Neuron Explorer

Qwen3-0.6B

digits inside numbers, not number words

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 reacts to a single digit in the middle of a number, such as the 5 in 45 or the zeros in 1,500 or 25,000. It goes quiet, or pushes the other way, on spelled-out large-number words like "hundreds" or "thousands". Turning it up or down did not give a clear new theme in the sample text.

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

    ing.⏎location of the weather station site to be surveyed in advance. Sites are listed by name, and give an approximate lat/long to the nearest⏎hundredth of a degree, but

  2. #2 peak

    Which medical conditions medical marijuana treats⏎- The impact smoking or vaping has on the lungs and other body systems⏎- Four engaging marijuana prevention units, each 45 minutes long:⏎- Marijuana and the brain⏎- Marijuana and the body⏎-

  3. #3 peak

    walking distance of this marker. Montrose Park (about 400 feet away, measured in a direct line); Watching the Flames (about 700 feet away); Dumbarton Oaks (about 700 feet away);

  4. #4 peak

    ,034.⏎Eureka is the largest coastal city between San Francisco and Portland, and the westernmost city of more than 25,000 residents in the 48 contiguous states. It is the regional center for government

  5. #5 peak

    reduced by 500 to 750 calories. In general:⏎- Eating plans that contain 1,200–1,500 calories each day will help most women lose weight safely.⏎- Eating plans that contain

  6. #6 peak

    members of the population consume no more than 2,300 mg of sodium.*** The American Heart Association recommends no more than 1,500 mg

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: The first time, it was a question of a single person, but if you have more than one, or three, or five, or even one, or two, or three, or four, or five, then it's something,

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

    Sun inside them! Stars only appear small in the nighttime sky because they are so far away.⏎In the long ago, astronomers once thought that stars numbered in the hundreds or thousands, instead of the now-estimated “billion trillions.” About

  2. #2 peak

    , impossibly large slice of silicon that does it all. Apple’s most powerful chip to date has 114 billion transistors packed into over a hundred processing cores dedicated to logic, graphics, and artificial intelligence, all of it connected

  3. #3 peak

    dots written in layers that are only a few nanometers thick that cover the surface of millions of saucer-sized spinning disks. These hard disks are stacked by the thousands in server farms worldwide.⏎In recent years, the technology for growing uniform magnetic layers

  4. #4 peak

    hundred twenty-four years after it was first proposed, Kenneth Appel and Wolfgang Haken cleverly reduced the problem to checking the properties of 1,936 maps by computer. The result was hundreds of pages of hand analysis combined with thousands

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. L8:41 suppresses
  2. L5:1109 excites
  3. L3:1129 suppresses
  4. L7:415 weakly suppresses
  5. L6:586 weakly suppresses
  6. L8:1939 weakly suppresses
  7. L5:1662 weakly suppresses
  8. L7:1313 weakly excites
  9. L7:1643 weakly suppresses
  10. L4:1263 weakly excites
  11. L4:1845 weakly excites
  12. L4:326 weakly suppresses
  13. L4:97 weakly suppresses
  14. L3:202 weakly suppresses
  15. L1:2110 weakly suppresses
  16. L5:1671 weakly excites
  17. L3:2436 weakly suppresses
  18. L7:2580 weakly excites
  19. L7:2343 weakly suppresses
  20. L8:780 weakly excites
  21. L8:2079 weakly excites
  22. L3:1771 weakly excites
  23. L7:1506 weakly excites
  24. L5:1432 weakly excites

Feeds later layers

  1. L10:2300 suppresses
  2. L12:2284 suppresses
  3. L13:2603 excites
  4. L12:2116 weakly suppresses
  5. L10:1031 weakly suppresses
  6. L12:217 weakly excites
  7. L23:104 weakly suppresses
  8. L21:2833 weakly suppresses
  9. L12:824 weakly suppresses
  10. L10:2791 weakly suppresses
  11. L11:1783 weakly suppresses
  12. L25:1436 weakly excites
  13. L13:2563 weakly suppresses
  14. L10:2499 weakly suppresses
  15. L13:2679 weakly suppresses
  16. L15:2708 weakly suppresses
  17. L12:947 weakly suppresses
  18. L14:2379 weakly suppresses
  19. L16:1410 weakly excites
  20. L11:1016 weakly excites
  21. L19:2291 weakly suppresses
  22. L14:2765 weakly suppresses
  23. L11:301 weakly suppresses
  24. L25:2573 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 -5.08 and -4.911 values between -4.04 and -3.871 values between -2.13 and -1.962 values between -1.44 and -1.262 values between -1.26 and -1.092 values between -1.09 and -0.922 values between -0.92 and -0.746 values between -0.74 and -0.5721 values between -0.57 and -0.40177 values between -0.40 and -0.22900 values between -0.22 and -0.051760 values between -0.05 and 0.12849 values between 0.12 and 0.30154 values between 0.30 and 0.4713 values between 0.47 and 0.6510 values between 0.65 and 0.822 values between 0.82 and 0.992 values between 0.99 and 1.171 values between 1.17 and 1.341 values between 1.34 and 1.511 values between 1.69 and 1.86offnegativepositiveoften

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. VA
  2. hip
  3. اظ
  4. AYOUT
  5. itti
  6. inea
  7. преимуществ
  8. @property
  9. па
  10. alance

Suppresses

  1. mers
  2. atology
  3. zens
  4. umd
  5. odic
  6. 团长
  7. zos
  8. 重重
  9. zell
  10. aptcha