Measuring Non-Adversarial Reproduction of Training Data in Large Language Models

Open in new window