Neuron Explorer

Qwen3-0.6B

"of" after a percentage or share

A neuron in layer 19 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 "of" right after a percentage or share, as in "2.5% of Earth's water" or "44% of these people." It does not react much to other uses of the word. Turning it up mostly made the model's text repeat words like "who who who," so the steering tells us little about its job.

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

    One in nine people on Earth don’t have access to safe water, and one in three don’t have access to a toilet. Only 2.5% of Earth’s water is freshwater (water that we can use), and we only have

  2. #2 peak

    .⏎In 2016, 27.6 million people ages 0–64 did not have health insurance. About 44% of these people were white, 33% were Hispanic, 15%

  3. #3 peak

    computer programming in high school.⏎Neither did most of the professionals who come to speak to her classes about their careers. Across the country, just 10 percent of high schools offer computer science classes, according to a website promoting Computer Science Education Week

  4. #4 peak

    33 percent) than were nonveterans (19 percent).⏎Among Gulf War-era II veterans in 2009, 31 percent of veterans aged 18 to 24 years were enrolled in school. As

  5. #5 peak

    of people who underwent a Roux-en-Y gastric bypass weight loss surgery developed an alcohol-abuse problem compared to the 11 percent of those who underwent gastric banding.⏎“We knew there was an increase in the number

  6. #6 peak

    nea include older age, male sex, and higher BMI. In a new study of 1448 people living with Parkinson's, 45% of participants experienced sleep apnea. In addition to the general risk factors, for people

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: who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who who

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

    circuit that you will be connecting into before beginning work. Double check to make sure the circuit is actually safe to work on by using a noncontact voltage tester. In many cases, more than one circuit may run through a lighting outlet box or through

  2. #2 peak

    . Medications, such as antihistamines, may be recommended to relieve symptoms only if it can be determined that the hives are caused by allergy. In cases where hives are not related to

  3. #3 peak

    5 from the number in your head. [Pause again]⏎- Now, think of the letter in the alphabet that corresponds with the number you are thinking about. For instance, if you are thinking of the number 1, it would be "

  4. #4 peak

    within the joint. Your doctor also will evaluate your knee's range of motion and will maneuver your knee to see whether your meniscus is sensitive to pressure. For example, in the "McMurray

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. L14:2223 suppresses
  2. L17:3002 suppresses
  3. L14:464 excites
  4. L18:1233 excites
  5. L18:1522 excites
  6. L17:2004 excites
  7. L14:2739 excites
  8. L16:1759 excites
  9. L15:2439 suppresses
  10. L10:1851 excites
  11. L15:2060 excites
  12. L18:1238 suppresses
  13. L18:172 suppresses
  14. L13:1829 suppresses
  15. L13:832 excites
  16. L17:1432 suppresses
  17. L12:1218 weakly excites
  18. L17:1944 weakly suppresses
  19. L18:2316 weakly excites
  20. L17:2681 weakly excites
  21. L15:2850 weakly excites
  22. L12:1446 weakly suppresses
  23. L18:1547 weakly excites
  24. L13:2122 weakly excites

Feeds later layers

  1. L20:1533 suppresses
  2. L21:858 suppresses
  3. L21:2325 suppresses
  4. L20:880 suppresses
  5. L20:1771 suppresses
  6. L20:2338 suppresses
  7. L20:2580 suppresses
  8. L21:519 suppresses
  9. L23:1742 suppresses
  10. L20:1737 suppresses
  11. L25:2146 suppresses
  12. L21:1977 suppresses
  13. L20:2066 suppresses
  14. L21:665 suppresses
  15. L21:1296 weakly suppresses
  16. L20:2028 weakly suppresses
  17. L23:2969 weakly suppresses
  18. L20:1747 weakly suppresses
  19. L21:2234 weakly suppresses
  20. L20:1837 weakly suppresses
  21. L20:1766 weakly suppresses
  22. L21:1550 weakly suppresses
  23. L26:1228 weakly suppresses
  24. L26:670 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 -9.76 and -9.362 values between -7.35 and -6.951 values between -6.55 and -6.151 values between -6.15 and -5.751 values between -5.75 and -5.352 values between -5.35 and -4.955 values between -4.95 and -4.551 values between -4.15 and -3.755 values between -3.75 and -3.359 values between -3.35 and -2.959 values between -2.95 and -2.5415 values between -2.54 and -2.1422 values between -2.14 and -1.7442 values between -1.74 and -1.3453 values between -1.34 and -0.94100 values between -0.94 and -0.54644 values between -0.54 and -0.142294 values between -0.14 and 0.26428 values between 0.26 and 0.66116 values between 0.66 and 1.0670 values between 1.06 and 1.4632 values between 1.46 and 1.8718 values between 1.87 and 2.2714 values between 2.27 and 2.6710 values between 2.67 and 3.071 values between 3.07 and 3.475 values between 3.47 and 3.873 values between 4.27 and 4.672 values between 4.67 and 5.071 values between 5.07 and 5.471 values between 5.87 and 6.27offnegativepositiveoften

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. ziehung
  2. 这三个
  3. 这两个
  4. bulunduğu
  5. 폰
  6. 基石
  7. 膺
  8. 专项斗争
  9. 学
  10. ;;;;;;;;;;;;;;;;

Suppresses

  1. 那些
  2. larger
  3. Larger
  4. small
  5. 较大的
  6. alleng
  7. smaller
  8. 不符
  9. 不符合
  10. SMALL