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

2020/04/30 by Sharan Narang, Colin Raffel, Narang, Sharan +10 · 105 citations
Computer Science · #Algorithm #Artificial intelligence #Artificial neural network #Code (set theory) #Computation and Language (cs.CL) #Computer science #Decoding methods #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Function (biology) #Language model #Leverage (statistics) #Machine Learning (cs.LG) #Machine learning #Natural Language Processing Techniques #Natural language #Natural language processing #Natural language understanding #Programming language #Rationalization (economics) #Set (abstract data type) #Topic Modeling #Training set #cs.CL #cs.LG

paper · pdf · doi:10.48550/arxiv.2004.14546

published in arXiv (Cornell University) (Cornell University)

arxiv created 2020/04/30 · openalex publication_date 2020/04/30 · arxiv updated 2020/05/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

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.

Citations

• Sequence to Sequence Learning with Neural Networks
• Online and Linear-Time Attention by Enforcing Monotonic Alignments
• An Overview of Multi-Task Learning in Deep Neural Networks
• Adafactor: Adaptive Learning Rates with Sublinear Memory Cost
• Multi-Task Deep Neural Networks for Natural Language Understanding
• A large annotated corpus for learning natural language inference

Cited by

• Reason2Decide: Rationale-Driven Multi-Task Learning
• Toward Trustworthy Difficulty Assessments: Large Language Models as Judges in Programming and Synthetic Tasks
• The Scaling Properties of Implicit Deductive Reasoning in Transformers
• Towards Transparent AI: A Survey on Explainable Language Models
• Interpreting Deep Learning Models in Natural Language Processing: A Review
• ExplaGraphs: An Explanation Graph Generation Task for Structured Commonsense Reasoning
• Adversarial Attacks on VQA-NLE: Exposing and Alleviating Inconsistencies in Visual Question Answering Explanations
• Are Training Resources Insufficient? Predict First Then Explain!
• mT5: A massively multilingual pre-trained text-to-text transformer
• Towards Large Language Models with Self-Consistent Natural Language Explanations
• Towards Transparent AI: A Survey on Explainable Large Language Models
• Investigating the Effect of Natural Language Explanations on Out-of-Distribution Generalization in Few-shot NLI
• Leakage-Adjusted Simulatability: Can Models Generate Non-Trivial Explanations of Their Behavior in Natural Language?
• Iterative Self-Improvement of Vision Language Models for Image Scoring and Self-Explanation
• A Necessary Step toward Faithfulness: Measuring and Improving Consistency in Free-Text Explanations
• You Can Do Better! If You Elaborate the Reason When Making Prediction
• Pushing the boundary on Natural Language Inference
• ChartQA-X: Generating Explanations for Visual Chart Reasoning

Related

• Learning to Faithfully Rationalize by Construction
• You Can Do Better! If You Elaborate the Reason When Making Prediction
• Are Training Resources Insufficient? Predict First Then Explain!
• Towards Explainable NLP: A Generative Explanation Framework for Text Classification
• Graph-Guided Textual Explanation Generation Framework