A Novel Semantic Manifold Alignment Attack against Embedding-to-Embedding Obfuscation in Privacy-Preserving LLMs
本文针对隐私保护大语言模型（LLM）推理中的嵌入到嵌入混淆（Embedding-to-Embedding Obfuscation, E2EO）机制提出了一种新的攻击方法——代理流形对齐（Proxy Manifold Alignment, P
#embedding-attack
共收录 1 条相关情报。
← 返回所有主题