Neuron Explorer

Qwen3-0.6B

'of' after part-whole nouns

A neuron in layer 12 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 word "of" in phrases like "a subsidiary of", "a division of" or "a subset of", where a bigger group is about to be named. Turning it up did not give a clearly related change in the writing.

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

    Mariana Mazzucato. Copyright © 2018. Available from PublicAffairs, an imprint of Perseus Books, LLC, a subsidiary of Hachette Book Group, Inc.⏎The January 2010

  2. #2 peak

    913, upon passage of the 16th Amendment (which created the federal income tax), total government spending at all levels was about 8% of gross domestic product. By 1940, it was approximately 1

  3. #3 peak

    Highway Administration, a division of the U.S. Department of Transportation, has released the newest numbers for miles driven

  4. #4 peak

    the inventory of words in a particular language.⏎It is now widely accepted that lexicography is a scholarly discipline in its own right and not a sub-branch of applied linguistics, as the chief object of study in lexicography is the

  5. #5 peak

    or Computer Vision(how Google Lens works).⏎Machine Learning is the field of study that will enable computers to learn without being explicitly programmed.⏎Machine Learning is a subset of Artificial Intelligence.⏎Common ML Techniques⏎There are broadly three types of machine learning;

  6. #6 peak

    the ocean are threatening to the ecosystem, but all is not gloomy. This is where BLUE BEAUTY comes to the rescue. It’s a sub-section of clean or green beauty but dominantly focuses on the well-being of the<|endoftext|>Memory

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 new and the old. The new and the old. The new and the old. The new and the old. The list. The new. The new. The new. The new. The new. The new. The new.

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

    spot and demonstrate limited growth are generally considered to be benign.⏎More dangerous, or malignant, tumors form when two things occur:⏎a cancerous cell manages to move throughout the body using the blood or lymph systems, destroying healthy tissue in a process called

  2. #2 peak

    -resistant container. Keep Ampicillin out of the reach of children and away from pets.⏎Do not use Ampicillin if

  3. #3 peak

    ecost, the Spirit of the Promise was poured out on the disciples, gathered "together in one place."92 While awaiting the

  4. #4 peak

    respiratory diseases. The aerosol also contains particles that can cause inflammation of the lungs.⏎The use of nicotine in vaping products can cause physical dependence. Nicotine crosses the bloodstream and activates 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. L11:437 suppresses
  2. L6:2217 suppresses
  3. L11:2547 suppresses
  4. L9:2566 excites
  5. L10:2362 suppresses
  6. L10:132 excites
  7. L6:2842 excites
  8. L11:1858 suppresses
  9. L7:1152 suppresses
  10. L6:800 excites
  11. L9:2545 suppresses
  12. L10:1429 suppresses
  13. L11:506 suppresses
  14. L10:2468 suppresses
  15. L10:1649 excites
  16. L6:1004 excites
  17. L10:2138 weakly suppresses
  18. L8:1428 weakly excites 'on' introducing a location or surface
  19. L5:1477 weakly excites
  20. L11:2860 weakly suppresses
  21. L9:2390 weakly excites
  22. L11:2207 weakly suppresses "of" after a noun in trademark and title phrases
  23. L9:1858 weakly excites
  24. L10:1603 weakly excites

Feeds later layers

  1. L15:1338 excites
  2. L15:1514 excites
  3. L16:1984 excites
  4. L14:1707 excites
  5. L20:1038 excites
  6. L16:1798 excites
  7. L16:3007 excites
  8. L16:1096 excites
  9. L14:1508 excites
  10. L16:1051 excites
  11. L17:2959 excites
  12. L16:2789 excites
  13. L13:2135 excites
  14. L15:1741 excites
  15. L15:1305 excites
  16. L17:2022 excites
  17. L17:189 excites
  18. L18:2573 excites
  19. L22:1993 excites
  20. L17:2620 excites
  21. L17:1287 weakly excites
  22. L16:1128 weakly excites
  23. L15:1127 weakly excites
  24. L17:1888 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 -2.16 and -2.032 values between -1.26 and -1.135 values between -1.13 and -1.001 values between -1.00 and -0.875 values between -0.87 and -0.746 values between -0.74 and -0.6116 values between -0.61 and -0.4860 values between -0.48 and -0.36243 values between -0.36 and -0.23784 values between -0.23 and -0.101682 values between -0.10 and 0.03797 values between 0.03 and 0.16198 values between 0.16 and 0.2939 values between 0.29 and 0.4223 values between 0.42 and 0.5511 values between 0.55 and 0.6710 values between 0.67 and 0.805 values between 0.80 and 0.935 values between 0.93 and 1.063 values between 1.06 and 1.192 values between 1.19 and 1.322 values between 1.45 and 1.583 values between 1.58 and 1.701 values between 1.83 and 1.961 values between 2.48 and 2.602 values between 2.60 and 2.731 values between 2.86 and 2.99offnegativepositiveoften

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. 总
  2. cum
  3. itle
  4. urs
  5. Amazon
  6. uded
  7. 總
  8. ified
  9. I
  10. ful

Suppresses

  1. lashes
  2. qx
  3. 砣
  4. grave
  5. 让她
  6. zac
  7. 痘
  8. quine
  9. qx
  10. 套房