Grafted artificial neurons lift a 135M model's multiplication accuracy from 21% to 83%
CatAstro_Piyush · x · 2026-11-08
cybermynd shares a "brain surgery" experiment on a tiny 135M-parameter model: by grafting artificial neurons directly into the network, they changed how it solves problems, lifting accuracy on unseen multiplication problems from 21% to 83% — with no training at all. It's a striking example of directly editing a model's internal computation instead of fine-tuning it.
More from Research
- Satire: Robotics Paper Works Backwards From a Viral Video Idea — Sentdex · 2026-11-08
- Meta's ABTest Agent: LLM Plans, Deterministic Code Computes Every Number — rohanpaul_ai · 2026-11-08
- MatAnyone-E cuts 4K video matting cost 13x, runs 28fps on an RTX 3090 — bdsqlsz · 2026-11-08
- Consciousness researchers argue in Nature that AI on silicon hardware won't become conscious — The Decoder · 2026-11-08
- DeepMind research lead says scientific discovery pace is hitting a real phase transition, results coming in weeks — Scobleizer · 2026-11-08
- MLIP Mena large atomistic model runs molecular simulation on a laptop — CatAstro_Piyush · 2026-11-08