mlsanityData to 7 Oct 2026

Task Contamination: Language Models May Not Be Few-Shot Anymore

arXiv 2312.16337 ↗First posted 26 Dec 2023cs.CLcited by 98

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) offer impressive performance in various zero-shot and few-shot tasks. However, their success in zero-shot and few-shot settings may be affected by task contamination, a potential limitation that has not been thoroughly examined. This paper investigates how zero-shot and few-shot performance of LLMs has changed chronologically over time. Utilizing GPT-3 series models and several other recent open-sourced LLMs, and controlling for dataset difficulty, we find that on datasets released before the LLM training data creation date, LLMs perform surprisingly better than on datasets released after. This strongly indicates that, for many LLMs, there exists task contamination on zero-shot and few-shot evaluation for datasets released prior to the LLMs' training data creation date. Additionally, we utilize training data inspection, task example extraction, and a membership inference attack, which reveal further evidence of task contamination. Importantly, we find that for classification tasks with no possibility of task contamination, LLMs rarely demonstrate statistically significant improvements over simple majority baselines, in both zero and few-shot settings.

Key facts. Cited by 98 papers in the map (data through 7 Oct 2026). Builds on 29 of its 47 references that are in the map. First posted on arXiv 26 Dec 2023.

Builds on

29 of 47 references.

References · 29

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Training language models to follow instructions with human feedbackcited by 15,712
  3. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  4. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  5. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,072
  6. 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,769
  7. 2022OPT: Open Pre-trained Transformer Language Modelscited by 3,464
  8. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,140
  9. 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,970
  10. 2022BLOOM: A 176B-Parameter Open-Access Multilingual Language Modelcited by 1,968
  11. 2019SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systemscited by 1,718
  12. 2021Did Aristotle Use a Laptop? A Question Answering Benchmark with Implicit Reasoning Strategiescited by 736
17 more
  1. 2023Is ChatGPT a General-Purpose Natural Language Processing Task Solver?cited by 486
  2. 2021Learning How to Ask: Querying LMs with Mixtures of Soft Promptscited by 376
  3. 2018WiC: the Word-in-Context Dataset for Evaluating Context-Sensitive Meaning Representationscited by 370
  4. 2023NLP Evaluation in trouble: On the Need to Measure LLM Data Contamination for each Benchmarkcited by 339
  5. 2023GPT4Tools: Teaching Large Language Model to Use Tools via Self-instructioncited by 245
  6. 2022Data Contamination: From Memorization to Exploitationcited by 234
  7. 2023Proving Test Set Contamination in Black Box Language Modelscited by 227
  8. 2023Don't Make Your LLM an Evaluation Benchmark Cheatercited by 194
  9. 2023Stop Uploading Test Data in Plain Text: Practical Strategies for Mitigating Data Contamination by Evaluation Benchmarkscited by 139
  10. 2023Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4cited by 123
  11. 2018Transforming Question Answering Datasets Into Natural Language Inference Datasetscited by 99
  12. 2020Few-Shot Text Generation with Pattern-Exploiting Trainingcited by 87
  13. 2022Iteratively Prompt Pre-trained Language Models for Chain of Thoughtcited by 75
  14. 2023Trillion Dollar Words: A New Financial Dataset, Task & Market Analysiscited by 73
  15. 2023Searching for Needles in a Haystack: On the Role of Incidental Bilingualism in PaLM's Translation Capabilitycited by 62
  16. 2023Estimating Contamination via Perplexity: Quantifying Memorisation in Language Model Evaluationcited by 47
  17. 2022Prompting Language Models for Linguistic Structurecited by 39

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 55; 2025: 26; 2026: 17.

98 papers.

Sort:

By area:

