AI Glossary

Reinforcement learning from human feedback

Also called RLHF

What does “Reinforcement learning from human feedback” mean?

Reinforcement learning from human feedback: Training method in which people compare or rate model outputs, and a reward model learns their preferences from those judgments. The language model is then optimized to produce answers that the reward model scores highly.

Reinforcement learning from human feedback visual guide: Training method in which people compare or rate model outputs, and a reward model learns their preferences from those judgments. The language model is then optimized to produce answers that the reward model scores highly.
Visual guide Open full size (new tab)

What is an example of “Reinforcement learning from human feedback”?

RLHF was a key part of turning raw language models into chat assistants that respond more helpfully and decline clearly harmful requests.

Keep learning

See also

Reward modelTermReinforcement learning from AI feedbackTermDirect preference optimizationTermReinforcement learningTerm

Same topic

AlignmentTermData labelingTermInstruction tuningTermReinforcement learning from verifiable rewardsTermSycophancyTerm