1 paper
Benjamin D. Kim, Wanrong Zhang, Weitong Ruan +2
We introduce a framework for multi-bit text watermarking with security defined directly through f-divergence from the base language model distribution. Unlike prior approaches th…