TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

CryptGPT: A Simple Approach to Privacy-Preserving LLMs Using Vigenere Cipher

10 点作者 diwank11 个月前

3 条评论

xrd11 个月前
Great article.<p>Can someone smarter than me explain why an embedding is or isn&#x27;t a poor man&#x27;s form of encryption? You can retrieve the closeness to something using cosine similarity, but you can&#x27;t go back to the original source (unless you store the embedding side by side, which is what most people do with those vectors).<p>But, isn&#x27;t using an embedding model a good way to cloak your original data and still make it comparable?
评论 #40694161 未加载
评论 #40694224 未加载
fsmv11 个月前
But this cypher is easily defeated, it even says how in the Wikipedia article.<p>I suspect it wouldn&#x27;t work with anything that actually prevents people from &quot;decrypting&quot; it.
评论 #40698736 未加载
评论 #40698747 未加载
tripplyons11 个月前
How does this preserve privacy when everything can be decrypted?
评论 #40694954 未加载
评论 #40694937 未加载