Datasets:
Formats:
parquet
Sub-tasks:
abstractive-qa
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
speech-recognition
missing data on test
#1
by
Splend1dchan
- opened
in github, test_SQUAD has 101 entries, test_OOD has 166 entries, but test.parquet only has 117 entries
Since the testing treat each answer as a unique data, instead in here we treat each question as a unique data, so our number will be smaller.