Neuron Explorer

Qwen3-0.6B

'which' after a preposition

A neuron in layer 16 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 'which' when it follows a small word like 'to' or 'from', as in 'a common purpose to which' or 'opium poppies, from which'. Turning it up gave text that did not show this pattern, so the link is only in the examples.

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

    teams are highly structured and are sometimes confused with the work group. Teamwork can be difficult to establish within a traditional team without the identification of a common purpose to which all workers can agree and commit.⏎The self-directed

  2. #2 peak

    to a garden, a theme to which all other parts relate. This may be a walk or a summer-house or a

  3. #3 peak

    and parsley. When crushed, wild chervil smells sweet and carrot-like.⏎Wild chervil has a more bitter flavor than garden chervil, to which it is related. As with garden chervil, cooks primarily use wild ch

  4. #4 peak

    medical course at the University of Tübingen, but the Franco-Prussian war interrupted his studies. After the war he returned to the university, from which he received the M.D. degree with honors in 1877.⏎

  5. #5 peak

    Italian botanists of the Renaissance devoted them-selves chiefly to interpreting the works of the classical writers on Natural History, and to the identification of the plants to which they referred. This came about quite naturally, from the fact that the Mediterranean flora

  6. #6 peak

    with fibromyalgia. Decongestants are stimulants which the doctor says can make the symptoms of co-morbid conditions (a category to which fibromyalgia belongs) worse.⏎There's an additional risk of an unintention

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 right way to do for you? for you to know how to help you with your to you for you to help with your for you to the way of for you to the for you to the (what is the right way

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

    three types of configuration files: machine, application, and security.⏎Configuration files contain elements, which are logical data structures that set configuration information. Within a configuration file, you use tags to mark the beginning and end of an element. For example,

  2. #2 peak

    41 patent models.⏎This sampling of patent models from the Textile Collection describes the two major groupings, textile machinery and sewing machines. In both groups, the examination of the models begins with the earliest of the inventions. In this early

  3. #3 peak

    ation: To ensure that an RTG infrastructure will be in place to meet requirements after the 1997 Cassini<|endoftext|>When building cascading deck stairs, it is important to make sure that the stairs are properly supported and that each stair

  4. #4 peak

    sections:⏎In addition, the Wiki contains state-by-state information on ratifications of international human rights treaties, communications/ complaints mechanisms and inquiry procedures.⏎On the Wiki, read:⏎Our eventual aim is to make all information about children's rights available in

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. L15:853 strongly suppresses
  2. L8:1052 suppresses
  3. L11:1781 suppresses
  4. L11:949 excites
  5. L14:520 suppresses
  6. L15:1468 excites
  7. L15:2521 excites
  8. L13:2780 excites reciprocal 'each other' or 'one another'
  9. L10:371 suppresses
  10. L13:2516 suppresses
  11. L9:361 excites
  12. L14:2854 suppresses
  13. L3:311 suppresses
  14. L4:1593 suppresses
  15. L11:2663 excites
  16. L4:23 weakly excites
  17. L4:1330 weakly suppresses
  18. L15:379 weakly excites
  19. L5:729 weakly suppresses
  20. L5:2424 weakly excites
  21. L5:224 weakly excites
  22. L10:1603 weakly excites
  23. L5:1734 weakly excites
  24. L10:289 weakly excites

Feeds later layers

  1. L17:2238 strongly excites
  2. L20:1912 excites
  3. L17:2878 excites
  4. L18:162 excites
  5. L17:2643 suppresses
  6. L20:242 suppresses
  7. L20:2146 excites
  8. L17:2162 weakly excites
  9. L18:740 weakly excites
  10. L20:386 weakly excites
  11. L17:2431 weakly suppresses
  12. L18:2476 weakly suppresses
  13. L17:2524 weakly suppresses
  14. L19:689 weakly excites
  15. L19:366 weakly excites
  16. L17:2690 weakly excites
  17. L25:987 weakly excites opening bracket before an acronym
  18. L19:1364 weakly excites
  19. L20:685 weakly suppresses
  20. L25:666 weakly suppresses
  21. L18:1007 weakly suppresses
  22. L21:1872 weakly suppresses
  23. L19:2455 weakly suppresses
  24. L25:281 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 -3.26 and -2.991 values between -1.63 and -1.362 values between -1.36 and -1.0924 values between -1.09 and -0.82190 values between -0.82 and -0.55856 values between -0.55 and -0.282337 values between -0.28 and -0.01453 values between -0.01 and 0.2731 values between 0.27 and 0.544 values between 0.54 and 0.814 values between 0.81 and 1.082 values between 1.08 and 1.351 values between 1.62 and 1.891 values between 2.71 and 2.981 values between 7.32 and 7.59offnegativepositiveoften

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. bú
  2. zę
  3. prostitution
  4. São
  5. 供电
  6. ходим
  7. brazil
  8. bildung
  9. rosso
  10. 阄

Suppresses

  1. 对她
  2. 给我们
  3. notin
  4. 对企业
  5. olia
  6. 共和
  7. 对孩子
  8. stone
  9. 出炉
  10. fur