e-SNLI: Natural Language Inference with Natural Language Explanations
This work addresses the need for interpretable AI by enabling models to learn from and output explanations, potentially benefiting researchers and practitioners in natural language processing.
The authors extended the Stanford Natural Language Inference dataset with human-annotated natural language explanations, creating e-SNLI, and demonstrated its use for tasks like providing model justifications and improving sentence representations.
In order for machine learning to garner widespread public adoption, models must be able to provide interpretable and robust explanations for their decisions, as well as learn from human-provided explanations at train time. In this work, we extend the Stanford Natural Language Inference dataset with an additional layer of human-annotated natural language explanations of the entailment relations. We further implement models that incorporate these explanations into their training process and output them at test time. We show how our corpus of explanations, which we call e-SNLI, can be used for various goals, such as obtaining full sentence justifications of a model's decisions, improving universal sentence representations and transferring to out-of-domain NLI datasets. Our dataset thus opens up a range of research directions for using natural language explanations, both for improving models and for asserting their trust.