1
Why do LLMs freak out over the seahorse emoji?
Investigating the seahorse emoji doom loop using logitlens.
Theia Vogel
Blog about linguistics, programming, and my projects
- Popular posts
- 8
- Total points
- 2,748
- HN comments
- 1,125
- Rank
- #535 personal
Visit vgel.me → On Hacker News Oct 2015 – Oct 2025
Software Engineering 6AI & Machine Learning 5Tools & Productivity 1Society, Policy & Culture 1Security & Privacy 1Math & Algorithms 1
2
Writing a C compiler in 500 lines of Python
Blog about linguistics, programming, and my projects
3
I made a transformer to predict a simple sequence manually
To better understand how transformers work, I hand-assigned all the weights to predict a simple sequence.
4
Representation Engineering: Mistral-7B on Acid
Playing around with the Representation Engineering paper, I made some interesting control vectors, and a Python package to make your own.
5
Signed distance functions in 46 lines of Python
A walkthrough of 46 lines of code that render a 3D ASCII donut using signed distance functions.
6
How to make LLMs go fast
Blog about linguistics, programming, and my projects
7
GPT-3 will ignore tools when it disagrees with them
Augmenting GPT-3 with tools is a popular technique right now. But is GPT-3 really using them?
8