Nick Cammarata: LayerNorm, not superposition, caused early polysemanticity
nickcammarata · x · 2026-10-16
Responding to critique from livgorton, Goodfire researcher Nick Cammarata clarifies his claim that prior interpretability work wasn't all wrong—rather, the field (including himself) misdiagnosed why early neuron/circuit study broke. The cause was ordinary architectural issues like LayerNorm, not an ontology problem: without changing model size, interpretability can go from 0% on a 5B model to 99%.
More from Research
- MathEconLib open-sourced: Axiom Math and top economists bring machine-checked proofs to economics in Lean 4 — soumitrashukla9 · 2026-10-23
- Ken Ono invites the community to build MathEcon on the new Lean 4 economics library — soumitrashukla9 · 2026-10-23
- IFM open-sources RL360: 8,192 concurrent sandboxes and ~1.3M rollouts for agentic RL — rupspace · 2026-10-23
- Science lifts expression of concern on 2024 AI debunking-conspiracies paper; conclusions unchanged — DG_Rand · 2026-10-23
- Embodied First Exam benchmarks coding agents on 1,090 tasks across 44 robots — ManlingLi_ · 2026-10-23
- Anima Anandkumar on neural operators: AI weather models beat supercomputers — AnimaAnandkumar · 2026-10-23