# Debiasing Masks: A New Framework for Shortcut Mitigation in NLU

arXiv: https://arxiv.org/abs/2210.16079 · HTML: https://mlsanity.com/paper/2210.16079/

## Key facts

- Cited by 16 papers in the map (data through 14 Oct 2026).
- Builds on 26 of its 28 references that are in the map.
- First posted on arXiv 28 Oct 2022.

## Builds on (most cited first, 26 of 26)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding](https://mlsanity.com/paper/1804.07461/) (2018)
- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](https://mlsanity.com/paper/1606.05250/) (2016)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [FEVER: a large-scale dataset for Fact Extraction and VERification](https://mlsanity.com/paper/1803.05355/) (2018)
- [BERT Rediscovers the Classical NLP Pipeline](https://mlsanity.com/paper/1905.05950/) (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)
- [Adversarial Examples for Evaluating Reading Comprehension Systems](https://mlsanity.com/paper/1707.07328/) (2017)
- [Underspecification Presents Challenges for Credibility in Modern Machine Learning](https://mlsanity.com/paper/2011.03395/) (2020)
- [Movement Pruning: Adaptive Sparsity by Fine-Tuning](https://mlsanity.com/paper/2005.07683/) (2020)
- [PAWS: Paraphrase Adversaries from Word Scrambling](https://mlsanity.com/paper/1904.01130/) (2019)
- [Don't Take the Easy Way Out: Ensemble Based Methods for Avoiding Known Dataset Biases](https://mlsanity.com/paper/1909.03683/) (2019)
- [Compressing BERT: Studying the Effects of Weight Pruning on Transfer Learning](https://mlsanity.com/paper/2002.08307/) (2020)
- [Adversarial Filters of Dataset Biases](https://mlsanity.com/paper/2002.04108/) (2020)
- [Towards Debiasing Fact Verification Models](https://mlsanity.com/paper/1908.05267/) (2019)
- [Well-Read Students Learn Better: On the Importance of Pre-training Compact Models](https://mlsanity.com/paper/1908.08962/) (2019)
- [Towards Debiasing NLU Models from Unknown Biases](https://mlsanity.com/paper/2009.12303/) (2020)
- [Adversarial Training for Large Neural Language Models](https://mlsanity.com/paper/2004.08994/) (2020)
- [Masking as an Efficient Alternative to Finetuning for Pretrained Language Models](https://mlsanity.com/paper/2004.12406/) (2020)
- [Competency Problems: On Finding and Removing Artifacts in Language Data](https://mlsanity.com/paper/2104.08646/) (2021)
- [Mind the Trade-off: Debiasing NLU Models without Degrading the In-distribution Performance](https://mlsanity.com/paper/2005.00315/) (2020)
- [Learning from others' mistakes: Avoiding dataset biases without modeling them](https://mlsanity.com/paper/2012.01300/) (2020)
- [Robust Machine Comprehension Models via Adversarial Training](https://mlsanity.com/paper/1804.06473/) (2018)
- [Avoiding Inference Heuristics in Few-shot Prompt-based Finetuning](https://mlsanity.com/paper/2109.04144/) (2021)

## Built on by (most cited first, 16 of 16)

- [Sociodemographic Bias in Language Models: A Survey and Forward Path](https://mlsanity.com/paper/2306.08158/) (2023)
- [Robust Natural Language Understanding with Residual Attention Debiasing](https://mlsanity.com/paper/2305.17627/) (2023)
- [Attention Pruning: Automated Fairness Repair of Language Models via Surrogate Simulated Annealing](https://mlsanity.com/paper/2503.15815/) (2025)
- [EqualizeIR: Mitigating Linguistic Biases in Retrieval Models](https://mlsanity.com/paper/2504.07115/) (2025)
- [Local Contrastive Editing of Gender Stereotypes](https://mlsanity.com/paper/2410.17739/) (2024)
- [Mitigating Shortcuts in Language Models with Soft Label Encoding](https://mlsanity.com/paper/2309.09380/) (2023)
- [Mitigating Shortcut Learning with InterpoLated Learning](https://mlsanity.com/paper/2507.05527/) (2025)
- [Do Students Debias Like Teachers? On the Distillability of Bias Mitigation Methods](https://mlsanity.com/paper/2510.26038/) (2025)
- [Domain Information Control at Inference Time for Acoustic Scene Classification](https://mlsanity.com/paper/2306.08010/) (2023)
- [Confounder Balancing in Adversarial Domain Adaptation for Pre-Trained Large Models Fine-Tuning](https://mlsanity.com/paper/2310.16062/) (2023)
- [A Note on Bias to Complete](https://mlsanity.com/paper/2402.11710/) (2024)
- [Not Eliminate but Aggregate: Post-Hoc Control over Mixture-of-Experts to Address Shortcut Shifts in Natural Language Understanding](https://mlsanity.com/paper/2406.12060/) (2024)
- [DBR: Divergence-Based Regularization for Debiasing Natural Language Understanding Models](https://mlsanity.com/paper/2502.18353/) (2025)
- [FairFlow: Mitigating Dataset Biases through Undecided Learning](https://mlsanity.com/paper/2503.17632/) (2025)
- [Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Models](https://mlsanity.com/paper/2603.05582/) (2026)
- [Fair Cognitive Impairment Detection Through Unlearning](https://mlsanity.com/paper/2606.18571/) (2026)

## Abstract (arXiv)

Debiasing language models from unwanted behaviors in Natural Language Understanding tasks is a topic with rapidly increasing interest in the NLP community. Spurious statistical correlations in the data allow models to perform shortcuts and avoid uncovering more advanced and desirable linguistic features. A multitude of effective debiasing approaches has been proposed, but flexibility remains a major issue. For the most part, models must be retrained to find a new set of weights with debiased behavior. We propose a new debiasing method in which we identify debiased pruning masks that can be applied to a finetuned model. This enables the selective and conditional application of debiasing behaviors. We assume that bias is caused by a certain subset of weights in the network; our method is, in essence, a mask search to identify and remove biased weights. Our masks show equivalent or superior performance to the standard counterparts, while offering important benefits. Pruning masks can be stored with high efficiency in memory, and it becomes possible to switch among several debiasing behaviors (or revert back to the original biased model) at inference time. Finally, it opens the doors to further research on how biases are acquired by studying the generated masks. For example, we observed that the early layers and attention heads were pruned more aggressively, possibly hinting towards the location in which biases may be encoded.