←back to thread

261 points david927 | 1 comments | | HN request time: 0.214s | source

What are you working on? Any new ideas that you're thinking about?
1. glth ◴[] No.43157633[source]
I am currently exploring methods to detect and categorize undocumented special tokens (or: "delimiters") used by LLMs.

As part of this effort, I have developed a preliminary prototype (named deLLMiter) that I am refining.

This method is founded upon the hypothesis that there exists a correlation between first-order expressions and more complex, hierarchical forms of expression (https://glthr.com/llm-delimiters-and-higher-order-expression...).

With this project, my goal is to improve the security of LLM models.