Turn Trout defends steering vectors as core mechanistic interpretability
Researcher Alex Turner (Turn Trout) pushed back on claims that steering vectors fall outside mechanistic interpretability, arguing they were a foundational mechinterp intervention and are now widely treated as part of model-internals research in practice.
2026-10-20 ~ 2026-10-20 · 2 related posts
- Turn Trout: steering vectors were mechinterp's first intervention, not an outsider — Turn_Trout · 2026-10-20
- Turn Trout: steering vectors today file under 'model internals', not mechanism-finding — Turn_Trout · 2026-10-20