Neuron Explorer

Qwen3-0.6B

percentages and shares in statistics

A neuron in layer 13 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up on the number in a statistic, like the 20 in '20 percent' or the 60 in '60 percent', especially where a share of a whole is being described. Turning it up made the model write in a more generic, repetitive style that didn't clearly relate to numbers.

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

    plaster, clay, or other substance to give it the colour and lustre of bronze. Dutch metal, an alloy of 80 percent copper and 20 percent zinc, is frequently used for bronzing.

  2. #2 peak

    years; 50.5% boys) and 38 who were unexposed (mean age, 6.8 years; 42.1% boys). The researchers used two 24-hour dietary

  3. #3 peak

    bigger than one third.⏎You arrange the data set in ascending order. You then find the observation such that a quarter of the observations are smaller than it and three quarters are bigger. That value is the lower quartile. Next find the observation such

  4. #4 peak

    when thinking about hackers and cyber criminals. While in 2015, 40 percent of attacks stemmed from ‘outsiders’, a surprising 60 percent were actually perpetrated by company insiders.⏎IBM, who produced the figures based on

  5. #5 peak

    , real estate markets are modeled as a stock/flow market. About 98% of supply consists of the stock of existing houses, while about 2% consists of the flow of

  6. #6 peak

    can exhibit soft tissue-like behavior. The nanotube arrays have a porosity of around 95 percent, meaning that carbon nanotubes only occupied five percent of the arrays while 95 percent of the array was occupied by air.⏎

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 can't have a similar effect, if I want to for the next time. So I need to make sure that this time, I can have a similar effect, if it's a different one. So a similar effect. So 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

    2 MHz⏎- Cancer 42 MHz⏎- Negative Thoughts lower our body’s frequency by 12 MHz⏎- Positive Thoughts raise our body’s frequency by 10 MHz⏎-

  2. #2 peak

    . Using these data, the author estimates the number of rental units and ownership units that would be affordable to prototypical households at specified income levels.⏎Previous Article | Next Article<|endoftext|>Scientists at Stanford University have created a way to produce thin, stretchable

  3. #3 peak

    45 right before landing: vx = 2cos25 = 1.8126 vy = 2sin25 = -0.845 3. The attempt at a solution Im not sure if my solution

  4. #4 peak

    Continuous splits are based on a threshold value, such as the split in a regression task where all instances below the threshold value are assigned to one class and all instances above the threshold value are

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. L12:455 suppresses
  2. L12:2518 suppresses
  3. L12:15 suppresses
  4. L12:1960 suppresses
  5. L10:466 excites
  6. L3:2125 weakly excites
  7. L12:2674 weakly excites
  8. L12:2503 weakly suppresses
  9. L3:2036 weakly suppresses
  10. L2:2812 weakly excites
  11. L12:2826 weakly suppresses
  12. L12:606 weakly suppresses
  13. L12:909 weakly excites leading digit of metric conversions
  14. L12:1493 weakly suppresses
  15. L11:1529 weakly suppresses
  16. L11:1220 weakly suppresses
  17. L0:278 weakly suppresses
  18. L12:2591 weakly suppresses
  19. L9:715 weakly suppresses
  20. L12:2968 weakly excites
  21. L12:2945 weakly suppresses
  22. L1:33 weakly suppresses
  23. L11:2392 weakly excites
  24. L12:133 weakly suppresses

Feeds later layers

  1. L15:1563 strongly excites
  2. L15:2446 strongly excites
  3. L19:1343 excites
  4. L17:1062 excites
  5. L17:2957 excites
  6. L14:2418 excites
  7. L14:2521 excites
  8. L16:1920 excites
  9. L18:2084 excites
  10. L14:2445 excites
  11. L17:1132 excites
  12. L15:1921 excites
  13. L14:1799 excites
  14. L15:250 suppresses
  15. L15:2112 excites
  16. L15:449 excites
  17. L15:1037 excites
  18. L14:2375 excites
  19. L14:1165 excites
  20. L15:3013 excites
  21. L19:1989 excites
  22. L15:2060 excites
  23. L14:2117 excites
  24. L15:1032 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 -1.05 and -0.982 values between -0.92 and -0.851 values between -0.85 and -0.793 values between -0.79 and -0.723 values between -0.72 and -0.669 values between -0.66 and -0.5915 values between -0.59 and -0.5322 values between -0.53 and -0.4647 values between -0.46 and -0.40104 values between -0.40 and -0.33145 values between -0.33 and -0.27263 values between -0.27 and -0.20400 values between -0.20 and -0.14554 values between -0.14 and -0.07713 values between -0.07 and -0.01713 values between -0.01 and 0.06444 values between 0.06 and 0.12251 values between 0.12 and 0.19120 values between 0.19 and 0.2561 values between 0.25 and 0.3221 values between 0.32 and 0.386 values between 0.38 and 0.455 values between 0.45 and 0.512 values between 0.51 and 0.582 values between 0.84 and 0.901 values between 1.49 and 1.55offnegativepositiveoften

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. 另一半
  2. addCriterion
  3. ева
  4. remaining
  5. complement
  6. 젯
  7. .swt
  8. )__
  9. vette
  10. وا

Suppresses

  1. stag
  2. ounced
  3. ANGED
  4. 進
  5. tober
  6. }`).
  7. izon
  8. regist
  9. Skill
  10. angu