Neuron Explorer

Qwen3-0.6B

quantifier 'all' mid-phrase

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 fires on the word 'all' when it comes after a verb or preposition, as in 'reuse any or all of this', 'stop all swarming' or 'clean all surfaces'. It goes quiet or flips negative on 'All' inside names and words. Turning it up made the text drift into lists and repeated numbers, which does not clearly reflect the trigger.

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

    . Formulation of regional guidelines may help reduce AOM burden.⏎- acute otitis media⏎Statistics from Altmetric.com⏎If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance

  2. #2 peak

    swarm it is a natural process where the colonies divide. In essence, it is nature’s way of trying to increase bee colony populations. Beekeepers can not stop all swarming. However, swarm prevention should be incorporated info a beekeeper’s

  3. #3 peak

    save lives. Good communication also reduces patient anxiety, enhances patient well-being, and reduces frustration and confusion. Our focus, then, should be to improve communication across all of these vital touch points:⏎1. Between physician and patient. When our patients

  4. #4 peak

    total of ten words, all consisting of nine letters each, broken down into three equal portions, and scrambled. Each portion is used only once. Try to reconstruct all of the portions and create all ten of the nine letter words!⏎TEL ONC

  5. #5 peak

    different types of anxiety disorders. These include generalized anxiety disorder, social anxiety disorder, and separation anxiety disorder. It is possible to have more than one, or possibly all three. No matter what type(s) of anxiety you are experiencing.⏎Examples of

  6. #6 peak

    Regents made the calamitous decision to eliminate affirmative action for its then nine universities, one of the biggest public university systems in America. California’s electorate voted to end all affirmative action programs a year later.⏎Since then, Asian American efforts for equity have

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 I am a student in a school, and I have a certain amount of time from the morning, like if you have an important thing to do in the morning, like what the whole day is in the day from the morning, I

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

    East Java; a green fashion show in Korea; and a project aimed at developing smart grid systems in Vietnam.⏎Young leaders can make a difference⏎Siti Nur Alliah, a British Council Climate Champion from Indonesia, is tackling a burning issue in

  2. #2 peak

    ik (1898), 32; Lee, The Microtomists Vade Mecum (London, 1900); Macallum, On the Detection and Localization of Phosphorus in Animal and Vegetable Cells

  3. #3 peak

    flowers, fruit and meat from devotees to appease Siva and his consort Durga, accompanied by the rhythmic ringing of bells.⏎Young girls in saffron and white satin wraparounds, with head-dresses of coconut palm fr

  4. #4 peak

    and hypothesized that the RNA of TMV is single-stranded. This conjecture was proven to be correct after her death and is now know to<|endoftext|>Sarah Allard and Kristin Jenkins⏎Mentor: Dr. Neal Williams⏎E

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. L1:66 excites
  2. L1:161 suppresses
  3. L16:119 excites
  4. L2:400 excites
  5. L4:308 excites
  6. L17:615 excites
  7. L15:82 suppresses
  8. L0:1578 excites
  9. L2:1803 excites
  10. L15:78 excites
  11. L16:96 excites
  12. L2:355 suppresses
  13. L17:1572 excites
  14. L15:114 excites
  15. L15:96 suppresses
  16. L15:80 suppresses
  17. L3:2187 suppresses
  18. L15:115 excites
  19. L17:517 weakly suppresses "but" in "not only X, but" phrases
  20. L15:89 weakly excites
  21. L16:14 weakly excites
  22. L10:2268 weakly suppresses
  23. L15:403 weakly excites period inside an abbreviation
  24. L16:95 weakly excites

Feeds later layers

  1. L19:574 strongly suppresses
  2. L21:2717 strongly suppresses
  3. L24:1354 weakly suppresses
  4. L20:1531 weakly excites
  5. L19:2293 weakly excites
  6. L20:800 weakly excites
  7. L19:2889 weakly excites
  8. L22:2274 weakly excites
  9. L20:3050 weakly suppresses
  10. L19:1788 weakly suppresses
  11. L19:233 weakly excites
  12. L24:1867 weakly excites
  13. L25:827 weakly excites
  14. L21:2472 weakly excites
  15. L21:1102 weakly excites 'one or more' before a count noun
  16. L25:572 weakly excites
  17. L23:2014 weakly excites
  18. L26:2116 weakly excites
  19. L26:1083 weakly excites
  20. L19:489 weakly excites
  21. L21:1840 weakly excites
  22. L23:2015 weakly suppresses
  23. L21:1306 weakly suppresses
  24. L23:927 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.

2 values between -22.42 and -21.591 values between -14.95 and -14.121 values between -14.12 and -13.291 values between -11.63 and -10.801 values between -10.80 and -9.971 values between -9.97 and -9.142 values between -9.14 and -8.312 values between -7.48 and -6.654 values between -6.65 and -5.822 values between -5.82 and -4.994 values between -4.99 and -4.163 values between -4.16 and -3.343 values between -3.34 and -2.5112 values between -2.51 and -1.6823 values between -1.68 and -0.85438 values between -0.85 and -0.023230 values between -0.02 and 0.82175 values between 0.82 and 1.652 values between 1.65 and 2.481 values between 9.94 and 10.77offnegativepositiveoften

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. ivity
  2. six
  3. Six
  4. seven
  5. Seven
  6. ten
  7. Space
  8. five
  9. thấy
  10. ộc

Suppresses

  1. ctor
  2. alic
  3. 士
  4. lli
  5. aption
  6. 您的孩子
  7. 书香
  8. usion
  9. 慕容
  10. gger