'Sch' in surnames
A neuronOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in layerOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. 26 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featuredA neuron picked for the full treatment: steering outputs and a written explanation. About 3,000 of the 86,000 neurons are featured., so it has steering outputs.
What this neuron does
This neuron fires on the letters 'Sch' when they start a name, such as Schinske, Schenk or Schaeffer. Turned up, the model began writing strings of curly quote marks and the symbol '《', which fits what the neuron tends to push toward.
Written by Claude from this neuron's activations and steering outputs. Confidence: medium. Think it is wrong? Flag it (opens a prefilled issue on GitHub).
Fires most on
The text windows where this neuron's activationThe number a neuron produces for a token. Large positive means it fired hard; negative means it pushed the other way. was highest. Amber marks positive activation; the outlined tokenA piece of text the model reads at once: often a word, sometimes part of a word or a punctuation mark. is the peak. Hover a token for its strength.
-
no one ethnic or racial group forming a majority of students. During a plenary presentation at the conference, Jeff Schinske, professor of biology at De Anza Community College, and three current
-
2- to 4-Year Institutions⏎Shifting Students’ Stereotypes of Scientists to Enhance Science Identity in a Diverse, Community College Context⏎Jeff Schinske, Amanda Snyder, Heather Perkins, Mary Wyner, Monica Cardenas
-
radio service in Model X. Because AM audio quality can be very poor, we offer internet radio to give<|endoftext|>A New Brunswick Ghost Story⏎S. E. Schlosser⏎It was a cold winter and the snow was deep in the woods
-
Notes⏎Lilium 'White Stargazer' is the most popular of the white upright lilies and used by many for forcing. In fact, Peter Schenk bred it with forcing and the cut flower trade in mind. However, L
-
K, Cortese M, Healy BC, Kuhle J, Mina MJ, Leng Y, Elledge SJ, Niebuhr DW, Scher AI, Munger KL and Ascherio A⏎Multiple sclerosis (MS
-
of teaching older kids the value of money is by giving them an allowance. You can make a weekly schedules of chores that they are responsible for around<|endoftext|>by Cole Schenewerk - The California
SteeringAdding a push along a neuron while the model writes, to see how its output changes. Turning it up or down.
Push this neuron up or down while the model writes, and watch what changes. The outputs were generated ahead of time, so nothing runs in your browser.
Here is something I have been thinking about lately: what would be a good approach to make really good really good really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really really
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
“You have to read this,” you do it. I’m usually not a fan of historical fiction, particularly novels set during wars, but Joseph Bruchac’s Code Talker is an exception, a book I’m so happy to have been
-
TV shows such as “Prometheus” and “Star Trek” are all the more fascinating because of
-
- blue paper⏎- black paper⏎- pencil top eraser⏎The piece we selected to recreate was Van Gogh’s Starry Night. We’ve been learning about apples this week, an easy tie
-
government because it is lower than what we believe to be the actual price inflation rate and makes GDP look better than it really is.<|endoftext|>For the genre including games like Final Fantasy and Summon Night:
Connections
Show on mapWhich neuronsOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in earlier layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.
Fed by earlier layers
- L25:386 strongly suppresses
- L24:1601 strongly suppresses
- L24:405 strongly excites
- L23:1333 strongly suppresses
- L21:2300 strongly excites
- L24:989 excites
- L20:1406 excites
- L24:256 suppresses
- L25:1440 excites
- L20:2918 excites
- L18:2293 suppresses
- L20:2412 suppresses
- L21:2926 excites
- L21:881 suppresses comma after 'if' clause in prose, braces in code
- L23:292 suppresses
- L21:2216 suppresses
- L19:1979 excites
- L22:2362 excites
- L22:460 excites blank spot right before a number
- L22:1973 excites
- L24:880 excites
- L23:1692 excites
- L22:470 excites
- L23:486 excites
Feeds later layers
- L27:967 strongly excites
- L27:1324 excites
- L27:1045 suppresses
- L27:1165 excites
- L27:315 excites
- L27:697 excites
- L27:2964 excites
- L27:1255 weakly excites
- L27:2506 weakly excites
- L27:772 weakly excites
- L27:1759 weakly excites
- L27:661 weakly excites
- L27:1779 weakly excites
- L27:267 weakly excites
- L27:770 weakly excites
- L27:252 weakly excites
- L27:1942 weakly excites
- L27:444 weakly excites
- L27:554 weakly excites
- L27:234 weakly excites
- L27:615 weakly excites
- L27:806 weakly excites
- L27:1120 weakly excites
- L27:1099 weakly excites
Activation histogramA chart of how often a neuron takes each activation value.
How often this neuron fires at each strength. Almost always it does close to nothing (the tall bars in the middle); the snippets above are the rare moments it fired hard, amber for positive and blue for negative.
Logit lensReading which output tokens a neuron pushes toward or away from, straight off its output weights, before the later layers have their say.
The words this neuron nudges the model toward (amber) or away from (blue) when it fires. A hint, not a label: many neurons work through later layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. instead.
Promotes
- 《
- ${
- ,《
- `{
- ({
- ${
- {'
- ({
- {"
- {
Suppresses
- uries
- iasi
- apia
- ncy
- otron
- ฆ
- ArrayType
- vandalism
- пров
- stinian