Datasets:
Sample mismatch v5 and v5.1 not reported in docs
#8
by
nikithasr08 - opened
We notice that the Sample IDs in v5 and v5.1 are entirely different in different splits of train/val/test. However this doesn't seem to be indicated in the dataset card. (Papers using the card might be comparing different test sets in that case)
Updated the card
kyamagu changed discussion status to
closed