biobert
biobert copied to clipboard
Split for JNLPBA dataset.
What is the split used for the JNLPBA dataset? And why do we have 4 splits i.e., train.tsv, train_dev.tsv, devel.tsv, test.tsv?