about the size of xnli
#4
by
Mabrouka
- opened
I have a confusion regarding the size of XNLI mentioned in the paper, it states that the dataset consists of 7500 sentence pairs in total, but I am observing it to be much larger. Can you explain this discrepancy?
Hi @Mabrouka , the paper says XNLI extends the MultiNLI dataset with 7500 development and test examples:
- validation: 2490
- test: 5010
albertvillanova
changed discussion status to
closed