# PairCFR: Enhancing Model Training on Paired Counterfactually Augmented Data through Contrastive Learning

arXiv: https://arxiv.org/abs/2406.06633 · HTML: https://mlsanity.com/paper/2406.06633/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Jun 2024.

## Built on by (most cited first, 7 of 7)

- [A Survey on Natural Language Counterfactual Generation](https://mlsanity.com/paper/2407.03993/) (2024)
- [CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triples](https://mlsanity.com/paper/2508.21083/) (2025)
- [iFlip: Iterative Feedback-driven Counterfactual Example Refinement](https://mlsanity.com/paper/2601.01446/) (2026)
- [RedNote-Vibe: A Dataset for Capturing Temporal Dynamics of AI-Generated Text in Lifestyle Social Media](https://mlsanity.com/paper/2509.22055/) (2025)
- [TypedCSIP: Typed Counterfactual Pretraining for Chinese Legislative Conflict Classification](https://mlsanity.com/paper/2605.25474/) (2026)
- [Gumbel Machine: Counterfactual Student Writing Generation via Gumbel Noise Steering](https://mlsanity.com/paper/2605.27249/) (2026)
- [Is This Evidence Decision-Critical? Learning to Verify Rule-Governed Decisions](https://mlsanity.com/paper/2609.39608/) (2026)

## Abstract (arXiv)

Counterfactually Augmented Data (CAD) involves creating new data samples by applying minimal yet sufficient modifications to flip the label of existing data samples to other classes. Training with CAD enhances model robustness against spurious features that happen to correlate with labels by spreading the casual relationships across different classes. Yet, recent research reveals that training with CAD may lead models to overly focus on modified features while ignoring other important contextual information, inadvertently introducing biases that may impair performance on out-ofdistribution (OOD) datasets. To mitigate this issue, we employ contrastive learning to promote global feature alignment in addition to learning counterfactual clues. We theoretically prove that contrastive loss can encourage models to leverage a broader range of features beyond those modified ones. Comprehensive experiments on two human-edited CAD datasets demonstrate that our proposed method outperforms the state-of-the-art on OOD datasets.