Abstract
Large pre-trained generative models are known to occasionally output undesirable samples, which undermines their trustworthiness. The common way to mi......
小提示：本篇文献需要登录阅读全文，点击跳转登录
Full Text Link
