mlsanityData to 7 Oct 2026

Explore Spurious Correlations at the Concept Level in Language Models for Text Classification

arXiv 2311.08648 ↗First posted 15 Nov 2023cs.CL cs.AIcited by 48

Not on the map. Why?

Abstract (arXiv)

Language models (LMs) have achieved notable success in numerous NLP tasks, employing both fine-tuning and in-context learning (ICL) methods. While language models demonstrate exceptional performance, they face robustness challenges due to spurious correlations arising from imbalanced label distributions in training data or ICL exemplars. Previous research has primarily concentrated on word, phrase, and syntax features, neglecting the concept level, often due to the absence of concept labels and difficulty in identifying conceptual content in input texts. This paper introduces two main contributions. First, we employ ChatGPT to assign concept labels to texts, assessing concept bias in models during fine-tuning or ICL on test data. We find that LMs, when encountering spurious correlations between a concept and a label in training or prompts, resort to shortcuts for predictions. Second, we introduce a data rebalancing technique that incorporates ChatGPT-generated counterfactual data, thereby balancing label distribution and mitigating spurious correlations. Our method's efficacy, surpassing traditional token removal approaches, is validated through extensive testing.

Key facts. Cited by 48 papers in the map (data through 7 Oct 2026). Builds on 48 of its 73 references that are in the map. First posted on arXiv 15 Nov 2023.

Builds on

48 of 73 references.

References · 48

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  5. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  6. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  7. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,018
  8. 2015Character-level Convolutional Networks for Text Classificationcited by 2,293
  9. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,140
  10. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  11. 2021Calibrate Before Use: Improving Few-Shot Performance of Language Modelscited by 1,371
  12. 2020Making Pre-trained Language Models Better Few-shot Learnerscited by 1,255
36 more
  1. 2021Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivitycited by 1,122
  2. 2020Beyond Accuracy: Behavioral Testing of NLP models with CheckListcited by 979
  3. 2020Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inferencecited by 934
  4. 2019Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inferencecited by 864
  5. 2023ChatGPT Outperforms Crowd-Workers for Text-Annotation Taskscited by 849
  6. 2019Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailmentcited by 663
  7. 2018Gender Bias in Coreference Resolution: Evaluation and Debiasing Methodscited by 657
  8. 2017Adversarial Examples for Evaluating Reading Comprehension Systemscited by 568
  9. 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 522
  10. 2016Ups and Downs: Modeling the Visual Evolution of Fashion Trends with One-Class Collaborative Filteringcited by 517
  11. 2019Learning the Difference that Makes a Difference with Counterfactually-Augmented Datacited by 428
  12. 2018Generating Natural Language Adversarial Examplescited by 422
  13. 2020An Investigation of Why Overparameterization Exacerbates Spurious Correlationscited by 310
  14. 2018Hypothesis Only Baselines in Natural Language Inferencecited by 291
  15. 2022Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlationscited by 171
  16. 2020Adversarial Filters of Dataset Biasescited by 159
  17. 2020Algorithmic Fairnesscited by 143
  18. 2020An Empirical Study on Robustness to Spurious Correlations using Pre-trained Language Modelscited by 133
  19. 2019Unlearn Dataset Bias in Natural Language Inference by Fitting the Residualcited by 119
  20. 2020Mind the Trade-off: Debiasing NLU Models without Degrading the In-distribution Performancecited by 85
  21. 2021Identifying and Mitigating Spurious Correlations for Improving Robustness in NLP Modelscited by 76
  22. 2023Large Language Models Can be Lazy Learners: Analyze Shortcuts in In-Context Learningcited by 69
  23. 2022CEBaB: Estimating the Causal Effects of Real-World Concepts on NLP Model Behaviorcited by 54
  24. 2023Measuring Inductive Biases of In-Context Learning with Underspecified Demonstrationscited by 49
  25. 2020Automatic Shortcut Removal for Self-Supervised Representation Learningcited by 43
  26. 2020Learning to Model and Ignore Dataset Bias with Mixed Capacity Ensemblescited by 43
  27. 2021Why Machine Reading Comprehension Models Learn Shortcuts?cited by 40
  28. 2021Low Frequency Names Exhibit Bias and Overfitting in Contextualizing Language Modelscited by 39
  29. 2020Evaluating Robustness to Input Perturbations for Neural Machine Translationcited by 36
  30. 2020Towards Robustifying NLI Models Against Lexical Dataset Biasescited by 34
  31. 2022Finding Dataset Shortcuts with Grammar Inductioncited by 15
  32. 2023Scalable Prompt Generation for Semi-supervised Learning with Language Modelscited by 10
  33. 2023CAT: Causal Audio Transformer for Audio Classificationcited by 10
  34. 2023C-Disentanglement: Discovering Causally-Independent Generative Factors under an Inductive Bias of Confoundercited by 7
  35. 2023Measuring Spurious Correlation in Classification: 'Clever Hans' in Translationesecited by 5
  36. 2022Investigating Ensemble Methods for Model Robustness Improvement of Text Classifierscited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 13; 2025: 23; 2026: 12.

48 papers.

