelling
Robert Miles AI Safety · 3 HN recommendations
Video Q&A
Ask AI to explain this scene or summarize the whole video. When an answer includes timestamps, select one to jump to that moment.
Try asking: What is the speaker explaining here?
Google sign-in is required. Review the maximum credit cost before confirming.
Robert Miles explains reward modelling, in which an AI learns what humans want from their preference comparisons instead of a hand-written reward function. He walks through the 2017 paper on deep reinforcement learning from human preferences.
Why HN recommends it: Seen as the clearest introduction to the idea behind RLHF, and also borrowed by commenters as a metaphor for human motivation.
“I find this video by Rob Miles still the best presentation of the ingenious original 2017(!) paper” Read on HN ↗