Neuron Explorer

Qwen3-0.6B

prepositions in fixed phrases

A neuron in layer 24 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up on the small word that completes a set phrase, like 'under a microscope', 'behind the wheel' or 'in front of a bright light'. Steering it up produced repetitive, broken text, so the effect on writing is hard to read.

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

    . By starting earlier it finishes its flow earlier. Thus solves the problem of too much melatonin in the morning. For some people the idea of sitting in front of a bright light first thing in the morning is unpleasant and simply not practical. There

  2. #2 peak

    the only thing we do that makes us all the same. Motor vehicles are the true equalizer. Remember driving is a privilege not a right, when you’re behind the wheel you are responsible for a 4000 lb. potential weapon

  3. #3 peak

    A dive, use a submersible vessel, or study marine life from shore. A marine biologist may work in a laboratory, where they may examine tiny creatures under a microscope, on a large research vessel

  4. #4 peak

    to take action and remove the clear and present danger of mobile phones on our roads.”⏎Have your say | Should drivers be banned from using hands-free mobile phones when behind the wheel? #RSGB— Road Safety GB (@Road_Safety_GB)

  5. #5 peak

    doctor will ask about your symptoms and medical history. A physical exam will be done. Tests may include:⏎- A sample of the breast discharge to look at under a microscope⏎- Blood<|endoftext|>Welcome to the Computer Science Program!⏎Tarleton State University

  6. #6 peak

    often perform tissue biopsy (a scientific procedure for collecting cell samples for closer examination) when a patient reports suspicious lump or symptoms.⏎Through examining the appearance of cells under a microscope, the presence of cancer can be determined. This process reveals the type of

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 support, not just to see, but to see, and to see how to see, to see how to see—so that we can see, not just to see, but to see, to see, to see, to

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

    8 to April 26. What about solar power? On April 19, the unveiling of the solar classroom provided by Con Edison is sure to be eye-opening. And who doesn’t want to see those marvelous birds of

  2. #2 peak

    "Black-haired and dark-skinned men and women were powdered a ghostly white as they swept the streets for a last fistful."<|endoftext|>Editor's note: Brain Basics from Scientific American Mind is a series of short video primers on the brain

  3. #3 peak

    ite woman and has children with her.⏎Now, hereís a hint as to why Moses included this in Genesis 3<|endoftext|>Sujata Gupta for Nature – Brain food: Clever eating:⏎Our ancestors’ consumption of meat is what made us

  4. #4 peak

    a million displaced. It generated an international response that was probably unprecedented in scale. As someone who regularly reads articles on underfunded international crisis appeals, I was heartened by the response to the tsunami. That it hit easily accessible coastal regions,

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. L5:2652 weakly excites
  2. L4:2299 weakly excites
  3. L13:239 weakly suppresses
  4. L16:395 weakly suppresses
  5. L0:2570 weakly suppresses
  6. L7:557 weakly excites
  7. L21:814 weakly excites
  8. L22:2518 weakly suppresses
  9. L7:2724 weakly excites
  10. L19:1241 weakly suppresses
  11. L2:3070 weakly suppresses
  12. L17:2822 weakly suppresses
  13. L10:3006 weakly excites
  14. L22:373 weakly suppresses
  15. L8:1004 weakly excites
  16. L5:510 weakly excites
  17. L0:2454 weakly suppresses
  18. L22:143 weakly excites
  19. L1:977 weakly suppresses
  20. L6:576 weakly excites
  21. L5:1197 weakly excites
  22. L6:615 weakly excites
  23. L3:312 weakly suppresses
  24. L6:2465 weakly excites

Feeds later layers

  1. L26:507 excites
  2. L26:2782 excites
  3. L27:127 weakly suppresses
  4. L27:797 weakly excites
  5. L27:425 weakly excites
  6. L25:490 weakly suppresses
  7. L27:792 weakly excites
  8. L25:336 weakly excites
  9. L25:750 weakly suppresses
  10. L26:1310 weakly suppresses
  11. L27:2084 weakly excites
  12. L27:1486 weakly suppresses
  13. L26:851 weakly suppresses
  14. L25:1586 weakly excites
  15. L25:530 weakly excites
  16. L25:2569 weakly excites
  17. L25:906 weakly suppresses
  18. L25:1335 weakly excites
  19. L25:1131 weakly excites
  20. L25:2910 weakly suppresses
  21. L27:2932 weakly excites capital F starting a word
  22. L27:1641 weakly excites
  23. L26:2348 weakly suppresses
  24. L25:135 weakly suppresses

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 -16.56 and -15.801 values between -14.27 and -13.511 values between -13.51 and -12.751 values between -11.22 and -10.461 values between -10.46 and -9.702 values between -9.70 and -8.931 values between -8.93 and -8.171 values between -8.17 and -7.415 values between -7.41 and -6.645 values between -6.64 and -5.885 values between -5.88 and -5.129 values between -5.12 and -4.3511 values between -4.35 and -3.5921 values between -3.59 and -2.8327 values between -2.83 and -2.0648 values between -2.06 and -1.30228 values between -1.30 and -0.542605 values between -0.54 and 0.23711 values between 0.23 and 0.9988 values between 0.99 and 1.7554 values between 1.75 and 2.5221 values between 2.52 and 3.2815 values between 3.28 and 4.0410 values between 4.04 and 4.8010 values between 4.80 and 5.575 values between 5.57 and 6.337 values between 6.33 and 7.093 values between 7.09 and 7.861 values between 7.86 and 8.623 values between 8.62 and 9.381 values between 9.38 and 10.152 values between 10.91 and 11.671 values between 11.67 and 12.443 values between 13.20 and 13.96offnegativepositiveoften

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. stay
  2. ersen
  3. ˢ
  4. umont
  5. ấm
  6. ertation
  7. رام
  8. The
  9. trzymał
  10. oral

Suppresses

  1. AttribPointer
  2. idos
  3. 高等学校
  4. shoulders
  5. 照明
  6. outube
  7. numbers
  8. tires
  9. arily
  10. less