RRaghavan T M·Jun 30RLHF: How ChatGPT Learned What You Actually Want Discover how the three-step RLHF process and reward models taught ChatGPT to provide surprisingly useful answers5 min read·changeofbasis.hashnode.dev