Neuron Explorer

Qwen3-0.6B

question marks ending a question

A neuron in layer 21 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 lights up on the question mark at the end of a question, especially a question asked in the middle of a sentence or paragraph. When it was turned up, the model's text was not clearly about anything. Turned down, the model fell into long strings of question marks.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    . Armstrong said, but by watching larger trends. “We are not pushing the<|endoftext|>Would you be able to live in a house that was just four feet wide? That’s right – four feet wide! Well, Polish architect Jakub Szcz

  2. #2 peak

    at their heart to do with science. But the journey from evidence to policy is far from simple, always lengthy, frequently divisive and often ineffective. Why is that? To get to the bottom of this question we first need to ask another — what

  3. #3 peak

    Denial Paradox. When a guy denies he's a crook, what goes through most people's minds is, "why does he have to deny it? There must be some⏎truth to the accusation."⏎But there's something deeper going

  4. #4 peak

    , would he please hush and go away, and stay away, if he could, for five minutes? By all means, yes, he would. He’d stay away indefinitely if she

  5. #5 peak

    . But by making sustainability apply to imports, could the vote paradoxically harm farmers in developing countries by imposing a sort of hidden tariff for entry into the Swiss market? Jonathan Hepburn, senior programme manager for agriculture at Geneva-based International Centre for Trade

  6. #6 peak

    nutrition experts.⏎Healthy eating includes making healthful food choices, which means knowing what and how much you eat. Do you know the difference between serving and portion sizes? While the terms serving and portion often are used interchangeably, they actually mean different

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: how to make?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?

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

    ask: “If you were forced to return to your home country by the U.S. government, would you bring your children born in the U.S. with you or would you leave them in this

  2. #2 peak

    Galilee has two famous inventions to his credit. What are they?⏎- Four inventions that were to change the history of the world came from China. What were they?⏎- Which great painter of the Renaissance was also a famous inventor?⏎- This

  3. #3 peak

    the Sinai Experience at some point during the prayer service so as to make sure that we don’t forget it. Yet we seemingly find no such reminder … or do we?⏎The truth is that the Rabbis did set in place a ritual in

  4. #4 peak

    to authentic Christianity. As a matter of fact, many who grew up in the church have the same question as those who didn’t: “What is Lent, anyway?”⏎What is Lent, Anyway?⏎Lent’s origin is hidden in the

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. L17:1069 strongly excites
  2. L16:1078 strongly excites
  3. L20:1371 strongly suppresses comma inside a question
  4. L20:1957 strongly excites
  5. L18:2812 strongly excites
  6. L20:63 excites
  7. L16:2817 suppresses
  8. L16:1123 excites
  9. L18:1002 excites
  10. L15:725 suppresses
  11. L15:2412 suppresses
  12. L19:1084 suppresses "and" joining clauses after a condition
  13. L11:296 suppresses
  14. L14:1800 suppresses
  15. L13:2080 suppresses
  16. L18:310 suppresses
  17. L5:1800 suppresses
  18. L20:491 excites
  19. L20:218 suppresses
  20. L20:373 weakly excites
  21. L12:27 weakly excites
  22. L16:111 weakly excites
  23. L10:69 weakly suppresses
  24. L14:247 weakly suppresses "of" before a field or department name

Feeds later layers

  1. L25:541 strongly suppresses
  2. L24:402 strongly suppresses
  3. L27:233 strongly suppresses
  4. L26:613 strongly suppresses
  5. L23:164 strongly suppresses
  6. L27:975 strongly suppresses
  7. L26:953 strongly suppresses
  8. L27:838 strongly suppresses
  9. L27:1080 excites
  10. L25:1025 suppresses
  11. L24:2628 suppresses
  12. L26:11 suppresses
  13. L23:2365 excites
  14. L25:683 suppresses
  15. L27:1576 excites
  16. L24:125 suppresses
  17. L27:319 excites
  18. L27:347 excites
  19. L23:46 excites
  20. L27:1801 weakly excites
  21. L22:328 weakly excites
  22. L22:282 weakly suppresses period then line break in lists
  23. L22:321 weakly excites
  24. L26:1018 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 -34.38 and -33.442 values between -30.63 and -29.691 values between -27.82 and -26.882 values between -20.32 and -19.392 values between -17.51 and -16.582 values between -13.76 and -12.833 values between -11.89 and -10.951 values between -10.95 and -10.022 values between -9.08 and -8.143 values between -8.14 and -7.215 values between -7.21 and -6.272 values between -6.27 and -5.332 values between -5.33 and -4.406 values between -4.40 and -3.4614 values between -3.46 and -2.5221 values between -2.52 and -1.5933 values between -1.59 and -0.65296 values between -0.65 and 0.292768 values between 0.29 and 1.23675 values between 1.23 and 2.1667 values between 2.16 and 3.10offnegativepositiveoften

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. .",
  4. ;",⏎
  5. .");
  6. .";
  7. .',
  8. ";
  9. ."),
  10. .",⏎

Suppresses

  1. ?
  2. ?
  3. ?⏎⏎
  4. ?⏎
  5. ?⏎⏎
  6. ?⏎
  7. ? ⏎
  8. ?"⏎
  9. ?)⏎
  10. ?”