Yang Liu

Microsoft researchers introduced G-Eval, a prompting framework that uses large language models to score open-ended text generation without requiring reference examples. The method combines precise rub…