Not All Metrics Are Guilty: Improving NLG Evaluation with LLM Paraphrasing

Open in new window