Closed Elfsong closed 10 months ago
Hi,
Sorry for missing your request--the processed NLI dataset can be accessed here. I believe the former Drive link led to a version of the dataset that was not shuffled properly, which affects the distributions in the train / val / test partitions.
Alternatively, you can re-generate the dataset by following the instructions from the original source here. You could do some light pre-processing to get the columns formatted as [premise
, hypothesis
, label
], in which the label
value should always be entailment
(or 1), and then shuffle all the examples.
Really appreciate! Thank you for making this solid work.
Hi there,
Could you share the processed version of the Bias-NLI dataset? it needs to request access, I submitted my request already. Thank you!
Best regards, Mingzhe