mlsanityData to 7 Oct 2026

Time Travel in LLMs: Tracing Data Contamination in Large Language Models

arXiv 2308.08493 ↗First posted 16 Aug 2023cs.CL cs.AI cs.CRcited by 175

Not on the map. Why?

Cited by 24 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Data contamination, i.e., the presence of test data from downstream tasks in the training data of large language models (LLMs), is a potential major issue in measuring LLMs' real effectiveness on other tasks. We propose a straightforward yet effective method for identifying data contamination within LLMs. At its core, our approach starts by identifying potential contamination at the instance level; using this information, our approach then assesses wider contamination at the partition level. To estimate contamination of individual instances, we employ "guided instruction:" a prompt consisting of the dataset name, partition type, and the random-length initial segment of a reference instance, asking the LLM to complete it. An instance is flagged as contaminated if the LLM's output either exactly or nearly matches the latter segment of the reference. To understand if an entire partition is contaminated, we propose two ideas. The first idea marks a dataset partition as contaminated if the average overlap score with the reference instances (as measured by ROUGE-L or BLEURT) is statistically significantly better with the completions from guided instruction compared to a "general instruction" that does not include the dataset and partition name. The second idea marks a dataset partition as contaminated if a classifier based on GPT-4 with few-shot in-context learning prompt marks multiple generated completions as exact/near-exact matches of the corresponding reference instances. Our best method achieves an accuracy between 92% and 100% in detecting if an LLM is contaminated with seven datasets, containing train and test/validation partitions, when contrasted with manual evaluation by human experts. Further, our findings indicate that GPT-4 is contaminated with AG News, WNLI, and XSum datasets.

Key facts. Cited by 175 papers in the map (data through 7 Oct 2026). Builds on 32 of its 55 references that are in the map. First posted on arXiv 16 Aug 2023.

Builds on

32 of 55 references.

References · 32

  1. 2017Attention Is All You Needcited by 59,932
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2020Language Models are Few-Shot Learnerscited by 32,199
  4. 2023GPT-4 Technical Reportcited by 18,132
  5. 2022Training language models to follow instructions with human feedbackcited by 15,712
  6. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  7. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  8. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  9. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  10. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,854
  11. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,747
  12. 2022Scaling Instruction-Finetuned Language Modelscited by 2,997
20 more
  1. 2023Sparks of Artificial General Intelligence: Early experiments with GPT-4cited by 2,579
  2. 2015Character-level Convolutional Networks for Text Classificationcited by 2,293
  3. 2020Extracting Training Data from Large Language Modelscited by 2,144
  4. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,734
  5. 2019SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systemscited by 1,718
  6. 2021Multitask Prompted Training Enables Zero-Shot Task Generalizationcited by 1,462
  7. 2018Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarizationcited by 1,308
  8. 2023A Survey on Evaluation of Large Language Modelscited by 1,227
  9. 2020BLEURT: Learning Robust Metrics for Text Generationcited by 1,196
  10. 2021GLaM: Efficient Scaling of Language Models with Mixture-of-Expertscited by 890
  11. 2023A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivitycited by 888
  12. 2022Quantifying Memorization Across Neural Language Modelscited by 755
  13. 2023The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Onlycited by 755
  14. 2023OpenAssistant Conversations -- Democratizing Large Language Model Alignmentcited by 712
  15. 2019SAMSum Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarizationcited by 530
  16. 2022Deduplicating Training Data Mitigates Privacy Risks in Language Modelscited by 309
  17. 2022Impact of Pretraining Term Frequencies on Few-Shot Reasoningcited by 142
  18. 2021Learning Compact Metrics for MTcited by 93
  19. 2023Can ChatGPT Reproduce Human-Generated Labels? A Study of Social Computing Taskscited by 88
  20. 2023ChatGPT Participates in a Computer Science Examcited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 10; 2024: 54; 2025: 53; 2026: 58.

175 papers.

