Neuron Explorer

Qwen3-0.6B

colon that introduces a statement

A neuron in layer 14 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 reacts to a colon that is about to introduce an explanation, quote or example, such as "I start to wonder:" or "here's one example:". Turning it down made the model output strings of punctuation, and turning it up gave rambling text.

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

    we have almost no idea about how powerful the military technologies of the aliens can be and whether they will be friendly to the human or not; I start to wonder: ‘Assuming there is only finite resource on the earth based on which we cannot

  2. #2 peak

    , many literature professors have taken an extreme position that authorial intent should never be considered, ever. To even look down that road is foolish.⏎But consider this: if you walk into any contemporary art museum you will, without fail, find on

  3. #3 peak

    the interruption would last. But with her intimate knowledge of volcanic ash, and a good idea of what it does to jet engines, she was sure of one thing: “I wouldn’t want to get on a plane right now.”⏎Admirably

  4. #4 peak

    benefits to agriculture.⏎For more information, please contact:⏎Dr. Norihito Nakamichi⏎Biodynamics Research Team⏎RIKEN Plant Science Center⏎Tel: +81

  5. #5 peak

    , Comparative Physiology also gave birth to an experimental approach that has come to be known as the “August Krogh Principle,” which states: “For many problems, there is an animal on which it can be most conveniently

  6. #6 peak

    time.⏎You should give examples from dynamic typed languages instead, e.g. clojure.⏎Here's one example:<|endoftext|>The commerce of the ancient world was gathered into the warehouses of Tyre.

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::::::::::::::::::::::::: : ::::::::::: :: is that??:? is this for:

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

    Why Lesson Planet?⏎Find quality lesson planning resources, fast!⏎Share & remix collections to collaborate.⏎Organize your curriculum with collections. Easy!⏎

  2. #2 peak

    poverty group. A family from this group with four members, which is the average family size in the<|endoftext|>Why Lesson Planet?⏎Find quality lesson planning resources, fast!⏎Share & remix collections to collaborate.⏎Organize your curriculum with collections. Easy!⏎

  3. #3 peak

    test, hemogram, ESR⏎with antibiotics and anti-inflammatory medications helps reduce tissue damage⏎from inflammation, lessen pain

  4. #4 peak

    ?, on Time's website.<|endoftext|>Start a 10-Day Free Trial to Unlock the Full Review⏎Why Lesson Planet?⏎Find quality lesson planning resources, fast!⏎Share & remix collections to collaborate.⏎Organize your curriculum with collections. Easy!⏎

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. L9:7 strongly excites
  2. L1:1 strongly excites
  3. L10:33 strongly suppresses
  4. L12:5 strongly excites
  5. L2:204 suppresses
  6. L3:3 suppresses
  7. L6:4 suppresses
  8. L8:2 excites
  9. L1:8 excites
  10. L6:0 excites
  11. L2:44 excites
  12. L5:5 suppresses digit after a label like paper or type
  13. L13:10 weakly excites
  14. L0:1224 weakly suppresses
  15. L8:30 weakly excites blank slot right before a number
  16. L10:86 weakly suppresses
  17. L0:1301 weakly suppresses
  18. L13:345 weakly excites
  19. L0:178 weakly suppresses
  20. L0:1864 weakly excites
  21. L13:95 weakly excites
  22. L1:106 weakly excites
  23. L9:2517 weakly suppresses
  24. L3:2331 weakly suppresses

Feeds later layers

  1. L15:123 strongly excites
  2. L18:48 strongly excites
  3. L17:38 strongly excites
  4. L16:170 strongly excites
  5. L15:44 strongly suppresses
  6. L21:173 strongly excites
  7. L20:863 strongly excites
  8. L20:8 strongly suppresses
  9. L17:62 strongly suppresses
  10. L19:131 strongly suppresses
  11. L25:118 strongly excites
  12. L23:35 strongly excites
  13. L22:262 strongly excites comma inside big numbers
  14. L21:195 strongly excites
  15. L17:137 strongly suppresses
  16. L16:241 strongly suppresses
  17. L25:36 excites
  18. L18:887 excites
  19. L27:143 excites
  20. L26:94 excites
  21. L18:246 excites
  22. L25:273 excites dollar sign before an amount
  23. L24:130 excites line break at the end of a line
  24. L20:836 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 -1.95 and -1.761 values between -1.76 and -1.571 values between -1.57 and -1.393 values between -1.39 and -1.202 values between -1.20 and -1.0281 values between -1.02 and -0.83596 values between -0.83 and -0.641290 values between -0.64 and -0.461048 values between -0.46 and -0.27326 values between -0.27 and -0.0863 values between -0.08 and 0.1010 values between 0.10 and 0.2923 values between 0.29 and 0.4823 values between 0.48 and 0.6635 values between 0.66 and 0.8527 values between 0.85 and 1.0334 values between 1.03 and 1.2230 values between 1.22 and 1.4130 values between 1.41 and 1.5951 values between 1.59 and 1.7831 values between 1.78 and 1.9737 values between 1.97 and 2.1538 values between 2.15 and 2.3428 values between 2.34 and 2.5230 values between 2.52 and 2.7112 values between 2.71 and 2.9013 values between 2.90 and 3.086 values between 3.08 and 3.2711 values between 3.27 and 3.466 values between 3.46 and 3.646 values between 3.64 and 3.836 values between 3.83 and 4.012 values between 4.01 and 4.202 values between 4.20 and 4.393 values between 4.39 and 4.571 values between 4.76 and 4.951 values between 5.32 and 5.50offnegativepositiveoften

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. 特别是
  2. anca
  3. 尤其是
  4. 特别是在
  5. merce
  6. idents
  7. rawler
  8. 榔
  9. ([]*
  10. 一门

Suppresses

  1. :
  2. (Y
  3. :y
  4. -packed
  5. :t
  6. :E
  7. (y
  8. (d
  9. =d
  10. (t