Neuron Explorer

Qwen3-0.6B

're' prefix fragment

A neuron in layer 26 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 responds to the 're' piece that begins a word like 'reusing', 'reclamation' or 'reimagined'. The steering outputs were scattered and did not reflect this trigger.

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

    mist concentrates. It is a strong and solid material, and it is likewise recyclable. Actually, noteworthy vitality investment funds and asset sparing can be accomplished through steel reusing.⏎

  2. #2 peak

    were fastest for species with widespread host plants or where host plants established well during restoration. Low mobility butterfly species took longer to colonise. We show that arable reversion is an effective tool for the management of butterflycommunities. We suggest that

  3. #3 peak

    material to radiate energy. When heat or light energy—typically from the sun or HVAC system—is absorbed by glass it is either shifted away by air movement or re-radiated by the glass surface.⏎Typically, a highly reflective material will

  4. #4 peak

    every state. With a widespread, shallow root system, it is ideal for thin soils where it will prevent soil erosion; it is often used for strip-mine reclamation projects barren due to mining debris and acid soils.⏎The tree has long thin

  5. #5 peak

    (Excerpt from Wall Street Journal).⏎Our team of designers, architects, engineers and scientists were brought together by Scott Francisco of Pilot Projects.⏎Brooklyn Bridge Forest re-imagines the bridge as an icon of climate action and social equity, improving

  6. #6 peak

    .⏎Recycling - An Important Part Of The Solution⏎The more we recycle, the less garbage winds up in our landfills and incineration plants. By reusing aluminum, paper, glass, plastics, and other materials, we can save

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: what if we can electric charge electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field electric field

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

    times atmospheric pressure (24 gigapascals) and reach 1.3 million times atmospheric pressure (13

  2. #2 peak

    1776. Washington went into winter quarters and waited for a spring that would bring new hardships. Most of his soldiers went home, to be replaced by unpromising recruits. But now, for the rest of the war, there would

  3. #3 peak

    MG 34 (shortened from German: ''Maschinengewehr 34'', or "machine gun 34") is a German⏎recoil-operated air-cooled machine gun⏎, first tested in 19

  4. #4 peak

    is an act of Congress⏎codified in 5 USC, section 552. This Act allows for the full or partial⏎disclosure of previously unreleased information and documents controlled⏎by the Federal Government. The Act defines agency records

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. L22:979 excites
  2. L23:1099 excites
  3. L22:2630 suppresses
  4. L22:1780 excites
  5. L16:110 excites
  6. L14:2568 weakly excites
  7. L23:491 weakly excites
  8. L19:372 weakly excites
  9. L16:17 weakly excites
  10. L19:322 weakly excites
  11. L18:178 weakly suppresses
  12. L23:2797 weakly excites
  13. L11:429 weakly suppresses
  14. L2:86 weakly excites
  15. L2:571 weakly excites
  16. L4:1374 weakly excites
  17. L8:2701 weakly suppresses
  18. L20:1298 weakly suppresses
  19. L24:459 weakly excites
  20. L21:1999 weakly excites 'well' in compound adjectives
  21. L16:1153 weakly suppresses
  22. L9:226 weakly suppresses 'to' starting an infinitive
  23. L14:1498 weakly excites
  24. L0:1012 weakly suppresses "of" after nouns like eve, family, modes

Feeds later layers

  1. L27:1528 weakly excites
  2. L27:2371 weakly suppresses
  3. L27:2646 weakly suppresses lowercase p starting a word
  4. L27:783 weakly suppresses
  5. L27:564 weakly suppresses
  6. L27:3041 weakly excites
  7. L27:2727 weakly excites
  8. L27:1994 weakly suppresses
  9. L27:1138 weakly excites
  10. L27:1935 weakly excites
  11. L27:1835 weakly excites
  12. L27:1894 weakly excites
  13. L27:1235 weakly suppresses
  14. L27:971 weakly suppresses
  15. L27:608 weakly suppresses
  16. L27:1560 weakly excites
  17. L27:2133 weakly suppresses
  18. L27:1319 weakly suppresses
  19. L27:2171 weakly suppresses
  20. L27:1253 weakly excites
  21. L27:1834 weakly excites
  22. L27:1825 weakly excites
  23. L27:2060 weakly suppresses
  24. L27:2785 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 -20.53 and -19.481 values between -18.43 and -17.381 values between -17.38 and -16.331 values between -16.33 and -15.281 values between -15.28 and -14.231 values between -14.23 and -13.183 values between -13.18 and -12.133 values between -12.13 and -11.098 values between -11.09 and -10.047 values between -10.04 and -8.996 values between -8.99 and -7.9413 values between -7.94 and -6.8921 values between -6.89 and -5.8421 values between -5.84 and -4.7945 values between -4.79 and -3.7476 values between -3.74 and -2.69112 values between -2.69 and -1.64282 values between -1.64 and -0.592579 values between -0.59 and 0.46591 values between 0.46 and 1.5154 values between 1.51 and 2.5623 values between 2.56 and 3.6119 values between 3.61 and 4.6614 values between 4.66 and 5.716 values between 5.71 and 6.764 values between 6.76 and 7.817 values between 7.81 and 8.862 values between 8.86 and 9.912 values between 9.91 and 10.961 values between 10.96 and 12.012 values between 12.01 and 13.061 values between 20.40 and 21.45offnegativepositiveoften

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. otion
  2. asionally
  3. ful
  4. y
  5. icial
  6. usa
  7. oupper
  8. صيد
  9. エン
  10. entlich

Suppresses

  1. .msg
  2. 祚
  3. .setCurrent
  4. ')");⏎
  5. �权
  6. .esp
  7. buff
  8. Terr
  9. 帘
  10. nie