Does Data Contamination Detection Work (Well) for LLMs? A Survey and Evaluation on Detection Assumptions
Open Access
- 1 January 2025
- conference paper
- Published by Association for Computational Linguistics (ACL)
- p. 5235-5256
Abstract
Yujuan Fu, Ozlem Uzuner, Meliha Yetisgen, Fei Xia. Findings of the Association for Computational Linguistics: NAACL 2025. 2025.Keywords
All Related Versions
This publication has 16 references indexed in Scilit:
- Data Contamination Quiz: A Tool to Detect and Estimate Contamination in Large Language ModelsTransactions of the Association for Computational Linguistics, 2025
- Scaling Trends for Data Poisoning in LLMsProceedings of the AAAI Conference on Artificial Intelligence, 2025
- Enhancing Chat Language Models by Scaling High-quality Instructional Conversations2023
- Stop Uploading Test Data in Plain Text: Practical Strategies for Mitigating Data Contamination by Evaluation Benchmarks2023
- Copyright Violations and Large Language Models2023
- Can we trust the evaluation on ChatGPT?2023
- Membership Inference Attacks on Machine Learning: A SurveyACM Computing Surveys, 2022
- Quantifying Privacy Risks of Masked Language Models Using Membership Inference Attacks2022
- Privacy Risks of General-Purpose Language Models2020
- Auditing Data Provenance in Text-Generation Models2019