# Learning to Learn to be Right for the Right Reasons

arXiv: https://arxiv.org/abs/2104.11514 · HTML: https://mlsanity.com/paper/2104.11514/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 24 references that are in the map.
- First posted on arXiv 23 Apr 2021.

## Builds on (most cited first, 19 of 19)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks](https://mlsanity.com/paper/1703.03400/) (2017)
- [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://mlsanity.com/paper/1909.11942/) (2019)
- [Matching Networks for One Shot Learning](https://mlsanity.com/paper/1606.04080/) (2016)
- [Unsupervised Domain Adaptation by Backpropagation](https://mlsanity.com/paper/1409.7495/) (2014)
- [SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems](https://mlsanity.com/paper/1905.00537/) (2019)
- [Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference](https://mlsanity.com/paper/1902.01007/) (2019)
- [Annotation Artifacts in Natural Language Inference Data](https://mlsanity.com/paper/1803.02324/) (2018)
- [SWAG: A Large-Scale Adversarial Dataset for Grounded Commonsense Inference](https://mlsanity.com/paper/1808.05326/) (2018)
- [Probing Neural Network Comprehension of Natural Language Arguments](https://mlsanity.com/paper/1907.07355/) (2019)
- [Meta-Learning Representations for Continual Learning](https://mlsanity.com/paper/1905.12588/) (2019)
- [Towards Debiasing Fact Verification Models](https://mlsanity.com/paper/1908.05267/) (2019)
- [Learning to Continually Learn](https://mlsanity.com/paper/2002.09571/) (2020)
- [What Makes Reading Comprehension Questions Easier?](https://mlsanity.com/paper/1808.09384/) (2018)
- [Does It Make Sense? And Why? A Pilot Study for Sense Making and Explanation](https://mlsanity.com/paper/1906.00363/) (2019)
- [When Choosing Plausible Alternatives, Clever Hans can be Clever](https://mlsanity.com/paper/1911.00225/) (2019)
- [On Adversarial Removal of Hypothesis-only Bias in Natural Language Inference](https://mlsanity.com/paper/1907.04389/) (2019)

## Built on by (most cited first, 4 of 4)

- [State-of-the-art generalisation research in NLP: A taxonomy and review](https://mlsanity.com/paper/2210.03050/) (2022)
- [Less Learn Shortcut: Analyzing and Mitigating Learning of Spurious Feature-Label Correlation](https://mlsanity.com/paper/2205.12593/) (2022)
- [Not quite Sherlock Holmes: Language model predictions do not reliably differentiate impossible from improbable events](https://mlsanity.com/paper/2506.06808/) (2025)
- [Model Guidance via Robust Feature Attribution](https://mlsanity.com/paper/2506.19680/) (2025)

## Abstract (arXiv)

Improving model generalization on held-out data is one of the core objectives in commonsense reasoning. Recent work has shown that models trained on the dataset with superficial cues tend to perform well on the easy test set with superficial cues but perform poorly on the hard test set without superficial cues. Previous approaches have resorted to manual methods of encouraging models not to overfit to superficial cues. While some of the methods have improved performance on hard instances, they also lead to degraded performance on easy instances. Here, we propose to explicitly learn a model that does well on both the easy test set with superficial cues and hard test set without superficial cues. Using a meta-learning objective, we learn such a model that improves performance on both the easy test set and the hard test set. By evaluating our models on Choice of Plausible Alternatives (COPA) and Commonsense Explanation, we show that our proposed method leads to improved performance on both the easy test set and the hard test set upon which we observe up to 16.5 percentage points improvement over the baseline.