Neuron Explorer

Qwen3-0.6B

'into' after a verb or 'well'

A neuron in layer 19 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 lights up on the word 'into' when it follows a verb or phrases like 'well into', as in 'vanishes into the past'. Turning it up did not produce any clear theme in the writing.

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

    Until well into the 20th century, knowledge of Jacques Le Moyne de Morgues

  2. #2 peak

    4 “Glory to God in the highest,⏎and on earth peace among those with whom he is pleased!”⏎15 When the angels went away from them into heaven, the shepherds said to

  3. #3 peak

    which seems to have been written by their friend, Hugo Dyson! When they rush to warn him, Hugo walks through a door in time - and vanishes into the past. And in that moment, the world begins to change.The Frontier,

  4. #4 peak

    and warm, southern ones. You can enjoy it for three whole seasons every year with varieties which flower as early as March, and others which produce flowers until well into the autumn. This makes

  5. #5 peak

    oughly 180 million years ago, during the height of the Jurassic period, the Earth's magnetic field flipped, bringing the magnetic north pole once again into the Northern Hemisphere.⏎This so-called van Zijl reversal, named for the

  6. #6 peak

    movements for our freedom. He has faced many struggles went into prison for many times. After his great struggle, we got freedom on 1

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: that in this 500% (100000000000000000000000000000000000000

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 American Academy of Child & Adolescent Psychiatry believes the problem is even more widespread, reporting that as many as 50 percent of all young people are bullied at least once during their school years, with about one in 10 suffering from

  2. #2 peak

    he shared:⏎The tragic Ghost Ship fire in Oakland has reminded us all that fire can kill very quickly. And it is not the only danger that we may face at home and in public places. A few basic precautions can save your life — but

  3. #3 peak

    nonetheless. Here are two important reasons why:⏎The odds of needing long-term care are high:⏎- Approximately 70% of people will need long-term care at some point during their lifetimes after reaching age 65.*⏎- Approximately

  4. #4 peak

    :⏎- Protected classes in California⏎- Anti-discrimination requirements for employers in California⏎- What to do if your protected class is discriminated against⏎Facing discrimination at your workplace? You came to the right place.⏎Protected Classes in California⏎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. L10:132 excites
  2. L18:135 excites
  3. L16:825 weakly excites
  4. L4:1406 weakly excites
  5. L12:2522 weakly excites
  6. L8:1627 weakly excites
  7. L4:328 weakly suppresses 'to' after 'so as' and infinitive cues
  8. L13:2167 weakly excites
  9. L18:1069 weakly suppresses
  10. L11:293 weakly suppresses
  11. L6:1149 weakly suppresses
  12. L15:317 weakly excites
  13. L2:1579 weakly suppresses
  14. L12:559 weakly suppresses
  15. L4:2899 weakly suppresses
  16. L5:2992 weakly excites
  17. L3:2929 weakly excites
  18. L4:1847 weakly suppresses
  19. L15:1328 weakly suppresses
  20. L10:944 weakly suppresses
  21. L6:515 weakly suppresses
  22. L3:687 weakly excites
  23. L5:3009 weakly excites 'to' after a verb or noun
  24. L0:2383 weakly excites

Feeds later layers

  1. L21:1935 excites
  2. L22:2262 excites
  3. L25:2648 suppresses
  4. L20:1139 excites 'in' after a past participle
  5. L24:893 weakly suppresses
  6. L21:2823 weakly excites
  7. L24:2584 weakly excites 'in' in set phrases like 'in mind'
  8. L20:83 weakly excites
  9. L23:2125 weakly suppresses
  10. L20:1722 weakly excites
  11. L22:2998 weakly excites
  12. L21:2170 weakly excites
  13. L24:1442 weakly excites
  14. L25:3070 weakly excites
  15. L27:1851 weakly suppresses capital K starting a word
  16. L23:2136 weakly excites 'in' after a noun phrase
  17. L21:1081 weakly excites
  18. L25:1865 weakly excites 'in' in 'keep in mind'
  19. L24:1957 weakly excites capital 'In' starting a word or name
  20. L21:1335 weakly suppresses
  21. L24:2163 weakly excites
  22. L25:1853 weakly excites
  23. L23:769 weakly suppresses
  24. L21:2772 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 -13.55 and -12.393 values between -11.24 and -10.091 values between -10.09 and -8.933 values between -7.78 and -6.638 values between -6.63 and -5.474 values between -5.47 and -4.3228 values between -4.32 and -3.1756 values between -3.17 and -2.01195 values between -2.01 and -0.862828 values between -0.86 and 0.30617 values between 0.30 and 1.4571 values between 1.45 and 2.6031 values between 2.60 and 3.7617 values between 3.76 and 4.9112 values between 4.91 and 6.068 values between 6.06 and 7.225 values between 7.22 and 8.375 values between 8.37 and 9.522 values between 9.52 and 10.681 values between 10.68 and 11.833 values between 11.83 and 12.983 values between 12.98 and 14.143 values between 14.14 and 15.291 values between 17.60 and 18.751 values between 18.75 and 19.911 values between 31.44 and 32.59offnegativepositiveoften

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. liğinde
  2. osphere
  3. ther
  4. ther
  5. nowhere
  6. oto
  7. odon
  8. ducted
  9. _collections
  10. BuildContext

Suppresses

  1. hammad
  2. 섭
  3. 人と
  4. sti
  5. งาม
  6. root
  7. Ctl
  8. -notch
  9. 泪
  10. #undef