Neuron Explorer

Qwen3-0.6B

narrator and speaker words

A neuron in layer 9 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up on words that name the one telling a story or giving an opinion, such as "narrator" or "speaker", and on a few words like "sense" and "wonders". Turning it up made the model's replies drift toward everyday personal topics, though that link is weak.

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

    hand (the broken wall), and reminding us about the mysterious Something that destroys this wall.⏎- The Something is very different from the hunters and their dogs (the speaker knows all about them). The Something is far more covert. In fact, it

  2. #2 peak

    century. In fact, the only common thread was a mutual childhood memory of playing on a wreck of varying descriptions – depending on the age and recall power of the narrator.⏎In spite of the odds against finding the vessel’s identity, the public imagination

  3. #3 peak

    finally, that level of self-communion in which the husband will seek to explain his wife’s behavior and their crumbling marriage to himself.⏎Last, there is the narrator’s ambiguous silence, looking on dispassionately, allowing things as well as

  4. #4 peak

    as “…kingdom of dead…” inferno 8, line 85, wondering how this stranger has managed to penetrate their territory without being killed. The narrator refers to his leader as a master and mentor all through the play meaning his role

  5. #5 peak

    ⏎Mary Oliver, a renowned American poet, has captivated readers with her profound exploration of nature and the human experience. Her poems often evoke a sense of wonder and contemplation, inviting readers to reflect on their own lives. In

  6. #6 peak

    ches of British-red.⏎One inquiring reader wants to know who controls the world's non-independent islands.⏎Detra K., from Los Gatos, California, wonders about the

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: the number of times as many as the number of times as many as the number of times as many as the number of times as many as the number of times as many as the number of times as many as the number of times as many

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

    ://dictionary.reference.com/browse/daemon Pronounces it DEE-mon Which is how I've always said it. (The Golden Compass audiob<|endoftext|>50 Reasons to Eat & Grow Local Food⏎Preparing and preserving local food preserves culinary culture. For

  2. #2 peak

    for your middle school-aged kids. Teach them that you expect them to work hard. In return, you will reward them with plenty of downtime to do things they enjoy,<|endoftext|>

  3. #3 peak

    where we practised massage, yoga and relaxation.⏎We teamed up with P7 for an afternoon in Seaton Park where we played rounders and football. Thank you to our parent helpers who joined us and took part in the games.⏎Soup and

  4. #4 peak

    the world’s leading scientific expert on gratitude. a professor of psychology and founding editor-in-chief of The Journal of Positive Psychology, has discovered when people regularly engage in gratitude, they

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. L6:2344 weakly suppresses
  2. L7:2429 weakly suppresses
  3. L4:2746 weakly suppresses
  4. L7:1426 weakly excites
  5. L3:1957 weakly excites
  6. L7:1106 weakly suppresses
  7. L8:1104 weakly excites
  8. L4:2335 weakly excites
  9. L8:425 weakly excites
  10. L3:3014 weakly excites
  11. L3:1630 weakly excites
  12. L6:6 weakly excites
  13. L8:1508 weakly excites
  14. L5:1320 weakly excites
  15. L8:1470 weakly suppresses
  16. L4:556 weakly suppresses
  17. L3:725 weakly excites
  18. L5:290 weakly suppresses
  19. L7:1051 weakly excites
  20. L7:1687 weakly excites
  21. L3:1259 weakly excites
  22. L7:63 weakly excites
  23. L8:4 weakly excites
  24. L1:2266 weakly excites

Feeds later layers

  1. L10:1910 weakly excites
  2. L11:431 weakly excites
  3. L12:2903 weakly excites
  4. L19:422 weakly excites
  5. L17:1337 weakly suppresses
  6. L11:3037 weakly excites
  7. L21:1095 weakly suppresses 'and' or ',' inside metric pairs
  8. L16:214 weakly suppresses
  9. L21:1861 weakly suppresses
  10. L13:2093 weakly excites
  11. L25:2141 weakly suppresses
  12. L18:2477 weakly suppresses
  13. L24:957 weakly suppresses
  14. L25:2595 weakly suppresses
  15. L19:2893 weakly suppresses
  16. L12:508 weakly suppresses
  17. L12:648 weakly suppresses
  18. L11:298 weakly excites
  19. L13:2687 weakly excites
  20. L13:1826 weakly suppresses
  21. L13:2989 weakly excites nouns for who or what a text is about
  22. L14:1748 weakly excites
  23. L11:2821 weakly suppresses
  24. L23:1406 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.

1 values between -2.59 and -2.482 values between -1.22 and -1.112 values between -1.11 and -1.011 values between -1.01 and -0.904 values between -0.90 and -0.798 values between -0.79 and -0.6914 values between -0.69 and -0.5816 values between -0.58 and -0.4831 values between -0.48 and -0.3756 values between -0.37 and -0.27171 values between -0.27 and -0.16532 values between -0.16 and -0.061687 values between -0.06 and 0.05829 values between 0.05 and 0.16345 values between 0.16 and 0.26120 values between 0.26 and 0.3742 values between 0.37 and 0.4719 values between 0.47 and 0.5810 values between 0.58 and 0.685 values between 0.68 and 0.792 values between 0.79 and 0.892 values between 0.89 and 1.002 values between 1.00 and 1.102 values between 1.21 and 1.322 values between 1.32 and 1.423 values between 1.53 and 1.63offnegativepositiveoften

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. وع
  2. chop
  3. lean
  4. Comment
  5. 提议
  6. comment
  7. وص
  8. ETCH
  9. besie
  10. Comment

Suppresses

  1. 转载请
  2. 好吗
  3. став
  4. 么
  5. 全校
  6. Bins
  7. ğer
  8. État
  9. 太多了
  10. oggle