Punctation Restoration Benchmark

指标

Overall F1-score:模型对3种标点（，。？）输出结果的平均F1-score值
Infer Time(s):模型处理10K条句子的纯计算时间的平均

采用wudao语料库里面的100K句子训练， 10K句子用于验证， 10K为测试集

Model_name	Ernie-1.0(baseline)	Ernie-3.0-mini-zh	Ernie-3.0-micro-zh	Ernie-3.0-nano-zh	Ernie-tiny
Overall F1-score	0.7519	0.6663	0.6203	0.6034	0.6651
Infer Time(s)	3022.643	638.165	422.562	338.509	1218.835