Machine Reading Comprehension-Enabled Public Service Information System: A Large-Scale Dataset and Neural Network Models

Table 2

The number of questions of each MRC dataset.

Dataset	Question #	Train Qu. #	Dev Qu. #	Test Qu. #	Percentage of the train set
SQuAD2.0	151,054	130,319	11,873	8,862	86.27
MCScript	13,939	9,731	1,411	2,797	69.81
TQA	26,260	15,154	5,309	5,797	57.71
SQuAD1.1	107,702	87,599	10,570	9,533	81.33
MovieQA	21,406	14,166	2,844	4,396	66.18
OpenBookQA	5,957	4,957	500	500	83.21
DREAM	10,197	6,116	2,040	2,041	59.98
WikiQA	3,047	2,118	296	633	69.51
ARC-E	5,197	2,251	570	2,376	43.31
ProPara	488	391	54	43	80.12
ARC-C	2,590	1,119	299	1,172	43.20
C-Pulse	20,000	16,000	2,000	2,000	80.00