mlsanityData to 14 Oct 2026

Quantifying Memorization and Detecting Training Data of Pre-trained Language Models using Japanese Newspaper

arXiv 2404.17143 ↗First posted 26 Apr 2024cs.CLcited by 2

Not on the map. Why?

Abstract (arXiv)

Dominant pre-trained language models (PLMs) have demonstrated the potential risk of memorizing and outputting the training data. While this concern has been discussed mainly in English, it is also practically important to focus on domain-specific PLMs. In this study, we pre-trained domain-specific GPT-2 models using a limited corpus of Japanese newspaper articles and evaluated their behavior. Experiments replicated the empirical finding that memorization of PLMs is related to the duplication in the training data, model size, and prompt length, in Japanese the same as in previous English studies. Furthermore, we attempted membership inference attacks, demonstrating that the training data can be detected even in Japanese, which is the same trend as in English. The study warns that domain-specific PLMs, sometimes trained with valuable private data, can ''copy and paste'' on a large scale.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 40 of its 64 references that are in the map. First posted on arXiv 26 Apr 2024.

Builds on

40 of 64 references.

References · 40

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2021On the Opportunities and Risks of Foundation Modelscited by 4,033
  3. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,411
  4. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
  5. 2020Extracting Training Data from Large Language Modelscited by 2,189
  6. 2022Large Language Models Encode Clinical Knowledgecited by 1,830
  7. 2022Red Teaming Language Models with Language Modelscited by 1,026
  8. 2016Deep Variational Information Bottleneckcited by 930
  9. 2018Adafactor: Adaptive Learning Rates with Sublinear Memory Costcited by 905
  10. 2021Membership Inference Attacks From First Principlescited by 878
  11. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 864
  12. 2021Deduplicating Training Data Makes Language Models Bettercited by 714
28 more
  1. 2022GPT-NeoX-20B: An Open-Source Autoregressive Language Modelcited by 713
  2. 2018Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidatescited by 639
  3. 2019Making AI Forget You: Data Deletion in Machine Learningcited by 465
  4. 2021Large Language Models Can Be Strong Differentially Private Learnerscited by 412
  5. 2022Deduplicating Training Data Mitigates Privacy Risks in Language Modelscited by 314
  6. 2023Membership Inference Attacks against Language Models via Neighbourhood Comparisoncited by 258
  7. 2022Data Contamination: From Memorization to Exploitationcited by 236
  8. 2020Auditing Differentially Private Machine Learning: How Private is Private SGD?cited by 218
  9. 2022Are Large Pre-Trained Language Models Leaking Your Personal Information?cited by 200
  10. 2021Adversary Instantiation: Lower Bounds for Differentially Private Machine Learningcited by 199
  11. 2023Stop Uploading Test Data in Plain Text: Practical Strategies for Mitigating Data Contamination by Evaluation Benchmarkscited by 142
  12. 2018The Adverse Effects of Code Duplication in Machine Learning Models of Codecited by 126
  13. 2021How much do language models copy from their training data? Evaluating linguistic novelty in text generation using RAVENcited by 124
  14. 2024Min-K%++: Improved Baseline for Detecting Pre-Training Data from Large Language Modelscited by 103
  15. 2022Preventing Verbatim Memorization in Language Models Gives a False Sense of Privacycited by 100
  16. 2021Does BERT Pretrained on Clinical Notes Reveal Sensitive Data?cited by 93
  17. 2021Large Scale Private Learning via Low-rank Reparametrizationcited by 88
  18. 2024DE-COP: Detecting Copyrighted Content in Language Models Training Datacited by 74
  19. 2022Do Language Models Plagiarize?cited by 73
  20. 2023Bag of Tricks for Training Data Extraction from Language Modelscited by 70
  21. 2021Membership Inference Attack Susceptibility of Clinical Language Modelscited by 52
  22. 2022Exploring the Limits of Differentially Private Deep Learning with Group-wise Clippingcited by 52
  23. 2020Formalizing Data Deletion in the Context of the Right to be Forgottencited by 50
  24. 2023Training Data Extraction From Pre-trained Language Models: A Surveycited by 45
  25. 2023Quantifying Association Capabilities of Large Language Models and Its Implications on Privacy Leakagecited by 28
  26. 2024Sampling-based Pseudo-Likelihood for Membership Inference Attackscited by 20
  27. 2015ReCon: Revealing and Controlling PII Leaks in Mobile Network Trafficcited by 9
  28. 2020KART: Parameterization of Privacy Leakage Scenarios from Pre-trained Language Modelscited by 7

Built on by

Cited by year

Citing papers per year: 2025: 2.

2 papers.

  1. 2025Analyzing Memorization in Large Language Models through the Lens of Model Attributioncited by 14
  2. 2025Fast-MIA: Efficient and Scalable Membership Inference for LLMscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025The Landscape of Memorization in LLMs: Mechanisms, Measurement, and Mitigation
  2. 2025Memorization in Fine-Tuned Large Language Models
  3. 2024Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data
  4. 2024Mitigating Memorization In Language Models
  5. 2023SoK: Memorization in General-Purpose Large Language Models
  6. 2024Undesirable Memorization in Large Language Models: A Survey
  7. 2021How BPE Affects Memorization in Transformers
  8. 2025Hubble: a Model Suite to Advance the Study of LLM Memorization