RLHF
Generative AI & LLMsChapter 4: Choosing the LLMTerm 76 of 253
Reinforcement Learning from Human Feedback: humans rank answers; model learns to prefer better ones.
At Sunrise School
Gopal asksWho decided which answers are “good”?
Bit answersRLHF: people ranked pairs of answers, a reward model learned their taste, and the model was tuned toward the preferred style.
In the real worldTeacher grading essays to improve writing.
Watch the short
Short coming soonSeason 2 · Day 17Follow @haigopala
Short coming soonSeason 2 · Day 17Follow @haigopala.telugu
Short coming soonSeason 2 · Day 17Follow @haigopala.hindi
Quick check
Quick check. In RLHF, do humans write every answer? Yes or no?
No! They rank answers; the model learns their taste.
Practise it
Try this term and all 253 others in the free Model Training Arena.
Open the Arena →Follow Bit daily: YouTube English · తెలుగు · हिंदी | Instagram English · తెలుగు · हिंदी