cypher_rlhf
Project Cypher, a language model evaluation initiative that involved crafting prompts designed to challenge the model's capabilities and expose failure points. Tasks included generating complex prompts, providing reference texts for context, and evaluating model responses based on dimensions such as instruction following, localization, truthfulness, verbosity, writing quality, and safety. Additionally, provided detailed justifications for evaluation scores to ensure alignment with project guidelines and quality standards