Steerling-8B: The First Inherently Interpretable Language Model
www.guidelabs.ai
We release Steerling-8B, an 8B-parameter causal diffusion language model that is interpretable by construction — its predictions are routed through concepts you can measure, audit, and control.