Study Finds Coherent Body Map Inside LLMs, Improving with Scale
An interpretability study with open-source code finds that LLMs carry a coherent internal body map when discussing touch, with precision improving as model scale increases, suggesting an implicitly learned spatial self-representation.
2026-11-11 ~ 2026-11-11 · 2 related posts
- Study finds LLMs hold coherent internal body maps that scale with model size — Sauers_ · 2026-11-11
- LLMs carry coherent internal body maps, precision scales with model size — Sauers_ · 2026-11-11