Neuron Explorer

Qwen3-0.6B

equals signs in maths and equations

A neuron in layer 9 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 equals sign in arithmetic and algebra, such as "24 = 4 times 6" or "15+19 =". Turning it up pushed the model toward number puzzles and counting, though the match is loose.

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

    , 24 = 4 ⋅ 6. So to solve the second part, on

  2. #2 peak

    - I'll do that 1/5 in purple -- this is going to be equal to 1/5 --⏎- I hope you don't hear that crow<|endoftext|>The

  3. #3 peak

    correct answer (ex: 15+19=??? or what is 12 divided by 3???). This game is

  4. #4 peak

    urgy celebrated?⏎- When is the liturgy celebrated?⏎- Where is the liturgy celebrated?<|endoftext|>A particle is moving along the parabola 4y = (x + 2)^2 in such a way that its x-coordinate is

  5. #5 peak

    ,” the report said.<|endoftext|>Determining if a Product is Positive or Negative⏎Recall the following properties of multiplication of signed numbers:⏎(positive)(positive) = positive⏎(positive)(negative) = negative⏎(negative)(negative) =

  6. #6 peak

    – 39 = 611 candles.⏎Using a bar diagram⏎In the bar diagram,

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: it's not a problem that you can be a computer is not a problem that you can be a person is not a problem that you can be a robot is not a problem that you can be a person is not a person is not a

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

    François Mitterand who, in 1954, defiantly pronounced that ‘L’Algérie, c’est la France’. Nicolas Sarkozy, on the other hand, dramatically distanced himself

  2. #2 peak

    world’s most prosperous countries with strong international trading links (its port is one of the world’s busiest in terms of tonnage handled) and with per capita GDP equal

  3. #3 peak

    Important numbers!⏎xs = [6, 9, 13, 42]⏎-- Without library⏎f :: Bool -> [Int]⏎f b = if b then xs else ⏎-- With library⏎g :: Bool -> [Int

  4. #4 peak

    according to a report by the United Nations Population Fund.⏎Baby No 7 Billion will probably be born in the Asia-Pacific region, where the population growth rate is the highest in the world.⏎China's family planning policy, which limits most mainland couples

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. L4:24 strongly excites
  2. L3:76 suppresses
  3. L0:423 suppresses
  4. L6:2237 excites
  5. L8:312 suppresses
  6. L4:2069 excites
  7. L8:469 suppresses
  8. L4:20 suppresses
  9. L4:1197 weakly suppresses
  10. L6:654 weakly excites
  11. L3:1213 weakly excites
  12. L7:1171 weakly excites
  13. L6:1037 weakly suppresses
  14. L6:1974 weakly excites
  15. L7:350 weakly excites
  16. L8:1650 weakly excites
  17. L1:1430 weakly suppresses
  18. L6:2065 weakly excites
  19. L6:777 weakly suppresses
  20. L8:137 weakly suppresses
  21. L5:1016 weakly excites
  22. L3:648 weakly suppresses
  23. L5:1030 weakly suppresses
  24. L6:564 weakly excites

Feeds later layers

  1. L17:1066 excites
  2. L10:2144 excites
  3. L10:647 excites
  4. L17:439 excites
  5. L18:841 excites
  6. L14:827 excites open bracket after a value or citation
  7. L11:2292 excites 'than' in comparisons
  8. L11:2066 excites
  9. L16:2028 excites
  10. L12:2171 excites
  11. L14:952 excites
  12. L13:287 suppresses
  13. L17:882 suppresses
  14. L15:1847 excites
  15. L11:681 excites
  16. L17:215 excites
  17. L12:2562 excites
  18. L10:205 suppresses
  19. L19:835 excites
  20. L10:206 excites colon or equals sign before a definition
  21. L14:924 excites
  22. L10:759 excites
  23. L14:1326 suppresses
  24. L12:2548 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.

16 values between -0.59 and -0.4770 values between -0.47 and -0.34351 values between -0.34 and -0.221270 values between -0.22 and -0.091961 values between -0.09 and 0.03216 values between 0.03 and 0.1616 values between 0.16 and 0.291 values between 0.29 and 0.413 values between 0.41 and 0.541 values between 0.54 and 0.661 values between 0.79 and 0.911 values between 1.91 and 2.041 values between 4.29 and 4.42offnegativepositiveoften

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. immers
  2. 一定程度
  3. 的一员
  4. ensual
  5. るもの
  6. immer
  7. 某种程度
  8. 于是我
  9. oyer
  10. 协会会员

Suppresses

  1. .ts
  2. 他是
  3. ��
  4. ADM
  5. 岸
  6. atatype
  7. ('@
  8. ature
  9. ucc
  10. co