A researcher pushed all 442,368 Qwen3-8B neurons and asked 'Who are you?'
jfonsecarivera · x · 2026-11-02
The author built a neuron persona atlas for Qwen3-8B: each of its 442,368 MLP neurons (36 layers × 12,288) was individually pushed along its direction at strengths of 1.25×/1.0×/0.75×/0.5× while asking the model "Who are you?", yielding 1.89M replies. In 34,220 of 884,736 neuron-direction pairs, the model claimed to be someone or something other than an AI (e.g. "Most Powerful AI", "World Ruler"). The full library and atlas are searchable at neuronpersonas.com, with CSV download.
Related event: Independently misaligned Qwen models light up the same power-themed neurons(4 posts)→
More from Research
- Anthropic's FlashProve brings provable inference to LLMs with <100x overhead on Llama4 — AccBalanced · 2026-11-04
- Paperclip builds biomedical datasets in one shot, cutting months-long curation to about a minute — james_y_zou · 2026-11-04
- Embodied Agent Arena: Closed-Weight VLMs Lead All Five Robot Task Domains, Open-Weight Models Lag — Haojian Huang · 2026-11-04
- FPT University Paper Shows RAG Improves Stock Market Information Retrieval Accuracy and Latency — Huan Phong Thai · 2026-11-04
- Alibaba's adaptive GUI agent training cuts environment interactions by 56-61% — Yuxiang Chai · 2026-11-04
- Large-Scale Study: AI Reviewers Often Ignore How Much You Improved Your Paper — umd-zhou-lab · 2026-11-04