Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback.
Yuntao BaiAndy JonesKamal NdousseAmanda AskellAnna ChenNova DasSarmaDawn DrainStanislav FortDeep GanguliTom HenighanNicholas JosephSaurav KadavathJackson KernionTom ConerlySheer El ShowkNelson ElhageZac Hatfield-DoddsDanny HernandezTristan HumeScott JohnstonShauna KravecLiane LovittNeel NandaCatherine OlssonDario AmodeiTom B. BrownJack ClarkSam McCandlishChris OlahBenjamin MannJared KaplanPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- motor skills
- test bed
- machine learning
- training process
- test set
- training samples
- training set
- function approximation
- training phase
- human operators
- training sessions
- state space
- relevance feedback
- dynamic programming
- human subjects
- support vector machine
- supervised learning
- multi agent systems
- human behavior
- multi agent
- human interaction
- data sets