Sort:

By area:

  1. 2024Qwen2 Technical Reportcited by 2,076
  2. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
  3. 2024From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judgecited by 480
  4. 2023Detecting Pretraining Data from Large Language Modelscited by 375
  5. 2024LiveBench: A Challenging, Contamination-Limited LLM Benchmarkcited by 242
  6. 2023Proving Test Set Contamination in Black Box Language Modelscited by 227
  7. 2025Generalizing Verifiable Instruction Followingcited by 178
  8. 2024The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Surveycited by 164
  9. 2023Factuality Challenges in the Era of Large Language Modelscited by 125
  10. 2024Black-Box Access is Insufficient for Rigorous AI Auditscited by 123
  11. 2025Preference Leakage: A Contamination Problem in LLM-as-a-judgecited by 119
  12. 2024BetterBench: Assessing AI Benchmarks, Uncovering Issues, and Establishing Best Practicescited by 112
38 more
  1. 2023DyVal: Dynamic Evaluation of Large Language Models for Reasoning Taskscited by 100
  2. 2024InstructRAG: Instructing Retrieval-Augmented Generation via Self-Synthesized Rationalescited by 100
  3. 2024Investigating Data Contamination for Pre-training Language Modelscited by 91
  4. 2025A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibilitycited by 89
  5. 2024A Systematic Survey and Critical Review on Evaluating Large Language Models: Challenges, Limitations, and Recommendationscited by 84
  6. 2024Follow My Instruction and Spill the Beans: Scalable Data Extraction from Retrieval-Augmented Generation Systemscited by 76
  7. 2024DE-COP: Detecting Copyrighted Content in Language Models Training Datacited by 72
  8. 2024Alice in Wonderland: Simple Tasks Showing Complete Reasoning Breakdown in State-Of-the-Art Large Language Modelscited by 67
  9. 2024From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examplescited by 55
  10. 2024Dynamic Evaluation of Large Language Models by Meta Probing Agentscited by 54
  11. 2023BAMBOO: A Comprehensive Benchmark for Evaluating Long Text Modeling Capacities of Large Language Modelscited by 53
  12. 2024Watermarking Makes Language Models Radioactivecited by 52
  13. 2024Conformal Prediction for Natural Language Processing: A Surveycited by 52
  14. 2024What's Wrong with Your Code Generated by Large Language Models? An Extensive Studycited by 52
  15. 2024NegotiationToM: A Benchmark for Stress-testing Machine Theory of Mind on Negotiation Surroundingcited by 49
  16. 2023Data Contamination Quiz: A Tool to Detect and Estimate Contamination in Large Language Modelscited by 46
  17. 2023Data Contamination Through the Lens of Timecited by 39
  18. 2025Neither Valid nor Reliable? Investigating the Use of LLMs as Judgescited by 37
  19. 2024ConStat: Performance-Based Contamination Detection in Large Language Modelscited by 33
  20. 2024Training on the Test Task Confounds Evaluation and Emergencecited by 33
  21. 2024Membership Inference Attacks against Large Vision-Language Modelscited by 31
  22. 2024Evading Data Contamination Detection for Language Models is (too) Easycited by 30
  23. 2024Not All LLM Reasoners Are Created Equalcited by 29
  24. 2023Denevil: Towards Deciphering and Navigating the Ethical Values of Large Language Models via Instruction Learningcited by 26
  25. 2024Data Contamination Can Cross Language Barrierscited by 26
  26. 2024DARG: Dynamic Evaluation of Large Language Models via Adaptive Reasoning Graphcited by 26
  27. 2024TRUCE: Private Benchmarking to Prevent Contamination and Improve Comparative Evaluation of LLMscited by 24
  28. 2024DataGen: Unified Synthetic Dataset Generation via Large Language Modelscited by 22
  29. 2024CodePori: Large-Scale System for Autonomous Software Development Using Multi-Agent Technologycited by 21
  30. 2024Biomedical Large Languages Models Seem not to be Superior to Generalist Models on Unseen Medical Datacited by 21
  31. 2024Benchmarking LLMs' Judgments with No Gold Standardcited by 19
  32. 2024Quantifying Generalization Complexity for Large Language Modelscited by 18
  33. 2024Raising the Bar: Investigating the Values of Large Language Models via Generative Evolving Testingcited by 17
  34. 2025Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmarkcited by 16
  35. 2024How Much Can We Forget about Data Contamination?cited by 14
  36. 2025Personalized Safety in LLMs: A Benchmark and A Planning-Based Agent Approachcited by 14
  37. 2024PertEval: Unveiling Real Knowledge Capacity of LLMs with Knowledge-Invariant Perturbationscited by 13
  38. 2024Exposing the Achilles' Heel: Evaluating LLMs Ability to Handle Mistakes in Mathematical Reasoningcited by 13

50 most cited of 175 All 175 (JSON)