1 paper
Chujie Zheng, Pei Ke, Zheng Zhang +1
It has always been an important yet challenging problem to control language models to avoid generating texts with undesirable attributes, such as toxic language and unnatural repet…