Fetching the paper…

Adaptive Pre-training Data Detection for Large Language Models via Surprising Tokens · Around