Generative AI editing other models' weights: tiny neuron edits do improve performance, but modestly
zakkohane · x · 2026-11-11
Nathan Park (Harvard DBMI) tested whether generative AI models can edit another model's weights to improve performance, on a growth disorder detection task. A tiny number of neuron edits did improve performance (Qwen outperformed Gemma), though the effect was small and results may vary. A rare empirical validation for a previously skeptical claim.
More from Models
- Indie dev gnukeith teases free self-trained LLM landing mid-January with a companion alignment model — gnukeith · 2026-11-11
- Yuanbao-Reranker: 0.6B/4B/8B Qwen3-based rerankers trained on synthetic preferences, open-sourced — _reachsumit · 2026-11-11
- Indie trainer gnukeith bets on sub-9B models: a 2B model can likely do 75% of your tasks — gnukeith · 2026-11-11
- Leaked chart claims Anthropic's internal Model 4 hit 178.76 AECI — haider1 · 2026-11-11
- Dev trains small CoreML decision model, beats his baselines by a wide margin — cyrilzakka · 2026-11-11
- Fable 5.2 audits an Opus 5.5-built codebase and finds issues Opus missed — CtrlAltDwayne · 2026-11-11