Persistent Homology Automates Detection of Circular Feature Manifolds in LLM Activations
Siddharth Setlur proposes using persistent homology combined with sparse autoencoders to automatically detect multi-dimensional feature manifolds, such as circular structures, in LLM activation spaces, improving on prior manual discovery methods.
2026-11-10 ~ 2026-11-10 · 2 related posts
- Persistent Homology Automates Detection of Circular Feature Manifolds in GPT-2 — burny_tech · 2026-11-10
- Using Persistent Homology to Automatically Find Multidimensional Features in LLMs — burny_tech · 2026-11-10