Sort:

By area:

  1. 2024Mementos: A Comprehensive Benchmark for Multimodal Large Language Model Reasoning over Image Sequencescited by 93
  2. 2024Enhancing Visual-Language Modality Alignment in Large Vision Language Models via Self-Improvementcited by 74
  3. 2025Which of These Best Describes Multiple Choice Evaluation with LLMs? A) Forced B) Flawed C) Fixable D) All of the Abovecited by 51
  4. 2025The Best Instruction-Tuning Data are Those That Fitcited by 48
  5. 2025Beyond Reward Hacking: Causal Rewards for Large Language Model Alignmentcited by 42
  6. 2024The Clever Hans Mirage: A Comprehensive Survey on Spurious Correlations in Machine Learningcited by 39
  7. 2025Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignmentcited by 32
  8. 2024Large Language Models and Causal Inference in Collaboration: A Surveycited by 26
  9. 2024A Survey on Natural Language Counterfactual Generationcited by 21
  10. 2024CausalChaos! Dataset for Comprehensive Causal Action Question Answering Over Longer Causal Chains Grounded in Dynamic Visual Scenescited by 13
  11. 2025MergeME: Model Merging Techniques for Homogeneous and Heterogeneous MoEscited by 13
  12. 2024Shortcut Learning in In-Context Learning: A Surveycited by 8
36 more
  1. 2024Emojis Decoded: Leveraging ChatGPT for Enhanced Understanding in Social Media Communicationscited by 7
  2. 2025BOAD: Discovering Hierarchical Software Engineering Agents via Bandit Optimizationcited by 7
  3. 2025DISCO Balances the Scales: Adaptive Domain- and Difficulty-Aware Reinforcement Learning on Imbalanced Datacited by 6
  4. 2025Rectifying Shortcut Behaviors in Preference-based Reward Learningcited by 6
  5. 2025Debiasing Reward Models by Representation Learning with Guaranteescited by 6
  6. 2024Mitigating Biases of Large Language Models in Stance Detection with Counterfactual Augmented Calibrationcited by 5
  7. 2024Multi-Stage Balanced Distillation: Addressing Long-Tail Challenges in Sequence-Level Knowledge Distillationcited by 5
  8. 2024Focus On This, Not That! Steering LLMs with Adaptive Feature Specificationcited by 5
  9. 2025Evaluate Bias without Manual Test Sets: A Concept Representation Perspective for LLMscited by 5
  10. 2025Stochastic Chameleons: Irrelevant Context Hallucinations Reveal Class-Based (Mis)Generalization in LLMscited by 5
  11. 2024CSRec: Rethinking Sequential Recommendation from A Causal Perspectivecited by 4
  12. 2025Skill Discovery for Software Scripting Automation via Offline Simulations with LLMscited by 3
  13. 2025When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMscited by 3
  14. 2026Retrieving Counterfactuals Improves Visual In-Context Learningcited by 3
  15. 2025Exploring Explanations Improves the Robustness of In-Context Learningcited by 2
  16. 2025LoRA Users Beware: A Few Spurious Tokens Can Manipulate Your Finetuned Modelcited by 2
  17. 2024From Adoption to Adaption: Tracing the Diffusion of New Emojis on Twittercited by 1
  18. 2025Syntactic Blind Spots: How Misalignment Leads to LLMs Mathematical Errorscited by 1
  19. 2026Factored Causal Representation Learning for Robust Reward Modeling in RLHFcited by 1
  20. 2026Towards Effective In-context Cross-domain Knowledge Transfer via Domain-invariant-neurons-based Retrievalcited by 1
  21. 2026Reason Analogically via Cross-domain Prior Knowledge: An Empirical Study of Cross-domain Knowledge Transfer for In-Context Learningcited by 1
  22. 2026Consistency Training Can Entrench Misalignmentcited by 1
  23. 2025Large Language Models Are Still Misled by Simple Bias Ensemblescited by 0
  24. 2025TARS: MinMax Token-Adaptive Preference Strategy for Hallucination Reduction in MLLMscited by 0
  25. 2025Hallucination-Resistant Relation Extraction via Dependency-Aware Sentence Simplification and Two-tiered Hierarchical Refinementcited by 0
  26. 2025CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triplescited by 0
  27. 2025CURE: Controlled Unlearning for Robust Embeddings -- Mitigating Conceptual Shortcuts in Pre-Trained Language Modelscited by 0
  28. 2025Expert-guided Clinical Text Augmentation via Query-Based Model Collaborationcited by 0
  29. 2025Debiasing Large Language Models in Thai Political Stance Detection via Counterfactual Calibrationcited by 0
  30. 2026What do your logits know?cited by 0
  31. 2026Models Know Their Shortcuts: Deployment-Time Shortcut Mitigationcited by 0
  32. 2026Robust Reward Modeling for Large Language Models via Causal Decompositioncited by 0
  33. 2026Does Topic Sentiment Cause Perceived Ideology? Comparing Human and LLM Annotations in Political News Articlescited by 0
  34. 2026Style or Content? Evaluating Style Classifiers with Controlled Content Overlapcited by 0
  35. 2026Shortcuts in the Tail: Debiasing via Post-Hoc Spectral Compression of Fine-Tuning Updatescited by 0
  36. 2026What do Reward Models Memorize?cited by 0