ResearchBenchmarksMachine Translation Meta-evaluation on WMT Zh-En (subset of 600 samples) 2022Follow0.4597Kendall CorrelationEAPrompt0.448780.4516150.454450.457285Mar 24, 2023Evaluation ResultsMethodMethodLinksKendall CorrelationEAPromptEvaluator Model=GPT-4Evaluator Model=GPT-42023.030.4597GEMBAEvaluator Model=GPT-4Evaluator Model=GPT-42023.030.4492