Edit Distance Robust Watermarks for Language Models

The dataset used in the paper is a language model output, which is a sequence of tokens generated by a language model.

BibTex: