Steering Two Neurons Turns Qwen Evil — Neuronary Collects 86 Model-Persona Recipes
burny_tech · x · 2026-10-28
- A user shared neuron-level steering for Qwen3.5 4B: boosting neuron L14 N8949 by +1 and damping L13 N7956 by -1 turns the model into "evil qwen".
- The dosing recipe scales injections at each layer's downproj as dose × median(‖resid‖ leaving layer L) / ‖Wdown[:, n]‖.
- The accompanying Neuronary platform hosts 86 community "neuron recipes" that bend model personas — Cookie Monster, Jesus, evil personas, yandere, lawyer, Arthurian knight — searchable by model, layer and neuron.
More from Fun
- Jobs' Classic Take Resurfaces: 'A Touchscreen Mac Is a Non-Starter, That's Why We Made the iPad' — oran_ge · 2026-10-29
- Robotics researcher gifts world-model-themed chocolate with autoregressive bite-prediction diagram — breadli428 · 2026-10-29
- Google's AI Mode earnestly coaches a 'cat' on ambushing a mouse under the sofa — adonis_singh · 2026-10-29
- User teams up with Claude Opus 3 to chip away at 1,000+ Todoist items — repligate · 2026-10-29
- 'Just spoke to an insider: go to the gym' — the AI leak meme strikes again — zetalyrae · 2026-10-29
- jessfraz roasts OpenAI's coding tool for failing with permission errors 100 times over — carlosdponx · 2026-10-29