Neuron Explorer

Qwen3-0.6B

"up" in "sign up"

A neuron in layer 3 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 "up" in the phrase "sign up", as in signing up for a newsletter or an account. Turning it up made the model write about registering for something new, which fits the same theme.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    been viewed as a resource to be conquered.⏎Elizabeth Flock Elizabeth Flock. Become a member of the book club by joining our Facebook group , or by signing up to our newsletter. Learn more about the book club here. Below are questions to

  2. #2 peak

    DRM protected ebook on your desktop or laptop you will need to have Adobe Digital Editions installed. It is a free software. We also strongly recommend that you sign up for an AdobeID at the Adobe website. For more details please see FAQ

  3. #3 peak

    DRM protected ebook on your desktop or laptop you will need to have Adobe Digital Editions installed. It is a free software. We also strongly recommend that you sign up for an AdobeID at the Adobe website. For more details please see FAQ

  4. #4 peak

    ledge of Allegiance, the Ten Commandments, the Boy Scouts, the institution of Marriage — and how these wrong decisions can be avoided and remedied. Sign up now while there is still space in this special course!⏎Lessons cover issues such

  5. #5 peak

    DRM protected ebook on your desktop or laptop you will need to have Adobe Digital Editions installed. It is a free software. We also strongly recommend that you sign up for an AdobeID at

  6. #6 peak

    views that are particularly detrimental for women.⏎But nothing — not even a conservative culture or fear of the Taliban<|endoftext|>This preview shows pages 1–2. Sign up to view the full content.⏎This preview has intentionally blurred sections. Sign up to

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 am really not sure I am. I am so sure I am. Which one is the most accurate and accurate? I need to know. So, I think. So, I think. So, I think

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 following decades, becoming the legal foundation for the reception in Germany of more than 4.5 million so-called “resettlers” (Aussiedler and Spätaussiedler) from Eastern Europe. In this contribution,

  2. #2 peak

    ic missions among the Guaranis - Relação abbreviada da republica, que os religiosos jesuitas estabelecerão no Paraguai, by Marquis of Pombal, Candide, by Voltaire

  3. #3 peak

    ain some of your losses. Don’t make applications too early, since volatilization losses could be high without rain to ensure incorporation of N.⏎Be wary

  4. #4 peak

    he not be buried until two days following his death.⏎To assuage this fear of being buried alive the safety coffin was invented. It is fitted with

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. L0:917 weakly suppresses
  2. L1:1370 weakly excites
  3. L2:2451 weakly excites
  4. L2:292 weakly excites
  5. L1:2052 weakly suppresses
  6. L2:3027 weakly excites
  7. L2:2392 weakly suppresses
  8. L2:2772 weakly excites
  9. L0:1336 weakly suppresses
  10. L0:1948 weakly excites
  11. L1:1354 weakly excites
  12. L0:2171 weakly excites
  13. L2:2529 weakly excites
  14. L2:2288 weakly suppresses
  15. L1:2109 weakly suppresses
  16. L2:2596 weakly suppresses
  17. L2:767 weakly excites
  18. L1:1081 weakly suppresses
  19. L2:1556 weakly suppresses
  20. L0:410 weakly excites
  21. L1:2288 weakly suppresses
  22. L1:2922 weakly excites
  23. L0:1 weakly suppresses
  24. L0:1507 weakly excites

Feeds later layers

  1. L16:2884 excites
  2. L5:1125 excites
  3. L25:3019 weakly suppresses
  4. L16:2496 weakly suppresses
  5. L18:2759 weakly suppresses
  6. L4:1754 weakly excites
  7. L6:2224 weakly suppresses
  8. L24:2337 weakly excites
  9. L4:2899 weakly suppresses
  10. L25:1654 weakly suppresses
  11. L11:922 weakly suppresses
  12. L25:1745 weakly suppresses
  13. L13:2668 weakly excites
  14. L5:1048 weakly suppresses
  15. L9:1209 weakly excites
  16. L17:2766 weakly suppresses
  17. L5:1585 weakly excites
  18. L10:987 weakly suppresses
  19. L5:2746 weakly suppresses
  20. L7:2356 weakly excites
  21. L15:1496 weakly excites
  22. L12:2214 weakly suppresses
  23. L16:419 weakly suppresses
  24. L16:2792 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 -1.24 and -1.161 values between -0.70 and -0.625 values between -0.47 and -0.398 values between -0.39 and -0.3271 values between -0.32 and -0.24211 values between -0.24 and -0.16591 values between -0.16 and -0.091052 values between -0.09 and -0.011152 values between -0.01 and 0.07531 values between 0.07 and 0.15193 values between 0.15 and 0.2259 values between 0.22 and 0.3016 values between 0.30 and 0.387 values between 0.38 and 0.451 values between 0.45 and 0.531 values between 0.53 and 0.613 values between 0.68 and 0.761 values between 0.84 and 0.911 values between 1.07 and 1.141 values between 1.37 and 1.451 values between 1.60 and 1.681 values between 1.76 and 1.83offnegativepositiveoften

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. 伏
  2. Using
  3. using
  4. 维
  5. ke
  6. ](
  7. ee
  8. ósito
  9. using
  10. oto

Suppresses

  1. arness
  2. arez
  3. rength
  4. happiest
  5. 钫
  6. 就够了
  7. indsight
  8. igne
  9. vetica
  10. Wohnung