Neuron Explorer

Qwen3-0.6B

'of' after a noun phrase, as in 'of course'

A neuron in layer 16 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 responds to the word 'of' when it follows a noun and sets up a phrase like 'of course', 'each of which', or 'one of each'. Turning it up or down did not produce a clear theme in the model's writing.

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

    at AATR to help us through cycling in the wind either on solo or small group rides.⏎Plan your route!⏎If you’re heading out for a ride of reasonable length, check the weather forecast and wind direction for the duration of the ride

  2. #2 peak

    has a population of 4,008,994. Popular nicknames for Philadelphia are Philly and The City of Brotherly Love, the latter of which comes

  3. #3 peak

    across disparate operating systems.⏎Kubernetes clusters are actually a set of nodes categorized into master nodes and worker nodes. Every cluster needs to have at least one of each. While the master node is meant to control the state of a cluster,

  4. #4 peak

    ating a gas balloon similar to the ones used by Lowe, and allowing visitors to interact with Lincoln, Lowe and Mary Henry themselves (portrayed by actors, of course). There will also be presentations about ballooning and espionage during the war,

  5. #5 peak

    of materials and weight and requires something that is much harder to come by: a propellant. You can make your own gunpowder, of course, but refining the process to create effective munitions is as tricky as building

  6. #6 peak

    9 determined that enslaved people in the state would be emancipated on July 4, 1827, but there were addenda and provisos, of course, so not everyone was instantly manumitted on the magic date, and

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: of of of, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and, and,

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

    why it is so significant that the angels described Jesus’ birth as “peace on earth.”⏎9 And an angel of the Lord appeared to them, and the glory of the Lord shone around them, and they were filled with great fear.

  2. #2 peak

    3,4 The voice of the LORD is upon the waters; The God of glory thunders, The LORD is over many waters. The voice of the LORD

  3. #3 peak

    content make, and they explicitly disclaim, any express or implied representations or warranties of any kind, including, without limitation, representations and warranties as to the functionality of the translation feature or the accuracy or completeness of the translations.⏎Translations are not retained

  4. #4 peak

    irling wheels⏎Four strange whirling wheels are associated with the cherubim.⏎Then I looked, and behold, in the expanse that was over the heads of the cherubim something like a sapphire stone, in appearance in the likeness

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. L3:315 excites 'of' after a noun phrase
  2. L4:4 suppresses
  3. L1:725 suppresses "to" in "referred to"
  4. L6:357 suppresses
  5. L9:2264 excites 'of' after superlatives and 'the X of'
  6. L15:785 suppresses 'of' in set phrases like 'of course'
  7. L5:363 excites
  8. L4:1686 excites
  9. L0:40 suppresses 'of' in 'sort of' hedges
  10. L0:2726 excites
  11. L0:1220 suppresses
  12. L2:895 suppresses
  13. L0:117 suppresses
  14. L6:13 excites
  15. L2:422 weakly suppresses
  16. L3:1379 weakly excites
  17. L8:16 weakly suppresses
  18. L1:2567 weakly excites
  19. L2:1795 weakly suppresses
  20. L2:202 weakly excites "of" after "many" or "which"
  21. L6:703 weakly suppresses
  22. L0:897 weakly suppresses
  23. L7:122 weakly excites
  24. L1:316 weakly suppresses

Feeds later layers

  1. L17:317 strongly excites 'of' after a noun phrase
  2. L17:1835 strongly excites
  3. L18:2808 excites
  4. L19:715 excites
  5. L17:1686 excites
  6. L22:229 excites "of" after a noun in "X of Y"
  7. L17:1177 excites
  8. L20:283 excites
  9. L17:2297 excites
  10. L19:2893 suppresses
  11. L25:231 suppresses
  12. L18:293 excites
  13. L17:963 excites
  14. L19:2138 weakly excites "of" in "matter of" phrases
  15. L19:1992 weakly excites
  16. L17:1507 weakly excites
  17. L20:61 weakly suppresses
  18. L17:1246 weakly excites
  19. L20:64 weakly suppresses
  20. L21:1287 weakly suppresses
  21. L22:496 weakly excites
  22. L18:2244 weakly excites "of" inside titles and names
  23. L27:146 weakly excites
  24. L20:1972 weakly excites 'of' in 'of course'

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.

932 values between -1.27 and -0.602787 values between -0.60 and 0.0846 values between 0.08 and 0.7516 values between 0.75 and 1.4310 values between 1.43 and 2.119 values between 2.11 and 2.784 values between 2.78 and 3.4612 values between 3.46 and 4.1315 values between 4.13 and 4.8111 values between 4.81 and 5.4910 values between 5.49 and 6.165 values between 6.16 and 6.848 values between 6.84 and 7.518 values between 7.51 and 8.191 values between 8.19 and 8.873 values between 8.87 and 9.546 values between 9.54 and 10.223 values between 10.22 and 10.893 values between 10.89 and 11.572 values between 11.57 and 12.254 values between 12.92 and 13.601 values between 13.60 and 14.272 values between 14.27 and 14.951 values between 14.95 and 15.631 values between 15.63 and 16.301 values between 16.30 and 16.981 values between 18.33 and 19.011 values between 19.01 and 19.681 values between 19.68 and 20.361 values between 21.03 and 21.711 values between 22.39 and 23.061 values between 23.74 and 24.411 values between 25.09 and 25.77offnegativepositiveoften

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. sorts
  2. screen
  3. 正しい
  4. screens
  5. contr
  6. urn
  7. fact
  8. _BOOL
  9. liche
  10. course

Suppresses

  1. стоя
  2. voke
  3. "]];⏎
  4. zsche
  5. ��
  6. ventory
  7. etest
  8. 谈起
  9. bies
  10. ons