[Retracted] Analyzing the Effect of Masking Length Distribution of MLM: An Evaluation Framework and Case Study on Chinese MRC Datasets

Table 3

Statistics of the proposed span extraction datasets.

	Short span extraction			Long span extraction
	Train set	Dev set	Test set	Train set	Dev set	Test set
Paragraph #	600	150	200	600	150	200
Passage #	15,265	2770	2484	7756	1163	1565
Question #	31,390	6774	8309	12,053	2376	5812
Max tokens in a context #	512	512	512	512	512	512
Max answer tokens #	6	6	6	9	9	9
Min answer tokens #	4	4	4	7	7	7