'one' in 'for one reason' and 'one another'
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. 19 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
The neuron responds to the word 'one' when it means 'a particular one' or 'one another', as in 'for one reason' or 'one of the greatest'. Its strongest negative reactions are on 'Children' in names like 'St. Jude Children's Research Hospital', which is a different kind of word. Turning it down made the model repeat the fragment "'s" over and over.
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.
-
less comfortable expressing themselves via words. Art Therapy is also known to help people with social skills, specifically, those that are withdrawn or shy, or who, for one reason or another, have a difficult time functioning within social situations.⏎Why art?
-
whole poem, but we're just getting into it now, thirty-nine lines in. Lowell illustrates how hard, and how steadfastly, the soldiers fought for one another, and for the Republic.<|endoftext|>By the end of the session, participants
-
8, the father confirmed that Yanela and her mother stayed together with the Washington Post (Selk, 2018). They were not separated from one another like hundreds of other families who crossed the border between April and June 2
-
914). As one historian characterized this trans-Atlantic movement, it attempted “to limit the socially destructive
-
molecule with a trick like this. And that’s part of the problem: it’s so obvious that it seems like it probably wouldn’t work at all. For one thing, how do you know if your
-
reef fish abundance and distribution trends between locations. Armed with the knowledge of where<|endoftext|>Edited by William Byron Forbush This is a book that will never die — one of the great English classics. . .
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: 's' 8's' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s' s
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
able alternative therapies for this common disease.⏎In today’s session, Drs. Kenneth Ataga of the University of North Carolina and Russell Ware of St. Jude Children’s Research Hospital will review existing and innovative therapies for SCD.⏎Dr. At
-
prescribe medications to take during an episode of vertigo, which will lessen the severity of an attack. If Meniere’s disease is causing balance issues, vestibular rehabilitation therapy can help.⏎Low Blood
-
meeting mad characters is easy but finding your way home all the while keeping your sanity is less so.⏎Carroll, Lewis. Alice in Wonderland. Macmillan Children's Books, 2015. 208 pp., ISBN
-
, might not be immediately obvious.⏎“It is not a part of our culture to think of heat as risky,” said Bernstein, a pediatrician at Boston Children’s Hospital and co-director of the Center for Climate, Health, and the
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
- L17:2369 excites
- L8:1541 excites
- L11:2179 excites
- L18:800 excites
- L18:1095 excites
- L18:569 excites
- L14:280 suppresses
- L18:1028 weakly excites
- L5:1823 weakly suppresses
- L18:305 weakly suppresses
- L8:79 weakly suppresses
- L6:2338 weakly excites
- L17:529 weakly suppresses
- L14:169 weakly excites
- L18:1103 weakly suppresses
- L3:59 weakly excites
- L16:951 weakly excites
- L16:1087 weakly suppresses
- L16:527 weakly excites
- L17:1535 weakly excites
- L15:202 weakly excites mixed, digits in URLs and comparisons
- L8:16 weakly suppresses
- L16:1683 weakly suppresses
- L12:336 weakly excites
Feeds later layers
- L22:397 strongly suppresses
- L26:765 strongly suppresses
- L22:576 strongly suppresses
- L25:2331 strongly suppresses
- L27:905 strongly suppresses
- L24:1505 strongly suppresses
- L21:1071 strongly suppresses
- L27:1064 strongly suppresses
- L20:1340 suppresses
- L20:1050 suppresses
- L22:1163 suppresses
- L25:1566 suppresses
- L27:2084 suppresses
- L25:127 excites
- L20:2278 suppresses
- L27:404 suppresses
- L20:519 suppresses
- L26:2244 suppresses
- L24:286 suppresses
- L20:1388 suppresses
- L27:355 suppresses
- L22:2151 excites
- L23:309 weakly suppresses
- L27:942 weakly suppresses
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
- ("-",
- 日讯
- _-
- ("{}
- {};
- ugu
- osa
- (()
- {};⏎
- /*#__
Suppresses
- 's
- ’s
- ´s
- ’S
- �s
- 的工作
- 'S
- 的手
- 的大
- 的城市