Datasets:
GEM
/

Languages:
English
Multilinguality:
unknown
Size Categories:
unknown
Language Creators:
unknown
Annotations Creators:
none
Source Datasets:
original
Tags:
data-to-text
License:
viggo / challenge_train_10_percent.csv

Commit History

Adds challenge sets: 5 subsets (20%,10%, 5%, 2%,1%) of the original training set sampled in a controlled way to ensure a reasonable distribution of all dialogue acts.
de33bf0

jjuraska commited on