Neuron Explorer

Qwen3-0.6B

word completing a technical term

A neuron in layer 10 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 fires on the word that finishes a specialised term, like 'fever' in 'hay fever', 'disorder' in 'psychiatric disorder', or 'constellation' in star charts. Those are mostly medical and astronomy phrases. Turning it up made the model's writing wander into questions about people and life, which doesn't clearly match the pattern.

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

    Ephemeris Service; put "Philplait" into the big text box and it will give you the coordinates of the asteroid (it's currently in the constellation of Aries).

  2. #2 peak

    cardiovascular health, 38.1% in intermediate health, and 7.8% in good health. One in four participants had some type of psychiatric disorder, which were more common in women (33.6%) than in men

  3. #3 peak

    not always have negative effects. However, the following individuals may be more sensitive to mould:⏎- pregnant women⏎- the elderly⏎- those with asthma, hay fever or other allergies⏎- those with weak immune system due to other medical conditions⏎

  4. #4 peak

    alternating current in the series RLC circuit at an instant are given by⏎v=Vm sin ωt and i = I m sin( ω t + φ)⏎where ϕ is the phase angle between υ and i. The

  5. #5 peak

    Power of a sinusoidal waveform⏎Consider an electrical AC sinusoidal signal which is characterized by its voltage V(t)=Vmax×sin(ωt+ΦV) and current I(t)=Imax×sin(ωt+

  6. #6 peak

    . Common symptoms may include:⏎- runny nose⏎- nasal and throat irritation⏎- aggravation of asthma⏎- allergic reaction⏎People with asthma, hay fever or other allergies could experience temporary respiratory symptoms when exposed to low amounts of common mould

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: I need to have a working tool. I need to have a tool that can be used to build a structure. I need to have a tool that can be used to build something. I need to have a tool that can be used 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

    out.org, a project of ESC member group, the Center for Biological Diversity.⏎You can read the whole piece on the Summit County Citizens Voice website.⏎0 comments on “Less lead ammunition & less lead poisoning in California condors”<|endoftext|>Throughout life

  2. #2 peak

    island group corresponds to its governing country. Most of these places are inhabited, but some of them contain more animals than people.⏎|Printer friendly Cite/link Email Feedback|⏎|Title Annotation:||British Empire|⏎|Article Type:||Brief Article

  3. #3 peak

    - Tumors of chest wall⏎- In newborns, high levels of circulating estrogen may result in enlarged breast tissue and secretion of milk⏎Galactorrhea is more common in women. Other factors that may increase your chance of galactor

  4. #4 peak

    that you're in the mirror.⏎A simple game of peekaboo might be absolutely fascinating to your baby. That's because he's starting to understand object permanence (that an object he can't see can still exist) and loves games 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. L9:539 excites
  2. L9:2305 excites
  3. L9:1638 suppresses
  4. L9:2570 suppresses
  5. L8:13 excites
  6. L9:244 excites
  7. L8:2744 excites
  8. L6:16 suppresses
  9. L6:2385 suppresses
  10. L9:521 suppresses
  11. L7:2333 excites
  12. L8:2247 suppresses
  13. L8:16 excites
  14. L9:1959 excites
  15. L6:2328 excites
  16. L8:3 weakly suppresses
  17. L9:5 weakly suppresses
  18. L4:1738 weakly excites
  19. L9:921 weakly excites
  20. L9:2922 weakly excites
  21. L9:2476 weakly excites
  22. L4:857 weakly excites
  23. L9:2627 weakly suppresses
  24. L9:778 weakly suppresses

Feeds later layers

  1. L11:756 suppresses
  2. L11:816 suppresses
  3. L12:676 suppresses
  4. L11:1377 excites
  5. L12:863 suppresses
  6. L16:554 suppresses
  7. L12:484 suppresses
  8. L12:426 suppresses
  9. L12:617 suppresses
  10. L12:809 suppresses
  11. L13:668 suppresses
  12. L16:618 suppresses
  13. L11:468 excites
  14. L16:687 suppresses
  15. L12:479 suppresses the word 'verb' in grammar explanations
  16. L16:446 suppresses
  17. L16:384 suppresses
  18. L11:2437 excites
  19. L12:278 suppresses
  20. L13:1018 suppresses
  21. L12:802 suppresses
  22. L16:563 suppresses
  23. L16:476 suppresses
  24. L11:861 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.

4 values between -1.25 and -1.133 values between -1.13 and -1.019 values between -1.01 and -0.9014 values between -0.90 and -0.7830 values between -0.78 and -0.6643 values between -0.66 and -0.5589 values between -0.55 and -0.43155 values between -0.43 and -0.31317 values between -0.31 and -0.20596 values between -0.20 and -0.08936 values between -0.08 and 0.04619 values between 0.04 and 0.16362 values between 0.16 and 0.27234 values between 0.27 and 0.39145 values between 0.39 and 0.5192 values between 0.51 and 0.6265 values between 0.62 and 0.7451 values between 0.74 and 0.8632 values between 0.86 and 0.9728 values between 0.97 and 1.0924 values between 1.09 and 1.2111 values between 1.21 and 1.3216 values between 1.32 and 1.445 values between 1.44 and 1.563 values between 1.56 and 1.674 values between 1.67 and 1.795 values between 1.79 and 1.916 values between 1.91 and 2.023 values between 2.02 and 2.142 values between 2.14 and 2.261 values between 2.38 and 2.491 values between 2.61 and 2.731 values between 2.73 and 2.841 values between 2.96 and 3.081 values between 3.31 and 3.43offnegativepositiveoften

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. pub
  2. C
  3. I
  4. F
  5. E
  6. H
  7. ——
  8. 剖
  9. iment
  10. con

Suppresses

  1. tempList
  2. HeaderComponent
  3. 鞮
  4. 缝隙
  5. enzhen
  6. (groupId
  7. ossier
  8. lıkları
  9. 之城
  10. 中心城区