Neuron Explorer

Qwen3-0.6B

"of" after a noun

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 'of' when it follows a noun, as in 'the words of', 'the names of' or 'the cry of'. It is weaker on other small words like 'for', 'against' or 'at'. When turned up, the model fell into repeating short phrases like 'the leader a leader', so the steering does not show a clear story.

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

    all the way which I command you, that it may be well with you.'⏎Matthew 7:24-27 “Therefore everyone who hears these words of Mine and acts on them, may

  2. #2 peak

    establishments and missionary colonies (all under the authority of Columba), until in time the Gospel light was ablaze on the hills of Albyn; and the names of St. Columba and Iona were on every tongue from Rome to the utmost

  3. #3 peak

    11 children. He led an exemplary life, bearing constantly in mind his mother’s words: “I would rather see you dead at my feet than guilty of a mortal sin.” His biographers have written of the long hours he spent in

  4. #4 peak

    et maîtriser les dépenses : propositions de l’Assurance Maladie pour 2014 – Rapport au Ministre chargé de la Sécurité Sociale et au Parlement sur l’évolution des charges

  5. #5 peak

    the simple story of a village life is depicted in an extra-ordinary way in this creation of his. The title of the story is kept by the name of the village only. So the title truly

  6. #6 peak

    case or motion being tried or heard and the laws applicable to it.⏎(2)Either the outline of a case made in preparation for a class, or a written argument for Legal Writing

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. The, I. I. 1. 1. 1 This: I. 1. 1 1 1 3. 3 3 3

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

    family studies⏎Langley, Elizabeth Dale, "Parenting Styles and Perceptions Across Christian Denominations" (2011). ETD Collection for Pace University. AAI3467374.⏎Remote User:

  2. #2 peak

    familiar with it, U.S. law defines human trafficking as “the use of force, fraud, or coercion to compel a person into commercial sex acts or labor against their will,” and it brings in an estimated $150 billion a year

  3. #3 peak

    in the town. The population density was 1.2 people per square mile (0.4/km²). There were 106 housing units at an average density of 1.1 per square mile (0.4/k

  4. #4 peak

    wrong<|endoftext|>External Web sites⏎Britannica Web sites⏎Articles from Britannica encyclopedias for elementary and high school students.⏎- Sen

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. L4:4 strongly suppresses
  2. L7:6 suppresses
  3. L10:132 excites
  4. L8:517 suppresses
  5. L5:763 suppresses
  6. L10:469 suppresses 'of' after a noun phrase
  7. L10:709 suppresses
  8. L1:403 excites
  9. L10:105 suppresses
  10. L9:2417 excites
  11. L6:101 suppresses
  12. L5:1053 suppresses
  13. L4:2275 suppresses
  14. L3:232 excites 'on' after a noun like 'impact on'
  15. L6:357 suppresses
  16. L5:5 suppresses digit after a label like paper or type
  17. L4:16 excites
  18. L12:2736 suppresses
  19. L6:655 excites
  20. L12:234 excites
  21. L9:1025 weakly excites
  22. L7:943 weakly suppresses
  23. L8:705 weakly excites
  24. L12:205 weakly suppresses 'of' after an abstract noun

Feeds later layers

  1. L17:385 strongly excites
  2. L17:303 strongly excites
  3. L17:1122 excites
  4. L18:663 excites
  5. L18:293 excites
  6. L17:1507 excites
  7. L20:283 excites
  8. L17:146 suppresses period after abbreviations and dates
  9. L18:1135 excites
  10. L16:2666 excites
  11. L25:286 excites
  12. L14:143 excites 'the' inside journal and institution names
  13. L16:412 excites
  14. L22:489 excites
  15. L15:316 excites
  16. L17:292 excites
  17. L17:175 excites 'to' after a noun or verb of motion
  18. L17:137 excites
  19. L15:381 excites
  20. L18:93 excites
  21. L19:1464 excites
  22. L17:2302 excites 'in' after a noun
  23. L22:160 excites
  24. L17:1835 excites

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 -2.61 and -2.342 values between -1.80 and -1.534 values between -1.53 and -1.269 values between -1.26 and -1.00164 values between -1.00 and -0.73965 values between -0.73 and -0.461079 values between -0.46 and -0.19708 values between -0.19 and 0.08299 values between 0.08 and 0.35160 values between 0.35 and 0.6279 values between 0.62 and 0.8861 values between 0.88 and 1.1535 values between 1.15 and 1.4236 values between 1.42 and 1.6934 values between 1.69 and 1.9628 values between 1.96 and 2.2327 values between 2.23 and 2.5029 values between 2.50 and 2.7630 values between 2.76 and 3.0324 values between 3.03 and 3.3020 values between 3.30 and 3.5720 values between 3.57 and 3.8411 values between 3.84 and 4.119 values between 4.11 and 4.3718 values between 4.37 and 4.648 values between 4.64 and 4.919 values between 4.91 and 5.184 values between 5.18 and 5.456 values between 5.45 and 5.7211 values between 5.72 and 5.994 values between 5.99 and 6.253 values between 6.25 and 6.526 values between 6.52 and 6.792 values between 6.79 and 7.062 values between 7.86 and 8.13offnegativepositiveoften

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. いう
  2. ゜
  3. compat
  4. >>::
  5. _));⏎
  6. })(
  7. mọi
  8. }}>
  9. mut
  10. ationToken

Suppresses

  1. CTION
  2. ibly
  3. PTION
  4. ulent
  5. worth
  6. fähig
  7. -through
  8. RIES
  9. 在整个
  10. 还会