Extracting memorized pieces of (copyrighted) books from open-weight language models

Open in new window