'to' after a verb or noun
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. 5 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 lights up on the small word 'to' when it follows a verb or noun, as in 'put to use', 'right to recall' or 'went to live'. Turning it up made the model's writing drift into loose, repetitive or off-topic text, with no clear theme.
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.
-
the 19th cent. when Pasteur noted the antagonistic effect of other bacteria on the anthrax organism and pointed out that this action might be put to therapeutic use. Meanwhile the German chemist Paul Ehrlich developed the idea
-
59; died at Madrid, 27 March, 1625. He was a great-grandson of the Tordesillas who was put to death by the Comuneros at Seville. He studied in Spain and Italy
-
istic reforms and translated the Bible into his native tongue.⏎Many Poles turned from the Roman Church, winning a measure of religious freedom. Some church buildings were put to Protestant use. Laws forbade the execution of Roman Church decrees
-
implemented the law of giving a right to recall to the employers of certain industries. This law has been enacted until the end
-
of the narrative “My first mistress was a woman of… To “l had resided but a short time in Baltimore” to “Very soon after I went to live with,” Douglass shows to his readers a clear time pattern, his whereabouts
-
like Pope Francis who at the World Day for Youth in Rio de Janeiro, Brazil, called for the "respect and protection of the Creation which God has confided to humankind," As educators, the Sisters read, research, study and teach about
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: I will say, for example, that the value of the expression is 2000. I will say, for example, that the value of the expression is 5. I will say, for example, that the value of
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
maintained a normal weight. (The reason for this difference remains a mystery.)⏎Cowley's team then removed the hypothalamus from the animals and analysed the response of cells from this brain region to leptin. When exposed to the hormone, the
-
Japanese public opinion and Japan–South Korea relations.⏎An invaluable resource for scholars of East Asian regional security issues.⏎Table of Contents⏎- Bringing People’s Voices Back in Japan–South Korea Relations⏎- South Korea-Japan FCR Crisis and Public
-
This was mainly because of the Friar.⏎The Capulets and Montagues were not on good terms with each other. Unfortunately though, Romeo and Juliet fell in love with each other regardless if they are from opposing families. Because of these families
-
men heard the shells edging closer. “And then bang!—one of the big shells struck the ground right in front of us and we all went up in the air. But we all come down again.”⏎York was showered with honors
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
- L0:2060 excites
- L4:2373 excites
- L0:1956 weakly suppresses
- L4:561 weakly excites
- L2:3058 weakly suppresses
- L4:1568 weakly suppresses
- L1:2801 weakly excites
- L4:1532 weakly suppresses
- L4:2649 weakly excites
- L4:1658 weakly excites
- L0:1027 weakly excites
- L4:1880 weakly suppresses
- L4:2122 weakly excites
- L2:1060 weakly excites
- L3:589 weakly suppresses
- L0:1217 weakly excites
- L3:799 weakly suppresses
- L4:1990 weakly excites
- L4:2146 weakly excites
- L1:750 weakly suppresses "of" after "kind" or "sort"
- L2:977 weakly excites
- L2:2423 weakly suppresses
- L3:2878 weakly suppresses
- L2:1157 weakly suppresses
Feeds later layers
- L6:338 suppresses
- L8:1627 excites
- L9:723 excites
- L8:45 excites
- L16:1168 excites 'to' after verbs of change
- L12:1851 excites hyphen inside dates
- L17:2821 excites
- L19:1345 excites
- L17:2527 weakly excites
- L19:858 weakly excites
- L10:132 weakly excites
- L10:2213 weakly excites 'to' or 'into' showing movement between two things
- L18:1001 weakly excites
- L7:14 weakly excites
- L10:297 weakly excites
- L13:2521 weakly suppresses
- L18:1857 weakly excites
- L6:1004 weakly excites
- L26:524 weakly suppresses
- L15:675 weakly suppresses
- L16:623 weakly excites
- L19:1963 weakly excites
- L6:368 weakly suppresses
- L10:272 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
- ppers
- pper
- gether
- 了他的
- iration
- .githubusercontent
- Brow
- olini
- gal
- eing
Suppresses
- пед
- -validate
- upe
- хват
- nels
- фик
- 要做
- 万美元
- pipe
- 扉