Goodfire's Nick Cammarata argues InceptionV1 polysemanticity wasn't caused by superposition
petrusenko_max · x · 2026-10-16
Nick Cammarata of interpretability startup Goodfire posted a contrarian argument on HN: polysemanticity in early model organisms like InceptionV1 may not have been caused by superposition, but by something more ordinary. If borne out, this would undercut part of the premise underlying the superposition-based interpretability research program. Discussion at the HN link.
More from Research
- AstraZeneca strikes $40 million AI drug discovery deal with insitro — HealthcareLdr · 2026-10-17
- Fizgig LoRA Isolation: 0/50 face bleed-through vs 48/50, likeness kept — shootthesound · 2026-10-17
- NVIDIA's open midtraining recipe lifts SWE-bench Multilingual from 55.8 to 62.8 with 500B tokens — dair_ai · 2026-10-17
- Audit of 160,860 citations: nearly 40% of links shown by Chinese AI search go unused — Sanbi_Ai · 2026-10-17
- Dogs show the Bouba–Kiki effect, study finds — abenitezburraco · 2026-10-17
- rasbt's Reasoning From Scratch Finale: Distilling Reasoning Models Hands-On — rasbt · 2026-10-17