Watermarking Diffusion Language Models with Learned Foresight and Balanced Bands
arXiv:2610.32205v1 Announce Type: new Abstract: Green-red watermarks favor tokens from a green list that a secret key derives from the preceding tokens. Diffusion language models, however, often choose a token before its preceding tokens,...
Read more: https://arxiv.org/abs/2610.32205
Scan with your phone
Scan with your phone