This dataset is a concatenation of the SNLI and MultiNLI datasets.
This is the same dataset as sentence-transformers/all-nli pair-class split;
however, the label ids are not identical, and teacher scores have been added from dleemiller/ModernCE-large-nli.
I have also added hashes for score lookup, since a lookup must be added into a custom loss function,
if using the sentence transformers CrossEncoder trainer.
The hashes were computed straightforwardly as… See the full description on the dataset page:
https://huggingface.co/datasets/dleemiller/all-nli-distill.
