Neuron Explorer

Qwen3-0.6B

'and' joining parallel phrases

A neuron in layer 13 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 fires on the word 'and' when it links two matching parts of a sentence, like two verbs or two nouns in an official or instructional text. When it was turned up, the model fell into repeating 'and' and similar filler words.

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

    ston. For more information about this artist and her work, visit Inuit Gallery of Vancouver.<|endoftext|>1. The problem statement, all variables and given/known data If an athlete jumped 2 feet high and left the ground

  2. #2 peak

    . Matuga⏎Measuring Housing Affordability⏎U.S. Department of Housing and Urban Development⏎The views expressed in this article are those of the author and do not represent the official positions or policies of

  3. #3 peak

    , to retain the full citation information.⏎- This article was written and prepared by U.S. Government employees on official time, and is therefore in the

  4. #4 peak

    ⏎PortraitDyan Mazurana is senior research fellow at the Feinstein International Famine Center, Tufts University. Angela Raven-Roberts is director of research and programs at the Feinstein Famine Center, Tufts University. Jane Parpart is

  5. #5 peak

    of speech: nouns/pronouns, verbs, adjectives, adverbs, conjunctions, and prepositions. Each section:⏎- Defines the part of speech and its usage⏎- Gives examples to illustrate the definition⏎- Provides a range of

  6. #6 peak

    -Cover Texts, and no Back-Cover Texts. A copy of the license is included in the page “GNU Free Documentation License”.⏎The copyright and license notices on this page only apply to

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 to of the people in the of the of the of of of of of of of of of... I OF... THIS, THE, THE, THE, THE, THE, THE, THE, THE, THE, ... THE,

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

    our institutions, particularly those dedicated to learning? Hitherto clear-cut boundaries to educational institutions - those that defined the very nature of schooling - are being challenged and some are becoming more permeable. The massive uptake of mobile technologies has challenged the physical

  2. #2 peak

    paths (-<|endoftext|>Simply begin typing or use the editing tools above to add to this article.⏎Once you are finished and

  3. #3 peak

    iseáil go slán. Ní dhéanann muid sonraí cártaí creidmheasa a stóráil agus níl aon rochtain againn ar shonraí a

  4. #4 peak

    factors compromise responses to climate changes – both ‘normal’ variation and shocks as well as new variability and increased uncertainty.⏎Third, local support systems are changing, and so is people’s ability to make use of them. The report shows that changes in

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. L9:92 strongly suppresses "and" or "or" joining two items
  2. L10:74 strongly excites
  3. L4:57 strongly suppresses
  4. L12:51 excites
  5. L1:1250 suppresses
  6. L5:880 excites
  7. L9:2010 suppresses
  8. L5:548 excites
  9. L10:1643 suppresses
  10. L4:108 excites "or" offering a choice in web text
  11. L4:2270 excites
  12. L8:70 excites 'and' between stacked superlatives
  13. L11:154 excites commas after citation numbers
  14. L0:313 excites "and" joining paired nouns
  15. L6:13 excites
  16. L0:114 suppresses
  17. L12:2732 weakly excites
  18. L1:1029 weakly excites
  19. L6:7 weakly suppresses
  20. L0:2146 weakly suppresses
  21. L10:554 weakly excites
  22. L8:13 weakly suppresses
  23. L8:7 weakly excites
  24. L5:75 weakly excites

Feeds later layers

  1. L16:376 strongly excites 'or' joining alternatives
  2. L17:572 strongly excites
  3. L18:870 strongly excites
  4. L17:2332 strongly excites 'and' joining two clauses
  5. L20:885 strongly excites 'and' joining paired words
  6. L19:2080 excites ampersand joining two items
  7. L22:198 excites
  8. L25:1694 excites 'and' joining paired names or fields
  9. L17:1856 excites
  10. L17:1537 excites
  11. L16:1669 excites
  12. L15:2765 excites
  13. L14:1116 excites 'and' inside lists
  14. L18:868 excites 'and' joining two nouns in a pair
  15. L18:2345 excites
  16. L15:727 excites
  17. L26:1007 excites
  18. L19:291 excites 'and' joining two full clauses
  19. L17:3061 excites
  20. L16:2049 excites
  21. L16:2577 suppresses
  22. L19:1754 excites
  23. L16:2442 excites
  24. L14:801 excites 'and' or 'or' joining list items

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.

273 values between -1.03 and -0.632408 values between -0.63 and -0.231080 values between -0.23 and 0.1619 values between 0.16 and 0.567 values between 0.56 and 0.964 values between 0.96 and 1.363 values between 1.36 and 1.752 values between 1.75 and 2.152 values between 2.55 and 2.953 values between 2.95 and 3.341 values between 3.34 and 3.743 values between 3.74 and 4.142 values between 4.14 and 4.543 values between 4.54 and 4.935 values between 4.93 and 5.334 values between 5.33 and 5.735 values between 5.73 and 6.135 values between 6.13 and 6.525 values between 6.52 and 6.9211 values between 6.92 and 7.323 values between 7.32 and 7.714 values between 7.71 and 8.117 values between 8.11 and 8.517 values between 8.51 and 8.916 values between 8.91 and 9.302 values between 9.30 and 9.707 values between 9.70 and 10.105 values between 10.10 and 10.504 values between 10.50 and 10.894 values between 10.89 and 11.293 values between 11.29 and 11.692 values between 11.69 and 12.093 values between 12.09 and 12.482 values between 12.88 and 13.282 values between 13.68 and 14.072 values between 14.47 and 14.87offnegativepositiveoften

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. /or
  2. /OR
  3. vice
  4. hence
  5. rogen
  6. thus
  7. rog
  8. rei
  9. thus
  10. 其它问题

Suppresses

  1. 理解和
  2. 和支持
  3. :“
  4. 企业和
  5. ary
  6. 能力和
  7. 生产和
  8. оп
  9. 质量和
  10. ylvania