News for hackers. Stuff that matters.

A comment on Language Models Are Injective and Hence Invertible.

Re:Language Models Are Injective and Hence Invertible by yosito on Thursday October 30, 2025 @08:13AM (#45759158)
In layman's terms, this seems to mean that given a certain unedited LLM output, plus complete information about the LLM, they can determine what prompt was used to create the output. Except that in practice this works almost never. Am I understanding correctly?
[ Parent | on HN ]
Re:Language Models Are Injective and Hence Invertible by ctenb on Thursday October 30, 2025 @08:20AM (#45759227)
No, it's about the distribution being injective, not a single sampled response. So you need a lot of outputs of the same prompt, and know the LLM, and then you should in theory be able to reconstruct the original prompt.
[ Parent | on HN ]
Re:Language Models Are Injective and Hence Invertible by eapriv on Thursday October 30, 2025 @08:21AM (#45759232)
No, it says nothing about LLM output being invertible.
[ Parent | on HN ]