WT5?! Training Text-to-Text Models to Explain their Predictions

Abstract

Neural networks have recently achieved human-level performance on various challenging natural language processing (NLP) tasks, but it is notoriously difficult to understand why a neural network produced a particular prediction. In this paper, we leverage the text-to-text framework proposed by Raffel et al.(2019) to train language models to output a natural text explanation alongside their prediction. Crucially, this requires no modifications to the loss function or training and decoding procedures -- we simply train the model to output the explanation after generating the (natural text) prediction. We show that this approach not only obtains state-of-the-art results on explainability benchmarks, but also permits learning from a limited set of labeled explanations and transferring rationalization abilities across datasets. To facilitate reproducibility and future work, we release our code use to train the models.

References in corpus (8)

• Sequence to Sequence Learning with Neural Networks
• Towards A Rigorous Science of Interpretable Machine Learning
• An Overview of Multi-Task Learning in Deep Neural Networks
• SmoothGrad: removing noise by adding noise
• Attention is not Explanation
• Adversarial Examples for Evaluating Reading Comprehension Systems
• Multi-Task Deep Neural Networks for Natural Language Understanding
• The (Un)reliability of saliency methods

Cited by in corpus (2)

• Interpreting Deep Learning Models in Natural Language Processing: A Review
• Adversarial NLI for Factual Correctness in Text Summarisation Models