  1. 2024Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Toolscited by 170
  2. 2024Benchmark Data Contamination of Large Language Models: A Surveycited by 138
  3. 2025Preference Leakage: A Contamination Problem in LLM-as-a-judgecited by 119
  4. 2024A Systematic Survey and Critical Review on Evaluating Large Language Models: Challenges, Limitations, and Recommendationscited by 84
  5. 2024Functional Benchmarks for Robust Evaluation of Reasoning Performance, and the Reasoning Gapcited by 79
  6. 2024ToMBench: Benchmarking Theory of Mind in Large Language Modelscited by 70
  7. 2024Alice in Wonderland: Simple Tasks Showing Complete Reasoning Breakdown in State-Of-the-Art Large Language Modelscited by 67
  8. 2024SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)cited by 54
  9. 2024NegotiationToM: A Benchmark for Stress-testing Machine Theory of Mind on Negotiation Surroundingcited by 49
  10. 2024DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Expertscited by 42
  11. 2024CharacterBox: Evaluating the Role-Playing Capabilities of LLMs in Text-Based Virtual Worldscited by 38
  12. 2024Elephants Never Forget: Memorization and Learning of Tabular Data in Large Language Modelscited by 34
38 more
  1. 2024Training on the Test Task Confounds Evaluation and Emergencecited by 33
  2. 2025A Survey on Data Contamination for Large Language Modelscited by 33
  3. 2024Evading Data Contamination Detection for Language Models is (too) Easycited by 30
  4. 2024Unveiling the Spectrum of Data Contamination in Language Models: A Survey from Detection to Remediationcited by 30
  5. 2024VarBench: Robust Language Model Benchmarking Through Dynamic Variable Perturbationcited by 29
  6. 2024Does Data Contamination Detection Work (Well) for LLMs? A Survey and Evaluation on Detection Assumptionscited by 26
  7. 2024Knowledge Distillation in Automated Annotation: Supervised Text Classification with LLM-Generated Training Labelscited by 25
  8. 2024TRUCE: Private Benchmarking to Prevent Contamination and Improve Comparative Evaluation of LLMscited by 24
  9. 2024ASR Error Correction using Large Language Modelscited by 23
  10. 2024MECD: Unlocking Multi-Event Causal Discovery in Video Reasoningcited by 23
  11. 2024Are LLMs Prescient? A Continuous Evaluation using Daily News as the Oraclecited by 19
  12. 2024Inference-Time Decontamination: Reusing Leaked Benchmarks for Large Language Model Evaluationcited by 18
  13. 2025Reality Check: A New Evaluation Ecosystem Is Necessary to Understand AI's Real World Effectscited by 18
  14. 2024Effects of diversity incentives on sample diversity and downstream model performance in LLM-based text augmentationcited by 17
  15. 2024Comparing Specialised Small and General Large Language Models on Text Classification: 100 Labelled Samples to Achieve Break-Even Performancecited by 17
  16. 2024Keeping Humans in the Loop: Human-Centered Automated Annotation with Generative AIcited by 16
  17. 2024On Catastrophic Inheritance of Large Foundation Modelscited by 15
  18. 2025PuzzleBench: A Fully Dynamic Evaluation Framework for Large Multimodal Models on Puzzle Solvingcited by 15
  19. 2024Promises and pitfalls of artificial intelligence for legal applicationscited by 14
  20. 2024VersiCode: Towards Version-controllable Code Generationcited by 14
  21. 2024PRESTO: Progressive Pretraining Enhances Synthetic Chemistry Outcomescited by 14
  22. 2024How Much Can We Forget about Data Contamination?cited by 14
  23. 2024Learning To See But Forgetting To Follow: Visual Instruction Tuning Makes LLMs More Prone To Jailbreak Attackscited by 13
  24. 2024The Sociolinguistic Foundations of Language Modelingcited by 13
  25. 2024A Comprehensive Survey of Contamination Detection Methods in Large Language Modelscited by 12
  26. 2024Feature contamination: Neural networks learn uncorrelated features and fail to generalizecited by 11
  27. 2024A Taxonomy for Data Contamination in Large Language Modelscited by 11
  28. 2025DCR: Quantifying Data Contamination in LLMs Evaluationcited by 10
  29. 2024The Mosaic Memory of Large Language Modelscited by 9
  30. 2025NetArena: Dynamic Benchmarks for AI Agents in Network Automationcited by 8
  31. 2025Evaluating GPT- and Reasoning-based Large Language Models on Physics Olympiad Problems: Surpassing Human Performance and Implications for Educational Assessmentcited by 7
  32. 2025Unlocking Post-hoc Dataset Inference with Synthetic Datacited by 7
  33. 2024Survey of Computerized Adaptive Testing: A Machine Learning Perspectivecited by 6
  34. 2024Prompting with Phonemes: Enhancing LLMs' Multilinguality for Non-Latin Script Languagescited by 6
  35. 2024Benchmarking zero-shot stance detection with FlanT5-XXL: Insights from training data, prompting, and decoding strategies into its near-SoTA performancecited by 5
  36. 2024Unsupervised End-to-End Task-Oriented Dialogue with LLMs: The Power of the Noisy Channelcited by 5
  37. 2024CogniDual Framework: Self-Training Large Language Models within a Dual-System Theoretical Framework for Improving Cognitive Taskscited by 5
  38. 2024Had enough of experts? Quantitative knowledge retrieval from large language modelscited by 4

50 most cited of 98 All 98 (JSON)