Neuron Explorer

Qwen3-0.6B

'not' in 'if not' hedges

A neuron in layer 13 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 "not" in phrases like "if not the most important" or "if not impossible", where a writer qualifies a claim by saying something stronger. It is one of the clearest patterns in this set, though turning it up made the model's text wander rather than show this 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

    is one of the most important, if not the most important, metrics in valuing the bond.⏎Now, let's look

  2. #2 peak

    eating plan.⏎Beyond promoting every nutrition expert’s mantra of “eat more vegetables and fruits”, the standard raw food diet is true to its name – most, if not all, foods need to be “raw”. This means dieters are prohibited from

  3. #3 peak

    0 years at one level of relative sea-level rise, when they should be building projects that can function for several generations as sea level rises twice as high, if not higher.”⏎Garret Graves, head of the state Coastal Planning and Protection Authority,

  4. #4 peak

    tanks, large flags, chest thumping marches, discipline-for-itself and martyr walls help inspire nationalism and patriotism? This becomes a particularly good question, if not a surprising one, because, historically speaking, ideas of nationalism and patriotism have never

  5. #5 peak

    <|endoftext|>Climate change is one of, if not the biggest challenge we all currently face. It can be difficult and confusing to know

  6. #6 peak

    and shouldn’t!) take my word for it – the following is from a very recent review by leaders in this field:⏎"...epigenetic inheritance is usually—if not always—associated with transposable elements, viruses, or transgenes and may be

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: another, more, another more, more is more, MORE AND MORE MORE. MORE AND MORE. MORE AND MORE. MORE AND MORE. MORE AND MORE NMORE. MORE AND MORE NMORE. MORE AN MORE AN MORE. MORE AN

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

    , religion, the right to hold public office and the right to vote. Also a place where government is by the people and for the people. The United Kingdom is a representative democracy. This means that everyone over

  2. #2 peak

    1988413967-1-768x146-1-300x57.png 0 0 admin https://

  3. #3 peak

    Mother of Sons⏎I will go into the Mother of Seven Sons first. As far as I can tell, they both sprang from the same War to End all Wars, against the Greek King Antiochus Epiphanes who decided

  4. #4 peak

    well as improve their athletic skills. They work cooperatively playing soccer, badmitton, frisbee, bean bag bowling and other games.⏎17 Views 20 Downloads<|endoftext|>In Key Stage 1 the children will:⏎The five enquiry

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. L12:2662 excites
  2. L10:2859 excites
  3. L8:2278 suppresses
  4. L12:2132 suppresses
  5. L12:1481 excites
  6. L12:2983 excites
  7. L9:1655 excites
  8. L6:923 suppresses
  9. L6:1332 excites
  10. L12:2418 suppresses
  11. L12:1244 weakly excites
  12. L12:2770 weakly excites
  13. L10:1387 weakly excites
  14. L3:1537 weakly excites
  15. L0:567 weakly excites
  16. L4:182 weakly suppresses
  17. L8:991 weakly suppresses
  18. L10:2314 weakly excites
  19. L12:136 weakly suppresses
  20. L5:2049 weakly suppresses
  21. L11:2684 weakly suppresses
  22. L3:716 weakly excites
  23. L0:60 weakly excites
  24. L0:2153 weakly suppresses

Feeds later layers

  1. L18:1882 strongly suppresses
  2. L16:1921 strongly suppresses
  3. L15:2186 suppresses
  4. L17:1679 suppresses
  5. L17:2572 suppresses
  6. L18:1765 suppresses
  7. L15:2808 suppresses
  8. L20:516 suppresses
  9. L15:2915 suppresses
  10. L18:2772 suppresses
  11. L15:1818 suppresses "own" in reflexive phrases
  12. L18:2113 suppresses
  13. L19:2642 suppresses
  14. L18:1572 weakly suppresses
  15. L15:3069 weakly suppresses
  16. L17:444 weakly suppresses
  17. L17:129 weakly suppresses
  18. L15:1173 weakly suppresses
  19. L15:1037 weakly suppresses
  20. L16:1408 weakly excites
  21. L17:1927 weakly suppresses
  22. L17:429 weakly excites
  23. L25:170 weakly suppresses
  24. L16:1868 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 -0.97 and -0.854 values between -0.74 and -0.6311 values between -0.63 and -0.5140 values between -0.51 and -0.40207 values between -0.40 and -0.29536 values between -0.29 and -0.171089 values between -0.17 and -0.061140 values between -0.06 and 0.06594 values between 0.06 and 0.17196 values between 0.17 and 0.2855 values between 0.28 and 0.4015 values between 0.40 and 0.514 values between 0.51 and 0.635 values between 0.63 and 0.742 values between 0.74 and 0.852 values between 0.85 and 0.971 values between 0.97 and 1.081 values between 1.08 and 1.191 values between 1.65 and 1.761 values between 2.22 and 2.331 values between 2.78 and 2.901 values between 3.47 and 3.58offnegativepositiveoften

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. otive
  2. oting
  3. ote
  4. 飙
  5. gest
  6. izontal
  7. itness
  8. ation
  9. isory
  10. ulated

Suppresses

  1. `;⏎
  2. `);⏎
  3. .`);⏎
  4. `);⏎⏎
  5. `;⏎⏎
  6. "`⏎⏎
  7. 涤
  8. 处处
  9. --;⏎⏎
  10. .${