Neuron Explorer

Qwen3-0.6B

'quite' after a negation or 'never'

A neuron in layer 0 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up on the word 'quite' when it follows words like 'never', 'could not' or 'can't', as in 'he never quite succeeds'. Turning it up did not produce a clear change in the model's replies.

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 activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

  1. #1 peak

    an antique shop, struggle to befriend prominent Japanese individuals as well as advance in a society now ordered by Japanese sensibilities. Despite his best efforts, he never quite succeeds. He is always on the outside looking in.⏎Another plot line follows those

  2. #2 peak

    restaurants and retail stores destroyed. The retail corridor, later renamed Cecil B. Moore Avenue after the NAACP leader who at the time cautioned for peace, would never quite recover from the riots.⏎↑ Independence Hall⏎It makes sense that the very building

  3. #3 peak

    form of robotics and remote-controlled machinery.⏎“Sometimes you can have combinations of things, autonomy most of the time and remote control for situations that the machine can’t quite do things,” he said.⏎For the rest of this article, click here:

  4. #4 peak

    and who would not forget the sight. As Ketchum writes, “The Americans’ revolution survived—survived in some mysterious way that no one could quite f

  5. #5 peak

    recognize the goodness of American foreign policy and the inherent democratic character of the United States. Bush increasingly appears somewhat like a frustrated Lyndon Baines Johnson who could not quite understand why “all” Americans did not recognize their prosperity during the democratic upheav

  6. #6 peak

    not quite true. Even though they represent one family of plants, there are some distinctive features

Steering

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.

turned down a lot
Prompt

Here is something I have been thinking about lately: I have a 3D and 3D is a 3D is a 3D, I have a 3D and 3D is a 3D, I have a 3D and 3D is

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

  1. #1 peak

    the questions what? or whom? Remember a pronoun replaces a noun, and in this case the noun is a direct object. Verbs that have a direct object are called transitive verbs.⏎Direct object pronouns in Spanish agree in number and

  2. #2 peak

    870, at the family farm, Bovenplaats, near Malmesbury, in the Cape Colony. His parents, Jacob<|endoftext|>The main purpose of the script is to compare maximum land surface temperature (LST) of a

  3. #3 peak

    of strict austerity or merely as a time of forgoing a few simple pleasures, Lent may seem like an empty, meaningless ritual in such cases. On the other hand, some grow up in church traditions where Lent is not observed at all. These

  4. #4 peak

    . S. I. Hayakawa, the semanticist and, later, senator from California, called slang “the poetry of everyday life.” On the other hand, Ambrose Bierce defined slang as “the grunt of the human hog,

Connections

Show on map

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Fed by earlier layers

None: this is the first layer.

Feeds later layers

  1. L3:84 suppresses
  2. L25:3065 weakly excites
  3. L5:53 weakly excites
  4. L14:203 weakly excites
  5. L9:1172 weakly excites
  6. L5:2466 weakly excites
  7. L19:1126 weakly excites
  8. L7:1374 weakly suppresses
  9. L23:714 weakly excites
  10. L9:634 weakly excites
  11. L5:1272 weakly excites
  12. L11:4 weakly suppresses
  13. L23:1868 weakly excites
  14. L24:384 weakly excites
  15. L1:2146 weakly excites
  16. L20:1529 weakly excites
  17. L16:997 weakly suppresses
  18. L16:1294 weakly excites
  19. L19:173 weakly suppresses
  20. L10:470 weakly suppresses
  21. L24:1687 weakly excites
  22. L27:1481 weakly excites
  23. L12:1322 weakly suppresses
  24. L24:1851 weakly suppresses

Activation histogram

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.

2 values between -0.61 and -0.563 values between -0.50 and -0.448 values between -0.44 and -0.398 values between -0.39 and -0.3317 values between -0.33 and -0.2723 values between -0.27 and -0.2235 values between -0.22 and -0.16100 values between -0.16 and -0.11337 values between -0.11 and -0.051879 values between -0.05 and 0.011100 values between 0.01 and 0.06240 values between 0.06 and 0.1269 values between 0.12 and 0.1838 values between 0.18 and 0.2316 values between 0.23 and 0.2910 values between 0.29 and 0.356 values between 0.35 and 0.407 values between 0.40 and 0.462 values between 0.46 and 0.525 values between 0.52 and 0.571 values between 0.57 and 0.631 values between 1.25 and 1.301 values between 1.58 and 1.64offnegativepositiveoften

Logit lens

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 layers instead.

Promotes

  1. 领域的
  2. andel
  3. .ComponentModel
  4. 备
  5. own
  6. ビュー
  7. oki
  8. CAS
  9. efined
  10. ータ

Suppresses

  1. hoe
  2. irut
  3. theses
  4. ilio
  5. ive
  6. ivo
  7. 如果有
  8. ison
  9. asto
  10. ruptcy