Neuron Explorer

Qwen3-0.6B

h starting a new word piece

A neuron in layer 25 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 letter h when it begins a new piece of a word, especially after a hyphen or a break, like the "-h" in "re-hydration" or "H" starting "Hypothesis". Turning it up gave odd, repeated words, so the steering does not show a clear theme.

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

    can. In-fact, these machine learning tools do out-predict individual traits better than work colleagues with only ten likes; family members, friends, and co-habitants with only 100 to 150 likes; and

  2. #2 peak

    transform atmospheric CO2 into organic molecules, plants can use the energy from⏎Transpiration decreases as air becomes drier.⏎De-hardening in spring involves gradual re-hydration of the cells, recovery of photosynthetic capacity and a tight control

  3. #3 peak

    any of these areas and would⏎like to participate in our research programs, please contact us.⏎ARTICHOKE SUPPLEMENTATION FOR DIABETES⏎HYPOTHESIS: Dietary supplementation⏎with Jerusalem artichoke (JA) significantly

  4. #4 peak

    loses some of its density) when coming in contact with acids. Saliva can dilute and neutralize acid so the tooth can re-mineralize (re-harden). This process isn’t instantaneous, however. While the enamel is soft,

  5. #5 peak

    OOK OF PHYSIOLOGY.⏎For Students and Practitioners of Medicine. By AUSTIN⏎A COURSE OF ELEMENTARY⏎PHYSIOLOGY AND HISTOLOGY. By Sir M. FOSTER, K. C.

  6. #6 peak

    the other pronoun, me, which is serves as a indirect object pronoun in me cuesta and as a reflexive pronoun in expresarme.¿Hablas el inglés? —Me cuesta trabajo hablarlo y expresarme,

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: as a child, as a child on a car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car car

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

    In the following example, hijos is a masculine, plural noun so the masculine plural direct object pronoun los is used to replace it in the second sentence.Son muy buenos hijos. No los cambiaría.⏎They are good

  2. #2 peak

    from 1472 to 1554. This work was printed at Madrid in 1601; reprinted by Juan de la Cuesta in 1615; revised and augmented by Andrés González

  3. #3 peak

    , World Bank says. Wealth remains unequal, and some countries are falling behind.⏎El Banco Mundial analiza la riqueza de los países, tomando en cuenta no solo el PIB, sino también el capital humano y el natural.⏎

  4. #4 peak

    above the border checkpoint at San Ysidro, California. “En nombre del cielo, les pido posada,” sings one group. “Pues no puede andar mi esposa am

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. L21:160 suppresses
  2. L22:245 suppresses
  3. L23:322 suppresses
  4. L24:346 excites
  5. L24:624 suppresses
  6. L24:375 suppresses
  7. L24:533 suppresses
  8. L23:268 weakly excites
  9. L24:231 weakly suppresses
  10. L21:266 weakly excites
  11. L22:322 weakly excites
  12. L19:388 weakly excites
  13. L13:1706 weakly excites
  14. L18:247 weakly suppresses
  15. L23:2095 weakly suppresses
  16. L13:1658 weakly suppresses
  17. L8:576 weakly excites
  18. L22:700 weakly suppresses
  19. L20:2338 weakly excites
  20. L24:482 weakly suppresses inside Italian names and places
  21. L16:2535 weakly excites
  22. L12:1269 weakly suppresses
  23. L16:1173 weakly suppresses
  24. L17:1759 weakly suppresses

Feeds later layers

  1. L27:600 weakly excites
  2. L27:723 weakly suppresses
  3. L26:878 weakly suppresses
  4. L27:855 weakly excites
  5. L26:829 weakly suppresses
  6. L26:2641 weakly suppresses
  7. L26:2398 weakly suppresses
  8. L27:667 weakly excites
  9. L27:1670 weakly suppresses
  10. L26:2040 weakly suppresses
  11. L27:2318 weakly excites
  12. L27:1483 weakly suppresses
  13. L27:321 weakly suppresses
  14. L26:2834 weakly excites
  15. L27:1085 weakly suppresses
  16. L26:546 weakly suppresses
  17. L27:2166 weakly suppresses
  18. L26:1429 weakly excites
  19. L26:431 weakly excites
  20. L27:2331 weakly suppresses
  21. L26:1843 weakly excites
  22. L27:636 weakly excites
  23. L27:1557 weakly excites
  24. L27:2678 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.

2 values between -8.06 and -7.501 values between -6.38 and -5.821 values between -5.82 and -5.261 values between -4.70 and -4.145 values between -3.03 and -2.471 values between -2.47 and -1.914 values between -1.91 and -1.3521 values between -1.35 and -0.79470 values between -0.79 and -0.232980 values between -0.23 and 0.33367 values between 0.33 and 0.8937 values between 0.89 and 1.452 values between 1.45 and 2.012 values between 2.01 and 2.571 values between 2.57 and 3.131 values between 3.13 and 3.693 values between 3.69 and 4.252 values between 4.25 and 4.812 values between 4.81 and 5.373 values between 5.37 and 5.931 values between 9.85 and 10.411 values between 13.77 and 14.33offnegativepositiveoften

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. ig
  2. iras
  3. 睢
  4. �
  5. esseract
  6. isposable
  7. un
  8. quis
  9. elize
  10. ervised

Suppresses

  1. 第一百
  2. palms
  3. Nd
  4. orges
  5. Hearth
  6. seventeen
  7. intention
  8. /cards
  9. Nhap
  10. cars