Cypher RLH
RLHF project focused on evaluating and improving large language model outputs. Tasks included response ranking, text labeling, rewriting for clarity and tone, and identifying factual issues, hallucinations, and guideline violations to generate high-quality training data.