mlsanityData to 7 Oct 2026

The Effect of Natural Distribution Shift on Question Answering Models

arXiv 2004.14444 ↗First posted 29 Apr 2020cs.LG cs.CL stat.MLcited by 104

Not on the map. Why?

Abstract (arXiv)

We build four new test sets for the Stanford Question Answering Dataset (SQuAD) and evaluate the ability of question-answering systems to generalize to new data. Our first test set is from the original Wikipedia domain and measures the extent to which existing systems overfit the original test set. Despite several years of heavy test set re-use, we find no evidence of adaptive overfitting. The remaining three test sets are constructed from New York Times articles, Reddit posts, and Amazon product reviews and measure robustness to natural distribution shifts. Across a broad range of models, we observe average performance drops of 3.8, 14.0, and 17.4 F1 points, respectively. In contrast, a strong human baseline matches or exceeds the performance of SQuAD models on the original domain and exhibits little to no drop in new domains. Taken together, our results confirm the surprising resilience of the holdout method and emphasize the need to move towards evaluation metrics that incorporate robustness to natural distribution shifts.

Key facts. Cited by 104 papers in the map (data through 7 Oct 2026). First posted on arXiv 29 Apr 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 18; 2022: 22; 2023: 33; 2024: 21; 2025: 6; 2026: 4.

104 papers.

Sort:

By area:

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2022Robust Speech Recognition via Large-Scale Weak Supervisioncited by 3,972
  3. 2021Robust fine-tuning of zero-shot modelscited by 785
  4. 2021Accuracy on the Line: On the Strong Correlation Between Out-of-Distribution and In-Distribution Generalizationcited by 248
  5. 2021Predicting with Confidence on Unseen Distributionscited by 116
  6. 2021Robustness Gym: Unifying the NLP Evaluation Landscapecited by 107
  7. 2021Measure and Improve Robustness in NLP Models: A Surveycited by 107
  8. 2023VisIT-Bench: A Benchmark for Vision-Language Instruction Following Inspired by Real-World Usecited by 100
  9. 2022Quality Not Quantity: On the Interaction between Dataset Design and Robustness of CLIPcited by 97
  10. 2022Model Ratatouille: Recycling Diverse Models for Out-of-Distribution Generalizationcited by 97
  11. 2023Revisiting Out-of-distribution Robustness in NLP: Benchmark, Analysis, and LLMs Evaluationscited by 96
  12. 2022Agreement-on-the-Line: Predicting the Performance of Neural Networks under Distribution Shiftcited by 94
38 more
  1. 2021Extending the WILDS Benchmark for Unsupervised Adaptationcited by 85
  2. 2021The Evolution of Out-of-Distribution Robustness Throughout Fine-Tuningcited by 74
  3. 2022Domain-Adjusted Regression or: ERM May Already Learn Features Sufficient for Out-of-Distribution Generalizationcited by 69
  4. 2022Investigating Selective Prediction Approaches Across Several Tasks in IID, OOD, and Adversarial Settingscited by 66
  5. 2023Benchmarking Distribution Shift in Tabular Data with TableShiftcited by 64
  6. 2023Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmarkcited by 62
  7. 2023Red Teaming Language Model Detectors with Language Modelscited by 55
  8. 2022Predicting Out-of-Distribution Error with the Projection Normcited by 50
  9. 2023EcomGPT: Instruction-tuning Large Language Models with Chain-of-Task Tasks for E-commercecited by 48
  10. 2021Flexible Clustered Federated Learning for Client-Level Data Distribution Shiftcited by 47
  11. 2022ID and OOD Performance Are Sometimes Inversely Correlated on Real-world Datasetscited by 44
  12. 2023Diagnosing Model Performance Under Distribution Shiftcited by 37
  13. 2024Stable Neural Stochastic Differential Equations in Analyzing Irregular Time Series Datacited by 35
  14. 2021Knowledge Enhanced Pretrained Language Models: A Compreshensive Surveycited by 33
  15. 2023Mind the instructions: a holistic evaluation of consistency and interactions in prompt-based learningcited by 31
  16. 2024A Survey on Data Augmentation in Large Model Eracited by 31
  17. 2023Benchmarking Robustness of Adaptation Methods on Pre-trained Vision-Language Modelscited by 30
  18. 2022Generative Language Models for Paragraph-Level Question Generationcited by 28
  19. 2024SimRAG: Self-Improving Retrieval-Augmented Generation for Adapting Large Language Models to Specialized Domainscited by 28
  20. 2024Learning to Generate Instruction Tuning Datasets for Zero-Shot Task Adaptationcited by 27
  21. 2023Gradient-Free Structured Pruning with Unlabeled Datacited by 26
  22. 2021Reliability Testing for Natural Language Processing Systemscited by 25
  23. 2024Do Large Language Models Perform the Way People Expect? Measuring the Human Generalization Functioncited by 24
  24. 2022Robustness Analysis of Video-Language Models Against Visual and Language Perturbationscited by 23
  25. 2022Exploring The Landscape of Distributional Robustness for Question Answering Modelscited by 23
  26. 2024Efficient Knowledge Injection in LLMs via Self-Distillationcited by 22
  27. 2021Do Question Answering Modeling Improvements Hold Across Benchmarks?cited by 20
  28. 2021Contrastive Domain Adaptation for Question Answering using Limited Text Corporacited by 20
  29. 2023On the Robustness of Aspect-based Sentiment Analysis: Rethinking Model, Data, and Trainingcited by 20
  30. 2023Post-Abstention: Towards Reliably Re-Attempting the Abstained Instances in QAcited by 20
  31. 2023A Survey on Out-of-Distribution Evaluation of Neural NLP Modelscited by 20
  32. 2024SemEval-2024 Task 2: Safe Biomedical Natural Language Inference for Clinical Trialscited by 20
  33. 2024Data Contamination Report from the 2024 CONDA Shared Taskcited by 19
  34. 2024Look, Listen, and Answer: Overcoming Biases for Audio-Visual Question Answeringcited by 18
  35. 2023An Empirical Comparison of LM-based Question and Answer Generation Methodscited by 15
  36. 2021MLDemon: Deployment Monitoring for Machine Learning Systemscited by 14
  37. 2021Learning to Perturb Word Embeddings for Out-of-distribution QAcited by 13
  38. 2023Effective Robustness against Natural Distribution Shifts for Models with Different Training Datacited by 13

50 most cited of 104 All 104 (JSON)