Neuron Explorer

Qwen3-0.6B

'and' joining two nouns in a pair

A neuron in layer 18 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 'and' when it links two people, places, or ideas in a pair, such as 'X and Y'. Turning it up made the text wander into loose, repetitive phrasing.

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

    amendments granting these regions formal recognition and autonomy.⏎In 1948 Belgium became a co-signatory to the Benelux Customs Union along with the Netherlands and Luxembourg. This later became the Benelux Economic Union in 195

  2. #2 peak

    opper showed early interest in art, particularly drawing, and went on to study illustration and painting. With their emphasis on truthful, contemporary subjects, his teachers Robert Henri and Kenneth Hayes Miller at the New York School of Art were vital to Hopper's

  3. #3 peak

    Southern states-Virginia, North Carolina, Arkansas, and Tennessee-joined the Confederacy once the war started. Secessionist governments were established in Missouri and Kentucky, two border states that officially remained in the Union, while the western counties

  4. #4 peak

    -Nicod within France's National Center for Scientific Research (CNRS) and a Global Distinguished Professor at New York University.⏎The combined team of linguists and primatologists analyzed alarm calls of Campbell's monkeys on two sites: the Tai

  5. #5 peak

    and can be used to add visual interest to an outfit. When choosing patterns, it is important to consider their scale and how they will work with other elements of your outfit. You can also mix and

  6. #6 peak

    an archaeologist with a strong background in anthropology and museums, passionate about outreach and education. Her research focuses on death and burial,

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: what I want to do with this as a a final, only final effort to build it up, and then then then then then and then and then and then and then and then and then and then and then and then and then and then

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

    that planets orbit at least 5 to 10 percent of all stars. But how many of these planets are Earth-size, and possibly Earth-like?<|endoftext|>& Tornado Alley⏎Jan Mikusinski developed an operational calculus which is relevant for

  2. #2 peak

    . It is a free software. We also strongly recommend that you sign up for an AdobeID at the Adobe website. For more details please see FAQ 1&2. To view this ebook

  3. #3 peak

    . It is a free software. We also strongly recommend that you sign up for an AdobeID at the Adobe website. For more details please see FAQ 1&2. To view this ebook on an iPhone, iPad or Android mobile device you

  4. #4 peak

    across the United States. Others are limited to specific regions of the country. Thankfully, however, most ticks are fairly easy to identify.⏎Genus & Species⏎& Conditions Caused⏎|American Dog Tick⏎||Dark brown with creamy-gray markings

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

  1. L13:131 excites 'and' joining parallel phrases
  2. L4:57 suppresses
  3. L9:92 suppresses "and" or "or" joining two items
  4. L15:114 excites
  5. L1:1250 suppresses
  6. L0:2776 suppresses
  7. L15:82 suppresses
  8. L3:447 excites
  9. L5:548 excites
  10. L15:115 excites
  11. L10:74 excites
  12. L15:78 weakly excites
  13. L16:177 weakly suppresses zero digit inside dates and codes
  14. L17:1950 weakly suppresses
  15. L16:135 weakly suppresses
  16. L16:208 weakly suppresses
  17. L16:503 weakly excites
  18. L16:376 weakly suppresses 'or' joining alternatives
  19. L15:89 weakly excites
  20. L15:80 weakly suppresses
  21. L15:122 weakly suppresses
  22. L16:282 weakly excites
  23. L8:7 weakly excites
  24. L16:34 weakly excites

Feeds later layers

  1. L22:134 suppresses
  2. L20:885 suppresses 'and' joining paired words
  3. L20:2504 excites
  4. L25:1694 suppresses 'and' joining paired names or fields
  5. L19:2080 suppresses ampersand joining two items
  6. L23:270 excites
  7. L20:1536 weakly suppresses the word 'well' in phrases
  8. L22:2880 weakly suppresses
  9. L19:2565 weakly excites
  10. L20:2371 weakly excites
  11. L22:198 weakly suppresses
  12. L23:821 weakly excites
  13. L19:334 weakly excites
  14. L21:1246 weakly excites
  15. L19:860 weakly excites
  16. L25:2143 weakly suppresses
  17. L21:921 weakly excites
  18. L20:533 weakly excites
  19. L27:2695 weakly suppresses
  20. L20:3043 weakly suppresses
  21. L21:1765 weakly suppresses
  22. L21:357 weakly excites
  23. L21:1546 weakly suppresses
  24. L23:2057 weakly suppresses ampersand inside names and abbreviations

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.

1 values between -34.91 and -33.991 values between -28.46 and -27.541 values between -22.93 and -22.011 values between -21.09 and -20.171 values between -20.17 and -19.252 values between -17.41 and -16.483 values between -16.48 and -15.561 values between -15.56 and -14.641 values between -14.64 and -13.723 values between -13.72 and -12.801 values between -12.80 and -11.881 values between -11.88 and -10.963 values between -10.96 and -10.041 values between -10.04 and -9.121 values between -9.12 and -8.197 values between -8.19 and -7.279 values between -7.27 and -6.358 values between -6.35 and -5.4314 values between -5.43 and -4.517 values between -4.51 and -3.5911 values between -3.59 and -2.6714 values between -2.67 and -1.7510 values between -1.75 and -0.831219 values between -0.83 and 0.102507 values between 0.10 and 1.0280 values between 1.02 and 1.94offnegativepositiveoften

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. че
  2. 客户提供
  3. 철
  4. ading
  5. ราม
  6. 遷
  7. ない
  8. 一层
  9. 出入
  10. 买单

Suppresses

  1. amp
  2. lt
  3. emsp
  4. raquo
  5. nbsp
  6. quot
  7. nds
  8. gt
  9. uum
  10. vice