Datasets:
Tasks:
Question Answering
Modalities:
Text
Formats:
parquet
Sub-tasks:
closed-domain-qa
Languages:
English
Size:
10K - 100K
ArXiv:
License:
Train split: 16 questions appear more than once with different correct answers (two are misspellings)
#6 opened 1 day ago
by
nootxlm
A benchmark dataset with Non-Commercial license? doesn't make sense
#5 opened 10 months ago
by
ceshiheike