Labeling all 1.1M MLP neurons in Qwen 27B cost just $5 with chain lens readouts
maxsloef · x · 2026-11-12
Inspired by @nickcammarata's "Neural Networks are Already Clean" and work by @treetowntree and @a9lim, @voooooogel applied chain lens readouts to neurons—and it worked: many readouts roundtrip and describe neuron steering behavior well, at a cost of only $5 for all 1.1M MLP neurons in Qwen 27B. @maxsloef proposed extending the labeling to Qwen3 8B to investigate a few oddly behaving neurons.
Related event: All 1.1M MLP Neurons of Qwen 27B Labeled for Just $5(2 posts)→
More from Research
- Talent vs. luck model reproduced: equal funding is ~9x more efficient than funding winners — ChenhaoTan · 2026-11-12
- Stanford HAI dashboards show employment in AI-exposed jobs down over 4% this year — StanfordHAI · 2026-11-12
- Shun-ichi Amari on how 'do as you like' freedom birthed the math behind AI's 60 years — FrnkNlsn · 2026-11-12
- Formal verification proves a 3,000-step Sokoban optimum, with a bounty on beating it — banteg · 2026-11-12
- Probing Qwen3-8B: one shared place code for touches on your body, a woman's, or a doll's — repligate · 2026-11-12
- Paper applies statistical mechanics to 16 literary works, finding word frequencies follow Rayleigh-Jeans thermalization — burny_tech · 2026-11-12