生成式嵌入反演攻击揭示句向量的信息泄露（英文版）

英文标题：Sentence Embedding Leaks More Information than You Expect: Generative Embedding Inversion Attack to Recover the Whole Sentence中文摘要：通过生成对抗网络逆向语言模型的句向量表示，揭示其可能存在的信息泄露，并提出一种更为高效的逆向分析方式。英文摘要：Sentence-level representations are beneficial for various natural languageprocessing tasks. It is commonly believed that vector representations cancapture rich linguistic properties. Currently, large language models (LMs)achieve state-of-the-art performance on sentence embedding. However, somerecent works