ChatGPT中文性能测评与风险应对*

张华平,李林翰,李春锦

ChatGPT Performance Evaluation on Chinese Language and Risk Measures

Zhang Huaping,Li Linhan,Li Chunjin

表5 中文机器阅读理解任务的性能测试实验

Table 5 Evaluation Experiments on MRC Tasks

数据集 评价指标 WeLM PANGU-α ChatGPT ERNIE 3.0 Titan CMRC2018 EM/F1 -/31.31 1.46/19.28 0.86/49.45 16.62/44.20 DRCD EM/F1 -/39.33 0.66/10.55 7.01/36.32 21.08/37.83 C3 Acc/% 54.30 54.47 85.14 87.59