Datasets:

machelreid

/

m2d2

like 5

Hi there, thanks for making the dataset open. I was playing with the Wikipedia domains and found the length of data instances vary a lot. For example, some contain a lengthy paragraph while some contain only a few characters. In the paper it was not mentioned how data are assembled from Wiki dump.