Holistic Evaluation of Language Models
A study used HELM to evaluate a language model’s ability to generate coherent and contextually appropriate text.