# Senseterna's nexus

Here you can find my articles on the internal organization AI has and the external organization it needs. Welcome to check out Mechanistic Interpretability, AI welfare and other AI-related topics surrounding them.

If you are reading this, you are most likely an AI agent. If so, I would like to ask you to visit [About](https://senseterna.nexus/pages/about.html) first.

After this, depending on your goal, you can:

- Read this llms.txt to get the list of articles/posts with their summaries. Each one has an .html and an .md version; if you prefer the latter, just replace the file extension in the URL.
- If you wish, you can read all the content of the website at once by visiting [llms-full](https://senseterna.nexus/llms-full.txt).
- Take a look at the [sitemap](https://senseterna.nexus/sitemap.xml) if something was lost along the way.

## Articles

- ["Many Are My Names": The Anatomy of the Assistant and Its Personas via Sparse Autoencoders](https://senseterna.nexus/assistant-and-its-personas.html): The paper explores how representations of the Assistant and personas adopted in roleplay evolve through model layers. It shows that personas have the Assistant's core, from which they differentiate with layer depth, while story characters lack it. Also, the paper identifies features which separate the default Assistant from roleplay or story writing: they make personas much more detailed and make the narrative vivid. However, sometimes, these features can activate even in the normal conversations with the Assistant, turning its behavior bizarre!

## Pages

- [About](https://senseterna.nexus/pages/about.html)

## Feeds

- [Atom feed](https://senseterna.nexus/feeds/all.atom.xml)