Table 1: 
Comparison of C3 and representative Chinese question answering and machine reading comprehension tasks. We list only one English counterpart for each Chinese dataset.
Chinese TaskDocument GenreQuestion TypeAnswer TypeQuestion SizeEnglish Counterpart
Question Answering 
QS (Cheng et al., 2016) N/A free-form multiple-choice 0.6K ARC (Clark et al., 2016) 
MCQA (Guo et al., 2017a) N/A free-form multiple-choice 14.4K ARC (Clark et al., 2016) 
MedQA (Zhang et al., 2018b) N/A free-form multiple-choice 235.2K ARC (Clark et al., 2016) 
GeoSQA (Huang et al., 2019) N/A free-form multiple-choice 4.1K DD (Lally et al., 2017) 
 
Machine Reading Comprehension 
PD (Cui et al., 2016) news cloze extractive 876.7K CNN/Daily (Hermann et al., 2015) 
CFT (Cui et al., 2016) books cloze extractive 3.6K CBT (Hill et al., 2016) 
CMRC 2018 (Cui et al., 2018b) Wiki free-form extractive 19.1K SQuAD (Rajpurkar et al., 2016) 
DuReader (He et al., 2017) web free-form abstractive ≈ 200K MS MARCO (Nguyen et al., 2016) 
ChID (Zheng et al., 2019) mixed-genre cloze multiple-choice 728.7K CLOTH (Xie et al., 2018) 
CM3(this work) mixed-genre free-form multiple-choice 10.0K RACE (Lai et al., 2017) 
CD3(this work) dialogue free-form multiple-choice 9.6K DREAM (Sun et al., 2019a) 
Chinese TaskDocument GenreQuestion TypeAnswer TypeQuestion SizeEnglish Counterpart
Question Answering 
QS (Cheng et al., 2016) N/A free-form multiple-choice 0.6K ARC (Clark et al., 2016) 
MCQA (Guo et al., 2017a) N/A free-form multiple-choice 14.4K ARC (Clark et al., 2016) 
MedQA (Zhang et al., 2018b) N/A free-form multiple-choice 235.2K ARC (Clark et al., 2016) 
GeoSQA (Huang et al., 2019) N/A free-form multiple-choice 4.1K DD (Lally et al., 2017) 
 
Machine Reading Comprehension 
PD (Cui et al., 2016) news cloze extractive 876.7K CNN/Daily (Hermann et al., 2015) 
CFT (Cui et al., 2016) books cloze extractive 3.6K CBT (Hill et al., 2016) 
CMRC 2018 (Cui et al., 2018b) Wiki free-form extractive 19.1K SQuAD (Rajpurkar et al., 2016) 
DuReader (He et al., 2017) web free-form abstractive ≈ 200K MS MARCO (Nguyen et al., 2016) 
ChID (Zheng et al., 2019) mixed-genre cloze multiple-choice 728.7K CLOTH (Xie et al., 2018) 
CM3(this work) mixed-genre free-form multiple-choice 10.0K RACE (Lai et al., 2017) 
CD3(this work) dialogue free-form multiple-choice 9.6K DREAM (Sun et al., 2019a) 
Close Modal

or Create an Account

Close Modal
Close Modal