Chaos NLI MNLI portion with gini coefficient pre-computed (from 0 to 1) High gini means unambiguous inference.

@inproceedings{xzhou2022distnli,
    Author = {Xiang Zhou and Yixin Nie and Mohit Bansal},
    Booktitle = {Findings of the Association for Computational Linguistics: ACL 2022},
    Publisher = {Association for Computational Linguistics},
    Title = {Distributed NLI: Learning to Predict Human Opinion Distributions for Language Reasoning},
    Year = {2022}
}