Interpretability Demo: One Neuron Steers Model Outputs Toward Green
An interpretability experiment shows that amplifying a single neuron (layer 30, neuron 2310) systematically steers model outputs toward green topics, biasing 45 of 50 responses.
2026-10-26 ~ 2026-10-26 · 2 related posts
- Steering one neuron makes the model see green: 45/50 replies hijacked — Sauers_ · 2026-10-26
- Interpretability curiosity: a researcher found the neuron that makes the model green — Sauers_ · 2026-10-26