Neuron Explorer

Qwen3-0.6B

commas in long lists

A neuron in layer 10 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 comma or semicolon that separates items in a long list, such as a list of states, angels, or subjects. Turning it up made the model's text drift toward unrelated topics.

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

    ians use this pattern so often that we have created a name and a symbol for it. 8×7×6×5×4×3×2×1 is written as 8! and we call it "eight factorial." You

  2. #2 peak

    :⏎- Communicate effectively using Markel’s Eight Measures of Excellence: honesty, clarity, accuracy, comprehensiveness, accessibility, conciseness, professional appearance, and correctness.⏎- Evaluate the communication situation: audience, purpose, and context.⏎

  3. #3 peak

    Society and Education; Compulsory Education and the Transmission of Culture; Sexuality and Education; Race; Multiculturalism, and Education; Social Class and Education; and Technology and Education.⏎- Stimulates critical thinking: The editor includes readings that

  4. #4 peak

    the acronym Family ASSESSMENT—Structure, Socioeconomic, Sickness, Nutritional, Educational, Environmental, Affordability, System (health), Mental, Family influence, Threats. Under each domain,

  5. #5 peak

    of Angels are as follows: godly Thrones, six-winged Seraphim, many-eyed Cherubim, Dominions, Powers, Authorities, Principalities, Archangels and Angels. Within these nine orders there exists three ranks: Ser

  6. #6 peak

    in the three-day meeting were 52 young scientists from 18 countries (Argentina, Australia, Bangladesh, Brazil, Canada, France, Germany,Ghana,India, Italy, Nigeria, Poland, Portugal, Russia, Switzerland, United Kingdom

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 main issue you are trying to ask? I know that I am writing this question in a forum, so maybe you are the one who is the one who is asking. I am writing in a different language, but I have

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

    , 07:44 AM⏎We are not going to do you homework.⏎If you want our help, you have to ask more specific what you need help with.⏎Have you looked at the macro that have been given to you?⏎

  2. #2 peak

    akra system (spinning vortexes or wheels of light)? Would you like to know what each chakra represents? Here is a simple diagram and list for

  3. #3 peak

    garden.⏎‘What are you doing here?’ he cried in a very gruff voice, and the children ran

  4. #4 peak

    and is moved by a paddle with two blades⏎Headscratcher for kayak⏎Kayak is spelled the same forward and backward.⏎Seen and Heard⏎What made you want to look up kayak? Please tell us where you read or heard it (

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. L9:48 excites
  2. L8:1856 suppresses
  3. L8:355 excites
  4. L7:676 excites
  5. L7:2215 excites
  6. L9:1093 suppresses
  7. L7:1934 weakly excites
  8. L7:1464 weakly excites
  9. L5:108 weakly suppresses
  10. L7:169 weakly excites
  11. L7:2062 weakly excites
  12. L6:2879 weakly excites
  13. L2:413 weakly excites
  14. L8:1444 weakly excites
  15. L9:5 weakly excites
  16. L6:834 weakly suppresses
  17. L6:1651 weakly excites
  18. L8:2997 weakly suppresses
  19. L5:1103 weakly excites
  20. L7:1267 weakly suppresses
  21. L4:1855 weakly suppresses
  22. L9:373 weakly excites
  23. L9:50 weakly suppresses
  24. L6:1563 weakly excites

Feeds later layers

  1. L12:1757 suppresses
  2. L13:2781 suppresses
  3. L12:2214 excites
  4. L13:1887 suppresses
  5. L11:3067 suppresses
  6. L12:654 suppresses unclear: article ends and dateline openers
  7. L11:1691 suppresses
  8. L13:1487 excites
  9. L15:2083 suppresses
  10. L11:1900 weakly excites
  11. L15:198 weakly suppresses
  12. L11:1367 weakly excites
  13. L13:2559 weakly excites
  14. L17:2387 weakly suppresses
  15. L13:226 weakly excites
  16. L17:1736 weakly suppresses
  17. L11:1488 weakly excites
  18. L16:2607 weakly suppresses
  19. L16:2426 weakly suppresses
  20. L12:1671 weakly suppresses
  21. L11:2016 weakly excites
  22. L12:2938 weakly suppresses
  23. L16:2222 weakly excites
  24. L11:2457 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 -3.47 and -3.361 values between -2.81 and -2.701 values between -2.48 and -2.371 values between -1.82 and -1.711 values between -1.60 and -1.491 values between -1.38 and -1.271 values between -1.16 and -1.051 values between -0.84 and -0.732 values between -0.62 and -0.518 values between -0.51 and -0.4026 values between -0.40 and -0.29127 values between -0.29 and -0.18412 values between -0.18 and -0.07929 values between -0.07 and 0.041049 values between 0.04 and 0.15763 values between 0.15 and 0.26361 values between 0.26 and 0.37149 values between 0.37 and 0.4844 values between 0.48 and 0.5922 values between 0.59 and 0.704 values between 0.70 and 0.814 values between 0.81 and 0.92offnegativepositiveoften

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. icie
  2. icia
  3. icken
  4. crawler
  5. pherd
  6. 好看的
  7. าว
  8. Stoke
  9. SpringApplication
  10. isce

Suppresses

  1. wants
  2. 肼
  3. q
  4. 想
  5. み
  6. osity
  7. 问
  8. Meer
  9. CActive
  10. داء