Neuron Explorer

Qwen3-0.6B

closing parenthesis after a place or name

A neuron in layer 5 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

The neuron fires on a closing bracket that follows a short name or place, such as '(Minnesota)' or '(French)'. It is mostly quiet elsewhere and turning it up did not produce a clear theme.

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

    holds a flag in both hands as he glances forward. The sculpture is signed "Gutzon Borglum 1929 (illegible) AKUNST FDY NYC". The back of the base is inscribed:

  2. #2 peak

    will be characterized with quantum optical metrology techniques. The second objective is to enhance the emission brightness and entanglement quality through integration<|endoftext|>UC DAVIS (US) — After a live videoconference with a specialist, rural emergency room physicians are more

  3. #3 peak

    (Minnesota) for three years before her marriage to Dr. Will Kennicott, of G

  4. #4 peak

    7 (2013) [Refereed Article]⏎Infectious diseases of wildlife reduce population size and

  5. #5 peak

    7 x 66 mm|⏎|Printer||American Bank Note Company (U.S.A.) – ABNC|⏎|Signatures||Ernesto (Che) Guevara Serna & Rolando Díaz Astaraín|⏎|

  6. #6 peak

    ankofa Day™, an annual national memorial in London’s Trafalgar Square to mark this day, celebrate the great Ayitian (Haitian) Revolution and remember and honour the victims of the Transatlantic Human Trafficking 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: I am a 3D model of a 3D in the shape of a cube in the 3D space. My name is in the shape of a cube. Let me know where I am. I think I can take a

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

    Glass” and their use of geometry as speculative and imaginative.⏎In trying to find more about this term, it appears that sadly the intelligent design crowd has laid claim to it. The most I could find (in an admittedly short search) was this

  2. #2 peak

    the French parliament and supreme court declared that claims to the French throne could not be passed through women. The French Estates chose Philip of

  3. #3 peak

    analysis is trustworthy. If the analysis is bulletproof, the next question is whether the evidence is substantial enough to claim a discovery. And if a discovery can be claimed, the final question is what, exactly, has been discovered? Scientists can have

  4. #4 peak

    the northerners, but two of the invading peoples settled on the coast of Palestine, between Gaza and Mount Carmel. The attempted invasion ended Egyptian pretensions to a Syro-Palestinian hegemony.⏎Two more years of peace ensued,

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. L4:1981 weakly excites
  2. L4:516 weakly suppresses
  3. L2:911 weakly suppresses
  4. L4:955 weakly excites
  5. L3:2787 weakly excites
  6. L3:57 weakly excites
  7. L1:118 weakly excites
  8. L4:2429 weakly excites
  9. L0:2822 weakly suppresses
  10. L2:2319 weakly excites
  11. L2:1937 weakly excites
  12. L4:2463 weakly excites
  13. L1:2117 weakly suppresses
  14. L4:2632 weakly excites
  15. L3:2989 weakly excites
  16. L4:775 weakly excites
  17. L0:2554 weakly suppresses
  18. L3:1949 weakly excites
  19. L2:2904 weakly suppresses
  20. L3:2409 weakly excites
  21. L3:2128 weakly excites
  22. L4:2143 weakly suppresses "in some" opening a qualifier
  23. L3:503 weakly suppresses
  24. L2:2581 weakly excites

Feeds later layers

  1. L9:828 suppresses
  2. L13:3016 excites
  3. L6:536 suppresses
  4. L8:2060 weakly suppresses
  5. L18:2842 weakly excites
  6. L9:2384 weakly excites
  7. L16:1207 weakly excites
  8. L6:1713 weakly suppresses
  9. L23:1713 weakly suppresses
  10. L11:2581 weakly suppresses
  11. L7:2906 weakly excites
  12. L19:441 weakly excites
  13. L22:578 weakly suppresses
  14. L13:1991 weakly suppresses
  15. L7:2211 weakly excites 'their' after work and manage verbs
  16. L18:974 weakly excites
  17. L6:1698 weakly suppresses
  18. L14:2167 weakly excites
  19. L12:821 weakly suppresses
  20. L24:1221 weakly excites
  21. L25:1786 weakly excites
  22. L12:1710 weakly excites
  23. L18:638 weakly excites
  24. L8:115 weakly 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.

3 values between -1.07 and -0.951 values between -0.95 and -0.832 values between -0.83 and -0.713 values between -0.71 and -0.608 values between -0.48 and -0.3655 values between -0.36 and -0.24326 values between -0.24 and -0.121365 values between -0.12 and -0.011577 values between -0.01 and 0.11447 values between 0.11 and 0.2384 values between 0.23 and 0.3517 values between 0.35 and 0.475 values between 0.47 and 0.583 values between 0.58 and 0.703 values between 0.82 and 0.941 values between 0.94 and 1.061 values between 1.06 and 1.171 values between 1.29 and 1.411 values between 1.53 and 1.642 values between 1.76 and 1.881 values between 2.59 and 2.711 values between 3.41 and 3.531 values between 3.53 and 3.65offnegativepositiveoften

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. {"
  2. {'
  3. ges
  4. ixer
  5. ESC
  6. drop
  7. {{
  8. 坏
  9. nomin
  10. 大宗

Suppresses

  1. ford
  2. ข
  3. ью
  4. unge
  5. .idea
  6. -chart
  7. TimeInterval
  8. =post
  9. 也不要
  10. Manning