CopyBench: Measuring Literal and Non-Literal Reproduction of Copyright-Protected Text in Language Model Generation

• https://arxiv.org/abs/2407.07087
• Tong Chen, Akari Asai, Niloofar Mireshghallah, Sewon Min, James Grimmelmann, Yejin Choi, Hannaneh Hajishirzi, Luke Zettlemoyer, Pang Wei Koh

Training data leakage and memorization in language models, Copyright