Neuron Explorer

Qwen3-0.6B

unclear: word fragments like 'de' in 'deities'

A neuron in layer 23 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on small pieces of longer words, most often the piece "de" in "deities". The pattern is weak and mixed, and the text it produced when turned up or down showed no clear theme.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    a drop of the heart's blood of Mother Earth creates fertility according to the Orient region. French jewelers called ruby 'the gem of gems' or the ‘dearly loved stone’. Ruby is spoken of in the book of Job, in

  2. #2 peak

    represent the venue itself in a traditional and respectful way. A theatre building that connects with something as ancient as Chinese acrobatics itself was the outcome.⏎As was the peak of Chinese architecture, during

  3. #3 peak

    about as you read?⏎Mar 07, Minutes Buy. Anthony Lukas Award A landmark work of science, history and reporting on the past, present and imperiled future of the Great Lakes. But they are under threat as never before,

  4. #4 peak

    belief 1n winged minor gods but the concept of the intermediary, wholly good angel was most fully developed in the West. Where there existed a thousand little deities, some kind, some mischievous, but all approachable and bri

  5. #5 peak

    repertory.⏎The Symphony might be regarded as a battle, a study of opposites in conflict: the stately versus the agitated, the singable diatonic melody against the sinewy chromatic theme, the foreboding

  6. #6 peak

    h-only" party, which insisted that no other god but the Lord must be worshiped, and the more cosmopolitan leaders who favored incorporating other national and tribal deities into public worship. In the north, a Yahweh-only coup had

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: if the end-tempered and the end-tempered if the end-tempered if the end-tempered if the end-tempered if the end-tempered if the end-tempered if the end-tempered

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

    New York factory, which was a one-piece brass measuring 33.5 meters (109’) tall. This flagpole is still the tallest in service at An

  2. #2 peak

    gent plate margin between the African Plate and the Eurasian Plate. It is the tallest active volcano in Europe, currently standing 3,329 m (10,922 ft) high, though this varies with summit eruptions

  3. #3 peak

    conflict between the shame and guilt associated with the act of rape. In conclusion, Williams observes that Lucrece may be best understood as Shakespeare's portrayal of the archetypal rape victim. Edward T.

  4. #4 peak

    cm⏎Possibly William Bryant, sale at Edward Foster’s, London, 15th May 1823 as ‘Holbein’⏎Archibald Alexander Leslie Melville, 13th Earl of Leven and

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. L22:151 weakly suppresses
  2. L19:647 weakly suppresses
  3. L13:1492 weakly excites
  4. L20:134 weakly suppresses
  5. L21:774 weakly excites
  6. L4:1924 weakly excites
  7. L15:2578 weakly suppresses
  8. L22:366 weakly excites
  9. L9:1925 weakly suppresses
  10. L7:799 weakly excites
  11. L3:108 weakly suppresses
  12. L8:2650 weakly excites
  13. L0:1727 weakly suppresses
  14. L7:887 weakly excites
  15. L2:492 weakly suppresses
  16. L21:654 weakly suppresses
  17. L22:1859 weakly excites
  18. L18:305 weakly excites
  19. L12:948 weakly suppresses
  20. L17:1703 weakly excites weak, mixed: 'on' and 'the' after publishing verbs
  21. L11:2553 weakly suppresses
  22. L17:3003 weakly suppresses
  23. L14:2829 weakly excites
  24. L14:1083 weakly excites

Feeds later layers

  1. L24:1898 weakly suppresses
  2. L24:2370 weakly suppresses
  3. L24:2645 weakly suppresses
  4. L24:1875 weakly suppresses
  5. L27:2462 weakly suppresses
  6. L24:2045 weakly excites acronym introductions
  7. L25:3042 weakly excites
  8. L26:682 weakly excites
  9. L26:488 weakly suppresses
  10. L24:1531 weakly suppresses
  11. L24:2857 weakly excites
  12. L24:776 weakly suppresses
  13. L25:2502 weakly excites
  14. L26:2550 weakly excites
  15. L27:178 weakly excites
  16. L27:2469 weakly suppresses
  17. L25:726 weakly suppresses
  18. L25:1514 weakly suppresses
  19. L26:2289 weakly excites
  20. L26:841 weakly excites
  21. L25:1886 weakly suppresses
  22. L24:2263 weakly suppresses
  23. L24:1665 weakly excites
  24. L26:2782 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.

1 values between -24.72 and -23.501 values between -14.99 and -13.771 values between -12.56 and -11.341 values between -11.34 and -10.121 values between -8.91 and -7.693 values between -7.69 and -6.472 values between -6.47 and -5.264 values between -5.26 and -4.0413 values between -4.04 and -2.8228 values between -2.82 and -1.61400 values between -1.61 and -0.393312 values between -0.39 and 0.8371 values between 0.83 and 2.0425 values between 2.04 and 3.2614 values between 3.26 and 4.4814 values between 4.48 and 5.696 values between 5.69 and 6.915 values between 6.91 and 8.121 values between 8.12 and 9.341 values between 9.34 and 10.562 values between 10.56 and 11.771 values between 15.42 and 16.641 values between 22.72 and 23.94offnegativepositiveoften

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. bidden
  2. 搏
  3. rous
  4. chner
  5. charges
  6. alar
  7. 新动能
  8. allo
  9. 限り
  10. 依法

Suppresses

  1. Ế
  2. эр
  3. #{@
  4. 份
  5. aturas
  6. (port
  7. 重伤
  8. itudes
  9. emet
  10. _PORT