Neuron Explorer

Qwen3-0.6B

'body' in immune and pain explanations

A neuron in layer 11 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron mostly reacts against the word 'body' when it is explaining how the body defends itself, such as in allergies or pain as a warning. Its positive side is scattered across unrelated words. Turning it up gave unrelated text, so the strongest clear pattern is the negative one.

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

    second is in which the possible causes of a person’s actions are related to his situation. This is the theory of attribution in social psychology. However, Allah’s Messenger said,⏎“No child is born except on al-fitra (Islam or primordial

  2. #2 peak

    the 17th century. In their separate times, each attempted to understand the relationship between the body and mind and how people perceive and experience reality. Descartes, having lived so many years after Plato may have been influenced by the latter

  3. #3 peak

    Klan members argued they would face harassment if they couldn’t wear masks. Likewise, political opponents of the Shah of Iran successfully argued that they needed masks to avoid reprisals from the shah’s security forces.⏎Opponents of anti-mask laws also

  4. #4 peak

    medical advice.⏎If you think you may have a medical emergency, call your doctor or (in the United States) 911 immediately. Always seek the advice of your doctor before starting or changing treatment. Medical professionals who provide responses to health

  5. #5 peak

    Hashem had spoken.” (Shemot 8:15)⏎Paroh’s magicians could not duplicate the plague of Lice. They told Paroh that this plague was the “finger of the L-rd”. Rashi seems

  6. #6 peak

    their boss, tend to gossip and spread toxicity among coworkers, bringing down morale. These individuals will likely jump at the chance of a new job should the opportunity present itself. In extreme cases,

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 sum of the numbers in the world, which I have shown you the world's favorite math problem." What is the probability of the world's favorite math problem, which has a certain pattern for the sum of the world's favorite math

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

    the body’s adaptation process that leads to increased strength or endurance as muscle recover and hypertrophy.⏎But while discomfort is natural if you push yourself, pain is the body’s protective mechanism, warning<|endoftext|>The advancement of solar technology has given us a new

  2. #2 peak

    also be affected by the pollination profile of a geographic area.⏎What Happens With an Allergy?⏎In simple terms, an allergy is the result of the body's overactive immune system trying to fight what it considers a dangerous invader.⏎

  3. #3 peak

    continue for varying periods of time.⏎There are two kinds of hives:⏎- Allergic urticaria, which we've discussed. These are caused by the body's allergic reaction to substances it encounters.⏎- Non-allergic urticaria.

  4. #4 peak

    rows. They can also be used to help scars heal. The collagen (derived from cow hide) is injected just beneath the skin where the body accepts it as its own, and allows it to fill the wrinkle or heal

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. L10:2089 excites
  2. L10:223 excites
  3. L7:2132 suppresses
  4. L7:1749 excites
  5. L10:2048 suppresses
  6. L9:2368 weakly excites
  7. L7:1301 weakly excites
  8. L8:971 weakly suppresses
  9. L5:3066 weakly excites
  10. L8:1352 weakly suppresses
  11. L7:2160 weakly excites
  12. L8:2984 weakly suppresses
  13. L6:229 weakly excites
  14. L9:2973 weakly excites
  15. L9:807 weakly excites
  16. L10:0 weakly excites
  17. L9:379 weakly suppresses
  18. L9:277 weakly excites
  19. L9:2449 weakly excites
  20. L4:2993 weakly suppresses
  21. L9:2007 weakly excites
  22. L10:1734 weakly excites
  23. L8:2007 weakly suppresses
  24. L4:2537 weakly suppresses

Feeds later layers

  1. L13:1545 suppresses
  2. L18:2655 suppresses
  3. L16:1625 suppresses
  4. L12:1458 suppresses
  5. L16:2594 suppresses
  6. L16:667 suppresses
  7. L13:2765 suppresses
  8. L12:2388 suppresses
  9. L18:1322 suppresses
  10. L18:3060 weakly suppresses
  11. L16:1894 weakly excites
  12. L19:2187 weakly suppresses
  13. L25:2879 weakly suppresses
  14. L22:1878 weakly suppresses
  15. L13:2900 weakly suppresses
  16. L16:2202 weakly suppresses
  17. L12:2462 weakly suppresses
  18. L18:3001 weakly suppresses
  19. L12:552 weakly excites
  20. L22:1254 weakly suppresses
  21. L12:1037 weakly suppresses
  22. L18:2952 weakly suppresses
  23. L15:1461 weakly suppresses
  24. L12:1787 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 -4.45 and -4.311 values between -4.16 and -4.021 values between -3.15 and -3.011 values between -1.86 and -1.711 values between -1.71 and -1.574 values between -1.57 and -1.421 values between -1.42 and -1.282 values between -1.28 and -1.132 values between -1.13 and -0.994 values between -0.99 and -0.843 values between -0.84 and -0.7013 values between -0.70 and -0.5618 values between -0.56 and -0.41114 values between -0.41 and -0.27497 values between -0.27 and -0.121843 values between -0.12 and 0.021024 values between 0.02 and 0.17299 values between 0.17 and 0.3145 values between 0.31 and 0.4619 values between 0.46 and 0.605 values between 0.60 and 0.743 values between 0.74 and 0.893 values between 0.89 and 1.032 values between 1.03 and 1.182 values between 1.18 and 1.32offnegativepositiveoften

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. enberg
  2. 行って
  3. Dann
  4. roat
  5. [d
  6. [next
  7. ��
  8. clusions
  9. etri
  10. bins

Suppresses

  1. ativa
  2. económica
  3. \Validator
  4. SystemService
  5. sez
  6. medida
  7. 的强大
  8. iais
  9. economy
  10. __.__