Neuron Explorer

Qwen3-0.6B

'mil' or 'bil' inside names

A neuron in layer 25 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 the short syllable 'Mil' or 'Bil' when it starts or sits inside a name, like Milosevic, Milner, Milford or Tbilisi. Turning it up did not produce any clear change in the text that followed.

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

    if Western history isn’t littered with cravens and rapscallions.⏎As John Gallagher notes in Romania After Ceausescu, leaders such as Milosevic and Ceausescu used nationalism in ways that “denied individual

  2. #2 peak

    celebrates its 58th anniversary this year (Nature, 2009). In 2015, Stephen Hawking and Russian billionaire Yuri Milner announced the start of a well-funded program, ‘Breakthrough Initiatives’,

  3. #3 peak

    script: Gender and Policing in Peace Operations Chapter 15 The Guatemalan Peace Accords: Critical Reflections Chapter 16 Les Femmes Aux Milles Bras: Building Peace in Rwanda Chapter

  4. #4 peak

    ized according to diseases and regions, scientifically studied and provided with possibilities for use in modern medicine.⏎The museum is a training-scientific base for students of Tbilisi State Medical University and other higher education institutions, an information and resource center for

  5. #5 peak

    boat is a monument to human tenacity — or desperation. So is Joal. Fifty years ago, it was a village of 6,546 millet farmers and five pirogues. Then a terrible drought wrung the

  6. #6 peak

    explored the country in the 19th century. He ended a wandering life (he had earlier fought under Garibaldi in Italy) when he settled at Milford Sound in

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: what is the number of elements on the set on the number on the number on the number on the number on the number on the number on the number on the number on the number on the number on the number on the number on the number

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

    protect the wheat land from Bedouins’ raids coming from the marginal lands to the east. This location is on the green line, the eastern border of the Moab Plateau.

  2. #2 peak

    دانشکده معماری و شهرسازی، دانشگاه علم و صنعت ایران، تهران، ایران

  3. #3 peak

    .⏎Metsäpohjainen piha paransi päiväkotilasten immuunijärjestelmää jo kuukaudessa<|endoftext|>Research Looks at Alternative

  4. #4 peak

    دانشکده معماری و شهرسازی، دانشگاه علم و صنعت ایران، تهران، ایران.⏎عنوان مقاله [English]⏎In the process of architecture

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. L22:2199 suppresses
  2. L24:1002 excites
  3. L20:253 excites
  4. L23:1334 weakly suppresses
  5. L21:121 weakly suppresses
  6. L22:1328 weakly suppresses
  7. L23:3029 weakly suppresses
  8. L1:711 weakly excites
  9. L3:301 weakly suppresses
  10. L4:796 weakly suppresses
  11. L3:822 weakly excites
  12. L10:1921 weakly suppresses
  13. L7:2948 weakly excites
  14. L11:2992 weakly suppresses
  15. L23:506 weakly suppresses
  16. L14:97 weakly suppresses
  17. L10:763 weakly suppresses
  18. L7:1586 weakly excites
  19. L9:1059 weakly suppresses
  20. L7:1967 weakly suppresses
  21. L20:2685 weakly excites
  22. L19:1944 weakly excites
  23. L22:870 weakly excites
  24. L17:376 weakly suppresses

Feeds later layers

  1. L27:1027 weakly suppresses
  2. L27:813 weakly excites
  3. L26:1427 weakly suppresses
  4. L26:1143 weakly excites
  5. L26:760 weakly suppresses
  6. L26:2594 weakly excites
  7. L27:578 weakly excites
  8. L26:2350 weakly suppresses
  9. L26:676 weakly excites
  10. L27:1241 weakly suppresses
  11. L27:984 weakly excites
  12. L27:2614 weakly suppresses
  13. L27:2382 weakly excites
  14. L26:481 weakly suppresses
  15. L27:2475 weakly suppresses
  16. L27:2101 weakly suppresses
  17. L26:713 weakly excites
  18. L26:1965 weakly suppresses
  19. L26:2371 weakly excites
  20. L27:299 weakly suppresses
  21. L27:2607 weakly excites
  22. L27:2585 weakly suppresses
  23. L27:70 weakly excites
  24. L27:729 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 -16.13 and -15.411 values between -15.41 and -14.701 values between -14.70 and -13.991 values between -11.86 and -11.151 values between -6.88 and -6.172 values between -4.74 and -4.031 values between -4.03 and -3.322 values between -3.32 and -2.616 values between -2.61 and -1.907 values between -1.90 and -1.19143 values between -1.19 and -0.483209 values between -0.48 and 0.24499 values between 0.24 and 0.9520 values between 0.95 and 1.667 values between 1.66 and 2.372 values between 2.37 and 3.081 values between 3.79 and 4.501 values between 4.50 and 5.221 values between 5.22 and 5.931 values between 8.77 and 9.481 values between 11.62 and 12.33offnegativepositiveoften

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. 给她
  2. 把她
  3. 她的
  4. 因为她
  5. 她
  6. 对她
  7. 让她
  8. herself
  9. 但她

Suppresses

  1. riers
  2. aviolet
  3. u
  4. uars
  5. chrom
  6. igans
  7. .u
  8. uced
  9. perial
  10. orian